A skin punch biopsy, 3 mm in diameter, was obtained from areas of affected skin from eight SSc patients at the Division of Rheumatology in Lund (Sweden) and from eight age-matched and sex-matched healthy control individuals. All SSc individuals fulfilled the American College of Rheumatology criteria for SSc 41 and had clinical features of early disease, and none was undergoing immunosuppressive therapy 42. Permission to perform this investigation was granted by the ethics committee. Informed consent was obtained from all individuals. Peripheral blood from healthy individuals was provided by the Blood Transfusion Center of the Geneva University Hospital (Switzerland).

Anti-CD3 OKT3 mAb and anti-human IL-4 mAb (clone 25D2) were from ATCC (Bethesda, MD, USA). Anti-human IFN-γ mAb was a gift from Dr G Garotta (Serono Biomedical Research Institute, Geneva, Switzerland). IgG₁ mAb (anti-TNP, clone G106HN) was a kind gift from S Izui (Department of Pathology and Immunology, Geneva School of Medicine, Geneva, Switzerland). Anti-CD40L (mAb 5c8) was a kind gift from P Lipski (University of Texas, Dallas, TX, USA). Human recombinant (r)IL-4 was from Schering Plough (Dardilly, France). Human rIL-2 was from Biogen (Cambridge, MA, USA). Human rIL-13 was from Sanofi (Montpellier, France). Human rIFN-γ was from Roussel Uclaf (Paris, France). Human recombinant transforming growth factor (rTGF)-β was from R&D (Minneapolis, MN, USA). Anti-TNF (recombinant-methionyl soluble TNF-type I pegylated receptor [sTNF-RI]) 43 and recombinant human IL-1 receptor antagonist 44 were from Amgen (Thousand Oaks, CA, USA). Recombinant human CD40L trimer/leucine zipper fusion protein was from Immunex (Seattle, WA, USA) 45. RPMI-1640, Dulbecco's modified Eagle medium (DMEM), glutamine, penicillin, streptomycin, trypsin, nonessential amino acids, sodium pyruvate and foetal calf serum (FCS) were from Gibco (Paisley, Scotland). Sucrose, phenyl methyl sulfonyl fluoride, pepstatin, EDTA, iodoacetamide, NP40 and indomethacin were from Sigma (St. Louis, MO, USA). Proteasome inhibitor I (PSI; Z-Ile-Glu [OtBu]-Ala-Leu-CHO), c-Jun NH₂-terminal kinase (JNK) inhibitor SP600125 (anthra [1,9-cd]pyrazol-6(2H)-one), and mitogen-activated protein kinase of extracellular signal-regulated kinase (MEK)1/2 inhibitor U-0126 (1,4-diamino-2,3-dicyano-1,4-bis [2-aminophenithiol]butadiene) were from Calbiochem (San Diego, CA, USA).

Prototypic Th1 and Th2 cell clones were generated from peripheral blood of normal individuals upon antigen activation and cloning by limiting dilution in RPMI-1640 medium supplemented with IL-2 (20 U/ml), penicillin (50 U/ml), streptomycin (50 μg/ml), 5% human AB serum, 10% FCS, and irradiated (3,500 rads) allogeneic peripheral blood mononuclear cells with phytohaemagglutinin (1 μg/ml) 46. Growing cells were further expanded and characterized with respect to their capacity to produce IFN-γ and IL-4 upon CD3 crosslinking. High IFN-γ/low IL-4 producers were defined as Th1, whereas low IFN-γ/high IL-4 producers were defined as Th2. In addition, we used SSc skin-derived polarized T cell clones (generation and characterization of which are described elsewhere 36). For the preparation of T cell membranes, T cells (8 × 10⁶) were activated in six-well trays, with plastic-adsorbed anti-CD3 mAb. Controls consisted of T cells cultured in medium alone. After six hours of culture the supernatants were collected and frozen for further cytokine determination, and cell membranes were prepared as described elsewhere 46.

