Biotechnol. J.

2012, 7, 117126

DOI 10.1002/biot.201100177

www.biotechnology-journal.com

Research Article

An in vitro model of glucose and lipid metabolism in a multicompartmentalbioreactor

3 2 Bruna Vinci1,*, Ellen Murphy2,*, Elisabetta Iori2, FrancescoMeduri , Silvia Fattori4, Maria Cristina Marescotti , 4 2 1 Maura Castagna , Angelo Avogaro and Arti Ahluwalia
1 2

Centro Interdipartimentale di Ricerca E. Piaggio, Faculty of Engineering, University of Pisa, Pisa, Italy Dipartimento di Medicina Clinica e Sperimentale, Divisione di Malattie del Metabolismo, University of Padua, Padua, Italy 3 Dipartimento di Chirurgia Generale Patologia Speciale, Azienda Ospedaliera di Padova, Padua, Italy 4 Dipartimento di Chirurgia, Anatomia Patologica III, Azienda Ospedaliero Universitaria Pisana, Pisa, Italy

The energy balance in vivo is maintained through inter-organ cross-talk involving several different tissues. As a first step towards recapitulating the metabolic circuitry, hepatocytes, endothelial cells and adipose tissue were connected in a multicompartmental modular bioreactor to reproduce salient aspects of glucose and lipid metabolism in vitro. We first examined how the two-way cellular interplay between adipose tissue and endothelial cells affects glucose and lipid metabolism. The hepatocyte cell line HepG2 was then added to the system, creating a three-way connected culture, to determine whether circulating metabolite concentrations were normalized, or whether metabolic shifts, which may arise when endothelial cells and adipose tissue are placed in connection, were corrected. The addition of hepatocytes to the system prevented the drop in the concentrations of glucose, L-alanine and lactate, and the rise in that of free fatty acids. There was no significant change in glycerol levels in either of the connected cultures. The results show that connected cultures recapitulate complex physiological systemic processes, such as glucose and lipid metabolism, and that the HepG2 hepatocytes normalize circulating metabolites in this in vitro environment in the presence of other cell types.

Received 26 March 2011 Revised 24 June 2011 Accepted 18 July 2011

Keywords:Adipose tissue Endothelial cells Hepatocytes In-vitro model Metabolism

1 Introduction
It is now widely accepted that current in-vitro cell culture systems are poorly representati ve of human or animal physiolog y. This has been generally attributed to the fact that the compl exity of the physiological environment is not replicated in con-

Correspondence Professor Arti Ahluwalia, Centro Interdipartimentale di : Ricerca E. Piaggio, Faculty of Engineering, University of Pisa, Via Diotisalvi 2, 56125 Pisa, Italy E-mail: arti.ahluwalia@centropiaggio.unipi.it Abbreviations: AT, adipose tissue; ECG M endothelia cell growth medium; , FBS, fetal bovine serum; FFA, free fatty acid; HepG2, human hepatocellular liver carcinoma cell line; HUVEC human umbilical vein endothelial cell; , LFC, laminar flow chamber; MCmB, multicompartmental modular bioreactor; PLG A poly(lactic-co-glycolic) acid ,

ventional culture s. In fact, all cells are exquisitely sensiti ve to their micro-e nvironment, which is rich with 3-D cues from the extracellular matrix, other cells and from mechanical stimuli due to flow, concentration gradients and mov ement. A variety of methods have been proposed to refine the simplistic in-vitro representations that are currently used in cell-culture laboratories worl dwid e. For example, since the cross-talk between different tissues is important in modulating and enhancing cell func- tion, the use of conditioned media or the addition of growth factors generated by stromal or other cells is increasingly common, as are coculture s. Recent approaches use microscale device s, combining cell

* These authors contributed equally to this work.

 2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

117

Biotechnolo gy Journal

Biotechnol. J. 2012, 7, 117126

culture and microfluidics to fabricate cells and or- gans-on-chips [1]. We have developed a modular multicompartmental bioreactor (MCmB) system that enables mi- crowell protocols to be transferred directly to the bioreactor modules without redesign of cellcul- ture experiment s. Each module can be addressed and interrogated separately and different cell types can be added stepwise to the system. As explained previously [2], the MCmB design principles are based on allometric scaling of cell numbers and the mean residence time of molecules in metabolic tis- sue s, as well as considerations on oxygen tension and shear stres s, which together can be combined to establish organ and system model s. Here we use the MCmB to construct an in-vitro model of glucose and lipid metabolism starting from three tissue types known to play a fundamental role in metabolic homeostasis: liver, endothelium and adipose tissue (AT). Over 40 years ago, it was proposed that free fatty acids (FFAs) might compete with glucose as the main energy substrate in some tissue s, leading to decreased glucose oxidation when FFAs are el evated [3, 4]. Clinical studies have suggested that circulating metabolites display a reciprocal relationship between hepatic FFA uptake and glucose/lactate flux, which may deri ve from intrahepatic substrate competition [5]. AT releases FFAs into the circulation through lipolysi s. The liver is important in processing FFAs; therefor e, adequate liver function is critical in shuttling excess FFAs out of the circulation to limit consequent damage to the endothelium. This compl ex interpl ay between tissues is impossible to recapitulate in current cell-culture systems. Indeed, while a great deal of attention has been paid to identifying path ways and mechanisms in single cells, our understanding of inte r-organ cross-talk lags far behind. Besides the coculture of hepatocytes with non-parenc hymal cells, the implementation of integrated in-vitro models to stu dy the biochemical and molecular path ways involved in the development of metabolic disorders is still scarc e, primarily due to the difficulty of reproducing the physiological interaction between tissues connected in the body by the bloodstream and the molecules transported through it. In a previous paper [6] we cultured the three different cell/tissue types separately in the MCmB bioreactors and demonstrated that flow plays an important role in modulating their metabolic profiles. Her e, endothelial cells (HUVECs), AT, and he- patocytes are connected together to investigate metabolic cross-talk between AT and endothelium and the role of hepatic cells in regulating circulat- ing metabolite s.

