Activation mechanism of a neuromodulator-gated pacemaker ionic current

Neuronal rhythmic activity is generated by either network-based or cell-autonomous mechanisms. In the pyloric network of decapod crustaceans, the activation of a neuromodulator-gated pacemaker current is crucial for the generation of rhythmic activity. This current is activated by several neuromodulators, including peptides and acetylcholine, presumably via metabotropic receptors. We have previously demonstrated a novel extracellular calcium-sensitive voltage-dependence mechanism of this current. We presently report that the activation mechanism depends on intracellular and extracellular calcium-sensitive components.

Abstract

The neuromodulator-gated current (I_MI) found in the crab stomatogastric ganglion is activated by neuromodulators that are essential to induce the rhythmic activity of the pyloric network in this system. One of these neuromodulators is also known to control the correlated expression of voltage-gated ionic currents in pyloric neurons, as well as synaptic plasticity and strength. Thus understanding the mechanism by which neuromodulator receptors activate I_MI should provide insights not only into how oscillations are initiated but also into how other processes, and currents not directly activated by them, are regulated. To determine what specific signaling molecules are implicated in this process, we used a battery of agonists and antagonists of common signal transduction pathways. We found that the G protein inhibitor GDPβS and the G protein activator GTPγS significantly affect I_MI amplitude, suggesting that its activation is mediated by G proteins. Interestingly, when using the more specific G protein blocker pertussis toxin, we observed the expected inhibition of I_MI amplitude but, unexpectedly, in a calcium-dependent fashion. We also found that antagonists of calcium- and calmodulin-associated signaling significantly reduce I_MI amplitude. In contrast, we found little evidence for the role of cyclic nucleotide signaling, phospholipase C (PLC), or kinases and phosphatases, except two calmodulin-dependent kinases. In sum, these results suggest that proctolin-induced I_MI is mediated by a G protein whose pertussis toxin sensitivity is altered by external calcium concentration and appears to depend on intracellular calcium, calmodulin, and calmodulin-activated kinases. In contrast, we found no support for I_MI being mediated by PLC signaling or cyclic nucleotides.

NEW & NOTEWORTHY Neuronal rhythmic activity is generated by either network-based or cell-autonomous mechanisms. In the pyloric network of decapod crustaceans, the activation of a neuromodulator-gated pacemaker current is crucial for the generation of rhythmic activity. This current is activated by several neuromodulators, including peptides and acetylcholine, presumably via metabotropic receptors. We have previously demonstrated a novel extracellular calcium-sensitive voltage-dependence mechanism of this current. We presently report that the activation mechanism depends on intracellular and extracellular calcium-sensitive components.

many neuromodulators act through G protein-coupled receptors (Clark et al. 2008; Doi and Ramirez 2008; Jaber et al. 1996; Olson et al. 2005; Wei and Wang 1990). They also regulate various parameters of specific ionic currents (Harris-Warrick et al. 1998; Marder 2012; Marder et al. 2014). In so doing, they impact cellular and network function in multiple important ways. For example, rhythmic activity can be activated (Bose et al. 2014; Golowasch and Marder 1992b; Marder 2012; Zhao et al. 2010) or modified (LeBeau et al. 2005; Marder 2012; Miles and Sillar 2011; Vilim et al. 2010; Wu et al. 2010; Xi et al. 2015), entire networks can be reconfigured to serve a new function (Fénelon et al. 1998; Hooper and Moulins 1989, 1990; Koch et al. 2011; Lieske et al. 2000; Ramirez 1998), synaptic function can be altered (Böhm et al. 2015; Johnson et al. 1993a, 1993b, 1995; Mackey et al. 1989; Small et al. 1992; Zhao et al. 2011), and plasticity itself can be regulated by neuromodulators (Lawrence et al. 2015; Pawlak et al. 2010; Zhou et al. 2007). Given the profusion of neuromodulators and their effects on so many signaling pathways and ion channels (Dickinson 2006; Doi and Ramirez 2008; Marder 2012), understanding how neuromodulator receptors activate their effectors is important, and is still incompletely understood for many peptide receptors. This is crucial in many cases, especially for neuronal networks in which neuromodulator effects may be particularly important to ensure stability of activity, due to the convergence of neuromodulator actions at the cellular level. Such convergence might provide resistance to overmodulation. It can also promote network stability due to occlusion of the effects of additional modulators acting on a common effector (Marder 2012; Swensen and Marder 2000) and due to neuromodulator-dependent conductance homeostasis mechanisms (Khorkova and Golowasch 2007; O’Leary and Marder 2016). Nevertheless, within the same network, these different neuromodulators can still express divergent effects in their patterns of activation if different subsets of modulators specifically target distinct cell types (e.g. Doi and Ramirez 2008; Marder 2012; Swensen and Marder 2001).

In this study we used a system that has been widely studied to understand the effects of neuromodulation, the pyloric network of the stomatogastric ganglion (STG) of the crab Cancer borealis. This system displays neuromodulator-specific patterns of activity (Hooper and Marder 1984; Marder 2012; Swensen and Marder 2001) and allows for experimental control of the release of endogenous neuromodulators. Furthermore, this network is known to be a target of a large number of peptidergic, aminergic, cholinergic and gaseous modulators (Marder 2012; Marder and Bucher 2007) through a variety of receptors. Yet, only very few receptors have been identified (Clark et al. 2008; Garcia et al. 2015; Spitzer et al. 2008). Neuromodulator receptors that activate the pyloric rhythm include five peptide receptors and the metabotropic ACh receptor, all of which ultimately activate the neuromodulator-dependent inward current (I_MI) (Golowasch and Marder 1992b; Swensen and Marder 2001, 2000) that acts as a pacemaker current for the network (Bose et al. 2014; Zhao et al. 2010). The lateral pyloric (LP) neuron, which is the focus of this study, is known to respond to all these neuromodulators (Golowasch and Marder 1992b; Swensen and Marder 2001; 2000), and a transcript of a close homolog of the crustacean cardioactive peptide (CCAP) receptor has been shown to be expressed by LP and other pyloric neurons (Garcia et al. 2015). At least one of these peptides (proctolin) is also known to control the coordinated expression of ionic currents, which is thought to participate in the homeostatic regulation of the stability of the pyloric activity (Golowasch 2015; Khorkova and Golowasch 2007). In addition, proctolin also regulates short-term synaptic dynamics in the network, and consequently synaptic strength and the short-term stability of network activity (Zhao et al. 2011). Finally, endogenous neuromodulators appear to suppress alternative pacemaking mechanisms in the system. This is revealed by the spontaneous but slow recovery of the pyloric activity after endogenous neuromodulation is persistently suppressed (decentralization) (Luther et al. 2003; Thoby-Brisson and Simmers 1998, 2000, 2002). This recovery is correlated with the up- and downregulation of a number of voltage-gated and synaptic ionic currents (Khorkova and Golowasch 2007; Thoby-Brisson and Simmers 2002) in a transcription-dependent manner (Thoby-Brisson and Simmers 2000). Given that this process is activity independent (Thoby-Brisson and Simmers 1998), it is assumed that deprivation of a subset of neuromodulators contained in the stomatogastric nerve initiates this process of recovery of activity.

Furthermore, we recently showed that the mechanism of voltage dependence of I_MI appears to be a particularly interesting one, because it seems to involve the active sensing of extracellular calcium levels via activation of a metabotropic calcium-sensing receptor (Gray and Golowasch 2016). Yet, despite the evident importance of the role of I_MI in pyloric network function, and the interesting properties of this current in general, nothing is known about the signaling mechanisms underlying its activation, or to what extent the voltage-dependence mechanism interacts with, or is independent from, the I_MI current-gating mechanism. Thus the focus of this study is to attempt to pharmacologically identify and characterize the signaling pathway(s) activated by two peptidergic neuromodulators, proctolin and CCAP, that activate I_MI in the LP neuron (Swensen and Marder 2000) This current is characterized by its strong voltage dependence, extracellular calcium dependence of its voltage dependence, divalent cation sensitivity, and monovalent cation permeability (Garcia et al. 2015; Golowasch and Marder 1992b; Gray and Golowasch 2016; Swensen and Marder 2000).

In this study, assuming that proctolin and CCAP converge on a common intermediate, and ultimately on I_MI in the LP neuron (Garcia et al. 2015), we employed a battery of agonists and antagonists of second messenger pathways targeting G proteins, cyclic nucleotides, phospholipase C, and general inhibitors of kinases and phosphatases. We found that proctolin-induced I_MI amplitude requires a pertussis-sensitive G protein but is insensitive to cyclic nucleotides, phospholipase C, and general kinase and phosphatase inhibitors. On the other hand, we found evidence for the involvement of intracellular calcium signaling pathways and calmodulin-activated proteins, as well as an intriguing dependence of pertussis toxin action on extracellular calcium levels.

