Lithium-Responsive Seizure-Like Hyperexcitability Is Caused by a Mutation in the Drosophila Voltage-Gated Sodium Channel Gene paralytic

Outcrossed Shu mutants exhibit morphological and behavioral abnormalities

In the original study by Williamson (1982), phenotypes of Shu mutants were examined using Shu/FM6 female flies, where FM6 is a balancer X chromosome with multiple inversions and marker mutations. It was therefore possible that genetic aberrations in the FM6 balancer chromosome contributed to the previously reported dominant phenotypes of Shu. To rule out this possibility, the following phenotypic analyses were carried out with the progeny of crosses between control males and Shu females with the X chromosome balancer (i.e., Shu/+ and Shu/Y). To minimize potential effects of unidentified genetic variations in the original Shu genome, we used a Shu mutant stock that had been backcrossed to CS for at least 26 generations. Outcrossed Shu males (Shu/Y) survived for only a few days after eclosion and were too sluggish to perform behavioral tasks. Under our standard rearing conditions, Shu males displayed severe defects in courtship behavior and rarely copulated. Consequently, homozygous adult Shu females (Shu/Shu) were not normally found in Shu mutant stocks. Our reassessment revealed morphological and behavioral defects in Shu mutants, some of which had been reported for the original Shu/FM6 stock in Williamson (1982); others are described for the first time in this study.

Shu males (Shu/Y) and heterozygous females (Shu/+, hereafter referred to as Shu or Shu females unless otherwise stated) were found to exhibit an abnormal wing posture (downturned wings) and a cuticular defect on the back (indented thorax), neither of which was observed in wild-type flies (Fig. 1A). These morphological phenotypes manifested shortly after eclosion and at high penetrance in both sexes. In Shu males, 88% and 69% exhibited downturned wings and an indented thorax, respectively (n = 272), whereas in Shu females 73% and 65% displayed the defects (n = 683). Similar morphological abnormalities have been reported in Shaker (Sh) mutants harboring a second genetic lesion in either ether-a-go-go (eag; Ganetzky and Wu, 1983; Engel and Wu, 1992) or receptor oscillation A (rosA; Wu and Wong, 1977) also known as inebriated (ine; Stern and Ganetzky, 1992), or when combined with a para duplication (Stern et al., 1990). Sh and eag encode the α-subunit of voltage-gated potassium channels K_v1 (Kamb et al., 1987; Papazian et al., 1987; Pongs et al., 1988) and K_v10 (Warmke et al., 1991), respectively, whereas rosA/ine encodes the neurotransmitter/osmolyte transporter (Burg et al., 1996; Soehnge et al., 1996). Double mutants for the gene quiver (qvr; Humphreys et al., 1996) and either eag or Hyperkinetic (Hk; Kaplan and Trout, 1969) show similar defects in wing posture and thoracic morphology (Wang et al., 2000). Hk encodes the auxiliary β-subunit for the Sh K_v1 channel (Chouinard et al., 1995), and mutations of Hk and qvr modify the Sh K current (Wang and Wu, 1996; Yao and Wu, 1999; Wang and Wu, 2010). These morphological abnormalities are thought to be caused by an increase in neuronal excitability that leads to hypercontraction of the relevant thoracic muscles (Engel and Wu, 1992; Huang and Stern, 2002).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Shu mutants exhibit abnormalities in morphology, behavior, and lifespan. A, Shu males (Shu/Y) and females (Shu/+) often display downturned wings (arrow head) and an indented thorax (arrow), which are not observed in wild-type flies (Cont). B, 0- to 10-d-old Shu (Shu/+) and wild-type (Cont) female flies were scored for the number of times in 2 min that a strong convulsive episode resulted in the fly being thrown onto its back. Wings of the mutant and control flies were removed shortly after eclosion to eliminate possible effects of downturned wings of Shu mutants on shuddering behavior. Data are mean ± SEM. n ≥ 30. C, The general activity and motor coordination of Shu mutants were assessed by a reactive climbing assay using a countercurrent apparatus. 0- to 10-d-old wild-type (Cont) and Shu (Shu/+) females were used. To ensure that the climbing defect was not simply due to the downturned wing phenotypes of the fly, wings of the mutant and control flies were removed shortly after eclosion. Data are mean ± SEM for 10 groups of 20 flies. D, Representative locomotion traces of wild-type (Cont) and Shu (Shu/+) females in a circular chamber. Compared with wild-type flies, which travel mostly along the edge of the chamber, Shu mutants spent significantly longer times in the center of the chamber. E, Head and hind leg movements under ether anesthesia were video recorded at 15 frames/s, and their velocities (pixels/frame) in wild-type (Cont), Shu (Shu/+), and Sh⁵ females are shown as mean ± SEM. Data for head and leg movements were analyzed using Kruskal–Wallis one-way ANOVA on ranks followed by Dunn’s test, and one-way ANOVA followed by Bonferroni test, respectively. The statistical significance of differences between wild-type (Cont) and each genotype (Shu/+ or Sh⁵) is shown. *p < 0.05; ***p < 0.001. n = 11. F, Survival of CS (Cont) and Shu (Shu/+) females under standard conditions at 25°C, 60–70% humidity. Flies were kept in groups of ∼20. Data are presented as the daily average percentage of surviving flies in each vial with SEM. n = 180 and 176 for CS and Shu, respectively. The median survival times were 72 and 30 d for wild-type and Shu. Two survival curves showed a statistically significant difference (p < 0.001, log-rank Kaplan–Meier survival analysis).

