INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

AH neurons, which function as intrinsic primary afferent neurons (IPANs) in the myenteric plexus of the small intestine, are characterized by prolonged hyperpolarizations, after firing one or a few action potentials (APs) in rapid succession (Furness et al. 1998; Hirst et al. 1974). The postexcitation afterhyperpolarization (AHP), which follows the undershoot of the action potential (AP), lasts from ~2 to 10 s and can increase the threshold for AP initiation to the point where these AH (afterhyperpolarization) neurons are unable to fire (Hirst et al. 1974, 1985; Morita et al. 1982; Vogalis et al. 2000b; Wood and Mayer 1978). The AHP is associated with an increase in the membrane conductance to K⁺ (Hirst et al. 1985) and is delayed in onset by 40-80 ms following repolarization of the AP. Ca²⁺ entry during the AP is critical for triggering the AHP (Hirst et al. 1985), and measurement of changes in intracellular Ca²⁺ using ratiometric Ca²⁺-sensitive fluorescent dyes in intact ganglia (Hanani and Lasser-Ross 1997) and in cultured myenteric neurons (Vogalis et al. 2000a) has shown that cytoplasmic [Ca²⁺] rises significantly during the AHP. These findings suggest that intracellular Ca²⁺ gates the AHP channels. However the relationship between Ca²⁺ entry and Ca²⁺ release from stores and the type of K⁺ channel that is activated to produce the AHP are unknown, although it is known that the K⁺ conductance underlying the AHP is not voltage dependent (North and Tokimasa 1987).

Ca²⁺-dependent AHPs are generated by other types of neurons including rabbit vagal afferent neurons (Cordoba-Rodriguez et al. 1999), rat vagal motoneurons (Sah 1995), and hippocampal pyramidal neurons (Sah and Isaacson 1995). In each of these, the immediate source of Ca²⁺ that is required to activate the Ca²⁺-activated K⁺ conductance (g_K-Ca) that underlies the AHP comes from internal Ca²⁺ stores. The AHP and underlying current (I_AHP) are decreased by treatment of cells with ryanodine, which inhibits Ca²⁺-release channels on the endoplasmic reticulum (Jobling et al. 1993). The role of Ca²⁺ stores in the AHP in AH myenteric neurons has not been investigated, although ryanodine-sensitive Ca²⁺ stores have been demonstrated in cultured myenteric neurons (Kimball et al. 1996). In addition, the identity of the voltage-gated Ca²⁺ channel(s) through which Ca²⁺ enters the cell to trigger unloading of Ca²⁺ stores is not properly defined. In guinea pig sympathetic neurons, a large fraction of the "trigger" Ca²⁺ enters through dihydropyridine-sensitive (L-type) Ca²⁺ (Davies et al. 1999) while in neurons of the rat superior cervical ganglion, Ca²⁺ entry associated with the AHP occurs through N-type Ca²⁺ channels (Davies et al. 1996). In hippocampal neurons, L-type Ca²⁺ channels are co-localized with small-conductance Ca²⁺-activated K ⁺ (SK) channels, allowing Ca²⁺ to directly stimulate SK channels without triggering secondary Ca²⁺ release (Marrion and Tavalin 1998). The characteristic delay in the onset of the AHP, according to this scheme, is due to the depolarization-evoked "delayed facilitation" of the opening of L-type Ca²⁺ channels at resting potentials (Cloues et al. 1997).

In the guinea pig myenteric plexus, the Ca²⁺ channel currents recorded from AH neurons that have Dogiel type II morphology and most of which stain positively for calbindin are generated by a mixture of high-voltage-activated (HVA) channels of the N and P/Q type (Starodub and Wood 1999; Zholos et al. 1999). The "hump" on the action potential, which is a characteristic feature of AH neurons, is abolished by N-type Ca²⁺ channel blockers such as -conotoxin GVIA (Furness et al. 1998), suggesting that the AHP is initiated by Ca²⁺ entry mainly through N-type Ca²⁺ channels, which then leads to the activation of the g_K-Ca. Although the AHP is largely due to an increase in g_K-Ca, activation of inwardly rectifying K⁺ (K_ir) channels by membrane hyperpolarization may contribute to the maintenance of the AHP (Zholos et al. 1999). The channels responsible for the increase in g_K-Ca in myenteric AH neurons have not been identified at the single-channel level. In vagal motoneurons (Sah 1995) and in hippocampal pyramidal neurons (Sah and Isaacson 1995), the unitary conductance of the AHP channels in these cells was estimated to be <10 pS using variance analysis of whole cell currents.

The AHP has been suggested to be important in controlling AH cell excitability (Wood 1994). The presence of the AHP enables the soma of multipolar AH cells to gate the conduction of APs from one process to the others (Wood and Mayer 1978). Modulation by neurotransmitters of the ability of AH cells to generate AHPs is a primary mechanism by which the excitability of AH neurons is regulated. In the intact plexus, IPANs form self-reinforcing networks by releasing transmitters that produce slow postsynaptic excitation and inhibit the AHP, thus amplifying the excitability of the network as a whole (Kunze et al. 2000).

