© 2006 The Authors

Genes to Cells (2006)

11

, 791–803

Journal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

791

DOI: 10.1111/j.1365-2443.2006.00979.x

Blackwell Publishing IncMalden, USAGTCGenes to Cells1365-2443© Blackwell Publishing Ltd? 2006117Original ArticleControl of cytoprotection by Bach1Y Yano et al.

Genetic ablation of the transcription repressor Bach1 leads to myocardial protection against ischemia/reperfusion in mice

Yoko Yano

1,2

, Ryoji Ozono

1,

*, Yoshihiko Oishi

1

, Masayuki Kambe

1

, Masao Yoshizumi

3

, Takafumi Ishida

4

, Shinji Omura

2

, Tetsuya Oshima

1

and Kazuhiko Igarashi

2,a

1

Departments of Clinical Laboratory Medicine,

2

Biomedical Chemistry,

3

Cardiovascular Physiology and Medicine, and

4

Medicine and Molecular Science, Hiroshima University Graduate School of Biomedical Sciences, 1-2-3 Kasumi, Minami-ku, Hiroshima 734-8551, Japan

Bach1 is a transcriptional repressor of heme oxygenase-1 gene (

Hmox-1

) and ββββ

-globin gene.Heme oxygenase (HO)-1 is an inducible cytoprotective enzyme that degrades pro-oxidant heme tocarbon monoxide (CO) and biliverdin/bilirubin, which are thought to mediate anti-inflammatoryand anti-oxidant actions of HO-1. In the present study, we investigated the role of Bach1 in tissueprotection against myocardial ischemia/reperfusion (I/R) injury

in vivo

using mice lacking theBach1 gene (

Bach1

−−−−

/

−−−−

) and wild-type (

Bach1

+/+

) mice. In

Bach1

−−−−

/

−−−−

mice, myocardial expression ofHO-1 protein was constitutively up-regulated by 3.4-fold compared to that in

Bach1

+/+

mice.While myocardial I/R induced HO-1 protein in ischemic myocytes in both strains of mice, the extentof induction was significantly greater in

Bach1

−−−−

/

−−−−

mice than in

Bach1

+/+

mice. Myocardial infarctionwas markedly reduced in size by 48.4% in

Bach1

−−−−

/

−−−−

mice. Pretreatment of

Bach1

−−−−

/

−−−−

mice withzinc-protoporphyrin, an inhibitor of HO activity, abolished the infarction-reducing effect of

Bach1

disruption, indicating that reduction in the infarct size was mediated, at least in part, byHO-1 activity. Thus, Bach1 plays a pivotal role in setting the levels of both constitutive andinducible expression of HO-1 in the myocardium. Bach1 inactivation during I/R appears to bea key mechanism controlling the activation level of cytoprotective program involving HO-1.

Introduction

Cells change gene expression in response to variousforms of damage. Fates of cells are often determineddepending on the balance between protective andsuicidal gene expression programs. Heme oxygenase(HO)-1 is a cytoprotective enzyme induced by diversecellular stresses, degrading heme to carbon monoxide,biliverdin and iron. These products are involved in theanti-oxidant and anti-inflammatory actions of HO-1in conditions of tissue injury (Poss & Tonegawa 1997).Thus, gene delivery of heme oxygenase-1 gene (

Hmox-1

)( Juan

et al

. 2001; Tulis

et al

. 2001; Yet

et al

. 2001; Melo

et al

. 2002; Vulapalli

et al

. 2002) is a possible therapeuticstrategy for ischemic disease and atherosclerosis. An

alternative approach may be to enhance HO-1 expressionby interfering with its transcriptional regulators. Recentprogress in elucidation of the mechanisms of

Hmox-1

regulation may provide the foundation for such anapproach.

The inducible enhancers of

Hmox-1

carry multipleMaf-recognition elements (MAREs). Heterodimers ofthe small Maf proteins and NF-E2-related factor 2 (Nrf2)activate

Hmox-1

through binding to MAREs (Itoh

et al

.1997; Alam

et al

. 1999; Ishii

et al

. 2000; Kataoka

et al

.2001). In contrast, heterodimers of small Maf and Bach1or Bach2 repress MARE-dependent transcription(Oyake

et al

. 1996). The ability of Nrf2 to activate

Hmox-1

expression is greatly reduced in the presence of Bach1(Sun

et al

. 2002, 2004). These observations suggest thatthe activity of Bach1 to repress

Hmox-1

is dominant overthe activity of

Hmox-1

activators such as Nrf2 undernormal conditions and that inactivation of Bach1 is a keymechanism of

Hmox-1

induction (Sun

et al

. 2002, 2004).It is known that

Hmox-1

is strongly induced by itssubstrate heme as well as by oxidative stress. Heme binds

Communicated by

: Masayuki M. Yamamoto*

Correspondence

: E-mail: ozono@hiroshima-u.ac.jp

a

Present address

: Department of Biochemistry, TohokuUniversity Graduate School of Medicine, 2-1 Seiryo, Aoba-ku,Sendai 980-8575, Japan

Y Yano

et al.

Genes to Cells (2006)

11

, 791–803

© 2006 The AuthorsJournal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

792

to Bach1 to inhibit its DNA binding activity and to induceits nuclear export (Ogawa

et al

. 2001; Sun

et al

. 2002,2004; Suzuki

et al

. 2004). Oxidative stressors such as cad-mium also induce nuclear export of Bach1 (Suzuki

et al

.2003). On the other hand, signals from reactive oxygenspecies or electrophilic insults are reported to inactivatethe ubiquitin-proteasome degradation pathway of Nrf2in the cytoplasm (Kobayashi & Yamamoto 2005), allowingthis transcriptional factor to translocate to the nuclei andactivate its target genes including

Hmox-1.

Taken together,both inactivation of Bach1 and activation of Nrf2 inresponse to heme and/or oxidative stress appear to bedecisive for deployment of the cytoprotective program,including activation of

Hmox-1

. By the same token,Bach1 may set a threshold for cell survival under stressfulconditions through inhibiting downstream target genesincluding

Hmox-1

. In the present study, we tested thishypothesis using a myocardial ischemic injury model

in vivo.

The data demonstrated a dramatic increase incardioprotection against ischemic stress in mice lackingBach1 (

Bach1

−

/

−

mice), providing new insights into thetranscriptional program of cytoprotection as well as thetherapy of ischemic disease.

