Common Genetic Variation of the Low-Density Lipoprotein Receptor-Related Protein 5 and 6 Genes Determines Fracture Risk in Elderly White Men

Authors


  • The authors have no conflict of interest

Abstract

Both LRP5 and LRP6 genes have been implicated to play a role in bone metabolism. In a large population-based study, we related common variation in both genes to bone parameters and fractures. LRP5 variation was associated to both BMD and frame size, whereas both LRP5 and 6 variations were associated with an increased fracture risk in males.

Introduction: The low-density lipoprotein receptor-related protein 5 (LRP5) gene has a clear role in rare BMD traits and also in normal variation in peak BMD. We examined whether common variation in LRP5 and its close homolog, LRP6, plays a role in BMD in old age and fractures, the main clinical endpoint of osteoporosis.

Materials and Methods: We analyzed four variants of LRP5 and one amino acid variant of the LRP6 gene in a large prospective population-based cohort study of elderly subjects.

Results and Conclusions: In men, the LRP5 1330-valine variant was associated with decreased BMD at the lumbar spine and the femoral neck with evidence for an allele-dose effect (p = 0.001 and 0.01, respectively). The Val allele was also associated with decreased vertebral body size and femoral neck width. Haplotype analysis of studied polymorphisms did not improve the association found and suggested that the 1330 variant was driving the association. We observed a borderline significant association of the LRP6Ile1062Val polymorphism with height and vertebral body size in males. Male carriers of the LRP5 1330-valine variant had a 60% increased risk for fragility fractures, and the LRP61062-valine allele also conferred a 60% higher risk. Carriers of both the risk alleles of LRP5 and 6 had a 140% (p = 0.004) higher risk compared with noncarriers of both risk alleles and accounted for 10% of the fractures in males. The fracture risks were independent of age, height, weight, and BMD. In women, all of these associations were weaker and less consistent compared with men. The polymorphisms that were found associated were both situated in potentially important domains of the receptor and show considerable evolutionary conservation, which is evidence for functional importance of these residues.

INTRODUCTION

OSTEOPOROSIS IS CHARACTERIZED by low BMD and microarchitectural deterioration leading to bone fragility and increased fracture risk. Heritability studies show that genetic factors may account for 50–80% of the variability in BMD. (1, 2) Whereas the genes that contribute to variation in risk for osteoporosis are largely unknown, it is thought that the risk to develop osteoporosis is dependent on several common gene variants with modest but real genetic effects. (3, 4) The most promising approach to identify the individual genetic factors for osteoporosis unto now is the candidate gene approach, which has been successfully applied for several genes (e.g., ESR1 and COLIA1).(5–7)

Osteoporosis candidate genes are identified, for example, because mutations in them lead to severe Mendelian bone phenotypes. An example of this concerns the low-density lipoprotein receptor-related protein 5 gene (LRP5; 11q12). Recently, it has been shown that loss-of-function mutations of the LRP5 gene in both human and mice lead to decreased bone mass, (8, 9) whereas point mutations in this same gene result in a high bone mass phenotype. (10, 11) LRP5 is a member of a family of cell surface receptors involved in diverse biological processes and acts as a co-receptor for Wnt proteins. (9, 12) The bone-related phenotype found in humans and mice lacking a normally functioning LRP5 protein are thought to be caused largely by an osteoblastic defect, (8, 9) which are the cells responsible for the formation of bone matrix.

A closely related homolog of LRP5, LRP6, is also capable to act as a co-receptor for Wnt ligands. LRP6-deficient mice display severe developmental abnormalities similar to several Wnt-deficient mice, and die before birth. (13) Mice studies showed that a point mutation in LRP6 led to abnormal formation of the axial skeleton and a low bone mass phenotype. (14) A recent report analyzed mice that were heterozygous carriers for mutations in both the LRP5 and LRP6 genes and showed that LRP5 and LRP6 genetically interact in limb development and BMD in mice. (15)

Common variations in the LRP5 and LRP6 genes might therefore contribute to normal population variance in bone metabolism. In this study, we examined the role of polymorphisms in the LRP5 and LRP6 genes in determining BMD, bone geometry, and fracture risk in a large population-based cohort of elderly white individuals.

