2414 3. Ruotsalainen V, Ljungberg P, Wartiovaara J et al. Nephrin is specifically located at the slit diaphragm of glomerular podocytes.

Proc Natl Acad Sci USA 1999; 96: 79627967 4. Boute N, Gribouval O, Roselli S et al. NPHS2, encoding the glomerular protein podocin, is mutated in autosomal recessive steroid-resistant nephrotic syndrome. Nat Genet 2000; 24: 349354 5. Donoviel DB, Freed DD, Vogel H et al. Proteinuria and perinatal lethality in mice lacking NEPH1, a novel protein with homology to NEPHRIN. Mol Cell Biol 2001; 21: 48294836 6. Liu G, Kaw B, Kurfis J et al. Neph1 and nephrin interaction in the slit diaphragm is an important determinant of glomerular permeability. J Clin Invest 2003; 112: 209221 7. Shih NY, Li J, Karpitskii V et al. Congenital nephrotic syndrome in mice lacking CD2-associated protein. Science 1999; 286: 312315 8. Huber TB, Hartleben B, Kim J et al. Nephrin and CD2AP associate with phosphoinositide 3-OH kinase and stimulate AKT-dependent signaling. Mol Cell Biol 2003; 23: 49174928 9. Kaplan JM, Kim SH, North KN et al. Mutations in ACTN4, encoding alpha-actinin-4, cause familial focal segmental glomerulosclerosis. Nat Genet 2000; 24: 251256 10. Reiser J, Polu KR, Moller CC et al. TRPC6 is a glomerular slit diaphragm-associated channel required for normal renal function. Nat Genet 2005; 37: 739744 11. Li H, Lemay S, Aoudjit L et al. SRC-family kinase Fyn phosphorylates the cytoplasmic domain of nephrin and modulates its interaction with podocin. J Am Soc Nephrol 2004; 15: 30063015 12. Jones N, Blasutig IM, Eremina V et al. Nck adaptor proteins link nephrin to the actin cytoskeleton of kidney podocytes. Nature 2006; 440: 818823 13. Johnstone DB, Holzman LB. Clinical impact of research on the podocyte slit diaphragm. Nat Clin Pract Nephrol 2006; 2: 271 282

A. D. Rule et al. 14. Lenkkeri U, Mannikko M, McCready P et al. Structure of the gene for congenital nephrotic syndrome of the finnish type (NPHS1) and characterization of mutations. Am J Hum Genet 1999; 64: 5161 15. Beltcheva O, Martin P, Lenkkeri U et al. Mutation spectrum in the nephrin gene (NPHS1) in congenital nephrotic syndrome. Hum Mutat 2001; 17: 368373 16. Gigante M, Monno F, Roberto R et al. Congenital nephrotic syndrome of the Finnish type in Italy: a molecular approach. J Nephrol 2002; 15: 696702 17. Aya K, Tanaka H, Seino Y. Novel mutation in the nephrin gene of a Japanese patient with congenital nephrotic syndrome of the Finnish type. Kidney Int 2000; 57: 401404 18. Sako M, Nakanishi K, Obana M et al. Analysis of NPHS1, NPHS2, ACTN4, and WT1 in Japanese patients with congenital nephrotic syndrome. Kidney Int 2005; 67: 12481255 19. Weber S, Gribouval O, Esquivel EL et al. NPHS2 mutation analysis shows genetic heterogeneity of steroid-resistant nephrotic syndrome and low post-transplant recurrence. Kidney Int 2004; 66: 571579 20. Ruf RG, Lichtenberger A, Karle SM et al. Patients with mutations in NPHS2 (podocin) do not respond to standard steroid treatment of nephrotic syndrome. J Am Soc Nephrol 2004; 15: 722732 21. Caridi G, Bertelli R, Di Duca M et al. Broadening the spectrum of diseases related to podocin mutations. J Am Soc Nephrol 2003; 14: 12781286 22. Kitamura A, Tsukaguchi H, Iijima K et al. Genetics and clinical features of 15 Asian families with steroid-resistant nephrotic syndrome. Nephrol Dial Transplant 2006; 21: 31333138 23. Maruyama K, Iijima K, Ikeda M et al. NPHS2 mutations in sporadic steroid-resistant nephrotic syndrome in Japanese children. Pediatr Nephrol 2003; 18: 412416 24. Grantham R. Amino acid difference formula to help explain protein evolution. Science 1974; 185: 862864 Received for publication: 28.3.08; Accepted in revised form: 2.3.09

Downloaded from ndt.oxfordjournals.org by guest on May 15, 2011

Nephrol Dial Transplant (2009) 24: 24142420 doi: 10.1093/ndt/gfp080 Advance Access publication 3 March 2009

Genome-wide linkage analysis for uric acid in families enriched for hypertension
Andrew D. Rule1 , Brooke L. Fridley2 , Steven C. Hunt3 , Yan Asmann4 , Eric Boerwinkle5 , James S. Pankow6 , Thomas H. Mosley7 and Stephen T. Turner1
1 Division of Nephrology and Hypertension, 2 Division of Biostatistics, Mayo Clinic, Rochester, MN, 3 Cardiovascular Genetics Division, University of Utah School of Medicine, Salt Lake City, UT, 4 Division of Bioinformatics, Mayo Clinic, Rochester, MN, 5 Human Genetics Center and Institute of Molecular Medicine, University of Texas-Houston Health Science Center, TX, 6 Division of Epidemiology and Community Health, University of Minnesota, Minneapolis, MN and 7 Department of Medicine, University of Mississippi Medical Center, Jackson, MS, USA

