University of South Florida
Scholar Commons
Graduate Theses and Dissertations
Graduate School
January 2014
Development of an efficient human hepatocyte
culture platform for assessing novel therapeutic
efficacy against Plasmodium liver parasites
Steven Patrick Maher
University of South Florida, [email protected]
Follow this and additional works at: http://scholarcommons.usf.edu/etd
Part of the Biology Commons, Biomedical Engineering and Bioengineering Commons, and the
Public Health Commons
Scholar Commons Citation
Maher, Steven Patrick, "Development of an efficient human hepatocyte culture platform for assessing novel therapeutic efficacy against
Plasmodium liver parasites" (2014). Graduate Theses and Dissertations.
http://scholarcommons.usf.edu/etd/5263
This Dissertation is brought to you for free and open access by the Graduate School at Scholar Commons. It has been accepted for inclusion in
Graduate Theses and Dissertations by an authorized administrator of Scholar Commons. For more information, please contact
[email protected].
Development of an Efficient Human Hepatocyte Culture Platform for Assessing Novel
Therapeutic Efficacy against Plasmodium Liver Parasites
by
Steven P. Maher
A dissertation submitted in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy in Public Health
with a concentration in Global Communicable Diseases
Department of Global Health
College of Public Health
University of South Florida
Co-Major Professor: John H. Adams, Ph.D.
Co-Major Professor: Wajeeh M. Saadi, Ph.D.
Dennis E. Kyle, Ph.D.
Thomas R. Unnasch, Ph.D.
Joe D. Cuiffi, Ph.D.
Date of Approval:
July 10, 2014
Keywords: Plasmodium, liver, hepatocyte, hypnozoite, multiplexed platform
Copyright © 2014, Steven P. Maher
Acknowledgments
I am incredibly grateful to Dr. John Adams, Dr. Dennis Kyle, Dr Wajeeh Saadi, Dr. Joe
Cuiffi, Dr. Tom Unnasch, and Dr. Michael Ferdig for their many contributions over the past 5-10
years. This project was the perfect opportunity when it came time to work toward a PhD, and it
was made possible by Dennis, Wajeeh, John and Joe. The Draper Lab Fellow program is a
wonderful program for higher education; this project was supported by a DLF and the Bill &
Melinda Gates Foundation (Grant OPP1023601). Thanks also to Lisa Checkley, Doug Shoue,
Dr. Bharath Balu, and Dr. Naresh Singh for teaching me how to do Plasmodium research.
Thanks to the many graduate students who came and went since we first moved to USF in July
2007; the parties, soccer games, and general comradely will be missed. The data in this
dissertation required thousands of hours of planning and execution—it would have been
impossible without the help of Alison Roth, Rick Crouse, Amy Conway, Emilee Bannister, Dr.
Anil Achyuta, Dr. Shulin Xu, Dr. Steve Sun, Sandi Kennedy, Sam Barnes and Lynn Dong.
Thanks to all the teachers in my life. Hundreds of people have imparted their dedication and
wisdom on me throughout the years, and their work is reflected in this dissertation. Most
importantly, thank you to my wife Julia, for staying patient and keeping me healthy and happy
despite the stress of graduate school. Thanks to my mother, father, sister and brother, for their
unwavering support.
Table of Contents
List of Tables…………………………….. .....................................................................................v
List of Figures……………………………….. .............................................................................. vi
Abstract……………………………………… ............................................................................ viii
Chapter One: Introductions…………….. ........................................................................................1
Malaria—An Ancient and Global Public Health Problem ..................................................1
A Brief History of Discovery...................................................................................3
Plasmodium Taxonomy and Evolutionary History..................................................4
The Plasmodium Life Cycle... .................................................................................8
Current Prophylaxis and Standard of Care ............................................................14
Control and Eradication Efforts .............................................................................16
Vector Control ...........................................................................................16
Drugs………….. ........................................................................................17
Vaccines………….. ...................................................................................18
Liver Stage Drug and Vaccine Development.. ......................................................20
Need for a More Efficient in vitro Plasmodium Liver Model.. .................22
Introduction to in vitro Liver Models……….. ..................................................................22
Role of Liver Models in Public Health.. ................................................................22
Animal-Free ADMET…. ...........................................................................23
Cytochrome P450 Enzymes. ..........................................................25
Infectious Liver Disease Culture and Research. ........................................27
Architecture of the Liver ........................................................................................27
Zone-Specific Liver Physiology… ............................................................29
Cell Types of the Liver.. ............................................................................29
Hepatic Cell Resources for Research…. ........................................31
Considerations for Designing in vitro Liver Models… .........................................34
Advanced Cell Culture Platform Fabrication ............................................38
Current Progress: Static Liver Systems .................................................................39
Current Progress: Perfused Liver Systems ............................................................42
Comparing and Contrasting Components and Systems .........................................42
Tailoring Liver System for Experimental Question...............................................44
Focus of Study………………….. .....................................................................................44
References……….................................. ............................................................................49
i
Chapter Two: Development of a Long-Term, Mimetic Human Liver Model ...............................67
Rationale of Study……………………..............................................................................67
Materials and Methods………………….. .........................................................................69
Hepatic Cell Culture…………. .............................................................................69
Live Cell Staining and Quantification ...................................................................71
Hepatic Cell Phenotyping…… ..............................................................................73
Co-culture Seeding and Culture .............................................................................74
Transwell Seeding and Culture ..............................................................................75
Microfluidic Device Fabrication ............................................................................75
Microfluidic Device Culture Initiation.. ................................................................76
Microfluidic Device Extended Use and Adaptation ..............................................77
Results………………………………… ............................................................................78
Hepatic Cell Candidate Genotypes and Phentoypes…. .........................................78
Generation and Optimization of Cell Tracking, Cell Quantification,
Phenotyping, and Co-culture Protocols.. .........................................................79
Assessment of Co-culture Contributions and Drawbacks .....................................81
Microfluidic Device Development and Optimization............................................82
First Generation Device Successes and Failures .......................................82
Device Perfusion, Short and Long-Term .......................................83
Device Vascularization ..............................................................................85
Third Generation Device Improvements.. ................................................ 85
Successful Long-Term Culture of Primary Human Hepatocytes ..........................86
Identification of Cell Compaction as Essential Characteristic ..................87
Further Optimizations on the Basic Liver Model ..................................................87
Discussion…………………………………. .....................................................................88
Weighing Primary versus Immortalized Cells For Liver Models ..........................88
Practicality of Co-cultures for Liver Models .........................................................89
Throughput of Transwells, Devices, Static and Perfused Systems ........................90
Dichotomy of Complexity and Scalability ............................................................91
Future Studies…………………… ........................................................................92
References………………………………………. ...........................................................108
Chapter Three: Identification of Critical Features Necessary for Plasmodium Liver Stage
Culture………………………………………........................................................................113
Rationale of Study……………………............................................................................113
Materials and Methods………………….. .......................................................................114
Propagation, Isolation and Cryopreservation of Plasmodium Sporozoites .........114
Sporozoite Quality Control…… ..........................................................................116
Coordinating Device Fabrication, Device Setup, and Sporozoite
Production ………………..............................................................................117
Device-Based Sporozoite Invasion and Development Assay ..............................117
Liver Stage Immunofluorescent Assay ....................................................118
Liver Parasite Qualitative and Quantitative Assessment .........................119
Results………………………………… ..........................................................................119
Reproduction of Standard P. berghei and P. falciparum Liver Stage
Assays…………………….. ..........................................................................119
ii
Sporozoite Quality Control Assays .....................................................................120
Protocol Development for Parasite Imaging ........................................................121
Identification of Binary and Regression-Based Parameters for Liver Stage
Invasion and Development ............................................................................122
Plasmodium falciparum and P. vivax Sporozoite Sourcing.....................122
Primary Hepatocyte Donor Screen ..........................................................123
Ascertaining Antimycotic Activity on Liver Stage Parasites ..................124
Exploring Nonparenchymal Effects on Liver Invasion Rates .................124
Establishment of High Throughput MOI .................................................127
Successful P. falciparum and P. vivax Liver Stage Development .......................127
Discussion…………………………………. ...................................................................128
Identification of Key Assay Factors ....................................................................128
Identification of Essential Liver Stage Growth Factors .......................................128
Future Studies…………………. .........................................................................129
References………………………………………. ...........................................................144
Chapter Four: Determination of Antiparasitic Therapeutic Efficacy within a Microfluidic
Plasmodium Liver Stage Culture Platform… ........................................................................148
Rationale of Study……………………............................................................................148
Materials and Methods………………….. .......................................................................149
Review of the Optimized Microfluidic Device Based Liver Stage Protocol .......149
Drug Assay Overview and Protocol ....................................................................150
Vaccine Assay Setup and Protocol ......................................................................152
Generation of mCherry-lucifersase-Expressing P. falciparum SporozoiteProducing Lines................. ............................................................................152
Results………………………………… ..........................................................................154
Basic Protocol Development with P. berghei-luciferase and Hepatic Cell
Lines…………………………. ......................................................................154
Inhibition of P. falciparum and P. vivax Development with Primaquine ............155
Inhibition of P. falciparum Development with Inhibitory Antibody...................156
Liver Culture Optimization for Multiplexing ......................................................157
Materials and Fabrication Selection.........................................................157
Identification of Confining Geometries ...................................................157
Development of a Self-Reporting P. falciparum Line .............................159
Rapid IFA Quantification via High Content Imaging..............................159
Discussion…………………………………. ...................................................................160
Importance of Optimization to Meet Parasite Requirements ...............................160
Efficiency of Assays…............ ............................................................................160
Multiplexed Device Requirements ......................................................................162
Logistics of Sporozoite Sourcing .........................................................................163
Future Studies………………………… ..............................................................164
References………………………………………. ...........................................................172
Chapter Five: Conclusions………………………. ......................................................................174
Summary of Findings…………………………………… ...............................................174
Recent Advances in Human Plasmodium Liver Stage Culture .......................................177
iii
Comparison of Drug and Vaccine Platforms .......................................................178
Implications of This Work……………….. .....................................................................178
References………………………….. ..............................................................................179
iv
List of Tables
Table 3.1. Binary and regression factors to be optimized for liver stage culture. ......................138
Table 3.2. Primary hepatocyte donor comparison ......................................................................140
Table 3.3. Summary of Multiplicity of Infection across all experiments ...................................142
v
List of Figures
Figure 1.1. Global distribution of P. falciparum and P. vivax malaria. ......................................46
Figure 1.2. The Plasmodium life cycle…………........................................................................47
Figure 1.3. Structure of the liver and Plasmodium infection ......................................................48
Figure 2.1. Third-generation microfluidic device fabrication and properties .............................93
Figure 2.2. Hepatocyte and endothelial phenotypes....................................................................94
Figure 2.3. Cell count optimizations—CellTiter-Glo®, Hoechst, and Trypsin ...........................95
Figure 2.4. Cell count optimizations—DNA quantification and PrestoBlue® ............................96
Figure 2.5. Generation of co-culture methods .............................................................................97
Figure 2.6. Generation of complex co-culture methods ..............................................................98
Figure 2.7. Co-culture effect on hepatocyte phenotype ..............................................................99
Figure 2.8. First generation device seeding and high resolution microscopy ...........................100
Figure 2.9. Continuous perfusion in device troubleshooting ....................................................101
Figure 2.10. Full device optimization for hepatocyte line perfused and static culture ..............102
Figure 2.11. Device vascularization…................ ........................................................................103
Figure 2.12. Device dimension and membrane optimization ......................................................104
Figure 2.13. Long-term culture of primary human hepatocytes..................................................105
Figure 2.14. Confinement leads to stable primary hepatocyte phenotype ..................................106
Figure 2.15. Further device optimization……… ........................................................................107
Figure 3.1. Sexual stage culture workflow…... .........................................................................131
vi
Figure 3.2. Full liver stage assay workflow…. .........................................................................132
Figure 3.3. Invasion and development quantification ...............................................................133
Figure 3.4. Reproduction of P. falciparum and P. berghei liver models ..................................134
Figure 3.5. Sporozoite traversal assay............ ...........................................................................135
Figure 3.6. Sporozoite gliding and flow cytometry invasion assay ..........................................136
Figure 3.7. IFA validation and optimized deconvolution imaging ...........................................137
Figure 3.8. Identification of candidate P. falciparum parasite line, stable parasite
sourcing, and effect of antimycotics on parasites ...................................................139
Figure 3.9. Testing regression parameters on liver stage growth. .............................................141
Figure 3.10. Plasmodium falciparum and P. vivax liver stage development ..............................143
Figure 4.1. mCherry-luciferase construct cloning strategy and transfection ............................165
Figure 4.2. Plasmodium berghei, P. falciparum, and P. vivax drug assays ..............................166
Figure 4.3. Antibody-specific inhibition of P. falciparum liver stage development.................167
Figure 4.4. Identification of channel geometries for stable NON phenotypes ..........................168
Figure 4.5. Preliminary studies on microwells for multiplexed liver platform .........................169
Figure 4.6. mCherry-luciferase construct and P. falciparum transformation ...........................170
Figure 4.7. Necessary optimization for high-throughput IFA readout ......................................171
vii
Abstract
Malaria is a critical and global public health problem, affecting over 200 million people
every year, resulting in over 500,000 deaths. A vaccine is not currently available and only one
drug, primaquine, is effective against the dormant stages of Plasmodium vivax. Preclinical
assessment of novel therapeutic drugs and vaccines is hampered by the lack of an in vitro liver
model for P. falciparum and P. vivax. To provide a stable human hepatocyte-based culture
platform for parasite development, we engineered a microfluidic bilayer device capable of both
simple and complex culture methods, including perfusion and co-culture, to better understand the
requirements of both hepatocytes as host cells and parasite liver stage development. As only the
mechanical compaction of the device channels was found necessary for stable hepatocyte
culture, and only a validated host cell lot was found necessary for improved parasite
development rates, we present an efficient and simple Plasmodium liver model capable of
supporting P. vivax dormant forms. Device liver platforms were used to generate kill curves of
P. vivax and P. falciparum after three days under primaquine drug pressure. Furthermore, an
anti-CSP antibody-based inhibition of development assay with P. falciparum successfully
demonstrated antibody-specific liver stage inhibition. Ongoing studies aim to identify the
minimal unit of the culture system necessary to be multiplexed in a fully functional and efficient
drug and vaccine discovery platform.
viii
Chapter One:
Introduction
Malaria—An Ancient and Global Public Health Problem
Parasitic protozoa have fundamentally affected humanity to the point that many details of
our evolution are the result of host-parasite interactions. Our interaction with Plasmodium dates
prior to the split between chimpanzees and hominids, and has resulted in more morbidity and
mortality than any other infectious disease (Silva et al., 2011). Over evolutionary history
endemic human populations have evolved to survive malaria through selection of red blood cell
disorders such as sickle cell anemia and specific expression of some antigens on the red cell
surface. In ancient history, accounts of fevers associated with floods and wetlands describe
malaria limiting the success and range of the Egyptians, Greeks, Romans, Chinese and many
other civilizations (Neghina et al. 2010). In modern history, malaria has altered the course of
world wars, survived a global eradication effort, evolved resistance to most front-line treatments,
and currently threatens over half of the world’s population. We have learned a tremendous
amount of information on Plasmodium and malaria since Charles Laveran first saw parasites in a
blood film in 1880; but taken into context, our research and control efforts over the past 135
years pale in comparison to the millennia Plasmodium has had to refine its host-parasite
interaction. However, the rapid pace of advance in science and technology has started the
conversation that, within our lifetime, we may be able to eradicate humanity’s worst disease
(Cox, 2010; malERA, 2011b; Roberts, p141).
1
According to the World Health Organization, there were between 125-287 million cases
of malaria in 2012, resulting in 473-627 thousands deaths. Malaria is a disease that in humans is
caused by five species of parasitic protozoa: Plasmodium falciparum, P. vivax, P. malariae, P.
ovale and zoonotic infection of P. knowlesi. Malaria is transmitted by several species of
Anopheles mosquitoes, which feed at night (WHO, 2014). While we understand the causes of
malaria and treatments for disease, malaria is endemic to many parts of the world where people
and governments do not have enough resources to control mosquito breeding or diagnose and
treat infected persons (Noble, p90). Improved surveillance, diagnosis, and public health
measures such as indoor spraying and insecticide-treated bed nets have resulted in a 40%
decrease in malaria deaths over the past 15 years, but a combination of continued surveillance,
vector control, and therapeutic licensing will be needed to continue malaria control efforts
(malERA, 2011b). In the United States, endemic malaria was eliminated by 1951, coordinated
by the Office of Malaria Control in War Areas (later becoming the Centers for Disease Control
in Atlanta, Georgia) by application of DDT insecticide to some 4.6 million homes and proper
drainage of swamp areas (CDC, 2012b). Today, over 3.4 billion people are at risk of contracting
malaria. Global malaria endemicity maps show worldwide distribution of P. falciparum, focused
in Sub-Saharan Africa and P. vivax, focused in South America and Southeast Asia (Gething et
al., 2012; Gething et al., 2011).
Morbidity from malaria is caused by one or more factors, including anemia due to lysis of
erythrocytes, pulmonary edema, hyperparasitemia, kidney failure, metabolic acidosis, and
hypoglycemia. Each type of malaria (that is, caused by a different species of Plasmodium) is
further characterized by specific symptoms, especially the periodicity of sudden fever brought on
by synchronous completion of the parasite’s blood cycle, culminating in erythrocyte lysis.
2
Plasmodium falciparum malaria, also known as malignant tertian malaria, accounts for about
half of all malaria cases and is characterized by a two day cycle of fever and can result in
cerebral and placental malaria. Cerebral malaria occurs when parasitized erythrocytes sequester
in the brain, leading to extreme fever, psychosis, convulsions, and death within hours. Placental
malaria occurs by a similar mechanism as cerebral malaria, but parasitized erythrocytes sequester
in the placenta and can lead to complications during pregnancy (Uneke, 2007). Plasmodium
vivax malaria, also known as benign tertian malaria, also features a two day cycle of fever, but is
able to initiate a second round of blood infection after treatment for a primary round of infection,
due to the parasite’s ability to arrest its initial liver infection. Plasmodium ovale malaria, also
known as mild tertian malaria, features a two day cycle of fever and can also arrest in the liver,
but is rare and is chiefly found in Southeast Asia and India. Plasmodium malariae malaria is
also known as quartan malaria, featuring a three day cycle of fever, and has been known to
maintain extremely low and asymptomatic infections arising years after the initial infection
(Roberts, p147-50).
A Brief History of Discovery. In 1847 Johann Meckel first described a black pigment in
the blood of malaria patients, but it was initially thought the pigment was derived from the body
during an infection of malaria. He did not realize the black pigment was hemozoin—the
byproduct of parasitized erythrocyte hemoglobin stored in the parasite’s food vacuole. Charles
Laveran, a French army physician, is regarded as the first person to witness and describe a
malaria parasite when, in November 1880, he observed exflagellation of Plasmodium gametes in
a blood smear. For his discovery, Laveran was awarded the Nobel Prize in 1907. In 1885
Camillo Golgi realized there were different forms of malaria, characterizing tertian and quartan
symptoms as well as differential production of invasive merozoites from a parasitized
3
erythrocyte, for which he was awarded the Nobel Prize in Medicine in 1906. Plasmodium vivax
and P. malariae were named by Giovanni Batista Grassi and Raimondo Filetti in 1890, while in
1897 William H. Welch named the parasite originally discovered by Laveran P. falciparum. At
the turn of the 20th century it was unknown how malaria was transmitted until Ronald Ross, of
the Indian Medical Service, noted formation of parasite oocysts in the midgut of Anopheles
mosquitoes on August 16th, 1987. His early work nearly ended in failure as he had originally
worked with Culex mosquitoes, which do not transmit mammalian Plasmodium parasites. He
won the Nobel Prize in Medicine in 1902 for his work. The parasite’s life cycle was completed in
1898 when Amigo Bignami and Giovanni Grassi allowed Anopheles claviger to feed on human
volunteers, resulting in transmission of P. falciparum, P. vivax, and P. malariae. Of particular
importance to this work, the complete life cycle including pre-erythrocytic liver stages were not
described until 1948 when H.C. Shortt and P.C.C. Garnham described liver forms of P.
cynomolgi in monkeys and P. vivax in humans. Perhaps one of the most important success
stories of malaria history involved William Gorgas’s saving of the Panama Canal through insect
control. When he was placed in charge of sanitation in the Panama Canal Zone, 21,000 of
26,000 builders had been hospitalized for malaria and yellow fever (CDC, 2010; Roberts, p14143).
Plasmodium Taxonomy and Evolutionary History. Taxonomy studies place the Genus
Plasmodium into Kingdom Chromalveolata, Superphylum Alveolata, Phylum Apicomplexa,
Class Aconoidasida, Order Haemosporida, Family Plasmodiidae; Plasmodium features over 200
known species infecting rodents, reptiles, birds, and primates. Plasmodium parasites may have
evolved from Dinoflagellates—a phylum of Alveolata to which marine plankton belong.
Dinoflagellates are characterized by a photosynthetic plastid surrounded by four membranes,
4
derived from a primary endosymbiosis of photosynthetic algae and a bacterium, followed by a
secondary endosymbiosis of the primary bacterium and another bacterium (Keeling, 2009). A
subgroup of parasitic, photosynthetic Dinoflagellate-like protozoa surviving in the intestinal
track of prehistoric marine animals may have eventually lost the ability to perform
photosynthesis as animals became more opaque, leading to the nonphotosynthetic plastid-like
apicoplast organelle found in Plasmodium (Ralph et al., 2004). The Dinoflagellate ability to
form dormant forms, or cysts, is reflected in Apicomplexans as many are also able to form
similar cysts (Markus, 2011). Apicomplexans are further characterized by a predominantly
haploid nucleus, spore formation, and an apical complex containing molecular machinery
necessary to invade host cells (described below). Aside from malaria, Apicomplexan parasites
cause babesiosis, cryptosporidiosis, cyclosporiasis, isosporiasis, and toxoplasmosis. The
Haemosporida family Plasmodiidae are thought to have split from the Haemosporida family
Leucocytozoidae. Leucocytozoidea are characterized as parasites infecting white blood cells and
liver; black flies serve as the invertebrate host (Bennett and Campbell, 1972).
Plasmodium spp. are theorized to have evolved approximately 130 million years ago in
conjunction with the rise of flowering plants supporting speciation of mosquitoes. Phylogenetic
analysis of four genes among several species of three subtypes of Haemosporida
(Leucocytozoon, Plasmodium, and Hepatocytsis), found parasite speciation may have occurred in
conjunction with shifts in the invertebrate host. For example, avian Plasmodium spp. use Culex,
Aedes and Culiseta mosquitoes for transmission but speciation of Plasmodium spp. to mammals
correlates with a switch to Anopheles mosquitoes as vectors. Thus, speciation of mosquitoes,
possibly resulting from changing climates and feeding preferences, resulted in new vertebrate
5
reservoir opportunities for Plasmodium spp. (Killick-Kendrick, 1978; Martinsen et al., 2008;
Valkiunas et al., 2005).
Explanations for the rise of human-specific malaria parasites have changed as genome
sequencing and analysis technology has developed. An early phylogenetic study of 10 genes
from human parasites P. falciparum, P. vivax, P. ovale, and P. malariae found all four to be
loosely related but possibly arriving in humans from four different lateral transfers from
primates, likely before the split between chimpanzees and hominids. The authors concluded P.
falciparum diverged from P. reichenowi approximately 8 million years ago, P. vivax is related to
P. simium and P. knowlesi, and P. malariae is related to a New World monkey parasite P.
brasilianum (Ayala et al., 1999). However, the authors were not able to confirm if the direction
of transfer for these related parasites was from human to monkey, vice versa, or both. Other
studies link the loss of cytidine monophosphate-N-acetylneuraminic acid hydroxlase (CMAH)
function in humans as the result of selection pressure from P. reichenowi and resulted in P.
reichenowi speciation toward P. falciparum. All mammals, except humans, complete synthesis
of sialic acid N-glycolylneuraminic acid (Neu5Gc) from N-acetylneuraminic acid (Neu5Ac) by
CMAH. This sialic acid is the preferred EBA-175 target ligand binding P. reichenowi merozoites
to erythrocytes. A few million years after divergence of the chimpanzee and human lineages, a
mutation abrogated CMAH function and was so strongly selected for it swept throughout the
entire human lineage—possibly due to conferred resistance to P. reichenowi. However, P.
falciparum may have speciated from P. reichenowi as P. falciparum shows preference for EBA175 binding to (incompletely processed) sialic acid Neu5Ac (Martin et al., 2005).
Another theory, coined the Malaria Eve hypothesis, states today’s P. falciparum
diversity is limited and, by measuring the rate of polymorphism in 13 microsatellite markers in
6
over 100 worldwide samples, P. falciparum could only have spread throughout Africa within the
past 10,000 years. The mechanism of this spread could involve the rise of agriculture, climate
change following the end of the most recent glacial period, and increased human interaction with
forests (Leclerc et al., 2004). Proponents of the Malaria Eve hypothesis agree P. falciparum
speciation could have occurred millions of years ago, but was restricted to Africa until this
relatively recent spread, resulting in the evidence for a recent population bottleneck (Rich et al.,
2009).
Much data in these studies were collected prior to the ushering in of the genomic age and
focused on a few dozen loci assessed between a small collection of samples. Completion of the
P. falciparum genome project and development of rapid, global genotyping methods such as
Single Nucleotide Polymorphism (SNP) microarrays have resulted in vastly different theories of
P. falciparum’s arrival in humans (Gardner et al., 2002; Gardner et al., 1999). One study looked
at over 400 polymorphic sites on chromosome 3 from both P. falciparum and P. reichenowi,
including microsatellites and SNP’s, and concluded the mutation rate of noted polymorphisms
placed P. falciparum divergence from P. reichenowi between 100,000 to 180,000 years ago (Mu
et al., 2002). Yet another study draws into question almost all earlier sequence data that P.
falciparum is divergent from P. reichenowi. When over 1000 genes sequenced from 3,000
Plasmodium –infected chimpanzee, gorilla, and human samples were used to form a new
phylogenetic tree, P. falciparum speciation could only have arisen from a recent, single lateral
transfer event from gorillas to humans. However, this study did not rule out the possibility of a
human-to-gorilla P. falciparum transfer event as the source of such similarity between human
and gorilla samples (Liu et al., 2010).
7
The origin of P. vivax infection in humans is also debated; alternative theories suggest P.
vivax split from a macaque or monkey parasite in either Southeast Asia or Africa. A comparison
45 orthologous genes of P. vivax and P. knowlesi suggest P. vivax radiated from P. knowlesi
approximately 7 million years ago and resided in Asian monkeys until a host-switch event when
hominids arrived in Asia (Silva et al., 2011). However, P. vivax’s effect on human evolution is
readily noted in that persons of African descent lack Duffy antigens on erythrocytes, conferring
partial protection against P. vivax erythrocyte invasion (reviewed in Langhi and Bordin, 2006).
Conversely, a comparison of wild chimpanzees and apes in Africa revealed widespread infection
with a P. vivax-like parasite demonstrating much less stringent species-specificity than that of
human parasites. These data suggest P. vivax drove selection for Duffy-negative individuals in
Africa before traveling with hominids to Asia and South America (Liu et al., 2014).
The Plasmodium Life Cycle. Apicomplexans share a generic life cycle: a zygote
produces sporozoites, sporozoites invade new hosts and form merozoites, merozoites continue
the host infection and generate gametes, and gametes sexually recombine to form a zygote.
Apicomplexans require unique molecular machinery for the dramatically different environments
characteristic of each point in the life cycle. These environments include intracellular,
extracellular, and independent habitation as well as interaction with several different types of
host cells featuring various temperatures, salt concentrations and pH. Apicomplexans use these
conditions to signal transformation from one life cycle stage to the next, a process recently
shown to be tightly controlled by a family of plant-derived Apetala (ApiAP2) transcription
factors (Painter et al., 2011).
In Plasmodium, Anopheles mosquitoes are the definitive host (where sexual reproduction
occurs) and vertebrates are the intermediate host (where asexual reproduction occurs—with the
8
caveat that gamete formation occurs in the vertebrate host). Plasmodium contain 14
chromosomes harboring approximately 5000 genes and spends nearly the entire life cycle as
haploid; the parasite is diploid for a matter of minutes after gametes fuse in the mosquito midgut
before meiosis generates haploid cysts. Merozoites, sporozoites, and oökinetes contain one
plastid and one mitochondrion, which are replicated along with the parasite’s nucleus before
being packaged together for cytokinesis (Roberts, p143). The Plasmodium life cycle is
biologically fascinating as it repeatedly utilizes a theme of invasion followed by rapid,
exponential reproduction of its genome and karyokinesis, followed by cytokinesis and
destruction of the host cell containing the developing form to release new invasive forms.
Figure 1.2 summarizes the Plasmodium life cycle. When an infected Anopheles mosquito
feeds, a few dozen sporozoites are released into the dermis, and only a small percentage make
their way to the blood; those that do not can begin transformation into liver forms in lymph
nodes (Amino et al., 2006). Sporozoite motility is required to travel from the bite location to a
nearby blood vessel, and entry into vessels is required for progression to liver invasion.
Disruption of microneme proteins sporozoite microneme protein essential for cell traversal
(SPECT)-1 and 2, phospholipase, or cell traversal protein for oökinetes and sporozoites
(CelTOS) results in skin arrest (Kaiser et al., 2004). Sporozoites travel in the circulation but are
targeted to the liver. Liver arrest has been heavily studied and different models for liver invasion
have been described; confirmation of these models is important as the cell and proteins
responsible for liver arrest and invasion likely contain vaccine targets. It has been shown that the
binding of circumsporozoite protein (CSP) and thrombospondin-related adhesive protein (TRAP)
to highly-sulfated liver heparin sulfate proteoglyans (HSPG) triggers release of microneme
cysteine proteases which cleave CSP to reveal a thrombospondin-like type I repeat (TSR)-cell
9
binding motif, priming the sporozoite for hepatocyte invasion (Coppi et al., 2011; Frevert et al.,
2006; Pradel et al., 2004). Prior to hepatocyte invasion, the sporozoite again uses traversal
activity to migrate through Küpffer cells, liver sinusoidal endothelium, and hepatocytes.
Traversal is characterized by internalization of sporozoites without formation of a
parasitophorous vacuole, leading to open holes in the cell membrane and release of cytoplasm—
this process can be observed by release of stained cytoplasm (Mota et al., 2001). Reports have
described traversal as obligatory to invasion, and also as merely a tool for escape from Küpffer
engulfment and traversal of the liver endothelium, if needed (Baer et al., 2007; Frevert et al.,
2006; Menard et al., 1997; Mota et al., 2002; Tavares et al., 2013). Regardless, the traversal
mechanism requires function of SPECT’s, TRAP and CelTOS. Sporozoites can survive
engulfment by Küpffer cells as CSP has been shown to inhibit Küpffer reactive oxygen species
(ROS) formation, prior to completing traversal escape of macrophages (Usynin et al., 2007).
Following liver targeting and either passive or active traversal to the hepatic side of the liver
sinusoid, invasion of a hepatocyte begins first by reversible, then nonreversible, attachment of a
sporozoite to the host membrane (Ejigiri and Sinnis, 2009). This attachment could be mediated
by apical membrane antigen (AMA)-1 and TRAP, as deletion of functional domains in the
proteins results in attachment but not invasion (Matuschewski et al., 2002; Silvie et al., 2004).
Hepatocyte invasion is driven by the parasite’s molecular motor and results in encasement in a
host-membrane derived parasitophorous vacuole (Keeley and Soldati, 2004).
Over 6-10 days liver forms grow into schizonts; multinucleated forms containing
thousands of merozoites (Figure 1.3b). Plasmodium vivax and P. ovale can also halt division
early in the liver stage and remain dormant as hypnozoites for weeks, months, or years. While
many studies have demonstrated possible confirmation of hypnozoites in in vivo and in vitro, an
10
accepted biological marker unique to hypnozoites has not been established, nor do we know
much about hypnozoite biology (Dembele et al., 2011; Mazier et al., 1984). The mechanism of
hypnozoite formation and reactivation has recently been suggested to occur through histone
modification and regulation (Dembele et al., 2014). The first P. vivax infection of life is thought
to lead to two populations of liver parasites, one completing liver development leading to
disease, and the other dormant until reactivation. However, continued re-infection with
Plasmodium can lead to reactivation of dormant forms from previous infections (Imwong et al.,
2012). Both hypnozoites and developing liver parasites are able to completely hijack hepatocyte
molecular machinery and mask Toll Like Receptor (TLR) immune responses. A yeast hybrid
study revealed a parasite protein upregulated in sporozoites (UIS)-3, which localized to the
parasitophorous vacuole membrane (PVM), is exposed to the out leaf of the membrane and binds
liver fatty acid binding protein (LFABP). This serves as evidence the parasite sequesters
macromolecules and nutrients (glucose) from the host cell to aid in replicating its genome and
membranes (Mikolajczak et al., 2007). Additional evidence for parasite interaction with the host
cell has been described by identification of exported protein (EXP)-2 in the PVM—this protein is
part of a well-described translocon found in the parasite’s blood stage used for shuttling of
parasite protein into the erythrocyte (Boddey et al., 2009; Vaughan et al., 2012). Transcriptional
profiling of infected hepatocytes reveals parasite-modified immune signaling, including
decreased peroxisome proliferator-activated receptors (PPAR) activity, tumor necrosis factor
(TNF) signaling, and Nitrous Oxide formation compared to uninfected hepatocytes; and infected
hepatocytes are apoptosis-inhibited by parasite signals (Albuquerque et al., 2009; Heussler and
Stanway, 2008; van de Sand et al., 2005). To achieve massive and rapid replication, merozoite
membrane formation is assisted by the parasite’s Type II fatty acid synthesis pathways, located
11
in the apicoplast (Vaughan et al., 2009). After complete merozoite formation, merozoite packets
called merosomes encased in host cell membrane are released to the bloodstream following
parasite-directed PVM breakdown (Vaughan et al., 2012).
The blood stages begin when a merozoite in circulation come into contact with an
erythrocyte, leading to a low-affinity attachment of the merozoite to the erythrocyte surface
mediated by merozoite surface proteins (MSP)-family proteins. Attachment is also mediated by
P. falciparum merozoite surface protein erythrocyte binding antigen (EBA)-175 to glycophorin
A; or P. vivax merozoite surface protein Duffy binding protein (DBP) to the Duffy blood group
on the surface of erythrocytes (Adams et al., 1992; Siddiqui et al., 2012; Tolia et al., 2005). The
merozoite rolls on the erythrocyte’s surface and reorients its apical end (containing secretory
organelles termed micronemes and rhoptries) toward the erythrocyte surface (Cowman and
Crabb, 2006). After reorientation, is has been shown by super-resolution microscopy that the
rhoptry necks secrete their contents, resulting in parasite protein rhoptry neck (RON)- 4 delivery
to the cytosolic side of erythrocyte membrane. RON4 serves as an anchor for AMA1, recently
secreted from micronemes. AMA1 has been shown to bind RON2, forming an irreversible tight
junction. The AMA1-RON2 complex transmits force generated by an actomyosin motor driving
the parasite into the erythrocyte. Concurrently, rhoptry bulbs secrete their contents, releasing
parasite proteins necessary for PVM formation (Bargieri et al., 2013; Lingelbach and Joiner,
1998). Following invasion the parasite’s dense granules release, beginning protein export to
erythrocyte and erythrocyte remodeling (Riglar et al., 2011).
Following erythrocyte invasion, the parasite endocytoses hemoglobin into a digestive
vacuole, degrades hemoglobin by families of the aspartic and cysteine proteases (plasmepsins
and falcipains), and scavenges the resulting amino acids (Moura et al., 2009). The byproduct of
12
hemoglobin digestion, hemozoin, remains in the digestive vacuole and is visible by light
microscopy. Intraerythrocytic parasites also sequester glucose and hypoxanthine from the host’s
circulation, necessary for glycolysis and purine synthesis for reproduction of the parasite’s
genome. Genome replication results in the formation of 8-30 daughter merozoites (Asahi et al.,
1996). Throughout the parasite’s 44-72 hour asexual life cycle, the parasite is able to transport
proteins into the erythrocyte cytosol or surface using a translocon termed PTEX (Riglar et al.,
2013). Such translocation results in shuttling of knob-associated proteins (i.e. KARP, PfEMP1)
to the exterior side of the erythrocytes membrane. Knobs function by binding endothelial cell
surface markers (i.e. ICAM-1, CD31, CD36, CSA) in the circulatory system to sequester infected
erythrocytes from splenic clearance—sequestration in the brain and placenta leads to
symptomatic cerebral and placental malaria (Tilley et al., 2011). Furthermore, knob proteins
provide the basis for blood stage antigenic variation. Approximately 60 var genes encode
aPfEMP1 protein variant; yet only one var is expressed at a given time. As PfEMP1 is extremely
immunogenic, antibody responses to PfEMP1 are ablated due to occasion switching of the
expressed var gene (Scherf et al., 2008). After 44 or 72 hours, blood stage parasites
synchronously lyse their host erythrocyte, partially due to the specific timing of release and
subsequent action of parasite proteins SUB1 and SERA5 (Balu et al., 2011). Synchronous
destruction of infected erythrocytes triggers a burst of TNF leading to the periodic fevers
associated with symptomatic malaria (Roberts, p150).
Immediately after blood infection (P. vivax) or after a stress-inducing parasitemia is
reached (P. falciparum), some blood stage parasites undergo gametocytogenesis. Regulation of
this decision most likely occurs in the asexual cycle preceding invasion of gametocyte-destined
merozoites as all merozoites from a schizont are committed to become either all male or female
13
gametocytes (Bruce et al., 1990; Silvestrini et al., 2000; Smith et al., 2000). Following ingestion
during a mosquito blood meal, both male and female gametocytes are released from their host
erythrocytes. Males undergo 3 rounds of DNA replication, segregating into 8 microgametes.
Microgametes are incredibly simple cells containing only a haploid nucleus and lack nearly all
organelles. Termed exflagellation, this process unfolds over 10 min and is triggered by a drop
from body to ambient temperatures and contact with xanthurenic acid in the mosquito midgut
(Billker et al., 1997). Female gametes contain the apicoplast and mitochondrion (Creasey et al.,
1993). After initial contact, gamete fusion occurs when the entire male moves into the
cytoplasm of the female and their two nuclei fuse to form a diploid genome (Liu et al., 2008).
Meiosis immediately ensues, resulting in oökinete formation (Sinden, 1985). Oökinetes traverse
the mosquito midgut wall, come into contact with the midgut basal lamina, and transform into an
oocyst. Oocyst undergo several rounds of mitotic division, forming packets of sporozoites which
are released into the mosquito hemolymph and are targeted to invade salivary glands in
preparation for transmission (Sinden, 2009).
Current Prophylaxis and Standard of Care. CDC recommendation for prophylactic
treatment of travelers visiting endemic areas varies based on the individual traveling and location
of travel. Malarone (atovaquone and proguanil), choloroquine, doxycycline, mefloquine, and
primaquine can be used for prophylactic treatment.
However, Malarone and doxycycline
should not be used by pregnant women, chloroquine and mefloquine may be ineffective due to
widespread resistance in certain parts of the world, and chloroquine, doxycycline, mefloquine,
and primaquine all harbor undesirable side-effects (CDC, 2012a).