In the present study, we investigated the sequence of events between the occurrence of a somatic action potential and the generation of the AHP. We used patch-clamp techniques to record whole cell currents and the channel activity from neurons within intact ganglia.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dissection and enzymatic cleaning of myenteric ganglia

Guinea pigs were stunned and killed by cutting the spinal cord followed by exsanguination. All procedures have been approved by the University of Melbourne Animal Experimentation Ethics Committee. A 1- to 2-cm-long segment of the proximal duodenum was removed and placed in a recording dish filled with oxygenated physiological salt solution (PSS) of the following composition (in mM): 118.1 NaCl, 4.8 KCl, 25 NaHCO₃, 1.0 NaH₂ PO₄, 1.2 MgSO₄, 11.1 glucose, and 2.5 CaCl₂. The PSS was maintained at room temperature (17-20°C) during dissection but was heated to 35-37°C during recording. Nicardipine (1 µM) was added to the PSS to block spontaneous muscle contraction. The myenteric plexus was exposed by dissecting away the mucosa, submucosal plexus, and circular muscle under a binocular microscope, and the preparation was then pinned out in a transparent recording dish that was mounted on the stage of an inverted microscope (Olympus CK40, Japan). The surface of one ganglion was exposed to 0.01-0.02% protease type X1V (Sigma, http://www.sigma-aldrich.com) dissolved in PSS, and the upper surfaces of neurons were cleaned by sweeping with a hair over a portion of the ganglion (Kunze et al. 2000). Conventional whole cell and cell-attached patch-clamp recordings were made. Voltage and current signals were low-pass filtered (4-pole Bessel filter) at 1-5 kHz and amplified using an Axopatch 200B (Axon Instruments). The Axopatch 200B amplifier was driven by Clampex 8 acquisition software through a Digidata 1200A AD/DA interface (Axon Instruments), running on a Pentium PC. Acquisition rates ranged from 2 to 10 kHz; for current recordings, acquisition rate was at least twice the analog filter cutoff frequency on the amplifier. Voltage and current signals were processed (filtered, digitally subtracted, averaged) using Clampfit 8 (PClamp, Axon Instruments). Patch electrodes were pulled from standard-wall-thickness borosilicate capillary tubing (GC150F-10, Harvard Apparatus) to have resistances of 4-10 M when filled with a high-K⁺ pipette filling solution of the following composition (in mM): 130 KCl, 10 NaCl, 1 MgCl₂, 0.5 CaCl₂, 10 HEPES, 1 EGTA, and 2 ATPK₂. This solution was titrated to a pH of 7.35 with 4 M KOH, which resulted in the addition of 11 mM of K⁺. This yielded a final [K⁺] in the pipette solution of 145 mM. Patch pipettes with resistances of 15-20 M were used for cell-attached recordings, and their tips were fire-polished using a microforge (Narishige, Japan). The [Ca²⁺] of the pipette solution was estimated to be ~85 nM at 35°C using Maxchelator (http://www.standford.edu/~cpatton/maxc.html). A junction potential of 3.4 mV was calculated to exist between the pipette solution and bathing solution (Clampfit); this value was not added to the recorded membrane potential.

Preparations were perfused continuously at ~1 ml/min with PSS preheated to 35-37°C. Drugs were added directly to the perfusate. The following compounds were used: caffeine (Sigma, St. Louis, MO), ryanodine (Sigma), niflumic acid (Sigma), charybdotoxin (Latoxan), tetraethylammonium chloride (Sigma), and -conotoxin GVIA (kindly provided by Dr. C. Wright, Department of Pharmacology, Univ. of Melbourne). All the drugs with the exception of ryanodine and niflumic acid were dissolved in distilled water; ryanodine was dissolved in DMSO at stock concentration of 10² M and niflumic acid was dissolved in DMSO at 10¹ M.

Analysis of data

Current records were analyzed using Clampfit (PClamp8). Mean current-variance analysis was performed on selected current traces of the I_AHP using IgorPro and user-defined macros (Wavemetrics). This entailed fitting the raw current trace with a polynomial curve that was subtracted from the raw I_AHP current to obtain the difference current. The fitted curve (or mean current) was divided into nonoverlapping segments. The variance (²) of each segment of difference current was plotted as a function of the average of the corresponding segment of mean current, following subtraction of the baseline variance and mean current. The data points were then fitted with the following equation (Heinemann and Conti 1992)

(1)

to obtain estimates for i, which is equal to the single-channel current, and N, which represents the maximum number of channels in the cell. I_mean is the mean whole cell current. Fitting was performed using a least-squares procedure in IgorPro.

We also performed power spectral density (PSD) analysis on time segments of the difference current of the I_AHP to determine shifts in the cutoff frequency (f_c) of the power spectral density (PSD) of the current at the peak of the I_AHP versus the baseline current. This was done using a built-in fast Fourier transform (FFT) function in IgorPro on 8,192-point segments of difference-current. Segment length was set to 512 points, and a Hanning filter window was used. The PSD (A²/Hz) calculated from the record was plotted as a function of frequency (Hz) on log-log plots and fitted with single Lorentzian functions (in IgorPro) of the form

(2)

where S_o is the maximum power asymptote value, f is the frequency, and f_c is the corner frequency. For a channel having only two states (i.e., closed and open), f_c is equal to ( + )/2 (Heinemann et al. 1992) where  is rate constant for open-to-closed transitions (and therefore 1/ is the mean open dwell time) and  is rate of closed-to-open transitions (and 1/ is the mean closed dwell time). I_mean has a linear dependence on _c [equal to 1/( + ), which approaches 1/ as I_mean approaches zero; the value of 1/ can be obtained by extrapolation when I_mean is zero (Valiante et al. 1997)].

Single-channel recordings in cell-attached patch

In the cell-attached configuration, we were able to record channel activity that followed a discharge of action potentials (recorded as action currents). Such recordings were digitally low-pass filtered off-line in Clampfit (200-300 Hz, acquired at 5 kHz) in an attempt to resolve small-amplitude single-channel current levels. All-points histograms were then constructed from filtered current traces [durations of 2-4 s immediately following the spike(s)], and the histograms were fitted with the sum of 6 Gaussians. The probability of channel opening (NP_o) was estimated by multiplying the area under each Gaussian component with the number of component simultaneous channel openings, and summing the products.

