Evaluation of membrane‐bound and soluble forms of human leucocyte antigen‐G in systemic sclerosis

Summary

Systemic sclerosis (SSc) is a complex disease characterized by immune dysregulation, extensive vascular damage and widespread fibrosis. Human leucocyte antigen‐G (HLA‐G) is a non‐classic class I major histocompatibility complex (MHC) molecule characterized by complex immunomodulating properties. HLA‐G is expressed on the membrane of different cell lineages in both physiological and pathological conditions. HLA‐G is also detectable in soluble form (sHLA‐G) deriving from the shedding of surface isoforms (sHLA‐G1) or the secretion of soluble isoforms (HLA‐G5). Several immunosuppressive functions have been attributed to both membrane‐bound and soluble HLA‐G molecules. The plasma levels of sHLA‐G were higher in SSc patients (444·27 ± 304·84 U/ml) compared to controls (16·74 ± 20·58 U/ml) (P < 0·0001). The plasma levels of transforming growth factor (TGF)‐β were higher in SSc patients (18 937 ± 15 217 pg/ml) compared to controls (11 099 ± 6081 pg/ml; P = 0·003), and a significant correlation was found between TGF‐β and the plasma levels of total sHLA‐G (r = 0·65; P < 0·01), sHLA‐G1 (r = 0·60; P = 0·003) and HLA‐G5 (r = 0·47; P = 0·02). The percentage of HLA‐G‐positive monocytes (0·98 ± 1·72), CD4⁺ (0·37 ± 0·68), CD8⁺ (2·05 ± 3·74) and CD4⁺CD8⁺ double‐positive cells (14·53 ± 16·88) was higher in SSc patients than in controls (0·11 ± 0·08, 0·01 ± 0·01, 0·01 ± 0·01 and 0·39 ± 0·40, respectively) (P < 0·0001). These data indicate that in SSc the secretion and/or shedding of soluble HLA‐G molecules and the membrane expression of HLA‐G by peripheral blood mononuclear cells (PBMC) is clearly elevated, suggesting an involvement of HLA‐G molecules in the immune dysregulation of SSc.

Introduction

Systemic sclerosis (SSc) is a connective tissue disease characterized by diffuse fibrosis and obliterative vascular lesions occurring in skin and internal organs 1. SSc is classified into limited (lSSc) and diffuse (dSSc) forms presenting different cutaneous and visceral involvement and specific autoantibody patterns 2. Although the causative factors remain to be characterized, three pathogenetic events underlie SSc development, namely vascular damage, immune dysregulation and fibroblast activation 3, 4, 5. In particular, endothelin‐1 (ET‐1) levels are increased, leading to activation of vascular endothelial cells, arteriolar hyperreactivity, myointimal proliferation and vascular occlusion 6, whereas vascular epidermal growth factor (VEGF) levels are decreased, contributing to the impaired vascularization of digital arteries 7. As far as the immune system is concerned, autoantibodies 8, high levels of B cell activating factor (BAFF) 9 and perturbations of dendritic cells 10 have been reported. Furthermore, alterations of the normal functional balance between proinflammatory subpopulations, in particular T helper type 17 (Th17) and regulatory T cell (T_reg) subpopulations, including both CD4⁺ and CD8⁺ T_reg subsets, have been demonstrated in patients affected by SSc 11, 12, 13. Of note, double‐positive CD4/CD8 T lymphocytes have been described in dermal lesions of patients with SSc; these cells secrete elevated amounts of interleukin (IL)‐4, which has profibrotic properties 14. Finally, connective tissue growth factor (CTGF), platelet‐derived growth factor (PDGF) and transforming growth factor (TGF)‐β levels are increased in SSc, thus favouring fibroblast activation and extracellular matrix synthesis 1, 15, 16.

