Background A major systemic lupus erythematosus (SLE) susceptibility locus lies within a common inversion polymorphism region (encompassing 3.8 – 4.5 Mb) located at 8p23. Initially implicated genes included FAM167A-BLK and XKR6, of which BLK received major attention due to its known role in B-cell biology. Recently, additional SLE risk carried in non-inverted background was also reported.
Objective and methods In this case –control study, we further investigated the ‘extended’ 8p23 locus (~ 4 Mb) where we observed multiple SLE signals and assessed these signals for their relation to the inversion affecting this region. The study involved a North American discovery data set (~ 1200 subjects) and a replication data set (> 10 000 subjects) comprising European-descent individuals.
Results Meta-analysis of 8p23 SNPs, with p < 0.05 in both data sets, identified 51 genome-wide significant SNPs (p < 5.0 × 10−8). While most of these SNPs were related to previously implicated signals (XKR6-FAM167A-BLK subregion), our results also revealed two ‘new’ SLE signals, including SGK223-CLDN23-MFHAS1 (6.06 × 10−9 ≤ meta p ≤ 4.88 × 10−8) and CTSB (meta p = 4.87 × 10−8) subregions that are located > 2 Mb upstream and ~ 0.3 Mb downstream from previously reported signals. Functional assessment of relevant SNPs indicated putative cis-effects on the expression of various genes at 8p23. Additional analyses in discovery sample, where the inversion genotypes were inferred, replicated the association of non-inverted status with SLE risk and suggested that a number of SLE risk alleles are predominantly carried in non-inverted background.
Conclusions Our results implicate multiple (known+novel) SLE signals/genes at the extended 8p23 locus, beyond previously reported signals/genes, and suggest that this broad locus contributes to SLE risk through the effects of multiple genes/pathways.
Statistics from Altmetric.com
Systemic lupus erythematosus (SLE) is a multisystem autoimmune disease that is highly heterogeneous in clinical presentation and also highly variable in prognosis. As a complex human disease predominantly affecting women of reproductive age, SLE susceptibility and manifestation are modulated by a complex interplay of heritable, hormonal and environmental factors. The complex genetic basis of SLE has been supported by a growing number of susceptibility loci/genes identified until now by candidate gene and/or genome-wide association studies (GWASs) in various ethnic groups.1–3
One of the major loci identified by initial high-density GWASs of SLE in European-descent subjects4–6 lies within a common large inversion polymorphism region (encompassing 3.8–4.5 Mb) on chromosome 8p237–9 and has since been replicated in multiple studies and in various ethnic groups.1–3 Initially implicated genes at 8p23 included FAM167A-BLK and XKR6, of which BLK (BLK proto-oncogene, Src family tyrosine kinase; http://www.ncbi.nlm.nih.gov/gene/640) has subsequently received major attention due to its known role in B-lymphocyte development and signalling. BLK was also shown to physically interact with BANK1, another SLE-associated gene product involved in B-cell biology.10 Following its first recognition for its involvement in SLE susceptibility, the 8p23 locus has also been shown to be associated with other autoimmune diseases.11–15
The objective of our study was to further investigate the ‘extended’ 8p23 locus (~4 Mb) where we observed multiple SLE signals and assess these signals for their relation to the inversion affecting this chromosomal region. The study involved a North American discovery data set (~1200 subjects) and an independent replication data set (>10 000 subjects) comprising European-descent subjects. Here, we report our results that implicate multiple (known+novel) genes/signals in this broad 8p23 region, which seems to contribute to SLE risk beyond the commonly studied BLK gene.
