Metabolic Control Through Hepatocyte and Adipose TissueCross-Talk in a Multicompartmental Modular Bioreactor

Maria Angela Guzzardi, Ph.D.,1,2 Claudio Domenici, Ph.D.,2 and Arti Ahluwalia, Ph.D.2,3

Physiological processes involve a complex network of signaling molecules that act through paracrinal or en-docrinal pathways; however, traditional in vitro models cannot mimic these interactions because of the lack of adynamic cross-talk between cells belonging to different tissues. The multicompartmental modular bioreactor is anovel cell culture system where hepatocytes and adipose tissue are shown to interact in a more physiologicalmanner. In the multicompartmental modular bioreactor, cells and tissues can be cultured in a common medium,which flows through the system acting as the bloodstream. Primary rat hepatocytes and adipose tissue werecultured separately and together in conventional conditions and in the bioreactor. Urea synthesis, albuminsecretion, glycerol, free fatty acid, and glucose concentrations were analyzed and compared. The dynamicconnected culture of adipose tissue and hepatocytes led to a significant enhancement of hepatic function in termsof increase of albumin and urea production with respect to conventional cultures. Interestingly, the glycerolgradually released from adipose tissue was buffered in the dynamic connected culture, manifesting a homeo-static-like control. These data show that the dynamic culture not only improves hepatocyte function, but alsoallows a cross-talk between tissues, leading to enhanced metabolic regulation in vitro.

Introduction

Many physiological processes involve the cross-talkbetween different tissues. In metabolic regulation, for

instance, liver, adipose tissue, pancreas, and muscles cooper-ate through several biochemical pathways triggered by hor-mones and small molecules, providing the body with the rightamount of energy it needs, and storing excess energetic sub-strates.1 In vivo studies provide evidence of the energy andmetabolic balance that characterize different physiological andpathophysiological conditions, but simpler in vitro models areneeded to achieve a more profound and mechanistic-basedunderstanding of the interactions between individual tissues.The liver, the main orchestrator of metabolism, is involved inmany of the most important functions promoting metabolichomeostasis in the whole body, such as glycogen storage,protein and lipid release, toxin removal, cytochrome P450activity, urea production, and metabolic processing.2

Primary hepatocytes are a tool for drug metabolism studiessince containing physiological concentrations of cofactorsand enzymatic systems, which make them much more ap-pealing than microsomal preparations.3,4 Moreover, severaltypes of bioreactors for hepatocyte in vitro culture have beendeveloped to capture some aspect of the in vivo physiology.Most of them have been realized with the aim of retaining

hepatocyte-specific functions, to provide temporary hepaticsupport to patients with acute liver failure. They are knownas bioartificial liver devices (BAL).5,6 Other bioreactors havebeen developed for carrying out physiological in vitrostudies, such as cancer cell metastasis and regeneration,mimicking the 3D liver parenchyma architecture7,8 or fortoxicological investigations, connecting hepatocytes andother types of cells in the same device.9 Moreover, althoughthe possibility has not been widely investigated, they couldbe used to establish in vitro models for studying the cross-talk between tissues involved in metabolic regulation and inglucose homeostasis. An in vitro model that reproduces andenhances the cross-talk between hepatocytes and adiposetissue would be interesting for studying glucose homeostasisand regulation in physiological and pathological conditions,such as type 2 diabetes.

To this end the concept of multicompartmental bioreactor(MCB) has been developed to study the cross-talk betweentissues belonging to different organs, with particular regardto the metabolic system. In previous reports we showed thatthe metabolic function of both primary rat hepatocytes andhuman HepG2 cells increased when they were cultured withendothelial cells (HUVEC) in the MCB.10,11 One of the mainconclusions of our studies was that to elicit physiologicallymeaningful responses, a connected culture system requires,

1Scuola Superiore Sant’Anna, Sector of Medicine, Piazza Martiri della Liberta’, Pisa, Italy.2Institute of Clinical Physiology, CNR, Pisa, Italy.3Faculty of Engineering, Interdepartmental Research Center ‘‘E. Piaggio,’’ University of Pisa, Pisa, Italy.

TISSUE ENGINEERING: Part AVolume 17, Numbers 11 and 12, 2011ª Mary Ann Liebert, Inc.DOI: 10.1089/ten.tea.2010.0541

1635

first, physiologically scaled cell numbers and ratios and,second, flow rates that do not cause shear stress-relateddamage to cells and that allow adequate residence timesin each compartment to enable cells to process metabolicsignals.