The following diagnosis and care guidelines were obtained from the WHO Guidelines for
the Treatment of Malaria (2010). Diagnostic guidelines vary by region and practicality of
14
methods. In general, microscopic identification of Plasmodium parasite presence and species
and identification by rapid tests (i.e. colormetric antibody-reaction cards) are the two most
widely used methods. Both methods require resources and costs sometimes unavailable in
endemic areas, such as microscopes, trained users, and equipment (i.e. PCR thermocyclers).
Regardless, proper identification of species and even strain can greatly assist in monitoring
parasite populations and the spread of antimalarial resistance.
Uncomplicated P. falciparum malaria should be treated by combination therapies
utilizing different modes of action as, given the huge biomass of parasites within an infected
individual, the chance for selection of resistant parasites to monotherapy is moderate. This
phenomenon can be explained as there are approximately 1011 parasites in an infected individual,
and the estimated rate of resistance-inducing mutations is 10-9, thus approximately 100 parasites
will survive a monotherapy and spread as resistant strains. The chance of a double mutant is (109 2
) , meaning the chance of a double mutant surviving a combination therapy is 1 in 10 million
(Hastings, 2011). Artemisinin-based combination therapies (ACT) are the current front line
treatment, with the artemisinin component given for at least 3 days (Greenwood et al., 2008).
Artemisinin derivatives include dihydroartemsinin, arteether, artesunate, and artemether.
Combination therapies include artemether plus lumefantrine, artesunate plus amodiaquine,
artesunate plus mefloquine, artesunate plus sulfadoxine-pyrimethamine, and dihydrortemisinin
plus piperaquine. Failed first line treatment may require second line treatment consisting of
artesunate plus tetracycline or doxycycline or clindamycin. Pregnant women should be treated
with quinine plus clindamycin in the first trimester, and ACTs in the second trimester, third
trimester, and during post-birth lactation. Young children should also be treated with ACTs.
Uncomplicated P. vivax, P. ovale and P. malariae malaria can still be effectively treated with
15
chloroquine, as resistance is less frequent in these species. In cases of resistant P. vivax, ACTs
and primaquine are suggested unless the individual is severely glucose-6-phosphate
dehydrogenase (G6PD) deficient. Primaquine is the only approved drug with activity against the
dormant forms of P. vivax and P. ovale.
Control and Eradication Efforts. Malaria control is dependent on monitoring,
diagnosis, vector control, and drug treatment methods. Such methods have drastically improved
malaria mortality rates over the past decade, but the rise of artemisinin resistance and
identification of an artemisinin resistance marker highlights the need for further drug
development to continue malaria control (Ariey et al., 2014; Noedl, 2005; White, 2014).
Malaria elimination from less affluent, endemic areas has only been achieved on a remote
Pacific Island, with a small human population, using stringent mass drug administration followed
by thorough monitoring of fever. P. falciparum was considered eliminated from the island after
five weeks, but P. vivax elimination required five years (Kaneko et al., 2000; von Seidlein et al.,
2003). Our current malaria toolbox is inadequate for similar elimination strategies in mainland
areas of Central and South America, Africa, and Southeast Asia as treatment areas are not likely
to be extended across borders and terrain where reservoirs of parasites would return to the treated
population. Elimination and, eventually eradication, is not practical without economic
development or licensing of an extremely efficacious vaccine, which currently does not exist
(WHO, 2010).
Vector Control. Vector control formed the basis of malaria elimination in the United
States in the 1940’s and 50’s—homes were sprayed with DDT and improved drainage drastically
reduced mosquito populations in endemic areas. In areas where people sleep accessible to
nocturnal mosquito feeding, distribution, maintenance, and use of long lasting insecticide-treated
16
bed nets (LLBN) and indoor residual spraying of insecticide (IRS) are critical for preventing
transmission. Aside from bite prevention, mosquito population control is extremely effective as
the parasite spends weeks developing in mosquitoes, providing ample time for targeting.
However, the efficacy of vector control is much greater in low transmission areas as, in high
transmission areas, an individual may receive hundreds of infective bites per year and vector
control may only reduce this bite rate 10-fold. In both settings, the rise of pyrethroid resistance
is currently hampering control efforts. Novel, long lasting, mosquito-specific, and inexpensive
insecticides are needed to continue vector control. Furthermore, alternative vector control
technologies, such as releasing genetically engineered sterile males into wild populations, could
prove essential if such methods prove environmentally safe and effective (review in malERA,
2011b).
Drugs. Many classes of drugs have been developed for treating malaria, some in use over
hundreds of years. Herbal use of artemisinin from Artemisia annua in China dates back to 340
A.D. and herbal use of quinine from cinchona bark in South America dates back to 1633 A.D.
(Ball, 2008; Wallace, 1996) . Other drugs include antibiotics (azithromycin, doxycycline)
targeting the apicoplast, atovaquone and tafenoquine targeting the mitochondrion, antifolates
targeting the cytosol, and falcipain inhibitors targeting the digestive vacuole (reviewed in
Greenwood et al., 2008). Due to widespread resistance to many of these classes, ACTs are the
‘last line’ of defense for complicated, resistant cases of malaria. For a complete cure of P. vivax
and P. ovale dormant forms, primaquine is the only licensed drug available. Primaquine use
must be closely monitored due to a high prevalence of G6PD alleles in populations located in
malaria-endemic areas. Several mutations leading to G6PD deficiency exist, exhibiting a range
of severity. Toxicity is thought to occur as G6PD individuals are less able to provide energy
17
through the pentose phosphate shunt for recycling of NADP to NADPH, needed for recycling of
glutathione disulfide to glutathione, which is needed to control oxidation within erythrocytes.
The activity of primaquine involves formation of free radicals, creating excessive oxidation in
erythrocytes of malaria patients with G6PD, causing hemolytic anemia (Bolchoz et al., 2001).
Success of a novel malarial drug is dependent on several characteristics. Desirable
characteristics include effectiveness as a prophylactic, absence of drastic side effects, suitability
for vulnerable groups (i.e. children, pregnant women, G6PD-deficient and similar individuals),
effectiveness against P. vivax and P. ovale dormant forms, a single dose complete cure, a short
half-life supplemented with sustained release, rapid elimination from the body, a long shelf life,
minimal storage requirements, and suitability for combination therapy (malERA, 2011b).
Recently, a Phase IIb clinical trial involving 329 patients from Brazil, Peru, India, and Thailand
demonstrated up to 92% anti-relapse efficacy of tafenoquine (an analog of primaquine) in
combination with chloroquine for treatment of P. vivax (Llanos-Cuentas et al., 2014). While
studies such as these are promising, more antimalarial compounds are needed to continue malaria
control.
Vaccines. An effective, licensed antimalarial vaccine is essential for continued malarial
control and hope for future worldwide malaria eradication. Several bottlenecks in the parasite’s
life cycle have been studied for potential vaccine targets. These bottlenecks are defined as
critical points where few parasites persist--or invasion steps featuring intricate host-parasite
interaction--where an effective vaccine could prevent life cycle completion and subsequent
disease or transmission. Studies of the sporozoite’s journey from a bite site to the liver have
revealed CSP as a prime vaccine target. Phase III trials are currently underway for RTS,S, a CSP
subunit vaccine against P. falciparum liver invasion. Phase II trials demonstrated 60% efficacy
18
in protecting children from malaria (Aide et al., 2010). Other sporozoite-based vaccines utilize
either radiation or genetically attenuated sporozoites to mediate immune response prior to
invasion and incomplete development in the liver (Epstein et al., 2011; van Dijk et al., 2005;
Wang et al., 2009). While early studies in mice are promising, the safety and efficacy of these
vaccine candidates is still unproven and much optimization and continued testing is needed.
That said, these works have revealed a CD8+ T cell response and IL2, IFNγ, IL12, TNFα, IL1
and IL6 production are necessary for innate and acquired immunity to sporozoites and liver
forms (Pombo et al., 2002). Blood stage targets include AMA1, MSP1, and EBA-175—all
designed to inhibit merozoite invasion. Other blood stage vaccine targets are to knob proteins or
PfEMP1 subunits, effective by inhibiting sequestration and allowing clearance of infected
erythrocytes by the spleen. Vaccination strategies against gametocytes and oökinetes have also
been suggested and tested. The ethical limitations of such vaccines as been questioned as they
would effectively prevent transmission, but not necessarily the illness of the already infected
individual.
Vaccine development is hampered by our incomplete understanding of protection
mechanisms in natural transmission settings. Most deaths from malaria occur in children under 5
years of age, as the primary infection is most virulent in immune-naive individuals. The adult
immune system is capable of coping with, but not preventing or clearing, Plasmodium infections.
The inability to prevent sporozoite infection may be due to low parasite numbers exposed to
extracellular conditions for a short period of time, followed by immune masking of an
intracellular developing liver parasite. In the blood stage, antigenic variation accounts for
continued infection as immune response are raised to short-lived epitopes on PfEMP1.
Furthermore, as one desirable characteristic of an effective vaccine is life-long protection, such
19
protection does not appear to occur as immunity raised to malaria in high-transmission setting is
lost upon removal from transmission (reviewed in Langhorne et al., 2008). Combined, these
factors demonstrate the difficulty of antimalarial vaccine research and the need for continued
efforts to prevent transmission.
Liver Stage Drug and Vaccine Development. Taking into account that the Plasmodium
liver stage is essential for subsequent steps in the parasite’s life cycle and disease progression,
that P. vivax and P. ovale hypnozoites are drug-resistant targets, and that the liver stages are
prime vaccine targets, murine and human Plasmodium liver models have been developed to learn
more about liver stage biology and to test the efficacy of novel therapeutics on liver forms.
These liver models serve as essential pre-clinical studies providing scientific support for novel
therapeutic concepts. After pre-clinical studies, millions of dollars and decades of time are
required to drive novel therapeutics through Good Manufacturing Practices and efficacy trials,
allowing for licensing and use. More valid preclinical studies, including in vitro and in vivo
models, show promise to better predict both positive and negative efficacy in human trials, thus
bringing more efficiency to drug and vaccine development (malERA, 2011c).
Due to rate-limiting difficulty of P. falciparum sporozoite production and as-yet
undeveloped in vitro sourcing of P. vivax sporozoite, P. berghei and P. yoelii rodent models have
been developed as surrogates for human in vivo and in vitro liver models. After decades of
development, mouse liver models now allow for rapid quantification of liver stage development
in both HepG2-populated microtiter plates and live animals using fluorescent proteins and
luciferase expression (de Koning-Ward et al., 1999; Franke-Fayard et al., 2004; Lacrue et al.,
2013). Such studies have provided for development of 384 and 1536-well high-throughput
compound screening, revealing hundreds of new ‘hits’ for anti-Plasmodium activity.
20
Furthermore, much of what we now know of Plasmodium liver stage biology was discovered in
rodent models. Our lack of a medium or high-throughput human Plasmodium liver model
requires that the next step of preclinical studies need be performed with human parasites in a
chimpanzee model. Such studies are extremely limited in throughput and require time and
resources, yet the hits from mouse models rarely correlate well to the chimpanzee model. This
may be due to a species-specific drug metabolism between mice and chimpanzees, or speciesspecific drug susceptibility between P. berghei and human parasites. Furthermore, rodent
models are incapable of modeling hypnozoite formation as rodent hypnozoites have not been
described.
To bridge the gap between mouse and chimpanzee preclinical studies, several studies
have been performed to stabilize human hepatocytes for long-term in vitro culture and infection
with human Plasmodium sporozoites. Plasmodium falciparum and P. vivax sporozoites show a
remarkable preference for primary human hepatocytes over hepatic cell lines; however primary
hepatocyte culture methods require complex features to ensure hepatocyte stability post-plating.
Mazier et al. (1985) successfully completed the P. falciparum and P. vivax liver stages in vitro
with a primary hepatocyte-stromal cell co-culture. This model was improved upon to allow for
extremely long-term culture by addition of a Matrigel® overlay—having the effect of stabilizing
hepatocytes for extended (21-35 d) long-term culture necessary for hypnozoite reactivation
(Dembele et al., 2014). Co-culture methods have also been improved upon by patterning
primary human hepatocytes in islands surrounded by fibroblasts in microtiter wells, forming a
multiplexed liver platform (March et al., 2013). To avoid difficulties associated with procuring
and maintaining primary human hepatocytes, more phenotypic (that is, less de-differentiated)
hepatocyte cell lines have been developed and shown to support liver stage invasion and
21
development (Karnasuta et al., 1995; Sattabongkot et al., 2006). The liver stages of P.
falciparum have also been completed in human hepatocytes engrafted in a living mouse liver,
generating perhaps the most robust model to date (Vaughan et al., 2012).
Need for a More Efficient in vitro Plasmodium Liver Model. The aforementioned
advancements of human Plasmodium liver models have undoubtedly advanced live stage
research, but they do not solve the problem of bridging the gap between mouse and chimpanzee
preclinical trials. Most of their shortcomings arise from a lack of efficiency. As mentioned, P.
falciparum and P. vivax sporozoites are extremely limited and expensive to obtain, and primary
hepatocyte lots are equally specialized, limited, and expensive. None of the models mentioned
above requires less than 10,000 hepatocytes per experimental replicate. Indeed, to achieve a
moderate (~0.1%) liver stage infection, these models require millions or tens of millions of
sporozoites. Such requirements are rate-limiting to the point of being prohibitive for drug and
vaccine screening. Thus, an efficient and robust human Plasmodium liver model should feature
primary human hepatocytes in long-term culture, complete development of P. falciparum and P.
vivax liver forms, P. vivax hypnozoite formation, in vivo-like drug metabolism, elevated
infection rates for statistical analysis of efficacy, and minimal demand of limited resources.
Such a model has not been described at the time of this writing, and development of such a
model is the focus of this research (reviewed in malERA, 2011a).
Introduction to in vitro Liver Models
Role of Liver Models in Public Health. The liver is primarily responsible for
homeostasis and maintains over 500 functions including glycogenesis, glycogenolysis,
gluconeogenesis, triglyceride oxidation, lipoprotein synthesis, cholesterol and phospholipid
synthesis, amino acid transamination, ammonia removal through urea synthesis, synthesis of
22
non-essential amino acids, synthesis of plasma proteins, storage and metabolism of copper and
iron, and clotting factor production. The liver is also chiefly responsible for metabolism of
exogenous compounds (drugs), including activation of, deactivation of, and excretion of
compounds from the body. Over 100 forms of liver disease have been identified with broad
implications in public health, including parasitic infection, viral hepatitis, obesity, alcoholism,
genetic disorders, autoimmune disorders, and cancer. Modeling the human liver, its response to
drugs, and possible disease states is essential to our understanding of the mechanisms of such a
complex and essential organ. To this end, many liver systems have been generated ranging from
cell-free microsomes, hepatic cell lines and primary hepatocytes in in vitro culture, threedimensional multi-cell type liver scaffolds, chimeric mice with humanized livers, bioartificial
livers (BAL), and a wide array of animal models (i.e. zebrafish, mice, dogs, and chimpanzees).
Such systems are tailored to the experimental questions to be answered, featuring variance in
throughput, species-specific responses, and complexity. Often, these factors are at odds with
each other (Arias, p1-15; reviewed in Godoy et al., 2013).
Animal-Free ADMET. In pharmacology the analyses and profile of a compound’s
absorption, distribution, metabolism, excretion, and toxicology is referred to as ADME,
ADMET, or ADMETox. An understanding of a compound’s ADME via preclinical
pharmacodynamic and pharmacokinetic studies is the foundation for an Investigational New
Drug Application (INDA). Pharmacodynamic studies aim to establish a what concentrationdependent effects a compound possesses, while pharmacokinetic studies aim to understand all
effect s of a drug, and the body’s response, from ingestion all the way through excretion. These
studies establish the efficacy and safety of compounds such that quality candidates are selected
for further testing in human trials. An accurate assessment of drug efficacy and safety is
23
essential to the drug production pipeline as up to $1 billion and over 10 years of research are
need to take a compound from discovery to licensing—resources lost when drugs fail (Singh,
2006). Since 1950, over 600 pharmaceutical products have been withdrawn or restricted due to
adverse drug reactions (ADR) and drug induced liver injury (DILI), with obvious negative
implications for the user’s health and both the user’s and drug company’s financial interests.
These failures highlight the importance of accurate preclinical and clinical testing (Materne et
al., 2013).
Animal testing is important to ADME studies as rat hepatocyte culture and rat drug
dosing are currently widely used for pharmacokinetic studies. Such studies are less practical
with human tissue or cells as human hepatocytes are notoriously difficult to maintain with full
functionality in short or long-term culture (Guguen-Guillouzo et al., 1983). Furthermore,
hepatocytes in the context of an organ section or a full organ better predict transport
phenomenon characteristic of liver function (LeCluyse, 2001). However, each species (and
individuals within species) possesses genotypically and phenotypically different repertoire of
exogenous-compound metabolizing enzymes and transport proteins. Such differences account
for the inability of rat hepatocytes to accurately predict methotrexate sensitivity in human cells
(Walker et al., 2000). Therefore preliminary, but high throughput and reproducible, studies to
assess the human metabolism of compounds use microsomes containing liver enzymes in cellfree reactions (Pybus et al., 2012). Again, these studies are informative but still leave a gap in
preclinical studies from animal testing to human trials, as cell-free systems do not accurately
predict the intra- and extracellular transport of compounds and their metabolites. To fill this gap,
several technologies (i.e. HepatopacTM from Hepragen and LiverChipTM from Zyoxel) have been
made commercially available featuring complex human hepatocyte culture systems aimed to
24
stabilize primary human hepatocytes in long-term, in vivo-like culture for drug toxicology
assessment (Chan et al., 2013; Li, 2007).
Cytochrome P450 Enzymes. Cytochrome P450 systems have been extensively reviewed
in Iyanagi (2007). To summarize, the most important proteins associated with ADMETox is the
cytochrome P450 enzyme family. This family fulfills other roles of the liver as well, including
elimination of cholesterol, regulation of blood homeostasis, and steroid metabolism. The ‘450’
name of P450 comes from a procedure whereby hepatocytes are ground up, centrifuged, and
separated. As P450 enzymes are located in mitochondria and endoplasmic reticulum (ER)
membranes, the supernatant containing these membranes absorbs light 450 nm. This absorption
is due to the enzyme’s basic structure: at the center of molecular lies a heme group connected to
the tertiary protein structure by 6 nitrogens. Twelve different mammalian P450 structures are
known: CYP1A2, 2A6, 2B4, 2C5, 2C8, 2C9, 2D6, 2R1, 3A4, 7A1, 8A1, and 46A1. The two
most active in drug metabolism are CYP2D6 and CYP3A4, as 3A4 metabolizes over 120 drugs
(other estimates suggest over half of all approved pharmaceuticals are metabolized by 3A4).
CYP 2D6, 3A4, and 1A1 have been implicated in the metabolism of many antimalarial
compounds (Bapiro et al., 2002; Pybus et al., 2012). The human P450 system includes 57 genes
from 18 families and 44 subfamilies. In the ER, P450 enzymes predominantly reside on the
cytomplasmic side of the membrane and are fed electrons by NADPH and NADPH cytochrome
P450 reductase. Alternatively, P450 enzymes found in the mitochondrial inner membrane are
fed electrons by NADPH and NADH, with ferredoxin controlling the final transfer of electrons
to the enzymes Drug metabolism occurs in three phases and P450 enzymes often perform phase
I. In phase I, a chemical modification results in addition of a functional group, often a hydroxyl,
to make the compound more water-soluble. For some compounds, this modification actually
25
generates the active species of the compound. In phase II, the added functional group is
conjugated to a glutathione, sulfate, glycine, or glucurinic acid by a large family of transferasses;
often resulting in deactivation of the compound. This step also tends to generate more polar
entities which increase compound transport across membranes. Finally, in phase III, the
deactivated compound is pumped into bile canaliculi by multidrug resistance proteins (MRP)—
an ATP-binding cassette (ABC) family of transporters found on the apical surfaces of
hepatocytes.
P450 enzymes are governed by a variety of regulatory mechanisms responding to
introduction of compounds into the body. Two example pathways involve cyclic AMP (cAMP)
and peroxisome proliferator activated receptor (PPAR) activity. In PPAR signaling, a drug binds
PPAR in the hepatocyte membrane, causing PPAR to translocate to the nucleus where it
heterodimerizes with an X receptor (RXR) on a gene regulatory element, resulting in P450
transcription. Drug-drug interactions occur when multiple drugs taken at the same time result in
improper P450 expression or repression. For example, itraconazole is antifungal that inhibits
CYP3A4. As CYP3A4 metabolizes hundreds of other drugs, the pharmacokinetics of a drug
taken along with itraconazole would be altered, possibly leading to liver damage or death
(Hauben and Hung, 2013).
Polymorphism in the P450 enzyme family can account for a range of phenotypes
(sensitivities) to drugs and are also a source of adverse drug reactions. Idiosyncratic liver injury
can result when toxicity of a drug is not found in trials as polymorphisms appear as rarely as 1 in
5000 to 1 in 100,000 persons. For example, a specific CYP2C19 polymorphism resulting in
sensitivity to mephenytoin is found in only 3% of the human population, but carries a prevalence
of 20% in Asian populations. Adverse drug reactions can also occur due to an overreliance on
26
animal testing—rodents have many more P450 enzymes than humans resulting in differential
drug metabolism. For example, humans have only one 2D and J2 enzymes while mice have 9
and 8, respectively; and humans have two 2C enzymes while mice have 15 (reviewed in Nelson,
1999; Nelson et al., 2004).
Infectious Liver Disease Culture and Research. Although a much more narrow field
than drug metabolism, Hepatitis A, B, and C viruses all infect the liver and Plasmodium parasites
spend an obligatory round of replication within hepatocytes after initial infection by sporozoites
and before blood-borne disease ensues. Hepatocyte cell lines, advanced primary hepatocyte
culture systems, and humanized mice featuring chimeric livers have recently been developed to
model these disease; as is the focus of this research (Dembele et al., 2014; Hantz et al., 2009;
March et al., 2013; Ploss et al., 2010; Vaughan et al., 2012). These models have found the use of
primary human hepatocytes in stable, long-term culture as essential for proper Plasmodium
infection and development.
Architecture of the Liver. The liver is the body’s largest organ and is closely
associated, in proximity and function, with the small intestine as it receives and processes
nutrient-rich blood from digestion. The liver is fed by both arterial and portal blood. Most
portal blood arrives from the gastrointestinal system, but also from the spleen, pancreas, and
gallbladder. Portal blood is characterized by very low pressure (10-12 mmHg), low oxygen
content, and an absence of pulsing from heart pumping. Arterial blood comes from the hepatic
artery which branches from the descending aorta. Arterial blood is characterized by high pressure
(120-180 mmHg), high oxygen content and pulsation. These two blood supplies extensively
branch and narrow before they meet at the level of the liver sinusoid; they mix in sinusoids, flow
into central veins are delivered back into circulation via the hepatic vein. The major blood
27
vessels, lymph vessels and bile ducts serving the liver all arrive at the liver at the hilus, located
almost in the center of the liver (Arias, p1-15).
The functional unit of the liver can be described as a lobule or acinus; geometric
formations of hepatocytes and supporting cell types repeated throughout the liver at the supercellular level (Figure 1.3). Lobules are organized into hexagons whereby six portal canals (at the
hexagonal points) deliver blood to the sinusoids, and drain into one central vein. Portal canals
contain three different vessels: hepatic arterioles and portal venules delivering blood to the
sinusoids, and bile ducts removing bile from bile canaliculi. From the portal canals to the central
vein, sheets of hepatocytes only one to two cells thick are separated by endothelial cell-lined
sinusoids. Tight junctions between hepatocytes in hepatic sheets form bile canaliculi on their
apical surface; thus hepatocytes in a single cell sheet harbor two sinusoid-facing basolateral
domains on either side of a bile canaliculi-facing apical belt (Dunn et al., 1991). These hexagon
structures geometrically compact into an array; thus a liver acinus is described as portal canals
centrally serving three central veins. In this view, gradients like glucose metabolism and
oxygenation form as blood leaves the portal canals and arrives in the central veins. These
gradients form three zones (periportal, pericentral, and midzonal) characterized by functionally
unique hepatocytes. Whole blood plasma, as well as proteins and small molecules in blood
plasma, are passively or actively transported across sinusoidal endothelial cells. Passive
transport can occur in either gaps in the endothelial lining or through sieve plates; bundles of tiny
(100-150 nm) fenestrae passing directly through sinusoidal endothelium. Plasma components
pass through the space of Disse before being taken up and processed by hepatocytes (Braet and
Wisse, 2002).
28
Zone-Specific Liver Physiology. The periportal zone (where blood enters the sinusoid) is
defined by high oxygen content, glycogen production from lactate, glucose release, glutathione
peroxidation, albumin formation, and bile formation. Ureagenesis occurs greatest at the
periportal zone and tapers off toward the perivenous zone. The perivenous zone is defined by
low oxygen content, glycogen production from glucose, glcolysis, bile production, P450
function, alpha 1-antitrypsin production, heme synthesis and glutamine production (reviewed in
Godoy et al., 2013; Materne et al., 2013). Hepatocyte contact with extracellular matrix (ECM) is
one factor driving differentiation towards zone-specific function. Throughout the sinusoid,
fibronectin is the main matrix component of the basal lamina between hepatocytes and
endothelial cells while collagen I and III support the sinusoidal areas in the periportal and
perivenous zones (Rhodes, 2007).
Cell Types of the Liver. The main cell types of the liver include parenchymal hepatocytes
and nonparenchymal stellate cells, sinusoidal endothelial cells, Küpffer cells (macrophages), and
cholangiocytes (Arias, p9-10). An adult liver contains about 1011 hepatocytes, each ~25 µm
across, making up 60% of all cells of the liver and 80% of the liver volume. One third of the
cell’s surface faces a liver sinusoid and this surface area is expanded by microvilli which can
stretch out through the space of Disse, past endothelial cells, and into the liver sinusoid. Half of
the hepatocyte surface is connected with five to six other hepatocytes through tight junctions,
intermediate junctions, and desmosomes. As previously mentioned, hepatocyte vesicular
transport and xenobiotic metabolism is of particular importance to this work. The hepatocyte
cell membrane is composed of basolateral surfaces facing the sinusoid and apical domains
facing other hepatocytes. The basolateral domains contain organic anion-transporting
polypeptide (OAT)-family transporters such as OAT 1B1, 1B3, 2B1, 2, and 7 while the apical
29
domains contain multidrug resistance (MDR)-family transporters such as MDR 1, 2, and 3
(reviewed in Godoy et al., 2013).
Liver endothelial cells are a unique type of endothelial cell, differing from other
endothelial cells in size, extracellular matrix, function, and ultrastructure (Wisse, 1972). An
adult liver contains approximately 3x1010 endothelial cells forming the liver sinusoids.
Sinusoids lack a true basal lamina but are surrounded by a rich mixture of extracellular matrix
including collagens I and III through VI, laminin, fibronectin, and heparin sulfate, dermatan
sulfate, and chondroitin sulfate proteoglycans. Endothelial cells possess a unique structuretermed sieve plates-made up of tiny (<200 nm) fenestrations allowing plasma to pass directly
through the cells (Sellaro et al., 2007). Sinusoidal endothelial cells carry immune function as
they constitutively express ICAM-1 and LFA-3 on their surface, recognize pathogen-associated
molecular patterns (PAMPs), recruit leukocytes, and scavenge biological waste materials from
the blood. Endothelial cells are also able to recognize IgG via Fcγ, triggering endocytosis of the
IgG molecular and antigen. Sinusoidal endothelial cells lack expression of typical endothelial
markers E, L, and P-selectin, VCAM-1, CD31 (PECAM), and CD34—thought to help maintain
an immune-tolerant state in the liver necessary as it is the collection and clearance point for
many foreign entities (Daneker et al., 1998).
Liver stellate cells are important for fat-soluble vitamin A metabolism and storage,
production of liver extracellular matrix, and response to liver injury. They reside between
hepatocytes and endothelial cells in the space of Disse. Stellates respond to liver injury as
stellate toll-like receptor (TLR) 9 detects dead hepatocyte DNA, triggering matrix deposition for
repopulation by hepatocytes (Friedman, 2008). Stellate and Küpffer cells can activate oneanother in response to liver injury or infection. Küpffer cells are the resident macrophages of the
30
liver, loosely attached to or sometimes embedded in liver endothelium, and function by
phagocytosing bacteria, old erythrocytes, and dead hepatocytes from sinusoidal flow. Activation
of Küpffer cells results in secretion of TNFα and IL1b and subsequent hepatocyte apoptosis, but
Küpffers also balance inflammation by secretion of anti-inflammatory IL10, IL18, and
transforming growth factor (TGF) β (Goldin et al., 1996). Küpffer cells can be identified as they
express high levels of F4/80, C3 and Fc receptors on their surface (Holt et al., 2008).
Hepatic Cell Resources for Research. At least three different sources of hepatic cells are
currently available for liver research. Primary cells are obtained from a liver biopsy or whole
liver perfusion with detergents to isolate free cells. Purification methods, such as column-based
affinity chromatography, can be employed to purify specific cell types from whole liver
(Guguen-Guillouzo et al., 1982). Hepatic cell lines are either isolated from liver tumors, cloned,
and characterized or generated through immortalization by serial passage or introduction of
SV40 virus antigens to induce unlimited propagation (Kim et al., 1998). Induced Embryonic
Stem Cells (iES), that is, stem cells derived from chemical treatment toward pluripotency, can be
propagated and then driven to differentiate down the hepatocyte lineage (Yu et al., 2012).
Each source of hepatic cell types, but especially hepatocytes, carries with it advantageous
and disadvantages. Primary cells often are the most differentiated and express hepatic
phenotypes at nearly in vivo levels, but this expression is short lived as very few (if any) culture
systems are capable of replicating the microenvironment of the liver at the level needed to
maintain biological cues for long-term culture. Human primary cells are often obtained from
commercial sources, and each company may use different protocols for isolation, with each
having positive and negative effects on suitability for culture. Furthermore, primary cells are
often extremely expensive, lot specific, and limited as the distribution of phenotypes in human
31
populations is not captured from a single isolation (Godoy et al., 2013; Strom et al., 1982). Cell
lines present solutions to the expense, limited availability, and lot specificity of primary cells but,
due to their nature as cancer-like cells, are prone to de-differentiation and loss of phenotypes.
Additionally, cell line proliferation often negates their use in long-term culture systems as cell
numbers can quickly outgrow the space provided. iES cells show promise as a renewable source
of primary-like cells but final differentiation into mature, parenchymal cell types is just outside
the reach of modern culture systems as research has shown mimicking the microenvironment of
organs is essential for in vivo-like function (Okura et al., 2010; Si-Tayeb et al., 2010).
Primary hepatocytes are available from Lonza, Sciencell, Bioreclamation IVT, Yecuris,
and others; hepatocyte cell lines HepG2, Huh7, HC-04 and HepaRG (and others) are publically
available, and iES cells are often derived from collaborations (Antherieu et al., 2010; Horwich et
al., 1990; Morris et al., 1982; Sattabongkot et al., 2006). These hepatocyte sources can be
characterized by many of their hundreds of functions, but often albumin production, urea
synthesis, glycogen synthesis, and P450 function are assessed to valid functions. Several studies
have described the various levels of function associated with these hepatocyte sources, but one
study in particular performed a direct comparison of HepG2, HepaRG, and primary human
hepatocytes. The authors concluded HepG2 were unable to properly respond to drug treatments
but HepaRG and primary hepatocyte P450 expression was strongly induced, indicating primary
phenotypes necessary for drug toxicity screening (Gerets et al., 2012; Wilkening et al., 2003).
Humanized mice present an attractive alternative to human hepatocyte sourcing issues as
human hepatocytes engrafted into a mouse liver matrix undergo an injury-response type
expansion, useful for in vivo studies or, after harvest, in vitro applications. At least one
humanized mouse model, FRG® KO, has been commercialized to produce and sell liver chimeric
32
mice and serially passages human hepatocytes (Strom et al., 2010). To generate the mice,
already immunocompromised mice were further genetically modified to be fumarylacetoacetate
hydrolase (FAH)-deficient. Removal of a compound from drinking water (NTBC) results in
buildup of FAH in hepatocytes and subsequent death. Human hepatocytes are injected into the
mouse, where they engraft into the space left behind by mouse hepatocytes, developing into a
chimeric organ over several months. While validation is needed, hepatocytes derived from
chimeric mice may significantly impact liver research (Maher et al., 2014).
As mentioned, iES hepatocytes may also advance liver research if full differentiation can
be achieved. Final differentiation into hepatocytes would require loss of fetal hepatocyte
markers alpha-fetoprotein and glutathione S-transferase P1, and primary-like function of P450
enzymes, albumin production, and alpha 1-antitrypsin synthesis (Nava et al., 2005). Only
recently have methods been developed to maintain iES cells without the support of
nonparenchymal feeder cells, which could complicate application to some liver studies. But, the
potential application for iES cells are incredible, as formation of a functional liver bud from iES
cells has recently been described (Takebe et al., 2013).
Compared to hepatocytes, relatively few sources exist for non-parenchymal cell types.
At least one company (Sciencell) has made fetal human liver sinusoid endothelial cells (HLSEC)
available and a tumor-derived liver endothelial cell type (HAEND) has been described (Hoover
et al., 1993). A history of research with HLSEC has revealed these cells rapidly de-differential
after removal from the body. The cells undergo a process termed endothelial-to-mesenchymal
transition resulting in loss of endothelial morphology, sieve plates, and functions (Arciniegas et
al., 2007; Elvevold et al., 2008). Similarly HAEND cells completely lack in vivo-like
morphology and do not form cobblestoned multicellular sheets. Primary human Küpffer cells
33
have recently been made commercially available (Life Technology), but suffer from the same
lot-specific difficulties of primary cells and often uncontrollable damage culture systems through
inflammation (Sunman et al., 2004). Human stellate cells have been immortalized by the SV40
method and are publically available (Xu et al., 2005).
Considerations for Designing in vitro Liver Models. Advanced, complex liver models
are designed to integrate in vivo-like cues into in vitro culture to recapitulate the liver
microenvironment, with the goal of maintaining hepatocyte function for liver assays. These cues
include, but are not limited to, phenotypic cell sources, cell co-culture, complex extracellular
matrix, advanced media formulations, media perfusion (and establishment of small molecule
gradients), and culture system architecture mimicking that of the liver sinusoid. In selection of
hepatic cells for advanced systems, it is important to account for the advantages and
disadvantages offered by primary cells, cell lines, and stem cells. Studies requiring predictive
metabolism, such as preclinical ADME studies, may demand primary hepatocytes as cell lines
and stem cells are not fully differentiated. Other studies requiring consistent and reproducible
phenotypes (and studies on tight budgets) may be able to take advantage of cell lines (Donato et
al., 2008). The longevity of the experiment must also be considered: both cell lines and primary
cells become increasingly more difficult to maintain in longer term research. The decision to
include multiple cell types is also critical. Co-cultures are well documented methods of
maintaining both cell line and primary phenotypes. For example, primary hepatocytes can be
maintained long term in a fibroblast co-culture, HepG2 function has been partially recovered in
an endothelial co-culture, and a stellate-Küpffer-hepatocyte triculture forms a mimetic in vitro
sinusoid (Khetani and Bhatia, 2008; Ohno et al., 2008; Ries et al., 2000; Thomas et al., 2005).
However, finding a universal media for co-culture, controlling the localization and movement of
34
co-culture component cells, preventing overgrowth of cell lines, monitoring cell activation states,
and the cost associated with using multiple primary cell types in experimentation cannot be
overlooked when planning co-culture models. For example, HLSEC may be ruled out of
advanced liver culture systems due to notorious instability, but HMEC and HUVEC endothelial
cell types are completely non-mimetic replacements for liver sinusoidal endothelial cells
(Daneker et al., 1998).
Selection of the type and number of extracellular matrix components (ECM) is also
critical in an advanced liver model design. Hepatocyte sensitivity to ECM has been
demonstrated by quantifying hepatocyte albumin production on 32 different combinations of
major liver matrix proteins, including collagen I, collagen III, collagen IV, laminin, and
fibronectin, revealing significant upregulation of phenotypes on collagen IV and fibronectincoated surface (Flaim et al., 2005). Extracellular matrix can also be used to maintain hepatocyte
polarization and morphology by overlaying gelled matrix on a hepatocyte monolayer
(‘sandwiching’) or completely embodying hepatocytes in a free-floating hydrogel spheroid.
(LeCluyse, 2001; Liu Tsang et al., 2007). Three-dimensional organ-like culture of hepatocytes
can also be achieved by seeding hepatocytes into decellularized animal liver matrix or artificial
liver scaffolds (Lin et al., 2004; Provin et al., 2009). Complex ECM mixes and threedimensional ECM culture systems can drastically alter hepatocyte polarity to maintain function,
but mixing and locating ECM components, or performing repeated functional studies with
engrafted liver scaffolds, can be irreproducible and cumbersome. An alternative strategy is to
supply a culture system with the cell types necessary for de novo ECM deposition, namely a
stellate cell co-culture for liver systems (Herrmann et al., 2007).
35
Complex medias have also proven useful for advanced liver models. Adding DMSO or
nicotinamide has been found to push primary hepatocytes through one or two rounds of
replication, having the effect of filling out and compacting an in vitro culture (Tateno and
Yoshizato, 1996). Since studies have shown liver regeneration is controlled by interleukin (IL)-1,
basic fibroblast growth factor (FGF), epidermal growth factor (EGF), and insulin-like growth
factor (IGF), commercial media kits have been developed to supplement standard media with
these growth factors and others, including ascorbic acid, hydrocortisone, transferring, and insulin
(Mohn et al., 1991). Although not typically added to culture media, morphogens (cytokines
active in organogenesis) wnt, bone morphogenic protein (BMP) and Hedgehog are known to
control liver acinus zonation, and could be carefully applied to further develop a sinusoid-mimic
model (Hendriks and Reichmann, 2002).
As liver organogenesis and regeneration are controlled by soluble factors, media
perfusion should also be considered in advanced liver models. Perfusion is not trivial as the
complexity of perfusing even a single culture dramatically increases with the introduction of
fluidics and pumps, thus lowering throughput that may be needed for larger liver assays. In
acinus zonation, wnt/β-catenin signaling gradients are essential, as β-catenin drives
differentiation toward pericentral phenotypes. Adenomatous polyposis coli (APC) is a negative
regulator of β-catenin. Normally APC forms a gradient whereby the concentration is greater in
the periportal zone. When the APC gene is knocked out, over-expression of β-catenin drives
pericentralization of liver sinusoids, and conversely a β-catenin knockout leads to
periportalization of the acinus (Benhamouche et al., 2006). While far-fetched, perfused liver
models could be designed to produce natural liver zonation in vitro—rapidly advancing the
accuracy of ADME and other studies.
36
A more practical application of media perfusion is to meet oxygen demands of primary
hepatocytes. In the liver sinusoid, a gradient of oxygen persists where the periportal area is well
oxygenated while the pericentral area is less oxygenated (Godoy et al., 2013). To mimic in vivo
liver oxygenation, we must first understand oxygen is poorly soluble in culture media, containing
about 7 mg/L (219 µM) at 37 ºC (Griffith and Swartz, 2006). For this reason, standard wellbased culture methods recommend not exceeding 2-3 mm of media—having the effect of
lowering diffusion distance from ambient oxygen, through media, toward the cell surface.
