Drosophila couch potato Mutants Exhibit Complex Neurological Abnormalities Including Epilepsy Phenotypes

Abstract

RNA-binding proteins play critical roles in regulation of gene expression, and impairment can have severe phenotypic consequences on nervous system function. We report here the discovery of several complex neurological phenotypes associated with mutations of couch potato (cpo), which encodes a Drosophila RNA-binding protein. We show that mutation of cpo leads to bang-sensitive paralysis, seizure susceptibility, and synaptic transmission defects. A new cpo allele called cpo^EG1 was identified on the basis of a bang-sensitive paralytic mutant phenotype in a sensitized genetic background (sda/+). In heteroallelic combinations with other cpo alleles, cpo^EG1 shows an incompletely penetrant bang-sensitive phenotype with ∼30% of flies becoming paralyzed. In response to electroconvulsive shock, heteroallelic combinations with cpo^EG1 exhibit seizure thresholds less than half that of wild-type flies. Finally, cpo flies display several neurocircuit abnormalities in the giant fiber (GF) system. The TTM muscles of cpo mutants exhibit long latency responses coupled with decreased following frequency. DLM muscles in cpo mutants show drastic reductions in following frequency despite exhibiting normal latency relationships. The labile sites appear to be the electrochemical GF-TTMn synapse and the chemical PSI-DLMn synapses. These complex neurological phenotypes of cpo mutants support an important role for cpo in regulating proper nervous system function, including seizure susceptibility.

RNA-BINDING proteins perform myriad crucial roles throughout the life of an RNA molecule in eukaryotes. Subsequent to their genesis in the nucleus during transcription, pre-messenger RNAs (pre-mRNA's) are bound by RNA-binding proteins, which mediate their maturation into mRNA's via processing reactions, such as splicing, editing, capping, and polyadenylation. RNA-binding proteins then assist in transporting mRNA's to the cytoplasm where they are instrumental in regulating the translation, stability, and localization of the transcripts. In addition to translated RNAs, the discovery of regulatory nontranslated RNA genes, termed micro RNA's because of their minuscule size (<100 nucleotides) suggests additional functions for RNA-binding proteins (Ambros 2001). Thus, RNA-binding proteins serve a most critical role in the control of gene expression, especially in the nervous system where extensive alternative splicing occurs and aberrations frequently result in neurological disease.

Many neurological disorders result when the performance of RNA-binding proteins goes awry, highlighting their importance in the maintenance of fundamental neuronal processes. For example, in the human neurological condition paraneoplastic opsoclonus myoclonus ataxia (POMA), patients lose inhibitory control of motor neurons in spinal cord and brainstem. POMA is associated with the ectopic expression of the NOVA family of RNA-binding proteins, which regulate neuron-specific alternative splicing (Jensen et al. 2000a,b). In humans with fragile X syndrome, impaired expression of the cytoplasmic RNA-binding protein, FMRP, leads to mental retardation, likely resulting from misregulation of mRNA transport or translation (Perrone-Bizzozero and Bolognani 2002).

Several neurological disorders that have been characterized in animal models with defective RNA-binding proteins include jerky and quaking in mice and pumilio in Drosophila. The jerky mice exhibit temporal lobe epilepsy analogous to the most common seizure disorder in human adults. The jerky gene encodes an RNA-binding protein postulated to regulate mRNA usage in neurons, which is inactivated in the mutant (Liu et al. 2002). The quaking mice exhibit tonic-clonic seizures and hypomyelination. An RNA-binding protein involved with mRNA nuclear export appears responsible for the “quaking” defects (Larocque et al. 2002). In Drosophila, pumilio mutants show defects in embryonic development and maintenance of neuronal excitability. The pumilio mutant exhibits increased rates of long-term facilitation at the larval neuromuscular junction (Schweers et al. 2002). The pumilio gene has been shown to encode an RNA-binding protein that acts as a translational repressor (Wreden et al. 1997).

The work presented here examines a Drosophila RNA-binding protein gene called couch potato (cpo) that, when impaired, causes several neurological abnormalities including epilepsy phenotypes. The cpo gene was originally identified in a screen for genes expressed in sensory organ precursor cells during peripheral nervous system (PNS) development (Bellen et al. 1992a). Cpo protein is localized to the nucleus and is expressed in the PNS and central nervous system (CNS) of embryos, larvae, and adults, as well as other tissues such as midgut, glia, and salivary glands (Bellen et al. 1992b). The protein contains an RNA recognition motif (RRM) and a nuclear localization sequence (Bellen et al. 1992b). The RRM domain of cpo shows homology to the hermes gene of Mus musculus, Gallus gallus, and Xenopus laevis; the Caenorhabditis elegans gene mec-8; and the human gene RBP-MS (Lundquist et al. 1996; Shimamoto et al. 1996; Gerber et al. 1999). The cpo gene has also been linked to the human neurodegenerative disorder spinocerebellar ataxia type 1 (SCA1). The defective human SCA1 gene causes neurodegeneration when expressed in Drosophila; this phenotype is enhanced by overexpressing cpo (Fernandez-Funez et al. 2000). Partial loss-of-function mutations of Drosophila cpo cause a variety of behavioral phenotypes including overall sluggishness and abnormal phototaxis, geotaxis, flight ability, ether recovery, and mating vigor (Bellen et al. 1992a; Hall 1994).

In this article, we identify a new cpo allele, cpo^EG1. Electrophysiological analysis shows that the cpo^EG1 mutation contributes to numerous defects in the giant fiber (GF) neural circuit. Additionally, the cpo^EG1 mutation contributes to increased seizure susceptibility manifested as seizure thresholds that are less than half that of wild-type flies. We have examined existing cpo alleles and have shown that they fail to complement the electrophysiological defects associated with cpo^EG1. Taken together, these findings suggest a neurological basis for the complex cpo behavioral defects described previously. In addition to the previously postulated role that cpo plays in PNS differentiation and normal adult behavior, this work provides evidence that RNA-binding proteins are also essential for the proper functioning of synapses in the adult CNS and for regulating susceptibility to seizure.

MATERIALS AND METHODS

Fly stocks:

The cpo gene is located on the third chromosome at map location 90D1-6 and encodes a putative RNA-binding protein (Bellen et al. 1992b). The mutant cpo allele, cpo^EG1, was identified in a screen that selected for bang-sensitive paralytic mutant phenotypes revealed in a sensitized genetic background provided by slamdance heterozygotes (sda/+) (Zhang et al. 2002). The screen utilized P-element hybrid dysgenesis with the transposon P[lacZ, w⁺]. In brief, mutants were isolated in a mating of X^X, 8:P[lacZ, w⁺]/Y; ry Sb P[ry⁺ delta2.3]/+ females crossed to w/Y; sda males. Exceptional w/Y; sda/+ male progeny that showed bang-sensitive paralysis were individually crossed to set up appropriate stocks. The cpo^EG1 allele has previously been called line N (Zhang et al. 2002). Five additional cpo alleles, cpo^cp1, cpo^cp2, cpo^v3, cpo^l2, and cpo^lΔcp11, were obtained as a generous gift (H. Bellen, Baylor College of Medicine). The cpo^cp1 and cpo^cp2 alleles are due to 17-kb P[lArB] insertions at scaffold position 12297|12298. The cpo^cp1, cpo^cp2, and cpo^v3 alleles have been reported as viable with delayed development (Bellen et al. 1992a); at present, they behave as recessive lethals. The cpo^v3 allele is due to a 10-kb P[lacZ, w⁺] insertion at 12305|12306. The cpo^l2 allele is due to a P[lacZ, w⁺] insertion at 12297|12298. The cpo^lΔcp11 allele is due to 200-bp deletion of the cpo promoter, causing a loss of cpo expression, except in the chordotonal neurons, which express Cpo later in development than wild type (Bellen et al. 1992a,b). The cpo^l2 and cpo^lΔcp11 alleles behave as recessive lethals. Note that scaffold position 12297|12298 appears to be a hot spot for transposon insertions since at least seven of the known cpo alleles map to this position. This hot spot is 38-bp upstream of the cpo transcription start site. The cpo^Δ125 allele was generated in this work as a precise excision of the P element in cpo^EG1 flies that reverts cpo^EG1 phenotypes to wild type. The slamdance (sda) gene maps to 97D and encodes an aminopeptidase N (Zhang et al. 2002). The sda allele utilized was sda^iso7.8, caused by a 2-bp insertion in the 5′-untranslated region. The sda mutation causes seizure sensitivity and bang-sensitive paralysis (Zhang et al. 2002). Duplication and deficiency mapping for cpo^EG1 was done using strains Df(3R)P-14=Df(3R) 90C2-D1;91A1-2 and Tp(3;Y)L58=Tp(3;Y) 88D;93D;Y obtained from the Bloomington Stock Center at Indiana University. Stocks were maintained on standard cornmeal-molasses medium at 21°–24°. Other markers and stocks are described in Lindsley and Zimm (1992).

Behavioral testing:

Testing for bang-sensitive (BS) paralysis was performed on flies 1–4 days posteclosion. BS paralysis was assayed by transferring the flies to a clean empty food vial (Applied Scientific) and then vortexing on a VWR vortex at maximum setting for 10 sec. Any fly that lay motionless following the vortex was scored BS. Flies were considered recovered when they were able to resume an upright standing position. Flies were rested for a minimum of 2 hr following CO₂ anesthesia prior to testing.

Electrophysiology:

Electrophysiological analysis was performed on male flies using methods previously described to stimulate and record GF-driven muscle potentials and seizures (Kuebler and Tanouye 2000). Briefly, the fly was removed from a food vial by sucking onto its head with a 23-gauge syringe needle attached to a vacuum line. Another syringe needle was then used to suck onto the abdomen, thereby completely immobilizing the fly. The fly was then mounted by gluing a tungsten wire across the fly's neck. In experiments, the GF was driven by brain stimulation via bipolar tungsten electrodes inserted through the ventral antennal margin and into the brain. The preparations were grounded by placing an electrode into the abdomen. Stimulating, recording, and ground electrodes were all made from uninsulated tungsten wire (WPI 0.075 mm), electrolytically sharpened to the desired diameter, usually <1 μm. Experiments requiring thoracic ganglion stimulation of dorsal longitudinal muscle (DLM) and tergotrochanteral muscle (TTM) motoneurons were conducted similarly except that stimulating electrodes were bent at 45° angles and inserted just through the anterior pre-episternum, near the base of the first coxa, as described in Kuebler and Tanouye (2000).

To assess GF circuit performance, single pulses of 0.2 msec were used. Latency and following frequency experiments were done at 21°–24° on flies that were ≤7 days posteclosion, unless otherwise noted. The latency was the time from the end of the stimulus pulse to the beginning of the evoked muscle response. Following frequency was the highest frequency at which the muscle responded to 19 of 20 pulses at stimulation intensity 1.4 times the GF threshold. Flies were allowed to rest at least 1 min between following frequency trials. GF thresholds were defined as the lowest voltage that elicits a stable, short-latency (∼1.4 msec) DLM response.

Seizures were elicited by delivering a short wavetrain (0.5-msec pulses at 200 Hz for 300 msec) of high-frequency electrical stimuli (HFS) to the brain. Electroconvulsive shock (ECS) is HFS of sufficient intensity (i.e., above threshold) to elicit seizure. Seizures consist of aberrant high-frequency activity in at least seven different muscle groups and >30 muscle fibers in the thorax (Kuebler and Tanouye 2000). This seizure-like activity in a particular muscle corresponds to activity in the motoneuron that innervates it. Seizure is followed by a period of synaptic failure in the GF pathway (Pavlidis and Tanouye 1995). In this work, seizures were monitored by DLM activity. Recovery time was assessed as the time required to elicit four consecutive stable DLM responses following a buzz using a GF stimulation rate of 0.5 Hz.

Molecular mapping of cpo^EG1:

Plasmid rescue of genomic DNA was used to determine the approximate insertion site of the P element in cpo^EG1 flies (Wilson et al. 1989). Genomic DNA (2–5 μg) was digested with EcoRI and fragments were self-ligated with T₄ DNA ligase. Ligated products were transformed into XL-1 Blue electrocompetent cells (Stratagene, La Jolla, CA) and the transformants were selected on kanamycin (10 mg/ml) plates. Plasmid DNA from positive clones (Amp+) was isolated and the genomic fragment DNA was sequenced using a primer complementary to a site near the end of the P-element sequence (CGACGGGACCACCTTATGTTATTTCATCATG). The exact insertion site was then determined by sequencing back toward the P element using a primer specific for the flanking genomic fragment (GCACGAGACGAGCAGCTA).

Anatomy of the GF system:

GF morphology was examined by X-gal staining, utilizing a second chromosome enhancer trap line P[GAL4]A307 that expresses Gal4 predominantly in the GFs (Phelan et al. 1996; Allen et al. 1998). Mutant cpo flies were generated with A307 in heterozygous combination with a second chromosome UAS-lacZ insertion to drive β-galactosidase expression in the GFs. The nervous systems of these A307/+ UAS-lacZ/+; cpo flies were dissected in phosphate-buffered saline (PBS) and fixed in 0.5% glutaraldehyde in PBS for 15–30 min. These whole-mount preparations were then washed in 0.1% Triton-X 100 in PBS (PBT) and stored for up to 1 day at 4°. For β-galactosidase detection, the tissues were equilibrated with staining solution [5 mm K₃Fe(CN)₆, 5 mm K₃Fe(CN)₆, 3 mm MgCl₂ in PBT] for 10 min at 37° and then incubated in freshly prepared 0.08% X-gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) in staining solution at 37° for ∼2 hr. Following staining, the tissues were washed twice in PBS and then dehydrated in a 50–100% ethanol series. The dehydrated tissues were then cleared and mounted using xylenes.

RESULTS

Isolation of the cpo^EG1 mutation:

The cpo^EG1 mutation was identified in a P-element mutagenesis screen for new seizure-sensitive mutants that utilized sda heterozygotes as a sensitized genetic background (Zhang et al. 2002). Subsequent mapping and complementation tests showed that cpo^EG1 is an allele of cpo. Behavioral and electrophysiological analysis showed that cpo^EG1 and other cpo alleles differ from canonical members of the BS paralytic class of Drosophila. The cpo mutations are pleiotropic: seizure sensitivity is only one of numerous neurological abnormalities.

