Monocyte chemoattractant protein 3 as a mediator of fibrosis: Overexpression in systemic sclerosis and the type 1 tight-skin mouse

Authors


Abstract

Objective

To determine the gene-expression profile in dermal fibroblasts from type 1 tight-skin (Tsk1) mice, and to examine the expression and potential fibrotic activity of monocyte chemoattractant protein 3 (MCP-3) in Tsk1 mouse and human systemic sclerosis (SSc) skin.

Methods

Complementary DNA microarrays (Atlas 1.2) were used to compare Tsk1 fibroblasts with non-Tsk1 littermate cells at 10 days, 6 weeks, and 12 weeks of age. Expression of MCP-3 protein was assessed by Western blotting of fibroblast culture supernatants, and localized in the mouse and human skin biopsy samples by immunohistochemistry. Activation of collagen reporter genes by MCP-3 was explored in transgenic mouse fibroblasts and by transient transfection assays.

Results

MCP-3 was highly overexpressed by neonatal Tsk1 fibroblasts and by fibroblasts cultured from the lesional skin of patients with early-stage diffuse cutaneous SSc. Immunolocalization confirmed increased expression of MCP-3 in the dermis of 4 of 5 Tsk1 skin samples and 14 of 28 lesional SSc skin samples, compared with that in matched healthy mice (n = 5) and human controls (n = 11). Proα2(I) collagen promoter–reporter gene constructs were activated by MCP-3 in transgenic mice and by transient transfection assays. This response was maximal between 16 and 24 hours of culture and mediated via sequences within the proximal promoter. The effects of MCP-3 could be diminished by a neutralizing antibody to transforming growth factor β.

Conclusion

We demonstrate, for the first time, overexpression of MCP-3 in early-stage SSc and in Tsk1 skin, and suggest a novel role for this protein as a fibrotic mediator activating extracellular matrix gene expression in addition to promoting leukocyte trafficking. This chemokine may be an important early member of the cytokine cascade driving the pathogenesis of SSc.

Systemic sclerosis (SSc) is a multisystem connective tissue disorder characterized by skin thickening and widespread, but variable, visceral fibrosis. Its pathogenesis involves immunologic activation and vascular dysfunction leading to excessive accumulation of extracellular matrix (ECM) in lesional tissues. A number of well-characterized animal models have been used to elucidate the pathogenesis of SSc. The type 1 tight-skin (Tsk1) mouse develops skin fibrosis reminiscent of human SSc, with increased synthesis and accumulation of collagen-rich matrix in the skin. Activated fibroblasts are implicated in the overproduction of ECM components, but the mechanisms by which these cells are activated in SSc remain incompletely understood.

Soluble profibrotic mediators, including transforming growth factor β1 (TGFβ1) and connective tissue growth factor (CTGF), have been shown to be up-regulated in SSc, and a hierarchical cascade of cytokines in which initial induction of proinflammatory cytokines leads to later expression of profibrotic mediators has been proposed (1). Inflammatory cell recruitment is an early feature of SSc and may be involved in the activation of fibroblasts. Vectorial migration of inflammatory cells into subendothelial tissues is a multistep process which is mediated by a series of cellular interactions that establish chemotactic gradients in the perivascular space. Both resident fibroblasts and extravascular leukocytes may provide the initial migratory trigger. Members of the chemokine family of proteins are implicated in the regulation of this process (2). Chemokines are small, basic peptides having a molecular weight of 8–11 kd. Like other cytokines, they may be produced constitutively or after induction, and exert their effect locally or in a paracrine or autocrine manner. They are classified into 4 groups: CC, CXC, C, and CX3C, according to the position of 2 highly conserved cysteine residues (3).

Monocyte chemoattractant protein (MCP) types 1, 2, 3, 4, and 5 constitute a subfamily within the CC(β) chemokines. MCP-1 was first identified as a monocyte-specific cytokine, followed by MCP-2, MCP-3, MCP-4, and MCP-5 (4–7). There is high amino-acid sequence homology among the 5 MCP chemokines (60–70%) compared with ∼40% homology between other, non-MCP CC chemokines. MCP chemokines exert their putative effects through activation of specific G protein–coupled 7-transmembrane receptors. They are proinflammatory in nature, and chemotactic for monocytes, eosinophils, and basophils. Chemokine expression is increased in a number of different pathologic processes, including synovial pannus of rheumatoid joints, autoimmune lesions of multiple sclerosis, affected mucosal surfaces in ulcerative colitis and Crohn's disease, lung inflammation in chronic bronchitis, sarcoidosis, and asthma, and the vascular inflammation that characterizes atherosclerosis (8–12). Overexpression of MCP-1 by SSc fibroblasts was first found to underlie the promotion of leukocyte migration across endothelial cell monolayers by these cells (13), and other studies have subsequently confirmed overexpression of MCP-1 in SSc tissues (14, 15) and by lesional fibroblasts in culture (16). Up-regulation of MCP-1 has also been demonstrated in several chronic inflammatory disorders (17).

In the present study, we describe, for the first time, the overexpression of another CC chemokine, MCP-3, in the Tsk1 mouse model using gene-expression profiling, and investigate whether this factor is also overexpressed in human SSc. Since inflammatory cell infiltration is not a feature of Tsk1 skin, and on the basis of analogy with MCP-1 (18, 19), we have investigated the potential for MCP-3 to promote collagen-gene expression and thereby operate as a novel mediator of fibrosis.

PATIENTS AND METHODS

Patients and controls.

