1 Two-signal requirement for growth-promoting function of Yap in 2 hepatocytes 3

4 Tian Su1,3, Tanya Bondar1,3, Xu Zhou1, Cuiling Zhang1, Hang He2 and Ruslan Medzhitov1

5

6 1Howard Hughes Medical Institute and Department of Immunobiology, Yale University School 7 of Medicine, New Haven, Connecticut, USA

8 2Peking-Yale Joint Center for Plant Molecular Genetics and Agro-Biotechnology, National 9 Laboratory of Protein Engineering and Plant Genetic Engineering, College of Life Sciences, 10 Peking University, Beijing, China

11 3These authors contributed equally to this work 12

13

14

15

16

17

18

19

20

21

22 Correspondence and requests for materials should be addressed to:

23 Ruslan Medzhitov 24 Department of Immunobiology 25 Yale University School of Medicine 26 PO Box 208011 27 300 Cedar Street 28 New Haven, CT 29 Phone: (203) 785-7541 30 e-mail: [email protected] 31

1

32 Abstract

33 The transcriptional coactivator Yes-associated protein (Yap) promotes proliferation and inhibits 34 , suggesting that Yap functions as an oncogene. Most oncogenes, however, require a 35 combination of at least two signals to promote proliferation. Here we present evidence that Yap 36 activation is insufficient to promote growth in the otherwise normal tissue. Using a mosaic mouse 37 model, we demonstrate that Yap overexpression in a fraction of hepatocytes does not lead to their 38 clonal expansion, as proliferation is counterbalanced by increased apoptosis. To shift the activity of Yap 39 towards growth, a second signal provided by tissue damage or inflammation is required. In response to 40 liver injury, Yap drives clonal expansion, suppresses hepatocyte differentiation and promotes a 41 progenitor phenotype. These results suggest that Yap activation is insufficient to promote growth in the 42 absence of a second signal thus coordinating tissue homeostasis and repair.

43

44 Introduction

45 The transcriptional coactivator Yap is an evolutionarily conserved mediator of cell fate decisions such 46 as proliferation, differentiation, and survival1. Together with a related protein Taz, Yap is the central 47 target of the Hippo pathway, a growth-suppressive network of kinase complexes that inactivates Yap 48 and Taz2. The Hippo pathway has been implicated in organ size control, as overexpression of Yap or 49 inactivation of the Hippo components leads to tissue overgrowth and organomegaly in Drosophila and 50 mouse models3-5. These effects are largely mediated by Yap-TEAD complexes that activate 51 transcription of genes promoting cell proliferation and inhibiting apoptosis6-9. Yap is interconnected with 52 RTK10, GPCR11, PI3K12, Wnt13 and TGF-beta14-16 signaling, and Yap co-regulates transcription by 53 interacting with Smads16, 17, TCF/LEF18, Tbx517, Runx219, FoxO120, and p7321, among others22.

54 The intestinal epithelium in both Drosophila23-26 and mice does not depend on Yap for homeostatic 55 tissue turnover27, but it does respond very strongly to Yap overexpression28 and requires Yap for tissue 56 repair29. Thus, in some tissues Yap may be dispensable for homeostasis but required specifically in 57 response to injury. The liver is one of the organs most responsive to excessive Yap activity. Transgenic 58 overexpression of Yap or inactivation of its upstream negative regulators causes a dramatic increase in 59 liver size, hepatocyte proliferation, progenitor cell expansion, and tumorigenesis4, 5, 30-34. In contrast, 60 deletion of Yap in the liver leads to defects in bile duct formation but no apparent defects in hepatocyte 61 number and function35, suggesting that Yap may be dispensable for hepatocyte homeostasis. However, 62 whether its function is required for hepatocyte homeostasis and response to injury remains to be 63 established36.

64 2

65 Yap activation promotes proliferation, survival, stemness, and tumor development in mouse models4, 5, 66 22, and is commonly observed in human cancers37, 38. Collectively, these data suggest that 67 hyperactivation of Yap abrogates organ size control mechanisms and drives tumorigenesis in a 68 seemingly unrestrained fashion. However, growth-promoting pathways are normally safeguarded by 69 tumor suppressive mechanisms39. For example, c-myc hyperactivation sensitizes cells to apoptosis40, 70 oncogenic Ras induces senescence41, and overexpression of Bcl-2 inhibits cell proliferation42. Whether 71 or not Yap activity is subject to a similar tumor suppressive regulation is currently unclear. While Yap is 72 known to interact with p73 and promote apoptosis in response to DNA damage in vitro21, 43, there is no 73 evidence that Yap can induce apoptosis in vivo.

74 Control of cell fate decisions at the tissue level is poorly understood, but it is likely to involve cell 75 contact-dependent regulation. Yap activity is regulated by adherens and tight junctions, cell polarity 76 complexes, and the actin cytoskeleton44. At high cell densities Yap is either directly recruited to 77 intercellular junctions or undergoes phosphorylation and cytosolic retention via the Hippo pathway, 78 which itself is also regulated in a cell contact-dependent manner. This feature makes Yap competent to 79 direct cell fate decisions depending on cell density and architecture. Thus, cell environment may be a 80 major determinant of the outcome of Yap activation. Moreover, evidence from Drosophila45 and 81 mammalian cell culture46 suggest that the outcome of Yap activation depends on Yap activity in 82 neighboring cells. The regulation of Yap activity by such mechanism has not been characterized in 83 mammalian tissues in vivo.

84 Here, we describe a mosaic model of Yap activation in the mouse liver, where a high level of Yap 85 expression is induced in a fraction of hepatocytes surrounded by normal tissue. We present evidence 86 that a high level of Yap in this context is insufficient to drive clonal expansion, as increased proliferation 87 of Yap-overexpressing hepatocytes is balanced by their increased susceptibility to apoptosis. Yap- 88 mediated clonal expansion requires a second signal provided by disruption of tissue homeostasis, such 89 as injury or inflammation. Upon returning to homeostasis, excessive Yap-overexpressing cells are 90 eliminated by apoptosis.

91

92 Results

93 Generation of mosaic Yap mouse model

94 To study the role of cell environment in the regulation of the mammalian Hippo pathway, we generated 95 a conditional mosaic mouse model of Yap overexpression (YapKI mice). We used Yap1 mutant S112A 96 which has enhanced nuclear localization and has been previously shown to cause a several-fold 97 increase in liver size when overexpressed47, 48 in the entire organ. A targeting construct driving 3

98 expression of the Yap1S112A-IRES-GFP downstream of a floxed transcriptional STOP cassete was 99 knocked into the Gt(ROSA)26Sor (Rosa26) locus. Upon expression of Cre recombinase, the 100 transcriptional STOP cassette is removed, allowing expression of Yap together with GFP (YapKI mice; 101 Figure 1 – figure supplement 1).

102 We used the Rosa26-IRES-GFP targeting construct which is expressed bimodally49 (likely due to 103 epigenetic regulation of the locus), generating GFPlow and GFPhigh populations in all cells of 104 hematopoietic lineages49 and solid tissues tested (liver, , muscle, intestine; Figure 1- figure 105 supplement 2 and data not shown). Both populations have a recombined STOP cassette, but the gene 106 of interest is expressed above physiological level only in GFPhigh cells. GFPlow cells serve as an internal 107 control population of the identical genetic background for the GFPhigh cells. In the absence of selective 108 pressure, the ratio of GFPlow to GFPhigh cells remains constant throughout life. Thus, crossing YapKI 109 mice to Albumin-Cre (Alb-Cre) yields mosaic Yap expression in the hepatic lineage, where individual 110 cells overexpressing Yap can be traced by GFP fluorescence (YapKIAlb-Cre mice). Similar to the 111 previously reported Rosa26-DNp53 mice49, the proportion of GFPlow to GFPhigh cells remained stable in 112 the blood of YapKICreER mice upon tamoxifen induction for many months (Figure 1 – figure supplement 113 2B-D). In the colons of YapKIVillin-Cre mice, crypts were either entirely GFPhigh or GFPlow (GFPlow 114 appearing as GFP- due to limited sensitivity of GFP detection by tissue immunostaining). There were 115 no crypts that have mixed populations, as would be expected if cells randomly switched expression 116 level (Figure 1 – figure supplement 2E). Collectively, these data illustrate stable inheritance of the 117 Rosa26 allele expression level.

118 First we verified the correlation of Yap and GFP expression by immunofluorescence staining of the 119 YapKIAlb-Cre liver sections. Indeed, Yap and GFP in hepatocytes displayed mosaic and overlapping 120 expression pattern, with cytosolic as well as nuclear Yap localization (Figure 1A). Utilizing the GFP 121 marker, we tested the recombination efficiency by flow cytometry. More than 90% of hepatocytes were 122 GFP-positive, indicating high deletion efficiency of the STOP cassette (Figure 1B). Recombination was 123 also confirmed by genomic qPCR (Figure 1 – figure supplement 3). The hepatocytes formed 2 distinct 124 populations differing in GFP brightness. To determine the level of Yap activity in these cells, we sorted 125 each of these populations by FACS. Since GFP fluorescence varied 10-fold within the GFPlow 126 population, we further subdivided it into GFPlow and GFPmed-low populations (Figure 1B). The exogenous 127 transcripts of Yap are higly expressed only in the GFPhigh population, but not in the GFPlow and GFPmed- 128 low populations. The small amount of the exogenous Yap mRNA in the GFPlow and GFPmed-low 129 populations is insignificant due to a potential negative feedback mechanism that reduces the 130 endogenous Yap expression, resulting in a similar level of total Yap mRNA compared to that of wild- 131 type (WT) (Figure 1 – figure supplement 4). Additionally, CTGF, a known Yap downstream target 132 gene9, was significantly upregulated only in the GFPhigh population, whereas its expression remained 4

133 unchanged in GFPlow and GFPmed-low populations, indicating normal levels of Yap activity in these cells 134 (Figure 1C). Thus, GFP and Yap protein levels correlate in YapKIAlb-Cre mice, and Yap is overexpressed 135 and has higher activity specifically in the GFPhigh hepatocytes (to which we will refer as Yaphigh cells).

