osteocyte-secreted wnt signaling inhibitor · jenna l aronson,2 xiaolong liu,2 jordan m spatz,2...

13

Upload: tranbao

Post on 18-Oct-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

Osteocyte-Secreted Wnt Signaling InhibitorSclerostin Contributes to Beige Adipogenesisin Peripheral Fat Depots

Keertik Fulzele,1 Forest Lai,1 Christopher Dedic,1 Vaibhav Saini,2 Yuhei Uda,1 Chao Shi,1 Padrig Tuck,2

Jenna L Aronson,2 Xiaolong Liu,2 Jordan M Spatz,2 Marc N Wein,2 and Paola Divieti Pajevic1

1Molecular and Cell Biology, Henry M. Goldman School of Dental Medicine, Boston University, Boston, MA, USA2Endocrine Unit, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA

ABSTRACTCells of the osteoblast lineage are increasingly identified as participants in whole-body metabolism by primarily targeting pancreatic

insulin secretion or consuming energy. Osteocytes, the most abundant bone cells, secrete a Wnt-signaling inhibitor called sclerostin.

Here we examined three mouse models expressing high sclerostin levels, achieved through constitutive or inducible loss of the

stimulatory subunit of G-proteins (Gsa inmature osteoblasts and/or osteocytes). Thesemice showed progressive loss ofwhite adipose

tissue (WAT)with tendency toward increased energy expenditure but no changes in glucose or insulinmetabolism. Interestingly beige

adipocyteswere increasedextensively inbothgonadal and inguinalWATandhad reduced canonicalb-catenin signaling. Todetermine

if sclerostin directly contributes to the increased beige adipogenesis, we engineered an osteocytic cell line lacking Gsawhich has high

sclerostin secretion. Conditioned media from these cells significantly increased expression of UCP1 in primary adipocytes, and this

effectwaspartially reduced after depletionof sclerostin from the conditionedmedia. Similarly, treatment ofGsa-deficient animalswith

sclerostin-neutralizing antibody partially reduced the increased UCP1 expression in WAT. Moreover, direct treatment of sclerostin to

wild-type mice significantly increased UCP1 expression in WAT. These results show that osteocytes and/or osteoblasts secrete factors

regulating beige adipogenesis, at least in part, through the Wnt-signaling inhibitor sclerostin. Further studies are needed to assess

metabolic effects of sclerostin on adipocytes and other metabolic tissues. © 2016 American Society for Bone and Mineral Research.

KEY WORDS: SCLEROSTIN; BEIGE ADIPOCYTES; OSTEOCYTES; LEAN PHENOTYPE

Introduction

Micewith geneticmanipulation of osteoblast lineage cells or

their intracellular signaling show altered bodymetabolism

and composition.(1–9) These studies indicate the presence of at

least two mechanisms through which osteolineage cells

participate in global energy metabolism. In the first mecha-

nisms, the osteoblast-specific factor undercarboxylated osteo-

calcin (ucOCN) has been shown to be a potent regulator of

insulin secretion and insulin sensitivity.(1,3,5)However, decreased

ucOCN does not explain reduced body fat in mice with induced-

osteoblast deficiency,(8) indicating the presence of other

ucOCN-independentmechanism(s) throughwhich osteolineage

cells regulate body fat accumulation. Inducible osteocyte

deficiency in adult mice also results in profound loss of white

adipose depots, which can be normalized by restoring

osteocytes.(9) Interestingly, the decrease in body adiposity is

present despite normal serum insulin, glucose, and osteocalcin

levels.(9) Recently, Frey and colleagues(10) identified LRP5-Wnt

signaling in osteoblast as capable of regulating body adiposity

independent of changes in glucosemetabolism or ucOCN levels.

Mice lacking LRP5 in osteoblasts have increased body adiposity

due to reduced LRP5-mediated energy consumption by

osteoblasts.(10) Here we report that G-protein coupled recep-

tor(s) (GPCR), signaling through Gsa, functions as another class

of signaling in osteolineage cells capable of influencing body

adiposity. We identify the secreted Wnt inhibitor sclerostin as

the molecule capable of inducing beige adipogenesis and we

identified osteocytes as important regulators of fat metabolism.

Beige adipocytes are brown-adipocyte-like uncoupling-pro-

tein 1 (UCP1)-positive cells in white adipose tissue (WAT) that

respond to cold-exposure and sympathetic control in a manner

similar to brown adipose tissue (BAT).(11,12) Wnt signaling is a

multiphasic regulator of white and brown adipogenesis. Wnt

ligands bind to Frizzled and LRP5/6 co-receptors to translocate

b-catenin to the nucleus and activate gene expression.

Activation of Wnt-b-catenin signaling in early adipocytes

inhibits both white and brown adipocyte differentiation and

vice versa.(13–16) However, in differentiated brown adipocytes,

Wnt signaling suppresses expression of UCP1 through repres-

sion of PGC1a.(17) This raises a possibility that Wnt signalingmay

also play a role in beige adipogenesis. Two previous studies

Received in original form June 7, 2016; revised form September 17, 2016; accepted September 20, 2016. Accepted manuscript online September 21, 2016.

Address correspondence to: Paola Divieti Pajevic, MD, PhD or Keertik Fulzele, PhD, 700 Albany Street, W201C, Boston, MA 02118, USA. E-mail: [email protected];

[email protected]

Additional Supporting Information may be found in the online version of this article.

ORIGINAL ARTICLE JJBMR

Journal of Bone and Mineral Research, Vol. 32, No. 2, February 2017, pp 373–384

DOI: 10.1002/jbmr.3001

© 2016 American Society for Bone and Mineral Research

373

support this notion. Although inactivating mutation in LRP6 in

patients is associated withmetabolic syndrome and diabetes,(18)

closer examination of LRP6þ/–mice show significantly increased

UCP1 expression in both brown and beige depots.(19) LGR4 is a

receptor for R-spondins that enhances canonical Wnt signaling

in cooperation with Frizzled.(20)Global LGR4-deficient mice have

increased brown and beige adipogenesis and UCP1 activity.(21)

Sclerostin is secreted specifically by osteocytes(22,23) and it

antagonizes Wnt signaling by binding to LRP5, LRP6, or both.(24)

Sclerostin is a negative regulator of bone formation(25) and a

sclerostin-neutralizing antibody is a promising treatment for

osteoporosis.(26) Paradoxically, high circulating sclerostin is

positively associated with higher bone density,(27,28) suggesting

that the effects of sclerostin are more complex. Serum sclerostin

is also positively correlatedwith higher gonadal fat and vertebral

bone marrow fat in humans.(28,29) Recently, continuous

sclerostin treatment has been shown to increase differentiation

of white adipogenic cell line 3T3-L1.(30) However, the role of

sclerostin on beige adipogenesis remains unknown.

