Donor and recipient genetic variants in NLRP3 associate with early acute rejection following kidney transplantation

NLRP3 (NOD-like receptor family, pyrin domain containing 3) is a member of the inflammasome family and is of special interest in renal disease. Experimental studies have shown that Nlrp3 plays a significant role in the induction of renal damage and dysfunction in acute and chronic renal injury. However, the role of NLRP3 in human renal disease is completely unknown. From a retrospective cohort study, we determined in 1271 matching donor and recipient samples if several NLRP3 single nucelotide polymorphisms (SNPs) were associated with primary non-function (PNF), delayed graft function (DGF), biopsy-proven acute rejection (BPAR) and death-censored graft and patient survival. NLRP3 gain-of-function SNP (rs35829419) in donors was associated with an increased risk of BPAR while NLRP3 loss-of-function SNP (rs6672995) in the recipient was associated with a decreased risk of BPAR in the first year following renal transplantation (HR 1.91, 95% CI 1.38–2.64, P < 0.001 and HR 0.73, 95% CI 0.55–0.97, P = 0.03 resp.). NLRP3 SNPs in both donor and recipient were not associated with PNF, DGF, graft survival or patient survival. We conclude that genetic variants in the NLRP3 gene affect the risk of acute rejection following kidney transplantation.

to the host's infiltrating leukocyte-associated NLRP3, respectively. So far it is known that NLRP3 SNPs with a gain-of-function (GOF) are related to several auto-inflammatory disorders due to an enhanced inflammatory state in these patients [13][14][15][16] . Currently, it is unknown if NLRP3 SNPs are related to human renal diseases or patient outcome after solid organ transplantation 5 . Therefore, we determined NLRP3 SNPs in a large cohort of matching donors and recipients and investigated their association with renal allograft and patient outcomes.

NLRP3 SNP distribution in donors and recipients.
Baseline characteristics of donors and recipients are displayed in Table 1. In our cohort, genotypic distribution of NLRP3-related SNPs in donor and recipient were comparable to previous general/Caucasian population distributions (Supplementary Table S1 and www.ensemble.org). None of the SNPs deviated significantly from a Hardy-Weinberg equilibrium (all P > 0.46 after Holm-Bonferroni correction). In the combined model (A/A vs. A/a + a/a) genotype distribution of NLRP3 SNPs were comparable between donors and recipient and SNPs were not considered as a potential risk factor for diseases of the native kidneys. SNPs with minor allele frequency less than 1% were left out from further analysis (N = 3).
NLRP3 SNPs are not associated with delayed graft function, primary non-function, graft survival or patient survival. Delayed graft function (DGF) occurred in 415/1271 patients (33%) of which 60/415 resulted in primary non-functioning (PNF) of the graft (14% of DGF, 5% of total patients). None of the 3 NLRP3 SNPs in either donor or recipient were correlated to the occurrence of DGF or PNF ( Table 2). After stratification by donor type (living and cadaveric), again none of the NLRP3 SNPs were significantly associated with DGF or PNF (Supplementary Table S2). Median overall graft survival was 5.5 years (interquartile range 2.9-8.9 years). The overall cumulative incidence of death-censored graft failure was 215/1271 (17%) of which 136/215 due to rejection (63% of graft failure, 11% of total). We did not observe a significant association between any of the NLRP3 SNPs, in either donor or recipient, with death-censored graft failure ( Table 2). The cumulative incidence of patient mortality was 220/1271 (17%). Donor and recipient NLRP3 SNPs were not associated with patient survival (Table 2).