Materials and methods
Discovery data set used for the investigation of the extended 8p23 locus
The chromosome 8p23 data were derived from a recent SLE GWAS conducted on Affymetrix Genome-Wide Human SNP Array 6.0, which included 1166 European-descent subjects after excluding those with cryptic relationship from ~1200 individuals initially genotyped.16 Patients with SLE (n=676) were 18 years or older (mean±SD age: 45.1±12.5 years; 97.3% women) and all met 1982 or revised 1997 American College of Rheumatology classification criteria for SLE.17 18 The controls (n=490) were 21 years or older (mean±SD age: 48.9±10.6 years; 100% women). Additional information on these samples can be found in previous publications.19–22 At recruitment, written informed consents were obtained from all study subjects for the genetic studies approved by the institutional review boards of participating centres.
After applying the traditional GWAS quality control (QC) filters on samples/markers and also performing population stratification analysis, the final association analysis comprised 1148 subjects (661 SLE cases and 487 controls) and included 1687 markers (minor allele frequency (MAF) ≥0.01, call rate ≥95%, Hardy-Weinberg equilibrium (HWE) p>1.0×10−6) located at the ‘extended’ 8p23 locus (~4 Mb). Logistic regression analysis was performed under an additive model using the recruitment site, sex, age and first four principal components (PCs) as covariates. PLINK was used for all single-site association analyses (http://pngu.mgh.harvard.edu/~purcell/plink/), while Haploview v.4.2 was used for pairwise linkage disequilibrium (LD), Tagger and haplotype analyses (https://www.broadinstitute.org/haploview/haploview).
Replication data set used for the investigation of the extended 8p23 locus
Data from another recent and independent European GWAS of SLE23 were used for the replication of SLE signals observed at the extended 8p23 locus in the discovery sample. The replication sample included 4036 patients with SLE (90% women) and 6959 controls (50% women; 5699 from the National Institutes of Health (NIH) Health and Retirement Study) genotyped on HumanOmni1-Quad BeadChip. The data were imputed to the density of the 1000 Genomes Project study (using the 1000 genomes phase I reference panel and IMPUTE V2.2.3) and the association analysis24 was performed under an additive model computed by SNPTEST using the first four PCs as covariates.
Meta-analysis of the discovery and replication data sets for relevant markers
The METAL software25 was used to perform the meta-analysis of discovery and replication data sets for the markers that showed p<0.05 in both data sets.
In silico assessment of putative functional effects of the top SLE-associated variants
RegulomeDB v1.1 (http://www.regulomedb.org/), a database of DNA features and regulatory elements in non-coding genomic regions in humans, was used to evaluate the regulatory potential of the variants of interest. The details of RegulomeDB variant scoring scheme are provided in online supplementary table S3 footnotes, which overall include the following main categories: category 1 (indicates the strongest functional evidence such as ‘alteration of transcription factor binding and a gene regulatory effect’) for ‘likely to affect binding and linked to expression of a gene target', category 2 for 'likely to affect binding ', category 3 for 'less likely to affect binding' and categories 4–6 for 'minimal binding evidence’.
Additionally, the cis-eQTL (expression quantitative trait locus) effects of the relevant SNPs were examined using an SQL database of genome-wide SNP associations with human monocyte expression traits.26 This database was created from the data on 1490 individuals of European origin obtained by using Affymetrix SNP Array 6.0 and Illumina Human HT12 BeadChip.26
Assessment of chromosome 8 inversion status and its relation to the top SLE signals at 8p23
Given that the SLE locus at 8p23 falls within a known common inversion polymorphism (8p23-inv) region and a recent study27 reported a new association between SLE and 8p23 non-inversion status, we also sought to infer the inversion genotypes in our discovery sample in order to assess the effect of inversion status on SLE susceptibility and its relation to SLE-associated SNPs. For this purpose, we adopted the same approach used by that recent SLE study27 in order to obtain comparable results in our study. Given the lack of perfect surrogate marker(s) for 8p23-inv (due to the absence of markers in complete LD), this approach involves a novel statistical method that is based on a principal component analysis (PCA) performed locally in the inversion region using unphased high-density SNP genotype data.28 In our post-QC GWAS data (1148 individuals), a total of 1687 SNPs (MAF ≥0.01, spread over a ~4 Mb region from position ~8.1 Mb to ~12.2 Mb on chromosome 8) were located in the predicted 8p23-inv region (spanning from 7.2 Mb to 12.4 Mb), of which 1489 were used to successfully merge our data with the HapMap-3 data (ftp://ftp.ncbi.nlm.nih.gov/hapmap/genotypes/hapmap3_r3/plink_format/) from 267 Caucasian individuals (165 CEU and 102 TSI subjects, including some previously typed by FISH assay).9 27 To infer the inversion status, PCs were calculated and the first PC was used to cluster the samples into three groups using the k-means algorithm and squared Euclidean distance as the similarity metric. Association analysis with the inversion status was performed using the same genetic model and covariates mentioned above for SNP association analysis performed in the discovery sample. The PLINK and R statistical software were used to conduct all necessary analyses.