Statistics

Averaged data are presented as the means ± SE where n is the number of cells. Statistical significance (P < 0.05) was determined using the unpaired t-test unless otherwise stated.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Whole cell recordings of the AHP and underlying current, I_AHP

About two-thirds of the cells that were patch-clamped in myenteric ganglia from the guinea pig duodenum were AH neurons. This was based on the presence of a prominent AHP after a single action potential (AP) or a compound AHP that followed a short volley (usually 3) of APs at 50 Hz (see Fig. 1B). The AHP peaked between 400 and 1,000 ms following the APs and decayed over the following 5-10 s. During depolarizing constant current steps (150-ms duration), AP firing in AH cells strongly accommodated within 50-100 ms (Fig. 1A) due to the onset of the AHP. In AH cells, hyperpolarizing current steps produced electrotonic potentials with a characteristic sag (Galligan et al. 1990). The input resistance (R_in) of AH cells under these conditions, determined from the magnitude of voltage deflections in response to a 40-pA hyperpolarizing current steps, was 308 ± 18 M (n = 67), and their resting potential (54 ± 0.9 mV, n = 108) was more negative than in the remaining population of cells (40 ± 1 mV, n = 55). The durations of APs at 50% peak amplitude in AH cells (2.3 ± 0.07 ms, n = 112) were longer than in the rest of the cells (1.3 ± 0.05 ms, n = 59; Fig. 1C). These properties of AH cells are similar to those reported using intracellular sharp microelectrodes with the exception that the amplitude of APs was greater and the plateau component was more pronounced in the current patch-clamp recordings as compared with microelectrode recordings of APs (Hirst et al. 1974). Two AH cells were filled with neurobiotin; both stained positively for calbindin and had multiple processes (Dogiel type II morphology).

View larger version (9K): [in this window] [in a new window]   Fig. 1. Electrical activity of AH neurons recorded in current-clamp mode in the whole cell patch configuration. A: injection of a 40-pA depolarizing current pulse triggered a burst of action potentials (APs). B: 3 10-ms depolarizing current pulses (0.2 nA) delivered at 50 Hz () triggered APs that were followed by a slow afterhyperpolarization (AHP). C: an AP recorded from an AH cell is shown on a fast time-base showing the "hump" on the falling phase of the action potential. - - -, the resting potential.

To record the ionic current produced by the increase in g_K-Ca, AH neurons were voltage-clamped in the whole cell mode with the holding potential set to 65 or 80 mV. The cells were then depolarized to test potentials positive of +35 mV for short durations (20-100 ms) to open voltage-gated Ca²⁺ channels and trigger Ca²⁺ entry, and then the membrane potential was returned to 55 mV to record the evolving outward AHP current (I_AHP). The baseline current at 55 mV in the absence of a test depolarizing stimulus was also recorded, over the same time period, for comparison and subtraction from the stimulus-evoked I_AHP (Fig. 2A). The I_AHP recorded at 55 mV activated with a characteristically slow onset that could be fitted over most of its activation phase with a single exponential function with a time constant (_on) of 326 ± 68 ms, n = 14. The value of _on did not vary with membrane potential (e.g., at 70 mV _on = 309 ± 72 ms, n = 12). The time course of the decay of the I_AHP was more variable. The late decay phase of the I_AHP recorded at 55 mV could be fitted with a single exponential with a time constant of 22.7 ± 7 s (n = 12).

View larger version (15K): [in this window] [in a new window]   Fig. 2. Reversal potential of the IAHP. A: in an AH cell held at 65 mV, the membrane potential was step-depolarized to +35 mV for 50 ms and then returned to a holding potential (Vh) of 55 mV to record the slowly developing outward AHP current (IAHP). Once the slow outward current had reached a plateau, a ramp depolarization was applied (120 to 40 mV). Superimposed is the baseline current recorded at 55 mV in the absence of the step depolarization to +35 mV showing the small net outward current at 55 mV (dashed line represents 0 current level). B: the difference ramp IAHP is plotted as a function of ramp potential between 120 and 40 mV (thin trace). The IAHP is zero at 82 mV and shows outward rectification. The IAHP was fitted with a modified GHK current equation of the form: IAHP = P*F/S** [[K+]out - [K+]in*exp(/S)]/[1  exp(/S)], where  = membrane potential, P is a permeability constant for the membrane of the whole cell, and S = RT/F, which is equal to 27 mV at 35°C. The fitted line (thick line) follows the IAHP over the entire voltage range. P was equal to 0.156*109 cm3/s.

The reversal potential of the I_AHP was determined by applying ramp depolarizations to the AH cells (120 to 40 mV, for 0.5-1 s), in the absence and during the plateau phase of the I_AHP (Fig. 2A). The reversal potential of the I_AHP was determined by subtracting the ramp current recorded in the absence of the I_AHP from the ramp current recorded during the I_AHP and plotting the difference current as a function of ramp potential (Fig. 2B). The ramp-evoked I_AHP was equal to 0 at 89 ± 2 mV (n = 22; not corrected for a 3.4 mV junction potential), indicating that it was generated mainly by an increase in the membrane permeability to K⁺; E_K was calculated to be 89 mV from the Nernst equation ([K]_o 4.8 mM: [K]_pipette 145 mM). The fact that the I_AHP could be fitted with a modified GHK current equation for asymmetric K⁺ concentrations (Fig. 2B) indicates that it was generated by a K⁺ conductance that is not voltage dependent over the voltage range of 120 to 40 mV.

