Cell Metabolism

Article

Human but Not Mouse AdipogenesisIs Critically Dependent on LMO3Josefine Lindroos,1 Julia Husa,1 GerfriedMitterer,1 Arvand Haschemi,1 Sabine Rauscher,1 Robert Haas,1 Marion Groger,1

Robert Loewe,2 Norbert Kohrgruber,1 Klaus F. Schrogendorfer,3 Gerhard Prager,3 Harald Beck,3 J. Andrew Pospisilik,5

Maximilian Zeyda,4 Thomas M. Stulnig,4 Wolfgang Patsch,6 Oswald Wagner,1 Harald Esterbauer,1,* and Martin Bilban1,*1Department of Laboratory Medicine2Department of Dermatology3Department of Surgery, Division of Plastic and Reconstructive Surgery4Christian Doppler Laboratory for Cardio-Metabolic Immunotherapy at the Department of Internal Medicine III

Medical University of Vienna, 1090 Vienna, Austria5Max Planck Institute of Immunobiology and Epigenetics, 79108 Freiburg, Germany6Institute of Pharmacology and Toxicology, Paracelsus Medical University, 5020 Salzburg, Austria

*Correspondence: harald.esterbauer@meduniwien.ac.at (H.E.), martin.bilban@meduniwien.ac.at (M.B.)

http://dx.doi.org/10.1016/j.cmet.2013.05.020This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use,

distribution, and reproduction in any medium, provided the original author and source are credited.

SUMMARY

Increased visceral fat is associated with a high risk ofdiabetes and metabolic syndrome and is in partcaused by excessive glucocorticoids (GCs). How-ever, the molecular mechanisms remain undefined.We now identify the GC-dependent gene LIM domainonly 3 (LMO3) as being selectively upregulated in adepot-specific manner in human obese visceral adi-pose tissue, localizing primarily in the adipocyte frac-tion. Visceral LMO3 levels were tightly correlatedwith expression of 11b-hydroxysteroid dehydroge-nase type-1 (HSD11B1), the enzyme responsible forlocal activation of GCs. In early human adipose stro-mal cell differentiation, GCs induced LMO3 via theGC receptor and a positive feedback mechanisminvolving 11bHSD1. No such induction was observedin murine adipogenesis. LMO3 overexpression pro-moted, while silencing of LMO3 suppressed, adipo-genesis via regulation of the proadipogenic PPARgaxis. These results establish LMO3 as a regulator ofhuman adipogenesis and could contribute a mecha-nism resulting in visceral-fat accumulation in obesitydue to excess glucocorticoids.

INTRODUCTION

The study of human white fat-cell development and distribution

has become an important issue in the past decades due to the

immense prevalence of obesity and related disease. High risk

of developing metabolic disease is associated with amplified

visceral (VI) obesity, whereas obese subcutaneous (SC) adipose

tissue (SAT) presents a smaller or no risk andmay even beprotec-

tive (Gabriely et al., 2002; Jensen, 2008; Tran et al., 2008). These

differences in contribution to disease and function could be

caused by regional variations of replication and adipogenic

potential. Depot-specific fat mass expansion mechanisms may

62 Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors

facilitate drug development targeting particular fat depots. Fat-

depot differences in human adipogenic potential (SC > VI [Tchko-

nia et al., 2005; Tchkonia et al., 2013; Tchoukalova et al., 2010])

have been attributed to depot-specific intrinsic gene expression

signatures (Gesta et al., 2006; Macotela et al., 2012; Perrini

et al., 2013; Tchkonia et al., 2007) such as higher levels of PPARg

and C/EBPa in differentiating SC human adipose stromal cells

(hASCs) and their superior response to troglitazones (TZDs)

(Hauner et al., 1988; Tchkonia et al., 2002). Other modulators

could be extrinsic factors such as glucocorticoids (GCs), which

areknown topotentiatehumanadipogenesis (Morton, 2010;Tom-

linson et al., 2006; Wiper-Bergeron et al., 2007). Clinically, GCs

are widely used as immunosuppressants directly regulating

transcription via the glucocorticoid receptor (GR). The enzyme

11-b-hydroxysteroid dehydrogenase type 1 (11bHSD1) plays a

crucial role in determining intracellular (prereceptor level) GC

levels by regenerating active GCs (cortisol) from inactive metabo-

lites (cortisone) and is highly expressed in visceral adipose tissue

(VAT) (Bujalska et al., 2008; Morton, 2010; Walker and Andrew,

2006). In humans, the adipogenic-enhancing properties of GCs

are most obvious in the truncal obesity of Cushing’s syndrome

(Arnaldi et al., 2010), as well as in patients on systemic immuno-

suppressive corticosteroid treatment (McDonough et al., 2008).

In common for these individuals are hypertension,VATexpansion,

and insulin resistance (Bjorntorp and Rosmond, 2000).

In this study, we investigated the role of GCs in human

adipocyte differentiation by analyzing the transcriptome of

GC-induced primary hASCs. This approach identified LMO3, a

member of the LIM-only proteins (LMOs), known to be involved

in cell-fate determination and neurogenesis (Dawid et al., 1998;

Zheng and Zhao, 2007), as a critical GC-responsive proadipo-

genic regulator. Further, LMO3 was among the earliest factors

induced in the course of human but not mouse white adipocyte

differentiation. We demonstrate that LMO3 exerts its activity at

the interface between GC action and peroxisome proliferator-

activated receptor g (PPARg). Importantly, LMO3 was upregu-

lated in VAT (as compared to SAT) in obese humans and tightly

correlated with 11b-hydroxysteroid dehydrogenase type-1

(HSD11B1) expression. These findings present LMO3 as a

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

proadipogenic factor that is (1) driven by GC action, (2) showing

high expression levels in human VAT, and (3) critical for human

adipocyte differentiation. Thus, LMO3 could provide an attrac-

tive target to interfere with human adipocyte differentiation in a

therapeutically relevant manner.

RESULTS

LMO3 Is A Target and Amplifier of GC Actionin Human VATOne of the genes emerging from our screen of the hASC tran-

scriptome in response to Dexamethasone (Dex; full target list

in Table S1, available online) that fulfilled our selection criteria,

(1) high adipocyte expression and (2) lack of known function in

adipogenesis, was the adaptor protein LMO3 (Figure 1A, verified

by quantitative PCR in Figure S1A). Several natural as well as

synthetic GR ligands, including Dex, hydrocortisone (HC), corti-

costerone (CC), prednisone (dehydrocortisone) (PRD), and the

fluorinated steroids clobetazol (CBTZ) and fluticazone (FLTZ),

all potently induced LMO3 messenger RNA (mRNA), an effect

that was blunted by the GR antagonist RU486, suggesting a

role of the GR for LMO3 induction (Figures 1B and S1B). We

further silenced the GR to confirm our results utilizing RU486.

Transfection of hASCs with a GR-specific small interfering RNA

(siRNA), or siGR, resulted in efficient silencing of GR mRNA

and protein (Figures 1C and 1D), in contrast to control

siRNA (siCtrl). Importantly, siCtrl-transfected hASCs displayed

a robust induction of LMO3 mRNA expression upon Dex treat-

ment, and no such induction was observed upon GR silencing

(Figure 1E). To further determine whether GCs upregulate

LMO3 via the GR, we performed transient transfection studies

with a luciferase reporter construct of the LMO3 promoter. GR

cotransfection resulted in an approximately 2.5-fold activation

of LMO3 promoter luciferase activity, further enhanced upon

treatment with GR ligand Dex. Importantly, LMO3 promoter

activity was blocked when 293FT cells were cotreated with

RU486 (Figure 1F). Of note, Dex failed to induce Lmo3 expres-

sion in murine adipose stromal cells (mASC) and 3T3-L1 preadi-

pocytes (Figure S1C).

The promoter regions of GR target genes typically contain one

or more GC response elements (GREs), defined by the canonical

sequence AGAACAnnnTGTTCT. Many variations of the

consensus sequence are possible, including GRE half sites,

which are sufficient to generate specific GR binding (van der

Laan and Meijer, 2008). Bioinformatic analysis of the proximal

promoter region of human LMO3 revealed two putative GRE

half sites, defined by the sequence �930TGTTTC�924 (GRE1)

and �742AAAACA�736 (GRE2) (Figure 1G). Deletion constructs

were created to identify which potential GR-binding sites were

involved in transactivation from the LMO3 promoter reporter

construct. Loss of both GRE1 and GRE2 resulted in diminished

transactivation, while constructs containing solely GRE2 or

both GREs showed potent transactivation (Figure 1G). Dex failed

to induce Lmo3 expression in mouse-derived adipocytes and

bioinformatic analysis revealed lack of GRE1 conservation in

the mouse Lmo3 promoter. Therefore, we verified the lack of

Lmo3 expression by exchanging the human with the noncon-

served murine GRE1 site (Figure 1G). Again, Dex significantly

increased luciferase reporter activity of the original human pro-

moter plasmid (pLMO3-Luc-Hs.), an effect markedly blunted in

the luciferase reporter plasmid harboring the murine GRE1 site

(pLMO3-Luc-Hs.>Mm.; Figure 1H).

As endogenous GCs need to be activated by 11bHSD1, we

tested whether LMO3 induction is dependent on this mechanism

by blocking 11bHSD1 activity with the pharmacologic inhibitor

carbenoxolone. This not only prevented PRD-induced LMO3

expression, but also reduced differentiation into mature adipo-

cytes (Figures 1I and S1D). To elucidate whether LMO3 feeds

back into HSD11B1 expression levels, we silenced LMO3

in hASCs. Significantly reduced HSD11B1 expression was

observed (Figure 1J), suggesting a positive functional correlation

between LMO3 and 11bHSD1 activity, thus comprising partially

the permissive role of GCs for adipogenesis. To identify compo-

nents of the adipogenic cocktail able to induce LMO3 expres-

sion, we assessed the individual components during the early

stages of differentiation (24–48 hr). Among the individual compo-

nents, solely GR ligands Dex and HC, significantly induced

LMO3 in both hASCs and Simpson Golabi Behmel syndrome

(SGBS) cells, that are biochemically and functionally similar to

human adipocytes (Wabitsch et al., 2001) (Figures 1K and

S1E). Again, cotreatment with RU486 potently blocked LMO3

expression in hASCs (Figure 1K).

