Modeling Glial Contributions to Seizures and Epileptogenesis: Cation-Chloride Cotransporters in Drosophila melanogaster

Flies carrying a kcc loss-of-function mutation are more seizure-susceptible than wild-type flies. The kcc gene is the highly conserved Drosophila melanogaster ortholog of K+/Cl− cotransporter genes thought to be expressed in all animal cell types. Here, we examined the spatial and temporal requirements for kcc loss-of-function to modify seizure-susceptibility in flies. Targeted RNA interference (RNAi) of kcc in various sets of neurons was sufficient to induce severe seizure-sensitivity. Interestingly, kcc RNAi in glia was particularly effective in causing seizure-sensitivity. Knockdown of kcc in glia or neurons during development caused a reduction in seizure induction threshold, cell swelling, and brain volume increase in 24–48 hour old adult flies. Third instar larval peripheral nerves were enlarged when kcc RNAi was expressed in neurons or glia. Results suggest that a threshold of K+/Cl− cotransport dysfunction in the nervous system during development is an important determinant of seizure-susceptibility in Drosophila. The findings presented are the first attributing a causative role for glial cation-chloride cotransporters in seizures and epileptogenesis. The importance of elucidating glial cell contributions to seizure disorders and the utility of Drosophila models is discussed.


Introduction
Glial cells are proposed to be important players in seizure disorders because of their critical role in maintaining extracellular ionic homeostasis in the nervous system [1][2][3][4][5]. Glial contributions to seizures, however, have not been studied as well as neuronal and synaptic signaling mechanisms. In this study, we examined cation-chloride cotransporter (CCC) function separately in neurons and in glia, and showed that a loss of CCC function in either cell class causes seizure-sensitivity in Drosophila.
CCCs are involved in a variety of physiological mechanisms in humans and experimental models [10], [11]. CCCs play cell-type specific roles in the control of several fundamental processes including cell volume homeostasis, cell migration, neural circuit development and neuronal excitability [12], [13]. For example, CCCs regulate cell volume with cell swelling countered by an efflux of Cl 2 , K + , and water, mediated by KCC3 and ion channels [14][15][16]. Cell shrinkage is countered by an influx of Cl 2 , Na + , K + and water, mediated by NKCC1 and ion exchangers [17], [18]. All cells tightly regulate their size and shape because even minor changes can be catastrophic to the myriad of cell and organ functions [19][20][21].
In the present experiments, CCC loss-of-function flies were generated by ectopic spatial and temporal restriction of RNA interference (RNAi) transgene expression. Seizure-sensitivity was observed in animals with kcc loss-of-function in either glia or neurons. ncc69 loss-of-function in glia, but not neurons, caused seizure-sensitivity. These neurological phenotypes correlated with cell and tissue volume abnormalities in larvae and adults.

Fly Strains
The genotypes of the Drosophila strains used in this study and their sources are listed in Table S1. Flies were maintained on standard cornmeal-molasses medium at 25uC in a humidified incubator unless otherwise indicated. The kcc DHS1 mutation is a 13 base pair insertion in intron 11 of kcc (kazachoc). The mutation leads to an approximate two-fold reduction in kcc transcript as determined previously by RT-PCR; and an approximate four-fold reduction in Kcc protein as determined by Western blot analysis [24]. Genetically, kcc DHS1 behaves as a hypomorph [24]. The kcc EY08304 embryonic lethal mutation is a P-element insertion at 2R:19,812,398, upstream of the sequence encoding the Cterminus tail predicted to be recognized by the Kcc antibody used in this study (Fig. S1). The kcc Ad4 embryonic lethal mutation is not molecularly mapped, but it fails to complement kcc EY08304 and is apparently a null mutation (see [24] and Fig. S1). Genotypes were constructed by standard Drosophila crossing schemes using common strains with balancer chromosomes as needed.