Role of Ca²+ entry and Ca²⁺ release in the I_AHP

The AHP and the I_AHP were dependent on Ca²⁺ entry through voltage-gated Ca²⁺ (Ca_V) channels because they were both blocked by adding Co²⁺ (2 mM) or Cd²⁺ (0.1 mM) to the bathing solution. -Conotoxin GVIA (GVIA; 0.6-1 µM) greatly attenuated the AHP in nine AH cells treated with this toxin (Fig. 3A). The suppression of the AHP caused AH cells to fire in a tonic manner when injected with depolarizing current (Fig. 3B, i and ii). In five of these AH cells in which recordings were obtained before and after application of GVIA, the AP_half-dur was significantly (P < 0.05, paired t-test) decreased from 2.46 ± 0.45 to 1.79 ± 0.30 ms. The RMP was unchanged following GVIA application (control, 61 ± 4 mV; GVIA, 57 ± 7 mV) but the I_AHP was significantly (P < 0.05) decreased from 532 ± 196 to 19 ± 14 pA at its peak (Fig. 3C). This suggests that the Ca²⁺, which is required for the generation of the AHP, enters through N-type Ca_V channels. Block of N-type Ca_V channels did not significantly change the resting potential of AH cells nor the R_in (control: 248 ± 44 M; GVIA, 225 ± 35 M; n = 2). However, both Cd²⁺ and Co²⁺ depolarized the resting potential of AH cells by 10-15 mV and increased their R_in from 160 to 375 M (Cd²⁺, 0.1 mM) and from 132 to 267 M (Co²⁺, 2 mM), as has been reported previously (North and Tokimasa 1987). These results indicate that Ca²⁺ entry at rest is mediated by non-N-type Ca²⁺ channels.

View larger version (23K): [in this window] [in a new window]   Fig. 3. Effect of -conotoxin GVIA on AH neurons. A: a depolarizing current step (500 ms, 120 pA) evoked APs and a prolonged AHP. Following wash in of GVIA (1 µM) the AHP was markedly decreased. B: responses during current injection before (i) and after (ii) addition of GVIA to the preparation. Note loss of adaptation and firing throughout the stimulus. Same cell as in A. C: GVIA profoundly decreased the generation of the IAHP recorded under voltage-clamp at 55 mV (same cell as in A and B).

Studies conducted on other neuronal types with similar slowly activating AHPs have suggested that the g_K-Ca is activated by Ca²⁺ released from intracellular stores by Ca²⁺ entry (Ca²⁺-induced Ca²⁺-release, CICR). To test whether CICR plays a part in the generation of the I_AHP in AH neurons, we used caffeine to block Ca²⁺ release channels on the endoplasmic reticulum. A typical set of recordings showing the effect of caffeine on the I_AHP is plotted in Fig. 4A. Within 10 min of wash-in, caffeine (5 mM) almost fully inhibited the I_AHP evoked by a test depolarization to +35 mV for 50 ms. The recording in control solution in Fig. 4A also illustrates the very slow decay of the I_AHP, which in this cell persisted for tens of seconds. The inhibitory action of caffeine was partially reversible, and the I_AHP began to recover on wash out of caffeine, although the magnitude of the I_AHP was not restored to its pretreatment level (Fig. 4A). On average, the peak I_AHP was decreased significantly (P < 0.05, paired t-test) from 436 ± 137 to 47 ± 28 pA (n = 9) following treatment with 2-5 mM caffeine. Wash-in of caffeine-containing solution was usually accompanied by a transient increase in I_AHP. However, we did not observe any large (>100 pA) transient increase in the outward current at 55 mV. Records taken in current-clamp mode showed that in the presence of caffeine the AHP was abolished (Fig. 4B) but caffeine had no effect on the AP half-duration (Fig. 4Ci). Caffeine treatment resulted in membrane depolarization (Fig. 4D) and in an increase in inward current at 55 mV (see Fig. 4A in the presence of caffeine) and a decrease in the input resistance (Fig. 4Cii).

View larger version (22K): [in this window] [in a new window]   Fig. 4. Inhibition of IAHP by caffeine. A: the cell was stimulated every 50 s with a 50-ms depolarizing step to +35 mV, from a holding potential (Vh) of 65 mV, and the slow outward IAHP was recorded at 55 mV (control). Ten minutes following wash in of caffeine (5 mM), IAHP was almost abolished. IAHP partially recovered following wash out of caffeine. B: caffeine decreased the AHP in another AH cell and depolarized the cell. C: caffeine had little effect on the APhalf-dur (i) but decreased the input resistance of the AH cell (ii), as measured by the voltage change in response to a constant-current hyperpolarizing pulse.

The action of ryanodine on the I_AHP was similar to the effect of caffeine. Ryanodine (10-20 µM) decreased the I_AHP significantly (P < 0.05, paired t-test) from 802 ± 167 to 106 ± 51 pA (n = 8; Fig. 5, A and D); similarly the AHP was decreased or suppressed by ryanodine from 8 to 1 mV in six cells tested (Fig. 5B). The effect of ryanodine was slow in onset and required 30 min of exposure to the drug before significant reduction in the I_AHP was apparent (Fig. 5B). In control solution, I_AHP showed little run-down over this period after establishing the whole-cell recording configuration. In the presence of ryanodine, the AP_half-dur was also decreased from 2.4 ± 0.4 to 1.6 ± 0.12 ms (n = 5; P = 0.052, unpaired t-test) while the resting potential of AH cells was increased from 58 ± 6 to 71 ± 2.4 mV (n = 6; P = 0.09, unpaired t-test).

View larger version (23K): [in this window] [in a new window]   Fig. 5. Suppression of IAHP by ryanodine. A: an AH neuron was stimulated to generate an IAHP at 60 mV by 50-ms step depolarization to +35 mV. Following 30-min of wash-in of ryanodine (20 µM), the amplitude of the IAHP was markedly reduced. - - -, zero current level. B: time course of the suppression of the IAHP by ryanodine (10-20 µM). Note slight run-up in IAHP on wash-in of ryanodine (10 µM). C: in the same cell depicted in A, the AHP triggered by a volley of 3 pulses at the beginning of the traces was almost abolished following treatment with ryanodine. , the resting potential (60 mV) prior to the addition of ryanodine. D: mean data. Ryanodine (10-20 µM) decreased the peak IAHP from 802 ± 167 (C, control) to 106 ± 51 pA (R, ryanodine) (n = 7).

Thapsigargin (1 µM) treatment also reduced the I_AHP in four AH cells tested from 217 ± 65 to 137 ± 83 pA (P = 0.08, unpaired t-test) over 20-30 min. This reduction in the I_AHP was associated with the development of an outward current at 55 mV from 25 to 7 pA, consistent with hyperpolarization of the resting potential of these cells.

A further indication that CICR was necessary for the generation of the I_AHP was the absence of the AHP in AH cells that were dialyzed internally with a pipette solution containing 10 mM EGTA and no added Ca²⁺. In three such cells tested, the AHP and associated I_AHP were absent. The AP_half-dur in these cells was longer (4.5 ms) than in AH cells that were perfused with standard internal solution, indicating that the repolarization of the action potential is dependent in part on a Ca²⁺-dependent K⁺ conductance.

The characteristically slow onset of the AHP and of the I_AHP following the step-depolarizing stimulus (e.g., see Fig. 2A) may reflect the time taken for Ca²⁺ entry to trigger Ca²⁺ release from stores and for the propagated Ca²⁺ wave to reach the AHP channels. Another contributing factor may involve concomitant activation of a post-excitation transient inward current following repolarization. Such an inward current was recorded in eight AH cells following pharmacological suppression of the I_AHP with either TEA (20 mM) or charybdotoxin (50-400 nM). TEA (20 mM) suppressed the I_AHP from 341 ± 31 to 15 ± 15 pA (n = 2), and this was associated with the development of an inward current (30 pA at 55 mV), consistent with membrane depolarization. As shown in Fig. 6A, following suppression of the I_AHP with TEA (20 mM), a post-depolarization inward current was evident immediately following repolarization and decayed within 1-2 s. Exponential fits to the decay of the current yielded time constants of the order of 300-500 ms. The reversal potential of the inward current was extrapolated to lie between 35 and 40 mV (n = 2; Fig. 6B), and the current was inhibited by niflumic acid (100 µM) in six cells tested. In two cells tested, this transient inward current was blocked by Cd²⁺ (0.2 mM), indicating that is was dependent on Ca²⁺ entry. These results suggest that this inward current may be activated under normal conditions following action potential firing in AH neurons and by summating with the I_AHP may contribute to the apparent delay in the onset of the AHP and to its slow rise time.

View larger version (17K): [in this window] [in a new window]   Fig. 6. Block of transient poststimulus inward current by niflumic acid. A: superimposed are poststimulus currents recorded at Vh = 55 mV under control conditions, following addition of TEA (20 mM), and following the addition of niflumic acid (100 µM) with TEA present. TEA, which blocks the IAHP and other K+ channel currents, unmasked a postdepolarization transient inward current that was blocked by the addition of niflumic acid. B: the peak poststimulus transient inward current is plotted as a function of Vh showing that the reversal potential was extrapolated to 36 mV.

K+ channels that open during the AHP

Both the lack of voltage dependence in the ramp I_AHP and in the activation time constant of the I_AHP suggest that the AHP channels are not gated by voltage but are gated primarily by intracellular Ca²⁺. To characterize these channels further, and given the wide range in conductance values of Ca²⁺-activated K⁺ (K_Ca) channels (Vergara et al. 1998), we determined the unitary conductance of the AHP channels by applying variance-mean analysis on I_AHP currents, of the type shown in Fig. 7A (see METHODS). The decay phase of these I_AHP currents was sufficiently slow to enable us to measure the variance of the current using digitization rates of 2-5 kHz. As shown in Fig. 7C, subtraction of the fitted mean current trace (Fig. 7B) from the raw I_AHP (Fig. 7A, recorded at a clamp voltage of 55 mV) yielded a difference current whose noise was greater at time-points just after the peak of the I_AHP. A plot of the mean variance of the difference current versus the average mean current over the corresponding time segments yielded a relationship that could be well fitted with Eq. 1 (Fig. 7D). The bell-shaped relationship between the mean current and the variance of the difference current in Fig. 7D indicates that the open probability of the AHP channels had exceeded 0.5 at the peak of the I_AHP. From such plots we obtained estimates of the single-channel current (i) and of the maximum number of AHP channels in AH cells (N). In 25 AH cells, the unitary current at 55 mV averaged 0.27 ± 0.02 pA, and the maximum number of AHP channels per cell was estimated to be 3,890 ± 625 channels. Assuming a reversal potential of 89 mV, this yields a unitary chord conductance for AHP channels of 7.9 ± 0.6 pS.

View larger version (22K): [in this window] [in a new window]   Fig. 7. Mean-variance analysis of the IAHP in AH cells. A: a typical IAHP evoked by a 50-ms step-depolarization to +50 mV and recorded at a holding potential (Vh) of 55 mV. The current peaks in 2-3 s and decays slowly over 25 s. The IAHP was low-pass filtered on-line (1 kHz, 4-pole Bessel filter) and acquired at 2 kHz. B: polynomial fit to the IAHP described the mean current. C: the difference current was obtained by subtracting the mean current (polynomial line, B) from the raw IAHP (A) and shows the decrease in current noise with time, corresponding to the decay of the IAHP. D: the variance of each of 20 current bins of the difference current is plotted as a function of the arithmetic average mean current of the corresponding level. Both the mean current and variance have been corrected for the baseline mean current and variance. The data points have been fitted with the mean-variance Eq. 1 (see METHODS) and yielded a value for the single channel current (i) of 0.2 pA and N (the number of channels) was equal to 1,844. This gives a single-channel conductance of ~6 pS.

We also analyzed the I_AHP difference-current by plotting its power spectral density spectrum (PSD). As shown in Fig. 8A, the noise in the difference current at the peak of the I_AHP was considerably greater than the noise of the baseline current. FFT of the peak difference-current yielded a PSD curve (Fig. 8B, ) that asymptoted at a higher energy level than the baseline current (Fig. 8B, ). Subtraction of the baseline PSD curve (representing the noise spectrum of background channel activity) from the PSD curve at the peak of the I_AHP yielded a PSD curve that could be fitted with a single Lorentzian function (Fig. 8D) that had a cutoff frequency (f_c) of 133 Hz. Another PSD curve constructed from a segment of the difference-current mid-way between the peak of the I_AHP and the baseline current, and corrected for baseline noise, was also fitted with a Lorentzian function and yielded a f_c of 61 Hz. Because f_c = ( + )/2, a decrease in f_c with the decay of the I_AHP indicates a change in one or both of the rate constants governing channel opening or closing. Assuming that the open dwell time of AHP channels is constant (i.e., the dissociation of Ca²⁺ from the channel is constant), then the change in f_c is attributable to changes in the mean closed dwell time or 1/. The mean open time (or 1/) of AHP channels can be estimated from the value of the x intercept of a linear regression line fitted to a plot of mean current level versus _c [i.e., 1/(2f_c)] (Valiante et al. 1997). In data collected from six cells, the value of  averaged 357 ± 82 ms¹, which corresponds to a mean open time of 2.8 ± 0.6 ms. In the cell depicted in Fig. 8D, _c is equal to ~1 ms at the peak of the I_AHP (peak of I_mean). Given that _c = 1/( + ) and that the mean open time was 2.5 ms in this neuron, then the mean closed dwell time (1/) is equal to _c/(1  _c) or 1/(1  1/2.8) = 1.55 ms. From these values, the open probability of an AHP channel at the peak of the I_AHP in the neuron depicted in Fig. 8 is estimated to be 2.5/(2.5 + 1.55) = 0.62. 

View larger version (32K): [in this window] [in a new window]   Fig. 8. Spectral analysis of IAHP. A: representative segments of IAHP difference current at the peak (i) and baseline (ii) current levels. B: power spectral density (PSD) distributions of the 2 difference currents shown in A showing higher power levels at all frequencies of the peak difference current (i). C: the PSD distributions of 2 difference currents, corresponding to the peak of the IAHP and to a period during the decay of the IAHP, corrected for baseline noise, have been fitted with Lorentzian functions (, see METHODS; fc is corner frequency). D: plot of the time-constant, c [=1/(2fc)] as a function of mean current. The data points were fitted with a linear regression line; the x intercept is equal to mean open dwell time of the channels contributing to the current noise, assuming a two-state kinetic model.

Single-channel recordings of SK channels from cell-attached patches

Based on the presence of ~4,000 AHP channels per AH cell, as estimated from variance-mean analysis, and an AH cell capacitance of ~40 pF (mean determined from capacitance cancellation circuitry on the amplifier, 45 ± 9 pF, n = 24), the density of AHP channels is expected to be ~1 channel/µm² (assuming an even distribution of AHP channels on the cell soma and a specific membrane capacitance of 1 µF/cm²). From this channel density, a typical cell-attached patch that has an area of 5-10 µm² (Kunze et al. 2000) should contain at least one AHP channel. Inspection of our current records in the cell-attached mode revealed that there was an increase in apparent channel activity following action currents, at trans-membrane potentials in the range of 10-30 mV positive of the RMP (Fig. 9A). In most cell-attached patch recordings, however, despite low-pass filtering of such current records with cutoff frequencies of 200-300 Hz, it was difficult to discern unitary current levels in patches where the net current increased by more than ~5 pA. In patches where the increase in patch current was <2-3 pA, it was possible to distinguish unitary current transitions that persisted for a similar period of time as the whole cell recorded AHPs (Fig. 9A). From such recordings (2-4 s in duration), the data points were binned into all-points histograms, which revealed approximately equidistant peaks that corresponded to the unitary levels on the current traces (Fig. 9Bi). All-point histograms constructed from baseline current recordings usually showed a large peak corresponding to the closed channel level and a smaller peak corresponding to one or two channels being open (Fig. 9Bii). The all-points histograms as in Fig. 9Bi were then fitted with multiple Gaussian functions, and from these fits, the probability of channel opening (P_o) and the amplitude of the unitary current (i) were calculated. In Fig. 9C, the mean unitary current level is plotted as a function of pipette potential and shows that the unitary current decreased as the pipette potential was made more negative. To determine the conductance of these channels, the data points were fitted with an equation for a straight line that was multiplied by a voltage-dependent term to account for the apparent inward current rectification (see Fig. 9C). By interpolation, the unitary current was zero when the pipette potential was approximately equal to the cell resting potential (67 mV). The unitary chord conductance at 20 mV was estimated to be 17 pS. In 16 patches, the mean pooled P_o at all patch potentials for these channels over the 2-4 s of recording following action currents was 0.424 ± 0.033; in the absence of action currents, the mean P_o of channels with approximately the same unitary currents, in the same patches, at the same potential, averaged 0.060 ± 0.014. The P_o of these channels following action currents did not differ at a pipette potential (V_p) of 40 mV (P_o = 0.39) from the activity recorded at V_p = 20 mV (P_o = 0.4). These data suggest that the AHP in AH cells is generated by the opening of small conductance voltage-insensitive Ca²⁺-activated K⁺ channels.

View larger version (26K): [in this window] [in a new window]   Fig. 9. Unitary currents through AHP channels recorded in cell-attached mode. A: cell-attached recording from a AH neuron at a pipette potential of 40 mV. The pipette was filled with high-K+ intracellular solution and contained 2 mM TEA and 2 mM Cs+ to block BK channels and If channels, respectively. Following a single action current (inset shows inflection on action current indicative of an AH cell), there is an increase in channel activity in the patch. Channel openings are downward. Single-channel current levels (o1-o4) have been drawn according to the unitary current level derived from the all-points histogram in B. B, i: all-points histogram of the extended 3-s trace shown in A and fitted with the sum of 4 Gaussians whose peaks were separated by ~0.6 pA. Assuming 4 AHP channels are present in the patch, Po was estimated to be 0.24. ii: all-points histogram was constructed from a 3-s current recording at the same potential, preceding the action current in A. The large peak corresponds to the closed channel current level (c) and there are some openings to the first channel level. C: pooled data showing the mean unitary current amplitude (i) plotted as a function of pipette potential (Vp). The data points have been fitted with a regression line multiplied by an exponential term, K = Ko*exp(d*Vp), to account for the rectification. Ko was equal to 0.825 and d was 0.0073. The 0 current was interpolated to occur at Vp = 67 mV.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have shown in the present study that the Ca²⁺-activated K⁺ current that is responsible for the slow AHP in myenteric AH neurons is dependent on release of Ca²⁺ from intracellular Ca²⁺ stores and is generated by the opening of K⁺ channels of relatively small conductance, ~10 pS, that can be classified as SK channels (Vergara et al. 1998). The delay between repolarization of the action potential and the onset of the AHP and the slow rise time of the I_AHP suggest that Ca²⁺ influx does not directly activate these AHP channels, although part of the delay may be due to the superimposition of a poststimulus transient inward current (discussed in the following text). CICR is also a prerequisite for activation of AHP channels in other types of neuron (Berridge 1998; Sah 1996).

Release of Ca²⁺ from caffeine- and ryanodine-sensitive Ca²⁺ stores in AH neurons was mainly triggered by Ca²⁺ entry through N-type voltage-gated Ca²⁺ channels because -conotoxin GVIA reduced the I_AHP by 80-90%. This toxin almost abolished the plateau component of the action potential in AH cells, confirming that the "hump" on the repolarization phase of the AP is generated primarily by the opening of N-type Ca²⁺ channels (Furness et al. 1998). On the other hand, Starodub and Wood (1999) reported that -conotoxin MVIIC, a blocker of P/Q type Ca_V channels blocked the Ca²⁺ current in AH neurons; this suggests that the Ca²⁺ channels in these neurons have an atypical pharmacology.

Our present study indicates that the AHP channels have a relatively small unit conductance (9-15 pS) and there are ~4,000 channels per AH cell. The unitary current of AHP channels showed weak inward rectification that is characteristic of SK-type channels (Kohler et al. 1996). Following a volley of action potentials, the open probability of AHP channels can reach 0.5 while at the peak of a typical whole cell I_AHP the open probability exceeds 0.5. At present, the Ca²⁺ sensitivity of AHP channels in AH neurons is not known. However SK channels in hippocampal neurons, which are capable of generating slow AHPs, are half-maximally activated when the [Ca²⁺] on the cytoplasmic surface reaches 560 nM (Hirschberg et al. 1999). This suggests that if AHP channels in AH neurons have a similar Ca²⁺ sensitivity, then cytoplasmic [Ca²⁺] may exceed 500 nM at the peak of the AHP. However, microelectrode studies on AH myenteric neurons that were loaded with Fura-2 to measure changes in [Ca²⁺] reported that the increase in global cytoplasmic [Ca²⁺] during the AHP was more modest, increasing from ~90 to 120 nM (Tatsumi et al. 1988). Simultaneous recordings of cytoplasmic [Ca²⁺] and AHP in AH neurons have also shown that the decay of the Ca²⁺ transient is faster than the decay of the AHP (Hanani et al. 1997; Tatsumi et al. 1988; Vogalis et al. 2000). This suggests that there may be differences in changes in [Ca²⁺] in the cytoplasm and the submembrane region, which cannot be determined accurately using Ca²⁺-sensitive dyes.

The slow rise time of the AHP current (time constant of ~300 ms in the present study) is considerably slower than the rise time of the Ca²⁺ transient recorded at the soma in AH neurons (Hanani et al. 1997). A similar discrepancy between the fast time-to-peak of the Ca²⁺ transient and the relatively slow-to-peak of the I_AHP is also seen in hippocampal neurons (Sah and Clements 1999). In the latter case, this was taken as evidence that the AHP channels have inherently slow kinetics of activation by Ca²⁺, which is in contrast to the very rapid activation kinetics (milliseconds) of cloned SK channels by Ca²⁺ (Hirschberg et al. 1999). These findings suggest that AHP channels may be different molecular entities from the SK channels isoforms cloned thus far (Vergara et al. 1998) although they have similar small unit conductances. Alternatively, the time course of the Ca²⁺ transient to which the AHP channels are exposed is different from that occurring in the bulk of the cytoplasm (Bond et al. 1999).

