This Article Abstract Full Text (PDF) All Versions of this Article: 91/9/3575    most recent Author Manuscript (PDF) Submit a related Letter to the Editor Purchase Article View Shopping Cart Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Stewart, T. L. Articles by Ralston, S. H. Search for Related Content PubMed PubMed Citation Articles by Stewart, T. L. Articles by Ralston, S. H. Related Collections Calcium and Bone Metabolism

Haplotypes Defined by Promoter and Intron 1 Polymorphisms of the COLIA1 Gene Regulate Bone Mineral Density in Women

Department of Medicine and Therapeutics (T.L.S., H.J., F.E.A.M., A.B., D.M.R., S.H.R.), University of Aberdeen Medical School, Aberdeen AB25 2ZD, United Kingdom; Rheumatic Diseases Unit (H.J., O.M.E.A., S.H.R.), School of Molecular and Clinical Medicine, University of Edinburgh, Edinburgh EH4 2XU, United Kingdom; and University of Barcelona (N.G.-G., D.G., S.B.), Barcelona, Spain

Address all correspondence and requests for reprints to: Professor Stuart H. Ralston, Rheumatic Diseases Unit, Western General Hospital, Edinburgh EH4 2XU, United Kingdom. E-mail: stuart.ralston{at}ed.ac.uk.

	Abstract

Top Abstract Introduction Subjects and Methods Results Discussion References

 
Context: The COLIA1 gene is a strong candidate for susceptibility to osteoporosis. The causal genetic variants are currently unclear, but the most likely are functional polymorphisms in the promoter and intron 1 of COLIA1.

Objective: The objective of the study was to determine whether promoter and intron 1 polymorphisms of COLIA1 or haplotypes defined by these polymorphisms regulate bone mineral density (BMD) in women.

Design: This was a population-based association study involving 3270 women from the United Kingdom who took part in a regional osteoporosis screening program.

Main Outcome Measures: BMD at the lumbar spine (LS-BMD) and femoral neck (FN-BMD) was measured on two occasions approximately 6 yr apart, in relation to polymorphisms and haplotypes defined by polymorphisms within the COLIA1 intron 1 (+1245G/T; rs1800012) and promoter (–1997G/T; rs1107946; –1663IndelT; rs2412298).

Results: The polymorphisms were in strong linkage disequilibrium, and three haplotypes accounted for more than 95% of alleles at the COLIA1 locus. The individual polymorphisms were associated with BMD, but the most consistent associations were with haplotypes defined by all three polymorphisms. Homozygote carriers of haplotype 2 (–1997G/–1663delT/+1245T) had reduced BMD at baseline (P = 0.007 for LS-BMD; P = 0.008 for FN-BMD), whereas homozygotes for haplotype 3 (–1997T/–1663insT/+1245G) had increased BMD (P = 0.007 for LS-BMD). Similar associations were observed at follow-up for haplotype 3, but the association with haplotype 2 was weaker due to increased uptake of hormone replacement therapy in homozygotes for this haplotype.

Conclusions: Two haplotypes defined by polymorphisms in the 5' flank of the COLIA1 regulate BMD in a bidirectional manner in women.

	Introduction

Top Abstract Introduction Subjects and Methods Results Discussion References

 
OSTEOPOROSIS IS A common disorder characterized by low bone mass, microarchitectural deterioration of bone tissue, and enhanced bone fragility leading to an increased incidence of fracture. Genetic factors are recognized to play an important role in the regulation of bone mass and other phenotypes relevant to the pathogenesis of osteoporosis such as bone turnover, bone geometry, and fracture itself (1). Many candidate gene polymorphisms have been implicated in the pathogenesis of osteoporosis, but one of the most widely studied is a polymorphism that affects Sp1 binding site in the first intron of the COLIA1 gene at position +1245 relative to the transcription start site (rs1800012) (2). The COLIA1 Sp1 binding site polymorphism has been associated with bone density (2, 3), fragility fractures (2, 3), and other osteoporosis-related phenotypes such as postmenopausal bone loss (4, 5), bone geometry (6), bone quality (7), and bone mineralization (8). Functional analysis has shown that the osteoporosis-associated T allele ("s") of the Sp1 polymorphism is associated with increased DNA-protein binding, increased transcription, and abnormally increased production of the collagen type I1 mRNA and protein (7). It is thought that the resulting imbalance between the type I collagen I and 2 chains contributes to impairment of bone strength and reduced bone mass in carriers of the T allele by subtly affecting bone mineralization (8). Retrospective meta-analyses of published studies concluded that carriage of the T allele is associated with reduced bone mineral density (BMD) at the lumbar spine and femoral neck and vertebral fractures (7, 9, 10). Moreover, homozygotes for the T allele of the Sp1 polymorphism were recently found to be associated with BMD and incident vertebral fractures in a large prospective meta-analysis of more than 20,000 participants of the Genetic Markers for Osteoporosis (GENOMOS) study, which included subjects from the population described here (11).

