Mesenchymal stem cells support hepatocyte function in engineered liver grafts

Abstract

Recent studies suggest that organ decellularization is a promising approach to facilitate the clinical application of regenerative therapy by providing a platform for organ engineering. This unique strategy uses native matrices to act as a reservoir for the functional cells which may show therapeutic potential when implanted into the body. Appropriate cell sources for artificial livers have been debated for some time. The desired cell type in artificial livers is primary hepatocytes, but in addition, other supportive cells may facilitate this stem cell technology. In this context, the use of mesenchymal stem cells (MSC) is an option meeting the criteria for therapeutic organ engineering. Ideally, supportive cells are required to (1) reduce the hepatic cell mass needed in an engineered liver by enhancing hepatocyte function, (2) modulate hepatic regeneration in a paracrine fashion or by direct contact, and (3) enhance the preservability of parenchymal cells during storage. Here, we describe enhanced hepatic function achieved using a strategy of sequential infusion of cells and illustrate the advantages of co-cultivating bone marrow-derived MSCs with primary hepatocytes in the engineered whole-liver scaffold. These co-recellularized liver scaffolds colonized by MSCs and hepatocytes were transplanted into live animals. After blood flow was established, we show that expression of adhesion molecules and proangiogenic factors was upregulated in the graft.

Introduction

Currently, the only curative treatment available for end-stage liver disease is organ transplantation. However, the shortage of cadaveric and living donor organs causes long waiting list times, resulting in the death from liver failure of many severely ill patients before a graft becomes available.^1,2 Alternative approaches to resolve this problem have been explored using several different techniques, for example, cell transplantation,³ tissue-engineered grafts,⁴ or cell-based extracorporeal devices.⁵ However, these treatments can only provide temporary partial restoration of organ function. The main problem is long-term retention of explanted primary hepatocytes, which need abundant and constant supplies of oxygen and nutrients to maintain viability, phenotype, and to perform unique liver functions (detoxification, synthesis of numerous factors, and regulation of metabolism).

Recently, a regenerative approach using decellularization and recellularization methods to create bio-artificial organs has been explored in small animal models, and applied to different organs such as liver,⁶ lung,⁷ heart⁸ and kidney.⁹ This method preserves the organ-specific three-dimensional extracellular matrix (ECM) which can be used as an acellular and transplantable native scaffold potentially providing an ideal microstructural cradle for cell attachment, differentiation and function.¹⁰ The preservation of the native ECM may also have a beneficial role in signaling and paracrine pathways for cell-to-cell and cell-to-matrix contacts by providing natural growth factors.^11-13 Possibly even more important for survival and long-term function of reseeded cells, original vascular networks may be re-vascularized for rapid oxygen and nutrient transport by connecting to vessels in situ.^6-9

For liver, this whole-organ decellularization and recellularization technology is envisioned as an alternative to a “bridging therapy” prior to liver transplantation, as well as a basic science model. However, to establish long-term culture systems using native scaffolds both in vitro and in vivo, the technology remains in need of further improvement to maximize hepatocyte-survival and function.^6,14-19 One solution is to use a non-parenchymal cell support, which is generally required to reconstitute blood vessel intraluminal surfaces, including the microvasculature.^6,16 This is particularly important for regaining vascular integrity and thus functionality of hepatic sinusoid structures and bile duct required for normal liver function. Recapitulation of the regenerative capacity of the liver also requires hepatic differentiation and reconstitution using a variety of stem cells in the native niche.^20,21 Although the optimal cell source for achieving long-term efficacy using liver-assisted devices has not yet been determined, one option is to use MSCs.

MSCs retain the potential to differentiate into hepatocytes²² and to modulate inflammation²³ through a combination of mechanisms involving soluble factors and cell-to-cell contact.²⁴ Therefore, MSCs are thought to provide a regenerative microenvironment at the site of damaged tissues and organs by virtue of their homing capacity.²⁴ Because of the poor understanding of molecular mechanisms regulating these desirable effects of MSCs and concerns about their safety in clinical applications, the translation of experimental rodent models into the clinical setting is still under investigation. A therapeutic approach using a combination of MSCs and hepatocytes as a composite liver-assisted device for treatment of rat acute liver failure has previously been reported to provide a long-term survival benefit.²⁵ Indeed, the MSC-composite device yielded improvement of hepatocellular metabolism, synthesis, and secretion, and decreased cell apoptosis, enhanced cell replication, as well as possessing immune-modulating effects via secretion of bioactive molecules.²⁵ More recently, very interesting work by Takebe et al. suggested the importance of MSCs for natural liver development.²⁶ On this basis, we hypothesized that MSCs would be strong candidates for supportive cells in transplantable whole-organ engineered grafts, which could result in long-term well-maintained hepatocyte function.