All patients and controls participated in the study after providing their informed consent, in accordance with local institutional guidelines. Dermal punch biopsy samples were obtained from 22 patients with diffuse cutaneous systemic sclerosis (dcSSc), 6 with limited cutaneous systemic sclerosis (lcSSc), and 11 healthy controls. Skin samples were immediately divided for tissue culture and snap frozen for histologic study. Patients' characteristics are summarized in Table 1. Samples were classified as having early dcSSc (n = 14) if obtained within the first 2 years of disease onset, as defined by the appearance of the first non–Raynaud's phenomenon symptom. Patients with established dcSSc (n = 8) had >2 years of disease.

Table 1. Clinical features of systemic sclerosis patients and healthy controls*
 dcSSclcSSc (n = 6)Controls (n = 11)
Early (n = 14)Established (n = 8)
  • *

    dcSSc = diffuse cutaneous systemic sclerosis; lcSSc = limited cutaneous systemic sclerosis.

  • Defined by <2 years' duration.

Characteristic    
 Sex, male/female, no.3/112/60/65/6
 Age, mean ± SD years49.3 ± 9.957.3 ± 12.444.3 ± 16.245 ± 14.4
 Duration of disease, mean ± SD months16.7 ± 4.269 ± 51.266 ± 58.2
Duration of Raynaud's phenomenon symptoms, mean ± SD months18.9 ± 6.675 ± 53.2154 ± 49.4
Organ involvement, %    
 Esophageal718883 
 Other gastrointestinal142533 
 Lung365017 
 Muscle141317 
 Joint1400 
 Renal21017 
 Cardiac700 
 Neurologic0130 
Serology, %    
 Antinuclear100100 
 Anticentromere067 
 Anti–topoisomerase I3217 
 Anti–RNA polymerase I/III90 
 Anti–nuclear RNP017 
 Other590 

All SSc patients fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) preliminary criteria for the classification of SSc (20). Patterns of organ involvement at the time of biopsy were assessed according to standard practice. Briefly, esophageal involvement was determined by the medical history, the barium swallow test, or scintigraphy. Lung fibrosis was assessed by high-resolution computed-tomography scanning in the presence of a restrictive pattern of pulmonary function abnormalities. Cardiac involvement was considered present if electrocardiogram alterations, impaired ventricular function, or pericardial effusion were evident on echocardiography. Pulmonary arterial hypertension (PAH) was determined by an elevated pulmonary artery peak systolic pressure on Doppler echocardiography, or by other echocardiographic hallmarks of PAH, and usually confirmed by right heart catheterization. A creatinine kinase level elevated more than 2-fold defined skeletal muscle involvement. Renal involvement was identified by a history of scleroderma renal crisis or significant impairment of creatinine clearance.

Patients were treated with vasodilators and, in most patients with dcSSc, by immunosuppression using anti–thymocyte globulin and mycophenolate mofetil (21). Biopsy samples of skin from patients with early-stage dcSSc were obtained prior to immunosuppressive treatment. Patients with lcSSc were receiving vasodilator therapy only at the time of biopsy. Among patients with established dcSSc, 88% (n = 7) were taking immunosuppressant therapy. Three patients with lcSSc had pulmonary, renal, and muscle involvement requiring treatment with low-dose corticosteroids and immunosuppression.

Tight-skin and transgenic mice.

To examine the profibrotic effects of MCP-3 on extracellular matrix gene expression, we used dermal fibroblasts cultured from transgenic mice harboring a bacterial β-galactosidase reporter gene regulated by a fibroblast-specific expression cassette (2kb-LacZ). A far upstream transcriptional enhancer between −17.1 kb and −15.1 kb of the transcription start site and linked to a minimal endogenous promoter shows a consistently high level of fibroblast-specific expression in embryonic development and postnatally. In these mice, basal high-level expression of the reporter gene correlates with expression of the type I collagen gene. Selective up-regulation of this reporter transgene has been demonstrated in vivo, using Tsk1 mice (22). Tsk1 heterozygote males were crossed with homozygous pallid females so that genotyping could be assessed by coat color from 7 days postnatally.

For gene-expression profiling and assessment of MCP-3 expression, skin biopsy samples were obtained from the interscapular region of Tsk1 (n = 5) and wild-type (pallid coat color, n = 5) littermate mice at between 3 days and 12 weeks of age. For most immunostaining experiments, 10-day- and 3-week-old mouse tissues were examined, since these time points gave consistent and reproducible expression patterns.

Fibroblast cultures.

Fibroblasts derived from 3 sources were used for these experiments. Tight-skin and non–tight-skin dermal fibroblasts were cultured from skin extracted from the upper back of same-sex littermate mice. For functional studies of the fibroblast response to MCP-3, reporter transgenic mouse fibroblasts were cultured from the 2kb-LacZ mouse line as described above. This allowed expression of the reporter transgene to be examined in tissue culture, and allowed the effect of recombinant MCP-3 on the level of transgene expression to be assessed. Human dermal fibroblasts were cultured from dcSSc patients and from age- and site-matched control human skin. Neonatal mouse fibroblasts were used for transient transfection experiments, since they show consistent responsiveness to recombinant cytokines.

Cells were grown by explant culture and maintained in Dulbecco's modified Eagle's medium (Gibco BRL, Grand Island, NY) supplemented with 10% fetal calf serum (Gibco BRL), 100 units/ml penicillin, and 100 mg/ml streptomycin, and cultured in a humidified atmosphere of 5% CO2 in air. For experiments, fibroblasts were generally used at the first or second passage, because later-passage cells showed greater interculture variability. All cultures were inspected at high power using a phase-contrast inverted microscope, to confirm absence of epithelial cells and a typical fibroblastic morphology.

