Variations in RANK gene are associated with adult height in Caucasians

Original Research Article

Variations in RANK Gene are Associated withAdult Height in Caucasians

YUAN CHEN,1,2,3{ DONG-HAI XIONG,2,3{ TIE-LIN YANG,4 FANG YANG,1,2,3 HUI JIANG,1,2,3

FENG ZHANG,2,3,4 HUI SHEN,5,6 PENG XIAO,2,3 ROBERT R. RECKER,2,3

AND HONG-WEN DENG1,4,5,6*1Laboratory of Molecular and Statistical Genetics, College of Life Science, Hunan Normal University,Changsha, Hunan 410081, People’s Republic of China2Osteoporosis Research Center, Creighton University, Omaha, Nebraska 681313Department of Biomedical Sciences, Creighton University, Omaha, Nebraska 681314The Key Laboratory of Biomedical Information Engineering of Ministry of Education and Institute ofMolecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University,Xi’an 710049, People’s Republic of China5Department of Orthopedic Surgery, School of Medicine, University of Missouri-Kansas City,Kansas City, Missouri 641086Department of Basic Medical Sciences, School of Medicine, University of Missouri-Kansas City,Kansas City, Missouri 64108

ABSTRACT Height is a complex trait significantly influenced by genetic factors, with herit-ability ranging from 48% to 98%. Previous studies have yielded a number of important genomicregions that may account for the variation of height in human populations. However, more‘height’ genes still wait for identification. Recent studies have revealed that tumor necrosis fac-tor receptor superfamily member 11a (RANK) is a vital factor for chondroclastic/osteoclastic dif-ferentiation and activity that influence the morphology of growth plates and linear bone growth.Despite its importance, little effort has been made to find out whether the RANK polymor-phisms are associated with adult height variation in normal populations. Herein, we performeda family based association test (FBAT) in 1873 white subjects from 405 nuclear families. Amongeighteen single nucleotide polymorphisms (SNPs) and seven blocks, SNP rs6567274 wasdetected to be significant even after multiple-testing correction. In corroboration with single-locus analysis, a major haplotype in block 5 bearing the variant ‘‘T’’ of rs6567274 was signifi-cantly associated with higher stature. Our findings firstly suggested the RANK polymorphismsmight contribute to adult height variation. Further researches need to be launched to replicatethe present results and further unravel the molecular mechanism underlying the significantassociations discovered. Am. J. Hum. Biol. 19:559–565, 2007. ' 2007 Wiley-Liss, Inc.

Height is a complex trait mainly determinedby genetic factors, with the heritability rang-ing from 48% to 98% (Brown et al., 2003; Denget al., 2002; Perola et al., 2001; Wu et al.,2003). Certain progress has been made inunderstanding genetic mechanisms underly-ing bone growth, such as whole genome link-age scan for height and detailed candidategene association studies (Bargerlux et al.,1995; Deng et al., 2002; Lehrer et al., 1994;Wang et al., 2004). Particularly, the Framing-ham Heart Study (Geller et al., 2003) reporteda LOD score of 1.73 in the 18q22.1 region thatharbors RANK (tumor necrosis factor receptorsuperfamily member 11a), a vital gene forchondroclastic/osteoclastic differentiation andactivity (Dougall et al., 1999; Henriksen et al.,2003; Li et al., 2000; Xu et al., 2006). However,currently no association study concerning the

contribution of RANK polymorphisms to heightvariation has been done.Tumor necrosis factor receptor superfamily

member 11a is a pivotal factor in the RANK/

{Yuan Chen and Dong-Hai Xiong contributed equally tothis work.

Contract grant sponsor: National Institute of Health,the State of Nebraska; Contract grant numbers: R01AR050496, K01 AR02170-01, R01 AR45349-01, and R01GM60402-01A1; Contract grant sponsor: Hunan ProvincialNatural Science Foundation of China; Contract grant num-bers: 05JJ40051, 04JJ1004.

*Correspondence to: Hong-Wen Deng, Ph. D., Laboratoryof Molecular and Statistical Genetics, College of Life Scien-ces, Hunan Normal University, Changsha, Hunan 410081,People’s Republic of China. E-mail: [email protected]

Received 2 October 2006; Revision received 22 November2006; Accepted 25 November 2006

Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ajhb.20619

AMERICAN JOURNAL OF HUMAN BIOLOGY 19:559–565 (2007)

VVC 2007 Wiley-Liss, Inc.

