TRPM4 non‐selective cation channels influence action potentials in rabbit Purkinje fibres

Key points

• The transient receptor potential melastatin 4 (TRPM4) inhibitor 9‐phenanthrol reduces action potential duration in rabbit Purkinje fibres but not in ventricle.

• TRPM4‐like single channel activity is observed in isolated rabbit Purkinje cells but not in ventricular cells.

• The TRPM4‐like current develops during the notch and early repolarization phases of the action potential in Purkinje cells.

Abstract

Transient receptor potential melastatin 4 (TRPM4) Ca²⁺‐activated non‐selective cation channel activity has been recorded in cardiomyocytes and sinus node cells from mammals. In addition, TRPM4 gene mutations are associated with human diseases of cardiac conduction, suggesting that TRPM4 plays a role in this aspect of cardiac function. Here we evaluate the TRPM4 contribution to cardiac electrophysiology of Purkinje fibres. Ventricular strips with Purkinje fibres were isolated from rabbit hearts. Intracellular microelectrodes recorded Purkinje fibre activity and the TRPM4 inhibitor 9‐phenanthrol was applied to unmask potential TRPM4 contributions to the action potential. 9‐Phenanthrol reduced action potential duration measured at the point of 50 and 90% repolarization with an EC₅₀ of 32.8 and 36.1×10⁻⁶ mol l⁻¹, respectively, but did not modulate ventricular action potentials. Inside‐out patch‐clamp recordings were used to monitor TRPM4 activity in isolated Purkinje cells. TRPM4‐like single channel activity (conductance = 23.8 pS; equal permeability for Na⁺ and K⁺; sensitivity to voltage, Ca²⁺ and 9‐phenanthrol) was observed in 43% of patches from Purkinje cells but not from ventricular cells (0/16). Action potential clamp experiments performed in the whole‐cell configuration revealed a transient inward 9‐phenanthrol‐sensitive current (peak density = −0.65 ± 0.15 pA pF^–1; n = 5) during the plateau phases of the Purkinje fibre action potential. These results show that TRPM4 influences action potential characteristics in rabbit Purkinje fibres and thus could modulate cardiac conduction and be involved in triggering arrhythmias.

Key points

• The transient receptor potential melastatin 4 (TRPM4) inhibitor 9‐phenanthrol reduces action potential duration in rabbit Purkinje fibres but not in ventricle.

• TRPM4‐like single channel activity is observed in isolated rabbit Purkinje cells but not in ventricular cells.

• The TRPM4‐like current develops during the notch and early repolarization phases of the action potential in Purkinje cells.

Abbreviations

9‐phe

9‐phenanthrol

AP

action potential

APA

Action potential amplitude

APD

action potential duration

APD_x

action potential duration at x % of repolarization

I_Ca

voltage‐gated Ca²⁺ current

I_K1

inward‐rectifier K⁺ current

I_Ks

slow K⁺ current

I_Na

voltage‐gated Na⁺ current

I_to

transient outward current

PC

Purkinje cell

PF

Purkinje fibre

RMP

resting membrane potential

TRPM4

transient receptor potential melastatin 4

V_m

transmembrane potential

V_max

maximum upstroke velocity of action potential during the depolarizing phase

Introduction

The cardiac conduction system, including the sino‐atrial node, atrioventricular node and His–Purkinje system, is responsible for the initiation and propagation of the cardiac electrical activity, which allows for coordinated heart contraction (Wang & Hill, 2010). The His–Purkinje system in the ventricles consists of a common bundle (His bundle), the left and right bundle branches (which arise from the His bundle) and a network of terminal Purkinje fibres (PFs) (which arise from the bundle branches). These fibres rapidly conduct the action potential (AP) to all parts of the ventricles to ensure simultaneous contraction (Boyett, 2009). Nevertheless, PF APs are different from ventricular APs: they exhibit a faster upstroke velocity (V _max) and higher amplitude (APA) (phase 0), a larger rapid transient repolarization (phase 1), a more negative plateau (phase 2) and a longer AP duration (APD) (Boyden et al. 2010). These differences are attributable, in part, to a greater contribution of neuronal voltage‐gated sodium channels that comprise the fast sodium current I _Na (Haufe et al. 2005), a larger contribution of the potassium component of the transient outward current I _to in phase 1 and smaller contributions of the inward‐rectifier potassium current I _K1 and the slow potassium current I _Ks during repolarization (phase 3) in Purkinje cells (PCs) (Cordeiro et al. 1998; Dumaine & Cordeiro, 2007). Additional currents such as that attributable to transient receptor potential melastatin 4 (TRPM4) ion channels may further account for these differences.

Indeed, the heart is known as a major TRPM4 mRNA expressing tissue (Launay et al. 2002; Nilius et al. 2003; Mathar et al. 2014 a; Kruse & Pongs, 2014; Guinamard et al. 2015). More specifically, the TRPM4 mRNA relative expression in non‐diseased human hearts indicates that it is expressed more in PF than in septum, atrium and ventricles (Kruse et al. 2009). Immunoblot experiments performed on the bovine heart confirmed that TRPM4 is mostly present in the conduction system (Liu et al. 2010). TRPM4 encodes a non‐selective cation channel that is equally permeable to Na⁺ and K⁺, but not to Ca²⁺ (permeability sequence: Na⁺ ∼ K⁺ > Cs⁺ > Li⁺ ≫ Ca²⁺) (Launay et al. 2002; Nilius et al. 2005). TRPM4 current is sensitive to [Ca²⁺]_i, which activates the channel, and to voltage, which evokes higher activity in the positive voltage range (Launay et al. 2002; Nilius et al. 2003). Functional TRPM4 currents were recorded in several cardiac preparations, particularly in the mouse sinus node belonging to the conduction system (Demion et al. 2007). In the sinus node of mouse, rat and rabbit, TRPM4 participates in pacemaking activity (Hof et al. 2013). TRPM4 also participates in atrial (Simard et al. 2013) and ventricular (Mathar et al. 2014 b) murine cardiac APs. Accordingly, it is conceivable that TRPM4 participates in PF APs as well.

