prickle modulates microtubule polarity and axonaltransport to ameliorate seizures in fliesSalleh N. Ehaideba,b,1, Atulya Iyengarc,1, Atsushi Uedad, Gary J. Iacobuccie, Cathryn Cranstonf, Alexander G. Bassukg,David Gubbh, Jeffrey D. Axelrodi, Shermali Gunawardenae, Chun-Fang Wua,c,d, and J. Robert Manaka,d,g,2

Interdisciplinary Graduate Programs in aGenetics, cNeuroscience, and fMolecular and Cellular Biology and Departments of dBiology and gPediatrics, Universityof Iowa, Iowa City, IA 52242; bKing Abdullah International Medical Research Center, King Abdulaziz Medical City, Riyadh 11426, Kingdom of Saudi Arabia;eDepartment of Biological Sciences, University at Buffalo, The State University of New York, Buffalo, NY 14260; hReponse Immunitaire et Developpment,Centre National de la Recherche Scientifique, 67084 Strasbourg Cedex, France; and iDepartment of Pathology, Stanford University School of Medicine,Stanford, CA 94305

Edited by Barry Ganetzky, University of Wisconsin–Madison, Madison, WI, and approved June 23, 2014 (received for review March 18, 2014)

Recent analyses in flies, mice, zebrafish, and humans showed thatmutations in prickle orthologs result in epileptic phenotypes, al-though the mechanism responsible for generating the seizureswas unknown. Here, we show that Prickle organizes microtubulepolarity and affects their growth dynamics in axons of Drosophilaneurons, which in turn influences both anterograde and retro-grade vesicle transport. We also show that enhancement of theanterograde transport mechanism is the cause of the seizure phe-notype in flies, which can be suppressed by reducing the level ofeither of two Kinesin motor proteins responsible for anterogradevesicle transport. Additionally, we show that seizure-prone pricklemutant flies have electrophysiological defects similar to other flymutants used to study seizures, and that merely altering the bal-ance of the two adult prickle isoforms in neurons can predisposeflies to seizures. These data reveal a previously unidentified path-way in the pathophysiology of seizure disorders and provide ev-idence for a more generalized cellular mechanism whereby Pricklemediates polarity by influencing microtubule-mediated transport.

epilepsy | planar cell polarity | spiny-legs | EB1-GFP | neurodegeneration

Epilepsy, which affects ∼1% of the population, is a disablingand debilitating disease characterized by recurrent seizures.

Coupling genetic analysis of human epilepsy cases with func-tional validation in animal models, we previously showed thatmutations in human PRICKLE1 and PRICKLE2, as well asmutations in mouse, zebrafish and fly prickle (pk) orthologs, in-crease susceptibility to seizures (1–3). We recently showed thatboth mouse and fly Prickle can associate with Synapsin (4).However, little is known regarding what cellular process con-nects Prickle with epilepsy. Products of the prickle gene work inconcert with a group of cytoplasmic and membrane-associatedproteins including Frizzled (Fz) and Dishevelled (Dsh) to es-tablish polarity of structures such as hairs and bristles in the flyepidermis (5–8). The fly pk locus expresses three alternatelyspliced isoforms, two of which, pksple (prickle-spiny-legs) andpkpk (prickle-prickle), are expressed in postembryonic animals(5, 7). Homozygous pkpk and pksple fly mutants show strongplanar cell polarity (PCP) phenotypes in the wing and in the legs,respectively, in addition to other regions of the adult body (6, 7).Prickle (Pk) protein isoforms are localized to the proximal end offly wing cells, whereas Fz and Dsh are localized distally (8).Here, we show that vesicle transport dynamics are altered in

neurons of Drosophila mutant for either adult isoform of prickle.Seizure-prone pksple heterozygotes show enhanced vesicle trans-port (along with electrophysiological defects similar to otherseizure-prone mutant flies), and the seizure phenotypes can berescued by lowering the dose of vesicle transport motor proteins.However, pkpk heterozygotes show severely reduced vesicletransport, with loss of both copies of pkpk showing a markeddecrease in viability. These data demonstrate that it is thebalance of Pk isoforms that provides normal vesicle transportfunctions in axons of fly neurons. Additionally, we show that

when the Pksple isoform predominates, microtubule (MT) po-larity can be radically altered in axons. However, when the Pkpk

isoform predominates, MT polarity is normal, whereas MTgrowth dynamics are altered such that there is an increase inplus-end MT polymerization. Our data establish that Prickleisoforms work in concert to promote the proper transport ofvesicles in neurons and demonstrate that Prickle function isnecessary for establishment of the correct polarity and growthdynamics of MTs within axons.

ResultsTo determine whether mutations in pk were solely responsiblefor the seizure phenotype we previously reported in pksple flies(2), we outcrossed both pksple1 and pkpk1 loss-of-function allelesinto an Oregon-R (OR) background and then subjected pksple/+and pkpk/+ flies to a modified bang-sensitivity behavioral assay,used to assess seizures in adult flies (Materials and Methods) (2,9). Given that homozygous pksple and pkpk loss-of-function allelesresult in strong PCP phenotypes on the adult fly body, whichmight interfere with their climbing ability, we used pksple and pkpk

heterozygotes that are devoid of the epidermal PCP phenotypes.As shown in Fig. 1A, pksple/+ flies are seizure-prone at all timepoints (Movie S1, clip 1), whereas pkpk/+ flies recover as wellas (or somewhat better than) the control OR flies (Movie S1,clip 2), and similar results were obtained using additional loss-of-function pksple and pkpk alleles (Fig. S1A). Additionally, pksple/+flies showed a seizure phenotype before outcrossing (2). Col-lectively, these data demonstrate that genetic background does

Significance

Mutations in prickle genes from flies to humans cause epilepsy,a disorder that affects approximately 1% of the population.Although Prickle has been studied for many years, its molecularfunction is unknown. We now show that Prickle modulatestransport of vesicles in fruit fly neurons. Additionally, we showthat prickle mutants have electrophysiological defects consis-tent with epilepsy, and merely altering the balance of Prickleisoforms causes seizures. Finally, we demonstrate that re-ducing the level of vesicle transport motor proteins can suppressprickle-mediated seizures, revealing a previously unidentifiedpathway in the pathophysiology of epilepsy.

