Nephrol Dial Transplant (2004) 19: 2198–2207
DOI: 10.1093/ndt/gfh399
Advance Access publication 20 July 2004
Original Article
Evaluation of long-term transport ability of a bioartificial
renal tubule device using LLC-PK1 cells
Nazira Ozgen1, Masuo Terashima1, Tun Aung1, Yoshinobu Sato2, Chie Isoe1, Takatoshi Kakuta2
and Akira Saito1
1
Department of Molecular and Cellular Nephrology, Institute of Medical Sciences, Tokai University, Japan and 2Division
of Nephrology, Tokai University School of Medicine, Kanagawa, Japan
Abstract
Background. Haemodialysis therapy does not provide
renal tubule function, such as active fluid and solute
transport, nor metabolic or endocrine action.
Moreover, this treatment is usually associated with
serious complications and high mortality. We constructed a bioartificial renal tubule device by using
renal tubule epithelial cells in an artificial membrane,
and evaluated transport properties of the device for
2 weeks.
Methods. A renal epithelial cell line, LLC-PK1
(Lewis-lung cancer porcine kidney), was seeded on
polysulfone hollow fibres in small and large modules.
We studied perfusion and leakage of urea nitrogen
(UN) and creatinine (Cr), as well as reabsorption of
water, glucose and sodium for a period of 2 weeks.
Results. Cell-lined hollow fibre membranes significantly reduced the leakage of UN and Cr throughout
the 2 week period. Reabsorption of water, glucose and
sodium were adequate from days 3 to 10 and gradually
decreased thereafter. LLC-PK1 cells actively transported these substances. Scanning electron microscopy
revealed that cells in the hollow fibres on day 8 became
completely confluent. However, they became multilayered and almost obstructed the hollow fibres on
day 13.
Conclusions. This bioartificial renal tubule device
functioned to reabsorb water, glucose and sodium for
10 days. This is the first report of successful longterm evaluation of a bioartificial renal tubule device.
This device, in combination with continuous haemofiltration, may provide treatment to prevent complications of dialysis and raise the quality of life in chronic
renal failure patients.
Correspondence and offprint requests to: Professor Akira Saito,
Department of Molecular and Cellular Nephrology, Institute of
Medical Sciences, Tokai University, Bohseidai, Isehara, Kanagawa
259-1193, Japan. Email: [email protected]
Keywords: bioartificial renal tubule device; chronic
renal failure; continuous haemofiltration; LLC-PK1
cells
Introduction
Haemodialysis therapy, also known as artificial kidney
treatment, saves the lives of end-stage renal failure
patients. However, associated with this are serious
complications and high mortality rates. Increases in
complications due to repeated and long-term intermittent dialysis are especially prevalent among patients
with chronic renal failure. Although conventional
haemodialysis therapy may replace haemofiltration
function, it does not perform the reabsorption, excretion, metabolic and endocrine functions of renal
tubules. Failure to reabsorb filtered extracellular
sodium during dialysis leads to impairment of intravascular fluid refilling and reduced blood volume,
resulting in hypotension [1]. Reduced vitamin D
activation may cause secondary hyperparathyroidism
and subsequent decrements in bone density, leading to
renal osteodystrophy [2]. An inability to metabolise and
remove the filtered low molecular-weight proteins, such
as b2-microglobulin, produces amyloidosis in long-term
haemodialysis patients [3] that is clinically manifested
as carpal tunnel syndrome and destructive arthropathy
associated with cystic bone lesions [4]. In addition,
uraemic patients have high serum levels of advanced
glycation end products (AGE) formed from nonenzymatic glycation and oxidation of proteins [5].
AGE levels are highly correlated with the levels of
carbonyl compounds [6], which are responsible for
carbonyl stress that causes amyloidosis [7] and atherosclerosis [8].
Saito et al. [9] reported that continuous haemofiltration therapy of 10 l/day in chronic renal failure patients
maintained a lower level of blood urea nitrogen (UN),
Nephrol Dial Transplant Vol. 19 No. 9 ß ERA–EDTA 2004; all rights reserved
A bioartificial renal tubule device using LLC-PK1 cells
creatinine (Cr) and b2-microglobulin, compared with
intermittent conventional haemodialysis therapy given
three times a week. To prevent the complications of
dialysis, we have been developing a portable bioartificial kidney using a continuous haemofilter and a
bioartificial renal tubule device, which is composed of
renal tubule epithelial cells and an artificial membrane
that purifies 10 l of blood per day.
Humes et al. [10] reported the use of a bioartificial renal tubule assist device that uses a hollow
fibre module and porcine renal tubule progenitor
cells. Their short-term study demonstrated that the
device created satisfactory fluid, bicarbonate, sodium,
glucose and other standard transport properties, as well
as adequate metabolic and endocrine activities, such
as ammonia production and vitamin D activation in
cells. In the current study, we evaluated long-term
transport abilities of a bioartificial renal tubule device
that used improved hollow fibres and LLC-PK1 porcine
renal proximal tubule epithelial cell lines in order
to reproduce the functions of in vivo renal tubules. We
first evaluated methods for uniformly seeding the
cells, for the transport of fluid and solutes, and for
the maintenance of a confluent monolayer in small
modules. Scaled-up experiments were then performed
with large modules for 2 weeks for studying potential
future clinical use.