Seizure sensitivity is associated with members of the BS paralytic class of behavioral mutants including sda (Pavlidis and Tanouye 1995; Kuebler and Tanouye 2000). BS mutants display a unique behavioral response to mechanical shock, such as a brief vortex. When subjected to such a mechanical stimulus, BS flies display bouts of hyperactivity, marked by wing flapping, leg shaking, and abdominal contractions. After a few seconds, this hyperactive episode gives way to a period of temporary paralysis, in which the fly lies completely motionless for a time of 30 sec to a few minutes depending on genotype (Benzer 1971; Ganetzky and Wu 1982; Pavlidis et al. 1994). Mutants of the bang-sensitive class include bang-sensitive (bas), bangsenseless (bss), easily shocked (eas), sda, and technical knockout (tko). The eas gene encodes an ethanolamine kinase, sda encodes an aminopeptidase N, and tko encodes a mitochondrial protein (Royden et al. 1987; Pavlidis et al. 1994; Zhang et al. 2002).

Heterozygotes of sda appear to be a useful tool for detecting new seizure-sensitive mutants. Seizure threshold measured electrophysiologically is 21.0 ± 0.6 V ECS for sda/+ flies, in between the threshold for sda homozygotes (6.2 ± 0.8 V ECS) and CS wild-type flies (30.1 ± 3.8 V ECS) (J. Tan, personal communication; Kuebler et al. 2001). Following mechanical stimulation, most sda/+ flies (>98%) exhibit normal behavior, but a few exhibit BS paralysis (Zhang et al. 2002). Our interpretation from this is that the sda/+ genetic background is sensitized for seizures, indicated especially by the few flies that display BS paralysis. We infer further that mutations increasing the percentage of flies paralyzed by mechanical stimulation would indicate new seizure-sensitive mutants. The cpo^EG1 mutation was isolated on the basis of this phenotype: a strain with the apparent genotype cpo^EG1/+ sda/+ displayed 33% BS paralysis following mechanical stimulation. The cpo^EG1 mutants were analyzed for BS behavior independent of the sda mutation. As heterozygotes, cpo^EG1/+ flies show no paralysis following mechanical stimulation. The behavior of cpo^EG1 homozygotes could not be tested directly since cpo^EG1 causes recessive lethality.

Molecular and genetic basis of cpo^EG1 mutation:

The cpo^EG1 mutation was mapped by both molecular and genetic methods. Molecularly, plasmid rescue was used to isolate genomic DNA flanking the insertion site of the P element responsible for cpo^EG1. This flanking DNA was then sequenced and compared to the information available in the Drosophila genome database using the Berkeley Drosophila Genome Project Blast program. This comparison identified the location of the cpo^EG1 insertion within the cpo gene in the 90D1-6 region of the third chromosome (Figure 1). By sequencing the flanking genomic DNA back toward the P element, the exact location of the insertion was identified. The P element responsible for the cpo^EG1 mutation is inserted in an intron 38 bp upstream of the transcription start site, between nucleotides 12297 and 12298 of genomic scaffold AE003720 (release 3), corresponding to nucleotides 415 and 416 of the Cpo 61.1 cDNA described in Bellen et al. (1992b).

Figure 1.—

The cpo locus at 90D on chromosome III. The cpo locus at 90D1-6 is large and complex, spanning >100 kb and encoding three different messages: cpo 61.1 (3.4 kb), cpo 61.2 (2.5 kb), and cpo 17 (3.0 kb). For simplicity, only the transcript corresponding to cDNA 61.1 is shown; the mutations described here apparently affect only this transcript (Bellen et al. 1992b). The exons are represented as numbered boxes. The solid boxes correspond to untranslated exons, while the hatched boxes correspond to protein-coding exons. The relationship of cpo 61.1 to the corresponding nucleotide numbers in genomic scaffold AE003720 (release 3) is indicated below the splicing diagram. All of the alleles examined in this study are P-element insertions in the promoter region of transcript 61.1 just 5′ to exon 1. Interestingly, the cpo alleles EG1, l2, cp1, and cp2 are all inserted at the same site between nucleotides 12297 and 12298 on scaffold AE003720. The v3 allele is a P-element insertion 8 nt 3′ of the other alleles. All of these insertions are positioned between enhancer-like sequences and the transcription start site for cpo 61.1 at 12335 (the transcription start site is indicated as a large boldface A in the expanded portion). This region has been shown to contain most of the key regulatory sequences for cpo expression (Bellen et al. 1992b).

The location of cpo^EG1 is situated between binding sites for several transcription factors and the transcription start site and appears to be a hot spot for P-element insertion. At least 10 cpo P-element alleles have been mapped to the same 100-bp region of 90D, especially between nucleotides 12297 and 12298 where cpo^EG1 is inserted. Interestingly, 6 other cpo alleles share the exact same insertion site as cpo^EG1: alleles cpo^cp1, cpo^cp2, cpo^l2, cpo^l3, cpo^l6, and cpo^v2. The different insertions produce different phenotypes apparently dependent on size, orientation, and type (Bellen et al. 1992a,b). The cpo^cp1 and cpo^cp2 mutations are hypomorphic 17-kb P[lArB] insertions that cause adult behavioral defects. The cpo^v2 mutation is a 10-kb P[lacZ, w⁺] insertion that produces no discernible phenotype and is inserted in the opposite orientation relative to the other alleles. The cpo^EG1 mutation appears identical to the recessive lethal mutations cpo^l2, cpo^l3, and cpo^l6. All are caused by 10-kb P[lacZ, w⁺] insertions at the exact same site and all result in homozygous lethality.

Complementation analysis and deficiency and duplication mapping also confirm that the cpo^EG1 mutation is a new allele of cpo. The cpo^EG1 mutation fails to complement lethality of cpo^l2 and cpo^lΔcp11. The cpo deletion Df(3R)P-14 fails to complement the lethality of cpo^EG1. Conversely, the cpo duplication Tp(3;Y)L58 complements cpo^EG1. Furthermore, cpo^EG1 phenotypes are reverted to wild type by precise excision of the P element, as seen in cpo^Δ125 flies.

Sequence analysis of the cpo gene:

Analysis of the cpo gene sequence reveals three putative ORFs that encode proteins containing a nuclear localization sequence, an OPA repeat, and a conserved RRM (Bellen et al. 1992b). Between residues 1 and 193 of Cpo 61.1 protein, cpo possesses a putative bipartite nuclear localization signal suggesting a role for cpo in the nucleus. This assertion is corroborated by the results of antibody staining, which show that Cpo protein is localized to the nucleus (Bellen et al. 1992b). Cpo protein also contains an OPA repeat between amino acids 194 and 296, implying a neural role for cpo that is supported by the behavioral, electrophysiological, and expression phenotypes of cpo (Bellen et al. 1992b). OPA repeats were first discovered in the neurogenic gene Notch, which is important for differentiation of the PNS (Wharton et al. 1985). They consist of glutamine-dense regions of CAG and CAA repeats and although their importance is not fully understood, they commonly appear in neural-specific proteins such as Notch. Similar long stretches of glutamine residues are also common in neural-specific genes such as Huntingtin, where they frequently show unstable expansion to pathological lengths, leading to neurodegenerative disorders (Reddy et al. 1999). Interestingly, cpo has been linked to the trinucleotide repeat disease spinocerebellar ataxia type 1. Overexpression of cpo enhances an SCA1-induced neurodegeneration phenotype in a Drosophila model of the disease (Fernandez-Funez et al. 2000).

