Potential conflict of interest: Nothing to report.
Alternative methods to whole liver transplantation require a suitable cell that can be expanded to obtain sufficient numbers required for successful transplantation while maintaining the ability to differentiate into hepatocytes. Mesenchymal stem cells (MSCs) possess several advantageous characteristics for cell-based therapy and have been shown to be able to differentiate into hepatocytes. Thus, we investigated whether the intrahepatic delivery of human MSCs is a safe and effective method for generating human hepatocytes and whether the route of administration influences the levels of donor-derived hepatocytes and their pattern of distribution throughout the parenchyma of the recipient's liver. Human clonally derived MSCs were transplanted by an intraperitoneal (n = 6) or intrahepatic (n = 6) route into preimmune fetal sheep. The animals were analyzed 56–70 days after transplantation by immunohistochemistry, enzyme-linked immunosorbent assay, and flow cytometry. The intrahepatic injection of human MSCs was safe and resulted in more efficient generation of hepatocytes (12.5% ± 3.5% versus 2.6% ± 0.4%). The animals that received an intrahepatic injection exhibited a widespread distribution of hepatocytes throughout the liver parenchyma, whereas an intraperitoneal injection resulted in a preferential periportal distribution of human hepatocytes that produced higher amounts of albumin. Furthermore, hepatocytes were generated from MSCs without the need to first migrate/lodge to the bone marrow and give rise to hematopoietic cells. Conclusion: Our studies provide evidence that MSCs are a valuable source of cells for liver repair and regeneration and that, by the alteration of the site of injection, the generation of hepatocytes occurs in different hepatic zones, suggesting that a combined transplantation approach may be necessary to successfully repopulate the liver with these cells. (HEPATOLOGY 2007.)
The maintenance of cellular homeostasis within the normal liver and during liver regeneration is provided by mature hepatocytes and cholangiocytes and by the intrahepatic (IH) stem cell compartment located in the canals of Hering and the intralobular bile ducts.1 Recently, another potential source of cells able to provide liver cell replacement has been identified. These stem/progenitor cells consist of hematopoietic stem cells (HSCs) and mesenchymal stem cells (MSCs) that reside within the bone marrow (BM) but can reach the liver through the circulatory system.2–12 Because it is clearly evident that the shortage of available human donor organs cannot meet the needs of all the patients awaiting liver transplantation, alternatives to whole-organ replacement are urgently needed. Cell-based treatments, with cells of either hepatic or extrahepatic origin that would be able to repopulate and re-establish a functional liver after administration, could serve as a possible alternative to whole-organ transplantation.
In order to attain this goal, several questions, such as the sources of the cells to be used, the dosage, and the optimal administration route, must first be answered with clinically relevant animal models. MSCs represent an optimal source for cell-based therapy, given the possibility of using an autologous source of cells, the ease of their isolation, their extensive expansion rate and broad differentiative potential, and their apparent hypoimmunogenicity and immunosuppressive effects.13–18 We and others have demonstrated that human MSCs are able to differentiate into hepatocytes in vitro and in vivo.12, 19–23 However, the levels of donor liver cells obtained after transplantation seems to depend on both the model used and the type of underlying liver disease and/or injury caused.24–28 These results suggest that the stimuli provided by the microenvironment and the signaling factors recruiting MSCs to the damaged tissues are important for MSC differentiation and the robustness of liver regeneration by the transplanted cells. In the fetal sheep model, the transplantation of human HSCs during the preimmune period of gestation generated, by a mechanism independent of cell fusion, a significant number of functional human hepatocytes in the livers of the transplanted animals.7 In this