Subjects and methods
2199
thick and had a molecular weight cut-off of 6000 Da. A small
module contained 40 hollow fibres of 17 cm in length, with an
effective intraluminal membrane surface area of 56 cm2 and
inner compartment volume of 0.5 cm3. The large one
contained 1600 hollow fibres of 34 cm in length with effective
intraluminal membrane surface area of 4000 cm2 and inner
compartment volume of 35 cm3. The inner surfaces of the
hollow fibres were coated with 40 mg/ml synthetic extra
cellular matrix (ECM) pronectin-F (Sanyo Chemical
Industries, Kyoto, Japan) before seeding with 1–2 107
LLC-PK1 cells/ml. To achieve a uniform distribution of
cells on the entire intraluminal surface of the hollow fibres,
cell-seeding was performed four times at intervals of 1 h each,
during which the module was rotated 90 . Culture medium
was circulated at a rate of 0.25 ml/min for the small module
and 20 ml/min for the large module on the outer compartment
of the module during this period to supply oxygen and
nutrients to the cells. At approximately 1 h after final seeding,
media was circulated in the inner compartment at a rate of
0.25 ml/min for the small module and 20 ml/min for the large
module. For small modules, a single-pass perfusion was
performed throughout the experiment. For the large modules,
a closed-circuit circulation perfusion was performed except
during transport studies when a single-pass perfusion was
performed. Medium was fed with 5% CO2 in air at a rate of 1
l/min through a hollow fibre oxygenator (UOXY 10,
membrane surface area 0.6 m2; Unisyn Technologies,
Hopkinton, MA). Oxygen content and pH of the medium
was monitored with a pH, DO controller (Mettler-Toledo
Process Analytical Inc., Wilmington, MA), and 100% oxygen
was fed when oxygen levels fell below 5.5 p.p.m. The culture
medium was replaced when pH fell below 7.2 (Figure 1).
Cell culture
We used the renal epithelial cell line, LLC-PK1, that
originated from American Type Culture Collection (ATCC,
Rockville, MD). The cells were purchased from the Health
Science Research Resources Bank (HSRRB, Osaka, Japan).
They were cultured at 37 C in Dulbecco’s modified Eagle
medium (glucose concentration 400 mg/dl; Gibco, Invitrogen
Corporation, Grand Island, NY), containing 10% fetal
bovine serum (FBS; Biosource International, Camarillo,
CA), 100 U/ml penicillin and 100 mg/ml streptomycin
(Gibco). The same medium was used as a basal medium in
all of the transport studies described below. The cells were
cultured in 225 cm2 tissue culture flasks (Iwaki, Asahi Techno
Glass Corporation, Chiba, Japan) or 1700 cm2 expanded
surface roller bottles (Corning Incorporated, Corning, NY) at
a density of 1–3 104 cells/cm2. Culture medium was
refreshed three times a week. The cells were removed by
rinsing the bottle once with phosphate-buffered solution
without magnesium and calcium (PBS; Takara Bio Inc.,
Shiga, Japan) before incubation with 0.1% trypsin-ethylenediaminetetraacetic acid (Cosmo Bio Co., Ltd, Tokyo, Japan).
After incubation for 10 min at 37 C, the reaction was stopped
by adding the medium containing 10% FBS. Third-passage
cells were used for module studies.
Preparation of modules
The small and large polysulfone hollow fibre modules were
kindly supplied by Nipro, Osaka, Japan. The hollow fibre
had an inner diameter of 300 mm. Its membrane was 100 mm
Transport studies
Small hollow fibre modules. At 5–6 h after the first seeding
with the cells, the inner compartment of the module was
perfused with the basal medium containing 50 mg/dl UN and
5.0 mg/dl Cr, at a flow rate of 0.25 ml/min. The outer
compartment was perfused with the basal medium without
UN and Cr at the same flow rate. At 2 days after cell seeding,
inlet bottles from both the compartments were replaced with
new medium, and leakage of UN and Cr, and reabsorption
of water, glucose and sodium were evaluated for 24 h.
Synthetic ECM pronectin-F-coated hollow fibres without
cells were used as a control. In another series of experiments,
2.5 g/dl of bovine serum albumin (Wako, Osaka, Japan) was
added to the basal medium perfusing the outer compartment
in order to study the influence of oncotic pressure on the
movement of substances. To study the effect of aquaporin
inhibition, cells were fixed for 5 minutes in ice-cold 0.15 mol/l
cacodylate buffer (pH 7.4) containing 1% glutaraldehyde. To
this, 1 mmol/l HgCl2 was then added to the medium in the
inner compartment and it was perfused through the inner
compartment for 5 min before the transport studies [11]. To
study the influence of active transport inhibition, 0.1 mmol/l
ouabain, an inhibitor of Naþ, Kþ-ATPase, was added to
the outer compartment medium. For all of these transport
studies, samples were taken before and after perfusion from
the medium bottles of inner and outer compartments, and
the media were also weighed. We measured concentrations
of UN, Cr, glucose and sodium, and calculated movement of
substances across the hollow fibre membrane.
