Inhibitory KIR3DL1 Alleles are Associated with Psoriasis

DEAR EDITOR

Genetic association studies have implicated over 40 genetic loci in the pathogenesis of psoriasis, with the largest signal observed at the MHC Class I locus for HLA-C; however, recent data also suggest a lesser but significant pathogenic role for HLA-B^1,2. Although both of these molecules participate in adaptive immunity through antigen presentation, they can also regulate the innate immune response through interaction with killer cell immunoglobulin-like receptors (KIR) expressed on the surface of both natural killer (NK) and a subset of T cells. KIRs are a family of receptors encoded by a cluster of 14 genes on chromosome 19q13.4. KIRs may encode activating or inhibitory receptors that bind the Bw4 motif on HLA-B or C1/C2 motif on HLA-C³.

Previous studies have shown that the presence of activating KIR3DS1⁴ and KIR2DS1⁵ are associated with increased susceptibility for psoriasis and psoriatic arthritis^6,7, results consistent with a model wherein expression of activating KIRs increases risk of developing immune-mediated diseases³. However, whether decreased expression of inhibitory KIRs is also associated with increased risk of psoriasis has not been fully explored.

The inhibitory KIR3DL1 is among the most polymorphic of all KIR loci (http://www.ebi.ac.uk/ipd/kir/stats.html). Cell surface expression of KIR3DL1 varies greatly between alleles and alleles can be classified into three allotypes: KIR3DL1*NULL (no cell surface expression: *004 and *019), KIR3DL1*LOW (low-expression: *005, *007, and *053), and KIR3DL1*HIGH (high-expression: *001, *002, *008, *009, *015, *020, *029, and *035)^8,9. The ligand for KIR3DL1 is the Bw4 epitope present on several HLA-B molecules such as HLA-B*57 and HLA-B*27, as well as HLA-A*24¹⁰.

To investigate the association of KIR3DL1 and HLA-Bw4 with psoriasis, we carried out genotyping of KIR3DL1 and HLA-B in a cohort of 314 subjects (203 psoriasis cases and 111 healthy controls) of European descent recruited from the United States. Diagnosis of psoriasis was established by a dermatologist. The average age of onset of psoriasis was 19 years, 54% were female, and approximately 23% were diagnosed with psoriatic arthritis. Controls were healthy volunteers without a history of autoimmune disease, with an average age of 39.0 and were 51.4% female.

KIR3DL1 typing was performed by sequencing exons 3, 4, 5, and 9 of KIR3DL1. Primers and PCR conditions are shown in Supplementary Table 3. KIRDL1 alleles were called from the sequencing data using Assign software (Conexio Genomics, Fremantle, Australia). HLA typing of cases was performed as previously described¹¹. HLA typing of controls was performed with the PCR-sequence based typing (PCR-SBT) method recommended by the 13th International Histocompatibility Workshop (available at: http://www.ihwg.org). HLA sequences were analyzed using Assign software (Conexio Genomics). In a minority of cases with missing HLA-B typing (n = 18), HLA-B genotypes were imputed with genome-wide association data (Supplementary Materials online).

Based on our previous work¹² and empirical data measuring cell surface expression of KIR3DL1¹³, individuals carrying at least one *HIGH allele but no *LOW allele were considered KIR3DL1*HIGH carriers, while individuals carrying at least one *LOW allele were considered KIR3DL1*LOW carriers. If an individual carried at least one *NULL allele and no copy of either a *LOW or *HIGH allele, they were considered a *NULL carrier. 32 individuals homozygous for KIR3DS1 (who do not carry any copies of KIR3DL1) were excluded from analysis. HLA-B genotypes were grouped into either HLA-Bw4 or HLA-Bw6.

First, we tested KIR3DL1*LOW, KIR3DL1*HIGH, and KIR3DL1*NULL functional genotypes for association with psoriasis using a univariate logistic regression model (Supplementary Materials online). All p-values were adjusted for multiple tests by the Benjamini-Hochberg false discovery rate (FDR) procedure. Statistically significant associations were observed with the KIR3DL1*LOW (FDR = 0.05) and *NULL (FDR = 0.04) genotypes (Table 1A). The *LOW genotype associated with increased risk of psoriasis (OR = 1.68) while the *NULL genotype associated with a protective effect (OR = 0.41).

