In Vivo Analysis of a Gain-of-Function Mutation in the Drosophila eag-Encoded K⁺ Channel

Abstract

Neuronal Na⁺ and K⁺ channels elicit currents in opposing directions and thus have opposing effects on neuronal excitability. Mutations in genes encoding Na⁺ or K⁺ channels often interact genetically, leading to either phenotypic suppression or enhancement for genes with opposing or similar effects on excitability, respectively. For example, the effects of mutations in Shaker (Sh), which encodes a K⁺ channel subunit, are suppressed by loss-of-function mutations in the Na⁺ channel structural gene para, but enhanced by loss-of-function mutations in a second K⁺ channel encoded by eag. Here we identify two novel mutations that suppress the effects of a Sh mutation on behavior and neuronal excitability. We used recombination mapping to localize both mutations to the eag locus, and we used sequence analysis to determine that both mutations are caused by a single amino acid substitution (G297E) in the S2–S3 linker of Eag. Because these novel eag mutations confer opposite phenotypes to eag loss-of-function mutations, we suggest that eag^G297E causes an eag gain-of-function phenotype. We hypothesize that the G297E substitution may cause premature, prolonged, or constitutive opening of the Eag channels by favoring the “unlocked” state of the channel.

NEURONAL excitability, the propensity for a neuron to fire an action potential in response to synaptic activity, is regulated in large part by K⁺ channels. Loss-of-function mutations in K⁺ channels in vivo confer phenotypes such as prolonged nerve terminal depolarization in response to nerve stimulation and spontaneous generation of action potentials (Ganetzky and Wu 1983; Wu et al. 1983; Warmke et al. 1991; Zhong and Wu 1991). Whereas it has been possible to overexpress and analyze in vivo-engineered versions of channels such as EKO (White et al. 2001), to date chromosomal K⁺ channel gain-of-function mutations have been identified only in organisms in which it is difficult to perform in vivo electrophysiological studies. Consequently, such mutant channels have been best characterized following expression in heterologous systems (Johnstone et al. 1997). As such, the effects of such mutations on behavior and neuronal function are unclear.

The first K⁺ channel gene cloned was the Drosophila Shaker (Sh) voltage-gated K⁺ channel. Sh was first identified on the basis of a behavioral phenotype in which Sh mutant adults shake their legs when under ether-induced anesthesia (Kaplan and Trout 1969). Sh mutants also exhibit increased neurotransmitter release at the Drosophila larval neuromuscular junction (nmj) (Jan et al. 1977) and an elimination of the rapidly inactivating K⁺ current I_A (Wu and Haugland 1985). The Sh gene has also recently been suggested to play a special role in controlling Drosophila sleep (Cirelli et al. 2005).

It was found previously that combining the Sh¹³³ mutation with other excitability mutations can result in either the suppression or the enhancement of the Sh¹³³-induced phenotypes (Ganetzky and Wu 1982, 1983; Wu et al. 1983; Stern et al. 1990). For example, combining Sh¹³³ with a loss-of-function mutation in a second voltage-gated K⁺ channel subunit encoded by ether-a-go-go (eag) enhances the Sh¹³³-induced phenotypes (Wu et al. 1983). The roles played by Eag channel subunits in K⁺ currents are complex: Whereas Eag subunits alone are sufficient to form functional currents, it has also been proposed that Eag can contribute to multiple K⁺ currents in the larval muscle and in heterologous systems by heteromultimerization with other subunits such as Sh and slow-poke (slo) (Zhong and Wu 1991, 1993; Chen et al. 1996, 2000). Hence, we will refer to Eag channels. However, it is not known if Eag affects more than one current at the nmj.

