Mitochondrial Ca²⁺ Uptake Regulates the Excitability of Myenteric Neurons

Abstract

We investigated the role of mitochondria in the regulation of intracellular Ca²⁺([Ca²⁺]_i) and excitability of myenteric neurons in guinea pig ileum, using microelectrodes and fura-2 [Ca²⁺]_i measurements. In AH/Type-II neurons, action potentials evoke ryanodine-sensitive increases in [Ca²⁺]_i that activate Ca²⁺-dependent K⁺ channels and slow afterhyperpolarizations (AH) lasting ∼15 sec. Exposure to the protonophore carbonyl cyanidep-(trifluoromethoxy)phenylhydrazone (FCCP; 1 μm) had no significant effect on the membrane potential or resting [Ca²⁺]_i. However, action potentials elicited in the presence of FCCP triggered a sustained (>5 min) increase in [Ca²⁺]_i and a compound hyperpolarization (13.4 ± 1.5 mV). The respiratory chain blockers antimycin A and rotenone (10 μm) had similar effects that developed more slowly. Depletion of the intracellular Ca²⁺ stores with thapsigargin (2 μm) or ryanodine (10 μm) greatly attenuated the hyperpolarization caused by FCCP. S/Type-I neurons that do not have AH were hyperpolarized by mitochondrial inhibition independently of action potentials. Blockade of the F₀F₁ ATPase by oligomycin (10 μm) had variable effects on myenteric neurons. The majority of AH/Type-II neurons were hyperpolarized by oligomycin, most likely by activating ATP-dependent K⁺ channels. This hyperpolarization was not triggered by action potential firing and not accompanied by an increase in [Ca²⁺]_i. MitoTracker staining revealed a dense mitochondrial network particularly in myenteric AH/Type-II neurons, supporting the importance of mitochondrial Ca²⁺ buffering in this subset of neurons. The data indicate that mitochondrial uptake of Ca²⁺released from the endoplasmic reticulum sets [Ca²⁺]_i and the activity of Ca²⁺-dependent conductances, thus regulating the excitability of myenteric neurons.

Mitochondria produce ATP via ATPases that run in reverse, driven by the proton electrochemical gradient across the mitochondrial membrane. This gradient is built up by electron transport during oxidative phosphorylation. Besides their role in supplying energy, mitochondria are also important in apoptosis, aging, and Ca²⁺ homeostasis. Although the latter has long been debated, recent studies have put mitochondria back on the Ca²⁺ scene as crucial players (Duchen, 2000; Friel, 2000; Rizzuto et al., 2000). Friel and Tsien (1994) described a Ca²⁺ store in bullfrog sympathetic neurons that was sensitive to the protonophore FCCP [carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone], and mitochondria also have been shown to alter depolarization-induced [Ca²⁺]_i responses and the gating properties of Ca²⁺-activated currents in a variety of neurons, indicating substantial Ca²⁺removal by these organelles (Thayer and Miller, 1990; Werth and Thayer, 1994; Kenyon and Goff, 1998; Nowicky and Duchen, 1998; Pivovarova et al., 1999; Friel, 2000; Vanden Berghe et al., 2002; Wang and Thayer, 2002). Quantitative measurements have revealed a double role for mitochondria: they sequester Ca²⁺ during depolarization to release it again during repolarization (Babcock and Hille, 1998; Colegrove et al., 2000). Several systems, including Ca²⁺ uniporters, Ca²⁺/H⁺exchangers, Na⁺-dependent and independent release mechanisms, and permeability transition pores, contribute to mitochondrial Ca²⁺ handling (Gunter et al., 2000). Further, mitochondria often are positioned closely to the endoplasmic reticulum (ER), and optimal ER function requires competent mitochondria (Landolfi et al., 1998).

Myenteric neurons are comprised in a ganglionated network in the intestinal wall. They include intrinsic sensory neurons, interneurons, and motor neurons that exert reflex activity and coordinate the propulsion of the intestinal contents (Smith et al., 1992). Electrophysiologically, these neurons have been classified in two or more groups (Nishi and North, 1973; Hirst et al., 1974; Wood, 1994;Smith et al., 1999). S/Type-I neurons have either phasic or tonic firing characteristics and their action potentials are TTX-sensitive. The AH/Type-II neurons are characterized by prolonged afterhyperpolarizations (AH) that follow action potential firing. Their action potentials have, besides the TTX-sensitive part, also a Ca²⁺ component (Morita et al., 1982; Hirst et al., 1985) that triggers Ca²⁺ release from the ER via the activation of ryanodine receptors (Hillsley et al., 2000; Vogalis et al., 2001). The increased intracellular Ca²⁺ concentration opens Ca²⁺-activated K⁺ channels responsible for the transient slow AH, a mechanism also observed in other neurons (Kawai and Watanabe, 1989; Sah and McLachlan, 1991; Yoshizaki et al., 1995; Moore et al., 1998; Shah and Haylett, 2000; Vogalis et al., 2002). The AH/Type-II neurons have been identified as intrinsic primary afferent neurons (Furness et al., 1998) and are therefore important in the organization of gastrointestinal motility. By inhibiting the throughput of sensory information during the afterhyperpolarization, they determine the frequency at which interneurons and motor neurons are driven. Because the AH is directly dependent on the [Ca²⁺]_i and because there is evidence that mitochondria play an important role in the Ca²⁺ handling in several other neurons, we aimed to investigate the role of mitochondria in the Ca²⁺ homeostasis and therefore the membrane excitability of myenteric neurons.

Parts of this work have been published previously as an abstract (Vanden Berghe and Smith, 2001).

