Molecular Genetic Studies of Gene Identification for Osteoporosis: The 2009 Update

Abstract

Osteoporosis is a complex human disease that results in increased susceptibility to fragility fractures. It can be phenotypically characterized using several traits, including bone mineral density, bone size, bone strength, and bone turnover markers. The identification of gene variants that contribute to osteoporosis phenotypes, or responses to therapy, can eventually help individualize the prognosis, treatment, and prevention of fractures and their adverse outcomes. Our previously published reviews have comprehensively summarized the progress of molecular genetic studies of gene identification for osteoporosis and have covered the data available to the end of September 2007. This review represents our continuing efforts to summarize the important and representative findings published between October 2007 and November 2009. The topics covered include genetic association and linkage studies in humans, transgenic and knockout mouse models, as well as gene-expression microarray and proteomics studies. Major results are tabulated for comparison and ease of reference. Comments are made on the notable findings and representative studies for their potential influence and implications on our present understanding of the genetics of osteoporosis.

This review aims to summarize the progress of molecular genetic studies of gene identification for osteoporosis published between October 2007 and November 2009. It is intended to constitute a sequential update of our previously published review covering the available data up to September 2007. The topics covered include genetic association and linkage studies in humans, transgenic and knockout mouse models, as well as gene-expression microarray and proteomics studies.

• I. Introduction

• II. Candidate Gene Association Studies

  • A. Receptors for calciotropic hormones
  • B. Cytokines and receptors
  • C. Bone matrix proteins
  • D. Limitations and improvements
• III. Genome-Wide Linkage (GWL) Studies

  • A. Univariate linkage analyses
  • B. Bivariate linkage analyses
• IV. Genome-Wide Association (GWA) Studies

  • A. Single nucleotide polymorphism (SNP) analyses
  • B. Copy number variation (CNV) analyses

• V. Transgenic/Knockout Mouse Models

• VI. Gene-Expression Microarray Studies

  • A. Regulation of osteoblast and osteoclast activity
  • B. Proliferation and differentiation of mesenchymal stem cells (MSCs)
  • C. Gene expression in healthy vs. diseased tissues
  • D. Effects of therapeutic agents on the healing of fractures
  • E. Endocrine regulation of bone remodeling
• VII. Proteomics Studies

  • A. Studying cultured cells using in vitro systems
  • B. Studying fresh cells, serum, or tissues
• VIII. Future Directions

  • A. Functional studies
  • B. Epigenetic variation
  • C. New phenotypes

• IX. Future Prospects for the Application of Genetic Risk Assessment in Osteoporosis Prediction and Treatment

• X. Summary

I. Introduction

Osteoporosis is a systemic skeletal disease characterized by low bone mineral density (BMD) and microarchitectural deterioration of bone tissue, with a consequent increase in susceptibility to fracture. Molecular genetic studies have been extensively performed to search for genes underlying osteoporosis. Since 2002, we have regularly published reviews that summarize the progress in molecular genetic studies of gene identification for osteoporosis (1,2,3). In this review, we continue these reviews with an update that captures the important and representative findings published from October 2007 to November 2009.

Similar to our previous updates (1,2,3), this article systematically reviews publications relevant to osteogenesis and osteoporosis that involve genetic association and linkage studies or functional genomics (including gene-expression microarray and proteomics) in human populations and transgenic and knockout animal models. The data presented in this review were collected from PubMed using the searching key words “BMD,” “osteoporosis,” or “bone” in combination with “association,” “polymorphisms,” “linkage,” “knock out,” “transgenic,” “microarray,” or “proteomics.” The results of important studies are incorporated in tables for clear comparison and ease of reference. Table 1 summarizes the major candidate genes subjected to association studies, classified by their functional relevance to bone and mineral metabolism. Table 2 reviews major results from approximately 120 reported candidate gene association studies. Genome-wide linkage (GWL) and genome-wide association (GWA) studies for osteoporosis-related phenotypes in humans are summarized in Tables 3 and 4, respectively. Table 5 highlights studies using transgenic and knockout mouse models relevant to osteoporosis, and Table 6 reviews approximately 110 gene-expression microarray studies on the pathogenesis of osteoporosis and other bone-related diseases. In this review, due to space limitations, we only comment on the most representative results that have had an immediate influence on our understanding and research of genetic mechanisms underlying osteoporosis.

Table 1.

Major candidate genes tested for association with osteoporosis-related phenotypes

Candidate genes	Protein	Chromosome location
Calciotropic hormones and receptors
AR	Androgen receptor	Xq11.2-q12
CASR	Calcium-sensing receptor	3q13
CRHR1	Corticotropin-releasing hormone receptor 1	17q12-q22
CTR	Calcitonin receptor	7q21.3
CYP17A1	Steroid 17-α-hydroxylase	10q24.3
CYP19	Cytochrome P450, family XIX	10q26
CYP19A1	Aromatase	15q21.1
CYP1B1	Aryl hydrocarbon hydroxylase	2p21
DBP	Vitamin D-binding protein	4q12-q13
ER-a	Estrogen receptor-a	6q25.1
ER-b	Estrogen receptor-b	14q23.2
ESRRA	Estrogen-related receptor α	11q13
GR	Glucocorticoid receptor	5q31.3
LHB	Luteinizing hormone β polypeptide	19q13.32
LHCGR	Luteinizing hormone-choriogonadotropin receptor	2p21
PRL	Prolactin	6p22.2-p21.3
PTH	Parathyroid hormone	11p15.3-p15.1
PTHLH	Parathyroid hormone-like hormone	12p12.1-p11.2
PTHR1	Parathyroid hormone receptor 1	3p22-p21.1
PTHR2	Parathyroid hormone receptor 2	2q33
SHBG	Sex hormone-binding globulin	17p13-p12
SRD5A2	Steroid-5-α-reductase, α polypeptide 2	2p23
TSHR	Thyroid-stimulating hormone receptor	14q31
VDR	Vitamin-D receptor	12q13.11
Cytokines, growth factors, and receptors
BMP2	Bone morphogenetic protein 2	20p12
BMP7	Bone morphogenetic protein 7	20q13
CD40	Tumor necrosis factor receptor superfamily member 5	20q12-q13.2
CNR2	Cannabinoid receptor 2	1p36.11
DKK2	Dickkopf homolog 2	4q25
FGFR1	Fibroblast growth factor receptor 1	8p11.2-p11.1
FGFR2	Fibroblast growth factor receptor 2	10q26
GDF5	Growth differentiation factor 5	20q11.2
GHRH	Growth hormone-releasing hormone	20q11.2
IGFBP2	Insulin-like growth factor binding protein 2	2q33-q34
IL-23	Interleukin 23	12q13.2
IL-23R	Interleukin 23 receptor	1p31.3
IL-6	Interleukin 6	7p21
IL-6R	Interleukin 6 receptor	1q21
IL-15	Interleukin 15	4q31
LEPR	Leptin receptor	1p31
LRP1	Low-density lipoprotein receptor-related protein 1	12q13-q14
LRP5	Low-density lipoprotein receptor-related protein 5	11q13.4
LRP6	Low-density lipoprotein receptor-related protein 6	12p11-p13
LTBP2	Latent transforming growth factor β binding protein 2	14q24
MSTN	Myostatin	2q32.2
NPY	Neuropeptide Y	7p15.1
OPG	Osteoprotegerin	8q24
RANK	Receptor activator of nuclear factor κ-β	18q22.1
RANKL	Receptor activator of nuclear factor κ-β ligand	13q14
TGF-b1	Transforming growth factor b-1	19q13.1
TNF-a	Tumor necrosis factor a	6p21.3
TNFRSF1B	Tumor necrosis factor receptor superfamily, member 1B	1p36.3-p36.2
TNF-β	Lymphotoxin α	6p21.3
VEGF	Vascular endothelial growth factor A	6p12
Bone matrix proteins
COL1A1	Collagen type I α-1	17q21.33
COL1A2	Collagen type I α-2	7q22.1
ITGA1	Integrin, α 1	5q11.2
MMP2	Matrix metallopeptidase 2	16q13-q21
OCIL	Osteoclast inhibitory lectin	12p13
SPARC	Osteonectin	5q31.3-q32
(Continued)
Candidate genes	Protein	Chromosome location
Miscellaneous
ADCY10	Adenylate cyclase 10	1q24
ALOX15	Arachidonate 15-lipoxygenase	17p13.3
ALOX5	Arachidonate 5-lipoxygenase	10q11.2
ALPL	Alkaline phosphatase	1p36.12
ANXA6	Annexin A6	5q32-q34
APC	Adenomatous polyposis coli	5q21-q22
ARHGEF3	Rho guanine nucleotide exchange factor 3	3p21-p13
BMPR1B	Bone morphogenetic protein receptor, type IB	4q22-q24
CA10	Carbonic anhydrase X	17q21.33
CA8	Carbonic anhydrase VIII	8q11-q12
CALM1	Calmodulin 1	14q24-q31
CLCN7	Chloride channel 7	16p13
COMT	Catechol-O-methyltransferase	22q11.21
CRTAP	Cartilage-associated protein	3p22.3
DMP1	Dentin matrix acidic phosphoprotein 1	4q21
ENPP1	Ectonucleotide pyrophosphatase/phosphodiesterase 1	6q22-q23
FABP3	Fatty acid-binding protein 3	1p33-p32
FLNB	Filamin β	3p14.3
FLT1	Fms-related tyrosine kinase 1	13q12
FOXC2	Forkhead box C2	16q22-q24
FZD1	Frizzled homolog 1	7q21
FZD6	Frizzled homolog 6	8q22.3-q23.1
HMGA2	High mobility group AT-hook 2	12q15
HOXA	Homeobox A cluster	7p15-p14
HSD11B1	Hydroxysteroid (11-β) dehydrogenase 1	1q32-q41
MDR1	Multidrug resistance 1	7q21.1
MTHFR	Methylenetetrahydrofolate reductase	1p36.3
NFATC1	Nuclear factor of activated T-cells, cytoplasmic, calcineurin-dependent 1	18q23
NOG	Noggin	17q21-q22
NOS3	Endothelial nitric oxidase synthase	7q36
NR1I3	Constitutive androstane receptor	1q23.3
P2X 7	Purinergic receptor P2X, ligand-gated ion channel, 7	12q24
PBX1	Pre-B-cell leukemia homeobox 1	1q23
PIR	Pirin	Xp22.2
PLOD	Procollagen-lysine 1, 2-oxoglutarate 5-dioxygenase 1	1p36.22
PPAR-g	Peroxisome proliferator-activated receptor γ	3p25
PTN	Pleiotrophin	7q33-q34
RIZ1	Retinoblastoma protein-interacting zinc finger protein	1p36.21
ROR2	Receptor tyrosine kinase-like orphan receptor 2	9q22
RUNX2	Runt-related transcription factor 2	6p21
SFRP1	Secreted frizzled-related protein 1	8p12-p11.1
SFRP2	Secreted frizzled-related protein 2	4q31.3
SOST	Sclerostin	17q11.2
SREBF1	Sterol regulatory element binding transcription factor 1	17p11.2
THSD4	Thrombospondin, type I, domain containing 4	15q23
THSD7	Thrombospondin, type I, domain containing 7A	7p21.3
TIMP2	Tissue inhibitor of metalloproteinase 2	17q25
TWIST1	Twist homolog 1	7p21.2
WISP3	WNT1 inducible signaling pathway protein 3	6q21
WNT10B	Wingless-type MMTV integration site family, member 10B	12q13
WNT3A	Wingless-type MMTV integration site family, member 3A	1q42
WNT7B	Wingless-type MMTV integration site family, member 7B	22q13

Adapted from Y. Guo et al.: Expert Rev Endocrinol Metab 3:223–267, 2008 (3), with permission from Expert Reviews Ltd. MMTV, Mouse mammary tumor virus.

Table 2.

Association studies for osteoporosis-related phenotypes in humans (published between October 2007 and November 2009)

