TRPM channels modulate epileptic-like convulsions via systemic ion homeostasis

Summary

Neuronal networks operate over a wide range of activity levels, with both neuronal and non-neuronal cells contributing to the balance of excitation and inhibition. Activity imbalance within neuronal networks underlies many neurological diseases, such as epilepsy [1]. The C. elegans locomotor circuit operates via coordinated activity of cholinergic excitatory and GABAergic inhibitory transmission [2]. We have previously shown that a gain-of-function mutation in a neuronal acetylcholine receptor, acr-2(gf), causes an epileptic-like convulsion behavior [3]. Here, we report that the behavioral and physiological effects of acr-2(gf) require the activity of the TRPM channel GTL-2 in non-neuronal tissues. Loss of gtl-2 function does not affect baseline synaptic transmission, yet can compensate for the excitation-inhibition imbalance caused by acr-2(gf). The compensatory effects of removing gtl-2 are counter-balanced by another TRPM channel GTL-1, and can be recapitulated by acute treatment with divalent cation chelators, including those specific for Zn²⁺. Together these data reveal an important role for ion homeostasis in the balance of neuronal network activity and a novel function of non-neuronal TRPM channels in the fine-tuning of this network activity.

Results

GTL-2 acts in non-neuronal tissues to modulate convulsive behaviors caused by acr-2(gf)

In the wild type C. elegans locomotor circuit, cholinergic innervation excites muscles to contract alternately on the ventral or dorsal side, while simultaneously activating GABAergic cross-inhibition to relax muscles on the opposite side [2]. A heteromeric acetylcholine receptor, ACR-2R, expressed in the cholinergic neurons plays a key role in the locomotor network [3]. A Valine to Methionine substitution in the transmembrane TM2 domain of the ACR-2 non-alpha subunit increases channel current and causes a gain-of-function effect in the locomotor circuit, manifested as overexcitation accompanied with decreased inhibition. The acr-2(gf) animals display a spontaneous convulsive behavior (Figure 1, supplemental movie 1). Similar amino acid substitutions in mammalian neuronal acetylcholine receptors are associated with epilepsy [4].

Figure 1. gtl-2 acts in non-neuronal tissues to modulate acr-2(gf) convulsions.

(A) Images of wild type, acr-2(gf) and gtl-2(lf);acr-2(gf) locomotion. (B) Quantification of convulsion frequency in mutants of genotype as indicated. **, p<0.01, compared to acr-2(gf), ANOVA and Dunnett’s post hoc test. (C) Transgenic rescue of the suppression activities in gtl-2(0);acr-2(gf) and gtl-2(lf);acr-2(gf). ***, p<0.001, ANOVA and Bonferroni post hoc test. (D) Confocal image of anti-GFP immunostaining of GTL-2::GFP in an adult shows expression in the excretory cell (arrowhead) and epidermis (arrow). (E) Confocal image of anti-GFP immunostaining of GTL-2::GFP in a L4 larva expressing Pdpy-7:gtl-2cDNA::gfp. (F) Cell specific rescue of gtl-2 shows requirement in epidermis and excretory cell. Cell types are as follows: Pdpy-30 – all cells, Prgef-1 – pan-neuronal, Pmyo-3 – body muscles, Psulp-4 – excretory cell, and Pdpy-7 – epidermis. **, p<0.01, compared to gtl-2(lf);acr-2(gf), ANOVA and Dunnett’s post hoc test.

