• RANK;
  • RANKL;


  1. Top of page
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

Over the past decade, many genome-wide association studies (GWAs) and meta-analyses have identified genes and regions involved in osteoporotic phenotypes. Nevertheless, the large majority of these results were not tested at any functional level. GWA-associated single-nucleotide polymorphisms (SNPs) near candidate genes such as RANK and RANKL suggest that these SNPs and/or other variants nearby may be involved in bone phenotype determination. This study focuses on SNPs along these two genes, which encode proteins with a well-established role in the bone remodeling equilibrium. Thirty-three SNPs, chosen for their location in evolutionary conserved regions or replicated from previous studies, were genotyped in the BARCOS cohort of 1061 postmenopausal women and tested for association with osteoporotic phenotypes. SNP rs9594738, which lies 184 kb upstream of the RANKL gene, was the only SNP found to be associated with a bone phenotype (dominant model: beta coefficient = –0.034, p = 1.5 × 10−4, for lumbar spine bone mineral density). Functional experiments exploring a distal region (DR) of 831 bp that harbors this SNP in a centered position (nt 470) demonstrated its capacity to inhibit the RANKL promoter in reporter gene assays. Remarkably, this DR inhibition was significantly reduced in the presence of vitamin D. In conclusion, the GWA-associated SNP rs9594738 lies in a region involved in transcription regulation through which vitamin D could be regulating RANKL expression and bone mineral density. © 2013 American Society for Bone and Mineral Research.


  1. Top of page
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

The cellular interactions between osteoblasts and osteoclasts, in which the RANK/RANKL/OPG system plays a key role, are essential for the regulation of bone remodeling.[1, 2] Any variation of this system that results in a negative balance between bone formation and resorption may lead to osteoporosis.[3]

Osteoblasts and stromal cells express the receptor activator of NF-κB (RANK) ligand (RANKL), which binds to RANK on the surface of osteoclasts and their precursors and leads to their differentiation and survival.[4] Single-nucleotide polymorphisms (SNPs) located near TNFSF11 (gene map locus 13q14), which encodes RANKL, and TNFRSF11A (gene map locus 18q22.1), which encodes RANK, have been associated with bone mineral density (BMD) and fractures in several genome-wide association studies (GWAs) and meta-analyses.[5-11] The majority of the recent GWAs used SNPs from the HapMap database[12] as markers along the genome to detect association with osteoporotic phenotypes. However, the actual functional variants remain largely undiscovered. The functionality of tag SNPs needs to be explored, together with the possible presence of other functional variants in the region, to better understand the overall implication of a given gene. In this setting, genotyping putative functional SNPs within or close to the genes suggested by the GWAs results may be appropriate.

In the case of RANKL, the BMD-associated SNPs at the genome-wide level, rs9594738 and rs9533090, are both located ∼184 kb upstream, in a noncoding region with scarce functional data. These SNPs belong to a haplotypic block located between the RANKL and the A kinase anchor protein 11 (AKAP11) genes, within a highly conserved region. Interestingly, in all species with an identifiable RANKL gene, AKAP11 lies upstream of RANKL, whereas the downstream gene varies.[13] The conservation of this large intergenic region suggests that RANKL expression might be regulated by a complex mechanism.[13] The ENCODE Project Consortium has predicted several functional genomic elements in this region in humans.[14] The regulatory hypothesis is also supported by experimental evidence in mice showing that elements located far upstream of the Rankl gene (including a vitamin D response element [VDRE] at –76 kb) do affect its expression.[13, 15-17] Moreover, a functional VDRE was identified by a Chip-seq assay in two human lymphoblastoid cell lines (GM10855 and GM10861) in the region nearby SNP rs9594738.[18]

To further explore the BMD- and fracture-related RANK and RANKL variants, in a cohort of Spanish postmenopausal women, we genotyped potentially functional SNPs located in highly conserved regulatory regions of the promoter and intron 1, in the exons, and in the 3′ UTR. We also replicated the BMD-associated SNPs from previous GWAs. Finally, we analyzed the relevance of a far upstream region in the regulation of RANKL expression. We cloned an 831-bp fragment containing rs9594738 upstream of the RANKL promoter and analyzed the effect of vitamin D treatment on the functional outcome.

Materials and Methods

  1. Top of page
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

Study subjects

BARCOS cohort participants were recruited from Hospital del Mar, Barcelona.[19, 20] All patients were consecutive, unselected, postmenopausal women attending the outpatient clinic. Patients were prospectively recruited regardless of BMD value (Table 1). Exclusion criteria were any history of metabolic or endocrine disease, chronic renal failure, chronic liver disease, malignancy (except superficial skin cancer), Paget's disease of bone, malabsorption syndrome, hormone-replacement therapy, antiresorptive or anabolic agents, oral corticosteroids, anti-epileptic drugs, and lithium, heparin, or warfarin treatments. In addition, women with early menopause (before the age of 40 years) were excluded for this analysis. Blood samples and written informed consents were obtained in accordance with the regulations of the Human Subjects Research Review Committee for Genetic Procedures, Hospital del Mar. Patients who declined the invitation to participate or did not give informed consent were excluded.

Table 1. Baseline Characteristics of the BARCOS Cohort
Patient characteristicsPhase 1Phase 2
Mean ± SDnMean ± SDn
Age at menopause (years)48.27 ± 3.9288448.41 ± 3.911061
BMI (kg/m2)26.35 ± 3.8788426.15 ± 3.841058
Breastfeeding (months)7.97 ± 13.248847.79 ± 12.821054
Age at LS densitometry (years)55.64 ± 8.5588455.98 ± 8.441057
Years since menopause LS7.37 ± 8.358847.59 ± 8.271057
LS BMD (g/cm2)0.853 ± 0.158840.854 ± 0.151060
Age at FN densitometry (years)57.71 ± 8.0380157.74 ± 7.95972
Years since menopause FN9.41 ± 7.908019.30 ± 7.87972
FN BMD (g/cm2)0.683 ± 0.118010.684 ± 0.11976
Age at menarche (years)12.91 ± 1.6087412.88 ± 1.601044
Fractures135 (15.3%)884145 (13.7%)1061

Bone mineral density measurement and fracture assessment

The BMD (g/cm2) was measured at the lumbar spine (LS) L1 to L4 and at the nondominant femoral neck (FN) using a dual-energy X-ray densitometer (QDR 4500 SL, Hologic, Waltham, MA, USA). In our center, the technique has an in vivo coefficient of variation (CV) of 1.0% for LS and 1.65% for FN measurements. Nonvertebral and clinical vertebral fractures were recorded. Nonvertebral fractures were validated from medical records, and spine X-ray was performed at baseline when there was a history of vertebral fracture diagnosis, height loss, or back pain. Fractures were defined as osteoporotic if they occurred after the age of 45 years and were the result of low-impact trauma (ie, fall from standing height). Fractures of the face, fingers, toes, and skull were excluded. Vertebral fractures were defined according to the semiquantitative criteria of Genant and colleagues.[21]

DNA extraction

The buffy coat of 3 mL of blood collected in EDTA tubes was stored at –20°C. Genomic DNA was obtained from leukocytes by a salting-out procedure[22] or by Autopure LS (Qiagen, Valencia, CA, USA), a robotic workstation for automated purification of genomic DNA using autopure chemistry, at LABS Laboratory Biomedical Support Services, IMIM, Barcelona, Spain. Samples were stored at –20°C.

