Multipotent mesenchymal stromal (MS) cells from adult bone marrow are a cell population that can be expanded to large numbers in culture. MS cells might be differentiated toward hepatocytes in vitro and thus are promising candidates for therapeutic applications in vivo. The efficacy of bone marrow-derived MS cells versus hepatocytes to contribute to liver regeneration was compared in a rat model of prolonged toxic hepatic injury. Liver damage was induced by injection of carbon tetrachloride (CCl4) or allyl alcohol (AA) with and without retrorsine (R) pretreatment. MS cells or hepatocytes of wild-type F344 rats were injected into dipeptidyl peptidase IV (DPPIV)-deficient syngeneic rats. Hepatocyte chimerism was higher after intraportal hepatocyte transplantation in the R/AA group (mean maximal cluster size [MCS] = 21 cells) compared with the R/CCl4 treatment group (MCS = 18). No hepatocyte engraftment was outlined following post-transplant CCl4 injection only, whereas mere AA injection resulted in small clusters of donor-derived hepatocytes (MCS = 2). Intraparenchymal injection of hepatocytes was associated with a MCS = 11 after R/AA treatment and a MCS = 6 after AA administration alone. Redistribution of MS cells to the liver was shown after intraportal and intraparenchymal injection. In contrast to hepatocyte transplantation, however, donor-derived DPPIV-positive cells could not be demonstrated in any recipient after MS cell transplantation. Data from the present study indicate that a well-defined population of MS cells obtained according to established standard protocols does not differentiate into hepatocytes in vivo when transplanted under regenerative conditions, in which the application of hepatocytes results in stable hepatic engraftment.
Committed stem cells from one adult tissue can give rise to differentiated cells of other tissues [1, , –4]. This phenomenon has been described as the plasticity of stem cells and is considered to play a key role in the development of novel stem cell therapies applicable in end stage organ failure.
The liver is especially interesting among the possible target organs for stem cell applications, since alternative therapies are urgently needed in the case of terminal hepatic dysfunction due to hepatitis, toxins, or metabolic disorders. At present, liver transplantation is very often the only therapeutic option for patients with deteriorating liver function. However, the continuing shortage of donor organs and the intrinsic risks associated with whole organ transplantation mark the clinical limits for this procedure. In addition, pharmacological immunosuppression required after liver transplantation represents a lifelong risk of side effects for the recipient.
Multipotent mesenchymal stromal (MS) cells are an intriguing stem cell population for the application in liver-directed cell transplantation, since they are readily obtained from bone marrow aspirates and can be expanded in culture, achieving a large number of potent cells . In contrast to MS cells, other stem cell populations with a more embryonic-like phenotype, such as multipotent adult progenitor cells , have been controversially discussed . Classic MS cells are well defined by function and phenotype [5, 8]. They can be obtained by plastic adherence and be cultivated in basic media supplemented with serum for several (but not indefinite) passages. These MS cells readily differentiate into cells of the mesoderm, such as adipocytes, chondrocytes, and osteoblasts, in vitro. Moreover, cultured MS cells can differentiate into hepatocytes when subjected to a combination treatment with hepaocyte growth factor (HGF) and oncostatin M . Furthermore, it was recently reported that human MS cells can give rise to hepatocyte-like cells in a xenogenic rat liver injury model in vivo . Thus, we wanted to compare the ability of syngeneic MS cells and hepatocytes to contribute to liver regeneration in an in vivo transplant model suitable to identify donor cell-derived hepatocytes independent from the underlying mechanism. Possible mechanisms of hepatic cell plasticity are direct differentiation, cellular fusion [11, 12], or repopulation with a diverse stem cell pool [3, 13, 14].