MATERIALS AND METHODS

Tissue preparation. Adult guinea pigs (250–350 gm) were asphyxiated by CO₂, followed by exsanguination (a method approved by the Animal Ethics Committee at the University of Nevada, Reno). A 5 cm segment of ileum was removed, flushed with cold Krebs' solution, and pinned flat in a Sylgard-lined dissection dish. The intestine was opened along the serosal border, and the mucosa, submucosa, and circular muscle layer were removed with fine forceps. During the dissection the preparation was perfused constantly with ice-cold Krebs' solution bubbled with a 3% CO₂/97% O₂ gas mixture. The longitudinal muscle myenteric plexus preparations were stretched and clipped on a stainless steel ring. The tissues were equilibrated in a recording dish with a coverslip bottom for at least 1 hr in warmed Krebs' solution (36°C; bubbled with 3% CO₂/97% O₂). Shortly after the temperature reached steady state, vigorous contractions were observed in the tissue preparations. To make intracellular recordings possible, we added 1–2 μm nicardipine and 1 μm atropine to the bath solution to relax the underlying longitudinal muscle layer (Hillsley et al., 2000).

Membrane potential recordings. Ganglia were identified easily under bright-field light microscopy (Nikon TE 300; Nikon, Tokyo, Japan). The neurons were impaled with glass microelectrodes pulled on a 767 Sutter (Novato, CA) Microelectrode puller, and the membrane potential was recorded. To perform [Ca²⁺]_imeasurements simultaneously, we filled the tips of the electrodes with 1–2 mm bis-fura-2 (K_D = 370 nm) hexapotassium salt dissolved in 1–2 m KCl. The rest of the electrode was filled with 2 m KCl. The resistance of the electrodes varied between 120 and 150 MΩ. The signals were amplified with an Axoclamp 2B amplifier (Axon Instruments, Union City, CA) and monitored continuously on a Tektronix R468 oscilloscope (Tektronix, Beaverton, OR). A bipolar electrode was used to stabilize the tissue against the coverslip bottom and to elicit synaptic potentials in the impaled neurons. Intracellular pulses and extracellular pulse trains were triggered and generated by a Master-8 pulse generator (A.M.P.I., Jerusalem, Israel).

Fura-2 measurements of [Ca²⁺]_i. Fura-2 was loaded iontophoretically into the neurons by applying a train of hyperpolarizing pulses (0.2–0.4 nA at 1–2 Hz) for 2–4 min. Fluorescent signals were recorded with an Ionoptix (Milton, MA) Fluorescence System. A xenon lamp was used as a UV light source. Differential excitation (340/380 nm) of the Ca²⁺ indicator resulted from a spinning chopper wheel (at 60 Hz) to obtain ratiometric values at 30 Hz. Images were recorded with a CCD camera (Ionoptix), and rectangular regions of interest were drawn in the image. The fura-2 ratio was calculated for that specific area. The membrane potential recordings were sampled at the maximal rate (1 kHz) simultaneously with the fura-2 signals by using the analog-to-digital board of the Ionoptix interface box and displayed and stored in the same data file. We chose to express the [Ca²⁺]_i signals as fura-2 ratios rather than calibrated Ca²⁺signals, because the R_min could not be obtained reliably from calibrations by using ionomycin and 0 mm Ca²⁺ solutions. This was in part due to the way the preparation was set up and also attributable to the fact that the underlying muscle layer interfered with exact calibration of the signals.

Immunochemical stainings and fluorescent markers. Myenteric plexus longitudinal muscle preparations were pinned flat in sterile Sylgard-lined dissection dishes. The tissues were incubated (37°C; 5% CO₂) during 1 hr in the presence of 0.1–1 μm ER-Tracker Blue DPX and/or 0.1–1 μmMitoTracker Green. The organotypic medium consisted of HAM/F12 culture medium supplemented with 2% fetal bovine serum, 1 μmnifedipine, and 1% penicillin–gentamycin antibiotic solution. The tissues were washed thoroughly in cold Krebs' solution, fixed in 4% paraformaldehyde (30 min), and washed in a 0.1 mPO₄³⁻ buffered 0.9% NaCl solution, pH 7.2. To identify calbindin-like immunoreactive neurons, we incubated the tissues (48 hr) in a 0.1 mPO₄³⁻ buffered 0.9% NaCl solution, pH 7.2, containing 1:100 mouse monoclonal anti-calbindin D-28K antibody. The secondary antibody was anti-mouse IgG coupled to Texas Red (1:150; 90 min). The tissues were washed and mounted on slides with Aquamount. An Olympus BX50 microscope with specific filter cubes (FITC: EX, BP470/490; DM, 505; EM, BA 535/50; and aminomethylcoumarin: EX, D360/40; DM, BS 400 dichroic longpass; EM, BA 435–485; Chroma, Brattleboro, VT) was used to visualize the MitoTracker and ER tracker, respectively. Confocal imaging was performed with Bio-Rad (Hercules, CA) MRC 600 (60× lens) and Nikon confocal microscopes (40× lens). Images were acquired via excitation wavelengths of 488 nm (for MitoTracker Green) and 568 nm (for Texas Red). Scion Image (Frederick, MD) was used to quantify MitoTracker intensity in composite images of z-series scans of five optical sections through a depth of 2.5 μm, using a confocal aperture <20% of maximum.

Calculations and statistics. Amplitudes and durations of the transient events were calculated as indicated in Figure 1. The time to 90% recovery was calculated as a measurement of duration (90% duration) of the afterhyperpolarization and the Ca²⁺ transient. Resting membrane potentials were measured before and during the application of mitochondrial blockers and compared with a paired Student'st test or ANOVA with a Bonferroni test as a post hoc analysis. To study the effect of mitochondrial blockers on the transient responses, we calculated both the 90% duration and the remaining amplitude of the AH and Ca²⁺transient 15 sec after action potential spiking. The average ± SEM duration and amplitudes of AH and Ca²⁺transients were compared with a paired Student's t test. Ap value of 0.05 was considered the cutoff for statistical significance; n values specify the number of neurons. Curve fittings and statistical analysis were performed in GraphPad Prism (San Diego, CA) or MS Excel (Microsoft, Redmond, WA).

