Systemic sclerosis (SSc) is an autoimmune disease that affects skin and multiple internal organs. TGF-β, a central trigger of cutaneous fibrosis, activates fibroblasts with the involvement of the stress-inducible chaperone heat shock protein 90 isoform α (Hsp90α). Available evidence supports overexpression and secretion of Hsp90α as a feature in profibrotic pathological conditions. The aim of this work is to investigate the expression and function of Hsp90α in experimental models of skin fibrosis such as human fibroblasts, C57BL/6 mice, and in human SSc. For this purpose, we generated a new experimental model based on doxorubicin administration with improved characteristics with respect to the bleomycin model. We visualized disease progression in vivo by fluorescence imaging. In this work, we obtained Hsp90α mRNA overexpression in human skin fibroblasts, in bleomycin- and doxorubicin-induced mouse fibrotic skin, and in lungs of bleomycin- and doxorubicin-treated mice. Hsp90α-deficient mice showed significantly decreased skin thickness compared with wild-type mice in both animal models. In SSc patients, serum Hsp90α levels were increased in patients with lung involvement and in patients with the diffuse form of SSc (dSSc) compared with patients with the limited form of SSc. The serum Hsp90α levels of patients dSSc were correlated with the Rodnan score and the forced vital capacity variable. These results provide new supportive evidence of the contribution of the Hsp90α isoform in the development of skin fibrosis. In SSc, these results indicated that higher serum levels were associated with dSSc and lung fibrosis.

Visual Abstract

Systemic sclerosis (SSc) is a prototypical orphan multiorgan disease characterized by autoimmunity, vasculopathy, and fibrosis of the skin and multiple internal organs. Among the different immune-mediated rheumatic diseases, SSc is one of the most disabling and potentially fatal disease (1), having in Europe a prevalence of ∼1 in 2000 of the general population (1, 2), and in the United States 16.4 per 100,000 person-years (3). Apart from clinical thickening of the skin and the development of digital ulcers, progressive deterioration of internal organs, especially the lungs, heart, kidneys, and gastrointestinal tract, as well as often underestimated polyarthritis, result in a high morbidity and mortality (4). The therapy of SSc is based on the use of immunosuppressive drugs (5), and there is a clear need for disease-modifying drugs targeting the fibrotic process, as well as markers for identification of the disease (6). Dysregulation of immune responses, which contributes to the overproduction of cytokines, chemokines, and adhesive molecules, leads to fibroblast activation and secretion of extracellular matrix (ECM) components, contributing to fibrogenesis (7, 8). TGF-β is a multifunctional cytokine that plays crucial pathophysiological roles in fibrosis (9). The role of TGF-β signaling in impaired ECM remodeling has been shown in various fibrotic models such as cultured dermal fibroblasts (10).

Increasing evidence points to chaperones as potential targets for reducing the deleterious effects of various human pathologies. In particular, in diseases characterized by uncontrolled cell proliferation, altered cell–cell interactions, tissue invasion, and/or genetic alterations, the administration of chaperone inhibitors resulted in improve outcomes (11). One of these molecular chaperones is heat shock protein 90 (Hsp90), whose inhibitors were already used as anticancer agents (12). Human cells contain two isoforms of cytoplasmic Hsp90, a constitutively expressed Hsp90β (Hsp90ab1) and a stress-inducible Hsp90α (Hsp90aa1) that maintain 86% amino acid sequence identity (13). Despite high conservation, the α and β isoforms show different functions and differential induction, with Hsp90α being involved in more adaptive roles (13) with increased expression and secretion into the extracellular space (14). Despite the classical role of Hsp90α as a de novo and misfolded protein folding helper, chaperones exhibit another cellular role by participating in functional complexes. Knowledge of the proteins involved in protein–protein interactions to form functional complexes is becoming increasingly important in the search for pharmacological targets with the aim of discovering new drugs (15) to reduce protein expression without altering key proteins for cell homeostasis. Hsp90 mainly functions as a chaperone that helps proteins to fold, preventing protein aggregation and reducing cellular stress (16), but also Hsp90 binds and acts in coordination with many factors to protect cell signaling cascades as in the case of TGF-β and its cascade-related proteins (17, 18). Examples are TGF-β degradation by Hsp90 inhibition that ameliorates pancreatic fibrosis (19) or involvement in TGF-β receptor stabilization in SSc (20).

Moreover, increased total Hsp90 levels have recently been described in patients with SSc (21), as well as in other fibrotic diseases also related to the TGF-β signaling cascade, such as renal fibrosis (22), myocardial fibrosis (18, 23), or liver fibrosis (24). However, studies focused on the specific inducible Hsp90α isoform are limited (11, 25), and none of them is related to scleroderma. The interest in studying this isoform stems from the hypothesis that Hsp90α expression is induced under pathological conditions and subsequently secreted into the extracellular milieu to participate in the profibrotic TGF-β signaling cascade (18, 23). Overexpression and secretion of Hsp90α may indicate that it could have a role as a biomarker. The main clinical significance of secreted Hsp90α stems from data indicating that its overexpression is associated with poor prognosis in different tumors and wound fibrosis (26, 27). SSc is associated with the development of potentially lethal manifestations that can reduce the quality of life. One of these manifestations is SSc-associated interstitial lung disease (ILD; SSc-ILD), which is a complication in most patients with SSc. In 25–30% of cases SSc-ILD progresses to life-threatening progressive ILD (28). The association of ILD to a specific marker could provide a step forward in the characterization of aggressive subtypes of the SSc.

The research objective of this work is to explore the role and study the expression of Hsp90α in mice models of cutaneous fibrosis, pulmonary fibrosis, and in human SSc.

The animal models generated so far recapitulate one or more aspects of SSc (29). Increasingly, they are becoming vital tools for identifying the molecular mediators and signaling pathways involved in pathogenesis, as well as for preclinical studies to test potential disease-modifying therapies. Unfortunately, no experimental model fully reproduces the pathophysiological spectrum of human SSc and no biomarker is able to discriminate disease severity. In this study, doxorubicin was used to generate skin fibrosis because of its dermal responsiveness (30) and its potential of monitoring. We hypothesized that circulating levels of Hsp90α have direct relationships with the severity of SSc, and that Hsp90α has cellular involvement in the development the disease. This experimental approach will advance the understanding of the role of Hsp90α in SSc and reveal whether it could be a potential biomarker for the disease.

Human skin fibroblasts were cultured by growing explants from healthy adult skin obtained during minor cosmetic surgery. Fibroblasts from six individuals were used between passages 4 and 6. Each batch was cultured on three different plates to generate replicates. Cells were cultured in 10% Dulbecco’s PBS/fecal bovine serum-DMEM in plastic flasks and activated with TGF-β (0.3 ng/ml, 24 h).

Immunofluorescence assays of primary fibroblasts were performed in paraformaldehyde-fixed cells by staining the cytoplasm with phalloidin (A12381, Thermo Fisher Scientific), the endoplasmic reticulum (ER) with ER tracker (E34251, Thermo Fisher Scientific), and Hsp90α Ab (PA3-013, Invitrogen). PureBlu Hoechst (33342, Bio-Rad) was used as nuclear staining dye. The immunofluorescence detection of representative images belonging to three independent assays was performed with the LSM 510 laser scanning microscope (Carl Zeiss, Jena, Germany) using a Plan Apo VC 60×/oil DIC (differential interference contrast) N2 objective. The fluorescent intensity quantification was performed with ImageJ software. The fluorescence intensity of four independent areas for each condition with at least two cells in each area was analyzed for detection of Hsp90α Ab.

Extracellular media were collected from the TGF-β–activated primary human fibroblasts described above and their corresponding TGF-β–free controls. Samples were concentrated (Centricon Plus tube filters, Corning Life Sciences) by centrifugation. To detect protein levels of Hsp90α, three independent assays were performed in triplicate with the hHsp90α FineTest ELISA kit using 100 μl of concentrated culture medium from cells treated with TGF-β (0.3 ng/ml), TGF-β (0.3 ng/ml) and Smad3 inhibitor, TGF-β (0.3 ng/ml) and SIS3 CAS 1009104-85-1 (10 µM) (SIS3), TGF-β (0.3 ng/ml) and TGF-βRI inhibitor SB431542 (10 µM) (SB), and TGF-β (0.3 ng/ml) and ERK inhibitor U0126 monoethanolate (10 µM) (U0126) for 24 h and the same volume of cell lysates.

The study followed the updated ARRIVE guidelines. Studies with live animals were approved by the Institutional Committee for the Care and Use of Laboratory Animals of the the University of Cantabria/Barcelona, according to the Guide for the Care and Use of Laboratory Animals (Institute for Laboratory Animal Research, 1985) and were performed in accordance with the European Directive for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes (European Communities Council Directive 86/606/EEC). Adult females of the C57BL/6 strain (12–16 wk old) and Hsp90α knockout (KO) (23) mice, housed in a 22°C room with a 12-h light/12-h dark cycle were provided with food, water ad libitum, and areas to hide from view as environmental enrichment. Animal body weights were recorded every 2 d during the course of the experiments. Signs of deterioration caused by surgery and/or reagent treatments were monitored. Animals were randomly distributed into two experimental groups (control and profibrotic) and control mice received saline administration.