Materialsand methods

2.1 Experimentaldesign
The MCmB is a modular system in which different bioreactor chambers are connected together by the flow of media. The configuration described here is composed of three module s. The hepatic and adi- pose tissue modules are the low-shear stres s, high- flow MCmB 2.0 chambers described previously [7], while the module used for the endothelial cell cul- ture is a laminar flow chamber (LFC) [8]. All the chambers are connected in a closed loop as shown in Figs. 1A and B. In addition, there is a mixing chamber for oxygenation and for the addition or sampling of medium, as well as a peristaltic pump (Ismatech IPC-4, Zurich, Switzerland) and silicone tubing, which connects the cell chamber s. To establish an in-vitro model of metabolism, we focused on the visceral compartment of the human abdomen to determine physiologically relevant cell ratios and transit time s. The visceral region was chosen firstly for its rel evance to shortterm metabolism, and secondly because the three tissues used make up a substantial fraction of the total cell number in this compartment. Using data available from the literature and physiological databases [2], the final hepatocyte:adipocyte:endothelial cell ratio was estimated to be 10:2:1. The number of cells added to each chamber at the beginning of the experiments was calculated to reach these ratios by the completion of each experiment. Shear stress is particularly critical for hepatocyte s, which may suffer damage at high flow rates; the flow rate was therefore fixed at 250 L/min, which we have shown to be optimal for rat and human hepatocytes [7, 9], and represents a fluid residence time of about 10 min in each modul e, comparable with the mean organ perfusion time of about 37 min in the human liver [10]. At this flow rate the wall shear stres s, calculated using finite element method s, was 0.002 Pa in the LFC and 105 Pa in the MCmB 2.0.

2.2 Cell/tissue sources

All reagents were from Sigma-Aldrich (Milan, Italy) unless otherwise specified. Omental AT was obtained, with informed consent, from surgical inter ventions in non-diabetic subjects free of known metabolic diseas e. After treatment with collagenase type II in Hanks balanced salt solution (HBSS), the partially digested tissue was placed on a 200-m mesh filter and rinsed with DMEM to re- move blood vessel s. Floating partially digested AT was then divided into aliquots and transferred to

118

2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

Biotechnol. J. 2012, 7, 117126

www.biotechnology-journal.com

Figure. 1. The multicompartmental bioreactor modules: 3-way connected culture of HepG2 hepatocytes and adipose tissue in the round chambers and endothelial cells in the rectangular LFC. The mixing chamber is a partially filled plastic or glass bottle into which media droplets are passively oxygenated as they return from the culture chambers. (A) Experimental set up; (B) Connection scheme showing chamber dimensions in mm.

DMEM:F12 with 20% fetal bovine serum (FBS) for subsequent experiment s. For each condition test- ed, 200 mg partially digested tissue containing ap- proximately 300 000 adipocytes was used. The par- tial digestion all owed us to concentrate adipocytes without destr oying the collagen matrix, which is important for preventing dedifferentiation in the chamber s. HUVECs from Promocell (Heidelberg, Germa ny) were used to m odel the endothelium. The cells were cultured in endothelial cell basal medium (ECGM, Promocell) supplemented with 10% FBS, 0.1 ng/mL epidermal growth facto r, 1.0 ng/mL basic fibroblast growth facto r, 0.4% endothelial cell growth supplement/heparin, 1.0 g/mL hydrocortisone (Promocell), 100 U/mL penicillin and 100 g/mL strepto mycin. HUVECs were used up to the 4th passag e. Hepatocytes were from the human hepatocellular liver carcinoma (HepG2) cell line. Although their xenobiotic metabolic functions are kn own to be compromised, their endogenous meta- bolic functions are largely intact. They were grown in Eagle s minimal essential medium (EMEM) with 1 g/L glucose supplemented with 5% FBS, 1% nonessential amino acid s, 1% EMEM vitamin s, 2 mM L-glutamin e, 100 U/mL penicillin and 100 g/mL strepto mycin and used up to passage 22.

architectur e. 3-D scaffolds of poly(lactic- co-glycolic) acid (PLGA 75:25, MW 270 000; Boehringer Ingelheim, Ingelheim, Germa ny) were microfabricated using a pressure assisted microsyringe (PAM) [11]. The scaffolds can host high-density function- al cultures of HepG2 hepatocytes for up to 1 week and their fabrication, preparation and characterization have been reported [12]. The PLGA scaffolds used in this work were composed of three layers of hexagonal elements with sides of 500 m and ov er- all dimensions of 1 cm 1 cm 100 m. Scaffolds were routinely assessed for degradation after the end of the experiment by measuring changes in mas s, morphology and elastic modulus [13]. No ob- servable changes were detected, as verified by sim- ilar studies on this copolymer [14].