Cardiac electrical perturbations suggest that TRPM4 contributes to PF activity. Trpm4 knock‐out mice exhibit bundle branch blocks with PR interval lengthening in the ECG. Both suprahisian and infrahisian conduction time are increased in Trpm4 knock‐out mice, which indicates that TRPM4 influences cardiac conduction (Demion et al. 2014). In a mouse model of hypoxia and re‐oxygenation‐induced ventricular arrhythmias, pharmacological inhibition of TRPM4 suppressed early after depolarizations (Simard et al. 2012) putatively triggered by the PF. In addition, TRPM4 mutations were found in families with autosomal dominant inherited conduction blocks, including bundle branch blocks (Kruse et al. 2009; Liu et al. 2010; Stallmeyer et al. 2012). In most cases a gain of function resulted from the mutations, which further suggests a role for TRPM4 in regulating conduction in PFs.

Despite these putative roles and high expression in the His–Purkinje system, the potential role(s) of TRPM4 channels in PF electrical activity has never been investigated directly. Here we examine whether a TRPM4 current is present in rabbit PCs using patch clamp and then evaluate its contribution to the PF AP using intracellular microelectrodes. We were able to reveal single channel TRPM4‐like currents from rabbit PCs. In addition, TRPM4 pharmacological inhibition reduced APD in rabbit PFs.

Methods

Ethical approval

Experiments were carried out in strict accordance with the European Commission Directive 2010/63/EU for animal care. This study was also conducted with authorization for animal experimentation #14‐98 from the local DDPP (Direction Départementale de la Protection des Populations). Rabbits were housed at 18–22°C in stainless steel cages on racks equipped with automatic watering systems and free access to food, at the animal care centre of Caen University and received daily care by trained staff. All procedures were performed humanely after general anaesthesia induction, and all efforts were made to minimize suffering.

Heart sampling

New Zealand female rabbits of 1.5–2 kg were killed under anaesthesia with sodium pentobarbital 125 mg kg⁻¹ i.v. (1000 USP kg⁻¹ heparin was added to the injection). The heart was removed and dissected in a cardioplegic solution containing (in mmol l⁻¹): 108 NaCl, 30 KCl, 1.8 CaCl₂, 1 MgCl₂, 1.8 NaH₂PO₄, 25 NaHCO₃ and 55 glucose. Ventricles were opened to expose the PFs. Ventricular samples containing PFs were isolated and pinned in a superfusion chamber perfused at a rate of 10 ml min⁻¹ with a physiological solution containing (in mmol l⁻¹): 108 NaCl, 4 KCl, 1.8 CaCl₂, 1 MgCl₂, 1.8 NaH₂PO₄, 25 NaHCO₃ and 11 glucose. All solutions were warmed to 37°C and maintained at pH 7.4 by bubbling with a 95% O₂ and 5% CO₂ mixture.

AP recordings in PFs

Isolated ventricular strips with PFs were stimulated at a rate of 2 Hz by square electric pulses using an SMP‐310 programmable stimulator (Bio‐Logic Science Instruments, Claix, France) connected to a bipolar silver‐wire electrode. After 2 h of stabilization, transmembrane potential and APs were recorded after cell impalement using a glass microelectrode filled with 3 mol l^–1 KCl and with a tip resistance of around 10×10⁶ Ω. Microelectrodes were coupled to the input stages of a home‐built impedance capacitance‐neutralizing amplifier. The recordings were displayed and analysed using cardiac AP acquisition software (iox 2, EMKA, Coventry, UK), which automatically analyses waveforms and thus provides the resting membrane potential (RMP), APA, AP duration at 50% (APD₅₀), 70% (APD₇₀) and 90% (APD₉₀) repolarization, as well as the maximum upstroke velocity of the AP during the depolarizing phase (V _max). To investigate the pharmacological modulation of PF APs, pharmacological tools (9‐phenanthrol or T16Ainh‐A01 from Sigma Aldrich, Poole, UK) were added to the superfusion solution. These molecules were previously dissolved in DMSO. The maximal final DMSO concentration in the superfused control solution did not exceed 0.1%.

