Estrogens counteract the profibrotic effects of TGF-β and their inhibition exacerbates experimental dermal fibrosis

Journal Pre-proof Estrogens counteract the profibrotic effects of TGF-β and their inhibition exacerbates experimental dermal fibrosis Jérôme Avouac, Sonia Pezet, Virginie Gonzalez, Léa Baudoin, Anne Cauvet, Barbara Ruiz, Gonçalo Boleto, Marie Laure Brandely, Manon Elmerich, Yannick Allanore PII:

S0022-202X(19)33206-3

DOI:

https://doi.org/10.1016/j.jid.2019.07.719

Reference:

JID 2132

To appear in:

The Journal of Investigative Dermatology

Received Date: 21 December 2018 Revised Date:

24 July 2019

Accepted Date: 25 July 2019

Please cite this article as: Avouac J, Pezet S, Gonzalez V, Baudoin L, Cauvet A, Ruiz B, Boleto G, Brandely ML, Elmerich M, Allanore Y, Estrogens counteract the profibrotic effects of TGF-β and their inhibition exacerbates experimental dermal fibrosis, The Journal of Investigative Dermatology (2019), doi: https://doi.org/10.1016/j.jid.2019.07.719. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 The Authors. Published by Elsevier, Inc. on behalf of the Society for Investigative Dermatology.

Estrogens counteract the profibrotic effects of TGF-β β and their inhibition exacerbates experimental dermal fibrosis Jérôme Avouac1,2, Sonia Pezet1, Virginie Gonzalez1, Léa Baudoin1, Anne Cauvet1, Barbara Ruiz1, Gonçalo Boleto1, Marie Laure Brandely3, Manon Elmerich1, Yannick Allanore1,2 1

Université Paris Descartes, Sorbonne Paris Cité, INSERM U1016 and CNRS UMR8104, Institut Cochin, Paris, France 2

Université Paris Descartes, Sorbonne Paris Cité, Service de Rhumatologie A, Hôpital Cochin, Paris, France.

3

GH Hôpitaux Universitaires Paris Centre, Service de Pharmacie Clinique, Hôpital Cochin, Paris, France.

Corresponding author: Pr. Jérôme Avouac Service de Rhumatologie A, Hôpital Cochin Université Paris Descartes 27 rue du Faubourg Saint-Jacques 75014 Paris, France Telephone: + 33 1 58.41.25.86 Fax: + 33 1 58.41.26.24 e-mail: [email protected] Short title: Role of estrogens in fibrosis Abbreviations: SSc: Systemic Sclerosis TGF-β β : Transforming Growth Factor-ß ERKO-α α: Estrogen Receptor-α Knock Out TSK-1: Tight Skin-1 i.p.: intraperitoneal COL1A1: Collagen 1A1 COL1A2: Colagen 1A2 α-SMA: α-Smooth Muscle Actin

1

ABSTRACT Systemic sclerosis (SSc) primarily affects women. This sex bias raises the role of female hormones on fibrosis, which is largely unknown. Our aim was to evaluate the effects of estrogens in the development of experimental dermal fibrosis, in the mouse models of bleomycin-induced dermal fibrosis and tight skin (Tsk-1) mice, and on the activation of dermal fibroblasts by transforming growth factor-β (TGF-β). Estrogen inhibition, obtained through gene inactivation for the estrogen receptor-α (ERKOα) or treatment with tamoxifen, exacerbated skin fibrosis in the bleomycin model and in Tsk-1 mice. In dermal fibroblasts, treatment with 17-β-estradiol significantly decreased the stimulatory effects of TGF-β on collagen synthesis and myofibroblast differentiation, decreased activation of canonical TGF-β signalling, and markedly reduced the expression of TGF-β target genes. Tamoxifen reversed the inhibitory effects of estrogens by restoring Smad2/3 phosphorylation and TGF-β-induced collagen synthesis. Our results demonstrate a beneficial effect of estrogens in dermal fibrosis. Estrogens reduce TGF-β-dependent activation of dermal fibroblasts and estrogen inhibition leads to a more severe experimental dermal fibrosis. These findings are consistent with the prominent development of SSc in postmenopausal women and the greater severity of the disease in men.

2

INTRODUCTION Fibrosis results from an imbalance between synthesis and degradation of extra-cellular matrix components and leads to scarring of various tissues. It is the hallmark of systemic sclerosis (SSc), which is a prototypical fibrotic disease affecting both the skin and internal organs (Gabrielli et al., 2009; Varga and Abraham, 2007; Wynn and Ramalingam, 2012). The accumulation of extracellular matrix components in SSc is caused by an increased production by activated fibroblasts (Varga and Trojanowska, 2008). The profibrotic cytokine transforming growth factor-ß (TGF-ß) is one of the key mediators of fibroblast activation in SSc (Garrett et al., 2017; Varga and Pasche, 2009; Varga et al., 2017). However, the intracellular signalling cascades by which this cytokine stimulate the production of extracellular matrix are incompletely understood. SSc shows a strong and well-established gender bias, with an overall ratio (women to men) of approximately 6:1, reaching 10:1 at childbearing period (Elhai et al., 2016, Whitacre, 2001). Although SSc is more frequent in women, data from the European Scleroderma Trial and Research (EUSTAR) group have shown that men with SSc have a more severe phenotype and the prognosis of SSc is worse in men (Elhai et al., 2016). Consistent with this observation, an analysis of the effect of gender on the severity of experimental dermal fibrosis showed that male Balb/C, C57BL/6, DBA/2 mice had a trend to develop more pronounced fibrosis than females (Ruzehaji et al., 2015). Despite the well-established sex bias in SSc, the molecular bases of this critical aspect remain poorly understood. In particular, the role of sex hormones in modulating fibrosis is not well defined. Indeed, data on the role of sex hormones in the development of fibrosis are complex and contradictory among the different tissues studied (Murase et al., 2012; Tofovic et al., 2009; Voltz et al., 2008; Zhang and Wu, 2013). More specifically in SSc, available data on the role of sex hormones in the promotion of fibrosis are scarce and heterogeneous. Some studies have shown that estrogens induce fibroblast

3

dysfunction and increase the production of extracellular matrix proteins, whereas some others have reported a decreased collagen production upon treatment with estradiol (Aida-Yasuoka et al., 2013; Kashnikova et al., 1991; Soldano et al., 2010). Thus, to explore in more details the implications of sex hormones in the pathogenesis of dermal fibrosis, our aim was to evaluate the effects of estrogens i) in the development of experimental dermal fibrosis and ii) on the pathological activation of dermal fibroblasts induced by TGF-β.

RESULTS Increased susceptibility of ERKO-α α mice to bleomycin-induced dermal fibrosis To evaluate the role of estrogen inhibition in fibrosis, ERKO-α mice and control C57BL/6 female mice were challenged with bleomycin. Skin architecture and dermal thickness did not differ between ERKO-α mice and C57BL/6 mice injected with NaCl (Figure 1a). Upon bleomycin injections, the mean increase in dermal thickness was 55% in C57BL/6 mice, as compared with 87% in ERKO-α mice. Thus, dermal thickening upon bleomycin challenge was increased by 21% in ERKO-α mice compared to C57BL/6 mice (P=0.034) (Figures 1a and 1b). Consistent with increased dermal thickening, increased accumulation of collagen upon bleomycin challenge was observed on trichrome-stained skin sections of ERKO-α mice (Figure 1c). In addition, the hydroxyproline content and the number of myofibroblasts in lesional skin of ERKO-α mice upon challenge with bleomycin were significantly increased by 39% (P=0.042) and 36% (P=0.010) respectively, as compared to the skin of C57BL/6 mice (Figures 1d-f). To evaluate whether estrogen inhibition influence the outcome of bleomycin-induced fibrosis by regulating infiltration of inflammatory cells, we quantified the number of T cells, B cells, and monocytes in fibrotic skin. T cell counts were significantly

4

increased in ERKO-α mice compared to wild-type mice (Figures S1a). Conversely, the number of B cells (Figures S1b) and monocytes (Figure S1c) were similar between ERKO-α and wild type mice, upon bleomycin challenge.

