High-throughput method for profiling genetically
characterized cancer cell lines with small molecules
Hanh H. Le1, Joshua A. Bittker1, Jaime H. Cheah1, Edmund V. Price1, Michelle Palmer1, Alykhan F. Shamji1, Stuart L. Schreiber1,2
1Broad Institute of Harvard and MIT, 7 Cambridge Center, Cambridge, MA 02142; 2Howard Hughes Medical Institute, Chemistry & Chemical Biology, Harvard University, Cambridge, MA 02142
Identification of the genetic
dependencies of small molecule
sensitivities provides tools for
1,000 CCLs
(genetic features)
relationships
between
genetic
features
cancer cell
measurement
identifying potential
therapeutic targets in cancers
CTD probe kit
8 (in duplicate)
with a given genotype. By
(high specificity)
concentrations
using highly specific probe
molecules and genetically
characterized cell lines,
hypotheses can be generated
roles of quantitative variables
by correlating sensitivities and Abby Bracha, Jaime Cheah, Daisuke Ito, Ke Liu, Edmund Price,
Giannina Schaefer, Alykhan Shamji
genetic features.
Manual cell
culture
Semi-automated
plating
The hypotheses generated by regression analyses are highly dependent on the
variability in genetic features among cell lines tested and variability in small
molecule efficacy. Therefore, it is essential to increase the rate at which
compound response curves can be measured in many cell lines to increase overall
statistical power.
384 well format
Cells passed,
counted, grown
in standard
incubators
BRD1218: HDAC6 inhibitor
(Weiping Tang)
Cells seeded using bulk
dispenser and adhered
in plates prior to
compound addition
Moderate density assay
plates used
O
O
Acoustic cmpd dispensing
Alternatives to improve reliability and automation of cell culture
are being implemented.
• A CompacT automated cell culture robot will automate the cell
expansion and plating of up to 90 lines concurrently.
• Cell culture on alginate carrier beads will enable cells to be
dispensed with no pre-adhesion necessary prior to compound
treatment. Rather, cells will be treated as a bulk reagent. Cells
are expanded in a bead suspension, with a 50 mL tube yielding
equivalent cell numbers to 10 traditional T175 flasks.
Cell-based assays can be
miniaturized to 1536 well density
assay plates. This poses challenges
in liquid handling and compound
transfer but quadruples the
throughput of assays.
• Compound transfer requires lowvolume steel pins or acoustic
dispensing
• For best throughput, a CCDbased reader is required.
• Higher throughput allows
doubling of dose curve range,
reducing need for compound
concentration re-optimization
In place of runtime contact dispensing,
compounds can be pre-printed via
acoustic dispensing into empty assay
plates to generate Assay-Ready Plates
(ARPs).
• Plates can be pre-generated and stored,
enabling fewer runs at assay time.
• Effects of compound treatment on
suspended adherent cells is unknown;
can be solved using bead-based cells.
GEM (Global
Eukaryotic
Microcarrier) alginate
culture bead and
Biolevitator
suspension culture
instrument (Global
Cell Soliutions &
Hamilton)
CompacT cell culture robot
(The Automation Partnership)
OCH3
Cell-line profiling workflow cycle
Plating, freezing, & compound treatment
A549 cells and CHOK1 cells were either grown in a flask or on beads. An aliquot
of beads was then frozen and recovered. In the case of CHOK1, recovered
beads were tryspinized and adhered to fresh beads. Flask-grown, GEM, or GEM
frozen & recovered cells were dispensed into 384 well plates, treated with varying
concentrations of staurosporine, and measured for viability (CellTiterGlo,
Promega) Z’ scores based on the top dose (3.33 uM) are indicated.
2000000
Flask
Compound selection &
acquisition
Cell scale-up (CompacT)
Compound formatting
GEM prep
Implementation & dependencies
Some of the proposed improvements are already in routine use in the Broad
Institute Chemical Biology screening platform, while others require development.
As indicated above, some technologies can only be implemented with the
successful development of other methods. We have therefore begun developing
one of the key enabling techniques, bead-based cell culture.
GEM
IC50s of the three
conditions were
within 2-fold (11
nM to 22 nM)
CHOK1
1800000
DMSO
1600000
Staurosporine
1400000
1200000
1000000
800000
600000
400000
200000
0.86
0.74
0.62
Flask
GEM
GEM-frozen
0
Assay readout
LIMS capture
Assay execution
Flask
GEMtrypsinized
Alginate bead-based tissue culture (GEM):
50 to 100 M cells / 50 mL tube
Can be frozen on bead, thawed & dispensed as reagent
GEM
banking
Cell treatment
23 additional cell lines tested
A549 cells were either dispensed on GEMs or grown on GEMs and trypsinized
for dispensing as a free suspension, and were dispensed either fresh or recovered
from frozen. GEM dispensing of A549 was highly variable, and frozen cells did
not recover well.
BioBanking
Data QC & Curve fitting
Additional data increases confidence in correlations
Potential yield was determined with CHOK1 cells, which were seeded in 50 mL
of GEM gelatin bead suspension in media. 70 million total cells were recovered,
equivalent to the yield of a 10-layer HyperFlask requiring 555 mL of medium.
This is an upper limit of the yield, as CHO cells have been well optimized for
GEM growth and other cells, such as fibroblasts, are less prolific on the beads.
Screening system (HiRes Biosolutions)
with dispenser, incubator, and Viewlux
CCD reader (Perkin Elmer)
Enables use of
Enables use of
Addition data identifies more significant correlations
To take advantage of the throughput of 1536-well plates, a CCD
reader that images the entire plate is
required.