large-animal noninjury model, when stem cell transplantation is performed, the liver is at a stage of development in which hepatocytic maturation is occurring, but the liver still contains hematopoietic niches conducive to the lodging of BM-derived cells that arrive through the portal circulation after intraperitoneal (IP) injection. It is also possible that stem cells initially migrate to and lodge within the BM after transplantation and that the differentiation into liver cells occurs later as the result of the engrafted BM stem/progenitor cells re-entering circulation and subsequently lodging within the liver. However, because the presence of MSCs in the peripheral blood correlates inversely with the gestational age,29 it is likely that, after transplantation in the fetal sheep model, few circulating MSCs are available to lodge in the liver and give rise to hepatocytes. It is also conceivable that the hepatocytes obtained after MSC transplantation could be, at least in part, derived from HSCs generated from the transplanted MSCs because in sheep fetuses, human MSCs are able to differentiate into hematopoietic cells.30
In the present studies, we used the fetal sheep model to compare the ability and efficiency of clonally derived human adult BM MSC populations to give rise to hepatocytes following IH or IP transplantation. We demonstrate that the direct delivery of clonally derived MSCs into the liver resulted in both a higher level of hepatocytes and a more widespread distribution of these hepatocytes throughout the parenchyma than IP injection. These results provide evidence that the IH delivery of human MSCs is a safe and efficacious method for generating human hepatocytes and that with this route of administration, it is possible to modulate the pattern of distribution of donor-derived hepatocytes throughout the parenchyma of the recipient's liver.
Isolation, Characterization, and Culture of Clonally Derived Human MSCs
Heparinized human BM was obtained from healthy donors after informed consent according to guidelines from the Office of Human Research Protection at the University of Nevada at Reno. Low-density BM mononuclear cells were separated with a Ficoll density gradient (1.077 g/mL; Sigma, St. Louis, MO) and washed twice in Iscove's modified Dulbecco's medium (Invitrogen Corp., Carlsbad, CA). BM mononuclear cells were first enriched for the Stro-1+ fraction with a Stro-1 antibody (R&D Systems, Minneapolis, MN) and magnetic bead cell sorting (Miltenyi Biotec, Inc., Auburn, CA). This was followed by the single cell deposition of Stro-1+CD45−Gly-A− cells (where Gly-A is glycophorin A) into fibronectin-coated 96-well plates with a FACSVantage (BD Biosciences, San Jose, CA). Twenty-four hours after the single cell deposition, the wells were visually inspected with phase contrast microscopy (IX70, Olympus, Melville, NY) for the presence of a single cell per well. Only wells containing single cells were used to establish the MSC clones employed in these studies. Single cell populations were expanded in vitro at 37°C in 5% CO2 humidified air in the presence of a prescreened MSC growth medium (Cambrex, Walkersville, MD). At a cell confluence of 60%–75%, MSC clones were detached with 0.25% trypsin (Invitrogen) for 3–5 minutes at 37°C. Trypsin was neutralized with a medium containing fetal bovine serum, and the cells were replated at a 1:3 ratio in gelatin-coated tissue culture flasks. The MSCs were cultured for approximately 7 passages, which corresponded to roughly 3 months, because these were clonal populations derived from a single cell, commencing in a 96-well plate and ending up in 75-cm2 culture flasks to obtain sufficient cells for transplantation. A phenotypic analysis of the cultured MSC clones was performed prior to transplantation by flow cytometry with a FACScan (BD Biosciences) and monoclonal antibodies against human CD34, CD45, CD51, CD31, CD13, CD90 (thymus cell antigen 1), CD102, CD117 (c-kit), CD38, CD106, CD29, human leukocyte antigen ABC, human leukocyte antigen DR, CD105, and CD10 (BD Biosciences) and by immunostaining with antibodies against chemokine (C-X-C motif) receptor 4 (CXCR-4), c-mesenchymal epithelial transition factor (c-met) receptor, stromal cell-derived factor 1 (SDF-1; R&D Systems), and vimentin (Sigma).
