Low pH_o boosts burst firing and catecholamine release by blocking TASK‐1 and BK channels while preserving Cav1 channels in mouse chromaffin cells

Abstract

Key points

• Mouse chromaffin cells (MCCs) generate spontaneous burst‐firing that causes large increases of Ca²⁺‐dependent catecholamine release, and is thus a key mechanism for regulating the functions of MCCs.

• With the aim to uncover a physiological role for burst‐firing we investigated the effects of acidosis on MCC activity.

• Lowering the extracellular pH (pH_o) from 7.4 to 6.6 induces cell depolarizations of 10–15 mV that generate bursts of ∼330 ms at 1–2 Hz and a 7.4‐fold increase of cumulative catecholamine‐release.

• Burst‐firing originates from the inhibition of the pH‐sensitive TASK‐1‐channels and a 60% reduction of BK‐channel conductance at pH_o 6.6.

• Blockers of the two channels (A1899 and paxilline) mimic the effects of pH_o 6.6, and this is reverted by the Cav1 channel blocker nifedipine.

• MCCs act as pH‐sensors. At low pH_o, they depolarize, undergo burst‐firing and increase catecholamine‐secretion, generating an effective physiological response that may compensate for the acute acidosis and hyperkalaemia generated during heavy exercise and muscle fatigue.

Abstract

Mouse chromaffin cells (MCCs) generate action potential (AP) firing that regulates the Ca²⁺‐dependent release of catecholamines (CAs). Recent findings indicate that MCCs possess a variety of spontaneous firing modes that span from the common ‘tonic‐irregular’ to the less frequent ‘burst’ firing. This latter is evident in a small fraction of MCCs but occurs regularly when Nav1.3/1.7 channels are made less available or when the Slo1β2‐subunit responsible for BK channel inactivation is deleted. Burst firing causes large increases of Ca²⁺‐entry and potentiates CA release by ∼3.5‐fold and thus may be a key mechanism for regulating MCC function. With the aim to uncover a physiological role for burst‐firing we investigated the effects of acidosis on MCC activity. Lowering the extracellular pH (pH_o) from 7.4 to 7.0 and 6.6 induces cell depolarizations of 10–15 mV that generate repeated bursts. Bursts at pH_o 6.6 lasted ∼330 ms, occurred at 1–2 Hz and caused an ∼7‐fold increase of CA cumulative release. Burst firing originates from the inhibition of the pH‐sensitive TASK‐1/TASK‐3 channels and from a 40% BK channel conductance reduction at pH_o 7.0. The same pH_o had little or no effect on Nav, Cav, Kv and SK channels that support AP firing in MCCs. Burst firing of pH_o 6.6 could be mimicked by mixtures of the TASK‐1 blocker A1899 (300 nm) and BK blocker paxilline (300 nm) and could be prevented by blocking L‐type channels by adding 3 μm nifedipine. Mixtures of the two blockers raised cumulative CA‐secretion even more than low pH_o (∼12‐fold), showing that the action of protons on vesicle release is mainly a result of the ionic conductance changes that increase Ca²⁺‐entry during bursts. Our data provide direct evidence suggesting that MCCs respond to low pH_o with sustained depolarization, burst firing and enhanced CA‐secretion, thus mimicking the physiological response of CCs to acute acidosis and hyperkalaemia generated during heavy exercise and muscle fatigue.

Key points

• Mouse chromaffin cells (MCCs) generate spontaneous burst‐firing that causes large increases of Ca²⁺‐dependent catecholamine release, and is thus a key mechanism for regulating the functions of MCCs.

• With the aim to uncover a physiological role for burst‐firing we investigated the effects of acidosis on MCC activity.

• Lowering the extracellular pH (pH_o) from 7.4 to 6.6 induces cell depolarizations of 10–15 mV that generate bursts of ∼330 ms at 1–2 Hz and a 7.4‐fold increase of cumulative catecholamine‐release.

• Burst‐firing originates from the inhibition of the pH‐sensitive TASK‐1‐channels and a 60% reduction of BK‐channel conductance at pH_o 6.6.

• Blockers of the two channels (A1899 and paxilline) mimic the effects of pH_o 6.6, and this is reverted by the Cav1 channel blocker nifedipine.

• MCCs act as pH‐sensors. At low pH_o, they depolarize, undergo burst‐firing and increase catecholamine‐secretion, generating an effective physiological response that may compensate for the acute acidosis and hyperkalaemia generated during heavy exercise and muscle fatigue.

Abbreviations

AP

action potential

BCC

bovine chromaffin cell

CA

catecholamine

CC

chromaffin cell

CFE

carbon fibre electrode

DHP

dihydropyridine

DMEM

Dulbecco's modified Eagle's medium

ISI

interspike interval

LJP

liquid junction potential

MCC

mouse chromaffin cell

RCC

rat chromaffin cell

Introduction

Chromaffin cells (CCs) of the adrenal medulla undergo spontaneous firing at rest and respond to sustained depolarization with trains of action potentials (APs) that rapidly adapt their firing to higher frequencies (Nassar‐Gentina et al. 1988; Martinez‐Espinosa et al. 2014; Vandael et al. 2015 a). The molecular components of this phenomenon have been identified in mouse CCs (MCCs). AP firing is generated by fast inactivating TTX‐sensitive Nav1.3/Nav1.7 sodium channels that sustain the AP upstroke, whereas, at the same time, slowly inactivating L‐type Ca²⁺ channels (Cav1) contribute to the slow depolarization phase during prolonged interspike intervals (Vandael et al. 2015 b). AP repolarization is ensured by voltage‐gated K⁺ channels and by the differential coupling of voltage‐gated Cav channels to BK and SK channels (Marcantoni et al. 2010; Vandael et al. 2010; Vandael et al. 2012). Opening of BK and SK channels sets the shape and frequency of APs, as well as their mode of adaptation during sustained depolarizations (Vandael et al. 2015 a).

Recently, we reported that a reduction of Nav1.3/Nav1.7 channel availability as a result of slow inactivation during sustained depolarizations or block by TTX cause a sudden switch from tonic to burst firing with a consequent increase of Ca²⁺ influx during the burst and marked rise of catecholamine (CA) release in MCCs (Vandael et al. 2015 b). Burst firing also occurs upon deletion of the Slo1β2 subunits responsible for the fast inactivation of voltage‐ and Ca²⁺‐dependent BK channels (Martinez‐Espinosa et al. 2014) and is also evident in a small fraction (10–15%) of resting control MCCs (Martinez‐Espinosa et al. 2014; Vandael et al. 2015 b). Thus, CCs possess spontaneous ‘neuron‐like’ firing modes that boost the non‐neurogenic Ca²⁺‐dependent release of CAs when specific membrane conductances are modulated. This endogenous burst behaviour may represent a simple mechanism by which CCs and other neuroendocrine cells potentiate Ca²⁺ entry and hormone release during specific physiological stimuli. Given that bursting pacemaker activity and other patterns of similar electrical activity may arise from a number of distinct ionic conductances (Marder & Taylor, 2011), it would be extremely interesting to identify the existence of other ionic mechanisms that are able to induce burst firing in MCCs, besides a reduction of Nav channel availability and the deletion of Slo1β2 subunits (Lingle, 2015). Accordingly, it would be of key importance to determine whether burst firing in MCCs also arises during physiological stimuli causing robust membrane depolarization, such as blood acidosis, hyperkalaemia, elevation of histamine and increased levels of muscarine induced by splanchnic nerve stimulation (Neely & Lingle, 1992; Inoue et al. 1998; Wallace et al. 2002; Inoue et al. 2008; Mahapatra et al. 2011).

In the present study, we show that isolated MCCs respond to low extracellular pH (pH_o) with sustained depolarization, burst firing and enhanced CAs exocytosis. Lowering pH_o from 7.4 to 7.0 and 6.6 induces robust depolarizations that switch spontaneous tonic firing into regular bursts. Burst firing at low pH_o is a result of the inhibition of pH‐sensitive TASK‐1 and TASK‐3 ‘leak’ channels (Cotten, 2013; Bayliss et al. 2015) and Ca²⁺‐dependent BK channels (Prakriya & Lingle, 1999). The other channels responsible of AP firing in MCCs (Kv, Nav, Cav and SK) are weakly or not affected by low pH_o. Mixtures of the TASK‐1 channel blocker A1899 (Streit et al. 2011) and paxilline (a blocker of BK channels) can reproduce the effects of pH_o 6.6. By contrast, nifedipine can either revert regular bursts into tonic firing or block the firing, suggesting that Cav1 channels play a key role in the generation of the plateau potential of bursts in MCCs (Vandael et al. 2015 b).

Using amperometry, we also show that burst firing induced by acidic pH_o or mixtures of A1899 and paxilline causes a marked increase of CA cumulative release, which is mainly the result of an elevated rate of vesicle release. The similar action of low pH_o and mixtures of TASK‐1 and BK channel blockers suggests that the acute action of protons on CA release in MCCs is mainly a result of ionic conductance changes that boost Ca²⁺ entry during bursts rather than the specific effects of protons on the secretory apparatus (Jankowski et al. 1993).

Our data provide new evidence suggesting that burst firings in CCs probably comprise an effective mechanism that regulates the feedback response of adrenal glands to acute blood acidosis and hyperkalaemia by increasing circulating CA (Cryer, 1980; Medbo & Sejersted, 1990).

Methods

Ethical approval

Ethical approval was obtained for all experimental protocols from the University of Torino Animal Care and Use Committee, Torino, Italy. All experiments were conducted in accordance with the National Guide for the Care and Use of Laboratory Animals adopted by the Italian Ministry of Health. Every effort was made to minimize animal suffering and the number of animals used. For removal of tissues, animals were deeply anaesthetized with CO₂ inhalation and rapidly killed by cervical dislocation.

