Cell, Vol. 123, 347–358, October 21, 2005, Copyright ©2005 by Elsevier Inc. DOI 10.1016/j.cell.2005.08.004

The Homeodomain Transcription FactorIrx5 Establishes the Mouse CardiacVentricular Repolarization Gradient

Danny L. Costantini,1,2,4,5 Eric P. Arruda,1,2,4,6

Pooja Agarwal,1,2,4,6,15 Kyoung-Han Kim,4,5

Yonghong Zhu,1,4 Wei Zhu,1 Melanie Lebel,2,6

Chi Wa Cheng,2,10 Chong Y. Park,11,16

Stephanie A. Pierce,11 Alejandra Guerchicoff,12

Guido D. Pollevick,12 Toby Y. Chan,13

M. Golam Kabir,4,13 Shuk Han Cheng,10

Mansoor Husain,4,5,7,9,13 Charles Antzelevitch,12

Deepak Srivastava,11,16 Gil J. Gross,1,3,4,8

Chi-chung Hui,2,6 Peter H. Backx,4,5,7,14,*and Benoit G. Bruneau1,2,4,6,*1Program in Cardiovascular Research2Program in Developmental Biology3Cardiology DivisionThe Hospital for Sick ChildrenToronto, Ontario M5G 1X8Canada4The Heart and Stroke/Richard Lewar Centre

of Excellence5Department of Physiology6Department of Molecular and Medical Genetics7Department of Medicine8Department of Pediatrics9McLaughlin Centre for Molecular MedicineUniversity of TorontoToronto, Ontario M5S 1A8Canada10Department of Biology and ChemistryCity University of Hong KongHong KongChina11Departments of Pediatrics and Molecular BiologyUniversity of Texas Southwestern Medical Center

at DallasDallas, Texas 7539012Masonic Medical Research LaboratoryUtica, New York 1350113Division of Cellular and Molecular BiologyThe Toronto General Hospital Research InstituteToronto, Ontario M5G 2C4Canada14Division of CardiologyUniversity Health NetworkToronto, Ontario M5G 2C4Canada

Summary

Rhythmic cardiac contractions depend on the orga-nized propagation of depolarizing and repolarizingwavefronts. Repolarization is spatially heterogeneous

*Correspondence: p.backx@utoronto.ca (P.H.B.); bbruneau@sickkids.ca (B.G.B.)

15 Present address: Cardiovascular Research Institute, University ofCalifornia, San Francisco, San Francisco, California 94143. 16 Present address: Gladstone Institute of Cardiovascular Disease,San Francisco, California 94158.

and depends largely on gradients of potassium cur-rents. Gradient disruption in heart disease may un-derlie susceptibility to fatal arrhythmias, but it is notknown how this gradient is established. We showthat, in mice lacking the homeodomain transcriptionfactor Irx5, the cardiac repolarization gradient is abol-ished due to increased Kv4.2 potassium-channel ex-pression in endocardial myocardium, resulting in aselective increase of the major cardiac repolarizationcurrent, Ito,f, and increased susceptibility to arrhyth-mias. Myocardial Irx5 is expressed in a gradient oppo-site that of Kv4.2, and Irx5 represses Kv4.2 expressionby recruiting mBop, a cardiac transcriptional repres-sor. Thus, an Irx5 repressor gradient negatively regu-lates potassium-channel-gene expression in the heart,forming an inverse Ito,f gradient that ensures coordi-nated cardiac repolarization while also preventing ar-rhythmias.

Introduction

Patterning of cardiac-gene expression underlies normalheart development. For example, longitudinal pat-terning along the anteroposterior axis of the heart es-tablishes the distinctions between atrial and ventricularchambers, while concentric patterning within chambersestablishes transmural cardiac growth and gene-expres-sion gradients (Bruneau, 2002; Habets et al., 2003). Animportant gene-expression gradient is that which de-fines the cardiac repolarization gradient (Antzelevitch,2004; Nerbonne and Guo, 2002; Oudit et al., 2001). Fol-lowing depolarization and ventricular contraction, repo-larization initiates cardiac relaxation. In all mammals,ventricular repolarization proceeds in a synchronizedwave advancing from the base of the heart to its apexand from epicardial to endocardial myocardium, whichis believed to ensure efficient pump function and main-tain an arrhythmia-free heart. However, neither how therepolarization gradient is established nor its preciserole in modulating the incidence of arrhythmias isknown.

For the orderly sequence of repolarization to occur,endocardial myocytes must have longer action-poten-tial durations (APDs) than epicardial cardiac myocytes.This is primarily achieved through differences in therates of repolarization, and, in several mammalian spe-cies, this is linked to regional differences in density ofthe fast component of the transient outward current,Ito,f (Nerbonne and Guo, 2002; Oudit et al., 2001). Thehighest density of Ito,f is seen in epicardial myocytes,whereas the lowest density is observed in endocardialmyocytes (Figure 1). Ito,f is formed by the heterotetra-meric assembly of pore-forming α subunits, Kv4.2 andKv4.3, in association with accessory (β) subunits suchas KChIP2 or frequenin/NCS-1 (Brunet et al., 2004; Guoet al., 2002b; Shibata et al., 2003). In small rodents suchas mice and rats, regional heterogeneity of Ito,f in ventri-cles parallels that of Kv4.2 (Brunet et al., 2004; Guo etal., 2002a; Wickenden et al., 1999a) and possibly K 4.3

v

Cell348

Figure 1. Repolarization Gradients in theMammalian Heart

The gradient of density of Ito,f and Kv4.2 pro-tein is shown as red dots on a diagram ofthe heart. Examples of outward currents andaction potential resulting from the high Ito,f/Kv4.2 in epicardial myocardium and low Ito,f/Kv4.2 in endocardial and septal myocardiumare shown. See text for details.

(Kaprielian et al., 2002; Wickenden et al., 1999a). Inlarger mammals such as human and dog, Kv4.3 is thepredominant Ito,f-encoding α subunit in the heart (Ner-bonne and Guo, 2002; Oudit et al., 2001), and a gradientof KChIP2 may be related to the graded expression ofIto,f (Deschenes et al., 2002; Rosati et al., 2003). Theseobservations suggest that spatial patterning of Ito,f istightly regulated in mammalian cardiac myocytes. How-ever, very little is known about the transcriptional regu-lation of the Ito,f components.

Ito,f downregulation and altered heterogeneity of re-polarization are hallmark features of diseased myocar-dium in humans (Antzelevitch, 2004; Kaab et al., 1998;Nerbonne and Guo, 2002), as well as small-animalmodels (Kaprielian et al., 2002). While increases in thedispersion of repolarization are linked to increasedsusceptibility to ventricular and atrial fibrillation (Antzel-evitch, 2004; Tomaselli and Zipes, 2004), disruption ofIto,f gradients may create substrates for local reentry(Guo et al., 2000; Kuo et al., 2001). For example, micedeficient in Ito due to the loss of KChIP2 or to a domi-nant-negative Kv4.2 transgene show a complete loss ofheterogeneity of repolarization and become suscepti-ble to the induction of polymorphic ventricular tachy-cardia (Guo et al., 2000; Kuo et al., 2001), although lossof current does not directly prove a requirement for acurrent gradient per se. Since ventricular tachycardiacan be lethal by directly impairing pump function or byinducing ventricular fibrillation, further understandingthe basis for regional heterogeneity of repolarization isclearly of physiological and pathophysiological interest.

