www.sciencetranslationalmedicine.org/cgi/content/full/10/454/eaan1230/DC1

Supplementary Materials for

TGFβ inhibition restores a regenerative response in acute liver injury by

suppressing paracrine senescence

Thomas G. Bird*, Miryam Müller, Luke Boulter, David F. Vincent, Rachel A. Ridgway, Elena Lopez-Guadamillas, Wei-Yu Lu, Thomas Jamieson, Olivier Govaere, Andrew D. Campbell, Sofía Ferreira-Gonzalez, Alicia M. Cole,

Trevor Hay, Kenneth J. Simpson, William Clark, Ann Hedley, Mairi Clarke, Pauline Gentaz, Colin Nixon, Steven Bryce, Christos Kiourtis, Joep Sprangers, Robert J. B. Nibbs, Nico Van Rooijen, Laurent Bartholin, Steven R. McGreal, Udayan Apte, Simon T. Barry, John P. Iredale, Alan R. Clarke, Manuel Serrano, Tania A. Roskams, Owen J. Sansom, Stuart J. Forbes

*Corresponding author. Email: t.bird@beatson.gla.ac.uk

Published 15 August 2018, Sci. Transl. Med. 10, eaan1230 (2018)

DOI: 10.1126/scitranslmed.aan1230

The PDF file includes:

Fig. S1. Senescence markers in human acute liver disease. Fig. S2. Senescence in the acute CCl4 model. Fig. S3. Senescence in the acute acetaminophen model. Fig. S4. Senescence in acute dietary models of liver injury. Fig. S5. Hepatocyte Mdm2 deletion model. Fig. S6. Non–cell-autonomous senescence marker induction. Fig. S7. Hepatocyte TGFβ pathway activation model. Fig. S8. Hepatocyte TGFβ pathway promotes hepatic TGFβ ligand production. Fig. S9. TGFβ pathway activity in the acetaminophen model. Fig. S10. Serial sections of the TGFβ pathway and senescent hepatocytes. Fig. S11. Macrophage recruitment, TGFβ secretion, and induced senescence. Fig. S12. TGFβR1 inhibition in acute and chronic CCl4 models. Fig. S13. Genetic deletion of hepatocyte TGFβR1 in the acetaminophen model. Fig. S14. Therapeutic TGFβR1 inhibition in the acetaminophen model. Fig. S15. Schematic representation of paracrine-induced senescence in acute liver injury. Fig. S16. Ductular reaction responses in murine models and human disease. Table S2. RNA-seq gene: Hallmarks. Table S3. RNA-seq GSEA: Selected ranked hallmarks and OIS signature.

Other Supplementary Material for this manuscript includes the following: (available at www.sciencetranslationalmedicine.org/cgi/content/full/10/454/eaan1230/DC1)

Table S1 (Microsoft Excel format). Source data for Fig. 2.

Table S4 (Microsoft Excel format). Source data for Fig. 3. Table S5 (Microsoft Excel format). Source data for Fig. 4. Table S6 (Microsoft Excel format). Source data for Fig. 5. Table S7 (Microsoft Excel format). Source data for Fig. 6.

(A) IHC detection of DcR2, γH2Ax and SA-Gal and comparison of p21/DcR2 and γH2Ax in serial sections in human liver specimens obtained from normal liver and from patients transplanted for FHF. (B) Representative images of IHC for p16INK4a and Ki67 in the cases series of human sub-massive necrosis described in Figure 1. Scale bars = 50µm.

Supplementary Materials:

Fig. S1. Senescence markers in human acute liver disease.

CCl4 induced acute liver injury; 1mg/kg CCl4 given i.p. to male C57BL/6J mice with harvest at indicated time points; BrdU administered 2 hours prior to harvest. (A) H&E and IHC detection of p21, γH2Ax and IL1α. (B) Dual IHC for p21 and BrdU and HNF4α; individual channel for images shown in Fig. 2. (C) Serial sections with consecutive IHC staining for γH2Ax, p21, BrdU and Ilα. Scale bars = 50µm.