Characterization of the Clonally Derived MSCs
The clonally derived MSCs expressed high levels of CD90, CD13, CD29, and SDF-1 and were negative for CD14, CD33, CD34, CD45, Gly-A, human leukocyte antigen DR, CXCR-4, and C-met receptor. These results indicate that the clonal MSC populations that we obtained and used in these studies possessed a phenotype consistent with what has been described for MSCs by other authors.31–34 Importantly, these cells did not express genetic markers of embryonic stem cells, such as octamer 4 and reduced expression 1, distinguishing them from the multipotent adult progenitor cell described by Verfaillie's group.34 Although all the clones expressed vimentin, only clones expressing high levels of this intermediate filament were used for transplantation studies.35 Prior to transplantation, clonal populations were also evaluated for their ability to fulfill the criteria of MSCs by undergoing osteogenesis and adipogenesis in vitro, as described in detail later. All clonal populations continued to express Stro-1 and were able to undergo differentiation along both of these mesenchymal lineages; this confirmed their status as MSCs.
After the expansion of the clonally derived Stro-1+ cells in a standard medium and passaging to obtain sufficient cell numbers, the medium was exchanged for a commercial medium designed to induce adipogenic differentiation (Cambrex). Adipogenic differentiation took place over the course of 2 weeks. To confirm successful adipogenic differentiation, cytoplasmic triglyceride lipid droplets were stained with the Oil Red O staining method.36
Clonally derived Stro-1+ cells were grown in a standard MSC medium until they reached approximately 60%–70% confluence. Subsequently, the medium was changed to a commercially available osteogenic induction medium (Cambrex). The cells were then maintained under these conditions for 2 weeks, the medium being changed twice weekly. At the end of this period, the cells were washed in phosphate-buffered saline (PBS) and fixed in acetone/methanol for 10 minutes at 4°C, and von Kossa staining was performed to confirm mineralization.
Transplantation of the Human MSCs into the Fetal Sheep Recipients
Each of the 3 MSC clones used in these studies was grown separately until enough cells were obtained for these studies. Cells from each clone were then split into aliquots of 5 × 105 cells in 0.3 mL of a QBSF serum-free medium (Quality Biological, Gaithersburg, MD) and transplanted into 55–60–day-old preimmune fetal sheep recipients (term of 145 days), either by IP injection as previously described7 or by IH ultrasound-guided injection with a transabdominal ultrasound apparatus (ALOKA SSD-1000) with a 5-MHz probe. Clone 1A was transplanted into 1 animal IP and 1 animal IH. Clone 1B was transplanted into 1 animal IP and 1 animal IH, and Clone 2 was transplanted into 4 animals IP and 4 animals IH. Thus, a total of 12 animals were transplanted: 6 IP and 6 IH. The number of fetuses, their viability, and their position were first determined, and this was followed by the identification of the right and left liver lobes. A 22-gauge spinal needle was then inserted through the skin and the uterine wall into the amniotic cavity and then into the right lobe of the liver of the fetuses under continuous ultrasound guidance with the freehand technique. After confirmation of the appropriate positioning of the needle, the cells were injected slowly during the withdrawal of the needle. The transplanted animals were then analyzed for the presence of human donor cell engraftment 56–70 days after transplantation, with nontransplanted, age-matched animals serving as controls. All animals received humane care, and all the procedures were executed according to protocols approved by the Institutional Animal Care and Use Committee at the University of Nevada at Reno.
Assessment of Human Donor Hematopoietic Cell Engraftment
The BM, peripheral blood, and livers of animals transplanted with each of the MSC clones were analyzed 56–70 days after transplantation for the presence of donor (human) hematopoietic cells by flow cytometry with monoclonal antibodies [directly conjugated with fluorescein isothiocyanate (FITC) or phycoerythrin] to CD3, CD7, CD10, CD13, CD20, CD34, CD45, and CD33 (BD Biosciences) and Gly-A (Beckman Coulter, Miami, FL) according to the manufacturers' recommendations. The flow cytometry analysis was performed with a FACScan (BD Biosciences), and the results were compared to those obtained when identical staining was performed with an age-matched, nontransplanted control animal.