136

137 Yap activation in hepatocytes does not lead to clonal expansion at steady state

138 The first question we adressed is whether liver size is affected by mosaic Yap overexpression. Previous 139 studies showed that hyperactivation of Yap in the entire liver leads to rapid and progressive 140 hepatomegaly: liver weight increases 5-fold in 1 month in Yap transgenic mice4, 5, and 4-fold by 3 141 months of age in Stk3, Stk4 (Mst1/2) double null mice31, 50 (where Yap is activated due to an absence of 142 upstream negative regulators). However, the liver/body weight ratio of the mosaic YapKIAlb-Cre mice was 143 only slightly elevated (Figure 2A). Hepatocyte numbers (Figure 2B) and liver morphology (Figure 2C) 144 also remained normal. Moreover, Yaphigh cells did not have a growth advantage over the WT neighbors, 145 since neither the average size of Yaphigh cells clusters in the liver sections (Figure 2 – figure supplement 146 1), nor the average frequency of Yaphigh hepatocytes increased over time (Figure 2D).

147 These results were unexpected, considering the dramatic increase in hepatocyte proliferation and 148 survival driven by Yap in the previously published whole-liver transgenic models4, 5. We therefore tested 149 whether Yap influenced proliferation and apoptosis in YapKIAlb-Cre liver. Indeed, more Yaphigh 150 hepatocytes incorporated BrdU than their GFPlow neighbors or hepatocytes in WT livers (Figure 2E). 151 Meanwhile, TUNEL staining indicated more apoptosis in the YapKIAlb-Cre e livers (Figure 2F). It is 152 noteworthy that only a minor fraction of the Yaphigh cells proliferated, and over 95% remained quiescent 153 throughout the 3-day BrdU labeling period. These data suggest that increased Yap levels sensitize 154 hepatocytes to both mitogenic and apoptotic signals but are insufficient to drive cell expansion, 155 resulting in slightly increased cell turnover without net growth advantage.

156 This insufficiency was not due to an overall lack of Yap activation, as Yap overexpression induced a 157 significant change in in our model. Sorted WT, GFPlow and GFPhigh hepatocytes were 158 compared by RNA-sequencing analysis. Gene expression of GFPlow cells isolated from YapKIAlb-Cre 159 livers was similar to that of the WT hepatocytes, whereas GFPhigh hepatocytes showed robust 160 transcriptional changes: 821 genes were significantly up-regulated (p ≤ 0.005 and fold change ≥ 2) in 161 Yaphigh hepatocytes compared to WT hepatocytes (Figure 2 – Figure Supplement 2A). 37% of these 162 genes overlapped with genes upregulated in the previously characterized Yap overexpression model5 163 (Figure 2 – Figure Supplements 2B, 3A-B), further validating our model. Dong et al5 used an ApoE 164 which is not restricted to hepatocytes, and assayed the whole liver that includes stromal, 165 endothelial and immune cells with microarray, whereas we assayed sorted hepatocytes in a model with 166 hepatocyte-specific Yap induction with RNA-sequencing. Thus, a complete overlap in gene profile 5

167 would not be expected. Yap-induced genes showed an enrichment in cell adhesion, extracellular matrix 168 (ECM), and cell-cell junction functional groups (Figure 2 – Figure Supplement 2C), according to DAVID 169 functional cluster analysis51, 52. In contrast, the category was not significantly enriched, likely 170 because only a small fraction of all Yaphigh hepatocytes were proliferating, as evidenced by BrdU 171 labeling.

172 It is well known that pro-growth effects of Yap overexpression are mediated through TEAD6-9, and the 173 dominant-negative TEAD mutant largely suppresses effects of Yap transgene in the liver53. To evaluate 174 how many of the Yap-induced genes are potentially regulated directly by Yap-TEAD complexes, we 175 compared these genes to genes with TEAD binding sites identified previously by ChIP-seq54. Indeed, 176 over 25% of all Yap-activated genes are associated with TEAD compared to only 9% of all genes in the 177 genome. Moreover, a large fraction of genes induced in Yaphigh hepatocytes and related to cell 178 adhesion (33%), ECM (36%), and cell-cell junctions (36%) contain TEAD binding regions (Figure 2 – 179 Figure Supplement 2D). The gene expression data shown here suggest that inability of Yap to drive cell 180 expansion in our model is not due to lack of transcriptionally competent Yap-TEAD complexes.

181 To test whether mosaic Yap activation can lead to clonal expansion in previously characterized mouse 182 models, we injected Stk4-/-;Stk3flox/floxmice with a low dose of adenovirus expressing Cre and GFP, and 183 compared patterns of BrdU incorporation and GFP+ hepatocytes 5 and 35 days later. At both time 184 points, we could only observe single GFP+ hepatocytes, BrdU negative (Figure 2- Figure Supplement 185 4A). Semi-quantitative PCR of the recombined Stk3 (∆Mst2) allele also did not show an increase at 35 186 days, and liver/body mass ratio remained normal (Figure 2- Figure Supplement 4B,C).

187 Taken together, these data suggest that clonal activation of Yap in hepatocytes is not sufficient to 188 overrule growth restricting mechanisms at steady state, but is sufficient to regulate transcription of a 189 large number of genes related to extracellular environment.

190

191 Yap promotes hepatocyte proliferation in response to injury

192 Previous studies demonstrated that Yap is activated during liver injury35, 55, 56 and protects mice from 193 oxidative stress-induced damage57. We therefore tested whether Yap induction provides selective

194 advantage during liver injury. We chose carbon tetrachloride (CCl4) as a damaging agent due to its 195 localized effects around the central vein (CV)58.

196 We first characterized the kinetics of liver damage after a single injection of CCl4. Blood ALT levels in 197 YapKIAlb-Cre and WT mice rose to similar levels, indicating comparable hepatocyte damage (Figure 3 – 198 Figure Supplement 13). By day 4 the acute damage phase ceased completely (Figure 3 – Figure

6

199 Supplement 1) and very few hepatocytes incorporated BrdU (Figure 3 – Figure Supplement 2). We 200 therefore chose day 4 to study the outcome of the tissue repair response.

high 201 In response to CCl4 damage Yap cells expanded dramatically within 4 days in the pericentral area, 202 comprising over 80% of all hepatocytes in this zone (Figure 3A,B). These pericentral Yaphigh cells low 203 incorporated significantly more BrdU during days 1-3 after CCl4 administration compared to Yap 204 hepatocytes in the same location (Figure 3C). In contrast, Yaphigh cells proliferated much less in the CV- 205 distal areas than in the pericentral area (Figure 3C). In fact, the BrdU index outside the pericentral 206 zone did not differ between Yaphigh and Yaplow hepatocytes within the same liver. Even though Yaphigh 207 cells expanded locally at the injury sites, the effect was strong enough to significantly increase total 208 percentage of Yaphigh cells in the liver (Figure 3D). Nevertheless, this expansion of Yaphigh cells did not 209 overrule normal liver size control, as the total number of hepatocytes was similar in YapKIAlb-Cre and WT 210 mice (Figure 3 – figure supplement 3). Liver/body weight ratios in both groups also returned to steady

211 state levels within a month after undergoing a similar increase in response to CCl4, (likely due to the 212 infiltration of immune cells during early phase of tissue repair) (Figure 3 – figure supplements 3-5). 213 These results indicate that Yaphigh cells undergo expansion in response to local tissue injury signals.

214 To investigate whether Yap activity in hepatocytes is required for tissue repair, we generated Yap 215 conditional knockout mice and crossed them to the AlbCre strain to obtain hepatocyte-specific Yap 216 knockout Yap1flox/flox;AlbCre (YapCKOAlb-Cre) mice (Figure 4 – figure supplements 1,2). These mice were

217 treated with CCl4 using the same protocol as described above. A similar increase in blood ALT level in 218 YapCKOAlb-Cre and WT mice indicated comparable liver damage in the first two days (Figure 4 – figure 219 supplement 3). By day 4, ALT levels returned to baseline (Figure 4 – figure supplement 3), normal 220 pericentral morphology was restored, and expression of the CV-proximal marker glutamine synthase 221 was reestablished in both groups (Figure 4A). However, YapCKOAlb-Cre hepatocytes proliferated 222 significantly less than their WT counterparts throughout the repair phase, as shown by BrdU labeling on 223 days 1- 3 (Figure 4B,C). Consistent with the defective regeneration, YapCKOAlb-Cre livers displayed 224 abnormal collagen deposition around the CV (Figure 4D). Interestingly, the steady state liver/body 225 mass ratio (Figure 4 – figure supplement 4) and BrdU incorporation rate (Figure 4 – figure supplement 226 5) were higher in the YapCKOAlb-Cre mice than in the WT mice, suggesting that Yap is not required for 227 homeostatic hepatocyte proliferation.

228 Thus, while Yap is not required for liver development and hepatocyte proliferation under normal 229 conditions, Yap function is required for hepatocyte proliferation during tissue repair. Additionally, 230 hepatocytes with a higher level of Yap have growth advantage during the early repair phase.

231

7

232 Yap promotes hepatocyte proliferation in response to inflammation

233 Given differential effects of Yap on hepatocyte proliferation during homeostasis versus injury, we 234 wondered how growth control mechanisms might differ under these conditions. Tissue injury is 235 accompanied by inflammation, which can promote proliferation by mechanisms distinct from 236 homeostasis. To test whether Yap cooperates with inflammatory signals to drive cell cycle progression, 237 we measured hepatocyte BrdU incorporation in YapKIAlb-Cre mice after 6 continuous days of 238 intraperitoneal injections of LPS. This chronic LPS treatment induced a massive expansion of Yaphigh

239 hepatocytes, similar to what was seen after CCl4 treatment (Figure 3D). Consistently, the BrdU 240 incorporation index indicated that Yaphigh hepatocytes proliferated significantly more than hepatocytes 241 in the WT livers (Figure 5A, B). YapKIAlb-Cre livers displayed significant hyperplasia, as visualized by 242 increased hepatocyte density (Figure 5C) and measured by total number of hepatocytes per liver 243 (Figure 5D).

244 We then tested whether IL-6, an inflammatory cytokine induced by LPS and by liver injury, can 245 cooperate with Yap in promoting hepatocyte growth. When IL-6 was produced by LLC cells growing as 246 subcutaneous tumors in YapKIAlb-Cre mice, it caused expansion of Yaphigh hepatocytes in the livers of 247 mice that had the highest levels of systemic IL-6 (Figure 5 – Figure Supplement 1A, B), suggesting that 248 IL-6 may cooperate with Yap to promote hepatocyte proliferation.