Fig. 1.

Schematic overview of some parameters calculated in this study. A, A slow afterhyperpolarization (sAH) and the matching Ca²⁺ transient. Amplitude and 90% duration were calculated as indicated with the dashed lines. B, Amplitude and duration of the fast afterhyperpolarization (fAH).

Drugs and solutions. All drugs and the antibody against calbindin were obtained from Sigma (St. Louis, MO). Bis-fura-2 hexapotassium salt, MitoTracker Green, and ER-Tracker Blue DPX were obtained from Molecular Probes (Eugene, OR). The anti-mouse IgG coupled to Texas Red was from Vector Labs (Burlingame, CA), and tissue culture supplies (HAM/F12, fetal bovine serum, and antibiotics) were obtained from Invitrogen (Grand Island, NY). The Krebs' solution used throughout this study contained (in mm): 120.3 NaCl, 5.9 KCl, 1.2 MgCl₂, 1.2 NaH₂PO₄, 15.5 NaHCO₃, 2.5 CaCl₂, and 11.5 glucose.

RESULTS

Microelectrode recordings were obtained from 102 neurons (75 animals), 54 of which were identified as AH/Type-II neurons on the basis of the prominent AH they displayed after action potential (AP) firing. Most of the hyperpolarizations consisted of both a fast (fAH) and a slow (sAH) component (Fig.1A,B). The average amplitude of the sAH, elicited by 2 APs, was 9.8 ± 0.6 mV, and the average 90% duration was 15.9 ± 2.9 sec (n = 10). fAH values were smaller (5.1 ± 0.3 mV) and shorter (43.5 ± 1.5 msec). The amplitudes of both fast and slow AH and the duration of the sAH increased linearly (r² > 0.9) with the number of action potentials preceding the AH (fAH, ∼2.5 mV/AP; sAH, ∼2 mV/AP; DUR_sAH, 2.2 sec/AP; n = 6). Single APs usually did not elicit a detectable fAH. The amplitude of the sAH was also linearly dependent (r² > 0.85) on the holding potential (current clamp) and increased ∼1 mV for each 10 mV the neuron was depolarized. In 13 AH neurons the [Ca²⁺]_i was monitored simultaneously. The sAH was always closely matched by a Ca²⁺ transient that started with the AP firing and reached a maximum 385 ± 81 msec (n = 7) after onset of the depolarizing pulse. The sAH itself reached its maximum 1115 ± 257 msec (n = 7) after the onset of AP firing (Fig. 1A). The amplitude and duration of the accompanying Ca²⁺ transient, like for those of the AH, increased for an increasing number of action potentials (Fig. 2A), and the recovery followed an exponential time course (τ = ∼2.4 sec) similar to that described in a previous report (Hillsley et al., 2000).

Fig. 2.

A typical example of the responses as observed in myenteric neurons. Top panels show the changes in [Ca²⁺]_i; bottompanels are the changes in membrane potential (V_m). A, Action potential firing in AH/Type-II neurons is followed by a transient afterhyperpolarization that is larger for an increasing number of action potentials, as shown in the insets. The afterhyperpolarization is matched closely by a [Ca²⁺]_i transient, which decays exponentially. B, In S/Type-I neurons several action potentials are needed to elicit a detectable Ca²⁺transient that is increasing and decaying gradually.

In phasic S/Type-I neurons the [Ca²⁺]_i increased gradually during the repetitive firing of single APs (up to 10) and decayed slowly afterward (Fig. 2B). Single APs never elicited a detectable transient in these neurons. The Ca²⁺ transients in tonically firing S/Type-I neurons increased in amplitude with the number of APs during the depolarizing pulse, whereas the short afterhyperpolarization (3.5 ± 0.3 sec) that sometimes was observed in these neurons (Shuttleworth and Smith, 1999) was independent of the number of APs.

Effect of mitochondrial inhibitors on membrane potential and [Ca²⁺]_i of AH/Type-II neurons

Membrane potential and resting [Ca²⁺]_i

To assess the effect of mitochondrial inhibition on neuronal excitability, we challenged nine AH neurons with the protonophore FCCP (1 μm). The membrane potential was measured in control conditions and in the presence of FCCP before and after stimulation. Exposure to FCCP (∼1.5 min) did not change the resting membrane potential significantly (−57.5 ± 4.2 to −59.2 ± 4.3 mV; ANOVA; Bonferroni post hoc analysis; p > 0.05; n = 9). However, when APs were elicited in the presence of FCCP, the AH triggered by these APs was followed by a significant compound hyperpolarization (−57.5 ± 4.2 to −71.6 ± 4.9 mV; ANOVA; Bonferroni post hoc analysis;p < 0.001; n = 9) (Fig.3A). In the majority (6 of 9) of AH neurons the membrane potential did not recover after the first AH and stayed hyperpolarized. The time to reach the maximum of the compound hyperpolarization was, on average, 146 ± 31 sec (n = 9). In the course of this hyperpolarization depolarizing current pulses no longer elicited action potentials (Fig.3A). In six AH neurons in which the [Ca²⁺]_i was monitored simultaneously, the AP-induced compound hyperpolarization in the presence of FCCP was accompanied by a ∼10% increase in fura-2 ratio (0.94 ± 0.13 to 1.04 ± 0.17; p = 0.04). This [Ca²⁺]_i rise was approximately twice the amplitude of the single AP-induced transients (Fig. 3A). Seven recordings were kept sufficiently long to investigate the reversibility of FCCP. After washout (5–10 min) the membrane potential recovered to control levels (−61.3 ± 6.7 compared with −57.5 ± 4.2 mV; ANOVA; Bonferroni post hoc analysis; p > 0.05; n = 7). Similarly, after washout of FCCP the [Ca²⁺]_i no longer was elevated significantly (ratio, 1.11 ± 0.23 to 1.05 ± 0.19; n = 4; p = 0.3) (Fig.3A).