Investigation of known and new SLE association signals at the extended 8p23 locus
The 8p23 locus was among the top three known SLE loci replicated in our discovery sample following the major histocompatibility complex locus at 6p21 and STAT4 at 2q32.16 At the ‘extended’ 8p23 locus (~4 Mb), over 500 SNPs were observed at p<0.05 in our discovery sample, of which over 80 had p<0.001.
A total of 425 SNPs with p<0.05 in the discovery sample were also present and significant in the replication sample. Meta-analysis of these 425 markers identified a total of 51 SNPs with genome-wide significance (meta p<5.0×10−8) as the top SLE-associated variants (table 1, figure 1: red dots represent the SNPs with meta p<5×10−8). Online supplementary figure S1 shows the LD and pairwise correlations among these 51 SNPs in the discovery sample. Tagger analysis of these 51 SNPs using an r 2 cut-off of 0.5 (to identify the SNP groups that were not in high LD) yielded a total of eight separate groups (table 1). While most of these SNPs/groups were related to previously implicated SLE signals in the XKR6-FAM167A-BLK subregion (groups 2–7; 1.35×10−19≤meta p≤4.53×10−8), we also identified two ‘new’ SLE signals located upstream (>2 Mb) and downstream (~0.3 Mb) from previously reported signals (table 1, figure 1: new SLE signals/subregions are marked by green arrows). The upstream ‘new’ SLE signal (group 1; seven SNPs, 6.06×10−9≤meta p≤4.88×10−8) was identified in the SGK223-CLDN23-MFHAS1 subregion and the downstream ‘new’ SLE signal (group 8; rs880632, meta p=4.87×10−8) was detected near the CTSB gene (located upstream from the beta-defensin genes).
The LD between new and known SLE signals/SNPs was either minimal (0.07≤r 2 ≤0.22 for the downstream ‘new’ signal) or weak to modest (0.18≤r2≤0.54 for the upstream ‘new’ signal) (online supplementary figure S1). One SNP (rs11998678) located further downstream at 8p23 (near beta-defensin genes) was in LD with other SNPs in group 2 (table 1).
The pairwise conditional analysis results for top 51 SNPs in the discovery and replication samples are shown in online supplementary tables S1 and S2, respectively. As expected, the results of conditional analysis were influenced by the extent of pairwise correlations and the strength of individual associations. In the discovery sample, both of the top SNPs representing the upstream and downstream new SLE signals (rs10087493 and rs880632, respectively) were among the four SNPs that remained nominally significant (2.01×10−2≤ conditioned p≤4.19×10−2) after conditioning on the most significant BLK SNP (rs1478897) in this sample (see the related column in online supplementary table S1—the last section/page of the table). Similarly in the replication sample, the rs880632 SNP that represents the downstream new SLE signal was the most significant SNP (conditioned p=1.61×10−2) after conditioning on the strongest BLK SNP (rs2736340) in this sample (see the related column in online supplementary table S2), followed by four SNPs with marginal p values (4.72×10−2≤ conditioned p≤5.12×10−2), including the top SNP (rs1567398) representing the upstream new SLE signal in this sample. Overall, BLK/rs2736340 and CTSB/rs880632 were the only two SNPs that survived all pairwise conditional analyses in the large replication sample (see the related rows in online supplementary table S2).