Correspondence and offprint requests to: Andrew D. Rule; E-mail: rule.andrew@mayo.edu

Abstract Background. Uric acid is heritable and associated with hypertension and insulin resistance. We sought to identify genomic regions influencing serum uric acid in families in which two or more siblings had hypertension. Methods. Uric acid levels and microsatellite markers were assayed in the Genetic Epidemiology Network of Arteriopa-

thy (GENOA) cohort (1075 whites and 1333 blacks) and the Hypertension Genetic Epidemiology Network (HyperGEN) cohort (1542 whites and 1627 blacks). Genome-wide linkage analyses of uric acid and bivariate linkage analyses of uric acid with an additional surrogate of insulin resistance were completed. Pathway analysis explored gene sets enriched at loci influencing uric acid.

C The Author [2009]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

Linkage analysis for uric acid

2415

Results. In the GENOA white cohort, loci influencing uric acid were identified on chromosome 8 at 135 cM [multipoint logarithm of odds score (MLS) = 2.4], on chromosome 9 at 113 cM (MLS = 3.7) and on chromosome 16 at 93 cM (MLS = 2.3), but did not replicate in HyperGEN. At these loci, there was evidence of pleiotropy with other surrogates of insulin resistance and genes in the fructose and mannose metabolism pathway were enriched. In the HyperGEN-black cohort, there was some evidence of a locus for uric acid on chromosome 4 at 135 cM (MLS = 2.4) that had modest replication in GENOA (MLS = 1.2). Conclusions. Several novel loci linked to uric acid were identified but none showed clear replication. Widespread diuretic use, a medication that raises uric acid levels, was an important study limitation. Bivariate linkage analyses and pathway analysis were consistent with genes regulating insulin resistance and fructose metabolism contributing to the heritability of uric acid.
Keywords: genomics; hypertension; insulin resistance; linkage analysis; uric acid

els rise to compensate [15]. Several recent genome-wide studies have found SLC2A9, a glucose, fructose and urate transporter, to be associated with uric acid levels [1618]. Hyperuricaemia may also lead to chronic kidney disease by activation of reninangiotensin systems, glomerular hypertrophy/hypertension, renal hypoperfusion and tubulointerstitial fibrosis [19,20]. Bivariate linkage analyses were performed between uric acid and other surrogates of insulin resistance (obesity, diabetes mellitus or triglycerides) and between uric acid and serum creatinine.
Subjects and methods
Subjects The GENOA cohorts recruited non-Hispanic white and black participants from two centres between 1996 and 2000. In Rochester, MN, the Rochester Epidemiology Project [21] was used to identify non-Hispanic white residents of the Olmsted County general population with a diagnosis of essential hypertension before the age of 60 years. In Jackson, MS, residents with essential hypertension were identified through the Atherosclerosis Risk in Communities cohort, a general population sample of 45- to 64-year-old non-Hispanic black residents. Probands with evident secondary hypertension were not recruited at either site (e.g. serum creatinine >2.0 mg/dL). The hypertensive proband and all siblings were invited to participate if at least one other sibling had essential hypertension. Between 2000 and 2004, 2721 (79%) of the 3434 original GENOA participants (1239 whites and 1482 blacks) returned for a second clinic visit [22]. Serum uric acid levels were measured in 1238 whites and 1468 blacks. Linkage analyses were based on the pedigree relationships of all GENOA participants and the genome-wide microsatellite markers available in 1075 whites and 1333 blacks with measured uric acid levels. The HyperGEN cohorts were recruited between 1996 and 1999 from five centres: Framingham, MA; Minneapolis, MN; Salt Lake City, UT; Forsyth County, NC; and Birmingham, AL. The NHLBI Family Heart Study was used to identify hypertensive sibships in Massachusetts, Minnesota, Utah and North Carolina. Additional black participants in North Carolina and Alabama were recruited through a variety of media and community sources. As with GENOA, hypertension was confirmed at the study visit by a systolic blood pressure >140 mm Hg or diastolic blood pressure >90 or treatment with anti-hypertensive medications before age 60 years [12]. Probands with evident secondary hypertension were excluded. In addition to the sibships (n = 2342 individuals), parents and unmedicated offspring of index siblings (n = 665 individuals) and a random sample of the community (mostly spouses, n = 918 individuals) were also recruited. Overall, 3925 individuals (1906 whites and 1926 blacks) participated in the visit. Serum uric acid levels were measured in 1832 whites and 1898 blacks. Linkage analyses were based on the pedigree relationships of all HyperGEN participants and the genome-wide microsatellite markers available in 1542 whites and 1627 blacks with measured uric acid levels. Measurements The clinic visit for both the GENOA and HyperGEN cohorts included a blood draw, measurement of height and weight to calculate body mass index (BMI), waist circumference, smoking history and current medications. Use of anti-gout medications was defined as the use of allopurinol, colchicine or probenecid in the last month. Other medications known to affect uric acid levels were also identified: diuretics, aspirin, warfarin, losartan and fenofibrate. Fasting glucose, creatinine, triglycerides and HDL cholesterol were assayed in serum samples. Hypertension was defined by a blood pressure >140/90 or use of anti-hypertensive medications. Diabetes was defined by a fasting glucose 126 mg/dL or use of oral hypoglycaemic agents or insulin. Serum uric acid was measured with a standard automated uricase enzymatic assay [23] at a separate central laboratory for each network. The GENOA cohort used a Hitachi 911 (Roche Diagnostics, Indianapolis, IN) analyser and the HyperGEN cohort used a Vitros 700 analyser (Johnson & Johnson Clinical Diagnostics, Rochester, NY). A set of 381 microsatellite markers for GENOA and 391 microsatellite markers for HyperGEN across the 22 autosomes were genotyped by the Mammalian Genotyping Center of the Marshfield Medical Research Foundation (Marshfield, WI) using a standard polymerase chain reaction.