2200
N. Ozgen et al.
Fig. 1. Schematic presentation of the large hollow fibre module experiment. Dotted line represents the closed-circuit circulation perfusion
when the single-pass perfusion study for fluid and substances transport was not performed.
Large hollow fibre modules. Evaluations of leakage of UN
and Cr, and reabsorption of water, glucose and sodium
through the hollow fibres were performed by single-pass
perfusion, at 24 h after cell seeding, and at every 2–3 days
thereafter up to 2 weeks. The basal medium perfusing the
inner compartment was added with UN and Cr at the same
concentration as in the study using small hollow fibre
modules, and the medium perfusing the outer compartment
was without UN or Cr. The flow rate was set at 20 ml/min for
both the inner and outer compartments. Inlet bottles of both
compartments were replaced with new medium and singlepass perfusion was performed for 90 min, and movement of
substances across the hollow fibres was evaluated. We used
synthetic ECM pronectin-F-coated hollow fibres without cells
as a control. To study the influence of oncotic pressure on the
movement of substances, 2.5 g/dl albumin was added to the
basal medium perfusing the outer compartment, as was done
with the small hollow fibre modules. In this experiment,
single-pass perfusion was performed for 45 min. Inhibition of
sodium-dependent glucose transport was studied by adding
1 mmol/l phlorizin to the basal medium perfusing the inner
compartment. In all experiments, samples were taken from
the medium bottles, before and after perfusion and the media
were also weighed (Figure 1). We measured concentrations
of glucose, sodium, UN and Cr using standard laboratory
assays and calculated movement of substances across the
hollow fibres.
Determination of leakage and reabsorption. The net
movement of substances from inner to outer compartment
was calculated as follows:
Net movement of the substance ¼ amount collected in the
outlet bottle from the outer compartment – amount that
passed to the outer compartment – amount in the dead space
Each amount of the substance was calculated by multiplying the concentration of the substance by the respective
volume:
Net movement of the substance ¼ CO VO CI (VI VR VD) CD VD
CO, concentration of the substance in the outlet bottle of the
outer compartment; VO, volume in the outlet bottle of
the outer compartment; CI, concentration of the substance
in the inlet bottle of the outer compartment; VI, volume
in the inlet bottle of the outer compartment at the start of
the experiment; VR, volume remaining in the inlet bottle of
the outer compartment at the end of the experiment; CD,
concentration of the substance in the dead space at the start
of the experiment; VD, volume of the dead space.
The volume of the dead space in the small module studies
was so small compared with the 24 h perfusate (360 ml) that
it was not included in the calculations.
Therefore, the net movement of substances from inner
to outer compartment in small modules was calculated as
follows:
Net movement of the substance ¼ CO VO CI(VI VR)
The net movement of UN or Cr from the inner to outer
compartment in the direction of the concentration and
pressure gradients was designated as the leakage. It was also
presented as a percentage of the total amount that had passed
from the inner compartment. Leakage of UN and Cr served
as an indicator of the confluency of LLC-PK1 cells on the
hollow fibre membranes. A smaller leakage indicated a
greater confluency of the cells.
The net movement of glucose or sodium from the inner
to outer compartment was designated as reabsorption by
LLC-PK1 cells, since their concentrations at the inlets of the
inner and outer compartments were almost the same.
A bioartificial renal tubule device using LLC-PK1 cells
2201
Table 1. Leakage and reabsorption of substances in small modules
Leakage
Control group (without cell)
Study group (with cell)
Reabsorption
UN (%)
Cr (%)
H2O (ml/day)
Glucose (mg/day)
Naþ (mEq/day)
40±3
8±3*
36±2
7±2*
NA
14±2
NA
27±2
NA
1.6±3
Experiments were performed on day 2. n ¼ 6 each. Values are expressed as means±SEM.
*P<0.05 vs control group (without cell). NA, negligible amount.
Table 2. Reabsorption of water and sodium in small modules under the influence of enhancer or inhibitors
Control group (base line)
Study group (enhancer/inhibitor)
H2O (ml/h)
H2O (ml/h)
Naþ (mEq/h)
0.5±0.1
1.0±0.2* (albumin)
0.5±0.1
0.2±0.1* (HgCl2)
0.06±0.02
0.03±0.02* (ouabain)
Albumin enhances the osmotic pressure, HgCl2 inhibits the aquaporin and ouabain inhibits the Naþ,Kþ-ATPase. Experiments were
performed on day 2. n ¼ 6 each. Values are expressed as means±SEM.
*P<0.05 vs control group.
Scanning electron microscopy. After 4, 8 and 13 days of
incubation, the modules were disconnected from the incubation system for examination by scanning electron microscopy.
