Q3

Q2

lable at ScienceDirect

Biochimie xxx (2015) 1e9

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354

55

BIOCHI4716_proof ■ 22 May 2015 ■ 1/9

Contents lists avai

Biochimie

journal homepage: www.elsevier .com/locate/biochi

565758596061626364

Research paper 656667686970

Mest and Sfrp5 are biomarkers for healthy adipose tissue

Magdalena Jura, Julia Jarosławska, Dinh Toi Chu, Leslie P. Kozak*

Institute of Animal Reproduction and Food Research, Polish Academy of Science, Olsztyn, Poland

7172

7374757677787980818283

a r t i c l e i n f o

Article history:Received 18 March 2015Accepted 9 May 2015Available online xxx

Keywords:Postnatal adiposity developmentObesityob/obC57BL/6JCaveolin

* Corresponding author. ul. Tuwima 10, 10-748 Olsz46 21, þ48 667 742 817.

E-mail address: l.kozak@pan.olsztyn.pl (L.P. Kozak

http://dx.doi.org/10.1016/j.biochi.2015.05.0060300-9084/© 2015 Published by Elsevier B.V.

84858687888990919293

Please cite this article in press as: M. Jura, e10.1016/j.biochi.2015.05.006

a b s t r a c t

Obesity depends on a close interplay between genetic and environmental factors. However, it is un-known how these factors interact to cause changes in the obese condition during the progression ofobesity from the neonatal to the aged individual. We have utilized Mest and Sfrp5 genes, two geneshighly correlated with adipose tissue expansion in diet-induced obesity, to characterize the obese con-dition during development of 2 genetic models of obesity. A model for the early onset of obesity waspresented by leptin-deficient mice (ob/ob), whereas late onset of obesity was induced with high-fat diet(HFD) consumption in C57BL/6J mice with inherent risk of obesity (DIO). We correlated obese anddiabetic phenotypes with Mest and Sfrp5 gene expression profiles in subcutaneous fat during pre-weaning, pre-adulthood and adulthood. A rapid development of obesity began in ob/ob mice immedi-ately after weaning at 21 days of age, whereas the obesity of DIO mice was not evident until after 2months of age. Even after 5 months of HFD treatment, the adiposity index of DIO mice was lower than inob/ob mice at 2 months of age. In both obesity models, the expression of Mest and Sfrp5 genes increasedin parallel with fat mass expansion; however, gene expression proceeded to decrease when the adiposityreached a plateau. The reduction in the expression of genes of caveolae structure and glucose metabolismwere also suppressed in the aging adipose tissue. The analysis of fat mass and adipocyte size suggeststhat reduction inMest and Sfrp5 is more sensitive to the age of the fat than its morphology. The balance offactors controlling fat deposition can be evaluated in part by the differential expression profiles of Mestand Sfrp5 genes with functions linked to fat deposition as long as there is an active accumulation of fatmass.

© 2015 Published by Elsevier B.V.

9495

96 979899

100101102103104105106107108109110111112

1. Introduction

Obesity, defined as an abnormal fat mass (FM) accumulation, isconsidered an epidemic problem leading to major health conse-quences in developing countries. Mechanisms controlling thedevelopment of adipose tissue are under intense study. Althoughmuch has been discovered about the adipocyte differentiation [1],the molecular regulation of adipose tissue expansion (ATE) is lessknown. The identification of genes controlling the size andenlargement of adipocytes could be crucial to the management ofobesity and insulin resistance.

Although the expression of many genes with putative roles intriglyceride (TG) synthesis, including lipoprotein lipase, fatty acidbinding protein 3 and steroyl CoA desaturase, occurs duringadipogenesis [2], establishing their roles in ATE as a consequence of

tyn, Poland. Tel.: þ48 89 523

).

113114115116

t al., Mest and Sfrp5 are bioma

an obesogenic environment is less definitive. There are genetic-developmental mechanisms of fat accumulation in adipocytesthat occur during the development of adipose tissue or the 3T3-L1pre-adipocyte cell culture model that show strong associationswith genes encoding proteins with putative functions in fat syn-thesis [3]. For example, the cytoplasmic glycerol-phosphate dehy-drogenase (GPDH) makes a transition from low expression of anembryonic isoform in the undifferentiated state to a 1000-foldaccumulation from an alternative gene locus during the differen-tiation process [4]. The location of GPDH in metabolic pathwayspredicts that the production of glycerol phosphate from glycolyticintermediates by this enzyme is important for accumulation oftriglycerides in adipose tissues, yet a spontaneous null mutation inthe BALB/cHeA subline does not cause a significant reduction inadipose levels and mice appear normal [5]. Furthermore, a trans-genic mouse with over-expression of GPDH does not have aphenotype with increased triglyceride levels, in fact is has reducedadiposity [6]. Therefore, despite a firm biochemical basis for a roleby GPDH in triglyceride synthesis and evidence for its accumulation

117118119

rkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/

M. Jura et al. / Biochimie xxx (2015) 1e92

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465

66676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130

BIOCHI4716_proof ■ 22 May 2015 ■ 2/9

during adipose tissue differentiation [7], over-expressing trans-genes and null mutations of GPDH do not support a role for GPDHin TG synthesis and ATE. These genes may be part of a fixed geneticdevelopmental program that is independent of environmentalmechanisms of fat accumulation, if indeed such mechanisms exist.

