Neuropathology in Drosophila Membrane Excitability Mutants

Abstract

Mutations affecting ion channels and neuronal membrane excitability have been identified in Drosophila as well as in other organisms and characterized for their acute effects on behavior and neuronal function. However, the long-term effect of these perturbations on the maintenance of neuronal viability has not been studied in detail. Here we perform an initial survey of mutations affecting Na⁺ channels and K⁺ channels in Drosophila to investigate their effects on life span and neuronal viability as a function of age. We find that mutations that decrease membrane excitability as well as those that increase excitability can trigger neurodegeneration to varying degrees. Results of double-mutant interactions with dominant Na⁺/K⁺ ATPase mutations, which themselves cause severe neurodegeneration, suggest that excitotoxicity owing to hyperexcitability is insufficient to explain the resultant phenotype. Although the exact mechanisms remain unclear, our results suggest that there is an important link between maintenance of proper neuronal signaling and maintenance of long-term neuronal viability. Disruption of these signaling mechanisms in any of a variety of ways increases the incidence of neurodegeneration.

GENETIC factors have been implicated in virtually every neurological disease, including degenerative disorders such as Alzheimer's, Huntington's, and Parkinson's diseases. Although several genes have been linked to neurodegenerative diseases, including those encoding α-synuclein, huntingtin, tau, APP, and parkin, the molecular mechanisms mediating neuronal loss remain largely unknown (Bonini and Fortini 2003; Driscoll and Gerstbrein 2003; Shulman et al. 2003). To elucidate the underlying molecular pathways, Drosophila has been used successfully to model various neurodegenerative conditions, including tauopathy, polyglutamine repeat diseases, Parkinson's disease, and Alzheimer's disease (reviewed by Bonini and Fortini 2003). In complementary studies, forward genetic approaches have been used to identify genes required for age-dependent maintenance of neuronal viability (Buchanan and Benzer 1993; Kretzschmar et al. 1997; Min and Benzer 1997, 1999; Palladino et al. 2002; Troulinaki and Tavernarakis 2005).

In similar screens, we found that mutations in ATPα, the gene encoding the α-subunit of the Na⁺/K⁺ ATPase, which cause stress-sensitive seizures, temperature-sensitive paralysis, and reduced life span, also cause pronounced neurodegeneration in the central nervous system (Schubiger et al. 1994; Palladino et al. 2002, 2003). Mutations of the ATPase β2 subunit in mice also cause neurodegeneration (Magyar et al. 1994; Molthagen et al. 1996). Pharmacological disruption of the Na⁺/K⁺ ATPase has been shown to cause membrane depolarization resulting in increased intracellular calcium, ultimately leading to necrosis (Chatterjee and Roy 1965; Lees et al. 1990). In addition, depletion of intracellular K⁺ levels, due to pharmacological impairment of pump activity, has been linked to apoptosis (Beauvais et al. 1995; Bortner et al. 1997; Yu et al. 1997). These results suggest that perturbations in membrane excitability and ion homeostasis may be important triggers for neuronal death. To investigate the significance of altered membrane excitability and ion homeostasis on neuronal viability, we utilized the existing set of mutations in Drosophila affecting voltage-gated Na⁺ and K⁺ channels alone and in combination with Na⁺/K⁺ ATPase mutations to examine their effect on life span and neuronal maintenance.

We found that many ion channel mutations are associated with an elevated occurrence of neurodegeneration. Mutations affecting eag or Sh K⁺ channels alone or in double-mutant combinations result in neuronal hyperexcitability but do not trigger striking neurodegeneration even though the adult life span of the double mutant is much reduced. In contrast, mutations in the sei K⁺ channel gene exhibit normal life spans but manifest significant neurodegeneration in the central nervous system (CNS). Mutations in para, which encodes a voltage-activated Na⁺ channel, decrease excitability but confer neurodegeneration, indicating that decreased neuronal activity can also impair neuronal viability. We examined double mutants of ATPα in combination with various ion channel mutations to test whether the severe neurodegeneration seen in ATPα mutants resulted from hyperexcitability-induced excitoxicity. The results of the genetic interactions that we observed suggest that mechanisms other than or in addition to excitotoxicity underlie neurodegeneration in ATPα mutants. Our results focus attention on the existence of an important link between maintenance of proper neuronal signaling and maintenance of neural viability. Molecular and physiological characterization of the relevant mechanisms should have important implications for understanding a broad array of disorders.

MATERIALS AND METHODS

Fly strains:

Fly stocks were cultured on cornmeal–molasses agar media at 22°–29°. ATPα alleles DTS1, DTS2, 2206, and DTS1R1 were used and are described elsewhere (Schubiger et al. 1994; Palladino et al. 2003). Other strains used in this study include para^ts1, para^ts115, para^st109, para^lk5, para^ld34, mle^napts (nap^ts1), Sh¹³³, eag¹, sei^ts1, sei^ts2, elav3A, EKO E323, and HS-GAL4. Controls for genetic modifiers of excitability were performed on Sh¹³³ in several genetic backgrounds, including w¹¹¹⁸ and CS, with all lines exhibiting similar and strong leg shaking upon exposure to ether. Unless otherwise stated, wild type and control refer to Canton-S, which is the original background for many of these mutants.

