Seizure control through genetic and pharmacological manipulationof Pumilio inDrosophila a key component of neuronal homeostasisWei-Hsiang Lin Carlo N G Giachello and Richard A Baines
ABSTRACTEpilepsy is a significant disorder for which approximately one-third ofpatients do not respond to drug treatments Next-generation drugswhich interact with novel targets are required to provide a betterclinical outcome for these individuals To identify potential noveltargets for antiepileptic drug (AED) design we used RNA sequencingto identify changes in gene transcription in two seizure models of thefruit fly Drosophila melanogaster The first model compared genetranscription between wild type (WT) and bangsenseless1 (parabss)a gain-of-functionmutant in the sole fly voltage-gated sodium channel(paralytic) The second model compared WT with WT fed theproconvulsant picrotoxin (PTX) We identified 743 genes (FDRle1)with significant altered expression levels that are common to bothseizure models Of these 339 are consistently upregulated and 397downregulated We identify pumilio (pum) to be downregulated inboth seizure models Pum is a known homeostatic regulator of actionpotential firing in both flies and mammals achieving control ofneuronal firing through binding to and regulating translation of themRNA transcripts of voltage-gated sodium channels (Nav) We showthat maintaining expression of pum in the CNS of parabss flies ispotently anticonvulsive whereas its reduction through RNAi-mediated knockdown is proconvulsive Using a cell-basedluciferase reporter screen we screened a repurposed chemicallibrary and identified 12 compounds sufficient to increase activity ofpum Of these compounds we focus on avobenzone whichsignificantly rescues seizure behaviour in parabss flies The mode ofaction of avobenzone includes potentiation of pum expression andmirrors the ability of this homeostatic regulator to reduce thepersistent voltage-gated Na+ current (INaP) in an identified neuronThis study reports a novel approach to suppress seizure andhighlights the mechanisms of neuronal homeostasis as potentialtargets for next-generation AEDs
KEY WORDS Anticonvulsant Drosophila Epilepsy PumilioSodium current Translational repression
INTRODUCTIONThe number of known contributory genetic loci to human seizureexceeds 500 which greatly increases the challenge of providingpersonalised medicine through tailoring treatments based on
individual gene mutation (Noebels 2015) An alternative is totarget treatment to common modifiers to which larger groupings ofindividual gene mutations contribute One obvious modifier isneuronal homeostasis which acts to stabilise neural circuit activitylevels through continual adjustment of neuron excitability(Turrigiano 2012) However this opportunity remains unexplored
Seizures in humans and Drosophila exhibit sufficient parallels toimplicate that the underlying neuronal abnormalities are highlysimilar This includes defined seizure thresholds common geneticmutations that modify seizure susceptibility spread of seizuresalong defined neuronal tracts and suppression of seizures byrecognised AEDs (Muraro and Baines 2008 Song and Tanouye2008) As in humans certain mutations in Drosophila genes resultin a seizure phenotype (collectively termed bang-sensitive)Seizures can also be induced in Drosophila by exposure toproconvulsants including picrotoxin (PTX) primarily throughblock of inhibitory GABAA receptors (Lin et al 2012)
To model seizure we used parabss which is a L1699F pointmutation that imparts a gain of function in the sole voltage-gatedsodium channel (NaV) of the fly genome (Parker et al 2011)Mutations in the human ortholog SCN1A are associated withseizure and intractable epilepsy (Escayg and Goldin 2010) Incomparison we also used exposure to PTX We exploited themolecular tractability of Drosophila to identify changes to genetranscription that occur during seizure to identify possible pathwaynodes exploitable for anticonvulsant therapy Comparison betweenthe two models identifies 743 common transcriptional changesincluding pum Pum is a translational repressor that binds mRNAtranscripts that normally (but not exclusively) contain an 8-nucleotide binding motif in their 3prime-UTR termed a Nanosresponse element [NRE also known as Pumilio response element(PRE)] (Gerber et al 2006) A particularly relevant Pum targetwith respect to seizure is Nav We have previously shown that thefly Nav ( paralytic) is translationally regulated by Pum and also thatrat Scn8a (Nav16) is similarly regulated by Pum2 (the closestmammalian homologue to pum) (Driscoll et al 2013 Mee et al2004 Muraro et al 2008) This mechanism forms part of a well-characterised homeostatic response that tunes action potential firingto match the changing level of synaptic excitation to which neuronsare exposed (Baines 2005 Weston and Baines 2007) Two recentstudies highlight the potential involvement of Pum in epilepsy Firsta Pum2 knockout mouse exhibits spontaneous seizures (Siemenet al 2011) and second PUM2 expression is reduced in humanpatients suffering temporal lobe epilepsy (TLE) (Wu et al 2015)
We show here that overexpression of pum in parabss flies ismarkedly anticonvulsant By contrast RNAi-mediated knockdownof pum exacerbates seizure The likely beneficial effect ofupregulation of Pum is through reduction of the voltage-gatedpersistent sodium current (INaP) in central neurons Thus our resultshighlight mammalian Pum2 as a potential target for the design ofnovel and possibly wide-spectrum AEDs To identify potentialReceived 8 July 2016 Accepted 5 December 2016
Division of Neuroscience and Experimental Psychology School of BiologicalSciences Faculty of Biology Medicine and Health University of ManchesterManchester M13 9PT UK
Author for correspondence (RichardBainesmanchesteracuk)
RAB 0000-0001-8571-4376
This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (httpcreativecommonsorglicensesby30) which permits unrestricted usedistribution and reproduction in any medium provided that the original work is properly attributed
141
copy 2017 Published by The Company of Biologists Ltd | Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
compounds that influence Pum activity andor expression weconstructed a luciferase reporter of Pum activity and screened acomprehensive library of approved compounds From 785compounds we identify 12 that potentiate Pum activity Furtheranalysis of one of these compounds avobenzone shows that itincreases transcription of pum reduces INaP in identifiedmotoneurons and is potently anticonvulsive in Drosophila
RESULTSRNA-sequencing identifies pum as downregulated in seizureIn order to determine changes to gene transcription that occur inseizure-prone CNSs we used RNA sequencing (RNA-seq) tocompare gene transcription in the CNS in two models of seizure agenetic model (parabss) and a chemical model (PTX) Using RNAextracted from the CNS of third instar larvae (L3) we identifiedtranscriptional change in 2246 and 1013 genes respectively using anFDRle1 in WT versus parabss and WT versus WT fed PTX (seeTables S1 S2) Comparison between data sets revealed that 743common genes exhibit significant change to expression (Fig 1Ainset see Table S3 for gene details) Of these 736 showed significantand consistent altered expression in both seizure models A log2 plotof fold-change (log2FC) showed that 339 (46) are significantlyupregulated and 397 (54) are significantly downregulated(P=0001 ANOVA) The remaining seven genes did not showconsistent direction of change (Fig 1A) Identified genes generated a
total of 130 functional clusters representing a wide array of functionsincluding predicted genes encoding ion channels and synapticproteins (detailed below) The top 20 enriched clusters are shown inFig S1 The top four clusters are for genes associated with pre-replicative complex assembly eukaryotic translation elongationfactor 1 complex negative regulation of neuroblast proliferationand translation repressor activity Genes associated with translationalrepression include minichromosome maintenance (orthologues 2 35 7) elongation factor 1α100E 1α48D and 1β anachronismprospero musashi embryonic lethal abnormal vision brain tumorand pum (Table S3) Twenty genes that we identify have beenpositively associated with human epilepsy (httpwwwinformaticsjaxorghumanDiseaseshtml) (red dots in Fig 1A and described inTable 1) Of these genes five were upregulated and 15 weredownregulated in the Drosophila seizure models These genesinclude paralytic (Nav) nicotinic Acetylcholine Receptor α5 Ihchannel and Shaker (K+ channels) in addition to Syntaxin Synapsinand unc-13 (synaptic proteins) Seven genes were identified that showparticularly large increases in transcription (gt3 log2FC blue dots inFig 1A) These genes are CG18331 (mucin 68Ca) CG34076(mitochondrial NADH-ubiquione oxidoreducatse chain 3)CG11205 (photorepair) CR41620CR40734 (rRNA genes) andCG7606CG32198 (unknowns)
Our attention was drawn to pum which was significantlydownregulated in both seizure models [meanplusmnsd WT 602plusmn14
Fig 1 Analysis of altered gene transcription in seizure models (A) Cross-comparison shows 743 changes are common to both seizure models Analysis ofdirection of log2 fold-change (log2FC) in transcription of the 743 common genes (main figure) shows that 339 are significantly (two-way ANOVA) upregulatedand 397 downregulated Seven genes show differential expression in the two models pumilio (pum) which is downregulated is identified by the orange dotGenes previously linked to human epilepsy are shown by red dots (described in Table 1) Blue dots highlight genes that show particularly large fold-changes(log2FCgt3) in expression levels in the seizure backgrounds (see Results text for identity) Inset analysis of the transcriptome by RNA-sequencing shows changeto transcription of 2246 genes in the parabss CNS compared with wild type (WT) Comparison of WT with WT fed picrotoxin (PTX) shows 1013 transcriptionalchanges (B) Analysis of pum transcript level in isolatedCNS from L3 shows a significant reduction inWT+PTX and parabss comparedwithWT controls (C) pum issignificantly reduced in adult heads in both seizure models compared with WT controls TheWT value has been set to 1 in each experimental condition Data aremeanplusmnsd for n=5 independent samples Ple005 Ple0001 (unpaired t-test)
142
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
vs WT fed PTX 405plusmn2 and parabss 381plusmn15 cpm (counts permillion) P=15times10minus5 n=3] This is because its homologue PUM2has been reported to be downregulated in humans sufferingtemporal lobe epilepsy (Wu et al 2015) Pum is a well-characterised translational repressor which we have previouslyreported regulates translation of Navs in both Drosophila and rat toachieve homeostatic control of neuron action potential firing(Driscoll et al 2013 Mee et al 2004 Muraro et al 2008) Weconsidered that manipulation of a homeostatic regulator mightrepresent a promising approach to control seizure To validateRNA-seq data we undertook RT-qPCR pum was significantlydecreased in WT larval CNS after exposure to PTX (083plusmn003)and in parabss (078plusmn003) compared with WT control (set as 1P=003 n=5 Fig 1B) We observed a similar and significantdownregulation of pum transcription in adult heads that containmostly brain tissue (WT fed PTX 041plusmn007 and parabss 058plusmn017) relative to WT control (set as 1 P=00002 n=5 Fig 1C)In addition we used RT-qPCR to validate the identification ofthe 20 genes that have been positively associated with epilepsy(red dots in Fig 1A) This validation was undertaken only for theparabss background We found consistent change for 14 of thegenes (representing a validation rate of 70) Two genes showedsignificant change by RT-qPCR but in the opposite direction toRNA-seq whereas four genes showed no significant change (seeTable 1)
Upregulation of Pum is anticonvulsiveWe have shown that Pum binds to mRNA encoding Navs in bothDrosophila and rat Binding subsequently reduces the density ofNav channels available in the neuronal membrane (Driscoll et al2013 Mee et al 2004 Muraro et al 2008) We predicted
therefore that maintaining pum expression in seizure backgroundswould be anticonvulsant Inducing seizure by vortexing of parabssYmale flies resulted in a recovery time (RT 114plusmn133 s n=3) that wassignificantly reduced by exposure to recognised AEDs (Parker et al2011) Vortexing WT flies by comparison resulted in a nearinstantaneous RT (53plusmn25 s n=3) This is the averaged time takenfor all flies in the vial (n=10) to regain a standing posture followingvortexing and does not imply that WT flies exhibit seizures Bycontrast expressing pum in a Cha-Gal4(19B) cholinergic neurondriver line (which are the predominant excitatory interneuron type inthe insect CNS Yasuyama and Salvaterra 1999) in parabss
( parabssY Cha-Gal4(19B)UAS-pum) flies significantly reducedseizure RT compared with control parabssY Cha-Gal4(19B)+ (7plusmn36 s vs 114plusmn133 s P=12times10minus5 n=3 Fig 2A) Indeed recoverytime following upregulation of pum was not significantly differentto WT controls (53plusmn25 s) indicative that seizures were completelysuppressed By contrast expression of pumRNAi using the sameCha-Gal4(19B) driver in the parabss background was stronglyproconvulsive (206plusmn304 s vs 114plusmn133 s P=00002 n=3Fig 2A) We observed the same outcome in L3 where seizurebehaviour was induced by electroshock (Fig 2B) Seizure RT wassignificantly reduced (134plusmn101 s P=0005 n=18) or increased(557plusmn255 s P=00004 n=20) by expression of either UAS-pum orUAS-pumRNAi respectively in Cha-Gal4(19B) cholinergic neuronsin the parabss background (control parabssY Cha-Gal4(19B)+324plusmn159 s n=20) Manipulation of pum levels pan-neuronally(using parabss elaV-Gal4) resulted in an identical effect toelectroshock-induced seizure in L3 (Fig 2C) Increasing pumexpression reduced seizure duration (133plusmn70 s P=0009 n=21) andRNAi-mediated knockdown increased seizure duration (255plusmn99 sP=002 n=42) compared with control (204plusmn88 s n=39)
Table 1 Identification of epilepsy-associated genes
CG number Drosophila geneparabss fold-change (log2)RNA-seqqPCR
Mouse and human homologue data from Mouse Genome Informatics HumanndashMouse Disease Connection database (httpwwwinformaticsjaxorghumanDiseaseshtml) + increased minus decreased mRNA levels compared with respective controls All numerical values shown are significant at Plt005 nsvalues not significantly different from control
143
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Manipulation of pum in a WT background resulted in a differentoutcome Both RNAi-mediated knockdown and particularlyoverexpression of pum resulted in an induction of a seizurephenotype [Cha-Gal4(19B)UAS-pumRNAi 187plusmn109 s P=18times10minus5n=23 Gal4(19B)UAS-pum 387plusmn77 s P=28times10minus26 n=20]compared with control [Cha-Gal4(19B)+ 97plusmn43 n=40Fig 2D] This paradoxical result is similar to the effect offeedingWTDrosophilaAEDs such as phenytoin that also result inseizure induction an effect that has also been observed in rat(Callaghan and Schwark 1980 Marley and Baines 2011Rundfeldt et al 1990)
Increased pum expression decreases INaP in motoneuronsOur previous work has shown that Pum regulates INa throughtranslational regulation of para (Mee et al 2004 Muraro et al2008) We recorded from parabssY L3 where the expression oftransgenic pum was selectively manipulated in only the aCC
motoneuron (using RRa-Gal4) Our choice to use this motoneuronis guided by the ability to combine genetics and electrophysiologya selective Gal4 driver exists to express UAS-transgenes in thisneuron which is also accessible to patch electrodes That INaP isgreater in amplitude in aCC motoneurons in seizure mutants(Marley and Baines 2011) is indicative that they share propertieswith central interneurons in human epilepsy which can also showincreased INaP (Stafstrom 2007)
Increased expression of pum in L3 parabss aCC resulted in astriking reduction of INaP (44plusmn41 pApF vs 126plusmn40 pApFP=49times10minus5 Fig 3ABD) but no change to INaT (Fig 3E)Analysis of the persistent-to-transient current ratio (PT) recorded inL3 aCC showed a marked reduction (200plusmn180 vs 510plusmn119P=50times10minus5 Fig 3F) A high PT ratio (gt40) in centralmotoneurons has been previously shown to be characteristic ofDrosophila seizure mutants and its reduction to be anticonvulsant(Lin et al 2015 Marley and Baines 2011) Thus we conclude thatupregulation of pum is anticonvulsant which is due at leastpartially to its ability to reduce INaP
RNAi-mediated downregulation of pum in L3 parabss aCCincreased INaT (313plusmn33 pApF vs 247plusmn45 pApF P=0005) butdid not affect INaP or the PT ratio (Fig 3C-F) Analysis of the effecton seizure behaviour following this more selective manipulation ofpum expression showed no significant differences to controls( parabssY RRa-Gal4+ data not shown) This is entirely expectedgiven the highly selective cell targeting used in these experimentsHowever a more widespread manipulation of pum [eg using Cha-Gal4(19B)] which is sufficient to alter seizure duration andorseverity probably acts via an identical mechanism throughalteration of INa
Increasing pum expression in aCC in a WT background resultedin essentially the same changes to INa as seen with manipulation inthe parabss background INaP was significantly reduced (24plusmn17 pApF vs 74plusmn49 pApF P=00028 Fig 3G) but no change toINaT was observed (188plusmn48 pApF vs 219plusmn27 pApF Fig 3H)By contrast downregulation using pumRNAi produced a differentoutcome compared with parabss INaP was significantly increased(110plusmn24 pApF vs 74plusmn49 pApF P=0032 Fig 3G) with noeffect on INaT (246plusmn47 pApF vs 219plusmn27 pApF Fig 3H)Analysis of the PT ratio however similarly only showed asignificant reduction following upregulation of pum expression inWT (147plusmn119 vs 333plusmn202 P=0016 Fig 3I)
On occasion we noted the appearance of multiple resurgent INaduring the INaP plateau in the para
bss background (Fig 4A indicatedby arrow) Moreover we observed a significant correlation betweenthe occurrence of resurgent INa and pum level (P=0002 Chi-squaretest Fig 4B) Thus resurgent INa was most often observedfollowing RNAi-knockdown and only rarely following expressionof pum The origin of these currents remains uncertain Analysis ofvoltage recordings (Fig 4A) showed no obvious issue of spaceclamp which suggests these currents are not occurring in distalunclamped regions of the neuron The averaged frequency of theresurgent currents was sim100 Hz which did not vary with level ofpum expression (Fig 4C) Resurgent currents are particularlyevident at holding potentials between minus50 to minus20 mV and exhibithighest frequency at minus30 mV (RRa-Gal4UAS-pumRNAi 10450plusmn3678 Hz RRa-Gal4+ 12000plusmn2016 Hz RRa-Gal4UAS-pum11500plusmn4093 Hz) Increased resurgent INa probably supportsincreased action potential firing consistent with our observationthat RNAi-mediated knockdown of pum is proconvulsant (Griecoet al 2005) Resurgent INa is only rarely observed (lt5) in WTaCC recordings (data not shown)
Fig 2 Expression of transgenic pum is anticonvulsive (A) Expression oftransgenic full-length pum lacking NRE motifs (UAS) in cholinergic neurons inparabss [Cha-Gal4(19B)gtpum] is sufficient to reduce recovery time (RT) frommechanical shock-induced seizure in adult flies compared with parabss alone(CTRL) By contrast further reduction of pum through RNAi-mediatedknockdown (RNAi) [Cha-Gal4(19B)gtRNAi] significantly lengthens seizure RTEach manipulation tested 10 flies per vial to produce an average value Thiswas repeated in triplicate and a final average calculated Data are meansplusmnsdn=3 (B) Identical manipulation of pum expression in cholinergic neurons in L3parabss had an identical effect on seizure duration when seizurewas evoked byelectroshock (C) Pan-neuronal manipulation of pum is also sufficient to affectelectroshock-induced seizure in L3 Upregulation (UAS parabss elaV-Gal4gtpum) reduces seizure and downregulation (RNAi parabss elaV-Gal4gtRNAi) increases seizure duration compared with control (parabss elaV-Gal4+) (D) Up- or downregulation of pum in aWT background using the Cha-Gal4(19B) cholinergic driver line results in induction of a seizure phenotypeData are meansplusmnsd n is stated in individual bars Ple005 Ple001Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
144
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
A screen to identify positive regulators of Pum activityUpregulation of Pum activity either through increased transcriptionor post-transcriptional modification might provide an effectivemeans to suppress seizures To identify possible lead compoundswith this mode of action we constructed a luciferase-based reporterof Pum activity for use in an in vitro S2R+ cell line suited to large-
scale screens (Lin et al 2015) Overexpression of pum is sufficientto repress luciferase activity (due to translational repression)whereas incubation with pum double-stranded RNA is sufficient toincrease luciferase activity by reducing endogenous Pum activityPCR analysis shows that pum is endogenously expressed in S2R+cells (Fig S2) Thus activity of the firefly-luciferase-NRE reporter
Fig 3 Expression of transgenic pum reduces INaP (A-C) Whole-cell patch recordings of INa from L3 aCC motoneurons in parabss (CTRL) parabss expressingtransgenic pum (UAS) or pumRNAi (RNAi) Transgene expression is limited to aCC motoneurons in these manipulations using RRa-Gal4 (DE) Expression oftransgenic pum (UAS) is sufficient to reduce the magnitude of INaP without change to INaT Expression of pumRNAi (RNAi) results in no change to INaP but asignificant increase in INaT (F) Persistent-to-transient (PT) current ratio for INa recorded in DE (GH) The effect of manipulating pum in a WT backgroundIncreasing expression (UAS) is sufficient to reduce INaP with no change to INaT whereas reduction (RNAi) increases INaP amplitude but has no effect on INaT(I) Analysis of the PT ratio in individual cells recorded in GH shows increased pum is sufficient to reduce the ratio Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
Fig 4 Occurrence of resurgent INa is related to level of pum (A) Resurgent INa (INaR arrow) is seen superimposed on repolarization of holding potential usedto evoke INaP Analysis of the voltage trace (lower trace) shows good control during this step (B) The occurrence of INaR in the parabss background is highestwhen pum is reduced (RNAi 82 14 from 17 cells) and lowest when increased (UAS 21 3 from 14 cells) Control (CTRL parabss 64 9 from 14 cells)Transgene expression was limited to aCC cells using RRa-Gal4 (C) Frequency of INaR oscillations is unaffected by expression level of pum Data are meansplusmnsd
145
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(FF-NRE) reflects the absolute level of Pum function in these cellsA second reporter which lacked an NRE-motif was also transfected[renilla (Ren)-luciferase] to allow detrimental effects to cellviability to be determined The final readout of the assay was aFFRen luciferase ratio that would be reduced followingupregulation of Pum activityWe screened 785 compounds from a repurposed library (see
Materials and Methods drugs screened are listed in Table S4) Weidentified 12 compounds that significantly reduced the FFRen ratioat 5 μM (Table 2) Based on structure andor known drug target thecompounds fall into one of four groupings those containing amethoxybenzaldehyde moiety (aniracteam and avobenzone) anti-cancer agents (cladribine gemcitabine floxuridine clofarabinebleomycin and docetaxel) mTOR inhibitors (temsirolimus andrapamycin) and topoisomerase II inhibitors (mitoxantrone andteniposide) Our attention was particularly drawn to avobenzonebecause unlike the other compounds it had no significant effect ontranscription of the control Ren-luciferase reporter (all othercompounds also reduced expression of this reporter in addition todecreasing the FFRen ratio) Thus we took avobenzone forward forfurther testing
