Calcified Tissue International

, Volume 74, Issue 1, pp 35–41

Association Between a Polymorphism Affecting an AP1 Binding Site in the Promoter of the TCIRG1 Gene and Bone Mass in Women

Authors

  • C. Sobacchi
    • Instituto di Tecnologie BiomedicheCNR, Via Fratelli Cervi, 93, 20090 Segrate, Milano
  • P. Vezzoni
    • Instituto di Tecnologie BiomedicheCNR, Via Fratelli Cervi, 93, 20090 Segrate, Milano
  • D. M. Reid
    • Department of Medicine and TherapeuticsUniversity of Aberdeen, Foresterhill, Aberdeen, AB25 2ZD Scotland
  • F. E. A. McGuigan
    • Department of Medicine and TherapeuticsUniversity of Aberdeen, Foresterhill, Aberdeen, AB25 2ZD Scotland
  • A. Frattini
    • Instituto di Tecnologie BiomedicheCNR, Via Fratelli Cervi, 93, 20090 Segrate, Milano
  • M. Mirolo
    • Instituto di Tecnologie BiomedicheCNR, Via Fratelli Cervi, 93, 20090 Segrate, Milano
  • O. M. E. Albhaga
    • Department of Medicine and TherapeuticsUniversity of Aberdeen, Foresterhill, Aberdeen, AB25 2ZD Scotland
  • A. Musio
    • Instituto di Tecnologie BiomedicheCNR, Via Fratelli Cervi, 93, 20090 Segrate, Milano
  • A. Villa
    • Instituto di Tecnologie BiomedicheCNR, Via Fratelli Cervi, 93, 20090 Segrate, Milano
    • Department of Medicine and TherapeuticsUniversity of Aberdeen, Foresterhill, Aberdeen, AB25 2ZD Scotland
Article

DOI: 10.1007/s00223-002-0004-2

Cite this article as:
Sobacchi, C., Vezzoni, P., Reid, D. et al. Calcif Tissue Int (2004) 74: 35. doi:10.1007/s00223-002-0004-2

Abstract

The TCIRG1 gene encodes a component of the osteoclast vacuolar proton pump and previous work has shown that inactivating mutations of the TCIRG1 cause autosomal recessive osteopetrosis. In order to determine whether allelic variation in TCIRG1 contributes to the regulation of bone mineral density (BMD) in normal individuals, we studied the relationship between polymorphisms of TCIRG1 and BMD in a population-based cohort of 739 perimenopausal women. Five common polymorphisms were identified: two in the promoter, a conservative change within exon 4, one within intron 4 and one within intron 11. One of the promoter polymorphisms (G-1102A) lay within a consensus recognition site for the AP1 transcription factor. There was a significant association between the G-1102A genotype and BMD at the lumbar spine (P = 0.01) and femoral neck (P = 0.03). The association remained significant after correcting for age, weight, height, menopausal status/HRT use and smoking (P = 0.008 for spine BMD and P = 0.03 for hip BMD), and homozygotes for the −1100 “G” allele had BMD values significantly higher than individuals who carried the −1100 “A” allele at both spine (P = 0.007) and hip (P = 0.047). Subgroup analysis showed that the association between G-1102A and BMD was restricted to premenopausal women who comprised 50.6% of the study group. None of the other polymorphisms or haplotypes were significantly associated with BMD in the study group as a whole or in any subgroup. Functional studies will need to be performed to determine the mechanisms that underlie this association, but we conclude that, in this relatively large population, allelic variation at the G-1102A site of TCIRG1 accounts for part of the heritable component of BMD in Scottish women, possibly by affecting peak bone mass.