METHODS

Animals.

Crabs of the species C. borealis were purchased from local fisheries and housed in saltwater aquaria at 8–12°C. Crabs were anesthetized on ice at least 30 min before dissection. The stomatogastric nervous system (STNS) was dissected out and pinned on Sylgard-lined dishes as previously described (Gutierrez and Grashow 2009; Maynard and Dando 1974). The isolated STNS was continuously perfused with chilled saline (12–14°C) composed of (in mM) 440 NaCl, 11 KCl, 13 CaCl₂, 26 MgCl₂, 5 maleic acid, and 11 Trizma base, pH 7.4–7.5. In low-calcium solutions, MgCl₂ was added in equimolar amounts to compensate for the reduced calcium. In all experiments, the somata of the STG neurons were exposed by desheathing and pinning down the surrounding connective tissue onto a Sylgard-lined petri dish to expose the neuropile. All data reported were obtained from LP neurons.

Electrophysiology.

Extracellular recordings were made by building Vaseline wells around a section of the lateral ventricular (lvn) or dorsal ventricular nerves (dvn) and placing stainless steel wires, connected to an A-M Systems 1700 differential AC amplifier (Carlsberg, WA), into the Sylgard lining the bottom of the recording dishes. Ground electrodes were either AgCl pellets (Molecular Devices, Sunnydale, CA) or chloride-coated silver wires (coated by submerging metallic silver wires in straight Clorox bleach-sodium hypochlorite solution for at least 10 min). All intracellular recordings, unless otherwise stated, were obtained with an Axoclamp 2B amplifier (Molecular Devices) and digitized with either a Digidata 1322A or 1440 (Molecular Devices) and recorded onto a computer using the pClamp 9 or 10.4 software suite (Molecular Devices). Currents were recorded in two-electrode voltage clamp (TEVC) and were passively filtered using an RC filter at 4-kHz cutoff frequency. A circuit diagram of this filter is available upon request. LP neurons were identified by both hyperpolarization at the soma and observation of their disappearance from extracellular recordings in the lvn and by matching intracellular recordings of the neuron’s action potentials to those recorded extracellularly on the lvn.

Electrodes were pulled on a Sutter P-97 puller (Navato, CA) with resistances of 15–25 MΩ for the voltage-recording microelectrode (ME1) and 10–20 MΩ for the current-passing microelectrode (ME2). All recording electrode solutions consisted of 20 mM KCl plus 0.6 M K₂SO₄. The same was used for current injection solutions except for pressure injection experiments, in which case we used a vehicle solution of 500 mM KCl with or without 20 mM tetraethylammonium (TEA). For pertussis toxin protomer A (PTx-A) injection experiments and their controls, solutions buffered with 10 mM HEPES at pH 7.2 were used.

Membrane-permeable drug experiments.

Membrane permeable drugs were applied for 30–40 min before measurement of I_MI. To build dose-response curves, we started at the lowest concentration and progressed to higher concentrations.

Pressure injection experiments.

To test membrane impermeable modulators of G protein function (GDPβS, GTPγS, and PTx-A), we used the same electrode (ME2) to pressure inject modulators and to inject current. We found that the higher resistance electrodes used for recording were prone to clogging. Therefore, pressure injection electrodes were made with much lower resistance (8–15 MΩ measured in standard intracellular K₂SO₄ recording solution). TEA (20 mM) was included in the electrode to monitor the efficacy of the pressure injection experiments by monitoring the inhibition by TEA of the high-threshold potassium current (I_HTK) (Golowasch and Marder 1992a; Zhao and Golowasch 2012). All experiments began in normal saline plus 0.1 μM tetrodotoxin (TTX) and were maintained in this saline throughout the experiments (note that external TEA was not added until pressure injection was completed and shortly before measurement of I_MI began). To assess whether pressure injection was working, we measured I_HTK once every hour during the pressure injection procedure. Pressure injections were applied for 20- to 30-min intervals; they began at 10 psi, with pressure gradually increased until I_HTK reduction was observed. Typical final pressure injection levels were between 20 and 30 psi, but this was highly variable because it depended on pipette characteristics and quality of impalement. When transient I_HTK was reduced by at least 50% of its original value, the final solution in which I_MI was measured was added and I_MI measurements were started (described below).

Neuromodulators.

Proctolin was purchased from American Peptide or Bachem. CCAP was initially purchased from American Peptide, but in winter 2013–spring 2014, this peptide did not produce I_MI comparable to that reported in the literature (Swensen and Marder 2000) and our own previous experiments. Therefore, all CCAP data from 2013 to 2014 were omitted. and all later experiments requiring CCAP were performed using CCAP obtained from Bachem. All other neuromodulators were obtained from Sigma Co.

Solutions and drugs.

The following were obtained from Sigma-Aldrich: N-(6-aminohexyl)-5-chloro-1-naphthalenesulfonamide hydrochloride (W7; all experiments except dynasore experiments), TEA, PTX, forskolin, pilocarpine, guanosine 5′-[γ-thio]triphosphate tetralithium salt (GTPγS), guanosine 5′-[β-thio]diphosphate tri-lithium salt (GDPβS), 8-bromo-adenosine-3′,5′-cyclic monophosphate sodium salt (8-Br-cAMP), 8-bromo-guanosine cyclic 3′,5′-monophosphate sodium salt (8-Br-cGMP), dynasore, N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide dihydrochloride (H89), 1-(5-iodonaphthalene-1-sulfonyl)-1H-hexahydro-1,4-diazepine hydrochloride (ML-7), okadaic acid, genistein, gallein, and neomycin. The following were obtained from Tocris Bioscience: W7 (dynasore experiments), 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid tetrakis(acetoxymethyl ester) (BAPTA-AM), cyclosporin A, and edelfosine (Et-18-OCH₃). Calmidazolium (CDZ), dantrolene, ryanodine, PTx-A, and staurosporine were purchased from Enzo Life Sciences. CCAP was purchased from Bachem. TTX was purchased from Calbiotech or Tocris Bioscience. BSA was purchased from Fisher Scientific. N-[2-[[[3-(4-chlorophenyl)-2-propenyl]methylamino]methyl]phenyl]-N-(2-hydroxyethyl)-4-methoxybenzenesulphonamide (KN-93) was obtained from EMD Bioscience. Dasatinib was purchased from Selleck Chemical.

Measurement of I_MI.

All I_MI recordings were generated in TEVC exactly as described in Gray and Golowasch (2016): the current was obtained by subtracting control currents, elicited with voltage ramps from +20 to −80 mV at 75 mV/s in the absence of the modulator that activates I_MI, from currents elicited with an identical voltage ramp in the presence of the activating neuromodulator, in this study either proctolin or CCAP. Unless otherwise stated, all recordings of I_MI were made in standard “I_MI recording saline,” which contained normal Cancer saline plus 0.1 μM TTX to block sodium currents, 20 mM TEA to block potassium currents, 10 μM PTX to block synaptic currents, 5 mM CsCl to block the H-current, and 200 μM CdCl₂ to block calcium currents. In some cells, spontaneous voltage oscillations were observed under these conditions; when this happened, TTX and PTX concentrations were transiently raised to 1 and 30 μM, respectively, until oscillations stopped or were attenuated. Standard I_MI recording saline was then resumed for at least 20 min before I_MI was measured. Both proctolin and CCAP were bath-applied at 1 μM for 1–3.5 min (perfusion rate of 3–4.5 ml/s).

In a previous study, we reported that I_MI amplitude was sensitive to application number and calcium concentration (Gray and Golowasch 2016). Therefore, in normal extracellular calcium, only applications 2–5 were included in data analysis. In experiments conducted with low extracellular calcium concentration, analysis was adjusted for covariate application number as described previously (Gray and Golowasch 2016).

We defined two properties of I_MI: slope conductance and amplitude. The slope conductance is the slope of the current-voltage relationship (I-V curve) between −20 and −75 mV; this value is used as our quantification of the voltage dependence of I_MI reported in Gray and Golowasch (2016). This quantification was robust and detected the previously reported reduction of voltage dependence of I_MI induced by lowering extracellular calcium concentration (Golowasch and Marder 1992b; Gray and Golowasch 2016; Swensen and Marder 2000). To quantify I_MI amplitude, we used the current level at −15 mV. We base this criterion on our previous results showing that extracellular calcium tends not to affect I_MI when quantified at this voltage (see Fig. 1 in Gray and Golowasch 2016). To distinguish whether a drug affected either I_MI activation or voltage dependence, we performed statistical analyses on these two properties of I_MI.