Female wild-type (WT) and Hsp90α KO group mice were age matched and divided into two groups, control littermates (n = 4 mice) and the profibrotic bleomycin and doxorubicin group.

In the bleomycin model (n = 6), under anesthesia, each mouse was administered 50 μl of bleomycin (30 ng/ml) s.c. every day during 21 d in the nape of the neck.

In the doxorubicin model (n = 10), under anesthesia, each mouse received an intradermal injection of doxorubicin (40 μl of 1 μg/g body weight) every other day for 1 wk in four areas of the ventral region.

Mice were monitored daily and showed no sign of distress. Mice were euthanized under anesthesia and skin samples were extracted and either directly frozen or fixed in a 4% paraformaldehyde, sucrose gradient and frozen at −80°C. Assays were performed double-blinded with respect to the mouse group.

Female C57BL/6 mice aged 3–4 mo and n = 6 mice per group were used. Mice were anesthetized by i.p. injection of ketamine (50 µg/g body weight), atropine sulfate (0.2 µg/g), and diazepan (4 µg/g). Mice were euthanized 7 d later and lungs were removed for histological and biochemical studies.

In the doxorubucin model, for intratracheal instillation of doxorubicin, under anesthesia, the trachea was located through a skin incision on the anterior face of the neck. Then, 75 μl of doxorubicin hydrochloride (1 mg/ml) was injected intratracheally with a 30G needle and then the skin was sutured with 3/0 silk.

In the bleomycin model, pulmonary fibrosis was induced by intranasal instillation with a single dose of bleomycin (2.5 U/kg) dissolved in sterile saline under anesthesia.

In vivo fluorescence images were obtained with the Xenogen in vivo imaging system (IVIS, PerkinElmer). Mice were anesthetized with 5 µl/g mouse of medetomidine (0.2 mg/ml), ketamine (16 mg/ml), and buprenorphine analgesic (0.018 mg/ml) and placed ventrally (doxorubicin group) or dorsally (bleomycin group). Excitation and emission settings for collagen (COL) detection were obtained with pure solid COL I at 440 and 520 nm, respectively. The excitation and emission wavelength of doxorubicin was set at 520/570 nm (wavelengths of highest excitation and emission of doxorubicin). The epifluorescence of fibrosis and doxorubicin was obtained after unmixing the images relativized to all controls of the different assays and expressed in units of relativized maximum radiant efficiency [photons/s]/[μW/cm2]. The exposure time was adjusted with “auto-expose,” and the acquired images were analyzed with Living Image 4.7.2 software. Images were obtained with IVIS for each group of individuals before and after the administration of bleomycin, doxorubicin, or saline in both WT and Hsp90α mice.

Longitudinal skin sections (4 μm) were obtained from the skin of WT and Hsp90α KO mice (bleomycin model, one dorsal area; doxorubicin model, four ventral areas) and the corresponding paired areas of control mice. Longitudinal skin sections were prefixed in 4% paraformaldehyde, followed by gradient alcohol dehydration and paraffin embedding. COL fibers were stained with Masson’s trichrome stain, and the skin thickness of each sample was assessed with digital image analysis using ImageJ 1.52i software. Samples for the H&E assay were stained in parallel in consecutive longitudinal skin sections (4 μm). Complete scanning of all sections was performed in bright field with the Zeiss Axioscan Z1 scanner using the ×10/0.45 Plan Apo objective. Images were exported to JPG with a pixel size of 0.884 µm/pixel and analyzed with ImageJ 1.52i software. The pieces of mouse skin analyzed for each group were n = 6 in the case of the bleomycin model and n = 10 in the case of the doxorubicin model, with three technical replicates per sample in all cases. Infiltrating immunogenic cells were detected and quantified on the previously described scanned sections of the H&E-stained skin samples. Three arbitrarily chosen fields in three consecutive skin sections of the tissue samples from each mouse were counted at ×400 magnification. Serum from the mice was obtained with standard heparinized capillary tubes using the retro-orbital bleeding procedure and analyzed with a mouse TGF-β1 ELISA kit (Quimigen) following the manufacturer’s instructions.

Expression of molecules related to cutaneous fibrosis

Human skin fibroblasts and skin from WT and Hsp90α KO mice were collected, pelleted, and RNA extracted. Total RNA was obtained with TRItidy G reagent (AppliChem). cDNA was prepared from 0.5 μg of total RNA by random primers using a first-strand cDNA synthesis kit (Promega). Data were corrected with the ΔΔCt method as target gene expression relative to 14S. Data were relativized to the control group as mean ± SD in triplicate for each replicate and condition.

The sequence of primers for the SYBR Green quantitative PCR (qPCR) of COL I, COL III, fibronectin, Hsp90aa1, Hsp90ab1, and gremlin-1 were as follows: COL Ia1, forward, 5′-TGGGGCAAGACAGTC ATCGAATA-3′, reverse, 5′-GGGTGGAGGGAGTTTACACG-3′; COL IIIa1, forward, 5′-ACCCCATGATGTGTTTTGTGGCA-3′, reverse, 5′-CAGGTCCTCGGAAGCCACTA-3′; fibronectin, forward, 5′-CACCCACATGGCAGCTCACA-3′, reverse, 5′-ATGGGAACCCTGAAGCCAGC-3′; Hsp90aa1, forward, 5′-ACTGGTGACATCCCCGTGCT-3′, reverse, 5′-CCTGTTAGCATGGGTCTGGG-3′; Hsp90ab1, forward, 5′-GGGAGGTCACCTTCAAGTCG-3′, reverse, 5′-CACTCTGACAGCCACGCGTA-3′; 14S, forward, 5′-AGT GACTGGTGGGATGAAGG-3′, reverse, 5′-CTTGGTCCTGTTTCCTCCTG-3′; gremlin-1, forward, 5′-GAGAGGAGGTGCTTGAGTCCAG-3′, reverse, 5′-CACTGGCCATAACAGAAGCGG-3′.

Expression of molecules related to pulmonary fibrosis

WT mice were divided into two control and profibrotic groups with six individuals per group. Total RNA from lung tissue was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA), and 0.5 μg was reverse transcribed into cDNA, according to the manufacturer’s instructions. PCR amplification was performed on a Peltier thermal cycler (Bio-Rad, Hercules, CA) using iQ SYBR Green supermix (Bio-Rad, Hercules, CA) and the primers corresponding to Hsp90aa1 and COL I (described in the preceding section). The mRNA expression levels were normalized with respect to GAPDH. The primers used for GAPDH were as follows: forward, 5-′CATCTGAGGGCCCACTGAAG-3′, reverse, 5′-TCGAAGGTGGAAGAGTGGGA-3′.

Data source

This is a retrospective cohort study based on samples and data from a prospective, observational cohort of patients with SSc. Patients were recruited from three public Hospitals with 215 participants in three Spanish cities (12 de Octubre, La Laguna, and Valdecilla hospitals). This multicenter study presents 57 controls who were matched for age and sex. The study of the patients complied with the Declaration of Helsinki. The Hospital 12 de Octubre centralized the approval of the study. The Clinical Research Ethics Committee approved the study with identification code 09/101. All participants in the registry gave written informed consent prior to participation.

Patient eligibility

Information was collected at the patient’s first visit. Eligible patients were adults (aged ≥18 y) with SSc and classified into two groups (limited SSc [lSSc] and diffuse SSC [dSSc]) on the basis of the criteria by LeRoy et al. (31). Detailed disease parameters and patient-reported outcomes are available on request. Available serum samples from the 215 patients under follow-up were analyzed. Patient characteristics are described in Table I. Detection of human Hsp90α ELISA protein was performed with the FineTest ELISA kit to detect Hsp90α levels using 100 μl of serum from controls and SSc patients.

Table I.

Baseline characteristics of multicenter study at date of blood collection

No. of SSc patients 215 
Sex (F/M) 191/24 
Age 62 ± 15 
Type of SSc: diffuse/limited 65/150 
ACA positive/negative 93/122 
ANA positive/negative 153/62 
ASCL70 positive/negative 49/166 
Pulmonary fibrosis 66/149 
Number of controls 57 
Sex of controls (F/M) 50/7 
Age of controls 64 ± 10 
No. of SSc patients 215 
Sex (F/M) 191/24 
Age 62 ± 15 
Type of SSc: diffuse/limited 65/150 
ACA positive/negative 93/122 
ANA positive/negative 153/62 
ASCL70 positive/negative 49/166 
Pulmonary fibrosis 66/149 
Number of controls 57 
Sex of controls (F/M) 50/7 
Age of controls 64 ± 10 

The disease was determined according to the criteria proposed by LeRoy et al. (32).