Tamoxifen enhances bleomycin-induced dermal fibrosis To further investigate the effects of estrogen inhibition observed with the gene inactivation strategy, we evaluated the effects of the ER antagonist tamoxifen in the model of bleomycininduced dermal fibrosis. In mice challenged with NaCl, tamoxifen did not increase dermal thickness and collagen content compared to mice treated i.p. with corn oil (data not shown). In mice challenged with bleomycin, tamoxifen significantly increased dermal thickening by 28% (P=0.008) compared to mice injected with corn oil (Figures 2a and 2b). Analyses of the hydroxyproline content and of myofibroblast counts confirmed increased accumulation of collagen by 23% (P=0.002) and increased activation of fibroblasts by 22% (P=0.024) upon treatment with tamoxifen (Figures 2c-f).

Estrogen inhibition enhances the activation of canonical Smad signalling in bleomycininduced dermal fibrosis Given hat estrogen inhibition led to a more severe dermal fibrosis, we aimed to assess whether these effects were mediated by increased TGF-β signalling. To this end, we analyzed the nuclear levels of phosphorylated Smad2 / Smad3 (pSmad2/3) signalling in the lesional skin of bleomycin-treated mice. As expected, levels of pSmad2/3 were increased upon challenge with bleomycin (Figure S2a-d). Estrogen inhibition obtained by gene inactivation (ERKO-α mice) or by the use of tamoxifen further stimulated canonical Smad signalling with increased nuclear levels of pSmad2/3 compared to wild-type mice (Figures S2a-b) or shamtreated controls (Figures S2c-d).

5

Tamoxifen enhances skin fibrosis in the tight skin mouse model Since we demonstrated some profibrotic effects of estrogen inhibition by tamoxifen in a model of inflammation-driven dermal fibrosis, we next aimed to confirm these effects in Tsk1 mice, which represents a model of later and less inflammatory stages of SSc with endogenous activation of fibroblasts. Compared to control pa/pa mice, hypodermal thickness was increased by 278% in sham-treated Tsk-1 mice (P<0.001) and by 364% in tamoxifentreated Tsk-1 mice (P<0.001). Thus, hypodermal thickening was increased by 31% in tamoxifen-treated Tsk-1 mice compared to sham-treated Tsk-1 mice (P=0.028) (Figures 3a and 3b). Hydroxyproline content was also significantly increased in tamoxifen-treated Tsk-1 mice by 16% (P=0.015) and 541% (P<0.001) compared to sham-treated Tsk-1 mice and pa/pa mice, respectively (Figure 3c).

Estrogens abrogate the profibrotic effects of TGF-β β in dermal fibroblasts Since enhanced TGF-β signalling was observed through estrogen inhibition in experimental dermal fibrosis, we next aimed to determine whether estrogens could antagonize the profibrotic effects directly induced by TGF-β. Incubation of SSc fibroblasts with 10µg/ml of 17-β-estradiol for 24 hours did not modify collagen production and differentiation of resting fibroblasts to myofibroblasts (Figure 4a-f). However, treatment of SSc fibroblasts with 17-βestradiol for 24 hours markedly decreased the stimulatory effects of TGF-β on COL1A1 and COL1A2 mRNA levels (Figures 4a-b). COL1A1 and COL1A2 mRNA levels were reduced by 47% (P<0.001) and 34% (P=0.025) respectively upon exposition to 17-β-estradiol. Consistent with decreased mRNA levels, the release of collagen in fibroblast culture supernatants was also significantly reduced by 43% (P=0.023) (Figure 4c). To study whether the effects of 17-β-estradiol were mediated by ER-α or ER-β, SSc dermal fibroblasts were treated with specific ER-α or ER-β inhibitors (MPP and PHTPP respectively)

6

at a concentration of 10-6M. Treatment of SSc fibroblasts with MPP or PHTPP prior to addition of 17-β-estradiol and stimulation with TGF-β was both associated with a similar nonsignificant reduction of collagen synthesis. These results support that the potent antifibrotic effects of 17-β-estradiol in dermal fibroblasts requires both ER-α and ER-β (Figure S3a-c). Regarding myofibroblast differentiation, the stimulatory effect of TGF-β on the expression of α-SMA mRNA was markedly reduced by 17-β-estradiol (Figure 4d). The inhibitory effects of 17-β-estradiol were confirmed at the protein level by immunofluorescence and western blot analyses after 24 hours of incubation with 17-β-estradiol (Figures 4e-g). Similar findings were observed in control dermal fibroblasts subjected to TGF-β stimulation. Indeed, treatment of control fibroblasts with 17-β-estradiol potently reduced TF-β-induced collagen release (Figure S4a) and expression of α-SMA (Figure S4b-d). In line with these findings, the expression of the estrogen receptors ERα and ERβ was not significantly different between SSc and control fibroblasts (Figure S4e-f).

Estrogens inhibit the activation of canonical Smad signalling and reduce TGF-β β target gene expression in SSc dermal fibroblasts Since we showed that 17-β-estradiol reduced the response of dermal fibroblasts to TGF-β stimulation, we next focused on the effect of 17-β-estradiol on canonical Smad-dependent, and non-canonical signalling pathways. Western blot and immunofluorescence experiments showed that TGF-β caused a robust upregulation of pSmad2/3 in human SSc dermal fibroblasts (Figures 5a and 6a-b). A marked decreased pSmad2/3 expression induced by 17β-estradiol was observed after incubation of human SSc fibroblasts with 17-β-estradiol for 24 hours, whereas levels of total Smad2/3 remained unchanged. (Figures 5a and 6b). In addition, in the presence of the proteasome inhibitor MG132 at a concentration of 20 µM, 17-

7

β-estradiol failed to reduce the TGF-β-induced expression of pSmad 2/3 (Figure S5a-b). Conversely, the phosphorylation of TGFBR1, p38 and p44/42 in SSc fibroblasts by TGF-β was not potently inhibited by treatment with17-β-estradiol (Figures 5b-d). Taken together, these results support that estrogens decrease the profibrotic effects of TGF-β by interfering specifically with the canonical Smad-dependent pathway through a proteasome-dependent degradation of pSmad2/3.. We also examined whether estrogens affect the mRNA levels of the TGF-ß ⁄ Smad target genes plasminogen activator inhibitor-1 (PAI-1) and matrix metalloproteinase-9 (MMP-9). Consistent with the effects of estrogens observed on psmad2/3, TGF-β-induced expression of PAI-1 and MMP-9 was significantly decreased in SSc dermal fibroblasts after incubation with17-β-estradiol for 24 hours by 49% (P=0.010) and 56% (P<0.001), respectively (Figures 6c and 6d).