MS cells or hepatocytes from F344 rats were injected into syngeneic dipeptidyl peptidase IV (DPPIV)-deficient rats subjected to toxic liver damage by CCl4 or AA. In addition, some animals were pretreated with retrorsine, an alkaloid that effectively induces a cell cycle arrest in hepatocytes and thereby inhibits hepatocyte mitosis. DPPIV was used as a marker system, whereby the de novo expression of DPPIV on recipient liver cells indicates MS cell to hepatocyte plasticity in this model. In summary, we show that syngeneic MS cells do not contribute to liver regeneration in our rat liver injury model applying various proliferation stimuli, whereas hepatocytes effectively engraft and proliferate under similar conditions.
Materials and Methods
Isolation and Culture of Multipotent Mesenchymal Stromal Cells
MS cells were isolated according to Pittenger et al.  with slight modifications. Bone marrow cells were collected by flushing the long bones with Hanks' balanced salt solution (HBSS) medium (Biochrom AG, Berlin, http://www.biochrom.de). Cells were then plated in 175-cm2 flasks and cultured in expansion medium, that is, high-glucose Dulbecco's modified Eagle's medium (DMEM) (with GlutaMAX, without pyruvate; Biochrom) supplemented with 10% fetal bovine serum (FBS) (Biochrom), 1% penicillin/streptomycin, and 1% glutamine. After 24, 48, and 72 hours, nonadherent cells were removed by changing the culture medium. Adherent cells were trypsinized (0.5% trypsin-EDTA), harvested, and replated into new flasks 2 weeks after starting the culture, or each time when cell confluence reached 60%–80%.
Differentiation of MS cells into adipocytes was achieved by plating MS cells into 24-well plates in expansion medium without FBS for 2 days after confluence had been reached in the expansion culture. Differentiation was induced by culturing MS cells in expansion medium without FBS supplemented with insulin (15 U/ml; sanofi-aventis, Paris, http://en.sanofi-aventis.com), dexamethasone (10−6 M; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), goat serum (5 ml/100 ml; PromoCell, Heidelberg, Germany, http://www.promocell.com) and 3-isobutyl-1-methylxanthin (0.1 mg/ml; Sigma-Aldrich) for 3 days. For maturation, cells were then cultured in expansion medium without FBS supplemented with insulin (15 U/ml; Aventis) for 5 days. Cell differentiation into adipocytes was confirmed by oil red O staining. Cells were washed in cold phosphate-buffered saline (PBS), fixed with 10% formaldehyde at 4°C overnight, and then incubated with 5 mg/ml oil red O solution for 2 hours at room temperature.
To induce chondrogenic differentiation, 200,000 MS cells were centrifuged in a 15-ml polypropylene tube (1,200 rpm, 5 minutes, 4°C) to form a pellet. Then, cells were treated for 16 days with chondrogenic medium consisting of high-glucose DMEM supplemented with 0.1 μM dexamethasone, 0.2 μM ascorbic acid, 0.2 μM sodium pyruvate, 10 ng/ml transforming growth factor-β1, and 1% ITS (insulin, transferrin, selenium) premix (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com). After 16 days pellets, were embedded in Tissue-tec and snap frozen. Chondrogenic differentiation was assessed by Alcian Blue staining (Bio-Optica, Milano, Italy, http://www.bio-optica.com) according to the manufacturer's instructions.
To induce osteogenic differentiation, MS cells were seeded in six-well plates at a density of 15.000 cells per cm2 in DMEM without FBS for 1 day. Cells were then treated with osteogenic medium for 2 weeks, changing the medium twice a week. Osteogenic medium consisted of DMEM supplemented with 0.1 μM dexamethasone, 0.3 mM ascorbic acid, and 10 mM β-glycerol phosphate. Osteogenic differentiation was assessed by von Kossa staining. Cells were covered with 5% silver nitrate solution for 40 minutes in bright light followed by an incubation step in UV light for 2 minutes. After rinsing with distilled water, cells were incubated for 5 minutes in 1% pyrogallol and rinsed again. Leftover silver nitrate was removed by washing the cells in 5% sodium thiosulfate for 5 minutes.