Keywords

TCIRG1 geneBMDPolymorphismsPremenopause

Genetic factors play an important role in the pathogenesis of osteoporosis—a common disease characterized by reduced bone mass, microarchitectural deterioration of bone tissue and increased susceptibility to fragility fractures [1]. Bone mineral density (BMD) is an important predictor of osteoporotic fracture risk, and evidence from twin and family studies suggests that between 50–85% of the variance in BMD is genetically determined [2, 3, 4]. Bone mineral density is a complex trait, which is regulated by an interaction between environmental factors such as diet and exercise and polymorphic variants in several different genes, each with modest effects on BMD. A wide variety of candidate genes have been studied so far in relation to BMD, including the vitamin D receptor [5], the estrogen receptor [6], and the COLIA1 gene [7]. Current evidence suggests that allelic variation in these genes accounts for a relatively small portion of the population variance in BMD [8] indicating that many of the genes that regulate BMD remain to be discovered. Linkage studies in humans have mapped three Mendelian traits that are associated with abnormalities of BMD to a region of chromosome 11q12-13. These are osteoporosis-pseudoglioma syndrome [9], autosomal recessive infantile osteopetrosis [10] and high bone mass [11]. Positional cloning studies have shown that inactivating mutations of the Lipoprotein Receptor-Related Protein 5 gene (LRP-5) are responsible for osteoporosis-pseudoglioma syndrome [12] whereas an activating mutation in the same gene accounts for high bone mass [13, 14]. The TCIRG1 gene, which encodes a 116 Kd subunit of the vacuolar proton pump, also maps to this region of chromosome 11q12-13. Previous studies have identified inactivating mutations of TCIRG1 in approximately 60% of individuals with infantile osteopetrosis [15, 16]. Moreover, recent work has shown evidence of linkage between a polymorphism at the TCIRG1 locus and femoral neck BMD in healthy premenopausal sib-pairs [17]. In this study we examined the relationship between several common polymorphisms of TCIRG1 and BMD in a population-based sample of 739 Scottish women aged 45–55.

Subjects and Methods

Subjects

The study group comprised 739 unrelated women aged 45–55 who were randomly selected from a large population-based BMD screening program for osteoporotic fracture risk [18]. This program originally involved 7,000 women who were identified using Community Health Index records (CHI) from a 25-mile radius of Aberdeen, a city with a population of about 250,000 in the Northeast of Scotland. Women were invited by letter to undergo BMD measurements between 1990–1994 and 5119 of the 7000 invited (73.1%) attended for evaluation. Blood samples were subsequently obtained for DNA extraction on 3069 (59.9%) of these individuals. Participants were weighed wearing light clothing and no shoes on a set of balance scales calibrated to 0.05 kg (Seca, Hamburg, Germany). Height was measured using a stadiometer (Holtain Ltd, Crymych, United Kingdom). Participants completed a questionnaire on menopausal status and use of hormone replacement therapy (HRT) and on the basis of this, were classified into five groups. They were classified as “premenopausal” if they were not on HRT and menstruating regularly (n = 374) and “perimenopausal” if they were not on HRT and menstruation was irregular and/or if up to 6 months had elapsed since their last period (n = 14), and “postmenopausal” if they were not on HRT and menstruation had ceased for 6 months or more (n = 144). The remaining two groups consisted of women who were currently receiving HRT at the time of study (n = 196) and those who previously had received HRT (n = 11). Current and previous HRT users were not further classified in terms of menopausal status. All participants gave written informed consent to being included in the study which was approved by the Grampian Joint Research Ethical Committee.

Bone Mineral Densitometry

Bone mineral density (BMD) measurements of the left proximal femur (the femoral neck, FN) and lumbar spine (LS) (L2-4) were performed by dual energy x-ray absorptiometry using one of two Norland XR26 or XR36 densitometers (Norland Corp, Wisconsin, USA). Calibration of the machines was performed daily, and quality assurance was checked by measuring the manufacturer’s lumbar spine phantom at daily intervals and a Hologic spine phantom at weekly intervals. The in vivo precision for the XR36 was 1.2% for the lumbar spine (LS), and 2.3% for the femoral neck (FN). Corresponding values for the XR26 were 1.95% and 2.31% (LS and FN, respectively).