Here we report the identification and characterization of two mutations (Sup³⁹ and Sup¹⁴⁶) that suppress the leg-shaking phenotype of Sh¹³³. From recombination mapping, restriction fragment length polymorphism (RFLP) analysis, and sequencing we show that both mutations result in the same amino acid substitution (G297E) in the eag-encoded channel. Therefore we have renamed these suppressor mutations eag^G297E. We assayed synaptic transmission at the nmj in eag^G297E and found that this mutation suppresses the Sh¹³³-induced increase in excitatory junctional potential (ejp) and that this effect is mediated presynaptically. Furthermore, eag^G297E suppresses the increased neuronal excitability conferred by the K⁺ channel-blocking drugs 4-aminopyridine (4-AP) and tetraethylammonium (TEA). As the phenotypes of the Sh¹³³ mutation are enhanced by eag loss-of-function mutations (Wu et al. 1983), we hypothesize that eag^G297E confers a gain-of-function phenotype. We further hypothesize that the G297E mutation might favor the unlocked state of the channel. The isolation of eag^G297E provides a unique opportunity to test the phenotypes of hyperactive K⁺ channels in vivo.

MATERIALS AND METHODS

Drosophila stocks:

All fly stocks were maintained on standard cornmeal/agar Drosophila media at room temperature. The Sh¹³³ allele is a dominant Sh allele described previously (Kaplan and Trout 1969; Jan et al. 1977) that produces a rapid leg-shaking phenotype when under ether anesthesia. para¹⁴¹ and para⁶³ are alleles of para previously identified and characterized as part of this mutagenesis screen and are otherwise isogenic to Sup¹⁴⁶ and Sup³⁹, respectively (Stern et al. 1990). C(1)DX ywf is a compound (attached) X chromosome that will be represented as ywf.

Behavioral tests:

Ether-induced leg shaking was assayed by exposing young adult flies to ether for ∼10 sec. Under these conditions, wild-type flies are immobilized except for occasional tarsal twitches; Sh¹³³-mutant flies exhibit a rapid shaking of all six legs.

Mutagenesis:

Sh¹³³ males were mutagenized by the addition of ethyl methanesulfonate (EMS) to their food according to Lewis and Bacher (1968), except that 0.25% EMS was used. Following mutagenesis, the males were mated to an equivalent number of ywf females with ∼25 mating pairs per bottle. After 4–5 days of mating, the males were removed and discarded. EMS was inactivated as described previously (Oshima and Takano 1980).

Mapping:

The map position of Sup³⁹ and Sup¹⁴⁶ was determined by recombination mapping, first between y (1A5) and f (15F4-7), then between g (12B4-6) and sd (13F1-4), and finally between two P[ry⁺] inserts located at 12E and 13B/C, respectively. The presence of Sup³⁹ and Sup¹⁴⁶ was scored in males by suppression of Sh¹³³-induced leg shaking.

Sequencing:

All sequence analysis and alignments were produced using the BioEdit sequence alignment editor (Hall 1999). Primer pairs flanking partial, complete, or pairs of eag exons were designed and used for PCR amplification. Sequencing was performed using the same PCR primers, or in the case of longer products, with additional internal sequencing primers. All putative mutations were resequenced at least twice from separate PCR amplifications.

Electrophysiology:

Larval dissections and muscle recordings were performed as described previously (Jan and Jan 1976; Ganetzky and Wu 1982; Stern and Ganetzky 1989; Stern et al. 1995; Huang and Stern 2002). Ventral lateral longitidunal peripheral nerves that innervate the body-wall muscles were cut immediately posterior to the ventral ganglion and were stimulated using a suction electrode. Intracellular muscle recordings were made using a microelectrode pulled on a Flaming/Brown micropipette puller to tip resistances of 30–60 MΩ and filled with 3 m KCl. All dissections and recordings were performed at room temperature in standard saline solution (0.128 m NaCl, 2.0 mm KCl, 4.0 mm MgCl₂, 0.34 m sucrose, 5.0 mm HEPES, pH 7.1, and CaCl₂ as specified in the text). Duration of the ejp was measured from half-maximal response to half-maximal response. Ventral longitudinal muscle cell 6 was used for the recording of all electrophysiological parameters. Quinidine, 4-aminopyridine, and tetraethylammonium were applied following dissection as described previously (Jan et al. 1977; Singh and Wu 1989).