2.4 Cell culture

All cultures were carried out using ECGM containing 10% FBS. Preliminary experiments were performed to verify that conditioned medium in the bioreactor system does not contribute significantly to the metabolic profile of cells and that the vitali- ty and morphology of all three cell types were con- ser ved with respect to their standard media [15]. In the two-way connected cultur e, the AT chamber was placed in series with the LFC containing the HUVEC s. In the three- way connected cultur e, the liver chamber was added in serie s. The components of the bioreactor system were sterilized using H2O2 gas plasma before each use and assembled under a

2.3 3-D scaffolds for HepG2 hepatocyte culture

One of the most important influence hepatocyte function in presence of a 3-D factors that vitro is the

 2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

119

Biotechnolo gy Journal

Biotechnol. J. 2012, 7, 117126

laminar flow hood, connecting the three cell chamber s, the pum p, and the mixing chamber via tubing as shown in Fig. 1. The AT was placed in the MCmB 2.0 and 1 mL medium was added to the chambe r. The top of the chamber was layered with a pr ewetted 200-m nylon mesh san dwiched with 240- m nylon mesh circles to pr event movement of adipocytes out of the chamber and into the tubing [6]. The HUVECs were seeded onto a glass coverslip and all owed to adher e. The coverslip was then placed in the bottom of the LFC. HepG2 hepatocytes were seeded on collagencoated PLGA scaffolds placed on 12-mm glass coverslips in 24-well microplates (BD Bioscience s, Buccinasc o, Italy) at a density of 100 000 cells per scaffold in 2 mL ECGM. At 24 h the scaffolds were moved to a new 24-multiwell plat e. After a further 48 h, when the cells had proliferated to about 250 000 cells per scaffold; the slides were carefully transferred to the MCmB 2.0 and coated with an al- ginate film consisting of 250 L 1% sodium alginate dissol ved in serum -free medium, cross-linked with 50 L 1% CaCl 2. Excess alginate was rem oved with a pipett e. The alginate coating was used to protect the cells from direct mechanical stres s, while allowing adequate nutrient diffusion [10]. When all cells had been added to their respective chamber s, the chambers were closed. Medium was added to the mixing chamber to bring the total medium volume to 15 mL and the pump was turned on. When medium had filled all chamber s, the flow was set to 250 L/min, after which the bioreactor was placed inside a 37C/5% CO2 incubator for 15, 24, or 48 h. After the designated time period, cells/tissues were obser ved under light and/or fluorescent microscopes to confirm cell viabilit y, and medium was collected and stored at 80C for even- tual metabolite dosin g. Each condition was run at least in triplicat e.

analyzed using a microscop e, and

2.5 Cell viabilityand metabolite dosing

Before and after incubation, an aliquot of AT was digested with collagenas e, 1 mg/mL, in HBSS at 37C for 30 min; DMEM medium containing 10% FBS was then added, and cells were centrifuged. Floating cells were suspended in PBS containing Hoechst 33258. Adipocytes with Hoechst-positi ve nuclei were obser ved using a fluorescent microscope Olympus AX70 (Olympus Italia, Milan), con- firming cell viability and cell numbe r. There was no visible change in the size or the appearance of the tissue before and after incubation, and cells re- mained free of contamination. HUVEC and HepG2 morphology was

the total cell number for every time point was eval- uated using a Burker chamber and trypan blue to exclude non-viable cells. The number of dead cells counted was always less than 2% of the total. Glycerol, D-lactat e, and L-alanine concentrations were determined by a modified Lloyd ass ay using an automated spectrophotometer Cobas Fara II (Roche) [16]. FFAs were measured by an enzymatic colorimetric method (NE FA C test- Wako Chemicals GmbH, Germa ny) and glucose was determined by a hexokinase-based commercial ass ay (Glucosio HK CP-HoribaABX, Italy). Finall y, albumin (a marker of hepatic function as well as a carrier for FFAs) was ass ayed for the con- nected culture conditions using an enzymelinked immunosorbent ass ay specific for human albumin (Bet hyl Laboratorie s, Montgomer y, TX, USA).

was performed in triplicat e. Statistical significance was set at p<0.05. Unless noted, the p values in the text refer to those obtained comparing metabolite values at 0 h and 48 h using Student s t-test s, and confirmed with Mann- Whitn ey analyse s.