Results

Hemodynamics before and after ischemic injury in the heart

Wild-type (

Bach1

+/+

) mice and

Bach1

−

/

−

mice showed nosignificant differences in body and heart weights, cardiachistology and cardiac geometry and function evaluatedby ultrasonography (Table 1). Likewise, they showed nodifference in basal systolic and diastolic blood pressuresor heart rate (Table 2). These observations suggest thatthe absence of Bach1 does not affect normal myocardialstructure and function. We induced myocardial ischemiain

Bach1

+/+

and

Bach1

−

/

−

mice by ligating the left anterior

descending artery (LAD) for 60 min followed by reper-fusion for 24 h (ischemia and reperfusion, I/R). Twenty-four hours after induction of the myocardial I/R injury,there was no significant difference in mortality between

Bach1

+/+

and

Bach1

−

/

−

mice (9.8% in

Bach1

+/+

and 7.7% in

Bach1

−

/

−

mice). I/R injury caused significant reductionin systolic and diastolic blood pressures in

Bach1

+/+

micebut not in

Bach1

−

/

−

mice (Table 2), suggesting that theimpact of ischemic myocardial damage on systemichemodynamics was smaller in the mutant mice. Heartrates were significantly increased in both strains of mice.

Effects of I/R on HO-1 protein in left ventricles (LVs)

We compared the effects of I/R on HO-1 protein levelsin the LVs of

Bach1

+/+

mice and

Bach1

−

/

−

mice 24 h after

Table 2 Hemodynamics before and after I/R

Bach1+/+ mice Bach1−/− mice

Sham(n = 15)

I/R(n = 15)

Sham(n = 15)

I/R(n = 15)

Blood pressure (mmHg)Systolic 110.2 ± 3.4 94.6 ± 2.9 112.0 ± 4.1 103.7 ± 7.3Diastolic 79.2 ± 2.7 69.3 ± 2.7 78.4 ± 2.8 75.2 ± 7.3

Heart rate (bpm) 426.6 ± 23.7 521.0 ± 20.8 414.6 ± 8.7 503.1 ± 15.8

Values are mean ± SEM.

Table 1 Basal measurements of cardiac function, body weightand cardiac weight

ParameterBach1+/+ mice (n = 7)

Bach1−/− mice (n = 7)

LVDd (mm) 2.80 ± 0.05 2.85 ± 0.06LVDs (mm) 1.18 ± 0.05 1.12 ± 0.06%FS (%) 57.60 ± 1.87 60.90 ± 1.84IVSTd (mm) 1.06 ± 0.05 1.08 ± 0.02LVPWTd (mm) 1.01 ± 0.05 1.07 ± 0.08Body weight (g) 22.20 ± 8.9 21.00 ± 0.6RVW (mg) 25.80 ± 0.3 24.60 ± 0.3LVW (mg) 130.00 ± 0.1 122.00 ± 0.1

LVDd, left ventricular endodiastolic dimension; LVDs, left ventricular endosystolic dimension; %FS, fractional shortening; IVSTd, endodiastolic interventricular septum thickness; LVPWTd, endodiastolic left ventricular posterior wall thickness; RVW, right ventricular weight; LVW, left ventricular weight.Values are means ± SEM.

Control of cytoprotection by Bach1

© 2006 The Authors Genes to Cells (2006) 11, 791–803Journal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

793

I/R by Western blotting (Fig. 1). LVs were dissected intonon-ischemic and ischemic regions. In Bach1+/+ mice, I/R had no effect on HO-1 level in the non-ischemicregion, whereas it caused an increase in HO-1 level inthe ischemic region (by two-fold vs. that in sham-operatedmice) (Fig. 1B). When Bach1+/+ mice were pretreated

with hemin, an established inducer of HO-1, I/R causedincreases in HO-1 levels not only in the ischemic region(by three-fold) but also in the non-ischemic region (by1.9-fold). In Bach1−/− mice, the basal HO-1 level (i.e. thatin sham-operated Bach1−/− mice) was 3.4-fold higher thanthat in Bach1+/+ mice (P < 0.01). I/R caused further

Figure 1 Effects of I/R on HO-1 proteinexpression and Bach1 expression in LVs.Twenty-four hours after I/R, LVs wereseparated into non-ischemic (n-isc) andischemic (isc) regions, and analyzed. (A)Western blot analysis of HO-1 protein. Theblotted membrane was first reacted withHO-1 antibodies and then re-probedwith actin as an internal standard. (B)Densitometric analysis of Western blots.(C) Real-time PCR analysis of Bach1expression in Bach1+/+ mice. Bach1 levelwas normalized by the level of hypoxanthinephosphoribosyl-transferase (HPRT). ZnPP:zinc-protoporphyrin. Values are means +/−SEM (n = 6 per group). *P < 0.05 vs.sham-operated Bach1+/+ mice group,†P < 0.05 vs. I/R (n-isc) groups of eachmouse group, ¶P < 0.05 vs. I/R (n-isc)group in the other two mouse groups,§P < 0.05 vs. I/R (isc) group in the othertwo mouse groups.

Y Yano et al.

Genes to Cells (2006) 11, 791–803 © 2006 The AuthorsJournal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

794

up-regulation of HO-1 in both the non-ischemic region(by 1.7-fold vs. the basal level) and ischemic region (by2.9-fold). Importantly, the absolute HO-1 protein levelsin the non-ischemic and ischemic regions of Bach1−/−

mice were significantly (P < 0.05) higher than those ofthe other two mouse groups, indicating that Bach1 playsan important role in setting the level of both constitutiveand inducible expression of HO-1. However, as shownin Fig. 1C, Bach1 expression level examined by real-timePCR in the heart of Bach1+/+ mice was not altered eitherby I/R or treatment of mice with hemin or zinc-protoporphyrin (ZnPP), an inhibitor of HO-1, suggestingthat the transcriptional control of Bach1 abundance is notlikely to be the major mechanism of HO-1 up-regulation.Increase in HO-1 level in the non-ischemic regionsof hemin-treated Bach1+/+ mice and Bach1−/− mice wasunexpected, but this finding may indicate that theinducibility of Hmox-1 was sensitized in these mice,leading to induction by non-ischemic stress such ashemodynamic overload resulting from partial myocar-dial loss.

HO activity was measured in LVs to determinewhether the changes in HO-1 protein levels accompaniedchanges in its activity (Fig. 2). In Bach1+/+ mice, I/R hadno effect on the HO activity in the non-ischemic region,but it caused an increase in HO activity in the ischemicregion (by 4.1-fold compared to the level in sham-operated animals). In hemin-treated Bach1+/+ mice, I/R

caused increases in HO activity in both the non-ischemic region (by 3.9-fold) and ischemic region (by6.4-fold). In Bach1−/− mice, the basal HO activity level(i.e. that in sham-operated Bach1−/− mice) was signifi-cantly higher (P < 0.01) than that in Bach1+/+ mice. I/Rcaused further increases in HO activity both in thenon-ischemic region (by 1.8-fold) and ischemic region(by 3.6-fold). These results indicate that difference inHO activity completely reflects the difference in HO-1protein level as described in the above section. Takentogether, Bach1 is a key repressor of Hmox-1 in the heartand its loss leads to higher expression of HO-1 beforeand after I/R.

Immunohistochemical localization of HO-1 protein in LVs

We carried out immunohistochemical analysis to identifycells expressing HO-1. HO-1 was only scarcely presentin the myocardium of sham-operated Bach1+/+ mice, butit was up-regulated mainly in cardiomyocytes in thepresumptive ischemic region following I/R (Fig. 3A).HO-1 staining was very weak in the vascular smoothmuscle layer (α-smooth muscle actin-positive) andendothelium (CD31-positive) before and after I/R inBach1+/+ mice (Fig. 3B). On the other hand, HO-1 wasabundant throughout the myocardium in sham-operatedBach1−/− mice and it was further up-regulated followingI/R (Fig. 3A). HO-1 was positive in the smooth musclelayer and endothelium of vasculature in Bach1−/− micebefore and after I/R (Fig. 3B). Interestingly, HO-1 wasclearly observed in microvessels in the infarcted area inboth Bach1+/+ and Bach1−/− mice (Fig. 3C). These resultssuggest that the regulation of HO-1 expression is differentin large vessels and microvessels. Bach1 may inhibit HO-1 expression in larger vessels but not in microvessels afterI/R injury in Bach1+/+ mice. Upon I/R, inflammatory cellsinfiltrate into the damaged area. HO-1 was abundant inCD68 positive- monocytes/macrophages in both Bach1+/+

and Bach1−/− mice (Fig. 3C). Taken together, the absenceof Bach1 caused marked up-regulation of HO-1 not onlyin cardiomyocytes but also in vascular smooth musclecells.

We have noted that the vascular smooth muscle layer,labeled by α-smooth muscle actin, was thinner in Bach1−/−

mice than in Bach1+/+ mice, which was confirmed by aquantitative morphometric analysis (1226 ± 19.3 vs.1840 ± 22.5 µm2 in Bach1−/− vs. Bach1+/+ mice, respectively;P < 0.05). This difference is thought to reflect a reducedability of Bach1-deficient vascular smooth muscle cellsto proliferate, which was found in our previous in vitroexperiments (Omura et al. 2005).

Figure 2 Effects of I/R on HO activity of LVs. After I/R, LVswere separated into non-ischemic (n-isc) and ischemic (isc)regions. Values are means +/− SEM (n = 4 per group). *P < 0.05vs. sham-operated Bach1+/+ mice group, †P < 0.05 vs. I/R (n-isc)group of each mouse group, ¶P < 0.05 vs. I/R (n-isc) group inthe other two mouse groups, §P < 0.05 vs. I/R (isc) group inthe other two mouse groups, #P < 0.05 vs. sham-operated, I/R(n-isc), and I/R (isc) groups of Bach1−/− mice.

Control of cytoprotection by Bach1

© 2006 The Authors Genes to Cells (2006) 11, 791–803Journal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

795

Figure 3 Immunohistochemical localization of HO-1 protein in the myocardium. (A) HO-1 protein expression in myocardium at lowpower magnification. Arrow indicates the presumptive boundary of ischemic and non-ischemic regions. Scale bar: 100 µm. (B) HO-1expression in vessel walls. Localization of endothelium and vascular smooth muscle cells are indicated by CD31 and α-SM actin in serialsections, respectively. Scale bar: 50 µm. (C) HO-1 expression in microvessels and monocytes/macrophages in the infarcted myocardialarea. HO-1 protein was present in CD31-possitive microvessels (red arrow) and infiltrated CD68-positive monocytes/macrophages(yellow arrowheads) of both strains of mice. Note that the number of infiltrated cells is smaller in Bach1−/− mice. DAB signal was enhancedby Nickel in CD31 and α-SM staining. *cardiomyocyte, scale bar: 20 µm.

Y Yano et al.

Genes to Cells (2006) 11, 791–803 © 2006 The AuthorsJournal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

796

Reduction of infarct size following I/R in Bach1−−−−/−−−− mice

Having established that the genetic ablation of Bach1leads to higher levels of HO-1 expression, we nextcompared the tissue damage following I/R injury. Themyocardial area at risk (AAR; i.e. perfusion area of theoccluded coronary artery) in the I/R and sham-operatedhearts was evaluated 24 h after surgery by Evans Bluestaining, and the size of myocardial infarction was deter-mined by triphenyl-tetrazolium chloride (TTC) staining(Fig. 4A), which delineates staining-resistant infarctedmyocardium out of red-stained viable myocardium. Asexpected, there was no significant difference in the AARamong the groups of mice (Fig. 4B). The infarct sizeswere compared after adjusting the respective AARs. Asshown in Fig. 4A,C, the size of myocardial infarction inBach1+/+ mice after I/R reached 46.9 ± 1.6% of the

AAR. Pretreatment of Bach1+/+ mice with hemin, aninducer of HO-1, reduced infarct size to 35.3 ± 1.4% ofthe AAR (Fig. 4A,C). In Bach1−/− mice, the infarct size wasmarkedly reduced to 24.3 ± 1.3% of the AAR (Fig. 4A,C).The infarct size in Bach1−/− mice was significantly smallerthan that in Bach1+/+ mice and hemin-treated Bach1+/+

mice (Fig. 4C).To investigate whether the observed effect of Bach1-

ablation was dependent on the catalytic activity of HO-1,we pretreated Bach1−/− mice with ZnPP, an inhibitor ofHO activity (Figs 2 and 4B). As expected, HO activitiesin LVs of ZnPP-treated Bach1−/− mice were diminishedto the level in sham-operated Bach1+/+ mice (Fig. 2).Accordingly, pretreatment of Bach1−/− mice with ZnPPresulted in infarct size comparable to that in Bach1+/+

mice (Fig. 4C), suggesting that the infarction-reducingeffect of Bach1-ablation was mediated, at least in part, bythe increase in HO activity.