We have shown that Gsa is a strong negative regulator of

sclerostin secretion by osteocytes(31) and mice lacking Gsa in

osteocytes have increased sclerostin expression.(32) Here, we

examined three mouse models of Gsa deletion in osteoblasts

and/or osteocytes that show dramatic increase in beige

adipogenesis and decreased body fat. These changes were

not associated with changes in glucose/insulin metabolism or

ucOCN levels. Here we show that the increased beige adipo-

genesis is, at least in part, via increased sclerostin secretion.

Materials and Methods

Mice

All animal experimental procedures were approved by the

Institutional Animal Care and Use Committee (IACUC) of

Massachusetts General Hospital and Boston University. DMP1-

Cre(þ):Gsa(flox/flox), OC-Cre(þ):Gsa(flox/flox), and DMP1-ERT2-GsaKO

mice were generated by mating Gsaflox/flox mice(33) (kindly

provided by Lee Weinstein) with 10 kb DMP1-Cre(34) (kindly

provided by Jerry Feng), OC-Cre,(35) and DMP1-ETR2-Cre mice,(36)

respectively. Littermates lacking Cre expression were used as

controls for all experiments. The genotyping primers and tissue-

specific DNA recombination details are described in the Support-

ingMethods. For postnatal Gsadeletion, DMP1-ERT2-GsaKOmice

were injected with 100 mL of tamoxifen solution from ages 10 to

18 weeks. For sclerostin neutralization, 16-week-old mice were

injected subcutaneously with antisclerostin antibody (100mg/kg,

two times per week; kindly provided by Michaela Kneissel,

Novartis) for 2 weeks. For treatment of wild-typemice, 8-week-old

to 10-week-old wild-type mice were injected intraperitoneally

with 100mg/kg body weight of mouse recombinant sclerostin

(R&D Systems, Minneapolis, MN, USA) every day for 3 weeks.

Body composition and metabolic status analysis

Noninvasive body composition was measured in anesthetized

mice by DXA (Lunar PIXImus II GE Healthcare) to obtain absolute

body fat and leanmass.Absolutemeasurementswerenormalized

to total body weights to account for decreased body weights in

the GsaKOmice. Absolute weights of individual organs (gonadal,

inguinal, and brown adipose depots, and pancreas) were

measuredduringnecropsy and normalized to total bodyweights.

Food intake, energy expenditure, and physical activity were

measured in metabolic cages followed by hyperinsulinemic-

euglycemic clamp measurements at the National Mouse

Metabolic PhenotypingCenter at theUniversityofMassachusetts.

Serum measurements

Blood glucose levels were measured using Precision Xceed Pro

monitor (Abbott, LakeBluff, IL, USA). ELISA assayswereperformed

tomeasure serum levels of insulin (CrystalChem; 90080, Downers

Grove, IL, USA), leptin (Millipore; MMHMAG-44K, Billerica, MA,

USA), Meteorin-like (R&D Systems; DY6679, Minneapolis, MN,

USA), irisin (Adipogen; AG-45A-0046, San Diego, CA, USA), and

sclerostin (R&D Systems; MSST00, Minneapolis, MN, USA).

IHC

IHC was performed as described in the Supporting Methods and

using antibodies for UCP1 (Abcam; ab10983, Cambridge, MA,

USA), insulin (Abcam; ab63820, Cambridge, MA, USA), irisin

(Aviscera; A00170-01-100, Santa Clara, CA, USA), sclerostin (R&D

Systems; BAF1589, Minneapolis, MN, USA), total b-catenin (Cell

Signaling; #9587, Danvers, MA, USA), and active (nonphosphory-

lated) b-catenin (Cell Signaling; #8814, Danvers, MA, USA).

Western blot

Western Blot analysis was performed as described in the

SupportingMethods and using antibodies against Gsa (Millipore;

06-237or Santa Cruz; sc-383, Santa Cruz, CA, USA). The sameblots

were probedwith GAPDH-HRP (Cell Signaling, Danvers, MA, USA)

or b-tubulin-HRP (Cell Signaling, Danvers, MA, USA) as loading

control. Ratios of Gsa-to-GAPDH protein densitometric levels

were calculated by ImageJ (https://imagej.nih.gov/ij/) analysis.

In vitro cultures

Osteocytic cell line Ocy454was generated frommouse long bones

as described(31) and a single-cell subclone of Ocy454 cells was

used.(37)GsaknockdownusingshRNAorknockoutusing “clustered

regularly interspacedshortpalindromic repeats” (CRISPR)/Cas9was

performed as described in the SupportingMethods. Primary beige

adipocyte differentiation and sclerostin depletion from condi-

tioned media was done as described in the Supporting Methods.

Quantitative PCR

Total RNAwas extracted from tissues or cells using Trizol reagent

(MRC). First-strand cDNAwas generated using the High Capacity

cDNA Reverse Transcription Kit (Life Technologies, Carlsbad, CA,

USA), and qPCR was performed using SYBR Green Master Mix in

a StepOnePlus Real-Time PCR system (Applied Biosystems,

Foster City, CA, USA). Target genes were normalized to one

housekeeping gene for in vitro samples or to at least three

housekeeping genes using geometric averaging methods.(38)

Primers used are listed in the Supporting Methods.

Results

Mice lacking Gsa in osteocytes have severely decreasedWAT mass

As reported,(39,40) the Cre-loxP recombination in DMP1-GsaKO

mice were observed in skeletal tissues as well as skeletal muscle

(Fig. 1A). Gsa mRNA expressions were significantly reduced in

osteocyte-enriched bones and skeletal muscle (Fig. 1B). Gsa

protein expressions were decreased in skeletal tissue but not in

skeletal muscle (Fig. 1C). Cre-loxP recombinations were absent

374 FULZELE ET AL. Journal of Bone and Mineral Research

Fig. 1. DecreasedWATmass inmice lacking Gsa in osteocytes. (A) Cre-loxP recombination on DNA from various tissues of DMP1-GsaKOmice. (n¼ 3). (B)

GsamRNA expression in osteocyte-enriched bones (OEBE), gastrocnemius muscle, GWAT, and InWAT from 21-week-old control and DMP1-GsaKOmice

(n¼ 5). (C) Gsa and GAPDH protein levels in osteocyte-enriched femurs and other metabolic tissues (n¼ 3). The bands were quantified as Gsa-to-GAPDH

ratio. (D) Growth curves of control, DMP1-GsaKO, and genetic backgroundmatched DMP1-Cre only mice (n� 10). (E) Nose to anus body length from 12-

week-old and 21-week-old control (Con) (n¼ 6) and DMP1-GsaKO (KO) mice (n¼ 6). DXA analysis showing normalized whole-body (F) fat and (G) lean

mass in 21-week-old control (Con) (n¼ 10) and DMP1-GsaKO (KO) (n¼ 7). (H) Serum leptin levels from 21-week-old control (Con) (n¼ 13) and DMP1-

GsaKO (KO) mice (n¼ 10). (I) Representative images of GWAT (top panel) and InWAT (bottom panel) from control (Con), and DMP1-GsaKO (KO) mice. (J)

Individual GWAT, InWAT, BAT, and pancreas organ weights normalized to total body weights from 21-week-old control (Con) (n¼ 7) and DMP1-GsaKO

(KO) mice (n¼ 7). All data: mean� SE, �p< 0.05.