Isolation of ventricular cardiomyocytes and PCs

Rabbit ventricular cardiomyocytes and PCs were isolated by an enzymatic digestion procedure modified from a previously published protocol (Brette et al. 2013). The heart was removed and mounted on a Langendorff apparatus and perfused retrogradely with a Hepes‐based isolation solution containing (in mmol l⁻¹): 130 NaCl, 5.4 KCl, 0.4 NaH₂PO₄, 1.4 MgCl₂, 1 CaCl₂, 10 Hepes, 10 glucose, 20 taurine and 10 creatine (pH 7.4 with NaOH). When the coronary circulation had cleared of blood, perfusion was continued with Ca‐free isolation solution (in which CaCl₂ was replaced with 0.1 mmol l⁻¹ EGTA) for 4 min, followed by perfusion for a further 8 min with isolation solution containing 224 i.u. ml⁻¹ collagenase (type I; Worthington Biochemical, Lakewood, NJ, USA), and 0.28 i.u. ml⁻¹ protease (type XIV; Sigma Aldrich). The heart was then opened, the PFs were dissected in the Ca‐free isolation solution and gently shaken at 37°C in 2 ml of enzyme‐containing solution. Isolated PCs were obtained within 20–60 min of digestion. To obtain ventricular myocytes, ventricles were minced and digested in the enzyme‐containing solution supplemented with 1% BSA and gently shaken at 37°C. Ventricular cells were filtered from this solution at 4 min intervals, gently centrifuged and suspended in two successive steps in isolation solution containing first 0.1 then 1 mmol l⁻¹ CaCl₂. The isolation solutions were continuously oxygenated during the digestion protocol.

Patch‐clamp experiments

Patch‐clamp recordings in PCs and ventricular isolated cells were performed at room temperature (20–25°C). An Axopatch 200B (Axon Instruments, Sunnyvale, CA, USA) amplifier controlled by a computer and connected via a Digidata 1322A A/D converter (Axon Instruments) was used to acquire and analyse the data in conjunction with pClamp software (Axon Instruments). Signals were filtered at 1 and 2 kHz for inside‐out and whole‐cell recordings respectively, using an eight‐pole Bessel low‐pass filter before digitization at 10 kHz and storage. Patch pipettes were made from TW150F‐3 glass (World Precision Instruments, Sarasota, FL, USA). Pipettes resistance for whole‐cell recordings was typically 1×10⁶ Ω for isolated ventricular cells and PCs and 6–8×10⁶ Ω for inside‐out recordings when filled with intracellular solutions (see below). The liquid junction potential was subtracted from V _pipette to calculate V _m.

Single channel recordings in isolated PCs and ventricular cardiomyocytes

Cells were preincubated, for at least 10 min, before patching, in the Hepes‐based isolation solution supplemented with 500 nmol l⁻¹ phorbol 12‐myristate 13‐acetate (PMA) (Sigma Aldrich), which enhances TRPM4 detection in human and rat cardiomyocytes (Guinamard et al. 2004, 2006). Pipette solution contained (in mmol l⁻¹): 145 NaCl, 1.2 MgCl₂, 10 glucose, 10 Hepes and 1 CaCl₂ (pH 7.4). The same solution, containing 10⁻³, 10⁻⁶ or 10⁻⁹ mol l⁻¹ CaCl₂ [using a Ca²⁺‐EGTA balance calculated by Maxchelator, Chris Patton, Standford University, http://maxchelator.standford.edu (Bers et al. 2010)] was perfused at the inside of the membrane to assess Ca²⁺ sensitivity of the channel. To estimate anion to cation selectivity of the current, the perfused solution contained (in mmol l⁻¹): 42 NaCl, 1.2 MgCl₂, 10 glucose, 10 Hepes and 1 CaCl₂ (pH 7.2), supplemented with sucrose to maintain osmolarity. To determine selectivity of monovalent cations, the perfused solution contained (in mmol l⁻¹): 145 KCl, 1.2 MgCl₂, 10 glucose, 10 Hepes and 1 CaCl₂ (pH 7.4).

Whole‐cell current recordings in isolated PCs

Representative PF APs recorded using the microelectrode technique were used as standard command voltage for action potential‐clamp experiments on PCs. This protocol was preceded by a 80 ms prepulse to −40 mV to fully inactivate I _Na. The second protocol consisted of two steps: an 80 ms prepulse at −40 mV to inactivate I _Na and a 300 ms step at 0 mV to measure I _Ca amplitude. Baseline membrane potential was clamped at −80 mV. Note that a junction potential of −7 mV has to be added to the clamp potential.

Native cells were bathed in the Hepes‐based isolation solution described above. Perfused solution contained (in mmol l⁻¹): 20 NaCl, 115 sodium acetate, 1.8 calcium acetate, 10 glucose and 10 Hepes (pH 7.4 with NaOH). Pipette solution contained (in mmol l⁻¹): 135 caesium acetate, 1 MgCl₂, 20 Hepes and 0.5 EGTA (pH 7.2 with CsOH).

Data analysis

Results are reported as mean ± SEM. For microelectrodes recordings n refers to the number of experiments and animals. For patch‐clamp recordings, n refers to the number of isolated cells examined. Paired or unpaired Student's t‐test or non‐parametric Wilcoxon test were used as appropriate to compare data. Statistical analyses were performed using Sigma Plot software v.11 (Systat Software Inc., Chicago, IL, USA). Statistically significant results were recognized at P < 0.05, which are indicated in the figures by asterisks.

Results

APs recorded from rabbit PFs and ventricle

Rabbit ventricular strips with intact PFs, as shown in Fig. 1, were superfused with oxygenated physiological solution. PFs were impaled with intracellular microelectrodes to record the transmembrane potential. Similarly, in separate experiments, ventricular cells were impaled. Typical PF APs were detected and were similar to those reported in previous studies (Rouet et al. 2012; Lemoine et al. 2014) and could be differentiated from ventricular APs according to their parameters (APA, APDs, V _max and RMP), which were all significantly different (Table 1).