MATERIALS AND METHODS

Study population

Subjects were participants of the Rotterdam Study, a prospective population-based cohort study of individuals ≥55 years of age. The study was designed to investigate the incidence and determinants of chronic disabling diseases. Rationale and design have been described previously. (16) The Medical Ethics Committee of Erasmus University Medical School approved the Rotterdam Study, and written informed consent was obtained from each subject. All 10,275 inhabitants ≥55 years of age of a district in Rotterdam, The Netherlands, were invited for baseline examination between August 1990 and June 1993. Of those, 7983 participated. For haplotype analysis of the LRP5 gene, a random set of individuals (n = 2293) were used. Subsequently, the complete study cohort (n = 6373) was genotyped for the selected polymorphisms. In addition, we determined allele frequencies in 130 white blood bank donors.

Clinical examination

At baseline, BMD (g/cm2) was measured at the femoral neck and lumbar spine (L2-L4) by DXA (Lunar DPX-L densitometer; Lunar Corp.) as described previously. Height and weight were measured in standing position in indoor clothing without shoes. Body mass index (BMI) was computed as weight in kilograms divided by height in meters squared (kg/m2).

Hip structural analysis

We used the hip structural analysis (HSA) software developed by Thomas J. Beck to measure hip bone geometry from the DXA scans. (17, 18) Analysis locations included the narrow-neck region across the narrowest point of the femoral neck. BMD, bone cross-sectional area (CSA), bone width (outer diameter), and cross-sectional moment of inertia (CSMI) and were measured directly. In addition, estimates of cortical thickness were obtained with simple models of the cross-sections that use measured dimensions and assumptions of cross-section shape. Calculations of section moduli (Z), an index of bending strength, and buckling ratios (BR), index of bone instability, were slightly modified from those of previous reports to account for shifts in the center of mass. Z was calculated as CSMI/ds, where ds is the maximum distance from the center of mass to the medial or lateral surface. Buckling ratios were computed as ds divided by estimated mean cortical thickness. In previous work, one-half the outer diameter was used instead of ds in calculations of Z and BR.

Vertebral fracture assessment

At a follow-up visit, between 1997 and 1999, thoracolumbar radiographs of the spine were obtained. The follow-up radiographs were available for 3241 individuals, who survived an average of 7.4 years after baseline center visit and who were still able to come to our research center. All follow-up radiographs were scored for the presence of vertebral fracture by the McCloskey/Kanis method(19) as described earlier. (20)

Assessment of incident fracture

Follow-up started either on January 1 1991 or, when later, at the time of inclusion into the study. For this analysis, follow-up ended either at January 1, 2002 or, when earlier, at the participant's death. For ∼80% of the study population, medical events were reported through computerized general practitioner diagnosis registers. For the remaining 20%, research physicians collected data from the general practitioners' medical records of the study participants. All collected fractures were verified by reviewing discharge reports and letters from medical specialists. Fracture events were coded independently by two research physicians according to the International Classification of Diseases, 10th revision (ICD-10). Finally, an expert in osteoporosis reviewed all coded events for final classification. Any fracture was used as an outcome measure to have sufficient power. All fractures that were considered not osteoporotic (fractures caused by cancer and all hand, foot, skull, and face fractures) were excluded. In addition, we considered separately all fragility fractures that occur at older age, which included hip, proximal humerus, and pelvis fractures.

Genotyping

Genomic DNA was isolated from peripheral leukocytes by standard procedures.

Polymorphism-containing regions in the LRP5 gene were PCR amplified from genomic DNA. Each PCR was carried out in a 10-μl reaction volume containing 5 ng of genomic DNA, 1.5 mM magnesium chloride, 0.2 mM deoxy-NTP, 2 pmol of each primer, 0.2 units of Taq polymerase (Promega), and 10× PCR buffer (Promega).