In silico functional assessment of the top SLE signals identified at the extended 8p23 locus
Initial functional assessment of top 51 SNPs (with meta p<5.0×10−8) using RegulomeDB (see online supplementary table S3 — left panel) revealed the following: (i) a number of SNPs with strong regulatory potential (27 SNPs with the score of ‘1’), (ii) a consistent cis-effect of most of the SNPs on the expression of CLDN23 and FAM167A (previously known as C8orf13) along with other regulatory functions and (iii) the indication for cell type-specific effects/functions.
Subsequent evaluation of these 51 SNPs using the data from a recently published ‘monocyte’ eQTL study26 (see online supplementary table S3 — right panel), which also used Affymetrix SNP Array 6.0 that was used in our discovery sample, further confirmed the consistent and most significant cis-effect of all evaluated SNPs (including BLK SNPs) on the expression of CLDN23, a gene located within the ‘upstream new SLE signal’ subregion. Although at a lesser degree, a widespread cis-effect of the evaluated SNPs was also observed on the expression of FAM167A (located near BLK), FDFT1 (located near CTSB) and SGK223 (located near CLDN23). On the other hand, the association with BLK expression was restricted to a subset of SNPs located within/near the previously implicated XKR6-FAM167A-BLK subregion (see online supplementary table S3 — right panel).
All seven SNPs located within/near the SGK223-CLDN23-MFHAS1 subregion (upstream new SLE signal) showed cis-effects on the expression of these three local genes and had RegulomeDB scores ranging from 1b to 6 (see online supplementary table S3). Moreover, several other SNPs located throughout the extended 8p23 locus also showed association with the expression of SGK223 and CLDN23 in human monocytes (as stated above) and a few distant cis-eQTL effects were also observed on MFHAS1 (see online supplementary table S3 — right panel). While the rs880632 SNP (downstream new SLE signal—near CTSB; RegulomeDB score=1f) showed a cis-effect on the expression of various genes (see online supplementary table S3), it was the only SNP (out of 51) that was associated with CTSB expression.
Effect of the 8p23 inversion polymorphism on SLE susceptibility and its relation to the top SLE signals at 8p23
We used a recently developed PCA-based statistical approach27 28 to infer the inversion genotypes in our discovery sample in order to assess the effect of inversion status on SLE susceptibility and its relation to known and novel SLE signals at 8p23. Figure 2 shows three clusters identified by the PCA of our European-descent discovery sample (n=1148, in blue) along with the HapMap-3 Caucasian sample (n=267, in red). The numbers indicate the inversion status: 1=inverted homozygous, 2=heterozygotes and 3=non-inverted homozygous.27 The distribution of 8p23-inv in our discovery sample was as follows: 28% inverted homozygous, 49% heterozygous and 22% non-inverted homozygous. The frequency of non-inversion allele was significantly higher in SLE cases than in controls (50.6% vs 41.6%; OR=1.47, p=8.9×10−5) and the ratio of SLE cases to the controls in each group was as follows: 0.95 in inverted homozygous group, 1.44 in heterozygous group and 1.94 in non-inverted homozygous group.
Online supplementary figure S2 demonstrates the LD and pairwise correlations between the inversion polymorphism and 51 SLE-associated top SNPs in our discovery sample. 8p23-inv showed the lowest correlations with a subset of SNPs located within/near BLK gene (rs2736340, rs2618476, rs998683, rs9329246, rs1478897; 0.33≤r 2 ≤0.39) and with the rs880632 SNP located near CTSB gene (downstream new SLE signal; r 2 =0.25).