Fig. 3.

Effect of mitochondrial blockers on the membrane potential and [Ca²⁺]_i of AH/Type-II neurons. A, Typical example of the effect of FCCP on an AH/Type-II neuron. The compound hyperpolarization induced by FCCP is triggered and amplified by action potential firing. Thenumber of action potentials that are elicited is indicated under each stimulus. The top trace represents the [Ca²⁺]_i changes, whereas the membrane potential is shown in the bottom panel ofA. The presence of FCCP (1 μm) is indicated with a horizontal bar. Theinset shows a membrane potential recording from another neuron in control conditions (black), in the presence of FCCP (dark gray), and during current clamp in the presence of FCCP (light gray). Although clamped to the control potential, the amplitude of the AH remains reduced in FCCP.B, The average amplitude ± SEM of the compound hyperpolarization induced by FCCP, antimycin A, and rotenone in AH/Type-II neurons. C, The average time ± SEM to reach the maximum hyperpolarization. The number of neurons per condition is indicated in the bars.

To check the specificity of FCCP, we used two respiratory chain blockers, antimycin and rotenone, to block mitochondrial function. Similar to the effect of FCCP, antimycin (10 μm) and rotenone (10 μm) induced a compound hyperpolarization that was enhanced by AP firing. Antimycin caused a 6.9 ± 2.2 mV (n = 4) hyperpolarization, reaching its maximal amplitude in 450 ± 30 sec, and rotenone hyperpolarized the membrane 9.8 ± 3.8 mV (n = 3) in 255 ± 25 sec (Fig. 3B,C). The hyperpolarization in the presence of the respiratory chain blockers, however, never developed as rapidly as in FCCP. None of the three mitochondrial inhibitors had a detectable effect on the shape and duration of the APs.

AH and Ca²⁺ transient characteristics

We examined the properties of the AH and associated Ca²⁺ transients during mitochondrial blockade; only those neurons were included with an AH elicited by the same number of APs in control and drug conditions. The amplitude of both the AH (11.6 ± 1.1 to 10.4 ± 1.2 mV; p= 0.01; n = 4) and the accompanying Ca²⁺ transient were reduced (ratio, 0.077 ± 0.016 to 0.053 ± 0.011; p = 0.01;n = 4) in the presence of FCCP. When clamped back to the predrug resting membrane potential, the AH amplitude remained smaller (Fig. 3A, inset). Although the 90% duration of the AH and Ca²⁺ transients was prolonged (n = 3), relevant statistical analysis of the AH duration was not possible because in four of seven neurons the AH did not recover and merged continuously into the compound hyperpolarization induced by FCCP. To quantify the change in recovery, we measured the hyperpolarization 15 sec (the average 90% duration of the AH) after action potential firing. The neurons were hyperpolarized significantly more in the presence of FCCP (2.0 ± 1.4 to 5.3 ± 1.8 mV; p = 0.04; n = 7), and the residual Ca²⁺ elevation was increased slightly (ratio, 0.005 ± 0.003 to 0.011 ± 0.003;p = 0.02; n = 4).

Antimycin A did not significantly alter the amplitude of the AH (8.7 ± 2 to 7.9 ± 1.8 mV; p = 0.19;n = 4), but the 90% duration was prolonged by ∼40%. Similarly, the amplitude of the AH was not altered by rotenone (7.2 ± 1.6 to 9.7 ± 0.9 mV; p = 0.37;n = 3), and the 90% duration increased, with one neuron that never recovered. Similar to the hyperpolarizing effect of the respiratory chain blockers, the effect on the AH was slower and less pronounced than that of FCCP. Also, unlike FCCP, a prolonged exposure to rotenone and antimycin was required for the drug to become fully effective. None of the mitochondrial blockers altered the properties of the fast AH.

Input resistance

The AH that follows an action potential in myenteric neurons is caused by K⁺ efflux through Ca²⁺-dependent K⁺ channels (Hirst et al., 1985; Vogalis et al., 2002). Depending on the neuron type, the Ca²⁺ responsible for opening of these channels originates directly from the extracellular space or indirectly from the intracellular Ca²⁺ stores (Berridge, 1998). In myenteric neurons the Ca²⁺-induced Ca²⁺ release from ryanodine-sensitive stores provides the bulk of Ca²⁺responsible for this afterhyperpolarization (Hillsley et al., 2000). The opening of K⁺ channels leads to a decrease in input resistance (R_in), which can be monitored by measuring the amplitude of the voltage deflection elicited by a constant hyperpolarizing current pulse (Fig.4A). We set out to monitor the changes in R_in during the AH, which is regulated by the opening and gradual closure of Ca²⁺-activated K⁺ channels. Single hyperpolarizing pulses were given every second before and during the AH. During the AH theR_in dropped to 70% of the restingR_in, whereas in the presence of FCCP (1 μm) it dropped to almost 50% (Fig.4B). Similarly, although the effect developed more slowly, the drop in R_in was enhanced by antimycin A and was more pronounced for longer exposure times (69.8, 60.8, 55.2, and 45.6% for 42, 115, 217, and 340 sec of exposure, respectively) (Fig. 4C). The input resistance also was measured at the time the compound hyperpolarization in the presence of FCCP reached its maximum. Surprisingly, this compound hyperpolarization was not accompanied by a decrease in generalR_in when measured at the maximal FCCP effect (113 ± 30 compared with 92 ± 18 MΩ;n = 6; p = 0.16), suggesting the inhibition of other conductances.