Cell culture

CCs were obtained from male C57BL/6J mice (Harlan, Correzzano, Italy) aged 2 months. Under sterile conditions, the abdomen was opened and the adrenal glands were isolated and transferred to ice cold Ca²⁺ and Mg²⁺ free Locke's buffer containing (in mm) 154 NaCl, 3.6 KCl, 5.6 NaHCO₃, 5.6 glucose and 10 Hepes (pH 7.4) (Marcantoni et al. 2009; Vandael et al. 2012). Under a dissecting microscope, the adrenal glands were decapsulated and subjected to an enzymatic dissociation with 20–25 units ml^–1 papain (Worthington Biochemical Corporation, Segrate, Italy) dissolved in Dulbecco's modified Eagle's medium (DMEM) (GIBCO, Invitrogen Life Technologies, Monza, Italy) supplemented with 1.5 mm of L‐cysteine, 1 mm of CaCl₂ and 0.5 mm of EDTA (Sigma‐Aldrich, Munich, Germany) for 25–30 min at 37 °C in a water saturated atmosphere with 5% CO₂. Afterwards, two washing steps were performed with DMEM supplemented with 1 mm CaCl₂ and 10 mg ml^–1 of BSA (Sigma‐Aldrich). Adrenal medullas were re‐suspended in DMEM containing 1% penicillin/streptomycin and 15% fetal bovine serum (both from Sigma‐Aldrich) and were mechanically dissociated with a fire polished Pasteur pipette. A drop (100 μl) of this concentrated cell suspension was plated on poly‐ornithine (1 mg ml^–1) and laminin (5 μg ml^–1) coated petri‐ dishes and, 30 min later, 1.9 ml of DMEM containing 1% penicillin/streptomycin and 15% fetal bovine serum (all from Sigma‐Aldrich) was added. The primary CC cultures were kept in an incubator at 37 °C and a water saturated atmosphere with 5% CO₂. Measurements were performed on cultured MCCs 2–5 days after plating.

APs and ion currents recordings

Macroscopic whole‐cell currents and APs were recorded in perforated‐patch conditions using a multiclamp 700‐B amplifier and pClamp, version 10.0 (Molecular Devices, Sunnyvale, CA, USA) (Marcantoni et al. 2010; Vandael et al. 2012). Traces were sampled at 10 kHz using a digidata 1440 A acquisition interface (Molecular Devices) and filtered using a low‐pass Bessel filter set at 1–2 kHz. Borosilicate glass pipettes (Kimble Chase Life Science, Vineland, NJ, USA) with a resistance of 2–3 MΩ were dipped in an Eppendorf tube containing intracellular solution before being back filled with the same solution containing 500 μg ml^–1 of amphotericin B (Sigma‐Aldrich), dissolved in DMSO (Sigma‐Aldrich) (Cesetti et al. 2003). Recordings were initiated after amphotericin B lowered the access resistance below 15 MΩ (5–10 min). Series resistance was compensated by 60–80% and monitored throughout the experiment. Fast capacitive transients during stepwise depolarizations (in voltage clamp mode) were minimized online by the use of the patch clamp analogue compensation. Uncompensated capacitive currents were further reduced by subtracting the averaged currents in response to P/4 hyperpolarizing pulses.

The normalized voltage‐dependent conductance of Nav channels (g _Na) was calculated with the equation: g _Na = I _Napeak/(V − V _rev), with V _rev equal to the reversal potential for Na⁺, and fitted with a Boltzmann function with variable V _1/2 (in mV) and k slope (in mV) (Carbone et al. 1997). The same procedure was carried out for g_Ca and g_BK.

Amperometric current recordings during low pH_o‐induced burst firing

Simultaneous detection of amperometric currents associated with CA release and AP recordings was performed using a HEKA EPC‐10 double amplifier. For amperometry, we used standard carbon fibre microelectrodes (CFEs) with a tip diameter of 5 μm polarized at +800 mV (ALA Scientific Instruments Inc., Westbury, NY, USA) (Carabelli et al. 2007; Marcantoni et al. 2009; Vandael et al. 2015 b) . For the current clamp AP recordings, we used glass pipettes and perforated patch conditions as described above. The CFE was first placed sidewise adjacent to the cell, taking care to leave part of the cell surface accessible to the glass pipette for recording APs. The glass pipette was positioned opposite to the CFE on the free available side of the cell. When recording amperometric signals, the spontaneous APs that appeared as ‘irregular tonic’ firing at pH_o 7.4 were converted to ‘burst’ firing by either lowering the pH_o to 6.6 or adding mixtures of A1899 (300 nm) and paxilline (300 nm) to the external solution. Amperometric currents were sampled at 4 kHz and low‐pass filtered at 1 kHz. Data were analysed using IGOR macros Quanta analysis (WaveMetrics, Lake Oswego, OR, USA) as described previously (Carabelli et al. 2007). The analysis of individual exocytotic events was performed by measuring the parameters: maximum oxidation current (I _max), spike width at half‐height (t _{1 / 2}), total charge of the spike (Q), cubic root of Q (Q ^1/3) and time to reach the spike (t_p). All experiments were performed at room temperature.

Solutions

Intracellular solution for current clamp and Na⁺ and K⁺ current measurements in voltage clamp or AP clamp mode comprised (in mm): 135 KAsp, 8 NaCl, 2 MgCl₂, 5 EGTA, 20 Hepes (pH 7.4) (with KOH; Sigma‐Aldrich). For Ca²⁺ current recordings, the intracellular solution contained (in mm): 135 Cs‐MeSO₃, 8 NaCl, 2 MgCl₂, 5 EGTA and 20 Hepes (pH_o 7.4) (with CsOH; Sigma‐Aldrich). The extracellular solution used for current clamp measurements is a physiological Tyrode's solution containing (in mm): 130 NaCl, 4 KCl, 2 CaCl₂, 2 MgCl₂, 10 glucose and 10 Hepes (pH_o 7.4) (with NaOH; Sigma‐Aldrich). The same solution was used to measure K⁺ currents. K_V currents were obtained by adding 500 μm Cd²⁺ to the external solution, whereas Ca²⁺‐activated BK currents were estimated by subtracting K_V from the total K⁺ currents. As noted previously (Vandael et al. 2015 b), residual Cd²⁺‐insensitive voltage‐dependent BK currents contribute little (< 5%) to the total BK currents at +20 to +30 mV (Berkefeld & Fakler, 2013). Thus, isolation of BK currents using 500 μm Cd²⁺ appears to be a reliable protocol. The extracellular solution used for Na⁺ current measurements comprised (in mm): 104 NaCl, 30 TEACl, 4 KCl, 2 CaCl₂, 2 MgCl₂, 10 glucose and 10 Hepes (pH_o 7.4) (with NaOH). The extracellular solution used for Ca²⁺ current measurements in voltage clamp configuration contained (in mm) 135 TEACl, 2 CaCl₂, 2 MgCl₂, 10 glucose and 10 Hepes (pH_o 7.4) (with TEA‐OH; Sigma Aldrich). The extracellular pH_o of the bath solutions was adjusted individually by adding HCl to reach the values: 7.2, 7.0, 6.8 and 6.6. The addition of HCl to obtain pH_o 6.6 increased the osmolality of the solution by < 2%. This small osmolality change was found to have no effect on cell excitability.

The liquid junction potential (LJP) of the solutions was calculated using JPCalcWin (Clampex, version 10.5; Axon Instruments, Foster City, CA, USA) that is derived from the original software package of Barry (1994). For the solutions used, the uncompensated LJP was +12.6 mV for the current clamp and K⁺ current, +13.5 mV for the Na⁺ current and +16.6 mV for the Ca²⁺ current measurements at 22 °C. These values should be further corrected for the Donnan equilibrium potential that generates at the perforated patch (V _pf) (Horn & Marty, 1988). V _pf was estimated to be in the order of 2.6 mV for the current clamp, K⁺ and Na⁺ currents and 3.4 mV for the Ca²⁺ currents, and was subtracted from the LJP. Following the subtraction, the uncompensated LJPs were +10, +10.9 and +13.2 mV for the three cases indicated above. Because the addition of the drug tested has almost no effect on the LJP and lowering of pH_o with HCl causes voltage changes of ∼1 mV, the membrane potentials of our current and voltage clamp recordings were not corrected for the different LJPs.

Testing the effects of a low pH_o required usually 1 min to reach steady‐state conditions. Washout was also rapid when testing the effects of pH_o on Nav, Cav, Kv and SK channels given the small effects observed. The recovery was significantly slower (3–6 min) when washing out the effects of pH_o on BK currents, probably because the inhibitory effects on these channels are mainly the result of a lowering of the intracellular pH (see Discussion) (Kume et al. 1990).

TASK‐1 and TASK‐3 blockers (PK‐THPP and A1899) (Streit et al. 2011; Coburn et al. 2012) were purchased from Aberjona Laboratories (Woburn, MA, USA), dissolved in DMSO and stored at −20 °C. Nifedipine and BayK8644 were obtained from Sigma‐Aldrich. The two dihydropyridines (DHPs) were dissolved, stored and used as described previously (Carabelli et al. 2001).

Statistical analysis

Data are given as the mean ± SEM for the number (n) of cells. Statistical significance was estimated using either paired/unpaired Student's t tests or one‐way ANOVA followed by a Bonferroni post hoc test in cases where two or multiple groups of measurements had to be compared. P ≤ 0.05 was considered statistically significant. Statistical analysis was performed with SPSS, version 20.0 (IBM Corp., Armonk, NY, USA). Off‐line data analysis was performed with pClamp and Origin (OriginLab Corporation, Northampton, MA, USA).

Results

Low extracellular pH induces cell depolarization and burst firing in MCCs

The first goal was to assess how lowering the pH_o from 7.4 to 6.6 altered both the resting potential and the spontaneous firing modes of MCCs in current clamp conditions with no current injection (Fig. 1 A). Slight changes in pH_o (from 7.4 to 7.0) were sufficient to evoke tonic firing of increased frequency (from 0.43 ± 0.07 Hz to 1.36 ± 0.33 Hz; n = 9, P < 0.01, paired Student's t‐test) (Fig 1 A and B), as associated with ∼7 mV cell depolarization of resting potential (V _rest) (Fig. 1 C). Lower pH_o caused increased cell depolarization and resting firing frequency. Mean frequency was 2.05 Hz at pH_o 6.8 and 2.52 Hz at pH_o 6.6 (Fig. 1 B). The pH‐dependence of V _rest followed a dose–response curve with −35.1 mV at pH_o 6.6, −51.0 mV at pH_o 7.6 and IC₅₀ = 7.2 (Fig. 1 C). Thus, a ∆pH excursion from 7.6 to 6.6 was sufficient to depolarize the MCCs by ∼16 mV (Fig. 1 C) and result in an increased rate of firing from 0.15 Hz to 2.5 Hz (Fig. 1 B). The effects of lowering the pH_o were always tested after recording spontaneous MCCs activity at pH_o 7.4 for 60 to 90 s to ensure stable cell activity. Resting depolarization and increased firing frequency induced by a low pH_o required usually 20–40 s to reach steady‐state values and were fully reversible after 2–3 min of continuous washing (Fig. 1 D). On average, MCC firing activity was stable for 5 to 6 min (occasionally for 10 min), which is a sufficient time lapse for testing the effects of a low pH_o. Cells with fluctuating resting potential (∆V _rest = ±8 mV), with APs below 0 mV or with AP firing often interrupted by silent periods longer than 20 s were disregarded from the analysis.