Tamoxifen does not directly activate SSc dermal fibroblasts, but reverses the inhibitory effects of estrogens on TGF-β β signalling and TGF-β β -induced collagen synthesis Since tamoxifen worsens dermal fibrosis in two complementary animal models of SSc, we next aimed to determine whether tamoxifen could contribute to the pathologic activation of SSc fibroblasts. Treatment of SSc dermal fibroblasts with 5 µg/ml of tamoxifen for 24 hours did not increase mRNA levels of COL1A1 and COL1A2 (Figures S6a-b). Consistent with this observation, release of collagen protein into cell culture supernatant was not modified upon treatment with tamoxifen (Figure S6c). Moreover, tamoxifen did not markedly amplify the profibrotic effects of TGF-β on collagen synthesis in SSc dermal fibroblasts (Figures S6a-c). In line with collagen production, tamoxifen failed to induce the differentiation of SSc fibroblasts into myofibroblasts. The mRNA levels (Figure S6d) and protein expression (Figures S6e-f) of α-SMA were not modified in SSc fibroblasts upon tamoxifen treatment. In

8

addition, tamoxifen did not significantly amplify the effects of TGF-β on α-SMA expression in SSc dermal fibroblasts (Figures S6d-f). Based on the effects observed with tamoxifen on experimental dermal fibrosis, and given that tamoxifen did not enhance the effects of TGF-β in dermal fibroblasts, we made the hypothesis that tamoxifen could reverse the inhibitory effects of estrogens on TGF-β signalling and TGFβ-induced collagen synthesis. We first observed that tamoxifen did not modify TGF-βinduced protein expression of pSmad2/3 and mRNA levels of PAI-1 and MMP-9 (Figures 6b-d). However, tamoxifen had the capacity to counteract the inhibitory effects of estrogens on TGF-β signalling. Incubation of SSc dermal fibroblasts with tamoxifen 4 hours prior to treatment with 17-β-estradiol restored TGF-β-induced phosphorylation of Smad2/3 (Figures 6a-b). In addition, the sensitivity of PAI-1 and MMP-9 to TGF-β was re-established in fibroblasts incubated with tamoxifen prior to treatment with 17-β-estradiol (Figures 6c and 6d). Following these results, we sought to determine whether tamoxifen had the capacity to reduce the inhibition mediated by estrogens on collagen synthesis. Indeed, 17-β-estradiol failed to decrease COL1A1 and COL1A2 mRNA levels in SSc dermal fibroblasts pre-treated with tamoxifen (Figures S7a-b). This result was confirmed with the measurement of the collagen release in cell culture supernatants (Figure S7c).

DISCUSSION Genetic or pharmacologic inhibition of estrogen signalling led to increased sensitivity to bleomycin-induced dermal fibrosis, with a similar extent for both strategies (20% increase of dermal thickness). This result is consistent with the prevention of bleomycin-induced fibrosis previously observed with 2-Methoxyestradiol (Zhu et al., 2015). Increased T-cell infiltrates observed in the lesional skin of ERKO-α mice support that antifibrotic effects of estrogens in

9

early stages of fibrosis are, at least partially, driven by inflammation. Pro-inflammatory effects of estrogen inhibition seem preferentially mediated by T cells, rather than B cells and macrophages, as observed in experimental autoimmune encephalomyelitis (Offner and Polanczyk, 2006). In line with our results, oestrogen inhibition was also associated with inflammation and fibrosis in the liver, leading to increased tumorigenesis (Wei et al., 2016). On the other hand, the herein exacerbation of skin fibrosis was also observed upon tamoxifen therapy in Tsk-1 mice, characterized by persistent overproduction of extracellular matrix proteins occurring in the absence of inflammatory infiltrates (Beyer et al., 2010). Additional mechanisms beyond inflammation may be suspected, such as a direct effect of estrogens on fibroblast activation (Zhu et al., 2015). Indeed, we observed that 17-β-estradiol inhibited the profibrotic effects of TGF-ß on collagen synthesis and myofibroblast differentiation, and these effects were mediated both by ER-α and ER-ß. Conversely, 17-β-estradiol did not alter collagen synthesis of resting SSc fibroblasts, which contrasts with the modulation of basal collagen synthesis by estrogens observed in two previous studies (Kashnikova et al., 1991; Soldano et al., 2010). 17-β-estradiol inhibited the profibrotic effects of TGF-β in dermal fibroblasts by repressing the TGF-β-induced canonical cascade, without interfering with Smad independent TGF-β signalling. A direct interaction between Smad2/3 and ER-α has been previously described, leading to enhanced degradation of Smad proteins via an ubiquitin-proteasome pathway (Ito et al., 2010; Matsuda et al., 2001). Consistent with this result, we observed that TGF-β response was not repressed by 17-β-estradiol after treatment of dermal fibroblasts with the proteasome inhibitor MG132. Tamoxifen potently prevented 17-β-estradiol-induced decreased phosphorylation of Smad2/3 in SSc dermal fibroblasts, and incubation of SSc dermal fibroblasts with tamoxifen prior to 17-β-estradiol treatment restored the profibrotic effects of TGF-β on collagen synthesis.

10

Despite the previously reported elevation of TGF-β production upon tamoxifen therapy (Thomas-Golbanov et al., 2003), tamoxifen alone failed to amplify TGF-β signalling or the transcription of TGF-β target genes, as previously reported (Goto et al., 2011), supporting that estrogen stimulation is required or tamoxifen to exert its action of competitive estrogen inhibitor. Taken together, our findings support the potential beneficial effects of estrogens in different stages of experimental dermal fibrosis. This is consistent with the improvement of hepatic fibrosis through estrogen therapy, with inhibition of hepatic stellate cell activation (Zhang and Wu, 2013). In lungs, exacerbation of bleomycin-induced pulmonary hypertension and fibrosis has been observed in estrogen-deficient rats (Tofovic et al., 2009) and male sex hormones have been shown to increase lung function impairment after bleomycin-induced pulmonary fibrosis (Voltz et al., 2008). On the other hand, oestrogen inhibition with tamoxifen improved several fibrotic diseases, like progressive renal disease, sclerosing cervicitis, or retroperitoneal fibrosis (Delle et al., 2012; Ruffy et al., 2006; Savelli et al., 1997; Thomas-Golbanov et al., 2003). Moreover, estrogens were implicated in left ventricular fibrosis and diastolic dysfunction exacerbation in a rat model of metabolic syndrome (Murase et al., 2012). Taken together, these data support that estrogens may differentially modulate fibrosis in different organs, and the heterogeneity of the results strongly suggest a tissue specificity of estrogens and sex hormones in fibrosis.

In conclusion, our results support that estrogens alleviate through ER-α stimulation the activated profile of SSc dermal fibroblasts by inhibiting TGF-β signalling (Figure S8). These antifibrotic effects of estrogens are repressed by the estrogen antagonist tamoxifen (Figure S8). These data are consistent with the amplification of TGF-β signalling observed in vivo through estrogen inhibition in the bleomycin mouse model and Tsk-1 mice, leading to the

11

exacerbation of fibrosis. Thus, our data indicate that estrogen inhibition amplifies the profibrotic effects of TGF-β in mouse models of skin fibrosis. Our results are consistent with the prominent occurrence of SSc in postmenopausal women and the greater severity of the disease in men.

MATERIAL AND METHODS Animals Six-week-old female ERKO-α mice on a C57BL/6 background were purchased from Jackson laboratory (Bar Harbor, ME). Five-week-old female tight skin-1 (Tsk-1) strains were also purchased from Jackson Laboratory and six-week old female wildtype C57BL/6 mice were purchased from Charles River (Chatillon-sur-Chalaronne, France). All animals were bred and maintained at the animal care facilities of Cochin institute (Cochin Hospital, France) and Montrouge dental university (Montrouge, France). All experimental procedures were conducted in compliance with Animal Health regulations and the local ethical committee approved all animal experiments.

Estrogen inhibition in the mouse model of bleomycin-induced experimental dermal fibrosis Skin fibrosis was induced in ERKO-α and C57BL/6 mice by local injections of bleomycin for 3 weeks; 100µl of bleomycin dissolved in 0.9% sodium chloride (NaCl) at a concentration of 0.5mg/ml was administered every other day by subcutaneous injections in defined areas of 1cm2 at the upper back. Subcutaneous injections of 100µl 0.9% NaCl were used as controls.

12

Bleomycin-induced dermal fibrosis in ERKO-α mice Four different groups, consisting of two groups of ERKO-α mice (total of 20 mice) and two groups of wild-type mice (total of 20 mice) were analyzed. One group of ERKO-α mice (n=12) and one group of wild-type mice (n=12) were challenged with bleomycin, whereas the remaining two groups (n=8 each) were injected with NaCl.

