Human Liver Stem Cells Improve Liver Injury in a Model of Fulminant Liver Failure

Maria Beatriz Herrera,1,2 Valentina Fonsato,1 Stefania Bruno,1 Cristina Grange,1 Nicholas Gilbo,3 Renato Romagnoli,3 Ciro Tetta,4 and Giovanni Camussi1

Liver transplantation is currently the only effective therapy for fulminant liver failure, but its use is limited by the scarcity of organs for transplantation, high costs, and lifelong immunosuppression. Here we investigated whether human liver stem cells (HLSCs) protect from death in a lethal model of fulminant liver failure induced by intraperitoneal injection of D-galactosamine and lipopolysaccharide in SCID mice. We show that injection of HLSCs and of HLSC-conditioned medium (CM) significantly attenuates mouse mortality in this model. Histopathological analysis of liver tissue showed reduction of liver apoptosis and enhancement of liver regeneration. By optical imaging we observed a preferential local- ization of labeled HLSCs within the liver. HLSCs were detected by immunohistochemistry in large liver vessels (at 24 hours) and in the liver parenchyma (after day 3). Fluorescence in situ hybridization analysis with the human pan-centromeric probe showed that positive cells were cytokeratin-negative at 24 hours. Coexpression of cytokeratin and human chro- mosome was observed at 7 and, to a lesser extent, at 21 days. HLSC-derived CM mimicked the effect of HLSCs in vivo. Composition analysis of the HLSC-CM revealed the presence of growth factors and cytokines with liver regenerative properties. In vitro experiments showed that HLSC-CM protected human hepatocytes from apoptosis and enhanced their proliferation. Conclusion: These data suggest that fulminant liver failure may potentially benefit from treatment with HLSCs or HLSC-CM. (HEPATOLOGY 2013;57:311-319)

Fulminant liver failure (FLF) is a life-threatening

disease for which liver transplantation is the

1
only definitive treatment, but the scarcity of

donor livers and the timing of available organs often precludes transplantation. Liver regeneration could also be facilitated by using a bioartificial liver, but this

Abbreviations: CM, conditional medium; FLF, fulminant liver failure; GalN, D-galactosamine; HLSCs, human liver stem cells; LPS, lipopolysaccharide; MSCs, mesenchymal stem cells.

From the 1Department of Internal Medicine and Molecular Biotechnology Center, University of Turin, Italy; 2Sis-Ter S.p.A., Palazzo Pignano (CR), Italy; 3Liver Transplantation Center, University of Torino, Torino, Italy; 4Fresenius Medical Care, Bad Homburg, Germany.

Received December 15, 2011; accepted July 16, 2012.

Funded by Regione Piemonte, Piattaforme Biotecnologiche, Pi-Stem project, Converging Technologies NanoIGT.
Address reprint requests to: Dr. Giovanni Camussi, Cattedra di Nefrologia, Dipartimento di Medicina Interna, Ospedale Maggiore S. Giovanni Battista, Corso Dogliotti 14, 10126, Torino, Italy. E-mail: giovanni.camussi@unito.it; fax: þ39-011-6631184.

Copyright VC 2012 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com.
DOI 10.1002/hep.25986
Potential conflicts of interest: Drs. Fonsato and Herrera Sanchez received

grants from Sis-Ter S.p.a. Drs. Camussi and Tetta is an employee of and received grants from Fresenius Medical Care.
Additional Supporting Information may be found in the online version of this article.

approach is limited by the lack of availability of viable hepatocytes, required by the bioreactor.2 Stem cells, which can be expanded in vitro and cryopreserved, could form the basis of an ideal therapy.1 Liver stem cells, or even stem cells derived from other tissues, could potentially provide a source of human hepato- cytes for regeneration of the injured liver.3,4

In particular, mesenchymal stem cells (MSCs), shown to be capable of in vitro differentiation into hepato- cytes,5 were investigated as a possible source of hepato- cytes for liver regeneration. In addition, it has been shown that secretion of trophic molecules by MSCs may favor regeneration following acute liver injury.6

In a previous study, we isolated a population of human adult liver stem cells (HLSCs) expressing MSC markers and certain embryonic and hepatic cell markers, and having multipotent differentiation capa-

7
bilities and regenerative properties. However, the ther-

apeutic potential of HLSCs and HLSC-conditioned medium (CM) in FLF has not yet been evaluated. In this study we investigated the effect of HLSCs and HLSC-derived CM in a lethal model of liver injury induced by D-galactosamine (GalN) and lipopolysac- charide (LPS) in SCID mice.

311

312 HERRERA SANCHEZ ET AL.

HEPATOLOGY, January 2013

Materials and Methods

Culture of HLSCs, MSCs, and Human Hepatocytes.