ATPα immunohistochemistry:

Goat monoclonal antibody (mAB) against the chicken Na⁺/K⁺ ATPase α (a5) was obtained from the Iowa hybridoma bank. Immunohistochemistry was performed on dissected third instar larvae to visualize Na⁺/K⁺ ATPase α localization in the motor axon bundles using standard procedures (Bellen and Budnik 2000). The a5 antibody was used (1:100) in PBTX on 4% paraformaldehyde fixed specimens. Dissected animals were uniquely tagged and processed together in the same microcentrifuge tube. Anti-goat fluorescein conjugate was used 1:500 as a secondary antibody. Identical imaging parameters were used to visualize fluorescein localization on a Zeiss confocal microscope. ImageJ was used to compare the fluorescence intensity in motor axon bundles.

Life-span analysis:

Life spans were measured at 29° according to standard protocols (Kretzschmar et al. 1997; Lin et al. 1998; Min and Benzer 1999; Kapahi et al. 2004) as previously described (Palladino et al. 2002, 2003). In brief, newly eclosed animals were separated by sex, placed in vials (<21/vial), and transferred daily to fresh vials, and the number of surviving flies in each vial was recorded. Animals that were lost or removed for analysis were subtracted from the total population in calculations. The age in days at which only 50% of the original population still survived (50% survivorship) was compared among different genotypes using the Mann-Whitney test.

Histology:

Histological analyses were performed as previously described (Palladino et al. 2002, 2003). Briefly, heads and bodies from adult flies were dissected and placed in freshly prepared Carnoy's fixative at room temperature for 1–2 days and then washed with 70% ethanol and processed into paraffin. Heads were embedded to obtain frontal sections. Serial 4-μm sections were stained with hematoxylin and eosin and examined under a light microscope (n > 10 for each genotype). The degree of neuropathology present for each genotype was determined by the examination of the frequency of vacuolar pathology in serial sections, according to the following rating scale. A score of 0 was assigned to brains exhibiting no gross neuropathology or a single small vacuolar structure (<12 μm in diameter), similar to aged wild type. Brain tissue presenting with small sporadic individual vacuolar structures (<12 μm in diameter) in multiple sections of a brain was given a score of 1. More frequent small individual vacuolar structures (<12 μm in diameter) appearing in most sections of each brain was scored as a 2. Heads exhibiting widespread small vacuolar pathology (<15 μm in diameter) affecting the majority of sections or large vacuolar or clustering vacuolar structures (15–20 μm in diameter) were assigned a value of 3. Brain tissue with numerous (>100) vacuolar structures 10–20 μm in diameter in individual sections and/or clustering vacuolar structures affecting an area >500 μm² were given a score of 4. A score of 5 indicates that a large proportion of brain tissue is missing (>40%).

RESULTS

Neuropathology in K⁺ channel mutants:

Mutations in a number of genes encoding various K⁺ channel subunits have been isolated in Drosophila. In general, loss of K⁺ channel function leads to neuronal hyperexcitability (Ganetzky and Wu 1986; Noebels 2003). Repeated or even single seizures have been reported to produce neuronal damage in mammals (Sutula et al. 2003). Thus, it was of interest to determine whether K⁺ channel mutants exhibit enhanced susceptibility to neurodegeneration. For these studies we examined mutant alleles of eag, Sh, and sei, which have been well characterized behaviorally and electrophysiologically. eag and Sh cause leg shaking under ether anesthesia (Kaplan and Trout 1969; Jan et al. 1977; Wu et al. 1983), whereas sei causes convulsive seizures when the mutants are placed at 37°–40° (Jackson et al. 1984). eag and Sh interact genetically in double mutants to produce flies that shake and have severe motor defects under all conditions owing to extreme hyperexcitability (Ganetzky and Wu 1983). As severe neurodegeneration is often associated with a strongly reduced life span (Buchanan and Benzer 1993; Kretzschmar et al. 1997; Palladino et al. 2002), we first determined the life span of eag, Sh, and sei mutants as well as eag Sh double mutants (Figure 1, Table 1). Only eag Sh double mutants, which become severely uncoordinated with age, consistent with previous observations (Ganetzky and Wu 1983), showed significantly shortened life span (7 days ± 2.1, P < 0.001).

Figure 1.

Life spans of individual mutant K⁺ channel strains are normal. Survival was determined daily for populations of various genotypes (5–10 vials/genotype). (A) Life spans for sei (diamonds), Sh (squares), and eag (triangles) K⁺ channel mutant strains. (B) No significant alteration in median age (T_50% survivorship) was observed for K⁺ channel mutants individually. eag Sh double mutants are severely uncoordinated and have significantly reduced life spans (open circles; *P < 0.001). Wild-type control strain is Canton-S (CS; solid circles). Each point represents mean survival ±SEM.

TABLE 1.