Complex culture systems are often much thicker than 2-3 mm, and if they are not made of
polydimethylsiloxane (PDMS, which is very gas permeable), perfusion may be required to
supply hepatocytes with adequate amounts of oxygen. This point has been demonstrated in a
perfused culture system featuring 500,000 hepatocytes where a flow rate of 250 µL/min is
required to maintain adequate oxygenation at 50 µM and 1 mL/min to keep dissolved oxygen at
a constant 100 µM (Domansky et al., 2010). A similar study utilizing cell lines found a flow rate
of above 5 µL/min maintained constant oxygenation, but dropping the flow rate to 0.03 µL/min
resulted in a 75% decrease in dissolved oxygen, below the demand of primary hepatocytes
(Mehta et al., 2007). These studies show media oxygenation may be a requirement of complex
culture systems, and if oxygenation is maintained by perfusion, culture systems require several
milliliters or even liters of media per day. Furthermore, additional oxygen tension requirements
may manifest from the final use of a culture system. For Plasmodium research, studies have
shown liver parasites may require the hypoxic conditions of the pericentral zone, as parasites
developing in this region are larger than those developing in the periportal zone (Ng et al., 2014).
The microphysical space and geometric features provided by an advanced culture system
should also be considered in its design. From a practical use aspect, device optical clarity and
37
thinness is essential for monitoring by light microscopy (Epshteyn et al., 2011). Also, membrane
and post features can be integrated into culture systems to allow for indirect perfusion and
indirect co-culture (Huh et al., 2010; Lee et al., 2007). In a parenchymal culture chamber,
hepatocyte phenotypes can be drastically altered by materials and shapes making up their
immediate environment. Primary hepatocytes cultured on hard surfaces such as glass and
plastics, or with ‘hard’ ECM proteins such as collagen I, de-differentiate, lose characteristic
morphology and polarity, and may even proliferate (Wells, 2008). These effects can be partially
reversed by addition of ECM proteins to the top of a hepatocytes monolayer in a sandwich
culture, or in a gel suspension (Underhill et al., 2007). Aside from the gradients formed by
perfused culture, shear force itself, generated from media flow, can have considerable effects on
hepatocytes and co-culture cells. (Morigi et al., 2001; Tilles et al., 2001). As hepatocytes are
located in compacted, single cell sheets featuring tight cell-cell junctions in vivo, replicating such
compaction has also been found advantageous to long-term primary hepatocyte culture (Wang et
al., 2013). It is thought that this compaction helps to maintain the cuboidal architecture and
zonal polarization characteristic of hepatocytes, as well as bile transport functions associated
with sheets of polarized hepatocytes (Maher et al., 2014).
Advanced Cell Culture Platform Fabrication. Selection of cell types, ECM proteins,
media, media perfusion, and culture geometry are biology-centric concerns for advanced liver
model design. However, often these requirements may contradict each other (i.e. thinness versus
dual-culture chambers, or complexity versus scalability) or fabrication methodology may not
exist to achieve complexities, creating engineering-centric concerns (Chao et al., 2009). As full
development of functional polystyrene (PS) culture systems demands many resources,
polydimethylsiloxane (PDMS) is often used to prototype advanced culture systems containing
38
microfeatures using efficient fabrication methods such as SU-8 molding, casting, and lamination
(reviewed in Ni et al., 2009). PDMS is also unique in that it allows rapid diffusion of oxygen,
benefitting some models. The efficiency of PDMS is offset by some of its negative properties,
as it is relatively porous and absorbs small molecules—possibly confounding drug and
metabolite measurements (Toepke and Beebe, 2006). Additionally, often PDMS surfaces need to
be extensively coated with ECM proteins or chemically modified to optimize cell attachment
(Nishikawa et al., 2008). Once a design is relatively optimized, designs can be built into PS
using hot embossing, laser etching, and injection modeling after investments into the necessary
molds and machines are made. Aside from being a hard surface and impermeable to gases, PS is
thought to be relatively inert for cell culture. Like PDMS, PS surface chemistry and
modification are a major concern as fabrication steps can adversely or positively affect suitability
for cell culture. For example, PS culture dishes sold commercially display a variety of surface
roughness that is known to affect cell spreading and function (Zeiger et al., 2013). In the near
future, rapid prototyping by 3D printing may prove as a balance between the prototyping abilities
of PDMS and biocompatibility of PS. Recently, 3D printers have been developed to print at
nearly 100 µm-detail necessary for microfeatures; and validation of 3D printing plastics and 3D
printed matrix proteins for cell culture is underway (Imani et al., 2012; Muller et al., 2013).
Current Progress: Static Liver Systems. Many types of advanced liver culture systems
possessing a myriad of the features described above can be found in the literature. While they
differ on many levels, the presence or absence of perfusion may be the most fundamental
difference as perfusion often dictates design and throughput and few systems are capable of both
static and perfused culture. Thus, advanced in vitro liver culture systems can be arbitrarily
grouped as either static or perfused. Regardless, development of these systems has revealed
39
much about the mechanisms of in vitro liver culture and such revelations should be considered
when designing a novel model.
Following are examples of advances in static liver systems, which primarily make use of
architecture, overlays, and co-cultures. Evenou et al. (2010) integrated micropillars in wells to
induce spheroid formations around pillars. They found the distance between micropillars
impacted longevity of primary hepatocytes; as the distance between pillars decreased, longevity
was improved. Similarly, Matsui et al. (2012) created sandwich cultures of hepatocytes in
microwells and found an ideal diameter of 60-80 µm per microwell as optimal for hepatocytes
confinement. While wells and pillar produce confinement on the hepatocyte lateral axis,
confinement can also be achieved on the vertical axis through sandwich culture. Dunn et al.
(1991) found that rat hepatocytes cultured between two layers of collagen maintained primary
cell functions for up to six weeks and Berthiaume et al. (1996) found a hepatocyte sandwich
culture led to bile canaliculi formation. Sellaro et al. (2010) compared both Matrigel® and
porcine-liver-matrix sandwich culture of primary hepatocytes and found both sandwich cultures
stabilize albumin and urea phenotypes. Matrigel® is commercially available and is widely used
to supplement all types of culture systems with complex ECM.
Co-culture of hepatocytes with nonparenchymal cells was established by GugenGuilluozo et al. (1983), where primary hepatocyte phenotypes were rescued after initiation of coculture with an epithelial cell type. Ries et al. (2000) expanded upon the co-culture phenomenon
by adding a third cell type, generating an endothelial-Küpffer-hepatocyte model capable of stable
glucagon and urea metabolism. Interestingly, this work found the addition of hepatocytes
resulted in loss of endothelial cells and formation of a Küpffer-hepatocytes cell-cell interaction.
Patterned co-cultures offer a means to control heterotypic cell interactions and have been
40
demonstrated to further enhance primary hepatocyte phenotypes, thus patterning methods have
been developed using chemical or physical patterning (Bhatia et al., 1998a). Patterning by
collagen printing has been developed by Bhatia et al. (1998b) to yield an array of hepatocyteseeded collagen island surrounded by fibroblasts. Perhaps one of the most well known models in
bioengineering, this system is capable of extremely stable long-term expression of hepatic
phenotypes, formation of bile canaliculi, and has been used to model drug metabolism, hepatitis
C infection, and Plasmodium infection (Chan et al., 2013; Khetani and Bhatia, 2008; March et
al., 2013; Ploss et al., 2010). Krause et al. (2009) establish a hepatocyte-stellate co-culture by
first patterning hepatocytes within a plastic ring, followed by ring removal and stellate seeding,
having the effect of stabilizing PKC activity. To achieve mixing of paracrine factors from two
cell types, but not mixing of the two cell types themselves, Lee et al. (2013) generated a twochamber culture system with stellate cells in one chamber and hepatocytes in the other; having a
slightly less drastic effect on hepatocyte phenotypes than direct co-culture.
Co-culture can also be directly and indirectly achieved vertically by addition of a nonparenchymal cell type to a hepatocyte monolayer or culture of two cell types across a permeable
barrier. Ohno et al. (2009a, b) allowed endothelial cells to form a confluent sheet on plastic and
then transferred the endothelium to a layer of HepG2 cells, resulting in increased expression of
integrin, decorin, and P450 enzymes. Similarly, Lee et al. (2004) cultured hepatocytes in
spheroid and then overlayed endothelial cells onto the outer layer of the spheroid, forming space
of Disse-like morphology and stabilizing phenotypes. Like indirect horizontal co-cultures,
transwells also allow for two cell types to be seeded on reverse sides of a membrane to allowing
media mixing but not cell mixing. Such a system was developed by Wagner et al. (2013) to
generate a liver-skin model for cosmetic testing.
41
Current Progress: Perfused Liver Systems. Perfusion adds both mechanical factors
(shear stress) and chemical factors (cytokine gradients) to advanced liver systems. However,
often shear or gradients are desired, but not both. When direct perfusion of hepatocytes is
undesirable, a mechanical barrier to flow can be employed to shield hepatocytes under perfusion.
For example, a simple channel defined by posts on both sides and a grooved lower surface for
hepatocyte compaction has been patented by Corning, Inc. (Goral et al., 2010). Perfusion through
channels juxtaposed through the post-defined center channel allowed for two-week culture of
primary hepatocytes, stable MRP2 transport function, and formation of a cord-like multicellular
structure. A similar device developed by Lee at al. (2007) has been partially multiplexed to
allow for toxicity testing of eight concentrations of a compound, generated by gradient tree
attached to a device array. Membranes between two channels, one perfused, and one holding
hepatocytes in static conditions, have also been described as capable of long-term stable culture
(Carraro et al., 2008; Kane et al., 2006). Combinations of co-culture and perfusion have also
resulted in mimetic liver models. Domansky et al. (2010) co-cultured endothelial cells and
hepatocytes in a perfused bioreactor whereby media is flowed through packets of hepatocytes.
This model has been well helpful for a better understand of oxygen requirements in liver
systems, and has been multiplexed and made commercially available.
Comparing and Contrasting Components and Systems. Each of the models described
above, and many others not mentioned, have been qualified by user-defined metrics of success.
These metrics are often repeated across many models, allowing for a comparison to understand
how incrementally more complex systems correlate towards greater similarity to in vivo liver
conditions. Some of the most commonly used metrics are P450 induction, albumin synthesis and
ureagenesis, as these functions are representative of hepatocyte health and function (Allen et al.,
42
2001). Albumin production and ureagenesis are often gathered by collection of media effluent
followed by quantification by enzyme-linked immunosorbent assay (ELISA), revealing a
concentration. Concentrations need to be converted to net quantity produced per unit cell per
time for comparison across models, as production is cell number and time-dependent.
Unfortunately, many reports either do not reveal cell quantification values or convert production
to odd units and do not explain the conversion well enough to allow back-calculation. While
direct comparisons may not always be possible, in general albumin production is confirmed
when an upward or stable trend is seen across several days (Bhatia et al., 1998b). Downward
trends are indicative of poor phenotype (Wagner et al., 2013).
Qualitative methods have also proven useful as hepatocytes can be stained for many
intrinsic features. Oxygen content can be quantified by ruthenium tris (2,2'-dipyridyl) dichloride
hexahydrate staining, glycogen synthesis can be imaged by acid-Schiff staining, and lipid
droplets can be stained by oil red O. Externally, hepatocytes should show apical staining for
MRP2, zona occludens protein (ZO)-1, claudin 1, and occludin in tight junctions as well as
CD26 in basolateral domains (Feracci et al., 1987; Godoy et al., 2013; Ploss et al., 2010; Thomas
et al., 2005). Alternatively, a global approach to assessing liver model performance incorporates
transcription profiling of some or all genes. Studies have shown transcription of at least 3000
genes can be altered by culture conditions, especially those responsible for key hepatic functions
(Kim et al., 2010; Zellmer et al., 2010). However, gene expression profiling is costly and, along
with many qualitative methods, is an endpoint assessment that results in discontinuation of
culture. In general, quantifying specific functions (albumin, P450) and qualitative imaging are
more cost-efficient and high throughput than microarray analysis, but methods throughout the
spectrum of simple to complex are informative.
43
Tailoring Liver System for Experimental Question. The many advanced in vitro liver
cultures methods described above, along with methods to compare them, open a conversation as
to which is the ‘best.’ On one end of this conversation, ‘best’ could be described as more
mimetic to in vivo conditions. Thus, a model integrating multiple cell types, complex ECM, and
sinusoid-like spaces in a perfused system may be termed ‘better’ than perhaps the simplest in
vitro liver model: cell-free microsomes. On the other end of the conversation, the addition of
many features results in loss of throughput and transferability--establishing merit to
reproducibility and simplicity. When solving a biological problem calls for use of advanced in
vitro liver culture, the minimum complexity and throughput needed to answer that biological
question must be established first to help down-select advanced culture options (reviewed in
LeCluyse, 2001).
Focus of Study
At the onset of this study, there was no mimetic, high-throughput, human liver platform
capable of assessing efficacy of new therapeutics to prevent transmission of malaria. To meet
this need, we set out to establish a novel, human in vitro Plasmodium liver stage drug and
vaccine discovery platform, with the end goal of actual use in drug and vaccine discovery
workplaces. We immediately identified critical requirements for such a platform: the platform
must support hepatocytes in long-term culture, the platform must allow high-resolution imaging
of liver stage parasites, the platform must be multiplexed to allow low to medium throughput, the
platform must support complete development of P. falciparum and P. vivax liver stages
(including hypnozoites), the platform must make use of limited hepatocyte and parasite
resources, and the platform must be easy to obtain and use with minimal training.
44
Preliminary studies in our lab successfully produced a bilayer device capable of highresolution imaging by integration of a coverglass into a dual-chamber culture system. This
system was modified and optimized to be quite flexible in its use: supportive of hepatocyte cell
line and primary hepatocytes, capable of monoculture and co-culture in direct and indirect
application, and capable of static or perfused culture. This flexibility was used to identify the
minimal features necessary for long-term static culture, and to understand the incremental
advantages of adding complexities such as co-culture and perfusion. After establishment of a
simple but mimetic long-term human liver model, we further optimized the model to understand
what details and complexities were needed for Plasmodium invasion and development. With
such an understanding, we describe further optimization of the model by multiplexing the
simplest advanced culture unit needed for efficient Plasmodium liver stage development, and
demonstrate utility of the model through actual determination of therapeutic efficacy.
45
Figure 1.1. Global distribution of P. falciparum and P. vivax malaria. a) Global distribution
of P. falciparum malaria, showing highest endemicity in Africa. b) Global distribution of P.
vivax malaria, demonstrating endemicity in South America and Southeast Asia.
*Source: Malaria Atlas Project (Gething et al., 2012; Gething et al., 2011), used with permission
46
Figure 1.2. The Plasmodium life cycle. 1) Infection of a new host begins when an infected
Anopheles mosquito feeds, releasing sporozoites into the dermis. Sporozoites travel to the liver,
traverse the liver endothelium, and invade a hepatocyte, encasing the sporozoite in a hepatocytederived membrane (2). Over 6-10 days liver forms grow into schizonts: multinucleated forms
containing thousands of merozoites (3). P. vivax and P. ovale can also halt division early in the
liver stage and remain dormant as hypnozoites for weeks, months, or years. 4) Completely
developed liver forms burst, releasing merosomes (packets of merozoites) encased in hepatocytes
membrane into the endothelium to initiate infection of erythrocytes or, for P. vivax, reticulocytes.
5) Merozoites invade erythrocytes by initial attachment, reorientation, and active invasion into
the cell, again encasing itself in a host cell membrane. Over 2 days (or 3 for P. malariae) blood
forms develop from trophozoites into schizonts by karyokinesis followed by cytokinesis, forming
8-36 daughter merozoites, which burst out of erythrocytes and re-invade other erythrocytes.
Rounds of replication yield an exponential increase in blood parasitemia, leading to the onset of
disease and diagnosis. After only a few days (for P. vivax) or a few weeks (P. falciparum), male
and female sexual forms, called gametocytes, develop from early trophozoites (7). Mature
gametocytes translocate to the peripheral circulation for ingestion during a mosquito blood meal
(8). Within only a few minutes in the mosquito midgut, male gametes exflagellate and form 8
microgametes which fertilize female macrogametes, undergo sexual recombination, and then
meiosis to form a haploid oökinete (10). After traversing the midgut, oökinetes attach to the
midgut exterior and form oocysts (11). Oocysts undergo karyokinesis and cytokinesis, forming
thousands of sporozoites which travel through the mosquito lymph before finding and invading
the mosquito salivary gland (12). Sporozoites pack into salivary glands to await invasion into the
next host.
*Source: http://www.cdc.gov/malaria/about/biology, used with permission
47
Figure 1.3. Structure of the liver and Plasmodium infection. a) The functional unit of the
liver, portal fields, contain an arteriole, portal venule, and bile ductile. Blood flows into a liver
lobule through the portal vein and hepatic arteriole, and leaves through the central vein (CV).
Bile is pumped into bile canaliculi between hepatocytes, which collect in to bile ductules which
flow out of portal fields. b) The liver sinusoid begins with the merging of the hepatic arteriole
and portal venule. Plasma filters through liver sinusoids (through sieve plates), where blood
components are exchanged with hepatocytes. Waste products collect into bile canaliculi between
hepatocytes. A Plasmodium infection begins when sporozoites (colored in red) enter the liver
and cross the sinusoid, invading hepatocytes.
*Source: Frevert et al., 2006, used with permission
48
References
Adams, J.H., Sim, B.K., Dolan, S.A., Fang, X., Kaslow, D.C., and Miller, L.H. (1992). A family
of erythrocyte binding proteins of malaria parasites. Proc Natl Acad Sci U S A 89, 70857089.
Aide, P., Aponte, J.J., Renom, M., Nhampossa, T., Sacarlal, J., Mandomando, I., Bassat, Q.,
Manaca, M.N., Leach, A., Lievens, M., et al. (2010). Safety, immunogenicity and
duration of protection of the RTS,S/AS02(D) malaria vaccine: one year follow-up of a
randomized controlled phase I/IIb trial. PLoS One 5, e13838.
Albuquerque, S.S., Carret, C., Grosso, A.R., Tarun, A.S., Peng, X., Kappe, S.H., Prudencio, M.,
and Mota, M.M. (2009). Host cell transcriptional profiling during malaria liver stage
infection reveals a coordinated and sequential set of biological events. BMC Genomics
10, 270.
Allen, J.W., Hassanein, T., and Bhatia, S.N. (2001). Advances in bioartificial liver devices.
Hepatology 34, 447-455.
Amino, R., Thiberge, S., Martin, B., Celli, S., Shorte, S., Frischknecht, F., and Menard, R.
(2006). Quantitative imaging of Plasmodium transmission from mosquito to mammal.
Nat Med 12, 220-224.
Antherieu, S., Chesne, C., Li, R., Camus, S., Lahoz, A., Picazo, L., Turpeinen, M., Tolonen, A.,
Uusitalo, J., Guguen-Guillouzo, C., and Guillouzo, A. (2010). Stable expression, activity,
and inducibility of cytochromes P450 in differentiated HepaRG cells. Drug Metab Dispos
38, 516-525.
Arciniegas, E., Frid, M.G., Douglas, I.S., and Stenmark, K.R. (2007). Perspectives on
endothelial-to-mesenchymal transition: potential contribution to vascular remodeling in
chronic pulmonary hypertension. Am J Physiol Lung Cell Mol Physiol 293, L1-8.
Arias, I.M., ed. (2009). The Liver Biology and Pathobiology (Chichester).
Ariey, F., Witkowski, B., Amaratunga, C., Beghain, J., Langlois, A.C., Khim, N., Kim, S., Duru,
V., Bouchier, C., Ma, L., et al. (2014). A molecular marker of artemisinin-resistant
Plasmodium falciparum malaria. Nature 505, 50-55.
Asahi, H., Kanazawa, T., Kajihara, Y., Takahashi, K., and Takahashi, T. (1996). Hypoxanthine:
a low molecular weight factor essential for growth of erythrocytic Plasmodium
falciparum in a serum-free medium. Parasitology 113 ( Pt 1), 19-23.
Ayala, F.J., Escalante, A.A., and Rich, S.M. (1999). Evolution of Plasmodium and the recent
origin of the world populations of Plasmodium falciparum. Parassitologia 41, 55-68.
49
Baer, K., Roosevelt, M., Clarkson, A.B., Jr., van Rooijen, N., Schnieder, T., and Frevert, U.
(2007). Küpffer cells are obligatory for Plasmodium yoelii sporozoite infection of the
liver. Cell Microbiol 9, 397-412.
Ball, P. (2008). History of science: quinine steps back in time. Nature 451, 1065-1066.
Balu, B., Maher, S.P., Pance, A., Chauhan, C., Naumov, A.V., Andrews, R.M., Ellis, P.D., Khan,
S.M., Lin, J.W., Janse, C.J., et al. (2011). CCR4-associated factor 1 coordinates the
expression of Plasmodium falciparum egress and invasion proteins. Eukaryot Cell 10,
1257-1263.
Bapiro, T.E., Andersson, T.B., Otter, C., Hasler, J.A., and Masimirembwa, C.M. (2002).
Cytochrome P450 1A1/2 induction by antiparasitic drugs: dose-dependent increase in
ethoxyresorufin O-deethylase activity and mRNA caused by quinine, primaquine and
albendazole in HepG2 cells. Eur J Clin Pharmacol 58, 537-542.
Bargieri, D.Y., Andenmatten, N., Lagal, V., Thiberge, S., Whitelaw, J.A., Tardieux, I., Meissner,
M., and Menard, R. (2013). Apical membrane antigen 1 mediates apicomplexan parasite
attachment but is dispensable for host cell invasion. Nat Commun 4, 2552.
Benhamouche, S., Decaens, T., Godard, C., Chambrey, R., Rickman, D.S., Moinard, C.,
Vasseur-Cognet, M., Kuo, C.J., Kahn, A., Perret, C., and Colnot, S. (2006). Apc tumor
suppressor gene is the "zonation-keeper" of mouse liver. Dev Cell 10, 759-770.
Bennett, G.F., and Campbell, A.G. (1972). Avian haemoproteidae. I. Description of
Haemoproteus fallisi n. sp. and a review of the haemoproteids of the family Turdidae.
Can J Zool 50, 1269-1275.
Berthiaume, F., Moghe, P.V., Toner, M., and Yarmush, M.L. (1996). Effect of extracellular
matrix topology on cell structure, function, and physiological responsiveness:
hepatocytes cultured in a sandwich configuration. FASEB J 10, 1471-1484.
Bhatia, S.N., Balis, U.J., Yarmush, M.L., and Toner, M. (1998a). Microfabrication of
hepatocyte/fibroblast co-cultures: role of homotypic cell interactions. Biotechnol Prog 14,
378-387.
Bhatia, S.N., Balis, U.J., Yarmush, M.L., and Toner, M. (1998b). Probing heterotypic cell
interactions: hepatocyte function in microfabricated co-cultures. J Biomater Sci Polym
Ed 9, 1137-1160.
Billker, O., Shaw, M.K., Margos, G., and Sinden, R.E. (1997). The roles of temperature, pH and
mosquito factors as triggers of male and female gametogenesis of Plasmodium berghei in
vitro. Parasitology 115 ( Pt 1), 1-7.
50
Boddey, J.A., Moritz, R.L., Simpson, R.J., and Cowman, A.F. (2009). Role of the Plasmodium
export element in trafficking parasite proteins to the infected erythrocyte. Traffic 10, 285299.
Bolchoz, L.J., Budinsky, R.A., McMillan, D.C., and Jollow, D.J. (2001). Primaquine-induced
hemolytic anemia: formation and hemotoxicity of the arylhydroxylamine metabolite 6methoxy-8-hydroxylaminoquinoline. J Pharmacol Exp Ther 297, 509-515.
Braet, F., and Wisse, E. (2002). Structural and functional aspects of liver sinusoidal endothelial
cell fenestrae: a review. Comp Hepatol 1, 1.
Bruce, M.C., Alano, P., Duthie, S., and Carter, R. (1990). Commitment of the malaria parasite
Plasmodium falciparum to sexual and asexual development. Parasitology 100 Pt 2, 191200.
Carraro, A., Hsu, W.M., Kulig, K.M., Cheung, W.S., Miller, M.L., Weinberg, E.J., Swart, E.F.,
Kaazempur-Mofrad, M., Borenstein, J.T., Vacanti, J.P., and Neville, C. (2008). In vitro
analysis of a hepatic device with intrinsic microvascular-based channels. Biomed
Microdevices 10, 795-805.
CDC (2010). The History of Malaria, an Ancient Disease. Retrieved from
http://www.cdc.gov/malaria/about/history/.
CDC (2012a). Choosing a Drug to Prevent Malaria. Retrieved from
http://www.cdc.gov/malaria/travelers/drugs.html
CDC (2012b). Elimination of Malaria in the United States (1947-1951). Retrieved from
http://www.cdc.gov/malaria/about/history/elimination_us.html
Chan, T.S., Yu, H., Moore, A., Khetani, S.R., and Tweedie, D. (2013). Meeting the challenge of
predicting hepatic clearance of compounds slowly metabolized by cytochrome P450
using a novel hepatocyte model, HepatoPac. Drug Metab Dispos 41, 2024-2032.
Chao, P., Maguire, T., Novik, E., Cheng, K.C., and Yarmush, M.L. (2009). Evaluation of a
microfluidic based cell culture platform with primary human hepatocytes for the
prediction of hepatic clearance in human. Biochem Pharmacol 78, 625-632.
Coppi, A., Natarajan, R., Pradel, G., Bennett, B.L., James, E.R., Roggero, M.A., Corradin, G.,
Persson, C., Tewari, R., and Sinnis, P. (2011). The malaria circumsporozoite protein has
two functional domains, each with distinct roles as sporozoites journey from mosquito to
mammalian host. J Exp Med 208, 341-356.
Cowman, A.F., and Crabb, B.S. (2006). Invasion of red blood cells by malaria parasites. Cell
124, 755-766.
51
Cox, F.E. (2010). History of the discovery of the malaria parasites and their vectors. Parasit
Vectors 3, 5.
Creasey, A.M., Ranford-Cartwright, L.C., Moore, D.J., Williamson, D.H., Wilson, R.J.,
Walliker, D., and Carter, R. (1993). Uniparental inheritance of the mitochondrial gene
cytochrome b in Plasmodium falciparum. Curr Genet 23, 360-364.
Daneker, G.W., Lund, S.A., Caughman, S.W., Swerlick, R.A., Fischer, A.H., Staley, C.A., and
Ades, E.W. (1998). Culture and characterization of sinusoidal endothelial cells isolated
from human liver. In Vitro Cell Dev Biol Anim 34, 370-377.
de Koning-Ward, T.F., Speranca, M.A., Waters, A.P., and Janse, C.J. (1999). Analysis of stage
specificity of promoters in Plasmodium berghei using luciferase as a reporter. Mol
Biochem Parasitol 100, 141-146.
Dembele, L., Franetich, J.F., Lorthiois, A., Gego, A., Zeeman, A.M., Kocken, C.H., Le Grand,
R., Dereuddre-Bosquet, N., van Gemert, G.J., Sauerwein, R., et al. (2014). Persistence
and activation of malaria hypnozoites in long-term primary hepatocyte cultures. Nat Med
20, 307-312.
Dembele, L., Gego, A., Zeeman, A.M., Franetich, J.F., Silvie, O., Rametti, A., Le Grand, R.,
Dereuddre-Bosquet, N., Sauerwein, R., van Gemert, G.J., et al. (2011). Towards an in
vitro model of Plasmodium hypnozoites suitable for drug discovery. PLoS One 6,
e18162.
Domansky, K., Inman, W., Serdy, J., Dash, A., Lim, M.H., and Griffith, L.G. (2010). Perfused
multiwell plate for 3D liver tissue engineering. Lab Chip 10, 51-58.
Donato, M.T., Lahoz, A., Castell, J.V., and Gomez-Lechon, M.J. (2008). Cell lines: a tool for in
vitro drug metabolism studies. Curr Drug Metab 9, 1-11.
Dunn, J.C., Tompkins, R.G., and Yarmush, M.L. (1991). Long-term in vitro function of adult
hepatocytes in a collagen sandwich configuration. Biotechnol Prog 7, 237-245.
Ejigiri, I., and Sinnis, P. (2009). Plasmodium sporozoite-host interactions from the dermis to the
hepatocyte. Curr Opin Microbiol 12, 401-407.
Elvevold, K., Smedsrod, B., and Martinez, I. (2008). The liver sinusoidal endothelial cell: a cell
type of controversial and confusing identity. Am J Physiol Gastrointest Liver Physiol
294, G391-400.
Epshteyn, A.A., Maher, S., Taylor, A.J., Holton, A.B., Borenstein, J.T., and Cuiffi, J.D. (2011).
Membrane-integrated microfluidic device for high-resolution live cell imaging.
Biomicrofluidics 5, 46501-465016.
52
Epstein, J.E., Tewari, K., Lyke, K.E., Sim, B.K., Billingsley, P.F., Laurens, M.B., Gunasekera,
A., Chakravarty, S., James, E.R., Sedegah, M., et al. (2011). Live attenuated malaria
vaccine designed to protect through hepatic CD8(+) T cell immunity. Science 334, 475480.
Evenou, F., Fujii, T., and Sakai, Y. (2010). Spontaneous formation of stably-attached and 3Dorganized hepatocyte aggregates on oxygen-permeable polydimethylsiloxane membranes
having 3D microstructures. Biomed Microdevices 12, 465-475.
Feracci, H., Connolly, T.P., Margolis, R.N., and Hubbard, A.L. (1987). The establishment of
hepatocyte cell surface polarity during fetal liver development. Dev Biol 123, 73-84.
Flaim, C.J., Chien, S., and Bhatia, S.N. (2005). An extracellular matrix microarray for probing
cellular differentiation. Nat Methods 2, 119-125.
Franke-Fayard, B., Trueman, H., Ramesar, J., Mendoza, J., van der Keur, M., van der Linden, R.,
Sinden, R.E., Waters, A.P., and Janse, C.J. (2004). A Plasmodium berghei reference line
that constitutively expresses GFP at a high level throughout the complete life cycle. Mol
Biochem Parasitol 137, 23-33.
Frevert, U., Usynin, I., Baer, K., and Klotz, C. (2006). Nomadic or sessile: can Küpffer cells
function as portals for malaria sporozoites to the liver? Cell Microbiol 8, 1537-1546.
Friedman, S.L. (2008). Hepatic stellate cells: protean, multifunctional, and enigmatic cells of the
liver. Physiol Rev 88, 125-172.
Gardner, M.J., Hall, N., Fung, E., White, O., Berriman, M., Hyman, R.W., Carlton, J.M., Pain,
A., Nelson, K.E., Bowman, S., et al. (2002). Genome sequence of the human malaria
parasite Plasmodium falciparum. Nature 419, 498-511.
Gardner, M.J., Tettelin, H., Carucci, D.J., Cummings, L.M., Smith, H.O., Fraser, C.M., Venter,
J.C., and Hoffman, S.L. (1999). The malaria genome sequencing project: complete
sequence of Plasmodium falciparum chromosome 2. Parassitologia 41, 69-75.
Gerets, H.H., Tilmant, K., Gerin, B., Chanteux, H., Depelchin, B.O., Dhalluin, S., and Atienzar,
F.A. (2012). Characterization of primary human hepatocytes, HepG2 cells, and HepaRG
cells at the mRNA level and CYP activity in response to inducers and their predictivity
for the detection of human hepatotoxins. Cell Biol Toxicol 28, 69-87.
Gething, P.W., Elyazar, I.R., Moyes, C.L., Smith, D.L., Battle, K.E., Guerra, C.A., Patil, A.P.,
Tatem, A.J., Howes, R.E., Myers, M.F., et al. (2012). A long neglected world malaria
map: Plasmodium vivax endemicity in 2010. PLoS Negl Trop Dis 6, e1814.
Gething, P.W., Patil, A.P., Smith, D.L., Guerra, C.A., Elyazar, I.R., Johnston, G.L., Tatem, A.J.,
and Hay, S.I. (2011). A new world malaria map: Plasmodium falciparum endemicity in
2010. Malar J 10, 378.
53
Godoy, P., Hewitt, N.J., Albrecht, U., Andersen, M.E., Ansari, N., Bhattacharya, S., Bode, J.G.,
Bolleyn, J., Borner, C., Bottger, J., et al. (2013). Recent advances in 2D and 3D in vitro
systems using primary hepatocytes, alternative hepatocyte sources and non-parenchymal
liver cells and their use in investigating mechanisms of hepatotoxicity, cell signaling and
ADME. Arch Toxicol 87, 1315-1530.
Goldin, R.D., Ratnayaka, I.D., Breach, C.S., Brown, I.N., and Wickramasinghe, S.N. (1996).
Role of macrophages in acetaminophen (paracetamol)-induced hepatotoxicity. J Pathol
179, 432-435.
Goral, V.N., Hsieh, Y.C., Petzold, O.N., Clark, J.S., Yuen, P.K., and Faris, R.A. (2010).
Perfusion-based microfluidic device for three-dimensional dynamic primary human
hepatocyte cell culture in the absence of biological or synthetic matrices or coagulants.
Lab Chip 10, 3380-3386.
Greenwood, B.M., Fidock, D.A., Kyle, D.E., Kappe, S.H., Alonso, P.L., Collins, F.H., and
Duffy, P.E. (2008). Malaria: progress, perils, and prospects for eradication. J Clin Invest
118, 1266-1276.
Griffith, L.G., and Swartz, M.A. (2006). Capturing complex 3D tissue physiology in vitro. Nat
Rev Mol Cell Biol 7, 211-224.
Guguen-Guillouzo, C., Campion, J.P., Brissot, P., Glaise, D., Launois, B., Bourel, M., and
Guillouzo, A. (1982). High yield preparation of isolated human adult hepatocytes by
enzymatic perfusion of the liver. Cell Biol Int Rep 6, 625-628.
Guguen-Guillouzo, C., Clement, B., Baffet, G., Beaumont, C., Morel-Chany, E., Glaise, D., and
Guillouzo, A. (1983). Maintenance and reversibility of active albumin secretion by adult
rat hepatocytes co-cultured with another liver epithelial cell type. Exp Cell Res 143, 4754.
Hantz, O., Parent, R., Durantel, D., Gripon, P., Guguen-Guillouzo, C., and Zoulim, F. (2009).
Persistence of the hepatitis B virus covalently closed circular DNA in HepaRG human
hepatocyte-like cells. J Gen Virol 90, 127-135.
Hastings, I. (2011). How artemisinin-containing combination therapies slow the spread of
antimalarial drug resistance. Trends Parasitol 27, 67-72.
Hauben, M., and Hung, E.Y. (2013). A quantitative analysis of the spontaneous reporting of
congestive heart failure-related adverse events with systemic anti-fungal drugs. J Clin
Pharmacol 53, 762-772.
Hendriks, B., and Reichmann, E. (2002). Wnt signaling: a complex issue. Biol Res 35, 277-286.
54
Herrmann, J., Gressner, A.M., and Weiskirchen, R. (2007). Immortal hepatic stellate cell lines:
useful tools to study hepatic stellate cell biology and function? J Cell Mol Med 11, 704722.
Heussler, V.T., and Stanway, R.R. (2008). Cellular and molecular interactions between the
apicomplexan parasites Plasmodium and Theileria and their host cells. Parasite 15, 211218.
Holt, M.P., Cheng, L., and Ju, C. (2008). Identification and characterization of infiltrating
macrophages in acetaminophen-induced liver injury. J Leukoc Biol 84, 1410-1421.
Hoover, M.L., Vetvicka, V., Hoffpauir, J.M., and Tamburro, C.H. (1993). Human endothelial
cell line from an angiosarcoma. In Vitro Cell Dev Biol 29A, 199-202.
Horwich, A.L., Furtak, K., Pugh, J., and Summers, J. (1990). Synthesis of hepadnavirus particles
that contain replication-defective duck hepatitis B virus genomes in cultured HuH7 cells.
J Virol 64, 642-650.
Huh, D., Matthews, B.D., Mammoto, A., Montoya-Zavala, M., Hsin, H.Y., and Ingber, D.E.
(2010). Reconstituting organ-level lung functions on a chip. Science 328, 1662-1668.
Imani, R., Hojjati Emami, S., Fakhrzadeh, H., Baheiraei, N., and Sharifi, A.M. (2012).
Optimization and comparison of two different 3D culture methods to prepare cell
aggregates as a bioink for organ printing. Biocell 36, 37-45.
Imwong, M., Boel, M.E., Pagornrat, W., Pimanpanarak, M., McGready, R., Day, N.P., Nosten,
F., and White, N.J. (2012). The first Plasmodium vivax relapses of life are usually
genetically homologous. J Infect Dis 205, 680-683.
Iyanagi, T. (2007). Molecular mechanism of phase I and phase II drug-metabolizing enzymes:
implications for detoxification. Int Rev Cytol 260, 35-112.
Kaiser, K., Camargo, N., Coppens, I., Morrisey, J.M., Vaidya, A.B., and Kappe, S.H. (2004). A
member of a conserved Plasmodium protein family with membrane-attack
complex/perforin (MACPF)-like domains localizes to the micronemes of sporozoites.
Mol Biochem Parasitol 133, 15-26.
Kane, B.J., Zinner, M.J., Yarmush, M.L., and Toner, M. (2006). Liver-specific functional studies
in a microfluidic array of primary mammalian hepatocytes. Anal Chem 78, 4291-4298.
Kaneko, A., Taleo, G., Kalkoa, M., Yamar, S., Kobayakawa, T., and Bjorkman, A. (2000).
Malaria eradication on islands. Lancet 356, 1560-1564.
55
Karnasuta, C., Pavanand, K., Chantakulkij, S., Luttiwongsakorn, N., Rassamesoraj, M.,
Laohathai, K., Webster, H.K., and Watt, G. (1995). Complete development of the liver
stage of Plasmodium falciparum in a human hepatoma cell line. Am J Trop Med Hyg 53,
607-611.
Keeley, A., and Soldati, D. (2004). The glideosome: a molecular machine powering motility and
host-cell invasion by Apicomplexa. Trends Cell Biol 14, 528-532.
Keeling, P.J. (2009). Chromalveolates and the evolution of plastids by secondary endosymbiosis.
J Eukaryot Microbiol 56, 1-8.
Khetani, S.R., and Bhatia, S.N. (2008). Microscale culture of human liver cells for drug
development. Nat Biotechnol 26, 120-126.
Killick-Kendrick, R. (1978). Recent advances and outstanding problems in the biology of
phlebotomine sandflies. A review. Acta Trop 35, 297-313.
Kim, S.H., Banga, S., Jha, K.K., and Ozer, H.L. (1998). SV40-mediated transformation and
immortalization of human cells. Dev Biol Stand 94, 297-302.
Kim, Y., Lasher, C.D., Milford, L.M., Murali, T.M., and Rajagopalan, P. (2010). A comparative
study of genome-wide transcriptional profiles of primary hepatocytes in collagen
sandwich and monolayer cultures. Tissue Eng Part C Methods 16, 1449-1460.
