Abstract
Na+ sensitivity is a unique feature of Na+-activated K+ (KNa) channels, making them naturally suited to counter a sudden influx in Na+ ions. As such, it has long been suggested that KNa channels may serve a protective function against excessive excitation associated with neuronal injury and disease. This hypothesis, however, has remained largely untested. Here, we examine KNa channels encoded by the Drosophila Slo2 (dSlo2) gene in males and females. We show that dSlo2/KNa channels are selectively expressed in cholinergic neurons in the adult brain, as well as in glutamatergic motor neurons, where dampening excitation may function to inhibit global hyperactivity and seizure-like behavior. Indeed, we show that effects of feeding Drosophila a cholinergic agonist are exacerbated by the loss of dSlo2/KNa channels. Similar to mammalian Slo2/KNa channels, we show that dSlo2/KNa channels encode a TTX-sensitive K+ conductance, indicating that dSlo2/KNa channels can be activated by Na+ carried by voltage-dependent Na+ channels. We then tested the role of dSlo2/KNa channels in established genetic seizure models in which the voltage-dependent persistent Na+ current (INap) is elevated. We show that the absence of dSlo2/KNa channels increased susceptibility to mechanically induced seizure-like behavior. Similar results were observed in WT flies treated with veratridine, an enhancer of INap. Finally, we show that loss of dSlo2/KNa channels in both genetic and pharmacologically primed seizure models resulted in the appearance of spontaneous seizures. Together, our results support a model in which dSlo2/KNa channels, activated by neuronal overexcitation, contribute to a protective threshold to suppress the induction of seizure-like activity.
SIGNIFICANCE STATEMENT Slo2/KNa channels are unique in that they constitute a repolarizing K+ pore that is activated by the depolarizing Na+ ion, making them naturally suited to function as a protective “brake” against overexcitation and Na+ overload. Here, we test this hypothesis in vivo by examining how a null mutation of the Drosophila Slo2 (dSlo2)/KNa gene affects seizure-like behavior in genetic and pharmacological models of epilepsy. We show that indeed the loss of dSlo2/KNa channels results in increased incidence and severity of induced seizure behavior, as well as the appearance of spontaneous seizure activity. Our results advance our understanding of neuronal excitability and protective mechanisms that preserve normal physiology and the suppression of seizure susceptibility.
Introduction
Neuronal hyperactivity and Na+ overload are associated with many neuropathological conditions, including Alzheimer's disease (Hartley et al., 1999; Palop et al., 2007; Busche et al., 2008, 2012; Kuchibhotla et al., 2008; Minkeviciene et al., 2009; J. T. Brown et al., 2011; Davis et al., 2014; Ping et al., 2015; Hahm et al., 2018), neuropathic pain (Calvo et al., 2019), amyotrophic lateral sclerosis (Kuo et al., 2004; Gunes et al., 2020), Fragile X Syndrome (Gibson et al., 2008; Deng and Klyachko, 2021), ischemia (J. M. Lee et al., 1999; Choi, 2020), traumatic brain injury (Huttunen et al., 2018), and epilepsy (Fisher et al., 2005). K+ channels offer a fast response element to repolarize the membrane potential and prevent overexcitation. Na+-activated K+ (KNa) channels, in particular, could provide the first line-of-defense against a sudden increase in intracellular Na+ that occurs during neuronal injury and disease. Indeed, physiologists initially reported that high levels of intracellular Na+ (20-180 mm) were required to activate KNa channels and suggested that they play a protective role during epileptic, ischemic, or hypoxic conditions (Kameyama et al., 1984; Luk and Carmeliet, 1990; Mitani and Shattock, 1992; Jiang and Haddad, 1993; Dryer, 1994; Yuan et al., 2003). Since the Slo2.2/Slack/KCNT1/KNa1.1 and Slo2.1/Slick/KCNT2/KNa1.2 genes were identified to encode KNa channels (Bhattacharjee et al., 2003; Yuan et al., 2003), referred to herein as Slo2/KNa channels, they have been found to be widely expressed in the CNS (Bhattacharjee et al., 2002, 2005; Rizzi et al., 2016). Consistent with a protective role against overexcitation, mutations in the human Slo2/KNa genes, KCNT1 and KCNT2, have been linked to multiple forms of epilepsy (Heron et al., 2012; Ishii et al., 2013; Møller et al., 2015; Gururaj et al., 2017; Ambrosino et al., 2018; Mao et al., 2020).
Structure-function studies have shown that cytoplasmic cofactors and post-translational regulation can potentially shift Na+ sensitivity of Slo2/KNa channels to a more physiologically relevant range (Yuan et al., 2003; Tamsett et al., 2009; Huang et al., 2013) and knock-down/-out studies of Slo2.1/KNa1.2 and Slo2.2/KNa1.1 in mice have demonstrated physiological functions of Slo2/KNa channels. For example, Slo2/KNa channels have been suggested to contribute to pain sensation and adaptation (M. R. Brown et al., 2008; R. Lu et al., 2015; Martinez-Espinosa et al., 2015; Evely et al., 2017; Tomasello et al., 2017), particular types of learning (Bausch et al., 2015; Quraishi et al., 2020), and locomotor (Quraishi et al., 2020) behaviors. Although a neuroprotective role against overexcitation and physiological roles in signaling are not mutually exclusive, the hypothesis that Slo2/KNa channels serve as a built-in mechanism against Na+ overload has remained largely untested.
Here, we use Drosophila as a model to test whether Slo2/KNa channels protect against overexcitation. We examine the expression of the single Drosophila Slo2 (dSlo2)/KNa gene, then show that dSlo2/KNa channels carry a TTX-sensitive K+ current in cholinergic neurons that protects against behavioral abnormalities and seizure-like activity associated with prolonged exposure to a cholinergic agonist. Since our results suggest that dSlo2/KNa channels are activated by the persistent Na+ current (INap), similar to Slo2/KNa in mammalian neurons (Budelli et al., 2009; Hage and Salkoff, 2012), we examine the role of dSlo2/KNa channels in genetic mutant and pharmacologically challenged backgrounds in which INap has been elevated, leaving the organism prone to seizure induction. We show that loss of dSlo2/KNa in these models increases the occurrence of induced seizure-like behavior, the severity of seizure-like behavior, as well as the number of individuals that exhibit spontaneous seizures.
Materials and Methods
Drosophila strains
w1118 was used as our WT control in this study. We used the previously generated mutations and transgenic insertions: ChAT-GAL4 (BDSC #6798, 6013), 30y-GAL4, UAS-CD8-GFP, Dα1-T2A-Gal4 (nAchRα1Mi00453-Gal4, BDSC #6670), para-RFP (Ravenscroft et al., 2020), ΔMdr65 (Denecke et al., 2017a), Mi13397 (Metaxakis et al., 2005; Venken et al., 2011), vas-cas9 (BDSC #51323), Generalized Epilepsy with Febrile Seizures Plus (GEFS+) (Sun et al., 2012) (gift from Diane O'Dowd), bang-senseless (bss1) originally denoted (basMW1; gift from Mark Tanouye) (Jan and Jan, 1978), julius seizureiso7.8 (jusiso7.8) previously known as slamdanceiso7.8 (sdaiso7.8) (Zhang et al., 2002; Horne et al., 2017) (gift from Mark Tanouye), easily shocked (easPC80) (Ganetzky and Wu, 1982) (gift from Mark Tanouye), SK- (Abou Tayoun et al., 2011) (gift from Patrick Dolph), Shks133(Ganetzky and Wu, 1982); the following stocks, ChAT-T2A-LexA (ChATMi04508-LexA), GAD1-T2A-LexA (GADMi09277-LexA), VGlut-T2A-LexA (VGlutMi04979-LexA), LexAop-TdTom-nls, UAS-2xEYFP, pC(lox2-attB2-SA-T2A-Gal4-Hsp70)3, hs-Cre, and vas-ΦC31, were all from Diao et al. (2015) (gift from Benjamin White). Unless noted as a gift, remaining stocks were obtained from the Bloomington Drosophila Stock Center (as noted). The following lines were generated and used in the current study: dSlo2/KNa-, dSlo2-T2A-Gal4, dSlo2-myc, UAS-dSlo2/KNa, UAS-dSlo2/KNa-DN.
dSlo2/KNa- null mutant
CRISPR-Cas9 approaches, described previously (Gratz et al., 2015a, b), were used to generate the null mutant dSlo2/KNa-. To target the pore region, sequence spanning the S5-S6 transmembrane region was inserted into the online tool: http://tools.flycrispr.molbio.wisc.edu/targetFinder. Primers that include two gRNA sequences (underlined) to target upstream (forward: 5′-CTT CGT GCT GGA TAC CAC AAA CGC-3′; reverse: 5′-AAA CGC GTT TGT GGT ATC CAG CAC-3′) and downstream (forward: 5′-CTT CGA TGA CCA TAT AAA GCT GCG-3′; reverse: 5′-AAA CGC CAG CTT TAT ATG GTC ATC-3′) sequences flanking the dSlo2/KNa pore. Primers were annealed, phosphorylated, and ligated into pU6-BbsI-chiRNA vector. A donor plasmid that would incorporate a screenable 3xP3≫DsRed marker into the deleted site was designed for homology directed repair. Primers for the left homologous arm (forward: 5′- TAC TCA CCT GCT ACT TCG CCA CTC AAC TGA CCC AAA TTC -3′; reverse: 5′- TAC TCA CCT GCT ACT CTA CCG CTG GGG ATA AAT AAA AAA -3′), and right homologous arm (forward: 5′- TAC TGC TCT TCA TAT AGC TTT ATA TGG TCA TCA TG -3′; reverse: 5′- TAC TGC TCT TCA GAC ATT ATT GGG CGG GTC CCA GA -3′). Genomic DNA was isolated by standard phenol-chloroform extraction methods, followed by isopropanol precipitation. PCR using the left homologous arm primers and right homologous arm primers generated two separate products, which were ligated into the pHD-DSRed-attp vector using AarI and SapI restriction sites, respectively. gRNA and donor plasmids were microinjected into vas-cas9 flies (Rainbow Transgenic Flies). Injected flies were crossed to w1118, and progeny were screened for integration of the donor plasmid. Lines with successful deletion of dSlo2/KNa pore sequence, indicated by expression of DsRed in the eye, were balanced by standard genetic approaches. dSlo2/KNa- null lines were verified by RT-PCR and DNA sequencing.