SNPs selection

Putative functional SNPs were chosen using the ENSEMBL (,[23] UCSC genome browser (,[24, 25] Entrez SNP (,[26] and HapMap ( databases.[12] The SNPs from the proximal promoter and intron 1 were selected according to their evolutionary conservation. To establish conserved regions, genomic sequences of Mus musculus, Rattus norvigicus, Canis familiaris, Bos taurus, and Homo sapiens (mouse, rat, dog, cow, and human, respectively) were compared. Using the ENSEMBL multiple alignment tool, we chose SNPs within conserved regions. Only those with a minor allele frequency (MAF) >0.1 were included in the study. SNPs lacking published MAF in Utah residents with ancestry from northern and western Europe (CEU) were validated in 20 chromosomes from the BARCOS cohort by means of PCR-RFLP or Sanger sequencing. For 3′ UTR SNPs, those with a published MAF >0.01 in one of the referenced databases were selected to ensure a good coverage of this region, relevant for miRNA regulation. Other SNPs were selected if there was either a previous report of RANK or RANKL SNP association with BMD or fracture risk, or exonic changes (either synonymous or nonsynonymous).


Genotyping was performed in two phases: the first one, in 884 women of the BARCOS cohort, included all SNPs in the two genes, except for those in the 3′ UTR. The second phase included the 3′ UTR SNPs and was performed in an extended (n = 1061) BARCOS cohort (Table 2). Polymorphism genotyping was carried out using the SNPLex System (Applied Biosystems, Carlsbad, CA, USA) at the CEGEN platform (Barcelona, Spain) in the first phase and KASPar v4.0 genotyping system at the Kbioscience facilities (Herts, England) in the second phase, using the Kraken allele-calling algorithm. Quality control was done by cross-genotyping four SNPs (∼12% of the results) using both platforms. The readings showed 99.67% concordance between the two techniques.

Table 2. List of the Genotyped Gene-Wide SNPs, Genotyping Efficiency, MAF, and p Values for Log-Additive Model
GeneSNP no.rsLocationnEfficiencyMAF BARCOSHWELSFNFractures
  • MAF = minor allele frequency; HWE = Hardy Weinberg equilibrium; LS = lumbar spine; FN = femoral neck.

  • Bold indicates p < 0.05.

  • a


  • b

    Because of low MAF for rs9562415, rs956700, rs8092336, and 78622775, the only available statistical model was the codominant model. For the rest of the SNPs, in case of lower significant p value in other model rather than log-additive, the results are given in parentheses.

  • c


  • d


  • e


  • f


RANKL1rs9594738184 kb upstream10610.970.440.803.4 × 10−40.070.56
        (1.5 × 10−4)a(0.02)a 
 2rs9594759104 kb upstream8840.870.490.150.130.810.37
 3rs17639305Proximal promoter8840.920.170.380.630.860.39
 4rs9533156Proximal promoter8840.890.460.340.530.970.16
 5rs9525641Proximal promoter8840.910.450.240.640.960.18
 6rs2296533Exon 18840.760.460.210.720.610.24
 7rs9594782Intron 18840.920.070.390.860.820.49
 8rs12427596Intron 18840.910.430.490.940.770.10
 9rs9525642Intron 18840.920.380.990.870.970.03
 10rs9533158Intron 18840.860.160.840.640.990.48
 11rs9533159Intron 18840.900.430.240.850.670.10
 12rs2277438Intron 18840.920.170.500.770.870.30
 13rs9562415bExon 510610.950.020.550.650.710.28
 14rs9567000b3′ UTR10610.970.020.570.770.500.08
 15rs10540163′ UTR8840.870.400.560.730.740.15
RANK1rs6567265Proximal promoter8840.830.300.380.660.850.25
 2rs7233419Proximal promoter8840.900.300.170.690.970.77
 3rs12457042Intron 18840.920.060.560.090.660.29
 4rs11152341Intron 18840.890.240.300.14 (0.037)c0.850.50
 5rs7233197Intron 18840.930.070.540.100.710.07
 6rs4941125Intron 18840.920.290.200.320.360.72
 7rs4941126Intron 18840.890.280.320.360.400.83
 8rs12150741Intron 110610.970.240.130.27 (0.029)c0.590.49 (0.035)d
 9rs35211496Exon 4e8840.830.220.930.550.450.48
 10rs1805034Exon 6f10610.970.420.440.730.570.15 (0.049)a
 11rs8092336bExon 98840.930.040.220.860.630.90
 12rs78622775b3′ UTR10610.980.010.730.300.340.99
 13rs124553233′ UTR10610.970.320.890.790.920.86
 14rs729336403′ UTR10610.970.130.0060.190.930.58
 15rs783264033′ UTR10610.970.080.280.770.430.05 (0.021)c
 16rs784599453′ UTR10610.970.080.400.970.340.18 (0.035)c
 17rs729336413′ UTR10610.970.
 18rs8842053′ UTR10610.970.190.360.590.690.07 (0.0087)d

Statistical methods

Hardy-Weinberg equilibrium (HWE) was calculated using the chi-square test. HWE p values were calculated using the Tufts University web site (∼mcourt01/Documents/Court%20lab%20-%20HW%20calculator.xls). Multivariate linear or logistic regression models were fitted to assess the association between genotyped SNPs and BMD or fractures, respectively. Potential confounders considered for adjustment were body mass index (BMI), age at menarche, years since menopause at the time of densitometry, and months of breastfeeding for the models where BMD was the outcome, and BMI[27] and age for fractures. All analyses were two-tailed, and p values < 0.05 were considered significant.

To calculate the new p target, we used the Cheverud approach,[28] modified by Li and Ji.[29] According to this approach, the effective number of independent markers was 24.002 and the effective significance threshold associated to keep type I error rate at 5% was 2.13 × 10−3 (new p target).

In the luciferase assays, pairwise statistical comparisons between constructs (with and without additional treatments) were calculated using the nonparametric Wilcoxon paired-sample test.