249 Yap upregulates progenitor markers and represses hepatocyte differentiation in 250 response to injury

251 4 days after CCl4 treatment in WT mice, normal liver tissue morphology was restored around the CV. In 252 contrast, we observed high cell density, strong cytoplasmic eosinophilia, round to ovoid nuclei, and Alb-Cre 253 increased nuclear/cytoplasmic ratio in the pericentral zones of livers from CCl4 treated YapKI mice 254 (Figure 6A). These morphological changes are often seen in hepatocellular carcinomas (HCC) and are 255 associated with dedifferentiation59, 60. Immunofluorescent staining indicated that these cells were GFP+ 256 and therefore originated from Yaphigh hepatocytes (Figure 3A). To determine whether Yap induction 257 affects differentiation status of hepatocytes, we compared the gene expression of hepatocytes sorted Alb-Cre 258 from YapKI mice at steady state and 4 days after CCl4 damage. The majority of liver-specific 259 transcripts61 (which roughly represent hepatocyte differentiation state) were strongly repressed in high low 260 Yap hepatocytes from CCl4-treated mice, whereas WT and Yap hepatocytes showed minimal 261 alterations and no bias towards decreased expression (Figure 6B). Immunofluorescent staining and 262 Western blot for hepatocyte-specific gene glutamine synthase also showed a strong downregulation of Alb-Cre 263 expression in YapKI livers after CCl4 damage (Figure 6C,D). Downregulation of other hepatocyte- 264 specific genes (acute phase protein Orm1 and urea cycle component Asl) was also confirmed by qPCR 265 (Figure 6 – figure supplement 1). To further verify RNA-sequencing data, and evaluate levels of Yap 8

266 activation in our model we performed qPCR analysis of the genes previously shown to be induced in 267 Yap-Tg liver by Pan’s group5 and found comparable effects (Figure 6 – figure supplement 1).

268 The transcriptional repression was not global: a number of genes expressed predominantly during 269 embryogenesis61 were induced in Yaphigh hepatocytes, and this effect was strongly enhanced by injury 270 (Figure 6E). In addition, several hepatocyte progenitor markers were upregulated in Yaphigh cells after 271 injury, and some of them were already elevated in the steady state (Figure 6F). FACS staining of liver - - + 272 cells isolated on day 4 post-CCl4 confirmed expansion of CD45 CD31 EpCam hepatic progenitors in 273 YapKI;AlbCre livers (Figure 6G,H). These data are in line with the recent report by Yimlamai et al. 62.

274 Altogether, these data suggest that Yap cooperates with injury signals to repress hepatocyte 275 differentiation and promote a progenitor phenotype.

276

277 Excessive Yaphigh cells are eliminated during repair resolution phase

high 278 The expansion of Yap cells induced by CCl4 did not persist, and by day 14 after the treatment, the 279 proportion of Yaphigh cells per liver returned to the average value of the untreated group (Figure 3D). 280 Yaphigh cells formed a ring of only 1-2 cell layers around CVs, in contrast to multiple layers observed on 281 day 4 (Figure 3A), indicating that the Yaphigh cell pool underwent contraction. TUNEL staining suggested 282 that the contraction was mediated by apoptosis: the number of TUNEL positive cells was markedly Alb-Cre 283 elevated in the YapKI livers specifically around the CV on day 4 after CCl4 treatment (Figure 7A). 284 Accordingly, RNA sequencing revealed a number of pro-apoptotic genes (e.g., Bok, Bax, NGFRAP1, 285 and Tnfrsf10b (Dr5)) (Figure 7B) upregulated as a consequence of Yap overexpression and further 286 induced after damage. We verified by qPCR that Dr5 gene expression was strongly upregulated in the high high 287 Yap cells after CCl4 (Figure 7C,D). Indeed, Yap cells sorted from CCl4-treated livers were more 288 sensitive to killing mediated by TRAIL in vitro, compared to WT or GFPlow counterparts (Figure 7E), 289 suggesting that Yap activation sensitizes hepatocytes to Dr5-mediated cell death. This higher sensitivity 290 to cell death was induced by a combination of injury and Yap overexpression, as it was not observed high 291 neither in Yap hepatocytes isolated from untreated mice nor in WT hepatocytes from CCl4 livers.

292 In summary, the expanded population of Yaphigh cells contracts to baseline at later stages of the tissue 293 injury response. This process is likely mediated, in part, by sensitization of Yaphigh cells to TRAIL- 294 induced apoptosis via upregulation of Dr5.

295

9

296 Discussion

297 Yap has emerged as a central mediator of signals promoting proliferation, survival, and stemness22. Not 298 surprisingly, an expanding number of pathways restricting the pro-growth potential of Yap have been 299 identified, including the Hippo pathway, the actin cytoskeleton and cell junction complexes44, C/EBPa, 300 and Trib263. Despite these multifactorial control mechanisms, mere overexpression of Yap in the whole 301 liver in mouse models leads to hyperplasia, organomegaly, dedifferentiation, and cancer4, 5, 47, raising a 302 question as to whether some layer of homeostatic growth control may be lacking in such models. 303 Organ-wide Yap activation in the liver does occur during embryonic development64 and adaptive 304 hepatomegaly32. Interestingly, these are also examples of physiologically relevant settings of liver size 305 increase. Transgenic Yap overexpression in the entire liver may thus be viewed as a model of these 306 physiological scenarios. In contrast, during response to a localized injury, only the cells within the 307 damage site need to proliferate, and Yap activation in this case should conform to the physiological 308 organ size limits. Our study suggests that under these circumstances, activation of a pro-growth gene 309 program by Yap requires an additional local signal (e.g., growth factor, cytokine, or release of contact 310 inhibition) provided by the injury. Our preliminary data implicate IL-6 as at least one of such signals 311 (Figure 5 – figure supplement 1). Proto-oncogenes typically require two signals in order to induce 312 proliferation. In case of Yap, which acts as a sensor of cell density, its overexpression mimics a 313 condition in which a cell has lost its contacts with the neighbors. This in itself should not induce 314 proliferation unless cell-extrinsic signals of tissue damage such as inflammatory cytokines are also 315 present. The mechanism of cooperation between activation of Yap and injury/inflammatory signals may 316 occur at the level of gene expression, as many proliferation genes have TEAD as well as NFkB binding 317 sites (Figure 5 – figure supplement 2). While it is also possible that cooperation is mediated by 318 increased Yap stability or recruitment of TEAD, this possibility is less likely as we observed induction of 319 high number of TEAD-dependent genes in Yap-overexpressing hepatocytes whereas proliferation 320 genes were not induced.

321 Our data suggest that the tissue environment imposes a selective pressure on cells that activate Yap. 322 The mosaic model of Yap induction described here reveals selective pressures that determine the fate 323 of Yaphigh hepatocytes: no growth advantage at steady state, positive selection during early stages of 324 tissue injury, and negative selection at later stages when injury induced signals subside. The 325 mechanisms opposing clonal expansion of Yaphigh hepatocytes under homeostatic conditions may have 326 many components, but one implicated by our data is apoptosis. Pro-apoptotic effects of Yap have been 327 extensively documented in vitro21, 43, 65, and increased apoptosis has been reported in Mst1/2-deficient 328 mouse livers31, but to our knowledge there has been no evidence of Yap promoting cell death in vivo. 329 We show that Yap activation in the mouse liver leads to increased apoptosis in vivo and sensitization to

10

330 TRAIL measured ex vivo. The induction of proliferation or apoptosis by Yap is modulated by tissue 331 damage and inflammatory signals. This is analogous to the well characterized functions of c-myc, which 332 is also known induce proliferation or apoptosis, depending on growth factor availability40, 66.These 333 examples may represent a general principle of mitogenic pathway design, aimed at elimination of 334 aberrant cells.

335 YapKIAlb-Cremice display many features described in other in vivo models of Yap induction, including 336 increased proliferation, dedifferentiation and activation of TEAD target genes. However, the unexpected 337 phenotype unique to YapKIAlb-Cremice is that the pro-growth effects of Yap are only manifested in a state 338 of altered homeostasis, such as tissue injury or inflammation. These differences are unlikely to be due 339 to an insufficient level of Yap expression, as a large number of genes are highly induced by Yap under 340 steady state conditions in YapKIAlb-Cremice. One relevant difference may be mosaic versus tissue-wide 341 Yap activation. We show that mosaic deletion of Mst2 in Mst1-/- background does not lead to clonal 342 expansion or liver size increase. Furthermore, in another mouse model recently reported by Camargo 343 and colleagues62 while this manuscript was under consideration, mosaic Yap activation in hepatocytes 344 did not induce proliferation of clonal expansion in the first weeks. Instead, it promoted hepatocyte 345 dedifferentiation into progenitors, which then expanded. Taken together, these results suggest that 346 activation of Yap in hepatic progenitor cells is sufficient for clonal expansion but in mature hepatocytes 347 it requires the second signal to drive proliferation. Another possible cause of different phenotypes may 348 lie in different levels of inflammation between our mouse model and the whole liver transgenics. Of 349 note, livers with a hepatocyte-specific ablation of Mst1/2 or Sav (both of which result in Yap activation) 350 have elevated expression of immune response genes, including TNFa and IL-631. Many variables may 351 contribute to differences in systemic inflammation. One likely contributor is the intestinal microbiome 352 composition, which varies greatly across mouse facilities67 and can induce inflammatory responses in 353 the liver68. Another source of inflammation in case of whole-liver Yap overexpression may be the 354 disruption of systemic metabolism due to Yap-mediated repression of liver-specific genes (62 and our 355 RNA-sequencing analysis). As liver-specific genes are still expressed at normal levels in a large 356 fraction of hepatocytes (Yaplow cells), systemic metabolism is not compromised in YapKIAlb-Cre mice. It is 357 therefore possible that the level of systemic inflammation is low in our mice, and this is why they require 358 an exogenous inflammatory signal to enable Yap pro-growth effects. Inflammation is an essential 359 component of carcinogenesis69, and anti-inflammatory treatments can inhibit aberrant proliferation and 360 delay tumor development70. Inflammation can promote tumor development through multiple 361 mechanisms69, including provision of a permissive signal for proliferation in the settings of disturbed 362 homeostasis70,71.