Purification of Hepatocytes
To obtain single-cell suspensions of hepatocytes, the inferior vena cava of each sacrificed animal was cannulated with a 26-gauge needle connected to a pump. Livers were perfused in retrograde (four-step protocol) with (a) 25 ml of HBSS medium (Gibco, Grand Island, NY, http://www.invitrogen.com) at 37°C, with (b) 25 ml of HBSS medium containing 0.5 mmol EDTA, with (c) another 25 ml of plain HBSS medium, and with (d) the same amount of HBSS medium containing 5 mmol of calcium hydrochloride and 0.5 mg/ml hepatocyte purification-grade collagenase (Sigma-Aldrich). The resulting cell suspension was centrifuged at 500 rpm/50g for 3 minutes at 4°C and washed in RPMI three times and in PBS once. Washes were followed by gradient centrifugation with 35% isotonic Percoll (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com) at 2,000 rpm for 10 minutes at 20°C to select for viable single cells.
Animals and Experimental Groups
F344 rats (MHC haplotype: RT1lllv1; Charles River, Germany) weighing approximately 200 g were maintained in the animal center at the University of Regensburg. A colony of DPPIV-deficient F344 rats was established with breeders initially obtained from Brown University (courtesy of Douglas C. Hixson). For cell transplantation, 2 million MS cells or hepatocytes were injected into the portal vein with a 27-gauge needle. To apply cells directly into the liver, the right liver lobe was surgically exposed, and 2 million cells were injected into the parenchyma. A suture loop was inserted near the injection site to ease later tracking of the infused cells. All animal procedures were approved by regional authorities and conducted under appropriate anesthesia.
Animals were divided into two treatment regimens as shown in Figure 1. The first group received doses of 0.62 mmol/kg of body weight (BW) AA 1 day prior to cell transplantation. On the next day, 2 × 106 F344 hepatocytes or MS cells were injected either directly into the liver or into the portal vein. Beginning with day 2, animals received 0.31 mmol/kg of BW AA every 3rd day until they were sacrificed. Before transplantation, some animals received, in addition, 30 mg/kg of BW of retrorsine 2 and 4 weeks before cell infusion. Animals were sacrificed on days 14, 30, and 60 in the MS cell group and on day 30 in the hepatocyte group. The second group was treated with CCl4 as a liver toxin and pretreated with 30 mg/kg of BW retrorsine twice on day −28 and day −14. Hepatocytes or MS cells were injected into the portal vein on day 0. On days 14 and 28, animals received 0.6 ml/kg of BW CCl4 dissolved in corn oil at a 1:1 ratio.
Flow Cytometry and (Immuno-) Histochemistry
For flow cytometry analysis, passage 5 MS cells were incubated with monoclonal antibodies at 4°C for 30 minutes with combinations of saturating amounts of purified, fluorescein isothiocyanate-conjugated or phycoerythrin-conjugated monoclonal antibodies against CD4, CD25 (Caltag, Burlingame, CA, http://www.caltag.com), CD45, CD49b, CD73, CD80, CD86, Thy1 (CD90), pan anti-rat MHC class II (RT1B), and CD26 (all from Becton Dickinson). Anti-rat RT1.Al was originally purified and kindly provided by K. Wonigeit (Hannover Medical School, Hannover, Germany). For quantification, a FACSCalibur laser flow cytometer (Becton Dickinson) was used, and living (propidium iodide-negative) cells were analyzed with WinMDI software.
For DPPIV histochemistry cryostat sections were fixed in 95% ethanol for 5 minutes. Air-dried slides were then incubated in a substrate made of 100 mmol/l Tris-maleate (pH 6.5), 100 mmol/l NaCl, 1 mg/ml Fast Blue BB salt (Sigma-Aldrich), and 0.5 mg/ml Gly-Pro-methoxy-β-naphthylamide (Sigma-Aldrich) for 20–30 minutes at room temperature. Slides were then washed with PBS, counterstained with hematoxylin, and mounted with DPX (1,3-diethyl-8-phenylxanthine; Sigma-Aldrich).