ACA, anticentromere Abs; ANA, antinuclear Abs; ASCL70, anti-topoisomerase Scl-70 Abs; F, female; M, male.

Basic descriptive statistics were calculated for all variables. The normality of the datasets was assessed using the Kolmogorov–Smirnov test. Continuous variables were compared using the two-tailed Student t test or the Mann–Whitney U test. Bivariate relationships between the variables under study were assessed using Spearman’s correlation coefficient. Linear regression analysis was used to predict patients’ Hsp90α levels and the predictors modified Rodnan skin score (mRSS) and forced vital capacity (FVC). Data are presented as the median, and GraphPad Prism 6.0 was used as the statistical package and graphics program for studies in cells and animal models. SPSS statistical software was used for human studies with GraphPad Prism 6.0 as the graphics program.

One of the processes that promotes Hsp90α overexpression and secretion is wound formation (32). Another process that promotes Hsp90α overexpression and is related to TGF-β is the activation of fibroblasts to produce COL (18, 33). Overexpression of COL gene in damaged regions is an event associated with cutaneous fibrosis (34). Knowing that variation in TGF-β concentrations could promote various cellular effects such as cell apoptosis, cell proliferation, or alterations of cell homeostasis, to analyze the expression and secretion of Hsp90α following TGF-β–mediated fibroblast activation, we first checked the presence of intracellular Hsp90α protein at different concentrations of TGF-β at 24 h, obtaining clear significant reductions only in the case of 0.3 ng/ml TGF-β (1.31 ± 0.33 ng/ml, **p < 0.005) and slight differences with 1.0 ng/ml TGF-β (1.31 ± 0.33 ng/ml *p < 0.05) compared with controls (control, 1.88 ± 0.37 ng/ml; vehicle, 1.68 ± 0.18 ng/ml). In the rest of the concentrations tested we did not obtained significant differences (TGF-β, 0.1 ng/ml: 1.44 ± 0.26 ng/ml; TGF-β, 0.5 ng/ml: 1.31 ± 0.33 ng/ml; and TGF-β, 1.5 ng/ml: 1.31 ± 0.33 ng/ml). Then, we checked cell ultrastructural changes of primary fibroblasts under TGF-β stimulation at the concentration of 0.3 ng/ml for 24 h. TGF-β–activated human skin fibroblasts showed an increase in ER size in response to TGF-β, pointing to increased protein synthesis and proper cell activation (Fig. 1A). When fibroblasts are activated by TGF-β, increased expression of the COL I gene was accompanied by increased expression of the Hsp90α gene (Fig. 1B, 1C). As the possibility of Hsp90α acting as a biomarker of fibrotic skin disease depends on its secretion into the medium after overexpression in fibroblasts, we next checked the secretion of Hsp90α protein to the extracellular space of cultured primary human skin fibroblasts that were TGF-β activated. We also found that TGF-β induction of Hsp90α secretion is mediated by Smad3 and ERK pathways. In (Fig. 1D, we show a significant increase of extracellular Hsp90α protein in TGF-β–stimulated fibroblasts (Fig. 1D) parallel to a reduction in its intracellular presence demonstrated by ELISA and immunofluorescence methodologies (Fig. 1E, Supplemental Fig. 1A, 1B). Reductions in the extracellular presence of Hsp90α protein are also shown in (Fig. 1D following inhibition of canonical and noncanonical TGF-β signaling pathways by inhibition of TGF-βRI (SB431542) or Smad3 (SIS3 CAS 1009104-85-1) or ERK (U0126). These reductions were paralleled by an increased intracellular presence of Hsp90α relative to TGF-β–stimulated cells (Fig. 1E).

FIGURE 1.

Secretion of Hsp90α from TGF-β–activated human skin fibroblasts. (A) Confocal microscopy images highlighting the increased ER signal after addition of TGF-β compared with control cells, showing a detail of two TGF-β–activated cells. (B and C) Gene expression assays (qPCR) showing overexpression of COL I and Hsp90aa1 mRNA. (D and E) ELISA assays showing extracellular (D) and intracellular (E) Hsp90α protein levels in control, TGF-β–activated cells, TGF-β–activated and Smad3 inhibitor (SIS3) treated, TGF-β–activated and TGF-βRI inhibitor (SB) or ERK inhibitor (U0126) treated. Cells were treated with 10 µM SIS3, SB431542, or UO126 or with the same amount of DMSO (0.1%) in the final dilution, vehicle (Veh) or without DMSO (Contr). Data are expressed as mean ± SD in arbitrary units (A.U.) for gene expression and ng/ml for protein expression (*p < 0.05, **p < 0.005, ***p < 0.0005, by Mann–Whitney test). The images are representative of three to four independent biological replicates that were performed with human skin fibroblasts and three technical replicates for each sample.

FIGURE 1.

Secretion of Hsp90α from TGF-β–activated human skin fibroblasts. (A) Confocal microscopy images highlighting the increased ER signal after addition of TGF-β compared with control cells, showing a detail of two TGF-β–activated cells. (B and C) Gene expression assays (qPCR) showing overexpression of COL I and Hsp90aa1 mRNA. (D and E) ELISA assays showing extracellular (D) and intracellular (E) Hsp90α protein levels in control, TGF-β–activated cells, TGF-β–activated and Smad3 inhibitor (SIS3) treated, TGF-β–activated and TGF-βRI inhibitor (SB) or ERK inhibitor (U0126) treated. Cells were treated with 10 µM SIS3, SB431542, or UO126 or with the same amount of DMSO (0.1%) in the final dilution, vehicle (Veh) or without DMSO (Contr). Data are expressed as mean ± SD in arbitrary units (A.U.) for gene expression and ng/ml for protein expression (*p < 0.05, **p < 0.005, ***p < 0.0005, by Mann–Whitney test). The images are representative of three to four independent biological replicates that were performed with human skin fibroblasts and three technical replicates for each sample.

Close modal

In other fibrosis studies, elevated levels of Hsp90 (presumably both isoforms α and β) were detected in different organs, including the skin (35), indicating its very likely involvement in the fibrotic process. Next, we sought to detect Hsp90α in the skin of fibrotic mice. We used two mouse models based on s.c. administered bleomycin (36), as well as intradermal doxorubicin administration. We generated a new doxorubicin model based on previous studies showing fibrotic skin reactions after its administration (30). In this study, we decided to administer the reagent doxorubicin intradermally to obtain localized, precise, and defined skin damage. One difference with the standard bleomycin model, which is administered s.c., is the formation of a less localized and less defined damaged area. We performed a titration of doxorubicin concentration (0, 0.12, 0.25, 0.60, 1.00, and 3.00 μg/g mouse body weight) to reach the optimal administration pattern to generate cutaneous fibrosis. In Supplemental Fig. 1C and 1D, the fibrotic spot appears at 1 μg/g mouse body weight. Although with 3.00 μg/g mouse body weight the damage was higher, with 1.00 μg/g mouse body weight the intensity of the fibrotic spot was as significant as the damage generated with 3.00 μg/g mouse body weight (Supplemental Fig. 1D). Therefore, the concentration chosen for the assays was kept at 1.00 μg/g mouse body weight, and 40 μl of doxorubicin was administered at that concentration. The emission wavelength of maximum radiant efficiency at which emission was maximal for detecting in vivo COL signal on the damaged skin spot was set at 450 nm (Supplemental Fig. 1E). In parallel, the reagent absorption tracking setting for the optimal emission wavelength was obtained measuring the autofluorescence of doxorubicin at 540 nm (Supplemental Fig. 1F). We further adjusted the optimal excitation and emission wavelengths for visualization of skin damage progression by in vivo fluorescence (IVIS) (Supplemental Fig. 1E, 1F). By measuring intrinsic COL formation (35) at 450 nm, increases in fibrosis were detected in vivo in the ventral regions where doxorubicin was administered. Supplemental Fig. 1G shows the progression of fibrosis formation in the damaged areas in the new doxorubicin model. Supplemental Fig. 2 shows the fluorescent signal of bleomycin and doxorubicin reagents before and after each administration. For nonfluorescent bleomycin, the maximum radiant efficiency was not elevated at any time point, as expected (Supplemental Fig. 2A); for doxorubicin, it showed a significant increment in the fluorescent signal after each injection (Supplemental Fig. 2B). In contrast to the bleomycin model, whose absorption could not be followed because of its lack of fluorescence (Supplemental Fig. 2A), the autofluorescence of doxorubicin was no longer observed at the site of administration and was therefore absorbed during the 2 d following each injection (Supplemental Fig. 1G). Thus, we set the administration schedule at 1.00 μg/g mouse body weight at 40 μl every other day and maintained this administration schedule for 6 d (Supplemental Fig. 1G). This observation indicated that administration can be performed every other day, rather than every day as for the bleomycin model, which cannot be fluorescently detected. Once the emission wavelengths of maximum radiant efficiency for COL and doxorubicin detection were established, fibrosis progression and reagent absorption were monitored in both the bleomycin (Fig. 2A) and doxorubicin (Fig. 2B) models. After 21 d in the bleomycin model or 7 d in the case of the doxorubicin model, significant COL detection appeared in the regions where the profibrotic reagents were administered (***p < 0.0005) (Fig. 2A, 2B). We observed the in vivo evolution of the disease (0, 7, and 21 d for the bleomycin model, and 0, 2, 4, 6, and 7 d for the doxorubicin model). COL accumulation in the doxorubicin model was more rapid and highly localized at the point of doxorubicin administration (images in (Fig. 2A and 2B, respectively). COL staining assays confirmed the increased dermal thickness of damaged tissue compared with healthy skin in both models (Fig. 2E). End point skin thickness quantification showed a significant increase (***p < 0.0005) for both models (bars in WT Control and WT Fibrotic of (Fig. 3D, 3F). We also corroborated the significant correlation (**p < 0.005) between the in vivo fluorescence signal detected for the COL autofluorescence and the measured skin thickness (Fig. 2C).

