Биоинженерия / Биомеханика_микрофлюидные_устройства / статьи / 69_baudoin2007

1. Introduction

The discussions on the REACH European directive tend to suggest the transfer of a large amount of toxicological studies to the chemical and pharmaceutical industries to quantify their product safety. In the addition, the 3R recommendations (reduce, refine, replace) propose to limit as much as possible animal testing. One of the challenges of in vitro toxicity studies is to be able to preserve the cell/tissue functions for the longest time possible to construct adequate long-term pharmaceutical screening models. Chronic toxicity of a substance is usually evaluated on a live animal during several months, up to several years, to investigate the effect of daily administrations. In vitro, chronic toxicity is evaluated only over a few days because the primary

* To whom correspondence should be addressed. CNRS UMR 6600, Laboratoire de BiomeÂcanique et GeÂnie BiomeÂdical, Universite de Technologie de CompieÁgne, France. Phone: 33 (0)3 44 23 79 43. E-mail: eric.leclerc@utc.fr.

² CNRS-SATIE/BIOMIS.

³ Centre Hospitalier Universitaire d'Amiens.

cellular models can only preserve their specific and high activity for these short periods (1-3).

The pharmacological industry and laboratories have today the imperative to carry out reliable in vitro tests over longer periods to refine their diagnosis. Thus, to improve the culture conditions, the conventional Petri in vitro models used in flasks or microwell plates need to be re-evaluated and modified (4, 5). Thanks to progress in engineering fields, such as microtechnologies and tissue engineering, new pertinent tools, ªcellular microchipsº, have been proposed for the pharmaceutical and toxicological studies as alternative supports (6, 7). The controls in the microfabrication technologies allow new approaches to investigate cell-to-cell or cell-material interactions in dynamic and 3D microenvironments mimicking in vivo situations (8-13).

In this frame, various types of cellular microchips using silicon or biocompatible polymers to culture mammalian cells have been developed (14-16). The cellular microchips can be understood as a reaction zone for the toxicological analysis in which the reactive fluids will be tested on the cultivated

1246

engineered tissue. Advances in the microfluidic field have enabled us to realize microscale liquid conduits suitable for a continuous supply of nutrition, oxygen, toxic loadings, and waste removal through perfused culture medium (10, 11). The microstructures associated with the microfluidic flows have also demonstrated a secondary function as the microstructures located in the microchips have shown to enhance the reorganization of the cells in 3D tissue like structures (15, 16).

In a previous work, we described a cellular microchip made with polydimethylsiloxane (PDMS) loaded with hepatocarcinoma cell lines (16). In the present study, we present a MDCK cell cultures in a similar PDMS based cellular microchips to demonstrate that the same type of microchip can be used for various cell applications. Second, a toxicity study is performed to demonstrate the potential of the microchips for large-scale toxicity studies. For that purpose, we used two ammonium chloride loadings that are reported to reduce mammalian cell proliferation (17). In addition, we modeled flow and mass transfer within the microchip to present some interpretations of the in vitro results.

2.Materials and Methods

2.1.Polymer. To fabricate the microchip, we used the PDMS polymer (Dow Corning, Sylgard 184). This silicone is ideally suited to microfabrication. PDMS has been widely used in emerging biotechnology fields to construct microchannels and microstructures with the submicron features (18, 19). This material is generally considered to be biocompatible and possesses high gas permeability (20, 21), which allows oxygenation of cells in culture. As PDMS is a transparent material, microchips fabricated with PDMS allow optical observation coupled with real time analysis of the morphological views of the cells. However, the silicone is also accepted to be unfavorable to cell growth and migration due to its low surface energy (12, 13). Fibronectin, an extracellular matrix protein, coated on PDMS, was used to solve this problem. Literature reported that MDCK attached more rapidly to collagen type I to IV, but it was also observed that, after 4 h hours of contact, any difference in adhesion of MDCK with collagen, laminin, and fibronectin was detected (22, 23). Therefore, after microfabrication steps, a surface treatment (described below) was performed.

2.2.Microfabrication. The fabrication details, based on replica molding and PDMS plasma bonding, have been reported previously (18, 19). In summary, the construction of the cell microchip includes a first PDMS layer with the microstructures for cell attachment, which contains a series of microchambers (300 300 100 ím) and microchannels (400 150 100 ím) inside a larger cell culture chamber (100 ím deep). The total depth of this layer was 200 ím and is based on a double photolithography process. A second PDMS layer with a fluidic network is used to close the cell culture chamber. This layer includes an inlet microchannel network and a chamber of 100 ím thick to distribute the culture medium homogenously throughout the cell culture chamber. The resulting microchip has a volume roughly estimated at 30 íL, with a maximal height of 300 ím. The total surface available for cell growth is about 2 cm2 (including a bottom surface of 1 cm2 and side walls of the microchannels, but excluding the top surface of the second PDMS layer).