UAS-dSlo2/KNa transgenic lines
dSlo2/KNa cDNA (kindly provided by Lawrence Salkoff) was digested from the pOX-dSlo2/KNa plasmid using KpnI and XhoI sites and ligated into the pENTR1A vector by Genewiz. The pENTR1A-dSlo2/KNa vector was recombined with the pTW vector using Gateway Technology LR Recombination, and confirmed by DNA sequencing. pTW-dSlo2 was injected into w1118 embryos (Rainbow Transgenic Flies). Transgenic lines were identified and isolated and by standard techniques. Positive UAS-dSlo2/KNa insertions were driven by elav-Gal4 and overexpression of dSlo2/KNa was confirmed by RT-PCR.
UAS-dSlo2/KNa-DN transgenic lines
dSlo2/KNa-DN sequence was generated, in which sequence encoding the pore was mutated from a glycine-tyrosine-glycine (5′-GGA TAC GGC-3′) to alanine-alanine-alanine (5′-GCC GCC GCC-3′) (GYG>AAA). A larger sequence encompassing this mutant pore sequence was generated (Genewiz), digested at endogenous DpnIII and ScaI restriction sites, ligated into the digested pENTR1A-dSlo2/KNa vector, and confirmed by DNA sequencing. pENTR1A-dSlo2-DN was recombined with the pTW vector using Gateway Technology LR Recombination to generate pTW-dSlo2/KNa-DN, which was injected into w1118 embryos (Rainbow Transgenic Flies). Transgenic lines were isolated and identified by standard techniques, and Gal4-driven expression of dSlo2-DN was confirmed by RT-PCR.
dSlo2-T2A-Gal4 line
The Minos-Mediated Integration Cassette line Mi13397 was used as a target for Trojan-mediated expression of Gal4 (T2A-Gal4) that would be under the control of endogenous dSlo2/KNa regulatory elements. Mi13397 is located in a 5′ intronic region of all predicted splice forms of dSlo2/KNa. Transgenic flies containing all three phases of T2A-Gal4, flanked by attB sites as well as loxP sites, were crossed to the Mi13397 line, as described by Diao et al. (2015). Progeny were crossed to a line expressing hs-Cre as well as vas-ΦC31-Integrase. Progeny from this cross were crossed to transgenic UAS-EYFP lines and screened for expression of EYFP, which would indicate successful insertion in the correct phase. EYFP-positive progeny were backcrossed to build a stable stock, and successful insertion of T2A-Gal4 was further verified by DNA sequencing.
dSlo2-T2A-LexA line
The T2A-LexA cassette was inserted into the Mi13397 site to create the dSlo2-T2A-LexA line. To allow for proper translation of LexA, the phase of the Mi13397 insertion in the dSlo2/KNa locus was predicted using the online tool (http://flypush.imgen.bcm.tmc.edu/pscreen/mimic.html) and confirmed to be Phase 2 by DNA sequencing. The plasmid pBS-KS-attB2-SA(2)-T2A-LexA:QFAD-Hsp70 (kindly provided by Ben White) (Diao et al., 2015) was microinjected with the vas-ΦC31 integrase plasmid into Mi13397 embryos. Individually injected flies were crossed to LexAop-TdTom-nls flies, and progeny were screened for TdTom fluorescence.
dSlo2-myc line
A position in Exon 5 (immediately following nucleotide 94 of Exon 5) near the 3′ end of dSlo2 was chosen for insertion myc. Optimal guide RNA (gRNA) target sites within Exon 5 were identified using the online tool (http://tools.flycrispr.molbio.wisc.edu/targetFinger/). Based on these results, a sense primer 5′-CTT CGT TCT GCT CGA ACA TCA ACC -3′ and antisense primer 5′- AAA CGG TTG ATG TTC GAG CAG AAC -3′ encoding the gRNA sequence (underlined) were annealed, phosphorylated, and ligated into the BbsI sites in pU6-BbsI-chiRNA. A single-stranded oligodeoxynucleotide (ssODN) repair template was designed such that the myc sequence (GAA CAA AAA CTC ATC TCA GAA GAG GAT CTG) was in-frame with dSlo2/KNa and flanked by homologous regions of ∼60 nt. Importantly, the myc sequence also disrupts the gRNA target sequence, preventing Cas9 digestion of the repair template. Both the pU6-BbsI-chiRNA-Exon5 plasmid and ssODN were injected into vas-Cas9 embryos (Rainbow Transgenic Flies). Individually injected flies were crossed to a double-balanced line, and progeny from these were backcrossed to each other. Once these flies began laying eggs, adults were pooled in groups of five and screened for successful insertion of the myc tag by PCR and restriction digest with EarI, a site located within the myc sequence. If digestion occurred, progeny from this vial were subsequently crossed individually to double-balancer lines. Once eggs were laid, the individual adult was screened for positive insertion. Successful insertion of in-frame myc sequence was ultimately confirmed by DNA sequencing.
Immunostaining and imaging
Larval body wall immunostaining
Larvae were fileted in cold PBS along the dorsal midline with slight cuts made perpendicular at each end. Corners of the body wall were pinned down on Sylgard, and tissue inside of the larvae was removed. The fileted prep was washed 3× in cold PBS, then fixed in 4% PFA in PBS for 30 min at room temperature. PFA was washed out 3 times with PBS. The prepared larval body wall was then blocked in PBS with 0.5% Triton-X 100 (PBT) and 5% BSA for 2-3 h, then incubated for ∼48 h at 4°C on rocker in PBT with 5% BSA and primary mouse anti-myc (Santa Cruz Biotechnology, 9E10) at 1:50. Body walls were washed 3 times for 30 min, each in PBT with 5% BSA, then incubated in secondary goat anti-mouse conjugated to AlexaFluor-647 at 1:2000 overnight at 4°C. Body walls were again washed 3 times for 30 min in blocking solution, washed 2 times in PBS, followed by fixation in 4% PFA for 10 min to secure antibodies to their targets. Body walls were then washed in PBS, then H2O, and mounted in VectaShield.
CNS immunostaining
Adult and larval Drosophila CNSs were dissected in cold PBS. Following dissection, CNSs were fixed in 4% PFA in PBS at room temperature for 20 min, washed 2 times in PBS, and 2 times in PBT. Primary antibodies used were mouse anti-myc (1:50, Santa Cruz Biotechnology), chicken anti-GFP (1:2000, Aves Labs), and rabbit anti-RFP (1:2000, Rockland Immunochemicals). Primary antibodies were diluted in PBT with 5% BSA or NGS and incubated with CNSs for 48-72 h at 4°C on rocker. CNSs were then washed 3× times in PBT for 10 min each. Alexa fluorophores, anti-chicken 488, anti-rabbit 568, and anti-mouse 647 from Invitrogen were used at a 1:2000 dilution in PBT with 5% BSA or NGS and incubated with CNS for 24-48 h at 4°C on rocker. CNSs were again washed three 3 times for 10 min each in PBT, washed 2 times in PBS, and fixed in 4% PFA for 10 min to secure antibodies to their targets. Following fixation, CNSs were rinsed in PBS, then H2O, and mounted using Vectashield or DABCO (90% glycerol in PBS and 2.5% DABCO). Larval brains were oriented with the dorsal side facing the coverslip, whereas the adult brains were oriented with the ventral side facing the coverslip.
Images were acquired on either an upright Zeiss LSM 880 fluorescence confocal microscope or an inverted Zeiss LSM 800 fluorescence confocal microscope (Carl Zeiss). Z stacks and image tiling on the Zeiss ZEN software were used to acquire images of whole brains. Images were processed in ImageJ, and maximum intensity Z projections were generated when compiling Z-stack images.