RANKL pathway and plays a key role inosteoclastogenesis, chemotaxis, function andsurvival (Dougall et al., 1999; Henriksenet al., 2003; Li et al., 2000; Xu et al., 2006).Animal model studies found that reducedchondroclastic/osteoclastic differentiation andactivity would delay ossification, alter themorphology of growth plates and impair lin-ear bone growth (Sanchez et al., 2000; Turneret al., 1994). Moreover, RANK deficient micealso presented reduction in the resorption ofnewly formed calcified cartilage within theprimary spongiosa and a widened hyper-trophic cartilage zone of the epiphyseal growthplate at the ends of long bones in the develop-ing skeleton (Li et al., 2000). Therefore, givenits important role in bone growth, we canhypothesize RANK as a ‘‘candidate gene’’ andits genetic variants may be associated withthe variation of adult height.Herein, using the robust family based asso-

ciation test (FBAT), we tested 18 single nucle-otide polymorphisms (SNPs) and seven haplo-type blocks covering the full transcript lengthof the RANK gene so as to make a good profil-ing of the role of RANK in genetic determina-tion of height in a large sample of 1,873 whitesubjects from 405 nuclear families.

MATERIALS AND METHODS

Patient recruitment and sample collection

The study was approved by the CreightonUniversity Institutional Review Board. Signedinformed-consent documents were obtainedfrom all study participants before theyentered the study. The design and samplingprocedures have been published before (Xionget al., 2005). All height measurements with-out shoes were made using a standard wallmounted statiometer in the clinic by nurses.In addition, the average of two repeated mea-surements for height was used as the finaldata of height. In brief, there were a total of1,873 subjects from 405 nuclear families inour sample, including 740 founders and 1,133nonfounders, all of them were US Caucasiansof European origin. Average family size was4.86 (3–12). To reduce the bias, specific crite-ria for enrollment in terms of osteoporosis,fracture, or BMD status were not set, exceptthat people with bone metabolic diseases wereexcluded. Our linkage disequilibrium (LD)and haplotype analysis focused on the 703unrelated parents from the 405 nuclear fami-lies, while the association tests were donein all of the subjects. The greater sample

size we adopted allowed for more confidentpredictions.

Genotyping

Genomic DNA was extracted from wholeblood using a commercial isolation kit (GentraSystems, Minneapolis, MN) following the pro-cedure listed on the kit. DNA concentrationwas assessed by a DU530 UV/VIS Spectropho-tometer (Beckman Coulter, Fullerton, CA).We selected single nucleotide polymorphisms(SNPs) mainly according to the public infor-mation available in dbSNP [genome build 34]and SNPper (http://snpper.chip.org/). A totalof 18 SNPs in and around RANK (tumor ne-crosis factor receptor superfamily member11a) were selected on the basis of the followingcriteria: (1) validation status, especially inCaucasians, (2) an average density of 1 SNPper 3 kb, (3) degree of heterozygosity, i.e.,minor allele frequencies (MAF)> 0.05, (4) func-tional relevance and importance, (5) reported todbSNP by various sources. All these SNPswere successfully genotyped to calculate allelefrequencies using the high-throughput Bea-dArray SNP genotyping technology of Illu-mina. (San Diego, CA). The average rate ofmissing genotype data was reported by Illu-mina to be *0.05% and the average genotyp-ing error rate estimated through blind dupli-cating was reported to be less than *0.01%.The SNPs were spaced*5 kb apart on averageand covered the full transcript length of theRANK gene.

Statistical analysis

As implemented in the program packageSOLAR (Sequential Oligogenic Linkage Anal-ysis Routines, available at http://www.sfbr.org/solar/), the univariate variance compo-nents approaches, considering the familystructure and kinships, were employed to esti-mate the narrow-sense heritability (h2) and toevaluate the effects of the two assumed cova-riates (sex, age) on the studied phenotype.

