Effects of multiple genetic loci on the pathogenesis from serum urate to gout

Gout is a common arthritis resulting from increased serum urate, and many loci have been identified that are associated with serum urate and gout. However, their influence on the progression from elevated serum urate levels to gout is unclear. This study aims to explore systematically the effects of genetic variants on the pathogenesis in approximately 5,000 Chinese individuals. Six genes (PDZK1, GCKR, TRIM46, HNF4G, SLC17A1, LRRC16A) were determined to be associated with serum urate (PFDR < 0.05) in the Chinese population for the first time. ABCG2 and a novel gene, SLC17A4, contributed to the development of gout from hyperuricemia (OR = 1.56, PFDR = 3.68E-09; OR = 1.27, PFDR = 0.013, respectively). Also, HNF4G is a novel gene associated with susceptibility to gout (OR = 1.28, PFDR = 1.08E-03). In addition, A1CF and TRIM46 were identified as associated with gout in the Chinese population for the first time (PFDR < 0.05). The present study systematically determined genetic effects on the progression from elevated serum urate to gout and suggests that urate-associated genes functioning as urate transporters may play a specific role in the pathogenesis of gout. Furthermore, two novel gout-associated genes (HNF4G and SLC17A4) were identified.

Gout is among the most common forms of inflammatory arthritis and affects approximately 1 to 6% of the population in various countries [1][2][3] . An elevated serum uric acid concentration (hyperuricemia, HUA) promotes the deposition of monosodium urate crystals in the joints, which then causes gout 4,5 . Hyperuricemia is a key risk factor for the pathogenesis of gout 6 , but only a quarter of people with hyperuricemia develop gout, suggesting that an elevated serum uric acid concentration is necessary but not sufficient for the pathogenesis of gout 7,8 . Recently, a large meta-analysis of genome-wide association studies (GWAS) identified 28 loci associated with serum urate concentration 9 ; however, this result only explained approximately 7% of the variance in serum urate concentrations, and only a portion of those loci were determined to be associated with the risk of gout 9,10 . Therefore, it is necessary to systematically analyze genetic effects on the progression from elevated serum urate to gout and to further identify novel candidate loci that affect the risk of HUA and gout. In addition, population groups have been reported to show common heterogeneity in the genetic contribution of serum urate concentrations and gout 1,11,12 , suggesting the need for transancestral studies to identify population-specific loci that affect the pathogenesis of gout.
To explore the genetic architecture of serum urate and gout mentioned above, 31 SNPs were selected based on predefined criteria and tested in 4914 Chinese individuals (582 gout patients, 1387 HUA patients and 2945 healthy subjects). For data analysis, genetic effects on different combinations (serum urate, gout vs. control, gout vs. HUA and gouty tophi vs. control) were tested to show the influence of genetic variants on the progression from elevated serum urate to gout. In addition, the differences in the transcription levels of candidate genes were measured in 213 male individuals (70 gout patients, 85 hyperuricemia patients and 58 healthy subjects) to further

Results
Loci associated with serum urate and gout. The process for selecting target loci is shown in Figure S1, and after filtration, 31 loci were tested in 4914 Chinese individuals (582 gout patients, 1387 HUA patients and 2945 healthy subjects).
In addition, twelve loci in nine genes (A1CF, PDZK1, GCKR, ABCG2, SLC2A9, TRIM46, HNF4G, SLC17A1 and SLC17A4) were found to influence the risk of gout (all P < 0.05) ( Table 1). After multiple correction, except for the genes PDZK1 and SLC17A4, the remaining ten genes still had effects on the risk of gout with a P FDR value less than 0.05. HNF4G is a novel gout-associated gene associated with the pathogenesis of gout (OR = 1.28, P FDR = 1.08E-03), and two other genes, A1CF and TRIM46, were identified to be associated with susceptibility to gout in the Chinese population for the first time (rs10821905 in A1CF: OR = 1.61, P FDR = 1.57E-03; rs4971101 in TRIM46: OR = 1.37, P FDR = 3.25E-04; rs2070803 in TRIM46: OR = 1.22, P FDR = 0.031). Because SLC17A4 did not affect the concentration of serum urate (P FDR = 0.108), the combined sample of HUA patients and healthy controls was treated as a larger sample control for the further analysis of gout. As a result, SLC17A4 was found to be a novel gout-associated gene affecting the risk of gout (OR = 1. 19 with controls, six genes (A1CF, NRXN2, GCKR, ABCG2, SLC17A1 and TRIM46) were found to influence the development of gouty tophi (all P < 0.05), and after multiple correction, three genes (A1CF, ABCG2 and TRIM46) were still associated with gouty tophi (all P FDR < 0.05) (Supplementary Material, Table S1). In addition, to avoid the heterogeneity of gender, logistic regression adjusted for gender was performed to confirm the association for HNF4G and SLC17A4, and the results also showed that those genes influence the risk of gout (P FDR = 8.92E-05 and 0.040, respectively). To further understand the pathogenesis of gout from serum urate, the comparative genetic effects in hyperuricemia patients and gout patients were analyzed. Six genes (A1CF, ABCG2, SLC17A1, TRIM46, ADRB3 and SLC17A4) were found to be associated with this pathogenesis (all P < 0.05) ( Table 1). After multiple correction, the association between two genes, ABCG2 and SLC17A4 (novel gout-associated gene), remained significant (OR = 1.56, P FDR = 3.68E-09; OR = 1.27, P FDR = 0.013, respectively).
