Specificity in the interaction of high-voltage-activated Ca²⁺ channel types with Ca²⁺-dependent afterhyperpolarizations in magnocellular supraoptic neurons

Abstract

Magnocellular oxytocin (OT) and vasopressin (VP) neurons express an afterhyperpolarization (AHP) following spike trains that attenuates firing rate and contributes to burst patterning. This AHP includes contributions from an apamin-sensitive, medium-duration AHP (mAHP) and from an apamin-insensitive, slow-duration AHP (sAHP). These AHPs are Ca²⁺ dependent and activated by Ca²⁺ influx through voltage-gated Ca²⁺ channels. Across central nervous system neurons that generate Ca²⁺-dependent AHPs, the Ca²⁺ channels that couple to the mAHP and sAHP differ greatly, but for magnocellular neurosecretory cells this relationship is unknown. Using simultaneous whole cell recording and Ca²⁺ imaging, we evaluated the effect of specific high-voltage-activated (HVA) Ca²⁺ channel blockers on the mAHP and sAHP. Block of all HVA channels via 400 μM Cd²⁺ inhibited almost the entire AHP. We tested nifedipine, conotoxin GVIA, agatoxin IVA, and SNX-482, specific blockers of L-, N-, P/Q-, and R-type channels, respectively. The N-type channel blocker conotoxin GVIA (1 μM) was the only toxin that inhibited the mAHP in either OT or VP neurons although the effect on VP neurons was weaker by comparison. The sAHP was significantly inhibited by N-type block in OT neurons and by R-type block in VP neurons although neither accounted for the entirety of the sAHP. Thus the mAHP appears to be elicited by Ca²⁺ from mostly N-type channels in both OT and VP neurons, but the contributions of specific Ca²⁺ channel types to the sAHP in each cell type are different. Alternative sources to HVA channels may contribute Ca²⁺ for the sAHP.

NEW & NOTEWORTHY Despite the importance of afterhyperpolarization (AHP) mechanisms for regulating firing behavior of oxytocin (OT) and vasopressin (VP) neurons of supraoptic nucleus, which types of high-voltage-activated Ca²⁺ channels elicit AHPs in these cells was unknown. We found that N-type channels couple to the medium AHP in both cell types. For the slow AHP, N-type channels contribute in OT neurons, whereas R-type contribute in VP neurons. No single Ca²⁺ channel blocker abolished the entire AHP, suggesting that additional Ca²⁺ sources are involved.

INTRODUCTION

High-voltage-activated (HVA) Ca²⁺ channels were classically categorized by their pharmacology and their subunit composition (Birnbaumer et al. 1994). Many neuronal cell types express L-, N-, P/Q-, and R-type HVA Ca²⁺ channels (Castelli and Magistretti 2006; Fisher and Bourque 1995, 1996; Foehring and Armstrong 1996; Lorenzon and Foehring 1995; Mermelstein et al. 1999; Pineda et al. 1998; Quinlan et al. 2008; Simms and Zamponi 2014; Stewart and Foehring 2000; Wang and Fisher 2014). This includes oxytocin (OT) and vasopressin (VP) magnocellular neurosecretory cells (MNCs). These cells control milk ejection and parturition during the reproductive cycle (OT) as well as water balance and vasoconstriction (VP) (Poulain et al. 1977). Depolarization attributable to action potentials (APs) triggers the voltage-dependent activation of these channels, allowing Ca²⁺ ions to flow inward. The existence of different HVA channel subtypes reflects a need to meet the complex Ca²⁺ requirements of neurons and other excitable cells (Snutch et al. 2013). A classic example in neurons is specificity in Ca²⁺ channel coupling to transmitter release in certain neurons (Catterall 2011; Cohen-Kutner et al. 2010; Satake and Imoto 2014; Snutch et al. 2013; Wheeler et al. 1994).

HVA Ca²⁺ channels are also involved in the activation of Ca²⁺-dependent K⁺ conductances underlying spike afterhyperpolarizations (AHPs) following spike trains. AHPs shape firing behavior and cause spike frequency adaptation in many cell types (Andrade et al. 2012; Meech 1978; Pineda et al. 1998; Sah 1996). MNCs demonstrate three distinct AHPs after spiking. The fast AHP (fAHP) follows a single spike and is involved in repolarization of the cell after an AP. We do not address the fAHP here. In MNCs, the medium AHP (mAHP) is a Ca²⁺-dependent conductance activated after one to three spikes, has an inactivation time constant (τ) of ~500 ms, and is blocked by the SK channel toxin apamin (Armstrong et al. 1994; Bourque and Brown 1987; Kirkpatrick and Bourque 1996; Teruyama and Armstrong 2005). The mAHP contributes strongly to spike frequency adaptation in MNCs but also contributes to burst length in VP neurons (Kirkpatrick and Bourque 1996). The slow AHP (sAHP) is also Ca²⁺ dependent but requires longer spike trains, lasts for seconds, is insensitive to apamin, and is markedly attenuated by muscarinic receptor activation (Ghamari-Langroudi and Bourque 2004); the underlying channel is unknown. The sAHP contributes to phasic patterning in VP neurons (Ghamari-Langroudi and Bourque 2004). MNCs fire in phasic bursting patterns, maximizing hormone release and minimizing secretory fatigue (Dutton and Dyball 1979). This phasic bursting is characterized by seconds of repetitive firing followed by seconds of quiescence.

Which Ca²⁺ channels couple to AHPs differs among neuronal cell types and is often different for the mAHP and sAHP within the same neuron. In myenteric neurons and vagal motoneurons, N-type currents are coupled to both mAHPs and sAHPs (Sah 1995; Vogalis et al. 2001). In contrast, L-type channels contribute to the sAHP in CA1, CA3, and sympathetic neurons (Martínez-Pinna et al. 2000; Moyer et al. 1992; Tanabe et al. 1998). An extensive review of the HVA Ca²⁺ channels coupling to sAHPs can be found in Andrade et al. (2012).

