Cornell University Method for isolating high quality RNA

1

2

3Q2

4Q3

5678

9

10111213141516

17181920212235

3637

38

39

40Q6

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

Bone xxx (2014) xxx–xxx

Q1

Q4

BON-10438; No. of pages: 5; 4C:

Contents lists available at ScienceDirect

Bone

j ourna l homepage: www.e lsev ie r .com/ locate /bone

Technical Note

A method for isolating high quality RNA from mouse cortical andcancellous bones☆

OFNatalie H. Kelly a,b, John C. Schimenti c, F. Patrick Ross d, Marjolein C.H. van der Meulen a,b,d,⁎

a Sibley School of Mechanical and Aerospace Engineering, Cornell University, 105 Upson Hall, Ithaca, NY 14853, USAb Department of Biomedical Engineering, Cornell University, 101 Weill Hall, Ithaca, NY 14853, USAc College of Veterinary Medicine, Cornell University, Ithaca, NY 14853, USAd Research Division, Hospital for Special Surgery, 541 East 71st St., New York, NY 10021, USA

☆ Financial support: National Institutes of Health grantGRFP (NHK).⁎ Corresponding author at: Sibley School of Mechanic

Cornell University, 219 Upson Hall, Ithaca, NY 14853, USAE-mail addresses: [email protected] (N.H. Kelly), jcs9

[email protected] (F. Patrick Ross), [email protected] (M.C.H

http://dx.doi.org/10.1016/j.bone.2014.07.0228756-3282/© 2014 Elsevier Inc. All rights reserved.

Please cite this article as: Kelly NH, et al, Ahttp://dx.doi.org/10.1016/j.bone.2014.07.0

O

a b s t r a c t

23

24

25

26

27 Q528

29

30

31

32

33

34

Article history:Received 30 June 2014Revised 18 July 2014Accepted 19 July 2014Available online xxxx

Edited by: Sundeep Khosla

Keywords:OsteoporosisAnimal modelsGene expressionCancellous boneCortical bone

TED P

R

The high incidence of fragility fractures in cortico-cancellous bone locations, plus the fact that individual skeletalsites exhibit different responsiveness to load and disease, emphasizes the need to document separately geneexpression in cortical and cancellous bones. A further confounding factor is marrow contamination since itshigh cellularity may effect gene expression measurements. We isolated RNA from cortical and cancellousbones of intact mouse tibiae, and also after marrow removal by flushing or centrifugation. RNA isolated fromcancellous bone by each method was sufficient for gene expression analysis. Centrifugation removed contami-nating cells more efficiently than flushing, as indexed by histology and decreased expression of Icam4, a highlyexpressed erythroid gene. In contrast, centrifuged cortical bone had 12- and 13- fold higher expression of thebone-related genes Col1a1 and Bglap, while levels in marrow-free cancellous bone were 30- and 31-fold higherwhen compared to bone where marrow was left intact. Furthermore, cortical bone had higher expression ofCol1a1 and Bglap than cancellous bone. Thus, RNA isolated by this novel approach can reveal site-specific changesin gene expression in cortical and cancellous bone sites.

© 2014 Elsevier Inc. All rights reserved.

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

UNCO

RRE

Introduction

Osteoporosis is a skeletal disease characterized by low bone massthat often results in fracture. A majority of osteoporotic fractures occurat cortico-cancellous sites such as the proximal femur, distal forearm,and vertebra [1,2]. At these sites bone is a composite tissue consistingof an outer cortical shell surrounding an inner cancellous network.Mouse models have been developed to understand the etiology of oste-oporosis and to identify possible treatments for osteoporosis-relatedbone decline. Biophysical stimuli, including mechanical loading, are apromising approach to maintain or recover bone mass. In vivo loadingoften elicits a greater response in cancellous than cortical bone sites,increasing metaphyseal cancellous bone volume fraction by 49–73%anddiaphyseal cortical area by 41% in themouse tibia [3,4]. But differentloading protocols can alter cancellous and cortical bone responsesselectively; some stimulate cancellous bone to a greater extent, whileothers mainly affect cortical bone [5], indicating that differencesin mechanoresponsiveness may exist in cortical and cancellous bones.