Recent microarray gene expression studies of obese mousemodels have identified two additional candidate genes encodingproteins, mesoderm specific transcript (Mest) and secretedfrizzled-related protein-5 (Sfrp5), that may drive ATE in an obe-sogenic environment [8e10]. In adult C57BL/6J (B6) mice withdiet-induced obesity high levels of expression of these genes showsignificant correlations with adiposity in both visceral and sub-cutaneous fat depots. The correlation is with the degree ofadiposity and is independent of the high-fat diet (HFD), unlike theinduction of Scd1 by HFD [11,12]. Importantly,Mestwas reported tobe highly expressed in embryos and at early neonatal stage of life[13] while Sfrp5 has very low levels of expression during the post-natal period of adipogenesis. Neither of these genes is correlatedwith adipogenesis in primary cell cultures [14]. On the other hand,Mest and Sfrp5 expression profiles were found to increase duringadulthood at the onset of positive energy balance [11], suggestingtheir role in ATE. Regarding SFRP5, a protein secreted by adipo-cytes, it has been associated with inflammation and insulin resis-tance in mouse models of obesity [15]. SFRP5, as other SFRPproteins, envelop Wnt proteins in the extracellular space andprevent binding of Wnt proteins to their receptors [16]. This actionby Sfrp5 was proposed to suppress oxidative metabolism by inhi-bition of Wnt signalling leading to a model inwhich suppression ofmitochondrial respiration by Sfrp5 provided conditions for lipidaccumulation in adipocytes [17]. Two mouse models with loss offunction for Mest or Sfrp5 genes, are resistant to diet-inducedobesity [7,17], suggesting a role for these genes in adipocyte hy-pertrophy [11]. The expression levels of Mest, but not Sfrp5, in pre-weaning mice could predict the propensity for obese phenotype inadult mice fed a HFD.

The multivariable nature of obesity predicts that differentpathways will be involved in its origin and development. Ourresearch describes how two genetic obesity models on the samegenetic background, that is, the C57BL/6J, provide two distinctobesity phenotypes. Adiposity phenotypes show variations thatresult from genetic and environmental conditions: a monogenicmodel of obesity is presented by the dominant effects of leptindeficiency in ob/obmice on food intake and fat metabolism, while apolygenic model of environmental obesity is represented by B6mice fed a HFD called diet-induced obesity group (DIO) [18e20].Obesity in polygenic B6 mice is a multifactorial trait, based uponallelic differences that condition susceptibility to HFD at a specifictime of development [21]. The multifactorial conditions promotingobesity in B6 mice may also have an impact on the adiposityphenotype in ob/ob mice [22]. Determining the mechanisms con-trolling ATE is important because of the association of ATE with thedevelopment of insulin resistance [23]. Our goal is to understand ifFM accumulation rate can be defined as a function of interactionsamong developmental mechanisms, allelic variation and environ-mental conditions. Thus, we hypothesize that different origins ofobesity represent different efficiency for FM storage, which affectsATE genes expression profile. To gain insights into the physiologicalrole of Mest and Sfrp5 genes, we conducted a time-course study toevaluate whether functional interdependences can be found be-tween obesity state (defined by adiposity index), Mest and Sfrp5gene expression. Comparing gene expression profiles betweendifferent types of obesity, as well as comparing obesity states tohealthy B6 mice fed low-fat standard diet (control group, ctr), willshowus phenotypic characteristics that distinguish obesity models,that is, early onset hyperphagic obesity in ob/ob mice vs, late onset

Please cite this article in press as: M. Jura, et al., Mest and Sfrp5 are biom10.1016/j.biochi.2015.05.006

diet-induced obesity in B6 wild-type mice. These models can pro-vide assessments on both genetic, developmental and environ-mental conditions that determine the potential roles of Mest andSfrp5 function on the capacity for white adipose tissue (WAT)accumulation.

2. Materials and methods

2.1. Animals

Breeding colonies of C57BL/6J (B6) were purchased from theJackson Laboratory (Bar Harbor, Maine USA) and B6.ob/þ breedingpairs were obtained from Dr. Martin Klingenspor. Mice werehoused under constant environmental conditions (12 h light/darkcycle at 23 �C) and fed ad libitumwith a low fat standard diet (STD)containing 13 kcal % (LabDiet 5053) or high-fat diet (HFD) with59 kcal % fat (TestDiet 21331). All experimental protocols, con-ducted at the Institute of Animal Reproduction and Food Research,Polish Academy of Sciences, Olsztyn, Poland, were approved by theLocal Committee for the Ethical Treatment of Experimental Ani-mals of WarmiaeMazury University (NR 38/2011). In the experi-mental protocol only male mice were used. All animals weresacrificed by decapitation, fat tissues (subcutaneous, epididymal,retroperitoneal fat pads) were rapidly removed and immediatelysnap-frozen in liquid nitrogen and stored in �80 �C until furtheranalysis.

2.2. Animal protocol

In the experimental protocol for diet-induced obesity B6 micewere fed STD for the first 2 months of age and then were dividedinto two groups: (i) mice were fed STD until 8 months of age (ctr),or (ii) mice were fed HFD from the 2nd until the 8th month of age(DIO). A third experimental group was represented by ob/ob micefed STD until the 5th month of age. Control and DIO mice weresacrificed at 10 and 21 days of age and 2, 5, 7 and 8 months of age.Additionally, ob/ob mice were sacrificed at 10 and 21 days of ageand at 2 and 5 months of age.

Another cohort of mice was generated to maximize theadiposity index in preweaning B6 wild-type and ob/ob mice tolevels of obesity found in adult micewith DIO, by feeding the damesand their litters with a HFD from birth to weaning.

2.3. Adiposity phenotype

For experimental animals' body weight and body compositionwas monitored on a weekly basis by nuclear magnetic resonance(NMR, Bruker, Germany). Adiposity index (fat mass/lean mass) is ameasure of the enrichment of fat mass relative to the lean mass; forhealthy, non-diabetic, adult mice it lies between 0.22 and 0.25.

2.4. RNA isolation and RT-PCR analysis

Adipose tissue was homogenized in TRI reagent according tomanufacturer's directions (Molecular Research Centre, USA). RNAwas isolated from homogenates of one fat pad and then storedat �80 �C in RNase-free water with SUPERase In (Ambion, USA)following the manufacturer's instructions. RNA quality waschecked by agarose gel electrophoresis by visualization of ribo-somal RNA subunits and RNA concentrationwasmeasured with theNanoDrop1000 spectrophotometer (Thermo Scientific, USA).Quantitative RT-PCR using TaqMan probes and primers (AppliedBiosystems, Foster City, CA, USA) was performed with standardcurves generated from pooled RNA isolated from interscapularbrown adipose tissue and white adipose tissue collected from adult

arkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/

M. Jura et al. / Biochimie xxx (2015) 1e9 3

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465

666768

BIOCHI4716_proof ■ 22 May 2015 ■ 3/9

B6 mice. Standards and unknown samples were run in duplicatesand normalized to Cyclophilin b. Sequences of probes and primersare available upon request.

