Objective: Cartilage oligomeric matrix protein (COMP) accumulates in systemic sclerosis (SSc) skin and is upregulated by transforming growth factor (TGF)β. To further characterise the response to TGFβ in SSc, we investigated TGFβ1 and COMP expression and myofibroblast staining in SSc skin.
Methods: Skin biopsies from patients with diffuse cutaneous SSc (dSSc), limited cutaneous SSc (lSSc) and healthy controls were evaluated for COMP mRNA expression using real-time PCR. COMP, α-smooth muscle actin (SMA) and TGFβ were assessed in skin sections and in cultured fibroblasts by immunohistochemistry. Clinical disease status was assessed by the modified Rodnan skin score (mRSS).
Results: Myofibroblasts expressing SMA and COMP were found coexpressed in many cells in dSSc dermis, but each also stained distinct cells in the dermis. Cultured SSc dermal fibroblasts also showed heterogeneity for COMP and SMA expression, with cells expressing SMA, COMP, both or neither. TGFβ treatment increased COMP and SMA-expressing cells. COMP mRNA expression in lesional skin from patients with dSSc correlated with the mRSS and TGFβ1 staining.
Conclusion: These findings suggest that TGFβ upregulation of COMP and/or SMA expression in subpopulations of fibroblasts contributes to different pathways of fibrosis and that multiple TGFβ regulated genes may serve as biomarkers for the degree of SSc skin involvement.
Statistics from Altmetric.com
Systemic sclerosis (SSc) is an autoimmune disease characterised by fibrosis of the skin and multiple internal organs and by the presence of inflammatory cells in involved tissues.1 The underlying causes and the relationship between inflammation, fibroblast activation and fibrosis are not well understood, though it is generally accepted that transforming growth factor (TGF)β is one of the most important cytokines stimulating the fibrotic response.2 Evidence supporting a profibrotic function for TGFβ have been shown in vivo and in vitro,3 but in spite of these observations its levels are generally not elevated in SSc sera, and its detection in SSc skin remains inconsistent.4 Indirect evidence for the role of TGFβ in pathogenesis of fibrotic processes is the presence of myofibroblasts, as TGFβ strongly induces myofibroblast differentiation. Myofibroblasts, specialised fibroblasts expressing α-smooth muscle actin (SMA), are considered to play an important role in the pathogenesis of fibrosis by producing high levels of collagens and other matrix proteins.5 Although potentially stimulated in patients with SSc by TGFβ, myofibroblasts in SSc skin may, alternatively, be derived from pericytes, circulating fibrocytes or other stem cell population, or epithelial cells (reviewed in Varga and Abraham).6
Among possible factors that might contribute to the fibrotic process, we have shown previously that cartilage oligomeric matrix protein (COMP, or thrombospondin-5) accumulates in lesional and non-lesional skin, and in cultured fibroblasts from patients with diffuse cutaneous SSc (dSSc).7 COMP is a non-collagenous glycoprotein expressed in the extracellular matrix of articular cartilage, tendon and ligaments,8 and is normally produced by chondrocytes, osteoblasts and synovial fibroblasts.9 COMP binds to several extracellular matrix proteins, including collagens type I, II and IX collagens and fibronectin,10 11 and catalyses collagen fibrillogenesis. We have recently shown that COMP expression is strongly induced in dermal fibroblasts by TGFβ.7
Although fibrosis is considered a hallmark of SSc, it remains difficult to follow clinically, with the modified Rodnan skin score (mRSS) the most widely used outcome measure for skin disease.12 Despite limitations with the mRSS, there are no generally accepted biomarkers to follow and measure the outcome and progression of dermal fibrosis. We have recently shown that myofibroblasts are increased in the deep reticular dermis of SSc skin, and correlate with the mRSS.13 To further understand the role of TGFβ in SSc skin and its potential utility as a biomarker of SSc skin disease, in this study we investigated COMP, SMA and TGFβ1 expression in disease subsets of SSc, including patients with different levels of disease activity and patients with dSSc compared to limited cutaneous SSc (lSSc).