Using these considerations, the multicompartmentalmodular bioreactor (MCmB), which stems from the MCB,and represents its evolution in a more flexible ‘‘system on aplate’’ bioreactor device, was developed. In the MCmB, dif-ferent types of cell or tissues can be cultured in differentchambers that are connected to each other through the flowof the medium, exchanging metabolites from one chamber tothe others. The detailed design and fabrication process of theMCmB is described in our previous work.12 Focusing onglucose and lipid homeostasis, hepatocytes and adipose tis-sue cells, two of the main cell types present in the abdominalregion and having a role in metabolic regulation, were cul-tured in the MCmB.

We define a ‘‘connected culture’’ as a static or dynamicculture where two cell types are cultured together, andwhere the cross-talk between them is mediated not by directcell–cell junctions, but by simple diffusion through the me-dium or laminar flow respectively. In this work we foundthat the flow-mediated connected culture of hepatocyteswith adipose tissue in the new MCmB not only promoteshepatocyte-specific functions with respect to cultures inconventional static conditions, but also enhances tissue cross-talk and a homeostasis-like metabolic regulation. Thus, thisrepresents the first step toward the realization of an in vitrometabolic model comprising liver, adipose tissue, pancreas,and muscle cells.

Materials and Methods

The MCmB bioreactor

The MCmB bioreactor is made of polydimethyldisiloxane(Dow Corning), and has the same dimensions as a24-microwell plate (13 mm diameter and 2 mL volume). Forthe purpose of this study, from one to three MCmB biore-actors were connected in series to a pump (Ismatech) and amixing chamber, which serves to facilitate assembly of theflow circuit, allows gaseous mixing and acts as a bubble trap(Fig. 1). The MCmB bioreactor was specifically designed toprovide a low shear stress to cells that are not directly ex-posed to tangential flow in vivo (such as hepatocytes andadipocytes), but receive nutrients through interstitial flow.The flow rate used (185 mL/min) is in the range found to beoptimal for hepatocyte function in Mazzei et al.12 The aver-age shear stress in the cell culture zone, calculated by fluiddynamic simulation, was about 10 - 5 Pa.

Materials

Trypsin, ethylenediaminetetraacetic acid, phosphate-buffered saline, and cell culture media were purchased fromCambrex Corporation. Fetal bovine serum (FBS) was fromLonza Bioscience. All other reagents and medium supple-ments were purchased from Sigma-Aldrich.

Two different hepatocyte culture media were obtainedfrom standard William’s E medium supplemented with50 mg/mL streptomycin sulfate, 50 mg/mL kanamycinmonosulfate, 10 mg/mL sodium ampicillin, and 7.3 IU/mL

benzyl penicillin. Medium named T1 was obtained afteraddition of 10% FBS, whereas medium called T2 contained0.02 mg/mL epidermal growth factor, 0.2% bovine serumalbumin, and 1mM dexametasone.

Adipose tissue culture medium consisted of Dulbecco’smodified Eagle’s medium supplemented with 100 U/mLpenicillin, 100 mg/mL streptomycin, and 5% FBS.

Hepatocyte isolation and culture

Primary hepatocytes were isolated from adult male Wistarrats weighing between 250 and 300 g using a modification ofthe two step collagenase perfusion method as described bySeglen.13 Briefly, rat liver was washed with Ca2 + -free Hank’sbuffer with 5.4 g/L D-glucose and 6 g/L HEPES, perfusedwith ethylene glycol tetraacetic acid (EGTA) buffer, andwashed again. Then, it was perfused with 0.5 mg/mL col-lagenase type IV solution with CaCl2 5 mM. Isolated hepa-tocytes were filtered and suspended in T1 medium. Cellviability ranged from 85% to 90%, as determined by thetrypan blue exclusion assay.

Hepatocytes were cultured in a modification of the collagensandwich configuration as described by Dunn et al.14,15 Glasscoverslips placed in 24-well plates were precoated with0.20 mL of collagen gelling solution prepared by mixing nineparts of collagen stock (1.25 mg/mL) with one part of10 · M199, and incubated at 37�C for 60 min for gel formationbefore cell seeding. Collagen stock solution was prepared fromrat-tail tendons using the procedure from Beken et al.16 Pri-mary hepatocytes were seeded with T1 medium at density of2 · 105 cell per coverslip and left at 37�C and 5% CO2. Fivehours after seeding, the medium was removed and replacedwith FBS-free T2 medium. The following day (day 1), themedium was removed again and a second layer of collagenwas allowed to gel over the cells. Medium was changed dailyand hepatocyte viability was assessed for 4 days after isolation.