2.3.Cell Cultures and Media. Renal cells (Madin Darby Canine Kidney, MDCK) were provided by the American Type Culture Collection (ATCC, ref number CCL-34) and were originally extracted from canine distal tubules. The medium, as recommended by ATCC, contained Minimum Essential

Biotechnol. Prog., 2007, Vol. 23, No. 5

Medium (MEM, Eagle), 2 mM L-glutamine, 0.1 mM nonessential amino acids, 1.0 mM sodium pyruvate, 10% of fetal bovine serum, and penicillin-streptomicin (100 units/mL). Batch cultures were performed in T75 flasks (Falcon, Merk Eurolab, Strasbourg, France) using 12 mL of medium. This allowed the confluence of the cells with about 5 105 cells/cm2 (equivalent to 3.3 106 cells/cm3).

2.4. Experiments. In the experiments, the cell cultures were perfused inside the microchips using a perfusion loop, including the culture medium tank, the peristaltic pump, and the microchip. The circuit was interconnected by a 1.5 mm interior diameter silicone tubing.

To enhance initial cell adhesion, the inner surface of the microchips was coated with fibronectin by the introduction of a 10 íg/mL fibronectin solution for about 40 min (24, 25). After washing with culture medium, the cells were inoculated inside the microchips and kept at rest overnight in a 5% CO2 incubator at 37 °C. Initial tested cell density was from 5 to 10 105 cells per microchip (corresponding roughly to 2.5 and 5 105 cells/cm2). After 24 h, the nonattached cells were rinsed and removed. Then the perfusion was started and completed inside the CO2 incubator. The culture medium in the medium tank was periodically changed. Flow rates of 10, 25, and 50 íL/min were used. The perfusion was applied for 72 h, leading to a total of 96 h of culture inside the microchips. For comparative purposes, static cell cultures were also done inside the microchips without perfusion of culture media. For culture experiments in these ªstaticº microchips, only 200 íL of new media were injected for a few seconds every 24 h.

In the case of ammonium chloride loading, a stock solution of 100 mM was prepared. Then, dilutions in the culture medium were performed to achieve concentrations of 5 and 10 mM. After cell adhesion inside the microchips (adhesion performed without ammonium chloride in the culture medium), the cells were exposed to the medium containing the ammonium chloride during the 72 h of perfusion.

2.5. Biochemical Assays and Cell Counting. To demonstrate the viability and the potential of the microchips, basic metabolism of the cells were monitored by measuring the glucose cell consumption and the production of ammonia by the cells in the medium via glutamine metabolism. Glucose and ammonia were measured using the Konelab 20 biochemical analyzer (Thermo Electron Corporation).

The ammonia located in the culture media was measured indirectly by a combination with the alpha-ketoglutarate to create L-glutamate in the presence of NADH. The conversion of NADH to NAD during the L-glutamate formation leads to a decrease of the absorbance at 340 nm that is proportional to the ammonia concentration in the samples.

The glucose was quantified using two enzymatic reactions. At first, the glucose is transformed in D-gluconate and in H2O2 by a glucose oxidase (GOD). Then, in the presence of peroxydase, the oxidation of the 4-aminoantipyrine and phenol, by the H2O2 previously produced, is performed. This led to quinoneimine production. Light intensity induced by quinone-imine production was measured at 510 nm.

The viability was monitored by trypan blue coloration and cytotoxocity was assessed by lactate deshydrogenase (LDH) controls. LDH analysis was performed using the CytoTox96, nonradioactive cytotoxicity assay (Promega kit). A total of 50 íL of a solution containing the buffer assay and substrate mixture was dropped in microwell plates. Then, 100 íL of the culture medium was added and incubated in the dark at room temperature for 30 min. The reaction was stopped by adding

ªthe stop solutionº of the kit. The color development was measured by absorbance spectroscopy at 490 nm filter.

To count the cells, they were detached with perfusion of conventional trypsine-EDTA solution in the microchips and then collected. Then they were stained by trypan blue dye and counted under an optical microscope. To be able to check the cell proliferation, we used, for each point, six microchips. Three were stopped after 24 h at rest to quantify the cell number at the beginning of the experiment. The three last ones were stopped at the end of the experiments (after the 72 h of perfusion for instance).

2.6. Fluid Flow and Mass Transfer Modeling. 2.6.1. Grid.

A portion of our microchip geometry was reproduced using Solidworks software. The geometry was meshed using Gambit software 2.2.3 (Fluent Inc., Lebanon, NH). A grid based on tetrahedral 334 000 elements was implemented in the geometry of the microchip. The mesh quality was controlled by EquiAngleSkew (only 0.18% of elements of the mesh had an EquiAngleSkew between 0.7 and 0.8 and 0% above 0.8) and the EquiSizeSkew (only 0.19% of element had values between 0.7 and 0.8, and 0% above 0.8). Edge ratios and aspect ratios were below 3 for all elements.