Measurement of potassium currents I_HTK and I_A.

To assess the efficacy of our pressure injection treatment, we measured the mean current during the 10 ms surrounding the peak of I_HTK at +20 mV. This current was measured in response to a current pulse from a holding voltage of −40 mV in the presence of 0.1 µM TTX (Golowasch and Marder 1992a; Zhao and Golowasch 2012). The transient (inactivating) potassium current I_A was measured by subtraction of currents elicited with depolarizing pulses from a holding voltage of −40 mV from those elicited from a holding potential of −80 mV.

Statistics and data analysis.

All calculations for I_MI, difference currents, and leak subtractions were performed as described in Gray and Golowasch (2016). All graphs making comparisons between conditions show mean ± SE unless noted otherwise. Statistical analysis was performed as described previously in Gray and Golowasch (2016). In brief, analysis of variance with post hoc Tukey tests were used for normal homoscedastic data. Data that did not pass these tests were ranked, and the same statistical procedures were performed on these rank scores. Two-tailed paired and unpaired Student t-tests were used for normal two-group comparisons. Mann-Whitney U-tests were used for non-normal two-group comparisons. Analysis of covariance was used for data in Table 2 that were obtained in low calcium concentration to account for previously observed I_MI desensitization in low calcium (Gray and Golowasch 2016). All ANOVAs and Student t-tests were performed with SigmaPlot 11.0. Analysis of covariance was carried out with IBM SPSS Statistics v22. Data differences were considered significant at P < 0.05.

Table 2.

Effects of signal transduction modulators on proctolin-induced I_MI

Target	Drug	Concentration	n	IMI(−15 mV), nA	Statistical Test	P Value
PLC	Edelfosine (normal calcium)	0 μM	6	−1.287 ± 0.294	One-way ANOVA	0.370
		10 μM	4	−0.912 ± 0.336
		20 μM	2	−0.515 ± 0.098
	Neomycin (low calcium)	0 μM	14	−0.892 ± 0.155	One-way ANOVA + covariate application no.a	0.093
		0.10 μM	3	−1.428 ± 0.310
		0.33 μM	5	−1.583 ± 0.259
		1.00 μM	3	−0.826 ± 0.333
Ser/Thr kinases (also Tyr)	Staurosporine (low calcium)	0.00 μM	17	−0.886 ± 0.108	One-way ANOVA + covariate application no.b	0.083
		0.10 μM	3	−0.217 ± 0.258
		0.33 μM	3	−0.754 ± 0.258
cAMP	8-Br-cAMP (normal calcium)	0 μM	4	−1.579 ± 0.486	t-Test	0.721
		500 μM	3	−1.839 ± 0.446
Adenylyl cyclase (cAMP)	Forskolin (normal calcium)	0 μM	4	−1.586 ± 0.491	t-Test	0.389
		10 μM	4	−0.922 ± 0.521
PKA	H89 (Normal calcium)	0 μM	6	−1.311 ± 0.360	t-Test	0.886
		20 μM	4	−1.224 ± 0.472
cGMP	8-Br-cGMP (normal calcium)	0 μM	7	−1.363 ± 0.352	One-way ANOVA	0.714
		500 μM	2	−0.942 ± 0.076
		1,000 μM	3	−1.683 ± 0.742
Phosphatases (general)	Okadaic acid (normal calcium)	0 nM	4	−0.530 ± 0.106	Paired t-test; IMI = okadaic acid	0.056
		10 nM	4	−0.223 ± 0.157
	Okadaic acid (low calcium)	0 nM	16	−0.672 ± 0.118	One-way ANOVA + covariate application no.c	0.319
		3.33 nM	4	−0.706 ± 0.259
		10 nM	4	−1.168 ± 0.235
		33 nM	3	−0.676 ± 0.292
Phosphatases (calcineurin)	Cyclosporin A (normal calcium)	0 μM	5	−1.603 ± 0.393	One-way ANOVA	0.459
		5 μM	2	−1.171 ± 0.457
		13.3 μMe	3	−1.041 ± 0.295
Tyrosine kinases (general; see also staurosporine)	Genistein (low calcium)	0 μM	8	−0.703 ± 0.173	One-way ANOVA + covariate application no.d	0.783
		100 μM	5	−0.783 ± 0.220
SRC kinases	Dasatinib (low calcium)	0 μM	4	−1.493 ± 0.378	t-Test	0.523
		100 μM	4	−1.194 ± 0.229

Values are means ± SE. In experiments conducted with low extracellular calcium, analyses was adjusted for covariate application number.

^aCorrected mean reported with covariate application no. = 3.08.

^bCorrected mean reported with covariate application no. = 3.83.

^cCorrected mean reported with covariate application no. = 4.30.

^dCorrected mean reported with covariate application no. = 3.31.

^eThe 13.3 μM cyclosporin A group consisted of 2 preparations at 10 μM and 1 preparation at 20 μM.

Low-calcium-induced depolarization of resting membrane potential is prevented by BSA.

As described previously (Gray and Golowasch 2016), we found that application of saline with calcium concentrations ≤2 mM significantly depolarized membrane potential and reduced input resistance at −50 mV, unless we included 0.5% BSA. Importantly, as illustrated in Table 1, no effect of BSA was observed on I_A at +20 mV [1-way ANOVA: F(4,20) = 2.22, P = 0.09], I_HTK at +20 mV [1-way ANOVA: F(4,20) = 1.45, P = 0.25], or proctolin-induced I_MI amplitude at −15 mV [1-way ANOVA: F(3,32) = 0.255, P = 0.857]. Therefore, experiments conducted with low calcium concentration, unless otherwise noted, were always supplemented with 0.5% BSA.

Table 1.

Effects of BSA on I_HTK, I_A, and I_MI

BSA, %	IHTK (+20 mV), nA	IA (+20 mV), nA	IMI (−15 mV), nA
0	132.7 ± 23.0 (5)	29.9 ± 3.3 (5)	−0.98 ± 0.22 (8)
0.01	124.9 ± 24.6 (5)	33.4 ± 5.0 (5)	−1.09 ± 0.15 (4)
0.1	131.5 ± 27.3 (5)	36.2 ± 4.2 (5)	−1.13 ± 0.13 (19)
0.5	137.0 ± 25.8 (5)	36.7 ± 3.4 (5)	−0.93 ± 0.24 (5)
1.0	128.1 ± 23.1 (5)	36.2 ± 2.8 (5)	−1.11 ± 0.23 (2)*
2.0	123.6 ± 22.9 (5)	40.7 ± 5.0 (5)	−1.19 ± 0.03 (2)*

Values are means ± SE; n is indicated parentheses. BSA does not affect I_HTK (at +20 mV), I_A (at +20 mV), or I_MI (at −15 mV) (I_HTK and I_A were measured in normal saline and 1 µM TTX).

^*Data were omitted from statistical analysis.

RESULTS

Following Swensen and Marder (2000), we hypothesize that some peptidergic neuromodulator receptors converge on a common intermediate that, in turn, activates I_MI. Expression of the putative G protein-coupled CCAP receptor in this system has been shown to correlate with CCAP-induced I_MI (Garcia et al. 2015). It is also well known that muscarinic agonists, such as pilocarpine, which activate I_MI in the crab STG (Swensen and Marder 2000), act via G protein-coupled receptors (GPCR) in other systems (Wei and Wang 1990). Proctolin, pilocarpine, and other peptides also activate I_MI in this system (Golowasch and Marder 1992b; Swensen and Marder 2000). Recently, a proctolin receptor that has the structural features of a GPCR was predicted from the related Homarus americanus transcriptome (Christie et al. 2015). In light of this, we hypothesize that the receptor to the peptide proctolin is a GPCR, and in this study we attempt to identify and characterize the putative G protein-coupled signaling pathway(s) that mediate(s) the activation of I_MI.

I_MI depends on a G protein with calcium-dependent pertussis toxin sensitivity.