Results

3.1 Glucose
In the two-way connection, the glucose concentration dropped by 3.00 mM in the culture medium ov er 48 h, from 5.20 0.85 mM to 2.20 0.01 mM (p<0.004) (Fig. 2A). This obser vation was similar to what we would expect if the glucose uptake in the two individual cultures were cumulati ve based on previous experiments carried out with monocultures in the MCmB, which showed a 2.60 mM drop in the HUVEC culture and a 0.27 mM drop in the AT culture (Fig. 3A and [6]). In the three- way connection, when the hepatic cell line was added to the system, the medium glucose concentration did not change significantly over 48 h, showing that HepG2 hepatocytes were able to preser ve normal glucose concentrations in the medium, which was not the case in the two-way connection.

2.6 Statistical analysis

Statistica Version 7 was used to carry out analysis of variance (ANOVA), Student s t-test s, and MannWhitn ey test s. Data were expressed as the mean SD. All monoculture and two-way experiments were carried out in triplicat e, while the three- way experiments were repeated six time s. The dosing

120

2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

Biotechnol. J. 2012, 7, 117126

www.biotechnology-journal.com

Figure. 2. Medium metabolites over time in the 2-way and 3-way connected cultures. In the 2-way connection the cell ratio was 2:1 (50 000 adipocytes: 25 000 endothelial cells) whereas the 3-way connection had a ratio of 10:2:1 (250 000 hepatocytes: 50 000 adipocytes: 25 000 endothelial cells). Medium volumes were 15 ml: Data are expressed as means SD (n=3 for the 2-way cultures, n=6 for the 3-way cultures). Time 0 represents the levels of metabolites in fresh medium. (A) Glucose concentration over time in culture; glucose after 48H was significantly different from time zero for the 2-way connected culture (p<0.0004). There was a significant difference between the 2-way and 3-way connected cultures for Glucose over 48 h (p<0.01) . (B) FFA concentration over time in culture; FFAs after 48 h were significantly different from time zero for the 2-way connected culture (p<0.002). There was a significant difference between the 2-way and 3-way connected cultures for FFAs over 48 h (p<0.0007). (C) Glycerol concentration over time in culture. (D) Lactate concentration over time in culture; lactate after 48 h was significantly different from time zero for the 3-way connected culture (p<0.0008). There was a significant difference between the 2-way and 3-way connected cultures for Lactate over 48H (p<0.009). Asterisks denote a significant difference from 0 h value (p<0.05).

3.2 FFAs
In the two- way connection between AT and HUVEC s, there was a 0.097 0.007 mM increase in medium FFA concentrations over 48 h (Fig. 2B), which was less than what would be expected if the FFA release from the two monocultures (from previous experiment s, see Fig. 3B) had been cumulative (0.070 +0.075 mM = 0.145 mM), suggesting a positi ve, but nonetheless blunted FFA release in the two-way connection. In the three- way connection, FFA concentrations did not change significantly over tim e, indicating that the addition of HepG2 prevented a rise in circulating FFAs. This obser vation, together with that of the HepG2 monoculture (Fig. 3B), which showed total FFA disposal ov er 48 h, suggests that the hepatocytes take up excess FFAs from the medium.

3.3 Glycerol
When AT was placed in connection with HUVEC s, there was no significant change in glycerol over 48 h (Fig. 2C). The lack of change ov er time in the two-way connection was surprising becaus e, in the monoculture experiment s, AT and HUVECs both released glycerol into the medium over 48 h, albeit to different degrees: 0.065 mM from AT (p<0.001), 0.022 mM from HUVECs (p<0.03), as reported in Fig. 3C. The obser vation of no net glycerol change in the two-way connected culture suggests a metabolic interpl ay between AT and endothelial cells that results in a shift tow ard cellular glycerol retention. The HepG2 monoculture took glycerol up from the medium over 48 h (0.039 mM, p<0.04) (Fig. 3C). Yet when hepatocytes were connected to

 2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

121

Biotechnolo gy Journal

Biotechnol. J. 2012, 7, 117126

Figure 3. Summary of the changes in medium metabolite concentrations in the MCmB system over 48 h. The bars refer respectively to AT, HUVEC and HepG2 cells in monoculture (n=3) and the 2-way AT-HUVEC (n=3) and 3-way AT-HUVEC-HepG2 connected cultures (n=6). Data are expressed as means SD. (A) Glucose concentrations after 48 h, (B) FFA concentrations after 48 h, (C) Glycerol concentrations after 48 h, and (D) Lactate concentrations after 48 h. Asterisks denote the cultures in which there was a significant difference between 0 h and 48 h (p<0.05).

the AT and HUVEC culture s, there change in glycerol over tim e.

was no net

3.4 Lactate
In the two-way connection, there was no significant change in lactate ov er 48 h despite a trend tow ards net lactate uptake (0.34 0.007 mM, Fig. 3D). There must be cross-talk between AT and endothelial cells that causes a shift in lactate movement becaus e, in contrast the monoculture s, both showed significant lactate release ov er time (Fig. 2D). In the three- way connection, the medium lactate concentrations increased significantly by 1.11 0.39 mM over 48 h, demonstrating net lactate release (p<0.008). HepG2 hepatocytes were able to partially ov erride the inhibition in lactate release in the two-way connected cultur e. Nonetheles s, lac-

tate release was less than what would be expected for a merely cumulati ve effect from the three individual cultures (1.26 + 0.96 + 3.87=6.09 mM) or the AT-HUVEC connection and HepG2 monoculture (0.34 + 3.87=3.53 mM), showing that cell-cell cross-talk is critical in determining circulating lactate levels.