Mutation Screening and Genotyping

Mutation screening was carried out by DNA sequencing of the promoter and intron-exon boundaries of the TCIRG1 gene in DNA extracted from peripheral venous blood samples from about 70 individuals using PCR-based methods, as previously described [15, 16]. Genotyping for polymorphisms was carried out by DNA sequencing of PCR-amplified fragments of genomic DNA. The PCR products for sequencing were generated using Qiagen Taq DNA polymerase, Q-solution and standard reaction buffer containing 1.5 mM MgCl2 according to the manufacturer’s recommendations. The PCR was carried out for 35 cycles with a melting temperature of 95°C, an annealing temperature of 60°C and an extension temperature of 72°C. The promoter polymorphisms (G-1102A and G-920A) were analyzed using the following primer pairs: forward—5′ ACAAGGCAGGCGCAGGACTCC and reverse—CGGGCCTGGAAACTGAGTCAC; the exon 4 (C3856T) and intron 4 (A3900G) polymorphisms were analyzed using the following primer pairs: forward 5′ TTGGGGCAGCAGGTGGGGCC 3′ and reverse—AGAGGAGAACCCCCTAGGGCTAG 3′; and the intron 11 polymorphism (G8645A) was analyzed using the following primer pairs: forward: GTTCGGGGATGTGGGCCAC 3′ and reverse: 5′ GCCCATAAGCAGGAGCAGG 3′ (for G8645A). The PCR products were treated with Exonuclease III and Shrimp Alkaline Phosphatase-(Exo-SAP-IT) (Amersham Pharmacia) according to the manufacturers instructions and sequenced using the forward and/or reverse primer as the sequencing primer using DYNamic ET sequencing chemistry on a MegaBace 1000 DNA sequencer (Amersham Pharmacia).

Statistical Methods

Statistical analysis was carried out using Minitab version 12 (Minitab Inc, Pennsylvania, USA). Differences in BMD between the genotypes were tested using one-way ANOVA and General Linear Model (GLM) analysis of variance (ANOVA) adjusting for height, weight, age, menopausal status/HRT use and smoking. Haplotypes were constructed from the population genotype data by the algorithm of Niu et al. [19] using the Haplotyper program. GLM ANOVA analysis was also used to test for allelic associations by combining data from the genotype groups and for haplotypes predicted by the Haplotyper program. Stepwise logistic regression was used to evaluate the relative contribution of genotype and other factors to the population variance in BMD. Linkage disequilibrium between polymorphisms was estimated by calculating D’ values using the 2BY2 program on output generated by the EH program [20]. Both programs were obtained from the Columbia University Website.

Results

We identified 5 common polymorphisms (those with allele frequency greater than 5%) in TCIRG1 on mutation screening of 70 normal subjects. (1) a C to T change at position 3856, (C3856T), (2) an A to G change at position 3900 (A3900G), and (3) a G to A change at position 8645 (G8645A) on sequence accession number AF033033. The C3856T change is within exon 4 of the osteoclast-specific transcript of the TCIRG1 gene but is a conservative change (CAC – CAT; both histidine). The A3900G polymorphism is within intron 4 of TCIRG1 and the G8645A is within intron 11 (G8645A). Two additional polymorphisms were discovered in the TCIRG1 promoter at positions 9326 (G9326A) and 9508 (G9508A) on sequence accession number AP002807. These polymorphisms are situated 1102 nucleotides upstream (G9326A) and 920 nucleotides upstream (G9508A) of the first nucleotide of the TCIRG1 mRNA start site which is at position 10428 on the sequence. Hereafter we refer to these polymorphisms as G-1102A and G-920A. We did not detect any of the exonic polymorphisms present in the SNP database cited by Carn et al. [17] with the exception of the C226T change which predicts an arginine to tryptophan amino change at codon 56 (this is at position 2827 of sequence accession number AF033033). This was rare, however, with an allele frequency of only 2% in the normal subjects used for mutation screening and was not analyzed further in the population-based study.

Details of age, BMD, height, weight, smoking history, menopausal status and HRT use in the study population are shown in Table 1. Menopausal status of the women was 50.6% premenopausal, 1.8% perimenopausal and 19.4% postmenopausal. The average time elapsed since menopause was 6.07 years in the postmenopausal group. Menopausal status was unclassified for 207 (28%) of subjects because the date of cessation of natural menstruation could not be accurately established because of current HRT use in 196 women (26.5%) and previous HRT use in 11 women (1.4%).