Statistical analyses were performed using SPSS for Windows version 13.0 (SPSS, Inc., Chicago, IL, USA) and R software version 2.13.2 with the SNPassoc, foreign, rms, epicalc, genetics, gdata, and LDcorSV packages.

DNA constructs

To generate a 2180-bp length promoter, the human RANKL region (ENSG00000120659; TNFSF11-001) comprising –2084/ + 96 was PCR amplified and cloned by blunt-end ligation into pUC19 SmaI (Fermentas), using the following primers: forward 5′-CCTGTGAAACAGCAGCAG-3′; reverse 5′-TCTTGTCTGCGGCCAACT-3′. The insert was excised with KpnI and BamHI and subsequently subcloned into the pGL3-Basic vector digested with KpnI and HindIII at the polylinker site. In parallel, a 999 bp fragment containing rs9594738 in position 470 of the segment (NCBI reference sequence: NT_024524.14) was amplified by PCR, using the following primers: forward 5′-TGTAAATTGTGATGATGTGAACG-3′; reverse 5′-TCACCAACTAGTGCCCATGA-3′. The fragment was cloned by blunt-end ligation into pUC19 SmaI. To integrate this fragment into the pGL3-RANKL promoter structure, both constructs were double digested: a KpnI - Bstz17I segment was eliminated from the 2180 bp length promoter to obtain the long promoter (LP) construct (2015 bp length, -1919/ + 96), and the 999-bp fragment was excised from pUC19 with KpnI and EcoRV to achieve the 831-bp fragment named DR. This DR was subcloned into the pGL3-Basic vector upstream of LP. The DR contains 6 SNPs, polymorphic in the white population (rs10507506, rs12871509, rs9594738, rs185241354, rs75341477, and rs17457484), and the ancestral allele was cloned for all of them (C, A, C, T, T and C, respectively) to generate the most frequent haplotype. To achieve the short promoter (SP) construct and in parallel to subclone DR upstream of this short promoter construct, the pGL3-RANKL promoter construct was double digested with KpnI and PvuII (Fig. 1A). To generate the SP_pUC construct, a 920-bp segment was purified from a KpnI and ScaI double digestion of pUC19 and subcloned into KpnI- and PvuII-digested pGL3-RANKL promoter construct. To generate a 2388-bp OPG promoter fused to the luciferase gene, the human gene region (NCBI reference sequence: NG_012202.1) comprising –2078/ + 310 was PCR-amplified and cloned by blunt-end ligation into pUC19. Primers used were: forward 5′-GTGCCCCAACCTGTCTCC-3′; reverse 5′-AACCTCAGGGGCTTGGAG-3′. The insert was excised with SacI and NheI and subcloned into the pGL3-Basic vector digested with the same enzymes at the polylinker site. A final digestion with KpnI and PvuII yielded the OPG short promoter (OPG_SP) –131/ + 310. The OPG_SP_DR promoter was generated by adding DR to OPG_SP (Fig. 1B). All constructs were verified by automatic sequencing.


Figure 1. Promoter constructs designed for gene reporter assays. (A) RANKL promoter constructs LP and SP, each cloned with and without DR. (B) Control constructs designed to characterize the DR effect: OPG short promoter with and without DR and RANKL SP with an additional 920-bp segment from pUC19.

Download figure to PowerPoint

Gene reporter assays and cell treatments

U2OS human osteosarcoma cells were grown in Dulbecco's modified Eagle medium (DMEM, Gibco-BRL, Paisley, Scotland, UK) supplemented with 10% fetal bovine serum (FBS, Biological Industries, Kibbutz Beit Haemek, Israel) and ascorbic acid 100 µg/mL (Sigma-Aldrich, St. Louis, MO, USA). We plated 8 to 12 × 104 cells per well in supplemented DMEM. At 60% to 80% confluence, 2 to 3 µg of each construct and 2 ng of Renilla control vector were co-transfected into the cultured cells using Lipofectamine LTX and Plus reagent (Invitrogen, Carlsbad, CA, USA). At 24 hours after transfection, firefly and Renilla luciferase activities were measured in an OrionII microplate luminometer (Berthold Detection Systems, Pforzheim, Germany) using Dual Luciferase Reporter Assay (Promega, Madison, WI, USA). In the case of FBS free-medium tests, the medium was changed from 10% FBS to DMEM containing 0.1% bovine serum albumin (BSA, Sigma Aldrich) 4 hours after transfection, and firefly and Renilla luciferase activities were measured 24 hours post-transfection. In the case of an additional treatment with hormones, transfection was performed in 10% FBS medium followed by up to 24 hours of incubation. The medium was changed to DMEM containing 0.1% BSA, and after 2 hours treatments were added. Final concentration for each treatment was of 100 nM for the hormones (vitamin D [Sigma-Aldrich], 17β-estradiol [Sigma-Aldrich], and parathyroid hormone [PTH, Sigma-Aldrich]). Firefly and Renilla luciferase activities were measured at 16 hours of treatment. To estimate the effect, each treated construct was compared with the equivalent nontreated one.

For each assayed construct or treatment, the number of transfection experiments (replicas) is given in the corresponding figure. Transfection replicas were performed with a number of different sets of minipreps or midipreps (Qiagen) and were tested in duplicate or triplicate, as appropriate. Although this strategy generated more inter-replica variability, it avoided any bias attributable to single clones of each construct. The results for RANKL promoter were normalized in reference to the LP construct. For the OPG experiments, the results were normalized in reference to the OPG_SP construct.


  1. Top of page
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

Association of RANK and RANKL SNPs with bone mineral density and fracture

In this study, the BARCOS cohort involved 1061 female patients, all of Spanish ancestry. Age, weight, height, age at menarche, age at menopause, years since menopause at the time of densitometry, months of breastfeeding, and history of prior fractures are shown in Table 1.

Twenty SNPs were found in evolutionary conserved regions encompassing the proximal promoter and intron 1 of the RANK and RANKL genes. Two additional SNPs previously associated with BMD,[6] located upstream of the RANKL gene, were also included for replication. In addition, 11 exonic SNPs (3 in RANKL and 8 in RANK) and 9 SNPs in the 3′ UTR with MAF >0.01 (2 in RANKL and 7 in RANK) were chosen. Of these, 17 SNPs lacking published MAF were validated in our facilities, and 9 were not polymorphic in our cohort (rs9533157, rs35443351, rs35858555, rs35114461, rs34151971, rs12721430, rs34945627, rs35184120, and rs35993683).