To search for novel regulators involved in the acr-2(gf) convulsive phenotype we isolated and characterized a genetic suppressor, n2618 (see supplemental methods, Table S1). Double mutants of n2618;acr-2(gf) show grossly wild type locomotion and a near complete elimination of spontaneous convulsions (Figures 1A, B, supplemental movie 2). By genetic mapping and whole-genome sequencing we found that n2618 animals had a single amino acid substitution in a conserved domain of the protein GTL-2, which is a member of the TRPM (for Melastatin family of Transient Receptor Potential) channels (Figure S1) [5]. TRPM channels are multifunctional non-selective cation channels. Those closely related to GTL-2 generally play a role in ion homeostasis in a tissue and organism-specific manner [6–10]. C. elegans has three TRPM channels GON-2, GTL-1 and GTL-2, which have previously been implicated in electrolyte Mg²⁺ homeostasis and intestinal Ca²⁺ oscillations [9, 10]. Presumed gtl-2 null mutations caused similar suppression of acr-2(gf), and failed to complement n2618 for the suppression activity (Figure 1B). In addition, the suppression effect of gtl-2(null) or gtl-2(n2618) on acr-2(gf) was fully rescued by transgenic expression of the genomic gtl-2 DNA (Figure 1C). We conclude that inactivation of gtl-2 antagonizes the locomotor circuit imbalance caused by acr-2(gf).

Other genetic suppressors of acr-2(gf) define genes required for the activity of the ACR-2 channel and which act in cholinergic neurons [3]. To determine in which tissues gtl-2 function was required for suppression, we generated a GTL-2::GFP fusion construct that completely rescued the suppression effects of gtl-2(lf) on acr-2(gf) (Figure 1C, Table S1, 2). We observed GTL-2::GFP on the epidermal cell surface (Figure 1D, E), as well as in the excretory cell and pharynx, as reported [9], but not in neuronal and muscle cells associated with locomotion. This expression pattern suggests that gtl-2 may influence the acr-2(gf) neuronal phenotype through non-neuronal means. To test this hypothesis we performed cell specific rescue experiments with a gtl-2cDNA::gfp construct (Table S2). Expression of gtl-2 in either the epidermis or the excretory cell, but not in neurons or muscles, showed a rescuing activity comparable to that of gtl-2 expressed ubiquitously or under the control of its own promoter (Figure 1F). The excretory cell and the epidermis are connected by gap junctions [11]. The ability of either excretory cell or epidermal expression of GTL-2 to rescue is consistent with functional coupling of these tissues and with our evidence below that GTL-2 regulates neuronal excitability via extracellular ion homeostasis.

GTL-2 specifically affects the activity imbalance caused by acr-2(gf)

To investigate how non-neuronal GTL-2 might affect motor neuron activity, we first analyzed neuronal anatomy and physiology in gtl-2 mutants. Overall synapse and neuronal morphology were grossly normal in these mutants (Figure S2A). gtl-2 mutants displayed normal sensitivity to the acetylcholinesterase inhibitor Aldicarb [12] (Figure 2A), and exhibited wild type locomotion (supplemental movie 3). Electrophysiological recordings of dissected body-wall muscle preparations showed that the frequencies and amplitudes of endogenous cholinergic and GABAergic activity in gtl-2(lf) was similar to that of wild type animals across a range of Ca²⁺ concentrations (Figures 2B, S2D). The expression levels of the postsynaptic GABA receptor UNC-49 and the response to exogenous GABA were also similar to those in wild type animals (Figures S2B,C). Thus, GTL-2 is not essential for most aspects of neuronal and muscle development or physiology.

Figure 2. Physiology of gtl-2(lf) and its effects on acr-2(gf).

(A) Paralysis response to 500 µM aldicarb in gtl-2(lf) compared to wild type. (B) Representative traces and frequencies of endogenous acetylcholine and GABA postsynaptic currents from wild type (n=25), gtl-2(lf) (n=12), acr-2(gf) (n=14), and gtl-2(lf);acr-2(gf) (n=12) animals in 1 mM external CaCl₂ bath solution. Error bars indicate SEM. (C) Paralysis response to 200 µM aldicarb in wild type, acr-2(gf), and gtl-2(0);acr-2(gf) animals. *, p<0.05, compared to acr-2(gf), two-way ANOVA and Bonferroni post hoc tests. (D) Summary of frequencies of endogenous acetylcholine (left) and GABA (right) release in CaCl₂ bath solutions of 0.1 mM (wild type (n=14), acr-2(gf) (n=9), and gtl-2(lf);acr-2(gf) (n=10)), 0.5 mM (wild type (n=12), acr-2(gf) (n=14), and gtl-2(lf);acr-2(gf) (n=11)), and 1 mM (wild type (n=25), acr-2(gf) (n=14), and gtl-2(lf);acr-2(gf) (n=12)). Error bars indicate SEM. Statistics in B, D used SigmaStat 3.5 (Aspire Software International): *, p<0.05, ***, p<0.001 by Student’s t-test, or the Mann-Whitney rank sum test for GABA data in 0.5 mM CaCl₂ according to the normality of datum distribution for data.

