Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis

Article

Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis Graphical Abstract

Authors Guangqi Song, Martin Pacher, Asha Balakrishnan, ..., Tobias Cantz, Michael Ott, Amar Deep Sharma

Correspondence [email protected] (M.O.), [email protected] (A.D.S.)

In Brief Sharma, Ott, and colleagues show that expression of a key set of four transcription factors can reprogram hepatic myofibroblasts to induced hepatocyte-like cells in vivo and reduce liver fibrosis, suggesting that direct in vivo reprogramming may be an effective treatment approach for chronic liver disease.

Highlights d

Transcription factor induction converts hepatic myofibroblasts to iHeps in vitro

d

Lineage tracing documents in vivo reprogramming of myofibroblasts into iHeps

d

iHeps induced in vivo closely resemble hepatocytes

d

In vivo induction of iHeps ameliorates chemically induced liver fibrosis

Song et al., 2016, Cell Stem Cell 18, 1–12 May 5, 2016 ª2016 Elsevier Inc. http://dx.doi.org/10.1016/j.stem.2016.01.010

Accession Numbers GSE76843

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

Cell Stem Cell

Article Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis Guangqi Song,1,2,3,13 Martin Pacher,1,4,13 Asha Balakrishnan,1,4 Qinggong Yuan,1,4 Hsin-Chieh Tsay,1,2,4 Dakai Yang,1,2 Julia Reetz,5 Sabine Brandes,1,4 Zhen Dai,1,2 Brigitte M. Pu¨tzer,5 Marcos J. Arau´zo-Bravo,6,7 Doris Steinemann,8 Tom Luedde,9 Robert F. Schwabe,10 Michael P. Manns,1 Hans R. Scho¨ler,11 Axel Schambach,12 Tobias Cantz,1,3 Michael Ott,1,4,14,* and Amar Deep Sharma1,2,14,* 1Department

of Gastroenterology, Hepatology, and Endocrinology, Hannover Medical School, Hannover 30625, Germany Research Group MicroRNA in Liver Regeneration, Cluster of Excellence REBIRTH, Hannover Medical School, Hannover 30625, Germany 3Translational Hepatology and Stem Cell Biology, Cluster of Excellence REBIRTH, Hannover Medical School, Hannover 30625, Germany 4Twincore Centre for Experimental and Clinical Infection Research, Hannover 30625, Germany 5Institute for Experimental Gene Therapy and Cancer Research, Rostock University Medical Center, Rostock 18057, Germany 6Group of Computational Biology and Systems Biomedicine, Biodonostia Health Research Institute, San Sebastia ´ n 20014, Spain 7IKERBASQUE, Basque Foundation for Science, Bilbao 48013, Spain 8Institute of Human Genetics, Hannover Medical School, Hannover 30625, Germany 9Division of Hepatobiliary Oncology, Department of Medicine III, University Hospital RWTH, Aachen 52074, Germany 10Department of Medicine, Columbia University, New York, NY 10032, USA 11Department of Cell and Developmental Biology, Max Planck Institute for Molecular Biomedicine, Mu ¨ nster 48149, Germany 12Institute for Experimental Hematology, Hannover Medical School, Hannover 30625, Germany 13Co-first author 14Co-senior author *Correspondence: [email protected] (M.O.), [email protected] (A.D.S.) http://dx.doi.org/10.1016/j.stem.2016.01.010 2Junior

SUMMARY

Direct induction of induced hepatocytes (iHeps) from fibroblasts holds potential as a strategy for regenerative medicine but until now has only been shown in culture settings. Here, we describe in vivo iHep formation using transcription factor induction and genetic fate tracing in mouse models of chronic liver disease. We show that ectopic expression of the transcription factors FOXA3, GATA4, HNF1A, and HNF4A from a polycistronic lentiviral vector converts mouse myofibroblasts into cells with a hepatocyte phenotype. In vivo expression of the same set of transcription factors from a p75 neurotrophin receptor peptide (p75NTRp)-tagged adenovirus enabled the generation of hepatocyte-like cells from myofibroblasts in fibrotic mouse livers and reduced liver fibrosis. We have therefore been able to convert pro-fibrogenic myofibroblasts in the liver into hepatocyte-like cells with positive functional benefits. This direct in vivo reprogramming approach may open new avenues for the treatment of chronic liver disease.

into neurons (Grande et al., 2013; Guo et al., 2014; Han et al., 2012; Kim et al., 2011; Lujan et al., 2012; Ring et al., 2012; Son et al., 2011; Su et al., 2014; Thier et al., 2012; Torper et al., 2013; Vierbuchen et al., 2010), cardiomyocytes (Ieda et al., 2010; Qian et al., 2012), and hepatocytes (Du et al., 2014; Huang et al., 2011, 2014; Morris et al., 2014; Sekiya and Suzuki, 2011) through overexpression of specific TFs has been demonstrated. For example, functional neurons were generated from mouse fibroblasts by ectopic expression of Ascl1, Brn2, and Myt1l (Vierbuchen et al., 2010). Similarly, a combination of three cardiacspecific TFs, Gata4, Mef2c, and Tbx5, directly reprogrammed mouse cardiac fibroblasts into cardiomyocyte-like cells in vitro and in vivo in a heart infarction model (Ieda et al., 2010; Song et al., 2012). Lineage reprogramming of mouse fibroblasts into hepatocytes has been achieved in vitro by ectopic expression of HNF4A plus Foxa1, Foxa2 or Foxa3 or by the combination of Gata4, Hnf1a, and Foxa3 and inactivation of p19(Arf) in culture (Huang et al., 2011; Sekiya and Suzuki, 2011). Recently, two other groups independently reprogrammed human fibroblasts into hepatocytes by forced ectopic expression of FOXA3, HNF1A, and HNF4A or the combination of HNF1A, HNF4A, and HNF6 together with the maturation factors ATF5, PROX1, and CEBPA (Du et al., 2014; Huang et al., 2014). The direct conversion of fibroblasts into hepatocyte-like cells in vivo remains to be investigated, and the question needs to be addressed whether such an approach could also ameliorate the degree of fibrosis in damaged livers.

INTRODUCTION RESULTS The utility of transcription factors (TFs) for the acquisition of novel cell fates has been unlocked through landmark studies of induced pluripotent stem cell (iPSC) reprogramming (Takahashi et al., 2007; Takahashi and Yamanaka, 2006). Recently, by circumventing the pluripotent cell state, direct reprogramming of fibroblasts

FOXA3, GATA4, HNF1A, and HNF4A Convert Hepatic Myofibroblasts into iHeps In Vitro Based on previously reported results (Huang et al., 2011; Iacob et al., 2011; Sekiya and Suzuki, 2011), we screened TFs Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc. 1

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

Figure 1. FOXA3, GATA4, HNF1A, and HNF4A Directly Reprogram Mouse Myofibroblasts into iHeps (A) Screening of transcription factors in myofibroblasts. Albumin secretion and CYP3A activity were measured as read out during screening of seven transcription factors. The data represent values from three independent experiments. (B) Lentiviral vector maps, the LV.4TF used to generate iHeps, and a reporter lentiviral vector, used to detect albumin-positive cells. (C) Schematic of iHep generation from myofibroblasts. To prepare myofibroblasts, primary HSCs were isolated from BALB/c mice and cultured in the presence of platelet-derived growth factor (PDGF) before transduction with LV.4TF. (D) Phase-contrast microscopy (3100) of myofibroblasts and iHeps at day 14 after lentiviral transduction with LV.4TF. Scale bars, 200 mM. (E) iHeps express dTomato while myofibroblasts, cultured in HCM for 14 days and transduced with reporter lentiviral vector, show absence of dTomato. Scale bars, 200 mM. (F) A representative FACS plot (n = 3 independent experiments) showing the percentage of albumin promoter-driven dTomato-positive cells. (G) Representative RT-PCR (n = 3 independent experiments) showing the expression of hepatic genes in iHeps, whereas fibroblast genes such as Acta1, Col1a1, and Col2a1 were downregulated. Murine primary hepatocytes (Pr. Hc) cultured for 24 hr on a collagen matrix were used as positive control. (H) 2D principal component analyses indicate that the global expression profiles of iHeps (n = 3) are distinct from myofibroblasts and more similar to primary mouse hepatocytes cultured for 24 hr. (I) Representative pictures from three independent experiments showing PAS staining and LDL uptake in iHeps. Scale bars, 200 mM. (J) Albumin secretion in the supernatant was measured in the absence of serum components. CYP1A2 and 3A activities in 24 hr cultured primary hepatocytes, iHeps, and myofibroblasts. The values shown are mean of three independent experiments. (K) aCGH-based karyotype analysis of myofibroblasts derived iHeps. The data shown in (G)–(K) are obtained from FACS-purified iHeps. See also Figure S1.

