Transcript
Biomaterials 32 (2011) 1229e1241
Contents lists available at ScienceDirect
Biomaterials journal homepage: www.elsevier.com/locate/biomaterials
A robust high-throughput sandwich cell-based drug screening platform Shufang Zhang a, b, c,1, Wenhao Tong a, c,1, Baixue Zheng a, d, Thomas A.K. Susanto e, Lei Xia a, f, Chi Zhang a, c, Abhishek Ananthanarayanan a, c, Xiaoye Tuo a, d, g, Rashidah B. Sakban f, Ruirui Jia f, Ciprian Iliescu a, Kah-Hin Chai h, Michael McMillian i, Shali Shen f, Hwaliang Leo a, e, *, Hanry Yu a, c, f, j, k, l, ** a
Institute of Bioengineering and Nanotechnology, A*STAR, The Nanos, #04-01, 31 Biopolis Way, Singapore 138669, Singapore Center for Stem Cell and Tissue Engineering, School of Medicine, Zi Jing Gang Campus, Zhejiang University, #388 Yu Hang Tang Road, Hangzhou 310058, China c NUS Graduate School for Integrative Sciences and Engineering, Centre for Life Sciences (CeLS), #05-01, 28 Medical Drive, Singapore 117456, Singapore d Singapore-MIT Alliance, National University of Singapore, E4 #04-10, 4 Engineering Drive 3, Singapore 117576, Singapore e Division of Bioengineering, National University of Singapore, 7 Engineering Drive, E3A #04-15, Singapore 117574, Singapore f Department of Physiology, Yong Loo Lin School of Medicine, National University of Singapore, #03-03, 2 Medical Drive, Singapore 117597, Singapore g Department of Burns and Plastic Surgery, The First Affiliated Hospital of PLA General Hospital, Beijing 100037, China h Industrial and Systems Engineering Department, National University of Singapore, 1 Engineering Drive 2, Singapore 117576, Singapore i Johnson & Johnson Pharmaceutical Research & Development, L.L.C., 920 Route 202, Raritan, NJ 08869, USA j NUS Tissue-Engineering Programme, DSO Labs, National University of Singapore, Singapore 117597, Singapore k Mechanobiology Institute, Temasek Laboratories, National University of Singapore, #05-01, 5A Engineering Drive 1, Singapore 117411, Singapore l Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 8 September 2010 Accepted 29 September 2010 Available online 23 October 2010
Hepatotoxicity evaluation of pharmaceutical lead compounds in early stages of drug development has drawn increasing attention. Sandwiched hepatocytes exhibiting stable functions in culture represent a standard model for hepatotoxicity testing of drugs. We have developed a robust and high-throughput hepatotoxicity testing platform based on the sandwiched hepatocytes for drug screening. The platform involves a galactosylated microfabricated membrane sandwich to support cellular function through uniform and efficient mass transfer while protecting cells from excessive shear. Perfusion bioreactor further enhances mass transfer and cellular functions over long period; and hepatoctyes are readily transferred to 96-well plate for high-throughput robotic liquid handling. The bioreactor design and perfusion flow rate are optimized by computational fluid dynamics simulation and experimentally. The cultured hepatocytes preserved 3D cell morphology, urea production and cytochrome p450 activity of the mature hepatocytes for 14 days. When the perfusion-cultured sandwich is transferred to 96-well plate for drug testing, the hepatocytes exhibited improved drug sensitivity and low variability in hepatotoxicity responses amongst cells transferred from different dates of perfusion culture. The platform enables robust high-throughput screening of drug candidates. Ó 2010 Elsevier Ltd. All rights reserved.
Keywords: Hepatotoxicity Drug screening Bioreactor Hepatocytes High-throughput Microfabrication
1. Introduction In vitro hepatocyte culture models are gaining increasing attention from the pharmaceutical industry for early stage drug toxicity screening [1]. Despite the advent of many in vitro liver models [2,3], adaptation of these models into industry-scale drug
* Corresponding author. National University of Singapore, Division of Bioengineering, 7 Engineering Drive, E3A #04-15, Singapore 117574, Singapore. Tel.: þ65 6516 5608; fax: þ65 6872 3069. ** Corresponding author. National University of Singapore, Physiology, YLL School of Medicine, NUHS, Singapore. Tel.: þ65 6516 3466; fax: þ65 6874 8261. E-mail addresses:
[email protected] (H. Leo),
[email protected] (H. Yu). 1 These authors contribute equally to this work. 0142-9612/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2010.09.064
testing application still faces many challenges. Beside re-establishing hepatocyte in vivo functions, the ideal in vitro liver model for industry-scale drug screening demands high reproducibility with minimal variations in screening results, high-throughput and automated processing capability [4]. Among the various in vitro models, such as perfusion liver slices [5], microsomes [6], cell lines [7] etc, isolated primary hepatocytes are the preferred model as they strike a balance between throughput and basic cellular architecture and functions [8]; but they lose differentiated functions rapidly under conventional culture configuration [9]. Isolated hepatocytes can be sandwiched between two layers of extracellular matrices (e.g. collagen) [10,11] to reestablish the differentiated hepatocyte functions such as urea and albumin secretion [11], biotransformation enzyme functions [12], polarity and transporter
1230
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
activities [13]; making it the model recommended by the US Food and Drug Administration for in vitro drug metabolism and toxicity studies [14]. Still, collagen sandwich culture suffers from batch to batch collagen variations [15], variable drug absorptions [16] and poor mass transfer across collagen gel especially in static culture over extended culture period [17]. Each of these factors contributes to variable drug toxicity measurements. Adapting collagen sandwich culture into perfusion system for drug testing is difficult because collagen gel overlaying the hepatocytes can be washed off by perfusion flow, resulting in increased drug access variability and possibly in the shear stress experienced by the hepatocytes. Perfusion culture can maintain and improve cell viability and metabolic activities of cultured hepatocytes [18] due to improved oxygen or nutrient delivery and waste removal [19]. Many hepatocyte perfusion culture systems developed [2,20,21] are low throughput [22] as they are not compatible with the industryscale multi-well plates for high-throughput drug screening. Multiwell plate perfusion culture systems that conform to standard multi-well culture plate dimension developed previously [22,23] rely upon online drug administration in fluidic flow that compromises on controls of shear stress, cellular functions and throughput. Here we report a high-throughput and robust druginduced hepatotoxicity screening platform (RoboTox) using perfusion sandwich-cultured hepatocytes. 2. Materials and methods 2.1. Materials 100 mm thick biaxially oriented polyethylene terephthalate (PET) films were purchased from Goodfellow Inc. (Cambridge, UK). The galactose ligand, 1-O-(60 -aminohexyl)-D-galactopyranoside (AHG, M.W. 279) was synthesized as described previously [24,25] and verified by NMR spectrum. All other chemicals were purchased from Sigma-Aldrich (Singapore) unless otherwise stated. Low stress porous Si3N4 membrane was designed, fabricated, surface functionalized and characterized as described previously [26,27]. 