We next addressed whether the behavioral suppression of acr-2(gf) by gtl-2(lf) could be accounted for by restoration of the excitation and inhibition balance in the locomotor circuit. acr-2(gf) animals exhibit hypersensitivity to aldicarb, in part due to increased acetylcholine release [3]; and gtl-2(lf) partially suppressed this aldicarb hypersensitivity (Figure 2C). We further investigated this effect by electrophysiological recordings of neuromuscular junctions (NMJ). We previously observed that acr-2(gf) disrupts the correlated activities between cholinergic and GABAergic transmission in a Ca²⁺ dependent manner [3]. Here, using a modified recording procedure [13], we recorded over a range of extracellular Ca²⁺ concentrations (see supplemental methods). acr-2(gf) animals at 0.1 mM extracellular Ca²⁺ showed an increase in both cholinergic and GABAergic transmission, compared to wild type animals (Figure 2D). This observation indicates that the GABAergic neurons in acr-2(gf) animals are healthy and capable of responding to synaptic inputs as predicted by their anatomical connectivity [2]. As the extracellular Ca²⁺ concentration was increased, wild type animals showed increases in both cholinergic and GABAergic activities. By contrast, in acr-2(gf) mutants, while the cholinergic activity continued to increase, the GABAergic activity began to decrease (Figure 2D), consistent with previous findings [3]. Thus, under physiological calcium levels over 1 mM [14, 15], the increased cholinergic activity caused by the acr-2(gf) mutation is accompanied by an anti-homeostatic inhibition of GABAergic neuron activity (Figure 2B), leading to a net over-excitation of the locomotor circuit. Strikingly, the gtl-2(lf) mutation not only enhanced GABA release in acr-2(gf), but also restored the Ca²⁺ dependent increase of GABAergic transmission, resulting in correlated cholinergic excitation and GABAergic inhibition in the acr-2(gf);gtl-2(lf) double mutant (Figures 2B, D). In these preparations the tissues expressing gtl-2 were disrupted, therefore the observed physiological effects most likely reflect long-lasting changes in GABAergic presynaptic release as the result of reducing gtl-2 function.

Although the recordings from the dissociated NMJ preparations did not detect significant effects of gtl-2(lf) on the cholinergic activities of acr-2(gf), we wondered whether gtl-2 might also modulate cholinergic neurons. To address this, we generated gtl-2(lf);acr-2(gf);unc-25/GAD(0) triple mutant animals in which GABAergic transmission was completely absent due to a null mutation in the glutamic acid decarboxylase UNC-25 [16]. These animals showed significant suppression of convulsions, compared to acr-2(gf);unc-25/GAD(0) (Figure 3A), indicating that loss of gtl-2 activity can alleviate hyperexcitation defects of acr-2(gf) in the absence of GABA. These observed in vivo effects of gtl-2(lf) may reflect acute changes on cholinergic transmission that are not evident in NMJ recordings due to the dissection of GTL-2 expressing tissues and consequent disruption of the extracellular milieu. It is also possible that GTL-2 might also influence ACR-2(gf) independent of both GABAergic and cholinergic activities. To assess whether gtl-2 modulated other types of neuronal over-excitation, we tested the ability of gtl-2(lf) to suppress mutants with increased cholinergic transmission, such as the G_oα protein mutant goa-1(lf) [17]. Unlike acr-2(gf), loss of gtl-2 function had no effect on locomotion nor on aldicarb sensitivity of goa-1(lf) mutants (Figures 3B, C). These analyses indicate that loss of gtl-2 function specifically modulates the anti-homeostatic defects in acr-2(gf).