In total, 33 genetic variants in the RANK and RANKL genes were genotyped in the BARCOS cohort (Fig. 2 and Table 2). All the SNPs except for rs72933640 were in HWE. Moreover, the MAF for rs72933640 in BARCOS (0.129) was similar to the MAF published by the National Center for Biotechnology Information (NCBI) for the CEU population (0.108), which suggests the absence of genotyping error or stratification. MAFs of all genotyped SNPs were ≥0.01 in our population.


Figure 2. Genotyped SNPs in the RANK and RANKL genes. Each SNP is represented by its corresponding number in Table 2. For each gene, its chromosomal location, the HapMap haplotypic blocks (Haploview software), and the evolutionary conserved profile (UCSC genome browser) are shown. Note that RANKL SNPs rs9594738 and rs9594759 (numbered 1 and 2 in Table 2) do not appear in this figure because of their far-upstream position.

Download figure to PowerPoint

For the RANKL gene, only SNP rs9594738 was significantly associated with LS BMD (Log additive model: beta coefficient = –0.021, p = 3.4 × 10−4; dominant model: beta coefficient = –0.034, p = 1.5 × 10−4; Table 2) and with FN BMD (Log additive model: beta coefficient = –0.008, p = 0.07; dominant model: beta coefficient = –0.015, p = 0.02]. Although this SNP replicates a previously reported BMD association study and therefore correction for multiple testing is not mandatory, the LS BMD result withstood the multiple-test correction.

No significant association with fractures was found for any of the RANKL SNPs studied, except for rs9525642 (Log additive model: odds ratio [OR] = 0.70, 95% confidence interval [CI] 0.51–0.97, p = 0.03; Table 2). However, this result did not withstand multiple-test correction.

For the RANK gene, SNPs rs11152341 and rs12150741, both in intron 1, were found to be associated with LS BMD in our cohort (overdominant model: beta coefficient = –0.021, p = 0.037 and overdominant model: beta coefficient = –0.019, p = 0.029, respectively). SNP rs12150741 and rs1805034 showed a trend to association with fracture risk. SNPs rs78326403, rs78459945, and rs884205, all in the 3′ region of the gene, were significantly associated with fracture prevalence (overdominant model: OR = 1.84, 95% CI 1.11–3.04, p = 0.021; overdominant model: OR = 1.77, 95% CI 1.06–2.96, p = 0.035; and recessive model: OR = 3.11, 95% CI 1.41–6.90, p = 8.7 × 10−3, respectively) (Table 2). SNPs rs78326403 and rs78459945 were found to be in linkage disequilibrium (LD) (D' = 0.999, R2 = 0.968). However, none of these associations withstood multiple-test correction (target p value = 2.13 × 10−3).

In silico study of the genomic region containing rs9594738

The white population haplotype structure across the intergenic region between AKAP11 and RANKL genes retrieved from HapMap Genome Browser (release #28) using the Haploview software version 4.2[30] (Fig. 3A) showed that rs9594738 (SNP 1 in Fig. 3A) belongs to a haplotypic block smaller than 1 kb, containing only 3 tag-SNPs. This block can be integrated into the larger (114 kb) neighboring telomeric block, which contains rs9594759 (SNP 2 in Fig. 3A). However, it is independent from the haplotypic block containing the RANKL gene and its proximal promoter. This architecture suggests an independent heritability of SNP rs9594738 and the RANKL gene. Given that rs9594759, located closer to the gene than rs9594738, is not significantly associated with BMD in our cohort, a putative functional region may be limited to the region upstream of rs9594759.


Figure 3. AKAP11-RANKL intergenic region containing SNPs rs9594738 (marked as 1), rs9594759 (marked as 2), and rs9533090. (A) Full genomic view. The black arrow marks the inset showing and enlarged view of the small haplotypic block containing rs9594738 (Haploview software). (B) A zoom-in on the region surrounding SNPs rs9594738 and rs9533090 (UCSC genome browser) showing H3K27Ac mark, chromatin state, DNaseI hypersensitivity, transcription factor ChIP-seq, and mammalian conservation. The 831-bp fragment functionally tested in this study (named DR) is depicted with a thick black line. A Genomatix MatInspector-predicted VDRE is indicated by a small black box.

Download figure to PowerPoint

An analysis of the 1-kb region surrounding the rs9594738 in the open UCSC genome browser (, taking into consideration the ENCODE data annotated there,[31] identified elements consistent with active chromatin, such as DNase hypersensitive areas, acetylation of H3K27, several transcription factor binding sites (Fig. 3B), and other enhancer- and promoter-associated histone marks (H3K4Me1 and H3K4Me3; not shown). Interestingly, an increase in the scores of evolutionary conservation was observed next to the SNP (Fig. 3B).

Functional analyses of RANKL proximal promoter and far-upstream sequences

Reporter gene assays of constructs containing RANKL promoter sequences up to –1919 bp (LP) or up to –234 bp (SP) transfected into U2OS osteosarcoma cells are presented in Fig. 4. The region upstream of the short promoter was found to contain regulatory elements that strongly reduced expression levels (mean: 4.87-fold, SP versus LP, Fig. 4A). A distal 831-bp region (DR), located about 184 kb upstream of RANKL and containing SNP rs9594738 in a central position, was linked to the 5′ end of the LP and SP promoter sequences to achieve constructs LP_DR and SP_DR (Fig. 1A), and these constructs were tested for luciferase activity. In cells cultured in FBS-free medium, the DR significantly inhibited both promoter constructs, with a large inhibitory effect on SP (SP versus SP_DR, mean: 4.62-fold) and a small effect on LP (LP versus LP_DR, mean: 1.19-fold) (Fig. 4A). In contrast, in cells cultured with 10% FBS, the DR effect observed was 2.37-fold on SP activity and nonsignificant on LP (Fig. 4B). To test the specificity of its inhibitory effect on the RANKL promoter, the DR region was also cloned upstream of the OPG short promoter (Fig. 1B). In this case, no regulatory effect was observed (Fig. 4C). To rule out that the DR effect on the RANKL short promoter might be the result of cloning any DNA upstream of the RANKL short promoter, we prepared a 920-bp segment derived from the pUC19 vector and cloned it upstream of SP (Fig. 1B). This nonspecific region failed to inhibit the reporter expression levels of the SP construct (Fig. 4D).


Figure 4. Reporter gene assays. Results are means (±SD). For A, B, and D, each construct was compared with LP, which was arbitrarily set at 1. (A) The RANKL LP and SP promoter constructs, with and without DR, were tested after overnight incubation in DMEM + 0.1% BSA. The number of replicates was 20. (B) The RANKL LP and SP promoter constructs, with and without DR, were tested after overnight incubation in DMEM + 10% FBS. The number of replicates was 9. (C) The OPG basal promoter, with and without the DR segment, tested after overnight incubation in DMEM + 10% FBS. The number of replicates was 5. OPG_SP_DR was compared with OPG_SP, which was arbitrarily set at 1. (D) SP construct with an additional 920-bp segment from pUC19 compared with SP, tested after overnight incubation in DMEM + 10% FBS. The number of replicates was 6. **p < 0.01, Wilcoxon test.