Figure 1. Photomicrographs of PFs on a ventricular strip .

For microelectrode AP recordings, microelectrodes were impaled in a fibre.

Table 1.

Action potential parameters from Purkinje fibres and ventricular strips

AP type	RMP (mV)	APA (mV)	APD50 (ms)	APD70 (ms)	APD90 (ms)	V max (V s−1)	n
Purkinje	−78  ± 2	111  ± 2	107  ± 8	138  ± 8	170  ± 8	520  ± 35	16
Ventricle	−88 ± 1	94  ± 2	84  ± 5	99  ± 6	111  ± 6	268  ± 24	8

RMP, resting membrane potential; APA, action potential amplitude; APD_x, action potential duration at x % of total repolarization; V _max, maximal upstroke velocity. All parameters are significantly different between PF and ventricle (P < 0.05).

Modulation of rabbit PF APs by 9‐phenanthrol

To unmask the contribution of the TRPM4 channel in APs, the effect of its inhibitor 9‐phenanthrol (Grand et al. 2008; Guinamard et al. 2014) was investigated. After the preparation stabilized under control conditions (2 h), 9‐phenanthrol was added to the superfusion chamber for 10 min at concentrations ranging from 3×10⁻⁶ to 10⁻⁴ mol l⁻¹, the additions being separated by 20 min washout periods. 9‐Phenanthrol reversibly shortened APD in a concentration‐dependent manner, starting at 10⁻⁵ mol l⁻¹ (all P < 0.05, n = 5). Figure 2 A and B shows the time course of 9‐phenanthrol's effect on APD in a representative experiment. Fitting APD reduction as a function of [9‐phenanthrol] to a single sigmoid curve predicts maximal reductions of 38.3, 35.6 and 30.2% (n = 5–6 rabbits), and an EC₅₀ at 32.8, 31.8 and 36.1×10⁻⁶ mol l⁻¹ for APD₅₀, APD₇₀ and APD₉₀, respectively (Fig. 2 C). These EC₅₀ values are in line with those reported for the shortening effect of 9‐phenanthrol on mice atrial APD (Simard et al. 2013) but also for the inhibitory effect of the molecule on the TRPM4 current expressed in HEK‐293 cells (Grand et al. 2008). By contrast, 9‐phenanthrol did not produce any effect on APA, V _max or RMP (Fig. 2 D), except at 10⁻⁴ mol l⁻¹.

Figure 2. Reduction of rabbit PF APD by 9‐phenanthrol .

APs were recorded in rabbit PFs using intracellular microelectrodes A, representative PF APs under control conditions, then in the presence of 3×10⁻⁵ mol l⁻¹ 9‐phenanthrol and after washout. B, continuous time course of mean APD₅₀, APD₇₀, APD₉₀ calculated every 10 s, in the control and during the superfusion of 3×10⁻⁵ mol l⁻¹ 9‐phenanthrol (9‐phe.), indicated by the bold line, corresponding to the experiment shown in A. C, concentration–response curve showing the effect of 9‐phenanthrol on the rabbit PF APD₅₀, APD₇₀ and APD₉₀ reductions. Values are expressed as the percentage of reduction of APD. Each point is the mean (± SEM) of 3–6 animals. Data points are fitted to a Hill equation. D, mean variation of AP parameters expressed as the percentage of reduction from control, induced by 10⁻⁵ mol l⁻¹ (white) and 3×10⁻⁵ mol l⁻¹ (black) 9‐phenanthrol (n = 5 and 6, respectively). Asterisks indicate a statistically significant difference from control (P < 0.05).

Lack of TMEM16a participation in 9‐phenanthrol's effect

9‐Phenanthrol has recently been shown to inhibit the TMEM16a Ca²⁺‐activated chloride channel in rat cerebral myocyte arteries (Burris et al. 2015). To evaluate whether 9‐phenanthrol diminished APD via TMEM16a inhibition, we perfused a specific TMEM16a inhibitor, T16Ainh‐A01, using the same protocol as 9‐phenanthrol. T16Ainh‐A01 at 10⁻⁵ and 3×10⁻⁵ mol l⁻¹ (n = 4) did not significantly modify PF APs (Table 2). These results suggest that the 9‐phenanthrol‐induced shortening of PF APs is not attributable to TMEM16a inhibition and further that TMEM16a does not participate in PF APs.

Table 2.

Action potential parameters from Purkinje fibres under control and with superfusion of T16Ainh‐A01 at 3×10⁻⁵ mol l⁻¹

	RMP (mV)	APA (mV)	APD50 (ms)	APD70 (ms)	APD90 (ms)	V max (mV)	n
Control	−75 ± 1	117 ± 5	104 ± 19	139 ± 14	166 ± 11	579 ± 10	4
T16	−76 ± 1	116 ± 5	100 ± 19	134 ± 14	162 ± 12	555 ± 17	4
Statistical significance	NS	NS	NS	NS	NS	NS

RMP, resting membrane potential; APA, action potential amplitude; APD_x, action potential duration at x % of total repolarization; V _max, maximal upstroke velocity; NS, not significant. No significant differences are observed between the two conditions (P > 0.05).