Analysis of gene expression by complementary DNA (cDNA) microarray.

To determine differences in the expression profiles of fibroblasts from Tsk1 and non-Tsk1 mice, total messenger RNA (mRNA) was prepared from confluent cultures of early-passage neonatal dermal fibroblasts using Trizol (Gibco BRL), in accordance with the manufacturer's protocol. Expression analysis was performed using the mouse Atlas 1.2 array (ClonTech, Palo Alto, CA), which incorporates oligonucleotides specific for 1,176 mouse-gene transcripts. A cDNA-synthesis gene-specific primer mix, used for probe synthesis, is enriched for sequences corresponding to the cDNA for the genes on the array. Hybridization was performed according to the manufacturer's instructions (ClonTech). Briefly, after DNAse I treatment of the total RNA to remove any genomic contaminants, 5 μg of each paired sample was incubated with the sequence-specific primer and reverse transcriptase. The resulting cDNA probes, labeled by incorporation of α32P-dATP, were hybridized with microarrays. The methods are discussed in detail on the manufacturer's Web site (http://www.clontech.com/techinfo).

Experiments were performed in parallel, and generally, 4 membranes were probed at the same time. Initial experiments suggested that this gave more reproducible data when comparing expression patterns between samples. After hybridization, membranes were washed and radioactivity was determined by phosphorimaging. Differential gene expression was assessed using AtlasImage software (ClonTech). Normalization of the expression data was achieved by using the sum of the global intensities of the arrays with the local background signal for each cDNA spot taken into account. This analysis used a default difference threshold of 7 for gene detection, and in order for the ratio of the 2 individual samples to be significant, a 2-fold difference in gene expression was selected, and this difference must have been observed on at least 3 occasions, including 2 independent experiments. To minimize the effect of genetic background and sex, littermate animals were examined and same-sex (male) mice were used.

Western blot analysis.

To extend the cDNA microarray data, fibroblasts from Tsk1 and non-Tsk1 and human SSc or control samples were used. After overnight incubation of skin samples in serum-free medium (200 μl per well), culture supernatants were collected and cell layers were lysed with Laemmli lysis buffer. Sodium dodecyl sulfate–polyacrylamide gel electrophoresis was performed on 18% Tris-glycine gels (Invitrogen, Paisley, UK), and the separated proteins were transferred onto nitrocellulose membranes at 30V for 90 minutes. Membranes were blocked by incubation for 1 hour with 5% nonfat milk in phosphate buffered saline (PBS) containing 0.2% Tween 20, and were stained for 1 hour using the following antibodies: goat anti-human MCP-1 antibody, goat anti-human MCP-3 antibody (100 μg/ml) (all from R&D Systems, Oxford, UK), and goat anti-mouse MCP-3 antibody (500 gm/ml) (Abcam, Cambridge, UK) at a dilution of 1:1,000. The nitrocellulose transfers were washed 3 times before being stained with biotinylated species-specific secondary antibody for 1 hour, and again washed 3 times before being stained with biotin substrate (Vectastain; Vector Laboratories, Peterborough, UK) and then chemiluminescent substrate (Amersham Pharmacia Biotech, Little Chalfont, UK), and developed against photographic film (Hyperfilm Enhanced Chemiluminescence; Amersham Pharmacia Biotech). Parallel blots of cell lysates from each culture were probed with a monoclonal antibody against β-actin (Sigma, St. Louis, MO) to control for variation in cell number between cultures.

Immunohistochemical staining of serial sections.

For detection of MCP-3 protein, immunohistochemistry analysis with a biotin/streptavidin-based amplifying system was performed on mice and human skin sections. Serial frozen sections (5 μm) were cut on a cryostat at −30°C and air-dried for an hour. Sections were fixed in ice-cold acetone and covered with 3% hydrogen peroxide for 10 minutes in the dark to block endogenous peroxidase activity. Slides were then blocked with 20% normal horse serum, and incubated with goat polyclonal anti-mouse MARC/MCP-3 IgG antibody and goat polyclonal anti-human MCP-3 IgG antibody (25 μg/ml in PBS; R&D Systems) for 1 hour at room temperature. After washing with PBS, sections were incubated with biotinylated horse anti-goat IgG diluted in PBS (7.5 μg/ml, BA-9500; Vector Laboratories) for 30 minutes, rinsed, and finally incubated with Vectastain Elite STR-ABC reagent (Vector Laboratories) for 30 minutes. After washing, sections were visualized using 3-amino-9-ethylcarbazole chromogen and H2O2 as substrate (SK-4200; Vector Laboratories). Sections were then washed in tap water, counterstained with Carrazzis hematoxylin, and mounted with Gelmount (Biomeda, Foster City, CA) for examination using an Olympus BH-2 photomicroscope. Controls included an exchange of primary antibodies with goat isotype-matched antibodies.

Transactivation of proα2(I) collagen reporter genes by recombinant MCP-3.

