Contribution of Functional LILRA3, but Not Nonfunctional LILRA3, to Sex Bias in Susceptibility and Severity of Anti–Citrullinated Protein Antibody–Positive Rheumatoid Arthritis

Authors


Abstract

Objective

Leukocyte immunoglobulin-like receptor A3 belongs to a family of receptors with inhibitory or activating functions. Since Caucasian individuals lacking LILRA3 have been found to be susceptible to multiple sclerosis and Sjögren's syndrome, we undertook this study to examine whether LILRA3 deletion is a novel genetic risk factor for rheumatoid arthritis (RA) (another autoimmune disease), whether there are sex-specific effects, and whether LILRA3 influences the subtype and severity of RA.

Methods

The LILRA3 deletion and its tagging single-nucleotide polymorphism rs103294 were genotyped in a Northern Han Chinese cohort (N-Han) (1,618 cases and 1,658 controls) and a Southern Han Chinese cohort (S-Han) (575 cases and 549 controls). Association analyses were performed on the complete data set and subsets. The effect of the nondeleted (functional) LILRA3 allele on radiographic severity and LILRA3 expression was evaluated.

Results

In the N-Han discovery cohort, we unexpectedly observed a higher frequency of the functional LILRA3 in RA patients compared with healthy individuals (10.1% versus 6.3%; P = 4.01 × 10−5, odds ratio [OR] 1.92). The association was replicated in the S-Han cohort and confirmed by meta-analysis (P = 5.63 × 10−6, OR 1.83). Functional LILRA3 conferred greater risk for RA in males (P = 1.09 × 10−6, OR 4.47), and was specifically associated with anti–citrullinated protein antibody (ACPA)–positive RA (P = 3.05 × 10−4, OR 1.75). Furthermore, functional LILRA3 was associated with higher radiographic scores in ACPA-positive patients with early RA (P = 9.70 × 10−3) and higher LILRA3 messenger RNA levels (P = 3.31 × 10−8).

Conclusion

Our study provides the first evidence that functional LILRA3 is a novel genetic risk factor for RA, especially in males. It appears to highly predispose to ACPA-positive RA and confers an increased risk of disease severity in patients with early RA.

Rheumatoid arthritis (RA) is an autoimmune disease characterized by chronic joint erosion and destruction. The etiology of RA remains largely unknown, but it is widely accepted that genetic factors play major roles in disease pathogenesis. The disease course of RA is heterogeneous, and genetic factors are believed to influence not only susceptibility, but also severity. Furthermore, a number of candidate genes, particularly the shared epitope–containing HLA–DRB1 alleles, have been shown to be specifically associated with anti–citrullinated protein antibody (ACPA)–positive RA ([1-5]).

RA is more prevalent in women, with a female-to-male ratio of ∼3:1. Although sex hormones have been considered as candidates responsible for the sex bias, sex-specific genetic factors also play a role in RA susceptibility. For example, sex differences in associations with the IFNG locus have been observed in RA, but in women only ([6]), whereas the FCGR3A 158F/V variant has shown increased association with RA in men ([7]).

The leukocyte immunoglobulin-like receptors reside in human chromosome 19q13.4. This multigene family of HLA class I–recognizing receptors consists of activating receptors (LILRA1–6) and inhibitory receptors (LILRB1–5) expressed mainly on myeloid cells. LILRA3 (OMIM 604818) is unique among the LILRs in that the gene exhibits a presence or absence variation due to a 6.7-kb deletion, comprising the first 6 of a total of 7 exons. The 6.7-kb deletion removes all of the Ig-like domains of the gene, suggesting that the putative truncated protein would not exhibit functional significance ([8-10]). The frequencies of the LILRA3 deletion polymorphism vary widely among different populations, being extremely high in Northeast Asians (0.56–0.84) as compared with Europeans (0.17) or Africans (0.10). Genotyping of the 4 HapMap populations reveals that LILRA3 alleles are in strong linkage disequilibrium (LD) with LILRB2 alleles in Northeast Asians, suggesting that natural selection occurred on LILRA3 ([11]). Furthermore, LILRA3 is the only member of the LILR family that has a termination codon in the stalk leading to a deletion of the transmembrane domain, and therefore, is likely expressed only as a soluble receptor ([12-14]).