697071727374757677787980818283848586878889

2.5. Western blot analysis

Protein expression of adipose tissues was analyzed by Westernblotting. RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mMEDTA, 0.1% SDS, 1% NP40, 0.25% sodium azide) contained proteaseinhibitor (phenylmethylsulfonyl fluoride, PMSF e Sigma, P7626),phosphatase inhibitor (PhosStop e Roche, 4906837001) and pro-teinase inhibitor cocktails (Sigma, P8340). After centrifugation su-pernatant was collected and protein concentrationwas determinedusing Bradford reagent (Sigma B6916). Western blot analysis wasperformed as described by Xue et al. [24] To normalize the proteinloading, a mouse monoclonal antibody against b actin (Santa CruzBiotech, Sc-47778) (1:10000) was used. Blots were incubated withpolyclonal antibodies anti-MEST (rabbit anti-MEST e kindly pro-vided by L. Nikonova [11]) (1:500), anti-SFRP5 (goat anti-SARP3SantaCruz) (1:400). Bands were developed and quantified by Od-yssey imaging system software (Li-Cor, USA) with fluorescent-labeled antibodies IRDye800 and IRDye700 (Rockland, USA) andnormalized against b actin band.

Fig. 1. Developmental study. A- Body weight, B- Fat mass, C- Lean mas, D- Adiposity index,mice (ob/ob) and C57BL/6 mice fed a STD diet (ctr) or a high-fat diet mice from 2 months of agto control and ($) significant differences between ob/ob and DIO mice *P < 0.05, $P < 0.05.

Please cite this article in press as: M. Jura, et al., Mest and Sfrp5 are bioma10.1016/j.biochi.2015.05.006

2.6. Adipocyte cell size

Subcutaneous fat tissues were fixed in 6% formalin, embeddedin paraffin and stained with hematoxylin and eosin for generalmorphological examination. Adipocyte cell area size was calculatedwith a use of cellSens imaging software.

2.7. Statistical analysis

The data are given as mean ± SD. Data comparison, statisticaldifferences was analysed by ANOVA test or by student's t-test withWelch's correction or ManneWhitney test. The level of significancewas set at p < 0.05. Data analyses were performed using Graph PadPrism 6.0 statistical software.

3. Results

An overview of the adiposity profile and expression of Mest andSfrp5 is shown in Fig. 1 for B6 control (ctr) and ob/ob mice (ob/ob)fed a standard diet and B6 mice fed a HFD beginning at 2 months ofage (DIO). The adiposity phenotypes are predictable, that is, nosignificant change in the adiposity of B6 mice fed a chow diet, arapid and robust induction of adiposity in ob/ob, as expected from

E- Mest gene expression profiles, F- Sfrp5 gene expression profiles for leptin-deficiente (DIO). Data showmean values for n ¼ 7e8; (*) significant differences for comparisons

90919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130

rkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/

M. Jura et al. / Biochimie xxx (2015) 1e94

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465

6667686970717273747576777879808182838485868788

BIOCHI4716_proof ■ 22 May 2015 ■ 4/9

the leptin deficiency, hyperphagic phenotype, and a gradual in-duction of adiposity in male DIO mice. However, the apparent lackof correlation of profiles of adiposity with the expression of Mestand Sfrp5 genes was not predicted. Below we examined the early,middle and late developmental phases of adiposity in more detail.

3.1. Phenotypes during pre-weaning period

Body weight (BW) did not differ between wild-type and ob/obmice at 10 days of age (Fig. 1A), nevertheless due to differences infat mass (FM) ob/ob mice already presented higher adiposity index(AI) (Fig. 1D). In the early stages of development, from 10 to 21 daysof age, both groups of mice significantly increased their BW, FM,lean mass (LM) and AI (Figs. 1 and 2AeD). At weaning, BW was thesame for all the groups, but ob/ob mice showed slightly higher FMand smaller LM in comparison to B6 mice, thus AI was greater forob/ob mice (Fig. 2F). The mRNA of Mest was highly expressed at 10days of age (Fig. 2B) and very low at 21days of age as previouslydescribed [11,14] (Fig. 2G), with greater values at both time-pointspresented by ob/obmice. Sfrp5mRNA levels were very lowat 10 and21 days of age; the expression level did not differ between groups at10 days of age (Fig. 2B), but it was higher for ob/ob mice in com-parison with B6 mice at 21 days of age (Fig. 2H). During their

Fig. 2. Developmental study of young C57BL/6 (B6) and leptin-deficient mice (ob/ob). For 10adiposity), B- Gene expression profiles of Mest and Sfrp5. And for 21 days old mice: C- Bodyshow mean values for n ¼ 8e12; * significant differences *P < 0.05, **P < 0.01, ***P < 0.00

Please cite this article in press as: M. Jura, et al., Mest and Sfrp5 are biom10.1016/j.biochi.2015.05.006

development mice also accumulate FM, but during the pre-weaning period mice did not achieve sufficient adiposity (AI>0.25) to promote higher expression of Sfrp5. On the other hand,Mest has very high expression at 5 days of age, even though the AIis < 0.25 [11], which gradually reduced until it was almost unde-tectable at 21 days of age. Therefore, Mest, which has expressionhighly correlated for young adults at the onset of positive energybalance [11], is completely un-correlated in the pre-weaningperiod.

We performed a second experiment where we sought to forceinduction of Mest and Sfrp5 by expanding FM to achieve an AIhigher than 0.25 by feeding B6 and ob/ob mice a HFD from birth toweaning. With this HFD protocol both B6 wild-type and ob/obmicedeveloped significantly higher BW, FM and AI although theadiposity was higher in ob/ob mice. Under this HFD protocol, MestmRNA levels increased in both mouse models as the AI augmented.It is also evident that the higher AI in the ob/ob mice led to higherlevels of Mest expression (Fig. 2G). During the pre-weaning period,Sfrp5 gene expression was increased by HFD in B6 mice anddecreased in ob/ob mice (Fig. 2H).