2.6.2.Model Assumptions and Boundary Conditions. For our models of the microfluidic microchips, it was assumed that the flow was steady, three-dimensional, and laminar all over the geometry. The walls were considered to be solid. Consequently, a no-slip condition was applied at the wall. Due to the geometry, two symmetries were defined by a zero continuous shear stress value on the side external walls. We adjusted the pressure gradient in the simulation to reach the experimental flow rate and to achieve a parabolic flow profile in the geometry.

2.6.3.CFD Package. The Navier-Stokes equations (momentum and mass conservation) governing fluid motion (velocity, pressure) in the aforementioned conditions were solved by a finite volume method implemented in Fluent 6.2.3 software (Fluent Inc., Lebanon, NH). To model the nutrient transport of various species, advection diffusion equations were added for each corresponding concentration treated as a new variable name ªscalarº ¼k in Fluent. A segregated algorithm was applied to solve the equations so as to lower the memory requirements imposed by the complex geometry. Used equations are reported below and nomenclature, parameters, and inputs values are summarized in Table 1.

The boundary conditions of eq 3 were described by wall flux. The sets of eqs 3.1 to 3.4 were used to describe the metabolic flux. In the case of oxygen transport, a diffusion flux was added to take into account the oxygen diffusion through the PDMS wall of the microchip (Fext). This was done to describe the gas exchange through the materials. Glucose consumption was also modeled using parameters extracted from the experimental results (see Discussion sections).

At the bottom layers of the microchip, describing the cell activity we have

In addition, we defined the convection time tC and the

The Damkohler number, which is the dimensionless molecule flux, induces by the cell uptake

2.6.4. Post-treatment. At the end of each run, the convergence criteria were checked. They concerned the relative errors in velocity, according to x,y,z coordinates and pressure. An additional validation was made by comparing the inlet and outlet flow rates, which were equal only when the problem was appropriately formulated in Fluent. The software associated after the treatment allowed quantification of the scalars quantities, as well as the wall shear rates along the microchannels, and the velocity profiles in selected cross-sections.

3.Results

3.1.Cell Morphology and Activity. 3.1.1. Perfusions without Ammonium Chloride. Cells attached in all experiments after 24 h at rest, as shown by Figure 1A. Cell proliferation was observed by the optical visualizations, as shown by Figures 1B-D. At first, cells have created a confluent monolayer on the bottom surface. Then, an overgrowth on the microchannel walls was observed. This led to create 3D-like tissue structures and resulted in covering the geometry of the microchannels. Despite the flow rate, any specific cell orientation along the flow in the microchannels was observed when compared with dynamic and static cultures. In all of those experiments, the viability was above 90% except at 50 íL/min (trypan blue results).

Biotechnol. Prog., 2007, Vol. 23, No. 5

Figure 2. Effect of the flow rates and the initial cell densities on the MDCK proliferation inside the cellular microchip.

The cell proliferation was not affected by the initial cell density for the tested cases (5 to 10 105 cells/microchip), as demonstrated for three cases by Figure 2. In all static cases (inside the microchips flow rate was 0 íL/min), the cell number remained similar between the beginning and the end of the cultures. When there was no applied flow, the cells did not proliferate. On the contrary, for flows of 10 íL/min, initial cell density multiplied roughly by 2.2 ( 0.1, and at 25 íL/min, the population increased by 1.7 ( 0.2, as shown by the Figure 2. The measured glucose consumption, which corresponds to the cell metabolism, was about 16 ( 2 íg/h/106 cells at a flow rate of 10 íL/min and 35 ( 5 íg/h/106 cells at 25 íL/min (Table 2). The maximum cell density achieved in the microchip was 6.6 107 cells/cm3 at 10 íL/min and 4.6 107 cells/cm3 at 25 íL/min (corresponding to 2.3 106 and 1.6 106 cells/ microchip, respectively). LDH levels were very low, demonstrating healthy culture conditions (Table 2). Ammonia production was found higher at 25 íL/min, with a production of 1.4 ( 0.3 ímol/day/106 cells when compared to 0.8 ( 0.4 ímol/ day/106 cells at 10 íL/min (Table 2).

At 50 íL/min, for all inoculated densities, the cells detached after 24 h of perfusion. Viability was found below 10%.

3.1.2. Experiments with Ammonium Chloride. When the ammonium chloride was added to the culture medium, we clearly observed a reduction of the cell proliferation, as shown by Figure 1E,F. Morphological observations have shown that the cell shape and size were bigger when the ammonium chloride was loaded, especially at 10 mM, as shown by Figure 1E,F. We estimated roughly that the cells were 1.4- and 5.9- fold bigger with the ammonium chloride loading (at 5 mM and 10 mM, respectively) when compared to none loaded experiments. The 5 mM concentration contributed to the reduction of the growth ratio to about 1.5- to 2-fold, whereas 10 mM concentrations reduced the cell growth ratio to about 2- to 4-fold at 10 íL/min, as shown by Figure 3. In addition, the reduction of the cell proliferation was found relatively more important when the initial cell density was lower, as shown by Figure 3.