Fig. S2. Senescence in acute . the CCl model4

C57BL/6J mice treated with 350mg/kg acetaminophen after ten hour of fasting were analysed by serology, qRT-PCR and IHC. (A) Serum Alanine transaminase (ALT) and total bilirubin following acetaminophen injury were analysed (n= 5 each time point; except for ALT where n = 3 as two samples were of insufficient volume for serial dilution over 2000U/L); p values = 0.031 and 0.008 day one vs. baseline for ALT and Bilirubin respectively using one way ANOVA with Dunn’s multiple comparison test. (B) p21 expression by qRT-PCR over the injury time course (n= 3 each time point). (C) IHC for hepatocellular p21 following acetaminophen. (D) IHC for p16, senescence-associated heterochromatic foci marker HMGA, and γH2Ax in acetaminophen treated mice compared to healthy controls. (E) SA-βgal in perinecrotic hepatocytes four days following acetaminophen. (F) IHC for IL1α 2 and four days versus following acetaminophen and untreated controls. (G) Individual channels for p21/BrdU and p21/HNF4α dual IHC for Fig. 2B. (H) Serial sections from liver two days following acetaminophen, with consecutive IHC staining for γH2Ax, p21, BrdU and Ilα. (I) IHC for p21 in p21KO mice following acetaminophen. Serum analysis for ALT, AST and Bilirubin, two days following acetaminophen induced liver injury (350mg/kg). Scale bars = 50µm.

Fig. S3. Senescence in acute acetaminophen model. the

Fig. S4. Senescence in acute dietary models of liver injury.

Dietary models of murine acute injury models: CDE diet (four days) or DDC diet (18 days) result in over-expression of p21 by hepatocytes as detected by IHC. Scale bars = 50µm.

(A) Efficiency of Mdm2 assessed by p53 IHC in response to βNF dose escalation in untreated and single 20, 40 and 80mg/kg βNF i.p. injection treated AhCre+ Mdm2fl/fl mice; arrows

highlight p53low hepatocytes. (B) In the Mdm2Hep model (AhCre+ Mdm2fl/fl mice, versus AhCre- Mdm2fl/fl control mice, given 3x80mg/kg βNF i.p.) hepatocellular expression of senescence markers p21 and γH2Ax, and SASP component IL1α are present after 2 days. (C) SA-βgal widely detectable after four days. (D) Functional assessment of hepatocyte senescence by application of a hepatocellular mitogen cocktail (HGF 250µg/kg and T3 4mg/kg) three days following βNF induction to both p21WT AhCre+ Mdm2fl/fl and p21KO AhCre+ Mdm2fl/fl mice, followed by tissue analysis the following day. Representative images of dual IHC for CYP2D6 (hepatocyte marker) and BrdU (proliferation) analysed by confocal microscopy. (E) Quantification of total BrdU+ hepatocytes performed in WT, p21WT AhCre+ Mdm2fl/fl and p21KO AhCre+ Mdm2fl/fl mice; results expressed as BrdU+/CYP2D6+ cells/total CYP2D6+ cells; n≥3 each group; p = 0.0065 One way ANOVA with p values shown representing Bonferroni’s multiple comparison. (F) Representative images of mitogen effects and p53 status of BrdU+

hepatocytes in WT, Mdm2Hep and p21KO Mdm2 animals; (channels shown in Fig. S5D). Proliferation of p53+ cells in the Mdm2 model as assessed by multicolour IHC (BrdU, p53,

CYP2D6 and DAPI). (G) ISH by RNAscope for TGFβR1 and TGFβ1 ligand in partial Mdm2 model, versus AhCreWT Mdm2fl/fl control. TGFβR1 expression by hepatocytes – highlighted areas. TGFβ1 ligand expression by non-parenchymal cells with a monocyte-like appearance – highlighted areas. Scale bars = 50µm. (H) Gene expression in whole liver quantified by qRT-

PCR for Tgfbr1 and its ligands Tgfb1 and Tgfb2 and Tgfb3 following partial Mdm2 and referenced to baseline control; P values = one way ANOVA, n= 6,3,3,5,5,5 for baseline, days as indicated, Bonferroni's Multiple Comparison Test performed versus baseline confirm P<0.05 at day five for each; others >0.05). (I) Quantification of liver TGFβ1 by ELISA (four day time point, n =6); p value = two tailed Mann-Whitney test.