Fig. 4.

The effect of FCCP (1 μm) and antimycin (10 μm) on the input resistance changes during the slow afterhyperpolarization in AH/Type-II neurons.A, The input resistance (R_in) is calculated by measuring the voltage deflection (X) for a fixed current step (0.2 nA) and is inversely dependent on the number of open channels in the membrane. B, The R_in(MΩ) during the first 15 sec of the AH in normal conditions (black) and in the presence of FCCP (gray). The data points are averages ± SEM of two to four trials in three neurons. Thearrow at time 0 sec indicates the time of the action potential. C, The R_in (MΩ) in control (black) and in the presence of antimycin A (gray). Individual traces were measured at different time points (as indicated) after the start of the antimycin application. The arrow at time 0 sec indicates the time of the action potential.

Role of intracellular Ca²⁺ stores

To investigate the contribution of Ca²⁺ release from the intracellular Ca²⁺ stores, we performed some experiments after the depletion of the Ca²⁺ stores with thapsigargin (Tg) and during the blockade of ryanodine (Ry) receptors, shown to be crucial for the myenteric AH (Hillsley et al., 2000; Vogalis et al., 2001). The blockade of the ATP-dependent Ca²⁺ pump with Tg did not have major effects on the membrane potential (∼5 mV depolarization in 30 min) but abolished the AH. The compound hyperpolarization caused by FCCP was attenuated greatly by Tg. Indeed, in three Tg-treated neurons FCCP did not cause a significant hyperpolarization (−59.6 ± 8.4 to −65.9 ± 6.8 mV; p = 0.07) even when APs were fired (Fig. 5A). In two neurons a small depolarization developed after 3 min of FCCP exposure during the next 10 min.

Fig. 5.

The involvement of intracellular Ca²⁺ stores in the effect of FCCP on AH/Type-II neurons. Drugs are added as indicated by the horizontal bars; the number of action potentials that are elicited is indicated under each stimulus. A, Effect of FCCP after blockade of the ATP-dependent Ca²⁺ pump with thapsigargin (2 μm). The AH is abolished almost completely after the thapsigargin treatment. Action potentials in the presence of FCCP induce only a small hyperpolarization.B, The effect of FCCP in the presence of ryanodine (10 μm). Ryanodine was applied 200 sec before the addition of FCCP. The FCCP-induced hyperpolarization develops slower than in the absence of ryanodine. C, The average amplitude ± SEM of the hyperpolarization induced by FCCP alone and in the presence of ryanodine. D, The average time ± SEM to reach the maximum hyperpolarization. In the presence of ryanodine the hyperpolarizing effect of FCCP is delayed significantly. The number of neurons per condition is indicated in the bars. Significant differences (p < 0.05) are indicated with an asterisk.

In preliminary experiments the AH neurons were perfused with Ry for at least 200 sec before exposure to FCCP. Here FCCP still induced a significant AP-dependent hyperpolarization (−54.0 ± 3.9 to −61.7 ± 5.5; n = 3; p = 0.03) (Fig. 5B). The amplitude of the compound hyperpolarization was not different from that of FCCP alone (Fig. 5C); however, the presence of Ry significantly delayed the effect (360 ± 121 compared with 146 ± 31 sec; p = 0.03) (Fig. 5D). This may indicate that Ca²⁺ release from Ry receptors is needed to accelerate the hyperpolarizing effect of FCCP. To test this hypothesis, we exposed neurons to Ry for >30 min, shown to reduce the AH to 40% (Hillsley et al., 2000). Similar to the Tg treatment, in Ry-treated neurons (n = 3) FCCP no longer evoked a hyperpolarization (−61.3 ± 3.9 to −67.0 ± 1.8 mV;p = 0.22) even in the presence of AP firing.

Effect of mitochondrial inhibitors on the membrane potential and [Ca²⁺]_i of S/Type-I neurons

S/Type-I neurons are heterogeneous in terms of firing characteristics, including phasic S neurons, having only 1 AP for a prolonged (100–500 msec) depolarization pulse, and tonic S neurons, with multiple spikes. The effect of mitochondrial inhibition was studied in 29 S/Type-I neurons (23 phasic and 6 tonic). All S/Type-I neurons were pooled because no consistent differences in the effect of mitochondrial inhibitors were observed between the two types of S neurons (Fig. 6A). FCCP (1 μm) significantly hyperpolarized S/Type-I neurons (−57.0 ± 3.9 to −67.9 ± 5.5 mV; p= 0.02; n = 6) in 189 ± 42 sec. However, this hyperpolarization differed from that observed in AH neurons in that it developed independently of AP firing (Fig. 6A). The respiratory chain blocker rotenone caused a similar hyperpolarization (−54.1 ± 3.6 to −65.4 ± 2.4 mV; p = 0.04;n = 4) (Fig. 6B), but it took longer for the hyperpolarization to develop fully (271 ± 73 sec) (Fig.6C). The complex I blocker antimycin hyperpolarized three of eight S/Type-I neurons (−43.0 ± 8.7 to −65.7 ± 4.6 mV;p = 0.03) in 280 ± 53 sec (Fig.6B,C) and had no effect on the other S/Type-I neurons (−51.8 ± 5.1 to −51.2 ± 5.9 mV; p = 0.5). Within the limits of detection none of the mitochondrial inhibitors had an effect on the shape of the APs of these neurons.

Fig. 6.