Behavioral Assays
Behavioral seizure-like activity, or bang-sensitive (BS) paralysis, was quantified as described previously [29]. Briefly, flies were collected ,1 day post-eclosion then placed in a 25uC humidified incubator to recover from CO 2 -induced anesthesia until tested for BS paralysis the following day. For scoring BS paralysis, 15 or fewer individual flies were tested per vial. Mechanical stimulations, or ''bangs'', were performed on vials using a VWR vortexer set to maximum speed for 10 s. Pools of data were then combined for each genotype (n.100 flies for each) to yield a final measure of %BS paralysis at 24-48 h-old. Movies of BS paralysis were acquired using a mobile cellular device and edited in Windows Live Movie Maker. Flies kept at 30uC for Temporal and Regional Gene Expression Targeting [30] experiments were moved to fresh vials after each BS paralysis test due to deterioration of vial conditions. Electrophysiology in vivo electrophysiological assays were from the adult giant fiber system [31]. Seizure threshold tests were described previously [29], [32][33][34]. For mounting, the fly was suctioned onto a hypodermic needle attached to a vacuum line, then stabilized in dental wax on a glass microscope slide. Stimulating and ground electrodes were uninsulated tungsten. Recording electrodes were glass micropipettes filled with 3 M KCl. Seizure-like activity was evoked by high-frequency electrical brain stimulation (0.5 ms pulses at 200 Hz for 300 ms) of varying voltage and monitored by dorsal longitudinal muscle recordings as described previously [29], [33], [34]. Flies were given a maximum of four high-frequency stimuli total, with 7-10 minutes of rest between each stimulus before being discarded. The giant fiber circuit was monitored continuously as a proxy for holobrain function. Evoked seizures were identified by the occurrence of aberrant high-frequency dorsal longitudinal muscle activity and subsequent failure of giant fiber stimulation to elicit a muscle response, indicating chemical synapse failure [29]. Grass 44S stimulus duration was controlled by a Molecular Devices Digidata 1440a commanded by custom protocols implemented in Strathclyde Electrophysiology Software, WinWCP V4.5.8. Pulse width, stimulation frequency, and voltage settings on the stimulator were set manually. Recordings were led to a WPI Electro 705 preamplifier and digitized using the Digidata 1440a with 50 Hz sampling. n.10 animals recorded for each genotype tested.

Histology
Whole mount Drosophila embryos were stained following Bossing's whole-mount protocol (Abcam). Third instar larvae were filleted in hemolymph-like solution [35] then stained according to standard procedures. Immuno-staining of 24-48 hold adult brains was performed as described previously [36]. Kcc antibody is rabbit polyclonal anti-Kcc (Pacific Immunology, Ramona, CA). The antibody is predicted to specifically recognize all seven Drosophila melanogaster Kcc isoforms (Kcc-A, -B, -C, -D, -E, -F, and -G) at their identical C-termini (RGGGREVITIYS) (BLAST; National Institutes of Health). Control experiments with kcc deletion mutations in embryos show antibody specificity to Kcc protein (Fig. S1) consistent with previous Western blot analysis [24]. Other primary antibodies used were: Ab13970 chicken

Imaging and Quantification
Images were acquired using the following confocal microscopes and software: Zeiss LSM 5-Live with LSM 5 software, and Zeiss LSM 780 with ZEN software. Cross-channel bleed-through for each tissue type was negligible as determined by single antibody staining and ZEN software linear unmixing (data not shown). Fluorescent micrographs were processed using ImageJ software (National Institute of Health). Embryo Kcc fluorescence intensities were averages of internal regions defined by perimeter signal for each genotype, after background subtraction. Calyx Kcc fluorescence intensities were averages of regions of interest defined by the GFP+ channel for each genotype in three slices (anterior, center, and posterior of calyx), after background subtraction. Mean Kcc fluorescence intensities of embryos and adults were normalized to control genotype intensities. Adult brain size was approximated as the volume of an ellipsoid with semi-axes along the dorsal-ventral axis (top to center of brain along the midline in the central optical slice), the anterior-posterior axis (half of encompassing Z-stack size), and the lateral-medial axis (lateral border of optic lobe to the midline in the central optical slice). Volume estimation of neuron somata (assumed spherical) was achieved by measuring the longest observable diameter for each cell. Quantification of larval peripheral nerve average cross-sectional area was performed by averaging fourteen evenly-distributed diameter measurements at the center-most GFP channel optical slice along 500 mm of each nerve beginning below the ventral nerve cord (assuming each nerve was a cylinder). Peripheral nerve sections with glial cell nuclei were excluded from measurements. Statistical tests were unpaired two-sample Student's t-tests. Confocal slice and Z-stack projection sizes were noted in figure legends. Image files were imported into Adobe Illustrator CS6 for figure composition.