Evaluation of the effects of tamoxifen in bleomycin-induced dermal fibrosis Following the use of a gene inactivation strategy with the use of mice deficient for ER-α, we used a complementary targeting molecular strategy using tamoxifen, an ER antagonist. Tamoxifen was diluted in corn oil and kept in dark glass tubes at 4°C. Tamoxifen was administrated by intraperitoneal (IP) injections, at a dose of 400µg in a volume of 200µl of corn oil, every other day, as previously described (Sthoeger et al., 2003). One group of 14 mice was challenged with bleomycin injections for three weeks and was treated in parallel with IP injections of tamoxifen at a dose of 400µg (n=7) every other day. Three control groups were used: one group challenged with bleomycin and injected IP with corn oil (n=7), one challenged with NaCl and injected with tamoxifen at a dose of 400µg (n=7) every other day, and one challenged with NaCl and injected with corn oil (n=7).

Effects of tamoxifen in the tight skin mouse model One group of 5 Tsk-1 mice received IP injections of 400 µg of tamoxifen every other day for 5 weeks while another group of 5 Tsk-1 mice received IP injections of corn oil. A third group of 5 pa/pa mice on the same genetic background not carrying the tsk1 mutation was used as controls; these mice also received IP injections of corn oil. Treatment was started at the age of 5 weeks and mice were sacrificed after 5 weeks of treatment.

13

Histological analysis and Hydroxyproline assay These techniques have been previously reported (Avouac et al., 2013; Avouac et al., 2011; Avouac et al., 2012, Avouac et al., 2014, Desallais et al., 2014, Woessner and Boucek, 1961), and a full description is available in the Online Supplement.

Immunohistochemistry for α-SMA, CD3, CD22, CD68 Myofibroblasts were identified by staining for α-SMA as previously described (Rawson et al., 1965). Cells positive for α-SMA in mouse skin sections were detected by incubation with monoclonal mouse anti-human antibodies against anti-α-SMA. To quantify the numbers of infiltrating T cells, B cells and monocytes, skin sections were incubated with polyclonal rabbit anti-mouse antibodies for CD3, CD22 (Abcam, Paris, France), or with monoclonal mouse anti-mouse antibodies against CD68 (Novus Biologicals, Littleton, CO).

Fibroblast cultures Fibroblast cultures were obtained from skin biopsies of 8 SSc female patients who signed a written consent form approved by the local institutional review boards. All SSc patients fulfilled the criteria for SSc subsets as suggested by LeRoy et al (LeRoy et al., 1988). Human fibroblasts were prepared by outgrowth cultures from skin biopsy specimens and cultured in Dulbecco’s modified Eagle’s medium (DMEM)–Ham’s F-12 containing 10% heat-inactivated fetal calf serum (FCS), 25mM HEPES, 100units/ml penicillin, 100µg/ml streptomycin, 2mM L-glutamine, and 2.5µg/ml amphotericin-B (all from Invitrogen, Karlsruhe, Germany) as described (Avouac et al., 2012). In selected experiments, fibroblasts were stimulated with recombinant TGF-β for 24 hours (10ng/ml, PeproTech, Hamburg, Germany). Fibroblasts from passages 4–8 were used for the experiments. In addition, we used

14

tamoxifen (5 µg/ml), solubilized in methanol, 17-β-estradiol (10µg/ml), solubilized in PBS, the proteasome inhibitor MG132 (10 or 20 µ) solubilized in DMSO, the ER-α inhibitor MPP (10-6M) and the ER-β inhibitor PHTPP (10-6M) solubilized in DMSO (all from Sigma Aldrich, Saint-Quentin Fallavier, France) to modulate estrogens in dermal fibroblasts.

Quantitative real-time polymerase chain reaction (PCR) and Western blots The full description of these techniques is available in the Online Supplement.

Quantification of soluble collagen protein Total soluble collagen in cell culture supernatants was quantified using the SirCol collagen assay (Biocolor, Belfast, Northern Ireland), as previously described (Tomcik et al., 2013).

Immunofluorescence for phosphorylated smad-2/3, and α-smooth muscle actin (α-SMA) The expression of pSmad2/3 and α-SMA was detected by staining with polyclonal goat antihuman antibodies against phospho-smad-2/3 (Santa Cruz Biotechnology, reacts with human and mouse pSmad2/3) and monoclonal mouse anti-human antibodies against anti-α-SMA antibody (clone 1A4; Sigma-Aldrich), respectively (Avouac et al., 2012). Donkey anti-goat Alexa Fluor 594 and donkey anti-mouse Alexa Fluor 594-conjugated antibodies were used as secondary antibodies. Counterstaining was performed with DAPI (Santa Cruz Biotechnology) for 10 minutes at room temperature. Slides were then viewed by microscopy using appropriate fluorescence filters.

15

Statistical analysis All data are expressed as mean values ± SEM. Multiple group comparisons were analyzed by using post hoc Dunnett’s test. Unpaired or paired t-test was used for a two-group comparison. P<0.05 (all two-sided) was considered significant.

DATA AVAILABILITY STATEMENT : - The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. - All data generated or analysed during this study are included in this published article [and its supplementary information files].

ACKNOWLEDGMENTS: Plateformes d’Immuno-biologie (M. Andrieu), de Morphologie / Histologie (M. Favier) et d’imagerie cellulaire (P. Bourdoncle) de l’Institut Cochin, Paris, France and Prof. Catherine Chaussain, EA2496, Université Paris Descartes, Faculté de Chirurgie Dentaire, Montrouge, France

AUTHOR CONTRIBUTIONS: Conceptualization: J.A and Y.A conceived the study, and supervised its analysis. Resources: M.L.B provided bleomycin Investigation and Formal Analysis: S.P, V.G, L.B, A.C, B.R, G.B and M.E performed experiments. J.A, S.P, V.G and Y.A analysed data. Writing – original draft : J.A wrote the manuscript Writing – review & editing : J.A, S.P, V.G, L.B, A.C, B.R, G.B, M.L.B, M.E, Y.A revisited the manuscript and gave the final approval of the manuscript.

COMPETING INTERESTS: J. Avouac has/had consultancy relationship and/or has received research funding in relationship with the treatment of systemic sclerosis from Actelion, Roche, Pfizer and Bristol-Myers Squibb. Y. Allanore has/had consultancy relationship and/or has received research funding in relationship with the treatment of systemic sclerosis from Actelion, Bayer, Biogen Idec, Bristol-Myers Squibb, Genentech/ Roche, Inventiva, Medac, Pfizer, Sanofi/Genzyme, Servier and UCB.

16

References Aida-Yasuoka K, Peoples C, Yasuoka H, Hershberger P, Thiel K, Cauley JA, et al. Estradiol promotes the development of a fibrotic phenotype and is increased in the serum of patients with systemic sclerosis. Arthritis Res Ther 2013;15(1):R10. Avouac J, Elhai M, Tomcik M, Ruiz B, Friese M, Piedavent M, et al. Critical role of the adhesion receptor DNAX accessory molecule-1 (DNAM-1) in the development of inflammation-driven dermal fibrosis in a mouse model of systemic sclerosis. Ann Rheum Dis 2013;72(6):1089-98. Avouac J, Furnrohr BG, Tomcik M, Palumbo K, Zerr P, Horn A, et al. Inactivation of the transcription factor STAT-4 prevents inflammation-driven fibrosis in animal models of systemic sclerosis. Arthritis Rheum 2011;63(3):800-9. Avouac J, Palumbo K, Tomcik M, Zerr P, Dees C, Horn A, et al. Inhibition of activator protein 1 signaling abrogates transforming growth factor beta-mediated activation of fibroblasts and prevents experimental fibrosis. Arthritis Rheum 2012;64(5):1642-52. Avouac J, Palumbo-Zerr K, Ruzehaji N, Tomcik M, Zerr P, Dees C, et al. The nuclear receptor constitutive androstane receptor/NR1I3 enhances the profibrotic effects of transforming growth factor beta and contributes to the development of experimental dermal fibrosis. Arthritis & rheumatology 2014;66(11):3140-50. Beyer C, Schett G, Distler O, Distler JH. Animal models of systemic sclerosis: Prospects and limitations. Arthritis Rheum. 2010 Oct;62(10):2831-44. Delle H, Rocha JR, Cavaglieri RC, Vieira JM, Jr., Malheiros DM, Noronha IL. Antifibrotic effect of tamoxifen in a model of progressive renal disease. J Am Soc Nephrol 2012;23(1):37-48.