Cardiac patterning is accomplished largely via tran-scription factors expressed in specific compartmentsof the developing heart (Bruneau, 2002; Habets et al.,2003). The Iroquois homeobox (Irx) genes encode aconserved family of transcription factors that specifythe identity of diverse territories of the body in mostmetazoans by establishing proper spatial and temporalpatterns of target genes (Cavodeassi et al., 2001). Theyencode proteins with a conserved homeodomain of thethree-amino acid length extension (TALE) superclassand a conserved 13 amino acid-residue motif, the Irobox, which is unique to the family (Burglin, 1997). Mam-

ias22be

mdicdgpmTpicnvr

R

EMvrtiteaH(crdat2

malian Irx genes show overlapping expression patterns

n the developing central nervous system, limbs, heart,nd skin, and all six Irx genes display specific expres-ion patterns in the developing heart (Bruneau et al.,000, 2001a; Christoffels et al., 2000; Cohen et al.,000; Mummenhoff et al., 2001). To date, only Irx4 haseen shown to have a role in heart development (Baot al., 1999; Bruneau et al., 2001a; Lebel et al., 2003).In the present study, we show that, in Irx5-deficientice, the cardiac repolarization gradient is flattenedue to increased Kv4.2 potassium-channel expression

n endocardial myocardium, resulting in a selective in-rease of Ito,f and susceptibility to arrhythmias. Myocar-ial Irx5 is expressed in an endocardial-to-epicardialradient in mouse and dog, and Irx5 can repress ex-ression of the gene encoding Kv4.2 (Kcnd2) via recruit-ent of the cardiac transcriptional repressor mBop.hus, a repressor gradient of Irx5 negatively regulatesotassium-channel-gene expression in the heart, form-

ng an inverse Ito,f gradient that ensures coordinatedardiac repolarization. This suggests a novel mecha-ism for the patterning of gene expression in the de-eloping heart and shows a requirement for the cardiacepolarization gradient to reduce the risk of arrhythmia.

esults

lectrophysiological Defects in Irx5-Deficient Miceice homozygous for a targeted deletion of Irx5 are

iable and fertile but have defects in differentiation ofetinal cone bipolar cells and are slightly smaller thanheir wild-type counterparts (Cheng et al., 2005). Todentify a potential role for Irx5 in cardiac form or func-ion, 8-week-old Irx5+/+, Irx5+/−, and Irx5−/− mice werexamined by histology and in vivo echocardiographynd hemodynamics, which revealed no abnormalities.owever, signal-averaged surface electrocardiography

SAECG, data not shown) and in vivo telemetric electro-ardiography in awake, free-moving mice (Figure 2A)evealed that, while heart rate, PR interval, and QRSuration were not affected in Irx5−/− hearts, there wassignificantly decreased amplitude of electrical signals

hat correlate to ventricular repolarization (Danik et al.,002; Liu et al., 2004), which we refer to as the T wave.

This was apparent on leads I and II of the SAECG. The

Irx5 Establishes the Cardiac Repolarization Gradient349

Figure 2. Absent T Wave and Inducible Arrhythmias in Irx5−/− Mice

(A) Representative ECGs in the lead II configuration recorded from awake, free-moving mice with the use of telemetric monitoring. Wild-typemice (+/+) show pronounced downward T wave deflections (arrows). No T waves are evident in ECG recordings of Irx5−/− mice (−/−).(B) Quantitation of T wave amplitude (mean ± SEM). n = 6–8; *p < 0.01.(C and D) Representative intracardiac ECG (IECG, red) and surface ECG (SECG, black) in the lead II configuration obtained from wild-type(Irx5+/+) and Irx5−/− mice.(C) Programmed ventricular stimulation at the right ventricular apex using two extra stimuli (“S2S3”) induced episodes of ventricular tachycar-dia (VT) in Irx5−/− mice, whereas no VTs could be induced in wild-type animals.(D) Rapid overdrive pacing in Irx5−/− mice also induced VTs of long duration.

absolute T wave amplitudes (Tamp) were measured fromthe isoelectric point to the most negative point of the Twave, showing that the T wave remained isoelectric inadult Irx5−/− mice, whereas a pronounced downward Twave deflection was identified in wild-type littermates(Figures 2A and 2B). In the mouse, activation and repo-larization follow a pattern similar to that in larger mam-mals (Liu et al., 2004). The inverted T wave in mice isthe summation of vectors reflecting short (8 ms) differ-ences in transmural action-potential durations in themouse heart (Liu et al., 2004), whereas positive T wavesin larger mammals may reflect larger voltage gradientson either side of the M cells during repolarization (Yanand Antzelevitch, 1998). The change in the T wave seg-ment of the ECG suggested a defect in cardiac repolar-ization in Irx5−/− mice.

Irx5−/− Mice Are Susceptible to Inducible ArrhythmiaDefects in repolarization often result in a predispositionto arrhythmias (Antzelevitch, 2004; Guo et al., 2000;Kuo et al., 2001). As no spontaneous arrhythmias orsudden deaths were observed during 48 hr telemetryrecordings, intracardiac programmed stimulation wasconducted to determine the susceptibility of Irx5−/−

mice to arrhythmia induction. Ventricular effective re-fractory periods were shorter in Irx5−/− mice (38 ± 5 ms,

n = 7) compared to wt mice (59 ± 6 ms, n = 6, p < 0.001).

Using a protocol in which two to four extra stimuli wereapplied at the end of a train of eight paced beats to theapex of the right ventricle, ventricular tachycardia (VT)could be reproducibly induced (>10 episodes) in 3 of 7Irx5−/− mice but in none of 6 wild-type controls (Figure2C). In Irx5−/− mice, induced VTs had a mean length of21 beats (range 19–30 beats) and duration of 926 ms(range 810–1218 ms). With overdrive pacing, in whichcontinuous rapid pacing is applied, VT was induced in4 of 7 Irx5−/− mice, the longest lasting for 180 beatsover 14 s (Figure 2D), but in none of the wild-type con-trols. The propensity for arrhythmias does not seem tocorrelate with any other measured parameter. There-fore, Irx5−/− mice are highly susceptible to inducibletachyarrhythmia. Similar to humans with genetic muta-tions that predispose to arrhythmia in only some cases(Roberts and Brugada, 2003), differences in geneticbackground or stochastic events may confer arrhyth-mia inducibility to some Irx5−/− mice but not others.

Shorter Action-Potential Duration in Irx5−/−

Endocardial MyocytesSince alterations in T wave configuration typically re-flect regional heterogeneity in the timing of ventricularrepolarization, action potentials (APs) were recordedusing whole-cell current-clamp techniques in epicardial

myocytes isolated from the apical region of the outer

Cell350

left ventricle (LV) free wall or endocardial myocytes iso-lated from the base of the interventricular septum. Asexpected, AP durations (APDs) assessed at 25%, 50%,and 90% repolarization (APD25, APD50, and APD90)were longer in endocardial myocytes compared to epi-cardial myocytes derived from wild-type mice (Figures3A and 3B). Remarkably, APD25, APD50, and APD90 ofendocardial myocytes from Irx5−/− mice were abbrevi-ated compared to wild-type endocardial myocytes (p <0.05) and were not different from those measured inepicardial myocytes from Irx5−/− or wild-type mice (Fig-ures 3A and 3B). Wild-type and Irx5−/− myocytes hadsimilar resting membrane potentials and AP amplitudes(data not shown). These results establish that electricalheterogeneity of repolarization is selectively abolishedin the Irx5−/− hearts, consistent with the observedECG changes.