When HFD treatment augmented the BW range, we foundstrong associations between AI and Mest expression, as measuredin subcutaneous fat for ob/ob mice (Fig. 3A). Conversely, the same

days of age mice: A- Phenotype characteristics (body weight, fat mass, lean mass andweight, D- Fat mass, E- Lean mass, F- Adiposity, G- Mest mRNA, H- Sfrp5 mRNA. Data

1, ****P < 0.0001.

8990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130

arkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/

Fig. 3. Correlation of Mest gene expression with adiposity index. A-ob/ob mice at 21days of age, mice were fed high-fat diet from birth, B- ob/ob mice at 5 month of age fed standarddiet from birth, C- C57BL/6 (B6) strain of mice at 21days of age, mice were fed high-fat diet (HFD) from birth, D- B6 mice at 5 month of age fed HFD from 2nd month of age (DIO).After expanding the range of adiposity index to the higher degree by high-fat diet, young B6 mice at 21 days of age showed no correlation between Mest mRNA level and adiposityindex, while young ob/ob showed strong correlation between those factors measured in subcutaneous fat. Reverse results were observed in adult animals. 21d n ¼ 10e12; 5Mn ¼ 7e9.

M. Jura et al. / Biochimie xxx (2015) 1e9 5

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465

66676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130

BIOCHI4716_proof ■ 22 May 2015 ■ 5/9

was not observed for B6mice (Fig. 3C). Accordingly, during this pre-weaning developmental period, despite the increased adiposityinduced by the HFD treatment in wild-type B6 mice, no correlationwas observed between AI and Mest and/or Sfrp5 genes.

3.2. Phenotypes during pre-adulthood from weaning to 2 months ofage

By 2 months of age (2M), ob/ob mice had significantly highervalues of FM and BW (Fig. 1A and B and Fig. 4AeC) in comparisonwith the wild-type B6 mice. In contrast to DIO mice, whichincreased their FM and BW progressively, ob/ob mice increasedtheir FM until the 3M, afterwards ob/ob had reached a plateauwithout further significant changes (Fig. 4C and F). At weaning ob/obmice rapidly increased their FM and BW due to their compulsiveovereating which continued for the duration of the experiment. Incontrast, DIO group in wild-type B6 mice started to increase FMuntil after 2M when they were introduced to HFD. At 3M DIO micewere at the onset of maximally increasing their FM, and at 3M DIOmice manifested significantly higher AI in comparison to controlmice fed STD (Fig. 4).

Between 21 days and 2M the AI did not increase above 0.25 forB6 mice fed a STD and, as predicted, the levels of Mest and Sfrp5mRNA did not increase. In contrast, as ob/obmice were undergoingtheir most dynamic accumulation of FM from weaning to 2M,during this time Mest and Sfrp5 genes reached their highest levelsof expression in ob/ob mice. The HFD treatment stimulated Mestand Sfrp5 expression and DIO mice showed progressively highertranscript levels (Fig. 1E and F). A different response was observedin ob/ob mice, Mest and Sfrp5 mRNA transcript levels reached amaximum at 2M and then started to decrease (Fig. 1E and F). Thesedata demonstrate that although both types of obesity models

Please cite this article in press as: M. Jura, et al., Mest and Sfrp5 are bioma10.1016/j.biochi.2015.05.006

accumulate fat at different stages of life, Mest and Sfrp5 expressionincreases in parallel with ATE.

3.3. Mature and obese animals

Body composition analysis showed that regardless of the originof obesity, that is, by genetically determined hyperphagia or hy-persensitivity to HFD,micewill reach a stable adiposity level, whichoccurs at a different age for our 2 models. The susceptibility tohyperphagia from leptin deficiency begins near weaning, whereasthe susceptibility to DIO begins at 2M. From 3M to 5M, ob/ob miceshow only a slight increase in adiposity, and changes in AI and FMstabilized after the 3M (Fig. 4C and F). DIO mice presented a verysharp increase in all adiposity parameters when initially exposed toa HFD at 2M (Fig. 1); however, their AI and FM did not change afterthe 5M (Fig. 4B). Moreover, no statistical differences were detectedon obesity phenotypes of control B6 mice fed a STD after the 2M(Fig. 4A and D). Even though ob/ob mice at 5M had significantlyhigher AI and FM levels than DIOmice (Fig. 1B and D), DIOmice hadmuch higher levels ofMest and Sfrp5mRNA expression at this time-point (Fig. 1E). This could be explained by the fact that Mest andSfrp5 mRNA levels in ob/ob sharply decreased between the 2M and5M, whereas for DIO mice the increase in both adiposity and Mestand Sfrp5 levels during this period of timewas maximal. However asimilar fate awaits B6 DIO mice at 8M, that is, declining geneexpression occurred despite a stable level of adiposity. The timecourse study showed that both types of obesity have similar geneexpression profiles, but in DIO mice the decline in Mest and Sfrp5mRNA levels is delayed.

During pre-weaning, mice fed a standard diet showed no sig-nificant correlation between Mest and AI (R2 ¼ 0.14, p ¼ 0.14 for B6mice, R2 ¼ 0.04, p ¼ 0.75 for ob/ob). Nevertheless, after the range of

rkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/

Fig. 4. Developmental study of adiposity (A,B,C) and of fat mass accumulation (D,E,F) in adult C57BL/6 (B6) mice fed standard diet (ctr) or fed high-fat diet (DIO) and leptin-deficientmice (ob/ob). Data show the mean values ±SD for n ¼ 7e8; (*) significant differences between following time-points, *P < 0.05, **P < 0.01, and ***P < 0.001 ****P < 0.0001.

M. Jura et al. / Biochimie xxx (2015) 1e96

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465

66676869707172737475767778798081828384858687888990919293949596979899

100101102103

BIOCHI4716_proof ■ 22 May 2015 ■ 6/9

adiposity was extended with a use of HFD during pre-weaning, asignificant positive correlation betweenMest expression and AI wasobserved for ob/ob group of mice (Fig. 3A and C). Since the mean AIfor ob/ob was 0.56 and that for B6 wild-type was 0.32, it suggestedB6 wild-type mice did not achieve a sufficiently high level ofadiposity. There was a broad range of adiposity among obese adultmice, however at 5M only DIO mice showed a significant positivecorrelation between Mest and AI (Fig. 3B and D).