75

76

77

78

79

80

s R01-AG028664 (MCHM), NSF

al and Aerospace Engineering,. Fax: +1 607 255 [email protected] (J.C. Schimenti),. van der Meulen).

method for isolating high qu22

To identify the nature and degree of change in these two bone tissuesrequires definitive data on which genes are altered at cortical and can-cellous sites following manipulation of test animals. To date this infor-mation is largely unknown, as most studies have used approaches thatare derived from homogenized bone tissues.

The cellular andmolecularmechanisms driving load-induced anabo-lism at the tissue level have been examined using both quantitative real-time polymerase chain reaction (qPCR) and microarray [6–9]. Thesestudies have examined tissues fromwhole bones, homogenizing corticaland cancellous bones. In contrast, only a single study examined the load-ing response in cancellous bone from the mouse vertebrae, but did notcompare the response to cortical bone [10]. Furthermore, marrowcells have been left in place [8], removed by flushing [7], centrifugation[6,9], or sequential collagenase digestion [10,11]. Collagenase digestionhas been shown to affect expression of bone-related genes [12], butthis effect may depend on the type of collagenase used; endotoxin-freecollagenase had a less detrimental effect on RNA quality [13]. However,failure to remove the highly cellular marrow component may result inextensive contamination of RNA isolated from unfractionated bone.The relative efficacy of flushing to centrifugation has not been studied,nor has the degree of bone marrow presence on gene expression.More importantly, gene expression has not been reported in minimallycontaminated cortical versus cancellous bone, a challenging endeavor inmurine models.

ality RNA from mouse cortical and cancellous bones, Bone (2014),

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122Q7

123

124

125

126

127

128

129

130

131

132

133

134

135

2 N.H. Kelly et al. / Bone xxx (2014) xxx–xxx

We describe a method to isolate high quality RNA from cortical and,separately, cancellous bonewith almost no contamination frommarrowcells. Both the quantity and quality of the RNA are suitable for tran-scriptome studies, providing the ability to compare cancellous andcortical tissues to elucidate differences in cellular andmolecularmecha-nisms at these two sites.

Materials and methods

Animals

Fifteen 11-week-old male C57Bl/6 mice were euthanized and theirtibiae rapidly dissected for histology (left leg) or RNA isolation and anal-ysis (right leg). The epiphysis, all soft tissues, periosteum, fibula, and thevery distal end were removed. In five mice, bone marrow of the tibiaewas not removed (no marrow removal, NMR group). In another fivemice, marrow was flushed with 1 mL of phosphate buffered saline(PBS) by inserting a 26-gauge needle into the distal tibia. The tibiae inthe third groupwere centrifuged for 20 s at 16,100 g at room temperaturein nestedmicrocentrifuge tubes, where the inner tube (0.2mL) had holescreated with a 26-gauge needle to allowmarrow to travel into the outertube (0.65mL). Tibiaewere cut approximately 5mmdistal to the growthplate to isolate the metaphysis. The metaphysis was held with forceps sothe proximal end was perpendicular to a 1 mm biopsy punch (Miltex,Integra LifeSciences Corp., Plainsboro, NJ, USA), which was used tomechanically separate a cancellous core from the cortical shell (Fig. 1).The IACUC of Cornell University approved all animal procedures.

Histology

Cancellous and cortical sampleswere placed in 10% formalin for 24 hfor fixation, and decalcified in 45% formic acid for 4 days with daily

UNCO

RRECT

Fig. 1. Schematic of cortical and cancellous bone sample preparation. Tibiae were rapidly dissewas left intact (group 1), flushedwith 1mL of PBS using a 26 g needle inserted distally (group 2cut approximately 5 mm distal to the proximal end to isolate the metaphysis, and a 1 mm bio

Please cite this article as: Kelly NH, et al, A method for isolating high quhttp://dx.doi.org/10.1016/j.bone.2014.07.022

ED P

RO

OF

exchange of the solution. Samples were processed and embedded inparaffin and sectioned longitudinally (6 μmsections, Leica RM2255, Buf-falo Grove, IL, USA). Sectioned samples were stained with hematoxylinand eosin (H&E).