Figure 4 Myocardial infarction assessedby Evans Blue and triphenyl-tetrazoliumchloride (TTC) staining. (A) RepresentativeEvans Blue and TTC staining of LV slices24 h after I/R. (B) Calculated AARs infour mouse groups. (C) Infarct sizes of fourmouse groups expressed as percentage ofinfarct area relative to the AAR. Values aremeans +/− SEM. (Bach1+/+ mice, n = 11;Bach1+/+ mice treated with hemin, n = 8;Bach1−/− mice, n = 16; and Bach1−/− micetreated with ZnPP, n = 9). *P < 0.05 vs.Bach1+/+ mice, †P < 0.05 vs. hemin-treatedBach1+/+ mice, ¶P < 0.05 vs. Bach1−/− mice.

Control of cytoprotection by Bach1

© 2006 The Authors Genes to Cells (2006) 11, 791–803Journal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

797

Inhibitions of apoptosis in Bach1−−−−/−−−− mice

To determine whether apoptosis is involved in themechanism of the cardioprotection, we examined cleav-age of procaspase-3 to caspase-3 in myocardial tissues.Caspase-3 was activated following I/R in Bach1+/+ miceas judged by the appearance of cleaved form, whereas thecleaved form of caspase-3 was not detected in Bach1−/−

mice (Fig. 5A). Consistent with this finding, terminaldeoxynucleotidyl transferase-mediated dUTP nick-endlabeling (TUNEL) assays revealed fewer apoptotic cellsin the LV of Bach1−/− mice than in LV of Bach1+/+ mice(Fig. 5B,C). Here, we focused on the ischemic regionsof LVs in the sections (see Experimental procedures).Judging from the histology, most of the TUNEL-positivecells appeared to be cardiomyocytes. These results indicatethat apoptosis of myocardial cells following I/R was

inhibited in Bach1−/− mice. These results suggest that theabsence of Bach1 caused overall suppression of apoptoticloss of myocardial cells after I/R, which may, at least inpart, explain the cytoprotection against I/R in Bach1−/−

mice.To gain an insight into the signaling pathways of

cardioprotection in Bach1−/− mice, we measured theactivation levels of key molecules implicated in cardio-protection/apoptosis after I/R. Namely, we examinedpresumptive survival signals including Akt and signaltransducer and activator of transcription 3 (STAT3). Wealso examined c-jun-NH2-terminal kinase ( JNK) andp38 mitogen-activated protein kinase (MAPK). Akt wassimilarly activated after I/R in both Bach1+/+ and Bach1−/−

mice (data not shown). Phosphorylation of JNK was notdetected after I/R in either strain of mice (data notshown). The phosphorylation level of p38 MAPK was

Figure 5 Induction of apoptosis inresponse to I/R injury. (A) Activation ofcaspase-3 in LVs was analyzed afterdissecting into non-ischemic (n-isc) andischemic (isc) regions as described inExperimental procedures. The blottedmembrane was re-probed with α-tubulinas a loading control. (B) TUNEL-positivecells in the ischemic myocardium followingI/R. Scale bar: 20 µm (C) Bar graphshowing the averaged percentage ofTUNEL-positive cells in the ischemicregions of LVs from Bach1−/− mice andBach1+/+ mice. Values are means +/− SEM(n = 5 per group). *P < 0.05 vs. sham-operated Bach1+/+ mice, †P < 0.05 vs.Bach1+/+ mice.

Y Yano et al.

Genes to Cells (2006) 11, 791–803 © 2006 The AuthorsJournal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

798

transiently increased after I/R injury in Bach1+/+ mice.In contrast, this was not apparent in Bach1−/− mice(Fig. 6A,B). Phosphorylation of STAT3 was induced inthe ischemic region of LV at 2 h and 24 h after I/Rinjury in Bach1+/+ mice (Fig. 6A,C). It was alreadydetectable in the sham-operated Bach1−/− mice and furtherincreased both in the non-ischemic and ischemic regionsat 2 h after I/R injury (Fig. 6A,C). However, whenBach1−/− mice were treated with ZnPP, I/R induced p38MAPK activation was restored and I/R induced STAT3activation was attenuated, although the increased basalactivity of STAT3 still remained (Fig. 6D). These resultssuggest that HO-1 modulates the signaling pathwaysto suppress I/R-induced p38 MAPK activation and toenhance I/R-induced STAT3 activation in Bach1−/−

mice.

DiscussionTranscription is often regulated by a combination ofactivators and repressors. Such a combination is expectedto generate an integrator of multiple signals in geneexpression. One of the major challenges in the field ofgene expression is to understand the function of suchintegrators in the maintenance of homeostasis in highereukaryotes. By using the myocardial I/R injury as amodel system for the function of Bach1 in vivo, werevealed two aspects of Bach1 in the context of Hmox-1regulation. First, as shown in Figs 1 and 3, the geneticablation of Bach1 led to constitutive expression of Hmox-1in virtually every myocardial cell even under normalconditions. This observation suggests that myocardialcells are under the influence of diverse stimuli that canpotentially result in Hmox-1 expression and that Bach1plays a critical role in vivo to counteract such inputs,maintaining the quiescence state of Hmox-1. Thus, itappears that the regulation of the activator function suchas Keap1-Nrf2 pathway (Kobayashi & Yamamoto 2005)is not sufficient to fine-tune the expression of Hmox-1in vivo. Second, the Hmox-1 expression in the Bach1-deficient mice was further up-regulated after I/R, whichmounted to the level as high as 4-5-fold of that in wild-typemice subjected to the same I/R injury (Fig. 1B). Thisobservation indicates that the Bach1 repressor functionremains to inhibit Hmox-1 to some extent even after I/R.It appears that Bach1 is not completely inactivated evenwhen cells are under extreme stresses such as I/R. Thus,Bach1 not only keeps Hmox-1 off under normal condi-tions but also determines the induced levels. Theseobservations have important clinical implications in lightof the cytoprotective role of HO-1 and are discussedfurther below.