Association between genetic variants and serum urate or gout in gender subgroups. Gender has been proven to be an important heterogeneity factor for serum urate and gout 1,9 . Thus, we further tested the above associations in subgroups of gender (Table 2). In the male subgroup, nine loci in six genes (TCF7L2, GCKR, ABCG2, TRIM46, HNF4G and SLC17A1) were determined to be associated with serum urate (all P < 0.05), and four genes (GCKR, ABCG2, HNF4G and SLC17A1) were still associated with serum urate after multiple correction (all P FDR < 0.05). Eight loci in six genes (TCF7L2, ABCG2, SLC2A9, TRIM46, ESR1 and SLC17A4) exhibited contributions to the level of serum urate in females, with a P value less than 0.05, and ABCG2 showed significant association after multiple corrections (rs1481012: P FDR = 7.47E-10).
Six genes (A1CF, ABCG2, SLC2A9, TRIM46, SLC17A1 and SLC17A4) exhibited effects on the development of gout from hyperuricemia in males (all P < 0.05), but none of them exhibited such effects in females (all P > 0.05) ( Table 2). After multiple correction, only the association of ABCG2 in males was still significant (OR = 1.69, P FDR = 1.55E-10).
In the male subgroup, nine genes (A1CF, PDZK1, GCKR, ABCG2, SLC2A9, TRIM46, HNF4G, SLC17A1 and SLC17A4) were determined to be associated with gout risk (all P < 0.05) ( Table 2), which was consistent with the results of the association in all samples ( Table 1). Six of the genes (A1CF, GCKR, ABCG2, TRIM46, SLC17A1 and HNF4G (novel gout-associated gene)) were still significantly associated with gout in males after multiple correction (all P FDR < 0.05). Because several loci did not affect the concentration of serum urate in males, the combined sample of hyperuricemia patients and healthy controls was treated as a larger sample control for further analysis. Consequently, another novel gout-associated gene, SLC17A4, was identified as a risk factor for the pathogenesis of gout in males (OR = 1.19, P FDR = 0.035) ( Table 2). In females, ABCG2 and TRIM46 contributed to the risk of gout (both P < 0.05), but neither was significant after multiple correction (both P FDR > 0.05).

Association between genetic variants and serum urate in BMI and smoking subgroups.
Previous studies have shown that obesity and cigarette smoking can influence serum uric acid levels [13][14][15] , while their effect on the association between genetic variants and urate was limited. Therefore, further analysis in the BMI and smoking status subgroups was performed.
When body mass index (BMI) was analyzed, ten loci in seven genes (TCF7L2, GCKR, ABCG2, SLC2A9, TRIM46, SLC17A1 and LRRC16A) were associated with serum urate in normal individuals (18.5  BMI < 25), with a P FDR value less than 0.05, and three of those genes (GCKR, ABCG2 and TRIM46) were associated with serum urate in overweight subjects (BMI ≥ 25) ( Table 3). To the best of our knowledge, LRRC16A was identified in the Chinese population for the first time. In the underweight subgroup, no significant associations were found after multiple correction.