Viability and neuropathology of various ion channel mutants

	50% ± SD	Pathology	50% ± SD	Pathology
Genotype	+/+	+/+	DTS1/+	DTS1/+
+/Y	38 ± 4.4	0/1	14 ± 0.8	4
eag/Y	33 ± 1.7	1	14 ± 2.1	3/4
Sh/Y	35 ± 2.6	1	16 ± 3.4	3
eag Sh/Y	7 ± 2.1	0	7 ± 0.7	NA
seits1/seits1	37 ± 0.5	2	NA	NA
napts1/napts1	27 ± 1.9	1	NA	NA
parats1/Y	37 ± 1.5	1/2	6 ± 1.1	0/1
+/+	38 ± 4.4	0/1	23 ± 3.4	4
eag/+	47 ± 0.7	NA	21 ± 1.4	NA
Sh/+	39 ± 16.2	NA	31 ± 6.1	NA
eag Sh/+	36 ± 4.2	NA	24 ± 3.5	3/4
seits1/+	39 ± 2.4	NA	23 ± 2.1	4
parats1/+	48 ± 3.3	2	16 ± 5.3	4

Viability is assessed as median age (50%). Error is standard deviation (SD). Genotype in column heading represents the third chromosome. DTS1 indicates ATPα^DTS1. Neuropathology was scored 0 (none) to 5 (marked).

Histological analysis of the CNS at 50% survivorship for each mutant strain revealed mild neurodegeneration in these mutants (Table 1). Vacuolar pathology in the central brain as well as in the optic lobes was detected to varying degrees in each of these mutants but was rarely seen in age-matched controls (Figure 2). Both sei alleles examined, sei^ts1 (Figure 2A) and sei^ts2 (data not shown), display substantial pathology, whereas eag Sh double mutants exhibited no significant neurodegeneration at 50% survivorship (7 days; Figure 2D).

Figure 2.

Neurodegeneration in K⁺ channel mutant animals. Frontal sections of mutants aged to median life span are shown. (A) At day 37 sei^ts1 mutants exhibit significant pathology presented as sporadic vacuoles throughout the central nervous system. The region designated by the square is enlarged at the right as A′–E′. Micrographs showing intermittent brain pathology in eag (B) and Sh (C) mutants at 33 and 35 days, respectively. (D) No gross pathology was ever observed in eag Sh double mutants, which die within 1 week. (E) Aged CS animals display little or no pathology. Bar, 50 μm.

To determine if repeated seizures exacerbate the pathology observed in sei mutants, seizures were evoked daily by exposure to 40° for 3 min. For all sei alleles tested, this treatment results in seizures that persist for up to a minute before the mutants become completely paralyzed. Wild-type flies are unaffected by this treatment. The seizure phenotype of both sei^ts1 and sei^ts2 progresses with age and becomes more severe with daily exposure to the restrictive temperature. However, wild-type control animals began to exhibit some locomotor impairment after ∼25 days of daily exposure to 40° that included temperature-dependent incoordination and inactivity with occasional partial paralysis. Daily induction of seizures did not result in significant changes in recovery time, seizure behavior, or life span when compared with controls. Although we observed no simple correlation with the frequency of seizure activity, our results show that loss or impairment of K⁺ channel function can result in elevated levels of neurodegeneration.

Shortened life span and neurodegeneration in Na⁺ channel mutants:

To examine the effects of decreased electrical activity on neuronal viability, we asked whether mutations that reduce Na⁺ channel activity or expression decrease life span and cause increased neurodegeneration. para and mle^napts mutations disrupt voltage-dependent Na⁺ channel function and result in reduced neuronal activity (Suzuki et al. 1971; Siddiqi and Benzer 1976; Wu et al. 1978; Wu and Ganetzky 1980; Loughney et al. 1989; Budnik et al. 1990; Kernan et al. 1991; Reenan et al. 2000). No significant changes in median life span were observed in the para mutants examined (Figure 3, Table 1; para^ts1 T_50% = 37 days, para^ts1/ts115 T_50% = 46 days, para^st109/ld34 T_50% = 39 days) (see also Palladino et al. 2002). However, the mle^napts strain exhibits significantly reduced life span (Figure 3; T_50% = 27 days, P < 0.001) as well as mild neuropathology with multiple regions of vacuolization present in most brain sections (Figure 4A). Varying degrees of neurodegeneration were observed in the allelic series of para mutants (Figure 4). The pathology was similar to that observed in K⁺ channel mutants with the appearance of sporadic vacuolization in the neuropil. However, in comparison with K⁺ channel mutants, neurodegeneration in para mutants exhibited greater penetrance and a more severe manifestation with a higher density of vacuolization in each brain section (Figure 4, B–D). Neurodegeneration in para mutants was scored as in Table 1 (histo-pathology score: para^ts1 = 1/2, para^st109/ld34 = 2, and para^ts1/ts115 = 1/3). The degree of neurodegeneration, measured at the midpoint of their respective survival curves, appears to increase with the life span of each para allele, suggesting that the mechanism underlying this pathology has a critical component that depends on absolute time. These results show that Na⁺ channel mutations known to reduce neuronal excitability are associated with age-dependent neurodegeneration, which is even more severe than that associated with the K⁺ channel mutants examined. Thus, impaired neuronal excitability as well as increased neuronal excitability can trigger neurodegeneration.

Figure 3.

mle^napts mutant animals exhibit shortened life spans. (A) Survivorship values for mle^napts (open diamonds) and para^ts1 (solid squares) are plotted with Canton-S as wild-type control (solid circles). (B) No significant change in life span was observed in any para mutant strain examined, including para^ts1 shown here (Palladino et al. 2002). A significant reduction in life span was found in the mle^napts strain (T_50% = 27 days, P < 0.001). Each point represents mean survival ±SEM.

Figure 4.

Na⁺ channel mutants exhibit age-dependent neuropathology. (A) mle^napts animals at 50% survivorship display mild neurodegeneration. Varying degrees of pathology are observed in para alleles, ranging from single sporadic vacuoles in neighboring brain sections to numerous holes per section. The region designated by the square is enlarged at the right as A′–E′. Sections from para^ts1 (B), para^ts1/para^ts115 (C), and para^st109/Df(1)LD34 (D) heads are shown, which were collected at the median age for each genotype (∼38 days). (E) An uncommon brain from a para^ts1 animal with sporadic massive neurodegeneration. Bar, 50 μm.

Ion channel mutations can exacerbate early adult lethality in ATPα mutants:

We previously found that mutations in the Na⁺/K⁺ ATPase result in severe neurodegeneration and early adult lethality (Palladino et al. 2003). Electrophysiological studies showed that these mutations also cause neuronal hyperexcitability perhaps by destabilizing the membrane resting potential. These results raise the possibility that an excitotoxic mechanism is responsible for neurodegeneration in these mutants. However, the precise mechanism leading to loss of neuronal tissue in these mutants remains unclear because integrity of septate junctions might also be disrupted (Genova and Fehon 2003; Paul et al. 2003). Metabolic processes could also be impaired in ATPα mutants through increased ATPase activity (Palladino et al. 2003). Therefore, we tested whether altered neuronal excitability was the primary cause of neurodegeneration in ATPα mutants by examining their phenotypic interactions in combination with ion channel mutations. Thus, if hyperexcitability-associated excitoxicity were a key factor in the mechanism of neurodegeneration and life-span reduction in ATPα mutants, we would expect that K⁺ channel mutations would further enhance these phenotypes. However, as measured by life-span duration, none of the K⁺ channel mutants examined increased phenotypic severity in double-mutant combinations with ATPα compared with the single mutants (Figure 5A, Table 1). Furthermore, even in eag Sh; ATPα^DTS1 triple mutants there was no significant reduction in life span compared with eag Sh controls.

Figure 5.

para mutants exhibit shortened life spans in the presence of dominant ATPα mutations. Analysis of the presence of ion channel mutations on ATPα mutant life spans was performed. Introduction of ion channel mutations resulted in significant interactions only with ATPα dominant mutations. (A) No genetic interactions were observed with K⁺ channels based on life-span analysis. The presence of eag and/or Sh did not significantly alter any ATPα mutant life spans. (B) Combination of para mutations with dominant ATPα alleles resulted in a striking reduction of survivorship. para^ts1/Y;;ATPα^DTS1/+ (diamonds), para^st109/Y;;ATPα^DTS1/+ (triangles), and para^ts115/Y;;ATPα^DTS1/+ (circles) animals die in ∼1 week, while +/+;;ATPα^DTS1/+ (solid squares) and para^ts1/Y;CyO/+ (solid diamonds) controls have a T_50% of 14 and 38 days, respectively. Each point represents mean survival ±SEM. (C) Median age of survivorship for para^ts1 and ATPα^DTS1 mutant animals. (D) Mean time to recovery after 3 min at 38°. para^ts mutants display immediate recovery from paralysis upon return to permissive temperature while both ATPα^DTS1 and para^ts1/Y;;ATPα^DTS1/+ mutant animals require at least several minutes (P < 0.001). (E) A representative section from a para^ts1/Y;;ATPα^DTS1/+ mutant animal, which exhibits mild to no neurodegeneration. Bar, 50 μm.

Conversely, we generated double-mutant combinations between para and ATPα alleles to test if a reduction in excitatory channels could suppress the neuropathological phenotypes of ATPα mutants. Surprisingly, ATPα^DTS1 and ATPα^DTS2 exhibited a striking reduction in life span in double-mutant combinations with various para alleles (Figure 5B, Table 1). We did not observe the same effect with the recessive ATPα²²⁰⁶ and ATPα^DTS1R1 alleles in combination with para mutations (data not shown). Thus, the dominant ATPα alleles appear to be more severe than ATPα null alleles when heterozygous, suggesting a dominant negative mode of action. Although para^ts1, para^ts115, and para^st109 all display essentially normal life spans individually, the median life span drops to <1 week in combination with ATPα^DTS1 (Figure 5B; P < 0.001 for each) or ATPα^DTS2 (data not shown). Despite the greatly reduced life span of para^ts/Y; ATPα^DTS1/+ mutants, we observed only minimal neuropathology at 50% survivorship (Figure 5C; Table 1), similar to age-matched ATPα^DTS1/+ controls. This finding suggests that overt neurodegeneration is not the cause of this unexpectedly early death, which may instead reflect some other type of severe neural impairment in the double mutants.