Cpo protein also contains a single RRM in the 61.1 and 61.2 cDNAs that shows significant homology (>65% identical residues) to RRM domains in the hermes gene of M. musculus, G. gallus, and X. laevis; the C. elegans gene, mec-8; and the human gene, RBP-MS (Lundquist et al. 1996; Shimamoto et al. 1996; Gerber et al. 1999). RRM domains consist of 80–100 amino acids and are usually present in one to four copies in proteins that bind pre-mRNA, poly(A) RNA, heterogeneous nuclear RNA (hnRNA), and small nuclear RNA (snRNA). An individual RRM domain contains two conserved sequences called RNP1 and RNP2, which are necessary for binding to RNA, and the overall secondary structure β₁α₁β₂β₃α₂β₄, where the RNP1 and RNP2 sequences correspond to β₁ and β₃, respectively. Specificity of binding is conferred by the amino acids in the loop connecting β₂ and β₃ and by the residues at the termini of the RRM (Burd and Dreyfuss 1994). Of the genes with homology to the RRM domain of cpo, hermes and RBP-MS contain a single RRM domain as does cpo, while mec-8 contains two RRM domains. The hermes protein shows nuclear and cytopolasmic localization and is proposed to play a role in heart development, possibly by regulating translation or mRNA stability (Gerber et al. 2002). The mec-8 protein is nuclearly localized and mutants show embryonic lethality and adult chemosensory and mechanosensory neuronal defects associated with impaired body muscle attachments, resulting from defective alternative splicing (Lundquist and Herman 1994; Lundquist et al. 1996). The RBP-MS gene was originally identified in the search for the gene responsible for the premature aging disease, Werner syndrome, which has since been identified and is not associated with RBP-MS. At this time, no function or expression data are known for RBP-MS, but computer simulations support the idea that RBP-MS can bind to RNA (Sahasrabudhe et al. 1998). Of these genes, mec-8 seems to align most closely with cpo in expression and phenotype. Both exhibit nuclear localization and both are associated with nervous system developmental defects. In addition, both cpo and mec-8 possess alanine and glutamine (AQ) rich regions that are also common in other neural-specific proteins with RRM domains, such as embryonic lethal abnormal vision (elav) and musashi (Robinow et al. 1988; Nakamura et al. 1994; Lundquist et al. 1996). It should be noted, however, that the homology of cpo to hermes, mec-8, and RBP-MS is limited to the RRM domains of these proteins. Regardless, the striking similarity between the RRM domains of cpo and these genes indicates that they may bind similar targets and may act in similar pathways. When taken as a whole, the sequence, localization, and phenotype of cpo seem to indicate an important role for cpo in regulating nervous system-specific transcripts required for proper nervous system function, possibly by mediation of alternative splicing or RNA export out of the nucleus.

The cpo BS paralytic behavior:

Mutants of cpo have been shown to display a number of behavioral abnormalities, including sluggishness and abnormal phototaxis, geotaxis, flight, ether recovery, and mating vigor (Bellen et al. 1992a; Hall 1994). Here, we show that cpo mutants also show BS paralytic behavior with viable heteroallelic combinations. We examined especially the viable genotypes cpo^EG1/cpo^cp1, cpo^EG1/cpo^cp2, cpo^EG1/cpo^v3, cpo^cp1/cpo^cp2, cpo^cp1/cpo^l2, and cpo^cp2/cpo^l2. We confirmed cpo^EG1/cpo^cp1 and cpo^EG1/cpo^cp2 behavioral abnormalities of sluggishness and abnormal phototaxis, geotaxis, flight, ether recovery, and mating vigor (data not shown). Furthermore, we confirmed that cpo^v3 complements cpo behavioral abnormalities (data not shown). These are consistent with the observations of Bellen et al. (1992a). The heteroallelic combinations cpo^EG1/cpo^cp1 and cpo^EG1/cpo^cp2 show BS paralysis in 32 and 18% of flies, respectively (Figure 2). In addition, other viable combinations of the cpo alleles tested showed some degree of paralysis following mechanical stimulation, except for combinations involving cpo^v3 (Figure 2). The ability of the cpo^v3 allele to complement the paralysis phenotype of cpo mutants correlates with previous behavioral analysis (Bellen et al. 1992a).

Figure 2.—

Bang-sensitive paralysis in cpo mutants. Various heteroallelic combinations were created and tested for bang-sensitive paralysis. Paralysis is greatest with the cpo^EG1/cpo^cp1 combination in which 32% of flies become paralyzed (n = 299). Flies with the genotype cpo^EG1/cpo^cp2 show the second highest degree of bang sensitivity with 18% of flies becoming paralyzed (n = 210). This bang-sensitive paralysis defect is also seen in viable cpo mutant combinations independent of cpo^EG1. Viable mutant combinations of cpo^cp1/cpo^cp2, cpo^cp1/cpo^l2, and cpo^cp2/cpo^l2 show paralysis in 16% (n = 281), 12% (n = 65), and 7% (n = 55) of flies, respectively. Interestingly, the cpo^v3 allele complements the bang-sensitive paralysis associated with the cpo^EG1 mutation as no cpo^EG1/cpo^v3 flies become paralyzed (n = 56). The apparent allelic series is as follows: cpo^EG1 ≥ cpo^l2 > cpo^cp1 ≥ cpo^cp2.

The BS paralysis seen in cpo flies shares some similarities and differences with members of the BS paralytic class of behavioral mutants. The paralytic phenotype of cpo flies shows incomplete penetrance, unlike canonical members of the BS paralytic class, such as eas, bss, and sda, in which 100% of flies become paralyzed following mechanical stimulation. Paralysis in cpo flies usually lasts ∼25–45 sec, during which time the flies lie completely motionless with occasional slight leg movement. This recovery time is similar to the time required for recovery in sda, which ranges from 30 to 60 sec, but it is much briefer than the recovery period observed in eas and bss, which ranges from 60 to 300 sec for both (Zhang et al. 2002). Unlike BS paralytics, which experience a hyperactive phase during recovery, cpo mutants do not display hyperactivity as they recover, but instead just right themselves to a standing position and resume normal behavior. Upon recovery, cpo flies do not exhibit a well-defined refractory period like BS mutants. Following BS paralysis, cpo mutants do not always become reparalyzed, even when tested a day later, probably owing to the incomplete penetrance of the BS phenotype. Conversely, cpo mutants that did not become paralyzed after vortexing sometimes became paralyzed when tested later. The incomplete penetrance of the paralysis phenotype coupled with the potential of all cpo flies to become paralyzed suggests that cpo flies have a seizure susceptibility that is close to the threshold for manifestation of the BS phenotype and that they consequently receive adequate stimulus to become paralyzed some of the time, while other times they do not.

cpo flies have increased seizure susceptibility:

All mutants showing BS paralytic behavioral phenotypes, including bss, eas, and sda, have been found to have low seizure thresholds in electrophysiology experiments. We examined seizure thresholds for cpo mutations in heteroallelic combinations to examine correlates with the BS behavioral defect. The seizure threshold here is defined as the minimum voltage required to elicit aberrant high-frequency DLM motoneuron activity by administration of an ECS to the brain of the fly. However, it should be noted that previous work has shown that seizures are not limited to the DLMs, but rather involve at least seven different muscle groups and >30 thoracic motoneurons. A seizure, which appears as intense, high frequency muscle activity, is followed by a brief period of synaptic failure through the GF pathway during which time brain stimulation fails to elicit DLM muscle potentials.

