Associations of interactions between NLRP3 SNPs and HLA mismatch with acute and extensive chronic graft-versus-host diseases

HLA matching is a well-known genetic requirement for successful bone marrow transplantation (BMT). However, the importance of non-HLA single-nucleotide polymorphisms (SNPs) remains poorly understood. The NLR family pyrin domain–containing 3 (NLRP3) inflammasome, a key regulator of innate immunity, is associated with multiple diseases. We retrospectively genotyped SNPs of NLRP1–3 and caspase recruitment domain family member 8 (CARD8), which are implicated in the interleukin 1β (IL-1β) signaling, in 999 unrelated BMT donor–recipient pairs. We identified an association of the interaction between the recipient NLRP3 SNP CC genotype and total HLA mismatches with grade 2–4 acute graft-versus-host disease (AGVHD), and an association of the interaction between the donor NLRP3 SNP T allele and HLA-C mismatch with extensive chronic GVHD (ECGVHD), in both adjusted and unadjusted regressions (P < 0.005). Importantly, the ECGVHD risk associated with HLA-C mismatch was not elevated when the donor NLRP3 genotype was CC. We also identified an association of the interaction between recipient NLRP3 SNP and donor cytomegalovirus seropositivity with overall survival in adjusted regressions (P < 0.005). These results suggest the importance of certain SNP–covariate interactions in unrelated BMT. The three identified interactions may be useful for donor selection or outcome prediction.

. Characteristics of 999 donor-recipient pairs who underwent unrelated BMT ! CMV, cytomegalovirus; Ara-C, arabinofuranosyl cytidine; GVH, graft-versus-host; HVG, host-versus-graft; N.A., not applicable. *Groups 1, 2, and 3 are BMT pairs of malignant-disease patients without previous transplantation, non-malignant disease patients without previous transplantation, and patients who underwent previous transplantation, respectively. †E.g., aplastic anemia. ‡These two subcategories are merged into one subcategory to enable/simplify multivariable analyses. §Tacrolimus was administered to 594 of the 596 CyA non-users and in 6 of the 225 CyA users in Group 1. ¶10 8 per kg body weight.  Sex  ABO blood type  Female  336  26  42  Female  260  22  33  Match  464  38  65  Male  486  39  70  Male  562  43  79  Mismatch  358  27  47 Age           Table 2 for other notations.            Table   2 for other notations.          The upper panel shows allelic discrimination plot from one representative run of the TaqMan assay for each SNP drawn by the analysis software (Applied Biosystems). The red, blue, and green dots indicate the two homozygous genotypes and the heterozygous genotype, respectively. The X and black box represent the failed samples and the no-DNA control, respectively. The lower panel shows confirmation of TaqMan assays by PCR direct sequencing. Eight to 11 samples of each of the three genotypes successfully determined by the TaqMan assays were subsequently re-genotyped by PCR direct sequencing. The arrowhead indicates the position of each SNP.

Details of SNP selection
We chose two NLRP3 SNPs (Supplementary Table S2). One is an intronic SNP, rs4612666, which is associated with susceptibility to food-induced anaphylaxis and aspirin-induced asthma in Japanese individuals 2 . The C allele creates a putative GATA-2 binding sequence, induces 1.2-fold higher expression of a downstream reporter gene, and affects an electrophoretic mobility shift assay pattern in THP-1 cells 2 . The other SNP is rs10925027, which is located downstream of both the NLRP3 and OR2B11 genes. This SNP was associated with relapse in an HLA-identical sibling HSCT study 4 .
Although the molecular function of this SNP, if any, remains unknown, it is in strong LD in JPT104 from the 1000 Genomes Project 5 with a known functional SNP, rs10754558, which affects the stability of NLRP3 mRNA 2 . The NLRP1 SNP rs11651270 causes a missense substitution that is functionally important for autoproteolysis and IL-1! secretion, and is also associated with autoimmune diseases 1,6 .
The CARD8 SNP rs2043211 causes a nonsense substitution that abrogates inhibition of NF-"B activity by CARD8 and is associated with disease severity in rheumatoid arthritis patients 3 . The NLRP2 SNP rs1043673 causes a missense substitution of unknown molecular functional consequence and is associated with the harmful effects of arsenic 7 . This SNP was associated with NRM in a univariable analysis in the sibling HSCT study, and is in 100% LD in JPT104 with rs1043684, which was also associated with NRM and OS in the same study 4 .

Details of SNP genotyping by direct sequencing
Short genomic regions spanning the two SNPs were amplified from 10 ng each of the DNAs by PCR using ExTaq HS (Takara), with the pairs of primers listed in Supplementary Table S2. The PCR products were treated with ExoSAP-IT (USB). DNA sequencing reactions were carried out using one of these primers and the BigDye Terminator v3.1 Cycle Sequencing kit, which was then read on the 3730xl DNA Analyzer (Applied Biosystems) at Eurofins Genomics (Tokyo). After visual inspection of each chromatogram, the ratio of peak amplitudes of the primary and secondary bases was calculated at each SNP nucleotide. The genotype was judged to be heterozygous for the primary and secondary bases if the ratio of the two peaks was less than 3, and otherwise judged as homozygous for the primary base.