Fig. S5. Hepatocyte deletion model. Mdm2

(A) Induction of partial Mdm2, with reduced dose βNF (20mg/kg), results in hepatic p21 up-regulation by qRT-PCR; logarithmic plot, n = 5 mice at each time point vs 6 at baseline, p = 0.018 one way ANOVA. IHC for p16 in AhCre+ Mdm2flox/+ and control mice treated with 20mg/kg βNF. (B) Induction of AhCre+ Mdm2fl/fl mice using reduced dose (10mg/kg βNF) followed by IHC and hepatocyte quantification by p53/p21; 25 fields per mouse, n = 4 each group. p53 up-regulation by hepatocytes with zonal clustering and associated zonal clustering of non-cell-autonomous p21 expression; arrowheads highlight p21low/p53low hepatocytes clustering in zones in which p53hi hepatocytes are absent. (C) Recombination of hepatocytes four days following AAV8-TBG-Cre administration (2.5x1011 GC) as assessed by nuclear p53 hyper-expression versus control AAV8-TBG-null vector, quantified in 10 fields in n = 4 mice each group. Expressed as % total hepatocytes p21 hyper-expression was detected in p53high and p53low hepatocytes in 73.3% and 12.6% respectively. Quantification of cell-autonomous and non-cell-autonomous p21 expression based upon p21/p53 IHC, results expressed as hepatocytes per field quantified by p53low/p21low, p53hi/p21hi and p53low/p21hi expression - no p53hi/p21low hepatocytes were observed, p = 0.015, Mann-Whitney; n = 4 mice each group. Individual and merged channels for p53 (green), p21 (red) and DAPI (blue) in AAV8 control (arrows highlight non-cell-autonomous p21 expression by hepatocytes in response to control vector) and AAV8-TBG-Cre treated animals (arrows highlight non-cell-autonomous p21

expression). (D) IHC for pSMAD3 in partial Mdm2Hep model with vehicle control or SB525334, see Fig. 3H for experimental details. (E) Quantification of p53+ hepatocytes

revealed no detectable effect by SB525335 upon recombination in the partial Mdm2Hep model; n = 3 vs. 4. Mean ± SEM. Scale bars = 50µm.

Fig. S6. Non-cell-autonomous senescence marker induction.

Fig. S7. Hepatocyte TGFβ pathway activation model.

(A) Hepatocellular TGFβR1 activity was induced using AAV8-TBG-Cre (2x1011 GC/mouse) mediated recombination of LSL-TGFβR1-CA; see Fig. 3F. Four days following AAV8 vector administration the livers were harvested and analysed histologically. H&E shows evidence of midlobular liver injury and inflammatory infiltrate dense with F4/80+ macrophages. RNAscope ISH for SMAD7 transcription shows up-regulation of this TGFβ target. IHC for pSMAD3 also shown. (B) Serum analysis for bilirubin and ALT from mice four days following AAV8-TBG-Cre injection (2x1011GC) to WT controls or LSL-TGFβR1-CAHet mice or AAV8-TBG-Null injection (2x1011GC) to LSL-TGFβR1-CAHet mice. p values = two-tailed Mann-Whitney. Scale bars = 50µm.

Fig. S8. Hepatocyte TGFβ pathway promotes hepatic TGFβ ligand production.

Analysis of livers four days following AAV8-TBG-Cre or AAV8-TBG-Null control vector (2x1011 GC/mouse) delivery to LSL-TGFβR1-CA mice. (A) Hepatic TGFβ1 ligand content assessed by ELISA; p value = two tailed Mann-Whitney, n = 3 vs. 7. (B) Detection of hepatic TGFβ1 ligand using RNAscope ISH liver. Up-regulation of TGFβ1 ligand production by non-parenchymal cells with morphology consistent with hepatocytes.

(A) Time course of SMAD7 expression in the pericentral area following acetaminophen injury compared to baseline; as assessed by ISH using RNAscope. (B) Serial section analysis for the relationship between p21 expression and TGFβR1 receptor in acute liver injury models (acetaminophen toxicity – day four and CDE diet – day 4) and healthy controls. (C) Immunohistochemical staining for expression of the senescence marker p21 and TGFβR1 in mouse liver 2 days after acetaminophen induced liver injury.

Fig. S9. TGFβ pathway activity in acetaminophen model. the

Fig. S10. Serial sections of TGFβ pathway and senescent hepatocytes.