The cpo^EG1/cpo^cp1 and cpo^EG1/cpo^cp2 mutants exhibit significant reductions in seizure threshold, with buzzes of 11.1 ± 3.7 V ECS (n = 10) and 13.3 ± 4.8 V ECS (n = 7), respectively, sufficient to elicit seizures (Figure 3; Table 1). These values are about two to five times higher than those in the other BS mutants, but still approximately two times lower than those in wild type, placing cpo flies in the unique position of being almost exactly intermediate between the known BS mutants and wild type with regard to seizure susceptibility. BS mutants have seizure thresholds that range from 3.2 ± 0.6 V (bss) to 6.2 ± 0.8 V (sda). Wild-type strains do not show seizure activity until given high-frequency stimuli ranging from 25.5 ± 3.7 V (Berlin) to 39.3 ± 6.6 V (Oregon-R) (Kuebler et al. 2001). Precise excision of cpo^EG1 in cpo^Δ125 flies reverts the seizure threshold to a wild-type level of 29.0 ± 4.5 V ECS (n = 10). This intermediacy in seizure threshold correlates with the BS paralytic behavior of cpo mutants, which shows only ∼30% penetrance compared with the 100% penetrance seen in the known BS mutants.

Figure 3.—

cpo mutants exhibit increased seizure susceptibility without a concomitant increase in single neuron excitability. (A) A representative seizure as recorded from the DLM following a high-frequency brain stimulus of 8 V ECS (depicted as a hatched box) in a cpo^EG1/cpo^cp1 fly. The DLM shows aberrant high-frequency firing approaching 50 Hz during the seizure phase. These high-frequency responses then give way to synaptic failure (not depicted) (calibrations, 10 mV and 200 msec). (B) A cpo^EG1/+ fly shows only a few spikes following an identical 8-V high-frequency stimulus to the brain, since this voltage is below its threshold for seizure. (C) However, seizures are elicited in cpo^EG1/+ flies following administration of 32 V ECS. (D) Comparison of seizure thresholds for various cpo genotypes shows increased seizure susceptibility for cpo mutants (n ≥ 6 for each genotype). (E) Comparison of the firing thresholds of the giant fiber in control flies and cpo mutants reveals no significant differences (n ≥ 6 for each genotype).

TABLE 1.

Performance of the giant fiber system incpo mutants and control flies

Genotype	Seizure threshold (V)	DLM latency (msec)	DLM following frequency (Hz)	TTM latency (msec)	TTM following frequency (Hz)
+/+	30.1 ± 3.8	1.25 ± 0.10	137 ± 14.7	0.81 ± 0.07	>100
cpoΔ125/cpoΔ125	29.0 ± 4.5	1.18 ± 0.08	128 ± 16.3	0.72 ± 0.10	191 ± 33.2
cpoEG1/+	28.4 ± 4.3	1.35 ± 0.12	124 ± 31.3	0.86 ± 0.10	216 ± 44.3
cpocp1/+	26.6 ± 7.4	1.31 ± 0.07	135 ± 30.6	0.89 ± 0.09	194 ± 40.3
cpocp2/+	34.6 ± 6.7	1.39 ± 0.12	138 ± 16.1	0.94 ± 0.15	185 ± 32.5
cpoEG1/cpocp1	11.1 ± 3.7	1.48 ± 0.09	22 ± 9.2	1.36 ± 0.05	<1
cpoEG1/cpocp2	13.3 ± 4.8	1.45 ± 0.05	14 ± 4.8	1.46 ± 0.24	<1

For each genotype, n ≥ 6.

The seizures in cpo flies appear qualitatively similar to those observed in other BS mutants with a few quantitative differences. Seizures in cpo flies exhibit the three phases that are characteristic of Drosophila seizures: initial seizure, synaptic failure, and recovery seizure. These three phases of seizure in cpo mutants differ quantitatively from those seen in other genotypes as would be expected since each of these characteristics is genotype dependent. The maximum response frequency during the initial seizure is reduced in cpo flies compared to other BS mutants. In cpo flies, the maximum response frequency during the seizure usually ranges between 20 and 40 Hz, whereas in other BS mutants the frequency can often approach 100 Hz. This frequency reduction scales with the impaired following frequency of the DLMs in cpo flies, which probably limits the attainable muscle firing frequency during seizure. The recovery period following seizure in cpo flies is consistent with the durations seen in other BS mutants. cpo^EG1/cpo^cp1 and cpo^EG1/cpo^cp2 flies show recovery times of 83.2 ± 23.8 sec (n = 5) and 98.6 ± 15.0 sec (n = 7), respectively, which is longer than the 48.9 ± 5.4-sec recovery for eas and shorter than the 136 ± 6-sec recovery for bss (J. Tan, personal communication). These recovery periods are decidedly longer than the 29.9 ± 1.6-sec recovery for wild-type Canton-S flies (J. Tan, personal communication) and the 34.4 ± 7.2-sec (n = 5) and 39.1 ± 5.5-sec (n = 10) recovery times for cpo^EG1/+ and cpo^Δ125/cpo^Δ125 flies, respectively. Also similar to other BS mutants, cpo flies always exhibit recovery seizures following synaptic failure. Finally, the decreased seizure threshold of cpo mutants is not accompanied by a change in GF response threshold, a phenomenon also observed in BS mutants (Figure 3).

cpo acts as an enhancer of seizure mutation:

The cpo^EG1/+ mutation was identified on the basis of genetic interaction with the BS paralytic mutation sda. We examined some general features of this interaction. In the behavioral screen that identified cpo^EG1, this allele was found to function as a dominant enhancer of the sda/+ BS paralytic phenotype. That is, ∼33% of cpo^EG1/+ sda/+ double heterozygotes showed BS paralysis (n = 425), a substantially greater number than that in sda/+ and cpo^EG1/+ parentals, ∼1% (n = 257) and 0% (n = 71), respectively. We examined the possibility that other cpo alleles also act as enhancers. The double heterozygote cpo^l2/+ sda/+ showed ∼44% BS paralysis (n = 119). The double heterozygote cpo^cp1/+ sda/+ showed ∼75% BS paralysis (n = 93). We examined the possibility that cpo^EG1 can enhance other BS mutations by testing for interaction with eas. The double heterozygote eas/+; cpo^EG1/+ showed ∼7% BS paralysis (n = 149), greater than the 0% BS paralysis (n = 103) that was shown for the eas/+ parental.

Interaction between cpo and sda is also observed in seizure threshold measured electrophysiologically. We have shown previously that seizure threshold for sda/+ heterozygotes is 21.0 ± 0.6 V ECS (J. Tan, personal communication). The cpo^EG1 allele acts as a dominant enhancer of seizure susceptibility: the cpo^EG1/+ sda/+ double heterozygote has a seizure threshold of 14.4 ± 3.1 V ECS (n = 10). The cpo^cp1 allele also acts as a dominant enhancer of seizure susceptibility: the cpo^cp1/+ sda/+ double heterozygote has a seizure threshold of 11.8 ± 1.6 V ECS (n = 9). Excision of the P element in cpo^EG1 eliminates the sda enhancement phenotype restoring the seizure threshold to 20.3 ± 1.1 V ECS (n = 10) in cpo^Δ125/+ sda/+ flies.

cpo flies exhibit synaptic defects in the GF system:

Mutants of cpo show alterations in GF-TTM response, but the basis for this defect is not known. In wild type, this response is via a monosynaptic pathway and has a characteristically short latency that is stable at high frequencies of stimulation. In cpo mutants, this short-latency response is replaced by a longer latency GF-TTM response that cannot follow even moderate frequencies of stimulation. This appears to be due to an alteration in the GF-TTM neurocircuit in the mutant. Possibilities for such a neurocircuit alteration include, for example: (1) an action potential conduction abnormality in the GF, (2) a synaptic transmission defect between the GF and the TTM motoneuron, (3) an action potential conduction abnormality in the TTM motoneuron, (4) a synaptic transmission defect at the TTM neuromuscular junction (NMJ), or (5) some other, more complex neurocircuit alteration. The abnormality in the GF-TTM pathway does not appear to be due to the TTM motoneuron or the TTM NMJ. This is indicated by direct electrical stimulation of the TTM motoneuron showing that the cpo mutant response is similar to wild type in latency and following frequency. The latency of the TTM response following direct motoneuron stimulation for cpo^EG1/cpo^cp1 is ∼0.4 msec, which is similar to the CS wild-type latency of 0.66 ± 0.05 msec (Tanouye and Wyman 1980). The following frequencies of the TTM response following direct motoneuron stimulation for cpo^EG1/cpo^cp1 and CS are both >100 Hz. The defect in the GF-TTM pathway is probably not due to an action potential conduction abnormality of the GF. This is inferred from an examination of the GF-DLM response in cpo mutants that is normal in latency and, although measurably impaired in following frequency, is markedly less impaired than the GF-TTM response (Figure 4; Table 1). From this we infer that the latency abnormality of the GF-TTM response is not due to the GF, which is common to both pathways. The latency abnormality and at least a portion of the following frequency defect must occur after the GF pathways to the TTM and DLM diverge. The implication from these observations is that the defect for the GF-TTM response occurs distal to the GF and proximal to the TTM motoneuron, that is, most likely at the synapse connecting the GF with the TTM motoneuron. Examination of GF anatomy in cpo^EG1/cpo^cp1 flies reveals no gross morphological abnormalities. In cpo flies, the GF appears to make the proper lateral bend in the mesothoracic neuromere to contact the tergotrochanteral motoneuron (TTMn)(data not shown).

Figure 4.—

Schematic of the GF neural circuit in Drosophila and transmission defects in the GF-TTM pathway of cpo adults. (A) The GF is a large interneuron that drives the escape jump in Drosophila in response to visual stimuli. The GF has a cell body in the brain and projects its axon down the cervical connective to the thoracic ganglion where it makes two major synaptic connections. One connection is with the tergotrochanteral motoneuron (TTMn), which in turn drives the tergotrochanteral muscle (TTM) that enables jumping. The second connection is with the peripherally synapsing interneuron (PSI), which contacts the five dorsal longitudinal motoneurons (DLMn's). These five DLMn's drive the six dorsal longitudinal muscle (DLM) fibers that are responsible for wing depression during flight. Each DLMn innervates a single DLM fiber except for DLMn5, which drives DLM fibers 5 and 6. For simplicity, DLMn's 1–3 and DLM fibers 1–3 are not included. Transmission through the monosynaptic GF-TTM pathway is fast, eliciting a response in the TTM in ∼0.8 msec. The disynaptic GF-DLM pathway produces a DLM response ∼1.2 msec following brain stimulation. In addition to producing responses with very short latencies, both pathways can follow stimuli at high frequencies. The GF-TTM pathway can follow stimulation exceeding 200 Hz, while the GF-DLM pathway exhibits following frequencies >100 Hz (Ikeda et al. 1980; Tanouye and Wyman 1980). The small “e” designations in the figure denote electrical synapses. All other synapses are chemical. The GF-TTMn synapse has been shown to have both electrical and chemical components (Blagburn et al. 1999). (B) Response to brain stimulation in a control fly (cpo^EG1/+). The DLM and TTM always fire together when the brain is stimulated at voltages above the GF threshold and always differ in latency by ∼0.4 msec (Tanouye and Wyman 1980) (calibrations, 10 mV and 1 msec). (C) Response to brain stimulation in a cpo fly (cpo^EG1/cpo^cp1). Shown are responses to three consecutive 0.2-sec pulses given at 0.5 Hz. The DLM latency never deviates while the TTM latency progressively increases before the muscle eventually fails. This indicates that the DLM response is capable of following at this frequency (0.5 Hz), but the TTM response is not (calibrations, 10 mV and 1 msec).

The GF-TTM synapse has been shown previously to be a bifunctional synapse with a component that transmits electrically via gap junction and a component that transmits via chemical transmitter (Blagburn et al. 1999). It is difficult from these experiments to determine whether it is the electrical or chemical component that carries the cpo mutant defect. However, in the GF-DLM pathway, there is a clearer separation of central synapses that transmit electrically [i.e., the synapse between the GF and the peripherally synapsing interneuron (PSI)] and chemically (i.e., the synapse between the PSI and the DLM motoneuron). The cpo mutant shows a following frequency defect in the GF-DLM pathway due to some labile site along the transmission pathway. We identified the probable location of this labile site. The defect in the GF-DLM pathway does not appear to be due to the DLM motoneuron or the DLM NMJ. This is indicated by direct electrical stimulation of the DLM motoneuron showing that the cpo mutant response is similar to wild type in latency and following frequency. The latency of the DLM response following direct motoneuron stimulation for cpo^EG1/cpo^cp1 is ∼0.6 msec, which is similar to the CS wild-type latency of 0.83 ± 0.06 msec (Tanouye and Wyman 1980). The following frequencies of the DLM response following direct motoneuron stimulation for cpo^EG1/cpo^cp1 and CS are both >100 Hz. The defect in the GF-DLM pathway does not appear to be due to GF action potential conduction, the synapse between the GF and the PSI, or PSI action potential conduction. This is inferred from comparison of the GF-DLM4 response with the GF-DLM5 response. DLM4 and DLM5 are innervated by different motoneurons that receive independent synaptic connections from a single PSI (King and Wyman 1980; Tanouye and Wyman 1980). That is, the GF pathway to DLM4 diverges from the pathway to DLM5 at the PSI to DLM motoneuron synapse (Figure 4). We assumed that if the cpo mutant labile site occurs after pathway divergence, transmission to the two DLM fibers will fail and recover independently. Conversely, we assumed that if the labile site is a circuit element common to the two muscle fibers, that is, prior to pathway divergence, transmission to the two DLM fibers would not fail or recover independently. From this, we propose that the defect in the GF-DLM pathway is distal to the PSI since the GF-DLM4 response and the GF-DLM5 response fail independently. That is, in cpo^EG1/cpo^cp2 mutants responses in both the DLM4 pathway and the DLM5 pathway fail at ∼14 ± 4.8 Hz. During the period of stimulation when failures are occurring, a trace-by-trace comparison reveals that, usually, DLM4 and DLM5 responses fail together. However, occasionally, an individual stimulus will drive a DLM4 response, but a failure in DLM5; another individual stimulus might occasionally show the opposite, a DLM4 failure coupled with a DLM5 response (Figure 5). These occasional stimuli indicate that DLM4 and DLM5 responses are capable of failing independently. A trace-by-trace analysis of the recovery period shows that the responses also recover independently. As control, we compared failure of GF-DLM5 response with the GF-DLM6 response. DLM5 and DLM6 are innervated by a single motoneuron, usually called motoneuron DLM5/6 (Harcombe and Wyman 1977; King and Wyman 1980; Tanouye and Wyman 1980). Unlike other DLM pathways, the GF-DLM5 pathway does not diverge from the GF-DLM6 pathway until the NMJ. In cpo^EG1/cpo^cp2 mutants, responses in both the DLM5 pathway and the DLM6 pathway fail at ∼14 ± 4.8 Hz. However, during the period of stimulation when failures are occurring, a trace-by-trace comparison reveals that responses in the two DLM fibers never fail independently. That is, every time a DLM5 response is seen, a response is always also seen in DLM6. Every time there is a failure of the DLM5 response, there is also a failure of the DLM6 response (Figure 5). A trace-by-trace analysis of the recovery period shows that the responses also do not recover independently. Our interpretation is that the labile site is the synapse between the PSI and the DLM5/6 motoneuron, a circuit element that is common in the pathway to the two muscle fibers. From these combined results, we infer that the labile site in the GF-DLM pathway lies distal to the PSI and proximal to the DLM motoneuron and is most likely the synapse between the PSI and the DLM motoneuron.