Table 1

Demographic details of study population

Number

739

 

Age

47.9 ± 1.53

 

Premenopausal

374 (50.6%)

 

Perimenopausal

14 (1.9%)

 

Postmenopausal

144 (19.5%)

 

Previous HRT users

11 (1.5%)

 

Current HRT users

196 (26.5%)

 

Years since menopause

   6.1 ± 5.3

 

Spine BMD (g/cm2)

   1.067 ± 0.15

 

Femoral neck BMD (g/cm2)

   0.890 ± 0.12

 

Weight

65.9 ± 11.8

 

Height

161.5 ± 6.1

 

Values are means ± SD or numbers and percentages

Years since menopause refers to the postmenopausal group only.

Significant linkage disequilibrium (LD) was observed between most of the polymorphisms identified. The strongest LD was between G-1102A and G-920A (D′ = 0.80, P < 0.0001). Other LD values ranged between 0.569 and 0.752 (all P < 0.001), with the exception of the C3856T polymorphism which showed significant LD only with A3900G. (D′ = 0.321; P < 0.001). Analysis using the Haplotyper program predicted 27 different haplotypes from the genotype data, but five common haplotypes were identified that accounted for 77.3% of alleles at the TCIRG1 locus. These are summarized in Figure 1, which also illustrates the position of the polymorphisms in relation, to the TCIRG1 gene structure.

https://static-content.springer.com/image/art%3A10.1007%2Fs00223-002-0004-2/MediaObjects/fig1.gif
Figure 1

TCIRG1, gene structure, location of polymorphisms, and common halplotypes. Common haplotypes with allele frequency greater than 5% are shown. The translation start site (CDS) of the osteoclast specific isoform is indicated.

We studied the relationship between genotypes at each site and BMD values before and after adjustment for age, height, weight, menopausal status/HRT use and smoking. The results of this analysis are shown in Table 2 for spine BMD and Table 3 for hip BMD. The genotype distributions of G-l102A, G-920A, A3900G and G8645A were as predicted by Hardy-Weinberg equilibrium, but for the C3856T polymorphism we found more C/T heterozygotes than expected (102 vs 64, P = 0.007).

Table 2

Lumbar spin BMD values in relation to TCIRG1 genotypes and alleles

 

Genotypes

Allele

ANOVA (genotypes)

ANOVA (alleles)

      

G-1102A

GG (n = 136)

GA (n = 309)

AA (n = 163)

GA/eAA (n = 472)

P-value

P-value

      

Unadjusted BMD

1.099 ± 0.173

1.052 ± 0.158

1.080 ± 0.140

1.062 ± 0.152

0.01

0.02

      

Adjusted BMD

1.074 ± 0.018

1.027 ± 0.016

1.049 ± 0.018

1.035 ± 0.015

0.008

0.007

      

G-920A

GG (n = 121)

GA (n = 333)

AA (n = 193)

GA/eAA (n = 526)

        

Unadjusted BMD

1.079 ± 0.181

1.059 ± 0.156

1.077 ± 0.146

1.065 ± 0.153

0.31

0.37

      

Adjusted BMD

1.061 ± 0.017

1.039 ± 0.014

1.052 ± 0.016

1.043 ± 0.014

0.28

0.24

      

C3856T

CC (n = 495)

CT (n = 102)

TT (n = 2)

CT/eTT (n = 104)

        

Unadjusted BMD

1.072 ± 0.157

1.043 ± 0.148

1.070 ± 0.092

1.044 ± 0.147

0.24

0.09

      

Adjusted BMD

1.049 ± 0.014

1.014 ± 0.019

1.101 ± 0.103

1.015 ± 0.020

0.08

0.04

      

A3900G

AA (n = 346)

AG (n = 213)

GG (n = 42)

AG/eGG (n = 255)

        

Unadjusted BMD

1.095 ± 0.200

1.054 ± 0.151

1.074 ± 0.150

1.074 ± 0.150

0.19

0.33

      

Adjusted BMD

1.054 ± 0.015

1.031 ± 0.016

1.072 ± 0.025

1.039 ± 0.015

0.11

0.20

      

G8645A

GG (n = 428)

GA (n = 160)

AA (n = 21)

GA/eAA (n = 181)

        

