| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Osteoporosis Research Center (Y.-Z.L., P.X., D.-H.X., L.-J.Z., H.S., Y.-J.L., V.D., J.-R.L., H.-Y.D., H.-W.D.), Creighton University Medical Center, and Department of Biomedical Sciences, Creighton University, Omaha, Nebraska 68131; The Key Laboratory of Biomedical Information and Engineering (H.-W.D.), Ministry of Education, and Institute of Molecular Genetics, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China; Laboratory of Molecular and Statistical Genetics (Y.-F.G., H.-W.D.), College of Life Sciences, Hunan Normal University, Changsha, Hunan 410081, People’s Republic of China; Department of Biological Sciences (V.D.), Kent State University, Kent, Ohio 44242; and Seattle Biomedical Research Institute (J.-L.L.), Seattle, Washington 98109
Address all correspondence and requests for reprints to: Hong-Wen Deng, Ph.D., Osteoporosis Research Center, Creighton University Medical Center, 601 North 30th Street, Suite 6787, Omaha, Nebraska 68131. E-mail: dengh{at}umkc.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: This study aimed to search for potential genomic regions that harbor interactive genes underlying human height.
Methods: Here with a sample containing 3726 Caucasians, the largest one ever obtained from a single population of the same ethnicity among genetic linkage studies of human complex traits, we performed variance component linkage analyses of height based on a two-locus epistatic model. We examined pairwise genetic interaction among three regions, 9q22, 6p21, and 2q21, which achieved significant or suggestive linkage signals for height in our recent whole genome scan.
Results: Significant genetic interaction between 6p21 and 2q21 was detected, with 2q21 achieving a maximum LOD score of 3.21 (P = 0.0035) under the epistatic model, compared with a maximum LOD score of 1.63 under a two-locus additive model. Interestingly, 6p21 contains a cluster of candidate genes for skeletal growth, suggesting a mechanism whereby 2q21 regulates height through 6p21.
Conclusion: By providing the first evidence for genetic interaction underlying human height variation, this study further delineated the genetic architecture of human height and contributed to the genetic dissection of human complex traits in general.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In normal populations, the variation of height is characterized by high inheritance. The heritability,1 generally above 0.75 (2, 3, 4), is higher than most other human complex traits, such as body mass index and bone mineral density. Yet the genetic factors underlying normal height variation are still largely unknown. Association studies have identified quite a few candidate genes for height, such as the collagen I A1 gene (5), the GH-1 gene (6), the vitamin D receptor gene (7), the peroxisome proliferator-activated receptor-
gene (8), and the PTH/PTHrP receptor gene (9). Through a genome-wide linkage2 approach, studies have suggested many genomic regions linked to height (10, 11, 12, 13, 14, 15, 16, 17) [for a brief review of identified regions linked to height, see Willemsen et al. (14)]. Interestingly, the 7q31 region has been identified in three independent studies (11, 13, 15), making it one of the promising regions containing genes for human height. In a series of three consecutive genome-wide linkage scans (10, 12, 18), we also identified 9q22 and Xq24 as two regions showing consistent signals for linkage to height.
The likely underlying genetic model for height, as for other complex traits/diseases,3 might involve multiple genes, which work additively or interactively to contribute to the trait. So far, genetic studies of height have mainly focused on identification of genes for their independent (additive) effects on height variation. No study has been performed to find genes or loci that interact to control height. Here with a large sample containing 3726 Caucasians from 434 pedigrees, the largest one ever obtained in the field of linkage studies of human complex traits, we attempted to identify interactive genomic regions underlying human height. We investigated genetic interaction between three loci, 9q22, 6p21, and 2q21. These three loci achieved relatively strong linkage signals among all the regions on autosomes in our recent whole-genome linkage study of height (19). We tested their pairwise interaction based on a two-locus epistatic4 model. Interestingly, significant interaction between 6p21 and 2q21 was detected, making them the first set of identified regions that may contain interactive genes for human height.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The study was approved by the Creighton University Institutional Review Board. All the study subjects signed informed-consent documents before entering the project. The study subjects came from an expanding database created for ongoing studies in the Osteoporosis Research Center of Creighton University to search for genes underlying common human complex traits, including height, body mass index, bone mineral density, etc. The sampling scheme and exclusion criteria have been detailed elsewhere (10). Briefly, patients with chronic diseases and conditions that may potentially affect the development of human height as well as other studied traits were excluded from the study. These diseases/conditions included chronic disorders involving vital organs (heart, lung, liver, kidney, and brain), serious metabolic diseases (diabetes, hypo- and hyperparathyroidism, hyperthyroidism, etc.), skeletal diseases (Paget’s disease, osteogenesis imperfecta, rheumatoid arthritis, etc.), chronic use of drugs affecting bone metabolism (corticosteroid therapy, anticonvulsant drugs), and malnutrition conditions (chronic diarrhea, chronic ulcerative colitis, etc.), etc.
