HMGA2 Is Confirmed To Be Associated with Human Adult Height

Authors

  • Tie-Lin Yang,

    1. Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, P. R. China
    2. School of Medicine, University of Missouri – Kansas City, Kansas City, MO 64108, USA
    Search for more papers by this author
  • Yan Guo,

    1. Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, P. R. China
    2. School of Medicine, University of Missouri – Kansas City, Kansas City, MO 64108, USA
    Search for more papers by this author
  • Li-Shu Zhang,

    1. School of Medicine, University of Missouri – Kansas City, Kansas City, MO 64108, USA
    Search for more papers by this author
  • Qing Tian,

    1. School of Medicine, University of Missouri – Kansas City, Kansas City, MO 64108, USA
    Search for more papers by this author
  • Han Yan,

    1. Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, P. R. China
    Search for more papers by this author
  • Yan-Fang Guo,

    1. Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, P. R. China
    Search for more papers by this author
  • Hong-Wen Deng

    Corresponding author
    1. Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, P. R. China
    2. School of Medicine, University of Missouri – Kansas City, Kansas City, MO 64108, USA
    3. Center of System Biomedical Sciences, Shanghai University of Science and Technology, Shanghai 200093, P. R. China
    Search for more papers by this author

Corresponding author: Hong-Wen Deng, Ph.D., Key Laboratory of Biomedical Information Engineering of Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, P. R. China. Tel: 816-235-5354; Fax: 816-235-6517; E-mail: dengh@umkc.edu

Summary

Recent genome-wide association studies have identified a novel polymorphism, rs1042725, in the HMGA2 gene for human adult height, a highly heritable complex trait. Replications in independent populations are needed to evaluate a positive finding and determine its generality. Thus, we performed a replication study to examine the associations between polymorphisms in HMGA2 and adult height in two US Caucasian populations (an unrelated sample of 998 subjects and a family-based sample of 8385 subjects) and a Chinese population (1638 unrelated Han subjects). We confirmed the association between rs1042725 in HMGA2 and adult height both in the unrelated and family-based Caucasian populations (overall P= 4.25 × 10−9). Another two SNPs (rs7968902 and rs7968682), which were in high linkage disequilibrium with rs1042725, also achieved the significance level in both Caucasian populations (overall P= 6.34 × 10−7, and 2.72 × 10−9, respectively). Our results provide strong support to the initial finding. Moreover, SNP rs1042725 was firstly found to be associated with adult height (P= 0.008) in the Chinese population, and the effect is in the same direction as in the Caucasian populations, suggesting that it is a common variant across different populations. Our study further highlights the importance of the HMGA2 gene's involvement in normal growth.

Introduction

Human adult height is an important physical index to reflect the processes of growth and development, and is associated with several common complex diseases, including cancers (Gunnell et al., 2001), cardiovascular disease (Davey Smith et al., 2000), and osteoporosis (Hemenway et al., 1995). As a classic, polygenic quantitative trait, adult height is under strong genetic influence with heritability estimated up to 90% (Silventoinen et al., 2000, 2003; Macgregor et al., 2006; Perola et al., 2007). Identifying the genetic determination for adult height will provide insights into the mechanism of growth and development, and the genetic architecture of other human disorders in general.

Recent advances in SNP genotyping technologies and analytic methods have provided new opportunities for genome-wide association studies (GWAS) to explore common variants for adult height. The first GWAS of height from 4921 European individuals identified a novel variant rs1042725 in the high mobility group-A2 (HMGA2) oncogene that is associated with variation in height (P= 4 × 10−8) (Weedon et al., 2007). Subsequently, another two GWAS provided confirmatory evidence for the association of this variant with height (Lettre et al., 2008; Weedon et al., 2008). Replication of genetic associations in independent populations is essential to reduce false-positive results and to further explore the role of these variants in the complex traits. However, these studies were performed by the same groups, and the samples were mainly from northern Europe. Replications in multiple additional populations by independent groups are needed to evaluate the credibility and generality of this finding.

Therefore, the aim of this study was to attempt to replicate the associations between polymorphisms in the HMGA2 gene and adult height both in unrelated and family-based US Caucasian populations. Since genetic and environmental backgrounds vary between different ethnic groups, we further examined the associations across ethnicities in a Chinese population to see whether the variants identified are common or ethnicity-specific.

