Ion channel-kinase TRPM7 is required for maintaining cardiac automaticity

Significance

Transient Receptor Potential Melastatin 7 (TRPM7) is a divalent-permeant channel-kinase of unknown function expressed in human atrial myocytes and fibroblasts and recently implicated in atrial arrhythmias. We show that TRPM7 is highly expressed in embryonic myocardium and sinoatrial node (SAN). Trpm7 disruption in vitro, in cultured embryonic cardiomyocytes, and in vivo in zebrafish and in mice impairs cardiac automaticity. We show that this occurs via reductions in Hcn4 mRNA and the pacemaker current, I_f, in SAN. We conclude that TRPM7 influences diastolic membrane depolarization and automaticity in SAN via regulation of Hcn4 expression.

Abstract

Sick sinus syndrome and atrioventricular block are common clinical problems, often necessitating permanent pacemaker placement, yet the pathophysiology of these conditions remains poorly understood. Here we show that Transient Receptor Potential Melastatin 7 (TRPM7), a divalent-permeant channel-kinase of unknown function, is highly expressed in embryonic myocardium and sinoatrial node (SAN) and is required for cardiac automaticity in these specialized tissues. TRPM7 disruption in vitro, in cultured embryonic cardiomyocytes, significantly reduces spontaneous Ca²⁺ transient firing rates and is associated with robust down-regulation of Hcn4, Ca_v3.1, and SERCA2a mRNA. TRPM7 knockdown in zebrafish, global murine cardiac Trpm7 deletion (KO^αMHC-Cre), and tamoxifen-inducible SAN restricted Trpm7 deletion (KO^HCN4-CreERT2) disrupts cardiac automaticity in vivo. Telemetered and sedated KO^αMHC-Cre and KO^HCN4-CreERT2 mice show episodes of sinus pauses and atrioventricular block. Isolated SAN from KO^αMHC-Cre mice exhibit diminished Ca²⁺ transient firing rates with a blunted diastolic increase in Ca²⁺. Action potential firing rates are diminished owing to slower diastolic depolarization. Accordingly, Hcn4 mRNA and the pacemaker current, I_f, are diminished in SAN from both KO^αMHC-Cre and KO^HCN4-CreERT2 mice. Moreover, heart rates of KO^αMHC-Cre mice are less sensitive to the selective I_f blocker ivabradine, and acute application of the recently identified TRPM7 blocker FTY720 has no effect on action potential firing rates of wild-type SAN cells. We conclude that TRPM7 influences diastolic membrane depolarization and automaticity in SAN indirectly via regulation of Hcn4 expression.

Sinus node dysfunction and atrioventricular node block (AVB) are common causes of bradyarrhythmias in patients, often requiring treatment with permanent pacemakers (1). However, the pathophysiology of sinus node failure and the physiology underlying sinoatrial node (SAN) automaticity remains incompletely understood. The currently held paradigm is that cardiac automaticity arises from the integrated activity of voltage-gated ionic currents (Hcn2/Hcn4, Ca_v3.1, Ca_v1.3), transporters (NCX), and sarcoplasmic reticulum (SR) Ca²⁺ release (2–4). In addition to these “classic” ion currents, some members of the transient receptor potential (TRP) superfamily of ion channels are also expressed in myocardium (TRPC1/3/4/6, TRPM4, TRPM7) (5, 6), but the contribution of these channels to myocardial function remains relatively unexplored.

Among these TRP channels, TRPM7 is especially abundant in both human and murine heart (5, 6) and is concentrated in myocardium during embryonic development (7). TRPM7, and its homolog TRPM6, are unique in that they are ion channels containing a carboxyl terminal kinase of unknown function. TRPM7 channel is divalent-permeant and forms an outwardly rectifying current that is inhibited by both cytoplasmic and extracellular Mg²⁺ (8). TRPM7 current was recently shown to be up-regulated in human atrial fibroblasts in the context of atrial fibrillation and was hypothesized to provide a Ca²⁺ influx pathway inducing TGF-β1–mediated fibroblast proliferation and differentiation, thereby contributing to atrial fibrosis in the pathogenesis of atrial fibrillation (9). In addition to atrial fibroblasts, ventricular fibroblasts also have a large TRPM7 current (10), and a TRPM7-like current has been recorded from human atrial myocytes (11). Thus, TRPM7 represents a heretofore unstudied ionic current and signaling molecule in cardiac biology that may participate in arrhythmogenesis in human heart disease or contribute to normal myocardial Ca²⁺ signaling.

In this study we show that TRPM7 is required for maintaining cardiac automaticity. TRPM7 current is largest in myocardial cells that exhibit automaticity, such as cultured embryonic ventricular cardiomyocytes (EVM) and isolated SAN cells, compared with the quiescent adult ventricular cardiomyocyte (AVM). Trpm7 deletion in both EVM and SAN slows spontaneous Ca²⁺ transient frequency, thereby disrupting cardiac automaticity in vitro. In vivo, Trpm7 loss of function slows heart rate in embryonic zebrafish and induces sinus pauses (SPs) and AVB in both global cardiac-targeted Trpm7 knockout mice (KO^αMHC-Cre) and in tamoxifen-inducible SAN/atrioventricular node cell (AVN)-restricted Trpm7 knockout mice (KO^Hcn4-CreERT2). We show that these effects are associated with shallowing of the diastolic depolarization slope (DDSL) due to TRPM7-dependent down-regulation of Hcn4 and pacemaker current, I_f.

Results

Trpm7 Deletion in Vitro Disrupts Cardiac Automaticity in Cultured Murine Embryonic Cardiomyocytes.