(FOXA1, FOXA2, FOXA3, GATA4, HNF1A, HNF4A, and CEBPA) for direct reprogramming of myofibroblasts derived from primary hepatic stellate cells into hepatocytes. FOXA3, GATA4, HNF1A, and HNF4A transcription factors (4TFs) were selected for further experiments, since the absence of each of these TFs substantially reduced albumin secretion and CYP3A activity in our screening experiments (Figure 1A). To overexpress the selected 4TFs simultaneously, we cloned the cDNAs into a polycistronic lentiviral vector (henceforth referred to as LV.4TF) (Figure 1B). 2 Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc.

In response to persistent inflammatory injuries large numbers of stellate cells, which have undergone ‘‘activation’’ to pro-fibrogenic myofibroblasts, accumulate in the liver. To harness the potential therapeutic effects of direct reprogramming in chronic liver disease, we first tested whether forced expression of 4TFs would induce a hepatocyte phenotype in cultured myofibroblasts (Figure 1C). The expression of 4TFs in transduced myofibroblasts was confirmed on day 4 by qRT-PCR (Figure S1A). Within 14 days after transduction, we observed profound changes in morphology of the transduced cells now resembling

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

an epithelial phenotype (Figure 1D). To confirm the generation of hepatocytes from myofibroblasts, we transduced cells at day 10 with a reporter lentiviral vector, which expressed dTomato under transcriptional control of the albumin promoter (Figure 1B). Approximately 12% of the cells expressed dTomato, indicating the generation of iHeps (Figures 1E and 1F). We enriched dTomato protein expressing iHeps by fluorescence-activated cell sorting (FACS) for further characterization (Figures 1G–1K). The iHeps showed expression, albeit at lower levels, of typical primary hepatocyte (Pr. Hc, cultured for 24 hr) markers and the downregulation of fibroblast genes (Figures 1G, 1H, and S1B). Our principal-component analysis (PCA) analyses suggested that although iHeps acquired a hepatic gene expression profile, they remained a distinct cell type when compared to primary hepatocytes cultured for 24 hr. Importantly, iHeps exhibited functional characteristics of hepatocytes as they stored glycogen, showed uptake of low-density lipoprotein (LDL), secreted albumin, and acquired cytochrome P450 (CYP1A2 and 3A) activities (Figures 1I and 1J). Genomic integrity was confirmed by arraybased comparative genomic hybridization (aCGH) analysis (Figure 1K). Thus, ectopic expression of FOXA3, GATA4, HNF1A, and HNF4A converts myofibroblasts into iHeps in vitro. Establishment of a Lineage-Tracing Model to Detect In Vivo Reprogramming We next examined whether 4TFs expression would facilitate iHep formation in vivo. To investigate 4TFs-mediated lineage reprogramming in vivo, we developed a mouse model to detect iHeps derived from non-parenchymal cells (Figure 2A). We used (Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP) mice (henceforth referred to as mT/mG). All cells including hepatocytes and non-parenchymal liver cells from these mT/mG mice express membrane-targeted tdTomato (mT) before Cre-mediated recombination (Muzumdar et al., 2007) (Figure 2B). To label endogenous hepatocytes, adult mT/mG mice were injected intrasplenically with 4 3 1011 adeno-associated virus (AAV) serotype 8 particles expressing Cre recombinase under transcriptional control of the liver-specific transthyretin (Ttr) promoter (Malato et al., 2011; Sharma et al., 2011). As a result, membranous EGFP fluorescence in hepatocytes and tdTomato membranous fluorescence in non-parenchymal cells of the liver were expressed four weeks after AAV-Ttr-Cre injection (Figure 2C). Then, we induced myofibroblasts in the liver of AAV-Ttr-Cre-injected mT/mG mice by intraperitoneal injections of carbon tetrachloride (CCl4) twice per week for a total of 8 weeks. These mice developed extensive fibrosis (Figure 2D) and accumulated myofibroblasts in the tissue. Notably, in these fibrotic livers, all non-parenchymal cells including myofibroblasts expressed tdTomato, whereas hepatocytes expressed EGFP (Figure 2E). To overexpress 4TFs in myofibroblasts, we designed a p75NTRp- tagged recombinant adenoviral vector (serotype 5), which expressed all 4TFs from a polycistronic transgene cassette. The adenoviral vector was modified to target mouse myofibroblasts through coupling of adenoviral fiber knobs with a peptide (single-chain antibody fragment) of the nerve growth factor (NGFp), which was selected for specific and high-affinity binding to the p75 neurotrophin receptor (p75NTR) present on hepatic stellate cells and myofibroblasts (Reetz et al., 2013). Efficiency of vector targeting was tested in cell culture, as previ-

ously published (Reetz et al., 2013) (Figure S2A). In vivo targeting of myofibroblasts by Ad.GFP-S11-NGFp was tested by injecting 5 3 109 adenoviral particles via the portal vein of fibrotic wildtype BALB/c mice as shown previously (Reetz et al., 2013). The amount of injected virus particles was sufficient to transduce 30% of stellate cells in normal and 20% of myofibroblasts in CCl4-induced fibrotic livers of BALB/c mice. Predominantly, myofibroblasts were found to express EGFP in the liver of mice injected with Ad.GFP-S11-NGFp (Figure 2F), whereas control adenovirus vector injected mice showed transduction of hepatocytes (Figure 2G). Co-staining of EGFP with albumin (hepatocyte marker), desmin (myofibroblast marker), cytokeratin (CK) 19 (bile duct and liver stem/progenitor cell marker), F4/80 (Kupffer cell marker), CD31 (endothelial cell marker), and CD45 (hematopoietic cell marker) revealed preferential transduction of myofibroblasts, but not of hepatocytes (except <0.05% rare double-positive cells), Kupffer cells, endothelial cells, bile duct cells, or liver/ stem progenitor cells (undetected) (Figure S2B). Therefore, injection of Ad.GFP-S11-NGFp allows preferential targeting of myofibroblasts. Accordingly, one week after cessation of the CCl4 treatment, we injected 5 3 109 p75NTRp-tagged Ad5.FOXA3. GATA4.HNF1A.HNF4A (henceforth referred to as Ad.4TF) via the portal vein of AAV-Ttr-Cre injected fibrotic wild-type BALB/c mice. This led to successful overexpression of 4TFs in sorted myofibroblasts from these mice within 4 days after injection (Figure 2H). Importantly, Ad.4TF administration in normal BALB/c mice neither affected liver function tests nor histology (Figure 2I). In Vivo iHep Formation via Direct Reprogramming After establishing a model that faithfully identifies iHep formation, we examined iHep formation in mT/mG mice injected with AAV-Ttr-Cre followed by CCl4 injections and subsequent administration of Ad.4TF. The control animals were treated exactly in the same manner, except for injection with empty adenoviral vector instead of Ad.4TF. In our lineage-tracing model, the endogenous hepatocytes would express membranous EGFP, while the iHeps could be detected by tdTomato positive membrane fluorescence (Figure 2J). To identify iHeps and distinguish them from endogenous hepatocytes, we stained liver tissues 30 days after Ad.4TF injection for albumin, major urinary protein (MUP), fumarylacetoacetate hydrolase (FAH), alpha-1 antitrypsin (AAT), and HNF4A, all characteristic hepatocyte markers in the liver. Immunofluorescence analyses of the liver harvested at 30 days after Ad.4TF injection showed albumin, MUP, FAH, AAT, and HNF4A expression mainly in endogenous hepatocytes with EGFP-positive membranes (Figures 2K and S2C). However, few cells with tdTomato-positive membranes also stained positive for albumin, MUP, FAH, AAT, and HNF4A in the liver of Ad.4TF-injected mice. These tdTomatopositive cells that also stained positive for hepatocyte markers, either in clusters or as single cells, indicate the presence of iHeps. The control animals (n = 10), injected with empty adenoviral vector, did not express albumin, MUP, FAH, and AAT in any of the tdTomato-positive cells. The efficiency of direct reprogramming in vivo was calculated from the average of tdTomato-expressing cells, which stained positive for albumin, MUP, FAH, AAT, and HNF4A relative to the total hepatocyte population (n = 10 mice). The percentage of in-vivo-generated iHeps among the total hepatocyte population ranged from Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc. 3

