website tracking


Genetics of Osteoporosis


Osteoporosis is a polygenic disease where multiple gene variants each having a small effect contribute to an individual’s increased susceptibility to the disease, although a major gene might also be involved (Cardon et al., 2000). Until now a number of studies have been performed in various populations to identify genetic variants that might be responsible for an increased risk of osteoporosis. Different approaches such as case-control association and family based linkage have been used to identify any association between phenotype and gene variants. The definition of the phenotype most commonly used in these studies was a continuous variable such as BMD and discrete phenotypes such as fractures. From these studies a number of chromosomal loci and SNPs found within candidate genes have been associated with BMD and/or fracture risk, but no major gene was identified so far (Huang et al., 2003; Liu et al., 2003; Peacock et al., 2002; Zmuda et al., 2006). Also animal models were occasionally used to perform linkage studies where in these models it is possible to control for heterogeneity and environmental factors (Klein et al., 2001). Information from these studies is useful for the identification of candidate genes found in syntenic loci in humans.


Linkage Studies       

Family based linkage studies were performed by a number of investigators to try to identify loci that might contain genes responsible for an increased susceptibility for osteoporosis. A number of chromosomal loci have been identified and confirmed to a quantitative trait locus (QTL) as BMD by genome-wide scans and by scanning candidate regions. Using microsatellite markers at loci where known genes are found, Duncan and co-workers observed strong evidence of linkage to the parathyroid hormone receptor type 1 (PTHR1) on chromosome 3 (Duncan et al., 1999). In this study, moderate evidence of linkage was also reported at loci where other candidate genes are found including the COL1A1, COL2A1/VDR, IL-6 and oestrogen receptor 1 (ESR1). A major limitation of this study was that it did not differentiate between the effects of genes that affect peak bone mass and those affecting bone loss. Candidate gene approaches are based upon current knowledge of physiology and therefore other genes found at distinct loci can easily be missed.

A genome-wide scan performed in two independent cohorts with similar ethnic backgrounds also observed evidence of linkage at locus 3p21 to lumbar spine (LS) BMD (Wilson et al., 2003), which agrees with the results of Duncan et al (1999). In this study, Wilson et al (2003) also confirmed linkage to locus 1p36 and showed evidence of linkage to other loci including 11p, 2q and 19q. Locus 1p36 was first associated with a low hip BMD when performing non-parametric linkage analysis in seven extended pedigrees (Devoto et al., 1998). A higher resolution scan using nine microsatellite markers at the candidate region was performed in an extended sample of forty two families. Results from this study supported the previous ones that a major QTL controlling femoral neck (FN) BMD was on chromosome 1p36 (Devoto et al., 2001). At this locus a number of genes involved in bone physiology are known to be found including those coding for tumour necrosis factor alpha receptor-2 (TNFR2) and lysyl hydroxylase (PLOD1). Positive associations of SNPs within these genes with LS and FN BMD further strengthen the importance of region 1p36 (Spotila et al., 2003). These results were replicated in an independent study carried out in the Amish population, which is a relatively genetically homogeneous population due to the closed lifestyle that they live because of cultural and religious reasons. A LOD of 2.02 for femoral neck BMD to locus 1p36 was obtained in a group of 593 women (Streeten et al., 2006). In this study only suggestive linkage to a number of other chromosomal loci including 3q26, 7q31, 21q22, 12q24, 11q22, 14q23 and 1q21, was observed.

Locus 11q12-13 was also confirmed several times by a number of investigators. In their first study, Koller and colleagues performed a linkage scan at this locus in pre-menopausal African Americans and Caucasians where linkage to FN BMD was confirmed showing the importance of this locus in peak bone mass (Koller et al., 1998). Two years later they performed a genome scan where they reconfirmed this locus to variations in FN BMD and reported linkage of LS and FN BMD to regions 1q21-23, 5q33-35 and 6p11-12 (Koller et al., 2000). Locus 11q12-13 is of particular interest since it was also linked to other bone diseases including autosomal recessive osteopetrosis (Heaney et al., 1998), high bone mass phenotype (Johnson et al., 1997) and osteoporosis pseudoglioma syndrome (Gong et al., 1996). Two interesting genes found in this locus are the T-cell immune regulator 1 (TCIRG1) and the lipoprotein receptor related protein-5 (LRP5), where mutations in the latter were found to be responsible for high bone mass (Boyden et al., 2002) and osteoporosis-pseudoglioma syndrome (Ai et al., 2005).