As reported by Williamson (1982), Shu mutants displayed sporadic, spontaneous tremors or “shuddering.” This behavioral abnormality was nearly absent during the first 8 h after eclosion. However, 24 h later, the frequency of severe shuddering episodes (resulting in the fly being thrown onto its back) increased significantly and remained at this level throughout the 10-d period of observation (Fig. 1B). In contrast, wild-type flies displayed these types of episodes only rarely when first introduced into the observation chamber. We also performed a reactive climbing assay using a countercurrent apparatus (Benzer, 1967) to assess the general activity and motor coordination of Shu mutants. Defects in this behavior were clearly evident in Shu mutants soon after eclosion and further deteriorated in an age-dependent manner (Fig. 1C).

In an effort to better quantify the behavioral abnormalities of Shu mutants (e.g., jerking, twitching, and uncoordinated locomotion), we also analyzed their movements using an automated system originally developed for Drosophila sleep analysis (Gilestro and Cirelli, 2009). When placed in a circular chamber, wild-type flies spent most of their time along the periphery (Besson and Martin, 2005), resulting in tracking patterns resembling a circle. In contrast, the uncoordinated movements and spontaneous jerking of Shu mutants led to their increased presence in the center part of the chambers, as represented by a typical tracking trace (Fig. 1D). During the 10-min recording period, Shu mutants spent a significantly longer time in the center than wild-type flies. Typical median percentages of time spent in the center of the chamber were 4.8% for wild-type flies (n = 63) and 39.5% (n = 53) for Shu mutants (P < 0.001).

Shu mutants display an ether-induced shaking phenotype and reduced lifespan

Shu mutants were reported to exhibit leg-shaking behavior under ether anesthesia (Williamson, 1982). This behavioral phenotype has previously been observed in several other mutants with enhanced neuronal excitability, such as the aforementioned Sh, eag, Hk, and qvr mutant strains (Kaplan and Trout, 1969; Wang et al., 2000). To analyze this behavioral phenotype in a semi-quantitative manner, we recorded ether-induced movements of the head and hind leg using a video-tracking system (see Materials and Methods). Shu mutants vigorously moved the head as well as the legs under ether anesthesia, resulting in a drastic increase in the velocities of their movements compared with wild-type flies (Fig. 1D). The ether-induced shaking phenotype of Shu was similar to, but distinct from, that of Sh⁵, a neomorphic mutant allele of Sh (Salkoff and Wyman, 1983; Haugland and Wu, 1990), in that Shu had a tendency to waggle the head more extensively than Sh (Fig. 1E), and that the leg movements of Shu were sporadic, whereas those of Sh were rather continuous. Moreover, legs severed from ether-anesthetized Sh continued to shake (Ganetzky and Wu, 1985) whereas those severed from Shu did not, suggesting that Shu’s ether-induced leg shaking phenotype likely has a central origin.

Shortened lifespan is a common feature of Drosophila mutants that display ether-induced leg-shaking, such as Sh, Hk, and qvr (Trout and Kaplan, 1970; Koh et al., 2008; Bushey et al., 2010). Similarly, the median survival time of Shu mutants was significantly reduced (30 d, n = 176) compared to that of wild-type flies (72 d, n = 180; p < 0.001, Fig. 1F).

DLMs of Shu mutants exhibit drastically enhanced spontaneous activity and ECS seizure discharges

The behavioral and morphological phenotypes of Shu mutants suggest that the excitability of their motor circuits is enhanced. The dorsal longitudinal muscles are particularly amenable to examination of changes in motor circuit function, because they serve as outputs of the motor neurons that integrate inputs, including those from the flight pattern generator (Harcombe and Wyman, 1977), and the GF circuit mediating the jump-and-flight escape reflex (Tanouye and Wyman, 1980; Fig. 2A). DLMs’ large size and isometric contraction enable prolonged monitoring of their action potentials (Ca²⁺ spikes). Importantly, this system has revealed alterations in basic physiological properties caused by a number of hypo- and hyper-excitable mutations (Siddiqi and Benzer, 1976; Elkins et al., 1986; Engel and Wu, 1992, 1998; Iyengar and Wu, 2014).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Shu mutants display hyperexcitable seizure-like discharges in the electrical output of an identified motor neuron. A, The dorsal longitudinal motor neuron (DLMn) receives input from a number of pattern generators, and its activity may be monitored via spiking activity in the dorsal longitudinal muscle (DLM). B, Example traces of spontaneous activity in Shu males (Shu/Y) and females (Shu/+; right) compared with wild-type control individuals (left). C, Scatterplot of spontaneous firing rates in control (+/Y, +/+) and Shu (Shu/Y, Shu/+) populations. Each gray circle represents a 1-min recording of activity, black circles (right) represent the median, and vertical bars represent the 25th–75th percentile interval (n > 10 individuals for each group). **p < 0.01; ***p < 0.001, Kruskal–Wallis ANOVA, rank-sum post hoc test. D, DLM firing, monitored as readout of stereotypic seizure-like discharges triggered by high-frequency, high-voltage electroconvulsive stimulation across the brain (ECS, see Materials and Methods). Sample ECS-triggered discharges in control (Cont) and Shu/+ individuals are displayed. E, Plot of ECS voltage and number of spikes per discharge, a measure of seizure intensity. Shu/+ individuals display significantly more intense seizure discharges at 60 and 80 V compared with control (Cont) flies. ***p < 0.001, Kruskal–Wallis ANOVA, rank-sum post hoc test.