Download figure to PowerPoint

Effect of vitamin D treatment on the DR and proximal RANKL promoter activities

With the aim of testing whether the VDRE identified by Ramagopalan and colleagues[18] lies in the DR, cell cultures transfected with LP, LP_DR, SP, and SP_DR were treated with vitamin D. Each treated construct was compared with the corresponding nontreated one (Fig. 5). Vitamin D significantly stimulated luciferase expression of all the tested constructs, except the SP. Therefore, the significant effect of vitamin D on the SP_DR construct is to be attributed to the DR region. Both LP and LP_DR were stimulated by vitamin D with no differences between the two. The results are consistent with the presence of VDREs both in the LP and the DR sequences. Other hormones that are well-known regulators of the RANKL/OPG system (PTH, 17β-estradiol) were tested to investigate the presence of additional response elements in the DR. None of them seemed to act on the DR (data not shown).


Figure 5. Reporter gene assay to test the effect of vitamin D. Luciferase expression values are relative to the nontreated LP construct, which was arbitrarily set at 1. Results are means (±SD) of luciferase activity. Each treated construct (gray bars) was compared with the corresponding nontreated one (black bars). The number of replicates was 12. **p < 0.01, Wilcoxon test.

Download figure to PowerPoint


  1. Top of page
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

In recent years, many GWAs have identified relationships between pathologic situations and genetic variants, but the majority of the associations were not further studied and their functional and biological effect remained unknown. In this study, we performed an in-depth analysis of genetic variations of RANK and RANKL, genes associated with bone phenotypes.[5-11] Putative functional SNPs in the RANK and RANKL genes, as well as genome-wide associated SNPs, were genotyped to evaluate their association with BMD and osteoporotic fractures. Only one SNP, rs9594738, which had been previously associated with BMD,[6, 10] was found to be significantly associated with a bone phenotype, lumbar spine BMD, in our cohort after multiple-test correction. This SNP is located in an upstream distal region (DR) of RANKL, and we present here the first functional study of this region in human cells. Our results demonstrate that DR has the ability to regulate RANKL transcription and also that it responds to vitamin D treatment.

RANKL does not contain missense SNPs with MAF above 5%. We chose SNPs located in evolutionary conserved regions (in mammalian species), within the proximal promoter and the transcription unit, and they failed to show association with bone phenotypes at the significance level of this study. It is quite surprising that only an SNP at a distance from the gene (rs9594738) was positively associated, and none within the gene itself. However, this result is consistent with the GWA meta-analysis by Estrada and colleagues [9] (see Supplemental Fig. 3BR in this reference), in which all the positive signals are located far upstream of the gene and none within the gene. Interestingly, there is a recombination hotspot between the positive signals and the gene (see Fig. 3A and Supplemental Fig. 3BR in Estrada and colleagues[9]). In this context, an important role for rare RANKL variants cannot be ruled out. With this in mind, we have explored available rare variants in the HGMD (Professional 2012.4 version, December 14, 2012) and the 1000 genomes (release 13, December 2012) databases. In the first database, three RANKL variants were listed, which are the cause of autosomal recessive osteopetrosis; in the other one, up to 83 different rare missense variants were found, for which no phenotypic data were available. Whether any of these many missense variants are relevant for osteoporosis remains an open question.

We have also analyzed several RANK SNPs, including two missense SNPs with high MAF, p.H141Y (rs35211496, MAF in BARCOS = 0.22) and p.A192V (rs1805034, MAF in BARCOS = 0.42). Only a borderline association was observed between rs1805034 and fracture risk, under a dominant model. This result is consistent with that of Zhang and colleagues,[32] who report a significant protective effect for the valine allele in relation to osteoporotic hip fracture in a group of elderly Chinese women. Browsing HGMD and 1000 genome databases for rare functional variants yielded several missense and nonsense changes involved in osteopetrosis and Paget's disease, as well as 74 missense variants for which no phenotypic data were available. Again, whether these are relevant for osteoporosis remains an open question. As we did for RANKL, we also chose SNPs located in evolutionary conserved regions within the promoter and intron 1 of RANK, and these failed to show association with any bone phenotype. Finally, we have analyzed several 3′ UTR SNPs, and three of them, rs78326403, rs78459945, and rs884205, were nominally associated with fracture risk, although this association did not withstand the new p target. SNP rs884205, previously related to osteoporotic phenotypes with GW significance,[8, 9, 33] also showed a trend to association with LS BMD in the adjusted recessive model (p = 0.06) in our study (not shown), which suggests an important role for this genetic variant or others in the 3′ UTR of the RANK gene. This latter hypothesis is supported by the highly significant association with LS-BMD observed in Estrada and colleagues[9] (see Supplemental Fig. 3BW of this reference) for several SNPs in this region that are not in LD among them.

Several factors suggest that the region containing rs9594738, a GWA-associated SNP and the only one with a significant association with BMD in the present study, is a functional one. First, its association with bone phenotypes cannot be explained by LD with functional SNPs in the RANKL promoter or gene because they belong to different haplotypic blocks (Fig. 3A). Second, it lies between AKAP11 and RANKL, genes that are located in the same relative position across evolution.[13] And finally, ENCODE information for the 1-kb region surrounding the rs9594738 shows the presence of elements that are consistent with active chromatin (Fig. 3B), such as DNase hypersensitive areas, histone acetylation, and several transcription factor binding sites.

Our work has, therefore, focused on the transcriptional relevance of an 831-bp region surrounding the SNP rs9594738 (DR) on RANKL expression regulation. First, gene reporter assays were performed with two different RANKL promoter segments (LP and SP), with and without DR. The results support the capability of DR to modulate both RANKL promoter structures in the U2OS cell line in FBS-free medium, with a major effect on SP. In 10% FBS medium, the DR inhibitory effect on SP decreased, suggesting that some FBS component might have a stimulatory effect on DR. The same effect was observed comparing LP and SP constructs. We were able to show that vitamin D may be such a component because it stimulated the LP, LP_DR, and SP_DR constructs but not the SP. Our results agree with in vitro studies demonstrating that vitamin D induces RANKL expression in osteoblast linage cells[34, 35] and with the functional VDRE sites defined in the human RANKL promoter at position –1570 to –1584[36] (corresponding to LP in our study). In addition, our results suggest that the VDRE identified in a recent ChIP-seq assay[18] falls in the DR, and supportive evidence is supplied by Matinspector of Genomatix,[37] which predicts a VDR/RXR binding site within the DR, 140 bp downstream of rs9594738.