Figure 3. gtl-2 specifically influences the overexcitation phenotype in acr-2(gf), but not of goa-1(lf) mutants with increased cholinergic transmission.

(A) Quantification of convulsion frequency in mutants of genotypes indicated. -, no mutation; x, has mutation. ***, p<0.001, compared to gtl-2(lf);acr-2(gf), ANOVA and Bonferroni post hoc test. (B) Images of animals of genotype indicated. gtl-2(lf) does not alter locomotion behavior of goa-1(ep275) animals. (C) Paralysis response to 200 µM aldicarb in animals of genotypes indicated. n=3 trials of 10 animals per genotype.

Manipulating ion homeostasis recapitulates the effects of gtl-2(lf)

GTL-2 has been reported to regulate fluid Mg²⁺ homeostasis, in conjunction with the other two C. elegans TRPM channels, GTL-1 and GON-2 [9, 18]. Specifically, GTL-1 and GON-2 function in the intestine to take up Mg²⁺, whereas GTL-2 acts in the excretory cell to excrete Mg²⁺. gtl-2(lf) mutants have increased systemic Mg²⁺, but also reduced systemic Ca²⁺ levels, suggesting other roles for GTL-2 in addition to Mg²⁺ export [9]. To address the basis of gtl-2 action on the locomotor circuit, we first investigated the roles of these other TRPM channels on acr-2(gf). Loss of gtl-1 function had no effect on acr-2(gf) alone, but completely rescued the suppression of convulsions in the gtl-2(lf);acr-2(gf) background (Figure 4A). Moreover, RNAi knockdown of gtl-1 in gtl-2(lf);acr-2(gf) animals in a genetic background in which the neurons are refractory to RNAi also resulted in rescuing of the suppression activity of gtl-2(lf) (Figure 4A), suggesting GTL-1 may also function in non-neuronal cells to modulate gtl-2 suppression of acr-2(gf). Unexpectedly, loss of gon-2 function attenuated the convulsion frequency of acr-2(gf), similar to mutating gtl-2 (Figure 4A). This observation suggests that the increased Mg²⁺ in gtl-2(lf) may not be directly responsible for the suppression effect on acr-2(gf). To test this, we cultured animals using plates containing varying Mg²⁺ concentration. While increasing Mg²⁺ had a significant effect on brood size in gtl-2 mutants (Figure S3A), consistent with the previous observations [9], it had no effect on acr-2(gf) convulsions nor on the suppression of acr-2(gf) by gtl-2(lf) (Figure S3B). Soaking animals in high concentrations of Mg²⁺ similarly failed to influence acr-2(gf) convulsions. We also performed NMJ recordings using extracellular bath solutions with varying Mg²⁺, but did not observe any significant differences on cholinergic and GABAergic activities in either wild type or acr-2(gf) (Figure S3C,D). Thus, the increased Mg²⁺ levels in gtl-2(lf) mutants do not account for the effects on the locomotor circuit. Instead, other divalent cations might be involved in suppression of excitation-inhibition imbalance.

Figure 4. Ion homeostasis plays a key role in the acr-2(gf) phenotype.