MATERIALS AND METHODS
Patient selection and skin scoring
Patients were recruited from the Boston University scleroderma programme. All patients met the American College of Rheumatology (ACR; formerly the American Rheumatology Association) criteria for the diagnosis of SSc,14 and gave informed consent to participate in this institutional review board approved study. Skin thickness was evaluated for each patients by experienced rheumatologists using the mRSS.15
A total of 90 human skin samples were obtained by 3–6 mm punch biopsies from lesional (with clinical skin disease from the dorsal mid or lower forearm of patients with dSSc and lSSc) and non-lesional skin (distant from any clinically detectable skin disease, generally the back, in dSSc). These samples were comprised of 9 lesional lSSc, 24 lesional and 12 non-lesional dSSc skin biopsies used for immunohistochemistry, and 30 lesional and 15 non-lesional dSSC skin biopsies used for mRNA studies. Eight skin biopsies from the dorsal forearm of healthy donors were obtained as controls.
Total RNA extraction
For RNA studies skin biopsies (1–3 mm) were collected in RNAlater (Qiagen, Carlsbad, California, USA), and stored at −20°C until processed. For RNA purification, tissues were transferred into 600 μl of RLT buffer (Qiagen), and disrupted using a polytron homogeniser. Homogenates were centrifuged at 12000 g for 5 min, and RNAs were purified from RLT buffer supernatants using the RNeasy total RNA kit (Qiagen). cDNAs synthesised from 0.5–1 μg of total RNA using Superscript II RNase H− reverse transcriptase (Invitrogen Life Technologies, Rockville, Maryland, USA) using random primers (Applied Biosystems, Branchburg, New Jersey, USA) were used as templates for quantitative real-time PCR.
The following primers were used for quantitative real-time PCR: COMP forward, 5′ AGC ACC GGC CCC AAG T 3′; reverse, 5′ GGT TGT GCC AAG ACC ACG TT 3′. COMP mRNA expression was measured using SYBR Green chemistry in an ABI Prism 7700 Sequence Detector (Applied Biosystems). To normalise for equal amounts of cDNA, the transcript levels of 18 S were assayed in the same samples (18 S TaqMan primers, Applied Biosystems).
For immunohistochemistry we used monoclonal rat anti-human COMP that recognises an epitope located between the two most N-terminal type II modules (1:100 dilution; MCA1455; Accurate Chemical, Westbury, New York, USA), monoclonal mouse anti-human SMA antibody (1:1000 dilution, clone IA4 DakoCytomation, Carpinteria, California, USA), or polyclonal rabbit anti-human TGFβ1 antibody (LC1–30, kindly provided by K Flanders, National Institutes of Health, Bethesda, Maryland, USA).16 Tissue sections were deparaffinised, rehydrated and blocked for 30 min with 3% bovine serum albumin in Tris-buffered saline (TBS) pH 7.4, and then incubated overnight with the primary antibody. For COMP staining, slides were incubated for 30 min with a biotinylated donkey anti-rat secondary antibody, and then incubated with streptavidin-linked alkaline-phosphatase (Jackson Immuno Research, West Grove, Pennsylvania, USA). SMA and TGFβ1 antibody-incubated slides were subsequently incubated with alkaline phosphatase-conjugated anti-mouse/anti-rabbit antibody (Evision, DakoCytomation, Carpinteria, California, USA) for 30 min at room temperature and detected with Fast Red (DakoCytomation). For COMP and SMA dual staining, COMP staining was followed by SMA staining, using horseradish peroxidase detection with peroxidase-linked antibody and 3,3′ diaminobenzidine (DAB) (DakoCytomation).
Scoring of immunohistochemical staining
COMP, SMA and TGFβ staining were evaluated using a 10 cm continuous visual analogue scale from 0 (no staining) to 100 (strong diffuse staining).13 17 COMP staining was calculated as an average of staining in papillary and the deep dermis. SMA and TGFβ scores were evaluated in the reticular dermis with the exclusion of vascular and glandular tissue for SMA and papillary dermis for TGFβ. Two experienced and blinded observers scored all sections independently and the mean was used for analysis.