Other studies have proposed a functional role for the AKAP11-RANKL intergenic region. Kim and colleagues[15, 16] identified multiple VDR/RXR interacting sequences in this region in mice. One of them was conserved in humans and shown to be functional. The authors hypothesized that a chromatin hub centered on the Rankl promoter allows the distant enhancers to act and regulate the gene expression. In this regard, Galli and colleagues[17] demonstrated that the deletion of a 2.3-kb segment from this region upstream of Rankl in mice, which they named DCR, reduced Rankl expression, yet did not affect the Akap11 expression that was used as a control.

Our study has several limitations. The relatively small sample (1061 women) limits the statistical power of the associations observed and, therefore, our ability to identify novel associations. Further studies in additional cohorts are needed to determine whether some of the associations we report are replicated. However, the SNP we found in association with BMD replicates previous findings and withstands the multiple-test correction. Therefore, its association is robust. Regarding the functional studies, although the artificial conditions of the experiments preclude a straightforward extrapolation to the physiological environment in human bone tissue, they provide a first approach to show functionality for the DR segment. Further in vitro experiments, closer to the physiological conditions, can be done by cloning larger or even the entire human intergenic region between RANKL and AKAP11, as demonstrated in mice.[38] Plasma levels of vitamin D were not available for most of our cases and, therefore, the possible effect of vitamin D on the association results could not be assessed. We also addressed the functionality of SNP rs9594738. Preliminary experiments failed to yield conclusive results. Further studies on this SNP and on five additional SNPs within the DR are in progress.

In conclusion, to the best of our knowledge, this is the first functional analysis in human cells of the RANKL distal region harboring rs9594738, a genetic variant associated with BMD in this and previous studies. The experimental findings in this study, combined with others, confirm the existence of a RANKL distal regulatory region between AKAP11 and RANKL that is stimulated by the presence of vitamin D, suggesting that it may play an important role in the RANK/RANKL/OPG equilibrium.


  1. Top of page
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References

This work was supported by grants from the Generalitat de Catalunya (DIUE; 2009 SGR 818, 2009 SGR 971) and the Red Temática de Investigación Cooperativa en Envejecimiento y Fragilidad (RETICEF) and FEDER funds. Grant FIS PI 060895 (Carlos III Health Institute, Science and Innovation Ministry), SAF2011-25431 and PIB2010AR-00473 (Science and Innovation Ministry), and the support from the Centro de Investigación Biomédica en Red de Enfermedades Raras (CIBERER, an initiative of ISCIII) are also acknowledged. The authors thank Elaine M Lilly, PhD, for helpful advice and critical reading of the manuscript.

Authors' roles: Study design: GY, NGG, DG, XN, SB, ADP. Data collection and procedures: GY, NGG, MRS, RU, RG, SAB, LM, XN. Data analysis and interpretation: GY, NGG, DPA, DG, MRS, SB, ADP. Drafting of the manuscript: GY, NGG, DPA, DG, SB, ADP. Critical review and approval: All the authors.