A number of loci were confirmed by linkage to BMD at different anatomical sites (Kammerer et al., 2003; Wynne et al., 2003), variation in femoral structure (Koller et al., 2001) and peak bone mass reached early in life (Econs et al., 2004).  From this study strong evidence of linkage was reported to chromosome 1q (LOD 4.3) when they increased the number of markers genotyped at this locus in 938 sister pairs. Their results complemented other results obtained from studies carried out in mice where linkage to peak BMD was reported in syntenic regions of the mouse genome (Klein et al., 2001).

Linkage and association were confirmed to chromosome 20p12 in a study carried out in the Icelandic population (Styrkarsdottir et al., 2003). In this study, discrete phenotypes were defined by using z-scores corrected for age and combined with fractures at different degrees of severity. Following the initial scan, linkage was observed at chromosome 20p although other smaller peaks were reported at other chromosomal loci. Linkage disequilibrium mapping was performed at this region using SNPs and additional families followed by sequencing of the gene coding for bone morphogenic protein-2 (BMP2) found in this region. A missense polymorphism and two SNP haplotypes in this gene were associated with osteoporosis in the Icelandic population and confirmed in the Danish population. In a study carried out in a population of European origin, suggestive linkage was observed to this region on chromosome 20 for both spine and hip BMD (Shen et al., 2004). Contrasting these observations, single nucleotide polymorphisms within the BMP-2 gene did not have any effect on BMD in the Dutch population (Medici et al., 2006). In the same study, evidence of linkage was observed to another region on chromosome 11 (11q23) and to Xq27, besides suggestive linkage several other loci.

In a linkage study carried out in 3658 subjects from eight European countries (the FAMOS study) a number of chromosomal loci were linked with BMD in a gender- specific, site-specific and age-specific manner (Ralston et al., 2005). When the analysis was performed on the whole group of participants no evidence or suggestive linkage was observed. Evidence of linkage was observed when analysis of different sub-groups was done according to age, sex and anatomical site. In men, evidence of linkage was observed between FN-BMD and chromosome 10q21, while in women suggestive linkage to LS-BMD was observed to loci 18p11 and 20q13. Five other QTLs were identified to other chromosomal loci.

            From a meta-analysis of nine whole-genome scans, a significant LS-BMD QTL was found on chromosome 1p13.3-q23 while other regions were also indicated to a lesser degree. These loci were on chromosomes 12q24.31, 3p25.3-p22.1, 11p12-q13.3, 18p11-q12.3 and 1q32-q42.3 for LS-BMD. For FN-BMD the strongest QTL was observed to chromosome 9q31.1-q33.3 (Ioannidis et al., 2007). In this study a genome-wide significance was not reached for any of the indicated loci.

Association Studies

During the last decade the number of association studies performed to identify genes responsible for osteoporosis increased dramatically. Different investigators studied large numbers of SNPs within a series of genes most of which are known to be involved in bone physiology or else linked to other bone diseases. Conflicting results were obtained from these studies. Candidate genes studied are those primarily involved in the various biological processes of bone physiology including those coding for various receptors (Zmuda et al., 2000; Ioannidis et al., 2002; Urano et al., 2004), cytokines (Chung et al., 2003), growth factors (Langdahl et al., 2003; Kim et al., 2003; Styrkarsdottir et al., 2003), structural proteins (Grant et al., 1996) and other bone diseases (Uitterlinden et al., 2004), together with other genes not directly involved in bone biology (Jorgenson et al., 2002; Spotila et al., 2003). The influence of gene variants might also be influenced by other variants within the same gene or by other gene-gene interactions as well as interactions with the environment and epigenetic effects such as DNA methylation (Friso et al., 2002). An overview of the major genes associated with osteoporosis or fracture risk that were studied in this thesis are given below.