Figure 5.—

Transmission defects in the GF-DLM pathway of cpo adults. (A) DLM4 and DLM5 responses to brain stimulation in a cpo fly (cpo^EG1/cpo^cp1). Shown are responses to three consecutive 0.2-sec pulses administered near the following frequency threshold of 22 ± 9.2 Hz in cpo^EG1/cpo^cp1 mutants. The response of DLM4 is invariant, while the DLM5 response shifts in latency and then fails independent of DLM4. Independent failure between DLM4 and DLM5 implicates a defect distal to the PSI as the likely labile site in the GF-DLM pathway. (B) DLM5 and DLM6 responses to brain stimulation in the same cpo fly (cpo^EG1/cpo^cp1). Three consecutive 0.2-sec pulses administered near the following frequency threshold do not produce independent failure between DLM5 and DLM6, which are innervated by the same motoneuron, DLMn5. Instead both show identical shifts in latency before eventually failing together. This synchronous firing and failing between DLM5 and DLM6 tends to rule out the DLM NMJ as the labile site in the GF-DLM circuit. Thus, the labile site in the GF-DLM pathway appears to be synapses between the PSI and DLM motoneurons (calibrations, 20 mV and 1 msec).

In sum, the GF neurocircuit apparently carries an alteration in the bifunctional synapse connecting the GF with the TTM motoneuron that causes an increase in transmission latency and a severe lability with transmission failure rapidly occurring with repeated stimulation. The GF neurocircuit apparently also carries a moderate alteration in the synapse connecting the PSI with the DLM motoneuron that causes a small change in following frequency without a change in transmission latency.

Interestingly, cpo flies still retain the ability to jump when exposed to a light-off stimulus, despite having a dysfunctional GF to TTM motoneuron connection. The TTM is the primary jump muscle in the fly and is driven by the GF in response to a light-off stimulus. For a wild-type fly, a light-off stimulus elicits an escape jump. However, some mutants with defective TTM pathways, such as bendless (ben), are unable to jump in response to visual stimuli. In ben flies, the GF fails to make the terminal bend to connect to the TTM motoneuron (Thomas and Wyman 1982). The TTM in ben flies exhibits a long latency response of ∼2.3 msec and a following frequency of <1 Hz (Thomas and Wyman 1984; Oh et al. 1994). Although the cpo electrophysiological phenotype resembles ben in its TTM following frequency and latency defects, the nature of the GF to TTM motoneuron synaptic defects in cpo flies must not be severe enough to totally inhibit GF-driven jump behavior.

DISCUSSION

Discovery of a new cpo allele:

Several observations show that the cpo^EG1 mutation is a new recessive lethal allele of cpo. Genetically, the lethal cpo alleles l2 and lΔcp11 fail to complement the recessive lethality of cpo^EG1, as well as a deficiency with cpo deleted. However, the lethality of cpo^EG1 is rescued by replacement of the cpo gene region using a duplication line. Behaviorally, cpo^EG1 mutants show defects, such as overall sluggishness and lethargy, consistent with previously characterized cpo mutants when put in heteroallelic combinations with viable cpo mutants. The cpo^EG1/cpo^cp1 and cpo^EG1/cpo^cp2 flies also show BS paralysis that is shared by other heteroallelic combinations without cpo^EG1, such as cpo^cp1/cpo^cp2 and cpo^cp1/cpo^l2. Electrophysiologically, cpo^EG1 mutants manifest seizure susceptibility increases and giant fiber synaptic defects when put in heteroallelic combination with cpo^cp1 and cpo^cp2. These abnormalities are not seen in cpo^EG1, cpo^cp1, and cpo^cp2 heterozygotes. Molecularly, comparison to the Drosophila genome database of the genomic DNA flanking the cpo^EG1 insertion uniquely identifies the location of cpo^EG1 as being at 90D in an intron of cpo at a site shared by several other previously characterized alleles of cpo.

P-element insertions in cpo show complex genetic interactions:

Of the numerous P-element alleles of cpo, at least seven (EG1, cp1, cp2, l2, l3, l6, and v2) are inserted at exactly the same site in an intronic region of cpo but with differing phenotypic consequences, thought to be due to differences in P-element type, size, and orientation. The existence of multiple independent alleles resulting from insertion of a transposable element in the exact same location is not a novel phenomenon. Several examples are known, such as at the white locus, the singed locus, and the ovo locus (O'Hare and Rubin 1983; Roiha et al. 1988; Dej et al. 1998). In addition, P elements do not insert randomly, but show preference for euchromatic sites, the 5′-ends of genes, and GC rich 8-bp target sequences that provide appropriate DNA secondary structure (Liao et al. 2000). The seven P-element insertions between nucleotides 12297|12298 meet each of these criteria: they are inserted in euchromatin in the 5′ region of cpo next to an 8-bp target sequence of GTTCAGGC, which closely approximates the sequence GTCCGGAC shown to be preferred by P elements (Liao et al. 2000).

P-element insertions in cpo exhibit complex genetic interactions. Most interestingly, the P elements inserted between 12297|12298 show different phenotypes on the basis of their type and size. Previously, Bellen et al. (1992b) noted that 17-kb P[lArB] insertions (cpo^cp1 and cpo^cp2) at this site yield viable flies with adult behavioral defects. In contrast, flies with 10-kb P[lacZ, w⁺] insertions in the same location (cpo^l2, cpo^l3, and cpo^l6) behave as recessive lethals (Bellen et al. 1992b). Bellen et al. (1992a) also documented that the cpo^cp1 and cpo^cp2 alleles partially complement the lethal cpo^l2, cpo^l3, and cpo^l6 alleles producing viable heteroallelic flies with behavioral defects. Differences in phenotype due to the type of transposable element have also been documented at the Notch locus, where different types of transposon insertions within a 3-bp span result in different eye and wing phenotypes (Kidd and Young 1986). This implies that the phenotypic differences between the P[lArB] and P[lacZ, w⁺] insertions in cpo may reflect the unique properties of the P elements themselves according to mechanisms we do not fully understand yet. Interestingly, an oppositely oriented 10-kb P[lacZ, w⁺] insertion 8-bp downstream at 12305|12306 (cpo^v3) is not lethal but produces viable adults with behavioral defects similar to those produced by the P[lArB] insertions. Surprisingly, this insertion fully complements other alleles of cpo (Bellen et al. 1992a,b), including the BS paralytic behavior associated with cpo^EG1. Bellen et al. (1992a) hypothesized that this intragenic complementation may result from disrupted somatic pairing of the chromosomes in heterozygous flies, leading to complex transvection-like phenomena. Clearly, the genetic interactions of P elements at the cpo locus show an uncommon degree of complexity depending on the nature of the individual P-element alleles.