Adipose tissue isolation and culture

Visceral adipose tissue (VAT) was obtained from the sameWistar rats used for hepatocyte isolation.

VAT was minced, capillaries were removed, and adipo-cytes isolated by collagenase treatment to estimate the

FIG. 1. In the dynamic connected culture configuration,three culture chambers, housing hepatocytes or visceraladipose tissue (VAT), were connected to a mixing chamberand to a peristaltic pump.

1636 GUZZARDI ET AL.

number of cells per milligram of tissue. Cell size was as-sessed using an optical microscope (AX70; Olympus Italy,Milan with a 40 · objective). Then, we calculated the numberof cells per mass of fat tissue by dividing the volume of eachpiece by the average size of fat cells, assuming that VAT iscomposed solely of adipocytes. At the beginning of the ex-periment VAT was placed in the bioreactor chamber or in aPetri dish for dynamic or static experiments, respectively,and coated with alginate gel as described previously.11 Thethin alginate coating prevents movement of VAT duringdynamic experiments.

Cell ratio in connected culture experiments

For the connected culture experiments the ratio betweenhepatocytes and adipocytes was estimated using data fromhttp://msis.jsc.nasa.gov/, which tabulates individual organmass and body fat percentiles. It was estimated that in anaverage adult male abdomen, which comprehends liver,visceral fat, endothelial cells, and pancreatic b-cells, hepato-cytes represent 78% of cellular mass, whereas adipocytes areabout 14.4%. Therefore, 20–25 mg of VAT (corresponding toabout 6.5 · 104 - 7.5 · 104 fat cells) was used with 4 · 105 he-patocytes seeded in each system (two coverslips per experi-ment) to achieve the calculated ratio.

Static cultures

Hepatocytes and VAT were cultured separately and to-gether (static connected culture) in static conditions. Experi-ments were carried out placing two collagen sandwichedhepatocyte coverslips or 20–25 mg of VAT or both about 1 cmapart in the center of a 10-cm-diameter Petri dish with 15 mLof T2 medium. Hepatocytes and adipocytes were not in di-rect contact, but they were allowed to exchange metabolitesby the diffusive flux of the medium in the same Petri dish.Medium samples were withdrawn after 2, 4, 6, and 24 h forbiochemical analysis, whereas cell viability was assessedat the end, after 24 h. All experiments were performed intriplicate.

Dynamic cultures

Dynamic experiments were carried out in the MCmB withhepatocytes or VAT or both in a combination called ‘‘dy-namic connected culture’’ using the same cell numbers andmedia volumes as the static cultures. One, two, or threechambers were connected together depending on the celltypes used: one chamber for adipose tissue, two chambersfor collagen sandwiched hepatocytes (one for each cover-slip), and three for the complete system. Bioreactor chamberswhere then connected to a mixing reservoir and to theperistaltic pump and filled with 15 mL of T2 medium.Withdrawal of media samples and cell viability assessmentwere performed at the same time intervals established forstatic cultures. All experiments were performed in triplicate.

Assays

Viability and metabolite concentrations were evaluated inboth dynamic and static cell cultures.

Cell counting and viability. Hepatocyte number and via-bility were assessed after isolation by the trypan blue ex-

clusion assay. Hepatocyte viability was assessed at the end ofall experiments by the CellTiter�Cell Viability Assay (Pro-mega). Five hundred microliters of medium and 50mL ofreagent were used per each well in a 24-well plate; the in-crease in fluorescence emission, read at 590 nm in 2 h timeframe, represented cell viability.

Biochemical analysis. Glucose, urea, glycerol, and freefatty acid (FFA) concentrations were assessed using colori-metric commercial kits (Glucose Test Kit, Megazyme Inter-national Ireland, Ltd.; Urea Kit, Sigma-Aldrich and BeckmanCoulter, Inc.; Free Fatty Acid Quantification Kit, Biovision,Inc.) in accordance with the manufacturer’s instructions. Ratalbumin concentration was determined by an ELISA im-munochemical assay (Bethyl Laboratories). All the resultswere normalized to the volume and the cell number in theculture system at each time point.

Morphology. Hepatocyte morphology and adipocyte sizewere analyzed using an optical microscope with a 20 · and40 · objective.