We also previously reported that the association between COLIA1 alleles and vertebral fracture in women from northeast Scotland was primarily driven by the Sp1 binding site polymorphism rather than other known polymorphisms at the COLIA1 locus (12). Since publication of this study, however, two other polymorphisms have been identified in the promoter of COLIA1, which are in strong linkage disequilibrium (LD) with the Sp1 binding site polymorphism (13). These are a G/T polymorphism at position –1997 relative to the transcription start site (–1997G/T; rs1107946) and an insertion/deletion polymorphism at position –1663 (–1663indelT; rs2412298). These polymorphisms were found to interact with each other and with the Sp1 polymorphism to regulate BMD in a small study of Spanish postmenopausal women (13) and to influence COLIA1 transcription in promoter-reporter assays (14). The –1997G/T promoter polymorphism has been studied in relation to BMD in other populations and family-based studies with mixed results, although most of these studies have been of limited sample size (15, 16, 17). In this study, therefore, we examined associations between BMD and the promoter and Sp1 binding site polymorphisms of the COLIA1 gene and haplotypes defined by these polymorphisms in a large, population-based study of 3270 predominantly postmenopausal women from the United Kingdom.

	Subjects and Methods

Top Abstract Introduction Subjects and Methods Results Discussion References

 
Subjects

The study subjects participated in a population-based screening study for osteoporotic fracture risk, involving 5119 unrelated women aged 45–54 yr who were recruited at random from the general population within a 32-km radius of Aberdeen by use of the Community Heath Index number as previously described (18). Participants underwent bone densitometry between 1990 and 1994 and completed a risk factor questionnaire including questions on menopausal status and use of hormone replacement therapy (HRT) as previously described (19). Weight was measured on a set of balanced scales calibrated to 0.05 kg (SECA, Hamburg, Germany), and height was measured with a stadiometer (Holtain, Ltd., Crymych, UK). Women were classified as premenopausal if they were menstruating regularly, perimenopausal if their menstruation was irregular and/or up to 6 months had elapsed since their last period, and postmenopausal if their menstruation had ceased for 6 months or more. After this assessment, the BMD results were disclosed to participants and their general practitioners, and subjects with BMD values in the lowest quartile of the population were advised to consider taking HRT therapy as prophylaxis against osteoporosis (18). Previous studies have shown that the uptake of HRT in participants of this study (approximately 30%) is 6% greater than in women from the same region who had not undergone BMD screening (18).

All participants were invited to undergo further assessment between 1997 and 1999, and a total of 3883 women of the original 5119 attended (75.8%). At this time, the baseline questionnaire was repeated, and additional information was collected on smoking habit and physical activity using a questionnaire that was previously used in the Scottish Heart Health Study (20), which gives a physical activity level (PAL) score, calculated from the number of hours in a 24-h period doing heavy, moderate, or light activities and how many hours are spent sleeping or resting in bed. Participants in the present study comprised 3270 women who consented to give blood samples for DNA extraction at the follow-up visit and were successfully genotyped for at least one of the polymorphisms studied.

Bone density measurements

BMD was measured at the femoral neck and lumbar spine (L2–4) by dual-energy x-ray absorptiometry using XR26 and XR36 densitometers (Norland, Fort Atkinson, WI). The majority of women were scanned using an XR26 machine, but 357 women (11.5%) were scanned using an XR36 machine. Cross-calibration of the machines was done using the manufacturer’s phantom, and this revealed a small difference (1.258%) in mean BMD value between the machines. This correction factor was used to convert the XR36 values to XR26-equivalent values. We previously validated this approach, and the use of the scanner-corrected BMD values has been shown to give similar results to analyses in which scanner type is entered into the statistical model as a random factor in assessing genotype-specific differences in BMD (21). There were no significant differences in the proportion of women scanned on each machine across different genotypes.

Genotyping

Genotyping for –1997G/T and –1663indelT polymorphisms was carried out by DNA sequencing of PCR-amplified fragments of genomic DNA, whereas genotyping for the COLIA1 Sp1 polymorphism was carried out predominantly by TaqMan analysis as previously described (22). All PCRs were performed in 96-well plates in a 25-µl volume containing 2.5 µl 10x PCR buffer (QIAGEN Ltd, Crawly, UK), 2.0 µl deoxynucleotide triphosphate (10 mM each), 5.0 µl Q-solution, 2.5 µl of oligonucleotide primer mix (5 µM each of forward and reverse primer), 0.625 U Taq DNA polymerase (QIAGEN), and approximately 50 ng genomic DNA. The PCRs were carried out on a Gene Engine thermocycler (MJ Research Inc., Waltham, MA). The PCR products were treated with ExoSAP IT (USB Corp., Cleveland, OH) according to the manufacturer’s instructions and sequenced using the DYEnamic ET dye terminator cycle sequencing kit (Amersham Pharmacia Biotech U.K. Ltd., Buckinghamshire, UK) according to the manufacturer’s instructions. The sequencing reactions were analyzed using a MEGABACE automated DNA sequencer (Amersham Pharmacia Biotech).