Avobenzone potentiates activity of PumWe first tested for anticonvulsant activity in L3 parabss mutantsLarvae raised in food containing avobenzone (04 mgml) showedsignificantly reduced RT in response to electroshock (avobenzone213plusmn124 s n=40 vs control 339plusmn83 s n=20 P=00004 Fig 5A)Similarly exposure of adult parabss flies to avobenzone (04 mgml)24 h before testing also resulted in significant reduction of seizureduration (avobenzone 61plusmn29 vs control 138plusmn29 s n=5 P=00002Fig 5B) Next we recorded INa from parabss aCC in L3 that hadbeen raised on food containing different concentrations ofavobenzone (01-04 mgml Fig 5C-G) Avobenzone reducedINaP from 139plusmn76 pApF in controls to 76plusmn62 pApF at 01 mgml(P=017) 54plusmn64 pApF at 02 mgml (P=003) and 35plusmn42 pApF at04 mgml (P=0002) (Fig 5D) Conversely avobenzone treatment atthese concentrations did not induce any detectable effect in INaT(Fig 5E) Analysis of the PT ratio for INa shows that exposure toavobenzone significantly reduced this value from493plusmn92 in control to 280plusmn232 at 01 mgml (P=009) 219plusmn269 at 02 mgml (P=003) and 121plusmn132 at 04 mgml(P=00004) (Fig 5F) which compares favourably with
overexpression of pum (cf Fig 3) We also observed a significantcorrelation between avobenzone concentration and the occurrence ofresurgent INa (P=0005 Chi-square test Fig 5G)
Our predicted mode of action for avobenzone is inconsistent withan immediate effect of this compound acting instead to potentiatePum which in turn downregulates Nav channels in the neuronalmembrane To test this we recorded from non-drug-exposed L3parabss aCC and used bath application of avobenzone (5 microM) Nochanges were observed in either component of INa (data not shown)and the PT ratio remained unaffected (Fig 5H) Higher doses(20 microM) or longer exposure times (10 min) similarly produced nodetectable effect (data not shown) This lack of acute effect isconsistent with our predicted mode of action Finally to directly testthis prediction we measured pum transcript abundance in parabss
L3 grown in the presence of avobenzoneWe observed a modest butstatistically significant increase in transcript abundance of sim20(12plusmn017 n=5 P=004 t-test vehicle control set as 1 Fig 5I)Thus we conclude that avobenzone acting to increase thetranscription andor transcript stability of pum is able to suppressseizure duration through downregulation of INaP Finally weobserved equally potent anticonvulsive activity of avobenzone intwo other bang-sensitive mutants easily-shocked (avobenzone142plusmn82 vs control 240plusmn120 s n=40 P=10times10minus5 L3electroshock) encoding an ethanolamine kinase (Pavlidis et al1994) and slamdance (avobenzone 178plusmn122 vs control 272plusmn108 s n=40 P=68times10minus5 L3 electroshock) encoding anaminopeptidase (Zhang et al 2002) indicative that increasingPum activity might be effective against a broad range of epilepsies
DISCUSSIONThe causes of seizure even in genetic epilepsies vary greatly andare not confined to genes with obvious contributions to ion fluxacross neuronal membranes This increases the challenge to identifyindividual mutations to determine the physiological role of both theWT and mutated protein and ultimately to design drugs tominimise the unwanted effect of the mutation In this study weidentify transcriptional changes that occur in the seizure-proneCNS We identify over 700 common genes that show alteredtranscription in two different seizure models It is noteworthy thatwe observed approximately double the number of genes showingaltered transcription in parabss flies compared with those treatedwith PTX The reason for this is unclear but might representaccumulated compensatory changes in the mutant line that haveoccurred in order to lessen the severity of seizure activity in parabss
mutants These additional genes warrant further investigation aspotential seizure suppressors
Many of the common transcriptional changes we identify and inparticular those that are upregulated (and thus open to inhibition bydrug exposure) might provide effective drug targets for novel AEDdesign However our attention was drawn to Pum which we havepreviously shown orchestrates homeostasis of action potential firingin both Drosophila and rat central neurons (Driscoll et al 2013Mee et al 2004) The degree of seizure suppression achieved byupregulating Pum in parabss flies is considerable and is onlymatched by the no-action-potential (napts) allele of the maleless(mle) locus in Drosophila which encodes an ATP-dependentdouble-stranded RNA (dsRNA) helicase (Ganetzky andWu 1982)This mutation causes a catastrophic change in splicing of theDrosophilaNav (Reenan et al 2000) The net effect of both of thesemanipulations increased Pum or the presence of napts is to reducethe availability of functional Nav expressed in central neurons Thedirection of change of pum in the two seizure models (that show
Table 2 List of compounds that reduce the fireflyRenilla (FFRen)luciferase ratio thus mimicking the activity of increased pumexpression (shown at bottom of table for reference)
All but avobenzone also reduce expression of the control Ren luciferasereporter that does not contain an NRE motif Luciferase values shown arenormalised such that 10 would represent no effect
146
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
reduced expression) might not be ideal with respect to drugdevelopment given that disruption of a gene or protein is often moreachievable Nevertheless we show that upregulation of pum in aDrosophila seizure mutant is potently anticonvulsive and furtherwe identify a potential lead anticonvulsive compound thatseemingly increases the level of expression of this homeostaticregulator This compound might catalyse the development of anovel class of AEDNeurons display an array of homeostatic mechanisms to maintain
action potential firing within pre-determined and physiologicallyappropriate limits (Davis 2013) Pum is a well-characterised RNA-binding protein that binds mRNA usually through a specific motiftermed the NRE Once bound Pum recruits additional cofactorsincluding Nanos and Brain tumor (Brat) to form a complex that issufficient to prevent translation (Wharton et al 1998) Our results inthis study indicate that increased expression of Pum might havetherapeutic benefit for seizure suppression However a potential
issue in this regard is that a genome-wide identification of RNAsbound to Pum in ovaries identifies upwards of 700 genes(FDRlt01) (Gerber et al 2006) This raises the problem ofspecificity of effect following global potentiation of level or activityof Pum This potential issue might however be overcome throughidentifying and targeting neuronal-specific regulators of Pum Onesuch alternative target might be the inhibition of Myocyte enhancerfactor 2 (Mef2)-induced expression of miR-134 in neurons that inturn inhibits translation of mammalian PUM2 (Fiore et al 2009)Additional possibilities include targeting of cofactors required forPum activity It is interesting in this regard that a loss-of-functionmutation in mei-P26 a homologue of Brat produces strong seizuresuppression in Drosophila bang-sensitive seizure mutants(Glasscock et al 2005)
Mammalian PUM2 binds transcripts encoding SCN1A (Nav11)and SCN8A (Nav16) (Driscoll et al 2013 Vessey et al 2010) Areduction in supply of Nav protein to the neuron membrane is
Fig 5 Avobenzone is anticonvulsant and selectively reduces INaP (A) parabss L3 raised in food containing 04 mgml avobenzone show significantly reducedrecovery time (RT) following electroshock compared with controls (CTRL parabss+DMSO) (B) Exposure of adult parabss flies to avobenzone (04 mgml) is alsopotently anticonvulsant compared with controls (CTRL parabss+DMSO) Each manipulation tested 10 flies per vial to produce an average value This wasrepeated five times and a final average calculated (C) Whole-cell patch recordings of INa from parabss L3 aCC raised in food containing 04 mgml avobenzoneshow reduced INaP (DE) Increasing concentrations of avobenzone (01 02 and 04 mgml) induced a proportional decrease of INaP (D) without affectingINaT (E) (F) Persistent-to-transient (PT) current ratio for INa recorded in aCC (G) The frequency of cells that exhibit resurgent INa correlates with avobenzoneconcentration (P=0005 Chi-square test) (H) PT ratio measured from parabss aCC before (CTRL) and after a 1 min bath application of 5 microM avobenzone(I) Analysis of pum transcript level in isolated CNS from parabss L3 raised on food containing avobenzone (04 mgml) shows a significant increase compared withparabss L3 raised on food containing an equal amount of vehicle (08 DMSO) The control value has been set to 1 Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (AH-I unpaired t-test D-F two-way ANOVA with Bonferronirsquos post hoc)
147
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
consistent with a reduction in action potential firing and a generalanticonvulsant effect (Mee et al 2004) Analysis of INa inmotoneurons indicates that a likely mechanism includes a markedreduction in INaP Increased INaP is associated with mutations inSCN1A that have been identified from individuals with epilepsy(Meisler and Kearney 2005) and is specifically reduced by AEDssuch as phenytoin valproate and lamotrigine (Stafstrom 2007) Inlight of this the anticonvulsant effect of increased pum expression isunderstandable That reducing pum expression through RNAi-mediated knockdown is proconvulsive is again both predictable andunderstandable However the effect of this manipulation on INa isnot so clear Rather than increasing INaP INaT is instead significantlyincreased together with a novel appearance of resurgent INa duringrepolarisation Increased INaT would be expected to reduce thethreshold for action potential firing (ie making firing more likely)whereas resurgent INa is associated with increased action potentialfiring frequency partly by reducing the refractory period (Griecoet al 2005) Although we have observed this current component inrecordings from seizure mutants (including parabss) it is rarelyobserved in WT or following expression of transgenic pumThe ability to manipulate Pum in vivo to determine its
anticonvulsive properties in rodent models of seizure will begreatly aided by the identification of chemical compounds thatdirectly potentiate either expression or activity state We report theuse of a suitable cell-based screen to identify such compounds andhighlight avobenzone as a potential lead compound for futuredevelopment The in vivo toxicity of avobenzone has not been wellestablished and although there are few reports of serious side effectsassociated with its use as an active ingredient of sunscreen itstendency to form free radicals might be a potential issue To ourknowledge this compound has not been used to treat neurologicaldisease and its mode of action in reducing seizure in Drosophilaremains to be determined Our observations that ingestion ofavobenzone result in increased expression of pum is indicative thatthis compound might mimic elements of the pathway that controlexpression of this homeostatic regulatorThe output of our screen also provides additional support for the
use of rapamycin to control seizure (Lasarge and Danzer 2014Russo et al 2013) indicative that this molecule might influenceneuronal homeostasis The identification of topoisomerase II as apotential target to control seizure also validates previousobservations reporting that inhibition of this class of nuclearprotein is anticonvulsant (Lin et al 2015 Song et al 2008)Finally that we identify that the increase in Pum activity byaniracetam might hint at an additional mode of action for this classof known anticonvulsants (Shiotani et al 2000) The relatedracetams levetiracetam and brivaracetam are currently in clinicaluse as AEDs exploiting their capability to bind and inhibit synapticvesicle protein 2A (SV2A) (Klitgaard et al 2016)In summary we present a description of transcriptional change
present in seizure-prone CNS We identify in particular that pumexpression is downregulated in both genetic and chemically inducedseizure models This mirrors the reported reduction in PUM2 inhuman TLE and in rats exposed to the proconvulsant pilocarpine(Wu et al 2015) It also provides a possible understanding for whyPum2 null mice exhibit spontaneous seizures (Siemen et al 2011)However it is perplexing that pum levels should decrease duringseizures given that the published model predicts an increase (Meeet al 2004) As reduced Pum levels are predicted to increaseneuronal excitability it seems that epileptic seizures are associatedwith a pathological dysregulation of pum expression We speculatethat this occurs because Pum can auto-regulate (the pum transcript
contains NRE motifs) Thus although the neuronal hyperactivityinduced by seizures will initially increase Pum expression theaccumulating Pum protein might feed back to downregulate its owntranscript (Gerber et al 2006) Sampling at later stages after seizureoccurrence might only report reduced Pum compared with non-seizure controls Indeed we have shown that upregulation of pum inthe Drosophila CNS through expression of a wild-type transgene(lacking NRE motifs) results in reduction of endogenous pumtranscript level (W-HL and RAB unpublished data) Preventionof this feedback achievable in this study through expression oftransgenic pum lacking an NRE or exposure to avobenzone holdssignificant promise for anticonvulsant therapy
MATERIALS AND METHODSFly stocksWild type (WT maintained in the Baines lab) was Canton-S parabss (bss1)which was obtained from Dr Kevin OrsquoDell (Institute of Molecular Cell andSystems Biology University of Glasgow UK) is detailed in Parker et al(2011) The parabss stock (and other transgenic lines used) were notbackcrossed to the CS stock Controls consisted of either untreated parabss
andor parental stocks (ie Gal4+ UAS+) and are stated in respectivefigure legends Slamdanceiso78 was obtained from Dr Mark Tanouye(Department of Environmental Science Policy and Management andDepartment of Molecular and Cell Biology University of CaliforniaBerkeley California USA) Easily-shocked2F was obtained from Dr KevinOrsquoDell RRa-Gal4 is expressed in only the aCC and RP2 motoneurons (Linet al 2012) We are able to discriminate between these neurons duringelectrophysiological recordings and use only the aCC neuron in this studyWe used Cha-Gal4(19B) to drive UAS-transgene expression in allcholinergic neurons which include excitatory premotor interneurons(Salvaterra and Kitamoto 2001) Pan-neuronal expression was achievedby combining elaV-Gal4 (Bloomington stock no 8760 3rd chromosomeinsert) with parabss UAS-pumRNAi was obtained from the ViennaDrosophila RNAi Center (stock no 101399) and UAS-pum is detailed inSchweers et al (2002) UAS-pum lacks NRE motifs that are present in the3prime-UTR of the endogenous pum gene All genetic crosses were maintained at25degC with the exception of overexpression of pum (larvae die as 1st or 2ndinstars) These experiments were maintained at 205degC Chemical-inducedseizure was achieved by raising WT larvae on food containing 025 mgmlPTX (P1675 Sigma Poole UK) until wall-climbing third instarabbreviated to L3 (Lin et al 2015)
Library construction and RNA sequencingCNSs were removed from 50 L3 (mixed sexes) and RNA extracted using theRNeasy mini kit (QIAGEN Hilden Germany) as described (Lin et al2015) RNA integrity and purity were determined using an Agilent 2200TapeStation system (Agilent Technologies Santa Clara CA) The RNA-sequencing library was created using an mRNA Seq library preparation kitas per manufacturerrsquos instructions (Illumina Inc San Diego CA) Thelibrary products were sequenced in paired-end reads using an IlluminaHiSeqTM 2000 RNA-sequencing data were analysed using edgeR(empirical analysis of digital gene expression in R) (Robinson et al2010) This analysis identified genes with altered levels of expression usinga threshold false discovery rate (FDR)le1 GO terms for BiologicalProcess Cellular Component Molecular Function and Kyoto Encyclopediaof Genes and Genomes (KEGG) pathway were used for annotations Weclassified differentially expressed genes using the Functional AnnotationCluster (FAC) tool available in the Database for Annotation Visualizationand Integrated Discovery (DAVID) (Huang et al 2009ab)
Validation of RNA-sequencing analysis by quantitative PCRQuantitative RT-PCR was performed using a SYBR Green I real-time PCRmethod (Roche LightCyclerreg 480 SYBR Green I Master MannheimGermany) as described in Lin et al (2015) RNAwas extracted from either 20adult heads (3 days old) or 20 L3 CNSs (mixed sexes) using the RNeasymicro kit (QIAGEN) Primer sequences (5prime to 3prime) used were actin-5C
148
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
compounds that influence Pum activity andor expression weconstructed a luciferase reporter of Pum activity and screened acomprehensive library of approved compounds From 785compounds we identify 12 that potentiate Pum activity Furtheranalysis of one of these compounds avobenzone shows that itincreases transcription of pum reduces INaP in identifiedmotoneurons and is potently anticonvulsive in Drosophila
RESULTSRNA-sequencing identifies pum as downregulated in seizureIn order to determine changes to gene transcription that occur inseizure-prone CNSs we used RNA sequencing (RNA-seq) tocompare gene transcription in the CNS in two models of seizure agenetic model (parabss) and a chemical model (PTX) Using RNAextracted from the CNS of third instar larvae (L3) we identifiedtranscriptional change in 2246 and 1013 genes respectively using anFDRle1 in WT versus parabss and WT versus WT fed PTX (seeTables S1 S2) Comparison between data sets revealed that 743common genes exhibit significant change to expression (Fig 1Ainset see Table S3 for gene details) Of these 736 showed significantand consistent altered expression in both seizure models A log2 plotof fold-change (log2FC) showed that 339 (46) are significantlyupregulated and 397 (54) are significantly downregulated(P=0001 ANOVA) The remaining seven genes did not showconsistent direction of change (Fig 1A) Identified genes generated a
total of 130 functional clusters representing a wide array of functionsincluding predicted genes encoding ion channels and synapticproteins (detailed below) The top 20 enriched clusters are shown inFig S1 The top four clusters are for genes associated with pre-replicative complex assembly eukaryotic translation elongationfactor 1 complex negative regulation of neuroblast proliferationand translation repressor activity Genes associated with translationalrepression include minichromosome maintenance (orthologues 2 35 7) elongation factor 1α100E 1α48D and 1β anachronismprospero musashi embryonic lethal abnormal vision brain tumorand pum (Table S3) Twenty genes that we identify have beenpositively associated with human epilepsy (httpwwwinformaticsjaxorghumanDiseaseshtml) (red dots in Fig 1A and described inTable 1) Of these genes five were upregulated and 15 weredownregulated in the Drosophila seizure models These genesinclude paralytic (Nav) nicotinic Acetylcholine Receptor α5 Ihchannel and Shaker (K+ channels) in addition to Syntaxin Synapsinand unc-13 (synaptic proteins) Seven genes were identified that showparticularly large increases in transcription (gt3 log2FC blue dots inFig 1A) These genes are CG18331 (mucin 68Ca) CG34076(mitochondrial NADH-ubiquione oxidoreducatse chain 3)CG11205 (photorepair) CR41620CR40734 (rRNA genes) andCG7606CG32198 (unknowns)
Our attention was drawn to pum which was significantlydownregulated in both seizure models [meanplusmnsd WT 602plusmn14
Fig 1 Analysis of altered gene transcription in seizure models (A) Cross-comparison shows 743 changes are common to both seizure models Analysis ofdirection of log2 fold-change (log2FC) in transcription of the 743 common genes (main figure) shows that 339 are significantly (two-way ANOVA) upregulatedand 397 downregulated Seven genes show differential expression in the two models pumilio (pum) which is downregulated is identified by the orange dotGenes previously linked to human epilepsy are shown by red dots (described in Table 1) Blue dots highlight genes that show particularly large fold-changes(log2FCgt3) in expression levels in the seizure backgrounds (see Results text for identity) Inset analysis of the transcriptome by RNA-sequencing shows changeto transcription of 2246 genes in the parabss CNS compared with wild type (WT) Comparison of WT with WT fed picrotoxin (PTX) shows 1013 transcriptionalchanges (B) Analysis of pum transcript level in isolatedCNS from L3 shows a significant reduction inWT+PTX and parabss comparedwithWT controls (C) pum issignificantly reduced in adult heads in both seizure models compared with WT controls TheWT value has been set to 1 in each experimental condition Data aremeanplusmnsd for n=5 independent samples Ple005 Ple0001 (unpaired t-test)
142
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
vs WT fed PTX 405plusmn2 and parabss 381plusmn15 cpm (counts permillion) P=15times10minus5 n=3] This is because its homologue PUM2has been reported to be downregulated in humans sufferingtemporal lobe epilepsy (Wu et al 2015) Pum is a well-characterised translational repressor which we have previouslyreported regulates translation of Navs in both Drosophila and rat toachieve homeostatic control of neuron action potential firing(Driscoll et al 2013 Mee et al 2004 Muraro et al 2008) Weconsidered that manipulation of a homeostatic regulator mightrepresent a promising approach to control seizure To validateRNA-seq data we undertook RT-qPCR pum was significantlydecreased in WT larval CNS after exposure to PTX (083plusmn003)and in parabss (078plusmn003) compared with WT control (set as 1P=003 n=5 Fig 1B) We observed a similar and significantdownregulation of pum transcription in adult heads that containmostly brain tissue (WT fed PTX 041plusmn007 and parabss 058plusmn017) relative to WT control (set as 1 P=00002 n=5 Fig 1C)In addition we used RT-qPCR to validate the identification ofthe 20 genes that have been positively associated with epilepsy(red dots in Fig 1A) This validation was undertaken only for theparabss background We found consistent change for 14 of thegenes (representing a validation rate of 70) Two genes showedsignificant change by RT-qPCR but in the opposite direction toRNA-seq whereas four genes showed no significant change (seeTable 1)
Upregulation of Pum is anticonvulsiveWe have shown that Pum binds to mRNA encoding Navs in bothDrosophila and rat Binding subsequently reduces the density ofNav channels available in the neuronal membrane (Driscoll et al2013 Mee et al 2004 Muraro et al 2008) We predicted
therefore that maintaining pum expression in seizure backgroundswould be anticonvulsant Inducing seizure by vortexing of parabssYmale flies resulted in a recovery time (RT 114plusmn133 s n=3) that wassignificantly reduced by exposure to recognised AEDs (Parker et al2011) Vortexing WT flies by comparison resulted in a nearinstantaneous RT (53plusmn25 s n=3) This is the averaged time takenfor all flies in the vial (n=10) to regain a standing posture followingvortexing and does not imply that WT flies exhibit seizures Bycontrast expressing pum in a Cha-Gal4(19B) cholinergic neurondriver line (which are the predominant excitatory interneuron type inthe insect CNS Yasuyama and Salvaterra 1999) in parabss
( parabssY Cha-Gal4(19B)UAS-pum) flies significantly reducedseizure RT compared with control parabssY Cha-Gal4(19B)+ (7plusmn36 s vs 114plusmn133 s P=12times10minus5 n=3 Fig 2A) Indeed recoverytime following upregulation of pum was not significantly differentto WT controls (53plusmn25 s) indicative that seizures were completelysuppressed By contrast expression of pumRNAi using the sameCha-Gal4(19B) driver in the parabss background was stronglyproconvulsive (206plusmn304 s vs 114plusmn133 s P=00002 n=3Fig 2A) We observed the same outcome in L3 where seizurebehaviour was induced by electroshock (Fig 2B) Seizure RT wassignificantly reduced (134plusmn101 s P=0005 n=18) or increased(557plusmn255 s P=00004 n=20) by expression of either UAS-pum orUAS-pumRNAi respectively in Cha-Gal4(19B) cholinergic neuronsin the parabss background (control parabssY Cha-Gal4(19B)+324plusmn159 s n=20) Manipulation of pum levels pan-neuronally(using parabss elaV-Gal4) resulted in an identical effect toelectroshock-induced seizure in L3 (Fig 2C) Increasing pumexpression reduced seizure duration (133plusmn70 s P=0009 n=21) andRNAi-mediated knockdown increased seizure duration (255plusmn99 sP=002 n=42) compared with control (204plusmn88 s n=39)
Table 1 Identification of epilepsy-associated genes
CG number Drosophila geneparabss fold-change (log2)RNA-seqqPCR
Mouse and human homologue data from Mouse Genome Informatics HumanndashMouse Disease Connection database (httpwwwinformaticsjaxorghumanDiseaseshtml) + increased minus decreased mRNA levels compared with respective controls All numerical values shown are significant at Plt005 nsvalues not significantly different from control
143
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Manipulation of pum in a WT background resulted in a differentoutcome Both RNAi-mediated knockdown and particularlyoverexpression of pum resulted in an induction of a seizurephenotype [Cha-Gal4(19B)UAS-pumRNAi 187plusmn109 s P=18times10minus5n=23 Gal4(19B)UAS-pum 387plusmn77 s P=28times10minus26 n=20]compared with control [Cha-Gal4(19B)+ 97plusmn43 n=40Fig 2D] This paradoxical result is similar to the effect offeedingWTDrosophilaAEDs such as phenytoin that also result inseizure induction an effect that has also been observed in rat(Callaghan and Schwark 1980 Marley and Baines 2011Rundfeldt et al 1990)
Increased pum expression decreases INaP in motoneuronsOur previous work has shown that Pum regulates INa throughtranslational regulation of para (Mee et al 2004 Muraro et al2008) We recorded from parabssY L3 where the expression oftransgenic pum was selectively manipulated in only the aCC
motoneuron (using RRa-Gal4) Our choice to use this motoneuronis guided by the ability to combine genetics and electrophysiologya selective Gal4 driver exists to express UAS-transgenes in thisneuron which is also accessible to patch electrodes That INaP isgreater in amplitude in aCC motoneurons in seizure mutants(Marley and Baines 2011) is indicative that they share propertieswith central interneurons in human epilepsy which can also showincreased INaP (Stafstrom 2007)
Increased expression of pum in L3 parabss aCC resulted in astriking reduction of INaP (44plusmn41 pApF vs 126plusmn40 pApFP=49times10minus5 Fig 3ABD) but no change to INaT (Fig 3E)Analysis of the persistent-to-transient current ratio (PT) recorded inL3 aCC showed a marked reduction (200plusmn180 vs 510plusmn119P=50times10minus5 Fig 3F) A high PT ratio (gt40) in centralmotoneurons has been previously shown to be characteristic ofDrosophila seizure mutants and its reduction to be anticonvulsant(Lin et al 2015 Marley and Baines 2011) Thus we conclude thatupregulation of pum is anticonvulsant which is due at leastpartially to its ability to reduce INaP
RNAi-mediated downregulation of pum in L3 parabss aCCincreased INaT (313plusmn33 pApF vs 247plusmn45 pApF P=0005) butdid not affect INaP or the PT ratio (Fig 3C-F) Analysis of the effecton seizure behaviour following this more selective manipulation ofpum expression showed no significant differences to controls( parabssY RRa-Gal4+ data not shown) This is entirely expectedgiven the highly selective cell targeting used in these experimentsHowever a more widespread manipulation of pum [eg using Cha-Gal4(19B)] which is sufficient to alter seizure duration andorseverity probably acts via an identical mechanism throughalteration of INa
Increasing pum expression in aCC in a WT background resultedin essentially the same changes to INa as seen with manipulation inthe parabss background INaP was significantly reduced (24plusmn17 pApF vs 74plusmn49 pApF P=00028 Fig 3G) but no change toINaT was observed (188plusmn48 pApF vs 219plusmn27 pApF Fig 3H)By contrast downregulation using pumRNAi produced a differentoutcome compared with parabss INaP was significantly increased(110plusmn24 pApF vs 74plusmn49 pApF P=0032 Fig 3G) with noeffect on INaT (246plusmn47 pApF vs 219plusmn27 pApF Fig 3H)Analysis of the PT ratio however similarly only showed asignificant reduction following upregulation of pum expression inWT (147plusmn119 vs 333plusmn202 P=0016 Fig 3I)
On occasion we noted the appearance of multiple resurgent INaduring the INaP plateau in the para
bss background (Fig 4A indicatedby arrow) Moreover we observed a significant correlation betweenthe occurrence of resurgent INa and pum level (P=0002 Chi-squaretest Fig 4B) Thus resurgent INa was most often observedfollowing RNAi-knockdown and only rarely following expressionof pum The origin of these currents remains uncertain Analysis ofvoltage recordings (Fig 4A) showed no obvious issue of spaceclamp which suggests these currents are not occurring in distalunclamped regions of the neuron The averaged frequency of theresurgent currents was sim100 Hz which did not vary with level ofpum expression (Fig 4C) Resurgent currents are particularlyevident at holding potentials between minus50 to minus20 mV and exhibithighest frequency at minus30 mV (RRa-Gal4UAS-pumRNAi 10450plusmn3678 Hz RRa-Gal4+ 12000plusmn2016 Hz RRa-Gal4UAS-pum11500plusmn4093 Hz) Increased resurgent INa probably supportsincreased action potential firing consistent with our observationthat RNAi-mediated knockdown of pum is proconvulsant (Griecoet al 2005) Resurgent INa is only rarely observed (lt5) in WTaCC recordings (data not shown)
Fig 2 Expression of transgenic pum is anticonvulsive (A) Expression oftransgenic full-length pum lacking NRE motifs (UAS) in cholinergic neurons inparabss [Cha-Gal4(19B)gtpum] is sufficient to reduce recovery time (RT) frommechanical shock-induced seizure in adult flies compared with parabss alone(CTRL) By contrast further reduction of pum through RNAi-mediatedknockdown (RNAi) [Cha-Gal4(19B)gtRNAi] significantly lengthens seizure RTEach manipulation tested 10 flies per vial to produce an average value Thiswas repeated in triplicate and a final average calculated Data are meansplusmnsdn=3 (B) Identical manipulation of pum expression in cholinergic neurons in L3parabss had an identical effect on seizure duration when seizurewas evoked byelectroshock (C) Pan-neuronal manipulation of pum is also sufficient to affectelectroshock-induced seizure in L3 Upregulation (UAS parabss elaV-Gal4gtpum) reduces seizure and downregulation (RNAi parabss elaV-Gal4gtRNAi) increases seizure duration compared with control (parabss elaV-Gal4+) (D) Up- or downregulation of pum in aWT background using the Cha-Gal4(19B) cholinergic driver line results in induction of a seizure phenotypeData are meansplusmnsd n is stated in individual bars Ple005 Ple001Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
144
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
A screen to identify positive regulators of Pum activityUpregulation of Pum activity either through increased transcriptionor post-transcriptional modification might provide an effectivemeans to suppress seizures To identify possible lead compoundswith this mode of action we constructed a luciferase-based reporterof Pum activity for use in an in vitro S2R+ cell line suited to large-
scale screens (Lin et al 2015) Overexpression of pum is sufficientto repress luciferase activity (due to translational repression)whereas incubation with pum double-stranded RNA is sufficient toincrease luciferase activity by reducing endogenous Pum activityPCR analysis shows that pum is endogenously expressed in S2R+cells (Fig S2) Thus activity of the firefly-luciferase-NRE reporter
Fig 3 Expression of transgenic pum reduces INaP (A-C) Whole-cell patch recordings of INa from L3 aCC motoneurons in parabss (CTRL) parabss expressingtransgenic pum (UAS) or pumRNAi (RNAi) Transgene expression is limited to aCC motoneurons in these manipulations using RRa-Gal4 (DE) Expression oftransgenic pum (UAS) is sufficient to reduce the magnitude of INaP without change to INaT Expression of pumRNAi (RNAi) results in no change to INaP but asignificant increase in INaT (F) Persistent-to-transient (PT) current ratio for INa recorded in DE (GH) The effect of manipulating pum in a WT backgroundIncreasing expression (UAS) is sufficient to reduce INaP with no change to INaT whereas reduction (RNAi) increases INaP amplitude but has no effect on INaT(I) Analysis of the PT ratio in individual cells recorded in GH shows increased pum is sufficient to reduce the ratio Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
Fig 4 Occurrence of resurgent INa is related to level of pum (A) Resurgent INa (INaR arrow) is seen superimposed on repolarization of holding potential usedto evoke INaP Analysis of the voltage trace (lower trace) shows good control during this step (B) The occurrence of INaR in the parabss background is highestwhen pum is reduced (RNAi 82 14 from 17 cells) and lowest when increased (UAS 21 3 from 14 cells) Control (CTRL parabss 64 9 from 14 cells)Transgene expression was limited to aCC cells using RRa-Gal4 (C) Frequency of INaR oscillations is unaffected by expression level of pum Data are meansplusmnsd
145
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(FF-NRE) reflects the absolute level of Pum function in these cellsA second reporter which lacked an NRE-motif was also transfected[renilla (Ren)-luciferase] to allow detrimental effects to cellviability to be determined The final readout of the assay was aFFRen luciferase ratio that would be reduced followingupregulation of Pum activityWe screened 785 compounds from a repurposed library (see
Materials and Methods drugs screened are listed in Table S4) Weidentified 12 compounds that significantly reduced the FFRen ratioat 5 μM (Table 2) Based on structure andor known drug target thecompounds fall into one of four groupings those containing amethoxybenzaldehyde moiety (aniracteam and avobenzone) anti-cancer agents (cladribine gemcitabine floxuridine clofarabinebleomycin and docetaxel) mTOR inhibitors (temsirolimus andrapamycin) and topoisomerase II inhibitors (mitoxantrone andteniposide) Our attention was particularly drawn to avobenzonebecause unlike the other compounds it had no significant effect ontranscription of the control Ren-luciferase reporter (all othercompounds also reduced expression of this reporter in addition todecreasing the FFRen ratio) Thus we took avobenzone forward forfurther testing
Avobenzone potentiates activity of PumWe first tested for anticonvulsant activity in L3 parabss mutantsLarvae raised in food containing avobenzone (04 mgml) showedsignificantly reduced RT in response to electroshock (avobenzone213plusmn124 s n=40 vs control 339plusmn83 s n=20 P=00004 Fig 5A)Similarly exposure of adult parabss flies to avobenzone (04 mgml)24 h before testing also resulted in significant reduction of seizureduration (avobenzone 61plusmn29 vs control 138plusmn29 s n=5 P=00002Fig 5B) Next we recorded INa from parabss aCC in L3 that hadbeen raised on food containing different concentrations ofavobenzone (01-04 mgml Fig 5C-G) Avobenzone reducedINaP from 139plusmn76 pApF in controls to 76plusmn62 pApF at 01 mgml(P=017) 54plusmn64 pApF at 02 mgml (P=003) and 35plusmn42 pApF at04 mgml (P=0002) (Fig 5D) Conversely avobenzone treatment atthese concentrations did not induce any detectable effect in INaT(Fig 5E) Analysis of the PT ratio for INa shows that exposure toavobenzone significantly reduced this value from493plusmn92 in control to 280plusmn232 at 01 mgml (P=009) 219plusmn269 at 02 mgml (P=003) and 121plusmn132 at 04 mgml(P=00004) (Fig 5F) which compares favourably with
overexpression of pum (cf Fig 3) We also observed a significantcorrelation between avobenzone concentration and the occurrence ofresurgent INa (P=0005 Chi-square test Fig 5G)
Our predicted mode of action for avobenzone is inconsistent withan immediate effect of this compound acting instead to potentiatePum which in turn downregulates Nav channels in the neuronalmembrane To test this we recorded from non-drug-exposed L3parabss aCC and used bath application of avobenzone (5 microM) Nochanges were observed in either component of INa (data not shown)and the PT ratio remained unaffected (Fig 5H) Higher doses(20 microM) or longer exposure times (10 min) similarly produced nodetectable effect (data not shown) This lack of acute effect isconsistent with our predicted mode of action Finally to directly testthis prediction we measured pum transcript abundance in parabss
L3 grown in the presence of avobenzoneWe observed a modest butstatistically significant increase in transcript abundance of sim20(12plusmn017 n=5 P=004 t-test vehicle control set as 1 Fig 5I)Thus we conclude that avobenzone acting to increase thetranscription andor transcript stability of pum is able to suppressseizure duration through downregulation of INaP Finally weobserved equally potent anticonvulsive activity of avobenzone intwo other bang-sensitive mutants easily-shocked (avobenzone142plusmn82 vs control 240plusmn120 s n=40 P=10times10minus5 L3electroshock) encoding an ethanolamine kinase (Pavlidis et al1994) and slamdance (avobenzone 178plusmn122 vs control 272plusmn108 s n=40 P=68times10minus5 L3 electroshock) encoding anaminopeptidase (Zhang et al 2002) indicative that increasingPum activity might be effective against a broad range of epilepsies
DISCUSSIONThe causes of seizure even in genetic epilepsies vary greatly andare not confined to genes with obvious contributions to ion fluxacross neuronal membranes This increases the challenge to identifyindividual mutations to determine the physiological role of both theWT and mutated protein and ultimately to design drugs tominimise the unwanted effect of the mutation In this study weidentify transcriptional changes that occur in the seizure-proneCNS We identify over 700 common genes that show alteredtranscription in two different seizure models It is noteworthy thatwe observed approximately double the number of genes showingaltered transcription in parabss flies compared with those treatedwith PTX The reason for this is unclear but might representaccumulated compensatory changes in the mutant line that haveoccurred in order to lessen the severity of seizure activity in parabss
mutants These additional genes warrant further investigation aspotential seizure suppressors
Many of the common transcriptional changes we identify and inparticular those that are upregulated (and thus open to inhibition bydrug exposure) might provide effective drug targets for novel AEDdesign However our attention was drawn to Pum which we havepreviously shown orchestrates homeostasis of action potential firingin both Drosophila and rat central neurons (Driscoll et al 2013Mee et al 2004) The degree of seizure suppression achieved byupregulating Pum in parabss flies is considerable and is onlymatched by the no-action-potential (napts) allele of the maleless(mle) locus in Drosophila which encodes an ATP-dependentdouble-stranded RNA (dsRNA) helicase (Ganetzky andWu 1982)This mutation causes a catastrophic change in splicing of theDrosophilaNav (Reenan et al 2000) The net effect of both of thesemanipulations increased Pum or the presence of napts is to reducethe availability of functional Nav expressed in central neurons Thedirection of change of pum in the two seizure models (that show
Table 2 List of compounds that reduce the fireflyRenilla (FFRen)luciferase ratio thus mimicking the activity of increased pumexpression (shown at bottom of table for reference)
All but avobenzone also reduce expression of the control Ren luciferasereporter that does not contain an NRE motif Luciferase values shown arenormalised such that 10 would represent no effect
146
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
reduced expression) might not be ideal with respect to drugdevelopment given that disruption of a gene or protein is often moreachievable Nevertheless we show that upregulation of pum in aDrosophila seizure mutant is potently anticonvulsive and furtherwe identify a potential lead anticonvulsive compound thatseemingly increases the level of expression of this homeostaticregulator This compound might catalyse the development of anovel class of AEDNeurons display an array of homeostatic mechanisms to maintain
action potential firing within pre-determined and physiologicallyappropriate limits (Davis 2013) Pum is a well-characterised RNA-binding protein that binds mRNA usually through a specific motiftermed the NRE Once bound Pum recruits additional cofactorsincluding Nanos and Brain tumor (Brat) to form a complex that issufficient to prevent translation (Wharton et al 1998) Our results inthis study indicate that increased expression of Pum might havetherapeutic benefit for seizure suppression However a potential
issue in this regard is that a genome-wide identification of RNAsbound to Pum in ovaries identifies upwards of 700 genes(FDRlt01) (Gerber et al 2006) This raises the problem ofspecificity of effect following global potentiation of level or activityof Pum This potential issue might however be overcome throughidentifying and targeting neuronal-specific regulators of Pum Onesuch alternative target might be the inhibition of Myocyte enhancerfactor 2 (Mef2)-induced expression of miR-134 in neurons that inturn inhibits translation of mammalian PUM2 (Fiore et al 2009)Additional possibilities include targeting of cofactors required forPum activity It is interesting in this regard that a loss-of-functionmutation in mei-P26 a homologue of Brat produces strong seizuresuppression in Drosophila bang-sensitive seizure mutants(Glasscock et al 2005)
Mammalian PUM2 binds transcripts encoding SCN1A (Nav11)and SCN8A (Nav16) (Driscoll et al 2013 Vessey et al 2010) Areduction in supply of Nav protein to the neuron membrane is
Fig 5 Avobenzone is anticonvulsant and selectively reduces INaP (A) parabss L3 raised in food containing 04 mgml avobenzone show significantly reducedrecovery time (RT) following electroshock compared with controls (CTRL parabss+DMSO) (B) Exposure of adult parabss flies to avobenzone (04 mgml) is alsopotently anticonvulsant compared with controls (CTRL parabss+DMSO) Each manipulation tested 10 flies per vial to produce an average value This wasrepeated five times and a final average calculated (C) Whole-cell patch recordings of INa from parabss L3 aCC raised in food containing 04 mgml avobenzoneshow reduced INaP (DE) Increasing concentrations of avobenzone (01 02 and 04 mgml) induced a proportional decrease of INaP (D) without affectingINaT (E) (F) Persistent-to-transient (PT) current ratio for INa recorded in aCC (G) The frequency of cells that exhibit resurgent INa correlates with avobenzoneconcentration (P=0005 Chi-square test) (H) PT ratio measured from parabss aCC before (CTRL) and after a 1 min bath application of 5 microM avobenzone(I) Analysis of pum transcript level in isolated CNS from parabss L3 raised on food containing avobenzone (04 mgml) shows a significant increase compared withparabss L3 raised on food containing an equal amount of vehicle (08 DMSO) The control value has been set to 1 Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (AH-I unpaired t-test D-F two-way ANOVA with Bonferronirsquos post hoc)
147
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
consistent with a reduction in action potential firing and a generalanticonvulsant effect (Mee et al 2004) Analysis of INa inmotoneurons indicates that a likely mechanism includes a markedreduction in INaP Increased INaP is associated with mutations inSCN1A that have been identified from individuals with epilepsy(Meisler and Kearney 2005) and is specifically reduced by AEDssuch as phenytoin valproate and lamotrigine (Stafstrom 2007) Inlight of this the anticonvulsant effect of increased pum expression isunderstandable That reducing pum expression through RNAi-mediated knockdown is proconvulsive is again both predictable andunderstandable However the effect of this manipulation on INa isnot so clear Rather than increasing INaP INaT is instead significantlyincreased together with a novel appearance of resurgent INa duringrepolarisation Increased INaT would be expected to reduce thethreshold for action potential firing (ie making firing more likely)whereas resurgent INa is associated with increased action potentialfiring frequency partly by reducing the refractory period (Griecoet al 2005) Although we have observed this current component inrecordings from seizure mutants (including parabss) it is rarelyobserved in WT or following expression of transgenic pumThe ability to manipulate Pum in vivo to determine its
anticonvulsive properties in rodent models of seizure will begreatly aided by the identification of chemical compounds thatdirectly potentiate either expression or activity state We report theuse of a suitable cell-based screen to identify such compounds andhighlight avobenzone as a potential lead compound for futuredevelopment The in vivo toxicity of avobenzone has not been wellestablished and although there are few reports of serious side effectsassociated with its use as an active ingredient of sunscreen itstendency to form free radicals might be a potential issue To ourknowledge this compound has not been used to treat neurologicaldisease and its mode of action in reducing seizure in Drosophilaremains to be determined Our observations that ingestion ofavobenzone result in increased expression of pum is indicative thatthis compound might mimic elements of the pathway that controlexpression of this homeostatic regulatorThe output of our screen also provides additional support for the
use of rapamycin to control seizure (Lasarge and Danzer 2014Russo et al 2013) indicative that this molecule might influenceneuronal homeostasis The identification of topoisomerase II as apotential target to control seizure also validates previousobservations reporting that inhibition of this class of nuclearprotein is anticonvulsant (Lin et al 2015 Song et al 2008)Finally that we identify that the increase in Pum activity byaniracetam might hint at an additional mode of action for this classof known anticonvulsants (Shiotani et al 2000) The relatedracetams levetiracetam and brivaracetam are currently in clinicaluse as AEDs exploiting their capability to bind and inhibit synapticvesicle protein 2A (SV2A) (Klitgaard et al 2016)In summary we present a description of transcriptional change
present in seizure-prone CNS We identify in particular that pumexpression is downregulated in both genetic and chemically inducedseizure models This mirrors the reported reduction in PUM2 inhuman TLE and in rats exposed to the proconvulsant pilocarpine(Wu et al 2015) It also provides a possible understanding for whyPum2 null mice exhibit spontaneous seizures (Siemen et al 2011)However it is perplexing that pum levels should decrease duringseizures given that the published model predicts an increase (Meeet al 2004) As reduced Pum levels are predicted to increaseneuronal excitability it seems that epileptic seizures are associatedwith a pathological dysregulation of pum expression We speculatethat this occurs because Pum can auto-regulate (the pum transcript