Author contributions: S.N.E., A.I., G.J.I., J.D.A., S.G., C.-F.W., and J.R.M. designed research;S.N.E., A.I., G.J.I., and C.C. performed research; D.G. and J.D.A. contributed new reagents/analytic tools; S.N.E., A.I., A.U., G.J.I., C.C., A.G.B., D.G., S.G., C.-F.W., and J.R.M. analyzeddata; and J.R.M. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1S.N.E. and A.I. contributed equally to this work.2To whom correspondence should be addressed. Email: john-manak@uiowa.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1403357111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1403357111 PNAS | July 29, 2014 | vol. 111 | no. 30 | 11187–11192

NEU

ROSC

IENCE

not significantly influence the seizure phenotype and that the pksple

mutation is solely responsible for the seizure phenotype. Moreover,considering that both the pksple/+ and pkpk/+ flies are devoid of overtPCP alterations, the epilepsy phenotype can be genetically sepa-rated from the PCP phenotypes (5). Generating a pkpk and pksple

heteroallelic combination (which restores the balance between thetwo isoforms) rescues the pksple/+ seizure phenotype (Fig. 1A).To explore whether pksple mutants exhibited characteristic

seizure-like activity in an intact neuronal circuit similar to what isobserved for other seizure-prone fly mutants (9–12), we used anelectroconvulsive stimulation (ECS) paradigm (9, 13, 14). High-frequency stimulation across the brain triggers a stereotypicseizure discharge across the whole nervous system. The dischargepattern can be faithfully monitored by recording the spikingactivity of the dorsal longitudinal flight muscle (DLM), whosespiking reflects the motor output of its identified motor neuron(Materials and Methods) (9, 13, 14) (Fig. 1 C–E). We found thatpksple/+flies showed a lower threshold required to trigger seizuredischarges compared to WT flies (mean thresholds of 50 ± 4.2 Vand 65 ± 3.7 V, respectively, P = 0.01). Furthermore, morespikes were observed during seizure discharges in pksple/+mutants compared to control (Mann–Whitney U, P < 0.05;Fig. 1 D and E, see legend for median spike counts). Although

pkpk/+ flies did not show any significant difference in spikingactivity during seizure discharges compared to controls, re-storing the balance of the isoforms in the pksple/+, +/pkpk het-eroallelic combination suppressed the spiking activity relative tothe pksple/+ flies (Fig. 1 D and E).Taken together, both the behavioral and electrophysiological

data suggest that the balance of Pk isoforms is critical for pro-ducing normal motor function and imply that tipping the balanceof isoforms so that Pkpk predominates might increase suscepti-bility to seizures. Thus, we used the GAL4/UAS system tooverexpress either the Pkpk isoform (UAS-pk) or the Pksple iso-form (UAS-sple) in neurons, or neurons plus additional tissues.As shown in Fig. 1B, overexpression of Pkpk (but not Pksple) inbrain, motor neurons, and muscles (using the A51-Gal4 driver;Movie S1, clip 3), or all neurons (using the elav-Gal4 driver; Fig.S1B), produced seizure-prone flies that took longer to recovercompared to controls. These results are consistent with what isobserved for the heterozygous mutant phenotypes whereby tip-ping the balance such that Pkpk predominates (as in the pksple/+flies) leads to a seizure phenotype, but tipping the balance suchthat Pksple predominates (as in the pkpk/+ flies) does not.A critical determinant of normal bristle and hair polarity in

cells of the Drosophila wing is the proper localization of Fz andDsh to the distal end of the cells, and vesicles have been shownto transport both of these proteins along MTs (15, 16). Becausepkpk homozygous mutants show altered bristle/hair polarity in thewing (5, 7), which would be predicted if Fz/Dsh vesicles werenot properly transported distally, we reasoned that Pkpk mightbe involved in directly influencing transport of the Fz/Dsh-containing vesicles in the wing cells and that perhaps similarmechanisms might underlie vesicle transport in neurons. We thusused an in vivo system (Materials and Methods) to assess axonalvesicle transport dynamics in Drosophila larvae and expressed aYFP-tagged synaptic vesicle protein (amyloid precursor protein-GFP, or APP-YFP) in third-instar larval segmental nerves incontrol, pkpk/+, and pksple/+ larvae, followed by fluorescencemicroscopy of larval fillet preps (Movie S2 and Table S1) (17–19). Removing just one functional copy of the pkpk isoformwas sufficient to severely perturb vesicle transport (Fig. 2B),and significant decreases in both anterograde and retrogradevesicle velocities were observed (Fig. 2 C, i), in addition to anincrease in vesicle pause duration (Fig. 2 C, ii) compared tocontrol. Moreover, a greater percentage of vesicles remainedstationary, whereas a decreased percentage of vesicles reverseddirection, compared to control (Fig. 2D).Given the severe reduction in vesicle transport (an essential

