Introduction

Systemic lupus erythematosus (SLE) is a complex autoimmune disease affecting multiple organs, characterized by a wide spectrum of clinical manifestations and laboratory findings. Multiple lines of evidence support a strong genetic contribution to the development of the disease, and it is well accepted that SLE occurs in genetically predisposed individuals exposed to certain environmental stimuli. Evidence of familial clustering was the first indication of a genetic susceptibility to SLE. On the basis of twin studies, SLE heritability (that is the relative contribution of genetic variation to the liability of developing the disease) has been estimated to be about 66 % based on the higher monozygotic twins concordance rates (24–56 %) compared to dizygotic twins (2–5 %) [1]. Familial aggregation for SLE, measured by the sibling recurrent risk ratio, varies from 8 to 29 depending upon the disease prevalence in the population used as reference [2].

Several linkage studies and genome-wide association studies have highlighted the impact of genetic polymorphisms on the risk of developing the disease [3–9]. The strong contribution of HLA region is widely known; the risk conferred by DRB1*1501 (HLA-DR2) and DRB1*0301 (HLA-DR3) genes is confirmed in many European populations. Outside the MHC region, IRF5 (interferon regulatory factor 5) is one of the most strongly and consistently SLE-associated loci; STAT4 (signal transducer and activator of transcription 4) has been found to associate with SLE in multiple studies in European or Asian populations [10, 11].

Only a few studies have addressed the influence of disease-predisposing genes on SLE severity and outcome. Fc Rs I, II, and III have been consistently associated with both susceptibility and severity of SLE. In the case of FcgammaRIIa and Fc gammaRIIIa, the low affinity allele is predisposing not only to SLE but also to lupus nephritis [12, 13], while homozygosity for the valine allele of Fc gammaRIIIa is a risk factor for the progression of renal involvement to end-stage renal disease [14]. On the whole, only limited information is available on the association of risk genes for SLE with specific disease phenotypes.

The GAPAID (Genes And Proteins for AutoImmunity Diagnostics) consortium was created within the European Union’s Seventh Framework Programme for Research and Technological Development (FP7), with the aim of developing a novel diagnostic/prognostic platform for patients affected by SLE, based on a genetic array, a serological protein array, and a software combining the clinical data with the genetic and serological information. In this context, the present study describes the genetic study performed on susceptibility to SLE, where 48 single nucleotide polymorphisms (SNP) from 40 different loci have been tested for SLE susceptibility and their association with disease subphenotypes.

Materials and methods

Ethic statement

This study was approved by the Ethics Committee of the University Hospital of Pisa (reference number: 45066/2012) and the Hungarian Scientific and Research Ethics Board (reference number: 24973-1/2012 EKU). The procedures followed were in accordance with the Helsinki Declaration of 1975. All the patients gave written informed consent.

SLE case-control population

A cohort of 208 SLE patients (cases) and 152 healthy blood donors (controls) was recruited between August 2012 and October 2013 from two centers, the Clinical Immunology Unit, Department of Clinical and Experimental Medicine of the University of Pisa (Italy) and the Department of Rheumatology and Immunology of the University of Pécs (Hungary) (Table 1). All SLE patients fulfilled the American College of Rheumatology (ACR) classification [15]. Gender and age data from all individuals were collected, as well as several clinical data retrospectively evaluated from SLE patients such as the age of patients at the onset of the disease, organ involvement, and secondary antiphospholipid syndrome (APS) (Table 1). The occurrence of arthritis (erosive or not) in clinical history was scored. Renal involvement was diagnosed on the basis of proteinuria, hematuria, and/or creatinine increase and/or hypertension and was in most cases (more than 80 % pts) confirmed by kidney biopsy. Hematological involvement was diagnosed in the presence of thrombocytopenia and/or hemolytic anemia and/or leucopenia. Neurological involvement was diagnosed in the presence of neuropsychiatric manifestations such as seizures, psychoses, cerebrovascular disease, and myelopathy. Skin involvement included generalized or malar rash, discoid lesions, and cutaneous vasculitis.