Statistical analysis

Results were calculated from at least three independentexperiments and expressed as means and standard deviationof the mean. Data were analyzed by two tails Student’s t-test.Statistical significance was set at p < 0.05.

Results

Biochemical assays

Urea secretion. Urea secretion is a marker of hepaticfunction. Hepatic urea synthesis was slightly higher in dy-namic cultures with respect to the corresponding staticconditions. However, no difference was found between he-patocyte monocultures and connected cultures (Fig. 2a, b).VAT, as expected, did not synthesize urea, as its amountremained constant over 24 h. Figure 2c shows the rate ofurea secretion at 24 h in monoculture and connected cultureboth in static conditions and in the bioreactor. As shownin the figure, the presence of VAT did not significantlyaffect urea secretion, which was increased by the flowboth in monoculture and in connected culture (4.98 – 0.97 vs.2.46 – 1.32 mg/h, p = 0.12 in hepatocyte monocultures and4.50 – 0.22 vs. 1.86 – 0.28 mg/h, p = 0.009 in connected cul-tures).

Albumin synthesis. Albumin is an important marker ofsynthetic function in hepatocytes. Figure 3a shows thatafter 24 h in static culture hepatic albumin secretion wasincreased in the presence of VAT with respect to hepato-cyte monoculture (0.48 – 0.01 mg/mL vs. 0.38 – 0.06 mg/mL,p = 0.11). However, the effect of the cross-talk between he-patocytes and VAT is more evident in the dynamic setting;here the albumin released in the connected culture was morethan twofold higher than in the hepatocyte monoculture(0.78 – 0.15 mg/mL vs. 0.29 – 0.13 mg/mL, p = 0.049, Fig. 3b).As expected, VAT did not produce albumin. The results to-gether show that hepatic albumin release was principallymodulated by the flow-mediated cross-talk with VAT in theMCmB connected culture. The graph shows no difference in

METABOLIC CONTROL THROUGH HEPATOCYTE AND ADIPOSE TISSUE CROSS-TALK 1637

FIG. 2. Urea synthesis in static (a) and dynamic (b) cultures. (c) Urea synthesis rate for hepatocyte monoculture andhepatocyte and VAT connected culture in static and dynamic condition over 24 h: *p < 0.01.

FIG. 3. Albumin production in static (a) and dynamic (b) cultures. (c) Albumin production rate for hepatocyte monocultureand hepatocyte and VAT connected culture in static and dynamic conditions over 24 h. *p < 0.05.

1638 GUZZARDI ET AL.

the first 6 h of the experiment; however, a major change be-tween static and dynamic culture is evident at the last timepoint (24 h).

Figure 3c highlights the effect of VAT on albumin syn-thesis showing that the production rate increased when theconnection of hepatocytes with adipose tissue was mediatedby the flow of the medium with respect to all the otherculture conditions considered.

Glycerol concentration. Glycerol, together with fatty ac-ids, is a by-product of triglyceride lipolysis. Glycerol wasanalyzed both in static cultures and in the MCmB. Hepato-cyte monocultures did not release glycerol in dynamic orstatic conditions. On the other hand, in dynamic cultureglycerol released from VAT increased over time from0.59 – 0.23 mM to 1.88 – 0.50 mM in 24 h, but when VAT andhepatocytes were present together, glycerol concentrationwas more stable (about 0.40 – 0.15, Fig. 4b). These resultssuggest that hepatocytes in the dynamic connected cultureserve as a buffer, preventing an increase in glycerol con-centration and maintaining it close to 0.3 mM, the rat phys-iological fasting state level,17 at least in the first 24 h. In allstatic cultures we found only a little glycerol release and nosignificant differences between VAT monoculture and con-nected culture (Fig. 4a). These results together suggest thatthe flow stimulates triglyceride lipolysis in VAT leading tothe release of glycerol, which is taken and metabolized byhepatocytes, in a sort of ‘‘in vitro homeostasis.’’

FFA concentration. FFA concentration was measured atthe end of the 24 h. In static cultures, FFA concentrations werenegligible and no difference was detected between connectedculture and either hepatocytes or VAT monoculture (Fig. 4c).In the dynamic setting, on the other hand, higher concentra-

tions were measured, suggesting an effect due to the flow ofthe medium. Both hepatocyte and VAT monoculture releasedFFAs (Fig. 4c); however, in the connected culture, their con-centration was comparable to that observed in the hepatocytemonoculture. It is likely that the FFA secretion rate, as afunction of the medium concentration, was modulated by theconcurrent release by both cell types. However, because FFAconcentration is regulated through release and uptakemechanisms in both cell types, we cannot indicate here whichbiochemical pathways and metabolic fluxes are up- ordownregulated in the two cell types, but the dynamic culturecould be useful to study the mechanisms involved.