Introduction
Uric acid is the main breakdown product of purine metabolism and is largely renally excreted since humans lack a uricase enzyme. Hyperuricaemia is common, particularly among men in the general community. Uric acid can crystalize at levels >6.7 mg/dL [1]. Gouty arthritis and nephrolithiasis are well-described complications of hyperuricaemia, but the vast majority of hyperuricaemic individuals are asymptomatic. Hyperuricaemia is a strong predictor of hypertension, insulin resistance, cardiovascular disease and chronic kidney disease [26]. Established risk factors for hyperuricaemia include factors that increase dietary purines (e.g. animal-based proteins) and decrease renal clearance of purines (e.g. diuretics and alcoholic beverages). Uric acid levels are also heritable [7,8], and recent genome-wide linkage analyses have identified quantitative trait loci (QTL) in white and Mexican American families [911]. These studies have also found evidence of pleiotropy in as much as some loci influencing uric acid also influence other surrogates of insulin resistance (e.g. obesity and diabetes). This study sought to compare these previously published studies with results of genome-wide linkage analyses in two biracial study populations: the Genetic Epidemiology Network of Arteriopathy (GENOA) and the Hypertension Genetic Epidemiology Network (HyperGEN), both within the Family Blood Pressure Program (FBPP) [12]. Genomewide linkage analyses of uric acid in these families enriched for hypertension may provide new insight into the genetics of uric acid since hypertension is both common [13] and associated with hyperuricaemia [2]. Hyperuricaemia is also associated with insulin resistance, perhaps via dietary fructose, a common sweetener that raises uric acid levels by depleting hepatic ATP with the conversion to fructose-1-P and the subsequent degradation of AMP to uric acid [14]. By raising uric acid levels, fructose impairs glucose uptake by skeletal muscle and insulin lev-

Downloaded from ndt.oxfordjournals.org by guest on May 15, 2011

2416 Table 1. Descriptive characteristics of the study participants Characteristic (frequency/meanSD) Sample size Sibships Sibling pairs Singletons Male (%) Age (years) Uric acid (mg/dL) Hyperuricaemiaa (%) Serum creatinine (mg/dL) BMI (kg/m2 ) Hypertension (%) Current smoker (%) Diabetes mellitus (%) Triglycerides, mg/dL HDL cholesterol, mg/dL Diuretic use (%) Anti-gout medicationb use (%) Regular aspirin use (%) Warfarin use (%) Losartan use (%) Fenofibrate use (%)
a Serum uric acid >7 mg/dL in men and b Allopurinol, colchicine or probenecid.

A. D. Rule et al.

GENOA whites 1075 415 1217 4 45 60 10 6.1 1.7 35 0.90 0.27 31 6 75 9.3 16 160 96 51 15 40 2.0 43 2.4 5.2 1.1 >6 mg/dL in women.

GENOA blacks 1333 552 1380 24 30 63 9 6.1 1.7 40 0.90 0.35 32 7 79 12.5 29 120 78 58 18 45 2.4 33 2.8 3.1 0.3

HyperGEN whites 1542 612 1248 180 47 57 13 5.9 1.6 27 0.98 0.28 30 6 80 9.0 15 165 104 49 14 30 3.4 20 3.4 3.2 0

HyperGEN blacks 1627 702 983 26 36 48 13 6.0 1.8 34 1.01 0.49 32 8 77 28 20 113 108 54 15 37 1.9 13 1.2 1.7 0