Effect of mitochondrial blockers on the membrane potential and [Ca²⁺]_i of S/Type-I neurons. A, Typical example of the effect of FCCP on an S/Type-I neuron. The hyperpolarization develops slowly and is independent of action potential firing. The top tracerepresents the [Ca²⁺]_i changes, whereas the membrane potential is shown in the bottom panel of A. The presence of FCCP (1 μm) is indicated with a horizontal bar.B, The average amplitude ± SEM of the hyperpolarization induced by FCCP, antimycin A, and rotenone in S/Type-I neurons. C, The average time ± SEM to reach the maximum hyperpolarization. The number of neurons per condition is indicated in the bars.

In 12 S/Type-I neurons the [Ca²⁺]_i was monitored simultaneously during the application of mitochondrial blockers. The link between [Ca²⁺]_i and membrane potential changes was not as clear in S/Type-I neurons as for AH/Type-II neurons (see also Shuttleworth and Smith, 1999; Hillsley et al., 2000). Although some changes in [Ca²⁺]_i were observed in the presence of FCCP, on average the [Ca²⁺]_i levels were not significantly different at the time the hyperpolarization reached its maximum (0.91 ± 0.11 to 0.95 ± 0.13;n = 5; p = 0.09). Similarly, antimycin and rotenone did not cause a significant [Ca²⁺]_i increase in S/Type-I neurons. The mitochondrial blockers did not alter significantly the AP-induced Ca²⁺transients elicited in S/Type-I neurons.

Effect of F₀F₁ ATPase blockade on the membrane potential and [Ca²⁺]_i in myenteric neurons

During mitochondrial blockade the ATP/ADP balance may change because of the lack of ATP production or ATP consumption by the depolarized mitochondria. Oligomycin (10 μm) blocks the mitochondrial F₀F₁ ATPase and therefore can be used to assess the contribution of ATP depletion. In five of seven AH neurons oligomycin induced a membrane hyperpolarization (−50.5 ± 3.9 to −57.1 ± 4.3 mV;p = 0.01) that was not influenced by AP firing (Fig.7A) and not accompanied by a [Ca²⁺]_i rise. This hyperpolarization was smaller (Fig. 7B) and developed significantly slower than the FCCP-induced compound hyperpolarization (800 ± 183 compared with 146 ± 31 sec; p = 0.005; n = 5) (Fig. 7C). In two other AH neurons a small depolarization (∼5 mV) was observed. Oligomycin did not alter the amplitude and duration of the AH (n = 5). Oligomycin had no detectable effect on the AP of AH/Type-II neurons. In four AH neurons that were hyperpolarized with FCCP, we added the K_ATP blocker glibenclamide (1 μm) at least 6 min after the addition of FCCP. The hyperpolarized neurons partially recovered (from −76.6 ± 6.4 to −67.9 ± 5.2 mV; ANOVA; p < 0.01 in FCCP; control, −66.6 ± 6.3 before FCCP), indicating the involvement of a K_ATP conductance. Blockade of the F₀F₁ ATPase in S/Type-I neurons had variable effects; four of seven neurons hyperpolarized slowly (−48.9 ± 3.7 to −58.8 ± 3.8 mV in 420 ± 100 sec), whereas in three others a depolarization (∼7 mV) was observed. Oligomycin did not induce a rise in [Ca²⁺]_i in S/Type-I neurons (n = 3).

Fig. 7.

The effect of the mitochondrial F₀F₁ ATPase blocker oligomycin (10 μm) and FCCP on the membrane potential of AH/Type-II neurons. Oligomycin is added as indicated by the horizontal bar; the number of action potentials that are elicited is indicated under each stimulus. A, Oligomycin hyperpolarizes the membrane potential of AH/Type-II neurons. The hyperpolarization develops slowly and is not triggered by action potential firing. B, The average amplitude ± SEM of the hyperpolarization. C, Average time ± SEM to reach the maximum hyperpolarization induced by oligomycin compared with the effect of FCCP. The number of neurons per condition is indicated in the bars. Significant differences (p< 0.05) are indicated with an asterisk.

MitoTracker, ER tracker, and immunohistochemical stainings

Longitudinal muscle myenteric plexus preparations were stained with ER tracker and MitoTracker (Fogarty et al., 2000; Buckman et al., 2001) to visualize the ER and the mitochondria in the myenteric neurons of the guinea pig ileum. Because the myenteric plexus consists of a monolayer of neurons, we used conventional fluorescence microscopy. Out-of-focus light, emerging from the continuous underlying muscle layer, equally contributed to the whole field and therefore did not affect the assessment of organelle tracker dyes in different myenteric neurons. Although ER tracker images showed an overall staining of the preparation, the myenteric neurons stained brighter than the underlying muscle layer (Fig. 8A). Unlike the ER tracker staining, the intensity of MitoTracker staining (Fig. 8B) was higher in a certain subset of neurons (Fig. 8B, arrows) in the myenteric plexus. To identify this subset of neurons, we used an antibody against calbindin, a calcium-binding protein that is a marker for AH/Type-II neurons in various tissues (Iyer et al., 1988; Song et al., 1991;Furness et al., 1998). Confocal microscopy was used to examine preparations that were double labeled with MitoTracker (Fig.8C) and the calbindin antibody (Fig. 8D). This revealed a colocalization between calbindin-like immunoreactivity and intense MitoTracker staining (Fig. 8E). In Figure8F the MitoTracker intensity is plotted as a histogram for both calbindin-positive and calbindin-negative neurons, clearly indicating that calbindin-positive neurons have a more intense MitoTracker staining (Fig. 8F).

Fig. 8.