The contribution of genetic effects to the pathogenesis of gout. Because most candidate genes only affected the concentrations of serum urate and the risk of gout in males (Table 2), further analysis for the candidate loci were only preformed in males. Across the 13 loci (rs10749127, rs10821905, rs12129861, rs1260326, rs1481012, rs16890979, rs2070803, rs2231137, rs2941484, rs3799352, rs4971101, rs780094 and rs9358890) identified above, for each additional effect allele in males, the odds ratio for gout showed positive linear correlation with the genetic effect on serum urate (R 2 = 0.855) (Fig. 1A). This result was consistent with the fact that an increased serum uric acid level is a key risk factor in the pathogenesis of gout 9 . For the genetic contribution of hyperuricemia to gout, we tested the correlation between the genetic effects associated with gout and hyperuricemia/control. The results showed a high correlation between those two associations (R 2 = 0.816), suggesting that the development of gout from hyperuricemia made a great contribution to the pathogenesis of gout (Fig. 1B).
Genetic urate risk score associated with hyperuricemia and gout. In males, the genetic urate score for the 13 loci identified above could explain an average of 4.76% of the serum urate variance and was strongly associated with hyperuricemia and gout (coefficients = 0.013, P < 2E-16; coefficients = 0.025, P < 2E-16, respectively). The scores ranged from − 70 to + 140, and the increased genetic urate score resulted in elevated proportions of hyperuricemia and gout in males (Supplementary Material, Figure S2). Furthermore, the male subjects Differential expression of candidate genes among groups. To further validate the results presented above, this study randomly selected 70 male gout patients, 85 male hyperuricemia patients and 58 healthy male individuals and tested the differences in the transcription levels of the candidate genes among them (Fig. 2). All candidate genes, except for SLC17A4, showed at least one significant difference in relative expression with a P value less than 0.05, indicating that the loci identified above might influence the risk of hyperuricemia and gout through changing the relative expression levels. Regarding SLC17A4, the synonymous mutation of rs9358890 might affect mRNA transport, splicing and translation 16 and thereby influence the pathogenesis of gout. In addition, the candidate loci (rs10821905, rs12129861, rs1260326, rs2070803, rs2231137, rs3799352, rs4971101, rs780094 and rs742132) were significantly expressed quantitative trait loci (eQTL) that were associated with the expression of one or more transcript in one or more tissue by querying two existing expression eQTL databases (Supplementary Material, Table S2).
Genetic variants influence the progression from hyperuricemia to gout. Across all of the genes identified above with nominal significance, a systematic analysis was conducted to determine the influence of genetic variants on the progression from elevated serum urate to gout (Fig. 3). As a result, a total of fourteen genes exhibited effects on the serum urate concentrations, seven genes on the development of gout from hyperuricemia, and nine genes on gout in all individuals or subgroups. Among the fourteen genes, six genes (ABCG2, SLC2A9, TRIM46, SLC17A1, A1CF and SLC17A4) affected both the elevation of the serum urate level and the development of gout from hyperuricemia and were identified as risk factors for gout. Interestingly, three genes (PDZK1, GCKR and HNF4G) associated with urate influenced the risk of gout, but the other five genes (TCF7L2, LRRC16A, ESR1, NRXN2 and ALPK1) did not, suggesting that elevated serum urate concentration is necessary but not sufficient for the pathogenesis of gout (Fig. 3). In summary, different genes played distinct roles in the pathogenesis of gout, and the loci associated with both serum urate and the development of gout from hyperuricemia were definitively identified as risk factors for gout.

Discussion
An elevated serum urate concentration is necessary but not sufficient for the pathogenesis of gout 7,8 . A total of 28 serum urate concentration-associated loci only explained approximately 7% of the variance in serum urate concentrations, and a portion of those loci were determined to contribute to the pathogenesis of gout 9,10 . Therefore, it is necessary to systematically analyze the genetic variants influencing the progression from elevated serum urate to gout and to identify novel candidate loci that affect the risk of gout. In addition, population-specific effects of serum urate concentrations and gout related genes have been proven before 1,11 , suggesting the need for transancestral studies in validating those loci in other population without association tests.
In this study, we systematically analyzed the genetic variants influencing the progression from elevated serum urate to gout and identified 2 novel gout-associated genes (HNF4G and SLC17A4), using genetic analysis and mRNA expression analysis. The loci from the association analysis were characterized in detail, including an analysis of the association between subgroups comprising gender, body mass index (BMI) and smoking status. For each effect allele identified in this study, the genetic effect on serum urate showed a positive linear correlation with the odds ratio for gout. Gene expression analysis further validated the associations for the loci identified above (except for SLC17A4).