Immunostaining embryonic cultures
Single embryos aged 5-6 h (at room temperature) were dissociated in drop cultures of 20 μl culture medium, as previously described (Tsunoda and Salkoff, 1995a, b; Eadaim et al., 2020). After overnight primary antibody incubation at 4°C, cultures were washed, and incubated with secondary. Cultures were fixed in 4% formaldehyde in PBS for 10 min, blocked (1% BSA or 5% goat serum, 0.1% saponin in PBS) for 30 min, then incubated with primary antibody overnight at 4°C. Antibodies against the following proteins were used at the indicated concentrations: anti-myc (1:50; c-Myc Antibody (9E10), Santa Cruz Biotechnology), anti-RFP (1:2000; Rockland). After 4 washes (0.1% saponin in PBS, 5 min each), we incubated cultures with goat-anti-rabbit/mouse AlexaFluor-568/-488 (1:500-1000; Invitrogen) secondary antibodies at room temperature for 2 h, washed again, then mounted them in p-phenylenediamine in 90% glycerol. We imaged cultures with a 60× oil-immersion objective, using Zeiss Apotome technology to obtain optical sections. Images were acquired with Zen Blue (2012) software (Zeiss) and processed for figures using Adobe Photoshop and Illustrator.
Imidacloprid feeding
Larval feeding was performed as previously described (Perry et al., 2008, 2012; Somers et al., 2015). An insecticide containing the active ingredient imidacloprid (“Compare N Save Systemic Tree and Shrub Insect Drench,” Ragan & Massey) was diluted 1:1000 in H2O. From this dilution, imidacloprid was added to standard cornmeal and agar food at the listed concentrations and allowed to solidify overnight in the dark. Embryos were collected from adult flies on egg laying plates of grape juice and agar. Embryos were aged for 24 h to the L1 larval stage. Twenty L1 larvae were gently collected from the egg laying plates using a brush and placed on the cornmeal and agar food containing different concentrations of the insecticides. Larvae were aged for 14-18 d in the dark to prevent degradation of the drug by exposure to light. The larvae that made it to adulthood were counted and averaged over three separate feeding trials, each with three vials (20 larvae/vial) per genotype.
Acute exposure assays following eclosion were performed similar to a previously described study (Denecke et al., 2017b), with noted adjustments. Media containing 5% sucrose and 1% agar were boiled to dissolve sucrose and agar, then imidacloprid was added to final concentration and stirred thoroughly; 4 ml of the food mixture was then aliquoted into separate vials and allowed to solidify overnight in the dark. Newly eclosed males (<24 h) were collected and aged for 5 d under 12 h light/12 h dark conditions. On the fifth day, flies were starved in empty vials for 4 h. Flies were then placed in the imidacloprid food mixture vials and left in the dark. Every 24 h, flies were analyzed for their ability to stand, walk, or groom. If flies were unable to stand or walk, or were dead, they were counted as “affected.”
For imidacloprid feeding before mechanical stimulation, powdered imidacloprid (Sigma-Aldrich, #138261-41-3) was dissolved in H2O to make a 2.36 mm stock solution and kept in the dark. Dissolved imidacloprid was added to media (5% sucrose + 1% agar) to a final concentration of 9 ppm. Male flies were aged for 3 d under 12 h light/12 h dark conditions. On the morning of the third day, flies were placed on the mock- or imidacloprid-food for 24 h under 12 h light/12 h dark conditions. Flies were then subjected to bang-sensitive assays blinded to the researcher. Groups of 5 flies were transferred to empty vials and allowed to acclimate for at least 5 min, then were subjected to 30 s of vortex at high speed. Immediately following vortex, flies were analyzed to determine the fraction that were paralyzed. Fractions affected were averaged across multiple trials.
Veratridine feeding
Newly eclosed male flies (5 per vial) rested, following CO2 exposure, for 24 h under 12 h light/12 h dark conditions. Flies were then starved for 24 h (25°C, ∼64% humidity) in empty vials containing only H2O-soaked filter paper. The H2O-soaked filter paper was replaced with filter paper soaked in 2% sucrose containing either 20 μm veratridine (V-110, Alomone Labs) in DMSO or the vehicle (“mock”; 0.05% DMSO) and left for 2 h. Flies were then tested in mechanically induced and spontaneous seizure assays.
Mechanically induced seizure assays
Newly eclosed male flies were aged for 2 d under 12 h light/12 h dark conditions, gently transferred individually to empty vials using a suction pipette to avoid the use of CO2. Flies were allowed to acclimate to the new vials for at least 5 min, then subjected to a 10 s vortex on high speed. Immediately following vortex, flies were videotaped for later analysis offline. The time to each behavioral characteristic was noted by the researcher, who was blinded to the genotype. These characteristics included the time to both the recovery seizure and full recovery; seizure was characterized by the inability to stand or walk and/or uncontrollable motor function. Additionally, flies were scored for a secondary paralysis period, characterized by a period of quiescence following the recovery seizure that lasted for at least 3 s. If flies exhibited a secondary paralysis, the time spent paralyzed was also measured. These characteristics were averaged across trials of ∼20 flies individually assayed for each genotype.
Heat-induced seizure assays
In experiments using the GEFS+ line, newly eclosed females were aged for 2-3 d. To induce seizures, 5 flies were placed in empty vials and allowed to acclimate for at least 5 min. Flies were then placed in a 40°C water bath, and behavior was videotaped. Analysis of recordings was done offline with the researcher blinded to the genotype of the flies. Seizure-like behavior, defined as uncontrolled motor behavior or paralysis, was scored as a fraction of the total flies every 10 s per vial and averaged across trials.
Spontaneous seizure assays
Newly eclosed male flies were placed in vials in groups of 10 per vial and aged for 2 d under 12 h light/12 h dark conditions. Before testing, flies were gently transferred to empty vials, allowed to acclimate to the new environment for 30 min, then videotaped for 15 min. Videos were analyzed offline, with the researcher blinded to the genotype. Seizures were characterized as random bouts of uncontrollable motor movement in which the flies were unable to stand or walk. The number of overall seizures exhibited in a vial were totaled over the 15 min period and averaged across genotypes.
Electrophysiological recordings
Whole-cell recordings were performed in perforated patch-clamp configuration by adding 400-800 µg/ml Amphotericin-B (Sigma-Aldrich) in the pipette, as described previously (Ping and Tsunoda, 2011; Ping et al., 2011). For Na+ and K+ current recordings, we used external solution (in mm) as follows: NaCl 140, KCl 2, HEPES 5, CaCl2 1.5, MgCl2 6, pH 7.2. For Na+-free external solution, CholineCl was substituted for NaCl. For Na+ current (only) recordings, we used Cs-based internal solution composition (in mm) as follows: CsCl 10, Cs-methanesulfonate 130, HEPES 10, EGTA 0.5, ATP-Mg 4, MgCl2 2, CaCl2 0.1, pH 7.2. Electrodes were filled with internal solution (in mm) as follows: K-gluconate 120, KCl 20, HEPES 10, EGTA 1.1, MgCl2 2, CaCl2 0.1, ATP-Mg 4, pH 7.2. All recordings were performed at room temperature. Gigaohm seals were obtained for whole-cell recordings. Data were acquired and analyzed using an Axopatch 200B amplifier, Axon Digidata 1440A, and pClamp 10.7 software (Molecular Devices). Recordings were digitized at 5 kHz and filtered at 2 kHz, using a lowpass Bessel filter. Additional analysis and presentation software used for all electrophysiological data include the following: Clampfit 10.7 (Molecular Devices), Origin (Microcal Software), and Illustrator (Adobe). Averaged data presented as mean ± SEM. p < 0.05 (Student's t test).
TTX (Sigma-Aldrich) and veratridine (Alomone Labs) were prepared in stock solutions of 2 and 50 mm, respectively. Veratridine was dissolved in DMSO for stock solution, then added to external solutions at final concentrations indicated in the text, and the vehicle concentrations did not exceed 0.01%. Drugs were applied using a rapid application system, termed the “Y-tube method,” which can change the local external solution within a few seconds, as described previously (Akaike et al., 1991; Min et al., 1996).
Experimental design and statistical analyses
In these studies, male Drosophila were used for all comparisons except for those made with bss-/+ mutants (see Figs. 5, 6, and 8); because the bss gene is on the X chromosome, heterozygous bss- required that we use females. Sample sizes and controls are indicated in the figure legends. All behavioral experiments were performed in multiple cohorts with genotypes blinded to the experimenter, as described in Materials and Methods for each experimental assay. Reported are mean ± SEM values, and the standard statistical test used was the unpaired two-tailed t test. The criterion used for statistical significance was p < 0.05; lesser values (e.g., p < 0.01) are also noted in the figure legends.