The cells were first rapidly washed with 0.1 mol/l PBS once
and prefixed with 2.5% glutaraldehyde in 0.05 mol/l PBS for
2 h at 4 C. After being washed once with 0.1 mol/l PBS, the
fibres were recovered from the module and sectioned into
pieces. Then, the samples were washed twice in 0.1 mol/l PBS
and fixed with 1% osmium tetroxide in 0.05 mol/l PBS for 1 h
at 4 C. After that, the samples were dehydrated in a graded
series of ethanol solutions, passed through three changes of
100% t-butyl alcohol, and finally dried in a freeze drying
device (JFC-310; JOEL, Tokyo, Japan). After being coated
with gold in an ion sputter coater (JFC-1100; JEOL), cellattached hollow fibres were examined with a scanning
electron microscope (JSM-840A; JOEL).
after seeding were 14±2 ml/day, 27±2 mg/day and
1.6±0.3 mEq/day, respectively (Table 1). When 2.5 g/dl
albumin was added to the basal medium perfusing the
outer compartment to study the influence of oncotic
pressure, reabsorption of water significantly increased
from 0.5±0.1 to 1.0±0.2 ml/hr (Table 2). In contrast,
HgCl2, an inhibitor of aquaporin, significantly
decreased the reabsorption of water from 0.5±0.1 to
0.2±0.1 ml/hr. Similarly, ouabain, an inhibitor of
Naþ,Kþ-ATPase, significantly decreased the reabsorption of sodium from 0.06 ± 0.02 to 0.03 ± 0.02 mEq/h
(Table 2), indicating that transport of water and
electrolytes was facilitated by aquaporin and Naþ,
Kþ-ATPase in the LLC-PK1 cells.
Statistical analysis. Data are shown as means±SEM.
Comparisons of data were done using two-tailed Student’s
t-tests, either paired or nonpaired, by employing Microsoft
Excel software. A P-value of 0.05 was considered statistically significant.
Leakage of UN and Cr as well as transport studies
in large hollow fibre modules
Results
Leakage of UN and Cr and transport studies in
small hollow fibre modules
To evaluate the confluency of LLC-PK1 cells in the
small hollow fibre module, UN and Cr were added to
the basal medium perfusing the inner compartment,
and their leakages to the outer compartment were
measured. Leakages of UN and Cr in small modules
without LLC-PK1 cells were 40±3 and 36±2%,
respectively, and were 8±3 and 7±2%, respectively,
in modules with cells on day 2 of cell-seeding (P<0.05)
(Table 1). The reabsorption of water, glucose and
sodium by LLC-PK1 cells in small modules at 2 days
UN and Cr leakage. To monitor the confluency of
LLC-PK1 cells in the hollow fibres of large modules,
UN and Cr were added to the basal medium perfusing
the inner compartment, and leakage of UN and Cr
through the hollow fibres was measured for 90 min
every 2–3 days. Leakages of UN and Cr in large
modules without LLC-PK1 cells were 44 ± 1.3 and
40 ± 2.4%, respectively, but were 10% in modules
with cells up to 13 days after cell-seeding (Figure 2). A
similar tendency was observed in the concentration
ratio of exiting tubular fluid to the presenting
ultrafiltrate (TF/UF) of the inner compartment and
for the absolute leak rate (Table 3).
Reabsorption of water. The reabsorption of water
from the inner to the outer compartment of large
modules increased gradually and reached a plateau
(100 ml/90 min) on day 6 after cell seeding. Reabsorption then decreased gradually to 80 ml/90 min on day 13
2202
N. Ozgen et al.
Fig. 2. Leakage of UN and Cr in large modules with or without cells. Mean ± SEM (n ¼ 4 for without cells; n ¼ 6 for with cells up to day
8; n ¼ 2 on days 10 and 13). *P<0.05 vs without cell.
Fig. 3. Reabsorption of water in large modules with or without albumin addition. Mean ± SEM (for albumin not added, n ¼ 6 up to day 8,
and n ¼ 2 on days 10 and 13; for albumin added, n ¼ 3 up to day 8, and n ¼ 2 on day 10). *P<0.05 vs albumin not added.
Table 3. TF/UF ratios and absolute leak rates of UN and Cr in large modules
Without Cell
With Cell (day)
1
3
6
8
10
13
Alb
0.7 ± 0.03
(327 ± 6.5)
0.84 ± 0.04
(143 ± 36)*
0.85 ± 0.04
(161 ± 10)*
0.82 ± 0.03
(168 ± 5)*
0.87 ± 0.03
(131 ± 20)*
0.80 ± 0.1
(190 ± 46)
0.76 ± 0.08
(125 ± 45)*
Albþ
0.66 ± 0.01
(353 ± 3.0)
0.74 ± 0.02
(25.5 ± 0)
0.80 ± 0.03*
(205 ± 44)
0.86 ± 0.05
(11 ± 4.0)
0.82 ± 0.02*
(211 ± 29)
0.87 ± 0.04
(12 ± 2.6)
0.82 ± 0.01*
(216 ± 55)
0.86 ± 0.02
(13 ± 0.4)
0.83 ± 0.02*
(196 ± 19)
0.91 ± 0.03
(11 ± 1.3)
0.86 ± 0.14
(242 ± 56)
0.83 ± 0.09
(16 ± 4.1)
0.80 ± 0.08
(11 ± 5.3)
0.71 ± 0.01
(29.7 ± 0.4)
0.82 ± 0.03
(20 ± 4.1)
0.85 ± 0.01*
(18 ± 2.6)
0.84 ± 0.02*
(18 ± 4.7)
0.85 ± 0.02*
(17 ± 0.9)
0.90 ± 0.2
(19 ± 7.1)
UN
Alb
Cr
Albþ
n ¼ 2 each for the group without cells, and for day 13 with cell group. n ¼ 3 each for the remaining groups with cells. Albþ and Alb imply
with or without albumin addition. Values are expressed as means ± SEM. Data in the parentheses are absolute leak rates in mg/dl.