Lack of contribution of TRPM4 to the rabbit ventricular AP

The effect of 9‐phenanthrol on rabbit ventricular AP parameters was evaluated using the same protocols as in PFs. Ventricular strips were superfused with a physiological and oxygenated solution and APs were recorded using intracellular microelectrodes. Superfusion of 3×10⁻⁵ mol l⁻¹ 9‐phenanthrol did not significantly modify ventricle AP parameters (Fig. 3 A, B, n = 5; Table 3). These results suggest that TRPM4 does not participate in rabbit ventricular AP.

Figure 3. 9‐Phenanthrol does not modulate ventricular APs .

AP were recorded on ventricular strips using intracellular microelectrodes. A, representative ventricular APs under control condition, then in the presence of 3×10⁻⁵ mol l⁻¹ of 9‐phenanthrol and after washout. B, continuous time course of mean APD₅₀, APD₇₀, APD₉₀ calculated every 10 s, in the control and during the superfusion of 3×10⁻⁵ mol l⁻¹ 9‐phenanthrol, indicated by the bold line, corresponding to the experiment shown in A.

Table 3.

Action potential parameters from ventricle under control and with superfusion of 9‐phenanthrol at 3×10⁻⁵ mol l⁻¹

	RMP (mV)	APA (mV)	APD50 (ms)	APD70 (ms)	APD90 (ms)	V max (mV)	n
Control	−86 ± 1	96 ± 4	82 ± 7	96 ± 9	107 ± 8	279 ± 42	5
9‐phe.	−87 ± 1	95 ± 5	84 ± 8	97 ± 8	110 ± 8	257 ± 53	5
Statistical significance	NS	NS	NS	NS	NS	NS

RMP, resting membrane potential; APA, action potential amplitude; APD_x, action potential duration at x % of total repolarization; V _max, maximal upstroke velocity; NS, not significant. No significant differences are observed between the two conditions (P > 0.05).

A TRPM4‐like current at the unitary level on isolated PCs

To investigate whether a functional TRPM4 channel is present in PCs, we probed for TRPM4 single‐channel currents in isolated PCs (Fig. 4 A shows photomicrographs of isolated ventricular cardiomyocytes and PCs). Compared to ventricular cells, PCs exhibit weaker striation and a statistically significant smaller size: 61.6 ± 4.7 vs. 135.0 ± 6.8 pF for Purkinje and ventricular cells, respectively (n = 41 for each cell type).

Figure 4. A TRPM4‐like single channel activity is detectable on isolated PCs but not on ventricular cells .

A, photomicrographs of typical isolated PCs and ventricular cells. Note that the scale is different between photomicrographs. B, single‐channel tracings recorded at various voltages (V _m) from an inside‐out patch. Solutions superfused at the internal side of the membrane (145 or 42 mmol l⁻¹) are indicated above the tracings. Dotted lines indicate the current level of closed channels. I–V relationships under different ionic conditions (middle panel) were reported and data points were fitted linearly. C, channel open probability (P _o) with 145 mmol l⁻¹ NaCl as a function of the transmembrane potential (V _m) from four experiments using the protocol shown in B. D, recordings of single channel activity at 10⁻³, 10⁻⁶ and 10⁻⁹ mol l⁻¹ [Ca²⁺]_i from the same patch, at + 40 mV with 145 mmol l⁻¹ NaCl. E, mean channel open probability (P _o) relative to the control (10⁻³ mol l⁻¹[Ca²⁺]_i) as in D. F, recording of single channel activity under control and during superfusion of 9‐phenanthrol at 3×10⁻⁵ mol l⁻¹. Detailed openings are shown in the insets. G, effect of superfusion of 9‐phenanthrol at 3×10⁻⁵ mol l⁻¹ on single channel open probability (P _o) on three different experiments. Mean is indicated by dark squares and large line. H, percentage of patches exhibiting TRPM4‐like channel activity in isolated PCs and ventricular cells.

Channel activity was first recorded using the inside‐out patch‐clamp configuration with the 145 mmol l⁻¹ NaCl standard solution (1 mmol l⁻¹ CaCl₂) on both sides of the membrane. Figure 4 B illustrates channel activity as a function of the membrane holding potential. Under symmetrical ionic conditions, the corresponding I–V relationship was linear with a slope conductance of 23.8 ± 0.3 pS and a reversal potential (E _R) at −0.5 ± 0.8 mV (n = 7). The anion to cation selectivity was evaluated by lowering NaCl in the internal side of the membrane from 145 to 42 mmol l⁻¹. That change shifted the reversal potential to more positive voltages (E _R = 20.0 ± 1.1 mV, n = 4) corresponding to a permeability ratio P _Na/P _Cl = 9.1, according to the Goldman–Hodgkin–Katz (GHK) equation (Fig. 4 B). Substitution of 145 mm NaCl by an equivalent concentration of KCl at the internal side of the membrane did not significantly change the reversal potential compared to symmetrical ionic conditions (E _R = 0.2 ± 0.8 mV, n = 5), which indicates that the channel does not differentiate among monovalent cations (P _K/P _Na = 0.98). Single‐channel conductance (g) in the presence of KCl was similar to that in the presence of NaCl (g = 23.4 ± 1.4 pS, n = 5). Mean channel open probability (P _o) plotted as a function of voltage from four experiments using a protocol similar to the one shown in Fig. 4 B indicates higher activity at depolarized voltages (Fig. 4 C). The data were fitted to a Boltzmann function with values of P _o,max = 0.60 ± 0.16 and V _0.5 = 47.3 ± 14.2 mV (n = 4).