Figure 1. Low pH_o effects on spontaneous firing in MCCs.

A, top: representative trace of a current clamped spontaneously firing MCC (no current injection) at pH_o 7.4, 7.0 and 6.6. Bottom: AP recordings on an expanded time scale corresponding to the grey window above. A decrease in pH results in depolarization and the switch of firing modes from tonic (pH_o 7.4, grey rectangle to the left) to mildly bursting (pH_o 7.0, grey rectangle in the centre), to sustained bursting (pH_o 6.6, grey rectangle to the right). Intermittent and sustained burst firing are accompanied by a net decrease of AP peak amplitude associated with the slow inactivation of Nav channels at depolarized potentials (Vandael et al. 2015 b). The dotted line indicates the 0 mV level. Dashed lines indicate V _rest at the differing pH_o values. V _rest was determined by averaging the slowly rising potential during the interspike interval. B, mean firing rate at differing pH_o values obtained from nine MCCs. C, mean V _rest vs. pH_o. The continuous curve is a dose–response best fit with equation: V = V _min + [(V _max – V _min)/$1+{10}^{(\mathrm{pH}-{\mathrm{IC}}_{50})\mathrm{n}}$] with V _min = −53 mV, V _max = −34 mV, IC₅₀ = 7.2 and Hill slope n = 2.2 (n = 9 cells). D, representative trace of the stability of the control firing recording and reversible effects of pH_o 6.6. The acidic solution was applied for 60 s to induce burst firing and then washout to rescue the initial tonic firing frequency. The V _m of current clamp recordings was not corrected for the LJPs (see Methods). [Color figure can be viewed at wileyonlinelibrary.com]

Lowering the pH_o did not cause only a simple increase of firing frequency. Starting from pH_o 7.0, we observed that spontaneous firing of MCCs switched from ‘irregular tonic’ to ‘burst’ firing as already reported during sustained depolarization or partial blockade of Nav1.3/Nav1.7 channels in MCCs (Vandael et al. 2015 a; Vandael et al. 2015 b). The probability of observing burst firing increased with a lowering of the pH_o to reach almost permanent burst‐like firing conditions at pH_o 6.6 (Fig. 1 A). The switch from tonic to burst firing was particularly evident in the distribution of the interspike interval (ISI) duration. At pH_o 7.4 and 7.2, the distribution had a broad range of ISI values (from 0.03 to 3 s) reflecting the irregular tonic firing of the cells (Fig. 2). At pH_o < 7.0, two separate Gaussian distributions with distinct peaks became evident. The first peak (brief durations) represents the ISI between two consecutive spikes within a burst (intra‐burst interval), whereas the second peak at longer times represents the ISI between consecutive bursts (inter‐burst interval) (Vandael et al. 2015 b). The mean intra‐burst interval was 32, 17 and 9 ms at pH_o 7.0, 6.8 and 6.6, whereas the Gaussian distribution of the inter‐burst intervals peaked at 394, 360 and 331 ms at pH_o 7.0, 6.8 and 6.6, respectively. When we plotted each ISI (ISI_i) against its successive ISI (ISI_i+1), we could distinguish a clear difference between pH_o 7.4 and pH_o < 7.0 (Fig. 2, bottom). In these ISI_i/ISI_i+1 graphs with the ISI_i plotted along the x‐axis and the ISI_i+1 plotted along the y‐axis, the irregular ‘tonic’ firing patterns gave rise to a random distribution of events, whereas a moderate burst firing gave rise to ‘L‐shaped’ distributions. This is clearly visible at pH_o 7.0 and pH_o 6.8, where the ‘L‐shaped’ distribution is best evident. At pH_o 6.6, burst firing was more sustained, leading to a shorter inter‐burst interval that results in a less well‐resolved ‘L‐shaped’ distribution.

Figure 2. Analysis of tonic and burst firing patterns at different pH_o values.

Top: representative AP recordings at the indicated pH_o. Middle: ISI distribution for the indicated pH_o from a number of cells varying from n = 22 (pH_o 7.4) to n = 7 at pH_o 7.2. The continuous curves are best fits with double Gaussian functions with means (first, second) indicated. Bottom: joint ISI graphs obtained by plotting each ISI on the x‐axis (ISI_i) against its successive ISI duration (ISI_i+1) on the y‐axis. Irregular firing at various pH_o leads to a cloudy pattern, whereas burst firing at very low pH_o leads to an L‐shaped distribution. The firing characteristics of the cells (i.e. the interspike interval, the AP frequency and the threshold) were calculated as described previously (Vandael et al. 2012, 2015 b).

Block of TASK‐1 and TASK‐3 leak channels causes cell depolarization and burst firing in MCCs

Recent work suggests that CC depolarization induced by low pH_o is mainly associated with the blockade of the two‐pore TASK‐1 and TASK‐3 K⁺ channels family (Inoue et al. 2008). Following this, we tested whether TASK‐1 and TASK‐3 blockers (PK‐THPP and A1899) induce MCCs sustained depolarization and generate burst firing similar to the low pH_o. PK‐THPP is reported to block TASK‐3 channels with high affinity (IC₅₀ = 35 ± 5 nm) (Coburn et al. 2012) and TASK‐1 with an almost 10‐fold lower affinity (IC₅₀ = 300 ± 20 nm), whereas A1899 is reported more selective for TASK‐1 channels (IC₅₀ = 35 ± 3.8 nm) and less selective toward TASK‐3 (IC₅₀ = 318 ± 30 nm) (Streit et al. 2011). Given this, we tested the blocking effects of both compounds at 300 nm, aiming to achieve an ∼50% blockade of the channel with lower affinity and almost complete blockade of the channel with higher affinity (Fig. 3). Specifically, in the case of 300 nm A1899, we expected 50% blockade of TASK‐3 and 100% blockade of TASK‐1 and the opposite with PK‐THPP. We found that 300 nm A1899 caused a mean depolarization of 4.5 mV (from −48.2 mV, n = 11 to −43.7 mV, n = 15; P < 0.001, one‐way ANOVA) (Fig. 3 C) and a net AP frequency increase from 0.83 to 1.74 Hz (P < 0.001, one‐way ANOVA) (Fig. 3 C) and brief periods of burst firing activity (Fig. 3 A, grey insets). PK‐THPP produced remarkably lower depolarizations and a less moderate increase of AP firing that were not statistically significant (P > 0.05, one‐way ANOVA). Prolonged applications of A1899 (300 nm) and the addition of PK‐THPP (300 nm) that lasted 2–3 min caused only slight increased depolarizations (−42.6 mV; n = 7) and firing frequency (1.89 Hz; n = 7) and was not statistically different from the effects of A1899 alone (P > 0.05, one‐way ANOVA) (Fig. 3 D). The reversibility of A1899 action was not tested systematically because the drug was never washed during all the experiments (see below). In a few cells, we found a significant recovery that could not be complete as a result of the deterioration of conditions in cells at the end of the experiment.

Figure 3. Effects of the TASK‐1 and TASK‐3 blockers A1899 and PK‐THPP on MCC firing.

A, representative traces of spontaneously firing MCCs under control conditions (pH_o 7.4) and during bath application of 300 nm PK‐THPP (upper trace) or A1899 (lower trace) as indicated by the horizontal bars. On top of each trace are shown 5 s of recordings on an expanded time scale corresponding to the grey window below. The dotted line indicates the 0 mV level. Dashed lines indicate V _rest at control (pH_o 7.4) and with the blocker. Note the difference between the effects on firing rates and modes of the two blockers. B and C, mean frequencies (Hz) and resting potential (mV) under control conditions (dark bars) during the addition of 300 nm PK‐THPP, 300 nm A1899 or simultaneous application of the two blockers (^** P < 0.01; ^*** P < 0.001; one‐way ANOVA followed by a Bonferroni post hoc test). D, representative trace of a spontaneously firing MCC at pH_o 7.4 to which A1899 (300 nm) was applied for 2 min alone and PK‐THPP (300 nm) was added later to test the simultaneous effects of the two drugs. On top are shown 5 s of recordings on an expanded time scale corresponding to the grey window below. The V _m of current clamp recordings was not corrected for the LJPs (see Methods).

Accurate analysis of ISI distribution showed that 300 nm A1899 or mixtures of 300 nm A1899 + 300 nm PK‐THPP caused burst firings that were similar to pH_o 7.0 but significantly different from pH_o 6.6 (see below). In conclusion, blockade of TASK‐1/TASK‐3 channels by either A1899, PK‐THPP or both does not account for the marked cell depolarization and sustained burst firing induced by lowering the pH_o below 7.4. We therefore hypothesized that low pH_o could induce marked resting depolarizations and sustained burst firing by also inhibiting other K⁺ channels (BK, SK and Kv), at the same time as preserving or mildly affecting the Cav or Nav channels that sustain the inward currents underlying the bursts (Marcantoni et al. 2010; Vandael et al. 2015 b). We thus studied the effects of pH_o on the ionic conductances sustaining the spontaneous AP firing of MCCs with the aim of identifying the most sensitive K⁺ channel whose reduced permeability or altered voltage‐ and Ca²⁺‐dependence activation could favour membrane depolarization and burst firing modes.

Nav, Cav and SK currents are little or not affected by lowering pH_o

We started by testing the effects of low pH_o on the two major ionic components controlling cell depolarization: Nav and Cav currents. We also decided to limit the pH_o test to 7.0, which is within the physiological pH range and the minimal ∆pH able to induce burst firing. As shown in Fig. 4 A, pH_o 7.0 caused almost no change in the amplitude and time course of inward Nav currents activated from −40 to +60 mV. The time difference to reach 90% and 10% of the peak amplitude t _(90%–10%), was taken as an estimate of channel activation and appeared almost identical between –20 and 0 mV (Fig. 4 A, left inset). The same was true for the time constant of inactivation (τ_inact) (Fig. 4 A, right inset), whose values correspond to the τ_inact values of Nav1.3 channels (Catterall et al. 2005). Mean peak Nav currents at 0 mV were also not significantly different: −1.89 ± 0.22 nA and −1.96 ± 0.19 nA at pH_o 7.4 and 7.0, respectively (Fig. 4 B, inset). The voltage‐dependence of normalized I/V and channel conductance curves at pH_o 7.0 was almost identical to pH_o 7.4 (mean V _½ = −17.6−mV at pH_o 7.4 and V _½ = –16.6 mV at pH_o 7.0) (Fig. 4 B and C).

Figure 4. Lowering pH_o to 7.0 has minimal effects on Nav currents in MCCs.

A, whole cell Nav currents evoked by depolarizing pulses lasting 10 ms in steps of 10 mV from −40 to +60 mV at pH_o 7.4 (black traces) and 7.0 (red traces). V _h was –70 mV. Insets: mean t _{(90% –10%)} and τ_inact at −20, −10 and 0 mV at pH_o 7.4 (black dots) and pH_o 6.6 (red dots). B and C, normalized I_Na current amplitudes and g _Na vs. voltage at pH_o 7.4 (black squares) and pH_o 7.0 (red squares) (n = 7). g _Na(V) was calculated as described in the Methods. In (B), the normalized I/V curves are continuous lines drawn through data points. Inset: mean I_Na peak values (n = 9) at 0 mV at pH_o 7.4 (black bar) and 7.0 (red bar). In (C), the two continuous curves are the results of the fit with two Boltzmann equations with half‐maximal values V _½ (in mV) and slope factors k (in mV) obtained from the fit: 17.6 and 4.9 mV (pH_o 7.4; black curve) and 16.6 and 4.8 mV (pH_o 7.0; red curve). The V _m of voltage clamp recordings was not corrected for the LJPs (see Methods).

Minimal gating changes occurred also on voltage‐gated Cav channels at pH_o 7.0. Cav currents elicited during pulses of 30 ms from −40 to +60 mV had almost the same activation time course, whereas the amplitude was slightly smaller during pulses <0 mV (Fig. 5 A) but equal or larger at positive potentials (not significantly different at all voltages). Mean peak amplitude at 0 mV was: −101.6 ± 24.0 pA and −114.6 ± 21.8 pA at pH_o 7.4 and 7.0, respectively (Fig. 5 B, inset). At pH_o 7.0, the voltage‐dependence of normalized I/V and channel conductance curves were shifted by ∼4 mV to the right, as expected from the Ca²⁺‐induced surface charge screening of high‐threshold Ca²⁺ channel activation described in other cells (Zhou & Jones, 1996) (Fig. 5 B and C). The shift increased proportionally to the pH_o value: it was ∼2 mV at pH_o 7.2, and ∼6 mV at pH_o 6.8.

Figure 5. Lowering pH_o to 7.0 has little effect on calcium current amplitude.

A, overlapped whole cell recordings of total Ca²⁺ currents at the test potentials indicated at pH_o 7.4 (black traces) and pH_o 7.0 (red traces; V _h = −70 mV). B, current–voltage relationship at pH_o 7.4 (black dots) and pH_o 6.6 (red squares) of normalized Ca²⁺ currents. Inset: mean I_Ca peak values (n = 8) at 0 mV at pH_o 7.4 (black bar) and 7.0 (red bar). C, normalized Ca²⁺ channels conductance fit with a Boltzmann function: V _1/2 = –16.2 mV, k = 7.1 mV at pH_o 7.4 (black dots and trace) and V _1/2 = −12.0 mV, k = 7.0 mV for pH_o 7.0 (red squares and trace; n = 8). Note the 4.2 mV shift of g _Ca(V) to the right at low pH_o. g _Ca(V) was calculated as described in the Methods. D, superimposed normalized whole‐cell recordings of total and L‐type calcium currents at −10 mV at pH_o 7.4 (black trace) and pH_o 6.6 (red trace) lasting 300 ms to estimate the time course of current inactivation. L‐type currents were obtained by subtraction of nifedipine (3 μm) resistant current from control traces. The continuous yellow curves within traces of total Cav currents are double‐exponential fits with the parameters: A _fast = −22.2 pA, A _slow = −24.2 pA, τ_fast = 16.4 ms, τ_slow = 229.3 ms, C = −32.4 pA at pH_o 7.4 and A _fast = −25.3 pA, A _slow = −23.5 pA, τ_fast = 18.6 ms, τ_slow = 208.6 ms, C = −29.5 pA at pH_o 7.0. The continuous yellow curves within traces of L‐type currents are single exponential fits with the parameters: A _fast = −19.5 pA, τ_fast = 39,9 ms at pH_o 7.4 and A _fast = −19.0, τ_fast = 32.5 ms, at pH_o 7.0. The V _m of voltage clamp recordings was not corrected for the LJPs (see Methods).

Given that burst firing is sustained mainly by slowly inactivating Cav1 channels carrying sufficient inward current during plateau potentials of 300 to 400 ms to −20 mV (Vandael et al. 2015 b), we also tested whether the total and L‐type (nifedipine‐sensitive) Ca²⁺ currents had altered kinetics during step depolarization of 300 ms at −20 and –10 mV. Inward Ca²⁺ currents had only a slight decrease (10–15%; not significant) at pH_o 7.0, as expected from the right shift of their voltage‐dependent activation, although they had almost the same time course of inactivation of control currents. This was evident by comparing the normalized traces of pH_o 7.4 and 7.0 at −10 and −20 mV. As shown in Fig. 5 D, total and L‐type current traces (estimated after subtraction for nifedipine‐insensitive currents) at pH_o 7.4 and 7.0 are almost undistinguishable at both potentials (black and red traces). Double‐exponential fits of the averaged normalized traces at control (continuous yellow curves in Fig. 5 D) indicate the presence of a fast and slow inactivating component with similar time constants and baseline at both pH_o. Single exponential fits of L‐type currents also had a similar amplitude, time constant and baseline. The amplitude and time constants of the fast (A _f, τ_f) and slow component (A _s, τ_s) and baseline values (C) of the curve fit at pH_o 7.4 and 6.6 are given in Fig. 5 D. Similar data were obtained in five other MCCs.

Low pH_o reduces BK channel conductance and has little effect on K_V and SK currents

Regardless of the little effects of pH_o on Cav currents, the Ca²⁺ and V‐dependent BK channel currents were markedly attenuated when lowering the pH_o (Fig. 6). To test the action of a low pH_o, we first determined the Ca²⁺‐dependence of BK channel activation by measuring the BK currents activated by 100 ms pre‐loading steps of variable voltage (from –60 to +120 mV by steps of 20 mV) to inject variable quantities of Ca²⁺ ions. BK currents were measured at +120 mV to induce maximal BK channel activation (Gavello et al. 2015). We first applied pH_o 7.4 and then pH_o 7.0 (Fig. 6 A). The addition of Cd²⁺ (500 μm) to block the BK current component and determine the size of Cd²⁺‐insensitive K_V currents was performed after full recovery of the current at pH_o 7.4 that required 3 min on average, most probably because of an equal lowering of intracellular pH (pH_i) (data not shown). Subtraction of the traces in Cd²⁺ from the traces at pH_o 7.4 and 7.0 ultimately led to the BK current (Fig. 6 B). As shown in the example in Fig. 6 A, lowering the pH_o from 7.4 to 7.0 caused an almost 40% reduction of BK currents at all Ca²⁺ pre‐loading steps injecting large amounts of Ca²⁺ ions into the cells (–20 to +40 mV). BK current peaks measured at pH_o 7.4 followed the bell‐shaped curve expected for BK channels (Marty & Neher, 1985; Neely & Lingle, 1992) and the same bell‐shape of lower amplitude was evident at pH_o 7.0 (Fig. 6 C). No shift of the peak values was evident, indicating that the effect of lowering pH_o cannot be attributed to a specific effect on L‐ or non‐L‐type Ca²⁺ channels.

Figure 6. Lowering pH_o to 7.0 reduces markedly the size of BK currents.

A, total K⁺ currents (K_V + BK) measured in accordance with the protocol shown on top, at pH_o 7.4 (black traces) and pH_o 6.6 (red traces). B, pulse protocol used to determine the size of BK currents from total K⁺ currents. The blue trace with 500 μm Cd²⁺ is obtained with no pre‐step depolarization to +20 mV and is used to determine the size of BK currents at pH_o 7.4 (black trace) and pH_o 7.0 (red trace) as indicated. C, dependence of BK currents on pre‐conditioning voltage at pH_o 7.4 (black dots) and pH_o 7.0 (red triangles). Note the typical bell‐shaped I/V curves expected for BK channels at both pH_o values. The V _m of voltage clamp recordings was not corrected for the LJPs (see Methods).

We then determined the V‐dependence of BK channel activation by measuring the BK currents after 100 ms pre‐pulse to +20 mV and test pulses of increasing amplitude from –60 to +120 mV in steps of 20 mV (Fig. 7 A). Lowering the pH_o caused a marked decrease of BK currents that was quantified by subtracting the K_V currents remaining after application of 500 μm Cd²⁺. Accordingly, the voltage‐dependence of the normalized BK conductance (g _BK) was determined and data were fit by a Boltzmann function. As expected, pH_o 7.0 decreased by ∼40%, which is the maximal conductance compared to control (n = 8; P < 0.01, paired Student's t test). There was also a steeper V‐dependence, with a decrease in the slope factor k in the Boltzmann equation from 33.8 mV (pH_o 7.4) to 28.4 mV (pH_o 7.0) for an e‐fold change and a marked left shift of the half‐maximal activation voltage (V _1/2 = −61.0 and −49.9 mV, respectively). The two effects probably originate from the interaction of protons with the intracellular Ca²⁺ bowl sensor (Hou et al. 2009) and compensate each other to give a percentage blockade of ∼40% at almost all potentials. Note that the left shift of g _BK at low pH_o is opposite to that of voltage‐gated Na⁺, Ca²⁺ and K⁺ channels, suggesting a different action of protons on BK channel gating (see Discussion).

Figure 7. Lowering pH_o reduces markedly the size of BK currents at different voltages with little effect on the size of KV and SK currents.