In order to further assess the relation of 8p23-inv with eight separate SNP groups described in the previous section (table 1), we performed additional analyses on 8p23-inv and one SNP selected from each group (the most significant one in the discovery sample where the inversion genotypes were also available) along with additional BLK SNPs that showed the strongest associations in the meta-analysis (meta p<5×10−15). Online supplementary figure S3 shows the LD and pairwise correlations between the inversion polymorphism and 10 selected SNPs in our discovery sample (eight SNPs representing eight separate groups+two additional BLK SNPs— italics: rs10087493-rs11783045-rs2736340-rs2618476-rs998683-rs2618443-rs1478897-rs17807624-rs880632-rs11998678). The LD findings (r 2 =0.60 for inversion and rs10087493 and 0.25 for inversion and rs880632) and the haplotype analysis results (see online supplementary table S4) have indicated that the upstream new SLE risk signal (represented by rs10087493 risk allele) is predominantly carried in the non-inverted background, while the downstream new SLE risk signal (represented by rs880632 risk allele) is similarly observed in both non-inverted (N) and inverted (I) backgrounds. Haplotype association analysis of 8p23-inv+10 selected SNPs (see online supplementary table S4) has revealed that the most significant common haplotype carried all the risk alleles for SLE (NAATGAATACA; 18.9% in cases vs 12.7% in controls, p=1.00×10−4), while the second most significant common haplotype carried all the protective alleles (IGGCAGGAGAG; 9.8% in cases vs 14.5% in controls, p=8.00×10−4). The common haplotype that carried the risk allele of the downstream new SLE signal (represented by rs880632) but the protective alleles of all of the remaining polymorphisms was no longer significant (IGGCAGGAGCG; 12.5% in cases vs 14.9% in controls, p=0.1086), neither was the fourth common haplotype that carried a combination of risk and protective alleles (IGACAGGAGCG; 6.0% in cases vs 6.5% in controls, p=0.6659). The results of pairwise conditional analysis of 8p23-inv+10 selected SNPs are shown in online supplementary table S5, where some SNPs maintained their significance after conditioning on 8p23-inv, which again seemed to be influenced by the extent of pairwise correlations and the strength of individual associations. Altogether, the above results have suggested that the (known+novel) signals that appear to contribute to SLE risk at the extended 8p23 locus appear to include a number of SNPs with risk alleles predominantly carried in non-inverted background as well as some SNPs acting more independent from the inversion.
In this study, we further investigated the ‘extended’ 8p23 locus (~4 Mb) where we identified multiple (known+novel) SLE signals (figure 1, table 1). We also assessed the functional significance of the identified SNPs/signals as well as their relation to the large inversion polymorphism (8p23-inv) affecting this chromosomal region.7–9
Known and new SLE association signals at the extended 8p23 locus and their putative functional effects
Of a total of 425 SNPs that had p<0.05 in both discovery and replication samples at the extended 8p23 locus, 51 reached genome-wide level of significance (p<5.0×10−8) in meta-analysis (table 1, figure 1: red dots) and thus we focused on these top SLE-associated variants for further evaluation and analyses. These 51 SLE-relevant SNPs were divided into eight separate groups after running a Tagger analysis in our discovery sample (table 1). While most of these SNPs/groups were related to previously implicated SLE signals at 8p23 (the XKR6-FAM167A-BLK subregion), we also identified two ‘new’ SLE signals located upstream (>2 Mb) and downstream (~0.3 Mb) from previously reported signals, falling within/near the SGK223-CLDN23-MFHAS1 and the CTSB subregions, respectively (table 1, figure 1: green arrows). The LD and pairwise correlations of these new SNPs/signals with the known SNPs/signals were either weak to modest (for upstream new SLE signal) or minimal (for downstream new SLE signal) (see online supplementary figure S1). The newly identified SLE signals have also been further supported by the results of our conditional regression analyses (see online supplementary table S1 and S2).