To date, little is known about the exact role of LILRA3 in the immune system, but studies have demonstrated that the 6.7-kb LILRA3 deletion was associated with autoimmune diseases, e.g., multiple sclerosis (MS) and Sjögren's syndrome (SS), in Caucasians ([15-17]). However, whether the nonfunctional LILRA3 gene is a common genetic risk factor for multiple autoimmune diseases besides MS and SS has not been investigated. In the present study we therefore aimed to determine whether the nonfunctional LILRA3 is a novel risk factor contributing to RA susceptibility, whether the genetic effect of LILRA3 is specific to any subsets of RA, whether LILRA3 influences disease phenotypes other than susceptibility, e.g., disease severity, and, finally, whether there is a sex-specific role. Unexpectedly, we found that functional LILRA3 was strongly associated with RA susceptibility and joint destruction, whereas nonfunctional LILRA3 was not. Specifically, our study revealed that functional LILRA3 was more strongly associated with RA in males, and conferred risk for the development of ACPA-positive RA.

PATIENTS AND METHODS

Study subjects

Two independent cohorts, consisting of 1,618 Northern Han Chinese RA patients and 1,658 unrelated healthy controls (N-Han cohort) and 575 Southern Han Chinese RA patients and 549 controls (S-Han cohort), were enrolled in the study. All patients satisfied the American College of Rheumatology 1987 revised criteria for RA ([18]). The populations of N-Han and S-Han were classified using the Yangtze River as a geographic boundary, as in previous studies ([19]). The N-Han cohort was examined as a discovery cohort, with associations confirmed in the S-Han cohort. Baseline characteristics of the patients and controls are shown in Table 1.

Table 1. Characteristics of the study cohorts*
 Northern Han cohortSouthern Han cohort
  1. RA = rheumatoid arthritis; ACPA = anti–citrullinated protein antibody; SHS = modified Sharp/van der Heijde score; NA = not available.
No. of patients/controls1,618/1,658575/549
Demographic characteristics  
Female, %  
Patients80.582.0
Controls74.876.7
Age, mean ± SD years  
Patients54.7 ± 13.250.5 ± 13.2
Controls38.3 ± 10.536.0 ± 11.9
Clinical characteristics (patients)  
Age at RA onset, mean ± SD years46.1 ± 14.343.9 ± 12.6
Disease duration, mean ± SD years8.7 ± 8.96.8 ± 6.3
ACPA positive, %81.480.3
SHS, mean ± SEM78.4 ± 63.8NA

In the N-Han cohort, the patients with RA were recruited from the Department of Rheumatology at Peking University People's Hospital, Xuanwu Hospital, and Peking University Third Hospital in Beijing. Of the 985 patients in this cohort with available information, 802 (81.4%) were ACPA positive, as determined and quantified using a second-generation anti–cyclic citrullinated peptide enzyme-linked immunosorbent assay kit (Euroimmun). Samples with antibody levels of >5 units/ml were classified as positive. Unrelated healthy controls were recruited from health care centers affiliated with People's Hospital. All patients and healthy controls in this cohort came from the traditional northern provinces of China, i.e., Beijing, Tianjin, Hebei, Shandong, Liaoning, Jilin, and Heilongjiang, and self-identified as northerners of Han ethnicity.

In the S-Han cohort, the patients with RA were recruited from the Department of Rheumatology, Number 1 Hospital, Anhui Medical University in Anhui province, and from the Department of Rheumatology, Gulou Hospital, Nanjing University in Jiangsu province. Of the 228 with available information, 183 (80.3%) were ACPA positive. The healthy controls were recruited from health care centers affiliated with the 2 hospitals. All patients and healthy controls in this cohort were from the traditional southern provinces of China, i.e., Jiangsu and Anhui, and self-identified as southerners of Han ethnicity.

The study was approved by the Medical Ethics Committee of Peking University People's Hospital. Written informed consent was obtained from all participants.

Determination of LILRA3-del by sequence-specific primer–polymerase chain reaction (PCR).