FIGURE 2.

Detection of skin fibrosis in bleomycin and doxorubicin mouse models. (A and B) Figures showing the relativized fluorescent signal in vivo for tracking fibrotic regions at 450 nm in the bleomycin model of the dorsal region shown by the black arrow in (A) (n = 6 independent experiments) and the doxorubicin model of ventral regions [black arrow in (B), n = 10 independent experiments]. (C) Significant correlation between skin thickness (independent assays n = 10 doxorubicin model and n = 6 bleomycin model) and in vivo measurements of relativized maximal radiation efficiency of fibrotic areas of mouse skin (independent assays n = 10 doxorubicin model and n = 6 bleomycin model). (D) Quantification of infiltrating immunogenic cells/area (n = 4 areas per condition). (E) Representative images of injured skin of WT animals treated with doxorubicin, bleomycin, and their corresponding controls. Last four images of (E) show representative images of infiltrating immunogenic cells (black arrows). (FK) Bar graphs showing the gene expression of fibronectin (FN) (F and I), Hsp90α (G and J), and Hsp90β (H and K) from fibrotic skin samples from bleomycin and doxorubicin models compared with their corresponding controls. In qPCR assays the number of biological replicates per group are bleomycin group (n = 4), doxorubicin group (n = 16), and control group (n = 4), with three technical replicates per sample. Data are expressed as mean ± SD in arbitrary units (*p < 0.05, **p < 0.005, ***p < 0.0005, by Mann–Whitney test). Bleo, bleomycin; Doxo, doxorubicin.

FIGURE 2.

Detection of skin fibrosis in bleomycin and doxorubicin mouse models. (A and B) Figures showing the relativized fluorescent signal in vivo for tracking fibrotic regions at 450 nm in the bleomycin model of the dorsal region shown by the black arrow in (A) (n = 6 independent experiments) and the doxorubicin model of ventral regions [black arrow in (B), n = 10 independent experiments]. (C) Significant correlation between skin thickness (independent assays n = 10 doxorubicin model and n = 6 bleomycin model) and in vivo measurements of relativized maximal radiation efficiency of fibrotic areas of mouse skin (independent assays n = 10 doxorubicin model and n = 6 bleomycin model). (D) Quantification of infiltrating immunogenic cells/area (n = 4 areas per condition). (E) Representative images of injured skin of WT animals treated with doxorubicin, bleomycin, and their corresponding controls. Last four images of (E) show representative images of infiltrating immunogenic cells (black arrows). (FK) Bar graphs showing the gene expression of fibronectin (FN) (F and I), Hsp90α (G and J), and Hsp90β (H and K) from fibrotic skin samples from bleomycin and doxorubicin models compared with their corresponding controls. In qPCR assays the number of biological replicates per group are bleomycin group (n = 4), doxorubicin group (n = 16), and control group (n = 4), with three technical replicates per sample. Data are expressed as mean ± SD in arbitrary units (*p < 0.05, **p < 0.005, ***p < 0.0005, by Mann–Whitney test). Bleo, bleomycin; Doxo, doxorubicin.

Close modal
FIGURE 3.

Hsp90aa1 gene deficiency reduces fibrosis in bleomycin and doxorubicin mice models. (A and B) In vivo detection of COL I autofluorescence relativized signal at 450 nm in Hsp90α KO mice with bleomycin (A) and doxorubicin (B) treatments. (C) Representative images of skin thickness differences (black vertical bar) in Masson’s trichrome–stained skins (top images) and H&E-stained (bottom images) skins from control (n = 4) and treated (doxorubicin, n = 10; bleomycin, n = 6) Hsp90α KO mice. Scale bars, 250 μm. (D and F) Comparison of the quantification of skin thickness measurements of bleomycin-treated (D) and doxorubicin-treated (F) WT and KO mice, control groups (n = 4) and treated (doxorubicin, n = 10; bleomycin, n = 6). (E and GI) Bar graphs comparing control gene expression of COL III (E and G) and gremlin (H and I) of bleomycin- and doxorubicin-treated WT and KO mice. n = 6 biological replicates of bleomycin, n = 16 of doxorubicin, and n = 4 of controls were assessed with three technical replicates per biological sample for the qPCRs. Data are expressed as mean ± SD (*p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.0001, by Mann–Whitney test). Bleo, bleomycin; Doxo, doxorubicin.

FIGURE 3.

Hsp90aa1 gene deficiency reduces fibrosis in bleomycin and doxorubicin mice models. (A and B) In vivo detection of COL I autofluorescence relativized signal at 450 nm in Hsp90α KO mice with bleomycin (A) and doxorubicin (B) treatments. (C) Representative images of skin thickness differences (black vertical bar) in Masson’s trichrome–stained skins (top images) and H&E-stained (bottom images) skins from control (n = 4) and treated (doxorubicin, n = 10; bleomycin, n = 6) Hsp90α KO mice. Scale bars, 250 μm. (D and F) Comparison of the quantification of skin thickness measurements of bleomycin-treated (D) and doxorubicin-treated (F) WT and KO mice, control groups (n = 4) and treated (doxorubicin, n = 10; bleomycin, n = 6). (E and GI) Bar graphs comparing control gene expression of COL III (E and G) and gremlin (H and I) of bleomycin- and doxorubicin-treated WT and KO mice. n = 6 biological replicates of bleomycin, n = 16 of doxorubicin, and n = 4 of controls were assessed with three technical replicates per biological sample for the qPCRs. Data are expressed as mean ± SD (*p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.0001, by Mann–Whitney test). Bleo, bleomycin; Doxo, doxorubicin.

Close modal

Once skin fibrosis and its relationship with COL signal was verified, we detected infiltrating immunogenic cells in the injured regions of the skin of the models under study (black arrows in (Fig. 2E). The assessment of the number of inflammatory cells showed the absence of significant differences between the doxorubicin model and the bleomycin model in the number of infiltrating leukocytes in lesional murine skin. However, the animals treated with bleomycin or doxorubicin showed significant differences with their corresponding untreated controls (Fig. 2D). We also observed that circulating levels of TGF-β were similar in both models (WT bleomycin at 71 ± 11 pg/ml; WT doxorubicin at 66 ± 11 pg/ml; Hsp90α KO bleomycin at 65 ± 15 pg/ml; Hsp90α KO doxorubicin at 58 ± 9 pg/ml) and were significantly augmented compared with controls (WT 42 ± 7 pg/ml and Hsp90α KO 36 ± 8 pg/ml, *p < 0.05). We next measured whether profibrotic proteins such as fibronectin, Hsp90aa1, and Hsp90ab1 were expressed in the fibrotic areas. Overexpression of fibronectin, Hsp90aa1, and Hsp90ab1 in these models showed significant values (**p < 0.005**) (Fig. 2F–K). Overexpression of the Hsp90α gene in fibrotic areas (**p < 0.005**) (Fig. 2G, 2H) could indicate its involvement in the fibrotic process, as we have previously described in other tissues (18). The overexpression of COL III and gremlin (profibrotic markers) was also analyzed and is compared with KO mice in the following section.

Hsp90α presents a key role in cell motility and cell survival (23, 32, 37) as well as in pathological conditions (35). The absence of Hsp90α is crucial in the progression of fibrosis in organs such as the heart (18). We show in (Fig. 1C that human skin fibroblasts experienced a clear increase (>10-fold) in Hsp90aa1 gene expression after stimulation by TGF-β. It is clear that fibrosis in WT mice is observed after treatment with bleomycin and doxorubicin. In addition, we analyzed the differences between WT and Hsp90α KO mice in the development of fibrosis. To observe the progression of fibrosis in the absence of Hsp90α (Hsp90α KO mouse), we performed in vivo fluorescence imaging for COL detection after administration of bleomycin (Fig. 3A) and doxorubicin (Fig. 3B). In Supplemental Fig. 3, we present fibrosis progression and doxorubicin uptake over time in fluorescence images before and after reagent administration at days 0, 2, 4, and 6. We observed that in Hsp90α KO mice, the fibrosis progression was delayed compared with that in WT mice. Fibrotic areas (*p < 0.05) appeared at day 7 in WT mice (Fig. 2A) and at day 21 (**p < 0.005) in the Hsp90α KO bleomycin model (Fig. 3A). In the doxorubicin Hsp90α KO model, in vivo monitoring revealed significant differences for COL accumulation compared with controls after 6 d (Fig. 3B). (Fig. 2B shows that WT mice took 4 d to show significant values (*p < 0.05), and thus the delayed effect (6 d) was observed in Hsp90α KO mice in both models.

