Novel PITX2 Homeodomain-Contained Mutations from ATRIAL Fibrillation Patients Deteriorate Calcium Homeostasis

Atrial fibrillation (AF) is the most common cardiac arrhythmia in the human population, with an estimated incidence of 1–2% in young adults but increasing to more than 10% in 80+ years patients. Pituitary Homeobox 2, Paired Like Homeodomain 2 (PITX2c) loss-of-function in mice revealed that this homeodomain (HD)-containing transcription factor plays a pivotal role in atrial electrophysiology and calcium homeostasis and point to PITX2 as a candidate gene for AF. To address this issue, we recruited 31 AF patients for genetic analyses of both the known risk alleles and PITX2c open reading frame (ORF) re-sequencing. We found two-point mutations in the homedomain of PITX2 and three other variants in the 5’untranslated region. A 65 years old male patient without 4q25 risk variants but with recurrent AF displayed two distinct HD-mutations, NM_000325.5:c.309G>C (Gln103His) and NM_000325.5:c.370G>A (Glu124Lys), which both resulted in a change within a highly conserved amino acid position. To address the functional impact of the PITX2 HD mutations, we generated plasmid constructs with mutated version of each nucleotide variant (MD4 and MD5, respectively) as well as a dominant negative control construct in which the PITX2 HD was lacking (DN). Functional analyses demonstrated PITX2c MD4 and PITX2c MD5 decreased Nppa-luciferase transactivation by 50% and 40%, respectively, similar to the PITX2c DN (50%), while Shox2 promoter repression was also impaired. Co-transactivation with other cardiac-enriched co-factors, such as Gata4 and Nkx2.5, was similarly impaired, further supporting the pivotal role of these mutations for correct PITX2c function. Furthermore, when expressed in HL1 cardiomyocyte cultures, the PITX2 mutants impaired endogenous expression of calcium regulatory proteins and induced alterations in sarcoplasmic reticulum (SR) calcium accumulation. This favored alternating and irregular calcium transient amplitudes, causing deterioration of the beat-to-beat stability upon elevation of the stimulation frequency. Overall this data demonstrate that these novel PITX2c HD-mutations might be causative of atrial fibrillation in the carrier.


Introduction
Atrial fibrillation (AF) is the most common cardiac arrhythmia in the human population, with an estimated incidence of 1-2% in young adults but increasing to more than 10% in 80+ years patients [1]. Thanks to genome wide association studies (GWAS) an increasing number of single nucleotide polymorphisms (SNPs) have been associated with risk of AF [2][3][4][5] with some of the most significant are located in 4q25 locus. Given the key developmental role of Pitx2 during cardiogenesis [6,7], and particularly its role in pulmonary vein formation [8], it has been postulated that PITX2c dysfunction might be responsible for AF [2]. Experimental evidence in distinct laboratories, including ours, have demonstrated that PITX2c loss of function predisposes to atrial arrhythmogenesis [9][10][11][12] and provokes disturbances in the calcium homeostasis [13]. In line with this, the 4q25 risk variant rs13143308 has recently been linked to defective calcium homeostasis [14], pointing to PITX2 as a candidate gene for AF.
In this study we have searched for mutations in AF patients in a small cohort of 31 patients with distinct degrees of AF; sporadic, recurrent and permanent. Assessment of known 4q25 AF risk variants (rs2200733, rs13143308, rs10033464) was run in parallel with direct DNA sequencing of the PITX2 open reading frame (ORF). We have identified two-point mutations in the homeodomain of PITX2 and three polymorphisms in the 5´untranslated region. To test if these novel PITX2 HD point mutations could underlie the onset of atrial fibrillation in this patient, we assessed their functional impact on transactivation of both reported constructs, on endogenous calcium regulatory genes, and on calcium homeostasis and beat-to-beat stability in transfected cardiomyocyte cultures. Functionally, both mutations impaired Nppa transactivation and Shox2 repression as well as endogenous calcium regulatory gene expression. This in turn deteriorated the calcium homeostasis and beat-to-beat stability in transfected cardiomyocyte cultures.