Genotyping for the presence or absence of LILRA3 was performed according to a previously reported method ([20]), with modified primers to optimize the PCR conditions in this study. The primers 5′-CCGCCCCAAAGCCCCTTCAT-3′ and 5′-TGCACTGTAGCCCCCGCTGA-3′ were used for detecting the presence of LILRA3 (670 bp), and the primers 5′-TGACCTGTGCCCAAGCATCCA-3′ and 5′-TGCACTGTAGCCCCCGCTGA-3′ were used for detecting the absence of LILRA3 (1,096 bp). The genotyping success rate was 99.1%. The genotypes obtained were subsequently validated and confirmed by sequencing the PCR products with an ABI 3700 Automated Sequencer (Applied Biosystems). Briefly, 26 individuals were sequenced for confirmation of the deletion of LILRA3 and 13 were sequenced for confirmation of the presence or absence of the LILRA3. The confirmation rate was 100%.

Single-nucleotide polymorphism (SNP) genotyping

One SNP (rs103294), known to be in strong LD with LILRA3 (r2 = 0.83), was genotyped using TaqMan genotyping assays (C_2942283_10; Applied Biosystems). Allelic discrimination was performed using an ABI 7300 Real-Time PCR system. The genotyping success rate for rs103294 was 92.4%.

Measurement of bone destruction

A total of 467 sets of hand radiographs from cases were available. All radiographs were graded chronologically, using the modified Sharp/van der Heijde score (SHS) ([21]), by an experienced professional reader (Xia Liu) who was blinded with regard to the clinical and genetic data. One hundred of the 467 sets were scored by a second professional reader (Nan Hong), and the interrater agreement rate was 0.99.

LILRA3 transcription quantification

A total of 62 RA patients were assessed for LILRA3 messenger RNA (mRNA) expression in peripheral blood mononuclear cells (PBMCs). All of these cases were derived from the N-Han cohort and had genotyping data on LILRA3 deletion. Cells were harvested and total RNA was extracted from PBMCs using an RNA extraction kit (Tiangen). Complementary DNA was synthesized using a RevertAid First Strand cDNA Synthesis Kit, according to the instructions of the manufacturer (Fermentas). For real-time PCR, 2-step PCR was performed using Power SYBR Green PCR Master Mix, according to the instructions of the manufacturer (Applied Biosystems). Primers used were as follows: GAPDH forward 5′-AAGGTGAAGGTCGGAGTCAA-3′, reverse 5′-AATGAAGGGGTCATTGATGG-3′; LILRA3 forward 5′-GCGCCAATCTCATAAGTACCAGGCTG-3′, reverse 5′-CCAGCCTTGGAGTCGGACTTGTT-3′ ([22]). The reaction was run on a 7300 Fast Real-time PCR System (Applied Biosystems). Gene expression was quantified relative to the expression of the housekeeping gene GAPDH, and normalized to control by standard 2−ΔΔCt calculation.

Power analysis

Power analyses were performed retrospectively for the available samples (cases and controls), using a fixed minor allele frequency (MAF) of 25%, a Type I error P value of 0.05, and an odds ratio (OR) of 1.4. PS software, version 3.0.14 (available at http://www.mc.vanderbilt.edu/prevmed/ps) was used for power calculation.

Meta-analysis

Meta-analysis using the Mantel-Haenszel method was performed with Review Manager 5 software (www.cc-ims.net/RevMan). A significant I2 statistic (I2> 30%, P < 0.05) indicated heterogeneity for ORs across studies. The fixed-effects model was applied in these metaanalyses.

Statistical analysis

Hardy-Weinberg equilibrium was assessed for each polymorphism, using Pearson's goodness-of-fit chi-square test. Pearson's chi-square test was used to calculate the significance of differences in allelic frequency between cases and controls. ORs and 95% confidence intervals (95% CIs) for alternative genetic model (recessive model) analysis were calculated using logistic regression, with adjustment for age and sex. LD and haplotype were calculated using Haploview, version 4.2 (http://www.broad.mit.edu/mpg/haploview/). The significance of differences in radiographic scores between genotypic groups (recessive model) was assessed by unpaired t-test. Association of LILRA3 expression with genotypic variants was analyzed by Kruskal-Wallis test. All statistical analyses were conducted using SPSS 13.0. P values less than 0.05 were considered significant.

RESULTS

Both variants (the 6.7-kb deletion and SNP rs103249) were in Hardy-Weinberg equilibrium (P > 0.05) in controls in both independent cohorts. In the 2 control groups, the allele frequencies of the LILRA3 deletion (74.6–78.0%) and rs103249 (T allele: 76.2–77.2%) were similar to the data from the HapMap CHB (Han Chinese in Beijing, China) population (http://hapmap.ncbi.nlm.nih.gov/) or the data from previous reports on Northeast Asians ([11, 20, 22]). The study had a statistical power of 0.999 to detect a modest effect size of OR 1.40, and a fixed MAF of 25% between LILRA3 and RA.