3.5

L-Alanine

L-Alanine

is a nonessential amino acid that is produced from pyru vate by transamination. Within the liver L-alanine is converted back to pyru vate, which is then a source of carbon atoms for gluconeogenesis. This form of energy production is called the alanine-glucose cycle, and it plays a major role in maintaining the bodys blood sugar balanc e. The amino acid was measured in the two- and three-

122

2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

Biotechnol. J. 2012, 7, 117126

www.biotechnology-journal.com

Figure. 4. (A) Medium L-Alanineover time in the 2-way (n=3) and 3-way connected cultures (n=6). Data are expressed as means SD. Time 0 represents the levels of L-Alaninein fresh medium. (B) 2-way and 3-way connected cultures: Summary of the changes in medium L-Alanineconcentrations over 48 h Data are expressed as means SD (n=3 for the 2-way cultures, n=6 for the 3-way cultures). Asterisks denote the cultures in which there was a significant difference between 0 h and 48 h (p<0.0001).

connected cultures to determine if the glucosealanine path way was involved in metabolic regulation in the system. As shown in Fig. 4A, in the twoway connection there was a significant decrease in L-alanine concentrations ov er time (from 0.132 0.003 mM to 0.070 0.003 mM, p<0.0001) demonstrating a net uptake (0.062 0.003 mM) (Fig. 4B). In contrast, in the three- way connection, medium alanine concentrations increased significantly over time (from 0.132 0.011 mM to 0.211 0.03 mM, p<0.0001). The L-alanine release (0.078 0.030 mM) ov er 48 h (Fig. 4B) may be due to both an increase of amino acid synthesis and the conversion of pyru vate by the hepatic cell s.

ser ved over time (an increase of 2.4 0.2 nM in 48 h, p<0.0002, and [6]). In the two-way connection, there was no change in albumin over time because neither AT nor HUVEC synthesize this protein, whereas in the three- way connection there was a significant increase (11.2 0.4 nM over 48 h, p<0.003), as shown in Fig. 5. This represents m ore than a fourfold increase compared to the hepato- cyte monocultur e, demonstrating that connection with AT and endothelial cells promotes albumin synthesis in the HepG2 cell s.

3.6 Albumin
In previous experiments using HepG2 hepatocyte monoculture s, a net increase in albumin was ob-

Discussion

4.1 Glucose concentrations are regulated to normal levels in three-waycultures

There are significant differences between the metabolic profiles in two- and three- way-connected culture s. In the two-way connection, medium concentrations of glucos e, lactate and L-alanine dropped ov er time compared with the three- way cultur e, whereas the FFA level increased. There was no significant change in the glycerol level. In the two-way connection, in contrast to AT and HUVEC monoculture s, FFA release was not associated with glycerol release [6], thus this connection provoked a shift in the association of these two metabolite s. In the three- way connection, increased albumin synthesis and lactate and L-alanine release were obser ved, whereas the other metabolites remained stabl e. The addition of the hepatic cell line prevented the drop in glucose and rise in FFA levels provoked by the two-way connection, demonstrating the capacity of the HepG2

Figure. 5. Human albumin concentration over time in the HepG2 hepatocyte monoculture, the 2-way connection (n=3) and the 3-way connection (n=6): Data are expressed as means SD.

 2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

123

Biotechnolo gy Journal

Biotechnol. J. 2012, 7, 117126

hepatocytes to maintain normal glucose concentrations and prevent any rise in FFAs provoked by the other two tissue s. In the two-way connection, glucose, lactate and L-alanine were taken up by cells. Our pr evious monoculture stu dy demonstrated that glucose uptake in both endothelial cells and HepG2 hepatocytes is relati vely insensiti ve to dropping glucose concentrations in the medium [6]. These monocultures continue to take up glucose even when medium concentrations drop to hypoglycemic levels. In contrast, the three- way connection medium maintained stabl e, normal glucose concentrations over 48 h. Regulation of glucose and other metabolites in two-way connections between HepG2 hepatocytes and endothelial cells, as well as rat hepatocytes and visceral AT, has also been reported pr eviously [17, 18]. These data demonstrate that cross-talk among the different cell types is required to prevent the drop of circu- lating glucose to hypoglycemic levels, and that this is at least partly through utilization of the glucosealanine path way and enhanced FFA uptak e. In the physiological cont ext, hepatocytes release glucose through gluconeogenesis or glycogenolysis to replace declining circulating glucose levels. This in-vitro model also manifests a similar homeostatic-like control of glucose concentration by hepatocytes in connected cultur e.