Results
dSlo2/KNa channels are expressed in excitatory neurons in the CNS
To first examine where dSlo2/KNa is expressed in the CNS, we set out to generate a transgenic line expressing the transcriptional activator Gal4 under the control of native dSlo2/KNa regulatory elements. Using the previously developed “Trojan exon” system (Diao et al., 2015), we aimed to introduce a “Trojan exon,” containing a splice acceptor site followed by the T2A-Gal4 sequence, into a coding intron within the dSlo2/KNa gene. When spliced into the dSlo2/KNa coding transcript, the viral-based T2A signal is predicted to terminate translation of the dSlo2/KNa transcript and promote translation of Gal4 as a second protein product from the same transcript (Diao and White, 2012; Diao et al., 2015; P. T. Lee et al., 2018). This results in Gal4 expression that should mimic endogenous dSlo2/KNa expression (this will also result in truncation of dSlo2/KNa just upstream of the S1 transmembrane domain, making it in effect a null mutant of dSlo2/KNa). To generate this line, ΦC31-mediated recombination was used to induce an in vivo exchange of an exogenously provided T2A-Gal4 cassette for a previously inserted Minos-Mediated Integration Cassette element (Venken et al., 2011), Mi13397; Mi13397 is present in a coding intron predicted to be incorporated into all splice forms of dSlo2. Successful exchange of the T2A-Gal4 cassette was screened for and verified by DNA sequencing; we refer to this new line as dSlo2-T2A-Gal4. We used dSlo2-T2A-Gal4 to drive expression of UAS-EYFP or UAS-CD8-GFP as a reporter for dSlo2/KNa expression in the larval and adult CNS.
In the CNS of third instar (L3) larvae, dSlo2-T2A-Gal4≫UAS-EYFP expression was most abundant in Kenyon cells (KCs) and cells in the ventral nerve cord (VNC) (Fig. 1A). In the adult CNS, we observed widespread dSlo2/KNa reporter expression, with especially strong expression in KCs, and in the morphologically distinct motor neurons of the VNC (Fig. 1B). KCs have been shown to play a key role in associated visual and olfactory learning and memory function (Heisenberg et al., 1985; McGuire et al., 2001; Tanaka et al., 2008; van Swinderen, 2009). dSlo2/KNa reporter expression was also found in second-order sensory processing cells, including the lamina and medulla of the optic lobe, and the antennal lobe that receives input from olfactory receptor neurons (Fig. 1B). Thus, dSlo2/KNa channels may function in regulating excitability in neurons involved in sensory processing, and learning and memory function.
dSlo2/KNa reporter expression in the larval and adult nervous system. A, Representative optical sections, 0.6-0.8 µm, from an L3 larval brain show the following: (Ai) dSlo2-T2A-Gal4≫UAS-EYFP expression in KCs (kc) and (Aii) dSlo2-T2A-Gal4≫UAS-mCD8-GFP expression in the VNC. B, Representative optical sections and Z projections from the adult brain are as follows: (Bi) Z projection of 27 confocal sections spanning 16.5 µm showing dSlo2-T2A-Gal4≫UAS-mCD8-GFP expression in KCs (kc), calyx (cx), and mushroom body lobes (α and β); (Bii) Z projection of 7 confocal sections spanning 3.5 µm showing dSlo2-T2A-Gal4≫UAS-mCD8-GFP expression in the antennal lobe (al); (Biii) optical sections, 0.6-0.8 µm, showing dSlo2-T2A-Gal4≫UAS-mCD8-GFP expression in motor neurons (mn); (Biv) optical sections, 0.6-0.8 µm, showing dSlo2-T2A-Gal4≫UAS-EYFP expression in the lamina (la) of the optic lobe; and (Bv) optical sections, 0.6-0.8 µm, showing dSlo2-T2A-Gal4≫UAS-EYFP expression in the medulla (md) of the optic lobe. Scale bars, 10 µm.
Since dSlo2/KNa expression appeared to be widespread, at least at low levels, we next investigated whether dSlo2/KNa is preferentially expressed in cholinergic, GABAergic, and/or glutamatergic neurons. To do this, we used ChAT-T2A-LexA, GAD1-T2A-LexA, and VGlut-T2A-LexA lines, respectively, which drive LexA expression in each of these respective neuronal subtypes. LexA transgenes and dSlo2-T2A-Gal4 were used to drive expression of different colored fluorophores in the same brains. dSlo2-T2A-Gal4 drove expression widely in cholinergic neurons, as expected, since the major excitatory neurotransmitter in the Drosophila brain is acetylcholine (Fig. 2A). Since neurons localized to the medulla cortex exhibit a mixture of cholinergic and GABAergic neurons (Strother et al., 2017), we used this region to examine whether dSlo2-T2A-Gal4 was expressed in excitatory and/or inhibitory neurons. Interestingly, all apparent dSlo2-T2A-Gal4≫UAS-EYFP-positive neurons were also positive for ChAT-LexA≫LexAop-TdTomato-nls (Fig. 2A). In contrast, dSlo2-T2A-Gal4≫UAS-EYFP and Gad1-LexA≫LexAop-TdTomato-nls were strikingly nonoverlapping (Fig. 2B), suggesting that dSlo2/KNa channels are expressed exclusively in cholinergic neurons and excluded from GABAergic neurons. dSlo2-T2A-Gal4≫UAS-CD8-GFP expression in glutamatergic cells appeared to be cell type-specific. For example, in the medulla cortex of the optic lobe, dSlo2-T2A-Gal4≫UAS-CD8-GFP and VGlutMi14979-LexA≫LexAop-TdTomato-nls expression showed no apparent overlap, whereas in the VNC, significant overlap was observed in motor neurons (Fig. 2C). The widespread expression in excitatory neurons in the brain, and in motor neurons, suggests that dSlo2/KNa channels might function to modulate neuronal, and perhaps behavioral, excitability.
dSlo2-T2A-Gal4 drives expression selectively in cholinergic neurons. A, B, Representative optical sections, 0.6-0.8 µm, of UAS-EYFP driven by dSlo2-T2A-Gal4 (dSlo2≫EYFP; green) through the medulla cortex of the adult brain. LexAop-TdTom-nls (magenta) was simultaneously driven in (A) cholinergic neurons with ChAT-T2A-LexA (ChAT≫TdTom) and (B) GABAergic neurons with Gad1-T2A-LexA (GAD≫TdTom). C, Representative optical sections, 0.6-0.8 µm, showing dSlo2-T2A-Gal4≫UAS-mCD8-GFP (dSlo2≫CD8-GFP, green) expression in glutamatergic neurons identified by expression of VGlut-T2A-LexA≫UAS-TdTom-nls (VGlut≫TdTom) in the medulla (top, Brain) and VNC (bottom). D, Representative optical sections, 0.6-0.8 µm, showing 30y-Gal4≫UAS-CD8-RFP (30y≫CD8-RFP; magenta) and dSlo2-myc (dSlo2-myc; green) coimmunolabeling in KCs (top, cell bodies [cb] and calyz [cx]) and in an optical cross-section of the peduncle (ped; dotted white outline). E, Z projection of 19 confocal sections spanning 18 µm showing coexpression of ppk-Gal4≫UAS-CD4-TdTom (ppk≫TdTom; magenta) and dSlo2-myc (green) in Class IV multidendritic neurons. Arrows indicate axonal projection from this multidendritic neuron. Scale bars, 10 µm.
To gain further insight into how dSlo2/KNa channels might function in neuronal excitability, we set out to characterize the subcellular localization of dSlo2/KNa channels. To do this, we generated a line in which DNA coding for myc was inserted into the endogenous dSlo2/KNa gene such that the translated myc epitope would be fused within the C-terminus of the channel subunits (see Materials and Methods); we refer to this line as dSlo2-myc. We found that dSlo2-myc was differentially localized to cell bodies and axonal/dendritic subcompartments in various neuronal subtypes. For example, in KCs, dSlo2-myc expression was localized to cell bodies, the calyx, and peduncle, a structure consisting of KC axons that project to the mushroom body lobes (Figs. 1B and 2D), but was absent from the axons in the mushroom body lobes themselves. Since the calyx consists of a mixture of dendritic processes from the KCs as well as axonal processes from other neurons, such as the projection neurons, it is unclear whether dSlo2-myc was present in the dendrites of the KCs and/or axonal terminals from projection neurons. We also examined the subcellular localization of dSlo2-myc in Class IV multidendritic neurons whose axonal and dendritic processes are clearly distinguishable. We found that dSlo2-myc channels were localized to axonal process of these neurons, but seemingly absent from dendrites and cell bodies (Fig. 2E). Thus, dSlo2-myc appears to be trafficked to different neuronal subcompartments, depending on cell identity, where these dSlo2/KNa channels may dampen neuronal excitability by different mechanisms.
Effects of a cholinergic agonist are exacerbated by the loss of dSlo2/KNa channels
Given the expression of dSlo2/KNa in excitatory cholinergic neurons in the brain, and the likelihood that dSlo2/KNa channels would be activated by increased excitation, our goal was to test the long-standing hypothesis in the field that Slo2/KNa channels dampen excitability when confronted with excessive excitation (Kameyama et al., 1984). We first investigated whether dSlo2/KNa channels might function in circumstances of increased excitation induced by feeding flies the cholinergic agonist, imidacloprid. Activation of nicotinic acetylcholine receptors (nAChRs), the major excitatory receptor in the Drosophila CNS, results in an influx of cations, including Na+, which could in turn activate dSlo2/KNa channels. Since we planned to feed flies imidacloprid, which has been shown to act as a partial agonist for Dα1 nAChRs (L. A. Brown et al., 2006; Perry et al., 2008; Ihara et al., 2018), we first confirmed that dSlo2/KNa channels and Dα1 receptors have overlapping expression. To do this, we generated and used dSlo2/KNa-T2A-LexA and Dα1-T2A-Gal4 lines to drive expression of LexAop-TdTomato-nls and UAS-CD8-GFP, respectively. Throughout the CNS, we found that virtually all Dα1-positive cells were also positive for dSlo2/KNa reporter expression (Fig. 3A).