Immunohistochemical Analysis of Human Donor Cell Engraftment in the Liver
At least 6 different sections from 3 different lobes (right lobe, left lobe, and quadrate lobe) of the liver of every transplanted and control sheep were analyzed on an Olympus BX60 microscope equipped with an Olympus DP70 charge-coupled device camera. Paraffin sections (4 μm thick) were prepared on Superfrost Plus slides (Fisher, Pittsburgh, PA) from formalin-fixed, paraffin-embedded liver tissue obtained from transplanted animals and age-matched, nontransplanted control animals. The slides were incubated at 60°C for 45 minutes and deparaffinated by two 5-minute incubations in xylene followed by rehydration through a graded ethanol series to deionized water. To facilitate antigen detection, the slides were incubated at 90°C for 10 minutes with a target retrieval solution (Dako, Carpentaria, CA). Human cells were subsequently detected with the anti-human CD45 antibody (Biogenex, San Ramon, CA; to assess hematopoietic engraftment within the liver), anti-human albumin (Sigma), anti-human hepatocytes (clone OCH1E5, Dako), and anti-Ki67 (Biogenex) with a previously published methodology.7
Enzyme-Linked Immunosorbent Assay (ELISA) To Detect Human Albumin in the Serum of Chimeric Sheep
Albumin secretion into the serum was measured as a parameter for the functionality of the human hepatocytes detected by immunohistochemistry. To this end, we used a human-specific amplified biotin/streptavidin-based ELISA that was capable of detecting trace amounts (<200 pg/mL) of human albumin in sheep serum (Cygnus Technologies, South Port, NC). This kit was designed specifically to detect trace contamination of human serum albumin in pharmaceutical and other biological products and exhibited absolute species specificity. The sera of chimeric and control (age-matched, nontransplanted) sheep were run three times and analyzed according to the manufacturer's instructions. A standard curve was created from which the albumin concentration in each of the transplanted animals was extrapolated on the basis of an average absorption value. The final absorbance values were obtained by the subtraction of the zero standard absorbance values from all readings. Control sheep serum consistently gave absorbance values that fell within the background range seen with blank wells.
In Situ Hybridization for Human Cells
Paraffin-embedded blocks of liver from transplanted and control nontransplanted sheep were sectioned at 4 μM, attached to positively charged glass slides, and baked at 55°C for 30 minutes. The sections were then dewaxed in 2 changes of xylene for 10 minutes each. The sections were then air-dried, fixed in methanol/acetic acid (3:1) for 45 seconds, and allowed to air-dry. They were then rehydrated through a reverse-graded ethanol series (100%, 95%, and 70%) to deionized H2O, incubating for 5 minutes in each solution. The slides were then incubated at room temperature in 100 mM glycine in PBS for 30 minutes followed by 15 minutes in 50 mM NH4Cl to quench autofluorescence. The slides were washed with PBS for 3 minutes, and this was followed by 15 minutes of antigen retrieval at 95°C with a 1× Dako citrate buffer at pH 6. The slides were then allowed to cool for 20 minutes and washed twice for 2.5 minutes each in H2O. The slides were then treated with a commercially available proteinase K solution (Biogenex) for 5 minutes at room temperature and were washed twice for 2.5 minutes each in H2O. The sections were treated with 0.2 N HCl for 8 minutes and were washed twice for 5 minutes in H2O. Fifteen microliters of an FITC-labeled pan-human centromere probe (Rainbow Scientific Inc., Windsor, CT) was added to each section, and the sections were cover-slipped and sealed with rubber cement. Once the rubber cement had dried, the probe and target sequences were codenatured by heating at 95°C for 12 minutes. Hybridization was then allowed to proceed overnight at 37°C. On the following day, the slides were subjected to a rapid high-stringency quick-dip wash in 55% formamide, 2× standard sodium citrate (SSC), and 0.1% Tween 20, which was followed by a 3-minute wash at room temperature in 2× SSC and 0.1% Tween, a 5-minute wash in 0.15 M NaCl and 0.1 M trishydroxymethylaminomethane-HCl (pH 7.5), and a 3-minute wash in 2× SSC. The sections were then counterstained with 4′,6-diamidino-2-phenylindole for 3 minutes (BioGenex), rinsed for 5 minutes in PBS, and cover-slipped with Clarion (Biomeda, Foster City, CA), and images were acquired on an Olympus IX81 Fluoview FV1000 confocal microscope.