The LRP5 polymorphisms Q89R, IVS5-4T/C, R353Q, V667M, IVS18-63C/T, and A1330V were genotyped using the single base extension (SBE) procedure using SBE primers of different lengths. The SBE reactions were performed according to details provided by the manufacturer (ABI Prism SNaPshot Multiplex Kit) with slight modifications. The genotypes thus generated were analyzed with software programs Gene Mapper 1.1 and Genotyper 3.7 (both from Applied Biosystems) and also checked by eye. In addition, in the large population, LRP5 A1330V and LRP6 I1062V were genotyped using the Taqman allelic discrimination assay. Primer and probe sequences were optimized using the single nucleotide polymorphism (SNP) assay-by-design service of Applied Biosystems. Reactions were performed on the Taqman Prism 7900HT 384-well format. All used primers and probes are available on request.

Linkage disequilibrium analysis

The linkage disequilibrium coefficient (D′) between markers was calculated using the program HAPLOXT. Haplotypes were reconstructed with the program PHASE. (21) Estimation of haplotypes was only reliable (>95% probability) when linkage disequilibrium between the polymorphisms was high (D′ > 0.8).

Functionality analysis of polymorphisms

To predict the possible functional consequences of the polymorphisms that changes the amino acid sequence, we used the software tool SIFT.

Statistical analysis of association

Subjects were grouped according to genotype. We grouped subjects by allele copy number (0, 1, and 2, corresponding to noncarrier, heterozygote carrier, and homozygote carrier, respectively) of the alleles.

We allowed for three possible genetic models to explain differences between groups (i.e., an allele dose effect, a dominant effect, or a recessive effect). Allele dose was defined as the number of copies of a certain allele in the genotype. In case of a consistent trend reflected as an allele-dose effect, we performed a linear regression analysis to quantify the association. In case of a dominant or a recessive effect of the test allele, analysis of (co)variance {AN(C)OVA} was performed to test for differences between two genotype groups. For dominant alleles, we compared test-allele carriers versus noncarriers, whereas for recessive effects, homozygous subjects for the test allele were compared with heterozygous carriers combined with noncarriers.

Hardy-Weinberg equilibrium (HWE) was calculated according to standard procedures using χ2 analysis. To estimate nonvertebral fracture risk, we used Cox proportional hazard models, thereby taking potential differences in follow-up time into account. To estimate the risk of vertebral fractures, ORs with 95% CIs were calculated using logistic regression models. All statistical analysis was performed using SPSS version 11.0 (SPSS, Chicago, IL, USA).

Population attributable risks were calculated by {P(RR − 1)/P(RR − 1) + 1) × 100, where P is the percentage population exposed, and RR is the relative risk.

RESULTS

LRP5 polymorphisms, allele frequencies, and linkage disequilibrium

We searched the literature, NCBI, and Celera databases for polymorphisms in the LRP5 gene, focusing on the coding region and flanking intronic regions. We selected six polymorphisms that either changed the amino acid sequence or were situated in an intron close to the intron-exon boundary; these are depicted in Fig. 1. Figure 1B shows the allele frequencies of the selected polymorphisms in 130 healthy white individuals. Two previously reported variants (Gln89Arg and Arg353Gln) were not polymorphic in our population (n = 130).

Figure FIG. 1..

Schematic overview of polymorphisms in the LRP5 gene. (A) The haplotype block structure taken from the HapMap project, Perlegen genome browser, and ABI SNP-browser. (B) The position of all the polymorphisms examined in previous studies and the seven polymorphisms analyzed in this study and their allele frequencies in 130 white individuals are depicted. Boxed polymorphisms were reported to be associated with BMD. (C) The four haplotypes observed in our population are depicted with the allele frequencies in 130 white blood bank donors. (D) A graphical overview of pairwise disequilibrium coefficients (D′) between the studied polymorphisms is shown. (E) The position of the LRP6 polymorphism examined.20

To study the pattern of linkage disequilibrium between LRP5 polymorphisms examined in this study, pairwise disequilibrium measures (D′) were calculated in 130 healthy whites (Fig. 1D). Strong linkage was observed between the polymorphisms Val667Met, C-IVS18-63T, and Ala1330Val, but linkage was low between T-IVS5-4C and the other SNPs. We constructed haplotypes for the three polymorphisms in strong linkage disequilibrium using the software program PHASE. Only four of the possible eight haplotypes were present in our white population, the haplotype frequencies of which are shown in Fig. 1C.