Krause, P., Saghatolislam, F., Koenig, S., Unthan-Fechner, K., and Probst, I. (2009). Maintaining
hepatocyte differentiation in vitro through co-culture with hepatic stellate cells. In Vitro
Cell Dev Biol Anim 45, 205-212.
Lacrue, A.N., Saenz, F.E., Cross, R.M., Udenze, K.O., Monastyrskyi, A., Stein, S., Mutka, T.S.,
Manetsch, R., and Kyle, D.E. (2013). 4(1H)-Quinolones with liver stage activity against
Plasmodium berghei. Antimicrob Agents Chemother 57, 417-424.
Langhi, D.M., Jr., and Bordin, J.O. (2006). Duffy blood group and malaria. Hematology 11, 389398.
Langhorne, J., Ndungu, F.M., Sponaas, A.M., and Marsh, K. (2008). Immunity to malaria: more
questions than answers. Nat Immunol 9, 725-732.
Leclerc, M.C., Durand, P., Gauthier, C., Patot, S., Billotte, N., Menegon, M., Severini, C., Ayala,
F.J., and Renaud, F. (2004). Meager genetic variability of the human malaria agent
Plasmodium vivax. Proc Natl Acad Sci U S A 101, 14455-14460.
LeCluyse, E.L. (2001). Human hepatocyte culture systems for the in vitro evaluation of
cytochrome P450 expression and regulation. Eur J Pharm Sci 13, 343-368.
56
Lee, K.W., Park, J.B., Yoon, J.J., Lee, J.H., Kim, S.Y., Jung, H.J., Lee, S.K., Kim, S.J., Lee,
H.H., Lee, D.S., and Joh, J.W. (2004). The viability and function of cryopreserved
hepatocyte spheroids with different cryopreservation solutions. Transplant Proc 36, 24622463.
Lee, P.J., Hung, P.J., and Lee, L.P. (2007). An artificial liver sinusoid with a microfluidic
endothelial-like barrier for primary hepatocyte culture. Biotechnol Bioeng 97, 1340-1346.
Lee, S.A., No da, Y., Kang, E., Ju, J., Kim, D.S., and Lee, S.H. (2013). Spheroid-based threedimensional liver-on-a-chip to investigate hepatocyte-hepatic stellate cell interactions and
flow effects. Lab Chip 13, 3529-3537.
Li, A.P. (2007). Human hepatocytes: isolation, cryopreservation and applications in drug
development. Chem Biol Interact 168, 16-29.
Lin, P., Chan, W.C., Badylak, S.F., and Bhatia, S.N. (2004). Assessing porcine liver-derived
biomatrix for hepatic tissue engineering. Tissue Eng 10, 1046-1053.
Lingelbach, K., and Joiner, K.A. (1998). The parasitophorous vacuole membrane surrounding
Plasmodium and Toxoplasma: an unusual compartment in infected cells. J Cell Sci 111
( Pt 11), 1467-1475.
Liu Tsang, V., Chen, A.A., Cho, L.M., Jadin, K.D., Sah, R.L., DeLong, S., West, J.L., and
Bhatia, S.N. (2007). Fabrication of 3D hepatic tissues by additive photopatterning of
cellular hydrogels. FASEB J 21, 790-801.
Liu, W., Li, Y., Learn, G.H., Rudicell, R.S., Robertson, J.D., Keele, B.F., Ndjango, J.B., Sanz,
C.M., Morgan, D.B., Locatelli, S., et al. (2010). Origin of the human malaria parasite
Plasmodium falciparum in gorillas. Nature 467, 420-425.
Liu, W., Li, Y., Shaw, K.S., Learn, G.H., Plenderleith, L.J., Malenke, J.A., Sundararaman, S.A.,
Ramirez, M.A., Crystal, P.A., Smith, A.G., et al. (2014). African origin of the malaria
parasite Plasmodium vivax. Nat Commun 5, 3346.
Liu, Y., Tewari, R., Ning, J., Blagborough, A.M., Garbom, S., Pei, J., Grishin, N.V., Steele, R.E.,
Sinden, R.E., Snell, W.J., and Billker, O. (2008). The conserved plant sterility gene
HAP2 functions after attachment of fusogenic membranes in Chlamydomonas and
Plasmodium gametes. Genes Dev 22, 1051-1068.
Llanos-Cuentas, A., Lacerda, M.V., Rueangweerayut, R., Krudsood, S., Gupta, S.K., Kochar,
S.K., Arthur, P., Chuenchom, N., Mohrle, J.J., Duparc, S., et al. (2014). Tafenoquine plus
chloroquine for the treatment and relapse prevention of Plasmodium vivax malaria
(DETECTIVE): a multicentre, double-blind, randomised, phase 2b dose-selection study.
Lancet 383, 1049-1058.
57
Maher, S.P., Crouse, R.B., Conway, A.J., Bannister, E.C., Achyuta, A.K., Clark, A.Y., Sinatra,
F.L., Cuiffi, J.D., Adams, J.H., Kyle, D.E., and Saadi, W.M. (2014). Microphysical space
of a liver sinusoid device enables simplified long-term maintenance of chimeric mouseexpanded human hepatocytes. Biomed Microdevices.
malERA (2011a). A research agenda for malaria eradication: basic science and enabling
technologies. PLoS Med 8, e1000399.
malERA (2011b). A research agenda for malaria eradication: drugs. PLoS Med 8, e1000402.
malERA (2011c). A research agenda for malaria eradication: vaccines. PLoS Med 8, e1000398.
March, S., Ng, S., Velmurugan, S., Galstian, A., Shan, J., Logan, D.J., Carpenter, A.E., Thomas,
D., Sim, B.K., Mota, M.M., et al. (2013). A microscale human liver platform that
supports the hepatic stages of Plasmodium falciparum and vivax. Cell Host Microbe 14,
104-115.
Markus, M.B. (2011). The hypnozoite concept, with particular reference to malaria. Parasitol Res
108, 247-252.
Martin, M.J., Rayner, J.C., Gagneux, P., Barnwell, J.W., and Varki, A. (2005). Evolution of
human-chimpanzee differences in malaria susceptibility: relationship to human genetic
loss of N-glycolylneuraminic acid. Proc Natl Acad Sci U S A 102, 12819-12824.
Martinsen, E.S., Perkins, S.L., and Schall, J.J. (2008). A three-genome phylogeny of malaria
parasites (Plasmodium and closely related genera): evolution of life-history traits and
host switches. Mol Phylogenet Evol 47, 261-273.
Materne, E.M., Tonevitsky, A.G., and Marx, U. (2013). Chip-based liver equivalents for toxicity
testing--organotypicalness versus cost-efficient high throughput. Lab Chip 13, 34813495.
Matsui, H., Takeuchi, S., Osada, T., Fujii, T., and Sakai, Y. (2012). Enhanced bile canaliculi
formation enabling direct recovery of biliary metabolites of hepatocytes in 3D collagen
gel microcavities. Lab Chip 12, 1857-1864.
Matuschewski, K., Nunes, A.C., Nussenzweig, V., and Menard, R. (2002). Plasmodium
sporozoite invasion into insect and mammalian cells is directed by the same dual binding
system. EMBO J 21, 1597-1606.
Mazier, D., Beaudoin, R.L., Mellouk, S., Druilhe, P., Texier, B., Trosper, J., Miltgen, F., Landau,
I., Paul, C., Brandicourt, O., and et al. (1985). Complete development of hepatic stages of
Plasmodium falciparum in vitro. Science 227, 440-442.
58
Mazier, D., Landau, I., Druilhe, P., Miltgen, F., Guguen-Guillouzo, C., Baccam, D., Baxter, J.,
Chigot, J.P., and Gentilini, M. (1984). Cultivation of the liver forms of Plasmodium vivax
in human hepatocytes. Nature 307, 367-369.
Mehta, G., Mehta, K., Sud, D., Song, J.W., Bersano-Begey, T., Futai, N., Heo, Y.S., Mycek,
M.A., Linderman, J.J., and Takayama, S. (2007). Quantitative measurement and control
of oxygen levels in microfluidic poly(dimethylsiloxane) bioreactors during cell culture.
Biomed Microdevices 9, 123-134.
Menard, R., Sultan, A.A., Cortes, C., Altszuler, R., van Dijk, M.R., Janse, C.J., Waters, A.P.,
Nussenzweig, R.S., and Nussenzweig, V. (1997). Circumsporozoite protein is required
for development of malaria sporozoites in mosquitoes. Nature 385, 336-340.
Mikolajczak, S.A., Jacobs-Lorena, V., MacKellar, D.C., Camargo, N., and Kappe, S.H. (2007).
L-FABP is a critical host factor for successful malaria liver stage development. Int J
Parasitol 37, 483-489.
Mohn, K.L., Melby, A.E., Tewari, D.S., Laz, T.M., and Taub, R. (1991). The gene encoding rat
insulinlike growth factor-binding protein 1 is rapidly and highly induced in regenerating
liver. Mol Cell Biol 11, 1393-1401.
Morigi, M., Galbusera, M., Binda, E., Imberti, B., Gastoldi, S., Remuzzi, A., Zoja, C., and
Remuzzi, G. (2001). Verotoxin-1-induced up-regulation of adhesive molecules renders
microvascular endothelial cells thrombogenic at high shear stress. Blood 98, 1828-1835.
Morris, K.M., Aden, D.P., Knowles, B.B., and Colten, H.R. (1982). Complement biosynthesis by
the human hepatoma-derived cell line HepG2. J Clin Invest 70, 906-913.
Mota, M.M., Hafalla, J.C., and Rodriguez, A. (2002). Migration through host cells activates
Plasmodium sporozoites for infection. Nat Med 8, 1318-1322.
Mota, M.M., Pradel, G., Vanderberg, J.P., Hafalla, J.C., Frevert, U., Nussenzweig, R.S.,
Nussenzweig, V., and Rodriguez, A. (2001). Migration of Plasmodium sporozoites
through cells before infection. Science 291, 141-144.
Moura, P.A., Dame, J.B., and Fidock, D.A. (2009). Role of Plasmodium falciparum digestive
vacuole plasmepsins in the specificity and antimalarial mode of action of cysteine and
aspartic protease inhibitors. Antimicrob Agents Chemother 53, 4968-4978.
Mu, J., Duan, J., Makova, K.D., Joy, D.A., Huynh, C.Q., Branch, O.H., Li, W.H., and Su, X.Z.
(2002). Chromosome-wide SNPs reveal an ancient origin for Plasmodium falciparum.
Nature 418, 323-326.
Muller, M., Becher, J., Schnabelrauch, M., and Zenobi-Wong, M. (2013). Printing
thermoresponsive reverse molds for the creation of patterned two-component hydrogels
for 3D cell culture. J Vis Exp, e50632.
59
Nava, S., Westgren, M., Jaksch, M., Tibell, A., Broome, U., Ericzon, B.G., and SumitranHolgersson, S. (2005). Characterization of cells in the developing human liver.
Differentiation 73, 249-260.
Neghina, R., Neghina, A.M., Marincu, I., Iacobiciu, I. (2010). Malaria, a journey in time: in
search of the lost myths and forgotten stories. Am J Med Sci 340, 492-498.
Nelson, D.R. (1999). Cytochrome P450 and the individuality of species. Arch Biochem Biophys
369, 1-10.
Nelson, D.R., Zeldin, D.C., Hoffman, S.M., Maltais, L.J., Wain, H.M., and Nebert, D.W. (2004).
Comparison of cytochrome P450 (CYP) genes from the mouse and human genomes,
including nomenclature recommendations for genes, pseudogenes and alternative-splice
variants. Pharmacogenetics 14, 1-18.
Ng, S., March, S., Galstian, A., Hanson, K., Carvalho, T., Mota, M.M., and Bhatia, S.N. (2014).
Hypoxia promotes liver-stage malaria infection in primary human hepatocytes in vitro.
Dis Model Mech 7, 215-224.
Ni, M., Tong, W.H., Choudhury, D., Rahim, N.A., Iliescu, C., and Yu, H. (2009). Cell culture on
MEMS platforms: a review. Int J Mol Sci 10, 5411-5441.
Nishikawa, M., Yamamoto, T., Kojima, N., Kikuo, K., Fujii, T., and Sakai, Y. (2008). Stable
immobilization of rat hepatocytes as hemispheroids onto collagen-conjugated polydimethylsiloxane (PDMS) surfaces: importance of direct oxygenation through PDMS for
both formation and function. Biotechnol Bioeng 99, 1472-1481.
Noble, E.R., ed. (1976). Parasitology: The Biology of Animal Parasites (Philadelphia: Lea &
Febiger).
Noedl, H. (2005). Artemisinin resistance: how can we find it? Trends Parasitol 21, 404-405.
Ohno, M., Motojima, K., Okano, T., and Taniguchi, A. (2008). Up-regulation of drugmetabolizing enzyme genes in layered co-culture of a human liver cell line and
endothelial cells. Tissue Eng Part A 14, 1861-1869.
Ohno, M., Motojima, K., Okano, T., and Taniguchi, A. (2009a). Induction of drug-metabolizing
enzymes by phenobarbital in layered co-culture of a human liver cell line and endothelial
cells. Biol Pharm Bull 32, 813-817.
Ohno, M., Motojima, K., Okano, T., and Taniguchi, A. (2009b). Maturation of the extracellular
matrix and cell adhesion molecules in layered co-cultures of HepG2 and endothelial cells.
J Biochem 145, 591-597.
60
Okura, H., Komoda, H., Saga, A., Kakuta-Yamamoto, A., Hamada, Y., Fumimoto, Y., Lee,
C.M., Ichinose, A., Sawa, Y., and Matsuyama, A. (2010). Properties of hepatocyte-like
cell clusters from human adipose tissue-derived mesenchymal stem cells. Tissue Eng Part
C Methods 16, 761-770.
Painter, H.J., Campbell, T.L., and Llinas, M. (2011). The Apicomplexan AP2 family: integral
factors regulating Plasmodium development. Mol Biochem Parasitol 176, 1-7.
Ploss, A., Khetani, S.R., Jones, C.T., Syder, A.J., Trehan, K., Gaysinskaya, V.A., Mu, K., Ritola,
K., Rice, C.M., and Bhatia, S.N. (2010). Persistent hepatitis C virus infection in
microscale primary human hepatocyte cultures. Proc Natl Acad Sci U S A 107, 31413145.
Pombo, D.J., Lawrence, G., Hirunpetcharat, C., Rzepczyk, C., Bryden, M., Cloonan, N.,
Anderson, K., Mahakunkijcharoen, Y., Martin, L.B., Wilson, D., et al. (2002). Immunity
to malaria after administration of ultra-low doses of red cells infected with Plasmodium
falciparum. Lancet 360, 610-617.
Pradel, G., Garapaty, S., and Frevert, U. (2004). Küpffer and stellate cell proteoglycans mediate
malaria sporozoite targeting to the liver. Comp Hepatol 3 Suppl 1, S47.
Provin, C., Takano, K., Yoshida, T., Sakai, Y., Fujii, T., and Shirakashi, R. (2009). Low O2
metabolism of HepG2 cells cultured at high density in a 3D microstructured scaffold.
Biomed Microdevices 11, 485-494.
Pybus, B.S., Sousa, J.C., Jin, X., Ferguson, J.A., Christian, R.E., Barnhart, R., Vuong, C., Sciotti,
R.J., Reichard, G.A., Kozar, M.P., et al. (2012). CYP450 phenotyping and accurate mass
identification of metabolites of the 8-aminoquinoline, anti-malarial drug primaquine.
Malar J 11, 259.
Ralph, S.A., Foth, B.J., Hall, N., and McFadden, G.I. (2004). Evolutionary pressures on
apicoplast transit peptides. Mol Biol Evol 21, 2183-2194.
Rhodes, J., ed. (2007). Textbook of hepatology: from basic science to clinicla practice (Malden:
Blackwell).
Rich, S.M., Leendertz, F.H., Xu, G., LeBreton, M., Djoko, C.F., Aminake, M.N., Takang, E.E.,
Diffo, J.L., Pike, B.L., Rosenthal, B.M., et al. (2009). The origin of malignant malaria.
Proc Natl Acad Sci U S A 106, 14902-14907.
Ries, K., Krause, P., Solsbacher, M., Schwartz, P., Unthan-Fechner, K., Christ, B., Markus,
P.M., and Probst, I. (2000). Elevated expression of hormone-regulated rat hepatocyte
functions in a new serum-free hepatocyte-stromal cell coculture model. In Vitro Cell Dev
Biol Anim 36, 502-512.
61
Riglar, D.T., Richard, D., Wilson, D.W., Boyle, M.J., Dekiwadia, C., Turnbull, L., Angrisano,
F., Marapana, D.S., Rogers, K.L., Whitchurch, C.B., et al. (2011). Super-resolution
dissection of coordinated events during malaria parasite invasion of the human
erythrocyte. Cell Host Microbe 9, 9-20.
Riglar, D.T., Rogers, K.L., Hanssen, E., Turnbull, L., Bullen, H.E., Charnaud, S.C., Przyborski,
J., Gilson, P.R., Whitchurch, C.B., Crabb, B.S., et al. (2013). Spatial association with
PTEX complexes defines regions for effector export into Plasmodium falciparuminfected erythrocytes. Nat Commun 4, 1415.
Roberts, L.S., ed. (2000). Foundation of Parasitology, 6th edn (Boston: McGraw-Hill Higher
Education).
Sattabongkot, J., Yimamnuaychoke, N., Leelaudomlipi, S., Rasameesoraj, M., Jenwithisuk, R.,
Coleman, R.E., Udomsangpetch, R., Cui, L., and Brewer, T.G. (2006). Establishment of a
human hepatocyte line that supports in vitro development of the exo-erythrocytic stages
of the malaria parasites Plasmodium falciparum and P. vivax. Am J Trop Med Hyg 74,
708-715.
Scherf, A., Lopez-Rubio, J.J., and Riviere, L. (2008). Antigenic variation in Plasmodium
falciparum. Annu Rev Microbiol 62, 445-470.
Sellaro, T.L., Ranade, A., Faulk, D.M., McCabe, G.P., Dorko, K., Badylak, S.F., and Strom, S.C.
(2010). Maintenance of human hepatocyte function in vitro by liver-derived extracellular
matrix gels. Tissue Eng Part A 16, 1075-1082.
Sellaro, T.L., Ravindra, A.K., Stolz, D.B., and Badylak, S.F. (2007). Maintenance of hepatic
sinusoidal endothelial cell phenotype in vitro using organ-specific extracellular matrix
scaffolds. Tissue Eng 13, 2301-2310.
Si-Tayeb, K., Noto, F.K., Nagaoka, M., Li, J., Battle, M.A., Duris, C., North, P.E., Dalton, S.,
and Duncan, S.A. (2010). Highly efficient generation of human hepatocyte-like cells
from induced pluripotent stem cells. Hepatology 51, 297-305.
Siddiqui, A.A., Xainli, J., Schloegel, J., Carias, L., Ntumngia, F., Shoham, M., Casey, J.L.,
Foley, M., Adams, J.H., and King, C.L. (2012). Fine specificity of Plasmodium vivax
Duffy binding protein binding engagement of the Duffy antigen on human erythrocytes.
Infect Immun 80, 2920-2928.
Silva, J.C., Egan, A., Friedman, R., Munro, J.B., Carlton, J.M., and Hughes, A.L. (2011).
Genome sequences reveal divergence times of malaria parasite lineages. Parasitology
138, 1737-1749.
Silvestrini, F., Alano, P., and Williams, J.L. (2000). Commitment to the production of male and
female gametocytes in the human malaria parasite Plasmodium falciparum. Parasitology
121 Pt 5, 465-471.
62
Silvie, O., Franetich, J.F., Charrin, S., Mueller, M.S., Siau, A., Bodescot, M., Rubinstein, E.,
Hannoun, L., Charoenvit, Y., Kocken, C.H., et al. (2004). A role for apical membrane
antigen 1 during invasion of hepatocytes by Plasmodium falciparum sporozoites. J Biol
Chem 279, 9490-9496.
Sinden, R.E. (1985). Gametocytogenesis in Plasmodium spp., and observations on the meiotic
division. Ann Soc Belg Med Trop 65 Suppl 2, 21-33.
Sinden, R.E. (2009). Malaria, sexual development and transmission: retrospect and prospect.
Parasitology 136, 1427-1434.
Singh, S.S. (2006). Preclinical pharmacokinetics: an approach towards safer and efficacious
drugs. Curr Drug Metab 7, 165-182.
Smith, T.G., Lourenco, P., Carter, R., Walliker, D., and Ranford-Cartwright, L.C. (2000).
Commitment to sexual differentiation in the human malaria parasite, Plasmodium
falciparum. Parasitology 121 ( Pt 2), 127-133.
Strom, S.C., Davila, J., and Grompe, M. (2010). Chimeric mice with humanized liver: tools for
the study of drug metabolism, excretion, and toxicity. Methods Mol Biol 640, 491-509.
Strom, S.C., Jirtle, R.L., Jones, R.S., Novicki, D.L., Rosenberg, M.R., Novotny, A., Irons, G.,
McLain, J.R., and Michalopoulos, G. (1982). Isolation, culture, and transplantation of
human hepatocytes. J Natl Cancer Inst 68, 771-778.
Sunman, J.A., Hawke, R.L., LeCluyse, E.L., and Kashuba, A.D. (2004). Küpffer cell-mediated
IL-2 suppression of CYP3A activity in human hepatocytes. Drug Metab Dispos 32, 359363.
Takebe, T., Sekine, K., Enomura, M., Koike, H., Kimura, M., Ogaeri, T., Zhang, R.R., Ueno, Y.,
Zheng, Y.W., Koike, N., et al. (2013). Vascularized and functional human liver from an
iPSC-derived organ bud transplant. Nature 499, 481-484.
Tateno, C., and Yoshizato, K. (1996). Long-term cultivation of adult rat hepatocytes that
undergo multiple cell divisions and express normal parenchymal phenotypes. Am J
Pathol 148, 383-392.
Tavares, J., Formaglio, P., Thiberge, S., Mordelet, E., Van Rooijen, N., Medvinsky, A., Menard,
R., and Amino, R. (2013). Role of host cell traversal by the malaria sporozoite during
liver infection. J Exp Med 210, 905-915.
Thomas, R.J., Bhandari, R., Barrett, D.A., Bennett, A.J., Fry, J.R., Powe, D., Thomson, B.J., and
Shakesheff, K.M. (2005). The effect of three-dimensional co-culture of hepatocytes and
hepatic stellate cells on key hepatocyte functions in vitro. Cells Tissues Organs 181, 6779.
63
Tilles, A.W., Baskaran, H., Roy, P., Yarmush, M.L., and Toner, M. (2001). Effects of
oxygenation and flow on the viability and function of rat hepatocytes cocultured in a
microchannel flat-plate bioreactor. Biotechnol Bioeng 73, 379-389.
Tilley, L., Dixon, M.W., and Kirk, K. (2011). The Plasmodium falciparum-infected red blood
cell. Int J Biochem Cell Biol 43, 839-842.
Toepke, M.W., and Beebe, D.J. (2006). PDMS absorption of small molecules and consequences
in microfluidic applications. Lab Chip 6, 1484-1486.
Tolia, N.H., Enemark, E.J., Sim, B.K., and Joshua-Tor, L. (2005). Structural basis for the EBA175 erythrocyte invasion pathway of the malaria parasite Plasmodium falciparum. Cell
122, 183-193.
Underhill, G.H., Chen, A.A., Albrecht, D.R., and Bhatia, S.N. (2007). Assessment of
hepatocellular function within PEG hydrogels. Biomaterials 28, 256-270.
Uneke, C.J. (2007). Impact of placental Plasmodium falciparum malaria on pregnancy and
perinatal outcome in sub-Saharan Africa: I: introduction to placental malaria. Yale J Biol
Med 80, 39-50.
Usynin, I., Klotz, C., and Frevert, U. (2007). Malaria circumsporozoite protein inhibits the
respiratory burst in Küpffer cells. Cell Microbiol 9, 2610-2628.
Valkiunas, G., Anwar, A.M., Atkinson, C.T., Greiner, E.C., Paperna, I., and Peirce, M.A. (2005).
What distinguishes malaria parasites from other pigmented haemosporidians? Trends
Parasitol 21, 357-358.
van de Sand, C., Horstmann, S., Schmidt, A., Sturm, A., Bolte, S., Krueger, A., Lutgehetmann,
M., Pollok, J.M., Libert, C., and Heussler, V.T. (2005). The liver stage of Plasmodium
berghei inhibits host cell apoptosis. Mol Microbiol 58, 731-742.
van Dijk, M.R., Douradinha, B., Franke-Fayard, B., Heussler, V., van Dooren, M.W., van
Schaijk, B., van Gemert, G.J., Sauerwein, R.W., Mota, M.M., Waters, A.P., and Janse,
C.J. (2005). Genetically attenuated, P36p-deficient malarial sporozoites induce protective
immunity and apoptosis of infected liver cells. Proc Natl Acad Sci U S A 102, 1219412199.
Vaughan, A.M., Mikolajczak, S.A., Wilson, E.M., Grompe, M., Kaushansky, A., Camargo, N.,
Bial, J., Ploss, A., and Kappe, S.H. (2012). Complete Plasmodium falciparum liver-stage
development in liver-chimeric mice. J Clin Invest 122, 3618-3628.
Vaughan, A.M., O'Neill, M.T., Tarun, A.S., Camargo, N., Phuong, T.M., Aly, A.S., Cowman,
A.F., and Kappe, S.H. (2009). Type II fatty acid synthesis is essential only for malaria
parasite late liver stage development. Cell Microbiol 11, 506-520.
64
von Seidlein, L., Walraven, G., Milligan, P.J., Alexander, N., Manneh, F., Deen, J.L., Coleman,
R., Jawara, M., Lindsay, S.W., Drakeley, C., et al. (2003). The effect of mass
administration of sulfadoxine-pyrimethamine combined with artesunate on malaria
incidence: a double-blind, community-randomized, placebo-controlled trial in The
Gambia. Trans R Soc Trop Med Hyg 97, 217-225.
Wagner, I., Materne, E.M., Brincker, S., Sussbier, U., Fradrich, C., Busek, M., Sonntag, F.,
Sakharov, D.A., Trushkin, E.V., Tonevitsky, A.G., et al. (2013). A dynamic multi-organchip for long-term cultivation and substance testing proven by 3D human liver and skin
tissue co-culture. Lab Chip 13, 3538-3547.
Walker, T.M., Rhodes, P.C., and Westmoreland, C. (2000). The differential cytotoxicity of
methotrexate in rat hepatocyte monolayer and spheroid cultures. Toxicol In Vitro 14,
475-485.
Wallace, D.J. (1996). The history of antimalarials. Lupus 5 Suppl 1, S2-3.
Wang, R., Smith, J.D., and Kappe, S.H. (2009). Advances and challenges in malaria vaccine
development. Expert Rev Mol Med 11, e39.
Wang, Y., Toh, Y.C., Li, Q., Nugraha, B., Zheng, B., Lu, T.B., Gao, Y., Ng, M.M., and Yu, H.
(2013). Mechanical compaction directly modulates the dynamics of bile canaliculi
formation. Integr Biol (Camb) 5, 390-401.
Wells, R.G. (2008). The role of matrix stiffness in regulating cell behavior. Hepatology 47,
1394-1400.
White, N.J. (2014). Malaria: a molecular marker of artemisinin resistance. Lancet 383, 14391440.
WHO (2010). Guidelines for the treatment of malaria, second edition. Retrieved from
http://www.who.int/malaria/publications/atoz/9789241547925/en/
WHO (2014). Malaria Fact Sheet. Retrieved from
http://www.who.int/malaria/media/world_malaria_report_2013/en/
Wilkening, S., Stahl, F., and Bader, A. (2003). Comparison of primary human hepatocytes and
hepatoma cell line Hepg2 with regard to their biotransformation properties. Drug Metab
Dispos 31, 1035-1042.
Wisse, E. (1972). An ultrastructural characterization of the endothelial cell in the rat liver
sinusoid under normal and various experimental conditions, as a contribution to the
distinction between endothelial and Küpffer cells. J Ultrastruct Res 38, 528-562.
65
Xu, L., Hui, A.Y., Albanis, E., Arthur, M.J., O'Byrne, S.M., Blaner, W.S., Mukherjee, P.,
Friedman, S.L., and Eng, F.J. (2005). Human hepatic stellate cell lines, LX-1 and LX-2:
new tools for analysis of hepatic fibrosis. Gut 54, 142-151.
Yu, Y., Liu, H., Ikeda, Y., Amiot, B.P., Rinaldo, P., Duncan, S.A., and Nyberg, S.L. (2012).
Hepatocyte-like cells differentiated from human induced pluripotent stem cells: relevance
to cellular therapies. Stem Cell Res 9, 196-207.
Zeiger, A.S., Hinton, B., and Van Vliet, K.J. (2013). Why the dish makes a difference:
quantitative comparison of polystyrene culture surfaces. Acta Biomater 9, 7354-7361.
Zellmer, S., Schmidt-Heck, W., Godoy, P., Weng, H., Meyer, C., Lehmann, T., Sparna, T.,
Schormann, W., Hammad, S., Kreutz, C., et al. (2010). Transcription factors ETF, E2F,
and SP-1 are involved in cytokine-independent proliferation of murine hepatocytes.
Hepatology 52, 2127-2136.
66
Chapter Two:
Development of a Long-Term, Mimetic Human Liver Model
Rationale of Study
While dozens of in vitro liver models have been described featuring a full spectrum of
divergent properties and capabilities, remarkably few have been designed and used to model an
infectious disease. Modeling an infectious disease in vitro requires specific host cells (dictated by
the infectious agent) in stable culture (Mazier et al., 1985; Ploss et al., 2010). Progress in liverrelated research has yielded several cell lines of each distinct cell type of the liver; and many
companies offer primary liver cell types. Biotechnology advances describe liver culture methods
featuring plates, channels, reactors, monocultures, sandwich cultures, co-cultures, and media
perfusion. (reviewed in LeCluyse, 2001). Taken together, hundreds of thousands of liver model
permutations exist, yet for a model to play host to an infectious agent, a specific mix of
conditions must be identified to create a precise in vivo-like environment. Thus, we began this
project by designing a novel and flexible liver model that could support and test many of the
candidate cell types and culture phenomena taken from the literature.
Liver models are hepatocyte-centric as it is the parenchymal cell of the liver. We
obtained three different hepatocyte lines (HC-04, HepG2, and Huh-7), all of which having been
used in some capacity as Plasmodium liver models. We also obtained primary hepatocytes from
four differ companies, and one source of hepatocyte-like cells from induced Embryonic Stem
cells (iES), because human Plasmodium parasites are notoriously selective for unknown host cell
67
characteristics possibly lost in cell lines (March et al., 2013). The parasite’s interaction with
nonparenchymal fractions of liver, specifically Küpffer cells, endothelium, and proteoglycans
has been wonderfully shown as essential steps in the host cell invasion process (Coppi et al.,
2007; Frevert et al., 2005; Menard et al., 1997). We obtained two types of human stellate cells
(which generate liver extracellular matrix protein), primary human Küpffer cells, and two types
of human liver endothelial cells. Cell types were characterized for differentiated phenotype and
were assessed for compliance in various co-cultures to elucidate which combinations were
possible.
Next, we engineered methodologies to fabricate, seed, and continue culture in transwells
and microfluidic devices. Because human Plasmodium liver stages require a week to initially
develop, and perhaps three or more weeks for dormant parasites to initiate development, we
generated protocols for long-term liver cell culture—a rare feature found among available liver
models. To test our models and the possible combinations of hepatocytes, nonparenchymal cells,
and physical factors, we developed a flexible liver model to measure hepatocyte phenotypes
throughout long-term assays. These assays revealed components essential to a long-term,
mimetic human liver model.
Because Plasmodium parasites have evolved to live in circulation during most of their
life cycle, and previous studies with Toxoplasma (another Apicomplexan species) demonstrate
shear stress is essential during endothelium traversal (Harker et al., 2014), we adapted our liver
model to be able to mimic blood flow by media perfusion at in vivo-like rates through the
devices. Perfused culture is inherently much more complicated than static culture and much
effort was needed to optimize a protocol for long-term perfused liver culture.
68
Using our versatile liver platform we established comparisons between cell lines and
primary cells, monocultures and co-cultures, and static versus perfused culture; looking for
dramatic upregulation of liver-specific phenotypes. Serendipitously, upon testing a new supplier
of primary hepatocytes, we found the liver device was able to maintain primary hepatocyte
morphology and function without the need for complexities like co-culture or flow. We
hypothesized that either compact space or a low volume of media was responsible for the
phenomena, and confirmed that microphysical space was responsible. This lead to a publication
describing the versatility of our liver model and how complex hepatic phenotypes can be
stabilized by a relatively simple concept (Maher et al., 2014). Our research produced a liver
model that can function on a simple level and then be modified to better understand what factors
are necessary for optimized infection by malaria parasites (Chapter Three).
Materials and Methods
Hepatic Cell Culture. Several liver cell types, both primary and cell lines, were
obtained, expanded, and cryobanked in Cell Freezing Media (Sigma Aldrich, St. Louis, MO,
USA). All cells were cultured in a humidified 5% CO2 incubator. Every cell type required
different culture media and treated surfaces, per manufactures’ instructions, but all hepatocytes
were cultured in 5 µg/cm2 collagen coated flasks and immortalized lines were subcultured 1:10
with Trypsin LE (Invitrogen/Life Technologies, Grand Island, NY, USA) prior to reaching
confluence. T-75 flasks were prepared by overnight addition of 100 µL stock rat tail collagen I
(BD/Corning Life Sciences, Tewksbury, MA, USA) mixed in 5 mL 0.02 N acetic acid (Thermo
Fisher Scientific, Waltham, MA, USA). Human hepatocyte line HepG2 [(Knowles et al., 1980),
(ATCC, Manassas, VA, USA)] were cultured in EMEM (ATCC) supplemented with 10% FBS
(Hyclone/GE Healthcare Bio-Sciences, Pittsburgh, PA, USA). Human hepatocyte line Huh-7
69
[(Pugh et al., 1988), (ATCC)] were cultured in RPMI 1640 (Sigma Aldrich) supplemented with
10% FBS and 0.25% NaHCO3. Human hepatocyte line HC-04 (Sattabongkot et al., 2006) were
cultured with a 1:1 mixture of F12 and MEMα (Life Technologies) supplemented with 10%
FBS, 0.03M HEPES (Life Technologies), and 2 mM L-Glutamine (Life Technologies). Vials of
cryopreserved primary human hepatocytes obtained from Lonza (Portsmouth, NH, USA),
Yecuris (Tualatin, OR, USA), and Bioreclamation IVT (Baltimore, MD, USA) were thawed
according to manufacturer’s instructions and cultured with daily media change with slightly
modified manufacturer’s recommended media. Upon finding that antimycotics could negatively
affect parasite growth, amphotericin B was omitted from Lonza Hepatocyte Complete Medium
(HCM) and Torpedo antibiotic mix was replaced with 1x Penicillin/Streptomycin (Life
Technologies) in Bioreclamation IVT CP Medium. iES Hepatocytes (provided by Dr. Hongkui
Deng through the Bill & Melinda Gates Consortium) were cultured in DMEM-F12 (Life
Technologies) supplemented with an HCM SingleQuotesTM kit (Lonza), 10 ng/mL Oncostatin M
(R&D Systems, Minneapolis, MN, USA) and 1 µM Dexamethasone (Sigma Aldrich). iES
Hepatocytes were collected from a shipping flask with Dispase and Cell Recovery Solution
(BD/Corning) and subcultured to a flask coated with 0.25 mg/mL Matrigel® (BD/Corning). iES
Hepatocytes tested positive for mycoplasma with the MycoAlertTM kit (Lonza) and cleaning was
attempted with a MycoZapTM kit (Lonza), however hepatocytes cultures did not survive
mycoplasma treatment.
All liver sinusoidal endothelial cell types were cultured on 2 µg/cm2 fibronectin coated
flasks. T-75 flasks were prepared by overnight addition of 150 µL stock bovine fibronectin
(Sigma) mixed in 5 mL PBS. Human Angiosarcoma Endothelial cells isolation from human
liver [(HAEND), (Hoover et al., 1993)] were cultured in RPMI 1640 (Life Technologies)
70
supplemented with 10% FBS and were subcultivated 1:3 prior to reaching full confluence.
Primary Human Liver Sinusoidal Endothelial Cells [(HLSEC), (Sciencell, Carlsbad, CA, USA)]
were cultured in manufacture’s complete medium with supplied endothelial growth supplements.
HLSEC’s could be subcultured up to three times (or about 8 cell duplications) using
manufacturer’s trypsin and protocol before overgrowth by contaminating fibroblasts abrogated
continued culture.
All liver nonparenchymal cell lines were cultured in collagen-coated T-75 flasks
(prepared as above) and subcultivated 1:10 before reaching full confluence. Human Stellate cell
lines LX-1 and LX-2 (Xu et al., 2005) were cultured in DMEM (Life Technologies)
supplemented with 10% FBS (LX-1 are optimized for low serum and were cultured in 1% FBS)
and 2 mM L-Glutamine. LX-2 tested positive for and were successfully cleared of mycoplasma
with above mentioned kits. Mouse fibroblast line NIH-3T3 [(Hoover et al., 1993), (ATCC)]
were cultured in DMEM (ATCC) supplemented with 10% FBS.
Live Cell Staining and Quantification. Several method were assessed for quantifying
both the number and health of cells in well, transwell, and device culture. Imaging was
performed on either a Zeiss Observer Z1 (Carl Zeiss Microscopy, Thronwood, NY, USA) or a
Deltavision® Core (GE Healthcare Bio-Sciences) featuring image deconvolution with the
Softworx® Image Processing Suite (GE Healthcare Bio-Sciences). Alternatively, an ImagePro
(Media Cybernetics, Rockville, MD, USA) image processing macro was developed to rapidly
quantify monolayers of Hoechst-stained nuclei. Live cell stains were added to cultures for 3060min at RT or 37 ºC followed by washing with PBS and media. Live cell stains included 10
µg/mL Hoechst 33342 (Sigma) in media or PBS (detecting live and dead DNA), 1:1000 SYBR®
Green I (Life Technologies) in media or PBS (detecting live and dead DNA), 10 µM
71
CellTrackerTM Vybrant® Green or Red (Life Technologies) in PBS (detecting and tracking live
cells over days), 500 nM MitoTracker® Red (Life Technologies) in basal media (detecting
mitochondria), 250 nM Tubulin TrackerTM Green (Life Technologies) mixed 1:1 with Pluronic
F-127 in PBS (detecting tubulin) in basal media, 30 nM Rhodamine Phalloidin (Life
Technologies) in PBS (detecting actin), 1:500 Draq® 5 (Biostatus, Leicestershire, UK) in media
(detecting live and dead nuclear DNA) in media, 1:100 Draq® 7 (Biostatus) in media (detecting
dead DNA only) in media, 1:2000 Calcein AM (Life Technologies) in media (staining live cell
cytoplasm green) in media and 1:500 Ethidium Homodimer (Life Technologies) in media
(staining dead nuclei red).