(A) Quantification of convulsion frequency of acr-2(gf) and gtl-2(lf);acr-2(gf) with genetic mutations in gon-2 and gtl-1 or treated with gtl-1 RNAi to reduce the function of different TRPM channels. *** = p<0.001, ANOVA and Bonferroni post hoc test. (B) Images of acr-2(gf) animals soaked in either M9 (control), EDTA or TPEN. (C) Convulsion frequency of acr-2(gf) and gtl-2(lf);acr-2(gf) soaked in the cation chelators EDTA (75 mM), EGTA (75 mM), DTPA (75mM), and TPEN (100 µM). Statistics in A, C, *** = p<0.001, ANOVA and Bonferroni post hoc test. (D) Model of the regulation of locomotor circuit by ion homeostasis and the three TRPM channels. GON-2 and GTL-1 act in the intestine (I) to allow divalent cation influx, whereas GTL-2 acts in the excretory cell (EC) for cation efflux [9](and this study). Neuronal activity (MN) likely influences local cation levels (dark blue oval). Our data suggest that both the systemic cation fluctuations due to the function of the three TRPM channels as well as local ion fluctuations involving GTL-2 in the epidermis (E) modulate the excitability of the locomotor circuit, hence contractions of muscles (M).

To further address the involvement of ion homeostasis in modulating the excitation-inhibition imbalance caused by acr-2(gf), we soaked acr-2(gf) and gtl-2(lf);acr-2(gf) animals in multiple cation chelators (Figure 4B,C). Bathing acr-2(gf) animals in the divalent cation chelators ethylenediaminetetraacetic acid (EDTA) resulted in significant suppression of convulsions and improved locomotor ability (supplemental movie 4). However, EDTA had no effects on gtl-2(lf);acr-2(gf), consistent with the above observation of culturing acr-2(gf) on high concentration Mg²⁺ plates and gtl-2(lf);acr-2(gf) on plates with no Mg²⁺ added. Treatment with ethylene glycol tetraacetic acid (EGTA) had no effect on acr-2(gf) single or gtl-2(lf);acr-2(gf) double mutants. As EGTA has relatively high specificity to Ca²⁺ [19], these results suggest that reducing extracellular cations, but not necessarily Ca²⁺, specifically mimicked the effects of reducing gtl-2 activity. As some TRPM channels show a high level of conductance for trace metals, particularly Zn²⁺ [20], we tested the heavy metal specific chelators dietylenetriaminepentaacetic acid (DTPA), a membrane impermeant chelator [21], and tetrakis(2-pyridylmethyl)ethlyenediamine (TPEN), a membrane permeant chelator [22], and found that both suppressed the acr-2(gf) phenotype comparable to EDTA treatment (Figures 4B,C). All together, these findings demonstrate that divalent cationic balance of heavy metals is critical for the expression of the convulsive behavior caused by acr-2(gf).

Discussion

Epilepsy is one of the most common neurological diseases today, affecting approximately one percent of the population [23]. Common to all forms of epilepsy is an imbalance of excitation and inhibition [1]. A distinct physiological feature of the acr-2(gf) mutation is that while increasing cholinergic excitation, it creates an anti-homeostatic decrease in GABAergic activity cell non-autonomously [3] (this study). How are TRPM channels, GTL-2, GON-2 and GTL-1 influencing this neuronal activity imbalance? Our data suggest two possible mechanisms that are not mutually exclusive, one involving local action of GTL-2 and the other involving systemic ion homeostasis (Figure 4D). Local ionic composition fluctuates as the result of neuronal activity. The epidermis is adjacent to motor neurons and their axons [2]. As GTL-2 is expressed in the epidermis, it could influence this local ionic balance (Figure 4D). The role of GTL-2 and the epidermis may be analogous to that of glial cells in the mammalian nervous system. Glial cells maintain normal levels of network activity through their roles in K⁺ buffering, glutamate uptake, and glia specific release of neurotransmitter, and have been implicated in epilepsy [24]. A number of TRPM channels are highly expressed in the brain, and TRPM1 and TRPM7 have been shown to play cell autonomous roles in neuronal function [25–28]. However, non-cell autonomous roles of TRPM channels in the nervous system remain to be examined.