Glucose concentration. According to energetic require-ments and substrate availability, primary hepatocytes canboth take up and release glucose switching from glycolysisand glycogenesis to gluconeogenesis and glycogenolysispathways and vice versa. Figure 5 shows glucose concen-tration profiles for the mono and the connected cultures inboth static and dynamic conditions. In VAT monocultureswe did not find any differences between static and dynamicconditions. In the hepatocyte monoculture and in the con-nected culture the net glucose concentration, resulting fromrelease and consumption processes, slightly increased over24 h in static condition, whereas it stayed stable in dynamic.The dynamic cultures have a higher metabolic rate, as shownby urea, albumin, glycerol, and FFA metabolism; however,this is not reflected in a significant increase in glucose con-sumption.

Discussion

The liver has a central role in major functions of the or-ganism. In fact it is responsible for the uptake, conversion,

FIG. 4. Glycerol concentrations in static (a) and dynamic (b) cultures. (c) Free fatty acid (FFA) concentrations in static anddynamic cultures at the end of the 24 h experiments. *p < 0.05.

METABOLIC CONTROL THROUGH HEPATOCYTE AND ADIPOSE TISSUE CROSS-TALK 1639

and bioavailability of many of the nutrients entering the di-gestive tract. It is involved in glucose and lipid homeostasis,and is also the principal organ involved in xenobiotic bio-transformation.18 Considerable effort has been made to es-tablish in vitro hepatocyte models for pharmacological testingor hepatocyte-based systems for therapeutic support. Despitethe tremendous progress, hepatocyte function is known to becompromised in vitro, and it is thought that improving the cellenvironment by coculture with parenchymal9 and non-parenchymal cells or using 3D interactions19–22 can help sta-bilize hepatocyte phenotype. Recently, Jindal et al.23 showedthat the effect of endothelial cells on hepatocyte function is notdue to cell–cell junctions, but it is mediated through solublesignaling, in particular by the amino acid proline released byendothelial cells and circulating in the medium.

In this work we explored a novel in vitro metabolic modelapplying both biomechanical and biochemical stimuli, withthe aim of improving the in vitro cross-talk between hepa-tocytes and VAT. This represents the first step toward therealization of a comprehensive metabolic model with liver,fat, pancreas, and muscle cells. Two features distinguish thestudy: the cell types involved, whose cross-talk in vitro hasnot been widely explored yet, and the type of cell–cell in-teraction, which is achieved by a dynamic flow-dependenttransport of molecules from one chamber to the others.

The cross-talk between hepatocytes and adipocytes wasexplored as liver and adipose tissue are involved in manyphysiological and pathological processes. In our experi-mental conditions, we found notable differences in metabolicand functional profiles between static and dynamic settingsand in mono- versus connected cultures.

Hepatocyte urea synthesis is principally modulated by theflow, since its synthesis rate in the dynamic cultures is morethan twofold higher with respect to the corresponding staticconditions (Fig. 2c). However, no further improvement dueto the cross-talk between the two cell types was observed. Inthe first 2 h, more urea is synthesized in the static cultureswith respect to the corresponding dynamic condition. Thedynamic cultures have an initial lag, probably due to celladaptation to the new dynamic setting, followed, however,by an increase in urea synthesis. Albumin secretion, on theother hand, is markedly increased by the dynamic cross-talkof hepatocytes and visceral fat tissue. Its concentration ismuch higher only at the last time point, probably as theresult of a gradual increase in the synthesis and release of theprotein, which appears in the medium after 6 or more hoursfrom the beginning of the experiment. After 24 h albumin

production in dynamic connected culture is about 2.5-foldhigher than in hepatocyte monoculture. Therefore, the flowin the MCmB enhances the cross-talk between hepatocytesand adipocytes, resulting in an increase in protein expressionwith respect to both dynamic monoculture and static con-nected culture.