Fibroblasts from skin biopsy were grown in DMEM medium supplemented with 10% FCS, penicillin, streptomycin, nonessential amino acids and sodium pyruvate, and split at confluence. All of the experiments were performed with fibroblasts at passages four to nine. In order to study chemokine production, we plated fibroblasts in 96-well trays at 2 × 10⁴ cells/well in 200 μl DMEM supplemented with 10% FCS, which were then cultured for 72 hours. The culture medium was then replaced by DMEM supplemented with 1% FCS. To assess the effect of T cell contact on chemokine production, 5 μl of T cell membranes equivalent to 2 × 10⁵ cells/well, unless otherwise stated, was added to fibroblasts in triplicate wells and cultured for 48 hours. In no instance were T cells syngeneic with fibroblasts. Supernatant was then harvested and frozen until chemokine determination. In blocking experiments anti-TNF (soluble TNF receptor I; 10^-8 mol/l), anti-IFN-γ (10 μg/ml) 40, IL-1 receptor antagonist (2 μg/ml) 44, anti-IL-4 (10 μg/ml) and irrelevant monoclonal IgG₁ (10 μg/ml), alone or in combination, were added to the wells 30 minutes before T cell membranes 4748. To evaluate the effect of cytokines on chemokine production by fibroblasts, TGF-β (10 ng/ml), IL-4 (10 ng/ml), IL-13 (10 ng/ml) and IFN-γ (1000 U/ml) were added to triplicate wells. When used in combination, TGF-β, IL-4 and IL-13 were used at 5 ng/ml each. In some experiments 20 × 10³ fibroblasts were cultured in the upper chamber of a semipermeable polyester membrane transwell and 10⁶ living T cells were cultured in the lower chamber, in which anti-CD3 mAb had been plastic adsorbed (Transwell #347-clear; Costar, Corning, NY, USA) for 48 hours. The total culture volume was 1 ml.

Fibroblasts (5 × 10⁵) were plated to confluence in 100 mm Petri dishes and serum starved overnight. T cell membranes to the equivalent of 8 × 10⁶ cells were then added to the cultures (typically 200 μl of cell membranes in 3 ml) and fibroblasts were cultured for four additional hours in 1% FCS medium. When used, intracellular signalling inhibitors (PSI [40 μmol/l], JNK-inhibitor [10, 20, and 50 μmol/l], U-0126 [40 μmol/l]) were added 1 hour before adding T cell membranes. Total fibroblast RNA was isolated with TRIzol™ reagent (Life Technologies, Invitrogen, Carlsbad, CA, USA). The levels of expression of IL-8, MCP-1, IP-10 and L-32 mRNAs were assessed by RiboQuant RPA using the hCK-5 multiprobe template set from Pharmingen (San Diego, CA, USA), in accordance with the supplier's instructions. After phosphor imaging (Typhoon 9400, Applied Byosistems, Foster City, CA, USA), signal intensity was determined by densitometry using ImageQuant software (Molecular Dynamics, NIH, Bethesda, MD, USA), and normalized values were used to determine the effect of T cells and signal inhibitors.

Levels of IL-8, MCP-1, IP-10, IL-4, IFN-γ and TNF-α (R&D Systems, Abingdon, UK), and IL-1α (Immunotech, Marseille, France) proteins were assessed using commercial enzyme-linked immunosorbent assay. T cell membranes were solubilized in 1% NP40 to detect cytokine content.

Student's t test was used with two-tailed P values or Mann–Whitney U test using StatView 5.0™ (SAS Institute Inc., Cary, NC) software on a Macintosh computer. P < 0.05 was considered statistically significant.

We first tested whether human dermal fibroblasts were able to produce IL-8, MCP-1 and IP-10 when activated in a contact-dependent manner by polarized T cell clones 36. To minimize the effect of soluble mediators, we used T cell membranes as effectors of contact-dependent activation. Upon T cell contact, dermal fibroblasts produced IL-8, MCP-1 and IP-10, with the production depending on whether T cells were resting or activated (CD3 crosslinking) and on the subset of Th cells (Figure 1a). To induce chemokine production, both Th1 and Th2 cells had to be activated because chemokine production was marginal in the presence of resting T cells (Figure 1a). Dose-dependent increases in IL-8, MCP-1 and IP-10 were observed in the presence of activated T cells. However, Th1 cells were less potent inducers of IL-8 and stronger inducers of IP-10 than were Th2 cells, although both subsets potently induced MCP-1 (Figure 1a).