Human leucocyte antigen (HLA)‐G is a human non‐classical major histocompatibility complex (MHC) molecule expressed mainly in membrane‐bound form at the fetal–maternal interface on the extravillous cytotrophoblast 17 and in placental tissue 18, where it contributes to the development of maternal tolerance to the semi‐allogeneic fetus 19, 20. It is also expressed physiologically in a few adult tissues, such as thymus 21, cornea 22, pancreas 23 and bronchial epithelial cells 24, as well as in different cell types such as activated monocytes and erythroid and endothelial precursors 25. The expression of HLA‐G antigens has been reported recently in some solid tumours, transplanted organs and cutaneous inflammatory diseases as well as on virally infected cells 26, 27. HLA‐G is also detectable in soluble form (sHLA‐G) in several body fluids, deriving from the secretion of soluble isoforms (HLA‐G5) and/or the shedding of proteolytically cleaved surface isoforms (sHLA‐G1) 28, 29, 30. Elevated levels of sHLA‐G molecules have been detected in plasma of patients affected by various pathological conditions 31, 32, 33, 34, 35, as well as in cerebrospinal fluid of patients with multiple sclerosis 36.

Several immune functions have been attributed to both membrane‐bound and soluble HLA‐G molecules 37, 38. It has been proposed recently that HLA‐G should be qualified as an ‘immune checkpoint’ molecule 39. Most functions of HLA‐G molecules are immunosuppressive, as they inhibit the cytolytic function of natural killer (NK) cells and CD8⁺ T lymphocytes 40, 41, the alloproliferative response of CD4⁺ T cells 42, maturation of dendritic cells 43 and activation of B cells 44. In addition, HLA‐G molecules are able to trigger apoptosis in antigen‐specific CD8⁺ T lymphocytes 41, 45, 46. HLA‐G also seems to be involved in the tuning of immune responses, as in‐vitro studies indicate that incubation of peripheral blood mononuclear cells (PBMC) with HLA‐G‐expressing cells favour a shift towards a Th‐2 cytokine profile, whereas incubation with sHLA‐G protein may have a counterbalancing effect by creating an anti‐inflammatory environment due to the release of IL‐10 47, 48. Finally, HLA‐G‐positive T_regs have been detected in peripheral blood and inflamed tissues 49, 50, 51, 52.

HLA‐G molecules have been detected in approximately 50% of skin biopsies from SSc patients, and their expression has been associated with a better clinical outcome 53. Furthermore, in a recently published paper it has been reported that the serum levels of sHLA‐G are lower in SSc patients than in normal subjects, although the difference was not statistically significant, and that, within SSc patients, the lowest HLA‐G levels were associated with a more severe or active disease 54.

The aims of the present study were: (i) to determine the plasma levels of sHLA‐G molecules in a cohort of SSc patients with the limited or diffuse form of the disease; (ii) to correlate sHLA‐G levels with TGF‐β; and (iii) to evaluate the expression of HLA‐G in PBMC.

Materials and methods

Patients

Thirty‐five patients affected by SSc (28 females and seven males, aged 40–89 years) followed at the Clinical Immunology Unit (Department of Internal Medicine, University of Genoa, Genoa, Italy) were included in the present study. The patients satisfied the European League Against Rheumatism/American College of Rheumatology (EULAR/ACR) 2013 classification criteria for SSc 55. Patients were subclassified as having lSSc (23 subjects) or dSSc (12 subjects), according to the LeRoy criteria 2. Disease duration was determined from the onset of the first non‐Raynaud's manifestation. Forty healthy subjects, matched for sex and age, were recruited as controls. This study was approved by the local Ethical Committee and all patients signed a written informed consent. The study was conducted in accordance with the principles of the Declaration of Helsinki

Plasma samples

Plasma from both SSc patients and healthy donors was obtained by centrifugation (1125 g for 15 min at room temperature) from heparin tubes and stored at −80°C until use. Plasma was preferred to serum, as the amount of soluble HLA‐G molecules is higher in plasma than in serum 56.

sHLA‐G determination

The determination of sHLA‐G molecules was performed by sandwich immunoenzymatic assays. A commercially available assay (Exbio, Vestec, Czech Republic), which employs the MEM‐G9 monoclonal antibody (mAb) as capture antibody, was utilized to determine the total sHLA‐G amount. In order to determine the specific amount of sHLA‐G1 or HLA‐G5, immunoenzymatic assays were performed using the 01G mAb (Exbio) or the 5A6G7 mAb (Exbio) as capture antibodies, respectively. The HRP‐conjugated B2M mAb (Exbio), which recognizes β2‐microglobulin, was employed as detection antibody in all assays. Plates were read with an ELX800 ELISA reader (BioTek Instruments, Inc., Winooski, VT, USA) and results were expressed as U/ml.