Journal of Bone and Mineral Research EFFECT OF SCLEROSTIN ON BEIGE ADIPOCYTES IN PERIPHERAL FAT 375

and Gsa mRNA-protein expressions were normal in other

metabolic tissues including gonadal WAT (GWAT), inguinal WAT

(InWAT), BAT, pancreas, and liver (Fig. 1A–C). It should be noted

that mice lacking Gsa in skeletal muscle do not have a body

composition phenotype.(41)

DMP1-GsaKO had significantly decreased body weight

compared to control littermates beginning at 5 weeks of

age (Fig. 1D). Mice carrying the Cre-only transgene (DMP1-Cre)

on the same genetic background of the DMP1-GsaKO mice

were indistinguishable from controls (Gsaflox/flox) (Fig. 1D).

Therefore, littermates lacking Cre expression DMP1-Cre(–):

Gsa(flox/flox) were used as controls for all subsequent experi-

ments. The body lengths of DMP1-GsaKO mice were similar at

12 weeks of age and were significantly decreased only at

21 weeks of age (Fig. 1E). DXA showed a significant decrease

in absolute (Supporting Fig. 1A) and normalized (Fig. 1F)

whole-body fat mass in DMP1-GsaKO mice. Lean mass was

significantly decreased in DMP1-GsaKO mice (Supporting

Fig. 1B), but was significantly increased after correcting for the

decreased body weight (Fig. 1G). Serum leptin, which

correlates with body adiposity,(42,43) was also significantly

decreased in the DMP1-GsaKO mice (Fig. 1H). Necropsy

showed marked reduction in WAT depots including GWAT and

InWAT depots (Fig. 1I, J). The organ weights of other

metabolic organs including BAT, pancreas (Fig. 1J), liver,

and skeletal muscles (data not shown) were not changed.

Similar to the female DMP1-GsaKO mice (Fig. 1), male DMP1-

GsaKO mice also had significant decrease in body weight,

GWAT, and InWAT (Supporting Fig. 1C, D). These data show

that loss of Gsa-signaling in osteolineage cells decreases body

adiposity.

Mice lacking Gsa in mature osteoblasts and osteocyteshave decreased WAT mass

Next we sought to validate the body fat changes in DMP1-

GsaKO mice using another similar mouse model of OC-GsaKO

(Fig. 2A). Littermates lacking Cre expression OC-Cre(–):Gsa(flox/

flox) were used as controls for all experiments because presence

of OC-Cre transgene does not alter body composition.(44) Cre-

loxP recombination (Fig. 2B) and decreased Gsa mRNA

expression (Fig. 2C) were observed only in skeletal tissues and

not in other metabolic tissues of OC-GsaKOmice (Fig. 2B, C). The

mice had significantly reduced body weight starting at 9 weeks

of age (Fig. 2D). These mice were shorter from early age and had

significantly reduced body length by 12 weeks of age (Fig. 2E).

DXA analysis showed that these mice were osteopenic (Fig. 2F),

and had significantly reduced absolute whole-body fat and lean

mass (Supporting Fig. 2A, B). After normalizing for the decreased

body weight, OC-GsaKO mice continued to have significant

reduction in body fat (Fig. 2G), but had significantly increased

lean mass (Fig. 2H). Serum leptin was significantly decreased in

OC-GsaKO mice (Fig. 2I). Gross appearances showed markedly

reduced sizes of GWAT and InWAT (Fig. 2J). Normalization of

absolute weights of GWATs and InWATs to respective body

weights were significantly decreased in the OC-GsaKO mice

compared to control littermates, whereas absolute or normal-

ized weights of BAT and pancreas were not altered (Fig. 2K).

Similar to females, male OC-GsaKO mice also had significant

decrease in body weight (Supporting Fig. 2C), reduced absolute

and normalized body fat (Supporting Fig. 2D), osteopenia

(Supporting Fig. 2E), and significantly reduced GWAT organ

weight (Supporting Fig. 2F, G).

Mice with postnatal loss of Gsa in osteocytes havereduced WAT mass

Next we examined body composition ofmice with inducible loss

of Gsa in osteocytes (DMP1-ERT2-GsaKO) by treating thesemice

with tamoxifen from 10 weeks of age until 18 weeks of age. The

DMP1-Cre-ERT2 mice expression has been shown to be specific

to osteocytes and a fewmature osteoblasts.(36) Before beginning

of tamoxifen treatment, body weight (Supporting Fig. 3A), BMD

(Supporting Fig. 3B), bone mineral content (Supporting Fig. 3C),

body lean mass (Supporting Fig. 3D), and fat mass (Supporting

Fig. 3E) were similar between the DMP1-ERT2-GsaKO and

control littermates. At the end of the 8-week tamoxifen

treatment, total body weights were not significantly changed

(Fig. 3A). As expected, these mice developed osteopenia (Fig. 3

B), indicating an essential role of Gsa signaling postnatally in

regulating bone mass.(45) DXA analysis showed a significant

decrease in absolute (Supporting Fig. 3F) and normalized

(Fig. 3C) whole-body fat mass in DMP1-ERT2-GsaKO mice. The

absolute (Supporting Fig. 3G) or corrected (Fig. 3D) lean mass

were not altered in tamoxifen-treated DMP1-ERT2-GsaKO mice.

Femur lengths at the end of the tamoxifen treatment were not

different from littermate controls (Fig. 3E), indicating that the

changes in whole-body fat in these mice are independent of

changes in body size. In accordance with the DXA analysis,

GWAT and InWAT gross appearances were markedly decreased

and organ weights normalized for total body weight were

significantly reduced in DMP1-ERT2-GsaKOmice (Fig. 3F, G). The

normalized organ weights of other metabolic organs including

pancreas (Fig. 3H) and BAT (Fig. 3I) were not changed. Taken

together, the three constitutive and inducible mouse models of

Gsa deletion in osteolineage cells indicate that osteolineage

cells contribute to body fat accumulation.

The lean phenotype of mice lacking Gsa in osteocytes isassociated with increase in energy expenditure

The decreased body adiposity in the mice lacking Gsa in

osteolineage cells could be due to altered energy balance.

Combined food intake, physical activity, or oxygen consumption

over a 24-hour period were not changed in DMP1-GsaKO mice

(Supporting Fig. 4A–C). Close examination showed that

although food intake (Supporting Fig. 4A) and physical activity

(Supporting Fig. 4C) were significantly reduced during the 12-

hour dark cycle, oxygen consumption was normal during the

dark cycle in DMP1-GsaKO mice and there was a tendency to

increased oxygen consumption during the light cycle (p¼ 0.064)

(Supporting Fig. 4C). Therefore, DMP1-GsaKO mice have a

sustained increase in energy expenditure that, at least in part,

drives the decrease in body fat.