Immunostaining was performed as previously described  using commercially available antibodies against CD26 (BD Pharmingen, San Diego, http://www.bdbiosciences.com/pharmingen) and CK18 (Abcam, Cambridge, U.K., http://www.abcam.com) in combination with appropriate peroxidase-conjugated secondary antibodies (BD Pharmingen or Roche Diagnostics [Basel, Switzerland, http://www.roche-applied-science.com]). Goat and rat sera were used to decrease nonspecific staining. 3,3′-Diaminobenzidine substrate (Sigma-Aldrich) was taken for color development. Staining was performed on 5-μm-thick cryostat sections (CD26) and on cytospins (CK18). Slides were counterstained with hematoxylin and mounted with DPX (Sigma-Aldrich). To analyze CD26-positive donor-derived cells, four slides of each animal were inspected. The diameter of the largest accumulation of CD26-positive cells was documented.
Isolation and Phenotypic Characterization of Rat MS Cells and Hepatocytes
Rat MS cells from adult bone marrow were purified by adherence to plastic culture flasks, as previously described . After expansion, adherent MS cells had a spindle-shaped fibroblastic morphology and showed colony-forming unit fibroblast activity (Fig. 2A). At least 17 passages could be obtained from each primary population. Culturing in adipocyte-differentiation media induced differentiation of MS cells into adipocytes, as verified by oil red O staining of intracellular fat deposits (Fig. 2B). MS cells were also differentiated into chondrocytes and osteoblasts, as confirmed by Alcian Blue and von Kossa staining, respectively (Fig. 2C, 2D). Phenotypic analysis of F344 (RT1lllv1) MS cells was performed by flow cytometry and immunohistochemistry on cytospins. MS cells expressed MHC class I molecules (RT1Al) and CD80 (dim expression), but they did not express MHC class II molecules (RT1B, nonpolymorphic determinant) or the costimulatory molecule CD86. MS cells were positive for Thy1 (CD90) and CD73, but they lacked the expression of CD26, CK18, CD4, CD25, CD45, and CD49b.
Hepatocytes of F344 rats were purified, and single-cell suspensions were analyzed by flow cytometry. Hepatocytes expressed MHC class I molecules (RT1Al) but did not express MHC class II molecules or costimulatory molecules (CD80 and CD86). Hepatocytes were positive for CD73 and negative for Thy1 (CD90), CD3, CD4, CD25, and CD45. In contrast to MS cells, hepatocytes clearly expressed CD26 and CK18. Fluorescence-activated cell sorting analysis was confirmed by immunohistochemistry of frozen sections of liver tissue and single-cell suspensions (cytospins).
Distribution of MS Cells After Intraportal Transplantation
To localize MS cells in the liver and to analyze their distribution over time, cultivated MS cells were labeled with 5-bromo-2-deoxyuridine (BrdU) 24 hours before injecting 2 × 106 MS cells into the portal vein. Three animals each were analyzed 2 hours, 2 days, and 7 days after cell transplantation. Transplanted MS cells were detected by immunohistochemistry, and all BrdU-positive cells of six different slides were counted per animal. Two hours after injecting MS cells, a median of 173 cells per slide was detected. Single cells were mostly located in the periportal area, although aggregations of cells were also found. After 2 days, only 7.5 cells per slide could still be detected within the liver. This declined further to 6.2 cells per slide after 7 days (Fig. 3). MS cells located within the vascular system were not assessed, since it was assumed that these cells had not integrated into the parenchyma.
