International Journal of Obesity (2000) 24, 206±210 ß 2000 Macmillan Publisher Ltd All rights reserved 0307±0565/00 $15.00 www.nature.com/ijo
Melanocortin 3 receptor (MC3R) gene variants in extremely obese women W-D Li1, E-J Joo1#, EB Furlong2, M Galvin2, K Abel2, CJ Bell2 and R Arlen Price1* 1
Department of Psychiatry, University of Pennsylvania, Philadelphia, PA, USA; and 2Axys Pharmaceuticals, La Jolla, CA, USA
OBJECTIVE: Following several reports of linkage of obesity related phenotypes to human chromosome 20q we sought to determine whether variations of the melanocortin 3 receptor (MC3R) gene are associated with obesity. DESIGN: We screened the MC3R gene coding region and approximately 2 kb of 50 and 30 ¯anking sequences for DNA variants in unrelated extremely obese women and average weight controls using polymerase chain reaction (PCR) single strand conformation polymorphism (SSCP) analysis and DNA sequencing. SUBJECTS: 124 unrelated extremely obese women (body mass index, (BMI) 40 kg=m2) and 85 average weight controls (BMI < 27 kg=m2). MEASUREMENTS: Radiation hybrid (RH) mapping was performed to localize the MC3R gene. 50 and 30 ¯anking sequences of MC3R gene were cloned. PCR-SSCP and DNA sequencing were used to detect mutations in the MC3R gene coding region and ¯anking sequences. RESULTS: RH mapping localized the MC3R gene to 20q13, between markers D20S100 and D20S149. 1083 bp 50 and 653 bp 30 ¯anking region of the MC3R gene were cloned. A missense mutation ( 241, codon 81 ATT=GTT, Ile ! Val) was found in the MC3R coding region. Four more variants were detected in the 50 ¯anking sequence: ÿ201(C ! G), ÿ239 (A ! G), ÿ762(A ! T) and ÿ769(T ! C). Compared with controls, no signi®cant allele frequency differences were found. Racial differences were found for the 241, ÿ201, ÿ239 and ÿ762 polymorphisms. CONCLUSIONS: Several sequence variants were found in the MC3R gene coding region and in 50 ¯anking sequences. However, none of the variants were associated with obesity phenotypes. The linkage of extreme human obesity on 20q13 is likely caused by genes other than MC3R. International Journal of Obesity (2000) 24, 206±210 Keywords: melanocortin 3 receptor; ¯anking region; sequence polymorphism; obesity
Introduction Obesity is a very common disorder, and the genetic pathogenesis is complex, with more than one hundred candidate genes or regions implicated.1 Recently, there has been some convergence of positive linkage results for human chromosome 20q13 and the homologous region for mouse chromosome 2.2 ± 6 The human melanocortin 3 receptor (MC3R) is a single exon gene encoding a protein of 360 amino acids (1083 nucleotides, Genbank L06155) that has been mapped to this region (20q13.2-q13.3) using ¯uorescence in situ hybridization (FISH).7 MC3R belongs to a family of G-protein coupled melanocortin receptors, ®ve of which have been cloned.8 ± 13 MC1R is expressed in skin (melanocytes) and pituitary, MC2R in adrenal cortex, and MC5R in several different organs including endocrine glands and adipose tissue. The MC3R is mainly expressed in brain and placenta, and MC4R is only expressed in *Correspondence: Dr R Arlen Price, Department of Psychiatry, Center for Neurobiology and Behavior, University of Pennsylvania, One Clinical Research Building, 135b, 415 Curie Blvd, Philadelphia, PA 19104, USA. Email:
[email protected] Received 25 May 1999; revised 16 August 1999; accepted 17 August 1999 # Current address: Department of Psychiatry, Eulji Medical School, Daejon, Korea
brain.14 The agouti signaling protein (ASIP) mutant and the wild type agouti related protein (AGRP), when ectopically expressed, produce obesity through their competitive antagonism of MC4R, blocking the anorectic action of a-melanocyte stimulating hormone (a-MSH).15 Several MC4R frameshift mutations have been found in human obesity.16,17 Recently, Yang et al 18 showed that ASIP is also an antagonist of MC3R. Based on its similar structure, MC3R may have a similar role normally, and thus mutations in the gene may play a role in the pathogenesis of obesity. Based on location as well as biology, MC3R is a plausible candidate gene for obesity. In order to test for a possible role for the MC3R in human obesity, we constructed a radiation hybrid (RH) map of the region, cloned approximately 1.3 Kb of 50 and 0.7 Kb of 30 ¯anking region and screened MC3R variants in extremely obese women and average weight female controls by polymerase chain reaction-single strand conformation polymorphism (PCR-SSCP) and DNA sequencing.
