Effects of two mesenchymal cell populations on hepatocytes and lymphocytes


  • Alejandro Gómez-Aristizábal,

    1. Faculty of Dentistry and Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • Corey Ng,

    1. Faculty of Dentistry and Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Joseph Ng,

    1. Faculty of Dentistry and Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
    • These authors contributed equally to this work.

  • John E. Davies

    Corresponding author
    1. Faculty of Dentistry and Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada
    2. Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada
    • Institute of Biomaterials and Biomedical Engineering, University of Toronto, 164 College Street, Room 407, Toronto, Ontario, Canada M5S 3G9
    Search for more papers by this author
    • Telephone: 416-978-1471

  • This study was supported by a Grant-in-Aid of Research from Tissue Regeneration Therapeutics, a Harron scholarship from the Faculty of Dentistry of the University of Toronto, and a doctoral completion award from the University of Toronto.


The inflammatory response to liver injury plays an important role in the onset of liver fibrosis, which may ultimately lead to liver failure. The attenuation of inflammation and hepatocyte rescue are, therefore, of the utmost importance for recovery. Mesenchymal stromal cells (MSCs) from adult bone marrow have been shown to rescue hepatocyte function. Here we explore a more plentiful source of neonatal MSCs: human umbilical cord perivascular cells (HUCPVCs). We cocultured HUCPVCs or bone marrow–derived mesenchymal stromal cells (BM-MSCs) with rat hepatocytes or human peripheral blood mononuclear cells in order to identify their effects on hepatocyte functionality and the proliferation of phytohemagglutinin-stimulated peripheral blood mononuclear cells (phaPBMCs). The expression of hepatotrophic factors by both types of MSCs in the presence of hepatocytes and the functional implications of blocking putative MSC anti-inflammatory factors were compared. Both types of MSCs improved albumin secretion, ureagenesis, hepatospecific gene expression, cytochrome P450 (CYP) activity, and functional hepatocyte mass maintenance. However, although HUCPVCs had an improved effect on the maintenance of ureagenesis, BM-MSCs had a strong effect on hepatocyte CYP activity. Additionally, each MSC type differentially expressed putative hepatotrophic factors, whereas phaPBMC proliferation was significantly decreased. Indoleamine 2,3-dioxygenase (IDO) was the main immunosuppressive mechanism used by both types of MSCs, but HUCPVCs exhibited higher expression of programmed death 1 ligands. However, the functional significance of the difference in anti-inflammatory factor expression still remains to be elucidated. Thus, both MSC types can serve as hepatocyte stromal cells and mitigate inflammation with IDO, but they present differences in the manner in which they affect hepatocytes and in the expression of both hepatotrophic and anti-inflammatory factors. Liver Transpl 18:1384–1394, 2012. © 2012 AASLD.

See Editorial on Page 1274

To date, the only successful way to treat liver failure is transplantation; however, this approach is available to only a select patient group. Although the incidence of acute liver failure (ALF) is very low, its mortality rate is 60% to 90%,1 and acute-on-chronic liver failure, which is much more common, accounted for approximately 1.1% of deaths in Canada between 2003 and 2007.2 Both ALF and cirrhosis, the main cause of acute-on-chronic liver failure, involve inflammation and cellular damage.3, 4 Upon liver injury, an inflammatory response is triggered to allow regeneration under normal conditions.3, 5 This requires the precise orchestration of hepatocyte and nonparenchymal (NP) cell proliferation, apoptosis suppression, angiogenesis, extracellular matrix remodeling, and immunomodulation.5-8 However, in some cases, inflammation does not abate, further injures the organ, and, depending on the intensity, leads to either ALF or chronic injury; in the latter case, the continuing regenerative response eventually loses equilibrium, and this leads to fibrogenesis, which may later develop into cirrhosis.3, 5 Therefore, diminishing inflammation and ameliorating the recovery of the liver cell content are promising alternatives to transplantation.

Mesenchymal stromal cells (MSCs) have been shown to secrete trophic factors that aid in healing9, 10; moreover, they have both anti-inflammatory and hepatotrophic properties. MSCs already have been used to treat inflammatory complications11 and are able to at least inhibit both T and B cell proliferation through different mediators such as the programmed death 1 (PD1) pathway and indoleamine 2,3-dioxygenase (IDO).10, 12-14 MSCs have also rescued hepatocytes in vitro, improved their survival, restored their polarity and functionality, and even induced their proliferation15 (Gómez-Aristizábal A, Davies JE. Human umbilical cord perivascular cells improve rat hepatocyte function ex vivo. Tissue Eng Part A. 2012). The ways in which MSCs affect hepatocytes include cytokine, extracellular matrix, and contact interactions.15 Thus, MSCs may be excellent candidates for supplying the therapeutic needs of ailing livers.