Cell cultures and immunocytochemistry
Cultured human fibroblasts grown on 2-well chamber slides (Lab-Tek; Nalge Nunc, Rochester, New York, USA) were incubated in the absence of serum for 24 h prior to the addition of 5 ng/ml TGFβ1 (rhTGFβ1; R&D systems, Minneapolis, Minnesota, USA). Cells were then were fixed in 4% paraformaldehyde for 10 min at room temperature, permeabilised in 0.2% Triton X-100, and COMP and SMA proteins were detected as described above for tissue sections.
Linear regression and correlation coefficients were calculated using the statistical functions of Excel (Microsoft, Redmond, Washington, USA). The statistical significance of the differences between experimental and control groups were determined by analysis of variance and Wilcoxon two sample test. p Values <0.05 were considered significant. Intraclass correlation was calculated using an on-line calculator (http://department.obg.cuhk.edu.hk/researchSupport/Intraclass_correlatio.asp).
Increased COMP expression is found in skin of patients with lSSc and dSSc, but increased SMA expression is found only in skin of patients with dSSc
A total of 33 patients, 24 with dSSc and 9 with lSSc were studied. Clinical and demographic characteristics of the patients were similar with no significant differences between age and disease duration in either of the two groups (table 1). As expected, the mRSS was significantly increased in patients with dSSc (18.2 (2.1) vs 4.67 (1.08), p<0.001). We stained sections for COMP, SMA and TGFβ1. Scoring of stained sections was highly reproducible with the interclass correlation for SMA, COMP and TGFβ scoring: 0.93, 0.86 and 0.71, respectively.
COMP protein was significantly increased in lesional skin of patients with dSSc (24/24; mean score (standard error of the mean (SEM)) 54 (4)) and in almost all lesional skin of patients with lSSc (8/9 lesional skin; mean score 38 (7); lesional/dSSc vs lesional/lSSc p = 0.097; lesional/lSSc vs control p<0.001; lesional/dSSc vs control p<0.001). COMP staining was also increased in non-lesional skin from dSSc compared to controls (non-lesional/dSSc vs control p<0.05) (fig 1A:a–c,B:a). COMP protein was primarily expressed in the matrix, including the papillary and reticular dermis. COMP was also detected in the cytoplasm of a large number of fibroblasts, especially in some skin sections of patients with dSSc (fig 1, arrow). COMP expressing fibroblasts were found primarily close to blood vessels or adjacent to adnexal structures. COMP was distributed similarly in lSSc and dSSc skin (fig 1).
We found increased staining for myofibroblasts in the deep reticular dermis of lesional skin in patients with dSSc (fig 1A:e,B:b; mean score (SEM) 39 (5.54)) compared to control skin, which was consistently negative for myofibroblasts (fig 1A:f,B:b). Strikingly, none of the patients with lSSc showed myofibroblast positive staining on lesional tissue (fig 1A:d,B:b). Myofibroblasts were also significantly increased in non-lesional skin of patients with dSSc compared to controls (SMA mean score (SEM) 15 (3.74); diffuse non-lesional/dSSc vs control p<0.001). COMP, but not SMA, was also detected in 2/9 biopsies on non-lesional skin of patients with lSSc (data not shown).
Increased TGFβ1 expression in diffuse and limited SSc skin
TGFβ1, detected as intracellular staining, was found significantly increased in lesional skin from patients with lSSc (fig 1A:g,h,B:c; lesional/lSSc vs control p<0.05) and dSSc (lesional/dSSc vs control p<0.001). Most cells staining for TGFβ1 had a fibroblast morphology and were predominantly localised in the deep reticular dermis.
COMP mRNA and myofibroblasts are markers of skin disease severity and both correlate with mRSS in lesional diffuse SSc skin
We did not find any correlation between COMP protein expression and the local skin score (data not shown, R2 = 0.14; p = NS) or the mRSS (fig 2C, R2 = 0.004; p = NS). Since COMP protein might be stably incorporated into skin matrix and thus change slowly over time, we also examined whether COMP mRNA expression might be a marker of the extent of SSc skin disease. COMP expression was analysed in RNA extracted from 30 lesional and 15 non-lesional skin biopsies of patients with dSSc. Clinical and demographic characteristics of patients studied for COMP mRNA expression were similar to the previous group evaluated by immunohistochemistry (table 1). COMP mRNA was significantly overexpressed in dSSc lesional compared to control skin (fig 2A; lesional/dSSc compared to control 7.44-fold increase; p<0.001) and correlated with the degree of skin involvement as assessed by the mRSS (fig 2B, R2 = 0.35, p<0.001). COMP expression was also increased in non-lesional skin from patients with dSSc (fig 2A; non-lesional/dSSc compared to control 3.06-fold increase; p<0.05). COMP mRNA expression did not correlate with pulmonary function tests (forced vital capacity or diffusing capacity, data not shown). As we described previously,13 myofibroblast staining correlated with the mRSS (fig 2C, 2D, R2 = 0.31 p<0.01). We did not observe any significant correlation between disease duration and COMP or SMA expression (COMP and disease duration, R2 = 0.005; SMA and disease duration, R2 = 0.025).