Gene	Marker locus	Population characteristics	Phenotype	P value	Ref.
ADCY10	16 SNPs	1,692 premenopausal Caucasian women and 715 Caucasian men	BMD at LS	<0.01	36
ALPL	3 SNPs	360 postmenopausal Hungarian women	BMD at LS, total hip and distal radius, non-vertebral fracture	NS	137
ALOX5	rs7084793, rs378090, rs3802548	1,688 premenopausal European-American sisters, 512 premenopausal African-American sisters, and 715 European-American brothers	BMD at LS and FN	≤0.05	138
ALOX15	rs2619112, rs916055	942 southern Chinese women, with either low or high BMD	High BMD in premenopausal women and low BMD in postmenopausal women	<0.05	139
	rs7220870 (G48924T)	6,752 Caucasian women recruited at four U.S. clinical centers and enrolled into the Study of Osteoporotic Fractures	Hip fracture	<0.05	12
ANXA6	rs9324679, rs9324677, rs10037814, rs11960458, and haplotypes	443 Korean patients with osteonecrosis of the femoral head and 273 Korean control	Osteonecrosis of the femoral head	<0.05	140
APC	rs459552, rs4705573, rs6594646	862 Caucasian men aged ≥65 yr	BMD at LS and FN	<0.05	141
AR	CAG repeat	299 Swedish men aged 41–76 yr, from the population register in the county of Uppsala	BMD at LS, FN, and total body modified by androgen levels	<0.05	142
		2,693 European men aged 40–79 yr	Ultrasound BMD	0.021	143
ARHGEF3	rs7646054	769 Caucasian women from 335 families, recruited in Australia and the UK	BMD at LS, FN, and total hip	0.0007–0.041	144
BMP2	rs2273073 (Ser37Ala)	192 Chinese including 57 OPLL patients and 135 controls	Occurrence of OPLL in the cervical spine	<0.05	145
	rs235764 (A125611G)	6,752 Caucasian women recruited at four U.S. clinical centers and enrolled into the Study of Osteoporotic Fractures	Vertebral fracture	<0.05	12
BMP7	rs17404303	920 European-Americans from 374 Diabetes Heart Study families	Quantitative CT at thoracic spine and LS	0.03, 0.02	146
BMPR1B	rs1434536, rs3796443	862 Caucasian men aged ≥65 yr	BMD at LS	<0.05	141
CA8	rs6984526	337 Japanese women with osteoporosis	BMD at LS and FN	0.00017, 0.00029	147
CA10	rs2106329	337 Japanese women with osteoporosis	BMD at FN	0.00002	147
CALM1	rs12885713	158 patients with idiopathic knee osteoarthritis and 193 controls in a Greek Caucasian population	Knee osteoarthritis	NS	148
CAR	rs2502815	548 healthy Japanese postmenopausal women	BMD at LS and total body	0.0185, 0.0416	149
CASR	rs1801725 (A986S)	176 premenarche Chinese girls aged 9–11.5 yr	Increased BMD at LS and BMD at Ward’s triangle	0.022, 0.049	150
CD40	rs1883832	602 postmenopausal Spanish women	BMD at LS and FN, osteopenia or osteoporosis at LS and FN	<0.05	151
CLCN7	6 SNPs and 1 VNTR	1,692 healthy premenopausal white sisters aged 33.1 ± 7.2 yr and 715 healthy white brothers aged 33.6 ± 10.9 yr	BMD at LS and FN	NS	152
CNR2	16 SNPs	574 Caucasians from 126 two- to four-generation pedigrees	BMD at radiographic hand breaking bending resistance index	0.007–0.008, 0.001–0.003	153
	rs2501431	1,243 Chinese subjects	BMD at hip and trochanter	0.02, 0.04	154
COL1A1	rs1800012 (Sp1, +1245G /T)	21 osteoporotic patients with proximal femur fracture and 21 controls	BMD at FN, Ward’s triangle, trochanter and total body	NS	155
	rs2412298 (−1663indelT), rs1800012 (Sp1, +1245G /T)	124 females aged 50 to 70 yr cases and 150 controls from Volga-Ural region	Traumatic fracture	<0.05	156
(Continued)
Gene	Marker locus	Population characteristics	Phenotype	P value	Ref.
	rs1107946, rs2412298 (−1663indelT), rs1800012 (Sp1, +1245G /T)	98 consecutive patients including 75 women and 23 men who were admitted to Aberdeen Royal Infirmary with low-trauma hip fractures; 3275 perimenopausal women from the Aberdeen Prospective Osteoporosis Screening Study and 143 postmenopausal women from the same region	Material density of the bone core, yield strength, and toughness	<0.0001	23
COL1A2	rs42524	2,004 elderly Swedish men	BMD at several skeletal sites	<0.05	24
COMT	rs4680 (Val158met)	2,822 Swedish men aged 75.4 ± 3.2 yr	Early fractures (≤50 yr of age)	<0.05	157
CRHR1	rs1876828	157 men long-term survived of ALL at St. Jude Children’s Research Hospital	BMD at LS	0.02	158
CRTAP	rs7623768, rs4076086-rs7623768 haplotype	1080 Chinese females	BMD at FN	0.009, 0.003	159
CTR	AluI	201 Korean men aged 51.6 ± 11.7 yr	QUS	<0.05	160
	AluI	301 healthy young Caucasian men aged 18–25 yr	BMD, BMC, and geometrical areas at specific skeletal sites of the forearm	NS	161
CYP1B1	rs1056836 (Val432Leu)	468 Caucasian postmenopausal women (220 from St. Louis, MO, mean age 63.5 ± 0.53 yr; and 248 from Palermo, Italy, mean age 72.9 ± 0.44 yr)	BMD at LS and FN	0.03, 0.03	162
	rs1056836 (L432V)	124 Japanese women, diagnosed with osteopenia or osteoporosis and taking hormone therapy for 12 months	Lumbar BMD and low-density lipoprotein cholesterol before and after hormone therapy	<0.05	163
CYP17A1	rs743572, rs743575	2,693 European men aged 40–79 yr	Ultrasound BMD	0.018, 0.025	143
CYP19	(TTTA)7-3/ER-α (TA)19 combination,	92 Croatian males aged 21–35 yr	BMD at LS	0.02	11
	(TTTA)7-3/ER-α (TA)21 combination		BMD at LS, FN, and total hip	0.02, 0.02, 0.008
CYP19A1	rs11575899	2,693 European men aged 40–79 yr	Ultrasound BMD	<0.01	143
	rs1062033	1,163 postmenopausal Spanish women	BMD at hip	<0.05	164
DBP	rs222029, rs222020	1,873 Caucasians from 405 nuclear families	CSI at FN	0.0019, 0.0042	131
DKK2	rs17037102, rs6827902, rs17037297	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD, BMC at total hip	<0.05	18
DMP1	rs1381632	862 Caucasian men aged ≥65 yr	BMD at FN	<0.05	141
ENPP1	rs1974201	1,513 unrelated subjects from the Framingham Offspring cohort	FNW	3.8 × 10−7	165
ER-α	rs2228480 (G2014A)	640 Mexican natives including 70 osteoporotic women, 70 nonosteoporotic women, and 500 subjects from the Mexican population	Osteoporosis	<0.05	13
	rs2234693 (PvuII)	228 premenarche Chinese girls aged 9–11.5 yr	BMD at total body and femoral intertrochanter	0.010, 0.038	9
	rs2234693 (PvuII), rs9340799 (XbaI)	158 healthy Chinese adolescent girls aged 12–14 yr	BMD and bone turnover markers	NS	166
	rs2234693 (PvuII), rs9340799 (XbaI)	146 Chinese boys aged 13–17 yr	BMD at total body, forearm and LS influenced by the first spermorrhea	<0.05	10
	rs2234693 (PvuII), rs9340799 (XbaI)	42 Turkish including 21 osteoporotic patients with proximal femur fracture and 21 controls	BMD at FN, Ward’s triangle, and trochanteric	NS	155
	rs2234693 (PvuII)–rs9340799 (XbaI) haplotype	691 postmenopausal Chinese women aged 45–65 yr	Decreased BMD at whole body, LS, and hip	<0.05	8
	rs2234693 (PvuII)- IL-6 rs2010963 (634C/G) interaction	228 premenarche Chinese girls aged 9–11.5 yr	BMD at total left hip and femoral intertrochanter	0.009, 0.007	9
	rs1801132, rs726282	2,693 European men aged 40–79 yr	Ultrasound BMD	0.002, 0.019	143
	rs3020314, rs1884051	350 Chinese with osteoporotic hip fractures and 350 Chinese controls	Hip fracture	0.0004, 0.0004	14
(Continued)
Gene	Marker locus	Population characteristics	Phenotype	P value	Ref.
	rs1801132	6,752 Caucasian women recruited at four U.S. clinical centers and enrolled into the Study of Osteoporotic Fractures	Vertebral fracture	<0.05	12
	(TA)19 repeat	92 unrelated healthy males aged 21–35 yr	BMD at LS, trochanter	0.006, 0.02	11
	(TA)21 repeat		BMD at LS, trochanter, and total hip	0.04, 0.02, 0.03
	rs2234693, rs1884052, rs3778099	425 Chinese adults	BMD	NS	71
ER-β	CA repeat	400 Chinese women including 78 FN PMO cases and 122 controls	Postmenopausal osteoporosis	0.001	167
		108 LS PMO cases and 92 controls		0.023
	rs1256063, rs1256049, rs1256031	2,693 European men aged 40–79 yr	Ultrasound BMD	NS	143
	rs960070	350 Chinese with osteoporotic hip fractures and 350 Chinese controls	Hip fracture	0.0070	14
ESRRA	23 bp repeat polymorphism in promoter	673 premenopausal women from the Toronto metropolitan area	BMD at LS	NS	168
FABP3	rs10914367	360 postmenopausal Hungarian women	BMD at total hip	0.028	137
FGFR1	rs6996321	360 postmenopausal Hungarian women	BMD at LS	0.002	137
FGFR2	rs7916940	862 Caucasian men aged ≥65 yr	BMD at FN	<0.05	141
FLNB	rs7637505, rs9822918, rs2177153, rs2001972	1,085 UK female twins and 1,315 Australian women	BMD at LS and FN	<0.05	169
	rs9828717	1,080 Chinese females	BMD at LS	<0.05	159
FLT1	rs1408245	862 Caucasian men aged ≥65 yr	BMD at FN	<0.05	141
FOXC2	rs3751797	862 Caucasian men aged ≥65 yr	BMD at LS	<0.05	141
FZD1	rs2232157, rs2232158	1,084 African men	BMD at FN, bone size at the radius, strength-strain index	<0.05	170
FZD6	rs3808553 (L345 M), rs12549394 (E664A)	371 postmenopausal Korean women	BMD at LS and FN	NS	171
GDF5	rs143383	2,487 European cases and 2,018 age-matched controls from the UK and Spain	Osteoarthritis	<0.05	172
	rs143383	6,365 Caucasian elderly subjects	Hip axis length and nonvertebral fractures	0.0004, 0.02	173
GHRH	rs4988492	498 men and 468 women aged 59–71 yr from the Hertfordshire Cohort Study	BMC and BMD at proximal femur and LS	<0.05	174
GR	rs1866388	400 Chinese women and 400 Chinese men	Extreme age-adjusted BMD at hip	0.028	175
HMGA2	rs1042725	1,680 Afro-Caribbean men aged ≥40 yr and 1,548 Caucasian American men aged ≥69 yr	Trabecular volumetric BMD	0.007	176
				0.0007
HOXA	rs6951180, rs6964896	862 Caucasian men aged ≥65 yr	BMD at FN and LS	<0.05	141
HSD11B1	6 SNPs	1,392 postmenopausal Korean women	BMD at FN and vertebral fracture	<0.05	177
IGFBP2	rs10932669	862 Caucasian men aged ≥65 yr	BMD at LS	<0.05	141
IL-6	rs603573 (−174G/C), CA repeat	640 Mexican natives including 70 osteoporotic women, 70 nonosteoporotic women, and 500 subjects from the Mexican population	BMD at LS	< 0.0001	178
	rs603573 (−174G/C)	267 postmenopausal women of Wielkopolska region aged 58.5 ± 5.9 yr	BMD at LS	NS	179
	rs603573 (−174G/C)	42 Turkish including 21 osteoporotic patients with proximal femur fracture and 21 controls	BMD at FN, Ward’s triangle, and trochanteric	NS	155
	rs2010963 (−634C/G)	228 premenarche Chinese girls aged 9–11.5 yr	Percentage accrual in BMD of total body and femoral trochanter	0.032, 0.048	9
	rs2010963 (−634C/G)	176 premenarche Chinese girls aged 9–11.5 yr	Increased BMD at total body and femoral trochanter	0.027, 0.028	150
	rs2010963 (−634C/G) interaction with ER-α PvuII	228 premenarche Chinese girls aged 9–11.5 yr	BMD at total left hip and femoral intertrochanter	0.009, 0.007	9
(Continued)
Gene	Marker locus	Population characteristics	Phenotype	P value	Ref.
IL6R	rs3887104, rs4845617, rs8192284	559 postmenopausal Spanish women	BMD at FN BMD at LS	0.011, 0.025, 0.038	180
IL-15	rs2857261, rs10519613, rs56245420, rs1057972	1,921 postmenopausal Korean women	BMD at LS and FN	<0.05	181
IL-23R	rs4655686, rs1569922, rs7539625	443 patients with osteonecrosis of the femoral head and 273 control subjects enrolled in Korea	Osteonecrosis of the femoral head	0.0198–0.0447	182
ITGA1	8 SNPs	946 postmenopausal Korean women	BMD at proximal femur and LS	0.009–0.05, 0.002–0.005	183
LEPR	rs1137100 (Lys109Arg)	145 premenopausal and 118 postmenopausal Korean women	BMD at total hip	0.044	184
LHB	rs2013040	2,693 European men aged 40–79 yr	Ultrasound BMD	NS	143
LHCGR	rs6545061	2,693 European men aged 40–79 yr	Ultrasound BMD	0.023	143
LRP1	rs4759044 and rs4759044-rs11172113 haplotype	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD, BMC at total hip and FN area	<0.05	18
LRP5	rs3781590	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD at total hip	0.0006	18
	rs3736228 (Ala1330Val)	6,082 men from three cohorts (MrOS Sweden: n = 3,014, aged 69–81 yr; MrOs Hong Kong: n = 2,000, aged >65 yr; Swedish GOOD study: n = 1,068, aged 18–20 yr)	BMD at LS and FN	NS	21
	rs3736228 (Ala1330Val)	739 postmenopausal Japanese women	BMD at total body	0.0026	185
	rs3736228 (Ala1330Val)	Meta-analysis	BMD at LS and FN	0.55, 0.05	186
	rs3736228 (Ala1330Val)	Meta-analysis	BMD at LS, FN, and trochanter	<0.001, <0.001, 0.053	187
	rs4988321 (Val667Met)	3,800 men from two cohorts (MrOS Sweden: n = 3,014, aged 69–81 yr; Swedish GOOD study: n = 1,068, aged 18–20 yr)	BMD at LS	<0.05	21
	rs4988321 (Val667Met)	673 premenopausal women from the Toronto metropolitan area	BMD at LS	0.015	168
	rs3736228 (Ala1330Val), rs4988321 (Val667Met)	589 physically active Caucasian men aged 20–30 yr	BMD at LS and whole body BMD at FN	<0.02, 0.04	188
	rs3736228 (Ala1330Val), rs4988321 (Val667Met)	37,534 subjects from 18 participating teams in Europe and North America	BMD at LS and FN, prevalence of all fractures and vertebral fractures	<10−7	20
	rs312009, rs2508836, rs729635, rs643892	964 postmenopausal Spanish women	BMD at LS and FN, fracture	<0.05	189
	rs682429, rs686921	286 young southern Chinese females, aged 22–44 yr with low BMD	BMD at LS and hip	<0.05	190
	rs4988330, rs4988331, rs312786	608 with OA (348 undergoing hip replacement surgery and 260 knee surgery) and 520 with hip fractures	Osteoporosis and osteoarthritis	0.008–0.03	191
	rs41494349 (266A>G)	800 healthy recently perimenopausal Danish women	Parameters of hip geometry	NS	192
	rs545382, rs2277268	126 prepubertal children (64 with premature adrenarche and 62 controls)	BMD at LS and FN	<0.05	193
	rs3736228	652 Slovenian subjects	BMD at LS and FN	NS	194
LRP6	rs2075241, rs11054704	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD and BMC at total hip	<0.05	18
	rs2302685 (Ile1062Val)	37,534 subjects from 18 participating teams in Europe and North America	BMD at LS and FN, prevalence of all fractures and vertebral fractures	NS	20
(Continued)
Gene	Marker locus	Population characteristics	Phenotype	P value	Ref.
	rs1181334	608 with OA (348 undergoing hip replacement surgery and 260 knee surgery) and 520 with hip fractures	Osteoporosis and osteoarthritis	<0.05	191
	rs2302685	652 Slovenian subjects	BMD at LS and FN	NS	194
LTBP2	16 SNPs	1,459 subjects in 306 southern Chinese pedigrees, 706 and 760 case-control subject pairs with extremely high and low trochanter and total hip BMD	BMD at total hip and fracture	0.0004, 0.01	195
MDR1	rs1045642 (C3435T)	127 Chinese systemic lupus erythematosus patients	Development of osteonecrosis of femoral head observed by magnetic resonance imaging	0.038	196
	rs2032582 (G2677T)			0.05
MMP2	5 SNPs	360 postmenopausal Hungarian women	BMD at LS, total hip and distal radius, nonvertebral fracture	NS	137
	rs243865 (C595T)	6,752 Caucasian women recruited at four U.S. clinical centers and enrolled into the Study of Osteoporotic Fractures	Vertebral fracture	<0.05	12
MSTN	3 SNPs	1,260 Chinese from 401 nuclear families	BMD at proximal femur and LS	<0.05	197
MTHFR	rs1801133 (677C>T)	3,196 children aged 9.9 yr from the Avon Longitudinal Study of Parents and Children	BMD at LS	<0.001	198
	rs1801133 (677C>T)	1,243 Chinese subjects	BMD at FN, hip, and trochanter	0.02–0.05	154
	rs1801133 (677C>T)	800 healthy recently perimenopausal Danish women	Parameters of hip geometry	NS	192
NFATC1	rs177820	862 Caucasian men aged ≥65 yr	BMD at LS	<0.05	141
NOG	7 SNPs	2,060 Afro-Caribbean men age 40 and older	BMD at the proximal femur and LS	NS	199
NOS3	18 SNPs and haplotypes	1,451 subjects from Framingham Offspring Cohort	Bone density/ultrasound and geometry	NS	200
	rs2070744 [T(−786)C]	167 Americans including 95 cases with femoral head necrosis and 72 controls	Osteonecrosis of femoral head	<0.05	201
	rs2070744 [T(−786)C]	305 postmenopausal Turkish females	BMD at femoral trochanter and LS	0.046, 0.005	202
NPY	rs16135, rs16123	1,113 randomly selected men of African ancestry	BMD at proximal femur	<0.05	203
OCIL	rs16914640 (Asn19Lys)	500 postmenopausal Spanish women	BMD at LS and FN	<0.05	204
OPG	rs2073618 (Lys3Asn)	69 premenopausal and 263 postmenopausal Spanish women	BMD at LS	<0.05	205
	rs2073618 (Lys3Asn)	6,695 Caucasian women aged 65 yr and older in the United States	BMD at calcaneus and LS, FN fracture	<0.05	16
	rs237025 (A163G)	87 postmenopausal Caucasian women	Calcaneal velocity of sound	0.0102	206
	G209A, T245G	42 Turkish including 21 osteoporotic patients with proximal femur fracture and 21 controls	BMD at FN, Ward’s triangle, and trochanter	NS	155
	rs2073618 (K3N)	239 osteoporotic and 228 nonosteoporotic postmenopausal, 57 premenopausal women, and 117 elderly men	BMD at LS, FN, and total hip	0.021, 0.041, 0.032	5
	rs1032129	964 postmenopausal Spanish women	BMD at FN	0.001	207
PBX1	rs2800791, rs9661977	720 southern Chinese subjects from 231 families; another independent unrelated southern Chinese cohort with a total of 835 extremely high and extremely low BMD subjects; 1,268 case-control Japanese subjects with 703 osteoporotic subjects and 565 normal controls	BMD at LS	0.004, 0.050	208
PIR	rs5935970	4,000 Chinese subjects	BMD at LS	<0.05	209
PLOD	rs7529452, rs2273291, rs7514577	1,243 Chinese subjects	BMD	<0.05	154
PPAR-γ	rs1801282	239 Korean women with mean age 51 yr	Serum OPG level	0.035	210
	rs1801282-rs3856806 haplotype			0.010
(Continued)
Gene	Marker locus	Population characteristics	Phenotype	P value	Ref.
	rs12486170 (−796A>G), rs1801282 (+34C>G), rs3856806 (+82466C>T)	448 patients with osteonecrosis of the femoral head and 336 control subjects	Osteonecrosis of the femoral head	NS	211
P2X7	rs3751143 (Glu496Ala), rs1653624 (Ile568Asn)	800 healthy recently perimenopausal Danish women	Parameters of hip geometry	NS	192
PRL	rs7739889 (T228C)	6,752 Caucasian women recruited at four U.S. clinical centers and enrolled into the Study of Osteoporotic Fractures	Vertebral and hip fracture	<0.05	12
PTH	3 haplotypes	1,044 women aged 75 yr from the Malmo Osteoporosis Prospective Risk Assessment study	Fractures	0.038	212
PTHRLH	rs805512, rs10492364, rs1268693	1,044 women aged 75 yr from the Malmo Osteoporosis Prospective Risk Assessment study	BMD at LS and FN, fracture	NS	212
	VNTR	234 young Finnish males	BMD, BUA, and SOS at LS	<0.0034	213
PTHR1	rs6442037, rs724449, rs7652849	1,044 women aged 75 yr from the Malmo Osteoporosis Prospective Risk Assessment study	BMD at LS and FN, fracture	NS	212
	rs724448-rs2242116 haplotype	1,080 Chinese females	BMD at LS and FN	0.02, 0.044	159
PTHR2	rs9288393, rs10497900, rs897083	1,044 women aged 75 yr from the Malmo Osteoporosis Prospective Risk Assessment study	BMD at LS and FN, fracture	NS	212
PTN	rs322297	862 Caucasian men aged ≥65 yr	BMD at FN	<0.05	141
RANK	+35966insdelC	467 postmenopausal Slovenian women	BMD at LS, FN, and hip	0.020, 0.034, 0.024	214
RANKL	−290C>T, −643C>T, −693G>C	404 postmenopausal Slovenian women aged 42–91 yr	BMD at LS	0.001, 0.041, 0.013	17
	CCG, TTC haplotypes			0.005, 0.007
	−290C>T,	239 osteoporotic and 228 nonosteoporotic postmenopausal, 57 premenopausal women, and 117 elderly men	BMD at LS	0.017	5
	−290C>T-K3N combination		BMD at FN and total hip	0.029, 0.041
	rs12585014, rs7988338, rs2148073	1,873 Caucasians from 405 nuclear families	CSI at FN	0.0007, 0.0007, 0.0005	130
RIZ1	Pro704 ins/del	2,424 men and 3,517 women from the Rotterdam study	BMD at LS and FN, fracture	NS	215
ROR2	19 SNPs	705 Caucasians from 212 nuclear families	Bone length and BMD for hand bones, proximal phalanges and metacarpal bones	<0.05	216
Runx2	4 SNPs and haplotypes	743 Caucasians from 212 nuclear families	Hand bone length and BMD	<0.05	217
	rs7771980	729 postmenopausal Korean women	BMD at LS, trochanter and total femur	0.02, 0.05, 0.04	218
SFRP1	rs921142, rs4736965, and their haplotype	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD and BMC at total hip	<0.005	18
	rs16890444, rs3242	931 Japanese women	BMD at LS, FN, and hip	0.037, 0.025, 0.027	219
SFRP2	4 SNPs	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD, BMC at total hip and FN	NS	18
SHBG	rs1799941, rs6259, TAAAA repeat in promoter	2,693 European men aged 40–79 yr	Ultrasound BMD	NS	143
	rs1799941, rs6259	213 healthy postmenopausal Caucasian women	BMD at the proximal femur sites	<0.05	220
(Continued)
Gene	Marker locus	Population characteristics	Phenotype	P value	Ref.
Smad6	rs755451	721 postmenopausal Japanese women	BMD at total body and LS	0.0004, 0.005	221
SOST	rs851054, rs851056	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD and BMC at total hip	0.004–0.008	18
	rs1230399	1,243 Chinese subjects with low BMD and high BMD	BMD at LS, FN, trochanter and, total hip, osteoporosis	0.004–0.03 0.005	222
	rs1534401, rs1877632	862 Caucasian men aged ≥65 yr	BMD at LS	<0.05	141
	rs851054, rs851056
	rs10534024	652 Slovenian subjects	BMD at LS and FN	NS	194
SPARC	1046C-1599C-1970T, 1046C-1599G-1970T haplotypes	115 middle-aged Caucasian men with idiopathic, low turnover osteoporosis (n = 56) and matched controls (n = 59)	BMD at LS, hip, and radius	<0.05	223
SRD5A2	rs632148, rs523349, rs765138, rs7571644, TA repeat in exon 5	2,693 European men aged 40–79 yr	Ultrasound BMD	NS	143
SREBF1	rs12601420 (IVS6-48 C>T), rs9925115 (IVS7+117A>G)	423 patients with osteonecrosis of the femoral head and 348 controls	Osteonecrosis of the femoral head	<0.0001	224
TGFB1	rs1800469 (1348 C>T), rs1800472 (788 C>T)	28,924 subjects from 10 European research studies	BMD at LS, vertebral fracture	<0.05	225
THSD4	rs10851839	Japanese women with osteoporosis aged 72.7 ± 7.3 yr	BMD at LS and FN	0.0092, 0.0046	226
THSD7A	rs12673692	337 Japanese women with osteoporosis aged 72.7 ± 7.3 yr	BMD at LS and FN	0.00017, 0.036	226
TIMP2	7 SNPs	360 postmenopausal Hungarian women	Nonvertebral fracture	0.0187	137
TNF-α	rs1800629 (−308G/A)	267 postmenopausal women of Wielkopolska region aged 58.5 ± 5.9 yr	BMD at LS	NS	179
	rs1800629 (−308 G/A)	159 Portuguese	BMD at LS and hip	NS	227
	rs1799964 (−1031 T>C)	377 postmenopausal Korean women	Osteoporosis	<0.05	228
TNF-β	rs909253 (A252G)	377 postmenopausal Korean women	Osteoporosis	<0.05	228
TNFRII	rs1061622 (T676G), rs3397 (1690T>C), A1663G	377 postmenopausal Korean women	Osteoporosis, bone turnover markers	<0.05	228
TNFRSF1B	rs976881	1,243 Chinese subjects	BMD	0.04–0.08	154
TSHR	rs1991517 (Asp727Glu)	4,934 elderly Caucasian men and women of the Rotterdam Study	BMD at FN	<0.05	229
TWIST1	BD0027 (+1871A>G)	729 postmenopausal Korean women	Osteoporosis, BMD at FN	0.02, 0.039	230
VEGF	rs2010963 (−634G>C)	814 Korean including 317 osteonecrosis of the femoral head cases and 497 controls	Osteonecrosis of the femoral head cases	0.015	231
	rs2010963 (−634 G>C), rs3025039 (+936 C>T)	252 postmenopausal Caucasian women aged 46–80 yr	BMD at LS	0.017	232
				0.05
VDR	rs11568820 (Cdx-2)	190 postmenopausal Brazilian women	BMD at LS, FN, great trochanter, and Ward’s triangle with different physical activity level	<0.05	6
	rs10735810 (Fokl)	192 postmenopausal active women	BMD at FN and Ward’s triangle	<0.05	233
	rs1544410, rs7975232, rs731236 (Bsml, ApaI, TaqI)	246 postmenopausal Turkey women including 100 osteoporosis and 146 controls	Osteoporosis	NS	234
	rs1544410, rs7975232, rs731236 (Bsml, ApaI, TaqI)	47 Caucasian patients with ulcerative colitis and 47 control subjects matched for age and gender	BMD at LS	<0.05	235
	rs1544410, rs7975232, rs731236 (haplotype BBAAtt)	147 postmenopausic Venezuelan women including 71 cases with osteoporosis and 76 controls	Osteoporosis	<0.05	236
(Continued)
Gene	Marker locus	Population characteristics	Phenotype	P value	Ref.
	rs1544410, rs7975232, rs731236, rs10735810 (BsmI, ApaI, TaqI, FokI)	238 Polish women including 75 premenopausal cases with Graves’ disease and 163 controls	BMD at LS and FN	NS	237
	rs1544410, rs7975232, rs731236, rs10735810 (BsmI, ApaI, TaqI, FokI)	126 Chilean elderly women aged 65–94 yr including 67 hip fractures and 59 controls	Hip fracture	NS	238
	rs1544410 (BsmI)	301 healthy young Caucasian men aged 18–25 yr	BMD, BMC and geometrical areas at specific skeletal sites of the forearm	NS	161
	rs1544410 (BsmI)	335 Korean women older than 65 yr with low calcium intake	BUA of calcaneus	0.013	239
	rs1544410 (BsmI)	42 Turkish including 21 osteoporotic patients with proximal femur fracture and 21 controls	BMD at FN, Ward’s triangle, and trochanteric	NS	155
	rs1544410 (BsmI)	68 healthy, 54 osteopenic, and 64 osteoporotic Argentine postmenopausal women	BMD at LS and FN	<0.05	240
	rs1544410 (BsmI) interaction with ER-α PvuII and XbaI	68 healthy, 54 osteopenic, and 64 osteoporotic Argentine postmenopausal women	BMD at FN	<0.05	240
	rs7975232, rs731236 (ApaI, TaqI )	40 Lactating Brazilian adolescents aged 15–18 yr	BMD at LS and BMC at total body	<0.05	241
	BsmI, FokI, Cdx2 and combination of K3N(OPG)-Cdx2	239 osteoporotic and 228 nonosteoporotic postmenopausal, 57 premenopausal women, and 117 elderly men	BMD at FN, LS, and total hip	<0.05	5
	rs7975232 (ApaI)	136 postmenopausal women	BMD at LS and hip	<0.05	242
WISP3	4 SNPs	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD, BMC at total hip and FN area	NS	18
WNT3a	rs708114, rs4653533, rs752107	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD at total hip	<0.05	18
WNT7b	8 SNPs	344 Australian women with extreme BMD, aged 55–80 and >5 yr postmenopausal	BMD at total hip	<0.05	18
WNT10b	rs1051886, rs3741627	1,035 Afro-Caribbean men aged ≥40 yr and 980 Afro-Caribbean men	Cross-sectional area, periosteal circumference, BMD at hip and BMC at the radius and cortical	<0.05	243
	7 SNPs	1,438 Caucasian postmenopausal women	BMD at LS and hip	NS	244