Loss of the Ventricular Transmural Gradientof Ito in Irx5−/− MiceAlterations in repolarization are primarily determined bychanges in K+-channel expression and function (Ner-bonne and Guo, 2002; Oudit et al., 2001). To assess theimpact of Irx5 deficiency on K+-current density, whole-

vgq(rotecwaawletlwIcpacw4

Figure 3. Shortened Endocardial Action Potentials in Irx5−/− Cardio-myocytes

(A) Representative action-potential traces from Irx5+/+ and Irx5−/−

cardiomyocytes from epicardium and endocardium. Irx5−/− endo-cardial cardiomyocytes demonstrate a shortening of the action po-tential (arrows).(B) Mean action-potential durations (APD) measured at 25%, 50%,and 90% repolarization following complete depolarization. n =6–14, *p < 0.05. Scale bars: 20 mV, 25 ms. Data are mean ± SEM.

ccsewb−sdti(I(4Idddbh+tcm2Gsmftaewoiem

cell patch-clamp experiments were conducted on epi-

ardial myocytes from LV apex and endocardial myo-ytes from the interventricular septal base. Figure 4Ahows representative outward K+-current waveforms inpicardial and endocardial myocytes isolated fromild-type and Irx5−/− hearts, with Ca2+ and Na+ currentslocked, recorded following depolarizing steps from40 mV. Corresponding peak current-voltage relation-hips are shown in Figure 4B. In mouse ventricle, theecaying phase of the outward K+ currents can be usedo identify four overlapping currents with distinct kinet-cs: rapidly inactivating transient outward K+ currentsIto), two slowly inactivating K+ currents (Ik,slow1 andk,slow2), and sustained noninactivating currents (Iss)Brunet et al., 2004; Xu et al., 1999). As shown in FigureC, no measurable differences in densities of Ik,slow1,

k,slow2, or Iss were observed between epicardial and en-ocardial myocytes in either wild-type or Irx5−/− mice,emonstrating that these currents do not contribute toifferences in APD and repolarization properties (Ner-onne and Guo, 2002; Oudit et al., 2001). On the otherand, Ito density measured following depolarization to60 mV (Figures 4B and 4C) and maximal Ito conduc-ance (Gto,max, Figure 4D) were greater (p < 0.05) in epi-ardial myocytes than endocardial cells of wild-typeice, consistent with previous work (Brunet et al.,

004; Xu et al., 1999). By contrast, Ito density andto,max in endocardial myocytes from Irx5−/− mice wereignificantly elevated relative to wild-type endocardialyocytes, resulting in values that were not different

rom epicardial myocytes from either Irx5−/− or wild-ype hearts (Figures 4B–4D). The Ito gradient was alsopparent in myocytes isolated from the epicardial andndocardial layers of the LV free wall, and this gradientas also lost in Irx5−/− mice (Figure 4F). Therefore, lossf Irx5 leads to pronounced and specific increases of Ito

n the endocardial myocardium, effectively conferringpicardial myocardium properties to the endocardialyocardium.Despite differences in Ito density and Gto,max, the acti-

ation-gating properties of Ito were identical amongroups, as assessed from estimates of the voltages re-uired for Ito to reach 50% of the maximal conductance

V1/2, data not shown). While this suggests that Ito cur-ents are identical between the different groups, previ-us studies have established that Ito can originate fromwo distinct currents, Ito,fast (Ito,f) and Ito,slow (Ito,s) (Ouditt al., 2001). Ito,f channels are expressed in most myo-ytes of the LV and recover quickly from inactivation,hile Ito,s is found primarily in the ventricular septumnd recovers 100-fold more slowly than Ito,f (Brunet etl., 2004; Guo et al., 1999; Xu et al., 1999). To determinehether disruption of the Ito gradient in Irx5−/− was re-

ated to changes in Ito,f or Ito,s, we examined the recov-ry-from-inactivation properties for Ito (see Figure S1 inhe Supplemental Data available with this article on-ine). As expected, Ito recovery in epicardial myocytesas dominated by a similar single rapid component.

n endocardial myocytes, recovery of Ito was biphasic,onsistent with the existence of both Ito,f and Ito,s. Im-ortantly, the amplitude of the fast component associ-ted with Ito,f was about 2-fold larger (p < 0.01) in endo-ardial myocytes from Irx5−/− versus wild-type hearts,hile the slow Ito,s component was not different (FigureE). Time constants were identical in wild-type and

Irx5−/− mice (data not shown). We conclude that Irx5−/−

Irx5 Establishes the Cardiac Repolarization Gradient351

Figure 4. The Transmural Gradient of Ito Is Eliminated in Irx5−/− Cardiomyocytes

(A) Whole-cell outward K+ currents were recorded from wild-type (+/+) and Irx5−/− (−/−) cardiomyocytes from epicardial (LV apex) and endocar-dial (septum) regions of the heart.(B) Mean ± SEM normalized peak Ito amplitudes are plotted as a function of test pulse (top, epicardium; bottom, endocardium).(C) Normalized current densities (pA/pF) for Ito, Ik,slow1, Ik,slow2, and Iss measured at +60 mV.(D) Maximum current-conductance values for Ito (Gmax).(E) Normalized current densities (pA/pF) for Ito,f and Ito,s measured at +60 mV.(F) Normalized current densities (pA/pF) for Ito measured in myocytes isolated from LV free wall epicardium or endocardium at +60 mV. Forall, n = 6–14, *p < 0.05. Scale bars: 5 nA, 500 ms. Data are mean ± SEM.

mice show a selective increase of Ito,f in endocardialmyocytes, thus flattening the ventricular repolarizationgradient.

Kv4.2 � Subunits Are Increasedin Irx5−/− EndocardiumMouse I channels reflect the heteromeric assembly

to,f

of Kv4.2 and Kv4.3 α subunits and accessory β subunits

such as KChIP2 (Guo et al., 2002a; Shibata et al., 2003).Kv1.5, in turn, underlies Ik,slow1 and has been suggestedto demonstrate regional differences in expression in themouse heart (Brunet et al., 2004; Xu et al., 1999). Asexpected from the electrophysiological data, Kv4.2levels were significantly (p < 0.05) higher in epicardialversus endocardial myocardium in wild-type mice and

were increased in endocardial myocardium of Irx5−/−

Cell352

mice, comparable to epicardial levels (Figures 5A and5B). There were no significant regional differences inthe mean relative densities of Kv4.3 or Kv1.5 in wild-type or Irx5−/− hearts (Figures 4A and 4B). There wasalso a marked increase (p < 0.05) in the levels of Kcnd2mRNA (encoding Kv4.2) in Irx5−/− endocardial myocar-dium, as well as a slight increase in Kcna5 mRNA (Fig-ure 4C). These results confirm that Kv4.2 determinesthe transmural gradient of Ito,f expression in the mouseheart and demonstrate that transcriptional upregulationof Kcnd2 in Irx5−/− mice results in increased expressionof Kv4.2-encoding ion channels and larger density ofIto,f in the endocardial myocardium, thereby eliminatingheterogeneity of repolarization.

Inverse Gradients of Irx5 and Kv4.2 acrossthe Ventricular WallTransverse sections of E14.5 and E16.5 embryos were

cihitscvdm6oipaotp

Figure 5. Kv4.2 Expression Is Increased inthe Endocardium of Irx5−/− Mice

(A) Representative Western blots, using spe-cific anti-Kv4.2, anti-Kv4.3, and anti-Kv1.5antibodies.(B) Quantitation of Western blot analysesshows increased Kv4.2 protein in Irx5−/− en-docardial myocardium.(C) Relative expression of Kcnd2, Kcnd3,Kcna5, and Kcnip2 in the hearts of wild-typeand Irx5−/− mice assessed by quantitativereal-time RT-PCR. mRNA levels (mean ±SEM) are relative to average wild-type epi-cardial values; n = 6–8, *p < 0.05 Irx5+/+ en-docardial myocardium (Endo) compared withIrx5+/+ epicardial myocardium (Epi), **p <0.05 Irx5−/− Endo compared with Irx5+/+

Endo. Data are mean ± SEM.

incubated with a polyclonal antibody specific to the

arboxyl terminus of the Irx5 peptide sequence. Irx5mmunoreactivity was clearly evident in the lungs andeart, showing predominant distribution throughout the

nterventricular septum and endocardial myocardium ofhe LV (Figures 6A, 6B, 6D, and 6E). Irx5−/− embryoshowed only background staining and autofluores-ence from red blood cells (Figures 6C and 6F). In adultentricular sections, robust expression of Irx5 was pre-ominantly observed in the septum and endocardialyocardium of the LV in wild-type hearts (Figures 6G,

I, and 6M). Thus, these results demonstrate a gradientf Irx5 in the mouse heart, with predominant expression

n septum and endocardial myocardium and lower ex-ression in epicardial myocardium. We could not detectn apex-to-base gradient, suggesting that either Irx5nly regulates transmural gradients or the gradient isoo shallow to detect. Western blots of fractionatedroteins from adult mouse hearts confirmed the pre-

dominant expression of Irx5 in endocardial regions and

Irx5 Establishes the Cardiac Repolarization Gradient353

Figure 6. Inverse Gradients of Irx5 and Kv4.2in the Mouse Heart

(A–F) Immunohistochemistry for Irx5 at E14.5(A and B) and E16.5 (D and E). Regions in(A) and (D) are magnified in (B) and (E). Onlybackground staining is apparent in Irx5−/−

embryos (C and F). lv, left ventricle; rv, rightventricle.(G and H) Irx5 (G) and Kv4.2 (H) expressionin adult myocardium.(I) Images in (G) and (H) were pseudocoloredgreen and red, respectively, and digitallymerged.(J) Western blot showing Irx5 expression innuclear extract from epicardial myocardium(lane 1), endocardial myocardium (lane 2),isolated myocytes from epicardial myocar-dium (lane 3), isolated myocytes from endo-cardial myocardium (lane 4), isolated neona-tal myocytes (lane 5), isolated myocytesfrom Irx5−/− epicardial myocardium (lane 6),and isolated myocytes from Irx5−/− endocar-dial myocardium (lane 7). GAPDH is shownas loading control.(K) Quantitation of Irx5 Western blot; n = 3,*p < 0.05.(L) Relative expression of Irx5 mRNA in dogheart; n = 5, *p < 0.05. Data are mean ± SEM.(M) Immunoreactivity of Irx5 and Kv4.2 in theventricles of adult wild-type (+/+) and Irx5−/−

mice (−/−).

lower expression in epicardial myocardium (Figures 6Jand 6K). Irx5 was detected in fractionated nuclear pro-teins from isolated ventricular cardiomyocytes (Figure6J), suggesting that Irx5 functions directly withincardiac myocytes. As the components of Ito, especiallythose that form the gradient, are not completely con-served between mouse and larger mammals, wewished to determine whether the gradient of Irx5 wasconserved. Irx5 mRNA levels in dog myocardium re-vealed a clear endocardial-to-epicardial gradient of Irx5transcript (Figure 6L), similar to that of the mouse.

Parallel tissue sections of adult wild-type and Irx5−/−

ventricles were also stained using an antibody to Kv4.2.In wild-type hearts, Kv4.2 immunoreactivity was ex-pressed in a steep gradient across the left ventricularfree wall and was of low abundance in the endocardialmyocardium and interventricular septum (Figures 6Hand 6M). Thus, the expression of Kv4.2 is a mirror imageof Irx5 protein distribution. Consistent with the Western

blot and RT-PCR results, in Irx5−/− ventricular sections,

homogeneous Kv4.2 immunoreactivity was detectedthroughout the entire ventricular myocardium, reflect-ing increased expression in endocardial myocardium(Figure 6M). Therefore, an endocardial-epicardial gradi-ent of Irx5 inversely correlates with the epicardial-endo-cardial gradient of Kv4.2.

Irx5 Represses Kcnd2, the Gene Encoding Kv4.2Irx5 is expressed in cardiac myocytes and thereforemay act directly on Kcnd2 in these cells. Irx proteinsact mainly as transcriptional repressors (Gomez-Skar-meta et al., 2001; Itoh et al., 2002; Kudoh and Dawid,2001; Matsumoto et al., 2004) and occasionally as acti-vators (Bao et al., 1999; Matsumoto et al., 2004). Thus,Irx5 may act to repress Kcnd2 in endocardial myocytes.To test this possibility, we examine the function of Irx5on the rat Kcnd2 promoter, which shares a high degreeof homology with mouse and human Kcnd2 (Jia andTakimoto, 2003). We cotransfected isolated neonatal

mouse cardiomyocytes with Kcnd2 reporter constructs

Cell354

and with an Irx5 expression construct. Consistent withthe hypothesis that graded levels of Irx5 regulate theKcnd2 gradient, increasing amounts of Irx5 dose-dependently repressed Kcnd2-luciferase activity (Fig-ure 7A). In contrast, Irx5 activated Kcnd2-luciferase innoncardiac COS7 and 10T1/2 cells (Figures 7B and 7D).There is currently no known consensus Irx binding site,and therefore we cannot determine whether Irx5 bindsdirectly or indirectly to the Kcnd2 promoter. We con-clude that Irx5 can dose-dependently repress the activ-ity of the Kcnd2 promoter and hypothesized that Irx5repressor activity in cardiac myocytes requires acardiac-specific corepressor protein.

Irx5 Can Interact with mBop, a Cardiac Corepressor,to Repress Kcnd2In a screen to identify cardiac transcription factors thatinteract with the MYND- and SET-domain muscle-restricted transcriptional repressor mBop (Gottlieb etal., 2002), we identified Irx4 as a strong interacting part-ner (C.Y.P. and D.S., unpublished data). Based on thehigh degree of similarity between Irx4 and Irx5, we hy-pothesized that Irx5 would also interact with mBop.Coimmunoprecipitation assays in COS7 cells demon-strated that both Irx4 and Irx5 interact with mBop (Fig-ure 7C). Coexpression of mBop in 10T1/2 cells resultedin a marked abrogation of the Irx5-dependent activa-tion of Kcnd2-luciferase (Figure 7D). As mBop-medi-ated repression is thought to rely on recruitment of his-tone deacetylases (HDACs) (Gottlieb et al., 2002), weexamined whether the Bop-mediated repressive effectoccurred via HDACs by using the HDAC inhibitortrichostatin A (TSA). Addition of TSA relieved the inhibi-tion by mBop of activation by Irx5 (Figure 7E). Struc-ture-function analysis of Irx5 (Figures 7F–7H) demon-strated that deletion of the homeodomain (Irx5�HD) orall residues following the homeodomain (Irx5�C1) pre-vented activation by Irx5, while removal of the 153C-terminal residues (Irx5�C2), which include the con-served Iro box (Burglin, 1997), did not affect activationbut prevented repression of activation by mBop. Allmutants localized to the nucleus, although Irx5�HDnuclear localization was impaired (Figure S2). Consis-tent with these observations, mBop interacted onlyweakly or not at all with C-terminal-deletion forms of Irx5(Figure 7H). Knockdown of mBop mRNA by RNA interfer-ence in cardiomyocytes slightly decreased expression ofKcnd2-luciferase and, importantly, eliminated the abilityof exogenous Irx5 to repress Kcnd2-luciferase (Figure 7I),indicating that endogenous mBop may be a critical factorfor the repressive actions of Irx5. These results demon-strate that Irx5 can repress Kcnd2, and this is likely tooccur via interaction with mBop and the recruitment ofHDACs, although other proteins may also be involvedin the repressive actions of Irx5. This provides a novelmechanism by which Irx transcription factors exert re-pressive effects during development (Figure 7J).