Patient Recruitment and Sample Collection
Patients referred to the Cardiology Unit of the Regional Hospital in Jaen, during a 24-month period, with a recorded history of atrial fibrillation or de novo assessment of paroxysmal AF were included in the study. Informed consent was obtained in all cases. Peripheral blood samples were collected from all enrolled patients in EDTA tubes and stored at −20 • C until DNA extraction was carried out. Control genomic DNAs were obtained from the Spanish DNA National Bank (Banco Nacional de ADN, Salamanca, Spain). Genomic DNA from all participants was extracted from blood lymphocytes with Wizard Genomic DNA Purification Kit (Promega; Madison WI, USA).

DNA Sequencing
Mutations of the PITX2 gene (NM_000325) were screened according to the following method. Each PITX2 exon and their splice sites were amplified by polymerase chain reaction (PCR) using intronic PCR primers as detailed in Table 1. For all PCR, genomic DNA (100 ng) was amplified using a Mastercycler personal thermal cycler (Eppendorf, Hamburg, Germany) in 25 µL reaction mixture containing 1.5 mM MgCl 2 , 1 mM of each primer, 0.2 mM deoxyribonucleoside triphosphate, and 0.5 IU of GoTaq_Flexi DNA polymerase (Promega, Madison, WI, USA). The resulting reaction mixtures were purified using the Genomics Millipore Montage ® PCR Centrifugal Filter Device (Millipore Corporate Headquarters, Billerica, USA) the PCR products were sequenced (Sistemas Genómicos, Valencia, Spain). Chromatogram files were processed using Chromas Pro 2.33 software (Technelysium Pty Ltd.) and Blast (www.ncbi.nlm.nih.gov). Sequence alignments with the reference PITX2 gene (NM_000325) were performed. In case of abnormality detection, PCR amplification and re-sequencing was repeated at least three times for further validation. Variants were described using the Human Genome Variation Society guidelines (http://www.varnomen.hgvs.org, accessed on 1 March 2020) for nomenclature [34]. Table 1. Primer sequences used for human PITX2 sequencing and for site-directed mutagenesis.
The identified mutation was introduced into the wild-type PITX2c construct using an iProof High Fidelity DNA polymerase (BioRad, Hercules, CA, USA) with a complementary pair of primers. The mutant was sequenced to confirm the desired mutation and to exclude any other sequence variations.
PCR-based site-directed mutagenesis was performed using Stratagene QuikChange Site-Directed Mutagenesis kit, but using the enzymes and buffers from the Bio-Rad iPROOF PCR kit. Amplification conditions were: 98 • C for 60 s followed by 14 cycles of 98 • C for 10 s, 55 • C for 30 s and 72 • C for 6 min followed by 72 • C for 10 min. All cloning PCRs were performed by using iProof High-Fidelity DNA Polymerase (Bio-Rad). Amplification conditions were: 98 • C for 30 s followed by 35 cycles of 98 • C for 10 s, 64 • C for 30 s and 72 • C for 2 min followed by 72 • C for 10 min.