In the static setting there is no glycerol release in any of thethree cultures, whereas in the dynamic setting glycerol iscontinuously released by adipose tissue but not by hepato-cytes. Further, in the connected culture its concentration sta-bilizes around 0.4 mM, close to the physiological value infasting rats,17 suggesting an interaction between the cell typesleading to homeostasis-like control. We hypothesize thatglycerol release is due to lipolysis of triglycerides in adiposetissue, which does not use glycerol circulating in the mediumfor re-esterification of fatty acid because of its constitutivelylow expression of the enzyme glycerol kinase.24 Thus, while inVAT monoculture its concentration increases over time, indynamic connected culture glycerol can be up taken by he-patocytes. Also FFA concentration is higher in dynamic ex-periments than in static conditions, where it is negligible(Fig. 4c). In vivo VAT controls FFA concentration regulatingboth in- and out-fluxes through triglyceride lipolysis, fatty acidb-oxidation, and esterification. In dynamic cultures FFA is re-leased by both hepatocytes and VAT; however, there is nofurther increase in the connected culture with respect to thehepatocyte monoculture. We hypothesize that the FFA loadsecreted by VAT leads to a deceleration of hepatic FFA release,and thus the connected system manifests a metabolic self-regulation. Moreover, the high FFA circulating level couldrepresent a cellular signal to enhance secretion of albumin, whichbinds and transports lipids (such as circulating FFAs) insidehepatic cells keeping their extracellular concentration stable.

Adipocytes cooperate in glucose homeostasis in physio-logical conditions by storing or releasing gluconeogenesissubstrates (such as glycerol), whereas hepatocytes play acritical role in glucose metabolism storing and producingglucose as necessary.25,26 In our experimental conditions, nosignificant differences in glucose concentration were foundbetween static culture and dynamic conditions, although indynamic cultures metabolic functions were enhanced interms of albumin secretion, urea synthesis, and glycogen andFFA homeostasis. This might be because the cells use othersources of energy available to maintain homeostasis.

In all our experiment we observed an upregulation ofmetabolic profiles in the presence of flow. Convective flowincreases both the rate of nutrient and oxygen supply, as well

FIG. 5. Glucose concentration in dynamic and static cell cultures. (a) Hepatocytes alone, (b) hepatocytes and adipocytes inconnection, and (c) adipocytes alone.

1640 GUZZARDI ET AL.

as the clearance of metabolic and catabolic products. In ad-dition, the cells are stimulated by fluid shear stress, at valuesclose to interstitial shear stresses on cells estimated by Ped-ersen et al.27 In a previous report we showed that flow rateslower than 80 mL/min do not increase metabolic functionwith respect to controls, whereas above 500mL/min theshear stress may damage cells as their viability decreasesrapidly.12 This suggests that below 80 mL/min there is nobenefit in convective flow, whereas above 500mL/min thebenefits of convection are outweighed by the damaging ef-fect of mechanical forces. We hypothesize that nutrientturnover and dynamic flow act synergically to improvemetabolic function in the MCmB.

Many studies report that hepatic insulin resistance could bedirectly related to visceral fat.25,28 Visceral fat deposits areanatomically close to the liver and substances derived fromvisceral adipocytes enter directly into the portal vein. Inter-action between the two tissues could be mediated by lowconcentrations of visceral fat factors as suggested by Wanget al.29 These factors play a major role in the induction ofinsulin resistance and impairment of glucose uptake. Growingevidence indicates that inflammatory adipokines, which in-clude tumor necrosis factor-alpha (TNF-a), interleukin-6, andinterleukin-1b, are involved in the pathways that lead to im-pairment of glucose regulation and metabolic disorders.30,31

An in vitro system where several cell types involved in ho-meostatic control are connected and interact in a physiologicmanner could fill the gap between in vitro and in vivo studies.

Conclusion

In this work we show that a connected culture with he-patocytes and VAT in a dynamic system not only improvesliver-specific functions but also results in an enhanced cross-talk between the two cells types involved. The dynamicconnected culture enables tissue interactions to be re-produced in vitro and can be used to explore liver and vis-ceral fat interaction in pathophysiological conditions.Moreover, it paves the way for the establishment of morecomplex in vitro tissue and organ models of the metabolicand other physiological systems.

Disclosure Statement

No competing financial interests exist.

References

1. Meyer, C., Dostou, J.M., Welle, S.L., and Gerich, J.E. Role ofhuman liver, kidney, and skeletal muscle in postprandialglucose homeostasis. Am J Physiol Endocrinol Metab 282,

E419, 2002.2. Kulig, K.M., and Vacanti, J.P. Hepatic tissue engineering.

Transpl Immunol 12, 303, 2004.3. Kern, A., Bader, A., Pichlmayr, R., and Sewing, K.F. Drug

metabolism in hepatocyte sandwich cultures of rats andhumans. Biochem Pharmacol 54, 761, 1997.

