| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
From The Clinical Research Centers |
Department of Genetics (B.D.M., S.A.C., S.J.I., E.A.R., J.B., J.W.M., J.E.H.), Southwest Foundation for Biomedical Research, San Antonio, Texas 78245-0549; and Department of Medicine (R.L.B.), University of Texas Health Science Center, San Antonio, Texas 78284-7879
Address correspondence and requests for reprints to: Braxton D. Mitchell, Ph.D., Department of Genetics, Southwest Foundation for Biomedical Research, P.O. Box 760549, San Antonio, Texas 78245-0549.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The heritability of OC levels in this population was 62 ± 8%. We detected significant evidence for linkage between a quantitative trait locus influencing serum OC levels and markers on chromosome 16q, and suggestive evidence for linkage of OC levels with markers on chromosome 20q. The multipoint lod scores peaked at 3.35 on chromosome 16 and 2.78 on chromosome 20, corresponding to P values of 0.00004 and 0.00017, respectively. A potential candidate gene for bone formation in the linked region on chromosome 20 is CDMP1, which encodes cartilage-derived morphogenetic protein 1. Future studies should evaluate whether variation in CDMP1 or in other genes in the linked regions on chromosomes 16 and 20 influence the rate of bone turnover.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Serum concentrations of osteocalcin (OC) have been used as a biochemical marker of bone turnover. Serum OC, also known as bone Gla protein, is a 49-amino acid protein synthesized by osteoblasts (6). The majority of secreted OC is incorporated into bone, where it becomes incorporated into the bone matrix (7). A fraction of newly synthesized OC is also released into circulation, where it can be detected by immunoassay. Additional evidence suggests a role for OC in the recruitment and differentiation of osteoclast precursors at the bone surface (8, 9). Serum OC levels are increased in children and young adults compared with older individuals (7) and are also elevated in conditions associated with rapid bone loss (10) and excessive bone turnover (6, 7, 11, 12).
Several studies have evaluated the impact of genetic factors on serum OC levels. Kelly et al. (1) reported that variation in OC levels among female twin pairs was substantially smaller in monozygotic compared with dizygotic twins, and from these results they estimated that genetic factors accounted for 82% of the total variance in circulating OC levels. Somewhat lower estimates were obtained by Garnero et al. (13), who reported a heritability of 37% in their study of postmenopausal twin pairs. Few, if any, studies have estimated the heritability of OC levels from large family-based samples, especially ones including both men and women.
Despite the important role that genetic effects are believed to play in bone turnover, the individual genes that mediate these effects are not known. With the exception of one early study that reported an association between polymorphisms in the vitamin D receptor and serum OC levels (14), no other single gene effects on OC levels have been reported.
The aim of our study was to identify, using quantitative trait linkage analysis, chromosomal regions that might contain genes influencing variation in serum OC levels in a randomly ascertained set of families. To accomplish this aim, we conducted a genome-wide scan using markers spaced at a 10 cM average density throughout the genome.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Participants received a medical examination in our clinic in the morning following a 12-h fast. By interview, we obtained information about demographics, dietary habits (including history of smoking and alcohol consumption), current medication use, and for women, reproductive history. Women were classified as postmenopausal if they reported having had both ovaries removed or if at least 12 months had elapsed since their last menstrual period. To our knowledge, there was no history of metabolic bone disease or hyperthyroidism in these subjects since the subjects were members of "randomly" ascertained families, although our survey did not specifically inquire about these conditions. Blood samples were obtained using standard venipuncture technique for DNA processing and measurement of blood chemistries. Serum was separated from clotted blood by centrifugation and then stored at -80 C until assayed. Diabetes was defined according to the plasma glucose criteria of the World Health Organization (16).
Serum concentrations of intact OC were measured in duplicate with the NovoCalcin(r) assay (Metra Biosystems, Mountain View, CA), an enzyme immunoassay in a microtiter stripwell format using a murine monoclonal anti-OC antibody (17). In our laboratory, the interassay coefficient of variation for serum OC is 6.6%.
DNA was isolated from lymphocytes for PCR and automated genotyping. The DNA was amplified with fluorescently labeled primer pairs from MapPairs Human Screening Set Version 6 (Research Genetics, Inc., Huntsville, AL) that detect highly polymorphic microsatellite markers. PCR reactions were performed according to the manufacturer’s protocol. Aliquots of the PCR reactions were pooled into multiplexed panels for genotyping with PE Applied Biosystems (Perkin-Elmer Corp., Foster City, CA) Model 377 DNA Sequencers and Genescan and Genotyper DNA Fragment Analysis software.