PedCheck (O’Connell and Weeks, 1998) wasused to check Mendelian consistency of SNPgenotype data and all inconsistent genotypeswere removed. Then the error checking optionin Merlin (Abecasis et al., 2002) was used toidentify and disregard the genotypes flankingexcessive recombinants so as to further reducegenotyping errors. The allele frequencies of allthe SNPs were calculated by allele countingand tested for departure fromHardy-Weinberg

560 Y. CHEN ET AL.

American Journal of Human Biology DOI 10.1002/ajhb

equilibrium using the PEDSTATS procedureembedded in Merlin. Population haplotypesand their frequencies were inferred usingPHASE v2.1.1 software among 703 unrelatedparents. LD structure was defined by GOLD(http://www.sph.umich.edu/csg/abecasis/GOLD/)to chart pairwise |D’| statistics derived fromhaplotype data. HaploBlockFinder (http://cgi.uc.edu/cgi-bin/kzhang/haploBlockFinder.cgi/)was used to identify block structures andselect haplotype-tagging SNPs (htSNPs) ofeach candidate gene. To infer haplotypesdefined by the htSNPs within each block forall of the subjects among 405 families, weadopted the algorithm of integer linear pro-gramming (ILP) implemented in PedPhaseV2.0 (http://www.cs.ucr.edu/*jili/haplotyping.html), which is based on LD assumption andable to recover phase information at eachmarker locus with great speed and accuracyeven in the presence of 20% missing data (Liand Jiang, 2005).Associations of single loci and height were

analyzed by the family-based association tests(FBAT, available online at http://www.biostat.harvard.edu/*fbat/default.html) and only in-formative families (i.e., families with a non-zero contribution to the test statistic) wereincluded. Since confounding covariates andgene/covariate interactions could not be di-rectly included in FBAT-statistics, regressionresiduals representing age-, sex-adjusted heightvalues were used in the following analyses toincrease statistical efficiency. Normality testsfor these residuals were performed by MINI-TAB (Minitab, State College, PA). Both addi-tive and non-additive models (recessive) wereused to perform FBAT and individual haplo-type, because segregation analysis suggestedthat some major recessive genes should regu-late adult height variation (Ginsburg et al.,1998; Li et al., 2004; Xu et al., 2002), with amodel that included a major recessive geneand residual polygenic effect best fitted heightdata (Xu et al., 2002). Global haplotype testsof association were performed under multial-lelic mode using the haplotype-based associa-tion test (HBAT) in the FBAT program, wherehaplotypes were treated as alleles of a multi-allelic marker. P values under nominal signifi-cance thresholds (P < 0.05) were furtherexamined using Monte-Carlo permutationprocedures implemented in HBAT (10,000permutations were conducted) under the nulldistribution of no linkage and no association.SNP spectral decomposition was used to cor-rect for multiple testing (Nyholt, 2004). There-

fore, the single-test threshold for experiment-wide significance to keep type I error rate at5% was set to 0.0043.

RESULTS

LD and haplotype analyses

All the 1,873 subjects were age >19 and hadreached (or nearly so) their final height whenrecruited (Xu et al., 2002). The mean height ofthe 1,124 women was 1.64 (SD ¼ 0.06), whilethe mean height of the 749 men was 1.78 (SD¼ 0.07). After adjusting for the significantcovariates of age and sex (P < 0.001), theresiduals of height of all the subjects followedthe normal distribution (P > 0.15) and thenarrow-sense heritability of height was about59%.The allele frequencies of all the studied

SNP markers in the parental population ofour Caucasian sample obeyed the Hardy-Weinberg equilibrium. As indicated in Figure 1,seven LD blocks were identified, ranging insize from 5 kb to 19 kb. Fourteen htSNPs(marked with asterisk) were needed to repre-sent the common haplotypes (with populationfrequencies >5%) of the overall seven LDblocks. Blocks 1 and 2 were located in intron1. Block 3 extended from intron 2 to intron 3,while block 4 was from intron 3 to intron 4.Block 5 ranged from intron 7 to intron 9, andblock 7 encompassed the 30-UTR. No strongLD was found between SNP15 (rs9646629)with any other SNPs, so it was designatedalone as block 6.