Mice lacking ucOCN are obese, which is associated with

decreased insulin secretion and pancreatic b-cell proliferation.(1)

Therefore, the lean phenotype of mice lacking Gsa in

osteolineage cells could be due to increased ucOCN and insulin

secretion. However, serum insulin levels were normal in DMP1-

GsaKO, OC-GsaKO, and DMP1-ERT2-GsaKO (Supporting

Fig. 4D–F). Pancreatic b-cell area, as determined by insulin

immunohistochemistry, was not changed in DMP1-GsaKO mice

(Supporting Fig. 4G, H). Further examination of glucose

metabolism and insulin action by hyperinsulinemic-euglycemic

clamps showed normal levels of glucose infusion rate (GIR),

basal and clamp hepatic glucose production, glucose turnover,

glycolysis, and glycogen synthesis in DMP1-GsaKO mice

(Supporting Fig. 4I). Serum osteocalcin (OC) levels were

376 FULZELE ET AL. Journal of Bone and Mineral Research

significantly decreased in DMP1-GsaKO (Supporting Fig. 4J) and

OC-GsaKO (Supporting Fig. 4L) mice. Serum ucOCN showed

decreasing trend in DMP1-GsaKO mice (Supporting Fig. 4K) and

was significantly decreased in OC-GsaKO mice (Supporting

Fig. 4M). Taken together, these data show that the reduction in

WAT in mice lacking Gsa in osteolineage cells is independent

from ucOCN.

Beige adipogenesis is increased in mice lacking Gsa inosteolineage cells

Histological examination ofWAT fromDMP1-GsaKO, OC-GsaKO,

and DMP1-ERT2-GsaKO mice (Fig. 4 A–C) showed markedly

smaller adipocytes with significantly reduced adipocyte surface

area (Supporting Fig. 5A). These smaller adipocytes appeared as

clusters of brown-adipocyte–like multilocular adipocytes or

beige adipocytes. Immunohistochemical staining for beige

adipogenic marker UCP1 showed marked increase in UCP1-

positive cells in GWATs from DMP1-GsaKO, OC-GsaKO, and

DMP1-ERT2-GsaKO (Fig. 4D–F). Lower magnification imaging

showed that the UCP1þ adipocytes are more widespread

throughout the WATs (Supporting Fig. 5B). Similarly, InWATs

from DMP1-GsaKO mice (Supporting Fig. 5C) and OC-GsaKO

(Supporting Fig. 5D) also showed a widespread increase in

UCP1þ beige adipocytes. Beige adipogenic gene markers

including UCP1, Cox5b, Dio2, Cidea, and/or Elovl3 were

Fig. 2. Mice lacking Gsa inmature osteoblasts and osteocytes have decreased body fat. (A) Schematic representation of OC-Cre targeted ablation of Gsa

ablation in osteoblastic lineage. (B) Cre-loxP DNA recombination in various tissues of OC-GsaKOmice (n¼ 3). (C) GsamRNA expression in bonemarrow–

depleted bones, gastrocnemius muscle, GWAT, and InWAT from 21-week-old control and OC-GsaKO mice (n¼ 5). (D) Growth curves of control and

OC-GsaKO mice (n� 6). (E) Nose to tail-base body length from 12-week-old control (Con) (n¼ 6) and OC-GsaKO (KO) mice (n¼ 5). Noninvasive DXA

showing (F) BMD, (G) normalized whole-body fat, and (H) normalized lean mass in 12-week-old control (Con) (n¼ 9) and OC-GsaKO (KO) mice (n¼ 7). (I)

Serum leptin levels from 21-week-old control (Con) (n¼ 9) and OC-GsaKO (KO) mice (n¼ 8). (J) Representative gross images showing sizes of GWAT and

InWAT from 15-week-old control and OC-GsaKO mice. (K) Individual GWAT, InWAT, BAT, and pancreas organ weights normalized to total body weights

from 21-week-old control (Con) (n¼ 8) and OC-GsaKO (KO) mice (n¼ 6). All data: mean� SE; �p< 0.05.

Journal of Bone and Mineral Research EFFECT OF SCLEROSTIN ON BEIGE ADIPOCYTES IN PERIPHERAL FAT 377

significantly increased in InWATs of DMP1-GsaKO, OC-GsaKO,

and DMP1-ERT2-GsaKO (Fig. 4G–I). We next examined the

expressions of white adipogenic genes. Gene expression of

master regulator of adipocyte differentiation PPARg but not the

co-regulator of differentiation CEBPa and SREBP1, were

significantly reduced in InWATs from both DMP1-GsaKO

(Fig. 4J) and OC-GsaKO (Fig. 4K). This was accompanied with

significant decrease in perilipin (PLIN), PPARa, and Acox1 (Fig. 4J,

K). Interestingly, these parameters of white adipocyte differenti-

ation were not significantly changed in InWATs from DMP1-

ERT2-GsaKO mice (Fig. 4L), suggesting that the increased beige

adipogenesis most likely precedes the decrease in white

adipogenesis in the DMP1-GsaKO and OC-GsaKO mice. We

next focused on mechanisms through which Gsa-deficient

osteocytes could increase beige adipogenesis.

Osteocyte-secreted Wnt signaling inhibitor sclerostincontributes to the increased beige adipogenesis

White and beige adipogenic differentiation of the stromal

vascular fraction (SVF) cells from InWATs of control and DMP1-

GsaKO mice were similar (Supporting Fig. 6A, B). Therefore, the

increased beige adipogenesis in thesemice is not intrinsic to the

adipose tissue but imposed by external factors. Recently skeletal

muscle-secreted factors meteorin-like(46) and irisin(47) have been

shown to increase beige adipogenesis, raising the possibility

that factors secreted from bone cells could alter the secretion of

these factors from skeletal muscle. Whereas the serum levels of

myokine meteorin-like were not changed (Supporting Fig. 6C),

serum irisin levels were significantly increased in DMP1-GsaKO,

OC-GsaKO, and DMP1-ERT2-GsaKOmice (Supporting Fig. 6D, E).

As described in the Introduction, activatedWnt-signaling is an

inhibitor of UCP1-expression in differentiated brown adipocytes

and mice with heterozygous loss of sclerostin receptor LRP6

show significantly increased UCP1 expression in both brown and

beige depots.(19) Interestingly, sclerostin-deficient mice had

significantly increased body fat mass,(48) suggesting that

sclerostin may play a role in beige adipogenesis. Serum

sclerostin levels were significantly increased in the DMP1-

GsaKO, OC-GsaKO, and DMP1-ERT2-GsaKO mice (Fig. 5A).