Interestingly, paired SAT and VAT samples of 55 obese study

participants (Todoric et al., 2011) (Table S2) revealed an

increased expression of LMO3 in VAT compared to SAT, a

finding in line with our observation that GCs induce LMO3 in

human fat cells (Figure 1L). To consider the possibility that mac-

rophages present in VAT of obese individuals could account for

increased VAT LMO3 expression, we measured the proportional

contribution fromdifferent cell types present in VAT. Analysis of a

previously published cohort of subfractionated VAT obtained by

cell sorting (Zeyda et al., 2007) confirmed the highest LMO3

mRNA expression in mature adipocytes compared with other

cell types including macrophages (Figure 1M). To account for

LMO3 expression arising from CD68-expressing adipose tissue

macrophages, we repeated our human paired SAT and VAT

comparison on a subset of obese nondiabetic study participants

that displayed adipose CD68 mRNA levels comparable to sam-

ples obtained from lean control subjects (Table S3). Importantly,

this approach further validated our finding obtained from the

entire study cohort and clearly showed that VAT displays sub-

stantial higher LMO3 levels than SAT (Figure S1F). Furthermore,

no correlation between LMO3 andCD68mRNA levels was found

in the ‘‘obese CD68LOW’’ subcohort (Figure S1G). In contrast to

cell culture models of adipocyte differentiation in which GCs

are added at the beginning of differentiation, white adipose

tissue (WAT) 11bHSD1 activity in vivo provides continuous

exposure of GCs. Thus, we investigated whether LMO3 and

HSD11B1 mRNA expression correlate in vivo in WAT and found

a strong positive correlation in VAT, but not SAT, in obese sub-

jects (Figure 1N), irrespective of CD68 expression (Figure S1H

and Table S3). To explore whether LMO3 functionally affects

VAT upon GC treatment, we performed gene expression

profiling on LMO3-silenced hASCs isolated from matched SAT

and VAT from several donors (Figures 1O and S1J–S1L and

Table S4). This analysis revealed (1) higher basal LMO3

levels in VAT- as compared with SAT-derived preadipocytes

(Figure S1I), (2) impaired HC-mediated gene induction in

Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors 63

(legend on next page)

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

64 Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

LMO3-siRNA treated VAT cells affecting 21% (181 genes) of all

HC-responsive genes (Figure S1J), (3) importantly, no differ-

ences in SAT-derived hASCs upon silencing of LMO3 (Figure 1O

left versus right panel), and (4) LMO3-dependent and VAT-spe-

cific enrichment of genes regulating cell differentiation, commu-

nication, and adhesion, as well as signal transduction. Of note,

several of these LMO3-enriched genes are well known GC tar-

gets, such as inhibitor of growth family, member 2 (ING2),

TBC1 domain family, member 2B (TBC1D2B), tumor necrosis

factor (ligand) superfamily, member 10 (TNFSF10), C-X-C motif

chemokine 5 (CXCL5), and ecto nucleotide pyrophosphatase/

phosphodiesterase 2 (ENPP2) (Bujalska et al., 2006; Lu et al.,

2007; Viguerie et al., 2012) (Figures S1K and S1L). These data

suggest that the (ligand-bound) GR is critical for GC-mediated

LMO3 expression. Further, LMO3 is a specific target and

amplifier of GC action in human adipocytes displaying a VI fat

pattern, i.e., high levels of VI expression and close linkage with

HSD11B1 levels. Thus, LMO3 is modulating GC-triggered

responses of human (pre)adipocytes in a depot-, i.e., VAT-,

specific manner.

LMO3 Is Induced in Human AdipogenesisIn order to better understand the biological function of LMO3 in

adipose tissue/adipocytes, we examined whether LMO3 is regu-

lated during fat-cell formation in vivo and in vitro. DNAmicroarray

analysis of hASCs treated with an adipogenic cocktail revealed

strong induction of LMO3, whereas other LMO family members

were unaffected (Figure 2A). The LMO3 expression pattern

was determined in hASCs isolated from SAT, but also holds

true for VAT-derived hASCs and for SGBS preadipocytes,

when induced to differentiate into mature adipocytes (Figures

2B and S2A). Furthermore, mature adipocyte markers ADIPOQ,

LPL, PLIN,CD36, and FABP4 and the transcriptional master reg-

ulators PPARG and CEBPA confirmed successful generation of

mature adipocytes (Figures 2B, S2A, and S2B). Adipose tissue

LMO3 mRNA expression was among the top 25% of all human

tissues examined (Figure 2C). Importantly, LMO3 protein levels

could be detected simultaneously with the master regulators

PPARg andCEBPa in hASCs induced to differentiate intomature

adipocytes (Figure 2D). Fractionation of WAT allowed us to iden-

tify mature adipocytes as the dominant site of LMO3mRNA and

Figure 1. LMO3 Expression Is Regulated by the GC Receptor

(A) hASCs were treated with Dex and mRNA isolated on indicated time points. F

(B) LMO3 expression 24 hr after addition of GCs to hASCs in growth medium ±R

(C and D) mRNA (C) and protein (D) of GR in transfected hASCs treated for 24 h

(E) LMO3 mRNA in transfected hASCs with Dex treatment for 0 and 24 hr.

(F–H) LMO3 promoter analysis. 293FT cells were cotransfected with pcDNA and

promoter luciferase reporter plasmid (‘‘LUC’’; promoter construct shown above) (F

GRE1 site (pLMO3-Luc-Hs.>Mm.) (H). Twenty-four hours after transfection, cells

point mutations representing the murine sequence. Deletion constructs are show

(I) LMO3 mRNA expression in hASCs treated for 24 hr with vehicle, carbenoxolo

(J) HSD11B1 mRNA during the indicated days of differentiation in transfected hA

(K) LMO3 mRNA in hASCs after adding individual components of the adipogenic

(L) Depot-specific expression of LMO3 mRNA in paired obese human SAT and V

(M) Human LMO3 mRNA expression in human VAT fractions normalized to Ubiq

(N) Linear regression analysis between LMO3 and HSD11B1 mRNA expression in

(O) RNA profiling of transfected hASCs derived from matched human SAT and VA

normalized mean fluorescence pooled from n = 3 for each treatment group. A su

All error bars represent means ± SEM. p values: ns, not significant; *p < 0.05, **p <

Table S4, Table S5, and Table S6.

protein expression within WAT (Figures 2E and 2F). Thus, we

show in patient fat biopsies and in two well-established in vitro

models that LMO3 expression increases throughout adipogene-

sis. To determine subcellular LMO3 protein expression during

adipogenesis, we isolated cytoplasmic and nuclear protein

extracts throughout differentiation of hASCs into mature adipo-

cytes. We demonstrate a strong increase in LMO3 expression

from day 3 to day 7 that was detectable in cytoplasmic but not

nuclear fractions (Figure 2G). Restriction of LMO3 expression

to the cytoplasm was further corroborated by confocal immuno-

fluorescence analysis of hASCs undergoing differentiation (Fig-

ures 2H and S2C) showing no LMO3 signal in preadipocytes

(day 0) but strong cytoplasmic expression in perilipin (PLIN)-

positive adipocytes (day 6). Occasionally, WAT-resident CD68-

positive macrophages also stained positive for LMO3 (Figure 2I

and S2D). Perhaps most intriguing, LMO3 induction was specific

for human adipogenesis, asmurine adipocyte differentiation was

not accompanied by enhanced Lmo3 expression. Lack of Lmo3

expression was confirmed in (1) WAT of a mouse tissue library

(Figure S2E), (2) the murine 3T3-L1 adipocyte cell model

throughout differentiation (Figures 2J and S2F), and (3) differen-

tiating murine primary preadipocytes (Figure 2K), as well as in (4)

ASCs and mature adipocytes isolated from chow-fed or high-fat

diet (HFD)-challenged mice (Figures 2L and S2G), (5) distinct SC

and VI WAT depots and BAT of mice fed a low-fat diet (LFD) or

HFD (Figures 2M and S2H), and (6) WAT obtained from a genetic

obesity mouse model (db/db) on both LFD and HFD (Figures 2N

and S2I). Lmo3 expression was highest in the brain and therefore

was used as a positive control throughout analysis with murine

samples (Figures 2J–2N and S2G–S2I). Thus, LMO3 is a marker

of human adipogenesis not applicable for mice.

LMO3 Promotes AdipogenesisTo clarify the role of endogenous LMO3 in adipocyte differentia-

tion, we silenced the expression of LMO3 in subconfluent hASCs

using siRNA and induced differentiation. To reduce the risk of

potential off-target effects, we applied two different siRNA oligo-

nucleotides (referred to as siLMO3 #1 and siLMO3 #2) targeting

different LMO3 exons. Figures 3A and 3B show efficient LMO3

mRNA and protein knockdown in hASCs. Cells with reduced

LMO3 demonstrated diminished adipogenic potential, including

old change is compared to day 0 of Dex treatment.

U486. LMO3 in the absence of RU486 in growth medium was set to 1 (n = 3).

r with Dex.

/or GC receptor expression plasmid (pGR), as well as human full-length LMO3

), various deletion constructs (G, right), or LMO3 construct featuring the murine

were treated with DMSO, Dex, or RU486. Underlined letters, GRE half sites; *,

n below homology plot (G, left).

ne, or PRD in growth medium.

SCs.

cocktail or full mix (FM) in growth medium for 24 hr. (n = 3).

AT biopsies (n = 55). Horizontal bars indicate the mean.

uitin-C mRNA (n = 4).

obese human VAT. p values were obtained from regression analysis (n = 45).

T treated with DMSO or HC for 24 hr. Genes are represented as lines and are

mmary is shown in Table S4.

0.001, and ***p < 0.0001. See also Figure S1 and Table S1, Table S2, Table S3,

Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors 65

Figure 2. Human, Not Mouse, LMO3 Is Induced during Adipocyte Differentiation

(A) LMO1-4 mRNA expression based on DNA microarray expression profiling. Fold change compared to day 0 of differentiation is shown.

(B) mRNA was determined in differentiating hASCs isolated from SAT or VAT as the mean fold change ± SEM (n = 3).

(C) LMO3 mRNA tissue distribution in humans. Each tissue is pooled from three donors.

(legend continued on next page)

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

66 Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

less lipid accumulation (Figures 3C and 3D) and diminished

expression of adipocyte marker genes, including FABP4,

PLIN1, and LPL (Figures 3B [perilipin blot] and 3E). Importantly,

neither cell viability nor cell proliferation varied between siCtrl-

and siLMO3-treated hASCs under the experimental conditions

for adipocyte differentiation (Figures S3A and B). We additionally

verified the hASC requirement of GCs to potentiate the adipo-

genic program by excluding GCs. This resulted in almost com-

plete lack of mature adipocyte formation also in LMO3-silenced

hASCs, as expected (Figure S3C), demonstrating that GCs are

indeed necessary to induce adipogenesis in human preadipo-

cytes. To study the role of LMO3 in regulation of adipogenesis

in vivo, we subdermally administered control- or LMO3-silenced

hASCs into severe combined immunodeficient (SCID) mice. The

transplants were collected and histologically examined after

6 weeks. We confirmed the human origin of the collected cells

by immunofluorescent stainingwithMAB1281, a human-specific

nuclear antibody that does not stain mouse nuclei (Figures 3F

and S3D). Mature adipocytes also stained positive for the adipo-

cyte marker perilipin. We observed a significantly higher propor-

tion of hASC-derived mature adipocytes in tissues collected

from animals that were transplanted with siCtrl-treated hASCs

as compared to mice transplanted with siLMO3-transfected

hASCs (Figure 3G), underlining strong proadipogenic effects of

LMO3 also in vivo.

Next, we employed LMO3 gain-of-function studies to mini-

mize the risk that the siLMO3-triggered reduction of adipocyte

differentiation is an unspecific side effect due to our experi-

mental manipulations interfering with a highly coordinated, and

thus sensitive, cellular program. LMO3mRNA and protein levels

were significantly increased in LMO3 cells transfected with a

LMO3 expression plasmid relative to cells transfected with a

control plasmid (Figures 3H and 3I, pCtrl versus pLMO3-V5). In

accordance with our loss-of-function data elucidating LMO3 as

a proadipogenic mediator, overexpression of LMO3 in these

cells significantly enhanced adipogenesis, shown by increased

oil red O staining of neutral lipids (Figures 3J and 3K) and signif-

icant overexpression of the adipocyte markers FABP4, LPL, and

PLIN1 (Figure 3L). As LMO3 overexpression promotes adipo-

genesis and expression of genes facilitating lipid accumulation

in hASCs, we investigated whether visceral LMO3 expression

is linked to body mass index (BMI), waist circumference, or

HOMA-IR in our nondiabetic lean and obese patient cohorts

(Tables S5 and S6). Interestingly, and despite the fact that

HSD11B1 levels were robustly increased in the VAT of obese

study subjects displaying high HOMA-IR (Table S5), no such dif-

ference could be found for LMO3 expression.

Of note, siRNA mediated knockdown of mouse Lmo3 failed

to interfere with adipogenesis in 3T3-L1 cells (Figures S3E

(D) Protein expression throughout differentiation in SAT-isolated hASCs.

(E and F) mRNA and (F) protein (F) expression in human SAT cell fractions (n = 9

(G) Protein expression in nuclear and cytosolic fractions in differentiating SAT-de

(H and I) Images are representative of multiple donors. Scale bars represent 20 mm

(I) are shown.