Kcc found in glial and neuronal cell membranes
Kcc protein was shown previously to be expressed throughout the adult brain of wild-type Drosophila with Western blot analysis showing a loss of immunoreactivity in kcc mutants [24]. In kcc DHS1 , a loss of Kcc protein causes seizure-sensitivity [24]; kcc DHS1 seizuresensitivity can be rescued by supplying Kcc back to neurons or back to glia [25]. Here, we extend these observations of Kcc immunoreactivity with a Kcc-specific antibody ( Figure S1) and confocal fluorescence microscopy to show expression is also present in the larval nervous system (Fig. 1). We find that both larval and adult nervous system staining shows Kcc in glial cells in addition to neurons.
In third instar larvae, prominent Kcc staining is present surrounding the ventral nerve cord and peripheral nerves, as well as in neuronal cell bodies of the ventral nerve cord (Fig. 1A). Highmagnification images of peripheral nerve displayed robust GFP signal in sensory and motor neuron axons of animals expressing membrane-bound GFP in all neurons (Fig. 1B1). Kcc protein is present throughout the nerve (Fig. 1B2), and colocalizes with neuronal GFP (Fig. 1B3). Strong Kcc immunoreactivity was present ensheathing the peripheral nerve that appeared to terminate before sensory neuron somata (data not shown). Taken together, our interpretation is that most of the Kcc at the periphery is in subperineurial glia (SPG) cells, surface blood-brain barrier glia which are thought to ensheath peripheral nerves and terminate before the somatic and synaptic regions. Peripheral Kcc may in part correspond to that of a second type of surface glia, the perineurial glia (PG), ensheathing SPG cells [35].
In the adult brain, Kcc immunoreactivity was present in glial, as well as neuronal cell types. For example, Fig. 1C1-1C5 depicts the Figure 1. Immunohistochemistry with confocal fluorescence microscopy reveals neuronal and glial Kcc in larvae and adults. (A) Larval nervous system Kcc in a 0.75 mm confocal slice is seen surrounding the ventral nerve cord (VNC) and peripheral nerves (PNs), and colocalizing with neuronal somata on the periphery of the VNC. CNS: central nervous system. High-magnification PN in a 2 mm slice reveals membrane-bound GFP of neuronal processes (B1) and Kcc (B2) surrounding and colocalizing with the processes (B3). Adult posterior brain Kcc in a 0.5 mm slice (C3) depicts the general localization pattern observed in several regions; Kcc is widespread and salient around cell somata (C4) and on brain surface/ edges (C5). Neuronal GFP in mushroom body somata and calyx membranes (C2) colocalizes with Kcc (C1), confirming that Kcc is in central brain neuropile and ubiquitous. Cyx: mushroom body calyx. Optic lobe Kcc in 0.5 mm slice (D3) conformed to the central brain pattern of neuronal membrane (D2) colocalization (D1). Optic medulla NC82 (D4), a neuropile marker, colocalized with Kcc (D5). Posterior brain Kcc in a 0.5 mm slice (E3) in animals expressing glial membrane-bound GFP (E2) colocalizes with GFP significantly (E1), probably in cortex (arrow heads) and surface (arrows) glia. Neuronal membrane (E4) colocalization with Kcc (E5) in the same slice is shown for comparison. High magnification micrographs of the optic lamina showed that Kcc in a 0.5 mm slice (F3) is present in glial (F2) and neuronal (F4) membranes of this brain structure (F1), (  1C1; mushroom body calyx, Cyx). A layer of non-neuronal Kcc expression appeared to be surrounding the mushroom body cell bodies ( Fig. 1C1; arrow). We interpret this Kcc expression as belonging to PG and/or SPG cells, ensheathing the central nervous system [36]. This general arrangement of staining was found in many brain regions: colocalization of neuronal and Kcc staining in neuronal cell bodies and neuropile with non-neuronal Kcc expression ensheathing cell body regions (see below and data not shown). DIC image of the same optical slice showed light granular regions of cell bodies, and dark regions of neuropile (Fig. 1C4). Colocalization of cell bodies was often associated with the undulated Kcc patterns found in adult central brain (Fig. 1C5). Similarly, in the optic medulla, membrane-bound GFP was expressed in a subset of neurons ( Fig. 1D2) and Kcc signal in the same optical slice (Fig. 1D3) colocalized with GFP (Fig. 1D1). Neuropile staining in the same medulla region (Fig. 1D4) showed robust colocalization with Kcc (Fig. 1D5). Kcc interlacing the medulla neuropile and surrounding cell bodies was observed as well (data not shown), most consistent with expression in glial membranes [37].