Downloaded from ndt.oxfordjournals.org by guest on May 15, 2011

Data cleaning to verify family relationships and identify sample mix-ups was performed separately at each network as described elsewhere [19,20]. Genotyping errors that could not be resolved were recoded as missing. Heritability and linkage analyses Mean levels of uric acid by gender and race were estimated within the GENOA and HyperGEN cohorts using mixed models, where a random sibship effect was included to model the dependence within a family [24]. The genetic analyses were performed separately in GENOA whites, GENOA blacks, HyperGEN whites and HyperGEN blacks. Uric acid levels were transformed using the van der Waerden transformation to meet the assumption of normality for variance component linkage analysis. Analyses were agesex-adjusted or multivariable-adjusted (age, sex, serum creatinine, body mass index, current smoker, diabetes, triglycerides, HDL cholesterol, anti-gout medications, diuretics, aspirin, warfarin, losartan and fenofibrate). Alcohol intake and C-reactive protein were available for GENOA; including these variables in models had no substantive effect on the results and is not shown. Excluding individuals on anti-gout medications also had no substantive effect on the results (data not shown). The heritability estimate was computed as the percent of variance in the quantitative measure of uric acid that is explain by the familial relation2 2 2 2 ship, h 2 = a ( a + ), where a is the variance due to shared familial 2 effects and is the residual variance from a polygenic model [25,26], after adjustment for age and sex. Quantitative linkage analysis using a variance component approach was implemented using the multic R library [2527]. Identity by descent matrices were computed using Simwalk2 [28,29]. Allele frequencies were estimated based on each study cohort. Test for linkage was based on the likelihood ratio test with LOD scores also calculated. For the genome-wide univariate linkage analyses, a MLS >3.00 (P < 0.0001) was considered statistically significant evidence of linkage and a MLS >2.00 (P < 0.001) was considered suggestive evidence of linkage [30]. The genome-wide linkage results were compared between the GENOA and HyperGEN cohorts to assess replication. Meta-analyses for both whites and blacks were also completed by combining the linkage results in the GENOA and HyperGEN networks using the GSMA (genome search meta analysis) method outlined by Wise et al. with age- and sex-adjusted LOD scores and 30 cM bins [31]. Bivariate linkage can have the advantage of greater statistical power than univariate linkage when a single locus influences multiple traits (i.e. pleiotropy). Pleiotropy of loci influencing uric acid and other surrogates of insulin resistance has been previously reported [911]. For this study, bivariate linkage analyses were performed between uric acid and each of

the following measures: BMI, diabetes, triglycerides and serum creatinine. The bivariate linkage analysis was completed using an extension of the univariate variance component model [2527,32], and a test for linkage was conducted using a likelihood ratio test [33] with the multic R library. Diabetes was modelled using a 01 indicator variable. For the genome-wide bivariate analyses, a MLS 4.00 (P < 0.0001) was considered statistically significant evidence of bivariate linkage and a MLS 2.87 (P < 0.001) was considered suggestive evidence of bivariate linkage [34]. Pathway analysis Pathway analysis was performed to explore uric acid as a polygenic trait affected by genes enriched in canonical pathways related to uric acid metabolism. Pathway analysis was performed separately in each cohort using all QTL with at least suggestive evidence of agesex-adjusted univariate linkage to uric acid. QTL were defined by the centimorgan (cM) range for a drop of 1.0 LOD on either side of the MLS. The National Center for Biotechnology Information (NCBI) genome (http://www.ncbi.nlm.nih.gov/mapview) build 36.2 was used to map the Marshfield markers to the physical locations closest to the start and end of the QTL. All genes between the two markers were identified. Canonical pathways associated with the set of genes located within the QTL were explored with Ingenuity Pathway Analysis (Ingenuity Systems, Inc., Redwood City, CA). With this tool, Fishers exact test was used to determine if a given pathway was enriched by the genes located within the QTL. For this exploratory analysis, statistical significance was defined as a nominal P < 0.05.

Results
Descriptive characteristics of the study populations are shown in Table 1. In the GENOA cohort, uric acid was 1.0 mg/dL higher (P-value < 0.0001) in men than in women after adjusting for age and race and 0.2 mg/dL (P-value = 0.01) higher in blacks than in whites after adjusting for age and sex. In the HyperGEN cohort, uric acid was 1.2 mg/ dL (P-value < 0.0001) higher in men than in women after adjusting for age and race and 0.3 mg/dL (P-value < 0.0001) higher in blacks than in whites after adjusting for age and sex. Age- and sex-adjusted heritability estimates in the GENOA cohorts were 0.41 0.07 for whites and

Linkage analysis for uric acid Table 2. Heritability (SE) of uric acid Agesexadjusted 0.41 0.07 0.55 0.07 0.35 0.06 0.35 0.06 Multivariableadjusted 0.48 0.07 0.51 0.07 0.39 0.06 0.31 0.06

2417

Cohort GENOA whites GENOA blacks HyperGEN whites HyperGEN blacks

Unadjusted 0.46 0.07 0.53 0.07 0.34 0.06 0.36 0.06

0.55 0.07 for blacks and in the HyperGEN cohorts were 0.35 0.06 for whites and 0.35 0.06 for blacks (Table 2). Figure 1 summarizes of the agesex-adjusted univariate linkage analysis for uric acid. Regions with at least suggestive evidence of linkage [i.e. multipoint logarithm of odds score (MLS) > 2] are presented in Table 3. Significant evidence of linkage (i.e. MLS > 3) on chromosome 9 (MLS = 3.7 at 113 cM) for GENOA whites and suggestive evidence of linkage on chromosome 8 (MLS = 2.4 at 135 cM) and chromosome 16 (MLS = 2.3 at 93 cM) for GENOA whites were not evident in HyperGEN whites (Figure 1A). These QTL were attenuated with multivariable adjustment, with only chromosome 9 in GENOA whites having at least suggestive evidence of linkage (MLS = 2.2 at 111 cM). With multivariable adjustment, suggestive evidence of linkage emerged on chromosome 2 for GENOA whites (MLS = 2.3 at 247 cM, agesex-adjusted MLS = 1.1 at 237 cM). Suggestive evidence of linkage on chromosome 4 (MLS = 2.4 at 135 cM) for HyperGEN blacks was seen with some evidence of replication at the same region in GENOA blacks (MLS = 1.2 at 145 cM) (Figure 1B). With multivariable adjustment, this QTL was attenuated (MLS = 1.0 at 153 cM), but suggestive evidence of linkage emerged at another loci on chromosome 4 (MLS = 2.1 at 33 cM) in HyperGEN blacks. The meta-analysis combining the results from the GENOA and HyperGEN cohorts showed no evidence of linkage (P > 0.01) at these QTL or any other region of the genome. Quantitative trait loci with at least suggestive evidence of bivariate linkage between uric acid and either BMI, diabetes, triglycerides or serum creatinine are presented in