Discussion

Our results demonstrate that a repressor gradient ofIrx5 is essential for regulating cardiac K+-channel-geneexpression, forming an inverse I gradient and ensur-

ivtittds

ITmpetwtlwacscd(lmasemuctKtfsgwd

v(tnppamitgwKetfcirh

to,f

ng the concordant propagation of repolarization in theentricular myocardium. These findings demonstratehe importance of an epicardial-to-endocardial repolar-zation gradient in prevention of potentially lethal ven-ricular tachycardia. This may have relevance to pa-ients with heart disease associated with gradientisruption, who are recognized as being at high risk ofudden cardiac death.

rx5 and the Ito,f Gradienthe electrophysiological composition of the ventricularyocardium is largely heterogeneous due to the ex-ression of distinct cardiac ion channels. Heterologousxpression of Kv4.2-, Kv4.3-, and Kv1.4-channel pro-eins, for example, has been shown to produce currentsith biophysical properties resembling, to varying ex-

ents, Ito measured in myocytes, indicating that they areikely the primary correlates of cardiac transient out-ard K+ currents (Nerbonne and Guo, 2002; Oudit etl., 2001). Manipulating the expression of putative K+-hannel genes in vivo has also allowed a better under-tanding of their role in generating contribution toardiac Ito,f and Ito,s. For example, overexpression ofominant-negative Kv4.2 α subunits attenuates Ito,f

Barry et al., 1998; Wickenden et al., 1999b), while theoss of the β subunit KChIP2, which is required for tetra-

eric channel assembly (Guo et al., 2002a; Shibata etl., 2003), abolishes Ito (Kuo et al., 2001). The presenttudy reveals a substantial increase in the functionalxpression of Ito,f in endocardial myocytes of Irx5−/−

ice, in conjunction with a selective and coordinatedpregulation of Kv4.2 mRNA and protein levels in endo-ardial myocardium. This provides conclusive evidencehat Kv4.2 is a major component of mouse Ito,f and thatv4.2 gradients are responsible for the Ito,f gradient in

he mouse heart. Since we did not detect regional dif-erences in the levels of other K+ currents, our findingsupport the notion that the primary determinant of re-ional heterogeneity of repolarization and peak out-ard K+ currents in mouse ventricular myocytes is theifferential expression of Ito,f.As in mouse, a Kcnd2 mRNA gradient exists in rat

entricles that parallels the transmural gradient of Ito,f

Wickenden et al., 1999a). However, it should be notedhat the formation of the transmural gradient of Ito,f isot entirely conserved between mammals. For exam-le, in contrast to mouse and rat, Kcnd2 is not ex-ressed in canine or human myocardium (Nerbonnend Guo, 2002; Oudit et al., 2001). Instead, in largerammals, a transcriptional gradient of Kcnip2 (encod-

ng KChIP2) across the ventricular wall is thought to behe primary determinant that underlies the transmuralradient of Ito,f expression (Rosati et al., 2003), althoughhether the Kcnip2 mRNA gradient is paralleled by aChIP2 protein gradient has been questioned (Deschenest al., 2002). Common among mammals, however, ishat Ito,f gradients form the basis for the transmural dif-erences in repolarization across the ventricular myo-ardium. As in mouse, Irx5 is expressed in a gradient

n dog heart, suggesting that it may be a regulator ofepolarization gradients in larger mammals, includingumans, perhaps via other genes such as Kcnip2.

Irx5 Establishes the Cardiac Repolarization Gradient355

Figure 7. Irx5 Directly Represses the Kcnd2 Promoter

(A) Kcnd2 −1094–+592-luciferase and Kcnd2 −432–+592-luciferase (but not Kcnd2 −3162–+592-luciferase) are strongly activated in neonatalcardiac myocytes. Addition of an Irx5 expression construct (Irx5) reduces the activity of Kcnd2 reporters. For this and all other panels: +, 100ng; ++, 250 ng; +++, 500 ng; ++++, 1000 ng Irx5 expression construct.(B) Irx5 activates Kcnd2-luciferase in COS cells.(C) mBop interacts with Irx4 and Irx5. Immunoprecipitation using anti-HA antibodies followed by immunoblotting against FLAG shows thatmBop (arrow) can interact with Irx4 and Irx5.(D) mBop prevents activation of Kcnd2 −1094–+592-luciferase by Irx5. Similar results were obtained with Kcnd2 −432–+592-luciferase.(E) Histone deacetylase inhibition by trichostatin A (TSA) relieves the inhibition of Irx5 activity by mBop. In (D) and (E) for mBop: +, 500 ng;++, 1000 ng expression constructs.(F) Diagram of Irx5 proteins used in (G) and (H). HD, homeodomain; Iro, Iro box.(G) Irx5�HD or Irx5�C2 no longer activates transcription, while Irx5�C1 activates but is not repressed by mBop.(H) Coimmunoprecipitations show that mBop cannot interact with Irx5�C1 or Irx5�C2.(I) mBop is required for Irx5-mediated repression in cardiac myocytes. siRNAs against mBop (+, 25 ng; ++, 50 ng) reduced expression ofKcnd2 −432–+592-luciferase and prevented Irx5-mediated repression. Data are mean ± SEM.(J) Model for the role of Irx5; see text for details.

An Inverse Repressor Gradient of Irx5 Patterns theKv4.2 Transmural Gradient by Recruiting mBOPThe graded transmural expression of Irx5 is necessaryfor maintaining the K 4.2 ventricular gradient by sup-

v

pressing Kv4.2 expression in regions where Irx5 ishighly expressed. Consistent with this hypothesis, Irx5dose-dependently inhibited the activity of a Kcnd2 pro-moter construct in cardiac myocytes. In contrast to its

Cell356

repressor activity in cardiac myocytes, Irx5 activatedthe Kcnd2 promoter construct in noncardiac cells.Based on these observations, we hypothesized that, incardiac cells, Irx5 associates with a corepressor. In-deed, we show that mBop, a cardiac corepressor(Gottlieb et al., 2002), can associate with Irx5 and re-press its activation of Kcnd2 in noncardiac cells. Fur-thermore, we show that endogenous mBop is importantfor the repressive activity of Irx5 in cardiac myocytes.We propose a model (Figure 7J) whereby Irx5 acts onthe Kcnd2 promoter and locally suppresses the expres-sion of Kv4.2 by recruiting mBop, which in turn recruitsHDACs to repress Kcnd2 transcription. The graded re-pressive effects of Irx5 would therefore be accom-plished by shifting the stoichiometry of transcriptionalactivator and repressor complexes toward a repressivestate with increasing amounts of Irx5. Together, ourfindings of an Irx5 repressor gradient via corepressorrecruitment demonstrate a novel mechanism for theformation of cardiac transcriptional gradients.

Other members of the Iroquois gene family are ex-pressed in the heart in unique spatiotemporal patterns,and insights into their various roles in tissue specifica-tion emphasize their importance for physiologicalcardiac function. For example, the expression of Irx4 inboth birds and mammals is confined to the ventriclesthroughout heart development, and gain-of-functionand loss-of-function studies demonstrate an essentialrole for Irx4 in regulating the expression of genes tomaintain the ventricular phenotype, in part via repres-sion (Bao et al., 1999; Bruneau et al., 2000, 2001a). AsIrx4 can also interact with mBop, it is likely that its re-pressive actions are also mediated by this interactionand that corepressor recruitment is a general feature ofgene regulation by Irx proteins.