GON-2 and GTL-1 function in fluid ion homeostasis through ion uptake in the intestine [9]. Thus, their effects on acr-2(gf) indicate a role of systemic ion regulation on neuronal function (Figure 4D). Although the three C. elegans TRPM channels have been linked to Mg²⁺ homeostasis [9] based on their effects on fertility and growth rate, our results suggest they influence neuronal excitability independent of Mg²⁺. In gtl-2(lf) animals, systemic Mg²⁺ levels are increased. However, chelation of divalent cations, including Mg²⁺, suppressed acr-2(gf), and did not reverse the suppression effects of gtl-2(lf) on acr-2(gf). Instead, treatment using heavy metal chelators DTPA and TPEN mimicked loss of gtl-2 function. These observations indicate that the action of these TRP channels influences cation homeostasis beyond simple import and export of Mg²⁺. In fact, Teramato et al. also showed Ca²⁺ levels were decreased in gtl-2(lf) mutants, which they attributed to the inhibition of GON-2 by high Mg²⁺ [9]. This interpretation could explain our observation that loss of GON-2 partially suppressed acr-2(gf) in a manner similar to gtl-2(lf).

Studies of several TRPM channels in other organisms, such as TRPM7 channels, show that they can be permeable to multiple divalent ions, including the heavy metal Zn²⁺ [20]. Additionally, independent studies on the Drosophila TRPM channel reported changes in larval Zn²⁺ and Mg²⁺ homeostasis [6, 7]. Interestingly, altered Zn²⁺ levels have been reported in epilepsy patients, and it is thought that disturbed Zn²⁺ homeostasis can be proconvulsant, whereas Zn²⁺ can act as an anticonvulsant when its homeostasis is maintained [29]. Manipulation of Zn²⁺ can also potentiate acetylcholine receptors and inhibit certain GABA receptors [30, 31]. We have shown here that altering homeostasis of heavy metals including Zn²⁺ by chelation has an anticonvulsant effect. We speculate that altered neuronal activity by ACR-2(gf) could cause local ionic imbalance, possibly involving Zn²⁺. Conversely, alteration in systemic Zn levels may directly affect the activity of acr-2(gf), restoring neuronal network balance.

Current treatments for epilepsy focus on normalizing the neuronal network excitation-inhibition imbalance by targeting mostly neuronal proteins [32]. Yet, many epilepsy patients are refractory to pharmacological treatments [33], prompting an increased effort to search for epilepsy treatments that go beyond direct effects on neurotransmission. One long-standing treatment that has proven effective in epilepsy patients who are refractory to traditional medications is the ketogenic diet [34]. Systemic metabolic changes can have long-lasting effects on neurotransmission [35]. Our study reveals a critical role for ion homeostasis in balancing neuronal excitation and inhibition in a network, as well as introducing TRPM channels as novel molecular players in this pathway.

Experimental Procedures

Genetics, molecular and transgenic analysis

C. elegans strains were grown on NGM plates at room temperature (20–22°C) as described [36]. Isolation and cloning of n2618 were detailed in supplemental methods. Double mutants were constructed using standard procedures, and genotypes were confirmed by allele sequence determination (Table S1). Molecular biology, transgenic analysis and RNAi were detailed in supplemental methods. Table S2 lists the information of all constructs.

Behavior, Pharmacology and Chelation analysis

Convulsion was quantified as detailed in supplemental methods [3]. All drugs were from Sigma-Aldrich, and manipulations were performed according to published procedures [3, 12, 38]. For chelator soaking assays L4 larvae were put into 200 µl of M9 solution containing the chosen chelator and concentrated bacteria. For each chelator, a range of concentration was tested, and the concentration that did not cause noticeable sickness to animals was used in the final quantification analysis. Soaking was performed overnight at room temperature. The treated animals were transferred to regularly seeded plates the following morning, and were then video recorded to quantify convulsions.

Electrophysiology

We adapted a recording procedure developed from previous studies [13, 39]. Detailed information is in the supplemental methods.

Highlights.

• TRPM channels are modulators of neuronal excitability

• Non-neuronal action of TRPM channel GTL-2 affects excitation-inhibition imbalance

• Epileptic-like activity can be modulated by extracellular divalent ion balance