For the –1663 and –1997 polymorphisms, we generated a 513-bp fragment using the following primers: 5'-TCACTAACCCTCATACTACCAAGC-3' and 5'-AAGATTCCATTGCCTCCCCC-3', and this was sequenced using these primers or an internal reverse primer 5'-CCTTTAATTATAGCCCCTGCA-3'. The thermal cycling protocol for both fragments consisted of an initial incubation for 5 min at 94 C, followed by 31 cycles of 60 sec at 94 C, 30 sec at 61 C, and 60 sec of 72 C followed by an extension step at 72 C for 5 min in the last cycle. A random sample of approximately 5% of subjects were genotyped twice for each polymorphism to assess the genotyping error rate, and this showed 100% concordance between genotyping runs. For the –1663indelT polymorphism, all subjects who were heterozygous or homozygous for the –1663delT allele were genotyped twice, with 100% concordance between genotyping runs. Genotyping was successfully performed for the COLIA1 Sp1 polymorphism in 3225 individuals, the –1997G/T polymorphism in 3011 individuals, and the –1663indelT polymorphism in 2986 individuals. Haplotypes were estimated from genotype data for individual participants using the PHASE program (23). For the purpose of statistical analysis, we only used haplotype data in which the probability of correct haplotype assignment by PHASE in an individual case was estimated as 95% or greater.

Statistical analysis

Statistical analyses were performed using Minitab (version 12.23; State College, PA). The ² test was used to test for Hardy-Weinberg equilibrium. Levene’s test was used to check for equality of variances between genotype and haplotype groups. Differences in BMD values between genotypes and haplotypes were analyzed by ANOVA at the baseline and follow-up visit using the general linear model routine of Minitab. For this analysis, genotypes and haplotypes were entered into the model along with other potential predictors such as body mass index (BMI), PAL score, age, smoking (current smoker or nonsmoker), menopausal status (premenopausal vs. perimenopausal and postmenopausal), and HRT use (never used HRT, previously used HRT, current use of HRT). A similar analysis was performed for BMD values at the baseline visit, but here only genotype or haplotype, age, BMI, HRT use, and menopausal status were entered into the model because information on smoking and PAL was not available at this time point. When significant differences were detected between genotype or haplotype groups using this procedure, the means were compared between groups using Tukey’s test. Analysis of the BMD data using Levene’s test showed no significant difference in variances for any of the individual genotype or haplotype groups studied. Multiple regression analysis and stepwise multiple regression analysis were used to evaluate associations between genotypes and haplotypes and BMD and estimate the effect size of the genotype and haplotypes in relation to other predictors of BMD. For this analysis, continuous variables (age, BMI, and PAL) were entered into the statistical model as covariates. For categorical variables, such as menopausal status, HRT use, and COLIA1 genotype and haplotype, dummy variables were constructed for use in the multiple regression analysis.

Although we performed multiple statistical tests of three polymorphisms (and haplotypes defined by these polymorphisms) in relation to BMD at baseline and follow-up at two skeletal sites, these factors were interrelated and therefore not independent. In view of this, we used a modified Bonferroni correction to adjust for the number of independent tests performed to take into account the fact that the polymorphisms were in LD and the BMD measurements at the spine and hip at baseline and follow-up were strongly correlated. To make this correction, we took an average of the D' values for all three polymorphisms (0.92) and subtracted this figure from 1 to get an indication of the number of independent polymorphisms tested [1 + (2 _* 1 – 0.92)]. According to this calculation, the number of independent polymorphisms tested was 1.16. We analyzed two skeletal sites (femoral neck BMD and lumbar spine BMD), but these were significantly correlated (r = 0.668), and taking this correlation into account, we effectively analyzed 1.66 independent sites [1 + (2 _* 1 – 0.668)]. Similarly, baseline and follow-up BMD were strongly correlated (r = 0.883), and therefore, the analysis of genotype in relation to baseline and follow-up BMD can be considered as 1.22 independent tests (1 + 2 _* 1 – 0.883). In total, therefore, we conducted 4.05 independent tests (1.16 + 1.66 + 1.22). Applying a Bonferroni correction to our data using this figure gives an adjusted threshold for significance of P = 0.012, equating to the conventional threshold for significance of P = 0.05. The sample size gave approximately 99% power of detecting a difference of 0.15 BMD Z-score units between genotype groups, assuming a population frequency of 0.20 for the rare allele.

Ethics

Written informed consent was obtained for all the women, and the study was approved by the Grampian Research Ethics Committee.