Behavioral phenotypes of the double mutants further indicated that the mechanism of paralysis in ATPα mutants also involves a component that does not appear to depend on neuronal activity. para^ts1; ATPα^DTS adults become instantly paralyzed upon exposure to 38°, as do para^ts1 mutants alone, consistent with the block of Na⁺-mediated action potentials (Suzuki et al. 1971; Wu and Ganetzky 1980; Loughney et al. 1989; O'Dowd et al. 1989). Upon being returned to 22° following a 3-min exposure to 38°, para^ts1 mutants recovery instantly. In contrast, para^ts1; ATPα^DTS adults do not recover instantly but require several minutes to regain normal activity, similar to ATPα^DTS mutants alone (Figure 5D). Thus, the mechanisms mediating temperature-sensitive seizures and paralysis in ATPα^DTS mutants appear to be epistatic to the rapid paralysis of para^ts mutants. Because neuronal activity is presumably blocked in para^ts mutants at elevated temperature, the slow recovery of double mutants suggests that the temperature-sensitive mechanism of ATPα^DTS is not strictly dependent on neuronal electrical activity, which, for example, could contribute to further disruption of ionic gradients when Na⁺/K⁺ ATPase activity is impaired. The difference in para and ATPα^DTS temperature-sensitive mechanisms becomes more revealing upon prolonged exposure to 38°. Following 15 min of paralysis at 38°, para^ts animals quickly recover upon returning to 22°. However, when ATPα^DTS mutants are exposed to 38° for the same length of time >60% never recover and survivors require >1 hr before exhibiting any movement. Thus, it appears that at elevated temperatures para^ts mutants simply cease neuronal activity while ATPα^DTS mutants cease neuronal activity and initiate a toxic mechanism, possibly disruption of ion homeostasis.

To test if these interactions are simply due to the summation of less viable phenotypes, we also examined various ATPα mutants with heterozygosity for several of these recessive ion channel mutations. Animals heterozygous for various channel mutations also revealed significant interactions with ATPα mutants (Figure 6; Table 1). ATPα^DTS1 mutant females heterozygous for K⁺ channel mutations did not show a difference in adult viability from single-mutant control animals (Figure 6A), consistent with the results that we obtained from our analysis of double mutants where the K⁺ channel mutations were hemizygous (Figure 5A). Analysis of heterozygosity for para alleles in combination with ATPα mutations gave different results, depending on the particular allele. para^ts1, para^ts115, and para^st109 all exhibit shorter life spans when heterozygous in an ATPα^DTS genetic background (Figure 6B). Surprisingly, heterozygosity for the null allele, para^lk5, or the deficiency Df(1)lD34 do not shorten either the ATPα^DTS1 or the ATPα^DTS2 life span. The differences in these para heterozygotes may result from modifiers in these genetic backgrounds. Histological analyses of para^ts1/+; ATPα^DTS1/+ animals show an increase in the severity of spongiform neuropathology compared with ATPα^DTS1/+ age-matched controls (Figure 6C). Behavioral testing of double-mutant animals revealed no significant changes in temperature sensitivity (data not shown). Thus, although their behavioral phenotypes are somewhat distinct, a link between the mechanisms underlying para and ATPα mutant pathologies seems to exist.

Figure 6.

Animals heterozygous for ion channel mutations interaction with ATPα^DTS1. (A) Heterozygous sei/+, eag/+, Sh/+, and eag Sh/+ + do not show any interaction with ATPα mutants. (B) Heterozygous para allele combinations ts1/+ (diamonds), st109/+ (open squares), and ts115/+ (triangles) with ATPα^DTS1/+ show a significant reduction in life span (P < 0.05). Surprisingly, the para deficiency Df(1)LD34 (circles) did not have any affect on ATPα^DTS1 life span. (C) para^ts1/+;;ATPα^DTS1/+ brain slice at day 16 shows more severe pathology than age-matched para^ts1 or ATPα^DTS1 alone. Each point represents mean survival ±SEM. Bar, 50 μm.

Normal expression and localization of sodium pump in dominant ATPα mutants:

Na⁺/K⁺ ATPases have a fundamental cellular role in establishing and regulating ionic gradients, which is especially important in neurons. Although defects in ion homeostasis could alter neural activity and lead to intracellular concentrations of ions that are toxic (e.g., Na⁺, K⁺, and, secondarily, Ca²⁺), protein trafficking defects that disrupt the endoplasmic reticulum (ER) secretory pathway and other cellular functions could also result in the observed neurodegeneration. Furthermore, a mutation in a vertebrate Na⁺/K⁺ ATPase that is similar to the dominant ATPα mutations in Drosophila has been shown to cause defects in processing of the mutant protein, resulting in reduced expression overall (Arguello et al. 1999).

If the ATPα^DTS mutations act in a dominant negative manner affecting the secretion of the tetramer, there could be a significant reduction in Na⁺/K⁺ ATPase trafficking to the membrane. To test if protein expression is reduced through aberrant trafficking or maturation, we performed immunohistochemical analysis. We saw no significant intracellular Na⁺/K⁺ ATPase α immunoreactivity and no colocalization with PDI∷GFP (an ER marker; Bobinnec et al. 2003) in vivo (data not shown). It is possible that the ATPα^DTS mutations do cause inefficient secretion of the protein but the proteins are rapidly degraded or their intracellular immunoreactivity is poor. To test this possibility, we quantified ATPα protein localized to nerve bundles in ATPα^DTS1 and wild-type animals. There was no reduction of ATPα immunoreactivity in motor neuron bundles in ATPα^DTS1 compared with wild-type animals (Figure 7). To assess the sensitivity of our assay, we also examined ATPα^DTS1R1 animals (heterozygous null mutation) for ATPα protein expression in nerve bundles. These mutants did show a significant reduction of ATPα protein (∼35%, P < 0.001), demonstrating that our assay is sufficiently sensitive to detect changes and the basis for ATPα haplo-insufficiency phenotypes. These data argue strongly against a simple protein-trafficking defect in dominant ATPα mutants and suggest that the protein dysfunction is at the membrane and likely compromises ion homeostasis.

