Intergenic Interactions of SBNO1, NFAT5 and GLT8D1 Determine the Susceptibility to Knee Osteoarthritis among Europeans of Russia

This study was conducted to examine the associations between genome-wide association studies (GWAS)-important single nucleotide polymorphisms (SNPs) and knee osteoarthritis (KOA) among Europeans of Russia. The present replicative study (“patient-control” design has been used) was carried out on 1000 DNA samples from KOA (n = 500) and KOA-free (n = 500) participants. Ten GWAS-important for KOA SNPs of eight candidate genes (LYPLAL1, GNL3, GLT8D1, SBNO1, WWP2, NFAT5, TGFA, GDF5) were studied. To assess the link between SNPs and KOA susceptibility, logistic regression (to establish independent SNP effects) and MB-MDR (to identify SNP–SNP interactions) were used. As a result of this genetic analysis, the associations of individual SNPs with KOA have not been proven. Eight loci out of ten tested SNPs interacted with each other (within twelve genetic models) and determined susceptibility to KOA. The greatest contribution to the disease development were made by three polymorphisms/genes such as rs6976 (C>T) GLT8D1, rs56116847 (G>A) SBNO1, rs6499244 (T>A) NFAT5 (each was included in 2/3 [8 out 12] KOA-responsible genetic interaction models). A two-locus epistatic interaction of rs56116847 (G >A) SBNO1 × rs6499244 (T>A) NFAT5 determined the maximum percentage (0.86%) of KOA entropy. KOA-associated SNPs are regulatory polymorphisms that affect the expression/splicing level, epigenetic modification of 72 genes in KOA-pathogenetically significant organs such as skeletal muscles, tibial arteries/nerves, thyroid, adipose tissue, etc. These putative KOA-effector genes are mainly involved in the organization/activity of the exoribonuclease complex and antigen processing/presentation pathways. In conclusion, KOA susceptibility among Europeans of Russia is mediated by intergenic interactions (but not the main effects) of GWAS-important SNPs.


Introduction
Osteoarthritis (OA) is a common chronic degenerative joint disease [1], resulting in pathological changes in all joint tissues (cartilage/subchondral bone/ligaments/menisci/ articular capsule/synovial membrane) [2]. More than 500 million people worldwide suffer from this disease; an especially high incidence of OA is observed among people over 65 years of age [3]. The prevalence of large-joint OA (knee and hip) worldwide is 3754.2 per 100 thousand population [4]. At the same time, knee joints are affected more often among the large joints [5,6]. The economic costs associated with OA are high, starting with the direct costs of care and treatment, and ending with the costs of total endoprosthetics [7]. Patients with knee OA (KOA) spend on average about USD 15,000 during their lifetime on direct medical expenses related to the disease [8]. At the same time, expensive endoprosthesis is not always effective; about 20% of patients report Life 2023, 13,405 3 of 15 such as parents, siblings, uncles/aunts, grandparents) was assessed during a survey of the subjects (cases and controls).
The blood of KOA/KOA-free individuals was collected invasively from the elbow vein of the subjects and subsequently used for DNA allocation (the protocol for receiving DNA was set out earlier [40]). The NanoDrop spectrophotometer was employed to verify the DNA concentration/cleanliness [41]. DNA was stored in the freezer at −60 • C and subsequently utilized for the individual's genotypic detection by real-time PCR (using CFX96 apparatus) [42,43]. Genotyping of DNA specimens was executed blindly by laboratory staff, and its accuracy was checked based on re-genotypic of several samples (≈7-10% case/control) [44,45], which showed a degree of 99% agreement.

Statistical Analysis
There was calculated the agreement of SNP genotype observed frequencies (KOA/KOAfree cohorts) to HWE [46,47]. To investigate the statistical dependence between individual SNP variants (allelic, additive, recessive, and dominant models have been employed [48]) and SNP interactions (simulation epistatic models were executed [49] with KOA using gPLINK [50] and MB-MDR [51,52] programs taking into account a number of confounders such as age, BMI, sex, hereditary, occupation-associated physical workload, leisure time physical activity, and the concomitant pathology availability (the musculoskeletal, cardiovascular, and endocrine systems) ( Table 1). Parameters OR and 95% CI (odd ratio and confidence intervals of OR, respectively) have been utilized for the SNP-KOA dependence strength estimates [53,54]. Permutation-corrected p values were used [55] such as p perm ≤ 0.0125 for individual SNPs (Bonferroni-corrected p value based on four calculated models was calculated additionally, 0.05/4 [56]) and p perm ≤ 0.05 for models of SNP interactions [57]. Importantly, amongst the SNP epistatic models for further permutation assessment were the several models appropriate to the Bonferroni-corrected p value (the recombination of ten reported loci was taken into consideration) such as ≤1.11 × 10 −3 (0.05/45 in 2-locus models), ≤4.17 × 10 −4 (0.05/120 in 3-locus) and ≤2.38 × 10 −4 (0.05/210 in 4-locus) [58]. The interaction map of high-confidence KOA effector SNPs was made with the MDR program [59].

