1. IntroductionSystemic sclerosis (SSc) is a chronic disease characterized by skin and internal organ fibrosis, vasculopathy and autoimmunity [1,2]. Chemokine (C-X-C motif) ligand 4 (CXCL4) is an early SSc biomarker, which predicts a worse disease prognosis and later complications, such as pulmonary fibrosis [3,4,5]. CXCL4 binds DNA and RNA and favors interferon (IFN)-α production by plasmacytoid dendritic cells (pDC), contributing to the type I IFN (IFN-I) signature present in about half of the SSc patients [5,6]. Importantly, it has been shown that presence of an IFN-I signature early in the disease is often associated with more aggressive disease at later stages and complications [3,7,8,9,10]. Thus, both CXCL4 and IFN-I seem to play a role in early stages and be associated with future complications. Importantly, CXCL4 and IFN-I are closely linked via the process of triggering the nucleic sensor TLRs that leads to the IFN-I signature. In addition to that, we have previously shown that CXCL4 also acts as B-cell autoantigen in SSc patients [11,12]. This could be the result of formation of large CXCL4-DNA complexes in SSc patients, which could contribute to the immunogenicity of CXCL4. Indeed, we have previously shown that CXCL4-DNA complexes circulate in SSc patients, correlate with IFN-I and are also present in SSc-affected skin where they co-localize with an IFN-I induced gene [5]. Anti-CXCL4 antibodies were higher in patients with pulmonary fibrosis and digital ulcers and were associated with disease activity [11,12].However, it is well known that CXCL4 is a heparin-binding protein [13,14], and heparin-dependent anti-CXCL4 antibodies have been described in another rheumatic disease, systemic lupus erythematosus (SLE) [12].Most importantly, heparin-dependent antibodies play a pathogenic role in heparin-induced thrombocytopenia (HIT), a rare but fatal autoimmune condition, which leads to massive activation of platelets and thrombi formation with subsequent depletion of platelets [15]. Importantly, the activation of platelets is also characteristic of SSc; thus, whether heparin-dependent antibodies are more frequent in SSc than in the healthy population is of interest [16]. Anti-CXCL4 antibodies that we have previously described in SSc were likely heparin-independent [11,12].It remains elusive whether antibodies to CXCL4 in SSc can recognize complexes formed by CXCL4 and heparin, what their frequency is, and which role they play in disease pathogenesis. The present research is a pilot study to analyze the occurrence of heparin-dependent autoantibodies in SSc and their possible relationship with heparin-independent antibodies and disease manifestations. We found that heparin-dependent antibodies circulate in SSc and are mainly present in patients with pulmonary arterial hypertension (PAH), and, like the heparin-independent antibodies, correlate with digital ulcers (DU). These observations are of potential interest, as heparin-dependent antibodies are not routinely measured in SSc. We suggest that these antibody types are worth investigating in this chronic condition, characterized by vasculopathy and activation of platelets, in addition to autoimmunity and fibrosis [1,16]. 4. DiscussionIn this pilot study, by analyzing a SSc cohort, we have examined the expression of heparin-dependent anti-CXCL4 antibodies, also known as HIT antibodies, and their relationship with disease in SSc patients [13]. CXCL4 is a marker of SSc, which especially predicts SSc progression and complications [3]. In our previous studies [11,12], we only assessed the presence of antibodies directed to CXCL4, but not to CXCL4 and heparin. However, heparin is a CXCL4 binding protein, and anti-CXCL4 heparin-dependent antibodies could be also generated in SSc. They have been found in SLE patients and are characteristic of the rare autoimmune disease called heparin-induced thrombocytopenia (HIT) [13,14].SSc patients that do not have significant amounts of anti-CXCL4 antibodies usually have HIT antibodies, at least in our cohort. If this is a general phenomenon, it needs to be studied in larger cohorts. This evidence may mean that the two antibody types are probably generated via different mechanisms. We hypothesize that generation of CXCL4-specific antibodies, which are heparin independent, can be due to formation of complexes between CXCL4 and nucleic acids, as in SSc there is a high expression of CXCL4 and circulating and tissue-deposited CXCL4-DNA complexes [5]. This may favor breach of tolerance to CXCL4. It is likely that, in a very inflamed context and in the presence of cell death, CXCL4 easily binds nucleic acids released in high quantities by dead cells. Antibodies raised may transport self-DNA inside the immune cells (pDC, B-cells). This has been suggested in our previous paper, where