*P<0.05 vs without cell group.
A bioartificial renal tubule device using LLC-PK1 cells
2203
Fig. 4. Reabsorption of glucose in large modules with or without albumin addition. Mean ± SEM (for albumin not added, n ¼ 6 up to day
8, and n ¼ 2 on days 10 and 13; for albumin added, n ¼ 3 up to day 8, and n ¼ 2 on day 10). *P<0.05 vs albumin not added.
Fig. 5. Reabsorption of glucose in large modules with or without phlorizin. Mean ± SEM (n ¼ 3 each). *P<0.02 vs without phlorizin.
(Figure 3). When albumin was added to the basal
medium perfusing the outer compartment, reabsorption of water significantly increased on day 1 and day 3
(160 ml/90 min and 170 ml/90 min, respectively) compared with the group not given albumin (Figure 3).
Reabsorption then decreased gradually to 120 ml/
90 min on day 10. Although reabsorption tended to be
higher than in the albumin group from day 6 onward
too, this difference was not statistically significant
(Figure 3).
or more above the group not given albumin (Figure 4).
This difference narrowed and lost statistical significance on the following days. When sodium-dependent
glucose reabsorption was inhibited by the addition
of phlorizin to the basal medium perfusing the
inner compartment, glucose reabsorption decreased
significantly from 260 to 125 mg/90 min (P<0.02)
(Figure 5), indicating that glucose reabsorption was
facilitated by the sodium-dependent glucose transporter (SGLT).
Reabsorption of glucose. The reabsorption of glucose
from the inner to the outer compartment of the large
module was 300 mg/90 min or more during days 3–10
(Figure 4). When albumin was added on day 1 to the
basal medium perfusing the outer compartment,
glucose reabsorption significantly increased to twice
Reabsorption of sodium. The reabsorption of sodium
was similar to that of water and glucose, and was at
10 mEq/90 min or more during days 3–13, with the peak
of 19 mEq/90 min on day 8 (Figure 6). When albumin
was added to the basal medium perfusing the outer
compartment, sodium reabsorption was significantly
2204
N. Ozgen et al.
Fig. 6. Reabsorption of sodium in large modules with or without albumin addition. Mean ± SEM (for albumin not added, n ¼ 6 up to day
8, and n ¼ 2 on days 10 and 13; for albumin added, n ¼ 3 up to day 8, and n ¼ 2 on day 10). *P<0.05 vs albumin not added. **P<0.05 vs
albumin added day 1.
greater on days 1 and 3 (25.5 mEq/90 min and
26.2 mEq/90 min, respectively, P<0.05 vs albumin not
added). It decreased thereafter on days 8 and 10 to
levels even lower in the group not given albumin. On
day 10, the albumin group had reabsorption that was
significantly lower than on day 1.
Scanning electron microscopy
Hollow fibres were recovered from the large modules
on days 4, 8 and 13, and were observed under a scanning
electron microscope (Figure 7). The hollow fibre
membrane was covered with a confluent monolayer
of cells on day 4 (Figure 7A). On day 8, the cells
became completely confluent and revealed a large
number of microvilli, although there were a few areas
of multi-layered cells (Figure 7B). However, the cells on
day 13 became multi-layered inside the hollow fibre,
and the microvilli were shorter and in smaller number
than those observed on day 8 (Figure 7C).
Discussion
The major obstacles in the development of an artificial
renal proximal tubule device are the obtaining of a large
number of viable renal tubule cells and an adequate
and even lining of the intraluminal surface of hollow
fibres of the modules with these cells in order to
maintain cell function over a longer period of time. To
examine the potential of using cells from xenogenic
origins, we compared three immortalized cell-lines,
including JTC-12 (Japan Tissue Culture-12, monkey
renal proximal tubule cell line), LLC-PK1 and MDCK
(Madin-Darby canine kidney cell line), to test their
ability for expansion, their compatibility with membranes, and their viability during manipulation [12].
Among these, the LLC-PK1 cell line was best characterized as having properties of renal epithelial cells
and that showed easy expansion into large numbers
in in vitro culture. LLC-PK1 cells have also been
reported to provide a reliable model of renal proximal
tubule cells in a number of studies [13,14]. They
displayed typical characteristics of renal proximal
tubule cells, including tight junctions, dome formation,
sodium-dependent phlorizin-sensitive glucose transport
and sodium-dependent amino acid and phosphate
transport. Moreover, LLC-PK1 cells maintained monolayer formation for a long period on membranes coated
with an ECM [15].