Endoplasmic reticulum (ER) tracker, MitoTracker, and calbindin-like immunostaining in whole-mount myenteric plexus preparations; scale bar, 20 μm. A, ER-Tracker Blue DPX was used to visualize the ER in the myenteric plexus. Note that ER-Tracker Blue DPX shows an overall staining of the ganglion and the underlying layers. B, MitoTracker Green was used to stain mitochondria in the same ganglion. The MitoTracker staining shows a different pattern, in that some neurons in the ganglion are brighter (arrows) than others (arrowheads).C–E, Confocal images of a MitoTracker (C) calbindin (D) double staining. The preparations were stained with MitoTracker Green and processed afterward with a primary antibody against calbindin and a secondary antibody coupled to Texas Red. E, The calbindin-like immunoreactive neurons (e.g., arrows) were stained intensely with MitoTracker, as clearly seen in the overlay of C and D. F, Distribution histogram of the MitoTracker intensity for calbindin-like immunoreactive neurons (red) and calbindin-negative neurons (gray). The MitoTracker signal is clearly higher in calbindin-like immunoreactive neurons.

DISCUSSION

In this study we combined intracellular recordings and Ca²⁺ imaging to assess the effect of mitochondrial blockers on resting membrane potential and excitability of myenteric neurons of the guinea pig ileum. Mitochondria are important organelles not only because they are the major source of energy but also because they play a crucial role in Ca²⁺ homeostasis. Here we showed that the resting membrane potential and therefore the excitability of myenteric neurons, in particular AH/Type-II neurons, are regulated in part by Ca²⁺ uptake via mitochondria.

The AH/Type-II neurons displayed a double-phased afterhyperpolarization after AP firing. The slow phase was mirrored by a transient increase and exponential decay in [Ca²⁺]_i, closely linked to the afterhyperpolarization in amplitude and duration (Tatsumi et al., 1988; Hillsley et al., 2000; Vogalis et al., 2000). Hirst et al. (1985) suggested that the delay observed at the onset of the slow AH was attributable to a Ca²⁺ priming period of the K⁺ channels that were involved, whereas Vogalis et al. (2001) attributed the delay to a transient inward current, unmasked by TEA and blocked by niflumic acid.

The blockade of mitochondria hyperpolarized both S/Type-I and AH/Type-II myenteric neurons. In AH/Type-II neurons the resting membrane potential was not altered significantly by FCCP until action potentials were elicited. Then the action potentials and the AH were followed by a compound hyperpolarization that lasted for several minutes. Simultaneous [Ca²⁺]_i recordings showed that the compound hyperpolarization in AH neurons was accompanied by a rise in [Ca²⁺]_i, suggesting that mitochondria sequester an important amount of Ca²⁺ during normal neuronal function. However, unlike the observations in rat hippocampal neurons (Nowicky and Duchen, 1998), the rise in [Ca²⁺]_i did not develop in the absence of activation. In line with other reports (Greenwood et al., 1997; Pivovarova et al., 1999) the FCCP effect was reversible, and both membrane potential and [Ca²⁺]_i recovered during washout. In contrast to AH/Type-II neurons, the FCCP induced hyperpolarization in S/Type-I neurons developed steadily and independently of AP firing. To test whether the effect of FCCP was specific, we also used other mitochondrial blockers, rotenone and antimycin A, blocking complex I and III, respectively, of the respiratory chain. Hyperpolarizations of similar amplitude were observed in both S/Type-I and AH/Type-II neurons. In AH/Type-II neurons AP firing enhanced the effect, similar to that of FCCP.

The effect of mitochondrial blockers on AH and [Ca²⁺]_i transient characteristics was studied in those neurons without an immediate compound hyperpolarization. FCCP reduced the AH and the accompanying Ca²⁺ transient. Several studies report increased [Ca²⁺]_iresponses to activation during mitochondrial blockade in different neuronal preparations (Khodorov et al., 1999; Pivovarova et al., 1999;Friel, 2000; Wang and Thayer, 2002). However, in cultured myenteric neurons mitochondrial blockade did not affect the amplitude of K⁺-induced Ca²⁺ transients but slowed the [Ca²⁺]_i decay (Vanden Berghe et al., 2002). In the present study, as well, mitochondrial blockade by FCCP significantly prolonged the duration of the Ca²⁺ transient. All of the effects on the Ca²⁺ transients invariably were reflected in effects on the AH. In the presence of FCCP the AH was smaller, but the time to reach the maximum and the total duration were increased. These findings indicate that hampered Ca²⁺ removal prolongs the [Ca²⁺]_i elevation and therefore delays the closure of Ca²⁺-activated channels. Similar observations for Ca²⁺-activated currents were made in smooth muscle cells (Greenwood et al., 1997), chick DRG neurons (Kenyon and Goff, 1998), and rat hippocampal neurons (Nowicky and Duchen, 1998). More evidence for this prolonged opening of Ca²⁺-activated K⁺ channels was found in theR_in measurements. Mitochondrial blockers amplified the drop in R_induring the AH. However, by the time the compound hyperpolarization reached its maximum, the R_in had recovered to control levels. Therefore, the compound hyperpolarization cannot be attributed simply to an increase in K⁺ conductance, unless another conductance is switched off accordingly. A possible candidate isI_h (Galligan et al., 1990; Rugiero et al., 2002), a cation current activated during hyperpolarization, which is reduced by metabolic inhibition and increases in [Ca²⁺]_i (Duchen, 1990).

Because mitochondria act both as a source and sink, the net effect on Ca²⁺ uptake is probably highly dependent on the kinetics of the blockade and the state of the mitochondria (Gunter et al., 2000). This also might explain why the AH and Ca²⁺ transient amplitude are affected differently by FCCP and rotenone. Also, the fact that the hyperpolarizing effect and the reduction inR_in by FCCP developed faster and more suddenly than for antimycin and rotenone was probably attributable to the mechanism of the action of the drugs. The latter two block the respiratory cycle and therefore inhibit mitochondrial function more slowly than the sudden collapse of mitochondrial potential induced by FCCP.