(J–N) Determination of protein expression in differentiating murine 3T3-L1s (J),

gonadal pads (n = 5) (K and I), the indicated cell subpopulations from C57Bl/6 p

perigonadal pads from db/db mice and littermate controls (db/+) (n = 3–4) (N).

All error bars represent the means ± SEM. p values: ns, not significant; **p < 0.0

and S3F). Interestingly, when Lmo3 was overexpressed, it

exerted the phenotype observed in hASCs, i.e., enhanced adipo-

genesis (Figures S3G and S3H), implying that murine cells can

utilize Lmo3, but that due to lack of conservation in the GRE1

site, it is not inducible. Thus, LMO3 is a prerequisite to unveil

the full adipogenic potential of human preadipocytes.

LMO3 Boosts a Proadipogenic PPARg ProgramWe next sought to investigate the mechanism by which LMO3

promotes adipogenesis. Thus, we profiled genome-wide

expression changes that occur in response to the adipogenic

cocktail, comparing patterns between siCtrl- or siLMO3-treated

human preadipocytes on day 6 of differentiation, integrating both

primary and secondary effects of LMO3. Figure 4A shows 1,892

genes from adipogenesis-induced preadipocytes that displayed

at least a 1.5-fold expression change relative to day 0 and were

therefore selected as adipogenesis-induced genes. Approxi-

mately 4.6% of the adipogenic gene signature was affected by

LMO3 knockdownwith two independent siRNA oligonucleotides

targeting LMO3 mRNA (Figure 4B). Hierarchical clustering parti-

tioned the 1,892-adipogenesis-induced genes into LMO3-inde-

pendent clusters 1 and 2 or LMO3-dependent clusters 3 and 4

(Figure 4C and Table S7). Pathway enrichment analysis of

LMO3-dependent cluster 3 revealed a highly significant enrich-

ment for PPAR signaling (Figure 4D). In agreement with the

pathway enrichment results, inspection of cluster 3 showed

that silencing of LMO3 (siLMO3) diminished expression levels

of several known PPARg target genes in hASCs (Figure 4E).

Therefore, we tested whether LMO3 modulates PPARg expres-

sion and/or activity. Western blotting of siCtrl- and siLMO3-

treated hASCs suggested a slight but nonsignificant impact on

PPARg protein expression (Figure 4F). We further investigated

whether PPARg activity is needed for the proadipogenic effects

of endogenous LMO3 on lipid accumulation. We silenced

PPARG (siPPARg) in hASCs overexpressing LMO3 (pLMO3-V5)

and evaluated lipid accumulation by oil red O staining in differen-

tiating hASCs 8 days later (Figures 4G–4I). As shown above

(Figures 3J and 3K), overexpression of LMO3 increased lipid

accumulation (Figures 4H and 4I, left panels). Of note, silencing

of the adipogenic master regulator PPARG (siPPARg) abolished

the proadipogenic effect of LMO3, suggesting that LMO3 acts

upstream of PPARg (Figures 4H and 4I, right panels). We next

determined whether LMO3 is able to modulate the transcrip-

tional activity of PPARg. We performed transfection assays in

293FT or 3T3-L1 cells with a reporter driven by isolated PPAR

response elements (PPREs). As expected, cotransfected

PPARG resulted in an increase in luciferase activity, in part

because of the ligand-independent activation function in its

amino terminus. Troglitazone (TZD) treatment further enhanced

).

rived hASCs.

. Stainings in differentiating SAT-derived hASCs (H) and human VAT sections

primary murine adipose stromal cells (mASCs) isolated from C57BL/6J peri-

erigonadal pads (n = 5) (L), WAT depots from C57BL/6J mice (n = 5) (M), and

01 and ***p < 0.0001. See also Figure S2.

Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors 67

Figure 3. RNA-Interference-Mediated Knockdown of LMO3 Suppresses, whereas Overexpression of LMO3 Promotes, Adipogenesis

(A, D, E, and G) Comparisons of control (siCtrl)- versus LMO3-silenced (siLMO3)-transfected cells.

(A and B) mRNA (A) and protein (B) verification of LMO3 silencing on the indicated days (blots day 6) of differentiation in transfected hASCs.

(C) Mature adipocytes (differentiation day 10) stained with oil red O. Microscopic views, magnifications 103.

(D) Quantification of oil red O staining in (C) (n = 3).

(E) RT-PCR analysis in transfected hASCs. (n = 3).

(F) Representative immunofluorescent staining of xenotransplanted SCIDmice. From left to right: top, bright-fieldmorphology of the transplant sections (the scale

bar represents 50 mm), zoomed bright-field image (the scale bar represents 20 mm), and immunofluorescent merge; bottom, DAPI (blue), MAB1281 (gray with

arrows), and perilipin (red) (immunofluorescent panels, the scale bar represents 20 mM).(legend continued on next page)

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

68 Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

PPARG activity in a dose-dependent manner. Cotransfection

with increasing amounts of LMO3 expression plasmid increased

PPARG activity, which was further enhanced in the presence of

TZD (Figure 4J). Thus, LMO3 drives adipogenesis through

increasing PPARg tone.

A critical step required during adipogenesis is the downregu-

lation of mitogen-activated protein kinase extracellular signal-

regulated kinases (MAPK-ERKs) mediated phosphorylation at

serine 112 (S112) in the N-terminal region of PPARg, which

blocks PPARg to activate the full proadipogenic gene program

(Adams et al., 1997; Camp and Tafuri, 1997; Hu et al., 1996;

Shao et al., 1998). Interestingly, we found that loss of LMO3

resulted in increased serum-induced ERK1/2 phosphorylation

(p-p44/42 MAPK; Figure 4K), further supporting the observation

that LMO3 acts upstream of PPARg. To test whether LMO3

could directly inhibit ERK-dependent signaling, we performed

transient transfection assays using reporter plasmids that read

out ERK-dependent activation of the transcription factor ELK1.

Cotransfection of a LMO3 expression vector diminished the abil-

ity of EGF to activate the ELK1 reporter (Figure 4L). Importantly,

compared with control cells, the LMO3 knockdown cells not only

showed increased ERK1/2 phosphorylation but also 1.7-fold

increased phosphorylation levels of endogenous PPARg at

S112 (Figure S3I). Next, we addressed whether p-S112 is

involved in LMO3-mediated effects by making use of a mutant

PPARg-S112A, which renders PPARg refractory to p-S112-

mediated inactivation (Camp and Tafuri, 1997; Hu et al., 1996;

Shao et al., 1998). We initially tested whether LMO3 increases

TZD-mediated PPARg activity via S112 when a reporter driven

by isolated PPREs (AOx-TK) was used. As reported, mutant

PPARg-S112A displayed increased PPARg activity (Adams

et al., 1997). Again, LMO3 cotransfection boosted PPARg pro-

moter activity. However, and in sharp contrast, LMO3 was

unable to further increase the transcriptional activity of mutated

PPARg-S112A (Figure 4M). Similarly, cotransfection with an acti-

vated allele of MEK, the upstream kinase of ERK1/2, prevented

the stimulating effect of LMO3 on PPARg activity (Figure 4M).

In line with our previous results, LMO3 knockdown reduced adi-

pogenesis and the expression of PPARg target genes in hASCs

expressing wild-type PPARg (Figures 4N). However, and impor-

tantly, these LMO3-dependent effects were lost in hASCs trans-

fected with the mutated form of S112 PPARg (PPARg-S112A) or

the presence of a constitutively active MEK (Figure 4N). Thus,

interference with MAPK-ERK phosphorylation of PPARg is one

possiblemechanism bywhich LMO3 regulates human adipocyte

differentiation.

DISCUSSION

Obesity is associated with many metabolic consequences,

where VI fat accumulation produces a greater risk of diabetes,

(G) Quantification of xenotransplanted SCID mice stainings.

(H, K, and L) Comparisons of control (pCtrl)- or LMO3 (pLMO3-V5)-transfected c

(H and I) mRNA (H) and protein (I) verification of LMO3 overexpressing transfecte

(J) Oil red O stain of overexpressing hASCs (differentiation day 10). Shown are m

(K) Quantification of oil red O stain in (J) (n = 3).

(L) RT-PCR analysis in pCtrl or pLMO3-V5 transfected hASCs. (n = 3).

All error bars represent the means ± SEM. p values: ns, not significant; *p < 0.05

dyslipidemia, and accelerated atherosclerosis (Kissebah et al.,

1982; Wajchenberg, 2000). In this study, we aimed to identify

GC target genes involved in the differentiation of human

adipocytes, on the basis of (1) a hitherto unknown function

in adipocyte biology, (2) a robust induction in human adipo-

cyte differentiation models, and (3) the potential to act

upstream of the adipogenic master regulator PPARg. Using

these criteria, we identified among the top-most regulated

genes LMO3.

LMO3 promoter studies, GR silencing, GR antagonist RU486,

and several natural and synthetic GCs showed that LMO3 is a

direct GR target gene. Interestingly, we found that LMO3 is not

only induced by GCs and HSD11B1, but is also part of a positive

feedback loop enhancing GC action on adipocytes, a finding

paralleled by our data showing tightly correlated LMO3 and

HSD11B1 levels in human VAT but not SAT. Overexpression of

LMO3 in hASCs enhanced adipogenesis and was reflected by

enhanced adipogenic gene expression and enhanced lipid

accumulation. Knockdown of LMO3 in hASCs suppressed fat

differentiation both in vitro and in vivo. This collectively renders

LMO3 as an essential factor linking extrinsic factors (GCs)

with specific molecular mediators, resulting in progressed adi-

pogenesis. To better understand the role of LMO3 in human

depot-specific responses to GCs, we silenced LMO3 expression

in hASCs from matched SAT and VAT. Upon GC treatment, an

LMO3-dependent gene expression signature was observed

solely in VAT- but not SAT-derived preadipocytes, a finding

that may be related—at least in part—to the higher basal

LMO3 and HSD11B1 expression in VAT-derived preadipocytes.

Of note, both fat depots are responsive to the actions of LMO3

in vitro; however, in vivo measurements and SAT/VAT compari-

sons revealed a clear preference of VI preadipocytes for LMO3-

dependent GC action. Some of the LMO3-dependent genes

expressed in VI but not SC preadipocytes were ENPP2,

TNFSF10 or TBC1 domain family member 2B (TBC1D2B) poten-

tially having direct influence on (VI) fat-cell growth ormetabolism.

The lysophospholipase ENPP2 and its product, lysophosphati-

dic acid, have established effects on preadipocyte proliferation

and fat-tissue expansion, and its expression is enhanced in a

depot-specific manner in obese/insulin-resistant individuals

(Rancoule et al., 2012), whereas the secreted protein TNFSF10

regulates adipocyte metabolism through cleavage of PPARg

(Keuper et al., 2013). The GC-induced signaling nexus TBC1D2B

may enhance insulin signaling in a manner reported for its close

paralogues, TBC1D1, TBC1D3, and TBC1D4 (Pehmøller et al.,

2012; Wainszelbaum et al., 2012) in a LMO3-dependent manner

in VAT-, but not SAT-, derived preadipocytes, adding to adipose

depot-specific actions of insulin (Hazlehurst et al., 2013). We

now add LMO3 to the growing list of developmental regulators

controlling (depot-specific) adipocyte differentiation (Macotela

et al., 2012).

ells.

d hASCs on the indicated days.

icroscopic views, magnifications 103 (n = 3).

, **p < 0.001, and ***p < 0.0001. See also Figure S3.

Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors 69

Figure 4. LMO3 Promotes Adipogenesis via Increasing PPARg Tone

(A) Flow chart with experimental DNA microarray setup and gene selection process.

(B) Schematic pie chart of LMO3-dependent and -independent genes.