Adapted from Y. Guo et al.: Expert Rev Endocrinol Metab 3:223–267, 2008 (3), with permission from Expert Reviews Ltd. ALL, Acute lymphoblastic leukemia; BMC, bone mineral content; BUA, broadband ultrasound attenuation; CSI, compression strength index; FNW, FN width; GOOD study, Gothenburg Osteoporosis and Obesity Determinants study; NS, not significant; PMO, postmenopausal osteoporosis; QUS, quantitative ultrasound; SOS, speed of sound; VNTR, variable number tandem repeat.

Table 3.

GWL scans for osteoporosis-related phenotypes in humans (published between October 2007 and November 2009)

Study subjects	No. of markers	Phenotype	Results	Candidate gene	Ref.
300 Mexican-Americans (>45 yr of age) from the San Antonio Family Osteoporosis Study	460	ΔBMD at 33% ulna	3p (81 cM), LOD = 1.9		30
		ΔBMD at hip	6q (103 cM), LOD = 1.75
103 Caucasian pedigrees (Network in Europe on Male Osteoporosis Family Study) ascertained through a male relative with low (Z-score ≤ −2) BMD values at either LS or FN	441	BMD at LS	1q42-43, LOD = 1.75		26
			11q12-13, LOD = 2.64	LRP5, TCIRG1
			12q23-24, LOD = 1.65
			17q21-23, LOD = 3.63	COL1A1, SOST
			21q22, LOD = 2.05	COL6, COL6A2
			22q11, LOD = 2.74
		BMD at FN	5q31-33, LOD = 1.53	RIL
			3q12-14, LOD = 2.71	RANKL
879 Caucasians of AFOS (mean age ± sd = 49.8± 16.1 yr, range 18–91 yr) from large multigenerational families	731	NN_ID	1p35.5, LOD = 2.57; 2p25, LOD = 2.37	ALKPL (1p35.5)	245
		NN_Z	1p36, LOD = 2.36; 14q23, LOD = 2.5	BMND3, MTHFR (1p36)
		S_BR	1q23-24, LOD = 2.65; 12q24.2, LOD = 2.63	SAC, Hypercalciuria (1q23-24)
		S_Z	4p11, LOD = 1.94
		S_ID	5q24-30, LOD = 2.59	LOX
		S_CSA	11q22.1, LOD = 2.3
		NN_CSA	2p11.2, LOD = 1.92; 6q22-23, LOD = 2.46	COL10A1, ENPP-1, WISP3 (6q22-23)
			17q11.2-12, LOD = 2.16	SOST
2522 Caucasian females from 414 pedigrees	410	BMD at FN/AAM	22q13, LOD = 3.33; 3q13, LOD = 2.31	EP300 (22q13)	35
		BMD at LS/AAM	22q13, LOD = 3.30; 15q13, LOD = 2.97
			3p25, LOD = 2.36; Xp22, LOD = 2.20	PPARG (3p25)
			Xq13, LOD = 2.18; Xq23, LOD = 2.53
			Xq27, LOD = 2.34
		BMD at UD/AAM	22q13, LOD = 3.12; 7p15, LOD = 2.44
2200 Caucasians aged 20–50 yr from 207 pedigrees	410	PBMD at wrist in total sample	2p13, LOD = 2.04		29
			10p14, LOD = 2.31
			14q23, LOD = 2.07	ER-β, BMP-4
			Xq27, LOD = 2.64
		PBMD at hip in total sample	2p12, LOD = 2.79	MGP
			22q13, LOD = 2.16	MCHR1
		PBMD at wrist in female	2p13, LOD = 2.64
			6q24, LOD = 1.91	ER-α
			11q13, LOD = 1.97	FRA-1, LRP5, TCIRG1
			18q21, LOD = 2.29
		PBMD at hip in male	15q26, LOD = 2.93
			7p21, LOD = 2.10	IL-6
4126 Caucasians for a composite osteoporosis phenotype that combines OF and low BMD	393	A composite osteoporosis phenotype that combines OF and low BMD in total sample	14q32, LOD = 2.61; 7p14, LOD = 2.42; 11q25, LOD = 2.09; 9p21, LOD = 1.26	BMP4	246
		A composite osteoporosis phenotype that combines OF and low BMD in female	14q22, LOD = 3.53; 9p21, LOD = 2.29; 7p14, LOD = 3.07	GLI3
		Imprinting analyses by assigning weights to allele sharing specific to parental origins among affected sibling pairs	1q42, LODFA = 2.12; 9q34, LODFA = 1.88; 7q22, LODMO = 1.67	COL1A2
(Continued)
Study subjects	No. of markers	Phenotype	Results	Candidate gene	Ref.
1473 Caucasians aged 31–96 yr in 323 pedigrees from the Framingham Osteoporosis Study	636	S_CSA/height	Chr. 2, LOD = 1.56; Chr. 4, LOD < 1.0 Chr. 6, LOD = 2.34; Chr. 9, LOD = 1.73	CALM2, FSHR, TGF α (Chr. 2); IL2 (Chr. 4)	247
		NN_CSA/height	Chr. 9, LOD = 1.04
		FNL/height	Chr. 15, LOD = 1.57
		NN_Z/height	Chr. 4, LOD = 2.05
		IT_OD/height	Chr. 6, LOD = 2.13	ESR1, IGF2R
		S_Z/height	Chr. 6, LOD = 2.08; Chr. 21, LOD = 2.45
		S_OD/height	Chr. X, LOD = 3.28
3899 Caucasians from 451 pedigrees	410	PC phenotype of BS at LS, hip, and wrist in total sample	7q34, LOD = 2.85	IRF 5, Leptin	248
		PC phenotype of BS at LS, hip, and wrist in male
		PC phenotype of BS at LS, hip, and wrist in total female	5q23, LOD = 2.39
			11p11, LOD = 2.82	EXT2
			7q34, LOD = 1.79; 8q24, LOD = 2.30
			21q21, LOD = 1.85
34 Caucasians in a single extended family	380	BMD at LS	1q36.3, LOD = 3.07	WDR8, EGFL3	27
4126 Caucasians from 451 pedigrees	410	aBMD/ABS at FN	20q11, LOD = 3.11; Xp11, LOD = 2.46	GDF5 (20q11); BMP15 (Xp11)	32
		aBMD/ABS at LS	Xq27, LOD = 4.30; 12p11, LOD = 2.86	BGN, IRAK1 (Xq27); LRP6 (12p11)
			17q21, LOD = 2.44	COL1A1, CHAD, HOXB, SOST
		aBMD/ABS at UD forearm	5q23, LOD = 3.01; Xq27, LOD = 2.43	IL4, LOX (5q23)
			7p15, LOD = 2.92; 12p11, LOD = 2.42	IL6, NPY, HOXA@, HOXB@ (7p15)
2582 Caucasian females from 451 pedigrees including 1486 premenopausal and 1096 postmenopausal women	410	BMD at LS in total sample	15q13, LOD = 3.67; 3p25, LOD = 3.06		28
			16p13, LOD = 1.94; Xq25, LOD = 2.95
		BMD at LS in postmenopausal women	15q13, LOD = 2.49	HERC2
		BMD at LS in premenopausal women	15q13, LOD = 1.52
		BMD at hip in total sample	Xp11.4, LOD = 1.95
4126 Caucasians from 451 pedigrees	410	BMD at LS/TBLM in total sample	7p22, LOD = 2.53; Xq25, LOD = 5.26	TWIST, IL6	33
			Xp22.3, LOD = 5.20; Xq13.3, LOD = 4.31
			Xp11.4-Xq11.1, LOD = 4.54
		BMD at hip/TBLM in female	Xq23-Xq24, LOD = 3.78; Xq27.1, LOD = 3.68
		BMD at LS/TBLM in female	7q32, LOD = 2.67	LEP
			15q13, LOD = 4.86; 7q32, LOD = 2.44	GREM1 (15q13)
			Xq13.3, LOD = 3.94; Xp22.3, LOD = 3.92
		BMD at hip/TBLM in male	Xq22.33, LOD = 3.92
		BMD at LS/TBLM in male	7q21, LOD = 2.52
			13p11, LOD = 3.25; Xq11.1, LOD = 3.78
(Continued)
Study subjects	No. of markers	Phenotype	Results	Candidate gene	Ref.
4126 Caucasians from 451 pedigrees	410	BMD at LS/BFM in total sample	7p22-p21, LOD = 2.69	IL6, RAC1	34
			6q27, LOD = 2.42	ESR1
		BMD at hip/BFM in total sample	6q27, LOD = 2.30
			2q32, LOD = 2.29	GDF8, STAT1
		BMD at wrist/BFM in total sample	11q13, LOD = 2.64	GAB2, FRA-1, LRP5
		BMD at hip, LS/BFM in total sample	6p21, LOD = 2.32	TNF α, RUNX2, HLA-A
		BMD at LS/BFM in female	6q27, LOD = 2.34
		BMD at wrist/BFM in female	15q13, LOD = 3.32
		BMD at LS/BFM in male	6p25-24, LOD = 3.15	BMP6
		BMD at hip/BFM in male	13q12, LOD = 3.23	KL
			7q21, LOD = 2.59	CTR, SERPINE1, PON1
2584 Caucasian females from 414 pedigrees	410	BMD at LS/AAM	1q44, LOD = 4.28; 2q37, LOD = 3.39		36
			8q24, LOD = 4.59; 13q33, LOD = 3.31	EXT1 (8q24)
			15q13, LOD = 4.80; 22q13, LOD = 4.60
		BMD at FN/AAM	8q11, LOD = 3.36; 22q13, LOD = 4.88
		BMD at UD/AAM	15q24, LOD = 5.25; 22q13, LOD = 4.79
4498 Caucasians from 451 pedigrees	410	BR/TBFM	20q12, LOD = 3.23		249
		CSA/TBFM	20p11, LOD = 2.47
		CT/TBFM	6q27, LOD = 3.19
		W/TBFM	20p12, LOD = 1.68
		Z/TBFM	7q11, LOD = 2.47
327 Mexican-Americans (aged 25–45 yr) from 32 extended pedigrees	460	ΔBMD at FNK BMD in total sample	1q23, LOD = 3.6	BGLAP, osteocalcin, IL6R, SLC39A1	250
		ΔBMD at 33% ulna	11p14-15, LOD = 2.5	CALCA, CALCB, CALCP
1346 Caucasians from 327 extended families of the Framingham Study	636	S_CSA / LLM	12p12.3-12p13.2, LOD = 3.49		251
		NSA/LLM	14q21.3-22.1, LOD = 3.77
3782 Caucasian females aged 18–80 yr	400	Length of spine	5q15-5q23.1, LOD = 3.00	IRX	31
		Length of femur	5q15-5q23.1, LOD = 2.19
348 Caucasian dizygotic healthy female twin volunteers	737	Lumbar degenerative disc disease	Chr. 1 (285 cM), LOD = 6.05		252
			Chr. 5 (175 cM), LOD = 6.05
			Chr. 19 (80 cM), LOD = 4.06
210 individuals from 39 families of predominantly British descent without mutations of SQSTM1	382	Paget’s disease of bone	10p13, LOD = 4.08		253
405 Caucasian pairs and 110 American black pairs of brothers aged 18–61 yr	402	aBMD at LS	7q34, LOD = 5.26		254
		aBMD at hip	4q21, LOD = 4.14; 14q32, LOD = 4.19	ESRβ (14q32)
			19p13, LOD = 5.77; 21q21, LOD = 4.63
			22q13, LOD = 4.71

Adapted from Y. Guo et al.: Expert Rev Endocrinol Metab 3:223–267, 2008 (3), with permission from Expert Reviews Ltd. Z, Section modulus; NN, narrowest section of the femoral neck; AFOS, Amish Family Osteoporosis Study; TR, trochanter; OF, osteoporotic fracture; IT, intertrochanteric; S, femoral shaft; NSA, neck-shaft angle; CSA, cross-sectional area; FNL, femoral neck length; PC, Principal component; aBMD, areal BMD; ABS, areal bone size; TBFM, total body fat mass; LLM, leg lean mass; ID, inner diameter; UD, ultradistal; W, sub-periosteal diameter; LOD, likelihood; Chr., chromosome; OD, outer diameter; BR, buckling ratio. The two traits for bivariate analysis are separated by a slash (/). Bivariate LOD scores listed here were converted into one degree of freedom, equivalent to univariate LOD score level.