In the present study, we sought to optimize the decellularization and recellularization technique using bone marrow-derived MSCs as a source of supportive cells to achieve a well-structured hepatic scaffold with better-controlled cell distribution and function. The spatial architecture of matrix cues for the respective cell types and for the microvascular networks were examined by assessing the distribution and expression of specific markers of the reseeded cells and in the re-perfused graft in situ. In addition, the influence of co-culturing rat hepatocytes with MSCs as a non-parenchymal cell source was evaluated to determine efficacy for sustaining hepatocyte viability and function inside the recellularized scaffold in this tissue engineering approach.

Results

Appropriate sequence and numbers of hepatocytes and MSCs infused into engineered native liver scaffolds

We had previously investigated the beneficial role of MSCs as supportive cells for primary hepatocyte generation to create a functional liver-assist device.²⁵ To determine whether MSCs had similar supportive effects on hepatic function when they were applied in a novel tissue engineering approach using decellularized native liver scaffolds, it is of paramount importance to determine the appropriate sequence of addition of the two different cell types. This is because the recellularization process of the native liver scaffolds requires vascular access for cell infusion which is totally different from conventional in vitro co-culture methods.

We therefore compared the outcome of sequential cell infusion as follows: hepatocytes first (Fig. 1A, left: Group 1), MSCs first (Fig. 1A, middle: Group 2) and hepatocytes and MSCs simultaneously (Fig. 1A, right: Group 3). Separate infusions of the two cell types (Groups 1 and 2), resulted in higher numbers of hepatocytes and MSCs stacked inside the vascular walls (shown as arrows), which reduced hepatocyte migration into and repopulation of the parenchymal space. However, when both cell types were infused at the same time (Group 3), MSCs were well-engrafted among clusters of hepatocytes as well as inside the decellularized vascular walls in the matrix. However, no significant differences were seen among Groups 1–3 at day 4 in the number of TUNEL-positive cells (Fig. 1B and C: P = 0.648).

Figure 1. Different reseeding methods for hepatocytes and MSCs in the engineered liver scaffold: the sequence and number of cell infusions. (A) H&E staining of different sequences of reseeding. Left: hepatocytes first (Group 1), middle: MSCs first (Group 2) and right: hepatocytes and MSCs simultaneously (Group 3). Arrows show intravascular stagnant hepatocytes and MSCs. (B) TUNEL staining of the apoptotic cells in each different sequence of reseeding, left: hepatocytes first (Group 1), middle: MSCs first (Group 2) and right: hepatocytes and MSCs simultaneously (Group 3). (C) Quantification of the apoptotic cells by TUNEL staining in Group 1–3. (D) Gross appearance of the co-recellularized scaffolds in this perfusion/culture system using different numbers of hepatocytes, left: 3 × 10⁸ hepatocytes, middle: 1 × 10⁸ hepatocytes and right: 5 × 10⁷ hepatocytes. (E) Albumin synthesis by each different number of hepatocytes. Scale bars: 100 μm.

Previously, we had also determined feasible numbers and ratios of hepatocytes and MSCs for constructing a functional liver-assist device.²⁵ We introduced these two cell types at the same ratio (5:1 hepatocytes: MSCs) into the decellularized scaffold, but with different numbers of hepatocytes. Using the median lobe scaffold to control the distribution of reseeded cells, maximum numbers of cells could be employed (Fig. 1D, left: 3 × 10⁸ hepatocytes; Fig. 1D, middle: 1 × 10⁸ hepatocytes, each with 20% MSCs) while lower number of cells were also tested (Fig. 1D, right: 5 × 10⁷ hepatocytes with 20% MSCs). We evaluated the outcome by measuring albumin production. Scaffolds co-recellularized at higher cell numbers showed loss of integrity of the scaffold, leakage of tightly compressed cells from the limited parenchymal space, and poorer albumin production than those seeded with lower cell numbers at day 4 (Fig. 1E: P = 0.029).