Serial sections from baseline, twelve hours and two days following 350mg/kg acetaminophen were analysed using RNAscope ISH for SMAD7 (section 1), TGFβR1 (section 3) and TGFβ1 ligand (section 4) and IHC for p21 (section 2). Following alignment perinecrotic areas were assessed for co-localization of markers in adjacent sections. Arrows indicate equivalent area in serial sections.

the

Fig. S11. Macrophage recruitment, TGFβ secretion and induced senescence.

(A) Expression of the chemokines CX3CL1 and CCL2 and leukocyte associated chemokines receptors CCR2 and CX3CR1 assessed by qRT-PCR in whole liver four days following

MdmHep previously; data normalized to WT animals; p values = two tailed T test, n= 6 vs. 5 each group. (B) Four days following i.v. injection of 2x1011GC AAV8-TBG-Cre to WT and LSL-TGFβR1-CAHet mice livers were assessed for TGFβ1 ligand expression by RNAscope. Upon activation of hepatocellular TGFβR1 signalling non-parenchymal cells with a monocyte-like appearance in the midlobular areas up-regulate TGFβ1 ligand expression. (C) Partial

,

MdmHep was performed followed by antibody mediated CCL2 inhibition. (D) IHC for p53

was quantified to assess MdmHep recombination efficiency; p = Mann-Whitney; n= 3 vs. 3

mice. (E) SA-βgal detection in CCL2 inhibitor and control partial MdmHep livers. (F)

Liposomal clodronate was used to deplete macrophages three days following partial MdmHep and analysis with dual IHC for macrophages (F4/80) and ductular expansion (panCK) was performed at day five confirming depletion of macrophages. Following IHC, non-cell-autonomous p21 expression by hepatocytes (p53-/p21+) were identified (G) together with proliferating (Ki67+) hepatocytes (H); p values = T test; n= 4 mice each group. Following

liposomal clodronate depletion of hepatic macrophages in the partial Mdm2 model hepatocellular proliferation remained restricted to the p53low hepatocyte population (I) and

MdmHep efficiency (IHC assessment of p53+ hepatocytes) was unaffected (J), p = two tailed T test; n= 4 each group. Data presented as mean ± SEM. Scale bars = 50µm. (K) Serum analysis two days following acetaminophen injection in LysMCre+ and LysMCreWT mice (both TGFβfl/fl); p values = two tailed Mann-Whitney; n=6 vs. 4 . (L) BrdU analysis at the same time point shows perinecrotic proliferation in the LysMCre+ TGFβfl/fl animals.

Fig. S12. TGFβR1 inhibition in acute and chronic

(A) Acute pericentral necrosis induced by high dose CCl4 was combined with pharmacological inhibition of TGFβR1 via AZ12601011 commencing twelve hours after CCl4 administration. Histological analysis four days following CCl4 by IHC for p21 (B) and BrdU (C) was quantified; ten fields per mouse, n = 3-7, p values = two way ANOVA and Mann-Whitney respectively. (D) Serum analysis for bilirubin following TGFβR1 inhibition; n = 3-7 each time point; p = two way ANOVA. (E) Weekly doses of CCl4 were given for eight weeks following

hepatocellular TGFβR1 in AhCre+ TGFβR1fl/fl mice induced with 3x80mg/kg βNF with analysis performed 24 hours after the final dose of CCl4. CCl4 vehicle treated WT and

TGFβR1 controls are also shown. Reduced senescent hepatocytes are observed by IHC for

p21 in TGFβR1 mice (F) and quantified together with increased Ki67+ (G) hepatocytes; mean 128 hepatocytes per high power field. p values = T test, n = 4 vs. 4 each group. Mean ± SEM. Scale bars = 50µm

CCl4 models.

Fig. S13. Genetic deletion of hepatocyte TGFβR1 in acetaminophen model.

(A) Hepatocyte-specific deletion of TGFβR1 by 80mg/kg βNF given to AhCre+ TGFβR1fl/fl mice five days prior to 350mg/kg acetaminophen. (B) Hepatocellular p21 expression assessed by IHC. p21 quantified two days following acetaminophen; p values = Mann-Whitney, ten fields in n = 3 vs. 3 each time point. (C) Proliferating perinecrotic hepatocytes (BrdU by IHC) were observed and quantified; p values = Mann-Whitney, n = 3 each group.Inset highlights perinecrotic BrdU+ hepatocyte. (D) Necrosis assessed by H&E over time showed equivalent

early necrosis but enhanced resolution in TGFβR1Hep animals; hepatocytes per field used as surrogate for necrosis; p value = T test, with mean of ten high power fields per mouse and n = 3 mice per group.

the

Fig. S14. Therapeutic TGFβR1 inhibition in acetaminophen model.