We previously showed that TRPM7 is predominantly expressed in the heart of embryonic day (E)9.5 embryos (7) and becomes ubiquitously expressed in later embryonic development (7) and into adulthood (5, 6). Because TRPM7 is first expressed in the developing embryonic heart (7), we examined TRPM7 current in cultured EVM isolated from E13.5–14 Trpm7^fl/fl (7) embryos and compared these with TRPM7 current in freshly dissociated AVM. A large TRPM7-like current is activated in EVM (I_Trpm7,EVM = 64.2 ± 16.7 pA/pF at +100 mV, n = 5), ∼eightfold larger than AVM (I_Trpm7,AVM = 8.3 ± 0.9 pA/pF, n = 5, P < 0.05; Fig. 1 A and B) under whole-cell conditions. These TRPM7-like currents are initially absent on “break-in” but then run up over the course of 10 min of cell dialysis to a steady-state level (Fig. S1). To confirm that these TRPM7-like currents are indeed TRPM7, we used Cre-loxP technology to genetically ablate Trpm7 in murine EVM. The conditional Trpm7 allele contained loxP sites flanking exon 17, and Cre-mediated recombination induces a frame shift that prevents expression of the ion channel and kinase domains of TRPM7 (7). We introduced Cre to cultured Trpm7^fl/fl EVM using adenoviruses (Ad) that drive Cre expression under either cytomegalovirus (CMV: ubiquitous) or troponin-T (TnT: cardiac-specific) promoters (Ad-CMV-Cre/Ad-TnT-Cre). An adenovirus expressing β-galactosidase (Ad-CMV-Lacz) served as a control. Using EVM isolated from E14 ROSA26^mTmG embryos, we confirmed high-efficiency Cre transduction in EVM (>95%) by both Ad-CMV-Cre and Ad-TnT-Cre (Fig. 1C, only Ad-TnT-Cre shown). Genetic ablation of Trpm7 exon 17 in Trpm7^fl/fl EVM is evident by PCR within 2 d after adenoviral transduction with both Ad-TnT-Cre (Fig. 1D, lane 1) and Ad-CMV-Cre (lane 2) but not with Ad-CMV-Lacz (lane 3), according to the presence of the expected size of the deletion product in EVM genomic DNA. Finally, TRPM7 current is largely abolished in Trpm7^fl/fl EVM treated with Ad-CMV-Cre when patch-clamped 4–5 d after adenoviral transduction (Fig. 1 E and F).

Fig. 1.

TRPM7 current in cultured murine embryonic ventricular myocytes. (A) Representative TRPM7 current measured in cultured EVM compared with AVM after full run-up. (B) Mean TRPM7 current density at +100 mV in AVM (I_{Trpm7, AVM} = 8.3 ± 0.9 pA/pF, n = 5) compared with EVM (I_{Trpm7, EVM} = 64.2 ± 16.7 pA/pF, n = 5). (C) Interference contrast (DIC, Left) and fluorescence GFP image (Right). (D) PCR across exon 17 from genomic DNA isolated from Trpm7^fl/fl EVM transduced with Ad-TnT-Cre (1), Ad-CMV-Cre (2), and Ad-CMV-Lacz (3), Trpm7^fl/fl fibroblasts (+), and Trpm7^fl/- tail (-). Black arrow, full-length exon 17. Red arrow, deleted exon 17. (E) Representative TRPM7 current measured in Trpm7^fl/fl EVM and Trpm7^fl/fl EVM 5 d after transduction with Ad-CMV-Cre. (F) Mean TRPM7 current density in Trpm7^fl/fl EVM treated with Ad-CMV-Cre (n = 3) compared with untreated Trpm7^fl/fl EVM (n = 5). *P < 0.05.

After 4 d in culture, Ad-CMV-Lacz–treated Trpm7^fl/fl EVMs contract rhythmically at a significantly higher rate than either Ad-CMV-Cre- or Ad-TnT-Cre–treated EVM. To quantify this difference, we loaded EVM with Fluo-4AM and measured intracellular Ca²⁺ transients in clusters of Ad-CMV-Lacz-, Ad-CMV-Cre-, and Ad-TnT-Cre–treated Trpm7^fl/fl EVM using high-speed laser scanning confocal microscopy (Fig. 2A). Ca²⁺ transient firing frequency is reduced equivalently ∼twofold in EVM upon deletion of Trpm7 by Ad-CMV-Cre (102 ± 18/min, n = 18, P < 0.01) and Ad-TnT-Cre (103 ± 17/min, n = 13, P < 0.01) compared with Ad-CMV-Lacz controls (206 ± 23/min, n = 18; Fig. 2B). Accordingly, cycle length increased ∼twofold after Trpm7 deletion (Fig. 2C). To control for possible off-target effects of Cre expression, we also compared cycle lengths of Trpm7^fl/+ EVM treated with Ad-CMV-Cre and Ad-TnT-Cre (Fig. S2). We find these no different from the Ad-CMV-Lacz controls. There is a small but statistically significant increase in peak Ca²⁺ transient amplitude upon Trpm7 deletion with Ad-CMV-Cre (F/F_o-_CMV-Cre = 2.2 ± 0.1, n = 19, P < 0.05), but this is not observed in Ad-TnT-Cre (F/F_o-TnT-Cre = 1.7 ± 0.1, n = 12) treated Trpm7^fl/fl EVM compared with Ad-CMV-Lacz controls (F/F_o-CMV-Lacz = 1.8 ± 0.1, n = 19; Fig. 2D). Additionally, Ca²⁺ transient duration is significantly lengthened in Trpm7-deleted EVM by both Ad-CMV-Cre (664 ± 96 ms, n = 20, P < 0.01) and Ad-TnT-Cre (471 ± 59 ms, n = 12, P < 0.05) compared with Ad-CMV-Lacz (291 ± 43, n = 19; Fig. 2E). Thus, TRPM7 deletion seems to slow spontaneous contractions and Ca²⁺ transient firing, in a cell autonomous fashion in cultured embryonic cardiomyocytes.

Fig. 2.

Trpm7 deletion in embryonic myocardium disrupts cardiac automaticity in vitro. (A) Ca²⁺ transients measured by high-speed line-scanning confocal microscopy in Trpm7^fl/fl EVM 4–5 d after transduction with Ad-CMV-Lacz (Top), Ad-CMV-Cre (Middle), and Ad-TnT-Cre (Bottom). (B) Mean Ca²⁺ transient frequency, (C) cycle length, C_L, (D) peak Ca²⁺ transients, F/F_o, and (E) Ca²⁺ transient length in Ad-CMV-Lacz, Ad-CMV-Cre, and Ad-TnT-Cre treated Trpm7^fl/fl EVMs. *P < 0.05, **P < 0.01.

Trpm7 Loss of Function in Vivo Disrupts Automaticity in Zebrafish and Induces Sinus Pauses and AVB in Mouse.

We next examined whether this effect of TRPM7 knockout on embryonic cardiac automaticity is also present in vivo in embryonic zebrafish. We find that the Trpm7 morpholino (MO) zebrafish recapitulate the previously described phenotype of melanocyte deficiency and loss of touch responsiveness (12, 13) compared with water-injected controls. In addition, heart rates are ∼25% lower in Trpm7 MO zebrafish (HR, _{Trpm7 MO} = 80 ± 2 bpm, n = 30 embryos, P < 0.001) vs. controls (HR, _H20 = 107 ± 2 bpm, n = 27 embryos; Fig. 3A), suggesting that the negative chronotropic effect of TRPM7 disruption is not restricted to cultured murine EVM but also occurs in intact zebrafish embryos.

Fig. 3.

Trpm7 deletion in vivo disrupts automaticity in zebrafish and induces SPs and AVB in mice. (A) (Left) Images of zebrafish embryos: water-injected (Upper) and Trpm7 MO-injected (Lower). (Right) Trpm7 MO zebrafish (n = 30) and water-injected zebrafish (n = 27) heart rates. (B) Normal sinus rhythm (p denotes P waves; atrial depolarization) with intact atrioventricular conduction in the ECG of a telemetered, conscious WT mouse. (C) Representative ECG showing an episode of SP observed in a KO^αMHC-Cre mouse. Solid arrows denote location of expected p waves. (D) Representative ECGs demonstrating AVB observed in KO^αMHC-Cre mice (broken black arrow, conducted QRS complexes; broken gray arrow, expected location of QRS complex). (E and F) Box plots with overlying data points showing the distribution of the frequency of (E) SPs and (F) AVB observed over 24 h of telemetric monitoring in WT (n = 7) and KO^αMHC-Cre (n = 8) mice. (G) Mean heart rates of WT and KO^αMHC-Cre over a 24-h period were not statistically different. In box plots, error bars represent the SD of the mean. Box height represents the SE. **P < 0.01, ***P < 0.001.

To determine whether TRPM7 is also important for automaticity in the adult mouse heart we turned to global cardiac TRPM7 knockout mice generated by crossing αMHC-Cre mice with Trpm7^fl/fl and/or Trpm7^fl/- mice (KO^αMHC-Cre). Conscious electrocardiograms recorded over a 24-h period in freely moving telemetered KO^αMHC-Cre mice reveal frequent SPs (Fig. 3 C and E and Fig. S3 B and F) and AVB (Fig. 3 D and F and Fig. S3D) compared with no SP or AVB in WT mice (Fig. 3B and Fig. S3 A and E). KO^αMHC-Cre mice also exhibit SP with atrial bigeminy (Fig. S3 C and F) and ectopic atrial foci (Fig. S3 B and G) in both conscious, telemetered mice as well as in sedated mice. Despite the presence of these bradyarrhythmias in KO^αMHC-Cre mice, the mean heart rates over a 24-h period are not statistically different from WT (Fig. 3G).

TRPM7 Is Highly Expressed in Murine Sinoatrial Nodal Cells and Is Required for Normal SAN Automaticity.

The SPs and AVB observed in KO^αMHC-Cre mice suggest that Trpm7 deletion affects SAN and AVN function, because these are the specialized myocardial cell types that exhibit automaticity in the adult heart. To determine whether this was due to a direct effect of TRPM7 activity in SAN, we next examined TRPM7 current in freshly isolated SAN and assessed whether TRPM7 is efficiently deleted in SAN cells of KO^αMHC-Cre mice. We first confirmed robust Cre expression in all observed SAN cells by crossing αMHC-Cre mice with the mT/mG reporter mouse line (ROSA26^mTmG), in which membrane-targeted green fluorescent protein expression (mG) is induced only after Cre-mediated recombination (14) (Fig. 4A, single SAN shown). Indeed, αMHC-Cre induces recombination in all myocardial cells (15). Next we confirmed deletion of Trpm7 exon 17 from genomic DNA isolated from SAN by PCR in KO^αMHC-Cre mice (Fig. 4B) according to the presence of the expected size of the deletion product. Finally we measured TRPM7 current from isolated SAN (I_{Trpm7, WT SAN} = 32.0 ± 6.2 pA/pF at +100 mV, n = 6) and found them to be ∼fourfold larger than TRPM7 in AVM (I_{Trpm7, AVM} = 8.3 ± 0.9 pA/pF, n = 5) and absent in KO^αMHC-Cre SAN (I_{Trpm7, KO SAN} = 0.4 ± 0.2 pA/pF, n = 4, P < 0.01; Fig. 4 C and D). Similar to EVM, the initial TRPM7 current on break-in before SAN dialysis is negligible and only activates fully after ∼10 min of cell dialysis (Fig. S4). Thus, measuring TRPM7 current in SAN, under these conditions, serves primarily as an assay to quantify the amount of functional TRPM7 channels in the membrane of a given cell, as opposed to showing the active current present under physiological conditions.

Fig. 4.