(C) Clusters of genes from the microarray screen. Red is up- and blue is downregulated genes.

(D) Pathway enrichment analysis shown as a Z score on LMO3-dependent genes (87 in total) with DAVID and EASE.

(E) Heatmap diagram of PPARg target genes within the LMO3-dependent gene signature. Experiments were performed in duplicates. *, independent validation

shown in Figure 3E. See Table S7 for LMO3-dependent target genes at day 6 of differentiation.

(legend continued on next page)

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

70 Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

Several observations led us to suggest that LMO3 enhances

adipogenesis via PPARg: (1) the most significant category

of genes suppressed after LMO3 silencing in differentiating

hASCswas PPARg target genes, (2) RNA-interference-mediated

silencing of PPARG during hASC differentiation abrogated

LMO3-enhanced lipid accumulation, and, most importantly, (3)

LMO3 overexpression enhanced PPARg transcriptional activity

in two cell models. These observations raised the question of

how cytoplasmic LMO3 enhances primarily nuclear PPARg

activity. One potential mechanism is the well studied inhibition

of PPARg activity by ERK1/2-mediated phosphorylation of

PPARg at serine 112 (Adams et al., 1997; Hu et al., 1996; Shao

et al., 1998). Interestingly, we observed that LMO3-deficient

cells displayed an increase in the amount of p-S112 PPARg

and p-ERK1/2, while LMO3-overexpressing cells display

reduced ERK1/2 pathway activity. Importantly, we were also

able to reverse the LMO3 knockdown-based phenotype by

PPARg -S112A transfection, a mutation that has been described

to render PPARg insensitive to pS112-inhibition (Adams et al.,

1997; Hu et al., 1996). Taken together, this supports our hypoth-

esis that LMO3 targeting of PPARg at serine 112 (via ERK1/2)

represents a major determinant altering adipocyte gene expres-

sion. Interestingly, a similar mechanism, i.e., lack of catalytic

activity and negative modulation of ERK1/2, has been reported

to underlie the proadipogenic effects of the cytoplasmic down-

stream of tyrosine kinase-1 (DOK1) (Hosooka et al., 2008).

Indeed, the LIM domains of LMO3 lack intrinsic catalytic (i.e.,

phosphatase) activity. However, as for DOK1, LIM proteins

mediate many biological processes acting as a docking site for

the assembly of multiprotein complexes (Kadrmas and Beckerle,

2004; Zheng and Zhao, 2007). Among others, potential LMO3

interaction partners include ERK activators MAPK kinases

(MAPKK/MEK) (Burgermeister et al., 2007), ERK1/2 themselves,

(Adams et al., 1997; Hu et al., 1996), phosphatases (Hinds et al.,

2011), and the signaling adaptor DOK1 (Hosooka et al., 2008).

The 11bHSD1/LMO3/ PPARg module provides differentiating

human (pre)adipocytes with a molecular switch, enabling the

cells to fine-tune their response to circulating GCs. 11bHSD1 is

expressed at high levels in VI fat depots (Bujalska et al., 1997,

2008; Morton, 2010; Walker and Andrew, 2006), and LMO3 is

also found at higher levels in VAT as compared to SAT. There-

fore, it is proposed that GCs drive human LMO3 expression

and thus its proadipogenic activities in a depot-specific, VI

manner. In such a scenario, low 11bHSD1 activities or its inhibi-

tion by pharmacologic targeting will result in (1) reduced VI LMO3

(F) PPARg protein expression on day 6 of differentiation in transfected hASCs. D

(G) Silenced PPARG mRNA verification in transfected hASCs at day 9 of differen

(H) Oil red O staining of cells treated as in (G), magnifications 103.

(I) Quantification of (H). pCtrl/siCtrl-transfected cells are set to 1 for comparison.

(J) LMO3 enhances PPARg activation of a luciferase reporter driven by minimal P

(bottom) cells stimulated for 24 hr with DMSO or TZD. pcDNA with DMSO, set to

(K) Transfected and day 6 differentiated hASCs were serum starved overnight fol

blots. The bottom panels show densitometric evaluation.

(L) Luciferase reporter activity analysis of GAL-ELK-1 constructs in 293FT cells.

(M) 293FT cells transfected with wild-type (WT) PPARg or S112A mutant were an

presence or absence of LMO3 (n = 3).

(N) mRNA of mature adipocyte markers in hASCs with WT PPARg transfected

cotransfected with mutated PPARg S112A plasmid (gray bars). RNA was isolate

All error bars represent the means ± SEM. p values: ns, not significant; *p < 0.05

levels, (2) reduced PPARg transcriptional activities, and, conse-

quently (3) reduced adipogenesis, specifically in VI fat.

An appealing question is why LMO3 can be averted in mouse

adipocytes andwhy there is no obvious functional consequence.

This is especially interesting, since ectopically expressed Lmo3

in murine 3T3-L1 cells enabled replication of the phenotype

observed in differentiated hASCs (i.e., enhanced adipogenesis).

As a direct consequence, Lmo3-dependent fine-tuning in mice

does not apply, not because mice cannot utilize Lmo3 to

enhance adipogenesis, but because the critical GC induction

site GRE1 is mutated in the mouse genome. Thus, LMO3 might

represent a mechanism by which—in contrast to mice—humans

can adapt and modulate the activity of the key adipogenic mas-

ter regulator PPARg. Although the basic molecular mechanisms

of fat-cell development are identical in rodents and humans—as

is the intact response of murine cells to reintroduced Lmo3—

many of the observed species-specific attributes likely stem

from when (and where) the products of the genes are made (Wil-

son et al., 2008) so that the timing of preadipocyte recruitment

and adipocyte differentiation is accessible to more subtle fine-

tuning mechanisms.

Interestingly, as opposed to HSD11B1, no correlations were

observed between LMO3 and BMI or HOMA-IR in our study

participants. However, the missing link of LMO3 expression

with our metabolic parameters needs to be interpreted with

caution since several circumstances may have blurred a poten-

tial relationship (specifically, considering observed HSD11B1

correlations with LMO3 in VAT of obese subjects; Table S6).

Among others, we cannot exclude the possibility that nonfat

cells in VAT may obscure the association of LMO3 with meta-

bolic parameters. Also, we cannot exclude that we missed the

best time point to collect our fat biopsies, i.e., it might have

been too late in the course of fat-cell recruitment and expan-

sion, especially in our obese study cohort. Further, GC circadian

rhythms may have masked pre-existing differences in LMO3

expression (Peckett et al., 2011). Further studies with highly

standardized measurements of circadian systemic and adipose

GC levels are needed to relate LMO3 with states of obesity and

diabetes.

Our current study (summarized in Figure 5) added LMO3 as a

proadipogenic protein and suggests that LMO3 modulates

human adipocyte differentiation by acting on PPARg, between

the early and late phase of adipocyte differentiation (Farmer,

2006). Our data also help to explain, at least in part, the long-

known but ill-defined effect of GCs on VAT. Finally, we propose

ensitometric evaluation is shown below. (n = 3.)

tiation.

PAR-responsive elements (3X-PPRE) (‘‘pAOx-TK’’) in 293FT (top) and 3T3-L1

1 for comparison. (n = 3.)

lowed by 1 hr 40% fetal bovine serum (FBS). The top panels are representative

The construct is depicted above.

alyzed for TZD-induced reporter activity using PPRE-luciferase (as in J) in the

with either siCtrl or siLMO3 oligo (white and black bars) and/or additionally

d from cells on day 8 of differentiation.

, **p < 0.001, and ***p < 0.0001. See also Figure S3 and Table S7.

Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors 71

Figure 5. LMO3 Is a Human Driver of Adipogenesis

Schematic model of the pathways controlling differentiation in hASCs.

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

that the preferential expression in VAT, the GC responsiveness,

and the functional location upstream of PPARg make LMO3 an

attractive target to interfere with human adipocyte differentiation

in a depot-specific, therapeutically relevant manner.

EXPERIMENTAL PROCEDURES

Human Samples and Clinical Parameters

Study subjects included 55 obese patients and seven nonobese controls

that underwent weight-reducing surgery or elective surgical procedures

such as cholecystectomy. Participants were included if they had fasting

plasma glucose levels <7.0 mmol/liter, no history of diabetes or use of

blood-glucose-lowering medications, no weight changes >3% during the

previous 2 months, and C-reactive protein (CRP) levels <20 mg/liter. All

study subjects provided informed consent, and study protocols were

approved by the local Ethics Committee. Tissue biopsies from SAT and

VAT, obtained during surgery, were stored at �80�C until further processing.

Plasma glucose, insulin, and CRP were determined as described (Todoric

et al., 2011).

Isolation of Preadipocytes

Human SAT was obtained from healthy individuals undergoing lipoaspiration.

A total of 47 donors (female, n = 34; male, n = 13) were used throughout the

study; Total age was 44.02 ± 14.7 years (female, 46.47 ± 14.2; male, 37.62 ±

14.5), and total mean BMI was 25.08 ± 4.3 (female, 24.87 ± 4.5; male,

24.36 ± 3.3). Matched SAT and VAT was obtained from three of the above

donors undergoing elective abdominal surgery. This study was approved by

the Medical University of Vienna’s ethics committee and the General Hospital

of Vienna (EK no. 1115/2010). All subjects gave written informed consent prior

to taking part in the study.

Mouse Studies

Mice were purchased from Charles River Laboratories. Male C57BL/6J,

BKS.Cg-Dock7m+/+ Leprdb/J diabetic (db/db) and nondiabetic (db/+) litter-

mates were used as detailed in the Supplemental Experimental Procedures.

Human and Murine Adipocyte Differentiation

Two-day-postconfluent ASCs were induced to differentiate for 10 to 13 days

with (the medium used is referred to as ‘‘full mix’’ in the text) Dulbecco’s modi-

72 Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors

fied Eagle’s medium (DMEM)/Ham’s F12, 10% FBS, 33 mM biotin, 17 mM pan-

tothenic acid, 1 nM triiodothyronine (T3), 870 nM human insulin, 5 mMTZD, and

1 mg/ml transferrin, and for the first 3 days 1 mM Dex and 500 mM isobutyl-

methylxanthine (IBMX) were included (all from Sigma). Two-day-postconfluent

3T3-L1 cells were differentiated with DMEM, 10% CS, 870 nM insulin, 1 mM

Dex, and 500 mM IBMX. On day 3 of differentiation, this medium was added

excluding IBMX and Dex for remaining differentiation. Additional compounds

used were 100 nM HC, 100 nM CC, 1 mM PRD, 5 mM CBTZ, 5 mM FLTZ,

100 mM carbenoxolone, and 1 mM RU486.

Gene Expression Profiling

Was performed as previously described (Bilban et al., 2009). For extended

information on GC target genes and LMO3 targets after GC stimulation, refer

to the Supplemental Experimental Procedures.

Real-Time PCR

Real-time PCR was performed as previously published (Todoric et al., 2011).

Primer sequences are listed in the Supplemental Experimental Procedures.

Adipose Tissue Fractionation

Human ASCs and MA were isolated as described above. ASCs were frac-

tioned by flow cytometry (FACSAria, BD Biosciences) as previously described

(Zeyda et al., 2007).

Luciferase Assays

Luciferase assayswere carried out as previously described (Bilban et al., 2009)

and as detailed in the Supplemental Experimental Procedures.

hASC Transfection

siRNA (100 nmol/liter) (listed in the Supplemental Experimental Procedures) (all

Invitrogen) or plasmids (1 mg) were delivered into hASCs (6 3 105) by Amaxa

nucleofection (Lonza Bioscience) according to manufacturer‘s recommenda-

tions. Cells were utilized 48–72 hr after transfection.

Western Blot Analyses

Western blot analyses were performed as described previously (Bilban et al.,

2009).

Confocal Immunofluorescence Microscopy

All immunofluorescence slides were mounted for imaging with confocal laser

scanning microscopy (LSM 700 Carl Zeiss) as detailed in the Supplemental

Experimental Procedures.

SCID Mouse Xenotransplant Model

All procedures were carried out in accordance with the Association for

Assessment and Accreditation of Laboratory Animal Care guidelines and

the Guide for the Care and Use of Laboratory Animals (US Department of

Health and Human Services, National Institutes of Health, publication no.

86–23). All experiments were approved by the ethics committee of the Med-

ical University of Vienna and by the Austrian government committee on

animal experimentation. For further information, see the Supplemental Exper-

imental Procedures.

Statistical Analysis

The significance of differences between means was assessed by two-tailed

Student’s t test or analysis of variance (ANOVA) with Bonferroni post test. Dif-

ferences between human adipose tissue depots were ascertained by ANOVA.

Correlations were tested by linear regression. Logarithmic transformations

were made if the equal variance and normality assumptions were rejected.

All measurements were adjusted for confounding effects as indicated. Error

bars are expressed as the mean ± SEM unless otherwise specified. p < 0.05

was considered significant.

ACCESSION NUMBERS

Data sets have been deposited in theGene Expression Omnibus (GEO) archive

as series GSE44636.

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

SUPPLEMENTAL INFORMATION

Supplemental Information includes Supplemental Experimental Procedures,

three figures, and seven tables and can be found with this article online at

http://dx.doi.org/10.1016/j.cmet.2013.05.020.

ACKNOWLEDGMENTS

This work was supported by the Austrian Science Fund (FWF, Project no.

W1205, CCHD PhD program). We thank Markus Jeitler for help with DNA

microarrays, Mike Mitchell for support with site-directed mutagenesis, and

Christoph J. Binder for fruitful discussions (all Medical University of Vienna).

We thank Mitchell Lazar (University of Pennsylvania) for kind provision of

PPARg-S112A expression plasmid.

Received: October 19, 2012

Revised: March 6, 2013

Accepted: May 17, 2013

Published: July 2, 2013

REFERENCES

Adams, M., Reginato, M.J., Shao, D., Lazar, M.A., and Chatterjee, V.K. (1997).

Transcriptional activation by peroxisome proliferator-activated receptor

gamma is inhibited by phosphorylation at a consensus mitogen-activated pro-

tein kinase site. J. Biol. Chem. 272, 5128–5132.

Arnaldi, G., Scandali, V.M., Trementino, L., Cardinaletti, M., Appolloni, G., and

Boscaro, M. (2010). Pathophysiology of dyslipidemia in Cushing’s syndrome.

Neuroendocrinology 92(Suppl 1 ), 86–90.

Bilban, M., Haslinger, P., Prast, J., Klinglmuller, F., Woelfel, T., Haider, S.,

Sachs, A., Otterbein, L.E., Desoye, G., Hiden, U., et al. (2009).

Identification of novel trophoblast invasion-related genes: heme oxygenase-1

controls motility via peroxisome proliferator-activated receptor gamma.

Endocrinology 150, 1000–1013.

Bjorntorp, P., and Rosmond, R. (2000). Obesity and cortisol. Nutrition 16,

924–936.

Bujalska, I.J., Kumar, S., and Stewart, P.M. (1997). Does central obesity reflect

‘‘Cushing’s disease of the omentum’’? Lancet 349, 1210–1213.

Bujalska, I.J., Quinkler, M., Tomlinson, J.W., Montague, C.T., Smith, D.M., and

Stewart, P.M. (2006). Expression profiling of 11beta-hydroxysteroid dehydro-

genase type-1 and glucocorticoid-target genes in subcutaneous and omental

human preadipocytes. J. Mol. Endocrinol. 37, 327–340.

Bujalska, I.J., Gathercole, L.L., Tomlinson, J.W., Darimont, C., Ermolieff, J.,

Fanjul, A.N., Rejto, P.A., and Stewart, P.M. (2008). A novel selective 11beta-

hydroxysteroid dehydrogenase type 1 inhibitor prevents human adipogenesis.

J. Endocrinol. 197, 297–307.

Burgermeister, E., Chuderland, D., Hanoch, T., Meyer, M., Liscovitch, M., and

Seger, R. (2007). Interaction with MEK causes nuclear export and downregu-

lation of peroxisome proliferator-activated receptor gamma.Mol. Cell. Biol. 27,

803–817.

Camp, H.S., and Tafuri, S.R. (1997). Regulation of peroxisome proliferator-

activated receptor gamma activity by mitogen-activated protein kinase.

J. Biol. Chem. 272, 10811–10816.

Dawid, I.B., Breen, J.J., and Toyama, R. (1998). LIM domains: multiple roles as

adapters and functional modifiers in protein interactions. Trends Genet. 14,

156–162.

Farmer, S.R. (2006). Transcriptional control of adipocyte formation. Cell

Metab. 4, 263–273.

Gabriely, I., Ma, X.H., Yang, X.M., Atzmon, G., Rajala, M.W., Berg, A.H.,

Scherer, P., Rossetti, L., and Barzilai, N. (2002). Removal of visceral fat pre-

vents insulin resistance and glucose intolerance of aging: an adipokine-medi-

ated process? Diabetes 51, 2951–2958.

Gesta, S., Bluher, M., Yamamoto, Y., Norris, A.W., Berndt, J., Kralisch, S.,

Boucher, J., Lewis, C., and Kahn, C.R. (2006). Evidence for a role of develop-

mental genes in the origin of obesity and body fat distribution. Proc. Natl. Acad.

Sci. USA 103, 6676–6681.

Hauner, H., Wabitsch, M., and Pfeiffer, E.F. (1988). Differentiation of adipocyte

precursor cells from obese and nonobese adult women and from different

adipose tissue sites. Horm. Metab. Res. Suppl. 19, 35–39.

Hazlehurst, J.M., Gathercole, L.L., Nasiri, M., Armstrong, M.J., Borrows, S.,

Yu, J., Wagenmakers, A.J., Stewart, P.M., and Tomlinson, J.W. (2013).

Glucocorticoids fail to cause insulin resistance in human subcutaneous

adipose tissue in vivo. J. Clin. Endocrinol. Metab. 98, 1631–1640.

Hinds, T.D., Jr., Stechschulte, L.A., Cash, H.A., Whisler, D., Banerjee, A.,

Yong, W., Khuder, S.S., Kaw, M.K., Shou, W., Najjar, S.M., and Sanchez,

E.R. (2011). Protein phosphatase 5 mediates lipid metabolism through recip-

rocal control of glucocorticoid receptor and peroxisome proliferator-activated

receptor-g (PPARg). J. Biol. Chem. 286, 42911–42922.

Hosooka, T., Noguchi, T., Kotani, K., Nakamura, T., Sakaue, H., Inoue, H.,

Ogawa, W., Tobimatsu, K., Takazawa, K., Sakai, M., et al. (2008). Dok1 medi-

ates high-fat diet-induced adipocyte hypertrophy and obesity through modu-

lation of PPAR-gamma phosphorylation. Nat. Med. 14, 188–193.

Hu, E., Kim, J.B., Sarraf, P., and Spiegelman, B.M. (1996). Inhibition of adipo-

genesis through MAP kinase-mediated phosphorylation of PPARgamma.

Science 274, 2100–2103.

Jensen, M.D. (2008). Role of body fat distribution and the metabolic complica-

tions of obesity. J. Clin. Endocrinol. Metab. 93(11, Suppl 1), S57–S63.

Kadrmas, J.L., and Beckerle, M.C. (2004). The LIM domain: from the cytoskel-

eton to the nucleus. Nat. Rev. Mol. Cell Biol. 5, 920–931.

Keuper, M., Wernstedt Asterholm, I., Scherer, P.E., Westhoff, M.A., Moller, P.,

Debatin, K.M., Strauss, G., Wabitsch, M., and Fischer-Posovszky, P. (2013).

TRAIL (TNF-related apoptosis-inducing ligand) regulates adipocyte

metabolism by caspase-mediated cleavage of PPARgamma. Cell Death Dis

4, e474.

Kissebah, A.H., Alfarsi, S., Evans, D.J., and Adams, P.W. (1982). Integrated

regulation of very low density lipoprotein triglyceride and apolipoprotein-B

kinetics in non-insulin-dependent diabetes mellitus. Diabetes 31, 217–225.

Lu, N.Z., Collins, J.B., Grissom, S.F., and Cidlowski, J.A. (2007). Selective

regulation of bone cell apoptosis by translational isoforms of the glucocorti-

coid receptor. Mol. Cell. Biol. 27, 7143–7160.

Macotela, Y., Emanuelli, B., Mori, M.A., Gesta, S., Schulz, T.J., Tseng, Y.H.,

and Kahn, C.R. (2012). Intrinsic differences in adipocyte precursor cells from

different white fat depots. Diabetes 61, 1691–1699.

McDonough, A.K., Curtis, J.R., and Saag, K.G. (2008). The epidemiology of

glucocorticoid-associated adverse events. Curr. Opin. Rheumatol. 20,

131–137.

Morton, N.M. (2010). Obesity and corticosteroids: 11beta-hydroxysteroid

type 1 as a cause and therapeutic target in metabolic disease. Mol. Cell.

Endocrinol. 316, 154–164.

Peckett, A.J., Wright, D.C., and Riddell, M.C. (2011). The effects of glucocor-

ticoids on adipose tissue lipid metabolism. Metabolism 60, 1500–1510.

Pehmøller, C., Brandt, N., Birk, J.B., Høeg, L.D., Sjøberg, K.A., Goodyear, L.J.,

Kiens, B., Richter, E.A., and Wojtaszewski, J.F. (2012). Exercise alleviates

lipid-induced insulin resistance in human skeletal muscle-signaling interaction

at the level of TBC1 domain family member 4. Diabetes 61, 2743–2752.

Perrini, S., Ficarella, R., Picardi, E., Cignarelli, A., Barbaro, M., Nigro, P.,

Peschechera, A., Palumbo, O., Carella, M., De Fazio, M., et al. (2013).

Differences in gene expression and cytokine release profiles highlight the

heterogeneity of distinct subsets of adipose tissue-derived stem cells in the

subcutaneous and visceral adipose tissue in humans. PLoS ONE 8, e57892.

Rancoule, C., Dusaulcy, R., Treguer, K., Gres, S., Guigne, C., Quilliot, D., Valet,

P., and Saulnier-Blache, J.S. (2012). Depot-specific regulation of autotaxin

with obesity in human adipose tissue. J. Physiol. Biochem. 68, 635–644.

Shao, D., Rangwala, S.M., Bailey, S.T., Krakow, S.L., Reginato, M.J., and

Lazar, M.A. (1998). Interdomain communication regulating ligand binding by

PPAR-gamma. Nature 396, 377–380.

Tchkonia, T., Giorgadze, N., Pirtskhalava, T., Tchoukalova, Y., Karagiannides,

I., Forse, R.A., DePonte, M., Stevenson, M., Guo, W., Han, J., et al. (2002). Fat

Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors 73

Cell Metabolism

Human Adipogenesis Is Critically Dependent on LMO3

depot origin affects adipogenesis in primary cultured and cloned human pre-

adipocytes. Am. J. Physiol. Regul. Integr. Comp. Physiol. 282, R1286–R1296.

Tchkonia, T., Tchoukalova, Y.D., Giorgadze, N., Pirtskhalava, T.,

Karagiannides, I., Forse, R.A., Koo, A., Stevenson, M., Chinnappan, D.,

Cartwright, A., et al. (2005). Abundance of two human preadipocyte subtypes

with distinct capacities for replication, adipogenesis, and apoptosis varies

among fat depots. Am. J. Physiol. Endocrinol. Metab. 288, E267–E277.