Statistics and Data Analysis
The experimental results are presented as the mean plus or minus the standard error of the mean. Comparisons between the experimental results were determined by a 2-sided, nonpaired Student t test analysis. A P value lower than 0.05 was considered statistically significant. The power of our statistical analysis was also validated by a 2-sided Wilcoxon rank sum test for independent samples
Transplantation of Clonally Derived Human MSCs by IH Injection Results in More Efficient Differentiation into Human Hepatocytes Than Administration by an IP Route
The IP transplantation of human BM MSCs results in no more than a few percent of donor (human)-derived hepatocytes. It is possible that in the fetal sheep model, BM MSCs transplanted IP do not preferentially migrate/lodge to the liver, and this results in fewer cells giving rise to hepatocytes. In the present study, we investigated whether directly delivering these cells to the hepatic parenchyma could increase the number of human MSCs reaching the liver and thus achieve higher levels of MSC differentiation into hepatocytes. To this end, we transplanted 55–60–day-old fetal sheep recipients with clonally derived MSCs by either an IH (n = 6) or IP (n = 6) route at a concentration of 5 × 105 cells per fetus, as described in the Materials and Methods section. The same single cell–derived MSC populations were transplanted IP and IH to ensure that the same cells were being analyzed, regardless of the route of administration.
At 56–70 days after transplantation, livers were collected from the transplanted animals, and the liver parenchyma was dissected into the different hepatic lobes. Six different tissue sections from 3 different lobes of the liver of every transplanted and control sheep were analyzed to assess the levels of donor-derived hepatocytes with a monoclonal antibody specific for human hepatocytes (hepatocyte paraffin [HEPAR]-1). As can be seen in Fig. 1, animals transplanted IH contained 12.5% ± 3.5% human hepatocytes, whereas animals transplanted IP had only 2.6% ± 0.4% human hepatocytes; this demonstrated that IH injection led to much more efficient generation of human hepatocytes than IP injection.
On the basis of the relatively high levels of donor (human)-derived hepatocytes seen with this detection method, we wished to confirm the specificity of our reagents prior to performing further analyses. To confirm that the cells being identified by the HEPAR antibody were indeed hepatocytes and were human in origin, we used 3 approaches throughout our studies. The first of these was to prepare serial sections from the transplanted sheep and perform staining for HEPAR and in situ hybridization with a human-specific centromeric probe on sequential sections. As can be seen in Fig. 2C,D, the nuclei of the cells staining with HEPAR hybridized to a commercially available human-specific probe, and this confirmed the human specificity of the HEPAR antibody. To ensure that these cells were hepatocytes, we used serial sections to perform HEPAR staining and immunohistochemistry with a human-specific albumin antibody. As can be seen in Fig. 2E,F, cells labeled with HEPAR (Fig. 2F) also were labeled with the human-specific albumin antibody (Fig. 2E), and this proved that they were hepatocytes. Finally, to confirm that the antibody against human albumin was in fact species-specific, we used the same approach described for HEPAR; that is, we prepared serial sections from the transplanted sheep and performed staining with the anti-human albumin antibody and in situ hybridization with a human-specific centromeric probe on adjacent sections. As can be seen in Fig. 2A,B, the nuclei of the cells staining with the anti-human albumin antibody hybridized to a commercially available human-specific probe, and this confirmed the human specificity of the anti-human albumin antibody.