The Vitamin D Receptor (VDR) Gene

The VDR is one of the nuclear receptor superfamily (Mangelsdorf et al., 1995) that specifically binds to 1,25-dihydroxy vitamin D in target tissues, to stimulate intestinal calcium and phosphate absorption, increases bone resorption and renal calcium/phosphate reabsorption. Vitamin D directly increases gene expression of receptor activator of NF-κβ ligand (RANKL) in osteoblasts and indirectly increases differentiation and activation of osteoclasts by cell to cell interactions (Khosla, 2001). It was also shown that the VDR- retinoid X receptor (RXR) heterodimer directly binds to functionally vitamin D responsive elements (VDRE) found in the promoter region of RANKL thus increasing gene expression (Kitazawa and Kitazawa, 2002).   The end result is an overall increase in calcium/phosphate ion concentration in the blood (Haussler et al., 1998). The functional domains within this receptor are those primarily involved in hormonal ligand binding, heterodimerization to RXR, DNA binding and activation of transcription. It was confirmed that a tryptophan residue at position 286, is essential for ligand binding and mutations at this position might affect heterodimerization and interactions with other proteins that eventually affect transcription (Solomon et al., 2001).

The VDR gene was localised to human chromosome 12cen-q12 by fluorescent in situ hybridization and consists of fourteen exons spanning a region of 75kb (Taymans et al., 1999; Miyamoto et al., 1997). The functional protein product is encoded by exons 2 – 9 while another 6 untranslated isoforms of exon 1 (1a – 1f) are alternatively spliced in a tissue specific manner under the control of multiple promoters (Crofts et al., 1998). The promoter region of the VDR gene does not contain a TATA box but contains a GC rich region with five binding motifs for the transcriptional factor Sp1 (Miyamoto et al., 1997). 

    A number of allelic variants found within the VDR gene have been described and associated with an increased risk of osteoporosis and other diseases (Zmuda et al., 2000). In 1994, Morrison and co-workers concluded that common variants found within the VDR gene were responsible for up to 75% of the genetic effect on BMD (Morrison et al., 1994). Genotyping errors were reported three years later but still this study increased the interest in genetic studies of osteoporosis (Morrison et al., 1997). Following this study, during the last decade there was a dramatic increase in the number of association studies to try and identify other genetic variants responsible for the variability in bone density and their role in bone biology (Uitterlinden et al., 2002).

Two years later, Gross and colleagues studied a novel polymorphism at the translation initiation site in exon 2 of the VDR gene, which can be detected by RFLP using endonuclease FokI (Gross et al., 1996). Gross et al (1996) reported that Mexican-American women having the TT genotype had 12.8% lower spinal BMD when compared to the wild-type CC genotype. Jurutka and co-workers described how the wild-type isoform of the VDR, although it is three amino acids shorter, interacts more efficiently with transcriptional factor IIB (TFIIB) and possessed elevated transcriptional activity (Jurutka et al., 2000). In a study carried out in the geographically isolated population of Lampedusa, this SNP was significantly associated with low BMD with allele frequencies possibly indicating founder effects in this population (Falchetti et al., 2007).

Other polymorphisms identified were those analysed by endonucleases BsmI, ApaI and TaqI at the 3` end of the gene (Morrison et al., 1994), others associated with diseases such as rickets (Malloy et al., 1990) and a novel SNP detected by TruI (Kajickova et al., 2003).  Another variant is a mononucleotide repeat (A)n polymorphism that varies in length from 13 to 24 adenosines (12 alleles) poly (A) occurs in the 3’ untranslated region of the VDR gene, possibly affecting RNA stability (Grundberg et al., 2003).  In this group of Swedish women, this variant was found to be in linkage disequilibrium with the BsmI polymorphism further upstream in this gene, and was found to affect both LS and FN BMD (Grundberg et al., 2003). Another interesting polymorphism is a G to A transition found within the caudal related homeobox (Cdx)-2 binding site in the promoter region of the VDR gene, which modulates intestinal-specific transcription (Arai et al., 2001; Yamamoto et al., 1999).