cellular function) when pkpk isoform levels were reduced (but notwhen pksple isoform levels were reduced, discussed below), wereasoned that loss of pkpk (but not pksple) would likely causea reduction in viability. We collected 1,000 homozygous mutantembryos for each genotype and monitored their developmentfrom embryos to white prepupae and then from pupae to adultflies compared to controls (Table S2). Only the pkpk mutantsshowed a significant reduction in the number of embryos thatreached the white prepupal stage (P = 0.0172), and we alsoobserved a reduction in the percentage of pkpk pupae thateclosed to reach adulthood (P = 0.0469). These data are con-sistent with the hypothesis that there is a high demand for effi-cient vesicular transport across fly development, most notablyduring the stages of the greatest growth in size (from embryoto pupae) when embryos undergo substantial cell and tissuedifferentiation events.In contrast, removing one functional copy of the pksple isoform

caused reduced vesicle pause frequencies in both anterogradeand retrograde directions compared to control, although onlythe anterograde reductions were statistically significant (Fig. 2 C,iii). We also observed a decrease in the percentage of vesiclesmoving in a retrograde direction but found no difference for

Fig. 1. Tipping the balance of pk isoforms predisposes flies to seizure. (Aand B) Modified bang-sensitivity behavioral assay. (A) pksple/+ flies (sple/+)are seizure-prone and take longer to recover compared to Oregon-Rcontrols (+/+); all time points (P < 10−5; Student’s t test). Error bars show SEM.(B) Overexpression (in brain, motor neurons, and muscles; A51-Gal4 driver)of pkpk (UAS-pk) but not pksple (UAS-sple) is sufficient to produce seizure-prone flies; all time points (P < 10−15; Student’s t test). (C–E) ECS stimulationparadigm. (C) High-frequency ECS applied across the brain triggers a ste-reotypic seizure discharge; panel adapted from ref. 9. (D) RepresentativeDLM spiking activity demonstrating that pksple/+ flies exhibit pronouncedspiking activity at lower voltages. (E) Box blots of spike counts triggered byECS (60 V) demonstrate that pksple/+ flies have a statistically significant in-crease in total spiking activity (Mann–Whitney U, P < 0.05). Median spikecounts: +/+ = 32; pksple/+ = 66, pkpk/+ = 11, and pksple/pkpk = 2.5. n = 10–14flies per genotype for ECS experiments. n =12 vials per genotype for be-havioral experiments, repeated a minimum of two times, with similar results.

11188 | www.pnas.org/cgi/doi/10.1073/pnas.1403357111 Ehaideb et al.

vesicles moving in the anterograde direction (Fig. 2D). However,all other transport parameters were similar to what was observedin neurons of control animals. Nonetheless, taken as a wholethese data demonstrate that overall anterograde vesicle move-ment is enhanced in the pksple/+ background owing to the com-bination of a net increase in anterograde vesicle distance traveledover time and a greater relative percentage of vesicles moving inthe anterograde direction.To determine whether Pkpk protein could colocalize with the

Kinesin-1 motor protein complex [which is involved in ante-rograde transport of the APP-YFP vesicles along MTs (17–20)],we performed immunohistochemistry on third-instar larval filetpreparations to visualize both a neuronally expressed eGFP-Pkpk

in addition to Kinesin heavy chain (Khc) and observed anoverlapping pattern of puncta throughout the axons (Fig. 3 A–C)as well as boutons (Fig. 3 A′–C′). It is important to note thateGFP-Pkpk, although localized to puncta along the axon, doesnot move through the axon during live imaging, in contrast tovesicle-associated Kinesins. We speculate that colocalization ofeGFP-Pkpk and Kinesin might be due to traffic jams (axonalblockages) of the Kinesin-associated vesicles that have accumu-lated at the sites of Pkpk function (17, 18) owing to the eGFP-Pkpk

overexpression (staining of control larvae for Kinesin does notreveal Kinesin-positive axonal blockages or puncta). It is im-portant to note that axonal blockages are oftentimes seen whengenes encoding proteins known to be involved in axonal trans-port are either reduced in dosage [such as amyloid precursor

protein-like (Appl), Khc, Kinesin light chain (Klc), and dynein heavychain (17, 19, 21)] or overexpressed [such as Appl (17)]. Therefore,our data are consistent with both Pkpk and Kinesin functioningat the level of the MT to help promote vesicle transport.Because the pksple/+ flies showed a net increase in anterograde

vesicle transport, we reasoned that reducing transport of vesiclesby reducing the dose of one of the Kinesin-1 anterograde motorsubunits (Klc) in the context of the pksple mutation might sup-press the seizure phenotype. As shown in Fig. 4A, pksple/+ fliesare strongly seizure-prone, whereas Klc heterozygotes recoversimilarly to the control. However, removal of one functional copyof Klc in the context of the pksple heterozygote (pksple/+; +/Klc)fully suppresses the seizure phenotype such that the trans-heterozygotes now recover as efficiently as controls (Movie S1,clip 4). Additionally, the seizure phenotype of pksple /+ flies couldbe similarly suppressed by removing a functional copy of anothersubunit of the anterograde motor complex, Khc (Fig. S1C).We then used the ECS protocol to examine whether reducing