IH Transplantation of Human MSCs Results in the Intraparenchymal Generation of Hepatocytes
A pronounced difference was found in the localization of the human hepatocytes generated within the sheep liver, depending on whether the animals had been transplanted IP or IH. Within the fetal sheep that received IP injections, human hepatocytes were distributed preferentially in the periportal area, in acinar zone 1 (Fig. 3B). In contrast, in animals transplanted IH and within the injected lobe, large numbers of human hepatocytes were widely distributed throughout the liver parenchyma (Fig. 3C), in acinar zone 2, with very few hepatocytes being generated around periportal spaces. Furthermore, in these same animals, when we analyzed the hepatic lobes distal to the site of IH injection, we also found the presence of considerable numbers of donor hepatocytes with an intralobular distribution, with a smaller number of donor hepatocytes surrounding perivascular areas (Fig. 3D). As in the previous experiments, the specificity of HEPAR was again confirmed by the staining of serial sections with an anti-human albumin antibody. As can be seen in Fig. 4, both HEPAR-positive cells near the periportal area (Fig. 4D) and those in the parenchyma (Fig. 4F) expressed human albumin (Fig. 4C and E, respectively). Although it is possible that the more efficient generation of donor-derived hepatocytes resulted primarily from the higher number of MSCs delivered upon the direct liver injection, we also examined whether the proliferative status of the liver cells played a role in this process. Figure 5 shows staining, with an antibody against Ki67, of a control nontransplanted fetal sheep liver at the time of gestation when the transplantation was performed (55 days). As can be seen, the great majority of the cycling cells were found within the hepatic parenchyma, whereas very few were seen in close association with the portal spaces. Thus, it is also possible that the higher levels of hepatocytes generated following IH injection are due, at least in part, to the delivery of the donor MSCs to the parenchyma in close proximity to the cycling endogenous host cells, which could provide an inductive environment with appropriate paracrine signals to facilitate the expansion of the human cells.
Higher Numbers of MSC-Derived Hepatocytes Generated in the Periportal Region Produce Human Albumin
It has been previously reported by several investigators that there is a preferential synthesis of certain plasma proteins, such as albumin, by hepatocytes localized in the periportal regions of the liver.37–39 These periportal hepatocytes possess the maximum albumin synthetic activity, and albumin synthesis decreases as the hepatocytes move radially from the periportal zone toward the central vein.37–39
Thus, the possibility existed that more of the human hepatocytes generated following IP injection would be albumin-synthesizing hepatocytes than those generated following IH injection, despite the higher numbers of HEPAR-positive hepatocytes found in the latter animals. To this end, we performed immunohistochemistry on sections of liver tissue from the different hepatic lobes of IP and IH transplanted animals, using a human-specific monoclonal antibody against human albumin. As can be seen in Fig. 4, sections of liver that contained human hepatocytes also contained human albumin–producing cells, demonstrating that the human hepatocytes generated in vivo by both IP and IH routes were functional. However, liver sections of animals transplanted IP (Fig. 6C) contained consistently larger and more numerous foci of human albumin–producing cells than animals injected via the IH route (Fig. 6B).
In order to confirm these results and to quantify the overall amount of human albumin synthesized by the livers of the different transplanted animals, we performed an ELISA assay on the serum of these animals that was specific for human serum albumin. As shown in Fig. 7, animals transplanted IP contained higher levels of human albumin (4 ± 0.48 ng/mL) in circulation than IH transplanted animals (1.2 ± 0.8 ng/mL; the normal level of fetal sheep albumin at this gestational age is 15 mg/mL). Our results demonstrate that by altering the site of injection, we can generate new hepatocytes in different hepatic zones that then have the ability to synthesize different protein/enzymes.