2.2. Primary rat hepatocyte isolation and culture Hepatocytes were harvested from male Wistar rats weighing 250e300 g by a two-step in situ collagenase perfusion method [28]. Animals were handled according to the Institutional Animal Care and Use Committee (IACUC) protocol approved by the IACUC of the National University of Singapore. Only hepatocyte with viability of >85% as determined by Trypan Blue exclusion assay was used. Freshly isolated hepatocytes were seeded onto collagen-coated or galactoseimmobilized PET films at 1 105 cells/cm2 in a 96-well plate. The cells were cultured in William’s E medium supplemented with 10 mM NaHCO3, 1 mg/ml BSA, 10 ng/ml EGF, 0.5 mg/ml insulin, 5 nM dexamethasone, 50 ng/ml linoleic acid, 100 units/ml pencillin and 100 mg/ml streptomycin; and incubated with 37 C, 5% CO2, 95% humidity for 24 h. Collagen sandwich was assembled by placing a collagen-coated polycarbonate membrane (IsoporeÔ, Milipore, USA) on top of the hepatocytes seeded on collagen-coated PET film. Si3N4 sandwich culture was assembled by placing galactose-immobilized porous Si3N4 membrane (20 mm pore size and 20 mm inter-pore distance, corresponding to 20% porosity) on top of the cells on galactoseimmobilized PET films. 2.3. RoboTox platform Hepatocytes were first seeded on galactose-immobilized PET film for 24 h, and then overlaid with galactose-immobilized porous Si3N4 membrane (Fig. 1A). Hepatocytes were cultured in this Si3N4 sandwich culture configuration for another 24 h before transferring into the RoboTox bioreactor. The bioreactor comprises of three subunits: an upper lock plate, a middle sieve plate and a lower perfusion bioreactor. It was machined to dimension from a polycarbonate block and sealed with o-rings and screws (Fig. 1B). The lower perfusion bioreactor was fabricated according to the dimension of a typical 96-well plate, except the wells are interconnected in series by a 3 mm diameter fluid channel. The wells of the middle sieve plate are slightly smaller than the wells of lower perfusion bioreactor so that they can fit into the bioreactor and conventional 96-well plate. The middle sieve plate holds the Si3N4 sandwich and facilitates easy transfer of the cells between the perfusion bioreactor and standard 96-well plate. The upper lock plate provides uniform pressure to seal the entire assembly. To improve the delivery of oxygen which is important for hepatocyte culture [29], circular openings corresponding to the bioreactor wells were drilled into the upper lock plate. An oxygen-permeable
membrane (Breathe-EasyÒ, Diversified Biotech, USA) was placed between the sieve plate and the upper lock plate to preserve the closed circuit while allow adequate oxygen diffusion at the same time. The bioreactor was connected to a recirculating closed perfusion loop so that it is isolated from the external environment (Fig. 1C). The perfusion loop consists of RoboTox bioreactor, medium reservoir, peristaltic pump (Ismatec SA, Switzerland), three-direction valves (Upchurch Scientific, USA), stopping valves (Upchurch Scientific, USA), connectors (Upchurch Scientific, USA) and oxygen-permeable silicone tubing (Ismatec SA, Switzerland), and placed in a 37 C incubator with 5% CO2 and 95% humidity. Online sampling and monitoring was achieved via three-direction valve. For high-throughput drug testing, the middle sieve plate holding the perfusion-cultured Si3N4 sandwiches were removed from the bioreactor and placed into the conventional 96-well plate for multiplex drug testing using robotic liquid handler (Fig. 1D). 2.4. Fluid flow modeling Finite Element Analysis (FEA) software, COMSOL Multiphysics 3.5a (COMSOL Inc., Burlington, MA) was used for the fluid dynamic simulations. A 3D model which represented the fluid body containing oxygen and hepatocytes was constructed. Customized meshing was used and resulted in 744578 tetrahedral elements. The model consisted of one subdomain in where the steady state incompressible NavierStokes application mode and the steady state convection and diffusion application mode were applied to simulate for the fluid motion and the oxygen mass transfer. The characteristic parameters used to define the models are summarized in Table 1. The governing equations for the subdomain are
rðu$VÞu ¼ Vr þ hV2 u V$ðDVcÞ ¼ u$Vc where u½m2 $s1 denotes the fluid velocity field and V is the standard del (nabla) operator. A fully developed flow condition was applied at the inlet boundary with oxygen partial pressure set at the atmospheric pressure. Outlet boundary conditions were zero pressure outflow and convective-only oxygen transport. The fluid parts of the bioreactor which are exposed to air were modeled as boundaries with the slip wall condition having atmospheric oxygen partial pressure. The no-slip wall boundary conditions were used along the wall of the model. The hepatocytes in the bioreactor were assumed to be uniformly distributed on the bases of the wells. They were represented as boundaries with oxygen out-flux condition. The magnitude of the flux is in accordance to the Oxygen Uptake Rate (OUR) of the hepatocytes as described in the equation c
OUR ¼ Vm c a a þ Km where c is the local oxygen concentration. This equation follows the MichaeliseMenten kinetics and has been widely used as a mathematical model to represents Oxygen Uptake Rate of many cells including hepatocytes [30]. 2.5. Mass transfer efficiency measurement Mass transfer efficiency was measured by limited diffusion of cell labeling agent CellTrackerÔ Green (Invitrogen, USA). Hepatocytes were first thoroughly labeled with 20 mM CellTrackerÔ Orange (Invitrogen, USA) in culture media. The media was replaced with 2 mM of CellTrackerÔ Green under static and perfusion condition for 2 h. The cells were rinsed with phosphate buffered saline (PBS) and fixed with 3.7% paraformaldehyde. Z-stack images from three independent experiments were taken with confocal microscope (Fluoview 300, Olympus, Japan). Quantification of the labeled cells was performed by Matlab (R2009a). After removing noise using a lowpass filter, z-stacks of binary masks were created by thresholding the red channel images. Zero values in the mask represent background; while red signals represent the space occupied by cells. The total cell area for one z-stack was defined to be the total number of all positive pixels in the corresponding masks. Mass transfer was represented by total intensity in the green channel. Total intensity was defined as the sum of all pixel intensities in the positive areas of the corresponding green channel. The mass transfer efficiency was calculated by total intensity/total cell area. 2.6. Cell viability measurement Cell viabilities of primary rat hepatocytes cultured in RoboTox system for 6 days at the flow rates of 0.1, 0.06, 0.03, 0.015 ml/min were assessed by live and dead staining of the cells with Calcein AM (Molecular Probes, USA) and propidium iodide (PI). Cells were washed with PBS and incubated with 5 mM of Calcein AM and 25 mg/ml of PI in culture medium at 37 C for 30 min. Cells were then washed with PBS, placed in fluorescent mounting medium (Dako, Denmark) and viewed by confocal microscopy (Fluoview 300, Olympus, Japan). Cell viability was also measured by CellTiter 96Ò AQueous One Solution Cell Proliferation Assay (MTS Assay, Promega, USA).