Table 4.

GWA study for osteoporosis-related phenotypes in humans (published between October 2007 and November 2009)

Genotyping methods	Total markers	Significant markers	Discovery subjects	Replication subjects	Phenotype	P value	Candidate gene	Ref.
Infinium HumanHap300; HumanCNV370 SNP chip; Centaurus platform (Nanogen)	301,019 SNPs with genotypic call rates ≥ 97%, HWE P ≥ 10−7	rs9594759, rs9594738	5,861 Icelandic persons, 87% of whom are women	4,165 other Icelandic persons, 74% of whom are women; 2,269 postmenopausal Danish women; 1,491 persons from the Australian Dubbo Osteoporosis Epidemiology Study cohort, 61% of whom are women	BMD at LS and hip, osteoporotic fracture	1.2 × 10−7 to 2.0 × 10−21	RANKL, OPG, ZBTB4, MHC, RANK, ESR1	39
		rs6469804, rs6993813
		rs7524102, rs6696981
		rs3130340, rs3018362
		rs9479055, rs4870044
		rs1038304, rs6929137
		rs1999805
		rs2306033, rs7935346					LRP4
		rs11898505					SPTBN1
Infinium assay; HumanHap 550 v3.0 array; Taqman system	314,075 SNPs with genotypic call rates ≥ 90%, HWE P ≥ 0.0001, and MAF ≥ 1%	rs4355801	2,094 women from the Twins UK cohort	6,463 people from three other cohorts in Western Europe (4081 from Rotterdam cohort, 1,692 from Twins UK replication cohort, and 690 from Chingford cohort)	BMD at LS and FN, osteoporosis	<5 × 10−8	TNFRSF11B	40
		rs3736228					LRP5
IlluminaHap300K	186 SNPs	rs1513670, rs7220711	6,865 Icelanders, 5,934 women, and 931 men	3,015 Icelanders, 3,884 Danish postmenopausal women, and 1,491 from the Australian cohort	BMD at hip or spine	1.8 × 10−9 to 1.3 × 10−7	SOST	42
		rs1107748
		rs2010281, rs10876432					MARK3, SP7
		rs3018362					TNFRSF11A
Affymetrix Human Mapping 500K Array Set	727 CNV regions	CNV 4q13.2	700 elderly Chinese individuals including 350 cases with homogeneous hip osteoporotic fractures and 350 controls	Chinese, 399 cases with hip osteoporotic fracture and 400 controls; 689 Chinese unrelated subjects, and 1,000 white unrelated subjects; 236 young Chinese males	Osteoporotic fracture, hip BMD and FN bone geometry, concentrations of serum testosterone and estradiol	2.0 × 10−4	UGT2B17	47
Affymetrix Mapping 250 k Nsp and Affymetrix Mapping 250 k Sty arrays; Illumina Infinium assay; Affymetrix Human Mapping 500 K array set	379,319 SNPs with genotypic call rates ≥ 95%, HWE P ≥ 0.001 and MAF ≥ 1%	rs7595412	1,000 homogeneous unrelated Caucasian subjects, including 501 females and 499 males	1,216 women of Caucasian European ancestry from the UK; a Chinese sample containing 403 female subjects including 226 with low trauma osteoporotic hip fracture, and 177 controls	Hip bone size	3.72 × 10−7	PLCL1	51
		rs892515				8.62 × 10−3
		rs9789480				2.44 × 10−3
		rs3771362				7.66 × 10−3
(Continued)
Genotyping methods	Total markers	Significant markers	Discovery subjects	Replication subjects	Phenotype	P value	Candidate gene	Ref.
Affymetrix Gene Chip Human Mapping 500K Array Set; Affymetrix 50K supplemental array	379,319 SNPs with genotypic call rates ≥ 95%, HWE P ≥ 0.001 and MAF ≥ 1%	rs11864477, rs11860781, rs16945612, rs11859065	1,000 homogeneous unrelated Caucasian subjects, including 501 females and 499 males	593 white U.S. families (1,972 subjects), a Chinese hip fracture sample (350 cases, 350 controls), a Chinese BMD sample (2,955 subjects), and a Tobago cohort of African ancestry (908 males)	BMD and hip fracture	2.56 × 10−5 to 2.13 × 10−8 (meta-analyses)	ADAMTS18	46
		rs17131547					TGFBR3
Illumina HumanHap317K SNP chip	317,504 SNPs with missing genotypes ≤5%, HWE P ≥ 10−7 and MAF ≥ 1%	rs4759021	1,518 children from the Avon Longitudinal Study of Parents and Children	3,692 children from the Avon Longitudinal Study of Parents and Children	BMD	5.8 × 10−4	Osterix	255
					BMC	1.7 × 10−3
					Bone area	3.8 × 10−3
Affymetrix Genome-Wide Human SNP array 5.0	352,228 SNPs with missing genotypes ≤5%, HWE P ≥ 10−6, and MAF ≥ 1%	rs7776725	8,842 Korea samples from population-based cohorts	7,861 independent Korean samples	BMD at radius, tibia, and heel	1.0 × 10−11 to 1.6 × 10−6	FAM3C	41
		rs1721400				1.4 × 10−7 to 2.2 × 10−3	SFRP4
Illumina HumanHap300; HumanHapCNV370	303,120 SNPs with missing genotypes ≤5%, HWE P ≥ 0.001, and MAF ≥ 1%	rs219780	3,773 kidney stone cases and 42,510 controls from Iceland and The Netherlands	1,520 Icelandic kidney stone cases and 4,726 controls, 746 kidney stone cases and 3,751 controls from The Netherlands	BMD at hip and spine	0.00039; 0.0077	CLDN14	43
Affymetrix Gene Chip Human Mapping 500K Array Set	342,854 SNPs with HWE P ≥ 0.0001 and MAF ≥ 5%	rs9630182, rs2036417, rs7125774; rs8057551, rs8061992, rs7199138	983 unrelated Caucasian subjects	A family-based sample of 2,557 Caucasian subjects	BMD at FN	3.98 × 10−7 to 6.74 × 10−3	PTH, IL21R	256
Meta-analysis	36,016 SNPs in 150 candidate genes	241 SNPs	BMD data were obtained from 19,195 participants from 5 populations of European origin; data on fractures were obtained from a prospective cohort (n = 5,974) from The Netherlands		BMD at LS and FN fracture risk	P < 2.39 × 10−6	ESR1, LRP4, ITGA1, LRP5, SOST, SPP1, TNFRSF11A, TNFRSF11B, TNFSF11	45
(Continued)
Genotyping methods	Total markers	Significant markers	Discovery subjects	Replication subjects	Phenotype	P value	Candidate gene	Ref.
Meta-analysis	2,543,686 SNPs	467 SNPs	19,195 participants from five populations of European origin, e.g., the Rotterdam Study (n = 4,987), Erasmus Rucphen Family Study (n = 1,228), Twins UK Study (n = 2,734), deCODE Genetics Study (n = 6,743), and Framingham Osteoporosis Study (n = 3,503)		BMD at LS and FN	P < 5 × 10−8	GPR177, MEPE, SPTBN1, CTNNB1, MEF2C, ZBTB40, STARD3NL, FLJ42280, LRP4, ARHGAP1, F2, DCDC5, SOX6, ESR1, FOXL1, HDAC5, CRHR1, TNFRSF11B LRP5, SP7, TNFSF11, TNFRSF11A	44

BMC, Bone mineral content; HWE, Hardy-Weinberg equilibrium; MAF, minor allele frequency.

Table 5.

Studies using transgenic and knockout mouse models relevant to osteoporosis (published between October 2007 and November 2009)

Candidate gene	Protein	Manipulation	Phenotype	Ref.
Adipoq	Adiponectin	Overexpression	Decreased BMC at femur and lumbar vertebra	257
AR	Androgen receptor	Knockout	Decreased trabecular bone mass in male mice	258
		AR and ER-α double-knockout	Additionally reduced cortical bone and muscle mass	258
Agtr2	Angiotensin II receptor, type 2	Treatment with AT2 receptor blocker	Enhanced levels of bone mass	259
BSP	Bone sialoprotein	Overexpression	Osteopenia and mild dwarfism	260
CB1	Type 1 cannabinoid receptor	CB1−/− mice	Increased peak bone mass due to reduced bone resorption, age-related osteoporosis with reduced bone formation, and accumulation of adipocytes in the bone marrow space	261
CCR2	C-C chemokine receptor-2	Knockout	High bone mass	262
CD200	CD 200 molecule	CD200−/− mice, produced by homologous recombination	Fewer osteoclasts and more bone	263
Cftr	Cystic fibrosis transmembrane conductance regulator	Knockout	Increased trabecular bone volume	264
COL1A2	Collagen, type I, α 2	Knock-in	Reduced body mass, areal BMD, and bone strength	265
Col6a1	Collagen, type VI, α 1	Knockout	Accelerated development of OA joint degeneration; delayed secondary ossification and reduced BMD	266
Col9a1	Short collagen IX	Col9a1−/− and col9a1+/− mice	Trabecular bone loss in young adult female col9a1 (−/−) and col9a1 (+/−) mice; trabecular bone architecture deterioration in both male and female heterozygous col9a1 (+/−) mice while aging	267
Cthrc1	Collagen triple helix repeat containing-1	Knockout and overexpression	Low bone mass in Cthrc1-null mice and high bone mass in Cthrc1 transgenic mice	55
Cxcr4	Chemokine (C-X-C motif) receptor 4	Overexpression	Increased cell trafficking to bone in ovariectomy mice	268
CYLD	Deubiquitinating enzyme	Knockout	Osteoporosis	269
CYP191A1	Aromatase	Osteoblast-specific overexpress aromatase	Increased total body BMD, trabecular BMD, cortical BMD, and cortical thickness associated with elevated osteoprotegerin mRNA levels and reduced number of osteoclasts	270
DKK1	Dickkopf-1	Mice with hypomorphic Dkk1d (doubleridge) alleles, express low amounts of Dkk1	Increased trabecular and cortical bone mass	271
Dlx5	Distal-less homeobox 5	Dlx5+/− mice	Lower BMD and reduction in cortical thickness of femoral midshafts	272
ER-α/ER-β	Estrogen receptor	Knockout/knockdown	ERKs are not activated by stretching in osteocytic and osteoblastic cells	273
Frzb	Frizzled-related protein	Knockout	Increased cartilage proteoglycan loss	274
GIP	Glucose-dependent insulinotropic peptide	Overexpression	Increased osteoblastic activity	56
Gpnmb	Osteoactivin	Mutation leading to the generation of a truncated osteoactivin protein	Reduced alkaline phosphatase activity and calcium deposition	275
GPR30	G protein-coupled receptor 30	GPR30−/− mice	Reduced bone growth	276
Gpr48	G-protein-coupled receptor 48	Gpr48−/− mice	Dramatic delay in osteoblast differentiation and mineralization	277
HIF-1α	Hypoxia inducible factor-1 α	Knockout	Reduced bone formation ability of osteoblasts	278
		Knockout	Reduced values of bone histomorphometry and BMD	279
HIP/RPL29	Ribosomal modulator of protein synthesis rate	HIP/RPL29-deficient mice	Increased bone fragility	280
HSD17B2TG	Hydroxysteroid (17β ) dehydrogenase 2	Express human hydroxysteroid (17β) dehydrogenase 2	Decreased bone formation rate at prepubertal age in male mice	281
(Continued)
Candidate gene	Protein	Manipulation	Phenotype	Ref.
IGF-I	Insulin-like growth factor 1	Knockout	Lower femur BMD	282
		Knockout	Reduced body length, areal BMD, and BMC	283
		Crossbred IGF-I transgenic mice with homozygous weaver mice	Improved BMD and BMC	284
IL-7	Interleukin 7	Overexpression	Age-related loss of trabecular bone in both axial and long bones	285
Irf8	Interferon regulatory factor-8	Deficient in Irf8	Osteoporosis	286
JunD	An activator protein-1 component	Knockout	Increased bone volume and sustained high bone mass even after estrogen depletion	287
Mecp2	Methyl CpG binding protein 2	Knockout	Decreased osteoblast activity	288
Nfatc1	Nuclear factor of activated T cells c1	Knockout	Osteopetrosis and inhibition of osteoclastogenesis	289
Nmp4	Nuclear matrix protein 4	Knockout	Greater PTH-induced acquisition of femoral trabecular bone	290
NMU	Neuromedin U	Knockout	Increased bone mass	291
OT	Oxytocin	Deletion of OT or the OT receptor (Oxtr) in mice	Osteoporosis	292
PAI-1	Plasminogen activator inhibitor-1	Overexpression	Increased mineralization and biomechanical properties in 32-wk females	293
PAPP-A	Pregnancy-associated plasma protein-A	Knockout	Insufficiency in mass, density, architecture, and strength	294
PECAM-1	Platelet/endothelial cell adhesion molecule 1	Knockout	Reduced trabecular bone volume and number of trabeculae in femoral and tibial long bones	295
PPAR-γ	Peroxisome proliferator-activated receptor-γ	PPAR-γ is deleted in osteoclasts but not in osteoblasts	Increased bone mass, reduced medullary cavity space, and extramedullary hematopoiesis in the spleen	296
PTHr1	Parathyroid hormone receptor 1	Expressing a constitutively active PTH receptor	Increased bone mass and elevated bone remodeling	297
Rac1	Rho-GTPase	Nf1+/−, Rac1−/− mice	Normalized Erk activation compared with Nf1+/− osteoclasts	298
Rac2	A member of the Rho family of small GTPases	Knockout	Increased bone mass and anabolic response to PTH	299
RANKL	Receptor activator of nuclear factor-κ B ligand	Deletion of the distal control region, located 76 kb upstream from the transcription start site	Increased bone mass and strength	300
Runx2	Runt related transcription factor 2	Runx2+/− mice	Decreased BMD and trabecular bone volume and delayed bone formation	301
Sca-1	Stem cell antigen 1	Knockout	Decreased BMD at femoral and whole body	302
Sfrp1	Secreted frizzled-related protein 1	Knockout	Reduction of the cartilage callous, increased intramembranous bone formation	303
Sfrp4	Secreted frizzled-related protein 4	Overexpressing in osteoblasts	Reduction of trabecular bone mass	304
SOST	Sclerostin	Knockout	High bone mass phenotype characterized by marked increases in BMD, bone volume, bone formation, and bone strength	59
		Short treatment with an antibody to sclerostin	Halted bone loss	305
Sp7	Osterix	Inactivated in all osteoblasts	Decreased BMD and bone-forming rate in lumbar vertebra, thinner cortical bone, and more porous with reduced bone length	306
Sparc	Osteonectin	Knockout	Low-turnover osteopenia	307
Spp1	Osteopontin	Lacking expression of the osteopontin	Lower resorptive capacity of osteoclast	308
SRC-1	Steroid receptor coactivator 1	Knockout	Decrease in trabecular volumetric BMD and impaired skeletal response to estrogen in female mice	57
SRC-2	Steroid receptor coactivator 2	Knockout	Increased bone mass, decrease in bone marrow adipocytes	309
(Continued)
Candidate gene	Protein	Manipulation	Phenotype	Ref.
Spp24	Secreted phosphoprotein 24 kDa	Constitutive expression of fl bovine spp24 (1-203)	Inhibition of ectopic bone formation in male mice and adverse affects on BMD and histological parameters related to bone mass and formation in female mice	310
Terc	Telomerase	Knockout	Osteoporosis	311
TIMP-1	Matrix metalloproteinases	Overexpression	Decrease in bone loss induced by estrogen deficiency	312
TNFr	Tumor necrosis factor receptor	Knockout	Reduced adverse effects of magnesium deficiency on bone	313
TNFR2	Tumor necrosis factor type 2 receptor	Knockout	Decrease by lipopolysaccharide on the BMD of tibiae, femurs, and lumber vertebrae	314
TRα1	Thyroid hormone receptor α1	TRα 1(+/m)β(+/−) mice, express a mutant thyroid hormone receptor α1 with dominant-negative properties due to reduced ligand-binding affinity	Osteosclerosis in adults and delayed ossification in juveniles	315
TRPV4	Transient receptor potential vanilloid 4	Knockout	Suppressed unloading-induced bone loss	316
uPAR	Urokinase receptor	Knockout	Increased BMD and osteogenic potential of osteoblasts; decreased osteoclast formation; altered cytoskeletal reorganization in mature osteoclasts	317
VDR	Vitamin D receptor	VDR−/− mice	Shorter latency to fall from the rotarod, smaller fall angle in the tilting box test, and aberrant poor swimming	318
Wnt10b	Wnt10b protein	Transgenic mice, in which mouse Wnt10b is expressed from the human osteocalcin promoter	Increased mandibular bone and impaired eruption of incisors during postnatal development, higher BMD, bone volume fraction, and trabecular number at femoral distal metaphyses	319
Zmpste24	Zinc metallopeptidase, STE24 homolog	Zmpste24-null progeroid mice	Age-related bone loss including lower osteoblast and osteocyte numbers and higher levels of marrow adipogenesis	320
5-HT(2B)R	Monoamine serotonin (2B) receptor	Knockout	Osteopenia that worsens with age in female mice	321

Adapted from Y. Guo et al.: Expert Rev Endocrinol Metab 3:223–267, 2008 (3), with permission from Expert Reviews Ltd. BMC, Bone mineral content.

Table 6.