HLSCs were isolated from human cryopreserved nor- mal hepatocytes and MSCs were obtained from Lonza (Basel, Switzerland) and were cultured as described in the online Supporting Information.7,8

Preparation of CM. Detailed protocols for the preparation of CM from HLSCs or MSCs9 are pro- vided in the online Supporting Information.

The CM was analyzed for specific proteins, using multiplex biometric immunoassay, Bioclarma (Bio-Plex Human Cytokine Assay; Bio-Rad Laboratories, Hercu- les, CA) and data were confirmed by enzyme-linked immunosorbent assay (ELISA).

SCID Mice Model of FLF. Studies were approved by the University of Torino Ethics Committee and conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Intramuscular injection of Zolazepam (0.2 mL/kg) and Xilazin (16 mg/kg) were used as anesthe- sia (40 mL/mouse). FLF was induced in male SCID mice (7-8 weeks old) (Charles River Laboratories, Mi- lan, Italy), by intraperitoneal injection of GalN (600 mg/kg body weight) and LPS (125 ng per animal).10 Injection of GalN and LPS induced liver injury caus- ing apoptosis and necrosis of hepatocytes with 100% lethality at 8 hours.

Thirty minutes after GalN/LPS administration, mice received different treatments. The following groups were studied: group 1, FLF mice intravenously injected with vehicle alone (n 1⁄4 18); group 2, healthy mice intraperitoneally injected with vehicle instead of GalN/LPS (n 1⁄4 6); group 3, FLF mice intravenously injected with 2 􏰀 106 HLSCs (n 1⁄4 9, 3.3 􏰀 105 cells given six times for a total number of 2 􏰀 106); group 4, FLF mice intravenously injected with 2 􏰀 106MSCs(n1⁄46,3.3􏰀105cellsgiven6timesfora total number of 2 􏰀 106); group 5, FLF mice intra- peritoneally injected with 30 􏰀 106 HLSCs (n 1⁄4 9); group 6, FLF mice intraliver parenchyma (LP) injected (intheleftlobe)with0.5􏰀106 (n1⁄45)or0.2􏰀 106 HLSCs (n 1⁄4 5); group 7, FLF mice intraperitone- ally injected with concentrated HLSC-CM (n 1⁄4 14); group 8, FLF mice intraperitoneally injected with con- centrated MSC-CM (n 1⁄4 5); group 9, FLF mice in- traperitoneally injected with vehicle (PBS; n 1⁄4 5); group 10, FLF mice intraperitoneally injected with concentrated HLSC-CM pretreated for 1 hour at 4􏰁C with anti-HGF blocking antibody (100 mg/mouse; GeneTex, Aachen, Germany) (n 1⁄4 5); group 11, FLF mice intraperitoneally injected with 25 ng/mouse of

rhHGF (PreproTech, Rocky Hill, NJ,) (n 1⁄4 5). The intravenous injection (120 mL) was performed in the vein of the tail within 30 seconds. In all experiments, cells cultured in T75 flasks until the 2-6 passage were detached by trypsin (0.5% w/v), washed, and resus- pended in phosphate-buffered saline (PBS). Before in vivo injection, HLSCs were stained with the CellTrace CFSE Cell Proliferation Kit (Molecular Probes, Life Technologies, Paisley, UK).11

Morphologic Studies and Fluorescence In Situ Hybridization (FISH). Full details are provided in the online Supporting Information.

Optical Imaging. Full details are provided in the online Supporting Information.

Apoptosis and Proliferation of Hepatocyte In Vitro. Full details are provided in the online Support- ing Information.

Statistical Analysis. Data were analyzed using t tests, analysis of variance (ANOVA) with Newmann- Keuls’ or ANOVA with Dunnett’s multicomparison tests as appropriate. For survival experiments, a log- rank test was conducted. P < 0.05 was considered significant.

Results

HLSCs Improved Survival in a Model of FLF in SCID Mice. The intraperitoneal administration of 600 mg/kg of GalN and 125 ng of LPS in SCID mice induced FLF with a 100% mortality rate within 8 hours (Fig. 1A). To evaluate whether HLSCs might rescue mice with FLF, the cells were administered 30 minutes after GalN/LPS-induced injury in different protocols: single intraperitoneal injection (3 􏰀 107 cells); six repeated intravenous injections (3.3 􏰀 105 cells, total: 2 􏰀 106 cells); LP injection at two concen- trations (0.2 and 0.5 􏰀 106). Survival rates were 77%, 100%, and 100% in the case of intraperitoneal, intra- venous, and LP injections, respectively (Fig. 1A,B). An equal number of MSCs was intravenously injected six times (3.3 􏰀 105 cells, total: 2 􏰀 106 cells) using the same protocol. The survival rate was 33% (Fig. 1A). Cell injection timings were chosen on the basis of pre- liminary experiments performed at 30 minutes, 1 and 4 hours, which showed that cell administration after 30 minutes did not improve survival.

