ISSN: 1524-4636 Copyright © 2009 American Heart Association. All rights reserved. Print ISSN: 1079-5642. Online

7272 Greenville Avenue, Dallas, TX 72514Arteriosclerosis, Thrombosis, and Vascular Biology is published by the American Heart Association.

DOI: 10.1161/ATVBAHA.109.193375 published online Sep 3, 2009; Arterioscler Thromb Vasc Biol

Harald Carlsen, Dorian O. Haskard, Justin C. Mason and Paul C. Evans Cuhlmann, Shahir S. Hamdulay, Rob Krams, Indika Edirisinghe, Irfan Rahman, Mustafa Zakkar, Kim Van der Heiden, Le Anh Luong, Hera Chaudhury, Simon

Proinflammatory State Activation of Nrf2 in Endothelial Cells Protects Arteries From Exhibiting a

http://atvb.ahajournals.org/cgi/content/full/ATVBAHA.109.193375/DC2 http://atvb.ahajournals.org/cgi/content/full/ATVBAHA.109.193375/DC1

Data Supplement (unedited) at:  

http://atvb.ahajournals.orglocated on the World Wide Web at:

The online version of this article, along with updated information and services, is

http://www.lww.com/reprintsReprints: Information about reprints can be found online at  

journalpermissions@lww.com410-528-8550. E-mail:

Fax:Kluwer Health, 351 West Camden Street, Baltimore, MD 21202-2436. Phone: 410-528-4050. Permissions: Permissions & Rights Desk, Lippincott Williams & Wilkins, a division of Wolters 

http://atvb.ahajournals.org/subscriptions/Biology is online at Subscriptions: Information about subscribing to Arteriosclerosis, Thrombosis, and Vascular

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

Activation of Nrf2 in Endothelial Cells Protects ArteriesFrom Exhibiting a Proinflammatory State

Mustafa Zakkar, Kim Van der Heiden, Le Anh Luong, Hera Chaudhury, Simon Cuhlmann,Shahir S. Hamdulay, Rob Krams, Indika Edirisinghe, Irfan Rahman, Harald Carlsen,

Dorian O. Haskard, Justin C. Mason, Paul C. Evans

Objective—Proinflammatory mediators influence atherosclerosis by inducing adhesion molecules (eg, VCAM-1) onendothelial cells (ECs) via signaling intermediaries including p38 MAP kinase. Regions of arteries exposed to high shearstress are protected from inflammation and atherosclerosis, whereas low-shear regions are susceptible. Here weinvestigated whether the transcription factor Nrf2 regulates EC activation in arteries.

Methods and Results—En face staining revealed that Nrf2 was activated in ECs at an atheroprotected region of the murineaorta where it negatively regulated p38–VCAM-1 signaling, but was expressed in an inactive form in ECs at anatherosusceptible site. Treatment with sulforaphane, a dietary antioxidant, activated Nrf2 and suppressed p38–VCAM-1signaling at the susceptible site in wild-type but not Nrf2�/� animals, indicating that it suppresses EC activation viaNrf2. Studies of cultured ECs revealed that Nrf2 inactivates p38 by suppressing an upstream activator MKK3/6 and byenhancing the activity of the negative regulator MKP-1.

Conclusions—Nrf2 prevents ECs at the atheroprotected site from exhibiting a proinflammatory state via the suppressionof p38–VCAM-1 signaling. Pharmacological activation of Nrf2 reduces EC activation at atherosusceptible sites andmay provide a novel therapeutic strategy to prevent or reduce atherosclerosis. (Arterioscler Thromb Vasc Biol. 2009;29:00-00.)

Key Words: Nrf2 � arterial endothelium � shear stress � sulforaphane � proinflammatory activation � p38� MKK3/6 � MKP-1

Early atherosclerotic lesions contain monocytes andT-lymphocytes which are recruited from the circulation

by adhesion to activated vascular endothelial cells (ECs).1

This process is triggered by proinflammatory mediators (eg,TNF�) which induce cellular adhesion molecules (eg,VCAM-1) via signaling intermediaries including p38mitogen-activated protein (MAP) kinase, which is activatedby phosphorylation by MAP kinase kinases 3 and 6 (MKK3/6).2,3 Vascular inflammation and atherosclerosis developpredominantly at distinct sites of the arterial tree located nearbranches and bends which are exposed to nonuniform bloodflow, which exerts relatively low shear stress on vascularendothelium, whereas regions of arteries that are exposed tounidirectional high shear stress are protected.4–6 Proinflam-matory activation of ECs is reduced at high-shear sitescompared to low-shear regions, thus providing a potentialexplanation for the distinct spatial localization of vascular

inflammation and lesion formation.5,7–10 Similarly, the appli-cation of unidirectional high shear stress can suppress proin-flammatory activation of cultured ECs, whereas low oroscillatory shear can act as a positive regulator of ECactivation.5,10–15

The molecular mechanisms underlying the antiinflamma-tory effects of shear stress are uncertain, but previous studiesof cultured cells have suggested a role for the transcriptionfactor Nrf2.16–20 In unstimulated cells, Nrf2 is suppressed bykelch-like ECH-associated protein 1 (Keap1) which targets itfor ubiquitination and proteasomal processing. Nrf2 can beactivated by shear stress, dietary antioxidants (eg, sulfora-phane) and other physiological stimuli which disrupt Keap1-Nrf2 interactions leading to stabilization and nuclear translo-cation of Nrf2.16–22 A previous study of cultured ECsrevealed that activated Nrf2 can induce numerousantioxidant-defense genes (eg, heme oxygenase-1 [HO-1])

Received February 16, 2009; revision accepted August 18, 2009.From the British Heart Foundation Cardiovascular Sciences Unit (M.Z., K.V.d.H., L.A.L., H.C., S.C., S.S.H., D.O.H., J.C.M., P.C.E.), National Heart

and Lung Institute, Imperial College London, UK; the Department of Bioengineering (S.C., R.K.), Imperial College London, UK; University of RochesterMedical Center (I.E., I.R.), Rochester, NY; and the Department of Nutrition (H.C.), University of Oslo, Norway.

M.Z. and K.V.d.H. contributed equally to this study.Correspondence to Dr Paul C. Evans, Senior Lecturer, British Heart Foundation Cardiovascular Sciences Unit, National Heart and Lung Institute,

Imperial College London, Hammersmith Campus, Du Cane Road, London W12 ONN, UK. E-mail paul.evans@imperial.ac.uk© 2009 American Heart Association, Inc.

Arterioscler Thromb Vasc Biol is available at http://atvb.ahajournals.org DOI: 10.1161/ATVBAHA.109.193375

1 at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

and suppress adhesion molecule expression in ECs by inhib-iting phosphorylation/activation of p38.20 However, the ef-fects of Nrf2 on EC activation have not previously beenassessed in vivo.

Materials and MethodsFull details of all methods can be found in the supplemental materials(available online at http://atvb.ahajournals.org).

In Vivo StudiesMale C57BL/6 or Nrf2�/� (C57BL/6)23 mice were treated withsulforaphane (5 mg/kg) 24 hours and 4 hours before experimenta-tion. Mice were treated with LPS (4 mg/kg) for 1 hour or 6 hoursbefore assessment of p38 phosphorylation or VCAM-1 expression,respectively. Expression levels of specific proteins were assessed inECs at regions of the inner curvature (susceptible site) and outercurvature (protected site) of aortae by en face staining asdescribed.9,10

Cell Culture StudiesHuman umbilical vein endothelial cells (HUVECs) were collectedand cultured as described previously.14 Confluent cultures wereexposed to unidirectional laminar shear (12 dynes/cm2) for 24 hoursusing a parallel-plate flow chamber as described previously14 or weretreated with sulforaphane (1 �mol/L). Gene-specific small interfer-ing (si)RNAs were used to silence Nrf2, MAP kinase phosphatase-1(MKP-1), or HO-1. Transcript levels were quantified by real-timePCR using gene-specific primers (supplemental Table) as describedpreviously.14 MKP-1 redox state was assessed by nonreducingSDS-PAGE using total cell lysates, as described.24 The expressionand intracellular localization of Nrf2 was assessed by immunostain-ing using anti-Nrf2 antibodies and Alexafluor 568-conjugated sec-ondary antibodies followed by confocal microscopy. IntracellularROS were measured in HUVECs loaded with the redox-sensitivefluorescent probe 5-(and-6)-carboxy-2�7�-dichlorodihydrofluoresce-in diacetate (H2DCFDA) by quantification of fluorescence byconfocal microscopy.

ResultsHigh Shear Stress Suppressed EC Activation at anAtheroprotected Region by Inducing Nrf2We examined whether the known spatial distribution ofshear stress in the murine aorta4 correlated with Nrf2activation in ECs. Nrf2 was localized predominantly in thenucleus of ECs in the outer curvature which is exposed tohigh shear stress and is protected from inflammation andatherosclerosis, but was localized predominantly in thecytoplasm of ECs in the inner curvature which is exposedto low shear and is atherosusceptible (supplemental FigureIA). Staining was not observed in parallel experimentsusing Nrf2�/� animals, indicating that staining was spe-cific for Nrf2. We confirmed that nuclear Nrf2 was activein the protected region because HO-1 (an Nrf2-targetmolecule) was expressed by ECs in the protected region inwild-type mice but not in Nrf2�/� mice (supplementalFigure IB). HO-1 was not detected in the susceptibleregion of wild-type mice, suggesting that cytoplasmic Nrf2is inactive at this site.

We determined whether Nrf2 regulates p38 signaling toVCAM-1 in protected regions of the aorta by performing

en face staining using antibodies that recognize activephosphorylated p38 or VCAM-1. In wild-type mice, phos-phorylated p38 and VCAM-1 were expressed at signifi-cantly lower levels in the protected site compared to thesusceptible site in both untreated and LPS-treated mice(supplemental Figures II and III). The suppression ofphosphorylated p38 in the protected region was not causedby reduced protein expression of p38 which was similarin susceptible and protected sites (supplemental FigureIV). We used Nrf2�/� mice to examine whether Nrf2 isnecessary for suppression of p38 activation and VCAM-1expression in the protected region. Genetic deletion ofNrf2 significantly enhanced p38 phosphorylation andVCAM-1 expression at the protected site in both untreatedand LPS-treated animals (supplemental Figures II and III),indicating that Nrf2 suppresses EC activation at the pro-tected region. In contrast, genetic deletion of Nrf2 hadlittle or no effect on EC activation at the susceptible site,which is consistent with our finding that Nrf2 is inactive atthis region in wild-type mice. Interestingly, levels ofphosphorylated p38 and VCAM-1 were higher in thesusceptible region compared to the protected region inNrf2�/� mice suggesting that other negative regulatorscooperate with Nrf2 to suppress EC activation at theprotected site.

It is likely that Nrf2 suppresses p38–VCAM-1 signaling atthe protected site as a consequence of shear stress becauseparallel studies performed under well defined conditions invitro indicated that high shear stress (12 dynes/cm2 for 24hours) significantly enhanced expression and nuclear local-ization of Nrf2 and simultaneously reduced expression ofVCAM-1 in HUVECs (supplemental Figure VA throughVD). Inhibition of Nrf2 activity using an adenovirus contain-ing a dominant negative version (Ad-Nrf2-DN) restoredVCAM-1 expression in sheared HUVECs (supplementalFigure VD) indicating that endogenous Nrf2 is essential forsuppression of EC activation by shear stress. Taken together,our in vitro and in vivo studies suggest that Nrf2 activation byshear stress suppresses p38 activation and VCAM-1 expres-sion at protected sites, which may be a key molecularmechanism that governs the spatial distribution of atheroscle-rotic plaques.

Sulforaphane Inhibits p38–VCAM-1 Signaling viaActivation of Nrf2We reasoned that sites exposed to low shear may besusceptible to endothelial activation and vascular inflam-mation because of the absence of Nrf2 activation in ECs.Therefore, strategies to activate Nrf2 in susceptible sitesmay suppress vascular inflammation. To address thishypothesis we used sulforaphane, an isothiocyanate de-rived from cruciferous vegetables (eg, broccoli) that canactivate Nrf2 by dissociating it from Keap1 and can induceantioxidants in cultured human ECs21 and in murine tissuesin vivo.22 We observed that sulforaphane treatment ofHUVECs did not influence Nrf2 mRNA expression (sup-plemental Table) but significantly enhanced expressionand nuclear localization of Nrf2 protein (Figure 1A and

2 Arterioscler Thromb Vasc Biol November 2009

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

1B) and simultaneously reduced the expression ofVCAM-1 at both the mRNA (Figure 1C and 1D) andprotein levels (Figure 1E). We concluded that endogenousNrf2 is essential for suppression of VCAM-1 expressionby sulforaphane because VCAM-1 expression insulforaphane-treated cells was restored by treatment withAd-Nrf2-DN (Figure 1C), and by gene silencing usingNrf2-specific siRNA (Figure 1D and 1E). Pulse-chaseexperiments revealed that sulforaphane did not influencethe stability of VCAM-1 mRNA in activated HUVECs(supplemental Figure VI), suggesting that sulforaphaneinhibits VCAM-1 at a transcriptional level. To furtherstudy the molecular mechanism underlying VCAM-1 sup-pression by Nrf2 we examined whether sulforaphaneregulates p38 or NF-�B which are essential for VCAM-1expression. We observed that pretreatment of ECs withsulforaphane suppressed activation of p38 and MKK3/6 in

response to TNF� (Figure 1F and 1G) but did not influenceNF-�B activation (Figure 1G). Ad-Nrf2-DN restored p38activation in sulforaphane-treated HUVECs (Figure 1F),indicating that sulforaphane inhibits p38 activity via Nrf2.Similarly, overexpression of Nrf2 in HUVECs (supple-mental Figure VII) using an adenovirus (A; Ad-Nrf2)suppressed induction of VCAM-1 (B) and activation ofp38 and MKK3/6 (C and D) by TNF� but did not influenceNF-�B (E).

Given that sulforaphane induces antioxidants, we hy-pothesized that it suppresses p38 activation by alteringcellular ROS. Consistent with this idea, treatment ofHUVECs with sulforaphane (Figure 1H) or Ad-Nrf2 (sup-plemental Figure VIIF) suppressed the induction of cellu-lar ROS. Moreover, screening of a panel of antioxidants byquantitative PCR revealed that sulforaphane inducesHO-1, aldo-keto reductase family 1 members C1 and C3,

Figure 1. Sulforaphane suppresses p38–VCAM-1 signaling in cultured ECs viaNrf2. HUVECs were transduced withAd-Nrf2-DN or with an empty adenovirus(Ad-empty; C and F), or were transfectedwith Nrf2-specific (D and E) or MKP-1–specific siRNA (I) or with a scrambledcontrol (Scr), or remained untreated. Cellswere then exposed to SFN or vehiclealone for 4 hours. A, Cytosolic or nuclearlysates were tested by Western blottingusing specific antibodies. Representativeof 3 experiments. B, Immunofluorescencestaining using anti-Nrf2 antibodies andnuclear counterstain (DNA). Representa-tive images and quantitation of nuclearNrf2 levels from multiple ECs (mean�SD)are shown. C through G, Cells were stim-ulated with TNF� for 15 to 30 minutes (G),1 hour (F), 4 hours (C and D), or 6 hours(E) or remained untreated. C and D, Levelsof VCAM-1 or Nrf2 transcripts were quan-tified by real-time PCR. Mean values(�SD) calculated from triplicate measure-ments are shown. Cell lysates were testedby Western blotting using anti-Nrf2 orantitubulin antibodies. Representative of 3independent experiments. E, VCAM-1 sur-face expression was assessed by immu-nostaining followed by flow cytometry.Data were pooled from 2 independentexperiments and mean RFI values (�SD)are shown. F and G, Lysates were testedby Western blotting using specific anti-bodies. phos-p38 indicates Phosphorylat-ed-p38; phos-MKK3/6, PhosphorylatedMKK3/6. Representative of 2 independentexperiments. H, HUVECs were loadedwith the redox-sensitive probe H2DCFDA,treated with H2O2 (50 �mol/L), and fluo-rescence was assessed at varying timesby confocal microscopy. Representativeimages and quantitation of fluorescentcells after 5 minutes stimulation(mean�SD; pooled from 3 experiments)are shown. I, Cells were stimulated withTNF� (4 hours) or remained untreated.Levels of MKP-1 or VCAM-1 transcriptswere quantified by real-time PCR. Datawere pooled from 2 independent experi-ments and mean values (�SD) are shown.

Zakkar et al Nrf2 Suppresses Endothelial Activation in Arteries 3

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

and thioredoxin reductase 1 (supplemental Table). As theinduction of HO-1 was most prominent, we examined itsrole in EC responses to SFN. Inhibition of HO-1 activityusing zinc protoporphyrin or silencing of HO-1 expressionusing prevalidated siRNA sequences25 did not influenceVCAM-1 suppression by sulforaphane (supplemental Fig-ure VIII), suggesting that HO-1 is not essential for theantiinflammatory effects of sulforaphane. It is plausiblethat the other Nrf2-induced antioxidants can compensatefor the absence of HO-1 activity in sulforaphane-treatedcells. We next examined whether sulforaphane suppressesp38 activation via MKP-1, an antiinflammatory negativeregulator of p38 that is expressed by ECs in atheropro-tected regions.10 Sulforaphane did not induce MKP-1expression (supplemental Table). However, silencing ofMKP-1 significantly elevated VCAM-1 expression insulforaphane-treated HUVECs (Figure 1I), indicating thatendogenous MKP-1 is involved in suppression ofVCAM-1 expression by sulforaphane. A previous studydemonstrated that ROS can inactivate MKP-1 by oxidizingcatalytic cysteine residues to sulfenic acid, thus promotingthe formation of disulfide bonds between MKP-1 andcellular proteins.24 Given that Nrf2 reduces ROS, weexamined whether Nrf2 can enhance MKP-1 activity bypromoting the reduced form. Analysis of cell lysates bySDS-PAGE in the absence of reducing agents revealedboth oxidized and reduced forms of MKP-1, which dis-played a marked difference in electrophoretic mobility(supplemental Figure VIIG), as described previously.24

Treatment of HUVECs with H2O2 induced a high-molecular-weight disulfide-linked form of MKP-1 (oxi-dized; compare 1 and 2) that could be reduced by incuba-tion of cell lysates with DTT (compare 2 and 4). Thereduced form of MKP-1 was promoted by overexpressionof Nrf2 using Ad-Nrf2 (compare 2 and 3), suggesting thatNrf2 can enhance MKP-1 catalytic activity by influencingits redox state.

Pharmacological Activation of Nrf2 Suppresses ECActivation in a Susceptible Region of theMurine AortaWe examined whether pharmacological activation of Nrf2using sulforaphane can reduce proinflammatory activation atatherosusceptible sites. Treatment of mice with sulforaphaneactivated Nrf2 in a susceptible region of the aorta (Figure 2).Sulforaphane treatment also reduced p38 activation andVCAM-1 expression at the susceptible site in wild-type mice(Figures 3 and 4, compare 5 with 7) but had no effect inNrf2�/� mice (compare 6 with 8), indicating that sulfora-phane suppresses p38–VCAM-1 signaling at the susceptiblesite by activating Nrf2. At the protected region of wild-typeanimals, Nrf2 was constitutively active and sulforaphanetreatment had relatively modest effects on Nrf2 activity(Figure 2), p38 phosphorylation (Figure 3), and VCAM-1expression (Figure 4).

DiscussionStudies using cultured ECs suggest that the molecularmechanism underlying the antiinflammatory effects of

shear stress involves the transcription factors KLF215,26,27

and Nrf2,16 –20 both of which can be activated by shearstress and can limit proinflammatory activation whenoverexpressed in ECs. Indeed KLF2 and Nrf2 can actsynergistically to modulate the transcriptome in culturedECs exposed to shear stress.28 The involvement of KLF2has been confirmed by studies of human, murine andzebrafish arteries which have demonstrated that KLF2 isexpressed at high shear regions.26,29,30 In contrast, theregulation and physiological role of Nrf2 in the vasculaturein vivo has thus far received little attention. Althoughprevious studies have revealed that Nrf2 is expressed in thedescending thoracic aorta19 and can be activated by oxi-dized phospholipids in the carotid artery,31 the capacity ofNrf2 to regulate proinflammatory activation of ECs inarteries has not been previously investigated. Here wedemonstrate that Nrf2 is constitutively active in ECs at ahigh-shear atheroprotected region of the murine aorta andthat it reduces proinflammatory activation at this site byinactivating p38 MAP kinase and suppressing VCAM-1expression. Thus Nrf2 activation by shear stress may be akey molecular mechanism that governs the spatial distri-bution of atherosclerotic plaques. Interestingly, we ob-served that genetic deletion of Nrf2 did not entirely restore

Figure 2. Sulforaphane activates Nrf2 in ECs at a susceptibleregion of the aorta. Nrf2 nuclear localization was assessed byen face staining in protected or susceptible regions in wild-typemice (treated with sulforaphane [SFN] or vehicle alone as a con-trol (n�3 per group). Endothelial marker (CD31); nuclear coun-terstain (DNA). Representative images and the proportion ofnuclear Nrf2 calculated from multiple animals (mean�SD) areshown.

4 Arterioscler Thromb Vasc Biol November 2009

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

proinflammatory activation at the protected site, suggest-ing that Nrf2 acts in concert with other antiinflammatorymolecules (eg, KLF215,26 –28) to inhibit proinflammatoryactivation in vascular endothelium exposed to high shear.

We observed that ECs at the atherosusceptible siteexpressed an inactive form of Nrf2 which was localized tothe cytoplasm and was incapable of suppressing proinflam-matory activation, and concluded that regions exposed tolow or oscillatory shear may be susceptible to inflamma-

tion attributable to the absence of Nrf2 activation in ECs.The corollary is that drugs that can activate Nrf2 indepen-dently of shear stress may be able to suppress inflamma-tion and atherosclerosis at susceptible sites. To test thisconcept we studied sulforaphane, a dietary antioxidant thatacts as a potent activator of Nrf2.21,22 Our studies ofcultured ECs revealed that a relatively low concentrationof sulforaphane (1 �mol/L) elevated Nrf2 expression at theprotein level but did not enhance Nrf2 transcript levels, a

Figure 3. Activation of Nrf2 by sulforaphane suppresses p38activation in a susceptible region of the aorta. Phosphorylatedp38 (phos-p38) levels were assessed by en face staining of pro-tected or susceptible regions in wild-type or Nrf2�/� micetreated with sulforaphane (SFN) or vehicle, and 24 hours laterchallenged with LPS (n�3). Endothelial marker (CD31); nuclearcounterstain (DNA). Representative images and quantitation ofp38 phosphorylation from multiple animals (mean�SD) areshown.

Figure 4. Activation of Nrf2 by sulforaphane suppressesVCAM-1 expression in a susceptible region of the aorta.VCAM-1 expression levels were assessed by en face staining ofprotected or susceptible regions in wild-type or Nrf2�/� micetreated with sulforaphane (SFN) or vehicle, and 24 hours laterchallenged with LPS (n�3). Endothelial marker (CD31); nuclearcounterstain (DNA). Representative images and quantitation ofVCAM-1 expression from multiple animals (mean�SD) areshown.

Zakkar et al Nrf2 Suppresses Endothelial Activation in Arteries 5

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

finding that is consistent with the observation that sulfora-phane can stabilize Nrf2 by inhibiting Keap1.32 We ob-served that sulforaphane suppressed p38 activation andVCAM-1 expression in cultured ECs and subsequentstudies, in which Nrf2 was either silenced or inactivated,revealed that Nrf2 was required for these antiinflammatoryeffects. These data are consistent with our observation thatsulforaphane suppressed endothelial activation at a suscep-tible site in wild-type mice but not in Nrf2�/� animals. Arecent study demonstrated that relatively high concentra-tions of sulforaphane can influence EC activation throughNrf2-independent mechanisms,33 and we therefore con-clude that sulforaphane may have concentration-dependenteffects on proinflammatory signaling pathways.

Our study demonstrates that Nrf2 suppresses endothelialactivation by targeting the MAP kinase signaling pathwayat 2 levels (ie, by reducing MKK3/6 signaling to p38 andby enhancing the activity of MKP-1, an antiinflammatorymolecule that inactivates p38 by dephosphorylation). Nrf2did not regulate MKP-1 at the level of expression. Instead,we provide evidence that Nrf2 can enhance the catalyticactivity of MKP-1 by promoting a reducing environmentvia the induction of multiple antioxidants. We have previ-ously demonstrated that MKP-1 is induced by shear stressand is preferentially expressed by ECs in a protectedregion of the murine aorta where it inhibits p38 –VCAM-1signaling.10 Thus we propose that shear stress suppressesEC activation at atheroprotected sites by inducing MKP-1and by simultaneously enhancing MKP-1 activity viaactivation of Nrf2. Our studies revealed that Nrf2 alsosuppresses the activation of MKK3/6, a signaling interme-diary that activates p38 by phosphorylation. The underly-ing mechanism for MKK3/6 suppression by Nrf2 mayinvolve redox regulation as well because ASK1, a MAPkinase kinase kinase that acts upstream from MKK3/6, isknown to be inhibited by reduced forms of glutathione andthioredoxin.34,35

We believe that sulforaphane is the first example of atherapeutic intervention that enhances the activity of anantiinflammatory transcription factor at atherosusceptibleregions, and that this compound may have clinical utilityfor the prevention or treatment of vascular inflammation.In addition, given that cruciferous vegetables are a sourceof sulforaphane it is plausible that their known beneficialeffects on cardiovascular health36 involve activation ofNrf2 at atherosusceptible regions. We administered sul-foraphane to mice i.p. because dosing this route can becontrolled more accurately than oral administration. Thedose used in our study was based in the efficacy ofsulforaphane established in previous studies.37,38 Furtheranimal studies are now required to assess directly whetherthe consumption of green vegetables can influence Nrf2activation and inflammation at susceptible sites. A recentstudy revealed that genetic deletion of Nrf2 reducesatherosclerosis in ApoE�/� mice, and demonstrated thatNrf2 exerts proatherogenic effects in cultured macro-phages by positively regulating the expression of CD36which is a scavenger receptor for modified LDL.39 How-ever, our current findings indicate that Nrf2 may exert

antiatherogenic effects in vascular endothelium by sup-pressing inflammation. Thus, activation of Nrf2 in endo-thelial cells and macrophages may exert opposing effectson lesion development. Studies using conditional knock-outs of Nrf2 are now required to assess the functionsof Nrf2 in specific vascular cells in relation toatherosclerosis.

Sources of FundingThe study was funded by the British Heart Foundation.

DisclosuresNone.

References1. Libby P. Inflammation in atherosclerosis. Nature. 2002;420:868–874.2. Ahmad M, Theofanidis P, Medford RM. Role of activating protein-1

in the regulation of the vascular cell adhesion molecule-1 geneexpression by tumor necrosis factor-alpha. J Biol Chem. 1998;273:4616 – 4621.

3. Pietersma A, Tilly BC, Gaestel N, deJong N, Lee JC, Koster JF, SluiterW. P38 mitogen activated protein kinase regulates endothelial VCAM-1expression at the post-transcriptional level. Biochem Biophys Res Comm.1997;230:44–48.

4. Suo J, Ferrara DE, Sorescu D, Guldberg RE, Taylor WR, Giddens DP.Hemodynamic shear stresses in mouse aortas - Implications for athero-genesis. Arterioscler Thromb Vasc Biol. 2007;27:346–351.

5. Dai GH, Kaazempur-Mofrad MR, Natarajan S, Zhang YZ, Vaughn S,Blackman BR, Kamm RD, Garcia-Cardena G, Gimbrone MA. Distinctendothelial phenotypes evoked by arterial waveforms derived from ath-erosclerosis-susceptible and -resistant regions of human vasculature. ProcNatl Acad Sci U S A. 2004;101:14871–14876.

6. Caro CG. Discovery of the role of wall shear in atherosclerosis.Arterioscler Thromb Vasc Biol. 2009;29:158 –161.

7. Passerini AG, Polacek DC, Shi CZ, Francesco NM, Manduchi E, GrantGR, Pritchard WF, Powell S, Chang GY, Stoeckert CJ, Davies PF.Coexisting proinflammatory and antioxidative endothelial transcriptionprofiles in a disturbed flow region of the adult porcine aorta. Proc NatlAcad Sci U S A. 2004;101:2482–2487.

8. Iiyama K, Hajra L, Iiyama M, Li HM, DiChiara M, Medoff BD, CybulskyMI. Patterns of vascular cell adhesion molecule-1 and intercellularadhesion molecule-1 expression in rabbit and mouse atheroscleroticlesions and at sites predisposed to lesion formation. Circ Res. 1999;85:199–207.

9. Hajra L, Evans AI, Chen M, Hyduk SJ, Collins T, Cybulsky MI. TheNF-kappa B signal transduction pathway in aortic endothelial cells isprimed for activation in regions predisposed to atherosclerotic lesionformation. Proc Natl Acad Sci U S A. 2000;97:9052–9057.

10. Zakkar M, Chaudhury H, Sandvik G, Enesa K, Luong LA, Cuhlmann S,Mason JC, Krams R, Clark AR, Haskard DO, Evans PC. Increasedendothelial mitogen-activated protein kinase phosphatase-1 expressionsuppresses proinflammatory activation at sites that are resistant to ath-erosclerosis. Circ Res. 2008;103:726–732.

11. Yamawaki H, Lehoux S, Berk BC. Chronic physiological shear stressinhibits tumor necrosis factor-induced proinflammatory responses inrabbit aorta perfused ex vivo. Circulation. 2003;108:1619–1625.

12. Sheikh S, Rainger GE, Gale Z, Rahman M, Nash GB. Exposure tofluid shear stress modulates the ability of endothelial cells to recruitneutrophils in response to tumor necrosis factor-alpha: a basis forlocal variations in vascular sensitivity to inflammation. Blood. 2003;102:2828 –2834.

13. Chiu JJ, Lee PL, Chen CN, Lee CI, Chang SF, Chen LJ, Lien SC, Ko YC,Usami S, Chien S. Shear stress increases ICAM-1 and decreasesVCAM-1 and E-selectin expressions induced by tumor necrosisfactor-alpha in endothelial cells. Arterioscler Thromb Vasc Biol. 2004;24:73–79.

14. Partridge J, Carlsen H, Enesa K, Chaudhury H, Zakkar M, Luong L,Kinderlerer A, Johns M, Blomhoff R, Mason JC, Haskard DO, Evans

6 Arterioscler Thromb Vasc Biol November 2009

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

PC. Laminar shear stress acts as a switch to regulate divergentfunctions of NF-kappa B in endothelial cells. FASEB J. 2007;21:3553–3561.

15. Fledderus JO, van Thienen JV, Boon RA, Dekker RJ, Rohlena J, VolgerOL, Bijnens APJJ, Daemen MJAP, Kuiper J, van Berkel TJC, PannekoekH, Horrevoets AJG. Prolonged shear stress and KLF2 suppress consti-tutive proinflarnmatory transcription through inhibition of ATF2. Blood.2007;109:4249–4257.

16. Chen XL, Varner SE, Rao AS, Grey JY, Thomas S, Cook CK, Was-serman MA, Medford RM, Jaiswal AK, Kunsch C. Laminar flowinduction of antioxidant response element-mediated genes in endothelialcells - A novel anti-inflammatory mechanism. J Biol Chem. 2003;278:703–711.

17. Hosoya T, Maruyama A, Kang MI, Kawatani Y, Shibata T, Uchida K,Itoh K, Yamamoto M. Differential responses of the Nrf2-Keap1 system tolaminar and oscillatory shear stresses in endothelial cells. J Biol Chem.2005;280:27244–27250.

18. Warabi E, Takabe W, Minami T, Inoue K, Itoh K, Yamamoto M, Ishii T,Kodama T, Noguchi N. Shear stress stabilizes NF-E2-related factor 2 andinduces antioxidant genes in endothelial cells: Role of reactive oxygen/nitrogen species. Free Radic Biol Med. 2007;42:260–269.

19. Dai G, Vaughn S, Zhang Y, Wang ET, Garcia-Cardena G, Gimbrone MA.Biomechanical forces in atherosclerosis-resistant vascular regionsregulate endothelial redox balance via phosphoinositol 3-kinase/akt-dependent activation of Nrf2. Circ Res. 2007;101:723–733.

20. Chen XL, Dodd G, Thomas S, Zhang XL, Wasserman MA, Rovin BH,Kunsch C. Activation of Nrf2/ARE pathway protects endothelial cellsfrom oxidant injury and inhibits inflammatory gene expression. Am JPhysiol Heart Circ Physiol. 2006;290:H1862–H1870.

21. Xue MZ, Qian QW, Adaikalakoteswari A, Rabbani N, Babaei-Jadidi R,Thornalley PJ. Activation of NF-E2-related factor-2 reverses biochemicaldysfunction of endothelial cells induced by hyperglycemia linked tovascular disease. Diabetes. 2008;57:2809–2817.

22. Thimmulappa RK, Mai KH, Srisuma S, Kensler TW, Yamamato M,Biswal S. Identification of Nrf2-regulated genes induced by the chemo-preventive agent sulforaphane by oligonucleotide microarray. CancerRes. 2002;62:5196–5203.

23. Itoh K, Chiba T, Takahashi S, Ishii T, Igarashi K, Katoh Y, Oyake T,Hayashi N, Satoh K, Hatayama I, Yamamoto M, Nabeshima Y. An Nrf2small Maf heterodimer mediates the induction of phase II detoxifyingenzyme genes through antioxidant response elements. Biochem BiophysRes Comm. 1997;236:313–322.

24. Kamata H, Honda S, Maeda S, Chang LF, Hirata H, Karin M. Reactiveoxygen species promote TNF alpha-induced death and sustained JNKactivation by inhibiting MAP kinase phosphatases. Cell. 2005;120:649–661.

25. Ali F, Zakkar M, Karu K, Lidington EA, Hamdulay SS, Boyle JJ, Zloh M,Bauer A, Haskard DO, Evans PC, Mason JC. Induction of the cytopro-tective enzyme heme oxygenase-1 by statins is enhanced in vascularendothelium exposed to laminar shear stress and impaired by disturbedflow. J Biol Chem. 2009;284:18882–18892.

26. Dekker RJ, van Soest S, Fontijn RD, Salamanca S, de Groot PG,VanBavel E, Pannekoek H, Horrevoets AJG. Prolonged fluid shear stress

induces a distinct set of endothelial cell genes, most specifically lungKruppel-like factor (KLF2). Blood. 2002;100:1689–1698.

27. Senbanerjee S, Lin ZY, Atkins GB, Greif DM, Rao RM, Kumar A,Feinberg MW, Chen ZP, Simon DI, Luscinskas FW, Michel TM,Gimbrone MA, Garcia-Cardena G, Jain MK. KLF2 is a novel transcrip-tional regulator of endothelial proinflammatory activation. J Exp Med.2004;199:1305–1315.

28. Fledderus JO, Boon RA, Volger OL, Hurttila H, Yla-Herttuala S,Pannekoek H, Levonen AL, Horrevoets AJ. KLF2 primes the antioxidanttranscription factor Nrf2 for activation in endothelial cells. ArteriosclerThromb Vasc Biol. 2008;28:1339–1346.

29. Parmar KM, Larman HB, Dai GH, Zhang YH, Wang ET, Moorthy SN,Kratz JR, Lin ZY, Jain MK, Gimbrone MA, Garcia-Cardena G. Inte-gration of flow-dependent endothelial phenotypes by Kruppel-like factor2. J Clin Invest. 2006;116:49–58.

30. Lee JS, Yu C, Shin JT, Sebzda E, Bertozzi C, Chen M, Mericko P,Stadtfeld M, Zhou D, Cheng L, Graf T, Macrae CA, Lepore JJ, Lo CW,Kahn ML. KIf2 is an essential regulator of vascular hemodynamic forcesin vivo. Dev Cell. 2006;11:845–857.

31. Jyrkkanen HK, Kansanen E, Inkala M, Kivela AM, Hurttila H, HeinonenSE, Goldsteins G, Jauhiainen S, Tiainen S, Makkonen H, Oskolkova O,Afonyushkin T, Koistinaho J, Yamamoto M, Bochkov VN, Yla-HerttualaS, Levonen AL. Nrf2 regulates antioxidant gene expression evoked byoxidized phospholipids in endothelial cells and murine arteries in vivo.Circ Res. 2008;103:e1–e9.

32. Eggler AL, Liu G, Pezzuto JM, van Breemen RB, Mesecar AD. Mod-ifying specific cysteines of the electrophile-sensing human Keap1 proteinis insufficient to disrupt binding to the Nrf2 domain Neh2. Proc NatlAcad Sci U S A. 2005;102:10070–10075.

33. Chen X-L, Dodd G, Kunsch C. Sulforaphane inhibits TNF-alpha-inducedactivation of p38 MAP kinase and VCAM-1 and MCP-1 expression inendothelial cells. Inflamm Res. 2009;58:513–521.

34. Hojo Y, Saito Y, Tanimoto T, Hoefen RJ, Baines CP, Yamamoto K,Haendeler J, Asmis R, Berk BC. Fluid shear stress attenuates hydrogenperoxide-induced c-jun NH2-terminal kinase activation via a glutathionereductase-mediated mechanism. Circ Res. 2002;91:712–718.

35. Liu YM, Yin GY, Surapisitchat J, Berk BC, Min W. Laminar flowinhibits TNF-induced ASK1 activation by preventing dissociation ofASK1 from its inhibitor 14-3-3. J Clin Invest. 2001;107:917–923.

36. Yochum L, Kushi LH, Meyer K, Folsom AR. Dietary flavonoid intakeand risk of cardiovascular disease in postmenopausal women. Am JEpidem. 1999;149:943–949.

37. Zhao X, Sun G, Zhang J, Strong R, Dash PK, Kan YW, Grotta JC,Aronowski J. Transcription factor Nrf2 protects the brain fromdamage produced by intracerebral hemorrhage. Stroke. 2007;38:3280 –3286.

38. Tanito M, Masutani H, Kim Y-C, Nishikawa M, Ohira A, Yodoi J.Sulforaphane induces thioredoxin through the antioxidant-responsiveelement and attenuates retinal light damage in mice. Invest Opthalmol VisSci. 2005;46:979–987.

39. Sussan TE, Jun J, Thimmulappa R, Bedja D, Antero M, Gabrielson KL ,.Polotsky VY, Biswal S. Disruption of Nrf2, a key inducer of antioxidantdefenses, attenuates ApoE-mediated atherosclerosis in mice. PLoS ONE.2008;3:e3791.

Zakkar et al Nrf2 Suppresses Endothelial Activation in Arteries 7

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

SUPPLEMENT MATERIAL

MATERIALS AND METHODS

Animals Male C57BL/6 mice between 2 and 3 months of age were used. The Nrf2

knockout mouse strain (Nrf2-/- (C57BL/6))1 was generously supplied by Prof. Masayuki

Yamamoto, University of Tsukuba, Japan via the RIKEN BioResource Center,

Tsukuba, Japan. Mice were treated with sulforaphane (5 mg/kg in 10% corn-oil/ 90%

PBS) by intraperitoneal injections 24h and 4h prior to experimentation. Mice were

treated with LPS (4mg/kg) by intraperitoneal injection for 1 hour or 6 hours prior to

assessment of p38 phosphorylation or VCAM-1 expression, respectively. All

experiments were performed within guidelines set out by the Federation of European

Laboratory Animal Science Associations.

Reagents and antibodies Human recombinant TNFα and LPS (R&D) and anti-

phosphorylated-MKK3/6 (Cell signalling Technology), anti-phosphorylated-p38

Tyr180/Thr182 (Cell signalling Technology), anti-p38 (Cell Signaling Technology),

anti-RelA (Santa Cruz), anti-CD31-FITC (BD Biosciences Pharmingen), anti-Nrf2

(Santa Cruz Biotechnology), anti-HO-1 (Santa Cruz Biotechnology), anti-tubulin

(Sigma Aldrich) and anti-lamin B (Affinity BioReagents) antibodies were obtained

commercially. TNFα was used at a final concentration of 10ng/ml. The generation of

anti-VCAM-1 (1.G11) antibodies was described previously2. Anti-MKK3/6 antibodies3

were generously supplied by Dr Jonathan Dean, Imperial College London, UK.

Adenoviruses containing wild-type Nrf2 (Ad-Nrf2) or a dominant negative version that

lacks the transactivation domain (Ad-Nrf2-DN) were generously provided by Prof.

Jeffrey A. Johnson, University of Wisconsin-Madison, USA and have been described

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

previously4. Other reagents were purchased from Sigma Aldrich unless otherwise

stated.

En face staining The expression levels of specific proteins were assessed in EC at

regions of the inner curvature (susceptible site) and outer curvature (protected site) of

murine aortae by en face staining as described previously5,6. Animals were sacrificed by

CO2 inhalation. Aortae were perfused in situ with PBS (at a pressure of approximately

100mm Hg) and then perfusion-fixed with 2% formalin prior to harvesting. Fixed

aortae were tested by immunostaining using specific primary antibodies and

Alexafluor568-conjugated secondary antibodies (red). EC were identified by co-

staining using anti-CD31 antibodies conjugated to the fluorophore FITC (green). Nuclei

were identified using a DNA-binding probe with far-red emission (To-Pro-3;

Invitrogen). Stained vessels were mounted prior to visualization of endothelial surfaces

en face using confocal laser-scanning microscopy (Zeiss LSM 510 META). Isotype-

matched monoclonal antibodies raised against irrelevant antigens or pre-immune rabbit

sera were used as experimental controls for specific staining (data not shown). The

expression of particular proteins at each site was assessed by quantification of

fluorescence intensity for multiple cells (at least 100 per site) using LSM 510 software

(Zeiss) and calculation of mean fluorescence intensities with standard deviations.

Endothelial cell culture Human umbilical vein endothelial cells (HUVEC) were

collected using collagenase and cultured as described previously7. Confluent HUVEC

cultures were exposed to high unidirectional laminar shear (12 dynes/cm2) for 24h using

a parallel-plate flow chamber (Cytodyne) as described previously7. A stock solution of

sulforaphane (1 mM in DMSO) was used to treat HUVEC to achieve a final

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

concentration of 1 μM sulforaphane. Alternatively, HUVEC were treated with an

equivalent volume of DMSO alone as a control.

Gene silencing. RNA interference was carried out using small interfering (si)RNAs that

specifically target Nrf2 (5'-AAGGATTATTATGACTGTTAA-3'; Qiagen), MAP kinase

phosphatase-1 (MKP-1) (5’-AAGCUGGACGAGGCCUUUGAGUU-3’; Dharmacon)

or HO-1 (5’-UGCUGAGUUCAUGAGGAACUU-3’; Dharmacon). Alternatively, cells

were treated with non-targeting scrambled controls (Silencer Negative Control no. 1

siRNA; Ambion, Foster City, Calif, or non-targeting siRNA no.1; Thermo Fisher

Scientific). Cell cultures that were 80% to 90% confluent were transfected with siRNA

(5 μM final concentration) by microporation (Digital BioTechnology, Seoul, Korea)

following the manufacturer’s instructions and then incubated in growth medium without

antibiotics for 48h before analysis.

Comparative real time PCR Transcript levels were quantified by comparative real-

time PCR using gene-specific primers (Supplementary Table). Extraction and reverse

transcription of total RNA and real-time PCR were carried out as described previously7.

Reactions were performed in triplicate. Relative gene expression was calculated by

comparing the number of thermal cycles that were necessary to generate threshold

amounts of product as described previously7.

Flow cytometry HUVEC were harvested and incubated on ice for 1 hour with mouse

IgG antibodies specific for VCAM-1. After washing and application of fluorescein

isothiocyanate-conjugated secondary antibody, cells were analysed by flow cytometry

after gating out dead cells. The relative fluorescence index (RFI) was calculated after

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

subtracting background fluorescence measured using secondary-only controls.

Western blotting Cytosolic or nuclear lysates were prepared using the Nuclear

Extraction Kit (Active Motif). MKP-1 redox state was assessed by non-reducing SDS-

PAGE using total cell lysates, as described8. Western blotting was carried out using

specific primary antibodies, horse radish peroxidase-conjugated secondary antibodies

and chemiluminescent detection.

Assay of NF-κB DNA binding Binding of RelA NF-κB sub-units to consensus

oligonucleotides was assessed by DNA-binding ELISA (Active Motif) using nuclear

lysates prepared using the NucBuster kit (Novagen).

Immunofluorescent staining of Nrf2 in HUVEC The expression and intracellular

localization of Nrf2 in cultured HUVEC was assessed by immunostaining of methanol-

fixed cells using anti-Nrf2 antibodies and Alexafluor 568-conjugated secondary

antibodies followed by laser-scanning confocal microscopy (LSM 510 META; Zeiss).

Nuclei were identified using the DNA-binding probe To-Pro-3. Image analysis was

performed using Zeiss LSM 510 META software to calculate average fluorescence

values after subtracting background fluorescence values from cells stained with

secondary antibody alone.

Measurement of intracellular reactive oxygen species (ROS). Intracellular ROS were

measured in HUVEC cultured on glass slide chambers. They were incubated with the

redox-sensitive fluorescent probe 5-(and-6)-carboxy-2´7´-dichlorodihydrofluorescein

diacetate (H2DCFDA, 10 μM; Invitrogen) for 45 minutes and then washed with

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

phosphate-buffered saline prior to experimentation. Fluorescence was quantified in live

cells using a laser-scanning confocal microscope with a specialized stage that

maintained cultures at 37oC in a 5% CO2 environment (LSM 510 META; Zeiss).

Statistics Differences between samples were analysed using an unpaired Student’s t-

test or one-way ANOVA with Bonferroni adjustment (*p<0.05, **p<0.01,

***p<0.001).

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

REFERENCES FOR SUPPLEMENT MATERIAL

(1) Itoh K, Chiba T, Takahashi S, Ishii T, Igarashi K, Katoh Y, Oyake T, Hayashi N,

Satoh K, Hatayama I, Yamamoto M, Nabeshima Y. An Nrf2 small Maf heterodimer

mediates the induction of phase II detoxifying enzyme genes through antioxidant

response elements. Biochem Biophys Res Comm. 1997;236:313--322.

(2) Thornhill MH, Wellicome SM, Mahiouz DL, Lanchbury JSS, Kyanaung U, Haskard

DO. Tumor-Necrosis-Factor Combines with Il-4 Or Ifn-Gamma to Selectively Enhance

Endothelial-Cell Adhesiveness for T-Cells - the Contribution of Vascular Cell-

Adhesion Molecule-1-Dependent and Molecule-1-Independent Binding Mechanisms. J

Immunol 1991; 146:592-8.

(3) Lasa M, Brook M, Saklatvala J, Clark AR. Dexamethasone destabilizes

cyclooxygenase 2 mRNA by inhibiting mitogen-activated protein kinase p38.

Molecular and Cellular Biology 2001; 21(3): 771-80.

(4) Kraft AD, Johnson DA, Johnson JA. Nuclear factor E2-related factor 2-dependent

antioxidant response element activation by tert-butylhydroquinone and sulforaphane

occurring preferentially in astrocytes conditions neurons against oxidative insult. J

Neurosci 2004;24(5):1101-12.

(5) Hajra L, Evans AI, Chen M, Hyduk SJ, Collins T, Cybulsky MI. The NF-kappa B

signal transduction pathway in aortic endothelial cells is primed for activation in

regions predisposed to atherosclerotic lesion formation. Proc Natl Acad Sci USA.

2000;97:9052--9057.

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

(6) Zakkar M, Chaudhury H, Sandvik G, Enesa K, Luong LA, Cuhlmann S, Mason JC,

Krams R, Clark AR, Haskard DO, Evans PC. Increased endothelial mitogen-activated

protein kinase phosphatase-1 expression suppresses proinflammatory activation at sites

that are resistant to atherosclerosis. Circ Res. 2008;103:726--732.

(7) Partridge J, Carlsen H, Enesa K, Chaudhury H, Zakkar M, Luong L, Kinderlerer A,

Johns M, Blomhoff R, Mason JC, Haskard DO, Evans PC. Laminar shear stress acts as

a switch to regulate divergent functions of NF-kappa B in endothelial cells. FASEB J.

2007;21:3553--3561.

(8) Kamata H, Honda S, Maeda S, Chang LF, Hirata H, Karin M. Reactive oxygen

species promote TNF alpha-induced death and sustained JNK activation by inhibiting

MAP kinase phosphatases. Cell. 2005;120:649--661.

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

LEGENDS FOR SUPPLEMENTARY TABLE AND FIGURES

Supplementary Table

Effects of sulforaphane on gene expression. HUVEC were treated with SFN (1 μM)

or with vehicle alone and incubated for 4 hours. Levels of particular transcripts were

quantified by real-time PCR using gene-specific primers (sequences indicated) and

were normalised by measuring β-actin transcript levels. Mean values (+/- SD)

calculated from two independent experiments are shown.

Supplementary Fig. I

Nrf2 is activated in EC at a protected site in the aorta. Nrf2 expression levels and

intracellular localization (A) or HO-1 expression levels (B) were assessed in EC at

protected (P) or susceptible (S) regions of the aorta in wild-type or Nrf2-/- mice by en

face staining using anti-Nrf2 or anti-HO-1 antibodies. Endothelial marker (CD31);

nuclear counterstain (DNA). Representative images and quantitation of nuclear Nrf2

(A) and HO-1 (B) from multiple animals are shown.

Supplementary Fig. II

Nrf2 suppresses p38 activation in EC in a protected region of the aorta.

Phosphorylated p38 (phos-p38) levels were assessed by en face staining of protected or

susceptible regions in wild-type or Nrf2-/- mice (treated with LPS or untreated; n=4 per

group). Endothelial marker (CD31); nuclear counterstain (DNA). Representative

images and quantitation of p38 phosphorylation from multiple animals are shown.

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

Supplementary Fig. III

Nrf2 suppresses VCAM-1 expression in EC in a protected region of the aorta.

VCAM-1 expression levels were assessed by en face staining of protected or

susceptible regions in wild-type or Nrf2-/- mice (treated with LPS or untreated; n=4 per

group). Endothelial marker (CD31); nuclear counterstain (DNA). Representative

images and quantitation of VCAM-1 expression from multiple animals are shown.

Supplementary Fig. IV

Expression levels of p38 are similar in protected and susceptible regions of wild-

type and Nrf2-/- mice. p38 expression levels were assessed by en face staining of

protected or susceptible regions in wild-type or Nrf2-/- mice. Endothelial marker

(CD31); nuclear counterstain (DNA). Representative images and quantitation of p38

expression from multiple animals are shown.