Garcia et al. (2015) provided indirect evidence that CCAP-induced I_MI may be mediated by G proteins by showing that CCAP receptor mRNA cloned for C. borealis correlates well with a cell’s ability to produce CCAP-induced I_MI (Garcia et al. 2015). However, even if a GPCR exists, it does not necessarily imply that it signals through G proteins, as has been shown, for instance, for the activation of the sodium leak current (NaLCN) through the NKR1 receptor (Lu et al. 2009) in ventral tegmental area and hippocampal cells, and cholinergic agonists in substantia nigra neurons (Lutas et al. 2016). In these cases, NaLCN channels are activated by a non-G protein pathway dependent on a serine family kinase (SFK)-based mechanism, even though the NKR1 and muscarinic receptors are known GPCRs. Therefore, in the present study we attempted to establish whether proctolin-induced I_MI was dependent on G proteins, and if so, the particular G protein’s identity.

GDPβS inhibits I_MI.

To test the hypothesis that proctolin-induced I_MI is mediated by activation of a G protein, we pressure injected either 20 mM TEA plus 500 mM KCl, or the same with the addition of 10 mM GDPβS, a nonphosphorylatable broad-spectrum inhibitor of G proteins (Eckstein et al. 1979), in either normal (13 mM) or low (2 mM)-calcium saline. Consistent with I_MI being mediated by a G protein, as shown in Fig. 1, GDPβs significantly altered proctolin-induced I_MI amplitude at −15 mV in both high- and low-calcium saline [2-way ANOVA, calcium: F(1,30) = 5.99, P = 0.02; GDPβS: F(1,30) = 15.034, P = 5.3 × 10⁻⁴; interaction: F(1,30) = 0.376, P = 0.544]. A post hoc Tukey test showed that, in both normal calcium (P = 0.003) and low calcium (P = 0.035), GDPβS significantly attenuated I_MI amplitude relative to pressure injection alone (Fig. 1C). Similar results were obtained for proctolin-induced I_MI amplitude with pressure injection of 10 mM GDPβS with only 500 mM KCl in the pipette (no TEA): normal calcium control (n = 3), −0.81 ± 0.4 nA; normal calcium GDPβS (n = 2), −0.11 ± 0.5 nA; low-calcium control (n = 3), −1.74 ± 0.4 nA; and low-calcium GDPβS (n = 2), −0.14 ± 0.5 nA [2-way ANOVA, calcium: F(1,6) = 1.102, P = 0.334; GDPβS: F(1,6) = 6.263, P = 0.046; interaction: F(1,6) = 0.929, P = 0.372]. These results suggest that proctolin-induced I_MI signaling requires G proteins.

Fig. 1.

Proctolin-induced I_MI activation requires G proteins: effect of GDPβS. GDPβS (gray) or control solutions (black) were pressure injected in normal (13 mM CaCl₂; solid curves) or low extracellular (2 mM CaCl₂) calcium levels (dotted curves). A: averaged I-V curves of proctolin-induced I_MI in normal calcium. B: averaged I-V curves of proctolin-induced I_MI in 2 mM (low) extracellular calcium. C: quantification of the effect of GDPβS on I_MI: both GDPβs and calcium significantly altered I_MI amplitude at −15 mV [2-way ANOVA, calcium: F(1,30) = 5.99, P = 0.02; GDPβS: F(1,30) = 15.034, P = 5.3 × 10⁻⁴; interaction: F(1,30) = 0.376, P = 0.544]. Tukey comparisons: *P < 0.05; **P < 0.01; ns, not significant. Error bars are SE. Solid bars indicate normal extracellular Ca²⁺ (control); hatched bars indicate low calcium level.

GTPγS occludes I_MI.

One approach to study the putative role of G proteins in proctolin-induced I_MI is to perform occlusion experiments, in which an activator of G protein, such as GTPγS, is applied before activation of the receptor with the natural ligand (i.e., proctolin). It has previously been shown that proctolin can occlude the response to CCAP in crab LP neurons (Garcia et al. 2015; Swensen and Marder 2000). Therefore, we predicted that if proctolin-induced I_MI is G protein dependent, then pressure injection of G protein activators should occlude proctolin-induced I_MI. As shown in Fig. 2, GTPγS, but not calcium, was capable of occluding I_MI amplitude at −15 mV [2-way ANOVA; calcium: F(1,22) = 2.136, P = 0.158; GTPγS: F(1,22) = 13.562, P = 0.001; interaction: F(1,22) = 0.0817, P = 0.778]. A post hoc Tukey test showed that proctolin-induced I_MI was significantly occluded in both normal (P = 0.02) and low calcium (P = 0.014). Consistent with the effects of GDPβS reported above, these findings suggest that proctolin-induced I_MI requires G protein signaling.

Fig. 2.

Proctolin-induced I_MI activation requires G proteins: effect of GTPγS. GTPγS was pressure injected in a vehicle solution of 500 mM KCl + 20 mM TEA. Control pressure injections used the vehicle solution only. A: averaged I-V curves of proctolin-induced I_MI after pressure injection with either control (black) or 10 mM GTPγS (gray) solution in normal (13 mM) calcium. B: same as A but in saline containing low (2 mM) calcium. C: quantification of the effect of GTPγS on I_MI amplitude at −15 mV. A 2-way ANOVA showed that GTPγS, but not calcium, was capable of occluding I_MI amplitude [calcium: F(1,22) = 2.136, P = 0.158; GTPγS: F(1,22) = 13.562, P = 0.001; interaction: F(1,22) = 0.0817, P = 0.778]. Tukey post hoc test: *P < 0.05; ***P < 0.001; ns, not significant. Error bars are SE. Solid bars indicate normal extracellular Ca²⁺ (control); hatched bars indicate low calcium level.

We attempted to directly measure the activation of I_MI by G protein activators, but that proved unsuccessful. This is most likely due to the fact that the point of injection (the soma) is very far away from the area where neuromodulator receptors are expressed (the neuropile; cf. Golowasch and Marder 1992b). The consequence of this is that it took more than an hour for the pressure-injected activators to become effective, and in that amount of time, small amplitude changes of any of the background currents can obfuscate the recordings of the very small I_MI. In addition, during the injections, we needed to avoid exposing the cells to the various blockers we normally used to measure I_MI, which were applied only shortly before those measurements (see methods). Given that I_MI can only be detected by subtraction, this denies us the opportunity to directly measure the effects of GTPγS.

Pertussis toxin (PTx-A) inhibits I_MI in a calcium-dependent manner.

To identify the particular G protein that could be mediating proctolin-induced I_MI, we pressure-injected the specific G protein inhibitor PTx-A together with 20 mM TEA + 500 KCl + 10 mM HEPES (pH 7.2), or the vehicle (no PTx-A) as control. PTx-A is known to inhibit only trimeric G proteins that contain an ADP-ribosylation site, i.e., G proteins whose α-subunit is of either the α_i, α_o, or α_t types (Katada 2012). As illustrated in Fig. 3, PTx-A significantly reduced I_MI amplitude at −15 mV [2-way-ANOVA, calcium: F(1,22) = 20.630, P = 1.6 × 10⁻⁴; PTx-A: F(1,22) = 11.735, P = 0.002; interaction: F(1,22) = 3.541, P = 0.073], even though, surprisingly, low calcium levels also augmented I_MI amplitude in the absence of PTx-A (P < 0.001). A post hoc Tukey test showed that PTx-A inhibited I_MI only at low calcium levels (P = 0.002; in normal calcium P = 0.268), which is supported by the high F value for the interaction between these factors. These results were unexpected, because it was assumed that PTx-A sensitivity should not depend on extracellular calcium, but are consistent with the results observed with GDPβS (including the calcium sensitivity) and GTPγS. These results thus indicate that I_MI is mediated by a PTx-A-sensitive G protein, with higher sensitivity to PTx-A at lower than at normal extracellular calcium levels.

Fig. 3.

The specific G protein inhibitor pertussis toxin (PTx-A) inhibits proctolin-induced I_MI. Control pressure injection solutions of 500 mM KCl + 20 mM TEA (black) or the same solution plus 10 μg/ml PTx-A (gray) were pressure injected in either normal (13 mM) or low (2 mM) extracellular calcium. All recording electrodes used 20 mM KCl + 0.6 M K₂SO₄. A: averaged I-V curves of proctolin-induced I_MI after pressure injection of control solution (black) or 10 μg/ml PTx-A solution (gray). B: same as A but in saline containing low (2 mM) calcium. C: quantification of effect of PTx-A on proctolin-induced I_MI amplitude. A 2-way ANOVA showed that both calcium and PTx-A were capable of altering I_MI amplitude at −15 mV [calcium: F(1,22) = 20.630, P = 1.61 × 10⁻⁴; PTx-A: F(1,22) = 11.735, P = 0.002; interaction: F(1,22) = 3.541, P = 0.073]. Post hoc Tukey test: *P < 0.05; ***P < 0.001; ns, not significant. Error bars are SE. Solid bars indicate normal extracellular Ca²⁺ (control); hatched bars indicate low calcium level.