4.2 Systemic metabolic regulation in vitro through the FFA-glucosebalance

In all culture s, with the sole exception of the twoway connection, glycerol movement mimicked FFA movement, albeit to varying degree s. The parallel movement of these two metabolites might be explained by lipolysi s. In the two-way connection, FFAs increased while glycerol remained stable; further experiments will be required to identify the mechanism responsible for this. One possibility is that lipolysis associates with cellular glycerol re- tention, perhaps through altered aquaporin activi- ty [19]. Alternati vely endothelial lipase may be ac- tivated to cle ave FFAs from HUVEC membrane phospholipid s, releasing FFAs but not glycerol [20]. HUVECs may be sensiti ve to extracellular FFA concentrations such that when a certain threshold concentration is reached, there is a metabolic shift to limit FFA releas e. In the two-way connection, a peak medium FFA concentration of 161 M was reached at 15 h, after which it dropped down to 137 M by 48 h. A similar trend of a slowdown in release had been obser ved in pr evious experiments with HUVEC monoculture [6] but not with the AT monocultur e, in which FFAs continued to rise with time. In HUVEC s, such a slowdown in FFA release

may be related to the simultaneous decrease in medium glucos e, such that FFAs could pr ovide an alternate fuel source for endothelial cells as glucose becomes less availabl e. In our two-way connection, we obser ved net FFA release in the presence of net glucose uptak e, indicating the system s preference for glucose over FFAs as an energy substrat e. The data also suggest that the increase in extracellular FFAs in the two- way culture does not interfere with the basal glu- cose uptake of HUVEC s. Although it is well estab- lished that FFAs inhibit insulindependent glucose uptake [21], our results suggest that FFAs do not in- hibit insulin-independent glucose uptake in HUVEC s. In contrast, both glucose and FFA con- centrations remained stable in the three- way con- nection over tim e, presumably due to hepatic glu- cose release or synthesis as well as enhanced FFA uptake by HepG2 cell s. Clinical studies have demonstrated that high circulating glucose con- centrations in vivo are associated with defecti ve hepatic FFA uptake by competing with FFAs as an energy substrate [5]. Although the current studies were not carried out at high glucose concentra- tion s, the obser vations from the twoand three- way connections are nonetheless consistent with the hypothesis that there is a competition between glucose and FFAs as energy substrates in the cul- tures studied. In the three- way connection, by in- cluding the hepatic cell lin e, it was possible to maintain stabl e, normal glucose concentrations over 48 h, suggesting that enhanced hepatic FFA uptake may be a crucial part of the mechanism by which these cells maintain stable glucose concentrations in the medium. In fact the two-way connection exemplifies that AT and HUVECs are not capable of maintaining normal circulating glucose concentration s, whereas the three- way connection shows that HepG2 hepatocytes can restore normal glucose levels. Additionall y, the three- way connection, in which hepatocytes were introduced to the system, had reduced circulating FFAs relati ve to the two- way connection and this was associated with lactate releas e, as well as normalization of circulating glucose concentration s. Glycerol did not change over time and this may be due to glycerol exchange among AT, HUVEC s, and HepG2s rather than static ov erall glycerol retention. FFAs may compete with glucose for uptake by HepG2, given that the hepatocytes presumably rem oved FFAs from the medium and provided glucose to compen- sate for that taken up by the other cell type s. There- fore, these studies suggest that FFAs may be a mas- ter-regulator of metabolism, signaling the HepG2 cells to release glucose when circulating levels dro p.

124

2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

Biotechnol. J. 2012, 7, 117126

www.biotechnology-journal.com

4.3 Metabolic FFA-lactate,FFA-albumin, lactateglycerol and glucose-L-alanine balance

The two-way connection represented a reversal in the association between FFAs and lactat e, since AT and HUVEC monoculture FFA release was associ- ated with lactate releas e, whereas the twoway con- nection exhibited FFA release and lactate uptak e. Morand et al. [22] demonstrated that FFAs can down-regulate lactate release in rat hepatoc yes. Data from the two-way connection suggest that in- creased FFAs may down-regulate lactate release in AT and/or HUVEC s. Of all the cultures studied her e, the HepG2 hepatocyte m onoculture had the lowest final FFA concentration and the highest lac- tate releas e, findings which support the hypothesis that FFAs may be important regulators of lactate releas e. In addition, when FFAs are completely re- moved from the circulation, as in the HepG2 monocultur e, lactate release is enhanced. Although the three- way connected culture showed lactate release in the presence of low FFA levels, total release was less than the sum of the three monoculture s, demonstrating nonetheless that lactate regulation cannot be completely explained by circulating FFAs. Gi ven that albumin is the major carrier for circulating fatty acid s, the relationship between al- bumin and FFA concentrations may provide insight into FFA metabolism. The three- way connection was characterized by stable FFA concentrations and net albumin release over time. FFAs released by AT and HUVECs may have stimulated both albumin release and FFA uptake in HepG2 cells; al- bumin release could prevent a rise in FFAs by bind- ing the FFAs released into the circulation through lipolysis [23]. However, since albumin was detected in the nanomolar range (maximum detected 11.2 nM in the three- way connection), whereas FFAs were detected in the micromolar range (max- imum detected 137 M in the two-way connection), it is unlikely that this relati vely small albumin re- lease could bind enough FFAs to explain the re- duced FFAs obser ved at 48 h in the three- way con- nection. A more likely explanation implicates in- creased hepatic FFA uptak e. In the two-way connection, the rise in the level of FFAs is associated with lactat e, glycerol and Lalanine retention. It has been shown that FFAs can inhibit lactate release [22], and that lactate levels can regulate glycerol release [24]. What we obser ved in the two-way connection is consistent with these pr evious findings and supports the role for FFAs as important regulators of multiple metabo- lism s. In the absence of insulin and at the normal physiological concentrations studied her e, there is, however, no evidence that FFAs interfere with glu-