Supplementary Fig. V

Nrf2 is essential for suppression of pro-inflammatory activation by shear stress in

cultured EC. HUVEC were exposed to unidirectional shear stress for 24 hours or were

cultured under static conditions. (A) Levels of Nrf2 transcripts were quantified by real-

time PCR. Mean values (+/- SD) calculated from three independent experiments are

shown. (B) Cytosolic or nuclear lysates were tested by Western blotting using anti-

Nrf2, anti-tubulin or anti-lamin B antibodies. Representative of three independent

experiments. (C) Nrf2 expression and intra-cellular localization was assessed by

immunofluorescence staining using anti-Nrf2 antibodies and nuclear counterstaining

(DNA). Representative images and mean nuclear Nrf2 levels (+/- SD) calculated from

multiple EC are shown. (D) HUVEC were transduced with Ad-Nrf2-DN or with an

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

empty adenovirus (Ad-empty) or remained untreated and were incubated for 24 hours.

Cells were then either stimulated with TNFα for 4 hours or remained untreated. Levels

of VCAM-1 transcripts were quantified by real-time PCR. Mean values (+/- SD)

calculated from three independent experiments are shown.

Supplementary Fig. VI

Sulforaphane does not influence VCAM-1 mRNA stability

The effect of sulforaphane (SFN) on the stability of VCAM-1 mRNA was assessed by

pulse-chase experiments. HUVEC were stimulated with TNFα for 4 hours or remained

untreated. Activated cells were subsequently incubated with actinomycinD (50 μM) to

block transcription, either in the presence of SFN (1 μM) or vehicle alone. VCAM-1

transcript levels were then measured at various time points by real-time PCR. Mean

values (+/- SD) calculated from two independent experiments are shown.

Supplementary Fig. VII

Nrf2 suppresses p38-VCAM-1 signalling

(A) The effect of an Nrf2-containing adenovirus (Ad-Nrf2) on HO-1 expression levels

was determined. HUVEC were transduced with Ad-Nrf2 or with an empty adenovirus

(Ad-empty) and incubated for 24 hours. Cytosolic lysates were tested by Western

blotting using anti-Nrf2, anti-HO-1 or anti-tubulin antibodies. Representative of three

independent experiments. (B, C, D, E, F) The effect of Nrf2 overexpression on pro-

inflammatory signalling and on the induction of ROS was determined. HUVEC were

transduced with Ad-Nrf2, Ad-Nrf2-DN or Ad-empty and were incubated for 24 hours,

or were untreated. Cells were then stimulated with TNFα for 30 minutes (D), 4 hours

(B) or varying times (C, E). (B) Levels of VCAM-1 transcripts were quantified by real-

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

time PCR. Mean values (+/- SD) calculated from triplicate measurements are shown.

Representative of three independent experiments. (C, D) Cell lysates were tested by

Western blotting using antibodies that recognise phosphorylated p38 (phos-p38), total

p38, phosphorylated MKK3/6 or total MKK3/6. Representative of three independent

experiments. (E) NF-κB activation was assessed by DNA-binding ELISA. Mean optical

densities (450 nm) calculated from triplicate wells are shown with standard deviations.

Representative of three independent experiments. (F) HUVEC were loaded with the

redox-sensitive probe H2DCFDA, treated with H2O2 (50 μM) and fluorescence was

assessed at varying times by confocal microscopy. Representative images and

quantitation of fluorescent cells after 5 minutes stimulation (mean +/- SD; pooled from

3 experiments) are shown. (G) The effect of Nrf2 expression on MKP-1 redox state was

determined. Cells were treated with 0.3 μM H2O2 for 10 minutes or remained untreated

as a control. Total cell lysates were analysed by SDS-PAGE in the presence or absence

of DTT (as indicated) followed by Western blotting using anti-MKP-1 antibodies.

Oxidised and reduced forms of MKP-1 are indicated. NS, non-specific band.

Supplementary Fig. VIII

HO-1 activity is not required for suppression of VCAM-1 expression by

sulforaphane. The role of HO-1 in EC responses to SFN was assessed. (A) HUVEC

were treated with zinc protoporphyrin (ZnPP; 10 μM for 24 hours) which is a

pharmacological inhibitor of HO-1 or with vehicle alone. (B) Alternatively, HUVEC

were transfected with HO-1 specific siRNA or with a scrambled control (Scr). (A, B)

Cells were then exposed to SFN (1 μM) or vehicle alone for 4 hours, and then

stimulated with TNFα for 4 hours or remained untreated. Levels of HO-1 or VCAM-1

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

transcripts were quantified by real-time PCR. Mean values (+/- SD) calculated from

triplicate measurements are shown.

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

SUPPLEMENTARY TABLE

Vehicle SFN Forward primer Reverse primer

HO-1 * 1.48 ± 0.71 6.60 ± 1.56 5’-CGAAACAAGCAGAACCCAGTCT-3’ 5’-AGCCCTTCGGTGCAGCT-3’

Aldo-keto reductase fam 1 C3 *** 1.00 ± 0.00 2.69 ± 0.07 5’-ATTAACATGCAGGGCTTTGC-3’ 5’-AGTCGCTAAACAGGACGGAT-3’

Aldo-keto reductase fam 1 C1 * 1.19 ± 0.28 2.45 ± 0.07 5’-AACAAGCCAGGGCTACAAGT-3’ 5’-GACTTGCAGAAATCCAGCAG-3’

Thioredoxin reductase 1 * 1.08 ± 0.14 1.79 ± 0.14 5’-CCACTGGTGAAAGACCACGTT-3’ 5’-AGGAGAAAAGATCATCACTGCTGAT-3’

Aldo-keto reductase fam 1 C2 1.34 ± 0.50 1.99 ± 0.42 5’-GCTGGATTTCTGCAAGTCAA-3’ 5’-GTCCACCCATGGTTCTTCTC-3’

UDP-Glucose dehydrogenase 1.13 ± 0.21 1.42 ± 0.35 5’-CATCCAGGTGTTTCAGAGGA-3’ 5’-GCAAATAACAACAGCATGGG-3’

serum- and glucose-regulated kinase 1 1.14 ± 0.21 1.39 ± 0.00 5’-GCTCGTTTCTATGCTGCTGA-3’ 5’-AGGACAATGTGTCCCTGTGA-3’ Ferritin light chain 1.05 ± 0.07 1.38 ± 0.35 5’-GCGCGAGGGCTACGAG-3’ 5’-CTATTGGCTGGAGGGAGAGG-3’

γ-glutamylcysteine synthetase heavy chain

1.14 ± 0.21 1.19 ± 0.78 5’-TGAGGCCAACATGCGAAAAC-3’ 5’-AAATCACTCCCCAGCGACAATC-3’

Thioredoxin 1.19 ± 0.28 1.23 ± 0.35 5’-AGGGACAAAAGGTGGGTGAAT-3’ 5’-AGTTTAAATAGCCAATGGCTGGTT-3’

Microsomal epoxide hydrolase 1.09 ± 0.14 1.09 ± 0.14 5’-CTGGCCCGGCTCTTTCTACGA-3’ 5’-CAGCAGCTCCACATCCCTCTCA-3’

NAD(P)H:quinone oxidoreductase 1 0.88 ± 0.14 1.07 ± 0.28 5’-TCATTCTCTGGCCAATTCAGAG-3’ 5’-GGAGTGTGCCCAATGCTATAT-3’

Ferritin heavy chain 0.91 ± 0.14 1.08 ± 0.07 5’-CGACCGCGTCCACCTCG-3’ 5’-CTTTCATTATCACTGTCTCCC-3’

Manganese superoxide dismutase 1.07 ± 0.07 1.00 ± 0.42 5’-CGACCTGCCGTACGACTAT-3’ 5’-TTCAGGTTGTTCACGTAGGC-3’

glutathione S-transferase P1-1 0.92 ± 0.14 0.82 ± 0.14 5’-GCTCTATGGGAAGGACCAG-3’ 5’-CTCAAAAGGCTTCAGTTGC-3’

Nrf2 1.00 ± 0.00 0.86 ± 0.13 5’-TACTCCCAGGTTGCCCACA-3’ 5’-CATCTACAAACGGGAATGTCT-3’

MKP-1 1.00 ± 0.00 0.66 ± 0.14 5’-CAGCTGCTGCAGTTTGAGTC-3’ 5’-AGGTAGCTCAGCGCACTGTT-3’