RNA isolation from bone

RNA isolation was performed using Trizol (Life Technologies,Carlsbad, CA, USA) and RNeasy Mini kit (Qiagen, Germantown, MD,USA) as described previously [8]. Briefly, cancellous and corticalsamples (n = 30) were pulverized in liquid nitrogen-cooled flasks(Mikro-dismembrator S, Sartorius Stedim Biotech, Bohemia, NY, USA).Following pulverization, Trizol (Life Technologies) was added to theflasks and the powdered bone/Trizolmixwas incubated at room temper-ature for 45 min. 300 μL of chloroform was added to samples, vortexedfor 15 s, and decanted into phase lock gel tubes (PLG, heavy, 5 Prime,Gaithersburg, MD, USA). Samples were centrifuged for 15 min at 4 Cand 11,500 rpm, to separate the nucleic acid phase (~600 μL), whichwas removed and added to an equal volume of 70% ethanol. Thismixturewas applied to purification columns (RNeasy Mini kit, Qiagen) followingthe manufacturer's instructions, including a DNase digestion (RNase freeDNase kit, Qiagen). A final volume of 30 μL of RNAwas eluted, and purityand quantity were tested using a spectrophotometer (Nanodrop 1000,Thermo Scientific, Wilmington, DE, USA) and RNA Quality Number(RQN) using a fragment analyzer (Advanced Analytical Technologies,Inc., Ames, IA, USA).

RNA isolation from marrow

Marrow from centrifuged bones (n= 3)was placed into RNAlater(Qiagen) at 4 C until RNA isolation. Samples were centrifugedfor 5 min, RNAlater was removed, and 600 μL of lysis buffer with

cted and soft tissues, periosteum, epiphysis, fibula, and distal end were removed. Marrow), or centrifuged in nestedmicrocentrifuge tubes for 20 s at 16,100 g (group 3). Tibiaewerepsy punch mechanically separated a cancellous core from the cortical shell.

ality RNA from mouse cortical and cancellous bones, Bone (2014),

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

3N.H. Kelly et al. / Bone xxx (2014) xxx–xxx

β-mercaptoethanol (Buffer RLT, Qiagen) was added (as describedin RNeasy Mini instructions). Samples were homogenized by passingthrough an 18-gauge needle, centrifuged, and the supernatant(~600 μL) was added to ethanol and applied to columns for RNApurification as described above.

Gene expression

RNA was reverse transcribed to cDNA following the manufacturer'sinstructions (High Capacity cDNA Reverse Transcription kit, Life Tech-nologies) and brought to 5 ng/μL with RNase-free water. A final volumeof 20 μL containing 2× SYBR Green (Perfecta SYBR Green Fastmix, withROX, Quanta Biosciences, Gaithersburg, MD) was assayed by triplicateqPCR using 40 cycles of denaturing (95 °C, 15 s) and annealing/elonga-tion (60 °C, 60 s) (7300 Real-Time PCR System, Life Technologies).Primers for genes that are highly expressed in bone (Collagen type 1alpha 1, Col1a1, and bone gamma-carboxyglutamic acid-containing pro-tein, Bglap) andmarrow (intracellular adhesionmolecule 4, Icam4)weredesigned (available on request). Cycles to threshold (CT) was deter-mined for each gene and compared to the reference gene Gapdh. NMR,flushed and centrifuged groups and marrow samples are presented asrelative expression compared to Gapdh (2−ΔC

T). Fold-change to corticaland cancellous NMR groups was calculated by the 2−ΔΔC

T method [14].

Statistics

To test differences in relative gene expression between cancellousand cortical bone sites, and fold-change compared to NMR, a two-

UNCO

RRECT

Fig. 2. Centrifugation removes marrow more effectively than flushing in cortical and cancelloubones stained with H&E of (A,D) bones with no marrow removal, (B,E) bones where bone mTop row within each bone site is lower magnification and black rectangle outlines region of hi

Please cite this article as: Kelly NH, et al, A method for isolating high quhttp://dx.doi.org/10.1016/j.bone.2014.07.022

PRO

OF

factor ANOVA (main effects: marrow removal and site)with interaction(marrow removal ∗ site) was used with Tukey's HSD post-hoc (JMP Pro10, SAS, Cary, NC, USA). To test for differences in gene expressionbetween marrow and cancellous and cortical bone with marrow, aone-way ANOVA was used with Tukey's HSD post-hoc. Significancewas set at p b 0.05.