Association analyses

The basic information of 18 studied SNPs islisted in Table 1, mainly including minor allelefrequency (MAF), map location, and polymor-phism information content (PIC). Table 2shows the results of FBAT analysis for 14htSNPs and additional 4 SNPs. Under the re-cessive model, three successive SNPs in intron9 near the 30 UTR of RANK all yielded P values<0.05. Among them, SNP12 (i.e., rs6567274,P ¼ 0.0032) and SNP14 (i.e., rs4426449, P ¼0.0047) were highly significant for adult heightvariation. Nominally significant association(P values �0.05) was also observed for SNP13(P¼ 0.0299). Under the additive model, SNP12and SNP14 remained nominally significant(P ¼ 0.030 and 0.036, respectively), while thecorresponding P value for SNP13 was 0.095.Table 3 tabulated the HBAT results regardingthe haplotype association tests for height underthe two models. Using the recessive model, a

561RANK VARIATIONS AND ADULT HEIGHT


significant association with height was foundfor one major haplotype in block 5, namelyhap1 (i.e., ‘‘T-G-T’’ with frequency of 0.326, P¼ 0.0073), which showed preferential trans-mission to the taller offspring (Z ¼ 2.681).

Under the additive model, block 5-hap1 wasnominally significant for height variation (Pvalue & 0.05). The results of global haplotypetest by HBAT are summarized into Table 4.Haplotype block 5 was still marginally signifi-

TABLE 1. The information of all of the studied 18 SNPs

Marker Nucleotide MAF Regiona Phys_loc (bp) Distance (bps) PIC

rs4941125 A/G 0.3132 Intron 1 58148664 0 0.38rs7235803 A/G 0.3372 Intron 1 58151359 2,695 0.4rs4436867 A/C 0.2363 Intron 1 58154624 3,265 0.33rs8086340 C/G 0.4436 Intron 1 58157958 3,334 0.44rs12956925 A/G 0.1691 Intron 1 58164620 6,662 0.26rs3826619 A/G 0.1055 Intron 2 58166730 2,110 0.18rs11664594 A/T 0.348 Intron 3 58169186 2,456 0.4rs3826620 G/T 0.2811 Intron 3 58172484 3,298 0.36rs12969194 A/T 0.2702 Intron 4 58175045 2,561 0.36rs4303637 C/T 0.3214 Intron 7 58182743 7,698 0.39rs17069904 A/G 0.097 Intron 7 58183929 1,186 0.17rs6567274 G/T 0.3455 Intron 9 58188780 4,851 0.4rs12959396 G/T 0.4766 Intron 9 58190289 1,509 0.44rs4426449 C/T 0.3447 Intron 9 58193797 3,508 0.4rs9646629 C/G 0.3572 Intron 9 58202179 8,382 0.41rs884205 G/T 0.2602 30 UTR 58205837 3,658 0.35rs2957127 C/T 0.4507 30 UTR 58210345 4,508 0.43rs3017365 A/G 0.4908 30 UTR 58214321 3,976 0.44

MAF, minor allele frequency.Phys_loc, physical location.Distance, physical distance between two adjacent SNPs.PIC, polymorphism information content, calculated by the online calculator available at http://www.agri.huji.ac.il/*weller/Hayim/parent/PIC.aNo extronic SNP is listed since they are rare (<1%) or in strong LD with other selected SNP.

Fig. 1. Schematic representation of RANK SNPs analyzed in this study, together with the LD structures ofRANK defined by these SNPs. The 14 htSNPs (haplotype-tagging SNPs) were marked with asterisk. [Color figurecan be viewed in the online issue, which is available at www.interscience.wiley.com.]

562 Y. CHEN ET AL.


cant for height variation in the global testing,under both the additive and recessive models(P value ¼ 0.061 and 0.066, respectively).However, except for the association for SNP12

under recessive model, other associations didnot remain significant after correction formultiple testing based on the SNP spectraldecomposition approach.

TABLE 3. Haplotype-specific tests for association of RANK with height under the recessive and additive models

Block Hap no Haplotype Frequency ZR PR* ZA PA*

Block 1 Hap1 A-C 0.454 1.457 0.145 0.413 0.680Hap2 G-C 0.320 0.436 0.663 0.464 0.642Hap3 A-A 0.225 0.007 0.994 0.04 0.968

Block 2 Hap1 G-G 0.441 0.573 0.567 0.498 0.618Hap2 C-G 0.383 1.418 0.156 0.487 0.626Hap3 C-A 0.168 0.990 0.322 0.329 0.741Hap4 G-A 0.008

Block 3 Hap1 G-T 0.527 1.368 0.171 1.318 0.187Hap2 G-A 0.365 0.210 0.834 1.051 0.293Hap3 A-T 0.108 0.183 0.855 0.452 0.652