Strong association of functional LILRA3, but not nonfunctional LILRA3, with RA susceptibility in 2 independent Han Chinese cohorts

We first sought to determine whether there was an association between LILRA3 deletion and RA, by genotyping of the 6.7-kb deletion polymorphism in 1,618 cases and 1,658 control subjects in the N-Han cohort. The distribution of both allele and genotype frequencies is shown in Table 2. Unexpectedly, we observed significantly higher frequencies of the nondeleted (functional) LILRA3 in patients with RA compared with healthy controls, both in the allele model (27.2% versus 23.9%; P = 1.93 × 10−3, OR 1.19 [95% CI 1.07–1.33]) and in the genotype model (recessive model [+/+ versus +/− and −/−] 10.1% versus 6.3%; P = 4.01 × 10−5, OR 1.92 [95% CI 1.41–2.62]). To confirm the association of functional LILRA3 with RA, we genotyped one SNP known to be in strong LD with LILRA3 (rs103294; r2 = 0.83) ([22]). As seen in Table 2, SNP rs103294 displayed significant association with RA (allele model P = 5.17 × 10−3, OR 1.18 [95% CI 1.05–1.32]; genotype model P = 3.17 × 10−3, OR 1.65 [95% CI 1.18–2.29]).

Table 2. Analysis of association of LILRA3 and rs103294 with RA in 2 independent cohorts and in a combined meta-analysis, with adjustment for sex and age*
 ControlsRAPOR (95% CI)
  1. Values are the number (% [of 2n for the allele model and of n for the genotype model]); n values are the number of controls/number of rheumatoid arthritis (RA) patients. OR = odds ratio; 95% CI = 95% confidence interval; N-Han = Northern Han Chinese; − = 6.7-kb deletion; + = nondeletion; S-Han = Southern Han Chinese.
N-Han (discovery cohort)    
LILRA3 (n = 1,658/1,618)    
Allele model    
2,524 (76.1)2,355 (72.8)1.93 × 10−31.19 (1.07–1.33)
+792 (23.9)881 (27.2)  
Genotype model    
+/− and −/−1,553 (93.7)1,455 (89.9)4.01 × 10−51.92 (1.41–2.62)
+/+105 (6.3)163 (10.1)  
rs103294 (n = 1,636/1,618)    
Allele model    
T2,525 (77.2)2,401 (74.2)5.17 × 10−31.18 (1.05–1.32)
C747 (22.8)835 (25.8)  
Genotype model    
TC and TT1,543 (94.3)1,483 (91.7)3.17 × 10−31.65 (1.18–2.29)
CC93 (5.7)135 (8.3)  
S-Han (replication cohort)    
LILRA3 (n = 549/575)    
Allele model    
819 (74.6)778 (67.7)2.89 × 10−41.40 (1.17–1.69)
+279 (25.4)372 (32.3)  
Genotype model    
+/− and −/−516 (94.0)512 (89.0)4.00 × 10−21.66 (1.02–2.70)
+/+33 (6.0)63 (11.0)  
rs103294 (n = 380/465)    
Allele model    
T579 (76.2)631 (67.8)1.57 × 10−41.51 (1.22–1.88)
C181 (23.8)299 (32.2)  
Genotype model    
TC and TT356 (93.7)407 (87.5)3.92 × 10−21.80 (1.03–3.13)
CC24 (6.3)58 (12.5)  
Han Chinese (combined) (meta-analysis)    
LILRA3 (n = 2,207/2,193)    
Allele model    
3,343 (75.7)3,133 (71.4)4.65 × 10−61.29 (1.18–1.42)
+1,071 (24.3)1,253 (28.6)  
Genotype model    
+/− and −/−2,069 (93.7)1,967 (89.7)5.63 × 10−61.83 (1.41–2.37)
+/+138 (6.3)226 (10.3)  
rs103294 (n = 2,016/2,083)    
Allele model    
T3,104 (77.0)3,032 (72.8)1.15 × 10−51.25 (1.13–1.38)
C928 (23.0)1,134 (27.2)  
Genotype model    
TC and TT1,899 (94.2)1,890 (90.7)5.32 × 10−51.78 (1.35–2.36)
CC117 (5.8)193 (9.3)  