	Results

Top Abstract Introduction Subjects and Methods Results Discussion References

 
Characteristics of population

The average ± SD age of the participants at the baseline visit was 48.4 ± 2.3 yr (range 44–56 yr) and 54.8 ± 2.3 yr at the follow-up visit. The average time between baseline and follow-up visits was 6.3 ± 0.85 yr and did not differ between genotype or haplotype groups (data not shown). At the baseline visit, 44.9% of participants were premenopausal, and 16.9% were current users of HRT. At the follow-up visit, only 3.2% of participants were premenopausal, and the proportion of current HRT users had roughly doubled to 37.5%. Genotype frequencies for all three polymorphisms were similar to those previously reported in other Caucasian populations (13, 17). All three polymorphisms were in strong LD, and D' values ranged from 0.88 to 0.96 (Fig. 1). Genotypes for the –1997 G/T polymorphism and Sp1 polymorphism were in Hardy-Weinberg equilibrium (P = 0.22 and P = 0.48, respectively), but genotypes for the –1663indelT polymorphism deviated significantly from Hardy Weinberg equilibrium (P = 0.0003) due to overrepresentation of homozygotes for the –1663InsT allele and underrepresentation of heterozygotes.

View larger version (15K): [in this window] [in a new window]   FIG. 1. Common haplotypes in the promoter and intron 1 of COLIA1. Common haplotypes and their frequencies at the 5' flank of the COLIA1 gene are shown in relation to the COLIA1gene structure. Shaded boxes represent exons. LD values (as D') between SNP are depicted in black diamonds at the bottom of the figure.

Association between individual polymorphisms and BMD

Relevant characteristics of the study population at baseline and follow-up in relation to the individual polymorphisms are shown in Table 1. The BMD values shown are least square means adjusted for age, BMI, HRT use, and menopausal status at the baseline assessment and age, BMI, HRT use, menopausal status, smoking, and physical activity score at the follow-up assessment.

At the follow-up visit, there was a highly significant difference in the proportion of HRT users in relation to the –1663indelT (P = 0.02) and Sp1 polymorphisms (P = 0.003). For these polymorphisms, the proportion of HRT users in heterozygotes was less than in the other genotype groups, and for the Sp1 polymorphism, there was an excess of HRT users among TT homozygotes. Genotype-specific associations with BMD at the follow-up visit for the –1997G/T polymorphism were similar to the baseline visit (P = 0.003), except that the association with femoral neck BMD at follow-up was nominally significant (P = 0.039). For the –1663indelT and Sp1 polymorphisms, the associations with BMD at follow-up were weaker than at baseline, and nominal significance was observed only for the Sp1 polymorphism in relation to lumbar spine BMD (P = 0.037). When we performed multiple regression analysis entering genotypes at all three polymorphisms into the regression model, assuming a recessive mode of inheritance, we found that homozygosity for the –1997T allele was an independent predictor of follow-up lumbar spine BMD (P = 0.001), accounting for 0.35% of the variance and femoral neck BMD (P = 0.027), accounting for 0.21% of the variance. Other nongenetic predictors of femoral neck and lumbar spine BMD at follow-up were similar to those at the baseline visit (data not shown).

We also analyzed the relationship between individual polymorphisms and unadjusted BMD values at each visit by ANOVA, and this yielded broadly similar results to the adjusted BMD values (data not shown).

Association between COLIA1 5' haplotypes and BMD

Relevant characteristics of the study population at baseline and follow-up in relation to the three commonest haplotypes are shown in Table 2. There was no significant association between any haplotype and age, BMI, menopausal status, or HRT use at the baseline visit, and there was no significant difference in adjusted lumbar spine BMD or femoral neck BMD values at baseline in relation to haplotype 1. There was a significant association between haplotype 2 and baseline BMD at the lumbar spine (P = 0.007) and femoral neck (P = 0.008), with reduced BMD values in homozygote carriers of haplotype 2. There was also a significant association between haplotype 3 and baseline BMD at the lumbar spine (P = 0.007), but the association with baseline femoral neck BMD was not significant (P = 0.224).

At the follow-up visit, there was no difference in age, BMI, smoking, or physical activity score in relation to haplotype (data not shown), but there was a significant difference in the proportion of HRT users in relation to carriers of haplotype 2, with a reduced number of HRT users in heterozygotes and an increased number of HRT users in haplotype 2 homozygotes (P = 0.002). There was no significant difference in adjusted lumbar spine BMD or femoral neck BMD values at follow-up in relation to haplotype 1. There was no significant association between haplotype 2 and follow-up BMD at the lumbar spine, contrasting with the significant association at baseline. Moreover, the association between haplotype 2 and femoral neck BMD at follow-up was weaker than at baseline (P = 0.033). For haplotype 3, the association with lumbar spine BMD at follow-up remained significant (P = 0.004), with a similar effect size as at the baseline visit. Haplotype 3 was associated with femoral neck BMD at follow-up with a nominally significant P value (P = 0.030).