Earlier studies on the effect of 8p23-inv on local gene expression have suggested that inversion-associated expression patterns and inversion-eQTLs are predominantly mediated by the effects of specific SNP/allele configurations maintained in the inversion background.8 9 Our in silico functional assessment of 51 SLE-relevant SNPs identified in the 8p23inv region using two public databases (see online supplementary table S3) has revealed additional new and interesting findings. More than half of these SLE-relevant SNPs showed a RegulomeDB score of ‘1’, suggesting a strong regulatory potential. Moreover, a consistent and strong cis-effect on CLDN23 expression in human monocytes was observed for these SLE-relevant SNPs (including the BLK SNPs) throughout the extended 8p23 locus, while the effect on BLK expression was restricted to a subset of SNPs located within/near the previously implicated XKR6-FAM167A-BLK subregion.
All seven SNPs representing the upstream new SLE signal showed cis-effects on the expression of three local genes (SGK223, CLDN23 and MFHAS1) residing in that 8p23 subregion (see online supplementary table S3 —right panel). CLDN23 (http://www.ncbi.nlm.nih.gov/gene/137075) encodes a member of the claudin family of integral membrane proteins, which are structural and functional components of tight junctions that are involved in the regulation of paracellular permeability and signal transductions in endothelia/epithelia.29 Skin involvement constitutes a major component of SLE and epidermal tight junctions were proposed to be relevant for the pathogenesis of various inflammatory and neoplastic cutaneous conditions.30 Consistent expression of tight junction proteins was also reported in human leucocytes, primarily in lymphocytes and monocytes, and they were suggested to play a role in immune activity and autoimmunity.31 Tight junction proteins (eg, CLDN19) were also proposed among possible biomarkers for refractory lupus nephritis.32 SGK223 (http://www.ncbi.nlm.nih.gov/gene/157285) encodes a member of the tyrosine protein kinase family and, in recent years, tyrosine kinases have been increasingly targeted for new drug development for inflammatory and autoimmune diseases.33–35 MFHAS1 , a.k.a. MASL1 (http://www.ncbi.nlm.nih.gov/gene/9258), encodes a ROCO protein that is believed to interact with the cell cycle and it was also shown to regulate the Toll-like receptor (TLR)-dependent signalling.36–38
The rs880632 SNP that represents the downstream new SLE signal (near CTSB) showed a cis-effect on the expression of the CTSB gene (see online supplementary table S3 —right panel) in addition to be associated with the expression of CLDN23, FAM167A (near BLK) and FDFT1 (near CTSB). CTSB (http://www.ncbi.nlm.nih.gov/gene/1508) encodes a lysosomal cysteine proteinase (cathepsin B) that plays various roles in protein turnover and cell processes and it has also been implicated in autoimmunity and arthritis.39–41 Cathepsin B was shown to regulate various immune functions; for example, persistence of memory T cells, antigen-presenting function of dendritic cells, monocyte and macrophage necrosis/necroptosis and TLR signalling.42–46
The publicly available data on human monocytes and lymphoblastoid cell lines (see online supplementary table S3) have suggested cell type-specific effects/functions of the 8p23 SNPs as also supported by a recent BLK study.47 Guthridge et al 47 performed a comprehensive analysis of the FAM167A-BLK subregion, which has revealed two functional variants that regulate alternative promoter activities in cell type-specific and developmental stage-specific manners. Moreover, a recent SLE study48 reported widely divergent transcriptional patterns in sorted immune cell populations, further emphasising the importance of investigating multiple cell types in gene expression studies of SLE. Taking together the previous and current observations, the ‘extended’ 8p23 locus appears to contribute to SLE risk through multiple signals affecting the expression/function of various 8p23 genes in various cell populations in cell type-specific and developmental stage-specific manners and thus a comprehensive functional evaluation of this important broad SLE locus in various cell types is warranted.