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

A CCl4 injection (twice weekly)

day 0

30

Relative expression

H

R NT

5

P7

U/L

I 50 40 30 20 10 0 Co

.4

Ad

90

1.2 1.0 0.8 0.6 0.4 0.2 0.0

FOXA3

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

GATA4

1.2 1.0 0.8 0.6 0.4 0.2 0.0 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

HNF1A

i TF

97

on

cti

nje

AAV-Ttr-Cre injection

De

te

n ctio

of

iHe

p

B

mT/mG mice before AAV-Ttr-Cre injection

C

mT/mG mice after AAV-Ttr-Cre injection

D

F Sirius red

F

Non-parenchymal cells Hepatocytes

G Ad.GFP-S11-NGFp

Ad.GFP

127

J Endogenous hepatocytes (eHep)

***

Reprogrammed hepatocytes (iHep)

K Control EGFP

tdTomato

***

Ad.4TF Hepatocyte marker

Merged

EGFP

tdTomato

Hepatocyte marker

Merged

ALB ***

HNF4A MUP ***

e ive yte sit oc po epat R H T an 5N P7 h um . Pr

ne

ga

t iv

AST ns

40 30 20 10 0

ALT ns

F ol TF ol 4 ntr d.4T ntr Co Ad. A Ad.4TF Control

FAH

AAT

l

HE

Desmin

L

% of iHep / total hepatocytes / mouse Ad.4TF 0.4 1.2 0.7 0.2 0.5 1.0 0.4 0.8 0.7 0.5 Control 0

0

0

0

0

0

0

0

0

0

Figure 2. Establishment of a Lineage-Tracing Model to Detect In Vivo iHep Formation (A) Schematic of the model. (B) Representative pictures showing native tdTomato membranous fluorescence (red) is present in all liver cells including hepatocytes of adult mT/mG mice (n = 4). Nuclei (blue) are stained with DAPI. Scale bar, 100 mM. (C) All hepatocytes express membranous EGFP in liver of 4 3 1011 AAV-Ttr-Cre injected mT/mG mice (n = 4), while non-parenchymal cells retain tdTomato expression. Nuclei (blue) are stained with DAPI. Arrowheads indicate non-parenchymal cells that retained tdTomato. Scale bar, 100 mM. (D) Representative picture of Sirius red staining of liver (n = 5) after 16 intraperitoneal CCl4 injections (twice per week for 8 weeks). Scale bar, 200 mM. (E) Cells in the vicinity of emerging fibrous septa express native tdTomato fluorescence in CCl4-treated AAV-Ttr-Cre-injected mT/mG mice (n = 5). Scale bar, 200 mM. (F) Preferential in vivo transduction of HSC and myofibroblasts in liver of mice (n = 4) by Ad.GFP-S11-NGFp adenoviral vector. Scale bar, 200 mM. (G) In contrast, injection of control adenoviral vector Ad.GFP leads to the homogenous expression of GFP in hepatocytes (n = 4 mice). Scale bar, 200 mM. (H) In vivo overexpression of 4TFs in myofibroblasts of CCl4-treated BALB/c mice that were injected with Ad.4TF was confirmed by qRT-PCR. Four days after injection, myofibroblasts were sorted by staining with P75NTR antibody followed by FACS. Primary human hepatocytes were used as positive control. The data represent values from three independent experiments. (I) Similar levels of serum transaminases, H&E staining, and desmin staining suggest normal liver functions and absence of any histological abnormality in uninjured BALB/c mice (n = 4) injected with Ad.4TF. (J–L) p75NTR tagged Ad5.FOXA3.GATA4.HNF1A.HNF4A induces hepatocyte-specific gene expression in fibrotic mice. (J) Schematic illustration shows that endogenous hepatocytes can be identified by EGFP expression, whereas reprogrammed hepatocytes are identified by tdTomato expression. (K) Immunofluorescence staining with antibodies (blue) for hepatocyte markers such as albumin, MUP, FAH, and AAT on cryosections obtained from mT/mG (Gt(ROSA) 26Sortm4(ACTB-tdTomato,-EGFP)) mice injected with 4 3 1011 AAV-Ttr-Cre. These confocal figures are representative of livers obtained from ten mice in each group. Scale bars, 100 mM. (L) Table shows percentage of average tdTomato-expressing cells that stained positive for ALB, MUP, FAH, and AAT in the total hepatocyte population. Ten random sections were stained for each antigen per mouse (n = 10 mice). See also Figure S2.

0.2% to 1.2% in Ad.4TF-injected mice, whereas control mice did not show any reprogrammed cells (Figure 2L). Of note, we did not detect iHeps when Ad.4TF was administered in uninjured mice (data not shown). Since the number of myofibroblasts in fibrotic livers at 1 week after cessation of 8 weeks of 4 Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc.

CCL4 treatment is between 15.4% and 21.7% of the total number of liver cells (Figure S2D), it can be estimated that 0.2% to 1.2% of iHeps detected in AD.4TF-injected mice is equivalent to a reprogramming efficiency of less than 4%. It is noteworthy to mention that reprogramming efficiency may be even lower

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

Figure 3. Functional Characterization of In-Vivo-Reprogrammed iHeps and Evidence for Amelioration of Liver Fibrosis in Ad.4TFInjected Mice (A) Schematic of the FACS sorting and analyses of fibrosis. (B) Isolation of in-vivo-generated iHeps (tdTomato positive) and eHeps (EGFP positive) by FACS sorting. Representative pictures and FACS sorting (n = 6 mice) are shown. (C–G) The tdTomato-positive cells were pooled together and data are shown from technical triplicates. Scale bars, 200 mM. (C) Albumin ELISA revealed comparable levels of secreted albumin in iHeps and eHeps. (D) Urea synthesis was also found to be similar in iHeps and eHeps. (E) The iHeps show ICG uptake, oil red O staining, and PAS staining. Scale bars, 200 mM. (F) The iHeps and eHeps showed activities for CYP3A, CYP1A1, CYP2C9, and CYP1A2. (G) Evidence for drug response in iHeps was demonstrated by elevated levels of Cyp1a1, Abcc2, Ugt1a1, and Oatp. (H–K) Amelioration of liver fibrosis in Ad.4TF-injected mice. (H) Reduced levels of Col1a1 mRNA in Ad.4TF-injected mice (n = 9) compared to control mice (n = 9). (I) Hydroxyproline assay showed decreased levels of entire collagen content, measured in whole liver. (J) H&E, Sirius red, and immunohistochemical staining for desmin and p75NTR showed less fibrosis in Ad.4TF injected mice (n = 9) than respective controls (n = 9). Scale bars, 100 mM for H&E and desmin and 100 mM for Sirius red and p75NTR staining. (K) Quantification of Sirius red, desmin, and p75NTR stainings shown in (J). (L) Reduced levels of serum transaminases suggest improved liver functions in Ad.4TF-administered, CCl4-treated, AAV-Ttr-Cre-injected mT/mG mice (n = 3). (M) A representative photograph of H&E staining after an 8-month follow-up study of Ad.4TF-administered and CCl4-treated AAV-Ttr-Cre-injected mT/mG mice (n = 3) showing normal histology and no tumor formation.

than 4% if iHeps proliferate between the time when they appear and 30 days after Ad.4TF administration. Thus, our data indicate that expression of 4TFs in myofibroblasts facilitates the formation of iHeps in chronic liver disease. Functional Analyses of In-Vivo-Generated iHeps One of the prerequisites for in-vivo-generated iHeps to be considered as hepatocytes is to possess the functions of mature hepatocytes. To characterize iHeps at the functional level, we isolated iHeps by Percoll density-gradient centrifugation followed by FACS from the liver of Ad.4TF-injected mice (Figures