Unadjusted BMD

1.064 ± 0.153

1.066 ± 0.167

1.128 ± 0.161

1.073 ± 0.167

0.19

0.53

      

Adjusted BMD

1.038 ± 0.014

1.042 ± 0.016

1.102 ± 0.034

1.048 ± 0.016

0.17

0.39

      

Unadjusted BMD values are mean ± SD in g/cm2. Adjusted BMD values are least squares mean ± SD BMD values in g/cm2 adjusted for age, weight, height, menopausal status/HRT use and smoking

Table 3

Femoral neck BMD values in relation to TCIRG1 genotypes and alleles

 

Genotypes

Alleles

ANOVA (genotypes)

ANOVA (alleles)

      

G-1102A

GG (n = 136)

GA (n = 309)

AA (n = 164)

GA/eAA (n = 473)

P-value

P-value

      

Unadjusted BMD

0.907 ± 0.144

0.879 ± 0.119

0.904 ± 0.123

0.888 ± 0.121

0.04

0.12

      

Adjusted BMD

0.896 ± 0.014

0.866 ± 0.013

0.886 ± 0.014

0.873 ± 0.012

0.03

0.05

      

G-920A

GG (n = 121)

GA (n = 333)

AA (n = 194)

GA/eAA (n = 527)

        

Unadjusted BMD

0.902 ± 0.147

0.880 ± 0.121

0.900 ± 0.121

0.888 ± 0.121

0.12

0.25

      

Adjusted BMD

0.895 ± 0.014

0.871 ± 0.011

0.884 ± 0.013

0.875 ± 0.011

0.11

0.10

      

C3856T

CC (n = 496)

CT (n = 102)

TT (n = 2)

CT/eTT (n = 104)

        

Unadjusted BMD

0.894 ± 0.126

0.873 ± 0.117

0.865 ± 0.087

0.873 ± 0.117

0.30

0.12

      

Adjusted BMD

0.874 ± 0.011

0.848 ± 0.015

0.897 ± 0.082

0.849 ± 0.015

0.13

0.05

      

A3900G

AA (n = 347)

AG (n = 213)

GG (n = 42)

AG/eGG (n = 255)

        

Unadjusted BMD

0.886 ± 0.123

0.888 ± 0.122

0.909 ± 0.145

0.892 ± 0.126

0.55

0.61

      

Adjusted BMD

0.891 ± 0.011

0.905 ± 0.013

0.932 ± 0.026

0.872 ± 0.012

0.47

0.79

      

G8645A

GG (n = 429)

GA (n = 160)

AA (n = 21)

GA/eAA (n = 181)

        

Unadjusted BMD

0.895 ± 0.121

0.876 ± 0.132

0.914 ± 0.130

0.881 ± 0.132

0.19

0.21

      

Adjusted BMD

0.878 ± 0.011

0.862 ± 0.012

0.900 ± 0.027

0.865 ± 0.012

0 18

023

      

Unadjusted BMD values are mean ± SD in g/cm2. Adjusted BMD values are least squares mean ± SD BMD values in g/cm2 adjusted for age, weight, height, menopausal status / HRT use and smoking

There was a significant association between G-1102A polymorphism and both spine and hip BMD. The differences were significant for unadjusted and adjusted BMD values. When data were combined for the G/A heterozygotes and A/A homozygotes, the difference between groups was also significant at the spine and hip for adjusted BMD. A nonsignificant trend for association between the C3856T polymorphism and adjusted spine BMD values was observed (P = 0.079) and this became significant when the C/T and T/T genotypes were combined (P = 0.036). None of the other polymorphisms was associated with BMD, nor did we find a significant association between any of the TCIRG1 haplotypes predicted by the Haplotyper program and BMD, except when haplolypes were grouped according to the presence or absence of the A allele at the G-1102A site (data not shown). There was no association between TCIRG1 genotype and age, weight, height, smoking or menopausal status (data not shown).