Distribution of hepatocytes and MSCs in decellularized liver scaffolds and their functional differences

H&E staining showed that hepatocytes were aligned in a manner similar to the original hepatic cords from the portal vein through the central vein but integrated better when co-recellularized with MSCs at day 4 (Fig. 2A, right), compared with the hepatocyte single-cultured group (Fig. 2A, left). Fluorescent dye cell tracking revealed that when MSCs were co-infused with hepatocytes, they engrafted together in the parenchymal space from day 2 to day 4 (Fig. 2B), compared with the single-cultured group (data not shown). Nonetheless, some of the co-engrafted MSCs were seen to be present around the portal area and intraluminal surface of the vessels, in particular (Fig. S1A). Enumerated values were 50 ± 2.5 cells per unit area staining for albumin in the hepatocyte/MSC co-cultured group (Fig. 2C, middle), whereas this figure was 42.5 ± 2 in the single-cultured group (Fig. 2C, left) at day 2 (Fig. 2D: P = 0.144 between the single- and co-cultured groups). CD90/Thy1 was clearly positive (22 ± 4 positive cells) in the samples from the co-cultured group (Fig. 2E, middle), while it was not detected in the single-cultured group (Fig. 2E, left) (Fig. 2F: P = 0.032 between the single- and co-cultured groups). Albumin and CD90/Thy1 expressions were compared with the rat normal liver as control, which showed 340 ± 20 and negative cells, respectively (Fig. 2C, right; Fig. 2E, right). TUNEL-positive cells increased from day 2 to day 4 by 4 ± 4% to 60 ± 8% in the single-cultured group (Fig. 2G), and by 4 ± 3% to 50 ± 14% in the co-cultured group (Fig. 2H). This slightly lower value in the co-cultured group did not achieve significance at day 4 (Fig. 2I: P = 0.435). Hepatocellular functions shown by albumin production and urea synthesis were higher in the co-culture group but the difference between the two groups was again not statistically significant (Fig. 2J: albumin production: P = 0.609; Fig. 2K: urea synthesis: P = 0.505).

Figure 2. Distribution of hepatocytes and MSCs in the decellularized liver scaffold and their functional differences. (A) H&E staining (left: single-cultured group, right: co-cultured group). (B) Cell tracking of the co-cultured group, with hepatocytes in red and MSCs in green. (C) Immunohistochemical staining of albumin (left: single-cultured group, middle: co-cultured group, right: normal liver) and (D) Number of albumin-positive cells per unit area. (E) Immunohistochemical staining of CD90/Thy1 (left: single-cultured group, middle: co-cultured group, right: normal liver) and (F) Number of CD90/Thy1-positive cells per unit area. (G) TUNEL staining of the single-cultured group, (H) TUNEL staining of the co-cultured group and (I) Quantification of the apoptotic cells by TUNEL staining. (J) Albumin synthesis. (K) Urea secretion. Scale bars: 100 μm.

Expression of adhesion molecules and angiogenic factors is enhanced by MSCs in the engineered liver scaffold