Conflicting results were obtained from association studies of these SNPs performed in pre and postmenopausal women (Cheng & Tsai, 1999; Vidal et al., 2003; Zajickova et al., 2002; Douroudis et al., 2003) as well as in men (Pottelbergh et al., 2002) and adolescents (Strandberg et al., 2003), of different ethnic backgrounds. VDR polymorphisms were also associated with bone turnover, fracture risk (Langdahl et al., 2000; Moffett et al., 2007) and with other parameters such as height and bone size in children and young adults (Van der Sluis, 2003). From a meta-analysis performed by Thakkinstian and co-workers concluded that the B allele of the VDR BsmI polymorphism followed a recessive model where the BB genotype was associated with a lower BMD (Thakkinstian et al., 2004). This meta-analysis was performed using thirty nine association studies performed on pre and postmenopausal women. Thakkinstian et al (2004) emphasised the importance of performing this kind of analysis to identify methodological problems in molecular studies such as heterogeneity, effects of confounders such as calcium and vitamin D intake, as well as interactions of the gene involved with other genes and the environment. Confounding factors such as calcium intake were observed to have an effect on the influence of various polymorphisms on BMD, including that of the B allele for the BsmI polymorphism, which was correlated with a low BMD only in the presence of low calcium intake (MacDonald et al., 2006). The same observations were made for the G allele (with low BMD) of the Cdx-2 polymorphism in the promoter region.

Large scale association studies such as the GENOMOS study failed to find a significant association between VDR polymorphisms and BMD, showing that the effects of these SNPs on BMD are minimal (Uitterlinden et al., 2006). In this study 26,242 participants from all over Europe were included and genotyped using various techniques which were cross-validated between the different members of the consortium. Meta-analysis was performed and adjustments of various confounding factors including age, height, weight, use of hormone replacement therapy and menopausal status were done. Besides BMD, these SNPs were tested for correlation with increased fracture risk, where only the Cdx-2 polymorphism was found to have a very mild effect and thus increasing the risk of fractures. The protective role of the A allele for fractures was also reported earlier especially in older women, in a smaller study of almost 3,000 white men and women (Fang et al., 2003). Conversely, the conclusion that SNPs at the 3`end of the VDR gene does not have any effect on BMD or fracture risk contradicts earlier findings from a study carried out on 6,148 samples from Rotterdam (Fang et al., 2005). In this study VDR haplotypes at the 5` promoter region and 3`UTR were studied for an association with fracture risk and functional studies were done to understand the effects of these variants at the molecular level. A modest effect of 15% - 48% of VDR polymorphisms on fracture risk was observed corresponding to a 15% difference in VDR mRNA observed between the genotypes. Also a strong effect of one subtype of the BsmI-ApaI-TaqI haplotype was observed on fracture risk.  On the other hand, another meta-analysis this time on published data failed to detect any association of the BsmI and TaqI polymorphisms and increased risk to fracture (Fang et al., 2006).

In another study, Uitterlinden and colleagues observed that interactions between VDR and COL1A1 polymorphisms might predict a risk for osteoporotic fractures that is independent of BMD (Uitterlinden et al., 2001). This highlights the importance of genetic markers as predictors of fracture risk that cannot be predicted by measuring BMD alone. Similar inter-locus interactions that increased the risk to fractures were later reported, this time between ER1 and VDR polymorphisms (Colin et al., 2003). Gene to gene interactions between ER1 and VDR were also reported to affect BMD in a study carried out in prepubertal females (Willing et al., 2003). VDR and COL1A1 gene variants were observed to play a significant role in the determination of BMD in those subjects having a low calcium intake, showing that environmental factors can modify the effect of these genes on the phenotype (Brown et al., 2001).



Oestrogen Receptor Gene-α (ER1)

 The oestrogen receptor (ER) is another member of the nuclear receptor superfamily that consists of several domains that are important for hormone binding and initiation of transcription. There are two isoforms of oestrogen receptors known as α (1) and β (2) that are expressed in different tissues and are encoded by two distinct genes (Gustafsson, 1999). When analysing developing human bone for expression of these two ERs, it was observed that ER1 was predominantly expressed in cortical bone while ER2 showed higher levels of expression in cancellous bone (Bord et al., 2001). 

The gene encoding human ER1 is located on chromosome 6q25.1 and spans a region of more than 140kb of the genome. The gene is made up of eight exons with its introns showing a high degree of conservation when compared with the chicken’s progesterone receptor (Ponglikitmongkol et al., 1988). The hormone binding domain is encoded by an assembly of five exons. When performing experiments on mice, Lindberg and co-workers reported that the expression of many important genes involved in the regulation of trabecular BMD such as TGF-β are controlled by oestrogen via ER1 and not ER2 (Lindberg et al., 2002). The oestrogen system and the ER gene are interesting candidates for genetic studies of diseases that are triggered by the onset of menopause including osteoporosis, osteoarthritis, Alzheimer’s and coronary artery disease (Massart et al., 2001).