363 Our RNA-sequencing analysis has revealed two important functions of Yap in hepatocytes: First, the 364 largest category of Yap-regulated genes is related to ECM and cell adhesion. These transcriptional 11

365 targets of Yap appear to be the most common across many cell types and contexts of Yap activation; 366 for example, the gene widely used as a hallmark of Yap activation is a matricellular growth factor 367 CTGF. Since loss of cell contacts is known to activate Yap, it makes sense that Yap induces a 368 restorative transcriptional program. Second, Yap strongly suppresses liver-specific genes in response 369 to injury. Hepatocytes are known to downregulate liver-specific transcripts as they enter the cell cycle 370 during liver regeneration72. Proliferation and tissue-specific activities are often segregated between 371 progenitor and differentiated cells, especially in tissues with high cell turnover. This suggests that 372 extensive proliferation may be generally incompatible with specialized tissue functions.

373 In conclusion, tissue remodeling, cell cycle progression, and repression of tissue-specific functions all 374 need to be orchestrated in response to injury, and our findings suggest Yap as a key coordinator of 375 these activities during tissue repair. Furthermore, our data highlight the role of tissue microenvironment 376 in the outcome of Yap activation and argue that during homeostasis, growth-promoting function of Yap 377 in differentiated cells requires an additional signal, which may be provided by inflammation or injury.

378 Figure legends

379 Figure 1: YapKIAlb-Cre strain is a hepatocyte-specific mosaic mouse model of Yap 380 activation.

381 A, Detection of exogenous Yap and GFP by immunofluorescent staining of YapKIAlb-Cre liver sections. B, 382 Flow cytometric analysis of GFP fluorescence in primary hepatocytes. The plots are gated on 383 hepatocytes by forward and side scatter and on live cells by excluding 7-AAD positive events. C, 384 Expression of Yap and CTGF was determined by qPCR in primary hepatocyte populations sorted 385 based on GFP levels as shown in B.

386 Figure 1 – figure supplement 1: The design of the YapKI mice.

387 The targeting construct carrying mouse S112A mutant Yap1 followed by IRES-GFP was integrated into 388 Gt(ROSA)26Sor (Rosa26) locus. STOP denotes a strong transcriptional termination signal, removable 389 by loxP recombination.

390 Figure 1- figure supplement 2: stable mosaic expression of Rosa26 allele in multiple tissues of 391 YapKI mice.

392 A, Flow cytometry analysis of GFP populations in the peripheral blood of YapKICreER mice 4 weeks after 393 tamoxifen induction. CD19 is a marker of B lymphocytes. B, The proportion of GFP+ B cells (within 394 CD19+ gate); C, the proportion of GFPhigh cells within GFP+CD19+ gate; D, the proportion of GFPlo 395 cells within GFP+ CD19+ gate in the peripheral blood of YapKICreER after a single tamoxifen injection 396 was determined by flow cytometry weekly for one year. Numbers on the right correspond to mouse

12

397 eartags. E, Colons of YapKIVillin-Cre mice and WT littermates were stained with GFP antibody (green) 398 and DAPI (blue). Green cells in the lamina propria are autofluorescent macrophages.

399 Figure 1 – figure supplement 3: Verification of the correct targeting, expression and 400 recombination of the YapKI allele.

401 A, Southern blot verifying correct integration of the YapKI targeting construct into the Gt(ROSA)26Sor 402 locus. B, The level of Yap overexpression in YapKIAlbCre mice as determined by Western blot. Higher 403 migrating band corresponds to the exogenous Yap (due to the triple flag tag). C, qPCR on genomic 404 DNA isolated from YapKI (lox-STOP-lox) and YapKIAlbCre (lox-STOP-lox + Cre) hepatocytes with 405 primers that amplify only unrecombined (STOP-cassette-containing) region.

406 Figure 1 – figure supplement 4: Effect of the YapKI allele on Yap mRNA level.

407 Yap levels were measured by qPCR in YapKIAlbCre hepatocytes sorted based on the GFP levels as 408 described in Figure 1B with primers amplifying the total (A), endogenous (B) or exogenous (C) Yap 409 mRNA.

410

411 Figure 2: Yap overexpression in hepatocytes does not induce hepatomegaly or clonal 412 expansion at steady state.

413 A, Liver/body weight ratios of 1-3 month old YapKIAlb-Cre mice (KI) and littermate controls (WT), n≥11. 414 ****, p≤0.0001 B, Total hepatocyte numbers were determined by quantitative flow cytometry of primary 415 hepatocytes isolated from 1-3 month old mice of the indicated genotypes, n≥14. C, A representative 416 image of H&E staining performed on liver sections of 6 weeks old control (WT) and YapKIAlb-Cre mice 417 (KI). D, Primary hepatocytes were isolated by collagenase perfusion from YapKIAlb-Cre e mice of 418 indicated age groups, and percentage of GFPhigh hepatocytes was determined by flow cytometry; n≥11. 419 E, Mice were injected with BrdU for 3 consecutive days and liver sections were stained with BrdU and 420 GFP antibodies. Percent of BrdU+ hepatocytes was quantified in liver sections of the controls (WT) and 421 within Yaplow (GFP-) and Yaphigh (GFP+) populations of the YapKIAlb-Cre livers. n≥8. F, TUNEL-positive 422 nuclei were quantified on liver sections of control (WT) and YapKIAlb-Cre (KI) mice. 3 mice were used for 423 each group and 4 images were taken for each mouse. Each dot represents cell count from each image. 424 **, p≤0.01.

425 Figure 2 – figure supplement 1: Yaphigh hepatocyte cluster size does not change over time. 426 Representative images and quantification of GFP cluster size distribution in 1-5 month old YapKIAlb-Cre 427 mouse livers. Similar clone size distribution at different ages illustrates lack of clonal expansion of 428 Yaphigh cells.

13

429

430

431 Figure 2 – figure supplement 2: Overexpressed Yap induces robust transcription in hepatocytes 432 at steady state.

433 A, Heatmap showing RNA-sequencing data of genes induced in flow cytometry-sorted Yaphigh 434 hepatocytes compared to the WT (p  0.005 and  2-fold). Two biological replicates were analyzed for 435 WT, Yaplow and Yaphighhepatocytes. The expression fold change was calculated relative to the mean of 436 the WT. B, Diagram presenting the comparison between Yap-induced genes in RNA-sequencing data 437 on sorted hepatocytes generated in this study and the genes previously reported to be induced in the 438 whole liver of Yap transgenic mice5 . C, showing value of gene function analysis for Yap-induced genes 439 shown in A. D, Charts showing the percentages of TEAD binding in different functional categories of 440 genes. Gene function groups were identified from Yap induced genes shown in A. p<0.0001 for all 441 categories.

442

443 Figure 2 – figure supplement 3: Comparison of Yap-induced genes identified by the RNA 444 sequencing of Yaphigh hepatocytes (this study) and a microarray of Yap-transgenic liver5.

445 A. Heatmap showing the expression fold change of genes induced by Yap in this study as compared to 446 the data reported by Dong et al5. Genes are sorted based on their fold change in5. B, P-value of 447 Pearson correlation between the reported fold change in Dong et al5 and the fold change calculated 448 with the RNA-sequencing data. Mean of the indicated samples and the mean of wild type between 449 duplicates were used.

450 Figure 2 – figure supplement 4: Mosaic Mst1/2 loss does not induce hepatocyte expansion or 451 liver size increase. Stk4-/-;Stk3flox/flox (Mst1/2) mice were injected with a low dose adenovirus 452 expressing Cre and GFP. A, Livers were harvested 5 and 35 days after Ad-Cre-GFP injection; BrdU 453 was injected one day before harvesting. Liver sections were stained with antibodies to GFP and BrdU. 454 B, Liver/body weight ratio of Mst1/2 mice injected with Ad-Cre-GFP at the dose shown in A (5x) or 5- 455 fold lower dose (1x). C, Semi-quantitative analysis of the deleted Stk3 (∆Mst2) allele in liver genomic 456 DNA of Mst1/2 mice harvested 5 or 35 days after low dose Ad-Cre-GFP infection. Rag2 genomic PCR 457 was used as a loading control. The triangle denotes serial dilutions of the template DNA (4-fold). 458 Representative images are shown (repeated twice; n>3/group in each).

459

14

460 Figure 3: Injury induces proliferation and clonal expansion of Yap-overexpressing 461 hepatocytes.

462 A, Liver sections of control (WT) and YapKIAlb-Cre (KI) mice isolated at steady state (SS) or on days 4

463 and 14 after CCL4 treatment were stained with antibodies to GFP and BrdU. BrdU was injected in all 464 groups for 3 consecutive days before harvesting the livers. B, The percentage of Yaphigh (GFP+) cells 465 among all hepatocytes was quantified within the pericentral zone (“damage site”) and outside the Alb-Cre 466 pericentral zone (“outside”) and in the liver sections of YapKI mice on day4 after CCl4 treatment. 3- 467 4 mice were used for each group and 3-4 images were taken for each mouse. Each dot represents cell 468 count from each image. ****, p0.0001. C, The percentages of BrdU+ hepatocytes among GFP- and 469 GFP+ hepatocytes were quantified outside the CV zone (“outside”) and around CV (“damage site”) in Alb-Cre 470 liver sections of YapKI mice on day4 after CCl4 treatment. 3-4 mice were used for each group and 471 3-4 images were taken for each mouse. Each dot represents a cell count from each image. ****, 472 p0.0001. D, Hepatocytes were isolated by collagenase perfusion from 1-2 month old YapKIAlb-Cre mice

473 at steady state (SS), at day 4 (D4) and day 14 (D14) after CCl4 treatment, or after a 6-day LPS 474 treatment (LPS), and percentage of GFPhigh cells determined by flow cytometry.

475 Figure 3 – figure supplement 1: Kinetics of CCl4-induced liver damage in the YapKIAlb-Cre mice.

Alb-Cre 476 ALT blood level of control (WT) and YapKI (KI) mice measured at the indicated days after CCl4 477 treatment. One representative experiment is shown of the two independent repeats, n=3 per group in 478 each.

479 Figure 3 – figure supplement 2: Proliferation rates of WT and YapKIAlb-Cre hepatocytes are similar

480 on day 4 after CCl4 treatment.

481 Mice were injected with BrdU 1 hour before livers were harvested on day 4 after CCl4 treatment. Liver 482 sections were stained with antibodies to GFP (green), glutamine synthase (blue) and BrdU (red).

483 Figure 3 – figure supplement 3: Hepatocyte numbers undergo similar changes in response to Alb-Cre 484 CCl4 in the WT and YapKI mice.

485 Hepatocyte numbers were measured by quantitative flow cytometry in the pellet fractions of cells

486 isolated from collagenase-perfused livers at indicated times after CCl4 treatment.