Block 4 Hap1 G-T 0.444 0.385 0.700 0.132 0.895Hap2 T-T 0.287 1.104 0.269 0.427 0.669Hap3 G-A 0.269 0.129 0.370 0.304 0.761Hap4 T-A 0.001

Block 5 Hap1 T-G-T 0.326 2.681 0.0073 (0.0067) 2.153 0.031 (0.034)Hap2 C-G-G 0.318 0.884 0.377 0.024 0.981Hap3 T-G-G 0.252 0.954 0.340 2.630 0.009 (0.009)Hap4 T-A-G 0.101 0.013 0.990 0.316 0.752Hap5 C-G-T 0.002Hap6 C-A-G 0.001

Block 6 Hap1 G 0.646 0.119 0.905 0.182 0.855Hap2 C 0.354 0.465 0.642 0.182 0.855

Block 7 Hap1 G-A 0.488 1.695 0.090 1.075 0.282Hap2 G-G 0.256 0.022 0.983 0.873 0.383Hap3 T-G 0.252 0.198 0.843 0.311 0.755Hap4 T-A 0.004

R, recessive model; A, additive model.The empty cells indicate that the testing cannot be conducted because the number of informative families was too small to allow for ameaningful test.*P values under nominal significance thresholds are indicated in bold and confirmed with 10,000 permutations, numbers in paren-thesis indicate empirical P values.

TABLE 2. FBAT for association of RANK SNPs with height

SNP SNP no Allele Fam noR Fam noA ZR PR* ZA PA*

rs4941125 SNP1 G 106 261 0.679 0.497 �0.185 0.853rs7235803 SNP2 G 113 272 0.262 0.793 �0.173 0.863rs4436867 SNP3 A 60 209 0.162 0.871 0.093 0.926rs8086340 SNP4 G 175 280 0.696 0.486 �0.634 0.526rs12956925 SNP5 A 36 195 1.511 0.131 �0.051 0.959rs3826619 SNP6 A 22 139 0.749 0.454 �0.352 0.725rs11664594 SNP7 A 135 282 0.135 0.893 �0.914 0.361rs3826620 SNP8 T 94 262 1.265 0.206 �0.301 0.763rs12969194 SNP9 A 70 248 0.791 0.429 0.236 0.814rs4303637 SNP10 C 93 260 0.974 0.330 0.127 0.899rs17069904 SNP11 A 18 131 0.348 0.728 0.08 0.937rs6567274 SNP12 T 108 265 2.945 0.0032 (0.0036) 2.173 0.030 (0.030)rs12959396 SNP13 G 184 289 2.171 0.0299 (0.0367) 1.671 0.095rs4426449 SNP14 T 110 265 2.824 0.0047 (0.0054) 2.101 0.036 (0.044)rs9646629 SNP15 C 109 271 0.464 0.643 �0.182 0.856rs884205 SNP16 T 67 229 0.008 0.993 �0.411 0.681rs2957127 SNP17 T 171 287 0.265 0.791 �0.328 0.743rs3017365 SNP18 A 179 288 1.571 0.116 1.067 0.286

All alleles given in the ‘Allele’ column represent the minor allele of each SNP. SNPs are listed in order (30?50).Fam no, the number of informative families; R, recessive model; A, additive model.*P values under nominal significance thresholds are indicated in bold and further confirmed with 10,000 permutations, numbers inparenthesis indicate empirical P values.



DISCUSSION

Receptor activator of nuclear factor k B,also named RANK/TRANCER/EOF, is a mem-ber of TNF receptor gene family. It is presentin osteoclast precursors and plays the pivotalrole in the RANK/RANKL pathway, which issignificantly associated with endochondralossification and bone growth (Dougall et al.,1999; Henriksen et al., 2003; Li et al., 2000).Studies using different approaches indicatedRANK as an important bone candidate gene(Geller et al., 2003; Sanchez et al., 2000;Turner et al., 1994). However, little has beenknown about the influence of RANK polymor-phisms on the variation of adult height. Toexplore such relationship, we used 18 SNPscomprehensively covering RANK in a largeset of family samples and analyzed the datawith FBAT software.This software is an extension of the trans-

mission disequilibrium test (TDT) and candeal with any kind of pedigree structure,including incomplete nuclear-families and allpatterns of missing marker allele information(Horvath et al., 2001). In the FBAT analysis,the direction of a genetic effect can be deter-mined by the sign of the test statistic, withnegative (positive) Z scores indicating geno-types that are associated with trait valuesless (more) than the population means (Hor-vath et al., 2001). Our FBAT results providedsuggestive evidence that polymorphisms inthe RANK gene were associated with height.We found that before correction for multipletesting, SNP12 and SNP14 showed positiveassociation with adult height under both therecessive and additive models. So was SNP13under the recessive model. Especially, SNP12

was found to be most positively associatedwith height variation under recessive model,and met the experiment-wide stringent crite-rion (P < 0.0043). This locus could be a start-ing point for the future replication studies inother populations or for the further molecularand functional research.