TGF‐β determination

The determination of TGF‐β was performed by immunoenzymatic assay utilizing a commercially available kit (Bender MedSystems, Vienna, Austria) and results were expressed as pg/ml.

Flow cytometric analysis

The expression of cell membrane antigens by peripheral blood mononuclear cells (PBMC) was analysed by direct immunofluorescence incubating 100 µl of peripheral blood from each individual with the fluorochrome‐conjugated anti‐HLA‐G mAb MEM‐G9 (Exbio), which reacts with the native form of HLA‐G1, and with the fluorochrome‐conjugated anti‐CD3, ‐CD4, ‐CD8, ‐CD14 and ‐CD45 mAbs (Beckman Coulter Europe, Cassina de'Pecchi, Italy) at 4°C for 30 min in the dark. Fluorochrome‐conjugated isotype matched antibodies were used as controls. After red blood cell lysis, analysis was performed by flow cytometry using a Navios flow cytometer equipped with kaluza software (Beckman Coulter Europe).

Statistical analysis

Values are expressed as mean ± standard deviation. Comparisons between sHLA‐G plasma levels in SSc patients and controls were performed by unpaired Student's t‐test. Correlations between sHLA‐G plasma levels and TFG‐β levels were performed by Pearson's correlation coefficient. Comparisons between the percentages of different PBMC subpopulations were performed by Mann–Whitney U‐test. A P‐value < 0·05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism version 6.03 for Windows (GraphPad Software, San Diego, CA, USA).

Results

sHLA‐G and TGF‐β plasma levels

The plasma levels of total sHLA‐G molecules were significantly higher in SSc patients (444·27 ± 304·84 U/ml) compared to controls (16·74 ± 20·58 U/ml; P < 0·0001) (Fig. 1a). In SSc patients, plasma levels of sHLA‐G1 and HLA‐G5 isoforms were comparable (264·66 ± 226·95 U/ml and 181·44 ± 130·12 U/ml, respectively) (Fig. 1b). In addition, no significant differences were detected in total sHLA‐G, sHLA‐G1 and HLA‐G5 plasma levels between dSSc and lSSc patients (data not shown). As expected, the levels of sHLA‐G1 and HLA‐G5 correlated with those of total sHLA‐G (r = 0·92; P < 0·0001 and r = 0·72; P < 0·0001, respectively) (Fig. 1c,d). No correlation was found between the level of sHLA‐G molecules and the disease features of each single patient (e.g. lung function tests, pulmonary artery pressure, autoantibodies pattern, nailfold videocapillaroscopy pattern, disease's form and duration). The plasma levels of TGF‐β were significantly higher in SSc patients (18 937 ± 15 217 pg/ml) compared to controls (11 099 ± 6081 pg/ml; P = 0·003) (Fig. 2a). A significant correlation was found between the plasma levels of TGF‐β and the plasma levels of total sHLA‐G (r = 0·65; P < 0·01), sHLA‐G1 (r = 0·60; P = 0·003) and HLA‐G5 (r = 0·47; P = 0·02) (Fig. 2b–d).

Figure 1.

Plasma levels of soluble human leucocyte antigen‐G (sHLA‐G) molecules. (a) Total sHLA‐G concentrations in systemic sclerosis (SSc) patients and healthy donors. (b) HLA‐G1 and sHLA‐G5 concentrations in SSc patients. (c,d) Correlations between total sHLA‐G concentration and sHLA‐G1 and sHLA‐G5 plasma levels in SSc patients.

Figure 2.