487 Figure 3 – figure supplement 4: Liver/body weight ratios undergo similar changes in response Alb-Cre 488 to CCl4 in the WT and YapKI mice.

489 Liver/body weight ratio of YapKIAlb-Cre and littermate control mice was determined at the indicated days

490 after CCl4 injection.

15

491 Figure 3 – figure supplement 5: Hematopoietic cell numbers undergo similar changes in Alb-Cre 492 response to CCl4 in the WT and YapKI mice.

493 CD45+ cell numbers were measured by quantitative flow cytometry in the supernatant fractions of cells

494 isolated from collagenase-perfused livers at indicated times after CCl4 treatment.

495 Figure 4: Yap function in hepatocytes is required for tissue repair.

496 A, expression of glutamine synthase detected by immunofluorescent staining of littermate controls (WT) Alb-Cre 497 and YapCKO YapCKOlivers on day4 after CCl4 treatment. CV, central vein. B, BrdU staining

498 illustrating lack of hepatocyte proliferation in Yap-deficient hepatocytes in response to CCl4-induced 499 injury. The results are quantified in Figure 5C. C, Total BrdU+ hepatocytes in WT, YapKIAlb-Cre (KI) and 500 YapCKOAlb-Cre (CKO) mice were quantified outside the CV zone (“outside”) and around CV (“damage

501 site”) in liver sections on day4 after CCl4 treatment. 3-4 mice were used for each group and 3-4 images 502 were taken from each mouse. Each dot represents a cell count from each image. D, Excessive collagen 503 deposition in YapCKOAlb-Cre mice on day 4 after CCl4 treatment revealed by Sirius Red staining.

504 Figure 4 – figure supplement 1: The design of the Yap1 conditional knockout targeting 505 construct.

506 Exon 2 is floxed and the neo cassette is removable by frt recombination.

507 Figure 4 – figure supplement 2: Verification of the correct targeting, expression and 508 recombination of the Yap1 conditional knockout allele.

509 A, Southern blot verifying correct integration of the targeting construct. B, qPCR on genomic DNA 510 isolated from Yap1flox/flox (F/F) and Yap1flox/flox;AlbCre (F/F + Cre) hepatocytes with primers that amplify 511 only the unrecombined region. C, Decreased Yap mRNA levels in YapCKOAlb-Cre mouse liver as 512 measured by qPCR. D, Decreased Yap protein levels in YapCKOAlb-Cre mouse liver as measured by 513 Western blot.

Alb-Cre 514 Figure 4 – figure supplement 3: Kinetics of CCl4-induced liver damage in the YapCKO mice.

Alb-Cre 515 ALT blood level of control (WT) and YapCKO (KO) mice measured at indicated days after CCl4

516 treatment. The control group was injected with peanut oil (the carrier used for CCl4).

517 Figure 4 – figure supplement 4: Increased liver weight of the YapCKOAlb-Cre mice. Liver/body 518 weight ratio of YapCKOAlb-Cre and littermate control mice was determined in the untreated 6-8 weeks old 519 mice. ****, p0.0001.

520 Figure 4 – figure supplement 5: No defect in proliferation of the Yap-deficient hepatocytes at 521 steady state.

16

522 Representative images and quantification of the BrdU+ hepatocytes of the 1.5 month old YapCKOAlb-Cre 523 and littermate control mice at steady state. Mice were injected with BrdU for 3 consecutive days and 524 liver sections were stained with BrdU antibody. The number of BrdU+ hepatocytes per a 20x field was 525 quantified and normalized to the average values of the WT. 2-3 mice were used for each group and at 526 least 20 images were taken for each mouse. Each dot represents cell count from each image. ****, 527 p0.0001.

528

529 Figure 5: Inflammation induces proliferation of Yap-overexpressing hepatocytes.

530 A, Control (WT) or YapKIAlb-Cre mice were coinjected with BrdU and LPS for 6 days. Percentage of 531 BrdU+ hepatocytes was quantified. 4-6 mice were used for each group and 3-30 images were taken for 532 each mouse. Each dot represents cell count from each image. B, A representative image of chronic 533 (6xLPS) LPS injected livers used for the quantification in A. Arrows indicate BrdU+ hepatocyte nuclei. 534 C, H&E representative images of the chronic LPS-treated livers. D, Hepatocyte numbers of the chronic 535 LPS-treated livers was measured by quantitative flow cytometry.

536 Figure 5 – figure supplement 1: Role of IL-6 in expansion of Yap-overexpressing hepatocytes. A, 537 1-2 month old YapKIAlb-Cre were injected s.c. with LLC cells stably expressing IL-6 (KI+LLC-IL-6), or 538 empty vector (KI+LLC), and IL-6 levels in the serum were measured 14 to 20 days later. Untreated 539 YapKI and WT mice were bled as controls. B, Percentage of GFPhigh cells determined by flow cytometry 540 in hepatocytes from 1-2 month old YapKIAlb-Cre mice from (A) grouped based on serum IL-6 levels).

541 Figure 5 – figure supplement 2: Correlation of TEAD and NFkB binding sites in the promoters of 542 genes regulating proliferation. Genes with DNA sequence motifs of NF-kB (p65) and/or TEADs were 543 identified with Motifmap (NF-kB M00208, TEADs M01305), using mouse genome MM9 multiz30way. 544 NF-kB motifs were found within 1 kb of gene transcription start sites (TSSs) and TEAD motifs were 545 found within 10 kb of TSSs. Genes related to cell proliferation were identified with David 546 GO:0042127 for regulation of cell proliferation. Statistical significance was calculated with Chi-square 547 test.

548

549 Figure 6: Yap activation cooperates with tissue injury to repress hepatocyte 550 differentiation and promote progenitor phenotype.

Alb-Cre 551 A, H&E staining of liver sections of control (WT) and YapKI (KI) mice isolated on days 4 CCL4 552 treatment. B, Heatmap of RNA-sequencing data comparing expression of liver specific genes (see 553 methods). Hepatocytes were sorted from wild type livers (WT) or from YapKIAlb-Cre livers based on GFP

17

554 levels (Lo and Hi), at steady state (SS) or on day 4 after CCl4 treatment (CCl4). The fold change is 555 calculated between the indicated samples and WT in steady state. Data represent the mean of the Alb-Cre 556 duplicates. C, Liver sections of control (WT) and YapKI (KI) mice isolated on days 4 CCL4 557 treatment were stained with antibodies to glutamine synthase. D, Glutamine synthase (GS), Yap and

558 beta-actin protein levels in whole-liver protein lysates prepared on day 4 after CCl4 treatment were 559 determined by Western blotting. Higher migrating band in the middle panel corresponds to the 560 exogenous Yap (due to the in-frame triple flag tag). E, F, Heatmaps showing the RNA-sequencing data 561 for genes enriched in embryonic tissues (E) or progenitor markers (F). Hepatocytes were sorted from 562 wild type livers (WT) or from YapKIAlb-Cre livers based on GFP levels (Lo and Hi), at steady state (SS) or

563 on day 4 after CCl4 treatment (CCl4). The fold change is calculated between the indicated samples and 564 wild type in steady state. G, H, Flow cytometric analysis of primary hepatocytes isolated by collagenase Alb-Cre 565 perfusion from CCl4 treated control (WT) or YapKI (KI) mice. G, Representative flow cytometry 566 plots gated of CD45- CD31- population. The numbers on the plots indicate the percentages of the gated 567 population (EpCam+ progenitors). H, The results from (G) were combined with the total hepatocyte 568 numbers to calculate the number of hepatic EpCam+ progenitors per liver.

569

570 Figure 6 – figure supplement 1: Verification of the RNA-sequencing results.

571 Primary hepatocytes of WT and YapKIAlb-Cre (KI) mice were sorted by flow cytometry based on GFP

572 levels from steady state livers (SS) or on day4 after CCl4 treatment (CCL4) and gene expression 573 determined by qPCR. Dots represent individual mice. 574

575 Figure 7: Yap sensitizes hepatocytes to TRAIL-mediated apoptosis.

576 A, Cell death as reflected by TUNEL-positive cell numbers quantified outside the CV zone (“outside”) Alb-Cre 577 and around CV (“damage site”) in liver sections of YapKI mice on day4 after CCl4 treatment. 3-4 578 mice were used for each group and 3-4 images were taken from each mouse. Each dot represents cell 579 count from each image. B, RPKM data from RNA-sequencing illustrating expression of apoptosis- high 580 related genes in WT or Yap (Hi) hepatocytes sorted from untreated (SS) livers or on day 4 after CCl4 581 treatment (CCl4). C, D, Dr5 mRNA levels in WT and YapKIAlb-Cre livers (KI) at indicated time points after

582 CCl4 treatment were determined by qPCR in whole liver extracts (C), or in primary hepatocytes sorted

583 by flow cytometry based on GFP levels from steady state livers (SS) or on day4 after CCl4 treatment 584 (CCL4) (D). 2-3 mice were used for each group. *, p0.05. E, Primary hepatocytes sorted as in (D) 585 were cultured on collagen coated plates with or without TRAIL, and cell viability measured by 586 CellTiterBlue the next day. 3-4 mice were used for each group and 3-4 wells were seeded with 587 hepatocytes from each mouse. Each dot represents reading from each well. 18

588

589 Methods

590 Generation of YapKI and YapCKO mice

591 To generate the YapKI and YapCKO lines, targeting vectors were designed and constructed as shown 592 in Figure 1- Figure Supplement1A and Figure 4 – Figure Supplement 1. NM-001171147 mouse Yap1 593 isoform with introduced S112A mutation (corresponding to S127A of the human protein) was used to 594 generate YapKI allele. After the successful construction of the targeting vectors, the purified plasmid 595 DNA was linearized and transfected into the C57BL/6-derived Bruce4 ES cells. Correct recombination 596 and replacement of the DNA were verified by Southern blots. Selected clones were expanded and 597 submitted to the Yale Transgene Facility for injections into BALB/c blastocysts to generate chimeric 598 mice. In case of the YapCKO line, mice with the germline-transmitted Yap floxed allele were then 599 crossed to Flp-deleter transgenic mice54 to remove the neomycin cassette.

