Introduction

Type 1 diabetes (T1D) is a common, complex autoimmune disease arising from the destruction of the insulin-producing pancreatic β cells. Multiple genetic and environmental risk factors contribute to T1D. Although genome-wide association studies (GWAS) have discovered over 40 chromosomal regions where there is significant statistical evidence of association with T1D [1, 2], it remains challenging to pinpoint the causative genes and variants located in most of these regions. Furthermore, the majority of lead GWAS single-nucleotide polymorphisms (SNPs) for autoimmune diseases, including T1D, are noncoding, and their functions in disease pathogenesis are largely unknown. To tackle these problems, we herein investigated a group of disease-relevant, common noncoding SNPs in UBASH3A, the most likely candidate gene for the T1D risk locus on chromosome 21q22.3 [1,2,3,4,5].

UBASH3A is expressed primarily in T cells, and plays a broad role in autoimmunity; five different common autoimmune diseases have significant reported associations with SNPs in UBASH3A [1,2,3,4,5,6,7,8,9,10,11]. We previously showed that UBASH3A inhibits T-cell receptor-induced nuclear factor-κB (NF-κB) signaling and thereby suppresses IL2 expression, a target gene of NF-κB, and that the minor alleles of rs80054410 (GRCh37 chr21:g.43836010T>C) and rs11203203 (GRCh37 chr21:g.43836186G>A)—two credible causative variants located in a potential enhancer in intron 4 of UBASH3A [2]—enhance UBASH3A expression in human primary T cells while diminishing IL2 expression upon T-cell receptor stimulation [12]. These findings provide a mechanistic explanation for the association of UBASH3A with autoimmunity, as interleukin-2 (IL-2) is a crucial cytokine for the differentiation and function of T cells.

In the current study, we examine the role of another noncoding SNP in UBASH3A, rs1893592 (GRCh37 chr21:g.43855067A>C). The minor allele of rs1893592 protects against three autoimmune diseases, rheumatoid arthritis, celiac disease, and primary sclerosing cholangitis [8, 10, 11, 13], and it has been proposed to affect splicing, rather than expression, of UBASH3A [13, 14]. Our study reconciles these disparate mechanistic roles for autoimmune-associated SNPs in UBASH3A and highlights the combinatorial effects of noncoding UBASH3A variants on T1D risk.

Materials and methods

Frozen peripheral blood mononuclear cells from healthy subjects of European ancestry were obtained from the Type 1 Diabetes Genetics Consortium and from STEMCELL Technologies (Vancouver, BC, Canada). Genotype and phenotype data used in this study are available in dbGaP (www.ncbi.nlm.nih.gov/gap; Study Accession: phs000911.v1.p1). All biospecimens and data were represented by only non-identifying codes. This study was approved by the University of Florida Institutional Review Board.

Quantitative PCR and isolation and stimulation of human primary CD4+T cells were performed as previously described [12]. Primers used for quantitative PCR are provided in Supplementary Table 1.

IL-2 in culture supernatants was measured using the LEGENDplex kit (BioLegend, San Diego, CA, USA), following the manufacturer’s protocol.

The Family-Based Association Test program (v2.0.3) was used for single-marker association tests and haplotype analyses [15]. Conditional analysis was conducted using the UNPHASED program (v3.1.7). The Prism software (Graphpad v7.0) was used to perform Student’s t-tests and linear regression analysis.

Results

The minor allele of rs1893592 protects against T1D

To examine the effect of rs1893592 on risk of T1D, we conducted single-marker family-based association tests using data from 10,796 individuals from 2732 T1D-affected sibling pairs and trio families of European ancestry. The minor allele of rs1893592 was associated with reduced risk for T1D (P = 2.6 × 10−5, Table 1). rs11203202 (GRCh37 chr21:g.43825357C>G) was previously identified as the index SNP for the 21q22.3 risk locus in a large-scale fine-mapping study for T1D [2]. As expected, the minor alleles of rs11203202 (P = 2.7 × 10−4), rs80054410 (P = 4.3 × 10−4), and rs11203203 (P = 4.0 × 10−4) were each found to be associated with increased risk for T1D (Table 1) [2, 12].

Table 1 Single-marker family-based association tests of noncoding variants in the UBASH3A gene
Full size table

To explore the relationship of rs11203202, rs80054410, rs11203203, and rs1893592, we performed family-based haplotype association analysis and identified three T1D-associated haplotypes—H2, H3, and H9 (Table 2). The common H3 haplotype carrying the minor, protective allele of rs1893592 and the major alleles of rs11203202, rs80054410, and rs11203203 were associated with reduced risk for T1D (P = 3.8 × 10−6). The common H2 haplotype carrying the major allele of rs1893592 and the minor, risk alleles of rs11203202, rs80054410, and rs11203203 conferred risk for T1D (P = 0.0018). The rare H9 haplotype (frequency = 0.1%) carrying the minor alleles of all four SNPs also conferred risk for T1D (P = 0.012, Table 2). Even after conditioning on rs80054410 and rs11203203, rs1893592 exhibited a significant association with T1D (P = 0.026).

Table 2 Family-based haplotype association analysis of UBASH3A
Full size table

rs1893592 modulates the expression of UBASH3A and IL2 in human CD4+ T cells upon stimulation

rs1893592 is located at the +3 position of intron-9 of UBASH3A (NM_001001895.2 and NG_029750.1), and could affect splicing of UBASH3A. Our prior RNA-seq analysis of human primary T cells uncovered an alternative UBASH3A transcript that runs into intron-9 of UBASH3A and contains a premature stop codon [16]. We verified this transcript (ENST00000473381.1, hereafter referred to as intron-9 UBASH3A) by RT-PCR and Sanger sequencing.