The effect of recombinant murine MCP-3 was examined using tissue-culture cells. Initially, transgenic fibroblasts from the 2kb-LacZ line were used, since these cells provide a highly physiologic assay for transactivation of a genomic β-galactosidase reporter for which endogenous expression in vivo recapitulates that of endogenous type I collagen genes. Concentrations between 1 ng/ml and 500 ng/ml of recombinant chemokine were evaluated in early experiments, and 100–400 ng/ml gave consistent effects and were used in the majority of later studies. Transient transfection was used to assess MCP-3 responsiveness of other proα2(I) collagen promoter constructs. Analyses were performed in 24-well plates using Lipofectamine-plus, as described by the manufacturer (Life Technologies, Gaithersburg, MD). Briefly, 1 μg of collagen reporter plasmid was cotransfected with 0.1 μg of pCMVluc control plasmid to allow correction of β-galactosidase expression for transfection efficiency. The reporter constructs used are shown in Figure 1. These included minimal promoter sequences of mouse or human proα2(I) collagen as well as constructs that incorporate sequences from the far-upstream fibroblast-specific enhancer previously defined in transgenic mice. This element operates as a lineage-specific transcriptional enhancer in vivo and is a target for activation in Tsk1 mice (22). Reporter gene expression was determined using the Galactolight assay kit (Tropix, Bedford, MA).

Figure 1.

Constructs used to examine proα2(I) collagen gene activation. The minimal promoter sequences were previously defined in transient transfection, from −354 bp and −378 bp upstream of the transcription start site for the mouse and human proα2(I) collagen genes, respectively. Larger constructs incorporated additional sequences from the evolutionarily conserved far-upstream transcriptional enhancer that has previously been shown to be a target for activation in fibrosis. The 6kb-LacZ construct includes the region from −19.5 kb to −13.5 kb upstream of the transcription start site, and the 2kb-LacZ incorporates 2 kb of the upstream region linked to a minimal murine −354-bp promoter. Construction of the human 5kb-LacZ contains sequences −22.8 kb to −17.5 kb, which is linked to a minimal −378-bp promoter.

TGFβ neutralization studies.

To examine the potential role of TGFβ isoforms as downstream effectors of collagen gene activation by MCP-3, we used a pan-specific anti-TGFβ monoclonal antibody (1D11; R&D Systems) to block the effects of recombinant MCP-3 on neonatal mouse fibroblasts harboring a minimal COL1A2 promoter–reporter construct. Initial experiments suggested that a concentration of 50 μg/ml blocked the maximal effect of recombinant TGFβ1 (data not shown), and this concentration was selected for neutralization experiments using MCP-3.

Statistical analysis.

For quantitative variables, the mean ± SEM results from replicate samples, or from combined independent experiments where between-experiment variation allowed reliable combination of raw data, were compared. Means were compared by Student's paired or unpaired t-test, and a P value less than 0.05 was taken as statistically significant.

RESULTS

Gene-expression analysis identifies overexpression of MCP-3.

Parallel assessment of gene expression was performed using cDNA microarrays, comparing fibroblasts from Tsk1 and wild-type (control) littermate skin biopsy samples at 3 time points: neonatal, 6 weeks old, and 12 weeks old. Neonatal fibroblasts expressed 244 (21%) of the 1,176 genes represented on the array. At 6 and 12 weeks, 147 genes (13%) and 94 genes (8%), respectively, were expressed. Representative portions of the cDNA expression arrays for neonatal fibroblasts are shown in Figure 2A. Differential expression was defined by 2-fold differences in hybridization between the control and Tsk1 mRNA samples. Thus, 51 genes were differentially expressed neonatally, whereas the number of differentially expressed transcripts decreased among fibroblasts from adult mice. No significant differences were observed between the 6- and 12-week expression profiles.

Figure 2.

Overexpression of monocyte chemoattractant protein 3 (MCP-3) in type 1 tight-skin (Tsk1) neonatal mouse fibroblasts. A, Representative neonatal fibroblast gene-expression profiles are shown using portions of 2 mouse Atlas 1.2 cDNA arrays, corresponding to cytokine and growth factor or extracellular matrix transcripts. Analysis using AtlasImage software reveals significant up-regulation (shown in red) of tissue inhibitor of matrix metalloproteinase 3 (TIMP3), laminin β2 subunit (LAMC1), fibronectin 1 (FN1), thrombospondin 1 (TSP1), MCP-3, cytoplasmic dynein light chain 1 (dynein), Bcl-2 interacting protein (NIP3), inhibin βA subunit (inhibin), insulin-like growth factor binding protein 6 (IGFBP6), and cholecystokinin A receptor (CCKR). Conversely, there was down-regulation (shown in blue) of cordon bleu protein (COBL), Cell Division Cycle 46 (CDC46), RAD23 ultraviolet excision repair protein (RAD23), laminin γ1 subunit (laminin), CD28 precursor, and interleukin-1 receptor (IL1R). Genes that are similarly expressed in both samples are indicated in green. Half green/half red and half green/half blue indicate that although a difference (lower portion) was observed between the defined samples, the ratio (upper portion) is nonsignificant. The data are representative of triplicate experiments. B, Increased MCP-3 protein secretion by Tsk1 fibroblasts. The most highly overexpressed transcript in neonatal Tsk1 fibroblasts in replicate experiments was MCP-3. Overexpression was confirmed by Western blot analysis of Tsk1 and control (wild-type [Wt]) fibroblast culture supernatants. Cell-layer β-actin was used as a protein-loading control. Protein levels were quantified in a representative experiment as shown with the corresponding histogram.

