Báo cáo khoa học: AdipoR2 is transcriptionally regulated by ER stress-inducible ATF3 in HepG2 human hepatocyte cells pdf

We found that the hepatic expression of adiponectin receptor 2 AdipoR2 was lower, but the expression of markers of the ER stress pathway, 78 kDa glucose-regu-lated protein GRP78 and acti

stress-inducible ATF3 in HepG2 human hepatocyte cells In-uk Koh1,2, Joo H Lim1, Myung K Joe1, Won H Kim1, Myeong H Jung3, Jong B Yoon2and Jihyun Song1

1 Division of Metabolic Disease, Department of Biomedical Science, National Institutes of Health, Seoul, South Korea

2 Department of Biochemistry, College of Science, Yonsei University, Seodaemoon-Gu, Seoul, South Korea

3 School of Korean Medicine, Pusan National University, Yangsan-si, Gyeongnam, South Korea

Introduction

Obesity and/or obesity-linked insulin resistance, one of

the key features of type 2 diabetes, are regarded as risk

factors for metabolic syndrome and atherosclerosis [1]

Adiponectin, which is abundantly expressed in adipose

tissue, is a circulating peptide hormone with direct insulin-sensitizing activity This adipokine has amelio-rative effects on insulin resistance in peripheral tissues, and plays a central role in the regulation of energy

Keywords

AdipoR2; ATF3; ER stress; insulin

resistance; obesity

Correspondence

M H Jung, School of Korean Medicine,

Pusan National University, 30 Beom-eo ri,

Mulguem-eup, Yangsan-si, Gyeongnam

609-735, South Korea

Fax: +82 51 510 8437

Tel: +82 51 510 8468

E-mail: jung0603@pusan.ac.kr

J B Yoon, Department of Biochemistry,

College of Science, Yonsei University, 134

Shinchon-Dong, Seodaemoon-Gu, Seoul

120-749, South Korea

Fax: +82 2 392 3488

Tel: +82 2 2123 2704

E-mail: yoonj@yonsei.ac.kr

J Song, Division of Metabolic Disease,

Department of Biomedical Science, National

Institutes of Health, 5 Nokbun-dong,

Eunpyung-gu, Seoul 122-701, South Korea

Fax: +82 2 354 1057

Tel: +82 2 380 1530

E-mail: jhsong10@korea.kr

(Received 22 September 2009, revised 22

February 2010, accepted 9 March 2010)

doi:10.1111/j.1742-4658.2010.07646.x

Adiponectin acts as an insulin-sensitizing adipokine that protects against obesity-linked metabolic disease, which is generally associated with endo-plasmic reticulum (ER) stress The physiological effects of adiponectin on energy metabolism in the liver are mediated by its receptors We found that the hepatic expression of adiponectin receptor 2 (AdipoR2) was lower, but the expression of markers of the ER stress pathway, 78 kDa glucose-regu-lated protein (GRP78) and activating transcription factor 3 (ATF3), was higher in the liver of ob/ob mice compared with control mice To investi-gate the regulation of AdipoR2 by ER stress, we added thapsigargin, an

ER stress inducer, to a human hepatocyte cell line, HepG2 Addition of the ER stress inducer increased the levels of GRP78 and ATF3, and decreased that of AdipoR2, whereas addition of a chemical chaperone, 4-phenyl butyric acid (PBA), could reverse them Up- or down-regulation of ATF3 modulated the AdipoR2 protein levels and AdipoR2 promoter activ-ities Reporter gene assays using a series of 5¢-deleted AdipoR2 promoter constructs revealed the location of the repressor element responding to ER stress and ATF3 In addition, using electrophoretic mobility shift and chro-matin immunoprecipitation assays, we identified a region between nucleo-tides )94 and )86 of the AdipoR2 promoter that functions as a putative ATF3-binding site in vitro and in vivo Thus, our findings suggest that the

ER stress-induced decrease in both protein and RNA of AdipoR2 results from a concomitant increase in expression of ATF3, which may play a role

in the development of obesity-induced insulin resistance and related ER stress in hepatocytes

Abbreviations

AdipoR2, adiponection receptor 2; ATF3, activating transcription factor 3; GRP78, 78 kDa glucose-regulated protein; PBA, 4-phenyl butyric acid.

homeostasis [2–4] In obesity, adiponectin activity

declines as a result of decreased adiponectin expression

and/or a defect in downstream adiponectin signalling

The combined actions of genetic factors such as

single-nucleotide polymorphisms in the adiponectin gene

and environmental factors such as a high-fat diet and

sedentary lifestyle promoting obesity are thought to be

one of the mechanisms leading to insulin resistance [5]

Several drugs are known to affect adiponectin levels

in the plasma and expression in tissues Plasma

adiponectin levels increase in response to the PPAR

(peroxisome proliferator-activated receptor) agonists,

thiazolidinediones, but decrease in response to

anti-HIV drugs and the well-known endoplasmic reticulum

(ER) stressor, thapsigargin ER stress is a malfunction

of the organelle caused by the influx of immature

pro-teins and/or depletion of calcium ions [6–8], and this

perturbation of ER homeostasis occurs in diet-induced

and genetic models of obesity [9] Studies of the

regu-lation of obesity-linked insulin resistance have led to

the suggestion that ER stress plays a role in diabetic

insulin resistance [10]

ER dysfunction and the integrated stress response

could lead to abnormal activation of c-Jun N-terminal

kinase (JNK) and/or activating transcription factor 3

(ATF3), which in turn modify the lipogenic pathway,

insulin signaling and the expression levels of genes

related to insulin action, such as insulin receptor

sub-strates and adiponectin [8,9], resulting in obesity-related

insulin resistance [11–13] In an in vitro model of ER

stress induced by proteasome inhibition, the stress

induced transcription of the transcription factors

GADD153, ATF4 and ATF3, and regulation of

lipo-genic pathways by the ER stress response was also

shown in hepatocytes [14] ER stressors such as

thapsi-gargin or tunicamycin reduced insulin signaling by

ser-ine phosphorylation of insulin receptor substrate 1

[9,15] We recently demonstrated that an agent causing

ER stress activated JNK and consequently induced

ATF3, with a reduction in adiponectin transcription [8]

Nakatani et al [16] identified a molecular chaperone

that protects cells from ER stress and its effect on

insulin sensitivity in the liver The chemical chaperones

4-phenyl butyric acid (PBA) and tauroursodeoxycholic

acid, which have the ability to decrease ER stress, act

as potent anti-diabetic agents [10] Because ER is

abundant in hepatocytes, and the liver is a primary

target organ of insulin and adiponectin [3,14,17], we

focused on regulation of the adiponectin receptor

under ER stress-induced conditions in liver cells

In humans, adiponectin receptors 1 and 2 (AdipoR1

and AdipoR2, respectively) serve as receptors for

globu-lar and/or full-length adiponectin AdipoR1 is

ubiqui-tously expressed and is particularly abundant in skeletal muscle, whereas AdipoR2 is expressed primarily in liver [17] These receptors have seven transmembrane domains, and share 67% amino acid homology In con-trast to G protein-coupled receptors, their N-terminus is intracellular [5] Although the intracellular adaptor pro-tein APPL1 (adaptor propro-tein, phosphotyrosine interac-tion, PH domain and leucine zipper containing 1) has recently been proposed as a modulator of insulin action

by binding to adiponectin receptors [18], the overall mechanism of adiponectin signaling is largely unknown

As expression of adiponectin receptors, as well as adipo-nectin itself, is known to be decreased in obesity-related insulin resistance but increased by PPAR agonists [1,19– 21], both adiponectin and its receptors are regarded as potential therapeutic targets for control of obesity-linked insulin resistance [22,23] However, there are very few studies that have examined a specific agent or the mechanisms responsible for regulating adiponectin receptor expression [17,24–26]

In this study, we demonstrate that AdipoR2 is nega-tively regulated in liver cells in response to ER stress or induced expression of ATF3 In addition, by analyzing the promoter region of the AdipoR2 gene, we have identified a putative ATF3-binding site in the 5’ flanking region, suggesting that direct binding of this tran-scription factor might negatively regulate AdipoR2 expression We hypothesize that ER stress-inducible ATF3 plays an important role in regulating AdipoR2 in the liver

Results Down-regulation of AdipoR2 with up-regulation

of ATF3 expression under increased ER stress conditions

As shown in previous studies of obesity and ER stress [9,10], the expression of the ER stress marker proteins GRP78 and ATF3 was increased 1.4- and 2.0-fold, respectively, in ob/ob mice compared with lean con-trols, whereas that of AdipoR2 was decreased 0.8-fold (Fig 1A) AdipoR2 is the major hepatic receptor for adiponectin [5] ER stress-induced disruption of adipo-nectin action and increased insulin resistance in he-patocytes could be attributed to down-regulation of the AdipoR2 level, and thus we studied the effect of

ER stress on the AdipoR2 level in human hepatocytes When HepG2 cells were exposed to 1.0 lm thapsi-gargin, a well-known inducer of ER stress, for 24 h, both the ER molecular chaperone GRP78 and also ATF3, which has been shown to repress transcription

of the adiponectin gene in adipocytes [8], were induced

Interestingly, the level of AdipoR2 protein was

decreased coincidentally with the increase of ATF3

(Fig 1B and Table S1)

To determine whether the observed changes in

protein expression were caused by thapsigargin-induced

ER stress, cells were pre-incubated with 20 mm PBA for 24 h prior to thapsigargin exposure In cells pre-incubated with PBA, thapsigargin-induced increases

in GRP78 and ATF3 protein levels did not occur, and the ER stress-induced decrease in AdipoR2 level was

1.5 2

ATF3

GRP78

A

*

*

0 0.5

1

AdipoR2

Actin

2 3

4 +

PBA (20 m M )

Thap (1.0 µ M )

GRP78

adipor2 atf3

C Thap Thap/PBA

gapdh

*

* *

*

0.5 1 1.5 2

0 1 2

AdipoR2

ATF3

Actin

*

* *

*

0

+ – + +

Ad YFP

siATF3 Thap

– +

GRP78 ATF3 AdipoR2

atf3 adipor2

Control Thap Thap/PBA

– –

– – –

(MOI)

AdipoR2

ATF3

Neg.RNAi ATF3

1.5 2

2.5

* *

* *

0 0.5

ATF3 AdipoR2

Fig 1 Changes in expression of AdipoR2 under conditions of ER stress and ATF3 over-expression (A) Relative levels of GRP78, ATF3 and AdipoR2 protein in C57BL6/J and ob/ob mice (n = 3 for each group; 30 lg protein per lane) (B) Relative levels of GRP78, ATF3 and AdipoR2 protein in HepG2 cells treated with or without pre-incubation in 20 m M PBA for 24 h prior to treatment with 1.0 l M thapsigargin (30 lg pro-tein per lane; b-actin as control) (C) Relative levels of ATF3 and AdipoR2 mRNA in treated cells Semi-quantitative RT-PCR analysis was per-formed using GAPDH as the internal control and the values were normalized to control (untreated) (D) Changes in expression of AdipoR2 after infection with ATF3-expressing adenovirus HepG2 cells were infected with an adenoviral vector expressing human ATF3 (Adv-ATF3) at

a multiplicity of infection of 2–10 and incubated for 48 h HepG2 cells infected with Adv-YFP at a multiplicity of infection of 5 were used as control (E) Changes in expression of endogenous AdipoR2 upon thapsigargin-induced ER stress with or without silencing of ATF3 siATF3

or Neg.RNAi was introduced to the cells 24 h prior to treatment with 1.0 l M thapsigargin For western blot analysis, b-actin was used as a protein loading control The asterisks indicate a P value < 0.05 for the bracketed comparisons.

specifically rescued (Fig 1B, lane 3) In addition, we

measured the mRNA levels for ATF3 and AdipoR2

in HepG2 cells with or without thapsigargin or PBA

treatment, and the results showed a trend similar to the

protein level changes (Fig 1C)

We next examined the effect of ATF3

over-expres-sion in hepatocytes on changes in the AdipoR2 protein

level We introduced an adenoviral vector carrying

recombinant ATF3 (Adv-ATF3) into HepG2 cells,

and analyzed the resulting protein expression

using western blotting As expected, transduction of

Adv-ATF3 resulted in a dose-dependent increase in

the ATF3 protein level The increase in ATF3 protein

level resulted in a decrease in the AdipoR2 protein

level, but in a non-dose-dependent way (Fig 1D)

The absence of dose dependence for the reduction of

AdipoR2 may be due to cellular systemic utilization of

the proteins

In order to further investigate whether ATF3 plays

an important role in ER stress-induced

down-regula-tion of AdipoR2, we assessed the effect of knocking

down ATF3 on the AdipoR2 level Within 48 h after

introducing siRNA against ATF3 (siATF3) to HepG2

cells, ATF3 was mostly repressed, but the endogenous

AdipoR2 level was relatively increased The expected

changes in ATF3 and AdipoR2 levels as a result of

thapsigargin treatment were significantly ameliorated

by siATF3 (Fig 1E and Table S2) These changes were

not observed in cells treated with control siRNA

(Neg.RNAi) These data confirm the negative

regula-tory effect of transcription factor ATF3 on AdipoR2

levels

Localization of a repressor element in the

AdipoR2 promoter

To further investigate the changes in AdipoR2

expres-sion as a result of increased ER stress in hepatocytes,

we examined the AdipoR2 promoter activities in

HepG2 cells using the reporter gene construct

AR2P()1974), comprising nucleotides )1974 to +0

Exposure of cells transfected with AR2P()1974) to

1.0 lm thapsigargin caused an approximately 80%

repression of transcription from the promoter region

of AdipoR2 in 24 h (Fig 2A)

In addition, to assess the effect of ATF3, a known ER

stress-induced transcriptional repressor in adipocytes

[8], on AdipoR2 regulation in hepatocytes, we measured

the transcriptional activity of the AdipoR2 promoter

when AR2P()1974) was co-transfected with an

ATF3-expressing vector (ATF3/pcDNA3.1) As for

thapsigar-gin exposure (Fig 2A), ATF3 expression in HepG2 cells

down-regulated the promoter activity in a

dose-depen-dent manner (Fig 2B) Compared with Neg.RNAi treatment, silencing of ATF3 reduced the repressive effect of thapsigargin on the promoter activity of AdipoR2(Fig 2C) To investigate whether ATF3 affects AdipoR2 expression directly, in other words to locate the repressor element in the AR2P()1974) promoter region, as suggested by the above results, we analyzed the promoter activity of 5¢ serially deleted human AdipoR2 promoter constructs in pGL3-Basic vector (Fig 2D,E) Four plasmid constructs containing portions of the promoter region of various lengths were transfected into HepG2 cells with or without co-trans-fection of the ATF3-expressing vector (ATF3/ pcDNA3.1) As shown in Fig 2D, ATF3 co-transfec-tion repressed the promoter activities of the transfected AdipoR2 reporter constructs AR2P()1974), AR2P ()870) and AR2P()343) However, the activity of the shortest construct AR2P()72) was as low as that in the control (pGL3) group In another experiment, various amounts of ATF3/pcDNA3.1 (0, 0.2 and 0.4 lg) were co-transfected with AR2P()343) or AR2P()72), and significant dose-dependent repression by ATF3 was observed in cells transfected with AR2P()343) but not

in those transfected with AR2P()72) (Fig 2E) ATF3 co-transfection with this shortest construct AR2P()72) showed a tendency to decrease the reporter activity (approximately 50%) but without statistical significance (P = 0.15) (Table S3 and Fig S1)

Given that AR2P()72) was not responsive to ATF3,

we presume that more than 72 nucleotides of promoter region are required for the expression of AdipoR2, and that at least one of the repressive elements of AdipoR2is located between nucleotides)343 and )72

ATF3 binds to the AdipoR2 promoter in vitro and

in vivo

To confirm that the above results are an effect of ATF3 on AdipoR2 expression, we searched for a puta-tive repressor binding site by observing sequences with-out the aid of computer software between nucleotides )343 and )72 of the human AdipoR2 gene and using TESS analysis (http://www.cbil.upenn.edu/cgi-bin/tess/ tess) with TRANSFAC database version 6.0 (available online; http://www.gene-regulation.com) We isolated the sequence 5¢-TGCGCGTCA-3¢ located at nucleo-tides)94 to )86 (Fig 3A), which is similar to the con-sensus palindromic ATF/CRE site (TGACGTCA) to which members of the ATF3 family are known to homo- or heterodimerize for DNA binding and tran-scriptional regulation [27]

We performed electrophoretic mobility shift assays (EMSAs) using nuclear extracts from HepG2 cells and

22 bp radiolabeled DNA probes (nucleotides )79 to

)100) containing the putative ATF3-binding site

5¢-TGCGCGTCA-3¢ to determine whether ATF3

directly interacts with the AdipoR2 promoter The

EMSA results revealed that this oligonucleotide formed

a DNA–protein complex with the hepatocyte nuclear

extracts (Fig 3B, C) A specific interaction between

the putative ATF3-binding site and the repressor ATF3

was confirmed by competition with unlabeled

oligonu-cleotides (Fig 3C) and by dose-dependent inhibition

by antibody against ATF3 (Fig 3D) As a negative con-trol for binding of the bZIP (basic leucine zipper) tran-scription factor ATF3 to the putative binding site, we used probe ‘X’, containing a sequence that recruits one

of the zinc-finger DNA-binding transcription factor, also known to interact with the CREB-binding protein In the competition EMSA shown in Fig 3C, a

100 x excess of non-specific probe ‘X’ did not showed

100 120

A

*

100

100

20 40 60 80

20 40 60 80

20 40 60 80

pGL3

AR2P(–1974)

Thap

0

+ – –

+ – +

– + –

– + +

Neg.RNAi siATF3

AR2P(–1974)

0

ATF3

0

– –

+ –

+ +

+ ++

AR2P(–1974)

60 80 100 120

75 100 125

0 20 40 60

AR2P(–1974)

0 25 50 75

Luciferase activity (%) Luciferase activity (%)

(

NS

AR2P(

AR2P(–870)

AR2P(–343)

AR2P(–72)

ATF3

AR2P(–343) AR2P(–72)

ATF3

–

+

+

+

– +

Fig 2 Changes in the promoter activity of AdipoR2 under conditions of ER stress and ATF3 over-expression (A) Activity of the AdipoR2 pro-moter in HepG2 human liver cells with ER stress induction by 1.0 l M thapsigargin for 24 h The pGL3-Basic-derived reporter construct compris-ing nucleotides )1974 to +0 of the AdipoR2 promoter [AR2P(–1974)] was transfected into HepG2 cells, followed by treatment with thapsigargin (B) Activity of the AdipoR2 promoter in HepG2 cells with ATF3 over-expression by co-transfection of 0.2 or 0.4 lg of ATF3 expres-sion vector (C) Activity of the AdipoR2 promoter in HepG2 cells upon thapsigargin-induced ER stress with or without silencing of ATF3 siATF3

or Neg.RNAi was introduced to the cells 24 h prior to treatment with 1.0 l M thapsigargin Luciferase activity values were measured in triplicate and expressed as arbitrary units (D) Promoter activities of reporter gene constructs containing 0.6 lg of various lengths of 5¢ deleted fragments

of the promoter region subcloned into the pGL3-Basic plasmid vector and transfected with or without 0.4 lg of ATF3-expressing vector (E) ATF3-dose-dependent repression of the promoter activity upon co-transfection of 0, 0.2 or 0.4 lg of ATF3-expressing plasmids with 0.6 lg of reporter plasmid into HepG2 cells The asterisks indicate a P value < 0.05 for the bracketed comparisons NS, not significant.

competition in binding to ATF3, but assays using a

100 x excess of cold/unlabeled )94/)86 or the ATF/

CRE positive control probe did show competition with

tested probes containing the putative)94/)86 site,

indi-cating the specificity of this binding assay

To determine the physiological relevance of ER

stress and/or stress-related expression of ATF3 on

formation of the protein–DNA complex in vitro, we

increased the expression of ATF3 in HepG2 cells by

treatment with 1.0 lm thapsigargin or transduction

with ATF3-expressing adenovirus, Adv-ATF3

(Fig 4A,B, upper panel) More protein–DNA complex

was formed between radiolabeled oligonucleotides

containing the putative ATF3-binding site, or the

ATF/CRE consensus sequence, and nuclear extracts of

cells when the cells were thapsigargin-treated Nuclear

extracts of the cells adenovirally over-expressing ATF3

also formed more DNA–protein complex with both

the ATF/CRE consensus sequence and the )94/)86

oligonucleotide probe (Fig 4A,B)

To further investigate this interaction in vivo, we validated the predicted ATF3-binding site in the regulatory region of the human AdipoR2 gene using chromatin immunoprecipitation (ChIP) This showed specific in vivo binding of ATF3 to the putative ATF3-binding element at nucleotides )94/)86 of the promoter region of AdipoR2 (Fig 4C) In addition to the EMSA results (Figs 3 and 4A,B), showing that recruitment of ATF3 was increased by treatment with thapsigargin in a time-dependent manner, these results confirm that the transcription factor ATF3 binds to the promoter of AdipoR2 both in vitro and in vivo

Decreased responsiveness as a result of deletion

of the putative ATF3-responsive repressor element in the nucleotide)343/)72 region of the promoter

As the putative ATF3-binding site 5¢-TGCGCGTCA-3¢ from the promoter region of human AdipoR2 showed

:

NE: HepG2 cells

–94/–86 of promoter Consensus ATF/CRE

Competitor

NE: HepG2 cells

NE: HepG2 cells

Competitor + Ab

N w ATF3 Ab

*

A

Fig 3 ATF3 binds to the promoter of AdipoR2 in vitro (A) Comparison of the sequence of the EMSA probe containing the putative ATF3/ CRE-binding site (TGCGCGTCA) from the promoter of AdipoR2 with that of the palindromic consensus ATF3-binding sequence (TGACGTCA) (B–D) The putative ATF3-binding site exhibited specific binding with the HepG2 nuclear extract Nuclear extracts were prepared from HepG2 cells, and 5.0 mg of extract was used in EMSA reactions with 100 ng of radiolabeled double-stranded probe containing the putative ATF3-binding site between nucleotides )94 and )86 (B) Competition EMSAs were performed with a 10- or 100-fold excess of the unlabeled wild-type nucleotide )94/)86 probe, a 100-fold excess of the consensus ATF/CRE-binding site sequence, or a 100-fold excess of the non-specific probe (B,C) Competition assays with ATF-specific antibody (1–4 lg) were also performed (D).

binding ability in EMSA and ChIP experiments, we

generated a mutant promoter construct lacking a

22 bp fragment of the promoter between nucleotides

)94 and )86 (Fig 5A), to investigate whether this

region responds to ATF3 and ER stress and plays a

role in the transcriptional regulation of AdipoR2

The activity of this construct, AR2P()343D), was then analyzed with or without co-transfection of ATF3/pcDNA3.1 (Fig 5B) Co-transfection with ATF3 dramatically decreased the promoter activity of the wild-type promoter construct [AR2P()343)] to one-tenth that of untreated cells, but attenuated the

Thap (1.0 µ M )

ATF3

ββ-Actin

β-Actin

ATF3

Ext a a E xt. a a

NE: HepG2 cells

Ext ATF3 k E xt. ATF3 k NE: HepG2 cells

ATF/CRE –94 / –86

Adv- Mock No

Adv- Mock

ATF/CRE –94 / –86

Labeled probes Labeled probes

+ –

Exon1

( – 94/ – 86)

IgG

Input

ATF3

Input

C

Fig 4 Binding of ATF3 to the AdipoR2 promoter region increases under conditions

of ER stress and/or ATF3 over-expression

in vitro and in vivo (A) ATF3 expression was increased in HepG2 cells exposed to 1.0 l M thapsigargin for 24 h compared to control The amount of protein–DNA complex formed with the ATF/CRE consensus sequence and nucleotide )94/)86 double-stranded oligonucleotide probes was higher for thapsigargin-exposed samples.

(B) Transfection of an ATF3-expressing adenoviral vector (Adv-ATF3) resulted in over-expression of recombinant human ATF3 in HepG2 cells compared to control (Adv-YFP) Adenoviral vectors were infected

at a multiplicity of infection of 5 for each sample Nuclear extracts from cells over-expressing ATF3 from Adv-ATF3 showed increased binding affinity for both the putative ATF3-binding site and the consen-sus ATF/CRE-binding site (C) ChIP assays were performed with anti-ATF3 antibody (ATF3) or without it (IgG) PCR was used to amplify immunoprecipitated DNA fragments from HepG2 cells exposed to 1.0 l M thapsigargin for 0–12 h, showing a time-dependent increase in recruitment of ATF3

to the putative binding element as a result

of ER stress.

activity of the mutant construct without the )94/)86

putative binding element [AR2P()343D)] to only half

that of untreated cells (Fig 5B) As shown in Fig 5C,

the ER stress inducer thapsigargin had less of a

repres-sor effect on the mutant reporter construct ()56%)

than on the wild-type construct ()85%)

Discussion

Many groups have confirmed the

anti-diabetic/insulin-sensitizing effect of adiponectin, and thus plasma

adiponectin and its receptors in peripheral organs have

been proposed as therapeutic targets for the treatment

of diabetes and obesity-linked insulin resistance

[2,28,29] The action of adiponectin is known to be

transduced via regulation of AMP-activated protein

kinase (AMPK) function, and, given the report of a

putative adaptor protein that interacts with adiponectin

receptors, insulin and adiponectin signaling are now

considered to be linked in the peripheral organs of

insulin action, such as the liver and skeletal muscle

[16,30,31] Despite the fact that the action and plasma

level of adiponectin have been reported to be reduced

in diseases associated with obesity, including peripheral

insulin resistance and related ER stress cascades

[2,3,8,9], the relationship between obesity-related ER

stress and the consequent reduction in adiponectin

action is obscure We found that hepatic expression of AdipoR2 was lower but expression of the markers of the ER stress pathway, GRP78 and ATF3, was higher

in the liver of ob/ob mice compared with control mice

To determine the molecular mechanisms of this rela-tionship, we studied the regulation of human AdipoR2

in ER stress-induced hepatocytes ATF3, a member of the ATF/CREB family of transcription factors, is known to be a transcriptional repressor that is induced

by many stress signals, including ER stress [31,32], and has also been proposed to play a role in liver dysfunc-tion involving defects in glucose homeostasis [33] We and other investigators have also reported that adipo-nectin is negatively regulated by ATF3 and by the ER stress-mediated protein CHOP (C/EBP homologous protein) under obesity-related hypoxic conditions in adipocytes [8,34] In particular, in a transcriptional context, ATF3 functioned in response to thapsigargin-induced ER stress as a negative regulator of adiponec-tin expression by direct binding to the promoter [8] These reports imply that a relationship exists between the decreased transcriptional activity of AdipoR2 and subsequently-induced ATF3 in ER stress (Fig 1B)

In thapsigargin-treated hepatocytes, AdipoR2 expres-sion was inversely correlated with the induction of GRP78 and/or ATF3 by ER stress (Fig 1) Mean-while, in cells pre-treated with PBA, the rescued

Luc

(–343 bp)

Luc

(–343 bp)

AR2P(–343 Δ)

Luc Luc

(–72 bp) (–94/–86)

AR2P(–343) AR2P(–72)

120

A

120

*

40 60 80 100

40 60 80 100

0 20

0 20

AR2P(–343 Δ)

AR2P(–343) ATF3

pGL3

AR2P(–343) Thapsigargin –

ATF3 Thapsigargin –

– – – – + +

AR2P(–343 Δ)

Fig 5 Decreased responsiveness by deletion of the putative ATF3-responsive repressor element in the nucleotide )343/)72 region of the promoter (A) Sequences of the deletion mutant construct used in the luciferase reporter assay In the AR2P( )343D) reporter construct, the putative ATF3/CRE-binding sequence (nucleotides )94/)86: TGCGCGTCA) and six 5¢ and seven 3¢ flanking nucleotides are deleted (B) Reduced ATF3-induced repression of the promoter activity was observed for the deletion mutant promoter construct AR2P( )343D) lacking the putative ATF3/CRE-binding site at nucleotides )94/)86 Reporter construct derivatives (0.6 lg) were transfected into HepG2 cells with

or without 0.4 lg of ATF3-expressing vector (C) Rescued ER stress-induced repression of promoter activity for AR2P( )343D) Cells were treated with 1.0 l M thapsigargin for 24 h to induce ER stress The asterisks indicate a P value < 0.05 for the bracketed comparisons All luciferase assays were performed in triplicate, and error bars indicate the SEM of 3 or 6 experiments.