All the study subjects were Caucasians of European origin. The sample contains a total of 4102 phenotyped subjects from 434 pedigrees (see Table 1
for their basic characteristics), of whom 3726 subjects were genotyped. The sample was mainly made up of pedigrees of median to large size, i.e. with 10–451 genotyped subjects per family (see Table 2 for the distribution of pedigree size). Such a structure of the sample was advantageous by providing us a large number of relative pairs informative for linkage analyses, with a total of 158,456 (see Table 3 for the number of relative pairs contained in the sample).
|
|
|
DNA extraction was performed according to a protocol in our previous whole-genome studies (10, 18). The 393 autosomal microsatellite markers from the Marshfield screening set 14 were successfully genotyped by Marshfield Center for Medical Genetics, with a genotyping error rate around 5%. The markers have an average population heterozygosity of 0.75 ± 0.06 and spaced on average 8.9 cM. The detailed description, including name, chromosomal location, heterozygosity, and allele size range, of the whole panel of microsatellite markers genotyped in our study is listed in the table accessible via the link http://research.marshfieldclinic.org/genetics/sets/Set14FinalForWeb.xls. The detailed protocol for genotyping is available at http://research.marshfieldclinic.org/genetics/Lab_Methods/methods.html. A genetic database management system (GenoDB) (20) was employed to manage the phenotype and genotype data for linkage analyses. GenoDB was also used for allele binning (including setting up allele binning criteria and converting allele sizes to distinct allele numbers), data quality control, and data formatting for PedCheck (21) and linkage analyses. PedCheck was performed to ensure that the genotype data conform to a Mendelian inheritance pattern at all the marker loci.
Statistical analyses
Variance component linkage analyses (22, 23, 24) for quantitative traits were performed using sequential oligogenic linkage analysis routines (SOLAR) (24). Multipoint and two-point LOD scores were calculated for chromosomes 1–22. Tests for pairwise epistatic interactions between the three loci, 9q22, 6p21, and 2q21, were performed in a series of three models: 1) two separate single-locus models with marker effects due to each marker locus individually, 2) a two-locus additive model with marker effects for two loci simultaneously but without interaction term, and 3) a two-locus epistatic model with marker effects for both loci as well as an epistatic term. The significance associated with the increase of LOD scores under the epistatic model over that under the additive model was determined by using a 
2 test with 1 df. Twice the difference in loge likelihoods of these two models yields a test statistic that is asymptotically distributed as a 1/2:1/2 mixture of a 2 variable (1 df) and a point mass at zero (25).
In linkage analyses, age, sex, and age-sex interaction were used as covariates to adjust for raw height data.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Based on a two-locus (i.e. 6p21 and 2q21) additive model, a maximum LOD score (MLS) of 1.63 was achieved on 2p21. The LOD score increased to 3.21 under the two-locus epistatic model. According to a 2 test with 1 df, the nominal P value for the epistatic effect is 0.0035. Figure 1 shows the linkage signals on chromosome 2 under the single-locus, two-locus additive, and two-locus epistatic models. No significant epistasis was detected between 9q22 and 6p21 or between 9q22 and 2q21.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Our epistatic analyses for height were focused on three regions, 9q22, 6p21, and 2q21. They represent important regions for height in our sample because they were the only ones on autosomes that achieved significant or suggestive linkage signals in our recent whole-genome study of height (12). Limiting our study to the three regions may also help minimize the problem of multiple testing.
Interestingly, epistasis was observed between 2q21 and 6q21, as evidenced by a considerable increase of MLS on 2q21 under the epistatic model compared with under the additive model. The LOD increase, with a nominal P value of 0.0035, was significant even after Bonferroni correction (in total, three pairs of regions were tested for epistasis). According to SOLAR, the fraction of height variation due to the epistasis between the two regions is approximately 20%.
A clear advantage of our study is its large sample size. To the best of our knowledge, in the field of genetic linkage studies of complex traits, our study sample was the largest one ever obtained from a single study population of the same ethnicity. More importantly, the major structure of our sample is made up of large pedigrees (Table 2
). As a result, our sample contains a large number of relative pairs informative for linkage analyses (Table 3), which may give our study higher statistical power than most other linkage studies of human complex traits. Because a linkage study of low statistical power may suffer from a low positive predictive value, the likelihood that an observed linkage is true (31), the high statistical power of our study ensures a high likelihood that our findings for the epistatic effects between 6p21 and 2q21 are robust and reliable.