3A and 3B). We did not find iHeps in control mice by FACS as well (Figure S2E). We specifically sorted for single-tdTomatopositive cells, because double-positive cells may represent endogenous cells that have not silenced tdTomato completely. The iHeps that were tdTomato positive, cultured for 24 hr, showed albumin secretion and urea synthesis similar to EGFPpositive endogenous hepatocytes (eHeps) (Figures 3C and 3D). The iHeps showed the ability to uptake indocyanine green (ICG); stained positive for oil red O, indicating the presence of triglycerides and lipids; stained positive for PAS, thus demonstrating the ability to store glycogen (Figure 3E); and exhibited Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc. 5

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

Figure 4. Administration of Ad.4TF during Ongoing Injury in AAV-Ttr-Cre-Injected mT/mG Mice Also Generates iHeps and Leads to a Reduction in Fibrotic Markers (A) Schematic of the experimental design. CCl4 was injected twice weekly for 4 weeks before administration of Ad.4TF. Mice were injected again for 4 more weeks. (B) Immunofluorescence staining with antibodies (blue color) for hepatocyte markers, albumin, and MUP. The figures are representative of livers obtained from four mice in each group. Scale bars, 100 mM. (C) Reduced levels of Col1a1 mRNA in Ad.4TF-injected mice (n = 4) compared to control mice (n = 4). (D) Hydroxyproline assay showed decreased levels of entire collagen content, measured in whole liver. (E) Sirius red and immunohistochemical staining for desmin showed less fibrosis in Ad.4TF-injected mice (n = 4) compared to respective controls (n = 4). Scale bars, 200 mM for Sirius red and 100 mM for desmin staining. Right, quantifications of Sirius red and desmin stainings are shown. See also Figure S3.

cytochrome activity (CYP3A, 1A1, 2C9, and 1A2) similar to eHeps (Figure 3f). Furthermore, we tested whether isolated iHeps are drug inducible. The treatment of iHeps with phenobarbital and rifampicin led to induction of Cyp1a1 (phase 1), Ugt1a1 (phase 2), Abcc2, and Oatp (Figure 3G). Importantly, the in-vivogenerated iHeps showed chromosomes with no obvious numerical or structural aberrations (Figure S2F) and the absence of exogenous 4TFs (Figure S2G), thus indicating the stable reprogramming of myofibroblasts into iHeps. In addition, Ki67 staining indicated that iHeps have the ability to proliferate in response to two-thirds partial hepatectomy in vivo (Figure S2H). Therefore, our thorough characterization suggests that in-vivo-generated iHeps possess functional properties of hepatocytes. Overexpression of 4TFs Ameliorates Chemical-Induced Liver Fibrosis To examine whether in vivo reprogramming ameliorates chronic liver disease, we determined the extent of liver fibrosis in CCl4-in6 Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc.

jected mice. First, qPCR showed lower levels of Col1a1 in the liver of Ad.4TF-injected mice, suggesting less liver fibrosis compared to respective controls (Figure 3H). Second, we performed hydroxyproline assay, which measures the entire collagen content of a liver. The Ad.4TF-injected mice showed significantly reduced levels of hydroxyproline, indicating decreased liver fibrosis (Figure 3I). In addition, histological grading of fibrosis, Sirius red staining, immunohistochemical staining of desmin and p75NTR, as well as serum levels of aminotransferases confirmed decreased liver fibrosis in Ad.4TF-injected mice compared to control mice (Figures 3J–3L). In addition, an 8-month follow-up study of mice injected with Ad.4TF showed normal histology and no signs of liver tumors (Figure 3M). Next, we tested whether injection of Ad.4TF during ongoing liver injury can lead to in vivo iHep formation and a reduction of liver fibrosis (Figure 4A). Indeed, we detected the presence of iHeps and less liver fibrosis in mice injected with Ad.4TF during an ongoing injury (Figures 4B–4E). Hence, our results indicate

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

C

A

Cell types and Comparison

Number of upregulated genes

Number of downregulated genes iHep

In vitro myofibroblast-derived iHep and PH24 (mouse hepatocytes that were 24 hours in culture)

Missing genes from CellNet Liver GRN = 22 (gene regulatory network)

Missing genes from CellNet Liver GRN = 128 (gene regulatory network) Fibroblast GRN

Common genes from CellNet fibroblast GRN = 156 (gene regulatory network)

Second principal component (18%)

B Myofibroblasts In vivo Myofibroblasts-derived iHep

In vitro Myofibroblasts-derived iHep

In vivo myofibroblast-derived iHep and eHep (endogenous hepatocytes)

Liver GRN

1576

iHep

Myofibroblasts In vitro myofibroblasts-derived iHep In vivo eHep In vivo myofibroblasts-derived iHep

iHep

Liver GRN

1281

Liver GRN

Missing genes from CellNet Liver GRN = 8 (gene regulatory network) Genes: Agpat4, Cd34, Edn1, Foxc1, Hoxa5, Oxct1, Pdk3, Podxl.

iHep

Liver GRN

iHep

Fibroblast GRN

43

Missing genes from CellNet Liver GRN = 5 (gene regulatory network) Genes: Arhgef19, Cyp4a14, Cyp7a1, Dpys, Onecut1 iHep

Common genes from = 94 CellNet fibroblast GRN (gene regulatory network)

Fibroblast GRN

Common genes from CellNet fibroblast GRN = 28 (gene regulatory network)

iHep

4 26

iHep

Fibroblast GRN

Common genes from CellNet fibroblast GRN = 0 (gene regulatory network)

In vivo eHEP

First principal component (56%)

Figure 5. Microarray Analyses of Myofibroblast-Derived iHeps (A) Whole-transcriptome heatmap demonstrates clustering of in vivo iHeps (n = 3), eHeps, myofibroblasts, and in vitro iHeps. (B) 2D principal component analyses indicate that the global expression profiles of in vivo iHeps (n = 3) resembles freshly isolated eHeps compared to in vitro iHeps. (C) Comparison of our in vitro and in vivo iHeps with liver GRN or fibroblasts GRN obtained from CellNet.

that 4TFs-induced in vivo reprogramming ameliorates liver fibrosis in mice injected with CCl4 for 8 weeks. CCl4 administration for up to 8 weeks induces a significant amount of liver fibrosis; however, it is still reversible. We therefore examined whether the forced expression of 4TF is capable of ameliorating liver injury in mice injected for 12 weeks with CCl4 resembling an irreversible cirrhosis model (Figure S3A). Although, we detected the presence of iHeps, a beneficial effect on liver fibrosis was not observed in mice injected with Ad.4TF (Figures S3B–S3F). This might be due to the fact that the number of Ad.4TF reprogrammed iHeps in mice injected for 12 weeks with CCl4 remained similar to those in mice injected for 8 weeks, whereas total myofibroblast numbers further increased after prolonged CCl4 administration. This phenomenon can be explained, at least in part, by a modest but similar number of myofibroblasts transduced by Ad.4TF vector in both 8-week and 12-week CCl4 models, but these were most likely outnumbered by the massive increase of fibrotic cells in the 12-week CCl4 model. Characterization of In-Vivo-Generated iHeps To further prove the hepatic lineage identity of in vivo iHeps, we performed mRNA microarrays after pooling several livers isolated from different animals and compared them with myofibroblasts, myofibroblast-derived iHeps in vitro, and endogenous hepatocytes (eHeps, serving as positive control) in vivo. In silico analyses, such as hierarchical clustering and PCA, demonstrated that in vivo iHeps are more similar to eHeps than to in vitro iHeps