Effects of the cholinergic agonist, imidacloprid, are exacerbated in dSlo2/KNa- mutants. A, Representative optical sections showing UAS-CD8-GFP and LexAop-TdTom-nls driven by Dα1-Gal4 and dSlo2-T2A-LexA, respectively (Dα1≫CD8-GFP [green] and dSlo2≫TdTom [magenta]), in KC bodies of the adult brain. Scale bar, 10 µm. B, Mean percentage of adults that eclosed from groups of 20 L1 larvae raised on food with containing 0, 0.6, 1.2, or 1.8 ppm imidacloprid. Genotypes are as indicated: WT, dSlo2/KNa- (dSlo2-), dSlo2-T2A-Gal4, and ΔMdr65. n = 6-9 groups per genotype. C, D, Mean percentage of flies scored as “affected” (inability to stand or walk; see Materials and Methods) was averaged each day following placement of flies on food containing either 18 ppm (left) or 55 ppm (right) imidacloprid. Genotypes are as indicated: WT and dSlo2/KNa- (C), and dSlo2-T2A-Gal4 and dSlo2-T2A-Gal4≫UAS-dSlo2 (D). n = 8-10 groups of 5 flies per group for each genotype. E, Fraction of flies paralyzed following a 30 s mechanical stimulation (see Materials and Methods) after 24 h of feeding either vehicle (Mock) or 9 ppm imidacloprid (IMI). n = 12-16 trials of 5 flies per group for each condition and genotype. Data are mean ± SEM. *p < 0.05, mutant data compared with WT (Student's t test).
We then examined the toxic effects of feeding WT larvae imidacloprid. We placed WT first instar (L1) larvae on food mixed with imidacloprid, which they consumed for the rest of development, then scored for the number of viable adults that eclosed. Consistent with earlier reports (Perry et al., 2008; Denecke et al., 2017b), we found that the fraction of larvae that survived to adulthood was progressively reduced as the concentration of imidacloprid was increased (Fig. 3B). We also tested ΔMdr65 mutants, which contain a deletion of the Mdr65 gene that encodes an ATP-binding cassette transporter. These mutants were previously shown to be highly susceptible to imidacloprid treatment because of an inability to effectively efflux the compound from the CNS (Denecke et al., 2017a), and have been used here as a positive control for imidacloprid toxicity. Indeed, even at 0.6 ppm imidacloprid, only 23% of the ΔMdr65 larvae eventually eclosed as adults, compared with 78% of WT; at 1.2 ppm imidacloprid, only 1% of ΔMdr65 adults eclosed (Fig. 3B). Previous studies have shown that this imidacloprid toxicity is because of its hyperactivation of Dα1 nAChRs (Perry et al., 2008).
Hyperactivation of Dα1 nAChRs would likely result in an overload of intracellular Na+, and perhaps activation of dSlo2/KNa channels as a protective measure. To test whether the loss of dSlo2/KNa channels would affect imidacloprid toxicity, we generated a dSlo2/KNa- null mutant line (see Materials and Methods) and examined whether dSlo2/KNa- mutants exhibited any difference in imidacloprid toxicity compared with WT. We found that indeed dSlo2/KNa- mutants were more sensitive to imidacloprid. At 1.2 and 1.8 ppm imidacloprid, 50% and 35% of WT flies eclosed, respectively, while only 18% and 6% of dSlo2/KNa- mutants eclosed, respectively (Fig. 3B). We also tested the dSlo2-T2A-Gal4 line, which, as described above, is predicted to essentially be a null mutant of dSlo2/KNa. We found that dSlo2-T2A-Gal4 flies were also more susceptible to imidacloprid than WT flies (Fig. 3B). Our results show that imidacloprid toxicity during development is more detrimental in the absence of dSlo2 channels, suggesting that dSlo2 channels act protectively against imidacloprid-induced hyperactivity.
We next examined whether adult flies acutely fed imidacloprid exhibit detectable behavioral alterations. For these experiments, flies were fed food mixed with imidacloprid, then scored each day as “affected” if they were unable to walk, stand, or groom. We found that, after 3 d of feeding on 18 ppm imidacloprid or after 1 d of feeding on 55 ppm imidacloprid, an increasing number of WT flies were affected, with 100% of the flies affected after 4 d feeding on 55 ppm imidacloprid (Fig. 3C). When dSlo2/KNa- mutants were examined in concurrent assays, we found that significantly more dSlo2/KNa- mutants were affected than WT. This can be seen at 4-5 d on 18 ppm imidacloprid, and 2-3 d on 55 ppm imidacloprid (Fig. 3C). To test whether this increased susceptibility to imidacloprid was because of the loss of dSlo2/KNa channels, we examined whether exogenous expression of dSlo2/KNa in a dSlo2/KNa- mutant background would rescue the percent of flies affected by imidacloprid. To do this, we used the dSlo2-T2A-Gal4 line as a dSlo2/KNa loss-of-function background. Indeed, like the dSlo2/KNa- mutant, the dSlo2-T2A-Gal4 line showed high sensitivity to imidacloprid (Fig. 3D); it is unclear why the dSlo2-T2A-Gal4 line was even more sensitive to imidacloprid than dSlo2/KNa-. We generated UAS-dSlo2/KNa transgenic lines to test for rescue of this increased sensitivity. We used dSlo2-T2A-Gal4 to drive expression of UAS-dSlo2/KNa (dSlo2-T2A-Gal4≫UAS-dSlo2/KNa) and tested for imidacloprid susceptibility. We found that the percent of flies affected after 2-3 d on imidacloprid was indeed significantly reduced with expression of UAS-dSlo2/KNa (Fig. 3D), suggesting that dSlo2/KNa channels function protectively against imidacloprid-induced hyperactivity.
Since other hyperactive lines have been characterized to exhibit paralysis induced by mechanical stress (Parker et al., 2011b), we also tested whether mock- and imidacloprid-fed WT and dSlo2/KNa- flies exhibited any difference in mechanically induced paralytic behavior. We found that WT mock-treated flies and flies fed 9 ppm imidacloprid for 24 h showed no significant change in behavior after mechanical stimulation (Fig. 3E). In contrast, 80% of imidacloprid-fed dSlo2/KNa- flies exhibited paralysis following mechanical stimulation (Fig. 3E). Together, our results using imidacloprid to overactivate nAChRs suggest that dSlo2/KNa channels function to dampen this excessive activity and protect against detrimental developmental and behavioral consequences.
dSlo2 channels are activated by a TTX-sensitive Na+ current in neurons
Mammalian Slo2/KNa channels have been shown to be activated by Na+ influx through TTX-sensitive Na+ channels, and in particular, by the persistent current Na+ current (INap) that follows the transient Na+ current component (Budelli et al., 2009; Hage and Salkoff, 2012). To test whether there is also a TTX-sensitive K+ current present in Drosophila neurons, we assayed primary neurons dissociated from ChAT-Gal4≫UAS-GFP embryos, which express GFP in ChaT-positive neurons, and grown in culture for 9-10 DIV. In whole-cell perforated-patch mode, we applied voltage-clamp protocols to elicit voltage-dependent Na+ and K+ currents from ChAT-positive neurons. From a holding potential of –90 mV, we took 200 ms voltage jumps to potentials from –50 to 50 mV, in 10 mV increments. Based on extensive characterization of these neurons (Tsunoda and Salkoff, 1995a, b; Ping et al., 2011), this voltage protocol is expected to elicit a Na+ current encoded by the single voltage-dependent Na+ channel gene para, a rapidly activating transient A-type K+ current encoded by the Kv4/Shal gene, and delayed-rectifier type K+ currents encoded by the Kv2/Shab and Kv3/Shaw genes. In the first 100 ms of the voltage jump, Na+ and Kv4/Shal channels are likely to be activated and carry currents in opposing directions, making the visualization of either incomplete. In the last 100 ms of the voltage jump, however, the transient Na+ current component will be completely inactivated and only INap and sustained K+ currents will be represented. Thus, we measured the sustained current component, by averaging the last 60 ms, before and after TTX (1 μm) bath perfusion; any TTX-sensitive current was isolated by subtracting the whole-cell current after TTX application from the whole cell current before TTX application. We found a small outward current that could be measured from ∼20 mV (Fig. 4A,D). The outward nature of this current suggests the presence of a TTX-sensitive K+ current. To confirm that this TTX-sensitive K+ current is Na+-dependent, we also removed Na+ from the extracellular bath solution and performed the same experimental protocol. Indeed, the small outward TTX-sensitive current was absent under these conditions (Fig. 4B,E).