Although both OT and VP MNCs exhibit prominent mAHPs and sAHPs (Teruyama and Armstrong 2005), the associated Ca²⁺ channels contributing to their activation are unknown. Previous work in our laboratory has demonstrated a stark mechanistic difference in that both the mAHP and sAHP are dependent on phosphatidylinositol 4,5-bisphosphate (PIP₂) in OT but not VP neurons (Kirchner et al. 2017). It is possible that some of the observed differences could be explained by a difference in coupling of AHPs to Ca²⁺ channels between the cell types. Here we present a direct comparison of the Ca²⁺ channels involved in the generation of AHPs between OT and VP neurons.

METHODS

Slice preparation.

These data were obtained from 56 cells from 30 wild-type Sprague-Dawley rats’ supraoptic neurons (SONs). Coronal brain slices (250 µm) were prepared from randomly cycling, virgin adult Sprague-Dawley female rats (Harlan Laboratories, Indianapolis, IN) weighing between 150 and 220 g. We used randomly cycling females as in our previous studies. This ensured that any phenomena uncovered would not be dependent on estrous state. We recorded from between one and four cells per rat. We restricted these studies to females because of the laboratory’s extensive characterization of the differences between VP and OT neurons in this sex (Kirchner et al. 2017; Stern and Armstrong 1996; Teruyama and Armstrong 2005, 2007). The University of Tennessee Health Science Center institutional animal care and use committee review board approved all experiments. Animals were on an ad libitum diet. For use in experiments, rats were deeply anesthetized with ketamine (100 mg/kg)/xylazine (10%) and perfused through the heart with cold artificial cerebrospinal fluid (aCSF) with NaCl replaced by 210 mM sucrose. The rats were decapitated via guillotine. The brains were then removed and subsequently sliced in ice-cold aCSF with 210 mM sucrose replacing NaCl, using a Leica VT1000S vibrating microtome. After being sliced, the tissue was transferred to an aCSF-filled holding chamber and warmed for 15–20 min at 32°C. aCSF was continuously bubbled with 95% O₂-5% CO₂ and contained (in mM) 20.00 D-glucose, 0.45 ascorbic acid, 2.50 KCl, 1.00 MgSO₄, 1.25 NaH₂PO₄.H₂O, 26.00 NaHCO₃, 125.00 NaCl, and 2.00 CaCl₂. Slices were then transferred to aCSF at room temperature, where they remained for at least 40 min before the recording.

Electrophysiology.

Slices were placed in the well of a Plexiglas chamber attached to a modified stage on an Olympus BX51WI upright microscope and perfused with aCSF containing 5 mM CsCl to block the slow depolarizing afterpotential (Ghamari-Langroudi and Bourque 1998; Teruyama and Armstrong 2005, 2007) and 10 µM 6,7-dinitroquinoxaline-2,3-dione (DNQX; Sigma-Aldrich, St. Louis, MO), 40 µM 2R-amino-5-phosphonovaleric acid (Tocris, Minneapolis, MN), and 100 µM picrotoxin (Tocris) to block fast synaptic currents. The aCSF was bubbled constantly with 95% O₂-5% CO₂, warmed to 32°C ± 1°C, and flowed at ~2 ml/min. For recordings in which we applied Cd²⁺ to the cells, we used phosphate-free aCSF consisting of the following (in mM): 121.0 NaCl, 1.3 MgCl₂, 3.0 KCl, 26.0 NaHCO₃, 20.0 D-glucose, and 2.5 CaCl₂ to prevent Cd²⁺ precipitation. Whole cell and current-clamp recordings were obtained using an Axon Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA). Traces were digitized using an Axon 1440A Digitizer at 10 kHz on a Dell desktop computer running Clampex 9 software (Molecular Devices).

Recording pipettes (4–8 MΩ) were pulled from borosilicate glass (1.5-mm OD) using a P-1000 flaming/brown horizontal micropipette puller (Sutter Instruments, Novato, CA). The pipette internal solution consisted of the following (in mM): 135.0 KMeSO₄, 8.0 NaCl, 10.0 HEPES, 2.0 Mg-ATP, 0.3 Na-GTP, 0.1 leupeptin, 6.0 phosphocreatine, and 0.2 EGTA with pH 7.2–7.4 and 285–295.0 mOsm (kg/H₂O). Biocytin (0.1%) (Sigma-Aldrich) was added on the day of the experiment for visualization during immunochemical identification of cell type (Horikawa and Armstrong 1988). The liquid junction potential for the KMeSO₄ internal was approximately −10 mV and was not corrected.

AHPs measured in current clamp were evoked from a resting membrane potential of −55 mV by injecting 300 pA of current via a 20-spike, 20-Hz pulse train. AHP amplitudes correlate with spike count, so, to isolate the mAHP, we injected a shorter train of 5 spikes of the same amplitude and frequency (Ghamari-Langroudi and Bourque 2004). For Ca²⁺ channels, reagents included 5.0 µM nifedipine (Nif; Sigma-Aldrich) to block L-type channels, 1.0 µM ω-conotoxin GVIA (CnTx GVIA; Peptides International, Louisville, KY or Alomone Laboratories, Jerusalem, Israel) to block N-type channels, 500.0 nM agatoxin-IVA (AgTx IVA; Alomone Laboratories) to block P/Q-type channels, and 0.3 μM SNX-482 (Alomone Laboratories) to block R-type channels. Because SNX-482 also blocks A-type K⁺ currents (Liu and Bean 2014), 4 mM 4-aminopyridine (4-AP; Sigma-Aldrich) was used to block A-current (I_A) and other voltage-gated K⁺ channels contributing to spike width before administering SNX-482. Cells whose series resistance exceeded 20 MΩ and/or changed by >20% during the recording were discarded. Traces were averaged over two or more runs for each individual cell.

Immunochemistry.

Slices were fixed in 4.0% paraformaldehyde and 0.2% picric acid in PBS and stored at 4°C. Biocytin-labeled SONs were processed for double labeling with either anti-OT or anti-VP neurophysins. The anti-VP neurophysin is a rabbit polyclonal antibody provided by Alan Robinson (UCLA, Professor Emeritus) and was used at 1:20,000. The anti-OT neurophysin antibody (PS36) is a mouse monoclonal antibody provided by Harold Gainer (National Institutes of Health, Emeritus) and was used at 1:500. All antibodies and labeling reagents were dissolved in PBS + 0.5% Triton X-100 (PBST). After 36–72 h of incubation at 4°C, the slices were washed with PBST and incubated in a cocktail of secondary antibodies including Alexa Fluor 488 goat anti-rabbit immunoglobulin G (IgG) and Alexa Fluor 594 goat anti-mouse IgG (1:200) along with Avidin-AMCA (1:200) for reaction with the biocytin. Double staining of biocytin and one antibody complemented the negative staining of the other identified neurons as OT or VP.