Table 4. All three QTLs for univariate uric acid linkage in GENOA whites had significant evidence of linkage in the bivariate linkage analysis (Figure 2). In particular, there was bivariate linkage between uric acid and diabetes on chromosome 8 (MLS = 5.0 at 121 cM) and chromosome 9 (MLS = 5.3 at 129 cM) and between uric acid and BMI on chromosome 16 (MLS = 4.3 at 84 cM). There was suggestive evidence of linkage in HyperGEN blacks between uric acid and BMI on chromosome 4 (MLS = 3.0 at 166 cM) and in GENOA blacks between uric acid and triglycerides on chromosome 14 (MLS = 3.0 at 123 cM). There was also at least suggestive evidence of bivariate linkage between uric acid and serum creatinine at several loci in the GENOA and HyperGEN cohorts. Ingenuity pathway analysis was performed in each cohort using all QTL with at least suggestive evidence of agesex-adjusted linkage to uric acid. For GENOA whites, the physical location of the three QTL were chromosome 8, 120 M bp to 129 M bp; chromosome 9, 93 M bp to 112 M bp; and chromosome 16, 57 M bp to 80 M bp. In these regions a total of 437 genes were identified. Pathway analysis using these focus genes identified six enriched pathways and three of these pathways contained genes at two different QTLs: fructose and mannose metabolism (P = 0.013), inositol metabolism (P = 0.020) and aminoacyl-tRNA biosynthesis (P = 0.045). For HyperGEN blacks, the physical location of the QTL was chromosome 4, 97 M bp to 158 M bp; from which 339 genes were identified. There were eight enriched pathways: seven pathways involving the alcohol dehydrogenase superfamily of genes and the coagulation system pathway. The purine metabolism pathway was borderline significant for enrichment (P = 0.11).

Downloaded from ndt.oxfordjournals.org by guest on May 15, 2011

Discussion
Among family members with similar serum uric acid levels, there were genomic regions where family members were more likely to share the same alleles (i.e. linkage). For GENOA whites, genome-wide significant evidence

Fig. 1. Summary of linkage analysis of serum uric acid by each chromosomal location (y-axis) and LOD score (x-axis) for (A) whites and (B) blacks (GENOAsolid, HyperGENdashed)

2418 Table 3. Regions with at least suggestive evidence of linkage (MLS 2.0) for uric acid (agesex or multivariable-adjusted)

A. D. Rule et al.

Agesex-adjusted Cohort GENOA whites GENOA whites GENOA whites GENOA whites HyperGEN blacks HyperGEN blacks
a Results

Multivariable-adjusted Loci (cM) 128 111 91 247 153 33 LOD 0.5, P = 0.06 2.2, P = 0.0008 0.5, P = 0.06 2.3, P = 0.0006 1.0, P = 0.02 2.1, P = 0.001

Any evidence of replication (LOD 0.5 with agesex-adjusted uric acid linkage) Cohort HyperGEN whites Framingham HyperGEN whites HyperGEN whites GENOA blacks HyperGEN whites Framingham NHLBI-FHS GENOA blacks GENOA blacks Loci (cM) 135 135 113 93 110 260 256 228 145 39 LOD <0.5 1.2a <0.5 <0.5 1.2, P = 0.008 1.1, P = 0.02 1.1a 1.9 1.2, P = 0.008 <0.5

Chromosome 8 9 16 2 4 4

Loci (cM) 135 113 93 237 135 39

LOD 2.4, P = 0.0005 3.7, P < 0.0001 2.3, P = 0.0005 1.1, P = 0.01 2.4, P = 0.0004 1.7, P = 0.003

obtained from Figures 1 and 2 in the Framingham Heart Study cohort [9].

Table 4. Regions with at least suggestive evidence of bivariate linkage (MLS 2.87) and any evidence of univariate uric acid linkage at the same loci (LOD 0.5) Univariate (uric acid) LOD score 3.7, P < 0.0001 2.0, P = 0.0013 0.5, P = 0.06 1.3, P = 0.008 1.8, P = 0.002 3.6, P < 0.0001 1.7, P = 0.003 0.7, P = 0.03 3.7, P < 0.0001 0.9, P = 0.020 1.6, P = 0.0033 0.9, P = 0.021 Univariate (other trait) LOD score 0.3, P = 0.10 2.9, P = 0.0001 0.03, P = 0.36 4.0, P < 0.0001 3.6, P < 0.0001 0.5, P = 0.06 0.2, P = 0.17 0.4, P = 0.09 1.7, P = 0.0023 2.0, P = 0.0012 2.2, P = 0.0007 1.7, P = 0.0024

Bivariate phenotype Uric acid and BMI Uric acid and diabetesa Uric acid and triglycerides Uric acid and serum creatinine

Cohort GENOA whites HyperGEN blacks GENOA whites GENOA whites GENOA Jackson GENOA whites GENOA blacks HyperGEN whites HyperGEN blacks

Chromosome 9 16 4 8 9 9 16 14 9 5 17 7

Loci (cM) 113 84 166 121 129 113 123 123 113 85 79 155

Bivariate LOD score 3.8, P = 0.0001 4.3, P = 0.0001 3.0, P = 0.0009 5.0, P < 0.0001 5.3, P < 0.0001 3.8, P = 0.0001 2.9, P = 0.0001 3.0, P = 0.0009 4.6, P < 0.0001 3.4, P = 0.0004 3.5, P = 0.0003 3.2, P = 0.0006

Downloaded from ndt.oxfordjournals.org by guest on May 15, 2011

a Qualitative

linkage analysis used for the univariate diabetes phenotype.