RESULTS AND DISCUSSION

Isolation of mutations that suppress the Sh leg-shaking phenotype:

To identify novel mutations that suppress the leg-shaking phenotype of Sh-mutants, males carrying Sh¹³³ were mutagenized with EMS, as described in materials and methods, and crossed to ywf females, and their sons were scored for an absence of leg shaking. Of 35,000 sons tested, 9 failed to shake their legs when etherized. Five of these suppressor mutations have previously been shown to be new alleles of para, which encodes a sodium channel subunit (Stern et al. 1990). Two additional suppressor mutations, Sup³⁹ and Sup¹⁴⁶, are reported here.

Mapping of Sup³⁹ and Sup¹⁴⁶:

The segregation patterns of these nine suppressors indicated that the mutations were all X-linked. Initial recombination mapping using the leg-shaking phenotype placed both Sup³⁹ and Sup¹⁴⁶ between the garnet (12B4-6) and scalloped (13F) loci and then to a region between two P[ry⁺] elements at positions 12E and 13B/C. Finally, we used RFLP analysis to localize Sup¹⁴⁶ centromere-proximal to a PstI polymorphism in exon 1 of eag. The frequency of crossover between the leg-shaking suppression phenotype and the RFLP suggested that Sup¹⁴⁶ is ∼30 kb from eag exon 1, which would place it within the eag locus (see Figure 1A).

Figure 1.

Map position of the Sup³⁹ and Sup¹⁴⁶ mutations. (A) Mapping of the suppressor mutations between P[ry⁺] elements at positions 12E and 13B. Direct genomic sequencing identified the substitution of a glutamate for a glycine at position 297, indicated by the pair of asterisks. Sup⁺ represents sequence from eag⁺ para⁶³ and eag⁺ para¹⁴¹f. para⁶³ and para¹⁴¹ are siblings of Sup³⁹ and Sup¹⁴⁶, respectively, generated by mutagenesis from the same isogenic wild-type X chromosome. (B) Schematic depicting the topology of the Eag protein. S1–S6 represent the six transmembrane domains, P the pore domain, and cNBD the cyclic nucleotide binding domain. The asterisk indicates the location of the G297E substitution identified in Sup³⁹ and Sup¹⁴⁶. Having shown Sup³⁹ and Sup¹⁴⁶ to be mutations in eag, we have subsequently renamed them eag^G297E.

Sup³⁹ and Sup¹⁴⁶ are new alleles of eag:

To test the possibility that Sup³⁹ and Sup¹⁴⁶ are new alleles of eag, we sequenced genomic DNA from the 15 exons of eag from the suppressor mutations as well as from isogenic Sup⁺ flies. The eag sequence of Sup³⁹ yielded only a single amino acid substitution: a glycine-to-glutamate change at position 297 (G297E; see Figure 1B). Interestingly, the eag sequence from Sup¹⁴⁶ flies showed the identical G297E mutation as well as two additional mutations (A1088T and I1142T). To test the possibility that these two additional mutations were generated after the mutagenesis, while the fly line was being maintained, we sequenced these regions from a second Sup¹⁴⁶ stock, which was split off from the first stock ∼6 months after Sup¹⁴⁶ was obtained. We found that the G297E mutation was retained in this second Sup¹⁴⁶ stock, but that the two additional mutations were absent. We conclude that Sup¹⁴⁶, like Sup³⁹, is a G297E mutation in eag; we will refer to both of them as eag^G297E. We are not aware of any previous study that has assigned a function to G297. Previous studies have shown that eag loss-of-function mutations enhance the phenotypes of the Sh¹³³ mutation (Wu et al. 1983). In contrast, eag^G297E suppresses the phenotypes of the Sh¹³³ mutation. We thus hypothesize that eag^G297E is a gain-of-function allele (see Table 1).

TABLE 1.

Genetic interactions among Na⁺ and K⁺ channel mutations

Genotype	Enhances Sh133	Suppresses Sh133
Para−	X
Dp para+	X
eag−	X
eagG297E		X

The hypoexcitable para mutations suppress the Sh¹³³-induced leg shaking (Stern et al. 1990), whereas hyperexcitable eag mutations and Dp para⁺ enhance the Sh¹³³-induced leg shaking (Stern et al. 1990). eag^G297E suppresses the Sh¹³³-induced leg shaking, leading us to hypothesize that eag^G297E is a gain-of-function mutation. Table was adapted from Huang and Stern (2002).