1249

Figure 3. Effects of the ammonium chloride loading on the cell proliferation at 10 íL/min.

Figure 4. Effects of the ammonium chloride loading and the flow rates on the cell proliferation with an inoculated cell density of 0.75 106 cells/microchip.

At 5 mM and 10 mM, 5 105 cells multiplied respectively by 1 and 0.5, whereas 8 105 cells multiplied respectively by 1 and 1.4. However, in terms of proliferation, Figure 4 shows that the ammonium chloride effect was more important at 10 íL/min than at 25 íL/min.

Metabolism analyses clearly show that the ammonium chloride increases the glucose consumption of the cells and reduces the ammonia production in perfusion, as shown by Figures 5 and 6. Despite the ammonium chloride loading, the viability of the collected cells was above 90% and LDH levels remained low.

3.2. Model Outputs. The inlet flow rate of 10 and 25 íL/ min used experimentally created shear stress values of 2-3 10-3 and 5-6 10-3 Pa, respectively. The averaged velocity was calculated to 100 and 220 ím/s (with a maximal value in the Poiseuille velocity profile of 170 ím/s and 400 ím/s, respectively), as shown by Figure 7A,B (10 íL/min case). Because the microstructures of the geometry are located at the bottom of the microchip, the design of the microchip allowed

Figure 5. Glucose consumption inside the cellular microchip according to the flow rates and the ammonium chloride loadings.

Figure 6. NH3 production, estimated resulting from the glutamine metabolism, inside the cellular microchip according to the flow rates and the ammonium chloride loadings. Production at 10 mM could not be detected.

uniform flow at the upper portion of the cell cultures (as shown by the arrow in Figure 7B). In addition, the microstructures behaved as obstacles, improving fluid resistance. This allowed a reduction of the fluid velocity at the bottom surface where the cells were attached.

In terms of nutrient transport, the time of convection, giving the residence time of a molecule driving by the flow inside the microchip, was estimated to be about 60 s and 24 s at 10 and 25 íL/min. Diffusion lengths, LD, of the same molecule during that time, were estimated by relation (4.2). As the maximum interior height of our microchips is 300 ím, this indicates that at a 10 íL/min flow rate, the molecules such as glucose (LD ) 250 ím), ammoniac (LD ) 440 ím), or oxygen LD ) 350 ím) have roughly the time to migrate from the fluid to the cells between the inlet and outlet of the microchip. The situation is less favorable at 25 íL/min (glucose, LD ) 250 ím; ammoniac, LD ) 290 ím; oxygen, LD ) 220 ím).

In Figure 7C, we present the gradient of concentration of glucose along the microchip due to the cell consumption (for 106 cells/microchip at 10 íL/min and using experimental uptake). In those conditions, we found that the cells between the inlet and the outlet of the microchip would consume 20% of the glucose. Those results clearly demonstrate that glucose could be quickly consumed and be in shortage with respect to the total volume of the perfusion. When ammonia was added, we found that 70% of the glucose would disappear between the inlet and the outlet of the microchips (Figure 7D). The oxygen supply originates from the dissolved molecules in the culture medium and the diffusion flux through the PDMS layer. The diffusion calculations show that we do not have hypoxia in our culture conditions if the oxygen diffusion through the PDMS is taken into account (for 106 cells/microchip at 10 íL/

Figure 7. Results of the numerical simulations inside the microchip at 10 íL/min: (A) shear stress contours; (B) velocity profiles; (C) glucose consumption along the microchip based on the data measured and reported in Figure 6; (D) glucose consumption when 10 mM ammonium chloride is added to the culture medium, based on the data measured and reported in Figure 6; (E) oxygen consumption inside the cellular microchip taken into account diffusion through the PDMS;

(F) oxygen consumption inside the cellular microchip without taking into account diffusion through the PDMS.

min). The total length of the cell culture chamber being 1 cm, the dissolved oxygen in the culture medium was consumed in the first 0.25 cm, resulting in a main oxygenation via the PDMS in 75% of the microchip, as shown by Figure 7E.