Channel sensitivity to internal Ca²⁺ was evaluated by lowering [Ca²⁺] at the internal side ([Ca²⁺]_i) of the membrane from 10⁻³ mol l⁻¹ to 10⁻⁶ and 10⁻⁹ mol l⁻¹ (Fig. 4 D). The P _o relative to the control ([Ca²⁺]_i = 10⁻³ mol l⁻¹) was reduced to 0.49 ± 0.13 (n = 3) and 0.04 ± 0.03 (n = 4) for 10⁻⁶ and 10⁻⁹ mol l⁻¹, respectively (Fig. 4 E).

Application of 9‐phenanthrol at a concentration of 3×10⁻⁵ mol l⁻¹ significantly reduced P _o to 0.06 ± 0.03 from its control value of 0.31 ± 0.03 (Fig. 4 F and G, n = 3). These measurements match the previously reported characteristic properties of cloned TRPM4 channels heterologously expressed in HEK‐293 cells (Grand et al. 2008). They also match those of native TRPM4 currents reported in human, mouse and rat cardiomyocytes (Zhainazarov et al. 2003; Guinamard et al. 2004; Simard et al. 2013).

TRPM4‐like channel currents were detected in 43 % of inside‐out patches from PCs (n = 56). In contrast, similar experiments in isolated ventricular cardiomyocytes revealed no such channel currents (n = 16) (Fig. 4 H).

Whole‐cell 9‐phenanthrol‐sensitive current in rabbit isolated PCs

To determine whole‐cell TRPM4 current shape in a physiological context, we performed AP clamp experiments on isolated PCs, using PF APs recorded with a microelectrode as the standard voltage command. 9‐Phenanthrol at 3×10⁻⁵ mol l⁻¹ was used to inhibit TRPM4 channels and thus reveal TRPM4 contributions by subtraction.

The caveat of these experiments was spontaneous I _Ca run‐down during recordings. Because I _Ca reduction could wrongly be attributed to a 9‐phenanthrol‐inhibited current, cells presenting I _Ca variations during 9‐phenanthrol superfusion were excluded from further analyses. Note that we verified that 9‐phenanthrol has no effect on I _Ca at 3×10⁻⁵ mol l⁻¹ in specific experiments (data not shown).

A protocol to reveal the 9‐phenanthrol‐sensitive whole‐cell current in PCs during the AP is shown in Fig. 5 A: (1) the whole‐cell current is recorded in control conditions under an AP clamp protocol; (2) a two‐step protocol is applied to measure I _Ca amplitude; (3) 9‐phenanthrol (3×10⁻⁵ mol l⁻¹) is superfused for 2 min; (4) the whole‐cell current is recorded in the presence of 9‐phenanthrol during an AP clamp protocol; and (5) the two‐step protocol is repeated to remeasure I _Ca amplitude. If I _Ca amplitude is reduced compared to step 2 above, then the cell is discarded from further analysis.

Figure 5. Whole‐cell patch‐clamp experiments conducted on isolated PCs reveal a 9‐phenanthrol‐sensitive current in the early phases of PF AP .

A, diagram illustrating the sequence of the protocols. Numbers (1, 2, 3, 4 and 5) above the arrow indicate the different stages of the experiments where currents are recorded. B, currents recorded at different stages of the experiments (top panels) under different voltage‐clamped protocols (bottom panels). Numbers (1, 2, 4 and 5) correspond to the stages illustrated in A in which the current has been recorded. C, left panel: overlay of current tracings recorded in stages 1 and 4 corresponding to control and during 9‐phenanthrol superfusion. Middle panel: difference in current density (control – 9‐phenanthrol) for the experiment shown in B. Right panel: mean ± SEM difference in current density from five experiments similar to the one described in B. D, mean ± SEM I _Ca density during stages 2 and 5 from cells used in fig 5C right (n = 5).

Figure 5 B shows a representative experiment with no change in I _Ca. The current difference for this experiment, corresponding to the 9‐phenanthrol‐sensitive current, is provided in Fig. 5 C (middle panel). The mean ± SEM density of the 9‐phenanthrol‐sensitive current from five experiments is provided in Fig. 5 C (right panel). It shows a transient inward current that develops during phase 2 of the AP and extends to mid‐phase 3. It presents a maximal amplitude of −0.65 ± 0.15 pA pF^–1 (n = 5). Note that the mean I _Ca density from the five experiments conserved for the analysis did not change during the protocol, as presented in Fig. 5 D.

Discussion

TRPM4 contributes to PF APs. Our data demonstrate that the TRPM4 inhibitor 9‐phenanthrol shortens rabbit PF APs and inhibits an inward current during phase 2 and the first half of phase 3. In addition, a typical TRPM4‐like current is detectable at the single channel level in PCs.

Model limitations

Extrapolation of results obtained in animal models to humans depends on how well one can map the measurements in the model system to physiological parameters of human tissue. Regarding cardiac APs, there is a great discrepancy between the short‐lasting AP recorded in laboratory rodents (around 75 ms in rat) and the longer‐lasting AP in humans (around 300 ms) (Janse et al. 1998). The rabbit model appears to be a good compromise because its AP duration is longer than the rodent model but the animal care is still reasonably efficient to facilitate widespread experimentation. Cardiac AP parameters were reported to be qualitatively similar between rabbit, dog and human (Rudy et al. 2008). Furthermore, this animal is a commonly used and relevant model for studies evaluating Purkinje‐originated arrhythmias (Lu et al. 2001; Puddu et al. 2011).