To validate the association of functional LILRA3 with RA, we conducted a replication study in an independent case–control cohort, by genotyping LILRA3 deletion and rs103294 in 575 cases and 549 control subjects in the S-Han cohort. As shown in Table 2, both functional LILRA3 and rs103294 showed consistent association with RA in the replication panel, both in the allele model (LILRA3 P = 2.89 × 10−4, OR 1.40 [95% CI 1.17–1.69]; rs103294 P = 1.57 × 10−4, OR 1.51 [95% CI 1.22–1.88]) and in the genotype model (LILRA3 P = 4.0 × 10−2, OR 1.66 [95% CI 1.02–2.70]; rs103294 P = 3.92 × 10−2, OR 1.80 [95% CI 1.03–3.13]). A meta-analysis of the discovery and replication panels provided compelling evidence that the functional LILRA3 conferred high risk for RA, under a fixed-effects model (allele model P = 4.65 × 10−6, ORcombined 1.29 [95% CI 1.18–1.42]; genotype model P = 5.63 × 10−6, ORcombined 1.83 [95% CI 1.41–2.37]). There was no evidence of heterogeneity between the 2 sample sets (allele model Phet = 0.30, I2 = 5%; genotype model Phet = 0.33, I2 = 0%) (data not shown).

Similar results were obtained by 2-marker haplotype analysis (r2 = 0.73): both haplotype C–non-del and haplotype T–non-del (rs103294–LILRA3) displayed strong risk effects contributing to RA susceptibility (P = 6.37 × 10−5, OR 1.24 [95% CI 1.11–1.37] and P = 8.55 × 10−4, OR 1.51 [95% CI 1.18–1.93], respectively). In contrast, haplotype T-del displayed a strong protective effect against RA (P = 2.08 × 10−7, OR 0.78 [95% CI 0.70–0.85]) (Table 3).

Table 3. Frequencies of rs103294–LILRA3 (non)deletion haplotypes*
HaplotypeControls, %RA patients, %χ2POR (95% CI)
  1. Variant order: rs103294–LILRA3 (non)deletion. See Table 2 for definitions.
C–non-del21.324.916.0156.37 × 10−51.24 (1.11–1.37)
T–non-del2.74.111.1288.55 × 10−41.51 (1.18–1.93)
T-del74.268.727.0322.08 × 10−70.78 (0.70–0.85)

Profound effect of functional LILRA3 on risk of RA susceptibility in males

We next investigated whether LILRA3 confers a sex-specific effect on RA susceptibility. Interestingly, following stratification by sex, we observed a much stronger association of functional LILRA3 with RA in males than in females, despite the decrease in statistical power for the analysis of male-only subjects. In the N-Han cohort, functional LILRA3 was associated with much higher risk in male patients compared with female patients (genotype model P = 4.58 × 10−7, OR 6.58 [95% CI 3.16–13.67] versus P = 5.16 × 10−2, OR 1.41 [95% CI 1.00–2.00]) (Table 4). In the S-Han cohort, the frequency of the functional LILRA3 allele among male patients was also higher than among controls (9.5% versus 5.6%), although the difference did not reach statistical significance due to the loss of power after sex stratification. Nevertheless, the significant male effect of functional LILRA3 was further confirmed in the combined analysis, by the much higher OR in male versus female subjects (genotype model P = 1.09 × 10−6, OR 4.47 [95% CI 2.45–8.17], versus P = 1.01 × 10−2, OR 1.46 [95% CI 1.09–1.95]) (Table 4).