TGFβ-regulated COMP and SMA expression show different but overlapping fibroblast populations in the skin
Dual staining for COMP protein and SMA on lesional skin showed that these proteins appear to be expressed by the same cells in some regions of fibrotic dermal tissues as observed by the presence of COMP staining around myofibroblasts (fig 3A). However, in other regions COMP and SMA could be found in distinct skin regions, with COMP staining in some areas not associated with myofibroblasts (fig 3A,B). In general, COMP protein was expressed more in the extracellular matrix (ECM), from papillary to reticular dermis. It was also expressed in cells with a fibroblast morphology, though generally sparing myofibroblasts, which were confined mainly to the deep reticular dermis. In spite of this, we found a positive correlation between COMP protein in the deep dermis and SMA expression (fig 3C:a; R2 = 0.44, p<0.005). In addition, SMA and COMP staining each correlated strongly with TGFβ1 staining (SMA: fig 3C:b R2 = 0.38, p<0.01; and COMP: fig 3C:c; R2 = 0.25, p<0.05).
TGFβ-regulated COMP and SMA expression in different but overlapping fibroblast populations in vitro
To determine whether the same population of dermal fibroblasts that express SMA also express COMP after TGFβ treatment, normal and lesional dSSc dermal fibroblasts, treated with TGFβ or left untreated, were stained for COMP and SMA. Notably, fibroblasts isolated from SSc and control skin biopsies stained for SMA, COMP, both or neither in the absence of treatment with TGFβ (fig 4A:a,c,e). We saw an increased number of SSc fibroblasts expressing SMA alone or COMP alone compared to fibroblasts from healthy control skin (fig 4A: compare a,c with e). In fibroblasts from control biopsies TGFβ increased the percentage of cells staining for SMA alone, COMP alone and SMA and COMP (fig 4A:e,f,B:c). In fibroblasts cultures from patients with SSc TGFβ increased primarily cells staining for SMA and COMP (fig 4A:b,d,B:a,b).
COMP and SMA were also stained in cultured fibroblasts from lesional and non-lesional skin biopsies derived from the same patients. COMP and/or SMA were expressed in a high percentage of cells in the absence of TGFβ, and TGFβ induced expression of both markers in additional cells (fig 5). Although there was considerable heterogeneity in the percentage of COMP and SMA staining cells between patients, there was no statistical difference between lesional/non-lesional-fibroblast staining for SMA or COMP, before or after TGFβ induction.
In this study we show that skin fibrosis in patients with SSc is consistently associated with increased COMP expression in the ECM, as COMP protein shows increased expression in lesional and non-lesional skin from patients with diffuse disease, as well as lesional skin from patients with limited disease. Increased COMP expression in non-lesional dSSc skin was not due to subtle, clinically apparent changes in non-lesional skin, as non-lesional skin biopsies were taken distant from lesional skin.