Ten genes previously reported to be associated with serum urate and/or gout [8][9][10][17][18][19] , were confirmed in this study and further analyzed by mRNA expression analysis. To the best of our knowledge, of these genes, six urate concentration-associated genes (PDZK1, GCKR, TRIM46, HNF4G, SLC17A1 and LRRC16A) and two gout-associated genes (A1CF and TRIM46) were identified in the Chinese population for the first time. In addition, two novel gout-associated genes, HNF4G and SLC17A4, were found to affect the risk of gout in our study.
HNF4G was reported as a urate-concentration gene and showed no evidence of association with gout in previous studies 9,10 , although a strong trend towards the association of HNF4G with gout had been found in Europeans (P = 0.058) 10 . HNF4G is a transcription factor responding to nutrient signals, and its overexpression in bladder tumors can significantly increase tumor cell viability, colony formation rate, and invasion, while HNF4G knockdown can achieve the reverse effects 20 . In addition, HNF4G can constitutively bind to endogenous fatty acids 21 .
Our study shows that HNF4G gene expression to be lower in gout patients than in healthy individuals, most likely explaining the mechanism of its effect on the pathogenesis of gout. The effect mechanism should be further studied in future work. SLC17A4 (NPT homologue), an intestinal organic anion exporter, belongs to the NPT subfamily, and its mRNA is expressed mainly in the pancreas, liver, colon, and intestine 22,23 . Togawa et al. found that SLC17A4 actually exists in the apical membrane of the small intestine and transports various types of organic anions, such as urate 22 . Urate is synthesized predominantly in the liver, and nearly two-thirds of daily urate excretion occurs via the kidneys 24 . The remaining urate may be excreted from the intestines, resulting from the biological function of urate excretion for SLC17A4 and BCRP, which are expressed in the intestines 22 Table 3. Association between genetic variants and serum urate in subgroups of BMI and smoking status. A1, allele 1, effect allele. Subgroup of BMI: 1, Underweight (BMI < 18.5); 2, Normal (18.5  BMI < 25); 3, Overweight (BMI ≧ 25). Subgroup of smoke: 1, non-smokers; 2, former smokers; 3, current smokers. β values for SNP in serum urate were calculated by linear regression adjusted age and gender. P values for SNP in serum urate were calculated by deviance analysis for linear regression adjusted age and gender. P FDR value for SNP was multiple corrected by FDR method. When P FDR < 0.05, it would be considered as significant and showed in bold. Because of rs1137070 and rs5953210 in Chromosome X, they cannot be analyzed in this table. 28,141 individuals identified an additional larger region including the SLC17A1, SLC17A3 and SLC17A4 genes that influences the serum urate level 18 . In this study, SLC17A4 was found to be associated with both gout and the development of gout from hyperuricemia, which partially explains the mechanism of the progression from hyperuricemia to gout. Environmental factors, including gender, BMI and smoking status, commonly act as heterogeneity factors for the association of genetic variants and serum urate/gout 6,14,15,26,27 . For example, our previous study showed gender was a source of heterogeneity for the association between ABCG2 variant and gout risk in both meta-regression and subgroup analyses, and the OR values in men and women were significantly different 1 . In this study, HNF4G, SLC17A1 and GCKR played an important role in serum urate concentrations and gout risk in males but not in  HUA, hyperuricemia. All genes identified above with nominal significance were used in this analysis. Each gene was associated with a special phenotype with a P FDR value less than 0.05, as shown in bold. ABCG2, SLC2A9, SLC17A1 and SLC17A4 could encode urate transport. females, also suggesting the different contributions of genetic effects between different genders. Heterogeneity analysis was shown to determine the potential reasons for the equivocal results of the associations seen in our previous study 6 . For instance, a meta-analysis of genome-wide association studies suggested an association between uric acid and rs742132, a common variant in LRRC16A 18 ; in contrast, a replication study with 7795 subjects showed no significant association between this locus and serum urate concentrations 28 . In our study, we also found that rs742132 exhibited no significant effect on serum urate. However, when analyzing by BMI subgroup, this locus showed a significant influence on serum urate concentrations in individuals with normal weight, indicating that the association between rs742132 and serum urate might be modified by BMI, and BMI might thus be a heterogeneity factor leading to the observed discrepancies in results. Above all, the subgroup analysis of heterogeneity factors was helpful to find associations concealed in complex data and to partially explain the biological mechanism of gout incidence via the interaction of genetic variants and environment factors.