the Klc dose in the pksple/+ flies would ameliorate their seizuresusceptibility (9, 12). Only pksple/+, but not Klc/+, flies showeda significant reduction (P = 0.01) in ECS intensity requiredto trigger a seizure (50 V ± 4.2 and 59 V ± 4.8, respectively)compared to controls (65 V ± 3.7). However, pksple/+; +/Klctransheterozygotes showed a normal spiking distribution (Fig. 4B and C) and no significant difference in seizure threshold(63 V ± 4.5) compared to controls. Comparable results wereobtained when we reduced the dose of Khc in the pksple/+ fliessuch that the pksple/Khc transheterozygotes showed similar spik-ing activities (Fig. S2) as well as seizure thresholds comparedto controls (62 V ± 6.0 vs. 65 V ± 3.7), as did the Khc/+ flies(69 V ± 6.0). Taken together, these data demonstrate that low-ering the dose of either of two Kinesin-1 anterograde motorproteins critical for axonal vesicle transport can suppress theseizure phenotype of pksple/+ flies in both seizure behavioral aswell as electrophysiological assays.Having identified a variety of microtubule-mediated vesicle

transport defects in the segmental nerves of the pk mutants, wetested whether Pk isoforms might be directly acting at the levelof MT organization to influence transport. We imaged motorneurons of larvae overexpressing (i) end binding protein 1 (EB1)-GFP (an MT plus-end tracking protein), (ii) EB1-GFP and the

Fig. 2. Heterozygous pk mutations alter axonal transport in larval neurons.(A) Diagram of segmental nerve illustrating vesicle movements. (B) Kymo-graphs of APP-YFP–tagged vesicles show that APP-containing vesicle trans-port is strongly reduced in pkpk/+ flies. Rightward and leftward descendinglines represent anterograde- and retrograde-moving vesicles, respectively.Vertical lines represent stationary vesicles. (C) Quantification of vesiclemovement shows alterations of transport parameters in pkmutant larvae. (i)pkpk/+ larvae show decreases in duration-weighted segemental velocities ofboth anterograde and retrograde traveling vesicles. (ii) pkpk/+ larvae showincreases in normalized total pause duration of both anterograde and ret-rograde traveling vesicles. (iii) pksple/+ larvae show a decrease in bothanterograde and retrograde vesicle corrected pause frequencies, althoughonly the anterograde pause frequencies were statistically significant. (D)pksple/+ larvae show a reduction in vesicles moving in a retrograde directionas well as an increase in number of stationary vesicles. pkpk/+ larvae show anincrease (greater than fourfold) in the number of stationary vesicles anda decrease in the number of reversing vesicles. White bars represent Oregon-R(+/+) larvae, red bars represent pksple/+ larvae, and green bars representpkpk/+ larvae. n = 5 larvae. Error bars show SEM. See Table S1 for P values.

Fig. 3. Pkpk colocalization with a Kinesin motor protein. Confocal images ofthird-instar larval NMJ preparations show that a neuronally expressed eGFP-Pkpk (green) colocalizes (arrows) with Kinesin heavy chain (red) in bothsegmental nerves (A–C) (scale bar, 5 μM) and NMJ terminal boutons (A′–C′)(scale bar, 1 μM).

Ehaideb et al. PNAS | July 29, 2014 | vol. 111 | no. 30 | 11189

NEU

ROSC

IENCE

Pksple isoform (thus mimicking the isoform imbalance in pkpk/+larvae where transport is inhibited), and (iii) EB1-GFP and thePkpk isoform (thus mimicking the isoform imbalance in the seizure-prone pksple/+ flies where transport is enhanced). In both controlneurons and neurons expressing the Pkpk isoform we observed98% of the EB1-GFP comets moving toward the nerve terminal(reflecting a unified, canonical polarity of MTs; Fig. 4D, MovieS3, clips 1 and 2, respectively, and Table S3); however, the Pkpk-expressing larvae showed a significant increase in the length of theEB1-GFP comets compared to control larvae (Mann–Whitney U,P < 0.0001; Fig. 4E), demonstrating that there is an increase in

MT plus-end polymerization when the balance is tipped in favorof the Pkpk isoform (22, 23). In neurons expressing the Pksple

isoform, a large fraction of the MTs were actually reversed inorientation, with 36% of the comets moving toward the cell bodyand 64% of the comets moving toward the nerve terminal (Fig.4D, Movie S3, clip 3, and Table S3). Collectively, these datasuggest that Pkpk promotes and/or stabilizes the proper polarityof axonal MTs in addition to modulating MT growth dynamics,whereas Pksple disrupts the unified MT polarity, perhaps bypromoting/stabilizing reverse polarity or disrupting a polarizingfunction of the Pkpk isoform.

DiscussionSimilar to what has been observed regarding the relationship ofthe two Prickle isoforms while influencing planar cell polarity[with more severe hair and bristle phenotypes observed whenone or the other isoform is eliminated and less severe pheno-types observed when both isoforms are eliminated (5)], Pkpk andPksple must also be in balance in neurons to promote normalvesicle transport and proper neuronal function. Tipping thebalance in one direction where Pksple predominates createsmixed MT polarity along with severe reductions in transport(and significant reductions in viability but no seizure phenotype),whereas tipping the balance in the other direction where Pkpk

predominates has the opposite effect such that overall ante-rograde transport is enhanced, with an increase in seizure sus-ceptibility and alteration in MT growth dynamics.Our data are consistent with a mechanism whereby Prickle