As paracrine mechanisms of adult stem cells were related to their secreted factors, we investigated and compared the effects of HLSC-CM and MSC-CM to that of HLSCs. As shown in Fig. 1C, the intraperito- neal injection of HLSC-CM induced an 80% survival. Conversely, MSC-CM did not protect mice from FLF.

HEPATOLOGY, Vol. 57, No. 1, 2013 HERRERA SANCHEZ ET AL. 313

Fig. 1. Effect of HLSCs on a SCID model of FLF induced by GalN/LPS. (A) After 30 minutes of injury induction, mice received intraperitoneal injection of vehicle (n 1⁄4 6), intravenous injection of 2 􏰀 106 HLSCs (n 1⁄4 9), intravenous injection of 2 􏰀 106 of MSC (n 1⁄4 6), intraperitoneal injection of 30 􏰀 106 HLSCs (n 1⁄4 9), and intravenous injection of vehicle alone (n 1⁄4 6). (B) Mice also received LP injections of 0.2 􏰀 106 (n 1⁄4 5), 0.5 􏰀 106 (n 1⁄4 5) of HLSCs and vehicle (n 1⁄4 5). (C) Mice were inoculated intraperitoneally with vehicle, 250 mL of concentrated CM derived from HLSCs (n 1⁄4 10) and 250 mL of concentrated CM derived from MSC (n 1⁄4 5). Data were analyzed by a log-rank test: *P < 0.05 MSC intravenously, HLSCs intravenously versus vehicle (A); HLSCs 0.2 􏰀 106 and 0.5 􏰀 106 LP versus vehicle (B); HLSC-CM intraperitoneally, rhHGF versus HLSC-CM plus blocking antibody anti-HGF and vehicle (C).

All controls treated with vehicle alone (PBS) died within 8 hours, suggesting that the effect on survival did not depend on mice hydration (Fig. 1C).

HLSCs Inhibited Liver Necrosis, Apoptosis and, Enhanced Liver Regeneration. Histological analysis of mice treated with GalN/LPS showed extensive necrosis (Fig. 2A) and apoptosis (Fig. 3A) of the liver after 7 hours, along with a significant increase in liver enzyme levels measured in the peripheral blood (Fig. 2B,C). Treatment with intravenous injections of HLSCs led to a significant reduction of apoptosis and necrosis in surviving mice (Fig. 2A), despite the increase in liver enzymes at 7 hours. In mice surviving to injury, a sig- nificant decrease of liver enzymes was observed 3 days after HLSC injection, subsequently reaching normal values (Fig. 2B,C). A significantly lower concentration of ammonium was detected in serum of FLF mice (GalN/LPS) treated with HLSCs compared to vehicle alone (Fig. 2D). In HLSC-treated mice, normal liver morphology was reestablished after 7 to 21 days of treatment (not shown).

As shown in Fig. 3, HLSCs significantly inhibited liver apoptosis in FLF mice (GalN/LPS) compared to vehicle alone. Proliferating cell nuclear antigen (PCNA)-positive cells detected at 7 hours expressed human leukocyte antigen (HLA) or CFSE, indicating that they were derived from the injected HLSCs (Fig. 3C). However, after 3 days in mice treated with HLSCs, PCNA-positive cells were mainly negative for

HLA and CFSE, indicating that most proliferating cells were of murine origin (Fig. 3C).

Intravenous- and LP-Injected HLSCs Localized in the Liver of FLF Mice. Liver cell localization was evaluated by IVIS using DiD-labeled HLSCs.12,13 As shown in Fig. 4A,B, after intravenous injection HLSCs preferentially accumulated in livers of mice with FLF but not in livers of healthy mice (Fig. 4A,B). Fluorescence signals, expressed as average radi- ance, increased until day 7 in livers of mice with FLF but not in those of healthy mice. In LP-injected mice, no difference in liver accumulation of HLSCs between FLF and healthy mice was observed (Sup- porting Fig. 1S,B).

By immunohistochemistry, CFSE-labeled HLSCs were mainly detected in large liver vessel walls after 24 hours and within the liver parenchyma after days 7 and 21 in intravenously injected surviving FLF mice (Fig. 5A,D,E). In these mice, CFSE-labeled HLSCs were transiently detected after 24 hours in lungs and spleens, decreasing thereafter (Fig. 5B,C; Fig. 2S). When HLSCs were intravenously injected in healthy mice treated with vehicle alone, liver accumu- lation was significantly lower than in FLF mice (Fig. 5A). In LP-injected mice, CFSE-labeled HLSCs were detected in the liver parenchyma at days 7 and 21 following injection (Fig. 1S,B), but there was no accumulation in lungs or spleens at any timepoint (not shown).