Cell Culture and Luciferase Transactivation Assays
Mouse fibroblast (3T3) cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 200 mM L-Glu as well as 1000 units/mL penicillin and 10 µg/mL streptomycin. 3T3 cells were cultured 24 h prior transfection. The rat Nppa-pGluc reporter and the mouse Shox2-pGluc reporter constructs [11], with the corresponding internal control reporter plasmid psV40-pCluc were used in transient transfection assays to explore the transactivation activity of different PITX2c constructs, alone or in distinct combinations. Briefly, 3T3 cells were transfected with either 400 ng of wild-type PITX2c-pcDNA, PITX2cDN-pcDNA, the corresponding mutant PITX2c-pcDNA construct (MD4, carrying Gln103His mutation; MD5 carrying Glu124Lys mutations; MD4+5 carrying both Gln103His and Glu124Lys mutations), 400 ng of Nppa-pGluc reporter construct and 100 ng of psV40-pCluc control reporter vector using Lipofectamine 2000 Transfection Reagent (Invitrogen, Carlsbad, CA, USA). In addition, co-transfection experiments with Gata4, Nkx2.5 or both expression vectors was performed as previously reported [38]. Transfected cells were incubated for 24 h and cell culture supernatant were harvested and assayed for reporter activities. Cypridina luciferase and Gaussia luciferase activities were measured with the BioLux Cypridina Luciferase Assay Kit and BioLux Gaussia Luciferase Assay Kit (NEBL). The activity of the promoter was presented as fold activation of Gaussia luciferase relative to Cypridina luciferase. Three independent experiments were performed at minimum for each experimental condition and the corresponding mean and standard deviation error are represented.
HL-1 cells (6 × 10 5 cells per well) were transfected described above with increasing doses of PITX2c mutant constructs and the wild type PITX2c construct to assess the endogenous transcript expression (ion channels and transcription factors). A 1:1, 1:2 and 1:4 ratio (wt vs. mutant) was assayed and further processed of RNA isolation. Similarly as detailed above, three independent experiments were performed and the corresponding mean values and standard deviation errors were plotted.
In addition to test the rat Nppa and mouse Shox2 promoter, the basal human NPPA promoter was also assayed in HL1 cardiomyocytes. HL-1 cells were seeded in 96-well plates and cultured as described above. Cells were transfected with the proximal NPPA promoter cloned into pLightswitch vector (100 ng; Active Motif, Carlsbad, CA, USA), which contains firefly luciferase as a reporter. Moreover, an empty vector, WT, a dominant negative or mutants of PITX2c were co-transfected (50 ng). All transfections were performed by using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Luciferase activity assays were performed 24 h after transfection using the LightSwitch™ Luciferase Assay Kit (Active Motif, Carlsbad, CA, USA) and a Berthold Luminometer. Luciferase activity was normalized to sample protein concentration. All reporter assays were performed in triplicate.

qRT-PCR Analyses
Transfected cells were collected and total RNA from cultured 3T3 cells was extracted with the TriPure isolation reagent (Roche Diagnostics, Rotkreuz, Switzerland), treated with RNase-free DNase (Roche) for 1 h at 37 • C and purified using a standard phenol-chloroform protocol. RNA yield and purity was determined spectrophotometrically, and RNA integrity was verified by running samples on 1.2% agarose gels and staining with ethidium bromide. Resulting RNA preparations were resolved in nuclease-free water and kept at −80 • C. One microgram aliquots of individual total RNA were reverse transcribed using Maxima First Strand cDNA Synthesis Kit for RT-qPCR (Thermo Scientific) according to the manufacturer's instructions. Negative control reactions were performed in the same conditions without reverse transcriptase.
All qRT-PCR experiments followed MIQE guidelines [39]. Transcript (mRNA) real time PCR experiments were performed with 1 µL of cDNA, SsoFast EvaGreen mix and corresponding primer sets as described on Table 2. All qPCRs were performed using a CFX384TM thermocycler (Bio-Rad) following the manufacturer's recommendations. The relative level of mRNA synthesis was calculated as described by Livak and Schmittgen [40] using housekeeping genes coding for glyceraldehyde-3-phosphate dehydrogenase (Gapdh) and β-D-glucuronidase (Gusb) as internal controls. Each PCR reaction was carried out in triplicate and repeated in at least three distinct biological samples to obtain representative means.