TABLE 1.

A. Association analysis of KIR3DL1 genotypes
Genotype	OR (95% CI)	P value	FDR	Ncases (%)1	Ncontrols (%)2
3DL1*LOW	1.68 (1.02–2.79)	0.04	0.05	78/203 (38.4%)	30/111 (27.0%)
3DL1*HIGH	0.92 (1.02–2.79)	0.71	0.71	109/203 (53.7%)	62/111 (55.9%)
3DL1*NULL	0.41 (1.02–2.79)	0.01	0.04	16/203 (7.9%)	19/111 (17.1%)

B. Association analysis of HLA-B genotypes (adjusted for HLA-C*0602)
Genotype	OR (95% CI)	P value	FDR	Ncases (%)1	Ncontrols (%)2
HLA-Bw4	1.27 (0.70–2.27)	0.43	0.57	140/178 (78.7%)	68/102 (66.7%)
HLA-Bw6	0.79 (0.44–1.42)	0.43	0.57	38/178 (21.3%)	34/102 (33.3%)

C. Association analysis of compound KIR3DL1 and HLA-Bw4 genotypes (adjusted for HLA-C*0602)
Genotype	OR (95% CI)	P value	FDR	Ncases (%)1	Ncontrols (%)2
3DL1*LOW + Bw4	2.50 (1.32–5.00)	6.80×10−3	0.02	56/178 (31.5%)	14/102 (13.7%)
3DL1*HIGH + Bw4	0.73 (0.43–1.22)	0.23	0.48	72/178 (40.4%)	45/102 (44.1%)
3DL1*NULL + Bw4	0.65 (0.25–1.71)	0.37	0.48	12/178 (6.7%)	9/102 (8.8%)

Abbreviations: OR, odds ratio; CI, confidence interval.

¹Proportion of cases carrying given genotype.

²Proportion of controls carrying given genotype.

Next, we tested the ligand for KIR3DL1, HLA-Bw4, for association with psoriasis with and without adjustment for HLA-C*0602, the strongest known psoriasis risk allele in the MHC region¹. In the unadjusted analysis, Bw4-positive genotypes were associated with increase in risk for psoriasis (OR = 1.79, FDR = 0.05), while Bw6 genotypes (those lacking Bw4) were associated with reduced risk of psoriasis (OR = 0.56, FDR = 0.05). However, after adjustment for HLA-C*0602, which is commonly found on the same haplotype as the Bw4-positive allele HLA-B*5701⁴, neither Bw4 nor Bw6 were significantly associated with psoriasis (Table 1B).

Finally, we performed association testing of the compound genotypes 3DL1*LOW + HLA-Bw4, 3DL1*HIGH + HLA-Bw4, and 3DL1*NULL + HLA-Bw4 with adjustment for HLA-C*0602. This revealed a statistically significant association with the 3DL1*LOW + HLA-Bw4 compound genotype, with a significant increase in risk for psoriasis (OR = 2.50, FDR = 0.02) (Table 1C). A multivariate logistic regression model was used to test for evidence of multiplicative interaction between the 3DL1 genotypes and the HLA-Bw4 genotypes and found suggestive evidence for epistasis between 3DL1*LOW and -Bw4 (β = 1.14, p = 0.07).

Overall, our results suggest that low cell-surface expression of KIR3DL1 in the presence of the HLA-Bw4 epitope is associated with an increased risk for developing psoriasis. This result is consistent with a model whereby a reduction in inhibitory signal in NK or T cells results in a heightened immune cell response³. We also found that individuals with the KIR3DL1*NULL genotype were protected from psoriasis irrespective of HLA-Bw4 status, suggesting that *NULL alleles may have intracellular effects independent of cell surface expression. A genetic study of ankylosing spondylitis also found that KIR3DL1*NULL alleles were protective of disease development¹⁴. While our sample size is relatively small, a power calculation revealed that our study has 80% power to detect an odds ratio of 2.0 or higher under a dominant model (see Supplementary Materials online). Our work—while requiring replication—provides additional evidence that NK cell receptors and ligands may be important in psoriasis susceptibility, adding to a body of literature that has suggested the importance of other NK cell receptors such as KIR2DS1 and KIR3DS1, as well as ligands such as MICA¹⁵.