Until now a number of polymorphisms have been identified in the ER1 gene and studied for an association with BMD and/or fracture risk. Two most commonly studied SNPs are caused by transitions, C – T (PvuII) and A – G (XbaI), found within the first intron only 50bp apart. These two SNPs were first studied by Kobayashi and colleagues in Japanese postmenopausal women where they observed that the Px haplotype increased the risk of a low BMD (Kobayashi et al., 1996). Following this study a number of investigators carried out similar studies and reported positive correlation with BMD, fracture risk and rate of bone loss in postmenopausal women (Kobayashi et al., 2002; Albagha et al., 2005) and adolescent boys (Lorentzon et al., 1999). From a meta-analysis Ioannidis and colleagues concluded that the XX genotype played a protective role on BMD and risk of fractures when compared to the x allele (Ioannidis et al., 2002). Other investigators also studied these SNPs in relation to BMI (Deng et al., 2000), menopausal symptoms (Malacara et al., 2004) and as markers to predict the onset of natural and surgical menopause (Weel et al., 1999). Strong linkage disequilibrium between PvuII and XbaI SNPs was always reported. An association was reported in Chinese between these two SNPs in the first intron and variations in height but not with bone size (Lei et al, 2005).

A TA repeat polymorphism found within the promoter region of the ER1 gene was also associated with low BMD and an increased risk of fractures by two independent investigators (Becherini et al., 2000; Langdahl et al., 2000). Both studies were concordant where individuals with low numbers of TA repeats had lower BMD and an increased risk of fracture. Langdahl et al (2000) suggested that since this polymorphism was previously associated with familial premature ovarian failure this might explain why these individuals have a low BMD. Other explanations might be that this polymorphism is in LD with another functional polymorphism or else it might directly affect gene expression. In the same study, Langdahl and co-workers also analysed a G261-C polymorphism located in the first exon where no correlation was found. The TA repeat was also associated with responsiveness to HRT treatment in Korean postmenopausal women (Yim et al., 2005). 

A G2014A SNP located in exon 8, six nucleotides upstream from the stop codon was also associated with postmenopausal osteoporosis in a study carried out in Thailand. However these results have yet to be confirmed by other investigators (Ongphiphadhanakul et al., 2001).

Interactions of ER1 polymorphisms with other SNPs found in other genes and their influence on BMD were also reported by a number of investigators. Such associations were reported between haplotypes in the ER1 gene with COL1A1 (Bustamante et al., 2007), oestrogen receptor-β (Rivadeneira et al., 2006) and the aromatase gene (Riancho et al., 2006), among others (Colin et al., 2003).  A meta-analysis showed that XbaI variant might have an effect on fractures risk, with XX genotyping playing a protective role and in a way that is independent of BMD (Ioannidis et al., 2004).



The TNFRSF11B (OPG) Gene

Figure 1. Structure of TNFRSF11B gene (OPG)

Osteoprotegerin (OPG), a secreted glycoprotein of the tumour necrosis factor (TNF) receptor superfamily, is made up of 401 amino acids with four cysteine rich domains and is known to play a very important role in the regulation of bone mass (Simonet et al., 1997). Transgenic mice expressing high levels of OPG were observed to suffer from severe osteopetrosis due to a decrease in osteoclast differentiation (Simonet et al., 1997) while knock out mice were observed to have severe osteoporosis and an increased risk of fractures (Bucay et al., 1998). OPG controls osteoclastogenesis by binding to RANKL on preosteoblast/stromal cells preventing cell to cell interactions with RANK present on osteoclast precursors, thus preventing osteoclast differentiation, activation and promoting apoptosis (Hofbauer et al., 2000). Macrophage colony stimulating factor (M-CSF) binding to its receptor, c-Fms, on osteoclast precursors is also essential for activation. Administration of exogenous (recombinant) OPG in postmenopausal women was found to be safe, well tolerated and had a very positive effect as an antiresorptive agent (Hofbauer & Heufelder, 2000).