600 Mouse treatments

601 Animals were maintained at the Yale Animal Resources Center. All animal experiments were performed 602 with approval by the Institutional Animal Care and Use Committee of Yale University. All mice were on 603 C57BL/6 background. AlbCre mice were from Jackson Laboratory. For BrdU staining, daily 604 intraperitoneal (i.p.) injections of BrdU (Sigma, 100 ug/mouse) were performed for 3 days before liver

605 dissection. Liver damage was induced by i.p. injection of CCl4 (Sigma, 1ul/g, diluted 1:10 with peanut 606 oil). For the LLC-IL-6 experiment, 1 million LLC cells stably expressing mouse IL-6 or empty vector 607 were injected into each flank of 1-1.5 month old YapKIAlbCre mice. 2-3 weeks later when tumors reached 608 approximately 1cm in diameter, the mice were eye-bled for IL-6 ELISA and sacrificed for hepatocyte 609 cell preparation. For the in vivo deletion of Mst2flox/flox allele, AdCMVCre-RSV-GFP adenovirus from 610 KeraFast was amplified in 293A cells, purified by ViraPur kit, dialized against PBS and injected into the 611 tail vein of the 1-2 month old Mst1-/- Mst2flox/flox mice.

612 Immunofluorescence microscopy

613 Liver lobes were fixed in 4% PFA overnight and embedded in paraffin. Sections were deparaffinized 614 and heated for 20 min at 95’C for antigen retrieval in Tris-EDTA buffer pH 9.

615 TUNEL staining was performed using In Situ Cell Death Detection Kit (Roche).

616 For GFP, Yap, and GS immunofluorescence staining, samples were then blocked in TBS-T with 3% 617 BSA for 2 hours (staining buffer), incubated with the primary antibody overnight, and with the secondary 618 antibody for 1 hour. For BrdU staining, after the described steps, the samples were fixed in 4% PFA in 619 PBS for 20 minutes at RT. DNA was then denatured in 2 N HCl for 30 min at 37’C. After neutralization 19

620 with 0.1 M sodium tetraborate, the sections were stained with mouse anti-BrdU antibodies for 2 hours 621 and visualized with flourophore-conjugated anti-mouse secondary antibodies. The slides were mounted 622 in Vectashield anti-fade media containing Hoescht dye to counterstain the nuclei. 20 to 30 20x tiled 623 images were obtained using Zeiss Axioplan microscope and Axiovision software. Cells were counted 624 manually in AxioVison software. Only hepatocytes (as determined by morphology and GFP expression 625 where applicable) were scored for BrdU quantification.

626 Hepatocyte FACS

627 Primary hepatocytes were prepared by collagenase perfusion at Yale Liver Center. Hepatocytes were 628 washed twice with FACS buffer (2mM EDTA, 2% FBS in PBS), stained with 7-Aminoactinomycin D 629 (7AAD) or propidium iodide, and analyzed on BD Accuri or LSRII, or sorted on MoFlo cytometers. 630 Samples were gated on live hepatocytes based on forward and side scatter and live dye exclusion.

631 RNA sequencing analysis

632 Duplicate hepatocyte samples were prepared in several independent experiments. Each of the 633 duplicates contained material pooled from multiple mice. Hepatocytes were isolated by flow cytometry, 634 and RNA was extracted using RNeasy kit (Qiagen). 1-10 ug of each sample was submitted to the Yale 635 Center of Genome Analysis.

636 Sequencing libraries were constructed and were sequenced by Illumina Hiseq 2500 with 76 bp single- 637 end reads, which generated 20 million raw reads per sample on average. After removing the low-quality 638 reads (around 0.18% of all reads) and low-quality portions (Q value < 30) of each of the raw reads, the 639 RNA-sequencing data for each sample were mapped to the GRCm38/mm10 mouse reference first. The 640 GRCm38 reference was downloaded from the Mouse Genome Informatics (MGI). The mapping was 641 performed using TopHat v2.0.8, allowing 2 mismatches. The percentages of mapped reads were 642 71.16% on average (from 70.00% to 73.79%). After mapping, Cufflinks v2.0.2 was applied to assemble 643 and quantify the transcripts and discover the differentially expressed genes, with the annotated gene 644 information from the MGI. The gene expression values were calculated for each sample based on the 645 number of fragments per kilobase of exon per million reads mapped (FPKM). The significance of 646 differential expression of genes was detected using Cuffdiff for all comparisons of every two samples.

647 To select differentially expressed genes, the average of two biological replicates was compared 648 between the conditions examined. To select genes with significant activation or repression in any 649 comparison, the p-value of 0.005 was used together with a fold-change threshold of 2-fold. A 650 pseudocount of 0.01 was added to RPKM value of each gene as a sequencing background, to avoid 651 inflated fold change caused by lowly expressed genes.

20

652 Gene functional analysis

653 Gene functional analysis was performed with the DAVID functional annotation tools51, 52 654 (http://david.abcc.ncifcrf.gov). The p-value was adjusted with Benjamini methods for multiple hypothesis 655 testing.

656 Analysis of tissue specific genes

657 Tissue specific genes were obtained from Pattern Gene Database61 658 (http://bioinf.xmu.edu.cn/PaGenBase). Genes expressed specifically (only in one tissue) and selectively 659 (only in a group of samples) were pooled together as the tissue specific genes in our analysis. Liver 660 specific genes are identified from mouse liver sample, and mouse embryo specific genes are identified 661 from mouse day 9.5 embryo sample.

662 Transcription factor binding analysis

663 Genome-wide binding of Tead4 (identified by ChIP-sequencing) was retrieved from the ENCODE 664 project54. The uniform peaks determined in all available samples (HepG2 cell line, K562 cell line and 665 hESC) were combined, and genes whose TSS is within 10kb distance from any Tead4 uniform peak 666 were defined as Tead4-associated genes. Hg19 genome annotation was downloaded from the UCSC 667 database73

668 TRAIL sensitivity assay

669 Primary hepatocytes were plated in collagen coated wells in Williams’ E medium (WEM) supplemented 670 with 5% FCS, 10mM HEPES, 2mM L-glutamine, 10mM penicillin/streptomycin, 8 ug/ml gentamicin, 100 671 ug/mL chloramphenicol, 100 nM dexamethasone, and 1nM Insulin. 4 hours after plating, the medium 672 was changed to the supplemented WEM without chloramphenicol and dexamethasone. On the second 673 day cells were treated with recombinant TRAIL (R&D Systems) overnight. Viable cell number was 674 measured by the CellTiterBlue assay (Sigma) according to the manufacturer protocol.

675 Western blotting

676 Whole liver tissue was snap-frozen in liquid nitrogen, homogenized and resuspended in TNT buffer (0.1 677 M Tris-HCL pH 7.5, 150 mM NaCl, 0.1% Tween20) supplemented with the Complete protease 678 inhibitors (Roche) and cleared by centrifugation at 100 g. Protein concentration in the supernatants was 679 normalized using Bradford reagent. 15 ug of the extracts were separated on a gradient SDS-PAGE gel 680 and transferred to PVDF membrane. After blocking with 3% BSA in TBST buffer, the membranes were 681 probed with antibodies to Yap (Cell Signaling), glutamine synthase (Genscript), and beta-actin (Sigma).

21

682 ALT test

683 ALT tests were performed using ALT activity assay (Sigma) according to the manufacturer’s protocol.

684 Real-time PCR

685 RNA was extracted with RNA-bee, and cDNA synthesis were performed using SMART MMLV reverse 686 transcriptase according to the manufacturers’ protocols. Real time PCR was performed in triplicates 687 using Quanta SYBR reagent on C1000 Thermal Cycler (BioRad). Primer sequences are listed in the 688 Supplementary File1.

689 Liver genomic PCR

690 Three 1 mm3 pieces of liver tissue per mouse were digested with Pronase overnight, and genomic DNA 691 isolated by NaCl/ethanol precipitation. The three DNA samples for each mouse were pooled and used 692 at 4-fold dilutions to amplify the Rag2 or recombined Stk3flox/flox or alleles using Taqman polymerase.

693 Enzyme linked immunosorbent assay (ELISA)

694 Paired antibodies against IL-6 were purchased from BD Biosciences to perform ELISA.

695 Statistical analysis

696 All statistical analysis was performed by 2 tailed unpaired student t test, unless specified otherwise.

697 Accession numbers

698 The RNA sequencing data have been deposited to GEO database under the accession number 699 GSE65207.

700 Acknowledgements

701 We thank Kathy Harry from Yale Liver Center for primary hepatocyte preparation, Gouzel Tokmoulina 702 for assistance with hepatocyte sorting, Timothy Nottoli for help with generation of transgenic mice, 703 Charles Anicelli, Sophie Cronin and Steven Zhu for technical assistance with mouse treatments and 704 colony maintenance, Raj Chovatiya for the critical reading of the manuscript, and all members of the 705 Medzhitov lab for helpful discussions. This work was supported by the Howard Hughes Medical 706 Institute and a grant from NIH to RM, and by the National Institute of Diabetes and Digestive and 707 Kidney Diseases of the National Institutes of Health under Award Number P30KD034989 (Yale Liver 708 Center).