Results
The phenotypic characteristics (common biological parameters, occupation-related/ leisure time physical activity, alcohol/smoking habits, hereditary burden, and concomitant pathology) of the examined KOA/control groups are presented in Table 1. The participants' ages ranged from 40 to 70 years in the KOA group and 40-68 years in the KOA-free (control) group. The KOA patients did not differ from the control group in age, sex, height, smoking or alcohol status (p > 0.05). At the same time, KOA patients had a higher BMI (p < 1 × 10 −6 ), hereditary burden (p = 0.0005), obesity rate (p = 0.0005), and high incidence of the cardiovascular (p = 0.0005), endocrine (p = 0.02) and musculoskeletal (p = 0.0005) diseases compared to the control group (Table 1). In the KOA cohort, the individuals number with a high level of occupational physical activity, as well as with low physical activity in their free time, is significantly higher compared to the control group (p = 0.0005 and p = 0.0005, respectively). In turn, the control subjects registered a greater number of individuals with irregular (1.24 times) and regular (2.33 times) physical activity in their free time compared with the KOA patients (p = 0.04 and p = 0.0007, respectively) ( Table 1). Additionally, in the control group, the percentage of individuals with low occupational physical activity was significantly higher (2.12 times) than in the KOA subjects (p = 0.0005). The above parameters were used as factor-confounders in the evaluation of SNP-disease associations.
The statistical calculations showed that all examined GWAS loci in KOA/KOA-free participants were HWE-concordant (p HWE ≥ 0.117 in the KOA group and p HWE ≥ 0.059 in the KOA-free cohort) (Supplementary Table S3). As a result of the genetic analysis, associations of individual SNPs with KOA have not been proven (p bonf > 0.0125) ( Table 2). The present MB-MDR analysis indicated intergenic interaction of the eight GWAS loci of 10 examined SNPs (with the exclusion of rs3771501 (G>A) TGFA and rs143384 (A>G) GDF5) within twelve genetic (epistatic) models as KOA-risk/protective factors ( Table 3). The greatest contribution to the disease development has been made by three polymorphisms/genes: rs6976 (C>T) GLT8D1, rs56116847 (G>A) SBNO1, and rs6499244 (T>A) NFAT5 (each was included in 2/3 [8 out 12] KOA-responsible genetic interaction models). The highest Wald index (this statistic for high risk category equaled 47.56) was in a four-level genetic model: (Table 3). Very important protective potential (−1.07 < beta value < −1.60 and p = 0.00003-0.00009) was found for some SNPs genotypic combinations such as TT(rs6499244)NFAT5 × AG(rs56116847)SBNO1 × CT(rs6976)GLT8D1, Table S4).    The results were obtained using the MB-MDR method with adjustment for covariates; NH: number of significant high risk genotypes in the interaction; beta H: regression coefficient for high risk exposition in the step2 analysis; WH: Wald statistic for high risk category; NL: number of significant low risk genotypes in the interaction; beta L: regression coefficient for low risk exposition in the step2 analysis; WL: Wald statistic for low risk category; pperm: permutation p-value for the interaction model (1000 permutations).
It is important to underline that almost all KOA-associated polymorphic variants (with the exception of rs1060105) and a majority of strongly-linked loci exhibit their outstanding epigenetic effects in KOA-impact cell cultures and tissues such as primary osteoblast cells, chondrocytes, adipocytes, etc. For example, SNPs of genes such as LYPLAL1 (rs2820436 and rs2820443), GNL3 (rs11177), GLT8D1 (rs6976), WWP2 (rs34195470) and
Changes in the gene quantitative traits (transcription and splicing) in different organs (including KOA pathophysiology-involved organs such as skeletal muscles, tibial arteries and nerves, adipose tissue, thyroid, etc.) due to the allele-specific effects of the eight studied SNPs and high-LD loci (283 out 309 SNPs  Tables S11 and S12) may be changed due to the apparent eQTL and sQTL influences of KOA-correlated SNPs.
The outcome of this work demonstrated involvement in the KOA genetic architecture (within gene-gene mutual influence) of rs6499244 (T>A) NFAT5. between rs6499244 (T>A) NFAT5 and KOA risk was shown [35]. The potential epigenetic "authority" of this genome region was also known [15,72,73]. Rice et al. presented data on associations of the rs7359336 locus, which is in LD with rs6499244 NFAT5 (r 2 = 0.91), with an increased level of DNA methylation in the WWP2 gene in patients with OA [72]. Additionally, the authors noted the connection of this functionally-active genome sequence with five cartilage-expressed genes such as PDXDC2P, CLEC18A, NOB1, NFAT5, IL34 [72]. The nuclear factor gene of activated T cells 5 (NFAT5) encodes a transcription factor that manages the quantitative expression traits of genes implicated in the regulation of osmoprotective and inflammatory reactions [73][74][75][76][77]. There is evidence of the role of NFAT5 in the formation of innate immunity by activating gene expression in macrophages during TLR (Toll-like receptor) ligation [75,78,79]. TLR2 and TLR4 are known to be highly expressed in synovial fluid macrophages and are responsible for macrophage activation [80]. NFAT5 also plays an important role in the proliferation of synoviocytes themselves [81]. A higher NFAT5 expression was found in the synovial membrane [81]. NFAT5 indirectly affects the migration of myoblasts during skeletal muscle myogenesis via the CYR61-dependent pathway [82]. Transcription factors NFAT 1, NFAT2, NFAT3 have been an essential function in the OA-pathobiology [83][84][85].