Haplotype based association analysis wasalso adopted to test the association betweenRANK and adult height. The results revealedthe significant transmission disequilibriumfor one major haplotype (frequency of 32.6%),namely the block 5-hap1 containing the threeloci combination ‘‘T-G-T’’. Noticeably, this hap-lotype contained the allele ‘‘T’’ of SNP12 andthus corroborated prior single SNP FBATanalysis. The fact that SNP12 was locatednear the 30-UTR may suggest its regulatoryrole in RANK expression, possibly by affectingthe splicing mechanisms. However, the signifi-cant associations of SNP12 and block 5-hap1with height may also be explained by their LDwith other truly functional SNPs in or nearthe RANK block 5 region. Molecular and func-tional studies are necessary to reveal theunderlying mechanism of the associations weobserved.

LITERATURE CITED

Abecasis GR, Cherny SS, Cookson WO, Cardon LR. 2002.Merlin-rapid analysis of dense genetic maps using sparsegene flow trees. Nat Genet 30:97–101.

Bargerlux MJ, Heaney RP, Hayes J, Deluca HF, JohnsonML, Gong G. 1995. Vitamin-D-receptor gene polymorphism,bone mass, body-size, and vitamin-D-receptor density. CalcifTissue Int 57:161–162.

Brown WM, Beck SR, Lange EM, Davis CC, Kay CM, Lan-gefeld CD, Rich SS. 2003. Age-stratified heritability esti-mation in the Framingham Heart Study families. BMCGenet 4(Suppl 1):S32.

TABLE 4. Global (multi-haplotype) tests for association of RANK with height usingHBATunder the recessive and additive models

Block Model htSNPs Allele DF w2 P

Block 1 R SNP1-3 3 3 2.445 0.485A 2 0.240 0.887

Block 2 R SNP4-5 4 3 3.473 0.324A 3 5.139 0.162

Block 3 R SNP6-7 3 3 1.917 0.590A 2 1.738 0.419

Block 4 R SNP8-9 4 3 2.351 0.503A 2 0.198 0.906

Block 5 R SNP10-11-12 6 4 8.821 0.066A 4 9.017 0.061

Block 6 R SNP15 2 2 0.245 0.884A 1 0.033 0.855

Block 7 R SNP16-18 4 3 2.898 0.408A 3 1.290 0.731

R, recessive model; A, additive model; DF, degree of freedom.w2: chi-square FBAT test value.

564 Y. CHEN ET AL.


Deng HW, Xu FH, Liu YZ, Shen H, Deng HY, Huang QY,Liu YJ, Conway T, Li JL, Davies KM, Recker RR. 2002. Awhole-genome linkage scan suggests several genomicregions potentially containing QTLs underlying the vari-ation of stature. Am J Med Genet 113:29–39.

Dougall WC, Glaccum M, Charrier K, Rohrbach K, BraselK, De Smedt T, Daro E, Smith J, Tometsko ME, Malis-zewski CR, Armstrong A, Shen V, Bain S, Cosman D,Anderson D, Morrissey PJ, Peschon JJ, Schuh J. 1999.RANK is essential for osteoclast and lymph node develop-ment. Genes Dev 13:2412–2424.

Geller F, Dempfle A, Gorg T. 2003. Genome scan for bodymass index and height in the Framingham Heart Study.BMC Genet 4(Suppl 1):S91.

Ginsburg E, Livshits G, Yakovenko K, Kobyliansky E.1998. Major gene control of human body height, weightand BMI in five ethnically different populations. AnnHum Genet 62(Part 4):307–322.

Henriksen K, Karsdal M, Delaisse JM, Engsig MT. 2003.RANKL and vascular endothelial growth factor (VEGF)induce osteoclast chemotaxis through an ERK1/2-de-pendent mechanism. J Biol Chem 278:48745–48753.