Fig. 2. Multipoint LOD scores of agesex-adjusted linkage plots in GENOA whites for (A) chromosome 8, (B) chromosome 9 and (C) chromosome 16. Thin curve is for univariate linkage to uric acid. Dashed curve is for univariate linkage to the surrogate of insulin resistance. Thick curve is for the bivariate linkage to uric acid and the surrogate of insulin resistance. BMIbody mass index.

of linkage for uric acid was found on chromosome 9 (113 cM) and suggestive evidence of linkage on chromosomes 8 (135 cM) and 16 (93 cM), but these QTL did not replicate in HyperGEN whites. All three QTL showed genome-wide significant evidence of linkage (MLS > 4.0) in bivariate analysis with uric acid and a surrogate of insulin resistance (Figure 2). For HyperGEN blacks, there was suggestive evidence of linkage for uric acid on chromosome 4 (135 cM) that showed some evidence of replication in GENOA blacks. Other investigators have performed genome-wide linkage analysis for uric acid levels. Evidence for replication of

the findings in this study with the findings by other investigators is shown in Table 3. In the Framingham Heart Study, there was evidence of linkage (MLS = 3.3) on chromosome 15 (50 cM) [9], but this QTL was not replicated in the GENOA or HyperGEN cohorts. The National Heart, Lung, and Blood Institute Family Heart Study found evidence of linkage (MLS = 3.3) on chromosome 2 (240 cM) for a composite measure of uric acid and seven other surrogates of insulin resistance [11]. This QTL did show some evidence of replication in both GENOA and HyperGEN whites (Table 3) and in the Framingham Heart Study (MLS = 1.1 at 256 cM) [9]. At this locus (2q37.3), positional cloning

Linkage analysis for uric acid

2419

implicated the CAPN10 gene as a type 2 diabetes susceptibility gene [35,36]. Other studies have not confirmed this association and the mechanism for diabetes risk is unknown [37]. The San Antonio Family Heart Study (Mexican Americans) found evidence of linkage (MLS = 3.3) for uric acid on chromosome 6 (133 cM) [10]. There was some evidence of replication of this locus in GENOA whites (MLS = 0.9 at 138 cM), but not in HyperGEN whites. In this study, the loci with the most statistically significant evidence for uric acid linkage were on chromosomes 8, 9 and 16 in the GENOA white cohort. These QTL also showed evidence of pleiotropy between uric acid and surrogates of insulin resistance, and the QTL on chromosome 9 also showed pleiotropy between uric acid and serum creatinine (Table 4). Genes in the fructose and mannose metabolism pathway were enriched at these QTL. The enriched genes were aldolase B (9q21.3q22.2), fructose-1,6-bisphosphatase 1 (9q22.3), fructose1,6-bisphosphatase (9q22.3) and fucokinase (16q22.1). Fructose is the only dietary sugar that raises uric acid levels in humans [38]. The development of high-fructose corn syrup may be a major contributor to the increased prevalence of obesity, diabetes and hypertension [15,39,40]. By raising uric acid levels, fructose interferes with glucose uptake by muscles and contributes to insulin resistance [15]. The liver responds to a fructose load by sequestration of inorganic phosphate in fructose-1-phosphate leading to decreased ATP synthesis. Decreased ATP synthesis leads to increased inosine, a precursor to uric acid [14]. Of the four genes enriched in the fructose and mannose metabolism pathway, aldolase B is known to have an associated genetic disorder leading to hyperuricaemia. Aldolase B deficiency is a rare autosomal recessive disorder that presents clinically with failure to thrive and fast-induced hypoglycaemia. Heterozygotes for aldolase B deficiency have a greater rise in uric acid levels after ingestion of fructose than that in controls [41]. It is plausible that heterozygotes for aldolase B deficiency or other aldolase B variants explain the QTL found on chromosome 9. There was suggestive evidence of linkage of uric acid to a locus on chromosome 4 at 135 cM in HyperGEN blacks with evidence of replication in GENOA blacks. This QTL also showed suggestive evidence of pleiotropy with a surrogate of insulin resistance (BMI). While only borderline significant (P = 0.11) for enrichment, purine metabolism is a biologically plausible pathway to account for this QTL since uric acid is the end-product of purine metabolism. The purine metabolism pathway genes were GUCY1A3, GUCY1B3, PAPSS1, PDE5A and SMARCA5, but none have a reported association with hyperuricaemia or insulin resistance. Pathways containing alcohol dehydrogenase genes were enriched at this QTL, in part, because seven members of this gene family are contiguous at this locus. There was also suggestive evidence of multivariableadjusted linkage to uric acid on chromosome 4 at 33 cM in HyperGEN blacks, but without evidence of replication in GENOA blacks. The SLC2A9 gene, a glucose, fructose and uric acid transporter gene associated with serum uric acid levels [1618], is located near this locus on chromosome 4 at 21 cM [agesex-adjusted log of the odds (LOD) score = 0.9, P = 0.02; multivariable-adjusted LOD score = 1.2,