dSlo2/KNa encodes a TTX-sensitive outward current in neurons. A-C, Na+ and K+ currents elicited from representative WT neurons in Na+-containing extracellular solution (A, WT+Na+) or Na+-free extracellular solution (B, WT-Na+), and dSlo2/KNa- neurons in Na+-containing extracellular solution (C, dSlo2-) at 9 DIV in response to voltage jumps to −50 to 50 mV, in 10 mV increments, before (Control) and after 1 μm +TTX application. Current traces from the +TTX condition were subtracted from the Control condition to give TTX-sensitive current trace shown (KNa). Calibration: 100 pA, 10 ms. D-F, Plotted are current densities for Ctrl, +TTX, and KNa currents, as described for A-C at indicated test potentials for WT+Na+ (D), WT-Na+ (E), and dSlo2/KNa- mutant (F) neurons. n = 12-14 cells. G, Calculated current densities for the TTX-sensitive (KNa) current in WT+Na+, WT-Na+, and dSlo2- mutant neurons at a test potential of 40 mV; n = 12-14 cells. D-G, Data are mean ± SEM. *p = 0.018 (Student's t test). H, Optical sections through a representative cluster of neuronal cell bodies (top) and extending processes (bottom) of primary neurons cultured (8-10 DIV) from para-mCherry;dSlo2-myc embryos. Cells are coimmunostained for dSlo2-myc (green) and para-mCherry (red). Scale bar, 10 µm. I, Representative optical sections showing coimmunostaining of para-RFP (magenta) and dSlo2-myc (green) in KC somas (top; scale bar, 5 µm) and axons in the peduncle (bottom, dotted white outline; scale bar, 10 µm) from the intact adult brain.
We next tested whether this small sustained TTX-sensitive K+ current was encoded by the dSlo2/KNa gene. To do this, we recorded from ChAT-positive neurons in primary cultures made from the dSlo2/KNa- null mutant line. We found no outward TTX-sensitive current present in dSlo2/KNa- mutant neurons (Fig. 4C,F); instead, the TTX-sensitive current in dSlo2/KNa- mutant neurons was a small inward current that likely represents INap (Fig. 4F). These results suggest that dSlo2/KNa encodes a small sustained K+ current in excitatory Drosophila neurons that is activated by Na+ influx carried by TTX-sensitive Na+ channels.
To examine whether dSlo2/KNa channels are localized in close proximity to voltage-dependent Na+ channels, we immunostained similarly cultured and aged primary neurons from a Drosophila line expressing endogenously labeled dSlo2-myc and para-mCherry (Ravenscroft et al., 2020). We found that dSlo2-myc and para-mCherry exhibited significant overlap in neuronal cell bodies and processes at 8-10 DIV (Fig. 4H). In KCs of the adult brain, we observed that dSlo2-myc and para-RFP overlap in somas and in the peduncle, a structure housing axonal segments proximal to the KC cell body layer (Fig. 4I), but both were seemingly absent from more distal axonal regions in the mushroom body lobes. These results suggest that dSlo2/KNa and voltage-dependent Na+ channels often colocalize in similar neuronal subcompartments, where INap influx through Na+ channels could activate nearby dSlo2/KNa channels.
dSlo2/KNa- mutant neurons exhibit excitability similar to WT neurons under basal conditions
We first examined whether dSlo2/KNa- null mutant neurons exhibit altered excitability under normal physiological conditions. In current-clamp mode, we ramped current injection from 0 to 150 pA over 1000 ms and recorded membrane voltage changes. We measured the charge transfer required to elicit the first action potential (AP) firing, the rate of change in membrane potential in the first 50 ms (slope), amplitude of the first AP, time to first AP peak, and the interval between the first and second APs. We found no differences between WT and dSlo2/KNa- mutants for all of these measurements (Fig. 5), with the exception of an increase in the interval between the first and second AP (Fig. 5F). Overall, however, there was no significant change in neuronal excitability under basal conditions. Membrane resistance, cell capacitance, and resting membrane potentials were also not different between WT and dSlo2/KNa- mutants (Table 1). These results are consistent with behavioral data in this study which show no difference between WT and dSlo2/KNa- mutants unless challenged with conditions of global overexcitation.
dSlo2/KNa- mutant neurons exhibit excitability similar to WT neurons under basal conditions. A, Representative membrane voltage responses from WT and dSlo2/KNa- (dSlo2-) neurons to ramped current injection from 0 to 150 pA over 1000 ms; the first 165 ms are shown. The following measurements were made: B, charge transfer required to elicit the first AP; C, rate of membrane potential change in the first 50 ms of ramp stimulation; D, peak amplitude of the first AP; E, time to the first AP peak; F, interval between peaks of the first and second AP. n = 17 or 18 cells per genotype. Data are mean ± SEM. *p < 0.05 (Student's t test).
Passive electrical properties of WT and dSlo2/KNa- neurons are similar
Loss of dSlo2/KNa in GEFS+ seizure mutants exacerbate heat-induced seizure-like behavior
We next set out to test whether dSlo2/KNa channels might dampen excitation under conditions of hyperexcitation created by elevated levels of INap. In Drosophila, there are multiple seizure models that exhibit an enhanced INap current. First, we examined the GEFS+ mutant model. This model was designed to contain a genomic point mutation in the para Na+ channel gene that is analogous to a human mutation that causes an elevated INap current and underlies the disease, GEFS+ (Abou-Khalil et al., 2001; Sun et al., 2012). Similar to effects in humans, Drosophila GEFS+ mutants exhibit an enhanced INap current and seizure-like activity in response to increased temperature (Sun et al., 2012) (Fig. 6A). In contrast, WT and dSlo2/KNa- mutants did not induce any seizure-like activity at 40°C (Fig. 6A). We next examined whether the loss of dSlo2/KNa channels in GEFS+ mutants would affect seizure susceptibility. Since GEFS+ mutants exhibit an enhanced INap current, and INap activates dSlo2/KNa channels, we reasoned that dSlo2/KNa channels might provide some initial protection against hyperactivity. If this hypothesis is correct, we expect that the loss of dSlo2/KNa channels would result in greater seizure sensitivity. We compared the GEFS+ mutation with and without the dSlo2/KNa- null mutation. We found that, while 50% of GEFS+ mutants exhibited seizure-like activity after ∼100 s at 40°C, GEFS+; dSlo2/KNa- double mutants were significantly more sensitive to heat-induced seizure-like activity, with 50% of the flies seizing after only ∼70 s (Fig. 6A). These results suggest that dSlo2/KNa channels function to dampen hyperactivity and suppress seizure-like behavior.
The absence of dSlo2/KNa channels in multiple INap-affected seizure models results in increased seizure susceptibility following heat/mechanical stimulation. A, Mean fraction of flies exhibiting seizure-like behavior shown for WT, dSlo2/KNa-, GEFS+, and GEFS+;dSlo2/KNa- while subjected to 40°C for the times indicated. n = 10-12 groups of 5 individuals per genotype. B, Time to full recovery following 10 s mechanical stimulation for WT and dSlo2/KNa- (dSlo2-). C, Diagram showing typical sequence of seizure behavior in “bang-sensitive” models following mechanical stimulation. D-F, Mean time in primary paralysis, fraction of flies exhibiting secondary paralysis, time in secondary paralysis, and time to full recovery for individually assayed jus- (D), eas- (E), and bss-/+ (F) mutants with and without the dSlo2/KNa- mutation, as indicated, immediately following 10 s mechanical stimulation. n = 17-19 flies per genotype. Data are mean ± SEM. Double mutants compared with jus-, eas-, and bss-/+ backgrounds. **p < 0.01 (Student's t test).
Loss of dSlo2/KNa in INap-affected “bang-sensitive” seizure models results in increased susceptibility and prolonged seizure-like behavior
We next examined three “bang-sensitive” mutants that exhibit an elevated INap current and consequent susceptibility to mechanically induced seizure-like activity. These mutants, julius seizureiso7.8 (jus-) (previously known as slamdanceiso7.8), easily shockedPC80(eas-), and bang-senseless1 (bss-), have been well characterized as models of human epilepsy (Zhang et al., 2002; Song and Tanouye, 2008; Marley and Baines, 2011; Parker et al., 2011a, b). jus-, eas-, and bss- mutants all exhibit seizure-like activity induced by mechanical stress; this seizure-like activity has been well characterized both behaviorally and electrophysiologically (Benzer, 1971; Ganetzky and Wu, 1982; Pavlidis et al., 1994; Pavlidis and Tanouye, 1995; Kuebler et al., 2001; J. Lee and Wu, 2002; Song and Tanouye, 2008; Marley and Baines, 2011; Parker et al., 2011a). We set out to examine whether the dSlo2/KNa- null mutation exacerbates these seizure-like phenotypes. Because homozygous bss- mutants have been reported to exhibit such a strong phenotype, it was not clear that we would be able to detect an exacerbated phenotype in this background; we therefore used bss-/+ heterozygotes, which have been shown to exhibit a similar, albeit less severe, seizure-like phenotype (Kuebler and Tanouye, 2000; Parker et al., 2011b).