Although there is some debate about the role of MSCs in the generation of liver stroma in healthy and diseased individuals,16-19 there is increasing evidence for the potential of MSCs to treat liver disease. Murine bone marrow–derived mesenchymal stromal cells (BM-MSCs) have rescued injured livers with acute inflammation and prevented the onset of fibrosis4; also, nonhematopoietic cells from bone marrow (BM), which contains MSCs, have ameliorated carbon tetrachloride (CCl4)–induced liver fibrosis in mice.20 In rats, infused BM-MSCs have a significant effect in reversing liver fibrosis induced by CCl4 and dimethylnitrosamine.21 Furthermore, in a model of ALF in rats, BM-MSC-secreted factors diminished liver inflammation, improved survival, increased liver regeneration, and improved liver function.22 Kharaziha et al.23 showed that 8 patients with cirrhosis who were treated with autologous BM-MSCs had improvements in their disease with no adverse effects. In addition, human Wharton's jelly MSCs were shown by Tsai et al.24 to protect rat livers from fibrosis caused by CCl4 and heal them while diminishing liver inflammation.

As previously discussed, most research has shown the potential of adult BM-MSCs for liver therapy, although neonatal MSCs have shown a similar capacity14, 24 (Gómez-Aristizábal A, Davies JE. Human umbilical cord perivascular cells improve rat hepatocyte function ex vivo. Tissue Eng Part A. 2012). In our laboratory, MSCs were found in Wharton's jelly surrounding umbilical cord vessels; we call these cells human umbilical cord perivascular cells (HUCPVCs). HUCPVCs have many similarities to BM-MSCs, including an immunosuppressive phenotype, but they exhibit a higher proliferation rate and a higher colony forming unit/fibroblast frequency at harvest in comparison with BM-MSCs.14, 25, 26 Interestingly, Wharton's jelly MSCs have recently been shown to exhibit greater immunosuppression than BM-MSCs.27

Thus, we wished to directly compare BM-MSCs and HUCPVCs in light of their potential for liver therapeutics. In this article, we first highlight their similarities and focus on their capacity as putative hepatocyte stromal cells; then, we compare their anti-inflammatory profiles and mechanisms of inhibition of peripheral blood mononuclear cell (PBMC) proliferation.



Male Wistar Rats weighing 200 to 300 g (Charles River Laboratories, Montreal, Canada) were used for hepatocyte isolation. The animals were cared for in accordance with the guidelines set by the animal research ethics board of the University of Toronto and the National Institutes of Health.

Hepatocyte Isolation and Culture

Rat hepatocytes were isolated by 2-step collagenase perfusion of the liver as described previously by Moldéus et al.28 The cells were further purified by Percoll centrifugation (10 minutes at 50g with a final density of approximately 1.06 g/mL).29 The acquired hepatocytes were plated at a concentration of 2.83 × 104 hepatocytes/cm2, and they were maintained in hepatocyte culture medium (HCM), which consisted of Dulbecco's modified Eagle's medium (Sigma, Oakville, Canada) supplemented with 10% fetal bovine serum (FBS), 1 μM dexamethasone (Sigma), 1× Insulin-Transferrin-Selenium supplement (MP Biomedicals, Solon, OH), 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen, Burlington, Canada).

HUCPVC Culture and Source

HUCPVCs were provided by Tissue Regeneration Therapeutics, Inc. (Toronto, Canada). The cells were expanded in alpha minimal essential medium (α-MEM; Invitrogen) supplemented with 10% FBS and antibiotics (167 U/mL penicillin G, 0.3 mg/mL amphotericin B, and 50 mg/mL gentamicin; Sigma). The medium was changed 2 to 3 times a week, and cells were passaged at 80% confluency. For coculture experiments, HUCPVCs were pooled from at least 5 different donors.

Rat NP Liver Cell Isolation and Culture

After liver perfusion, hepatocytes were left to settle for 10 minutes. The supernatant resulting from this settling was taken and centrifuged at 300g for 5 minutes, and the pelleted cells were seeded with Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen). Although there were some hepatocytes in the population, after 3 passages, only NP cells remained according to morphology. Thus, only cells after passage 3 were used for experiments. For coculture experiments, NP cells were pooled from 3 different rats.

Human BM-MSC Culture and Source

BM-MSCs were also provided by Tissue Regeneration Therapeutics. They were cultured in the same way described for HUCPVCs. For coculture experiments, BM-MSCs were pooled from at least 5 different donors.

Hepatocyte Coculture

On the day after the seeding of hepatocytes, medium samples were collected; then, HUCPVCs, BM-MSCs, or NP cells were laid on top of the hepatocytes at a concentration of 1 × 104 cells/cm2. After this, the medium was changed daily, and samples were collected for albumin measurements and stored at −20°C. The medium used for both cocultures and monocultures was HCM, as described previously.