P = 0.01]. Functional polymorphism in the SLC2A9 gene may have caused the linkage in HyperGEN blacks near this locus. This study and other studies that have identified QTL with linkage to uric acid or QTL with pleiotropic effects between uric acid and other surrogates of insulin resistance have had limited replication. Unlike prior studies, the cohorts used in this study were composed of families characterized by two or more siblings with hypertension. The reason for lack of replication of the findings between GENOA whites and HyperGEN whites is not known. There were differences between the two networks with respect to how the hypertensive subjects were identified, inclusion of normotensive siblings and the multigenerational structure of the families. Instead of large effects of a few genes, small effects of many genes with differences in geneenvironment and genegene interactions between networks may have contributed to lack of replication [37]. Restricting the biological pathways that influence the phenotype, such as studying the urinary fractional excretion of urate, may improve replication between studies. Strengths of this study include linkage analysis in two separate networks allowing independent assessment for evidence of replication. In addition, heritability and linkage analysis of uric acid was performed in both non-Hispanic black and non-Hispanic white cohorts. Family-based studies using non-elderly hypertensive probands are particularly relevant given the potential link between hyperuricaemia and the early development of primary hypertension [42]. The pathway analysis provided a systematic approach to identifying candidate genes under uric acid QTLs. There were also several limitations to this study. Dietary data were not available, and dietary purines are known to have a large influence on serum uric acid levels. Diuretic medication use was common, as expected in families enriched for hypertension, but adjustment for diuretic use in the analyses may not have adequately accounted for all the variability in uric acid due to different diuretic types and doses. The genetic effects on uric acid over a lifetime may not be adequately captured with a single uric acid level. The QTL for uric acid identified in GENOA whites and HyperGEN blacks were attenuated with multivariable adjustment. Both agesex-adjusted and multivariable-adjusted uric acid phenotypes are relevant. Several disease pathways that may be affected by uric acid levels can be overadjusted for, including insulin resistance, hypertension, cardiovascular disease and kidney disease [3]. In conclusion, QTLs for uric acid with evidence of pleiotropic effects on other surrogates of insulin resistance or serum creatinine were identified on chromosomes 8, 9 and 16 in white families. Genes in the fructose metabolism pathway were enriched at these loci, providing a plausible biological mechanism that may underlie these findings. Quantitative trait loci for uric acid on chromosome 4 were found in blacks, with one locus that showed some evidence of replication and another locus in the region of the SLC2A9 gene. These loci could be explored further with positional cloning or candidate gene association studies. However, given the overall lack of replication of loci for uric acid identified with linkage analysis, alternative approaches may be needed. Genome-wide association studies have shown

Downloaded from ndt.oxfordjournals.org by guest on May 15, 2011

2420

A. D. Rule et al. 18. Vitart V, Rudan I, Hayward C et al. SLC2A9 is a newly identified urate transporter influencing serum urate concentration, urate excretion and gout. Nat Genet 2008; 40: 437442 19. Nakagawa T, Mazzali M, Kang DH et al. Uric acida uremic toxin? Blood Purif 2006; 24: 6770 20. Sanchez-Lozada LG, Tapia E, Santamaria J et al. Mild hyperuricemia induces vasoconstriction and maintains glomerular hypertension in normal and remnant kidney rats. Kidney Int 2005; 67: 237247 21. Maradit Kremers H, Crowson CS, Gabriel SE. Rochester Epidemiology Project: a unique resource for research in the rheumatic diseases. Rheum Dis Clin North Am 2004; 30: 81934, vii 22. Turner ST, Kardia SL, Mosley TH et al. Influence of genomic loci on measures of chronic kidney disease in hypertensive sibships. J Am Soc Nephrol 2006; 17: 20482055 23. Rock RC, Walker WG, Jennings CD. Nitrogen metabolites and renal function. In: TIetz NW (ed.). Fundamentals of Clinical Chemistry, 3rd edn. Philadelphia: Saunders, 1987, 684686 24. McCulloch CE, Searle SR. Generalized, Linear, and Mixed Models. New York: Wiley, 2001 25. Amos CI. Robust variance-components approach for assessing genetic linkage in pedigrees. Am J Hum Genet 1994; 54: 535543 26. Almasy L, Blangero J. Multipoint quantitative-trait linkage analysis in general pedigrees. Am J Hum Genet 1998; 62: 11981211 27. Hopper JL, Mathews JD. Extensions to multivariate normal models for pedigree analysis. Ann Hum Genet 1982; 46: 373383 28. Sobel E, Lange K. Descent graphs in pedigree analysis: applications to haplotyping, location scores, and marker-sharing statistics. Am J Hum Genet 1996; 58: 13231337 29. Sobel E, Sengul H, Weeks DE. Multipoint estimation of identityby-descent probabilities at arbitrary positions among marker loci on general pedigrees. Hum Hered 2001; 52: 121131 30. Morton NE. Significance levels in complex inheritance. [Erratum appears in Am J Hum Genet 1998;63(4):1252]. Am J Hum Genet 1998; 62: 690697 31. Wise LH, Lanchbury JS, Lewis CM. Meta-analysis of genome searches. Ann Hum Genet 1999; 63: 263272 32. Williams JT, Van Eerdewegh P, Almasy L et al. Joint multipoint linkage analysis of multivariate qualitative and quantitative traits: I. likelihood formulation and simulation results. Am J Hum Genet 1999; 65: 11341147 33. Self S, Liang KY. Asymptotic properties of maximum likelihood estimators and likelihood ratio tests under nonstandard conditions. J Am Stat Assoc 1987; 82: 605610 34. Turner ST, Kardia SL, Boerwinkle E et al. Multivariate linkage analysis of blood pressure and body mass index. Genet Epidemiol 2004; 27: 6473 35. Hanis CL, Boerwinkle E, Chakraborty R et al. A genome-wide search for human non-insulin-dependent (type 2) diabetes genes reveals a major susceptibility locus on chromosome 2. Nat Genet 1996; 13: 161166 36. Horikawa Y, Oda N, Cox NJ et al. Genetic variation in the gene encoding calpain-10 is associated with type 2 diabetes mellitus. Nat Genet 2000; 26: 163175 37. Barroso I. Genetics of Type 2 diabetes. Diabet Med 2005; 22: 517535 38. Emmerson BT. Effect of oral fructose on urate production. Ann Rheum Dis 1974; 33: 276280 39. Schulze MB, Manson JE, Ludwig DS et al. Sugar-sweetened beverages, weight gain, and incidence of type 2 diabetes in young and middle-aged women. [See comment]. JAMA 2004; 292: 927934 40. Gaby AR. Adverse effects of dietary fructose. Altern Med Rev 2005; 10: 294306 41. Oberhaensli RD, Rajagopalan B, Taylor DJ et al. Study of hereditary fructose intolerance by use of 31P magnetic resonance spectroscopy. Lancet 1987; 2: 931934 42. Feig DI, Johnson RJ. Hyperuricemia in childhood primary hypertension. Hypertension 2003; 42: 247252 Received for publication: 16.7.08; Accepted in revised form: 9.2.09