The human TNFRSF11B gene is a single copy gene found on chromosome 8q23-24 which consists of five exons and spans a region of 29 kb (Morinaga et al., 1998). Exons 4 and 5 encode for two death domains that are present in tandem and from their amino acid sequence it is evident that exon 4 is produced by duplication of a part of exon 5.  It is known that the expression of OPG and RANKL is controlled by various cytokines, hormones and growth factors such as transforming growth factor (TGF)-b (Thirunvakkarasu et al., 2001), insulin-like growth factor (IGF)-I (Rubin et al., 2002), BMP (Wan et al., 2001), oestrogen (Bord et al., 2003) and Cbfa1 (Thirunvakkarasu et al., 2000), and it is the ratio of OPG/RANKL that determines the pool size of active osteoclasts (Hofbauer et al., 2000). Experiments on mice also showed that higher expression of RANKL on osteoblasts mediates enhanced osteoclastogenesis highlighting the importance of osteoblasts in the regulation of bone turnover (Kiviranta et al., 2005).

A number of SNPs within the TNFRSF11B gene were analysed by a few investigators for any association with BMD and/or fracture risk. Variations within this gene were also positively associated with other human diseases such as Paget’s disease of bone (Wuyts et al., 2001; Daroszewska et al., 2004), idiopathic hyperphosphatasia (Cundy et al., 2002) and vascular disease (Hofbauer & Schoppet, 2002; Soufi et al., 2004). Twelve polymorphisms were identified in the TNFRSF11B gene, two of which were observed to be more frequently found in patients with vertebral fractures (Langdahl et al., 2002). The G allele of the T245-G polymorphism located in the promoter region was also associated with a low BMD in Slovenian postmenopausal women (Arko et al., 2002) and Japanese women but not in men (Yamada et al., 2003). These results were concordant with those reported by Langdahl et al (2002). In the same study, an association with the rare allele of the A163-G polymorphism and increased fracture risk was reported, which was later confirmed by Jorgensen et al (2004) in Danish women.

Ohmori and colleagues, in a study carried out in Japanese families and postmenopausal women did not find any evidence of linkage with the osteoprotegerin gene but reported an association of the T950-C polymorphism with BMD, where the T allele was associated with low BMD (Ohmori et al., 2002). A similar trend was also reported by other investigators although statistical significance was not reached (Langdahl et al., 2002; Wynne et al., 2002; Brandstrom et al., 2004; Vidal et al., 2006). In the Irish and Slovenian populations, a polymorphism that results in an amino acid change from lysine to asparagine in the signal peptide of OPG (G1181-C) was significantly associated with low BMD at both the LS and FN (Wynne et al., 2002; Arko et al., 2005).



Collagen Type 1α1 (COL1A1) Gene

Type I collagen is the most abundant and ubiquitously expressed of the collagen superfamily of proteins that are the most abundant proteins in the human body. Type I collagen is composed of two α1 and one α2 chains forming a triple helical structure, secreted as a pro-peptide followed by cleavage of the N and C telopeptides by proteases. Further post-translational modifications of collagen include hydroxylation of proline and lysine residues, cross-linking and glycosylation. Collagen is able to maintain the integrity of the tissues via its interactions with a large number of molecules that make up the extra cellular matrix and cell surfaces. More than fifty molecules are known to interact with collagen such as integrins, fibronectin, thrombospondin and matrix metalloproteinases (MMPs) (Di Lullo et al., 2002).

The gene encoding for α1(I) chain (COL1A1) is located at 17q21.3-q22.1, spans 18kb of the genome having a total of 52 exons, most of which are 54bp in length or an exact multiple of 9bp encoding Glycine – X – Y triplets. A number of SNPs and mutations have been identified and associated with diseases such as osteoporosis and osteogenesis imperfecta, especially if these variants lie in potential binding sites with other molecules (Dalgleish, 1997; Di Lullo et al., 2002).