Increased spontaneous nonflight DLM activity is a basic phenotype associated with hyperexcitable mutants, including the double potassium channel mutant, eag Sh (Engel and Wu, 1992). Figure 2 illustrates this spontaneous activity in Shu mutant DLMs. We found that heterozygous Shu females displayed a significantly increased spike frequency compared with wild-type females (0.99 ± 0.24 Hz vs. 0.50 ± 0.10 Hz, p < 0.01; Fig. 2B, C). In hemizygous Shu males, this phenotype was much more striking compared with wild-type males (5.56 ± 0.93 Hz vs. 0.53 ± 0.08 Hz, p < 0.001) and even greater than that previously observed in eag Sh double mutants (∼3.5 Hz; Engel and Wu, 1992). In contrast to these findings in spontaneous activities, analysis of the GF pathway–mediated responses to electrical stimulation across the brain revealed relatively mild effects of the Shu mutation. We examined both male and female Shu individuals for DLM spike initiation latency and the interpolated stimulus frequency at which 50% DLM responses failed (FF₅₀). We found only a slightly retarded DLM spike latency in Shu males and a modest increase in FF₅₀ in Shu females compared with wild-type flies (Table 1).

The same recording configuration was used to examine the ECS-triggered stereotypic seizure-like discharge pattern. High-intensity, high-frequency ECS (up to 80 V, 200 Hz) across the brain evokes seizure activities characteristic of wild-type and a variety of mutant flies (Pavlidis and Tanouye, 1995; Lee and Wu, 2002, 2006; Ueda et al., 2008; Parker et al., 2011; Ehaideb et al., 2014). We quantified the spike patterns immediately after the ECS (initial discharge; Lee and Wu, 2002) and found that in Shu females’ ECS-triggered discharges were far more intense than those of wild-type females in both number of spikes per discharge (Fig. 2D, E) and duration of spike discharges. This duration was consistently >5 s in Shu females (8 of 10 flies), which was rarely observed in wild-type flies. However, the ECS intensity threshold did not differ significantly between Shu and wild-type females (∼60 V; see Materials and Methods for details).

Shu maps to the Drosophila voltage-gated sodium channel gene paralytic

The Shu mutation was previously mapped to the X chromosome, at the genetic map position 55.1 with respect to the vermilion and forked loci (Williamson, 1982), leaving us with a rough estimate of where the lesion might be located. We undertook a recombination-based mapping approach using molecularly defined P-element insertions to narrow down the chromosomal position of the Shu mutation (Zhai et al., 2003). To do this we recombined an eye color marker, the white mutation (w), into the Shu-carrying chromosome and scored nearly 50,000 F2 flies for the recombination events between Shu mutation and six nearby P-element insertions containing the w⁺ mini gene (Table 2). Calculation of the recombination rates revealed that the P-element insertions EY05861 and EY00852, nearest the gene para (CG9907), provided the lowest recombination rates (0.18 and 0.22%, respectively), whereas insertions more distant in either direction from cytological region 14E resulted in higher rates of recombination (Table 2). Based on the molecular distance between the defined P-element insertion sites and the recombination rates between these markers and the Shu mutation, the mutation site was estimated between the insertion sites of two particular P-elements used in our screen: P[EPgy2]CG9902^EY05861 and P[EPgy2]Rbp2^EY00852. All combinations pairing one insertion distal, and one proximal, to cytological region 14E resulted in an estimated Shu mutation site in the 72-kb para locus (Fig. 3A), which encodes for the Drosophila voltage-gated sodium channel (Loughney et al., 1989).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Shu maps to the voltage-gated sodium channel gene paralytic. A, Mapping positions of the Shu mutation on the X chromosome. Red triangles and horizontal lines represent pairs of molecularly defined P-element insertions used to estimate the Shu mutation site; red X’s indicate the mutation sites deduced from the recombination rates between the corresponding P-element insertion and Shu. The estimated sites all reside within the para locus (CG9907). Boxes designate annotated genes near the para locus (based on FlyBase). B, DNA sequencing chromatogram identifying a G-to-A transition mutation (arrowheads) in the Shu genome at the position corresponding to the nucleotide 4249 in the para-RE cDNA (FlyBase). This mutation results in a methionine-to-isoleucine substitution at the amino acid position 1327. C, Schematic structural diagram of a Drosophila voltage-gated sodium channel. Arrow indicates the Shu mutation in the transmembrane segment S2 in homology domain III. The para^GEFS+ mutation K1330T, which corresponds to a SCN1A mutation K1270T causing GEFS+ in humans (Sun et al., 2012), lies three codons away from that of Shu. Also shown are the para^DS mutation S1291R and the para^bss1 mutation L1676F. D, Amino acid sequence alignment of Na_v channels of different animal species. Note that the methionine residue, which is mutated to isoleucine in Shu, is present in all Na_v channels.

Shu carries a missense mutation in the paralytic coding region

Because our recombination mapping indicated that the Shu mutation is located within the para locus, we sequenced all 31 para coding exons using genomic DNA from extensively backcrossed (40 generations) hemizygous Shu males. Our results revealed a single nucleotide change in Shu, a G-to-A transition in para exon 24 (Fig. 3B) that results in the substitution of an isoleucine for a methionine at amino acid 1327 (based on para-PE, FlyBase). We confirmed that this mutation was also present in the genomic DNA isolated from the original Shu stock obtained from R. Williamson, but not in the wild-type strain used to isogenize the mutant stock. M1327I lies in the predicted transmembrane segment S2, within homology domain III of the sodium channel (Fig. 3C). Although various splicing variants are produced from the para locus, the exon containing the Shu mutation is constitutively spliced (Olson et al., 2008; Lin et al., 2009). Therefore, this amino acid replacement must be present in all Na_v isoforms in Shu mutants. In addition, the methionine mutated in Shu represents a highly conserved residue located in the Na_v channel α subunit proteins of evolutionarily diverse animal species (Fig. 3D), including all nine human Na_v channel α subunit isoforms (Yu and Catterall, 2003), strongly indicating that it is important for Na_v channel function.

Targeted expression of the Shu Na_v channel partially phenocopies Shu phenotypes

To directly examine whether the identified Shu mutation in para is sufficient to cause the mutant phenotypes, we generated multiple fly lines carrying a para cDNA transgene with or without the Shu mutation (UAS-Shu and UAS-para⁺). When five independent UAS-Shu or UAS-para⁺ lines were crossed to the pan-neuronal driver elav-Gal4, the progeny did not show obvious morphological or behavioral abnormalities resembling those of Shu mutants. However, when UAS-Shu was crossed with C164-Gal4, an enhancer trap Gal4 line that drives gene expression widely in the nervous system including in motor neurons (Torroja et al., 1999; Jepson et al., 2012), a significant portion of the resulting progeny showed morphological defects similar to those seen in Shu mutants (Fig. 4A). Namely, 24 and 39% of the progeny of UAS-Shu lines 2 and 4, respectively, exhibited downturned wings. Such defects were rarely seen in any progeny of UAS-para⁺ lines. The progeny of C164-Gal4 and UAS-Shu with the morphological abnormalities were further subjected to behavioral analysis. They were found to display spontaneous jerking and twitching, similar to that observed in Shu mutants. In general, their locomotion was disorganized, as demonstrated by the video tracking method (Fig. 4B). The severity of the phenotype varied considerably among individuals, and only the progeny of UAS-Shu line 2 showed a statistical difference in this behavioral parameter (i.e., time spent in the center) compared with wild-type flies when the Mann–Whitney U-test was applied (p < 0.001). However, the progeny of UAS-Shu line 4 also exhibited abnormal behavior based on the observation that 14% of the tested progeny of UAS-Shu line 4 (9 of 66) spent >20% of their time in the center, whereas no control flies ever passed this time threshold (0 of 40). These results demonstrated that targeted expression of the Shu Na_v channel can mimic, at least partially, the morphological and behavioral abnormalities observed in Shu mutants.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Targeted expression of the Shu Na_v channel partially phenocopies Shu phenotypes. A, Effects of expression of the wild-type (UAS-para⁺) and Shu (UAS-Shu) Na_v channels using C164-Gal4 on morphological phenotypes. Indicated are percentages of flies with downturned wings (closed) or an indented thorax (open). The progeny of four of five independent UAS-Shu lines displayed downturned wings or an indented thorax, whereas the progeny of UAS-para⁺ lines rarely displayed the morphological defects. B, Effects of expression of the Shu Na_v channels on behavioral phenotypes. Percentages of time spent in the center of the chamber are shown for the progeny of C164-Gal4 and UAS-Shu lines. Sample numbers (n) are indicated in parentheses.

Functional interactions between Shu and other para mutant alleles

To infer the functional nature of the Shu mutation in para, we examined genetic interactions between Shu and other para mutant alleles, para^ts1, para^ts115, para^bss1, and para^GEFS+. Previous studies demonstrated that para^ts1 and para^ts115 exhibit recessive, temperature-induced paralysis (Suzuki et al., 1971; Song and Tanouye, 2007), whereas para^bss1 shows semidominant bang-sensitive paralysis (Parker et al., 2011). Generalized epilepsy with febrile seizures plus (GEFS+) is a common childhood-onset genetic epilepsy syndrome, often caused by a mutation in the human Na_v channel gene SCN1A (Scheffer and Berkovic, 1997). para^GEFS+ was generated by introducing a para mutation (K1330T) mimicking the GEFS+-causing human SCN1A K1270T mutation (Fig. 3C) and was found to cause a semidominant, heat-induced seizure phenotype when exposed to 40°C (Sun et al., 2012).

Shu mutant females were crossed to para^ts1, para^ts115, para^bss1, or para ^GEFS+ males, and the female progeny trans-heterozygous for Shu and another para mutant allele (i.e., Shu/para^mutant) were subjected to phenotypic analysis. We found that the defect in wing posture was significantly suppressed in Shu females trans-heterozygous for either para^ts1 or para^ts115 (Fig. 5A). Consistently, Shu/para^ts1 and Shu/para^ts115 flies showed improved behavioral phenotypes, spending considerably less time in the center of the chamber than Shu females (p < 0.001; Fig. 5B). Because a previous report demonstrated that para^ts1 and para^ts115 show hypomorphic function even at permissive temperature (Ganetzky, 1984), these results indicated that a general suppression of para function reduces the severity of the morphological and behavioral phenotypes of Shu mutants. On the other hand, the phenotypes of Shu mutants were not suppressed when Shu was combined with para^GEFS+ (Fig. 5A, B). Rather, Shu/para^GEFS+ flies displayed a more severe behavioral phenotype than Shu alone (p < 0.001, Fig. 5B). A previous study had reported that the knock-in GEFS+ mutation para^GEFS+ studied here reduces the activation threshold regardless of temperature (Sun et al., 2012). The worsening effect of para^GEFS+ on Shu phenotypes is thus consistent with the hypomorphic alleles, para^ts1 and para^ts115, having a suppressive effect. In this context, it was unexpected that Shu/para^bss1 displayed significantly suppressed morphological (Fig. 5A) and behavioral (p < 0.001; Fig. 5B) phenotypes relative to Shu/+ flies, because, unlike para^ts1 or para^ts115, para^bss1 was reported to be a gain-of-function allele (Parker et al., 2011).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Functional interactions among Shu and other mutant alleles of para. A, Effects of different para mutations on Shu morphological phenotype. Indicated are percentages of wild-type (+/+), Shu (Shu/+), and trans-heterozygous (Shu/para^mutant) females exhibiting the normal wing posture (gray) and the downturned wing phenotype (black). B, Effects of different para mutations on Shu behavioral phenotype. Percentages of time spent in the center of the chamber are shown for wild-type (+/+), Shu (Shu/+), and trans-heterozygous (Shu/para^mutant) females. The Mann–Whitney U test was used to assess the effect of each para allele on the behavioral phenotype of Shu. The statistical significance and sample numbers (n) are indicated. ***p < 0.001.