We also studied the relationship between TCIRG1 genotypes in relation to menopausal status and HRT use. This analysis was restricted to premenopausal women, postmenopausal women and current HRT users in view of the small number of subjects in the perimenopausal and previous HRT-user groups. There was no significant association among G-920A, C3856T, A3900G or G8645A genotypes and BMD in any of these subgroups, nor was there an association between TCIRG1 haplotypes and BMD (data not shown). The G-1102A polymorphism was significantly associated with BMD in the subgroup of women who were premenopausal, but there was no association between G-1102A and BMD in postmenopausal women or HRT users (Table 4).

Table 4

TCIRG1 G-1102A alleles and BMD in relation to HRT use and menopausal status

 

G-1102A Genotypes

G-1102A Alleles

ANOVA (genotypes)

ANOVA (alleles)

      

Spine BMD

GG

GA

AA

GA/eAA

P-value

P-value

      

Premenopausal (n = 308)

(n = 67)

(n = 157)

(n = 83)

(n = 240)

        

Adjusted BMD

1.142 ± 0.020

1.075 ± 0.013

1.099 ± 0.019

1.082 ± 0.012

0.01

0.006

      

Postmenopausal (n = 119)

(n = 29)

(n = 52)

(n = 37)

(n = 89)

        

Adjusted BMD

1.085 ± 0.026

1.065 ± 0.021

1.028 ± 0.024

1.049 ± 0.016

0.23

0.23

      

HRT users (n = 164)

(n = 32)

(n = 94)

(n = 37)

(n = 141)

        

Adjusted BMD

1.025 ± 0.025

1.016 ± 0.014

1.064 ± 0.024

1.028 ± 0.013

0.21

0.86

      

Femoral neck BMD

GG

GA

AA

GA/eAA

        

Premenopausal (n = 308)

(n = 67)

(n = 157)

(n = 84)

(n = 240)

        

Adjusted BMD

0.934 ± 0.015

0.882 ± 0.010

0.903 ± 0.014

0.888 ± 0.009

0.01

0.008

      

Postmenopausal (n = 119)

(n = 29)

(n = 52)

(n = 37)

(n = 89)

        

Adjusted BMD

0.896 ± 0.020

0.904 ± 0.016

0.888 ± 0.018

0.897 ± 0.013

0.79

0.97

      

HRT users (n = 164)

(n = 32)

(n = 94)

(n = 37)

(n = 141)

        

Adjusted BMD

0.869 ± 0.021

0.869 ± 0.012

0.900 ± 0.020

0.877 ± 0.011

0.38

0.71

      

Adjusted BMD values are least squares mean ± SD BMD values (g/cm2) adjusted for age, weight, height, menopausal status/HRT use and smoking

Analysis of the data by stepwise multiple regression identified three independent predictors of spine BMD, which together accounted for 13.3% of the variance in spine BMD. These were body weight (9.41% of the variance, P < 0.0001); menopausal status/HRT use (3.16% of the variance, P < 0.0001); and the G-1102A allele (1.00% of the variance, P = 0.017). For femoral neck BMD, we identified two independent predictors which accounted for 11.1% of the variance. These were body weight (13.6% of the variance, P < 0.0001) and menopausal status/HRT use (0.85% of the variance, P = 0.009).

Discussion

The TCIRG1 gene on chromosome 11q12 encodes a 116 Kd subunit of the osteoclast-specific vacuolar proton pump. This is a component of the Vacuolar ATPase complex expressed in the osteoclast ruffled border and is responsible for transport of H+ ions into the resorption lacuna, where the low pH plays an essential role in dissolving hydroxyapatite crystals [21]. The importance of TCIRG1 in bone metabolism is demonstrated by the fact that targeted inactivation of the gene in mice causes osteoclast-rich osteopetrosis [22]. A deletion involving the 5′ region of TCIRG1 is also known to be responsible for osteopetrosis in oc/oc mice [23] and several homozygous inactivating mutations of the gene have been identified in patients with autosomal recessive infantile osteopetrosis [15]. It is currently considered that between 50–60% of individuals with autosomal recessive osteopetrosis have mutations in the TCRIG1 gene [16].