Four different methods were tested to rapidly quantify live cells utilizing either a
SynergyTM HT plate reader (Biotek, Winooski, VT, USA) or NanoDropTM DNA Spectrometer
(Thermo Fisher Scientific) for phenotype assay cell-number standardization: Hoechst staining,
CellTiter-Glo® (Promega, Madison, WI, USA), PrestoBlue® (Life Technologies) and DNA
quantification by DNeasy® column (Qiagen, Valencia, CA, USA). HepG2 cells were diluted
into 12 microtiter wells ranging from 5,000 to 90,000 cells per well and subsequently stained
with Hoechst, or incubated with CellTiter-Glo® Reagent per manufacturers protocol. Light
emission values were collected at 505 nm (CellTiter-Glo®) or 481 nm (Hoechst) by plate reader.
Similarly, a dilution of HepG2 cells across 12 microtiter wells were subjected to trypsinization
and digestion by DNeasy® kit per manufacture’s protocols. DNA recovered from each well was
measured by spectrometer for quantification. Because PrestoBlue® was the only method found
useful for non-endpoint quantification, it was the only method tested on microfluidic devices.
Devices seeded to confluence with HepG2 cells were flowed at 100 µL/hr for 1 hr with 1:10
72
PrestoBlue® in media and effluent was collected and 590 nm fluorescence measured by plate
reader.
Hepatic Cell Phenotyping. Well and device-cultured hepatocyte albumin production
was measured by periodic collection of 50-100 µL media during daily media exchanges.
Samples were stored at -20 ºC until the final timepoint, when samples were stained and imaged
for cell number. Samples were diluted 5-50x in dilution buffer supplied by either of two human
albumin-specific ELISA kits used (Bethyl Labs, Montgomery, TX, USA or AssayPro, St.
Charles, MO, USA). Factor IX was measured by ELISA kit (AssayPro) from the same samples
as albumin production after 5x dilution in kit-supplied dilution buffer. Results were analyzed by
multiple student’s T tests in Prism (Graphpad, La Jolla, CA, USA)
Hepatocyte CYP 3A4 induction was measured by well and device-cultured hepatocyte
treatment with 25 µM Rifampin (MP Biomedical, Santa Ana, CA, USA) or 1:1000 DMSO in
media for 72 hrs followed by a slight modification to the P450-Glo® protocol (Promega); after
addition of Luciferin-IPA, effluent was collected and then Detection Reagent was flowed into
devices or wells and then added to the already collected Luciferin-IPA effluent.
For characterization of alpha-fetoprotein, primary hepatocytes (Yecuris) cultured within
devices for 3 weeks and HepG2 cells in devices were fixed with ice-cold methanol for 5 min,
permeabilized with 0.3% Triton X100 for 15 min, blocked with 3% BSA for 15 min, stained
with 1:200 mouse anti-alpha-fetoprotein for 1 hr, and then washed and stained with 1:1000 goatanti-mouse 568 (Live Technologies) for 1 hr. For staining of bile canaliculi, primary hepatocytes
(Yecuris) were cultured within devices for 3 weeks and then stained with Hoechst (as above) and
2 µg/mL 5-(and-6)-Carboxy-2’,7’-Dichlorofluorescein Diacetate (Life Technologies) for 10 min.
73
CFDA is metabolized by hepatocytes to yield fluorescein, which is then pumped into bile
canaliculi for image analysis (Khetani and Bhatia, 2008; Zamek-Gliszczynski et al., 2003).
HLSEC and HAEND endothelial cells were cultured on polycarbonate transwells and
stained (as above) with 1:100 anti-von Wildebrand factor (factor VIII, Abcam), 1:100 anti-Eselectin (Abcam), or 40 µg/mL anti-ICAM-1 (Abcam) following activation by 100ng/mL Tumor
Necrosis Factor α (TNFα) treatment in media for 6 hrs at 37 ºC (Daneker et al., 1998).
Endothelial cells were assessed for ability to internalize Low Density Lipoprotein (LDL) by
addition of 10 µg/mL Alexa Fluor®-LDL (Life Technologies) to culture media for 4hr at 37ºC,
followed by Hoechst staining and live cell imaging of Alexa Fluor® 488 within endothelials.
Co-culture Seeding and Culture. Fibroblasts, HAEND and LX-2 cell lines were
cultured under differential media by adding quarter fractions of alternative media to the base
media, increasing the quarter fraction per week until after four weeks cells survived and
propagated in 100% alternative media; these cell were then cryobanked. These lines were
optimized to survive in both HC-04 media and HCM (Lonza) in anticipation of being
nonparenchymal co-culture components. Otherwise, co-cultures were refreshed with a 1:1
mixture of the representative medias of cells in culture. Unpatterned co-cultures between
hepatocyte lines or primary hepatocytes with fibroblasts or stellate cells were obtained by mixing
the parenchymal to non-parenchymal fractions 1:1, 1:3, 1:5, or 1:10 (depending on the
experimental question) and adding the mixture collagen coated wells. Patterned well-cultures of
hepatocyte lines and primary hepatocytes in co-culture with fibroblasts or stellate cells were
obtained by dropping 10 µL 0.02 N acetic acid containing 0.74 µg rat collagen I (BD/Corning)
into a microtiter plate well and allowing the collagen to adsorb to the plate surface for 1 hr.
Adsorption was confirmed by immunofluorescent assay for collagen I (Abcam). The collagen
74
solution was then washed off and primary or hepatocyte cell lines were concentrated to achieve a
seeding value of 150,000 cells/cm2 and each cell type was allowed to adhere to the culture
surface before subsequent washing and continuous culture. After initial studies demonstrated
nonparenchymal cell types overgrow hepatocytes, fibroblasts were treated with 20 µg/mL
mitomycin C in complete media for 3 hrs prior to use in co-cultures, having the effect of ceasing
cell division (Chen et al., 2006).
Transwell Seeding and Culture. Transwells were seeded in a two step process. The
transwell side eventually facing the culture plate surface was seeded first in indirect co-culture
studies, by first inverting the transwell and seeding hepatocytes at 150,000 cells/cm2 in a 100 µL
media droplet (Figure 2.5a). Provided the transwell was collagen or fibronectin coated and dried
prior to cell seeding, droplets did not flow through transwells, presumably due to surface tension.
After 1-3 hrs (depending on cell type), transwells were flipped upright and the opposite layer was
seeded with the nonparenchymal cell type at 150,000 cells/cm2.
Microfluidic Device Fabrication. A detailed summary of device fabrication has been
published (Maher et al., 2014). In brief, three generations of microfluidic devices were used in
this study. The first generation was the longest (approximately 3 cm in channel length) and was
assembled using silane chemistry (Epshteyn et al., 2011). The second generation was shortened
to about 1 cm in length and was assembled with silicon glue. After careful optimization of
polydimethylsiloxane (PDMS) membrane fabrication and further understanding of diffusion
limitation within the channels, a third generation device was designed with an even shorter
channel (4.1 mm) and fabricated using oxygen plasma treatment. All devices were fabricated
from sandwiched components—the bottom consisted of a No. 1 glass coverslip bonded to a 100
µm-thick slice of PDMS with a 250 µm-wide and 7.4 mm-long laminated channel. The third
75
generation device incorporated either a 10 µm-thick polyethylene (PET) membrane with a total
10% porosity of random 8 µm diameter pores or a 10 µm-thick spun PDMS membrane with a
7.25% total porosity of evenly-patterned 10 µm-diameter pores. This membrane separated the
lower and upper layers and in doing so formed the bilayer feature. At the top a 500 µm-thick
PDMS block incorporating a 500 µm-wide and 7.4 mm-long cast channel. After originating in
separate areas the upper and lower channel are juxtaposed for about 4 mm (Fig. 2.1). Interface
with channels was accomplished with a #2 biopsy punch into either channel’s inlet and outlet,
followed by insertion of 0.02in ID (inside diameter), 0.06in OD (outside diameter) tubing into
biopsy holes. Then, channels could be filled with extracellular matrix protein, cells, and media
by a 50 µL glass and steel 22 gauge blunt end syringe (Thermo Fisher Scientific) inserted into
the tubing.
Microfluidic Device Culture Initiation. After ethylene oxide sterilization, devices were
placed in standard tall petri dishes (Corning) and 2 mL of sterile water was added to the dish to
control local humidity, as PDMS is highly permeable and devices were prone to drying. All
seeding steps were performed by controlling flow through the device by clipping tubing channel
ports with stainless steel binder clips. Devices were coated by injecting a 1.4 µg/µL solution of
rat collagen I in 0.02 N acetic acid into the lower channel, or a 0.6 µg/µL solution of bovine
fibronectin in PBS into the upper channel, or both components combined in one solution into
both channels and allowing adsorption overnight, thus achieving channels coated with 5 µg/cm2
collagen and/or 2 µg/cm2 fibronectin. Adsorption was confirmed by immunofluorescent assay
for collagen I (Abcam). Coating solutions were then washed out of device channels with media
and cells were thawed (per manufacturer’s instructions) or trypsinized and concentrated to either
15,000 cells/µL (for hepatocyte lines), 10,000 cells/µL (for primary hepatocytes), or 2,400
76
cells/µL (for endothelials cell types) and then injected into either the upper (endothelial) or lower
(parenchymal) channel. Hepatocyte cell lines adhered to the device surface after about 1 hr while
primary cell types required 3-4 hrs to adhere.
Microfluidic Device Extended Use and Adaptation. Following seeding, tubing was
removed from devices with sterile forceps and the lower 1 cm of custom-cut and sterilized 1 mL
pipette tips was inserted into channel ports, forming a media reservoir above each channel port
containing approximately 40 µL of media each. Alternatively, perfused culture was initiated by
inserting tubing connected to a syringe and pump (Harvard Apparatus, Holliston, MA, USA) into
the channel ports and flowing at a variety of flow rates; notably 2.5 µL/hr (for hepatocyte cell
lines) or 137 µL/hr (for primary hepatocytes). Shear stresses were calculated using an Excel
macro developed by Dr. Anil Achyuta. HLSEC required overnight orbital shaking at 1,000 rpm
prior to device seed (to align cells to flow) and two days to reach confluence in the device upper
channel before perfusion. Long-term culture (over 21 days) was accomplished with daily media
change of static systems and periodic syringe changes in perfused systems.
To achieve indirect co-culture across the device membrane, after initially seeding the
upper channel with endothelial cells, sterile wetted tubing was re-seated into device channel
ports and the parenchymal cell type was injected into the lower channel as described above.
Direct co-culture in the lower channel was similarly achieved by a two step seeding after the first
cell type was adhered. A collagen or Matrigel® sandwich culture could also be initiated by
injection of cold (wet ice) collagen gel (generated by mixing 100 µL 10x PBS, 13 µL 1 N NaOH,
341 µL cold water, and 546 µL cold stock collagen to a cold tube) or 2 mg/mL Matrigel® in cold
media into the lower channel.
77
Results
Hepatic Cell Candidate Genotypes and Phenotypes. All three hepatocyte lines
obtained were characterized for 24 hr production of albumin and CYP 3A4 (Figure 2.2a). HC-04
demonstrated higher differentiation than HepG2 and Huh-7, as expected, given it is the only cell
line reported to also support Plasmodium development—a presumably very high-order
phenotype (Sattabongkot et al., 2006). Due to a lack of provenance for both HepG2 and HC-04,
and concerns that the two lines may have been accidentally switched at some point in their
dissemination, a SNP chip was performed to genotype the two lines. Analysis revealed that the
two lines differed in 5.7% of ~1,000,000 SNP’s analyzed, whereas four random individuals
from the HapMap project differ at 38% of SNP’s (Figure 2.2b). These data suggest the two cell
lines may be divergent; however, in conjunction with phenotype data, they are not the same.
HC-04 were designated the leading candidate for further studies to optimize both the liver model
and Plasmodium liver model, while HepG2 were also used for some optimizations. Primary
hepatocytes obtained from Lonza, Yecuris, and Bioreclamation IVT differed significantly in
viability and platability. Several Lonza-distributed lots were only 10-20% viable post-thaw,
while Yecuris were 70-80% viable and Bioreclamation IVT were >90% viable. Further
characterization of primary cells is covered in Chapter 3.
Endothelial cells (HLSEC and HAEND) were also characterized for several phenotypes
typical of endothelial and liver sinusoidal endothelial cells. Most endothelial cells express von
Wildebrand Factor, while liver endothelial cells uptake Low Density Lipoprotein and express
ICAM-1 and E-Selectin constitutively—expression can increase after stimulation by TNFα
(Daneker et al., 1998). The two endothelial cell types did stain as expected for these factors, yet
HAEND present drastically different morphology than HSLEC (Figure 2.2c). HLSEC form tight
78
sheets of cobblestone morphology after a few days in culture, however HAEND avoid cell-cell
contact and spread in low confluence cultures. Furthermore, when seeded into channels and
perfused at various flow rates, HAEND did not tolerate shear while HLSEC could be
conditioned to survive 0.1 dyn/cm2 of shear after overnight shaking at 1000 rpm in a starter flask.
Generation and Optimization of Cell Tracking, Cell Quantification, Phenotyping,
and Co-culture Protocols. In order to compare and contrast different complex culture methods
in efforts to generate a more complete liver model, phenotypic assays such as albumin
production should be corrected for by cell number as cell number directly affects net
metabolism. While many organ-on-chip models are too opaque or not thin enough for direct
observation (Chao et al., 2009; Domansky et al., 2010), the devices used in this work are
observable by basic and high resolution microscopy, thus both imaging-based and effluent-based
assays were considered viable options.
Counting assays can be split into two groups: endpoint assays, where the culture being counted is
destroyed by the count, and non-endpoint assays, where the culture is healthy after the method is
complete and future sample and counts can continue to be collected. We examined both
endpoint (Live/Dead stain, trypsin-based cell collection, and DNA spectrometry) as well as nonendpoint (Hoechst staining and PrestoBlue®) assays. Standard curves associating signal to cell
number were successfully generated from well-based cultures of hepatocytes cell lines for the
CellTiter-Glo®, Hoechst, DNeasy® and PrestoBlue® methods (Figures 2.3 and 2.4). Attempts to
retrieve cells via trypsinization (Bhatia et al., 1998) for future counting by plate reader or flow
cytometry failed due to clogging of freed cells in a small channel (Figure 2.3e,f). While a
protocol utilizing a DNeasy® kit for device counting has been described (Donato et al., 2008),
devices used in this study harbored 10-fold fewer hepatocytes, thus the quantity of DNA
79
recovered was below the detection limit of the assay. Similar detection limits also ruled out the
CellTiter-Glo® kit; and both the DNease and CellTiter-Glo® methods required cell lysis—
abrogating potential imaging data. With PretoBlue®, we discovered a relationship between
perfused devices and a set of standardization wells could not be established as perfused cultures
experience greater mixing through convection and metabolized Resazurin more efficiently
(Figure 2.4f). The Live/Dead assay, using microscopy as an endpoint, was found not only to
provide numerical data but also revealed cell morphology, and was therefore selected for use in
future assays. Coupling Live-Dead images with counting software was necessary to improved
time efficiency (Figure 2.3b).
The albumin ELISA kits offered through Bethyl Labs and AssayPro were confirmed to
be specific to human albumin; some kits also reacted to bovine albumin which is present in FBS
in media and confounded results. Protocols were generated and optimized for assessment of
albumin and factor IX production, which incorporated identification of proper daily culture
methods, collection methodologies for up to hundreds of samples per experiment, and endpoint
staining for cell morphology and quantification to adjust production by cell number. Thus,
results were made readily comparable across experimental conditions and for comparison to
results found in the literature. Our cytochrome P450 3A4 assay was selected as CYP3A4
metabolizes more than 30% of approved pharmaceuticals and a substrate-based assay reveals
true metabolism, as opposed to measuring CYP transcription levels, (Rodriguez-Antona et al.,
2001). We successfully optimized the assay for use with wells and devices, requiring a modified
protocol to recovery signal from devices using flow.
Methods to seed different versions of co-culture, including random, patterned, and
transwell co-cultures, were generated to measure co-culture affect among candidate cell types.
80
Co-culture assembly is not simply the addition of two cell types to one culture system, as each
cell type is highly adapted to a specific media and each cell type propagates at different rates,
ranging from no propagation (for primary hepatocytes) to rapid propagation (hepatocyte cell
lines and fibroblast). We began by testing different seeding orders, with either the parenchymal
or non-parenchymal fraction being seeded first, and found that allowing hepatocyte cell lines to
propagate prior to adding fibroblasts resulted in generation of hepatocyte islands (Figure 2.5) not
dissimilar from previously published co-culture models (Bhatia et al., 1997; Rojkind et al.,
1995). Fibroblasts tended to rapidly overgrow cultures, thus mitomycin C (a chemotherapy drug
which inhibits cell division by cross-linking nuclear DNA) treatment was tested and found to be
extremely toxic for long-term culture (Figure 2.5a). To better organize co-cultures for long-term
studies, and in an attempt to circumvent cell line overgrowth, we next explored patterned and
transwell co-culture methods. A droplet collagen-coating method was developed (Figure 2.6a)
which, in combination with washing after attachment of hepatocytes to collagen surfaces but
before attachment to non-coated surfaces, resulted in a circular pattern of cells which could then
be backfilled in with nonparenchymal cells (Figure 2.6c,d). Transwell co-culture was
accomplished by selectively coating the upper and lower surfaces of the transwells with
fibronectin and collagen (respectively) and seeding HLSEC and HC-04 on opposite surfaces
(Figure 2.6e,f).
Assessment of Co-culture Contributions and Drawbacks. After co-culture protocols
were optimized, we began investigating the effect of co-culture on hepatocyte albumin
production. Initial studies with HepG2 cells and fibroblasts revealed toxicity in short-term (24
hr) co-culture, possibly due to incomplete media conditioning (Figure 2.7a). As phenotypic
effects, such as gene expression regulation, of co-culture can take days to manifest, we
81
completed conditioning of LX1 stellate cells to hepatocyte media for extended culture with both
primary hepatocytes and HC-04 cells. One week of extended culture resulted in loss of
phenotype for primary hepatocytes, and no gain-of-function from HC-04 cells (Figure 2.7d).
Previous studies have shown a patterned co-culture is superior to random co-culture (Bhatia et
al., 1998), yet patterned co-culture with primary hepatocytes was found difficult to reproducible
achieve, possibly due to low out-of-the-tube viability of primary hepatocytes, and incomplete
attachment (Figure 2.6d). Initial studies with transwells demonstrated extremely low throughput
and limited imaging capabilities (membranes of transwells are not flat, disrupting phase
microscopy). Further optimizations were not attempted after efforts shifted to microfluidic
devices culture, as devices feature optical clarity, organized co-culture and simplified use over
well and transwell co-culture.
Microfluidic Device Development and Optimization.
First Generation Device Successes and Failures. Due to extremely small channel
dimensions (7.5 mm2 lower channel surface area and 0.75 µL volume) collagen and cell seeding
solutions were concentrated to achieve sufficient protein coating and confluent seeding (Figure
2.8a,b). For immediate confluence, cell lines were seeded at 50,000 cells/µL (or 15,000 cells/µL
to allow cell lines to propagate). Devices were designed such that the entire upper and lower
culture area could be accessed by a 100x objective (Epshteyn et al., 2011), demonstrated with
live cell stains and high resolution imaging (Figure 2.8c,d). Despite these advances, the first
generation device was inadequate for several reasons. First, it became apparent that static culture
was necessary, at least temporarily, for seeding and co-culture seeding steps. Static culture was
found difficult to achieve as fluidic resistance provided by the long (3 cm) channel length
inhibited bulk media flow by differential head, resulting in channel media evaporation. Second,
82
the device length rendered it very fragile and prone to fracture with anything less than delicate
handling. Third, high resolution imaging eventually became required only through the glass
bottom, thus additional steps to produce thin top slabs were unnecessary. Fourth, the device was
assembled with silane chemistry which was excessively lot-specific and possibly cell toxic (Lee
and Ram, 2009).
Device Perfusion, Short and Long-Term. After understanding how to seed hepatocytes
into devices, shortening the device to about 1 cm (the second generation device), and modifying
device construction to feature glued assembly instead of silane, we next attempted continuously
perfused culture at a variety of flow rates. The microcirculatory and submicrocirculatory
systems flow at about 1-4 and 1 dyn/cm2, respectively, and previous studies have shown the liver
sinusoids experience 0.1-0.5 dyn/cm2 of shear, and a shear of 0.33 dyn/cm2 is suitable for
hepatocytes (Tilles et al., 2001). Top channel perfusion at 2.5, 100, and 300 µL/hr was
calculated to induce cell-surface shears of 0.001, 0.01, and 0.2 dyn/cm2, respectively. However,
multiple attempts to culture hepatocyte cell lines in the device lower channel under upper
channel perfusion (at rates listed above) resulted in culture death (Figure 2.9a). Possible
explanations, including oxygen starvation, lack of nutrients or buildup of waste, shear stress, and
autocrine factor removal were considered. Oxygen starvation was not experimentally assessed
given PDMS is extremely oxygen permeable and the device layers are thin (Ochs et al., 2014;
Roy et al., 2001). We first examined the diffusion properties of albumin (a large molecule with a
slow diffusion coefficient of 61 µm2/s in water) through the device membrane by culturing
hepatocytes in the lower channel and measuring output through the upper and lower channel
(Figure 2.9b). About one third of total albumin produced was captured through top channel
sampling, demonstrating that the membrane does increase the lower-to-upper diffusion gradient
83
in the device, but is not a drastic barrier to diffusion as most nutrients and waste metabolites (i.e.
glucose at 600 µm2/s in water) diffuse an order of magnitude faster than albumin.
Early observations of cells in device culture indicated the membrane may not block flow
crossing from one channel to the other, thus the device dimensions and properties (membrane
porosity and flow rates) were modeled in COMSOL (Maher et al. 2014). Models revealed the
membrane did not completely block crossflow and a percentage of flow did travel through the
lower channel during upper channel perfusion, proportional to the cross sectional area of the two
channel (Figure 2.9c,d). We tested two solutions to prevent crossflow: modifying the lower
channel geometry to induce ‘dead zones’ and filling the lower channel with a collagen gel. The
modified geometries did not mitigate crossflow (data not shown). Collagen gel did protect
hepatocytes from pulling off the lower channel glass after two days of shear at 20 µL/hr, but was
insufficient for hepatocyte survival (Figure 2.9e,f); indicating other possible causes of death
should be considered.
We next determine the upper limit of shear tolerated by HC-04 hepatocytes and tested if
autocrine/paracrine factor removal by flow were causes of hepatocyte death. Simple channels
were designed such that hepatocytes could propagate to confluence in a channel, followed by
perfusion at various flow rates, under live-cell video microscopy. Flow rates corresponding to
above 0.003 dyn/cm2 killed the hepatocyte line in less than 12 hrs, demonstrating the upper limit
of shear. This shear level correlated to a flow rate of 2.5 µL/hr in bilayer-device culture. To test
if constantly supplying fresh media has negative effects on hepatocytes, conditioned media
(media harvested from an HC-04 flask after 24 hrs) was perfused at 2.5 µL/hr over hepatocytes
in standard and collagen-embedded culture in the bottom channel of the device; having the effect
of stable, long-term (three week) culture of hepatocytes (Figure 2.10a-c). Difficulties with
84
perfused device culture prompted development of a static culture method by installing media
reservoirs instead of tubing into channel ports. Three-week hepatocyte line culture under static
conditions was also accomplished, but post-experimental Live/Dead staining revealed a dead
zone in the middle of the device; possibly due to diffusion limitations from media reservoirs at
device channel ends to the device center (Figure 2.10d). These results, especially HC-04
intolerance to shear and overestimation of proper channel length, proved critical during design of
the third generation device, described below.
Device Vascularization. Steps to vascularize devices and culture endothelial cells under
flow were taken to provide a more stable and mimetic liver model. HLSEC were thawed and
allowed to grow to confluence before overnight shaking (Kulig et al., unpublished results).
Conditioned endothelial cells were seeded into devices and allowed to completely cover the
vascular upper channel (Figure 2.11). However, multiple attempts to perfuse endothelial cells in
long-term culture were unsuccessful as post experimental Live/Dead staining revealed the cells
had completely pulled off the membrane surface. Vascularization was found important for
initiation of perfused co-culture with hepatocyte cell lines and primary hepatocytes in the lower
channel, as endothelials likely blocked membrane holes and effectively mitigated crossflow
throughout the early stages of experiments.
Third Generation Device Improvements. A third generation device was designed to meet
several requirements gleaned from experimental results from the first and second generations.
First, a protocol to produce PDMS membranes was developed to allow oxygen plasma bonding
of the entire stack assembly. Oxygen plasma bonding is covalent, unlike assembly with glue.
While cells often leaked from channels in the first two generations of device, the PDMS device
channels were found to be sealed. Fortuitously, the PDMS membrane was found to be much
85
more optically transparent than both PET and PC, allowing for visualization of cells in both
channels (Figure 2.12b). PDMS membrane holes were confirmed to be completely punched
through by scanning electron microscopy, confirming optimized fabrication (Figure 2.12e).
Second, the device channels were shortened to run together for about 4 mm, due to findings that
lengths longer than 5 mm resulted in diffusion limitations (Figure 2.1, Figure 2.10d and 2.12a).
Lastly, a comparison between PET and PDMS membrane devices with primary hepatocytes
revealed significantly improved hepatocyte health in the PDMS device, possibly due to the
absence of glue or greater permeability of oxygen through PDMS than PET (Figure 2.12f).
Successful Long-Term Culture of Primary Human Hepatocytes. After identification
of high-quality sources of primary human hepatocytes in Yecuris, Inc. and Bioreclamation IVT,
Inc., and because of increasing evidence that cell lines are not amendable to long-term culture
(due to overgrowth and poor phenotype expression), we focused on optimizing primary
hepatocyte culture in devices. High post-thaw hepatocyte viability allowed for dense seeding in
the device lower channel. Over the first week of culture, hepatocyte retracted from the original
cell monolayer and became taller, as was evident by phase microscopy (Figure 2.13d). After
about two weeks in culture, hepatocytes remodeled into multicellular unit—changes in
morphology correlated well with stabilization of albumin, factor IX, and CYP 3A4 phenotypes
(Figure 2.13a-c). While hepatocyte spreading is associated with loss of cellular architecture, loss
of polarization, and loss of transporter function (Ploss et al., 2010; Turncliff et al., 2006), the
cuboidal morphology noted in device cultures featured proper formation of bile canaliculi
between hepatocytes (Figure 2.13e). Primary hepatocytes also stained negative for alphafetoprotein, a phenotype of fetal and de-differentiated hepatocytes like HepG2 [(Figure 2.13f),
(Nakagawa et al., 2012)]. While co-culture with HLSEC over 1 week did not appear to affect
86
primary hepatocyte albumin production, albumin and CYP 3A4 may have been optimally
expression without culture as expression levels compared well with other culture methods
[(Figure 2.13h), (Bhatia et al., 1998; Kim and Rajagopalan, 2010; Weigand and Alpert, 1981)].
Identification of Cell Compaction as Essential Characteristic. To better understand
what characteristics of the device prompted primary hepatocyte longevity, we generated several
devices with decreasingly fewer device features. In a comparison of wells, bilayer devices,
channel devices, walled devices and patterned hepatocytes in an open well, only systems with
walls (wells, bilayer devices, channels, and walled devices) resulted in stabilized phenotype.
This experiment was repeated between walled and patterned systems, confirming that vertical
geometries confining primary hepatocytes help to maintain their architecture and phenotypes in
long-term culture. By culturing hepatocytes in a microenvironment that prevents them from
spreading, and instead encouraging formation of units several cells across and 50-100 cells long,
it is possible confinement is effectively re-creating the spatial features of a liver acinus. While
the exact mechanism of action of the device requires further investigation, this demonstrates that
the spatial constriction of the channel alone, without perfusion, co-cultures, elaborate
extracellular matrices or media components, is sufficient to prevent hepatocyte spreading. This
understanding, especially its simplicity, is invaluable in the design of both micro-scale liver
models, as well as larger liver-assist type devices (LeCluyse et al., 2001).
Further Optimizations on the Basic Liver Model. To further enhance long-term culture
phenotypes, we tested higher seeding densities, different extracellular matrix (ECM) proteins,
and collagen gel sandwich culture within the device, as these methods have been previously
described as successful (Dunn et al., 1991; Flaim et al., 2005; Lee et al., 2007). While collagen
87
gel sandwich cultures in the device lower channel did quickly form multicellular structures, no
quantitative advantage was noted in albumin production (Figure 2.15a). Similarly, no notable
increase in albumin production result from doubling the initial seed density to 16,000 cells/µL;
and this result was confirmed in well-based culture. As albumin production is a fundamental
process of hepatocytes, we next tested a higher-order process, CYP 3A4 induction, to see if
alternative ECM treatments affected primary phenotypes. Again, several interesting
morphologies resulted, as a mixture of collagen IV and laminin resulted in cord formation, but
no quantitative difference in CYP 3A4 induction developed (Figure 2.15d).
Discussion
Weighing Primary versus Immortalized Cells For Liver Models. Hepatocyte cell
lines are often dismissed as inadequate for a mimetic liver model due to poor bio-functions and
morphology. Previous studies have attempted to address this problem by isolating less dedifferentiated lines such as HepaRG and using complex culture systems to rescue cell line
phenotypes (Narayan et al., 2009; Ohno et al., 2008; Wilkening et al., 2003). Our data confirm
cell lines lack key hepatic phenotypes, such as albumin production and CYP 3A4 induction, but
attempts to rescue phenotypes using gel and endothelial overlays, perfusion, and co-culture
resulted in no significant differentiation. Furthermore, for a long-term liver model, cell line
overgrowth is a significant problem as, over weeks in culture, cells propagated into spheroids,
formed a necrotic core, and made phase microscopy difficult. Aside from initial loss of
differentiation through immortalization and activation of tumor pathways (as is evident by
expression of alpha-fetoprotein), dozens of passages in static, flask-based culture rendered cell
lines almost completely intolerant to shear. Despite their drawbacks, cell lines proved essential
88
in this study for initial prototyping and protocol development experiments. Advantages to using
cell lines for protocol optimization include continuous availability (no need to focus studies
around thaws of single vials), inexpensive and unlimited sourcing of cells, and the ability to
compare historical experiments.
Primary hepatocytes proved essential in meeting our requirement of a mimetic liver
model, but their practicality was dependent on several factors. First, device seeding protocols
were optimized with several round of trial-and-error with cell lines; this process would have been
extremely expensive with primary vials. Second, no two lots of primary hepatocytes attached to
our culture systems with the same efficiency—trying to troubleshoot a sourcing and protocol
issue at the same would lead to negative results with positive factors if all necessary factors were
not used in the same experiment, leading to improper conclusions. And third, while some liver
models require hundreds of thousands of cells per culture system, our device’s efficient surface
area allowed experimentation with multiple factors in dozens of devices per experiment. Thus,
use of both cell lines and primary hepatocytes was essential for this study.
Practicality of Co-cultures for Liver Models. The concept of parenchymal and nonparenchymal cell co-culture is logical as organ function is derived from complex intercellular
communication. Several methods of direct and indirect co-culture have been developed, most
notably the HepatoPacTM and LiverChipTM systems (Chan et al., 2013; Domansky et al., 2010).
However, we found cell co-culture demands much optimization, such that the effort to optimize
is dependent on knowledge of which candidate cell sources are the correct for the model’s
purpose. The issue of universal media is not trivial as cell lines are highly conditioned to the
media they are subcultivated with. Initially we experimented with three hepatocyte cell lines and
three nonparenchymal cell lines, meaning nine different mixed media-conditioning experiments
89
would be necessary to test which co-culture combinations were fruitful. Primary cells are a
solution to overgrowth issues precluding cell lines from long-term culture, but their use is also
not straightforward. Previous studies have shown that multicellular islands of hepatocytes are
required for long-term stable function; yet the technology to pattern collagen is not widespread
and is dependent on a proprietary, contact-inhibited fibroblast line (Bhatia et al., 1998; Bhatia et
al., 1997). Furthermore, a co-culture between two primary cell types is prohibitively expensive.
Despite these drawbacks to co-culture, we attempted to repeat previous studies using co-culture
to rescue cell line phenotypes, and maintain primary hepatocyte and endothelial phenotypes,
without success. Taken together, our findings demonstrate few co-culture permutations are
beneficial, previously described co-culture models are highly optimized and poorly transferrable,
and their complexity dramatically hinders their throughput.
Throughput of Transwells, Devices, Static and Perfused Systems. Throughput is
critical feature of our liver system as our overarching goal is to create a drug screening platform;
the platform needs to be only complex enough to support in vivo-like invasion and development
of Plasmodium parasites. Transwell, static, and perfused devices offer the ability to test threedimensional architectures and ‘four’-dimensional (perfused) effects on host cells and parasites.
However, these features come at dramatic cost to throughput. Transwells were found to be
burdensome to double seed and image and were not examined in greater detail. Furthermore, a
recent publication touting long-term transwell culture showed data with downward trends in
metabolic activity and albumin production—trends not amendable to a mimetic model (Wagner
et al., 2013).
Long-term perfusion of hepatocyte cell lines and primary hepatocytes was accomplished
after much optimization. We found maintaining signaling factor levels was essential for
90
hepatocyte health during perfusion; these autocrine and paracrine factors have been found
essential in other perfused liver models and may necessitate complex, closed loop perfusion
methods (Griffith and Swartz, 2006). Multiplexing perfused systems is complicated with the
need to independently access culture vessels, thus previously described multiplexed culture
systems have only managed 12-24 wells at a time (Domansky et al., 2010; Lee et al., 2007; Toh
et al., 2009). These multiplexed perfused models also lack interfaces to pre-existing drug
dilution plates through 96-well pitch and robotic handling; characteristics essential for drug
screening.
We successfully cultivated a layer of HLSEC’s in the device upper channel, however,
attempts to extend culture time beyond a few days and to flow the monolayer were unsuccessful.
Previous studies have described rapid de-differentiation of liver endothelial cells without the
proper environmental cues in vitro (Sellaro et al., 2007). A revealing study of in vitro
microvasculature demonstrated nonparenchymal paracrine signaling, extracellular matrix
protein, and endothelial-cell dependent tubule formation as essential components to generating
functional, perfused microvessels (Kim et al., 2013). While remarkable, this level of complexity
is not practical for a high throughput liver model, thus continued attempts to vascularize our liver
model were not pursued. However, short term vascularization for parasite invasion was still
useful and feasible in our model.
Dichotomy of Complexity and Scalability. A consistent theme of in vitro liver models
is complexity negates scalability (Epshteyn et al., 2011). Recent advances to produce highthroughput and complex liver models have produced a 24-well and 96-well plate assays for drug
discovery (Dembele et al., 2014; Khetani and Bhatia, 2008). Another interesting avenue is the
development of liver ‘pucks’ which, after production, could be cryostored and shipped to the end
91
user for use in a variety of plate throughputs (Li et al., 2014). However, these models both
require proprietary cell lines and commercial plate sources. The novelty of our work lies in
long-term maintenance of primary hepatocytes using a simple and transferrable concept:
compact microspace. While this phenomenon has been previously described, here we present
using microspace alone in a high-throughput liver model for both basic and Plasmodium research
(Wang et al., 2013).
Future Studies. The liver is responsible for hundreds of biological functions whereas we
only measured albumin, factor IX, CYP 3A4 and bile canaliculi formation. Analysis of more
phenotypes, especially dozens of other Cytochrome P450 and transporter functions, will better
characterize our liver model. As the primary focus of our research is optimization of a
Plasmodium liver stage drug discovery platform, we were unable to look into the underlying
effects of flow and co-culture within our liver model. An investigation of more complex
phenotypes or gene expression profiles may reveal incremental improvements from complexity
which are difficult to see in broad and basic phenotypes.
92
Figure 2.1. Third generation microfluidic device fabrication and properties. a) Summary of
device fabrication steps. b) Schematic of device features and cutaway cross-sectional view. c)
Device setup for perfusion and seated into stage adapter (black). d) High resolution microscopy
is made possible by incorporating coverglass (blue = DNA, red = mitochondria, green = tubulin).
Scale bar, 5 µm.
*Source: Maher et al. 2014, used with permission
93
Figure 2.2. Hepatocyte and endothelial phenotypes. a) HC-04 produce more albumin and
express higher levels of CYP 3A4 than HepG2 and Huh-7, but CYP 3A4 is not inducible for any
of the 3 hepatocyte cell lines. b) SNP chip of HC-04 and HepG2, red arrows depict similar
genotypes, suggesting the two cell lines are actually one divergent cell line. c) von-Willebrand
Factor, Low Density Lipoprotein, ICAM-1 expression and E-Selectin expression of HLSEC and
HAEND endothelial cells; both cells express typical liver sinusoidal markers.
94
Figure 2.3. Cell count optimizations—CellTiter-Glo®, Hoechst, and Trypsin. a) Live/Dead
Stain is an endpoint assay to identify live and dead cells in culture via microscopy. Scale bar, 20
µm. b) Cell counting software identifies and counts nuclei in lower channel of device. c)
CellTiter-Glo® quantification of various cell densities in microplate wells. d) Hoechst staining
and quantification by plate reader is non-endpoint method of cell quantification. e) Schematic of
trypsin-based cell removal and quantification. f) Trypsin removes most cells from channel after
200 µl/hr flow for 1hr. g) Cells can block channels and be lost within devices.
95
Figure 2.4. Cell count optimizations—DNA quantification and PrestoBlue®. a) Schematic of
PrestoBlue® and (b) DNeasy® workflow. c) DNeasy® well standard curve, the device seed value
is at the lower end. d) PrestoBlue® standard curve, the device seed value is more reliably
detected from 96 well plates. e) When PrestoBlue® is run through the top vs. bottom channel (in
duplicate) Resazurin in metabolized at different rates, indicating slow mixing between channels.
f) Device (green dot) and wells (blue dots) demonstrate differential metabolism of Resazurin at
40,000 cells due to absence and presence of convection, making comparisons to plate controls
impractical.
96
Figure 2.5. Generation of co-culture methods. a) 40x image of fibroblasts 3 days after addition
of mitomycin C, resulting in cell death. b) Co-culture initiation by seeding hepatocytes lines,
allowing hepatocyte islands to form (red circles), followed by addition of fibroblasts between
islands. c) Alternative seeding strategy where stellate cells are seeded first (right), followed by
hepatocytes (left) to achieve random co-culture. d) Co-culture of primary hepatocytes (which do
not propagate, arrows) with fibroblasts (left).
97
Figure 2.6. Generation of complex co-culture methods. a) Schematic of collagen droplet
collagen-patterning method. b) Anti-collagen I staining with labeled antibody demonstrated
patterned adsorption. c) Patterned co-culture of HepG2 cells (center) and fibroblasts (outside). d)
Patterned seeding of primary hepatocytes to collagen by washing prior to full attachment. e)
Schematic of transwell based culture of endothelial cells and HC-04 cells in a 24 well plate. f)
Projection of Z-stack imaging through transwell co-culture with cells stained with CellTrackerTM
red or green. Scale bars, 100 µm.
98
Figure 2.7. Co-culture effect on hepatocyte phenotype. a) 26hour albumin production of mono
and co-cultures of HepG2 cells and fibroblasts, at a 3k:12k seed ratio. b) Utilization of Hoechst
quantification to standardize data from (a) by cell number, revealing mixed media and co-culture
toxicity issues. c) Primary hepatocytes in random co-culture with stellate cells show loss of
phenotype. d) HC-04 in co-culture with stellate cells does not rescue cell line phenotype. Bars
represent standard deviation.