One aim of this study was to systematically analyze genetic effects on the progression from hyperuricemia to gout. All the urate transporter-coding genes, ABCG2, SLC2A9, SLC17A1 and SLC17A4, showed association with both the serum urate concentration and progression from hyperuricemia to gout and could affect the risk of gout (Fig. 3). However, only some of the other urate associated genes, which did not code urate transporters, were found to influence the development of gout. Thus, the difference in biological functions of urate-associated genes might be part of the reason that only a quarter of individuals with hyperuricemia can develop gout. In addition, we speculated that the urate-associated genes that function as urate transporters played a certain role in the pathogenesis of gout.
However, there were several limitations in this study. First, this study focused on only common SNPs and did not consider the contributions of rare variants, such as ALDH16A1 29 . Second, because SNPs that did not satisfy the requirements for selection criteria were replaced by other SNPs in the same genes or deleted directly, some SNPs enrolled in this study were not identical to the SNPs published in previous studies. In addition, the female sample is limited as the lower frequency of gout in females, the study might not have enough power to detect significant associations in females. Finally, although gender, BMI and smoking status were considered in our analysis, other environmental factors associated with uric acid and gout were not assessed. Therefore, further studies of these loci should be performed.
In conclusion, this study systematically revealed the genetic effects on the pathogenesis of gout from elevated serum urate and identified two novel gout-associated genes (HNF4G and SLC17A4). The loci associated with increased levels of uric acid were also associated with an increased risk of gout. This study suggests that the differences in biological function of urate-associated genes might be the reason that only a quarter of individuals with hyperuricemia develop gout. We also speculate that the urate-associated genes that function as urate transporters played a certain role in the pathogenesis of gout. These findings strongly support the hypothesis that genetic variants in urate transport genes are the key factors affecting the concentration of serum urate and the risk of gout, suggesting potential implications for the prevention, prediction and treatment of hyperuricemia and gout.

Materials and Methods
Experimental Design. Although genome-wide association studies have identified many genes which play a role in serum urate and gout, our understanding of the effect of genetic variants on the pathogenesis of hyperuricemia and gout is quite limited. Here, we systematically analyzed the genetic variants influencing the progression from elevated serum urate to gout using genetic analysis and mRNA expression analysis. By examining approximately 5,000 Chinese individuals, we attempted to identify novel gout-associated genes and systematically analyzed the genetic effects on the pathogenesis of gout. Further analysis for the transcription levels of the candidate genes also was used to study the association of candidate genes with gout. In addition, we also analyzed the effect of heterogeneity factors (gender, body mass index and smoking status) on the association between genetic variants and serum urate concentrations.

Study subjects. This study was approved by the Ethical Committees of the School of Life Sciences of Fudan
University and was conducted in accordance with the guidelines and regulations of the Declaration of Helsinki. All participants provided written informed consent to this study. A total of 582 gout patients and their clinical information were collected from Changhai Hospital, Taixing People's Hospital and Taizhou People's Hospital. All gout patients enrolled in this study were clinically diagnosed with primary gout (OMIM: #138900) according to the American College of Rheumatology diagnostic criteria 30 . All patients did not have urate-lowering drugs two weeks before sample collection. Among these patients, 174 gout patients were recorded with gouty tophi, which are deposits of uric acid crystals and pathognomonic for the disease gout.
In addition, 4332 subjects with no history of gout were recruited from the Taizhou Longitudinal Study 31 . Those individuals were divided into subgroups according to smoking status, as recorded in questionnaires, and body mass index (BMI) values following the categories of the World Health Organization (WHO) 32 . Smoking status included non-smokers, former smokers and current smokers. For BMI, three subgroups (underweight: BMI < 18.5; normal weight: 18.50 ≤ BMI < 25; overweight: BMI ≥ 25) contained 99, 2024 and 1825 individuals were used in this study. Among them, 1387 subjects with high serum urate (> 417 umol/L) were treated as hyperuricemia patients 33 , and the rest of the patients were treated as healthy controls. The characteristics of the participants in this study are shown in Supplementary Material, Table S3 and Table S4. DNA extraction. Peripheral blood was collected from all participants enrolled in this study. Genomic DNA was isolated from whole blood using a QIAamp DNA Blood Mini kit (QIAGEN, Germany) and then stored at − 20 °C immediately. The concentration and quality of DNA (including optical density (OD) 260/280 and 260/230 measurements) was determined using a Nanodrop Lite spectrophotometer (Thermo Fisher's Scientific, Waltham, MA, USA).