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
1231
Fig. 1. Schematic diagram of RoboTox platform for hepatocytes sandwich culture and drug testing. (A) Hepatocytes sandwich culture was assembled by sandwiching primary hepatocytes between silicon nitride (Si3N4) membranes and galactosylated PET films. (B) After 24 h, the hepatocytes sandwiches were transferred into RoboTox bioreactor with an upper lock plate, membrane oxygenator, middle sieve plate and lower perfusion chamber. The Si3N4 sandwiches were placed in the wells of the sieve plate that facilitates easy
1232
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
Table 1 Characteristic parameters used to define fluid flow model. Parameters
Value
Description
D Km
2.88 109 m2/s 5.6 mmHg
a
1.19 nmol cm3 mm Hg1
Vm
0.27 nmol s1 (million cells)1 0.6915 103 kg m1s1 1000 kg/m3
Medium oxygen diffusivity [67] Half-saturation constant for monolayer rat hepatocytes [68] Solubility of oxygen in plasma at 37 C [69] Maximum rat hepatocytes oxygen consumption rate [70] Medium viscosity [71] Medium density
h r
2.7. Hepatocyte function measurement Urea production in the culture media was measured using the Urea Nitrogen Kit (Stanbio Laboratory, Boerne, Texas). CYP450 enzymatic activity was measured via 3-cyano-7-hydroxycoumarin (CHC) production upon incubation with metabolic substrate 3-cyano-7-ethoxycoumarin (CEC). 50 mM of CEC in culture medium was added to static and perfusion culture. After incubation for 4 h, the medium was collected and the product measured by capillary electrophoresis method developed previously [31]. All functional data are normalized to 106 cells. 2.8. Drug-induced hepatotoxicity testing Hepatocytes were treated with acetaminophen (APAP) or diclofenac to assess their responses to drug-induced hepatotoxicity. Drug stock solution dissolved in DMSO was diluted with culture medium to different concentrations with the final concentration of DMSO in the medium was kept at less than 1%. Hepatocytes were exposed to drugs for 48 h and the cell viability measured by MTS assay. For IC50 measurement, hepatocytes were treated with APAP, diclofenac, ketoconazole, chlorpromazine, flutamide or quinidine for 24 h and cell viability was measured using MTS assay. Robotic liquid handling system (Fig. 1D; customized by Bio Laboratories, Singapore) was used to perform drug testing. Detailed configuration of the system and protocol are described in Supplementary materials and methods. 2.9. Statistical methods Data from at least two independent experiments were analyzed and values were represented as mean standard error of means (s.e.m.). The Student t-test was used to analyze the statistical significance of specific pairs of the data. Result with a p value less than 0.05 were considered statistically significant. One-way ANOVA test was used to analyze the significance of differences among multiple groups of data. P value less than 0.05 was considered significantly different.
3. Results 3.1. Design and optimization of RoboTox platform The RoboTox platform consists of Si3N4 sandwich culture (Fig. 1A), RoboTox bioreactor (Fig. 1B), recirculating closed perfusion loop (Fig. 1C) and transferring of cells to 96-well plate and robotic liquid handler for drug testing (Fig. 1D). The Si3N4 membranes used for the Si3N4 sandwich cultures have uniform pores for efficient mass transfer and are conjugated with galactose ligands for interaction with asialoglycoprotein receptor on hepatocytes to enhance cellular functions [26,32]. The Si3N4 sandwich cultures were placed in the wells of middle sieve plate with defined distance from the fluid flow channel to minimize shear stress on hepatocytes. The middle sieve plate held the Si3N4 sandwich cultures and facilitated easy transfer of the cells from the perfusion bioreactor to standard 96-well plate for drug testing. The platform was designed to adhere to standard well plate dimension for compatibility with commercial robotic liquid handling operations. The perfusion culture condition of the RoboTox platform was optimized using computational fluid dynamics (CFD) and experiments. Increasing the perfusion flow rate enhances the delivery of
nutrients and removal of waste, but too high of a flow rate may induce excessive wall shear stress which is detrimental to the hepatocytes [33]. To select the optimal flow rate, computational fluid dynamics (CFD) models were developed to evaluate the flow velocity, wall shear stress and oxygen tension in the bioreactor. Four different flow conditions (0.1, 0.06, 0.03, and 0.015 ml/min) were simulated. Simulation shows that the flow velocity was significantly lower and relatively more uniform at the bottom of the well as compared to the flow velocity in the fluid flow channel (Fig. 2A). Consequently, fluid-induced shear stress at the bottom of the well, where hepatocytes were seeded, was minimized. At the cell seeding regions, the maximum wall shear stress was found around the center (Fig. 2B), with a value of 0.046 mPa at 0.1 ml/min, 0.028 mPa at 0.06 ml/min, 0.014 mPa at 0.03 ml/min, and 0.007 mPa at 0.015 ml/min. All four values were much lower than the critical values of 33 mPa that can be tolerated by hepatocytes as previously reported [33]. The oxygen tension is one of the most important parameter for hepatocytes culture [29]. Hepatocytes are highly dependent on oxidative phosphorylation to generate energy for metabolism [34], but excessive oxygen tension stimulates free radicals production and deteriorates the cellular functions [35]. Our simulation shows that the pericellular oxygen tension is the highest at the flow rate of 0.1 ml/min, averaging 110.27 mmHg. This is followed by 101.97 mmHg, 88.20 mmHg and 72.65 mmHg, at the flow rate of 0.06 ml/min, 0.03 ml/min and 0.015 ml/min respectively (Table 2). This suggests that the flow rate of 0.015 ml/min is probably the best condition as its oxygen tension is closer to the physiological pericellular oxygen tension of hepatocytes of w20 mmHge65 mmHg [36,37]. To optimize the flow rate experimentally, the viability of the hepatocytes cultured under different flow rates (0.015, 0.03, 0.06, 0.1 ml/min) was compared after 8 days of culture. With the exception of the hepatocytes perfused at the high flow rate of 0.1 ml/min, hepatocytes perfused at other flow rates remained viable after 8 days of culture (Fig. 2C). Urea production of hepatocytes cultured under the viable flow rates (0.015, 0.03, 0.06 ml/min) during the 8-day culture period was measured to optimize flow rate for hepatocyte functions (Fig. 2D). Urea production increased significantly after perfusion initiation (day 4-day 8, >30 mg/million cells/day) compared to urea production before perfusion initiation (day 2, <10 mg/million cells/day) for all flow conditions. Urea production for hepatocytes cultured using the flow rate of 0.015 ml/min was consistently the highest among the three flow conditions throughout the culture period. This was followed by hepatocytes perfused at 0.03 ml/min. Hepatocytes perfused at 0.06 ml/min showed the lowest urea production among the three flow conditions. 0.015 ml/min was chosen as the optimal flow rate used in all subsequent experiments.
3.2. Mass transfer efficiency in the optimized RoboTox platform We compared the mass transfer efficiency between the culture medium and the sandwiched hepatocytes cultured in collagen sandwich, Si3N4 sandwich and RoboTox perfusion platform. The cell labeling dye (2 mM CellTracker Green) was introduced into the culture medium for 2 h and the degree of cell labeling (fluorescence intensity) was quantified as a measure of the mass transfer efficiency (Fig. 3). Hepatocytes cultured in collagen sandwich exhibited the least fluorescence intensity (ie. lowest mass transfer efficiency) as the dye penetration might be partially blocked by the collagen gel.
transfer to standard 96-well plate. (C) The bioreactor was connected to a recirculating closed perfusion loop for perfusion culture of hepatocytes. (D) After perfusion culture, the middle sieve plate holding the sandwich cultures was disassembled from the bioreactor and placed on a standard 96-well plate for drug testing using a robotic liquid handler that was customized as shown, including a biological safety cabinet, robotic arm, pipettor, fluorescence and absorbance reader, incubator and other accessories.
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
1233
Fig. 2. Flow rate optimization of RoboTox platform. The computational fluid dynamics simulated distribution of (A) flow velocity and (B) wall shear stress in RoboTox platform shows that the flow rate of 0.015 ml/min exhibited the most uniform flow velocity throughout the well, with the lowest and most uniformly distributed wall shear, compared to other flow rates. (C) Live (Calcein AM, green) and dead (PI, red) staining indicates that the primary hepatocytes were viable at the flow rate of 0.06, 0.03 and 0.015 ml/min, but not at 0.1 ml/min after 8 days of culture in RoboTox platform. (D) The urea production of hepatocytes cultured at the flow rate of 0.015 ml/min was significantly higher than others throughout 8 days of culture. -: 0.06 ml/min, : 0.03 ml/min, : 0.015 ml/min. Data plotted represent the mean s.e.m of 3 independent experiments. *: p > 0.05.