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

Fluo

resc

ence

inte

nsity

Nuclear Nrf2

Supplementary Fig. I

Protected (P)Susceptible (S)A

mergeDNA

CD31 Nrf2

mergeDNA

CD31 Nrf2

mergeDNA

CD31 HO-1

Nrf2 -/-Wild-type

125 -

100 -

75 -

50 -

25 -

0 -

***

Wild-type Nrf2-/-S P P

Fluo

resc

ence

inte

nsity

mergeDNA

CD31 HO-1Protected (P)

mergeDNA

CD31 HO-1Susceptible (S) Protected (P)B

Nrf2 -/-Wild-type

Protected (P)

mergeDNA

CD31 Nrf2

20

15

10

0

5

25

Wild-type Nrf2-/-S P P

***

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

mergeDNA

CD31 phos-p38 7

mergeDNA

CD31 phos-p38 3

mergeDNA

CD31 phos-p38 4

mergeDNA

CD31 phos-p38 8

+ LPS

mergeDNA

CD31 phos-p38 1

mergeDNA

CD31 phos-p38 2

mergeDNA

CD31 phos-p38 5

mergeDNA

CD31 phos-p38 6

Prot

ecte

d

Nrf2 -/-

untreated

Wild-type

Supplementary Fig. II

Susc

eptib

le

+ LPS

untreated

Fluo

resc

ence

inte

nsity

LPS

0

50

100

150

200

250 Wild typeNrf2 -/-

1 2 3 4 5 6 7 8

***

***

- - + + - - + +Protected Susceptible

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

mergeDNA

CD31 VCAM-1 1

5

mergeDNA

CD31 VCAM-1 7

mergeDNA

CD31 VCAM-1 2

mergeDNA

CD31 VCAM-1 3

mergeDNA

CD31 VCAM-1 4

mergeDNA

CD31 VCAM-1

mergeDNA

CD31 VCAM-1 6

mergeDNA

CD31 VCAM-1 8

Nrf2 -/-

+ LPS

untreated

Wild-type

Supplementary Fig. III

Fluo

resc

ence

inte

nsity

0

50

100

150

200

250 Wild typeNrf2 -/-

***

***

1 2 3 4 5 6 7 8LPS - - + + - - + +

Protected Susceptible

+ LPS

untreated

Prot

ecte

d Su

scep

tible

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

Supplementary Fig. IV

mergeDNA

CD31 p38 1

mergeDNA

CD31 p38 3

mergeDNA

CD31 p38 2

mergeDNA

CD31 p38 4

Susceptible

Protected

Nrf2 -/-Wild-type

Fluo

resc

ence

inte

nsity

0

10

20

30 Wild typeNrf2 -/-

Protected Susceptible1 2 3 4

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

shear

Ad-Nrf2-DNAd-empty

0 -

1 -

3 -

5 -

7 -

**

++ −−− +

+− +

D

VCAM

-1 m

RN

A le

vels

A

lamin B

Nrf2

shear

B

nuclear

tubulin

Nrf2cytosolic

0 -

1.5 -

*

3 -

Supplementary Fig. V

CNrf2

DNA

merge

Nrf2

DNA

merge

shearstatic

0

50

100

150

200

250

static shear

Fluo

resc

ence

inte

nsity

Nuclear Nrf2

Nrf

2 m

RN

A le

vels

static shear static

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

Supplementary Fig. VI

0

200

400

600

800

1000

1200

DMSOSFN

VCA

M-1

mR

NA

leve

ls600

800

1000

400

200

0TNFα − + + + + +

Chase time (min) 0 30 45 60 90

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

Supplementary Fig. VII

C

TNFα (min)Ad-Nrf2-DNAd-empty

0 10 15 30 60 90 0 10 15 30 60 90 0 10 15 30 60 90 0 10 15 30 60 90

p38

phos-p38

Ad-Nrf2No adenovirus

HO-1

Nrf2

MOI 500 100 200 400 500

tubulin

Ad-empty Ad-Nrf2

A

TNFα + +− −

Ad-empty Ad-Nrf2

160 -

80 -

0 -

VCA

M-1

mR

NA

leve

ls

**

B

E

0

1

2

3

NF-

κB D

NA

bind

ing

TNFα (min)

Ad-empty

0 15 15

+ + -Ad-Nrf2 - - +

+ -- +

30 30F

Tim

e (m

in)

Ad-empty Ad-Nrf2

0

2

4

80

50

100

150

200

250

Ad empty Nrf2

**

Posi

tive

cells

TNFαAd-empty

phos-MKK3/6

- + ++ + -

Ad-Nrf2 - - +

MKK3/6

D

G97 kD

64 kD

51 kD

39 kD

Ad-emptyAd-Nrf2

H2O2

DTT

- oxidised

- reduced

- ns

- ns

+ +- -

- ++ -

- + + +- - - +1 2 3 4

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

ATVB/2009/193375/R1

Supplementary Fig. VIII

Data 3

vcam-1 data with ho-1 sirna

0

25

50

75

100

125

100

150

200

250

300

VCAM

-1 m

RN

A le

vels *A

B

0

25

50

75

100

125

0

50

TNFα − + + +SFN − − − +

ZnPP − − + +

TNFα

ScrHO-1

−+ + +

− − − +

+ ++−SFN + +−−

0

10

20

VCAM

-1 m

RN

A le

vels

HO

-1 m

RN

A le

vels

**

*

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

Nrf2

lamin B

SFNvehicle

A

nuclear

tubulin

Nrf2cytosolic

vcam-1-nrf2sirna data for paper

Scr-tnf

Scr+tnf

Scr+tnf+1

sfn

Sirna+

tnf+1sfn

010

20

30

40

50

60

VCA

M-1

mR

NA

lev

els

C

TNFα

Ad-Nrf2-DNAd-empty

- + +SFN - - +

- + +- - +

*

Scr -+- +Nrf2

0

0.5 -

1-

Nrf

-2 m

RN

A l

evel

s*

D

0

20

40

60

80

100

TNFαSFN

ScrNrf2

VCA

M-1

mR

NA

lev

els *

*

-+ + +- - - +

+ +--- + + +

E

0

1

2

3

4

5

*

*

TNFαSFN

Scr

+ +--- + + +

-+ + +

F

TNFα

Ad-Nrf2-DNAd-empty

p38

phos-p38

- + +SFN - - +

- + +- - +

G

TNFα

Scr

Data 2

0

6

9

MK

P-1

mR

NA

lev

els

2

4

8 *I

+ + -0

100

200

300

VCA

M-1

mR

NA

lev

els

*

*

TNFαSFN

ScrMKP-1

+ + -+- - +-+ + +-- + +-

MKP-1 - - +- + +

VCA

M-1

exp

ress

ion

(RFI

)

Tim

e (m

in)

vehicle SFN

0

2

4

8

H

0

20

40

60

80

100

vehicle SFN

Pos

itive

cel

ls

**

B SFNvehicle

0

1020

304050

6070

Vehicle SFN

Nrf2

DNA

merge

Nrf2

DNA

merge

*

Fluo

resc

ence

inte

nsity

Nuclear Nrf2

NF-κB

lamin B

TNFα (mins)SFN - + -

0 15 15 30 30- +

phos-MKK3/6

MKK3/6Nrf2 - - - +

Nrf2

Tubulin

nuclear

total

total

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

Fig. 2

mergeDNA

CD31 Nrf2

mergeDNA

CD31 Nrf2

mergeDNA

CD31 Nrf2

mergeDNA

CD31 Nrf2

vehicle

SFN

ProtectedSusceptible

Fluo

resc

ence

inte

nsity

Nuclear Nrf2***

010203040506070

SFN - + - +susceptible protected

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

Fig. 3

mergeDNA

CD31 phos-p38 2

mergeDNA

CD31 phos–p38 1

mergeDNA

CD31 phos-p38 4

mergeDNA

CD31 phos-p38 3

mergeDNA

CD31 phos-p38 6

mergeDNA

CD31 phos–p38 5

mergeDNA

CD31 phos-p38 8

mergeDNA

CD31 phos-p38 7

Nrf2 -/-Wild-type

vehicle

SFN

vehicle

SFN

Fluo

resc

ence

inte

nsity

050

100150

200

250***

SFN

Wild type

Nrf2 -/-

1 2 3 4 5 6 7 8- - + + - - + +

Protected Susceptible

Prot

ecte

d Su

scep

tible

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from

Fig. 4

Nrf2 -/-Wild-type

vehicle

SFN

vehicle

SFN

mergeDNA

CD31 VCAM-1 1

mergeDNA

CD31 VCAM-1 3

mergeDNA

CD31 VCAM-1 2

mergeDNA

CD31 VCAM-1 4

mergeDNA

CD31 VCAM-1 5

mergeDNA

CD31 VCAM-1 7

mergeDNA

CD31 VCAM-1 6

mergeDNA

CD31 VCAM-1 8

Fluo

resc

ence

inte

nsity

***Wild type

Nrf2 -/-

***

0

50

100

150

SFN1 2 3 4 5 6 7 8- - + + - - + +

Protected Susceptible

Prot

ecte

d Su

scep

tible

at VA MED CTR BOISE on May 10, 2011 atvb.ahajournals.orgDownloaded from