Clinical Implications of Alterationin the Repolarization GradientArrhythmias are the leading cause of sudden death inpatients with heart failure or cardiomyopathies (Toma-selli and Zipes, 2004). Altered patterns of repolarizationare important aspects of heart failure that are thoughtto contribute its arrhythmogenicity (Antzelevitch, 2004;Tomaselli and Zipes, 2004). Genetic diseases affectingthe repolarization properties of the heart, such as longand short QT syndrome, are also important causes ofsudden death (Roberts and Brugada, 2003). In long QTsyndrome (LQTS), increased spatial dispersion of repo-larization associated with delayed repolarization pro-vides a substrate for triggered arrhythmia (Antzelevitch,2004). Similarly, in short QT syndrome (SQTS), ac-celerated repolarization contributes to the substrate forventricular tachycardia and sudden death (Extramianaand Antzelevitch, 2004; Gaita et al., 2003). Although anaccurate assessment of the QT interval in Irx5−/− micewas complicated by the absence of a well-defined Twave, loss of Irx5 yields a remarkable gain of functionof Ito,f that reproduces the pathogenesis associatedwith SQTS.

The repolarization gradient exists in all mammalianspecies, and therefore must have a critical role in nor-mal heart function. The differential expression of Ito,f

across the various cell layers of the heart ensures spa-

tth2osra2tmitiditwtldlamrdrmedsmp

CWlpmtgIgtemppf

E

AIgcf

PEepemba

ial heterogeneity of APD and refractory periods,hereby synchronizing cardiac repolarization and en-ancing electrical stability of the heart (Antzelevitch,004; Nerbonne and Guo, 2002). Spatial heterogeneityf repolarization has also been shown to assist in theynchronization of Ca2+ release from the sarcoplasmiceticulum, leading to enhanced mechanical stabilitynd pump efficiency (Kaprielian et al., 2002; Sah et al.,002). Although increased heterogeneity of repolariza-ion has been shown in larger species to be proarrhyth-ic, in Irx5−/− mice, the loss of heterogeneity of repolar-

zation results in increased susceptibility to ventricularachycardia. As the flattened Ito,f gradient in Irx5−/− mices associated with a marked abbreviation of endocar-ial APD and refractoriness, this may demonstrate the

mportance of a prolonged refractoriness within the ven-ricular myocardium, which serves to increase theavelength (product of refractory period and conduc-

ion velocity) of the reentrant wave beyond the pathength available in the mouse heart, thus preventing theevelopment of reentry. Indeed, the ready inducibility of

ife-threatening polymorphic VT/VF in SQTS has beenttributed in part to abbreviation of refractoriness of theyocardium (Extramiana and Antzelevitch, 2004). Our

esults therefore suggest that the loss of the Kv4.2 gra-ient results in an arrhythmogenic substrate and thus

eveal the importance of the repolarization gradient inaintaining an arrhythmia-free myocardium. Knowl-

dge of the mechanisms regulating repolarization gra-ients in the mammalian heart represent an importanttepping stone toward potential therapies for arrhyth-ogenic substrates by targeting the Irx5/mBop/Ito,f

athway.

onclusionse have shown that Irx5 establishes the cardiac repo-

arization gradient by its repressive actions on the Kv4.2otassium-channel gene. The susceptibility to arrhyth-ias in Irx5−/− mice provides compelling evidence that

he repolarization gradient per se is an important safe-uard against reentrant arrhythmias. The gradient of

rx5 in the mouse heart is analogous to the Brinker (Brk)radient in Drosophila, whereby graded levels of theranscriptional repressor Brk establish patterned genexpression that serves to transduce the gradient of theorphogen Decapentaplegic (Muller et al., 2003). Weropose that the Irx5 repressor gradient acts via core-ressor recruitment, demonstrating a novel mechanism

or the formation of cardiac transcriptional gradients.

xperimental Procedures

nimalsrx5+/− mice, maintained on a mixed CD-1 strain background, wereenerated as described elsewhere (Cheng et al., 2005) and were inter-rossed to generate Irx5−/− and Irx5+/+ mice. All animals were caredor according to institutional animal-care requirements.

hysiological Measurementschocardiography, in vivo LV physiological measurements, andlectrophysiological analysis of adult mice (8 to 12 weeks old) wereerformed as previously described (Bruneau et al., 2001b; Mungruet al., 2002). In vivo electrophysiology studies were performed inice aged 4 to 6 months and anesthetized with sodium pentobar-ital (0.033 mg/g i.p.) (Zhu et al., 2003). All studies were performednd analyzed by a blinded operator.

Irx5 Establishes the Cardiac Repolarization Gradient357

Myocyte Isolation and ElectrophysiologyVentricular myocytes were dissociated from the ventricular apexand septal base, or from left ventricular free wall epicardium andendocardium, from adult male mice (8 to 12 weeks old) using pro-cedures previously developed to distinguish regional differences inK+-current expression (Brunet et al., 2004; Xu et al., 1999). Actionpotentials and K+ currents were recorded at room temperature(20°C–23°C) with the whole-cell patch-clamp technique under cur-rent-clamp and voltage-clamp mode, respectively (Sun et al., 2004).A modified double-pulse protocol was used to determine the re-covery rate of Ito from steady-state inactivation (Wickenden et al.,1999a). The action potentials and current recordings were analyzedusing pClamp software (Clampfit 9.0, Axon). The decay phase ofoutward K+ currents was rigorously fit with a triexponential functionto yield estimates of four kinetically distinct K+ currents (Ito, IK,slow1,IK,slow2, Isus) using the AMC maximum-likelihood procedure. (Sun etal., 2004). Monoexponential or biexponential fits were used to fitrecovery-from-inactivation data (Wickenden et al., 1999a).

Analysis of mRNA and Protein LevelsRNA was isolated from LV apex and septal base from adult (8- to11-week-old) mice and epicardial and endocardial sections fromthe LV of hearts from mongrel dogs (weighing 20–25 kg). Quantita-tive real-time RT-PCR was performed with assay-on-demand Taq-Man probes (Applied Biosystems): Kcnd2 (Mm00498065_m1),Kcnd3 (Mm00498260_m1), Kcna5 (Mm00524346_s1), Kcnip2(Mm000518914_m1), and Gapdh (rodent GAPDH control). Se-quences for custom dog Irx5 Taqman probes were: forward primer,5#-GCAAGGGCGACTCCGA-3#; reverse primer, 5#-CGCAGCCGCCTTCTG-3#; TaqMan probe, 6-FAM 5#-TCCGCTCCTCCTGCTTC-3#.Western blot analysis was performed on 80 to 100 �g of nuclear ormembrane protein using rabbit polyclonal anti-Kv4.2 (1:200), anti-Kv4.3 (1:200), anti-Kv1.5 (1:200, all from Chemicon), rabbit anti-mouseGAPDH (1:5000, Amersham Biosciences), or anti-Irx5 antisera(1:1500). Affinity-purified Irx5 antibodies were from rabbit poly-clonal antisera raised against the carboxyl terminus of Irx5 fusedto GST.

ImmunohistochemistryTissues were fixed in 4% paraformaldehyde and embedded in par-affin. For Irx5, sections were incubated with Irx5 antiserum (1:100)overnight at 4°C and then with an anti-rabbit secondary antibodycoupled to biotin. For Kv4.2, the Mouse-On-Mouse kit (Vector) wasused. Sections were incubated with monoclonal Kv4.2 antibodies(K57/27, Dr. J. Trimmer) (1:10) overnight with anti-mouse IgG (H+L)secondary antibody coupled to biotin. The Vectastain ABC-AP kit(Vector) and the red substrate kit (Vector) were used to visualizethe signal.

Coimmunoprecipitation and Luciferase Reporter Gene AssaysTransfections of COS7 cells, 10T1/2 cells, or neonatal mouse ven-tricular myocytes were performed as previously described (Bru-neau et al., 2001b), using Fugene6 (Roche) or Lipofectamine 2000(Invitrogen). Luciferase assays and coimmunoprecipitations wereperformed as previously described (Bruneau et al., 2001b). mBopsiRNA (Dharmacon) sequences were sense 5#-UCACAAGAACGAGUGCGCUTT-3#, antisense 5#-AGCGCACUCGUUCUUGUGATT-3#.