For 13 genes, there was differential expression at either the neonatal or 12-week time point or at both of these time points. These genes, together with the normalized expression relative to the basal wild-type neonatal fibroblast levels, are listed in Table 2. More than 40 of the 51 genes that were differentially expressed neonatally were not expressed consistently in adult samples, and these were not analyzed further. Conversely, 3 genes that were not differentially expressed in neonatal fibroblasts were found to be differentially expressed in adult cells, and these are included. Some genes, including thrombospondin 1, osteopontin, hypoxia inducible factor 1α, tissue inhibitor of matrix metalloproteinase 3, and TGFβ1, showed sustained overexpression at all time points (prelesional through established). Others, includ-ing biglycan, were overexpressed neonatally but suppressed in fibroblasts derived from established fibrotic skin.

Table 2. Summary of consistently differentially expressed genes in neonatal or adult type 1 tight-skin (Tsk1) mouse fibroblasts*
Gene namePrelesional (neonatal)Established (12 weeks)GenBank accession no.
PallidTsk1RatioPallidTsk1Ratio
  • *

    Genes listed are those with consistent differential expression either in early or late-stage Tsk1 in 3 experiments. Values for expression data are relative to the mean signal for the total signal intensity of expressed genes, after correction for local background signal.

  • Differential gene expression in Tsk1 fibroblasts compared with non-Tsk1 (pallid) littermate cells.

Monocyte chemoattractant protein 3 (MARC-1, CCL7)0.22.715.32.11.40.7S71251
Monocyte colony-stimulating factor 10.33.410.90.40.82.2X05010
Cluster differentiation antigen 440.10.74.90.61.11.8M27129
Thrombospondin 11.86.63.72.14.22.0M87276
Hypoxia inducible factor 1α0.20.73.00.40.22.0U59496
Tissue inhibitor of matrix metalloproteinase 30.20.63.00.40.72.0L19622
Fibronectin 13.811.12.96.27.31.2X82402
Osteopontin1.84.32.46.88.61.3J04806
Biglycan5.110.12.08.44.20.5L20276
Transforming growth factor β10.40.72.00.20.32.0M13177
Integrin β10.40.61.61.02.12.2Y00769
Keratinocyte growth factor (FGF7)2.52.00.80.40.10.2Z22703
Decorin0.71.21.61.50.10.1X53929

In neonatal cultures, MCP-3 was the most overexpressed gene with more than 15-fold greater expression in neonatal Tsk1 fibroblasts, compared with non-Tsk1 fibroblasts, although sustained overexpression was not seen at later time points despite protein up-regulation in Tsk1 samples at later time points.

Western blot analysis of MCP-3 protein.

Transcript levels may not reflect changes in the gene product, since many genes are also regulated posttranscriptionally. Therefore, Western blotting of fibroblast culture supernatants was used to assess changes at the protein level. Up-regulation of MCP-3 was observed in 6-week-old Tsk1 samples in a series of 3 independent experiments (mean ± SEM 235 ± 26% compared with the non-Tsk1 controls; P = 0.002) (Figure 2B). Similar results were observed in parallel analyses of dermal fibroblast culture media derived from early-stage dcSSc skin compared with matched human controls. The level of MCP-3 immunoreactivity in SSc fibroblast supernatants (mean ± SEM relative density units [RDU] 4.92 ± 1.12 in controls versus 8.27 ± 0.5 in SSc; P = 0.02) was greater than that observed for MCP-1 (RDU 3.84 ± 0.73 versus 7.87 ± 0.14, respectively; P = 0.01), as shown in Figure 3.

Figure 3.

Overexpression of monocyte chemoattractant protein 1 (MCP-1) and MCP-3 by lesional systemic sclerosis (SSc) human skin fibroblasts. A, Fibroblast culture media from representative normal or SSc dermal fibroblast cultures were analyzed using antibodies to MCP-3 and MCP-1. Cell-layer β-actin staining was used as a loading control, and recombinant human MCP-1 and MCP-3 (rH) were used to confirm primary antibody specificity. B, Densitometric analysis summarizing 3 independent experiments shows the mean and SEM protein levels for MCP-3 and MCP-1 in relative density units (RDU).

Effect of MCP-3 on collagen reporter gene expression.

Analysis of dermal fibroblasts cultured from 2kb-LacZ transgenic mice suggested that transactivation of type I collagen by MCP-3 occurs in tissue culture. Using a highly physiologic bioassay, a series of independent experiments (n = 5) showed that MCP-3 increased the transactivation of the reporter transgene, with a maximal mean (±SEM) change above baseline levels of 80 ± 31% (P = 0.01). Similar up-regulation was observed with TGFβ1-treated cells (70 ± 24% above baseline) (Figure 4). Maximum induction of transgene expression was at 24 hours, with a threshold concentration that varied between experiments but was generally between 30 ng/ml and 250 ng/ml of MCP-3 and showed very similar dose-response characteristics to that previously observed for the related chemokine MCP-1 (19). Results from a series of duplicate experiments revealed that initial induction occurred at 6 hours with up-regulation of 34 ± 15%, which subsequently rose and then declined to 45 ± 22% above baseline at 48 hours.

Figure 4.

Activation of murine proα2(I) collagen (Col1a2) gene expression by monocyte chemoattractant protein 3 (MCP-3). A, Fibroblasts cultured from the skin of transgenic mice with high-level fibroblast-specific expression of β-galactosidase (2kb-LacZ) were used to examine MCP-3 activation of extracellular matrix gene expression. Values are the mean and SEM of triplicate samples and are representative of 5 independent experiments. ∗ = P < 0.05 as determined by Student's unpaired t-test. B, Transactivation of the murine Col1a2 gene in transient transfection is shown as reporter gene expression in 3 independent experiments for the minimal promoter construct. Activation is similar to maximum induction by recombinant transforming growth factor β1 (TGFβ) (10 ng/ml) at 24 hours. ∗ = P < 0.05; ∗∗ = P < 0.01, by Student's unpaired t-test. RLU = relative luminescence units.