An alignment of the Eag protein sequence from several species (Figure 2) indicates that position 297 and the surrounding residues are highly conserved in all members of the Eag channel subfamily, but not in the Eag-like (Elk) or Eag-related (Erg) subfamilies. Interestingly, this region of Eag is close to the site at which Mg²⁺ ions interact with the channel to regulate its activity, a phenomenon not seen in either the Elk or the Erg subfamilies (Tang et al. 2000).

Figure 2.

Sequence conservation in the S2–S3 loop of Eag. An alignment of the amino acid sequences corresponding to the S2 domain, the S3 domain, and the S2–S3 loop of Eag from a number of species. The S2 and S3 domains are boxed. Also indicated is the Dm–Eag position 297. Dm–Eag, Drosophila melanogaster (Warmke et al. 1991); Dp–Eag, Drosophila pseudobscura (Richards et al. 2005); M–Eag, mouse (Warmke and Ganetzky 1994); R–Eag1 (Ludwig et al. 1994), rat; B–Eag, Bos Taurus (bovine) (Frings et al. 1998); Ag–Eag, Anopheles gambiae (mosquito) (Mongin et al. 2004), H–Erg, human Eag-related gene (Warmke and Ganetzky 1994); Elk, human Eag-like K⁺ channel (Warmke and Ganetzky 1994). The Eag family has three branches: Eag, Erg, and Elk, all with slightly different properties. At the position indicated within the S2–S3 linker, a residue with a small, nonpolar side chain is conserved with the Eag subfamily.

eag^G297E suppresses the increased amount and duration of neurotransmitter release conferred by Sh¹³³:

In Sh¹³³ mutants, motor axon excitability is increased, which leads to action-potential-triggered Ca²⁺ influx in the nerve terminal that is larger and more prolonged than that in wild type (Jan et al. 1977). This increased Ca²⁺ influx, in turn, leads to increased neurotransmitter release from the motor neuron and subsequent muscle depolarization that is increased in amplitude and duration compared to wild type. We hypothesize that eag^G297E suppresses the Sh¹³³-induced leg-shaking phenotype by reducing neuronal excitability, which would compensate for the increased excitability conferred by Sh¹³³. If so, then eag^G297E might suppress the increased neurotransmitter release conferred by Sh¹³³. To test this possibility, we used the larval nmj preparation (Jan et al. 1977) to compare synaptic transmission at the larval nmj in eag^G297E Sh¹³³ and eag⁺ Sh¹³³. With this preparation an action potential is induced in the motor axon and the consequent neurotransmitter release at the nmj elicits a depolarization in the muscle, the ejp, which is monitored with an intracellular recording electrode.

As neurotransmitter release is dependent upon Ca²⁺ influx into the nerve terminal, at low external [Ca²⁺] (such as 0.1 mm) at most only a single quantum of neurotransmitter is released into a wild-type neuromuscular junction following nerve stimulation. As a consequence, only failures or low amplitude ejps are observed (Jan and Jan 1976). The Sh¹³³ mutant exhibits a prolonged nerve terminal depolarization, leading to prolonged neurotransmitter release and an ejp of increased duration and amplitude even at low [Ca²⁺] (Jan et al. 1977).

To determine if eag^G297E can suppress the defects in synaptic transmission conferred by the Sh¹³³ mutation, we compared ejps from eag⁺ Sh¹³³ with ejps from eag^G297E Sh¹³³ (Figure 3). We found that the eag^G297E allele from either the Sup³⁹ or the Sup¹⁴⁶ line partially suppresses the Sh¹³³ phenotype in a dosage-dependent manner. Significant decreases in both ejp amplitude (P < 0.001) and duration (P < 0.001 for eag^G297E from Sup³⁹ and P < 0.05 for Sup¹⁴⁶) were observed when the suppressor mutations were homozygous. When heterozygous, a moderate decrease in the Sh¹³³-induced ejp amplitude, but not ejp duration, was observed (P < 0.02 for Sup¹⁴⁶; Figure 3). The smaller suppressive effect in the heterozygotes indicates that dosage of eag^G297E controls the degree of suppression observed.