314 HERRERA SANCHEZ ET AL. HEPATOLOGY, January 2013

Fig. 2. Effects of HLSCs on hepatic histomorphology and function in FLF mice. Representative light microscopy micrographs of liver histology at 7 hours and day 3 after induction of FLF in SCID mice treated or not with intravenous injection of 2 􏰀 106 HLSCs. (A) H&E staining showing extensive cellular loss due to apoptosis and necrosis in FLF mice (GalN/LPS) at 7 hours with respect to control (ctrl). Cellular loss was reduced in FLF mice treated with HLSCs at 7 hours and especially after day 3. (B) Aspartate aminotransferase and alanine aminotransferase (C) were measured as bio- markers of liver cell injury in serum of control SCID mice (ctrl; black bar), in mice after 7 hours of FLF induction (gray bar) and in FLF mice treated with intravenous injection of 2 􏰀 106 HLSCs and sacrificed 7 hours and day 3 after injury induction (white bar), and expressed as U/L. Data are expressed as mean 6 standard deviation (SD) relative quantity of six different SCID per group. ANOVA with Newman-Keuls multicomparison test was performed; *P < 0.05 GalN/LPS after 7 hours versus ctrl; #P < 0.05 GalN/LPS injured SCID mice after 7 hours and day 3 of HLSC treatment versus GalN/LPS injured SCID mice. (D) Ammonium was measured in serum of vehicle-treated SCID mice (ctrl; white bar), in FLF mice after 7 hours (black bar), and in FLF mice treated with intravenous injection of 2 􏰀 106 HLSCs and sacrificed at 7 hours (gray bar). Data are expressed as mean 6 SD relative quantity. ANOVA with Newman-Keuls multicomparison test was performed; *P < 0.05 GalN/LPS after 7 hours versus ctrl; #P < 0.05 GalN/LPS injured SCID mice after 7 hours of HLSC treatment versus GalN/LPS injured SCID mice.

In Vivo Expression of Mature Hepatic Markers by HLSCs. To assess whether HLSCs engrafted in the liver expressed mature hepatic markers, we investigated coexpression of human antigen HLA and mature he- patic markers such as cytokeratin 8 and 18 by confocal analysis (Fig. 6). At day 7 the majority of HLA-posi- tive cells expressed cytokeratin 8 and 18 (Fig. 6A,B). At day 21, 􏰂50% of HLA-positive cells expressed cy- tokeratin 8/18 (Fig. 6A,B). FISH analysis with the human pan-centromeric probe showed that at 24 hours positive cells were pan-cytokeratin-negative, whereas at days 7 and 21 they coexpressed pan-cyto- keratin and human chromosome (Fig. 6B). At day 21 several pan-centromeric-positive cells did not express pan-cytokeratin, suggesting that some undifferentiated HLSCs were still present in liver parenchyma (Fig. 6B).

HLSC-CM Inhibited Liver Necrosis and Apoptosis and Enhanced Liver Regeneration. In the GalN/LPS model of FLF, HLSC-CM showed a similar protective activity on liver function and morphology compared to HLSCs (Figs. 1C, 7A,B). The increased presence of PCNA-positive cells demonstrated liver regeneration in

mice treated with HLSC-CM (Fig. 7C) and apoptosis was significantly reduced compared to mice treated with vehicle alone (Fig. 7D).

In vitro studies confirmed that HLSC-CM at doses as low as 29 mg protein/mL protected human hepato- cytes from GalN-induced apoptosis (Fig. 7E). More- over, increasing concentrations of HLSC-CM enhanced proliferation of human hepatocytes (Fig. 7F). Similar results were obtained with murine hepatocytes isolated from both normal and FLF mice (Fig. 3S)

As shown in Table 1 and Fig. 4S, HLSC-CM con- tained several growth factors/cytokines potentially involved in liver protection and regeneration, the most represented of which were interleukin (IL)-6, IL-8, vascular endothelial growth factor (VEGF), hepatocyte growth factor (HGF), and MSP, reconfirmed by single ELISA. The concentration of HGF in HLSC-CM was 􏰂50-fold higher than in MSC-CM. Treatment of HLSC-CM with neutralizing antihuman HGF anti- body abrogated the protective effect of HLSC-CM (Fig. 1C), and the survival rate of mice treated with rhHGF was 􏰂40%, suggesting a relevant role of HGF in the hepatoprotective effect of HLSC-CM (Fig. 1C).