The expression of Integrin β1²⁴ an important adhesion molecule, and VEGF, an important proangiogenic factor in the sinusoid^27-29 was analyzed immunohistochemically, based on the previous finding from histological evaluation of co-recellularized tissue which demonstrated high integration of hepatocytes and perivascular repopulation of MSCs in the scaffold. The number of Integrin β1-positive cells in the single-cultured group was significantly lower (Fig. 3A: 7.5 ± 1.5 per unit area) than in the co-cultured group (Fig. 3B: 22 ± 4 per unit area) (Fig. 3D: P = 0.028 between the single- and co-cultured groups). Integrin β1 expression was compared with the rat normal liver as control, which showed 73 ± 3 positive cells (Fig. 3C). Interestingly, in the co-cultured group, Integrin β1 expression was strong in the peripheral region of the confluent hepatocytes, relative to the inner region. The negative stain of CD90/Thy1 on these Integrin β1 positive cells, which were found in the peripheral region of the confluent formation, indicated that instead of MSCs, some of the co-cultured hepatocytes expressed Integrin β1 (Fig. S1B). In addition, no VEGF-positive cells were detected in the single-cultured group (Fig. 3E) but 13 ± 3 VEGF-positive cells per unit area were seen in the co-cultured group (Fig. 3F) (Fig. 3H: P = 0.049 between the single- and co-cultured groups). VEGF expression was compared with the rat normal liver as control, which showed 36 ± 4 positive cells (Fig. 3G). More interestingly, VEGF was co-expressed with CD90/Thy1-positive cells in the co-cultured graft. CD31, identifying endothelial cells,³⁰ was also evaluated in the co-cultured group at days 2 and 6. Little CD31 was expressed in either the single- or co-cultured groups at day 2 (0 ± 0 and 2 ± 1 per unit area, respectively) (Fig. S1C: single-cultured group; Fig. S1D: co-cultured group). However at day 6, increased numbers of CD31-positive cells were seen in the co-cultured group (Fig. 3J: 36 ± 6 per unit area), significantly more than in the single-cultured group (Fig. 3I: 2.8 ± 1.5 per unit area) (Fig. 3L: P = 0.034 between the single- and co-cultured groups). CD31 expression was compared with the rat normal liver as control, which showed 30 ± 5 positive cells (Fig. 3K). As with the pattern of VEGF expression, CD31 was mostly co-expressed with CD90/Thy1 in the co-cultured graft.

Figure 3. Expression of adhesion and vascular molecules in the co-recellularized engineered liver. (A) Immunohistochemical staining of Integrin β1 at day 2 in the single-cultured group and (B) in the co-cultured group: Integrin β1, CD90/Thy1 and Merged with DAPI from left to right. (C) Normal liver control of Integrin β1 merged with DAPI. (D) Number of Integrin β1-positive cells per unit area. (E) Immunohistochemical staining of VEGF at day 2 in the single-cultured group and (F) in the co-cultured group: VEGF, CD90/Thy1 and Merged with DAPI from left to right. (G) Normal liver control of VEGF merged with DAPI. (H) Number of VEGF-positive cells per unit area. (I) Immunohistochemical staining of CD31 at day 6 in the single-cultured group and (J) in the co-cultured group: CD31, CD90/Thy1 and Merged with DAPI from left to right. (K) Normal liver control of CD31 merged with DAPI. (L) Number of CD31-positive cells per unit area. Scale bars: 100 μm.

Effects of short-term portal flow after implantation of co-recellularized engineered liver scaffolds

The gross appearance of both types of implanted grafts indicated satisfactory perfusion (Fig. 4A: single-cultured graft; Fig. 4B: co-cultured graft) with no evidence of leakage from the surface or vasculature of the grafts. However, an inappropriate efflux from the drainage tube was seen in both, revealed by H&E staining as massive coagulation inside the vasculature and parenchymal space of the matrix (Fig. 4C, left: single-cultured graft; Fig. 4D, left: co-cultured graft). The amounts of albumin-positive hepatocytes and CD90/Thy1-positive cells per unit area after 60 min of blood perfusion were evaluated. There were 90 ± 10 albumin-positive cells per unit area (Fig. 4C, middle) and, as expected, no CD90/Thy-positive cells in the single-cultured graft without MSCs (Fig. 4C, right). In contrast, there were 90 ± 5 albumin-positive cells (Fig. 4D, middle) and 42.5 ± 2.5 CD90/Thy1-positive cells (Fig. 4D, right) in the co-cultured graft (Fig. 4E: albumin: P = 1.00 and Fig. 4F: CD90/Thy1: P = 0.029). Additionally, in the hepatocyte single-cultured graft, 14 ± 4 cells per unit area were stained for Integrin β1 (Fig. 4G) but none for VEGF (Fig. S1E), whereas in the co-cultured graft, these values were 52 ± 4 (Fig. 4H), and 30 ± 2 (Fig. S1F), respectively (Fig. 4I: Integrin β1: P = 0.022; Fig. S1G: VEGF: P = 0.042). Consistent with the data from the ex vivo experiments, the MSC co-cultured graft showed stable expression of cell specific markers as well as Integrin β1 and VEGF, suggesting the importance of MSCs as supportive cells enhancing angiogenic and cell adhesion capacity. Finally, we investigated the expression of laminin in the co-cultured graft. This is a ligand of Integrin β1 and a key component of the perivascular niche regulating the adherence of stem cell populations.^21,31 Laminin was found to be co-expressed with Integrin β1 (Fig. 4J) mostly separated on CD90/Thy1-positive cells in the co-cultured graft (Fig. 4K). Laminin was also expressed in the rat normal liver as control (Fig. S1H).