HUCPVCs and BM-MSCs were cultured on 6-well plates under previously described conditions. When approximately 80% confluency was reached, the RNA was isolated with TRI reagent (Sigma), and it was later purified with the RNeasy MinElute cleanup kit (Qiagen, Montreal, Canada) according to the manufacturer's instructions. The RNA purity and yield were determined with the NanoDrop 1000 (Thermo Fisher Scientific, Wilmington, DE), and the quality was determined with the Agilent 2100 bioanalyzer (Agilent Technologies, Mississauga, Canada). Eight HUCPVC biological replicates and 7 BM-MSC biological replicates were used for the microarray analysis with the GeneChip Human Gene 1.0 ST array (Affymetrix, Santa Clara, CA) according to the manufacturer's instructions. The microarray analysis was performed as described in the supporting information.


Immunofluorescence detection was performed after the samples were fixed with zinc formalin. A sheep anti-rat albumin antibody (1:500; Bethyl Laboratories, Montgomery, TX) was used as the primary antibody, and Alexa Fluor 555 donkey anti-sheep immunoglobulin G (1:600; Invitrogen) was used as the secondary antibody. Nuclei were also stained with Hoechst (1:2000). A more detailed description of the procedure as well as the method used to quantify the percentage of the area covered by albumin-positive cells (mature hepatocytes) can be found in the supporting information.

Hepatocyte Functional Assays

Albumin Enzyme-Linked Immunosorbent Assay

Daily samples stored at −20°C were tested for albumin levels by means of a rat albumin enzyme-linked immunosorbent assay (Bethyl Laboratories) according to the manufacturer's instructions.

Urea Secretion and Cytochrome P450 (CYP) Activity Measurements

On selected days, both cocultures and monocultures were incubated at 37°C with 5% carbon dioxide for 1 hour while they were rocked (200 μL in a 24-well plate) in serum-free HCM supplemented with 1 mM ammonium chloride (400 μM stock solution) and 5 μM Vivid BOMR substrate (Invitrogen; 2 mM stock solution in acetonitrile). After incubation, 100-μL samples were transferred to 96-well plates, and fluorescence (excitation at 530 nm and emission at 605 nm) was read at 37°C for the detection of CYP activity. Urea was quantified with the QuantiChrom urea assay kit (Bioassay Systems, Hayward, CA).

Proliferation Assay

HUCPVCs or BM-MSCs (1 × 104) were seeded in 96-well plates and were left to adhere overnight. Then, fresh carboxyfluorescein succinimidyl ester (CFSE)–stained PBMCs (1 × 105) that had been isolated from the buffy coats of peripheral blood samples (collected from normal healthy volunteers with informed consent) were added in 100 μL of a complete medium with phytohemagglutinin and inhibitors. The complete medium consisted of Roswell Park Memorial Institute 1640 medium with L-glutamine (80%), α-MEM (Invitrogen) (10%), 1 mM sodium pyruvate, and FBS (10%). In induced wells, phytohemagglutinin (Sigma) was used at a concentration of 10 μg/mL. Antibodies against human programmed death 1 ligand 1 (PD-L1), programmed death 1 ligand 2 (PD-L2), transforming growth factor β (TGF-β), hepatocyte growth factor (HGF), human leukocyte antigen G (HLA-G), leukemia inhibitory factor (LIF), and interleukin-10 (IL-10; 5 μg/mL each), IDO inhibitor 1-methyl-DL-tryptophan (1-MT; 1 mM), prostaglandin E2 inhibitor indomethacin (10 μM), and nitric oxide synthase inhibitor Nω-nitro-L-arginine methyl ester hydrochloride (L-NAME; 1 mM) were added as indicated. Phytohemagglutinin-stimulated peripheral blood mononuclear cell (phaPBMC) division on day 5 was identified via the halving of the CFSE fluorescence intensity with flow cytometry. A mouse immunoglobulin G1 isotype control was used to identify nonspecific effects of antibodies.

Priming With Interferon γ (IFN-γ)

HUCPVCs and BM-MSCs were cultured in α-MEM (Invitrogen) supplemented with 10% FBS, 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen). Two groups were set for the analysis of the effect of IFN-γ on the protein expression of the PD1 ligands and the up-regulation of IDO gene expression and activity: group 1 was the control group with no IFN-γ supplementation, and group 2 was stimulated with 500 U/mL recombinant human IFN-γ (2 × 107 IU/mg; US Biological, Swampscott, MA). All groups were seeded at the same cell concentration and were stimulated for 70 hours. The IDO expression and activity were then determined by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) and the tryptophan-to-kynurenine conversion of cell extracts, respectively (for further information, see the supporting information). The surface expression of PD-L1 and PD-L2 was determined with flow cytometry; anti–PD-L1 (5 μg/mL; clone MIH1, eBioscience, San Diego, CA) and anti–PD-L2 (2.5 μg/mL; clone MIH18, eBioscience) were used as the primary antibodies, and fluorescein isothiocyanate anti-mouse immunoglobulin G (1:200; Sigma) was used as the secondary antibody.