Mechanical stress was invoked by a 10 s vortex of single flies in empty vials. While WT and dSlo2/KNa- mutant flies immediately recovered from this mechanical stimulation (Fig. 6B), eas-, and bss-/+ mutants all exhibited the expected stereotypical sequence of seizure activity: a “primary paralysis” phase following an “initial seizure” that occurs during this mechanical stimulation, followed by a “recovery seizure,” and finally, “full recovery” (as diagrammed in Fig. 6C; Fig. 6D–F), as previously reported (for review, see Parker et al., 2011a). Seizure-like activity was characterized by uncontrollable motor movement and an inability to stand or walk. We then tested whether and how the absence of dSlo2/KNa affected this behavioral sequence in jus-, eas-, and bss-/+ mutant backgrounds. We measured the time flies spent in primary paralysis, as well as the total time to full recovery. We found that the loss of dSlo2/KNa channels in all bang-sensitive lines resulted in a significant increase in times of paralysis and times to full recovery (Fig. 6D–F). For example, seizure mutants' times to full recovery were lengthened by 57%-163% when combined with the dSlo2/KNa- mutation (Fig. 6D–F). Interestingly, in jus- and eas- mutants, the absence of dSlo2/KNa often caused a “secondary paralysis” period following the recovery seizure (diagrammed in Fig. 6C; Fig. 6D–F); this tendency was largely absent in the jus- and eas- mutant backgrounds alone, although routinely observed in bss-/+ mutants (Fig. 6D–F). We quantified the duration of these secondary paralysis periods and found that the loss of dSlo2/KNa also prolonged these periods in all mutant backgrounds (Fig. 6D–F).
We then tested whether dSlo2/KNa channels play this protective role in excitatory neurons of the CNS, as suggested by our expression studies (Fig. 2A). To do this, we generated a transgenic line expressing a dominant-negative (DN) dSlo2/KNa subunit under the control of UAS (UAS-dSlo2/KNa-DN). In this dSlo2/KNa-DN sequence, the highly conserved K+ channel pore sequence, glycine-tyrosine-glycine (GYG), was mutated to three alanines (GYG>AAA); this mutation has previously been shown to produce subunits that, when multimerized with WT subunits, renders K+ channels nonfunctional (Perozo et al., 1993; Barry et al., 1998; Abou Tayoun et al., 2011; Ping et al., 2011). Because we had observed widespread expression of dSlo2/KNa in most cholinergic neurons throughout the adult brain, we used ChAT-Gal4 to drive expression of UAS-dSlo2/KNa-DN in cholinergic neurons of jus-, eas-, and bss-/+ mutant lines. We found that ChAT-Gal4≫UAS-dSlo2/KNa-DN expression increased the time from mechanical stimulation to full recovery in each of these seizure lines (Fig. 7). Together, our results suggest that dSlo2/KNa channels protect excitatory neurons from hyperactivity and aid in the prevention of induced seizure-like behavior.
Expression of dominant-negative dSlo2/KNa subunit (dSlo2/KNa-DN) in cholinergic neurons prolongs induced seizure-like behavior in INap-affected models. Shown are mean times that individual flies took to full recovery after mechanical stimulation. These times were measured in jus- (A), eas- (B), and bss-/+ (C) mutants with ChAT-Gal4 (Gal4), UAS-dSlo2/KNa-DN (UAS), or ChAT-Gal4≫UAS-dSlo2/KNa-DN (dSlo2-DN), as indicated. Data for jus-, eas-, and bss-/+ alone are from Figure 6 and shown for comparison. Data are mean ± SEM. *p < 0.05 (Student's t test).
dSlo2/KNa channels protect flies from mechanically induced seizure-like behavior primed by pharmacological enhancement of INap
We next tested whether pharmacological enhancement of the INap current component would also prime the nervous system for induction of seizure-like behavior, and whether the loss of dSlo2/KNa would exacerbate this behavior. To do this, we considered veratridine, which has been shown in multiple mammalian cell types to prolong the open state of Na+ channels, thereby increasing the INap current (Barnes and Hille, 1988; Hage and Salkoff, 2012). We tested veratridine in primary cultured Drosophila neurons and found that bath perfusion of veratridine (5 μm) indeed increased the INap current component of the Na+ current, with no significant effect on the transient Na+ current component (Fig. 8A).
dSlo2/KNa- mutants fed veratridine, an enhancer of INap, exhibit increased susceptibility to mechanically induced seizure-like behavior. A, Representative whole-cell Na+ current recorded from a WT neuron (9 DIV) from a holding potential of −90 mV to a test potential of −20 mV before (Control; black trace) and after 5 µm veratridine (+veratridine; red trace) application. Transient and persistent components of the Na+ current are indicated. Calibration: 20 pA, 10 ms. B, Normalized Na+ current densities for the transient current component (left; transient) and persistent INap current component (right; persistent) from Control and veratridine-treated neurons. n = 9 cells. C, The mean fraction of flies exhibiting seizure-like behavior immediately following a 10 s mechanical stimulation. WT, dSlo2/KNa-, SK-, and ShKS133 (Sh-) flies were fed either vehicle (0.05% DMSO, Mock) or 20 μm veratridine (Vtd) for 2 h before mechanical stimulation. n = 10-23 groups of 5 flies/vial. Data are mean ± SEM. *p < 0.05 (Student's t test).
We then fed flies either vehicle (mock-treated) or veratridine (20 μm) for 2 h and subjected them to mechanical stress with a 10 s vortex. While this mechanical stress did not induce the stereotypical sequence of seizure-like behavior exhibited by bang-sensitive flies (Fig. 6C), it did result in seizure-like behavior immediately following mechanical stimulation. To test whether dSlo2/KNa channels function protectively against this seizure-like behavior, we compared mock- or veratridine-fed WT and dSlo2/KNa- flies. We found that a significantly higher fraction of veratridine-fed dSlo2/KNa- mutants displayed mechanically induced seizure-like behavior, compared with mock-fed dSlo2/KNa- mutants or veratridine-fed WT flies (Fig. 8B). To test the specificity of this effect, we also mock- and veratridine-fed null mutants of the small-conductance Ca2+-activated K+ channel gene (SK-) and the voltage-dependent Kv1/Shaker K+ channel gene (ShKS133). SK- and ShKS133 mutants, whether mock- or veratridine-fed, did not exhibit any more seizure-like behavior than WT (Fig. 8B). Our results suggest that dSlo2/KNa channels protect neurons from seizure induction also in this veratridine-primed model, in which the system is confronted with a global increase in INap.
The loss of dSlo2/KNa reveals a spontaneous seizure phenotype in INaP-affected seizure models
Bang-sensitive mutants are well known to require mechanical stress to induce seizure-like behavior, and this was largely true for our veratridine-primed model. However, we noticed that, when these seizure models contained the dSlo2/KNa- mutation, a significant number of flies appeared to exhibit spontaneous seizure-like behavior. One possibility is that dSlo2/KNa channels contribute to the threshold for seizure induction, and when absent in these seizure models, threshold is significantly reduced, thereby enabling spontaneous seizures to occur. To test this, we quantified the number of seizures observed in groups of 10 flies over a 15 min period; we first tested jus-, eas-, and bss-/+ mutants with and without dSlo2/KNa. As expected, jus-, eas-, and bss-/+ flies all exhibited little to no spontaneous seizure-like activity, similar to WT (Fig. 9A). In contrast, dSlo2/KNa-; jus- and bss-/+; dSlo2/KNa- double mutants both displayed a significant number of spontaneous seizures (Fig. 9A); dSlo2/KNa-;eas- double mutants, however, were similar to WT (Fig. 9A), and it is unclear why the eas- mutant background was less susceptible to spontaneous seizures than the other mutants. dSlo2/KNa- mutants themselves displayed minimal spontaneous seizure activity (Fig. 9A), suggesting that conditions for spontaneous seizure-activity is primed by an enhancement in INap, and revealed by a lack of dSlo2/KNa channels. To test whether this effect is specific to the loss of dSlo2/KNa channels in cholinergic neurons, we drove expression of the dominant-negative UAS-dSlo2/KNa-DN transgene exclusively in cholinergic neurons in the jus- mutant, and assayed for spontaneous seizure-like activity. Similar to the dSlo2/KNa-;jus- double mutant, ChAT-Gal4≫UAS-dSlo2/KNa-DN;jus- flies exhibited a significantly increased number of spontaneous seizures, compared with cholinergic expression of dSlo2-DN alone, or the jus mutation combined with the ChAT-Gal4 or UAS-dSlo2/KNa-DN alone (Fig. 9B).
Absence of dSlo2/KNa channels in multiple INap-affected seizure models results in an increased incidence of spontaneous seizures. A-C, Data are mean numbers of spontaneous seizures observed in groups of 10 flies over a 15 min period of time. A, Mean numbers of spontaneous seizures for WT, dSlo2/KNa-, and jus-, eas-, bss-/+ mutations alone and in combination with dSlo2/KNa-, as indicated. n = 7 or 8 groups for WT and single mutants; n = 14-16 groups for double-mutant genotypes with dSlo2-. B, Mean numbers of spontaneous seizures for ChAT-Gal4≫UAS-dSlo2/KNa-DN (dSlo2-DN), and jus- alone or in combination with ChAT-Gal4 (Gal4), UAS-dSlo2/KNa-DN (UAS), or dSlo2-DN, as indicated. n = 7 or 8 groups per genotype. C, Mean numbers of spontaneous seizures for WT, dSlo2/KNa-, SK-, and ShKS133 (Sh-) flies fed either vehicle (0.05% DMSO, Mock) or 20 μm veratridine for 2 h before assay. n = 8-10 groups per genotype. Data are mean ± SEM. *p < 0.05 (Student's t test). D-F, Shown, for descriptive purposes, are histograms of durations of spontaneous seizures grouped into indicated bins for indicated genotypes and conditions.