(Figures 5A and 5B). This suggests that the in vivo hepatic environment further contributes to the maturation of in vivo iHeps. The identity of reprogrammed cells can be faithfully determined by the CellNet platform with very high accuracy (Cahan et al., 2014; Morris et al., 2014). We therefore extracted a list of genes comprising either the liver-typical or fibroblast-typical gene regulatory networks (GRNs) from CellNet and compared our in vivo or in vitro myofibroblast-derived iHeps (Figure 5C). In iHeps derived from myofibroblasts in vitro, 1,281 genes were upregulated by more than 4-fold and 1,576 genes were downregulated by more than 4-fold compared to 24-hr-cultured primary mouse hepatocytes. 22 (of 1,281) upregulated genes and 128 (of 1,576) downregulated genes were in common with the set of genes comprising the CellNet liver GRN. To assess genetic memory of myofibroblasts, we compared in vitro myofibroblast-derived iHeps with fibroblast GRN. Deregulated genes common with the fibroblast GRN included 184 (156 upregulated and 28 downregulated) genes. When in vivo iHeps were compared to in vivo eHeps (endogenous hepatocytes), a total of 426 genes were upregulated by more than 4-fold and 43 were downregulated by more than 4-fold. Only 8 (of 426) upregulated genes and 5 (of 43) downregulated genes were in common with the set of genes comprising the CellNet liver GRN. Deregulated genes common with the fibroblast GRN include 94 (94 upregulated and 0 downregulated) genes. Deregulated genes (>4-fold up or down) in the in vitro iHeps derived from myofibroblasts thus represent 35% of the CellNet Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc. 7

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

A

day 0

F .4T

Ad

60 67

n

tio

ec

CCl4 injection (twice weekly)

inj

te De

ctio

no

f iH

ep

C Control tdTomato

EGFP

Ad.4TF

Hepatocyte marker

Merged

tdTomato

EGFP

Hepatocyte marker

Merged

97 ALB

B

Endogenous hepatocytes (eHep) Reprogrammed hepatocytes (iHep)

D

MUP % of iHep / total hepatocytes / mouse

Ad.4TF

1.6

2

2.5

0.9

Control

0

0

0

0

E

Unsorted hepatocytes

FAH

eHep (dTOMATO positive)

AAT

FACS Sorting

iHep

EGFP

CYP3A

6000 5000 4000 3000 2000 1000 0

*

CYP1A1

eHepiHep Urea 14 * 12 10 8 6 4 2 0 eHep iHep

Oil red O

ns

PAS

ns 45 40 35 30 25 20 15 10 5 0 eHep iHep

CYP2C9 3500 3000 2500 2000 1500 1000 500 0

RLU/ 10,000cells

eHep 250000 ns 200000 150000 iHep 100000 50000 0 eHep iHep

G Urea µg/24h/10000 cells

Albumin (pg/24h/10000 cells)

Albumin

RLU/ 10,000cells

ICG uptake

1.56%

F

I

CYP1A2 3500 3000 2500 2000 1500 1000 500 0

ns

eHep iHep

J

Abcc2 *** 8.0 7.0 6.0 5.0 *** 4.0 3.0 2.0 1.0 0.0 -+ - + - + - + eHep iHep eHep iHep Cyp1a1

Relative mRNA expression

eHep

RLU/ 10,000cells

dTOMATO

H iHep (EGFP positive)

CYP1A1 (pmol/10,000 cells /2hour)

95.8%

eHep iHep

3.0 2.5 2.0 1.5 1.0 0.5 0.0

5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0

**

*

Ugt1a1 * *

- + - + eHep iHep

10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0

Phenobarbital

Oatp **

***

- + - + Rifampicin eHep iHep

Figure 6. Demonstration of iHep Formation and Their Functional Characterization in LratCre-mT/mG Mice (A) Schematic of the model. (B) In LratCre-mT/mG model, eHeps are tdTomato positive whereas iHeps are EGFP positive. (C) Immunofluorescence staining with albumin, MUP, FAH, and AAT antibody (blue) on cryosections obtained from LratCre-mT/mG mice showed the presence of iHeps (green) only in mice injected with Ad.4TF. The figures are representative of livers obtained from four mice in each group. Scale bars, 100 mM. (D) Percentage of iHeps in the total hepatocyte population (n = 4 mice). Ten sections were stained for each antigen per mouse (n = 4 mice). (E) Isolation of in-vivo-generated iHeps (EGFP positive) and eHeps (tdTomato positive) by FACS sorting. Representative pictures and FACS (n = 4 mice) are shown. Scale bars, 200 mM. (F–J) The iHeps were pooled and data are shown from triplicates. (F) Albumin secretion and (G) urea synthesis were similar in iHeps and eHeps. (H) The iHeps show ICG uptake, oil red O staining, and PAS staining. Scale bars, 100 mM for ICG uptake, 50 mM for oil red O, and 200 mM for PAS staining. (I) iHeps and eHeps showed activity for CYP3A, CYP1A1, CYP2C9, and CYP1A2. (J) Evidence for drug response in iHeps was demonstrated by elevated levels of Cyp1a1, Abcc2, Ugt1a1, and Oatp. See also Figures S4–S7.

listed genes comprising the liver GRN. In contrast to the in-vitrogenerated iHeps, the in vivo iHeps show only 3.3% deregulated genes among the genes comprising the liver GRN. From these data, together with our thorough analyses of various liver metabolic functions, we conclude that the in vivo iHeps provide a similar liver GRN with only minor differences in 13 of 421 genes. In addition, we also provide evidence for (limited) genetic memory in a number of (myo)fibroblast GRNs. Confirmation of iHep Formation in a Direct LineageTracing Model Although, the AAV-Ttr-Cre-injected mT/mG model in combination with Ad.4TF is an appropriate model to detect iHeps in vivo, we examined the generation of iHeps in a second independent 8 Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc.

mouse model. To specifically detect myofibroblast-derived iHeps, we used mice expressing Cre under the transcriptional control of the lecithin-retinol acyltransferase (Lrat) promoter, which also contain the mT/mG transgenes (Figure 6A). LratCrebased lineage tracing has been shown to effectively trace myofibroblasts derived from the hepatic stellate cell (HSC) lineage (Mederacke et al., 2013). We first confirmed the labeling of HSCs and myofibroblasts expressing EGFP in LratCre-mT/mG mice, consistent with a previous report (Mederacke et al., 2013) (Figures S4A and S4B). We then examined the generation of iHeps after administration of Ad.4TF in mice injected for 8 weeks with CCl4. Again, we conclusively observed iHeps (either in clusters or as single cells) expressing hepatocyte markers in mice injected with Ad.4TF with a percentage of total

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

Figure 7. Demonstration of iHep Formation in DDC-Induced Liver Fibrosis (A) Schematic of the experimental design. Mice were kept on a DDC diet for a total of 4 weeks. Ad.4TF was administered 2 weeks after beginning the DDC diet. (B) Immunofluorescence staining with antibodies (blue) for hepatocyte markers, albumin, MUP, FAH, and AAT. The figures are representative of livers obtained from three mice in each group. Scale bars, 100 mM. (C) Table shows percentage of average EGFP-positive iHeps among the total hepatocyte population. (D) Reduced levels of Col1a1 mRNA in Ad.4TF injected mice (n = 4) compared to control mice (n = 4). (E) Hydroxyproline assay showed decreased levels of entire collagen content, measured in whole liver. (F) Sirius red and immunohistochemical staining for desmin showed less fibrosis in Ad.4TF-injected mice (n = 4) compared to respective controls (n = 4). Scale bars, 200 mM for Sirius red and 100 mM for desmin staining. Right, quantification of Sirius red and desmin stainings are shown. (G–I) The iHeps were pooled, and data are shown from triplicates. (G) The iHeps show ICG uptake, oil red O staining, and PAS staining. Scale bars, 100 mM for ICG uptake, 50 mM for oil red O, and 200 mM for PAS staining. (H) The iHeps and eHeps showed activity for CYP3A, CYP1A1, CYP2C9, and CYP1A2. (I) Evidence for a drug response in iHeps was demonstrated by elevated levels of Cyp1a1, Abcc2, Ugt1a1, and Oatp.