Data from transgenic mice were confirmed using transient transfection of wild-type fibroblasts using the same 2kb-Col1α2-LacZ construct (see Figure 1 for details). As shown in Figure 4B, later experiments evaluating constructs harboring only a minimal Col1α2 promoter showed a similar dose-dependent up-regulation of reporter gene expression with a mean (±SEM) activation above baseline of 72 ± 51% and 85 ± 30% at 200 ng/ml and 400 ng/ml of MCP-3, respectively, from a series of 3 independent experiments (P = 0.02). These data suggest that upstream elements previously implicated in selective activation of reporter transgenes in Tsk1 mice are not essential for the effect of MCP-3.

There is substantial sequence conservation of both proximal and distal regulatory elements of the mouse and human proα2(I) collagen genes (23). We therefore confirmed our data on mouse constructs and also tested analogous human promoter–reporter constructs. Figure 5A summarizes the results from 3 independent experiments using a human minimal promoter construct, with maximal induction of collagen reporter gene expression of 52 ± 14% (mean ± SEM above basal expression) (P = 0.02). Further transient transfection experiments showed no greater induction with larger human 5-kb and murine 6-kb upstream enhancer constructs, suggesting that lineage-specific upstream enhancer sequences are not important for activation in transient transfection (data not shown).

Figure 5.

Activation of a human proα2(I) collagen (COL1A2) gene reporter by MCP-3 via a TGFβ-dependent mechanism. A, Up-regulation of the COL1A2-regulated reporter gene by MCP-3 is expressed as the mean and SEM percentage of basal expression from 3 independent experiments and corrected for transfection efficiency. B, Anti-TGFβ antibody significantly reduces MCP-3–induced activation of human COL1A2 promoter constructs at 24 hours. Data are representative of 3 independent experiments and expressed as the mean and SEM of triplicate samples. ∗ = P < 0.05 by Student's paired t-test. See Figure 4 for other definitions.

The time course for stimulation of collagen gene expression (between 16 and 24 hours) suggested that a second mediator may be elaborated by MCP-3–activated fibroblasts. Since TGFβ is a potent stimulator of ECM gene expression in vitro, the effect of anti-TGFβ antibody on this stimulation of collagen reporter gene expression was examined. The results show that MCP-3 stimulation of fibroblast collagen synthesis was partially inhibited in the presence of anti-TGFβ antibody. In addition, there was a dose-dependent response with the antibody, which displayed a maximal neutralization effect at 50 μg/ml, as shown in Figure 5B.

MCP-3 expression in SSc skin.

MCP-3 was detected abundantly in the dermal region in skin biopsy samples from Tsk1 mice at 10 days and 3 weeks old. Generalized perifollicular staining was observed using anti–MCP-3 antibody in both non-Tsk1 control and Tsk1 skin sections. Despite this chemokine overexpression, there was no demonstrable inflammatory infiltrate in the skin sections from Tsk1 mice (Figure 6) at all ages examined, from 3 days to 12 weeks.

Figure 6.

Up-regulation of monocyte chemoattractant protein 3 expression in type 1 tight-skin (Tsk1) mouse skin. Frozen skin sections from 3-week-old healthy non–tight-skin (pallid) mice (A and B) and Tsk1 mice at 3 weeks (C) and 10 days (D) old were immunostained. Healthy adult mice show perifollicular (p with arrow) staining, whereas there is additional dermal expression (d with arrow) in Tsk1 samples. (Original magnification × 240 in A and C; × 480 in B and D.)

In human skin samples, there was a variable amount of specific immunostaining for MCP-3 in the lower epidermal layer, in both healthy control and SSc biopsy samples. However, as with Tsk1 skin, SSc lesional skin sections showed dermal expression of MCP-3 as well as additional strong MCP-3 expression in and around the blood vessels at sites of mononuclear cell infiltrates (Figures 7A–C). The majority of the inflammatory mononuclear cells in scleroderma skin were positive for CD68 (Figure 7D). Consistent with the data on mRNA and protein expression, additional immunostaining studies using anti-AS02 suggested that MCP-3 chemokines localized in dermal cells express this fibroblast-specific marker well away from vascular or epithelial structures (data not shown).

Figure 7.

Increased monocyte chemoattractant protein 3 (MCP-3) expression in early-stage diffuse cutaneous systemic sclerosis (dcSSc) skin biopsy samples. In skin sections from patients with early dcSSc (A–D), vascular MCP-3 localization with dermal immunostaining is observed in A and B, respectively, and MCP-3 protein is detected in the inflammatory perivascular infiltrate (C); most of these samples are positive for the monocyte/macrophage marker CD68 (D). There is a variable amount of epidermal staining (e with arrow) in skin sections from patients with established dcSSc (E) and in clinically uninvolved early dcSSc skin (F) or limited cutaneous SSc (lcSSc) skin (G) and in skin from healthy controls (H). Minimal vascular MCP-3 immunostaining is detected in uninvolved skin from patients with established dcSSc (E) and lcSSc (G). (Original magnification × 240.)