Calcium imaging.

Ca²⁺ imaging was performed alongside patch-clamp electrophysiology on a single computer using a Polychrome V monochromater (TILL Photonics, Planegg, Germany) or on two computers using a Zeiss LSM 7MP 2 photon system (Carl Zeiss, Oberkochen, Germany). Regardless of imaging hardware, the pipette internal solution for analyzing AHP tail currents was as previously described, except that 0.2 mM EGTA was replaced with 0.1 mM fura-2. Measurements were made from the soma, excluding the nucleus. We excited fura-2 at 380 nm and report all data as %ΔF/F (Callaway et al. 1993; Teruyama and Armstrong 2005). Fura-2 is a highly sensitive indicator that has little toxicity, unlike other fluorescein-based indicators. Although fura-2 allows rationing to determine [Ca²⁺]_i, single-wavelength imaging allows greater temporal resolution, and the changes we measure represent proportional changes in [Ca²⁺]_i as we start from a hyperpolarized state (resting) and allow time for [Ca²⁺]_i to recover between trials (Roper et al. 2003). These data were analyzed either with the custom acquisition program described below or with Igor Pro 7.0 (Wavemetrics, Portland, OR).

In the case of the Polychrome V, Ca²⁺ transients were acquired using a custom Windows-based program (CCD32; written by Dr. J. Callaway, UTHSC, Memphis, TN) based on software developed by Lasser-Ross et al. (1991). Images were obtained with a CCD Imago Sensicam camera at 380 nm using a USHIO UXL-150MO 150 W Xenon arc lamp. We measured fluorescence changes at an emission wavelength of 520 ± 40 nm. Photobleaching was corrected by subtracting a Ca²⁺ signal from a control sweep of equal length, in which the cell was not stimulated and held at a hyperpolarized holding potential (−70 mV) to minimize Ca²⁺ entry. We subtracted background autofluorescence by using a reference area near the cell.

When we used the two-photon system, Ca²⁺ transients were acquired with Windows-based software, Zeiss Zen 2010. Cells were imaged on a Zeiss LSM 7MP 2 photon system equipped with a Coherent Chameleon Vision-S laser (Titanium-Sapphire). We excited fura-2 at 800 nm. Cells were imaged using a raster scan at 30–40 Hz depending on the region of interest size. This equipment was used to supplement data obtained on the polychrome V while the latter was under repair.

Statistics and analysis.

In MNCs, the fAHP is a brief event lasting <15 ms (Dopico et al. 1999) and was not evaluated here. The mAHP decay τ is ~500 ms in MNCs (Teruyama and Armstrong 2005), whereas the sAHP has a decay τ of 1–2 s (Ghamari-Langroudi and Bourque 2004). With these considerations, we operationally defined the mAHP and sAHP in a 20-pulse current-clamp protocol as the amplitude of the AHP at peak (~100 ms after train) and 1,000 ms after the stimulus (sAHP), respectively. Although measurement of the AHP at the peak likely contains contributions from the sAHP, it is dominated by the mAHP because of the slower-onset kinetics of the sAHP (Ghamari-Langroudi and Bourque 2004; Kirchner et al. 2017; Teruyama and Armstrong 2007). To further evaluate the mAHP, we isolated the peak hyperpolarization after a shorter (5 pulses) train in current clamp that activates the mAHP without the sAHP. We operationally define the sAHP as the current amplitude 1 s after the pulse, a time point when the mAHP has mostly inactivated (Teruyama and Armstrong 2005).

All electrophysiological traces were analyzed in ClampFit 10.2 (Molecular Devices) or Igor Pro (Wavemetrics). Measurements of amplitude were averaged over a 30-ms segment of current (30 points). All statistics were performed in PRISM or SPSS. Data were evaluated using a two-way repeated-measures ANOVA to evaluate the within-sample effects of channel blockers on the cells and between-sample effects of OT vs. VP neurons. We used a Sidak multiple-comparisons test to evaluate toxin effects on specific cell types (OT or VP). The exception to this is the 20-spike Ca²⁺ data for the SNX-482 experiments, which did not meet the criteria for normality (Fig. 6). Therefore, we used a Friedman’s test for this data. We used a Bonferroni-Dunn post hoc analysis to make comparisons in the Friedman’s test. We used a repeated-measures ANOVA with a Bonferroni post hoc analysis to evaluate AP half-width (Fig. 5C), as well as a one-way ANOVA to evaluate differences in spike broadening during trains (Fig. 5D). APs are truncated in the figures. Data in the figures are marked as statistically significant depending on the number of asterisks associated with a difference between groups, *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 6.