Figure 7.

ATPα immunolocalization in axon bundles. Immunohistochemistry with anti-Na⁺/K⁺ ATPase α mAB and a fluorescence-conjugated secondary antibody was performed on larvae of the indicated genotypes. Confocal microscopy was performed under identical conditions and image acquisition parameters for all genotypes. Representative images of motor axon bundles are shown for (A) wild type, (B) ATPα^DTS1/+, and (C) ATPα^DTS1R1/+. ImageJ software was used for intensity analysis. Wild type had an average intensity of 509.11 ± 40.45, ATPα^DTS1 had an average intensity of 558.01 ± 33.40, and ATPα^DTS1R1 had an average intensity of 330.17 ± 12.28 (error given is SEM). Protein expression was not different between ATPα^DTS1 and control flies, whereas ATPα^DTS1R1 flies demonstrated a significant 35% reduction in ATPα expression (P < 0.001, indicated by an asterisk). N = 14 motor neurons evaluated from three independent animals.

Early lethality in dominant ATPα mutants may not result from excitotoxicity alone:

Using another method to test whether the neuropathological phenotypes in ATPα mutants are due to hyperexcitability, we asked whether these phenotypes could be suppressed by expressing the inhibitory electrical knockout (EKO) K⁺ channels (White et al. 2001) in these flies. Expression of the UAS–EKO E323 strain was driven in ATPα^DTS1 flies using the second chromosome heat-shock (HS) and third chromosome elav–GAL4 drivers. The HS–GAL4 driver is weakly expressed in all tissues at the constant temperature of 29° and the elav3A line expresses GAL4 at moderate levels throughout the nervous system. Use of neuronal drivers with greater levels of expression were lethal due to severe neuronal silencing (White et al. 2001; data not shown). We observed no significant change in life span of ATPα^DTS1 when EKO was expressed under the control of either driver (26 ± 7 days; P = 0.438). An identified mechanism for excitotoxicity in vertebrates leading to neuronal death is prolonged activation of the N-methyl-d-asparate (NMDA) glutamate receptor (Schanne et al. 1979; Siesjo and Bengtsson 1989; Choi and Rothman 1990). MK801 is an antagonist of the NMDA receptor and a strong blocker of excitotoxicity (Foster et al. 1987; Foster and Wong 1987). NMDA receptors have been shown to be expressed in the adult fly brain (Ultsch et al. 1993; Chiang et al. 2002) and MK801 has been demonstrated to block central pattern generators in Drosophila larvae (Cattaert and Birman 2001). We fed various concentrations of the drug in a yeast paste (Cattaert and Birman 2001) to ATPα^DTS mutants. None of the concentrations administered (0, 0.1, 0.5, and 1 mg/ml) were successful in significantly extending the life span of ATPα^DTS1 mutants (data not shown). Consistent with what we observed using Na⁺ channel mutations to limit excitability in ATPα mutants, the results of these experiments suggest a distinct mechanism, independent from increased excitability or excitotoxicity, which may be contributing to the early death of ATPα^DTS1 mutants.

DISCUSSION

Ion channel mutants exhibit progressive neuron loss:

Although a number of ion channel mutations exist in Drosophila and other organisms that perturb neuronal signaling mechanisms in a variety of ways, there are no systematic studies on how these functional impairments affect neuronal viability. In this study, we examined a collection of Drosophila mutants with altered membrane excitability to determine if they had an increased incidence of neurodegeneration. Mutations in eag, Sh, and sei K⁺ channels all cause increased neuronal excitability (Ganetzky and Wu 1982a, 1983; Jackson et al. 1984; Elkins and Ganetzky 1990). Although we did not find any striking reduction in adult life span in any of these mutants, we did commonly observe varying degrees of neurodegeneration. Among these mutants, neurodegeneration was most severe in sei. At elevated temperatures, sei mutants display substantially increased and continuous neural activity in the flight motor pathway that parallels their behavioral seizures (Kasbekar et al. 1987; Elkins and Ganetzky 1990). Sh and eag mutations have different effects on this same circuit (Engel and Wu 1992). Conversely, Sh and eag have profound effects on motor neuron activity at the larval neuromuscular junction, whereas sei looks normal (Jan et al. 1977; Ganetzky and Wu 1982b, 1983, 1986). The different neuropathological consequences of these mutants thus may be directly related to the precise ways in which neuronal activity and/or synaptic transmission are perturbed in each mutant. These mechanisms could be strongly dependent on important differences in levels of expression and subcellular localization of the different K⁺ channel types. Sh⁵, not used in our studies, was previously reported to be short lived at 25° (Trout and Kaplan 1970), and recent analysis revealed that numerous Sh mutations caused striking reductions in normal sleep and displayed shortened life spans when outcrossed to remove accumulated modifiers (Cirelli et al. 2005). It is possible that unidentified modifiers in the backgrounds of some of the mutants that we have examined reduce the severity of the neurodegenerative phenotypes, although we observed no changes in the life span of outcrossed Sh¹³³ mutants. Another possible complication is that individual neurons express a large number of different K⁺ channel subtypes, some of which may share substantial functional overlap (Ganetzky and Wu 1983; Sutherland et al. 1999; Misonou and Trimmer 2004). As a consequence, eag Sh double mutants show striking synergistic effects on neuronal hyperexcitability in comparison with either single mutant. In addition, eag Sh double mutants are severely uncoordinated under all conditions and die within a week of eclosion, although they exhibit no detectable neurodegeneration during their shortened life spans. If there is a time-dependent component to the onset and progression of neurodegeneration, it is possible that these extremely hyperexcitable double mutants die from other causes too quickly to manifest severe neuropathological lesions.