Visualization of Hsp90α KO skin thickness (Fig. 3C) and its quantification and comparison with WT mice in both models (Fig. 3D, 3F) indicated thickening after bleomycin and doxorubicin treatments compared with controls with a significance of ***p < 0.005 for both models (bleomycin and doxorubicin) and in both WT and KO mice (Fig. 3D, 3F), as well as a clear reduction of the skin engrossment in Hsp90α KO mice compared with the WT mice in both models, that is, *p < 0.05 for the bleomycin model and ***p < 0.0005 for the doxorubicin model (Fig. 3D, 3F). The lower thickening and longer time to reach significant fibrosis in the Hsp90α KO mice pointed to an active role of Hsp90α in the fibrotic process in vivo. The overexpression of two more profibrotic markers such as gremlin-1 and COL III were observed in the skin of WT mice treated with doxorubicin in both WT and KO mice (Fig. 3G, 3I). The bleomycin model did not show those increments (Fig. 3E, 3H). Interestingly, COL III and gremlin-1 do not appear to be good profibrotic markers for the bleomycin model. Moreover, significant reductions in the gene expression of COL III and gremlin of KO mice were found compared with WT mice in the doxorubicin model (Fig. 3G, 3I); and in bleomycin model, control Hsp90α KO mice exhibited lower COL III expression than did WT mice (Fig. 3E).

The above results could indicate molecular differences in fibrosis formation according to the model used or that the much more localized damaged area of the doxorubicin model allows better identification of the markers. In any case, we will devote future studies to clarify this point.

To confirm whether Hsp90α is upregulated in patients with SSc and its increase is associated with the form of the disease that develops more fibrosis, we determined Hsp90α protein levels in serum of a multicenter cohort of patients with SSc (see Materials and Methods and Table I for clinical characteristics of patients and controls). A recent study has shown that plasma levels of total Hsp90 are significantly elevated in both subsets of SSc patients (lSSc and dSSc) compared with healthy controls (38). In this study, we observed higher Hsp90α levels in dSSc patients compared with lSSc patients (****p < 0.0001) and controls (*p < 0.05), as well as higher Hsp90α levels in patients with lung affectation (ILD) compared with patients without ILD (***p < 0.0005) (Fig. 4A). In addition, exponentiation of the coefficient B (see Supplemental Table I to check univariate statistics analysis) showed a clear categorization of Hsp90α into groups with significant differences (Fig. 4B, Supplemental Table I). The proportion of patients with dSSc/lSSc increased accordingly in the higher quartiles of Hsp90α (Table II). These increases inferred an association of higher Hsp90α values with the more severe subtype of the disease (dSSc). Thus, the Hsp90α 5C category included the highest number of dSSc patients (>65%, Table II).

FIGURE 4.

Serum Hsp90α levels in patients with SSc and their association with disease parameters. Expression of profibrotic proteins in the lung of fibrotic mice was determined. (A) Bar graph showing Hsp90α protein levels of lSSc, dSSc, and SSc patients with and without ILD. (B) Graph bar showing exponentiation of Hsp90α coefficient B values as a categorical variable. (C and D) Pearson’s analysis for correlation of Hsp90α with the mRSS (C) and with ILD (D) in patients with dSSc. (EJ) Overexpression of mRNA levels in mouse lung of the following genes: Hsp90aa1 in the bleomycin model (E), Hsp90aa1 in the doxorubicin model (H), COL I in the bleomycin model (F), COL I in the doxorubicin model (I), gremlin-1 in the bleomycin model (G), and gremlin-1 in the doxorubicin model (J). Six biological samples per condition (control, bibrosis) were used. Three technical replicates were performed per biological sample, and data are expressed as mean ± SD in arbitrary units (A.U.) (*p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.0001, by Mann–Whitney U test). Bleo, bleomycin; Doxo, doxorubicin.

FIGURE 4.

Serum Hsp90α levels in patients with SSc and their association with disease parameters. Expression of profibrotic proteins in the lung of fibrotic mice was determined. (A) Bar graph showing Hsp90α protein levels of lSSc, dSSc, and SSc patients with and without ILD. (B) Graph bar showing exponentiation of Hsp90α coefficient B values as a categorical variable. (C and D) Pearson’s analysis for correlation of Hsp90α with the mRSS (C) and with ILD (D) in patients with dSSc. (EJ) Overexpression of mRNA levels in mouse lung of the following genes: Hsp90aa1 in the bleomycin model (E), Hsp90aa1 in the doxorubicin model (H), COL I in the bleomycin model (F), COL I in the doxorubicin model (I), gremlin-1 in the bleomycin model (G), and gremlin-1 in the doxorubicin model (J). Six biological samples per condition (control, bibrosis) were used. Three technical replicates were performed per biological sample, and data are expressed as mean ± SD in arbitrary units (A.U.) (*p < 0.05, **p < 0.005, ***p < 0.0005, ****p < 0.0001, by Mann–Whitney U test). Bleo, bleomycin; Doxo, doxorubicin.

Close modal
Table II.

Cross table with Hsp90α category and subtype of SSc

Hsp90α CategorylSSc, N (%)dSSc, N (%)dSSc/lSSc
Hsp90α 5C 70 (83.3) 14 (16.7) 0.2 
Hsp90α 5C(1) 47 (73.4) 17 (26.6) 0.4 
Hsp90α 5C(2) 15 (62.5) 9 (37.5) 0.6 
Hsp90α 5C(3) 11 (52.4) 10 (47.6) 0.9 
Hsp90α 5C(4) 7 (33.3) 14 (66.7) 2.0 
Hsp90α CategorylSSc, N (%)dSSc, N (%)dSSc/lSSc
Hsp90α 5C 70 (83.3) 14 (16.7) 0.2 
Hsp90α 5C(1) 47 (73.4) 17 (26.6) 0.4 
Hsp90α 5C(2) 15 (62.5) 9 (37.5) 0.6 
Hsp90α 5C(3) 11 (52.4) 10 (47.6) 0.9 
Hsp90α 5C(4) 7 (33.3) 14 (66.7) 2.0 

Number (N), percentage, and proportion (dSSc/lSSc) of patients with the limited and diffuse subtypes of SSc in each Hsp90α category. Hsp90α was defined as a categorical variable divided into five categories of the following concentration range: Hsp90α 5C, 0–19 ng/ml; Hsp90α 5C(1), 20–33.9 ng/ml; Hsp90α 5C(2), 34–39.9 ng/ml; Hsp90α 5C(3), 40–54.9 ng/ml; and Hsp90α 5C(4), ≥55 ng/ml (see Supplemental Table III for the univariate statistical analysis).

To assess whether fibrosis severity correlates with serum Hsp90α level, we investigated the correlation between the semiquantitative score for the assessment of skin involvement in patients with SSc (the mRSS) and serum Hsp90α levels. A significant positive correlation (*p < 0.05) between the mRSS and Hsp90α levels was confirmed (Fig. 4C). Pulmonary fibrosis involvement in SSc patients is commonly associated with the diffuse subtype of the disease. SSc patients with lung affectation or ILD are known to show a prevalence of 1.2 per 100,000 person-years (3). Moreover, predictors of progression in SSc patients with ILD are under development because it is a leading cause of death in SSc disease (39). In this regard, FVC measurements in SSc patients affected by ILD prove the presence of shortness of breath and impaired lung function as an important diagnostic tool (39). FVC of the patients under study was correlated with serum Hsp90α levels, showing a significant inverse correlation (*p < 0.05) (Fig. 4D), indicating the relationship between Hsp90α and another variable related to the most severe subtype of the disease.

ILD usually stabilizes during the first 4–6 y after the onset of scleroderma, and spirometry is not sensitive or specific enough to determinate the progression of the disease (40). Finding a marker of the disease is necessary to treat ILD before it becomes a severe problem, which will improve the quality of life of patients. In the preceding section, we showed that Hsp90α levels were higher in patients with ILD compared with patients without ILD (Fig. 4A). This is consistent with observations from previous studies describing increased expression of total Hsp90 in the plasma of patients with SSc (38). This information would include Hsp90α as a potential biomarker of disease severity and would point to it as a possible target. To test the relationship between ILD and the expression of Hsp90α, we studied the expression of Hsp90α in two models of pulmonary fibrosis. For this purpose, we analyzed the involvement of Hsp90α in two mouse models of pulmonary fibrosis, that is, those in which mice were treated with bleomycin and doxorubicin, respectively. Lung samples from both models showed overexpression of Hsp90aa1 (Fig. 4E, 4H), COL I (Fig. 4F, 4I), and gremlin-1 (Fig. 4G, 4J) genes compared with controls. We are aware that the generation of lung fibrosis in SSc does not appear as aggressively as in the models we used in this study (intrathecal or nasal administration of profibrotic reagents). However, these models gave elevated Hsp90α values that could be related to manifestations associated with the dSSc, such as ILD. Moreover, the overexpression of gremlin-1 is in line with the description of gremlin-1 as a differential biomarker of ILDs, in particular idiopathic pulmonary fibrosis (41).