Transforming growth factor (TGF)‐β plasma concentration and correlation with soluble human leucocyte antigen‐G (sHLA‐G) levels. (a) TGF‐β levels in systemic sclerosis (SSc) patients and healthy controls. (b,c,d) Correlations between TGF‐β concentrations and total sHLA‐G, sHLA‐G1 and sHLA‐G5 plasma levels in SSc patients.

Determination of HLA‐G expression in PBMC

The percentage of HLA‐G‐positive monocytes was significantly higher in SSc patients (0·98 ± 1·72) than in controls (0·11 ± 0·08; P < 0·0001). The percentage of CD4⁺ and CD8⁺ cells expressing HLA‐G molecules was significantly higher in PBMC from SSc patients (0·37 ± 0·68 and 2·05 ± 3·74, respectively) compared to controls (0·01 ± 0·01; P < 0·0001 and 0·01 ± 0·01; P < 0·0001, respectively). The frequencies of T lymphocytes subpopulations were comparable between controls and SSc patients (the mean of CD3⁺CD4⁺ lymphocytes was 70·5 and 71·3% and the mean of CD3⁺CD8⁺ lymphocytes was 26·5 and 27·4% in controls and SSc patients, respectively). The mean of monocytes (CD45⁺CD14⁺) was 8·2 and 5·4% in controls and SSc patients, respectively. Moreover, SSc patients displayed a higher, although not significantly different, percentage of CD4⁺CD8⁺ double‐positive (DP) cells (4·95 ± 8·31) in comparison with controls (2·45 ± 1·79). In the context of DP cells, SSc patients showed a significantly higher percentage of HLA‐G⁺ cells compared to controls (14·53 ± 16·88 and 0·39 ± 0·40, respectively; P < 0·0002). Furthermore, among DP cells a subpopulation of CD4^dullCD8^high cells was detectable in both SSc patients and controls (1·43 ± 3·58 and 1·27 ± 0·65, respectively). Of interest, CD4^dullCD8^high cells from SSc patients presented a high percentage of HLA‐G⁺ cells (30·47 ± 26·75) whereas HLA‐G expression was virtually absent in the same subpopulation from healthy controls. If the limited and diffuse forms of disease were compared, no significant differences were observable in the percentage of HLA‐G‐positive cells among DP and CD4^dullCD8^high populations. Representative examples of HLA‐G expression by monocytes, CD4⁺, CD8⁺, DP and CD4^dullCD8^high cells from a SSc patient and a healthy control are shown in Figs 3 and 4. Of interest, a significant correlation was found between sHLA‐G plasma levels and the percentage of HLA‐G expressing monocytes (r = 0·4, P < 0·01).

Figure 3.

Surface expression of human leucocyte antigen‐G (HLA‐G) molecule on peripheral blood mononuclear cells from systemic sclerosis (SSc) patients and healthy controls.

Figure 4.

Surface expression of human leucocyte antigen‐G (HLA‐G) molecule on CD4^highCD8^dull and CD4^dullCD8^high subpopulations from systemic sclerosis (SSc) patients and healthy controls.

Discussion

In the present investigation we have shown that the plasma levels of sHLA‐G molecules were significantly higher in SSc patients than in healthy subjects. Both sHLA‐G1 and HLA‐G5 isoforms accounted for this increase, and their levels correlated with those of total sHLA‐G. The plasma levels of total sHLA‐G, sHLA‐G1 and HLA‐G5 molecules did not differ significantly between patients affected by the diffuse or the localized form of disease. Moreover, no correlation was found between clinical characteristics (e.g. lung function tests, pulmonary artery pressure, autoantibody pattern, nailfold videocapillaroscopy pattern, disease form and duration) and the level of sHLA‐G molecules. Accordingly, we argue that the increase of sHLA‐G levels could be considered an intrinsic characteristic of the disease itself that is not related to specific clinical features of the single patient. Nevertheless, we cannot exclude that the lack of difference observed could be explained by the low number of subjects assessed. These findings are in disagreement with those published recently by other authors who did not find significant differences in sHLA‐G molecule levels between SSc patients and healthy donors 54. A possible explanation of this discrepancy might be the difference in the immunoenzymatic methods utilized by each research group for the quantitative determination of sHLA‐G molecules.