The relation of the 8p23 inversion polymorphism to SLE and to top SLE signals at 8p23
Given that this extended 8p23 locus lies within a known inversion polymorphism (8p23-inv) region and a recent study27 reported the association of non-inverted status as an additional risk factor for SLE, we also implemented the reported PCA-based approach to infer and assess the inversion genotypes in our discovery sample. The distribution of inversion genotypes in our sample and the ratio of SLE cases to controls in each genotype group were very similar to those reported previously,27 and the non-inversion allele was also found to be associated with increased SLE risk in our sample. Also consistent with the reported observation,27 the correlation of the inversion with some BLK SNPs was low in our sample (0.33≤r 2 ≤0.39); however, it was the lowest with the newly identified CTSB signal/SNP (r 2 =0.25) and relatively higher with the newly identified SGK223-CLDN23-MFHAS1 signal/SNPs (0.54≤r 2 ≤0.71) (see online supplementary figure S2). Overall, the results from various analyses in our discovery sample (see online supplementary figure S3, tables S4-S5) have suggested that multiple (known+novel) signals that appear to contribute to SLE risk at the extended 8p23 locus seem to be driven by multiple SNPs with risk alleles predominantly residing in non-inverted background as well as by some SNPs/alleles acting more independent from the inversion. Hence, the recently reported27 additional SLE risk carried in non-inverted background may be related to some of the additional signals/SNPs identified in our study (table 1). Given that inversions are known to influence the expression of multiple genes in affected regions,8 9 27 the SLE protective alleles carried in the inversion background might be executing their beneficial effects by altering the expression of multiple genes. However, for those SNPs that show high correlation with the inversion, it is statistically difficult to separate their individual effects from the inversion.
In conclusion, our results support the recent report of additional SLE risk carried in non-inverted background at 8p23 and furthermore implicate new SLE signals at the ‘extended’ 8p23 locus. Although the role of BLK in immune regulation is now well documented and that of newly implicated CTSB is also increasingly recognised, much remains to be learnt about other newly implicated and relatively understudied SLE-relevant genes (SGK223, CLDN23 and MFHAS1) and their role in (or interactions with) the immune system. Overall, our results suggest that the extended 8p23 locus contributes to SLE risk through cell type-specific effects of multiple (known+novel) signals/genes involved in various (known+novel) biological pathways/mechanisms. Some of the newly implicated pathways/mechanisms include possible involvement of the paracellular transport pathway/tight junctions (CLDN23) and the protein turnover/cathepsins (CTSB). Our findings need to be further confirmed in independent large studies and also warrant a comprehensive analysis of this ‘extended’ 8p23 region beyond the frequently studied FAM167A-BLK subregion. Given that previous SLE studies on the FAM167A/BLK subregion47 49 50 have identified both common and uncommon (low-frequency/rare) functional variants, it is reasonable to hypothesise that the ‘extended’ 8p23 locus harbours both common and uncommon SLE-relevant functional variants throughout the implicated region. Hence, a comprehensive follow-up analysis (sequencing + functional studies) of this important ‘broad’ locus at 8p23 is necessary to identify all common/uncommon causal variants, characterise their functions and unravel the full spectrum of affected pathways/mechanisms, with the ultimate goal of improving our understanding of SLE pathogenesis and uncovering new targets for future therapeutic interventions.
Contributors FYD and MIK planned the study, conducted research, evaluated/interpreted the data/results and drafted the manuscript. XW and EF conducted research, performed statistical analysis and/or provided statistical expertise and evaluated/interpreted the data/results. SB, CP, AC, RR-G and SM conducted research, provided the samples/data and clinical expertise. DLM and TJV conducted research, performed the analyses in the replication sample and provided the replication data and expertise. All authors contributed to the research and to the critical review of the manuscript.
Funding This work was supported by grants from the US NIH (HL092397, HL088648, AR057028, AR046588, AR057338, HD066139, AR002138, AR030692, AR064464 and TR000150) and by grants from Wellcome Trust (Ref 085492) and Arthritis Research UK (Ref 19289).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.