Materials and Methods

Subjects

The study was approved by the required Institutional Review Board or Research Administration of the involved institutions. Signed informed-consent documents were obtained from all study participants before entering the study.

Unrelated Caucasian Sample

A random sample of 998 unrelated healthy subjects (500 women and 498 men) was identified from our established and expanding database containing more than 10,000 subjects. All of the identified subjects were US Caucasians of Northern European origin, living in Omaha, Nebraska, and its surrounding regions in Midwestern USA. The height for each subject was measured without shoes using a standard wall mounted stadiometer in the clinic by nurses.

Family-based Caucasian Sample

The family-based sample came from the Framingham Heart Study (FHS), which is a longitudinal study of 14,277 phenotyped subjects to identify the risk factors for cardiovascular disease. Details and descriptions about the FHS have been reported before (Dawber et al., 1951, 1963). Subjects eligible for this investigation are drawn from the FHS SNP Health Association Resource (SHARe) project (Cupples et al., 2007), for which genotyping is conducted in over 9300 subjects from three generations of subjects (including over 900 families). We have the adult height measurements on 8385 phenotyped Caucasian subjects, 1307 from the Original cohort (521 men and 786 women), 3189 from the second generation cohort (1491 men and 1698 women), and 3889 from the third generation cohort (1821 men and 2068 women). The original cohort participants had height measured at exam 1, the second generation cohort participants were measured at exam 5/6, and the third generation cohort participants were measured at exam 1.

Chinese Sample

The Chinese sample consisted of 1638 unrelated subjects including 810 males and 828 females. The subjects were recruited from northern Chinese Han adults living in Xi’an City, Shanxi province. Height measurements were made as for the US Mid-West Caucasian sample.

Genotyping

Genomic DNA was extracted from peripheral blood leukocytes using standard protocols. For the unrelated Caucasian and Chinese samples, SNP genotyping was performed using the Affymetrix Human Mapping 500K array set (Affymetrix, Santa Clara, CA, USA), which has been described in our previous experiments (Liu et al., 2008; Yang et al., 2008). The experiment procedure strictly followed the Affymetrix protocol. For the family-based FHS sample, genotyping was performed using approximately 550,000 SNPs (Affymetrix 500K mapping array plus Affymetrix 50K supplemental array). Nineteen SNPs including rs1042725 within the HMGA2 gene were successfully genotyped in all samples. SNPs that deviated from Hardy-Weinberg equilibrium (HWE, P < 0.01) and had a minor allele frequency (MAF) < 0.01 were discarded in each sample set. Thus, 16 SNPs for each of the sample sets were included separately for subsequent association analyses.

Statistical Analyses

For the unrelated samples, a linear regression implemented in PLINK (Purcell et al., 2007) was fitted to test for association assuming an additive inheritance model. We used the genotype as an additive covariate and height as a response, simultaneously including age and gender as covariates in the regression model. For the family-based sample, significant covariates like age and gender were used to adjust for the raw height data. Height residuals were normally distributed and were used to perform the association tests using the QFAM method implemented in PLINK. We performed 10,000,000 permutation procedures to generate the empirical P values. Multiple testing was adjusted by adopting the conservative Bonferroni correction. The significance threshold was set at a P value of less than 3.13 × 10−3 (0.05/16 as 16 SNPs were used in association analyses).

Meta-analysis statistics were generated using the weighted Z-scores (a standard normal deviate, the statistic associated with a P value) to quantify the overall evidence for association with adult height variation. The individual Z-score was weighted by the square root of the sample size of each study. We added together the individual weighted Z-scores derived from each sample and divided by the square root of the sum of the sample sizes to obtain an overall Z-score and an associated combined P value (Rosenthal, 1991).

Haploview v4.0 (Barrett et al., 2005) was utilised to characterise the linkage disequilibrium (LD, r2) pattern and to plot the haplotype block map.