We extended these findings by testing a large panel of polarized T cell clones at a fixed ratio of T cells to fibroblasts of 10:1. Under these settings, Th1 cells induced statistically significantly higher amounts of IP-10 than did Th2 cells, and Th2 induced higher amounts of IL-8 than did Th1 cells, with no differences being observed in MCP-1 levels (Figure 1b). Of note, the levels of IP-10 produced by fibroblasts were positively correlated (R₂ = 0.813) with the potential for IFN-γ production by T cell clones. No correlation was observed with the levels of the other chemokines, or were chemokine levels correlated with the levels of IL-4.

Finally, we tested whether differences existed between dermal fibroblasts generated from normal skin and those from SSc skin. Although no differences were observed in response to T cell contact between control and SSc fibroblasts, differences in the levels of cytokines were striking depending on the origin of T cells, with Th1 cells being more potent inducers of IP-10 and Th2 being stronger inducers of IL-8 (data not shown). These findings indicate that dermal fibroblasts upregulate their production of chemokines when activated by T cells and that the type of T cell dictates which chemokine is preferentially produced.

In order to further explore possible differences between SSc and control fibroblasts, we assessed their capacity to produce chemokines in response to prototypic Th cytokines or to profibrotic TGF-β (Figure 2). None of the cytokines tested induced IL-8, but all of them (IFN-γ, IL-4, IL-13, TGF-β and combinations of IL-4 plus TGF-β and IL-13 plus TGF-β) induced MCP-1 production. Of interest, IFN-γ induced significantly higher levels of MCP-1 than did IL-4 or IL-13 (P < 0.005) and IL-4 or IL-13 higher levels than did TGF-β (P < 0.005). However, MCP-1 was synergistically induced by TGF-β added together with IL-4 or IL-13. Predictably, IFN-γ was a potent stimulus whereas all of the other cytokines inhibited IP-10 production. It is noteworthy that, although SSc fibroblasts tended to produce higher amounts of chemokines, the only statistically significant difference from control fibroblasts was the higher IP-10 production in the presence of IFN-γ (Figure 2).

In order to identify some of the mediators present in T cell membranes that induce chemokine production by fibroblasts, we used neutralizing reagents to block the biological activity of several cytokines. IL-8 induction was dependent on IL-1 and TNF-α, which exhibited additive effects in both Th1 and Th2 cells (Figure 3a,b). Because Th2 cells were more potent inducers of IL-8 production than were Th1 cells, we tested the hypothesis that IFN-γ could act as a partial inhibitor. Indeed, after exogenous addition of IFN-γ to fibroblasts activated by Th2 cells, IL-8 production was inhibited by more than 50%, and this inhibition was abrogated by IFN-γ neutralization (Figure 3c). Similarly, IFN-γ neutralization resulted in enhanced IL-8 production when fibroblasts were activated by Th1 cells (Figure 3c). MCP-1 induction by activated Th1 and Th2 cells was dependent on the synergistic effect of IL-1 and TNF-α. With Th1 cells, further reduction in MCP-1 production was observed when IFN-γ was neutralized (Figure 3a). Unsurprisingly, IP-10 induction by Th1 cells was mostly dependent on IFN-γ and, to a small extent, on TNF-α and IL-1, whereas the very poor IP-10 production induced by Th2 cells was essentially due to TNF-α with a marginal contribution from IL-1.