Materials and methods Subjects
One-hundred and twenty-four unrelated extremely obese women (BMI 40 kg=m2) were selected from
MC3R polymorphisms W-D Li et al
obesity families, including those used for a genome scan.4 Most of these women had at least one ®rst degree relative with a BMI 30 kg=m2.19 In addition, DNA from 85 average-weight (BMI < 27) control subjects was examined. This control sample was selected without regard to family history of obesity. Both obese and control groups contained Caucasians as well as African-Americans (Table 1). All subjects gave informed consent and the protocol was approved by the Committee on Studies Involving Human Beings at the University of Pennsylvania. Radiation hybrid mapping of the MC3R gene
The MC3R gene was mapped using the G3 radiation hybrid (RH) panel from the Stanford Human Genome mapping Center. RH DNAs were typed in duplicate by PCR (0.2 mM primers, 200 mM dNTPs, and 1 U Taq DNA polymerase, 95 C 30 s, 60 C 45 s and 72 C extension 30 s for 32 cycles) using primers spanning bases 23±278 of protein coding sequence: MC3Rf, 50 TGGAGGGAGATTTTGTCTTTCCTG; MC3Rr, 50 TTTTCCAGCAGACTGACGATGC. Cloning and sequencing of the 50 ¯anking region and 30 UTR of the MC3R gene
To design a genomic PCR assay including the entire coding region, and to survey potential regulatory elements, 50 and 30 genomic sequences ¯anking the MC3R coding region were obtained. Primers from the 50 coding end of MC3R gene (50 to 30 : TGGAGGGAGATTTTGTCTTTCCTG; TTTTCCAGCAGACTGA CGATGC) were used to screen pooled DNAs of a human BAC clone library (Research Genetics, Inc. CITB Human BAC Library). Two BACs were identi®ed in 25 ml PCR reactions using 20 mM primers and 0.03 U AmpliTaq (Perkin-Elmer). Cycling conditions were: denature at 95 C for 3 min, 32 cycles of 94 C for 30 s, 60 C for 30 s, 72 C for 30 s, followed by 72 C extension for 5 min. 50 and 30 ¯anking sequences were identi®ed by direct sequencing on maxiprepped (QIAGEN, Valencia, CA) BAC DNA using primers derived from the coding region sequence. Flanking sequences were then used to design primers for amplifying the complete MC3R coding region. DNA extraction
DNA was extracted from blood samples using a high salt method.20 20 ng DNA was used for PCR ampli®cation.
207
PCR-SSCP
PCR primers were designed according to the MC3R gene coding sequences10 as well as 50 and 30 ¯anking sequences. The primers were end-labeled using T4 polynucleotide kinase (New England Biolabs, Beverley, MA) and 33P-ATP (Amersham±Pharmacia, Arlynton Heights, IL). Ten ml PCR reactions (20 pmol primers, 1.5 mM MgCl2, 0.625 U Taq DNA polymerase, Sigma) were performed using the following conditions: denature at 95 C 5 min, followed by 94 C 1 min, 55 C 2 min, 72 C 2 min for 2 cycles, then 94 C 30 s, 55 C 45 s, 72 C 30 s for 34 cycles, and 72 C extension 7 min. After denaturing for 5 min, 3 ml PCR products were loaded onto a 0.5X MDE gel (FMC, Rockland, ME) and subject to 2 W electrophoresis at room temperature for 14 h. Selected samples with shifted bands were re-ampli®ed in 100 ml volume, PCR products were puri®ed using QIAquick PCR Puri®cation Kit (QIAGEN, Valencia, CA) and sequenced with an ABI 377 automatic DNA sequencer (by the Genetics Core Facility at the University of Pennsylvania).