Table 4. Association of LILRA3 with RA, stratified by sex, with adjustment for age*
 ControlsRAPOR (95% CI)
  1. Values are the number (% [of 2n for the allele model and of n for the genotype model]); n values are the number of controls/RA patients. See Table 2 for definitions.
Female    
N-Han (n = 1,226/1,101)    
Allele model    
1,852 (75.5)1,632 (74.1)0.271.08 (0.94–1.23)
+600 (24.5)570 (25.9)  
Genotype model    
+/− and −/−1,139 (92.9)1,001 (90.9)5.16 × 10−21.41 (1.00–2.00)
+/+87 (7.1)100 (9.1)  
S-Han (n = 422/457)    
Allele model    
634 (75.1)616 (67.4)3.59 × 10−41.46 (1.19–1.80)
+210 (24.9)298 (32.6)  
Genotype model    
+/− and −/−396 (93.8)405 (88.6)0.111.55 (0.90–2.65)
+/+26 (6.2)52 (11.4)  
Total (n = 1,648/1,558)    
Allele model    
2,486 (75.4)2,248 (72.1)2.81 × 10−31.19 (1.06–1.33)
+810 (24.6)868 (27.9)  
Genotype model    
+/− and −/−1,535 (93.1)1,406 (90.2)1.01 × 10−21.46 (1.09–1.95)
+/+113 (6.9)152 (9.8)  
Male    
N-Han (n = 411/267)    
Allele model    
642 (78.1)377 (70.6)1.79 × 10−31.49 (1.16–1.91)
+180 (21.9)157 (29.4)  
Genotype model    
+/− and −/−396 (96.4)234 (87.6)4.58 × 10−76.58 (3.16–13.67)
+/+15 (3.6)33 (12.4)  
S-Han (n = 126/105)    
Allele model    
183 (72.6)144 (68.6)0.341.22 (0.81–1.82)
+69 (27.4)66 (31.4)  
Genotype model    
+/− and −/−119 (94.4)95 (90.5)0.261.91 (0.62–5.87)
+/+7 (5.6)10 (9.5)  
Total (n = 537/372)    
Allele model    
825 (76.8)521 (70.0)1.17 × 10−31.42 (1.15–1.75)
+249 (23.2)223 (30.0)  
Genotype model    
+/− and −/−515 (95.9)329 (88.4)1.09 × 10−64.47 (2.45–8.17)
+/+22 (4.1)43 (11.6)  

Functional LILRA3 confers increased risk of development of ACPA-positive RA, especially in males

Following stratification by ACPA status, we found a significant association of functional LILRA3 with ACPA-positive RA (genotype model P = 3.05 × 10−4, OR 1.75 [95% CI 1.27–2.43]) (Table 5). No association between functional LILRA3 and ACPA-negative RA was observed (P = 0.28, OR 1.40 [95% CI 0.77–2.55]). Interestingly, after stratification by sex, we further observed that the association of functional LILRA3 with ACPA-positive RA appeared only in males, despite loss of power (P = 2.55 × 10−7, OR 5.95 [95% CI 3.02–11.71]), and was not observed in females (P = 0.19, OR 1.28 [95% CI 0.88–1.86]). Our data thus provide evidence of a significant male-specific effect of functional LILRA3 with regard to ACPA-positive RA.

Table 5. Association of LILRA3 with RA, stratified by ACPA status, with adjustment for sex and age*
Group (n)+/− and −/−+/+POR (95%CI), +/+ vs. +/− and −/−
  1. Values are the number (%); n values are the number of controls/RA patients. ACPA = anti–citrullinated protein antibody (see Table 2 for other definitions).
All    
Controls (2,207)2,059 (93.7)138 (6.3)
ACPA-positive RA (985)883 (89.6)102 (10.4)3.05 × 10−41.75 (1.27–2.43)
ACPA-negative RA (228)209 (91.7)19 (8.3)0.281.40 (0.77–2.55)
Female    
Controls (1,648)1,535 (93.1)113 (6.9)
ACPA-positive RA (770)699 (90.8)71 (9.2)0.191.28 (0.88–1.86)
ACPA-negative RA (180)163 (90.6)17 (9.4)0.291.43 (0.75–2.63)
Male    
Controls (537)515 (95.9)22 (4.1)
ACPA-positive RA (210)179 (85.2)31 (14.8)2.55 × 10−75.95 (3.02–11.71)
ACPA-negative RA (43)41 (95.3)2 (4.7)0.741.42 (0.18–11.58)

Functional LILRA3 confers increased risk of joint destruction in early ACPA-positive RA