cose uptake in the two-way connection. L-Alanine was also taken up by cells in the two-way connection, suggesting either its utilization or degradation, when glucose levels drop in the medium (Fig. 4). Glucose oxidation produces pyru vate, which can undergo transamination to alanin e, which is a glucogenic amino acid since it is con- verted to glucose by the liver. In the three- way con- nection, the glucose-alanine cycle may occur in the hepatic cells to replac e, at least in part, consumed glucos e. However, L-alanine levels increased over time, suggesting a shift towards amino acid produc- tion, as corroborated by the enhanced albumin re- lease in the system.

4.4 Conclusion
These experiments demonstrate that a three- way connected culture of AT, endothelial cells and a he- patocyte cell line in the MCmB reproduces several characteristics of in-vi vo glucose and lipid metab- olism. They also underline that traditional in-vitro studies of lipid or glucose metabolism in cells pres- ent an incomplete picture of the different regulato- ry mechanisms involved in cellular homeostasi s. This is the first example of an in-vitro model of glu- cose and lipid metabolism and paves the way for further and

detailed studie s. For exampl e, the abil- ity to add or subtract specific cell types from con- nected cell culture systems makes it possible to better understand the specific contributions that each cell type makes to systemic metabolism. In addition, the current experiments were carried out without insulin and, therefor e, explore metabolic cross-talk among different cell types under basal condition s, which is a prerequisite for stu dying simulations of pathop hysiological condition s. This stu dy explored only one two-way connection, that of AT and HUVEC s. Other studies have been carried out by our group examining the two- way connection between hepatocytes and endothelial cells [17, 25]. Jindal et al. [26] showed that the effect of endothelial cells on hepatocyte synthetic function is mediated through soluble signaling, in particular by the amino acid prolin e, released by endothelial cells and circulating in the medium. We have also shown that the metabolic cross-talk be- tween AT and HepG2 hepatocytes leads to meta- bolic self regulation in vitro [18]. The findings from the various combinations studied here and in oth- er laboratories will provide new insights into how glucose and lipid metabolism are regulated systemically and may provide clues to explain what is obser ved in vivo.

2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H

125

Biotechnolo gy Journal

Biotechnol. J. 2012, 7, 117126

The autho rs would like to thank Dr.ssa Santina Quarta from the Laboratory of Molecular Hepatology in Department of Clinical and Experimental Medicine at the Unive rsity of Padua for providing the HepG2 cells. B.V. performed the experimental wo rk, E.M. performed the statistical analysis and wrote the manuscript, E.I. analyzed the metabolic data and interpreted the results, F.M., S.F. and M.C. provided samples, and M.C.M. dosed metabolites. A. Avoga ro inte rpreted the data and designed experiments, A.Ahluwalia developed the conc ept and wrote the manuscript. The autho rs have declared no conflict of inte rest.

[12]

[13]

[14]

[15]

[16]

5 References
[1] Sung, J. H., Esch, M. B., Shule r, M. L., Integration of in silico and in vitro platforms for pharmacokinetic pharmaco dy- namic modelin g. Expert Opin. Drug Meta b. Toxicol . 2010, 6, 10631081. Sbrana, T., Ahlu walia, A., Engineering Quasi- VivoTM in-vitro Organ Model s. In: Balls, M., Combe s, R. Bhogal, N., (Eds) New Technologies for Toxicity Testing , Landes Bioscienc e, Austin 2011. Randl e, P. J., Garland, P. B., Hake s, C. N., Newsholme, E. A. et al. The glucose fatty-acid cycle. Its role in insulin sensitivi- ty and the metabolic disturbances of diabetes mellitu s. Lancet 1963, 1, 785789. Boden, G., Free fatty acids-the link between obesity and insulin resistanc e, Endoc r. Pract 2001, 7, 4451. Iozzo, P., Turpeinen, A. K., Takala, T., Oikonen, V. et al., De- fecti ve liver disposal of free fatty acids in patients with im- paired glucose toleranc e. J. Clin. Endocrinol. Meta b. 2004, 89, 34963502. Vinci, B., Murp hy, E., Iori, E., Marescotti, M. C. et al., Flowregulated glucose and lipid metabolism in adipose tissu e, endothelial cell and hepatocyte cultures in a modular bioreacto r. Biote chnol. J. 2010, 5, 618626. Mazzei, D., Guzzardi, M. A., Giusti, S., Ahlu walia, A., A low shear stress m odular bioreactor for connected cell culture under high flow rate s. Biote chnol. Bioeng . 2010, 106 , 127137. Mazzei, D., Vozzi, F., Cisternin o, A., Vozzi, G. et al., A highthroughput bioreactor system for simulating physiological environment s. IEEE Trans. Ind. Elect ron. 2008, 55, 3273 3280. Vinci, B., Duret, C., Kliebe r, S., Gerbal-Chaloin, S. et al., Mod- ular bioreactor for primary hepatocyte culture: Medium flow stim ulates expression and activity of det oxification gene s. Biote chnol. J. 2011, 6, 554564. Lindstedt, S. L., Schaeffe r, P. J., Use of allometry in predict- ing anatomical and physiological param eters of mammal s. Lab. Anim. 2002, 36, 113. Vozzi, G., Previti, A., De Rossi, D., Ahlu walia, A., Microsyringe-based deposition of two-dimensional and three-di[17]