Results Subjects
As shown in Table 1, the mean BMI of extremely obese women was over 50 kg=m2, and for averageweight women, less than 23 kg=m2. Within the extremely obese group, 100 were Caucasians and 24 were African-Americans. In the control (average weight) group, 58 were Caucasians and 27 were AfricanAmericans. Sequences of the 50 ¯anking region and 30 UTR of the MC3R gene
We cloned and sequenced 1083 bp 50 and 653 bp 30 ¯anking sequences of the MC3R gene. The sequence is presented in Figure 1. Radiation hybrid (RH) mapping
RH mapping data was submitted for analysis to the Stanford Human Genome Center (SHGC) RH server using the URL: http://wwwshgc.stanford.edu/ RH=rhserverformnew.html. RH mapping con®rmed the localization of MC3R to distal chromosome 20q,
Table 1 Description of sample characteristics in extremely obese (BMI 40 kg=m2) and average weight women (BMI < 27 kg=m2)
Age (y) Body mass Index (kg=m2)
Caucasian extremely obese n 100
Control n 58
African American extremely obese n 24
Control n 27
39 6 (23±53) 50.7 9.1 (40.1±97.1)
23 14 (19±67) 21.8 2.2 (17.2±26.1)
36 9 (18±54) 48.1 9.2 (40.1±78.4)
29.7 10 (19±63) 22.5 3 (17.6±26.9)
Values for age and BMI are given as mean s.d. (range). International Journal of Obesity
MC3R polymorphisms W-D Li et al
208
Table 2 Frequencies of MC3R sequence variants in extremely obese (BMI 40 kg=m2) and average-weight Caucasian women (BMI < 27 kg=m2)
Variants# 241(A=G) 7769(T=C) 7762(A=T) 7201(C=G) 7239(A=G)
Obese
Control
(n 100)( 7
(n 58) 7=77
11 14 9 16 0
89 84 89 83 99
7=1* 7 10 13 1
Chi-squared P-value
50 51 48 44 57
0.27 0.15 2.2 1.05 1.71
0.604 0.696 0.138 0.306 0.191
*denotes cases with heterozygous mutations=cases with homozygous mutations. # location of variants (normal published sequence by Gantz et al=sequence variants). 7: sequence varaint; : normal published sequence (Gantz et al, 1993, has been labeled in Figure 1 in bold).
Table 3 Frequencies for MC3R gene variants for extremely obese (BMI 40 kg=m2) and control African-American women (BMI < 27 kg=m2) Obese Variants# 241(A=G)# ÿ769(T=C) ÿ762(A=T)# ÿ201(C=G)# ÿ239(A=G) Figure 1 Genomic DNA sequence of human MC3R gene and ¯anking regions, including 71336 to 71 50 ¯anking sequence, 11083 coding region, and 653 bp 30 of the TAG stop codon. Upper case letters denote coding sequences and lower case letters denote ¯anking sequence. Nucleotides in bold denote start=stop codons (ATG=TAG) and sequences variants: 81 codon (nucleotide 241) ATT=GTT Ile=Val, 7769 (T=C), 7762 (A=T), 7239 (A=G) and 7201 (C=G).
between markers D20S100 and D20S149, with closest linkage to marker D20S775 (6 cR_10,000; LOD 16). This localization places MC3R close to one of three peak linkages in 20q13, near the marker D20S149.
Mutation in MC3R coding region
A common missense mutation (nucleotide 241, codon 81, ATT ! GTT, Ile ! Val) was found in the MC3R coding region. We found heterozygous and homozygous variants in both extremely obese and control women. (Table 2 and Table 3) 0
Variability in 5 ¯anking region
Four more sequence variants were detected in 50 ¯anking sequences: 7201(C ! G), 7239(A ! G), 7762(A ! T) and 7769(T ! C). No sequence variants were found in the 30 region. The frequencies of these polymorphisms are shown in Table 2. No signi®cant allele frequency differences (P < 0.05) were International Journal of Obesity
Control
Chi-squared P-value
(n 24) (n 27) 7=77 7=77 12/2.* 2. 14. 0. 2.