Intracellular Calcium Recordings in HL-1 Myocytes
Cultured atrial HL-1 myocytes used for calcium imaging experiments were expanded and maintained in a Claycomb Medium (JRH Biosciences, Lenexa, KS, USA) supplemented with 10% fetal bovine serum (FBS) (Invitrogen Corporation, Carlsbad, CA, USA), 100 µM norepinephrine, 100 units/mL penicillin, 100 µg/mL streptomycin, and L-Glutamine 2 mM (Sigma Chemical Company, St. Louis, MO, USA) in plastic dishes, coated with 12.5 µg/mL fibronectin and 0.02% gelatin, in a 5% CO 2 atmosphere at 37 • C. HL-1 cells were transfected as described above with PITX2c mutant constructs or the wild type PITX2c construct to assess the impact on intracellular calcium homeostasis. Two days after transfection, cultures were transferred to norepinephrine-free Tyrode solution containing (in mM): NaCl 136, KCl 4, NaH 2 PO 4 0.33, NaHCO 3 4, CaCl 2 2, MgCl 2 1.6, HEPES 10, Glucose 5, pyruvic acid 5, (pH = 7.4) and left for at least three hours before calcium imaging was performed. The experimental protocol consisted in field stimulation of the myocyte cultures at 2x the voltage necessary for eliciting a full calcium transient. The stimulation frequency was gradually increased from 0.5 to 1, 1.3, 2 and 4 HZ at intervals allowing the calcium transient to reach steady-state at each stimulation frequency.
To visualize changes in the intracellular calcium concentration, HL-1 cultures were loaded with 2 µM Cal-590 for 20 min, followed by at least 30 min wash and de-esterification. Confocal calcium images of 512 × 512 pixels were recorded at room temperature at a frame rate of 30 Hz with a resonance-scanning confocal microscope (Leica SP5 AOBS) and a HCX PL APO CS 20.0 × 0.70 IMM UV objective. Cal-590 was excited at 543 nm and fluorescence emission collected between 570 and 750 nm. Laser power was set to 20% of 1 mW and then attenuated to 6% (of the 20%). Fluorescence emission was measured with a Leica Hybrid detector in standard mode with a gain of 30%. Several image fields were recorded in each culture dish. Transfection with the different PITX2c constructs was successfully done in 9 different HL1 cultures. Non-confluent myocyte cultures with low cell density and cultures with poor calcium transients in reference conditions (WT or pcDNA) were discarded.
Image analysis was performed using a custom made program that allowed identification of myocytes with a detectable calcium signal and the beat-to-beat response in the whole image field and in individual myocytes with calcium signals as previously described [41]. Briefly, this program identifies all myocytes with calcium transients and assigns them a number. Subsequently, calcium transients from each myocyte is analyzed and features such as amplitude, duration, and decay time were determined at 0.5 Hz where preparations present a uniform response. In this condition, the error of the measurement of the full duration of the calcium transient at half maximum is less than 7%. The beat-to-beat response is determined by an algorithm based on expert classification of the beat-to-beat response at stimulation frequencies ranging from 0.2 to 4 Hz in more than 1000 individual myocytes from image fields recorded in at least 10 different HL-1 cultures. Unless otherwise stated, whole field responses were analyzed to achieve optimal signal to noise ratio and best signal classification responses were separated into the following categories: Uniform response; alternation of the amplitude with the extreme case being blockade where every second, 2 of 3 or 3 of 4 transients were suppressed; Irregular response; and calcium waves, a response where response to field stimulation was interrupted by spontaneous calcium waves with a duration that extended across several stimulation pulses.

Statistical Analysis
Values presented are expressed as mean ± S.E.M. All comparisons between groups were performed using an unpaired Student's t test or Wilcoxon's rank test for data that did not have a normal distribution. ANOVA was used for multiple comparisons of normally distributed data. Differences were considered statistically significant for p value ≤ 0.05.