The data for rs56116847 (G>A) SBNO1 as a KOA risk locus were first established by Tachmazidou et al. in GWAS on samples of European origin from UK Biobank and Arthritis Research UK Osteoarthritis Genetics (arcOGEN) [35]. According to our data, this locus, interacting with rs6499244 (T>A) NFAT5, makes the greatest contribution to KOA susceptibility (0.86% of the disease entropy) among Europeans of Russia. The literature data indicate a serious epigenetic potential (relationship with the level of methylation in cartilage) of the locus rs56116847 (G>A) SBNO1 [15]. The information obtained by us in silico also confirms the position of this polymorphism in enhancers in primary T-regulatory and T-helper cells of peripheral blood (E044(Epigenome ID)/BLD.CD4.CD25.CD127M.TREGPC(Mnemonic) and E043(Epigenome ID)/BLD.CD4.CD25M.TPC(Mnemonic), respectively), primary peripheral blood monocytes (E029(Epigenome ID)/BLD.CD14.PC(Mnemonic)), its correlation with eQTL of eight genes (ABCB9, ARL6IP4, C12orf65, CDK2AP1, KMT5A, MPHOSPH9, OGFOD2 and RILPL2) and sQTL of four genes (ABCB9, KMT5A, MPHOSPH9 and RILPL1) in fibroblasts cell culture, tibial nerve, skeletal muscles, whole blood, etc. The KMT5A gene (also called PR-SET7, SET8) is closely related to the regulation of various biological processes such as DNA replication, chromosome condensation and activation of DNA replication checkpoints, cell proliferation, etc. [86]. KMT5A regulates the transcription of Ras, p53 and Wnt genes [86], whose protein products of the same name are involved in the OA-pathophysiology [87][88][89]. For instance, Wnt is well known for its participation in osteogenesis due to activation of the signaling pathways such as Wnt/calcium, Wnt/cyclic adenosine monophosphate (cAMP), Wnt/c-Jun NH(2)-terminal protein kinase (JNK) and Wnt/β-catenin [87]. ERa plays responsible role in cartilage degradation [89]. The inhibitory importance of KMT5A in the proliferation and metastasis of osteosarcoma cells through the transmission of β-catenin signals is also known [90].
The GLT8D1 gene, located in the 3p21.1 region, encodes a protein from the glycosyltransferase family. Glycosyltransferases are a family of enzymes that catalyze the biosynthesis of oligosaccharides, polysaccharides and glycoconjugates [100]. To date, very little is known about the physiological and pathological functions of the GLT8D1 gene and its corresponding protein. The GNL3 gene encodes a guanine-nucleotide-binding protein (nucleostemin), which plays an important role in many processes occurring in the cell, including participation in stem cell proliferation, regulation of the cell cycle, maintenance of telomerase activity [101][102][103]. GNL3 is expressed in mesenchymal stem cells, from which chondrocytes originate [101], shows association with chondrogenic differentiation and may participate in genomic regulation as an RNA-binding protein [104]. In a study by Louka et al., it was found that the GNL3 gene expression in synovial tissue/fluid samples was significantly higher in the group of patients with primary OA compared with the control cohort [105]. The nucleostemin level was also increased in chondrocytes in patients with OA [31].
This study has certain limitations, which include the following: (a) only ten SNP of KOA candidate genes have been studied in the work and the inclusion of more loci in the analysis may cause a "shift" in the estimates of the intergenic interactions role of the examined loci in the formation of KOA; (b) the results of this work were not replicated on another sample from the same population or samples of another ethnic group; (c) this study did not take into account all possible risk factors for KOA (immune factors, bone metabolism, diet features, etc.) and, accordingly, their effects as covariates were not taken into account when evaluating associations; and (d) much broader experimental evidence of the functional effects of SNPs determining susceptibility to KOA is required (only in silico data were used in this work).

Conclusions
In the current study it has been revealed that the KOA susceptibility among Europeans of Russia is mediated by intergenic interactions (but not the independent effects) of GWASimportant SNPs.
Supplementary Materials: The following supporting information can be downloaded at: https: