Cellular and Molecular

Protein Array Reveals Differentially ExpressedProteins in Subcutaneous Adipose Tissue inObesityMartina Skopkova,* Adela Penesova,* Henrike Sell,† Zofia Radikova,* Miroslav Vlcek,* Richard Imrich,*Juraj Koska,* Jozef Ukropec,* Jurgen Eckel,† Iwar Klimes,* and Daniela Gasperıkova*

AbstractSKOPKOVA, MARTINA, ADELA PENESOVA,HENRIKE SELL, ZOFIA RADIKOVA, MIROSLAVVLCEK, RICHARD IMRICH, JURAJ KOSKA, JOZEFUKROPEC, JURGEN ECKEL, IWAR KLIMES, ANDDANIELA GASPERIKOVA. Protein array revealsdifferentially expressed proteins in subcutaneous adiposetissue in obesity. Obesity. 2007;15:2396–2406.Objective: Many adipokines, inflammatory cytokines, andother proteins produced by adipose tissue have been shownto be involved in the development of obesity-related insulinresistance. Nevertheless, new factors that play an importantrole in these processes are still emerging. Therefore, wescreened the level of 120 different proteins in biopsies ofsubcutaneous adipose tissue (ScAT) of lean and obese sub-jects.Research Methods and Procedures: All studied volunteers(12 obese with BMI �30 and 6 lean with BMI �25kg/m2) were young, clinically healthy, and drug-naivemales with normal glucose tolerance. The ScAT wasobtained by a needle biopsy from the umbilical region.Protein levels were assessed in adipose tissue lysatesusing protein arrays; mRNA levels were determined withthe aid of real-time reverse transcription-polymerasechain reaction (RT-PCR).Results: The obese subjects had higher fasting plasma glu-cose (although within the normal range) and insulin levels,

increased high sensitivity C-reactive protein (hsCRP) incirculation, and decreased in vivo insulin action. Using theprotein array technique, it was shown that of 120 proteinsmeasured, 27 showed higher levels (leptin, HGF, EGF-R,FGF-6, IGF-1sR, Fas/Apo-1, ENA-78, PARC, lymphotac-tin, HCC-4, IL-10, IL-1a, IL-1R1, IL-1R4, IL-12p70, an-giopoietin-2, Axl, Dtk, MIF, MIP-1a, �1b, �3b, MSP-a,osteoprotegerin, TECK, TIMP-1, -2) and only one (RAN-TES) showed a lower level in ScAT of obese subjects whencompared with the lean controls (p � 0.05). The real-timeRT-PCR confirmed the results of protein arrays for leptin,MIF, MIP-1a, TIMP-2, adiponectin, IL-6, and TNF-� butnot for RANTES.Discussion: To our knowledge, this is the first protein arraydata on a very early dysregulation of ScAT protein levels ininsulin-resistant obese, but apparently healthy, subjects withnormal glucose tolerance.

Key words: cytokines, inflammation, insulin resistance,protein

IntroductionObesity is one of the main contributing factors to the

development of type 2 diabetes mellitus. However, detailedmechanisms by which obesity contributes to insulin resis-tance are not yet fully understood. New information fromthe last couple of years shows that obesity is associated withsubclinical inflammation, as manifested by increased levelsof high sensitivity C-reactive protein (hsCRP)1 and var-ious cytokines in circulation and by enhanced macro-phage infiltration into adipose tissue (1–3). Many studieshave already shown that stimulation of inflammatorypathways directly and/or indirectly inhibits insulin sig-

Received for review October 9, 2006.Accepted in final form February 20, 2007.The costs of publication of this article were defrayed, in part, by the payment of pagecharges. This article must, therefore, be hereby marked “advertisement” in accordance with18 U.S.C. Section 1734 solely to indicate this fact.*Institute of Experimental Endocrinology Slovak Academy of Sciences, Bratislava, Slova-kia; and †German Diabetes Center, Institute of Clinical Biochemistry and Pathobiochemis-try, Dusseldorf, Germany.Address correspondence to Daniela Gasperıkova, Institute of Experimental Endocrinology,EU Centre of Excellence, Slovak Academy of Sciences, Vlarska 3, SK-83306 Bratislava,Slovak Republic.E-mail: daniela.gasperikova@savba.skCopyright © 2007 NAASO

1 Nonstandard abbreviations: hsCRP, high sensitivity C-reactive protein; TNF-�, tumornecrosis factor-alpha; IL-6, interleukin-6; PCR, polymerase chain reaction; OGTT, oralglucose tolerance test; EHC, euglycemic hyperinsulinemic clamp; HOMA, homeostasismodel assessment; RT-PCR, reverse transcription-PCR; O, obese group; C, control group.

2396 OBESITY Vol. 15 No. 10 October 2007

naling (4). Thus, it is now recognized that subclinicalinflammation is involved in the development of insulinresistance in the obese state (5).

There are several cytokines and adipokines produced byadipose tissue that have been extensively studied in relationto inflammation in obesity [e.g., adiponectin, leptin, resistin,tumor necrosis factor-alpha (TNF-�), interleukin-6 (IL-6)](6–8). Adipose tissue produces, however, an additionallarge number of biologically active molecules, includingseveral cytokines that might potentially play a role in thedevelopment of obesity-related insulin resistance. There-fore, we used a pre-designed high-throughput protein array,which permits simultaneous quantification of 120 cytokinesand related proteins, to screen their levels in subcutaneousadipose tissue of lean and obese subjects. In addition, wemeasured expression of genes for selected cytokines in thistissue to confirm the results obtained by the protein array.The volunteer recruitment criteria were set to identifyyoung, healthy, obese men without diabetes or pre-diabetes,who were compared with a group of young, healthy, leanindividuals. Thus, the primary aim of this study was toprovide new information about very early abnormalities inthe adipose tissue proteomic profile and to link obesity toinflammation and early signs of insulin resistance in obese,though yet clinically healthy, subjects.

Research Methods and ProceduresSubjects and Study Design

The nature and potential risks of the study were explainedto all subjects before obtaining their written informed con-sent. The study was approved by the Ethics Committee ofthe Derer’s Faculty Hospital in Bratislava and conforms tothe ethical guidelines of the Declaration of Helsinki asrevised in 2000 (9).

White males were recruited using the following inclusioncriteria: 1) age 20 to 45 years, 2) BMI either 19 to 25 kg/m2

(lean group) or 30 to 40 kg/m2 (obese group), 3) normalfasting glucose (� 5.6 mM) and normal glucose tolerance(� 7.8 mM 2 hours after oral glucose administration), 4) nodietary restrictions, 5) no treatment with drugs that mayalter glucose tolerance, lipid metabolism, or blood pressure,and 6) no known acute or chronic disease other than obesitybased on history and physical examination. Thus, 18 appar-ently healthy men, 6 lean and 12 obese without any treat-ment, were included in the study. Their clinical character-istics are shown in Table 1. An extended set of volunteers(31 subjects: 15 lean and 16 obese) were analyzed withreal-time polymerase chain reaction (PCR), and their clini-cal characteristics did not differ from the subset used for theprotein array.

To identify obese subjects with normal glucose tolerance,each individual underwent a standard oral glucose tolerance

Table 1. Clinical and laboratory characteristics of lean and obese subjects

Lean Obese p

N 6 12Age (yrs) 28.7 � 1 26.8 � 2 0.543BMI (kg/m2) 21.5 � 0.5 33.5 � 0.9 �0.001Height (cm) 183 � 4 181 � 3 0.711Waist circumference (cm) 81 � 1.7 110.8 � 2.5 �0.001Systolic blood pressure (mm Hg) 118 � 3 129 � 3 0.038Diastolic blood pressure (mm Hg) 69 � 4 71 � 2 0.573Fasting plasma glucose (mM) 4.7 � 0.1 5.2 � 0.1 0.003Plasma glucose in 120 minutes OGTT (mM) 6.1 � 0.5 6.5 � 0.2 0.441Fasting plasma insulin (�U/mL) 2.5 � 0.4 10.3 � 1.5 0.002Ins AUCOGTT (�U � mL�1 � min) 3542 � 448 7410 � 707 0.002HOMA-IR 0.91 � 0.1 2.9 � 0.4 0.005M/I (mg � kg�1 � min�1/�U � mL�1) 0.16 � 0.02 0.05 � 0.01 �0.001TG (mM) 0.9 � 0.2 1.6 � 0.2 0.057HDL (mM) 1.2 � 0.1 0.9 � 0.1 0.027hsCRP (ng/L) 0.3 � 0.1 2.6 � 0.4 0.001

OGTT, oral glucose tolerance test; AUC, area under the curve; HOMA-IR, homeostasis model assessment of insulin resistance; M/I, wholebody insulin sensitivity; TG, triglyceride; HDL, high-density lipoprotein; hsCRP, high sensitivity C-reactive protein.

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

OBESITY Vol. 15 No. 10 October 2007 2397

test (OGTT). Whole-body insulin sensitivity was also mea-sured in each subject, using the euglycemic hyperinsuline-mic clamp (EHC) technique. The two investigations, OGTTand EHC, were carried out at least 2 days apart to eliminatepotential effects of these interventions on metabolic vari-ables. A needle biopsy of subcutaneous adipose tissue wastaken before the OGTT.

OGTTAfter an overnight fast, an indwelling catheter (Surflo-W

Terumo, Belgium) was placed into an antecubital vein forblood sampling. Blood samples were drawn before (0 min-utes) and after (30, 60, 90, and 120 minutes) ingestion of75 g glucose for determination of plasma glucose and insu-lin levels. In addition, an estimate of insulin sensitivity wasobtained by the homeostasis model assessment (HOMA)score as calculated using the formula of Matthews et al.(10): fasting plasma insulin (�U/mL) � glucose (mM)/22.5.

Whole-Body Insulin SensitivityMeasurements of whole-body insulin sensitivity were

carried out with aid of the EHC technique after an overnightfast (11). Two indwelling catheters (Surflo-W Terumo)were inserted in antecubital veins, one for infusion of insu-lin and glucose and the other to obtain venous blood formeasurement of glucose concentrations. Regular humaninsulin (Actrapid; Novo Nordisk, Denmark) was infused ina primed-continuous fashion. The rate of the continuousinsulin infusion was 1 mU per kg body weight per min for3 hours. Euglycemia was maintained by adjusting the rate ofa 20% glucose infusion based on plasma glucose measure-ments from venous blood every 5 minutes. Whole bodyinsulin sensitivity (M/I) was determined from the glucoseinfusion rate (M) required to maintain euglycemia between30 and 180 minutes and the steady state insulin levels (I)during this period.

Analytical ProceduresPlasma glucose concentrations were measured with the

glucose oxidase method (Hitachi 911, Hitachinaka, Japan).Plasma insulin concentrations were measured by the IRMAmethod (Immunotech, Marseille, France). Serum high-den-sity lipoprotein cholesterol and triglyceride concentrationswere measured with enzymatic kits from Roche Diagnosticsusing an autoanalyzer (Roche Diagnostics Hitachi 911; Hi-tachi, Tokyo, Japan). Serum CRP concentrations were mea-sured by an immunoturbidimetric method using a highsensitivity test (Randox, U.K.).

Adipose Tissue BiopsyAbdominal subcutaneous adipose tissue was taken from

the umbilicus region by aspiration with a bioptic needle(Medin, Nove Mesto n. Morave, Czech Republic) underlocal subcutaneous anesthesia (1% Mesokain; Leciva, Pra-

gue, Czech Republic) after an overnight fast before theOGTT. The sample was quickly washed in saline to elimi-nate blood and connective tissue, immediately snap frozenin liquid nitrogen, and stored at �80 °C until analysis.

Protein ArrayProtein levels in subcutaneous adipose tissue were mea-

sured using the RayBioHuman Cytokine Antibody Arrays Cseries 1000.1 (RayBiotech) in 18 subjects (6 lean and 12obese) using one array per tissue sample (total of 18 arrays).The full list of measured proteins is shown in Table 2 andcan also be found online at http://www.raybiotech.com/map/C_Series_1000.pdf. Tissue lysates were prepared from70 to 150 mg of powdered subcutaneous adipose tissueusing cell lysis buffer (RayBiotech) with addition of theprotease inhibitor cocktail Complete (Roche, Switzerland).After 2 hours of lysis at 4 °C and centrifugation (10,000g,15 minutes, 4 °C), the protein content in the supernatant wasmeasured using the BioRad Protein Assay (BioRad); 200�g of proteins was used for analysis. The protein arrayswere processed as recommended by the producer. The arraysensitivity data are available at http://www.raybiotech.com/human_array_sensitivity.pdf.

The intensities of signals were quantified by densitometrywith the aid of the LumiImager device (Roche, Switzerland)using the LumiAnalyst software (Roche, Switzerland). Theresults were normalized to an internal positive control pro-vided on each membrane. The relative expression levels(target gene signal/positive control signal) were used tocompare the subjects. Proteins of interest were alwayspresent in duplicates, and positive controls (n � 8) were intwo different locations of the array. Due to the obviouslimitations of the protein array technology, the most prom-inent results were confirmed by real-time PCR.

Total RNA PreparationPowdered frozen tissue samples (200 mg) were homog-

enized using Ultra-Turax T8 homogenizator (IKA,Labortechnik, Germany) in 4 mL QIAzol Lysis Reagent(Qiagen, Hilden, Germany). Total RNA was isolated usingan RNeasy lipid tissue mini kit (Qiagen) according to themanufacturer’s instructions, including a DNase treatmentstep (RNase-free DNase Set, Qiagen). RNA concentrationand purity were measured spectrophotometrically, and RNAsample integrity was verified on agarose gels (2%) stainedwith ethidium bromide. Isolated RNA was stored at �80 °Cuntil quantification of target mRNAs.

Quantification of Relative mRNA ConcentrationsGene expression was measured by real-time reverse tran-

scription (RT)-PCR. First, 1 �g of total RNA was used forreverse transcription to synthesize the first-strand cDNAusing the GeneAmp RNA PCR kit (Applied Biosystems)containing MuLV reverse transcriptase and random hexam-

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

2398 OBESITY Vol. 15 No. 10 October 2007

Tab

le2.

Alp

habe

tical

list

ofcy

toki

nes

onth

eR

ayB

ioH

uman

Cyt

okin

eA

ntib

ody

Arr

ayC

seri

es10

00

Abb

revi

atio

nF

ull

nam

e/ot

her

nam

esA

bbre

viat

ion

Ful

lna

me/

othe

rna

mes

Abb

revi

atio

nF

ull

nam

e/ot

her

nam

es

Acr

p30

Adi

pone

ctin

HG

FH

emat

opoi

etic

grow

thfa

ctor

s,he

pato

cyte

grow

thfa

ctor

MC

P-4

Mon

ocyt

ech

emoa

ttrac

tant

P4

AgR

PA

gout

i-re

late

dpr

otei

nM

-CSF

Mac

roph

age-

colo

nyst

imul

atin

gfa

ctor

Am

phir

egul

inA

mph

ireg

ulin

I-30

9I-

309

AN

GA

ngio

geni

nIC

AM

-1In

terc

ellu

lar

adhe

sion

mol

ecul

e1

MD

CM

acro

phag

e-de

rive

dch

emok

ine

Ang

iopo

ietin

-1A

ngio

poie

tin-1

ICA

M-3

Inte

rcel

lula

rad

hesi

onm

olec

ule

3A

XL

Apr

otei

nty

rosi

neki

nase

(ark

)IF

N-g

amm

aIn

terf

eron

gam

ma

MIF

Mac

roph

age

mig

ratio

nB

DN

FB

rain

-der

ived

neur

otro

phic

fact

orIG

FBP-

1In

sulin

-lik

egr

owth

fact

orbi

ndin

gpr

otei

n1

Inhi

bito

ryfa

ctor

bFG

FB

asic

fibr

obla

stgr

owth

fact

orM

IGM

onok

ine

indu

ced

byga

mm

ain

terf

eron

BL

CB

-lym

phoc

yte

chem

oattr

acta

ntIG

FBP-

2In

sulin

-lik

egr

owth

fact

orbi

ndin

gpr

otei

n2

BM

P-4

Bon

em

orph

ogen

etic

prot

ein

4M

IP-1

�M

acro

phag

ein

flam

mat

ory

P-1�

BM

P-6

Bon

em

orph

ogen

etic

prot

ein

6IG

FBP-

3In

sulin

-lik

egr

owth

fact

orbi

ndin

gpr

otei

n3

MIP

-1�

Mac

roph

age

infl

amm

ator

yP-

1�

b-N

GF

Ner

vegr

owth

fact

or-b

eta

MIP

-1�

Mac

roph

age

infl

amm

ator

yP-

1�B

TC

Bet

acel

lulin

IGFB

P-4

Insu

lin-l

ike

grow

thfa

ctor

bind

ing

prot

ein

4M

IP-3

�M

acro

phag

ein

flam

mat

ory

P-3�

CC

L28

CC

K-1

MIP

-3�

Mac

roph

age

infl

amm

ator

yP-

3�C

kbe

ta8–

1C

hem

okin

e-be

ta-8

IGFB

P-6

Insu

lin-l

ike

grow

thfa

ctor

bind

ing

prot

ein

6M

SP-a

Mac

roph

age

stim

ulat

ing

prot

ein

a-ch

ain

CN

TF

Cili

ary

neur

onot

roph

icfa

ctor

CT

AC

KC

TA

C(c

utan

eous

T-c

ell

attr

actin

gch

emok

ine)

;sk

inki

neIG

F-I

Insu

lin-l

ike

grow

thfa

ctor

-1N

AP-

2N

eutr

ophi

lac

tivat

ing

pept

ide

2

IGF-

ISR

Insu

lin-l

ike

grow

thfa

ctor

1so

lubl

ere

cept

orN

T-3

Neu

rotr

ophi

n-4

Dtk

Dev

elop

men

tal

tyro

sine

kina

seN

T-4

Neu

rotr

ophi

n-3

EG

FE

pide

rmal

grow

thfa

ctor

IL-1

aIn

terl

euki

nI

alph

aO

PGO

steo

prot

eger

inE

GF-

RE

pide

rmal

grow

thfa

ctor

rece

ptor

IL-1

bIn

terl

euki

nI

beta

OSM

Onc

osta

tinM

EN

A-7

8E

pith

elia

lne

utro

phil-

activ

atin

gpr

otei

n78

IL-1

raIn

terl

euki

n1

rece

ptor

anta

goni

stPA

RC

Pulm

onar

yan

dac

tivat

ion-

regu

late

dch

emok

ine

IL-1

R1

Inte

rleu

kin

1re

cept

or-r

elat

edpr

otei

n1

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

OBESITY Vol. 15 No. 10 October 2007 2399

Tab

le2.

Con

tinue

d

Abb

revi

atio

nF

ull

nam

e/ot

her

nam

esA

bbre

viat

ion

Ful

lna

me/

othe

rna

mes

Abb

revi

atio

nF

ull

nam

e/ot

her

nam

es

Eot

axin

Eot

axin

IL-1

R4/

ST2

Inte

rleu

kin

1re

cept

or4,

ST2

PDG

F-B

BPl

atel

et-d

eriv

edgr

owth

fact

orE

otax

in-2

MPI

F-2

(mye

loid

prog

enito

rin

hibi

tory

fact

or-2

)IL

-2In

terl

euki

n2

PlG

FPl

acen

tagr

owth

fact

or

IL-2

R�

Inte

rleu

kin

2so

lubl

ere

cept

or�

RA

NT

ES

Reg

ulat

edup

onac

tivat

ion,

norm

alT

-cel

lex

pres

sed

Eot

axin

-3M

IP-4

-alp

ha(m

acro

phag

ein

flam

mat

ory

prot

ein-

4-al

pha)

IL-3

Inte

rleu

kin

3

IL-4

Inte

rleu

kin

4SC

FSt

emce

llfa

ctor

Fas

Fas/

TN

FRSF

6/A

po-1

/CD

95IL

-5In

terl

euki

n5

SDF-

1St

rom

alce

ll-de

rive

dfa

ctor

FGF-

4Fi

brob

last

grow

thfa

ctor

-4IL

-6In

terl

euki

n6

sgp1

30So

lubl

egl

ycop

rote

in13

0FG

F-6

Fibr

obla

stgr

owth

fact

or-6

IL-6

RIn

terl

euki

n6

rece

ptor

sTN

FR

ISo

lubl

eT

NF

rece

ptor

IFG

F-7

Fibr

obla

stgr

owth

fact

or-7

IL-7

Inte

rleu

kin

7sT

NF

RII

Solu

ble

TN

Fre

cept

orII

FGF-

9Fi

brob

last

grow

thfa

ctor

-9IL

-8In

terl

euki

n8

TA

RC

Thy

mus

and

activ

atio

n-re

gula

ted

chem

okin

eFl

t-3

Lig

and

fms-

like

tyro

sine

kina

se-3

Lig

and

(ST

K-1

ligan

d)IL

-10

Inte

rleu

kin

10

IL-1

1In

terl

euki

n11

TE

CK

Thy

mus

-exp

ress

edch

emok

ine

Frac

talk

ine

FKN

orFK

IL-1

2p4

0In

terl

euki

n12

p40

GC

P-2

Gra

nulo

cyte

chem

otac

ticpr

otei

n2

IL-1

2p70

Inte

rleu

kin

12p7

0T

GF-

beta

1T

umor

necr

osis

fact

orbe

ta1

IL-1

3In

terl

euki

n13

TG

F-be

ta3

Tum

orne

cros

isfa

ctor

beta

3G

CSF

Gra

nulo

cyte

-col

ony

stim

ulat

ing

fact

orIL

-15

Inte

rleu

kin

15T

IMP-

1T

issu

ein

hibi

tor

ofm

etal

lopr

otei

nase

1IL

-16

Inte

rleu

kin

16T

IMP-

2T

issu

ein

hibi

tor

ofm

etal

lopr

otei

nase

2G

DN

FG

lial-

deri

ved

neur

otro

phic

fact

orIL

-17

Inte

rleu

kin

17T

NF-

aT

umor

necr

osis

fact

or-a

lpha

I-T

AC

Inte

rfer

on-i

nduc

ible

Tce

llT

NF-

bT

umor

necr

osis

fact

or-b

eta

GIT

RG

luco

cort

icoi

din

duce

dtu

mor

necr

osis

fact

orre

cept

orA

lpha

chem

oattr

acta

ntT

POT

hrom

bopo

ietin

Lep

tinL

eptin

TR

AIL

R3

TN

F-re

late

dap

opto

sis-

indu

cing

ligan

dre

cept

or3

GIT

R-L

igan

dG

luco

cort

icoi

din

duce

dtu

mor

necr

osis

fact

orre

cept

orL

IGH

TL

IGH

T

Lym

phot

actin

Lym

phot

actin

TR

AIL

R4

TN

F-re

late

dap

opto

sis-

indu

cing

ligan

dre

cept

or4

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

2400 OBESITY Vol. 15 No. 10 October 2007

ers. The reaction conditions were set according to the man-ufacturer’s instructions. Real-time RT-PCR reaction wasperformed in 20 �L containing 1� PCR buffer, 5.5 mMMgCl2, 0.2 mM dATP, dGTP, and dCTP, 0.4 dUTP, 0.5 UUNG, 1.25 U Taq polymerase (Fermentas), and 1 �L ofpre-designed TaqMan Gene Expression Assay (AppliedBiosystems) containing primers and FAM-labeled probe.This technique was used to determine expression of adi-ponectin (Hs00605917_m1), leptin (Hs00174877_m1),TNF-� (Hs00174128_m1), IL-6 (Hs00174131_m1), MIF(Hs00236988_m1), MIP-1a (Hs00234142_m1), TIMP-2(Hs00234278_m1), CD14 (Hs00169122_g1), and CD68(Hs00154355_m1). Expression of 18S rRNA(Hs99999901_s1) was used as an internal reference to cal-culate the relative expression of the aforementioned genes.The real-time RT-PCR was performed with the aid of theRotorGene 2000 real-time cycler (Corbett Life Science,Sydney, Australia) in universal conditions for all genes.After initial incubation at 50 °C for 2 minutes and denatur-ation at 95 °C for 5 minutes, the amplification conditionswere as follows: 55 cycles consisting of denaturation for 15seconds and a unique annealing and extension step at 60 °Cfor 75 seconds. Data were obtained as threshold cycle(Ct) values, and for each target the PCR efficiency (eff)was assessed from standard curve slope. The results areexpressed relative to 18S rRNA using the formulaeffCt 18S rRNA/eff Ct target and were normalized to a valueobtained for calibrator (mix of all of the samples) in eachrun.

Statistical AnalysesStatistical analyses were performed using SigmaStat 3.10

(Systat Software, Inc., San Jose, CA). Group differenceswere determined by Student’s t test. Spearman’s correlationcoefficients were calculated for relationships between cyto-kine expression levels and individual metabolic variables.All data are presented as means � standard error, and theminimal level of significance was set at p � 0.05.

ResultsClinical characteristics of the subjects studied are shown

in Table 1. All subjects had normal fasting glucose levelsand normal glucose tolerance (as proven by a standard 75-gOGTT). Nevertheless, the obese group (O) had significantlyhigher fasting plasma glucose levels (although still withinthe normal range) when compared with the lean group (C).Moreover, their normal glucose tolerance was achievedonly at the expense of a markedly increased insulin secre-tion after oral glucose administration (Figure 1).

During EHC, the glucose infusion rate (M) required tomaintain euglycemia during hyperinsulinemia was lower inthe obese group (O: 4.8 � 0.6, C: 9.2 � 0.6 mg � kg�1 �min�1; p � 0.001) in the presence of higher steady stateT

able

2.C

ontin

ued

Abb

revi

atio

nF

ull

nam

e/ot

her

nam

esA

bbre

viat

ion

Ful

lna

me/

othe

rna

mes

Abb

revi

atio

nF

ull

nam

e/ot

her

nam

es

GM

-CSF

Gra

nulo

cyte

-mac

roph

age

colo

nyst

imul

atin

gfa

ctor

MC

P-1

Mon

ocyt

ech

emoa

ttrac

tant

prot

ein

1uP

AR

Uro

kina

sepl

asm

inog

enac

tivat

orre

cept

orG

RO

Gro

wth

-rel

ated

onco

gene

MC

P-2

Mon

ocyt

ech

emoa

ttrac

tant

prot

ein

2V

EG

FV

ascu

lar

endo

thel

ial

grow

thfa

ctor

GR

O- �

Gro

wth

-rel

ated

onco

gene

-�H

CC

-4H

emof

iltra

teC

Cch

emok

ine

4M

CP-

3M

onoc

yte

chem

oattr

acta

ntpr

otei

n3

VE

GF-

DV

ascu

lar

endo

thel

ial

grow

thfa

ctor

-D

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

OBESITY Vol. 15 No. 10 October 2007 2401

levels of plasma insulin (I) (O: 99.0 � 7.0, C: 59.7 � 5.7�U/mL; p � 0.002). The whole-body insulin sensitivity ascorrected to steady-state insulin levels (M/I) was 3-foldlower (p � 0.001) in obese than in lean subjects (Table 1).

The subjects in the obese group had higher systolic bloodpressure and lower levels of high-density lipoprotein cho-lesterol. Serum hsCRP concentrations were �8 times higher

in the obese than in the lean group, which strongly indicatesthe presence of subclinical inflammation associated solelywith obesity in the otherwise young and healthy individuals.

Of the 120 cytokines and related proteins measured inlysates of subcutaneous adipose tissue (full list in Table 2),28 showed changed levels in obese subjects in comparisonwith the lean controls (p � 0.05) (Figure 2). These proteins

Figure 1: Glycemia (A) and insulinemia (B) during OGTT. * p � 0.05; ** p � 0.01.

Figure 2: Relative levels of proteins that were changed in subcutaneous adipose tissue of obese subjects measured with protein array.* p � 0.05; ** p � 0.01.

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

2402 OBESITY Vol. 15 No. 10 October 2007

included hormones (leptin), growth factors and their recep-tors (angiopoietin-2, HGF, EGF-R, FGF-6, IGF-1sR, MSP-a), a number of chemokines (ENA-78, PARC, lymphotac-tin, HCC-4, MIP-1a, MIP-1b, MIP-3b, TECK, RANTES),cytokines and their receptors (IL-10, IL-1a, IL-1R1, IL-1R4, IL-12p70, MIF), as well as other receptors and en-zymes (Fas/Apo-1, Axl, Dtk, osteoprotegerin, TIMP-1 andTIMP-2). Interestingly, adipose tissue of both lean andobese individuals expressed comparable amounts of adi-ponectin (C: 339.7 � 22.7, O: 306.1 � 21.4 AU; p � 0.3),TNF-� (C: 0.74 � 0.13. O: 0.71 � 0.13 AU; p � 0.9), andIL-6 (C: 0.56 � 0.13, O: 0.62 � 0.11 AU; p � 0.8).

Results of the gene expression analysis for leptin, adi-ponectin, IL-6, and TNF-�, the markers of changes inadipose tissue associated with obesity and insulin sensitiv-ity, were in accordance with the results of protein arrays,indicating that elevated leptin expression might be consid-ered an early marker of isolated obesity in otherwise met-abolically healthy individuals. Gene expression analysiswas extended to the group of the adipokines (MIF, MIP-1a,TIMP-2, RANTES) with the difference in protein levels inthe adipose tissue at the highest level of significance (Table3). Three of the adipokines elevated at the protein levels(MIF, MIP-1a, TIMP-2) also showed higher gene expres-sion. The decrease of the RANTES protein in obese adiposetissue was not confirmed at the level of gene expression byreal-time PCR.

DiscussionThe volunteer recruitment criteria were set to identify a

group of young, healthy, obese men without diabetes or

pre-diabetes who were compared with a group of young,healthy, lean individuals. None of the subjects was on anykind of medical treatment or under dietary restrictions.Therefore, results of this study provide information aboutvery early abnormalities in the adipose tissue proteomicprofile and link obesity to inflammation and early signs ofinsulin resistance in obese, although yet clinically healthy,subjects.

Increased levels of hsCRP in the obese group stronglyindicate the presence of subclinical inflammation associatedsolely with obesity in the otherwise young and healthyindividuals. Indeed, a broad spectrum of literature datademonstrates that an excess of fat tissue in the body triggersthe onset of subclinical inflammation and cellular stress andthat both situations interfere with insulin signaling, leadingthen to insulin resistance (4,12). Therefore, we focused oncytokine levels in adipose tissue to compare their abundancein the above described cohort of healthy obese and leansubjects.

The proteins, which were differentially expressed in sub-cutaneous adipose tissue of obese subjects, are involved ina wide range of biological processes (e.g., in enhancementor suppression of inflammation and recruitment of immunecells, angiogenesis, apoptosis, cell growth, and energy ho-meostasis) (Figure 3). Some of these proteins are wellknown to have changed levels in obesity. Nevertheless, tothe best of our knowledge, many of them (angiopoietin-2,EGF-R, FGF-6, IGF-1sR, MSP-a, ENA-78, PARC, lym-photactin, HCC-4, MIP-1b, MIP-3b, TECK, RANTES, IL-10, IL-1a, IL-1R1, IL-1R4, IL-12, Fas/Apo-1, Axl, Dtk,osteoprotegerin, TIMP-2) have not yet been studied in hu-man adipose tissue in the context of obesity.

A number of pro-inflammatory cytokines and chemo-kines that we have found to be more abundant in thesubcutaneous adipose tissue of obese subjects (Figure 3)confirm the presence of an inflammatory state. Surprisingly,levels of the chemokine RANTES were decreased in obesesubjects. As this protein is a chemoattractant and its circu-lating levels have been described to be increased in type 2diabetes (13), we would rather have expected its elevation inthe obese state. However, this decrease was not confirmedat the mRNA level. Thus, more experiments would benecessary before drawing any conclusion.

Nevertheless, the majority of the changes in chemokinelevels implicate a recruitment of different types of immunecells into the adipose tissue. Indeed, we measured higherexpression of the macrophage markers CD68 and CD14 inadipose tissue of the obese subjects (Table 3). Expression ofmany of the proteins found in this study to be increased inthe adipose tissue of obese subjects has not yet been studiedin isolated cell fractions. Thus, it is possible that a part ofthe variability in their content in adipose tissue could also bedue to variability in the degree of macrophage infiltration.Indeed, only adiponectin and leptin were shown to be pri-

Table 3. Results of the gene expression analysis ofthe subcutaneous adipose tissue of obese and leansubjects

Lean(n � 15)

Obese(n � 16) p

Leptin 0.48 � 0.13 1.11 � 0.22 0.021MIF 1.05 � 0.11 1.44 � 0.14 0.038MIP-1a 1.97 � 0.59 14.22 � 3.17 �0.001TIMP-2 1.21 � 0.14 1.65 � 0.11 0.019RANTES 1.64 � 0.28 1.45 � 0.21 0.582Adiponectin 1.14 � 0.21 1.75 � 0.14 0.131TNF-� 1.47 � 0.25 1.65 � 0.18 0.572IL-6 0.56 � 0.16 0.88 � 0.22 0.260CD68 0.61 � 0.12 1.67 � 0.22 �0.001CD14 0.88 � 0.15 1.92 � 0.31 �0.001

TNF-�, tumor necrosis factor-alpha; IL-6, interleukin 6. Data areexpressed by the relative value of gene expression calculated as aratio of target gene and 18S rRNA.

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

OBESITY Vol. 15 No. 10 October 2007 2403

marily secreted by adipocytes; thus, the majority of releasedinflammatory interleukins and other cytokines originate inthe non-fat cells of adipose tissue (14–16). Cytokines andchemokines produced by macrophages decrease insulin sen-sitivity (4,5), stimulate further cytokine and chemokineproduction in adipocytes, and inhibit formation of matureadipocytes by downregulating gene expression of adipo-genic and lipogenic markers (17). Therefore, it is conceiv-able that inflammation provoked by obesity is a process bywhich the body defends itself against an excess of adiposetissue (4,17). On the other hand, changes in the anti-inflam-matory proteins (IL-10, MPS-a) indicate that the possibledefensive reaction of fat tissue to inflammation is present inyoung, healthy but obese individuals.

Angiogenesis is a process needed for growth of adiposetissue. Indeed, we found that levels of several growth fac-tors and enzymes known to modulate angiogenesis wereincreased in the adipose tissue of the obese subjects (Figure3). Interestingly, growth factors, in general, are mitogensinvolved in a variety of biological processes, includingmodulation of adipogenesis (18), or other processes that arelinked to obesity, e.g., atherosclerosis (19) or carcinogenesis(20).

We also found differences in receptors modulating apo-ptosis (Figure 3). Fas/Apo-1 is a well-known receptor in-ducing apoptosis. It has been suggested that the chronicinflammation present in obesity might predispose obeseindividuals to cancer due to the overexpression of Fas/Apo-1 receptor in leukocytes (21). But overexpression ofFas/Apo-1 in adipocytes would present another mechanism

for limiting adipose tissue expansion in obese individuals byapoptosis. In accordance with this, Tchoukalova et al. (22)have described that preadipocytes of upper-body-obesewomen were more susceptible to apoptotic stimuli thanpreadipocytes of lower-body-obese women. Therefore, itwould be very interesting to determine the cell type over-expressing Fas/Apo-1 in adipose tissue. Signaling throughFas/Apo-1 can also stimulate the inflammatory pathways(23) and contribute to the inflammatory state.

Axl and Dtk are receptor tyrosine kinases of the samereceptor family. They have anti-apoptotic properties (24)and could, thus, be another link to increased carcinogenesisin obesity. But they are involved also in regulation ofadipogenesis. Mice deficient for Gas-6, a ligand for thisfamily of receptor tyrosine kinases, are resistant to high-fatdiet-induced obesity (25), and overexpression of Axl inmyeloid lineages in mice has been shown to result in obe-sity, hyperglycemia, hyperinsulinemia, and insulin resis-tance (26).

Interestingly, protein levels of several adipokines, whichwere shown to be different in obesity, correlated signifi-cantly also with the phenotypes of insulin sensitivity andinflammation (Table 4). This indicates that some of thesephenotypes could be associated with the regulation of adi-pokine expression. However, establishing the independentcontribution of each relevant variable [BMI, waist circum-ference, HOMA of insulin resistance, glucose infusion ratevalue, fasting plasma glucose/insulin level (within the phys-iological range), and hsCRP] to the adipokine expression

Figure 3: Schematic view of increased production (decreased for RANTES) of cytokines and other proteins as measured in subcutaneousfat of obese subjects and classified by their functions.

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

2404 OBESITY Vol. 15 No. 10 October 2007

was precluded by the small sample size and by the highlevel of multicolinearity among the aforementioned pheno-types.

In summary, to our knowledge, this is the first informationon complex changes of adipose tissue adipokines, which occurin young, obese, insulin-resistant, but clinically healthy, maleswith normal glucose tolerance. These proteins are likely to beengaged in a wide range of biological processes involved in theinterplay of obesity, subclinical inflammation, and regulationof glucose metabolism. To solve this puzzle, further studieswith careful clarification of the exact cellular origin within theadipose tissue, as well as assessment of biological activity ofthese proteins, will be necessary.

AcknowledgmentsThis work was supported by the APVV Grant 51-040602,

the EU COST Action B17, and a research grant from theSlovak Diabetes Association. The authors thank Iveta Wac-zulıkova for help with the statistical data analysis and AlicaMitkova and Brigita Kramplova for skillful technical assis-tance.

References1. Rajala MW, Scherer PE. Minireview. The adipocyte: at the

crossroads of energy homeostasis, inflammation, and athero-sclerosis. Endocrinology. 2003;144:3765–73.

2. Weisberg SP, McCann D, Desai M, Rosenbaum M, Leibel

Table 4. Correlation analysis between the protein levels in subcutaneous adipose tissue measured with proteinarray and parameters related to obesity insulin sensitivity and inflammation

Protein BMI Waist HOMA M FPG FPI hsCRP

Leptin 0.625† 0.561* 0.650†IL-1a 0.510* 0.722‡ 0.529*IL-1 R1 0.487* 0.557* 0.501* 0.668† 0.726‡IL-10 0.485* 0.718‡ 0.515* 0.711‡ 0.738‡IL-12 p70 0.494*MIF 0.499* 0.629† 0.598*ENA-78 0.494* 0.503* �0.488* 0.627† 0.484* 0.510*PARC 0.520* 0.474* 0.472* 0.552* 0.490*Lymphotactin 0.571* 0.618† 0.658† 0.784‡HCC-4 0.513* 0.538* 0.733‡ 0.519* 0.559*MIP-1a 0.571* 0.523* 0.792‡ 0.684†MIP-1b 0.557* 0.521* 0.673† 0.538* 0.630†MIP-3b 0.547* 0.727‡ 0.705†TECK 0.505* 0.542* 0.610† 0.645†HGF 0.504* 0.470* 0.634† 0.472* 0.581*EGF-R 0.538* 0.480*FGF-6 0.642†IGF-1 SR 0.465* 0.624† 0.539*MSP-a 0.506* 0.527* 0.620†Fas/TNFRSF6 0.487* 0.467* 0.588*Axl 0.563* 0.590† 0.644† �0.500* 0.634† 0.524* 0.590*Dtk 0.583* 0.610† 0.545* �0.483* 0.745‡ 0.533* 0.676†Osteoprotegerin 0.496* 0.549* 0.531* �0.492* 0.574* 0.544* 0.586*TIMP-1 0.474* 0.597†TIMP-2 0.535* 0.521* 0.565* 0.573* 0.608†RANTES �0.480*

HOMA, homeostasis model assessment; M, glucose infusion rate; FPG, fasting plasma glucose; FPI, fasting plasma insulin; hsCRP, highsensitivity C-reactive protein. R values indicate potency of the association (n � 18).* Significant (p � 0.05).† Significant (p � 0.01).‡ Significant (p � 0.001).

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

OBESITY Vol. 15 No. 10 October 2007 2405

RL, Ferrante AW Jr. Obesity is associated with macrophageaccumulation in adipose tissue. J Clin Invest. 2003;112:1796–808.

3. Xu H, Barnes GT, Yang Q, et al. Chronic inflammation infat plays a crucial role in the development of obesity-relatedinsulin resistance. J Clin Invest. 2003;112:1821–30.

4. Wellen KE, Hotamisligil GS. Inflammation, stress, and dia-betes. J Clin Invest. 2005;115:1111–9.

5. Kolb H, Mandrup-Poulsen T. An immune origin of type 2diabetes? Diabetologia. 2005;48:1038–50.

6. Kershaw EE, Flier JS. Adipose tissue as an endocrine organ.J Clin Endocrinol Metab. 2004;89:2548–56.

7. Scherer PE. Adipose tissue: from lipid storage compartmentto endocrine organ. Diabetes. 2006;55:1537–45.

8. Guerre-Millo M. Adipose tissue and adipokines: for better orworse. Diabetes Metab. 2004;30:13–9.

9. No authors listed. World Medical Association Declaration ofHelsinki: ethical principles for medical research involvinghuman subjects. JAMA. 2000;284:3043–5.

10. Matthews DR, Hosker JP, Rudenski AS, Naylor BA,Treacher DF, Turner RC. Homeostasis model assessment:insulin resistance and beta-cell function from fasting plasmaglucose and insulin concentrations in man. Diabetologia.1985;28:412–9.

11. DeFronzo RA, Tobin JD, Andres R. Glucose clamptechnique: a method for quantifying insulin secretion andresistance. Am J Physiol. 1979;237:E214–23.

12. Wellen KE, Hotamisligil GS. Obesity-induced inflammatorychanges in adipose tissue. J Clin Invest. 2003;112:1785–8.

13. Herder C, Haastert B, Muller-Scholze S, et al. Associationof systemic chemokine concentrations with impaired glucosetolerance and type 2 diabetes: results from the CooperativeHealth Research in the Region of Augsburg Survey S4(KORA S4). Diabetes. 2005;54(Suppl 2):11–7.

14. Curat CA, Wegner V, Sengenes C, et al. Macrophages inhuman visceral adipose tissue: increased accumulation in obe-sity and a source of resistin and visfatin. Diabetologia. 2006;49:744–7.

15. Fain JN, Madan AK, Hiler ML, Cheema P, Bahouth SW.Comparison of the release of adipokines by adipose tissue,

adipose tissue matrix, and adipocytes from visceral and sub-cutaneous abdominal adipose tissues of obese humans. Endo-crinology. 2004;145:2273–82.

16. Fain JN. Release of interleukins and other inflammatory cyto-kines by human adipose tissue is enhanced in obesity and pri-marily due to the nonfat cells. Vitam Horm. 2006;74:443–77.

17. Lacasa D, Taleb S, Keophiphath M, Miranville A, ClementK. Macrophage-secreted factors impair human adipogenesis:involvement of proinflammatory state in preadipocytes. En-docrinology. 2006;148:868–77.

18. Adachi H, Kurachi H, Homma H, et al. Epidermal growthfactor promotes adipogenesis of 3T3–L1 cell in vitro. Endo-crinology. 1994;135:1824–30.

19. Lamb DJ, Modjtahedi H, Plant NJ, Ferns GA. EGF medi-ates monocyte chemotaxis and macrophage proliferation andEGF receptor is expressed in atherosclerotic plaques. Athero-sclerosis. 2004;176:21–6.

20. Renehan AG, Frystyk J, Flyvbjerg A. Obesity and cancerrisk: the role of the insulin-IGF axis. Trends EndocrinolMetab. 2006;17:328–36.

21. Cottam DR, Schaefer PA, Shaftan GW, Angus LD. Dysfunc-tional immune-privilege in morbid obesity: implications and ef-fect of gastric bypass surgery. Obes Surg. 2003;13:49–57.

22. Tchoukalova Y, Koutsari C, Jensen M. Committed subcu-taneous preadipocytes are reduced in human obesity. Diabe-tologia. 2006;50:151–7.

23. Wajant H, Pfizenmaier K, Scheurich P. Non-apoptotic Fassignaling. Cytokine Growth Factor Rev. 2003;14:53–66.

24. Hafizi S, Dahlback B. Signalling and functional diversitywithin the Axl subfamily of receptor tyrosine kinases. Cyto-kine Growth Factor Rev. 2006;17:295–304.

25. Maquoi E, Voros G, Carmeliet P, Collen D, Lijnen HR.Role of Gas-6 in adipogenesis and nutritionally induced adi-pose tissue development in mice. Arterioscler Thromb VascBiol. 2005;25:1002–7.

26. Augustine KA, Rossi RM, Van G, et al. Noninsulin-dependent diabetes mellitus occurs in mice ectopically ex-pressing the human Axl tyrosine kinase receptor. J CellPhysiol. 1999;181:433–47.

Subcutaneous Adipose Tissue in Obesity, Skopkova et al.

2406 OBESITY Vol. 15 No. 10 October 2007