RNA Isolation, Reverse Transcription, and Relative qRT-PCR

Total RNA was isolated with TRI Reagent (Sigma), and then genomic DNA elimination and reverse transcription were performed with the QuantiTect reverse-transcription kit (Qiagen) according to the company's recommended procedures. Each sample was run in triplicate with SYBR Green JumpStart Taq ReadyMix (Sigma) according to the manufacturer's recommendations at annealing temperatures of 60°C and 78°C for fluorescence measurements. All rat primer pairs (Supporting Table 1) were designed and tested (ie, they were run on HepG2 complementary DNA) so that they would not cross-react with human RNAs. Human primer pairs (Supporting Table 2) were tested on hepatocytes, and none of them had any reactions except for fibronectin (FN1); fibronectin reacted, but the levels were approximately 40-fold less than the levels presented by pure HUCPVCs. Quantitation was performed through the normalization of the levels of all the genes to the housekeeping genes with Pfaffl's method30: β-actin for rats and β-2-microglobulin for humans.

Statistical Analysis

The statistical analysis of isolated time points was performed with the Wilcoxon rank-sum test or Student t test analysis after logarithmic transformation. Each figure legend indicates the technique used, the number of samples, and the P value that was considered to be significant.


1-MT, 1-methyl-DL-tryptophan; α-MEM, alpha minimal essential medium; ALF, acute liver failure; AMNION-MSC, amnion membrane–derived mesenchymal stromal cell; AT, adipose tissue; AT-MSC, adipose tissue–derived mesenchymal stromal cell; BM, bone marrow; BM-HSC, bone marrow–derived hematopoietic stem cell; BM-MSC, bone marrow–derived mesenchymal stromal cell; CCl4, carbon tetrachloride; CFSE, carboxyfluorescein succinimidyl ester; COL1A1, collagen type I α1; CYP, cytochrome P450; DCN, decorin; EGF, epidermal growth factor; ES, embryonic stem cell; FBS, fetal bovine serum; FIBROBLASP, primary dermal fibroblast; FIBROBLAST, dermal fibroblast cell line; FN1, fibronectin; GJA1, connexin 43; HCM, hepatocyte culture medium; HGF, hepatocyte growth factor; HLA, human leukocyte antigen; HUCPVC, human umbilical cord perivascular cell; HUVEC, human umbilical vein endothelial cell; IDO, indoleamine 2,3-dioxygenase; IFN-γ, interferon γ; IL, interleukin; JAG1, jagged 1; KITLG, stem cell factor; LAMB1, laminin β1; LIF, leukemia inhibitory factor; L-NAME, Nω-nitro-L-arginine methyl ester hydrochloride; MHC, major histocompatibility complex; mRNA, messenger RNA; MSC, mesenchymal stromal cell; NP, nonparenchymal; PBMC, peripheral blood mononuclear cell; PD1, programmed death 1; PD-L1, programmed death 1 ligand 1; PD-L2, programmed death 1 ligand 2; phaPBMC, phytohemagglutinin-stimulated peripheral blood mononuclear cell; qRT-PCR, quantitative reverse-transcription polymerase chain reaction; TAT, tyrosine aminotransferase; TERT, telomerase reverse transcriptase; TGF-β, transforming growth factor β; TNF, tumor necrosis factor; TO, tryptophan 2,3-dioxygenase; V-SMC, vascular smooth muscle cell.


HUCPVC and BM-MSC Share a Similar Phenotype

HUCPVCs, as previously mentioned, have shown many properties indicating that they belong to the MSC type. To further this knowledge, we performed a direct comparison of the transcriptome of HUCPVCs with the transcriptome of BM-MSCs and other cell types and tissues from public data. Using hierarchical clustering for all samples, we confirmed the validity of the clustering by observing the clustering of known related samples (ie, hepatocytes with liver tissue) and the lack of aggrupation of cells of different origins. Thus, the clustering showed that both HUCPVCs and BM-MSCs belonged to a group containing only spindle-shaped mesenchymal cells; this group included BM-MSCs and other MSCs acquired and cultured by other groups (Fig. 1A). Hierarchical clustering based on the expression profile of putative MSC hepatotrophic factors15 showed aggrupation similar to the whole transcriptome analysis, but in this case, a subgroup was formed with most MSCs; amnion membrane–derived mesenchymal stromal cells (AMNION-MSCs) were excluded, and vascular smooth muscle cells (V-SMCs) were included. High expression of IL-6 and overall higher expression of the putative hepatotrophic profile in comparison with the other cells were characteristic of this subgroup (Fig. 1B).

Figure 1.

Similarities between HUCPVCs and BM-MSCs. (A) Hierarchical clustering of different cell types and tissues according to the transcriptome. (B) Hierarchical clustering of different cell types according to the expression of putative hepatotrophic factors. White indicates high levels, and black indicates low levels. (C) Telomerase activity on HUCPVCs as measured with TRAPeze. HeLa protein extracts were used as positive controls. Error bars indicate standard deviations (n = 6). (D) Number of MHC class II molecules in HUCPVCs with higher, similar, or lower expression in comparison with BM-MSCs at the mRNA level. Genes were considered different when they differed more than 1.5 times between cell types, and they were significantly different at α = .005. For parts A, B, and D, data were obtained with the Affymetrix Human ST 1.0 microarray.