In 1996, a G to T transversion at the Sp1 binding site of the COL1A1 gene promoter was identified and associated with low bone mass and an increased risk of vertebral fracture (Grant et al., 1996). Following this study a number of investigators studied this polymorphism for any association with BMD and/or fracture risk. The T allele was, however, absent in East Asian populations such as those of Shanghai, China (Lei et al., 2003). In association studies, the T allele was always associated with a low LS and FN BMD and with an increased risk of fractures (Gerdhem et al., 2004; Efstathiadou et al., 2001). Weichetova and colleagues observed that COL1A1 Sp1 genotyping might be a very useful predictor for an increased risk of wrist fractures, where TT homozygotes had a 2.8 times risk when compared to the wild-type GG homozygotes (Weichetova et al., 2000). They concluded that the increased risk of fractures was independent of BMD, an observation that was later confirmed by other similar studies (Bernad et al., 2002) and by a meta-analysis (Mann and Ralston, 2003). A reason for this might be that the Sp1 polymorphism alter gene expression and increases the production of collagen α1 chain and thus disrupts the ratio with α2(I) resulting in decreased bone quality (Mann et al., 2001).  The Sp1 polymorphism was also associated with an increased rate of bone loss especially at the lumbar spine (MacDonald et al., 2001) and with differences in femoral neck geometry, that might be another reason for an increased risk of fracture independent of BMD (Qureshi et al., 2001). In a large scale association study, as part of the European project GENOMOS, the Sp1 polymorphism was significantly associated with reduced BMD at both the lumbar and femoral sites, with TT homozygotes having the lowest BMD. These results remained the same even after adjustments for confounding factors. In the same study no significant association was found with fractures, but only a modest association was observed with the incidence of vertebral fractures that was independent of BMD (Ralston et al., 2006).

Another polymorphism in the promoter region of the COL1A1 gene that was described is the -1997 G/T that was in LD with the Sp1 variant and was also associated with BMD (Garcia-Giralt et al., 2002). This polymorphic site is thought to bind single-stranded DNA binding proteins that may be involved in the regulation of transcription. When studying these sites, Liu et al (2004) reported that the variation in BMD observed in elderly Caucasians might be a result of the interactions of these two polymorphic sites. Interactions between the -1997 polymorphism and another close by polymorphism (-1663) were found to affect transcription and possibly also affecting binding of osteoblast nuclear proteins (Garcia-Giralt et al., 2005).

Polymorphisms within the COL1A1 gene were found to affect BMD when interacting with other genes such as the ER1, TGF-β (Bustamante et al., 2007) and VDR (Uitterlinden et al., 2001). A recent study showed evidence of interactions (epistasis) between the COL1A1 gene locus (17q21.3) and loci where the IL-6 (7p15.3) and TNFRSF1B (MIM191191), encoding for tumour necrosis factor receptor 2 (1p36.2), genes are found (Yang et al., 2007). In this study a linkage study approach was used, where a number of microsatellite markers were analysed spanning the interval containing a number of candidate genes. The authors described how both IL-6 and TNFRSF1B are involved in osteoclast differentiation and how interactions of these genes with COL1A1 might have an effect on femoral neck BMD. These results also replicated the involvement of locus 1p36.2, already indicated by previous whole-genome scans (Devoto et al., 2001).



Methylenetetrahydrofolate Reductase (MTHFR) Gene

Another important gene of special interest in bone physiology is the methylenetetrahydrofolate reductase (MTHFR) gene. The gene is localised to 1p36.3 and is composed of 11 exons with a promoter that lacks a TATA box but contains CpG islands and several Sp1 binding sites (Gaughan et al., 2000).

Previous studies showed that high serum homocysteine concentrations might interfere with cross-linking of collagen, thus affecting bone quality (Lubec et al., 1996). Increased plasma homocysteine concentration has been observed in individuals with a low folate status that are homozygous for C677T variant in the methylenetetrahydrofolate reductase (MTHFR) gene (McLean et al., 2003). This polymorphism results in a missense substitution from alanine to valine giving rise to a thermolabile variant of MTHFR (Kang et al., 1988).  This common variant has been associated with various diseases including coronary artery disease (Kang et al., 1988), neural tube defects (van der Put et al., 1998) and recently with bone mineral density (BMD) (Miyao et al., 2000; Jorgensen et al., 2002).

LS BMD was observed to be significantly lower in TT homozygotes when compared to those having the wild-type allele (Miyao et al., 2000). Similar observations were reported in a study carried out in the Danish population where the T allele was also associated with a low BMD and increased fracture risk (Abrahamsen et al., 2003). These results differed from those of Jorgensen et al (2002) where an increased risk of fracture was only reported in homozygotes for the T allele. Also, Golbahar et al (2004) did not found any association of the C677T polymorphism with BMD but it was observed that hyperhomocysteinaemia, as a result of folate deficiency, is a more likely cause for the pathogenesis of osteoporosis. Levels of vitamins such as riboflavin, B12 and B6 are also thought to improve BMD in individuals that are homozygotes for the mutant allele (Abrahamsen et al., 2005). In a study in British postmenopausal women an association was found between folate and plasma homocysteine levels and BMD but not with vitamins B6 and B12 (Baines et al., 2007). Homozygotes TT for the MTHFR C677T polymorphism had significantly higher homocysteine levels when compared to the other genotypes. The same trend was not observed with folate or BMD.