Effect of R-type blocker SNX-482 (0.3 μM) on afterhyperpolarizations (AHPs) and their corresponding Ca²⁺ transients in oxytocin (OT) (n = 5) (A–D) and vasopressin (VP) (n = 5) (E–H) neurons. 4-Aminopyridine (4-AP) was preapplied to block Kv4 channels and isolate the effect of SNX482 on R-type Ca²⁺ channels. A: example AHP after a 20-Hz, 20-spike train from an OT neuron and its corresponding somatic Ca²⁺ signal. B: example AHP after a 20-Hz, 5-spike train from an OT neuron and its corresponding somatic Ca²⁺ signal. C: summary data for OT neuron AHP measurements at the 20-spike peak amplitude (medium and slow AHP, mAHP + sAHP), at 1 s after the (sAHP), and 5-spike AHPs at the peak amplitude (mAHP). Red points are individual cells, and black points are group averages. After 20 spikes, 4-AP significantly increased the AHP at peak (*P < 0.05) and 1 s (****P < 0.0001) in 20-spike AHPs. Subsequent application of SNX-482 did not significantly reduce the AHP compared with 4-AP. 5-spike AHPs were unaffected. D: summary data for OT neuron peak Ca²⁺ transients during AHP stimulation in 20-spike AHPs and in 5-spike AHPs. 4-AP increased Ca²⁺ transients significantly in 20-spike and 5-spike trains (*P < 0.05), and successive SNX-482 significantly inhibited the Ca²⁺ signal compared with 4-AP after both 20 and 5 spikes (*P < 0.05). E: example AHP after a 20-spike train from a VP neuron and its corresponding somatic Ca²⁺ signal. F: example AHP after a 5-spike train from a VP neuron and its corresponding somatic Ca²⁺ signal. G: summary data for VP neuron AHP measurements at the 20-spike peak amplitude (mAHP + sAHP), at 1 s after the train (sAHP), and 5-spike AHPs at the peak amplitude (mAHP). Green points are individual cells, and black points are group averages. 4-AP significantly increased the 20-spike AHP at 1 s (***P < 0.001). Subsequent SNX-482 significantly reduced the 20-spike peak and 1-s AHP compared with 4-AP alone (**P < 0.01). 5-spike AHPs were unaffected. H: summary data for VP neuron peak Ca²⁺ transients in 20-spike AHPs and 5-spike AHPs. 4-AP significantly increased Ca²⁺ in the 20-spike (*P < 0.05) and 5-spike (**P < 0.01) protocol. Subsequent SNX-482 significantly decreased Ca²⁺ transients in both 20- (**P < 0.01) and 5-spike AHPs (***P < 0.001). All data were evaluated with a 2-way repeated-measures ANOVA with the exception of the 20-spike Ca²⁺ imaging data. For this data, we used a Friedman’s test because this data did not present a normal distribution.

Fig. 5.

SNX-482 blocks transient outward rectification and broadens spikes [oxytocin (OT) and vasopressin (VP) n = 5]. A, i: example single action potential (AP) from an OT neuron treated with SNX-482 (0.3 μM) showing obvious broadening of the AP. A, ii: example single AP from a different OT neuron than Ai, treated first with 4 mM 4-aminopyridine (4-AP), followed by subsequent application of SNX-482. B: step injections of depolarizing current reveal a fast transient outward rectifier that is reduced by SNX-482 in the same cell as A, i. Positive steps (25 pA) were generated from cells at −80 mV. C: in 20-spike trains, we measured AP half-width of the 1st spike. In OT and VP neurons, 4-AP results in significant spike broadening (OT, *P < 0.05; VP, **P < 0.01). Subsequent application of SNX-482 produced no further broadening (P > 0.05) (repeated-measures ANOVA). D: changes in spike broadening between the 1st and 20th spike in a train represented as the ratio of AP half-width between the 2 spikes. Values >1 indicate spike-dependent broadening, whereas values <1 indicate narrowing. Application of 4-AP significantly reduced spike-dependent broadening within the 20-spike train in both cell types (*P < 0.05). Subsequent application of SNX-482 does not significantly reduce the spike-broadening index compared with 4-AP alone (1-way ANOVA).

RESULTS

To determine which specific Ca²⁺ channel subtypes contributed Ca²⁺ that elicited the mAHP and sAHP in SONs, we measured AHPs in current clamp before and after the application of pharmacological channel blockers. We used a train of 5 or 20 spikes stimulated at 20 Hz from a resting potential of −55 mV to control for effects on spike frequency adaptation and ensure all neurons received the same stimuli. We simultaneously monitored changes in bulk somatic Ca²⁺ ([Ca²⁺]_i) during stimulation. This measurement reflects Ca²⁺ entry plus extrusion and buffering. Measuring this allowed us to confirm the efficacy of toxin block on Ca²⁺ entry through its respective Ca²⁺ channel.

Cd²⁺ block of HVA Ca²⁺ channels inhibits most of the AHP.

We first used the inorganic Ca²⁺ channel blocker Cd²⁺ to test whether activation of the AHP in both MNC cell types required Ca²⁺ entry. Cd²⁺ (400 µM) blocked 80 ± 18% of the mAHP and 81 ± 7% of the sAHP in OT neurons (n = 6) (Fig. 1). In VP neurons, it blocked 79 ± 17% of the mAHP and 89 ± 2% of the sAHP (n = 6). These data are consistent with previous reports that inorganic blockers such as Cd²⁺, Mn²⁺, and Co²⁺ block most of the AHP in MNCs (Bourque et al. 1985; Ghamari-Langroudi and Bourque 2004; Kirkpatrick and Bourque 1996). Even stronger block can be achieved by chelating intracellular Ca²⁺, highlighting that activation of AHPs is dependent on elevated intracellular Ca²⁺ (Andrew and Dudek 1984).

Fig. 1.

400 µM Cd²⁺ application to magnocellular neurosecretory cells inhibited most of the afterhyperpolarization (AHP) in current clamp. AHPs were evoked using a pulse train of 20 spikes and measured at peak (medium and slow AHP, mAHP + sAHP) and 1 s (sAHP) following the pulse. A: example of an oxytocin (OT) neuron before and after Cd²⁺ application of an AHP in current clamp. The only remaining portion of the AHP is small (B and C). Summary data for Cd²⁺ inhibition at the peak current and 1 s following the pulse train in both OT (n = 6) (B) and vasopressin (VP) (n = 6) (C) neurons are shown. Cd²⁺ significantly inhibits AHPs in OT neurons at peak (***P < 0.001) and 1,000 ms (**P < 0.01). Cd²⁺ significantly inhibits AHPs in VP neurons at peak (***P < 0.001) and 1,000 ms (*P < 0.05). All data were evaluated with a 2-way repeated-measure ANOVA.

L-type channels.