Figure 3.

Dosage-dependent suppression of Sh¹³³ by eag^G297E. (A) Averaged intracellular muscle recordings from larvae of the indicated genotypes in response to nerve stimulation. (B) Average amplitudes of ejps evoked by nerve stimulation in the presence of low bath [Ca²⁺] (0.1 mm) in larvae of the genotypes indicated. Under these low [Ca²⁺] conditions most successful wild-type ejps result from the release of at most a single vesicle of neurotransmitter. (C) Average duration of evoked ejps. Values are presented as the mean ± SEM; data were collected from at least five larvae for each genotype. 1, eag⁺ Sh⁺; 2, eag⁺ Sh¹³³; 3, eag^G297E (from Sup³⁹) Sh¹³³/eag⁺ Sh¹³³; 4, eag^G297E (from Sup³⁹) Sh¹³³; 5, eag^G297E (from Sup¹⁴⁶) Sh¹³³/eag⁺ Sh¹³³; 6, eag^G297E (from Sup¹⁴⁶) Sh¹³³. *P < 0.05 by Student's t-test, **P < 0.02 by Student's t-test, ***P < 0.01 by Student's t-test vs. eag⁺ Sh¹³³.

Similar suppression of Sh¹³³-induced phenotypes has been observed previously by mutations in para and mle^nap, which decrease the number of Na⁺ channels (Ganetzky and Wu 1985; Stern et al. 1990) and thus create hypoexcitable neurons. These observations support our hypothesis that eag^G297E reduces neuronal excitability, a phenomenon that is presumably the result of conferring a gain-of-function phenotype on the Eag channels. A suppression of Sh¹³³ could possibly be produced by increasing the number of Eag-containing channels in the neuronal membrane, similar to the enhancement of Sh¹³³ seen with Dp para⁺ (Table 1). An alternative possibility is that this gain-of-function phenoptype is the result of premature, prolonged, or constitutive opening of the Eag channels.

When performed at higher [CaCl₂] (0.4 mm), eag^G297E does not decrease the amplitude or duration of Sh¹³³ mutant ejps (data not shown). This result is not unexpected as the evoked ejp at this [CaCl₂] is not significantly different in Sh mutants as compared to wild type (Jan et al. 1977). The absence of an effect by eag^G297E in the presence of high [CaCl₂] does suggest that muscle is not impaired in its response to neurotransmitters at the nmj.

eag^G297E suppresses the enhancement of Sh-mutant ejp by quinidine:

The addition to the extracellular bath of 0.1 mm quinidine, a drug that selectively blocks I_K in Drosophila larval muscles (Singh and Wu 1989), enhances the amplitude and duration of the action-potential-evoked ejp in Sh¹³³ mutant larvae (Wu et al. 1989). To determine if eag^G297E is able to suppress this enhancement of the Sh¹³³ phenotype, ejp recordings were performed at 0.1 mm [Ca²⁺] and in the presence of 0.1 mm [quinidine]. As shown in Table 2, when homozygous, eag^G297E is able to partially suppress the quinidine-enhanced Sh¹³³ ejp phenotype.

TABLE 2.

Suppression of the Sh¹³³-induced increase in neurotransmitter release by eag^G297E: effects of quinidine

	Excitatory junctional potential with quinidine
Genotype	Amplitude (mV)	Duration (msec)
eag+Sh133	23.4 ± 4.6	73.9 ± 10.8
eagG297E Sh133/eag+Sh133 (from Sup39)	13.9 ± 2.8	61.6 ± 15.9
eagG297E Sh133 (from Sup39)	4.7 ± 0.7***	41.5 ± 8.2*
eagG297E Sh133/eag+Sh133 (from Sup146)	11.0 ± 2.0*	52.6 ± 12.5
eagG297E Sh133 (from Sup146)	4.4 ± 1.5***	43.2 ± 10.8

Mean amplitudes and durations of ejps evoked by nerve stimulation in the presence of low bath [Ca²⁺] (0.1 mm) and 0.1 mm [quinidine] in larvae of the genotypes indicated. Values are presented as the mean ± SEM; data were collected from at least four larvae for each genotype. *P < 0.05, ***P < 0.01 vs. eag⁺ Sh¹³³.