Calmodulin-dependent signaling.

Swensen and Marder (2000) showed that the calmodulin inhibitor W7 increased I_MI amplitude at both −40 and −80 mV. In a previous paper, we showed that this observed augmentation was due to loss of voltage dependence, rather than enhancement of the current (Gray and Golowasch 2016). In the present study we found that when quantifying proctolin-induced I_MI amplitude at −15 mV, W7 produces a decrease of I_MI amplitude.

We applied varying concentrations of the calmodulin inhibitor W7 while measuring proctolin-induced I_MI amplitude. As shown in Fig. 4A, W7 reduced proctolin-induced I_MI amplitude in a concentration-dependent manner [1-way repeated-measures ANOVA, F(4,19) = 3.24, P = 0.035], although a post hoc Tukey test showed that this inhibition was only significant at 10 μM W7 (P = 0.026). This result supports an activating role (albeit small and perhaps nonlinear) for calmodulin in proctolin-induced I_MI amplitude. This result is consistent with previous observations of an increase in slope conductance by W7 (Gray and Golowasch 2016; Swensen and Marder 2000) and suggests that both voltage dependence and activation of I_MI involve a calmodulin-dependent pathway.

Fig. 4.

Role of calmodulin in proctolin-induced I_MI amplitude. Proctolin-induced I_MI is shown at different concentrations of calmodulin inhibitors W7 (A) and calmidazolium (CDZ; B). A: quantification of W7 data. A 1-way repeated-measures ANOVA showed a small but significant reduction of proctolin-induced I_MI amplitude by W7 [F(4,19) = 3.24, P = 0.035]. B: proctolin-induced I_MI at different concentrations of CDZ. A 1-way repeated-measures ANOVA showed that CDZ reduced proctolin-induced I_MI amplitude [F(3,9) = 6.38, P = 0.013]. Error bars are SE. Tukey test: *P < 0.05.

To confirm that the effects of W7 on I_MI were due to calmodulin, we used a non-napthalenesulfonamide calmodulin inhibitor, calmidazolium (CDZ), an inhibitor that is effective in crustacean systems (Goudeau and Goudeau 2001; Obukhov et al. 1998). As shown in Fig. 4B, CDZ also reduced proctolin-induced I_MI amplitude in a concentration-dependent manner [1-way repeated-measures ANOVA, F(3, 9) = 6.38, P = 0.013], with a post hoc Tukey test showing that this inhibition was significant at 1 μM CDZ (P = 0.015) (and perhaps higher) but not at lower concentrations (Fig. 4B). This result, along with the results using W7 presented above, suggests that proctolin-induced I_MI is dependent on calmodulin signaling. However, although all calmodulin inhibitors had an effect on I_MI slope (Gray and Golowasch 2016), the nonlinear relation of W7 on proctolin-induced I_MI amplitude (maximum effect at 10 μM) suggests that this may not be a simple relationship.

Upstream of calmodulin: calcium signaling.

If activated calmodulin mediates I_MI signaling, it follows that a source of calcium is needed for I_MI signaling. A possible source is intracellularly sequestered calcium. Previously, we tested BAPTA-AM to attempt to reduce intracellular calcium levels and discovered that it is not effective in modifying either I_MI amplitude or the high-threshold potassium current (I_HTK), which is dominated by a calcium-dependent potassium current (Gray and Golowasch 2016). The lack of effect on I_HTK is also consistent with the previously reported lack of effect of BAPTA-AM on the same current in the cardiac ganglion in the same species (Ransdell et al. 2012). It is not clear at this point why BAPTA-AM is not effective, but we think that it may either be the result of tight compartmentalization of intracellular BAPTA after deesterification or that STG neurons lack the necessary esterases to convert BAPTA-AM to BAPTA. To overcome this problem, direct iontophoretic injection of BAPTA into the cells was attempted. However, this resulted in a confounding activity-dependent reduction of background potassium currents (see Golowasch et al. 1999) that complicated the interpretation of the results. Pressure injection of BAPTA was also not without significant problems, because the cells sustained too much damage during the long injections needed to obtain an effect. With no good direct way of confirming whether calcium activator was working at the desired level, we used other pharmacological manipulations of intracellular calcium.

Dantrolene, first characterized in crustacean muscle fibers (Hainaut and Desmedt 1974), is an antagonist of ryanodine receptors, which reduces calcium-induced calcium release (Zhao et al. 2001). As shown in Fig. 5, 3.33 μM dantrolene significantly decreased proctolin-induced I_MI amplitude at −15 mV [paired t-test: t(5) = −3.502, P = 0.017]. This is consistent with the hypothesis that I_MI activation requires calcium-induced calcium release.

Fig. 5.

The ryanodine receptor antagonist dantrolene inhibits proctolin-induced I_MI. Proctolin-induced I_MI is shown before (open bar) and after (filled bar) application of 3.33 μM dantrolene. A paired t-test showed that dantrolene significantly decreased proctolin-induced I_MI amplitude at −15 mV [t(5) = −3.502, P = 0.017]. Error bars are SE. *P < 0.05.

Downstream of calmodulin: calmodulin kinase signaling.

As our results suggest that intracellular calcium and calmodulin play a role in I_MI activation, we examined whether calmodulin-dependent kinases may be involved. We have reported previously that the CaMKII inhibitor KN-93, an inhibitor that has been successfully used in crustacean neurons (Uzdensky et al. 2007; Withers et al. 1998), significantly altered proctolin-induced I_MI voltage dependence (Gray and Golowasch 2016). We now report that KN-93 reduces proctolin-induced I_MI amplitude. Because we had difficulty in obtaining large sample numbers for a large range of antagonist concentrations, for statistical analysis we pooled our data into a low-dose condition consisting of an average KN-93 concentration of 4.2 μM (2 μM, n = 1; 4 μM, n = 1; 5 μM, n = 3) and a high-dose condition with an average concentration of 15 μM KN-93 (10 μM, n = 1; 20 μM, n = 1). As illustrated in Fig. 6A, KN-93 significantly decreased proctolin-induced I_MI amplitude at −15 mV [1-way repeated-measures ANOVA, F(2,5) = 34.09, P = 0.001]. A post hoc Tukey test shows that both the low dose (P = 0.013) and high dose (P = 0.003) were significantly different from control, and from one another (P = 0.034). Similar results were obtained with overnight incubations in 20 μM KN-93 (KN-93, n = 3; control, n = 4; data not shown). These results, combined with the finding that an isoform of CaMKII has been identified in the STNS of the related lobster species Panulirus interruptus (Withers et al. 1998), suggests that CaMKII may play a role in proctolin-induced I_MI activation.

Fig. 6.

Kinase inhibitors reduce I_MI amplitude. A: proctolin-induced I_MI in different concentrations of KN-93. For statistical analysis, KN-93 was grouped into “low-dose” (2–5 μM range, average concentration 4.2 μM), and “high-dose” groups (10–20 μM range, average concentration 15 μM). A 1-way repeated-measures ANOVA showed that KN-93 significantly decreased proctolin-induced I_MI amplitude at −15 mV [KN-93: F(2,5) = 34.09, P = 0.001]. B: proctolin-induced I_MI at different concentrations of the MLCK inhibitor ML-7. A 1-way repeated-measures ANOVA showed that ML-7 decreased proctolin-induced I_MI amplitude at −15 mV [ML-7: F(3,26) = 4.468, P = 0.012]. Post hoc Tukey test: *P < 0.05; **P < 0.01; ns, not significant. Error bars are SE.

We also tested the specific myosin light chain kinase (MLCK) inhibitor ML-7 to examine the hypothesis that MLCK, also a calmodulin-sensitive kinase, may mediate the activation of I_MI. As shown in Fig. 6B, ML-7 decreased proctolin-induced I_MI amplitude at −15 mV in a concentration-dependent manner [1-way repeated-measures ANOVA; ML-7: F(3,26) = 4.468, P = 0.012], with a post hoc Tukey test indicating that this effect is significant at concentration 1 µM and above. These results suggests that both calmodulin-dependent kinases CaMKII and MLCK, and thus, putatively, phosphorylation of I_MI channels or other components of their activation pathway, may play a role in the activation of I_MI.

Role of calcium-sensing receptor.

