Pitx2 prevents susceptibility to atrial arrhythmias by inhibiting left-sided pacemaker specification

Pitx2c Is Expressed in the Postnatal Left Atrium and Pulmonary Vein.

To determine if Pitx2 is expressed in the postnatal heart, we generated a Pitx2 LacZ knockin allele, Pitx2^LacZ, to follow Pitx2 expression efficiently (Methods). LacZ staining in postnatal day 3 (P3) pups indicated that Pitx2 was expressed in the left atrium and pulmonary veins (Fig. 1 A–C). Sections confirmed that Pitx2^LacZ directed LacZ activity in the left atrium and pulmonary veins (Fig. 1 D–G). Pitx2 expression also was detected in the right ventricle, although this expression was weak (Fig. 1B). By P42, Pitx2 expression was detected in the left atrial myocardium, although at diminished levels (Fig. 1 H–J). At 1 year, Pitx2 expression was found only in rare left atrial myocardial cells (Fig. 1 K–N). To gain insight into postnatal Pitx2 isoform expression, we investigated LacZ expression in a transgenic line, Pitx2c^tg3K, in which LacZ is directed by Pitx2c regulatory elements (16). Because this Pitx2c transgene is influenced by copy number and position effect, this experiment provided only qualitative information about Pitx2 isoform expression in left atrium. In 3-month-old Pitx2c^tg3K transgenic mice, we found LacZ activity in right ventricle and left atrium, supporting the notion that Pitx2c was expressed in the postnatal left atrium and right ventricle (Fig. S1 A–D). RT-PCR experiments also indicated that Pitx2c was expressed in left atrium (Fig. 1O). Quantitative RT-PCR indicated that Pitx2c expression levels in a 3-month-old left atrium were ≈25% of that found in the embryo 13.5 days postconception (dpc) (Fig. 1P). Together, these findings indicate that Pitx2c is expressed in postnatal left atrium, pulmonary vein, and right ventricle and that Pitx2c expression in left atrium is silenced during aging.

Baseline Electrophysiological Features of Pitx2^{null +/−} Mice.

To test the hypothesis that Pitx2c is a susceptibility gene for atrial arrhythmias such as AF and AFL, we first used conventional six-lead surface ECG and four-lead bipolar intracardiac electrograms to record simultaneously in WT and Pitx2^{null +/−} mice (Fig. S2). Under baseline conditions, there were no differences between Pitx2^{null +/−} and WT mice in heart rate, interval between two consecutive R wave peaks (RR), interval from beginning of the P wave to the peak of the R wave (PR), or in the time elapsing from the beginning of the QRS complex to the end of the T wave (corrected Q-T interval) on surface ECGs (Table 1), with the exception of a small but significant prolongation of the QRS interval in Pitx2^{null +/−} mice. These findings were corroborated using bipolar intracardiac electrogram recordings. Endocardial programmed electrical stimulation was used to determine atrial effective refractory period, atrioventricular nodal effective refractory period, and sinus node recovery time at a basic cycle length of 100 ms. There were no significant differences between Pitx2^{null +/−} and WT mice (Table 1). During these electrophysiological studies, no episodes of spontaneous AF were observed in any of the mice studied.

Pacing-Induced Atrial Arrhythmias in Pitx2^{null +/−} Mice.

Although Pitx2^{null +/−} mice were in normal sinus rhythm at baseline, we wanted to test the notion that Pitx2c haploinsufficiency predisposed mice to atrial arrhythmias. There is precedent for this idea, because a mouse model for a gain-of-function mutation in ryanodine receptor type II (RyR2) provided a substrate for AF that was insufficient to produce spontaneous arrhythmias. RyR2 mutants required a second arrhythmogenic triggering event, in this case activation of Ca²⁺/calmodulin-dependent protein kinase II, which was uncovered by increased heart rate (17).