In this study, we have explored the involvement of the inducible Hsp90α isoform in the development of cutaneous and pulmonary fibrosis, and its potential role as a biomarker of the extent of fibrosis in human SSc.

Patients with scleroderma usually present a specific fibrosis-related phenotype characterized by excessive COL deposition and increased responsiveness to different profibrotic factors, in particular TGF-β (21), PDGF (platelet-derived growth factor) (42), or endothelin (ET-1) (43, 44). TGF-β is a key profibrotic factor that induces ECM synthesis and remodeling in SSc disease (45). Hsp90α is related to the TGF-β cascade in SSc (38) and is considered to be one of the extracellular partners of TGF-β (binding TGF-βRI) during profibrotic signaling in other fibrosis-related diseases, as pointed out by us previously (18, 23). The increased Hsp90α we have seen secreted by TGF-β–activated human skin fibroblasts compared with control cells would indicate a possibility that Hsp90α could behave as a biomarker for SSc. Specifically, we consider that Hsp90α might reflect the fibrotic stage of the disease because Hsp90α secretion was observed when the human skin fibroblasts were activated by TGF-β. Moreover, reduction of the profibrotic signal of fibroblasts by different secondary messenger inhibitors showed a reduction of Hsp90α secretion, as we previously demonstrated in other types of fibroblasts (23), which would indicate that fibroblasts return to a basal state in which Hsp90 secretion is not necessary. This demonstrates that the extracellular presence of Hsp90α might have the ability to reflect increased cutaneous fibrosis, similar to what has been previously demonstrated in studies available in other profibrotic environments such as cardiac fibrosis (23) or cancer progression (46). During fibrosis progression, as a part of the stress response, Hsp90α is overexpressed to stabilize, fold, or interact with TGF-β receptors promoting an efficient profibrotic signaling response (20, 23). Our results using fibrotic skin samples showed overexpression of Hsp90α in two murine models. Overexpression of Hsp90α was accompanied by increased transcript levels of ECM genes, such as COL I and fibronectin in the two murine models of cutaneous fibrosis studied. We interpreted these results as an indication of the involvement of Hsp90α in the profibrotic signaling cascade (18). These data further support the important role of Hsp90α as a participant in fibroblast activation that is responsible for cutaneous fibrosis.

With regard to fibrosis models and their monitoring, we contributed to the field with the development of a new in vivo fibrosis model (the intradermal doxorubicin model). This new model could contribute to find predictors of which very little is known. In this model, fibrosis progression can be monitored and detected in more localized areas and in less time than in previous models. We compared and studied the fibrotic process in the widely used bleomycin model and in the doxorubicin model that we present here as a new model to study the generation of SSc-associated skin fibrosis. Intradermal administration was the route of administration for this chemotherapeutic agent (doxorubicin), which provided additional technical and ethical advantages, such as better localization of lesions as well as a greater number of small lesions per mouse, reducing the number of mice per assay. We observed no difference in the number of infiltrating cells or circulating TGF-β levels between the bleomycin and the doxorubicin models, so inflammatory or systemic TGF-β–derived events are not increased in the new model. Autofluorescence of doxorubicin and COL allowed in vivo monitoring of reagent absorption and fibrosis formation, respectively.

In the model we describe herein, lesion formation is faster (1 wk) than the fibrosis formed in the bleomycin model, and fewer mice are needed because four lesions can be made per mouse instead of one that is made in the bleomycin model. In addition, because the robust doxorubicin method includes the ability to perform in vivo fluorescent monitoring of drug uptake (the drug will be absorbed when fluorescence is lost at the injection site), it allows prediction of the timing of administration once the previous dose has been absorbed. We confirmed the involvement of Hsp90α in the disease development with the results obtained in Hsp90α-deficient mice, in which skin thickness in damaged tissue was lower compared with WT mice in both models as demonstrated by histological quantification. Furthermore, fibrosis started later in the Hsp90α KO mice (in vivo imaging data) and resulted in reduced skin fibrosis. The absence of significant differences in Hsp90 KO animals compared with WT animals (doxorubicin model) shown between controls and fibrotic animals in relationship to the myofibroblast transition regulator, gremlin-1 (47), would indicate a less profibrotic environment in Hsp90α KO mice. The slight differences observed in vivo between WT and KO mice in the doxorubicin model make this in vivo fluorescent technique a complement supporting histological and gene expression assays in terms of quantifying fibrotic differences. The idea of utilizing in vivo monitoring for the COL formation is to help researchers visualize its formation. We conclude that the overexpression of Hsp90α was necessary for the progression of the skin fibrotic process in these animal models of SSc.

Although significant effort has been made to develop clinical and laboratory tools to assess or predict fibrosis progression in SSc, molecular detection of SSc remains a major challenge (48). Some serum biomarkers, such as CXCL4, TLR S100A8/A9, growth differentiation factor 15, lysyl oxidase, or type I IFN, have already been described and may help toward this purpose (4). Our data described the direct involvement of the stress-inducible Hsp90α isoform in the fibrotic process of SSc and its potential role as a biomarker of the most severe type of the disease. The use of ready-made commercial kits for the Hsp90α isoform is a breakthrough in terms of potential diagnostic applications. In SSc, Hsp90α levels clearly differentiated lSSc and dSSc phenotypes and were correlated with dSSc-associated scores (mRRS and FVC). It is known that this disease presents a high number of patients with pulmonary involvement (49) and that the most severe cases of SSc are usually associated with ILD. In a previous prospective analysis of patients with SSc-ILD, total plasma Hsp90 levels predicted the deterioration or stabilization of lung disease as measured by FVC and diffusing capacity for carbon monoxide (37). The previous association of increased total Hsp90 in SSc patients with impaired lung function (38) and increased total Hsp90 expression in lung fibrosis with elevated Hsp90 ATPase activity in fibroblasts isolated from fibrotic lung lesions (50) are in line with the hypothesis that Hsp90 plays an important role in the development of SSc-related pulmonary fibrosis. In our study, we go a step further to describe the importance of the stress-inducible isoform Hsp90α. Results obtained from mRNA from the lungs of pulmonary fibrosis models showed an increased presence of Hsp90α. As in other fibrotic diseases, the cellular role of Hsp90α in the fibrotic process of SSc could be related to the chaperone function of Hsp90α stabilizing TGF-β–related client proteins and participating in protein–protein interactions promoting maximum efficiency to the profibrotic TGF-β cascade. This dual function increases the importance of Hsp90α in the process. The association of Hsp90α with a patient’s FVC and the overexpression of Hsp90α fibrotic lungs of mice treated with bleomycin or doxorubicin indicated the potential of this isoform of Hsp90 and also of the gremlin-1 protein to act as markers of the pulmonary manifestations associated with the more severe form of the disease.

Limitations of this study include the absence of results in skin from patients with SSc. However, we verified that activation of human fibroblasts with TGF-β presented correct COL I overexpression, as known to occur in the primary skin fibroblasts of SSc patients. We showed the secretion of Hsp90α from these activated cells that would occur in fibroblasts from SSc patients. As for the in vivo assays, we used the bleomycin model as used in the literature, administering the reagent s.c. in the dorsal area of the subjects under study. This could be a potential source of bias in the comparison of the doxorubicin model in which the reagent was administered intradermally. Probably, if bleomycin was administered following the doxorubicin administration protocol, we could have obtained similar results for both models. Thus, the novelty of the doxorubicin model is related to the fluorescent detection of the profibrotic reagent that allows the monitoring of its absorption, the experimental design of the dosing schedule, and the route of administration used. Therefore, we recommend intradermal administration of the reagents in other SSc models.

Therefore, these results open the possibility of a new biomarker for the progression and extent of fibrosis in SSc. Hsp90α as a biomarker would present potential value in the evaluation of prognosis or therapeutic response in these patients. Its clinical value to classify or monitor progression of SSc will require further longitudinal and prospective studies.

We thank Dr. Victor Campa for valuable help.

This work was supported by Spanish Ministerio de Economía, Industria y Competitividad, Gobierno de España Grant RTI2018-095214-B-I00, as well as by the Instituto de Formación e Investigación Marqués de Valdecilla IDIVAL (InnVal 17/22; InnVal 20/34), 2020UCI22-PUB-0003 Gobierno de Cantabria (to A.V.V.), SAF2016-75195-R (to J.M.), SAF2017-82905-R (to R.M.), and (NextVal 18/14) to A.P.