Mutation Screening of PITX2c Reveals in AF Patients Revealed Two Novel HD Mutations
PITX2 has been recently associated as candidate gene to underlay AF by GWAS. We have conducted a genetic screening on the PITX2 ORF (NM_000325_5) in 31 patients diagnosed with AF according to the ESC Guidelines for treatment of AF referred to the Cardiology Unit of the Regional Hospital "Ciudad de Jaen". Male patients represented 77% of the enrolled cases. Hypertension was diagnosed in 45% of all patients while type II diabetes was found in 19% of the cases. Among all AF patients 7/31 (~22,5%) were diagnosed as having paroxysmal AF, 10/31 (~32%) had recurrent AF and 12/31 (~39%) permanent AF. The remaining two (2/31;~6,5%) had just shown the first episode of AF. After informed consent was obtained from each patient, a peripheral blood sample was obtained and processed for DNA isolation using standard procedures. Genetic screening of the PITX2 coding sequence identified two nucleotide substitutions at NM_000325.5:c.309G>C. (Gln103His) and NM_000325.5:c.370G>A (Glu124Lys), respectively, corresponding to distinct amino acids forming part of the homeodomain region of the PITX2c transcription factor ( Figure 1A). Importantly, these nucleotide changes lead to amino acid substitution Gln103His and Glu124Lys, respectively. Comparative analyses of the conservation of Gln103His and Glu124Lys demonstrate that these amino acids are highly conserved in different species, increasing thus the likelihood that they might be detrimental ( Figure 1B). Both variants were classified as potentially damaging PITX2c protein function using in silico prediction software MutationTaster (http://www.mutationtaster.org, accessed on 1 February 2020). Both NM_000325.5:c.309G>C. (Gln103His) and NM_000325.5:c.370G>A (Glu124Lys) variants were absent in the control populations of both Exome and 1000 genome projects (https://www.internationalgenome.org/1000-genomes-browsers/, accessed on 1 February 2020). Curiously, both variants were identified in a 65 years old male patient with recurrent AF who did not carry any 4q25 risk variants corresponding at any of the rs2200733, rs13143308 or rs10033464 SNPs. In addition to these two nucleotide substitutions, three additional nucleotide substitutions were identified in the 5´untranslated region of the PITX2c transcription factor ( Figure 1C), corresponding to the following positions: NM_000325.5:g.572C>A, NM_000325.5:g.601G>A and NM_000325.5:g.552A. C572A nucleotide substitution was identified in a 70 years old male patient with permanent FA, type II diabetes but no hypertension and with the presence of risk variants at rs13143308 and rs10033464 but not at rs2200733. G601A was identified in a 65 years old male patient with the first episode of AF, without hypertension or type II diabetes. Risk variants at rs13143308 and rs2200733 but not at rs10033464 were identified in this patient. Finally, G552A variant was identified in a 78 years old male AF patient with paroxysomal AF, hypertension but not type II diabetes. This patient carried no risk variants for any of the three AF-associated SNPs. NM_000325.5:g.601G>A and NM_000325.5:g.552A variants, but no NM_000325.5:g.572C>A, were classified as potentially damaging PITX2c protein function (plausible splice site alterations) using in silico MutationTaster prediction software.

Gln103His and Glu124Lys Mutations Impairs PITX2c Transactivation Activity
In order to test if the identified nucleotide substitutions in the homeodomain of the PITX2c transcription factor have a functional impact, we constructed site-directed mutants carrying the corresponding Gln103His and Glu124Lys mutations alone or in combination. In addition, we also engineered a dominant-negative PITX2c (DN) mutant by selective deletion of the homeodomain region. Co-transfection of these plasmids with the rat Nppa-pGLuc construct demonstrated that wild-type PITX2c is capable of transactivating 5 fold the Nppa basal promoter, in line with previous reports [32,33], while PITX2c DN reduced by half such transactivation capacity (Figure 2A). Importantly, constructs carrying Gln103His (MD4) and Glu124Lys (MD5) mutations lead to 50% and 40% reduction in the transactivation capacity, respectively. Interestingly, PITX2c constructs carrying both mutations (MD4+5) display similar impaired transactivation capacity of the Nppa-pGLuc promoter as those carrying individual mutations.
To add further support to this data, we conducted luciferase gene expression reporter assays by using HL-1 cells transfected with the proximal promoter region of the human atrial natriuretic peptide. Transfection of PITX2c WT (50 ng) resulted in a 2.5-fold increase of the luciferase activity compared to that measured in the presence of an empty vector (Control) ( Figure 2B). The specificity of the PITX2c effect was supported by the absence of changes produced by the PITX2c dominant negative. Interestingly, MD4, MD5 or MD4+5 constructs failed to modify luciferase activity confirming that these are loss-offunction mutations.
To further validate the transcriptional impairment of these novel PITX2 HD mutations, we performed co-transfection assays using the Shox2 promoter ( Figure 2C). As previously reported, wild-type PITX2c over-expression represses the Shox2 promoter [11]. Interestingly, both Gln103His (MD4) and Glu124Lys (MD5) mutations leads to enhanced activation of the Shox2 promoter as compared to wild type PITX2c co-transfections, which is more pronounced for the MD5 as compared to MD4 constructs. In line with these findings, double mutants (MD4+5) display similar transactivation properties as the MD5 construct.
The Nppa promoter is modulated not only by PITX2c but also by other cardiac-enriched transcription factors, such as Gata4 and Nkx2.5 [33]. We therefore tested if Gln103His and Glu124Lys mutations also impaired the synergistic transactivation of PITX2c with these transcription factors ( Figure 2D-F). Co-transfection experiments demonstrate that Gln103His and Glu124Lys mutations impaired Nppa promoter transactivation within a similar range. Interestingly, double mutants also displayed similar transactivation impairment in both conditions, when co-transfected with either Nkx2.5 or Gata4.
Overall these data suggest that both novel PITX2 mutations alter the transactivation properties of PITX2c similarly when acting as a transcriptional activator, whereas some differences occur in cases where PITX2c acts as a transcriptional repressor. Furthermore, these assays indicate that no synergistic effects are observed by the presence of both mutations within the same gene.