Inducibility of atrial arrhythmias, determined twice in each mouse, was measured with surface leads and an intraatrial catheter using the stimulation protocol previously described (17, 18). Episodes of AFL/tachycardia, defined as rapid but regular atrial rhythm lasting >1,000 ms (19), were observed frequently in Pitx2^{null +/−} mice (Fig. 2 A–C). Based on a-wave morphology and rate on the atrial electrogram, episodes of pacing-induced atrial arrhythmias were observed more frequently in Pitx2^{null +/−} (100%, six of six mice) than in WT mice (one of four mice, Fisher's exact test, P = 0.033). One of the Pitx2^{null +/−} mice showed an atrial arrhythmia only once with an episode lasting for at least 16 s, whereas five of six Pitx2^{null +/−} mice developed arrhythmias following both burst-pacing stimuli. Mean duration of atrial arrhythmias in Pitx2^{null +/−} mice, confirmed by rapid and regular a-wave on atrial electrograms, was 164 ± 131 s. In addition, we observed an episode of AFL/tachycardia in one of the WT mice (although this episode only lasted 6.8 s). Taken together, our findings indicate that Pitx2c haploinsufficiency predisposes mice to atrial arrhythmias, including AFL/tachycardia.

Pitx2c Inhibits the Pacemaker Gene Program in Left Atrium.

To gain insight into the mechanisms underlying predisposition to AFL/tachycardia, we performed microarray analysis on Pitx2^null-mutant and control hearts (Methods and Fig. S3). Interestingly the microarray study revealed that the SAN genes Shox2 and Tbx3, as well as a number of channel genes, were up-regulated in the Pitx2^null-mutant embryos when compared with controls (Fig. 3A). Among the up-regulated channel genes was Kcnq1, a potassium-channel gene that has been implicated in familial AF through a gain-of-function mutation (20).

We next used quantitative RT-PCR and whole-mount in situ hybridization to gain further insight into the mechanisms underlying predisposition to AFL/tachycardia in Pitx2c mutants. We examined expression of Hcn4, a hyperpolarization-activated, cyclic nucleotide-gated (HCN) channel that has a critical role in SAN automaticity (21, 22). In 12.5-dpc control embryos, Hcn4 was expressed in the large caval veins posterior to the heart, as well as in the forming SAN region (Fig. 3Ca and Fig. S4 A and B). In Pitx2^null+/− mutants, Hcn4 appeared to be modestly up-regulated in the left superior caval vein that also expresses Pitx2c (Fig. 3Cb and Fig. S4 C and D) (23). In Pitx2^{null −/−} embryos, the anatomy of the junction between the large veins and the atrium was disrupted (24, 25), and Hcn4 was expressed continuously in the abnormal vein–atrial junction in Pitx2^null−/−-mutant embryos and in the posterior wall of the mutant left atrium (Fig. 3Cc and Fig. S4 E and F). Quantitative RT-PCR indicated that Hcn4 was quantitatively up-regulated in both Pitx2^null+/− and Pitx2^null−/− mutant embryos (Fig. 3B).

Tbx3 is a T-box gene that is expressed in SAN progenitors and is required for correct separation of the SAN from working atrial myocardium (26). In Pitx2^{null +/−} mutants, Tbx3 expression was more prominent in the left atrioventricular canal, whereas in Pitx2^{null −/−} mutants Tbx3 was expanded and up-regulated in the left atrioventricular canal and in the midline cells at the base of the common venous sinus (Fig. 3 B and Cd–f).

Shox2, a homeodomain containing transcription factor expressed in the developing SAN, has been shown recently to be required for SAN development (27, 28). At both 12.5 and 13.5 dpc in Pitx2^{null +/−} embryos, Shox2 was expressed similarly to control SAN progenitors, although expression was up-regulated in left superior caval vein and interatrial septum of Pitx2^null+/− embryos (Fig. 3Cg, h, j, and k and Fig. S4 G–J and M–P). Shox2 up-regulation was confirmed using quantitative RT-PCR (Fig. 3B). In Pitx2^{null −/−} embryos, Shox2 was expanded in the posterior wall of the left atrium and left atrial myocardium (Fig. 3 Ci and l and Fig. S4 K, L, Q, and R). Quantitative RT-PCR also indicated that Shox2 was up-regulated in Pitx2^null−/− embryos (Fig. 3B).

In addition to genes identified in the microarray, we also investigated candidate genes in the Pitx2 mutants. Nppa, encoding the atrial natriuretic peptide hormone that regulates intravascular volume, has been reported to be a Pitx2 target and has been implicated in familial AF (29, 30). Whole-mount in situ hybridization and quantitative RT-PCR indicated that Nppa expression was up-regulated in Pitx2^{null +/−} and Pitx2^{null −/−}-mutant embryos (Fig. 3 B and Cm–o).