Of the alleles examined in this study, the cpo^EG1 allele appears most similar to cpo^l2 (and presumably to cpo^l3 and cpo^l6, which were unavailable for this study). Both cpo^EG1 and cpo^l2 exhibit homozygous lethality due to insertions of the same P[lacZ, w⁺] element in the same location at 12297|12298. Both also yield viable progeny with bang-sensitive paralytic defects in heteroallelic combinations with cpo^cp1 and cpo^cp2. Therefore, the cpo^EG1 mutation and the alleles examined in this study apparently form the allelic series: cpo^lΔcp11 > cpo^EG1 ≥ cpo^l2 > cpo^cp1 ≥ cpo^cp2 (see Figure 2).

cpo mutants display complex neurological deficits resembling human pathologies:

The neurological phenotypes of cpo are complex, perhaps more complex than any other behavioral mutant described for Drosophila. Previous descriptions have noted a number of behavioral phenotypes including sluggishness and abnormal phototaxis, geotaxis, flight ability, ether recovery, and mating vigor (Bellen et al. 1992a; Hall 1994). In this study, we describe in detail additional abnormalities: BS paralytic behavior, seizure sensitivity, and synaptic transmission defects in the GF circuit. The BS paralytic behavior and seizure-sensitivity phenotypes of cpo invite comparisons with other BS mutants. Unlike cpo mutants, the canonical BS mutants, such as sda, eas, and bss, usually show completely normal behavior: walking, flying, eating, mating, and grooming activities are all normal. However, in response to mechanical stimulation, the canonical BS mutants all show 100% paralysis unlike cpo mutants that show only ∼30% paralysis. Thus, the canonical BS mutants show far fewer overall abnormalities than cpo, but the paralytic defect itself is far more extreme. In addition, although canonical BS mutants show far fewer neurological syndromes and no GF synaptic defects, they are much more sensitive to seizures than cpo flies. The seizure threshold for cpo flies is 11–14 V ECS, whereas seizure thresholds for canonical BS mutants are 3–7 V ECS.

The GF system phenotypes of cpo invite comparisons with other GF system mutants. The canonical GF system mutants are ben and passover (pas). Unlike cpo mutants, the canonical GF system mutants usually show relatively normal behavior. The pas mutant shows a weak leg-shaking behavioral defect under ether anesthesia and ben shows a photochoice behavioral phenotype in choosing visible over UV light, as well as some lethargy. However, both ben and pas show an extreme GF-system behavioral defect: they cannot mount an escape jump response to a lights-off visual stimulus. Interestingly, cpo mutants do show escape jump behavior. GF system synaptic defects also appear more extreme for canonical mutants than for cpo. While cpo DLMs have only following frequency deficits, pas flies are incapable of any GF-driven DLM responses. Similar to cpo mutants, pas and ben flies have abnormal TTM responses marked by long latencies and following frequencies below 1 Hz. However, the TTM latency defect is more severe in ben flies than in cpo flies with a latency of 2.3 msec (Thomas and Wyman 1984; Oh et al. 1994). In addition, the ben GFs fail to bend to contact the TTMn in the mesothoracic neuromeres, whereas cpo GFs appear to bend normally. Interestingly, mutants of Shak-B², an allele of pas, show an extremely high seizure threshold at 94.7 ± 10.2 V ECS (Kuebler et al. 2001). In addition, the Shak-B² mutation is a potent suppressor of seizures. Both of these characteristics set apart Shak-B² flies as distinct from cpo mutants, which have lowered seizure thresholds and behave as enhancers of seizure. Thus, the canonical GF system mutants show far fewer overall abnormalities than cpo, but the GF system-specific defects are more extreme.

Although the complex neurological phenotype of cpo mutants makes it difficult to assign them to a particular mutant class, it also makes them a more realistic model of neurological disorders in humans and mice, which tend to be more pleiotropic. In humans, epilepsy is often one of a set of symptoms found in people with diseases such as cerebral palsy, neurofibromatosis, autism, tuberous sclerosis, and Landau-Kleffner syndrome. In each of these human diseases, seizures are one of a number of other neurological disorders composing the disease (Frazin 2001). Numerous mouse models of epilepsy also exhibit additional neurological conditions. Some examples include lethargic and quaking mice. The lethargic mouse exhibits absence epilepsy accompanied by ataxia, hypoactivity, paroxysmal dyskinesis, and reduced conduction velocity and prolonged distal latency in peripheral motor nerves (Herring et al. 1981; Khan and Jinnah 2002). The quaking mouse exhibits tonic-clonic seizures accompanied by generalized tremors and hypomyelination defects (Hogan and Greenfield 1984). Interestingly, the deficits in quaking mice result from a defective RNA-binding protein that regulates nuclear export of myelin basic protein mRNA (Larocque et al. 2002). Thus, the defects in cpo mutants seem to approximate pathologies common in more complex organisms.

A role for cpo in regulating nervous system function:

Several possibilities exist for how cpo regulates proper nervous system function and seizure susceptibility. Because of its localization to the nucleus, the interaction of cpo with RNA may involve participation in splicing or export of transcripts out of the nucleus. Possibly cpo regulates proper nervous system function by altering the balance of transcripts involved with neuronal excitation and inhibition via one of these processes. Several animal models with defects in nuclear export of transcripts and splicing display neuronal excitability defects. Some examples include the previously mentioned quaking mice, which exhibit defects resulting from a mutation affecting the RNA-binding protein QKI, which regulates nuclear export of myelin basic protein mRNA (Hogan and Greenfield 1984; Larocque et al. 2002). Another example is seen in Nova-1 mice, which display progressive motor dysfunction, marked by action-induced tremor and overt motor weakness, culminating in death 7–10 days after birth (Jensen et al. 2000a,b). These defects in Nova-1 mice result from deletion of the RNA-binding protein, Nova-1, which regulates alternative splicing of neural-specific pre-mRNA's, such as the inhibitory GABA_A receptor γ2 pre-mRNA (Dredge and Darnell 2003). The NOVA family of RNA-binding proteins was first identified as the target antigens in the human disease paraneoplastic opsoclonus myoclonus ataxia, a neurological disorder characterized by loss of inhibitory control of motor neurons in the spinal cord and brainstem (Musunuru and Darnell 2001). The examples of cpo, quaking, and Nova-1 show that mutation of RNA-binding proteins can have serious consequences for nervous system function and behavior.

Although the exact mechanism underlying the seizure susceptibility of cpo flies is unknown, some possible explanations for the altered seizure threshold in cpo mutants seem unlikely. Because of the unchanged GF threshold of cpo flies and their overall sluggish demeanor, increased overall neuronal excitability as an explanation for the seizure sensitivity of cpo mutants appears as an untenable hypothesis. One explanation for a decreased seizure threshold could be that the neurons in cpo mutants are hyperexcitable with a lowered stimulation threshold, which would allow for abnormal supernumerary recruitment of neurons upon administration of an electrical buzz, facilitating the temporal and spatial summation known to occur in seizures. At least in the case of the GF interneuron, a decreased neuronal stimulation threshold is not observed in cpo flies. However, this does not exclude the possibility that other unexamined neurons critical for seizure genesis in the cpo nervous system are hyperexcitable with decreased thresholds for stimulation. It should also be noted that seizures are not GF driven since they can never be elicited at the GF threshold voltage level. Thus, recruitment of other unidentified neurons is required for seizure manifestation. In addition, cpo flies do not seem to be good candidates for exhibiting increased temporal summation as is characteristic of seizures. Their severely decreased following frequencies would be expected to hamper the ability of their nervous system to temporally summate neuronal responses because they occur with lesser frequency. Because increased excitatory character does not appear responsible for the increased seizure susceptibility of cpo flies, maybe they exhibit unidentified deficits in inhibition, which could hinder their nervous system's ability to resist seizure following insult.