There was significantly more MCP-3 expression in the skin of patients with early-onset diffuse disease compared with that from patients with established diffuse SSc or limited cutaneous SSc, suggesting that MCP-3 is associated with inflammatory or progressive skin disease. Negative controls with IgG isotype–matched antibodies showed no staining. Table 3 summarizes the staining patterns observed in human and mouse skin biopsy specimens. Immunostaining for MCP-3 was associated with an inflammatory infiltrate that showed vascular localization in 10 of the 14 skin sections from patients with early dcSSc as well as more diffuse dermal staining. Diffuse dermal staining was absent in the skin sections from patients with established dcSSc and lcSSc.

Table 3. Immunostaining patterns for human or mouse skin using specific anti–monocyte chemoattractant protein (MCP-3) antibody*
PatternHuman skinMouse skin
Control (n = 11)Early dcSSc (n = 14)Established dcSSc (n = 8)lcSSc (n = 6)Tsk1 (n = 5)Pallid (n = 5)
  • *

    Values are the no. of samples displaying the respective staining patterns for MCP-3. dcSSc = diffuse cutaneous systemic sclerosis; lcSSc = limited cutaneous systemic sclerosis; Tsk1 = heterozygous type 1 tight-skin mouse; pallid = homozygous pallid (non–tight-skin) mouse.

  • Defined by <2 years' duration.

Inflammatory infiltrate0122000
Vascular0102200
Diffuse dermal0100040
Perifollicular000045
Epidermal8146500

DISCUSSION

Identification of MCP-3 as a highly overexpressed gene in fibroblasts from neonatal Tsk1 mice raised the possibility that this chemokine might have a role in the pathogenesis of fibrosis in these mice. Because the related CC chemokine MCP-1 has been shown to be overexpressed in SSc fibroblasts, and since the Tsk1 strain is regarded as a model for human SSc, we extended our study to examine skin biopsy specimens and fibroblasts cultured from a series of SSc patients. We found overexpression of MCP-3 in SSc, and confirmed that it may operate as a profibrotic mediator with the potential to activate ECM gene expression, at least in part via induction of TGFβ1, in addition to promoting an inflammatory cellular response. Although multiple mediators are likely to be involved in the complex, multifaceted pathology of SSc, the identification of MCP-3 as a soluble factor capable of regulating 2 key aspects of pathology, leukocyte migration and ECM overproduction, is intriguing. It is plausible that early mediators such as MCP-3 might induce other factors such as TGFβ1, and perhaps other downstream candidates such as CTGF or platelet-derived growth factor are induced later (1). The ability of fibroblasts to secrete these factors supports a model of autocrine or paracrine local regulatory pathways in pathogenesis. Potential consecutive induction of fibroblast-derived mediators also emphasizes the importance of disease stage–specific approaches to targeted molecular therapies in SSc.

There are a number of approaches for determining broad differences in gene expression in different tissue or cell samples. Other studies have used a variety of technologies including differential-display reverse transcription–polymerase chain reaction, ribonuclease protection assay, and Northern blotting to examine gene-expression profiles in connective tissue diseases, including SSc. Candidate genes suggested by these approaches have included fibronectin, protease nexin 1, interleukin-1α, and CTGF (24, 25). However, MCP-3 was not identified in these earlier studies. This may reflect intrinsic differences in the SSc skin biopsy samples examined or in the fundamental properties of each of the methods used for examining differential gene expression, and emphasizes the value of using multiple approaches to address the same question.

Overexpression of chemokines and related receptor genes, including MCP-3, has been described in gene-cluster analysis of data from high-density gene-chip experiments studying whole-lung samples in the bleomycin-induced model of murine pulmonary fibrosis (26). Also, in a recent analysis by Luzina et al (27) of gene-expression profiles in bronchoalveolar lavage cells from SSc patients with lung inflammation, there was increased expression of chemokines and chemokine receptor genes associated with a greater risk for lung fibrosis. These studies suggest that cDNA microarray approaches are valuable as a screening approach to identify potential targets for further study. However, results must be interpreted with caution, and repetition of studies is important, including validation with independently prepared samples.

Homogenous populations of dermal fibroblasts from Tsk1 and wild-type mice were used to reduce the biologic variability, which may otherwise limit data analysis. However, even with this starting material, there was substantial variation when independently prepared samples were compared, presumably due to differences in cell cycle and other biologic parameters. Thus, many of the differentially expressed genes initially seen in neonatal fibroblasts were observed in only some littermate pairs. Replication of data is essential to reduce background differences in gene expression that are independent of fundamental differences in the cell lines, even in genetically homogeneous samples such as sex-matched littermates of inbred mouse lines. Only a few genes were consistently differentially expressed, and low-density microarrays such as those used for screening in this study have relatively few transcripts. More complete assessment using high-density gene chips may be more amenable to formal statistical and bioinformatic analyses.

In addition to prompting the present study of MCP-3, our gene-expression analysis of Tsk1 skin provides some additional information regarding the pathogenesis of the tight-skin mouse phenotype. Representing one of the best-characterized genetically determined animal models for SSc (28), the Tsk1 mouse model demonstrates cutaneous hyperplasia and abnormal connective tissue architecture in the dermis and visceral organs (29). Although heterozygous Tsk1 mice are normal at birth, in the second week of life, skin tightness develops in the interscapular region with increased numbers of high collagen-expressing fibroblasts. The histologic appearances of marked thickening of the dermis and excessive deposition of thick collagen fibers extending into the subdermal adipose tissue are clearly apparent from 3 weeks of age. Generalized dermal fibrosis with adherence to subcutaneous tissues develops by 12 weeks.