Neuron-specific knockdown of para leads to suppression of the Shu phenotypes

The observation that para hypomorphic alleles partially rescued the phenotypes of Shu mutants (Fig. 5) prompted us to examine the effect of reduced para expression on the Shu phenotypes. Shu mutants were subjected to knockdown of para expression using UAS-para-RNAi lines obtained from Vienna Drosophila Resource Center (GD6131 and GD6132). Flies expressing GD6131 para-RNAi in neurons exhibited slightly delayed eclosion, but the adults appeared healthy. Interestingly, this genetic manipulation in Shu mutants resulted in nearly complete rescue of the downturned wing phenotype (Fig. 6A). The behavioral phenotype was drastically improved as well (Fig. 6B). Expression of para-RNAi in muscle cells using the muscle-specific driver Mhc-Gal4 did not lead to improvement of any aspects of the mutant defects examined. In support of the results with GD6131, we observed similar rescue effects when using another para-RNAi line, GD6132, to suppress para expression in Shu mutants (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Neuron-specific expression of para RNAi significantly improves the Shu mutant phenotypes. A, Effects of neuron-specific knockdown of para expression on Shu morphological phenotypes. Indicated are percentages of control and para RNAi-expressing flies displaying the normal wing posture (gray) and the downturned wing phenotype (black). B, Effects of neuron-specific knockdown of para expression on Shu behavioral phenotypes. Percentages of time spent in the center of the chamber are shown for control and para RNAi-expressing flies. The Mann–Whitney U test was used to assess the effect of each para allele on the behavioral phenotype of Shu. elav, neuron-specific elav-Gal4; Mhc, muscle-specific Mhc-Gal4. The statistical significance and sample numbers (n) are indicated. *p < 0.05; **p < 0.01; ***p < 0.001.

Lithium reduces the severity of Shu’s behavioral phenotypes

One of the remarkable observations by Williamson (1982) was that the behavioral phenotypes of Shu/FM6 flies were significantly suppressed by feeding the mutants lithium-containing food. To examine the effect of lithium on outcrossed Shu mutants, we fed the adult flies food supplemented with different concentrations of LiCl. Consistent with the previous report (Williamson, 1982), a 5-d treatment with 100 mm LiCl during adulthood reduced the frequency of shuddering (Fig. 7A). In addition, food containing lower concentrations of LiCl (25 and 50 mm) was also effective in suppressing shuddering behavior (Fig. 7A). The reduction in shuddering in response to lithium is not due to a general suppression of motor activity, because the same treatment considerably increased coordinated motor activity of Shu mutants in a reactive climbing assay (Fig. 7B). This improvement was observed in a dose-dependent manner, with higher climbing indices correlating with an increase in LiCl concentration (25, 50 and 100 mm) in the food. In sharp contrast to the normalizing effects of lithium on Shu mutant behavior, wild-type flies remained largely unaffected. However, after 5 d of receiving the highest concentration of lithium tested, wild-type flies began to display subtle signs of retarded locomotion (Fig. 7B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Lithium reduces the severity of behavioral phenotypes of Shu mutants. A, Shuddering behavior in wild-type (open) and Shu females (closed) after 5-d treatment with food containing 0, 25, 50, or 100 mm LiCl. Data are mean ± SEM. n ≥ 30. The statistical significance of differences between the control and each lithium treatment was assessed using the Mann–Whitney U test. *p < 0.05; **p < 0.01; ***p < 0.001. B, Reactive climbing ability of wild-type (open) and Shu females (closed) after the same LiCl treatment. Data are mean ± SEM of 10 groups of 20 flies. The statistical significance of differences between the control and each lithium treatment was assessed using the Student’s t-test. ***p < 0.001. C, Percentages of time spent in the center of the chamber by wild-type (Cont), Shu (Shu/+), and Sh⁵ females with or without lithium treatment. Kruskal–Wallis one-way ANOVA on ranks was performed followed by post hoc analysis using Dunn’s method. Sample numbers (n) and the statistical significance between lithium-treated and control flies are indicated. ***p < 0.001; NS, not significant (p > 0.05).