MSCs Contribute Directly to the Generation of Human Hepatocytes
Because in the fetal sheep model human MSCs are able to differentiate into hematopoietic cells30, 35 and we have previously demonstrated that human HSCs are able to efficiently give rise to hepatocytes upon transplantation in the sheep model,7 the possibility existed that the hepatocytes generated after MSC transplantation could be derived from the hematopoietic cells produced from the transplanted MSCs. To address this possibility, we evaluated the recipients of the clonally derived MSCs for hematopoietic engraftment after IP and IH transplantation. The flow cytometry analysis of BM and peripheral blood showed that animals transplanted with clonal MSCs via the IP route exhibited higher levels of human CD45+ cells in their BM (CD45 = 0.73% ± 0.1%) than the IH transplanted animals (CD45 = 0.35% ± 0.1%; P < 0.01); this showed that MSCs transplanted directly into the liver did not reach the BM to produce blood cells as efficiently as MSCs transplanted IP. Furthermore, in animals transplanted IP, multilineage hematopoietic engraftment was observed with the presence of CD7 (1.97%–6.2%), Gly-A (0.1%–4.08%), and CD3 (0.1%–2.2%), and the BM of these animals also contained CD34 cells (0.25%–0.26%). In contrast, in sheep fetuses transplanted by IH injection, some of the same MSC clones either were unable to produce multilineage engraftment or did so at lower levels than their counterparts injected by an IP route.
Because the production of human hepatocytes was significantly higher in IH transplanted animals that contained lower levels of donor hematopoietic cell engraftment, these results suggest that MSCs can give rise to hepatocytes without first having to differentiate into hematopoietic cells in the BM.
We also performed immunohistochemistry for human CD45 on tissue sections of the livers of animals transplanted IP and IH. As can be seen in Fig. 8, despite our ability to detect CD45-positive cells in human fetal liver sections (Fig. 8E,F), no human CD45-positive cells were found in the liver parenchyma of animals transplanted either IP or IH with the clonally derived MSC populations (Fig. 8A–D). These results collectively provide strong support for the conclusion that the hepatocytes were being generated by the direct differentiation of human MSCs without the need for a CD45+ hematopoietic intermediary step.
The liver is an organ with a high regenerative capacity provided by both mature hepatocytes and cholangiocytes and by the oval cells that compose the IH stem cell compartment. Despite this regenerative potential, however, many of the pathologies affecting this organ still require whole-organ transplantation to re-establish function. A preferable alternative to whole liver transplantation would be a cell-based treatment that, after administration, would lead to repopulation of the liver with normal cells and re-establishment of functionality. Unfortunately, in human clinical trials, the transplantation of hepatocytes has very rarely produced therapeutic effects, first because hepatocytes are difficult to expand in vitro and second because the number of viable hepatocytes obtained from a single donor for transplantation is too low to achieve a biologically relevant effect.40, 41 The use of oval cells for the restoration of liver cell mass has been attempted in several murine models.42 However, human oval cells are difficult to isolate, and similarly to the transplantation of mature hepatocytes, the modification of the host liver environment is required in order to achieve significant levels of liver repopulation by the transplanted cells. Thus, a somatic stem cell that could be used from an autologous source and could be expanded sufficiently to obtain the numbers required for successful transplantation, while maintaining its ability to differentiate into hepatocytes, could serve as a valuable source for cell therapy in liver diseases. Although HSCs can be considered a potential source of cells able to provide liver cell replacement,4, 6, 7, 43–45 MSCs possess several advantageous characteristics for cell-based therapy. These include their ease of isolation, their ability to expand extensively while maintaining their full differentiative potential, and their apparent hypoimmunogenicity and immunosuppressive effects upon transplantation.46, 47 In the present studies, we used an MSC population that was isolated on the basis of the expression of Stro-1 and that, even at the clonal level, exhibited a uniform phenotype and differentiative potential in vivo.35 Because the IP transplantation of human MSCs in the fetal sheep model did not result in more than a few percent of donor (human)-derived hepatocytes, we hypothesized that human MSCs may not have reached the