Statistical AnalysisStatistical comparisons were performed by Student’s t test or one-way ANOVA. p < 0.05 was considered significant.

Supplemental DataSupplemental Data include two figures and can be found with thisarticle online at http://www.cell.com/cgi/content/full/123/2/347/DC1/.

Acknowledgments

We thank J.N. Wylie for assistance with luciferase assays, K. Taki-moto for the Kcnd2 reporter constructs, and J. Trimmer for Kv4.2antiserum. This work was supported by grants from the CanadianInstitutes of Health Research (CIHR) to B.G.B, C.-c.H., M.H., and

P.H.B.; a CIHR collaborative grant (B.G.B. and M.H.); the NIH/NHLBI(D.S.); and the Research Grants Council of the Hong Kong SAR,China (Project # CityU 1164/02M) to S.H.C. D.L.C. and P.A. wererecipients of National Science and Engineering Council of CanadaScholarships, M.L. holds a CIHR Scholarship, and E.P.A. was partlysupported by the Hospital for Sick Children Research Training Pro-gram. M.H. is a CIHR Clinician-Scientist, M.H. and P.H.B. are CareerInvestigators of the Heart and Stroke Foundation of Ontario,C.-c.H. is a National Cancer Institute of Canada Scholar, and B.G.B.holds a Canada Research Chair in Developmental Cardiology.

Received: February 14, 2005Revised: July 4, 2005Accepted: August 1, 2005Published: October 20, 2005

References

Antzelevitch, C. (2004). Cellular basis and mechanism underlyingnormal and abnormal myocardial repolarization and arrhythmogen-esis. Ann. Med. 36 (Suppl. 1), 5–14.

Bao, Z.-Z., Bruneau, B.G., Seidman, J.G., Seidman, C.E., andCepko, C.L. (1999). Irx4 regulates chamber-specific gene expres-sion in the developing heart. Science 283, 1161–1164.

Barry, D.M., Xu, H., Schuessler, R.B., and Nerbonne, J.M. (1998).Functional knockout of the transient outward current, long-QT syn-drome, and cardiac remodeling in mice expressing a dominant-negative Kv4 alpha subunit. Circ. Res. 83, 560–567.

Bruneau, B.G. (2002). Transcriptional regulation of vertebratecardiac morphogenesis. Circ. Res. 90, 509–519.

Bruneau, B.G., Bao, Z.Z., Tanaka, M., Schott, J.J., Izumo, S.,Cepko, C.L., Seidman, J.G., and Seidman, C.E. (2000). Cardiac ex-pression of the ventricle-specific homeobox gene Irx4 is modulatedby Nkx2–5 and dHand. Dev. Biol. 217, 266–277.

Bruneau, B.G., Bao, Z.Z., Fatkin, D., Xavier-Neto, J., Georgako-poulos, D., Maguire, C.T., Berul, C.I., Kass, D.A., Kuroski-de Bold,M.L., de Bold, A.J., et al. (2001a). Cardiomyopathy in Irx4-deficientmice is preceded by abnormal ventricular gene expression. Mol.Cell. Biol. 21, 1730–1736.

Bruneau, B.G., Nemer, G., Schmitt, J.P., Charron, F., Robitaille, L.,Caron, S., Conner, D., Gessler, M., Nemer, M., Seidman, C.E., andSeidman, J.G. (2001b). A murine model of Holt-Oram syndrome de-fines roles of the T-box transcription factor Tbx5 in cardiogenesisand disease. Cell 106, 709–721.

Brunet, S., Aimond, F., Guo, W., Li, H., Eldstrom, J., Fedida, D.,Yamada, K.A., and Nerbonne, J.M. (2004). Heterogeneous expres-sion of repolarizing, voltage-gated K+ currents in adult mouse ven-tricles. J. Physiol. 559, 103–120.

Burglin, T.R. (1997). Analysis of TALE superclass homeobox genes(MEIS, PBC, KNOX, Iroquois, TGIF) reveals a novel domain con-served between plants and animals. Nucleic Acids Res. 25, 4173–4180.

Cavodeassi, F., Modolell, J., and Gomez-Skarmeta, J.L. (2001). TheIroquois family of genes: from body building to neural patterning.Development 128, 2847–2855.

Cheng, C.W., Chow, R.L., Lebel, M., Sakuma, R., Cheung, H.O.,Thanabalasingham, V., Zhang, X., Bruneau, B.G., Birch, D.G., Hui,C.-c., et al. (2005). The Iroquois homeobox gene, Irx5, is requiredfor retinal cone bipolar development. Dev. Biol., in press.

Christoffels, V.M., Keijser, A.G., Houweling, A.C., Clout, D.E., andMoorman, A.F. (2000). Patterning the embryonic heart: identifica-tion of five mouse Iroquois homeobox genes in the developingheart. Dev. Biol. 224, 263–274.

Cohen, D.R., Cheng, C.W., Cheng, S.H., and Hui, C.C. (2000). Ex-pression of two novel mouse Iroquois homeobox genes during neu-rogenesis. Mech. Dev. 91, 317–321.

Danik, S., Cabo, C., Chiello, C., Kang, S., Wit, A.L., and Coromilas,J. (2002). Correlation of repolarization of ventricular monophasicaction potential with ECG in the murine heart. Am. J. Physiol. HeartCirc. Physiol. 283, H372–H381.

Cell358

Deschenes, I., DiSilvestre, D., Juang, G.J., Wu, R.C., An, W.F., andTomaselli, G.F. (2002). Regulation of Kv4.3 current by KChIP2 splicevariants: a component of native cardiac I(to)? Circulation 106,423–429.

Extramiana, F., and Antzelevitch, C. (2004). Amplified transmuraldispersion of repolarization as the basis for arrhythmogenesis in acanine ventricular-wedge model of short-QT syndrome. Circulation110, 3661–3666. Published online November 29, 2004. 10.1161/01.CIR.0000143078.48699.0C

Gaita, F., Giustetto, C., Bianchi, F., Wolpert, C., Schimpf, R., Ric-cardi, R., Grossi, S., Richiardi, E., and Borggrefe, M. (2003). ShortQT Syndrome: a familial cause of sudden death. Circulation 108,965–970. Published online August 8, 2003. 10.1161/01.CIR.0000085071.28695.C4

Gomez-Skarmeta, J., de La Calle-Mustienes, E., and Modolell, J.(2001). The Wnt-activated Xiro1 gene encodes a repressor that isessential for neural development and downregulates Bmp4. Devel-opment 128, 551–560.

Gottlieb, P.D., Pierce, S.A., Sims, R.J., Yamagishi, H., Weihe, E.K.,Harriss, J.V., Maika, S.D., Kuziel, W.A., King, H.L., Olson, E.N., et al.(2002). Bop encodes a muscle-restricted protein containing MYNDand SET domains and is essential for cardiac differentiation andmorphogenesis. Nat. Genet. 31, 25–32.

Guo, W., Xu, H., London, B., and Nerbonne, J.M. (1999). Molecularbasis of transient outward K+ current diversity in mouse ventricularmyocytes. J. Physiol. 521, 587–599.

Guo, W., Li, H., London, B., and Nerbonne, J.M. (2000). Functionalconsequences of elimination of i(to,f) and i(to,s): early afterdepolar-izations, atrioventricular block, and ventricular arrhythmias in micelacking Kv1.4 and expressing a dominant-negative Kv4 alpha sub-unit. Circ. Res. 87, 73–79.

Guo, W., Li, H., Aimond, F., Johns, D.C., Rhodes, K.J., Trimmer,J.S., and Nerbonne, J.M. (2002a). Role of heteromultimers in thegeneration of myocardial transient outward K+ currents. Circ. Res.90, 586–593.