Histological evaluation of subcutaneous adipocyte cell size at5M demonstrated that ob/obmice presented the biggest adipocytes

Fig. 5. Age and diet effect on adipocyte cell size in subcutaneous fat. A- Average adipocyteexpression. D- Representative hematoxylin and eosin images were taken at �10 magnificatiostandard diet (ctr 5M) or fed high-fat diet (DIO 5M) and leptin-deficient mice (ob/ob 5M), and(DIO 7M). Data are expressed as means ± SE, n ¼ 2e3. Significant differences *P < 0.05, **

Please cite this article in press as: M. Jura, et al., Mest and Sfrp5 are biom10.1016/j.biochi.2015.05.006

area (Fig. 5). It is noteworthy that even after long duration of anobesogenic environment leading to adiposity development, DIOmice at 7M presented the highestMest expression level, but did notreach the adipocyte size of ob/ob mice at 5M.

3.4. Other markers of obese state

We observed significantly lower levels of Glucose transporter 4(Glut4) mRNA expression for ob/ob in comparison to other experi-mental groups (Fig. 6A). The decreased expression levels may

area calculated for 5 representative areas. B- Mest mRNA expression. C- Sfrp5 mRNAn, where scale bar represents 100 mm. At 5 month of age in adult C57BL/6 (B6) mice fedafter prolonged high-fat diet treatment when DIO mice stabilized their adiposity index

P < 0.01, and ***P < 0.001, ****P < 0.0001.

104105106107108109110111112113114115116117118119120121122123124125126127128129130

arkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/

Fig. 6. At 5 months of age, obesity origin affects expression profiles of genes involved in fat accumulation in mice, A- Glucose transporter 4 (Glut4), B- Fatty acid synthase (FAS), C-Caveolin1 (Cav1), D- Caveolin2 (Cav2), E- Polymerase I and transcript release factor (Cavin), F- Proliferation cell nuclear antigen (PCNA), G- S-phase kinase-associated protein 2 (Skp2)and inflammatory biomarkers H- Interleukin6 (IL-6), I- Tumor necrosis factor alpha (TNFa), for leptin-deficient mice (ob/ob) and C57BL/6 mice fed a STD diet (ctr) or high-fat dietmice from 56 days of age (DIO). Data show mean values for n ¼ 7e8; significant differences for comparisons to control *P < 0.05, **P < 0.01, and ***P < 0.001 ****P < 0.0001.

M. Jura et al. / Biochimie xxx (2015) 1e9 7

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465

66676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130

BIOCHI4716_proof ■ 22 May 2015 ■ 7/9

indicate the development of insulin resistance [25]. Moreover ourresults showed very low levels of fatty acid synthase (Fasn) tran-scription in control and ob/obmice, but high for DIOmice indicatingthat only this group is effectively accumulating fat (Fig. 6B) [26].Additional genes reported to be associated with ATE are Caveolin1(Cav1) and Caveolin2 (Cav2) [27]. At 5M only DIO mice had highlevels of Cav1, Cav2 and Cavin transcripts (Fig. 6CeE), suggestingintensified free fatty acid transport and fat accumulation for thisgroup. Moreover, characteristic for obesity is an increase in thenumber of cells as a result of cell growth and cell division. Prolif-eration markers such as Proliferation cell nuclear antigen (Pcna) [28]and S-phase kinase-associated protein 2 (Skp2) [29] showed thatthere is no difference between DIO and ob/ob mice at 5M (Fig. 6Fand G). Additionally, obesity is often associated with increasedcirculating pro-inflammatory cytokines such as TNFa and Il-6 [30].At 5M, both macrophage markers showed significantly higherlevels for ob/obmice in comparisonwith control group (Fig. 6H andI). Mild increase of TNFa and Il-6 was observed also in DIO group,but differences did not reach significance. DIO group at 8M showedcomparable results to DIO mice at 5M, with exception of Cav1 andCav2 genes showing smaller values at 8M.

3.5. Mest and Sfrp5 mRNA levels do not predict protein levels

At 21D of age, protein level of MEST showed no differencesbetween dietary protocols in ob/ob mice (Fig. 7A), despite thechanges detected for mRNA (Fig. 2G). SFRP5 protein at 21d was verylow irrespective of the dietary conditions, in fact it is not certainwhether any differential expression of SFRP5 protein occurs as afunction of age, diet or genotype (compare Fig. 7AeC). At 2M, MESTprotein achieved its highest levels in ob/ob mice (Fig. 7B), but wasundetected in B6 mice fed STD (Fig. 7B). After 2M of age, MESTprotein levels altered as the obesity phenotype progressed betweenmodels. In ob/ob mice MEST protein decreased from 2M to 5M(Fig. 7B), while the introduction of DIO in B6 mice at 2M promptedMEST protein to be expressed at the highest level at 5M (DIO 5M,Fig. 7B). In DIO model, MEST protein decreased afterwards in 7Mand 8M (Fig. 7B). The SFRP5 protein in adult ob/ob mice wasmodestly expressed at 2M and 5M (Fig. 7C). In the DIO model, theintroduction of HFD at 2M stimulated SFRP5 protein expression

Please cite this article in press as: M. Jura, et al., Mest and Sfrp5 are bioma10.1016/j.biochi.2015.05.006

with the highest values at 5M that subsequently declined in 7M and8M DIO mice (Fig. 7C).

4. Discussion

The important factors for the regulation of lipogenesis, lipolysisor adipogenesis are well understood; however, the factors andmechanisms that set the limits of ATE are poorly described. In adultwild-type B6 mice, fed a chow diet, the quantity of fat is approxi-mately 20%, and the variance is low; however, a HFD inducesobesity with wide variability [31]. The differences in obesity be-tween genetically identical animals in a seemingly indistinguish-able environment highlights the delicate balance of factorscontrolling fat deposition. Also in humans a large variation inadipocyte size was evidenced within and between individuals [32].Thus, using two different genetic models for obesity generation, weshowed that the balance of factors affecting and controlling fatdeposition may be associated with differential regulation of Mestand Sfrp5. High levels of expression define the capacity to accu-mulate fat in a state of positive energy balance, and low expressionsreflects limitation of FM accumulation. This may hold in matureadult mice, but not in mice from the pre-weaning period of life, norin aged mice which have different use or no use for their fatdeposits.

