lable at ScienceDirect

Atherosclerosis 234 (2014) 391e400

Contents lists avai

Atherosclerosis

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

Downregulation of the glucocorticoid-induced leucine zipper (GILZ)promotes vascular inflammation

Rebecca T. Hahn a,1, Jessica Hoppstädter a,1, Kerstin Hirschfelder a, Nina Hachenthal a,Britta Diesel a, Sonja M. Kessler a, Hanno Huwer b, Alexandra K. Kiemer a,*aDepartment of Pharmacy, Pharmaceutical Biology, Saarland University, Saarbruecken, GermanybDepartment of Cardiothoracic Surgery, Völklingen Heart Centre, Völklingen, Germany

a r t i c l e i n f o

Article history:Received 1 August 2013Received in revised form28 February 2014Accepted 23 March 2014Available online 5 April 2014

Keywords:Laminar shear stressmRNA stabilityAtherosclerosisVein graft failureToll-like receptor 2 (TLR2)Tristetraprolin (TTP, ZFP36)NF-kB

* Corresponding author. Saarland University, CampD-66041 Saarbruecken, Germany. Tel.: þ49 681 3057302.

E-mail address: pharm.bio.kiemer@mx.uni-saarlan1 These authors contributed equally.

http://dx.doi.org/10.1016/j.atherosclerosis.2014.03.0280021-9150/� 2014 Published by Elsevier Ireland Ltd.

a b s t r a c t

Objective: Glucocorticoid-induced leucine zipper (GILZ) represents an anti-inflammatory mediator,whose downregulation has been described in various inflammatory processes. Aim of our study was todecipher the regulation of GILZ in vascular inflammation.Approach and results: Degenerated aortocoronary saphenous vein bypass grafts (n ¼ 15), which exhibitedinflammatory cell activation as determined by enhanced monocyte chemoattractrant protein 1 (MCP-1,CCL2) and Toll-like receptor 2 (TLR2) expression, showed significantly diminished GILZ protein andmRNA levels compared to healthy veins (n ¼ 23). GILZ was also downregulated in human umbilical veinendothelial cells (HUVEC) and macrophages upon treatment with the inflammatory cytokine TNF-a in atristetraprolin (ZFP36, TTP)- and p38 MAPK-dependent manner. To assess the functional implications ofdecreased GILZ expression, we determined NF-kB activation after GILZ knockdown by siRNA and foundthat NF-kB activity and inflammatory gene expression were significantly enhanced. Importantly, ZFP36 isinduced in TNF-a-activated HUVEC as well as in degenerated vein bypasses. When atheroprotectivelaminar shear stress was employed, GILZ levels in HUVEC increased on mRNA and protein level. Laminarflow also counteracted TNF-a-induced ZFP36 expression and GILZ downregulation. MAP kinase phos-phatase 1 (MKP-1, DUSP1), a negative regulator of ZFP36 expression, was distinctly upregulated underlaminar shear stress conditions and downregulated in degenerated vein bypasses.Conclusion: Our data show a diminished expression of the anti-inflammatory mediator GILZ in theinflamed vasculature and indicate that GILZ downregulation requires the mRNA binding protein ZFP36.We suggest that reduced GILZ levels play a role in cardiovascular disease.

� 2014 Published by Elsevier Ireland Ltd.

1. Introduction

Atherosclerosis isa chronic inflammatorydisease characterizedbyaccumulation of inflammatory cells and lipids in the vascular wall ofarteries. Vein graft remodeling is also characterized by inflammatoryevents [1,2]. The inflammatory activation of endothelial cells (EC) andmacrophages (MF), triggered by different mediators such as tumornecrosis factor (TNF)-a, plays a central role within this process [3,4].

Generally, atherosclerotic plaques are localized in curvatures orbifurcations of vessels where static conditions as well as low andoscillatory shear stress occur [5]. These inflammatory conditions

us C2 2, P.O. Box 15 11 50,2 57301; fax: þ49 681 302

d.de (A.K. Kiemer).

promote the formation of atherosclerotic lesions by modification ofgene and protein expression. In straight vessels the laminar bloodflow is known as amain atheroprotective factor, which is importantfor the physiological function of the endothelium [5] via mecha-nisms called mechanotransduction [6].

Besides laminar shear stress as a physical inhibitor of vascularinflammation, other regulators antagonizing vascular inflammationare as yet poorly investigated. We hypothesized thatglucocorticoid-induced leucine zipper (GILZ, synonymousTSC22D3) might represent a promising candidate. GILZ, an anti-inflammatory protein inducible by glucocorticoids, was shown tobe expressed in various cells of the immune system as well as in EC[7e9]. GILZ induction has been suggested to result in an inhibitionof nuclear factor (NF)-kB and activator protein (AP)-1, therebyleading to diminished cytokine transcription [7,8].

We recently reported that the downregulation of GILZ upon TLRactivation is critically involved in inflammatory macrophage (MF)

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400392

activation [9]. In stimulated MF, GILZ was actively downregulatedvia GILZ mRNA destabilization, a process, which required the mRNAbinding protein tristetraprolin (TTP, ZFP36). These results implicatean inflammatory function of ZFP36 as does the finding that ZFP36 isexpressed in EC and foam cells of atherosclerotic lesions [10].Moreover, ZFP36 expression is inhibited downstream of the anti-inflammatory dual-specificity protein phosphatase 1 (DUSP1, MKP-1) [11], a potent inhibitor of p38 mitogen-activated protein kinase(MAPK) [12]. The role of DUSP1 in the pathophysiology of athero-sclerotic plaques is controversial, as both anti-inflammatory [13,14]as well as pro-atherosclerotic actions have been suggested [15,16].

Our data suggest a profound decrease of GILZ expression inhuman EC and vein graft stenosis under inflammatory conditions.GILZ downregulation is mediated by ZFP36 upregulation and leadsto an activation and translocation of NF-kB. These findings suggestthat the disappearance of GILZ plays a key role in the developmentof vascular disease.

2. Materials and methods

2.1. Materials

For Western blot analyses, anti-GILZ (sc-26518), anti-p65 (sc-109), anti-p50 (sc-114), and anti-MKP-1 (sc-1199) antibodies werepurchased from SantaCruz (Heidelberg, Germany), anti-ZFP36(T5327) and anti-Tubulin (T9026) were obtained from Sigma(Taufkirchen, Germany), and anti-TLR2 (Cat # 3268-1) from Epi-tomics (Burlingame, USA). The IRdye-labeled secondary antibodiesgoat anti-mouse, goat anti-rabbit, and donkey anti-goat were fromLI-COR Biosciences (Bad Homburg, Germany), and anti-GILZ anti-body for IHC (FL-134) was obtained from SantaCruz (Heidelberg,Germany). RNAlater, Qiazol and the RNeasy mini kit were fromQiagen (Hilden, Germany). All primers, probes, and oligonucleo-tides were obtained from MWG (Ebersfeld, Germany). 5� HOTFIREPol� EvaGreen� qPCR Mix Plus was from Solis BioDyne (Tartu,Estonia). siGILZ, siZFP36 (siGENOME SMARTpool) and siControlwere from Dharmacon (Nidderau, Germany).

pcDNA3-p38a-dn and pcDNA3-p38b2-dn were a gift from Prof.Dr. Jian-Dong Li, University of Rochester Medical Center, USA.pGL4.32[luc2P/NF-kB-RE/Hygro] containing 5 repetitive elementsof the NF-kB consensus sequence GGGAATTTCC was from Promega(Heidelberg, Germany).

The plasmids pZeo-hTTP-sense (ZFP36-V) and pZeo-hTTP-antisense (Co-V) were a kind gift from Hartmut Kleinert (Depart-ment of Pharmacology, Johannes Gutenberg University, Mainz,Germany).

All other materials were purchased from Sigma (Taufkirchen,Germany), Roth (Karlsruhe, Germany), MP Biomedicals (Heidel-berg, Germany), and Merck (Darmstadt, Germany).

2.2. Vessel specimens

Human healthy saphenous veins and degenerated aortocoronarysaphenous vein bypass grafts were obtained from patients under-going coronary bypass surgery and immediately transferred intoRNA stabilization solution (RNAlater). All samples were obtainedwith the consent of patients, and permission was given by the localethics committee (ref #102/09). Total RNA and proteins were iso-lated using Qiazol according to the manufacturer’s protocol.

2.3. Immunohistochemistry

Paraffin-embedded sections of murine femoral arteries werestained for GILZ with the CSA II Kit (Dako, Hamburg, Germany)[17,18].

2.4. Cell culture

Preparation, cultivation, and characterization of HUVEC hasbeen described previously [19,20]. HUVEC were used in passagethree or four and grown in 6-well plates or on collagen (Roche,1179179, 30 mg/ml in 0.2% acetic acid) coated glass slides in 4- wellplates until confluence.

Cultivation and preparation of primary human alveolar MFshave been described previously [9,21].

2.5. Shear stress

Confluent HUVEC grown on coated glass slides were exposed tolaminar shear stress of 20 dyn/cm2 for 22 h in a parallel flowchamber modified after (Frangos et al., 1988) [22] and manufac-tured by upag AG (Vollersode, Germany). TNF-a (10 ng/ml) wasadded to the flow medium and flow was continued for another 2 hor 3.5 h.

Flow was produced by a peristaltic pump (403U/VM purple/white, Watson Marlow), and flow rates were adjusted to match ashear stress of 20 dyn/cm2. Silicon tubes were purchased fromVWR(Darmstadt, Germany), and silicon mats for gasket constructionwere purchased from rfQ Medizintechnik (Tuttlingen, Germany).

2.6. Transfections

HUVEC were grown until 80% confluency and transfected withsiRNA (1 mM) or 20 mg/ml (luciferase vectors, p38dn) to 50 mg/ml(ZFP36 overexpression) plasmid DNA using Amaxa� Nucleofector�

Technology according to the manufacturer’s instructions (Lonza,Basel, Switzerland). Experiments were performed 20 h or 24 h aftertransfection.

2.7. Luciferase reporter gene assay

After transfection of HUVEC with pGL4.32[luc2P/NF-қB-RE/Hygro] and siGILZ or siControl, luciferase activity was measured asdescribed previously [9,19].

2.8. Detection of mRNA

Total RNA of cultivated cells was extracted with Qiazol or theRNEasy Mini kit. After residual DNA was removed (DNA-free kit,Applied Biosystems, Darmstadt, Germany), reverse transcriptionwas carried out using the high capacity cDNA reverse transcriptionkit (Applied Biosystems, Darmstadt, Germany) and 0.25e1 mg totalRNA. Real-time RT-PCR was performed in a Bio-Rad Cycler usinggene-specific primers, dual-labeled probes (see online-only DataSupplement) or 5� HOT FIREPol� EvaGreen� qPCR Mix Plus [23].cDNA cloned into pGEM-T Easy (Promega, Heidelberg, Germany) ora cDNA were used as a standard dilution series as described pre-viously [21]. All samples and standards were analyzed in triplicateon each plate.

2.9. Western blot analysis

Isolation of proteins of whole cell extracts was performed usingQiazol according to the manufacturer’s protocol or as reportedpreviously [9,24]. Nuclear and cytosolic protein extracts were pre-pared as described [25]. Protein concentrations were determinedby Bradford assay (Bio-Rad, Munich, Germany) or Pierce BCA Pro-tein Assay Kit (Thermo Scientific, Bonn, Germany).

Equal protein amounts were separated using 12% or 15% SDS/PAGE gels. After electroblotting onto a PVDF membrane (MilliporeGmbH, Schwalbach/Ts., Germany) GILZ, ZFP36, TLR2, DUSP1, p65,

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400 393

and p50 were detected using specific antibodies and the ODYSSEY�

Infrared Imaging System (LI-COR�, LI-COR Biosciences, Bad Hom-burg, Germany) as described previously [26]. Relative proteinamounts were determined using either Odyssey or ImageJsoftware.

2.10. Statistical analysis

All experiments were performed with HUVEC or MF from atleast two different donors and cell preparations. All other experi-ments were repeated at least three times. Data are expressed asmean þ SEM and statistical significance was determined by Stu-dent’s t-test.

3. Results

3.1. GILZ expression in degenerated vein bypasses

Degenerated vein bypasses were identified as inflamed tissuedue to significantly increased mRNA expression of the inflamma-tory markers CCL2 (MCP1) [18,19] and TLR2 [20] compared tohealthy veins (Fig. 1A). Importantly, inflamed veins showed asignificantly reduced GILZ mRNA expression (Fig. 1A). Similar re-sults were observed when analyzing GILZ and TLR2 on protein level(Fig. 1B). In order to identify GILZ expressing cells, we stainedmurine histological samples and detected a distinct localization ofGILZ protein in the endothelial layer of vessels (Fig. 1 in online-onlyData Supplement).

3.2. Inflammatory response in EC and MF

Both immunohistochemistry as well as data previously pre-sented by ourselves and others suggested a distinct expression ofGILZ in EC and MFs [8,9]. As shown in Fig. 1C and D, GILZ mRNA aswell as proteinwas downregulated under inflammatory conditions,i.e. after TNF-a treatment in both cell types. Inflammatory cellactivation was confirmed by an induction of TLR2 and CCL2 (MCP1)mRNA in HUVEC (Fig. 1E).

Interestingly, the mRNA binding protein ZFP36, known todestabilize GILZ mRNA in MFs [9], was strongly induced by TNF-a(Fig. 2A). Early after TNF-a treatment, ZFP36 was present in its low-phosphorylated, low molecular weight form, which is known to bethe active, but instable variant. At later time points, the phos-phorylated high-molecular weight isoform of ZFP36, which isinactive but stable [27], predominated (Fig. 2A). In order to examinethe functional links between GILZ and ZFP36 expression in HUVEC,an siRNA-mediated knockdown of ZFP36 was performed. In HUVECtransfected with control siRNA (siCo), ZFP36 was upregulated uponTNF-a treatment, which was paralleled by reduced GILZ expression.In contrast, GILZ downregulation was abrogated in siZFP36-transfected cells (Fig. 2BeD). To support these findings, we tran-siently overexpressed ZFP36 in HUVEC (Fig. 2E), which indeedresulted in reduced GILZ levels. Taken together, these results sug-gest a key role for ZFP36 in the regulation of GILZ expression.

3.3. Regulation of GILZ and ZFP36 by anti-inflammatory laminarshear stress

While inflammatory conditions downregulated GILZ, laminarshear stress as an anti-inflammatory and anti-atheroscleroticstimulus elevated GILZ mRNA levels in HUVEC (Fig. 3A). The samefinding was observed on protein level (Fig. 3B). The anti-inflammatory activation state was confirmed by elevated HO1(heme oxygenase 1) mRNA expression (Fig. 3C).

While ZFP36 was induced by TNF-a on the transcriptional level,its expression tended to decrease during laminar flow (Fig. 3D). Acombination of laminar flow and TNF-a completely abrogated GILZdownregulation, which we typically observed upon TNF-a treat-ment in statically cultured HUVEC (Fig. 3E). Concordantly, ZFP36induction by TNF-a was abrogated in cells exposed to laminar flow(Fig. 3F), suggesting that the lack of ZFP36 induction contributes tothe elevated GILZ expression in TNF-a-treated shear stressed cells.Interestingly, ZFP36 levels were elevated in degenerated, inflamedvenous bypasses (Fig. 3G).

3.4. Mechanisms of GILZ downregulation in inflammation

Our data suggest an inverse regulation of ZFP36 and GILZ ininflammation and under anti-inflammatory conditions in HUVEC.In fact, ZFP36 has been reported by ourselves to be a destabilizer ofGILZmRNA inmacrophages [9] and has been shown to be regulatedby DUSP1 [11], which inhibits mitogeneactivated protein kinases,most importantly p38 MAPK [12].

To assess the influence of p38 MAPK activation on ZFP36 andGILZ expression, p38 phosphorylation in HUVEC was inhibited bypretreatment with the p38 MAPK inhibitor SB203580 prior to TNF-a challenge. p38 inhibition antagonized TNF-a-mediated ZFP38induction both onmRNA and protein level (Fig. 4A, B and E). Similardata were obtained after transfection of HUVEC with a dominantnegative mutant of p38 (Fig. 2 in online-only Data Supplement).Reduced ZFP36 expression was accompanied by an abrogation ofGILZ downregulation (Fig. 4C, D and F), indicating that p38 inhi-bition enhances GILZ expression by reducing ZFP36 levels.

Interestingly, a significant downregulation of DUSP1 proteinexpression in degenerated vein bypasses could be detected(Fig. 4G). In cultivated HUVEC, an upregulation of DUSP1 mRNAlevels by laminar shear stress and downregulation by TNF-a wasobserved (Fig. 4H).

3.5. Functional implications of GILZ downregulation

Finally, we aimed to determine whether GILZ downregulationhas functional implications in inflammatory activation of HUVEC.GILZ knockdownwas shown to increase MF activation, as assessedby TNF-a induction and NF-kB activation [9], whereas GILZ over-expression inhibited NF-kB activation in EC [8].

We therefore knocked down GILZ in HUVEC by siRNA resultingin reduced GILZ mRNA and protein levels (Fig. 5A, B). GILZ knock-down induced the nuclear translocation of the NF-kB subunits p65and p50 (Fig. 5C). Using a luciferase reporter gene under an NF-kBpromoter, we showed that GILZ knockdown significantly increasedNF-kB activity compared to control transfected cells (Fig. 5D). Theabsence of GILZ after double-transfection with GILZ siRNA and theluciferase plasmid was confirmed usingWestern blot analysis (datanot shown), whereas functionality of the luciferase assay wasverified measuring TNF-a-induced NF-kB activity (Fig. 3 in online-only Data Supplement). These findings showed that disappearanceof GILZ can liberate NF-kB and induce its activation andtranslocation.

We then assessed whether downregulation of GILZ had anyinfluence on TLR2 [20] and adhesion molecule expression [28]. Wefound indeed that the TNF-a-induced expression of TLR2, ICAM1and SELE (E-selectin) was enhanced in siGILZ-transfected HUVECcompared to equally treated siCo-transfected controls (Fig. 5E),suggesting that the absence of GILZ drives a proinflammatoryresponse.

Fig. 1. GILZ mRNA and protein expression in clinical samples, macrophages, and EC. Panel A: GILZ, TLR2, and CCL2 mRNA expression in human veins. mRNA expression in saphenousveins (n ¼ 23) and degenerated aortocoronary saphenous vein bypass grafts (n ¼ 15) was measured by real-time RT-PCR using ACTB for normalization. Data are presented asindividual values (black squares) as well as 25th and 75th percentiles as boxes within geometric medians (line), arithmetic medians (square), 10th and 90th percentiles as whiskers,and ends of values (cross). Panel B: GILZ and TLR2 protein expression in human veins. Equal protein amounts were assessed by Western blot analysis using tubulin as loading control.One representative blot out of 4 independent experiments with 11 healthy and 12 degenerated samples is shown. Signal intensities were measured relative to tubulin values, andvalues for healthy controls were set as one. Panel C-E: TNF-a response. Primary human MF (C) or HUVEC (D, E) were treated with 10 ng/ml TNF-a for 2 h or for the indicated timepoints. Protein levels were measured byWestern blot analysis using tubulin as loading control. mRNA levels were determined by real-time RT-PCR using ACTB for normalization. ForHUVEC, results of three independent experiments are shown (duplicates). For MF, data represent means of two independent experiments performed in triplicate.

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400394

4. Discussion

The anti-inflammatory mediator GILZ is inducible by glucocor-ticoids and acts as an inhibitor of NF-kB and AP-1 by direct bindingand inhibition of their nuclear translocation [7]. Expression of GILZ

has been described for many human tissues and cell types [7e9].These include EC andMF, suggesting a potential role for GILZ in thepathogenesis of atherosclerosis.

We previously reported that TLR activation leads to GILZdownregulation in primary human MF and lung tissue of LPS-

Fig. 2. ZFP36-dependent GILZ downregulation in EC. Panel A: Induction of ZFP36 by TNF-a. HUVEC were treated with 10 ng/ml TNF-a for the times indicated and protein levels ofZFP36 were assessed by Western blot analysis using tubulin as loading control (n ¼ 3, duplicates). Values for untreated cells were set as one hundred percent or one, *p < 0.05,**p < 0.01, ***p < 0.001 compared to untreated cells. Panel B-D: Influence of ZFP36 knockdown on TNF-a-induced GILZ downregulation. HUVEC were transfected with ZFP36 siRNA(siZFP36) or control siRNA (siCo). 20 h after transfection, cells were treated with TNF-a (10 ng/ml). ZFP36 and GILZ protein expression were analyzed by Western blot using tubulinas loading control. One representative blot out of 3 independent experiments (A) and data of three independent experiments (duplicates) are shown (BeD). Panel E: Influence ofZFP36 overexpression on GILZ expression. HUVEC were transfected with either control (Co-V) or ZFP36 expression (ZFP36-V) vector and harvested after 24 h. ZFP36 and GILZexpression were assessed by western blot analysis. Tubulin served as a loading control. Values indicate relative signal intensities of 3 independent experiments performed induplicate after normalization to tubulin values. Values for Co-V transfected cells were set as one. *p < 0.05, **p < 0.01, compared to Co-V transfected cells.

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400 395

treated mice. [9] Here, we present evidence that diminished GILZexpression is a deleterious response in degenerated vein bypasses.A GILZ downregulation or even absence in inflammatory diseases,such as chronic rhinosinusitis, Crohn’s disease, or tuberculosis hasbeen reported in the literature [9,29,30], indicating that theabsence of GILZ is a general phenomenon in inflammation.

GILZ was recently shown to exert anti-inflammatory actions indifferent animal models of inflammatory disease [31e33], althoughthe observed effects were primarily attributed to interference withleukocyte functions. In addition to leukocytes, EC play a key role invascular inflammation [8]. Investigations on the role of GILZ in ECwere mostly carried out after GILZ overexpression. In this context,

Fig. 3. Inflammatory responses in HUVEC and aortocoronary saphenous vein bypass grafts. Panel A, B, C, D, E, F: GILZ, HO1, and ZFP36 expression under inflammatory and anti-inflammatory conditions. HUVEC were treated with 10 ng/ml TNF-a under static conditions or exposed to 24 h laminar flow (20 dyn/cm2) as indicated. GILZ (A), HO1 (C), andZFP36 (D) mRNA levels were determined by real-time RT-PCR using ACTB for normalization. Values for untreated cells were set as one, **p < 0.01, ***p < 0.001 compared tountreated cells under static conditions. Data were obtained from 4 independent experiments performed in duplicate. GILZ (B, E) and ZFP36 (F) protein levels were measured byWestern blot analysis using tubulin as loading control ((B, F) n ¼ 6; (E) n ¼ 8 derived from 4 independent experiments). Values for untreated cells were set as one as indicated,**p < 0.01, ***p < 0.001, þp ¼ 0.051 compared to untreated cells. Data represent means of 4 independent experiments performed in duplicate. Panel G: ZFP36 expression in humanveins. Equal protein amounts were assessed by Western blot analysis using tubulin as loading control. One representative blot out of 3 independent experiments with 8 healthy and10 degenerated samples is shown. Signal intensities were measured relative to tubulin values, and values for healthy samples were set as one.

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400396

GILZ was shown to reduce the capacity of EC to support leukocyteinteractions, i.e. rolling, adhesion, and transmigration [8]. Althougha growing body of evidence has accumulated regarding theimportance of GILZ in EC, little is known about the role of endog-enous GILZ in EC and the mechanisms regulating its expression.Herein, we confirm a substantial expression of GILZ in EC [8,34,35]and demonstrate that GILZ expression is considerably decreased inTNF-a-challenged HUVEC and MF. In contrast, atheroprotectivelaminar shear stress induced GILZ mRNA and protein in EC whencompared to static culture conditions, as previously suggested by amicroarray approach [35].

Blood flow influences atherosclerosis by exerting shear stress onvascular endothelium, which differs in magnitude and directiondepending on the vascular anatomy and blood pressure. Shearstress alters the phenotype of EC, which respond to it via mecha-noreceptors that translate mechanical distortions (see DataSupplement, Fig. 4) into various molecular signals [36]. Corre-spondingly, EC were shown to respond to laminar shear stress byglucocorticoid receptor translocation [37]. As multiple glucocorti-coid responsive elements are present in the GILZ promoter [7], GILZinduction by laminar shear stress might be a result of glucocorti-coid receptor activation.

Fig. 4. Mechanisms of ZFP36 regulation. Panels A-F: ZFP36 and GILZ expression after inhibition of p38 MAPK activity. HUVEC were pretreated with solvent control DMSO or SB203580(10 mM), followed by treatment with 10 ng/ml TNF-a for 2 h (AeB, EeF) or 4 h (CeD). Protein levels were measured by Western blot analysis using tubulin as loading control (AeD).mRNA levels were determined by real-time RT-PCR using ACTB for normalization (EeF). Values for cells pretreated with the solvent control DMSO, either in the presence (B, E) orabsence (D, F) of TNF-a were set as one hundred percent. Data show means of three (AeD) or two (EeF) independent experiments performed in triplicates, *p < 0.05, **p < 0.01,***p < 0.001, n.s: not significant. Panel G: DUSP1 protein expression in human veins. Equal protein amounts were assessed by Western blot analysis using tubulin as loading control.One representative blot out of 4 independent experiments with 11 healthy and 12 degenerated samples is shown. Signal intensities were measured relative to tubulin values, andvalues for samples from healthy tissues were set as one. Panel H: DUSP1 expression under pro- and anti-inflammatory conditions. HUVEC were treated with 10 ng/ml TNF-a understatic conditions or exposed to 24 h laminar flow (20 dyn/cm2) as indicated. mRNA levels were determined by real-time RT-PCR using ACTB for normalization. Values for untreatedcells under static conditions were set as one, **p < 0.01, ***p < 0.001 compared to untreated cells. Data represent means of 4 independent experiments performed in duplicate.

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400 397

Most interestingly, a TNF-a challenge failed to diminish GILZlevels in HUVEC exposed to laminar shear stress. We previouslyreported a correlation between GILZ downregulation and upregu-lation of the mRNA-binding protein ZFP36 in MF [9]. In line withthese findings, GILZ downregulation was paralleled by a rapid andextensive induction of ZFP36 in TNF-a-treated HUVEC, suggesting apossible role of ZFP36 in GILZ repression. Both ZFP36 over-expression and knockdown experiments confirmed this

assumption. Accordingly, ZFP36 induction was also seen inatherosclerotic vessels by others and ourselves [10], but not underanti-inflammatory conditions, i.e. laminar shear stress. ZFP36upregulation is usually considered to be an anti-inflammatoryfeedback loop, since inflammatory cytokines, such as TNF-a,represent the most extensively studied ZFP36 targets [38,39].Therefore, ZFP36 upregulation has been suggested to be an athe-roprotective process [10]. It has to be noted, however, that ZFP36-

Fig. 5. GILZ downregulation drives pro-inflammatory responses in EC. Panel A, B: GILZ knockdown. HUVEC were transfected with GILZ siRNA (siGILZ) or control siRNA (siCo). Cellswere harvested after 20 h (A, B) or 24 h (A). GILZ protein expression was analyzed byWestern blot using tubulin as loading control (A) and mRNA expression was quantified by real-time RT-PCR using ACTB for normalization (B). One representative blot out of 4 independent experiments (A) and data of three independent experiments (duplicates) are shown. (B)Panel C: NF-kB translocation after GILZ knockdown. HUVEC were nucleofected with GILZ siGILZ or siCo and harvested after 20 h. Equal protein amounts were assessed byWestern blotanalysis using tubulin as loading control. Results of 4 independent experiments are shown. *p < 0.05, #p ¼ 0.065 compared to siCo transfected cells. Panel D: NF-kB activity after GILZknockdown. HUVEC were transfected with either siCO or siGILZ and an NF-kB driven luciferase reporter construct. Cells were harvested 20 h post transfection. NF-kB activity wasdetermined by measuring luciferase activity. Data represent means of 4 independent experiments performed in quinticate. Values for siCo were set as one, ***p < 0.001 compared tosiCo transfected cells. Panel E: Impact of GILZ knockdown in HUVEC on TNF-a-induced expression of TLR2 and adhesion molecules. HUVEC were transfected with siGILZ or siCo. 20 h aftertransfection, TNF-a was added for an additional 4 h mRNA expression was quantified by real-time RT-PCR using ACTB for normalization. Values obtained for siCo transfected cellswere set as one hundred percent. *p < 0.05, ***p < 0.001 compared to siCo transfected cells.

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400398

associated mRNAs encode a broad spectrum of proteins engaged invarious cellular processes, including the anti-inflammatory medi-ator IL-10 [40]. Therefore, additional factors might be needed toorchestrate ZFP36 actions. These trigger factors may include othermRNA-binding proteins, whose binding might be further modu-lated by miRNAs [41]. Up to date, however, the identity of othermRNA-binding proteins and/or miRNAs interacting with GILZmRNA remains elusive. In addition, we can not rule out the possi-bility that GILZ expression might also be affected by other mech-anisms, such as proteasomal degradation [42].

The p38 MAPK pathway has been shown to promote ZFP36upregulation in human and murine MFs by increasing ZFP36expression [11,43,44]. In line with these reports, we found that p38inhibition also markedly reduced ZFP36 levels in HUVEC, whereasGILZ downregulation upon TNF-a-treatment was abrogated, indi-cating that p38 regulates GILZ expression via a mechanisminvolving ZFP36.

DUSP1, which can be induced in EC by laminar shear stress, hasbeen suggested to protect arteries from inflammation, mainly by

suppressing the activities of p38 and JNK MAP kinases in thevascular endothelium [14].

Investigations on LPS-induced ZFP36 expression inMF obtainedfrom DUSP1 knockout mice or in epithelial and macrophage-likecell lines after siRNA-mediated DUSP1 silencing recently revealeda reverse correlation between DUSP1 and ZFP36 expression: lack ofDUSP1was shown to result in enhanced ZFP36 expression, whereasp38 inhibition had opposing effects. This led to the conclusion thatDUSP1 suppressed ZFP36 expression by abrogating p38 activity[11]. Our data showing DUSP1 induction in HUVEC exposed tolaminar shear stress, paralleled by diminished ZFP36 levels andelevated GILZ expression, suggest a similar mechanism. In line withthis finding, we found that DUSP1was expressed in healthy, but notin atherosclerotic vessels.

To assess functional implications of GILZ downregulation underinflammatory conditions, we used siRNA to knockdown GILZ inHUVEC and subsequentlymeasured NF-kB activity. NF-kB is amajorpro-inflammatory transcription factor, whose translocation resultsin the expression of various cytokines, growth factors, and adhesion

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400 399

molecules [28,45]. Endothelial cell-specific NFekB inhibition hasbeen reported to protect mice from atherosclerosis and vascularremodeling [46,47], emphasizing the critical role of NF-kB in thepathogenesis of atherosclerosis.

We found that the absence of GILZ enhanced NF-kB activity, i.e.nuclear translocation of p65 and p50 and NF-kB-dependent tran-scription. Correspondingly, NF-kB activation has been reported tobe diminished after GILZ overexpression in several cell types[30,48e50]. Moreover, GILZ knockdown has been shown to activateairway epithelial cells by increased cytokine expression [50] and toamplify NF-kB activation in LPS-treated MFs [9].

A recent report by Cheng et al. [8] contrasts our findings byshowing that GILZ siRNA did notmodify TNF-a-induced endothelialcell responses, i.e. leukocyte rolling, transmigration, and IL-6release. However, these results can hardly be compared to ourstudies, since both treatment schemes and the readout parametersdiffer. The fact that TNF-a treatment alone results in diminishedGILZ expression might complicate matters further, as it might behard to distinguish between the effects of the naturally occurringGILZ downregulation and the siRNA-mediated GILZ knockdown.Our data on GILZ siRNA-transfected HUVEC support a functionalimportance of enhanced NF-kB signaling. We found indeed thatseveral pro-inflammatory mediators, including TLR2, SELE andICAM1 [20,28], were upregulated in TNF-a-treated siGILZ-transfected HUVEC compared to equally treated siCo-transfectedcontrols. These findings indicate that GILZ downregulation mightbe a critical step in atherogenesis.

5. Conclusion

Taken together, our data show that the expression of GILZ inhuman EC prevents vascular inflammation by suppressing NF-kBactivation. This assumption is supported by the decrease of GILZlevels found in atherosclerotic vessels, suggesting upregulation ofGILZ as a potential target for the treatment of the inflamedendothelium.

Acknowledgments

Wewould like to thank Theo Ranssweiler for excellent technicalassistance and Nicolas Frank as well as Susanne Schütz for help inthe preparation of histological samples.

We also thank PD Dr. Dieter Mink from the Klinikum Saar-brücken for supply of umbilical cords and Dr. Matthias Engel forproviding the NF-kB plasmid. Finally, we thank Christian andAlexander Hahn for support in development and construction ofparallel flow chambers. The work was supported, in part, by theDeutsche Forschungsgemeinschaft (KI 702).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atherosclerosis.2014.03.028.

References

[1] Karper JC, de Vries MR, van den Brand BT, Hoefer IE, Fischer JW,Jukema JW, et al. Toll-like receptor 4 is involved in human and mouse veingraft remodeling, and local gene silencing reduces vein graft disease inhypercholesterolemic APOE*3Leiden mice. Arterioscler Thromb Vasc Biol2011;31:1033e40.

[2] McPhee JT, Nguyen LL, Ho KJ, Ozaki CK, Conte MS, Belkin M. Risk prediction of30-day readmission after infrainguinal bypass for critical limb ischemia. J VascSurg 2013;57:1481e8.

[3] McKellar GE, McCarey DW, Sattar N, McInnes IB. Role for TNF in atheroscle-rosis? Lessons from autoimmune disease. Nat Rev Cardiol 2009;6:410e7.

[4] Feng Y, Gordts SC, Chen F, Hu Y, Van CE, Jacobs F, et al. Topical HDL admin-istration reduces vein graft atherosclerosis in apo E deficient mice. Athero-sclerosis 2011;214:271e8.

[5] Cunningham KS, Gotlieb AI. The role of shear stress in the pathogenesis ofatherosclerosis. Lab Invest 2005;85:9e23.

[6] Davies PF. Hemodynamic shear stress and the endothelium in cardiovascularpathophysiology. Nat Clin Pract Cardiovasc Med 2009;6:16e26.

[7] Ayroldi E, Riccardi C. Glucocorticoid-induced leucine zipper (GILZ): a newimportant mediator of glucocorticoid action. FASEB J 2009;23:3649e58.

[8] Cheng Q, Fan H, Ngo D, Beaulieu E, Leung P, Lo CY, et al. GILZ overexpressioninhibits endothelial cell adhesive function through regulation of NF-kappaBand MAPK activity. J Immunol 2013;191:424e33.

[9] Hoppstädter J, Diesel B, Eifler LK, Schmidt T, Brüne B, Kiemer AK. Glucocor-ticoid-Induced Leucine Zipper is downregulated in human alveolar macro-phages upon Toll-like receptor activation. Eur J Immunol 2012;42:1e13.

[10] Zhang H, Taylor WR, Joseph G, Caracciolo V, Gonzales DM, Sidell N, et al.mRNA-binding protein ZFP36 is expressed in atherosclerotic lesions and re-duces inflammation in aortic endothelial cells. Arterioscler Thromb Vasc Biol2013;33:1212e20.

[11] Huotari N, Hommo T, Taimi V, Nieminen R, Moilanen E, Korhonen R. Regu-lation of tristetraprolin expression by mitogen-activated protein kinasephosphatase-1. APMIS 2012;120:988e99.

[12] Kiemer AK, Weber NC, Furst R, Bildner N, Kulhanek-Heinze S, Vollmar AM.Inhibition of p38 MAPK activation via induction of MKP-1: atrial natriureticpeptide reduces TNF-alpha-induced actin polymerization and endothelialpermeability. Circ Res 2002;90:874e81.

[13] Kim HS, Ullevig SL, Zamora D, Lee CF, Asmis R. Redox regulation of MAPKphosphatase 1 controls monocyte migration and macrophage recruitment.Proc Natl Acad Sci U S A 2012;109:E2803e12.

[14] Zakkar M, Chaudhury H, Sandvik G, Enesa K, Luong lA, Cuhlmann S, et al.Increased endothelial mitogen-activated protein kinase phosphatase-1expression suppresses proinflammatory activation at sites that are resistantto atherosclerosis. Circ Res 2008;103:726e32.

[15] Imaizumi S, Grijalva V, Priceman S, Wu L, Su F, Farias-Eisner R, et al. Mitogen-activated protein kinase phosphatase-1 deficiency decreases atherosclerosisin apolipoprotein E null mice by reducing monocyte chemoattractant protein-1 levels. Mol Genet Metab 2010;101:66e75.

[16] Shen J, Chandrasekharan UM, Ashraf MZ, Long E, Morton RE, Liu Y, et al. Lackof mitogen-activated protein kinase phosphatase-1 protects ApoE-null miceagainst atherosclerosis. Circ Res 2010;106:902e10.

[17] Tybl E, Shi FD, Kessler SM, Tierling S, Walter J, Bohle RM, et al. Overexpressionof the IGF2-mRNA binding protein p62 in transgenic mice induces a steatoticphenotype. J Hepatol 2011;54:994e1001.

[18] Simon Y, Kessler SM, Bohle RM, Haybaeck J, Kiemer AK. The insulin-likegrowth factor 2 (IGF2) mRNA-binding protein p62/IGF2BP2-2 as a promoterof NAFLD and HCC? Gut 2014;63:861e3.

[19] Weber NC, Blumenthal SB, Hartung T, Vollmar AM, Kiemer AK. ANP inhibitsTNF-alpha-induced endothelial MCP-1 expressioneinvolvement of p38 MAPKand MKP-1. J Leukoc Biol 2003;74:932e41.

[20] Diesel B, Ripoche N, Risch RT, Tierling S, Walter J, Kiemer AK. Inflammation-induced up-regulation of TLR2 expression in human endothelial cells is in-dependent of differential methylation in the TLR2 promoter CpG island.Innate Immun 2012;18:112e23.

[21] Kiemer AK, Senaratne RH, Hoppstädter J, Diesel B, Riley LW, Tabeta K, et al.Attenuated activation of macrophage TLR9 by DNA from virulent mycobac-teria. J Innate Immun 2009;1:29e45.

[22] Frangos JA, McIntire LV, Eskin SG. Shear stress induced stimulation ofmammalian cell metabolism. Biotechnol Bioeng 1988;32:1053e60.

[23] Kessler SM, Pokorny J, Zimmer V, Laggai S, Lammert F, Bohle RM, et al. IGF2mRNA binding protein p62/IMP2-2 in hepatocellular carcinoma: antiapoptoticaction is independent of IGF2/PI3K signaling. Am J Physiol Gastrointest LiverPhysiol 2013;304:G328e36.

[24] Kiemer AK, Vollmar AM. Autocrine regulation of inducible nitric-oxide syn-thase in macrophages by atrial natriuretic peptide. J Biol Chem 1998;273:13444e51.

[25] Tiegs G, Küsters S, Künstle G, Hentze H, Kiemer AK, Wendel A. Ebselen pro-tects mice against T cell-dependent, TNF-mediated apoptotic liver injury.J Pharmacol Exp Ther 1998;287:1098e104.

[26] Dünschede F, Tybl E, Kiemer AK, Dutkowski P, Erbes K, Kircher A, et al. Bcl-2upregulation after 3-nitropropionic acid preconditioning in warm rat liverischemia. Shock 2008;30:699e704.

[27] Brook M, Tchen CR, Santalucia T, McIlrath J, Arthur JS, Saklatvala J, et al.Posttranslational regulation of tristetraprolin subcellular localization andprotein stability by p38 mitogen-activated protein kinase and extracellularsignal-regulated kinase pathways. Mol Cell Biol 2006;26:2408e18.

[28] Kiemer AK, Weber NC, Vollmar AM. Induction of IkappaB: atrial natriureticpeptide as a regulator of the NF-kappaB pathway. Biochem Biophys ResCommun 2002;295:1068e76.

[29] Berrebi D, Bruscoli S, Cohen N, Foussat A, Migliorati G, Bouchet-Delbos L, et al.Synthesis of glucocorticoid-induced leucine zipper (GILZ) by macrophages: ananti-inflammatory and immunosuppressive mechanism shared by glucocor-ticoids and IL-10. Blood 2003;101:729e38.

[30] Zhang XH, Lu X, Long XB, You XJ, Gao QX, Cui YH, et al. Chronic rhinosinusitiswith and without nasal polyps is associated with decreased expression ofglucocorticoid-induced leucine zipper. Clin Exp Allergy 2009;39:647e54.

R.T. Hahn et al. / Atherosclerosis 234 (2014) 391e400400

[31] Beaulieu E, Ngo D, Santos L, Yang YH, Smith M, Jorgensen C, et al. Glucocor-ticoid-induced leucine zipper is an endogenous antiinflammatory mediator inarthritis. Arthritis Rheum 2010;62:2651e61.

[32] Cannarile L, Cuzzocrea S, Santucci L, Agostini M, Mazzon E, Esposito E, et al.Glucocorticoid-induced leucine zipper is protective in Th1-mediated modelsof colitis. Gastroenterology 2009;136:530e41.

[33] Sima AV, Stancu CS, Simionescu M. Vascular endothelium in atherosclerosis.Cell Tissue Res 2009;335:191e203.

[34] Gomez M, Raju SV, Viswanathan A, Painter RG, Bonvillain R, Byrne P, et al.Ethanol upregulates glucocorticoid-induced leucine zipper expression andmodulates cellular inflammatory responses in lung epithelial cells. J Immunol2010;184:5715e22.

[35] McCormick SM, Eskin SG, McIntire LV, Teng CL, Lu CM, Russell CG, et al. DNAmicroarray reveals changes in gene expression of shear stressed humanumbilical vein endothelial cells. Proc Natl Acad Sci U S A 2001;98:8955e60.

[36] Tzima E, Irani-Tehrani M, Kiosses WB, Dejana E, Schultz DA, Engelhardt B,et al. A mechanosensory complex that mediates the endothelial cell responseto fluid shear stress. Nature 2005;437:426e31.

[37] Ji JY, Jing H, Diamond SL. Shear stress causes nuclear localization of endo-thelial glucocorticoid receptor and expression from the GRE promoter. CircRes 2003;92:279e85.

[38] Carballo E, Lai WS, Blackshear PJ. Feedback inhibition of macrophage tumornecrosis factor-alpha production by tristetraprolin. Science 1998;281:1001e5.

[39] Lai WS, Carballo E, Strum JR, Kennington EA, Phillips RS, Blackshear PJ. Evi-dence that tristetraprolin binds to AU-rich elements and promotes thedeadenylation and destabilization of tumor necrosis factor alpha mRNA. MolCell Biol 1999;19:4311e23.

[40] Stoecklin G, Tenenbaum SA, Mayo T, Chittur SV, George AD, Baroni TE, et al.Genome-wide analysis identifies interleukin-10 mRNA as target of triste-traprolin. J Biol Chem 2008;283:11689e99.

[41] George AD, Tenenbaum SA. MicroRNA modulation of RNA-binding proteinregulatory elements. RNA Biol 2006;3:57e9.

[42] Delfino DV, Spinicelli S, Pozzesi N, Pierangeli S, Velardi E, Bruscoli S,et al. Glucocorticoid-induced activation of caspase-8 protects theglucocorticoid-induced protein Gilz from proteasomal degradation andinduces its binding to SUMO-1 in murine thymocytes. Cell Death Differ2011;18:183e90.

[43] Ronkina N, Menon MB, Schwermann J, Arthur JS, Legault H, Telliez JB, et al.Stress induced gene expression: a direct role for MAPKAP kinases in tran-scriptional activation of immediate early genes. Nucleic Acids Res 2011;39:2503e18.

[44] Stoecklin G, Stubbs T, Kedersha N, Wax S, Rigby WF, Blackwell TK, et al. MK2-induced tristetraprolin:14-3-3 complexes prevent stress granule associationand ARE-mRNA decay. EMBO J 2004;23:1313e24.

[45] Hayden MS, Ghosh S. Shared principles in NF-kappaB signaling. Cell2008;132:344e62.

[46] Gareus R, Kotsaki E, Xanthoulea S, van dM I, Gijbels MJ, Kardakaris R, et al.Endothelial cell-specific NF-kappaB inhibition protects mice from athero-sclerosis. Cell Metab 2008;8:372e83.

[47] Saito T, Hasegawa Y, Ishigaki Y, Yamada T, Gao J, Imai J, et al. Importance ofendothelial NF-kappaB signalling in vascular remodelling and aortic aneu-rysm formation. Cardiovasc Res 2013;97:106e14.

[48] Ayroldi E, Migliorati G, Bruscoli S, Marchetti C, Zollo O, Cannarile L, et al.Modulation of T-cell activation by the glucocorticoid-induced leucine zipperfactor via inhibition of nuclear factor kappaB. Blood 2001;98:743e53.

[49] Di Marco B, Massetti M, Bruscoli S, Macchiarulo A, Di VR, Velardi E, et al.Glucocorticoid-induced leucine zipper (GILZ)/NF-kappaB interaction: role ofGILZ homo-dimerization and C-terminal domain. Nucleic Acids Res 2007;35:517e28.

[50] Eddleston J, Herschbach J, Wagelie-Steffen AL, Christiansen SC, Zuraw BL. Theanti-inflammatory effect of glucocorticoids is mediated by glucocorticoid-induced leucine zipper in epithelial cells. J Allergy Clin Immunol 2007;119:115e22.