Figure 4. Implantation of the co-recellularized engineered liver. (A) Gross appearance of the implanted grafts: left to right: pre-implant scaffold in perfusion/culture system, 30 min and 60 min after portal blood reperfusion of the single cultured and (B) the co-cultured graft. (C, left: single-cultured; D, left: co-cultured) Explanted single- and co-cultured graft histology 60 min after implantation: H&E staining, (C, middle: single-cultured; D, middle: co-cultured) Immunohistochemical staining of albumin and (C, right: single-cultured; D, right: co-cultured) CD90/Thy1. (E) Number of albumin- and (F) CD90/Thy1-positive cells per unit area. (G) Immunohistochemical staining of Integrin β1 in the single-cultured group and (H) in the co-cultured group: DAPI, Integrin β1, CD90/Thy1 and Merge from left to right. (I) Number of Integrin β1-positive cells per unit area. (J) Immunohistochemical staining of Integrin β1 and laminin of the hepatocyte/MSC co-cultured graft: DAPI, Integrin β1, laminin and Merge from left to right. (K) Immunohistochemical staining of laminin and CD90/Thy1 of the co-cultured graft: DAPI, laminin, CD90/Thy1 and Merge from left to right. Scale bars: 100 μm .

Discussion

Since the first report of liver decellularization,⁶ many studies have been devoted to re-establishing the architecture of the hepatic microenvironment inside a framework naturally originating from the liver in the form of an acellular transparent scaffold.^14-19 At the same time, many studies were investigating methods for better and longer maintenance of hepatocytes by modifying decellularization¹⁰ and recellularization approaches¹⁴ or by using non-parenchymal cells such as endothelial cells.^6,16 Among these cell sources, stem-cells are the most important entities for hepatic regeneration and for constructing autologous grafts for the permanent replacement of failed livers in vivo.^32,33 However, the behavior of the stem cell in the decellularized liver matrix has not been well-evaluated because of difficulties in keeping the matrix sufficiently viable to be able to study the fate of reseeded cells either ex vivo or in vivo.^34,35 We believe that re-establishment of the hepatic sinusoid structure, which consists of specific cellular components with unique characteristics, might require the support of stem cells.^28,36,37 The combination of stem cell technology and organ-specific three-dimensional ECM scaffolds will lead to the further development of hepatic tissue engineering and facilitate scientific and clinical use in future.^34,35

Bone marrow-derived MSCs have been reported to have a beneficial effect on other cell types in co-cultures,^38,39 including hepatocytes.⁴⁰ However, none documented their efficacy in more clinically-relevant models such as engineered liver scaffolds. Here, we demonstrated that bone marrow-derived MSCs are beneficial as supportive cells for the functional efficacy of this engineered liver graft. We have documented their positive effects on hepatic regeneration. In our MSC co-cultured grafts, VEGF and CD31 were to a great extent co-expressed on MSCs, the latter showing a marked increase at perfusion/culture day 6. We speculate that MSCs might be trans-differentiated into the endothelial lineage via interactions between co-cultured hepatocytes and the microenvironment of the hepatic post-decellularized ECM, with its altered secretory profile including many cytokines, chemokines and growth factors such as VEGF.^24,41 Previous studies suggested that VEGF played an important role in sinusoid remodeling with increased expression of CD31.^28,42 VEGF is one of the vital factors for liver morphogenesis and Matsumoto et al. reported that morphogenesis of the liver bud in flk-1 (VEGFR-2) mutant embryos arrests at the stage prior to liver vascular formation. They found that endothelial cells, hepatic cells and nascent vascular structures, prior to the formation of functioning vessels, are needed to promote hepatic morphogenesis.³⁰ Our results support the notion that the angiogenic potential of MSCs is beneficial for hepatic regeneration in the native liver scaffold, providing a reliable means to engineer these grafts for use in the clinical setting.

