Pitx2c increases in atrial myocytes from chronic atrial ﬁbrillation patients enhancing I Ks and decreasing I Ca,L

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Keywords


Introduction
Atrial fibrillation (AF) is the most frequent arrhythmia and the main risk factor associated with myocardial-related cerebrovascular events. 1 Nowadays, treatment of AF is clearly suboptimal, 2 mainly due to rapid changes in the electrical properties of the atria (electrical remodelling) induced by the arrhythmia itself. 1 This electrical remodelling promotes the maintenance and recurrence of AF 3 and is characterized by a marked shortening of the atrial action potential duration (APD) and refractoriness as a consequence of alterations in the expression and function of L-type Ca 2+ and inward rectifier and voltage-dependent K + channels. 4 -6 In chronic AF (CAF) patients, the L-type Ca 2+ current (I Ca,L ) decrease 4 and the slow delayed rectifier K + current (I Ks ) increase 5 critically contribute to the APD shortening. 7 Pitx2 is a transcription factor that plays a critical role in directing cardiac asymmetric morphogenesis. 8 In humans, the PITX2 gene generates three main isoforms (Pitx2a, Pitx2b, and Pitx2c), the latter being the predominant isoform in the heart. 8 Increasing evidence pointed to a role of Pitx2c in the pathophysiology of AF. Indeed, it has been described that single nucleotide variants on human chromosome 4q25, ≈170 kb distal to the PITX2 gene, are associated with an increased risk of familial AF. 9 Pitx2c expression can be increased 10 or decreased 11 in atrial appendage samples obtained from AF patients, compared with those from patients in sinus rhythm (SR). However, the putative role of Pitx2c in electrical remodelling in humans has not been explored as data available thus far have been obtained in mouse models. 11 -15 Therefore, this study was undertaken to measure Pitx2c expression in human atrial myocytes from SR and CAF patients and to determine the effects of Pitx2c on I Ks and I Ca,L . Importantly, the results obtained demonstrated that Pitx2c expression is augmented in atrial myocytes from CAF patients. Furthermore, Pitx2c increases I Ks by enhancing expression of Ks channel encoding genes, whereas it decreases I Ca,L by a post-translational effect.

Methods
This study was approved by the Investigation Committee of the Hospital Universitario Gregorio Marañó n (CNIC-13) and conforms to the principles outlined in the Declaration of Helsinki. Each patient gave written informed consent. Clinical data of the patients are included in Supplementary material online, Table S1.

Analysis of the mRNA expression in human atrial myocytes
Real-time quantitative polymerase chain reaction (qPCR) was performed using TaqMan gene expression assays (Life Technologies, USA). The cycle to threshold (Ct) values were normalized to 18S rRNA. To compare CAF vs. SR expression differences, the respective data were transformed from DCt values to equivalent fold differences using the following equation: Fold Difference (mean DCtSR -mean DCtCAF) ¼ 2 (mean DCtSR -mean DCtCAF) . 7,16

Patch clamping
Outward K + currents and I Ca,L were recorded in human atrial myocytes and HL-1 cells at room temperature using the whole-cell patch-clamp technique (micropipette resistance ,3.5 MV). 5,7,16 Human atrial myocytes were isolated from right atrial appendages obtained from SR (n ¼ 10) and CAF (n ¼ 10) patients as described. 5,7,16,17 HL-1 cells were cultured as described previously 16,18 and transfected with CMV-Pitx2c (2 mg) by using Lipofectamine 2000. In HL-1 cells, I Ca,L was recorded by using Ba 2+ as charge carrier to increase current density and to eliminate Ca 2+dependent inactivation. 16 Dofetilide (1 mM) and HMR-1556 (1 mM) (Sigma, UK and Thermo Fisher, USA) were used to inhibit the rapid delayed rectifier current (I Kr ) and I Ks , 19 respectively. Series resistance was compensated manually and usually ≥80% compensation was achieved. Thus, no significant voltage errors (,5 mV) due to series resistance were expected with the micropipettes used.

Luciferase gene expression reporter assays
Luciferase activity assays were conducted in CHO cells transfected with the corresponding promoters 20 cotransfected with an empty vector or with Pitx2c (100 ng). Luciferase activity was measured 48 h after transfection and normalized to sample protein concentration. All reporter assays were performed in triplicate.

Statistical analysis
Results are expressed as mean + SEM. Unpaired t-test or one-way analysis of variance (ANOVA) followed by Newman -Keuls test was used where appropriate. In small-sized samples (n , 15), statistical significance was confirmed by using non-parametric tests. Comparisons between categorical variables were performed using Z-test. To take into account repeated sample assessments, data were analysed with multilevel mixed-effects models. A value of P , 0.05 was considered significant.
Additional details are presented in Supplementary material.

Pitx2c expression increases in atrial myocytes from CAF patients
The expression of Pitx2c was measured by qPCR in myocytes enzymatically isolated from atrial appendages obtained from SR (n ¼ 10) and CAF (n ¼ 10) patients. Comparison of DCt values demonstrated that Pitx2c expression was significantly larger in CAF myocytes ( Figure 1A). Indeed, transformation of DCt values to fold differences demonstrated that Pitx2c expression was approximately two times larger in CAF than in SR myocytes ( Figure 1B).
3.2 Pitx2c expression levels correlate with I Ks and I Ca,L densities in SR and CAF myocytes An increase in I Ks and a decrease in I Ca,L densities are hallmarks of CAF-induced electrical remodelling. 4,5,7,16 Therefore, we analysed a putative correlation between Pitx2c levels and I Ks and I Ca,L densities.
In these experiments, myocytes enzymatically isolated from each atrial appendage were separated in two fractions: one for Pitx2c expression assay and another for I Ca,L or I Ks recordings. Interestingly, association studies demonstrated significant correlations of Pitx2c expression with I Ks density at +60 mV ( Figure 1C) and I Ca,L density at +10 mV ( Figure 1D), showing that the larger the Pitx2c expression the higher I Ks and the lower I Ca,L densities. To test whether these correlations were driven by differences between SR and CAF groups, a partial correlation was calculated. Pitx2c expression and I Ks or I Ca,L densities are correlated in the absence of the influence of the group effect ( Figure 1E and F). Moreover, a positive correlation was observed between Pitx2c mRNA levels and the expression of KCNQ1 and KCNE1 mRNA. However, Pitx2c mRNA expression was not correlated with CACNA1C mRNA levels (Supplementary material online, Figure S1).
The existence of such correlations suggested that Pitx2c may modulate the expression of the channels that generate these currents. To test this hypothesis, we first analysed the properties of I Ks and I Ca,L recorded in SR and CAF myocytes. As demonstrated previously, 5,7 cell capacitance of CAF myocytes was greater than that of SR myocytes (96.9 + 6.2 vs. 56.7 + 2.2 pF, P , 0.0001; n ¼ 194). In the presence of 4-aminopyridine (2 mM), used to inhibit the transient outward (I to1 ) and the ultrarapid component of the delayed rectifier (I Kur ) K + currents, a current whose time-and voltage-dependent properties are consistent with those of I Ks was recorded. 5,7 Figure 2A and B shows current traces and current density-voltage curves for I Ks recorded in SR and CAF myocytes. I Ks was measured as the difference between the amplitudes at the end and beginning of 4 s pulses and normalized by cell capacitance to obtain current density. Confirming previous results, 5,7 in myocytes obtained from CAF patients, I Ks density increased approximately two-fold compared with myocytes from SR patients (2.3 + 0.2 vs. 1.2 + 0.2 pA/pF at +60 mV, P , 0.01; n ¼ 40 and 86, respectively) and accordingly, KCNQ1 and KCNE1 mRNA expressions were also increased (Supplementary material online, Figure S1). Moreover, the time course of current activation was significantly faster, and the conductance curve was shifted to more hyperpolarized potentials in CAF myocytes relative to SR myocytes ( Table 1). I Ca,L was measured as the difference between the peak amplitude and the amplitude at the end of the pulses and normalized by cell capacitance to obtain the current density. As expected, 4,7,16 CACNA1C mRNA expression (Supplementary material online, Figure S1) and I Ca,L density in CAF myocytes (22.0 + 0.3 vs. 23.8 + 0.3 pA/pF at +10 mV, P , 0.01; n ¼ 27 and 41, respectively) ( Figure 2C and D) were significantly smaller than in SR myocytes. The midpoint of the I Ca,L inactivation curve in CAF was significantly shifted to more positive potentials compared with SR myocytes ( Table 1). However, no differences were observed in the activation (tact ¼ 1.2 + 0.1 vs. 1.1 + 0.1 ms at +10 mV) and inactivation kinetics and voltage dependence of activation ( Table 1).

Pitx2c increases I Ks density in HL-1 cells
The results suggested that Pitx2c could be involved in the CAF-induced I Ks increase and I Ca,L decrease. Next, we analysed the molecular mechanisms underlying the putative Pitx2c-induced I Ks increase and I Ca,L decrease in the mouse atrial-derived HL-1 cardiomyocyte-like cell line. Cells non-transfected with Pitx2c randomly patched (n ¼ 32) could be separated into three groups, depending on the main outward K + current recorded by applying 4 s pulses from 280 mV to potentials ranging from 280 to +40 mV. About 36% of the cells ( Figure 3A Figure S2). I Kr was completely absent in ≈23% of the cells, in which a slow-activating noninactivating dofetilide-resistant current with biophysical properties similar to those of I Ks was recorded (I Ks -predominant). In the rest of the cells (≈41%), both dofetilide-sensitive and -resistant currents could be recorded (I Kr + I Ks ). Transfection of Pitx2c did not modify cell capacitance (50.6 + 8.7 pF, P . 0.05) but significantly changed celltype distribution (n ¼ 41). Pitx2c decreased the percentage of the cells that exhibited the I Kr -predominant pattern, whereas it increased the percentage of the cells with the I Ks -predominant pattern (P , 0.05) ( Figure 3B). I Kr was recorded in I Kr -predominant and I Kr + I Ks cells. The results demonstrated that Pitx2c transfection did not significantly modify I Kr density, the voltage dependence of activation, or the time dependence of current activation or deactivation ( Table 2 and Supplementary material online, Figure S2). In cells with I Ks -predominant and I Kr + I Ks patterns perfused with dofetilide, I Ks was recorded by applying 4 s pulses from 280 mV to potentials ranging between 280 and +60 mV, followed by repolarizing pulses to 230 mV to record the tail currents ( Figure 3C and D). Under these conditions, the timedependent current was completely inhibited by HMR-1556 (1 mM) (inset of Figure 3D). Importantly, Pitx2c significantly increased the current density (from 2.6 + 0.4 to 6.1 + 1.1 pA/pF at +60 mV; P , 0.05) ( Figure 3D and E) and accelerated the time course of activation ( Table 2). It also significantly increased the tail current density ( Figure 3F) and shifted the activation curves to more hyperpolarized potentials ( Table 2). Densitometric analysis from western blots conducted in HL-1 cells ( Figure 4A) demonstrated that Pitx2c increased by approximately two-fold Kv7.1 and minK protein expression   Table 1 Electrophysiological properties of I Ks and I Ca,L recorded in isolated myocytes from atrial appendages obtained from SR and CAF patients   Pitx2c and human atrial currents ( Figure 4B and C ). These results suggested that Pitx2c regulates the expression of the ion channels responsible for I Ks .
It has been previously demonstrated that Pitx2c binds the DNA sequence TAATCC by means of a lysine residue located in position 9 of helix 3 in its homeodomain. 21 Therefore, the consensus sequence TAATCC was searched near the transcription start site (TSS) of KCNQ1 and KCNE1 genes. In KCNQ1, a single consensus sequence was found at a distance of 2 kb upstream of the TSS ( Figure 5A). In KCNE1, two consensus sequences were found at 1.4 and 1.2 kb upstream of the TSS ( Figure 5C). To determine whether Pitx2c may increase KCNQ1 and KCNE1 transcription, luciferase activity assays were conducted in CHO cells. A construction carrying the KCNQ1 proximal promoter and including the Pitx2c-binding site (22186/0) was cloned in the pLightSwitch_Prom luciferase expression reporter vector (KCNQ1_WT) and transiently transfected. The same construction but with the Pitx2c-binding site mutated to AAAAAA (KCNQ1_mut) was also tested ( Figure 5A). Cotransfection of Pitx2c significantly increased luciferase activity in cells transfected with KCNQ1_WT ( Figure 5B), indicating that Pitx2c increased KCNQ1 transcription. However, Pitx2c did not increase luciferase activity in cells transfected with KCNQ1_mut, revealing that the presence of the consensus sequence TAATCC was critical for the Pitx2c effect on KCNQ1 transcription ( Figure 5B). To determine the putative interaction of Pitx2c with KCNE1, a construction carrying the KCNE1 proximal promoter and including both Pitx2c-binding sites (21429/+16) was cloned in the PGL3-Basic luciferase expression reporter vector (F2) 20 and transfected in CHO cells. A shorter construction (F7) without the Pitx2cbinding sites (2311/+16) was also tested ( Figure 5C). Pitx2c produced a 1.8-fold increase in luciferase activity in cells transfected with F2, demonstrating that Pitx2c was able to increase KCNE1 transcription.
Conversely, it did not increase luciferase activity in cells transfected with F7, indicating that the presence of the TAATCC sequence is critical for the Pitx2c effects on KCNE1 transcription. It is known that atrial natriuretic peptide (ANP) encoded by the NPPA gene is one of the main targets of Pitx2c and is frequently used to evaluate transcriptional activity of Pitx2c as the gold standard. 21 Pitx2c increased NPPA transcription by interacting with several Pitx2c-binding sites within its promoter region. 22,23 For this reason, CHO cells expressing the pLightSwitch_ Prom luciferase expression reporter vector carrying NPPA minimal promoter were used as a positive control. As expected, in both groups of experiments, transfection of Pitx2c significantly increased luciferase activity generated by the NPPA promoter ( Figure 5B and D).

Pitx2c decreases I Ca,L density in HL-1 cells
In HL-1 cells, both I Ca,L and I Ca,T can be recorded. 16 To determine the effects of Pitx2c on I Ca,L , currents were recorded using Ba 2+ as the charge carrier (I Ba ) and the holding potential was fixed to 230 mV to fully inactivate I Ca,T . Under these conditions, 80% of the cells exhibited a measurable I Ba that reached its maximum value at +20 mV when 500 ms pulses to potentials ranging between 240 and +50 mV were applied (n ¼ 49) ( Figure 6A). Transfection of Pitx2c decreased the proportion of cells with measurable current to 57% (P , 0.01, n ¼ 53), without modifying cell capacitance (40.7 + 8.3 pF; P . 0.05). Pitx2c markedly decreased I Ba density at potentials between 0 and +50 mV (n ¼ 23, P , 0.05) ( Figure 6A and B). Moreover, it shifted the midpoint of the inactivation curve to more positive potentials ( Figure 6C) and slowed recovery from inactivation ( Figure 6D and Table 3). Conversely, it did not modify the voltage dependence of   Pitx2c and human atrial currents activation ( Figure 6C), kinetics of activation and inactivation, and the reversal potential ( Table 3). Additionally, western blot analyses demonstrated that Pitx2c did not significantly modify Cav1.2 expression ( Figure 7A and B), suggesting that Pitx2c does not regulate CACNA1C transcription, despite the presence of a TAATCC sequence 100 bp downstream of the TSS of CACNA1C. In contrast, neither CACNB2 nor CACNA2D minimal promoters exhibit the consensus sequence for Pitx2c binding. From this result, we surmised that the inhibition could be due to post-translational modifications. As shown earlier, Pitx2c activates the NPPA promoter. Moreover, it was previously described that ANP inhibits I Ca,L in atrial myocytes through a mechanism involving an increase in cyclic GMP levels. 24 Furthermore, Pitx2c significantly increased ANP concentration in the culture medium of HL-1 cells (n ¼ 6, P , 0.05) ( Figure 7C). Therefore, we hypothesized that the effects of Pitx2c on I Ca,L could be mediated by ANP. To preliminarily answer this question, the effects of Pitx2c transfection on I Ba were analysed in HL-1 cells incubated or not with the ANP type A receptor antagonist A71915 (0.5 mM) for 24 h. 25 As shown in Figure 7D, incubation with the antagonist completely prevented the inhibitory action induced by Pitx2c on I Ba , suggesting that ANP was responsible for this effect. It was described that isoproterenol-induced I Ca,L potentiation is larger in atrial myocytes from rats with heart failure than in control myocytes. This result was associated to higher levels of ANP in heart failure animals and due to an ANP-induced effect on cGMP-dependent phosphodiesterases. 24 We tested whether the Pitx2c-induced increase in ANP expression would modulate isoproterenol effects on I Ba recorded in HL-1 cells. In cells non-transfected with Pitx2c, isoproterenol (50 nM) increased I Ba recorded at +20 mV by 26.8 + 7.7% (n ¼ 7), whereas in Pitx2c-transfected cells, the I Ba increase induced by isoproterenol was significantly higher (65.0 + 11.3% at +20 mV, P , 0.05, n ¼ 7) (Figure 7E and F, inset).