promotes cell polarity by influencing MT-dependent vesicletransport, in part by modulating both MT polarity and MTgrowth dynamics. Shimada and colleagues (15) previouslyreported that in wing cells Fz is transported in vesicles towardthe plus ends of MTs, eventually accumulating at the distal endsof the cells. pkpk mutants show disruptions in hair and bristlepolarity in the wing (5, 7), and recent evidence (16) shows thatprickle influences vesicle transport in wing cells in a mannersimilar to what we have observed in neurons. Given the similareffects on MT polarity in two different cell types, we thereforehypothesize that Prickle plays a more generalized role in affectingvesicle transport in many cell types, with neurons being partic-ularly sensitive to perturbations in vesicle transport given thesignificant lengths vesicles must traverse in many axons. It is thusnot surprising that epilepsy was the first disease to be associatedwith human PRICKLE mutations.Our data show that vesicle velocity, directional change, re-

covery from a stall, and the overall ability to move are severelycompromised in pkpk heterozygotes, thus resulting in an overalldecrease in transport. We speculate that when Pksple predom-inates (as in pkpk/+) the directed MT polarization is lost, creatinga network of antiparallel MTs that would significantly impedethe sustained transport of a vesicle motor complex along thelength of an axon, producing the multitude of vesicle transportdefects that we observe. Nonetheless, a comparable percentageof vesicles in the pkpk heterozygotes compared to controls wastransported in both anterograde and retrograde directions (al-beit at much slower velocities and increased pause durations;Fig. 2D), and thus there must be sufficient vesicle transport tosupport life and allow a reasonable level of cellular function inanimals when the pkpk isoform is lost. Although reduction of pkpk

was not associated with a seizure phenotype, we did observe astrong reduction in viability of pkpk homozygotes compared topksple homozygotes or controls (Table S2), suggesting thatthere is a developmental cost to compromising efficient cellularvesicle transport.In contrast to what we observed in pkpk heterozygotes, vesicles

in the seizure-prone pksple heterozygotes show a pronouncedreduction in pause frequencies, most notably in the anterogradedirection, even though vesicle velocities, vesicle pause durations,

Fig. 4. Reducing the dose of a Kinesin motor subunit suppresses the pksple/+seizure phenotype. (A) Modified bang-sensitivity behavioral assay demon-strating that reduction of the dose of Klc in combination with the het-erozygous pksple/+ mutation suppresses the pksple/+ behavioral seizurephenotype. Error bars show SEM. n = 6 vials per genotype, repeateda minimum of two times with similar results. (B) Representative DLM spikingactivity demonstrating that reduction of the dose of Klc in combination withthe heterozygous pksple mutation suppresses the seizure discharge spikingactivity of the pksple /+ flies. (C) Spike counts after ECS (60 V) demonstratingthat the increase in the number of spikes per seizure discharge in the pksple /+flies (Mann–Whitney U, P < 0.05) is suppressed by reducing the dose of Klc.Median spike counts: +/+ = 32, pksple/+ = 66, Klc/+ = 31, and pksple/+; +/Klc =18.5. n = 10–14 flies per genotype. (D) Bar graph representation of EB1-GFPcomet directionality in the control, UAS-sple, and UAS-pk experimentallarval motor neurons demonstrating that a significant fraction of micro-tubules are reversed in polarity when the Pksple isoform predominates. Pink(MT plus ends toward nerve terminal) and blue (MT plus ends toward cellbody). n = 8 larvae per genotype. (E) Length of EB1-GFP comets in thecontrol, UAS-sple, and UAS-pk experimental larval motor neurons demon-strating that when Pkpk predominates the EB1-GFP comets are longer(Mann–Whitney U, P < 0.0001). EB1-GFPmedian comet length: control, 2.19 μm;UAS-sple, 1.83 μm; and UAS-pk, 4.15 μm. n = 4 larvae per genotype; ∼15comets were tracked per larva. All UAS constructs were driven by D42-Gal4.

11190 | www.pnas.org/cgi/doi/10.1073/pnas.1403357111 Ehaideb et al.

and vesicle reversals are unaffected. Additionally, although theoverall percentage of vesicles that are either stationary or movingin a retrograde direction is roughly the same in OR (∼21%) orpksple/+ (∼22%), the relative percentages in each category showan inverse relationship between the genotypes; specifically, inOR, ∼15% of these vesicles move in a retrograde directionwhereas ∼6% are stationary. In contrast, in pksple/+ animals,∼14% of the vesicles are stationary whereas only ∼8% move ina retrograde direction. The reason for the reduction in retro-grade traffic in the pksple heterozygotes is unclear, but these datasuggest that a significant proportion of vesicles normally desig-nated for retrograde transport are now stationary in the pksple/+mutant, and that reducing the pksple isoform partially com-promises retrograde traffic. Taken together, the reduced ante-rograde pause frequencies as well as the decreased retrogradetransport would both serve to enhance the overall anterogradetransport relative to retrograde transport, which we believe is theprincipal cause of the seizure phenotype. Consistent with thismodel, reducing the level of either of two Kinesin subunits (Klcor Khc), both of which are anterograde motor proteins, fullysuppresses the seizure phenotype, perhaps by (i) reestablishingthe proper balance of anterograde versus retrograde traffic (suchthat the reduction in the percentage of vesicles moving in a ret-rograde fashion in pksple/+ animals is balanced by a specific re-duction in anterograde transport in pksple/+, +/Klc, or pksple/Khc)or (ii) suppressing the enhanced anterograde traffic caused byreduced pause frequencies in the pksple/+ animals, or a combi-nation of both i and ii. Perhaps Pkpk biases transport in theanterograde, plus-end direction, which would also be consistentwith the requirement of Pkpk function in wing cells to transportplus-end-directed Fz-containing vesicles to the distal ends of thecells (16). It is intriguing to note that causing Pkpk to pre-dominate in motor neurons results in altered MT growth dy-namics, with MTs showing increased polymerization at their plusends as evidenced by longer EB1-GFP comets. We speculate thatreduced pause frequencies observed in the pksple/+ mutants arelinked to this change in growth dynamics; perhaps an increase inpolymerization results in an overall increase in MT length, thusallowing vesicle motor complexes to move a greater distancebefore encountering pauses when reaching the end of one MTbefore transitioning to the next. Further work is required tosort out the precise molecular mechanisms controlled by thePrickle isoforms.Although this is to our knowledge the first direct genetic evi-