The importance of AH/Type-II neuronal intracellular Ca²⁺ stores was assessed with ryanodine and thapsigargin, an irreversible blocker of the ATP-dependent Ca²⁺ pump. Depletion of the ER with thapsigargin prevented the AP-induced compound hyperpolarization in the presence of FCCP. Similarly, prolonged exposure to Ry, shown to reduce the AH significantly (Hillsley et al., 2000), eliminated the compound hyperpolarization. Even short treatment with Ry already delayed the FCCP hyperpolarization without affecting the amplitude. The Ca²⁺ release from ryanodine-sensitive Ca²⁺ stores seems to be prerequisite for FCCP to exert its effect, proving that mitochondria normally buffer substantial amounts of the Ca²⁺ release from the ER in myenteric AH/Type-II neurons.

Although the maximal hyperpolarization induced by mitochondrial blockade was similar in S/Type-I neurons and AH/Type-II, in the former the hyperpolarization was independent of AP firing. This suggested that another mechanism is responsible for the hyperpolarization in S/Type-I neurons and maybe also in unstimulated AH neurons. To assess the contribution of the putative changes in cytosolic ATP/ADP ratio attributable to the mitochondrial blockade, we blocked the F₀F₁ ATPase with oligomycin. Oligomycin itself hyperpolarized the majority of AH neurons and had variable effects on S neurons. The hyperpolarization developed independently of AP firing rather than in FCCP. Also, the [Ca²⁺]_i did not rise during oligomycin, which led to the idea that a Ca²⁺-independent mechanism, probably based on the activation of ATP-sensitive K⁺channels, was involved (Otero and Carrasco, 1984; Roper and Ashcroft, 1995; Hyllienmark and Brismar, 1996). Glibenclamide, a blocker of K_ATP, partially restored the membrane potential in myenteric neurons that were hyperpolarized in FCCP. Liu et al. (1999) also presented evidence for such a K_ATPcurrent in myenteric neurons by showing that glucose deprivation induced a tolbutamide-sensitive hyperpolarization. They also showed that the subunits of the K_ATP channel (Kir_1.6 and sulfonylurea receptor) were present in both calbindin-positive and other myenteric neurons. It is not clear what determines the different responses to oligomycin. Especially for S/Type-I neurons, different subtypes (phasic/tonic) and underlying differences in energy requirement may cause the variable effects. In AH neurons a shortage of ATP also may, besides activating K_ATP, slow down the ATP-dependent Ca²⁺ uptake by the ER (Landolfi et al., 1998).

In this study we also used fluorescent tracker molecules to trace mitochondria and ER in the myenteric plexus. The ER tracker stained the whole tissue, with a more extensive staining of the myenteric plexus. MitoTracker showed a more patterned staining, in which some neurons were brighter than others. In a double-labeling experiment the neurons with intensive MitoTracker staining were identified as calbindin-positive AH/Type-II neurons. This suggested that AH/Type-II neurons had a higher number or a denser mitochondrial network. Electron microscopy data showing that calbindin-positive neurons, unlike the calbindin-negative neurons, had numerous electron-dense mitochondria (Pompolo and Furness, 1988) corroborate our findings. These data tie in well with the physiological observations that in AH/Type-II neurons the effect of mitochondrial blockers is more pronounced than in S/Type-I neurons.

In conclusion, the blockade of mitochondria triggers a series of events in AH/Type-II neurons involving several currents (Fig.9). First, the Ca²⁺ released from ER is not buffered efficiently by the impaired mitochondria, slowing the deactivation of Ca²⁺-dependent K⁺ conductances and leading to a prolongation of the AH. Second, the glibenclamide data imply that ATP-sensitive K⁺ channels also are activated in the course of the compound hyperpolarization. This mechanism is likely to explain the oligomycin-induced hyperpolarization that develops without a [Ca²⁺]_i rise. Further, the recovery of R_in at the peak of the compound hyperpolarization implies the inhibition of a nonselective cation current. The contribution of these different conductances probably varies throughout the course of the compound hyperpolarization. In an early stage, action potential firing and consecutive ER Ca²⁺ release trigger the Ca²⁺-dependent component of the compound hyperpolarization. However, the ATP-dependent component also might be set off because, in an attempt to remove the excess Ca²⁺ ions, activated neurons might consume the remaining ATP more rapidly. Although mitochondrial blockade hyperpolarizes both types of myenteric neurons, the effects on S/Type-I neurons are less pronounced and not triggered by activation. It has to be noted that the S/Type-I neurons are a mixed population, comprising descending and ascending interneurons and inhibitory and excitatory motor neurons. Although the link between their function and the electrical properties is not established clearly, it is likely that functional differences are reflected in their electrical and Ca²⁺ handling properties. Mitochondria are highly sensitive to oxidative stress and therefore may be important in the pathological responses during inflammation. Reactive oxygen species induced fluctuations in the membrane potential of colonic AH myenteric neurons (Wada-Takahashi and Tamura, 2000). In view of the results in this study, it is likely that mitochondrial dysfunction and therefore disabled Ca²⁺ sequestration underlie the initial hyperpolarizations.

Fig. 9.

Schematic representation of the Ca²⁺ handling in myenteric neurons. Action potential firing causes Ca²⁺ entry through voltage-activated Ca²⁺ channels. The increase in [Ca²⁺]_i is amplified by the release of Ca²⁺ from the ER via activation of ryanodine receptors (Hillsley et al., 2000). Functional mitochondria sequester Ca²⁺ and thereby lower the [Ca²⁺]_i. A blockade of the mitochondria results in a prolonged elevation of the [Ca²⁺]_i, which in turn causes a prolonged activation of the Ca²⁺-dependent conductances. On the other hand, ATP depletion, occurring during mitochondrial blockade or the inhibition of the F₀F₁ ATPase by oligomycin, may activate K_ATP channels, thus leading to a Ca²⁺-independent hyperpolarization.