Sclerostin expression in cortical bones of OC-GsaKO (Fig. 5B) and

DMP1-ERT2-GsaKO mice (Fig. 5C) were also increased similar to

previous findings in DMP1-GsaKO animals.(32) Examination of

Wnt signaling in WAT showed significant reduction in expres-

sion of Axin2, a canonical Wnt signaling target gene, in DMP1-

GsaKO mice (Fig. 5D). At the protein level, total b-catenin

(Fig. 5E, F) and nonphosphorylated active b-catenin (Fig. 5G, H)

immunostainings were strikingly reduced in the beige adipo-

cytes in DMP1-GsaKO, OC-GsaKO, and DMP1-ERT2-GsaKO mice

relative to the control cells and endothelial cells lining blood

Fig. 3. Inducible loss of Gsa in osteocytes in adult mice also leads to decreasedWAT. (A) Total body weight, (B) whole-body BMD, (C) normalized whole-

body fat, (D) normalized whole-body lean mass, and (E) femur length in 18-week-old control (Con) (n¼ 6) and DMP1-ERT2-GsaKO (KO) (n¼ 7) treated

with tamoxifen from 10 to 18 weeks of age. Individual organ weights of (F) GWAT, (G) InWAT, (H) pancreas, and (I) BAT normalized to total body weights

from 18-week-old control (Con) (n¼ 6) and DMP1-ERT2-GsaKO (KO) (n¼ 6) mice treated with tamoxifen from 10 to 18 weeks of age. All data: mean� SE,�p< 0.05.

378 FULZELE ET AL. Journal of Bone and Mineral Research

Fig. 4. Beige adipogenesis is increased in mice lacking Gsa in osteolineage cells. Representative H&E staining of GWATs from (A) DMP1-GsaKO, (B) OC-

GsaKO, and (C) DMP1-ERT2-GsaKOmice with corresponding control littermates (n¼ 4). UCP1 IHC staining from (D) DMP1-GsaKO, (E) OC-GsaKO, and (F)

DMP1-ERT2-GsaKO mice (n¼ 3). Gene expressions of beige adipogenesis markers PRDM16, Cox5b, Dio2, Cidea, Elovl3, and UCP1 in InWAT from (G)

DMP1-GsaKO, (H) OC-GsaKO, and (I) DMP1-ERT2-GsaKOmice with corresponding control littermates (n¼ 5). Expressions of white adipogenesis markers

PPARg, CEBPa, SREBP1, PPARa, PLIN, Acox1, and Acls1 in InWAT of (J) DMP1-GsaKO, (K) OC-GsaKO, and (L) DMP1-ERT2-GsaKOmice with corresponding

control littermates (n¼ 5). All data: mean� SE; �p< 0.05.

Journal of Bone and Mineral Research EFFECT OF SCLEROSTIN ON BEIGE ADIPOCYTES IN PERIPHERAL FAT 379

vessels (Supporting Fig. 6F). Immunostaining for nonphos-

phorylated active b-catenin was present in nucleus and

cytoplasm of adipocytes from control mice (Fig. 5F). By contrast,

the active b-catenin staining was markedly absent from most of

the beige adipocytes from the KO mice (Fig. 5G). The co-

occurrence of decreased canonical Wnt signaling in beige

adipocytes coupled with increased circulating serum sclerostin

suggested a role of sclerostin in regulating beige adipogenesis.

Next we examined whether sclerostin could directly induce

expressions of beige adipogenic genes. SVF undergoing brown

adipogenic differentiation were treated with sclerostin (0.2 to

20 ng/mL) after their initial differentiation commitment for

5 days in lower incubator temperature of 33°C. Sclerostin dose-

dependently increased expression of UCP1 and PGC1a in

primary adipocytes (Fig. 6A). Sclerostin treatment (200 ng/mL) of

beige adipogenic SVF significantly rescued the reduction of

Fig. 5. Decreased Wnt signaling in beige adipocytes of mice lacking Gsa in osteolineage cells. (A) Serum sclerostin levels in DMP1-GsaKO, OC-GsaKO,

and DMP1-GsaKOmice compared to respective littermate controls (n¼ 8, each genotype). Sclerostin immunostaining in long bones from (B) OC-GsaKO

and (C) DMP1-ERT2-GsaKO mice (n¼ 4). (D) Wnt target gene Axin2 gene expression in adipose tissues from control and DMP1-GsaKOmice (n¼ 5, each

genotype). Immunohistochemical staining and the staining quantification for (E, F) total b-catenin and (G, H) nonphosphorylated active b-catenin in

InWAT from DMP1-GsaKO, OC-GsaKO, and DMP1-ERT2-GsaKO mice (n¼ 4). All data: mean� SE; �p< 0.05.

380 FULZELE ET AL. Journal of Bone and Mineral Research

PGC1a and UCP1 induced by Wnt-3a treatment (Fig. 6B). To

examine the in vivo effects of Sclerostin, wild-type mice were

administered 100mg/kg of sclerostin every day for 3 weeks.

Although sclerostin treatment did not alter total body weight

(Supporting Fig. 7E), it significantly reduced normalized tissue

weights of WAT (Fig. 6C). The weights of other tissues including

BAT and pancreas were not changed (Supporting Fig. 7E).

Quantitative PCR showed that the continuous sclerostin

treatment significantly increased UCP1 expression (Fig. 6D).

We have previously shown that Gsa-deficient osteocytic cells

have significant increase in expression of SOST/sclerostin.(31)

Here, we engineered an osteocytic cell line Ocy454 lacking Gsa

using the CRISPR-Cas9 technique (Supporting Fig. 7A). The

GsaKO Ocy454 cells had a time-dependent increase in sclerostin

expression over control cells (Supporting Fig. 7B). Sclerostin was

also significantly increased in conditioned media (CM) of GsaKO

Ocy454 cells (Fig. 6E). Using a sclerostin antibody we could

deplete sclerostin from the CM of GsaKOOcy454 cells by 10-fold

(Fig. 6E). Primary adipocytes from wild-type mice undergoing

beige adipogenic differentiation were treated with 20% (vol/vol)

CM from GsaKO Ocy454 cells. Treatment with 20% (vol/vol) CM

fromOcy454 cells significantly increased expression of UCP1 and

PGC1a in primary adipocytes undergoing beige adipogenic

differentiation compared to no treatment control group

(Fig. 6F). CM from GsaKO Ocy45 cells significantly increased

UCP1 and Elovl3 expression compared to the control group

(Fig. 6F). Sclerostin depletion significantly reduced expression of

UCP1 and Elovl3 induced by Gsa-KO Ocy454 cells CM (Fig. 6F).

CM from Gsa-KO Ocy454 cells modestly but significantly

increased expression of PPARg in primary adipocytes

Fig. 6. Increased sclerostin secretion from Gsa-deficient osteocytes contributes to beige adipogenesis. UCP1 and PGC1a expressions in SVF from wild-

type InWAT undergoing beige adipogenesis in response to (A) varying doses of sclerostin (n¼ 3), and (B) Wnt3a and/or sclerostin treatment (n¼ 3). (C)

GWAT organ weight and (D) UCP1 gene expression in GWAT in wild-type mice treated with 100mg/kg sclerostin every day for 3 weeks. (E) Sclerostin

secretion into CM in CRISPR-mediated Gsa knockout cells and after sclerostin depletion using antibody (n¼ 3). (F) UCP1, PGC1a, and Elovl3 expression in

primary SVF after treatment with CM from Ocy454 cells. Sclerostin-neutralizing antibody treatment of control and DMP1-GsaKO mice for 2 weeks

showing (G) serum sclerostin changes and (H) gene expression changes in InWAT tissues (n¼ 4, each group). All data: mean� SE; �p< 0.05 compared to

NT controls; þþp< 0.05 compared to Wnt-3a treatment.