A total of 376 microsatellite markers from 22 autosomes were included in the analysis. The mean (±SD) heterozygosity of these markers in this population was 0.74 ± 0.08. The distances between markers were computed from our data using CRI-MAP(18). The average spacing between markers was 10.0 cM, and the largest spacing was 27 cM (on chromosome 9).
Heritability of serum OC levels was defined in the "narrow sense," as the proportion of the total trait variance attributable to the additive effects of genes. Generally speaking, heritability is estimated from the degree of resemblance between relatives and corresponds to the observed correlation between relatives expressed as a proportion of the correlation that would be found if the trait were completely inherited (19). In pedigree structures, the covariances among different classes of relatives are generally not independent, and we, therefore, applied the method of maximum likelihood to obtain a pedigree-based estimate of heritability (20, 21). This method is more efficient than classical quantitative genetic methods that are based on pairs of relatives since all pedigree information is considered jointly. Briefly, estimates are obtained by modeling the observed covariances between two individuals within a pedigree as having an expected value given by the product of their coefficient of relationship (which is equal to two times their kinship coefficient), the heritability, and the phenotypic variance of the trait (conditional on covariate effects). Based on this simple model, the likelihood of the pedigree data is computed assuming multivariate normality as a working model. Parameter estimation is performed by finding those values of the parameters (including the heritability) that yield the maximum ln likelihood, which is summed over all pedigrees.
We tested for linkage with the genetic markers using a multipoint linkage analysis procedure based on variance component methodology (22). In this approach, the expected genetic covariances between relatives is modeled as a function of the probability that the relatives have inherited both alleles at the marker locus from a common ancestor. The total phenotypic variance is then partitioned into effects due to covariates, effects of a specific locus, and residual additive genetic effects. The variance attributable to the quantitative trait locus (QTL) is parameterized by the probability of allele sharing (i.e. the identity-by-descent matrix) at a specific locus, whereas the variance attributable to the residual heritability is parameterized as a function of the relationship among individuals (i.e. the kinship matrix). The hypothesis of linkage is evaluated by testing whether the variance attributable to the QTL is significantly greater than zero. Model parameters are obtained using maximum likelihood estimates, and the likelihood of the pedigree data is compared for nested models using the likelihood ratio test. For multipoint analysis, the probability of allele sharing was estimated at arbitrary chromosomal locations (e.g. at 1-cM intervals across the chromosome) using the multipoint approximation approach of Fulker et al. (23), generalized for use in extended families. All analyses were conducted using the SOLAR software package (22).
We derived the distribution of nominal logarithm of odds (lod)
scores under the null hypothesis of no linkage empirically by Monte
Carlo simulation. To generate this distribution, we simulated an
unlinked marker locus with five equifrequent alleles, assigned
genotypes to each founder, and then dropped genotypes down through the
pedigree based on Mendelian expectations and the founder genotypes. The
simulated unlinked marker had approximately the same information
content (i.e. heterozygosity 
80%) as the markers
used in the genome scan. We then conducted linkage analysis of serum OC
levels with the simulated unlinked marker. The unlinked marker locus
was simulated using the PAP4 software (24), and the linkage analysis on
each simulated data set was carried out using the SOLAR software
program (22). We conducted 20,000 replicates and defined the
probability of obtaining a false positive result as the proportion of
replicates for which we obtained a specified lod score or higher. These
probabilities were converted into lod scores, and it is these
simulation-derived lod scores that are presented.
We initially performed a single locus analysis, estimating the QTL effect (i.e. lod score) at each 1-cM interval along the chromosome. We then performed an oligogenic (or multilocus) analysis, including in the analysis all loci for which the initial QTL effect achieved at least marginal levels of statistical significance (e.g. a lod score of 2.0 or greater). The locus with the highest lod score was included in the model first, and then the analysis for the second locus was repeated, conditional on the first locus (and covariates) being included in the model. The conditional lod score was calculated as the difference in likelihoods between the two-loci model and the one-locus model.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
. The additive genetic heritability of
serum OC in this population was 62 ± 8%.
|
. In both men and women, the
distribution of serum OC was U-shaped with respect to age, that is OC
levels were higher in subjects younger than 31 yr and older than 60 yr
than in subjects between the ages of 30 and 60 yr. In general, women
had lower OC levels than men before age 45, at which point the
difference between the sexes disappeared.