A summary of the DNA microarray studies on osteogenesis and bone-related diseases (published between October 2007 and November 2009)

Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
Human trabecular bone	Trabecular bone from the IT region of the proximal femur from individuals with OP, OA, and control	Comparing the gene expression profiles of OA, control, and OP bone samples to identify altered gene expression in OP fracture bone	150			TREM2, ANXA2, CD14, CCR1, ADAM9, ST14, CTSB, FST, PRRX1, IL10, CTGF, KLF10, COL4A1, CCL2, SPP1, ANGPTL4, PEA15, MARCO, AEBP1		SPARC, DOK4, SLC14A1, ADM, TAZ	82
	Trabecular bone from the IT region of the proximal femur from individuals with OP, OA, and control	Comparing the gene expression profiles of OA, control and OP bone samples to identify altered gene expression in OA bone	150	Targets component or modulating genes of either the WNT signaling pathway or the TGF-β/BMP signaling pathway		MMP25, S100A4, SMAD3, WNT5B		IBSP, TWIST1, TIMP4, ADAMTS4, ADM, GADD45B, MEPE, COL4A1	81
Human MG63 cells	MG63 cell lines stably expressing pcDNA3-Flag-VP16C-SXR or the empty vector Flag-tagged pcDNA3	Vehicle (0.1% ethanol) or MK-4 (10 μm) treatment	85		85	GDF15, STC2, TRIB3, GDF8,MGC4504, ASNS, SLC7A11, COL15A1, TNFAIP6, DDIT4, PCK2, FGF2, PSAT1, HMOX1			322
	Human MG63 cells	Bio-Oss and PerioGlas				NOG, CALCR, COL11A2, COMP, TNFRSF11B, CASR, HOXA13, DLX5			59
	Human MG63 cells	Titanium						FGFR3, IGF1, MSX1, GDF10, CALCA, BMP1, BMP7, COL1A1, PHEX, CASR, COL11A1	60
	MG63 osteoblast cells	1.035 MHz pulsed ultrasound with three different acoustic pressures	377	Genes involved with cellular membranes, regulation of transcription		LRP5, BMP1, COL1A1, CD151, PTRF, HOXB8		EZH2, BMP2K	323
	Human MG63 cells	PerioGlas and P-15				NOG, COL11A2, IGF1, COL11A1, CLO2A1, CHRD, HOXD13, ALPL, BMP1, COMP, CALCR		GDF10, ANXA2,	324
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
	Human MG63 cells	Zirconium oxide				EGR2, FKHL18, SEP1, PRDM2, DLX2, FGF5, SP110, TOP3A, PKD2, PTPRG, PTPN6, KCNK4, LEPR, NR4A1, PDCD8, DUSP16, CD79B		GABRA6, RORB, GRCA, CLTCL1, PCBP4, WDR33, CDKN3, F11R, AAMP, VPS18, SEC23B, ILF2	325
	Human MG63 cells	Mixed a type I collagen solution with human osteoblast-like MG-63 cells and intermittent compression, gels loaded in wells without mechanical stimulation were used as controls	43		30	COX-2, MMP-3, ODC	13	MIL-1RAcP	326
	Human MG63 cells	PerioGlas		Genes involved in bone formation and skeletal and cartilage development		ZNF547		NOG, EN1, CHRD, GHRHR, COMP	58
	Human MG63 cells	High magneto-gravitational environment		Adducing 3, WIPF1, coactosin-like 1, filamin A, SORBS3, CDC42BPB, talin 1, paxillin, WASF2, tropomodulin 3, SPTBN1, supervillin, plectin 1					327
	Human MG63 cells	Cultured with porous polyethylene						CHRD, EN1, NOG, ADAMTS4, GHRHR, OSTF1, MGP, PTH, LECT1	63
	Human MG63 cells	Anatase coating				PRDX1, COL9A2, ADAMTS4, SHOX ALPL, AMBN, TUFT1		PHEX, FBN1, IGFBP4, CALCA PTH, TFIP11	61
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
	Human MG63 cells	Peptide-15				TFIP11		NOG, HOXD13, AEBP1, SHOX, EN1, COMP, IGF1, MATN1, SUFU	328
	Human MG63 cells	Zirconium oxide		NOG, SHOX, IGF1, BMP1, FGFR1		TRAPPC2		COMP, CHRD, BMP1, SHOX, IGF1, FGFR1, GHRHR, PMF1, AHSG, AMBN	65
	Human MG63 cells	Titanium and ZDCs				BMP7		NOG, FGFR1, COMP, MSX1	62
	Human MG63 cells	Demineralized freeze-dried bone allograft		Genes involved in cell cycle regulation, immunity, vesicular transport, bone remodeling, production of cytoskeletal elements					67
	Human MG63 cells	Calcium sulfate				BMP1, BMP7, PTH, FGFR1, CALCA, SHOX, EXT2		MSX1, EBP, EN1, INHBA, CMKLR1, COMP, NOG	64
	Human MG63 cells	Anatase coating				ERBB3, TSPAN-2, CRK, HYAL3, WDR10, MYOC, ANK3, LILRB4		CCR2, MYH11, CLTB, RTN4, MECP2, GLI2	329
	Human MG63 cells	Danggui Buxue Tang	883	CCL-2, CCL-7, CCL-8, galectin-9					330
	Human MG63 cell and sub-line model	Cisplatin, doxorubicin, and etoposide		ABCG2, ADD3, NMT2, WNTSa, PTN					331
Human osteoblast	Human osteoblast	Osteoblasts cocultured with U-266 myeloma	38	Genes involved in bone metabolism and other cell functions	7	E4BP4, UNC5C, VCAM1, UGP2, PRKAR2A, Follistatin precursor, enzyme similar to tissue plasminogen activator	31	BMPs, Runx2, COX-2, Coronin, Osterix, osteocalcin, MTRR, PMS2L9, NRIF3, CYP3A43, LIC2, Type XVIII collagen	70
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
	Human osteoblasts	Cyclosporin A		Wnt signaling molecules and the cytoskeletal and focal adhesion cascade					332
	Human osteoblasts isolated from trabecular bone tissue of FN and proximal femur	Comparing the gene expression profiles of osteoblasts from osteoporotic vs. nonosteoporotic bone tissues	1606		144	IBSP, CXCL2	208	PTN, COL15A1	333
	Human osteoblasts isolated from cancellous bone fragments	Shock wave	94			PTHrP, prostaglandin E2-receptor EP3, chordin, BMP-2 inducible kinase		Matrilin 3, cartilage oligomeric matrix protein	334
	Human osteoblasts derived from alveolar bone	TAK-778				ALP, osteocalcin, Msh homeobox 2, RANKL adhesion molecule 1			335
	ODHPSCs and normal osteoblasts derived from human dental pulpar and alveolar bone fragments	Identify genes that are differently regulated in ODHPSC in comparison to normal osteoblasts		Genes involved in cell differentiation, developmental maturation, cell adhesion, and production of cytoskeleton elements		SYNE1, LAMA1, MMP2, PPP1R9B NEO1, CDH1, AGT, MAP4K4, CDC2L5, GPC3, FYN, GNE, ATP2A2, TTN		MARK2, BAIAP2, EPB42, SDK1, PGM5, LRRN5, CSF3R, CNTN2, CXCL12, MSN, DNM1, ACTA2, SNTA1, MMP24, COL6A1, PLOD1, ARMCX3, PRDM2	336
	Human MC3T3-E1 preosteoblastic cells	Hydroxyapatite			11	SOX9, GUGA1B, HAPLN 2, RRAGA	6	Calpain 9, PI3KS, AGTR 2, OGFR, TSLRP	337
	Human nonunion osteoblasts	Evaluated global gene expression in human osteoblasts and human nonunion osteoblasts	281		200	HBEGF, VEGFB	81	IGF-2, TGF-β2, FGF-1, FGF-R2, BMP-4, PDGF, WISP2, WISP3	338
	Primary cultures of osteoblastic cells differentiated from the human bone marrow	Polymethylmethacrylate, hydroxyapatite, bioactive glass 45S5, titanium and stainless steel				IGF binding protein 4		BMP5, osteocalcin	66
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
	Osteoblasts differentiated from human MSCs	siRNAs generated from PDE7 and PDE8 PCR products			473	β-catenin, osteocalcin, caspase-8, CREB-5, BMP-5	156	BMP-1, osteoflycin, 1,25-dihydroxyvitamin D3 receptor	80
	Tumor chemonaive osteoblastic populations	Comparing the expression profiles of osteoblasts from osteosarcoma patients vs. controls				EBF2			339
Human MSCs	Human MSCs	Treated with lactate for different time periods		Genes involved in wound healing		Interleukin-6, heat shock protein, hypoxia-inducible factor-1 α		Superoxide dismutase 2, BCL2- associated X protein	340
	Human MSCs	Two osteogenic nanoscale topographies (pitted surface vs. raised islands) are compared with cells treated with dexamethasone				Integrin αM, TGFR1, MMP8, collagens, ALPL, integrinα1			341
	Human MSCs	Cell culture						Osteogenic and adipogenic differentiation capacity of hMSCs in late stage of the culture	75
	Human MSCs	Growth medium and osteoblastogenesis induction medium		Interferon γ-inducible genes		ALP, OCN, TGFBRII, PDGFR, BSPII, OPN, LRP5, Runx2			342
	Human MSCs	Extracellular matrix substrates						MCAM, angiopoietin-1	343
	Human MSCs	Dexamethasone and 1,25-(OH)2D3	2095		961	c-Myc			344
	Human MSCs and ASCs	Osteogenic differentiation medium	41	Extracellular matrix-related genes				COL1A2, COL3A1, COL4A1, COL5A2, COL15A1, PHR, COL2A1, COL6A1, COL9A1, INT-β 3, INT-β 1, osteopontin, osteonectin	345
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
	Human MSCs and osteoblasts	Comparing gene expression in osteoblasts vs. MSCs from the same donor	170		120	COL15A1, CDKNIA, DST, GADD45A, MACF1, COMP, COL8A2, STAT1, TNFRSF10B		MK167, PARP3, RAD54L, ECM1, NRP1, GSTP1	346
Human unrestricted somatic stem cells	Human unrestricted somatic stem cells	Has-mir-135b						IBSP, osterix	347
Human monocytes	Monocytes from premenopausal women with extremely low vs. high PBMD		49	GBP1, STAT1, CXCL10					71
	Monocytes from premenopausal women with extremely low vs. high BMD		13		6	STAT1, IFI44L, CXCL10, IFI44, GPB1, GPB2			348
Human hFOB1.19 cells	Immortalized human osteoblast cells hFOB1.19	Two-stage cell transformation using MNNG and TPA treatments	10	H19, MKRN3, NDN, CDKN1C, PHLDA2, MEST, CD81, GRB10, SLC22A18, SLC22A3	6	H19, PHLDA2, SLC22A18	4	CD18	349
	Human osteoblastic hFOB1.19 cells	Oral administration of collagen tripeptide			169	Osterix			68
Human STRO-1+ skeletal stem cell	STRO-1+ cell from bone marrow of hip	Cells cultured in basal, osteogenic, chondrogenic and adipogenic lineage-specifying culture conditions	38		16	COL111A1, FKBP5, LOC388610, MT1E, MT1F, MT1X, LBP, MT1H/MT1P2, XTLT1, C1Oorf10, CXCL6, NPY2R, C22orf16, PTGFR, FAM89A, STC2	22	GATA6, LDB2, GREM1,QSOX1, CLDN11, KCTD12, RBMS3, ZFP36L1, RBM39, ROCK1	350
Human HeLa and NIH3T3 cells	Human HeLa and NIH3T3 cells	Transfected with 5–10 μg siRNA vector			2	OPG, VEGF-A			351
Human myeloma cell lines	KMS-26 cell line	Enzastaurin	94		62	CXCL12, CXCR4, CTSB, TRAF5, BCL2L1, IGF1, GADD45A, CDC20	32	MYC, IRF4, MX11	352
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
Human MNCs	MNCs from bone marrow and umbilical cord blood	Chondrogenic differentiation		Receptors for fibronectin, vitronectin, osteopontin, and collagens		Ilk, CD47, Integrin a5b1, components of the vitronectin/osteopontin-receptors (avb5), collagens (a1b1, a2b1, a3b1)			353
Human chondrogenic cell line OUMS27	Human OUMS27 cells	Compared mRNA levels in OUMS27 cells adenovirally transfected with KLF5 and the control empty vector				MMMp9			354
Human aortic valve interstitial cells	Human aortic valve interstitial cells	LPS stimulation or BGP treatment; media alone was used as positive control				BMP2, PDGFA, FGF2, MMP1			355
Human OA cases and controls	Serum samples	Human OA cases compared to controls	16			MMP-7, IL-15, sVAP-1		PAI-1, ICAM-1	121
	Bone samples	Patients suffering from hip OA compared to younger donors undergoing spinal arthrodesis	83	CCL2, FOS, OMD, PRSS11, DVL2, AKT1, CA2, BMP6, MMP2, TGFBR3, FLT1, TNFRS11B, BMP1					96
Human cartilage	Chondrocytes from human knee cartilage	IL-1β				CXCL1, CXCL2, CXCL3, CXCL5, CXCL6, CXCL8, CCL2, CCL3, CCL4, CCL5, CCL8, CCL20, CCL3L1, CX3CL1, LIF, IL-6, GCSF3, ELAM1, MMP-13, BMP-2		COL2A1, aggrecan	356
	Human OA cartilage	Notch signaling was blocked by DAPT treatment				IGFiBP7, TNFSF15, IGF2, TNFRSF10c, FGFR2, CALCR1		BMP2, IL8	95
Human bone tissue	Human femoral bone tissue from non-osteoporosis controls and cases with osteoporosis			IGSF4, FABP3, FABP4, FKBP2, TIMP2, TRIB, TMSB4X					357
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
Human OPLL cells and non-OPLL cells	Human OPLL cells and non-OPLL cells	Osteogenic induction and siRNA-mediated knockdown of Runx2	47		24	Stathmin-like 3, Sestrin 2	23	COMP, CTGF, angiopoietin-1	93
Human RBM cell line	Human RBM cell line	Compared gene expression between two human RBM tissues and patient-matched primary RCC tissues				MIP-1δ			74
Human dental pulp cells and human MSCs	Human dental pulp cells and human MSCs	Inorganic polyphosphate				MMP1, OPN, OC, osteoprotegerin			79
Human periodontal ligament cells	Human periodontal ligament cells	Static tensional forces for 2 h			2856	IL-8, MMP2, PDGF-A	2574	ALP, COL1A2	358
Human osteosarcoma cell line Saos-2	Human RANK-positive Saos-2 osteosarcoma cells	RANKL	69		21	ROCK1, EMA3A	48	GDF15	89
Human sarcoma cells	Human sarcoma cells from osteosarcoma patients	Comparing gene expression profiles of sarcoma cells vs. human MSCs				ROR2			359
Human MDNSCs	Human adult MDNSCs	Investigated the expression profile of cancer-related genes in MDNSCs by comparing with that in fresh normal human adult bone marrow depleted of red blood cells			63	MYC, MMP2, Notch2, STC1, ITGA3, Wnt1, STAT5b, RhoC			360
Human chondrocytes	Human chondrocytes in osteoarthritic cartilage	Compared gene expression profiles of the divided 3 zones of the cartilage	198						94
Human cultured discs and cells	Human discs and anulus cells cultured in 3-dimensional					IGFBP-2, IGFBP-4, IGFBP-5			361
Mouse RAW 264.7 cell line	Mouse RAW264.7 cell line	RANKL				MMP-9, TRAP, cathepsin K, MST1R, integrin b3, NFATc1, calcitonin receptor		CD14 antigen, toll-like receptor-6, TNF, STAT-1, Fc receptor 11b, TNF receptor 1B	362
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
Mouse RAW 264.7 cells	Mouse RAW 264.7 mature osteoclasts					SPHK1, Wnt10b, BMP6		Sclerostin	363
Mouse RAW 264.7 cell and BMC	Mouse RAW 264.7 cell and BMC	RANKL				NHA-oc/NHA2			73
Mouse MC3T3 E1 cells	Mouse osteoblasts MC3T3 E1 cells, clone 4	Three dishes for 1-h thapsigargin treatment at 10 nm and the other three dishes as control				ATF4, eIF2α-p, FOS, FGF-9, BMP-2			364
	Mouse MC3T3-E1 preosteoblasts	Histone deacetylase inhibitors, such as trichostatin A, MS-275, or valproic acid				Slc9a3r1, sorbitol dehydrogenase 1, glutathione S-transferase α 4		Proteasome subunit, β type 10, adaptor-related protein complex AP-4 σ 1	365
	Mouse osteoblastic cell line MC3T3-E1 subclone 4	Mechanical straining	674	Fos, Ptgs2, Rgs2, Pthlp					366
Mouse bone marrow cells	Mouse osteoclast progenitors	IL-6 and RANKL		Genes related to MAPK phosphatases and the ubiquitin pathway		MKP 1, MKP 7		Senp 2, Cul4A	367
	Primary bone marrow cells	IGF2				CXCL7, SDF1			72
Mouse C3H10T1/2 cells	Mouse C3H10T1/2 cells	Runx2 or osterix adenovirus						Wnt4, Bglap1, BMP7	69
Mouse C3H/HeJ and C57BL/6J bone	Mouse C3H/HeJ and C57BL/6J femur	Rosiglitazone				Genes associated with PPARG signaling pathway and fatty acid metabolism		Genes associated with cell cycle	368
Mouse C57BL/6 tissues	Tibiae from wild-type mice	Mechanical loading for 3 h	642		324	Osteopontin, Postn, Ostn, Dlx5, Bmp4, Bmp10, Sost, Timp1, Timp2, Ctgf, Esr1	318		369
	Bone samples from the tibiae	Ankle load			242	C-fos, Egr1, Atf3, MMP3	199	TNF, CRLF1, BARX2	370
Mouse breast adenocarcinoma cell lines	Tumor-bone interface and the tumor alone area	Genetic expression at the tumor-bone interface was compared with the tumor alone area				Cathepsin G, cathepsin K, MMP9, MMP13			92
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
Mouse MSC line, Kusa 4b10 cells	Differentiating osteoblasts	PTH or PTHrP	4675	1444 genes were significantly changed by both treatments, 1384genes changed specifically by PTH, and 403 genes specifically by PTHrP at all time points		Genes belonging to the family of ephrins and their receptors, such as ephrinB2		PDZ-RGS3	87
Mouse ES cells	Mouse ES cells	Investigate the differentiation of mouse ES cells cultured under three differentiation conditions, embryoid body, gelatin, matrigel				Spp1, Csf1, Gsn, Bmp8b, Crlf1			371
Mouse chondrogenic cell line ATDC5	Mouse chondrogenic cell line ATDC5	Dexamethasone	96			CTGF, integrinα10, SGCK, DMP1, lipocalin 2		SFRP, IGF-1, Lumican	372
	Mouse chondrogenic cell line ATDC5	Leucine and rapamycin		1,571 genes affected by leucine restriction and 535 genes affected by rapamycin					373
Mouse VSMCs	Mouse VSMCs	MK 4				DT-diaphorase		Osteoprotegrin	374
Mouse 129SvEv	129SvEv Sparctm1cam null, male vs. 129Sv/Ev wild-type, male, femoral midshaft							Sparc, Zfp162, Bysl, E2F4	307
Swiss Webster mouse BVECs and MVECs	VECs from trabecular bone regions and central marrow cavity regions of mouse long bones	Compared the gene expression in BVECs and MVECs			5	ALDH3A1, SMOC-2, MMP-13, C/EBP-b, ANX8	2	Spa, MGP	375
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
Mouse TM40D and TM40D-MB cells	Mouse TM40D and TM40D-MB cells	Compared the gene expression profile of low (TM40D) vs. high metastatic (TM40D-MB) tumor cells	20			COX2, Fgf7, Ptx 3		CD1d1 antigen, Cxcl1	376
Mouse osteoblasts	Osteoblasts	Wnt3a and FGF alone or in combination		70% of the genes induced by Wnt3a were down-regulated by combined FGF treatment					377
	Calvarial osteoblasts from 7-d-old CD1 mice	Unit gravity (1 g) vs. modeled microgravity (0.0008 g)		Genes involved in osteoblast differentiation, function, and osteoblast-osteoclast cross-talk	45	IL-6, Lcn2, Nqo1	88	Penk1, Tnmd, Aspn, Cdh11, Ogn, Wisp2, Sfrp2	378
	Primary osteoblast cultures isolated from wild-type and Arrb2 −/− mouse calvaria	Intermittent PTH			215	Slc11a1, Unc93b1, Pla2g7, Lgmn, Cebpd	200	Sept7, Ttc3, Steap4, Bmpr1a, Sh3bgrl, Hltf	379
	Osteoblasts and osteocytes from mouse neonatal calvaria	Comparing genes expression between osteocytes and osteoblasts	385		249	Col15a1, Ank, Gnas, ENpp1, Enpp6, Notch1, Dlk1, Ptprz1, Bmp4, Gdf10, Tgfβ3, Fgf1	136	Col16a1, Mmp9, Mmp23, Adamts18, Fzd1	380
Mouse osteocyte Y4 Cells	Osteocyte Y4 cells from the long bone	10−7m risedronate and alendronate for 48 h				Genes encoding zinc ion binding proteins			381
Mouse osteocyte	Mouse osteocyte	Comparing the gene expression profiles of GFP-positive vs. GFP-negative cells			269	DMP1, Sost			382
Mouse ST2 cells	Mouse ST2 osteoblastic/stromal cells	P. gingivalis ATCC33277, gingipain-mutants						Cyclin D, Cyclin E	383
Other parts or tissues of mouse	Calvarial cells of Wnt5a−/− and wild-type mice	Gene expression profiles of the Wnt5a−/− calvarial cells as compared to wild-type cells were evaluated	3528		2002	Fb1, Fst, Cccng1, Wisp1	1526	Runx2, Osterix, ALP, Sfrp2, Igf2, Dhh, Col2a1	384
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
	Suture-associated dura mater	Compare genomic changes in the dura mater underlying the posterior frontal and sagittal sutures of mice	6057	2529 genes at postnatal d 5, increasing to 3439 genes at postnatal d 10 and 6057 genes at postnatal d 20		Igf2, Tgfb2, Vegf-C, cadherin 1, Collagen 1		Mmp13	385
	Cartilage of the distal femur of neonatal mice	Dicer-deficient				Hmga 2			386
	p85α−/− and wild-type OCs	M-CSF and RANKL					94	MITF, JDP2, cathepsin K, TRAP, MMP-9, integrin β3	387
	Diabetic mice and control mice	Insulin						FUNX2, RUNX2-related osteogenic genes	388
	Mouse femurs and tibiae excluding joints and primary metaphyses	Glucocorticoid treatment				Csf1, c-fms, Ibsp, Itgb3, disintegrin, Adam8, Trem2, Oscar, PlCγ, c-Fos, Nfatcl, c-Src, Syk, Vav3, ATPase, Ctsk		TGFβ1, BMP-2, LEF1, Akp2, MAPK	88
	Region of regeneration with the underlying dura mater in skeletally immature and mature mice with injured calvaria	Parietal bone defects were created by a 4-mm trephine bit			25	BMP-2, BMP-4, BMP-7, IGF-2, FGFR-1, Ptn, Acp5, Ctsk, Mmp2, Mnp14, Ttr, Ptgds			389
	Mouse calvarial bone	Examined the effect of the col9al-null mutation on the expression of osteoblastic genes	1	Col9a1			1	Col9a1	267
	Femur of C3H and B6 strains mice	Compared the effect of rosiglitazone in the C3H and B6 strains			7 gene sets	PPARG signaling pathway and fatty acid metabolism in both C3H and B6 strains with no significant difference between the two strains		Genes associated with cell cycle in the C3H strain	368
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
	Mouse wounds	HOXA3	837		332	Lrp1, MCP-1/Ccl2	505	Tnf-α, Myd88, Tollip	390
Rats of Sprague-Dawley	Kidney of female Sprague-Dawley rats	Ovariectomy and herbal formula	64		22	HSD17B2, rGK-4, PDK4, HMGCS2	42	PTGER3, NQO1, YC2, UGT1A6, GSTP2, AKR7A3, SREBP-1	391
	Both pairs of femora and tibiae of Sprague-Dawley female rats	Three different PTH peptides, PTH (1-34), (1-31), and (3-34)		Slpi, tfpi2, socs3, gro1		RANKL, c-fos, phex, Gprc5c		CXCR4	86
	Fractured femur of male Sprague-Dawley laboratory rats	TP580				PI3KR1,CAMKK1, Fibronectin, NFATC1, NFATC2, VHL, COX2		Dynamin 2, caveolin 1	84
	BMSCs of Sprague-Dawley rats	Exposed to simulated microgravity and static controls	413		206	ABI1, PTTG1, PMP22, ETS1	207	CCND1,CDK5, CDC5L, CATNB, FZD1, GPNMB, WNT5A, WISP1	78
	Growth plate chondrocytes from the PC and RZ of Sprague-Dawley rats	Captured the unique features of the PC and RZ		8 transcripts showing high expression unique to the PC and RZ, respectively		Tpt1, Ctgf, Morf4l1, Gas6, Col5a2, Frzb, Gdi2, Jund; Col9a1, Hapln1, Col1a2, LOC689064, Smoc2, Cast, Rp137, LOC497729			392
	BMSCs of Sprague-Dawley	α-Minimal essential medium containing β-glyceraldehyde-3-phosphate, l-ascorbic acid, dexamethasone and 1,25-(OH)2D3			12	Pla2g2a, Expi, Pcp4, Cx3cl1, GluAP, Fcna, Mmp9, Lcn2, Il10, Fmo1	1		76
	Cultured bone marrow MSCs of Sprague-Dawley rats	Electromagnetic field	19		6	Bmp1, Bmp7	13	Egf, Egfr	77
Rats of Wistar strain	Femurs	Titanium implants			86	Collagenous and noncollagenous extracellular matrix-related genes, proteoglycans and bone resorption-related genes			85
	Abdominal aorta tissues	High phosphorus diet	53				53	SFRP1, SFRP2, SFRP4, cathepsin K	393
(Continued)
Subject or sample categorization	Cell type or tissues	Treatments	Differentially expressed genes (n)	Important differentially expressed genes	Up-regulated genes (n)	Important up-regulated genes	Down-regulated genes (n)	Important down-regulated genes	Ref.
Rats of Fisher 344	Irradiated and nonirradiated primary osteoblast cultures	HBO treatments			26	Integrin β1, β-tubulin, ADP-ribosylation factor 1, stearoyl-coenzyme A desaturase 2, aldolase A, HSP90 β, peroxiredoxin	7		394
Rats of Fisher 344 and LEW strains	Proximal femora	Copenhagen 2331 and Dark Agouti were used as a negative control	99	VEGF, FGF2, IGF2, IGF2BP3, TNF					395
Rats of Fischer 344, LEW, COP, and DA strains	Femoral bone tissues	Comparing the mRNA sequence of QTL regions contributing to the variation in lumbar vBMD with the NCBI database	285	Akap1, Asgr2, Esd, Fam101b, Irf1, Lcp1, Ltc4s, Mdp-1, Pdhb, Plxdc1, Rabep1, Rhot1, Slc2a4, Xpo4					396