A, BK currents measured in accordance with the protocol shown at pH_o 7.4 (black traces) and pH_o 7.0 (red traces) after subtracting the K_V currents remaining after 500 μm Cd²⁺ application. B, current peak amplitudes were used to calculate the conductance (g _BK) vs. voltage at pH_o 7.4, 7.0, 6.8 and 6.6, as indicated. g _BK(V) was calculated as described in the Methods. Data were fit by a Boltzmann equation with V _1/2 = 61.0 mV and k = 33.8 mV at pH_o 7.4, V _1/2 = 49.9 mV and k = 28.4 mV at pH_o 7.0, V _1/2 = 39.2 mV and k = 22.5 mV at pH_o 6.8, V _1/2 = 36.9 mV and k = 21.0 mV at pH_o 6.6. C, pH_o dependence of BK currents normalized at pH_o 7.4, calculated from the maximal g _BK values at +120 mV. The continuous curve is a dose–response best fit with equation: % I _BK = % I _BKmax/$1+{10}^{(\mathrm{pH}-{\mathrm{IC}}_{50})\mathrm{n}}$ with % I _BKmax = 126, IC₅₀ = 7.0 and Hill slope n = 1.26. Data points are normalized to 7.4. D, K_V currents recordings at pH_o 7.4 (black trace) and pH 7.0 (red trace) at +90 mV in the presence of 300 nm TTX, 500 μm Cd²⁺, 1 μm paxilline and 200 nm apamin to block Nav, Cav, BK and SK channels. E, current peak amplitudes were used to calculate the conductance (g _K) vs. voltage at pH 7.4 (black squares) and 7.0 (red dots). F, SK tail currents elicited by the protocol shown on top at pH_o 7.4 (black trace) and pH_o 7.0 (red trace) and during the addition of 200 nm apamin (blue trace). Inset at bottom: mean amplitude of SK tail currents at pH_o 7.4, pH_o 7.0 and with 200 nm apamin as indicated (n = 6 cells). Tail currents amplitude was estimated 20 ms after the onset of the hyperpolarization to –100 mV, when K_V channels are deactivated. The V _m of voltage clamp recordings was not corrected for the LJPs (see Methods).

Given the strong effect of pH_o on g _BK, we also investigated its pH_o sensitivity over a wider range of pH_o (from 7.8 to 6.6). As shown in Fig. 7 C, BK currents decrease with lowering pH_o with an IC₅₀ at pH 7.0 and a Hill coefficient of 1.26, indicating one proton binding to the site controlling g _BK permeability. At pH_o 6.6, the BK current blockade is 65% at almost all potentials, with an ∼−24 mV left shift of V _1/2. This suggests the strong depressive action of a low pH_o on BK channels probably because of a dual action on channel permeability (Brelidze & Magleby, 2004), as well as the Ca²⁺‐sensitivity of channel opening (Hou et al. 2009) through the titration of amino acid residues with an IC₅₀ near neutral pH_o.

To complete the test on most expressed ion channels contributing to AP firing, we also checked the action of a low pH_o on voltage‐gated K_V currents measured in the presence of Nav, Cav, BK and SK channel blockers (Fig. 7 D). At variance with BK, the K_V currents exhibited steep V‐dependence between −20 and +30 mV and were only slightly reduced at pH_o 7.0. On average, K_V currents were depressed by ∼10% at voltages of maximal activation (+60 mV), with no evident shift to the voltage of half‐maximal activation (Fig. 7 E).

As previously shown, SK channels are functionally coupled to Cav1.3 channels in MCCs and the control resting membrane potential, firing frequency and terminate burst firing (Vandael et al. 2012). Given that total Cav currents and, particularly, Cav1 currents were affected only slightly at pH_o 7.0 (Fig. 5 D), we tested whether SK channels also undergo gating changes at acidic pH_o. Fig. 7 F shows that, when fully activated by Ca²⁺‐loading steps of 250 ms to +20 mV, the amplitude and time course of the slowly decaying inward tail SK currents during step repolarization to −100 mV were almost unchanged at pH_o 7.0. This suggests that lowering the pH_o does not alter the number of functioning SK channels and their Ca²⁺‐dependence.

Mixtures of BK and TASK‐1 channel blockers mimic the action of low pH_o

Given the strong blocking effects of low pH_o on g _BK (Figs 6 and 7) and the key role that BK channels play in the repolarization phase of APs in MCCs (Marcantoni et al. 2010; Martinez‐Espinosa et al. 2014), we hypothesized that combined blockade of BK and TASK channels (Fig. 3) could account for most of the action of low pH_o on burst firings. To confirm this, we first determined the concentration of paxilline that blocks 50–70% of BK channels (Zhou & Lingle, 2014), thus reproducing the effects of low pH_o. Using increasing concentrations (0.1 to 1.0 μm), we found that, by holding MCCs at –70 mV, 100 and 300 nm, paxilline induced 70% and 100% blockade of BK currents after almost 180 and 60 s, respectively (Li & Cheung, 1999). Given that paxilline blocks BK channels exclusively in their closed state (Zhou & Lingle, 2014) and that the rate and degree of block is largely attenuated at more positive voltages, we expected that 100 and 300 nm could well fit the blocking conditions of BK channels at pH_o 6.6 when MCCs are free to fire between a V _rest of –48 mV and mean overshoots of +20 mV. We then tested whether mixtures of paxilline and A1899 could mimic the sustained burst firing induced by low pH_o. We found that the best burst firing conditions were obtained when 300 nm paxilline was added to 300 nm A1899. Figure 8 shows how the addition of 300 nm A1899 to a spontaneously tonic firing MCC (Fig. 8, top left, grey) causes mild cell depolarization (∼6 mV) and intermittent burst firing activity (Fig. 8, top middle). Further addition of 300 nm paxilline converts the intermittent bursts into more regular, well‐organized bursts (Fig. 8, top right). As reported previously (Marcantoni et al. 2010), paxilline caused only small depolarization when applied on spontaneously firing MCCs (2–3 mV). These small depolarizations that sum to the depolarizing effects of TASK‐1/TASK‐3 blockers may account only partially for the net increase of burst firing activity. The major action of paxilline on cell firing is probably associated with the blockade of BK currents that sustain AP repolarization and counter balance the inward currents sustaining the plateau potential of bursts.

Figure 8. Mixtures of TASK‐1 and BK channel blockers favor the switch from tonic to sustained burst firing in MCCs.

Bottom: representative current clamp trace of a spontaneously firing MCCs before and during bath application of 300 nm A1899 (TASK‐1 blocker) alone and after addition of 300 nm paxilline (BK channel blocker). The black dotted line indicates the 0 mV level. White dashed lines indicate V _rest at control (pH_o 7.4) and with 300 nm A1899. Top: time expanded recordings corresponding to the grey windows indicated below. The switch from tonic (left) to intermittent (centre) and sustained burst firing (right) is evident. The V _m of current clamp recordings was not corrected for the LJPs (see Methods).

Quantitative analysis of burst firing at pH_o 6.6, with A1899 and paxilline alone or mixtures of the two blockers at pH_o 7.4 allowed direct comparison of the different firing conditions. Therefore, we estimated the mean ± SEM of six parameters that best define burst firing properties: number of events per burst, burst duration, peak amplitude of first AP, peak amplitude of last AP, plateau amplitude and bursts frequency (Fig. 9). Restrictive tests of significance (P < 0.01, one‐way ANOVA) of all the parameters at pH_o 6.6 compared to those estimated for 300 nm A1899, 1 μm paxilline and 300 nm A1899 + 300 nm paxilline indicate that the best mimicking firing conditions are those induced by the mixture of the two blockers (Fig. 9, second column to the left). Under these conditions, five out six parameters were not significantly different from those at pH_o 6.6, whereas paxilline and A1899 alone changed more parameters (two out of six parameters changing significantly). In conclusion, blocking of TASK or BK channels alone is sufficient to produce burst firing similar to that induced by sustained cell depolarization. This latter induces a slow inactivation of the Nav channels that is sufficient to induce burst firing in MCCs (Vandael et al. 2015 b). However, simultaneous blockade of TASK and BK channels ensures more stable and regular burst firings. Critical to this is the blockade of BK channels that regulate the strength and duration of the fast after‐hyperpolarization phase of APs (Prakriya & Lingle, 1999; Marcantoni et al. 2010; Vandael et al. 2010; Martinez‐Espinosa et al. 2014).

Figure 9. Comparative analysis of burst firing parameters induced by low pH_o and TASK‐1 plus BK channel blockers.

Top: examples of bursts recorded under the conditions indicated below for n = 10 to 5 cells. The number of events in a burst, as well as the first and last peak amplitude, were derived directly from the data analysis software. The mean plateau amplitude was estimated by calculating the half‐amplitude of the after‐hyperpolarization that was assumed to increase linearly from the first to the last spike. Burst duration was calculated from the initial fast rising of the first AP to the end of the fast repolarization of the last AP, just before the onset of the slower repolarization phase. One‐way ANOVA followed by Bonferroni post hoc tests were made by comparing the values at pH_o 6.6 (last column) with the values in each other condition (^** P < 0.01). P < 0.01 was considered statistically significant.

Cav1 channels contribute to burst firing at low pH_o: nifedipine attenuates and blocks, whereas BayK8644 favours the bursts

The plateau phase of AP bursts in MCCs has been shown to be the result of a balance between Ca²⁺ entry through Cav channels and K⁺ exit through Kv, BK and SK channels (Vandael et al. 2015 b). We also showed that Cav currents during bursts are of lower amplitude compared to those passing during an AP but persist for the entire duration of the burst, thus sustaining enhanced neurotransmitter release during bursts (Vandael et al. 2015 b). An unresolved issue of burst firing in MCCs is how critical are Cav1 channels in burst production given that they sustain spontaneous firing in MCCs. This is particularly true for the Cav1.3 isoform, which activates at more negative potentials than any other high‐threshold Ca²⁺ channel and inactivates very slowly in MCCs (Marcantoni et al. 2010; Vandael et al. 2010). With the purpose of highlighting the role of Cav1 channels in burst generation, we tested the effects of increasing concentrations of nifedipine (0.1, 0.3, 1 and 3 μm) on burst firing. We found that even 100 nM nifedipine, which blocks ∼50% of Cav1 currents in MCCs at −40 mV resting potential (Mahapatra et al. 2011), was sufficient to convert the bursts into irregular tonic firing. Increasing doses of nifedipine (300 nm) accelerated the conversion and lowered the tonic firing frequency. Full blockade of the activity occurred at 3 μm. Figure 10 recapitulates these observations. The current clamped MCC displays a mild burst firing at control (Fig. 10, top left), which was converted in an irregular burst firing after applying 300 nm A1899 + 300 nm paxilline, following a weak cell depolarization (from −43 to −40 mV). The small depolarization observed in this cell, most probably derives from a weak expression of TASK‐1/TASK‐3 channels that may drive the resting cell near to burst firing. Regardless of this, the two blockers (mainly paxilline) induced sustained irregular bursts lasting 300–1000 ms followed by profound hyperpolarizations sustained by a robust activation of outward SK currents (Fig. 10, arrows in top middle, left). The addition of 300 nm nifedipine first converted the burst into an irregular tonic firing (Fig. 10, top middle, right). After a mild depolarization, the firing stopped. Burst firing block persisted in the presence of 3 μm nifedipine (Fig. 10, top right). The addition of 3 μm nifedipine always (n = 5) blocked the firing regardless of any small depolarization or hyperpolarization originating from the different functional coupling and expression density of L‐type and SK channels (Vandael et al. 2012).