Humes et al. [10] constructed a bioartificial renal
tubule assist device by seeding porcine renal
proximal tubule cells into the intraluminal spaces
of the hollow fibres and then studied the short-term
metabolic and endocrine functions of these tubule
cells. They found that the cell-seeded module
required at least 14 days of incubation to obtain
the confluent layer of primary cells before starting
their study. In our study, LLC-PK1 cells reached
confluency within 24 h after seeding, which saved
time, labour and expense. This decreased time to
confluency was probably due to an improved
polysulfone hollow fibre surface. Our surface was
hydrophobic since it was not coated with polyvinyl
pyrrolidone (PVP), and as a result the cells showed
easy and firm adhesion. Our hollow fibres had a
molecular weight cut-off of 6000 Da and wall thicknesses of 100 mm, and those by Humes et al. [10]
were 45 000–50 000 Da and 40–70 mm. The smaller
molecular weight cut-off with thicker-walled hollow
fibres can more efficiently shut out non-self proteins
and other unwanted particles from entering the
body of the patient, thus ensuring better safety in
future clinical uses. The inner diameter of 300 mm
A bioartificial renal tubule device using LLC-PK1 cells
Fig. 7. Scanning electron micrographs of hollow fibres retrieved
from the large modules on day 4 (A), day 8 (B) and day 13 (C).
Arrows indicate the hollow fibres. Cracks and fissures (arrowheads)
seen on the surface of the cell layer are artefacts that appear during
preparation of the samples. Scale bars ¼ 20 mm.
was wide enough for the culture medium to circulate uniformly in the hollow fibres and to supply
sufficient nutrients to the cells.
Substances may move across a membrane covered
with a cell layer according to three pathways: by
transport through the cell membranes (transcellular
route), by transport through the junctional spaces
between the cells (paracellular route), and by leakage
through areas where cells fail to cover. For an ideal
2205
bioartificial renal tubule, cells should completely cover
the intraluminal surface of the hollow fibres with
a confluent monolayer. This suppresses leakage to
a minimum and sustains this condition and function
for a long period. We assessed the confluency of the
cell layer through the movement of UN and Cr across
the hollow fibre membrane, as both were an indicator
of leakage. Urea is partly reabsorbed by renal tubules
although Cr is not reabsorbed under physiological
conditions [16]. In a previous study, we found that tight
junctions between the LLC-PK1 cells of a confluent
monolayer were well preserved when cultured on a
membrane filter (Transwell, Corning), and that there
was significantly less movement of UN and Cr from
the apical to the basolateral side of the cells under
15–20 mmHg hydrostatic pressure plus 100 mOsm
osmotic pressure for 2 h compared with the control
group without cells [12]. Inulin, which is not physiologically excreted or reabsorbed by renal tubule cells,
may be a more appropriate indicator of leakage.
However, the molecular weight of inulin (5000) was
as large as the molecular weight cut-off of our hollow
fibre membrane. Moreover, inulin leakage was as low
as 10% through the hollow fibre membrane without
cells in our preliminary study. Therefore, leakage of
relatively low molecular weight UN and Cr was
designated to represent the confluency status of cells
on the intraluminal surfaces of hollow fibres in the
current study.
Leakage and reabsorption through the hollow
fibre membrane were first studied using the small
modules (effective surface area 56 cm2), which were
made of the same materials as the large modules
(effective surface area 4000 cm2). The leakage of UN
and Cr, as well as the reabsorption of water,
glucose and sodium by LLC-PK1 cells in the small
module indicated that cell confluency was achieved
and the cells actively transported fluid and substances (Table 1). Moreover, after treatment with
HgCl2, an aquaporin inhibitor, reabsorption of
water decreased significantly to 40% of that
without HgCl2 treatment. To avoid toxic effects of
HgCl2 on living cells, the cells were pre-treated with
glutaraldehyde, which preserves water permeability
characteristics of urinary bladders and renal descending vasa recta [11]. In addition, in the presence
of ouabain, a Naþ,Kþ-ATPase inhibitor, reabsorption of sodium significantly decreased to 50% of
that without ouabain (Table 2). These results
reconfirmed that aquaporin-facilitated water reabsorption and sodium-dependent active transport by
LLC-PK1 cells occurred inside the small module
[13,14]. After achieving the confluent monolayer of
LLC-PK1 cells in the small module by the method
of cell seeding and the materials employed, we
scaled-up the large module studies while examining
possibilities of future clinical use.
The leakage of UN and Cr in large modules
with LLC-PK1 cells was significantly lower than in
modules without cells throughout the 2 week period
(Figure 2), indicating that cell confluency inside the
2206
large modules was achieved. There was, however,
some UN and Cr leakage (Figure 2 and Table 3).
Tight junctions between proximal tubule cells are
weaker than those between distal tubule cells, and it
was likely that UN and Cr diffused through the
intercellular pathways followed by the osmotic movement of water. Moreover, 50% of urea that is filtered
into renal tubules is passively reabsorbed in kidneys
under physiological conditions [16].
The reabsorption of water and glucose gradually
increased from day 3, reached a plateau on days 6–10,
and tended to decrease thereafter (Figures 3 and 4).