While several studies have established Hmox-1 as atarget gene of Bach1 (Sun et al. 2002, 2004), there maybe additional downstream target genes. Because identityof other downstream genes and their functions are notknown at present, it has been unclear how cells andtissues respond to various stresses in the absence ofBach1 in vivo. If some of the downstream genes are pro-apoptotic, for example, their expressions are expected toincrease in the Bach1-deficient cells, resulting in increasedapoptosis in response to stress. In the present study, wedemonstrated that the genetic ablation of Bach1 causeda marked reduction in myocardial infarction after I/R(Fig. 4). Thus, the net effect of Bach1-ablation in themyocardial cells and tissues was to increase resistanceagainst stresses derived from I/R injury. Because theinfarction-reducing effect of Bach1-ablation was abolishedby pretreating mice with ZnPP, it was mediated, at leastin part, by the increased levels of HO-1 expression andHO activity in Bach1−/− mice. We have previouslyreported that Bach1-deficient mice showed reducedneointimal formation in the cuff injury model forarteriosclerosis, in which smooth muscle cells from Bach1-deficient mice expressed higher levels of HO-1 comparedwith wild-type mice (Omura et al. 2005). Thus, resultsof our previous and present studies suggest Bach1 as adetrimental factor in atherosclerosis and myocardial I/Rinjury. Although oxidative stress associated with suchdisease conditions is expected to inhibit Bach1, residuallevels of Bach1 function may suffice to preclude fulldeployment of downstream target genes such as HO-1.

Interestingly, the cardioprotection in the absence ofBach1 was associated with inhibition of the p38 MAPKpathway, activation of the STAT3 pathway, and inhibi-tion of apoptosis (Fig. 5). It has been suggested thatactivation of p38 MAPK is a pro-apoptotic signal in theheart under ischemic conditions as demonstrated incultured cardiomyocytes subjected to ischemic stress(Mackay & Mochly Rosen 1999), in I/R model ofisolated heart (Ma et al. 1999), and in in vivo I/R modelof mice (Kaiser et al. 2004). Thus, attenuation of p38

Figure 6 Effect of I/R on phosphorylations of p38 MAPK andSTAT3. (A) Western analysis with indicated antibodies wascarried out 2 h and 24 h after I/R injury. LVs were dissected intoischemic (isc) and non-ischemic (n-isc) regions as described inExperimental procedures. (B) Densitometric analysis of I/R-induced activations of p38 MAPK. (C) Densitometric analysis ofI/R-induced activations of STAT3. (D) Effects of zinc-protoporphyrin (ZnPP) on phosphorylations of p38 MAPK andSTAT3 in Bach1−/− mice. Values in (B) and (C) are means +/−SEM (n = 6 per group). *P < 0.05 vs. sham-operated Bach1+/+

mice group.

Control of cytoprotection by Bach1

© 2006 The Authors Genes to Cells (2006) 11, 791–803Journal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

799

Y Yano et al.

Genes to Cells (2006) 11, 791–803 © 2006 The AuthorsJournal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

800

MAPK may have contributed to the reduced levels ofapoptosis in the Bach1-deficient mice. On the otherhand, the activation of STAT3 pathway promotes cyto-protection and inhibits apoptosis in ischemic conditions(Negoro et al. 2000; Xuan et al. 2001; Stephanou 2004).Accordingly, conditional knockout mice harboring acardiomyocyte-restricted deletion of STAT3 display alarger infarct size and increased apoptosis after I/Rcompared with the control mice ( Jacoby et al. 2003;Hilfiker-Kleiner et al. 2004). Taking our observationstogether with these reports, we envisage that the cardio-protection in Bach1−/− mice was mediated, at least inpart, by activation of STAT3. In contrast, because thegenetic ablation of Bach1 did not significantly affect thephosphorylation levels of Akt during I/R injury, thissurvival pathway appears less affected by the presence orabsence of Bach1.

The molecular mechanisms that led to reducedphosphorylation of p38 MAPK and increased activationof STAT3 in the Bach1−/− hearts are not clear at present.Administration of ZnPP in Bach1−/− mice reversed theseeffects of Bach1-ablation on STAT3 activation andp38 MAPK inactivation (Fig. 6D), suggesting that thesechanges were secondary to the increased activity ofHO-1. Accordingly, phosphorylation of p38 MAPK wasinhibited by CO-releasing molecules in endothelin-stimulated cardiomyocytes in culture (Tongers et al.2004), suggesting that CO, one of the reaction productsof HO-1 reaction, regulates p38 MAPK activation in theheart. Alternatively, biliverdin and/or its reduced productbilirubin may be responsible for the reduction in p38MAPK activation because they are potent anti-oxidant(Stocker et al. 1987). Since the STAT3 is activated by theinterleukin (IL)-6 family of cytokines including IL-6(Kukielka et al. 1995), cardiotrophin-1 (Pennica et al.1995; Sheng et al. 1997), and leukemia inhibitory factor(Kunisada et al. 1996) and by granulocyte-colony-stimulating factor (G-CSF) (Harada et al. 2005), theactivation of STAT3 we observed in the Bach1−/− micemay reflect modulation of such cytokine network byHO-1. Much remains to be understood about this inter-esting coordination of p38 MAPK and STAT3 pathwaysvia HO-1 in cardiomyocytes.

Recently, HO-1 gene delivery has been shown toresult in amelioration of not only ischemic heart disease(Yet et al. 2001; Melo et al. 2002; Vulapalli et al. 2002)but also atherosclerosis ( Juan et al. 2001; Tulis et al.2001), renal I/R injury (Blydt-Hansen et al. 2003),hypoxia-induced lung injury (Otterbein et al. 1999) andliver I/R injury (Amersi et al. 1999). The results of thisstudy suggest a novel way to maximize the cytoprotectiverole of HO-1 in disease settings. Importantly, the

induced levels of HO-1 were found much higher inhearts of Bach1−/− mice. Furthermore, stress-regulatedinduction of HO-1 was somehow retained in theabsence of Bach1. Thus, because inhibition of Bach1 isexpected to result in higher levels of HO-1 expression,Bach1 may be an excellent therapeutic target against I/R injury. Further studies of Bach1 may lead to a detailedunderstanding of the gene expression programs ofoxidative stress response.

Experimental proceduresAnimals

We used pairs of adult male Bach1−/− and Bach1+/+ littermates agedof 8–10 weeks in this study (Sun et al. 2002). Heterozygous micewere intercrossed to obtain homozygous Bach1-deficient mice aswell as control wild-type mice. Congenic Bach1-deficient micewere obtained by repeatedly backcrossing with C57BL/6 J miceat least up to 12 generations. All experimental procedures wereapproved and carried out in accordance with the Guidelines ofHiroshima University Graduate School of Biomedical Sciences.