Study conception and design, A.V.V., V.M.-T., J.L.P., J.M., R.M., and A.S.-M.; acquisition of data, J.R., P.C., E.T., F.D.-G., D.M., A.P., M.L.-H., A.V.V., and A.S.-M.; analysis and interpretation of data, P.M., P.C., A.V.V., V.M.-T., J.R., J.L.P., and A.S.-M. All authors provided critical revision of the manuscript and final approval of submission.

The online version of this article contains supplemental material.

Abbreviations used in this article:

COL

collagen

dSSc

diffuse SSc

ECM

extracellular matrix

ER

endoplasmic reticulum

FVC

forced vital capacity

Hsp90α

heat shock protein 90 isoform α

ILD

interstitial lung disease

KO

knockout

lSSc

limited SSc

mRSS

modified Rodnan skin score

qPCR

quantitative PCR

SSc

systemic sclerosis

SSC-ILD

SSc-associated ILD

WT

wild-type

1.
Smith
V.
,
C. A.
Scirè
,
R.
Talarico
,
P.
Airo
,
T.
Alexander
,
Y.
Allanore
,
C.
Bruni
,
V.
Codullo
,
V.
Dalm
,
J.
De Vries-Bouwstra
, et al
2018
.
Systemic sclerosis: state of the art on clinical practice guidelines.
RMD Open
4
(
Suppl 1
):
e000782
.
2.
Allanore
Y.
,
R.
Simms
,
O.
Distler
,
M.
Trojanowska
,
J.
Pope
,
C. P.
Denton
,
J.
Varga
.
2015
.
Systemic sclerosis.
Nat. Rev. Dis. Primers
1
:
15002
.
3.
Fan
Y.
,
S.
Bender
,
W.
Shi
,
D.
Zoz
.
2020
.
Incidence and prevalence of systemic sclerosis and systemic sclerosis with interstitial lung disease in the United States.
J. Manag. Care Spec. Pharm.
26
:
1539
1547
.
4.
Denton
C. P.
,
D.
Khanna
.
2017
.
Systemic sclerosis.
Lancet
390
:
1685
1699
.
5.
Young
A.
,
D.
Khanna
.
2015
.
Systemic sclerosis: a systematic review on therapeutic management from 2011 to 2014.
Curr. Opin. Rheumatol.
27
:
241
248
.
6.
Chung
L.
,
C. P.
Denton
,
O.
Distler
,
D. E.
Furst
,
D.
Khanna
,
P. A.
Merkel
;
Scleroderma Clinical Trials Consortium
.
2012
.
Clinical trial design in scleroderma: where are we and where do we go next?
Clin. Exp. Rheumatol.
30
(
2
,
Suppl 71
):
S97
S102
.
7.
Kania
G.
,
M.
Rudnik
,
O.
Distler
.
2019
.
Involvement of the myeloid cell compartment in fibrogenesis and systemic sclerosis.
Nat. Rev. Rheumatol.
15
:
288
302
.
8.
Novikova
N.S.
,
Diatlova
A.S.
,
Derevtsova
K.Z.
,
Korneva
E.A.
,
Abraham
L.
,
Quinn
S.
,
Segal
Y.
,
Churilov
L.P.
,
Fedotkina
T.V.
,
Ostrinski
Y.
, et al
.
2019
.
Tuftsin-phosphorylcholine attenuate experimental autoimmune encephalomyelitis. [Published erratum appears in 2020 J. Neuroimmunol. 339: 577137.]
J. Neuroimmunol.
337
:
577070
.
9.
Verrecchia
F.
,
A.
Mauviel
.
2007
.
Transforming growth factor-β and fibrosis.
World J. Gastroenterol.
13
:
3056
3062
.
10.
Semkova
M. E.
,
J. J.
Hsuan
.
2021
.
TGFβ-1 induced cross-linking of the extracellular matrix of primary human dermal fibroblasts.
Int. J. Mol. Sci.
22
:
984
.
11.
Bellaye
P. S.
,
O.
Burgy
,
S.
Causse
,
C.
Garrido
,
P.
Bonniaud
.
2014
.
Heat shock proteins in fibrosis and wound healing: good or evil?
Pharmacol. Ther.
143
:
119
132
.
12.
Li
L.
,
N. N.
Chen
,
Q. D.
You
,
X. L.
Xu
.
2021
.
An updated patent review of anticancer Hsp90 inhibitors (2013–present).
Expert Opin. Ther. Pat.
31
:
67
80
.
13.
Sreedhar
A. S.
,
E.
Kalmár
,
P.
Csermely
,
Y. F.
Shen
.
2004
.
Hsp90 isoforms: functions, expression and clinical importance.
FEBS Lett.
562
:
11
15
.
14.
Eguchi
T.
,
C.
Sogawa
,
Y.
Okusha
,
K.
Uchibe
,
R.
Iinuma
,
K.
Ono
,
K.
Nakano
,
J.
Murakami
,
M.
Itoh
,
K.
Arai
, et al
2018
.
Organoids with cancer stem cell-like properties secrete exosomes and HSP90 in a 3D nanoenvironment.
PLoS One
13
:
e0191109
.
15.
Feng
Y.
,
Q.
Wang
,
T.
Wang
.
2017
.
Drug target protein-protein interaction networks: a systematic perspective.
BioMed Res. Int.
2017
:
1289259
.
16.
Sontag
E. M.
,
R. S.
Samant
,
J.
Frydman
.
2017
.
Mechanisms and functions of spatial protein quality control.
Annu. Rev. Biochem.
86
:
97
122
.
17.
Taipale
M.
,
D. F.
Jarosz
,
S.
Lindquist
.
2010
.
HSP90 at the hub of protein homeostasis: emerging mechanistic insights.
Nat. Rev. Mol. Cell Biol.
11
:
515
528
.
18.
Cáceres
R. A.
,
T.
Chavez
,
D.
Maestro
,
A. R.
Palanca
,
P.
Bolado
,
F.
Madrazo
,
A.
Aires
,
A. L.
Cortajarena
,
A. V.
Villar
.
2018
.
Reduction of cardiac TGFβ-mediated profibrotic events by inhibition of Hsp90 with engineered protein.
J. Mol. Cell. Cardiol.
123
:
75
87
.
19.
Peng
Y. F.
,
H.
Lin
,
D. C.
Liu
,
X. Y.
Zhu
,
N.
Huang
,
Y. X.
Wei
,
L.
Li
.
2021
.
Heat shock protein 90 inhibitor ameliorates pancreatic fibrosis by degradation of transforming growth factor-β receptor.
Cell. Signal.
84
:
110001
.
20.
Beyer
C.
,
J. H.
Distler
.
2013
.
Tyrosine kinase signaling in fibrotic disorders: translation of basic research to human disease.
Biochim. Biophys. Acta Mol. Basis Dis.
1832
:
897
904
.
21.
Tomcik
M.
,
P.
Zerr
,
J.
Pitkowski
,
K.
Palumbo-Zerr
,
J.
Avouac
,
O.
Distler
,
R.
Becvar
,
L.
Senolt
,
G.
Schett
,
J. H.
Distler
.
2014
.
Heat shock protein 90 (Hsp90) inhibition targets canonical TGF-β signalling to prevent fibrosis.
Ann. Rheum. Dis.
73
:
1215
1222
.
22.
Noh
H.
,
H. J.
Kim
,
M. R.
Yu
,
W. Y.
Kim
,
J.
Kim
,
J. H.
Ryu
,
S. H.
Kwon
,
J. S.
Jeon
,
D. C.
Han
,
F.
Ziyadeh
.
2012
.
Heat shock protein 90 inhibitor attenuates renal fibrosis through degradation of transforming growth factor-β type II receptor. [Published erratum appears in 2022 Lab. Invest. 102: 212.]
Lab. Invest.
92
:
1583
1596
.
23.
García
R.
,
D.
Merino
,
J. M.
Gómez
,
J. F.
Nistal
,
M. A.
Hurlé
,
A. L.
Cortajarena
,
A. V.
Villar
.
2016
.
Extracellular heat shock protein 90 binding to TGFβ receptor I participates in TGFβ-mediated collagen production in myocardial fibroblasts.
Cell. Signal.
28
:
1563
1579
.
24.
Myung
S. J.
,
J. H.
Yoon
,
B. H.
Kim
,
J. H.
Lee
,
E. U.
Jung
,
H. S.
Lee
.
2009
.
Heat shock protein 90 inhibitor induces apoptosis and attenuates activation of hepatic stellate cells.
J. Pharmacol. Exp. Ther.
330
:
276
282
.
25.
Whitesell
L.
,
S. L.
Lindquist
.
2005
.
HSP90 and the chaperoning of cancer.
Nat. Rev. Cancer
5
:
761
772
.
26.
Li
W.
,
D.
Sahu
,
F.
Tsen
.
2012
.
Secreted heat shock protein-90 (Hsp90) in wound healing and cancer.