4.Discussion

4.1.Glucose Metabolism. Glucose is one of the energy sources for the cells, and its consumption was found to increase with increasing flow rate. These results were consistent with the observations of different cell types such as erythrocytes, CHO, when submitted to shear stress (26, 27). In addition, the glucose consumption dramatically increased and the proliferation decreased when we added ammonium chloride in the medium, as reported in the literature (28-30). Metabolism of ammonia transport in the kidney may result mainly from two pathways,

the NH3 passive diffusion through the cell membranes and the active ionic transport of NH4+ through the K+ pathways (such as Na+-K+-ATPase ionic channels, the Na+- H+ exchangers, and the tight junctions (31, 32)). Mechanisms of toxicity of ammonia or ammonium chloride can be linked to the perturbation of intra/extracellular pH and the electrochemical gradients (16). NH3 is a small, uncharged, lipophilic molecule that readily diffuses across cellular membranes. MDCK is reported to have a high NH3 permeability (about 0.7 10-2 cm/s (33)). Therefore, it is reported that the NH3 diffusion will follow the gradient of the chemical potential of NH3 and thus rapidly equilibrate any transmembrane gradient of NH3. As the protonation is extremely fast, the pH will be reconstituted immediately (16). This diffusion will modify the intracellular

Biotechnol. Prog., 2007, Vol. 23, No. 5

cytoplasm pH that will result in an increasing mitochondria pH level (16). As the main reactions of the energy metabolism are located in the mitochondria, the pH modification in the mitochondria may modify the cell behavior. On the other hand NH4+ transport is very slow and is in competition with other ions (K+ for instance, via their ionic pathways). Na+-K+- ATPase pathways have a high-energy demand (16, 34) and competition between K+ and NH4+ is also energy costly (35)

4.2. Glutamine Metabolism. Glutamine is reported to be another energy source. Metabolism of glutamine leads to the production of ammonia by the cells. The ration Ramm/glu (between produced ammonia and available glutamine) in our experiments was found around 0.25 at 10 íL/min and about 0.5 at 25 íL/ min (when there is no ammonium chloride loadings), which appear weaker than the 1.06 data reported for MDCK growth cultures in other type of bioreactors (36). With a concentration of 2 mM of glutamine in our culture medium, we produce about 0.5 to 1 mM of ammonia. Ammonia is reported to be an inhibitory molecule for the cell proliferation. Investigation of the sensibility of MDCK cells have shown that concentrations of ammonium chloride below 4 mM were not effective on the MDCK, whereas values between 7 and 10 mM reduced the growth rate about 50% (30). However, Butler found that a 2 mM accumulation of ammonia stops the MDCK proliferation (37). Schneider reports that the glutamine metabolism and production of ammonia by the cells decrease the pH in the mitochondria. Therefore, glutamine metabolism helps in diffusing NH3 from the mitochondria to the cytoplasm and then to the culture medium. Our measurements have shown that ammonia production values of 0.5 and 1 mM were obtained after 2 days of cultures. However, the culture medium was periodically changed, and this may explain why we did not observe such growth inhibition (in the experiments without ammonium chloride loading). Furthermore, it is also reported that cell growth values have shown a higher tolerance to ammonia under continuous culture conditions, when compared to static batch cultures (29, 38). Perfusion may contribute to dilute and delay the ammonia concentration and therefore enhance the cell proliferation inside the bioreactors (28). However, additional should be taken into account as far as we observed better proliferation at 10 íL/min flow rate when compared to that of the 25 íL/min one.

In addition, when ammonium chloride was loaded, we found as reported in the literature that the glutamine metabolism was reduced. The ration Ramm/glu in our experiments was found around 0.14 ( 0.025 and 0 (0 ) below our detection limit) at 10 íL/min and about 0.11 ( 0.05 to 0 (below our detection limit) at 25 íL/min for 5 mM and 10 mM loadings, respectively. This was attributed by Miller (38) to a reduction of the flux of glutamate to alpha-ketoglutamate via the glutamate dehydrogenase during the deamination and transamination reactions of the glutaminolysis, whereas Schneider reported that this may result from an effect of the perturbation of intra/extracellular pH and electrochemical gradient (16).

4.3. Cell Density and Microchip Design. A close link between the glucose and the glutamine metabolisms has been shown and it appears that they replace each other when there is a shortage of one of those elements (16). Rapid aerobic glycolis and glutaminolysis with subsequent ammonia excretion appear to be a characteristic of fast growing mammalian cells. This appears consistent with our measurements and morphological observations. In addition, as reported in the literature, we developed a microchip with microstructures to promote a high cell density with respect to the surface/volume ratio. The

1251

maximal reached values were 6.6 107 cells/cm3 and corresponding to 1.15 106 cells/cm2. These values were higher than our results in conventional petri cultures (5 105 cells/ cm2 or 3.3 106 cells/cm3). These tendencies were consistent with other data reported for MDCK cultures in perfusion using microcarrier bioreactors in which 107 cells/mL were reached in the bioreactors, whereas 106 cells/mL was obtained in batch cultures (39). However, thanks to the microtopography, we achieved about a 7-fold higher cell density in our microchip. Due to the geometry, the cell grew up over the microstructures creating 3D like tissue structures. The introduction of microstructures created nonuniform flow fields, protecting the cells from the main flow and offering surface. Literature reported the benefit of such microstructures to protect the cells from mainstream flow for various types of cell culture in microchips (14, 15). In addition, the microstructures of the geometry, because they are located at the bottom of the microchip, allowed permanent cell feedings despite overgrowth over the microchannels walls.