709

22

710 References 711 1. Yu, F.X. & Guan, K.L. The Hippo pathway: regulators and regulations. Genes & development 27, 355-371 712 (2013). 713 2. Hong, W. & Guan, K.L. The YAP and TAZ transcription co-activators: key downstream effectors of the 714 mammalian Hippo pathway. Seminars in cell & developmental biology 23, 785-793 (2012). 715 3. Lin, J.I., Poon, C.L. & Harvey, K.F. The Hippo size control pathway--ever expanding. Science signaling 6, 716 pe4 (2013). 717 4. Camargo, F.D. et al. YAP1 increases organ size and expands undifferentiated progenitor cells. Current 718 biology : CB 17, 2054-2060 (2007). 719 5. Dong, J. et al. Elucidation of a universal size-control mechanism in Drosophila and mammals. Cell 130, 720 1120-1133 (2007). 721 6. Zhao, B. et al. TEAD mediates YAP-dependent gene induction and growth control. Genes & development 722 22, 1962-1971 (2008). 723 7. Zhang, L. et al. The TEAD/TEF family of transcription factor Scalloped mediates Hippo signaling in organ 724 size control. Developmental cell 14, 377-387 (2008). 725 8. Wu, S., Liu, Y., Zheng, Y., Dong, J. & Pan, D. The TEAD/TEF family protein Scalloped mediates 726 transcriptional output of the Hippo growth-regulatory pathway. Developmental cell 14, 388-398 (2008). 727 9. Zhao, B. et al. TEAD mediates YAP-dependent gene induction and growth control. Genes & development 728 22, 1962-1971 (2008). 729 10. Reddy, B.V. & Irvine, K.D. Regulation of Hippo signaling by EGFR-MAPK signaling through Ajuba family 730 proteins. Developmental cell 24, 459-471 (2013). 731 11. Yu, F.X. et al. Regulation of the Hippo-YAP pathway by G-protein-coupled receptor signaling. Cell 150, 732 780-791 (2012). 733 12. Fan, R., Kim, N.G. & Gumbiner, B.M. Regulation of Hippo pathway by mitogenic growth factors via 734 phosphoinositide 3-kinase and phosphoinositide-dependent kinase-1. Proceedings of the National 735 Academy of Sciences of the United States of America 110, 2569-2574 (2013). 736 13. Bernascone, I. & Martin-Belmonte, F. Crossroads of Wnt and Hippo in epithelial tissues. Trends in cell 737 biology 23, 380-389 (2013). 738 14. Attisano, L. & Wrana, J.L. Signal integration in TGF-beta, WNT, and Hippo pathways. F1000prime reports 739 5, 17 (2013). 740 15. Mullen, A.C. Hippo tips the TGF-beta scale in favor of pluripotency. Cell stem cell 14, 6-8 (2014). 741 16. Ferrigno, O. et al. Yes-associated protein (YAP65) interacts with Smad7 and potentiates its inhibitory 742 activity against TGF-beta/Smad signaling. Oncogene 21, 4879-4884 (2002). 743 17. Beyer, T.A. et al. Switch enhancers interpret TGF-beta and Hippo signaling to control cell fate in human 744 embryonic stem cells. Cell reports 5, 1611-1624 (2013). 745 18. Konsavage, W.M., Jr. & Yochum, G.S. Intersection of Hippo/YAP and Wnt/beta-catenin signaling 746 pathways. Acta biochimica et biophysica Sinica 45, 71-79 (2013). 747 19. Zaidi, S.K. et al. Tyrosine phosphorylation controls Runx2-mediated subnuclear targeting of YAP to 748 repress transcription. The EMBO journal 23, 790-799 (2004). 749 20. Shao, D. et al. A functional interaction between Hippo-YAP signalling and FoxO1 mediates the oxidative 750 stress response. Nature communications 5, 3315 (2014). 751 21. Strano, S. et al. Physical interaction with Yes-associated protein enhances p73 transcriptional activity. 752 The Journal of biological chemistry 276, 15164-15173 (2001). 753 22. Barry, E.R. & Camargo, F.D. The Hippo superhighway: signaling crossroads converging on the Hippo/Yap 754 pathway in stem cells and development. Current opinion in cell biology 25, 247-253 (2013). 755 23. Shaw, R.L. et al. The Hippo pathway regulates intestinal stem cell proliferation during Drosophila adult 756 midgut regeneration. Development 137, 4147-4158 (2010). 757 24. Karpowicz, P., Perez, J. & Perrimon, N. The Hippo tumor suppressor pathway regulates intestinal stem 758 cell regeneration. Development 137, 4135-4145 (2010).

23

759 25. Ren, F. et al. Hippo signaling regulates Drosophila intestine stem cell proliferation through multiple 760 pathways. Proceedings of the National Academy of Sciences of the United States of America 107, 21064- 761 21069 (2010). 762 26. Staley, B.K. & Irvine, K.D. Warts and Yorkie mediate intestinal regeneration by influencing stem cell 763 proliferation. Current biology : CB 20, 1580-1587 (2010). 764 27. Zhou, D. et al. Mst1 and Mst2 protein kinases restrain intestinal stem cell proliferation and colonic 765 tumorigenesis by inhibition of Yes-associated protein (Yap) overabundance. Proceedings of the National 766 Academy of Sciences of the United States of America 108, E1312-1320 (2011). 767 28. Barry, E.R. et al. Restriction of intestinal stem cell expansion and the regenerative response by YAP. 768 Nature 493, 106-110 (2013). 769 29. Cai, J. et al. The Hippo signaling pathway restricts the oncogenic potential of an intestinal regeneration 770 program. Genes & development 24, 2383-2388 (2010). 771 30. Zheng, T., Wang, J., Jiang, H. & Liu, L. Hippo signaling in oval cells and hepatocarcinogenesis. Cancer 772 letters 302, 91-99 (2011). 773 31. Lu, L. et al. Hippo signaling is a potent in vivo growth and tumor suppressor pathway in the mammalian 774 liver. Proceedings of the National Academy of Sciences of the United States of America 107, 1437-1442 775 (2010). 776 32. Kowalik, M.A. et al. Yes-associated protein regulation of adaptive liver enlargement and hepatocellular 777 carcinoma development in mice. Hepatology 53, 2086-2096 (2011). 778 33. Lee, K.P. et al. The Hippo-Salvador pathway restrains hepatic oval cell proliferation, liver size, and liver 779 tumorigenesis. Proceedings of the National Academy of Sciences of the United States of America 107, 780 8248-8253 (2010). 781 34. Zhang, N. et al. The Merlin/NF2 tumor suppressor functions through the YAP oncoprotein to regulate 782 tissue homeostasis in mammals. Developmental cell 19, 27-38 (2010). 783 35. Bai, H. et al. Yes-associated protein regulates the hepatic response after bile duct ligation. Hepatology 784 56, 1097-1107 (2012). 785 36. Yu, F.X., Meng, Z., Plouffe, S.W. & Guan, K.L. Hippo Pathway Regulation of Gastrointestinal Tissues. 786 Annual review of physiology (2014). 787 37. Fernandez, L.A. et al. YAP1 is amplified and up-regulated in hedgehog-associated medulloblastomas and 788 mediates Sonic hedgehog-driven neural precursor proliferation. Genes & development 23, 2729-2741 789 (2009). 790 38. Wang, K., Degerny, C., Xu, M. & Yang, X.J. YAP, TAZ, and Yorkie: a conserved family of signal-responsive 791 transcriptional coregulators in animal development and human disease. Biochemistry and cell biology = 792 Biochimie et biologie cellulaire 87, 77-91 (2009). 793 39. Hahn, W.C. & Weinberg, R.A. Rules for making human tumor cells. The New England journal of medicine 794 347, 1593-1603 (2002). 795 40. Evan, G.I. et al. Induction of apoptosis in fibroblasts by c-myc protein. Cell 69, 119-128 (1992). 796 41. Serrano, M., Lin, A.W., McCurrach, M.E., Beach, D. & Lowe, S.W. Oncogenic ras provokes premature cell 797 senescence associated with accumulation of p53 and p16INK4a. Cell 88, 593-602 (1997). 798 42. O'Reilly, L.A., Huang, D.C. & Strasser, A. The cell death inhibitor Bcl-2 and its homologues influence 799 control of cell cycle entry. The EMBO journal 15, 6979-6990 (1996). 800 43. Lapi, E. et al. PML, YAP, and p73 are components of a proapoptotic autoregulatory feedback loop. 801 Molecular cell 32, 803-814 (2008). 802 44. Boggiano, J.C. & Fehon, R.G. Growth control by committee: intercellular junctions, cell polarity, and the 803 cytoskeleton regulate Hippo signaling. Developmental cell 22, 695-702 (2012). 804 45. Chen, C.L., Schroeder, M.C., Kango-Singh, M., Tao, C. & Halder, G. Tumor suppression by cell competition 805 through regulation of the Hippo pathway. Proceedings of the National Academy of Sciences of the United 806 States of America 109, 484-489 (2012). 807 46. Norman, M. et al. Loss of Scribble causes cell competition in mammalian cells. Journal of cell science 808 125, 59-66 (2012). 24

809 47. Schlegelmilch, K. et al. Yap1 acts downstream of alpha-catenin to control epidermal proliferation. Cell 810 144, 782-795 (2011). 811 48. Zhang, H., Pasolli, H.A. & Fuchs, E. Yes-associated protein (YAP) transcriptional coactivator functions in 812 balancing growth and differentiation in skin. Proceedings of the National Academy of Sciences of the 813 United States of America 108, 2270-2275 (2011). 814 49. Bondar, T. & Medzhitov, R. p53-mediated hematopoietic stem and progenitor cell competition. Cell stem 815 cell 6, 309-322 (2010). 816 50. Song, H. et al. Mammalian Mst1 and Mst2 kinases play essential roles in organ size control and tumor 817 suppression. Proceedings of the National Academy of Sciences of the United States of America 107, 818 1431-1436 (2010). 819 51. Huang da, W., Sherman, B.T. & Lempicki, R.A. Systematic and integrative analysis of large gene lists using 820 DAVID bioinformatics resources. Nature protocols 4, 44-57 (2009). 821 52. Huang da, W., Sherman, B.T. & Lempicki, R.A. Bioinformatics enrichment tools: paths toward the 822 comprehensive functional analysis of large gene lists. Nucleic acids research 37, 1-13 (2009). 823 53. Liu-Chittenden, Y. et al. Genetic and pharmacological disruption of the TEAD-YAP complex suppresses 824 the oncogenic activity of YAP. Genes & development 26, 1300-1305 (2012). 825 54. Consortium, E.P. A user's guide to the encyclopedia of DNA elements (ENCODE). PLoS biology 9, 826 e1001046 (2011). 827 55. Anakk, S. et al. Bile acids activate YAP to promote liver carcinogenesis. Cell reports 5, 1060-1069 (2013). 828 56. Wang, C. et al. Differences in Yes-associated protein and mRNA levels in regenerating liver and 829 hepatocellular carcinoma. Molecular medicine reports 5, 410-414 (2012). 830 57. Wu, H. et al. The Ets transcription factor GABP is a component of the hippo pathway essential for growth 831 and antioxidant defense. Cell reports 3, 1663-1677 (2013). 832 58. Weber, L.W., Boll, M. & Stampfl, A. Hepatotoxicity and mechanism of action of haloalkanes: carbon 833 tetrachloride as a toxicological model. Critical reviews in toxicology 33, 105-136 (2003). 834 59. Balani, S., Malik, R., Malik, R. & Kapoor, N. Cytomorphological variables of hepatic malignancies in fine 835 needle aspiration smears with special reference to grading of hepatocellular carcinoma. Journal of 836 cytology / Indian Academy of Cytologists 30, 116-120 (2013). 837 60. Chang, O. et al. The cytological characteristics of small cell change of dysplasia in small hepatic nodules. 838 Oncology reports 23, 1229-1232 (2010). 839 61. Pan, J.B. et al. PaGenBase: a pattern gene database for the global and dynamic understanding of gene 840 function. PloS one 8, e80747 (2013). 841 62. Yimlamai, D. et al. Hippo pathway activity influences liver cell fate. Cell 157, 1324-1338 (2014). 842 63. Wang, J. et al. TRIB2 acts downstream of Wnt/TCF in liver cancer cells to regulate YAP and C/EBPalpha 843 function. Molecular cell 51, 211-225 (2013). 844 64. Septer, S. et al. Yes-associated protein is involved in proliferation and differentiation during postnatal 845 liver development. American journal of physiology. Gastrointestinal and liver physiology 302, G493-503 846 (2012). 847 65. Tomlinson, V. et al. JNK phosphorylates Yes-associated protein (YAP) to regulate apoptosis. Cell death & 848 disease 1, e29 (2010). 849 66. Harrington, E.A., Bennett, M.R., Fanidi, A. & Evan, G.I. c-Myc-induced apoptosis in fibroblasts is inhibited 850 by specific cytokines. The EMBO journal 13, 3286-3295 (1994). 851 67. Bleich, A. & Hansen, A.K. Time to include the gut microbiota in the hygienic standardisation of laboratory 852 rodents. Comparative immunology, microbiology and infectious diseases 35, 81-92 (2012). 853 68. Henao-Mejia, J. et al. Inflammasome-mediated dysbiosis regulates progression of NAFLD and obesity. 854 Nature 482, 179-185 (2012). 855 69. Ben-Neriah, Y. & Karin, M. Inflammation meets cancer, with NF-kappaB as the matchmaker. Nature 856 immunology 12, 715-723 (2011). 857 70. Pribluda, A. et al. A senescence-inflammatory switch from cancer-inhibitory to cancer-promoting 858 mechanism. Cancer cell 24, 242-256 (2013). 25