In a different set of experiments, green fluorescent protein (GFP)-expressing MS cells (kindly provided by C. Lange, Hamburg, Germany) were injected into the portal vein of syngeneic rats. Two animals were analyzed by fluorescence microscopy 2 hours, 2 days, and 7 days after transplantation. Five slides per animal were analyzed after 2 hours, and a mean value of 339 cells was counted per slide. After 2 days, a mean of 1.4 cells per slide could still be identified, whereas no GFP-positive cells were detected after 7 days (Fig. 3). The liver exhibits strong autofluorescence, presumably being responsible for the different cell counts in the 2-hour analysis. The absence of GFP-expressing cells after 7 days is most likely due to the immunogenicity of GFP . Although some BrdU-positive MS cells were found after 30 days of applying AA in one animal (data not shown), we cannot rule out the possibility that diminished GFP expression or fading presence of BrdU is responsible for detecting only a few cells after 7 days.
Hepatocytes Engraft and Proliferate After Transplantation in Distinct Models of Liver Injury
To compare the dynamics of hepatocyte engraftment with that of MS cells, hepatocytes were transplanted in the DPPIV−/− rat model subjecting recipients to several toxic stimuli. CD26-positive hepatocytes were injected into the portal vein of retrorsine-pretreated recipients with no further toxic damage (n = 6). After 30 days, hepatocytes had engrafted and proliferated throughout the host liver in a uniform pattern of distribution. Here, mostly small clusters of cells, two to four cells in diameter, were detected. Next, we analyzed whether toxic liver damage increased the engraftment rate of transplanted hepatocytes. When recipients were subjected to CCl4 after retrorsine pretreatment, DPPIV-positive donor-derived cell clusters with diameters ranging from 3 to 18 cells were detected throughout the whole host liver (n = 4). However, when liver damage was induced by the injection of CCl4 only, no hepatocyte engraftment was observed (n = 3).
Donor-derived hepatocytes were found even without retrorsine pretreatment when using AA as a toxic stimulus (Fig. 4A; Table 1). Here, mainly cell doublets were detected after 30 days (n = 2) and 60 days (n = 1). When combining the application of retrorsine and AA, the clusters were significantly larger, with diameters ranging from 4 to 21 cells (n = 3), compared with AA treatment only (intraportal injection; Fig. 4B). When donor hepatocytes were injected directly into the liver, without retrorsine pretreatment, small clusters with diameters up to six cells were found (n = 3). Interestingly, no cell clusters were identified directly at the injection site, but donor cells were evenly distributed throughout the whole liver (Fig. 4C). Thus, results did not depend upon the site of injection. When combining injection of AA with retrorsine pretreatment and intraparenchymal injection, the clusters were again larger (Fig. 4D). Here, cell doublets and cell clusters with diameters ranging from 6 to 11 cells were found after 30 days (n = 3). One animal of this group showed cell clusters near the site of injection only.
Table Table 1.. Summary of animal groups
In summary, we show that hepatocytes engraft within the livers of syngeneic recipients when host livers are damaged by retrorsine and multiple injections of CCl4 or allyl alcohol. Application of AA without retrorsine pretreatment results in robust hepatocyte engraftment, whereas retrorsine is necessary for hepatocyte engraftment when liver damage is induced by CCl4. Hepatocyte cell clusters are evenly distributed throughout the liver after cell transplantation.
No Evidence of MS Cell Engraftment in Various Liver Injury Models
After having established a robust model of hepatocyte transplantation, engraftment, and proliferation, we applied cultured MS cells to this model. MS cells were transplanted under the same regenerative conditions that led to stable hepatocyte engraftment. When wild-type MS cells were injected into the portal vein of syngeneic retrorsine-pretreated DPPIV−/− recipient rats without further toxic liver damage, no donor-derived cells were found after 60 days (n = 6). Using AA to damage the liver without the additional application of retrorsine, donor-derived cells were not found after 30 days (n = 7) or 60 days (n = 3). Retrorsine pretreatment could not change this finding after 30 days (n = 3). When MS cells were injected into the portal vein of recipients treated with both CCl4 and retrorsine, again no DPPIV-positive donor-derived cells were found after 60 days (n = 5). After injecting MS cells directly into the right liver lobe, no donor-derived CD26-positive cells were found using AA to damage the liver. Here, animals were analyzed after 14 days (n = 2), 30 days (n = 9), and 60 days (n = 3). To screen for MS cell-derived hepatocytes in these experiments with the highest possible rigor, we analyzed more than 50 slides from different liver lobes for each recipient using both immunohistochemistry for CD26 (DPPIV) and histochemistry for DPPIV.