  1. Top of page
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgments
  9. References
  • 1
    Cao X. Targeting osteoclast-osteoblast communication. Nat Med. 2011; 17(11):13446.
  • 2
    Trouvin AP, Goeb V. Receptor activator of nuclear factor-kappaB ligand and osteoprotegerin: maintaining the balance to prevent bone loss. Clin Interv Aging. 2010; 5:34554.
  • 3
    Teitelbaum SL. Bone resorption by osteoclasts. Science. 2000 Sep 1; 289(5484):15048.
  • 4
    Liu C, Walter TS, Huang P, Zhang S, Zhu X, Wu Y, Wedderburn LR, Tang P, Owens RJ, Stuart DI, Ren J, Gao B. Structural and functional insights of RANKL-RANK interaction and signaling. J Immunol. 2010 Jun 15; 184(12):69109.
  • 5
    Paternoster L, Lorentzon M, Vandenput L, Karlsson MK, Ljunggren O, Kindmark A, Mellstrom D, Kemp JP, Jarett CE, Holly JM, Sayers A, St Pourcain B, Timpson NJ, Deloukas P, Davey Smith G, Ring SM, Evans DM, Tobias JH, Ohlsson C. Genome-wide association meta-analysis of cortical bone mineral density unravels allelic heterogeneity at the RANKL locus and potential pleiotropic effects on bone. PLoS Genet. 2010 Nov; 6(11):e1001217.
  • 6
    Styrkarsdottir U, Halldorsson BV, Gretarsdottir S, Gudbjartsson DF, Walters GB, Ingvarsson T, Jonsdottir T, Saemundsdottir J, Center JR, Nguyen TV, Bagger Y, Gulcher JR, Eisman JA, Christiansen C, Sigurdsson G, Kong A, Thorsteinsdottir U, Stefansson K. Multiple genetic loci for bone mineral density and fractures. N Engl J Med. 2008 May 29; 358(22):235565.
  • 7
    Styrkarsdottir U, Halldorsson BV, Gretarsdottir S, Gudbjartsson DF, Walters GB, Ingvarsson T, Jonsdottir T, Saemundsdottir J, Snorradottir S, Center JR, Nguyen TV, Alexandersen P, Gulcher JR, Eisman JA, Christiansen C, Sigurdsson G, Kong A, Thorsteinsdottir U, Stefansson K. New sequence variants associated with bone mineral density. Nat Genet. 2009 Jan; 41(1):157.
  • 8
    Rivadeneira F, Styrkarsdottir U, Estrada K, Halldorsson BV, Hsu YH, Richards JB, Zillikens MC, Kavvoura FK, Amin N, Aulchenko YS, Cupples LA, Deloukas P, Demissie S, Grundberg E, Hofman A, Kong A, Karasik D, van Meurs JB, Oostra B, Pastinen T, Pols HA, Sigurdsson G, Soranzo N, Thorleifsson G, Thorsteinsdottir U, Williams FM, Wilson SG, Zhou Y, Ralston SH, van Duijn CM, Spector T, Kiel DP, Stefansson K, Ioannidis JP, Uitterlinden AG. Twenty bone-mineral-density loci identified by large-scale meta-analysis of genome-wide association studies. Nat Genet. 2009 Nov; 41(11):1199206.
  • 9
    Estrada K, Styrkarsdottir U, Evangelou E, Hsu YH, Duncan EL, Ntzani EE, Oei L, Albagha OM, Amin N, Kemp JP, Koller DL, Li G, Liu CT, Minster RL, Moayyeri A, Vandenput L, Willner D, Xiao SM, Yerges-Armstrong LM, Zheng HF, Alonso N, Eriksson J, Kammerer CM, Kaptoge SK, Leo PJ, Thorleifsson G, Wilson SG, Wilson JF, Aalto V, Alen M, Aragaki AK, Aspelund T, Center JR, Dailiana Z, Duggan DJ, Garcia M, Garcia-Giralt N, Giroux S, Hallmans G, Hocking LJ, Husted LB, Jameson KA, Khusainova R, Kim GS, Kooperberg C, Koromila T, Kruk M, Laaksonen M, Lacroix AZ, Lee SH, Leung PC, Lewis JR, Masi L, Mencej-Bedrac S, Nguyen TV, Nogues X, Patel MS, Prezelj J, Rose LM, Scollen S, Siggeirsdottir K, Smith AV, Svensson O, Trompet S, Trummer O, van Schoor NM, Woo J, Zhu K, Balcells S, Brandi ML, Buckley BM, Cheng S, Christiansen C, Cooper C, Dedoussis G, Ford I, Frost M, Goltzman D, Gonzalez-Macias J, Kahonen M, Karlsson M, Khusnutdinova E, Koh JM, Kollia P, Langdahl BL, Leslie WD, Lips P, Ljunggren O, Lorenc RS, Marc J, Mellstrom D, Obermayer-Pietsch B, Olmos JM, Pettersson-Kymmer U, Reid DM, Riancho JA, Ridker PM, Rousseau F, Lagboom PE, Tang NL, Urreizti R, Van Hul W, Viikari J, Zarrabeitia MT, Aulchenko YS, Castano-Betancourt M, Grundberg E, Herrera L, Ingvarsson T, Johannsdottir H, Kwan T, Li R, Luben R, Medina-Gomez C, Th Palsson S, Reppe S, Rotter JI, Sigurdsson G, van Meurs JB, Verlaan D, Williams FM, Wood AR, Zhou Y, Gautvik KM, Pastinen T, Raychaudhuri S, Cauley JA, Chasman DI, Clark GR, Cummings SR, Danoy P, Dennison EM, Eastell R, Eisman JA, Gudnason V, Hofman A, Jackson RD, Jones G, Jukema JW, Khaw KT, Lehtimaki T, Liu Y, Lorentzon M, McCloskey E, Mitchell BD, Nandakumar K, Nicholson GC, Oostra BA, Peacock M, Pols HA, Prince RL, Raitakari O, Reid IR, Robbins J, Sambrook PN, Sham PC, Shuldiner AR, Tylavsky FA, van Duijn CM, Wareham NJ, Cupples LA, Econs MJ, Evans DM, Harris TB, Kung AW, Psaty BM, Reeve J, Spector TD, Streeten EA, Zillikens MC, Thorsteinsdottir U, Ohlsson C, Karasik D, Richards JB, Brown MA, Stefansson K, Uitterlinden AG, Ralston SH, Ioannidis JP, Kiel DP, Rivadeneira F. Genome-wide meta-analysis identifies 56 bone mineral density loci and reveals 14 loci associated with risk of fracture. Nat Genet. 2012; 44(5):491501.
  • 10
    Duncan EL, Danoy P, Kemp JP, Leo PJ, McCloskey E, Nicholson GC, Eastell R, Prince RL, Eisman JA, Jones G, Sambrook PN, Reid IR, Dennison EM, Wark J, Richards JB, Uitterlinden AG, Spector TD, Esapa C, Cox RD, Brown SD, Thakker RV, Addison KA, Bradbury LA, Center JR, Cooper C, Cremin C, Estrada K, Felsenberg D, Gluer CC, Hadler J, Henry MJ, Hofman A, Kotowicz MA, Makovey J, Nguyen SC, Nguyen TV, Pasco JA, Pryce K, Reid DM, Rivadeneira F, Roux C, Stefansson K, Styrkarsdottir U, Thorleifsson G, Tichawangana R, Evans DM, Brown MA. Genome-wide association study using extreme truncate selection identifies novel genes affecting bone mineral density and fracture risk. PLoS Genet. 2011 Apr;7(4):e1001372.
  • 11
    Richards JB, Kavvoura FK, Rivadeneira F, Styrkarsdottir U, Estrada K, Halldorsson BV, Hsu YH, Zillikens MC, Wilson SG, Mullin BH, Amin N, Aulchenko YS, Cupples LA, Deloukas P, Demissie S, Hofman A, Kong A, Karasik D, van Meurs JB, Oostra BA, Pols HA, Sigurdsson G, Thorsteinsdottir U, Soranzo N, Williams FM, Zhou Y, Ralston SH, Thorleifsson G, van Duijn CM, Kiel DP, Stefansson K, Uitterlinden AG, Ioannidis JP, Spector TD. Collaborative meta-analysis: associations of 150 candidate genes with osteoporosis and osteoporotic fracture. Ann Intern Med. 2009 Oct 20; 151(8):52837.
  • 12
    The International HapMap Project. Nature. 2003 Dec 18; 426(6968):78996.
  • 13
    O'Brien CA. Control of RANKL gene expression. Bone. 2010 Apr; 46(4):9119.
  • 14