Results

Histology

To determine if centrifugation was more effective than flushing,we performed histology with H&E of longitudinal sections of cancellousand cortical bones (Fig. 2). Cortical (top panels) and cancellous bones(bottom panels) contained significant marrow that was removed in-completely by flushing. In marked contrast, marrow was barely detect-able in centrifuged samples.

RNA quantity/quality

Wealsodetermined if sufficient, high quality RNA could be extractedfrom cortical and cancellous bones separately and the effect of marrowcontamination onRNAquantity. Intact bonehadmore RNA thanflushedor centrifuged bone. Themean (±SD) concentration of cortical sampleswas 447 (±70), 279 (±13) and 45 (±37) ng/μL for NMR, flushed andcentrifuged bone, respectively, while those for analogous cancelloussamples was 186 (±54), 132 (±33) and 28 (±7) ng/μL. The averageconcentration for marrow samples was 191 (±130) ng/μL. The 260/

ED

s bones. Representative longitudinal 6 μm sections of (A–C) cortical and (D–F) cancellousarrow has been flushed, and (C,F) bones that have been centrifuged to remove marrow.gher magnification shown on bottom row. Scale bar = 400 μm.

ality RNA from mouse cortical and cancellous bones, Bone (2014),

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

4 N.H. Kelly et al. / Bone xxx (2014) xxx–xxx

280 ratio waswithin an acceptable range (1.8–2.1) for all samples. Can-cellous samples were of higher quality than their cortical counterpartswith a mean (±SD) RQN of 6.3 (±0.7) and 4.4 (±0.4), respectively.

Gene expression

We examined bone- andmarrow-related gene expression in corticaland cancellous bones as well as bone marrow and the effect of bonemarrow contamination. Centrifuged cortical bone had 12- and 13-foldhigher expression of Col1a1 and Bglap than bone with marrow, whilethe analogous data for centrifuged cancellous bone were 30-fold and31-fold higher, respectively (Figs. 3A/B). Flushing also increased bone-related gene expression when compared to no marrow removal, butthis effectwas not significant (2-fold and 3-fold, Col1a1 and Bglap in cor-tical and cancellous bones, respectively). The absolute expression ofCol1a1 and Bglapwas lower in cancellous compared to cortical bone. Fi-nally, marrow samples had extremely low expression of bone-relatedgenes, ranging from 71-fold to 930-fold less than bone with marrow.

Centrifugation, but not flushing, decreased the expression of a genethat is highly expressed in bone marrow (Fig. 3C). Icam4 expressionin centrifuged cortical and cancellous boneswas 3- and 2-fold lower, re-spectively, than intact bone and similar in both bone tissues. Flushingdid not alter Icam4 expression.Marrow samples and cortical and cancel-lous bones with marrow had similar expression of Icam4.

Discussion

We initiated these experiments based on our desire to performmoredetailed analysis beyond our well-established tissue-level detailing ofchanges in cortical and cancellous bones following in vivo loading. More-over, our recent report on the sexually-dimorphic and tissue-specific

UNCO

RRECT

Fig. 3. Centrifugation resulted in higher expression of bone related genes and lower expressiobottom panels are fold-change to NMR group (2−ΔΔC

T). (A) Col1a1was 12- and 30-foldmore hto NMR. Col1a1 expression in flushed bone was not significantly different than bone with marrthan cortical and cancellous bones with marrow, respectively (values above bar). (B) Bglapwaspectively, when compared to NMR. Bglap expression was not significantly changed in flushedthat was 261- and 71-fold lower than cortical and cancellous bones with marrow, respectivelylous boneswhen compared to NMR. Icam4 expression was similar in NMR, flushed bone, andmexpression to bone with marrow. Data are mean ± SD, n = 3–5 per group. *Main effect of sno marrow removal or marrow and flushed different from centrifuged. a–cGroups with differecancellous NMR and marrow groups by one-way ANOVA.

Please cite this article as: Kelly NH, et al, A method for isolating high quhttp://dx.doi.org/10.1016/j.bone.2014.07.022

D P

RO

OF

responses following loading ofmice lacking estrogen receptor-alpha pro-vided further impetus to pursue cellular- and molecular-based stud-ies [15]. Finally, a review of the literature revealed no publicationsdocumenting the goals and techniques described here.