4. Donato, M.T., Gomez-Lechon, M.J., and Castell, J.V. A mi-croassay for measuring cytochrome P450IA1 and P450IIB1activities in intact human and rat hepatocytes cultured on96-well plates. Anal Biochem 213, 29, 1993.

5. Kobayashi, N. Life support of artificial liver: development ofa bioartificial liver to treat liver failure. J HepatobiliaryPancreat Surg 16, 113, 2009.

6. Fiegel, H.C., Kaufmann, P.M., Bruns, H., Kluth, D., Horch,R.E., Vacanti, J.P., et al. Hepatic tissue engineering: fromtransplantation to customized cell-based liver directedtherapies from the laboratory. J Cell Mol Med 12, 56, 2008.

7. Park, J., Li, Y., Berthiaume, F., Toner, M., Yarmush, M.L.,and Tilles, A.W. Radial flow hepatocyte bioreactor usingstacked microfabricated grooved substrates. BiotechnolBioeng 99, 455, 2008.

8. Powers, M.J., Janigian, D.M., Wack, K.E., Baker, C.S., BeerStolz, D., and Griffith, L.G. Functional behavior of primaryrat liver cells in a three-dimensional perfused microarraybioreactor. Tissue Eng 8, 499, 2002.

9. Viravaidya, K., and Shuler, M.L. Incorporation of 3T3-L1cells to mimic bioaccumulation in a microscale cell cultureanalog device for toxicity studies. Biotechnol Prog 20, 590,2004.

10. Vozzi, F., Heinrich, J.M., Bader, A., and Ahluwalia, A.D.Connected culture of murine hepatocytes and HUVEC in amulticompartmental bioreactor. Tissue Eng Part A 15, 1291,2009.

11. Guzzardi, M.A., Vozzi, F., and Ahluwalia, A.D. Study of thecrosstalk between hepatocytes and endothelial cells using anovel multicompartmental bioreactor: a comparison be-tween connected cultures and cocultures. Tissue Eng Part A15, 3635, 2009.

12. Mazzei, D., Guzzardi, M.A., Giusti, S., and Ahluwalia, A. Alow shear stress modular bioreactor for connected cell cul-ture under high flow rates. Biotechnol Bioeng 106, 127, 2010.

13. Seglen, P.O. Preparation of isolated rat liver cells. MethodsCell Biol 13, 29, 1976.

14. Dunn, J.C., Tompkins, R.G., and Yarmush, M.L. Long-termin vitro function of adult hepatocytes in a collagen sandwichconfiguration. Biotechnol Prog 7, 237, 1991.

15. Dunn, J.C., Tompkins, R.G., and Yarmush, M.L. Hepatocytesin collagen sandwich: evidence for transcriptional andtranslational regulation. J Cell Biol 116, 1043, 1992.

16. Beken, S., Vanhaecke, T., De Smet, K., Pauwels, M., Ver-cruysse, A., and Rogiers, V. Collagen-gel cultures of rat he-patocytes: collagen-gel sandwich and immobilizationcultures. Methods Mol Biol 107, 303, 1998.

17. Gambert, S., Helies-Toussaint, C., and Grynberg, A. Extra-cellular glycerol regulates the cardiac energy balance in aworking rat heart model. Am J Physiol Heart Circ Physiol292, H1600, 2007.

18. Guillouzo, A., Morel, F., Langouet, S., Maheo, K., and Rissel,M. Use of hepatocyte cultures for the study of hepatotoxiccompounds. J Hepatol 26 Suppl 2, 73, 1997.

19. Bhatia, S.N., Balis, U.J., Yarmush, M.L., and Toner, M. Mi-crofabrication of hepatocyte/fibroblast co-cultures: role ofhomotypic cell interactions. Biotechnol Prog 14, 378, 1998.

20. Bhatia, S.N., Balis, U.J., Yarmush, M.L., and Toner, M. Effectof cell-cell interactions in preservation of cellular phenotype:cocultivation of hepatocytes and nonparenchymal cells.FASEB J 13, 1883, 1999.

21. Dunn, J.C., Yarmush, M.L., Koebe, H.G., and Tompkins,R.G. Hepatocyte function and extracellular matrix geometry:long-term culture in a sandwich configuration. FASEB J 3,

174, 1989.22. Nishikawa, M., Kojima, N., Komori, K., Yamamoto, T., Fujii,

T., and Sakai, Y. Enhanced maintenance and functions of rathepatocytes induced by combination of on-site oxygenationand coculture with fibroblasts. J Biotechnol 133, 253, 2008.