To explore nervous system Kcc expression further, specimens were prepared with triple labeling of glial cell membrane, neuronal membrane, and Kcc protein (Fig. 1E1-1F5). Imaging of the posterior central brain, for example, further confirmed the presence of Kcc in glial cells, apparently in surface glia and cortex glia (arrows and arrowheads, respectively, Fig. 1E1-1E3). In some instances, we were also able to infer likely Kcc-positive glial subtypes by manually tracing structures in GFP/HRP and Kcc/HRP merge Z-stacks, and basing identification on stereotyped anatomical relationships between glial subtypes and neurons ( [36] and Fig. 1E4,1E5). High-magnification images of the optic lamina exposed a non-uniform distribution of Kcc in glia and neurons ( Fig. 1F1-1F5). Neuronal processes contained Kcc foci (Fig. 1F5), while glial Kcc revealed in this region appear to be astrocyte-like glial processes mingling with synapses (Fig. 1F1). In summary, brain Kcc was membrane bound and found in both neurons and glia. Neuronal cell bodies, cortex glia, and surface glia all appear to have prominent Kcc immunoreactivity.

Neuronal expression of kcc RNAi transgenes causes behavioral seizure-like activity
Loss of Drosophila K + /Cl 2 cotransporter function, such as in the kcc DHS1 hypomorphic mutant, raises seizure-susceptibility through a reduction in kcc expression. The kcc DHS1 mutant displays seizuresensitivity as evidenced by a lowered threshold to evoked electrophysiologically recorded seizure-like activity, and bangsensitive behavioral seizure-like activity and paralysis (BS paralysis) [24], [25]. Here, we dissected spatial aspects of induced seizure- sensitivity using different GAL4 drivers and UAS-kcc-RNAi transgenes to generate kcc loss-of-function in subsets of cells [38]. Briefly, flies with promoter-determined GAL4 transcription factor expression (driver) were crossed to flies carrying a UAS-kcc-RNAi transgene (kcc RNAi under control of Upstream Activation Sequence; UAS) to drive promoter-determined kcc knockdown in F1 progeny (i.e. promoter-GAL4.UAS-kcc-RNAi progeny). We tested a total of 64 different GAL4 drivers in combination with two different kcc RNAi transgenes for BS paralysis. We found that this was a robust method of inducing seizure-sensitivity, particularly using UAS-kcc-RNAi-B (Bloomington stock center #34584, referred to hereafter as kcc-RNAi-B), which appeared to be more effective than UAS-kcc-RNAi-V (Vienna Drosophila RNAi Center #101742, referred to hereafter as kcc-RNAi-V) ( Table  S2). Twenty-eight of the 64 GAL4 drivers tested induced lethal and/or BS paralysis phenotypes. The other 36 GAL4 drivers gave rise to viable flies without BS paralysis phenotypes.
Several GAL4/UAS combinations resulted in lethal phenotypes, especially using the ubiquitous driver Act5C and the panneuronal driver elav c155 with either kcc-RNAi transgene ( Fig. 2A). Third instar larvae with kcc-RNAi-V driven by elav c155 and several other GAL4/UAS genotypes often burrowed to the bottom of their vials before perishing (data not shown). Cholinergic neuron drivers caused semi-lethality and 100% BS paralysis in escapers with kcc-RNAi-V, and produced late pupal lethality with kcc-RNAi-B (Table S2). In addition, kcc-RNAi-V flies driven with cholinergic neuron drivers are sterile and have a completely penetrant juvenile-wing phenotype identical to that of flies with perturbed CCAP neuropeptide-producing/-receptive neurons (data not shown). Taken together, these observations demonstrate further the criticality of kcc functions throughout the Drosophila nervous system.
For several neuronal GAL4 drivers, we consistently observed interesting, albeit somewhat surprising results comparing kcc-RNAi-V and kcc-RNAi-B. These GAL4 drivers were: the ellipsoid body driver, c507, the motor neuron driver, OK6, and the mushroom body drivers, 201Y, MB247, OK107, and c772: for each, the weaker kcc-RNAi-V is largely ineffective in inducing BS paralytic behavior. In contrast, in combination with the stronger kcc-RNAi-B, the mushroom body drivers induce complete (100%) or nearly complete (.90%) BS paralysis, and c507 and OK6 drivers cause lethality. These data suggest that a threshold of kcc RNAi in a particular neuronal population must be reached in order for phenotypes to manifest. Despite this unusual observation between the two RNAi transgenes, the mushroom body GAL4 drivers did not appear to be markedly more effective at inducing BS paralysis phenotypes in these experiments than any other neural drivers.