We show that RNA can be successfully isolated from centrifugedcortical and cancellous tissues of the mouse tibia separately and usedfor gene expression analysis. The ability to measure gene expressionseparately in cortical and cancellous bones is an important achieve-ment. For example, we discovered that the expression of two highly-expressed bone genes varies between the two tissue types. While bothmethods decreased marrow contamination, centrifugation removedmarrow more completely than flushing in both cortical and cancelloussites when compared to intact bone. Quantitatively, this removal ofcontaminating marrow cells allowed bone transcripts to be more easilydetected by qPCR. Bone genes were as much as 31-fold more highlyexpressed in centrifuged bone, while the expression of a marrow genewas decreased as much as 3-fold.

The ability to examine transcriptional activity in cortical and cancel-lous bones separately from a mouse is a significant achievement thathas the potential to impact osteoporosis and mechanotransduction re-search. Cancellous bone, the major modulator of osteoporosis-associatedfragility fractures, may respond differently to mechanical loading thanthe adjacent cortical component; this subject remains completely unex-plored. While several studies in rodents have examined gene expressionin cancellous bone [10,16,17], none have compared expression to the cor-tical bone at the same location in themouse. Importantly, here cancellousbone from unmanipulated wild type mice had lower expression levels ofbone-related genes than cortical bone, demonstrating the importance ofexamining these sites separately.Moreover, these site-specific differencesare likely to vary with age, sex, genetic background or exposure to a widerange of environmental, lifestyle or drug treatment-related stimuli.

n of a gene in bone marrow. Top panels show relative expression to GAPDH (2−ΔCT) and

ighly expressed in centrifuged cortical and cancellous bones, respectively, when comparedow. Marrow had very low relative expression of Col1a1 that was 932- and 307-fold lowers 13- and 31-fold more highly expressed in centrifuged cortical and cancellous bones, re-bone compared to bone with marrow. Marrow had very low relative expression of Bglap(values above bar). (C) Icam4 expression was decreased in centrifuged cortical and cancel-arrow as indicated by dashed line at a fold-change value of one.Marrow had similar Icam4ite: cancellous different from cortical. Brackets indicate main effect of marrow removal:nt letters are significantly different by Tukey HSD post-hoc. †Cortical NMR different from

ality RNA from mouse cortical and cancellous bones, Bone (2014),

T

241

242Q8

243Q9

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265Q10

266

267

268

269

270

271

272Q11273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310Q12311312313314315316317

318

5N.H. Kelly et al. / Bone xxx (2014) xxx–xxx

C

Although we have shown that centrifugation is an efficient methodof marrow removal, heterogeneous cells are likely to still exist in thebone samples: for example, adherent hematopoietic and endothelialcells and a mixture of osteoblasts and osteocytes. Further work shoulddetermine the exact cellular makeup of our bone samples, including amarker of endothelial cells, such as CD31. The amount of RNA obtainedfrom these bone sites after centrifugation was relatively small, limitingthe number of genes that can be examined with qPCR. However, theRNAwas of high quality and is appropriate for gene expression profilingwith RNA sequencing, which requires much less input material.

Conclusions

Cancellous and cortical bones can be isolated separately from themouse tibia for RNA isolation and used for gene expression studies. Inall future studies examining bone gene expression centrifugationshould be adopted for removing bone marrow from bone tissue sinceit ismore efficient thanflushing. The separate examination of cancellousand cortical bones has the potential to elucidate the differential re-sponse tomechanical loading aswell as geneticmanipulations resultingin different bone phenotypes in these sites.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgments

We thank Adrian McNairn, PhD, and Mary Lou Norman for thetechnical assistance and the Cornell CARE staff. Funding for this studywas provided by: NIH R01-AG028664 and NSF GRFP.

Authors' roles: Study design: NHK, JCS, FPR, MCHM. Study conduct:NHK. Data collection: NHK. Data analysis: NHK. Data interpretation:NHK, JCS, FPR, MCHM. Drafting manuscript: NHK. Revising manuscriptcontent: JCS, FPR, MCHM. Approving final version of manuscript: allauthors. NHK takes responsibility for the integrity of the data analysis.