hepatocytes ranging from 0.9% to 2.5% (Figures 6B–6D and S4C). Notably, we did not find iHeps in control LratCre-mT/mG mice, as all of the isolated hepatocytes expressed tdTomato only (Figure S4D). All EGFP-positive cells were myofibroblasts (Figure S4E). Furthermore, sorted in vivo iHeps exhibited functional characteristics similar to endogenous hepatocytes (Figures 6E–6J). In addition, isolated iHeps expressed hepatocyte markers (Fah, Ck18, Hnf4a, Hnf1a, Gjb1, Ck8, and Apoa1), but not the myofibroblast markers (Acta1, Col1a1, and Col2a1) or an intestinal marker (Cdx2) (Figure S4F). Importantly, in vivo iHeps possess the ability to proliferate in response to stimuli such as epidermal growth factor, in vitro (Figure S4G), or after two-thirds partial hepatectomy in vivo (Figure S4H). These iHeps regulated expression of genes involved in glucose metabolism such as G6pc and Pck1 (Figures S4I and S4J), and more impor-

tantly, glucose levels upon treatment with glucagon and insulin (Figures S4K and S4L). Thus, our findings in the LratCremT/mG model confirm that forced expression of 4TFs in myofibroblasts generates functional iHeps in vivo. Demonstration of iHep Formation in a CholestasisInduced Liver Fibrosis Model To rule out whether formation of in vivo iHeps is restricted to only CCl4-induced liver fibrosis, we investigated this phenomenon in a well-established cholestasis-induced liver fibrosis model. To address this, we analyzed iHep formation in LratCre-mT/mG mice fed an 3.5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) diet (Figure 7A). Similar to the CCl4 model, we detected iHeps that express EGFP and hepatocyte markers in mice injected with Ad.4TF (Figures 7B and 7C). We then Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc. 9

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

evaluated the effect of overexpression of 4TFs on DDC-induced fibrosis and characterized those iHeps similar to CCl4-induced liver fibrosis. Col1a1 expression analyses, hydroxyproline content, and Sirius red and desmin staining revealed the reduced fibrosis in 4TFs overexpressing mice (Figures 7D–7F). Furthermore, iHeps isolated from the DDC model also showed the functional characteristics of hepatocytes (Figures 7G–7I). Therefore, in vivo iHep formation can be induced by overexpression of 4TFs in the DDC model, and 4TFs overexpression ameliorates DDC-induced fibrosis. In-Vivo-Generated iHeps Lack Any Fibrolytic Activity In order to determine whether in-vivo-generated iHeps exert any direct fibrolytic effect, we co-cultured sorted eHeps or iHeps from LratCre-mT/mG mice with the Col-GFP hepatic stellate cell line. Unchanged mRNA expression of myofibroblast markers, such as Acta1, Col1a1, and Col2a1, suggests that in-vivo-generated iHeps were unable to suppress activation of stellate cells (Figure S4M). It further indicates that the observed reduction in fibrosis was due to decreased number of myofibroblasts rather than a fibrolytic effect of in-vivo-generated iHeps. Ad.4TF-Based Reprogramming Does Not Give Rise to Cells Other Than iHeps To assess whether other types of liver cells are produced via in vivo reprogramming, we stained livers for SOX9 (a liver stem/progenitor cell marker), CK19 (a bile duct and liver stem/progenitor cell marker), and MIC1 (an oval cell marker). Immunofluorescence analyses showed the absence of positive staining for any of the above-mentioned markers in cells with EGFP-positive membranes in the Ad.4TF-administered, CCl4treated, LratCre-mT/mG model (Figure S5). Thus, we confirmed that Ad.4TF injection in fibrotic livers leads to formation of iHeps in vivo without forming cholangiocytes or liver stem/progenitor cells. Since p75NTR is also expressed in tissues other than fibrotic livers, we also considered the possibility whether Ad.4TF administration converts cells of other organs into iHeps. However, our immunofluorescence staining for albumin and HNF4A revealed absence of positive cells in the brain, heart, lung, and kidney (Figure S6). Thus, our results indicate the absence of iHeps formation in organs other than the liver. In-Vivo-Generated iHeps Are Not the Result of Cell Fusion Next, we sought to investigate whether in-vivo-generated iHeps could be a result of cell fusion. To address this, we stained unsorted iHeps and eHeps, isolated from LratCre-mT/mG mice, for p75NTR, a myofibroblast-specific marker. By analyzing immunofluorescence stainings on more than 1,000 iHeps, we could not detect any iHeps, which were also positive for p75NTR (Figure S7). Furthermore, our aCGH analyses of invivo-generated iHeps did not show any increase in ploidy compared to equal DNA amounts of eHeps used as the reference sample (Figure S2F), thus suggesting absence of cell fusion. Together, these data suggest that in-vivo-generated iHeps are the result of direct reprogramming rather than cell fusion events. 10 Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc.

DISCUSSION Our findings indicate that simultaneous expression of FOXA3, GATA4, HNF1A, and HNF4A can reprogram mouse myofibroblasts into cells with a hepatocyte-like phenotype. Targeted, cell-type-specific expression of the TFs in mice results in a phenotype change of liver myofibroblasts into hepatic cells, which triggers tissue remodeling in chronic liver failure. Degradation of extracellular matrix, a decrease in myofibroblast numbers, and hepatocyte proliferation are considered necessary steps for effective resolution of fibrosis and recovery of the liver (Friedman et al., 2013; Kendall et al., 2009). Hence, the conversion of profibrogenic myofibroblasts into hepatocyte-like cells would not only attenuate the development of fibrosis but also improve liver function in chronic liver failure. Indeed, we provide evidence for amelioration of liver fibrosis following up to 8 weeks of CCl4 treatment in mice with iHeps. The observed amelioration of liver fibrosis by reprogramming of myofibroblasts into iHeps in vivo can be explained by two synergistic effects. On one hand, conversion of certain numbers of myofibroblasts, the major contributor of liver fibrosis, into iHeps, reduces the number of pro-fibrogenic myofibroblasts. In fact, reduced numbers of myofibroblasts can indeed ameliorate liver fibrosis (Puche et al., 2013). Notably, in both lineage-tracing mouse models, iHeps often appeared near the portal vein or central vein regions. Therefore, it is possible that a few iHeps may be generated from perivascular mesenchymal cells, including portal fibroblasts, smooth muscle cells around the portal vein, and fibroblasts around the central vein. On the other hand, generation of numerous new iHeps, which possess functional characteristics of primary hepatocytes, restores the deteriorating liver function that is frequently observed during liver fibrosis. There are a few unanswered questions that would require future studies to address them. First, it remains unclear why iHeps were detected only in fibrotic livers and not in normal mice. A plausible explanation is that quiescent HSCs present in normal liver express p75NTR at very low levels compared to its very high expression in myofibroblasts seen in fibrotic livers. This may, in turn, lead to suboptimal expression of 4TF in quiescent HSCs compared to their optimal expression in myofibroblasts. Additionally, the fibrotic milieu may also facilitate generation of iHeps from myofibroblasts. Second, it remains a possibility that iHeps are formed via an intermediate stage that resembles liver progenitor cells. A comprehensive characterization of iHeps between the time when they first appear and 30 days after Ad.4TF administration would not only answer whether iHeps are generated via a progenitor stage but also uncover key regulatory pathways involved in iHep formation. Third, we cannot rule out the possibility that overexpression of 4TFs in endogenous hepatocytes may contribute to reduction in fibrosis, since overexpression of HNF4A in hepatocytes has recently been suggested to revert long-term CCl4 induced damage in a rat model (Nishikawa et al., 2015). Our data, however, indicate that the direct conversion of myofibroblasts into hepatocyte-like cells is the more prominent factor in our experimental setup, since the delivery of 4TFs was preferentially targeted to myofibroblasts. The modest transduction efficiency of myofibroblasts in our in vivo experiments might explain the observation