Surgical procedure

We induced myocardial ischemia by ligating the LAD as previ-ously described (Oishi et al. 2003). The LAD was ligated at1.0 mm distal from the tip of the left appendix for a period of60 min, and this was followed by releasing the ligation and closingthe chest. Hearts were excised at 2 and 24 h after surgery orsham-operation and subjected to the following analyses.

Assessment of cardiac geometry and function by ultrasonography

Cardiac geometry and function were evaluated using an echo-cardiographic system (Toshiba SSA 550A) equipped with a 14-MHzlinear transducer as previously described (Oishi et al. 2003). LVend diastolic- and end systolic-dimensions (LVDd and LVDs) weremeasured at the distal level of the papillary muscle using short-axisM-mode images. Three beats were averaged for each measurement.Percent fractional shortening (%FS) was calculated as:

[(LVDd − LVDs)/LVDd] × 100.

Quantification of myocardial infarct after I/R

After excision, hearts were perfused with 1.0% TTC (Sigma-Aldrich) in phosphate buffer (pH 7.4) to stain viable myocardium.The LAD was then re-occluded, and each heart was perfused with10% Evans Blue (Sigma-Aldrich) to delineate the AAR, i.e.perfusion area of the LAD. The hearts were frozen, and each LVwas cut into five transverse slices. The images of the slices weredigitally analyzed for infarct area, ischemic area and total LV area

Control of cytoprotection by Bach1

© 2006 The Authors Genes to Cells (2006) 11, 791–803Journal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

801

using NIH image software. The ratio of AAR/total LV area andthe ratio of infarct area/AAR were calculated and expressed aspercentages as previously described (Maekawa et al. 2002).

Administration of an inducer or inhibitor of HO-1

We injected hemin (30 mg/kg, Sigma-Aldrich) intraperitoneally24 h prior to the coronary ligation in order to induce HO-1. Forthe purpose of blockade of HO activity, we injected ZnPP(10 mg/kg, Sigma-Aldrich) 24 h prior to the coronary ligation aspreviously described (Hangaishi et al. 2000). ZnPP was dissolvedin phosphate buffer with 0.1 m NaOH and neutralized with 0.1 mHCl immediately before administration and pH was maintained at7.4.

Western blot analysis

For detection of HO-1 protein, LVs were excised 24 h after I/R andhomogenized in a buffer containing 50 mm Tris (pH 7.5), 150 mmNaCl, 1 mm PMSF, 0.25 m sucrose, 0.5% deoxycholate, 2 µg/mLleupeptin and 2 µg/mL aprotinin. Before homogenization, LVswere dissected into non-ischemic and ischemic myocardialregions. The perfusion area of the LAD, which was almostconstant in mice (Fig. 4B), was considered to be the ischemicregion. For analyses of phosphorylation signals and procaspase-3cleavage, LVs were similarly dissected into non-ischemic andischemic regions and homogenized in a buffer containing 25 mmTris-HCl, 150 mm NaCl, 5 mm EDTA, 25 mm NaF, 25 mm Na-pyrophosphate, 1% Triton X-100, 0.1% SDS, 0.5% deoxycholate,10% glycerol, 1 mm vanadate and 100 mm Benzamisin. Protein(100 µg) of each heart homogenate was incubated with anti-HO-1polyclonal antisera (a gift from Dr S. Taketani), as previouslydescribed (Oishi et al. 2003). Antibodies for protein kinase Akt,phospho-Akt (Ser473), p38 MAPK, phospho-p38 (Thr180/Tyr182),STAT3 and phospho-STAT3 (Tyr705) were purchased from CellSignaling Technology. Antibody for JNK was from Santa CruzBiotechnology and that for phospho-JNK (pTPpY) was fromPromega. Anti-caspase-3 was purchased from Sigma. Thisantibody also recognized procaspase-3. To investigate the amountof phosphorylation levels, the hearts were analyzed at 2 and 24 hafter reperfusion. The blotted membrane was re-probed with actin(DAKO, clone 1A4) or α-tubulin (Santa Cruz Biotechnology) asan internal standard.

Real-time PCR analysis for Bach1 expression

Total RNA was extracted from the cardiac tissues using Trizol,and cDNA was prepared as previously described (Sun et al.2002). Real-time PCR was carried out using the LightCyclersystem (Roche), as previously described (Sun et al. 2002;Omura et al. 2005). Primers to amplify Bach1 cDNA were5′-TGATGTGACTGTCCTGGTGG-3′ and 5′-AAATCCTT-TAACCGTTACCTCTTC-3′. Primers to amplify cDNA ofHO-1 were 5′-GGGTGACAGAAGAGGCTAAG-3′ and 5′-GTGTCTGGGATGAGCTAGTG-3′. Primers to amplifycDNA of hypoxanthine phosphoribosyl-transferase (HPRT), a

house keeping gene, were 5′-CTCGAAGTGTTGGATACAGG-3′and 5′-AACTTGCGCTCATCTTAGG-3′.

Determination of HO activity

HO activity in microsomes was determined in LV tissues aspreviously described (Imai et al. 2001). The activity was expressedas nmol bilirubin formed per hour per mg of protein. The proteinconcentration was determined by a dye binding assay (Bio-Rad).

Immunohistochemistry

Immunohistochemical analyses of HO-1, endothelium, smoothmuscle cells and monocytes were performed in frozen and paraffinsections. Sections were incubated overnight at 4 °C with a 1 : 500dilution of a polyclonal rabbit anti-HO-1 antibody (Stressgen),a 1 : 200 dilution of a polyclonal rat anti-CD31 antibody(BD Pharmingen) for the endothelium, a 1 : 200 dilution of amonoclonal anti-α-smooth muscle actin (α-SM) antibody(DAKO, clone 1A4) for vascular smooth muscle cells and a 1 : 200dilution of a monoclonal anti-CD68 antibody (Serotec) formonocytes. Signals for HO-1, CD31 and α-SM were visualizedby the avidin-biotin immunoperoxidase method (ABC Elite kit,Vector Laboratories) using diaminobentizin as a substrate. Signalsfor CD68 were visualized using alkaline phosphatase as a substrate(Vector Laboratories).