UNCO

RRE

Please cite this article as: Kelly NH, et al, A method for isolating high quhttp://dx.doi.org/10.1016/j.bone.2014.07.022

ED P

RO

OF

References

[1] Bone health and osteoporosis: a report of the surgeon general; 2004 [Rockville, MD].[2] Cummings SR, Melton LJ. Epidemiology and outcomes of osteoporotic fractures.

Lancet 2002;359:1761–7.[3] Lynch ME, Main RP, Xu Q, Schmicker TL, Schaffler MB, Wright TM, et al. Tibial com-

pression is anabolic in the adult mouse skeleton despite reduced responsivenesswith aging. Bone 2011;49:439–46.

[4] Lynch ME, Main RP, Xu Q, Walsh DJ, Schaffler MB, Wright TM, et al. Cancellous boneadaptation to tibial compression is not sex dependent in growingmice. J Appl Physiol(1985) 2010;109:685–91.

[5] Holguin N, Brodt MD, Sanchez ME, Kotiya AA, Silva MJ. Adaptation of tibial structureand strength to axial compression depends on loading history in both C57BL/6 andBALB/c mice. Calcif Tissue Int 2013;93:211–21.

[6] Armstrong VJ, Muzylak M, Sunters A, Zaman G, Saxon LK, Price JS, et al. Wnt/beta-catenin signaling is a component of osteoblastic bone cell early responses to load-bearing and requires estrogen receptor alpha. J Biol Chem 2007;282:20715–27.

[7] Bonnet N, Standley KN, Bianchi EN, Stadelmann V, Foti M, Conway SJ, et al. Thematricellular protein periostin is required for sost inhibition and the anabolic re-sponse tomechanical loading and physical activity. J Biol Chem2009;284:35939–50.

[8] Silva MJ, Brodt MD, Lynch MA, Stephens AL, Wood DJ, Civitelli R. Tibial loadingincreases osteogenic gene expression and cortical bone volume in mature andmiddle-aged mice. PLoS One 2012;7:e34980.

[9] Zaman G, Saxon LK, Sunters A, Hilton H, Underhill P, Williams D, et al. Loading-related regulation of gene expression in bone in the contexts of estrogen deficiency,lack of estrogen receptor alpha and disuse. Bone 2010;46:628–42.

[10] Wasserman E, Webster D, Kuhn G, Attar-Namdar M, Muller R, Bab I. Differentialload-regulated global gene expression in mouse trabecular osteocytes. Bone 2013;53:14–23.

[11] Xiong J, Onal M, Jilka RL, Weinstein RS, Manolagas SC, O'Brien CA. Matrix-embeddedcells control osteoclast formation. Nat Med 2011;17:1235–41.

[12] Ayturk UM, Jacobsen CM, Christodoulou DC, Gorham J, Seidman JG, Seidman CE,et al. An RNA-seq protocol to identify mRNA expression changes in mouse diaphy-seal bone: applications in mice with bone property altering Lrp5 mutations. J BoneMiner Res 2013;28:2081–93.

[13] Fujita K, Roforth MM, Atkinson EJ, Peterson JM, Drake MT, McCready LK, et al. Isola-tion and characterization of human osteoblasts from needle biopsies without in vitroculture. Osteoporos Int 2014;25:887–95.

[14] Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-timequantitative PCR and the 2(−Delta Delta C(T)) Method. Methods 2001;25:402–8.

[15] Melville KM, Kelly NH, Surita G, Buchalter DB, Schimenti JC, Main RP, et al. Estrogenreceptor alpha in osteoblast lineage cells regulates bone mass differentially in bothsexes and attenuates bone's response to mechanical loading in female mice. J BoneMiner Res 2014 [in review].

[16] Macias BR, Aspenberg P, Agholme F. Paradoxical Sost gene expression response tomechanical unloading in metaphyseal bone. Bone 2013;53:515–9.

[17] Zheng D, Peng S, Yang SH, Shao ZW, Yang C, Feng Y, et al. The beneficial effect of Icariinon bone is diminished in osteoprotegerin-deficient mice. Bone 2012;51:85–92.

ality RNA from mouse cortical and cancellous bones, Bone (2014),