If 0.1 mm [quinidine] blocks I_K completely, how can eag^G297E reduce the sensitivity of the motor neuron to quinidine? There are several possible explanations. First, while it is known that 0.1 mm [quinidine] nearly eliminates I_K in Drosophila larval body-wall muscles (Singh and Wu 1989), it is not clear if it blocks I_K specifically and completely in the motor neuron. Second, the G297E mutation might render Eag less sensitive to quinidine than eag⁺. This possibility is supported by the observation that mutations in human Eag can alter the binding characteristics of quinidine and other antiarrhythmic agents (Gessner et al. 2004). Third, the K⁺ channels distinct from I_K in which Eag participates (I_A, I_CS, and I_CF; Ganetzky and Wu 1983; Wu et al. 1983; Warmke et al. 1991; Zhong and Wu 1991) are not sensitive to quinidine, and it is possible that Eag^G297E exerts its effects through one of these channels.

eag^G297E suppresses the neuronal excitability conferred by K⁺ channel-blocking drugs:

The eag^G297E mutation does not appear to confer any obvious behavioral abnormality in an otherwise wild-type background. No temperature-sensitive paralysis is observed, the flies appear well coordinated, and flight appears normal. To determine if eag^G297E confers a defect in synaptic transmission in an otherwise wild-type background, we made measurements of ejp amplitude and duration at 0.15 mm [CaCl₂]. No significant decrease in either ejp amplitude or duration was observed (data not shown). Similarly, the onset rate of long-term facilitation (Jan and Jan 1978) was also unaffected by eag^G297E (data not shown). The observation that eag^G297E conferred no detectable excitability phenotype in a Sh⁺ background raised the possibility that the suppression of Sh¹³³ by eag^G297E might be an allele-specific restoration of Sh function. This possibility is supported by the observation that Eag and Sh subunits can co-assemble in a channel complex (Chen et al. 1996, 2000). To test this possibility, we examined the effects of eag^G297E on the hyperexcitability conferred by two K⁺ channel-blocking drugs: 4-AP, which is a specific Sh channel blocker, and TEA, which is a more general voltage-gated K⁺ channel blocker (Ganetzky and Wu 1983; Yamamoto and Suzuki 1989).

We found that the amplitude of ejps evoked at 0.1 mm [CaCl₂] and 2 mm [4-AP] was significantly suppressed by eag^G297E (Table 3). Similarly, we found that large amplitude and prolonged ejps evoked at 0.1 mm [CaCl₂] by addition of 10 mm [TEA] were easily induced in all wild-type larvae; in contrast, in eag^G297E larvae it was possible to evoke such an ejp in only about half of the larvae tested (data not shown). Furthermore, as shown in Table 3, even when an ejp was successfully elicited, eag^G297E reduced the amplitude and duration of the ejp in comparison with eag⁺. This suppression of the increased ejp conferred by 4-AP and TEA suggests that the G297E mutation increases the activity of the voltage-gated K⁺ channels that remain active following the addition of 4-AP and TEA and thus that the suppression of the Sh-induced increased ejp is not allele specific. Interestingly, it appears that eag^G297E manifests its most obvious phenotypes under conditions that prolong the action potential.

TABLE 3.

Decreased neuronal excitability by eag^G297E: effects of 4-AP and TEA

	Excitatory junctional potential
4-AP		TEA
Genotype	Amplitude (mV)	Duration (msec)	Amplitude (mV)	Duration (msec)
eag+	19.8 ± 3.2	29.9 ± 5.8	14.6 ± 3.0	490.5 ± 37.7
eagG297E (from Sup146)	10.3 ± 2.3*	39.8 ± 12.6	6.2 ± 1.4**	181.8 ± 64.7***

Mean amplitudes and durations of ejps evoked by nerve stimulation in the presence of low bath [Ca²⁺] (0.1 mm) and either 2 mm [4-AP] or 10 mm [TEA] in larvae of the genotypes indicated. Values are presented as the mean ± SEM; data were collected from at least five larvae for each genotype. *P < 0.05, **P < 0.002, ***P < 0.01 vs. eag⁺.