We therefore attempted to verify whether IL-1α, TNF-α, IFN-γ and IL-4 could be detected in the T cell membranes of the clones used. This was indeed the case. IFN-γ was present in the membranes of all Th1 and in none but one of the Th2 cell membranes tested (Th1, n = 6; Th2, n = 6), IL-4 was present in none of the Th1 and all Th2 cell membranes tested, TNF-α was detectable in all membranes, and IL-1α was detectable in three out of six Th1 and in four out of six Th2 cell membranes tested (Figure 4). Thus, although IFN-γ and IL-4 were clearly mutually exclusive in polarized T cell membranes, this was not the case for TNF-α and IL-1α, which were present in both subsets.

It has been reported that fibroblasts express CD40 and that they may be activated via interaction with CD40L (CD154) 3. In our experimental conditions we could not observe any blocking effect of an anti-CD40L mAb when fibroblasts were cultured in the presence of T cell membranes whether from Th1 or Th2 clones, nor did we observe chemokine production by fibroblasts in response to recombinant human trimeric CD40L (not shown). We therefore tested whether fibroblasts would respond to cytokine produced by activated T cells in the absence of cell–cell contact. To this end, fibroblasts were cultured in the upper chamber of a semipermeable transwell, and living T cells were put in the lower chamber and activated by CD3 cross-linking. The results reinforce those observed when T cell membranes were used to activate fibroblasts (Figure 5). Fibroblasts exposed to cytokines produced by Th1 cells produced IL-8 and high levels of MCP-1 and IP-10, whereas those exposed to cytokines produced by Th2 cells produced high levels of IL-8 and MCP-1 but marginal amounts of IP-10. Furthermore, neutralization of TNF-α and IL-1, particularly when neutralizing reagents were used jointly, resulted in inhibition of induction of IL-8 and MCP-1 by Th1 and Th2 cells. In addition, neutralization of IFN-γ greatly enhanced IL-8 and abrogated IP-10 production induced by Th1 cells, with no effect on fibroblast responses induced by Th2 cells. Soluble TNF-α in the lower chamber, used as positive control, induced substantial amounts of IL-8 and MCP-1 but very little IP-10, and these effects were abrogated by its neutralization.

All together these findings indicate that TNF-α and IL-1, whether released in the supernatants or associated with T cell membranes, play a major role in inducing chemokine production by dermal fibroblasts. Differential effects observed with Th1 and Th2 cells are due to the role played by IFN-γ, which is produced only by Th1 cells, in that it inhibits IL-8 and stimulates IP-10 production.

Consistent with the data obtained when evaluating protein production, the steady state levels of IL-8, MCP-1 and IP-10 mRNA were upregulated in fibroblasts activated by T cell contact, as compared with resting fibroblasts (Figure 6a). The relative intensity of the bands we observed was dependent on the T cell subset used to activate fibroblasts. Thus, IL-8 mRNA levels were higher and IP-10 mRNA levels lower in the presence of Th2 cells than in the presence of Th1 cells, whereas MCP-1 mRNA levels were similar (Figure 6a). To further document the capacity of polarized T cells to induce distinct patterns of chemokine production in dermal fibroblasts, we used pharmacological inhibitors in a series of five experiments in which we explored some intracellular signalling pathways used in response to T cell contact.

IL-8 mRNA levels were differently affected by the inhibitors tested. The inhibitor of the MEK/ERK pathway U-0126 strongly inhibited IL-8 mRNA induced by both Th1 and Th2 cells (Figure 6a,b). Inhibition of nuclear factor-κB (NF-κB) pathway by PSI and of JNK by SP600125 resulted in a selective and dose dependent (JNK inhibitor) decrease in IL-8 mRNA levels induced by Th1 but not by Th2 cells (Figure 6a,b). Of further interest, MCP-1 mRNA induced by Th1 cells was not significantly affected by any of the inhibitors used, but the MCP-1 levels induced by Th2 cells were affected strongly by both U-0126 and PSI and weakly by the JNK inhibitor (Figure 6a,c). Finally, IP-10 mRNA levels induced by Th1 cells were decreased by U-0126 and PSI, and dose-dependently by JNK inhibitor. Overall, these findings indicate that T cell contact triggers the simultaneous activation on fibroblasts of several intracellular signalling pathways and that their role in regulating chemokine transcription may vary according to the particular chemokine and subset of polarized T cells.