the addition of an anti-CXCL4 antibody to pDC treated with CXCL4-DNA complexes implemented pDC activation [11]. Whether HIT antibodies cross-recognize anti-CXCL4-DNA complexes is unclear for the moment. However, if the HIT bind primarily CXCL4-heparin complexes, it is likely that they will favor uptake of heparin in pDC and other immune cells. Here we show not only that CXCL4-heparin complexes do not stimulate pDC, but also that heparin blocks the interferogenic activity of CXCL4-DNA complexes. This can give an explanation of why only heparin-independent anti-CXCL4 antibodies correlate with IFN-I. That heparin can antagonize the TLR9 stimulatory ability of CXCL4 and DNA had been already observed with B-cells in our previous paper [11]. We concentrated our attention on the capacity of both CXCL4-DNA and anti-CXCL4 antibodies to up-regulate IFN-α because immune complexes formed by CXCL4-DNA or CXCL4-DNA-antibody are likely to render self-DNA immunogenic and favor activation of pDC. We do not expect these immune complexes to stimulate (at least in pDC) other cytokines such as, for instance, TNF-α. Indeed, human DNA fails to induce significant production of TNF-α by pDC [18], even when in a complex with specific antimicrobial peptides that also bind, like CXCL4, the DNA, protecting it from degradation. This is also true for CXCL4, as shown here, where the complex with DNA of human origin only stimulate IFN-α but not TNF-α unlike artificial DNA mimics, such as CpG molecules, in our experiment CPGA. Indeed, we found no correlation between anti-CXCL4 antibodies and TNF-α, a cytokine also involved in SSc [19], and expressed in plasma of SSc patients, which reinforces the role of CXCL4 itself and anti-CXCL4 antibodies in participating, in particular, to sustain an IFN-I signature [11,12].On the other hand, we already published that CXCL4-DNA, but not CXCL4-heparin complexes, can directly stimulate memory B-cells to become antibody-secreting plasma cells [11,12]. The production of IFN-α, which can be at least partially ascribed to the presence of CXCL4-DNA (and perhaps CXCL4-RNA) complexes in SSc, can increase the capacity of B-cells to produce autoantibodies [11]. Thus, the relationship between anti-CXCL4 antibodies and IFN-α can be explained in an additional way: the creation of a loop in which CXCL4-DNA complexes stimulate pDC and IFN-α, the IFN-α stimulates B-cells to produce more antibodies to CXCL4, and the latter, in turn, further amplify IFN-α production.This picture reinforces the role of IFN-I in SSc, a cytokine that when present at disease onset is prognostic of a more severe disease at later stages [7,8,9].Regarding the apparent reciprocal expression of CXCL4 and HIT antibodies, it is interesting to report a mechanism that could also explain this mutually exclusive expression. Sachais et al. [20] showed that some HIT antibodies also bind to CXCL4 alone, but with much lower affinity than to CXCL4-heparin complexes. Antibodies with such a behavior were shown to induce clustering of CXCL4. In other words, they acted like heparin, inducing CXCL4 oligomerization, which increases immunogenicity. This action promotes the generation of further epitopes by cross-linking CXCL4 tetramers. Given the high likelihood of generating anti-CXCL4 antibodies in SSc, as CXCL4 is over-expressed, this mechanism could be frequently operative in SSc and favor the production of HIT antibodies.Both antibody types were instead found to correlate with DU. In this paper and in previous studies, we have shown that anti-CXCL4 antibodies correlated with DU [11,12]. We reasoned that this could be due to the fact that anti-CXCL4 antibodies favor IFN-α production, in that DU-positive patients usually show a higher IFN-α signature. However, HIT antibodies also seem to be frequent in DU-positive patients. In this case, the role of IFN-α could be important, but so could the presence of HIT. Two different mechanisms could promote DU: one linked to IFN-α, or at least to more general TLR-stimulation induced by CXCL4 itself and CXCL4-directed antibodies, and one mediated by the HIT antibodies, which are more linked to platelets’ activation and thrombi formation [13]. It is interesting that HIT antibodies were especially present in patients with PAH, which may likely be linked to the intrinsic pro-thrombotic effect mediated by the HIT antibodies.The limitations of this study are the small cohort studied and the fact that HIT antibodies should also be studied for their capacity to activate platelets in functional assays [21,22].

Nevertheless, we believe that HIT antibodies, which are increased in SSc as compared to HD, deserve more attention with respect to their clinical relevance as markers of specific disease features, and to prevent complications in SSc patients due to massive platelet activation.