[18]

[2]

[19]

[3]

[4] [5]

[20] [21]

[22]

[6]

[23]

[7]

[24]

[8]

[25]

[9]

[26]

[10]

mensional polymer scaffolds with a well-defined geometry for application to tissue engineerin g. Tissue Eng. 2002, 8, 10891098. Vinci, B., Cavallon e, D., Vozzi, G., Mazzei, D. et al., In-vitro liver model using microfabricated scaffolds in a modular bioreactor. Biote chnol. J. 2010, 5, 232241. Mariani, M., Rosatini, F., Vozzi, G., Previti, A. et al., Characterization of tissue-engineered scaffolds microfabricated with PAM. Tissue Eng. 2006, 12, 547557. Lu, L., Pete r, S. J., Lyman, M. D., Lai, H. L. et al., In vitro and in vivo degradation of porous poly( DL -lactic-coglycolic acid) foam s. Biomaterials 2000, 21, 18371845. Vinci, B., Bio reacto rs and tissue engineering for the realization of a dynamic in-vit ro model of glucose metabolism . PhD thesi s, 2008. Uni versity of Pisa. http://etd.adm .unipi.it/theses/ available/etd-02072008-141845/. Lloyd, B., Burrin, J., Smythe, P., Alberti, K. G., Enzymic fluorometric continuous-fl ow ass ays for blood glucos e, lactat e, pyru vate, alanin e, glycerol, and 3- hydr oxybutyrat e. Clin. Chem . 1978, 24, 17241729. Vozzi, F., Heinrich, J. M., Bade r, A., Ahlu walia, A. D., Connected culture of murine hepatocytes and HUVEC in a multicompartmental bioreacto r. Tissue Eng. A 2009, 15, 1291 1299. Guzzardi, M. A., Domenici, C., Ahlu walia, A., Metabolic control through hepatocyte and adipose tissue cross-talk in a modular multi-compartmental bioreacto r. Tissue Eng. A 2001, 17, 16351642. Walke r, C. G., Holnes s, M. J., Gibbon s, G. F., Sugden, M. C., Fasting-induced increases in aquaporin 7 and adipose triglyceride lipase mRNA expression in adipose tissue are attenuated by per oxisome proliferato r-acti vated receptor alpha deficienc y. Int. J. Obes . 2007, 31, 11651171. Cohen, J. C., Endothelial lipase: Direct evidence for a role in HDL metabolism. J. Clin. Invest. 2003, 111 , 318321. Boden, G., Shulman, G. I., Free fatty acids in obesity and type 2 diabetes: Defining their role in the development of insulin resistance and beta-cell dysfunction. Eur. J. Clin. Invest . 2002, 32 (Suppl 3), 1423. Morand, C., Remes y, C., Demign e, C., Fatty acids are potent modulators of lactate utilization in isolated hepatocytes from fed rat s. Am. J. Physiol . 1993, 264, E816E823. Ockne r, R. K., Weisige r, R. A., Gollan, J. L., Hepatic uptake of albumin-bound substances: albumin receptor concept . Am. J. Physiol. Gast rointest. Liver Physiol. 1983, 245, G13G18. Issekutz, B. Jr., Shaw, W. A., Issekutz, T. B., Effect of lactate on FFA and glycerol turn over in resting and exercising dogs. J. Appl. Physiol. 197, 39, 349353. Guzzardi, M. A., Vozzi, F., Ahlu walia, A., Study of the crosstalk between hepatocytes and endothelial cells using a novel multi-compartmental bioreacto r, a comparison between connected cultures and co-culture s. Tissue Eng. A 2009, 15, 36353644. Jindal, R., Nahmia s, Y., Tille s, A. W., Berthiaum e, F. et al., Amino acid-mediated heterotypic interaction governs performance of a hepatic tissue model. FASEB J. 2009, 23, 22882298.

[11]

126

2011 Wiley-VC Verlag GmbH & Co. KGaA, Weinheim H