Journal of Bone and Mineral Research EFFECT OF SCLEROSTIN ON BEIGE ADIPOCYTES IN PERIPHERAL FAT 381

undergoing white adipogenic differentiation (Supporting

Fig. 7D), but UCP1 expression was undetectable in these cells.

This data suggests that the increased beige adipogenesis inmice

lacking Gsa in osteolineage cells is in part mediated by sclerostin

effects on WAT. Finally, we treated DMP1-GsaKOmice with anti-

sclerostin neutralizing antibody for 2 weeks to investigate

whether normalizing sclerostin level in these animals could

rescue the adipose tissue phenotype. Anti-sclerostin antibody

significantly decreased serum sclerostin levels in DMP1-GsaKO

mice (Fig. 6G) and did not alter serum insulin or serum irisin

levels in these animals (Supporting Fig. 7C). In InWAT tissues, this

treatment significantly reduced the expression of UCP1 and

PGC1a in the DMP1-GsaKO mice compared to the IgG-antibody

treatment (Fig. 6H), showing that the increased beige adipo-

genesis in these mice is partly mediated by direct effect of

Sclerostin on WAT.

Discussion

Here we report that the osteocyte-secreted factor sclerostin

contributes to increased beige adipogenesis in vivo and in vitro.

Presence of increased beige adipogenesis in the three genetic

mouse models DMP1-GsaKO, OC-GsaKO, and DMP1-ERT2-

GsaKO with high circulating sclerostin supports this notion.

Although DMP1-Cre expressed in a small population of skeletal

muscle cells, in addition to osteocytes, the body composition

phenotypes of DMP1-GsaKO mice are not due to loss of Gsa

signaling in muscle because mice with Gsa disruption in skeletal

muscle have a normal body composition.(41) The OC-GsaKO and

DMP1-ERT2-GsaKO, in which OC-Cre or DMP1-ERT2-Cre are

expressed only in skeletal tissues, also display similar lean and

beige adipogenic phenotypes, showing that the phenotypes are

driven by Gsa ablation in osteogenic cells. Mice withWAT,b-cell,

or liver-specific homozygous Gsa deletion are also lean(49);

however, there was no Cre-loxP recombination observed in

these tissues from either of the OC-GsaKO, DMP1-GsaKO, or

DMP1-ERT2-GsaKO mice. Taken together, these data indicate

that the lean phenotypes of mice lacking Gsa in osteolineage

cells are direct effect of Gsa deficiency inmature osteoblasts and

osteocytes.

Altered body size between comparison groups adds com-

plexity in evaluating the metabolic or body composition

phenotypes, especially with altered skeletal phenotype.

Smaller-sized mice are expected to face higher thermoregula-

tory demands due to their increased heat loss from a relatively

greater surface area. Smaller mice have been shown to cope

with this by decreasing their core body temperature or possibly

by increasing the insulation quality of their fur.(50) In most dwarf

mice this thermoregulatory adjustment leads to positive energy

balance and obesity. Although OC-GsaKO mice were also

shorter by 5 weeks of age, body lengths of DMP1-GsaKO mice

were not different compared to controls by 12 weeks of age.

DMP1-GsaKOmice show a significantly reduced body weight by

5 weeks of age, indicating that the lean phenotype precedes the

reduced body length. Furthermore, conditional ablation of Gsa

in adult DMP1-ERT2-GsaKO mice upon tamoxifen injections

from 10 to 18 weeks of age resulted in a lean phenotype

although the body length remained unaltered. These data

clearly indicate that the lean phenotypes due to Gsa deficiency

in osteolineage cells are independent of body size.

Our previous analysis has shown increased circulating

myeloid cells in DMP1-GsaKO mice(32) and cytokines secreted

from hematopoietic lineage cells have been shown to increase

beige adipogenesis and thermogenesis.(46,51) It is possible that

increased circulating myeloid cells or cytokines in DMP1-GsaKO

mice can contribute to the increased beige adipogenesis.

However, mice with postnatal deletion of Gsa in osteocytes—

DMP1-ERT2-GsaKO mice—do not show changes in hematopoi-

etic lineages (data not shown) but still have increased beige

adipogenesis, indicating that factors independent of immune

cells are contributing to the beige adipogenesis. Moreover, our

in vitro CM experiments show that factors secreted from

osteocytes, and specifically sclerostin, can induce UCP1 and

other beige adipogenic markers in SVF. Similarly, increased

sympathetic tone is a well-known contributor to increased beige

adipogenesis. Although we cannot exclude a relative contribu-

tion of both the sympathetic tone or circulating cytokines in our

in vivo model, our in vitro experiments showed that osteocytes

indeed directly regulate adipogenesis through secreted factors.

As a negative regulator of Wnt signaling, high sclerostin levels

are expected to correlate with lower BMD and increased risk of

fractures. However, multiple studies show that sclerostin levels

do not necessarily follow an expected pattern with BMD, bone

turnover markers, and fracture risk.(52) Potential sources of

biological variability, clearance rate from circulation, and even

assay standardization may confound these findings.(52) None-

theless, sclerostin neutralizing antibody is a potent anabolic

treatment for postmenopausal osteoporosis,(26) highlighting the

clinical utility of sclerostin for disease prevention. Our findings

suggest that sclerostin may contribute negatively to obesity by

increasing beige adipogenesis without affecting glucose

metabolism. Serum sclerostin is increased in non-obese diabetic

patients(53) and during aging,(54) conditions characterized by

decreased metabolism and tendency to increased body fat.

However, similar to bone fractures, correlation of sclerostin and

body fat is confounding due to the presence of other

biologically variable factors. In postmenopausal older Japanese

women, serum sclerostin levels were positively correlated with

abdominal and gonadal fat.(55)However, sclerostin levels did not

change in obese young adult patients withmetabolic syndrome,

or after weight loss inducing bariatric surgery.(56,57) Moreover,

assessment of beige adipocytes in normal adults or patients has

not been widely examined at this time.(58) Therefore, changes in

beige adipogenesis in physiological conditions with high

circulating sclerostin levels or after anti-sclerostin antibody

treatment remain to be examined. Previous studies have shown

that circulating irisin and sclerostin were highly correlated in

human adults.(59) However, reducing circulating sclerostin to

normal levels in DMP1-GsaKOmice using sclerostin-neutralizing

antibody did not change circulating irisin levels in these mice

and yet had reduced expression of beige adipogenic markers.

This indicates that the increased beige adipogenesis effect in

these studies is primarily from increased sclerostin levels.

Further studies are needed to understand the contribution of

the Wnt signaling inhibitor sclerostin on white, brown, or beige

fat development.

Disclosures

All authors state that they have no conflicts of interest.

