A Molecular Pathway Including Id2, Tbx5, and Nkx2-5 Required for Cardiac Conduction System Development

Ivan P.G. Moskowitz,1,2,9 Jae B. Kim,1,3 Meredith L. Moore,1 Cordula M. Wolf,1 Michael A. Peterson,1,2 Jay Shendure,1 Marcelo A. Nobrega,4 Yoshifumi Yokota,5 Charles Berul,6 Seigo Izumo,7 J.G. Seidman,1,8 and Christine E. Seidman1,3,8,* 1 Department of Genetics, Harvard Medical School, Boston, MA 02115, USA 2 Cardiac Registry, Department of Pathology, Children’s Hospital, Boston, MA 02115, USA 3 Howard Hughes Medical Institute and Cardiovascular Division, Brigham & Women’s Hospital, Boston, MA 02115, USA 4 Department of Human Genetics, The University of Chicago, Chicago, IL 60637, USA 5 Department of Molecular Genetics, School of Medicine, University of Fukui, 23-3 Shimoaizuki, Matsuoka, Fukui 911-1193, Japan 6 Department of Cardiology, Children’s Hospital, Boston, MA 02115, USA 7 Novartis Institutes for BioMedical Research, 100 Technology Square, Suite 8402, Cambridge, MA 02139, USA 8 These authors contributed equally to this work. 9 Present Address: Department of Pediatrics and Pathology, Institute of Molecular Pediatric Sciences, The University of Chicago, 5841 South Maryland Ave., Rm. 314B, Chicago, IL 60637, USA. *Correspondence: [email protected] DOI 10.1016/j.cell.2007.04.036

SUMMARY electrophysiological activities. Disorders of the cardiac conduction system occur commonly, either in isolation The cardiac conduction system is an anatomi- or in the context of generalized heart disease, and often cally discrete segment of specialized myo- produce life-threatening arrhythmias. Medical treatments cardium that initiates and propagates electrical for conduction system disease are limited and primarily in- impulses to coordinate myocardial contraction. volve antiarrhythmic medications and pacemaker implan- To define the molecular composition of the tation. mouse ventricular conduction system we used The anatomic structures of the mammalian conduction system were first described 100 years ago (Tawara, 1906). microdissection and transcriptional profiling Atrial components of this complex structure include the by serial analysis of gene expression (SAGE). sinoatrial and atrioventricular nodes; these have been Conduction-system-specific expression for conserved throughout vertebrate development (reviewed Id2, a member of the Id gene family of transcrip- in Moorman et al., 1998). Ventricular components have tional repressors, was identified. Analyses of evolved more recently (Sedmera et al., 2003b) and in Id2-deficient mice demonstrated structural mammals include an atrioventricular bundle and the left and functional conduction system abnormali- and right bundle branches. The ventricular conduction ties, including left bundle branch block. A 1.2 kb system accounts for rapid spread of electrical impulses fragment of the Id2 promoter proved sufficient from the atrioventricular node at the atrial base to the ven- for cooperative regulation by Nkx2-5 and Tbx5 tricular apex, which is essential for apical contraction initi- in vitro and for conduction-system-specific ation. Coordinated cardiac contraction shifts from a linear pattern to an apex-initiated pattern between E11 and E14 gene expression in vivo. Furthermore, com- (Morley and Vaidya, 2001). The transition to a mature pound haploinsufficiency of Tbx5 and Nkx2-5 pattern of cardiac contraction is coincident with the differ- or Tbx5 and Id2 prevented embryonic specifica- entiation of myocytes in the ventricular conduction system tion of the ventricular conduction system. We lineage, including the onset of molecular conduction conclude that a molecular pathway including system marker expression (e.g., minK), implying estab- Tbx5, Nkx2-5, and Id2 coordinates specification lishment of a formal ventricular conduction system (Gour- of ventricular myocytes into the ventricular con- die et al., 1995; Cheng et al., 1999; Moorman et al., 1998; duction system lineage. Kondo et al., 2003). Whether the ontogeny of the atrial and ventricular conduction components requires the same INTRODUCTION molecular pathways or whether they are independently generated and physically coupled is unknown. The conduction system is a specialized structure within Tbx5 and Nkx2-5, two transcription factor genes the heart responsible for establishing and maintaining expressed broadly in the heart, have roles in the

Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. 1365 maintenance of conduction system structure and func- Presumptive left bundle branches—discrete, thin mem- tion. Dominant mutations in these genes cause congenital branes of tissue—were dissected intact from surrounding heart defects and conduction system abnormalities, most myocardium of the left ventricular septum of juvenile wild- notably atrioventricular block, in adult humans and mice type and minKlacZ/+ mouse hearts (Figures 1D and 1E). To (Basson et al., 1994; Li et al., 1997; Schott et al., 1998, confirm that these tissues were comprised of conduction Benson et al., 1999; Bruneau et al., 2001; Prall et al., system cells, isolated specimens from minKlacZ/+ mice 2002; Jay et al., 2004; Moskowitz et al., 2004; Pashmfor- were stained for b-galactosidase expression and DAPI. oush et al., 2004). Because electrophysiological defects b-galactosidase expression was observed in approxi- can occur in the absence of cardiac structural malforma- mately 90% of cells (Figure 1F), indicating a high degree tions (Basson et al., 1994; Benson et al., 1999; Schott of conduction system enrichment. et al., 1998, Pashmforoush et al., 2004), Tbx5 and Nkx2-5 are hypothesized to have roles in conduction system de- Transcriptional Profiling of the Ventricular Cardiac velopment independent of their roles in cardiac morpho- Conduction System genesis. Consistent with this model, Tbx5 and Nkx2-5 Left bundle branches from 60 19-day-old wild-type SvEv are transcribed at higher levels in the ventricular conduc- mice were pooled for SAGE library construction (Velcu- tion system than in surrounding myocardium (Thomas lescu et al., 1995). For comparison, a SAGE library was et al., 2001; Moskowitz et al., 2004; Harris et al., 2006). also constructed from the left ventricle of two 19-day- However, more precise understanding of the role of old SvEv mice. More than 60,000 SAGE tags were Tbx5, Nkx2-5, and other transcription factors in ventricular sequenced from each library, unique tags were annotated, conduction system development is lacking. and total tag counts between the libraries were Neither a molecular description of conduction system normalized. myocytes nor a definition of the features that distinguish To verify that starting material for the left bundle branch these from nonconduction myocytes has been described, library was enriched in conduction system cells, tags cor- in part due to the difficulty of isolating pure conduction responding to previously defined molecular markers of the system tissue, given its complex three-dimensional struc- conduction system were assessed (Figures 1G and 1H). ture and intimate association with nonconduction myo- Tags corresponding to the minK gene, expression of cytes. We developed and verified a ventricular conduction which guided microdissections, were specific for the con- system microdissection technique and defined the tran- duction system library: 11/60,000 tags were in the left bun- scriptome of the left bundle branch by serial analysis of dle branch library compared to 0/60,000 tags in the left gene expression (SAGE). The ventricular conduction sys- ventricular library (p < 0.001). Tags corresponding to a-3 tem shared RNAs expressed in both myocyte and neuron Na+,K(+)-ATPase and Anf (atrial natriuretic factor), genes transcriptomes. SAGE identified Id2, a member of the Id shown by in situ hybridization to have increased expres- family of transcriptional repressors, as having conduc- sion in the conduction system (Zeller et al., 1987; Zahler tion-system-specific expression. We characterized Id2 et al., 1992), were also significantly more abundant in expression and evaluated the consequences of Id2 defi- the left bundle branch library than in the left ventricle ciency on conduction system function. Analysis of the library (Figure 1G). We concluded that the left bundle Id2 promoter demonstrated that conduction-system- branch SAGE library was significantly enriched in tran- specific expression of Id2 is dependent on Nkx2-5 and scripts from conduction system cells. Tbx5. These data define a molecular pathway, including The left bundle branch conduction system transcrip- Nkx2-5, Tbx5, and Id2, that is required for ventricular con- tome was further evaluated by assessment of transcripts duction system development. with significantly different levels of expression from the left ventricle SAGE library (p < 0.02). Interrogation of RESULTS SAGE libraries identified 13 genes encoding proteins of the myocyte sarcomere complex (Figure 1H). Each myo- Isolation of Ventricular Conduction System Tissue cyte sarcomere gene was represented by lower tag We characterized postnatal expression of the conduction counts in the left bundle branch library compared to the system marker minK:lacZ (Kupershmidt et al., 1999) to de- left ventricle library (Figure 1H). fine maturation of the ventricular conduction system. The left bundle branch of adult mice (age > 12 weeks) was con- Id2 Expression in the Ventricular Cardiac solidated into thin discrete bands on the surface of the in- Conduction System terventricular septum; however, in juvenile mice (age 19 The left ventricular bundle SAGE library was informative days) this structure covered the entire basal portion of for identifying the expression of genes with previously the left ventricular septal surface (Figure 1A versus 1B). unknown roles in the conduction system. SAGE tags cor- Transverse sections of the interventricular septum re- responding to Id2 (Figure 1G), a member of a gene family vealed that left bundle branch cells, expressing nuclear encoding helix-loop-helix-containing transcriptional re- minK:lacZ, were tightly adherent to one another and pressors (Id1–Id4), were specifically represented in the segregated from underlying nonconduction myocytes left bundle branch library (6/60,000 tags) and were absent (Figure 1C). (0/60,000 tags) from the left ventricle library (p < 0.02).