1234
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
3.3. Long term 3D morphology and functional maintenance of hepatocytes
Table 2 Pericellular oxygen tension in RoboTox platform with different flow rate. Flow rate
Maximum (mmHg)
Minimum (mmHg)
Average (mmHg)
0.1 min/ml 0.06 min/ml 0.03 min/ml 0.015 min/ml
135.2 130.0 121.3 110.4
82.7 72.7 57.8 44.4
110.3 101.9 88.2 72.7
Physiological pericellular oxygen tension is between 20 mmHg to 65 mmHg [36,37].
Si3N4 sandwiched hepatocytes in static culture showed higher mass transfer efficiency than the cells in collagen sandwich [26]. The hepatocytes cultured in RoboTox platform exhibited the highest mass transfer efficiency among the three configurations. Quantification of the fluorescence intensity revealed that mass transfer efficiency was 2.5 times higher in Si3N4 sandwich static culture; and 5.5 times higher in RoboTox culture compared to the collagen sandwich culture control.
A
i.
The F-actin distribution of hepatocytes cultured in collagen sandwich, Si3N4 sandwich and RoboTox, versus the collagen-coated coverslip (2D monolayer) were compared after 14 days of culture (Fig. 4A). Cortical F-actin concentrated at the cell edge in collagen sandwich, Si3N4 sandwich and RoboTox cultures confirm the preservation of 3D cell morphology in contrast to the collagen-coated 2D monolayer culture with extensive F-actin stress fibers. The similarity of F-actin distribution of hepatocytes in RoboTox to those of Si3N4 sandwich and collagen sandwich supports that perfusion flow does not adversely affect the 3D hepatocyte morphology. Urea production of primary hepatocytes culture in RoboTox was significantly higher (140e200 mg/million cells/day) than urea production of hepatocytes in collagen sandwich and Si3N4 sandwich in static culture (<50 mg/million cells/day, Fig. 4B). The urea production of RoboTox-cultured hepatocytes at day 14 was maintained at a level similar to earlier culture days (day 4eday 8), albeit
iii.
ii.
Fluorescence intensity/ cell area
B
1000
*
*
800
600
400
200
0 Collagen Sandwich
Si3N4 Sandwich
RoboTox
Fig. 3. Higher mass transfer efficiency in RoboTox platform. (A) Confocal images of (i) collagen sandwich, (ii) Si3N4 sandwich and (iii) RoboTox-cultured hepatocytes incubated with 2 mM CellTrackerÔ Green for 2 h shows that fluorescence intensity was the highest in RoboTox-cultured hepatocytes, indicating the highest mass transfer efficiency among the three culture configurations. (B) The quantification of fluorescence intensity shows that mass transfer efficiency was 2.5 times higher in Si3N4 sandwich and 5.5 times higher in RoboTox culture hepatocytes compared to collagen sandwich. Data plotted represent the mean s.e.m of 3 independent experiments. *: p < 0.05.
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
1235
a slight drop was observed on days 10 and 12. In comparison, the urea production level of hepatocytes cultured in collagen sandwich and Si3N4 sandwich dropped irreversibly from day 10 onwards. Cytochrome p450 enzymes are involved in the phase I metabolism of drugs and were measured for the highly fluorescent 3-cyano-7-hydroxycoumarin (CHC) production upon incubation with metabolic substrate 3-cyano-7-ethoxycoumarin (CEC) by dealkylation of CEC [6,38]. The CHC production of hepatocytes in RoboTox started at 35 mmol/million cells on days 4 and 6 of culture and increased to 80 mmol/million cells from day 8 (Fig. 4C). The high level of CHC production was maintained at similar level till day 14. In comparison, the CHC productions from collagen sandwich and Si3N4 sandwich in static culture were much lower (<20 mmol/ million cells) and fluctuated throughout the 14 days culture period. CYP1A1 and CYP1A2 enzymatic activities therefore were comparatively more stable and maintained at higher level in RoboTox. The F-actin staining, urea production and CHC conversion results show that hepatocytes cultured in RoboTox maintained 3D morphology and hepatocytes differentiated functions over a longer period. 3.4. Higher sensitivity and reduced variation in hepatotoxicity responses Different concentrations of APAP or diclofenac were added into the hepatocytes cultured in collagen sandwich, Si3N4 sandwich and RoboTox for 48 h and cell viability were evaluated by MTS assay. For APAP treated cultures, the viability of hepatocytes in RoboTox were lower than those in the collagen sandwich and Si3N4 sandwich in static conditions, especially at high concentrations (40 mM) of APAP (p < 0.05) (Fig. 5A). For diclofenac treated cultures, the viability of hepatocytes in RoboTox were lower than those in the collagen sandwich and Si3N4 sandwich in static conditions for both 0.3 and 0.6 mM (p < 0.05) (Fig. 5B). These drug treatment results show that hepatocytes in RoboTox exhibit higher sensitivity towards druginduced hepatotoxicity than hepatocytes in collagen sandwich and Si3N4 sandwich in static conditions when exposed to the same concentration of drugs. We also compare the drug-induced hepatotoxicity response using hepatocytes from different days of culture to investigate the variation of drug screening response throughout long term culture period. Hepatocytes cultured in RoboTox were treated with APAP on day 4, day 8 or day 14 and cell viability was assessed. The cell viability of RoboTox-cultured hepatocytes after APAP treatment on different days was comparable (Fig. 6), indicating consistent response to drug-induced hepatotoxicity throughout 14-day culture period. In comparison, cell viabilities in collagen sandwich and Si3N4 sandwich culture varied when 40 mM APAP was treated, indicating significant variation towards APAP-induced hepatotoxicity throughout the 2-week culture period. 3.5. High-throughput hepatotoxicity testing with robotic liquid handler Since the RoboTox system is designed for parallel highthroughput drug testing, the uniformity of the cells cultured across different wells in the bioreactor is important as the data from each well has to be cross-compared. In the bioreactor, 8 wells are connected in series for each perfusion loop. To investigate the uniformity of nutrient access among the serially-connected wells, cell
Fig. 4. Long-term 3D morphology and functional maintenance of hepatocytes in RoboTox platform. (A) Cortical F-actin was observed at cell periphery with little stress fibers formation for hepatocytes cultured in (ii) collagen sandwich (iii) Si3N4 sandwich and (iv) RoboTox, but not (i) collagen monolayer after 14 days of culture. (B) Urea
production and (C) p450 enzymatic activity, as indicated by CHC production, were significant higher in RoboTox-cultured hepatocytes compared to collagen sandwich and Si3N4 sandwich-cultured hepatocytes and maintained over 14 days of culture. -: collagen sandwich, ,: Si3N4 sandwich, : RoboTox. Data plotted represent the mean s.e.m of 3 independent experiments. *: p < 0.05.