dence demonstrating that mutations in an ortholog of a knownhuman epilepsy gene produce seizures through altered vesicletransport, mutations in genes encoding motor or motor-associatedproteins (such as Kinesin-1 family member KIF5A or lissencephaly 1(LIS1), a gene encoding a factor that clamps dynein to MTs) havebeen shown to produce seizure activity in mice, worms, and some-times humans (24–29), suggesting that vesicle transport defects,normally linked to a variety of neurodegenerative diseases, can alsobe associated with seizures. In the case of KIF5A, it was shownthat a conditional knockout of the gene led to impaired GABAAreceptor-mediated synaptic transmission in inhibitory neurons,thus producing epileptic phenotypes, and KIF5A (but neitherKIF5B nor KIF5C, the two other Kinesin-1 family members) wasshown to directly interact with a GABAA receptor-associatedprotein (26). Additionally, rescue experiments showed that ex-pression of KIF5A (but neither KIF5B nor KIF5C) was able torescue the GABAA receptor transport defect, demonstrating thatonly one of the three Kinesin-1 family members was required forthis specific transport function. In the case of LIS1, a presynapticdefect was observed such that synaptic vesicles were improperlydistributed in GABAergic neurons (25). Although such datapoint toward motor-associated transport problems as beingcorrelated with seizures, defects in the ability to properly recyclesynaptic vesicles via dynamin/shibire in mice/flies, or amphiphysin

in mice, can also cause seizure activity, suggesting that overallcontrol of vesicle trafficking in neurons is critical for suppressingseizures (30–32).How, then, might alteration of vesicle transport dynamics

observed in the pksple/+ flies be increasing seizure susceptibility?The first possibility is that neuronal and/or axonal migrationdefects might play a role, because it has been observed that LIS1mutations are associated with migration defects (33), althoughthis has not been observed in worms (25). While we cannotfirmly rule this out, we do not observe obvious defects in neu-ronal positioning, nor do we see defects in axonal pathfinding tothe somatic musculature in pksple mutant larval filet preparations.A second possibility is that the enhanced anterograde transportcoupled with reduced retrograde transport is perturbing theproper distribution of proteins within the neurons, whether atthe nerve terminal or elsewhere. Improper distribution/concen-tration of voltage-gated ion channels, for example, could haveprofound effects on neuronal excitability. A corollary to this isthat specific classes of neurons might be particularly sensitiveto such vesicle transport defects, similar to what is seen forGABAergic neurons in the KIF5A mutants (26). Intriguingly,sodium and potassium channel-containing vesicles have beenshown to be transported in axons by Kinesin-1 (34, 35), which isthe Kinesin that we have focused on in this study, raising thepossibility that transport defects of such channels might underliethe seizure phenotype in the pksple/+ mutant. These channelsassociate directly with Khc (34, 35), yet we are able to suppressthe seizure phenotype by reducing the dose of either Klc or Khc;collectively, these data suggest that Klc might be working to-gether with Khc to modulate channel binding or transport, andthat lowering the dose of either Kinesin component suppressesthe aberrant transport of the channels.Alternatively, the seizure phenotype might be associated with

improper transport of a different set of neuronal componentssuch as those necessary for synaptic maturation and function.The Kinesin-3 motor protein Imac (encoded by immaculateconnections) is the Kinesin in flies known to transport many ofthese components to the nerve terminal, including Synapto-tagmin and Vesicular glutamate transporter (36); however, syn-aptic components have been shown to be transported by Kinesin-1in rats, including Syntaxin-1 and Bassoon (37), suggesting thataltered transport of an as-yet undetermined synaptic componentmight underlie the seizure phenotype in the pksple/+ flies. Experi-ments are currently underway to identify differentially localizedproteins in the pksple mutants, as well as to test whether pksple-altered vesicle transport in different classes of neurons increasesseizure susceptibility. Nonetheless, the full suppression of theseizure phenotype, both behaviorally and electrophysiologically,through reduction of either of two Kinesin motor protein subunits(for which the only known function is MT-mediated transport)firmly demonstrates that altered vesicle transport underlies theseizure phenotype.It is worth mentioning that Noebels (38) recently presented

compelling evidence linking epilepsy and Alzheimer’s disease(AD), and several reports describe connections between genesassociated with neurodegeneration and seizures (38–40). Wenote that two of the AD-associated proteins, amyloid precursorprotein and presenilin 1 (APP and PSEN1; Appl and Psn inflies), are components of the same vesicle in mammalian cells,and that Appl and Psn mutations in flies both affect vesicletransport, with reduction in Psn levels linked to increased netvesicle transport, similar to what we observe for the APP-YFPvesicles in the seizure-prone pksple/+ flies (17, 19, 38–42). APPbinds to a Kinesin subunit (43), and its transport is dependent onKinesin (44). Thus, alterations in axonal transport may underliethe connection between these disease processes. Additionalstudies are warranted to investigate the efficacy of drugs developed

Ehaideb et al. PNAS | July 29, 2014 | vol. 111 | no. 30 | 11191

NEU

ROSC

IENCE

for AD and other neurodegeneration disorders as potential anti-epileptic therapies.