17

Desallais L, Avouac J, Frechet M, Elhai M, Ratsimandresy R, Montes M, et al. Targeting IL-6 by both passive or active immunization strategies prevents bleomycin-induced skin fibrosis. Arthritis Res Ther 2014;16(4):R157. Elhai M, Avouac J, Walker UA, Matucci-Cerinic M, Riemekasten G, Airo P, et al. A gender gap in primary and secondary heart dysfunctions in systemic sclerosis: a EUSTAR prospective study. Ann Rheum Dis. 2016;75(1):163-9. Gabrielli A, Avvedimento EV, Krieg T. Scleroderma. N Engl J Med 2009;360(19):19892003. Garrett SM, Baker Frost D, Feghali C. The mighty fibroblast and its utility in scleroderma research. J Scleroderma relat disord. 2017;2(2):100-7. Goto N, Hiyoshi H, Ito I, Tsuchiya M, Nakajima Y, Yanagisawa J. Estrogen and antiestrogens alter breast cancer invasiveness by modulating the transforming growth factor-beta signaling pathway. Cancer Sci 2011;102(8):1501-8. Ito I, Hanyu A, Wayama M, Goto N, Katsuno Y, Kawasaki S, et al. Estrogen inhibits transforming growth factor beta signaling by promoting Smad2/3 degradation. J Biol Chem 2010;285(19):14747-55. Kashnikova LN, Grozdova MD, Panasiuk AF. [Significance of estradiol in the development of fibrosis in systemic scleroderma (bases of sexual predisposition to the disease)]. Biull Eksp Biol Med 1991;111(4):365-7. LeRoy EC, Black C, Fleischmajer R, Jablonska S, Krieg T, Medsger TA, Jr., et al. Scleroderma (systemic sclerosis): classification, subsets and pathogenesis. J Rheumatol 1988;15(2):202-5. Matsuda T, Yamamoto T, Muraguchi A, Saatcioglu F. Cross-talk between transforming growth factor-beta and estrogen receptor signaling through Smad3. J Biol Chem 2001;276(46):42908-14.

18

Murase T, Hattori T, Ohtake M, Nakashima C, Takatsu M, Murohara T, et al. Effects of estrogen on cardiovascular injury in ovariectomized female DahlS.Z-Lepr(fa)/Lepr(fa) rats as a new animal model of metabolic syndrome. Hypertension 2012;59(3):694-704. Offner H, Polanczyk M. A potential role for estrogen in experimental autoimmune encephalomyelitis and multiple sclerosis. Ann N Y Acad Sci 2006;1089:343-72. Rawson AJ, Abelson NM, Hollander JL. Studies on the Pathogenesis of Rheumatoid Joint Inflammation. Ii. Intracytoplasmic Particulate Complexes in Rheumatoid Synovial Fluids. Ann Intern Med 1965;62:281-4. Ruffy MB, Kunnavatana SS, Koch RJ. Effects of tamoxifen on normal human dermal fibroblasts. Arch Facial Plast Surg 2006;8(5):329-32. Ruzehaji N, Avouac J, Elhai M, Frechet M, Frantz C, Ruiz B, et al. Combined effect of genetic background and gender in a mouse model of bleomycin-induced skin fibrosis. Arthritis Res Ther 2015;17:145. Savelli BA, Parshley M, Morganroth ML. Successful treatment of sclerosing cervicitis and fibrosing mediastinitis with tamoxifen. Chest 1997;111(4):1137-40. Soldano S, Montagna P, Brizzolara R, Sulli A, Parodi A, Seriolo B, et al. Effects of estrogens on extracellular matrix synthesis in cultures of human normal and scleroderma skin fibroblasts. Ann N Y Acad Sci 2010;1193:25-9. Sthoeger ZM, Zinger H, Mozes E. Beneficial effects of the anti-oestrogen tamoxifen on systemic lupus erythematosus of (NZBxNZW)F1 female mice are associated with specific reduction of IgG3 autoantibodies. Ann Rheum Dis 2003;62(4):341-6. Thomas-Golbanov CK, Wilke WS, Fessler BJ, Hoffman GS. Open label trial of tamoxifen in scleroderma. Clin Exp Rheumatol 2003;21(1):99-102.

19

Tofovic SP, Zhang X, Jackson EK, Zhu H, Petrusevska G. 2-methoxyestradiol attenuates bleomycin-induced pulmonary hypertension and fibrosis in estrogen-deficient rats. Vascular pharmacology 2009;51(2-3):190-7. Tomcik M, Zerr P, Pitkowski J, Palumbo-Zerr K, Avouac J, Distler O, et al. Heat shock protein 90 (Hsp90) inhibition targets canonical TGF-beta signalling to prevent fibrosis. Ann Rheum Dis. 2014 Jun;73(6):1215-22. Varga J, Abraham D. Systemic sclerosis: a prototypic multisystem fibrotic disorder. J Clin Invest 2007;117(3):557-67. Varga J, Pasche B. Transforming growth factor beta as a therapeutic target in systemic sclerosis. Nat Rev Rheumatol 2009;5(4):200-6. Varga J, Trojanowska M, Kuwana M. Pathogenesis of systemic sclerosis: recent insights of molecular and cellular mechanisms and therapeutic opportunities. J Scleroderma relat disord. 2017;2(3):137-52. Varga JA, Trojanowska M. Fibrosis in systemic sclerosis. Rheum Dis Clin North Am 2008;34(1):115-43; vii. Voltz JW, Card JW, Carey MA, Degraff LM, Ferguson CD, Flake GP, et al. Male sex hormones exacerbate lung function impairment after bleomycin-induced pulmonary fibrosis. Am J Respir Cell Mol Biol 2008;39(1):45-52. Wei T, Chen W, Wen L, Zhang J, Zhang Q, Yang J, et al. G protein-coupled estrogen receptor deficiency accelerates liver tumorigenesis by enhancing inflammation and fibrosis. Cancer Lett 2016;382(2):195-202. Whitacre CC. Sex differences in autoimmune disease. Nature immunology 2001;2(9):777-80. Woessner JF, Jr., Boucek RJ. Connective tissue development in subcutaneously implanted polyvinyl sponge. I. Biochemical changes during development. Arch Biochem Biophys 1961;93:85-94.

20

Wynn TA, Ramalingam TR. Mechanisms of fibrosis: therapeutic translation for fibrotic disease. Nat Med 2012;18(7):1028-40. Zhang B, Wu ZY. Estrogen derivatives: novel therapeutic agents for liver cirrhosis and portal hypertension. Eur J Gastroenterol Hepatol. 2013 Mar;25(3):263-70. Zhu L, Song Y, Li M. 2-Methoxyestradiol inhibits bleomycin-induced systemic sclerosis through suppression of fibroblast activation. J Dermatol Sci 2015;77(1):63-70.