We next examined whether functional LILRA3 confers a risk of joint destruction in RA, according to genotype, with stratification by disease duration and ACPA status. A total of 467 patients with available SHS data were divided into 3 groups, according to disease duration (<2 years, 2–10 years, or >10 years). Within the groups of ACPA-positive patients with disease duration of 2–10 years or >10 years, there was no difference in SHS according to the genotype of LILRA3 (recessive model). However, in the ACPA-positive group with disease duration of <2 years, the SHS was significantly higher among patients who were homozygous for the functional LILRA3 than among carriers of the nonfunctional LILRA3 (P = 0.0097) (Figure 1A). In general, the radiographic scores among ACPA-negative patients were moderate (mean SHS 55.0); however, there were too few ACPA-negative patients who were homozygous for functional LILRA3 (n = 6) to enable comparison of RA severity status according to LILRA3 genotype and disease duration in this subgroup. The effect of functional LILRA3 on radiographic severity was observed in both female and male patients with early ACPA-positive RA, but was more pronounced in males (P = 0.025 versus P = 0.052 [data not shown]). Our results suggest that functional LILRA3 is associated with radiographic severity of RA, especially in patients with early ACPA-positive disease.

Figure 1.

Association of the functional LILRA3 with joint destruction and LILRA3 mRNA expression in patients with rheumatoid arthritis (RA). A, Radiographic severity in anti–citrullinated protein antibody (ACPA)–positive RA patients grouped according to disease duration and LILRA3 genotype (+/+ versus −/− and +/−), as assessed with the modified Sharp/van der Heijde score (SHS). The modified SHS was significantly increased among patients with early RA (<2 years) who were homozygous for functional LILRA3. B, Expression of LILRA3 mRNA in freshly isolated peripheral blood mononuclear cells from ACPA-positive RA patients grouped according to LILRA3 genotype, as assessed by real-time quantitative polymerase chain reaction. Expression levels of LILRA3 were significantly increased among patients who were homozygous for functional LILRA3. Symbols represent individual patients; bars show the mean. P values were determined by unpaired t-test (A) or Kruskal-Wallis test (B).

Association of the functional LILRA3 allele with LILRA3 mRNA expression

Given the above-described findings, we further investigated whether the functional LILRA3 allele was associated with LILRA3 mRNA expression in RA patients. As shown in Figure 1B, levels of LILRA3 expression were significantly increased among individuals who were homozygous for functional LILRA3, compared with carriers of the nonfunctional LILRA3 (P = 3.31 × 10−8 by Kruskal-Wallis test). In accordance with this, following stratification by ACPA status a similar association pattern was observed among patients with ACPA-positive RA (P = 0.0015 by Kruskal-Wallis test [data not shown]). There were too few ACPA-negative patients who were homozygous for functional LILRA3 (n = 3) to enable comparison of LILRA3 mRNA expression according to LILRA3 genotype in this subgroup.

DISCUSSION

Although genetic associations of nonfunctional LILRA3 (6.7-kb deletion) with MS and SS have been reported previously, to our knowledge this is the first report of an association of functional LILRA3 with an autoimmune disease, i.e., RA, with confirmation in a replication cohort. SNP rs103294, in strong LD with LILRA3, was also associated with RA. Specifically, our study clearly showed that functional LILRA3 is highly associated with increased susceptibility to RA in males and confers risk for development of ACPA-positive RA. The significant elevation of radiographic scores in patients with early RA who were homozygous for functional LILRA3, as well as the notably increased level of LILRA3 transcripts in LILRA3 carriers, support the notion of a functional role of LILRA3 in regulation of RA susceptibility and severity.

As noted above, associations of nonfunctional LILRA3 with susceptibility to MS and SS in Caucasians have been demonstrated previously ([15, 17]). However, despite the much higher frequency of the nonfunctional LILRA3 allele in Northeast Asians, the prevalence of MS in this population is much lower than in Caucasians ([23]). Of note, a recent genome-wide association study showed that the functional LILRA3 allele, but not the nonfunctional allele, was significantly associated with susceptibility to prostate cancer in the Han population. The functional LILRA3 allele was more common in cases (0.307) than in controls (0.252) in that study ([22]). This may indicate that in our population, the leukocyte immunoglobulin-like receptor A3 (LILRA3) molecule might serve as a risk factor, rather than as a modulating factor, contributing to the pathogenesis of chronic inflammation. However, in the present study we showed that SNP rs103294, tagging LILRA3, was significantly associated with RA as well. Thus, we currently cannot exclude the possibility that rs103294 might be an independent risk factor for RA, though the SNP might be a nonfunctional variant since it is not annotated as any promoter, enhancer, repressor, or distant regulatory element (http://www.ncbi.nlm.nih.gov; MAF source 1000 Genomes). Further functional studies will hopefully elucidate the role of the functional LILRA3 allele and SNP rs103294 in autoimmune diseases.