Figure 10. Increasing doses of nifedipine revert burst firing to tonic firing and then block.

Bottom: representative current clamp trace of a spontaneously firing MCCs before and during bath application of 300 nm A1899 + 300 nm paxilline (BK channel blocker). The cell displays an initial mild burst firing under control conditions (pH_o 7.4). The addition of saturating doses of the two blockers (300 nm A1899 + 300 nm paxilline) converts the firing into sustained long lasting bursts of 0.3 to 1 s followed by profound hyperpolarizations (arrows, top middle to the left). The addition of 300 nm nifedipine stops the bursts and the subsequent addition of 3 μm nifedipine causes a slight depolarization and blocks the firing. The black dotted line indicates the 0 mV level. White dashed lines indicate V _rest at control and with A1899 + paxilline. Top: time expanded recordings, corresponding to the grey windows of different duration indicated below. The V _m of current clamp recordings was not corrected for the LJPs (see Methods).

Nifedipine (3 μm) was also very effective at blocking burst firing induced by low pH_o (7.0 to 6.6) regardless of whether the bursts were continuously generated near rest (–40 to −50 mV) as in Fig. 10 or after brief depolarizations in cells maintained silent with steady hyperpolarizations (V _h –70 mV). Figure 11 shows an example of an MCC maintained at –70 mV that undergoes normal tonic firing at pH_o 7.4 during brief current injection of 100 ms (black trace). Tonic firing stops on cell repolarization (Fig. 10, top right inset), whereas the cell undergoes burst firing at pH_o 6.6 that persists after the brief depolarization (Fig. 10, red trace). The addition of 3 μm nifedipine stops the bursts and the cell repolarizes back to V _h (blue trace). We observed this in 12 cells, regardless of whether we were lowering pH_o (n = 7) or applying mixtures of the two blockers (n = 5) (not shown).

Figure 11. Nifedipine blocks burst firing evoked by brief step depolarizations.

Three overlapped AP recordings induced by brief (100 ms) step depolarization of 20 pA from V _h = −70 under control conditions (pH_o 7.4; black trace), during pH_o 6.6 application (red trace) and during the addition of 3 μm nifedipine to the pH_o 6.6 solution (blue trace). The dotted line indicates the 0 mV level. Inset: time expanded recording corresponding to the indicated grey window to the left. Nifedipine is effective at blocking the burst firing that persists after the cell is hyperpolarized to −70 mV at the same time as preserving the tonic firing during the brief depolarization. The V _m of current clamp recordings was not corrected for the LJPs (see Methods).

Given the involvement of Cav1 channels in burst firing generation, we tested whether increasing Cav1 channel currents with BayK8644 was sufficient to induce burst firing in MCCs. We previously reported that the addition of 1 μm BayK8644 caused resting cell hyperpolarization and increased firing frequency (Marcantoni et al. 2010), although we did not specifically investigate whether potentiation of Cav1 currents could generate tonic or burst firing in MCCs. We thus tested whether increasing doses of BayK8644 (0.1, 0.3 and 1 μm) favour burst firing in MCCs at pH_o 7.4. We found that, in the majority of MCCs (7 out of 11) with resting irregular tonic activity, BayK8644 induced burst firing even at concentrations as low as 0.1 μm, whereas, in the remaining four cells, the DHP agonist induced only a marked frequency increase with no bursts. Figure 12 shows two examples of MCCs responding differently to BayK8644. In Fig. 10 A, BayK8644 induces burst firing whose duration progressively increases with an increasing concentration. In Fig. 10 B, BayK8644 progressively increases the AP firing frequency without inducing burst firings. Interestingly, in five MCCs that exhibited mild spontaneous burst firing at rest, BayK8644 converted the firing into sustained bursts, causing depolarization blocks of several seconds in some cases (not shown). In conclusion, Cav1 channels appear to be critical not only in regulating pacemaking in CCs (Marcantoni et al. 2010; Vandael et al. 2010; Vandael et al. 2012), but also in contributing to burst firing.

Figure 12. Effects of BayK8644 on spontaneous AP firing.

A, representative current clamp trace of a spontaneously firing MCCs that switches from tonic to burst firing with increasing BayK8644 concentrations (0.1, 0.3 and 1 μm) applied sequentially. The time‐expanded recordings (top grey windows) illustrate the increase in burst duration with increasing concentrations of BayK8644. B, representative current clamp trace of a spontaneously firing MCC in which BayK8644 did not enhance burst firing. In this example, the DHP agonist increases only the rate of AP firing. Dotted lines indicate the 0 mV level. The V _m of current clamp recordings was not corrected for the LJPs (see Methods).

Low pH_o‐induced burst firing potently increases CA secretion in MCCs

A straightforward expectation of the resting depolarization and spontaneous burst firing induced by a low pH_o is a net increase of Ca²⁺ entry through Cav channels that probably drives a marked increase of CA release from MCCs. We have previously shown this to occur when burst firing is induced by application of TTX that reduces Na⁺ currents through Nav1.3/Nav1.7 channels in MCCs (Vandael et al. 2015 b). Under these conditions, we have demonstrated that the oxidative charges accumulated during CA release increase more or less with the same proportion of Ca²⁺ charges increase through Cav channels (∼ 3.5‐fold) (Vandael et al. 2015 b). We thus expected that a low pH_o exerts a similar boosting action on exocytosis and tested whether sustained burst firings induced by a low pH_o or mixtures of TASK and BK channel blockers induce marked increases of CA release. Accordingly, we combined current clamp recordings with CFE amperometry to reveal fast quantal release of CAs and determined how different firing patterns of 60 s at control (pH_o 7.4), low pH_o (6.6) or with application of 300 nm A1899 + 300 nm paxilline affect CA secretion (Fig. 13).

Figure 13. Burst firing induced by low pH_o and mixtures of A1899 + paxilline boosts MCC exocytosis.

A, example of simultaneous recordings of APs (bottom trace) and amperometric events (top trace) by carbon fibre amperometry at pH_o 7.4. B, the same recording conditions as in (A) with a perfusing solution of pH_o 6.6. C, the same recording conditions as in (A) and (B) in the presence of A1899 (300 nm) and paxilline (300 nm). To the right are shown the time expanded recordings of amperometric spikes and APs are indicated to the left. D, comparison of amperometric spikes parameters (see top representation), frequency and cubic root of cumulative charge (Q ^1/3) between pH_o 7.4 (black bars), pH_o 6.6 (blue bars) and A1899 + paxilline (red bars) (^* P < 0.05, ^** P < 0.01 and ^*** P < 0.001, n = 8; one‐way ANOVA followed by Bonferroni post hoc tests). E, overlap of cumulative secretion plots derived from amperometric measurements shown in (A), (B) and (C) for control (pH_o 7.4; black trace), pH_o 6.6 (blue trace) and A1899 + paxilline (red trace). The V _m of current clamp recordings was not corrected for the LJPs (see Methods).

Given the critical conditions of simultaneously recording AP firing (with a patch‐pipette) and amperometric signals (with a CFE; see Methods), experiments at pH_o 6.6 and with TASK and BK channel blockers were carried out on different cells. MCCs were bathed in control solution (pH_o 7.4) and then perfused with the test solution (pH_o 6.6 or blockers), having taken the precaution of changing the bath without moving the CFE. Figure 13 A–C shows three examples of recordings in MCCs maintained at control (pH_o 7.4, black traces), pH_o 6.6 (blue traces) or in the presence of 300 nm A1899 + 300 nm paxilline (red traces). At pH_o 7.4, the spontaneous tonic firing of MCCs in 2 mm extracellular Ca²⁺ induces the basal release of CA in forms of amperometric spikes of very low frequency. A similar basal release in 2 mm Ca²⁺ has been observed, both in isolated bovine chromaffin cell (BCCs) and MCCs of adrenal gland slices (Picollo et al. 2016). In 10 MCCs, all spontaneously firing, three cells had no amperometric spike activity, whereas the remaining had rare spike events, casually distributed during a recording period of 60 s. Mean spike frequency was 0.02 ± 0.01 Hz (Fig. 13 D). The frequency of spike events increased drastically at pH_o 6.6 (0.11 ± 0.3 Hz; P < 0.01, n = 8) and in the presence of TASK and BK channel blockers (0.18 ± 0.3 Hz; P < 0.001, n = 8).

Despite the marked increase in the rate of release, low pH_o and mixtures of the two blockers had no significant effects on the waveform of amperometric spikes (Fig. 13 D). The parameters associated with the peak amplitude (I _max), time to peak (t _p), half‐width (t _1/2), total quantity of charge released (Q) and cubic root of Q (Q ^1/3) (as an estimate of vesicle size) remained unchanged with respect to control (P > 0.05). The significantly larger increase in I _max observed only with the two blockers (P < 0.05) (Fig. 13 D, middle left) may derive from different mechanisms, which could be the result of: (i) an increased probability of double fusion of secretory events favoured by the 8‐fold increased rate of vesicle release; (ii) an increased probability of fused vesicles of larger size that coexist with a second population of smaller size vesicles in MCCs (Grabner et al. 2005; Marcantoni et al. 2009); and (iii) a not well‐identified interaction of protons with chromogranin A, a major protein in the vesicle that regulates the fraction of CA bound to the matrix (Jankowski et al. 1993). This latter effect, however, occurs only at very acidic pH_o (5.5). Interestingly, because of the increased rate of release, the time course of cumulative secretion (Fig. 13 E, black trace) exhibited a steeper rise at low pH_o (Fig. 13 E, blue trace) that increased further with A1899 + paxilline (Fig. 13 E, red trace). The mean quantity of cumulative charges recorded during 60 s recordings increased by 7.4‐ and 11.6‐fold with respect to control, respectively (Fig. 13 D).