Figure 1A also shows the high linkage blocks in the LRP5 gene in white individuals according to the hapmap (http://www.hapmap.org) project, Perlegen (http://genome.perlegen.com/browser/index.html), and according to the SNP browser (http://www.appliedbiosystems.com) from ABI). In addition, polymorphisms described in previous reports on association studies of LRP5 with BMD are depicted in Fig. 1A. (22–26) This summary of the literature indicates that most of the studies have examined association of the 3′ linkage block with BMD, although the exact polymorphisms analyzed across the studies are different.

For polymorphisms in the LRP6 gene, we searched databases and found an amino acid variant at position 1062, which changes an isoleucine to a valine (Fig. 1E). Predictive analysis by SIFT software estimated that this change is not tolerated well, which suggests a deleterious effect of the 1062-Val allele.

LRP5 haplotypes and association

For a random selection of 2293 subjects (980 men and 1213 women), we genotyped the four SNPs that were polymorphic in our population. In addition we inferred haplotypes over the SNPs that were in high linkage disequilibrium (Val677Met, C-IVS18-63-T, and Ala1330Val) and performed an association study with the separate SNPs and the four haplotype alleles that were found in the population (Table 1). We observed the Val allele of the Ala1330Val variant to be significantly associated with lumbar spine BMD in men. No association was seen with the other single SNPs. Both haplotype alleles that contained the Val allele of the Ala1330Val polymorphism (haplotypes B and D; Fig. 1), showed similar trends toward lower BMD.

Table Table 1.. Difference in Mean Height and Lumbar Spine BMD Between Men Carrying the Test LRP5 Allele and Men Not Carrying the Test Allele*
original image

LRP5 and bone characteristics

We analyzed bone characteristics by the LRP5 Ala1330Val genotype in the complete study population. When stratified by genotype for Ala1330Val, we observed in men a mean age that was 0.7 years higher for every copy of the Val allele (Table 2), but in women, age did not differ by genotype. Bone characteristics by genotypes are shown in Table 2. The 1330Val allele showed a significant association with decreased lumbar spine BMD with evidence for an allele-dose effect. BMD decreased with 0.13 SD (p = 0.001) and 0.10 SD (p = 0.01) per copy of the Val allele for men and women, respectively. The 1330Val allele was associated with femoral neck BMD in men only, with a difference of 0.2 SD (p = 0.02) between extreme genotypes. Adjustment of the association between Ala1330Val and BMD by height and weight did not change the results substantially. No association was seen between bone loss and the genotype (data not shown).

Table Table 2.. Baseline Anthropometrical and Bone Characteristics According to LRP5 A1330V Genotype
original image

The 1330Val allele was also associated with bone geometry measures. In men, the 1330Val allele was associated with decreased lumbar spine bone area (0.17 SD difference between extremes; p = 0.02) and decreased femoral neck width (0.15 SD between extremes; p = 0.06), whereas a nonsignificant trend toward decreased height was seen. In addition, the Val allele was also associated with a decreased section modulus (a measure for bone strength; p = 0.03; results not shown). No relation between the Ala1330Val genotype and buckling ratio (a measure for bone instability) was seen. Adjustment by height diminished the associations of the 1330Val allele with the bone geometry measures. In women, no association was seen between bone geometry measures and the Ala1330Val genotype.

We examined whether the associations between 1330Val and bone characteristics was age-dependent by stratifying the population in 10-year categories. No age dependency of the association was found (results not shown).