Consistently, Shu mutants fed food supplemented with 100 mm lithium spent significantly less time in the center of the chamber compared with those maintained on standard food (p < 0.001; Fig. 7C). The potassium channel mutant, Sh⁵, also exhibited jerking and twitching, albeit less frequently and less intensely than Shu. Like Shu, Sh⁵ mutants spent more time in the center than wild-type flies (control vs. Shu⁵, no lithium, p < 0.001). However, unlike Shu, these same parameters in Sh⁵ were not significantly altered by lithium treatment (Fig. 7C).

As shown in Figure 1F, Shu mutants were short-lived. Although lithium had a rescue effect on Shu behavioral phenotypes (Fig. 7), it did not positively impact their longevity. Rather, Shu lifespan was decreased when adults were maintained on food containing 50 and 100 mm LiCl. Furthermore, even at the lower concentrations of lithium supplementation (i.e., 1 and 10 mm), the drug failed to extend the lifespan of the mutants. The median survival times of Shu mutants in this set of experiments were 13 (n = 100), 22 (n = 150), 36 (n = 172), 30 (n = 147), and 44 (n = 139) d for 100, 50, 10, 1, and 0 mm LiCl.

Shu mutants accumulate higher levels of lithium

Although lithium treatment partially rescued behavioral phenotypes of Shu mutants, it had little or even a worsening effect on the behaviors of wild-type flies (Fig. 7A–C). To determine whether there was any difference between the two genotypes with respect to physiological responses to lithium, the internal concentrations of lithium in flies fed lithium-containing food for 5 d were analyzed. Shu mutants were found to accumulate higher levels of lithium than wild-type flies when fed food containing either 25 or 50 mm LiCl (p < 0.05; Table 3). A similar trend was observed when 100 mm LiCl was used, although the difference between the genotypes was not statistically significant. Our wild-type data were comparable to internal lithium concentrations determined in previous studies under similar conditions (Padiath et al., 2004; Dokucu et al., 2005).

Different para mutants are affected by lithium in an allele-specific manner

As mentioned above, the para mutation in a Drosophila GEFS+ model (para^GEFS+) is located near the site of the Shu mutation (Fig. 3C; K1330T vs. M1327I), presenting the possibility that Shu might also exhibit heat-induced seizures. Therefore, we tested Shu mutants and found that they respond to increased temperature in a manner similar to that previously observed in GEFS+ flies—albeit in a much more severe manner. Seizure-like behaviors were induced in almost all Shu mutants within 30 s of exposure to 40°C, whereas during the same time frame, only a few para^GEFS+ heterozygotes and <50% of para^GEFS+ homozygotes showed the phenotype (Fig. 8A). Unlike GEFS+ flies, Shu mutants exhibited seizures even at 37°C (Fig. 8B). As was the case for other behavioral abnormalities of Shu mutants, the heat-induced phenotype was alleviated when the flies were fed food containing 100 mm lithium (Fig. 8B). Furthermore, lithium had a comparable rescue effect on para^GEFS+ homozygotes, significantly reducing the severity of their heat-induced seizure phenotype (Fig. 8C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Shu mutants and GEFS⁺ flies exhibit a heat-induced seizure phenotype that is suppressed by lithium administration. A, Individual flies were put into glass vials that were submerged in a water bath (40°C) for 2 min. Each fly was examined for seizure status (seizing or not seizing) every 5 s. Three independent experiments were carried out, and 10–30 flies were analyzed in each experiment. The average percentages of seizing flies at each time point (±SEM) are shown for Shu (Shu/+) and para^GEFS+ mutants (homozygotes and heterozygotes). Shu mutants showed a heat-induced seizure phenotype similar to, but much more robust than, that of para^GEFS+ mutants. The average percentages of seizing flies at each time point (± SEM) are shown for Shu mutants at 37°C (B), para^GEFS+ mutants (homozygotes) at 40°C (C), and para^DS mutants (homozygotes) at 38°C (D) after treatment with food with (+) or without (–) 100 mm LiCl. Three independent experiments were carried out, and 27–30 flies were analyzed in each experiment. Control flies (CS or the genetic background strain for para^GEFS+ and para^DS flies) did not seize under these conditions. LiCl treatment significantly suppressed the heat-induced seizure phenotype of Shu and GEFS⁺ mutants (two-way repeated-measures ANOVA, Holm–Sidak multiple comparisons, ***p < 0.001). E, Time required for control (n = 73) and 100 mm lithium–treated (n = 46) para^ts1 mutants to become paralyzed at 34°C. LiCl treatment significantly enhanced the heat-induced paralysis phenotype. F, Time required for control (n = 51) and 100 mm lithium–treated (n = 56) para^bss1 mutants to recover from mechanical shock-induced paralysis. LiCl-treated para^bss1 mutants displayed reduced recovery time. Mann–Whitney U test. ***p < 0.001.

Because lithium feeding suppressed the phenotype of para^GEFS+, we wondered whether phenotypic severity of other para mutants was similarly improved by lithium. para^DS is a fly model of human epilepsy generated by Schutte et al. (2014). This knock-in allele carries the mutation in para (S1291R) that corresponds to the human SCN1A mutation (S1231R) causing a severe seizure disorder termed Dravet syndrome (DS). Drosophila para^DS mutants exhibit heat-induced seizures with onset temperature lower than para^GEFS+ mutants (Schutte et al., 2014). We examined the heat-induced phenotype of para^DS with or without feeding 100 mm lithium and found that lithium had no rescue effect (Fig. 8D). Likewise, lithium treatment did not lead to phenotypic improvement in para^ts1 mutants. Rather, para^ts1 mutants paralyzed in shorter time at high temperature when they were treated with lithium (Fig. 8E). In contrast, para^bss1 mutants were found to recover from mechanical shock–induced paralysis more rapidly when they were fed with lithium-containing food (Fig. 8F). In summary, among para mutant alleles examined in this study, lithium improved the phenotypes of Shu, para^GEFS+, and para^bss1 but not those of para^DS and para^ts1.