Although K⁺ channel mutations lead to hyperexcitability, the neurodegeneration observed in these mutants might be a consequence of neuronal dysfunction, rather than excitotoxicity per se. Consistent with this idea, we have found that mutations affecting Na⁺ channels, which decrease neuronal membrane excitability, also cause neurodegeneration. In fact, the severity of the neurodegenerative phenotypes in Na⁺ channel mutants appears to be greater than that for K⁺ channel mutants. The fact that the neuropathological phenotypes reported here do not correlate with simple changes in membrane excitability or with current models of excitotoxicity is intriguing. We observed neurodegeneration not only for mutations of para, which encodes voltage-gated Na⁺ channels, but also for the mle^napts mutation, which reduces expression of para-encoded Na⁺ channels through a post-transcriptional splicing defect (Reenan et al 2000). The more striking neurodegeneration observed in mle^napts corresponds with a more severe reduction in neuronal activity at permissive temperatures (Ganetzky and Wu 1982a; Ganetzky 1984).

The effect of mle^napts on life span and neurodegeneration appears to be more severe than that of para mutants themselves, even though the temperature-dependent paralysis and conduction block in mle^napts is mediated primarily through reduction in para expression (Wu et al. 1978; Kernan et al. 1991; Reenan et al. 2000). Despite these similarities, as only a single allele was assayed here, background effects cannot be discounted. Nonetheless, because of its role in post-transcriptional processing and editing of mRNA, it is likely that the mle^napts mutation also affects the expression of other, as yet unidentified, proteins in addition to Na⁺ channels. Similarly, recent studies have shown that numerous nervous system transcripts undergo RNA editing by dADAR (Hoopengardner et al. 2003) and mutations in dADAR result in striking age-dependent neurodegeneration (Palladino et al. 2000). Our finding that mutations in a variety of individual ion channel genes can produce neurodegeneration suggests that the more severe neuropathology observed in dADAR mutants may result from cumulative neuronal dysfunction owing to defects in RNA editing of transcripts for multiple different ion channels and other proteins required for neurotransmission.

Mechanisms of neurodegeneration in ATPα mutants:

Establishment and maintenance of the ionic gradients that produce a normal membrane resting potential are essential for proper neural function and are critically dependent on the activity of the Na⁺/K⁺ ATPase. Indeed, most of the ATP expended by neurons is used to fuel the Na⁺/K⁺ pump. Not surprisingly, mutations affecting the activity of the Na⁺/K⁺ ATPase cause profound electrophysiological and behavioral perturbations. These mutations provide another window for examining the relationship between neuronal activity and neuronal viability. We previously found that a 50% reduction in ATPα expression (heterozygosity for a null allele) results in shortened life span, conditional paralysis, and neurodegeneration (Palladino et al. 2003). Heterozygotes for dominant mutations of ATPα have similar but more severe phenotypes. Both the null mutations and the dominant mutations cause lethality when homozygous (Palladino et al. 2003; Paul et al. 2003). Although the residues altered in the dominant mutations have been implicated in maturation of the protein (Arguello et al. 1999), we found no significant change in protein levels or localization in ATPα^DTS1 mutant nerves. Thus, overt defects in trafficking or localization of the Na⁺/K⁺ pump do not account for the dominant phenotypes of ATPα^DTS mutants.

Instead, our data support the idea that both the dominant and recessive ATPα mutations cause loss of Na⁺/K⁺ ATPase function that alters ion homeostasis with resultant defects in membrane excitability and neuronal signaling. Na⁺/K⁺ ATPase activity is the primary means of K⁺ uptake in neurons. Its failure leads to depletion of intracellular K⁺ and accumulation of intracellular Na⁺ and Ca²⁺ (Archibald and White 1974; DiPolo and Beauge 1991; Xiao et al. 2002). Low intracellular K⁺ and high intracellular Ca²⁺ have been linked to apoptotic- and necrotic-mediated cell death, respectively (Hughes et al. 1997; Hughes and Cidlowski 1999; Yu and Choi 2000). Ouabain, a specific antagonist of Na⁺/K⁺ ATPase activity, also causes dose-dependent hyperexcitability and neural degeneration (Bignami and Palladini 1966; Lees and Leong 1994, 1995, 1996) similar to the effects of ATPα mutants. To account for the more severe phenotype associated with the dominant ATPα mutations in comparison with haplo-insufficiency, we propose that the dominant mutations represent antimorphic alleles that interfere with the activity of the normal polypeptide. Thus, the residual Na⁺/K⁺ ATPase activity in heterozygotes would be even <50%. Although this interpretation is consistent with all of the available data, we cannot rule out the possibility that the dominant ATPα mutations cause a novel gain-of-function that is responsible for the more severe neurodegeneration caused by these mutations.