1236
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
Fig. 5. Improved drug sensitivity exhibited by RoboTox-cultured hepatocytes. Higher drug sensitivity was displayed by RoboTox-cultured hepatocytes when dosed with different concentration of (A) APAP and (B) Diclofenac. -: collagen sandwich, ,: Si3N4 sandwich, : RoboTox. Data plotted represent the mean s.e.m of 3 independent experiments. *: p < 0.05.
labeling dye was allowed to perfuse through the system for 2 h, and images of cells from each well were acquired and analyzed. Fluorescence intensity per cell area among the cells from different wells show no significant difference (p value ¼ 0.20476; Fig. 7A). We also measured the viability of hepatocytes cultured in different wells of the RoboTox for 2 days using MTS assay. The p value from the oneway ANOVA analysis is 0.33196, indicating that there is no significant difference in cell viability for cells from different wells (Fig. 7B). To investigate the operational compatibility of RoboTox for highthroughput drug testing with robotic liquid handling system, the IC50 values of 6 paradigm compounds (APAP, ketoconazole, diclofenac, chlorpromazine, flutamide and quinidine) were evaluated. The drugs were chosen on the criteria that they are well characterized and they elicit toxicity by different mechanisms and are metabolized by different cytochrome p450 (Table 3) [39]. The IC50 values obtained from RoboTox was compared to those from collagen sandwich culture [16] and with the reported IC50 values from literatures (Table 3). The IC50 values are in agreement with
Fig. 6. Reduced variation in APAP-induced hepatotoxicity response for RoboToxcultured hepatocytes on different days of culture. Hepatocytes cultured in collagen sandwich, Si3N4 sandwich and RoboTox were exposed to (A) 10 mM APAP and (B) 40 mM APAP on days 4, 8 and 14 of culture. RoboTox-cultured hepatocytes exhibit similar hepatotoxicity response to 40 mM APAP on different days of culture; whereas hepatotoxicity responses for collagen sandwich and Si3N4 sandwich-cultured hepatocytes varied when 40 mM APAP was administered to cells from different days of culture. -: day 4, ,: day 8, : day 14. Data plotted represent the mean s.e.m of 3 independent experiments. *: p < 0.05.
those reported in the literatures and the values from RoboTox are generally the lowest indicating high drug sensitivity of the hepatocytes cultured in RoboTox (Fig. 8). 4. Discussion Our results demonstrate a robust, high-throughput drug screening platform that potentially meet industry-scale drug screening needs. In vitro liver model for drug screening has different set of criteria compared to liver tissue engineering for other purposes. In addition to good cellular functions, the liver tissue culture must be handled in large quantity in parallel (ie. high-throughput), and in an efficient and reproducible manner (ie. robust) [4]. We used the sandwich culture configuration to reestablish cellular functions of isolated hepatocytes. Sandwich culture has been shown to reestablish and maintain xenobiotic biotransformation capacity which is crucial for drug testing [11]. It also forms and maintains the cell polarity [10] which is important for drug-transporter interactions
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
Table 3 IC50 values measured using RoboTox, collagen sandwich culture and published IC50 values for 6 paradigm hepatotoxic drugs and their reported mechanisms of action.
1100 1000 900 800 700 600 500 400 300 200 100 0
Drugs
RoboTox Collagen Published Mechanism sandwich IC50 (mm) of action (mm) (mm)
APAP
7000
30000
25
70
180
400
Chlorpromazine
23
110
Flutamide
30
90
75
Quinidine
170
250
457
Ketoconazole
8 ut le t O
7
6
5
4
3
2
1
Diclofenac
In le t
Fluorescence intensity/ cell area
A
1237
14000 62.3 263 45.5
Formation of NAPQI mediated by CYP2e1 [52] Oxidative stress by glutathione depletion [72] Formation of benzoquinone and imine by CYP2c [73] T-cell mediated killing, mitochondrial balance and CYP1a2 inducer [39] Formation of 4-hydroxy radical mediated by CYP1a2 [74] Hepatitis and formation of sulphate radical mediated by CYP3a4 [75]
Well in series
t le
8
ut
7
6
5
4
3
2
1
O
In
le
t
Viability
B
1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
Well in series Fig. 7. High uniformity among hepatocytes cultured across serially-connected wells in RoboTox system. The uniformity of (A) mass transfer efficiency and (B) cell viability of hepatocytes cultured in 8 serially-connected wells were evaluated. No statistical significance difference was observed for cells cultured in different wells in terms of mass transfer efficiency (One-way ANOVA, p ¼ 0.20476) and cell viability (One-way ANOVA, p ¼ 0.33196). Data plotted are the mean s.e.m of 3 independent experiments.
investigations. However, the batch to batch variation of collagen isolation [40], intrinsic instability of collagen gel [24], uneven drug absorption and drug access of the collagen gel [16] lead to increase in drug testing variations. To create a robust system, we minimized such variations by replacing overlaying collagen gel with galactosylated microfabricated silicon nitride (Si3N4) membranes. Silicon nitride membrane has good chemical inertness, high fracture toughness and high wear resistance [41], allowing it to stay intact in the perfusion culture. This is important to stabilize the sandwich culture in the perfusion system and provide secure shield for hepatocytes from perfusion shear stress. The membrane was fabricated using Micro-Electro-Mechanical Systems (MEMS) technology which is highly reproducible with micrometer scale precision [42]. The fabricated membrane is ultra-thin (3 mm), highly porous (20% porosity) and has uniformly distributed pores, which facilitates efficient and uniform mass transfer while minimizes drug absorption. Galactose ligands were stably conjugated to PET films and Si3N4 membranes via chemical reactions [43,44], maintaining the differentiated hepatocytes functions via interactions with asialoglycoprotein receptors [24,45] throughout the perfusion culture periods.
The perfusion culture system further enhances hepatocyte functions; specifically, CYP1A1 and CYP1A2 activities were significantly higher for hepatocytes cultured in RoboTox platform compared to those in static culture (Fig. 4C). CYP1A1 and CYP1A2 are part of the cytochrome p450 (CYP450) family, which are involved in drug metabolism [46] by converting relatively nonpolar compound to polar compound for excretion. Some of the intermediate metabolites from drugs metabolism are chemically reactive and play a key role in drug-induced hepatotoxicity [47]. We tested the toxicity responses of 2 model CYP450-mediated hepatotoxicity drugs, APAP and diclofenac. Hepatocytes cultured in RoboTox platform displayed higher sensitivity towards APAP and diclofenac-induced hepatotoxicity (Fig. 5). Further work is required to understand the mechanism of toxicity through the determination of drug sensitivity in the presence of CYP450 inhibitors such as Furafylline (CYP1A inhibitor) [48] or ketoconazole (CYP3A inhibitor) [49]. Demonstrating reduced sensitivity to other drugs such as Caffeine [50] that are rendered non-toxic by hepatocytes can also shed light into the mechanism of toxicity. We also tested the ability of perfusion culture in maintaining hepatocyte functions over longer period. CYP2E1 is one of the CYP450s that will deteriorate rapidly within a week even when the hepatocytes are cultured using sandwich configuration [51]. APAP forms NAPQI mainly through the CYP2E1 pathway [52]. We tested the CYP2E1-mediated hepatotoxicity on days 4, 8 and 14 of culture and found that the toxicity responses displayed by RoboTox-cultured hepatocytes were comparable throughout the three time points while static-cultured hepatocytes display significant different responses over different days (Fig. 6). This shows that basal CYP450s activity is maintained for longer period, which is important for minimizing drug screening variation caused by functional fluctuation or deterioration. High-throughput is another basic requirement for in vitro liver models in industry-scale drug screening [53]. Although perfusion culture is able to maintain and promote hepatocytes functions in vitro, incorporating it into high-throughput testing platform is challenging as the total amount of setup time increases exponentially with the increasing number of tests performed. A conventional perfusion culture system that uses the same fluidic flow for culture medium delivery and drug delivery creates a coupled design where a design parameter (ie. perfusion culture flow) has contradicting effects on 2 different functional requirements: positive effects for maintaining cell functions while negative effects for scalability or vice versa. A coupled design which does not satisfy Suh’s independence axiom compromises the overall functionality [54], eg. high-throughput capacity. To resolve the conflicts, the
1238
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
Fig. 8. Incorporation of robotic liquid handler into RoboTox system for IC50 measurement. Hepatotoxicity responses to (A) APAP, (B) Ketoconazole, (C) Diclofenac, (D) Chlorpromazine, (E) Flutamide and (F) Quinidine were evaluated using RoboTox-cultured hepatocytes versus collagen sandwich-cultured hepatocytes. Drugs were serially diluted into the cell culture using robotic liquid handler. After 24 h, MTS assays were performed by the robotic liquid handler to assess the cells viabilities (detailed protocols are listed in : Collagen sandwich, -: RoboTox. Data plotted represent the mean s.e.m of 3 independent experiments. Supplementary information).