We performed quantitative PCR to further characterize intron-9 UBASH3A. Our results revealed that human primary CD4+ T cells, regardless of their genotype at rs1893592, expressed intron-9 UBASH3A, albeit at low levels (Fig. 1a). In addition, the amount of intron-9 UBASH3A mRNA was positively correlated with that of total UBASH3A mRNA in both stimulated and mock-stimulated CD4+ T cells (Fig. 1a).

Fig. 1
figure 1

Effects of rs1893592 on expression of UBASH3A and IL2 in human primary CD4+ T cells upon stimulation. a–d Human primary CD4+ T cells from healthy subjects of European ancestry were stimulated with anti-CD3 plus anti-CD28, or with culture medium, for 6 h. Each data point represents one subject, and the data are pooled from four quantitative PCR experiments. In a, the best-fit line generated by linear regression is represented by solid line, and statistics for correlation and significance are shown. In b–d, all subjects are homozygous for the major alleles of rs80054410 (A) and rs11203203 (G). The mean and SEM values are indicated by solid lines and error bars, respectively. P-values from unpaired two-tailed Student’s t-tests are shown. In b, the relative mRNA levels of total UBASH3A and intron-9 UBASH3A are shown. ns, not significant. In c, the ratio of intron-9 UBASH3A mRNA to total UBASH3A mRNA is shown. In d, the relative mRNA level of IL2 is shown. e Human primary CD4+ T cells from healthy subjects of European ancestry were stimulated with anti-CD3 plus anti-CD28 for 24 h. IL-2 concentrations in the culture supernatants are shown, and the data are pooled from two LEGENDplex assays. Each data point represents one subject, and P-value from unpaired one-tailed Student’s t-test is shown

Full size image

To assess the effect of rs1893592, we controlled for genotypes at rs80054410 and rs11203203 by selecting subjects homozygous for the major alleles of these two SNPs. We found that in stimulated CD4+ T cells, the minor, protective C allele of rs1893592 was associated with a significant reduction in the total amount of UBASH3A mRNA but not the amount of intron-9 UBASH3A transcript (Fig. 1b). As a result, the proportion of intron-9 UBASH3A mRNA relative to total UBASH3A mRNA was twice as large in rs1893592 A/C subjects as in rs1893592 A/A subjects (Fig. 1c). In addition, CD4+ T cells from rs1893592 A/C subjects, upon stimulation, produced more IL2 transcripts than those from rs1893592 A/A subjects (Fig. 1d).

To further test the hypothesis that the minor allele of rs1893592 protects against T1D by enhancing IL-2 production, we measured IL-2 secretion by primary CD4+ T cells from healthy subjects upon 24-h stimulation with anti-CD3 plus anti-CD28, without controlling for genotypes at other T1D-associated SNPs in UBASH3A or other risk loci. We found that CD4+ T cells from rs1893592 A/C subjects secreted five times more IL-2 than rs1893592 A/A subjects (Fig. 1e).

Discussion

In this study, we sought to define how the minor allele of the intronic SNP rs1893592 in UBASH3A protects against T1D in the presence of risk alleles at other SNPs in the same gene. The results of our haplotype analysis suggest an additive model, at least at the population level, where T1D risk results from the sum of the individual effects of the four noncoding UBASH3A SNPs. This is consistent with the notion—predicted by quantitative genetics theory and supported by empirical data—that most genetic variance in a population is additive [17]. Furthermore, this model is supported by our mechanistic findings that these UBASH3A SNPs affect, albeit in different directions, the same molecular pathway—the expression of UBASH3A, and thereby IL2 expression via the NF-κB signaling pathway [12].

UBASH3A is primarily expressed in T cells with little or no expression in other cell types, including B cells. Previous studies that assayed UBASH3A mRNA in human whole blood or B-lymphoblastoid cell lines concluded that rs1893592 was a splice QTL whose minor allele promoted the retention of intron-9 of UBASH3A [13, 14]. In contrast, Raj et al. [18] identified rs1893592 as the most significant cis-eQTL in UBASH3A in primary CD4+ T cells from European Americans, and the minor allele was found to be associated with decreased expression of UBASH3A. Our results in stimulated human CD4+ T cells both reconcile and extend these prior findings in several important ways. First, our study reveals that intron-9 UBASH3A is a normally produced transcript. Second, the impact of rs1893592 is two-fold: its minor allele reduces overall transcript levels of UBASH3A, while increasing alternative splicing, favoring the production of intron-9 UBASH3A (Fig. 1c). Third, since rs1893592 does not significantly alter the absolute levels of intron-9 UBASH3A, the most likely mechanism whereby variation at rs1893592 affects T1D risk is via its effects on overall UBASH3A mRNA levels rather than through the production of a novel protein isoform from the intron-9 UBASH3A transcript.

We showed that a significant outcome of reduced total amount of UBASH3A mRNA, resulting from the minor allele of rs1893592, is enhanced transcription and production of IL-2 in stimulated CD4+ T cells. This explains, at least partially, the protective effect of the minor allele of rs1893592, as IL-2 production is impaired in some subjects with T1D or rheumatoid arthritis [19, 20]. We noticed that the amounts of secreted IL-2 by CD4+ T cells varied markedly among rs1893592 A/C subjects (Fig. 1e). This is likely due to the facts that we grouped the subjects based only on their genotype at rs1893592, and that IL-2 production can be affected by variants in genes other than UBASH3A, such as those in IL2RA and PTPN2, two T1D-associated genes involved in the IL-2 pathway [2]. Nevertheless, our results show that rs1893592 has a considerable impact on IL-2 production, which may facilitate future IL-2 clinical trials for T1D.