Application of the dihydropyridine L-type channel blocker Nif (5 µM) did not alter either the mAHP or sAHP in OT (Fig. 2, A–C; n = 6, P > 0.05) and VP (Fig. 2, E–G; n = 6, P > 0.05) neurons. However, Nif reduced the peak [Ca²⁺]_i, suggesting reduced Ca²⁺ entry. In OT neurons, we observed a 40 ± 1% reduction in the peak Ca²⁺ signal after 20 spikes (Fig. 2D; n = 5, P < 0.01) and a 34 ± 4% reduction after 5 spikes (Fig. 2D; n = 5, P < 0.05). In VP neurons, we observed a 36 ± 2% reduction in the peak Ca²⁺ signal after 20 spikes (Fig. 2H; n = 5, P < 0.01) and a 43 ± 1% reduction after 5 spikes (Fig. 2H; n = 5, P < 0.05). The reduction in Ca²⁺ signal is consistent with previous observations in which 5–10 µM Nif blocked a similar proportion of whole cell Ca²⁺ currents in acutely dissociated SONs (Fisher and Bourque 1995; Foehring and Armstrong 1996).

Fig. 2.

Effect of L-type blocker 5 µM nifedipine (Nif) on afterhyperpolarizations (AHPs) and corresponding Ca²⁺ transients in oxytocin (OT) (n = 5) (A–D) and vasopressin (VP) (n = 5) (E–H) neurons. A: example AHP after a 20-Hz, 20-spike train from an OT neuron treated with Nif and the corresponding somatic Ca²⁺ signal. B: example AHP after a 20-Hz, 5-spike train from an OT neuron treated with Nif and its corresponding somatic Ca²⁺ signal. C: summary data for OT neuron AHP measurements at the 20-spike peak amplitude (medium and slow AHP, mAHP + sAHP), at 1 s after the train (sAHP), and 5-spike AHPs at the peak amplitude (mAHP). Red points are individual cells, and black points are group averages (paired t-test; P > 0.05 for all 3 measurements). D: summary data for OT neuron Ca²⁺ transients. Nif significantly reduced peak %ΔF/F in 20-spike AHPs (**P < 0.01) and 5-spike AHPs (*P < 0.05). E: example AHP after a 20-Hz, 20-spike train from a VP neuron treated with Nif and its corresponding somatic Ca²⁺ signal. F: example AHP after a 20-Hz, 5-spike train from a VP neuron treated with Nif and its corresponding somatic Ca²⁺ signal. G: summary data for VP neuron AHP measurements at the 20-spike peak amplitude (mAHP + sAHP), at 1 s after the train (sAHP), and 5-spike AHPs at the peak amplitude (mAHP). Green points are individual cells, and black points are group averages (paired t-test; P > 0.05 for all 3 measurements). H: Nif significantly reduced peak %ΔF/F from 20-spike trains (**P < 0.01) and 5-spike trains (*P < 0.05). All data were evaluated with a 2-way repeated-measures ANOVA.

N-type channels.

We next tested for effects of the N-type channel blocker GVIA (1 µM). In OT neurons, GVIA strongly reduced AHP amplitude (Fig. 3, A–D; n = 8). GVIA inhibited the 20-spike peak AHP by 39 ± 7% (P < 0.0001), at 1 s after the pulse by 38 ± 7% (P < 0.0001), and the 5-spike peak AHP by 61 ± 11% (P < 0.0001). GVIA inhibited the corresponding peak Ca²⁺ signals by 61 ± 4% for 20 spikes (P < 0.0001) and 69 ± 5% for 5 spikes (P < 0.0001).

Fig. 3.

Effect of N-type blocker 1 µM conotoxin GVIA (CnTx GVIA) on afterhyperpolarizations (AHPs) and corresponding Ca²⁺ transients in oxytocin (OT) (n = 8) (A–D) and vasopressin (VP) (n = 7) (E–H) neurons. A: example AHP after a 20-Hz, 20-spike train from an OT neuron treated with CnTx GVIA and its corresponding somatic Ca²⁺ signal. B: example AHP after a 20-Hz, 5-spike train from an OT neuron treated with CnTx GVIA and its corresponding somatic Ca²⁺ signal. C: summary data for OT neuron AHP measurements at the 20-spike peak (medium and slow AHP, mAHP + sAHP; ****P < 0.0001), at 1 s after the train (sAHP; ****P < 0.0001), and 5-spike AHPs at the peak amplitude (mAHP; ****P < 0.0001). Red points are individual cells, and black points are group averages. D: summary data for OT neuron Ca²⁺ transients during AHP stimulation. CnTx GVIA significantly reduced peak %ΔF/F in 20-spike AHPs (****P < 0.0001) and reduced %ΔF/F in 5-spike AHPs (****P < 0.0001). E: example AHP after a 20-Hz, 20-spike train from a VP neuron treated with CnTx GVIA and its corresponding somatic Ca²⁺ signal. F: example AHP after a 20-Hz, 5-spike train from a VP neuron treated with CnTx GVIA and its corresponding somatic Ca²⁺ signal. G: summary data for VP neuron AHP measurements at the 20-spike peak amplitude (mAHP + sAHP; **P < 0.01), at 1 s after the train (sAHP; P > 0.05), and 5-spike AHPs at the peak amplitude (mAHP; ***P < 0.001). Green points are individual cells, and black points are group averages. H: CnTx GVIA significantly reduced peak %ΔF/F in 20-spike AHPs (paired t-test; ****P < 0.0001) and 5-spike AHPs (****P < 0.0001). All data were evaluated with a 2-way repeated-measures ANOVA.

In VP neurons (n = 7), GVIA inhibited the 20-spike peak AHP by 25 ± 9% (P < 0.01), at 1 s after the pulse by 29 ± 5% (P > 0.05), and the 5-spike peak AHP by 25 ± 12% (P < 0.001). GVIA strongly inhibited the corresponding peak Ca²⁺ signals by 60 ± 4% for 20 spikes (P < 0.0001) and 67 ± 6% for 5 spikes (P < 0.0001) in VP neurons.

Furthermore, there was a significant interaction between cell type and GVIA effect for AHP measurements at 20-spike peaks (P < 0.001), 20-spike 1-s (P < 0.01), and 5-spike peaks (P < 0.05). There was no interaction for [Ca²⁺]_i peaks by GVIA between OT and VP neurons (P > 0.05). Thus, although N-type channels were found to be a major contributor to the AHP in both cell types, the N-type contribution was stronger for OT neuron AHPs. This happens despite no differences in decrease in [Ca²⁺]_i by GVIA between OT and VP neurons.