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241
design must be decoupled so that different or even contradicting optimization process can be implemented to achieve the optimal high-throughput capability within each requirement constraints [55,56]. Here decoupling was achieved by distinguishing and separating media culture flow and the drug delivery flow of RoboTox platform. For cell culture, perfusion flow is crucial for maintaining hepatocyte differentiated functions while individual fluidic flow is less important. Therefore, the wells of the perfusion bioreactor are interconnected in series by a single fluidic channel, allowing one fluidic setup to provide for multiple Si3N4 sandwiches. This decreases the total number of fluidic connections needed and increases the scalability of the design [57]. There was no significant difference in terms of the nutrient accessibility and cell viability among the cells cultured in series (Fig. 7). For high-throughput drug testing, individual fluidic flow becomes important for multiplex testing while perfusion flow is less crucial as drug exposure period are short. The perfusion-cultured Si3N4 sandwiches were transferred to conventional 96-well plate so that each Si3N4 sandwich will now has a disconnected fluidic flow addressing to each of them. This allows different drugs and concentrations to be added to different units. To further improve the processing capability of the platform, RoboTox was also designed according to the standard 96well plate dimensions so that it can be incorporated into commercial robotic liquid handler platform. To demonstrate the RoboTox platform’s high-throughput capability, IC50 of 6 hepatotoxicants were evaluated. We found that the RoboTox-cultured hepatocytes displayed higher sensitivity towards all 6 hepatotoxicants (Table 3, Fig. 8), probably due to the improved activity of drug metabolism enzymes as shown earlier. Transition of tissue engineering discovery from laboratory to industry, either for therapeutics or pharmaceutical screening, is challenging. With so much efforts focusing on the biological or materials aspects, many discoveries do not address the downstream technical issues such as cost, usability, scalability and manufacturability for industry uses [58]. Such situation is also partly due to the traditional research paradigm that assumes transition from discovery to usability to manufacturing scalability is linear and sequential [59], where scientists focus on fundamental researches while engineers focus on product and process development. However, tissue engineering products are extremely sensitive to changes, making subsequent modifications to improve scalability or usability extremely difficult if not impossible [60,61]. This creates a gap between tissue engineering research and tissue engineering products. One approach to overcome the gap is to conduct tissue engineering research with specific aim in mind [62,63] so that research design can be guided by having downstream considerations and refinements incorporated into the earlier stages of research. The RoboTox platform is an attempt to implement this approach: (1) the final use of the in vitro liver culture platform is specified at the very beginning of the research, laying down the requirements and boundaries for the research. (2) Learning from product development literatures [64e66], special emphasis are placed on translating end user requirements into design guidelines. We have identified 3 important criteria for liver tissue culture engineered for drug screening purposes: good hepatocytes differentiated functions, robust/low variability, and high-throughput. To recapitulate hepatocytes functions, primary hepatocytes were cultured in galactosylated Si3N4 sandwich configuration. To minimized variations, collagen gel was replaced by microfabricated Si3N4 membrane and perfusion culture was used for stable maintenance of cell functions. To enable highthroughput drug screening, the bioreactor was designed specifically to decouple the perfusion culture flow and drug delivery flow in order to meet the different high-throughput requirements of the culture stage and the drug testing stage. The platform was also
1239
designed according to the standard multi-well culture plate dimension to incorporate robotic liquid handler for large scale drug testing. (3) The bioreactor optimizations, cell function validations and drug testing were conducted in a format closely resembled the final form. This enables smoother transition from research to industry once all characterizations and process optimization are completed. 5. Conclusions We have developed a robust and high-throughput drug-induced hepatotoxicity screening platform using perfusion sandwichcultured hepatocytes. The platform minimized variation by replacing collagen gel with galactosylated microfabricated membrane; and by maintaining hepatocyte functions with perfusion culture. High-throughput testing was enabled by having flow-decoupling and 96-well plate compliant designs that allow incorporation of robotic liquid handler. The bioreactor design and perfusion flow rate were optimized by computational fluid dynamics simulation and experimentally. The cultured hepatocytes preserved 3D cell morphology, urea production and cytochrome p450 activity functions of the differentiated mature hepatocytes for 14 days. Hepatocytes cultured in the system exhibited improved drug sensitivity and low variability in hepatotoxicity testing results amongst hepatocytes transferred from different dates of perfusion culture. The platform enables robust high-throughput screening of drug candidates. Acknowledgements We thank members of the Cell and Tissue Engineering Laboratory for technical supports and stimulating scientific discussions. This work is supported in part by funding from the Institute of Bioengineering and Nanotechnology, Biomedical Research Council, Agency for Science, Technology and Research (A*STAR) of Singapore; and grants from Jassen Cilag Grant (R-185-000-182592), Singapore-MIT Alliance Computational and Systems Biology Flagship Project (C-382-641-001-091), SMART BioSyM and Mechanobiology Institute of Singapore (R-714-001-003-271) to HYU. SZ, WHT, BZ, CZ and AA are National University of Singapore Research Scholars. Appendix Figures with essential color discrimination. Figs. 1e4 in this article are difficult to interpret in black and white. The full color images can be found in the on-line