eag^G297E acts presynaptically to reduce the ejp amplitude and duration:

The suppression of the Sh¹³³-induced increased amplitude ejp conferred by eag^G297E could be a consequence of a suppression of the increased neurotransmitter release of Sh¹³³ mutants, of reduced sensitivity of the muscle membrane glutamate receptors to the neurotransmitter l-glutamate, or of increased voltage-dependent or voltage-independent K⁺ currents in the muscle membrane. To distinguish among these possibilities, we monitored synaptic transmission while holding the muscle membrane potential to −60 mV with a voltage clamp, which prevents the opening of voltage-gated ion channels such as Eag and Sh in the muscle membrane. If the eag^G297E-dependent suppression of the Sh¹³³-induced increase in ejp size is due to suppression of the increased neurotransmitter release, then we would expect that excitatory junctional currents (ejcs) of reduced amplitude would be observed in eag^G297E Sh¹³³ compared with eag⁺ Sh¹³³. In contrast, if the suppression of the ejp amplitude phenotype is a result of altered muscle voltage-dependent K⁺ currents, then ejc amplitude in eag^G297E Sh¹³³ is expected to be the same as that in eag⁺ Sh¹³³. Figure 4 shows that the ejc in eag^G297E Sh¹³³ is significantly smaller than that in eag⁺ Sh¹³³, which is consistent with the notion that eag^G297E acts presynaptically to suppress Sh¹³³ phenotypes by decreasing the amount of neurotransmitter released. The proportional reduction in ejc amplitude (69%) is similar to the proportional reduction in ejp amplitude (73%).

Figure 4.

The eag^G297E mutation acts presynaptically to suppress the phenotypes of Sh¹³³. (A) Typical ejcs in response to nerve stimulation in the presence of 0.1 mm [Ca²⁺]. Holding potential was −60 mV. Simultaneous intracellular recording of voltage (V) and current (I) were performed. (B) Average amplitude of evoked ejcs. eag^G297E from Sup³⁹. Values presented as mean ± SEM, n = 4. *P < 0.05 by Student's t-test.

To exclude the possibility that an altered voltage-independent current or altered sensitivity to l-glutamate is responsible for the suppression of the ejp phenotypes, we compared the amplitude of spontaneous miniature ejps (mejp) in eag⁺ Sh¹³³ and eag^G297E Sh¹³³. If the eag^G297E-dependent suppression of the Sh¹³³-induced increased ejp amplitude is due to decreased muscle response to the neurotransmitter l-glutamate or to increased voltage-independent K⁺ currents, then mejps of reduced amplitude will be observed in eag^G297E Sh¹³³ compared to eag⁺ Sh¹³³. We found that the mejp amplitudes recorded in eag^G297E Sh¹³³ and eag⁺ Sh¹³³ are almost identical (0.79 mV ± 0.11 mV and 0.81 mV ± 0.14 mV, respectively), suggesting that eag^G297E does not affect voltage-independent currents or the muscle response to l-glutamate. We conclude that eag³⁹ acts presynaptically to suppress the effects of Sh¹³³.

Suppression by eag^G297E of the Sh¹³³-induced increase in neuronal excitability requires extracellular Mg²⁺:

The S2 and S3 transmembrane domains of Eag contain three aspartic acid residues that coordinate the binding of a Mg²⁺ ion when the channel is in its closed state (Silverman et al. 2000; Tang et al. 2000; Schönherr et al. 2002). Mg²⁺ ions slow channel activation in a concentration- and voltage-dependent manner in bovine, mouse, and Drosophila Eag (Terlau et al. 1996; Schönherr et al. 1999; Silverman et al. 2000; Tang et al. 2000). The observation that the G297E substitution falls within the highly conserved S2–S3 linker raised the possibility that G297E acts by eliminating or reducing the effect of Mg²⁺. If so, then in the absence of extracellular Mg²⁺, neuronal excitability is predicted to be the same in eag^G297E and eag⁺. To test this possibility, ejp recordings were performed in eag^G297E Sh¹³³ and eag⁺ Sh¹³³ at low [Ca²⁺] in the absence of extracellular Mg²⁺ (Table 4). Table 4 shows that the ejp amplitude and duration are not as significantly different in eag⁺ Sh¹³³ as in eag^G297E Sh¹³³. This observation supports the hypothesis that the regulatory effects of Mg²⁺ upon Eag have been lost in eag^G297E and is consistent with the possibility that eag^G297E increases Eag activity by reducing the affinity of Eag for Mg²⁺.