Results

The basic characteristics of the study subjects are presented in Table 1. We summarised the association results of HMGA2 with adult height variation in Table 2. Significant association of rs1042725 with height was successfully replicated both in the unrelated (P= 1.58 × 10−3) and family-based Caucasian samples (P= 3.0 × 10−7), even after adjusting for multiple testing (significance threshold: P < 2.94 × 10−3). The allele effect was in the same direction as in the previous studies (Weedon et al., 2007, 2008; Lettre et al., 2008). Each copy of the C allele at rs1042725 was associated with an increase in height of ∼0.9 and 0.7 cm, respectively in the unrelated and family-based Caucasian samples. Besides rs1042725, another two common variants, rs7968902 and rs7968682, achieved the significance level both in the unrelated and family-based Caucasian samples (rs7968902: P= 1.13 × 10−3 and 3.4 × 10−5, respectively; rs7968682: P= 1.55 × 10−3 and 2.0 × 10−7, respectively). These two variants were in strong LD with rs1042725 (r2 of 0.71 and 0.89 estimated in the US Mid-West Caucasian sample, Fig. 1). The association of rs7968682 with height was also found in the original report (Weedon et al., 2007). The phenotype differences among subjects with different genotypes for these three SNPs are depicted in Figure 2. When we combined all Caucasian subjects together, the statistical evidence in favour of association greatly increased for these three SNPs (rs1042725: P= 4.25 × 10−9; rs7968902: P= 6.34 × 10−7; and rs7968682: P= 2.72 × 10−9). The proportion of variance in height explained by these three SNPs was about 1%, 1.07%, and 1.06%, respectively, estimated in the unrelated Caucasians after adjustment for age and gender.

Table 1.  Characteristics of the study subjects.
 Unrelated CaucasiansFamily-based CaucasiansChinese
  1. Note: Data are shown as mean (standard deviation, SD).

Number99883851638
Gender (M/F)498/5003833/4552810/828
Age (years)50.3 (18.3)45.5 (11.4)34.5 (13.2)
Weight (kg)80.2 (17.8)76.5 (17.5)60.2 (10.5)
Height (cm)170.8 (9.8)168.7 (9.5)164.3 (8.2)
Table 2.  Summary association results for 19 SNPs in HMGA2.
SNPPhysical positionGenic positionAllelesUnrelated CaucasiansFamily-based CaucasiansChinese
MAFP valueEffect size (cm)MAFP valueEffect size (cm)MAFP valueEffect size (cm)
rs7977687645023405’UTRG/A0.0410.1710.9750.0450.1330.4850.1960.1200.647
rs227204664510728intron2G/T0.0280.4250.6990.0250.0410.8400.0880.5710.192
rs1231031264521025intron3A/G0.0450.0841.1720.0620.0300.6190.2120.0290.524
rs258393764521731intron3C/T0.0690.9310.0510.0970.8040.0560.1140.6180.154
rs227204764523002intron3C/T0.0490.0841.1470.0580.0270.6410.2110.0300.241
rs261206064529769intron3G/A0.0680.8230.1320.0990.9230.0220.1150.7880.082
rs1710183964531835intron3T/C0.0240.1151.5000.0490.0230.754---
rs1117594464542662intron3C/G0.0640.0451.1490.0893.65 × 10−30.6930.2110.0180.566
rs795939664543214intron3C/A0.0400.2450.8380.0440.0570.6270.2110.0180.569
rs797357464558999intron3G/A0.0280.5200.5550.0462.49 × 10−31.1410.0360.4120.434
rs1087834664607140intron3A/G0.2260.2230.4200.2461.0 × 10−60.9740.1480.2670.308
rs115609564613521intron3G/A0.1670.6380.185---0.0320.2320.661
rs1117597364620042intron3G/T--- -----
rs148046864628384intron3A/G--- -----
rs1117597864629135intron3T/C---0.0525.0 × 10−61.9930.0920.0250.763
rs1042725646446143’ UTRT/C0.4991.58 × 10−30.9060.4863.0 × 10−70.6830.8030.0080.614
rs7968902646493373’ UTRT/G0.4291.13 × 10−30.9380.4143.4 × 10−50.5330.1530.0330.586
rs1480464646568383’downstreamT/C0.1960.1530.5180.1900.3790.1420.2520.0640.418
rs7968682646581473’downstreamG/T0.4991.55 × 10−30.9390.4682.0 × 10−70.7610.1070.1670.438
Figure 1.