4.4.Physiological Flows. Physiological data for the distal renal tubules reported a diameter of 25 ím to 35 ím, allowing a flow rate of about 10 to 30 nL (40). Therefore, in the distal tubule, the average velocity can be estimated at about 200 to 600 ím/s. If the viscosity of the filtrate in the tubule is taken close to the water one (10-3 Pa.s) or close to the blood (3-4 10-3 Pa s), the local shear stress can be estimated between 0.003 and 0.04 Pa. Thanks to the microgeometry, we could approach in vitro flow conditions as shown by numerical simulations in which we achieved shear stress values between 2 and 6 10-3 Pa inside 100 ím square type microchannels. However, in our experiments using 50 íL/min, resulting in a shear stress of 0.01 Pa, we have found a limitation of the cell proliferation and a large number of dead cells. This observation was consistent with literature reports on investigation of MDCK under shear stress of 9 10-3 Pa, demonstrating a large reduction of the viable cells after 12 h of perfusion (41). Because most of the reabsorption in the kidney is achieved in the proximal tubules, the fluid flow (velocity and shear stress) is reduced in the distal tubule (40). Therefore, MDCK, as distal cells, might be unable to adapt in vitro to higher shear stress values, as already reported (41).

4.5.Molecule Transports in the Model. The choice of the governing equations in the model was consistent with other works reported in the literature (10, 42, 43). However, we could find a large panel of parameters. For instance, oxygenation is a critical issue to maintain healthy tissue and culture conditions. Based on a cell oxygen uptake of 2 10-16 mol O2/s/cell used in the literature for MDCK (44) (and also for other mammalian cells (45)), we noticed that oxygenation resulting from the culture medium alone cannot be sufficient to oxygenate about 1 106 cells. PDMS permeability may allow additional oxygen permeation that can provide oxygenation as shown by the Figure 7E,F. In these cases, the Damkohler number, Da, which is the dimensionless molecule flux induced by the cell uptake, was found to be about 0.75. This means that the cells consume roughly the oxygen at the same rate as the diffusion of the molecules to the cells. However, because the cell number was increased up to 2 106 cells, as reached in some experiments, the model shows that we would be in weak hypoxia conditions (data not shown). In this case, Da was found to be about 1.5, which is however still acceptable. To achieve a fully oxygenated culture, we should work in the model with an oxygen uptake of 1 10-16 mol O2/s/cell. This uptake rate was found

1252

experimentally for chondrocyte cultures (43, 46, 47), resulting therefore in Da ) 0.75.

5. Conclusions

The results have demonstrated the MDCK proliferation during dynamic culture inside the PDMS microchip. Reduction of the proliferation by ammonium chloride, as reported in the literature, demonstrated the potential of the microchip for future in vitro drug analysis. Models for toxicology using fewer animals and less expense, avoiding an intensive labor, should be developed. The challenge of the studies carried out is to preserve the cell functions in culture in dynamic conditions over long time periods to build adequate models for drug screening and pharmacological discovery. The microchips offer the advantage of including a whole new set of technologies. Indeed, they offer the possibility of dynamic cultures and of kinetic studies on microstructured tissues simulating the cellular organizations that are met in vivo more closely than conventional petri dishes.

Acknowledgment

This work was granted by ªle fond social EuropeÂenº and ªla reÂgion Picardieº. This project is included in the regional research program ªIBF-Bioº.

References and Notes

(1)Williams, P. D. The application of renal cells culture in studying drug-induced nephrotoxicity. In Vitro Cell DeV. Biol. 1989, 25 (5), 800-805.

(2)Hayes, A. W. Principles and methods of toxicology, 4th edition; Taylor and Francis: Philadelphia, PA, U.S.A., 2001.

(3)Guillouzo, A. Liver cell models in in vitro toxicology. EnViron. Health Perspect. 1998, 106 (2), 511-532.

(4)Koh, H. L.; Yau, W. P.; Ong, P. S.; Hedge A. Current trends in modern pharmaceutical analysis for drug discovery. Drug DiscoVery Today 2003, 8, 19.

(5)Roberts, S. A. Drug metabolism and pharmacokinetics in drug discovery. Curr. Opin. Drug DiscoVery DeV. 2003, 6, 66-80.

(6)Viravaidya, K.; Sin, A.; Shuler, M. L. Development of a Microscale Cell Culture Analog To Probe Naphthalene Toxcity. Biotechnol. Prog. 2004, 20, 316-323.

(7)Griffith, L.; Naughton, G. Tissue enginneringsCurrent challenges and expanding opportunities. Science 2002, 295, 1009-1014.