Table 1 Clinical data of individuals included in the study
Full size table

SNP selection and genotyping process

A total of 48 SNPs from 42 loci previously associated with SLE, in European ancestry populations, have been included in GAPAID project (Table 2). Genetic markers were principally selected from candidate gene, genome wide and replication association studies published at the time of the project. Some of the previously reported subphenotype-specific associations regarding those clinical features evaluated in the present study were also considered. The final list included several well known susceptibility loci such as 1q25.1, PTPN22, TNFSF4, STAT4, PXK, BANK1, HLA, TNFAIP3, IRF5, BLK, IRF8, or ITGAM, along with some less studied or controversial ones (e.g., IL10, LY9, IL21, LYN, TRAF6, ICAM region).

Table 2 Selected SNPs
Full size table

DNA from the buffy coat of all collected samples was purified by NucleoSpin 96 Blood Core Kit Macherey-Nagel). DNA quantity (ng/ul) and quality (260/280 and 260/230 absorbances) were checked with Qubit fluorometer and NanoDrop 8000 Spectrophotometer, respectively, before the genotyping process. Genotyping of selected SNP was performed by BioMark™ HD System (Fluidigm), based on the 5’ exonuclease activity of the polymerase. For each array, 2 negative controls and 46 unknown samples were included. Fluidigm SNP Genotyping Analysis Software v.3 was used for allele assignation.

Before statistical analyses, three quality criteria were checked with PLINK v.2.050 software [34]: SNP call rate (min. 95 %), sample call rate (min. 95 %), and conformity of genotype proportions to Hardy-Weinberg equilibrium (HWE) in the overall population.

Statistical analyses

Regression analyses were performed with the abovementioned software with the aim of detecting SLE susceptibility loci (case-control analysis) or specific polymorphisms for any of the measured subphenotypes (case-case analyses). All analyses were carried out under additive, dominant, and recessive genetic models. As differences in age and gender distribution between analyzed populations could limit the results of the study, both factors were included as covariates in all regression analyses. In the same manner, the origin of individuals was also considered as covariate in order to control the possible effect of genetic ancestry in the study. Finally, the results from the case-case analyses for renal, central nervous system (CNS), and skin involvement and presence of arthritis or secondary APS were also adjusted for the age of patients at the onset of the disease. The genetic model with the best P value has been chosen in each case. Corrections for multiple testing implemented in PLINK v.2.050 software have been performed and significant associations were considered when adjusted P values < 0.05.

Results

After the genotyping process, SNPs rs396991 (FCGR3A), rs231775 (CTLA4), rs11568821 (PDCD1), rs6568431 (PRMD1/ATG5), and rs4963128 (PHRF1) were removed for subsequent statistical analyses due to their low call rate (<95 %). The same threshold was applied to remove 9 individuals (5 cases and 4 controls). In addition, SNP rs2187668 (HLA-DQA1) did not fit HWE in the overall population (P value < 0.001). Thus, a total of 42 SNPs and 351 individuals (203 cases and 148 controls) were included in the final case-control and case-case analyses.

Results from the regression analysis focused on the identification of SLE susceptibility loci (case-control analysis) are shown in Table 3. Two SNPs located in the HLA region (rs3131379 and rs1270942) appear significantly associated with the disease (adjusted P value < 0.05). Five additional SNPs in IRF5 (rs729302 and rs2070197), BLK (rs2736340), IRF8 (rs4843869), and ITGAM (rs1143679) loci show also a trend to be related with SLE susceptibility (nominal P value < 0.05).

Table 3 Results from regression analyses on the overall SLE case-control population
Full size table

Two significant subphenotype-specific associations (adjusted P value < 0.05) have been found between SNP rs5754217 (UBE2L3) and skin involvement, and SNP rs3093030 (ICAM1-ICAM4-ICAM5) and hematological manifestations (Table 4). Other 32 suggestive associations have also been observed among all the analyzed clinical features (nominal P value < 0.05).