Materials and MethodsModified Bang-Sensitivity Behavioral Assay. Flies were collected immediatelyafter eclosion, aged for 7 d, and mechanically stimulated with a vortex mixerfor 20 s. After vortex, flies were digitally recorded and their recovery wasassessed for up to 25 s; only flies that climbed off the bottom of the vial wereconsidered as recovered flies (2).

Electroconvulsive Stimulation Triggered Seizures. Action potentials wererecorded from flight muscles as previously described (9). ECS, a train of0.1-ms pulses delivered at 200 Hz for 2 s, was delivered across the brainwith voltages ranging from 30–80 V (Supporting Information).

Immunofluorescence. Third instar larvae were dissected and fixed with 4%(vol/vol) paraformaldehyde for 20 min and stained with rabbit anti-GFP andmouse anti-kinesin SUK4 (Supporting Information).

In Vivo Imaging and Vesicle Analysis. Larvae expressing YFP-tagged vesicleswere dissected and videos were taken at five frames per second at 200 ms for30 s using MetaMorph software (Supporting Information).

Viability Study. For each genotype, 1,000 embryos were grown on standardcornmeal media at 25 °C and monitored every 8 h to assess developmentalstages (Supporting Information).

EB1-GFP Comet Analysis. Larvae expressing EB1-GFP were dissected to ex-pose the segmental nerve and imaged using confocal microscopy. EB1-GFPcomets were manually counted/tracked and measured using ImageJ soft-ware (Supporting Information).

ACKNOWLEDGMENTS. We thank J. Comeron and M. Parida (University ofIowa) for help with statistical analyses. We thank D. Wadkins for assistingwith the viability study as well as G. Banker (Oregon Health and ScienceUniversity) and K. Shen (Stanford University) for invaluable discussions. Thiswork was supported by start-up funds (to J.R.M.), National Institutes of Health(NIH) Grant GM 088804 (to C.-F.W.), and NIH Fellowship NS 082001 (to A.I.).

1. Bassuk AG, et al. (2008) A homozygous mutation in human PRICKLE1 causes anautosomal-recessive progressive myoclonus epilepsy-ataxia syndrome. Am J HumGenet 83(5):572–581.

2. Tao H, et al. (2011) Mutations in prickle orthologs cause seizures in flies, mice, andhumans. Am J Hum Genet 88(2):138–149.

3. Mei X, Wu S, Bassuk AG, Slusarski DC (2013) Mechanisms of prickle1a function inzebrafish epilepsy and retinal neurogenesis. Dis Model Mech 6(3):679–688.

4. Paemka L, et al. (2013) PRICKLE1 interaction with SYNAPSIN I reveals a role in autismspectrum disorders. PLoS ONE 8(12):e80737.

5. Gubb D, et al. (1999) The balance between isoforms of the prickle LIM domainprotein is critical for planar polarity in Drosophila imaginal discs. Genes Dev13(17):2315–2327.

6. Tree DR, et al. (2002) Prickle mediates feedback amplification to generate asymmetricplanar cell polarity signaling. Cell 109(3):371–381.

7. Lin YY, Gubb D (2009) Molecular dissection of Drosophila Prickle isoforms dis-tinguishes their essential and overlapping roles in planar cell polarity. Dev Biol 325(2):386–399.

8. Gray RS, Roszko I, Solnica-Krezel L (2011) Planar cell polarity: Coordinating morpho-genetic cell behaviors with embryonic polarity. Dev Cell 21(1):120–133.

9. Lee J, Wu CF (2002) Electroconvulsive seizure behavior in Drosophila: Analysis of thephysiological repertoire underlying a stereotyped action pattern in bang-sensitivemutants. J Neurosci 22(24):11065–11079.

10. Tanouye MA, Ferrus A (1985) Action potentials in normal and Shaker mutant Dro-sophila. J Neurogenet 2(4):253–271.

11. Zhang H, et al. (2002) The Drosophila slamdance gene: A mutation in an ami-nopeptidase can cause seizure, paralysis and neuronal failure. Genetics 162(3):1283–1299.

12. Parker L, Padilla M, Du Y, Dong K, Tanouye MA (2011) Drosophila as a model forepilepsy: bss is a gain-of-function mutation in the para sodium channel gene thatleads to seizures. Genetics 187(2):523–534.

13. Pavlidis P, Tanouye MA (1995) Seizures and failures in the giant fiber pathway ofDrosophila bang-sensitive paralytic mutants. J Neurosci 15(8):5810–5819.

14. Lee J, Wu CF (2006) Genetic modifications of seizure susceptibility and expression byaltered excitability in Drosophila Na(+) and K(+) channel mutants. J Neurophysiol96(5):2465–2478.

15. Shimada Y, Yonemura S, Ohkura H, Strutt D, Uemura T (2006) Polarized transport ofFrizzled along the planar microtubule arrays in Drosophila wing epithelium. Dev Cell10(2):209–222.

16. Olofsson J, Sharp KA, Matis M, Cho B, Axelrod JD (2014) Prickle/Spiny-legs isoformscontrol the polarity of the apical microtubule network in planar cell polarity.Development 141(14)2866–2874.

17. Gunawardena S, Goldstein LS (2001) Disruption of axonal transport and neuronalviability by amyloid precursor protein mutations in Drosophila. Neuron 32(3):389–401.