TRPM7 is highly expressed in murine SAN. (A) Confocal images of an isolated SAN freshly dissociated from adult αMHC-Cre-ROSA26^mTmG heart: differential interference contrast (Left) and fluorescence GFP image (Right). (Scale bar, 50 μm.) (B) PCR across exon 17 from genomic DNA isolated from dissected KO^αMHC-Cre and WT SAN. Tail DNA from Trpm7^fl/- (-) serves as a positive control for Trpm7 exon 17 deletion. Black arrow, full-length exon 17. Red arrow, deleted exon 17. (C) Representative TRPM7 current-voltage traces measured in WT SAN, WT AVM, and KO^αMHC-Cre SAN. (D) Mean TRPM7 current density in WT SAN (I_{Trpm7, WT SAN} = 32.0 ± 6.2 pA/pF, n = 6), WT AVM (I_{Trpm7, AVM} = 8.3 ± 0.9 pA/pF, n = 5), and SAN (I_{Trpm7, KOαMHC-Cre SAN} = 0.4 ± 0.2 pA/pF, n = 4, P < 0.01).

As with EVM, we next assessed automaticity in WT (Fig. 5A) and KO^αMHC-Cre SAN (Fig. 5B) using confocal microscopy to measure spontaneous Ca²⁺ transients. Ca²⁺ transient frequency is significantly diminished in KO^αMHC-Cre SAN (Ca²⁺ transient rate _KO = 32 ± 5 /min, n = 34) compared with WT (Ca²⁺ transient rate _WT = 72 ± 11, n = 26, P < 0.01; Fig. 5C), and Ca²⁺ transients are lengthened (Fig. 5D), as observed in TRPM7-depleted EVM (Fig. 2). Isoproterenol (ISO) increases the Ca²⁺ transient frequency in both WT and KO^αMHC-Cre cells, but the maximal rate reached by KO^αMHC-Cre cells in ISO remains slower than in WT counterparts. Similar results, although less marked, are obtained by measuring the frequency of Ca²⁺ transients in individual pacemaker cells within the intact, undissociated SAN upon TRPM7 deletion (Fig. S5A). Despite these clear differences in Ca²⁺ transient firing frequency, peak Ca²⁺ transients are no different in KO^αMHC-Cre SAN compared with WT (Fig. S5B), consistent with unchanged SR Ca²⁺ load as assessed by 10 mM caffeine application (Fig. S5C).

Fig. 5.

Trpm7 deleted SAN exhibit impaired automaticity with slowed diastolic Ca²⁺ rise. Ca²⁺ transients measured by high-speed line-scanning confocal microscopy in (A) WT SAN and (B) KO^αMHC-Cre SAN under basal conditions (Top) and after stimulation with ISO: 2 nM (Middle) and 100 nM (Bottom). (C) Mean Ca²⁺ transient frequency and (D) Ca²⁺ transient length in WT SAN compared with KO^αMHC-Cre SAN under basal conditions and after stimulation with 2 nM and 100 nM ISO. (E) Representative Ca²⁺ transients from WT (Left) and KO^αMHC-Cre (Right) showing the slope of diastolic Ca²⁺ rise (Ca²⁺ ramp). (F) Mean slope of the diastolic Ca²⁺ ramp in WT and KO^αMHC-Cre under increasing ISO concentrations. *P < 0.05, **P < 0.01.

Cardiac automaticity is determined by diastolic depolarization arising from the combined effects of voltage-gated ion channels (Hcn2/Hcn4/Ca_v3.1/Ca_v1.3) and intracellular Ca²⁺ cycling (diastolic Ca²⁺ and Ncx2) (3). Thus we next measured the rate of rise of diastolic Ca²⁺ leading up to each Ca²⁺ transient and found this to be significantly blunted in KO^αMHC-Cre SAN (Fig. 5E, Right) compared with WT (Fig. 5E, Left) and also only minimally responsive to ISO stimulation (Fig. 5F). These data suggest that TRPM7 is required for maintenance of normal automaticity in murine SAN and that it can influence diastolic Ca²⁺ release, either directly or indirectly, and thereby contribute to diastolic depolarization in SAN (3).

DDSL Is Diminished in KO^αMHC-Cre SAN Cells: Hcn4 and I_f Are Down-Regulated.

To further examine the mechanism by which Trpm7 deletion impairs SAN cell automaticity we measured spontaneous action potentials (APs) in freshly isolated SAN using the perforated patch-clamp technique. Consistent with the Ca²⁺ imaging experiments, AP firing rates are significantly diminished in KO^αMHC-Cre SAN compared with WT SAN (Fig. 6 A and B) and largely reversed by ISO. By measuring membrane potential, we quantified the DDSL (Fig. 6C), an important determinant of SAN firing rate (4), in WT and KO^αMHC-Cre SAN. DDSL was significantly lower in KO^αMHC-Cre (DDSL _KO = 0.03 ± 0.01 mV/ms, n = 5, P < 0.05) compared with WT (DDSL _WT = 0.09 mV/ms ± 0.01, n = 7), and this difference was reversed by 100 nM ISO (Fig. 6 C and D and Table S1).

Fig. 6.

Trpm7 deletion slows spontaneous AP firing and diastolic depolarization in murine SAN. (A) Spontaneous APs recorded in WT SAN (Upper) and KO^αMHC-Cre SAN (Lower) under basal conditions (Left), 2 nM ISO (Center), and 100 nM ISO (Right). (B) Mean AP firing rate. (C) Representative AP from WT SAN compared with KO^αMHC-Cre SAN showing slowed DDSL in KO^αMHC-Cre SAN. (D) Mean DDSL in WT SAN compared with KO^αMHC-Cre SAN under basal conditions and with 2 nM and 100 nM ISO. *P < 0.05, **P < 0.01.