Effects of lithium on abnormal seizure-like discharge phenotypes of Shu mutants

The apparent behavioral improvement after lithium treatment prompted us to examine DLM activity phenotypes of Shu mutants in this context to determine how each was affected. We compared flies fed 50 mm LiCl to those receiving control food and found that LiCl feeding influenced only selected electrophysiological phenotypes. For instance, it did not significantly reverse the characteristic spontaneous DLM activity seen in Shu flies (mean spike rates for control vs. LiCl fed: 0.97 Hz, n = 8, vs. 0.93 Hz, n = 10, p = 0.36, rank sum test). However, for ECS-triggered discharges, the number of DLM spikes per discharge was reduced in LiCl-fed Shu flies compared with controls (Fig. 9A). At a stimulus intensity of 80 V, LiCl-fed Shu flies displayed discharges that were significantly milder in terms of spike count compared with control Shu flies (Fig. 9B; p < 0.05). A tendency of reduced spike numbers was observed at 60-V stimulation, although the difference did not reach statistical significance. Therefore, our electrophysiological observations were consistent with the behavioral evidence that LiCl feeding can improve aspects of the seizure phenotypes of Shu flies.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Chronic and acute effects of lithium on the severity of electrophysiological phenotypes of Shu mutant adults and larvae. A, Sample traces of ECS discharges (stimulation voltage as indicated) in control Shu/+ flies (0 mm) and lithium-treated Shu/+ flies (50 mm). B, Feeding Shu/+ flies LiCl-supplemented food reduced ECS-triggered spikes per discharge in Shu/+ flies. At 60-V stimulation, the trend was not significant (p = 0.24), whereas at 80 V, the ECS discharges were significantly reduced in lithium-treated Shu/+ flies (*p < 0.05, Kruskal–Wallis one-way ANOVA, rank-sum post hoc test). C, Acute extracellular LiCl application in wild-type larvae did not evoke spontaneous EJP activities (<10 min). 1 and 70 mm LiCl was added, with NaCl levels adjusted to maintain ionic strength. All larval recordings were done with extracellular Ca²⁺ at 0.5 mm. D, Shu homozygous larvae displayed spontaneous discharges of EJPs, associated with motor axon firing (5–20 Hz). Inset, Individual EJPs were coupled with a motor axon action potential. E, Representative traces of Shu neuromuscular activity upon extracellular application of Li⁺ at 1 and 70 mm. The spontaneous firing phenotype did not weaken upon application of LiCl. Inset, Expanded EJP traces at the onset of spontaneous activities.

Behavioral suppression of adult Shu mutants required feeding of lithium for a few days to observe chronic improvements (Williamson, 1982). Similarly, DLM physiology also indicated a lack of acute effects immediately after lithium administration. To directly assess the acute effects of lithium application, we used the larval neuromuscular preparation and observed action potential and neuromuscular transmission. In contrast to wild-type larvae, which did not display spontaneous EJPs with motor axons severed from the CNS (Fig. 9C), the surviving third-instar Shu/Shu larvae displayed extreme hyperexcitability, as indicated by spontaneous discharges of EJPs, correlated with bursting of motor axon action potentials, known to be mediated by para Na_v channels (Fig. 9D). Addition of 1 mm LiCl to the saline did not cause acute suppression in excitability (abnormal spontaneous EJPs persisted up to 20 min; Fig. 9E). This concentration of lithium is approximately comparable to the internal concentration after LiCl feeding to adult flies with 100 mm in the medium (Padiath et al., 2004; Dokucu et al., 2005). Effective serum lithium concentrations for treating patients with BPD are also in a range of ∼1 mm (Severus et al., 2008). In classic studies, Li⁺ is known to block the Na⁺/K⁺ ATPase pump but can also serve as a highly efficient charge carrier through Na_v channels, with permeability higher than Na⁺ (Hille, 2001). We replaced Na⁺ with Li⁺ in saline (70 mm) and found that this drastic treatment did not cause any substantial acute effect on hyperexcitability. Further, even after prolonged incubation (up to 20 min), it only led to a new pattern of extreme hyperexcitability, i.e., giant plateaued EJPs (not shown), reminiscent of those in eag Sh hyperexcitable K⁺ channel double mutants caused by high-frequency bursting of motor axon action potentials (Ganetzky and Wu, 1982). Consistently, LiCl at 1 or 70 mm did not produce any acute effect on wild-type larvae to generate spontaneous EJPs (Fig. 9C).

Effects of lithium treatment on gene expression in the adult heads

Our findings suggested that lithium’s suppressive effect on Shu phenotypes is not due to acute actions of lithium but rather through long-term physiological adjustments to lithium treatment. Such adjustments possibly involve alterations in gene expression. To explore this possibility, we performed microarray analysis and investigated the effect of the Shu mutation and lithium treatment on genome-wide gene expression in adult heads. Gene expression profiles were compared between genotypes (wild-type and Shu) and treatment (with or without lithium treatment) using Affymetrix GeneChip Drosophila Genome 2.0 Arrays as described in our previous study (Kasuya et al., 2009b). Three biological replicates were tested for each condition and showed high correlation coefficients (R > 0.93), indicating that the experimental data were sufficiently reproducible and reliable.