|
Sex, sex-specific age terms (both age and age2),
and oral contraceptive use were included as covariates in the linkage
analysis. Information on oral contraceptive use was unavailable for
three women, and so the linkage analyses were, therefore, based on 429
subjects. Fig. 2 shows the maximum lod
scores observed on each chromosome. The highest lod score observed was
3.35, occurring on chromosome 16q, approximately 77 cM from pter. The
second highest lod score was 2.78, occurring on chromosome 20q,
approximately 58 cM from pter. The P values corresponding to
these lod scores were 0.00004 (at chromosome 16q) and 0.00017 (at
chromosome 20q). We detected only one other lod score as high as 1.50
elsewhere in the genome, with that occurring on chromosome 18p
(lod = 1.74), approximately 24 cM from pter.
|
and 4 show the multipoint lod scores for
serum OC on chromosomes 16 and 20, respectively. On chromosome 16, peak
evidence for linkage occurred in the interval flanked by markers
D16S753 and D16S771 in cytogenetic band 16q11.1–16q13. The one lod
unit support interval encompassed a 19-cM region extending from 70 cM
from pter to 89 cM from pter (16q11.1–16q22.1). The simulation-derived
two-point lod scores for the markers in this region were: D16S769
(lod = 0.0), D16S753 (lod = 0.94), D16S771 (lod = 0.84),
D16S3253 (lod = 0.44), D16S2624 (lod = 2.05), and D16S518
(lod = 0.13). On chromosome 20, peak evidence for linkage occurred
between markers D20S477 and D20S107 in the region of 20p11.1–20q13.12.
Fig. 4 also shows the locations of three bone formation candidate genes
on chromosome 20 (see Discussion). The one lod unit support
interval extended for 15 cM from 50–65 cM from pter
(20p11.21–20q13.12). The two-point lod scores for the markers in this
region were: D20S898 (lod = 0.32), D20S470 (lod = 1.12),
D20S477 lod = 2.30), D20S107 (lod = 0.70), D20S478 (lod
= 0.48), and D20S481 (lod = 0.02).
|
|
Finally, we conducted a second simulation study to evaluate the power to detect linkage in this set of families. For this study, we simulated a two-allele QTL that accounted for specific proportions of the total phenotypic variation in serum OC levels (ranging from 15–30%), and then dropped a tightly linked (recombination value = 0%) highly polymorphic marker down through the pedigrees. The total trait heritability was assumed to be 60%, and so the amount of the genetic variation not accounted for by the QTL was partitioned into a residual heritability. In our sample, the power to detect a QTL effect (i.e. lod score = 3.0) that accounted for 20% of the total trait variability was 33%. For a QTL accounting for 25% of phenotypic variation, the power was 59%, and the power to detect a QTL accounting for 30% of phenotypic variation was 85%.
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Our results suggest that among the genes influencing variation in serum OC levels may be two located on chromosomes 16 and 20. On chromosome 16, the peak lod score was 3.35, occurring in the interval flanked by markers D16S753 and D16S771 in a region that maps to 16q11.1–16q22.1. The peak lod score on chromosome 20 was 2.78, occurring in the interval flanked by markers D20S477 and D20S107 in a region that maps to 20p11.21–20q13.12. Our statistical approach for lod score estimation was robust not only to outliers, since we down-weighted the influence of extreme observations, but also to violation of the multivariate normality assumption, since P values and lod scores were based on an empirical null distribution generated by Monte Carlo simulation. For only one of the 20,000 unlinked markers that we simulated did we obtain evidence for linkage as high as that which we observed on chromosome 16q.
The human OC gene has been mapped to chromosome 1q21-q23 (26, 27). One recent case-control study has reported an association between allelic variation at an anonymous marker tightly linked to the OC gene and bone mineral density, although the relationship between this polymorphism and OC levels was not reported (27). In our data, there was no evidence for linkage of OC levels to any markers in this region (multipoint lod = 0.30).
There are no obvious candidate genes that might potentially influence
bone turnover in our region of linkage on chromosome 16. In contrast,
there is one potential candidate gene for bone formation whose location
coincides very closely with our region of peak linkage on chromosome
20. This gene (CDMP1) encodes cartilage-derived
morphogenetic protein 1, a protein that is closely related to bone
morphogenetic proteins 5–7 (28). Like the bone morphogenetic proteins,
CDMP1 is also a member of the transforming growth factor ß
superfamily, a large group of signaling molecules that mediate a
variety of activities, including the induction of bone and cartilage
formation (28, 29). Allelic mutations in CDMP1 that result
in reduced or absent activity of the protein are associated with
inherited syndromes of acromesomelic dysplasia, characterized by
skeletal abnormalities restricted to the limbs and limb joints (30, 31). CDMP1 falls within the interval flanked by markers
D20S477 and D20S107 on our map, making its likely position within 5 cM
of our peak signal for linkage. There are, in addition, at least
several other bone turnover candidate genes that have been mapped to
chromosome 20, including bone morphogenetic proteins 2 and 7
(BMP2 and BMP7), and collagen, type IX, 
3
(COL9A3). However, BMP2 maps to the region of
20p12, approximately 30–40 cM to the pter side of our region of peak
linkage, whereas COL9A3 maps to 20q13.3, approximately
20–30 cM to the qter side of our region of peak linkage. Although
BMP7 has been assigned cytogenetically to chromosome 20, to
our knowledge it has not yet been localized.