Rising diastolic Ca²⁺ is thought to contribute, in part, to diastolic depolarization in SAN via forward mode Na⁺-Ca²⁺ exchange current and calcium-induced Ca²⁺ release (3, 16), but other membrane currents, including the hyperpolarization-activated pacemaker current (I_f, encoded by Hcn1/Hcn2/Hcn4), T-type Ca²⁺ current (I_Ca,T, encoded by Ca_v3.1), and L-type Ca²⁺ current (I_Ca,L, encoded by Ca_v1.3/Ca_v1.2) (2–4, 17) also participate. In fact, acute blockade of I_f alone in isolated SAN, using the selective I_f blocker ivabradine (18), was recently shown to be capable of slowing intracellular Ca²⁺ cycling kinetics and prolonging the period of spontaneous local Ca²⁺ releases occurring during diastolic depolarization (19). Thus, much of what we observe with respect to Ca²⁺ cycling in KO^αMHC-Cre SAN may be explained via a mechanism of Hcn4 and I_f reduction. Accordingly, we examined the change in relative expression levels of a panel of genes implicated in cardiac automaticity, both in vitro in cultured Trpm7-deleted EVM as well as in dissected SAN from WT and KO^αMHC-Cre hearts. In Trpm7-deleted EVM, Hcn4, Ca_v3.1, and SERCA2a mRNA are robustly down-regulated eightfold, sixfold, and sixfold respectively, relative to WT ECM, whereas Trpm4, Hcn2, and Ca_v1.3 are not significantly differentially expressed (Fig. 7A). As expected, Trpm7 mRNA is 19-fold reduced in these Trpm7-deleted cells. Similarly, in SAN dissected from KO^αMHC-Cre hearts, Hcn4 mRNA is reduced twofold, and Trpm7 mRNA is threefold reduced relative to WT SAN. However, neither Hcn2, Trpm4, Ca_v3.1, Ca_v1.3, nor Serca2a mRNA are significantly down-regulated in KO^αMHC-Cre SAN (Fig. 7B). We next measured the hyperpolarization-activated current, I_f, in freshly isolated SAN to establish whether these reductions in Hcn4 mRNA indeed result in diminished functional I_f current (Fig. 7 C and D). We find I_f to be on average significantly diminished in KO^αMHC-Cre SAN (I_{f, KO}α_MHC-Cre = −11.9 ± 2.6 pA/pF @ −160 mV, n = 27) compared with WT (I_{f, WT} = −28.7 ± 6.6 pA/pF @ −160 mV, n = 14, P < 0.01). Furthermore, functional mosaicism is observed in which some KO^αMHC-Cre SAN have virtually no I_f (9/27 = 33% < −3 pA/pF at −160 mV), whereas other KO^αMHC-Cre SAN have nearly normal I_f current densities (Fig. 7E). On the other hand, no WT SAN (0 of 14) have I_f current densities below 3 pA/pF.

Fig. 7.

Hcn4 mRNA and pacemaker current are down-regulated upon Trpm7-deletion in EVM and SAN cells. mRNA expression profile of genes implicated in automaticity in (A) cultured Trpm7-deleted EVM (n = 10 samples: 6 Ad-TnT-Cre and 4 Ad-CMV-Cre) relative to WT EVM (n = 5 samples: Ad-Lacz-Cre) and in (B) KO^αMHC-Cre (n = 4 samples, 4–6 pooled SAN/sample) relative to WT SAN (n = 3 samples, 4–6 pooled SAN/sample). (C) Representative hyperpolarization-activated “funny” current traces, I_f, from WT SAN (Middle) and KO^αMHC-Cre SAN (Bottom). Voltage protocols are shown above current traces. (D) I_f current–voltage relationship from WT SAN (n = 14) compared with KO^αMHC-Cre SAN (n = 27). (E) Box plots with overlying data points showing the mean and distribution of I_f current densities at −160 mV from WT SAN (n = 14) compared with KO^αMHC-Cre SAN (n = 27). (F) Representative Ca²⁺ current traces of total Ca²⁺ (I_CaT+L, Left) and L-type Ca²⁺ current (I_Ca,L, Right) from WT SAN (Middle) and KO^αMHC-Cre SAN (Bottom). Voltage protocols are shown above current traces. (G) Current–voltage relationship of T-type Ca²⁺ currents (I_Ca,T: obtained by subtracting I_Ca,L from I_CaT+L) from WT SAN (n = 6) compared with KO^αMHC-Cre SAN (n = 10). (H) Box plots with overlying data points showing the mean and distribution of heart rates after ivabradine (Left: 3 mg/kg; Right: 6 mg/kg) in WT mice (n = 5) compared with KO^αMHC-Cre mice (n = 7). In box plots error bars represent the SD of the mean. Box height represents the SE. In current–voltage relationships, error bars represent SEs. Statistical significance in A and B are based on ΔC_t values (C_t of gene − C_t GAPDH) and SD for each gene in Trpm7-deleted or KO compared with WT, using an unpaired Student t test. *P < 0.05, **P < 0.01, ***P < 0.001.

Because Ca_v3.1 mRNA is also significantly down-regulated upon Trpm7 deletion in cultured EVM but not in isolated KO^αMHC-Cre SAN, we examined whether L-type Ca²⁺ current (I_Ca,L) or T-type Ca²⁺ current (I_Ca,T) are altered in isolated KO^αMHC-Cre SAN, because these currents are also thought to contribute to both diastolic Ca²⁺ influx and diastolic depolarization in SAN. We measured I_Ca,T by subtracting I_Ca,L (Fig. 7F, Right) from total Ca²⁺ current, I_Ca,T + I_Ca,L (Fig. 7F, Left) and generated the current–voltage relationship for I_Ca,T shown in Fig. 7G. I_Ca,L, I_Ca,T + I_Ca,L, and I_Ca,T are not significantly different in KO^αMHC-Cre compared with WT SAN at all membrane potentials examined. Collectively these data suggest that the negative chronotropic effects of Trpm7 deletion in SAN are mediated largely via down-regulation of Hcn4 and I_f, which decreases the slope of diastolic depolarization and thereby slows automaticity. Consistent with this notion, KO^αMHC-Cre mice are less sensitive than WT counterparts, in a dose-dependent fashion, to the negative chronotropic effects of the ivabradine (Fig. 7H). Furthermore, the recently identified TRPM7 blocker FTY720 has no effect on SAN firing rates and AP durations when acutely applied to WT SAN cells at a concentration shown to nearly completely block endogenous TRPM7 in human atrial fibroblasts (500 nM) (20) (Fig. S6).