99
Figure 2.8. First generation device seeding and high resolution microscopy. a) Overnight
coating with collagen and confirmation by IFA. b) Device lower channel seeded with HC-04 at
50,000 cells/µL. c) To demonstrate device imaging capabilities, primary hepatocytes and HepG2
were seeded into either channel, stained, and imaged with an 100x objective. Scale bars, 5 µm.
d) 3D-rendering of a Z-stack through the device membrane with HC-04 on both sides (left), 3Drendering of a nucleus and surrounding mitochondria of an HC-04 cell in the bottom channel
(right).
*(c) Source: Epshteyn et al., 2011, used with permission
100
Figure 2.9. Continuous perfusion in device troubleshooting. a) Preliminary experiments
resulted in rapid loss of hepatocyte cell line phenotype across several attempts and flow rates.
b) Albumin measurements from top or bottom channel sampling, indicating modest diffusion
limitations, especially at higher flow rates. c) Modeling of flow through device revealed
crossflow from the upper to lower channel. d) Modeling of different channel dimensions
revealed crossflow is dependent on channel aspect ratios; the device used in this study utilized an
aspect ratio of 4. e) Devices lower channel designs to mitigate crossflow. f) Testing collagen
matrix to protect cells from crossflow by blocking free space in lower channel; devices were
flowed at 20 µL/hr. Flow was blocked but cells did not survive. Scale bars, 60 µm.
*(c,d) Source: Maher et al., 2014, used with permission
101
Figure 2.10. Full device optimization for hepatocyte line perfused and static culture. a-e)
Panels across devices showing post-experimental Live/Dead staining. a) Perfusion of
conditioned media promotes HC-04 survival over 6 days in standard and (b) collagen-embedded
culture. (a,b) Scale bars, 60 µm. c) HC-04 after 6 days of low (2.5 µL/hr) flow of conditioned
media. d) HC-04 after 21 days of static culture, including a dead zone in the device center. e)
HC-04 after 21 days of static culture in the shortened, third generation device. f) 40x image of
(e) after 21 days of culture, forming a cord of hepatocytes due to overgrowth. g) HC-04 after 21
days of perfused culture featuring a low flow rate, conditioned media, and a vascularized top
channel. (c-g) Scale bars, 100 µm.
*(d,e,g) Source: Maher et al., 2014, used with permission
102
Figure
2.11. Device vascularization. a) HLSEC in confluent monolayer before and (b) after overnight
orbital shaking at 1000 rpm, conditions endothelials for shear prior to device seed. b) Z-stack
through a double-seeded device with HC-04 in the lower channel and HLSEC in the upper
channel. c) 3D rendering of (b). d) HLSEC morphology in a simple channel at (l to r) 1hr, 24hr,
48hr. Scale bars, 100 µm.
*(b,c) Source: Maher et al., 2014, used with permission
103
Figure 2.12. Device dimension and membrane optimization. a) Endothelial cells cultured in
the center of the device upper channel demonstrate a dead zone in the second generation device
during static culture, possibly due to diffusion limitations to reservoirs (green half circles, green
= live, red = dead). b) Optical properties of (l to r) PDMS, PET and PC membranes. White
arrows indicate cells leaking out of glued PET and PC membrane devices. c) Scanning electron
microscopy of custom-manufactured PDMS membrane, demonstrating thinness (10 µm) and
through-punched holes. f) Long-term albumin production by primary hepatocytes in PDMS
devices outperforms PET devices. Error bars represent standard deviation.
*Source: Maher et al., 2014, used with permission
104
Figure 2.13. Long-term culture of primary human hepatocytes. a) Stable long-term (three
week) expression of albumin and (b) factor IX. c) Stable two week expression of CYP 3A4
induction. d) Primary hepatocyte culture at 5 hrs, 13 days, and 21 days. Hepatocytes form
multicellular structures (dotted lines) and bile networks (e). Scale bar, 50 µm. f) Primary
hepatocytes stain negative for alpha-fetoprotein after three weeks in device culture, while HepG2
are positive for de-differentiation (g). Scale bars, 20 µm. h) Comparison of device (‘Yecuris’)
hepatocyte albumin production in co-culture, and a comparison to previous studies. All error
bars represent standard deviation.
*(a-e) Source: Maher et al., 2014, used with permission
105
Figure 2.14. Confinement leads to stable primary hepatocyte phenotype. a) A screen of
several device geometries identified a difference in albumin production between walled and open
devices, but not between culture systems with confinement. b) A repeat of experiment from a)
with more replicates. Confined and patterned primary hepatocyte cultures were established.
Patterned hepatocytes spread beyond their original seeding area (d) and subsequently loss
primary phenotypes (c). Error bars represent standard deviation. Scale bars, 100 µm.
*(b-d) Source: Maher et al., 2014, used with permission
106
Figure 2.15. Further device optimization. a) Collagen gel and higher seed densities of primary
hepatocytes has no effect on device or (b) well culture. c) Collagen gel does result in rapid
formation of cuboidal hepatocyte in a sandwich configuration. Scale bars, 100 µm. d) A screen
for advantageous ECM protein layers yielded no significant results after one week in culture.
Error bars represent standard deviation. Scale bars, 60 µm.
*(b) Source: Maher et al., 2014, used with permission
107
References
Bhatia, S.N., Balis, U.J., Yarmush, M.L., and Toner, M. (1998). Probing heterotypic cell
interactions: hepatocyte function in microfabricated co-cultures. J Biomater Sci Polym
Ed 9, 1137-1160.
Bhatia, S.N., Yarmush, M.L., and Toner, M. (1997). Controlling cell interactions by
micropatterning in co-cultures: hepatocytes and 3T3 fibroblasts. J Biomed Mater Res 34,
189-199.
Chan, T.S., Yu, H., Moore, A., Khetani, S.R., and Tweedie, D. (2013). Meeting the challenge of
predicting hepatic clearance of compounds slowly metabolized by cytochrome P450
using a novel hepatocyte model, HepatoPac. Drug Metab Dispos 41, 2024-2032.
Chao, P., Maguire, T., Novik, E., Cheng, K.C., and Yarmush, M.L. (2009). Evaluation of a
microfluidic based cell culture platform with primary human hepatocytes for the
prediction of hepatic clearance in human. Biochem Pharmacol 78, 625-632.
Chen, T., Kunnavatana, S.S., and Koch, R.J. (2006). Effects of mitomycin-C on normal dermal
fibroblasts. Laryngoscope 116, 514-517.
Coppi, A., Tewari, R., Bishop, J.R., Bennett, B.L., Lawrence, R., Esko, J.D., Billker, O., and
Sinnis, P. (2007). Heparan sulfate proteoglycans provide a signal to Plasmodium
sporozoites to stop migrating and productively invade host cells. Cell Host Microbe 2,
316-327.
Daneker, G.W., Lund, S.A., Caughman, S.W., Swerlick, R.A., Fischer, A.H., Staley, C.A., and
Ades, E.W. (1998). Culture and characterization of sinusoidal endothelial cells isolated
from human liver. In Vitro Cell Dev Biol Anim 34, 370-377.
Dembele, L., Franetich, J.F., Lorthiois, A., Gego, A., Zeeman, A.M., Kocken, C.H., Le Grand,
R., Dereuddre-Bosquet, N., van Gemert, G.J., Sauerwein, R., et al. (2014). Persistence
and activation of malaria hypnozoites in long-term primary hepatocyte cultures. Nat Med
20, 307-312.
Domansky, K., Inman, W., Serdy, J., Dash, A., Lim, M.H., and Griffith, L.G. (2010). Perfused
multiwell plate for 3D liver tissue engineering. Lab Chip 10, 51-58.
Donato, M.T., Lahoz, A., Castell, J.V., and Gomez-Lechon, M.J. (2008). Cell lines: a tool for in
vitro drug metabolism studies. Curr Drug Metab 9, 1-11.
Dunn, J.C., Tompkins, R.G., and Yarmush, M.L. (1991). Long-term in vitro function of adult
hepatocytes in a collagen sandwich configuration. Biotechnol Prog 7, 237-245.
108
Epshteyn, A.A., Maher, S., Taylor, A.J., Holton, A.B., Borenstein, J.T., and Cuiffi, J.D. (2011).
Membrane-integrated microfluidic device for high-resolution live cell imaging.
Biomicrofluidics 5, 46501-465016.
Flaim, C.J., Chien, S., and Bhatia, S.N. (2005). An extracellular matrix microarray for probing
cellular differentiation. Nat Methods 2, 119-125.
Frevert, U., Engelmann, S., Zougbede, S., Stange, J., Ng, B., Matuschewski, K., Liebes, L., and
Yee, H. (2005). Intravital observation of Plasmodium berghei sporozoite infection of the
liver. PLoS Biol 3, e192.
Griffith, L.G., and Swartz, M.A. (2006). Capturing complex 3D tissue physiology in vitro. Nat
Rev Mol Cell Biol 7, 211-224.
Harker, K.S., Jivan, E., McWhorter, F.Y., Liu, W.F., and Lodoen, M.B. (2014). Shear forces
enhance Toxoplasma gondii tachyzoite motility on vascular endothelium. MBio 5,
e01111-01113.
Hoover, M.L., Vetvicka, V., Hoffpauir, J.M., and Tamburro, C.H. (1993). Human endothelial
cell line from an angiosarcoma. In Vitro Cell Dev Biol 29A, 199-202.
Khetani, S.R., and Bhatia, S.N. (2008). Microscale culture of human liver cells for drug
development. Nat Biotechnol 26, 120-126.
Kim, S., Lee, H., Chung, M., and Jeon, N.L. (2013). Engineering of functional, perfusable 3D
microvascular networks on a chip. Lab Chip 13, 1489-1500.
Kim, Y., and Rajagopalan, P. (2010). 3D hepatic cultures simultaneously maintain primary
hepatocyte and liver sinusoidal endothelial cell phenotypes. PLoS One 5, e15456.
Knowles, B.B., Howe, C.C., and Aden, D.P. (1980). Human hepatocellular carcinoma cell lines
secrete the major plasma proteins and hepatitis B surface antigen. Science 209, 497-499.
LeCluyse, E.L. (2001). Human hepatocyte culture systems for the in vitro evaluation of
cytochrome P450 expression and regulation. Eur J Pharm Sci 13, 343-368.
Lee, K.S., and Ram, R.J. (2009). Plastic-PDMS bonding for high pressure hydrolytically stable
active microfluidics. Lab Chip 9, 1618-1624.
Lee, P.J., Hung, P.J., and Lee, L.P. (2007). An artificial liver sinusoid with a microfluidic
endothelial-like barrier for primary hepatocyte culture. Biotechnol Bioeng 97, 1340-1346.
Li, C.Y., Stevens, K.R., Schwartz, R.E., Alejandro, B.S., Huang, J.H., and Bhatia, S.N. (2014).
Micropatterned Cell-Cell Interactions Enable Functional Encapsulation of Primary
Hepatocytes in Hydrogel Microtissues. Tissue Eng Part A.
109
Maher, S.P., Crouse, R.B., Conway, A.J., Bannister, E.C., Achyuta, A.K., Clark, A.Y., Sinatra,
F.L., Cuiffi, J.D., Adams, J.H., Kyle, D.E., and Saadi, W.M. (2014). Microphysical space
of a liver sinusoid device enables simplified long-term maintenance of chimeric mouseexpanded human hepatocytes. Biomed Microdevices.
March, S., Ng, S., Velmurugan, S., Galstian, A., Shan, J., Logan, D.J., Carpenter, A.E., Thomas,
D., Sim, B.K., Mota, M.M., et al. (2013). A microscale human liver platform that
supports the hepatic stages of Plasmodium falciparum and vivax. Cell Host Microbe 14,
104-115.
Mazier, D., Beaudoin, R.L., Mellouk, S., Druilhe, P., Texier, B., Trosper, J., Miltgen, F., Landau,
I., Paul, C., Brandicourt, O., and et al. (1985). Complete development of hepatic stages of
Plasmodium falciparum in vitro. Science 227, 440-442.
Menard, R., Sultan, A.A., Cortes, C., Altszuler, R., van Dijk, M.R., Janse, C.J., Waters, A.P.,
Nussenzweig, R.S., and Nussenzweig, V. (1997). Circumsporozoite protein is required
for development of malaria sporozoites in mosquitoes. Nature 385, 336-340.
Nakagawa, T., Moriwaki, K., Terao, N., Miyamoto, Y., Kamada, Y., and Miyoshi, E. (2012).
Analysis of polarized secretion of fucosylated alpha-fetoprotein in HepG2 cells. J
Proteome Res 11, 2798-2806.
Narayan, R., Gangadharan, B., Hantz, O., Antrobus, R., Garcia, A., Dwek, R.A., and Zitzmann,
N. (2009). Proteomic analysis of HepaRG cells: a novel cell line that supports hepatitis B
virus infection. J Proteome Res 8, 118-122.
Ochs, C.J., Kasuya, J., Pavesi, A., and Kamm, R.D. (2014). Oxygen levels in thermoplastic
microfluidic devices during cell culture. Lab Chip 14, 459-462.
Ohno, M., Motojima, K., Okano, T., and Taniguchi, A. (2008). Up-regulation of drugmetabolizing enzyme genes in layered co-culture of a human liver cell line and
endothelial cells. Tissue Eng Part A 14, 1861-1869.
Ploss, A., Khetani, S.R., Jones, C.T., Syder, A.J., Trehan, K., Gaysinskaya, V.A., Mu, K., Ritola,
K., Rice, C.M., and Bhatia, S.N. (2010). Persistent hepatitis C virus infection in
microscale primary human hepatocyte cultures. Proc Natl Acad Sci U S A 107, 31413145.
Pugh, J.C., Yaginuma, K., Koike, K., and Summers, J. (1988). Duck hepatitis B virus (DHBV)
particles produced by transient expression of DHBV DNA in a human hepatoma cell line
are infectious in vitro. J Virol 62, 3513-3516.
Rodriguez-Antona, C., Donato, M.T., Pareja, E., Gomez-Lechon, M.J., and Castell, J.V. (2001).
Cytochrome P-450 mRNA expression in human liver and its relationship with enzyme
activity. Arch Biochem Biophys 393, 308-315.
110
Rojkind, M., Novikoff, P.M., Greenwel, P., Rubin, J., Rojas-Valencia, L., de Carvalho, A.C.,
Stockert, R., Spray, D., Hertzberg, E.L., and Wolkoff, A.W. (1995). Characterization and
functional studies on rat liver fat-storing cell line and freshly isolated hepatocyte
coculture system. Am J Pathol 146, 1508-1520.
Roy, P., Baskaran, H., Tilles, A.W., Yarmush, M.L., and Toner, M. (2001). Analysis of oxygen
transport to hepatocytes in a flat-plate microchannel bioreactor. Ann Biomed Eng 29,
947-955.
Sattabongkot, J., Yimamnuaychoke, N., Leelaudomlipi, S., Rasameesoraj, M., Jenwithisuk, R.,
Coleman, R.E., Udomsangpetch, R., Cui, L., and Brewer, T.G. (2006). Establishment of a
human hepatocyte line that supports in vitro development of the exo-erythrocytic stages
of the malaria parasites Plasmodium falciparum and P. vivax. Am J Trop Med Hyg 74,
708-715.
Sellaro, T.L., Ravindra, A.K., Stolz, D.B., and Badylak, S.F. (2007). Maintenance of hepatic
sinusoidal endothelial cell phenotype in vitro using organ-specific extracellular matrix
scaffolds. Tissue Eng 13, 2301-2310.
Tilles, A.W., Baskaran, H., Roy, P., Yarmush, M.L., and Toner, M. (2001). Effects of
oxygenation and flow on the viability and function of rat hepatocytes cocultured in a
microchannel flat-plate bioreactor. Biotechnol Bioeng 73, 379-389.
Toh, Y.C., Lim, T.C., Tai, D., Xiao, G., van Noort, D., and Yu, H. (2009). A microfluidic 3D
hepatocyte chip for drug toxicity testing. Lab Chip 9, 2026-2035.
Turncliff, R.Z., Tian, X., and Brouwer, K.L. (2006). Effect of culture conditions on the
expression and function of Bsep, Mrp2, and Mdr1a/b in sandwich-cultured rat
hepatocytes. Biochem Pharmacol 71, 1520-1529.
Wagner, I., Materne, E.M., Brincker, S., Sussbier, U., Fradrich, C., Busek, M., Sonntag, F.,
Sakharov, D.A., Trushkin, E.V., Tonevitsky, A.G., et al. (2013). A dynamic multi-organchip for long-term cultivation and substance testing proven by 3D human liver and skin
tissue co-culture. Lab Chip 13, 3538-3547.
Wang, Y., Toh, Y.C., Li, Q., Nugraha, B., Zheng, B., Lu, T.B., Gao, Y., Ng, M.M., and Yu, H.
(2013). Mechanical compaction directly modulates the dynamics of bile canaliculi
formation. Integr Biol (Camb) 5, 390-401.
Weigand, K., and Alpert, E. (1981). Human albumin synthesis via an albumin precursor in liver
tissue slices. Experientia 37, 1145-1147.
Wilkening, S., Stahl, F., and Bader, A. (2003). Comparison of primary human hepatocytes and
hepatoma cell line Hepg2 with regard to their biotransformation properties. Drug Metab
Dispos 31, 1035-1042.
111
Xu, L., Hui, A.Y., Albanis, E., Arthur, M.J., O'Byrne, S.M., Blaner, W.S., Mukherjee, P.,
Friedman, S.L., and Eng, F.J. (2005). Human hepatic stellate cell lines, LX-1 and LX-2:
new tools for analysis of hepatic fibrosis. Gut 54, 142-151.
Zamek-Gliszczynski, M.J., Xiong, H., Patel, N.J., Turncliff, R.Z., Pollack, G.M., and Brouwer,
K.L. (2003). Pharmacokinetics of 5 (and 6)-carboxy-2',7'-dichlorofluorescein and its
diacetate promoiety in the liver. J Pharmacol Exp Ther 304, 801-809.
112
Chapter Three:
Identification of Critical Features Necessary for Plasmodium Liver Stage Culture
Rationale of Study
The two principal components of a Plasmodium liver model are invasive sporozoites and
stable hepatocytes. In Chapter 2, we described the development of a mimetic human liver model
and maintenance of primary hepatocyte phenotypes using compaction. To infect the model, we
needed to refine Plasmodium falciparum gametocyte culture protocols as, at the onset of the
study, sporozoite production was highly irregular. We screened several parasite lines for
production of viable sporozoites, and identified several critical steps responsible for production
of viable sporozoites, to be integrated into our workflow. Sourcing of P. vivax sporozoite was
accomplished through a complicated mosquito-shipping collaboration with Mahidol University
in Bangkok, Thailand and by liver assays performed at Mahidol University.
As no self-reporting, liver-capable P. vivax parasite lines exist, and a self-reporting, livercapable P. falciparum line has only been recently published (Talman et al., 2010; Vaughan et al.,
2012a), we needed to identify immunofluorescent assay (IFA) reagents and protocols for
reproducible liver stage identification. Additionally, preliminary studies in our lab found
primary hepatocyte susceptibility to Plasmodium infection to be rare, thus we screened several
donor lots for susceptibility. By securing sporozoites, identifying susceptible primary
hepatocytes, and optimizing protocols to detect liver forms in device culture, we obtained a basic
liver model to compare further optimizations against.
113
Several reports have described the sporozoite invasion pathway in the liver featuring
critical steps such as sporozoite surface protein modification and organelle ejection, reviewed in
Lindner et al. (2012). These steps occur as the parasite come into contact with nonparenchymal
liver fractions, such as endothelial cells, Küpffer cells, and matrix proteins. We hypothesized that
if we could recapitulate the in vivo invasion pathway with the device, we could drastically
improve historically poor invasion and development rates. We set about identifying
nonparenchymal cells responsible for proper sporozoite invasion by utilizing device flexibility to
seed and infect several different co-cultures. We also proposed and confirmed the small number
of hepatocytes and microphysical space of the device creates a highly efficient mixing
environment during sporozoite invasion, leading to more invasion events per hepatocytes than
previous models. Lastly, as this liver model is intended for drug efficacy studies against P.
falciparum and P. vivax liver forms, including hypnozoites, we searched for and found probable
hypnozoites in device culture.
Materials and Methods
Propagation, Isolation and Cryopreservation of Plasmodium Sporozoites. Asexual
blood stage culture of P. falciparum line KF7 was performed with modification to previously
published protocols (Trager, 1971, 1977). Custom made RPMI containing L-Glutamine, 25 mM
HEPES, and 50 mg/L hypoxanthine (KD Medical #CUS-0645, Columbia, MD, USA) was
supplemented with 10% pooled human AB sera (Interstate Blood Bank, Memphis, TN, USA)
quality controlled by passage though a 0.2 µm membrane filter, and 0.1875% Sodium
Bicarbonate (Mediatech/Corning, Manassas, VA, USA). Cultures were gassed in closed-top
flasks with 5% O2 / 5% CO2 / balance N2 (Airgas, Tampa, FL, USA) and never gassed in a selfgassing incubator with vented cap flasks. Cultures were maintained in 5% washed O+ human
114
erythrocytes (Interstate Blood Bank) and maintained at 1-10% parasitemia with daily media
change. Sexual blood stage culture was initiated with 5-10% parasitemia stock cultures,
confirmed by counting parasites in at least 1000 erythrocytes in Giemsa thin blood smears, by
adding V stock to 2 mL 50% washed erythrocytes and 15 mL media to each of two T75 flasks
(G1A and B), where
V = ((25 mL) (0.5%)/parasitemia%).
Stock cultures were then cut to exactly 1% parasitemia by adding V stock to 20 mL media and 2
mL 50% erythrocytes, where
V= ((20 mL)(1%)/parasitemia%).
Three days after initiation of G1A and B, G2A and B were initiated by repeating the procedure
for G1, and G1 flask volume was increased to 25 mL. When performed on Mondays and
Fridays, this cycling generated a stable production line of asexual cultures with key steps on
weekdays. Continued workflow and mixing of asexual cultures is described in Figure 3.1.
After 15 days of sexual culture progression, exflagellation slides of G1 were prepared
daily by pelleting 200 µL of culture for 1 min at 2000 RCF in a microcentrifuge, mixing 10 µL
pellet with 10 µL sera, and waiting 10-15 min under microscopic observation. Exflagellation
events per field of view (FOV) were tracked to identify peak exflagellation of G1. Typically
around day 18, G1 exflagellation events were decreasing, indicating formation of female
gametocytes and fully mature culture. Mosquito feeding was performed by first heating a 50 mL
conical tube, pooled sera, washed erythrocytes, and Hemotek feeder discs (Discovery
Workshops, Lancashire, UK) to 37 ºC. G1 and G2 cultures were combined in the warm tube and
centrifuged for 2 min at 8000 RCF with half brake. After removal of supernatant (without
disturbing the black gametocytes layer above erythrocytes), 4 mL of a 3:1 mixture of warm sera
115
and packed erythrocytes was added per 2.5 mL gametocyte culture pellet, gently mixed by
pipette, and loaded into Hemotek discs. Discs were then attached to a Hemotek heater and fed to
cartons of 3 hr sugar-starved, 3-5 day old Anopheles stephensi or A. freeborni mosquitoes for 15
min. Mosquitoes were monitored for oocyst formation by midgut dissection 7 days post feed,
and for sporozoite packing in salivary glands at 18 days post feed.
Mosquitoes confirmed positive for P. falciparum sporozoites, or P. vivax–infected- A.
dirus mosquitoes fed on Thai patient blood and shipped from Dr. Jetsumon Sattabongkot’s lab at
Mahidol University, were prepared for aseptic salivary gland dissection by dipping in 70%
EtOH, then 100x Pen Strep (Gibco/Life Technologies, Grand Island, NY, USA) in PBS, and then
in 100x Fungizone® (Life Technologies) in PBS. Methods of isolating salivary glands by needle
dissection were optimized to omit collection mosquito debris, essential for controlling mosquito
flora overgrowth in subsequent hepatocyte cultures. Isolated salivary glands were collected in 50
µL RPMI and dissociated with 100 movements of a 100 µL pipette. Sporozoites were quantified
by hemocytometer, assessed for gliding motility, and either used for liver stage assays or
cryopreserved using published protocols (Singh et al., in preparation).
Sporozoite Quality Control. After sporozoite dissection or thaw, 20,000 sporozoites
were applied to a glass coverslip coated with 10 µg/mL mouse anti-CSP 2A10 antibody, and
allowed to glide for 30 min at 37 ºC (Nardin et al., 1982). Coverslips were then fixed with 4%
paraformaldehyde for 20 min, blocked with 1% BSA in PBS overnight at 4 ºC, stained with 10
µg/mL of mouse anti-CSP 2A10 for 1hr at RT, washed, and stained with 1:1000 goat anti-mouse
Alexa Fluor® 488 (Life Technologies) for 1 hr at RT. Stained coverslips were washed, mounted
to a slide, and gliding trails longer >3 µm quantified by microscopy. To quantify traversal, HC04 hepatocytes cultured in 8 chamber slides were stained with 1:1000 CellTrackerTM Green,
116
washed, infected with 24,000 freshly dissected P. falciparum sporozoites, and traversal events
captured using live-cell time-lapse imaging at 2 sec centers (Deltavision® Core, GE Healthcare
BioSciences, Pittsburgh, PA, USA). To quantify invasion, two tubes containing 100,000 freshly
dissected sporozoites were stained with 1:1000 CellTrackerTM Green for 15 min at 37 ºC,
sporozoites in one tube were inactivated by 65 ºC heat shock for 15 min, and stained sporozoites
allowed to invade HC-04 hepatocytes in 8 chamber slides. Four hours after invasion, HC-04
were counterstained with 1:5000 Ethidium Bromide and trypsinized following a modified
version of a previously published liquid IFA protocol (Kaushansky et al., 2012).
Coordinating Device Fabrication, Device Setup, and Sporozoite Production. Fresh
sporozoites develop at about eight-fold higher rates compared to cryopreserved sporozoites
(March et al., 2013). However, use of fresh sporozoites requires a tightly regulated and wellcommunicated schedule of events. Liver assay reagents require 36 days to become available and
the assay itself runs for up to 21 days, thus the entire assay is 57 days long (Figure 3.2).
Successful use of fresh shipments of P. vivax-infected mosquitoes from Thailand required
immediate notification of positive feeds in order to prepare reagents in a timely manner.
Device-Based Sporozoite Invasion and Development Assay. Three types of seeding
conditions were utilized to identify critical factors of sporozoite invasion and development.
Indirect device co-culture was initiated by seeding endothelial cells into the upper channel and
primary hepatocytes or HC-04 cells into the bottom channel, while direct co-culture was initiated
by seeding Küpffer cells on top of primary hepatocytes in the device lower channel (see Chapter
2 for protocols). To avoid negative, and possibly confounding, effects of unoptimized co-culture,
a series of experiments were also run whereby sporozoites were incubated with non-parenchymal
117
cells in an 8 chamber slide for 30 min at 37 ºC prior to being recollected and injected into devicebased hepatocyte monocultures.
Device-based infections were performed by isolating 1,000-50,000 sporozoites per device
(by fresh harvesting from mosquitoes or cryovial thaw) in 50 µL media per device, most often
using an inoculum of 10,000 sporozoites in 50 µL per device. The 50 µL solution was pipetted
into either upper channel, lower channel, or both inlet reservoirs and initiating temporary flow by
removing 50 µL media from the channel outlet reservoirs. Four hours after infection and every
day thereafter, media was changed to control for contamination introduced with sporozoites.
Devices were checked daily by microscopy for growth of bacteria and fungus.
Liver Stage Immunofluorescent Assay. Both direct IFA and indirect IFA were used to
image parasites. Device-based IFA was optimized by pipetting reagents into inlet reservoirs and
inducing flow by aspirating reagents from outlet reservoirs. Devices were fixed by washing
media with 1x PBS and replacing the device volume with 4% paraformaldehyde for 10 min at
RT, followed by washing with PBS. Devices were then permeabilized with 0.1% Triton X 100
for 3 min at RT, washed with PBS, and blocked with 3% BSA for 30 min at 37 ºC. For indirect
P. falciparum assays, liver parasites were stained with 1:1000 rat anti-PfCSP (JA9) and 1:1000
rabbit anti-GRPBiP [(Roobsoong et al., 2014), (JA18)], and 1:1000 mouse anti-HSP70
(courtesy of Dr. Stefan Kappe’s lab) overnight in blocking buffer. Following wash, devices
were counterstained with 1:1000 Hoechst, 1:1000 goat anti-rabbit Alexa Fluor® 488 (Life
Technologies), 1:1000 goat anti-mouse Alexa Fluor® 568 (Life Technologies), and 1:1000
chicken anti-rat Alexa Fluor® 647 (Life Technologies). Indirect P. vivax assays were stained
similarly, but replacing the anti-CSP JA9 antibody with 1:1000 mouse anti PvCSP210 or 247
(Wirtz et al., 1990), omitting GRPBiP and HSP70 stains, and staining with 1:1000 rabbit anti118
UIS4 [(Mueller et al., 2005), (courtesy of Dr. Stefan Kappe lab)]. Direct P. falciparum IFA was
made possible by direct conjugation of PfCSP (JA9) with Alexa Fluor® 647 and GRPBiP with
Alexa Fluor® 488 with conjugation kits, per manufacturer’s protocols (Life Technologies).
Liver Parasite Qualitative and Quantitative Assessment. For invasion assays, invaded,
fixed, and stained sporozoites were quantified by imaging of the entire device lower and upper
channels, resulting in 12 images per channel using a 16x objective on a Deltavision® Core.
Nonparenchymal and hepatocyte nuclei were identified by positive Hoechst staining while
sporozoites were identified by CSP staining. Counts were performed by image intensity
thresholding using a Softworx® image processing software [(Figure3.3a,b), (GE Healthcare
Biosciences)]. For development assays, device lower channels were scanned at 100x. Stained
liver forms were assessed for shape (sporozoite shaped versus round), size (diameter in µm) and
absence or presence of all stained colors (4 colors for P. falciparum and 3 colors for P. vivax).
Of particular note, four populations were persistently found in cultures: 3 and 4 color
sporozoites-shaped forms, and 3 and 4 color round forms (Figure 3.5c-f). A few P. vivax liver
forms were identified to have a single protrusion in the PVM (stained for with UIS),
characteristic of hypnozoites (personal communication, unpublished data).
Results
Reproduction of Standard P. berghei and P. falciparum Liver Stage Assays.
Published P. berghei and P. falciparum liver stage protocols were reproduced in our lab to gather
experience working with mosquitoes, salivary glands, sporozoites, and hepatic cultures.
Furthermore, preliminary experiments were performed to check for co-culture effects on P.
berghei development and to confirm liver stage culture was viable in microfluidic devices. We
first cultured HepG2 cells alone and in co-culture with LX1 stellate cells and fibroblasts before
119
infection with P. berghei-ANKA GFP sporozoites. Wide field microscopy revealed extensive
invasion and development per unit area (Figure 3.4a), but non-synchronous completion of
development as over 33% of P. berghei liver forms failed to complete development after 96 hrs
(Figure 3.4b). Synchronous development was not rescued by co-culture with stellate cells or
fibroblasts (Figure 3.4e). Over 18 experiments, the average form-to-sporozoite invasion rate was
0.39% and strongly correlated with sporozoite inoculum. The average form-to-sporozoite
invasion rate of P. falciparum into HC-04 was 0.47%--on par with published results for invasion
and 10-fold higher for development (Sattabongkot et al., 2006). Likewise, P. falciparum invasion
rate strongly correlated with sporozoite inoculum (Figure 3.4f). These are important as
understanding that, with confluent cultures, invasion is dependent on sporozoite number helps to
establish proper Multiplicity of Infection (MOI) in devices. We also took advantage of P.
berghei liver stage culture to confirm sporozoites could pass through an 8-10 µm diameter pore
by infecting HepG2 through a transwell membrane (Figure 3.4c). Perfused devices were also
confirmed to support HepG2 infection with P. berghei GFP sporozoites through the device
membrane, under constant perfusion, followed by three days of development (Figure 3.4g-i).
These studies were essential in formation of protocols for human Plasmodium liver stage device
culture.
Sporozoite Quality Control Assays. As gliding motility is necessary but not
sufficient for liver stage development, and only loosely correlates to sporozoite viability for
assessing cryopreservation and liver stage development quality control, protocols to measure
sporozoite traversal and invasion were also developed. Sporozoites maintain two methods of cell
intrusion, defined by two complete sets of invasion machinery. ‘Traversal’ involves perforin-like
disruption of a cell membrane followed by complete travel through a cell, while ‘invasion’
120
results in encapsulation in a parasitophorous vacuole membrane (PVM) derived from the host
cell membrane. Traversal mechanisms feature parasite proteins SPECT1/2 and CelTOS and are
used to cross the liver endothelium, escape from Küpffer cell engulfment, and/or travel through
hepatocytes (Ishino et al., 2005; Ishino et al., 2004; Kariu et al., 2006; Mota and Rodriguez,
2001). Traversal primes sporozoite invasion machinery, featuring microneme secretion,
proteolytic cleavage of CSP, and TRAP (Coppi et al., 2011; Coppi et al., 2005; Mota et al.,
2002). Initial studies demonstrated successful, quantifiable traversal, but the assay was deemed
too cumbersome and irreproducible to be regularly performed (Figure 3.5). Invasion rates were
also qualified for P. falciparum-HC-04 invasion and liquid IFA, but invasion rates derived from
flow and microscopy did not compare well. Flow-measured invasion rates were an order of
magnitude higher than expected possibly due to leaching of stain from sporozoites during
traversal and externally-sticking, uninvaded sporozoites appearing as false positives in flow
analysis (Figure 3.6b,c). Consequently, gliding was used to quality-control most liver stage
assays (Figure 3.6a).
Protocol Development for Parasite Imaging. Validation of an IFA protocol for
Plasmodium liver stage detection is essential for protocol optimization, as such validation is the
only method to ensure against false negative results. Many antibodies currently exist for human
Plasmodium liver stages, but their specificity for P. falciparum and/or P. vivax is often
uncharacterized. Furthermore, intricate details, such as fixation with methanol or
paraformaldehyde, can be the difference between positive and negative results as target moieties
can be opened or destroyed by specific fixatives. Running repetitive IFA protocols can be time
consuming; thus we conjugated PfCSP and GRP-BiP to fluorophores to eliminate the need for
extra wash and stain steps; and to protect against background staining common with secondary
121
antibodies used in indirect IFA. We successfully identified reagents for reproducible P.
falciparum and P. vivax liver stage assays (Figure 3.7).
Staining for two or three parasite proteins, along with parasite DNA, dramatically
decreases the chance of detecting false positive artifacts in cultures compared to staining with
only one antibody (Dembele et al., 2014; March et al., 2013). This, in combination with high
resolution microscopy, enables detection of cytoplasmic and membrane bound parasite proteins,
resulting in more information gleaned from IFA than from other methods of detection, such as
RT-PCR (Zou et al., 2013). Indeed, morphology gathered from P. vivax liver-PVM staining and
high resolution, deconvolution microscopy reveals a slight protrusion as a possible biomarker for
hypnozoites; practically no hypnozoites biomarkers have been published to date (Figure 3.7c).
Identification of Binary and Regression-Based Parameters for Liver Stage Invasion
and Development. Similar to identification of candidate cell types for building a mimetic liver
platform, thousands of combinations of factors exist to optimize a human Plasmodium liver
platform. We identified over 20 steps in the >100 steps of the liver stage assay protocols where
one of several options needed to be selected. As time and funding requirements ruled out trial
and error, we instead identified factors critical to liver stage development and labeled them
‘binary,’ leaving other factors which were thought to have an increasing positive or negative
effect on liver stage culture as ‘regression.’ Binary factors included the parasite line, the host
hepatocyte lot, and the use of antimycotics in culture; these factors were optimized first (Table
3.1).
Plasmodium falciparum and P. vivax Sporozoite Sourcing. Few P. falciparum labadapted lines are capable of forming viable gametocytes, mosquito forms, or sporozoites as
extended in vitro culture selects for rapid blood stage growth and loss of mosquito and liver stage
122
genetic elements (Vaughan et al., 2012a). In particular, a family of 27 Apetela/Ethylen Response
Factors (AP2) tightly control the parasite’s life cycle; loss of AP2 function has resulted in
formation of nonviable gametocytes in P. berghei and similar mechanisms may occur in P.
falciparum (Painter et al., 2011; Sinha et al., 2014). Field isolates are, by definition, wild type
and fully capable of gametocytogenesis but much time and effort is needed to adapt them for lab
culture. To standardize sourcing of P. falciparum sporozoites, we tested lab-adapted lines which
have been passed through mosquitoes, chimps or humans, and back into in vitro blood stage
culture, having the effect of enriching the culture for competent parasites. After many
unsuccessful attempts to produce P. falciparum sporozoites from NF-54, 3D7, and MRA1000,
we obtained a line from Dr. Stefan Kappe, cloned it by limiting dilution, and characterized clone
F7 (KF7) for robust sporozoite production in A. stephensi (Figure 3.8a). Consistent, monthly
shipments of A. dirus mosquitoes fed with P. vivax-infected patient isolate blood from Mahidol
University, Bangkok, Thailand were found to be inconsistent for sporozoite production—
shipments were tailored to contain two different isolate feeds, ensuring at least one feed per
shipment produced sporozoites. Positive feeds resulted in generation of several million
sporozoites for experimentation and cryopreservation (Figure 3.8c).
Primary Hepatocyte Donor Screen. Our preliminary studies revealed intrinsic features of
primary hepatocyte lots, especially viability, platability, and susceptibility to infection, were
divergent and in need of characterization for suitability in our liver model. Similar studies
recently published revealed only two in eight human donor lots were highly permissive to P.
falciparum sporozoite infection (March et al., 2013). Viability above 80% and tight platability
was found necessary to achieve a primary hepatocyte monolayer in the device lower channel,
and a primary hepatocyte donor screen revealed consistent results with Celsis donor NON.
123
Across several experiment with NON, it proved not only permissive to P. falciparum and P.
vivax, but also more permissive than most donors from other lots and companies [(Table 3.2),
(personal communication, unpublished data)]. Primary hepatocyte variance in terms of
permissiveness to infection has been hypothesized to be a result of the timing, protocol, and
quality of hepatocyte isolation from surgically removed liver sections; protocol optimization and
quality control has not yet been standardized (personal communication). Identifying NON as a
candidate donor lot, and obtaining several dozen vials for experimentation, was essential as this
protects experimental results and conclusions against false negative results.
Ascertaining Antimycotic Activity on Liver Stage Parasites. Previous work in our lab
revealed Amphotericin B features antiplasmodial activity, yet this potent antimycotic was found
helpful in controlling yeast contamination from mosquito debris in extended device culture.
Several experiments were performed to assess the effect of Amphotericin in our model. The
addition of Torpedo Mix to Celsis hepatocyte media and Amphotericin B as part of Lonza’s
hepatocyte media bullet kit did result in nearly statistically-significant (p=0.08) reduction in liver
form quality and quantity (Figure 3.8b). In particular, we noted many round parasites lacking
DNA staining in Amphotericin B-treated cultures; possibly indicating death after invasion. In
combination with well-rehearsed clean-dissection protocols and consistent, daily washing of
infected devices during media change, Amphotericin B was found unnecessary for extended
device culture.