Following therapeutic use of SB525334 commenced twelve hours post acetaminophen (350mg/kg) administration (see Fig. 8A) liver tissue was assessed for pSMAD3 and pSMAD2 (A) showing reduction in perinecrotic hepatocytes versus SB525334 vehicle treated controls. (B) Histological analysis and p21 IHC performed at day two following acetaminophen in SB525334 and vehicle treated controls. (C) Serum analysis of Aspartate Transaminase (AST), Alkaline Phosphatase (Alk Phos) and Albumin; p values = two way ANOVA with Bonferroni post-test; n= 8 mice each group. (D) IHC for γH2Ax shows localisation around areas of necrosis 48 hours following acetaminophen administration with reduction in senescence markers in SB525334 treated mice. (E) Therapeutic use of AZ12601011 commenced twelve hours post acetaminophen (450mg/kg) reduced expression of pSMAD3, pSMAD2 and p21. (F) Serum transaminases following AZ12601011 treatment after 450mg/kg acetaminophen; n = 9 and 5 (ALT and AST respectively) each group, p value = two tailed T test. Mean ± SEM. Scale bars = 50µm.

the

Fig. S15. Schematic representation of paracrine induced senescence in acute liver injury.

Liver injury leads to both a regenerative stimulus and also induction of local p21 mediated senescence. These processes are proportional to the severity of liver injury. The p53 pathways

is hyper-activated by MdmHep, resulting in p21 activation and injury. Injury-induced senescence is associated with local TGFβ ligand production, principally by monocyte/macrophages and both reinforces and spreads senescence via TGFβR1. Senescence and regenerative stimuli then play opposing roles in affecting liver regeneration. p21 is a canonical target of TGFβ. However we also observe TGFβ stimulated liver injury in our models.

-

Fig. S16. Ductular reaction responses in murine models and human disease.

Ductular reaction was quantified using IHC for panCK+ ductular cells in three models of

manipulation of hepatocellular senescence in vivo. (A) The partial Mdm2Hep model followed by treatment with the TGFβR1 inhibitor SB525334 vs. vehicle control; n=4 p value=Mann-

Whitney, see Fig. 3. (B) Following iterative CCl4 and TGFβR1Hep vs TGFβR1WT controls; n=4 p value=Two tailed T test, see Fig. S12. (C) Following macrophage depletion, using

liposomal clodronate vs. PBS control, in the Mdm2Hep model; n=4 p value=Two tailed T test, see Fig. S5. (D) In the human case series of submassive necrosis (Fig. 1), the cases were also assessed for expansion of CK19+ ductular expansion, * denotes p<0.05.