21

FIGURE LEGENDS: Figure 1: Estrogen inhibition by gene inactivation exacerbates bleomycin-induced dermal fibrosis. a, Representative images of hematoxylin and eosin-stained sections (scale bar 100µm). b, Dermal thickness quantification. c, Representative images of Masson Trichrome-stained sections (scale bar 100µm). d, Hydroxyproline content of the skin. e, Representative images of α-SMA immunohistochemistry (scale bar 20 µm). f, Myofibroblast count. Control mice were injected with NaCl, and the value for these mice was defined as 1, the results from the other groups were normalized to this value. A total of 40 mice were used and analyzed (n=8 in NaCl groups and n=12 in bleomycin groups). Values are the mean ± SEM. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001 determined by one-way ANOVA followed by Dunnett’s multiple comparison test.

Figure 2: Estrogen inhibition by tamoxifen exacerbates bleomycin-induced dermal fibrosis. a, Representative images of hematoxylin and eosin-stained sections (scale bar 100µm). b, Dermal thickness quantification.c, Representative images of Masson Trichromestained sections (scale bar 100µm). d, Hydroxyproline content of the skin. e, Representative images of α-SMA immunohistochemistry (scale bar 20 µm). f, Myofibroblast count. Control mice were injected with NaCl, and the value for these mice was defined as 1, the results from the other groups were normalized to this value. A total of 21 mice were used and analyzed (n=7 per group). Of note, 1 mouse could not be analyzed for hydroxyproline concentrations in the NaCl/oil group. Values are the mean ± SEM. * p<0.05, ** p<0.01 determined by one-way ANOVA followed by Dunnett’s multiple comparison test.

22

Figure 3: Estrogen inhibition by tamoxifen exacerbates skin fibrosis in tight skin mice. a, Representative images of hematoxylin and eosin-stained sections (scale bar 200µm) b, Hypodermal thickness quantification. c, Hydroxyproline content of the skin. pa/pa mice (control mice for Tsk-1 mice) were defined as 1.0 and all other results are normalized to this value. A total of 15 mice were used and analyzed (n=5 per group). Values are the mean ± SEM, * p<0.05, *** p<0.001, **** p<0.0001 determined by one-way ANOVA followed by Dunnett’s multiple comparison test.

Figure 4: Estrogens abrogate the profibrotic effects of TGF-β β in SSc dermal fibroblasts. a-f, SSc dermal fibroblasts: mRNA levels of Col1a1 (a), Col1a2 (b), and collagen release (c) upon TGF-β stimulation (10ng/ml) and treatment with17-β-estradiol (10 µg/ml) for 24 hours (n=5 for each). d, α-SMA mRNA levels upon TGF-β stimulation (10ng/ml) and treatment with17-β-estradiol (10 µg/ml) for 24 hours (n=5 for each). e, Representative images of immunofluorescence for α-SMA in SSc dermal fibroblasts upon TGF-β stimulation (10ng/ml) and treatment with17-β-estradiol (10 µg/ml) for 24 hours (Scale bar 20µm). Nuclei are stained with DAPI (blue). f, amount of fluorescence quantified with the image J software (n=5 for each). g, Cell extracts from cultured SSc fibroblasts, stimulated with TGF-β (10ng/ml) and treated with17-β-estradiol (10 µg/ml) for 24 hours, were immunoblotted for αSMA. All data are shown as the mean ± SEM of 3 independent experiments of each group. * p<0.05, ** p<0.01, *** p<0.001, determined by Student’s t test.

23

Figure 5:

Effect of estrogens on TGF-β-induced phosphorylation of downstream

signaling molecules: a-d, Cell extracts from cultured SSc fibroblasts, stimulated with TGF-β (10ng/ml) and treated with17-β-estradiol (10 µg/ml) for 24 hours, were immunoblotted for phosphor-Smad2/3 (pSmad2/3) and total Smad2/3 (a), phospho-TGFBR1 (pTGFBR1) and total TGFBRI cphospho-p38 (p-p38) and total p38 (p38) and (C) phosphorylated p44/42 (pp44/42) and total p44/42 (p44/42) (d) (n=5 for each). All data are shown as the mean ± SEM of 3 independent experiments of each group; *p<0.05 determined by Student’s t test.

Figure 6: Estrogens inhibit the activation of canonical Smad signalling and reduce TGFβ target gene expression in SSc dermal fibroblasts; these effects being efficiently reversed by tamoxifen. a-b: SSc dermal fibroblasts: a, Representative images of immunofluorescence for pSmad2/3 in SSc dermal fibroblasts upon TGF-β stimulation (10ng/ml) and treatment with 17-βestradiol (10 µg/ml) and/or tamoxifen (5 µg/ml) (Scale bar 10µm). Tamoxifen was added 4 hours before the addition of TGF-β and 17-β-estradiol for 24 hours.b, amount of fluorescence quantified with the image J software (n=5 for each). c and d, The mRNA levels of the TGF-β ⁄ Smad target genes plasminogen activator inhibitor-1 (PAI-1) and matrix metalloproteinase-9 (MMP-9) upon TGF-β stimulation (10ng/ml) and treatment with17-β-estradiol (10 µg/ml) and/or tamoxifen (5 µg/ml) (n=5 for each). Tamoxifen was added 4 hours before the addition for 24 hours of TGF-β and 17-β-estradiol. All data are shown as the mean ± SEM of 3 independent experiments of each group. * p<0.05, ** p<0.01, *** p<0.001 determined by Student’s t test.

24

Supporting informations Estrogens counteract the profibrotic effects of TGF-β and their inhibition exacerbates experimental dermal fibrosis Jérôme Avouac1,2, Sonia Pezet1, Virginie Gonzalez1, Léa Baudoin1, Anne Cauvet1, Barbara Ruiz1, Gonçalo Boleto1, Marie Laure Brandely3, Manon Elmerich1, Yannick Allanore1,2 1

Université Paris Descartes, Sorbonne Paris Cité, INSERM U1016 and CNRS UMR8104, Institut Cochin, Paris, France 2

Université Paris Descartes, Sorbonne Paris Cité, Service de Rhumatologie A, Hôpital Cochin, Paris, France.

3

GH Hôpitaux Universitaires Paris Centre, Service de Pharmacie Clinique, Hôpital Cochin, Paris, France.

1

Additional Methods Animals Six-week-old female ERKO-α mice on a C57BL/6 background were purchased from Jackson laboratory (Bar Harbor, ME). These mice are characterized by a complete lack of ERα proteins, leading to rudimentary reproductive structures and infertility. Five-week-old female tight skin-1 (Tsk-1) strains were also purchased from Jackson Laboratory and six-week old female wildtype C57BL/6 mice were purchased from Charles River (Chatillon-surChalaronne, France). The Tsk-1 phenotype is caused by a dominant mutation in the fibrillin-1 gene. Tsk-1 mice are characterized by accumulation of collagen fibers in the hypodermis resulting in progressive hypodermal thickening. In contrast to bleomycin-induced fibrosis, inflammatory infiltrates are absent and the aberrant activation of fibroblasts is not caused by the release of inflammatory mediators from leukocytes. Similar to SSc fibroblasts, fibroblasts from Tsk-1 are endogenously activated with increased release of collagen that persists for several passages in vitro.

Histological analysis Lesional skin areas were excised, fixed in 4% formalin and embedded in paraffin. Five µm thick sections were stained with hematoxylin and eosin. The dermal thickness was analyzed at 100-fold magnification by measuring the distance between the epidermal-dermal junction and the dermal-subcutaneous fat junction at four sites from lesional skin of each mouse (Avouac et al., 2012, Avouac et al., 2014, Desallais et al., 2014). The hypodermal thickness in Tsk-1 mice was determined by measuring the thickness of the subcutaneous connective tissue beneath the panniculus carnosus at 4 different sites of the upper back in each mouse (Avouac et al., 2011b; Desallais et al., 2014). Two independent examiners blinded to the treatment of the mice performed the evaluation. (JA and LB).