Focusing on HUCPVCs and BM-MSCs, we found similarities in factors that are relevant for cell therapy. One of them was telomerase expression: HUCPVCs had levels of telomerase reverse transcriptase (TERT) messenger RNA (mRNA) similar to those of BM-MSC (P > 0.1). This was also confirmed by the fact that HUCPVCs lacked telomerase activity (Fig. 1C), as we have previously reported.31

Also, it is known that both types of MSCs are nonalloreactive and lack the expression of major histocompatibility complex (MHC) class II molecules14, 32; thus we decided to investigate in more detail the relative expression of MHC class II and the costimulatory molecules CD80 and CD86. We considered only differences that were greater than 1.5-fold with a P value less than 0.005. We found that the expression of 11 of the 14 analyzed MHC class II molecules was similar between the 2 cell types, but the expression of 3 molecules was lower in HUCPVCs (Fig. 1D). The expression of costimulatory molecules was similar; the expression of CD80 and CD86 was 0.7-fold (P = .002) and 0.87-fold (P = .015), respectively, in HUCPVCs versus BM-MSCs.

Furthermore, when the expression of factors that could serve liver healing was compared, we found that most were similarly expressed under basal conditions: 8 of the 11 analyzed hepatotrophic factors were similar, only 1 of 7 antiapoptotic factors was different, all the angiogeneic and antifibrotic cytokines were similarly expressed, 13 of 16 anti-inflammatory factors showed similarities (Supporting Table 3), and 15 of 22 analyzed chemokines did not show significant differences (Supporting Fig. 1).

Despite the many similarities found, 3 chemokines were expressed at significantly lower levels in HUCPVCs versus BM-MSCs, whereas 4 chemokines (all proangiogeneic33) were expressed at significantly higher levels (Supporting Fig. 1).

Differentially Expressed Hepatotrophic Factors

Under basal conditions, 3 hepatotrophic genes were expressed at significantly lower levels in HUCPVCs versus BM-MSCs: the cytokine stem cell factor (KITLG), the proteoglycan decorin (DCN), and the notch ligand jagged 1 (JAG1). HGF also showed lower expression, but this was significant only at P < 0.01 (Supporting Table 3).

Because of the differences under basal conditions, we decided to explore whether they were maintained upon coculturing with hepatocytes because of cell-cell crosstalk. Using qRT-PCR, we found that the HUCPVC/BM-MSC gene expression ratio was significantly changed. Although we were unable to achieve confident quantification of collagen type I (COL1A1), epidermal growth factor (EGF), and tumor necrosis factor α (TNF-α) because of their low expression with respect to the housekeeping gene, most of the other genes showed variations with respect to the basal conditions: DCN was similarly expressed in the 2 cell types, whereas laminin (LAMB1), HGF, IL-6, and connexin 43 (GJA1) showed higher expression in HUCPVCs (fold difference > 2.5 and P < 0.05; Fig. 2). Thus, under coculture conditions, HUCPVCs showed higher expression of 4 hepatotrophic factors, whereas BM-MSCs showed higher expression of 2 hepatotrophic factors (KITLG and JAG1); the profile was significantly different from the one observed under basal conditions.

Figure 2.

mRNA expression of selected hepatotrophic factors (detected by qRT-PCR) in HUCPVCs and human BM-MSCs after coculturing with hepatocytes. The results were normalized to BM-MSCs and β-2-microglobulin. The t test after logarithmic transformation was used. Error bars indicate standard deviations from 3 cocultures, each containing hepatocytes from a different subject. *P < 0.05.

HUCPVCs Support Hepatocytes Differently Than BM-MSCs

Because the 2 cell types have different hepatotrophic profiles both under basal conditions and in cocultures with hepatocytes and can support hepatocytes in vitro, we directly compared their effects on hepatocytes in addition to effects caused by rat liver NP cells.

Using fluorescent emissions from albumin-expressing (albumin-positive) cells to quantify the area covered by mature hepatocytes (which could be correlated to the number of functional hepatocytes), we found that mature hepatocytes in cocultures with HUCPVCs or BM-MSCs covered similar areas at the end of the culture (P > 0.1). Also, the area covered by mature hepatocytes in MSC cocultures was higher than the areas covered by the monoculture and the NP-hepatocyte coculture (P < 0.01; Fig. 3A). In comparison with the monoculture, the NP-hepatocyte coculture had a higher area covered by mature hepatocytes (P < 0.01), although it was significantly lower than the area on day 1 (Fig. 3A). On the other hand, both types of MSCs had increases in the area covered by mature hepatocytes (P < 0.05; Fig. 3A) and in the hepatocyte colony size (Fig. 3B).

Figure 3.