version, at doi:10.1016/j. biomaterials.2010.09.064. Appendix. Supplementary data The supplementary data associated with this article can be found in the on-line version at doi:10.1016/j.biomaterials.2010.09.064. References [1] Hewitt NJ, Lechon MJ, Houston JB, Hallifax D, Brown HS, Maurel P, et al. Primary hepatocytes: current understanding of the regulation of metabolic enzymes and transporter proteins, and pharmaceutical practice for the use of hepatocytes in metabolism, enzyme induction, transporter, clearance, and hepatotoxicity studies. Drug Metab Rev 2007;39:159e234. [2] Sivaraman A, Leach JK, Townsend S, Iida T, Hogan BJ, Stolz DB, et al. A microscale in vitro physiological model of the liver: predictive screens for drug metabolism and enzyme induction. Curr Drug Metab 2005;6:569e91. [3] Khetani SR, Bhatia SN. Microscale culture of human liver cells for drug development. Nat Biotechnol 2008;26:120e6. [4] Pӧrtner R, Giese C. An overview on bioreactor design, prototyping and process control for reproducible three-dimensional tissue culture. In: Marx U, Sandig V,
1240
[5] [6]
[7] [8]
[9]
[10]
[11]
[12]
[13]
[14]
[15] [16]
[17] [18]
[19]
[20]
[21]
[22] [23]
[24]
[25] [26]
[27]
[28] [29]
[30] [31]
[32]
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241 editors. Drug testing in vitro: breakthrough and trends in cell culture technology. New Jersey: Wiley; 2007. p. 53e78. Vickers AE, Fisher RL. Organ slices for the evaluation of human drug toxicity. Chem Biol Interact 2004;150:87e96. Donato MT, Jimenez N, Castell JV, Gomez-Lechon MJ. Fluorescence-based assays for screening nine cytochrome P450 (P450) activities in intact cells expressing individual human P450 enzymes. Drug Metab Dispos 2004;32:699e706. Donato MT, Lahoz A, Castell JV, Gomez-Lechon MJ. Cell lines: a tool for in vitro drug metabolism studies. Curr Drug Metab 2008;9:1e11. Cross DM, Bayliss MK. A commentary on the use of hepatocytes in drug metabolism studies during drug discovery and development. Drug Metab Rev 2000;32:219e40. Wanson JC, Drochmans P, Mosselmans R, Ronveaux MF. Adult rat hepatocytes in primary monolayer culture. Ultrastructural characteristics of intercellular contacts and cell membrane differentiations. J Cell Biol 1977;74:858e77. Dunn JC, Yarmush ML, Koebe HG, Tompkins RG. Hepatocyte function and extracellular matrix geometry: long-term culture in a sandwich configuration. FASEB J 1989;3:174e7. Dunn JC, Tompkins RG, Yarmush ML. Long-term in vitro function of adult hepatocytes in a collagen sandwich configuration. Biotechnol Prog 1991;7:237e45. Mingoia RT, Nabb DL, Yang CH, Han X. Primary culture of rat hepatocytes in 96-well plates: effects of extracellular matrix configuration on cytochrome P450 enzyme activity and inducibility, and its application in in vitro cytotoxicity screening. Toxicol in Vitro 2007;21:165e73. Hoffmaster KA, Turncliff RZ, LeCluyse EL, Kim RB, Meier PJ, Brouwer KL. Pglycoprotein expression, localization, and function in sandwich-cultured primary rat and human hepatocytes: relevance to the hepatobiliary disposition of a model opioid peptide. Pharm Res 2004;21:1294e302. Bjornsson TD, Callaghan JT, Einolf HJ, Fischer V, Gan L, Grimm S, et al. The conduct of in vitro and in vivo drug-drug interaction studies: a pharmaceutical research and manufacturers of America (PhRMA) perspective. Drug Metab Dispos 2003;31:815e32. Kim BS, Mooney DJ. Development of biocompatible synthetic extracellular matrices for tissue engineering. Trends Biotechnol 1998;16:224e30. Hewitt NJ, de Kanter R, LeCluyse E. Induction of drug metabolizing enzymes: a survey of in vitro methodologies and interpretations used in the pharmaceutical industryedo they comply with FDA recommendations? Chem Biol Interact 2007;168:51e65. Du Y, Han R, Wen F, Ng San San S, Xia L, Wohland T, et al. Synthetic sandwich culture of 3D hepatocyte monolayer. Biomaterials 2008;29:290e301. Gebhardt R, Mecke D. Perifused monolayer cultures of rat hepatocytes as an improved in vitro system for studies on ureogenesis. Exp Cell Res 1979;124:349e59. Langsch A, Bader A. Longterm stability of phase I and phase II enzymes of porcine liver cells in flat membrane bioreactors. Biotechnol Bioeng 2001;76:115e25. Xia L, Ng S, Han R, Tuo X, Xiao G, Leo HL, et al. Laminar-flow immediateoverlay hepatocyte sandwich perfusion system for drug hepatotoxicity testing. Biomaterials 2009;30:5927e36. De Bartolo L, Salerno S, Curcio E, Piscioneri A, Rende M, Morelli S, et al. Human hepatocyte functions in a crossed hollow fiber membrane bioreactor. Biomaterials 2009;30:2531e43. Domansky K, Inman W, Serdy J, Dash A, Lim MH, Griffith LG. Perfused multiwell plate for 3D liver tissue engineering. Lab Chip 2010;10:51e8. Gebhardt R, Hengstler JG, Muller D, Glockner R, Buenning P, Laube B, et al. New hepatocyte in vitro systems for drug metabolism: metabolic capacity and recommendations for application in basic research and drug development, standard operation procedures. Drug Metab Rev 2003;35:145e213. Ying L, Yin C, Zhuo RX, Leong KW, Mao HQ, Kang ET, et al. Immobilization of galactose ligands on acrylic acid graft-copolymerized poly(ethylene terephthalate) film and its application to hepatocyte culture. Biomacromolecules 2003;4:157e65. Findeis MA. Stepwise synthesis of a galnac-containing cluster glycoside ligand of the asialoglycoprotein receptor. Int J Pept Protein Res 1994;43:477e85. Zhang S, Xia L, Kang CH, Xiao G, Ong SM, Toh YC, et al. Microfabricated silicon nitride membranes for hepatocyte sandwich culture. Biomaterials 2008;29: 3993e4002. Iliescu C, Tay F, Wei J. Low stress and high deposition rate of PECVDeSiNx layers using high power and high frequency for MEMS applications. J Micromech Microeng 2006;16:869e74. Seglen PO. Preparation of isolated rat liver cells. Methods Cell Biol 1976;13:29e83. Kim SS, Sundback CA, Kaihara S, Benvenuto MS, Kim BS, Mooney DJ, et al. Dynamic seeding and in vitro culture of hepatocytes in a flow perfusion system. Tissue Eng 2000;6:39e44. Lin SH. Oxygen diffusion in a spherical cell with nonlinear oxygen uptake kinetics. J Theor Biol 1976;60:449e57. Toh YC, Zhang J, Khong YM, Du Y, Sun W, Yu H. Integrating sensitive quantification of hepatic metabolic functions by capillary electrophoresis with laser-induced fluorescence detection. Analyst 2008;133:326e30. Cho CS, Seo SJ, Park IK, Kim SH, Kim TH, Hoshiba T, et al. Galactose-carrying polymers as extracellular matrices for liver tissue engineering. Biomaterials 2006;27:576e85.