Adapted from Y. Guo et al.: Expert Rev Endocrinol Metab 3:223–267, 2008 (³), with permission from Expert Reviews Ltd. IT, Intertrochanteric; OP, osteoporosis; ZDCs, zirconium dioxide ceramics; ODHPSC, osteoblasts derived from human pulpar stem cell; PDE, phosphodiesterases; TPA, 12-O-tetradecanoyl phorbol-13-acetate; MNNG, N-methyl-N′-nitro-N-nitrosoguanidine; MNC, mononuclear cell; KLF, Kruppel-like factor; LPS, lipopolysaccharide; BGP, β-glycerol phosphate; DAPT, N-[N-(3,5-diflurophenylacetate)-L-alanyl]-(S)-phenylglycine t-butyl; RCC, renal cell carcinoma; RBM, RCC bone metastasis; MDNSCs, marrow-derived neural stemlike cell; ES, embryonic stem; VSMC, vascular smooth muscle cell; MK-4, menaquinone-4; BVEC, bone-derived vascular endothelial cell; VECs, vascular endothelial cells; MVEC, marrow-derived vascular endothelial cell; M-CSF, macrophage colony stimulating factor; OC, osteoclast; PC, perichondral zone; RZ, reserve zone; HBO, hyperbaric oxygen: ASCs, adipose tissue-derived stromal cells; GFP, green fluorescent protein; siRNA, small interfering RNA; vBMD, volume BMD.

II. Candidate Gene Association Studies

Candidate gene association analysis has the advantages of higher statistical power and easier sample recruitment compared with the linkage approach. During the period from October 2007 to November 2009, a series of genes catalogued on the basis of biological functions in Table 1 have been recognized and tested as candidates to osteoporosis-related traits by association studies. Several classical candidate genes, such as vitamin D receptor (VDR), estrogen receptor (ER), and low-density lipoprotein receptor-related protein 5 (LRP5), have been most widely studied (Table 2).

A. Receptors for calciotropic hormones

1. VDR

1,25-Dihydroxyvitamin D₃ [1,25-(OH)₂D₃], the biologically active metabolite of vitamin D, is required for calcium and phosphorus homeostasis, for normal skeletal development, and for maintenance of skeletal architecture. The action of 1,25-(OH)₂D₃ is mediated by VDR, which is a member of the superfamily of steroid/thyroid hormone/retinoid receptors. VDR was the first candidate gene to be studied in relation to osteoporosis (4). Frequently studied markers of VDR include BsmI, ApaI, TaqI, Cdx2, and FokI. For example, Cdx2 polymorphism was associated with femoral neck (FN) BMD in a study of 239 osteoporotic postmenopausal women carried out by Mencej-Bedrac et al. (5). In another study performed by Gentil et al. (6), Cdx2 polymorphism did not influence BMD in postmenopausal women by itself, but actually affected the BMD response to physical activity. The interpretation of VDR polymorphisms is currently hindered by the fact that most studies were performed with relatively small sample sizes and investigated only limited polymorphisms (e.g., BsmI, ApaI, TaqI, Cdx2, and FokI), which largely have unknown effects. Whole-gene analyses that exhaustively explore all potential sequence variations within/around the VDR gene in samples of larger size are critical for identifying potential functional variants.

2. ER-α and ER-β

Estrogens play an important role in regulating bone homeostasis, bone turnover, and maintenance of bone mass. The effects of estrogens on skeletal structure are mediated through binding to two different ERs, which are encoded by the ER-α and ER-β genes. Both receptors are highly expressed in bone (7). Although some studies showed significant associations between the three major polymorphisms [e.g., PvuII (T>C) and XbaI (A>G) in intron 1, and the TA repeat in the promoter region] of ER-α and osteoporosis-related phenotypes, the sample sizes for most of these studies were relatively small (8,9,10,11). Rs1801132, another marker for ER-α, was found to be associated with osteoporotic fractures in 6752 Caucasian women (12). Other positive results included associations between G2014A polymorphisms and osteoporosis (13) and between rs3020314 and rs1884051 and hip fractures (14). For ER-β, polymorphisms of rs960070 were associated with susceptibility to fractures (14). The effects of individual polymorphisms of ER-α and ER-β on osteoporosis warrant further confirmatory studies. These results are important and have the potential to impact therapy of osteoporosis through the development of estrogen replacement therapy and selective ER modulators that may be based on individual genetic makeup.

B. Cytokines and receptors

1. OPG-RANKL system

Two cytokines, osteoprotegerin (OPG) and receptor activator of nuclear factor κ-β ligand (RANKL), have been identified as important mediators in the pathogenesis of osteoporosis. OPG is a decoy receptor for RANKL. Its binding to RANKL blocks interaction of the latter with the receptor activator of nuclear factor κ-β (RANK) on the osteoclast surface, thereby inhibiting bone resorption (15). Moffett et al. (16) investigated the association between the Lys3Asn polymorphism (G-to-C polymorphism at codon 3 in exon 1) in the OPG gene and both BMD and the risk of fractures in 6695 white women. Women homozygous for the G (Lys) allele had significantly lower BMD at the intertrochanter, distal radius, and lumbar spine (LS) than those with the C (Asn) allele. Additionally, compared with women with the G/G (Lys-Lys) genotype, those with the C/C (Asn-Asn) genotype had a 26% increased risk of hip fractures (95% confidence interval, 1.02–1.54) and 51% increased risk of FN fractures (95% confidence interval, 1.13–2.02) (16). In postmenopausal women, the promoter polymorphisms of RANKL, including −290C>T, −643C>T, −693G>C, and two common haplotypes, CCG and TTC, showed association with LS BMD (17). Interestingly, the −290C>T (RANKL)-K3N (OPG) combination was associated with total hip BMD and FN BMD in 239 osteoporotic and 228 nonosteoporotic postmenopausal women, respectively (5), suggesting the presence of gene-gene interactions between RANKL and OPG on BMD. Further confirmation of their combined influence in larger cohorts is needed.

2. LRP5

LRP5 functions as a cell-membrane coreceptor for Wnt proteins in the canonical Wnt signaling pathway (18). Several lines of evidence suggest that LRP5 may be a key determinant of bone mass (19). Val667Met and Ala1330Val polymorphisms of the LRP5 gene have been widely studied, and both of these polymorphisms were consistently associated with LS BMD, FN BMD, and fracture risk across different Caucasian populations (20). Grundberg et al. (21) also showed an association between the Val667Met polymorphism and LS BMD by a meta-analysis in 3800 young and elderly men from Swedish cohorts. The approach of meta-analyses, by combining results across studies, is helpful in resolving problems of underpowered studies, revealing unexpected sources of heterogeneity, and resolving discrepancies in genetic studies (22).

C. Bone matrix proteins

1. COL1A1 and COL1A2

Collagen type I is the most abundant protein in connective tissue and is essential for normal bone function. The collagen I triple helix consists of two α1(I) chains and one α2(I) chain encoded by the collagen α1(I) (COL1A1) and collagen α2(I) (COL1A2) genes, respectively. Genetic variants of these two genes may cause normal variation in BMD and bone strength. A common noncoding polymorphism lying in the transcription factor Specificity protein 1 (Sp1) binding site, named Sp1 polymorphism, is the most widely studied polymorphism in the COL1A1 gene. Jin et al. (23) found that Sp1 was associated with biomechanical properties of bone and reduced bone quality. This polymorphism is one of the few polymorphisms that has been repeatedly associated with fractures and BMD (24). Genotypes of a common coding polymorphism in the COL1A2 gene, rs42524, were tested for associations with bone phenotypes in 2004 elderly Swedish men (24). Associations were observed between rs42524 genotype and BMD at several skeletal sites, and elderly Swedish men heterozygous for rs42524 were found to have lower BMD than homozygous subjects (24).

D. Limitations and improvements

Candidate gene association studies have substantial limitations for detecting the genetic basis of osteoporosis because this approach relies on selection of the correct genes for association studies based on either a biological hypothesis or the location of a particular gene in implicated linkage regions. In addition, most current candidate gene association studies in the bone field have generated inconsistent/inconclusive results. This limitation has been well addressed in our previous review (1). Several steps can be taken to increase the probability of discovering genes important to the pathogenesis of osteoporosis using candidate gene association studies, including: 1) controlling population stratification and enlarging sample size to increase statistical power; 2) using more efficient and reasonable methods, such as the single nucleotide polymorphism (SNP) spectral decomposition method, to correct for multiple testing by estimating the effective test number (25); 3) performing whole-gene analyses, rather than analyses for limited polymorphisms in the target genes, to explore all potential sequence variations within and around the gene of interest; and 4) taking into account the influence of gene-gene epistasis and gene-environment interactions on osteoporosis.

III. GWL Studies

Generally, family-based GWL studies are conducted to examine whether any markers from panels of microsatellite markers spaced about uniformly throughout the entire human genome cosegregate with phenotypic traits. In contrast to candidate gene association studies, the GWL approach is robust with respect to population admixture/stratification. Because GWL studies do not rely on linkage disequilibrium among genes or markers in adjacent genomic regions, it is a promising approach for identifying genomic regions contributing to relatively large variations in complex traits, without any prior knowledge about the potentially important function of specific genomic regions. Both univariate linkage analyses and bivariate linkage analyses were used extensively in GWL studies between October 2007 and November 2009. Specific details regarding these two types of GWL studies are outlined in Table 3.

A. Univariate linkage analyses

Most of the univariate GWL studies focused on BMD variations. Quantitative trait loci (QTLs) were identified on chromosomes 1q42-43, 11q12-13, 12q23-24, 17q21-23, 21q22, 22q11 (26), 1q36 (27) and 15q13 (28) for LS BMD; on chromosomes 5q31-33, 13q12-14 (26) for FN BMD; on chromosomes 12p12 and 15q26 for hip peak BMD (PBMD); and on chromosomes 2p13 and Xq27 for wrist PBMD (29). To identify genetic factors influencing bone loss, Shaffer et al. (30) measured the 5-yr change in BMD in 300 Mexican-Americans and found that chromosomes 6q and 3p were linked to BMD changes at the hip and the distal third of the ulna, respectively. Another frequently studied phenotype is bone size (BS). Preliminary evidence shows that spine length is linked to chromosome 5 (31). It is important to recognize that the univariate linkage analyses discussed above have limited statistical power and the linkage regions identified are relatively broad.