The above observations led us to investigate the possibility that bone mass in the general population may be associated with allelic variation at the TCIRG1 locus. Mutation screening resulted in the identification of 5 common polymorphisms over a 13 Kb region at the TCIRG1 locus. Two polymorphisms were situated in the promoter (G-1102A and G-920A); one conservative change within exon 4 (C3856T), one within intron 4 (A3900G) and one within intron 11 (G8645A). These showed strong and highly significant linkage disequilibrium with each other in our population, with the exception of C3856T where linkage disequilibrium was only observed with A3900G. We observed a significant association between BMD and allelic variation at the G-1102A site. The association was highly significant at the spine (P = 0.007) and at the femoral neck (P = 0.03), after correcting for potential confounding factors including age, height, weight, menopausal status/HRT use and smoking. Though we observed an association between G-1102A in the whole population, subgroup analysis revealed that the effect was primarily driven by an association in the premenopausal population. This would be consistent with a model whereby the TCIRG1 G-1102A allele affects peak bone mass rather than postmenopausal bone loss, although further work in other populations will be required to confirm this. The GA heterozygotes and AA homozygotes at the G-1102A site had lower BMD values at both skeletal sites than GG homozygotes, consistent with a recessive effect of the G allele on BMD. G-1102 alleles were found to be a significant and independent predictor of BMD, but allelic variation at this site accounted for only 1% of the variance in spine BMD, suggesting that TCIRG1 is not a major gene for determining BMD.

The mechanisms underlying the association we have described here will require further investigation. It is of interest, however, that the G-1102A polymorphism is situated at a consensus recognition sequence for the transcription factor AP1 (http://transfac.gbf.de/ ). In the presence of the G-nucleotide, a consensus AP1 site is present on the reverse strand (TCACGGC) whereas in the presence of the A nucleotide, the consensus sequence is altered (TCATGGC). Although these data are of potential interest, it is currently unclear whether this AP1 site is functional, and further work is needed to assess this. Our findings are of interest in relation to the work of Carn et al. [17], who demonstrated evidence of linkage between a polymorphism within intron 10 of the TCIRG1 gene and femoral neck BMD. In that study, the authors found no evidence of association between BMD and the intron 10 polymorphism, but they did not study the G-1102A polymorphism described here. We did not study the intron 10 polymorphism since it lies deep within a large intron and we considered it unlikely to be functional. Though our data showed the strongest association with the G-1102A polymorphism, it remains possible that this effect could be due to linkage disequilibrium with functional polymorphisms elsewhere in the TCRIG1 gene. In this regard, a coding polymorphism in TCIRG1 has been described (at position 2827 on AF033033) that causes an arginine to tryptophan amino acid change at codon 56 (R56W). While we observed this polymorphism in our population, it was rare (allele frequency 0.02) and therefore unlikely to explain the effect observed. Linkage disequilibrium with other polymorphisms in the TCIRG1 coding region also seems unlikely since we have failed to detect such polymorphisms by DNA sequencing of hundreds of chromosomes in normal and osteopetrotic subjects. Linkage disequilibrium with polymorphisms in other genes nearby such as LRP-5 or Fra-1 [12, 13, 14, 24] is possible but unlikely in view of the fact that LRP-5 is over 300 Kb and Fral 500 Kb distant from TCIRG1. Our data therefore suggest that the association may truly be explained on the basis of a functional polymorphism at the G-1102A site in the promoter, but studies in other populations will be required to confirm the findings reported here, since we cannot exclude the possibility of a false-positive result due to population stratification.

In conclusion, the data presented here suggest that allelic variation at the G-1102A polymorphism in the promoter of the TCIRG1 gene may be a genetic determinant of BMD in premenopausal women of Scottish descent.

Acknowledgements

We are grateful to R. Dulbecco, A. Albertini, P. Raineri and L. Rossi Bernardi for their encouragement. The technical assistance of L. Susani, E MacLeod and G Taylor is thankfully acknowledged. This work was partially supported by grants from MIUR-FIRB to P.V.; an MRC co-operative group grant to SHR and DMR; an Integrated Clinical Arthritis Centre grant from the Arthritis Research Campaign to DMR and SHR and a grant from the Grampian Osteoporosis Trust to SHR and DMR. This work is manuscript no. 59 of the Genoma 2000/ITBA Project funded by CARIPLO.

Copyright information

© Springer-Verlag 2003