HEPATOLOGY, Vol. 57, No. 1, 2013 HERRERA SANCHEZ ET AL. 315

Fig. 3. Effect of HLSCs on liver cell apoptosis and proliferation in FLF mice. (A) Quantification of apoptotic liver cells by TUNEL per high power field (hpf, 40􏰀) in control mice injected with vehicle (white bars), in GalN/LPS mice after 7 hours (black bar), in GalN/LPS mice treated with in- travenous injection of 2 􏰀 106 HLSCs and sacrificed at 7 hours (gray bar) or day 3 (shaded bar). Data are expressed as mean 6 SD relative quantity of three different mice per group. ANOVA with Newman-Keuls multicomparison test was performed; *P < 0.05 GalN/LPS injured SCID mice after 7 hours versus vehicle; §P < 0.05 GalN/LPS mice treated with intravenous injection HLSCs at 7 hours or 3 days versus GalN/LPS injured SCID mice after 7 hours. (B) Quantification of PCNAþ liver cells per hpf (40􏰀) in control mice injected with vehicle (white bars), GalN/ LPS injured SCID mice after 7 hours (black bar), in GalN/LPS mice treated with intravenous injection of 2 􏰀 106 HLSCs and sacrificed at 7 hours (gray bar) or day 3 (shaded bar). Data are expressed as mean 6 SD relative quantity of three different mice per group. ANOVA with New- man-Keuls multicomparison test was performed; #P < 0.05 GalN/LPS mice treated with intravenous injection HLSCs at 7 hours or day 3 versus GalN/LPS mice after 7 hours. (C) Representative immunofluorescence showing coexpression of HLA and PCNA at 7 hours and 3 days (d). Origi- nal magnification 400􏰀.

We also evaluated the amount of human and murine VEGF, IL-6, and HGF in SCID mice with FLF after intravenous injection with HLSCs. Human HGF and IL-6 levels in mouse serum peaked at 7 hours after HLSC injection (hIL6 1⁄4 596 6 390 pg/mL; hHGF 1⁄4 482 6 370 pg/mL), decreased after 24 hours (hIL6 1⁄4 46 6 19 pg/mL; hHGF 1⁄4 2 6 1.6 pg/mL), and were undetectable after day 7. Human VEGF was undetect- able. In mice injected with HLSCs, murine HGF was not significantly different from untreated controls (respectively 101 6 49 pg/mL versus 107 6 19 pg/ mL). An increase of murine HGF (159 6 106 pg/mL) was observed after 24 hours, decreasing thereafter. Lev- els of murine VEGF and IL-6 did not increase at any time after HLSC administration.

Discussion

The results of our study can be summarized as fol- lows: (1) HLSCs and HLSC-CM strikingly improved survival and reduced plasma levels of liver enzymes and ammonium in mice with FLF, and (2) the protective effect of HLSCs and HLSC-CM was due to reduced apoptosis, necrosis, and enhanced proliferation.

Cell therapy based on hepatocyte transplantation is a potential treatment option in patients with acute liver failure.14 Transplanted hepatocytes may replace liver functions allowing endogenous regeneration. However, the availability of a suitable cellular source is a limiting factor, but stem cells could represent an ideal cell source for liver regeneration. The capacity of

316 HERRERA SANCHEZ ET AL.

HEPATOLOGY, January 2013

Fig. 4. Liver localization of HLSCs by IVIS. (A) Quantification of flu- orescence intensity, calculated in region of interest (ROI), expressed as the mean of average radiance 6 SD of mice injected with DiD-labeled HLSCs after induction of FLF (black bar) or not (white bars). Vehicle represents the intensity of liver of mice treated with PBS instead of HLSCs. Data are expressed as mean 6 SD of two SCID mice injected with HLSCs after induction of FLF and 6 SD of four SCID mice injected with HLSCs without FLF. (B) Representative fluorescence images of mice with or without FLF injury injected with DiD-labeled HLSCs and sacrificed after 5, 24 hours, or day 7.

stem cells to reverse liver injury has been shown in dif- ferent experimental models of liver failure, e.g., in a liver fibrosis model induced by carbon tetrachloride (CCl4), mice treated with MSCs presented less fibrosis, better liver function, and a significant improvement of survival compared to untreated mice.15,16 In addition, MSCs were shown to protect the liver against hepato- cyte apoptosis induced by ischemia-reperfusion injury, and to enhance liver regeneration.9

Recently, Parekkadan et al.6 demonstrated that intra- venous injection of MSC-CM or extracorporeal perfu- sion in a bioreactor containing MSCs had a significant survival benefit in treatment of FLF in rats. Zagoura et al.17 reported that human amniotic fluid-derived MSCs also led to liver repair in a model of CCl4- induced acute hepatic injury. In particular, repair was increased when MSCs were initially differentiated in vitro into hepatic progenitor-like cells. Despite the observation of engraftment of the injected cells, a rele- vant role for secreted molecules was suggested.17

In the present study we showed that HLSCs may have a therapeutic potential in a lethal model of FLF in SCID mice. Enhanced survival and the improved histopathological findings were associated with signifi- cantly lower plasma serum transaminases and

Fig. 5. Detection of HLSCs after in vivo injection. (A-C) Quantification of CFSEþ cells per hpf (40􏰀) in livers, spleens, and lungs of SCID mice injected with 2 􏰀 106 of CFSE-labeled HLSCs after induction of FLF (black bars) or not (white bars). Data are expressed as mean 6 SD of relative quantity of cells in three different mice per group. (D,E) Representative micrographs of livers paraffin section of SCID mice injected with 2 􏰀 106 of CFSE-labeled HLSCs and stained with anti-CFSE antibody (brown staining). HLSCs were detectable within liver large vessels after 24 hours (D) and after day 7 (E) and 21 (not shown) within the liver parenchyma.