Exploring Nonparenchymal Effects on Liver Invasion Rates. After optimizing our
Plasmodium liver stage assay protocol for reproducible three-day development, we experimented
with direct and indirect co-cultures, as well as treating sporozoites with nonparenchymal liver
124
fractions. We hypothesized that if several known steps in the sporozoite’s invasion pathway to
the liver are missing from standard well-based liver models, and we provided those fractions in
our model, then we could recover in vivo-like invasion and development rates necessary for drug
study statistical power. Prior to obtaining a usable development model, we utilized the
reproducibility of HC-04 hepatocytes in the lower channel of a second generation device for
short-term (less than 4 hr) invasion studies with LX1 stellate cells and HLSEC endothelial cells.
Aside from previously published benefits of stellate-hepatocyte co-culture, stellate cells are
responsible for laying down a majority of liver extracellular matrix proteins (Krause et al., 2009;
Loreal et al., 1993; Watanabe et al., 2003). Detailed studies of P. berghei sporozoite liver
invasion have revealed a strong interaction between the parasite’s coat protein, circumsporozoite
protein (CSP), and highly-sulfated liver heparin n sulfate proteoglycans (HSPG), trigger proper
cleavage of CSP and sporozoite invasion (Coppi et al., 2005; Coppi et al., 2007; Pradel et al.,
2004). Directly co-culturing HC-04 hepatocytes and LX1 stellate cells, followed by an invasion
assay, revealed a possible 2-3 fold increase in invasion rate (Figure 3.9a). However, these
studies are not conclusive as true invasion was not ascertained (by a method such as insideoutside assay), and the studies need to be repeated.
The parasite’s interaction with the liver endothelium and resident macrophages (Küpffer
cells) has also been well-studied (Frevert et al., 2005; Frevert et al., 2006; Tavares et al., 2013).
As mentioned previously, the parasite maintains cell-traversal activity which has been witnessed
in both Küpffer and endothelial cells. However, the route of liver escape, with obvious
implications for vaccine targets, has not been definitively shown to occur through Küpffer cells,
endothelial cells, or both. Preliminary invasion studies with HC-04 hepatocytes and P.
falciparum sporozoites in second generation devices revealed that a device vascularized with
125
HLSEC did not significantly increase or decrease traversal into the lower channel (Figure 3.9b).
This finding is interesting as we found the HLSEC monolayer to be confluent, inferring HLSEC
traversal may have been required for parasites to reach device lower channels. Attempts to
witness HLSEC traversal by live cell microscopy were inconclusive (data not shown). However,
the membrane itself is only 10% porous without a monolayer of cells on it, and is far less porous
with a HLSEC monolayer, thus active recruitment of sporozoites by endothelial cells is one
possible explanation for no decrease in membrane traversal with a HLSEC monolayer. To look
for a possible invasion pathway-trigger effect of HLSEC interaction with P. falciparum
sporozoites, we incubated sporozoites with HLSEC prior to being introduced into the device.
This series-type experiment was designed to preclude possible issues with HSLEC co-culture
with primary hepatocytes, and to avoid possible confounding effects of HLSEC cells blocking
membrane pores. Again, no difference in endothelial treated versus untreated sporozoite
invasion rate was found (Figure 3.9c).
To test if Küpffer cell traversal is an essential step of the sporozoite’s invasion pathway,
we directly co-cultured Küpffer cells with primary hepatocytes in the device lower channel and
infected devices with P. falciparum and P. vivax sporozoites. No increase of invasion or
development rate was noted compared to controls (Figure 3.9d), but we did notice a Küpffer
effect on hepatocytes and, by association, liver forms. The Küpffer cells were seeded and
cultured on top of primary hepatocytes, having the effect of a sandwich culture. Sandwich
cultures establish with ECM overlays and cell overlays have been described to maintain primary
hepatocyte structures (Dunn et al., 1991; Zinchenko et al., 2006). In our sandwich co-culture, we
noticed remarkably improved host hepatocyte health and enhanced liver stage parasite
morphology (Figure 3.9e). We attempted to repeat this result with Matrigel® as the overlay
126
(because Küpffer cells are extremely limited and expensive) and found a possible correlation
between parasite size and a gel overlay (Figure 3.9f). A similar result of using a ECM gel
overlay to enhance liver parasite development has only been recently described, and further
studies are needed to understand the effect of sandwich culture on Plasmodium live models
(Dembele et al., 2014).
Establishment of High Throughput MOI. Throughout our optimization and co-culture
experiments, we also tested a range of parasite inoculums, termed MOI (Multiplicity of
Infection). The in vitro field-standard of MOI is 1 sporozoite per 1 hepatocyte in culture, or 1:1
(Zou et al., 2013). For most liver models, both basic and complex, this MOI requires 50,000 to 1
million sporozoites per well. As a single dissection of 50 P. falciparum –infected mosquitoes
yields anywhere between 50,000 and 250,000 sporozoites, and a single dissection of P. vivaxinfected mosquitoes yields 250,000-20 million sporozoites, this level of throughput is
insufficient for testing multiple dilutions and repeats in a drug assay. We found that perfusing a
small number of sporozoites (1,000-50,000) over a small number of hepatocytes (200-800) in the
device lower channel directly results in at least 10-fold greater per-hepatocytes invasion and
development rates than published in the literature (March et al., 2013; Sattabongkot et al., 2006).
We also found perfusing 10,000 sporozoites through the device, fresh or cryopreserved, was
sufficient for obtaining statistically adequate numbers of developing parasites (Table 3.3).
Successful P. falciparum and P. vivax Liver Stage Development. Protocol
optimizations to use competent parasites, clean dissections, infect sporozoites into NON
hepatocytes, and omit Amphotericin B in culture resulted in highly reproducible culture of P.
falciparum and P. vivax liver forms, including possible hypnozoites, for at least three days
(Figure 3.10). We found NON cell stability over one week in device culture differed from that
127
of primary cell lots used to optimized our original liver model (Chapter 2), thus complete
development was not attained. Stabilization of NON for long-term culture is covered in Chapter
4.
Discussion
Identification of Key Assay Factors. Several logic concerns quickly arose when we
began de novo protocol development for an assay as time consuming, complex, and intertwined
as a human Plasmodium liver stage assay. First, there was an issue of unreliable production of
sporozoites; a principal component for the assay. Securing a sporozoite producing line and
reproducible protocol was essential for all optimization steps. Likewise, communication and
coordination of shipments of P. vivax infected mosquitoes from Thailand were essential for
proper experimental planning and setup. Second, human Plasmodium liver assays are not well
described or standardized and several steps lacked positive controls to rule out false negative
results. In particular, we initially lacked a positive control for sporozoite viability, IFA staining,
the hepatocyte donor screen, and positive identification of liver forms, requiring ‘blind’
experiments for testing each condition. Confirmation of reproducible parasite invasion and
development in a basic version of our liver model allowed for proper interpretation of positive
and negative results throughout further optimization.
Identification of Essential Liver Stage Growth Factors. In this study we describe a
series of experiments designed to identify critical factors missing from standard Plasmodium
liver assays. Endothelial and Küpffer cell effects on sporozoites were not identified, but we did
find a possible doubling of P. falciparum invasion rate when LX1 Stellate cells were co-cultured
with HC-04 hepatocytes. One tantalizing explanation for this result is that LX1 produced matrix
proteins essential for liver invasion in the device’s lower channel, resulting in better activation of
128
sporozoite invasion pathways. Future experiments designed to repeat the assay with primary
hepatocytes, quantify proteoglycan content in the device lower channel, and pull down
sporozoite target molecules for identification will help describe a mechanism of action.
Regardless, we obtained drastically improved P. falciparum and P. vivax invasion rates with both
fresh and cryopreserved sporozoite lots, without the need for co-cultures. Although we have not
yet confirmed or identified a liver invasion pathway in our device, understanding that co-cultures
are not necessary vastly improves the efficiency (in terms of time and throughput) of liver stage
drug assays.
Future Studies. To extend our Plasmodium liver model to 7-10 days to allow complete
liver form development and to 21 days for hypnozoites reactivation and development, we need to
further stabilize NON primary hepatocytes in device culture. Preliminary studies have indicated
NON benefit from tighter confinement than our primary cell lot initially used to develop the liver
model (Chapter 2). Extending the model should allow formation of merosomes, invasion of liver
merozoites into erythrocytes, and completion of the life cycle back through the blood stage
(Vaughan et al., 2012b). Extending the model will also allow for a deeper understanding of the
effect of drug and vaccine studies, whereby we could observe differential morphology and hostparasite interactions under live cell imaging. An extended P. vivax hypnozoite model is required
to test hypnocidal drug efficacy, as only primaquine and tafenoquine are known to eliminate
these dormant forms, emphasizing the need for new antimalarial compounds. A P. vivax assay
would include use of atovaquone, which kills developing liver forms but not hypnozoites, to
detect and observe delayed parasite development as hypnozoite reactivation (Dembele et al.,
2011; Mazier et al., 2009).
129
Our optimized platform provides a novel avenue for future studies into the biology of
Plasmodium liver invasion and development. Device-derived flexibility with perfusion and coculture, in combination with high-resolution imaging and fully validated controls, will
undoubtedly help us to understand the liver stages of Plasmodium. Of particular interest are the
mechanisms behind hypnozoite dormancy and reactivation. Only recently has a study been
published looking into histone modification as a possible trigger for hypnozoites activation; yet
the model used was not capable of confirming hypnozoite morphology (Dembele et al., 2014).
The ability of our device to capture subcellular details through high resolution microscopy has
revealed hypnozoite are multinucleated (Figure 3.10). Once thought to be dormant, it appears
that hypnozoites actually undergo a couple rounds of genome replication before the signals for
dormancy take effect. One possible explanation is that sporozoites are relatively non-responsive
to foreign stimuli; their active molecular machinery is designed for travel and invasion-- most
mRNA needed for function is pre-packed during sporozoite formation. This ‘priming’ for
invasion and immediate development explains why the parasite is able to undergo rapid and
radical morphological changes at several points in the life cycle (Curtidor et al., 2011; GomesSantos et al., 2011; Paton et al., 1990). Thus, the signaling behind the decision to enter
dormancy may take several hours to process after receiving cues during the first hours of the
hepatocytes stages. Interrogation of such mechanisms is currently underway.
130
Figure 3.1. Sexual stage culture workflow. a) Daily workflow of starting and combining four
gametocyte cultures over 18 days, generating male and female gametocytes for mosquito feeding
and infection. b) Micrographs of workflow. High parasitemias during days 3-6 ‘stresses’
parasites and triggers gametocytogenesis. Stage II and III gametocytes appear around day 9 and
mature, round-ended ‘banana’ Stage V gametocytes appear from days 16-18.
131
Figure 3.2. Full liver stage assay workflow. Production and preparation of mosquitoes,
gametocyte cultures, sporozoite harvesting, device production, and device seeding must be
tightly controlled for proper alignment of critical timepoints.
132
Figure 3.3. Invasion and development quantification. a) IFA-stained P. falciparum
sporozoite in HC-04 culture at 3 hrs post infection. b) Imaging for CSP in devices reveals CSPpositive events in lower channel (white dots), quantifiable with image processing software. Scale
bars, 50 µm. c) A sporozoite-like artifact in device lower channel. Artifacts show no parasite
staining patterns and always appear as half-circles. d) A three-color round P. falciparum liver
form. e) A four-color P. falciparum sporozoite. f) A multinucleated, four-color round P.
falciparum form. (c-f) Parasites at 3 days post invasion. Scale bars, 5 µm.
133
Figure 3.4. Reproduction of P. falciparum and P. berghei liver models. a) P. berghei-GFP
imaged with a 10x objective in a HepG2 microtiter culture at 48 hrs post invasion and (b) 96 hrs
post invasion. c) PbGFP liver schizont in HepG2 on transwell membrane at 72 hrs. Scale bar, 10
µm. d) Correlation of PbGFP development rate to sporozoite inoculum at two days. e) PbGFP
forms noted 12 hrs post complete development in co-cultures of HepG2 (H) and LX1 stellates
(S) or NIH3T3 fibroblasts (F). Error bars represent standard deviation. f) Correlation of P.
falciparum development rate to sporozoite inoculum in HC-04 8 chamber slide culture at day 3.
g) Live cell imaging PbGFP parasites at 10 hrs, 30 hrs (h), and 60 hrs (i) post invasion into
HepG2 cells in device bottom channel (images are of three different parasites). Scale bars, 5 µm.
134
Figure 3.5. Sporozoite traversal assay. a) Still frame from time-lapse prior to hepatocyte
traversal by a P. falciparum sporozoite, arrows point to gliding sporozoites (top), box indicates
cell to be traversed full of green cytoplasm (bottom). b) Thirty three seconds after (a), a
sporozoite is captured traversing an individual cell, disrupting cell morphology (top) and causing
release of green cytoplasm (bottom). Scale bars, 5 µm.
135
Figure 3.6. Sporozoite gliding and flow cytometry invasion assay. a) P. falciparum
sporozoite gliding assay demonstrating sporozoite migration across glass b) HC-04-flow
invasion assay with unstained, stained, and stained/heat-shocked sporozoites. c) Flow cytometry
gating for HC-04 infection rate with unstained (left), stained (center) and stained/heat-shocked
sporozoites (right) used for data collection for (b).
136
Figure 3.7. IFA validation and optimized deconvolution imaging. a) A P. falciparum
sporozoite imaged at 100x before (left) and after (right) deconvolution. b) Example P.
falciparum sporozoite, rounding, and round forms in HC-04 stained with directly conjugated
antibodies. c) Day 4 P. vivax liver forms at high and low magnification, demonstrating loss of
fine and important details under low magnification and without deconvolution. All scale bars, 5
µm.
137
Table 3.1. Binary and regression factors to be optimized for liver stage culture.
Factor
Binary or
Regression
Comment
Literature Reference
Pf parasite line &
Pv case, fresh or
cryopreserved
Binary and
Regression
No 2 lines or cases are equal.
Dissection
conditions
Regression
Buffer, pH, temp, time all affect spz
viability/infectability
(Hegge et al., 2010)
Host Hepatocyte
Lot
Binary and
Regression
HHF07007 from Yecuris and NON
from Celsis
(March et al., 2013), (Zou et al.,
2013), (Albuquerque et al., 2009)
Antibiotics/
Binary and
Regression
Antimicrobials are important for
nonsterile culture work but can directly
affect parasites
Co-culture
Regression
Küpffers/Endothelials/Fibroblasts affect
spz and heps
(Mazier et al., 1985), (Frevert et
al., 2005), (March et al., 2013),
(Dembele et al., 2014)
ECM
Regression
Collagens, Laminins, Proteoglycans
affect spz and heps
(Pradel et al., 2004), (Sinnis and
Coppi, 2007), (Dembele et al.,
2014)
Incubation
Conditions
Regression
Collaborators have shown hypoxia is
beneficial to liver stages parasites
(Ng et al., 2014)
Multiplicity of
Infection (MOI)
Regression
Too many/few spz essential to obtain
statistically sig. rates
Media
components (ie
HGF)
Regression
Host growth factors can influence
development
USF Pf lines, fresh & cryo Pv show +
results
Antimycotics
138
(Tao et al., 2014),
Figure 3.8. Identification of candidate P. falciparum parasite line, stable parasite sourcing,
and effect of antimycotics on parasites. a) Five different parasite lines characterized for
formation of liver forms. F7 (KF7) is the parent to G4-MchLuc. b) Addition of Torpedo Mix
and/or Fungizone® to liver stage culture to control yeast has deleterious effect on parasite growth
(p=.08). c) Consistent shipments of P. vivax- infected mosquitoes from Thailand and production
of P. falciparum sporozoites are essential for repeated studies.
139
Table 3.2. Primary hepatocyte donor comparison.
Primary Source
Lot
Experiments
Results
Yecuris
HHM11003
Pf (n=1), Pv (n=3)
possible Pv forms
Yecuris
HHF01008
Pf (n=1), Pv (n=1)
no development
Yecuris
HHF07007
Pf (n=3)
day 2 round forms
Yecuris
SHS
Pv (n=2)
no development
Yecuris
HHM10002
Pv (n=1)
no development
Yecuris
LTS
Pf (n=1)
no development
Celsis
NON
Pf (n=8)
day 3 round forms
140
Figure 3.9. Testing regression parameters on liver stage growth. a) Preliminary experiments
with a HC-04 and LX1 co-culture in second generation devices, using invasion as readout, reveal
possible doubling of P. falciparum infection rate in the present of co-culture. b) An indirect
HLSEC and HC-04 co-culture, and triculture, in second generation devices shows no effect on
invasion rate. Error bars represent standard deviation. c) Treating P. falciparum sporozoites
with HLSEC has no effect on development rate. d) A direct primary hepatocyte and Küpffer coculture has no effect on development rate. e) Parasites in the primary hepatocyte-Küpffer coculture do show enhanced development over three days. Scale bar, 5 µm. f) Comparison of
parasite diameters from primary hepatocyte and hepatocyte-Matrigel® overlay cultures indicated
development may be enhanced by overlays.
141
Table 3.3. Summary of Multiplicity of Infection across all experiments.
Fresh
Parasite source Fresh Pf Fresh Pf Fresh Pf Cryo Pf Pv
Cryo
Pv
Spz innoculum 1,000
5,000
10,000 14,000 10,000 10,000 16,000 50,000
per device
6.3
4.0
2.7
14.5
3.17%
2.00%
1.37%
7.25% 8.25% 5.00%
3.75% 29.00%
0.63%
0.08%
0.03%
0.07% 0.17% 0.10%
0.05% 0.12%
Results in per #
parasites… hepatocytes
(day 3)
Cryo
Cryo Pv Pv
per #
sporozoites
16.5
10.0
7.5
*Data from controls of optimization, drug, and vaccine studies (Chapters 2 and 3).
142
58.0
Figure 3.10. Plasmodium falciparum and P. vivax liver stage development. a) Representative
3 day P. falciparum round liver forms in device culture. b) Representative 3 day P. vivax round
forms in device culture. c) Three possible P. vivax hypnozoites after 3 days of device culture,
showing a protrusion in the PVM (white arrow). All scale bars, 5µm.
143
References
Albuquerque, S.S., Carret, C., Grosso, A.R., Tarun, A.S., Peng, X., Kappe, S.H., Prudencio, M.,
and Mota, M.M. (2009). Host cell transcriptional profiling during malaria liver stage
infection reveals a coordinated and sequential set of biological events. BMC Genomics
10, 270.
Coppi, A., Natarajan, R., Pradel, G., Bennett, B.L., James, E.R., Roggero, M.A., Corradin, G.,
Persson, C., Tewari, R., and Sinnis, P. (2011). The malaria circumsporozoite protein has
two functional domains, each with distinct roles as sporozoites journey from mosquito to
mammalian host. J Exp Med 208, 341-356.
Coppi, A., Pinzon-Ortiz, C., Hutter, C., and Sinnis, P. (2005). The Plasmodium circumsporozoite
protein is proteolytically processed during cell invasion. J Exp Med 201, 27-33.
Coppi, A., Tewari, R., Bishop, J.R., Bennett, B.L., Lawrence, R., Esko, J.D., Billker, O., and
Sinnis, P. (2007). Heparan sulfate proteoglycans provide a signal to Plasmodium
sporozoites to stop migrating and productively invade host cells. Cell Host Microbe 2,
316-327.
Curtidor, H., Patino, L.C., Arevalo-Pinzon, G., Patarroyo, M.E., and Patarroyo, M.A. (2011).
Identification of the Plasmodium falciparum rhoptry neck protein 5 (PfRON5). Gene 474,
22-28.
Dembele, L., Franetich, J.F., Lorthiois, A., Gego, A., Zeeman, A.M., Kocken, C.H., Le Grand,
R., Dereuddre-Bosquet, N., van Gemert, G.J., Sauerwein, R., et al. (2014). Persistence
and activation of malaria hypnozoites in long-term primary hepatocyte cultures. Nat Med
20, 307-312.
Dembele, L., Gego, A., Zeeman, A.M., Franetich, J.F., Silvie, O., Rametti, A., Le Grand, R.,
Dereuddre-Bosquet, N., Sauerwein, R., van Gemert, G.J., et al. (2011). Towards an in
vitro model of Plasmodium hypnozoites suitable for drug discovery. PLoS One 6,
e18162.
Dunn, J.C., Tompkins, R.G., and Yarmush, M.L. (1991). Long-term in vitro function of adult
hepatocytes in a collagen sandwich configuration. Biotechnol Prog 7, 237-245.
Frevert, U., Engelmann, S., Zougbede, S., Stange, J., Ng, B., Matuschewski, K., Liebes, L., and
Yee, H. (2005). Intravital observation of Plasmodium berghei sporozoite infection of the
liver. PLoS Biol 3, e192.
Frevert, U., Usynin, I., Baer, K., and Klotz, C. (2006). Nomadic or sessile: can Küpffer cells
function as portals for malaria sporozoites to the liver? Cell Microbiol 8, 1537-1546.
144
Gomes-Santos, C.S., Braks, J., Prudencio, M., Carret, C., Gomes, A.R., Pain, A., Feltwell, T.,
Khan, S., Waters, A., Janse, C., et al. (2011). Transition of Plasmodium sporozoites into
liver stage-like forms is regulated by the RNA binding protein Pumilio. PLoS Pathog 7,
e1002046.
Hegge, S., Kudryashev, M., Barniol, L., and Frischknecht, F. (2010). Key factors regulating
Plasmodium berghei sporozoite survival and transformation revealed by an automated
visual assay. FASEB J 24, 5003-5012.
Ishino, T., Chinzei, Y., and Yuda, M. (2005). A Plasmodium sporozoite protein with a
membrane attack complex domain is required for breaching the liver sinusoidal cell layer
prior to hepatocyte infection. Cell Microbiol 7, 199-208.
Ishino, T., Yano, K., Chinzei, Y., and Yuda, M. (2004). Cell-passage activity is required for the
malarial parasite to cross the liver sinusoidal cell layer. PLoS Biol 2, E4.
Kariu, T., Ishino, T., Yano, K., Chinzei, Y., and Yuda, M. (2006). CelTOS, a novel malarial
protein that mediates transmission to mosquito and vertebrate hosts. Mol Microbiol 59,
1369-1379.
Kaushansky, A., Rezakhani, N., Mann, H., and Kappe, S.H. (2012). Development of a
quantitative flow cytometry-based assay to assess infection by Plasmodium falciparum
sporozoites. Mol Biochem Parasitol 183, 100-103.
Krause, P., Saghatolislam, F., Koenig, S., Unthan-Fechner, K., and Probst, I. (2009). Maintaining
hepatocyte differentiation in vitro through co-culture with hepatic stellate cells. In Vitro
Cell Dev Biol Anim 45, 205-212.
Lindner, S.E., Miller, J.L., and Kappe, S.H. (2012). Malaria parasite pre-erythrocytic infection:
preparation meets opportunity. Cell Microbiol 14, 316-324.
Loreal, O., Levavasseur, F., Fromaget, C., Gros, D., Guillouzo, A., and Clement, B. (1993).
Cooperation of Ito cells and hepatocytes in the deposition of an extracellular matrix in
vitro. Am J Pathol 143, 538-544.
March, S., Ng, S., Velmurugan, S., Galstian, A., Shan, J., Logan, D.J., Carpenter, A.E., Thomas,
D., Sim, B.K., Mota, M.M., et al. (2013). A microscale human liver platform that
supports the hepatic stages of Plasmodium falciparum and vivax. Cell Host Microbe 14,
104-115.
Mazier, D., Beaudoin, R.L., Mellouk, S., Druilhe, P., Texier, B., Trosper, J., Miltgen, F., Landau,
I., Paul, C., Brandicourt, O., and et al. (1985). Complete development of hepatic stages of
Plasmodium falciparum in vitro. Science 227, 440-442.
Mazier, D., Renia, L., and Snounou, G. (2009). A pre-emptive strike against malaria's stealthy
hepatic forms. Nat Rev Drug Discov 8, 854-864.
145
Mota, M.M., Hafalla, J.C., and Rodriguez, A. (2002). Migration through host cells activates
Plasmodium sporozoites for infection. Nat Med 8, 1318-1322.
Mota, M.M., and Rodriguez, A. (2001). Migration through host cells by apicomplexan parasites.
Microbes Infect 3, 1123-1128.
Mueller, A.K., Camargo, N., Kaiser, K., Andorfer, C., Frevert, U., Matuschewski, K., and
Kappe, S.H. (2005). Plasmodium liver stage developmental arrest by depletion of a
protein at the parasite-host interface. Proc Natl Acad Sci U S A 102, 3022-3027.
Nardin, E.H., Nussenzweig, V., Nussenzweig, R.S., Collins, W.E., Harinasuta, K.T., Tapchaisri,
P., and Chomcharn, Y. (1982). Circumsporozoite proteins of human malaria parasites
Plasmodium falciparum and Plasmodium vivax. J Exp Med 156, 20-30.
Ng, S., March, S., Galstian, A., Hanson, K., Carvalho, T., Mota, M.M., and Bhatia, S.N. (2014).
Hypoxia promotes liver-stage malaria infection in primary human hepatocytes in vitro.
Dis Model Mech 7, 215-224.
Painter, H.J., Campbell, T.L., and Llinas, M. (2011). The Apicomplexan AP2 family: integral
factors regulating Plasmodium development. Mol Biochem Parasitol 176, 1-7.
Paton, E.B., Vudmaska, M.I., Zolotukhin, S.B., Zhivolun, A.N., and Krupskaia, I.V. (1990).
[Highly conserved primary structure of the L10 protein from Salmonella typhimurium in
comparison to Escherichia coli assures the possibility of their regulation of expression of
genes in the rplJL operon of E. coli]. Bioorg Khim 16, 1280-1282.
Pradel, G., Garapaty, S., and Frevert, U. (2004). Küpffer and stellate cell proteoglycans mediate
malaria sporozoite targeting to the liver. Comp Hepatol 3 Suppl 1, S47.
Roobsoong, W., Maher, S.P., Rachaphaew, N., Barnes, S.J., Williamson, K.C., Sattabongkot, J.,
and Adams, J.H. (2014). A rapid sensitive, flow cytometry-based method for the
detection of Plasmodium vivax-infected blood cells. Malar J 13, 55.
Sattabongkot, J., Yimamnuaychoke, N., Leelaudomlipi, S., Rasameesoraj, M., Jenwithisuk, R.,
Coleman, R.E., Udomsangpetch, R., Cui, L., and Brewer, T.G. (2006). Establishment of a
human hepatocyte line that supports in vitro development of the exo-erythrocytic stages
of the malaria parasites Plasmodium falciparum and P. vivax. Am J Trop Med Hyg 74,
708-715.
Sinha, A., Hughes, K.R., Modrzynska, K.K., Otto, T.D., Pfander, C., Dickens, N.J., Religa, A.A.,
Bushell, E., Graham, A.L., Cameron, R., et al. (2014). A cascade of DNA-binding
proteins for sexual commitment and development in Plasmodium. Nature 507, 253-257.
Sinnis, P., and Coppi, A. (2007). A long and winding road: the Plasmodium sporozoite's journey
in the mammalian host. Parasitol Int 56, 171-178.
146
Talman, A.M., Blagborough, A.M., and Sinden, R.E. (2010). A Plasmodium falciparum strain
expressing GFP throughout the parasite's life-cycle. PLoS One 5, e9156.
Tao, D., King, J.G., Tweedell, R.E., Jost, P.J., Boddey, J.A., and Dinglasan, R.R. (2014). The
acute transcriptomic and proteomic response of HC-04 hepatoma cells to hepatocyte
growth factor and its implications for Plasmodium falciparum sporozoite invasion. Mol
Cell Proteomics 13, 1153-1164.
Tavares, J., Formaglio, P., Thiberge, S., Mordelet, E., Van Rooijen, N., Medvinsky, A., Menard,
R., and Amino, R. (2013). Role of host cell traversal by the malaria sporozoite during
liver infection. J Exp Med 210, 905-915.
Trager, W. (1971). A new method for intraerythrocytic cultivation of malaria parasites
(Plasmodium coatneyi and P. falciparum). J Protozool 18, 239-242.
Trager, W. (1977). Cultivation of Plasmodium falciparum. Arch Pathol Lab Med 101, 277-278.
Vaughan, A.M., Mikolajczak, S.A., Camargo, N., Lakshmanan, V., Kennedy, M., Lindner, S.E.,
Miller, J.L., Hume, J.C., and Kappe, S.H. (2012a). A transgenic Plasmodium falciparum
NF54 strain that expresses GFP-luciferase throughout the parasite life cycle. Mol
Biochem Parasitol 186, 143-147.
Vaughan, A.M., Mikolajczak, S.A., Wilson, E.M., Grompe, M., Kaushansky, A., Camargo, N.,
Bial, J., Ploss, A., and Kappe, S.H. (2012b). Complete Plasmodium falciparum liverstage development in liver-chimeric mice. J Clin Invest 122, 3618-3628.
Watanabe, T., Shibata, N., Westerman, K.A., Okitsu, T., Allain, J.E., Sakaguchi, M., Totsugawa,
T., Maruyama, M., Matsumura, T., Noguchi, H., et al. (2003). Establishment of
immortalized human hepatic stellate scavenger cells to develop bioartificial livers.
Transplantation 75, 1873-1880.
Wirtz, R.A., Rosenberg, R., Sattabongkot, J., and Webster, H.K. (1990). Prevalence of antibody
to heterologous circumsporozoite protein of Plasmodium vivax in Thailand. Lancet 336,
593-595.
Zinchenko, Y.S., Schrum, L.W., Clemens, M., and Coger, R.N. (2006). Hepatocyte and Küpffer
cells co-cultured on micropatterned surfaces to optimize hepatocyte function. Tissue Eng
12, 751-761.
Zou, X., House, B.L., Zyzak, M.D., Richie, T.L., and Gerbasi, V.R. (2013). Towards an
optimized inhibition of liver stage development assay (ILSDA) for Plasmodium
falciparum. Malar J 12, 394.
147
Chapter Four:
Determination of Antiparasitic Therapeutic Efficacy within a Microfluidic Plasmodium
Liver Stage Culture Platform
Rationale of Study
In Chapters 2 and 3 we describe a novel liver model and further optimization of that
model for Plasmodium liver stage development. Although continued efforts to complete liver
stage development are required, most Plasmodium liver stage therapeutic assays do not measure
complete development as they run for less than three days. Thus we set about optimizing the
model and workflow for vaccine and drug efficacy studies. As previous research in our lab has
generated data on the P. berghei- HepG2 liver model, we optimized a similar luciferase-based
assay with human parasites and microfluidic devices by constructing mCherry-luciferase
expression vectors and transgenic P. falciparum lines. The red-fluorescent protein mCherry was
selected because parasites expressing self-reporting fluorescent markers throughout the life cycle
allow for unique microscopic interrogation and selection (i.e. via FACS sorting) for basic
biological discovery, while luciferase is a preferred viability signal because of its precision and
rapid, simple quantification in inhibition assays.
After identification of NON primary hepatocytes as susceptible to P. falciparum and P.
vivax invasion and development, we designed and executed assays to confirm Primaquine
efficacy against both parasites, as well as an inhibition of invasion and development assay
against P. falciparum sporozoites. However, NON structure and phenotypes demonstrated
instability in long-term culture, resulting in studies to further optimize device geometries in an
148
attempt to rescue the phenotype. Current efforts focus on information gleaned from geometry
experiments to supplement what we know is required of a hypnozoite-capable Plasmodium liver
stage drug discovery platform to design a multiplexed liver model.
Materials and Methods
Review of the Optimized Microfluidic Device Based Liver Stage Protocol. Detailed
sporozoite production protocols can be found in Chapter 3. Plasmodium falciparum KF7 line
(and, after completion, KF7 mCherry-luciferase subclones) were subjected to gametocyte culture
and fed to cartons of Anopheles stephensi mosquitoes by Hemotek membrane feeding.
Mosquitoes were checked for midgut oocysts at day 8-9 and sporozoites beginning at day 18
post-feed. Prior to dissection, mosquitoes were cleaned by dipping into 2 mL 70% EtOH, 2 mL
100x PenStrep, and 2 mL 100x Fungizone®. Salivary glands were collected by needle method in
PBS and collected into 50 mL RPMI on ice. Approximately 36-72 mosquitoes were dissected
per Plasmodium falciparum collection, resulting in 1,000 to 50,000 sporozoites. Similar
dissections of P. vivax from A. dirus shipped from Thailand resulted in collection of 250,000 to 2
million sporozoites. Gliding motility was assessed by sandwich IFA, and remaining sporozoites
were used for fresh liver assays or cryostored for future use.
Detailed device fabrication and seed protocols can be found in Chapters 2. Microfluidic
devices were seeded at least two days prior to an expected mosquito dissection or sporozoite
cryovial thaw date. After fabrication, PDMS-membrane, third generation devices were placed in
petri dishes and packaged for overnight ethylene oxide sterilization. Device lower channels were
coated with 5 µg/cm2 rat tail collagen I in 0.02 N acetic acid in PBS at 37 ºC overnight in a
humidified incubator, with 2 mL of sterile water in the petri dish to prevent evaporation. Coated
devices were washed with 50 µL complete Bioreclamation IVT Plating media (containing 1x
149
PSN only) by first pushing the acetic acid solution out of the lower channel, clipping off the
lower channel with binder clips, and then wetting the upper channel. A vial of Bioreclamation
IVT human hepatocytes, lot NON, were rapidly thawed for 2 min in a 37 ºC water bath and then
transferred into 10 mL media. Hepatocytes were spun at 100 RCF for 5 min, supernatant
removed, and pellet suspended in 500 µL media for counting by Trypan Blue exclusion.
Hepatocyte concentration was re-established to 10,000 hepatocytes/µL, then hepatocytes were
injected into the device lower channel with a sterile HPLC syringe. After 4 hours, device tubing
elements were removed and replaced with sterilized tip reservoirs, prior to being filled with 100
µL media. Seeded devices were quality checked for monolayers and absence of overseeding.
For P. berghei assays, HepG2 and HC-04 hepatocytes were infected instead of primary human
hepatocytes.
Drug Assay Overview and Protocol. Two types of liver drug assays using three
different Plasmodium species, both cryopreserved and freshly dissected, were confirmed to
develop within the the device. As a proof of concept for luciferase-based detection we infected
10,000 freshly dissected P. berghei-GFP-luciferase sporozoites into each of 15 devices seeded
with HepG2 cells (Franke-Fayard et al., 2008; Franke-Fayard et al., 2004). Three hours postinfection, devices were washed thrice with media before drug ICI56,780 diluted to 10, 1, 0.1, or
0.01 ng/mL was added to devices in triplicate, including three DMSO controls. Devices were
left undisturbed in a humidified 5 % CO2 incubator for 48 hrs. Liver stage growth was assessed
by removing almost all media from devices, followed by addition of 50 µL 1x Luciferase Cell
Culture Lysis Buffer (Promega, Madison, WI, USA) for 30 min at 37 ºC. Following lysis, all
liquid was removed from devices and placed into an opaque OptiplateTM microtiter plate
(PerkinElmer, Waltham, MA, USA), 20 µL Detection Reagent LAR II (Promega) was added to
150
each well, and the plate read on a Top Count NXTTM (PerkinElmer) with a 10 sec integration
time.
A P. falciparum drug assay was performed by infecting 10,000 freshly dissected KF7
sporozoites into each of 12 devices seeded with primary hepatocytes. Three hours post infection,
devices were washed thrice with media before addition of 0.1, 1, or 10 ng/mL primaquine
bisphosphate in triplicates, including three DMSO controls. Media was replaced daily with
drugged media. After three days, devices were fixed by washing media with 1x PBS and
replacing the device volume with 4% paraformaldehyde for 10 min at RT, followed by washing
with PBS. Devices were then permeabilized with 0.1% Triton X 100 for 3 min at RT, washed
with PBS, and blocked with 3% BSA for 30 min at 37 ºC. Devices were stained with 1:1000
Alexa Fluor® 647-conjugated-antiCSP (JA9), 1:1000 Alexa Fluor® 488-conjugated-antiGRPBiP
(JA18), and 1:1000 Hoechst in PBS for 1 hr at 37 ºC, followed by washing. Device lower
channels were scanned at 100x with a Deltavision® Core (GE Healthcare Biosciences,
Pittsburgh, PA, USA) and liver form images deconvoluted with a Softworx® image processing
package (GE Healthcare Biosciences).
Plasmodium vivax drug assays were performed with fresh and cryopreserved sporozoites.
Cryopreserved sporozoites were used to explore the effect of Matrigel® overlay on both
hepatocyte metabolism and liver form development. P. vivax sporozoites were rapidly thawed in
a 37 ºC waterbath and pelleted at 12,000 RCF for 5 min at 4 ºC. The pellet was suspended in
media, counted by hemocytometer, and 16,000 sporozoites were added to each of 21 devices
seeded with primary human hepatocytes. Three hours post infection, devices were washed thrice
with media. Nine device lower channels were overlaid with 2 mg/mL Matrigel® in cold media
by adding tubing to channel ports, clipping top channel tubes, and injecting cold Matrigel®
151
solution. Thirty minutes post overlay, device tubing was removed and replaced with sterile
reservoirs. Devices were drug treated by adding 0.1, 1, or 10 ng/mL primaquine bisphosphate to
each of three Matrigel® overlayed and three non-overlayed devices, including three DMSO
controls. Media was replaced daily with drugged media. After three days, devices were fixed
and prepared for IFA as above, and then stained with 1:1000 mouse anti-HSP70 and 1:1000
rabbit anti-UIS4 [(Mueller et al., 2005), (courtesy of Dr. Stefan Kappe lab)] in PBS for 1 hr at 37
ºC. Following wash, devices were counterstained with 1:1000 Hoechst, 1:1000 goat anti-rabbit
Alexa Fluor® 488 (Life Technologies, Grand Island, NY, USA), and 1:1000 goat anti-mouse
Alexa Fluor® 568 (Life Technologies). Devices were images as described above.
A fresh P. vivax liver stage drug assay was performed similarly, infecting 12 devices with
12,000 sporozoites each, without any overlays. All data were analyzed to generate dose-response
curves and IC50 values in Prism (Graphpad, La Jolla, CA, USA).
Vaccine Assay Setup and Protocol. A P. falciparum inhibition of invasion and
development assay was performed with cryopreserved sporozoites by thawing as above, adding
20,000 sporozoites to each of 7 tubes, and adding inhibitory antibody 2A10 or noninhibitory
antibody 2F2 to achieve 25, 2.5, or 0.25µg/mL antibody per tube, with 1 tube as a no antibody
control (Nardin et al., 1982). Sporozoites were treated with antibody solution for 30 min at RT
prior to inoculation into primary hepatocyte-seeded devices in triplicate. Three hours post
infection, devices were washed thrice with media, followed by daily media change. At three
days post infection, devices were fixed, stained, and imaged as above.
Generation of mCherry-lucifersase-Expressing P. falciparum Sporozoite-Producing
Lines. Creation and characterization of reporter constructs was a collaborative effort with Dr.