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

that amelioration of liver fibrosis (8-week CCl4), but not of cirrhosis (12-week CCl4), was detectable. This latter finding can be attributed to a higher total number of myofibroblasts in cirrhotic livers, from which only a limited number (similar to that in fibrotic livers) was efficiently transduced. In summary, our study demonstrates the direct conversion of pro-fibrogenic myofibroblasts in vivo into hepatocyte-like cells in the liver and shows that this can indeed ameliorate fibrosis in damaged livers. This approach provides a promising therapeutic potential for the treatment of chronic liver disease. Therefore, future studies should be aimed to enhance the efficiency of iHep generation through improvement in targeted transduction in order to provide beneficial therapeutic effects through iHep formation in more advanced liver fibrosis. EXPERIMENTAL PROCEDURES Animals 12-week-old BALB/c and mT/mG (Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP) mice were obtained from Charles River Laboratories and Jackson Laboratory, respectively. Lrat-Cre transgenic mice were kindly provided by Robert Schwabe (Columbia University, New York, NY). To induce liver fibrosis, mice were injected with 4 ml/g 10% CCl4 dissolved in corn oil. To induce cholestasis-induced liver fibrosis, LratCre-mT/mG mice were fed a DDC diet. The use of animals for this study was approved by the Institutional Animal Care and Use Committee of the Hannover Medical School. Lentivirus Expression Plasmids Third-generation self-inactivating VSV-G lentiviral vectors were derived from the vector backbone pRRL.PPT.SF.pre* (Maetzig et al., 2011; Schambach et al., 2006). The human cDNAs of the TFs FOXA3, GATA4, HNF1A, and HNF4A were cloned together into the pRRL.PPT.SF.pre* plasmid. The cDNAs were separated by proteolytic cleavage sites. All 2A sites and the HNF4A cDNA were codon optimized according to favored human tRNA codons (Genscript). This process improves translatability and mRNA half-life and removes cryptic splice and poly(A) sites. For the reporter construct, a dTomato reporter gene was cloned under the transcriptional control of minimal albumin promoter/enhancer sequence. Lineage Reprogramming 1 3 105 primary mouse myofibroblasts cells were transduced with lentiviral vector encoding for 4TFs at MOI 5. One day after transduction, complete DMEM medium was replaced by HCM medium. On day 10, cells were transduced with reporter lentiviral vector before trypsinization and reseeding them on collagen-coated dishes the next day. Afterward, cells were maintained in HCM medium. Medium was changed every 2 days. 14 days after first transduction, the cells were tested for protein secretion and CYP activity or harvested for mRNA isolation. Generation of Recombinant Adeno-Associated Virus and Adenovirus Vectors AAV8.Ttr.Cre vector was prepared as described previously (Sharma et al., 2011). Adenovirus serotype-5-derived wild-type vector (Ad.GFP) expressing GFP and Ad5.FOXA3.GATA4.HNF1A.HNF4A expressing the 4TFs were generated by homologous recombination following cotransfection with pAdEasy1 in E. coli BJ5183. Ad vectors were propagated in HEK293 cells, purified by CsCl buoyant density centrifugation, and measured at an optical density of 260. Peptide coupling and titration of the vectors were performed as previously described (Reetz et al., 2013).

solved in 1.2% saturated aqueous picric acid solution, all from Sigma-Aldrich) and incubated for 60 min. Sections were rinsed with water, dehydrated, and mounted in xylene. Immunofluorescence stainings for albumin (Abcam, 19196), MUP (Santa Cruz, 21856), FAH (Abcam, 81087), AAT (Abcam, 117307), p75-NTR (Abcam, 8874), desmin (Thermo Scientific, RB-9140), CD45 (BioLegend, 103106), CD31 (Abcam, 56299), F4/80 (Abcam, 6640), SOX9 (Millipore, AB5535), HNF4A (Santa Cruz, 6556), CK 19 (Abcam15463), MIC1 (Thermo Scientific, MA5-16136), and Acta1 (Abcam, 5694) were performed on frozen sections following a standardized protocol. Quantification of immunofluorescence or immunohistochemical staining was performed using ImageJ software in a blinded manner. Global Gene Expression Analysis Whole Mouse Genome Oligo Microarray v2 (4x44K) (Agilent Technologies) was used to characterize global gene expression profiles of iHeps compared to myofibroblasts and primary mouse hepatocytes. All microarrays were performed at the Research Core Unit Transcriptomics of the Hanover Medical School. Briefly, total RNA was used to prepare the aminoallyl-UTP-modified (aaUTP) cRNAs (Amino Allyl MessageAmp II Kit, #AM1753; Life Technologies) as directed by the company. The aaUTP-cRNAs were labeled with Alexa Fluor 555 Reactive Dye (#A32756; Life Technologies). Prior to the reverse transcription reaction, 1 ml of a 1:5,000 dilution of Agilent’s One-Color spike-in Kit stock solution (#5188-5282, Agilent Technologies) was added to 100 ng total RNA of each analyzed sample. The cRNA fragmentation, hybridization, and washing steps were carried out according to Agilent’s One-Color Microarray-Based Gene Expression Analysis Protocol V5.7, except that 500 ng of each labeled cRNA sample was used for hybridization. Slides were scanned on the Agilent Micro Array Scanner G2565 CA (pixel resolution 5 mm, bit depth 20). Data extraction was performed with the Feature Extraction Software V10.7.3.1. Statistical Analyses Significance was determined with two-tailed, two-sample equal variance Student’s t test. A p value of < 0.05 was considered as significant. Error bars represent ± SEM. *p < 0.05, **p < 0.005, and ***p < 0.0005. ACCESSION NUMBERS The accession number for the gene expression data reported in this paper is GEO: GSE76843. SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures and seven figures and can be found with this article online at http://dx.doi. org/10.1016/j.stem.2016.01.010. AUTHOR CONTRIBUTIONS M.O. and A.D.S. conceived the idea, designed experiments, provided the conceptual framework for the study, and wrote the manuscript. G.S., A.B., and T.C. further contributed to manuscript preparation. G.S., M.P., A.B., Q.Y., H.-C.T., Z.D., and D.S. performed the experiments. G.S., M.P., and Q.Y. analyzed the data. D.Y. and S.B. prepared AAV-Ttr-Cre virus. J.R. and B.M.P. provided S11-NGFp peptide and Ad5.FOXA3.GATA4.HNF1A.HNF4A for targeting of myofibroblasts in vivo. A.B. and M.A.-B. analyzed microarray data. A.S. provided lentiviral vector for overexpression of 4TFs. R.F.S. provided LratCre mice. T.L., R.F.S., M.P.M., H.R.S., and T.C. contributed to conceptual evaluation of the project. ACKNOWLEDGMENTS

Histology, Immunohistochemistry, and Immunofluorescence Liver tissues were fixed with 4% formalin, embedded in paraffin, and cut into 5-mm-thick sections for histological and immunohistochemical analysis. For Sirius red staining, following deparaffinization, the sections were stained with Picro-Sirius red solution (0.1% direct red 80 plus 0.1% fast green dis-

The study was financed by the Deutsche Forschungsgemeinschaft (DFG SH640/1-2, DFG EXC 62/2, DFG 188/9-1, and SFB-738), Gilead Sciences International Research Scholars Program in Liver Diseases, and the Bundesministerium fu¨r Bildung und Forschung (Biodisc 6, START-MSC II). G.S. and Z.D.

Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc. 11

Please cite this article in press as: Song et al., Direct Reprogramming of Hepatic Myofibroblasts into Hepatocytes In Vivo Attenuates Liver Fibrosis, Cell Stem Cell (2016), http://dx.doi.org/10.1016/j.stem.2016.01.010

are sponsored by the China Scholarship Council. The authors wish to thank Xiaokun Liu for technical support. Received: August 16, 2015 Revised: December 15, 2015 Accepted: January 15, 2016 Published: February 25, 2016 REFERENCES Cahan, P., Li, H., Morris, S.A., Lummertz da Rocha, E., Daley, G.Q., and Collins, J.J. (2014). CellNet: network biology applied to stem cell engineering. Cell 158, 903–915. Du, Y., Wang, J., Jia, J., Song, N., Xiang, C., Xu, J., Hou, Z., Su, X., Liu, B., Jiang, T., et al. (2014). Human hepatocytes with drug metabolic function induced from fibroblasts by lineage reprogramming. Cell Stem Cell 14, 394–403. Friedman, S.L., Sheppard, D., Duffield, J.S., and Violette, S. (2013). Therapy for fibrotic diseases: nearing the starting line. Sci. Transl. Med. 5, 167sr1. Grande, A., Sumiyoshi, K., Lo´pez-Jua´rez, A., Howard, J., Sakthivel, B., Aronow, B., Campbell, K., and Nakafuku, M. (2013). Environmental impact on direct neuronal reprogramming in vivo in the adult brain. Nat. Commun. 4, 2373. Guo, Z., Zhang, L., Wu, Z., Chen, Y., Wang, F., and Chen, G. (2014). In vivo direct reprogramming of reactive glial cells into functional neurons after brain injury and in an Alzheimer’s disease model. Cell Stem Cell 14, 188–202. Han, D.W., Tapia, N., Hermann, A., Hemmer, K., Ho¨ing, S., Arau´zo-Bravo, M.J., Zaehres, H., Wu, G., Frank, S., Moritz, S., et al. (2012). Direct reprogramming of fibroblasts into neural stem cells by defined factors. Cell Stem Cell 10, 465–472. Huang, P., He, Z., Ji, S., Sun, H., Xiang, D., Liu, C., Hu, Y., Wang, X., and Hui, L. (2011). Induction of functional hepatocyte-like cells from mouse fibroblasts by defined factors. Nature 475, 386–389. Huang, P., Zhang, L., Gao, Y., He, Z., Yao, D., Wu, Z., Cen, J., Chen, X., Liu, C., Hu, Y., et al. (2014). Direct reprogramming of human fibroblasts to functional and expandable hepatocytes. Cell Stem Cell 14, 370–384. Iacob, R., Ru¨drich, U., Rothe, M., Kirsch, S., Maasoumy, B., Narain, N., Verfaillie, C.M., Sancho-Bru, P., Iken, M., Popescu, I., et al. (2011). Induction of a mature hepatocyte phenotype in adult liver derived progenitor cells by ectopic expression of transcription factors. Stem Cell Res. (Amst.) 6, 251–261.