We next evaluated gene expression in the left atrium of P42 Pitx2^{null +/−} mice that were prone to atrial arrhythmias as shown in Fig. 2. Microarray analysis followed by quantitative RT-PCR verification revealed that genes expanded in the adult heart were similar to those up-regulated in Pitx2^null-mutant embryos. A number of potassium-channel genes, including Kcnq1, were up-regulated in the Pitx2^{null +/−} P42 left atrium, as well as the SAN genes Shox2 and Hcn4 (Fig. 3 D and E).

Pitx2c Directly Represses Shox2.

To understand the underlying molecular mechanisms in the predisposition to atrial arrhythmia in Pitx2^{null +/−} mice, we generated an allele of Pitx2, Pitx2^Flag (Fig. 4 A–C). We found that Pitx2^Flag directed Pitx2c expression in embryoid bodies cultured from 7–10 days (Fig. 4D). In the P0 heart, we observed strong expression of Pitx2c protein in mice that were heterozygous for the Pitx2^Flag allele (Fig. 4D).

Bioinformatics analysis revealed a Pitx2c recognition element in the Shox2 gene that was conserved in the mouse, human, dog, and cow genomes (Fig. 4E and Methods). Moreover, in vivo ChIP indicated that the Pitx2c recognition element, located in Shox2 intron 2, was bound by Pitx2c in both embryoid bodies and the adult left atrium (Fig. 4F). Transfection experiments using a 2-kb Shox2 genomic fragment that included the Pitx2c binding site indicated that Pitx2c repressed Shox2 expression in P19 cells and that this repression was lost when the Pitx2c binding site was mutated (Fig. 4G). Together, our findings support the hypothesis that Shox2 is a direct target for Pitx2c repression in the embryonic and adult left atrium.

Pitx2c Directly Inhibits Pacemaker Specification.

Our in vivo ChIP data and transfection experiments indicate that Pitx2c normally silences Shox2 in the left atrium. Previous work has shown that Shox2-mutant mice have defective SAN development with reduced expression of Hcn4 and Tbx3, indicating that Shox2 positively regulates the SAN genetic program. Shox2 functions, at least in part, by inhibiting Nkx2.5 expression in SAN progenitors (Fig. S5) (27, 28). Nkx2.5 promotes the atrial working myocardial program and represses the SAN genetic pathway (31, 32). Transfection experiments revealed that Shox2 directly represses the Nkx2.5 5′ flanking region and that this repression was independent of Shox2 DNA binding activity (28). Although this genetic program is intact on the right side of the embryo, on the left side Pitx2c disrupts the negative cascade by inhibiting Shox2 expression (Fig. S5).

Tbx3 also promotes the SAN phenotype while inhibiting the working atrial myocardial phenotype through both direct and indirect mechanisms. There is strong evidence that Tbx3 binds to and inhibits atrial genes, such as Cx43. Tbx3 also promotes the expression of SAN genes, most likely through a poorly understood indirect mechanism (26). Moreover, recent findings indicate that Tbx3 is genetically downstream of Shox2 (28). Together, our data support a model whereby Pitx2c directly represses Shox2 and inhibits the SAN genetic program in left atrium.

Our findings support the notion that Pitx2c loss of function promotes ectopic automaticity in the left atrium. Under normal conditions, the left atrium should be protected from impulse generation. In Pitx2^null+/− mutants, our data reveal that left atrial myocardium expresses genes that are characteristic of the SAN such as Hcn4. We suggest that the Pitx2^null+/− null mutant left atrium provides an arrhythmogenic substrate that would enhance other pathologic triggers for atrial arrhythmias.

Atrial Fibrillation and Left–Right Asymmetry: A Direct Molecular Connection.

Our findings indicate that the Pitx2c-mediated LRA signaling cascade inhibits SAN specification by disrupting a negative regulatory transcriptional cascade. Previous work showed that Pitx2c directs formation of left and right anatomic characteristics that separate the systemic and pulmonic circulation that is vital for normal cardiac function. Morphologic abnormalities in Pitx2^null homozygous mutants include right atrial isomerism with bilateral venous valves, deficiency of the primary interatrial septum, and anomalous pulmonary venous drainage (31, 33). In Pitx2c ^−/−-mutant embryos, caval and pulmonary veins drain into an abnormal medial venous sinus. Moreover, there was a deficiency in contribution of the Pitx2-expressing lineage to pulmonary veins in Pitx2c homozygous mutant embryos (25) because of a defect in pulmonary vein precursors (24).