18. Gunawardena S, Goldstein LS (2004) Cargo-carrying motor vehicles on the neuronalhighway: Transport pathways and neurodegenerative disease. J Neurobiol 58(2):258–271.

19. Gunawardena S, Yang G, Goldstein LS (2013) Presenilin controls kinesin-1 and dyneinfunction during APP-vesicle transport in vivo. Hum Mol Genet 22(19):3828–3843.

20. Mochizuki H, et al. (2011) Unc-51/ATG1 controls axonal and dendritic developmentvia kinesin-mediated vesicle transport in the Drosophila brain. PLoS ONE 6(5):e19632.

21. Power D, Srinivasan S, Gunawardena S (2012) In-vivo evidence for the disruption ofRab11 vesicle transport by loss of huntingtin. Neuroreport 23(16):970–977.

22. Pawson C, Eaton BA, Davis GW (2008) Formin-dependent synaptic growth: Evidencethat Dlar signals via Diaphanous to modulate synaptic actin and dynamic pioneermicrotubules. J Neurosci 28(44):11111–11123.

23. Tortosa E, Galjart N, Avila J, Sayas CL (2013) MAP1B regulates microtubule dynamics

by sequestering EB1/3 in the cytosol of developing neuronal cells. EMBO J 32(9):

1293–1306.24. Xia CH, et al. (2003) Abnormal neurofilament transport caused by targeted disruption

of neuronal kinesin heavy chain KIF5A. J Cell Biol 161(1):55–66.25. Williams SN, Locke CJ, Braden AL, Caldwell KA, Caldwell GA (2004) Epileptic-like

convulsions associated with LIS-1 in the cytoskeletal control of neurotransmitter sig-

naling in Caenorhabditis elegans. Hum Mol Genet 13(18):2043–2059.26. Nakajima K, et al. (2012) Molecular motor KIF5A is essential for GABA(A) receptor

transport, and KIF5A deletion causes epilepsy. Neuron 76(5):945–961.27. Trokter M, Surrey T (2012) LIS1 clamps dynein to the microtubule. Cell 150(5):877–879.28. Sebe JY, Bershteyn M, Hirotsune S, Wynshaw-Boris A, Baraban SC (2013) ALLN rescues

an in vitro excitatory synaptic transmission deficit in Lis1 mutant mice. J Neurophysiol

109(2):429–436.29. Dobyns WB (2010) The clinical patterns and molecular genetics of lissencephaly and

subcortical band heterotopia. Epilepsia 51(Suppl 1):5–9.30. Boumil RM, et al. (2010) A missense mutation in a highly conserved alternate exon of

dynamin-1 causes epilepsy in fitful mice. PLoS Genet 6(8):e1001046.31. Di Paolo G, et al. (2002) Decreased synaptic vesicle recycling efficiency and cognitive

deficits in amphiphysin 1 knockout mice. Neuron 33(5):789–804.32. Salkoff L, Kelly L (1978) Temperature-induced seizure and frequency-dependent

neuromuscular block in a ts mutant of Drosophila. Nature 273(5658):156–158.33. Reiner O, Sapir T (2013) LIS1 functions in normal development and disease. Curr Opin

Neurobiol 23(6):951–956.34. Barry J, et al. (2014) Ankyrin-G directly binds to kinesin-1 to transport voltage-gated

Na+ channels into axons. Dev Cell 28(2):117–131.35. Xu M, Gu Y, Barry J, Gu C (2010) Kinesin I transports tetramerized Kv3 channels

through the axon initial segment via direct binding. J Neurosci 30(47):15987–16001.36. Pack-Chung E, Kurshan PT, Dickman DK, Schwarz TL (2007) A Drosophila kinesin re-

quired for synaptic bouton formation and synaptic vesicle transport. Nat Neurosci

10(8):980–989.37. Cai Q, Pan PY, Sheng ZH (2007) Syntabulin-kinesin-1 family member 5B-mediated

axonal transport contributes to activity-dependent presynaptic assembly. J Neurosci

27(27):7284–7296.38. Noebels J (2011) A perfect storm: Converging paths of epilepsy and Alzheimer’s de-

mentia intersect in the hippocampal formation. Epilepsia 52(Suppl 1):39–46.39. Chauvière L, et al. (2009) Early deficits in spatial memory and theta rhythm in ex-

perimental temporal lobe epilepsy. J Neurosci 29(17):5402–5410.40. Verret L, et al. (2012) Inhibitory interneuron deficit links altered network activity and

cognitive dysfunction in Alzheimer model. Cell 149(3):708–721.41. Kamal A, Almenar-Queralt A, LeBlanc JF, Roberts EA, Goldstein LS (2001) Kinesin-

mediated axonal transport of a membrane compartment containing beta-secretase

and presenilin-1 requires APP. Nature 414(6864):643–648.42. Pigino G, et al. (2003) Alzheimer’s presenilin 1 mutations impair kinesin-based axonal

transport. J Neurosci 23(11):4499–4508.43. Kamal A, Stokin GB, Yang Z, Xia CH, Goldstein LS (2000) Axonal transport of amyloid

precursor protein is mediated by direct binding to the kinesin light chain subunit of

kinesin-I. Neuron 28(2):449–459.44. Kaether C, Skehel P, Dotti CG (2000) Axonal membrane proteins are transported in

distinct carriers: a two-color video microscopy study in cultured hippocampal neurons.

Mol Biol Cell 11(4):1213–1224.

11192 | www.pnas.org/cgi/doi/10.1073/pnas.1403357111 Ehaideb et al.