1366 Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. Figure 1. Microdissection of Cardiac Conduction System Tissue (A) The adult mouse heart left bundle branch is a discrete structure (black arrowhead), marked by minK:lacZ expression (253). (B) The juvenile mouse heart left bundle branch (black arrowhead) is a broad structure, cover- ing the entire basal surface of the left ventricular septum (403). (C) A cross-section of the left side of the interventricular septal surface demonstrated tight association of the left bundle branch cells, expressing nuclear minK:lacZ (2003). (D) The broad juvenile left bundle branch could be discretely removed from the left ventricle, usually as a single intact sheet (black arrowhead; 403). (E) The microdissection generated thin sheets of tissue with no visible adherent myocardium (403). (F) Left bundle branch microdissected from minklacZ/+ hearts and subsequently costained with X-gal and DAPI demonstrated that microdissected tissue was predominantly composed of minK:lacZ-positive cells (2003). (G and H) SAGE libraries were generated from left bundle branches (LBB) and whole left ventricle tissues. Tags corresponding to MinK, Na+/K+ ATPase a3, and Anf, genes with known preferential conduction system expression, were signiﬁcantly enriched in the left bundle branch library (G). Id2, a gene with previously unknown conduction system expression, was also preferentially represented in the left bundle branch library (G). SAGE tags corresponding to sarcomere protein genes with signiﬁcantly (p < 0.02) different counts in the left bundle branch versus left ventricular libraries were uniformly higher in the left ventricular library (H).

Tags corresponding to other Id family members were not Id2 Is Required for Ventricular Cardiac Conduction differentially expressed (data not shown). System Function and Structure To confirm and characterize Id2 expression in the con- Id2 null (Id2/) mice (Yokota et al., 1999) were studied by duction system, in situ hybridization was performed for surface electrocardiography (ECG). The length of the QRS each Id gene. At E12.5, Id2 was expressed in the nascent interval, which corresponds to conduction time through atrioventricular bundle, which was demarcated by the ventricular conduction system, was measured in minKlacZ/+ expression (Figure 2A). Id2, but not Id1 or Id3, both newborn and 12-week-old mice. Telemetric electro- was expressed in the same region as minKlacZ/+ at this cardiogram recordings on adult mice were also used to developmental time (Figure 2C versus 2B and 2D). At confirm conduction intervals in the absence of anesthesia. E16.5, Id2 conduction system expression remained Both newborn and adult Id2/ mice displayed ventricular strong. minKlacZ/+ expression also marked the atrioven- conduction delay, with a widened QRS complex com- tricular bundle as well as the bundle branches at E16.5 pared to the wild-type controls (newborn p-value < (Figure 2E), and Id2, but not Id1 or Id3, was expressed 0.001; adult p-value = 0.01; Figures 3A–3C). Id2/ mice in this same region (Figure 2G versus 2F and 2H). Expres- also displayed a characteristic RsR0 pattern in lead I, sion of Id1, Id2, and Id3 was also detected in the aVL, and precordial lead V6 on surface six-lead ECGs, atrioventricular endocardial cushions at E12.5 (Figures indicative of a failure of left ventricular branch function, 2B–2D) and nascent atrioventricular valves at E16.5 (Fig- e.g., left bundle branch block (Figure 3C). ures 2F–2H), consistent with previous reports (Fraiden- Specific functional defects of the ventricular conduction raich et al., 2004). Id4 transcripts were not identified in system were also observed in adult Id2/ mice by in vivo the heart, either by SAGE or in situ hybridization (data electrophysiology. Intracardiac recordings showed a not shown). broad and split ventricular spike in all adult Id2/ mice

Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. 1367 Figure 2. Id2 Is Expressed in the Cardiac Conduction System (A–P) Cross-sections of embryonic mouse hearts at E12.5 (A–H) or E16.5 (I–P) analyzed for conduction system gene expression. MinK:lacZ expression marked the developing atrioventricular bundle at E12.5 ([A] and [E], blue arrowhead). In situ hybridization demonstrated Id2 expression in the atrioventricular bundle at E12.5 ([C] and [G], blue arrowhead), and in the atrioventricular bundle and bundle branches at E16.5 ([K] and [O], blue arrowhead). Id1, Id2, and Id3 are each expressed in endocardial cushions at E12.5 ([B–D] and [F–H], black arrowhead) and the developing valves at E16.5 ([J–L] and [N–P], black arrowhead). Neither Id1 ([B, F, J, and N]) or Id3 ([D, H, L, and P]), were expressed in the conduction system. Cardiac expression of Id4 was absent at either time point (data not shown). ([A–D] and [I–L], 403; [E–H] and [M–P], 1003). in contrast to the narrow ventricular spike seen on intra- abnormal morphology of the atrioventricular bundle and cardiac recordings of wild-type mice (Figure 3C). Together left bundle branch in Id2//minKlacZ/+ mice appeared with widened QRS complexes observed in Id2/ mice similar to that observed in adult mice with Tbx5 haploin- (Figure 3), this finding suggested abnormal intraventricular sufficiency (Tbx5+/; Figure 3F; Moskowitz et al., 2004). conduction within bundle branches. Spontaneous atrio- In each case, the atrioventricular bundle and left bundle ventricular bundle depolarizations, always present in branch were present but displayed reproducible pattern- wild-type mice, were also absent from four out of five ing abnormalities consistent with the observed functional Id2/ mice. abnormalities. The requirement of Id2 for ventricular conduction system structure was analyzed. Normally, a discrete atrioven- In Silico Search for Tbx5 and Nkx2-5 Target tricular bundle physically separates the atrioventricular Genes Identifies Id2 node from the left bundle branch, as is observed in The similar ventricular conduction system abnormalities minKlacZ/+ mice (Figure 3D, red arrowhead). However, a observed in Tbx5+/ and Id2/ mice suggested that discrete atrioventricular bundle was absent from Id2// Tbx5 and Id2 could be required in the same molecular minKlacZ/+ mice. Instead the left bundle branch was un- conduction system pathway. As Tbx5 and Nkx2-5 expres- characteristically broad and appeared to originate along sion is enriched in the conduction system and because the entire length of the ventricular crest (Figure 3E). The these transcription factors cooperatively regulate other

1368 Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. Figure 3. Id2(/) Mice Have Morphologic and Conduction Abnormalities (A) Conduction intervals from wild-type and Id2(/)mice. QRS intervals (measured in unanesthetized mice by surface ECGs) of adult and newborn Id2(/) mice were signiﬁcantly prolonged compared to wild-type mice. Means are ± SD with p-values determined from Stu- dent’s t test. (B) Representative tracings from wild-type mouse demonstrated normal QRS interval on surface ECG and intracardiac recordings. (C) Representative tracings from Id2(/) mouse demonstrated complete left bundle branch block with widened QRS complex and a characteristic RsR0 pattern in lead I and precordial lead V6 on surface ECG. The split morphology of the ventricular spike indicates intraventricular conduction delay. (D–F) MinK-lacZ expression marked the atrioventricular bundle and left bundle branch in 12-week-old wild-type/minKlacZ/+ (D), Id2(/)/ minKlacZ/+ (E), and Tbx5+//minKlacZ/+ (F) mouse hearts. In wild-type hearts, a discrete atrioventricular bundle (red arrowhead) separated the atrioventricular node (black arrow) from the left bundle branch (black arrowhead). In the Id2(/) and Tbx5+/ hearts, the left bundle branch was abnormally broad and origi- nated from the entire ventricular crest. In each case, the atrioventricular node (black arrow) and left bundle branch (black arrowhead) were present, but a discrete atrioventricular bundle was absent, and the left bundle branch appeared broad and immature ([D–F], 253).

genes (Bruneau et al., 2001, Hiroi et al., 2001), we hypoth- Tbx5 and Nkx2-5 Are Necessary and Sufficient esized that Tbx5 and Nkx2-5 might coregulate Id2 expres- for Conduction System-Specific Id2 Expression sion. We analyzed human-mouse sequence alignments of The ability of Tbx5 or Nkx2-5 to activate Id2 transcription the Id2 promoter for canonical Tbx5-binding sites was functionally assessed in a heterologous cell culture (GGTGTG, GTGACA) or Nkx2-5-binding sites (AAGTG). system. A 1052 bp fragment of the Id2 promoter, including Five Tbx5- and two Nkx2-5-binding sites were found three Tbx5 and one Nkx2-5 putative binding sites, was within 3 kb upstream and 1 kb downstream of the Id2 tran- cloned and inserted upstream of the firefly luciferase re- scriptional start site (Figure 4A). To evaluate the signifi- porter gene. CV-1 cells were cotransfected with the Id2- cance of multiple Tbx5- and Nkx2-5-binding sites within luciferase reporter construct and expression vectors for the Id2 promoter, we assessed the number of Tbx5- and Tbx5, Nkx2-5, both, or neither. Compared to empty Nkx2-5-binding sites found in 8,347 promoter regions vector, cotransfection with Nkx2-5 vector produced no in- (deduced by human and mouse genomic sequence align- crease in luciferase expression, and cotransfection with ments) in 5,717 genes from the OMGProm database (Pal- Tbx5 vector produced a 2-fold enhancement (Figure 4B). aniswamy et al., 2005). Numbers of conserved binding- In contrast, cotransfection with both Tbx5 and Nkx2-5 site motifs in these promoters were counted, giving four vectors generated robust cooperative transcriptional times as much weight to 6 bp motifs relative to 5 bp motifs, activation of the Id2 promoter (approximately 7-fold; and promoters were rank ordered. The Id2 promoter Figure 4B). ranked 10 out of 5717 for numbers of Tbx5-orNkx2-5- Tbx5 interactions with the putative binding sites in the binding sites, within the top 0.2% of all gene promoters Id2 promoter were analyzed by electrophoretic mobility searched. The rank order of promoters with Tbx5-or shift assays. Only the site located 921 bp 50 of the Id2 Nkx2-5-binding sites was also crossreferenced with transcriptional start site (GTGACA; blue, Figure 4A) dem- genes preferentially expressed in the conduction system onstrated specific Tbx5 binding (lane2). Binding was SAGE library. Id2 ranked second among 323 conduction competed by excess cold wild-type probe (lanes 3–5). system-enriched genes. Mutation of two conserved residues in the putative

Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. 1369 Figure 4. Id2 Expression Is Cooperatively Regulated by Nkx2-5 and Tbx5 (A) Id2 promoter regions with significant human/mouse sequence similarity are shown with canonical Nkx2-5- and Tbx5-binding sites, highlighted. The black horizontal bar denotes the Id2 promoter fragment used in Id2-luciferase reporter assays, including three Tbx5-binding sites (top row) and one Nkx2-5-binding site (bottom row). The Tbx5-binding site at position 921 (GTGACA; blue) demonstrated specific Tbx5 binding in the electro- mobility shift assay (Lane 1: Id2 oligo alone, Lane 2: Id2 oligo plus Tbx5 protein). Binding was competed by excess cold wild-type probe (Lane 3: 100-fold excess; Land 4: 200-fold excess; and Lane 5: 300-fold excess). Mutation of two conserved residues in the putative Tbx5-binding site (resulting in GCGAAA) eliminated the ability of the cold probe to abrogate Tbx5 protein/DNA interaction (Lane 6: 200-fold excess mutant oligo). (B) Cooperative activation of the Id2 promoter by Nkx2-5 and Tbx5. A 1052 bp fragment of the Id2 promoter was analyzed in transient transfection assays. The Id2-luciferase construct was transfected alone or with Tbx5 or Nkx2-5 expression vectors, singly and in combination. Luciferase activity promoted by the wild-type Id2 construct (blue) and Id2 construct with the mutated Tbx5-binding site at position 921 (pink) is shown. (C–F) The 1052 bp Id2 promoter fragment produced LacZ expression in transient transgenic mouse embryos. (C) Whole-mount Id2:LacZ cardiac expression at E14.5 illuminated the atrioventricular bundle (blue arrowhead) and bundle branches (black arrowheads). (D and E) Sagital sections of Min- K:lacZ (D) and Id2:lacZ (E) E14.5 embryos demonstrated LacZ expression specifically in the atrioventricular bundle (blue arrowhead) and bundle branches. Mutation of two conserved residues in the Tbx5-binding site at position 910 (resulting in GCGAAA) eliminated LacZ conduction system expression (F). ([C and F]: 253; [E and F]: 1003). (G–J) In situ hybridization of Id2 in wild-type (G and I) and Tbx5+//Nkx2-5+/ (H and J) E12.5 mouse hearts. Sagital sections demonstrated Id2 expression in the developing ventricular conduction system in the wild-type ([G], blue arrowhead) but not in the Tbx5+//Nkx2-5+/ heart. ([G and H]: 1003; [I and J]: 2003).

Tbx5-binding site (resulting in GCGAAA) eliminated the The Id2 promoter proved sufﬁcient for conduction sys- ability of the cold probe to abrogate Tbx5 protein/DNA tem gene expression in vivo. Transient transgenic mice interactions (lane 6). expressing the 1052 bp fragment of the Id2 promoter To determine if the Tbx5 binding at the 921 site affected fused to lacZ were generated. Two of four transient trans- cooperative transcriptional activation of the Id2 promoter genic Id2:lacZ embryos demonstrated b-galactosidase by Tbx5 and Nkx2-5, the mutated 921 site was introduced activity in the ventricular conduction system (Figures 4C into the Id2-luciferase reporter construct and cotrans- and 4E). In each case, cardiac transgene expression fected into CV-1 cells with Tbx5 and Nkx2-5 expression was selective for the ventricular conduction system and constructs, alone and in combination. Mutation of the included the atrioventricular bundles and bundle 921 site abolished the synergistic activation by Tbx5 and branches. Mutation of the Tbx5-binding site at position Nkx2-5 that was observed with the wild-type Id2-lucifer- 921 abolished conduction system transgene expression ase construct (Figure 4B, p < 107). (0/12 mutant transgenic embryos, Figure 4F), indicating

1370 Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. that conduction system expression was Tbx5 dependent. In contrast, extracardiac expression of the mutant Id2 transgene, e.g., mammary tissue (data not shown), was unaltered compared to the wild-type Id2 transgene. The roles of Tbx5 or Nkx2-5 on endogenous Id2 expression in the conduction system were assessed by in situ hybridization in the hearts from Tbx5+/, Nkx2-5+/ and compound Tbx5+//Nkx2-5+/ embryos. Id2 expression was found at the tip of the interventricular septum in the region of the developing ventricular conduction system at E12.5 in wild-type, Tbx5+/, and Nkx2-5+/ hearts (Fig- ures 4G and 4I and data not shown) but was absent from compound Tbx5+//Nkx2-5+/ hearts (Figures 4H and 4J). We concluded that Tbx5 and Nkx2-5 cooperatively activate Id2 transcription in the developing ventricular conduction system of the mouse.

Nkx2-5 and Tbx5 Are Required for Ventricular Conduction System Specification Hypothesizing that Tbx5 and Nkx2-5 affected development throughout the ventricular conduction system, we analyzed compound Tbx5+//Nkx2-5+/ mice for functional and/or structural deficits. Ventricular cardiac conduction system function was assessed in nonanesthetized newborn mice by measuring conduction intervals from electrocardiograms. The QRS intervals (Figure 5A) of Tbx5+//Nkx2-5+/ mice were significantly prolonged compared to wild-type and single heterozygote Tbx5+/ and Nkx2-5+/ mice (wild-type:Tbx5+//Nkx2-5+/,p< 0.001; Tbx5+/:Tbx5+//Nkx2-5+/, p < 0.001 Nkx2-5+/: Tbx5+//Nkx2-5+/, p < 0.001). The morphology of the ventricular conduction system was studied in Tbx5+//Nkx2-5+//minKlacZ/+ mice. The atrioventricular bundle and proximal left and right bundle branches were evident in minKlacZ/+ (n = 15; Figures 5C and 5D), Tbx5+//minKlacZ/+ (n = 14; Figures 5E and 5F), and Nkx2-5+//minKlacZ/+ mice (n = 11; Figures 5G and 5H). However, in Tbx5+//Nkx2-5+//minKlacZ/+ mice,

The QRS intervals, indicative of conduction through the ventricular conduction system, were signiﬁcantly longer in compound Tbx5+// Nkx2-5+/ and Tbx5+//Id2+/ mice than in wild-type or single Tbx5+/, Nkx2-5+/,orId2+/ mutant mice. Numbers in parentheses indicate the number of animals tested for each genotype. (B) Model depicting locations (yellow) of cardiac sagital sections (gray) used to analyze the atrioventricular node (posterior) and atrioventricular bundle and proximal bundle branches (anterior). (C–L) MinK-lacZ expression marked the atrioventricular node (C, E, G, I, and K) or atrioventricular bundle and proximal bundle branches (D, F, H, J, and L) in sagital sections from newborn wild-type/minKlacZ/+ (C and D), Nkx2-5+//minKlacZ/+ (E and F), Tbx5+//minKlacZ/+ (G and H), and Tbx5+//Nkx2-5+//minKlacZ/+ (I and J) and Tbx5+//Id2+// minKlacZ/+ (K and L) mouse hearts. The atrioventricular node (blue arrow in [C]) was well formed in each case. Well-formed atrioventricular bundles ([D], black arrow) and proximal bundle branches ([D], black Figure 5. Conduction System Defects in Tbx5+//Nkx2-5+/ arrowheads) were present in wild-type/minKlacZ/+ (D), Nkx2-5+// Mice minKlacZ/+ (F), and Tbx5+//minKlacZ/+ (H) hearts, but were absent in (A) Cardiac conduction intervals in newborn wild-type, Tbx5+/, compound Tbx5+//Nkx2-5+//minKlacZ/+ (J) and Tbx5+//Id2+// Nkx2-5+/, and Tbx5+//Nkx2-5+/, Id2+/, and Tbx5+//Id2+/ mice. minKlacZ/+ (L) mutant hearts. ([C–L], 1003).

Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. 1371 lacZ expression was completely absent from the atrioven- was completely absent from the interventricular septum tricular bundle and bundle branch regions (n = 9; of Tbx5+//Id2+//minKlacZ/+ mice (n = 6; Figure 5L). We Figure 5J). A well-formed atrioventricular node was uni- concluded that haploinsufficiency of either Tbx5 and formly present in every genotype (Figures 5C, 5E, 5G, Nkx2-5 or Tbx5 and Id2 abrogated ventricular conduction and 5I). Tbx5+//Nkx2-5+//minKlacZ/+ mice were analyzed system development. for the expression of two additional ventricular conduction system markers, Cx40 and CCS:lacZ (Delorme et al., Conduction System Cycle Defect in Tbx5- 1995; Rentschler et al., 2001). Cx40 and CCS:lacZ expres- and Nkx2-5-Deficient Mice sion were absent from the region of the atrioventricular To further test the hypothesis that genetic interactions bundle and the bundle branches in Tbx5+//Nkx2-5+/ among Tbx5, Nkx2-5, and Id2 were critical for differentia- mice (data not shown), suggesting a generalized failure tion of ventricular conduction system myocytes, we uti- of ventricular conduction system development. lized cell-cycle exit as an early indicator of conduction A time course was performed to determine when the system lineage specification. Myocytes of the ventricular ventricular conduction system defects appeared in conduction system exit the cell cycle between E12 and Tbx5+//Nkx2-5+/ mice. The atrioventricular node is ob- E14, unlike nonconduction embryonic ventricular myo- served as early as E10.5 and the developing ventricular cytes that show continued cell-cycle progression (Cheng conduction system at E12.5, as marked by minK:lacZ ex- et al., 1999; Sedmera et al. 2003a). We evaluated conduc- pression (Kondo et al., 2003). At E10.5 and E12.5, all mice tion system cell-cycle exit in wild-type and mutant mice had comparable minK expression in the developing atrio- using ventricular conduction system birthdating. Wild- ventricular node, implying that haploinsufficiency of Tbx5, type hearts exposed to BrdU from E11 to E16, beginning Nkx2-5, or both did not impair specification or differentia- prior to ventricular conduction system specification, had tion of the atrioventricular node (Figures S1A–S1H). How- uniform BrdU uptake throughout ventricular myocardium ever, at E12.5, minK expression was completely absent indicative of uniform cell cycling (Figure 6A). However, from the ventricles of compound heterozygote Tbx5+// wild-type hearts exposed to BrdU from E13 to E16, a pe- Nkx2-5+//minKlacZ/+ mice (Figure S1L). Ventricular ex- riod coincident with ventricular conduction system speci- pression of minK was absent from Tbx5+//Nkx2-5+// fication, demonstrated localized absence of BrdU uptake minKlacZ/+ hearts analyzed at E14.5 and E16.5 as well (cell-cycle exit) in the region of the ventricular conduction (data not shown). system (Figure 6B). To confirm that the nondividing cells We next considered whether absence of molecular con- were conduction system, BrdU birthdating was performed duction system markers from Tbx5+//Nkx2-5+/ ventri- in minKlacZ/+ mice. When exposed to BrdU from E13 to cles reflected a loss of conduction system precursor cells E16, hearts from minKlacZ/+ mice showed absence of or a failure to specify conduction system from myocytes BrdU staining in cells with minK expression, and, con- present in the embryo. Presumptive ventricular conduc- versely, BrdU staining in cells without minK expression tion system cells express Nkx2-5 at a higher level than sur- (Figure 6C). rounding myocardium (Thomas et al., 2001). We moni- Ventricular conduction system birthdating demon- tored Nkx2-5 expression via b-galactosidase activity strated a conduction system specification defect in mice from an Nkx2-5lacZ allele (Nkx2-5 protein sequences haploinsufficient for both Tbx5 and Nkx2-5. Tbx5+// deleted; Jay et al., 2004). b–galactosidase expression in Nkx2-5+//minKlacZ/+ mice exposed to BrdU at E13 embryonic ventricles from Nkx2-5lacZ/+ and Tbx5del/+/ showed uniform cell-cycle entry throughout embryonic Nkx2-5lacZ/+ mice was compared at E13.5 and E16.5 (Fig- ventricle, including the crest of the interventricular sep- ure S2). In both genotypes, higher levels of Nkx2-5 expres- tum. Myocytes in this region could not be distinguished sion were observed in presumptive conduction system from those located elsewhere in the ventricles, suggesting precursors. We concluded that conduction system a failure of cell-cycle exit of the conduction system anlage precursors were present but failed to adopt a conduction (Figure 6F). In contrast, haploinsufficiency of either Tbx5 or system fate in mice haploinsufficient for both Tbx5 and Nkx2-5 alone did not affect cell-cycle exit of the develop- Nkx2-5. ing conduction system (Figures 6D and 6E). We concluded that the activity of Tbx5 and Nkx2-5 in specified ventricular Id2 and Tbx5 Cooperate during Conduction myocardial cells resulted in cell-cycle exit and adoption of System Development a conduction system fate. To further explore genetic interactions between Tbx5 and Id2, compound mutant Tbx5+//Id2+/ mice were evalu- DISCUSSION ated. The QRS interval of Tbx5+//Id2+/ mice was se- verely prolonged compared to wild-type, Tbx5+/, and We describe the first conduction system transcriptome Id2+/ mice (Figure 5A) but was comparable to Tbx5+// and report a critical transcriptional network required for Nkx2-5+/ mice, demonstrating a significant functional de- development of the cardiac electrophysiological system. fect of the ventricular conduction system by surface ECG. Three lines of evidence suggested that specification of Ventricular conduction system morphology was assessed the ventricular cardiac conduction system was dependent in Tbx5+//Id2+//minKlacZ/+ mice. MinK:lacZ expression on combinatorial activity of Nkx2-5 and Tbx5. First, rapid

1372 Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. Figure 6. Cell-Cycle Conduction System Defect in Tbx5+//Nkx2-5+/ Mice (A–F) E16.5 heart sections were labeled with daily BrdU exposure beginning at E11 (A) or E13 (B–F). Sagital sections from wild-type hearts with BrdU introduced at E11 demonstrated uniform BrdU uptake (A), indicative of uniform myocardial cell division. Sagital sections from wild-type heart with BrdU beginning at E13 demonstrated a discrete region without BrdU uptake at the tip of the interventricular septum ([B], blue arrow), consistent with cell- cycle exit of the developing ventricular conduction system. Sagital sections from wild-type/ minKlacZ/+ heart with BrdU introduced at E13 demonstrated a discrete region without BrdU uptake, which colocalized with minK expression ([C], top), conﬁrming cell-cycle exit by conduction system cells. Hematoxylin and eosin staining of an adjacent section (5 microns) demonstrated uniform nuclear density in the conduction and nonconduction myocardium. Sagital sections from Nkx2-5+/ (D) and Tbx5+/ (E) hearts with BrdU introduced at E13 also demonstrated cell-cycle exit of minK- expressing cells. Sagital sections from a Tbx5+//Nkx2-5+//minKlacZ/+ heart with BrdU introduced at E13 showed uniform BrdU uptake (F), demonstrating failure of cell-cycle exit. (Upper panels, 1003; lower panels, 2003).