Postdevelopmental, SAN-Restricted Trpm7 Deletion Recapitulates the Phenotype Observed in KO^αMHC-Cre Mice.

Because αMHC-Cre is expressed and recombines in cardiomyocytes as early as E12–14 (21), it is possible that these findings in adult KO^αMHC-Cre SAN result from developmental effects of Trpm7 on SAN maturation or differentiation, consistent with our previous findings in lymphocytes (7) and embryonic tissues (15, 22). Furthermore, αMHC-Cre deletes globally in heart, so it is also conceivable that KO^αMHC-Cre SAN are affected in a paracrine fashion by Trpm7 deletion in neighboring, non-SAN cells. To address these alternative mechanisms, we crossed Hcn4-CreERT2 mice (23) with Trpm7^fl/fl or Trpm7^fl/- mice to generate a line of mice (KO^Hcn4-CreERT2) that provides tamoxifen (Tm)-inducible, SAN/AVN-restricted Trpm7 deletion (Fig. 8A). Using Hcn4-CreERT2-ROSA26^mTmG mice, we demonstrate efficient recombination that is restricted to the SAN region (green, mGFP) in 6-wk-old Hcn4-CreERT2-ROSA26^mTmG mouse right atrium (red, mTomato) (Fig. 8B) after Tm gavage (40 mg/kg) × 4 d (Fig. 8A). In addition, genomic DNA isolated from KO^Hcn4-CreERT2 SAN reveals efficient deletion of Trpm7 exon 17 according to the presence of the expected size of the deletion product on PCR (Fig. 8C). Finally, TRPM7 current measured by whole-cell patch clamp of KO^Hcn4-CreERT2 SAN is eliminated (I_{Trpm7, KO,Hcn4-CreERT2 SAN} = 0.1 ± 0.1 pA/pF, n = 4, P < 0.01; Fig. 8 D and E). Having established effective, postdevelopmental, SAN-restricted Trpm7 deletion in KO^Hcn4-CreERT2 mice, we next performed telemetric studies to assess for bradyarrhythmias, as observed in KO^αMHC-Cre mice. Strikingly, KO^Hcn4-CreERT2 mice also exhibit SPs with ectopic atrial escape rhythms (Fig. 8 F and G, Left) as well as AVB (Fig. 8G, Right), whereas the mean heart rate over 24 h is unchanged, similar to KO^αMHC-Cre (Fig. 3 C–F and Fig. S3). Likewise, I_f in patch-clamped KO^Hcn4-CreERT2 SAN is significantly diminished (I_{f, KOHcn4-CreERT2} = −6.9 ± 2.2 pA/pF, n = 25) relative to WT SAN (I_{f, WT SAN} = −21.8 ± 4.9 pA/pF, n = 11, P < 0.01, Fig. 8 H and I) and comparable to KO^αMHC-Cre (I_{f, KO}α_MHC-Cre = −11.9 ± 2.6 pA/pF @ −160 mV, n = 27, P = 0.14 compared with I_{f, KOHcn4-CreERT2}). Also similar to KO^αMHC-Cre SAN was the percentage of KO^Hcn4-CreERT2 SAN with near-zero I_f amplitudes (14 of 25 or 56% < −3 pA/pF at −160 mV) vs. (0 of 11) in WT SAN. Taken together, these data show that Trpm7 is required for the maintenance of normal cardiac automaticity via transcriptional regulation of Hcn4 and I_f in adult murine SAN and embryonic myocardium.

Fig. 8.

Postnatal SAN/AVN restricted Trpm7 deletion recapitulates the phenotype of global cardiac Trpm7 KO. (A) SAN-restricted Trpm7 deletion (KO^Hcn4-CreERT2) was achieved in Hcn4-CreERT2 × Trpm7^fl/fl/Trpm7^fl/- mice at 6–8 wk of age by treating with tamoxifen (40 mg/kg) by gavage for 4 d, followed by >2-wk washout period before telemetry and SAN studies (B) (Upper) Transmitted light + GFP fluorescence image of Hcn4-CreERT2-ROSA26^mTmG right atrium after tamoxifen treatment. (Lower) Fluorescence mTomato/mGFP image showing recombination (mGFP) restricted to SAN region. (Scale bar, 1 mm.) (C) PCR across exon 17 from genomic DNA isolated from dissected KO^Hcn4-CreERT2 and WT SAN. Tail DNA from Trpm7^fl/- serves as a positive control for Trpm7 exon 17 deletion (-). (D) Representative TRPM7 current–voltage traces measured in WT SAN and KO^Hcn4-CreERT2 SAN. (E) Mean TRPM7 current density in WT SAN and KO^Hcn4-CreERT2 SAN. (F) Representative ECG showing an episode of SP observed in a telemetered KO^Hcn4-CreERT2 mouse. Solid arrows denote location of expected p waves. Red dashed circles and Insets above show change in p-wave morphology after SP, indicative of ectopic atrial focus. (G) Box plots with overlying data points showing the distribution of the frequency SPs (Left) and AVB (Right) observed over 24 h of telemetric monitoring in WT (n = 7) and KO^Hcn4-CreERT2 (n = 5) mice. (H) I_f current–voltage relationship from WT SAN (n = 11) compared with KO^Hcn4-CreERT2 SAN (n = 25). (I) Box plots with overlying data points showing the mean and distribution of I_f current densities at −160 mV from WT SAN (n = 11) compared with KO^Hcn4-CreERT2 SAN (n = 25). *P < 0.05, **P < 0.01.