LRP5 polymorphism and fracture risk

Table 3 shows the number of fractures and fracture risks according to 1330Val carrier status. No associations were seen in women. We found no association of the 1330Val allele with risk for prevalent vertebral fractures, but a modest but nonsignificant increased risk for incident osteoporotic fracture in men. We did additional separate analyses for different types of nonvertebral fracture, such as hip, upper humerus, and wrist, and we found that in men the Val allele carriers had a higher risk for the “fragility” fractures occurring at advanced ages (hip, proximal humerus, and pelvis fractures). The separate relative risks are 1.6 (95% CI, 0.9-2.6) for hip fracture and 1.6 (95% CI, 0.7-3.7) for upper humerus fractures; when all type of fragility fracture fractures were combined, the risk was 1.6 (95% CI, 1.0-2.6; p = 0.04). Adjustment for age, height, weight, BMD, femoral neck width, or section modulus did not change these risks substantially.

Table Table 3.. Fracture Risk According to Ala1330Val Genotype
original image

Interaction between LRP5 and LRP6 polymorphisms

We analyzed anthropometric and bone characteristics by the LRP6 Ilel062Val genotype in the complete cohort stratified by gender (Table 4). In men, mean age was 0.7 years higher for every copy of the Val allele. In addition, we found a borderline significant association with height (p = 0.04 for carriers of the Val allele versus noncarriers) and lumbar spine bone area. We went on to analyze fracture risk according to the Ile1062Val genotype (Table 5) and found in men that the Val allele carriers had a 60% higher risk for fragility fractures. In addition, we found a nonsignificant trend toward a higher risk for vertebral fractures. Adjustment of the risk estimates by age, height, weight, and BMD did not change the risk estimates. In women, no association was found.

Table Table 4.. Baseline Anthropometrical and Bone Characteristics According to LRP6 I1330V Genotype
original image
Table Table 5.. Fracture Risk According to LRP6 Ile1062Val Genotype
original image

We analyzed the interaction between the polymorphisms in LRP5 and LRP6. We stratified the subjects according to carriership of the LRP5 1330Val and LRP6 1062Val alleles into four groups: a reference group not carrying a risk allele (n = 1179 men and n = 1819 women), a group only carrying the LRP5 Val allele (n = 435 men and n = 669 women), a group only carrying the LRP6 Val allele (n = 708 men and n = 930 women), and a group carrying both the LRP5 and LRP6 risk alleles (n = 260 men and n = 342 women). We observed that men carrying both risk alleles were at higher fracture risk compared with the reference group or compared with men carrying only a single risk allele. This was seen for both the fragility fractures and for the vertebral fractures (p = 0.004 and p = 0.02, respectively, for the double carriers against the reference group (Figs. 2A and 2B). Male double risk allele carriers had a 90% higher risk vertebral fracture (p = 0.008) against all other individuals. Women that carried both risk alleles showed a 30% higher risk for both fragility fractures and vertebral fractures compared with all other women; however, this trend failed to reach significance. Adjustment of the risk estimates by age, height, weight, and BMD did not change the results. The population attributable risk percent for carrying both the LRP5 and LRP6 risk alleles was 10% in the male population for both fragility fractures and vertebral fractures.

Figure FIG. 2..

(A) Fragility fracture risks and (B) vertebral fracture risks with 95% confidence limits, according to the combined genotype of LRP5 A1330V and LRP6 I1062V.20

Conservation analysis

To examine whether the LRP5 and LRP6 amino acid variants might have consequences for protein function, conservation was studied. Figure 3A shows the conservation across species of the second LDL receptor-like binding domain of LRP5 in which the Ala1330Val polymorphism is situated. The 1330Ala shows considerable conservation at the same position in LRP5 across different species and is situated next to three highly conserved amino acids in this domain. The Ile1062Val polymorphism lies within a domain that is predicted to form a structure resembling a propeller with six blades. The residue lies within the third blade of the fourth propeller. The Ile1062 shows high evolutionary conservation (Fig. 3B). Isoleucine is found at the same position in the third blade of the first, second, and fourth propeller of LRP6 and LRP5 in human, chicken, mice, and xenopus. In addition, isoleucine is present at this position in a wide range of other proteins that also include a β-propeller structure including other LDL receptor-related proteins, epidermal growth factor precursor, and very-low-density lipoprotein receptor in a number of species.