The slow rise time of the I_AHP may also reflect the time necessary for Ca²⁺ entry to trigger regenerative Ca²⁺ release. Another possible explanation is that AHP channels at rest are insensitive to Ca²⁺ and require a priming pulse of Ca²⁺ to convert them to an activatable state (Hirst et al. 1985). Given that the gating of cloned SK channels by Ca²⁺ is mediated by constitutively bound calmodulin (CaM) (Xia et al. 1998), the dissociation of Mg²⁺, which is 10,000-fold more concentrated than Ca²⁺ in the cytoplasm at rest, from the Ca²⁺ binding sites on CaM (Malmendal et al. 1998), may also contribute to the delay in the onset of the I_AHP. A further cause for the slow onset of the I_AHP may be the concomitant activation of an opposing post-excitation transient inward current that overlaps with the I_AHP. We have found that such a current can be unmasked after blockade of the I_AHP. Further experiments are required to determine the ionic, voltage, and Ca²⁺ dependence of this current. Although this current was inhibited by high concentrations of niflumic acid, suggesting that it is generated by the opening of Ca²⁺-dependent Cl channels (Hogg et al. 1994; Kenyon and Goff 1998), the involvement of Ca²⁺-activated cation channels, which are also inhibited by niflumic acid, cannot be excluded (Gogelein et al. 1990).

The AHP in myenteric AH neurons is insensitive to apamin and is partially sensitive to charybdotoxin and iberiotoxin (Kunze et al. 1994). This suggests that the channels underlying the AHP are not typical of the SK channels that have been cloned from rat and human brain, which are apamin sensitive but charybdotoxin insensitive (Kohler et al. 1996). Although the mean-variance analysis that we used to estimate the unitary conductance of the AHP channels assumes that they obey a simple two-state gating scheme (i.e., they are either closed or open), the activation of AHP channels by Ca²⁺ may involve multiple closed and open states. Single-channel recordings of SK channels in cultured hippocampal neurons indicate that SK channels have at least two open dwell times (2.8 and 68 ms) and at least three closed dwell times (Selyanko et al. 1998). The shorter open dwell time corresponds with the mean open dwell time estimated from our spectral analysis of the I_AHP in AH neurons (2.8 ms). The Lorentzian functions fitted to the PSD curves indicated that the cutoff frequencies of the power spectrum decreased as the I_AHP decayed. This suggests that the open probability of AHP channels is determined largely by changes in their closed dwell time (Valiante et al. 1997).

Using the number of SK channels that were activated following action currents in cell-attached recordings from intact AH myenteric neurons, we estimated that there are ~4,000 AHP channels on the soma of these neurons. This estimate is based on a patch area of 5 µm² and a cell-surface area of ~4,000 µm² (estimated from a typical AH cell capacitance ~40 pF and a specific membrane capacitance of 1 µF/cm²). This density of AHP channels is similar to that derived from mean-variance analysis of the whole cell I_AHP. In the majority of cell-attached recordings, there was evidence of multiple channel openings (up to ~5 channels) that followed action currents, and these openings persisted for similar periods of time as the duration of typical whole cell AHPs (i.e., seconds). Earlier studies have suggested that the conductance associated with the AHP is partly active at rest (Grafe et al. 1980; North and Tokimasa 1987). Consistent with this, we found evidence for ongoing channel openings in the absence of action currents. If it is assumed that these channels are the same channels that are activated during the AHP, then the resting conductance attributable to these channels is ~2.4 nS, given an open probability of 0.06 and the presence of 4,000 channels per cell with a unitary conductance of 10 pS. This conductance would increase to ~17 nS following an action potential discharge. The magnitude of the estimated resting conductance is similar to the contribution made by the persistent Ca²⁺-sensitive K⁺ current to the resting conductance in AH cells (3 nS) that was reported by North and Tokimasa (1987). The fact that the channels that we recorded in the absence of activity had a similar conductance to those that were activated following action potentials discharges suggests that only one population of channels is involved in both the AHP and the resting Ca²⁺-dependent K⁺ conductance. This agrees with the mutually occlusive relationship between the persistent Ca²⁺-sensitive K⁺ current and the AHP current that was reported in AH neurons previously (North and Tokimasa 1987). Suppression of the opening of these channels may contribute to slow excitatory postsynaptic potentials in AH neurons, and to an increase in R_in that can be mimicked by inhibition of resting Ca²⁺ entry (Grafe et al. 1980). However, the change in R_in caused by complete suppression of the persistent Ca²⁺-sensitive K⁺ current can only partly account for the increase elicited by agonists (North and Tokimasa 1987), suggesting that the slow synaptic excitation in AH neurons may involve suppression of additional K⁺ conductances that are active at rest.

The immediate source of Ca²⁺ that is required to maintain the Ca²⁺-sensitive K⁺ channels active at rest is unclear. It is possible that noninactivating non-N-type Ca_V channels are also active at rest and that Ca²⁺ entering AH cells through these channels has directly activates the persistent Ca²⁺-sensitive K⁺ channels that may be spatially coupled to the Ca_V channels. However, because these Ca²⁺-sensitive K⁺ channels were active in cell-attached patches that were exposed to Cd²⁺ to block co-localized Ca_V channels, their direct activation by Ca²⁺ entry may not be obligatory. Our results suggest that Ca²⁺ entry at rest is diverted to a Ca²⁺ storage compartment from which Ca²⁺ is slowly released in the vicinity of AHP channels, in a manner suggested by North and Tokimasa (1987). The increase in outward current at holding potentials between 50 and 55 mV elicited by ryanodine and thapsigargin is consistent with an ongoing leak of Ca²⁺ from stores to activate AHP channels, as both treatments interfere with sequestration of cytoplasmic [Ca²⁺].

In summary, we have shown that a high proportion of myenteric neurons in intact ganglia in the guinea pig duodenum generate prolonged AHPs following AP firing and that APs have inflections on their falling phases. The AHP in AH neurons is generated by the opening of SK channels in response to Ca⁺ entry through N-type Ca²⁺ channels. Activation of these SK channels is dependent on release of Ca²⁺ from ryanodine and caffeine-sensitive stores. Because the AHP is critically important in regulating the excitability of AH neurons (Wood 1994), which are the intrinsic primary afferent neurons in the enteric nervous system, modulation of these channels may be an important avenue for manipulating the motility of the intestine.