Acknowledgments

This work was supported by NIH grants AR060221, AR059655,

and DK079161 to PDP. This work was also partially supported by

382 FULZELE ET AL. Journal of Bone and Mineral Research

a GAP award from ASBMR to KF, a P&F award from Boston

University Clinical and Translational Science Institute (CTSI) (NIH

1UL1TR001430) to KF, a P&F award from Center for Skeletal

Research (CSR) of Massachusetts General Hospital (NIH NIAMS

P30 AR066261), and a P&F award from Boston Area Diabetes

Endocrinology Research Center (BADERC) to PDP. We thank

Michaela Kneissel and Novartis for providing the anti-sclerostin

antibody.

Authors’ roles: KF and PDP conceived the project, designed

the experiments, and wrote the manuscript. KF, FL, CD, VS, YU,

CS, PT, JG, and XL performed the experiments. KF, FL, CD, YU, CS,

and PDP analyzed the data. JMS and MW contributed materials.

References

1. Lee NK, Sowa H, Hinoi E, et al. Endocrine regulation of energymetabolism by the skeleton. Cell. 2007;130(3):456–69.

2. Yoshizawa T, Hinoi E, Jung DY, et al. The transcription factor ATF4regulates glucose metabolism in mice through its expression inosteoblasts. J Clin Invest. 2009;119(9):2807–17.

3. Rached MT, Kode A, Silva BC, et al. FoxO1 expression in osteoblastsregulates glucose homeostasis through regulation of osteocalcin inmice. J Clin Invest. 2010;120(1):357–68.

4. Riddle RC, Frey JL, Tomlinson RE, et al. Tsc2 is amolecular checkpointcontrolling osteoblast development and glucose homeostasis. MolCell Biol. 2014;34(10):1850–62.

5. Wei J, Hanna T, Suda N, Karsenty G, Ducy P. Osteocalcin promotesbeta-cell proliferation during development and adulthood throughGprc6a. Diabetes. 2014;63(3):1021–31.

6. Ferron M, Lacombe J. Regulation of energy metabolism by theskeleton: osteocalcin and beyond. Arch Biochem Biophys. 2014 Nov1;561:137–46.

7. Fulzele K, Riddle RC, DiGirolamo DJ, et al. Insulin receptor signalingin osteoblasts regulates postnatal bone acquisition and bodycomposition. Cell. 2010;142(2):309–19.

8. Yoshikawa Y, Kode A, Xu L, et al. Genetic evidence points to anosteocalcin-independent influence of osteoblasts on energymetabolism. J Bone Miner Res. 2011;26(9):2012–25.

9. Sato M, Asada N, Kawano Y, et al. Osteocytes regulate primarylymphoid organs and fat metabolism. Cell Metab. 2013;18(5):749–58.

10. Frey JL, Li Z, Ellis JM, et al. Wnt-Lrp5 signaling regulates fatty acidmetabolism in the osteoblast. Mol Cell Biol. 2015;35(11):1979–91.

11. Loncar D, Afzelius BA, Cannon B. Epididymal white adipose tissueafter cold stress in rats. I. Nonmitochondrial changes. J UltrastructMol Struct Res. 1988;101(2–3): 109–22.

12. Cousin B, Cinti S, Morroni M, et al. Occurrence of brown adipocytes inrat white adipose tissue: molecular and morphological characteri-zation. J Cell Sci. 1992;103(Pt 4):931–42.

13. Kang S, Bennett CN, Gerin I, Rapp LA, Hankenson KD, MacdougaldOA. Wnt signaling stimulates osteoblastogenesis of mesenchymalprecursors by suppressing CCAAT/enhancer-binding protein alphaand peroxisome proliferator-activated receptor gamma. J BiolChem. 2007;282(19):14515–24.

14. Christodoulides C, Laudes M, Cawthorn WP, et al. The Wntantagonist Dickkopf-1 and its receptors are coordinately regulatedduring early human adipogenesis. J Cell Sci. 2006;119(Pt 12):2613–20.

15. Qiu W, Andersen TE, Bollerslev J, Mandrup S, Abdallah BM, KassemM. Patients with high bone mass phenotype exhibit enhancedosteoblast differentiation and inhibition of adipogenesis of humanmesenchymal stem cells. J Bone Miner Res. 2007;22(11):1720–31.

16. Christodoulides C, Lagathu C, Sethi JK, Vidal-Puig A. Adipogenesisand WNT signalling. Trends Endocrinol Metab. 2009;20(1):16–24.

17. Kang S, Bajnok L, Longo KA, et al. Effects of Wnt signaling on brownadipocyte differentiation and metabolism mediated by PGC-1alpha.Mol Cell Biol. 2005;25(4):1272–82.

18. Mani A, Radhakrishnan J, Wang H, et al. LRP6 mutation in a familywith early coronary disease and metabolic risk factors. Science.2007;315(5816):1278–82.

19. Liu W, Singh R, Choi CS, et al. Low density lipoprotein (LDL)receptor-related protein 6 (LRP6) regulates body fat and glucosehomeostasis by modulating nutrient sensing pathways andmitochondrial energy expenditure. J Biol Chem. 2012; 287(10):7213–23.

20. Carmon KS, Gong X, Lin Q, Thomas A, Liu Q. R-spondins function asligands of the orphan receptors LGR4 and LGR5 to regulate Wnt/beta-catenin signaling. Proc Natl Acad Sci U S A. 2011;108(28):11452–7.

21. Wang J, Liu R, Wang F, et al. Ablation of LGR4 promotes energyexpenditure by driving white-to-brown fat switch. Nat Cell Biol.2013;15(12):1455–63.

22. Poole KE, van Bezooijen RL, Loveridge N, et al. Sclerostin is a delayedsecreted product of osteocytes that inhibits bone formation. FASEBJ. 2005;19(13):1842–4.

23. van Bezooijen RL, Roelen BA, Visser A, et al. Sclerostin is anosteocyte-expressed negative regulator of bone formation, but nota classical BMP antagonist. J Exp Med. 2004;199(6):805–14.

24. Baron R, Kneissel M. WNT signaling in bone homeostasis anddisease: from human mutations to treatments. Nat Med. 2013;19(2):179–92.

25. Li X, Ominsky MS, Niu QT, et al. Targeted deletion of the sclerostingene inmice results in increased bone formation and bone strength.J Bone Miner Res. 2008;23(6):860–9.

26. McClung MR, Grauer A, Boonen S, et al. Romosozumab inpostmenopausal women with low bone mineral density. N Engl JMed. 2014;370(5):412–20.

27. Sheng Z, Tong D, Ou Y, et al. Serum sclerostin levels were positivelycorrelated with fat mass and bone mineral density in central southChinese postmenopausal women. Clin Endocrinol. 2012;76(6):797–801.

28. Amrein K, Amrein S, Drexler C, et al. Sclerostin and its associationwith physical activity, age, gender, body composition, and bonemineral content in healthy adults. J Clin Endocrinol Metab.2012;97(1):148–54.

29. Ma YH, Schwartz AV, Sigurdsson S, et al. Circulating sclerostinassociated with vertebral bone marrow fat in older men but notwomen. J Clin Endocrinol Metab. 2014;99(12):E2584–90.