Interestingly, an additional effect was found in terms of significantly augmented expression of Integrin β1 on hepatocytes after co-engraftment with MSCs. Integrins are vital molecules for liver development⁴³ and heterodimeric ECM receptors play critical roles in modulating the composition of the ECM in the cell environment, which regulates cell fate.⁴⁴ The regulation of integrin binding as well as reorganization of adhesion ligands is also important for controlling stem cell fate. Indeed, integrins regulate differentiation of stem cells including MSCs into specific cell types by influencing the rigidity of the surrounding ECMs.^45,46 Furthermore, laminin, one of the ligands of Integrin β1, has been shown to promote the adhesion of progenitor cells in the periportal area of the liver. Therefore, we speculate that the upregulation of laminin in the engineered liver may promote liver regeneration^21,31,47based on the intensified signals of laminin and Integrin β1 only in the MSC co-cultured graft. In addition, we found that a tightly confluent formation was formed by hepatocytes under the influence of the co-cultured MSCs, related to Integrin β1 expression, in the peripheral area ex vivo. This phenomenon might be considered as due to the cell-to-cell contact between hepatocytes and MSCs rather than cell-to-matrix interactions which occur in hepatocyte-only attachment to the matrix.^25,34,35 This could itself improve the engraftment of the hepatocytes into the parenchymal space of the scaffold. Tight confluence formation would be expected to protect the central area containing hepatocytes from the harsh portal pressure by keeping the mass from breaking up, while the single-cultured hepatocytes diffused throughout the peripheral areas with poor cellular confluence formation. This co-engraftment effect seemed to contribute to the extended period of hepatocyte viability in the central part at day 6 of co-culture (Fig. S1I).

Liver regeneration is regulated by myriads of signals and humoral factors in the systemic environment. The finding of higher expression of albumin, CD90/Thy1, VEGF, CD31, and Integrin β1 and its ligand laminin, in the hepatocyte/MSC co-cultured group, suggests that MSCs played a role in the interaction of parenchymal cells in the three-dimensional hepatic ECM when exposed to portal blood flow. Such increased expression suggested that the portal flow might be an essential factor closely related to hepatic maturation and regeneration, as reported in previous studies.^48-50 Regulation of portal exposure of the co-recellularized scaffold to achieve long-term function implies that anti-coagulation therapy is vital to maintain constant hemodynamic flow through the matrix.^34,35 Endothelialization of the vascular canals and coating of the bare surface of empty matrix space might be a solution to the vascular coagulation problem noted here and for further reducing the shear-stress on the engrafted hepatocytes.^6,16-19 In any event, the formation of a functional vasculature is clearly vital for hepatic maturation.^26,30

In conclusion, we believe that we have developed an appropriate scaffold for effective introduction of two different cell sources, hepatocytes and MSCs. This rat engineered liver graft model represents an improved design among the many decellularized liver models reported so far with the optimum number and sequence of cell infusions. The assessed potential of hepatocyte/MSC co-cultures in this model included improved cellular engraftment with greater expression of Integrin β1 together with its ligand laminin on hepatocytes associated with MSCs, and VEGF and CD31 by MSCs in the perivascular area. These phenomena might be catalyzed by cellular and structural factors in the co-perfusion/cultured matrix containing mixtures of hepatocytes and MSCs^21,25with augmentation of hemodynamic stimulation. Our results suggest that administration of MSCs into the decellularized liver scaffold enriched the microenvironment for factors related to hepatic regeneration and served as supportive cells for hepatocyte maintenance and protein production. Such appropriate modifications of the engineered liver model simply by determining the number and ratio of MSCs relative to an optimized number of hepatocytes will facilitate new insights into liver regeneration and future clinical application as a bioartificial liver in the form of a transplantable graft.