Scientific RepoRts | 7:43614 | DOI: 10.1038/srep43614 RNA Isolation, cDNA Synthesis, and Real-time qPCR. We randomly collected RNA from 70 male gout patients, 85 male hyperuricemia patients and 58 male healthy individuals. RNA was extracted from blood cells using TRIzol reagent according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA). Complementary DNA (cDNA) was synthesized through RNA reverse transcription using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's protocol. Real-time quantitative polymerase chain reaction (qPCR) was performed using SYBR Premix Ex Taq (TakaRa Biotech, Tokyo, Japan) with an ABI Prism 7900 Detector System (Applied Biosystems). The data obtained from the assays were analyzed using the SDS 2.3 software (Applied Biosystems). The human housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control.
Target loci selected. The process for selecting target SNPs was as follows (Supplementary Material, Figure S1). First, SNP association studies were downloaded from the PubMed database (http://www.ncbi.nlm. nih.gov/pubmed/). Second, text-mining technology was used to search for SNPs associated with serum urate levels and/or gout and recorded the frequencies of reported associations for each SNP (frequency 1). Third, the same text-mining method was used to calculate the frequencies of associations with phenotypes other than serum urate levels and gout for each of the above SNPs (frequency 2). Fourth, each of the selected SNPs was considered as a candidate if its frequency 1 was significantly different with its frequency 2 with a P value less than 0.05 in Chi-squared test. In addition, those candidate SNPs were manually verified. Other reported urate/gout-associated SNPs in published reviews were also enrolled in our study. In addition, other important candidate SNPs in transporter genes and hypertension-or diabetes-related genes were included. All selected SNPs were annotated by SNPnexus (http://www.snp-nexus.org/) and filtered by their SNP functions (i.e. SNPs in 5′-upstream, 5′-utr, coding, intronic, 3′ -utr and 3′ -downstream were selected). Then, the SNPs were evaluated by the requirements of SNPscan, the genotyping technology used in our study. SNP that did not satisfy the requirements were replaced by other SNP in the same gene or deleted directly. Finally, after filtration, 31 SNPs were treated as target SNPs for further analysis.
Genetic analysis. Peripheral blood was collected from all the individuals investigated in this study. Genomic DNA was isolated from whole blood using the QIAamp DNA Blood Mini kit (QIAGEN, Germany) and stored at − 20 °C. The DNA concentration and quality (including optical density (OD) 260/280 and 260/230 measurements) were determined using a Nanodrop Lite spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Genotyping of all target SNPs was performed using SNPscan (TianHao, China).
Genetic urate risk score analysis. To analyze the cumulative effects of the loci identified from the association studies for urate and gout, we multiplied, for each locus, the number of effect alleles each person carried (0-2) by the beta coefficient from the genetic analysis and added the results to obtain a genetic urate score 19 . The genetic urate score equation is as follows: rs10749127(N) × beta + rs10821905(N) × beta + rs12129861(N) × beta + rs1260326(N) × beta + rs1481012(N) × beta + rs16890979(N) × beta + rs2070803(N) × beta + rs2231 137(N) × beta + rs2941484(N) × beta + rs3799352(N) × beta + rs4971101(N) × beta + rs780094(N) × beta + rs 9358890(N) × beta, where N of each SNP denotes the number of copies of that allele carried by each subject, and beta value is the effect size per allele in serum urate. The association between genetic urate score and HUA or gout were tested in males by logistic regression with adjustment for age. Linear regression was used to analyze the relationship between the score and serum urate concentration in males.

Statistical analysis.
The genotype data of the loci were checked for deviation from the Hardy-Weinberg equilibrium. The beta values for serum urate loci were calculated by linear regression adjusted for age and gender. P values for serum urate loci were calculated by deviance analysis for linear regression with adjustment for age and gender. All P values for gout loci were calculated by Fisher's exact test in the addition model and logistic regression adjusted for gender. Furthermore, subgroups based on gender, BMI and smoking status were used in this study. P values for the loci were multiply corrected by the FDR method (P FDR ), and values below 0.05 were considered statistically significant.
Data on mRNA expression are illustrated as boxplots with all outliers cleared. The differences in mRNA expression of candidate genes among gout patients, hyperuricemia patients and healthy controls were analyzed by Student's t-test. The differences among different genotypes were also tested in this study. P values below 0.05 were considered statistically significant. In addition, we also queried two existing expression quantitative trait locus (eQTL) databases (Geuvadis data browser (http://www.ebi.ac.uk/Tools/geuvadis-das/) 34 and Genotype-Tissue Expression Data Portal (http://www.gtexportal.org/home/)) 35 to analyze the association of candidate loci with transcript expression.
All statistical analyses were performed using R (Version 3.0.2: www.r-project.org/).