Multiple regression analysis showed that haplotype 3 was a significant predictor of lumbar spine BMD at follow-up. Homozygosity for haplotype 3 accounted for 0.50% of the variance in BMD (P = 0.004). Other predictors were BMI (P < 0.001; 7.49% of the variance), age (P < 0.001; 2.563% of the variance), menopausal status (P < 0.001; 1.55% of the variance), and HRT use (P < 0.001; 1.1% of the variance). Together these variables predicted 13.2% of the variance in follow-up lumbar spine BMD. For follow-up femoral neck BMD, multiple regression analysis showed that homozygosity for haplotype 2 was a significant predictor of BMD (P = 0.021), accounting for 0.21% of the variance. Haplotype 3 was also a significant predictor of follow-up femoral neck BMD (P = 0.05), accounting for 0.16% of the variance. Other predictors were BMI (P < 0.001; 8.47% of the variance), age (P < 0.001; 2.12% of the variance), and menopausal status (P < 0.023; 0.18% of the variance). Together these variables predicted 10.93% of the variance in follow-up femoral neck BMD. Because haplotypes 2 and 3 had opposing effects on BMD, we stratified participants according to combinations of these two haplotypes. This showed a striking difference in BMD between the extreme genotypes at both baseline and follow-up as illustrated in Fig. 2. Subjects who were homozygous for haplotype 3 had the highest BMD, and those homozygous for haplotype 2 had the lowest BMD. The difference between haplotypes was greatest for lumbar spine BMD at baseline (P = 0.001) and follow-up (P = 0.008) but was also evident and statistically significant for femoral neck BMD at baseline (P = 0.02) and follow-up (P = 0.012). The difference in BMD between the extreme haplotypes (homozygotes for haplotype 2 vs. homozygotes for haplotype 3) was 0.094 g/cm² at the lumbar spine, which is equivalent to about 0.58 BMD Z-score units, and 0.042 g/cm² at the femoral neck, which is equivalent to about 0.33 BMD Z-score units. At follow up, the difference in BMD between the extreme genotypes 2/2 and 3/3 was 0.10 g/cm² at the lumbar spine, which is equivalent to about 0.81 BMD Z-score, and 0.061 g/cm² at femoral neck, which is equivalent to about 0.50 BMD Z-score units.

View larger version (22K): [in this window] [in a new window]   FIG. 2. Association between COLIA1 5' haplotypes and BMD. The BMD values shown are adjusted for confounding factors by ANOVA-general linear model in relation to carriage of haplotypes and 5 in the 5' flank of the COLIA1 gene. A and B, Results for lumbar spine BMD (LS-BMD) at baseline and follow-up, respectively. C and D, Results for femoral neck BMD (FN-BMD) at baseline and follow-up, respectively. Columns are least square means, and the vertical bars are SE values. 2/2, Homozygotes for haplotype 2; 2/0, heterozygotes for haplotype 2 who were not carriers of haplotype 3; 3/0, heterozygotes for haplotype 3 who were not carriers of haplotype 2; 3/3, homozygotes for haplotype 3; 0/0, subjects who were not carriers of either haplotype 2 or haplotype 3.

Genotype and haplotype associations with bone loss

There was no significant association between genotype or haplotype and bone loss at either the lumbar spine or femoral neck in the study population. Multiple regression analysis showed that the major predictors of lumbar spine bone loss were HRT use (P < 0.001; 14.59% of the variance); menopausal status (P < 0.001; 4.48% of the variance); BMI (P < 0.001; 3.04% of the variance); and age (P = 0.01; 0.19% of the variance). Together these variables accounted for 22.3% of the variance in bone loss. Similar analysis for femoral neck bone loss showed that the major predictors were HRT use (P < 0.001; 7.85% of the variance); menopausal status (P < 0.001; 2.17% of the variance); and BMI (P = 0.01; 0.4% of the variance). Together these variables accounted for 10.4% of the variance in femoral neck bone loss.

	Discussion

Top Abstract Introduction Subjects and Methods Results Discussion References

 
The COLIA1 gene is a strong functional candidate for the genetic regulation of bone mass and susceptibility to fragility fractures because it encodes the 1-chain of type I collagen, the most abundant protein in bone. Most previous studies of COLIA1 alleles in relation to osteoporosis-related phenotypes have focused on the intronic Sp1 binding site polymorphism (2). The promoter polymorphisms that we studied here were originally identified by Garcia-Giralt et al. (13) on mutation screening of the proximal COLIA1 promoter in 265 postmenopausal Spanish women. In that study, the –1997G/T polymorphism was the one most strongly associated with BMD, and lumbar spine BMD values were reduced in TT homozygotes, compared with the other genotype groups (13). Whereas the –1663InsT/DelT polymorphism was not significantly associated with BMD in Spanish women, regression analysis showed that alleles at the –1663 and Sp1 site interacted with alleles at the –1997 site to regulate lumbar spine BMD. The relation between the –1997G/T polymorphism and BMD has also been investigated in other cohorts. In a study of 308 postmenopausal Caucasian women from the United States, haplotypes defined by the –1997 and Sp1 polymorphisms were found to be associated with BMD such that higher values were found in carriers of the G allele at both polymorphic sites (17). Another family-based study in Chinese subjects showed evidence of an association between the –1997G/T polymorphism and hip BMD using a quantitative transmission disequilibrium testing-based approach (15), whereas a population-based study of unrelated Chinese women showed no association between –1997G/T alleles and BMD (25). A further study of 1100 Japanese postmenopausal women reported a significant association between –1997G/T alleles and hip BMD with lower BMD values in GG homozygotes (16).