B. Bivariate linkage analyses

Bivariate linkage analyses may improve statistical power considerably and facilitate the identification of QTLs whose effects are too small to be detected by univariate linkage analyses for genetically correlated traits. Efforts have been made to explore the correlations between BMD and BS, total body lean mass (TBLM), body fat mass (BFM), and age at menarche (AAM). Significant linkage was observed on chromosome Xq27 for BMD and BS (32) and on chromosomes Xq25 and 15q13 for spine BMD and TBLM in 4126 Caucasian individuals from 451 pedigrees (33,34). A strong linkage signal influencing both BMD and AAM was found on chromosome 22q13 in 2522 females from 414 Caucasian pedigrees (35).

In total, more than 60 QTLs have been identified, and they have been found on all chromosomes, with the exception of chromosome Y. A number of QTLs have been replicated by at least two studies, such as 7p21-22 (29,33,34), 11q12-13 (26,29,34), 15q13 (28,33,34,35,36), and Xq27 (29,32,35). 7p21-22 and 11q13 are of particular interest because these regions harbor two strong candidate genes, IL6 in 7p21-22 and LRP5 in 11q13. However, many of the QTLs have not been replicated in independent studies. A significant limiting factor in replicating these linkages is genetic heterogeneity, especially when the sample size is relatively small or participants from various ethnic origins are included. In addition, other factors such as magnitude of the genetic effect, density of markers, definition and assessment of phenotypes, and statistical approaches might contribute to difficulties in replicating the findings of GWL scans. Despite these concerns and limitations, significant progress has been made in the last 2 yr. As the number of GWL studies continues to grow and more new GWA studies start to emerge, it is anticipated that some previously identified genomic regions will be replicated for linkages and that some genomic regions will eventually be proven to be falsely positive.

IV. GWA Studies

In contrast to candidate gene studies that select genes for study based on known or suspected disease mechanisms, GWA study is a non-hypothesis-driven approach in which a large number of subjects are genotyped for dense genetic markers covering the genome, usually in the form of SNPs and copy number variations (CNVs), in an unbiased fashion. Consequently, GWA studies have the potential to identify totally novel genes/genomic loci with modest effects on human complex diseases/traits (37). However, GWA studies have their own issues that involve multiple considerations in study design and data analyses, such as sample size, level of statistical significance, correction for multiple testing, population stratification, marker density, and replication of results by independent studies. Shortly after the initial GWA study on osteoporosis (38), results of 12 more GWA studies were published (Table 4). Nine of these were published in the New England Journal of Medicine (NEJM) (39), The Lancet (40), Nature Genetics (41,42,43,44), Annals of Internal Medicine (45), and The American Journal of Human Genetics (AJHG) (46,47).

A. Single nucleotide polymorphism (SNP) analyses

In the NEJM paper, sequence variants in five genomic regions were significantly associated with spine BMD in the discovery set (5861 Icelandic subjects) and were confirmed in the replication sets (3750 Icelandic, Danish, and Australia subjects) (39). Three of these five regions are close to, or within, candidate genes (RANKL, OPG, ESR1) for osteoporosis that were previously implicated by candidate gene association studies. The other two regions are close to the zinc finger and BTB (Bric-a-brac, Tramtrack, Broad-complex) domain containing 40 (ZBTB40) gene (1p36) and the major histocompatibility complex region (6p21). The ZBTB40 gene, of unknown function, is expressed in bone (UniGene accession no., Hs.418966), indicating a potentially unidentified role in bone biology. By enlarging the discovery sample size from 5861 to 6865 Icelanders and the follow-up replication sample size from 3750 to 5375 subjects of European descent, two new genome-wide significant loci for hip BMD were found near the SOST and MAPK3 genes (42). The SOST gene encodes sclerostin, which interferes with the Wnt pathway by disrupting Wnt-induced frizzled-Lrp complex formation. The MAPK3 gene encodes a protein kinase that phosphorylates microtubule-associated proteins and plays a role in determining cell polarity. However, 20p12 and BMP2, which the authors found to be associated with osteoporosis in their previous study of the same Icelandic population, were not confirmed in their current GWA papers. In the Lancet paper, GWAs with hip BMD and LS BMD were found for two SNPs: rs4355801 on chromosome 8, close to the OPG gene, and rs3736228 on chromosome 11 in the LRP5 gene (40). Further analyses in the replication cohort (6463 people from three other cohorts in Western Europe) corroborated the findings in discovery subjects (2094 women from the Twins UK cohort). The relationships of these two genes with bone metabolism have been well established in previous candidate gene association studies. In the Nature Genetics paper (41), rs7776725 within the FAM3C gene and rs1721400 mapping close to the SFRP4 gene showed association with BMD at the radius, tibia, and heel in a sample of 8842 Korean subjects from population-based cohorts and were replicated in an independent sample of 7861 Korean subjects. Previous evidence that the FAM3C gene is related to bone is rare, suggesting that this gene might be a novel candidate gene for osteoporosis. SFRP4 belongs to the SFRP family, members of which are well known for their involvement in bone formation and resorption and act as soluble modulators of the Wnt pathway (48).

In each of the GWA studies mentioned above, the tested subjects had similar ancestry, e.g., in the NEJM paper, all the participants were whites of European descent. Thus, the effects of the identified candidate genes are of interest to be investigated in individuals from other ancestries. In the AJHG paper (46), two genes, ADAMTS18 and TGFBR3, were associated with BMD in Caucasian U.S. subjects. Importantly, this association was replicated in Chinese samples and a Tabogo cohort of African ancestry (46). Furthermore, National Center for Biotechnology Information (NCBI) Gene Expression Omnibus expression profiles showed that decreased expression levels of ADAMTS18 and increased expression levels of TGFBR3 in vivo may potentially contribute to the nonhealing of skeletal fractures (46). Results of previous studies have suggested that the genetic basis for osteoporosis may be gender specific (49,50), so testing sex differences in GWA studies is encouraged. In the first GWA study of hip BS, Liu et al. (51) identified a novel gene, PLCL1, that had four SNPs associated with hip BS in female subjects. The importance of this gene to hip BS was replicated in an independent UK cohort, and the relevance to hip fractures was also observed in a Chinese sample (51). In male subjects from the Caucasian sample, two SNPs in SOX6 gene were identified to be associated with both hip BMD and body mass index by a bivariate GWA study (52), the approach of which has the potential to find pleiotropic genes underlying two genetically correlated diseases, like osteoporosis and obesity.

Due to the relatively small sample size in each dataset alone, the power to identify variants with small effects on osteoporosis is insufficient. Meta-analysis involves multidatasets, enlarges the sample size, and is designed to improve the power to detect more associations and investigate the consistency of associations across diverse study populations (53). Rivadeneira et al. (44) conducted a large-scale collaborative meta-analysis involving 19,195 participants from five study populations of European origin. A total of 20 loci showed associations with BMD at the genome-wide significance level, of which 13 were new for BMD and the remaining seven have been reported previously. For osteoporosis, it is important to further investigate whether the candidate genes associated with BMD are also associated with fractures. In a meta-analysis involving 19,195 participants with BMD data and 5,974 subjects with fracture data (45), of the nine candidate genes associated with BMD, four (LRP5, SOST, SPP1, and TNFRSF11A) were also significantly associated with risk for fractures. Potential noise may be introduced by including multiple populations that may be heterogeneous in environments and genetics factors and their interaction in influencing the study traits. The usefulness of meta-analyses may partially depend on the noise introduced relative to the genetic signal enhanced in larger samples. Therefore, findings from even meta-analyses should be replicated in various individual populations for their robustness.

Although the GWA studies reviewed above have brought us new insights into the genetic mechanisms of osteoporosis, the clinical value of these GWA studies is not immediately obvious due to the small variation (namely, ∼1–4%) in BMD that is accounted for by each of the implicated SNPs. It is likely that many other sequence variants imparting smaller effects on osteoporosis will be uncovered, but the primary value of these studies is to identify targets for future functional studies that enhance our knowledge of the molecular pathogenesis of osteoporosis, which will, in turn, facilitate the development of novel therapeutic modalities targeting these molecular pathways.

Approximately 30,000 genes are coded by the human genome, and currently, GenBank contains only about 16,000 known genes. Thus, SNPs associated with osteoporosis in future GWA studies may fall into genomic regions with known and predicted genes or into genomic regions that are more poorly characterized. For SNPs that are located within genes or within regulatory regions close to known genes, public databases (e.g., HapMap, NCBI, University of California Santa Cruz Genome Browser, and Ensembl) or private databases (e.g., Applied Biosystems SNPbrowser, Celera, and Perlegen) can be used to characterize the linkage disequilibrium block structure around those positive SNPs and then to identify the genes annotated on or near the linkage disequilibrium blocks. The strategy for identifying unknown genes in these genomic regions is to blast the Expression Sequence Tag databases and then perform computational gene prediction using gene prediction software (Genscan, GrailEXP, Michael Zhang’s Exon Finder, GeneMark) and Vista analysis in combination with experimental confirmation by real-time RT-PCR in bone-related cells and tissues. Due to false-positives and false-negatives in gene prediction using current software, new and better gene prediction tools are being developed and are likely to become available. In addition, databases containing information on known and predicted genes will be continuously updated. The new tools for gene prediction and enhanced information about existing genes will enhance our capacity to identify genes that reside around positive association signals, and RT-PCR can be used to confirm any unknown candidates in major bone cells and bone-related tissues.

B. Copy number variation (CNV) analyses

CNVs are newly appreciated structural genomic variations that include duplication or deletion of genomic segments. In contrast to SNPs, which only represent a single nucleotide, CNVs represent a relatively large segment of DNA, with variations in size ranging from 1 kilobase to several megabases. Using the Affymetrix 500K Array Set, Yang et al. (47) conducted case-controlled genome-wide CNV analyses in elderly Chinese individuals. A variant of UGT2B17 in CNV 4q13.2 showed association with osteoporotic fractures that was confirmed in replication cohorts. Because the UGT2B17 gene encodes a key enzyme catabolizing steroid hormones, the relationship between serum levels of sex hormones and UGT2B17 gene copy number was further assessed. Subjects with homozygous deletions of UGT2B17 were found to have significantly higher concentrations of total testosterone and estradiol than subjects with one or two copies of this gene. Compared with the well-developed resources available for SNP association studies, we are still in the early phases of incorporating structural genetic variation in GWA studies. GWA CNV analysis generates a distinct set of statistical and technical challenges, and the list of CNVs available for detection and analysis is undoubtedly incomplete (54). Over the next few years, however, a much more global understanding of the extent and precise location of CNVs will likely be achieved as new platforms and novel technical, statistical methods become available to accurately capture CNV information. These advances will undoubtedly benefit the complex phenotype studies, providing a more comprehensive understanding of the role of CNVs in the pathogenesis of osteoporosis. Despite the many obstacles yet to be overcome, we expect analyses of CNVs to continue to gain importance in osteoporosis studies.

The variants themselves found in candidate gene association and GWA studies may not directly cause higher risk to osteoporosis. They may just be closely linked to the actual causal variants. Therefore, further analyses, such as sequencing DNA base pairs in the implicated genome, are needed to identify the exact genetic change responsible for the development of osteoporosis.

V. Transgenic/Knockout Mouse Models

Transgenic mice contain artificially introduced inducible or tissue-specific transgenes that have been added to the genome to study the function and regulation of those genes. These transgenes often confer a gain of function by producing a new protein, increasing expression levels of an existing protein, or presenting the protein in a different type of cells. However, if the transgene disturbs the expression of an existing gene, it will cause loss of function, which is a strategy that has been used to generate knockout mice. Knocking out the activity of a specific gene can offer valuable clues about the function of that gene. Because humans share many genes with mice, detailed study of transgenic/knockout mice has been used to provide a better understanding of how similar genes may contribute to similar complex traits in humans. Knockout and transgenic mouse models have been widely used to confirm the contribution of many well-known or novel genes to abnormal bone phenotypes. From October 2007 to November 2009, approximately 70 new studies (summarized in Table 5) using transgenic and knockout mouse models were reported in the osteoporosis field. For instance, in a study using knockout and overexpression manipulation, bone mass in Cthrc1-null mice was significantly lower than in Cthrc1 transgenic mice (55). Using overexpression manipulation, Ding et al. (56) observed increased osteoblastic activity in glucose-dependent insulinotropic peptide transgenic mice. Transgenic animal models also help isolate and highlight the functions of a particular gene, providing further evidence to support the gene’s relevance to osteoporosis. For example, to determine the role of steroid receptor coactivator (SRC)-1 in mediating the maintenance effects of estrogen on bone, Mödder et al. (57) used SRC-1 knockout mice with gonadectomy and estrogen replacement (10 μg/kg/d) to show that the skeletal response to estrogen was impaired in female, but not in male, SRC-1 knockout mice. It is important to note that there are also some important limitations to the use of knockout mouse technology, as reviewed by the National Institutes of Health (http://www.genome.gov/12514551). Generally, approximately 15% of the genes that have been knocked out are developmentally lethal, so these genes cannot be studied using a knockout model. Additionally, despite the genetic similarities of mice and humans, there are still important differences, so knocking out a gene in mice may have effects that are not indicative that genes function in humans. Despite these limitations, knockout mice give us one of the most powerful methods now available for studying gene functions in vivo.

VI. Gene-Expression Microarray Studies

Because osteoporosis is largely determined genetically, the etiology of osteoporosis cannot be explicitly and comprehensively revealed unless the mechanisms of gene action underlying osteoporosis are established. This problem has been partially addressed by gene expression studies of osteoporosis. Gene expression studies explore relationships between diseases and certain genes at the mRNA level rather than at the DNA level. This approach can shed light on gene function and can also be used to elucidate some of the intermediate biochemical processes leading to a disease. In contrast, genetic epidemiology studies at the DNA level obviate the study of biological processes. Gene-expression microarray has proven to be a powerful tool for assessing gene expression for tens of thousands of genes simultaneously. Approximately 110 gene-expression microarray studies of osteoporosis and other related traits have been published since our 2007 review. These studies have focused primarily on a few key topics such as regulation of osteoblast and osteoclast activity, differentiation of mesenchymal stem cells (MSCs), gene expression comparing healthy and diseased tissues, effects of therapeutic agents on the healing of fractures, and endocrine regulation of bone remodeling. The results from these studies are summarized in Table 6.

A. Regulation of osteoblast and osteoclast activity

Regulation of osteoblast and osteoclast activity has been a major focus of microarray experiments in the past 2 yr and has provided novel insights into the mechanisms of bone formation and resorption. The activity of osteoblasts and osteoclasts is regulated by a variety of specific biomaterials, transcription factors, growth differentiation factors, and cytokines.

Several important biomaterials have been identified that promote bone formation by regulating the activity of osteoblasts. For example, Palmieri et al. (58) used microRNA (miRNA) microarray techniques to investigate translation regulation in human MG63 cells cultured with PerioGlas, an alloplastic material that is effective as an adjunct to conventional surgery for the treatment of intrabony defects. There were 10 up-regulated and two down-regulated miRNAs that in turn regulate the expression of several genes by either enhancing catabolism or repressing translation of their mRNA targets. The vast majority of these genes are down-regulated by miRNA, some of which are homeobox genes like NOG, EN1, and CHRD. Other down-regulated genes include receptors like GHRHR and extracellular matrix proteins like COMP. Among the genes that were down-regulated by PerioGlas, NOG, and COMP were up-regulated by Bio-Oss, a deproteinized sterilized bovine bone material containing calcium-deficient carbonate apatite (59). Palmieri et al. (60,61,62,63,64,65) also explored osteoblast regulation by several other biomaterials including zirconium oxide, calcium sulfate, porous polyethylene, titanium, and anatase. These materials were found to alter expression of several genes related to osteogenesis and bone remodeling, such as SHOX, IGF-I, GHRHR, BMP1, and FGFR1. Additional biomaterials like hydroxyapatite (HA), bioactive glass, stainless steel, polymethylmethacrylate (66), and demineralized freeze-dried bone allografts (67) were also used to study early osteogenesis. Results of the gene expression studies reviewed above have enhanced our understanding of the molecular mechanism underlying osteoblast function in bone regenerative procedures. This new knowledge can aid in the development of surgical techniques and biomaterials that are more suitable for use in the treatment of fractures in osteoporotic patients.

The recent discoveries of transcription factors and signal transduction pathways critical for osteoblast differentiation have opened up new approaches to understanding the pathogenesis of osteoporosis. The identification of the critical role for Osterix/Sp7, a member of the Sp1 transcription factor family, in bone formation and osteoblastogenesis is of particular interest. Using Affymetrix GeneChip arrays with collagen tripeptide-treated human osteoblastic hFOB1.19 cells, Tsuruoka et al. (68) identified 169 genes that were up-regulated more than 2.12-fold. Among them, Sp7 transcription factor showed the greatest change in expression. Sp7 was found to be regulated by BMP2 signaling through Msx2 and Runx2 during osteoblast differentiation (69). A novel inhibitory pathway of osteoblast function, namely the cyclooxygenase-2 (COX-2) pathway, was described by Silvestris et al. (70). Their study supports the role of E4BP4 as a negative osteoblast transcriptional regulator of the COX-2 pathway, by negatively regulating the suppressive COX-2 pathway, thus up-regulating the expression of both Runx2 and Sp7. In view of these findings, it will be quite interesting to determine whether polymorphisms of these transcription factors are associated with osteoporosis.

Gene expression studies of osteoclastogenesis have also provided important new insights into the maintenance of bone homeostasis. Circulating monocytes, which serve as early progenitors of osteoclasts, were collected from human premenopausal subjects with extremely low vs. high peak bone mass, and the expression of three genes (GBP1, CXCL10, and STAT1) was significantly different in these two populations (71). SDF1 and CXCL7 were identified as osteoclast enhancers in primary mouse bone marrow cells treated with IGF-II (72). Battaglino et al. (73) identified a novel gene, NHA-oc/NHA2, that was strongly up-regulated during RANKL-induced osteoclast differentiation in vitro and in vivo. NHA-oc/NHA2 is a new member of the cation-proton antiporter and is the first mitochondrial NHA characterized to date. In addition, Kominsky et al. (74) discovered a novel osteoclast stimulating factor, macrophage inflammatory protein-1 δ, which is secreted by renal cell carcinomas that metastasize to bone. Collectively, these studies have helped fill important gaps in our knowledge regarding regulation of osteoclastogenesis, thereby enhancing our fundamental understanding of the pathogenesis of osteoporosis.