Effects of HUCPVCs, BM-MSCs, and NP cells on mature hepatocyte masses. HUCPVC, BM-MSC, and NP cocultures with hepatocytes were used. (A) Quantification of the area occupied by albumin-positive hepatocytes with respect to the whole culture area. The Wilcoxon rank-sum test was used. Error bars indicate standard errors from 3 biological replicates. *P < 0.01 versus the monoculture. (B) Merged phase contrast and fluorescent images on day 10. Albumin-secreting hepatocytes were revealed by anti-rat albumin antibodies (white). The field width was 904 μm.

MSC-hepatocyte cocultures also showed a higher effect at the functional level. Although the secreted albumin levels of monocultures continuously decreased, an increase was seen in cocultures after an initial period of decrease. Noteworthy was the fact that MSC-hepatocyte cocultures increased in albumin secretion much more than NP-hepatocyte cocultures. However, by the last day of the culture, HUCPVC cocultures were outperformed by BM-MSCs because the albumin levels suddenly decreased (Fig. 4A). Despite that, both MSC-hepatocyte cocultures showed improved expression (P < 0.01) of the hepatospecific proteins albumin and tryptophan 2,3-dioxygenase (TO). Tyrosine aminotransferase (TAT) was, however, up-regulated significantly (P < 0.05) only in HUCPVC cocultures (Fig. 4B).

Figure 4.

Effects of HUCPVCs, BM-MSCs, and NP cells on the hepatocyte function and phenotype. (A) Secretion of albumin by hepatocytes over time. (B) Relative albumin, TO, and TAT gene expression normalized to actin B and expression on the first day. Error bars indicate standard deviations from 3 biological replicates. The t test after logarithmic transformation was used. (C) Urea in the medium with respect to day 1 (3.7 mg/dL with a standard error of 0.1 mg/dL and 0.4 mL/well). (D) CYP activity quantified by the production of resorufin from BOMR. The dashed line indicates the level on day 1. For parts A, C, and D, the Wilcoxon rank-sum test was used; the error bars indicate standard errors of the mean from 3 biological replicates and 2 sample replicates. *P < 0.01 versus the monoculture.

The maintenance of ureagenesis was improved in HUCPVC cocultures versus the monoculture for the last 2 measurements; however, BM-MSCs showed significance in the last measurement only, and NP cells failed to show any improvement versus the monoculture (P < 0.01; Fig. 4C). On the other hand, only BM-MSC cocultures showed an increase in CYP activity (Fig. 4D).

Because some reports have shown that MSCs have the capacity to transdifferentiate into hepatocytes15 and thus interfere with our measurements of ureagenesis and CYP activity, we assayed for differentiation in the cocultures. Using reverse-transcription polymerase chain reaction, we looked for human albumin mRNA in cocultures. We found none in either HUCPVC or BM-MSC cocultures (data not shown).

Differences in the Expression of Anti-Inflammatory Factors

Because inflammation plays an important role in the onset and continuation of liver disease, it is also highly relevant to characterize the differences in the expression of anti-inflammatory factors. From the microarray data, we identified 3 factors that were highly expressed by HUCPVCs versus BM-MSCs. Two of these factors were the PD1 ligands, and the other factor was IL-11. We decided to reconfirm the difference in the PD1 ligands because their expression had previously been directly linked to MSC immunosuppression.12, 34 qRT-PCR confirmed that PD-L1 expression was higher in HUCPVCs versus BM-MSCs, which had almost undetectable expression (Supporting Fig. 2A). According to flow cytometry, PD-L1 and PD-L2 were constitutively expressed in HUCPVCs. However, upon IFN-γ stimulation, PD-L1 was up-regulated in both cell types, and PD-L2 showed a marginal increase in comparison with native BM-MSCs (Fig. 5).

Figure 5.

PD1 ligands have higher expression in HUCPVCs, and PD-L1 is up-regulated in both types of MSCs by IFN-γ. HUCPVCs and BM-MSCs were cultured in the absence or presence of IFN-γ (500 IU/mL) for 70 hours. The values were taken from flow cytometry analysis. Dead cells were excluded with propidium iodide. On the left, mean fluorescence intensities are shown in bar graphs. The t test after logarithmic transformation was used. *P < 0.05, **P < 0.01, and ***P < .001. Error bars indicate standard errors from more than 4 biological replicates.

phaPBMC Proliferation Is Reduced by Both MSCs Mainly Through IDO

Thus, we decided to test the effects of the differential expression of the PD1 ligands along with other MSC-relevant anti-inflammatory factors on PBMC proliferation. Using antibodies or chemical inhibitors for each of the systems, we found that both HUCPVCs and BM-MSCs were able to significantly suppress phaPBMC proliferation under all conditions except when the action of IDO was inhibited by 1-MT (Fig. 6A).

Figure 6.