By reducing the number of runtime
steps (dispense cells to pre-printed
plates, incubate, read) the entire
assay can be fully automated on a
small robotic system.
Cell growth condition
ARP printing
& plate map generation
Addition data reduces significant correlations
CCD-reader and
automated integration
GEMfrozen
OCH3
BRD22983
Photo-multiplier based reader
measures wells individually
Acoustic
dispensing of
compound
using Echo
555 (Labcyte)
1536-well cyclo-olefin polymer
assay plate (Nexus Biosciences)
Reagent prep
OCH3
N
Steel pins used to transfer nL
volumes from DMSO stock plate
1536-well plates
OCH3
OCH3
BRD22983
Adhering conditions & yield
Automated & bead-based cell culture
OCH3
N
Global Eukaryotic Microcarriers (GEMs) are 75-150um alginate beads with an
iron core for magnetic manipulation and are available with a variety of cell
adherent matrix coatings. They enable growth of a large number of cells in
reduced media volumes and can theoretically be frozen down and thawed with
cells attached. However, the performance and optimal conditions are highly
variable depending on the cell line. Low-volume dispensing of a bead suspension
also requires development.
Four cancer cell lines from the CTD2 profiling set were mixed with GEMs with a
variety of coatings for 6 or 24 hours. The percentage of cells adhering to the
beads was determined by washing the beads, trypsinizing and counting attached
beads, and comparing to the input number of cells.
O
O
PMT reader
Pin tool probe addition
2
Analytical power depends on highthroughput data generation
Adapting instrumentation and methods commonly used in ultra High-Throughput
Screening (uHTS) would allow the miniaturization or automation of different steps in
the process, increasing the number of cell lines that can be screened in each batch.
Compound
profiling
Hypothesis
generation
Compound
optimization
Genomic
data
Data capture/analysis
The application of the above methods improves different steps of the cell-line
profiling workflow, from reagent preparation to assay throughput to compound
optimization. Furthermore, some improvements such as the ability to freeze
bead-adhered cells or pre-print compound plates, allow effort to be spread out
over time and maximize efficiency of the assay execution.
Assay ready plates (ARPs):
Assay plates pre-spotted with compound set
(currently used for biochemical assays only)
Enables use of
Automated screening system:
Minimize device requirements
Uses only bulk dispensers, incubator, & CCD reader
Integration with existing HTS LIMS
Cell reagent banking
Consistency of reagents & QC
through re-analysis
Routinely used
Alternate assays:
Assay setup compatible with
other analytical reagents
Not currently used
This work is supported by the NCI Cancer Target Discovery and Development Network grant (5 RC2 CA148399-02) and a Broad Institute SPARC grant.
GEM
Despite noise, IC50s
of all 3 conditions
were within 2-fold
(4-8 nM)
100000
A549
80000
DMSO
60000
Staurosporine
40000
20000
0.65
-0.99
Flask
GEM
0.45
0.59
-0.96
-1.04
0
0000
Equivalent numbers of A549 cells were also
dispensed into 1536 well plates, as a free suspension
or frozen and recovered on beads, using a Biotek
MultiFlo. GEM plating showed high variability
visible even to the naked eye.
GEM
Flask frozen GEM frozen GEM frozen
trypsinized
& trypsinized
Cell growth condition
1536 cell plating
CellTiterGlo Luminescence RLU
Relating small-molecule sensitivity to
genetic features of cancer
The current workflow for cell-line profiling uses semi-automated screening techniques.
This requires significant manual labor for several steps and limits the number of cell
lines that can be screened simultaneously.
Development of alginate carrier beads
CellTiterGlo Luminescence RLU
Small-molecule sensitivity profiling of large numbers of cancer cell lines, when
correlated with genetic features, enables the generation of hypotheses about what
targets and pathways may be promising for therapeutic intervention in cancers
with a given genotype. However, the power of this analysis and the confidence of
the conclusions are heavily dependent on the number of cell lines used. As part
of the CTD2 network, the Broad Institute has tested a set of ~200-250 smallmolecule probes in over 150 cell lines towards a goal of 1000 lines. Currently the
time needed to characterize the sensitivity of this many lines exceeds a year, which
limits both the inclusion of newly identified probes and the ability to generate
complete datasets. We have therefore begun investigating the application of
additional high-throughput screening (HTS) methods to cell-line profiling. These
include automated tissue culture robotics, acoustic dispensing of pre-printed
compounds in assay plates, and the use of bead-based 3-dimensional cell culture.
The latter method may be particularly valuable in enabling the handling of cell
lines as a common reagent, circumventing a major bottleneck in the current
process. The ability to handle a variety of bead-based cell-lines as a screening
reagent makes profiling more similar to traditional HTS and enables use of many
well-developed screening techniques. We anticipate that application of these
techniques would increase profiling throughput several fold, greatly reducing the
time needed to profile >1000 cell lines and enabling improved generation and
analysis of data.
Cell-line profiling: from semi-automated to uHTS
CellTiterGlo Luminescence RLU
Increasing the throughput of smallmolecule cancer cell-line profiling
140000
120000
100000
80000
60000
40000
20000
0
200 cells
400 cells
400 cells frozen 400 cells frozen
GEM plate 1
GEM plate 2
We have begun developing several new methods to increase the throughput
of cell-line profiling. One key enabling technology, bead-based adherent
cell culture, has shown promise with some cell lines and requires additional
development to realize its full potential.