cells as dominant contributors to liver fibrosis independent of its aetiology. Nat. Commun. 4, 2823. Morris, S.A., Cahan, P., Li, H., Zhao, A.M., San Roman, A.K., Shivdasani, R.A., Collins, J.J., and Daley, G.Q. (2014). Dissecting engineered cell types and enhancing cell fate conversion via CellNet. Cell 158, 889–902. Muzumdar, M.D., Tasic, B., Miyamichi, K., Li, L., and Luo, L. (2007). A global double-fluorescent Cre reporter mouse. Genesis 45, 593–605. Nishikawa, T., Bell, A., Brooks, J.M., Setoyama, K., Melis, M., Han, B., Fukumitsu, K., Handa, K., Tian, J., Kaestner, K.H., et al. (2015). Resetting the transcription factor network reverses terminal chronic hepatic failure. J. Clin. Invest. 125, 1533–1544. Puche, J.E., Lee, Y.A., Jiao, J., Aloman, C., Fiel, M.I., Mun˜oz, U., Kraus, T., Lee, T., Yee, H.F., Jr., and Friedman, S.L. (2013). A novel murine model to deplete hepatic stellate cells uncovers their role in amplifying liver damage in mice. Hepatology 57, 339–350. Qian, L., Huang, Y., Spencer, C.I., Foley, A., Vedantham, V., Liu, L., Conway, S.J., Fu, J.D., and Srivastava, D. (2012). In vivo reprogramming of murine cardiac fibroblasts into induced cardiomyocytes. Nature 485, 593–598. Reetz, J., Genz, B., Meier, C., Kowtharapu, B.S., Timm, F., Vollmar, B., Herchenro¨der, O., Abshagen, K., and Pu¨tzer, B.M. (2013). Development of adenoviral delivery systems to target hepatic stellate cells in vivo. PLoS ONE 8, e67091. Ring, K.L., Tong, L.M., Balestra, M.E., Javier, R., Andrews-Zwilling, Y., Li, G., Walker, D., Zhang, W.R., Kreitzer, A.C., and Huang, Y. (2012). Direct reprogramming of mouse and human fibroblasts into multipotent neural stem cells with a single factor. Cell Stem Cell 11, 100–109. Schambach, A., Mueller, D., Galla, M., Verstegen, M.M., Wagemaker, G., Loew, R., Baum, C., and Bohne, J. (2006). Overcoming promoter competition in packaging cells improves production of self-inactivating retroviral vectors. Gene Ther. 13, 1524–1533. Sekiya, S., and Suzuki, A. (2011). Direct conversion of mouse fibroblasts to hepatocyte-like cells by defined factors. Nature 475, 390–393. Sharma, A.D., Narain, N., Ha¨ndel, E.M., Iken, M., Singhal, N., Cathomen, T., Manns, M.P., Scho¨ler, H.R., Ott, M., and Cantz, T. (2011). MicroRNA-221 regulates FAS-induced fulminant liver failure. Hepatology 53, 1651–1661. Son, E.Y., Ichida, J.K., Wainger, B.J., Toma, J.S., Rafuse, V.F., Woolf, C.J., and Eggan, K. (2011). Conversion of mouse and human fibroblasts into functional spinal motor neurons. Cell Stem Cell 9, 205–218.

Ieda, M., Fu, J.D., Delgado-Olguin, P., Vedantham, V., Hayashi, Y., Bruneau, B.G., and Srivastava, D. (2010). Direct reprogramming of fibroblasts into functional cardiomyocytes by defined factors. Cell 142, 375–386.

Song, K., Nam, Y.J., Luo, X., Qi, X., Tan, W., Huang, G.N., Acharya, A., Smith, C.L., Tallquist, M.D., Neilson, E.G., et al. (2012). Heart repair by reprogramming non-myocytes with cardiac transcription factors. Nature 485, 599–604.

Kendall, T.J., Hennedige, S., Aucott, R.L., Hartland, S.N., Vernon, M.A., Benyon, R.C., and Iredale, J.P. (2009). p75 Neurotrophin receptor signaling regulates hepatic myofibroblast proliferation and apoptosis in recovery from rodent liver fibrosis. Hepatology 49, 901–910.

Su, Z., Niu, W., Liu, M.L., Zou, Y., and Zhang, C.L. (2014). In vivo conversion of astrocytes to neurons in the injured adult spinal cord. Nat. Commun. 5, 3338.

Kim, J., Efe, J.A., Zhu, S., Talantova, M., Yuan, X., Wang, S., Lipton, S.A., Zhang, K., and Ding, S. (2011). Direct reprogramming of mouse fibroblasts to neural progenitors. Proc. Natl. Acad. Sci. USA 108, 7838–7843. Lujan, E., Chanda, S., Ahlenius, H., Su¨dhof, T.C., and Wernig, M. (2012). Direct conversion of mouse fibroblasts to self-renewing, tripotent neural precursor cells. Proc. Natl. Acad. Sci. USA 109, 2527–2532. Maetzig, T., Brugman, M.H., Bartels, S., Heinz, N., Kustikova, O.S., Modlich, U., Li, Z., Galla, M., Schiedlmeier, B., Schambach, A., and Baum, C. (2011). Polyclonal fluctuation of lentiviral vector-transduced and expanded murine hematopoietic stem cells. Blood 117, 3053–3064. Malato, Y., Naqvi, S., Schu¨rmann, N., Ng, R., Wang, B., Zape, J., Kay, M.A., Grimm, D., and Willenbring, H. (2011). Fate tracing of mature hepatocytes in mouse liver homeostasis and regeneration. J. Clin. Invest. 121, 4850–4860. Mederacke, I., Hsu, C.C., Troeger, J.S., Huebener, P., Mu, X., Dapito, D.H., Pradere, J.P., and Schwabe, R.F. (2013). Fate tracing reveals hepatic stellate

12 Cell Stem Cell 18, 1–12, May 5, 2016 ª2016 Elsevier Inc.

Takahashi, K., and Yamanaka, S. (2006). Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell 126, 663–676. Takahashi, K., Tanabe, K., Ohnuki, M., Narita, M., Ichisaka, T., Tomoda, K., and Yamanaka, S. (2007). Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell 131, 861–872. Thier, M., Wo¨rsdo¨rfer, P., Lakes, Y.B., Gorris, R., Herms, S., Opitz, T., Seiferling, D., Quandel, T., Hoffmann, P., No¨then, M.M., et al. (2012). Direct conversion of fibroblasts into stably expandable neural stem cells. Cell Stem Cell 10, 473–479. Torper, O., Pfisterer, U., Wolf, D.A., Pereira, M., Lau, S., Jakobsson, J., Bjo¨rklund, A., Grealish, S., and Parmar, M. (2013). Generation of induced neurons via direct conversion in vivo. Proc. Natl. Acad. Sci. USA 110, 7038–7043. Vierbuchen, T., Ostermeier, A., Pang, Z.P., Kokubu, Y., Su¨dhof, T.C., and Wernig, M. (2010). Direct conversion of fibroblasts to functional neurons by defined factors. Nature 463, 1035–1041.