A weakness of this model is the lack of specific molecular tools against rabbit TRPM4. We performed immunofluorescence labelling on rabbit cardiac slices using a goat polyclonal antibody targeting N‐terminus protein sequences of human, rat and mouse TRPM4 (Santa Cruz Biotechnology t‐20, ref. sc 27539; Santa Cruz, CA, USA). While this antibody recognized hTRPM4 in HEK293‐transfected cells, we did not observe any labelling in rabbit tissue. Additionally, we performed Western‐blotting on rabbit heart and HEK293 hTRPM4‐transfected cell lysates as a positive control, using the same antibody. Again, it revealed TRPM4 protein in HEK293‐transfected cells but not in rabbit tissue. We thus tested another goat polyclonal antibody targeting C‐terminus protein sequences of rat and mouse TRPM4 (Osense, Keswick, Australia, ref. OST00345W) but, similarly, did not reveal any TRPM4 labelling on rabbit tissue. As a final attempt, we performed a reverse transcriptase PCR (RT‐PCR) approach to search for TRPM4 mRNA on rabbit tissues (Purkinje, ventricle, atrium) and on human fibroblasts as a positive control. As the rabbit Trpm4 gene has not yet been cloned, we designed six pairs of primers to target conserved sequences in human, mouse and rat genes. None of them was able to amplify any mRNA transcript in rabbit. All these negative data probably arise from the absence of appropriate tools to reveal TRPM4 expression in rabbit. In addition, they suggest that the rabbit Trpm4 sequence might be significantly different from human, mouse or rat sequences.

TRPM4 as the 9‐phenanthrol target

Two main parameters have to be considered to identify 9‐phenanthrol targets in the modulation of PF APs: first the identity of ion channels known to be targeted by 9‐phenanthrol and second the presence or absence of these channels in PCs.

9‐Phenanthrol is a widely used TRPM4 inhibitor that unmasks channel activity in various tissues (Guinamard et al. 2014). A similar 9‐phenanthrol‐inducced APD reduction was observed in the atrium of WT mice but not in Trpm4 knock‐out mice, indicating that in mouse atrium, TRPM4 is the 9‐phenanthrol target (Simard et al. 2013). In the present study, 9‐phenanthrol was mainly used at 3×10⁻⁵ mol l⁻¹ to inhibit the TRPM4 channel. In contrast to experiments performed on mouse sinoatrial node or atria (Simard et al. 2013; Hof et al. 2013), no transgenic rabbits are currently available to definitely identify TRPM4 as the drug target in our model. Thus, our conclusions can only be based on the specificity of 9‐phenanthrol as established in rodent models. In addition to TRPM4, 9‐phenanthrol had been tested on several other channels, including TRPM5 (Grand et al. 2008), TRPC3, TRPC6 (Gonzales et al. 2010), TRPM7, the Ca²⁺‐activated K⁺ current (BK_Ca), the inward‐rectifier K⁺ current (I _K1), the voltage‐dependent K⁺ current (K _v) and the voltage‐dependent Ca²⁺ current (VD_Ca) and L‐type Ca²⁺ currents (I _Ca,L) (Gonzales et al. 2010). 9‐Phenanthrol has no effect on any of these targets at a concentration of 3×10⁻⁵ mol l⁻¹. In our experiments, 9‐phenanthrol at 3×10⁻⁵ mol l⁻¹ did not modify ventricular and PF APA and V _max, which suggests that I _Na is not affected. In addition, in a previous study, we showed that 9‐phenanthrol at 10⁻⁵ mol l⁻¹ does not affect I _CaL and I _K in mouse ventricular cells whereas 10⁻⁴ mol l⁻¹ inhibits both (Simard et al. 2012). Therefore, it was uncertain whether 9‐phenanthrol at 3×10⁻⁵ mol l⁻¹ could inhibit these currents. Nevertheless, 9‐phenanthrol in rabbit ventricular APs at 3×10⁻⁵ mol l⁻¹ had no effect, which precludes a lack of effect on I _CaL and I _K. Moreover, the specific experiments that we performed on isolated rabbit ventricular cells showed that 9‐phenanthrol at 3×10⁻⁵ mol l⁻¹ does not modulate I _Ca.

A recent study reported that 9‐phenanthrol inhibits the TMEM16a Ca²⁺‐activated Cl⁻ channel in the same range as TRPM4 (Burris et al. 2015). However, the TMEM16a channel inhibitor T16Ainh‐A01 did not modulate PF APs, which indicates that this channel is not a targeted channel in the present model. Another recent study also showed that 9‐phenanthrol at 2×10⁻⁵ mol l⁻¹ activates K_Ca 3.1, a Ca²⁺‐activated potassium channel in arterial smooth muscle (Garland et al. 2015), which causes membrane hyperpolarization. Activation of such a channel in PF would also shorten APD. However, K_Ca 3.1c hannels have never been reported in heart. In addition, the absence of an effect of 9‐phenanthrol at 3×10⁻⁵ mol l⁻¹ on ventricular APs makes the activation of K_Ca 3.1 by 9‐phenanthrol unlikely to occur on cardiac cells.