contains NRE motifs) Thus although the neuronal hyperactivityinduced by seizures will initially increase Pum expression theaccumulating Pum protein might feed back to downregulate its owntranscript (Gerber et al 2006) Sampling at later stages after seizureoccurrence might only report reduced Pum compared with non-seizure controls Indeed we have shown that upregulation of pum inthe Drosophila CNS through expression of a wild-type transgene(lacking NRE motifs) results in reduction of endogenous pumtranscript level (W-HL and RAB unpublished data) Preventionof this feedback achievable in this study through expression oftransgenic pum lacking an NRE or exposure to avobenzone holdssignificant promise for anticonvulsant therapy
MATERIALS AND METHODSFly stocksWild type (WT maintained in the Baines lab) was Canton-S parabss (bss1)which was obtained from Dr Kevin OrsquoDell (Institute of Molecular Cell andSystems Biology University of Glasgow UK) is detailed in Parker et al(2011) The parabss stock (and other transgenic lines used) were notbackcrossed to the CS stock Controls consisted of either untreated parabss
andor parental stocks (ie Gal4+ UAS+) and are stated in respectivefigure legends Slamdanceiso78 was obtained from Dr Mark Tanouye(Department of Environmental Science Policy and Management andDepartment of Molecular and Cell Biology University of CaliforniaBerkeley California USA) Easily-shocked2F was obtained from Dr KevinOrsquoDell RRa-Gal4 is expressed in only the aCC and RP2 motoneurons (Linet al 2012) We are able to discriminate between these neurons duringelectrophysiological recordings and use only the aCC neuron in this studyWe used Cha-Gal4(19B) to drive UAS-transgene expression in allcholinergic neurons which include excitatory premotor interneurons(Salvaterra and Kitamoto 2001) Pan-neuronal expression was achievedby combining elaV-Gal4 (Bloomington stock no 8760 3rd chromosomeinsert) with parabss UAS-pumRNAi was obtained from the ViennaDrosophila RNAi Center (stock no 101399) and UAS-pum is detailed inSchweers et al (2002) UAS-pum lacks NRE motifs that are present in the3prime-UTR of the endogenous pum gene All genetic crosses were maintained at25degC with the exception of overexpression of pum (larvae die as 1st or 2ndinstars) These experiments were maintained at 205degC Chemical-inducedseizure was achieved by raising WT larvae on food containing 025 mgmlPTX (P1675 Sigma Poole UK) until wall-climbing third instarabbreviated to L3 (Lin et al 2015)
Library construction and RNA sequencingCNSs were removed from 50 L3 (mixed sexes) and RNA extracted using theRNeasy mini kit (QIAGEN Hilden Germany) as described (Lin et al2015) RNA integrity and purity were determined using an Agilent 2200TapeStation system (Agilent Technologies Santa Clara CA) The RNA-sequencing library was created using an mRNA Seq library preparation kitas per manufacturerrsquos instructions (Illumina Inc San Diego CA) Thelibrary products were sequenced in paired-end reads using an IlluminaHiSeqTM 2000 RNA-sequencing data were analysed using edgeR(empirical analysis of digital gene expression in R) (Robinson et al2010) This analysis identified genes with altered levels of expression usinga threshold false discovery rate (FDR)le1 GO terms for BiologicalProcess Cellular Component Molecular Function and Kyoto Encyclopediaof Genes and Genomes (KEGG) pathway were used for annotations Weclassified differentially expressed genes using the Functional AnnotationCluster (FAC) tool available in the Database for Annotation Visualizationand Integrated Discovery (DAVID) (Huang et al 2009ab)
Validation of RNA-sequencing analysis by quantitative PCRQuantitative RT-PCR was performed using a SYBR Green I real-time PCRmethod (Roche LightCyclerreg 480 SYBR Green I Master MannheimGermany) as described in Lin et al (2015) RNAwas extracted from either 20adult heads (3 days old) or 20 L3 CNSs (mixed sexes) using the RNeasymicro kit (QIAGEN) Primer sequences (5prime to 3prime) used were actin-5C
148
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
vs WT fed PTX 405plusmn2 and parabss 381plusmn15 cpm (counts permillion) P=15times10minus5 n=3] This is because its homologue PUM2has been reported to be downregulated in humans sufferingtemporal lobe epilepsy (Wu et al 2015) Pum is a well-characterised translational repressor which we have previouslyreported regulates translation of Navs in both Drosophila and rat toachieve homeostatic control of neuron action potential firing(Driscoll et al 2013 Mee et al 2004 Muraro et al 2008) Weconsidered that manipulation of a homeostatic regulator mightrepresent a promising approach to control seizure To validateRNA-seq data we undertook RT-qPCR pum was significantlydecreased in WT larval CNS after exposure to PTX (083plusmn003)and in parabss (078plusmn003) compared with WT control (set as 1P=003 n=5 Fig 1B) We observed a similar and significantdownregulation of pum transcription in adult heads that containmostly brain tissue (WT fed PTX 041plusmn007 and parabss 058plusmn017) relative to WT control (set as 1 P=00002 n=5 Fig 1C)In addition we used RT-qPCR to validate the identification ofthe 20 genes that have been positively associated with epilepsy(red dots in Fig 1A) This validation was undertaken only for theparabss background We found consistent change for 14 of thegenes (representing a validation rate of 70) Two genes showedsignificant change by RT-qPCR but in the opposite direction toRNA-seq whereas four genes showed no significant change (seeTable 1)
Upregulation of Pum is anticonvulsiveWe have shown that Pum binds to mRNA encoding Navs in bothDrosophila and rat Binding subsequently reduces the density ofNav channels available in the neuronal membrane (Driscoll et al2013 Mee et al 2004 Muraro et al 2008) We predicted
therefore that maintaining pum expression in seizure backgroundswould be anticonvulsant Inducing seizure by vortexing of parabssYmale flies resulted in a recovery time (RT 114plusmn133 s n=3) that wassignificantly reduced by exposure to recognised AEDs (Parker et al2011) Vortexing WT flies by comparison resulted in a nearinstantaneous RT (53plusmn25 s n=3) This is the averaged time takenfor all flies in the vial (n=10) to regain a standing posture followingvortexing and does not imply that WT flies exhibit seizures Bycontrast expressing pum in a Cha-Gal4(19B) cholinergic neurondriver line (which are the predominant excitatory interneuron type inthe insect CNS Yasuyama and Salvaterra 1999) in parabss
( parabssY Cha-Gal4(19B)UAS-pum) flies significantly reducedseizure RT compared with control parabssY Cha-Gal4(19B)+ (7plusmn36 s vs 114plusmn133 s P=12times10minus5 n=3 Fig 2A) Indeed recoverytime following upregulation of pum was not significantly differentto WT controls (53plusmn25 s) indicative that seizures were completelysuppressed By contrast expression of pumRNAi using the sameCha-Gal4(19B) driver in the parabss background was stronglyproconvulsive (206plusmn304 s vs 114plusmn133 s P=00002 n=3Fig 2A) We observed the same outcome in L3 where seizurebehaviour was induced by electroshock (Fig 2B) Seizure RT wassignificantly reduced (134plusmn101 s P=0005 n=18) or increased(557plusmn255 s P=00004 n=20) by expression of either UAS-pum orUAS-pumRNAi respectively in Cha-Gal4(19B) cholinergic neuronsin the parabss background (control parabssY Cha-Gal4(19B)+324plusmn159 s n=20) Manipulation of pum levels pan-neuronally(using parabss elaV-Gal4) resulted in an identical effect toelectroshock-induced seizure in L3 (Fig 2C) Increasing pumexpression reduced seizure duration (133plusmn70 s P=0009 n=21) andRNAi-mediated knockdown increased seizure duration (255plusmn99 sP=002 n=42) compared with control (204plusmn88 s n=39)
Table 1 Identification of epilepsy-associated genes
CG number Drosophila geneparabss fold-change (log2)RNA-seqqPCR
Mouse and human homologue data from Mouse Genome Informatics HumanndashMouse Disease Connection database (httpwwwinformaticsjaxorghumanDiseaseshtml) + increased minus decreased mRNA levels compared with respective controls All numerical values shown are significant at Plt005 nsvalues not significantly different from control
143
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Manipulation of pum in a WT background resulted in a differentoutcome Both RNAi-mediated knockdown and particularlyoverexpression of pum resulted in an induction of a seizurephenotype [Cha-Gal4(19B)UAS-pumRNAi 187plusmn109 s P=18times10minus5n=23 Gal4(19B)UAS-pum 387plusmn77 s P=28times10minus26 n=20]compared with control [Cha-Gal4(19B)+ 97plusmn43 n=40Fig 2D] This paradoxical result is similar to the effect offeedingWTDrosophilaAEDs such as phenytoin that also result inseizure induction an effect that has also been observed in rat(Callaghan and Schwark 1980 Marley and Baines 2011Rundfeldt et al 1990)
Increased pum expression decreases INaP in motoneuronsOur previous work has shown that Pum regulates INa throughtranslational regulation of para (Mee et al 2004 Muraro et al2008) We recorded from parabssY L3 where the expression oftransgenic pum was selectively manipulated in only the aCC
motoneuron (using RRa-Gal4) Our choice to use this motoneuronis guided by the ability to combine genetics and electrophysiologya selective Gal4 driver exists to express UAS-transgenes in thisneuron which is also accessible to patch electrodes That INaP isgreater in amplitude in aCC motoneurons in seizure mutants(Marley and Baines 2011) is indicative that they share propertieswith central interneurons in human epilepsy which can also showincreased INaP (Stafstrom 2007)
Increased expression of pum in L3 parabss aCC resulted in astriking reduction of INaP (44plusmn41 pApF vs 126plusmn40 pApFP=49times10minus5 Fig 3ABD) but no change to INaT (Fig 3E)Analysis of the persistent-to-transient current ratio (PT) recorded inL3 aCC showed a marked reduction (200plusmn180 vs 510plusmn119P=50times10minus5 Fig 3F) A high PT ratio (gt40) in centralmotoneurons has been previously shown to be characteristic ofDrosophila seizure mutants and its reduction to be anticonvulsant(Lin et al 2015 Marley and Baines 2011) Thus we conclude thatupregulation of pum is anticonvulsant which is due at leastpartially to its ability to reduce INaP
RNAi-mediated downregulation of pum in L3 parabss aCCincreased INaT (313plusmn33 pApF vs 247plusmn45 pApF P=0005) butdid not affect INaP or the PT ratio (Fig 3C-F) Analysis of the effecton seizure behaviour following this more selective manipulation ofpum expression showed no significant differences to controls( parabssY RRa-Gal4+ data not shown) This is entirely expectedgiven the highly selective cell targeting used in these experimentsHowever a more widespread manipulation of pum [eg using Cha-Gal4(19B)] which is sufficient to alter seizure duration andorseverity probably acts via an identical mechanism throughalteration of INa
Increasing pum expression in aCC in a WT background resultedin essentially the same changes to INa as seen with manipulation inthe parabss background INaP was significantly reduced (24plusmn17 pApF vs 74plusmn49 pApF P=00028 Fig 3G) but no change toINaT was observed (188plusmn48 pApF vs 219plusmn27 pApF Fig 3H)By contrast downregulation using pumRNAi produced a differentoutcome compared with parabss INaP was significantly increased(110plusmn24 pApF vs 74plusmn49 pApF P=0032 Fig 3G) with noeffect on INaT (246plusmn47 pApF vs 219plusmn27 pApF Fig 3H)Analysis of the PT ratio however similarly only showed asignificant reduction following upregulation of pum expression inWT (147plusmn119 vs 333plusmn202 P=0016 Fig 3I)
On occasion we noted the appearance of multiple resurgent INaduring the INaP plateau in the para
bss background (Fig 4A indicatedby arrow) Moreover we observed a significant correlation betweenthe occurrence of resurgent INa and pum level (P=0002 Chi-squaretest Fig 4B) Thus resurgent INa was most often observedfollowing RNAi-knockdown and only rarely following expressionof pum The origin of these currents remains uncertain Analysis ofvoltage recordings (Fig 4A) showed no obvious issue of spaceclamp which suggests these currents are not occurring in distalunclamped regions of the neuron The averaged frequency of theresurgent currents was sim100 Hz which did not vary with level ofpum expression (Fig 4C) Resurgent currents are particularlyevident at holding potentials between minus50 to minus20 mV and exhibithighest frequency at minus30 mV (RRa-Gal4UAS-pumRNAi 10450plusmn3678 Hz RRa-Gal4+ 12000plusmn2016 Hz RRa-Gal4UAS-pum11500plusmn4093 Hz) Increased resurgent INa probably supportsincreased action potential firing consistent with our observationthat RNAi-mediated knockdown of pum is proconvulsant (Griecoet al 2005) Resurgent INa is only rarely observed (lt5) in WTaCC recordings (data not shown)
Fig 2 Expression of transgenic pum is anticonvulsive (A) Expression oftransgenic full-length pum lacking NRE motifs (UAS) in cholinergic neurons inparabss [Cha-Gal4(19B)gtpum] is sufficient to reduce recovery time (RT) frommechanical shock-induced seizure in adult flies compared with parabss alone(CTRL) By contrast further reduction of pum through RNAi-mediatedknockdown (RNAi) [Cha-Gal4(19B)gtRNAi] significantly lengthens seizure RTEach manipulation tested 10 flies per vial to produce an average value Thiswas repeated in triplicate and a final average calculated Data are meansplusmnsdn=3 (B) Identical manipulation of pum expression in cholinergic neurons in L3parabss had an identical effect on seizure duration when seizurewas evoked byelectroshock (C) Pan-neuronal manipulation of pum is also sufficient to affectelectroshock-induced seizure in L3 Upregulation (UAS parabss elaV-Gal4gtpum) reduces seizure and downregulation (RNAi parabss elaV-Gal4gtRNAi) increases seizure duration compared with control (parabss elaV-Gal4+) (D) Up- or downregulation of pum in aWT background using the Cha-Gal4(19B) cholinergic driver line results in induction of a seizure phenotypeData are meansplusmnsd n is stated in individual bars Ple005 Ple001Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
144
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
A screen to identify positive regulators of Pum activityUpregulation of Pum activity either through increased transcriptionor post-transcriptional modification might provide an effectivemeans to suppress seizures To identify possible lead compoundswith this mode of action we constructed a luciferase-based reporterof Pum activity for use in an in vitro S2R+ cell line suited to large-
scale screens (Lin et al 2015) Overexpression of pum is sufficientto repress luciferase activity (due to translational repression)whereas incubation with pum double-stranded RNA is sufficient toincrease luciferase activity by reducing endogenous Pum activityPCR analysis shows that pum is endogenously expressed in S2R+cells (Fig S2) Thus activity of the firefly-luciferase-NRE reporter
Fig 3 Expression of transgenic pum reduces INaP (A-C) Whole-cell patch recordings of INa from L3 aCC motoneurons in parabss (CTRL) parabss expressingtransgenic pum (UAS) or pumRNAi (RNAi) Transgene expression is limited to aCC motoneurons in these manipulations using RRa-Gal4 (DE) Expression oftransgenic pum (UAS) is sufficient to reduce the magnitude of INaP without change to INaT Expression of pumRNAi (RNAi) results in no change to INaP but asignificant increase in INaT (F) Persistent-to-transient (PT) current ratio for INa recorded in DE (GH) The effect of manipulating pum in a WT backgroundIncreasing expression (UAS) is sufficient to reduce INaP with no change to INaT whereas reduction (RNAi) increases INaP amplitude but has no effect on INaT(I) Analysis of the PT ratio in individual cells recorded in GH shows increased pum is sufficient to reduce the ratio Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
Fig 4 Occurrence of resurgent INa is related to level of pum (A) Resurgent INa (INaR arrow) is seen superimposed on repolarization of holding potential usedto evoke INaP Analysis of the voltage trace (lower trace) shows good control during this step (B) The occurrence of INaR in the parabss background is highestwhen pum is reduced (RNAi 82 14 from 17 cells) and lowest when increased (UAS 21 3 from 14 cells) Control (CTRL parabss 64 9 from 14 cells)Transgene expression was limited to aCC cells using RRa-Gal4 (C) Frequency of INaR oscillations is unaffected by expression level of pum Data are meansplusmnsd
145
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(FF-NRE) reflects the absolute level of Pum function in these cellsA second reporter which lacked an NRE-motif was also transfected[renilla (Ren)-luciferase] to allow detrimental effects to cellviability to be determined The final readout of the assay was aFFRen luciferase ratio that would be reduced followingupregulation of Pum activityWe screened 785 compounds from a repurposed library (see
Materials and Methods drugs screened are listed in Table S4) Weidentified 12 compounds that significantly reduced the FFRen ratioat 5 μM (Table 2) Based on structure andor known drug target thecompounds fall into one of four groupings those containing amethoxybenzaldehyde moiety (aniracteam and avobenzone) anti-cancer agents (cladribine gemcitabine floxuridine clofarabinebleomycin and docetaxel) mTOR inhibitors (temsirolimus andrapamycin) and topoisomerase II inhibitors (mitoxantrone andteniposide) Our attention was particularly drawn to avobenzonebecause unlike the other compounds it had no significant effect ontranscription of the control Ren-luciferase reporter (all othercompounds also reduced expression of this reporter in addition todecreasing the FFRen ratio) Thus we took avobenzone forward forfurther testing
Avobenzone potentiates activity of PumWe first tested for anticonvulsant activity in L3 parabss mutantsLarvae raised in food containing avobenzone (04 mgml) showedsignificantly reduced RT in response to electroshock (avobenzone213plusmn124 s n=40 vs control 339plusmn83 s n=20 P=00004 Fig 5A)Similarly exposure of adult parabss flies to avobenzone (04 mgml)24 h before testing also resulted in significant reduction of seizureduration (avobenzone 61plusmn29 vs control 138plusmn29 s n=5 P=00002Fig 5B) Next we recorded INa from parabss aCC in L3 that hadbeen raised on food containing different concentrations ofavobenzone (01-04 mgml Fig 5C-G) Avobenzone reducedINaP from 139plusmn76 pApF in controls to 76plusmn62 pApF at 01 mgml(P=017) 54plusmn64 pApF at 02 mgml (P=003) and 35plusmn42 pApF at04 mgml (P=0002) (Fig 5D) Conversely avobenzone treatment atthese concentrations did not induce any detectable effect in INaT(Fig 5E) Analysis of the PT ratio for INa shows that exposure toavobenzone significantly reduced this value from493plusmn92 in control to 280plusmn232 at 01 mgml (P=009) 219plusmn269 at 02 mgml (P=003) and 121plusmn132 at 04 mgml(P=00004) (Fig 5F) which compares favourably with
overexpression of pum (cf Fig 3) We also observed a significantcorrelation between avobenzone concentration and the occurrence ofresurgent INa (P=0005 Chi-square test Fig 5G)
Our predicted mode of action for avobenzone is inconsistent withan immediate effect of this compound acting instead to potentiatePum which in turn downregulates Nav channels in the neuronalmembrane To test this we recorded from non-drug-exposed L3parabss aCC and used bath application of avobenzone (5 microM) Nochanges were observed in either component of INa (data not shown)and the PT ratio remained unaffected (Fig 5H) Higher doses(20 microM) or longer exposure times (10 min) similarly produced nodetectable effect (data not shown) This lack of acute effect isconsistent with our predicted mode of action Finally to directly testthis prediction we measured pum transcript abundance in parabss
L3 grown in the presence of avobenzoneWe observed a modest butstatistically significant increase in transcript abundance of sim20(12plusmn017 n=5 P=004 t-test vehicle control set as 1 Fig 5I)Thus we conclude that avobenzone acting to increase thetranscription andor transcript stability of pum is able to suppressseizure duration through downregulation of INaP Finally weobserved equally potent anticonvulsive activity of avobenzone intwo other bang-sensitive mutants easily-shocked (avobenzone142plusmn82 vs control 240plusmn120 s n=40 P=10times10minus5 L3electroshock) encoding an ethanolamine kinase (Pavlidis et al1994) and slamdance (avobenzone 178plusmn122 vs control 272plusmn108 s n=40 P=68times10minus5 L3 electroshock) encoding anaminopeptidase (Zhang et al 2002) indicative that increasingPum activity might be effective against a broad range of epilepsies
DISCUSSIONThe causes of seizure even in genetic epilepsies vary greatly andare not confined to genes with obvious contributions to ion fluxacross neuronal membranes This increases the challenge to identifyindividual mutations to determine the physiological role of both theWT and mutated protein and ultimately to design drugs tominimise the unwanted effect of the mutation In this study weidentify transcriptional changes that occur in the seizure-proneCNS We identify over 700 common genes that show alteredtranscription in two different seizure models It is noteworthy thatwe observed approximately double the number of genes showingaltered transcription in parabss flies compared with those treatedwith PTX The reason for this is unclear but might representaccumulated compensatory changes in the mutant line that haveoccurred in order to lessen the severity of seizure activity in parabss
mutants These additional genes warrant further investigation aspotential seizure suppressors
Many of the common transcriptional changes we identify and inparticular those that are upregulated (and thus open to inhibition bydrug exposure) might provide effective drug targets for novel AEDdesign However our attention was drawn to Pum which we havepreviously shown orchestrates homeostasis of action potential firingin both Drosophila and rat central neurons (Driscoll et al 2013Mee et al 2004) The degree of seizure suppression achieved byupregulating Pum in parabss flies is considerable and is onlymatched by the no-action-potential (napts) allele of the maleless(mle) locus in Drosophila which encodes an ATP-dependentdouble-stranded RNA (dsRNA) helicase (Ganetzky andWu 1982)This mutation causes a catastrophic change in splicing of theDrosophilaNav (Reenan et al 2000) The net effect of both of thesemanipulations increased Pum or the presence of napts is to reducethe availability of functional Nav expressed in central neurons Thedirection of change of pum in the two seizure models (that show
Table 2 List of compounds that reduce the fireflyRenilla (FFRen)luciferase ratio thus mimicking the activity of increased pumexpression (shown at bottom of table for reference)
All but avobenzone also reduce expression of the control Ren luciferasereporter that does not contain an NRE motif Luciferase values shown arenormalised such that 10 would represent no effect
146
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
reduced expression) might not be ideal with respect to drugdevelopment given that disruption of a gene or protein is often moreachievable Nevertheless we show that upregulation of pum in aDrosophila seizure mutant is potently anticonvulsive and furtherwe identify a potential lead anticonvulsive compound thatseemingly increases the level of expression of this homeostaticregulator This compound might catalyse the development of anovel class of AEDNeurons display an array of homeostatic mechanisms to maintain
action potential firing within pre-determined and physiologicallyappropriate limits (Davis 2013) Pum is a well-characterised RNA-binding protein that binds mRNA usually through a specific motiftermed the NRE Once bound Pum recruits additional cofactorsincluding Nanos and Brain tumor (Brat) to form a complex that issufficient to prevent translation (Wharton et al 1998) Our results inthis study indicate that increased expression of Pum might havetherapeutic benefit for seizure suppression However a potential
issue in this regard is that a genome-wide identification of RNAsbound to Pum in ovaries identifies upwards of 700 genes(FDRlt01) (Gerber et al 2006) This raises the problem ofspecificity of effect following global potentiation of level or activityof Pum This potential issue might however be overcome throughidentifying and targeting neuronal-specific regulators of Pum Onesuch alternative target might be the inhibition of Myocyte enhancerfactor 2 (Mef2)-induced expression of miR-134 in neurons that inturn inhibits translation of mammalian PUM2 (Fiore et al 2009)Additional possibilities include targeting of cofactors required forPum activity It is interesting in this regard that a loss-of-functionmutation in mei-P26 a homologue of Brat produces strong seizuresuppression in Drosophila bang-sensitive seizure mutants(Glasscock et al 2005)
Mammalian PUM2 binds transcripts encoding SCN1A (Nav11)and SCN8A (Nav16) (Driscoll et al 2013 Vessey et al 2010) Areduction in supply of Nav protein to the neuron membrane is
Fig 5 Avobenzone is anticonvulsant and selectively reduces INaP (A) parabss L3 raised in food containing 04 mgml avobenzone show significantly reducedrecovery time (RT) following electroshock compared with controls (CTRL parabss+DMSO) (B) Exposure of adult parabss flies to avobenzone (04 mgml) is alsopotently anticonvulsant compared with controls (CTRL parabss+DMSO) Each manipulation tested 10 flies per vial to produce an average value This wasrepeated five times and a final average calculated (C) Whole-cell patch recordings of INa from parabss L3 aCC raised in food containing 04 mgml avobenzoneshow reduced INaP (DE) Increasing concentrations of avobenzone (01 02 and 04 mgml) induced a proportional decrease of INaP (D) without affectingINaT (E) (F) Persistent-to-transient (PT) current ratio for INa recorded in aCC (G) The frequency of cells that exhibit resurgent INa correlates with avobenzoneconcentration (P=0005 Chi-square test) (H) PT ratio measured from parabss aCC before (CTRL) and after a 1 min bath application of 5 microM avobenzone(I) Analysis of pum transcript level in isolated CNS from parabss L3 raised on food containing avobenzone (04 mgml) shows a significant increase compared withparabss L3 raised on food containing an equal amount of vehicle (08 DMSO) The control value has been set to 1 Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (AH-I unpaired t-test D-F two-way ANOVA with Bonferronirsquos post hoc)
147
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
consistent with a reduction in action potential firing and a generalanticonvulsant effect (Mee et al 2004) Analysis of INa inmotoneurons indicates that a likely mechanism includes a markedreduction in INaP Increased INaP is associated with mutations inSCN1A that have been identified from individuals with epilepsy(Meisler and Kearney 2005) and is specifically reduced by AEDssuch as phenytoin valproate and lamotrigine (Stafstrom 2007) Inlight of this the anticonvulsant effect of increased pum expression isunderstandable That reducing pum expression through RNAi-mediated knockdown is proconvulsive is again both predictable andunderstandable However the effect of this manipulation on INa isnot so clear Rather than increasing INaP INaT is instead significantlyincreased together with a novel appearance of resurgent INa duringrepolarisation Increased INaT would be expected to reduce thethreshold for action potential firing (ie making firing more likely)whereas resurgent INa is associated with increased action potentialfiring frequency partly by reducing the refractory period (Griecoet al 2005) Although we have observed this current component inrecordings from seizure mutants (including parabss) it is rarelyobserved in WT or following expression of transgenic pumThe ability to manipulate Pum in vivo to determine its
anticonvulsive properties in rodent models of seizure will begreatly aided by the identification of chemical compounds thatdirectly potentiate either expression or activity state We report theuse of a suitable cell-based screen to identify such compounds andhighlight avobenzone as a potential lead compound for futuredevelopment The in vivo toxicity of avobenzone has not been wellestablished and although there are few reports of serious side effectsassociated with its use as an active ingredient of sunscreen itstendency to form free radicals might be a potential issue To ourknowledge this compound has not been used to treat neurologicaldisease and its mode of action in reducing seizure in Drosophilaremains to be determined Our observations that ingestion ofavobenzone result in increased expression of pum is indicative thatthis compound might mimic elements of the pathway that controlexpression of this homeostatic regulatorThe output of our screen also provides additional support for the
use of rapamycin to control seizure (Lasarge and Danzer 2014Russo et al 2013) indicative that this molecule might influenceneuronal homeostasis The identification of topoisomerase II as apotential target to control seizure also validates previousobservations reporting that inhibition of this class of nuclearprotein is anticonvulsant (Lin et al 2015 Song et al 2008)Finally that we identify that the increase in Pum activity byaniracetam might hint at an additional mode of action for this classof known anticonvulsants (Shiotani et al 2000) The relatedracetams levetiracetam and brivaracetam are currently in clinicaluse as AEDs exploiting their capability to bind and inhibit synapticvesicle protein 2A (SV2A) (Klitgaard et al 2016)In summary we present a description of transcriptional change
present in seizure-prone CNS We identify in particular that pumexpression is downregulated in both genetic and chemically inducedseizure models This mirrors the reported reduction in PUM2 inhuman TLE and in rats exposed to the proconvulsant pilocarpine(Wu et al 2015) It also provides a possible understanding for whyPum2 null mice exhibit spontaneous seizures (Siemen et al 2011)However it is perplexing that pum levels should decrease duringseizures given that the published model predicts an increase (Meeet al 2004) As reduced Pum levels are predicted to increaseneuronal excitability it seems that epileptic seizures are associatedwith a pathological dysregulation of pum expression We speculatethat this occurs because Pum can auto-regulate (the pum transcript
contains NRE motifs) Thus although the neuronal hyperactivityinduced by seizures will initially increase Pum expression theaccumulating Pum protein might feed back to downregulate its owntranscript (Gerber et al 2006) Sampling at later stages after seizureoccurrence might only report reduced Pum compared with non-seizure controls Indeed we have shown that upregulation of pum inthe Drosophila CNS through expression of a wild-type transgene(lacking NRE motifs) results in reduction of endogenous pumtranscript level (W-HL and RAB unpublished data) Preventionof this feedback achievable in this study through expression oftransgenic pum lacking an NRE or exposure to avobenzone holdssignificant promise for anticonvulsant therapy
MATERIALS AND METHODSFly stocksWild type (WT maintained in the Baines lab) was Canton-S parabss (bss1)which was obtained from Dr Kevin OrsquoDell (Institute of Molecular Cell andSystems Biology University of Glasgow UK) is detailed in Parker et al(2011) The parabss stock (and other transgenic lines used) were notbackcrossed to the CS stock Controls consisted of either untreated parabss
andor parental stocks (ie Gal4+ UAS+) and are stated in respectivefigure legends Slamdanceiso78 was obtained from Dr Mark Tanouye(Department of Environmental Science Policy and Management andDepartment of Molecular and Cell Biology University of CaliforniaBerkeley California USA) Easily-shocked2F was obtained from Dr KevinOrsquoDell RRa-Gal4 is expressed in only the aCC and RP2 motoneurons (Linet al 2012) We are able to discriminate between these neurons duringelectrophysiological recordings and use only the aCC neuron in this studyWe used Cha-Gal4(19B) to drive UAS-transgene expression in allcholinergic neurons which include excitatory premotor interneurons(Salvaterra and Kitamoto 2001) Pan-neuronal expression was achievedby combining elaV-Gal4 (Bloomington stock no 8760 3rd chromosomeinsert) with parabss UAS-pumRNAi was obtained from the ViennaDrosophila RNAi Center (stock no 101399) and UAS-pum is detailed inSchweers et al (2002) UAS-pum lacks NRE motifs that are present in the3prime-UTR of the endogenous pum gene All genetic crosses were maintained at25degC with the exception of overexpression of pum (larvae die as 1st or 2ndinstars) These experiments were maintained at 205degC Chemical-inducedseizure was achieved by raising WT larvae on food containing 025 mgmlPTX (P1675 Sigma Poole UK) until wall-climbing third instarabbreviated to L3 (Lin et al 2015)
Library construction and RNA sequencingCNSs were removed from 50 L3 (mixed sexes) and RNA extracted using theRNeasy mini kit (QIAGEN Hilden Germany) as described (Lin et al2015) RNA integrity and purity were determined using an Agilent 2200TapeStation system (Agilent Technologies Santa Clara CA) The RNA-sequencing library was created using an mRNA Seq library preparation kitas per manufacturerrsquos instructions (Illumina Inc San Diego CA) Thelibrary products were sequenced in paired-end reads using an IlluminaHiSeqTM 2000 RNA-sequencing data were analysed using edgeR(empirical analysis of digital gene expression in R) (Robinson et al2010) This analysis identified genes with altered levels of expression usinga threshold false discovery rate (FDR)le1 GO terms for BiologicalProcess Cellular Component Molecular Function and Kyoto Encyclopediaof Genes and Genomes (KEGG) pathway were used for annotations Weclassified differentially expressed genes using the Functional AnnotationCluster (FAC) tool available in the Database for Annotation Visualizationand Integrated Discovery (DAVID) (Huang et al 2009ab)
Validation of RNA-sequencing analysis by quantitative PCRQuantitative RT-PCR was performed using a SYBR Green I real-time PCRmethod (Roche LightCyclerreg 480 SYBR Green I Master MannheimGermany) as described in Lin et al (2015) RNAwas extracted from either 20adult heads (3 days old) or 20 L3 CNSs (mixed sexes) using the RNeasymicro kit (QIAGEN) Primer sequences (5prime to 3prime) used were actin-5C
148
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Manipulation of pum in a WT background resulted in a differentoutcome Both RNAi-mediated knockdown and particularlyoverexpression of pum resulted in an induction of a seizurephenotype [Cha-Gal4(19B)UAS-pumRNAi 187plusmn109 s P=18times10minus5n=23 Gal4(19B)UAS-pum 387plusmn77 s P=28times10minus26 n=20]compared with control [Cha-Gal4(19B)+ 97plusmn43 n=40Fig 2D] This paradoxical result is similar to the effect offeedingWTDrosophilaAEDs such as phenytoin that also result inseizure induction an effect that has also been observed in rat(Callaghan and Schwark 1980 Marley and Baines 2011Rundfeldt et al 1990)
Increased pum expression decreases INaP in motoneuronsOur previous work has shown that Pum regulates INa throughtranslational regulation of para (Mee et al 2004 Muraro et al2008) We recorded from parabssY L3 where the expression oftransgenic pum was selectively manipulated in only the aCC
motoneuron (using RRa-Gal4) Our choice to use this motoneuronis guided by the ability to combine genetics and electrophysiologya selective Gal4 driver exists to express UAS-transgenes in thisneuron which is also accessible to patch electrodes That INaP isgreater in amplitude in aCC motoneurons in seizure mutants(Marley and Baines 2011) is indicative that they share propertieswith central interneurons in human epilepsy which can also showincreased INaP (Stafstrom 2007)
Increased expression of pum in L3 parabss aCC resulted in astriking reduction of INaP (44plusmn41 pApF vs 126plusmn40 pApFP=49times10minus5 Fig 3ABD) but no change to INaT (Fig 3E)Analysis of the persistent-to-transient current ratio (PT) recorded inL3 aCC showed a marked reduction (200plusmn180 vs 510plusmn119P=50times10minus5 Fig 3F) A high PT ratio (gt40) in centralmotoneurons has been previously shown to be characteristic ofDrosophila seizure mutants and its reduction to be anticonvulsant(Lin et al 2015 Marley and Baines 2011) Thus we conclude thatupregulation of pum is anticonvulsant which is due at leastpartially to its ability to reduce INaP
RNAi-mediated downregulation of pum in L3 parabss aCCincreased INaT (313plusmn33 pApF vs 247plusmn45 pApF P=0005) butdid not affect INaP or the PT ratio (Fig 3C-F) Analysis of the effecton seizure behaviour following this more selective manipulation ofpum expression showed no significant differences to controls( parabssY RRa-Gal4+ data not shown) This is entirely expectedgiven the highly selective cell targeting used in these experimentsHowever a more widespread manipulation of pum [eg using Cha-Gal4(19B)] which is sufficient to alter seizure duration andorseverity probably acts via an identical mechanism throughalteration of INa
Increasing pum expression in aCC in a WT background resultedin essentially the same changes to INa as seen with manipulation inthe parabss background INaP was significantly reduced (24plusmn17 pApF vs 74plusmn49 pApF P=00028 Fig 3G) but no change toINaT was observed (188plusmn48 pApF vs 219plusmn27 pApF Fig 3H)By contrast downregulation using pumRNAi produced a differentoutcome compared with parabss INaP was significantly increased(110plusmn24 pApF vs 74plusmn49 pApF P=0032 Fig 3G) with noeffect on INaT (246plusmn47 pApF vs 219plusmn27 pApF Fig 3H)Analysis of the PT ratio however similarly only showed asignificant reduction following upregulation of pum expression inWT (147plusmn119 vs 333plusmn202 P=0016 Fig 3I)
On occasion we noted the appearance of multiple resurgent INaduring the INaP plateau in the para
bss background (Fig 4A indicatedby arrow) Moreover we observed a significant correlation betweenthe occurrence of resurgent INa and pum level (P=0002 Chi-squaretest Fig 4B) Thus resurgent INa was most often observedfollowing RNAi-knockdown and only rarely following expressionof pum The origin of these currents remains uncertain Analysis ofvoltage recordings (Fig 4A) showed no obvious issue of spaceclamp which suggests these currents are not occurring in distalunclamped regions of the neuron The averaged frequency of theresurgent currents was sim100 Hz which did not vary with level ofpum expression (Fig 4C) Resurgent currents are particularlyevident at holding potentials between minus50 to minus20 mV and exhibithighest frequency at minus30 mV (RRa-Gal4UAS-pumRNAi 10450plusmn3678 Hz RRa-Gal4+ 12000plusmn2016 Hz RRa-Gal4UAS-pum11500plusmn4093 Hz) Increased resurgent INa probably supportsincreased action potential firing consistent with our observationthat RNAi-mediated knockdown of pum is proconvulsant (Griecoet al 2005) Resurgent INa is only rarely observed (lt5) in WTaCC recordings (data not shown)
Fig 2 Expression of transgenic pum is anticonvulsive (A) Expression oftransgenic full-length pum lacking NRE motifs (UAS) in cholinergic neurons inparabss [Cha-Gal4(19B)gtpum] is sufficient to reduce recovery time (RT) frommechanical shock-induced seizure in adult flies compared with parabss alone(CTRL) By contrast further reduction of pum through RNAi-mediatedknockdown (RNAi) [Cha-Gal4(19B)gtRNAi] significantly lengthens seizure RTEach manipulation tested 10 flies per vial to produce an average value Thiswas repeated in triplicate and a final average calculated Data are meansplusmnsdn=3 (B) Identical manipulation of pum expression in cholinergic neurons in L3parabss had an identical effect on seizure duration when seizurewas evoked byelectroshock (C) Pan-neuronal manipulation of pum is also sufficient to affectelectroshock-induced seizure in L3 Upregulation (UAS parabss elaV-Gal4gtpum) reduces seizure and downregulation (RNAi parabss elaV-Gal4gtRNAi) increases seizure duration compared with control (parabss elaV-Gal4+) (D) Up- or downregulation of pum in aWT background using the Cha-Gal4(19B) cholinergic driver line results in induction of a seizure phenotypeData are meansplusmnsd n is stated in individual bars Ple005 Ple001Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
144
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
A screen to identify positive regulators of Pum activityUpregulation of Pum activity either through increased transcriptionor post-transcriptional modification might provide an effectivemeans to suppress seizures To identify possible lead compoundswith this mode of action we constructed a luciferase-based reporterof Pum activity for use in an in vitro S2R+ cell line suited to large-
scale screens (Lin et al 2015) Overexpression of pum is sufficientto repress luciferase activity (due to translational repression)whereas incubation with pum double-stranded RNA is sufficient toincrease luciferase activity by reducing endogenous Pum activityPCR analysis shows that pum is endogenously expressed in S2R+cells (Fig S2) Thus activity of the firefly-luciferase-NRE reporter
Fig 3 Expression of transgenic pum reduces INaP (A-C) Whole-cell patch recordings of INa from L3 aCC motoneurons in parabss (CTRL) parabss expressingtransgenic pum (UAS) or pumRNAi (RNAi) Transgene expression is limited to aCC motoneurons in these manipulations using RRa-Gal4 (DE) Expression oftransgenic pum (UAS) is sufficient to reduce the magnitude of INaP without change to INaT Expression of pumRNAi (RNAi) results in no change to INaP but asignificant increase in INaT (F) Persistent-to-transient (PT) current ratio for INa recorded in DE (GH) The effect of manipulating pum in a WT backgroundIncreasing expression (UAS) is sufficient to reduce INaP with no change to INaT whereas reduction (RNAi) increases INaP amplitude but has no effect on INaT(I) Analysis of the PT ratio in individual cells recorded in GH shows increased pum is sufficient to reduce the ratio Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
Fig 4 Occurrence of resurgent INa is related to level of pum (A) Resurgent INa (INaR arrow) is seen superimposed on repolarization of holding potential usedto evoke INaP Analysis of the voltage trace (lower trace) shows good control during this step (B) The occurrence of INaR in the parabss background is highestwhen pum is reduced (RNAi 82 14 from 17 cells) and lowest when increased (UAS 21 3 from 14 cells) Control (CTRL parabss 64 9 from 14 cells)Transgene expression was limited to aCC cells using RRa-Gal4 (C) Frequency of INaR oscillations is unaffected by expression level of pum Data are meansplusmnsd
145
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(FF-NRE) reflects the absolute level of Pum function in these cellsA second reporter which lacked an NRE-motif was also transfected[renilla (Ren)-luciferase] to allow detrimental effects to cellviability to be determined The final readout of the assay was aFFRen luciferase ratio that would be reduced followingupregulation of Pum activityWe screened 785 compounds from a repurposed library (see
Materials and Methods drugs screened are listed in Table S4) Weidentified 12 compounds that significantly reduced the FFRen ratioat 5 μM (Table 2) Based on structure andor known drug target thecompounds fall into one of four groupings those containing amethoxybenzaldehyde moiety (aniracteam and avobenzone) anti-cancer agents (cladribine gemcitabine floxuridine clofarabinebleomycin and docetaxel) mTOR inhibitors (temsirolimus andrapamycin) and topoisomerase II inhibitors (mitoxantrone andteniposide) Our attention was particularly drawn to avobenzonebecause unlike the other compounds it had no significant effect ontranscription of the control Ren-luciferase reporter (all othercompounds also reduced expression of this reporter in addition todecreasing the FFRen ratio) Thus we took avobenzone forward forfurther testing
Avobenzone potentiates activity of PumWe first tested for anticonvulsant activity in L3 parabss mutantsLarvae raised in food containing avobenzone (04 mgml) showedsignificantly reduced RT in response to electroshock (avobenzone213plusmn124 s n=40 vs control 339plusmn83 s n=20 P=00004 Fig 5A)Similarly exposure of adult parabss flies to avobenzone (04 mgml)24 h before testing also resulted in significant reduction of seizureduration (avobenzone 61plusmn29 vs control 138plusmn29 s n=5 P=00002Fig 5B) Next we recorded INa from parabss aCC in L3 that hadbeen raised on food containing different concentrations ofavobenzone (01-04 mgml Fig 5C-G) Avobenzone reducedINaP from 139plusmn76 pApF in controls to 76plusmn62 pApF at 01 mgml(P=017) 54plusmn64 pApF at 02 mgml (P=003) and 35plusmn42 pApF at04 mgml (P=0002) (Fig 5D) Conversely avobenzone treatment atthese concentrations did not induce any detectable effect in INaT(Fig 5E) Analysis of the PT ratio for INa shows that exposure toavobenzone significantly reduced this value from493plusmn92 in control to 280plusmn232 at 01 mgml (P=009) 219plusmn269 at 02 mgml (P=003) and 121plusmn132 at 04 mgml(P=00004) (Fig 5F) which compares favourably with
overexpression of pum (cf Fig 3) We also observed a significantcorrelation between avobenzone concentration and the occurrence ofresurgent INa (P=0005 Chi-square test Fig 5G)
Our predicted mode of action for avobenzone is inconsistent withan immediate effect of this compound acting instead to potentiatePum which in turn downregulates Nav channels in the neuronalmembrane To test this we recorded from non-drug-exposed L3parabss aCC and used bath application of avobenzone (5 microM) Nochanges were observed in either component of INa (data not shown)and the PT ratio remained unaffected (Fig 5H) Higher doses(20 microM) or longer exposure times (10 min) similarly produced nodetectable effect (data not shown) This lack of acute effect isconsistent with our predicted mode of action Finally to directly testthis prediction we measured pum transcript abundance in parabss
L3 grown in the presence of avobenzoneWe observed a modest butstatistically significant increase in transcript abundance of sim20(12plusmn017 n=5 P=004 t-test vehicle control set as 1 Fig 5I)Thus we conclude that avobenzone acting to increase thetranscription andor transcript stability of pum is able to suppressseizure duration through downregulation of INaP Finally weobserved equally potent anticonvulsive activity of avobenzone intwo other bang-sensitive mutants easily-shocked (avobenzone142plusmn82 vs control 240plusmn120 s n=40 P=10times10minus5 L3electroshock) encoding an ethanolamine kinase (Pavlidis et al1994) and slamdance (avobenzone 178plusmn122 vs control 272plusmn108 s n=40 P=68times10minus5 L3 electroshock) encoding anaminopeptidase (Zhang et al 2002) indicative that increasingPum activity might be effective against a broad range of epilepsies
DISCUSSIONThe causes of seizure even in genetic epilepsies vary greatly andare not confined to genes with obvious contributions to ion fluxacross neuronal membranes This increases the challenge to identifyindividual mutations to determine the physiological role of both theWT and mutated protein and ultimately to design drugs tominimise the unwanted effect of the mutation In this study weidentify transcriptional changes that occur in the seizure-proneCNS We identify over 700 common genes that show alteredtranscription in two different seizure models It is noteworthy thatwe observed approximately double the number of genes showingaltered transcription in parabss flies compared with those treatedwith PTX The reason for this is unclear but might representaccumulated compensatory changes in the mutant line that haveoccurred in order to lessen the severity of seizure activity in parabss
mutants These additional genes warrant further investigation aspotential seizure suppressors