10 22 10 24 22
10=4* 5 11 2 2
13 24 18 27 27
0.21 0.89 2.15 1.69 0.04
0.646 0.345 0.142 0.194 0.845
#
denotes racial differences between Caucasians and AfricanAmericans (P < 0.05). *denotes cases with heterozygous mutations=cases with homozygous mutations. ÿ: sequence variant; : normal published sequence (Gantz et al, 1993).
found between extremely obese women and those of average weight. The 7762 polymorphism was the only one with a marginal difference between cases and controls, however, it did not reach signi®cance. Racial differences
Four variants (241, 7201, 7239 and 7762) showed differences in allele frequencies between races. (Tables 3 and 4) 241, 7239 and 7762 polymorphisms occurred more frequently in AfricanAmericans, and the 7201 polymorphism was more common in Caucasians. Table 4 Racial differences of sequence variants on human MC3R gene in obese women (BMI 40 kg=m2) Variants 241(A=G)# 7769(T=C) 7762(A=T)# 7239(A=G)# 7201(C=G)* #
Caucasians African-Americans (n 100) (n 24) Chi-square 11 14 9 0 16
14 2 14 2 0
26.72 0.59 30.19 8.32 4.42
occurred more frequently in African Americans. *occurred more frequently in Caucasians.
P-value > 0.0001 0.4407 > 0.0001 0.0039 0.0355
MC3R polymorphisms W-D Li et al
Discussion Linkage studies have identi®ed a region containing one or more obesity susceptibility genes on human chromosome 20q12 to 20q13.3. Lembertas et al 2 and Mehrabian et al 3 have localized a quantitative trait locus (QTL) for obesity on mouse chromosome 2 (Mob5), and sibling pair analyses of Quebec families found a suggestive linkage in the interval between the ADA and MC3R genes in the homologous region of human chromosome 20q.2 A genome scan and subsequent saturation mapping by our group4 found linkage to several markers on 20q13. There are at least two other mouse obesity models which map to this same region.5,6 Our RH mapping indicated the MC3R gene is located between D20S100 and D20S149, with the closest marker being D20S775. This localization places the gene close to one of our peak linkages in 20q. Several mutations have been found by others in melanocortin receptors. Two nucleotide substitutions in MC1R can cause melanoma due to UV-induced skin damage.21,22 Several other missense and a nonsense mutations in MC2R yielded a glucocorticoid de®ciency phenotype.23 ± 27 Recently, two groups reported that frameshift mutations of the MC4R gene were associated with dominantly inherited human obesity.16,17 Another linkage study found moderate associations between MC4R and MC5R and obesityrelated phenotypes in Quebec families.28 In the MC3R coding region, we found one missense mutation (nucleotide 241, codon 81, ATT ! GTT), causing an Isoleucine ! Valine amino acid change. The MC3R gene has 7 transmembrane regions, and codon 81 is located in the ®rst transmembrane helix. Isoleucine and Valine are both neutral and hydrophobic and have side chains differing by only one methyl group. It appears unlikely that this polymorphism would have functional consequences. Moreover, there were no frequency differences between cases and controls. Unlike MC4R, no frame-shift mutation was found in this study. Four other polymorphisms were found in the 50 ¯anking region. The region we examined was suf®ciently long (1336 bp) to include promoter sequence. We used the MatInspector program (V2.2)29 to predict potential binding sites in the 50 ¯anking sequence. The 7239 polymorphism may affect a GATA box, however, the site may be too far upstream to be a functional transcription factor binding site, and functional assays will be required to evaluate its signi®cance. The 7239 variant is also relatively rare and more samples will be required (especially homozygous subjects) to assess its possible contribution to obesity. Although our genome scan4 and RH mapping placed the MC3R gene in the critical region of chromosome 20 linkage, and several other melanocortin receptor mutations were found associated with
obesity, the variants we identi®ed in the MC3R gene are unlikely to be key factors in human body weight control in extremely obese women. We cannot rule out the possibility that some of these polymorphisms may in¯uence the normal range of adiposity. We also note that our sample size has limited power for detecting associations to quantitative trait loci, which may have relatively small effects on phenotype. We found signi®cant race differences in the frequencies of four of the ®ve polymorphisms we detected. Three of the four were more common in African-Americans, consistent with the greater variability expected in African groups [see for example, reference 30]. These differences should reinforce the importance of controlling for race and ethnicity in association studies, since population strati®cation for genotype and phenotype can produce spurious associations.31 In summary, we found one coding region mutation and four 50 ¯anking region variants in extremely obese women, as well as normal controls. It is suggested that these sequence variants may not contribute to the extreme obesity phenotype.