Discussion
The NLRP3 inflammasone is of special interest in renal diseases because of its expression pattern in murine and human kidneys and its detrimental role in experimental models of acute and chronic renal injury [8][9][10][11][12] . Thus far, nothing is known about the role of NLRP3 in human renal diseases or solid organ transplantation. Determination of NLRP3 SNPs in donor and recipient DNA offers a unique opportunity to characterize the differential effects of transplant-and the host's leukocyte-associated NLRP3 inflammasome. We quantified several NLRP3 SNPs in a large cohort consisting of > 1200 matched donors and recipients. We found that the NLRP3 gain-of-function SNP rs35829419 in donors associated with an increased risk of acute rejection, whereas on the contrary that the NLRP3 loss-of-function SNP rs6672995 in recipients associated with a reduced risk of rejection, particularly those episodes that occurred within the first year after transplantation. NLRP3 SNPs became of particular interest in human disease because of their association and causal role in several autoinflammatory disorders. Gain-of-function mutations in the NLRP3 gene (R260W and Q705K) lead to enhanced basal IL-1β production and are linked to Muckle-Well syndrome, familial cold autoinflammatory syndrome and chronic infantile neurologic cutaneous and articular syndrome [13][14][15][16] . Moreover, NLRP3 SNP Q705K is associated with Crohn's disease, celiac disease and late-onset Alzheimer disease [17][18][19] . In contrast, OR2B11/NLRP3 SNP rs4353135 and NLRP3 SNP rs6672995 lead to a reduction in basal NLRP3 expression and rs6672995 homozygous variant a/a affects IL-1β production. Both SNPs are therefore considered to cause a loss of the inflammasome's function 20 . So far, knowledge about the role of NLRP3 in renal diseases is largely acquired from murine studies, where an important role for both tissueand myeloid-related Nlrp3 in murine IRI was observed [8][9][10] . The literature suggests that the pathophysiological role of Nlrp3 in murine renal IRI is in part inflammasome-independent 8,9,[21][22][23] and might also be related to TGF-β signaling 24 .
We observed an increased risk of rejection with the donor gain-of-function NLRP3 variant Q705K (rs35829419) and a reduced risk of rejection in recipients carrying the NLRP3 A71T variant (rs6672995), especially within the first year following transplantation. It is known that episodes of acute rejection that occur early after transplantation are less detrimental to the graft than those that occur at later time-points. Chronic rejection, especially the antibody-mediated variant, is more challenging to treat. Since we observed an association of NLRP3 SNPs with early, but not late BPAR, this could explain why we did not identify an association with graft loss. Little is known about the role of NLRP3 in adaptive immunity. Activation of the NLRP3 inflammasome is crucial in IL-1β and IL-18 maturation that in turn, besides their role in innate immunity, play an important part in priming of adaptive (allo)immunity. Both cytokines are involved in development and differentiation of CD4 + T-helper lymphocytes (reviewed in refs 4 and 25). Nlrp3 is known to boost adaptive immunity in an inflammasome-dependent manner in respiratory failure 26 , but Nlrp3 deficiency does not affect alum-mediated adjuvant activity in mice challenged with human serum albumin vaccination 27 . To date, it is unknown if NLRP3 is differentially involved in cellular compared to humoral immunity in humans. How gain-and loss-of-function variants in the NLRP3 gene affect acute rejection remains speculative and is an area of future study. A limitation of our study is the lack of a standardized assays for the determination of donor-specific antibodies over the years and we therefore could not investigate whether the association between the NLRP3 variants differed for T cellor antibody-mediated rejection. We were also unable to directly measure the effect of the NLRP3 variants on inflammasome-related cytokine production. The current study is the first to describe the association between the NLRP3 variants and acute rejection and we therefore acknowledge that other cohort studies should validate the generalizability and transportability of our findings.
Interestingly, the association of the NLRP3 gain-of-function specifically in the donor and NLRP3 loss-of-function specifically in the recipient with the development of rejection suggests that NLRP3 could be a double-edged sword in renal transplantation. NLPR3 is known to be expressed in human myeloid and lymphoid cells, but also in cells of the renal parenchyma, including tubular epithelial cells 9,11,28,29 . Tubular epithelial cells express NLRP3, IL-1β and IL-18 and a gain-of-function of NLRP3 could result in elevated levels of these cytokines leading to (local) chronic inflammation and chronic kidney injury 11,[13][14][15][16]30 . Additionally, tubular epithelial NLPR3 could also contribute to kidney fibrosis directly. In experimental murine studies for instance, tubular epithelial Nlrp3 was involved in epithelial-mesenchymal transmission (EMT) via TGF-β signaling, which is believed to precede the development of interstitial fibrosis and tubular atrophy 11 . However, besides parenchymal cells, one should keep in mind that also immune cells like resident dendritic cells of donor origin are being transplanted and subsequently present alloimmune epitopes in draining lymph nodes and a gain-of-function of NLRP3 in these cells could enhance direct allorecognition, resulting in early rejection. NLRP3 loss-of-function in recipients on the other hand could alter IL-1β , IL-18 and IL-33 production by (infiltrating) leukocytes upon activation, which was shown to be the case for IL-1β 20 . This could lead to a dampened alloimmune response after transplantation. Further experimental studies are required to determine how NLRP3 is involved in alloimmune recognition.
Given the important role of Nlrp3 in murine models of IRI, one would expect a role for the human NLRP3 gene in delayed graft function. However, neither donor nor recipient NLRP3 SNPs associated with delayed graft function or primary nonfunction. This lack of association could be explained by several aspects. Although murine ischemia-reperfusion injury is also characterized by tubular necrosis, there are some distinct differences with delayed graft function, including differences in circulating leukocyte composition, a slightly different renal anatomy, hemodynamic response and the lack of immunosuppressive drugs in the murine IRI model, making a direct extrapolation of murine data to the human clinical setting not always possible 31 . In addition, delayed graft function is a clinical term describing the need for additional renal replacement therapy, which could have other causes   besides IRI-related acute kidney injury rejection, drug toxicity and hyperkalemia 32 making it a heterogeneous surrogate end-point. In our study, we grouped A/a and a/a genotype together based on low numbers of the latter group but we do not know if A/a is phenotypically different from a/a and how it would affect our study. In donors, we did identify patients with a homozygous recessive rs35829419 variant. For this variant, Verma et al. showed that in vitro, the rs35829419 mutant monocytes had a significantly higher level of IL1-beta protein at baseline or after stimulation with the pro-pyroptotic substance alum (adjuvant) 16 . Additionally, Roberts et al. observed that in the homozygous recessive allele combination, there was an increased plasma concentration of IL1-beta 33 . It is difficult to extrapolate these results to our the findings in our cohort: (1) there was no heterozygous testing in the in vitro experiments by Verma et al., (2) they observed that the excretion of IL1-beta was only partly rescued by a pan-caspase inhibitor, indicating potential non-canonical, pyroptotic signalling in their experiments, (3) both studies measured the full IL1-beta and IL18 protein and not the active forms that are cleaved by the inflammasome complex. The loss of function NLRP3 variant rs6672995 was shown to correlate with the NLRP3 expression on the mRNA level 34 . They did not find an association with plasma concentrations of IL1-beta in patients with this variant. In line with this report, Villani et al. found that patients who were homozygous for the rs6672995 variant had a lower  production of IL1-beta as compared to heterozygous patients 20 . Again no active IL1-beta (or other inflammasome related cytokines), pyroptosis and in this case no patients with a homozygous dominant variant were analyzed and we can therefore not extrapolate these results to our cohort. In summary, we showed that various gain-and loss-of-function genetic variants in the NLRP3 gene differentially associate with biopsy-proven acute rejection, specifically within the first year after renal transplantation. These findings imply a complex role for NLRP3 in early allorecognition and rejection, which makes timing of interventions that aim at promoting or reducing NLRP3 inflammasome activity crucial.