(8)Falconnet, D.; Csus, G.; Grandin, M.; Textor, M. Surface engineering approaches to micropattern surfaces for cell-based assays. Biomaterials 2006, 27 (16) , 3044-3063.

(9)Bhatia, S. N.; Yarmush, M. L.; Toner, M. Controlling cell interactions by micropatterning in co-cultures: Hepatocytes and 3T3 fibroblasts. J. Biomed. Mater. Res. 1997, 34, 189-199.

(10)Allen, J. W.; Khetani, S. R.; Bhatia, S. N. In vitro zonation and toxicity in a hepatocyte bioreactor. Toxicol. Sci. 2005, 84, 110- 119.

(11)Tilles, A. W.; Baskaran, H.; Roy, P., Yarmush, M. L.; Toner, M. Effect of oxygenation and flow on the viability and function of rat hepatocytes cocultured in a microchannel flat-plate bioreactor.

Biotechnol. Bioeng. 2001, 73 (5), 379-389.

(12)Leclerc, E.; Corlu, A.; Griscom, L.; Baudoin, R.; Legallais, C. Guidance of liver and kidney organotypic cultures inside rectangular silicone microchannels. Biomaterials 2006, 27, 4109-4119.

(13)Baudoin. R.; Duval, J. L.; Griscom, L.; Baudoin, R.; Legallais, C.; Leclerc, E. Selective control of liver and kidney cells migration during organotypic co-cultures inside fibronectin coated rectangular silicone microchannels. Biomaterials 2007, 28, 1820-1829.

(14)14) Borenstein, J. T.; Terai, H.; King, K.; Weinberg, C.; Kaazempur-Mofrad, M.; Vacanti, J. P. Microfabrication technology for vascularized tissue engineering. Biomed. MicrodeVices 2002, 4 (3), 167-175.

(15)Powers, M. J.; Domansky, K.; Udapadhia, A.; Kazempur-Mofrad, M. R.; Kursawky, P.; Janigan, D.; Wack, K. E.; Stolz, D. B.; Kamm, R.; Griffith, L. A microarray perfusion bioreactor for 3D liver culture.

Biotechnol. Bioeng. 2001, 78, 257.

Biotechnol. Prog., 2007, Vol. 23, No. 5

(16)Leclerc, E.; Sakai, Y.; Fujii, T. Perfusion culture of fetal human hepatocytes in microfluidic environments. Biochem. Eng. J. 2004, 20, 143-148.

(17)Schneider, M.; Marison, I. W.; Von Stockar, U. The importance of ammonia in mammalian cell culture. J. Biotechnol. 1996, 46, 161-185.

(18)Xia, Y.; Whitesides, G. M. Extending microcontact printing as a microlithographic technique. Langmuir 1997, 13, 2059-2067.

(19)Anderson, J. R.; Chiu, D. T.; McDonald, J. C.; Jackman, R. J.; Cherniavskaya, O.; Wu, H.; Whitesides, S.; Whitesides, G. M. Fabrication of topologically complex three-dimensional microfluidic systems in PDMS by rapid prototyping. Anal. Chem. 2000, 72, 3158-3164.

(20)Zanzotto, A.; Szita, N.; Boccazzi, P.; Lessard, P.; Sinskey, A.; Jensen, K. Membrane-Aerated Microbioreactor for high throughput bioprocessing. Biotech. Bioeng. 2004, 87 (2), 243-254.

(21)Charati, S. G.; Stern, S. A. Diffusion of gases in silicone polymer: Molecular dynamics simulations. Macromolecules 1998, 31, 5529-5535.

(22)Kanai, N.; Fujita, Y.; Kakuta, T.; Saito, A. The effects of various extracellular matrices on renal cell attachment to polymer surfaces during the development of bioartificial renal tubules. Artif. Organs 1998, 23 (1), 114-118.

(23)Salas, P. J.; Vega-Salas, D. E.; Rodriguez-Boulan, E. Collagen receptors mediate early events in the attachment of epithelial (MDCK) cells. J. Membr. Biol. 1987, 98 (3), 223-236.

(24)Lee, J. H.; Khang, G.; Lee, J. W.; Lee, H. B. Interaction of different types of cells on polymer surfaces with wettability gradient.

J. Colloid Interface Sci. 1998, 205, 323-330.

(25)Chen, H.; Brook, A. M.; Sheardown, H. Silicone elastomers for reduced protein adsorption. Biomaterials 2002, 25 (12), 2273-2282.

(26)Keane, J. T.; Ryan, D.; Gray, P. P. Effect of shear stress on expression of a recombinant protein by Chinese hamster ovary cells.

Biotechnol. Bioeng. 2003, 20 (81), 211-220.

(27)Kodicek, M.; Mircevova, L.; Marik, T. Energy requirements of erythrocytes under mechanical stress. Biomed. Biochim. Acta 1987, 3, 103-107.