Many of the common transcriptional changes we identify and inparticular those that are upregulated (and thus open to inhibition bydrug exposure) might provide effective drug targets for novel AEDdesign However our attention was drawn to Pum which we havepreviously shown orchestrates homeostasis of action potential firingin both Drosophila and rat central neurons (Driscoll et al 2013Mee et al 2004) The degree of seizure suppression achieved byupregulating Pum in parabss flies is considerable and is onlymatched by the no-action-potential (napts) allele of the maleless(mle) locus in Drosophila which encodes an ATP-dependentdouble-stranded RNA (dsRNA) helicase (Ganetzky andWu 1982)This mutation causes a catastrophic change in splicing of theDrosophilaNav (Reenan et al 2000) The net effect of both of thesemanipulations increased Pum or the presence of napts is to reducethe availability of functional Nav expressed in central neurons Thedirection of change of pum in the two seizure models (that show
Table 2 List of compounds that reduce the fireflyRenilla (FFRen)luciferase ratio thus mimicking the activity of increased pumexpression (shown at bottom of table for reference)
All but avobenzone also reduce expression of the control Ren luciferasereporter that does not contain an NRE motif Luciferase values shown arenormalised such that 10 would represent no effect
146
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
reduced expression) might not be ideal with respect to drugdevelopment given that disruption of a gene or protein is often moreachievable Nevertheless we show that upregulation of pum in aDrosophila seizure mutant is potently anticonvulsive and furtherwe identify a potential lead anticonvulsive compound thatseemingly increases the level of expression of this homeostaticregulator This compound might catalyse the development of anovel class of AEDNeurons display an array of homeostatic mechanisms to maintain
action potential firing within pre-determined and physiologicallyappropriate limits (Davis 2013) Pum is a well-characterised RNA-binding protein that binds mRNA usually through a specific motiftermed the NRE Once bound Pum recruits additional cofactorsincluding Nanos and Brain tumor (Brat) to form a complex that issufficient to prevent translation (Wharton et al 1998) Our results inthis study indicate that increased expression of Pum might havetherapeutic benefit for seizure suppression However a potential
issue in this regard is that a genome-wide identification of RNAsbound to Pum in ovaries identifies upwards of 700 genes(FDRlt01) (Gerber et al 2006) This raises the problem ofspecificity of effect following global potentiation of level or activityof Pum This potential issue might however be overcome throughidentifying and targeting neuronal-specific regulators of Pum Onesuch alternative target might be the inhibition of Myocyte enhancerfactor 2 (Mef2)-induced expression of miR-134 in neurons that inturn inhibits translation of mammalian PUM2 (Fiore et al 2009)Additional possibilities include targeting of cofactors required forPum activity It is interesting in this regard that a loss-of-functionmutation in mei-P26 a homologue of Brat produces strong seizuresuppression in Drosophila bang-sensitive seizure mutants(Glasscock et al 2005)
Mammalian PUM2 binds transcripts encoding SCN1A (Nav11)and SCN8A (Nav16) (Driscoll et al 2013 Vessey et al 2010) Areduction in supply of Nav protein to the neuron membrane is
Fig 5 Avobenzone is anticonvulsant and selectively reduces INaP (A) parabss L3 raised in food containing 04 mgml avobenzone show significantly reducedrecovery time (RT) following electroshock compared with controls (CTRL parabss+DMSO) (B) Exposure of adult parabss flies to avobenzone (04 mgml) is alsopotently anticonvulsant compared with controls (CTRL parabss+DMSO) Each manipulation tested 10 flies per vial to produce an average value This wasrepeated five times and a final average calculated (C) Whole-cell patch recordings of INa from parabss L3 aCC raised in food containing 04 mgml avobenzoneshow reduced INaP (DE) Increasing concentrations of avobenzone (01 02 and 04 mgml) induced a proportional decrease of INaP (D) without affectingINaT (E) (F) Persistent-to-transient (PT) current ratio for INa recorded in aCC (G) The frequency of cells that exhibit resurgent INa correlates with avobenzoneconcentration (P=0005 Chi-square test) (H) PT ratio measured from parabss aCC before (CTRL) and after a 1 min bath application of 5 microM avobenzone(I) Analysis of pum transcript level in isolated CNS from parabss L3 raised on food containing avobenzone (04 mgml) shows a significant increase compared withparabss L3 raised on food containing an equal amount of vehicle (08 DMSO) The control value has been set to 1 Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (AH-I unpaired t-test D-F two-way ANOVA with Bonferronirsquos post hoc)
147
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
consistent with a reduction in action potential firing and a generalanticonvulsant effect (Mee et al 2004) Analysis of INa inmotoneurons indicates that a likely mechanism includes a markedreduction in INaP Increased INaP is associated with mutations inSCN1A that have been identified from individuals with epilepsy(Meisler and Kearney 2005) and is specifically reduced by AEDssuch as phenytoin valproate and lamotrigine (Stafstrom 2007) Inlight of this the anticonvulsant effect of increased pum expression isunderstandable That reducing pum expression through RNAi-mediated knockdown is proconvulsive is again both predictable andunderstandable However the effect of this manipulation on INa isnot so clear Rather than increasing INaP INaT is instead significantlyincreased together with a novel appearance of resurgent INa duringrepolarisation Increased INaT would be expected to reduce thethreshold for action potential firing (ie making firing more likely)whereas resurgent INa is associated with increased action potentialfiring frequency partly by reducing the refractory period (Griecoet al 2005) Although we have observed this current component inrecordings from seizure mutants (including parabss) it is rarelyobserved in WT or following expression of transgenic pumThe ability to manipulate Pum in vivo to determine its
anticonvulsive properties in rodent models of seizure will begreatly aided by the identification of chemical compounds thatdirectly potentiate either expression or activity state We report theuse of a suitable cell-based screen to identify such compounds andhighlight avobenzone as a potential lead compound for futuredevelopment The in vivo toxicity of avobenzone has not been wellestablished and although there are few reports of serious side effectsassociated with its use as an active ingredient of sunscreen itstendency to form free radicals might be a potential issue To ourknowledge this compound has not been used to treat neurologicaldisease and its mode of action in reducing seizure in Drosophilaremains to be determined Our observations that ingestion ofavobenzone result in increased expression of pum is indicative thatthis compound might mimic elements of the pathway that controlexpression of this homeostatic regulatorThe output of our screen also provides additional support for the
use of rapamycin to control seizure (Lasarge and Danzer 2014Russo et al 2013) indicative that this molecule might influenceneuronal homeostasis The identification of topoisomerase II as apotential target to control seizure also validates previousobservations reporting that inhibition of this class of nuclearprotein is anticonvulsant (Lin et al 2015 Song et al 2008)Finally that we identify that the increase in Pum activity byaniracetam might hint at an additional mode of action for this classof known anticonvulsants (Shiotani et al 2000) The relatedracetams levetiracetam and brivaracetam are currently in clinicaluse as AEDs exploiting their capability to bind and inhibit synapticvesicle protein 2A (SV2A) (Klitgaard et al 2016)In summary we present a description of transcriptional change
present in seizure-prone CNS We identify in particular that pumexpression is downregulated in both genetic and chemically inducedseizure models This mirrors the reported reduction in PUM2 inhuman TLE and in rats exposed to the proconvulsant pilocarpine(Wu et al 2015) It also provides a possible understanding for whyPum2 null mice exhibit spontaneous seizures (Siemen et al 2011)However it is perplexing that pum levels should decrease duringseizures given that the published model predicts an increase (Meeet al 2004) As reduced Pum levels are predicted to increaseneuronal excitability it seems that epileptic seizures are associatedwith a pathological dysregulation of pum expression We speculatethat this occurs because Pum can auto-regulate (the pum transcript
contains NRE motifs) Thus although the neuronal hyperactivityinduced by seizures will initially increase Pum expression theaccumulating Pum protein might feed back to downregulate its owntranscript (Gerber et al 2006) Sampling at later stages after seizureoccurrence might only report reduced Pum compared with non-seizure controls Indeed we have shown that upregulation of pum inthe Drosophila CNS through expression of a wild-type transgene(lacking NRE motifs) results in reduction of endogenous pumtranscript level (W-HL and RAB unpublished data) Preventionof this feedback achievable in this study through expression oftransgenic pum lacking an NRE or exposure to avobenzone holdssignificant promise for anticonvulsant therapy
MATERIALS AND METHODSFly stocksWild type (WT maintained in the Baines lab) was Canton-S parabss (bss1)which was obtained from Dr Kevin OrsquoDell (Institute of Molecular Cell andSystems Biology University of Glasgow UK) is detailed in Parker et al(2011) The parabss stock (and other transgenic lines used) were notbackcrossed to the CS stock Controls consisted of either untreated parabss
andor parental stocks (ie Gal4+ UAS+) and are stated in respectivefigure legends Slamdanceiso78 was obtained from Dr Mark Tanouye(Department of Environmental Science Policy and Management andDepartment of Molecular and Cell Biology University of CaliforniaBerkeley California USA) Easily-shocked2F was obtained from Dr KevinOrsquoDell RRa-Gal4 is expressed in only the aCC and RP2 motoneurons (Linet al 2012) We are able to discriminate between these neurons duringelectrophysiological recordings and use only the aCC neuron in this studyWe used Cha-Gal4(19B) to drive UAS-transgene expression in allcholinergic neurons which include excitatory premotor interneurons(Salvaterra and Kitamoto 2001) Pan-neuronal expression was achievedby combining elaV-Gal4 (Bloomington stock no 8760 3rd chromosomeinsert) with parabss UAS-pumRNAi was obtained from the ViennaDrosophila RNAi Center (stock no 101399) and UAS-pum is detailed inSchweers et al (2002) UAS-pum lacks NRE motifs that are present in the3prime-UTR of the endogenous pum gene All genetic crosses were maintained at25degC with the exception of overexpression of pum (larvae die as 1st or 2ndinstars) These experiments were maintained at 205degC Chemical-inducedseizure was achieved by raising WT larvae on food containing 025 mgmlPTX (P1675 Sigma Poole UK) until wall-climbing third instarabbreviated to L3 (Lin et al 2015)
Library construction and RNA sequencingCNSs were removed from 50 L3 (mixed sexes) and RNA extracted using theRNeasy mini kit (QIAGEN Hilden Germany) as described (Lin et al2015) RNA integrity and purity were determined using an Agilent 2200TapeStation system (Agilent Technologies Santa Clara CA) The RNA-sequencing library was created using an mRNA Seq library preparation kitas per manufacturerrsquos instructions (Illumina Inc San Diego CA) Thelibrary products were sequenced in paired-end reads using an IlluminaHiSeqTM 2000 RNA-sequencing data were analysed using edgeR(empirical analysis of digital gene expression in R) (Robinson et al2010) This analysis identified genes with altered levels of expression usinga threshold false discovery rate (FDR)le1 GO terms for BiologicalProcess Cellular Component Molecular Function and Kyoto Encyclopediaof Genes and Genomes (KEGG) pathway were used for annotations Weclassified differentially expressed genes using the Functional AnnotationCluster (FAC) tool available in the Database for Annotation Visualizationand Integrated Discovery (DAVID) (Huang et al 2009ab)
Validation of RNA-sequencing analysis by quantitative PCRQuantitative RT-PCR was performed using a SYBR Green I real-time PCRmethod (Roche LightCyclerreg 480 SYBR Green I Master MannheimGermany) as described in Lin et al (2015) RNAwas extracted from either 20adult heads (3 days old) or 20 L3 CNSs (mixed sexes) using the RNeasymicro kit (QIAGEN) Primer sequences (5prime to 3prime) used were actin-5C
148
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
A screen to identify positive regulators of Pum activityUpregulation of Pum activity either through increased transcriptionor post-transcriptional modification might provide an effectivemeans to suppress seizures To identify possible lead compoundswith this mode of action we constructed a luciferase-based reporterof Pum activity for use in an in vitro S2R+ cell line suited to large-
scale screens (Lin et al 2015) Overexpression of pum is sufficientto repress luciferase activity (due to translational repression)whereas incubation with pum double-stranded RNA is sufficient toincrease luciferase activity by reducing endogenous Pum activityPCR analysis shows that pum is endogenously expressed in S2R+cells (Fig S2) Thus activity of the firefly-luciferase-NRE reporter
Fig 3 Expression of transgenic pum reduces INaP (A-C) Whole-cell patch recordings of INa from L3 aCC motoneurons in parabss (CTRL) parabss expressingtransgenic pum (UAS) or pumRNAi (RNAi) Transgene expression is limited to aCC motoneurons in these manipulations using RRa-Gal4 (DE) Expression oftransgenic pum (UAS) is sufficient to reduce the magnitude of INaP without change to INaT Expression of pumRNAi (RNAi) results in no change to INaP but asignificant increase in INaT (F) Persistent-to-transient (PT) current ratio for INa recorded in DE (GH) The effect of manipulating pum in a WT backgroundIncreasing expression (UAS) is sufficient to reduce INaP with no change to INaT whereas reduction (RNAi) increases INaP amplitude but has no effect on INaT(I) Analysis of the PT ratio in individual cells recorded in GH shows increased pum is sufficient to reduce the ratio Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (two-way ANOVA with Bonferronirsquos post hoc)
Fig 4 Occurrence of resurgent INa is related to level of pum (A) Resurgent INa (INaR arrow) is seen superimposed on repolarization of holding potential usedto evoke INaP Analysis of the voltage trace (lower trace) shows good control during this step (B) The occurrence of INaR in the parabss background is highestwhen pum is reduced (RNAi 82 14 from 17 cells) and lowest when increased (UAS 21 3 from 14 cells) Control (CTRL parabss 64 9 from 14 cells)Transgene expression was limited to aCC cells using RRa-Gal4 (C) Frequency of INaR oscillations is unaffected by expression level of pum Data are meansplusmnsd
145
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(FF-NRE) reflects the absolute level of Pum function in these cellsA second reporter which lacked an NRE-motif was also transfected[renilla (Ren)-luciferase] to allow detrimental effects to cellviability to be determined The final readout of the assay was aFFRen luciferase ratio that would be reduced followingupregulation of Pum activityWe screened 785 compounds from a repurposed library (see
Materials and Methods drugs screened are listed in Table S4) Weidentified 12 compounds that significantly reduced the FFRen ratioat 5 μM (Table 2) Based on structure andor known drug target thecompounds fall into one of four groupings those containing amethoxybenzaldehyde moiety (aniracteam and avobenzone) anti-cancer agents (cladribine gemcitabine floxuridine clofarabinebleomycin and docetaxel) mTOR inhibitors (temsirolimus andrapamycin) and topoisomerase II inhibitors (mitoxantrone andteniposide) Our attention was particularly drawn to avobenzonebecause unlike the other compounds it had no significant effect ontranscription of the control Ren-luciferase reporter (all othercompounds also reduced expression of this reporter in addition todecreasing the FFRen ratio) Thus we took avobenzone forward forfurther testing
Avobenzone potentiates activity of PumWe first tested for anticonvulsant activity in L3 parabss mutantsLarvae raised in food containing avobenzone (04 mgml) showedsignificantly reduced RT in response to electroshock (avobenzone213plusmn124 s n=40 vs control 339plusmn83 s n=20 P=00004 Fig 5A)Similarly exposure of adult parabss flies to avobenzone (04 mgml)24 h before testing also resulted in significant reduction of seizureduration (avobenzone 61plusmn29 vs control 138plusmn29 s n=5 P=00002Fig 5B) Next we recorded INa from parabss aCC in L3 that hadbeen raised on food containing different concentrations ofavobenzone (01-04 mgml Fig 5C-G) Avobenzone reducedINaP from 139plusmn76 pApF in controls to 76plusmn62 pApF at 01 mgml(P=017) 54plusmn64 pApF at 02 mgml (P=003) and 35plusmn42 pApF at04 mgml (P=0002) (Fig 5D) Conversely avobenzone treatment atthese concentrations did not induce any detectable effect in INaT(Fig 5E) Analysis of the PT ratio for INa shows that exposure toavobenzone significantly reduced this value from493plusmn92 in control to 280plusmn232 at 01 mgml (P=009) 219plusmn269 at 02 mgml (P=003) and 121plusmn132 at 04 mgml(P=00004) (Fig 5F) which compares favourably with
overexpression of pum (cf Fig 3) We also observed a significantcorrelation between avobenzone concentration and the occurrence ofresurgent INa (P=0005 Chi-square test Fig 5G)
Our predicted mode of action for avobenzone is inconsistent withan immediate effect of this compound acting instead to potentiatePum which in turn downregulates Nav channels in the neuronalmembrane To test this we recorded from non-drug-exposed L3parabss aCC and used bath application of avobenzone (5 microM) Nochanges were observed in either component of INa (data not shown)and the PT ratio remained unaffected (Fig 5H) Higher doses(20 microM) or longer exposure times (10 min) similarly produced nodetectable effect (data not shown) This lack of acute effect isconsistent with our predicted mode of action Finally to directly testthis prediction we measured pum transcript abundance in parabss
L3 grown in the presence of avobenzoneWe observed a modest butstatistically significant increase in transcript abundance of sim20(12plusmn017 n=5 P=004 t-test vehicle control set as 1 Fig 5I)Thus we conclude that avobenzone acting to increase thetranscription andor transcript stability of pum is able to suppressseizure duration through downregulation of INaP Finally weobserved equally potent anticonvulsive activity of avobenzone intwo other bang-sensitive mutants easily-shocked (avobenzone142plusmn82 vs control 240plusmn120 s n=40 P=10times10minus5 L3electroshock) encoding an ethanolamine kinase (Pavlidis et al1994) and slamdance (avobenzone 178plusmn122 vs control 272plusmn108 s n=40 P=68times10minus5 L3 electroshock) encoding anaminopeptidase (Zhang et al 2002) indicative that increasingPum activity might be effective against a broad range of epilepsies
DISCUSSIONThe causes of seizure even in genetic epilepsies vary greatly andare not confined to genes with obvious contributions to ion fluxacross neuronal membranes This increases the challenge to identifyindividual mutations to determine the physiological role of both theWT and mutated protein and ultimately to design drugs tominimise the unwanted effect of the mutation In this study weidentify transcriptional changes that occur in the seizure-proneCNS We identify over 700 common genes that show alteredtranscription in two different seizure models It is noteworthy thatwe observed approximately double the number of genes showingaltered transcription in parabss flies compared with those treatedwith PTX The reason for this is unclear but might representaccumulated compensatory changes in the mutant line that haveoccurred in order to lessen the severity of seizure activity in parabss
mutants These additional genes warrant further investigation aspotential seizure suppressors
Many of the common transcriptional changes we identify and inparticular those that are upregulated (and thus open to inhibition bydrug exposure) might provide effective drug targets for novel AEDdesign However our attention was drawn to Pum which we havepreviously shown orchestrates homeostasis of action potential firingin both Drosophila and rat central neurons (Driscoll et al 2013Mee et al 2004) The degree of seizure suppression achieved byupregulating Pum in parabss flies is considerable and is onlymatched by the no-action-potential (napts) allele of the maleless(mle) locus in Drosophila which encodes an ATP-dependentdouble-stranded RNA (dsRNA) helicase (Ganetzky andWu 1982)This mutation causes a catastrophic change in splicing of theDrosophilaNav (Reenan et al 2000) The net effect of both of thesemanipulations increased Pum or the presence of napts is to reducethe availability of functional Nav expressed in central neurons Thedirection of change of pum in the two seizure models (that show
Table 2 List of compounds that reduce the fireflyRenilla (FFRen)luciferase ratio thus mimicking the activity of increased pumexpression (shown at bottom of table for reference)
All but avobenzone also reduce expression of the control Ren luciferasereporter that does not contain an NRE motif Luciferase values shown arenormalised such that 10 would represent no effect
146
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
reduced expression) might not be ideal with respect to drugdevelopment given that disruption of a gene or protein is often moreachievable Nevertheless we show that upregulation of pum in aDrosophila seizure mutant is potently anticonvulsive and furtherwe identify a potential lead anticonvulsive compound thatseemingly increases the level of expression of this homeostaticregulator This compound might catalyse the development of anovel class of AEDNeurons display an array of homeostatic mechanisms to maintain
action potential firing within pre-determined and physiologicallyappropriate limits (Davis 2013) Pum is a well-characterised RNA-binding protein that binds mRNA usually through a specific motiftermed the NRE Once bound Pum recruits additional cofactorsincluding Nanos and Brain tumor (Brat) to form a complex that issufficient to prevent translation (Wharton et al 1998) Our results inthis study indicate that increased expression of Pum might havetherapeutic benefit for seizure suppression However a potential
issue in this regard is that a genome-wide identification of RNAsbound to Pum in ovaries identifies upwards of 700 genes(FDRlt01) (Gerber et al 2006) This raises the problem ofspecificity of effect following global potentiation of level or activityof Pum This potential issue might however be overcome throughidentifying and targeting neuronal-specific regulators of Pum Onesuch alternative target might be the inhibition of Myocyte enhancerfactor 2 (Mef2)-induced expression of miR-134 in neurons that inturn inhibits translation of mammalian PUM2 (Fiore et al 2009)Additional possibilities include targeting of cofactors required forPum activity It is interesting in this regard that a loss-of-functionmutation in mei-P26 a homologue of Brat produces strong seizuresuppression in Drosophila bang-sensitive seizure mutants(Glasscock et al 2005)
Mammalian PUM2 binds transcripts encoding SCN1A (Nav11)and SCN8A (Nav16) (Driscoll et al 2013 Vessey et al 2010) Areduction in supply of Nav protein to the neuron membrane is