P/Q-type channels.

To evaluate P/Q-type channel contribution, we used AgTx IVA (0.5 µM). P- and Q-type variants are often distinguished by their sensitivity to AgTx IVA (Mintz et al. 1992); lower concentrations of AgTx IVA (10–100 nM) have been shown to block only P-type variants, whereas 0.5 µM blocks both P- and Q-type (Foehring and Armstrong 1996; Zhang et al. 1993). Both P- and Q-type currents are generated by the CACNA1A gene, and splice variants may explain the phenotypic differences (Bourinet et al. 1999; Nimmrich and Gross 2012). We evaluated AgTx IVA at 0.5 µM to determine contribution by P/Q channels. In OT (n = 6) and VP (n = 5) neurons, application of 0.5 µM AgTx IVA had no significant effect on either the mAHP or sAHP (Fig. 4; P > 0.05), suggesting that P/Q-type currents do not couple to the AHPs. Simultaneous Ca²⁺ imaging revealed a decrease in peak somatic Ca²⁺. In OT neurons, AgTx IVA inhibited the 20-spike Ca²⁺ peak by 16 ± 6% (P < 0.05) and the 5-spike Ca²⁺ peak by 25 ± 6% (P > 0.05) (Fig. 4D). In VP neurons, AgTx IVA significantly inhibited the 20-spike Ca²⁺ peak by 33 ± 5% (P < 0.05). The 5-spike Ca²⁺ peak decreased by 20 ± 5%, but this was not significant (P > 0.05) (Fig. 4H). Given the lack of effect at 0.5µM, we did not test lower concentrations of AgTx IVA to sort P from Q channel contributions.

Fig. 4.

Effect of P/Q-type blocker 0.5 µM agatoxin-IVA (AgTx IVA) on afterhyperpolarizations (AHPs) and corresponding Ca²⁺ transients in oxytocin (OT) (n = 6) (A–D) and vasopressin (VP) (n = 5) (E–H) neurons. A: example AHP after a 20-Hz, 20-spike train from an OT neuron treated with AgTx IVA and its corresponding somatic Ca²⁺ signal. B: example AHP after a 20-Hz, 5-spike train from an OT neuron treated with AgTx IVA and its corresponding somatic Ca²⁺ signal. C: summary data for OT neuron AHP measurements at the 20-spike peak amplitude (medium and slow AHP, mAHP + sAHP), at 1 s after the train (sAHP), and 5-spike AHPs at the peak amplitude (mAHP). Red points are individual cells, and black points are group averages (paired t-test; P > 0.05 for all 3 measurements). D: summary data for OT neuron Ca²⁺ transients. AgTx IVA significantly reduced peak %ΔF/F after 20-spike AHPs (*P < 0.05) but not 5-spike AHPs (P > 0.05). E: example AHP after a 20-Hz, 20-spike train from a VP neuron treated with AgTx IVA and its corresponding somatic Ca²⁺ signal. F: example AHP after a 20-Hz, 5-spike train from a VP neuron treated with AgTx IVA and its corresponding somatic Ca²⁺ signal. G: summary data for VP neuron AHP measurements at the 20-spike peak amplitude (mAHP + sAHP), at 1 s after the train (sAHP), and 5-spike AHPs at the peak amplitude (mAHP). Green points are individual cells, and black points are group averages (paired t-test; P > 0.05 for all 3 measurements). H: AgTx IVA significantly reduced peak %ΔF/F in 20-spike AHPs (*P < 0.05) but not 5-spike AHPs (P > 0.05). All data were evaluated with a 2-way repeated-measures ANOVA.

R-type channels.

We also tested the R-type blocker SNX-482 (0.3 µM). Although a potent blocker of R-type channels, SNX-482 also blocks I_A with high affinity (1–10 nM) (Liu and Bean 2014). Block of I_A increases AP width in MNCs, increasing Ca²⁺ influx during spike trains, and consequently altering the AHP amplitude and time course (Bourque 1988; Hlubek and Cobbett 2000). We initially observed that SNX-482 greatly broadened single spikes (Fig. 5A, i). This effect mimicked that of 4-AP (4 mM), which blocks I_A, as well as other Kv channels. After prior 4-AP (4 mM) application, SNX-482 did not widen spikes further (Fig. 5A, ii). Second, we observed a decrease in the transient outward rectifier (Fig. 5B) after SNX-482, which has previously been shown to be blocked by 4-AP application and attributable to I_A in MNCs (Bourque 1988; Fisher and Bourque 1998; Stern and Armstrong 1997). With these considerations, we quantified the spike broadening of spikes by measuring the half-width of the first AP of each 20-spike train (OT n = 5; VP n = 5; Fig. 5C). Neurons dosed with 4-AP (4 mM) showed significantly larger AP half-widths in both OT (P < 0.05) and VP (P < 0.01) neurons. Subsequent SNX-482 did not further increase spike width (OT P > 0.05; VP P > 0.05). These data demonstrate that 4-AP blocks almost all of I_A and that SNX-482 effects on AHPs cannot be attributed to changes in spike width. Additionally, 4-AP application increased spike widths more in VP neurons, consistent with previous reports of VP neurons having a larger I_A (Fisher and Bourque 1998; Stern and Armstrong 1997).

Finally, we evaluated spike broadening from the first to the 20th spike during the train. Repetitive activity produces spike frequency-dependent broadening (Andrew and Dudek 1985; Bourque and Renaud 1985; Hlubek and Cobbett 2000), which has been shown to increase Ca²⁺ influx, further complicating SNX-482 effects (Jackson et al. 1991). We measured the ratio of the 20th and 1st spike in each train as an index of frequency-dependent broadening (Fig. 5D). In both OT and VP neurons, spike broadening we observed during the train was significantly inhibited by 4-AP (OT and VP P < 0.05). In 4-AP, the spike-broadening index was 1.00 ± 0.10 for OT and 0.96 ± 0.10 for VP, indicating that there was almost no change in AP half-width during trains in 4-AP. This suggests that a majority of the spike broadening in MNCs is the result of I_A inactivation. Further application of SNX-482 to block R-type channels did not significantly alter the spike-broadening index (P > 0.05). Because changes in AHP amplitude could reflect both changes in spike broadening and consequent Ca²⁺ entry as well as reduction of Ca²⁺ channels, we tested the R-type channel effects of SNX-482 on the AHP after preapplied 4-AP.