HEPATOLOGY, Vol. 57, No. 1, 2013 HERRERA SANCHEZ ET AL. 317

Fig. 6. In vivo differentiation of HLSCs. (A) Number of HLSCs engrafted in liver of GalN/LPS mice treated with intravenous injection of 2 􏰀 106 HLSCs and sacrificed at 24 hours, 7, and 21 days detected as HLAþ cells (black columns), or HLAþCK8/18þ cells (white columns). (B) Representative confocal micrographs showing in the upper panel the intraparenchymal localization of HLSCs by the expression of HLA class I (green) and of cytokeratin 8 and 18 (red) in liver sections of SCID mice after induction of FLF treated with HLSCs or not and sacrificed 24 hours, 7, or 21 days later. Ctrl 1⁄4 isotypic control. Original magnification: 400􏰀. Representative confocal micrographs showing in the lower panel FISH analysis performed with human pan-centromeric probe (red spots) costained with pan-cytokeratin (green fluorescence). Ctrl 1⁄4 FISH analysis on normal murine liver. Nuclei were counterstained with Hoechst dye. Original magnification: 630􏰀.

ammonium levels. HLSCs are liver-resident MSCs that are already committed to a hepatic lineage, thus do not require in vitro differentiation.7 Immunofluores- cence tracing as well as FISH analyses using a human pan-centromeric probe showed some HLSC engraft- ment within the liver. Coexpression of pan-cytokeratin and human centromere indicated that at day 7 the ma- jority of engrafted HLSCs expressed pan-cytokeratin, with a significantly reduced expression at day 21. This suggests the persistence of an undifferentiated, small population of human HLSCs. However, our hypothe- sis is that recovery by HLSCs is attributed to a para- crine mechanism, and not by the substitution of the injured parenchyma. In fact, repopulation of the liver was mainly dependent on proliferation of murine he- patocytes. Crucially, HLSC-CM, containing several cytokines with proproliferative and antiapoptotic prop- erties mimicked the effect of HLSCs. Contrary to other experimental models of liver injury, MSC-CM was totally ineffective in the present model. Compar- ing the composition of HLSC-CM and MSC-CM, HGF was found to be 􏰂50-fold higher in HLSC-CM than in MSC-CM, which prompted us to perform in vivo experiments using rhHGF or HGF blockade, demonstrating that the beneficial effect of HLSC-CM depended, at least partly, on HGF.

These results suggest that in liver regeneration, he- patocyte proliferation is sustained by soluble factors. In this context, Strick-Marchand et al.18 recently showed a beneficial crosstalk between the immune sys-

tem and liver stem cells that operates through the release of cytokines that could promote tissue regenera- tion following acute liver damage. Moreover, Van Poll et al.19 demonstrated a direct antiapoptotic and promi- totic effect of MSC-CM in vitro and Parekkadan et al.6 demonstrated that the MSC-induced improve- ment in survival was attributed to antiinflammatory chemokine release in a rat model of GalN-induced FLF.

In the present study we found that some HLSCs, unlike MSCs, persisted in the liver after days 7 and 21. However, the observation that cell-free HLSC-CM mimicked the HLSC effects suggests a paracrine action by the release of cytokines and growth factors. Interest- ingly, HLSC-CM inhibited in vitro hepatocyte death and stimulated proliferation, and ELISA analysis of the HLSCs-CM showed the presence of several growth factors/cytokines, potentially involved in liver regenera- tion. In particular, HLSC-CM contained liver protec- tive factors, such as IL-10 (an antiinflammatory cyto- kine, recently identified as a mediator of the hepatoprotective action of amniotic fluid-derived he- patic progenitor cells), IL-1ra, MCP-1, and IL-1 beta.6,17 HLSCs also secrete growth factors such as VEGF, which facilitates angiogenesis and is involved in tissue repair,21,22 and IL-8, a chemokine with proan- giogenic activity.23 In HLSC-CM, we also found growth factors with known hepatoprotective properties, such as HGF, IGF-1, and IL-6.24 However, the growth factor of greatest relevance is HGF in HLSC-CM