209
Acknowledgements
This work was supported by the National Institutes of Health (NIH; R01DK44073 and R01DK48095), and Axys Pharmaceuticals and Glaxo Wellcome (RAP, PI). Collection of mental health controls was supported by P01MH43880 (Raquel E. Gur, PI). We acknowledge the cooperation of our subjects. We thank Guang Ming Yuan, Robin L. Kilker, Nicholas Guido, Andrew C. Krakowski, Balasahib Shinde, Sami Weisberg, Christa A. Mayr, Elizabeth Joe, Kruti Quazi, Karynn Henry, Paula Kown, Timothy Gasperoni, Richard Joseph, Duc Bach and sequencing technical assistance from Axys. Drs. Joseph H. Lee and Danielle R. Reed commented on earlier drafts. References
1 Perusse L, Chagnon Y, Weisnagell J, Bouchard C. The human obesity gene map: the 1998 update. Obesity Res 1999; 7: 111 ± 129. 2 Lembertas AV, Perusse L, Chagnon YC, Fisler JS, Warden CH, Purcell-Huynh DA, Dionne FT, Gagnon J, Nadeau A, Lusis AJ, Bouchard C. Identi®cation of an obesity quantitative trait locus on mouse chromosome 2 and evidence of linkage to body fat and insulin on the human homologous region 20q. J Clin Invest 1997; 100: 1240 ± 1247. 3 Mehrabian, M, Wen, P-Z, Fisler, J, Davis, RC, Lusis, AJ. Genetic loci controlling body fat, lipoprotein metabolism, and insulin levels in a multifactorial mouse model. J Clin Invest 1998; 101: 2485 ± 2496. 4 Lee JH, Reed DR, Li W-D, Xu W, Joo E-J, Kilker RL, Nathakumar E, North M, Sakul H, Bell C, Price RA. Genone scan for human obesity and linkage to markers in 20q13 Am J Hum Genet 1999; 64: 196 ± 209. 5 Pomp D. Genetic dissection of obesity in polygonic animal models. Behav Genet 1997; 27: 285 ± 306. 6 Taylor BA, Phillips SJ. Obesity QTLs on mouse chromosomes 2 and 17. Genomics 1997; 43: 249 ± 257. International Journal of Obesity
MC3R polymorphisms W-D Li et al
210
7 Gantz I, Tashiro T, Barcroft C, Konda Y, Shimoto Y, Miwa H, Glover T, Munzert G, Yamada T. Localization of the genes encoding the melancortin-2 (adrenocorticotropic hormone) and melanocortin-3 receptors to chromosomes 18p11.2 and 20q13.2-q13.3 by ¯uorescence in situ hybridization. Genomics 1993; 18: 166 ± 167. 8 Gantz I, Shimoto Y, Konda Y, Miwa H, Dickinson CJ, Yamada T. Molecular cloning, expression, and characterization of a ®fth melanocortin receptor. Biochem Biophys Res Commu 1994; 200: 1214 ± 1220. 9 Mountjoy KG, Robbins LS, Mortrud MT, Cone RD. The cloning of a family of genes that encode the melanocortin receptors. Science 1992; 257: 1248 ± 1251. 10 Gantz I, Konda Y, Tashiro T, Shimoto Y, Miwa H, Munzert G, Watson SJ, DelValle J, Yamada T. Molecular cloning of a novel melanocortin receptor. J Biol Chem 1993; 268: 8246 ± 8250. 11 Gantz I, Miwa H, Konda Y, Shimoto Y, Tashiro T, Watson SJ, DelValle J, Yamada T. Molecular cloning, expression, and gene localization of a fourth melanocortin receptor. J Biol Chem 1993; 268: 15174 ± 15179. 12 Chhajlani V, Muceniece R, Wikberg JE. Molecular cloning of a novel human melanocortin receptor [published erratum appears in Biochem Biophys Res Comm 1996 Jan 17; 218: 638]. Biochem Biophys Res Commun 1993; 195: 866 ± 873. 13 Chowdhary BP, Gustavsson I, Wikberg JE, Chhajlani V. Localization of the human melanocortin-5 receptor gene (MC5R) to chromosome band 18p11.2 by ¯uorescence in situ hybridization. Cytogenet Cell Genet 1995; 68: 79 ± 81. 14 Chajlani V. Distribution of cDNA for melanocortin receptor subtypes in human tissues. Biochem Mol Biol Int 1996; 38: 73 ± 80. 15 Marsh DJ, Hollopeter G, Huszar D, Laufer R, Yagaloff KA, Fisher SL, Burn P, Palmiter RD. Response of melanocortin-4 receptor-de®cient mice to anorectic and orexigenic peptides. Nature Genet 1999; 21: 119 ± 122. 16 Yeo GSH, Farroqi IS, Aminian S, Halsall DJ, Stanhope RG, O'Rahilly S. A frameshift mutation in MC4R associated with dominantly inherited human obesity. Nature Genet 1998; 20: 111 ± 112. 17 Vaisse C, Clement K, Guy-Grand B, Froguel P. A frameshift mutation in human MC4R is associated with a dominant form of obesity. Nature Genet 1998; 20: 113 ± 114. 18 Yang YK, Tompson DA, Dickinson CJ, Wilken J, Barsh GS, Kent SB, Gantz I. Characterization of Agouti-related protein binding to melanocortin receptors. Mol Endocrinol 1999; 13: 148 ± 155. 19 Price RA, Reed DR, Lee JH. Obesity related phenotypes in families selected for extreme obesity and leanness. Int J Obes Relat Metab Disord 1998; 22: 406 ± 413.