In AHP measurements from both OT and VP neurons, 4-AP enhanced 20-spike AHP size, amplitude, spike width, and the corresponding Ca²⁺ transients (Figs. 5 and 6), suggesting that the broader spikes increased Ca²⁺ entry. In OT neurons, the AHP enhancement was significant for 20 spikes at peak (20 ± 10% increase; P < 0.05) and 1 s (71 ± 11% increase; P < 0.0001) (Fig. 6C). In VP neurons, the measurement at 1 s after 20 spikes was significant (60 ± 12% increase; P < 0.001; Fig. 6G), whereas the peak measurement was not significant (16 ± 26% increase; P > 0.05; Fig. 6G). The 5-spike AHP was unaffected by 4-AP in either cell type (P > 0.05; Fig. 6).

The corresponding OT Ca²⁺ transients were significantly increased by 4-AP application in 20-spike AHPs by 15 ± 5% (control vs. 4-AP, P < 0.05; Fig. 6D) and 5-spike AHPs by 31+ 8% (control vs. 4-AP, P < 0.05; Fig. 6D). Likewise, the corresponding VP Ca²⁺ transients in 20- and 5-spike peaks significantly increased after 4-AP by 21 ± 6% and 38 ± 2%, respectively (20 spikes: control vs. 4-AP, P < 0.05; 5 spikes: control vs. 4-AP, P < 0.01; Fig. 6H). These data were also consistent with the notion that broader spikes after 4-AP increased Ca²⁺ entry.

Whereas OT neurons demonstrated no significant inhibition of the AHP after subsequent application of SNX-482, the response of VP neurons was different (Fig. 6, C and D). Here SNX-482 reduced 20-spike AHPs at peak by 15 ± 8% and 1 s by 26 ± 11% after 4-AP application (20-spike peak: 4-AP vs. 4-AP + SNX, P > 0.05; 20-spike 1 s: 4-AP vs. 4-AP + SNX, P < 0.01; Fig. 6G). This was despite the further increase in AP width in SNX-482 after 4-AP (see above). SNX-482 did not significantly affect 5-spike peaks (P > 0.05; Fig. 6).

These differential responses may be due to different Ca²⁺ channel contribution to Ca²⁺ entry. Application of SNX-482 in OT neurons inhibited the 20-spike Ca²⁺ peak by only 15 ± 5% (4-AP vs. 4-AP + SNX, P < 0.05; Fig. 6D) and the 5-spike Ca²⁺ peak by only 22 ± 10% (4-AP vs. 4-AP + SNX, P < 0.05; Fig. 6D). In contrast, in VP neurons, SNX-482 inhibited the 20-spike Ca²⁺ peak by 45 ± 6% (4-AP vs. 4-AP + SNX, P < 0.01; Fig. 6H) and the 5-spike Ca²⁺ peak by 45 ± 5% (4-AP vs. 4-AP + SNX, P < 0.001; Fig. 6H).

DISCUSSION

We tested which HVA Ca²⁺ channels elicit the medium and slow phases of AHPs in OT- and VP-releasing neurons of the SON. Our laboratory has previously observed distinct mechanistic differences in AHP generation between OT and VP neurons (Kirchner et al. 2017), spurring interest in the possibility of cell-type differences in coupling of Ca²⁺ channel types to AHPs. The present study demonstrates that Ca²⁺ channel contributions differ between OT and VP cells.

We observed no significant differences in baseline AHP size or [Ca²⁺]_i signals between OT and VP neurons in any of the experiments in which a Ca²⁺ channel blocker was evaluated. Furthermore, complete block of HVA Ca²⁺ channels with Cd²⁺ drastically inhibits the AHP in both cell types, as has been shown previously. The small remaining AHP after Cd²⁺ may reflect a Ca²⁺-independent K⁺ current, such as voltage-gated K⁺ current or the Na⁺-activated K⁺ current recently reported (Bansal and Fisher 2016).

The major contribution of N-type channels to the AHP.

In both cell types studied, N-type channels appear to be the most influential of the HVA types for Ca²⁺-dependent AHP generation. OT neurons show especially strong reduction in the mAHP and sAHP after GVIA application. The coupling to the mAHP was the strongest, but there is also a major contribution to the sAHP. Thus N-type channels provide the primary source of Ca²⁺ for both the mAHP and sAHP in OT neurons. It is important to note that some sAHP remains in OT neurons after N-type channel block, suggesting that other sources of Ca²⁺ exist for OT sAHPs as well. Block of the other HVA Ca²⁺ channel types had no effect on the AHP in these cells, however.

N-type channels also couple to mAHPs in VP neurons; GVIA significantly reduces the mAHP. However, the inhibition by GVIA is significantly less in VP neurons compared with OT despite similar inhibition of Ca²⁺ transients between the cell types. Like OT neurons, GVIA is the only HVA channel toxin that we tested that affected VP mAHPs. These results suggest that N-type Ca²⁺ entry is a primary source of Ca²⁺ for mAHPs in both OT and VP neurons.

Corresponding [Ca²⁺]_i signals in MNCs.

Regardless of whether or not an HVA Ca²⁺ channel blocker was able to affect the AHP, we consistently observed a significant reduction in cytoplasmic [Ca²⁺]_i after each application, indicating that the toxins were effectively blocking the channels. We acknowledge that summation of toxin effects on [Ca²⁺]_i from all experiments exceeds 100%. There are a few reasons for this. First, although these toxins are reasonably specific (Pringos et al. 2011), the most parsimonious explanation based on our previous studies of calcium currents in dissociated cells with good space clamp is that these agents are not perfectly selective (Fisher and Bourque 1995; Lorenzon and Foehring 1995). This is consistent with other studies showing that, although Ca²⁺ channel toxins are highly selective, there is some overlap in targets (Foehring and Armstrong 1996; Gandía et al. 1995). Second, we compared populations of cells with variation in expression levels of HVA channel types among cells. Third, because preapplication of 4-AP increases [Ca²⁺]_i, percent reduction by SNX-482 will be skewed because Ca²⁺ inhibition is not being compared with the cell at the control peak [Ca²⁺]_i value. Finally, it is possible that toxin application lessens the dendritic depolarization caused by the evoked APs, thereby diminishing activation of Ca²⁺ channels not targeted by the toxin. However, it is highly unlikely that proximal dendritic Ca²⁺ itself diffuses far along the dendrite enough to influence our somatic measurements (Wilson and Callaway 2000).