Given the opportunity of simultaneously recording AP bursts and secretory events, we tested specifically for a direct correlation between burst firing and amperometric spikes and found no specific links between the two signals. For example, we were unable to detect any frequency correlation between AP firing and secretory events. In the case of bursts, the AP frequency was uniform during 60 s of recordings (Fig. 13 C, red burst firing trace), whereas, in the case of spike events, the frequency was rather irregular, with periods of high activity alternating with long periods of no activity (Fig. 13 C, red spike events trace). As reported previously, this suggests a weak correlation between AP and amperometric events under both tonic and burst firing (Zhou & Misler, 1995).

Discussion

In the present study, we have provided evidence to suggest that lowering the pH_o causes a marked resting membrane depolarization, a switch of spontaneous firing from tonic to burst and a 7.4‐fold increase of cumulative CA release in MCCs. The consequence of lowering pH_o is thus a non‐neurogenic large increase of adrenaline and noradrenaline that is mainly driven by the increased Ca²⁺ entry during the plateau potential of bursts. From a functional point of view, we have shown that, by directly sensing a decrease of pH_o, MCCs act as pH sensors and secrete large amounts of CAs. This is the typical body response with respect to recovering from blood acidosis and hyperkalaemia‐induced muscle fatigue (Clausen, 1983) during heavy exercise (Medbo & Sejersted, 1990).

Specifically, we have shown that the marked resting membrane depolarization and burst firing induced by lowering pH_o from 7.4 to 6.6 is not exclusively associated with the blockade of pH‐sensitive TASK‐1 and TASK‐3 K⁺ channels but involves a pH‐induced blockade of Ca²⁺‐ and V‐dependent K⁺ conductances (BK and Kv) that contributes further to the MCC depolarization at low pH_o. A blocking effect of low pH_o on the M‐current (I _M) and TRPM4 channels that are expressed in CCs of some species and potentially contribute to the resting potential (Wallace et al. 2002; Mathar et al. 2010) cannot also be excluded. In a preliminary series of experiments, we found that blockade of Kv7 channels by increasing doses of the selective blocker XE991 (0.3, 1 and 3 μm) (Wang et al. 1998) caused partial hyperpolarization followed by cell depolarization and an increased firing frequency. This suggests that MCCs express Kv7 (I _M) channels that could contribute to cell depolarization when blocked by protons. The possibility that K_V7 channels play a role in spontaneous firing of MCCs is of great interest and further experiments are currently in progress.

Our findings clearly show that blockade of TASK‐1 and TASK‐3 is not the only mechanism supporting membrane depolarization and burst firing in MCCs. We have previously shown that burst firing in MCCs derives from the fine equilibrium between inward and outward currents flowing during AP repolarization (Vandael et al. 2015 b). Because inward currents are mainly carried by Nav1.3/Nav1.7 and Cav1/Cav2 channels and outward currents by BK, SK and Kv channels in MCCs, any significant increase of the former and decrease of the latter may potentially induce burst firing and increased Ca²⁺ entry. In the case of low pH_o, we found that a marked blockade of BK channels accompanied by a small attenuation of Cav channels is probably the cause of driving spontaneous tonic AP firing into regular burst firing. We have also shown that, by blocking BK channels with paxilline in a cell already depolarized with saturating doses of A1899, we could induce regular burst firing, whereas, by adding increasing doses of the Cav1 blocker nifedipine 0.1 to 1 μm), we could revert the firing to a tonic mode. A critical involvement of Cav1 channels in burst firing is also supported by the potentiating effects of BayK8644, which, in a large fraction of MCCs, induces burst firing of increased duration in a dose‐dependent manner.

Block of TASK channels does not account for the full effect of low pH_o on MCC excitability

Our findings show clearly that widely used selective blockers of the two‐pore TASK‐1 and TASK‐3 background channels (A1899 and PK‐THPP) (Cotten, 2013; Bayliss et al. 2015; Chokshi et al. 2015; Dadi et al. 2015) cause a net depolarization of resting MCCs. The depolarization is accompanied by an increased rate of AP firing and a moderate burst firing activity (Fig. 3). Comparing these effects with the selectivity of the two blockers, we can safely conclude that MCCs express more functional TASK‐1 than TASK‐3 ‘leak’ channels and we cannot exclude the possibility that the attenuated effects of PK‐THPP on resting potential could derive from a partial blockade of the TASK‐1 isoform. That TASK‐1 channels are the most probable two‐pore leak channels expressed in MCCs is also suggested by the IC₅₀ of the pH‐induced resting depolarization (IC₅₀ 7.2) (Fig. 1 C), which is very close to the pK _a of proton blockade of homodimeric TASK‐1 channels (pK _a 7.4) and quite different from the pKa of homodimeric TASK‐3 channels (pK _a 6.7) reported by Bayliss et al. 2015. MCCs probably also express heterodimeric TASK‐1:TASK‐3 channels. This is suggested by the increased depolarization induced by the simultaneous application of A1899 and PK‐THPP (Fig. 3 D), as well as by the pK _a of proton blockade of heterodimeric TASK‐1:TASK‐3 channels (pKa 7.2) that is similar to the IC₅₀ of pH‐induced MCC depolarization.

It is interesting to note that TASK‐3 are weakly or not expressed in rat CCs (RCCs) (Inoue et al. 2008) and BCCs (Enyeart et al. 2004), whereas TASK‐1 channels are highly expressed and mediate the muscarine‐induced resting depolarization that induces the increased CA secretion in RCCs (Inoue et al. 2008). Concerning the size of the mean depolarization (∼5 mV) induced by A1899 (Fig. 3 C), this is probably an underestimation of the true depolarizing effects of TASK‐1/TASK‐3 channel blockade. This is a result of the presence of highly expressed Ca²⁺‐dependent SK and BK channels (Vandael et al. 2015 a) that could partially attenuate the depolarizing effects of TASK‐1/TASK‐3 channel blockade. An increased Ca²⁺ influx during cell depolarization, which may occur at rest through open Cav1.3 channels, would induce partial hyperpolarization through the activation of SK and BK channels. Both channels contribute to set the resting potential of MCCs. It is worth noting that selective blockade of SK channels by apamin (Vandael et al. 2012) and BK channels by paxilline (Marcantoni et al. 2010) induces net depolarizations of 2–3 mV at rest, in each case.

Given the existence of several K⁺ channels regulating the resting membrane potential in MCCs (SK, BK, K_V) (Vandael et al. 2015 a), it is evident that the main role of pH‐sensitive TASK channels at low pH_o is to trigger a sufficient cell depolarization increasing the rate of AP firing in spontaneously active MCCs. Further depolarizations and corresponding changes in regular burst firing are caused by the pH_o‐induced inhibition of BK and partially of Kv rather than SK channels. These latter changes are almost unaffected by low pH_o (Fig. 7) and are thus expected to counteract rather than support the pH_o‐induced depolarization, mostly because of the increased Ca²⁺‐entry induced by the blockade of TASK channels. SK channels are fundamental in MCCs to terminate the bursts and allow the cell to recover Cav1 and TTX‐sensitive Nav1.3/1.7 channels and initiate the following burst (Vandael et al. 2015 b).

Our data indicate that, besides TASK‐1 blockade, a second major action of low pH_o is on BK channels, whose conductance is effectively inhibited at pH_o between 7.0 and 6.6 (Peers & Green, 1991) and by similar changes of pH_i in various tissues (Cook et al. 1984; Kume et al. 1990). In the case of MCCs, a smaller contribution of BK currents to the repolarization phase of AP generation is critical for the generation of the plateau potential on top of which AP burst develops. To support this idea, we showed that the effects of pH_o 6.6 on firing modes can be mimicked by simply adding nearly saturating doses of paxilline to the TASK‐1 blocker A1899. Indeed, the quantity of paxilline used (300 nm) is apparently higher than the quantity necessary to block 50–60% of BK channels, as occurs at pH_o 6.6. However, paxilline block is state‐dependent because it binds with high affinity to the closed state and with low affinity to the open state of the channel (Zhou & Lingle, 2014). Paxilline blockade is thus highly dependent on membrane potential, being strong at very negative potentials when channels are fully closed (IC₅₀ 10 nM) and very weak at very positive potentials when channels are fully open (IC₅₀ 10 μm). Given this, we expect that, in spontaneously firing cells in which the membrane potential fluctuates between −50 and +30 mV for most of the time, 300 nm paxilline would block ∼50–60% of the BK channels. The higher amount of paxilline used would also account for the partial blockade of Cav channels that partially reduces the amount of Ca²⁺ entry and thus the number of activated BK channels (Prakriya & Lingle, 1999). Regardless of these considerations, there is still good correspondence between the effects of low pH_o and the two channel blockers. This combination turns out to be essential for separating the effects of low pH_o on cell excitability (mimicked by the two blockers) from the action of protons on the molecular and subcellular components regulating the Ca²⁺‐driven CA release (see below).

The role of BK channels in burst firing

Our data clearly show that, in addition to the blocking effects on TASK‐1 and TASK‐3 channels, low pH_o also has marked effects on the BK channel conductance in the range between pH_o 7.0 and 6.6. Taking pH_o 7.0 as a reference pH_o, in the present study, we demonstrate that the other voltage‐ and Ca²⁺‐gated channels contributing to AP generation in MCCs (Nav, Kv, Cav and SK) are only slightly or mildly affected at pH_o 7.0 compared to BK channels. The size and kinetics of Na⁺ currents carried by TTX‐sensitive Nav channels in MCCs (Vandael et al. 2015 b) are only slightly affected at pH_o 7.0. This is in line with the common notion that proton blockade of the negatively charged glutamic and aspartic acid residues in the outer pore selectivity filter of Na⁺ channels requires a pH_o lower than 6.0 to significantly reduce channel permeability (Catterall, 2000). The same is true for the Ca²⁺ currents carried by Cav1.2, Cav1.3, Cav2.1, Cav2.2 and Cav2.3 in MCCs (Marcantoni et al. 2010; Mahapatra et al. 2011) that are mildly attenuated at pH_o 7.0. The selectivity filter of Cav1 and Cav2 channels is formed by a ring of four negatively charged glutamic acid residues (Heinemann et al. 1992), whose partial protonation requires a more acidic pH_o to effectively reduce Ca²⁺ permeability. Also, the protonation of membrane surface charges responsible for the right shift of g_Ca(V) requires a more acidic pH to produce large effects (Zhou & Jones, 1996). The voltage‐dependent activation of Cav channels is only affected slightly (Fig. 5 A) and the same is true for the time‐course of the Ca²⁺‐dependent inactivation of L‐type channels, which is not significantly altered at pH_o 7.0 during depolarizations that mimic the mean burst duration (300 ms).