In summary, we show that transplanted MS cells do not differentiate into DPPIV-positive hepatic cells under conditions in which transplanted hepatocytes engraft and proliferate.
End-stage liver disease can only be treated definitively by liver transplantation. However, the ongoing shortage of donor organs and the serious side effects of lifelong immunosuppression set the clinical limits for whole organ transplantation. Therefore, considerable efforts have been made during recent years to transplant hepatocyte suspensions only, particularly in patients with metabolic diseases (or liver failure in absence of cirrhosis). Some of these therapeutic approaches have had moderate or short-term success [18, 19]. Nevertheless, hepatocyte transplantation did not evolve as an appropriate alternative to liver transplantation, mainly since high-quality hepatocytes require collection from the same donor livers that might as well be transplanted. In addition, the in vitro culture of hepatocytes for transplantation is demanding and inefficient. Stem cells are therefore considered to be a valuable alternative for liver-directed cell therapies, since it has been reported that various kinds of stem cells were able to differentiate into functioning cells of a variety of mature tissues [1, 20], including hepatocytes [21, 22]. Multipotent MS cells, which originate from the adult bone marrow, have recently gained increasing attention in this respect. MS cells are easily harvested from bone marrow aspirates and can be expanded to large numbers in vitro—two prerequisites for their use in regenerative therapies. The differentiation potential of MS cells remains stable during long-term culture , and they permanently distribute to a variety of different tissues . By definition, MS cells give rise to bone, cartilage, and fatty tissue [5, 24]. Moreover, it was reported that human and rat MS cells can differentiate into hepatocyte-like cells in vitro [6, 9, 25, 26], indicating a possible transition to functioning liver cells. It remains a controversial question, however, whether MS cells can be differentiated into functioning hepatocytes in vitro, since hepatic differentiation succeeds mostly when using bone marrow-derived cells cultured with additional growth factors under nonstandard conditions  or when using cells passaged over a long period of time .
In the present study, the therapeutic potential of MS cells was compared with that of differentiated hepatocytes in a rat model of liver regeneration upon injury. By in vitro differentiation into adipocytes, chondrocytes, and osteoblasts and by demonstrating a MS cell-specific set of surface molecules, it was demonstrated that the population of MS cells used in the present experiments met the criteria defined by the International Society for Cellular Therapy [8, 27]. A genetic marker model was established making use of DPPIV-deficient rats, which, as a result of a natural hypermutation, do not express DPPIV (CD26) on any cell [28, –30]. Phenotypical analysis revealed that wild-type MS cells did not express CD26, whereas normal wild-type hepatocytes were ubiquitously positive for this marker. Application of carbon tetrachloride (CCl4) or allyl alcohol (AA) was used to damage recipient livers and thereby to induce liver regeneration. Transplanted hepatocytes engrafted in recipient livers and proliferated when animals were treated with the potent mito-inhibitor retrorsine prior to cell transplantation, as has previously been shown by Laconi et al [28, Clusters were larger when allyl alcohol–30]. Applying retrorsine, no further liver damage was necessary for hepatocyte engraftment . When CCl4 was used to damage the liver, transplanted hepatocytes did not engraft without retrorsine pretreatment. In contrast, hepatocytes demonstrated robust engraftment and proliferation without retrorsine when AA was used, suggesting that AA has a different mechanism of liver destruction compared with CCl4 in this setting. There was no obvious difference when hepatocytes were injected directly into the liver compared with intraportal application of the same cells. Presumably, hepatocytes injected directly into the liver parenchyma still penetrate the liver sinusoids and distribute throughout the whole organ. Consequentially, no accumulation of donor cells was found nearby the injection site after 30 and 60 days, indicating that injected hepatocytes did not form a linked cell matrix with established cell-cell contacts.