Material and Methods
Patient genotyping. Samples were collected from a retrospective cohort as described before 35,36 . Between March 1993 and February 2008, 1271 matching donor and recipient peripheral blood mononuclear cells (PBMCs) were obtained from all available patients who underwent kidney transplantation at the University Medical Center Groningen, The Netherlands. The exclusion criteria were: cases of re-transplantation, combined kidney/pancreas or kidney/liver transplantation, technical problems during surgery, the unavailability of DNA and loss of the patients on follow-up (Fig. 2). The institutional ethical review board of the University Medical Center Groningen approved the study (METc 2014/077). Written informed consent was obtained from all patients. None of the living transplant donors were from a vulnerable population and all living donors provided written informed consent. In case of deceased donation, the donors provided informed consent when they registered their donation status and by law, no additional consent was needed. The study was conducted according to the principles of the declaration of Helsinki. Non-synonymous NLRP3 SNPs were chosen from NCBI (2010). In addition, we chose the NLRP3/OR2B11 rs4353135 SNP, located downstream of NLRP3 and upstream of olfactory receptor gene OR2B11, which has previously been shown to influence NLRP3 expression levels 20 . Genotyping of the selected SNPs (Supplementary Table 1) was performed using the Illumina VeraCode GoldenGate Assay kit (Illumina, San Diego, CA, USA) according to the manufacturer's instructions. Genotype clustering and calling were performed using Beadstudio Software (Illumina). The selected SNPs were not in linkage disequilibrium (r 2 < 0.10; SNP Annotation and Proxy Search, Broad institute).
DNA isolation and quality control. Peripheral blood mononuclear cells were used to acquire donor and recipient DNA. DNA samples were validated for DNA concentrations using absorbance at 260 nm with a spectrophotometer (ND-1000, NanoDrop) and purity was assessed by 260/280 and 260/230 absorbance ratios. In case of impurity of the sample, repeated isolation attempts were conducted.
Study end-points. The end-points used in this study were: delayed graft function (DGF, defined as the requirement for dialysis within the first week after transplantation), primary non function (PNF, defined as nonfunctioning of the allograft from transplantation on), time to the first episode of clinical biopsy-proven acute rejection (BPAR), death-censored graft failure (defined as the need for dialysis or re-transplantation) and patient mortality.
Statistics. Statistical analyses were performed using R version 2.15.2 for Macintosh (www.r-project.org).
Two-sided P-values below 0.05 were considered statistically significant. As a routine data quality control, the Hardy-Weinberg equilibria for the SNPs were tested. Univariate P-values were subjected to Holm-Bonferroni correction for multiple testing. For the association with DGF and PNF, logistic regression models were constructed. For the association with BPAR, (rejection-mediated) death-censored graft failure and patient survival, time-to-event models were created by Cox proportional hazards regression. Data were considered missing (completely) at random and we therefore performed complete-case multivariable regression on 1014/1271 (79.8%) of patients with complete multivariate data available. To validate the effect of missing values on our multivariate models, we reperformed the Cox regression analyses on M = 20 imputed datasets by multiple imputations by chained equations (MICE). The imputation results from predictive mean matching were visually checked by Kernel density plots. Multivariate Cox models were subjected to 10x internal cross-validation and the decline in slope in the validation cohort was considered as an overestimation measure. Kaplan-Meier survival curves were generated to visualize the association between the genotypes of each SNP in donor and recipient and the time to BPAR. Due to the low numbers of patients with homozygous recessive (a/a) variants, we used co-dominant models in which we compared homozygous dominant (A/A) versus heterozygous + homozygous recessive (A/a + a/a) individuals.