To our knowledge, our study represents the first genome-wide scan
undertaken to identify regions containing genes influencing variation
in serum OC levels. However, several previous studies have been carried
out to investigate whether polymorphisms in selected bone formation
candidate genes are associated with variation in serum OC levels.
Morrison et al. (14) reported a significant association
between OC levels and variants in the Vitamin D receptor gene (on
chromosome 11), although later studies have failed to replicate this
finding (32, 33, 34, 35). Association studies of serum OC levels and gene
variants have been performed for several other candidate genes,
including type I collagen 1 gene (chromosome 17q) (36), the
estrogen receptor gene (chromosome 6q) (37, 38), and the
transforming growth factor ß I gene (chromosome 19q) (39), although
few, if any, associations have been reported. We detected no evidence
for linkage to any of these gene regions in our analysis. It is
possible that the relative effects of different genes on phenotypic
variation differ among populations, and results (including linkage)
obtained from our Mexican American population may not be generalizable
to others.
In our analysis, we parameterized the genetic effects on OC as additive genetic effects. However, the basic variance components model for linkage can be extended to incorporate parameters that account for additional sources of genetic variation, such as dominance effects, where the genotypic mean of heterozygotes does not fall halfway between that of the two homozygotes (22). In practice, however, estimation of dominance effects requires a large number of sib-pairs because this is the only type of relative pair that has the potential for sharing both zero and two alleles identical-by-descent. We did not incorporate dominance QTL effects in our model because of the low power to detect such effects in our sample. However, if dominance effects do exist at the QTL, then failure to account for them should underestimate the evidence for linkage.
In summary, our study provides evidence that genes influencing serum OC levels may be located on chromosomes 16 and 20. CDMP1, which has been mapped to the region of 20q11.2, is believed to play a role in bone formation, and its location corresponds to our region of peak linkage on chromosome 20. Further investigation of CDMP1 and other genes in these regions may enhance our understanding of the factors influencing bone remodeling.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received July 29, 1999.
Revised December 7, 1999.
Accepted January 11, 2000.
|
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]
This article has been cited by other articles:
 |
|
 |
|
  H.-W. Deng, F.-H. Xu, Q.-Y. Huang, H. Shen, H. Deng, T. Conway, Y.-J. Liu, Y.-Z. Liu, J.-L. Li, H.-T. Zhang, et al. A Whole-Genome Linkage Scan Suggests Several Genomic Regions Potentially Containing Quantitative Trait Loci for Osteoporosis J. Clin. Endocrinol. Metab., November 1, 2002; 87(11): 5151 - 5159. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M. Kammerer, D. L. Rainwater, L. A. Cox, J. L. Schneider, M. C. Mahaney, J. Rogers, and J. L. VandeBerg Locus Controlling LDL Cholesterol Response to Dietary Cholesterol Is on Baboon Homologue of Human Chromosome 6 Arterioscler. Thromb. Vasc. Biol., October 1, 2002; 22(10): 1720 - 1725. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. I Steinle, W.-C. Hsueh, S. Snitker, T. I Pollin, H. Sakul, P. L St Jean, C. J Bell, B. D Mitchell, and A. R Shuldiner Eating behavior in the Old Order Amish: heritability analysis and a genome-wide linkage analysis Am. J. Clinical Nutrition, June 1, 2002; 75(6): 1098 - 1106. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H.-W. Deng, F.-H. Xu, T. Conway, X.-T. Deng, J.-L. Li, K. M. Davies, H. Deng, M. Johnson, and R. R. Recker Is Population Bone Mineral Density Variation Linked to the Marker D11S987 On Chromosome 11q12-13? J. Clin. Endocrinol. Metab., August 1, 2001; 86(8): 3735 - 3741. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. M. Kammerer, L. A. Cox, M. C. Mahaney, J. Rogers, and R. E. Shade Sodium-Lithium Countertransport Activity Is Linked to Chromosome 5 in Baboons Hypertension, February 1, 2001; 37(2): 398 - 402. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