Figure FIG. 3..

(A) The conservation of the second LDL receptor-like ligand binding domain (L2) of LRP5 across species (h, human; m, mice; rb, rabbit; rt, rat; x, xenopus; ch, chicken; darr, LRP5 homolog in drosophila melanogaster) and other LDL receptor-like binding domains in the LRP5 (L1 and L3). The 1330-valine is situated next to three highly conserved amino acids. (B) Conservation of the 1062-isoleucine in the propeller structures. A partial amino acid sequence of the third blade (B3) of a number of propellers of the YWTD family is shown. Isoleucine is highly conserved at this position in the first two (P1 and P2) and fourth propeller (P4) of LRP5 and 6 in human (h), mice (m), and xenopus (x), as well as the fourth propeller of the drosophila melanogaster (d) LRP6 and LRP5 homolog Arrow (arr).20

DISCUSSION

This study shows an association between the valine allele of the Ala1330Val polymorphism in the LRP5 gene and the valine allele of the Ile1062Val polymorphism in the LRP6 gene with increased fracture risk in men. Men that carried both risk alleles had a 2.4 and 1.9 times higher risk for a fragility and vertebral fracture, respectively, compared with men not carrying a risk allele. The attributable fraction for double risk allele carriers in male subjects in this study is 10%, suggesting that the LRP51330Val allele and the LRP6 1062Val allele account for one-tenth of the fracture cases in men.

We showed that the LRP5 1330Val allele was associated with decreased BMD, decreased bone size, and decreased height in a large population-based cohort of elderly men and women. These results are in line with recently published studies that also found evidence for association of genetic variation of the LRP5 gene with BMD. In a relatively small case control study of Japanese women, a synonymous SNP in exon 10 was found to be associated with BMD. (22) In addition, a study using family-based and case control approaches reported association of the 1330Val allele with low BMD(23, 26) and showed association of the Val677Met SNP in exon 9 with BMD and height in three relatively small cohorts of children, adolescents, and (premenopausal) adults. To our knowledge, ours is the first study that examined osteoporotic fracture risk (as well as BMD) according to the LRP5 genotype in a large population-based and homogeneous cohort of elderly white men and women.

We observed the associations between the LRP5 and LRP6 variants and bone parameters to be stronger and more consistent in men compared with women. This finding is in line with the findings of a previous study(23) that also reported the association between the LRP5 polymorphism and bone parameters to be stronger in men compared with women, whereas these individuals were considerably younger (children and premenopausal adults) than our study population (postmenopausal adults). This suggests that the differences between Ala1330Val genotype groups in frame size and BMD are achieved early in life. This is further corroborated by the fact that the polymorphism was not associated with bone loss in our study. Until now, we did not have a good explanation for the gender difference. It is known that differences in bone width and stature are established during puberty, so we hypothesized that the differences in sex-specific hormones (such as androgens and estrogens) during puberty might be involved in the regulation of LRP5: therefore, gender differences can occur. Interestingly, a recent report indicate a functional interaction between the estrogen receptor and the Wnt-signaling system. (27)

The associations we observe seemed real, given the size of the population and their consistency (effects at several bone parameters including BMD, geometry, and fractures). We acknowledge the potentially difficulty with multiple comparisons in this study, and we cannot exclude the possibility that some of our significant associations are false positive. When p values are adjusted for multiple testing (using Bonferroni correction), the threshold of significance is 0.01 (0.05/4 = 0.013; considering that we tested two independent SNPs in the complete population for two major outcomes: BMD and fracture). However, in genetic association studies, simply adjusting for the number of independent tests is now considered to be an overconservative adjustment to control for false positives, and alternative methods are becoming available. (28) In particular, it is difficult to define what is an independent test, considering that SNPs are not always independent (in linkage disequilibrium), and clinical outcomes might be related (such as BMD at different sites and BMD and fracture). An alternative approach to conclude an association is “real” is to search for consistency in the pattern of associations and to seek replication in similar cohorts. Finally, a careful meta-analysis will help to assess heterogeneity and true effect size of these polymorphisms. (29)