The volume of vascular smooth muscle layer was quantified bymorphometry in the LV sections stained with α-smooth muscleactin. The digital images were analyzed using Scion Imagesoftware program, in which the areas of outer and inner profilesof intracardiac arteries were calculated and the difference wasregarded as the vascular area. We examined slides from 5 Bach1−/−

mice and 5 Bach1+/+ mice (sham-operated groups).

In situ detection of apoptosis

Apoptotic myocardial cells were detected in paraffin sectionsby the TUNEL technique with ApopTag Plus (Invitrogen)according to the manufacturer’s instructions. Since most of theTUNEL positive cells are observed in the ischemic region of LV,we evaluated only the ischemic myocardium in the sections. Weestimated the perfusion area of LAD, i.e. the ischemic LV region,by macro anatomy in reference to right ventricle, and the borderwas marked on the slides with a marker. The labels of slides wereblinded and sections of sham-operated animals were similarlyexamined. We sampled ten microscopic fields (at ×1000 magnifi-cation) within the ischemic LV region avoiding overlap. Thispolicy invariably resulted in examining most of the ischemicregion of LV. The numbers of TUNEL-positive and negative cellswere counted in each field and the percentage of TUNEL-positivecells was calculated, then the data from the ten fields were aver-aged. Each field contained approximately 50–70 cells includingcardiomyocytes and non-myocytes. Cardiomyocytes were distin-guished from non-myocytes by microscopic appearance; thatis, well-shaped, elongated, and striated cells. We examined slidesfrom 5 Bach1−/− mice and 5 Bach1+/+ mice (I/R groups).

Y Yano et al.

Genes to Cells (2006) 11, 791–803 © 2006 The AuthorsJournal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

802

Statistical analysis

All results are expressed as mean ± SEM. Comparison betweentwo groups was made by Student’s t-test, and comparison amongfour or five groups was made by analysis of variance followed byScheffe’s posthoc analysis. Statistical significance was accepted at avalue of P < 0.05.

AcknowledgementsWe thank Dr. T. Morita for advice on the enzymatic assay ofHO and reagents, Dr. N. Maekawa for advice on the surgicaltechnique and TTC staining technique, and Dr. S. Taketanifor providing HO-1 antiserum. This study was supported byGrants-in-Aid for Scientific Research to T.O., R.O. and K.I.from the Ministry of Education, Culture, Sports, Science andTechnology of Japan, Grants from the Charitable Trust ClinicalPathology Research Foundation of Japan, and Grant fromKurozumi Medical Foundation of Japan.

ReferencesAlam, J., Stewart, D., Touchard, C., Boinapally, S., Choi, A.M. &

Cook, J.L. (1999) Nrf2, a Cap’n’Collar transcription factor,regulates induction of the heme oxygenase-1 gene. J. Biol.Chem. 274, 26071–26078.

Amersi, F., Buelow, R., Kato, H., et al. (1999) Upregulation ofheme oxygenase-1 protects genetically fat Zucker rat liversfrom ischemia/reperfusion injury. J. Clin. Invest. 104, 1631–1639.

Blydt-Hansen, T.D., Katori, M., Lassman, C., et al. (2003) Genetransfer-induced local heme oxygenase-1 overexpressionprotects rat kidney transplants from ischemia/reperfusioninjury. J. Am. Soc. Nephrol. 14, 745–754.

Hangaishi, M., Ishizaka, N., Aizawa, T., et al. (2000) Induction ofheme oxygenase-1 can act protectively against cardiacischemia/reperfusion in vivo. Biochem. Biophys. Res. Commun.279, 582–588.

Harada, M., Qin, Y., Takano, H., et al. (2005) G-CSF preventscardiac remodeling after myocardial infarction by activatingthe Jak-Stat pathway in cardiomyocytes. Nat. Med. 11, 305–311.

Hilfiker-Kleiner, D., Hilfiker, A., Fuchs, M., et al. (2004) Signaltransducer and activator of transcription 3 is required formyocardial capillary growth, control of interstitial matrixdeposition, and heart protection from ischemic injury. Circ.Res. 95, 187–195.

Imai, T., Morita, T., Shindo, T., et al. (2001) Vascular smoothmuscle cell-directed overexpression of heme oxygenase-1elevates blood pressure through attenuation of nitricoxide-induced vasodilation in mice. Circ. Res. 89, 55–62.

Ishii, T., Itoh, K., Takahashi, S., et al. (2000) Transcription factorNrf2 coordinately regulates a group of oxidative stress-induciblegenes in macrophages. J. Biol. Chem. 275, 16023–16029.

Itoh, K., Chiba, T., Takahashi, S., et al. (1997) An Nrf2/small Mafheterodimer mediates the induction of phase II detoxifying

enzyme genes through antioxidant response elements. Biochem.Biophys. Res. Commun. 236, 313–322.

Jacoby, J.J., Kalinowski, A., Liu, M.G., et al. (2003) Cardiomyocyte-restricted knockout of STAT3 results in higher sensitivity toinflammation, cardiac fibrosis, and heart failure with advancedage. Proc. Natl. Acad. Sci. USA 100, 12929–12934.

Juan, S.H., Lee, T.S., Tseng, K.W., et al. (2001) Adenovirus-mediated heme oxygenase-1 gene transfer inhibits the develop-ment of atherosclerosis in apolipoprotein E-deficient mice.Circulation 104, 1519–1525.

Kaiser, R.A., Bueno, O.F., Lips, D.J., et al. (2004) Targetedinhibition of p38 mitogen-activated protein kinase antagonizescardiac injury and cell death following ischemia-reperfusion invivo. J. Biol. Chem. 279, 15524–15530.

Kataoka, K., Handa, H. & Nishizawa, M. (2001) Induction ofcellular antioxidative stress genes through heterodimerictranscription factor Nrf2/small Maf by antirheumatic gold(I)compounds. J. Biol. Chem. 276, 34074–34081.

Kobayashi, M. & Yamamoto, M. (2005) Molecular mechanismsactivating the Nrf2-Keap1 pathway of antioxidant generegulation. Antioxid. Redox Signal. 7, 385–394.

Kukielka, G.L., Smith, C.W., Manning, A.M., Youker, K.A.,Michael, L.H. & Entman, M.L. (1995) Induction ofinterleukin-6 synthesis in the myocardium. Potential rolein postreperfusion inflammatory injury. Circulation 92, 1866–1875.