International Journal of Obesity
20 Lahiri D, Nurnberger J. A rapid non-enzymatic method for the precipitation of HMW DNA from blood for RFLP studies. Nucleic Acids Res 1991; 19: 5444. 21 Valverde P, Healy E, Sikkink S, Haldane F, Thody AJ, Carothers A, Jackson IJ, Rees JL. The Asp84Glu variant of the melanocortin 1 receptor (MC1R) is associated with melanoma. Hum Mol Genet 1996; 5: 1663 ± 1666. 22 Frandberg PA, Doufexis M, Kapas S, Chhajllani V. Human pigmentation phenotype: a point mutation generates nonfunctional MSH receptor. Biochem Biophys Res Comm 1998; 245: 490 ± 492. 23 Tsigos C, Arai K, Hung W, Chrousos GP. Hereditary isolated glucocorticoid de®ciency is associated with abnormalities of the adrenocorticotropin receptor gene. J Clin Invest 1993; 92: 2458 ± 2461. 24 Tsigos C, Arai K, Latronico AC, DiGeorge AM, Rapaport R, Chrousos GP. A novel mutation of the adrenocorticotropin receptor (ACTH-R) gene in a family with the syndrome of isolated glucocorticoid de®ciency, but no ACTH-R abnormalities in two families with the triple A syndrome. J Clin Endocrinol Metab 1995; 80: 2186 ± 2189. 25 Clark AJ, McLoughlin L, Grossman, A. Familial glucocorticoid de®ciency associated with point mutation in the adrenocorticotropin receptor. Lancet 1993; 341: 461 ± 462. 26 Weber A, Toppari J, Harvey RD, Klann RC, Shaw NJ, Ricker AT, Nanto-Salonen K, Bevan JS, Clark AJ. Adrenocorticotropin receptor gene mutations in familial glucocorticoid de®ciency: relationships with clinical features in four families. J Clin Endocrinol Metab 1995; 80: 65 ± 71. 27 Naville D, Barjhoux L, Jaillard C, Faury D, Despert F, Esteva, Durand P, Saez JM, Begeot M. Demonstration by transfection studies that mutations in the adrenocorticotropin receptor gene are one cause of the hereditary syndrome of glucocorticoid de®ciency. J Clin Endocrinol Metab 1996; 81: 1442 ± 1448. 28 Chagnon YC, Chen WJ, Perusse L, Chagnon M, Nadeau A, Wilkison WO, Bouchard C. Linkage and association studies between the melanocortin receptors 4 and 5 genes and obesityrelated phenotypes in the Quebec Family Study. Mol Med 1997; 3: 663 ± 673. 29 Quandt K, Frech K, Karas H, Wingender E, Werner T. MatInd and MatInspector ± New fast and versatile tools for detection of consensus matches in nucleotide sequence data. Nucleic Acids Res 1995; 23: 4878 ± 4884. 30 Ayala FJ. The myth of eve: molecular biology and human origina. Science 1995; 270: 1930 ± 1936. 31 Ewens W, Speilman RS. The transmission=disequilibrium test: history, subdivision, and admixture. Am J Hum Genet 1995; 57: 455 ± 464.