Mammals are provided with energy storing WAT [33], and Hanet al. showed that adipocytes of subcutaneous fat contained lipidcomponents already on postnatal day one [34]. During childhoodobesity, ATE arises from combined adipocyte hyperplasia and hy-pertrophy, but during adulthood there is a fixed number of adi-pocytes and changes in FM are primarily determined by changes infat-cell volume [35]. A positive energy balance triggers theresponse of Mest and Sfrp5 genes, since their increased expressionis highly correlated with ATE and adipocyte hypertrophy [11,17].The surrounding environment of the adipocyte is important for theATE phenotype. We have previously shown an association of genesencoding components of the cytoskeleton, including vimentin andtubulins with ATE [27]. Moreover loss of function of particularcomponents of the extracellular matrix such as collagen VI alpha 3,Laminin alpha 4 or osteonectin results in resistance to diet-inducedobesity [36e38].Mest and Sfrp5 knockouts are also resistant to diet-

rkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/

Fig. 7. MEST and SFRP5 protein expression levels. A- MEST and SFRP5 protein for ob/obmice at 21days of age fed STD or fed HFD, B- MEST protein for adult B6 fed STD till 2nd monthof age (B6 2M), adult B6 mice fed high-fat diet till 5th, 7th and 8th month of age (DIO 5M, DIO7M, DIO8M), adult ob/obmice fed STD until 2nd (ob/ob 2M) and 5th (ob/ob 5M) monthof age. C- SFRP5 protein for adult B6 fed STD till 2nd month of age (B6 2M), adult B6 mice fed high-fat diet till 5th, 7th and 8th month of age (DIO 5M, DIO7M, DIO8M), adult ob/obmice fed STD until 2nd (ob/ob 2M) and 5th (ob/ob 5M) month of age. Immunoblot analysis of MEST and SFRP5 proteins were performed in subcutaneous fat. Data show the meanvalues ±SD for n ¼ 4; (*) significant differences *P < 0.05, **P < 0.01, and ***P < 0.001 ****P < 0.0001.

M. Jura et al. / Biochimie xxx (2015) 1e98

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465

66676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130

BIOCHI4716_proof ■ 22 May 2015 ■ 8/9

induced obesity, but these genes are highly expressed inside thecell [14], so they might work paracrinally and interfere withextracellular matrix components. Although there are alreadyidentified transcription factors associated with changes occurringin adipose tissue such as early B cell factor 1 [39], the specific genesinvolved in defining adipose tissue morphology are still largelyunknown.

In our experiment, polygenic model of environmental obesitywas presented by wild-type B6 mice which were chronicallyexposed to HFD for up to 8M. Monogenic obesity was presented byob/ob mice, a mouse model characterized by marked hyperphagiaand early-onset obesity. Our results on phenotype data showed thatBW of ob/ob mice began to plateau by 3M supporting Lagathu et al.data [40]. Following the phenotypic profile of FM accumulation ofboth mouse models the kinetics of ATE based on hyperphagia in ob/ob mice is much more rapid than that resulting from the complexdifferences in and lipid and energy metabolism underlying DIO inB6mice. At 8MDIO B6mice did not reach the AI levels of ob/obmiceat 2M and this difference may be associated with the phenotypes ofMest and Sfrp5.

Both Mest and Sfrp5 have previously been reported to be WATmarkers responsible for ATE [10,27]. Together they presentedsimilar profiles in DIO during adulthood; however, no previousreport has characterized the time dependent expression of thesemarkers in both models. Our results support the data obtained byKozak et al. thatMest is highly expressed at neonatal stage and thenat the onset of positive energy balance [27] supporting its role inATE during adulthood. An unexpected decrease of these genes at8M when AI was stable suggests an end to the involvement of Mestand Sfrp5 in establishing the capacity for maximal ATE. It waspreviously shown that it is not obesity per se, but adipose cell size

Please cite this article in press as: M. Jura, et al., Mest and Sfrp5 are biom10.1016/j.biochi.2015.05.006

that leads to elevated expression of Mest mRNA [11,13]. However,our data showing that severely obese 5M ob/ob mice have muchlarger adipocytes than 7M DIO mice, but lower Mest expression,suggests that the regulation of Mest mRNA levels is dependent onadditional unknown factors. Furthermore, immunoblot data onMEST indicates that translational regulation of MEST and SFRP5 hasfurther complexity. Thus, whileMestmRNA expression is related toadipocyte size during the dynamic phase of fat accumulation, ageand the metabolic origins of obesity can introduce additional reg-ulatory factors.

It has been proposed that MEST is an acyl transferring enzymecatalyzing acylation of a biological molecule [13]. Both DIO and ob/ob mice at 5M differ in Mest mRNA levels but have the same MESTprotein levels, it is plausible that MEST protein is not translatedanymore or is degraded as its function is not needed. Since MESTprotein levels are the highest for DIO mice at 5M and diminish withage the protein may no longer be required for lipid synthesis inadipocytes, especially after maximal FM accumulation (Fig. 7).

Although Sfrp5 has been studied intensively in models of mouseobesity, its pivotal role in ATE remains unclear. The mRNA of Sfrp5has been shown to be up-regulated [17] and down-regulated [15],and Sfrp5 KO being sensitive [15] or not [17] to diet-induced obesity.Our results are in line with the findings that ob/ob mice [17,40] aswell as DIOmice [8] have elevated Sfrp5 expression inWAT.We alsoobserved that during pre-weaning period Sfrp5 mRNA was almostundetectable regardless of the leptin status of the mouse model orthe composition of a diet, but with obese development in pre-adultmice a robust induction in Sfrp5 mRNA levels occurred. Ouchi et al.previously proposed that Sfrp5 controls the microenvironment ofWAT under conditions of metabolic stress [15]. In an earlier studySfrp5was highly correlated withMest and AI in B6mice fed 8 weeks

arkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/

Q1

M. Jura et al. / Biochimie xxx (2015) 1e9 9

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465

66676869707172737475767778798081828384858687888990919293949596979899

100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130

BIOCHI4716_proof ■ 22 May 2015 ■ 9/9

a HFD [8]. Generally, the pattern of Mest and Sfrp5 expression inaging mice are similar. It is not clear why we observed no corre-lation between these two genes, possibly the number of mice wastoo low. Whether the suppression of Mest and Sfrp5 in tissues withelevated expression of the macrophage markers, IL-6 and TNFa, hasfunctional significance to adipose tissue inflammation is unclear[15].