859 71. Bondar, T. & Medzhitov, R. The origins of tumor-promoting inflammation. Cancer cell 24, 143-144 860 (2013). 861 72. Ito, Y. et al. Depression of liver-specific gene expression in regenerating rat liver: a putative cause for 862 liver dysfunction after hepatectomy. The Journal of surgical research 51, 143-147 (1991). 863 73. Karolchik, D. et al. The UCSC Genome Browser database: 2014 update. Nucleic acids research 42, D764- 864 770 (2014).

865

866

867

26

868 List of figures

869

870 Figure 1: YapKIAlb-Cre strain is a hepatocyte-specific mosaic mouse model of Yap 871 activation. 872 Figure 1 – figure supplement 1: The design of the YapKI mice.

873 Figure 1- figure supplement 2: Stable mosaic expression of the YapKI allele in multiple tissues 874 of YapKI mice.

875 Figure 1 – figure supplement 3: Verification of the correct targeting, expression and 876 recombination of the YapKI allele.

877 Figure 1 – figure supplement 4: Effect of the YapKI allele on Yap mRNA level.

878 Figure 2: Yap overexpression in hepatocytes does not induce hepatomegaly or clonal 879 expansion at steady state. 880 Figure 2 – figure supplement 1: Yaphigh hepatocyte cluster size does not change over time.

881 Figure 2 – figure supplement 2: Overexpressed Yap induces robust transcription in hepatocytes 882 at steady state.

883 Figure 2 – figure supplement 3: Comparison of Yap-induced genes identified by the RNA 884 sequencing of Yaphigh hepatocytes (this study) and a microarray of Yap-transgenic liver.

885 Figure 2 – figure supplement 4: Mosaic Mst1/2 loss does not induce hepatocyte expansion or 886 liver size increase.

887 Figure 3: Injury induces proliferation and clonal expansion of Yap-overexpressing 888 hepatocytes.

889 Figure 3 – figure supplement 1: Kinetics of CCl4-induced liver damage in the YapKIAlb-Cre mice.

890 Figure 3 – figure supplement 2: Proliferation rates of WT and YapKIAlb-Cre hepatocytes are

891 similar on day 4 after CCl4 treatment.

892 Figure 3 – figure supplement 3: Hepatocyte numbers undergo similar changes in response to 893 CCl4 in the WT and YapKIAlb-Cre mice.

894 Figure 3 – figure supplement 4: Liver/body weight ratios undergo similar changes in response to

895 CCl4 in the WT and YapKI;AlbCre mice.

896 Figure 3 – figure supplement 5: Hematopoietic cell numbers undergo similar changes in

897 response to CCl4 in the WT and YapKI;AlbCre mice.

27

898 Figure 4: Yap function in hepatocytes is required for tissue repair.

899 Figure 4 – figure supplement 1: The design of the Yap conditional knockout targeting construct.

900 Figure 4 – figure supplement 2: Verification of the correct targeting, expression and 901 recombination of the YapCKO allele.

902 Figure 4 – figure supplement 3: Kinetics of CCl4-induced liver damage in the YapCKOAlb-Cre 903 mice.

904 Figure 4 – figure supplement 4: Increased liver weight of the YapCKOAlb-Cre mice.

905 Figure 4 – figure supplement 5: No defect in proliferation of the Yap-deficient hepatocytes at 906 steady state.

907 Figure 5: Inflammation induces proliferation of Yap-overexpressing hepatocytes.

908 Figure 5 – figure supplement 1: Role of IL-6 in expansion of Yap-overexpressing hepatocytes.

909 Figure 5 – figure supplement 2: Correlation of TEAD and NFkB binding sites in the promoters of 910 genes regulating proliferation,

911 Figure 6: Yap activation cooperates with tissue injury to repress hepatocyte 912 differentiation and promote progenitor phenotype.

913 Figure 6 – figure supplement 1: Verification of the RNA-sequencing results.

914 Figure 7: Yap sensitizes hepatocytes to TRAIL-mediated apoptosis. 915 Supplementary File 1 – List of Primer Sequences

916

917

28

Figure 1

A)

DAPI GFP Yap Merge

50 um

B) C) Gene expression in sorted Hepatocytes (qPCR)

25 0.035% 6.83% Hi WT 0.00% 30.6% Med 20 0.035% 42.4% Lo Neg 99.0% 10.8% Neg 15 Lo Med 10 Hi 5 GFP

mRNA fold induction 0 Yap Ctgf 7AAD Figure 2

A) Liver/Body weight ratio B) Total hepatocyte number **** N.S. 8 150 n=11 n=33 n=24 6 100 4 n=14 50 2

Liver/body weight (%) 0 0 WT KI WT KI

Hepatocyte number/liver (million)

C) D) Change of GFPhi fraction over time (flow cytometry) N.S. N.S. WT KI 40 n=11 n=19 n=11 30

20

100 um 10

0 % of GPFhi hepatocytes GPFhi % of 123 Age (Month)

E) F) Proliferation (BrdU incorporation) Cell death (TUNEL assay)

* ** 10 * 2.5 n=16 8 2.0 6 N.S. 1.5 4 n=16 1.0 n=8 2 0.5 Fold (v.s. Ave. WT) 0.0 % of total Hepatocytes 0 WT KI WT Lo Hi KI WT KI GFP BrdU Figure 3 A) B) %GFPhi cell at different locations (IHC) CCl4 150 SS Day 4 Day 14 **** 100

50 % GFP+ hepatocytes % GFP+ 0

Outside Damage site

C) D) GFP+ vs. GFP- Change of GFPhi cell fraction (BrdU incorporation) (Flow cytometry) 80 **** ** N.S. 60 60 ** ** n=5 n=11 40 N.S. 40 n=31 n=11

20 20

% of Brdu+ hepatocytes Brdu+ % of 0 0 % of GFP high hepatocytes % of GFP- GFP+ GFP- GFP+ SS D4 D14 LPS

Damage site Outside CCl4 Figure 4

A) B) WT YapKO WT YapKO Gs Gs BrdU

CV CV

100um 100um

C) D) Proliferation after CCl4 damage Sirius Red staining at different locations (BrdU incorporation) WT YapKO

80 **** **** **** 60 CCl4 **** 40 * 100 um * 20 % of Brdu+ hepatocytes Brdu+ % of 0 SS WT KI KO WT KI KO Damage site Outside Figure 5

A) B) BrdU Proliferation (BrdU incorporation) GFP

**** 50 WT ****

40 20 um 30 **

20

10 % of total Hepatocytes KI 0 WT Lo Hi

C) D) Hepatocyte number

50 WT KI * n=4 40 n=4 30

20

100 um 10

0 Hepatocytes number / liver (million) WT KI Figure 6

Liver specific A) H&E B) SS CCl4 WTLo Hi WT Lo Hi

KI

200um XP 100

WT

200um XP 200

C) D) 300 4 days after damage 4d after CCl4 WT KI

WT GS 400

Yap YapKI b-Actin

í í 0 1 2

Log2(Expression/WT)

E) F) G) Embryo specific Progenitor markers WT KI SS CCl 4 SS CCl4 0.4% 3.% WTLo Hi WT Lo Hi WTLo Hi WT Lo Hi Cd44 Afp Gpc3

Prom1 EpCam Igdcc4 GFP 100 Epcam H) Krt7 Quantification Krt19 ** Bex1 1000000 n=6 Cd24a 800000 í í 0 1 2 600000

Log2(Expression/WT) í 0  Log (Expression/WT) 400000 2 n=8 200000

Cell number/liver 0 WT KI Figure 7 A) B) Cell death on day4 after CCl4 damage (TUNEL assay) Gene upregualted after liver damage 25 10 **** WT 20 8 Hi WT CCl4 15 Hi 6

10 4 N.S.

5 2 level Relative mRNA

Positive cell per 20x view per cell Positive 0 0 WT KI WT KI Dr5 Bok Bax Damage site Outside Ngfrap1

C) D)

Dr5 mRNA levels Dr5 mRNA levels (qPCR, whole liver tissue) (qPCR, sorted hepatocytes) 0.010 0.008 * WT 0.008 * Lo 0.006 Hi 0.006 0.004 0.004 KI * 0.002 0.002

mRNA fold (v.s. ActB) WT mRNA level (v.s. ActB) ActB) (v.s. mRNA level 0.000 0.000 02468 SS CCl4, day4 Days after CCl4

E) In vitro Trail treatment (Cell titer blue, Sorted hepatocytes)

1.5 High

Low CCl4 1.0 WT

High 0.5 Cell viability **** Low SS WT 0.0 0 500 Trail dose (ng/ml)