Table 4 Results from regression analyses on the different subphenotypes
Full size table

Discussion

SLE is an autoimmune disease affecting predominantly women characterized by a loss of tolerance to self-antigens, inflammation, and dysregulated immune responses leading to multi-organ damage [35]. Epidemiological studies suggest a strong contribution of genetic factors in the pathogenesis of the disease; many SLE risk loci have been identified in several genome-wide association studies (GWASs) and other association studies in the last decades [36]. In the present study, the strongest association has been detected for HLA region, concordantly with previous GWASs [4, 9]. Encoding more than 200 genes and subdivided into class I, II, and III regions, HLA complex was the first SLE susceptibility locus identified [37]. The two SNPs significantly associated with SLE in this study, rs3131379 and rs1270942, are located in MSH5 and CFB immune genes, respectively, in the class III HLA region. The first one, previously reported as an SLE-associated gene in UK families and also in a study on SLE patients with African American background [38, 39], has recently also been related with cutaneous SLE in European populations [40].

The results found for IRF5, BLK, IRF8, and ITGAM could be also highlighted. Although the association level between these loci and SLE susceptibility does not reach the statistical significance in the present study, probably due to a limited sample size, all of them are well-known SLE susceptibility genes. The three SNPs from IRF5 analyzed here (rs729302, rs2070197, and rs10954213) represent three haplotype blocks with an already described independent effect on SLE risk [41]. Regarding the related IRF8, Cunninghame Graham et al. [19] identified for the first time this gene associated with the disease in a European population, which was later robustly established with the independent effect of the three SNPs analyzed in the present study [8]. There are several studies reporting also the implication of BLK locus in SLE susceptibility, which encodes a tyrosine protein kinase involved in the proliferation, differentiation, and tolerance of B cells. The T allele from SNP rs2736340 has been associated with a major risk of the disease, as detected in the present study [42, 43]. Finally, the non-synonymous variant from ITGAM associated in the present study (rs1143679, R77H) has been proposed as one of the causal variants in this locus affecting the numerous ligand binding activities of ITGAM in monocytes, neutrophils, and dendritic cells and impairing C3-mediated phagocytosis [44, 45].

The analysis of SLE subphenotypes performed in the present study reveals two major findings. The association between UBE2L3 gene and skin involvement is the strongest one. UBE2L3, encoding a ubiquitin conjugating enzyme, was suggested to be a SLE susceptibility locus for European populations by Harley et al. [4] and later confirmed in a large-scale replication study [7]. Both studies showed positive results for the intronic SNP rs5754217 which has been included in a SLE risk haplotype leading to a higher UBE2L3 mRNA expression [46]. Furthermore, T allele of rs5754217 has been associated with the presence of several autoantibodies in African-American and European SLE patients [47, 48], but this is the first study reporting its implication in organ involvement during the disease; this T allele appears to be associated with a higher risk of skin manifestations. The second novel association in the present study is that reported between ICAM1-ICAM4-ICAM5 and hematological disorders. The ICAM locus, encoding intercellular adhesion molecule proteins that are expressed in vascular endothelium, macrophages, lymphocytes, red blood cells, and brain, confers susceptibility to SLE in multiple ancestry populations [31], but no association with disease subphenotypes has been so far reported.

Other several suggestive associations have been observed in this study near to the statistical significance. XKR6 gene, encoding a transmembrane protein of the Kell blood group of antigens and related only to SLE nephritis up to now [28], appears to be linked with CNS involvement. On the contrary, the implication of IL21 with hematological disorders has already been described in Europeans [20]. Other subphenotype-specific tendencies observed here match also previously reported results. Furthermore, the present paper is the first one suggesting some of these associations in European SLE patients, such as the relation between IKZF1 and IRF5 loci and lupus nephritis, IRF8 and hematological disorders, or NCF2 locus and arthritis, all of them already described in Chinese populations [49–52]. However, these new suggestive subphenotype-specific associations with moderate significance values should be taken with caution and replicated in independent populations.

Overall, and despite the possibility of some study design limitations, results from the present study are consistent with previously established associations for the SLE susceptibility loci HLA, IRF5, BLK, ITGAM, and IRF8. The analysis of disease subphenotypes shows new specific associations such as those between UBE2L3 and skin involvement, and ICAM1-ICAM4-ICAM5 and hematological manifestations, in addition to other several suggestive associations which need to be replicated in an independent and larger populations.