We also tested whether veratridine-enhancement of INap would similarly induce spontaneous seizure-like activity that would be exposed with the lack of dSlo2/KNa channels. WT and dSlo2/KNa- flies, as well as SK- and ShKS133 K+ channel mutants, were mock- or veratridine-fed for 2 h. Then, groups of 10 were observed over a 15 min period for any seizure-like activity. We found that the number of seizures was very low and not significantly different in WT, SK-, and ShKS133 flies, whether mock- or veratridine-fed (Fig. 9C). In contrast, veratridine-fed dSlo2/KNa- mutants exhibited a significantly elevated number of spontaneous seizures compared with veratridine-fed WT flies or dSlo2/KNa- mock-fed flies (Fig. 9C). We also examined whether there were differences in the duration of observed seizures. dSlo2/KNa-;jus- and bss-/+;dSlo2/KNa- double mutants exhibited a wider distribution of seizure durations than single mutant lines (Fig. 9D,E). Similarly, veratridine-fed dSlo2/KNa- mutants displayed a wider range of seizure durations compared with veratridine-fed WT flies (Fig. 9F); all seizure-like events observed in WT flies lasted <3 s, while seizure events in dSlo2/KNa- mutants lasted from 0.7 to 89 s. Because of the low numbers of seizures seen in control lines, however, we were not able to perform meaningful statistical analyses on these distributions. Together, our results suggest that, when dSlo2/KNa channels are absent in models already susceptible to seizure, seizure thresholds are lowered, resulting in an increased incidence of induced as well as spontaneous seizure behavior.
Discussion
In this study, we show that, when dSlo2/KNa channels are absent in genetic and pharmacological models that increase Na+ influx carried by either nAChRs or voltage-gated Na+ channels, effects such as seizure-like behavior, paralysis, and death, are even worse, suggesting that a role of dSlo2/KNa channels is to help prevent these outcomes when the nervous system is overexcited. In particular, our data suggest that dSlo2/KNa channels in Drosophila, like their mammalian counterparts, are activated by the INap current carried by voltage-dependent Na+ channels. We examined four genetic seizure models and a newly introduced pharmacological (veratridine) model with enhanced INap currents. The mechanism underlying the enhanced INap in each of these genetic seizure mutants, however, is different. jus- and eas- mutations affect a novel transmembrane protein of unknown function (Horne et al., 2017) and ethanolamine kinase (Pavlidis et al., 1994), respectively, which have each been reported to affect alternative splicing of the para Na+ channel gene, increasing the inclusion of “exon-L,” which has been shown to be associated with increased INap (Lin et al., 2012). In contrast, the increase in INap in the bss- and GEFS+ lines are because of point mutations in the para gene itself (Parker et al., 2011b; Sun et al., 2012). Notably, mutations that increase INap in mammalian systems have been associated with multiple types of human epilepsy (Kearney et al., 2001; Lossin et al., 2002; Vanoye et al., 2006; Chen et al., 2011). And in all four genetic seizure models and veratridine-fed WT flies, we showed that the loss of dSlo2/KNa channels results in increased seizure susceptibility and severity. The observation that the loss of dSlo2/KNa channels similarly affects multiple INap-affected Drosophila models bolsters the suggestion that the protective role of dSlo2/KNa channels is indeed because of activation by the enhanced INap current.
Interestingly, the degree to which the dSlo2/KNa- null mutation and dSlo2/KNa-DN dominant-negative transgene exacerbated seizure phenotypes of the bang-sensitive mutants, jus-, eas-, and bss-, seems to be inversely related to the reported severity of each mutant phenotype. For example, the jus- mutant has been reported to exhibit the least severe phenotype of the three bang-sensitive mutants we tested (Kuebler and Tanouye, 2000), yet it was in this background that dSlo2/KNa- and dSlo2/KNa-DN resulted in the greatest percent increase in seizure indicators (e.g., times of paralysis, times to full recovery, frequency of seizures, etc.) in both induced and spontaneous seizure-like behavior. In contrast, we saw the mildest changes in seizure indicators with the bss- mutant, which has been reported to have the lowest seizure-threshold and most severe seizure phenotype of the bang-sensitive mutants (Kuebler and Tanouye, 2000). The reason for this trend is unclear, but one speculation is that dSlo2/KNa channels may function protectively as a first line of defense that raises the threshold for seizure induction, but as sustained and more severe overexcitation persists, dSlo2/KNa channels can do little to alleviate ensuing seizure activity. This is a point that may need to be considered when evaluating conditions under which Slo2/KNa channels contribute to seizure threshold in mammalian systems. Little has been reported about the role of Slo2/KNa channels in seizure; in a recent report, the KO of Slo2.2/Slack/KNa1.1 has been shown to result in a decrease in the electric shock stimulus needed to induce a seizure in a hindlimb extension assay (Quraishi et al., 2020).
Although roles for Slo2/KNa channels have been identified in nociceptive, thermal, and mechanical sensation (R. Lu et al., 2015; Martinez-Espinosa et al., 2015; Evely et al., 2017; Tomasello et al., 2017), as well as in reversal and motor learning (Bausch et al., 2015; Quraishi et al., 2020), there is a clear link between Slo2/KNa channels and nonphysiological hyperactivity. For example, there are numerous mutations in human Slo2/KNa KCNT1/KCNT2 genes, which have been associated with multiple forms of epilepsy, including malignant migrating partial seizures of infancy (MMPSI), autosomal dominant nocturnal frontal lobe epilepsy (ADNFLE), Ohtahara syndrome, sudden unexplained death in epilepsy (SUDEP), multifocal epilepsy, leukoencephalopathy, Brugada syndrome, and epilepsy of infancy with migrating focal seizures (EIMFS) (Heron et al., 2012; Ishii et al., 2013; Møller et al., 2015; Gururaj et al., 2017; Ambrosino et al., 2018; Mao et al., 2020). Some of these human mutations have been recapitulated in Xenopus oocytes (Barcia et al., 2012; Kim et al., 2014; Milligan et al., 2014; Tang et al., 2016; McTague et al., 2018), CHO cells (Rizzo et al., 2016; Mao et al., 2020), HEK cells (Ambrosino et al., 2018), as well as human iPSC-derived neurons (Quraishi et al., 2019), revealing both loss- and gain-of-function mutations. An enduring question in the field has been as follows: how do the many gain-of-function KNa channel mutations lead to hyperactivity? Genetically tractable systems, such as Drosophila, may provide in vivo modeling to eventually address this question.
Multiple factors contribute to the complexity of how Slo2/KNa channels affect neuronal excitability. For example, we found that the subcellular distribution of dSlo2/KNa channels varied from one cell type to another, suggesting that dSlo2/KNa channels likely affect excitability differently, depending on cell type. In addition, various factors, including Cl– and NAD+, have been reported to regulate Na+ sensitivity of Slo2/KNa channels (Yuan et al., 2000, 2003 ,Bhattacharjee et al., 2003;; Tamsett et al., 2009). Slo2/KNa activation has also been reported downstream of multiple sources of Na+, including glutamate receptors, H-channels, and cyclic nucleotide gated channels (Nanou and El Manira, 2007; S. Lu et al., 2010; Aoki et al., 2018); little, however, is known about the physical interaction/proximity of Slo2/KNa channels to these channels or Na+ channels, which would greatly influence the responsiveness and role of Slo2/KNa channels. Finally, we found that enhancement of seizure indicators by the dSlo2/KNa- mutation in seizure prone backgrounds was difficult to rescue with exogenous expression of dSlo2/KNa (data not shown); one speculation is that neurons are very sensitive to expressed levels of dSlo2/KNa, making it difficult to titrate the right amount of dSlo2/KNa expression to rescue the dSlo2/KNa- mutant phenotype without overexpressing dSlo2/KNa above WT levels. Regulating precise expression levels of dSlo2/KNa in vivo may be an additional challenge important to understanding how dSlo2/KNa channels affect neuronal excitability.
Our current study supports a model in which one role of dSlo2/KNa channels is to provide a reserve K+ conductance that suppresses seizure induction when the nervous system is confronted with excessive excitation. It will be interesting for future studies to explore what factors distinguish this protective role of dSlo2/KNa channels from roles in sensory processing and learning/memory function.
Footnotes
This work was supported by the National Institutes of Health R01GM083335 to S.T. and R03NS109495 to S.T. We thank Drs. Mark Tanouye, Fengqui Diao, Benjamin White, Patrick Dolph, Lawrence Salkoff, and Diane O'Dowd for plasmids and fly lines.
The authors declare no competing financial interests.
- Correspondence should be addressed to Susan Tsunoda at susan.tsunoda{at}colostate.edu