When gene expression profiles were compared between Shu mutants and genetically matched wild-type flies, 17 genes displayed a significant difference (p < 0.05, Welch t-test) in transcript levels with a fold change >2 (Table 4). Among them, 14 and three genes were up- and downregulated in Shu mutants, respectively. Intriguingly, seven of the 14 upregulated genes are directly involved in the innate immune response (Table 4). These include six genes encoding for antimicrobial peptides (AMPs; Lemaitre and Hoffmann, 2007): Diptericin B (DptB), Attacin A (AttA), Attacin B (AttB), Attacin C (AttC), Cecropin B (CecB), and Cecropin C (CecC). PGRP-SB1, another upregulated gene in Shu, encodes for a peptidoglycan recognition protein that functions upstream of the signaling cascades regulating the systemic production of AMPs (Lemaitre and Hoffmann, 2007). In addition to these bona fide immune genes, CG42807 and CG32368 were significantly upregulated in Shu. Although the function of these genes is not known, they were reported as two of the most upregulated genes in response to the endogenous presence of microbiota. They are tentatively annotated to encode small peptides, potentially serving as novel immune effectors (Broderick et al., 2014). Furthermore, one of the three calcineurin isoforms, calcineurin A at 14F (CanA-14F) was upregulated in Shu. CanA-14F, along with the other two calcineurin genes in Drosophila, has been shown to play a role in innate immune signaling (Dijkers and O’Farrell, 2007; Li and Dijkers, 2015). Upregulation of immune gene expression was also observed in two seizure-prone para knock-in mutants, para^DS and para^GEFS+. They displayed elevated expression of AttC, CecC, DptB, and PGRP-SB1. However, unlike Shu, these mutants did not show higher expression of AttA and CanA-14F compared with genetically matched control flies (Fig. 10).

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

Expression of innate immune response genes is upregulated in para^GEFS+ and para^DS mutants. Expression levels of the selected innate immune response genes were examined in control flies and two para knock-in mutants, para^GEFS+ and para^DS, by quantitative real-time RT-PCR. Data are normalized transcript levels presented as mean ± SEM of five to eight independent experiments. *p < 0.05; **p < 0.001.

We next examined whether lithium treatment has any effect on elevated levels of immune gene expression in Shu. As shown in Figure 11A and B, treatment with lithium had a general propensity to normalize expression of immune-related genes, although CanA-14F transcript levels in Shu remained higher than those in wild-type flies even after lithium treatment. Quantitative analysis of RT-PCR results confirmed that AMP genes and PGRP-SB1 were down-regulated by lithium treatment in Shu mutants but not in the wild-type flies (Fig. 11C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 11.

Lithium suppresses the increased expression of innate immune response genes in Shu mutants. A, Expression levels of innate immune response genes and CanA-14F in wild-type females from microarray analysis were compared with those in Shu mutants with or without 50 mm lithium treatment for 24 h. Data are presented as normalized average signal intensities of three biological replicates for each condition. Expression levels of the selected innate immune response genes were examined by RT-PCR in wild-type and Shu mutants with or without 50 mm lithium treatment for 24 h. A representative agarose gel (B) and ratios between transcript levels in control and lithium-treated flies based on pixel intensity of the bands (C) are shown (CS, open bar; Shu/+, closed bar). Data are mean ± SEM of three independent experiments. *p < 0.05; **p < 0.001; NS, not significant.

We analyzed microarray data to further assess the effect of lithium on genome-wide gene expression in Shu. When a comparison was made between Shu mutants with and without lithium treatment, 20 and 5 genes were up- and down-regulated by lithium with p < 0.05 and a fold change >2 (Table 5). Using similar criteria, we previously reported that 11 genes were most significantly up-regulated by lithium treatment in wild-type flies (Kasuya et al., 2009b). Noticeably, 9 of the 11 upregulated genes in wild-type flies were also found to be upregulated in Shu. These include genes potentially involved in amino acid transport and metabolism (Lithium-inducible SLC6 transporter/List and CG1673), detoxification and stress response (CG5999 and Activity-regulated cytoskeleton associated protein 1/Arc1), folate-dependent metabolism and ROS regulation (NAD-dependent methylenetetrahydrofolate dehydrogenase/Nmdmc), and oxidation-reduction process (Cyp309a1; Flybase). Although lithium-inducible genes are largely shared by Shu mutants and wild-type flies, none of the significantly downregulated genes were found to be in common (Kasuya et al., 2009b). The five genes downregulated by lithium in Shu include two antibacterial or antifungal peptide genes (Diptericin B/DptB and Metchnikowin/Mtk) and two oxidoreductase genes (deadhead/dhd and CG5653; Flybase).

We also applied a DAVID functional annotation tool (https://david.ncifcrf.gov/tools.jsp) to identify biological pathways likely associated with changes in lithium-induced gene expression. Using less stringent criteria (false discovery rate <0.5 and fold change >1.2) than the aforementioned analysis, 158 genes were identified as differentially regulated genes in Shu with and without lithium treatment. Analysis using the DAVID functional annotation chart program identified 19 partially redundant gene ontology (GO) term entries as highly enriched in a list of lithium-responsive genes with p < 10⁻⁵ (Table 6). Interestingly, amino acid metabolism represented by multiple GO terms (e.g., 0008652, 0046394, 0006564, 0009066) stood out as the biological process most significantly affected by lithium treatment. Genes involved in carbohydrate metabolism and redox reactions were also found to be strongly affected when Shu mutants were treated with lithium.