Discussion

We sought to determine the contribution of TRPM7 to myocardial function by studying the consequences of cardiac-targeted Trpm7 deletion in vitro and in vivo. We show that TRPM7, when activated, forms a significant outwardly rectifying current in several different myocardial subtypes: with highest current densities in EVM > SAN > AVM. TRPM7 deletion in cultured EVM slows cardiac automaticity in vitro, as does TRPM7 knockdown in zebrafish. Moreover, in both global cardiac (KO^αMHC-Cre) and inducible SAN-restricted (KO^Hcn4-CreERT2), TRPM7 deletion disrupts automaticity of isolated SAN and intact SAN (and likely AVN), manifesting as SPs and AVB in the adult mouse heart. Given that TRPM7 is known to conduct Ca²⁺ (8), an obvious potential mechanism is that TRPM7 provides a Ca²⁺ influx pathway important for cardiac automaticity. However, it important to note that TRPM7 currents are typically recorded under nonphysiological conditions, becoming fully activated after ∼10 min of cell dialysis (i.e., Mg²⁺ chelation with EDTA/Na₂ATP), in nominally free external Mg²⁺. Under these conditions, TRPM7 current density is used to assay the total amount of functional protein at the cell membrane as opposed to physiologically relevant current magnitudes. Under more physiological conditions, TRPM7 current is essentially inactive in SAN and EVM, as demonstrated by the initial currents on “break-in” (Figs. S1 and S4). Thus, it is unlikely that TRPM7 contributes in a meaningful way to diastolic Ca²⁺ influx, under basal conditions, given the strong outward rectification characteristic of TRPM7 and its further inhibition by physiological concentrations of extracellular Mg²⁺. Finally, TRPM7 blockade using FTY720 had no effect on WT SAN firing rates under basal physiological conditions (Fig. S6).

There are a number of other ion channels, Hcn4, Hcn2, Ca_v3.1, and Ca_v1.3, nonetheless, that are considered important for cardiac automaticity (4). Indeed, expression analysis for these “automaticity” genes revealed significant reductions in Hcn4, Ca_v3.1, and SERCA2a mRNA in EVM upon TRPM7 deletion, whereas in SAN only Hcn4 mRNA was reduced. Consistent with these findings, the pacemaker current, I_f (encoded by Hcn4), is significantly reduced in isolated SAN and associated with slowed diastolic depolarization. Furthermore, KO^αMHC-Cre mice were less sensitive than WT mice to the negative chronotropic effects of the I_f blocker ivabradine. Thus we conclude that TRPM7 is required for maintenance of cardiac automaticity via transcriptional regulation of Hcn4 expression (via an as yet unidentified pathway).

It is well established that the hyperpolarization-activated pacemaker current, I_f, (encoded by Hcn4) contributes to basal automaticity in mouse SAN (24–26) and embryonic heart (27). However, there is controversy as to the extent of this contribution. Hcn4 knockout studies have demonstrated phenotypes ranging from profound bradycardia with AV block (24) to modest SPs (25), with no significant effect on mean heart rates in telemetered mice, similar to what we observe. Also consistent with our results, both of these studies show clear effects on firing rates and diastolic depolarization of isolated SAN cells. Furthermore, with maximal β-adrenergic stimulation (100 nM ISO), KO^αMHC-Cre SAN largely overcome impairments in automaticity. This finding is consistent with the notion that I_f current is an important determinant of spontaneous activity under basal conditions (25) but does not play an exclusive role in β-adrenergic regulation of pacemaking (26). With β-adrenergic stimulation, the membrane potential rises during late depolarization, and I_f contributes less, allowing other ionic mechanisms, such as Ca_v3.1, Ca_v1.3, and SR Ca²⁺ release, to dominate. It is also noteworthy that in both KO^αMHC-Cre and KO^Hcn4-CreERT2 mice, I_f is reduced in a mosaic fashion such that there is a small population of SAN cells that contain near-normal I_f current densities. Therefore, the KO phenotype of SPs with no significant change in mean HR over 24 h is consistent with a mechanism in which a population of SAN cells, with near-normal I_f, supports “normal” automaticity by overdrive pacing the I_f-deficient SAN cells. However, given the lower number of “normal” SAN cells, this overdrive pacing occasionally fails and results in SPs.

Another member of the TRPM family, Trpm4, has been recently found to be mutated in forms of conduction system disease in humans, including progressive familial heart block I (28, 29). These are thought to be gain-of-function mutations resulting in impaired TRPM4 endocytosis secondary to constitutive SUMOylation. The mechanism for this gain of function was deduced on the basis of heterologous expression studies in HEK cells and the observation that TRPM4 is enriched in human (28) and bovine (29) Purkinje cells. However, the mechanisms underlying TRPM4 gain-of-function contributions to conduction block has yet to be established. Because TRPM4 is a Ca²⁺-activated, sodium-selective current, one possibility is that increased TRPM4 current density results in further membrane depolarization in Purkinje cells, inactivating excitatory voltage-gated ionic currents and reducing the contribution of hyperpolarization-activated currents (I_f) (28). To date, no gain-of-function mouse model data are available, and the Trpm4^−/− mouse, aside from hypertension (30), has not yet been shown to develop other cardiovascular phenotypes.