The reduction in [Ca²⁺]_i after N-type block is the greatest by any of the blockers in both cell types, consistent with MNCs containing a large population of N-type channels, as reported previously (Fisher and Bourque 1996; Foehring and Armstrong 1996; Lemos et al. 2012).

As discussed earlier, SNX-482 blocks I_A in addition to R-type channels, resulting in larger AP half-widths. This AP broadening resulted in increased in [Ca²⁺]_i and posed a potential confound for our evaluation of SNX-482 alone on the Ca²⁺-dependent AHP. After 4-AP block of I_A, we observed a significant reduction in [Ca²⁺]_i after SNX-482 despite no change in the AP half-width or spike-broadening index (Fig. 5), demonstrating its ability to block R-type channels independent of changes in spike width.

Strict coupling of mAHPs to Ca²⁺ microdomains.

mAHPs in other cell types are often exclusively coupled to a specific HVA Ca²⁺ channel (Sah 1995; Vogalis et al. 2001). This strict coupling is consistent with SK channel activation by microdomains of Ca²⁺, suggesting that Ca²⁺ channels must be in relatively close proximity to the SK channels and that bulk cytoplasmic Ca²⁺ would be a poor indicator of [Ca²⁺]_i at the membrane (Fakler and Adelman 2008; Hallworth et al. 2003; Neher 1998). In neocortical pyramidal neurons, the relationship between the mAHP and bulk Ca²⁺ is poor (Abel et al. 2004), consistent with its activation from microdomains concentrated near the membrane. Our results are also consistent with this idea, as GVIA is the only toxin to inhibit mAHPs in OT and VP neurons, despite the ability of the other toxins to reduce [Ca²⁺]_i.

Coupling of Ca²⁺ to the sAHP in VP neurons.

Cd²⁺ effectively blocks both mAHPs and sAHPs in both cell types. However, of subtype-specific HVA blockers, only block of N-type channels resulted in a decisive reduction in OT mAHPs or sAHPs. N-type channels are coupled to mAHPs in VP neurons as well, but only after R-type block (Fig. 6) was the sAHP inhibited in VP neurons. Although N-type (OT) or R-type (VP) channels may be the preferred Ca²⁺ source for the sAHP, it is possible that block of the preferred Ca²⁺ channels could reveal other Ca²⁺ channels with looser coupling to the sAHP. This could be facilitated by the slow kinetics of the sAHP. In other cell types, there is lower specificity in the coupling of Ca²⁺ channels to the sAHP compared with the mAHP (Ghamari-Langroudi and Bourque 2004; Lorenzon and Foehring 1992). This may also explain why Cd²⁺ is such a potent blocker of the AHP compared with GVIA or SNX-482.

It has been proposed that there is an intermediate step between Ca²⁺ entry and activation of the sAHP [i.e., Ca²⁺ does not directly bind to the K⁺ channel, reviewed in Andrade et al. (2012)]. In this model, cytoplasmic Ca²⁺ binding proteins act as the calcium sensor for the sAHP. Evidence consistent with a cytoplasmic Ca²⁺ sensor for the sAHP includes experiments in which pharmacological and genetic downregulation of cytoplasmic calcium-binding proteins (e.g., visinin-like calcium binding and calcineurin) resulted in substantial sAHP inhibition of myenteric and hippocampal neurons (Tzingounis et al. 2007; Villalobos and Andrade 2010; Vogalis et al. 2004). Besides HVA Ca²⁺ channels, another possible source for Ca²⁺ is from intracellular stores, which contribute to sAHPs in hippocampal neurons (Torres et al. 1996; van de Vrede et al. 2007) and immature neocortical pyramidal neurons (Pineda et al. 1998). It is possible that the sAHP in VP neurons is activated by Ca²⁺-induced Ca²⁺ release from internal stores, resulting in AHP resistance to block by specific Ca²⁺ channel blockers. Another possible source could be transient, low-voltage-activated (LVA) T-type Ca²⁺ channels. Although the presence of T-channels in MNCs has been reported (Erickson et al. 1993; Fisher and Bourque 1995), others report little low-threshold Ca²⁺ current (Foehring and Armstrong 1996; Luther and Tasker 2000; Luther et al. 2002). An LVA-type current sensitive to Ni²⁺ was shown to contribute to the depolarizing afterpotential in guinea pig SONs (Erickson et al. 1993). Finally, the differences in the response of these cells to Ca²⁺ could be differences in the sensitivity of the AHP channel to Ca²⁺ or to the differential expression of K⁺ AHP channels themselves.

Conclusions.

These studies revealed the HVA channel sources of Ca²⁺ to the AHP in MNCs. Briefly, in OT and VP neurons of the SON, it appears that mAHP generation reflects activation of only N-type channels although to a significantly lesser degree in VP compared with OT neurons. We observed cell type-specific preferential coupling of the sAHP with specific Ca²⁺ channels, N-type for OT neurons and R-type for VP neurons. This reinforces mechanistic differences of AHP generation between cell types, some of which might underlie the differential modulation of PIP₂ of AHPs by affecting HVA currents (Kirchner et al. 2017).

GRANTS

This work was supported by grants R01NS-044163 and R01HD-072056.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.K.K., R.C.F., and W.E.A. conceived and designed research; M.K.K. and J.C.C. performed experiments; M.K.K. analyzed data; M.K.K., R.C.F., J.C.C., and W.E.A. interpreted results of experiments; M.K.K. prepared figures; M.K.K., R.C.F., and W.E.A. drafted manuscript; M.K.K., R.C.F., J.C.C., and W.E.A. edited and revised manuscript; M.K.K., R.C.F., J.C.C., and W.E.A. approved final version of manuscript.