It has been previously shown that LLC-PK1 cells
reabsorb glucose by a SGLT as they became
confluent, and that phlorizin inhibited the SGLT
[17,18]. In agreement, we found that addition of
phlorizin to the medium perfusing the inner compartment caused a significant decrease in glucose
reabsorption from 260 to 125 mg/90 min (P<0.02)
(Figure 5), indicating that glucose reabsorption
by the LLC-PK1 cells lining the hollow fibres was
facilitated by SGLT. Sodium reabsorption gradually
increased and reached a plateau on day 8 (Figure 6).
These findings suggested that the expression of
Naþ, Kþ-ATPase increased in parallel with the
formation of tight junctions between the cells after
they became confluent. They further indicated that
after day 10, overgrowth of the cells appeared
to form multiple layers, resulting in the blockage of
substance transport. At that time, even the addition
of albumin failed to cause increases in sodium
reabsorption. Although immortalized cell lines such
as LLC-PK1 tend to exhibit overgrowth after
confluency, they have the advantage of higher
viability and expandability in in vitro cultures,
compared with primary cells. Ip et al. [15] reported
that LLC-PK1 cells cultured on membranes coated
with ECM containing the RGD (Arg-Gly-Asp) amino
acid sequence maintained their monolayer and
retained their ability for active glucose transport
over a long term. Here, we also coated the
intraluminal surfaces of the hollow fibres with
pronectin-F, an ECM containing the RGD sequence.
However, the cells became over-confluent and formed
multiple layers over time (Figure 7C). We believe that
this over-confluency produced the untoward effects
on the transport of water, glucose and sodium. These
adverse responses were more evident under the
influence of increased osmotic pressure. When albumin was added to the medium perfusing the outer
compartment, the reabsorptions of water, glucose and
sodium were significantly higher than that in the
group not given albumin on day 1 and/or day 3
(Figures 3, 4 and 6). However, this difference
narrowed with time, which was probably due to the
dense overgrowth of cells that partly blocked the
movement of fluid and substances, even under
the influence of osmotic pressure by day 10. The
use of tissue-engineered cells that have contact inhibition between cells may prevent overgrowth and thus
prolong the efficiency of this module.
N. Ozgen et al.
We calculated that this module with a surface area of
0.4 m2 could reabsorb 2.4 l of water/day when
albumin was added. The module with a surface area
of 1 m2 could reabsorb 6 l of water out of 10 l of
continuous haemofiltration/day, an amount that would
satisfy our preliminary requirements [9]. The insufficient reabsorption of 4 l could be covered by oral intake
of meals and drinks. Normal blood contains a greater
amount of proteins than was present in the albuminadded medium used in this study. Therefore, normal
blood would cause more fluid reabsorption due to high
colloid osmotic pressure in clinical use. Recently, we
demonstrated that aquaporin 1-transfected LLC-PK1
cells had approximately two times higher transcellular
osmotic water permeability than untransfected cells
[19]. Employment of these highly efficient transfected
cells may lead to availability of smaller portable
bioartificial renal tubule devices.
A number of clinical trials recently reported the
use of human renal proximal tubule cells in renal
tubule devices [20–22]. These devices, used for
treating patients with acute renal failure (ARF),
and patients with multiorgan failure and sepsis for a
maximum of 24 h, displayed satisfactory outcomes.
These reports have encouraged the development and
improvement of bioartificial renal tubule device for
clinical uses. In contrast with their limited critical
period for management of ARF, we intend to
develop bioartificial renal tubule devices that function for longer periods in chronic renal failure
patients. To our knowledge, this is the first in vitro
2 week continuous evaluation of a bioartificial renal
tubule device. We found that this device functioned
to reabsorb water, glucose and sodium for 10
days. In a previous study [23] we demonstrated that
a renal proximal tubule cell line played a key role
in the disposal of AGE, indicating that it may
prevent long-term complications in uraemic patients,
such as b2-microglobulin amyloidosis and atherosclerosis. Although not evaluated in this study, these
findings indicate that this bioartificial renal tubule
device may provide other metabolic functions of
renal tubules which are absent in haemodialysis
therapy. In combination with haemofiltration, this
device may perform most renal functions. With the
addition of necessary peripheral devices, such as
motors, batteries, pressure gauges and safety alarms,
this module may be used in portable settings.
Although some improvements will be necessary,
this artificial renal tubule device has potential for
patients with chronic renal failure to prevent the
complications of dialysis. The ability to perform
long-term therapy at their residence would also raise
quality of life.
Acknowledgements. The authors thank Mr Masayoshi Tokunaga
for excellent technical support on scanning electron microscopy.
This work was supported in part by a Grant-in-Aid for Scientific
Research (B, No. 2-12470213) from the Ministry of Education,
Culture, Sports, Science and Technology of Japan.
Conflict of interest statement. None declared.
A bioartificial renal tubule device using LLC-PK1 cells
2207
References
1. Van Stone JC, Bauer J, Carey J. The effect of dialysate sodium
concentration on body fluid distribution during hemodialysis.