2

Hydroxyproline assay The collagen content in lesional skin samples was evaluated by the hydroxyproline assay (Avouac et al., 2013, Avouac et al., 2011b, Woessner and Boucek, 1961). After digestion of punch biopsy specimens (3 mm diameter) in 6 M HCl for 3 hours at 120°C, the pH of the samples was adjusted to 7. Afterwards, samples were mixed with 0.06 M chloramine T and incubated for 20 minutes at room temperature. Then, 3.15 M perchloric acid and 20 % pdimethylaminobenzaldehyde were added and samples were incubated for an additional 20 minutes at 60°C. The absorbance was determined at 557 nm with a spectrophotometer. For direct visualization of collagen fibers, trichrome staining was performed.

Immunohistochemistry for α-SMA, CD3, CD22, CD68 For immunohistochemistry, skin sections were deparaffinized, followed by incubation with 5% bovine serum albumin in phosphate-buffered saline for 1 hour to block non-specific binding and incubation with 3% H2O2 for 10 minutes to block endogenous peroxidase activity. Staining was visualized with aminoethylcarbazole, using a peroxidase substrate kit (Vector, Burlingame, CA). Myofibroblasts were identified by staining for α-SMA as previously described (Rawson et al., 1965). Cells positive for α-SMA in mouse skin sections were detected by incubation with monoclonal mouse anti-human antibodies against anti-α-SMA. Polyclonal rabbit anti-mouse antibodies labeled with horseradish peroxidase (HRP) were used as secondary antibodies for 1 hour at room temperature. Irrelevant isotype-matched antibodies served as controls. The number of myofibroblasts was determined at 200-fold magnification using a Nikon Eclipse 80i microscope. Myofibroblasts were defined as single, α-SMA-positive spindle-shaped cells. Two experienced examiners (JA and LB) performed myofibroblast quantification in six

3

randomly selected high-power fields for each section. Three sections were evaluated for each slide (Avouac et al., 2011b, Avouac et al., 2013). To quantify the numbers of infiltrating T cells, B cells and monocytes, skin sections were incubated with polyclonal rabbit anti-mouse antibodies for CD3, CD22 (Abcam, Paris, France), or with monoclonal mouse anti-mouse antibodies against CD68 (Novus Biologicals, Littleton, CO). T cells, B cells and monocytes were counted in a blinded manner, by two examiners (SP and JA), in six different sections of lesional skin of each mouse.

Fibroblast cultures Fibroblast cultures were obtained from skin biopsies of 8 SSc patients who signed a written consent form approved by the local institutional review boards. All SSc patients fulfilled the criteria for SSc subsets as suggested by LeRoy et al (LeRoy et al., 1988). The median age of SSc patients was 48 years (range: 31-68 years) and their median disease duration was 7 years (range: 1-10 years); 3 had limited cutaneous disease and 5 diffuse cutaneous SSc. None of them was treated with immunosuppressive or other potentially disease-modifying drugs. Human fibroblasts were prepared by outgrowth cultures from skin biopsy specimens and cultured in Dulbecco’s modified Eagle’s medium (DMEM)–Ham’s F-12 containing 10% heat-inactivated fetal calf serum (FCS), 25mM HEPES, 100units/ml penicillin, 100µg/ml streptomycin, 2mM L-glutamine, and 2.5µg/ml amphotericin-B (all from Invitrogen, Karlsruhe, Germany) as described (Avouac et al., 2008). In selected experiments, fibroblasts were stimulated with recombinant TGF-β for 24 hours (10ng/ml, PeproTech, Hamburg, Germany). This latter concentration represents the standard concentration used for the stimulation of dermal fibroblasts and is based on the serum levels in SSc patients (Avouac et al., 2014). Fibroblasts from passages 4–8 were used for the experiments. In addition, we used

4

tamoxifen, solubilized in methanol, and 17-β-estradiol, solubilized in PBS (both from Sigma Aldrich, Saint-Quentin Fallavier, France), to modulate estrogens in dermal fibroblasts.

Quantitative real-time polymerase chain reaction (PCR) Total RNA was isolated from dermal fibroblasts with the RNeasy mini kit according to the instructions of the manufacturer (Qiagen, Courtaboeuf, France). First-strand complementary DNA was synthetized (from 2 µg of total RNA) using random primers and Superscript II reverse transcriptase (200 units/µl; Invitrogen, Carlsbad, CA). TaqMan quantitative reverse transcriptase-PCR (RT-PCR) was performed, as previously described (Avouac et al., 2008). We used primer sequences (inventoried by Life Technologies, Saint Aubin, France) for human

α1(I)

procollagen

(col1a1)

(Hs00164004_m1),

α2(I)

procollagen

(col1a2)

(Hs00164099_m1), plasminogen activator inhibitor-1 (PAI-1) (Hs01126606_m1) and matrix metalloproteinase-9 (MMP-9) (Hs 00234579_m1). Hypoxanthine guanine phosphoribosyltransferase was used as a housekeeping control (Hs99999909_m1). All PCRs were performed in triplicates. Relative mRNA levels for each sample were quantified using the Ct approach (fluorescence threshold), normalized to HPRT RNA as the standard.

Western blot analysis Western blots were performed as previously described (Avouac et al., 2011a, Avouac et al., 2008). SSc dermal fibroblasts were homogenized and sonicated in RIPA buffer containing protease and phosphatase inhibitors and 30 µg of protein was used. To detect, immunoblots were incubated overnight at 4°C with monoclonal rabbit anti-human or mouse anti-human antibodies against α-SMA, Smad2/3, phosphorylated Smad-2/3 (pSmad2/3), p38, phosphop38, p44/42, phospho-p44/42 (Cell Signaling, Leiden, The Netherlands), TGFBR1 (Abcam, Paris, France) and phosphorylated TGFBR1 (pTGFBR1, ThermoFischer, Waltham, MA).

5

Horseradish peroxidase (HRP)–conjugated goat anti-rabbit or rabbit anti-mouse (Dako, Glostrup, Denmark) were used as secondary antibodies.

References Avouac J, Cagnard N, Distler JH, Schoindre Y, Ruiz B, Couraud PO, et al. Insights into the pathogenesis of systemic sclerosis based on the gene expression profile of progenitorderived endothelial cells. Arthritis Rheum 2011a;63(11):3552-62. Avouac J, Elhai M, Tomcik M, Ruiz B, Friese M, Piedavent M, et al. Critical role of the adhesion receptor DNAX accessory molecule-1 (DNAM-1) in the development of inflammation-driven dermal fibrosis in a mouse model of systemic sclerosis. Ann Rheum Dis 2013;72(6):1089-98. Avouac J, Furnrohr BG, Tomcik M, Palumbo K, Zerr P, Horn A, et al. Inactivation of the transcription factor STAT-4 prevents inflammation-driven fibrosis in animal models of systemic sclerosis. Arthritis Rheum 2011b;63(3):800-9. Avouac J, Palumbo K, Tomcik M, Zerr P, Dees C, Horn A, et al. Inhibition of activator protein 1 signaling abrogates transforming growth factor beta-mediated activation of fibroblasts and prevents experimental fibrosis. Arthritis Rheum 2012;64(5):1642-52. Avouac J, Palumbo-Zerr K, Ruzehaji N, Tomcik M, Zerr P, Dees C, et al. The nuclear receptor constitutive androstane receptor/NR1I3 enhances the profibrotic effects of transforming growth factor beta and contributes to the development of experimental dermal fibrosis. Arthritis & rheumatology 2014;66(11):3140-50. Avouac J, Wipff J, Goldman O, Ruiz B, Couraud PO, Chiocchia G, et al. Angiogenesis in systemic sclerosis: impaired expression of vascular endothelial growth factor receptor 1 in endothelial progenitor-derived cells under hypoxic conditions. Arthritis Rheum 2008;58(11):3550-61. Desallais L, Avouac J, Frechet M, Elhai M, Ratsimandresy R, Montes M, et al. Targeting IL-6 by both passive or active immunization strategies prevents bleomycin-induced skin fibrosis. Arthritis Res Ther 2014;16(4):R157. LeRoy EC, Black C, Fleischmajer R, Jablonska S, Krieg T, Medsger TA, Jr., et al. Scleroderma (systemic sclerosis): classification, subsets and pathogenesis. J Rheumatol 1988;15(2):202-5. Rawson AJ, Abelson NM, Hollander JL. Studies on the Pathogenesis of Rheumatoid Joint Inflammation. Ii. Intracytoplasmic Particulate Complexes in Rheumatoid Synovial Fluids. Ann Intern Med 1965;62:281-4. Woessner JF, Jr., Boucek RJ. Connective tissue development in subcutaneously implanted polyvinyl sponge. I. Biochemical changes during development. Arch Biochem Biophys 1961;93:85-94.