Previous studies based upon microarray analysis of geneexpresssion in subcutaneous fat identified Cav1 and Cav2 as geneshighly correlated with Mest gene expression [27]. Caveolins areprincipal components of caveolae structure, which function insignal transduction and are believed to play a role in free fatty acidtransport [41]. Our results showed significantly higher expressionprofiles for both Cav1, Cav2 in DIO mice at 5M in comparison to 5Mob/ob mice. These data suggest that mice fed HFD still have highpotential for FM accumulation and increased free fatty acidstransport, whereas ob/ob mice have lost this feature [42]. Cavin,known as polymerase I and transcript release factor, is also critical forcaveolae formation [42]. Additionally Cavin was reported to beassociated with limited ATE [43] and it was suggested that cavinmay be an adipocyte marker for predicting pathological obesity[43]. Here we report significantly higher cavin expression level forDIO group of mice at 5M and at 8M in comparison to ob/ob mice at5M, indicating that DIO mice at 8M are still accumulating fat.Comparable expression levels of genes whose products are impli-cated in fatty acid uptake observed for ob/ob and control miceindicate that the FM accumulation processes were slowed down atthis age for genetically obese mice.

In summary, previous studies have shown strong positive cor-relations between ATE, adipocyte size and mRNA levels for Mestand Sfrp5 [10,11,13,27]. The data in this paper show exceptions tothis trend, the adipocytes in ob/ob mice are much larger that thosein B6 mice with DIO at a similar age; however the levels of geneexpression are much higher in mice with DIO at all times. The in-crease in gene expression in ob/ob mice is restricted to the periodwhen adiposity is increasing, that is, 2e3 months of age and thengene expression falls to almost base line levels by 5 months of age.The absence of leptin could be a contributing factor given thestrong suppression of several genes of carbohydrate and lipidmetabolism in 5M ob/ob, but high expression in 5M DIO mice.Nevertheless, gene expression in the DIOmice is not immune to thesuppression of gene expression as it also begins to be observed by8M. The delay in this suppressed phenotype may be due to themore gradual development of maximal obesity in DIO mice. Weinterpret the relationships of the genes we have analyzed to theattainment of adiposity phenotypes. In B6 mice in energy balancewith an AI of 0.25 the expression of Mest and Sfrp5 is almost un-detectable; however, in mice in an obesogenic environment or withhyperphagia, gene expression is robust. Importantly there is a limitto the gene expression, once maximal adiposity is reached, insulinresistance and glucose intolerance develops as well as macrophageinfliltration and tissue inflammation, characteristics of unhealthyadipose tissue. At this pathological stage expression of Mest andSfrp5, as well as other metabolic genes, diminishes.

Acknowledgments

This work was supported by a grant to LPK from the Foundationfor Polish Science, programme WELCOME, no. WELCOME/2010-4/3entitled “Nutrition and ambient temperature during early devel-opment can reduce susceptibility to obesity” financed by EUStructural Funds in Poland within the Innovative Economy Pro-gramme and REFRESH project (FP7-REGPOT-2010-1-264103). Wethank Agnieszka Korytko for outstanding support in the manage-ment of the mouse colony.

Please cite this article in press as: M. Jura, et al., Mest and Sfrp5 are bioma10.1016/j.biochi.2015.05.006

References

[1] C.M. Smas, H.S. Sul, Biochem. J. 309 (Pt 3) (1995) 697e710.[2] P.J. Smith, L.S. Wise, R. Berkowitz, C. Wan, C.S. Rubin, J. Biol. Chem. 263 (1988)

9402e9408.[3] O.A. MacDougald, M.D. Lane, Annu. Rev. Biochem. 64 (1995) 345e373.[4] L.S. Wise, H. Green, J. Biol. Chem. 254 (1979) 273e275.[5] M. Prochazka, U.C. Kozak, L.P. Kozak, J. Biol. Chem. 264 (1989) 4679e4683.[6] L.P. Kozak, U.C. Kozak, G.T. Clarke, Genes. Dev. 5 (1991) 2256e2264.[7] L. Lefebvre, S. Viville, S.C. Barton, F. Ishino, E.B. Keverne, M.A. Surani, Nat.

Genet. 20 (1998) 163e169.[8] R.A. Koza, L. Nikonova, J. Hogan, J.S. Rim, T. Mendoza, C. Faulk, J. Skaf,

L.P. Kozak, PLoS Genet. 2 (2006) e81.[9] A. Soukas, P. Cohen, N.D. Socci, J.M. Friedman, Genes. Dev. 14 (2000) 963e980.

[10] A. Voigt, K. Agnew, E.M. van Schothorst, J. Keijer, S. Klaus, Mol. Nutr. Food Res.57 (2013) 1423e1434.

[11] L. Nikonova, R.A. Koza, T. Mendoza, P.M. Chao, J.P. Curley, L.P. Kozak, FASEB J.Off. Publ. Fed. Am. Soc. Exp. Biol. 22 (2008) 3925e3937.