Tchkonia, T., Lenburg, M., Thomou, T., Giorgadze, N., Frampton, G.,

Pirtskhalava, T., Cartwright, A., Cartwright, M., Flanagan, J., Karagiannides,

I., et al. (2007). Identification of depot-specific human fat cell progenitors

through distinct expression profiles and developmental gene patterns. Am.

J. Physiol. Endocrinol. Metab. 292, E298–E307.

Tchkonia, T., Thomou, T., Zhu, Y., Karagiannides, I., Pothoulakis, C., Jensen,

M.D., and Kirkland, J.L. (2013). Mechanisms and Metabolic Implications of

Regional Differences among Fat Depots. Cell Metab. 17, 644–656.

Tchoukalova, Y.D., Koutsari, C., Votruba, S.B., Tchkonia, T., Giorgadze, N.,

Thomou, T., Kirkland, J.L., and Jensen, M.D. (2010). Sex- and depot-depen-

dent differences in adipogenesis in normal-weight humans. Obesity (Silver

Spring) 18, 1875–1880.

Todoric, J., Strobl, B., Jais, A., Boucheron, N., Bayer, M., Amann, S., Lindroos,

J., Teperino, R., Prager, G., Bilban, M., et al. (2011). Cross-talk between inter-

feron-g and hedgehog signaling regulates adipogenesis. Diabetes 60, 1668–

1676.

Tomlinson, J.J., Boudreau, A., Wu, D., Atlas, E., and Hache, R.J. (2006).

Modulation of early human preadipocyte differentiation by glucocorticoids.

Endocrinology 147, 5284–5293.

Tran, T.T., Yamamoto, Y., Gesta, S., and Kahn, C.R. (2008). Beneficial effects

of subcutaneous fat transplantation on metabolism. Cell Metab. 7, 410–420.

van der Laan, S., and Meijer, O.C. (2008). Pharmacology of glucocorticoids:

beyond receptors. Eur. J. Pharmacol. 585, 483–491.

Viguerie, N., Picard, F., Hul, G., Roussel, B., Barbe, P., Iacovoni, J.S., Valle, C.,

Langin, D., and Saris, W.H. (2012). Multiple effects of a short-term dexameth-

74 Cell Metabolism 18, 62–74, July 2, 2013 ª2013 The Authors

asone treatment in human skeletal muscle and adipose tissue. Physiol.

Genomics 44, 141–151.

Wabitsch, M., Brenner, R.E., Melzner, I., Braun, M., Moller, P., Heinze, E.,

Debatin, K.M., and Hauner, H. (2001). Characterization of a human preadipo-

cyte cell strain with high capacity for adipose differentiation. Int. J. Obes. Relat.

Metab. Disord. 25, 8–15.

Wainszelbaum, M.J., Liu, J., Kong, C., Srikanth, P., Samovski, D., Su, X., and

Stahl, P.D. (2012). TBC1D3, a hominoid-specific gene, delays IRS-1 degrada-

tion and promotes insulin signaling bymodulating p70 S6 kinase activity. PLoS

ONE 7, e31225.

Wajchenberg, B.L. (2000). Subcutaneous and visceral adipose tissue: their

relation to the metabolic syndrome. Endocr. Rev. 21, 697–738.

Walker, B.R., and Andrew, R. (2006). Tissue production of cortisol by 11beta-

hydroxysteroid dehydrogenase type 1 and metabolic disease. Ann. N Y Acad.

Sci. 1083, 165–184.

Wilson, M.D., Barbosa-Morais, N.L., Schmidt, D., Conboy, C.M., Vanes, L.,

Tybulewicz, V.L., Fisher, E.M., Tavare, S., and Odom, D.T. (2008). Species-

specific transcription in mice carrying human chromosome 21. Science 322,

434–438.

Wiper-Bergeron, N., Salem, H.A., Tomlinson, J.J., Wu, D., and Hache, R.J.

(2007). Glucocorticoid-stimulated preadipocyte differentiation is mediated

through acetylation of C/EBPbeta by GCN5. Proc. Natl. Acad. Sci. USA 104,

2703–2708.

Zeyda, M., Farmer, D., Todoric, J., Aszmann, O., Speiser, M., Gyori, G.,

Zlabinger, G.J., and Stulnig, T.M. (2007). Human adipose tissue macrophages

are of an anti-inflammatory phenotype but capable of excessive pro-inflamma-

tory mediator production. Int J Obes (Lond) 31, 1420–1428.

Zheng, Q., and Zhao, Y. (2007). The diverse biofunctions of LIM domain pro-

teins: determined by subcellular localization and protein-protein interaction.

Biol. Cell 99, 489–502.

1

Cell Metabolism, Volume 18

Supplemental Information

Human but Not Mouse Adipogenesis

Is Critically Dependent on LMO3 Josefine Lindroos, Julia Husa, Gerfried Mitterer, Arvand Haschemi, Sabine Rauscher, Robert Haas, Marion Gröger, Robert Loewe, Norbert Kohrgruber, Klaus F. Schrögendorfer, Gerhard Prager, Harald Beck, J. Andrew Pospisilik, Maximilian Zeyda, Thomas M Stulnig, Wolfgang Patsch, Oswald Wagner, Harald Esterbauer, and Martin Bilban

2

Figure S1. GC Induced LMO3 Expression in Human Adipose Depots, SGBS- and 3T3-L1 Cells, Related to Figure 1 All error bars represent means ± SEM. P values; ns=not significant; *P< 0,05, **P< 0,001 and ***P< 0,0001. (A) qPCR verification of LMO3 mRNA expression during

3

hASC differentiation time-course. Differentiation induced in 2 days post-confluent cells. (B) LMO3 mRNA levels measured in hASC after GC treatment for indicated hours. Dex, PRD and HC, were added in growth medium (GM) at indicated concentrations. Dex was additionally added in serum-free medium (SFM) at indicated concentrations. (C) Upper panel: Plotted amplification curves for respective gene upon Dex (1 µM) treatment in 3T3-L1 cells. The brain RNA lysates are from C57Bl/6J lean mice and have not been stimulated with Dex. Lower panel: Relative mRNA levels upon Dex (1 µM) stimulation throughout indicated timepoints in 3T3-L1 cells and primary mouse adipose stromal cells (mASC). nd = non detectable for day 5 3T3-L1 and mASC. (D) Human ASC treated with vehicle or carbenoxolone (100 µM) in the full mix throughout the entire differentiation. Cells were fixed and oil red O (ORO) staining was performed on day 10. Representative wells upper panel and quantification lower panel. (E) Individual components of the adipogenic cocktail were added to confluent Simpson-Golabi-Behmel Syndrome (SGBS) cells. Cells were harvested 24 hrs later for RT-PCR analysis. Full Mix (FM) refers to the differentiation cocktail. (n=3). Isobutyl-methylxanthine (IBMX); troglitazone (TZD); triiodothyronine (T3); RU486 (1 µM). All components were added in growth medium with concentrations listed under experimental procedures. (F) LMO3 mRNA levels from 25 obese females from the matched SAT and VAT biopsies shown in figure 1L, with low CD68 mRNA expression (G) Correlation plot shows lack of significant correlation between LMO3 and CD68 mRNA in the 25 subjects plotted in (F). (H) Correlation plot between LMO3 and HSD11B1 mRNA is apparent in the 25 subjects with low CD68 mRNA expression seen in (F). (I) Basal mRNA expression of indicated genes in matched SC and VI matched hASC. n=6. (J) Pie chart depicting LMO3-dependent gene expression in VI preadipocytes by microarray analysis of matched SAT and VAT preadipocytes related to Figure 1O. Approximately 21 % (181 genes), were differentially regulated upon LMO3-siRNA treatment referred to as LMO3-dependent genes. (K) Pathway enrichment analysis on LMO3-dependent genes in VAT (181 in total) with DAVID and EASE. Enrichment is shown as a Z-score. (L) RT-PCR analysis of LMO3-dependent genes in VAT. Colour coding matches the gene pathway enrichment graph shown in (K).

4

Figure S2. LMO3 Expression during Adipogenesis Is Human Specific, Related to Figure 2 All error bars represent means ± SEM. P values; ns=not significant; *P< 0,05, **P< 0,001 and ***P< 0,0001. (A) 2 day post-confluent SGBS cells were induced to differentiate into adipocytes. Total RNA was isolated at the indicated times post induction. mRNA levels of respective gene were measured by RT-PCR and normalized to the amounts of RPLP0. Bar graphs indicate mean fold change ± SEM of three independent experiments. (B) Verification of successful hASC differentiation, based on proadipogenic marker mRNA expressions analyzed by DNA Microarray expression profiling. Shades of red and blue indicate distinct degrees of gene activation. Fold-change activation from day 0 of differentiation is shown to the right of the heatmaps. (C) to (D) Scale bars = 20 µm. (C) Specificity of staining (seen in Figure 2H) is verified by negative isotype controls for LMO3 and perilipin in hASC on day 0 and 6. (D) Specificity of staining (seen in Figure 2I) is verified by negative isotype controls for LMO3, CD68 and perilipin in human WAT. (E) Lmo3 mouse tissue distribution pattern evaluated by a commercially available pool of murine multiple tissue RNA library by RT-PCR analysis and normalized to the amounts of Rplp0 in mouse. Each tissue is a pool of at least 3 donors. (F) Differentiation timecourse of 3T3-L1 cells indicating Lmo3 and Pparg mRNA expression. (G) to (I) Ponceau staining as loading controls for blots shown in Figures 2K to N. Total lane intensity was quantified to brain lane.

5

Figure S3. Impact of Silencing and Overexpressing LMO3 in hASC and Murine 3T3-L1 Cells and Phosphorylation Status of S112 PPARγ in Differentiated hASC, Related to Figures 3 and 4 All error bars represent means ± SEM. P values; ns=not significant; *P< 0,05, **P< 0,001 and ***P< 0,0001. (A) BrdU incorporation in hASC (cell proliferation) was initiated 48 hrs post transfection. Untreated cells are non-transfected. Differentiation timepoints chosen to match Figure 3. (B) MTT assay evaluating cell viability of untransfected (untreated) and siOligo transfected hASC. Assay was initiated 48 hrs post transfection. Differentiation timepoints chosen to match Figure 3. (C) Human ASC were subjected to siOligos; siCtrl or siLMO3 then post 48 hrs the FM with or without GCs (Dex and HC, both at 1 µM), throughout the entire differentiation. Cells

6

were fixed and lipid droplets stained with Oil red O on day 18 (representative wells upper panel, magnification 10X) and quantified (lower panel). (D) Specificity of staining (seen in Figure 3F) is verified by negative isotype controls for human nuclei specific marker MAB1281 and perilipin in both siCtrl- and siLMO3 injected SCID mice. Scale bar = 20 μM. (E) RT-PCR evaluating Lmo3 mRNA levels. siCtrl or siLmo3 murine 3T3-L1 cells were harvested on day 2 of differentiation (initiated 48 hrs post transfection). (F) Oil red O staining on siCtrl or siLmo3 transfected 3T3-L1 cells. Differentiation was initiated 48 hrs post transfection and cells fixed on day 7 (representative wells upper panel, magnification 10X) and quantified (lower panel). (G) Protein analysis of empty pMMP vector or retrovirus expressing LMO3 transduced 3T3-L1 preadipocytes. (H) Oil red O staining on pMMP or pLmo3 expressing 3T3-L1 cells. Differentiation was initiated 48 hrs post transfection and cells fixed on day 10 (representative wells upper panel, magnification 10X) and quantified (lower panel). (I) Transfected and day 6 differentiated hASCs were serum starved overnight then replenished for 1 hr with 40 % FBS. Top panels are representative blots from at least three donors. Bottom panels: Densitometric evaluation.