[33] Tilles AW, Baskaran H, Roy P, Yarmush ML, Toner M. Effects of oxygenation and flow on the viability and function of rat hepatocytes cocultured in a microchannel flat-plate bioreactor. Biotechnol Bioeng 2001;73:379e89. [34] Kidambi S, Yarmush RS, Novik E, Chao P, Yarmush ML, Nahmias Y. Oxygenmediated enhancement of primary hepatocyte metabolism, functional polarization, gene expression, and drug clearance. Proc Natl Acad Sci U S A 2009;106:15714e9. [35] Martin H, Sarsat JP, Lerche-Langrand C, Housset C, Balladur P, Toutain H, et al. Morphological and biochemical integrity of human liver slices in long-term culture: effects of oxygen tension. Cell Biol Toxicol 2002;18:73e85. [36] Sullivan JP, Gordon JE, Bou-Akl T, Matthew HW, Palmer AF. Enhanced oxygen delivery to primary hepatocytes within a hollow fiber bioreactor facilitated via hemoglobin-based oxygen carriers. Artif Cells Blood Substit Immobil Biotechnol 2007;35:585e606. [37] Kekonen EM, Jauhonen VP, Hassinen IE. Oxygen and substrate dependence of hepatic cellular respiration: sinusoidal oxygen gradient and effects of ethanol in isolated perfused liver and hepatocytes. J Cell Physiol 1987;133: 119e26. [38] Stresser DM, Turner SD, Blanchard AP, Miller VP, Crespi CL. Cytochrome P450 fluorometric substrates: identification of isoform-selective probes for rat CYP2D2 and human CYP3A4. Drug Metab Dispos 2002;30:845e52. [39] Sanderson JP, Naisbitt DJ, Park BK. Role of bioactivation in drug-induced hypersensitivity reactions. AAPS J 2006;8:E55e64. [40] Eschenhagen T, Didie M, Munzel F, Schubert P, Schneiderbanger K, Zimmermann WH. 3D engineered heart tissue for replacement therapy. Basic Res Cardiol 2002;97(Suppl. 1):I146e52. [41] Iliescu CWJ, Chen B, Ong OL. Silicon nitride membrane for cell culturing. Rom J Inf Sci Tech 2008;11:167e76. [42] Ni M, Tong WH, Choudhury D, Rahim NA, Iliescu C, Yu H. Cell culture on MEMS platforms: a review. Int J Mol Sci 2009;10:5411e41. [43] Davis DH, Giannoulis CS, Johnson RW, Desai TA. Immobilization of RGD to < 1 1 1 > silicon surfaces for enhanced cell adhesion and proliferation. Biomaterials 2002;23:4019e27. [44] Du Y, Chia SM, Han R, Chang S, Tang H, Yu H. 3D hepatocyte monolayer on hybrid RGD/galactose substratum. Biomaterials 2006;27:5669e80. [45] Lu HF, Lim WS, Wang J, Tang ZQ, Zhang PC, Leong KW, et al. Galactosylated PVDF membrane promotes hepatocyte attachment and functional maintenance. Biomaterials 2003;24:4893e903. [46] Guengerich FP. Cytochrome p450 and chemical toxicology. Chem Res Toxicol 2008;21:70e83. [47] Walgren JL, Mitchell MD, Thompson DC. Role of metabolism in drug-induced idiosyncratic hepatotoxicity. Crit Rev Toxicol 2005;35:325e61. [48] Burke MD, Thompson S, Weaver RJ, Wolf CR, Mayer RT. Cytochrome P450 specificities of alkoxyresorufin O-dealkylation in human and rat liver. Biochem Pharmacol 1994;48:923e36. [49] Kotegawa T, Laurijssens BE, Von Moltke LL, Cotreau MM, Perloff MD, Venkatakrishnan K, et al. In vitro, pharmacokinetic, and pharmacodynamic interactions of ketoconazole and midazolam in the rat. J Pharmacol Exp Ther 2002;302:1228e37. [50] Worboys PD, Bradbury A, Houston JB. Kinetics of drug metabolism in rat liver slices. III. Relationship between metabolic clearance and slice uptake rate. Drug Metab Dispos 1997;25:460e7. [51] Farkas D, Tannenbaum SR. Characterization of chemically induced hepatotoxicity in collagen sandwiches of rat hepatocytes. Toxicol Sci 2005;85:927e34. [52] Reid AB, Kurten RC, McCullough SS, Brock RW, Hinson JA. Mechanisms of acetaminophen-induced hepatotoxicity: role of oxidative stress and mitochondrial permeability transition in freshly isolated mouse hepatocytes. J Pharmacol Exp Ther 2005;312:509e16. [53] Sundberg SA. High-throughput and ultra-high-throughput screening: solution- and cell-based approaches. Curr Opin Biotechnol 2000;11:47e53. [54] Suh NP. The principles of design. Oxford: Oxford University Press; 1990. [55] Tate D, Nordlund M. A design process roadmap as a general tool for structuring and supporting design activities. In: Proceedings of the second world conference on integrated design and process technology 1996 Dec 1e4. Austin, USA. Birmingham: The Society for Design and Process Science; 1996. p. 97e104. [56] Eble A, Wagner G. Product design fundamentals. In: Bröckel U, Meier W, Wagner G, editors. Product design and engineering: best practices. Weinheim: Wiley-VCH; 2007. p. 317e40. [57] Stoll H. Design for manufacture: an overview. Appl Mech Rev 1986;39:9. [58] Archer R, Williams DJ. Why tissue engineering needs process engineering. Nat Biotechnol 2005;23:1353e5. [59] Godin B. The linear model of innovation e the historical construction of an analytical framework. Sci Technol Hum Values 2006;31:639e67. [60] Ratcliffe A, Niklason LE. Bioreactors and bioprocessing for tissue engineering. Ann N Y Acad Sci 2002;961:210e5. [61] Naughton GK. From lab bench to market: critical issues in tissue engineering. Ann N Y Acad Sci 2002;961:372e85. [62] Hollister SJ. Scaffold engineering: a bridge to where? Biofabrication 2009;1:1e14. [63] Stokes DE. Pasteurs quadrant: basic science and technological innovation. Washington DC: Brookings Institution Press; 1997. [64] Ulrich KT, Eppinger SD. Product design and development. 4th ed. New York: McGraw-Hill; 2004.
S. Zhang et al. / Biomaterials 32 (2011) 1229e1241 [65] Crawford M, Di Benedetto A. New products management. 9th ed. New York: McGraw-Hill; 2008. [66] Kahn KB, editor. PDMA handbook of new product development. 2nd ed. New Jersey: John Wiley & Sons; 2005. [67] Consolo F, Fiore GB, Truscello S, Caronna M, Morbiducci U, Montevecchi FM, et al. A computational model for the optimization of transport phenomena in a rotating hollow-fiber bioreactor for artificial liver. Tissue Eng Part A 2008; 15:41e55. [68] Foy BD, Rotem A, Toner M, Tompkins RG, Yarmush ML. A device to measure the oxygen uptake rate of attached cells: importance in bioartificial organ design. Cell Transplant 1994;3:515e27. [69] Hay PD, Veitch AR, Gaylor JD. Oxygen transfer in a convection-enhanced hollow fiber bioartificial liver. Artif Organs 2001;25:119e30. [70] Rotem A, Toner M, Tompkins RG, Yarmush ML. Oxygen uptake rates in cultured rat hepatocytes. Biotechnol Bioeng 1992;40:1286e91.
1241
[71] Lieberthal W, Wolf EF, Merrill EW, Levinsky NG, Valeri CR. Hemodynamic effects of different preparations of stroma free hemolysates in the isolated perfused rat kidney. Life Sci 1987;41:2525e33. [72] Amin A, Hamza AA. Effects of Roselle and Ginger on cisplatin-induced reproductive toxicity in rats. Asian J Androl 2006;8:607e12. [73] Tang W, Stearns RA, Bandiera SM, Zhang Y, Raab C, Braun MP, et al. Studies on cytochrome P-450-mediated bioactivation of diclofenac in rats and in human hepatocytes: identification of glutathione conjugated metabolites. Drug Metab Dispos 1999;27:365e72. [74] Kostrubsky SE, Strom SC, Ellis E, Nelson SD, Mutlib AE. Transport, metabolism, and hepatotoxicity of flutamide, drug-drug interaction with acetaminophen involving phase I and phase II metabolites. Chem Res Toxicol 2007;20: 1503e12. [75] Ishak KG, Zimmerman HJ. Drug-induced and toxic granulomatous hepatitis. Baillieres Clin Gastroenterol 1988;2:463e80.