Biochim. Biophys. Acta Mol. Cell Res.
1823
:
730
741
.
27.
Zou
M.
,
A.
Bhatia
,
H.
Dong
,
P.
Jayaprakash
,
J.
Guo
,
D.
Sahu
,
Y.
Hou
,
F.
Tsen
,
C.
Tong
,
K.
O’Brien
, et al
2017
.
Evolutionarily conserved dual lysine motif determines the non-chaperone function of secreted Hsp90alpha in tumour progression.
Oncogene
36
:
2160
2171
.
28.
Khanna
D.
,
Nagaraja
V.
,
Tseng
C.H.
,
Abtin
F.
,
Suh
R.
,
Kim
G.
,
Wells
A.
,
Furst
D.E.
,
Clements
P.J.
,
Roth
M.D.
, et al
.
2015
.
Predictors of lung function decline in scleroderma-related interstitial lung disease based on high-resolution computed tomography: implications for cohort enrichment in systemic sclerosis-associated interstitial lung disease trials.
Arthritis Res. Ther.
17
:
372
.
29.
Adamson
I. Y.
1976
.
Pulmonary toxicity of bleomycin.
Environ. Health Perspect.
16
:
119
126
.
30.
Reyes-Habito
C. M.
,
E. K.
Roh
.
2014
.
Cutaneous reactions to chemotherapeutic drugs and targeted therapies for cancer: part I. Conventional chemotherapeutic drugs.
J. Am. Acad. Dermatol.
71
:
203.e1
203.e12
,
quiz 215–216
.
31.
LeRoy
E. C.
,
C.
Black
,
R.
Fleischmajer
,
S.
Jablonska
,
T.
Krieg
,
T. A.
Medsger
Jr.
,
N.
Rowell
,
F.
Wollheim
.
1988
.
Scleroderma (systemic sclerosis): classification, subsets and pathogenesis.
J. Rheumatol.
15
:
202
205
.
32.
Bhatia
A.
,
K.
O’Brien
,
J.
Guo
,
V.
Lincoln
,
C.
Kajiwara
,
M.
Chen
,
D. T.
Woodley
,
H.
Udono
,
W.
Li
.
2018
.
Extracellular and non-chaperone function of heat shock protein-90α is required for skin wound healing.
J. Invest. Dermatol.
138
:
423
433
.
33.
Ihn
H.
,
K.
Yamane
,
Y.
Asano
,
M.
Jinnin
,
K.
Tamaki
.
2006
.
Constitutively phosphorylated Smad3 interacts with Sp1 and p300 in scleroderma fibroblasts.
Rheumatology (Oxford)
45
:
157
165
.
34.
Ota
Y.
,
Y.
Kawaguchi
,
K.
Takagi
,
H.
Ichida
,
T.
Gono
,
M.
Hanaoka
,
T.
Higuchi
,
H.
Yamanaka
.
2013
.
Ghrelin attenuates collagen production in lesional fibroblasts from patients with systemic sclerosis.
Clin. Immunol.
147
:
71
78
.
35.
Du
Y.
,
J.
Wu
,
L.
Luo
.
2018
.
Secreted heat shock protein 90α attenuated the effect of anticancer drugs in small-cell lung cancer cells through AKT/GSK3β/β-catenin signaling.
Cancer Contr.
25
:
1073274818804489
.
36.
Avouac
J.
2014
.
Mouse model of experimental dermal fibrosis: the bleomycin-induced dermal fibrosis.
Methods Mol. Biol.
1142
:
91
98
.
37.
Dong
H.
,
M.
Zou
,
A.
Bhatia
,
P.
Jayaprakash
,
F.
Hofman
,
Q.
Ying
,
M.
Chen
,
D. T.
Woodley
,
W.
Li
.
2016
.
Breast cancer MDA-MB-231 cells use secreted heat shock protein-90alpha (Hsp90α) to survive a hostile hypoxic environment.
Sci. Rep.
6
:
20605
.
38.
Štorkánová
H.
,
Oreská
S.
,
Špiritović
M.
,
Heřmánková
B.
,
Bubová
K.
,
Komarc
M.
,
Pavelka
K.
,
Vencovský
J.
,
Distler
J.H.W.
,
Šenolt
L.
, et al
.
2021
.
Plasma Hsp90 levels in patients with systemic sclerosis and relation to lung and skin involvement: a cross-sectional and longitudinal study.
Sci. Rep.
11
:
1
.
39.
Distler
O.
,
S.
Assassi
,
V.
Cottin
,
M.
Cutolo
,
S. K.
Danoff
,
C. P.
Denton
,
J. H. W.
Distler
,
A. M.
Hoffmann-Vold
,
S. R.
Johnson
,
U.
Müller Ladner
, et al
2020
.
Predictors of progression in systemic sclerosis patients with interstitial lung disease.
Eur. Respir. J.
55
:
1902026
.
40.
Goh
N. S.
,
S. R.
Desai
,
S.
Veeraraghavan
,
D. M.
Hansell
,
S. J.
Copley
,
T. M.
Maher
,
T. J.
Corte
,
C. R.
Sander
,
J.
Ratoff
,
A.
Devaraj
, et al
2008
.
Interstitial lung disease in systemic sclerosis: a simple staging system.
Am. J. Respir. Crit. Care Med.
177
:
1248
1254
.
41.
Aoshima
Y.
,
Y.
Enomoto
,
S.
Muto
,
S.
Meguro
,
H.
Kawasaki
,
I.
Kosugi
,
T.
Fujisawa
,
N.
Enomoto
,
N.
Inui
,
Y.
Nakamura
, et al
2021
.
Gremlin-1 for the differential diagnosis of idiopathic pulmonary fibrosis versus other interstitial lung diseases: a clinical and pathophysiological analysis.
Lung
199
:
289
298
.
42.
Luchetti
M. M.
,
G.
Moroncini
,
M.
Jose Escamez
,
S.
Svegliati Baroni
,
T.
Spadoni
,
A.
Grieco
,
C.
Paolini
,
A.
Funaro
,
E. V.
Avvedimento
,
F.
Larcher
, et al
2016
.
Induction of scleroderma fibrosis in skin-humanized mice by administration of anti-platelet-derived growth factor receptor agonistic autoantibodies.
Arthritis Rheumatol.
68
:
2263
2273
.
43.
Leask
A.
2008
.
Targeting the TGFβ, endothelin-1 and CCN2 axis to combat fibrosis in scleroderma.
Cell. Signal.
20
:
1409
1414
.
44.
Jing
J.
,
T. T.
Dou
,
J. Q.
Yang
,
X. B.
Chen
,
H. L.
Cao
,
M.
Min
,
S. Q.
Cai
,
M.
Zheng
,
X. Y.
Man
.
2015
.
Role of endothelin-1 in the skin fibrosis of systemic sclerosis.
Eur. Cytokine Netw.
26
:
10
14
.
45.
Gyftaki-Venieri
D. A.
,
D. J.
Abraham
,
M.
Ponticos
.
2018
.
Insights into myofibroblasts and their activation in scleroderma: opportunities for therapy?
Curr. Opin. Rheumatol.
30
:
581
587
.
46.
Bohonowych
J. E.
,
M. W.
Hance
,
K. D.
Nolan
,
M.
Defee
,
C. H.
Parsons
,
J. S.
Isaacs
.
2014
.
Extracellular Hsp90 mediates an NF-κB dependent inflammatory stromal program: implications for the prostate tumor microenvironment.
Prostate
74
:
395
407
.
47.
Duffy
L.
,
J.
Henderson
,
M.
Brown
,
S.
Pryzborski
,
N.
Fullard
,
L.
Summa
,
J. H. W.
Distler
,
R.
Stratton
,
S.
O’Reilly
.
2021
.
Bone morphogenetic protein antagonist Gremlin-1 increases myofibroblast transition in dermal fibroblasts: implications for systemic sclerosis.
Front. Cell Dev. Biol.
9
:
681061
.
48.
Martyanov
V.
,
M. L.
Whitfield
.
2016
.
Molecular stratification and precision medicine in systemic sclerosis from genomic and proteomic data.
Curr. Opin. Rheumatol.
28
:
83
88
.
49.
Bredemeier
M.
2020
.
Correspondence on “Safety and tolerability of nintedanib in patients with systemic sclerosis-associated interstitial lung disease: data from the SENSCIS trial”.
Ann. Rheum. Dis.
DOI: 10.1136/annrheumdis-2020-218911
.
50.
Sontake
V.
,
Y.
Wang
,
R. K.
Kasam
,
D.
Sinner
,
G. B.
Reddy
,
A. P.
Naren
,
F. X.
McCormack
,
E. S.
White
,
A. G.
Jegga
,
S. K.
Madala
.
2017
.
Hsp90 regulation of fibroblast activation in pulmonary fibrosis.
JCI Insight
2
:
e91454
.

The authors have no financial conflicts of interest.

Supplementary data