Interestingly, 6p21 contains a cluster of genes (e.g. COL11A2, CCD, RUNX2, and RXRB) that are functionally important to longitudinal growth or skeletal development. The COL11A2 (collagen, type XI, 
-2) gene (MIM 120290) is involved in human cartilage development (32). Mutations of the gene lead to chondrodysplasia or osteochondrodysplasia (33, 34), common conditions causing short stature. The CCD (cleidocranial dysplasia) gene (MIM 119600) is responsible for syndromes characterized by generalized skeletal dysplasia and growth retardation (35). The RUNX2 (runt-related transcription factor 2) gene (MIM 600211) encodes core-binding factor 1, a well-known transcription factor regulating osteoblast differentiation (36). The factor is essential to osteogenesis by controlling the integration, organization, and assembly of nucleic acids and regulatory factors for skeletal gene expression (37). The RXRB gene (MIM 180246) encodes retinoid X receptor-ß, which heterodimerizes with thyroid hormone and vitamin D receptor, two important factors for longitudinal growth, thereby increasing their DNA binding and transcriptional function (38). Because 2q21 does not contain genes known to be directly involved in skeletal growth and given the epistasis between 2q21 and 6p21 detected in our study, it may be reasonable to hypothesize that 2q21 may regulate human height mainly through its interaction with 6p21.
Noticeably, the region of 7q31 has been shown to be linked to height in three previous studies (11, 13, 15) yet did not achieve significant linkage signals in our study. Given the high statistical power of our sample (12), the negative linkage result of our study may indicate that the region’s effects on height might be very weak in our study population. Indeed, a recent linkage exclusion study by our group (39) has excluded the region for height at a relative effect size of 5% or greater (P < 0.0018), suggesting that the region may contribute very little to height variation in our Caucasian population. Genetic heterogeneity may be one of the reasons underlying the different results on the 7q31 region’s linkage to height. This is supported by the fact that the study population in our work was made up of Caucasians from the midwestern United States, whereas the samples for the studies by Hirschhorn et al. (11) and by Perola et al. (13), which detected 7q31, were mainly from a northern European population, which is well known for its different genetic makeup compared with the general population because of a long history of genetic isolation. The study by Wu et al. (15) combined several samples for reanalysis of linkage to height. Notably, the 7q31 region was not detected initially in any of the samples alone. Combining the samples together made a very large dataset containing 6752 subjects. Even with such a huge sample size and, most likely, very high statistical power, the linkage signals of the region (LOD = 2.46) did not achieve the genome-wide significance level for linkage (i.e. LOD 
3.0), suggesting that the region’s effects on height indeed might be very small in size in the general population (which is in fact consistent with what was implicated in our study).
In summary, our study provided the first evidence for significant genetic epistasis underlying human height. The results demonstrated that taking into account epistasis in linkage analyses of human complex traits may enhance linkage detection for regions that are otherwise insignificant in regular linkage analyses. With two new interactive regions for height identified, our study further delineated the genetic basis of human height.
|
Footnotes |
|---|
Disclosure statement: The authors have nothing to disclose.
First Published Online July 18, 2005
Abbreviations: MLS, Maximum LOD score; SOLAR, sequential oligogenic linkage analysis routines.
1 Phenotypic variation of a trait, such as height and body mass index, in a population results from the joint effects of both environmental and genetic factors. The proportion of phenotypic variation that is due to genetic factors is the heritability of the trait. 
2 Two genetic loci are in linkage (or linked) if they are transmitted together from parents to offspring more often than expected under independent inheritance. Loci that are physically close to each other in a chromosome are more likely to be in linkage than those that are more far apart. Linkage mapping takes advantage of this phenomenon by tracing the cosegregation of a genetic marker with a trait or a disease in human families. Through linkage mapping with multiple markers, genes underlying a trait or a disease may be localized to a relatively narrow chromosomal region for additional fine mapping and positional cloning.
3 A trait/disease is determined by actions and interactions of both genetic and environmental factors. Therefore, genetically, a complex trait/disease may not follow strict Mendelian inheritance. Typical examples include human height, body mass index, diabetes, hypertension, schizophrenia, and many types of cancers.
4 Epistasis, or gene-gene interaction, is broadly defined as the interaction between alleles of different loci in the human genome. Biologically, epistasis refers to the phenomenon where a gene’s effect is masked by or dependent on another gene. For quantitative traits, such as height, the term is defined as the deviation from additivity in the effects of alleles of different loci with respect to their contribution to a phenotype.
Received February 16, 2006.
Accepted July 10, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
1 Sp1 polymorphism, bone mass, and bone turnover in healthy French premenopausal women: the OFELY study. J Bone Miner Res 13:813–817[CrossRef][Medline]3 gene promoter affects height and lipid metabolism in a French population. Arterioscler Thromb Vasc Biol 23:289–2942(XI) collagen (COL11A2) chain. Molecular cloning of cDNA and genomic DNA reveals characteristics of a fibrillar collagen with differences in genomic organization. J Biol Chem 264:13910–13916Read all eLetters
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