30. Ukita M, Yamaguchi T, Ohata N, Tamura M. Sclerostin enhancesadipocyte differentiation in 3T3-L1 cells. J Cell Biochem. 2016Jun;117(6):1419–28.

31. Spatz JM, Wein MN, Gooi JH, et al. The Wnt-inhibitor sclerostin is up-regulated by mechanical unloading in osteocytes in-vitro. J BiolChem. 2015 Jul 3;290(27):16744–58.

32. Fulzele K, Krause DS, Panaroni C, et al. Myelopoiesis is regulated byosteocytes through Gsalpha-dependent signaling. Blood.2013;121(6):930–9.

33. Sakamoto A, Chen M, Kobayashi T, Kronenberg HM, Weinstein LS.Chondrocyte-specific knockout of the G protein G(s)alpha leads toepiphyseal and growth plate abnormalities and ectopic chondro-cyte formation. J Bone Miner Res. 2005;20(4):663–71.

34. Lu Y, Xie Y, Zhang S, Dusevich V, Bonewald LF, Feng JQ. DMP1-targeted Cre expression in odontoblasts and osteocytes. J Dent Res.2007;86(4):320–5.

35. ZhangM, Xuan S, Bouxsein ML, et al. Osteoblast-specific knockout ofthe insulin-like growth factor (IGF) receptor gene reveals an essentialrole of IGF signaling in bone matrix mineralization. J Biol Chem.2002;277(46):44005–12.

36. Powell WF Jr, Barry KJ, Tulum I, et al. Targeted ablation of the PTH/PTHrP receptor in osteocytes impairs bone structure and homeo-static calcemic responses. J Endocrinol. 2011;209(1):21–32.

37. Wein MN, Spatz J, Nishimori S, et al. HDAC5 controls MEF2C-drivensclerostin expression in osteocytes. J Bone Miner Res. 2015;30(3):400–11.

38. Vandesompele J, De Preter K, Pattyn F, et al. Accurate normalizationof real-time quantitative RT-PCR data by geometric averaging of

Journal of Bone and Mineral Research EFFECT OF SCLEROSTIN ON BEIGE ADIPOCYTES IN PERIPHERAL FAT 383

multiple internal control genes. Genome Biol. 2002;3(7):RESEARCH0034.

39. Cui Y, Niziolek PJ, Macdonald BT, et al. Lrp5 functions in bone toregulate bone mass. Nat Med. 2011;17(6):684–91.

40. Kalajzic I, Matthews BG, Torreggiani E, Harris MA, Divieti Pajevic P,Harris SE. In vitro and in vivo approaches to study osteocyte biology.Bone. 2013;54(2):296–306.

41. Chen M, Feng HZ, Gupta D, et al. G(s)alpha deficiency in skeletalmuscle leads to reduced muscle mass, fiber-type switching, andglucose intolerance without insulin resistance or deficiency. Am JPhysiol Cell Physiol. 2009;296(4):C930–40.

42. Considine RV, Sinha MK, Heiman ML, et al. Serum immunoreactive-leptin concentrations in normal-weight and obese humans. N Engl JMed. 1996;334(5):292–5.

43. Rosenbaum M, Nicolson M, Hirsch J, et al. Effects of gender, bodycomposition, and menopause on plasma concentrations of leptin.J Clin Endocrinol Metab. 1996;81(9):3424–7.

44. Lloyd SA, Lewis GS, Zhang Y, Paul EM, Donahue HJ. Connexin 43deficiency attenuates loss of trabecular bone and preventssuppression of cortical bone formation during unloading. J BoneMiner Res. 2012;27(11):2359–72.

45. Sinha P, Aarnisalo P, Chubb R, et al. Loss of Gsalpha in the postnatalskeleton leads to low bonemass and a blunted response to anabolicparathyroid hormone therapy. J Biol Chem. 2016;291(4): 1631–42.

46. Rao RR, Long JZ, White JP, et al. Meteorin-like Is a hormone thatregulates immune-adipose interactions to increase beige fatthermogenesis. Cell. 2014;157(6):1279–91.

47. Bostrom P, Wu J, Jedrychowski MP, et al. A PGC1-alpha-dependentmyokine that drives brown-fat-like development of white fat andthermogenesis. Nature. 2012;481(7382):463–8.

48. Krause A, Speacht T, Govey P, et al. Sarcopenia and increased bodyfat in sclerostin deficient mice. J Bone Miner Res. 2014;29 Suppl 1.[Presented orally at: Annual Meeting American Society for Bone andMineral Research (ASBMR); 2014 Sep 12–15; Houston, TX, USA;PresentationNumber: 1021]. Available from: http://www.asbmr.org/

education/AbstractDetail?aid=7daf67da-7cd2-44ea-9140-173cafb662d6

49. Weinstein LS, Xie T, Qasem A, Wang J, Chen M. The role of GNAS andother imprinted genes in the development of obesity. Int J Obes(Lond). 2010;34(1):6–17.

50. Meyer CW, Klingenspor M, Rozman J, Heldmaier G. Gene or size:metabolic rate and body temperature in obese growth hormone-deficient dwarf mice. Obes Res. 2004;12(9):1509–18.

51. Lee MW, Odegaard JI, Mukundan L, et al. Activated type 2 innatelymphoidcellsregulatebeigefatbiogenesis.Cell.2015;160(1–2):74–87.

52. Clarke BL, Drake MT. Clinical utility of serum sclerostin measure-ments. Bonekey Rep. 2013;2:361.

53. Garcia-Martin A, Rozas-Moreno P, Reyes-Garcia R, et al. Circulatinglevels of sclerostin are increased in patients with type 2 diabetesmellitus. J Clin Endocrinol Metab. 2012;97(1):234–41.

54. Modder UI, Hoey KA, Amin S, et al. Relation of age, gender, and bonemass to circulating sclerostin levels in women and men. J BoneMiner Res. 2011;26(2):373–9.

55. Urano T, Shiraki M, Ouchi Y, Inoue S. Association of circulatingsclerostin levels with fat mass and metabolic disease—relatedmarkers in Japanese postmenopausal women. J Clin EndocrinolMetab. 2012;97(8):E1473–7.

56. Popovic DS, Stokic E, Tomic-Naglic D, et al. Circulating sclerostin levelsand cardiovascular risk in obesity. Int J Cardiol. 2016;214:48–50.

57. Grethen E, Hill KM, Jones R, et al. Serum leptin, parathyroid hormone,1,25-dihydroxyvitamin D, fibroblast growth factor 23, bone alkalinephosphatase, and sclerostin relationships in obesity. J Clin Endo-crinol Metab. 2012;97(5):1655–62.

58. Cypess AM, Lehman S, Williams G, et al. Identification andimportance of brown adipose tissue in adult humans. N Engl JMed. 2009;360(15):1509–17.

59. KlangjareonchaiT,NimitphongH,SaetungS,etal.Circulatingsclerostinand irisin are related and interactwith gender to influence adiposity inadults with prediabetes. Int J Endocrinol. 2014;2014:261545.

384 FULZELE ET AL. Journal of Bone and Mineral Research