7

Table S2. Anthropometric and Metabolic Characteristics of Human Study Subjects, Related to Figure 1

Trait Non-obeseA ObeseA Individuals from obese cohort CD68LOW

A,B

N 7 55 25

Male/Female (n) 0/7 15/40 0/25

Age (years) 46.4 (33-57) 38.3 (17-59) 43.4 (30-56)

BMIC (kg/m2) 25.2 (20.3-28.1) 43.2 (33.0-60.5) 42.2 (33.5-60.5)

Glucose, fasting (mmol/L) 5.0 (4.3-5.7) 5.2 (4.1-6.1) 5.1 (4.1-5.9)

Insulin, fasting (pmol/L) 47.5 (15.3-84.7) 128.1 (24.3-403.5) 98.4 (24.3-243.1)

CD68 mRNA (AUC) 3.7 (1.89-6.25) 4.1 (1.28-9.89) 3.3 (1.28-5.98)

AData are numbers of observations or means (range). BCD68LOW, obese female study subjects displaying CD68 mRNA values within the minimum to

maximum range of healthy, non-obese females. CAbbreviations: BMI, body mass index; AU, arbitrary unit.

8

Table S3. Correlation of LMO3 and HSD11B1 mRNA Expression Levels in Fat Depots from Obese and Obese, CD68LOW Study subjects, Related to Figure 1 Obese Study Subjects (n=55)

SATA VATA

Adjusted Beta/RB PB Adjusted Beta/RB PB Adjusted for

-0.110 0.426 0.268 0.048 -

-0.150 0.285 0.314 0.025 sex, age

-0.150 0.290 0.312 0.026 sex, age, BMIA

Obese, CD68LOWC Study Subjects (n=25)

SATA VATA

Adjusted Beta/RB PB Adjusted Beta/RB PB Adjusted forD

0.178 0.292 0.364 0.025 -

0.152 0.416 0.395 0.022 Age

0.151 0.435 0.372 0.028 age, BMIA

0.133 0.538 0.427 0.044 age, CD68

0.131 0.553 0.397 0.049 age, BMIA, CD68

AAbbreviations: SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue; BMI, body mass

index. BStandardized beta-coefficients and P values obtained from regression analysis. Linear regression

adjusted for indicated variables. All variables normally distributed before or after log10-transformation. CCD68LOW: obese female study subjects displaying CD68 mRNA values within the minimum to

maximum range of healthy, non-obese females. DNo sex adjustment as all study subjects are female.

9

Table S5. Characteristics of the Non-obese, Obese Low HOMA-IR and Obese High HOMA-IR Study Participants, Related to Figure 1

Trait Non-Obese

Obeselow HOMA-

IR

Obesehigh HOMA-

IR

P* P† P‡

Male/female (n) 0/7 4/18 8/18 Age (years) 46.2 (3.5) 39.6 (2.2) 37.1 (2.3) n.s. n.s. n.s.BMI (kg/m2) 25.2 (1.2) 41.8 (1.4) 44.4 (0.9) n.d. n.d. n.s.HOMA-IR 1.6 (0.4) 1.5 (0.1) 7.2 (0.5) n.d. n.s. n.d.Waist circumference (cm) 88.5 (4.5) 119.5 (3.1) 132.0 (3.0) n.d. n.d. 0.033

Insulin (pmol/L) 47.5 (10.0) 47.0 (2.2) 210.6 (14.0) n.d. n.s. 0.000

Glucose (mmol/L) 5.0 (0.2) 4.8 (0.1) 5.4 (0.1) n.d. n.s. 0.000

LMO3mRNA, VAT (AU) 8.72 (2.90) 8.32 (1.32) 8.05 (0.80) n.s. n.s. n.s.

LMO3 mRNA, SAT (AU) 0.49 (0.13) 2.06 (0.82) 0.77 (0.18) n.s. n.s. n.s.

HSD11B1 mRNA, VAT AU)

0.35 (0.07) 0.39 (0.06) 0.75 (0.06) 0.000 n.s. 0.000

HSD11B1 mRNA, SAT AU)

0.46 (0.13) 0.73 (0.12) 0.63 (0.08) 0.000 n.s. 0.000

Data are numbers of observations or unadjusted and untransformed means (SEM). Two-sided P

values obtained from ANOVA (Scheffé-Test for subgroup comparisons adjusted for age and sex).

Heteroskedasticity tested by the Breusch-Pagan/Cook-Weisberg test, normal distribution of

quantitative traits tested by the Shapiro-Francia normality test. Non-normal distributed parameters

transformed (log10 transformation for HOMA-IR, insulin, LMO3 mRNA and HSD11B1 mRNA);

*comparison among all groups; †non-obese controls vs. obese low HOMA-IR; ‡obese low HOMA-IR

vs. obese high HOMA-IR; participants (total) n = 55 (7/22/26; non-obese/obese low HOMA-IR/obese

high HOMA-IR); low HOMA-IR, HOMA-IR <=2.0; high HOMA-IR, HOMA-IR >=5.0; n.s., not significant;

n.d., not determined; VAT., Visceral Adipose Tissue; SAT, Subcutaneous Adipose Tissue, AU,

arbitrary units.

10

Table S6. Correlations between Visceral Adipose LMO3 mRNA and Anthropometric and Metabolic Characteristics of Non-obese and Obese Study Subjects, Related to Figure 1 Trait Beta/R* P* Beta/R† P†

BMI (kg/m2) -0.157 n.s. -0.032 n.s.

HOMA-IR 0.092 n.s. 0.179 n.s.

Waist circumference (cm) -0.140 n.s. n.d.

Insulin (pmol/L) 0.092 n.s. 0.180 n.s.

Glucose (mmol/L) 0.048 n.s. 0.094 n.s.

HSD11B1 mRNA, VAT. (AU) 0.284 0.031 0.359 0.010

Standardized beta-coefficients (Beta/R) and two-sided P values obtained from regression analysis.

Normal distribution of traits tested by the Shapiro-Francia normality test. Non-normal distributed

parameters transformed (log10 transformation for HOMA-IR, insulin, LMO3 mRNA and HSD11B1

mRNA).

*adjusted for age and sex; †adjusted for age, sex and waist circumference; participants (total) n = 62

(7 non-obese/55 obese); n = 59 (4/55) for waist circumference and waist circumference adjusted;

Abbreviations: n.s., not significant; n.d., not determined; VAT., Visceral Adipose Tissue., visceral; AU,

arbitrary units.

11

SUPPLEMENTAL EXPERIMENTAL PROCEDURES Isolation of Preadipocytes Extended Procedure Minced adipose tissue was washed in DMEM (PAA) thereafter digested with 1 mg/ml collagenase type II (Worthington) and 50 units/ml deoxyribonuclease I (Sigma) in DMEM, supplemented with 50 µg/ml gentamicin (Invitrogen) for 1 hr at 37°C with constant shaking. Cells were filtered through 100 µm nylon filters and centrifuged. Floating cells were taken as mature adipocytes. Erythrocytes were lysed in hypotonic buffer (Qiagen), thereafter centrifuged and cell pellet resuspended in growth medium: DMEM/Ham's F12 (PAA), 50 µg/ml gentamicin and 10 % fetal bovine serum (PAA). After filtration using a 20 µm nylon mesh, cells were incubated at 37°C with 5% CO2 for 24-48 hrs. Mouse Studies Extended Information C57BL/6J mice were fed chow, LFD or HFD for 10 weeks; db/db and db/+ mice LFD or HFD for 20 weeks (LFD D12450B, HFD D12492; Research Diets Inc.) (diets administered ad libitum).Mice were sacrificed and depots excised and snap frozen in liquid N2 prior to further processing. Weights: C57BL/6J LFD: 33.3 ±1.6 g, chow: 35.9 ±5.0 g, HFD: 47.9 ±3.6 g; db/db LFD 46 ±0.7 g, HFD 54.7 ±1 g, and db/+ LFD: 29.3 ±0.5 g, HFD 31.1 ±0.8 g. RT-PCR RT-PCR was performed as described in main article under experimental procedures. Primer sequences are listed in Table S9. All RT-PCR data are normalized to amounts of acidic ribosomal phosphoprotein P0 (RPLP0), unless otherwise stated. An in-house made mouse tissue library was used to analyze the tissue distribution of mouse Lmo3 expression. Western Blot Analyses Extended Information PVDF membranes (GE healthcare) were incubated with following antibodies: LMO3 (Abnova), LMO3 (Santa Cruz), TopoIIβ (BD Biosciences), GAPDH (Santa Cruz), V5 antibody (Invitrogen), p-p44/42 MAPK (pERK1/2) (Cell Signaling), p44/42 MAPK (ERK1/2) (Cell Signaling); HSP70 (Cell Signaling), PPARγ (Santa Cruz), p-S112-PPARγ (Abcam), CEBPβ (Santa Cruz), Perilipin, (Cell Signaling), MAC2 (Cedarlane) and CEBPα (Santa Cruz). Hrp-conjugated IgG secondary antibodies were used (Cell Signaling) and blots were developed with ECL Plus Western Blotting Detection System (GE Healthcare). TotalLabQuant software (TotalLab Limited) was used for densitometric quantification. Gene Expression Profiling Extended Information GC target genes were identified with a time-course in hASC cells treated with Dex . Genes with signal intensities below 100 were eliminated. Genes exhibiting ±3-fold change relative to control treatment (in at least three out of five comparisons) were selected for further analysis. This approach identified 121 up- and 90 downregulated genes in response to Dex treatment in hASC. LMO3 target genes following GC stimulation were identified with duplicate well RNA pooled for GeneChip analysis (1 chip per time point) following a published 2-step protocol (Asada et al., 2011): First, genes responsive to the adipogenic cocktail in siCtrl-treated (control) hASC with mean ±1.5-fold expression day 6 vs. 0 of differentiation, yielded a set of 1892 adipogenesis-regulated genes. Next, from these 1892 genes we identified LMO3-dependent genes by comparing fold-changes (day 6 vs.0) of siCtrl with LMO3 knockdown (siLMO3 #1 or siLMO3 #2) hASC. Genes were