In a previous paper, we claimed that the calcium-sensing receptor (CaSR), a GPCR, plays a key role in mediating I_MI voltage dependence (Gray and Golowasch 2016). In the present study we tested the effects of the specific CaSR antagonist NPS-2143 and the CaSR agonist R568 on I_MI amplitude. Whereas the CaSR antagonist did not affect proctolin-induced I_MI amplitude at concentrations up to 1 mM [1-way repeated-measures ANOVA, NPS-2143: F(4,22) = 1.085, P = 0.388; data not shown], the agonist R568 induced a significant increase in proctolin-induced I_MI (Fig. 7). A two-way ANOVA for factors R568 and calcium showed significant changes in proctolin-induced I_MI amplitude at −15 mV [calcium: F(1,20) = 18.301, P = 3.7 × 10⁻⁴; R568: F(1,20) = 23.447, P = 9.9 × 10⁻⁵; interaction: F(1,20) = 5.962, P = 0.024). A post hoc Tukey test showed that R568 did not significantly increase I_MI amplitude in normal calcium (P = 0.09), but it significantly increased it in low calcium (P < 0.001). Consistent with our previous findings, when R568 was not present, calcium did not significantly modulate I_MI amplitude (P = 0.135), but when R568 was present, calcium significantly increased I_MI amplitude (P < 0.001). These results suggest that R568 increases I_MI amplitude, but the effect is stronger in low calcium than in normal calcium concentrations. This is consistent with what would be expected if the receptor was receiving saturating concentrations of calcium in the normal calcium condition, and thus having its maximal effect on the current’s voltage dependence (see discussion).

Fig. 7.

Role of the calcium sensing receptor (CaSR) in I_MI amplitude. Proctolin-induced I_MI in control (black) or in the presence of the CaSR agonist R568 (gray) was measured in either 13 mM (solid bars) or 2 mM CaCl₂ saline (hatched bars). A two-way ANOVA for factors R568 and calcium showed significant changes in proctolin-induced I_MI amplitude at −15 mV [calcium: F(1,20) = 18.301, P = 3.67 × 10⁻⁴; R568: F(1,20) = 23.447, P = 9.9 × 10⁻⁵; interaction: F(1,20) = 5.962, P = 0.024]. Post hoc Tukey test: ***P < 0.001; ns, not significant. Error bars are SE.

Negative results.

This work originally sought to identify an intermediate signaling molecule or pathway that enables neuromodulator receptors of the STG pyloric network to converge onto a common pathway that in turn activates I_MI. Although a large amount of evidence appears to implicate calmodulin- and calcium-dependent paths in I_MI activation, other pathways showed no evidence for being involved in I_MI signaling. For the sake of completion, we describe briefly our negative findings below.

I_MI activation is not sensitive to PLC or general serine-threonine kinase signaling.

Because phospholipase C (PLC) signaling often involves intracellular calcium concentration changes and some of the neuromodulators examined in this study are involved in PLC-mediated signaling elsewhere, we examined PLC signaling involvement in proctolin-induced I_MI activation. As summarized in Table 2, neither the PLC inhibitor edelfosine nor neomycin, both of which have been shown to be effective in crustacean systems (edelfosine on crustacean proteins in an expression system: Clark et al. 2004; neomycin in adult STG cells: Levi and Selverston 2006) were capable of altering proctolin-induced I_MI amplitude. Interestingly, the PKC and general kinase inhibitor staurosporine (Meggio et al. 1995) did not significantly alter I_MI amplitude either (Table 2). These results suggest that the proctolin receptor does not activate I_MI directly through PLC or a number of serine/threonine kinases given the known promiscuous effects of staurosporine on these kinases at the concentrations used (Meggio et al. 1995). This is also consistent with the findings of Swensen and Marder (2000), who reported that the PLC inhibitor U-73122 did not alter I_MI activation (Swensen and Marder 2000).

We further examined several specific and nonspecific inhibitors of phosphatases and kinases to determine whether they play a role in proctolin-induced I_MI activation. As summarized in Table 2, the general phosphatase inhibitor, okadaic acid did not significantly alter proctolin-induced I_MI amplitude. Although the inhibitory effect of okadaic acid in normal calcium saline was not strictly significant at our criterion for significance (P < 0.05), it was close to statistical significance (P = 0.056), and furthermore, it was quite large (63%). Interestingly also, there was no effect of okadaic acid in low-calcium saline (P = 0.319; discussed below). Consistent with the inefficacy of okadaic acid, cyclosporin A, an inhibitor of the phosphatase calcineurin, did not alter I_MI amplitude. Because both okadaic acid and cyclosporin A previously have been shown to be effective in this (Ransdell et al. 2012) and other crustacean species (Lin and Fu 2005; Nery et al. 1997), we interpret this as proctolin-induced I_MI having little or no dependence on phosphatase activity, although the effect of okadaic acid in normal calcium saline may reflect a complex participation of phosphatases in this process.

Consistent with our staurosporine findings, neither the general tyrosine kinase inhibitor genistein nor the SRC kinase inhibitor dasatinib was capable of altering I_MI amplitude (Table 2). Whereas genistein has been used previously in other crustacean systems (Chuo et al. 2005; Thiyagarajan et al. 2009), the use of dasatinib is, to our knowledge, unprecedented in crustacean systems. However, the findings that none of these three agents were capable of altering I_MI activation suggests that tyrosine kinase signaling is not involved in I_MI activation.

I_MI amplitude is not sensitive to cyclic nucleotides.

We tested many drugs that are presumed to alter cyclic AMP (cAMP) signaling or downstream kinases, but they had no effect on I_MI amplitude. As summarized in Table 2, neither the cAMP analog 8-Br-cAMP, the adenylyl cyclase agonist forskolin, nor the PKA inhibitor H89 produced significant changes in proctolin-induced I_MI, although they all have been used effectively at lower concentrations in crustacean systems (forskolin and 8-Br-cAMP: Ballo et al. 2010; Flamm et al. 1987; Spruston and Nusbaum 1991; Zhang et al. 2010; H89: Zhang et al. 2010). These results do not fully rule out the possibility that cAMP could be involved in long-term regulation of I_MI, but the lack of any effect during the application of these drugs suggests that cAMP signaling is not necessary for proctolin-induced I_MI.

Similar to results found for 8-Br-cAMP, we found no effect for the cyclic GMP analog 8-Br-cGMP, another drug that has been successfully utilized in crustacean systems (Michel and Ache 1992; Spruston and Nusbaum 1991). These results are consistent with those reported by Swensen and Marder (2000) and suggest that I_MI activation does not involve or require cyclic nucleotide signaling.

Does modifying I_MI signaling pathways affect neuronal activity?

I_MI modifies the activity of its target neurons. Because I_MI is thought to be the pacemaker current of the pyloric network, via its effects on the pacemaker neurons of the network (Bose et al. 2014; Zhao et al. 2010), it would be expected that modifying I_MI would affect the activity of the entire pyloric network. W7 and CDZ reduce both the amplitude of the current (Fig. 4) and its voltage dependence (Gray and Golowasch 2016; Swensen and Marder 2000). Therefore, modifying I_MI amplitude and voltage dependence would be expected to change the frequency of the pyloric rhythm. We applied 1 μM CDZ in preparations with their full complement of neuromodulators intact (n = 3) or in decentralized preparations bathed in 0.1 μM proctolin (n = 4) to activate I_MI enough to elicit rhythmic pyloric activity. Surprisingly, in neither case did we observe a statistically significant difference in pyloric rhythm frequency after ~60 min of application of CDZ compared with the control immediately before the application [non-decentralized/control: 1.26 ± 0.08 Hz; non-decentralized + CDZ: 1.26 ± 0.14 Hz; t(2) = 0.182, P = 0.972; decentralized + 0.1 μM proctolin/control: 0.40 ± 0.13 Hz; decentralized + 0.1 μM proctolin + CDZ: 0.36 ± 0.12 Hz; t(3) = 2.019, P = 0.137; paired Student t-tests]. A similar result was observed on two occasions with W7. Although these results were not what we initially expected, recent independent results (Golowasch et al. 2017) suggest an interesting mechanism by which this could occur (see discussion).

DISCUSSION

In this study we set out to dissect the mechanism by which the proctolin receptor activates its effector, the pacemaker current I_MI. Proctolin is known not only as an activator of the pacemaker current of the system (Bose et al. 2014; Zhao et al. 2010) but also as a regulator of the correlated expression of a number of voltage-gated currents other than I_MI (Khorkova and Golowasch 2007). Such correlated expression of multiple ionic currents in neurons is thought to be to be key to the generation of stable neuronal activity features (Bergquist et al. 2010; Golowasch 2015; Hudson and Prinz 2010; MacLean et al. 2005; Norris et al. 2011; O’Leary and Marder 2016; Zhao and Golowasch 2012). Thus a complete characterization of the proctolin signaling pathway is an important step in understanding the homeostatic regulation of activity in a central pattern generator such as the pyloric network, or neuronal networks in general.