6

Supplementary Figure legends Figure S1: Estrogen inhibition increases T cell infiltration into lesional skin: a, Representative images of lesional skin sections stained for CD3 (scale bar 20 µm) and positive CD3 T-cell counts. b, Representative images of lesional skin sections stained for CD22 (scale bar 20 µm) and positive CD22 B-cell counts. c, Representative images of lesional skin sections stained for CD68 (scale bar 20 µm) and positive CD68 monocyte counts. Control mice were injected with NaCl, and the value for these mice was defined as 1, the results from the other groups were normalized to this value. A total of 40 mice were used and analyzed (n=8 in NaCl groups and n=12 in bleomycin groups). Values are the mean ± SEM. * p<0.05, ** p<0.01, **** p<0.0001 determined by one-way ANOVA followed by Dunnett’s multiple comparison test. Figure S2: Estrogen inhibition enhances the activation of canonical Smad signaling in experimental fibrosis: a and c, Representative images of skin section of immunostained skin tissue for pSmad 2/3 in the bleomycin-induced dermal fibrosis mouse model (scale bar 50 µm) b and d, Quantification of the number of nuclei stained for pSmad 2/3 (n=5 for each). Control mice were injected with NaCl, and the value for these mice was defined as 1, the results from the other groups were normalized to this value. A total of 35 mice were used and analyzed. Values are the mean±SEM. * p<0.05, ** p<0.01, *** p<0.001 determined by oneway ANOVA followed by Dunnett’s multiple comparison test. BLM: bleomycin, TAMOX: tamoxifen. Figure S3: ER-α and ER-β cooperate to attenuate the profibrotic effects of TGF-β β : a-c, SSc dermal fibroblasts: mRNA levels of Col1a1 (a), Col1a2 (b), and collagen release (c) upon TGF-β stimulation (10ng/ml), treatment with17-β-estradiol (10 µg/ml), ER-α inhibitor MPP (10-6M) or ER-β inhibitor PHTPP (10-6M) for 24 hours (n=5 for each). MPP or PTHTPP were added 2 hours before the addition for 24 hours of TGF-β and 17-β-estradiol. All data are shown as the mean ± SEM of 3 independent experiments of each group. * p<0.05, ** p<0.01 determined by one-way ANOVA followed by Dunnett’s multiple comparison test. Figure S4: Estrogens abrogate the profibrotic effects of TGF-β β in control dermal fibroblasts: a, collagen release upon TGF-β stimulation (10ng/ml) and treatment with17-βestradiol (10 µg/ml) for 24 hours (n=5 for each) in control dermal fibroblasts. b, α-SMA mRNA levels upon TGF-β stimulation and treatment with17-β-estradiol (10 µg/ml) for 24 hours (n=5 for each) in control dermal fibroblasts. c, Representative images of immunofluorescence for α-SMA in control dermal fibroblasts upon TGF-β stimulation and treatment with17-β-estradiol (10 µg/ml) for 24 hours (Scale bar 20µm). Nuclei are stained with DAPI (blue). d, amount of fluorescence quantified with the image J software (n=5 for each). e-f, protein expression of ER-α (E) and ER-β (F) assessed by western blot in unstimulated SSc and control dermal fibroblasts (n=5 for each). All data are shown as the mean ± SEM of 3 independent experiments of each group. * p<0.05, ** p<0.01, determined by Student’s t test.

7

Figure S5: Estrogens interfere with the canonical Smad-dependent pathway through a proteasome-dependent degradation of pSmad2/3: a-b, Estrogen inhibition of TGFβ response in the presence of the proteasome inhibitor MG132. Cell extracts from cultured SSc fibroblasts, stimulated with TGF-β (10ng/ml), treated with17-β-estradiol (10 µg/ml) and the proteasome inhibitor MG132 (10 or 20 µM) for 24 hours, were immunoblotted for phosphorSmad2/3 (pSmad2/3) and total Smad2/3 (n=5 for each). MG132 (10 or 20 µM) were added 4 hours before the addition for 24 hours of TGF-β and 17-β-estradiol. All data are shown as the mean ± SEM of 3 independent experiments of each group; *p<0.05 determined by one-way ANOVA followed by Dunnett’s multiple comparison test.

Figure S6: Tamoxifen does not activate SSc dermal fibroblasts and does not enhance the profibrotic effects of TGF-β β : a-f, SSc dermal fibroblasts: mRNA levels of Col1a1 (a), Col1a2 (b), and collagen release (c) upon TGF-β stimulation (10ng/ml) and treatment with tamoxifen (5 µg/ml) for 24 hours (n=5 for each). d, α-SMA mRNA levels upon TGF-β stimulation (10ng/ml) and treatment with tamoxifen (5 µg/ml) for 24 hours (n=5 for each). e, Representative images of immunofluorescence for α-SMA in control dermal fibroblasts upon TGF-β stimulation (10ng/ml) and treatment tamoxifen (5 µg/ml) for 24 hours (Scale bar 20µm). Nuclei are stained with DAPI (blue). f, amount of fluorescence quantified with the image J software (n=5 for each). All data are shown as the mean ± SEM of 3 independent experiments of each group. Statistical test: Student’s t test. Figure S7: Tamoxifen reverses the inhibitory effects of estrogens on TGF-β β -induced collagen synthesis: a-c, SSc dermal fibroblasts: mRNA levels of Col1a1 (a), Col1a2 (b), and collagen release (c) upon TGF-β stimulation (10ng/ml) and treatment with17-β-estradiol (10 µg/ml) and/or tamoxifen (5 µg/ml) (n=5 for each). Tamoxifen was added 4 hours before the addition for 24 hours of TGF-β and 17-β-estradiol. All data are shown as the mean ± SEM of 3 independent experiments of each group. * p<0.05, ** p<0.01; n=5 per group determined by Student’s t test. Figure S8: Proposed model for the regulation of dermal fibrosis by controlling transforming growth factor (TGF)-β β signaling using estrogens and tamoxifen: In absence of tamoxifen, estrogens inhibit the profibrotic effects of TGF-β in dermal fibroblasts by repressing the TGF-β-induced canonical cascade. A putative direct interaction between pSmad2/3 and ER-α leads to enhanced degradation of Smad proteins via an ubiquitinproteasome pathway. The estrogen-induced inhibition of TGF-β canonical signaling leads to decreased collagen synthesis, reduced α-smooth muscle actin (α-SMA) and other TGF-β target gene expression. Tamoxifen induces a competitive inhibition of estrogens that prevents the putative interaction between estrogens and pSmad2/3, leading to enhanced collagen synthesis, increased αsmooth muscle actin (α-SMA) expression and transcription of other TGF-β target genes. ERα, Estrogen receptor α, P, phosphorylation, SBE, Smad binding element, RI/RII, Receptor type I/II, ECM, extracellular matrix. Data still pending, not directly demonstrated in our study are represented by dotted arrows. Data emerging from our study are represented by plain arrows.

8