[12] M.T. Flowers, J.M. Ntambi, Biochim Biophys Acta 1791 (2009) 85e91.[13] M. Takahashi, Y. Kamei, O. Ezaki, Am. J. Physiol. Endocrinol. Metab. 288 (2005)

E117eE124.[14] D.T. Chu, E. Malinowska, B. Gawronska-Kozak, L.P. Kozak, J. Biol. Chem. 289

(2014) 18478e18488.[15] N. Ouchi, A. Higuchi, K. Ohashi, Y. Oshima, N. Gokce, R. Shibata, Y. Akasaki,

A. Shimono, K. Walsh, Science 329 (2010) 454e457.[16] P. Bovolenta, P. Esteve, J.M. Ruiz, E. Cisneros, J. Lopez-Rios, J. Cell. Sci. 121

(2008) 737e746.[17] H. Mori, T.C. Prestwich, M.A. Reid, K.A. Longo, I. Gerin, W.P. Cawthorn,

V.S. Susulic, V. Krishnan, A. Greenfield, O.A. Macdougald, J. Clin. Invest. 122(2012) 2405e2416.

[18] T. Garland Jr., H. Schutz, M.A. Chappell, B.K. Keeney, T.H. Meek, L.E. Copes,W. Acosta, C. Drenowatz, R.C. Maciel, G. van Dijk, C.M. Kotz, J.C. Eisenmann,J. Exp. Biol. 214 (2011) 206e229.

[19] A. Sclafani, Int. J. Obes. 8 (1984) 491e508.[20] Y. Zhang, R. Proenca, M. Maffei, M. Barone, L. Leopold, J.M. Friedman, Nature

372 (1994) 425e432.[21] K. Almind, C.R. Kahn, Diabetes 53 (2004) 3274e3285.[22] S.C. Chua Jr., Behav. Genet. 27 (1997) 277e284.[23] U. Ozcan, Q. Cao, E. Yilmaz, A.H. Lee, N.N. Iwakoshi, E. Ozdelen, G. Tuncman,

C. Gorgun, L.H. Glimcher, G.S. Hotamisligil, Science 306 (2004) 457e461.[24] B. Xue, A. Coulter, J.S. Rim, R.A. Koza, L.P. Kozak, Mol. Cell. Biol. 25 (2005)

8311e8322.[25] D. Leto, A.R. Saltiel, Nat. Rev. Mol. Cell. Biol. 13 (2012) 383e396.[26] G. Shillabeer, J. Hornford, J.M. Forden, N.C. Wong, D.C. Lau, J. Lipid Res. 31

(1990) 623e631.[27] L.P. Kozak, S. Newman, P.M. Chao, T. Mendoza, R.A. Koza, PLoS One 5 (2010)

e11015.[28] Y. Takasaki, J.S. Deng, E.M. Tan, J. Exp. Med. 154 (1981) 1899e1909.[29] T. Sakai, H. Sakaue, T. Nakamura, M. Okada, Y. Matsuki, E. Watanabe,

R. Hiramatsu, K. Nakayama, K.I. Nakayama, M. Kasuga, J. Biol. Chem. 282(2007) 2038e2046.

[30] M.E. Starr, B.M. Evers, H. Saito, J. Gerontol. Ser. A Biol. Sci. Med. Sci. 64 (2009)723e730.

[31] J. Moitra, M.M. Mason, M. Olive, D. Krylov, O. Gavrilova, B. Marcus-Samuels,L. Feigenbaum, E. Lee, T. Aoyama, M. Eckhaus, M.L. Reitman, C. Vinson, GenesDev. 12 (1998) 3168e3181.

[32] P. Arner, S. Bernard, M. Salehpour, G. Possnert, J. Liebl, P. Steier, B.A. Buchholz,M. Eriksson, E. Arner, H. Hauner, T. Skurk, M. Ryden, K.N. Frayn, K.L. Spalding,Nature 478 (2011) 110e113.

[33] B. Cannon, J. Nedergaard, Physiol. Rev. 84 (2004) 277e359.[34] J. Han, J.E. Lee, J. Jin, J.S. Lim, N. Oh, K. Kim, S.I. Chang, M. Shibuya, H. Kim,

G.Y. Koh, Development 138 (2011) 5027e5037.[35] C. Christodoulides, C. Lagathu, J.K. Sethi, A. Vidal-Puig, Trends Endocrinol.

Metab. TEM 20 (2009) 16e24.[36] T. Khan, E.S. Muise, P. Iyengar, Z.V. Wang, M. Chandalia, N. Abate, B.B. Zhang,

P. Bonaldo, S. Chua, P.E. Scherer, Mol. Cell. Biol. 29 (2009) 1575e1591.[37] Y. Shen, Y. Zhao, L. Yuan, W. Yi, R. Zhao, Q. Yi, T. Yong, Acta Histochem. 116

(2014) 158e166.[38] M.K. Vaicik, J. Thyboll Kortesmaa, S. Moverare-Skrtic, J. Kortesmaa,

R. Soininen, G. Bergstrom, C. Ohlsson, L.Y. Chong, B. Rozell, M. Emont,R.N. Cohen, E.M. Brey, K. Tryggvason, PLoS One 9 (2014) e109854.

[39] H. Gao, N. Mejhert, J.A. Fretz, E. Arner, S. Lorente-Cebrian, A. Ehrlund,K. Dahlman-Wright, X. Gong, S. Stromblad, I. Douagi, J. Laurencikiene,I. Dahlman, C.O. Daub, M. Ryden, M.C. Horowitz, P. Arner, Cell. Metab. 19(2014) 981e992.

[40] C. Lagathu, C. Christodoulides, S. Virtue, W.P. Cawthorn, C. Franzin,W.A. Kimber, E.D. Nora, M. Campbell, G. Medina-Gomez, B.N. Cheyette,A.J. Vidal-Puig, J.K. Sethi, Diabetes 58 (2009) 609e619.

[41] T.M. Williams, M.P. Lisanti, Genome Biol. 5 (2004) 214.[42] A.G. Ostermeyer, J.M. Paci, Y. Zeng, D.M. Lublin, S. Munro, D.A. Brown, J. Cell.

Biol. 152 (2001) 1071e1078.[43] S. Perez-Diaz, L.A. Johnson, R.M. DeKroon, J.M. Moreno-Navarrete, O. Alzate,

J.M. Fernandez-Real, N. Maeda, J.M. Arbones-Mainar, FASEB J. Off Publ. Fed.Am. Soc. Exp. Biol. 28 (2014) 3769e3779.

rkers for healthy adipose tissue, Biochimie (2015), http://dx.doi.org/