Mommersteeg and colleagues (31) reported that Pitx2c homozygous mutants had a bilateral SAN that expressed SAN markers such as Hcn4. Our findings suggest that Pitx2c regulates the precise location of the SAN by directly inhibiting Shox2, a transcriptional regulator of SAN gene program. Moreover, we show that Pitx2 haploinsufficiency promotes the occurrence of atrial tachyarrhythmias.

Our data uncover a remarkably close connection between LRA and the SAN genetic program. Pitx2c is a direct target of the TGFβ superfamily member Nodal that is expressed transiently at early stages of development (10). Left-sided Pitx2c expression persists in organ primordia well after Nodal has been silenced (34, 35). Furthermore, the identification of Pitx2c as a susceptibility gene for atrial arrhythmias provides deeper insight into the genetic pathways that predispose to AF and AFL. Importantly, the Nodal-Pitx2c signaling cassette is evolutionarily conserved, raising the possibility that Nodal also is implicated in predisposition to atrial arrhythmias. Indeed, mutations in Nodal have been identified in human patients with LRA defects but have not been reported in familial AF (36).

Other Mechanisms Contributing to Pitx2c Predisposition to Atrial Arrhythmia.

In addition to the Shox2-mediated mechanism, we found other gene expression changes in Pitx2c mutants that suggest other contributory mechanisms for predisposition to atrial arrhythmias. Atrial natriuretic peptide, encoded by Nppa, regulates multiple ion channels in atrial cardiomyocytes. An atrial natriuretic peptide gain-of-function mutation that may promote arrhythmias within atrial myocardium has been identified in human patients (29). Our finding that Nppa was up-regulated in Pitx2c mutants suggests that similar mechanisms may contribute to the predisposition to atrial arrhythmia in Pitx2c mutants.

Previous work indicated that a gain-of-function substitution in a conserved serine of Kcnq1 (S140G) results in familial AF (20). Moreover, separate mutations in Kcnq1—a serine-to-proline (S209P) substitution and a valine-to-methionine (V141M) substitution—also were reported in familial AF (37, 38). In vitro experiments indicate that the gain-of-function mutations in Kcnq1 may be caused by defects in channel inactivation (39). Our data indicate that Kcnq1 is up-regulated in Pitx2c mutants. Together these findings suggest that Pitx2c likely has multiple targets that are relevant to the predisposition to atrial arrhythmia.

Mouse Alleles and Transgenic Lines.

The Pitx2^null allele, which removes function of all Pitx2 isoforms, and the Pitx2^tg3k transgenic line have been described previously (15, 16).

Real-Time Quantitative PCR.

Total RNA was isolated from embryonic hearts and adult hearts using RNeasy Micro Kit (QIAGEN), followed by RT-PCR using Super Script^TM II reverse transcriptase (Invitrogen). Real-time thermal cycling was performed using Mx3000P thermal cycler (Stratagene) with SYBR Green JumpStart^TM Taq ReadyMix^TM (Sigma). Sequences of primers are available upon request.

Cardiac Electrophysiology.

Cardiac electrophysiology was performed as described (40).

Microarray Studies.

Total RNA was extracted from embryonic hearts or P42 hearts and purified using either RNeasy Micro Kit (QIAGEN) for embryonic hearts or TRIzol Reagent (Invitrogen) for postnatal hearts. DNA microarray analysis was performed using the OneArray Mouse Whole Genome Array that interrogated 17,455 unique genes using 29,972 probes (Phalanx Biotech Group). Graphical presentations of the array data are shown in Fig S3.

ChIP.

ChIP analysis was performed using a ChIP assay kit (Upstate).

Generation of Reporter Constructs.

Using rVista2.0 (41) and Ensembl genomic alignments (http://www.ensembl.org/), we analyzed Shox2 genomic sequence and the genomic sequence5 kb upstream of Shox2 to identify conserved Pitx2 binding sites. This analysis was followed by the ChIP analysis, and the Pitx2 biding site “TAATCC” in intron between exon 2 and exon3 of Shox2 was validated.

Site-Directed Mutagenesis.

Site-directed mutagenesis of the Pitx2 binding sites in the Shox2 intron was achieved by using the QuikChange II site-directed mutagenesis kit (Stratagene).