This article reports a TRP channel influencing the expression level of another ion channel(s) in SAN and EVM. The connection between TRPM7 and Hcn4 expression in SAN and EVM remains an intriguing question. We showed recently that TRPM7 is required for early events in cardiogenesis, and perturbations in TRPM7 function during ventricular development impair myocardial function, atrioventricular conduction, and repolarization (15). Thus, we surmised that embryonic deletion of TRPM7 in SAN/AVN tissue in KO^αMHC-Cre mice might alter the maturation of SAN/AVN cells and consequently the expression of Hcn4. However, SAN/AVN-restricted TRPM7 deletion in adult KO^Hcn4-CreERT2 mice recapitulated the phenotype observed in KO^αMHC-Cre mice, suggesting that the effect of TRPM7 on Hcn4 expression and automaticity in SAN represents, instead, a postdevelopmental effect. In this regard, it is noteworthy that SAN cells are more fetal-like, in that they are smaller, mononucleated, and retain automaticity, similar to early embryonic myocardium and in contrast to the larger, quiescent, multinucleated adult ventricular cells. Perhaps it is retention of this “embryonic” nature of SAN cells that allows them to remain sensitive to the effects of TRPM7 after the developmental phase.

A number of studies now show that the T-box transcription factor Tbx3 is critical in mediating the developmental programs leading to formation and function of SAN, AVN, and conduction system in murine heart (31–34). Indeed, the most marked differentially expressed gene upon manipulation of Tbx3 expression levels in myocardium is Hcn4 (33). In addition, Tbx3-deficient mice exhibit both SAN and AVN dysfunction, manifested as SPs and AVB, respectively (35). Mef2 transcription factors have also been shown to directly regulate Hcn4 expression (36), and the Mef2a^−/− mouse develops sinus arrhythmia and conduction block (37). We speculate that TRPM7 may influence these transcriptional pathways in heart, either directly or indirectly, to affect cellular differentiation. We showed recently that TRPM7 C-terminal kinase is cleaved from the channel in LN-18 cells undergoing Fas-induced apoptosis, and this cleavage induced an increase in TRPM7 channel activity that is required for apoptosis (38). Indeed, cleaved TRPM7 C-terminal kinase has been observed in other cell types (38), potentially freeing it to interact with cellular partners beyond the plasma membrane to modify transcriptional programs. We are currently testing these hypotheses.

In summary, we showed that TRPM7 is most highly expressed in EVM and SAN myocytes, and genetic ablation of Trpm7 in these cells severely disrupts automaticity in vitro and in vivo. Impaired automaticity in Trpm7-deleted murine SAN cells arises from a lower and flatter slope of diastolic depolarization, associated with a slowed diastolic Ca²⁺ rise and reduced pacemaker current I_f (encoded by Hcn4). The next goal will be to determine how TRPM7 transcriptionally regulates genes in SAN and in embryonic myocardium to affect cardiac automaticity.

Methods

Refer to SI Methods for full methods.

Cardiac-Targeted Trpm7 Knockout Mice.

All animal procedures have been reviewed and approved by the Institutional Animal Care and Use Committee at Children’s Hospital Boston. Animals were housed under standard conditions and allowed access to food and water ad libitum. Cardiac-targeted knockout mice were generated by crossing Trpm7^fl/fl and Trpm7^fl/- mice described previously (7) with αMHC-Cre or Hcn4-Cre-ERT2 (23) mice. Mice were maintained on 129/SvEvTac mixed genetic background.

Embryonic Cardiomyocyte Isolation, Culture, and Adenoviral Transduction.

Trpm7^fl/fl embryonic hearts (6–11) were removed from E13.5–14.0 embryos obtained from pregnant Trpm7^fl/fl female mice mated with a Trpm7^fl/fl male. Adenoviruses, Ad-CMV-Lacz, Ad-CMV-Cre, and Ad-TnT-Cre, were kind gifts from William Pu, Boston Children's Hospital, Boston, MA. EVM were incubated differentiation media (DM) with adenovirus (multiplicity of infection 100) for 24 h and then washed two times with DM. Experiments were performed 2–5 d after viral transduction.

Ventricular Cardiomyocyte Isolation.

Ventricular myocytes were isolated by enzymatic digestion using either a solution of 0.07 mg/mL Liberase Blendzyme (Roche Diagnostics) or a mixture of 0.4 mg/mL Collagenase B (Roche), 0.3 mg/mL Collagenase D (Roche), and 0.025–0.05 mg/mL Protease XVI (Sigma), in nominally Ca²⁺-free Tyrode’s solution, as previously described (39).

Isolation of SAN Myocytes.

SAN myocytes were isolated as described by Marger et al. (40).

Zebrafish Morpholino.

Disruption of TRPM7 in zebrafish by MO injection was performed as previously described (41).

Ca²⁺ Imaging of Embryonic Cardiomyocytes.

EVMs plated on 10-mm glass-bottom coverslip culture dishes were loaded with 5–10 μM Fluo-4 for 30 min at 37 °C in DM. The cells were imaged using an FV1000 confocal microscope.

Ca²⁺ Imaging of SAN Myocytes.

Spontaneous [Ca²⁺]_i transients were recorded in both enzymatically isolated primary SAN cells and in individual cells of intact SAN tissue loaded with Fluo-4AM under control (Tyrode’s) or ISO at 36 °C as previously described (42).

Mouse Telemetry.

Mouse telemetry was performed as described previously (42).

Cellular Electrophysiology.

I_Trpm7, I_f, I_CaL, and I_CaT were measured in the whole-cell configuration as previously described (7, 40). Cellular automaticity was recorded under perforated-patch conditions using β-escin as previously described (40).

Quantitative RT-PCR Expression Analysis.

mRNA was quantified by quantitative RT-PCR from total RNA isolated from pooled SAN (four to six SAN per sample) dissected from KO or WT hearts, or from 35-mm dishes of beating cultured embryonic ventricular myocytes, as previously described (7).

Statistics.

All data are represented as means ± SEM, unless otherwise specified. All P values were calculated using the two-samples, independent Student t test, with the exception of the data from Fig. 3 E and F and Fig. 8G, which used the nonparametric Mann-Whitney test.