Trans Am Soc Artif Intern Organs 1980; 26: 383–386
2. Owda A, Elhwairis H, Narra S, Towery H, Osama S.
Secondary hyperparathyroidism in chronic hemodialysis
patients: prevalence and race. Ren Fail 2003; 25: 595–602
3. Schwalbe S, Holzhauer M, Schaeffer J, Galanski M, Koch
KM, Floege J. Beta-2 microglobulin associated amyloidosis: a
vanishing complication of long-term hemodialysis? Kidney Int
1997; 52: 1077–1083
4. Charra B, Calemard E, Uzan M et al. Carpal tunnel syndrome,
shoulder pain and amyloid deposition in long-term hemodialysis patients. Proc Eur Dial Transplant Assoc 1984; 21:
291–295
5. Miyata T, Ueda Y, Yoshida A et al. Clearance of
pentosidine, an advanced glycation end product, by different
modalities of renal replacement therapy. Kidney Int 1997; 51:
880–887
6. Miyata T, Ueda Y, Yasuda Y et al. Accumulation of carbonyls
accelerates the formation of pentosidine, an advanced glycation
end product: carbonyl stress in uremia. J Am Soc Nephrol 1998;
9: 2349–2356
7. Miyata T, van Ypersele de Strihou C, Kurokawa K, Baynes
JW. Alterations in nonenzymatic biochemistry in uremia: origin
and significance of ‘carbonyl stress’ in long-term uremic
complications. Kidney Int 1999; 55: 389–399
8. Miyata T, Ishikawa S, Asahi K et al. 2-Isopropylidenehydrazono-4-oxo-thiazolidin-5-ylacetanilide (OPB-9195) treatment
inhibits the development of intimal thickening after balloon
injury of rat carotid artery: role of glycoxidation and
lipoxidation reactions in vascular tissue damage. FEBS Lett
1999; 445: 202–206
9. Saito A, Takagi D, Sugiura K et al. Maintaining low
concentrations of plasma b2-microglobulin through continuous slow haemofiltration. Nephrol Dial Transplant 1995;
10 [Suppl 3]: 52–56
10. Humes HD, MacKay SM, Funke AJ, Buffington DA. Tissue
engineering of a bioartificial renal tubule assist device: in vitro
transport and metabolic characteristics. Kidney Int 1999; 55:
2502–2514
11. Deen PMT, Nielsen S, Bindels RJM, van Os CH. Apical
and basolateral expression of Aquaporin-1 in transfected
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
MDCK and LLC-PK cells and functional evaluation of their
transcellular osmotic water permeabilities. Eur J Physiol 1997;
433: 780–787
Terashima M, Fujita Y, Sugano K et al. Evaluation of water
and electrolyte transport of tubular epithelial cells under
osmotic and hydraulic pressure for development of bioartificial
tubules. Artif Organs 2001; 25: 209–212
Rabito CA. Sodium cotransport processes in renal epithelial
cell lines. Miner Electrolyte Metab 1986; 12: 32–41
Saier MH Jr, Boerner P, Grenier FC et al. Sodium entry
pathways in renal epithelial cell lines. Miner Electrolyte Metab
1986; 12: 42–50
Ip TK, Galletti PM, Aebischer P. Effects of attachment
substrates on the growth and differentiation of LLC-PK1 cells.
In Vitro Cell Dev Biol 1990; 26: 162–168
Guyton AC, Hall JE. Urine formation by the kidneys In:
Human Physiology and Mechanisms of Disease. WB Saunders
Company, Philadelphia, PA: 1997; 223–235
Mullin JM, Kleinzeller A. Sugar transport in the renal
epithelial cell culture. In: Taub M, ed. Tissue Culture of
Epithelial Cells. Plenum, New York, NY: 1985; 71–86
Misfeldt DS, Sanders MJ. Transepithelial transport in cell
culture and stoichiometry of Na/phlorizin binding and
Na/D-glucose cotransport. A two-step, two-sodium model of
binding and translocation. J Membr Biol 1983; 70: 191–198
Fujita Y, Terashima M, Kakuta T et al. Transcellular water
transport and stability of expression in aquaporin 1-transfected
LLC-PK1 cells in the development of a portable bioartificial
renal tubule device. Tissue Eng, 2004; 10: 711–722
Weitzel WF, Fissell WH, Humes HD. Initial clinical experience
with a human proximal tubule cell renal assist device (RAD). J
Am Soc Nephrol 2001; 12: 279A
Humes HD, Weitzel WF, Swaniker FC, Bartlett RH. Effect of
renal cell therapy on the gene expression profiles of
inflammation in patients with acute renal failure (ARF)
and multiorgan failure (MOF). J Am Soc Nephrol 2002; 13:
445A
Weitzel WF, Humes HD, Bartlett RH, Swaniker FC,
Paganini E. Early results with the bioartificial kidney in ICU
patients with acute renal failure. J Am Soc Nephrol 2002; 13:
642A
Asano M, Fujita Y, Ueda Y et al. Renal proximal tubular
metabolism of protein-linked pentosidine, an advanced glycation end product. Nephron 2002; 91: 688–694
Received for publication: 25.12.03
Accepted in revised form: 24.3.04