The present study of promoter and intron 1 polymorphisms at the COLIA1 locus is the largest and most comprehensive to be conducted so far and is the first study to examine the role of all three polymorphisms and haplotypes in relation to BMD. In agreement with the previous study of Spanish women (13), we found that all three polymorphisms were in strong LD and three common haplotypes accounted for more than 95% of alleles at the 5' flank of the COLIA1 locus. These were haplotype 1 (–1997G/–1666InsT/+1245G); haplotype 2 (1997G/–1663delT/+1245T); and haplotype 3 (–1997T/–1663InsT/+1245G). Whereas the individual polymorphisms were associated with BMD to an extent, we observed stronger and more consistent associations with BMD in relation to the carriage of the COLIA1 haplotypes and also significant differences in the strength of association by skeletal site and the time point at which participants were studied. For the –1997G/T polymorphism, homozygotes for the T allele had increased lumbar spine BMD at baseline and follow-up and had increased femoral neck BMD values at follow-up.

These observations are in broad agreement with the findings reported by Yamada et al. (16), who also reported higher BMD values in relation to the –1997 T allele in Japanese postmenopausal women. Our findings differ from the results reported in Spanish and U.S. postmenopausal women, however, where the –1997T allele was associated with reduced BMD (13, 17). The differences between these studies could be due to the fact that the polymorphisms studied here are in LD with other functional polymorphisms at the COLIA1 locus that regulate BMD and that the patterns of LD differ in these different populations. Another more likely explanation is due to a more accurate estimate of effect in the study by Yamada et al. (16) and this study, which, respectively, were about 3 and 10 times larger than the studies that reported an association between the –1997T allele and low BMD (13, 17).

For the –1663indelT and Sp1 polymorphism, we observed a strong and highly significant association with lumbar spine BMD values at baseline, even after correction for multiple testing, but these associations were weaker at follow-up. The difference that we observed between baseline and follow-up may have been influenced by differential use of HRT in the different genotype groups. This was most marked and statistically significant for the –1663indelT and Sp1 polymorphisms, for which there was a highly significant difference across genotype groups in terms of HRT use at follow-up, compared with no difference at baseline. This difference in HRT use may be accounted for by the fact that women in this cohort who had BMD values in the lowest quartile at baseline were advised to take HRT to protect against osteoporosis (18). Because BMD values were significantly lower in homozygotes for the rare allele at the –1663indelT and Sp1 sites at baseline, this could have accounted for higher rates of HRT uptake at follow-up in these genotype groups.

The association between haplotypes and BMD was stronger and more consistent across skeletal sites than the associations with individual genotypes. Haplotypes 2 and 3 were both strongly associated with lumbar spine BMD at baseline, and haplotype 2 was also strongly associated with femoral neck BMD at baseline. At follow-up, the association between haplotype 3 and lumbar spine BMD was similar to the baseline visit, whereas the association with haplotype 2 was weaker than at baseline and was not statistically significant. These differences were most likely due to differences in uptake of HRT in carriers of haplotype 2, reflected by the significant difference in the proportion of HRT users in this haplotype group at the follow-up visit as compared with baseline. As was the case with the individual polymorphisms, we speculate that this difference may be increased by the use of HRT in homozygote carriers of haplotype 2.

The P values for significance reported here should be interpreted in light of the fact that we studied associations among three polymorphisms and associated haplotypes in relation to BMD at two sites, measured on two occasions. However, taking into account the LD between polymorphisms and the fact that the traits were correlated, many of the associations we observed remained below the adjusted threshold for significance (P = 0.012), using a modified Bonferroni test to correct for the number of independent tests performed.

The haplotype associations that we report here are of interest in relation to a recent study using promoter-reporter assays in which it was shown that promoter constructs containing the –1997G-DelT allele (corresponding to haplotype 2) had higher rates of transcription than the –1997T/–1663InsT allele (corresponding to haplotype 3) (14). The effects of the COLIA1 Sp1 polymorphism on transcription were not investigated in the reporter studies mentioned above, but we have recently found that constructs containing the promoter and intron 1 alleles (corresponding to haplotype 2) also drive transcription at higher levels than constructs containing alleles of haplotype 3 (Jin, H., and S. H. Ralston, unpublished data). These observations suggest that there is an association between high levels of COLIA1 transcription and reduced BMD values, consistent with previous work that has suggested that the Sp1 T allele is associated with increased COLIA1 transcription, leading to an imbalance in the ratio of the collagen type I1 and collagen type I2 chains, resulting in a subtle impairment of bone mineralization (8). The present observations indicate that the previously reported association between the COLIA1 Sp1 T allele and increased allele-specific transcription (7) may actually be driven by an extended haplotype driven by both the intron 1 and promoter polymorphisms.