318 HERRERA SANCHEZ ET AL. HEPATOLOGY, January 2013

Fig. 7. In vivo and in vitro effect of HLSC-CM. Alanine aminotransferase (A) and aspartate aminotransferase (B) were measured in serum of vehicle-treated mice (ctrl; black bar), of GalN/LPS injured mice after 7 hours (gray bar) and GalN/LPS mice after 7 hours treated with intraperito- neal injection of 250 lL of concentrated HLSC-CM and sacrificed 7 hours and 3 days (white bar), and expressed as U/L. Data are expressed as mean 6 SD relative quantity of three mice per group. ANOVA with Newman-Keuls multicomparison test was performed; *P < 0.05 GalN/LPS af- ter 7 hours versus ctrl; #P < 0.05 HLSC-CM injected SCID mice after GalN/LPS injury versus GalN/LPS mice after 7 hours and day 3. (C) Quan- tification of PCNAþ liver cells per hpf (40􏰀) in control mice injected with vehicle (black bar), in GalN/LPS mice after 7 hours (gray bar), and in GalN/LPS mice treated with i.p injection of 250 lL of concentrated HLSC-CM and sacrificed after 7 hours and 3 days (white bars). Data are expressed as mean 6 SD relative quantity of three mice per group. ANOVA with Newman-Keuls multicomparison test was performed; *P < 0.05 HLSC-CM treated GalN/LPS injured SCID mice after 3 days versus GalN/LPS injured SCID mice. (D) Quantification of apoptoticþ liver cells per hpf (40􏰀) in control mice injected with vehicle (black bar), in GalN/LPS mice after 7 hours (gray bar), and in GalN/LPS mice treated with intra- peritoneal injection of 250 lL of concentrated HLSC-CM and sacrificed after 7 hours and 3 days (white bars); Data are expressed as mean 6 SD of relative quantity of cells in three mice per group. ANOVA with Newman-Keuls multicomparison test was performed; *P < 0.05 GalN/LPS injured SCID mice after 7 hours versus vehicle; #P < 0.05 HLSC-CM treated GalN/LPS mice after 7 hours and day 3 versus GalN/LPS mice. (E) In vitro apoptosis was evaluated by TUNEL assay in human hepatocytes incubated with vehicle alone (white bars) in the presence of GalN 5mM (black bar) and in hepatocytes incubated with 29 lg/mL of concentrated HLSC-CM (gray bars). Results are expressed as mean 6 SD of three different experiments performed in triplicate. ANOVA with Newman-Keuls multicomparison test was performed; *P < 0.05 hepatocytes incubated with GalN 5 mM versus vehicle alone; #P < 0.05 hepatocytes incubated with 29 lg/mL concentrated HLSC-CM versus hepatocytes incubated with GalN 5 mM. (F) Proliferation was evaluated by bromodeoxyuridine (BrdU) incorporation in hepatocytes cultured in the presence of vehicle alone (black bar) or with different concentration of concentrated HLSC-CM (29 lg/mL; 0.12-0.46-0.93 mg/mL) (gray bars). Results are expressed as mean 6 SD of three different experiments performed in triplicate. ANOVA with Newman-Keuls multicomparison test was performed; *P < 0.05 hepatocytes incubated with 0.12, 0.46, and 0.93 mg/mL of concentrated HLSC-CM versus hepatocytes incubated with vehicle alone.

(only expressed at low levels in MSC-CM), as blocking HGF significantly prevented the protective effect of HLSC-CM, and stimulation with rhHGF improved survival.

In conclusion, HLSCs and HLSC-CM may improve survival in a lethal mouse model mainly by paracrine mechanisms, and HLSCs may therefore represent a new cell source for FLF treatment.

HLSCs

MSCs

Table 1. Concentration of Cytokines (ng/mL)

2%FCS Cytokines ELISA

VEGF 0
HGF 0
IL-6 0.03460.01 IL-8 0
MCP1 0
MSP1 0

Bioclarma

ELISA

Bioclarma

ELISA

   

119656 119679 15865 15867 52 6 12 52 6 17 0.4660.2 0
38 6 4 37.9 6 6.2 30 6 3 30.4 6 4.9

207615 207620.5 228613 228617.7 5.760.2 5.6860.27 5.560.5 5.4960.66

34.4628 34.4639.6 0 0

HEPATOLOGY, Vol. 57, No. 1, 2013

HERRERA SANCHEZ ET AL. 319

Acknowledgment: We thank Federica Antico for technical assistance.