Horvath S, Xu X, Laird NM. 2001. The family based associ-ation test method: strategies for studying general geno-type–phenotype associations. Eur J Hum Genet 9:301–306.

Lehrer S, Rabin J, Stone J, Berkowitz GS. 1994. Associationof an estrogen-receptor variant with increased height inwomen. Horm Metab Res 26:486–488.

Li J, Jiang T. 2005. Computing the minimum recombinanthaplotype configuration from incomplete genotype dataon a pedigree by integer linear programming. J ComputBiol 12:719–739.

Li J, Sarosi I, Yan XQ, Morony S, Capparelli C, Tan HL,McCabe S, Elliott R, Scully S, VanG, Kaufman S, Juan SC,Sun Y, Tarpley J,Martin L, ChristensenK,McCabe J, Kos-tenuik P, Hsu H, Fletcher F, Dunstan CR, Lacey DL, BoyleWJ. 2000. RANK is the intrinsic hematopoietic cell surfacereceptor that controls osteoclastogenesis and regulation ofbone mass and calcium metabolism. Proc Natl Acad SciUSA 97:1566–1571.

Li MX, Liu PY, Li YM, Qin YJ, Liu YZ, Deng HW. 2004. Amajor gene model of adult height is suggested in Chinese.J Hum Genet 49:148–153.

Nyholt DR 2004. A simple correction for multiple testingfor single-nucleotide polymorphisms in linkage disequili-brium with each other. Am J Hum Genet 74:765–769.

O’Connell JR, Weeks DE. 1998. PedCheck: A program foridentification of genotype incompatibilities in linkageanalysis. Am J Hum Genet 63:259–266.

Perola M, Ohman M, Hiekkalinna T, Leppavuori J, Paju-kanta P, Wessman M, Koskenvuo M, Palotie A, Lange K,Kaprio J, Peltonen L. 2001. Quantitative-trait-locusanalysis of body-mass index and of stature, by combinedanalysis of genome scans of five Finnish study groups.Am J Hum Genet 69:117–123.

Sanchez CP, Kuizon BD, Abdella PA, Juppner H, SaluskyIB, Goodman WG. 2000. Impaired growth, delayed ossifi-cation, and reduced osteoclastic activity in the growthplate of calcium-supplemented rats with renal failure.Endocrinology 141:1536–1544.

Turner RT, Evans GL, Wakley GK. 1994. Reduced chondro-clast differentiation results in increased cancellous bonevolume in estrogen-treated growing rats. Endocrinology134:461–466.

Wang HJ, Geller F, Dempfle A, Schauble N, Friedel S,Lichtner P, Fontenla-Horro F, Wudy S, Hagemann S,Gortner L, Huse K, Remschmidt H, Bettecken T, Mei-tinger T, Schafer H, Hebebrand J, Hinney A. 2004. Ghre-lin receptor gene: identification of several sequence var-iants in extremely obese children and adolescents,healthy normal-weight and underweight students, andchildren with short normal stature. J Clin EndocrinolMetab 89:157–162.

Wu X, Cooper RS, Boerwinkle E, Turner ST, Hunt S, MyersR, Olshen RA, Curb D, Zhu X, Kan D, Luke A. 2003.Combined analysis of genomewide scans for adult height:results from the NHLBI Family Blood Pressure Program.Eur J Hum Genet 11:271–274.

Xiong DH, Liu YZ, Liu PY, Zhao LJ, Deng HW. 2005. Associa-tion analysis of estrogen receptor a gene polymorphismswith cross-sectional geometry of the femoral neck inCauca-sian nuclear families. Osteoporos Int 16:2113–2122.

Xu D, Wang S, Liu W, Liu J, Feng X. 2006. A novel receptoractivator of NF-kB (RANK) cytoplasmic motif plays anessential role in osteoclastogenesis by committing macro-phages to the osteoclast lineage. J Biol Chem 281:4678–4690.

Xu J, Bleecker ER, Jongepier H, Howard TD, KoppelmanGH, Postma DS, Meyers DA. 2002. Major recessivegene(s) with considerable residual polygenic effect regu-lating adult height: confirmation of genomewide scanresults for chromosomes 6, 9, and 12. Am J Hum Genet71:646–650.



Documents

Variations in RANK gene are associated with adult height in Caucasians