AdipoR2 level showed strong support for an ER

stress-based mechanism of AdipoR2 decrease in

human hepatocytes (Fig 1B) To determine the

mecha-nism of changes in AdipoR2 expression resulting from

ER stress and ATF3 over-expression or silencing in

the liver (Fig 1B–E), we measured the promoter

acti-vity of the AdipoR2 gene Up- or down-regulation of

ATF3 modulated the AdipoR2 promoter activity

(Fig 2B,C) By analysing the promoter region of the

AdipoR2 gene, we identified a putative ATF3-binding

sequence (Fig 3) As shown in Figs 3 and 4, the

decrease in AdipoR2 expression by ATF3 in ER stress

was mediated by this putative sequence which recruited

ATF3 in vitro and in vivo This result provides an

explanation for the role of ER stress and induced

ATF3 in obesity-linked insulin resistance through

regu-lation of adiponectin action

These transcription-repressing mechanisms of ER

stress-induced ATF3 have been shown to contribute to

the development of insulin resistance and type 2

diabe-tes Insulin receptor substrates 1 and 2 were found to be

repressed by ATF3 in myocytes [35] and pancreatic

b-cells [11], respectively The level of the insulin-sensitizing

hormone adiponectin was decreased by ATF3 in

adipo-cytes [8], and the major receptor in hepatoadipo-cytes,

Adi-poR2, was also negatively regulated The above effects

of ATF3 on insulin signaling and glucose homeostasis

involve the action of adiponectin in peripheral tissues

In particular, given that the cause and effect relationship

between adiponectin and insulin action is not fully

understood, the inappropriate actions of adiponectin in

obesity-linked insulin resistance are described as a

‘vicious cycle’ of adiponectin and insulin resistance [36]

For example, insulin receptor transgenic/knockout mice

exhibit decreased AdipoR2 levels in liver and muscle, as

well as decreased expression of the peroxisome

prolifera-tor-activated receptor gamma (PPARc) target genes of

fatty acid oxidation, showing that AdipoR2 defects are

relevant to diabetes susceptibility [37] In addition,

a decrease in levels of expression of adiponectin

recep-tors was reported to be associated with type 2 diabetes

[21], as well as reductions in plasma adiponectin levels in

various cases associated with insulin resistance [21, 38,

39] and alterations in the adiponectin gene [40–42]

On the other hand, despite decreased responsiveness

to thapsigargin and induced ATF3, a mutant AdipoR2

reporter construct lacking the putative ATF3-binding

site still showed repression of transcriptional activity

to some extent In addition, absence of the putative

ATF3-binding site reduced expression of the reporter

gene itself (Fig 5B,C) Co-transfection of ATF3

reduced the promoter activity of wild-type AdipoR2

dose-dependently, and mutant AdipoR2 to a lesser

degree (Fig S2 and Table S4), but co-transfection of ATF3 had a non-specific effect on the activity of the pGL3-basic control vector (Fig S3 and Table S5) Thus the decrease in promoter activity itself (Table S4) and the smaller but remaining responsiveness to ATF3 for the mutant reporter gene suggests that, in addition

to the ATF3-binding site ()94/)86), an indirect effect

of ATF3 on the promoter region of AdipoR2 may exist through an unidentified binding site In addition, this putative ATF3-binding site could recruit the transcription factor complex for dichotomous actions, possibly through the action of uncharacterized dimer-ization partner(s) of ATF3, such as ATF2, c-Jun, JunB, JunD, etc [43] Given the nature of the deleted

‘semi-palindromic’ sequence )94/)86 (TGCGCGTCA) and bZip transcription factors including ATF3 [27], these dimerization partner(s) of ATF3 may have very complicated transcription factor/co-factor relation-ships These possibilities must be studied further to clarify the ATF3-mediated negative effect on transcrip-tion of AdipoR2 under ER stress in the liver

In this study, exposure to the ER stress inducer thapsigargin and the accompanying induction of ATF3 were inversely correlated with changes in the expression level of AdipoR2 in human HepG2 cells, and this corre-lation was the result of direct transcriptional regucorre-lation

of AdipoR2 by the repressor ATF3 via the putative bind-ing site between nucleotides )94 and )86 of the pro-moter region This finding of decreased AdipoR2 levels

as a result of the regulation by ATF3 is noteworthy, and suggests that obesity-related ER stress may affect the development of hepatic insulin resistance, at least in part

by transcriptional repressing activity of ATF3

Experimental procedures Animals and materials

To compare the expression levels of ER stress markers and the adiponectin receptor in animals of various genetic back-grounds, ob/ob mice and age-matched lean control C57BL6/J mice (10 weeks, three mice per group) were purchased from the Jackson Laboratory (Bar Harbor, ME, USA) After overnight fasting, the mice were killed and liver was collected for further analysis The Animal Care and Use Committee of the National Institutes of Health and the Korean Food and Drug Administration approved all animal protocols The expression plasmid encoding ATF3 was kindly provided by Dr T Hai (Department of Molecular and Cellular Biochemistry, Ohio State Univer-sity, Columbus, OH, USA) Rabbit polyclonal antibodies against GRP78, ATF3 and AdipoR2 (sc-13968, sc-188 and sc-46754, respectively) and siRNA for ATF3 (sc-29758)

were purchased from Santa Cruz Biotechnology Inc (Santa

Cruz, CA, USA)

Cell culture and treatments

Human hepatocyte HepG2 cells and human embryonic

kidney HEK 293 cells (both American Type Culture

Collec-tion, Manassas, VA, USA) were cultured in Dulbecco’s

modified Eagle’s medium containing 4.5 gÆL)1glucose

(Invi-trogen, Carlsbad, CA, USA) and supplemented with 10%

fetal bovine serum (GibcoBRL, Gaithersburg, MD, USA)

To investigate the effect of ER stress, cells were treated

with 1.0 lm thapsigargin (Sigma, St Louis, MO, USA) in

Dulbecco’s modified Eagle’s medium supplemented with

10% fetal bovine serum for 24 h To reduce the effect of

ER stress, cells were pre-incubated for 24 h in culture

med-ium containing 20 mm 4-phenyl butyric acid (PBA)

(Calbio-chem, San Diego, CA, USA) prior to treatment with

1.0 lm thapsigargin

Over-expression of adenoviral ATF3

After PCR amplification, the ATF3 gene was ligated into

the adenovirus shuttle vector pShuttle-CMV (Stratagene,

La Jolla, CA, USA), which includes GFP (green fluorescent

protein) tagged to the C-terminus of the ATF3 protein

Recombinant adenoviral genomes were produced by

recom-bination between the shuttle vector constructed above and

the pAdEasy vector (Stratagene), according to the

manufac-turer’s protocol [44] The genomes were subsequently

trans-fected into HEK 293 cells using Lipofectamine reagent

(Invitrogen) ATF3-expressing adenovirus particles

(Adv-ATF3) were obtained as a viral mixture in culture medium

7–9 days after transfection, with the viral particle number

of the adenoviral mixture ranging between 1.0 and

2.0· 1010

IFU (inclusion-forming units)ÆmL)1 depending

on the sample The recombinant virus was propagated in

HEK 293 cells before transduction into HepG2 cells

Con-trol adenovirus (mock, Adv-YFP) was generated by the

same method using an empty adenoviral shuttle plasmid

HepG2 cells were infected with the adenoviral mixture at a

multiplicity of infection between 2 and 10 for

over-expres-sion of recombinant ATF3, while Adv-YFP was infected

into HepG2 at a multiplicity of infection of 5 as a control

(Figs 1C and 5B) To maximize ATF3 expression, cells

were lysed 48 h after infection

Knock-down of ATF3

Commercially available siRNA against ATF3 (siATF3,

Santa Cruz Biotechnology) was used HepG2 cells grown

in six-well plates were transfected with siATF3 using

LipofectAMINE reagent according to the manufacturer’s

protocol Briefly, the transfection reaction included

optimized amount of siATF3 (100 pm), 2· 106

cells and

4 lL of Lipofectamine reagent A possible non-specific gene silencing effect was assessed using a non-targeting negative control siRNA (46-2001; Invitrogen)

Promoter region constructs

Portions of the AdipoR2 promoter region (approximately

2 kb) were amplified using PCR with human genomic DNA as the template The AR2P()1974) primer pair

GAGG-3¢ and 5¢-ACTTCTTGGGAGCCACCGCTGAG-3¢ A series of deletion constructs of the AdipoR2 promoter were PCR-generated using pairwise combinations of the antisense primer 5¢-ACTGGCGGCCGCTCGAG-3¢ with one of the sense primers AR2P()870), AR2P()343) or AR2P()72) (5¢-GGTACCTTCCCCCTCCTACTGAATGT-3¢, 5¢-GGTACCCCTCCTCCTCAGCTCCAAAT-3¢ and 5¢-GGTACCTCGTGGGGGCGGGGAGA-3¢, respectively) Plasmids were constructed as derivatives of pGL3-Basic luciferase reporter vectors (Promega, Madison, WI, USA) using the KpnI and XhoI restriction sites AR2P()343D),

a deletion mutant lacking the putative ATF3-binding site, was PCR-generated from the AR2P()343) plasmid using the additional internal primers 5¢-GAGGCGGTTCGAG CCAATA-3¢ and 5¢-CGTGCGGTCGTGGGGG-3¢, which hybridized upstream and downstream, respectively, of the

22 bp promoter region containing the putative ATF3-bind-ing site at nucleotide positions)94 to )86

Luciferase activity assay

HepG2 cells were grown in six-well plates to 70% conflu-ence and then transfected with pGL3-Basic-derived reporter constructs containing the AdipoR2 promoter region and a pcDNA3.1-derived ATF3 expression plasmid using Lipofec-tAMINE reagent (Invitrogen) according to the manufac-turer’s instructions [45] b-galactosidase (CMV-b-gal) expression vectors were used to correct differences in trans-fection efficiency The cells were lysed 24 h after transfec-tion, and their luciferase activity was measured using a luciferase assay system (Promega)

Semi-quantitative RT-PCR

We used the following primers for RT-PCR: atf3-sense, 5¢-GGTTTGCCATCCAGAACAAG-3¢; atf3-antisense, 5¢-CC TCCCAGGAGAAGGTAAGC-3¢; adipor2-sense, 5¢-TAGC CTTTGGTTTGCTTTGG-3¢; adipor2-antisense, 5¢-CATAT CTCCAGGCGTCAACC-3¢; gapdh-sense, 5¢-ATGACATC AAGAAGGTGGTG-3¢; gapdh-antisense, 5¢-CCAAATTC GTTGTCATACCA-3¢ Total RNA was obtained from HepG2 cells using an RNeasy kit (Qiagen GmbH, Hilden, Germany) according to the manufacturer’s instructions