Gln103His and Glu124Lys PITX2 Mutations Alter Calcium Homeostasis
To further investigate if these novel PITX2 mutations can impair endogenous expression levels and function of PITX2 downstream genes, we performed co-transfection assays of wild-type PITX2c with distinct doses of the mutated PITX2c constructs within HL-1 atrial cardiomyocytes. We have previously reported that PITX2c impairment in mice leads to changes in ion channel expression and calcium homeostasis and we therefore used confocal calcium imaging in HL1 myocyte cultures to test how the mutated PITX2c constructs affect the properties and stability of the calcium transients elicited by electrical field stimulation.   D)), PITX2c MD5 (panel (E)) and PITX2c MD4+5, significantly impaired their transactivation capacity on the rat Nppa promoter (panel (F)). Experiments were carried out in triplicate and repeated in at least three distinct biological samples. # p < 0.05 statistical significance as compared to controls. * p < 0.05, *** p < 0.001 statistical significance as compared to corresponding PITX2c WT conditions. In the first set of experiments, the effects of several PITX2c constructs on the intracellular calcium transients were examined in HL-1 cardiomyocyte cultures. Transients were examined in cultures transfected with control (pcDNA), MD4, MD5, MD4+5 or DN PITX2c mutants. Figure 3A shows a 450 × 450µm field of HL-1 myocytes co-transfected with GFP (left panel). The intracellular calcium level was monitored simultaneously using the fluorescent calcium indicator Rhod-2 (central panel) and the overlay of the two channels are shown on the right. Calcium transients recorded at different stimulation frequencies are shown in Figure 3B. Because the PITX2c mutants affected the beat-to-beat stability, calcium transient properties were evaluated at 0.5 Hz, where all preparations responded to the electrical pulses. Figure 3C shows that while the calcium transient amplitude tended to decrease for the PITX2c mutants, the reduction was only significant for the MD5 and the DN mutant. The calcium transient duration at half maximum was not significantly affected by any of the mutants ( Figure 3D). To test if differences in the calcium transient amplitude were caused by alterations in SR calcium homeostasis, we first used a rapid caffeine application to release the SR calcium content into the cytosol. Figure 4A, shows the caffeine induced calcium transient and Figure 4B compares the peak amplitude for the different PITX2c constructs. Interestingly, all PITX2 mutants except the MD4 mutant displayed reduced SR calcium content compared with controls (pcDNA). These differences did not appear to be related to differential modulation of the Serca2 (Atp2a2) or phospholamban (Pln) mRNA levels by the different PITX2 mutants since there was a robust and parallel down-regulation of both Serca2 and Pln levels for MD4, MD5 and MD4+5 ( Figure 4C-E), suggesting that the calcium-dependency of the uptake rate is unaffected by the mutants while the maximal uptake rate may be blunted in all three mutants. Interestingly, the SR calcium buffering protein, calsequestrin (Casq2), was also reduced by more than 50% in the MD4+5 mutant ( Figure 4E).