Fig 5 Avobenzone is anticonvulsant and selectively reduces INaP (A) parabss L3 raised in food containing 04 mgml avobenzone show significantly reducedrecovery time (RT) following electroshock compared with controls (CTRL parabss+DMSO) (B) Exposure of adult parabss flies to avobenzone (04 mgml) is alsopotently anticonvulsant compared with controls (CTRL parabss+DMSO) Each manipulation tested 10 flies per vial to produce an average value This wasrepeated five times and a final average calculated (C) Whole-cell patch recordings of INa from parabss L3 aCC raised in food containing 04 mgml avobenzoneshow reduced INaP (DE) Increasing concentrations of avobenzone (01 02 and 04 mgml) induced a proportional decrease of INaP (D) without affectingINaT (E) (F) Persistent-to-transient (PT) current ratio for INa recorded in aCC (G) The frequency of cells that exhibit resurgent INa correlates with avobenzoneconcentration (P=0005 Chi-square test) (H) PT ratio measured from parabss aCC before (CTRL) and after a 1 min bath application of 5 microM avobenzone(I) Analysis of pum transcript level in isolated CNS from parabss L3 raised on food containing avobenzone (04 mgml) shows a significant increase compared withparabss L3 raised on food containing an equal amount of vehicle (08 DMSO) The control value has been set to 1 Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (AH-I unpaired t-test D-F two-way ANOVA with Bonferronirsquos post hoc)
147
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
consistent with a reduction in action potential firing and a generalanticonvulsant effect (Mee et al 2004) Analysis of INa inmotoneurons indicates that a likely mechanism includes a markedreduction in INaP Increased INaP is associated with mutations inSCN1A that have been identified from individuals with epilepsy(Meisler and Kearney 2005) and is specifically reduced by AEDssuch as phenytoin valproate and lamotrigine (Stafstrom 2007) Inlight of this the anticonvulsant effect of increased pum expression isunderstandable That reducing pum expression through RNAi-mediated knockdown is proconvulsive is again both predictable andunderstandable However the effect of this manipulation on INa isnot so clear Rather than increasing INaP INaT is instead significantlyincreased together with a novel appearance of resurgent INa duringrepolarisation Increased INaT would be expected to reduce thethreshold for action potential firing (ie making firing more likely)whereas resurgent INa is associated with increased action potentialfiring frequency partly by reducing the refractory period (Griecoet al 2005) Although we have observed this current component inrecordings from seizure mutants (including parabss) it is rarelyobserved in WT or following expression of transgenic pumThe ability to manipulate Pum in vivo to determine its
anticonvulsive properties in rodent models of seizure will begreatly aided by the identification of chemical compounds thatdirectly potentiate either expression or activity state We report theuse of a suitable cell-based screen to identify such compounds andhighlight avobenzone as a potential lead compound for futuredevelopment The in vivo toxicity of avobenzone has not been wellestablished and although there are few reports of serious side effectsassociated with its use as an active ingredient of sunscreen itstendency to form free radicals might be a potential issue To ourknowledge this compound has not been used to treat neurologicaldisease and its mode of action in reducing seizure in Drosophilaremains to be determined Our observations that ingestion ofavobenzone result in increased expression of pum is indicative thatthis compound might mimic elements of the pathway that controlexpression of this homeostatic regulatorThe output of our screen also provides additional support for the
use of rapamycin to control seizure (Lasarge and Danzer 2014Russo et al 2013) indicative that this molecule might influenceneuronal homeostasis The identification of topoisomerase II as apotential target to control seizure also validates previousobservations reporting that inhibition of this class of nuclearprotein is anticonvulsant (Lin et al 2015 Song et al 2008)Finally that we identify that the increase in Pum activity byaniracetam might hint at an additional mode of action for this classof known anticonvulsants (Shiotani et al 2000) The relatedracetams levetiracetam and brivaracetam are currently in clinicaluse as AEDs exploiting their capability to bind and inhibit synapticvesicle protein 2A (SV2A) (Klitgaard et al 2016)In summary we present a description of transcriptional change
present in seizure-prone CNS We identify in particular that pumexpression is downregulated in both genetic and chemically inducedseizure models This mirrors the reported reduction in PUM2 inhuman TLE and in rats exposed to the proconvulsant pilocarpine(Wu et al 2015) It also provides a possible understanding for whyPum2 null mice exhibit spontaneous seizures (Siemen et al 2011)However it is perplexing that pum levels should decrease duringseizures given that the published model predicts an increase (Meeet al 2004) As reduced Pum levels are predicted to increaseneuronal excitability it seems that epileptic seizures are associatedwith a pathological dysregulation of pum expression We speculatethat this occurs because Pum can auto-regulate (the pum transcript
contains NRE motifs) Thus although the neuronal hyperactivityinduced by seizures will initially increase Pum expression theaccumulating Pum protein might feed back to downregulate its owntranscript (Gerber et al 2006) Sampling at later stages after seizureoccurrence might only report reduced Pum compared with non-seizure controls Indeed we have shown that upregulation of pum inthe Drosophila CNS through expression of a wild-type transgene(lacking NRE motifs) results in reduction of endogenous pumtranscript level (W-HL and RAB unpublished data) Preventionof this feedback achievable in this study through expression oftransgenic pum lacking an NRE or exposure to avobenzone holdssignificant promise for anticonvulsant therapy
MATERIALS AND METHODSFly stocksWild type (WT maintained in the Baines lab) was Canton-S parabss (bss1)which was obtained from Dr Kevin OrsquoDell (Institute of Molecular Cell andSystems Biology University of Glasgow UK) is detailed in Parker et al(2011) The parabss stock (and other transgenic lines used) were notbackcrossed to the CS stock Controls consisted of either untreated parabss
andor parental stocks (ie Gal4+ UAS+) and are stated in respectivefigure legends Slamdanceiso78 was obtained from Dr Mark Tanouye(Department of Environmental Science Policy and Management andDepartment of Molecular and Cell Biology University of CaliforniaBerkeley California USA) Easily-shocked2F was obtained from Dr KevinOrsquoDell RRa-Gal4 is expressed in only the aCC and RP2 motoneurons (Linet al 2012) We are able to discriminate between these neurons duringelectrophysiological recordings and use only the aCC neuron in this studyWe used Cha-Gal4(19B) to drive UAS-transgene expression in allcholinergic neurons which include excitatory premotor interneurons(Salvaterra and Kitamoto 2001) Pan-neuronal expression was achievedby combining elaV-Gal4 (Bloomington stock no 8760 3rd chromosomeinsert) with parabss UAS-pumRNAi was obtained from the ViennaDrosophila RNAi Center (stock no 101399) and UAS-pum is detailed inSchweers et al (2002) UAS-pum lacks NRE motifs that are present in the3prime-UTR of the endogenous pum gene All genetic crosses were maintained at25degC with the exception of overexpression of pum (larvae die as 1st or 2ndinstars) These experiments were maintained at 205degC Chemical-inducedseizure was achieved by raising WT larvae on food containing 025 mgmlPTX (P1675 Sigma Poole UK) until wall-climbing third instarabbreviated to L3 (Lin et al 2015)
Library construction and RNA sequencingCNSs were removed from 50 L3 (mixed sexes) and RNA extracted using theRNeasy mini kit (QIAGEN Hilden Germany) as described (Lin et al2015) RNA integrity and purity were determined using an Agilent 2200TapeStation system (Agilent Technologies Santa Clara CA) The RNA-sequencing library was created using an mRNA Seq library preparation kitas per manufacturerrsquos instructions (Illumina Inc San Diego CA) Thelibrary products were sequenced in paired-end reads using an IlluminaHiSeqTM 2000 RNA-sequencing data were analysed using edgeR(empirical analysis of digital gene expression in R) (Robinson et al2010) This analysis identified genes with altered levels of expression usinga threshold false discovery rate (FDR)le1 GO terms for BiologicalProcess Cellular Component Molecular Function and Kyoto Encyclopediaof Genes and Genomes (KEGG) pathway were used for annotations Weclassified differentially expressed genes using the Functional AnnotationCluster (FAC) tool available in the Database for Annotation Visualizationand Integrated Discovery (DAVID) (Huang et al 2009ab)
Validation of RNA-sequencing analysis by quantitative PCRQuantitative RT-PCR was performed using a SYBR Green I real-time PCRmethod (Roche LightCyclerreg 480 SYBR Green I Master MannheimGermany) as described in Lin et al (2015) RNAwas extracted from either 20adult heads (3 days old) or 20 L3 CNSs (mixed sexes) using the RNeasymicro kit (QIAGEN) Primer sequences (5prime to 3prime) used were actin-5C
148
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(FF-NRE) reflects the absolute level of Pum function in these cellsA second reporter which lacked an NRE-motif was also transfected[renilla (Ren)-luciferase] to allow detrimental effects to cellviability to be determined The final readout of the assay was aFFRen luciferase ratio that would be reduced followingupregulation of Pum activityWe screened 785 compounds from a repurposed library (see
Materials and Methods drugs screened are listed in Table S4) Weidentified 12 compounds that significantly reduced the FFRen ratioat 5 μM (Table 2) Based on structure andor known drug target thecompounds fall into one of four groupings those containing amethoxybenzaldehyde moiety (aniracteam and avobenzone) anti-cancer agents (cladribine gemcitabine floxuridine clofarabinebleomycin and docetaxel) mTOR inhibitors (temsirolimus andrapamycin) and topoisomerase II inhibitors (mitoxantrone andteniposide) Our attention was particularly drawn to avobenzonebecause unlike the other compounds it had no significant effect ontranscription of the control Ren-luciferase reporter (all othercompounds also reduced expression of this reporter in addition todecreasing the FFRen ratio) Thus we took avobenzone forward forfurther testing
Avobenzone potentiates activity of PumWe first tested for anticonvulsant activity in L3 parabss mutantsLarvae raised in food containing avobenzone (04 mgml) showedsignificantly reduced RT in response to electroshock (avobenzone213plusmn124 s n=40 vs control 339plusmn83 s n=20 P=00004 Fig 5A)Similarly exposure of adult parabss flies to avobenzone (04 mgml)24 h before testing also resulted in significant reduction of seizureduration (avobenzone 61plusmn29 vs control 138plusmn29 s n=5 P=00002Fig 5B) Next we recorded INa from parabss aCC in L3 that hadbeen raised on food containing different concentrations ofavobenzone (01-04 mgml Fig 5C-G) Avobenzone reducedINaP from 139plusmn76 pApF in controls to 76plusmn62 pApF at 01 mgml(P=017) 54plusmn64 pApF at 02 mgml (P=003) and 35plusmn42 pApF at04 mgml (P=0002) (Fig 5D) Conversely avobenzone treatment atthese concentrations did not induce any detectable effect in INaT(Fig 5E) Analysis of the PT ratio for INa shows that exposure toavobenzone significantly reduced this value from493plusmn92 in control to 280plusmn232 at 01 mgml (P=009) 219plusmn269 at 02 mgml (P=003) and 121plusmn132 at 04 mgml(P=00004) (Fig 5F) which compares favourably with
overexpression of pum (cf Fig 3) We also observed a significantcorrelation between avobenzone concentration and the occurrence ofresurgent INa (P=0005 Chi-square test Fig 5G)
Our predicted mode of action for avobenzone is inconsistent withan immediate effect of this compound acting instead to potentiatePum which in turn downregulates Nav channels in the neuronalmembrane To test this we recorded from non-drug-exposed L3parabss aCC and used bath application of avobenzone (5 microM) Nochanges were observed in either component of INa (data not shown)and the PT ratio remained unaffected (Fig 5H) Higher doses(20 microM) or longer exposure times (10 min) similarly produced nodetectable effect (data not shown) This lack of acute effect isconsistent with our predicted mode of action Finally to directly testthis prediction we measured pum transcript abundance in parabss
L3 grown in the presence of avobenzoneWe observed a modest butstatistically significant increase in transcript abundance of sim20(12plusmn017 n=5 P=004 t-test vehicle control set as 1 Fig 5I)Thus we conclude that avobenzone acting to increase thetranscription andor transcript stability of pum is able to suppressseizure duration through downregulation of INaP Finally weobserved equally potent anticonvulsive activity of avobenzone intwo other bang-sensitive mutants easily-shocked (avobenzone142plusmn82 vs control 240plusmn120 s n=40 P=10times10minus5 L3electroshock) encoding an ethanolamine kinase (Pavlidis et al1994) and slamdance (avobenzone 178plusmn122 vs control 272plusmn108 s n=40 P=68times10minus5 L3 electroshock) encoding anaminopeptidase (Zhang et al 2002) indicative that increasingPum activity might be effective against a broad range of epilepsies
DISCUSSIONThe causes of seizure even in genetic epilepsies vary greatly andare not confined to genes with obvious contributions to ion fluxacross neuronal membranes This increases the challenge to identifyindividual mutations to determine the physiological role of both theWT and mutated protein and ultimately to design drugs tominimise the unwanted effect of the mutation In this study weidentify transcriptional changes that occur in the seizure-proneCNS We identify over 700 common genes that show alteredtranscription in two different seizure models It is noteworthy thatwe observed approximately double the number of genes showingaltered transcription in parabss flies compared with those treatedwith PTX The reason for this is unclear but might representaccumulated compensatory changes in the mutant line that haveoccurred in order to lessen the severity of seizure activity in parabss
mutants These additional genes warrant further investigation aspotential seizure suppressors
Many of the common transcriptional changes we identify and inparticular those that are upregulated (and thus open to inhibition bydrug exposure) might provide effective drug targets for novel AEDdesign However our attention was drawn to Pum which we havepreviously shown orchestrates homeostasis of action potential firingin both Drosophila and rat central neurons (Driscoll et al 2013Mee et al 2004) The degree of seizure suppression achieved byupregulating Pum in parabss flies is considerable and is onlymatched by the no-action-potential (napts) allele of the maleless(mle) locus in Drosophila which encodes an ATP-dependentdouble-stranded RNA (dsRNA) helicase (Ganetzky andWu 1982)This mutation causes a catastrophic change in splicing of theDrosophilaNav (Reenan et al 2000) The net effect of both of thesemanipulations increased Pum or the presence of napts is to reducethe availability of functional Nav expressed in central neurons Thedirection of change of pum in the two seizure models (that show
Table 2 List of compounds that reduce the fireflyRenilla (FFRen)luciferase ratio thus mimicking the activity of increased pumexpression (shown at bottom of table for reference)
All but avobenzone also reduce expression of the control Ren luciferasereporter that does not contain an NRE motif Luciferase values shown arenormalised such that 10 would represent no effect
146
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
reduced expression) might not be ideal with respect to drugdevelopment given that disruption of a gene or protein is often moreachievable Nevertheless we show that upregulation of pum in aDrosophila seizure mutant is potently anticonvulsive and furtherwe identify a potential lead anticonvulsive compound thatseemingly increases the level of expression of this homeostaticregulator This compound might catalyse the development of anovel class of AEDNeurons display an array of homeostatic mechanisms to maintain
action potential firing within pre-determined and physiologicallyappropriate limits (Davis 2013) Pum is a well-characterised RNA-binding protein that binds mRNA usually through a specific motiftermed the NRE Once bound Pum recruits additional cofactorsincluding Nanos and Brain tumor (Brat) to form a complex that issufficient to prevent translation (Wharton et al 1998) Our results inthis study indicate that increased expression of Pum might havetherapeutic benefit for seizure suppression However a potential
issue in this regard is that a genome-wide identification of RNAsbound to Pum in ovaries identifies upwards of 700 genes(FDRlt01) (Gerber et al 2006) This raises the problem ofspecificity of effect following global potentiation of level or activityof Pum This potential issue might however be overcome throughidentifying and targeting neuronal-specific regulators of Pum Onesuch alternative target might be the inhibition of Myocyte enhancerfactor 2 (Mef2)-induced expression of miR-134 in neurons that inturn inhibits translation of mammalian PUM2 (Fiore et al 2009)Additional possibilities include targeting of cofactors required forPum activity It is interesting in this regard that a loss-of-functionmutation in mei-P26 a homologue of Brat produces strong seizuresuppression in Drosophila bang-sensitive seizure mutants(Glasscock et al 2005)
Mammalian PUM2 binds transcripts encoding SCN1A (Nav11)and SCN8A (Nav16) (Driscoll et al 2013 Vessey et al 2010) Areduction in supply of Nav protein to the neuron membrane is
Fig 5 Avobenzone is anticonvulsant and selectively reduces INaP (A) parabss L3 raised in food containing 04 mgml avobenzone show significantly reducedrecovery time (RT) following electroshock compared with controls (CTRL parabss+DMSO) (B) Exposure of adult parabss flies to avobenzone (04 mgml) is alsopotently anticonvulsant compared with controls (CTRL parabss+DMSO) Each manipulation tested 10 flies per vial to produce an average value This wasrepeated five times and a final average calculated (C) Whole-cell patch recordings of INa from parabss L3 aCC raised in food containing 04 mgml avobenzoneshow reduced INaP (DE) Increasing concentrations of avobenzone (01 02 and 04 mgml) induced a proportional decrease of INaP (D) without affectingINaT (E) (F) Persistent-to-transient (PT) current ratio for INa recorded in aCC (G) The frequency of cells that exhibit resurgent INa correlates with avobenzoneconcentration (P=0005 Chi-square test) (H) PT ratio measured from parabss aCC before (CTRL) and after a 1 min bath application of 5 microM avobenzone(I) Analysis of pum transcript level in isolated CNS from parabss L3 raised on food containing avobenzone (04 mgml) shows a significant increase compared withparabss L3 raised on food containing an equal amount of vehicle (08 DMSO) The control value has been set to 1 Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (AH-I unpaired t-test D-F two-way ANOVA with Bonferronirsquos post hoc)
147
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
consistent with a reduction in action potential firing and a generalanticonvulsant effect (Mee et al 2004) Analysis of INa inmotoneurons indicates that a likely mechanism includes a markedreduction in INaP Increased INaP is associated with mutations inSCN1A that have been identified from individuals with epilepsy(Meisler and Kearney 2005) and is specifically reduced by AEDssuch as phenytoin valproate and lamotrigine (Stafstrom 2007) Inlight of this the anticonvulsant effect of increased pum expression isunderstandable That reducing pum expression through RNAi-mediated knockdown is proconvulsive is again both predictable andunderstandable However the effect of this manipulation on INa isnot so clear Rather than increasing INaP INaT is instead significantlyincreased together with a novel appearance of resurgent INa duringrepolarisation Increased INaT would be expected to reduce thethreshold for action potential firing (ie making firing more likely)whereas resurgent INa is associated with increased action potentialfiring frequency partly by reducing the refractory period (Griecoet al 2005) Although we have observed this current component inrecordings from seizure mutants (including parabss) it is rarelyobserved in WT or following expression of transgenic pumThe ability to manipulate Pum in vivo to determine its
anticonvulsive properties in rodent models of seizure will begreatly aided by the identification of chemical compounds thatdirectly potentiate either expression or activity state We report theuse of a suitable cell-based screen to identify such compounds andhighlight avobenzone as a potential lead compound for futuredevelopment The in vivo toxicity of avobenzone has not been wellestablished and although there are few reports of serious side effectsassociated with its use as an active ingredient of sunscreen itstendency to form free radicals might be a potential issue To ourknowledge this compound has not been used to treat neurologicaldisease and its mode of action in reducing seizure in Drosophilaremains to be determined Our observations that ingestion ofavobenzone result in increased expression of pum is indicative thatthis compound might mimic elements of the pathway that controlexpression of this homeostatic regulatorThe output of our screen also provides additional support for the
use of rapamycin to control seizure (Lasarge and Danzer 2014Russo et al 2013) indicative that this molecule might influenceneuronal homeostasis The identification of topoisomerase II as apotential target to control seizure also validates previousobservations reporting that inhibition of this class of nuclearprotein is anticonvulsant (Lin et al 2015 Song et al 2008)Finally that we identify that the increase in Pum activity byaniracetam might hint at an additional mode of action for this classof known anticonvulsants (Shiotani et al 2000) The relatedracetams levetiracetam and brivaracetam are currently in clinicaluse as AEDs exploiting their capability to bind and inhibit synapticvesicle protein 2A (SV2A) (Klitgaard et al 2016)In summary we present a description of transcriptional change
present in seizure-prone CNS We identify in particular that pumexpression is downregulated in both genetic and chemically inducedseizure models This mirrors the reported reduction in PUM2 inhuman TLE and in rats exposed to the proconvulsant pilocarpine(Wu et al 2015) It also provides a possible understanding for whyPum2 null mice exhibit spontaneous seizures (Siemen et al 2011)However it is perplexing that pum levels should decrease duringseizures given that the published model predicts an increase (Meeet al 2004) As reduced Pum levels are predicted to increaseneuronal excitability it seems that epileptic seizures are associatedwith a pathological dysregulation of pum expression We speculatethat this occurs because Pum can auto-regulate (the pum transcript
contains NRE motifs) Thus although the neuronal hyperactivityinduced by seizures will initially increase Pum expression theaccumulating Pum protein might feed back to downregulate its owntranscript (Gerber et al 2006) Sampling at later stages after seizureoccurrence might only report reduced Pum compared with non-seizure controls Indeed we have shown that upregulation of pum inthe Drosophila CNS through expression of a wild-type transgene(lacking NRE motifs) results in reduction of endogenous pumtranscript level (W-HL and RAB unpublished data) Preventionof this feedback achievable in this study through expression oftransgenic pum lacking an NRE or exposure to avobenzone holdssignificant promise for anticonvulsant therapy