successful replication of loci for uric acid and may be the optimal approach for identifying genomic regions that influence uric acid levels.
Acknowledgements. This study was supported by the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK073537, K23 DK078229). Conflict of interest statement. None declared.

References
1. Lin KC, Lin HY, Chou P. Community based epidemiological study on hyperuricemia and gout in Kin-Hu, Kinmen. J Rheumatol 2000; 27: 10451050 2. Forman JP, Choi H, Curhan GC. Plasma uric acid level and risk for incident hypertension among men. J Am Soc Nephrol 2007; 18: 287 292 3. Short RA, Tuttle KR. Clinical evidence for the influence of uric acid on hypertension, cardiovascular disease, and kidney disease: a statistical modeling perspective. Semin Nephrol 2005; 25: 2531 4. Johnson RJ, Segal MS, Srinivas T et al. Essential hypertension, progressive renal disease, and uric acid: a pathogenetic link? J Am Soc Nephrol 2005; 16: 19091919 5. Weiner DE, Tighiouart H, Elsayed EF et al. Uric acid and incident kidney disease in the community. J Am Soc Nephrol 2008; 19: 1204 1211 6. Obermayr RP, Temml C, Gutjahr G et al. Elevated uric acid increases the risk for kidney disease. J Am Soc Nephrol 2008; 19: 2407 2413 7. Wilk JB, Djousse L, Borecki I et al. Segregation analysis of serum uric acid in the NHLBI Family Heart Study. Hum Genet 2000; 106: 355359 8. Rice T, Borecki IB, Bouchard C et al. Commingling analysis of regional fat distribution measures: the Quebec Family Study. Int J Obes Relat Metab Disord 1992; 16: 831844 9. Yang Q, Guo CY, Cupples LA et al. Genome-wide search for genes affecting serum uric acid levels: the Framingham Heart Study. Metabolism 2005; 54: 14351441 10. Nath SD, Voruganti VS, Arar NH et al. Genome scan for determinants of serum uric acid variability. J Am Soc Nephrol 2007 11. Tang W, Miller MB, Rich SS et al. Linkage analysis of a composite factor for the multiple metabolic syndrome: the National Heart, Lung, and Blood Institute Family Heart Study. Diabetes 2003; 52: 2840 2847 12. Williams RR, Rao DC, Ellison RC et al. NHLBI family blood pressure program: methodology and recruitment in the HyperGEN network. Hypertension genetic epidemiology network. Ann Epidemiol 2000; 10: 389400 13. Wang Y, Wang QJ. The prevalence of prehypertension and hypertension among US adults according to the new joint national committee guidelines: new challenges of the old problem. Arch Intern Med 2004; 164: 21262134 14. Mayes PA. Intermediary metabolism of fructose. Am J Clin Nutr 1993; 58: 754S765S 15. Heinig M, Johnson RJ. Role of uric acid in hypertension, renal disease, and metabolic syndrome. Cleve Clin J Med 2006; 73: 1059 1064 16. Foley RN, Wang C, Ishani A et al. NHANES III: influence of race on GFR thresholds and detection of metabolic abnormalities. J Am Soc Nephrol 2007; 18: 25752582 17. Wallace C, Newhouse SJ, Braund P et al. Genome-wide association study identifies genes for biomarkers of cardiovascular disease: serum urate and dyslipidemia. Am J Hum Genet 2008; 82: 139149

Downloaded from ndt.oxfordjournals.org by guest on May 15, 2011