(28)Newland, M.; Greenfield, P. F.; Reid, S. Hybridoma growth limitations: The roles of energy metabolism and ammonia production. Cytotechnology 1990, 3, 215-229.

(29)Miller, W. M.; Wilke, C. R.; Blanch, H. W. Transient response of hybridoma cells to lactate and ammonia pulse and step changes in continuous culture. Bioprocess Eng. 1989, 3, 113-122.

(30)Glacken, M. W.; Fleischaker, R. J.; Sinskey, A. J. Reduction of waste product excretion via nutrient control, possible strategies for maximizing product and cell yields on serum in cultures of mammalian cells. Biotechnol. Bioeng. 1986, 26, 1376-1389.

(31)Knepper, M.; Packer, R.; Good, D. Ammonium transport in kidney. Physiol. ReV. 1989, 69 (1), 179-249.

(32)Chang, H.; Fujita T. a numerical model of acid base transport in rat distal tubule. Am. J. Physiol Renal Physiol. 2001, 281, F222-

F243.

(33)Golchini, K.; Kurzt, I. NH3 permeation through apical membrane MDCK cells is via the lipid pathway. Am. J. Physiology 1998, 255 (7), 135-141.

(34)Harris, S. I.; Balban, R. S.; Barett, L.; Mandel, L. J. Mitochondrial respiratory capacity and Na+ and K+ independent adenosine triphosphate mediated ion transport in the intact renal cell. J. Biol. Chem. 1981, 256, 319-328.

(35)Martinelle, K.; Haggstrom, L. Mechanism of ammonia and ammonium ion toxicity in animal cells: Transport across cell membranes. J. Biotechnol. 1993, 30, 339-350.

(36)Genzel, Y.; Behrendt, I.; KoÃnig, S.; Sann, H.; Reichl, U. Metabolism of MDCK cells during cell growth and influenza virus production in large scale micro carrier culture. Vaccine 2004, 22, 2202-2208.

(37)Butler, M.; Imamura, T.; Thomas, J.; Thilly, W. G. High yield from microcarrier cultures by medium perfusion. J. Cell Sci. 1983, 61, 353-363.

(38)Miller, W. M.; Blanch, H. W.; Wilke, C. R. A kinetic analysis of hybridoma growth and metabolism in batch and continuous suspension culture, effect of nutrient concentration, dilution rate and pH.

Biotechnol. Bioeng. 1988, 32, 947-965.

Biotechnol. Prog., 2007, Vol. 23, No. 5

(39)Butler, M.; Thilly, W. G. MDCK microcarrier cultures: Seeding density effects and amino acid utilization. In Vitro 1982, 18, 213- 219.

(40)Brenner, B. Brenner & Rector's the Kidney, 7th edition; W.B. Saunders Company: Philadelphia, PA, 2004.

(41)Bhat, V. D.; Windridge, W.; Cherry, R. S.; Mandel, L. J. Fluctuating shear stress effects on stress fiber architecture and energy metabolism of cultured renal cells. Biotechnol. Prog. 1995, 11, 596- 600.

(42)Sengers, B. G.; von Donkelaar, C. C.; Oomens, W. J.; Baaijens, F. P. T. Computational study of culture conditions and nutrient supply in cartilage tissue engineering. Biotechnol. Prog. 2005, 21, 1252- 1261.

(43)Allen, J. W.; Bhatia, S. N. Formation of steady state oxygen gradients in vitro, application to liver zonation. Biotech. Bioeng. 2003, 82 (3), 253-262.

(44)Bebok, Z.; Tousson, A.; Schwiebert, L.; Venglarik, C. Improved oxygenation promotes CFTR maturation and trafficking in MDCK monolayers. Am J. Physiol. 2001, 280, C135-C145.

1253

(45)Peachman, K. K.; Lyles, D. S.; Bass, D. A. Mitochondria in eosinophils: Functional role in apoptosis but not in respiration. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 1717-1722.

(46)Torzilli, P.; Askari, E.; Jenkins, J. Water content and solute diffusion properties in articular cartilage. In Biomechanics of Diarthrodial Joints; Mow, V., Ratcliffe, A., Woo, S. Y., Eds.; Springer: New York, 1990; 363-390.

(47)Allhands, R.; Torzilli, P.; Kallfeltz, F. Measurement of diffusion of uncharged molecules in articular cartilage. Cornell Vet. 1983, 74, 111-123.

(48)Foy, B. D.; Rotem, A.; Toner, M.; Tompkins, R. G.; Yarmush, M. L. A device to measure the oxygen uptake rate of attached cellss Importance in bioartificial organ design. Cell Transplant. 1994, 3, 515-527.

(49)Kemper, W. D. Solute diffusivity. In Methods of Soil Analysis, Part 1, 2nd ed.; Klute, A., Ed.; American Society of Agronomy and Soil Science Society of America: Madison, WI, 1986.

Received November 17, 2006. Accepted July 17, 2007.

BP0603513