Discussion
To our knowledge, this is the first study comparing Pitx2c expression in isolated atrial myocytes from SR and CAF patients and analysing the effects of Pitx2c on I Ks and I Ca,L . The results demonstrate that expression of Pitx2c increased in CAF myocytes and that this increase directly and inversely correlated with the densities of I Ks and I Ca,L , respectively.

Pitx2c and AF
In 2007, Gudbjartsson et al. 9 published the first genome-wide association study, describing that non-coding single nucleotide polymorphisms on chromosome 4q25, 170 kb distal to the PITX2 gene, were strongly associated with AF. This association has been further confirmed in populations of distinct ethnic backgrounds 26 and has also been reported for post-cardiac surgery AF 27 and left atrial dilatation 28 and responsiveness to anti-arrhythmic drug therapy. 29 The proposed mechanisms underlying this relationship are mainly based on its role in cardiac development, including alterations in the sinoatrial node genetic programme 13 and in the development of myocardial sleeves of the pulmonary veins. 14 However, recent evidences in mouse models suggest a role of Pitx2c in adult atria and, specifically, in the events that form part of the remodelling process associated with AF. 11,12,15 Chinchilla et al. 11 showed that Pitx2c mRNA expression was markedly decreased in right and left atrial appendages obtained from patients with AF when compared with those obtained from SR patients. More recently, it was demonstrated that Pitx2c expression increased in atrial tissue from patients in AF at the time of sample collection. 10 These discrepancies could be attributed to patient characteristics and AF progression. AF patients included in these two studies were not specifically classified as CAF patients, and the disease duration was not mentioned. More importantly, both studies were conducted using whole atrial appendage samples. It is known that the presence of fibroblasts and other non-myocyte cells in whole atrial samples may complicate the interpretation of gene expression analysis in myocytes. 16 To overcome these issues, we conducted qPCR experiments in isolated myocytes from atrial appendages obtained from patients in SR and patients diagnosed with CAF (.6 month at the time of surgery). Under these conditions, our qPCR experiments showed that Pitx2c mRNA expression was approximately two-fold greater in CAF than in SR myocytes.