Impact of Gln103His and Glu124Lys PITX2c Mutations on the Beat-to-Beat Response
While these alterations in the SR calcium content and calcium transient amplitude are expected to modulate shortening of the cardiomyocytes, these features may also affect the beat-to-beat response when the stimulation frequency is increased. As shown in Figure 5 a stepwise increase in the stimulation frequency from 0.5 to 4 Hz modified the calcium transients in the whole image field (panel 5B) as well as in individual myocytes (panel 5C-D). Comparison of the responses of individual myocytes within a field and the whole field response revealed that a uniform response could be maintained even though some individual myocytes did not respond uniformly at a given stimulation frequency (compare Figure 5B and 5C). Comparison of the whole field response for cultures transfected with wild type (WT) PITX2c and controls (pcDNA) showed that the intrinsic beating rate without electrical simulation and the ability to maintain a uniform response when the stimulation frequency was increased was similar (p = 0.22; ANOVA). Analysis of unstimulated cultures revealed no significant differences among different mutants. Inspection of the whole field beat-to-beat response of the different PITX2 mutant constructs upon electrical stimulation ( Figure 6) revealed that the fraction of uniform responses was similar for cultures transfected with MD5 and pcDNA (p = 0.65; ANOVA) while MD4 (p < 0.001), MD4+5 (p = 0.01) and DN (p = 0.002) mutants all reduced the fraction of myocytes that were able to maintain a uniform beat-to-beat response. Fitting the fraction of uniform responses versus the stimulation frequency, using the Hill equation, is shown in the right hand panels of Figure 6A and fits are superimposed in Figure 6B with indication of the frequency where 50% of the preparations had a uniform response. The slope factor for the fits was near -4 and did not vary significantly for the different constructs. Figure 6C compares this frequency for the different PITX2c constructs. Analysis of the whole field non-uniform beat-to-beat responses revealed that MD4, MD5, MD4+5 and DN all presented a strong increase in irregular beat-to-beat patterns that often contained spontaneous calcium waves ( Figure 7A,B). Interestingly, all constructs except the MD5 mutation showed alternating beat-to-beat responses or blockade ( Figure 7C,D) that were most prominent at intermediate stimulation frequencies (1.3-2 Hz), suggesting that the underlying mechanisms may be different for the MD4 and MD5 variants.

Discussion
Since atrial fibrillation was first associated with SNP risk variants at chromosome 4q25 that are located near PITX2 [2], research efforts have been undertaken to establish links between 4q25 risk variants and PITX2c activity [42,43], between PITX2c expression levels and AF [9,10,42,44] and to identify PITX2-mediated changes in the expression or activity of molecular mechanisms linked to AF [9,10,13]. Although it has been documented that the 4q25 variant rs2595104 reduces Pitx2 expression [43], the molecular events linking 4q25 risk variants, Pitx2 levels and AF still needs further clarification [42,44]. On the other hand, it has been documented that AF is associated with decreased Pitx2 expression [9] and that PITX2c loss of function predisposes to atrial arrhythmogenesis [9][10][11][12] and disturbances in the calcium homeostasis [13]. Moreover, the 4q25 risk variant rs13143308 has recently been linked to defective calcium homeostasis [14], pointing to PITX2 as a candidate gene for AF associated with defective calcium homeostasis. The present study explores further the links between PITX2c function and AF, affording new evidence that mutations in the PITX2 HD identified in patients with AF, affects altogether PITX2c activity, calcium homeostasis and the beat-to-beat stability when expressed in HL1 cardiomyocyte cultures.

Impact of AF Associated PITX2 Homeodomain Mutants on PITX2c Function
Mutants identified in patients with AF and the MD4 and MD5 mutants caused amino acid substitutions are highly conserved region of the PITX2c homeodomain, increasing thus the likelihood that they might be detrimental. Indeed, both HD-variants were classified as potentially damaging. Interestingly, the patient carrying the MD4 and MD5 mutations had no 4q25 risk SNPs at rs2200733, rs13143308 or rs10033464, suggesting that the observed loss of PITX2 function is unrelated to 4q25 risk SNPs, which have previously been proposed to modify Pitx2 function [2,43]. Furthermore our data demonstrate that both novel PITX2 mutations alter the transactivation properties of PITX2c, both alone and in combination with cardiac-enriched transcription factors, such as Nkx2.5 and/or Gata4. Curiously, PITX2c transactivation impairment is rather similar between distinct mutations (MD4, MD5) when acting as a transcriptional activator (Nppa promoter), whereas some differences occur when PITX2c acts as a transcriptional repressor (Shox2 promoter). Furthermore, our data also demonstrate that no synergistic effects are observed by the presence of both mutations (MD4+5).