SMA-expressing myofibroblasts are known to be a TGFβ differentiated fibroblast population, with an aggressive phenotype adapted to reorganise extracellular matrix during scar formation. Their role in scleroderma fibrosis has been long studied 2 18 and they are found in high proportions in SSc skin.19 20 We have shown that COMP is also a TGFβ inducible gene in cultured SSc fibroblasts.7 In the current study we find that COMP and SMA are induced in distinct, though overlapping, fibroblast subpopulations upon TGFβ treatment. After TGFβ treatment few fibroblasts expressed both, but the majority of fibroblasts expressed one or the other protein, suggesting that TGFβ is responsible for inducing different programs of gene expression by heterogeneous dermal fibroblast subpopulations in vitro and in vivo. Past studies have shown that fibroblasts are heterogeneous in terms of their functional properties, although the basis for this heterogeneity is incompletely defined.21 Earlier studies attributed fibroblast heterogeneity to differences in intracellular signalling following stimuli, as fibroblast clones showed variability in their rates of proliferation and collagen synthesis.22 23 Other studies showed that different fibroblast subpopulations can release widely different levels of mediators, such as prostaglandin E2 (PGE2), independent of the particular stimulus used, with clones releasing the highest levels of PGE2 manifesting this phenotype even in absence of a specific stimulus.24 Moreover, strains of fibroblasts producing high levels of one product (pGE2) in response to interleukin (IL)1 did not necessarily extend to other fibroblast IL1-inducible products (such as collagenase), suggesting that different fibroblast subpopulations secrete distinct metabolic components when activated.25 Fibroblast heterogeneity has also been found to be manifest in patients with SSc by altered susceptibility to apoptosis.18 Our study extends these observations to show that clonal heterogeneity also occurs in response to TGFβ. These results highlight the potential roles for distinct TGFβ-stimulated fibroblast subpopulations in SSc skin and suggest that heterogeneous fibroblast responses to TGFβ in SSc skin may contribute in different ways to tissue fibrosis.
The origin of myofibroblasts in SSc has been of considerable interest. Different studies have suggested that pericytes,26 27 epithelial cells,6 28 or circulating fibrocytes might be precursors of myofibroblasts.29 30 Our results showing that different subpopulations of fibroblasts express TGFβ-inducible genes in SSc skin and that similar effects can be reproduced in vitro in cultured fibroblasts from healthy controls, suggest that different resident fibroblasts subpopulations in SSc skin are being stimulated by TGFβ. Strikingly, these changes in COMP gene expression and SMA staining cells correlated with staining of TGFβ1 in the skin. Thus, these results suggest that myofibroblasts and COMP-expressing fibroblasts are induced by the local action of TGFβ on resident, heterogeneous fibroblast populations. Although, these results are not inconsistent with other hypothesis for myofibroblast origin, such hypotheses would have to be broadened to explain the presence of another subpopulation of TGFβ-responsive fibroblasts.
Surprisingly, unlike COMP, myofibroblasts appeared exclusively in the skin of patients with dSSc and not in the skin of patients with lSSc. Histological differences between limited and diffuse SSc disease subsets have not been studied extensively. The lack of myofibroblasts in lSSc skin might be because myofibroblasts differentiate only transiently in the most active skin disease, representing a marker of an aggressive fibroblast population that is not expressed in mild or late disease. Another possibility is that patients with lSSc respond differently to TGFβ, suggesting that the differences in disease phenotype between lSSc and dSSc might be related to differences in sensitivity of dermal fibroblast subpopulations to TGFβ.
The identification of biomarkers for assessing disease activity is an important goal in developing better outcome measures for clinical trials. In this study we show that dermal COMP mRNA expression correlates well with the extent of clinical skin disease as evaluated by the mRSS. COMP protein did not correlate well with the mRSS possibly because it persists longer when the disease is clinically inactive, as in limited disease. Recently another group has shown that COMP serum levels correlate with mRSS.31 Although our results do not show a correlation of COMP protein staining in the skin with the mRSS, serum COMP might be less stable than the matrix protein and thus serum COMP and COMP mRNA levels may be more indicative of ongoing TGFβ-regulated COMP production. We have recently shown that myofibroblasts also correlate with the mRSS, suggesting that multiple biomarkers of TGFβ correlate with the mRSS and thus might collectively serve as valuable biomarkers for disease activity.
Competing interests: None.
Funding: This study was supported by grants awarded to RL from the Scleroderma Foundation and the NIH: NIAMS Grant R01AR051089 and U01AR055063. This work was also supported by an NIH grant supporting the Boston University School of Medicine General Clinical Research Center Grant, M01 RR00533 and by an unrestricted grant from the American Society for Scleroderma Research.
Ethics approval: Ethics approval was obtained.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.