B. Proliferation and differentiation of mesenchymal stem cells (MSCs)

MSCs, which can differentiate into osteoblasts, chondrocytes, adipocytes, and myoblasts in vivo and in vitro, have been regarded as the most hopeful cell sources for bone tissue engineering, regenerative medicine, and gene therapy. Osteoporosis might be due to defects in MSCs that lead to reduced proliferation and osteoblast differentiation. However, the use of MSCs as seeding cells in bone tissue engineering is hampered by a lack of fundamental knowledge about the molecular mechanisms underlying osteogenic differentiation of MSCs. During the past 2 yr, several studies have explored the molecular mechanisms underlying osteogenic differentiation of MSCs (75). Evidence suggests that the capacity of MSCs for osteogenic differentiation was highly suppressed during late culture stages. The expression of four genes, EPHA5, NOV, NDN, and RUNX2, varied depending on culture stage, suggesting that these genes have the potential to act as stage-specific markers during osteogenic differentiation of MSCs (75). Purkinje cell protein 4 (Pcp4), a gene involved in the deposition of calcium and the modulation of calmodulin-dependent protein kinase, was found to be increased during osteoblastic differentiation of MSCs in vitro (76). Bmp1 and Bmp7 were up-regulated, whereas Egf and Egfr were down-regulated during osteogenesis of MSCs subjected to electromagnetic field treatment (77). In addition, simulated microgravity, inorganic polyphosphate, and phosphodiesterases all play important roles in the regulation of osteoblastic differentiation from MSCs by inhibiting population growth of MSCs or by activating fibroblast growth factor (FGF) signaling pathways to induce both proliferation and mineralization of MSCs (78,79,80). However, the specific signaling pathways controlling gene expression in the osteogenic differentiation process are still not clear. Further study of the molecular mechanisms underlying osteogenic differentiation of bone marrow cells will have important theoretical and practical significance in bone regeneration and repair.

C. Gene expression in healthy vs. diseased tissues

Comparative gene expression studies of healthy vs. diseased tissues can be quite informative because they provide immediate information about the differential regulation of genes that may contribute to the disease process. Two studies by Hopwood et al. (81,82) used microarray analysis to compare expression levels of thousands of genes from osteoporotic individuals to age-matched osteoarthritic or control individuals; for these studies, they investigated trabecular bone from the intertrochanteric region of the proximal femur. A substantial number of the top-ranking differentially expressed genes identified in osteoarthritic bone are known to play roles in osteoblasts, osteocytes, and osteoclasts. Many of these genes are targets of either the Wnt (TWIST1, IBSP, S100A4, MMP25, RUNX2, and CD14) or TGF-β/BMP signaling pathways (ADAMTS4, ADM, MEPE, GADD45B, COL4A1, and FST) (81).

D. Effects of therapeutic agents on the healing of fractures

An enhanced understanding of the molecular mechanisms by which therapeutic agents affect the healing of fractures has provided substantial motivation for performing expression profiling studies of bone repair. TP508, a 23-amino acid peptide that represents the thrombin-binding domain for a specific class of thrombin receptors, has a wide range of therapeutic effects on tissue repair (83). By comparing the translational profiles of fractured femurs treated with TP508 vs. saline controls, Li et al. (84) demonstrated that TP508 accelerated fracture healing by modulating expression levels of proteins primarily involved in the functional categories of cell cycle, cell growth, proliferation, and cell death. In another study of healing in rats treated by titanium implant vs. osteotomy, Kojima et al. (85) determined that 86 gene transcripts, including extracellular matrix-related, bone resorption-related, and proteoglycan gene transcripts, were up-regulated in at least one time point. Further studies are needed to identify the biological roles of the transcripts in osteointegration. These data not only provide new insights into fracture healing physiology, but also provide the rationale of potential new treatment strategies for fractures.

E. Endocrine regulation of bone remodeling

Systemic effects of hormones and growth factors play an important role in physiological and pathological mechanisms of bone remodeling. PTH, which is currently used in the therapy of osteoporosis, has been studied intensively. One important study of PTH and bone investigated the anabolic and catabolic effects of intermittent and continuous treatments with three different PTH peptides, PTH (1-34), (1-31), and (3-34) (86). This study identified and validated a large number of genes such as slpi, tfpi2, socs3, and gro1, which were previously not considered to be expressed in bone or to be regulated by PTH treatment. The precise function of these newly identified genes in the anabolic and catabolic bone state, if any, deserve further investigation. In using mouse whole-genome cDNA microarrays to assess the responses to PTH (1-34) and PTHrP (1-141) in the Kusa 4b10 mouse marrow stromal cell line, members of the ephrin and Eph family were identified as targets of PTH (1-34) and PTHrP (1-141) (87). Among the regulated genes, ephrinB2 mRNA was up-regulated in response to both PTH and PTHrP. EphrinB2 protein might act in a paracrine or autocrine manner on the osteoblast itself to stimulate osteoblast maturation and/or bone formation under the influence of local PTHrP or administered PTH (87). Glucocorticoid excess can induce alterations in bone metabolism that weaken bone structure and increase fracture risk. Using microarrays in glucocorticoid excess-treated mice, Yao et al. (88) revealed that glucocorticoid excess was associated with early activation of genes associated with osteoclastogenesis (csf1, c-fms, lbsp, and ltgb3) and adipogenesis (c/EBPα and PPARγ) and a later suppression of genes associated with osteogenesis and mineralization (TGFβ1, BMP2, LEF1, Akp2, and MAPK). These gene expression changes may correspond to alterations in bone metabolism with glucocorticoid exposure that result in rapid bone loss. These results enhance our understanding of glucocorticoid-induced bone loss by providing in vivo evidence supporting the concept that glucocorticoid excess directly or indirectly regulates the transcription of specific genes associated with bone physiology.

Besides the aforementioned studies, during the past 2 yr, fruitful results were also generated by microarray studies in several additional bone biology research fields, including osteosarcoma biology (89,90), spine fusion (91), regulation of osteolysis (92), and osteoblast- or chondrocyte-mediated pathological processes in bone diseases such as ossification of the posterior longitudinal ligament (OPLL) of the spine (93) and osteoarthritis (OA) (94,95,96). These studies will not only illuminate new methods for bone biology research but also are likely to provide significant insights that will contribute to the development of new therapeutic interventions in the near future.

As our previous review pointed out, most high-throughput gene expression studies have used cultured cell lines of humans, mice, and rats in vitro. The disadvantage of using cultured cell lines is that gene expression profiles undoubtedly change because cultured cells gradually lose their osteogenic potential due to lack of stimulation by factors that are available in vivo, but not in vitro. Given this significant limitation, studying fresh cell- or bone-related tissues may be a promising way to obtain data that more closely resemble in vivo conditions.

VII. Proteomics Studies

Because cell recognition and signal transduction pathways occur at the protein level (e.g., receptor/ligand interactions, antigen recognition, and cell adhesion), it is appropriate and important to utilize proteomics approaches to directly investigate protein expression. In addition, because of complicated processes such as alternative mRNA splicing and posttranslational modification of proteins, the correlation between expression levels of mRNA and proteins could be low (97,98). The Human Genome Project revealed that there are far fewer protein-coding genes in the human genome than there are proteins in the human proteome [∼35,000 genes give rise to ∼1,000,000 proteins (99)]. This profound discrepancy clearly indicates that the diversity of proteins cannot be fully characterized by gene-expression analysis alone. Consequently, to understand functional genomics, proteomics approaches have become an indispensable complement to mRNA expression microarrays. Combining the data generated by these two approaches provides much richer information than that provided by genomic studies alone.

A. Studying cultured cells using in vitro systems

Proteomics has only recently been applied to the bone field, so only a few relevant studies have been published in the past 2 yr. Most of these studies involve expression proteomics, in which samples from case and control groups are quantitatively analyzed to identify proteins that are differentially expressed between groups. The most common approach was to use in vitro cell culture systems, and several proteins that are important for the development of MSCs (100,101,102,103,104), osteoblasts (105,106,107,108,109), osteoclasts (110,111,112,113), chondrocytes (114,115), bone marrow osteoprogenitor cells (116), and osteosarcoma cells (117) were identified in this manner. For example, by applying two-dimensional liquid chromatography/matrix-assisted laser desorption/ionization mass spectrometry (MS) on MSC-derived fast-growing clones (with tripotential differentiation capacity) vs. MSC-derived slow-growing clones (with only unipotential differentiation capacity), Mareddy et al. (102) identified 11 proteins (e.g., calmodulin, tropomyosin, and caldesmon) with differential expression. Using iTRAQ-coupled two-dimensional-liquid chromatography-tandem MS/MS analysis, Xu et al. (105) found several cytoskeletal proteins, metabolic enzymes, signaling and cell growth proteins that were differentially expressed in human osteoblasts cultured on plane HA vs. osteoblasts cultured on carbon nanotube-reinforced HA. These findings of in vitro proteomics studies, however, may be limited by the fact that factors regulating gene expression in complex physiological/pathological environments in vivo can be changed or eliminated in vitro. Consequently, in vitro proteomics studies are likely to be compromised by this comparative lack of regulatory control.

B. Studying fresh cells, serum, or tissues

Compared with in vitro studies, ex vivo studies using fresh cells or tissues are more likely to provide data that represent protein expression under physiological/pathological conditions. Our group recently used two-dimensional electrophoresis (2-DE) coupled with MS to perform a comparative protein expression profiling study of circulating monocytes from premenopausal Chinese females with extremely high BMD vs. those with extremely low BMD. In total, 38 differentially expressed proteins were identified, and five of these proteins (GSN, FSU1, SOD2, GPX1, and P4HB) were confirmed by Western blotting (118). Moreover, several additional studies of serum and fresh tissues have been performed in the past 2 yr to investigate the global-scale molecular profiling of bone-related diseases, such as OA (114,119,120,121), osteonecrosis of the femoral head (122), and osteosarcoma (123). Using 2-DE, a comparative analysis of the proteins extracted from patients with OA cartilage vs. those with normal cartilage was performed (119). After further identification by linear ion trap-Fourier transform ion cyclotron resonance MS, 14 proteins associated with OA were unambiguously determined, including proteins involved in glycolysis and energy production (ADH, ADK, ENOA, KPYM, and FR), signaling (ANNX-I, PEBP, and TUB), Redox (PRDX3 and SODM), and cartilage matrix (COLL-I and COLL-VI). In another study using integrated genetic, bioinformatic, and proteomic approaches, five novel proteins (SOX11, FGF23, KLF6, WWOX, and GDF15) were implicated in the genesis of OA (120). Based on the methods of isoelectric focusing, 2-DE, and silver staining, as well as matrix-assisted laser desorption ionization time-of-flight mass spectrometry, levels of kininogen 1 variant, complement factor C3 precursor, and complement factor H were found to be increased, whereas levels of antithrombin III chain B, apolipoprotein A–IV precursor, and gelsolin isoform α precursor were decreased in osteonecrosis of the femoral head patients (122). By comparing the protein expression profiles of two distinct groups of osteosarcoma biopsy samples, a chemosensitive group vs. a chemoresistant group, Kawai et al. (123) identified 10 protein spots associated with the chemosensitivity of osteosarcoma to preoperative chemotherapy. The proteins represented by these 10 spots could potentially be new diagnostic or prognostic markers for osteosarcoma, or new therapeutic targets for the disease.

Proteomics represents a promising field that is poised to boost our understanding of the dynamic nature of protein expression, cellular and subcellular protein distribution, posttranslational modifications, and protein-protein interactions. However, the application of proteomics to the bone field is only in its initial stages. Current proteomics studies in osteoporosis focus mainly on expression proteomics. Because key proteins involved in osteoporosis development generally interact with other proteins, functional proteomics studies will be necessary to provide a global understanding of protein-protein interactions. To better understand and even predict the functions of proteins, additional knowledge regarding three-dimensional structures of the proteome is required. Structural proteomics may prospectively fulfill this goal by mapping out the structures of protein complexes or proteins in a specific cellular organelle.

The impact of advances in the field of proteomics on our knowledge of osteoporosis is relatively small at this point. As new techniques are developed and applied to bone, however, we anticipate that the abundance of new information generated will be crucial toward understanding the pathogenesis of osteoporosis and developing novel therapeutic strategies for treating this disease.

VIII. Future Directions

A. Functional studies

After genes for osteoporosis have been identified, it is important to perform functional studies to determine the influence of such genes on the differentiation of osteoblasts or osteoclasts and the effects of such genes on the variation of bone-related traits. The strategies may involve in vitro and/or in vivo assays to test the effects of the gene of interest in cell cultures or animal models. Strategies for in vitro assays often include overexpression or knockdown of gene expression in osteoblast or osteoclast precursor cells, which is followed by evaluation of alkaline phosphatase (ALP) or tartrate-resistant acid phosphatase (TRAP) activity to assess osteoblast or osteoclast differentiation. To determine the potential molecular mechanisms by which the expression of target genes are regulated through specific transcription factors, EMSAs and chromatin immunoprecipitation can also be employed. Should strong functional evidence be obtained in vitro, one will consider further in vivo investigations using knockout and/or transgenic approaches to examine how the target gene changes the bone-related phenotypes, e.g., BMD or BS. In vivo functional analyses are admittedly challenging for complex traits partially due to biological redundancy and the significant potential for knockdown of critical genes to prevent in utero development.

Furthermore, the specific impact of potential functional variants from the gene of interest will be ascertained through, e.g., promoter assay, RNA splicing assay, and gene activity assay. First, if SNPs change recognition sequences of potential transcriptional factors, binding affinity to the transcriptional factors may be modified, leading to changes in transcriptional efficiency. Investigators can perform EMSAs and luciferase reporter assays to examine this potential effect. Second, for SNPs located in the splice-site, one can detect alternative mRNA splicing using mRNA phenotyping protocols (124). Third, if SNPs lie in the coding region, constructs that differ in the target SNPs can be prepared and expressed in vitro. The activity of different protein products can then be compared using appropriate assays for determining the function of that specific protein.

B. Epigenetic variation

Epigenetics refers to reversible, heritable changes in gene regulation that occur without a change in DNA sequence. Epigenetic regulation has been implicated as a key regulatory mechanism in the etiology of human complex diseases (125). miRNA regulation, DNA methylation, and histone methylation are three common types of epigenetic modifications. Previous pioneering studies have shown that DNA methylation may be involved in the osteoclastogenesis (126), and acetylation of histone H3 and H4 may be involved in osteogenesis (127). Genome-wide measurement of epigenetic variation has recently been made possible using techniques such as Affymetrix Human Tiling 2.0R Array Set, which will allow us to progress toward a thorough understanding of the roles of epigenetics in osteoporosis.

C. New phenotypes

Active shape modeling has been used to create a template describing the outline of the hip joint from dual energy x-ray absorptiometry images. Using this method, Goodyear et al. (128) identified risk factors for hip fractures independent of BMD. Composite indices of FN strength that are also constructed from hip dual energy x-ray absorptiometry images, such as compression strength index (CSI), bending strength index, and impact strength index, have the potential to improve hip fracture risk assessment (129). Although two studies have made efforts in finding genetic variants for CSI (130,131), more genetic studies for these new phenotypes are necessary.

IX. Future Prospects for the Application of Genetic Risk Assessment in Osteoporosis Prediction and Treatment

Genes that have been implicated in osteoporosis by genetic studies only make minor contributions individually to bone density and fracture risk. For example, the five regions identified in the study of Styrkarsdottir et al. (39) accounted for only approximately 3% of the total variation in hip and spine BMD. Similarly, the contribution of the CNV of UGT2B17 gene to variations in BMD, cortical thickness (CT), and buckling ratio (BR) in Caucasians was only 0.67, 0.71, and 0.77%, respectively (47). It is thought that many genes, each with small effects, may be responsible for osteoporosis, rather than a small number of genes with large effects (132). Thus, osteoporosis predictive tests are likely to involve many genes, and tests for single genes are unlikely to be of clinical significance. Therefore, in theory, if most of the major genes that cause osteoporosis can be identified and their interaction with each other and with environmental factors can be understood, this information can be used to identify those who are at risk. However, the depth of our knowledge currently falls far short of this goal. Because molecular genetic studies in the field of osteoporosis are coming at a rapid rate, we are hopeful that, in the next one or two decades, sufficient genetic information will become available to develop genetic algorithms to assess the risk for osteoporosis.

Genetic studies have established a link between osteoporosis in humans and the Wnt, BMP-Smad, and 12/15-lipoxygenase pathways. Recently, the EphrinA-EphR pathway was found to be associated with FN bone geometry section modulus by pathway-based GWA analysis (133). It is likely that new pathways related to osteoporosis will continue to be identified by genetic studies, such as pathway-based GWA analysis. Furthermore, proteomics approaches could characterize the disease process directly by finding sets of proteins that work together to produce disease. Although our fundamental knowledge is fairly limited, there is a general feeling that bone-related pathways provide a substantial potential for developing novel ways to treat osteoporosis (134). Current emphasis in the development of new anabolic therapies for osteoporosis is directed at modifying the effects of Wnt pathway on osteoblast differentiation and bone formation (135). However, due to the pleiotropic function of these various pathways, there is substantial concern that it will be difficult to manipulate these pathways for osteoporosis treatment without producing unwanted side effects on other cells and tissues. In the coming years, basic and clinical efforts will be required to explore new pathways and modes of translating knowledge about the contribution of these pathways to the pathogenesis of osteoporosis into therapeutic applications.

X. Summary

In this article, we have updated the progress of genetic studies of osteoporosis published from October 2007 to November 2009 at three levels, corresponding to DNA, mRNA, and protein. Since our last update toward the end of September 2007, remarkable progress has been made in revealing the genetic basis of osteoporosis. A number of promising candidate genes, genomic regions, or proteins were identified, and several of them have been replicated by multiple studies. However, the majority of these findings are still inconclusive, pending further investigation. Efforts must be made in various aspects, including controlling population stratification and genetic heterogeneity, performing larger-scale studies in possibly hundreds of thousands of subjects, exploring new phenotypes, considering gene-gene and gene-environment interactions, deep resequencing of the genome, investigating the contribution of CNVs, DNA methylation, and histone modifications, and identifying the causal variants and their biological roles in osteoporosis. We expect that the results of these studies will be used to develop rational approaches toward predicting whether or not individuals are predisposed to developing osteoporosis and novel therapeutic approaches toward the treatment of this important disease.