The genetic basis for the phenotype has recently been identified as an in-frame genomic duplication within the fibrillin 1 gene (30), resulting in a mutant transcript that is 3 kb larger than the wild-type transcript. The link between expression of a mutant fibrillin 1 protein and generalized matrix overproduction is uncertain. Our data indicate major differences in expression profiles neonatally, with up-regulation of MCP-3 and monocyte colony-stimulating factor 1, although some of these differences were not sustained at later time points. A smaller number of genes, mostly related to ECM turnover, were consistently up-regulated, including TIMP3 and TGFβ, at all time points. There was significant similarity with the gene-expression profile of TGFβ1-stimulated wild-type fibroblasts (data not shown). Array data suggest that MCP-3 is overexpressed neonatally in Tsk1 skin and not up-regulated at 12 weeks. Immunostaining confirmed this in mouse tissue. These findings are consistent with the more frequently observed MCP-3 overexpression in early dcSSc skin as compared with biopsy samples of skin with established disease. This would suggest that MCP-3 may be an important initiator in the cascade of mediators leading to dermal fibrosis. This is similar to the characteristics of other profibrotic mediators, including TGFβ1, in skin samples, which often show little expression in established lesional skin of SSc.

The significance of the differences in staining patterns for MCP-3 in human and mouse skin is unclear. Diffuse dermal expression was observed in both mouse and human skin samples. Similar perifollicular staining and a variable amount of epidermal staining was seen in both control and lesional samples from mice and humans, and thus these patterns are unlikely to be of pathogenic significance, although it is apparently specific and therefore probably reflects a physiologic function for MCP-3. The vascular and perivascular staining observed in early-stage dcSSc may reflect a role for fibroblast-derived MCP-3 in mononuclear cell extravasation, and the colocalization with CD68-positive cells supports this. It is possible that MCP-3 overexpression recruits macrophages into the SSc skin, leading to the initiation of skin fibrosis and production of other inflammatory or fibrotic cytokines, as outlined above. The absence of mononuclear cell infiltrates in Tsk1 skin despite overexpression of a potent chemoattractant is surprising. A possible explanation may be that increased metalloproteinase activity that has previously been demonstrated in Tsk1 mouse skin samples might proteolytically cleave MCP-3 and render it inactive. Analogous modification has been identified for human MCP-3 and has been proposed to be a mechanism by which acute inflammation is down-regulated during wound healing or scar formation (31).

There is growing evidence that chemokines may affect the homeostasis of ECM, and excessive synthesis of type I collagen by fibroblasts in the dermis is a pathologic hallmark of SSc. Given the lack of inflammatory infiltrate in Tsk1 skin, we studied the effect of MCP-3 on ECM biosynthesis. Our data showed increased proα2(I) collagen promoter activity in response to MCP-3 stimulation with similar dose-response characteristics to those previously observed for MCP-1 (19). Although relatively high concentrations of recombinant cytokine were needed for maximal effect, the threshold for activation was often an order of magnitude lower, and likely to reflect expression levels in the pericellular space in vivo (8). For the promoter assay, we initially used transgenic mouse fibroblasts containing a 2-kb upstream enhancer fragment driving high-level fibroblast-specific expression to reporter genes. Such use of transgenic cells represents a highly physiologic approach to studies of collagen-gene activation in vitro, since the transgene is stably integrated in chromatin and is known to reflect endogenous collagen-gene expression in vivo.

Reporter genes regulated by different murine and human promoter sequences were also examined in transient transfection studies with early-passage fibroblast cultures. Transactivation of Col1a2 in Tsk1 skin has previously been shown to depend on far-upstream fibroblast-specific elements as well as sequences within the proximal promoter (22), with both proximal and distal regulatory sequences appearing to be responsive to TGFβ. In contrast, the present study suggests that MCP-3–induced Col1a2 activation may be independent of the upstream enhancer. This raises the possibility that the profibrotic effects of MCP-3 are only partly mediated via TGFβ1. Collagen-gene up-regulation in Tsk1 skin is likely to occur via a number of pathways that may depend both on TGFβ1 and on other factors. Antibody neutralization experiments showed that the stimulatory effect of MCP-3 on collagen-gene expression is partly dependent on TGFβ. Consistent with this observation is the time course of maximal induction, which was between 16 and 24 hours, with little effect before 6 hours. It nevertheless remains possible that TGFβ may be a co-factor as well as a potential downstream mediator of collagen-gene activation, especially since collagen-gene activation was not completely abrogated by a high concentration of anti-TGFβ antibody. Conversely, there are recent reports suggesting that overexpression of the CC chemokine pulmonary and activation-regulated chemokine (known as PARC) may directly activate ECM gene expression (32).

In summary, our findings implicate MCP-3 as a potential mediator of dermal fibrosis in SSc, although several issues remain unresolved, especially concerning the interplay in vivo between MCP-3 and other profibrotic factors. It is possible that different mediators are important at various stages or in different clinical subsets of SSc. A microsatellite polymorphism in the promoter region of MCP-3 has been described in association with certain patterns of multiple sclerosis, another heterogeneous chronic inflammatory and sclerotic disease (33), and it is possible that polymorphic variants may associate with the clinical phenotype of SSc or other conditions such as inflammatory bowel disease (34). Further studies to better define the regulation and function of MCP-3 overexpression in SSc are currently in progress.

Acknowledgements

The 2kb-LacZ transgenic mice were generated in collaboration with Dr. Benoit de Crombrugghe's laboratory at the University of Texas M. D. Anderson Cancer Center in Houston.

Ancillary