Guo, W., Malin, S.A., Johns, D.C., Jeromin, A., and Nerbonne, J.M.(2002b). Modulation of Kv4-encoded K(+) currents in the mamma-lian myocardium by neuronal calcium sensor-1. J. Biol. Chem. 277,26436–26443.

Habets, P.E., Moorman, A.F., and Christoffels, V.M. (2003). Regula-tory modules in the developing heart. Cardiovasc. Res. 58, 246–263.

Itoh, M., Kudoh, T., Dedekian, M., Kim, C.H., and Chitnis, A.B.(2002). A role for iro1 and iro7 in the establishment of an antero-posterior compartment of the ectoderm adjacent to the midbrain-hindbrain boundary. Development 129, 2317–2327.

Jia, Y., and Takimoto, K. (2003). GATA and FOG2 transcription fac-tors differentially regulate the promoter for Kv4.2 K(+) channel genein cardiac myocytes and PC12 cells. Cardiovasc. Res. 60, 278–287.

Kaab, S., Dixon, J., Duc, J., Ashen, D., Nabauer, M., Beuckelmann,D.J., Steinbeck, G., McKinnon, D., and Tomaselli, G.F. (1998). Mo-lecular basis of transient outward potassium current downregula-tion in human heart failure: a decrease in Kv4.3 mRNA correlateswith a reduction in current density. Circulation 98, 1383–1393.

Kaprielian, R., Sah, R., Nguyen, T., Wickenden, A.D., and Backx,P.H. (2002). Myocardial infarction in rat eliminates regional hetero-geneity of AP profiles, I(to) K(+) currents, and [Ca(2+)](i) transients.Am. J. Physiol. Heart Circ. Physiol. 283, H1157–H1168.

Kudoh, T., and Dawid, I.B. (2001). Role of the iroquois3 homeoboxgene in organizer formation. Proc. Natl. Acad. Sci. USA 98, 7852–7857.

Kuo, H.C., Cheng, C.F., Clark, R.B., Lin, J.J., Lin, J.L., Hoshijima,M., Nguyen-Tran, V.T., Gu, Y., Ikeda, Y., Chu, P.H., et al. (2001). Adefect in the Kv channel-interacting protein 2 (KChIP2) gene leadsto a complete loss of I(to) and confers susceptibility to ventriculartachycardia. Cell 107, 801–813.

Lebel, M., Agarwal, P., Cheng, C.W., Kabir, M.G., Chan, T.Y., Thana-balasingham, V., Zhang, X., Cohen, D.R., Husain, M., Cheng, S.H.,et al. (2003). The Iroquois homeobox gene Irx2 is not essential for

ni

LaoJ

MT(FN

MKd

MRM

MSmg7

Nvc4

OPwJ

RA

RMpc

Sroc

SSKdt

S(cr

Ti

W(sH

WBK1

Xdm

YTs

ZMemE

ormal development of the heart and midbrain-hindbrain boundaryn mice. Mol. Cell. Biol. 23, 8216–8225.

iu, G., Iden, J.B., Kovithavongs, K., Gulamhusein, R., Duff, H.J.,nd Kavanagh, K.M. (2004). In vivo temporal and spatial distributionf depolarization and repolarization and the illusive murine T wave.. Physiol. 555, 267–279.

atsumoto, K., Nishihara, S., Kamimura, M., Shiraishi, T., Otoguro,., Uehara, M., Maeda, Y., Ogura, K., Lumsden, A., and Ogura, T.2004). The prepattern transcription factor Irx2, a target of theGF8/MAP kinase cascade, is involved in cerebellum formation.at. Neurosci. 7, 605–612.

uller, B., Hartmann, B., Pyrowolakis, G., Affolter, M., and Basler,. (2003). Conversion of an extracellular Dpp/BMP morphogen gra-ient into an inverse transcriptional gradient. Cell 113, 221–233.

ummenhoff, J., Houweling, A.C., Peters, T., Christoffels, V.M., anduther, U. (2001). Expression of Irx6 during mouse morphogenesis.ech. Dev. 103, 193–195.

ungrue, I.N., Gros, R., You, X., Pirani, A., Azad, A., Csont, T.,chulz, R., Butany, J., Stewart, D.J., and Husain, M. (2002). Cardio-yocyte overexpression of iNOS in mice results in peroxynitriteeneration, heart block, and sudden death. J. Clin. Invest. 109,35–743.

erbonne, J.M., and Guo, W. (2002). Heterogeneous expression ofoltage-gated potassium channels in the heart: roles in normal ex-itation and arrhythmias. J. Cardiovasc. Electrophysiol. 13, 406–09.

udit, G.Y., Kassiri, Z., Sah, R., Ramirez, R.J., Zobel, C., and Backx,.H. (2001). The molecular physiology of the cardiac transient out-ard potassium current (I(to)) in normal and diseased myocardium.. Mol. Cell. Cardiol. 33, 851–872.

oberts, R., and Brugada, R. (2003). Genetics and arrhythmias.nnu. Rev. Med. 54, 257–267.

osati, B., Grau, F., Rodriguez, S., Li, H., Nerbonne, J.M., andcKinnon, D. (2003). Concordant expression of KChIP2 mRNA,

rotein and transient outward current throughout the canine ventri-le. J. Physiol. 548, 815–822.

ah, R., Ramirez, R.J., and Backx, P.H. (2002). Modulation of Ca(2+)elease in cardiac myocytes by changes in repolarization rate: rolef phase-1 action potential repolarization in excitation-contractionoupling. Circ. Res. 90, 165–173.

hibata, R., Misonou, H., Campomanes, C.R., Anderson, A.E.,chrader, L.A., Doliveira, L.C., Carroll, K.I., Sweatt, J.D., Rhodes,.J., and Trimmer, J.S. (2003). A fundamental role for KChIPs inetermining the molecular properties and trafficking of Kv4.2 po-

assium channels. J. Biol. Chem. 278, 36445–36454.

un, H., Oudit, G.Y., Ramirez, R.J., Costantini, D., and Backx, P.H.2004). The phosphoinositide 3-kinase inhibitor LY294002 enhancesardiac myocyte contractility via a direct inhibition of Ik,slow cur-ents. Cardiovasc. Res. 62, 509–520.

omaselli, G.F., and Zipes, D.P. (2004). What causes sudden deathn heart failure? Circ. Res. 95, 754–763.

ickenden, A.D., Jegla, T.J., Kaprielian, R., and Backx, P.H.1999a). Regional contributions of Kv1.4, Kv4.2, and Kv4.3 to tran-ient outward K+ current in rat ventricle. Am. J. Physiol. 276,1599–H1607.

ickenden, A.D., Lee, P., Sah, R., Huang, Q., Fishman, G.I., andackx, P.H. (1999b). Targeted expression of a dominant-negative(v)4.2 K(+) channel subunit in the mouse heart. Circ. Res. 85,067–1076.

u, H., Guo, W., and Nerbonne, J.M. (1999). Four kinetically distinctepolarization-activated K+ currents in adult mouse ventricularyocytes. J. Gen. Physiol. 113, 661–678.

an, G.X., and Antzelevitch, C. (1998). Cellular basis for the normalwave and the electrocardiographic manifestations of the long-QT

yndrome. Circulation 98, 1928–1936.

hu, W., Lepore, J.J., Saba, S., Joseph, S., Link, M.S., Homond,.K., Estes, N.A.M., Wang, P.J., and Leiden, J.M. (2003). Cardiac

lectrophysiologic abnormalities in the CREBA133 transgenicouse model of idiopathic dilated cardiomyopathy. J. Cardiovasc.lectrophysiol. 14, 982–989.