Materials and Methods

Animals

Female Lewis rats (200–250 g; Sankyo Research Laboratories) were used for liver harvesting to prepare decellularized matrices and to isolate hepatocytes to recellularize them. Male Lewis rats (450–500 g; Sankyo Research Laboratories) were used for transplantation of the recellularized matrices. All protocols were reviewed and approved by the local Ethics Committee of Keio University.

Whole liver harvest and decellularization

The rats were maintained under anesthesia using inhalation of 1.5–3.0% of isoflurane (Mylan). After the abdominal incision, we injected heparin (450 units) into the intracardiac space. The portal vein was cannulated with a 20G cannula, and 10–15 ml of PBS containing heparin (50 units) was injected. The IHVC was ligated and the whole liver was resected. The SHVC was resected without ligation. The resected whole liver was frozen at -80 °C for at least 24 h and was decellularized by Trypsin and Triton X-100 with EGTA, via SHVC, at the perfusion speed of 3 ml/hr, for 24 h until gross removal of all the cellular components had been accomplished based on the method reported by Soto-Gutierrez et al.¹⁴ We removed small caudate lobes, left lobe and the right small lobes from the native liver scaffold.

The perfusion/culture system

The system consisted of a peristaltic pump, silicone tubes, bubble traps, connection tubes (As One cooperation) and three-way stopcocks (TERUMO) and a chamber with O₂ transparent membrane (Tigers Polymer Corporation). The decellularized median lobe matrix was put in the chamber and hepatocyte culture medium (C+H medium: Dulbecco’s modified Eagle’s medium (GIBCO) supplemented with 10% fetal bovine serum (JR Scientific, Inc., Woodland, CA), 14 ng/mL glucagon, 0.5 U/mL insulin, 7.5 mg/mL hydrocortisone (SIGMA), 200 mg/mL streptomycin, and 200 U/mL penicillin (GIBCO) and 20 ng/mL epidermal growth factor (FUNAKOSHI) was perfused via the cannula which was inserted into the portal vein of the scaffold. The system was placed in an incubator for temperature control at 37 °C and 5% CO₂. The medium was changed daily during perfusion/culture.

Rat hepatocyte isolation and MSC culture

The induction and maintenance of anesthesia, incision of the abdominal cavity and heparin injection were as described above for whole liver harvest. We then used a two-step collagenase digestion method to isolate the hepatocytes. The portal vein was cannulated with a 20G cannula, and perfused with KRB isolation buffer (154 mmol/L NaCl, 5.6 mmol/l KCl, 5 mmol/L glucose, 25 mmol/L NaHCO3 (WAKO), and 20 mmol/L HEPES; pH 7.4 (Research Organics INC.) supplemented with 2 mmol/L EDTA (Invitrogen) through the vasculature at 20 ml/min, 400–500 ml in total, followed by collagenase (SIGMA) in KRB buffer with CaCl₂ (WAKO), 150 ml in total to lyse the extracellular components. The KRBs were diluted 5-fold with distilled water. Viability was assessed by trypan blue exclusion and was routinely > 95%. The yield of the isolated hepatocytes was 2–3 × 10⁷ cells per liver. Rat bone marrow derived MSCs (TAKARA HOLDINGS INC.) at passages 5–9 were cultured in this medium and used for experiments. Cell culture and expansion of human MSCs was performed using Dulbecco’s modified Eagle’s medium (GIBCO) supplemented with 10% fetal bovine serum (JR Scientific, Inc.), 100 U/ml penicillin, 100 μg/ml streptomycin, 0.1 mmol/l nonessential amino acids and 1 ng/ml of basic fibroblast growth factor (GIBCO).

Co-recellularization of hepatocytes and MSCs

Titrated numbers of 3 × 10⁸, 1 × 10⁸, and 5 × 10⁷ hepatocytes together with 20% of MSCs²⁵ were infused using a three-step method¹⁴ in sequential order of Group 1: hepatocytes first, Group 2: MSCs first, and Group 3: hepatocytes and MSCs simultaneously.