As inhibition of the Na⁺/K⁺ ATPase results in depolarization of the membrane potential (Spuler et al. 1988), induces convulsions, and can result in excitotoxic neuronal death (Lewin 1970; Lorenzo 1970; Lees and Leong 1994), we tested the idea that the neuronal hyperexcitability caused by ATPα mutants is directly responsible for neurodegeneration, perhaps through an excitotoxic mechanism. If this were the case, we expected the phenotype to become more severe in combination with K⁺ channel mutations that further increase membrane excitability and to be suppressed in combination with Na⁺ channel mutations that decrease membrane excitability. However, neither Sh nor eag mutations enhanced neurodegeneration caused by dominant or recessive ATPα mutations, suggesting that increased neuronal excitability by itself is probably not the direct cause of neurodegeneration. Nonetheless, since we have examined only the effect of a small subset of all K⁺ channels, it is possible that eliminating other K⁺ channels, depending on their pattern of expression and subcellular localization, would lead to substantially greater neurodegeneration in combination with ATPα mutations.

Numerous studies have demonstrated that Na⁺ channel inhibition protects against excitotoxicity (reviewed in Spedding and Lepagnol 1995; Obrenovitch 1997; Hemmings 2004). Thus, if ATPα mutants were causing excitotoxic-dependent neurodegeneration, we would expect that para mutations, which decrease neuronal excitability (Wu and Ganetzky 1980; Loughney et al. 1989; Stern et al. 1990), would partially rescue the phenotype. Instead, we found that dominant ATPα mutations in combination with para mutations resulted in flies that were uncoordinated, quickly manifested overt locomotor impairment, and had significantly reduced life spans. Although the life span of the double mutants was greatly reduced compared with ATPα^DTS mutants alone, no significant neuropathology was observed. Possibly, these double mutants die too soon for significant degeneration to develop. This result suggests that the early death of the double mutants is related to severe neural dysfunction, which may also trigger neurodegeneration, but neurodegeneration itself is not the cause of death. In any case, the interaction is different from what would be expected according to a simple excitotoxic model. The basis for the strong interaction between para and ATPα mutations is still unknown. One possibility is that ATPα^DTS mutations produce a partial depolarization that inactivates Na⁺ channels, similar to the mechanism occurring in periodic paralyses (Cannon 2000) and further reducing excitability in para^ts mutants. However, para-like reduced excitability cannot be the only component mediating dominant ATPα temperature sensitivity as prolonged exposures to nonpermissive paralysis-inducing temperatures can be lethal for ATPα^DTS1 animals while para^ts mutants recover quickly.

Decreases in K⁺ levels and increases in Ca²⁺ levels have been strongly linked to apoptotic and necrotic pathways, respectively. Both pathways have been associated with disruption of Na⁺/K⁺ ATPase function (Chatterjee and Roy 1965; Lees et al. 1990; Beauvais et al. 1995; Bortner et al. 1997; Yu et al. 1997), which recently has been found to induce “hybrid death” with concurrent signs of apoptosis and necrosis in the same cells (Xiao et al. 2002; Yu 2003). Lack of protection from introducing Na⁺ channel mutations and lack of aggravation from removing K⁺ channels, combined with failure of EKO expression and MK801 administration to rescue early lethality in ATPα alleles, all suggest that excitotoxicity is not the sole cause of the mutant pathology. This may be due to a “hybrid death” mechanism present in our mutants where, in the absence of excitotoxic-induced necrosis, decreased K⁺ levels may still induce apoptosis. We cannot rule out the possibility that both apoptotic and necrotic mechanisms act in parallel to produce neurodegeneration in the ATPα mutants. Our Na⁺ channel results suggest that multiple processes resulting in neuronal dysfunction and death and possibly activating interrelated injury cascades may be at work in these mutants (Stys 2005).

As mutations in ion channels and Na⁺/K⁺ ATPase α-subunits have been linked to a variety of cardiovascular and neuronal disorders (Jewell et al. 1992; Lingrel 1992; Renkawek et al. 1992; Lees 1993; Calandriello et al. 1995; James et al. 1999; Hubner and Jentsch 2002; Kullmann 2002; De Fusco et al. 2003; Vanmolkot et al. 2003), including rapid-onset dystonia with parkinsonism (de Carvalho Aguiar et al. 2004), characterizing the precise mechanisms underlying these pathologies is essential. Our data suggest that there may be specific “quality control” pathways monitoring the state of neuronal activity. Not only may these pathways initiate cell death upon reduction of K⁺ concentrations, but also specific mechanisms that act as checkpoints to distinguish normal activity from aberrant activity for a given neuronal type and subcellular domain may exist. These results provide the first evidence for neurodegeneration in animals containing mutations affecting either voltage-gated K⁺ or Na⁺ channels. Our data also suggest that excitotoxicity is not the sole mechanism inducing the massive neuronal death observed in animals containing dominant mutations in the Na⁺/K⁺ ATPase. These findings suggest that neuronal viability requires proper maintenance of neuronal signaling, which can be disrupted by either increasing or decreasing membrane excitability. Understanding the mechanisms initiating these degenerative pathways may ultimately have important therapeutic applications for prevention of neuronal death resulting from prolonged disruption of ionic concentrations under periods of ischemia, hypoxia, hypoglycemia, and trauma.