23. Jindal, R., Nahmias, Y., Tilles, A.W., Berthiaume, F., andYarmush, M.L. Amino acid-mediated heterotypic interaction

METABOLIC CONTROL THROUGH HEPATOCYTE AND ADIPOSE TISSUE CROSS-TALK 1641

governs performance of a hepatic tissue model. FASEB J 23,

2288, 2009.24. Guan, H.P., Li, Y., Jensen, M.V., Newgard, C.B., Steppan,

C.M., and Lazar, M.A. A futile metabolic cycle activatedin adipocytes by antidiabetic agents. Nat Med 8, 1122,2002.

25. Girard, J., and Lafontan, M. Impact of visceral adipose tissueon liver metabolism and insulin resistance. Part II: visceraladipose tissue production and liver metabolism. DiabetesMetab 34, 439, 2008.

26. Lafontan, M., and Girard, J. Impact of visceral adipose tissueon liver metabolism. Part I: heterogeneity of adipose tissueand functional properties of visceral adipose tissue. DiabetesMetab 34, 317, 2008.

27. Pedersen, J.A., Boschetti, F., and Swartz, M.A. Effects ofextracellular fiber architecture on cell membrane shear stressin a 3D fibrous matrix. J Biomech 40, 1484, 2007.

28. Carey, D.G., Jenkins, A.B., Campbell, L.V., Freund, J., andChisholm, D.J. Abdominal fat and insulin resistance innormal and overweight women: direct measurements reveala strong relationship in subjects at both low and high risk ofNIDDM. Diabetes 45, 633, 1996.

29. Wang, Z., Lv, J., Zhang, R., Zhu, Y., Zhu, D., Sun, Y., et al.Co-culture with fat cells induces cellular insulin resistance inprimary hepatocytes. Biochem Biophys Res Commun 345,

976, 2006.30. Wellen, K.E., and Hotamisligil, G.S. Obesity-induced in-

flammatory changes in adipose tissue. J Clin Invest 112,

1785, 2003.31. Coppack, S.W. Pro-inflammatory cytokines and adipose

tissue. Proc Nutr Soc 60, 349, 2001.

Address correspondence to:Maria Angela Guzzardi, Ph.D.Institute of Clinical Physiology

CNRVia Moruzzi 1

56124 PisaItaly

E-mail: mariguzz@gmail.com

Received: September 13, 2010Accepted: February 7, 2011

Online Publication Date: March 10, 2011

1642 GUZZARDI ET AL.

This article has been cited by:

1. Mr. Adam L Larkin, Mr. Richard R Rodrigues, Prof. T. M. Murali, Prof. Padmavathy Rajagopalan. Designing a Multi-cellularOrganotypic 3D Liver Model with a Detachable, Nanoscale Polymeric Space of Disse. Tissue Engineering Part C: Methods 0:ja. .[Abstract] [Full Text PDF] [Full Text PDF with Links]

2. Jong H. Sung, Mandy B. Esch, Jean-Matthieu Prot, Christopher J. Long, Alec Smith, James J. Hickman, Michael L. Shuler.2013. Microfabricated mammalian organ systems and their integration into models of whole animals and humans. Lab on a Chip13:7, 1201. [CrossRef]

3. Jong Hwan Sung, Michael L. Shuler. 2012. Microtechnology for Mimicking In Vivo Tissue Environment. Annals of BiomedicalEngineering . [CrossRef]

4. Andries D. van der Meer, Albert van den Berg. 2012. Organs-on-chips: breaking the in vitro impasse. Integrative Biology .[CrossRef]

5. Bruna Vinci, Ellen Murphy, Elisabetta Iori, Francesco Meduri, Silvia Fattori, Maria Cristina Marescotti, Maura Castagna,Angelo Avogaro, Arti Ahluwalia. 2011. An in vitro model of glucose and lipid metabolism in a multicompartmental bioreactor.Biotechnology Journal n/a-n/a. [CrossRef]

6. Federico Vozzi, Daniele Mazzei, Bruna Vinci, Giovanni Vozzi, Tommaso Sbrana, Leonardo Ricotti, Nicola Forgione, ArtiAhluwalia. 2011. A flexible bioreactor system for constructing in vitro tissue and organ models. Biotechnology and Bioengineeringn/a-n/a. [CrossRef]