We have shown that proctolin-induced I_MI does indeed appear to be mediated by a G protein and is, therefore, most likely a G protein-coupled receptor (GPCR). This is consistent with recent findings that a proctolin receptor with the structural hallmarks of a GPCR in the lobster H. americanus (predicted from its transcriptome; Christie et al. 2015) and of a CCAP receptor with similar structural properties in C. borealis (predicted from its transcriptome; Garcia et al. 2015). However, unexpectedly, and in contrast with most known GPCR-mediated effects, extracellular calcium appeared to affect the G protein utilized by the proctolin receptor, as demonstrated by the finding that the sensitivity of proctolin-induced I_MI amplitude to pertussis toxin appears to be modulated by extracellular calcium. Additionally, we have shown that I_MI activation is sensitive to agents that modulate intracellular calcium, calmodulin, and calmodulin-activated proteins. Furthermore, the dual action of ryanodine receptor antagonists, calmodulin inhibitors, and calmodulin-activated kinase inhibitors on I_MI suggests that the relationship between I_MI activation and its voltage dependence is not as clearly separable as we initially hypothesized.

I_MI activation does not depend on PLC, cyclic nucleotides or tyrosine kinases.

Despite the apparent involvement of intracellular calcium and calmodulin (presumably activated by intracellular calcium), we find little evidence to support any role on I_MI activation for phospholipase C (PLC), normally an enzyme regulated by GPCRs and involved in the regulation of intracellular calcium levels (Kadamur and Ross 2013). Similarly, no evidence was found for involvement of cyclic nucleotides or tyrosine kinases.

In this study, we found a large role for calmodulin and intracellular calcium that is consistent with previous studies (Gray and Golowasch 2016; Swensen and Marder 2000). These results set up the expectation that PLC, which typically increases intracellular calcium release through inositol 3,4,5-trisphosphate (IP₃) receptors (Portilla and Morrison 1986), could provide a likely mechanism by which the proctolin receptor is activated. Additionally, both proctolin (Baines et al. 1996; Groome and Watson 1989; Johnson et al. 2003b; Lange 1988; Mazzocco-Manneval et al. 1998; Philipp et al. 2006; Vezenkov and Danalev 2009) and muscarinic agonists (Heinrich et al. 2001; Trimmer 1995) have been associated with PLC signaling. Another reason we examined PLC signaling was the prevalence in vertebrates with similar voltage-dependent neuromodulator-activated cation currents being mediated by PLC signaling (Helliwell and Large 1997; Oh et al. 2003; Strübing et al. 2001). Additionally, many of the same neuromodulators that elicit I_MI have been shown to activate PLC pathways in other invertebrate preparations (proctolin: Baines et al. 1996; Groome and Watson 1989; Lange 1988; Mazzocco-Manneval et al. 1998; Philipp et al. 2006; Vezenkov and Danalev 2009; muscarinic: Heinrich et al. 2001; Trimmer 1995). Despite these expectations, neither neomycin nor edelfosine was capable of altering I_MI amplitude (Table 2). These results are consistent with the findings reported by Swensen and Marder (2000), who found that the potent PLC inhibitor U-73122 was also ineffective at modulating I_MI. Similarly, and also consistent with Swensen and Marder (2000), we found no evidence of cyclic nucleotides, phosphatases, or tyrosine kinases (Table 2) being involved in I_MI activation. The role for serine/threonine kinases, however, is somewhat less certain, and is discussed below.

I_MI activation is modulated by intracellular calcium and targets of calmodulin: a tentative mechanism of I_MI activation and amplification by calcium signaling.

We found that I_MI amplitude was significantly altered by the calmodulin inhibitors W7 and CDZ, the ryanodine receptor antagonist dantrolene, the CaMKII inhibitor KN-93, and the MLCK inhibitor ML-7. Because MLCK and KN-93 are both calmodulin-activated kinases, and these other compounds target either pathways upstream or downstream of calmodulin, a central role is suggested for calmodulin and calcium signaling in I_MI activation (Fig. 8). These results suggest that, at the very least, calcium-dependent proteins, likely including calmodulin-activated proteins, play a role in proctolin-induced I_MI activation. As reported previously, extended periods in low extracellular calcium tend to reduce I_MI amplitude (Gray and Golowasch 2016). This slow reduction in I_MI amplitude is consistent with the involvement of intracellular calcium in I_MI signaling. Interestingly, in “on” bipolar cells of the dogfish retina, intracellular calcium chelation or application of a CaMKII inhibitor reduces the voltage dependence of a cGMP/light-activated current (Shiells and Falk 2001).

Fig. 8.

Proposed mechanism of activation for proctolin-induced I_MI. According to this model, I_MI channels are activated by a neuropeptide GPCR using a pertussis toxin-sensitive pathway that is independent from a different GPCR-mediated voltage-dependence pathway (not shown; cf. Gray and Golowasch 2016). Calmodulin (CaM) via calmodulin-dependent kinases CamKII and MLCK, activated by exogenous calcium entering the cell via either the I_MI channels themselves (shown) or voltage-dependent calcium channels (not shown), amplifies the activation. This is aided by a further increase in intracellular calcium level due to calcium-induced calcium release mediated by RyRs. Blunt-ended lines show agents that inhibit the indicated paths; arrows indicate activating agents and pathways. ER, endoplasmic reticulum; ProcR, proctolin receptor; RyR ryanodine receptor.

Specifically, we propose the following working model: proctolin-induced I_MI is activated by a pertussis toxin-sensitive G protein, in turn coupled to peptide and likely also muscarinic Ach receptors (Swensen and Marder 2000). It appears that, in addition to monovalent cations (Golowasch and Marder 1992b), the I_MI channel may be partially permeable to calcium (see below). These calcium ions, and perhaps also additional calcium ions released from intracellular stores via calcium-induced calcium release, act on calcium-binding proteins such as calmodulin and calmodulin-binding proteins to amplify the proctolin-induced I_MI response (Fig. 8). In this model, the proctolin receptor (and likely other receptors such as CCAP and muscarinic ACh receptors) are GPCRs that activate I_MI via an unidentified G protein whose pertussis toxin sensitivity is modulated by external calcium concentration (see below).

Our suggestion that an elevated intracellular calcium level activates calmodulin, perhaps aided by calcium-induced calcium release, is supported by the finding that the calmodulin inhibitors W7 and CDZ both inhibit I_MI amplitude. Calmodulin, in turn, activates MLCK and CaMKII. These calmodulin-activated kinases may amplify I_MI activation, which is supported by our findings that ML-7 and KN-93 both reduce I_MI amplitude. We also suggest that calcium-sensing receptor (CaSR; not shown in Fig. 8) may play an enabling role in I_MI activation because 1) proctolin-induced I_MI amplitude decreases faster in low extracellular calcium than in normal calcium (Gray and Golowasch 2016), and 2) whereas CaSR agonists augment I_MI amplitude, CaSR antagonists do not reduce I_MI amplitude (Gray and Golowasch 2016), as would be expected if CaSR played a direct role in I_MI activation. This suggests that CaSR signaling may enhance I_MI but is not necessary to activate it.

Our analysis is based on manipulations of the extracellular calcium concentration under the assumption that when the extracellular concentration is low (i.e., 2 mM), no significant changes in intracellular calcium concentration occur. We think that intracellular calcium is significantly affected only in the long term for two reasons. First, at 2 mM extracellular calcium, synaptic input to the LP neuron from the pyloric dilator and anterior burster (pacemaker) neurons is not abolished (data not shown), suggesting that calcium influx of some type is occurring via either Cd²⁺-insensitive calcium channels (cf. Johnson et al. 2003a) or by release from intracellular stores. Second, in Gray and Golowasch (2016), we reported that extended periods in low calcium reduce I_MI amplitude. However, significant I_MI reduction is not observed until ~4 h in low-calcium solution. Because our measurements were never performed after 4 h in low-calcium solutions, we believe that no significant intracellular calcium changes occur in our low-calcium saline solution.

Is I_MI permeable to calcium?

Our model requires intracellular calcium-induced calcium release, which requires an initial source of calcium that could come from intracellular or extracellular sources. Zhao et al. (2011) have suggested that I_MI could be permeable to calcium itself. This would provide a plausible source for the initial calcium required for activating dantrolene-sensitive ryanodine receptors. It seems more likely that the initial calcium originates from an external rather than canonical internal source, such as PLC-induced intracellular calcium release, because we saw no modulation of proctolin-induced I_MI by PLC inhibitors (Table 2).

Another possible path for extracellular calcium influx is voltage-gated calcium channels. However, our experiments were conducted in the presence of 200 µM Cd²⁺, a concentration previously reported to block the majority of “observable” voltage-gated calcium currents in LP cells (Golowasch and Marder 1992a). Nevertheless, a calcium-imaging study in lobster STG neurons showed that 100 μM Cd²⁺ only reduces calcium peak activity by ~30% (Levi et al. 2003). Therefore, we cannot at this point discard the alternative hypothesis that influx through incompletely blocked voltage-gated calcium channels may trigger intracellular calcium-induced calcium release.

Proctolin-induced I_MI activation is modulated by G proteins whose pertussis toxin sensitivity is modulated by extracellular calcium.

We found that proctolin-induced I_MI is modulated by a G protein whose sensitivity to pertussis toxin appears to be modulated by extracellular calcium, with stronger inhibition at low than at normal (high) extracellular calcium levels (Fig. 3). Although unexpected, these calcium-dependent effects are not entirely unprecedented. Parathyroid hormone 1 receptors, for example, appear to use calcium as a weak agonist that can be switched to inverse agonist depending on extracellular calcium levels (Mitra et al. 2013). In addition, a large number of calcium-binding proteins regulate G protein receptor kinases (Sallese et al. 2000). Interestingly, our finding that okadaic acid may modulate I_MI in normal calcium, but not low calcium, may suggest that I_MI signaling is fundamentally altered in low vs. normal concentrations of extracellular calcium.

A possible explanation for the extracellular calcium modulation of pertussis toxin sensitivity derives from the findings that GPCR coupling to G proteins is more promiscuous than once thought (Gilman 1987; Knollman and Conn 2008; Lane et al. 2007; Wong and Ross 1994) and from the relatively recent finding that many regulators of G protein signaling (RGS) are inhibited by activated calmodulin (Popov et al. 2000). However, although the promiscuity of some Gα subunits, particularly of the Gα_i and Gα_o types, is not uncommon (Gilman 1987; Knollman and Conn 2008; Lane et al. 2007; Wong and Ross 1994), there are comparatively fewer available examples of this effect across pertussis toxin-sensitive (i.e., Gα_s and Gα_t) or pertussis-insensitive (i.e., Gα_q) types (Katada 2012; Knollman and Conn 2008; Lane et al. 2007; Offermanns and Simon 1995; Wong and Ross 1994). This important aspect of I_MI regulation deserves further in depth examination. A speculative model that may explain our observations of calcium-dependence of G protein signaling in I_MI activation could become the basis of this exploration: according to this model, two simultaneous I_MI activation pathways would exist. One pathway, which is pertussis toxin insensitive, would be active in normal calcium and would be amplified by calcium and calmodulin. A second, pertussis toxin-sensitive pathway, would dominate at low extracellular calcium levels. In normal calcium, a protein such as RGS10 (Popov et al. 2000), which undergoes inhibition by activated calmodulin and shows affinity for pertussis toxin-sensitive G protein (cf. Hepler 1999), would keep the second I_MI pathway relatively inactive. When extracellular calcium is decreased, activated calmodulin inhibition of the second pathway is relieved due to lower calcium influx, increasing the relative contribution of the second pathway to activation of I_MI. This would be consistent with our observation that the time course of I_MI peak after washout of the neuromodulator seems slower in normal calcium (I_MI is activated long after neuromodulator wash) relative to low calcium (I_MI fades very quickly after the neuromodulator is washed out; data not shown), which could be due to CaMKII activation in normal calcium, for example.

Separation of voltage dependence and activation mechanisms.

In this and our previous study (Gray and Golowasch 2016), we targeted a number of signal transduction processes associated with I_MI voltage dependence, and unexpected effects on I_MI activation were observed. One example of this was the apparent enabling role for the CaSR on I_MI activation. Although we propose that CaSR signaling enhances I_MI, it is not necessary for it. This was due to the apparent incongruence between the findings of the CaSR agonist R568 and the antagonist NPS-2143. Although NPS-2143 modulates I_MI voltage dependence (Gray and Golowasch 2016), it does so without altering I_MI amplitude. R568 increases both I_MI voltage dependence and I_MI amplitude in an extracellular calcium-dependent way (more so in low calcium than in normal calcium). This suggests that there could be more cross talk between these pathways than we had originally predicted. Another example of this was the finding that the MLCK inhibitor ML-7 and the ryanodine antagonist dantrolene both reduced I_MI voltage dependence (Gray and Golowasch 2016) and reduced I_MI amplitude. Although any inhibitor would be expected to significantly increase I_MI slope up to a slope of zero, an inhibitor would stop there and not increase slope further (e.g., GDPβS in normal calcium did exactly this; see Fig. 3). However, all three of these agents increased slope above zero, and the increase in I_MI slope observed with R568 is the opposite of the predicted direction of pure augmentation, because slope should become more negative with increasing current size when voltage dependence is held constant. These findings for ML-7, CDZ, dantrolene, and R568 are, therefore, suggestive that a relation between I_MI voltage dependence and activation exists. In light of these findings and the similar finding that the CaMKII inhibitor KN-93 has an effect on both I_MI activation and voltage dependence, it appears that I_MI activation and voltage dependence cannot be fully separated in these cases.

Why does modifying I_MI signaling pathways not affect neuronal activity?

We were initially surprised to observe no significant effect of CDZ or W7 on the frequency of the pyloric rhythm. Related work performed as part of a separate project (Golowasch et al. 2017), however, may provide an explanation. When injection of I_MI into the pacemaker neurons (PD neurons) in a decentralized (and thus not oscillating) pyloric network using the dynamic-clamp technique activates the pyloric rhythm (cf. Bose et al. 2014; Zhao et al. 2010) but very little dependence of the pyloric rhythm frequency on the amplitude of I_MI is observed (Fig. 2B in Golowasch et al. 2017), the rhythm appears to be either fully activated or not at all. In our present experiments we attempted to modify I_MI in all cells in the network with CDZ, either in pyloric networks activated by the full complement of naturally released neuromodulators (not decentralized) or in decentralized preparations with a 0.1 μM concentration of exogenously applied proctolin. In both cases, it is likely that the system is strongly activated and that I_MI is not depressed enough by CDZ (see Fig. 4) to turn the rhythmic activity off. Thus no change in pyloric frequency is precisely what one would expect to observe. It is also possible that CDZ affects I_MI differently in the pacemaker PD neurons than in the LP neurons studied in this work, thus explaining our observations. Therefore, we conclude that until a proper blocker of the ion channels that carry I_MI is discovered, this test is unlikely to be successful.

In conclusion, we find that the I_MI current, which acts as the pacemaker current in the pyloric network (Bose et al. 2014; Zhao et al. 2010), appears to be activated by a GPCR via a putatively calcium-dependent, pertussis toxin-sensitive G protein and is enhanced by intracellular calmodulin-dependent kinases such as CaMKII and MLCK. A full characterization of the entire signaling pathway requires a more in-depth examination, using more targeted and more specific molecular tools to manipulate individual pathway components that are not yet available. It will then be interesting to ask if the same pathway that activates I_MI is involved in the neuromodulator-dependent regulation of the correlated expression of other voltage-gated ionic currents in these cells (Khorkova and Golowasch 2007). Because the regulation of the correlated expression of ionic currents in neurons appears to be key to the maintenance of overall stable features of cellular and network activity (Bergquist et al. 2010; Golowasch 2015; Hudson and Prinz 2010; MacLean et al. 2005; Norris et al. 2011; O’Leary and Marder 2016; Zhao and Golowasch 2012), understanding the signaling pathways involved is important.

GRANTS

This work was supported by National Institutes of Health Grants R01 MH64711 and R56 NS085330 (to J. Golowasch) and by a Fellowship from the Behavioral and Neural Science Graduate Program, Rutgers University-Newark (to M. Gray).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

M.G. and J.G. conceived and designed research; M.G. and D.H.D. performed experiments; M.G. and D.H.D. analyzed data; M.G. and J.G. interpreted results of experiments; M.G. and J.G. prepared figures; M.G. and J.G. drafted manuscript; M.G. and J.G. edited and revised manuscript; M.G., D.H.D., and J.G. approved final version of manuscript.