Kunisada, K., Hirota, H., Fujio, Y., et al. (1996) Activation ofJAK-STAT and MAP kinases by leukemia inhibitory factorthrough gp130 in cardiac myocytes. Circulation 94, 2626–2632.

Ma, X.L., Kumar, S., Gao, F., et al. (1999) Inhibition of p38mitogen-activated protein kinase decreases cardiomyocyteapoptosis and improves cardiac function after myocardialischemia and reperfusion. Circulation 99, 1685–1691.

Mackay, K. & Mochly-Rosen, D. (1999) An inhibitor of p38mitogen-activated protein kinase protects neonatal cardiacmyocytes from ischemia. J. Biol. Chem. 274, 6272–6279.

Maekawa, N., Wada, H., Kanda, T., et al. (2002) Improvedmyocardial ischemia/reperfusion injury in mice lacking tumornecrosis factor-α. J. Am. Coll. Cardiol. 39, 1229–1235.

Melo, L.G., Agrawal, R., Zhang, L., et al. (2002) Gene therapystrategy for long-term myocardial protection using adeno-associated virus-mediated delivery of heme oxygenase gene.Circulation 105, 602–607.

Negoro, S., Kunisada, K., Tone, E., et al. (2000) Activation ofJAK/STAT pathway transduces cytoprotective signal in ratacute myocardial infarction. Cardiovasc. Res. 47, 797–805.

Ogawa, K., Sun, J., Taketani, S., et al. (2001) Heme mediatesderepression of Maf recognition element through directbinding to transcription repressor Bach1. EMBO J. 20, 2835–2843.

Oishi, Y., Ozono, R., Yano, Y., et al. (2003) Cardioprotective roleof AT2 receptor in postinfarction left ventricular remodeling.Hypertension 41, 814–818.

Omura, S., Suzuki, H., Toyofuku, M., Ozono, R., Kohno, N. &Igarashi, K. (2005) Effects of genetic ablation of bach1 uponsmooth muscle cell proliferation and atherosclerosis after cuffinjury. Genes Cells 10, 277–285.

Control of cytoprotection by Bach1

© 2006 The Authors Genes to Cells (2006) 11, 791–803Journal compilation © 2006 by the Molecular Biology Society of Japan/Blackwell Publishing Ltd.

803

Otterbein, L.E., Kolls, J.K., Mantell, L.L., Cook, J.L., Alam, J. &Choi, A.M. (1999) Exogenous administration of hemeoxygenase-1 by gene transfer provides protection againsthyperoxia-induced lung injury. J. Clin. Invest. 103, 1047–1054.

Oyake, T., Itoh, K., Motohashi, H., et al. (1996) Bach proteinsbelong to a novel family of BTB-basic leucine zipper transcriptionfactors that interact with MafK and regulate transcriptionthrough the NF-E2 site. Mol. Cell. Biol. 16, 6083–6095.

Pennica, D., King, K.L., Shaw, K.J., et al. (1995) Expressioncloning of cardiotrophin 1, a cytokine that induces cardiacmyocyte hypertrophy. Proc. Natl. Acad. Sci. USA 92, 1142–1146.

Poss, K.D. & Tonegawa, S. (1997) Reduced stress defense in hemeoxygenase 1-deficient cells. Proc. Natl. Acad. Sci. USA 94,10925–10930.

Sheng, Z., Knowlton, K., Chen, J., Hoshijima, M., Brown, J.H.& Chien, K.R. (1997) Cardiotrophin 1 (CT-1) inhibition ofcardiac myocyte apoptosis via a mitogen-activated proteinkinase-dependent pathway. Divergence from downstreamCT-1 signals for myocardial cell hypertrophy. J. Biol. Chem.272, 5783–5791.

Stephanou, A. (2004) Role of STAT-1 and STAT-3 in ischaemia/reperfusion injury. J. Cell. Mol. Med. 8, 519–525.

Stocker, R., Yamamoto, Y., McDonagh, A.F., Glazer, A.N. &Ames, B.N. (1987) Bilirubin is an antioxidant of possiblephysiological importance. Science 235, 1043–1046.

Sun, J., Brand, M., Zenke, Y., Tashiro, S., Groudine, M. &Igarashi, K. (2004) Heme regulates the dynamic exchange ofBach1 and NF-E2-related factors in the Maf transcriptionfactor network. Proc. Natl. Acad. Sci. USA 101, 1461–1466.

Sun, J., Hoshino, H., Takaku, K., et al. (2002) HemoproteinBach1 regulates enhancer availability of heme oxygenase-1gene. EMBO J. 21, 5216–5224.

Suzuki, H., Tashiro, S., Hira, S., et al. (2004) Heme regulates geneexpression by triggering Crm1-dependent nuclear export ofBach1. EMBO J. 23, 2544–2553.

Suzuki, H., Tashiro, S., Sun, J., Doi, H., Satomi, S. & Igarashi, K.(2003) Cadmium induces nuclear export of Bach1, a transcrip-tional repressor of heme oxygenase-1 gene. J. Biol. Chem. 278,49246–49253.

Tongers, J., Fiedler, B., Konig, D., et al. (2004) Heme oxygenase-1 inhibition of MAP kinases, calcineurin/NFAT signaling, andhypertrophy in cardiac myocytes. Cardiovasc. Res. 63, 545–552.

Tulis, D.A., Durante, W., Liu, X., Evans, A.J., Peyton, K.J. &Schafer, A.I. (2001) Adenovirus-mediated heme oxygenase-1gene delivery inhibits injury-induced vascular neointimaformation. Circulation 104, 2710–2715.

Vulapalli, S.R., Chen, Z., Chua, B.H., Wang, T. & Liang, C.S.(2002) Cardioselective overexpression of HO-1 prevents I/R.-induced cardiac dysfunction and apoptosis. Am. J. Physiol. HeartCirc. Physiol. 283, H688–H694.

Xuan, Y.T., Guo, Y., Han, H., Zhu, Y. & Bolli, R. (2001) Anessential role of the JAK-STAT pathway in ischemic precondi-tioning. Proc. Natl. Acad. Sci. USA 98, 9050–9055.

Yet, S.F., Tian, R., Layne, M.D., et al. (2001) Cardiac-specificexpression of heme oxygenase-1 protects against ischemia andreperfusion injury in transgenic mice. Circ. Res. 89, 168–173.

Received: 22 December 2005 Accepted: 10 April 2006