In summary, this study demonstrates that haplotypes, defined by the promoter and intron 1 of the COLIA1 gene, regulate BMD in postmenopausal women. One of the most interesting observations to emerge from the study was the much higher uptake of HRT in women who carried genotypes and haplotypes that were associated with low BMD at the baseline assessment. This example of a gene-environment interaction illustrates the challenges faced by investigators who seek to detect genetic variants that regulate BMD in populations such as this, in which systematic BMD screening has been followed by a specific therapeutic intervention. Whereas the mechanism by which COLIA1 variants regulate BMD remains to be fully elucidated, current evidence is consistent with an effect of these different haplotypes on COLIA1 transcription.

	Acknowledgments

 
We acknowledge the contribution of Mr. Stuart Bear, Ms. Rosie Farmer, Ms. Jay Wallace, and Mrs. Grace Taylor to the genotyping effort; Allan Walker for database support; and the staff of the Osteoporosis Unit (Woolmanhill Hospital, Aberdeen, UK) for data collection and bone densitometry.

	Footnotes

 
This work was supported by Grant GENOMOS-QLK6-CT-2002-02629 from the European Commission and Cooperative Group Grant G9823281 from the Medical Research Council, United Kingdom.

S.H.R. holds a patent on the use of the Sp1 polymorphism of COLIA1 as a genetic marker for osteoporosis. The other authors have nothing to declare.

First Published Online June 27, 2006

¹ T.L.S. and H.J. contributed equally to this work.

Abbreviations: BMD, Bone mineral density; BMI, body mass index; HRT, hormone replacement therapy; LD, linkage disequilibrium; PAL, physical activity level.

Received December 7, 2005.

Accepted June 16, 2006.

	References

Top Abstract Introduction Subjects and Methods Results Discussion References

 