    Myers RM, Stamatoyannopoulos J, Snyder M, Dunham I, Hardison RC, Bernstein BE, Gingeras TR, Kent WJ, Birney E, Wold B, Crawford GE. A user's guide to the encyclopedia of DNA elements (ENCODE). PLoS Biol. 2011 Apr; 9(4):e1001046.
  • 15
    Kim S, Yamazaki M, Zella LA, Shevde NK, Pike JW. Activation of receptor activator of NF-kappaB ligand gene expression by 1,25-dihydroxyvitamin D3 is mediated through multiple long-range enhancers. Mol Cell Biol. 2006 Sep; 26(17):646986.
  • 16
    Kim S, Yamazaki M, Zella LA, Meyer MB, Fretz JA, Shevde NK, Pike JW. Multiple enhancer regions located at significant distances upstream of the transcriptional start site mediate RANKL gene expression in response to 1,25-dihydroxyvitamin D3. J Steroid Biochem Mol Biol. 2007 Mar; 103(3–5):4304.
  • 17
    Galli C, Zella LA, Fretz JA, Fu Q, Pike JW, Weinstein RS, Manolagas SC, O'Brien CA. Targeted deletion of a distant transcriptional enhancer of the receptor activator of nuclear factor-kappaB ligand gene reduces bone remodeling and increases bone mass. Endocrinology. 2008 Jan; 149(1):14653.
  • 18
    Ramagopalan SV, Heger A, Berlanga AJ, Maugeri NJ, Lincoln MR, Burrell A, Handunnetthi L, Handel AE, Disanto G, Orton SM, Watson CT, Morahan JM, Giovannoni G, Ponting CP, Ebers GC, Knight JC. A ChIP-seq defined genome-wide map of vitamin D receptor binding: associations with disease and evolution. Genome Res. 2010 Oct; 20(10):135260.
  • 19
    Bustamante M, Nogues X, Agueda L, Jurado S, Wesselius A, Caceres E, Carreras R, Ciria M, Mellibovsky L, Balcells S, Diez-Perez A, Grinberg D. Promoter 2- 1025 T/C polymorphism in the RUNX2 gene is associated with femoral neck bmd in Spanish postmenopausal women. Calcif Tissue Int. 2007 Oct; 81(4):32732.
  • 20
    Bustamante M, Nogues X, Mellibovsky L, Agueda L, Jurado S, Caceres E, Blanch J, Carreras R, Diez-Perez A, Grinberg D, Balcells S. Polymorphisms in the interleukin-6 receptor gene are associated with bone mineral density and body mass index in Spanish postmenopausal women. Eur J Endocrinol. 2007 Nov; 157(5):67784.
  • 21
    Genant HK, Wu CY, van Kuijk C, Nevitt MC. Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res. 1993 Sep; 8(9):113748.
  • 22
    Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 1988 Feb 11; 16(3):1215.
  • 23
    Flicek P, Amode MR, Barrell D, Beal K, Brent S, Carvalho-Silva D, Clapham P, Coates G, Fairley S, Fitzgerald S, Gil L, Gordon L, Hendrix M, Hourlier T, Johnson N, Kahari AK, Keefe D, Keenan S, Kinsella R, Komorowska M, Koscielny G, Kulesha E, Larsson P, Longden I, McLaren W, Muffato M, Overduin B, Pignatelli M, Pritchard B, Riat HS, Ritchie GR, Ruffier M, Schuster M, Sobral D, Tang YA, Taylor K, Trevanion S, Vandrovcova J, White S, Wilson M, Wilder SP, Aken BL, Birney E, Cunningham F, Dunham I, Durbin R, Fernandez-Suarez XM, Harrow J, Herrero J, Hubbard TJ, Parker A, Proctor G, Spudich G, Vogel J, Yates A, Zadissa A, Searle SM. Ensembl 2012. Nucleic Acids Res. 2012 Jan; 40(Database issue):D8490.
  • 24
    Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, Haussler D. The human genome browser at UCSC. Genome Res. 2002 Jun; 12(6):9961006.
  • 25
    Meyer LR, Zweig AS, Hinrichs AS, Karolchik D, Kuhn RM, Wong M, Sloan CA, Rosenbloom KR, Roe G, Rhead B, Raney BJ, Pohl A, Malladi VS, Li CH, Lee BT, Learned K, Kirkup V, Hsu F, Heitner S, Harte RA, Haeussler M, Guruvadoo L, Goldman M, Giardine BM, Fujita PA, Dreszer TR, Diekhans M, Cline MS, Clawson H, Barber GP, Haussler D, Kent WJ. The UCSC Genome Browser database: extensions and updates 2013. Nucleic Acids Res. 2013 Jan 1; 41(D1):D649.
  • 26
    Sherry ST, Ward MH, Kholodov M, Baker J, Phan L, Smigielski EM, Sirotkin K. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001 Jan 1; 29(1):30811.
  • 27
    Prieto-Alhambra D, Premaor MO, Fina Aviles F, Hermosilla E, Martinez-Laguna D, Carbonell-Abella C, Nogues X, Compston JE, Diez-Perez A. The association between fracture and obesity is site-dependent: a population-based study in postmenopausal women. J Bone Miner Res. 2012 Feb; 27(2):294300.
  • 28
    Cheverud JM. A simple correction for multiple comparisons in interval mapping genome scans. Heredity (Edinb). 2001 Jul; 87(Pt 1):528.
  • 29
    Li J, Ji L. Adjusting multiple testing in multilocus analyses using the eigenvalues of a correlation matrix. Heredity (Edinb). 2005 Sep; 95(3):2217.
  • 30
    Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics. 2005 Jan 15; 21(2):2635.
  • 31
    Rosenbloom KR, Sloan CA, Malladi VS, Dreszer TR, Learned K, Kirkup VM, Wong MC, Maddren M, Fang R, Heitner SG, Lee BT, Barber GP, Harte RA, Diekhans M, Long JC, Wilder SP, Zweig AS, Karolchik D, Kuhn RM, Haussler D, Kent WJ. ENCODE data in the UCSC Genome Browser: year 5 update. Nucleic Acids Res. 2013 Jan; 41(Database issue):D5663.
  • 32
    Zhang YP, Liu YZ, Guo Y, Liu XG, Xu XH, Guo YF, Chen Y, Zhang F, Pan F, Zhu XZ, Deng HW. Pathway-based association analyses identified TRAIL pathway for osteoporotic fractures. PLoS One. 2011; 6(7):e21835.
  • 33
    Guo Y, Wang JT, Liu H, Li M, Yang TL, Zhang XW, Liu YZ, Tian Q, Deng HW. Are bone mineral density loci associated with hip osteoporotic fractures? A validation study on previously reported genome-wide association loci in a Chinese population. Genet Mol Res. 2012; 11(1):20210.
  • 34
    Suda T, Takahashi F, Takahashi N. Bone effects of vitamin D—discrepancies between in vivo and in vitro studies. Arch Biochem Biophys. 2012 Jul 1; 523(1):229.
  • 35
    Kitazawa R, Kitazawa S. Vitamin D(3) augments osteoclastogenesis via vitamin D-responsive element of mouse RANKL gene promoter. Biochem Biophys Res Commun. 2002 Jan 18; 290(2):6505.
  • 36
    Kitazawa S, Kajimoto K, Kondo T, Kitazawa R. Vitamin D3 supports osteoclastogenesis via functional vitamin D response element of human RANKL gene promoter. J Cell Biochem. 2003 Jul 1; 89(4):7717.
  • 37
    Cartharius K, Frech K, Grote K, Klocke B, Haltmeier M, Klingenhoff A, Frisch M, Bayerlein M, Werner T. MatInspector and beyond: promoter analysis based on transcription factor binding sites. Bioinformatics. 2005 Jul 1; 21(13):293342.
  • 38
    Fu Q, Manolagas SC, O'Brien CA. Parathyroid hormone controls receptor activator of NF-kappaB ligand gene expression via a distant transcriptional enhancer. Mol Cell Biol. 2006 Sep; 26(17):645368.