Table S2. RNA seq ene: Hallmarks

Table S1. RNAseq Gene: Hallmarks

Rank 12hr 24hr 36hr 48hr 72hr

1 TNFA_SIGNALING_VIA_NFKB IL6_JAK_STAT3_SIGNALING E2F_TARGETS E2F_TARGETS G2M_CHECKPOINT

2 IL6_JAK_STAT3_SIGNALING TNFA_SIGNALING_VIA_NFKB G2M_CHECKPOINT G2M_CHECKPOINT E2F_TARGETS

3 INFLAMMATORY_RESPONSE UNFOLDED_PROTEIN_RESPONSE MITOTIC_SPINDLE EPITHELIAL_MESENCHYMAL_TRANSITION MITOTIC_SPINDLE

4 MYC_TARGETS_V1 MYC_TARGETS_V1 EPITHELIAL_MESENCHYMAL_TRANSITION MITOTIC_SPINDLE EPITHELIAL_MESENCHYMAL_TRANSITION

5 UNFOLDED_PROTEIN_RESPONSE INFLAMMATORY_RESPONSE INFLAMMATORY_RESPONSE ALLOGRAFT_REJECTION INFLAMMATORY_RESPONSE

6 MYC_TARGETS_V2 MYC_TARGETS_V2 KRAS_SIGNALING_UP INFLAMMATORY_RESPONSE ALLOGRAFT_REJECTION

7 HYPOXIA DNA_REPAIR ALLOGRAFT_REJECTION INTERFERON_GAMMA_RESPONSE HEDGEHOG_SIGNALING

8 MTORC1_SIGNALING TGF_BETA_SIGNALING TNFA_SIGNALING_VIA_NFKB KRAS_SIGNALING_UP UV_RESPONSE_DN

9 TGF_BETA_SIGNALING COMPLEMENT COMPLEMENT INTERFERON_ALPHA_RESPONSE IL2_STAT5_SIGNALING

10 P53_PATHWAY KRAS_SIGNALING_UP CHOLESTEROL_HOMEOSTASIS IL2_STAT5_SIGNALING KRAS_SIGNALING_UP

11 PI3K_AKT_MTOR_SIGNALING MTORC1_SIGNALING ANGIOGENESIS ANGIOGENESIS MYOGENESIS

12 APOPTOSIS HYPOXIA MYOGENESIS COMPLEMENT WNT_BETA_CATENIN_SIGNALING

13 ALLOGRAFT_REJECTION IL2_STAT5_SIGNALING APICAL_JUNCTION APOPTOSIS NOTCH_SIGNALING

14 COMPLEMENT P53_PATHWAY IL2_STAT5_SIGNALING APICAL_JUNCTION COMPLEMENT

15 ESTROGEN_RESPONSE_EARLY ALLOGRAFT_REJECTION INTERFERON_ALPHA_RESPONSE IL6_JAK_STAT3_SIGNALING APICAL_JUNCTION

Key

Inflammation

TGFb pathway

Cell cycle arrest

Time point post acetaminophen 350mg/kg; Positive Enrichment (p<0.05)

- g .

Table S3. RNA seq GSEA: Selected anked allmarks and OIS signature

12 24 36 48 72

ES 0.28354234 0.25651422 0.2146408 0.2223561 0.18611553

NES 2.8616543 2.502138 2.0496526 2.125652 1.6695813

p value <0.001 <0.001 0.01012146 0.00211864 0.03529412

ES 0.35122243 0.2403069 0.12090696 0.13592513 0.11025173

NES 5.3631372 3.6312175 1.7938054 2.0131876 1.6209431

p value <0.001 <0.001 0.01405623 0.00567108 0.03080082

ES 0.3902897 0.34995708 0.1496695 0.1965174 0.13624752

NES 4.032234 3.6313157 1.5308425 2.0385005 1.3962431

p value <0.001 <0.001 0.05672269 0.00378788 0.11368015

ES 0.25194177 0.21970662 0.15730469 0.22905323 0.20062086

NES 3.7773032 3.1617396 2.23048 3.3589034 2.6742294

p value <0.001 <0.001 <0.001 <0.001 <0.001

ES 0.28010803 0.2704541 0.13054506 0.1760646 0.18272972

NES 2.2798328 2.2871246 1.0839112 1.4281753 1.549105

p value <0.001 <0.001 0.32806325 0.10395011 0.05040323

ES ‐0.15037447 0.08383791 0.43522075 0.4621551 0.25964147

NES ‐2.3362544 1.2552676 6.4885345 6.9313035 3.8531332

p value <0.001 0.1875 <0.001 <0.001 <0.001

ES ‐0.05900207 0.04774805 0.34803554 0.3575249 0.2900816

NES ‐0.901164 0.7092555 5.2078648 5.4151416 4.401901

p value 0.6051081 0.8330097 <0.001 <0.001 <0.001

ES 0.07861533 0.0805331 0.19944175 0.25522602 0.21748185

NES 1.2293746 1.2386999 3.0204787 3.9767787 3.319795

p value 0.20874752 0.19038077 <0.001 <0.001 <0.001

Highlights p>0.05

G2M_CHECKPOINT

MITOTIC_SPINDLE

Hallm

arks

Acetaminophen timepoint

IMR90 ER:KRAS (OIS)

TNFA_SIGNALING_VIA_NFKB

IL6_JAK_STAT3_SIGNALING

INFLAMMATORY_RESPONSE

TGF_BETA_SIGNALING

E2F_TARGETS

- r h.