propagation of electrical impulses from the atrioventricu- and Nkx2-5 of two other genes, Anf and Cx40 (connexin- lar node through the ventricles, an essential physiologic 40), has been reported (Bruneau et al., 2001, Hiroi et al., requirement for apex-initiated ventricular contraction, 2001). Interestingly, both genes were expressed at higher was absent in newborn Tbx5+//Nkx2-5+/ mice. Second, levels in the ventricular conduction system than in non- expression of molecular cardiac conduction system conduction ventricular myocardium (Zeller et al., 1987; markers was absent in compound mutant embryonic Delorme et al., 1995), consistent with the model that ven- (E12.5) and neonatal mice (e.g., minK, Figures 5 and S1). tricular conduction system gene expression is established Because cardiac morphology was identical at E12.5 in by the combinatorial activity of Tbx5 and Nkx2-5. each genotypic class tested, differences in conduction system differentiation could not be ascribed to cardiac The Role of Id2 in the Ventricular Conduction System morphologic abnormalities. Finally, exit from the cell cy- SAGE results indicated that conduction system cells are cle, a cellular marker for embryonic ventricular conduction less differentiated toward cardiac muscle than noncon- system specification, did not occur in Tbx5+//Nkx2-5+/ duction cardiomyocytes. Fewer tags were found in the embryos. Taken together, these results indicated that left bundle branch library for all 13 sarcomere protein specification of the ventricular conduction system failed genes, indicating that these genes are less abundantly ex- in mice haploinsufficient for both Nkx2-5 and Tbx5. pressed in the conduction system than in the left ventricle. Id2, a gene identified by SAGE as having selective ven- These data corroborated previous morphologic studies tricular conduction system expression, is a downstream that identified a poorly developed sarcomere apparatus target of Tbx5 and Nkx2-5. The Id2 promoter has func- in the conduction system (Lev and Thaemert, 1973). We tional binding sites for Tbx5, and cell culture studies and hypothesized that Id2 may help distinguish conduction in vivo analyses (Figure 4) demonstrated that Nkx2-5 and from nonconduction cardiomyocytes by limiting differenti- Tbx5 are necessary for Id2 expression in the specialized ation of conduction system cells into contractile myo- conduction system cells. Cooperative regulation by Tbx5 cytes. Id2 blocks the myogenic activity of MyoD, E12/E47,

Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. 1373 Figure 7. A Molecular Pathway for Con- duction System Development (A) Schematic of Tbx5 expression, Nkx2-5 expression, and their overlap. Tbx5 expression (shown in yellow) is high in the atria and endo- cardium but highest in the ventricular conduction system (Bruneau et al., 1999; Moskowitz et al., 2004). Nkx2-5 expression (shown in green) is present throughout the myocardium, but is higher in the ventricle than the atrium and is highest in the ventricular conduction system (Thomas et al., 2001). The overlap of Tbx5 and Nkx2-5 expression (shown in blue) is highest in the ventricular conduction system. We suggest that this overlap marks the domain of specialized ventricular conduction system gene expression. (B) Model for speciﬁcation of ventricular conduction system fate. High levels of Tbx5 and Nkx2-5 expression establish a regionally re- stricted program of ventricular conduction system gene expression. Tbx5 and Nkx2-5 act cooperatively on the promoters of conduction-system-expressed genes, including the Id2 promoter, with Tbx5-binding sites marked yellow and Nkx2-5-binding sites marked green, and, similarly, on the connexin-40 (Cx40) and atrial natriuretic factor (Anf) promoters, with schematic Tbx5- and Nkx2-5-binding sites. This program of conduction system gene expression recruits cells into the conduction system lineage, including exit from the cell cycle, adoption of functional fast conduction, and expression of conduction system markers.

and other myogenic basic helix-loop-helix (bHLH) proteins Specification of the Ventricular Conduction System (Liu et al., 2002), and ectopic Id gene expression inhibits by Tbx5, Nkx2-5, and Id2 differentiation of P19 cells into cardiomyocytes and We propose a model for ventricular conduction system decreases myosin heavy-chain expression in vitro (Ding specification predicated on the localized coordinated et al., 2006). Furthermore, ectopic expression of Id2 in activity of Tbx5 and Nkx2-5 (Figure 7A). The overlap of ectodermal precursors promotes differentiation toward expression of Tbx5 and Nkx2-5 within the ventricles is sig- a neural crest rather than epidermal lineage (Martinsen nificantly higher in presumptive conduction system cells and Bronner-Fraser, 1998). We suggest that in cardiac than in surrounding myocardium (Thomas et al., 2001; cells fated to become the ventricular conduction system, Moskowitz et al., 2004). Within this region, high levels of Id2 shifted the balance from promuscle towards proneural Tbx5 and Nkx2-5 expression establish a regionally re- differentiation by inhibiting cardiac muscle gene expres- stricted program of ventricular conduction system gene sion, including genes encoding sarcomere proteins and expression. Specifically, promoters directly sensitive to promoting gene expression appropriate for electrophysio- the coordinated activity of Tbx5 and Nkx2-5, including logic activities. Id2, are activated. Id2 and presumably others transcription factors inhibit sarcomere differentiation and promote Modular Development of the Cardiac conduction system differentiation with cell-cycle exit Conduction System in the ventricular conduction system (Figure 7B). Taken Our analyses of mutant mice indicated that Tbx5 and together, Nkx2-5, Tbx5, and Id2 define a critical transcrip- Nkx2-5 are cooperatively required for specification of the tional network for the differentiation of ventricular myo- ventricular conduction system, but not of the atrioventric- cytes into specialized cells of conduction system. Delinea- ular node. The atrioventricular node in Tbx5+//Nkx2-5+/ tion of the genes regulated by these transcription factors mice was structurally intact (based on minK expression) will promote future insights into the molecular mecha- and demonstrated appropriate electrophysiological prop- nisms by which these specialized cells function in cardiac erties. While these studies do not negate the potential in- electrophysiology. volvement of Tbx5 and Nkx2-5 in atrial conduction system components, our data suggested that these transcription factors are particularly critical for specifying the ventricular EXPERIMENTAL PROCEDURES electrophysiological system. This conclusion implies that Animal Handling the conduction system develops as modular units with All protocols conformed to the Association for the Assessment and subsequent joining of the atrial and ventricular compo- Accreditation of Laboratory Animal Care and the Harvard Medical nents. School Animal Care and Use Committee.

1374 Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. Serial Analysis of Gene Expression Cell Culture and Transfection SAGE libraries were constructed from sixty pooled microdissected left CV-1 cells (ATCC no. CCL-70) were maintained in Dulbecco modified bundle branches and three whole left ventricles by the micro sage Eagle Medium with 1% penicillin/streptomycin and 10% fetal calf se- technique (St Croix et al., 2000). The SAGE 3.0.1 program (courtesy rum. Plates were incubated at 37 in the presence of 95% O2 and of Victor Velculescu and Ken Kinzler, Johns Hopkins University School 5% CO2, trypsinized 24 hr before transfection, and plated at a density of Medicine, Baltimore, Maryland, United States) was used to extract of 2 3 105 cells/well in 6-well plates. Cells were transfected with SAGE tags and eliminate duplicate ditags. Identity of SAGE tags was Lipofectamine 2000 (Invitrogen) reagents in serum- and antibiotic- obtained from the National Center for Biotechnology Information free DMEM as per the manufacturer’s instructions. After 12 hr, media (NCBI) ‘‘reliable’’ tag map set for UniGene. Gene identification was was replaced with serum-containing DMEM for an additional 36 hr. made blinded to tag counts. Cotransfections included 1.45 ug reporter, 0.5 ug of Nkx2-5 or Tbx5 expression vectors, and 0.05 ug Renilla luciferase for transfection efficiency. The reporter was generated using pGL4 luciferase (Promega). Gene Expression Studies Firefly luciferase activity, normalized for transfection efficiency, was Hearts from mice with the minKlacZ/+ allele were stained for b-galacto- measured after 48 hr. Reporter activity was normalized to Renilla sidase activity as previously described (Moskowitz et al., 2004). Mouse luciferase. Data are presented as mean ± standard deviation of three heart in situ hybridizations were performed as previously described experiments in duplicate. (Moskowitz et al., 2004). Id1, Id2, and Id3 probe templates were made from plasmid accession numbers BE945568, AI843393, and Mouse In Vivo Transgenic Reporter Assay AI839283, respectively, from the BMAP library (Bonaldo et al., 2004). Using a Gateway-compatible reporter vector (Gateway-HSP68-LacZ) the Id2 promoter fragment was transferred into the destination reporter ECG vector. A NotI restriction fragment was purified from the vector back- Twelve-lead electrocardiograms (ECG) were obtained using subcuta- bone, diluted to a DNA concentration of 2.5 to 5 ng/ml, and used for neous electrodes at the four limbs. The ECG channels were filtered be- pronuclear injections of CD1 embryos. Embryos were harvested at tween 5 and 400 Hz, and the analog signal was digitized with 12-bit E14.5 and dissected in cold PBS, followed by staining in whole mount precision at a sampling rate of 2 kHz. Adult mice were anesthetized for b-galactosidase activity. with pentobarbital (0.033 mg/kg intraperitoneal) to obtain multilead ECG recordings. Unanesthetized neonatal mice underwent ECG re- Supplemental Data cordings at postpartum day 2, and heart rates were standardized at Supplemental Data include two figures and can be found with this 460 beats per minute (bpm). article online at http://www.cell.com/cgi/content/full/129/7/1365/ DC1/.