Low Density Lipoprotein Receptor-Related Protein 5 (LRP5)

LRP5 is a transmembrane protein of the low density lipoprotein (LDL) receptor family that acts as a co-receptor together with frizzled receptors for Wnt signalling via β-Catenin dependent and independent pathways (Zorn, 2001). The structure of LRP5 is made up of four epidermal growth factor (EGF) repeats and three LDL receptor repeats with an intracellular portion that interacts with other proteins such as axin (Mao et al., 2001). The important role of LRP5 in osteoblast function was observed from experiments performed in mice where disruption of LRP5 resulted in a low bone mass phenotype (Kato et al., 2002).

LRP5 gene is located on human chromosome 11q13.4 and is made up of 23 exons spanning a region of 100kb. This locus was indicated by a number of linkage studies (Gong et al., 1996; Koller et al., 1998) and several mutations within the LRP5 gene were identified and confirmed to be responsible for an increased bone density phenotype (Van Wesenbeeck et al., 2003) and osteoporosis pseudoglioma syndrome (Gong et al., 2001).

Several SNPs were also analysed for any association with BMD by a number of investigators. In a study carried out by Ferrari and co-workers it was observed that variants within the LRP5 gene were affecting changes in vertebral bone mass and size mostly in males and not females (Ferrari et al., 2004). A year later, variants within the LRP5 gene were also associated with idiopathic osteoporosis in men (Ferrari et al., 2005). Other investigators found positive association of other SNPs within this gene with BMD and/or fractures in postmenopausal women and in the general population (Urano et al., 2004; Koay et al., 2004; Mizuguchi et al., 2004; Bollerslev et al., 2005; Xiong et al., 2007). A number of variants within the LRP5 and LRP6 genes were studied and were significantly associated with fracture risk especially in males (van Meurs et al., 2006).

Three mutations found in exons 1, 12 and 14 were observed to be responsible for low bone mass and increased risk of fractures in paediatric cases (Hartikka et al., 2005). These mutations were found in the heterozygous state, where the frameshift mutation was also found in the proband’s father and brother while a missense mutation was found in mother and two brothers. All these individuals were osteoporotic and no other mutations were detected in the COL1A1 and COL1A2 genes in these individuals. In another study, another gain of function mutation was reported in a woman with high bone mass and that resulted in reduced inhibition of Wnt signalling by Dickkopf 1 (DKK1) (Balemans et al., 2007).



Other Candidate Genes

      A number of other genes involved in bone physiology and also with other bone diseases have also been studied for any association with BMD or with an increased risk of osteoporosis and fracture. Table 1-2 shows a short list of genes that were identified either by linkage or association studies and are known to play a role in bone physiology. These genes are either known to be found in loci indicated by linkage studies (TNFRSF1B, PLOD1, TNSALP) or else were already studied and associated with BMD and/or increased fracture risk (Klotho, TNSALP). Results obtained by a linkage and association study in Iceland for the BMP-2 gene were not replicated in an independent study carried out in Rotterdam (Medici et al., 2006). An association of polymorphisms found within the BMP2 gene and other BMPs was only observed with peak bone mass (Choi et al., 2006).


Table 1-2 Candidate Genes Associated with Bone Mass






Androgen Receptor


Yamada et al (2005)


Interleukin 6


Moffett et al (2004)


Bone gla Protein (osteocalcin)


Dohi et al (1998)


Tumour Necrosis Factor Receptor 2


Albagha et al (2002)


Lysyl Hydroxylase


Spotila et al (2003)


Apolipoprotein E


Schoofs et al (2004)


Transforming Growth Factor-β


Langdahl et al (2003)




Uitterlinden et al (2004)


Klotho Protein


Kawano et al (2002)


Insulin-like Growth Factor-I


Kim et al (2002)


T-cell Immune Regulator


Sobacchi et al (2004)


Bone Morphogenic Protein 2


Styrkarsdottir et al (2004)


Phosphodiesterase 4D


Reneland et al (2005)


Tissue-nonspecific alkaline phosphatase


Goseki-Sone et al (2005)