References

1. Russo FP, Parola M. Stem and progenitor cells in liver regeneration and repair. Cytotherapy. 2011;13:135-144.
2. Kobayashi N, Okitsu T, Tanaka N. Cell choice for bioartificial livers. Keio J Med 2003;52:151-157.
3. Oertel M, Shafritz DA. Stem cells, cell transplantation and liver repo- pulation. Biochim Biophys Acta 2008;1782:61-74.
4. Stock P, Bru ̈ckner S, Ebensing S, Hempel M, Dollinger MM, Christ B. The generation of hepatocytes from mesenchymal stem cells and engraftment into murine liver. Nat Protoc 2010;5:617-627.
5. Piryaei A, Valojerdi MR, Shahsavani M, Baharvand H. Differentiation of bone marrow-derived mesenchymal stem cells into hepatocyte-like cells on nanofibers and their transplantation into a carbon tetrachlor- ide-induced liver fibrosis model. Stem Cell Rev 2011;7:103-118.
6. Parekkadan B, van Poll D, Suganuma K, Carter EA, Berthiaume F, Tilles AW, et al. Mesenchymal stem cell-derived molecules reverse ful- minant hepatic failure. PLoS One 2007;2:e941.
7. Herrera MB, Bruno S, Buttiglieri S, Tetta C, Gatti S, Deregibus MC, et al. Isolation and characterization of a stem cell population from adult human liver. Stem Cells 2006;24:2840-2850.
8. Herrera MB, Fonsato V, Gatti S, Deregibus MC, Sordi A, Cantarella D, et al. Human liver stem cell-derived microvesicles accelerate hepatic regeneration in hepatectomized rats. J Cell Mol Med 2010;14: 1605-1618.
9. Kanazawa H, Fujimoto Y, Teratani T, Iwasaki J, Kasahara N, Tsur- uyama T, et al. Bone marrow-derived mesenchymal stem cells amelio- rate hepatic ischemia reperfusion injury in a rat model. PLoS One 2011;296:e19195.
10. Lehmann V, Freudenberg MA, Galanos C. Lethal toxicity of lipopoly- saccharide and tumor necrosis factor in normal and D-galactosamine- treated mice. J Exp Med 1987;165:657-663.
11. Herrera MB, Bussolanti B, Bruno S, Morando L, Mauriello-Romanazzi G, Sanavio F, et al. Exogenous mesenchymal stem cells localize to the kidney by means of CD44 following acute tubular injury. Kidney Int 2007;72:430-441.

12. 13.

14. 15. 16. 17.

18. 19.

20. 21. 22.

23. 24.

Boddington SE, Sutton EJ, Henning TD, Nedopil AJ, Sennino B, Kim A, et al. Labeling human mesenchymal stem cells with fluorescent con- trast agents: the biological impact. Mol Imaging Biol 2011;13:3-9. Sutton EJ, Boddington SE, Nedopil AJ, Henning TD, Demos SG, Baehner R, et al. An optical imaging method to monitor stem cell migration in a model of immune-mediated arthritis. Opt Express 2009; 17:24403-24413.

Dhawan A, Puppi J, Hughes RD, Mitry RR. Human hepatocyte trans- plantation: current experience and future challenges. Nat Rev Gastroen- terol Hepatol 2010;7:288-298.
Cho KA, Lim GW, Joo SY, Woo SY, Seoh JY, Cho SJ, et al. Transplan- tation of bone marrow cells reduces CCl4-induced liver fibrosis in mice. Liver Int 2011;31:932-939.

Sakaida I, Terai S, Yamamoto N, Aoyama K, Ishikawa T, Nishina H, et al. Transplantation of bone marrow cells reduces CCl4-induced liver fibrosis in mice. HEPATOLOGY 2004;40:1304-1311.
Zagoura DS, Roubelakis MG, Bitsika V, Trohatou O, Pappa KI, Kape- louzou A, et al. Therapeutic potential of a distinct population of human amniotic fluid mesenchymal stem cells and their secreted mole- cules in mice with acute hepatic failure. Gut 2012;61:894-906. Strick-Marchand H, Masse GX, Weiss MC, Di Santo JP. Lymphocytes support oval cell-dependent liver regeneration. J Immunol 2008;18: 2764-2771.

Van Poll D, Parekkadan B, Cho CH, Berthiaume F, Nahmias Y, Tilles AW, et al. Mesenchymal stem cell-derived molecules directly modulate hepatocellular death and regeneration in vitro and in vivo. HEPATOLOGY 2008;47:1634-1643.

Bao P, Kodra A, Tomic-Canic M, Golinko MS, Ehrlich HP, Brem H. The role of vascular endothelial growth factor in wound healing. J Surg Res 2009;153:347-358.
T€ogel F, Zhang P, Hu Z, Westenfelder C. VEGF is a mediator of the renoprotective effects of multipotent marrow stromal cells in acute kid- ney injury. J Cell Mol Med 2009;13:2109-2114.

Martin D, Galisteo R, Gutkind JS. CXCL8/IL8 stimulates vascular endo- thelial growth factor (VEGF) expression and the autocrine activation of VEGFR2 in endothelial cells by activating NFkappaB through the CBM (Carma3/Bcl10/Malt1) complex. J Biol Chem 2009;284:6038-6042. Hoek JB, Pastorino JG. Ethanol, oxidative stress, and cytokine-induced liver cell injury. Alcohol 2002;27:63-68.

Li WC, Ralphs KL, Tosh D. Isolation and culture of adult mouse he- patocytes. Methods Mol Biol 2010;633:185-196.