In the present study we have also shown that the plasma levels of TGF‐β were significantly higher in SSc patients than in controls and correlated with those of soluble HLA‐G molecules. This finding is in keeping with the results of in‐vitro experiments in which the production of TGF‐β1 by myelomonocytic cells was increased strongly after incubation with recombinant soluble HLA‐G molecules 57. Moreover, this observation supports the hypothesis that sHLA‐G may contribute to the immunopathogenesis of systemic sclerosis because TGF‐β has been identified as a major activator of fibroblasts in both mouse models and human disease 58, 59, 60, 61.

Further experiments have been performed in order to analyse the membrane expression of HLA‐G in PBMC from SSc patients. First, we have demonstrated that the percentage of HLA‐G positive monocytes is increased in SSc patients. This finding is in agreement with published data reporting that mononuclear phagocytes express HLA‐G mRNA and protein and that interferon (IFN)‐γ enhances HLA‐G expression selectively, which could influence inflammatory responses 62. Of note, the percentage of HLA‐G‐positive monocytes correlated significantly with the plasma levels of sHLA‐G molecules. Secondly, we found that the percentage of CD4⁺ cells and CD8⁺ cells expressing HLA‐G molecules was significantly higher in SSc patients than in controls. These CD4⁺HLA‐G⁺ and CD8⁺HLA‐G⁺ cells, that are different from the naturally occurring CD25⁺forkhead box protein 3 (FoxP3⁺) T_regs, may be involved in the modulation of immune responses in systemic sclerosis, as it has been reported that these HLA‐G‐expressing lymphocytes exert potent suppressive function and are present in inflamed tissues of patients affected by immune mediated disorders 50, 63, 64. Interestingly, HLA‐G molecules were detected in epidermal and dermal cells of skin biopsies from approximately 50% of patients with SSc, and HLA‐G expression was associated with a lower frequency of cutaneous ulcers, telangiectasias and polyarthralgias 53. Thirdly, we detected a significant increase in CD4⁺CD8⁺ DP cells in patients affected by SSc compared to healthy subjects. The function of circulating CD4⁺CD8⁺ DP T cells remains controversial, with conflicting reports describing cytotoxic activity in viral infections or immune suppressive roles in inflammatory disorders 65. Of note, CD4⁺CD8⁺ DP cells are present in the skin of patients with early active SSc and may contribute to the enhanced extracellular matrix deposition by fibroblasts through the production of high levels of IL‐4 14. Furthermore, CD4⁺CD8⁺ DP cells are found in other autoimmune diseases, such as systemic lupus erythematosus, where they exert a suppressive role in the production of autoantibodies 66. Finally, a subpopulation of CD4^dullCD8^high cells expressing HLA‐G‐positive cells was detected among CD4⁺CD8⁺ DP cells in SSc patients but not in healthy controls.

Collectively, these data indicate clearly that in SSc the secretion and/or shedding of soluble HLA‐G molecules and the membrane expression of HLA‐G by PBMC is elevated. A possible involvement of HLA‐G molecules in SSc pathogenesis might be suggested by the correlation between sHLA‐G and TFG‐β plasma levels leading to fibroblast activation and fibrosis development. Moreover, as the immunosuppressive role of HLA‐G is demonstrated widely 40, 41, 42, 43, 44, 45, 46, we suggest that the increased release of soluble HLA‐G forms and the up‐regulation of HLA‐G membrane expression by PBMC may reflect an attempt to control the immune derangement occurring in systemic sclerosis. This hypothesis is supported by previously published data indicating that the expression of HLA‐G in the skin is associated with a better disease prognosis 53, whereas low sHLA‐G serum levels associate with a worse clinical course 54. However, available data are scanty and sometimes conflicting, therefore further perspective basic and clinical investigations are required in order to define more clearly the modulatory role, if any, of HLA‐G molecules in SSc pathogenesis and clinical course.

Disclosure

The authors have no conflicting financial interests.