Pitx2c modulates I Ks and I Ca,L
Up-regulation of I Ks and down-regulation of I Ca,L are hallmarks of CAF-induced electrical remodelling in humans and critically contribute to the abbreviation of APD and atrial refractory period. 3 -5,7,16 Our results confirmed previous observations 5,7 and showed that in atrial myocytes from CAF patients, I Ks density was approximately two-fold higher, the activation kinetics was significantly faster, and the conductance curve was shifted to more hyperpolarized potentials when compared with SR myocytes. In contrast, in CAF myocytes, I Ca,L density was reduced and the inactivation curves were shifted to more positive potentials in comparison to SR myocytes. Importantly, in human atrial myocytes, increase in Pitx2c expression was positively and inversely correlated with the I Ks increase and I Ca,L decrease, respectively, suggesting that Pitx2c could be involved in the changes in these currents that characterize CAF-induced electrical remodelling. On the basis of our results, Pitx2c deletion should lead to APD prolongation. However, in mice heterozygous for Pitx2c 15   deficient mice, 11 atrial APD was shortened and unmodified, respectively, compared with wild-type animals. It is known that atrial action potential repolarization of mouse and human is different. 30 Indeed, mouse atrial myocytes lack functional Kv11.1 + miRP1 and Kv7.1 + minK channels and, thus, I Kr and I Ks , respectively, cannot be recorded. Therefore, changes in APD secondary to modulation of Kv7.1 + minK channels would not be expected in mouse. In contrast, repolarizing sustained K + currents in mice are generated by Kv1.5 and Kv2.1 channels. Therefore, it is possible that Pitx2c-deficient mice display altered expression of Kv2.1 channels or other unknown factors, which may compensate the effects on APD secondary to the I Ca,L increase upon Pitx2c deletion.
To elucidate the molecular mechanism underlying Pitx2c-induced effects, HL-1 cells were used as experimental model. These cells are mouse atrial-derived cells that can be maintained in culture and have been previously used to test Pitx2c actions. 11,31 Depending on the main outward K + current recorded, three groups of HL-1 cells were established: I Kr -predominant, I Ks -predominant, and intermediate (with . Each point represents the mean + SEM of more than 12 cells from at least 3 independent batches in each group. In (C ) and (D), *P , 0.05 vs. cells non-transfected with Pitx2c. I Ba density -voltage relationships for currents recorded in HL-1 cells transfected (F ) or not (E) with Pitx2c in the presence or absence of 50 nM isoproterenol. The inset shows I Ba density at +20 mV. Each point/bar represents the mean + SEM of seven cells from at least two independent batches in each group. In (E) and (F), *P , 0.05 vs. control and # P , 0.05 vs. cells non-transfected with Pitx2c. Unpaired t-test (B and C ), ANOVA followed by Newman-Keuls test (D -F, and inset), and multilevel mixed-effects model (D).
Pitx2c and human atrial currents I Kr and I Ks ) cells. Pitx2c decreased the percentage of I Kr -predominant cells, increased the percentage of I Ks -predominant cells, augmented I Ks density, and shifted Ks channel activation to more negative potentials. Both Kv7.1 and minK protein expressions were increased in the presence of Pitx2c, suggesting that expression of channel subunits was up-regulated by the transcription factor. Luciferase assays demonstrated that Pitx2c activated KCNQ1 and KCNE1 promoter regions, an effect that was blunted in the absence of the Pitx2c-binding motifs. In atrial samples from CAF patients, both increases and decreases in Kv7.1 and minK expressions have been described. 5,32,33 Our qPCR results demonstrated that Pitx2c produced a simultaneous increase in Kv7.1 and minK expressions.
Expression of Pitx2c in HL-1 cells increased the percentage of cells without measurable I Ca,L , reduced the current density, and shifted the voltage dependence of inactivation to positive potentials. Cav1.2 expression was not modified by Pitx2c, even when CACNA1C minimal promoter exhibits a TAATCC sequence. Therefore, it can be speculated that the presence of this sequence, 100 bp downstream of the TSS, is not enough for promoting an effect of Pitx2c on CACNA1C transcription. This result suggests that the Pitx2c-induced I Ca,L inhibition was not due to a direct regulation of channel expression, but involved a post-translational modification of channel function. We surmised that the effect could be produced by a mediator which is a Pitx2c target and is able to reduce Ca 2+ channel function. Pitx2c can target multiple genes; 8 however, we focused on the ANP as it meets both conditions: it is one of the main targets of Pitx2c 21 and inhibits I Ca,L . 24 As we and others demonstrate, Pitx2c markedly activates NPPA promoter, increases NPPA gene transcription, 21 and augments ANP production. In contrast, it has been described that ANP inhibits atrial I Ca,L via the accumulation of cGMP, which affected the phosphorylation/dephosphorylation balance of the Ca 2+ channel. 24 Our results demonstrated that the I Ca,L density reduction and the shift in the inactivation curve induced by Pitx2c were abolished by the incubation with an antagonist of the ANP receptor (A71915). Therefore, we suggest that Pitx2c regulates I Ca,L by a posttranslational mechanism involving ANP.

Study limitations
All samples came from right atrial appendages, which could not be representative of the rest of the atria. Furthermore, Pitx2c expression is higher in left than in right atria, 8 and thus it cannot be ruled out that the increase in Pitx2c is more relevant for left atrial myocytes. In this case, it is possible that Pitx2c would play a role in the exacerbation of electrophysiological heterogeneity that characterizes CAF. 5 Pitx2c expression and ion channel function could be influenced by age, sex, pharmacological treatment, and/or underlying cardiac diseases of the patients. Interestingly, I Ca,L density and plasma ANP levels are influenced by the clinical history of the donors, in such a way that the ANP levels are higher and I Ca,L density is smaller in patients with mitral valve disease or decreased left ventricular function. 34 In our sample, the proportion of patients with ischaemic heart disease alone and combined with valvular cardiomyopathy was equally distributed in both groups. However, more patients were in decompensated heart failure [New York Heart Association (NYHA) III and IV] in the CAF group than in the SR group. Interestingly, multiple linear regression analysis confirmed that CAF influenced both I Ca,L and I Ks densities, whereas the NYHA class and the presence of mitral valve disease did not (Supplementary material online, Tables S2 and S3).

Conclusions
We demonstrate that Pitx2c expression increases in human atrial myocytes from CAF patients, and this increase correlates with the I Ks increase and I Ca,L decrease that characterize CAF-induced electrical remodelling. We propose that through this mechanism Pitx2c is involved in the APD and refractory period shortening that enhances arrhythmia recurrence and maintenance.