Although the mechanism(s) underlying this genetic association remains elusive, LILR molecules are broadly expressed on immune cells and may play an important role in immune responses ([24]). Recombinant LILRA3 was able to induce T cell proliferation in response to tumor cells by inhibition of CD8+ regulatory T cells ([25]). Moreover, in a recent investigation, LILRA3 expression was significantly increased in patients with RA ([26]). In the present study, expression of LILRA3 was undetectable in RA patients who were homozygous for LILRA3 deletion and was reduced in heterozygous carriers, compared with RA patients with 2 copies of LILRA3. In addition, we have recently found that the functional LILRA3 allele is a novel genetic risk factor for systemic lupus erythematosus as well as primary SS; functional LILRA3 appears to highly predispose to certain phenotypes of both disorders in the Han population (Du Y, et al: unpublished observations). Thus, we speculate that the presence of LILRA3 alleles may lead to excessive activation of the immune system and, consequently, to increased susceptibility to autoimmune disorders and disease activity or severity. Future genetic association studies in other autoimmune diseases and functional characterization of this 6.7-kb deletion polymorphism are warranted in order to fully understand its contribution to autoimmune diseases.

Several sex-specific RA genetic associations have been reported earlier. Interestingly, in the present study we observed a profound impact of functional LILRA3 on RA susceptibility in males. The mechanism for this sex-specific effect is unknown, but the striking male effect of functional LILRA3 supports the notion that there may be specific genetic predisposition factors for RA in males ([7, 27]). An alternative hypothetical explanation for the present findings and above-mentioned findings of others could be that the functional LILRA3 allele is associated with susceptibility to certain chronic disorders, e.g., RA and prostate cancer, in a sex-specific manner in Asians. It would be interesting to investigate the contribution of functional LILRA3 to other diseases, especially those with sex-specific incidence and clinical manifestations, in Asian populations.

Identification of genetic risk factors responsible for clinical subsets of heterogeneous diseases is important for the understanding of disease pathogenesis. It has been demonstrated that certain genetic associations are restricted to ACPA-positive RA, e.g., HLA–DRB1 shared epitope–encoding alleles and the PTPN22 R620W variant. In the present study, we also observed an association of functional LILRA3 with ACPA-positive RA, especially in males. We further demonstrated that functional LILRA3 had an impact on joint destruction, especially in early ACPA-positive RA. Interestingly, Wisniewski et al have recently shown that patients carrying the functional LILRA3 allele developed MS at a significantly younger age than patients carrying the deleted LILRA3 allele in the Polish population ([28]). Our results indicate that the functional LILRA3 allele likely plays an additional role in disease progression, beyond its role in RA susceptibility. However, in the present retrospective study cohort, we were able to collect radiographic data on only 467 patients. Although the baseline characteristics of the patients with available radiographic records were similar to those of the entire case cohort (data not shown), given the modest sample size and the low frequency of functional LILRA3 allele the study may have been underpowered to detect minor effects on joint destruction. Additional investigations with larger sample sizes are needed to confirm the findings.

In summary, the present study provides the first evidence that the functional LILRA3 allele is a genetic risk factor for RA susceptibility, especially in males. Functional LILRA3 appears to highly predispose to ACPA-positive RA and to confer increased severity of early RA. These findings may improve the understanding of RA genetic susceptibility and provide clues for avenues to be pursued in further functional studies. Our findings also highlight the importance of genetic heterogeneity of complex diseases in different ethnic groups.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Drs. Guo and Li had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Guo, Li.

Acquisition of data. Du, Cui, Xia Liu, Hu, Yang, Wu, Xu Liu, Ma, Zuo, Sheng, Xiangyuan Liu, Xu, Zhu, Sun, Hong, Zhang, Guo, Li.

Analysis and interpretation of data. Du, Guo, Li.

Acknowledgments

We thank the N-Han Study Group members and staff for recruiting patients and healthy controls and the Department of Rheumatology and Immunology, Peking University People's Hospital for providing technical assistance. We also thank the S-Han Study Group members and staff for recruiting patients and healthy controls from the Department of Rheumatology, Number1 Hospital, Anhui Medical University and from the Department of Rheumatology, Gulou Hospital, Nanjing University, Jiangsu, China. We are grateful to the patients and healthy volunteers for their cooperation and for consenting to participate in the study.

Ancillary