MS cells were defined by their phenotype and their ability to differentiate into mature mesenchymal tissue prior to application to the present model. Wild-type MS cells, not expressing DPPIV, were injected into the portal vein of DPPIV-deficient rats. Only those MS cells that changed their phenotype to a more hepatocyte-like pattern would express DPPIV and could subsequently be identified. De novo expression of DPPIV was reported when multipotent adult progenitor cells were differentiated into hepatocytes in vitro . However, no DPPIV-positive cells were found after examining more than 500 slides of 38 animals. This clearly emphasizes that transplanted syngeneic MS cells did not differentiate into hepatocyte-like cells under conditions that led to stable engraftment and proliferation of transplanted hepatocytes, a finding that is in accordance with other reports questioning the pluripotency of MS cells [31, 32]. Although hepatocytes constantly expressed DPPIV and naïve MS cells did not, we cannot formally rule out the possibility that DPPIV is a late marker of hepatocyte development, and very few MS cells differentiated into hepatocyte-like cells not expressing DPPIV during the observation period. The distribution of MS cells throughout the liver after injection into the portal vein was analyzed by two different methods: BrdU labeling and GFP transduction. Two hours after the injection, both staining methods revealed the persistence of injected MS cells within the liver parenchyma. Two and 7 days after injection, however, the number of MS cells remaining in the liver was significantly reduced, indicating that the majority of MS cells were cleared from the liver early. Thus, it is possible that MS cells cannot enter the liver parenchyma as they lack important adhesion molecules. It may be that differentiation of MS cells into hepatocyte-like cells in vitro before injection could overcome this problem. A respective approach has been taken by Oyagi et al. , who describe contribution of HGF-treated MS cells to liver regeneration. These authors were able to outline that MS cell transplantation ameliorates liver function by showing reduction of liver enzyme levels and improvement of liver fibrosis. However, Oyagi et al.  did not demonstrate differentiated MS cells within recipient livers, implying that a bystander effect could have been responsible for the therapeutic effect of MS cells in their model.
Since the majority of MS cells were cleared within the first 2 days, we further analyzed whether direct injection of MS cells into the liver avoided their dispersion. However, again, no CD26-positive donor-derived cells were found. Sato et al. reported that MS cells differentiated into hepatocyte-like cells when injected directly into the liver, applying a model comparable to ours. In contrast to the present model, Sato et al.  used a xenogenic setting, injecting human MS cells into rats that were immunosuppressed by cyclosporine. It is therefore possible that human MS cells transplanted into rats have a different pattern of adhesion and distribution compared with syngeneic MS cells, especially when applied together with immunosuppressive drugs. Direct injection of cells into the liver in our model resulted in even spreading of MS cells throughout the liver, which is in contrast to the findings of Sato et al. , who reported that transplanted MS cells persisted as a dense cluster of cells after direct injection into the liver.
In the present study, we compared the ability of MS cells and hepatocytes to contribute to liver regeneration in different models of artificial liver injury. We could demonstrate that hepatocytes readily engraft and proliferate in recipient livers after transplantation. However, MS cell engraftment could not be observed under similar conditions. This indicates that differentiation of MS cells into hepatocytes does not occur during liver regeneration—at least not to an extent that is of potential clinical benefit. In contrast to MS cells (which can be easily obtained and be propagated in culture), hepatocytes (which cannot be easily obtained and cannot be cultivated long-term) turned out to be the more effective cell population for liver-directed cell therapy. Overall, the clinical implementation of liver stem cell therapy remains promising  but ambiguous so far.
The authors indicate no potential conflicts of interest.
We thank Irina Kucuk for excellent technical assistance.