Ambulatory Electrocardiogram Telemetry ACKNOWLEDGMENTS Adult mice (five Id2/ and five wild-type) underwent continuous telemetric ECG recordings obtained by implanted Holter devices. Radio- The authors thank Ivy Aneas, Vickas Patel, and Tharinda Rajapaksha frequency transmitters (DataSciences International, St. Paul, Minne- for excellent experimental assistance. This work was supported by sota, USA) were implanted into a subcutaneous pocket with leads grants from the Howard Hughes Medical Institute (C.E.S. and secured under the upper right and left limb in order to record lead I I.P.G.M.) and the National Institutes of Health (I.P.G.M., C.E.S., and ECG as previously described (Berul et al., 1996). After a 48 hr recovery J.G.S.). period, ECGs were recorded for one hour in awake, resting mice during daytime. Received: May 30, 2006 Revised: December 15, 2006 Electrophysiology Studies Accepted: April 10, 2007 In vivo EPS were performed in eight mice (four Id2+/ and four wild- Published: June 28, 2007 type) as previously described (Berul et al., 1996). Standard pacing protocols were used to assess atrial and ventricular conduction and re- REFERENCES fractoriness periods. Recording of a triphasic His-bundle electrogram was accomplished using simultaneous multielectrodes and persistent Basson, C.T., Cowley, G.S., Solomon, S.D., Weissman, B., Poznanski, catheter manipulation. Electrophysiologic parameters were measured A.K., Traill, T.A., Seidman, J.G., and Seidman, C.E. (1994). The clinical at baseline, and no drugs were administered during the study. All base- and genetic spectrum of the Holt-Oram syndrome (Heart-Hand Syn- line electrocardiographic and telemetry measurements were made for drome). New Engl. J. Med. 330, 885–891. three consecutive cardiac cycles by two experienced observers Benson, D.W., Silberbach, G.M., Kavanaugh-McHugh, A., Cottrill, C., blinded to mouse genotype. There was no significant interobserver Zhang, Y., Riggs, S., Smalls, O., Johnson, M.C., Watson, M.S., Seid- variability. man, J.G., et al. (1999). Mutations in the cardiac transcription factor NKX2.5 affect diverse cardiac developmental pathways. J. Clin. Invest. 104, 1567–1573. Electrophoretic Mobility Shift Assays Berul, C.I., Aronovitz, M.J., Wang, P.J., and Mendelsohn, M.E. (1996). Recombinant Tbx5 protein was incubated with 32P labeled double- In vivo cardiac electrophysiology studies in the mouse. Circulation 94, stranded oligonucleotide probes containing putative Tbx5-binding 2641–2648. sites from the Id2 promoter. Electrophoretic mobility shift assay (EMSA) reaction mixtures included 25 mM Hepes, 100 mM KCl, Bonaldo, M.F., Bair, T.B., Scheetz, T.E., Snir, E., Akabogu, I., Bair, J.L., 0.1% NP-40 (v/v), 1 mM DTT, 5% glycerol, and 50–100ng of poly(dIdC) Berger, B., Crouch, K., Davis, A., Eyestone, M.E., et al. (2004). 1274 as a nonspecific competitor. After a 10 min room-temperature incuba- full-open reading frames of transcripts expressed in the developing tion, 0.3 ng of radiolabeled probe was added, and the reaction was mouse nervous system. Genome Res. 10B, 2053–2063. incubated for an additional 20 min. When included, 100- to 300-fold Bruneau, B.G., Logan, M., Davis, N., Levi, T., Tabin, C.J., Seidman, excess cold wild-type or mutated probe was added to the first incuba- J.G., and Seidman, C.E. (1999). Chamber-specific cardiac expression tion. Protein-DNA complexes were separated on a 6% nondenaturing of Tbx5 and heart defects in Holt-Oram syndrome. Dev. Biol. 211, 100– polyacrylamide gel at 25C. 108.

Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc. 1375 Bruneau, B.G., Nemer, G., Schmitt, J.P., Charron, F., Robitaille, L., Morley, G.E., and Vaidya, D. (2001). Understanding conduction of Caron, S., Conner, D.A., Gessler, M., Nemer, M., Seidman, C.E., and electrical impulses in the mouse heart using high-resolution video im- Seidman, J.G. (2001). A murine model of Holt-Oram syndrome defines aging technology. Microsc. Res. Tech. 52, 241–250. roles of the T-box transcription factor Tbx5 in cardiogenesis and dis- Moskowitz, I.P., Pizard, A., Patel, V.V., Bruneau, B.G., Kim, J.B., ease. Cell 106, 709–721. Kupershmidt, S., Roden, D., Berul, C.I., Seidman, C.E., and Seidman, Cheng, G., Litchenberg, W.H., Cole, G.J., Mikawa, T., Thompson, R.P., J.G. (2004). The T-Box transcription factor Tbx5 is required for the and Gourdie, R.G. (1999). Development of the cardiac conduction sys- patterning and maturation of the murine cardiac conduction system. tem involves recruitment within a multipotent cardiomyogenic lineage. Development 131, 4107–4116. Development 126, 5041–5049. Palaniswamy, S.K., Jin, V.X., Sun, H., and Davuluri, R.V. (2005). OMG- Delorme, B., Dahl, E., Jarry-Guichard, T., Marics, I., Briand, J.P., Prom: a database of orthologous mammalian gene promoters. Bioin- Willecke, K., Gros, D., and Theveniau-Ruissy, M. (1995). Developmen- formatics 6, 835–836. tal regulation of connexin 40 gene expression in mouse heart corre- Pashmforoush, M., Lu, J.T., Chen, H., Amand, T.S., Kondo, R., Prader- lates with the differentiation of the conduction system. Dev. Dyn. vand, S., Evans, S.M., Clark, B., Feramisco, J.R., Giles, W., et al. 204, 358–371. (2004). Nkx2-5 pathways and congenital heart disease; loss of ventric- Ding, B., Liu, C.J., Huang, Y., Yu, J., Kong, W., and Lengyel, P. (2006). ular myocyte lineage specification leads to progressive cardiomyopa- p204 protein overcomes the inhibition of the differentiation of P19 thy and complete heart block. Cell 117, 373–386. murine embryonal carcinoma cells to beating cardiac myocytes by Id Prall, O.W., Elliott, D.A., and Harvey, R.P. (2002). Developmental par- proteins. J. Biol. Chem. 281, 14893–14906. adigms in heart disease: insights from tinman. Ann. Med. 34, 148–156. Fraidenraich, D., Stillwell, E., Romero, E., Wilkes, D., Manova, K., Rentschler, S., Vaidya, D.M., Tamaddon, H., Degenhardt, K., Sassoon, Basson, C.T., and Benezra, R. (2004). Rescue of cardiac defects in D., Morley, G.E., Jalife, J., and Fishman, G.I. (2001). Visualization and id knockout embryos by injection of embryonic stem cells. Science functional characterization of the developing murine cardiac conduc- 306, 247–252. tion system. Development 128, 1785–1792. Gourdie, R.G., Mima, T., Thompson, R.P., and Mikawa, T. (1995). Ter- Schott, J.J., Benson, D.W., Basson, C.T., Pease, W., Silberbach, G.M., minal diversification of the myocyte lineage generates Purkinje fibers Moak, J.P., Maron, B.J., Seidman, C.E., and Seidman, J.G. (1998). of the cardiac conduction system. Development 121, 1423–1431. Congenital heart disease caused by mutations in the transcription fac- Harris, B.S., Spruill, L., Edmonson, A.M., Rackley, M.S., Benson, D.W., tor NKX2-5. Science 281, 108–111. O’Brien, T.X., and Gourdie, R.G. (2006). Differentiation of cardiac Purkinje fibers requires precise spatiotemporal regulation of Nkx2–5 Sedmera, D., Reckova, M., DeAlmeida, A., Coppen, S.R., Kubalak, expression. Dev. Dyn. 235, 38–49. S.W., Gourdie, R.G., and Thompson, R.P. (2003a). Spatiotemporal pattern of commitment to slowed proliferation in the embryonic mouse Hiroi, Y., Kudoh, S., Monzen, K., Ikeda, Y., Yazaki, Y., Nagai, R., and heart indicates progressive differentiation of the cardiac conduction Komuro, I. (2001). Tbx5 associates with Nkx2-5 and synergistically system. Anat. Rec. 274, 773–777. promotes cardiomyocyte differentiation. Nat. Genet. 28, 276–280. Sedmera, D., Reckova, M., deAlmeida, A., Sedmerova, M., Biermann, Jay, P.Y., Harris, B.S., Maguire, C.T., Buerger, A., Wakimoto, H., M., Volejnik, J., Sarre, A., Raddatz, E., McCarthy, R.A., Gourdie, R.G., Tanaka, M., Kupershmidt, S., Roden, D.M., Schultheiss, T.M., O’Brien, and Thompson, R.P. (2003b). Functional and morphological evidence T.X., et al. (2004). Nkx2-5 mutation causes anatomic hypoplasia of the for a ventricular conduction system in zebrafish and Xenopus hearts. cardiac conduction system. J. Clin. Invest. 113, 1130–1137. Am. J. Physiol. Heart Circ. Physiol. 284, H1152–H1160. Kondo, R.P., Anderson, R.H., Kupershmidt, S., Roden, D.M., and St Croix, B., Rago, C., Velculescu, V., Traverso, G., Romans, K.E., Evans, S.M. (2003). Development of the cardiac conduction system Montgomery, E., Lal, A., Riggins, G.J., Lengauer, C., Vogelstein, B., as delineated by minK-lacZ. J. Cardiovasc. Electrophysiol. 14, 383– and Kinzler, K.W. (2000). Genes expressed in human tumor endothe- 391. lium. Science 289, 1197–1202. Kupershmidt, S., Yang, T., Anderson, M.E., Wessels, A., Niswender, K.D., Magnuson, M.A., and Roden, D.M. (1999). Replacement by ho- Tawara, S. (1906). Das Reizleitungssystem des Sa¨ ugetierherzen (Jena, mologous recombination of the minK gene with lacZ reveals restriction Germany: Verlag Von Gustav Fischer). of minK expression to the mouse cardiac conduction system. Circ. Thomas, P.S., Kasahara, H., Edmonson, A.M., Izumo, S., Yacoub, Res. 84, 146–152. M.H., Barton, P.J., and Gourdie, R.G. (2001). Elevated expression of Lev, M., and Thaemert, J.C. (1973). The conduction system of the Nkx-2.5 in developing myocardial conduction cells. Anat. Rec. 263, mouse heart. Acta Anat. (Basel) 85, 342–352. 307–313. Li, Q.Y., Newbury-Ecob, R.A., Terrett, J.A., Wilson, D.I., Curtis, A.R., Velculescu, V.E., Zhang, L., Vogelstein, B., and Kinzler, K.W. (1995). Yi, C.H., Gebuhr, T., Bullen, P.J., Robson, S.C., Strachan, T., et al. Serial analysis of gene expression. Science 270, 484–487. (1997). Holt-Oram syndrome is caused by mutations in TBX5, a mem- Yokota, Y., Mansouri, A., Mori, S., Sugawara, S., Adachi, S., Nishi- ber of the Brachyury (T) gene family. Nat. Genet. 15, 21–29. kawa, S.-I., and Gruss, P. (1999). Development of peripheral lymphoid Liu, C.J., Ding, B., Wang, H., and Lengyel, P. (2002). The MyoD-induc- organs and natural killer cells depends on the helix-loop-helix inhibitor ible p204 protein overcomes the inhibition of myoblast differentiation Id2. Nature 397, 702–706. by Id proteins. Mol. Cell. Biol. 22, 2893–2905. Zahler, R., Brines, M., Kashgarian, M., Benz, E.J., Jr., and Gilmore- Martinsen, B.J., and Bronner-Fraser, M. (1998). Neural crest specifica- Hebert, M. (1992). The cardiac conduction system in the rat expresses tion regulated by the helix-loop-helix repressor Id2. Science 281, 988– the alpha 2 and alpha 3 isoforms of the Na+,K(+)-ATPase. Proc. Natl. 991. Acad. Sci. USA 89, 99–103. Moorman, A.F., de Jong, F., Denyn, M.M., and Lamers, W.H. (1998). Zeller, R., Bloch, K.D., Williams, B.S., Arceci, R.J., and Seidman, C.E. Development of the cardiac conduction system. Circ. Res. 82, 629– (1987). Localized expression of the atrial natriuretic factor gene during 644. cardiac embryogenesis. Genes Dev. 7, 693–698.

1376 Cell 129, 1365–1376, June 29, 2007 ª2007 Elsevier Inc.