Examination of 49 GAL4/UAS combinations using mushroom body and other neuronal GAL4 drivers showed no obvious spatial tendencies in the induction of seizure-sensitivity by kcc RNAi. Drivers expressing in all or many neurons induced lethality and/or extreme BS paralysis ( Fig. 2A). Drivers thought to be expressed in small sets of neurons, including GABAergic, U/CQ, and Pdfexpressing neurons, did not elicit BS paralysis in combination with either kcc-RNAi-V or kcc-RNAi-B transgene (Table S2). Nonnervous system drivers, including those with robust hemocyte, salivary gland, fat body, muscle, tracheal, and adipokinetic hormone-secreting cell expression, always generated viable flies without BS paralysis phenotypes (Table S2). Cumulatively, it appeared that phenotypes were mainly associated with the number of neurons deficient in kcc function, and the magnitude of this deficiency.

BS paralysis caused by glial expression of kcc RNAi
Experiments using glial-specific GAL4 drivers indicate directly that loss of kcc function in this tissue causes seizure-sensitivity. repo-GAL4 drives expression in nearly all glia. When used to drive kcc-RNAi transgenes, we found that kcc-RNAi-V and kcc-RNAi-B flies showed substantial BS paralytic behavior: 70% and 85% BS paralysis, respectively ( Fig. 2A and Movie S1). These were surprising findings, particularly since expressing kcc + in glial cells was not remarkably effective in rescuing kcc DHS1 mutant BS phenotypes, although some rescue was observed [25].
Several glial subtypes are present in Drosophila [36], [37], [39], [40]. Particularly interesting are the SPG cells comprising the main blood-brain barrier cell type: pericellular diffusion being limited by septate junctions formed between the SPGs. GAL4 drivers apparently specific for cells that express septate junction genes, including SPG cells, were particularly effective at causing BS paralytic behavioral phenotypes in combination with kcc RNAi. Thus, Gli-GAL4 and Moody-GAL4 each driving kcc-RNAi-B in SPG gave rise to 100% and 65% BS paralysis, respectively ( Fig. 2A and Movie S2). Additional phenotypes were observed in the Gli-GAL4.kcc-RNAi-B flies. Flies often displayed malformed wings (data not shown) and were especially sensitive to behavioral seizure-like activity. Gentle handling of the vial was sufficient to trigger seizures; this appears to be the most severely seizuresensitive genotype that we have observed for Drosophila to date. Substantial (90%) BS paralysis was also observed with NP2222-GAL4 driving kcc-RNAi-B ( Fig. 2A). NP2222-GAL4 is reported to drive expression in cortex glia surrounding neuronal cell bodies [41]. We also observed BS paralysis in flies with kcc-RNAi-B driven with nrv2-GAL4 (cortex and neuropile glia, 88% BS paralysis), mz709-GAL4 (ensheathing glia, 60% BS paralysis) and alrm-GAL4 (astrocyte-like glia, 22% BS paralysis) ( Fig. 2A). Taken together, these results suggested that proper kcc function is required in glial cells, and that loss-of-function causes seizuresensitivity. Further, within the limitations of GAL4 glial driver specificity, several different glial subtypes appeared to contribute to the seizure-sensitive phenotype due to kcc loss-of-function.

Electrophysiologically recorded seizure-like activity with kcc RNAi
The BS paralysis phenotype is an indicator of seizure-sensitivity: flies that show a strong BS behavioral phenotype also display a reduced threshold to evoked seizure-like neuronal activity/ sustained discharges in electrophysiological tests (Fig. 2B), whereas mutations that decrease behavioral bang-sensitivity raise the threshold [29], [42]. Examination of kcc RNAi-induced seizuresusceptibility showed that the threshold to evoked seizure-like discharges of experimental A307-GAL4 (driving in a subset of neurons [43], [44]) flies (genotype: A307-GAL4.kcc-RNAi-V) was about one-fourth that of wild-type and control genotypes (Fig. 2C). Thus, A307-GAL4.kcc-RNAi-V flies showed a low threshold of 10.762.1 volts high-frequency stimulus (V HFS; see Materials and Methods) whereas kcc-RNAi-V control flies had a threshold of 37.862.7 V HFS, comparable with wild-type Canton-S flies (40.762.7 V HFS). For comparison, the threshold of A307-GAL4.kcc-RNAi-V flies was similar to that of kcc DHS1 mutant flies (13.164.67 V HFS, [24]).
Reducing kcc expression selectively in glial cells also generated flies with low thresholds for evoked seizure-like discharges (Fig. 2C), consistent with the notion that BS paralysis is a good indicator of seizure-sensitivity. Thus, repo-GAL4.kcc-RNAi-B flies (expressing RNAi in nearly all glia) had a low threshold of 17.561.5 V HFS. Thresholds were even lower for flies of the genotype Gli-GAL4. kcc-RNAi-B (9.961.9 V HFS) expressing RNAi in the SPG subset of glia. Representative traces of seizure-like discharges recorded from dorsal longitudinal muscles of the aforementioned genotypes are presented in Fig. 2D. These electrophysiological results confirmed behavioral observations indicating that reducing expression of kcc by RNAi in either nerve cells or glial cells increases seizure-sensitivity.
We frequently observed inner brain confocal image regions that appeared to be devoid of signal (GFP, Kcc or HRP) in genotypes with kcc RNAi expression. For example, the repo-GAL4.kcc-RNAi-B,mCD8::GFP (Fig. 3F1,3F5), and the OK107-GAL4.kcc-RNAi-B,mCD8::GFP (Fig. 3H3) calyces are surrounded by signal-devoid regions, compared to those of control genotypes (Fig. 3E1,3E5,3G3). Kcc fluorescence intensity of the OK107-GAL4.kcc-RNAi-B,mCD8::GFP calyx is significantly lower than that of the control calyx (Fig. 3I), however, the Kcc intensity of signal-devoid regions are even lower and comparable to intensities measured outside of the brain (data not shown).
The somata and processes of neurons expressing kcc-RNAi had relatively normal anatomical positions, but morphological differences were observed. Some identified cells, such as giant fiber and lateral pace-making neurons, were enlarged when expressing kcc RNAi transgenes (Fig. 4A2,4B2). Lateral pace-making neurons expressing kcc-RNAi-B had an average soma volume of 58.065.1 mm 3 compared to 18.561.8 mm 3 of controls (Fig. 4C). Diffuse and elongated dendrites lacking distinctive spines were also evident in enlarged giant fiber neurons expressing kcc-RNAi-V ( Fig. 4A2; insets) compared to wild-type giant fiber neurons ( Fig. 4A1; insets).
Knockdown of kcc in neurons also caused a nerve swelling phenotype, as seen in elav C155 .kcc-RNAi-V,mCD8::GFP larvae (expressing kcc RNAi and membrane-bound GFP in all neurons), albeit less severe than that of glial kcc knockdown (Fig. 5E1-5E5). Peripheral nerve size was significantly altered when compared to control elav C155 .mCD8::GFP larvae (Fig. 5D1-5D5,5G). Thus, the larval CCC loss-of-function results bolstered the connection between volume regulation abnormalities and seizure-susceptibil- ity; kcc or ncc69 knockdown in glia, and kcc knockdown in neurons, correlated with larval volume abnormalities, adult volume abnormalities, and adult seizure-sensitivity.

Discussion
In Drosophila, a combination of advanced genetic, behavior, and electrophysiology methods facilitate investigations of neurological dysfunction, especially seizure disorders. We showed here that RNAi, particularly kcc RNAi, is an especially effective approach for generating seizure-like phenotypes in flies. This, together with spatial and temporal control of expression, enables incisive investigations into possible mechanisms underlying seizure-susceptibility to be conducted. We found, surprisingly, that targeting kcc knockdown separately to either neurons or to glia induced seizure-sensitivity. Either kcc loss-of-function condition caused flies to exhibit robust seizure-like behaviors and electrophysiologically recorded seizure-like discharges. Immunohistochemistry showed Kcc was present in both neurons and glia at different stages of development, confirming the expectation that expression is present in both of these cell types.