HUCPVCs and BM-MSCs inhibit phaPBMC proliferation mainly by IDO. (A) phaPBMC proliferation as the percentage of cells belonging to generation 2 or higher (identified by the decrease in CFSE fluorescence with respect to generation). The data represent the results of triplicate experiments with the interactions of phaPBMCs from 1 donor with HUCPVCs from 4 umbilical cords and BM-MSCs from 4 donors (4 biological replicates). (B) IDO activity measured as tryptophan degradation to kynurenine. Data from 5 biological replicates were used. For parts A and B, the t test after logarithmic transformation was used. Error bars indicate standard errors. *P < 0.05 and ***P < .001.

The microarray revealed that BM-MSCs and HUCPVCs had similar IDO expression under basal conditions, and this was corroborated by the lack of expression and activity seen in unstimulated cells. However, IFN-γ priming significantly up-regulated IDO mRNA and activity in both cell types. Specifically, BM-MSCs showed higher up-regulation of the activity versus HUCPVCs, whereas they had similar upward tendencies at the mRNA level (Fig. 6B and Supporting Fig. 2B).


MSCs have shown considerable potential for the healing of numerous types of injuries and illnesses.9, 10, 15, 26, 31 In addition, their negligible immunogenicity has been the focus of much attention.9, 10, 31 The use of MSCs for extending the function of hepatocytes in bioartificial liver systems and for treating the liver per se is feasible because they have been shown to serve as putative hepatocyte stroma and as anti-inflammatory agents10, 13, 14 and can provide other trophic properties that facilitate healing.9

In previous reports, it has been shown that HUCPVCs and BM-MSCs possess similar characteristics with respect to the MSC phenotype, including their lack of telomerase.14, 25, 26, 31, 32 However, their transcriptomes had not been previously compared. We found that HUCPVCs and BM-MSCs cultured under the same conditions shared many similarities, although changing the culture conditions did introduce differences (Fig. 1A). This parallels the results of Wagner et al.,35 who showed that the BM-MSC morphology changed with the culture conditions employed. Similarly, the constitutive chemokine expression was generally similar in the 2 cell types, and this suggests that they are able to attract immune cells in order to exert their anti-inflammatory effects.36

Interestingly, among the 7 differentially expressed chemokines, the 4 that were up-regulated in HUCPVCs were angiogeneic,33 although the angiogeneic cytokines were not significantly different (Supporting Table 3). Hypoxia plays an important role in cirrhosis and other liver diseases, and despite neovascularization, the new vessels fail to resolve hypoxia because of their immaturity.37, 38 Because of the role of perivascular populations in the maturation of vessels,38, 39 it is possible that HUCPVC-induced angiogenesis may be accompanied by HUCPVC-facilitated maturation. Indeed, the high expression of vascular endothelial growth factor receptor 1 (FLT1) (29-fold versus BM-MSCs) and vascular endothelial growth factor receptor 2 (FLK1) (3-fold) suggests that HUCPVCs can react and sequester vascular endothelial growth factor more effectively and thus play an important role in the modulation of vessel formation.40, 41

In the liver, NP and parenchymal cells (hepatocytes) require a balanced interaction42, 43; this interaction is, however, lost in vitro, and their reunion is not able to completely restore the original homeostasis.44, 45 However, numerous reports have shown that the coculture of all or part of the NP cell population with hepatocytes helps to maintain a more differentiated state in the latter.43-45 Here, for the first time, we have shown that both HUCPVCs and BM-MSCs are able to provide a higher degree of support to hepatocytes than their native stromal cells. The reason that NP cells fail to provide strong support might be related to the fact that the main NP constituting cell types, stellate and sinusoidal endothelial cells, suffer from a change in the phenotype in vitro.42 MSCs, too, change phenotype,35 but it appears that their tendency to restore balance is greater, and thus they provide more adequate support to hepatocytes (an important property for therapeutics).

The support provided by HUCPVCs and BM-MSCs can be attributed to their hepatotrophic phenotype, the levels of which differ between the 2 cell types (Fig. 2 and Supporting Table 3). This in part explains why these 2 cell types show differences in their effects on both ureagenesis and CYP activity. These differences appear to suggest that because of the accelerated maintenance of ureagenesis by HUCPVCs, the latter might be preferable for systems that aim to eliminate hyperammonemia, which can cause critical complications in liver failure.46 On the other hand, the improvement in CYP activity in the presence of BM-MSCs indicates that they could be more appropriate for treating cases of xenobiotic-related liver injury and for creating more stable testing devices in pharmacology.