1. Albagha OM, Ralston SH 2003 Genetic determinants of susceptibility to osteoporosis. Endocrinol Metab Clin North Am 32:65–81[CrossRef][Medline]
2. Grant SFA, Reid DM, Blake G, Herd R, Fogelman I, Ralston SH 1996 Reduced bone density and osteoporosis associated with a polymorphic Sp1 site in the collagen type I 1 gene. Nat Genet 14:203–205[CrossRef][Medline]
3. Uitterlinden AG, Burger H, Huang Q, Yue F, McGuigan FE, Grant SF, Hofman A, van Leeuwen JP, Pols HA, Ralston SH 1998 Relation of alleles of the collagen type I 1 gene to bone density and risk of osteoporotic fractures in postmenopausal women. N Engl J Med 338:1016–1022[Abstract/Free Full Text]
4. MacDonald HM, McGuigan FEA, New SA, Campbell MK, Golden MH, Ralston SH, Reid DM 2001 COL1A1 Sp1 polymorphism predicts perimenopausal and early postmenopausal spinal bone loss. J Bone Miner Res 16:1634–1641[CrossRef][Medline]
5. Harris SS, Patel MS, Cole DE, Dawson-Hughes B 2000 Associations of the collagen type I 1 Sp1 polymorphism with five-year rates of bone loss in older adults. Calcif Tissue Int 66:268–271[CrossRef][Medline]
6. Qureshi AM, McGuigan FEA, Seymour DG, Hutchison JD, Reid DM, Ralston SH 2001 Association between COLIA1 Sp1 alleles and femoral neck geometry. Calcif Tissue Int 69:67–72[CrossRef][Medline]
7. Mann V, Hobson EE, Li B, Stewart TL, Grant SF, Robins SP, Aspden RM, Ralston SH 2001 A COL1A1 Sp1 binding site polymorphism predisposes to osteoporotic fracture by affecting bone density and quality. J Clin Invest 107:899–907[Medline]
8. Stewart TL, Roschger P, Misof BM, Mann V, Fratzl P, Klaushofer K, Aspden R, Ralston SH 2005 Association of COLIA1 Sp1 alleles with defective bone nodule formation in vitro and abnormal bone mineralisation in vivo. Calcif Tiss Int 77:113–118[CrossRef][Medline]
9. Efstathiadou Z, Tsatsoulis A, Ioannidis JP 2001 Association of collagen I1 Sp1 polymorphism with the risk of prevalent fractures: a meta-analysis. J Bone Miner Res 16:1586–1592[CrossRef][Medline]
10. Mann V, Ralston SH 2003 Meta-analysis of COL1A1 Sp1 polymorphism in relation to bone mineral density and osteoporotic fracture. Bone 32:711–717[Medline]
11. Ralston SH, Uitterlinden AG, Brandi ML, Balcells S, Langdahl BL, Lips P, Lorenc R, Obermayer-Pietsch B, Scollen S, Bustamante M, Husted LB, Carey AH, Diez-Perez A, Dunning AM, Falchetti A, Karczmarewicz E, Kruk M, van Leeuwen JP, van Meurs JB, Mangion J, McGuigan FE, Mellibovsky L, del Monte F, Pols HA, Reeve J, Reid DM, Renner W, Rivadeneira F, van Schoor NM, Sherlock RE, Ioannidis JP; GENOMOS Investigators 2006 Large-scale evidence for the effect of the COLIA1 Sp1 polymorphism on osteoporosis outcomes: the GENOMOS study. PLoS Med 3:e90
12. McGuigan FE, Reid DM, Ralston SH 2000 Susceptibility to osteoporotic fracture is determined by allelic variation at the Sp1 site, rather than other polymorphic sites at the COL1A1 locus. Osteoporos Int 11:338–343[CrossRef][Medline]
13. Garcia-Giralt N, Nogues X, Enjuanes A, Puig J, Mellibovsky L, Bay-Jensen A, Carreras R, Balcells S, Diez-Perez A, Grinberg D 2002 Two new single nucleotide polymorphisms in the COLIA1 upstream regulatory region and their relationship with bone mineral density. J Bone Miner Res 17:384–393[CrossRef][Medline]
14. Garcia-Giralt N, Enjuanes A, Bustamante M, Mellibovsky L, Nogues X, Carreras R, Diez-Perez A, Grinberg D, Balcells S 2005 In vitro functional assay of alleles and haplotypes of two COL1A1-promoter SNPs. Bone 36:902–908[Medline]
15. Zhang YY, Lei SF, Mo XY, Wang YB, Li MX, Deng HW 2005 The –1997 G/T polymorphism in the COLIA1 upstream regulatory region is associated with hip bone mineral density (BMD) in Chinese nuclear families. Calcif Tissue Int 76:107–112[CrossRef][Medline]
16. Yamada Y, Ando F, Niino N, Shimokata H 2005 Association of a –1997G–>T polymorphism of the collagen I1 gene with bone mineral density in postmenopausal Japanese women. Hum Biol 77:27–36[CrossRef][Medline]
17. Liu PY, Lu Y, Long JR, Xu FH, Shen H, Recker RR, Deng HW 2004 Common variants at the PCOL2 and Sp1 binding sites of the COL1A1 gene and their interactive effect influence bone mineral density in Caucasians. J Med Genet 41:752–757[Abstract/Free Full Text]
18. Torgerson DJ, Thomas RE, Campbell MK, Reid DM 1997 Randomized trial of osteoporosis screening. Use of hormone replacement therapy and quality-of-life results. Arch Intern Med 157:2121–2125[Abstract]
19. MacDonald HM, New SA, Campbell MK, Reid DM 2003 Longitudinal changes in weight in perimenopausal and early postmenopausal women: effects of dietary energy intake, energy expenditure, dietary calcium intake and hormone replacement therapy. Int J Obes Relat Metab Disord 27:669–676[CrossRef][Medline]
20. Tunstall-Pedoe H, Smith WC, Crombie IK, Tavendale R 1989 Coronary risk factor and lifestyle variation across Scotland: results from the Scottish Heart Health Study. Scott Med J 34:556–560[Medline]
21. Pettersson U, Albagha OM, Mirolo M, Mirolo M, Taranta A, Frattini A, McGuigan FE, Vezzoni P, Teti A, van Hul W, Reid DM, Villa A, Ralston SH 2005 Polymorphisms of the CLCN7 gene are associated with BMD in women. J Bone Miner Res 20:1960–1967[CrossRef][Medline]
22. McGuigan FE, Ralston SH 2002 Single nucleotide polymorphism detection: allelic discrimination using TaqMan. Psychiatr Genet 12:133–136[CrossRef][Medline]
23. Stephens M, Smith NJ, Donnelly P 2001 A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 68:978–989[CrossRef][Medline]
24. Wittke-Thompson JK, Pluzhnikov A, Cox NJ 2005 Rational inferences about departures from Hardy-Weinberg equilibrium. Am J Hum Genet 76:967–986[CrossRef][Medline]
25. Lau HH, Ng MY, Ho AY, Luk KD, Kung AW 2005 Genetic and environmental determinants of bone mineral density in Chinese women. Bone 36:700–709[Medline]

This article has been cited by other articles:

				A. L. Boskey Mineralization of Bones and Teeth Elements, December 1, 2007; 3(6): 385 - 391. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 91/9/3575    most recent Author Manuscript (PDF) Submit a related Letter to the Editor Purchase Article View Shopping Cart Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Stewart, T. L. Articles by Ralston, S. H. Search for Related Content PubMed PubMed Citation Articles by Stewart, T. L. Articles by Ralston, S. H. Related Collections Calcium and Bone Metabolism