12

classified as LMO3-dependent if the fold-change was ±1.5–fold different between siCtrl- or siLMO3 transfected hASC. DAVID and EASE algorithm was used to identify specific biological pathways. The results were visualized using GenespringGX (Agilent, Santa Clara, CA). Human and Murine Primers and Silencing Oligonucleotide-Sequences. siOligo sense sequences: LMO3 #1, 5‘-GGUAUCUUCUAAAGGCACUGGACAA-3‘;LMO3 #2, 5‘-CCCUGUACACUAAAGCUAAUCUUAU-3‘;GR, 5’-GCAUGUACGACCAAUGUAA-3’; PPARG, 5’-GGGCGAUCUUGACAGGAAATT-3’; Lmo3, 5’-GGAGACAAAUUUUUCUUAATT-’3. Genes forward- followed by reverse primer sequences: ADIPOQ, GCTGGGAGCTGTTCTACTGC, CGATGTCTCCCTTAGGACCA; BTK, GTCCTGGCATCTCAATGCATC, AATCACTGCGGCCATAGCTTC; CD36, GGGAAAGTCACTGCGACATGAT, ACGTCGGATTCAAATACAGCATAGA; CD68, GCTACATGGCGGTGGAGTACAA, ATGATGAGAGGCAGCAAGATGG; CEBPA, CTTGTGCCTTGGAAATGCAA, GCTGTAGCCTCGGGAAGGA; CXCL5, GACCACGCAAGGAGTTCATCC, AGGCTACCACTTCCACCTTGG; ENPP2, ATCTTCTCCGTGGATGGCTTC, GAGAGTGTGTGCCACAAGACC; FABP4, AAGGCGTCACTTCCACGAGAG, AATGCGAACTTCAGTCCAGGTC; FMN1, CCAATTCTAAGCCAGCGGATG, CAGCTTGAAGTCTGCCAGGAG; GPRC5A, CAAGGTGCAGGACTCCAACAG, ATGATGAAGGCGAAGGTGAGG; HSD11B1, TCATTCTCAACCACATCACCAACACT, CCAGCCAGAGAGGAGACGACAAC; ING2, CAGCAGCAGCAGCAACTGTAC, TCCACGCACTCAAGGTAGTCC; LMO3, AAGGTTGTGCTGGCTGCAAC, GGCACACTTCAGGCAGTCTTC; LPL, TGGAGGTACTTTTCAGCCAGGAT, TCGTGGGAGCACTCACTAGCT; PLCL1, CTGTCATCTCGGCTCATCACC, TCTGCACATCTGGAATTGCAC; PLIN1, GACAACGTGGTGGACACAGT, CTGGTGGGTTGTCGATGTC; PPARG1, AAGGCCATTTTCTCAAACGA, AGGAGTGGGAGTGGTCTTCC; PPARG2, CCATGCTGTTATGGGTGAAA, TCAAAGGAGTGGGAGTGGTC; PCDH18, CTCCAGTGTGCAGCCTTCTTC, CCATCATTGAGGTGGTTGAGC; PSPH, TGGTGCCACAGATATGGAAGC, GGCGTTATCCTTGACTTGTTGC; RPLP0, GTCATCCAGCAGGTGTTCGAC, CTCCAGGAAGCGAGAATGCAG; SLC7A14, ATCGCAGGCCTCTTCTTCATC, GAAGCATGTTGCTGCTCCTTG; TBC1D2B, AATGCAACTGCAGGTCCAGAG, CTGGAGCAGTCGAACAAGCTC; TNFSF10, GCTCCTGCAGTCTCTCTGTGTG, GTCCTGCATCTGCTTCAGCTC; TSPAN13, CTTCGCGTGTTCCAAGAACTG, CCACGCAGCAATTCCAATTAG; VNN1, CTGGTGGCACGCTACCATAAG, TCACAATCTCAGGCTCCTTGG; Lmo3, GCATGAGGACTGCCTGAAGTG, GCCTTCGTGTACAAGGTGGAG; Pparg, GCATGGTGCCTTCGCTGA, TGGCATCTCTGTGTCAACCATG; Cebpd, ATCGACTTCAGCGCCTACAT, GCTTTGTGGTTGCTGTTGAA; Rplp0, GCCAATAAGGTGCCAGCTGCTG, GAAGGAGGTCTTCTCGGGTCCTAG Luciferase Assays Extended Information For LMO3-promoter activity assays 293FT cells were co-transfected in 48-wells with either 50 ng of a LMO3 luciferase reporter plasmid (Switchgear Genomics) or a 5 point mutated LMO3 luciferase reporter plasmid within the human GRE1 site along with 250 ng of a GR expression plasmid (Addgene) or control DNA (pcDNA3.1, Invitrogen; pcDNA was added for a consistent amount of DNA and set to 1 for comparison). 10 ng of pCMV-β-gal expression plasmid was used as control for transfection efficiency. Point mutations to generate pLMO3-Luc-Hs.->Mm was performed by site-directed mutagenesis with the QuikChange Multi Site Directed Mutagenesis kit (Stratagene). The mutations within the human GRE1 sequence were

13

generated using a mutagenic primer: 5'-CCCATCCGCACAGCTCCCTgcTtgTCcACCCCACCTCTCTGCTAC-3' (upper strand sequence between −951 and −906 bp; the lowercase letters indicate nucleotide substitutions). DNA sequencing verified the subcloned fragment. For PPARG-reporter assays 250 ng PPARG-responsive reporter plasmid p(AOx)3-TKSL (provided by Christopher K Glass, University of California, San Diego, CA, USA), 250 ng PPARG2 expression plasmid (pPPARG) or 250 ng PPARG-S112A expression plasmid (pPPARG-S112A), 50, 100 or 250 ng pLMO3-V5 expression plasmid (Origene), or 250 ng pMCL-HA-MAPKK1-11A55B [S218E/S222D] expression plasmid (Addgene) and 10 ng Renilla luciferase control plasmid (ph-RL; Promega) were transfected into cells using Lipofectamine 2000 (Invitrogen). ELK-1 activity was measured using the luciferase reporter gene assay by transfecting 293FT cells with plasmids containing Gal4-UAS-luciferase (pFR-Luc) and a fusion of activation domain of ELK-1 and Gal4 DNA-binding domain (DBD) (pFA2-ELK-1). ELK-1 activation in the nucleus causes binding of Gal4-DBD to Gal4-UAS, leading to luciferase transcription. Cell were cotransfected with Renilla luciferase control plasmid (ph-RL; Promega). 293FT cells cultured on 24-well plates (plated at 5 × 105/well density) were cotransfected with pFR-Luc (0.3 μg/well), pFA2-ELK-1 (0.3 μg/well), phRL (0.02 μg/well), pLMO3 or control DNA (pcDNA3.1) using Lipofectamine 2000. 24 hr after transfection, cells were treated with EGF (Sigma,100 ng/ml) for 20 h. Luciferase activities were measured using the Dual-Luciferase Reporter Assay System (Promega) or the LightSwitch Luciferase Assay (Switchgear Genomics) and a multimode microplate reader (Synergy 2, BioTek Instruments). Luminescence values were normalized to Renilla luminescence levels or ß-Galactosidase activity to the ß-Galactosidase Enzyme Assay System (Promega), respectively. Oil Red O Staining Cells were fixed in 10% formalin, thereafter stained with Oil Red-O (Sigma) working solution (composed of 4 parts water and 6 parts 0.6% Oil red O dye in isopropanol). Quantification was performed by eluting stained cells and measuring optical density using a multimode microplate reader (Synergy 2, BioTek Instruments). Confocal immunofluorescence Microscopy Extended Information HASC were seeded in matrigel (BD Biosciences) pre-coated chamber slides and differentiated then fixed with cold methanol. Blocking: Antibody diluent with background reducing components (Dako) and human IgG (Jackson ImmunoResearch Laboratories). Antibodies: LMO3 (Abnova) and Perilipin (Cell Signaling) or respective IgG control (Sigma-Aldrich). Secondary fluorescent labeled antibodies were used for visualization (both Invitrogen). Human VAT was fixed in 10% formalin (Sigma) then paraffin embedded. Paraffin sections were deparaffinized followed by antigen retrieval by boiling in citrate buffer (Dako). Sections were blocked with 3% bovine serum albumin (BSA) then incubated with: LMO3 (Santa Cruz), Perilipin (Cell Signaling), CD68 (Dako) or respective IgG Control (all Sigma). Additional blocking step with goat serum (Invitrogen) prior to TRITC (Jackson Immunotech) was performed. Thereafter remaining secondary antibodies: Cy5 (Jackson Immunotech) and Alexa 488 (Invitrogen) followed by 4',6-diamidino-2-phenylindole (DAPI, Sigma). All antibodies were diluted in 3% BSA solution. SCID Mouse Xenotransplant Model Extended Information. Female, 6-week-old Crl:SHO-PrdkcscidHrhr mice were obtained from Charles River (Sulzfeld, Germany) and housed as previously described (Loewe et al., 2006). A

14

previously published protocol was followed with minor modifications (Ahfeldt et al., 2012). In brief, siCtrl- or siLMO3 transfected hASC were differentiated for 2 days. Thereafter injected subdermally into the flank of each mouse (n=5 per group; 150,000 cells per injection). 6 weeks post injection, mice were sacrificed and transplantation tissue-site was collected, paraffin embedded, sectioned and stained. 4 sections were chosen for each mouse on the basis of when the sections transversed the tissue (light microscopy evaluation). Sections were stained for 24 hrs at 4°C with MAB1281 and with Perilipin for the last hour. Thereafter secondary antibodies and DAPI was added. In total 120 fields per view were quantified (3 fields of view per section) by counting double positive stain for MAB1281 and perilipin in relation to positive perilipin stained cells in a blinded manner. MTT Cell Viability Assay Untreated, siCtrl, siLMO3 (#1) and (#2) hASCs were plated into 96-well culture plates at a density of 10,000 cells per well. 2 days post amaxa differentiation was initiated and cell viability was established using MTT Cell Viability Assay Kit (Biotium) according to manufacturer’s protocol on indicated days. In brief; 10 µl MTT solution was added to 100 µl culture medium. After 4 hrs of incubation at 37°C, 200 µl dimethylsulfoxide (DMSO) was added to each well, carefully pipetting up and down to dissolve formazan. Absorbance was measured on a multimode microplate reader (Synergy 2, BioTek Instruments) at 570 nm and a reference wavelength of 630 nm. BrdU Cell Proliferation Assay Untreated, siCtrl, siLMO3 (#1) and (#2) hASCs were plated into 96-well culture plates at a density of 10,000 cells per well. 2 days post amaxa differentiation was initiated and cell proliferation was established using the Cell Proliferation ELISA, BrdU kit (Roche) according to manufacturer’s protocol. Plates were subsequently kept at 4°C prior to harvest of day 10 and assay was then performed. In brief, 10 µl BrdU labelling solution was added to 100 µl differentiation medium per well. After 4 hrs of incubation at 37°C, labelling medium was removed and 200 µl/well FixDenat was added to the cells for 30 minutes at RT. Solution was removed by tapping, thereafter 100 µl/well anti-BrdU-POD working solution was added for 90 minutes at RT. Wells were rinsed with washing solution 3 times, then 100 µl/well substrate solution was added for 15 minutes at RT, followed by adding 1M H2SO4 stop solution 25 µl/well and absorbance was measured on a multimode microplate reader (Synergy 2, BioTek Instruments) at 450 nm and a reference wavelength of 690 nm. Retrovirus Preparation and Infection Retrovirus preparation and infection were performed as described (Bilban et al., 2008). Human LMO3 cDNA in pDEST-51 vector (Origene) was subcloned into the retroviral vector pMMP (kindly provided by Dr. Klaus Schmetterer, Department of Laboratory Medicine, Medical University of Vienna, Vienna, Austria) with 5´ HindIII and 3´ NotI (Invitrogen, Carlsbad, CA) restriction enzymes. Confirmation was verified by restriction site analysis and sequencing. Briefly, pMMP empty vector or pMMP vector containing LMO3 cDNA, along with vectors containing reverse transcriptase (gag-pol) and VSV-G-expressing plasmids, was transfected into 293FT packaging cells with Lipofectamine 2000 (Invitrogen). Viral supernatant was collected 48 hrs after transfection, filtered through 0.45 µm filters, and added to target cells for 12 hrs along with 8 g/ml Polybrene. GFP+ cells were sorted on a FACSAria (Becton Dickinson, San Jose, CA, USA) to make stable lines and were maintained in media containing appropriate antibiotics.

15

SUPPLEMENTAL REFERENCES

Ahfeldt,T., Schinzel,R.T., Lee,Y.K., Hendrickson,D., Kaplan,A., Lum,D.H., Camahort,R., Xia,F., Shay,J., Rhee,E.P., Clish,C.B., Deo,R.C., Shen,T., Lau,F.H., Cowley,A., Mowrer,G., Al-Siddiqi,H., Nahrendorf,M., Musunuru,K., Gerszten,R.E., Rinn,J.L., and Cowan,C.A. (2012). Programming human pluripotent stem cells into white and brown adipocytes. Nat. Cell Biol. 14, 209-219.

Asada,M., Rauch,A., Shimizu,H., Maruyama,H., Miyaki,S., Shibamori,M., Kawasome,H., Ishiyama,H., Tuckermann,J., and Asahara,H. (2011). DNA binding-dependent glucocorticoid receptor activity promotes adipogenesis via Kruppel-like factor 15 gene expression. Lab Invest 91, 203-215.

Loewe,R., Valero,T., Kremling,S., Pratscher,B., Kunstfeld,R., Pehamberger,H., and Petzelbauer,P. (2006). Dimethylfumarate impairs melanoma growth and metastasis. Cancer Res. 66, 11888-11896.