Hitoshi Otsuki, Dr. Shulin Xu, and Ms. Alison Roth. The P. berghei EF1α bidirectional
152
promoter has previously been shown to drive expression throughout the parasite life cycle
(Franke-Fayard et al., 2004; Vaughan et al., 2012). To enable rapid detection of liver stage
growth, we inserted firefly luciferase into pL-BACII-bEDMH—a piggyBac construct previous
built in our lab, containing a P. berghei EF1α promoter driving a dihydrofolate reductase cassette
and mCherry reporter. We PCR-amplified the luciferase coding region from P. falciparum 3D7luciferase (Kyle et al., in preparation), ligated the PCR product to a TA cloning vector, and
confirmed the sequence by sequencing. We then digested and the luciferase cassette and
pEDMH with StuI and blunt-end ligated it into bEDMH, thus obtaining bEDMH-Luc (Roth et
al., in preparation). After selection of a bacterial colony with proper luciferase directionality
confirmed by PCR, we prepared DNA of the vector for transfection (Figure 4.1a).
Transfection was performed using the piggyBac system (Balu et al., 2005). Plasmodium
falciparum line KF7 was cultured in human erythrocytes at 5% hematocrit in gametocyte culture
medium without antibiotics, using the closed-flask method described in Chapter 3. Blood stage
schizonts were purified by MACS magnetic column (Miltenyi Biotec) and 1 million purified
parasites were added to erythrocytes loaded with 300 μg of the transposon plasmid and 150 μg of
the transposase plasmid DCTH (Balu et al., 2009), beginning a 5 mL transfected culture.
Transformed parasites were selected with 5 nM WR99210 in gametocyte media at 3 days post
infection, and after recovery of parasites about two weeks later. Individual mutant clones were
obtained by limiting dilution of parasites post-drug selection (Figure 4.1b). Luciferase assays
were performed using the Luciferase Assay System (Promega). Specifically, mosquito assays
were performed by adding a mosquito to 100 µL lysis buffer, followed by freezing at -80 ºC for
2 hrs. Mosquito solution was then ground with a pestle and pelleted at 15,000 rpm for 1 min.
The supernatant was collected and 100 µL of Steady-Glo® Reagent was added to the supernatant
153
and read on an opaque plate reader as above. For sporozoite assays, sporozoites were collected
in 100 µL lysis buffer and frozen at -80 ºC for 2 hrs. Following thaw, 100 µL LARII reagent
was added to the solution in an opaque microtiter plate and read on a plate reader as above (Roth
et al. in preparation). Blood, gametocyte, mosquito, and sporozoite stages were confirmed for
mCherry signal by live cell imagining with a Deltavision® Core.
Results
Basic Protocol Development with P. berghei-luciferase and Hepatic Cell Lines.
Identification of accurate drug IC50 values is dependent on statistical power. Values are better
identified when multiple drug dilutions and repeats of drug dilutions are incorporated into the
assays. To understand how many devices are needed to gain statistical power and to optimize
workflow, and to demonstrate a proof-of-concept that luciferase can be utilized in a device liver
growth assay, we tested experimental compound IC56,780 on P. berghei liver stages in
HepG2and HC-04 seeded devices. Several rounds of assays were needed to identify the proper
hepatocyte seed conditions and MOI necessary for signal detection. Experimental design was
optimized to infect devices with 10,000 sporozoites into 12 devices, which were drugged with
three dilutions and a DMSO control. While the controls and drug dilutions below the IC50
value resulted in a wide range of luciferase readings, devices drugged above the IC50 were
reproducibly attenuated (Figure 4.2a). The IC50 value detected was about 10-fold higher than
that obtained by comparable well-based assays at 0.02 ng/mL (Dong et al., unpublished data).
We suspect the small molecule-absorbing properties of PDMS may be responsible for the loss of
drug activity. Remarkably, after sporozoite dissection, a total of about 2 hrs was needed to infect
devices and read growth via luciferase, demonstrating the efficiency of a luciferase assay over
IFA imaging.
154
Inhibition of P. falciparum and P. vivax Development with Primaquine. Primaquine
activity was studied against P. falciparum and P. vivax to demonstrate primary hepatocyte device
culture as a drug discovery platform. Across three experiments IC50 values of 0.39, 0.55 and
0.40 µM were identified in strong agreement with another, recently published human
Plasmodium liver assay (March et al., 2013). We were able to consistently obtain 12-17 liver
forms developing in control devices from parasite inoculums ranging from 10,000 to 16,000
(Figure 4.2b-d). Additionally, while cryopreservation has been described as decreasing liver
parasite formation 8-fold, we found sporozoites cryopreserved with a new protocol were half as
viable as freshly dissected sporozoites (March et al., 2013, Singh et al., in preparation). The
assay workflow proceeded as efficiently as well-based cultures—and improved efficiency was
found during microscopic detection of liver forms by IFA. As the surface area of each device
(each replicate) is less than 2 mm2 and could be covered by 100x high-resolution microscopy in
about 15 min, replicates could be counted notably faster in comparison to 8 chamber slides
harboring a 0.70 cm2 culture area. These studies also confirmed efficient use of sporozoite
resources. In a representative experiment, 15 liver forms were noted in about 200 hepatocytes
from a 10,000 sporozoite inoculum. With a liver form-to-hepatocyte ratio of 2.5%, we achieved
a 14-fold increase in three-day development rate compared to other models.
Previous reports have associated enhanced primary hepatocyte phenotypes with IC50
value curve shifts toward greater activity in in vitro assays (Chan et al., 2013; March et al.,
2013). These studies highlight the importance of mimetic P450 enzyme expression and function.
As previously noted, NON hepatocytes were found to be less stable in long-term culture than
other primary hepatocytes. Thus, a Matrigel® overlay protocol was deployed to recover rescue
long-term maintenance of primary phenotypes in a P. vivax drug study (Dunn et al., 1991,
155
Dembele et al., 2014). While the study needs to be repeated, preliminary results did indeed show
a kill curve shift toward greater activity in Matrigel® overlay devices (Fig 4.2c).
Inhibition of P. falciparum Development with Inhibitory Antibody. Recent studies
have shown that anti-CSP antibodies demonstrate a concentration-dependent inhibition of
sporozoite invasion and development (Zou et al., 2013). To demonstrate device usefulness for
similar assays, we performed an inhibition of liver stage development assay (ILSDA) by treating
P. falciparum sporozoites with inhibitory and non-inhibitory monoclonal antibody (mAB),
followed by quantifying growth by IFA after three days in device culture. In devices with no
antibody present there was an average of 14.5 liver-stage parasites detected at the end of the 3day assay. In these P. falciparum ILSDA studies the sensitivity and specificity of an inhibitory
and a non-inhibitory mAB were compared, using anti-PfCSP mAB 2A10 compared to antiPvCSP mAB 2F2. Anti-PfCSP mAB 2A10 inhibited P. falciparum sporozoite infection and
development in primary human hepatocytes in a concentration-dependent manner in contrast to
anti-PvCSP mAB 2F2 that had no measurable effect on P. falciparum development at any
concentration (Fig. 4.3). The higher concentrations of anti-PfCSP mAB 2A10 of 2.5 & 25 µg/ml
led to a significant (P = 0.0198) reduction in parasite development with 1.3 and 2.3 liver-stage
parasites per device, respectively. None of the concentrations of anti-PvCSP mAB 2F2 resulted
in significantly different P. falciparum LS development from the no antibody control group
(Figure 4.3a).
Historically, ILSDA and inhibition of sporozoite invasion (ISI) assays measure growth
inhibition after 1-24 hours, however, as Zou et al. (2013) notes, an extended assay endpoint of
24-96 hrs reveals inhibitory activity otherwise overlooked. As mentioned previously, our devicebased ILSDA was extended to 3 days—exposing an antibody-specific phenomenon’s not
156
previously reported. Sporozoites inhibited with 2A10 before inoculation showed restriction to the
inlets of the device channels, whereas 2F2 treated parasites were detected throughout the device.
This implies that bound anti-CSP antibody reduced sporozoite-gliding motility when in static
culture and holds potential as an additional method to measure mechanism of action.
Furthermore, immunofluorescence staining with P. falciparum-specific antibodies detected
altered morphological appearance of transformed sporozoites treated with mAB 2A10 (Figure
4.3b). The observed activity suggests that mAB 2A10 interacts with sporozoites, interferes with
normal growth mechanisms, and suppresses parasite development after hepatocyte invasion.
Liver Culture Optimization for Multiplexing
Materials and Fabrication Selection. PDMS is a useful material for prototyping, but is
not practical for use in a refined cell culture platform. While PDMS possesses beneficial
properties such as optical clarity and oxygen transport, it is highly porous and can absorb small
molecules, possibly confounding drug and vaccine assays (Toepke and Beebe, 2006). Indeed,
we did note a 10-fold reduction of IC56,780 activity against P. berghei in our PDMS-based liver
devices. We next tested polystyrene for biocompatibility with NON hepatocytes and confirmed
that the hepatocytes grow proficiently for several days on pieces of PS. Ongoing efforts are
focused on optimizing embossing protocols for fabrication of multiplexed liver platforms.
Identification of Confining Geometries. During long-term liver stage drug assays we
noticed degeneration of NON hepatocyte architecture after a week in culture. This result differs
from results in Chapter 2, where a Yecuris primary cell lot formed multicellular structures
featuring bile canaliculi after about a week in device culture. A three week albumin assay
confirmed loss of primary hepatic levels of albumin expression from NON hepatocytes in
157
typical, Matrigel®-overlaid, and collagen gel-overlaid device cultures (Figure 4.4a). However,
Matrigel® and collagen gel did help NON maintain morphology over the course of the assay.
We speculated that NON require addition levels of confinement including, but not limited to,
smaller device dimensions and confinement on the lateral and vertical dimensions. To test our
hypothesis, a series of channels were constructed with widths varying from 50 to 2000 µm, and
the series repeated with channel heights of 40, 70, and 95 µm. To compare the effect of different
widths, albumin production quantified after week 1 and week 2 was standardized by channel
width, assuming that width correlated with cell number. Results show a drastic decrease in
albumin production as channel dimensions expand to 500 µm and beyond, highlighting a
possible threshold where the microphysical space of the channel is no longer affects hepatocyte
morphology (Figure 4.4d). Additionally, after two weeks of NON culture, the 95 µm tall
channels showed a decrease in albumin expression from week 1 to week 2, but the 40 and 70 µm
tall channels maintained expression, indicating an effect of vertical confinement. Such an effect
has been recently characterized and is attributed to a sandwich culture-type phenomenon (Wang
et al., 2013). Hepatocytes are confined in almost every dimension in vivo: on two axes they
contact other hepatocytes, forming apical surfaces, and on their third axis contacting liver
sinusoids, forming basolateral surfaces (Feracci et al., 1987). Confinement in at least two
dimensions may help to mimic liver structure in vitro.
As channels are simple to fabricate but difficult to access and multiplex, we also tested if
NON could be confined within open microwells, which are more practical for high throughput
production and use. Microwell devices are characterized by one or many microfeatures of
various sizes (ranging from 50 to 2000 µm across) and shapes (rows, squares, and circles)
grouped within a macrowell. While cell seeding and media change are handled in the macro
158
feature, the wells still contain micro features to direct cell culture phenomena. We designed a
novel type of microwell with an extremely small dead space between wells, as preliminary
studies revealed localization of hepatocytes to microwells and not macrofeatures is impractical.
Remarkably, when testing circular and square geometries, we found that NON hepatocytes form
stable multicellular structures in circular, but not square, microwells (Figure 4.5). Ongoing
studies are aimed to better understand this preference.
Development of a Self-Reporting P. falciparum Line. Even further drug and ILSDA
optimization can be achieved with the integration of the P. falciparum mCherry-luciferase
reporter line. A piggyBac construct, pL-BACII-bEDMH-Luc, was made in pL-BACII-bEDMH,
which carries mCherry and hDHFR (human dihydrofolate reductase) as a selection marker with
the P. berghei EF1α bidirectional promoter. Multiple independent clones were isolated with
several expressing high levels of luciferase and mCherry signals through all stages of the malaria
life cycle. The clone KF7G4 was ultimately selected once confirmed to infect mosquitoes, form
viable sporozoites, and express both luciferase and mCherry in mosquito stages. Additionally,
KF7G4 showed a linear correlation of luciferase production with oocyst number in mosquito
midguts, along with sporozoite number in salivary glands. The current amount of liver-stage
parasites demonstrated in the devices is ample to quantitatively assay luciferase activity and
therefore establish reliable throughput for rapid readout (Figure 4.6). Additional studies are
planned to validate liver form expression of mCherry, detection by luciferase, and incorporate
this self-reporting parasite to enhance the speed and sensitivity of the assay.
Rapid IFA Quantification via High Content Imaging. While the exact architecture of a
multiplexed liver platform are undetermined, we do know the platform must be amendable to
high-resolution microscopy such that we can capture drug and vaccine-induced interactions, and
159
identify P. vivax liver forms, as no self-reporting P. vivax parasites are currently available. To
this end, we plan to optimize devices for detection by a high content imaging system, similar to a
Deltavision® Core with a microtiter stage adapter (Figure 4.7a) and PerkinElmer Operetta® HighContent screener. Initial testing with the Operetta® system have resulted in protocols for
medium-resolution imaging of the entire device bottom channel. Ongoing studies aim to use the
Harmony software to automatically find and quantify liver forms (Figure 4.7b).
Discussion
Importance of Optimization to Meet Parasite Requirements. In this study we have
identified requirements for a medium-to-high throughput Plasmodium liver stage drug and
vaccine platform. These requirements are derived from device manufacturing capabilities, liver
model stability, the liver model’s ability to support Plasmodium infections, and the practicality of
using this liver model for multiplexed screening. We have weighed these requirements against
our overarching goal: to reproducibly generate Plasmodium liver forms in mimetic in vitro
culture, especially the production of P. vivax hypnozoites capable of reactivation, to test efficacy
of drugs and vaccines on this required, but asymptomatic form of the parasite. Characteristics of
the parasite and current advances in the field require 1) medium-to-high resolution microscopy for
detection of hypnozoites, 2) stable liver culture for at least three weeks for detection of
hypnozoite reactivation and 3) production of enough hypnozoites in a single condition to provide
statistical significance. These requirements were incorporated in designs for a multiplexed
device.
Efficiency of Assays. One fundamental advantage of our model over other published
reports is our efficiency in terms of hepatocytes used, parasites used, reagents required, and time
needed to run assays. Hepatocytes are an extremely limited resource. A single cadaver liver
160
perfusion typically generates a few hundred vials of primary hepatocytes, available for $500$1000 per vial containing 1-10 million viable cells. An initial screen of donor lots for suitability
in an assay can cost thousands of dollars, adding expense to hepatocyte sourcing. As drug study
results are donor-specific, a single donor should be identified for comparable studies. The FRG®
KO chimeric mouse model is one possible solution to hepatocyte sourcing issues, but this model
is also expensive and not fully proven to provide reproducible results when hepatocytes are
serially passed (personal communication). A single vial of NON hepatocytes contains 8.5
million cells, and a populated device channels contain between 200-800 primary hepatocytes.
Such efficiency ameliorates hepatocytes as a limiting reagent and, coupled with integration of
luciferase into Plasmodium liver stage assays, expands single experimental throughput from
dozens to hundreds or thousands of conditions.
While parasite-free liver models must focus on rate limits from hepatocytes, antimalarial
liver models have an added burden of rate limiting due to parasite sourcing. Stable production of
P. falciparum sporozoites is labor-intensive, requiring a fully operation insectary producing
thousands of mosquitoes per week and constant gametocyte culture work. Obtaining P. vivax
sporozoites is dependent on transmission seasons in source countries, such as Thailand, and
cooperation of at least two laboratories, civil businesses, and government authorities to allow
shipment of Plasmodium-infected mosquitoes across international borders on commercial
aircraft. While we have demonstrated such efforts are possible, repeated drug and vaccine
studies are only economically feasible with efficient use of 1-20 million sporozoites per lot.
Here we demonstrate a Plasmodium liver model requiring merely a few hundred thousand
sporozoites for a small drug study. In tandem with the idea that one vial of hepatocytes could
161
seed 1000 devices, such an experiment requires only 1 million sporozoites, thus we have
matched hepatocyte and sporozoite efficiency.
As long as microscopy is needed to quantify P. falciparum and P. vivax liver stage
growth, resources must be placed on scanning the culture surface area for parasites as they occur
in about 1 in 100 hepatocytes in vitro. The hepatocyte culture area in devices is about 32 times
smaller than that of a microtiter wells, necessitating proportionally less time to analyze compared
to other reports (Dembele et al., 2014). These advantages are strengthened by our advances in
per-hepatocyte infection rate. By directing parasites into confined spaces, we have presumably
enhanced sporozoite-hepatocyte mixing, leading to a drastically improved infected hepatocyte
rate in our model. Taken together, these advances in efficiency enable more drug and vaccine
candidates per experiment, more dilutions per experiments, and more replicates per experiments.
Multiplexed Device Requirements. Current multiplexed design efforts are focused on
understanding and incorporating known requirements into a first-generation multiplexed
platform. Platform design efforts have focused on incorporating parasite efficiency
requirements, and have attempted to mesh them with practical fabrication methods. From a
manufacturing standpoint, there exists a chasm between PDMS prototyping and the production
capabilities of injection molding and stamping. In particular, manufacturing the extremely thin
bottom element necessary for high resolution microscopy is not trivial, nor is the interface
between fluidics handling equipment and devices in a multiplexed model. An open-well type
geometry is more amendable to multiplexing, as it lacks features like closed channels that require
multi-step assemblies and bonding. Furthermore, technology transfer of a working multiplexed
liver model is much more practical when fluidics can be handled by robots and multichannel
162
pipettes; many companies already have drug libraries in microtiter format (malERA, malERA,
2011). For these reasons, we are developing an open microwell format for multiplexing.
Recent experiments to better understand what types of microphysical spaces are
necessary for NON stability in long-term culture have yielded interesting, but not unprecedented,
geometry results. We found compaction within a 200 and 300 µm-diameter circular microwell
promotes stable NON phenotypes, but similar compaction in a channel or square well may be
less stable as identified by Live/Dead staining (Figure 4.5). One study using posts to drive
hepatocyte spheroid formation tested the effects of spacing between the posts and found an
increase in spacing from 20 to 200 µm resulted in less consistent spheroid formation (Evenou et
al., 2010). In another study, microcavities ranging from 40 to 200 µm were filled with
hepatocytes and covered by an ECM gel to initiate bile production. Quantification of bile
production revealed an ideal cavity width of 60 µm, but bile formation was noted across all
conditions (Matsui et al., 2012). These results confirm what has been shown in the development
of a micropatterned co-culture model featuring islands of hepatocytes surrounded by fibroblasts.
In that study, hepatocyte islands of 36, 100, 490, 6800, and 17800 µm diameters were compared
for albumin and urea production, with the 36 µm island producing significantly more albumin
and urea than other sizes in a size-dependent nonlinear fashion (Bhatia et al., 1998). All these
data taken together suggest an ideal culture system width between 35-200 µm, or about 2-10
hepatocytes wide. It is feasible that the cell-cell and cell-wall interactions help to resist
hepatocyte spreading; widths larger than a few hundred microns lead to a sudden loss of
phenotype.
Logistics of Sporozoite Sourcing. Sporozoite sourcing may be the single most teamoriented, essential task for drug and vaccine assays. Future studies may incorporate seasonal
163
assessment of P. vivax therapeutics by collection and cryobanking of sporozoites in the Thai
summer, followed by drug and vaccine experiments in the low transmission season. This
workflow would require adequate cryopreservation techniques, which we have begun validating.
Future studies. Ongoing efforts aim to fully optimize and validate our transgenic, selfreporting P. falciparum line, well-based longevity of NON, and our multiplexed model. Future
efforts will focus on scale-up and technology transfer of the liver model to enable its use in front
line drug-discovery laboratories
164
Figure 4.1. mCherry-luciferase construct cloning strategy and transfection. a) A luciferase
cassette from 3D7-luc was PCR amplified and cloned into bEDMH, forming bEDMH-Luc. b)
Schematic of transfection from Balu et al. (2005), in which the piggyBac transposase inserts
ITR1-ITR2 randomly into TTAA sites in the P. falciparum genome, following transfection by
loaded erythrocyte method.
*(b) Source: Balu et al., 2005, Copyright (2005) National Academy of Sciences, U.S.A.
165
Figure 4.2. Plasmodium berghei, P. falciparum, and P. vivax drug assays. a) An IC50 value of
0.24 µM is identified from a 15-device P. berghei assay featuring 4 drug dilutions, using
luciferase as a readout. b) An IC50 value of 0.39 µM is identified from a 12-device P.
falciparum assay featuring 3 drug dilutions, using IFA as readout. c) An IC50 curve shift is
noted in Matrigel® sandwich cultures infected with cryopreserved P. vivax in a 24-device assay
featuring 3 drug dilutions and IFA as readout. IC50 = 0.55µM. d) An IC50 value of 0.40 µM
confirms results from (c), but with freshly dissected P. vivax sporozoites in a 12-device assay
featuring 3 drug dilution and IFA as readout. All error bars represent standard deviation.
166
Figure 4.3. Antibody-specific inhibition of P. falciparum liver stage development. a) An
ILSDA reveals inhibition of P. falciparum liver stage development with 2.5 µg/mL antibody
2A10 but no inhibition with 2F2. b) Representative images of immunofluorescent stained P.
falciparum parasites after exposure to various concentrations of anti-PfCSP (2A10), anti-PvCSP
(2F2), and no antibody control. Scale bars, 5 µm.
167
Figure 4.4. Identification of channel geometries for stable NON phenotype. a) A three week
albumin assay reveals NON do not maintain albumin expression levels, even with collagen and
Matrigel® overlays. b) Schematic of different channel width seeded with NON. Channels were
created 40, 70, and 95 µm tall for each width. c) Two week albumin data stratified by channel
height, showing 95 µm tall channels do not maintain the phenotype. d) Two week albumin data
stratified by channel width and standardized to cell number showing smaller geometries maintain
the phenotypes, with a possible threshold below the 500 µm width. Error bars represent standard
deviation.
168
Figure 4.5. Preliminary studies on microwells for multiplexed liver platform. a) Schematic
of microwell devices where an array of microwells of different diameters reside in a macrowell,
holding media. b) A three week albumin assay revealing maintenance of NON phenotype in
various round, but not square, geometries. c) Live-dead staining of hepatocytes after three weeks
in culture show multicellular structures in round, but not square, geometries. Scale bars, 100
µm.
169
Figure 4.6. mCherry-luciferase construct and P. falciparum transformation. a)
Diagram of piggyBac expression vector featuring a dual-direction P. berghei EF1α
promoter, firefly luciferase, a dihydrofolate reductase selection cassette, and an mCherry
reported gene. b-d) Luciferase signal from blood stages, oocysts, and sporozoites,
respectively. e) mCherry signal in asexual and sexual blood stages. f,g) mCherry signal
in oocyst and from two clones, G4 and H7. h) mCherry signal continues to be expressed
in salivary gland sporozoites. Scale bars, 5 µm.
170
Figure 4.7. Necessary optimization for high-throughput IFA readout. a) A microtiter-stage
adapter for the Deltavision® Core can enable rapid, high-resolution quantification of P. vivax
liver forms in multiplexed device culture. b) Merged image from an Operetta® device scan,
demonstrating the relative surface area of a device lower channel compared to parasite size at
day 3 post infection (white arrow). Scale bar, 5 µm.
171
References
Balu, B., Chauhan, C., Maher, S.P., Shoue, D.A., Kissinger, J.C., Fraser, M.J., Jr., and Adams,
J.H. (2009). piggyBac is an effective tool for functional analysis of the Plasmodium
falciparum genome. BMC Microbiol 9, 83.
Balu, B., Shoue, D.A., Fraser, M.J., Jr., and Adams, J.H. (2005). High-efficiency transformation
of Plasmodium falciparum by the lepidopteran transposable element piggyBac. Proc Natl
Acad Sci U S A 102, 16391-16396.
Bhatia, S.N., Balis, U.J., Yarmush, M.L., and Toner, M. (1998). Probing heterotypic cell
interactions: hepatocyte function in microfabricated co-cultures. J Biomater Sci Polym
Ed 9, 1137-1160.
Chan, T.S., Yu, H., Moore, A., Khetani, S.R., and Tweedie, D. (2013). Meeting the challenge of
predicting hepatic clearance of compounds slowly metabolized by cytochrome P450
using a novel hepatocyte model, HepatoPac. Drug Metab Dispos 41, 2024-2032.
Dembele, L., Franetich, J.F., Lorthiois, A., Gego, A., Zeeman, A.M., Kocken, C.H., Le Grand,
R., Dereuddre-Bosquet, N., van Gemert, G.J., Sauerwein, R., et al. (2014). Persistence
and activation of malaria hypnozoites in long-term primary hepatocyte cultures. Nat Med
20, 307-312.
Dunn, J.C., Tompkins, R.G., and Yarmush, M.L. (1991). Long-term in vitro function of adult
hepatocytes in a collagen sandwich configuration. Biotechnol Prog 7, 237-245.
Evenou, F., Fujii, T., and Sakai, Y. (2010). Spontaneous formation of stably-attached and 3Dorganized hepatocyte aggregates on oxygen-permeable polydimethylsiloxane membranes
having 3D microstructures. Biomed Microdevices 12, 465-475.
Feracci, H., Connolly, T.P., Margolis, R.N., and Hubbard, A.L. (1987). The establishment of
hepatocyte cell surface polarity during fetal liver development. Dev Biol 123, 73-84.
Franke-Fayard, B., Djokovic, D., Dooren, M.W., Ramesar, J., Waters, A.P., Falade, M.O.,
Kranendonk, M., Martinelli, A., Cravo, P., and Janse, C.J. (2008). Simple and sensitive
antimalarial drug screening in vitro and in vivo using transgenic luciferase expressing
Plasmodium berghei parasites. Int J Parasitol 38, 1651-1662.
Franke-Fayard, B., Trueman, H., Ramesar, J., Mendoza, J., van der Keur, M., van der Linden, R.,
Sinden, R.E., Waters, A.P., and Janse, C.J. (2004). A Plasmodium berghei reference line
that constitutively expresses GFP at a high level throughout the complete life cycle. Mol
Biochem Parasitol 137, 23-33.
malERA (2011). A research agenda for malaria eradication: drugs. PLoS Med 8, e1000402.
172
March, S., Ng, S., Velmurugan, S., Galstian, A., Shan, J., Logan, D.J., Carpenter, A.E., Thomas,
D., Sim, B.K., Mota, M.M., et al. (2013). A microscale human liver platform that
supports the hepatic stages of Plasmodium falciparum and vivax. Cell Host Microbe 14,
104-115.
Matsui, H., Takeuchi, S., Osada, T., Fujii, T., and Sakai, Y. (2012). Enhanced bile canaliculi
formation enabling direct recovery of biliary metabolites of hepatocytes in 3D collagen
gel microcavities. Lab Chip 12, 1857-1864.
Mueller, A.K., Camargo, N., Kaiser, K., Andorfer, C., Frevert, U., Matuschewski, K., and
Kappe, S.H. (2005). Plasmodium liver stage developmental arrest by depletion of a
protein at the parasite-host interface. Proc Natl Acad Sci U S A 102, 3022-3027.
Nardin, E.H., Nussenzweig, V., Nussenzweig, R.S., Collins, W.E., Harinasuta, K.T., Tapchaisri,
P., and Chomcharn, Y. (1982). Circumsporozoite proteins of human malaria parasites
Plasmodium falciparum and Plasmodium vivax. J Exp Med 156, 20-30.
Toepke, M.W., and Beebe, D.J. (2006). PDMS absorption of small molecules and consequences
in microfluidic applications. Lab Chip 6, 1484-1486.
Vaughan, A.M., Mikolajczak, S.A., Camargo, N., Lakshmanan, V., Kennedy, M., Lindner, S.E.,
Miller, J.L., Hume, J.C., and Kappe, S.H. (2012). A transgenic Plasmodium falciparum
NF54 strain that expresses GFP-luciferase throughout the parasite life cycle. Mol
Biochem Parasitol 186, 143-147.
Wang, Y., Toh, Y.C., Li, Q., Nugraha, B., Zheng, B., Lu, T.B., Gao, Y., Ng, M.M., and Yu, H.
(2013). Mechanical compaction directly modulates the dynamics of bile canaliculi
formation. Integr Biol (Camb) 5, 390-401.
Zou, X., House, B.L., Zyzak, M.D., Richie, T.L., and Gerbasi, V.R. (2013). Towards an
optimized inhibition of liver stage development assay (ILSDA) for Plasmodium
falciparum. Malar J 12, 394.
173
Chapter Five:
Conclusions
Summary of Findings
Malaria is a significant global health problem with no licensed vaccine and only one class
of drug able to kill dormant liver stages. To prevent transmission, drugs activate against liver
stages and vaccines preventing liver invasion are strongly desired (malERA, 2011a, b).
However, human-specific Plasmodium liver models are often omitted from preclinical drug and
vaccine assessment because human malaria parasites can only infect and fully develop in human
and monkey hepatocytes, and neither hepatocyte is stable in standard culture methods (GuguenGuillouzo et al., 1983; Sattabongkot et al., 2006). Furthermore, no in vitro model capable of
supporting P. vivax hypnozoite reactivation is currently in use preclinical use; detrimental to
drug validation as hypnozoites are often refractory to drug treatment (Dembele et al., 2014). The
goal of this work was to bridge the gap between mouse-based preclinical drug and vaccine
studies and chimpanzee or human trials. To this end, we identified several requirements of
novel, advanced liver model useful for preclinical assessment of new antimalarial therapeutics.
These requirements include:
1. High-resolution imaging capabilities
2. Long-term culture (at least 21 days) of human hepatocytes
3. Confirmation of in vivo- like P450 expression of human hepatocytes in long-term
culture
174
4. Reproduction of natural in vivo Plasmodium sporozoite infection rates in vitro
5. Complete development of P. falciparum and P. vivax liver stages, including P. vivax
hypnozoites
6. Efficient use of hepatocytes and sporozoites
7. Confirmation of drug and vaccine efficacy on P. falciparum and P. vivax liver stages
8. Reactivation of P. vivax hypnozoites
9. Multiplexed operations allowing for manipulation of several culture systems at once,
preferably with widespread robotic and multichannel pipettes
10. Efficient use of detection components, including time to collect data
11. Ease of technology transfer to other facilities.
To generate a novel liver model meeting these requirements, we first identified several
sources of hepatic cells, including hepatocytes, liver endothelium, stellate cells, and Küpffer
cells, and assessed changes of function in various combinations of co-culture. We also
reproduced published protocols for initiation of patterned co-cultures and transwell culture to
look for advantageous arrangements. With little success with these relatively simple models, we
designed and fabricated an optically-clear bilayer device and used it to culture hepatocytes.
Several studies identifying ideal conditions for perfused media found an endothelial cell-lined
flow channel was necessary for long-term hepatocyte culture. Device modifications for static
culture and introduction of primary hepatocytes isolated from chimeric mice revealed the device
itself, without co-culture or perfusion, maintained primary hepatocytes phenotypes for 21 days in
culture. Repeating these results in an array of different device geometries identified channel
compaction as the mechanism for hepatocyte longevity. Thus, requirements 1, 2, and 3 were met
175
by our initial round of experimentation; and although not a requirement, we showed primary
hepatocytes could be extensively sourced from chimeric mice.
To turn our liver model into a Plasmodium liver stage model, P. falciparum sporozoite
production methods were better established and a sporozoite-producing line (KF7) was
identified. Likewise, P. vivax sporozoites were sourced from collaborators and shipped from
Thailand. To detect and quantify invaded and developing parasites, immunofluorescent assay
reagents and protocols were compared and validated. Finally, to complete the most basic version
of our complex liver model, we screened several sources and donors of primary human
hepatocytes, identifying a donor (NON) highly susceptible to invasion and development. As
human Plasmodium sporozoites feature a much higher invasion rate in vivo compared to in vitro,
we next used the device design’s inherent flexibility to test if nonparenchymal fractions of the
liver could recapitulate the sporozoite invasion pathway and properly activate invasion
machinery, potentially leading to a dramatic increase in invasion and development rates. Several
studies with stellate, endothelial, and Küpffer co-cultures revealed only a positive effect with
stellate cells, possibly explained by their ability to lay down liver-specific extracellular matrix
(Pradel et al., 2004). Studies with P. vivax liver stages yielded never-before seen details of
possible hypnozoites under 100x imaging. We determined the lower and upper limit of
sporozoite inoculum and found as few as 10,000 sporozoites are necessary to achieve over 10
developing forms per device. In our second round of experimentation, we partially met
requirements 4, 5, 6 and 7, with ongoing studies aimed at completing these requirements.
To demonstrate utility of our Plasmodium liver model for preclinical studies, we
inoculated devices with P. berghei, P. falciparum, and P. vivax sporozoites and successfully
determined IC50 values of two drugs, using luciferase and three-day development detected by
176
IFA as a readout. We also demonstrated antibody-specific inhibition of sporozoite invasion by
inoculation of P. falciparum sporozoites with inhibitory antibody at various concentrations,
assessing inhibition after three days of development. To enhance the efficiency of future
studies, we generated a mCherry-luciferase expressing P. falciparum line (KF7G4) capable of
sporozoite production and signal expression throughout the life cycle. Ongoing studies are
aimed at validating efficacy of using KF7G4 signals to determine liver stage growth, and
identification of which geometries are necessary for multiplexed operations. Our current
progress shows promise of fulfilling the remaining requirements.
Recent Advances in Human Plasmodium Liver Stage Culture
Over the term of this project, other groups have also produced advanced liver models for
human Plasmodium liver stage culture, some funded by the Bill & Melinda Gates Foundation, in
a unique consortium geared toward co-discovery. A FRG® KO chimeric mouse model has been
recently described robustly reproducing the timing and morphology of in vivo development, such
that Plasmodium liver forms reach 25 µm and larger in diameter and release merozoites in a
synchronous, precise timeline post-infection. Furthermore, hypnozoite formation has been
identified in the model, as well as confirmation of some blood-stage protein expression
hypothesized to also occur in the liver stages (Vaughan et al., 2012). Another group utilizing a
primary hepatocytes micropatterned co-culture have demonstrated similar complete development
(albeit with liver forms smaller than the FRG® KO mouse) and hypnozoite formation, as well as
multiplexed drug and vaccine application (March et al., 2013). Another group not part of the
Gates consortium, who started in vitro human Plasmodium liver stage research over 30 years ago
has recently shown long-term primary monkey hepatocyte culture in a Matrigel®-sandwich
177
model , capable of extremely long-term culture and possible hypnozoite reactivation (Dembele et
al., 2014; Mazier et al., 1985).
Comparison of Drug and Vaccine Platforms. The above mentioned models, including
our own, differ in mechanism and capabilities such that their place in the preclinical study
pipeline may not be the same. While the FRG® KO model produces more robust (larger) liver
parasites than in vitro culture, the model is extremely expensive and not multiplexed, such that
an individual mouse and 100,000 sporozoites are needed for each drug or vaccine concentration
to be tested. Furthermore, post-experimental assessment for P. vivax is dependent on animal
surgery, liver sectioning, and microscopic examination of several square cm of area. The
micropatterned co-culture model is also advanced compared to our device in that they have
completed full development, but like the chimeric mouse model, microtiter plate culture requires
at least 100,000 sporozoites per well and many hours of post-experimental microscopic scanning
for liver forms. While the Matrigel® overlay model is perhaps the only in vitro model to ever
demonstrate hypnozoite reactivation, the need for a gel overlay significantly complicates drug
and vaccines studies as the gel itself can sequester reagents and confound results, and requires
careful handling. Our model is unique compared to other because of its higher efficiency. A
single device culture area can be covered by four images at 10x microscopy, and future iterations
of the system are designed to be amendable to 384-well plate workflow. Efficiency in our
system is also evident in remarkably enhanced invasion rates (up to 1-4 %), 10-fold higher than
most models.
Implications of This Work
Extended use of our device-based liver system lends itself to discovery on at least three
different levels. First, the model’s flexibility is helpful for future studies comparing and
178
contrasting various cell types and culture components, like extracellular matrix and perfusion, to
understand how they add to the model. Most liver models are obligated to function in either
static or flowed conditions, for example, that hinder similar experimentation. Second, the
model’s flexibility is essential to confirm the actual invasion pathway of Plasmodium
sporozoites. To date, several routes of entry to the liver involving Küpffer cells, endothelial
cells, and open endothelial space have been implicated, but none confirmed to be essential or
persuasive. Past invasion, the liver stage is the most ambiguous of a parasite defined by
ambiguity: over half of the parasite’s 5000 genes are of unknown function. Our device could
advance our understanding of the presence and roles of live-stage proteins, especially those
responsible for hypnozoite formation and activation. Lastly, if all requirements above are met,
our liver model could become an essential step in the path taking novel therapeutic from
preclinical to clinic trials.
References
Dembele, L., Franetich, J.F., Lorthiois, A., Gego, A., Zeeman, A.M., Kocken, C.H., Le Grand,
R., Dereuddre-Bosquet, N., van Gemert, G.J., Sauerwein, R., et al. (2014). Persistence
and activation of malaria hypnozoites in long-term primary hepatocyte cultures. Nat Med
20, 307-312.
Guguen-Guillouzo, C., Clement, B., Baffet, G., Beaumont, C., Morel-Chany, E., Glaise, D., and
Guillouzo, A. (1983). Maintenance and reversibility of active albumin secretion by adult
rat hepatocytes co-cultured with another liver epithelial cell type. Exp Cell Res 143, 4754.
malERA (2011a). A research agenda for malaria eradication: drugs. PLoS Med 8, e1000402.
malERA (2011b). A research agenda for malaria eradication: vaccines. PLoS Med 8, e1000398.
March, S., Ng, S., Velmurugan, S., Galstian, A., Shan, J., Logan, D.J., Carpenter, A.E., Thomas,
D., Sim, B.K., Mota, M.M., et al. (2013). A microscale human liver platform that
supports the hepatic stages of Plasmodium falciparum and vivax. Cell Host Microbe 14,
104-115.
179
Mazier, D., Beaudoin, R.L., Mellouk, S., Druilhe, P., Texier, B., Trosper, J., Miltgen, F., Landau,
I., Paul, C., Brandicourt, O., and et al. (1985). Complete development of hepatic stages of
Plasmodium falciparum in vitro. Science 227, 440-442.
Pradel, G., Garapaty, S., and Frevert, U. (2004). Küpffer and stellate cell proteoglycans mediate
malaria sporozoite targeting to the liver. Comp Hepatol 3 Suppl 1, S47.
Sattabongkot, J., Yimamnuaychoke, N., Leelaudomlipi, S., Rasameesoraj, M., Jenwithisuk, R.,
Coleman, R.E., Udomsangpetch, R., Cui, L., and Brewer, T.G. (2006). Establishment of a
human hepatocyte line that supports in vitro development of the exo-erythrocytic stages
of the malaria parasites Plasmodium falciparum and P. vivax. Am J Trop Med Hyg 74,
708-715.
Vaughan, A.M., Mikolajczak, S.A., Wilson, E.M., Grompe, M., Kaushansky, A., Camargo, N.,
Bial, J., Ploss, A., and Kappe, S.H. (2012). Complete Plasmodium falciparum liver-stage
development in liver-chimeric mice. J Clin Invest 122, 3618-3628.
180