Strong linkage disequilibrium was observed between polymorphisms in exon 9, intron 17, and exon 18. Construction of haplotypes across the linked polymorphisms showed that the Ala1330Val SNP is probably driving the association found, because both haplotypes containing the 1330Val variant showed an association with low BMD and other bone-related endpoints, whereas the other haplotypes did not. Using data from the HapMap project, we showed that, although not every study examined the same polymorphisms, most of the previously associated LRP5 polymorphisms lie within the same haplotype block. Together these data suggest a very consistent association between this LRP5 haplotype block and BMD. In particular, our study now suggests that the A1330V polymorphism is the driving force behind this association, but final proof will have to come from a large meta-analysis and functional analysis of the polymorphisms.

LRP5 is thought to be important for the establishment of peak bone mass. (9) Because peak bone mass is reached around age 25, our population of relatively old subjects (55 years and older), might not be optimal to examine the relationship between LRP5 and LRP6 polymorphisms and variation in BMD. Age-related bone loss strongly influences BMD at older ages, so our study might underestimate the effect of LRP5 on peak bone mass. On the other hand, osteoporosis, and the clinically important osteoporotic fracture, is occurring most frequently in the older population, which is the reason why we examined the role of LRP5 polymorphisms in this older population.

Genetic association studies can be influenced by population heterogeneity. In this cohort study, all subjects were Dutch whites; to our knowledge, no systematic differences were present with respect to the part of The Netherlands in which this study was performed. We also observed HWE for the polymorphism analyzed, and therefore, our study population might be considered as an ethnically homogeneous and representative sample of the Dutch subjects.

Concerning functionality of the polymorphisms, we note that the A1330V polymorphism lies within a second LDL repeat of LRP5. Although the function of this region in LRP5 is not clear, similar domains in the LDL receptor are involved in binding the ligand to the receptor. (30) Therefore, it is conceivable that variations in the LDL repeats, such as Ala1330Val, alter the function and signaling of the protein. The alanine to valine amino acid conversion is considered chemically as a relatively mild change. However, such a subtle alteration in the MTHFR enzyme has been reported to have functional consequences, (31) as well as for identical mutations (alanine to valine) in the recently identified thyroid hormone transporter MTC8. (32) Therefore, functionality of this variation in the LRP5 gene is conceivable. Indeed, conservation studies of the 1330Ala amino acid showed a mild conservation of this residue at this position in the protein across species and is situated adjacent to a number of highly conserved amino acids in this domain, which indicates that it might be functionally important. Nevertheless, direct consequences for ligand binding and LRP5 activity is unclear, and further research will have to elucidate whether this is true. The Ile1062Val polymorphism in LRP6 lies within the fourth YWTD β-propeller domain, which is located in the dickkopf-binding region. (33) Alignment of close homologs of LRP6 showed a strong conservation of the 1062 isoleucine at this specific position. The evolutionary conservation of the isoleucine is strong evidence for a functional importance of this residue. Of course, true functional effects of both of the polymorphisms will have to be examined by doing in vitro testing on Wnt-signaling efficacy.

In conclusion, we have shown that both the LRP5Ala1330Val and the LRP6 Ile1062Val variant are associated to several bone parameters, and more importantly, to fracture risk. All of the associations found were stronger and more consistent in men compared with women.

Acknowledgements

We thank Dr M Warman for sharing sequence information of the LRP5 gene in an early stage of this project and Dr Thomas J Beck for providing us with the hip structure analysis of DXA data. This project was funded by European Commission Grants QLK6-2002-00491 (NEMO) and QLK6-CT-2002-02629 (GENOMOS).

Ancillary