TABLE 4.

Suppression of the Sh¹³³-induced increase in neurotransmitter release by eag³⁹ and eag¹⁴⁶ requires extracellular Mg²⁺

	ejp
Genotype	Amplitude (mV)	Duration (msec)
eag+Sh133	9.4 ± 1.7	66.6 ± 21.3
eagG297E Sh133 (from Sup39)	7.6 ± 2.4	87.1 ± 10.5
eagG297E Sh133 (from Sup39)	6.7 ± 1.4	107.0 ± 29.2

Mean amplitudes and durations of ejps evoked by nerve stimulation in the absence of extracellular MgCl₂ and in the presence of low bath [Ca²⁺] (0.1 mm) in larvae of the genotypes indicated. Values are presented as the mean ± SEM; data were collected from at least four larvae for each genotype.

The model of Mg²⁺ action upon Eag proposed by Schönherr et al. (2002) suggests that Mg²⁺ is involved in the switch between the resting (“locked”) and activated (“unlocked”) states. We suggest that the G297E mutation alters the configuration of the channel such that it favors the “unlocked” over the “locked” state. In vitro studies of this mutant channel would be required to test this possibility.

Our results suggest that the leg-shaking and electrophysiological phenotypes caused by the Sh¹³³ mutation are suppressed in a dosage-dependent manner by eag^G297E. Because phenotypes of eag^G297E are opposite to those of eag loss-of-function mutants (Wu et al. 1983; Warmke et al. 1991; Bruggemann et al. 1993), we propose that these new mutations confer a gain-of-function phenotype to the Eag channels. Thus, eag activity regulates neuronal excitability: reduction in eag activity confers a hyperexcitable neuron, whereas increases in eag activity confer a hypoexcitabile neuron. A similar phenomenon occurs with the para-encoded Na⁺ channel: decreasing or increasing channel number produces hypoexcitable or hyperexcitable neurons, respectively (Stern et al. 1990). Given that eag^G297E acts presynaptically, we hypothesize that the neuronal hypoexcitability effect is mediated by the premature, prolonged, or constitutive opening of the Eag channels in the motor neuron membrane, resulting in an attenuation of the action potential. Furthermore, our experiments raise the possibility that the reduction or elimination of the response to extracellular Mg²⁺ may be responsible for this gain-of-function phenotype of eag^G297E.

Defects in K⁺ channel function cause a wide range of diseases termed “potassium channelopathies.” These channelopathies encompass diseases such as episodic ataxia (D'Adamo et al. 1999), myokymia (Dedek et al. 2001), epilepsy (Schroeder et al. 1998), arrhythmia (Sanguinetti 1999), and deafness (Van Hauwe et al. 2000). Furthermore, loss of normal K⁺ channel function can enhance tumor growth (Meyer and Heinemann 1998; Meyer et al. 1999; Pardo et al. 1999; Smith et al. 2002; Suzuki and Takimoto 2004). K⁺ channel gain-of-function mutations have been identified in patients with diseases such as familial atrial fibrillation (Chen et al. 2003) and have also been associated with the low prevalence of diseases such as diastolic hypertension in some populations (Fernandez-Fernandez et al. 2004). Polymorphisms such as eag^G297E, if they exist in humans, might be of particular therapeutic importance because an individual carrying such a polymorphism might exhibit no overt abnormalities and yet show an aberrant sensitivity to particular therapeutic drugs. Consequently, understanding how K⁺ channels function, how they are regulated, and how they interact with other membrane components is of substantial importance. The identification of a gain-of-function mutation in a K⁺ channel gene that can be studied further both in vivo and in vitro provides a unique opportunity to obtain new knowledge of K⁺ channel regulation.