Linkage disequilibrium diagram of the HMGA2 gene.
Pairwise LD, measured as r2, was calculated from unrelated Caucasians and Chinese data using Haploview program.

Figure 2.

Comparison of height values among subjects with different genotypes for the three significant SNPs in Caucasians and Chinese.
For the family-based Caucasian sample, we selected 1,518 unrelated subjects including the parents from each family to plot this figure. Error bars denote standard error. P values were calculated by the linear model.

For the Chinese sample, SNP rs1042725 in HMGA2 was detected as nominally significant to adult height (P= 0.008). Although the allelic frequency (allele C: 0.197) was quite different from the frequency in the Caucasian samples (allele C: 0.501), the effect was in the same direction as in the Caucasian samples. Each copy of the C allele was estimated to be associated with an increase in height of ∼0.6 cm. SNP rs7968902 showed a marginally significant association (P= 0.033). However, no significant association was found for rs7968682. As shown in Figure 1, rs1042725 and rs7968902 were in strong LD with each other (r2 of 0.67), whereas the LD was relatively weak between rs1042725 and rs7968682 (r2 of 0.43).

Discussion

Recent GWAS reports have revealed a novel variant, rs1042725, in the HMGA2 gene for human adult height in European populations (Lettre et al., 2008; Weedon et al., 2008, 2007). Replication of GWAS findings can be quite difficult, especially for a quantitative trait. In this study, we successfully replicated this finding both in the unrelated and family-based US Caucasian populations, which demonstrated the validity of the initial finding. We also identified another two SNPs (rs7968902 and rs7968682) in HMGA2, which were in high LD with rs1042725, significantly associated with adult height. HMGA2 was considered as a strong biologic candidate for height as a rare severe mutation in this gene alters body size in mice (Zhou et al., 1995) and in humans (Ligon et al., 2005). All of these three SNPs are located within the 3′-UTR region of HMGA2, which may directly or indirectly influence the mRNA stability of HMGA2.

It would be worth evaluating the association in populations of different ancestry from that of the initial report, since genomic variation is greater when compared across populations, and this may increase the credibility of the findings. In this study, we successfully replicated the association between rs1042725 in HMGA2 and adult height in a Chinese population and the effect direction was consistent despite the different allele frequencies. Our findings suggest that this SNP might be a common variant for adult height across different populations.

The statistical power of our study is estimated by using the program Genetic Power Calculator (http://pngu.mgh.harvard.edu/~purcell/gpc/cc2.html). Assuming that a marker has a minor allele frequency of 0.25 and is in strong LD (D′= 0.9) with a functional mutation that accounts for ∼2% variation of height, under the conservative significance level of P= 0.0029 (Bonferroni correction threshold), our Caucasian and Chinese samples can achieve >85% statistical power, which is large enough to detect a genetic variant under the additive model.

In summary, using data from ∼11,000 subjects, we have confirmed the recently found association between HMGA2 and human adult height even across ethnic boundaries, in both Caucasian and Chinese populations, which highlights the importance of this gene's involvement in normal growth. Follow-up molecular and functional studies are warranted to elucidate its detailed roles and to identify the true functional variant.

Acknowledgements

This work was partially supported by the National Institutes of Health [R01 AR050496, R21 AG027110, R01 AG026564, P50 AR055081 and R21 AA015973]. The study also benefited from grants from the National Science Foundation of China, Huo Ying Dong Education Foundation, HuNan Province, Xi’an Jiaotong University, and the Ministry of Education of China. The work is also partially supported by Shanghai Leading Academic Discipline Project, Project Number: S30501. The Framingham Heart Study and the Framingham SHARe project are conducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with Boston University. The Framingham SHARe data used for the analyses described in this manuscript were obtained through dbGaP (phs000007.v3.p2, phs000008.v3.p2). This manuscript was not prepared in collaboration with investigators of the Framingham Heart Study and does not necessarily reflect the opinions or views of the Framingham Heart Study, Boston University, or the NHLBI.

Competing Interest Statement

The authors declare that they have no competing financial interests.

Ancillary