liver effectively. It has been demonstrated that the in vitro migration of MSCs cells is regulated by SDF-1/CXCR-4 and hepatocyte growth factor/c-met networks.48, 49 The immunostaining of Stro-1+ clonally derived MSCs with antibodies against CXCR-4, c-met receptor, and SDF-1 showed that, prior to transplantation, all MSC clones were positive for SDF-1 but negative for both CXCR-4 and c-met receptor. Because during fetal liver development CXCR-4 is not expressed by hepatocytes and, in the adult liver, this chemokine is expressed only close to the central vein at the site of more mature, zone 3 hepatocytes,50 it is possible that the lack of this crucial factor involved in MSC migration resulted in diminished numbers of MSCs reaching the liver. Thus, in these studies, we compared 2 different routes of administration, IP and IH, in the hopes of increasing the levels of donor-derived hepatocytes upon MSC transplantation. We found that the IH injection of human MSCs resulted in more efficient generation of hepatocytes (12.5% ± 3.5% versus 2.6% ± 0.4%), as demonstrated by HEPAR-1 staining. These results are in contrast to other studies in which human MSCs directly injected into the liver resulted in very low levels of differentiation into hepatocytes.51 However, we have also shown that animals that received an IP injection exhibited a preferential periportal distribution of human hepatocytes and a higher percentage of albumin-producing hepatocytes. Because there is a preferential synthesis of certain plasma proteins, such as albumin, in hepatocytes localized in the periportal regions of the liver,37–39 it is possible that the hepatocytes generated by IP injection around the periportal area are more suited to producing albumin than hepatocytes that are generated in another lobular zone. However, even in the animals with a higher percentage of albumin-producing hepatocytes, the amounts of human albumin detected within the serum were very low. This is in agreement with other studies of murine models51, 52 in which the levels of hepatic cells generated, as assessed by immunohistochemistry, did not correlate with the amount of human albumin observed in the serum. Several possibilities can be envisioned that could account for this discrepancy, the first of which, as discussed by Nolta and colleagues,52 is the possibility that the hepatocyte-like cells that have formed within the xenogeneic recipient liver exist in several stages of differentiation, only the final one of which is capable of secreting albumin into the circulation. Thus, although all developmental stages of hepatocyte-like cells would be detected with HEPAR, only a percentage of those that had differentiated sufficiently would be detected with the anti-albumin immunohistochemistry, and only a small percentage of those would have differentiated sufficiently to possess the capacity to both produce and secrete albumin into the bloodstream. Another possibility that has not, to the best of our knowledge, been discussed in this context comes from studies of so-called ghost albumin, or immunounreactive albumin, that is found in the urine of patients with diabetes.53, 54 This albumin appears to be intact but is invisible to the antibodies routinely used to detect serum albumin, likely because of either abnormal processing or complexing with fatty acids, glucose, or other ligands masking epitopes recognized by these antibodies. It is thus conceivable that higher levels of human albumin are in fact being produced, but the albumin is being processed differently within the context of the xenogeneic sheep liver or has become complexed to various ligands within the sheep circulation; thus, key epitopes are masked, and it is rendered undetectable with the antibody in the ELISA kit that we employed.
In summary, our results provide evidence that altering the site of injection enables the de novo generation of hepatocytes in different hepatic zones and suggest that a combined transplantation approach may be necessary in order to successfully repopulate the liver. These studies also show that MSCs can give rise to hepatocytes without the need to first migrate/lodge to the BM and that, even though clonally derived MSCs were able to generate blood cells in both IP and IH transplanted animals, the contribution of donor-derived HSCs to hepatocyte generation was not significant. Importantly, our studies demonstrate, in a model that because of its intrinsic fetal characteristics does not use fusion as a mechanism for hepatocytic formation,7 that if optimal cell populations and administrative routes are employed, MSCs could be a valuable source of cells for liver repair and regeneration.
Dr. Dee Harrison-Findik is thanked for her helpful discussion and criticisms provided for this article.