Concerning the effects of pH_o 7.0 on SK channels, we found also no evidence for a blockade of SK currents, which is consistent with the notion that SK channels are voltage independent and their open probability is steeply dependent on [Ca²⁺]_i, with an EC₅₀ of ∼0.5 μm (Fakler & Adelman, 2008). Such high affinity for Ca²⁺ is a result of the presence of calmodulin Ca²⁺‐binding sites, which regulate the Ca²⁺‐calmodulin‐dependent conformational changes driving SK channel open probability (Keen et al. 1999). To compete effectively with Ca²⁺ for the occupancy of these sites, the intracellular proton concentrations must be at least 10‐ to 100‐fold higher than the EC₅₀ for Ca²⁺. Given this, even under a sizable proton permeation through the plasmalemma, the pH_i should have to drop below 6.0 to have any sizeable effect on SK open probability.

At variance with the other channels, low pH_o has a marked inhibitory effect on BK channels conductance. The inhibition is characterized by a marked current depression at almost all potentials when BK channels are either activated by a fixed (Fig. 7 B) or variable Ca²⁺ loading (Fig. 6 C). Inhibition of BK channels by protons occurs through a reduction of g_BK(V) at all potentials and a marked shift of the curve toward negative potentials with no significant changes of the V‐dependence (Fig. 7 B). This is in agreement with the reported effects of low pH_o on the BK channels of type‐I carotid body cells (Peers & Green, 1991) and low pH_i on the BK channels of rabbit trachea smooth muscle (Kume et al. 1990). Our data are also in line with the widely accepted notion that low pH_o (or low pH_i) acts on BK channels by mainly decreasing their open probability in the pH range between 7.8 and 6.4 in variety of cells (Cook et al. 1984; Christensen & Zeuthen, 1987; Schubert et al. 2001). On this issue, there is general agreement that hydrogen ions decrease the open probability of BK channels by shortening their open state, although how this occurs is still a matter of discussion (Hou et al. 2009). We have not specifically tested whether changes of pH_o could induce comparable changes to pH_i, although two key considerations suggest that this is probably the case. A lowering of pH_o in rat CCs of the adrenal gland causes a rapid fall of intracellular pH_i (Fujiwara et al. 1994). In our experiments on BK channels, the onset the effects of low pH_o usually required 20–40 s to reach steady‐state, whereas washing out was complete in no less than 3 min (see Methods). Blockade of BK channel permeability by protons also occurs, although this is at a significantly more acidic pH (pK _a 5.1) (Brelidze & Magleby, 2004).

Cav1 channels contribute to the low pH_o‐induced burst firing

Cav1.3 channels possess all the gating properties to regulate the resting potential and sustain burst firing in CCs. They activate at relative low membrane voltages in 2 mm Ca²⁺ and deactivate slowly (Koschak et al. 2001; Marcantoni et al. 2010), are effectively coupled to BK channels (Prakriya & Lingle, 1999; Vandael et al. 2010), and probably contribute significantly to the inward Ca²⁺ current that critically sustains burst durations of 200–500 ms at depolarized potentials (−30 to  −20 mV) (Vandael et al. 2015 b). In the present study, we have shown that low doses of nifedipine (100–300 nm), which produce only partial blockade of L‐type channels (40–60% at resting potentials) (Mahapatra et al. 2011), are sufficient to revert the bursts into tonic firing. Nifedipine is also effective at preventing bursts that are generated by step depolarization from a very negative V _h (−70 mV). Under these conditions, nifedipine blocks the burst that persists when the cell is repolarized but not the tonic firing during the depolarization that is sustained by the available Nav1.3/Nav1.7 and Cav2 channels recruited at negative V _h. Finally, a role for Cav1 channels in burst firing is also suggested by the potentiating effects of BayK8644 that are very effective at inducing burst firing in a fraction of MCCs probably possessing higher densities of Cav1 channels or lower densities of BK and SK channels. Both conditions are at the basis of increased MCC excitability either in form of higher tonic firing frequencies (Vandael et al. 2012) or increased burst firing and depolarization blockade (Vandael et al. 2010).

Tonic‐to‐burst firing is the main ‘motor’ of the increased CA secretion at low pH_o

An important finding of the present study is that low pH_o causes a marked increase of CA release during resting conditions in MCCs (Fig. 13 E). This is in good agreement with previous observations in RCCs (Inoue et al. 2008), where a lowering of pH_o is reported to induce an increased rate of AP firing and a consequent increase of released CAs. Our data are also in good agreement with reports on MCCs (Vandael et al. 2015 b) and RCCs (Zhou & Misler, 1995; Duan et al. 2003) in which burst firing is shown more effective at potentiating CA release than increasing the rate of AP firing. In the case of MCCs, burst firing was induced by blocking Nav1.3/Nav1.7 channels with TTX, and produced a 3.7‐fold increase of cumulative charges associated with released CAs, which is in good correlation with the 3.5‐fold increase of Ca²⁺ entering the cells during bursts. In the case of pH_o 6.6, the increase of cumulative charges associated with CA secretion is even more marked (7.4‐fold). This larger value probably derives from the different protocols used to generate APs rather than the specific effects of pH_o on secretion. In the present study, we maintained the cells at their physiological resting potential (−45 to –50 mV) when measuring amperometric signals. Previously, we first hyperpolarized the cells to  −70 mV to recruit most available Ca²⁺ channels and then depolarized the cell to −50 mV with square current pulses (Vandael et al. 2015 b). Thus, with the present protocol, there are less available Ca²⁺ channels at rest and the basal secretion under control conditions is significantly smaller with respect to the previous protocol. A second reason for the larger increase of cumulative charges is that pH_o 6.6 and the TASK and BK channel blockers induce more regular burst firing during the time of amperometric recordings. Bursts at pH_o 6.6 are not interposed with APs of high frequency as in the case of Nav1.3/Nav1.7 block (Vandael et al. 2015 b).

Our data also help to clarify how much of the marked enhancement induced by low pH_o is associated with the switch from tonic to burst and how much is the result of an interaction of protons with the secretory apparatus (vesicle loading, kinetics of vesicle fusion and CA release). Our findings show that the enhanced secretion by low pH_o is attributable mainly to a switch from tonic to burst firing rather than to an effect of pH_o on the secretory apparatus itself. Application of A1899 and paxilline that mainly act on ionic conductances produces even larger increases of cumulative charges associated with released CAs (11.6‐fold). In addition to this, the effects of low pH_o compare well with those of the two blockers. Low pH_o and blockers significantly increase the rate of amperometric events, whereas they have almost no effect on the parameters associated with the shape of amperometric spikes, with the exception of I _max with the two blockers that we have previously emphasized. Thus, the common cause of enhanced secretion is the marked increase of Ca²⁺ entering the cells during the sustained bursts during either a low pH_o or A1899 + paxilline application.

A final interesting finding of the present study is that, even under sustained burst firing induced by low pH_o or TASK and BK channel blockade, there is no clear evidence of synchronism between AP firing and amperometric spikes. In other words, most Cav channels are not co‐localized with secretory vesicles in CCs and Ca²⁺ has to diffuse inside the cell to trigger most of the exocytotic events (Klingauf & Neher, 1997).

The physiological role of sensing low pH_o in the regulation of blood acidosis and hyperkalaemia

Adrenal CCs respond to acidosis and hyperkalaemia with increased firing activity and the release of adrenaline (Kao et al. 1991; Fujiwara et al. 1994; Inoue et al. 2008; Mahapatra et al. 2011). This occurs typically during heavy exercise and, together with the CAs released from sympathetic nerve terminals, is at the basis of the key body response to recovering from muscle fatigue (Medbo & Sejersted, 1990). Besides increasing heart beat rates and adapting all other functions involved in the ‘fight‐or‐flight’ body response, the elevation of circulating CAs increases the levels of cAMP in fatigued skeletal muscles via β₂‐adrenergic receptor stimulation (Clausen et al. 1993). Elevated cAMP levels increase the phosphorylation of Na⁺/K⁺ ATPase pumps with a consequent decrease of the K⁺ concentration outside the sarcolemma. The body response to acidosis and hyperkalaemia is fast, with restoration of the plasma [K]_o to its physiological value (3.5 mm) within a few minutes after the interruption of intense muscle exercise, whereas pH_o remains low (7.2 to 6.9) (Medbo & Sejersted, 1990). pH_o returns to its normal plasma level (7.4) after ∼1 h from the end of the exercise.

In the present study, we have shown that, at the basis of the body response to acidosis, there is probably an effective pH‐sensing mechanism involving a marked increase of cell excitability and CA release from the CCs of the adrenal medulla. CCs undergo pronounced depolarization and switch their tonic firing into bursts, which boosts Ca²⁺ entry and Ca²⁺‐driven CA release. A key point of the present study is that this non‐neurogenic response is probably mainly controlled by changes of ion conductances and cell excitability rather than by specific effects of pH_o on the secretory machinery (i.e. altered vesicle loading) (Borges et al. 2010), formation of the fusion pore and secretion of vesicle content (Lindau & Alvarez de Toledo, 2003) or alteration of bound CA in the vesicle matrix (Jankowski et al. 1993). The reported data also represent a good example of how typical neuron‐like burst firings can be exploited by CCs to effectively control key physiological functions (Lingle, 2015).

Additional information

Competing interests

The authors declare that they have no competing interests.

Author contributions

LG and DHFV contributed to the data collection for the whole‐cell and amperometric experiments. VC contributed to the design, analysis and interpretation of the amperometric measurements. LG and EC contributed to the conception and design of the experiments, data analysis and the drafting of the article, as well as its critical revision for important intellectual content involving the input of all co‐authors. All authors have approved the final version of the manuscript submitted for publication.

Funding

This work was supported by the Italian MIUR (PRIN 2010/2011 project 2010JFYFY2); the University of Torino and Telethon Foundation (grant # GGP15110).