Nevertheless, despite the benefits that MSCs can provide to hepatocyte maintenance, it can be argued that the improved hepatotrophicity might not be exclusive to MSCs. Indeed, it has been shown that other cells can provide support to hepatocytes.22, 43 Yagi et al.22 directly compared BM-MSCs and NIH 3T3-J2 fibroblasts (a highly hepatotrophic cell line47), and their results indicated that the 2 types of cells had similar supportive capabilities under normal conditions. However, BM-MSCs showed improved hepatoprotection and support of ureagenesis in comparison with fibroblasts in the presence of serum from ALF subjects. Furthermore, BM-MSCs had a marked capacity to rescue rats from ALF, in part by reducing acute inflammation.22

Our results showed that the 2 types of MSCs had similar anti-inflammatory profiles (Supporting Table 3). However, 3 interesting differences were identified. IL-11 had higher expression in HUCPVCs. IL-11 has been regarded as an anti-inflammatory cytokine because of its capacity to reduce the secretion of inflammatory cytokines and reduce colitis48; however, it has not been shown to be implicated in BM-MSC immunosuppression.49 The other observed differences were related to the higher expression in HUCPVCs of the 2 PD1 ligands, which have been shown to play roles in MSC immunosuppression.12, 34 Our results, however, differed from those of Tipnis et al.34 and Augello et al.12 Augello et al. showed that the 2 ligands were involved in the inhibition of T and B cell proliferation by murine BM-MSCs at a ratio of 1:1, whereas we saw no difference in whole human PBMC proliferation at a 1:10 ratio. Tipnis et al., on the other hand, tested their hypothesis on isolated human T cells that were not stimulated by any agent, and they analyzed their results only on the basis of the CD4+ portion and thus tested for immunoprivilege instead of immunosuppression. Even though we tested this model with PBMCs (a more relevant physiological model), we saw no effect by blocking the PD1 ligands, IDO, or prostaglandin E2 (Supporting Fig. 3). Thus, the relevance of the constitutive expression of PD1 ligands by HUCPVCs still remains to be elucidated. It is possible that because PD-L1 has been recently found to be a modulator of angiogenesis, HUCPVCs may be involved in angiogeneic modulation.50

IFN-γ priming of both cell types showed up-regulation of PD-L1 and IDO. IDO was up-regulated to higher levels in BM-MSCs versus HUCPVCs, and this result corroborates the findings of Prasanna et al.,27 who compared Wharton's jelly MSCs and BM-MSCs. However, this difference was not reflected in a difference in PBMC proliferation when IDO was blocked in phaPBMC-MSC cocultures. Although our results showed that IDO was crucial for the inhibition of PBMC proliferation by both cell types in our model, it is important to keep in mind that IDO can have a cooperative effect on the establishment of B cell–mediated inflammatory responses,51 and thus other mechanisms9 may also be important.

Interestingly, both the PD1 and IDO pathways appear to be involved in the modulation of liver inflammation,8, 52, 53 and this indicates a putative mechanism for MSC mediation of liver inflammation. The fact that HUCPVCs and BM-MSCs have different levels of expression of the factors involved in the 2 pathways suggests that they could also have differential effects on acute and chronic inflammation and the induction of tolerance in transplantation. Still, we know to date only that the 2 cell types can similarly regulate PBMC proliferation through IDO.14 It is possible that a comparative analysis of the effects of the 2 MSC types in specific immune cell populations (ie, T and B cells, dendritic cells, and macrophages) could shed some light on the importance of the differential expression of IDO and PD1 ligands. Similarly, experiments in which these 2 MSC types are directly compared in the ways in which they affect ALF, chronic hepatitis, and liver allograft tolerance are required to elucidate the possible differences at the functional level.

As previously mentioned, others have shown that BM-MSCs can reverse liver fibrosis,21 although a study by Carvalho et al.54 showed no beneficial therapeutic effect of MSCs in a rat model of severe chronic liver injury. Positive findings were reported by Tsai et al.,24 who employed human Wharton's jelly cells in a rat cirrhosis model. Stellate cells, the perivascular stromal cells of the liver, are derived from mesoderm,55 and some authors have demonstrated that in models of cirrhosis, BM-MSCs can home to the liver and contribute to the fibrous stroma. Thus, both Russo et al.18 and Li et al.16 clearly demonstrated that fibrotic hepatic stroma in mice was mesenchymal in origin, although Dalakas et al.17 also provided evidence that hematopoietic BM stem cells (CD34+ BM cells) may also contribute to the stromal compartment in hepatitis patients. All this evidence indicates that different BM-MSC populations may play different roles in fibrogenesis.

We have been able to show at the single cell clonal level that a hierarchy of MSC phenotypes exist in which different MSCs have diminishing multilineage differentiation capacity,26 but we do not yet know whether cytokine secretion also varies within such heterogeneous populations. Taken together, our current findings indicate that MSCs of 2 different tissue origins can serve as better hepatocyte stromal cells than isolated liver stromal cells, and they mitigate inflammation with IDO. However, they present differences in the manner in which they affect hepatocytes and in the expression of both hepatotrophic and anti-inflammatory factors, and this suggests that MSCs from different tissue sources may have differential therapeutic effects.


The authors thank Ms. Katie Chan for her technical expertise with hepatocyte isolation, Dr. Rahim Moineddin for his advice on statistics, Dr. Vanessa Mendes for her aid with the hepatocyte isolation procedure, and Dr. Yimin Wang for culturing and processing the samples for microarrays.