Impact of the PITX2 Mutants on Intracellular Calcium Homeostasis
The impact of loss of PITX2c function on ion channel expression and calcium homeostasis has previously been addressed in a mouse model of conditional atrial-specific PITX2c deletion [9,13]. Interestingly, we here found that the MD5, MD4+5 and DN mutants all reduced that SR calcium content while the content was similar for WT, pcDNA and the MD4 mutant. This observation is opposite to the mouse model where the SR calcium content was twice as high in left atrial myocytes from Pitx2 -/than from WT mice [13]. Similarly, Serca2, Pln and Casq mRNA levels were all increased in both left and right atria from Pitx2 -/mice [13] while we here observe a parallel reduction in Serca2 and Pln levels in MD4, MD5 and MD4+5 mutants as well as a reduced Casq level in the MD4+5 mutant. Nevertheless, the observed changes in SR calcium content are consistent with the observed differences in the calcium transient amplitude, which is reduced in the MD5 but not the MD4 variant. On the other hand, previous studies have associated paroxysmal AF [45] loss of PITX2c function [13] or 4q25 risk variants [14] with increased Serca2 expression and SR calcium content as well as a higher incidence of calcium waves, suggesting that another mechanism might be linking the PITX2 mutations to AF. In this regard, ventricular and atrial arrhythmia have also been associated with irregularities in the beat-to-beat response such as action potential and calcium transient alternans [46][47][48][49].

Impact of the PITX2 Mutants on the Rate-Dependency of the Beat-To-Beat Response
To test the latter hypothesis, we determined how the PITX2 mutations affected the beatto-beat response of the calcium transient when cardiomyocyte cultures were subjected to increasing stimulation frequencies. Interestingly, only the MD4, MD4+5 and DN constructs induced a significant reduction in the fraction of uniform responses between 0.5 and 4 Hz, lowering the frequency where the fraction of uniform responses was reduced to 50%. Further analysis of the non-uniform beat-to-beat responses revealed that all PITX2c mutants significantly increased the incidence of irregular beat-to-beat responses while the MD5 mutation repressed alternating responses or blockade observed for the other PITX2c constructs as well as for cultures transfected with pcDNA. Considering that all mutants induce loss of PITX2c function (Figure 2A,B) while only the MD5 completely impairs PITX2c-dependent repression of Shox-2, it is conceivable that PITX2c insufficiency underlies the observed increase in irregular beat-to-beat responses in all PITX2c mutants. On the other hand, the absence of alternating responses in MD5 mutants might be related to the inability of this variant to repress Shox-2 activity. Because both MD4 and MD5 variants are associated with AF, these finding either suggests that the promotion of irregular beat-to-beat responses is the key mechanisms contributing to the development of AF in the carrier of the MD4 and MD5 mutations. In support of this notion, human atrial myocytes from patients with AF have previously been shown to present a higher incidence of irregular beat-to-beat responses than myocytes from patients without AF [41]. Alternatively, several mechanisms may be acting in parallel to promote AF in the carrier of these mutations. In this regard, alternating responses have previously been linked to a high incidence of spontaneous calcium release events in human atrial myocytes [50], and RyR2 loss of function mutations causing severe ventricular arrhythmia have recently been linked to RyR2 inactivation and premature induction of calcium alternans [46,47].
In conclusion, we have identified two point mutations in the homeodomain of PITX2 and three polymorphisms in the 5´untranslated region. Moreover, we document that the PITX2 HD point mutations cause loss of PITX2c function, impair Shox2 promoter repression as well as co-transactivation with other cardiac-enriched co-factors such as Gata4 and Nkx2.5, pointing to a pivotal role for these mutations for correct PITX2c function. The mutations also impaired endogenous expression of calcium regulatory proteins and alterations in sarcoplasmic reticulum (SR) calcium accumulation that favored alternating and irregular calcium transient amplitudes, causing deterioration of the beat-to-beat stability upon elevation of the stimulation frequency. Together, this data demonstrate that these novel PITX2 mutations may contribute to the development of atrial fibrillation in the patient carrying them.  Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee at Hospital Regional de Jaén.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: Not applicable.