Seizure-sensitivity in kcc DHS1 mutants was previously shown to be mediated by Drosophila GABA A receptors, thereby linking kcc loss-of-function seizures with dysfunction of the GABAergic inhibitory system [24], [25]. This suggested further that seizuresensitivity via kcc dysfunction is caused by, or exacerbated by, excitatory GABA signaling due to neuronal intracellular Clmisregulation. This interpretation is similar to the mechanism thought to be responsible for seizures due to reduced KCC2 function in mice and humans [46][47][48][49].
Excitatory GABA signaling is an unlikely mechanism for seizure-sensitivity following kcc knockdown in glia. The similarity of seizure-sensitive phenotypes due to both kcc knockdown and ncc69 knockdown suggests the possibility of a common mechanism, as these two transporters are closely associated in many tissues, albeit generally with reciprocal functions in terms of Cl 2 flux [12]. Both kcc and ncc69 knockdown in glia also caused larval nerve swelling [27]. There are, therefore, several salient questions related to these transporters: 1) how do Kcc and Ncc69 function in glia? 2) how does CCC loss in glia cause phenotypes of neuronal dysfunction? and 3) is there a causal relationship between nervous system phenotypes, for example, are the structural phenotypes of swelling responsible for the seizure-sensitivity phenotypes? We attempted to partially answer these questions by assessing the developmental requirement for kcc RNAi-related seizures using the heat-shock GAL4 and TARGET temporally-controlled expression systems [30]. Behavioral seizure-like activity was never observed in adult flies undergoing several kcc RNAi induction paradigms spanning multiple days, having developed as wild-type (data not shown). Thus, the questions related to CCCs in glia require more investigation to be answered.
Glial contributions to seizure-susceptibility in humans and mammalian models are not well-studied, but are believed to arise primarily from a dysfunction in ionic homeostasis, mainly a failure of glial cells to adequately buffer extracellular K + (reviewed in [2]). This is a particular problem in post-traumatic models of epilepsy highlighted by studies showing that acute osmotic disruption of the blood-brain barrier causes seizures in human patients and porcine models [50]. Volume changes to neurons are also known to increase seizure-sensitivity and increase the severity of epileptiform activity (reviewed in [21]). Understanding the ways in which CCCs are involved in epilepsies may aid in the design of less toxic cures for seizures related to CCC dysfunction. The Drosophila experimental model for CCC loss-of-function studies can provide unique insight into glial CCC mechanisms. Although biochemical properties and loss-of-function phenotypes of glial CCCs have been explored previously in vitro, we are unaware of other studies exclusively targeting glial CCC functions in vivo besides the work presented here and that of Leiserson et al. in Drosophila [16], [17], [26], [51], [52]. The conservation of glial functions and the currently unmatched genetic tools for manipulating them in Drosophila make the fly an indispensable model for elucidating glial seizure mechanisms [41], [53]. Figure S1 Rabbit polyclonal anti-Kcc is largely Kccspecific. Representative images of 20-22 h-old F1 embryos of self-crossed kcc Ad4 /Cyo,Act-GFP flies (with ubiquitous cytoplasmic GFP expression in all cells) stained with anti-GFP (A1) and anti-Kcc (A2). Faint auto-fluorescence in GFP-negative embryos arising from developing gut can be seen in the GFP channel confirming that they, and other similarly scored specimens, were not unfertilized eggs. GFP-negative embryos were presumably kcc Ad4 /kcc Ad4 . Marginal Kcc signal was detected surrounding all embryos of any genotype suggesting a negligible degree of nonspecific staining or presence of persistent maternally-deposited Kcc. (A3) GFP-negative embryos were almost always Kcc-negative (n = 42 out of 51). Thus, kccAd4 appears to be a null mutation. Representative images of 20-22 h-old F1 embryos of self-crossed kcc EY08304 /Cyo,Act-GFP flies stained with anti-GFP (B1) and anti-Kcc (B2). GFP-negative embryos are presumably kcc EY08304 / kcc EY08304 . (B3) GFP-negative embryos were almost always Kccnegative (n = 28 out of 33). Thus, kcc EY08304 appears to be a null mutation. (C) Quantification of highly significant Kcc loss in kcc Ad4 /kcc Ad4 mutant embryos compared to controls. (D) Quantification of highly significant Kcc loss in kcc EY08304 /kcc EY08304 mutant embryos compared to controls. Error bars are S.E.M. and significance for Student's t-tests is: *** = p,0.001. Scale bars are in microns. (TIF) Table S1 The genotypes and sources of the Drosophila strains used in this study.