Immunofluorescence staining

We stained 4 μm paraffin-embedded sections of the recellularized matrix with primary antibodies as follows: albumin (ab53435, Abcam), CD90/Thy1 (ab92574, ab225, Abcam), Integrin β1 (ab52971, ab95623, Abcam), laminin (ab11575, Abcam), VEGF (sc-7269, Santa Cruz Biotechnology) and CD31 (ab24590, Abcam). For secondary antibodies, for albumin we used Alexa Fluor 488 goat anti-mouse; for CD90/Thy1 Alexa Fluor 488 goat anti-mouse for the single staining, Alexa Fluor 594 goat anti-mouse for the double staining with Integrin β1 and laminin, and Alexa Fluor 568 goat anti-rabbit for the double staining with VEGF and CD31; for Integrin β1 Alexa Fluor 488 goat anti-rabbit for the double staining with CD90/Thy1 and Alexa Fluor 488 goat anti-rat for the double staining with laminin; for laminin Alexa Fluor 488 goat anti-rabbit for the double staining with CD90/Thy1 and Alexa Fluor 568 goat anti-rabbit for the double staining with Integrin β1; for VEGF and CD31 Alexa Fluor 488 goat anti-mouse (Invitrogen). All specimens were finally stained for nuclear DNA with 4´-6-diamidino-2-phenylindole (DAPI Fluoromount-G, Southern Biotech).

Cell tracking

MSCs and hepatocytes were labeled using the Cell Linker Kit to identify their locations in the recellularized matrix. Hepatocytes were labeled red by PKH26 Red Fluorescent Cell Linker Kit, and MSCs were labeled green by PKH67 Green Fluorescent Cell Linker Kit (SIGMA).

TUNEL assay

For apoptotic cell detection, TUNEL staining was used (DeadEnd Fluorometric TUNEL System, Promega), following the manufacturer’s instructions. After all procedures, we counterstained the specimen with DAPI (DAPI Fluoromount-G, Southern Biotech). Apoptotic nuclei were identified when stained green in contrast to cells when without DNA fragmentation showing only blue nuclear staining.

Detection of albumin and UREA secretion from co-recellularized hepatocytes

The spent medium samples stored at -80 °C were thawed at 4 °C for 12 h and then centrifuged at 3000 rpm for 5 min. Supernatants were collected, and the concentration of urea and albumin in each dialyzed extract was determined by the rat albumin (Bethyl Laboratories, Inc.) and Urease ELISA Kit (BioAssay Systems) following the manufacturer’s protocol. Results are expressed as mean – standard error. P value of < 0.05 was considered significant.

Implantation of the engineered liver

We prepared the recellularized matrix with (1) hepatocytes single-cultured and (2) hepatocytes and MSCs co-cultured by the same method as ex vivo culture. The catheters of portal vein and SHVC, 20G and 14G each, were replaced by heparin-coated 3.5 Fr and 5 Fr tubes (Carmeda) which were cut into lengths of 2 cm. The inside of the recellularized matrix was gently flushed with heparin (total 2500 U) from the portal vein and SHVC tubes before transplantation. We injected 500 U heparin into the intracardiac space before connecting the tubes to the recipient’s left portal vein and left renal vein. The grafts were kept in the abdomen for 60 min before harvesting. The median lobe decellularized graft is seen in Figure S1J.

Statistical analyses

The mean values for cell apoptosis, albumin and urea production, and positive marker staining per area were compared by the Student’s t test. The mean values for more than 3 samples were compared by ANOVA. Fisher’s protected least significant difference test was used. P value of < 0.05 was considered significant.

Glossary

Abbreviations:

min

minute

TUNEL

terminal deoxynucleotidyl transferase–mediated nick-end labeling

VEGF

vascular endothelial growth factor

H&E staining

haematoxylin and eosin

PBS

phosphate-buffered saline

IHVC

inferior hepatic vena cava

SHVC

superior hepatic vena cava

G

gauge

EGTA

ethylene glycol tetraacetic acid

KRB

Krebs-Ringer bicarbonate solution

HEPES

4-(2-HydroxyEthyl)-1-PiperazineEthaneSulfonic acid

EDTA

ethylenediaminetetraacetic acid

DAPI

4',6-diamidino-2-phenylindole

DNA

deoxyribonucleic acid

rpm

revolution per minute

Eliza

Enzyme-linked immuno-sorbent assay

Fr

French

U

unit

ANOVA

analysis of variance

HEP

hepatocyte

10.4161/org.27879