The lack of an effect of 9‐phenanthrol in rabbit ventricular APs, while it is effective in PF APs, indicates that the target is absent in ventricular cells but present in PCs. These data match our single channel data indicating the absence of TRPM4‐like currents in rabbit ventricular cells. It is also commensurate with human measurements that show TRPM4 mRNA more highly expressed in conductive tissue than in ventricles (Kruse et al. 2009) and in bovine models showing higher immunolabelling of TRPM4 protein in conductive tissue than in ventricles (Liu et al. 2010). Even if we failed to detect TRPM4 labelling by immunofluorescence studies, most probably due to the lack of convenient tools, the recording of a TRPM4‐like current constitutes strong evidence that rabbits express TRPM4 in heart.

Altogether, these data support TRPM4 as the most probable target of 9‐phenanthrol with respect to modulation of PF APs.

A TRPM4 current in PCs

TRPM4 channels have well‐established characteristics including 20–25 pS conductance, linear I–V relationship, equal permeability for Na⁺ and K⁺, voltage sensitivity with higher activity at positive voltages, and activation by internal Ca²⁺ (Launay et al. 2002; Nilius et al. 2003; Mathar et al. 2014 a). The single channel currents recorded from PCs in our study match these properties. Properties such as these could also be attributable to TRPM5, but the PC current is inhibited by 9‐phenanthrol, similar to TRPM4, whereas TRPM5 is not attenuated by 9‐phenanthrol (Grand et al. 2008).

The shape of the whole‐cell TRPM4 current is also well known in non‐excitable cells. TRPM4 produced an outward rectifying current (Launay et al. 2002; Nilius et al. 2003). However, because TRPM4 is sensitive to internal Ca²⁺, the waveform of the current is more difficult to unravel in an excitable cell such as PC. Indeed, [Ca²⁺]_i depends on voltage‐gated activated Ca²⁺ channels and subsequent release of Ca²⁺ stores, leading to the Ca²⁺ transient. During an AP, both voltage changes and Ca²⁺ transients modulate TRPM4 activity. In our whole‐cell recordings, the reversal potential for monovalent cations was +8 mV. Three properties have to be considered to predict the whole‐cell current during an AP in PCs: (1) because membrane potential trajectory remains mainly below 0 mV during the AP – including the plateau phase – the expected TRPM4 current is inward; (2) the voltage sensitivity of the channel predicts enhanced activation when the cell depolarizes, i.e. during the plateau; and (3) TRPM4 activates during subsarcolemmal [Ca²⁺] increase, which occurs in a PC at the early phases of the AP and decreases rapidly (Cordeiro et al. 2001). According to these arguments, the TRPM4 current would be an inward current that develops during the Ca²⁺ transient and decreases during phase 3 when Ca²⁺ transient decreases and the cell repolarizes. This scenario corresponds to the shape of the whole‐cell 9‐phenanthrol‐inhibited current that we recorded on PCs, which develops rapidly during phase 2 and decreases slowly during phase 3.

Modulation of AP shape by TRPM4 in PCs

Our data indicate that 9‐phenanthrol reduces APD without affecting other AP parameters. As discussed above, an inward 9‐phenanthrol‐inhibited current can be unmasked during phase 2 and the first half of phase 3 of AP in PCs. This inward current contributes to membrane depolarization and thus prolongs the AP.

PFs are a major substrate of ventricular arrhythmias, for example during ischaemia (Wang & Hill, 2010). By significantly altering Ca²⁺ handling and oxidative phosphorylation, myocardial infarction leads to an increase in [Ca²⁺]_i and a decrease in [ATP]_i, thereby altering ionic currents (Carmeliet, 1999). Because it is activated by increases in subsarcolemmal [Ca²⁺], is inhibited by [ATP]_i (Nilius et al. 2004) and is highly expressed in the conduction system (Kruse et al. 2009), TRPM4 is a good candidate for triggering ventricular arrhythmias originating in PFs. In a previous study, 9‐phenanthrol was shown to reduce hypoxia‐reoxygenation‐induced arrhythmias in mouse ventricular preparations including the septum. These arrhythmias may arise from PFs and be transmitted to the ventricle. Because TRPM4 was found to be predominantly active in phase 2 and beginning of phase 3 PC APs, it would trigger early afterdepolarizations, as observed in the mouse ventricular model submitted to hypoxia‐reoxygenation (Simard et al. 2012).

Our results suggest that this Ca²⁺‐activated non‐selective cation channel should be integrated into mathematical models to improve modelling of PF APs. In addition, because the conductive tissue is a main source of cardiac arrhythmias, targeting TRPM4 to modulate PF activity could provide new cardioprotective therapies. However, the design of specific TRPM4 modulators remains a major caveat, and the broad expression profile of TRPM4 in cardiac and other physiologically significant tissues might generate relevant side effects.

Additional information

Conflict of interest

The authors state no conflict of interest.

Author contributions

T.H., R.R. and J.A. performed microelectrode recordings. T.H. and L.S. performed cardiac cell isolation. T.H. and R.G. performed single channel patch‐clamp recordings. T.H. and L.S. performed whole‐cell recordings. T.H., L.C. and R.R. performed biochemical experiments. T.H., R.G., L.S. and A.M. designed the experiments and wrote the paper. All authors approved the final version of the manuscript.

Funding

T.H. is a recipient of a fellowship from the French Ministère de l'Enseignement et de la Recherche. R.R. is a recipient of a fellowship from Région Basse‐Normandie. L.C., L.S., R.G. and R.R. are employed by the French Ministère de l'Enseignement et de la Recherche. A.M. and J.A. are employed by the French Ministère de la santé.