MATERIALS AND METHODSFly stocksWild type (WT maintained in the Baines lab) was Canton-S parabss (bss1)which was obtained from Dr Kevin OrsquoDell (Institute of Molecular Cell andSystems Biology University of Glasgow UK) is detailed in Parker et al(2011) The parabss stock (and other transgenic lines used) were notbackcrossed to the CS stock Controls consisted of either untreated parabss
andor parental stocks (ie Gal4+ UAS+) and are stated in respectivefigure legends Slamdanceiso78 was obtained from Dr Mark Tanouye(Department of Environmental Science Policy and Management andDepartment of Molecular and Cell Biology University of CaliforniaBerkeley California USA) Easily-shocked2F was obtained from Dr KevinOrsquoDell RRa-Gal4 is expressed in only the aCC and RP2 motoneurons (Linet al 2012) We are able to discriminate between these neurons duringelectrophysiological recordings and use only the aCC neuron in this studyWe used Cha-Gal4(19B) to drive UAS-transgene expression in allcholinergic neurons which include excitatory premotor interneurons(Salvaterra and Kitamoto 2001) Pan-neuronal expression was achievedby combining elaV-Gal4 (Bloomington stock no 8760 3rd chromosomeinsert) with parabss UAS-pumRNAi was obtained from the ViennaDrosophila RNAi Center (stock no 101399) and UAS-pum is detailed inSchweers et al (2002) UAS-pum lacks NRE motifs that are present in the3prime-UTR of the endogenous pum gene All genetic crosses were maintained at25degC with the exception of overexpression of pum (larvae die as 1st or 2ndinstars) These experiments were maintained at 205degC Chemical-inducedseizure was achieved by raising WT larvae on food containing 025 mgmlPTX (P1675 Sigma Poole UK) until wall-climbing third instarabbreviated to L3 (Lin et al 2015)
Library construction and RNA sequencingCNSs were removed from 50 L3 (mixed sexes) and RNA extracted using theRNeasy mini kit (QIAGEN Hilden Germany) as described (Lin et al2015) RNA integrity and purity were determined using an Agilent 2200TapeStation system (Agilent Technologies Santa Clara CA) The RNA-sequencing library was created using an mRNA Seq library preparation kitas per manufacturerrsquos instructions (Illumina Inc San Diego CA) Thelibrary products were sequenced in paired-end reads using an IlluminaHiSeqTM 2000 RNA-sequencing data were analysed using edgeR(empirical analysis of digital gene expression in R) (Robinson et al2010) This analysis identified genes with altered levels of expression usinga threshold false discovery rate (FDR)le1 GO terms for BiologicalProcess Cellular Component Molecular Function and Kyoto Encyclopediaof Genes and Genomes (KEGG) pathway were used for annotations Weclassified differentially expressed genes using the Functional AnnotationCluster (FAC) tool available in the Database for Annotation Visualizationand Integrated Discovery (DAVID) (Huang et al 2009ab)
Validation of RNA-sequencing analysis by quantitative PCRQuantitative RT-PCR was performed using a SYBR Green I real-time PCRmethod (Roche LightCyclerreg 480 SYBR Green I Master MannheimGermany) as described in Lin et al (2015) RNAwas extracted from either 20adult heads (3 days old) or 20 L3 CNSs (mixed sexes) using the RNeasymicro kit (QIAGEN) Primer sequences (5prime to 3prime) used were actin-5C
148
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
reduced expression) might not be ideal with respect to drugdevelopment given that disruption of a gene or protein is often moreachievable Nevertheless we show that upregulation of pum in aDrosophila seizure mutant is potently anticonvulsive and furtherwe identify a potential lead anticonvulsive compound thatseemingly increases the level of expression of this homeostaticregulator This compound might catalyse the development of anovel class of AEDNeurons display an array of homeostatic mechanisms to maintain
action potential firing within pre-determined and physiologicallyappropriate limits (Davis 2013) Pum is a well-characterised RNA-binding protein that binds mRNA usually through a specific motiftermed the NRE Once bound Pum recruits additional cofactorsincluding Nanos and Brain tumor (Brat) to form a complex that issufficient to prevent translation (Wharton et al 1998) Our results inthis study indicate that increased expression of Pum might havetherapeutic benefit for seizure suppression However a potential
issue in this regard is that a genome-wide identification of RNAsbound to Pum in ovaries identifies upwards of 700 genes(FDRlt01) (Gerber et al 2006) This raises the problem ofspecificity of effect following global potentiation of level or activityof Pum This potential issue might however be overcome throughidentifying and targeting neuronal-specific regulators of Pum Onesuch alternative target might be the inhibition of Myocyte enhancerfactor 2 (Mef2)-induced expression of miR-134 in neurons that inturn inhibits translation of mammalian PUM2 (Fiore et al 2009)Additional possibilities include targeting of cofactors required forPum activity It is interesting in this regard that a loss-of-functionmutation in mei-P26 a homologue of Brat produces strong seizuresuppression in Drosophila bang-sensitive seizure mutants(Glasscock et al 2005)
Mammalian PUM2 binds transcripts encoding SCN1A (Nav11)and SCN8A (Nav16) (Driscoll et al 2013 Vessey et al 2010) Areduction in supply of Nav protein to the neuron membrane is
Fig 5 Avobenzone is anticonvulsant and selectively reduces INaP (A) parabss L3 raised in food containing 04 mgml avobenzone show significantly reducedrecovery time (RT) following electroshock compared with controls (CTRL parabss+DMSO) (B) Exposure of adult parabss flies to avobenzone (04 mgml) is alsopotently anticonvulsant compared with controls (CTRL parabss+DMSO) Each manipulation tested 10 flies per vial to produce an average value This wasrepeated five times and a final average calculated (C) Whole-cell patch recordings of INa from parabss L3 aCC raised in food containing 04 mgml avobenzoneshow reduced INaP (DE) Increasing concentrations of avobenzone (01 02 and 04 mgml) induced a proportional decrease of INaP (D) without affectingINaT (E) (F) Persistent-to-transient (PT) current ratio for INa recorded in aCC (G) The frequency of cells that exhibit resurgent INa correlates with avobenzoneconcentration (P=0005 Chi-square test) (H) PT ratio measured from parabss aCC before (CTRL) and after a 1 min bath application of 5 microM avobenzone(I) Analysis of pum transcript level in isolated CNS from parabss L3 raised on food containing avobenzone (04 mgml) shows a significant increase compared withparabss L3 raised on food containing an equal amount of vehicle (08 DMSO) The control value has been set to 1 Data are meansplusmnsd for n independent cellsstated in individual bars Ple005 Ple001 Ple0001 (AH-I unpaired t-test D-F two-way ANOVA with Bonferronirsquos post hoc)
147
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
consistent with a reduction in action potential firing and a generalanticonvulsant effect (Mee et al 2004) Analysis of INa inmotoneurons indicates that a likely mechanism includes a markedreduction in INaP Increased INaP is associated with mutations inSCN1A that have been identified from individuals with epilepsy(Meisler and Kearney 2005) and is specifically reduced by AEDssuch as phenytoin valproate and lamotrigine (Stafstrom 2007) Inlight of this the anticonvulsant effect of increased pum expression isunderstandable That reducing pum expression through RNAi-mediated knockdown is proconvulsive is again both predictable andunderstandable However the effect of this manipulation on INa isnot so clear Rather than increasing INaP INaT is instead significantlyincreased together with a novel appearance of resurgent INa duringrepolarisation Increased INaT would be expected to reduce thethreshold for action potential firing (ie making firing more likely)whereas resurgent INa is associated with increased action potentialfiring frequency partly by reducing the refractory period (Griecoet al 2005) Although we have observed this current component inrecordings from seizure mutants (including parabss) it is rarelyobserved in WT or following expression of transgenic pumThe ability to manipulate Pum in vivo to determine its
anticonvulsive properties in rodent models of seizure will begreatly aided by the identification of chemical compounds thatdirectly potentiate either expression or activity state We report theuse of a suitable cell-based screen to identify such compounds andhighlight avobenzone as a potential lead compound for futuredevelopment The in vivo toxicity of avobenzone has not been wellestablished and although there are few reports of serious side effectsassociated with its use as an active ingredient of sunscreen itstendency to form free radicals might be a potential issue To ourknowledge this compound has not been used to treat neurologicaldisease and its mode of action in reducing seizure in Drosophilaremains to be determined Our observations that ingestion ofavobenzone result in increased expression of pum is indicative thatthis compound might mimic elements of the pathway that controlexpression of this homeostatic regulatorThe output of our screen also provides additional support for the
use of rapamycin to control seizure (Lasarge and Danzer 2014Russo et al 2013) indicative that this molecule might influenceneuronal homeostasis The identification of topoisomerase II as apotential target to control seizure also validates previousobservations reporting that inhibition of this class of nuclearprotein is anticonvulsant (Lin et al 2015 Song et al 2008)Finally that we identify that the increase in Pum activity byaniracetam might hint at an additional mode of action for this classof known anticonvulsants (Shiotani et al 2000) The relatedracetams levetiracetam and brivaracetam are currently in clinicaluse as AEDs exploiting their capability to bind and inhibit synapticvesicle protein 2A (SV2A) (Klitgaard et al 2016)In summary we present a description of transcriptional change
present in seizure-prone CNS We identify in particular that pumexpression is downregulated in both genetic and chemically inducedseizure models This mirrors the reported reduction in PUM2 inhuman TLE and in rats exposed to the proconvulsant pilocarpine(Wu et al 2015) It also provides a possible understanding for whyPum2 null mice exhibit spontaneous seizures (Siemen et al 2011)However it is perplexing that pum levels should decrease duringseizures given that the published model predicts an increase (Meeet al 2004) As reduced Pum levels are predicted to increaseneuronal excitability it seems that epileptic seizures are associatedwith a pathological dysregulation of pum expression We speculatethat this occurs because Pum can auto-regulate (the pum transcript
contains NRE motifs) Thus although the neuronal hyperactivityinduced by seizures will initially increase Pum expression theaccumulating Pum protein might feed back to downregulate its owntranscript (Gerber et al 2006) Sampling at later stages after seizureoccurrence might only report reduced Pum compared with non-seizure controls Indeed we have shown that upregulation of pum inthe Drosophila CNS through expression of a wild-type transgene(lacking NRE motifs) results in reduction of endogenous pumtranscript level (W-HL and RAB unpublished data) Preventionof this feedback achievable in this study through expression oftransgenic pum lacking an NRE or exposure to avobenzone holdssignificant promise for anticonvulsant therapy
MATERIALS AND METHODSFly stocksWild type (WT maintained in the Baines lab) was Canton-S parabss (bss1)which was obtained from Dr Kevin OrsquoDell (Institute of Molecular Cell andSystems Biology University of Glasgow UK) is detailed in Parker et al(2011) The parabss stock (and other transgenic lines used) were notbackcrossed to the CS stock Controls consisted of either untreated parabss
andor parental stocks (ie Gal4+ UAS+) and are stated in respectivefigure legends Slamdanceiso78 was obtained from Dr Mark Tanouye(Department of Environmental Science Policy and Management andDepartment of Molecular and Cell Biology University of CaliforniaBerkeley California USA) Easily-shocked2F was obtained from Dr KevinOrsquoDell RRa-Gal4 is expressed in only the aCC and RP2 motoneurons (Linet al 2012) We are able to discriminate between these neurons duringelectrophysiological recordings and use only the aCC neuron in this studyWe used Cha-Gal4(19B) to drive UAS-transgene expression in allcholinergic neurons which include excitatory premotor interneurons(Salvaterra and Kitamoto 2001) Pan-neuronal expression was achievedby combining elaV-Gal4 (Bloomington stock no 8760 3rd chromosomeinsert) with parabss UAS-pumRNAi was obtained from the ViennaDrosophila RNAi Center (stock no 101399) and UAS-pum is detailed inSchweers et al (2002) UAS-pum lacks NRE motifs that are present in the3prime-UTR of the endogenous pum gene All genetic crosses were maintained at25degC with the exception of overexpression of pum (larvae die as 1st or 2ndinstars) These experiments were maintained at 205degC Chemical-inducedseizure was achieved by raising WT larvae on food containing 025 mgmlPTX (P1675 Sigma Poole UK) until wall-climbing third instarabbreviated to L3 (Lin et al 2015)
Library construction and RNA sequencingCNSs were removed from 50 L3 (mixed sexes) and RNA extracted using theRNeasy mini kit (QIAGEN Hilden Germany) as described (Lin et al2015) RNA integrity and purity were determined using an Agilent 2200TapeStation system (Agilent Technologies Santa Clara CA) The RNA-sequencing library was created using an mRNA Seq library preparation kitas per manufacturerrsquos instructions (Illumina Inc San Diego CA) Thelibrary products were sequenced in paired-end reads using an IlluminaHiSeqTM 2000 RNA-sequencing data were analysed using edgeR(empirical analysis of digital gene expression in R) (Robinson et al2010) This analysis identified genes with altered levels of expression usinga threshold false discovery rate (FDR)le1 GO terms for BiologicalProcess Cellular Component Molecular Function and Kyoto Encyclopediaof Genes and Genomes (KEGG) pathway were used for annotations Weclassified differentially expressed genes using the Functional AnnotationCluster (FAC) tool available in the Database for Annotation Visualizationand Integrated Discovery (DAVID) (Huang et al 2009ab)
Validation of RNA-sequencing analysis by quantitative PCRQuantitative RT-PCR was performed using a SYBR Green I real-time PCRmethod (Roche LightCyclerreg 480 SYBR Green I Master MannheimGermany) as described in Lin et al (2015) RNAwas extracted from either 20adult heads (3 days old) or 20 L3 CNSs (mixed sexes) using the RNeasymicro kit (QIAGEN) Primer sequences (5prime to 3prime) used were actin-5C
148
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
consistent with a reduction in action potential firing and a generalanticonvulsant effect (Mee et al 2004) Analysis of INa inmotoneurons indicates that a likely mechanism includes a markedreduction in INaP Increased INaP is associated with mutations inSCN1A that have been identified from individuals with epilepsy(Meisler and Kearney 2005) and is specifically reduced by AEDssuch as phenytoin valproate and lamotrigine (Stafstrom 2007) Inlight of this the anticonvulsant effect of increased pum expression isunderstandable That reducing pum expression through RNAi-mediated knockdown is proconvulsive is again both predictable andunderstandable However the effect of this manipulation on INa isnot so clear Rather than increasing INaP INaT is instead significantlyincreased together with a novel appearance of resurgent INa duringrepolarisation Increased INaT would be expected to reduce thethreshold for action potential firing (ie making firing more likely)whereas resurgent INa is associated with increased action potentialfiring frequency partly by reducing the refractory period (Griecoet al 2005) Although we have observed this current component inrecordings from seizure mutants (including parabss) it is rarelyobserved in WT or following expression of transgenic pumThe ability to manipulate Pum in vivo to determine its
anticonvulsive properties in rodent models of seizure will begreatly aided by the identification of chemical compounds thatdirectly potentiate either expression or activity state We report theuse of a suitable cell-based screen to identify such compounds andhighlight avobenzone as a potential lead compound for futuredevelopment The in vivo toxicity of avobenzone has not been wellestablished and although there are few reports of serious side effectsassociated with its use as an active ingredient of sunscreen itstendency to form free radicals might be a potential issue To ourknowledge this compound has not been used to treat neurologicaldisease and its mode of action in reducing seizure in Drosophilaremains to be determined Our observations that ingestion ofavobenzone result in increased expression of pum is indicative thatthis compound might mimic elements of the pathway that controlexpression of this homeostatic regulatorThe output of our screen also provides additional support for the
use of rapamycin to control seizure (Lasarge and Danzer 2014Russo et al 2013) indicative that this molecule might influenceneuronal homeostasis The identification of topoisomerase II as apotential target to control seizure also validates previousobservations reporting that inhibition of this class of nuclearprotein is anticonvulsant (Lin et al 2015 Song et al 2008)Finally that we identify that the increase in Pum activity byaniracetam might hint at an additional mode of action for this classof known anticonvulsants (Shiotani et al 2000) The relatedracetams levetiracetam and brivaracetam are currently in clinicaluse as AEDs exploiting their capability to bind and inhibit synapticvesicle protein 2A (SV2A) (Klitgaard et al 2016)In summary we present a description of transcriptional change
present in seizure-prone CNS We identify in particular that pumexpression is downregulated in both genetic and chemically inducedseizure models This mirrors the reported reduction in PUM2 inhuman TLE and in rats exposed to the proconvulsant pilocarpine(Wu et al 2015) It also provides a possible understanding for whyPum2 null mice exhibit spontaneous seizures (Siemen et al 2011)However it is perplexing that pum levels should decrease duringseizures given that the published model predicts an increase (Meeet al 2004) As reduced Pum levels are predicted to increaseneuronal excitability it seems that epileptic seizures are associatedwith a pathological dysregulation of pum expression We speculatethat this occurs because Pum can auto-regulate (the pum transcript
contains NRE motifs) Thus although the neuronal hyperactivityinduced by seizures will initially increase Pum expression theaccumulating Pum protein might feed back to downregulate its owntranscript (Gerber et al 2006) Sampling at later stages after seizureoccurrence might only report reduced Pum compared with non-seizure controls Indeed we have shown that upregulation of pum inthe Drosophila CNS through expression of a wild-type transgene(lacking NRE motifs) results in reduction of endogenous pumtranscript level (W-HL and RAB unpublished data) Preventionof this feedback achievable in this study through expression oftransgenic pum lacking an NRE or exposure to avobenzone holdssignificant promise for anticonvulsant therapy
MATERIALS AND METHODSFly stocksWild type (WT maintained in the Baines lab) was Canton-S parabss (bss1)which was obtained from Dr Kevin OrsquoDell (Institute of Molecular Cell andSystems Biology University of Glasgow UK) is detailed in Parker et al(2011) The parabss stock (and other transgenic lines used) were notbackcrossed to the CS stock Controls consisted of either untreated parabss
andor parental stocks (ie Gal4+ UAS+) and are stated in respectivefigure legends Slamdanceiso78 was obtained from Dr Mark Tanouye(Department of Environmental Science Policy and Management andDepartment of Molecular and Cell Biology University of CaliforniaBerkeley California USA) Easily-shocked2F was obtained from Dr KevinOrsquoDell RRa-Gal4 is expressed in only the aCC and RP2 motoneurons (Linet al 2012) We are able to discriminate between these neurons duringelectrophysiological recordings and use only the aCC neuron in this studyWe used Cha-Gal4(19B) to drive UAS-transgene expression in allcholinergic neurons which include excitatory premotor interneurons(Salvaterra and Kitamoto 2001) Pan-neuronal expression was achievedby combining elaV-Gal4 (Bloomington stock no 8760 3rd chromosomeinsert) with parabss UAS-pumRNAi was obtained from the ViennaDrosophila RNAi Center (stock no 101399) and UAS-pum is detailed inSchweers et al (2002) UAS-pum lacks NRE motifs that are present in the3prime-UTR of the endogenous pum gene All genetic crosses were maintained at25degC with the exception of overexpression of pum (larvae die as 1st or 2ndinstars) These experiments were maintained at 205degC Chemical-inducedseizure was achieved by raising WT larvae on food containing 025 mgmlPTX (P1675 Sigma Poole UK) until wall-climbing third instarabbreviated to L3 (Lin et al 2015)
Library construction and RNA sequencingCNSs were removed from 50 L3 (mixed sexes) and RNA extracted using theRNeasy mini kit (QIAGEN Hilden Germany) as described (Lin et al2015) RNA integrity and purity were determined using an Agilent 2200TapeStation system (Agilent Technologies Santa Clara CA) The RNA-sequencing library was created using an mRNA Seq library preparation kitas per manufacturerrsquos instructions (Illumina Inc San Diego CA) Thelibrary products were sequenced in paired-end reads using an IlluminaHiSeqTM 2000 RNA-sequencing data were analysed using edgeR(empirical analysis of digital gene expression in R) (Robinson et al2010) This analysis identified genes with altered levels of expression usinga threshold false discovery rate (FDR)le1 GO terms for BiologicalProcess Cellular Component Molecular Function and Kyoto Encyclopediaof Genes and Genomes (KEGG) pathway were used for annotations Weclassified differentially expressed genes using the Functional AnnotationCluster (FAC) tool available in the Database for Annotation Visualizationand Integrated Discovery (DAVID) (Huang et al 2009ab)
Validation of RNA-sequencing analysis by quantitative PCRQuantitative RT-PCR was performed using a SYBR Green I real-time PCRmethod (Roche LightCyclerreg 480 SYBR Green I Master MannheimGermany) as described in Lin et al (2015) RNAwas extracted from either 20adult heads (3 days old) or 20 L3 CNSs (mixed sexes) using the RNeasymicro kit (QIAGEN) Primer sequences (5prime to 3prime) used were actin-5C
148
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
(CG4027) CTTCTACAATGAGCTGCGT and GAGAGCACAGCCTGG-AT pum (CG9755) GCAGCAGGGTGCCGAGAATC and CGCGGCGA-CCCGTCAACG (forward and reverse respectively) Relative gene expressionwas calculated as the 2minusΔCt where ΔCt was determined by subtracting theaverage actin-5C Ct value from that of pum
Luciferase reporter constructionA region of the 3primeUTR (NM_1692332 2390-2650) of hunchbackcontaining two pum-binding motifs (NRE1 and NRE2) (Gupta et al2009) was subcloned from UAS-firefly-NREpUAST (a gift from Dr KevinMoffat University of Warwick UK) by releasing the DNA fragment usingEcoRI and XhoI sites and ligating it into pAc51 vector (Invitrogen) Renillaluciferasewas subcloned from pRL-CMV vector (Promega) by releasing theDNA fragment using NheI (filling the sticky end to blunt end with Klenow)and XbaI sites and ligating it into EcoRV and XbaI sites of pAc51 vector(Invitrogen)
Compound library screenS2R+ cells (15times104 cells in 15 microl of Schneiderrsquos Drosophila MediumGibco) were treated with 5 microl drug (final concentration 5 microM with 05DMSO) in 384-well plates (Selleckchem) for 48 h followed by co-transfection (Effectene QIAGEN) of firefly-NRE and renilla luciferasereporters (10 ng each) for a further 48 h The transfection procedure is asdescribed in the manufacturerrsquos instructions (QIAGEN) S2R+ cells werelysed with 035 Triton X-100 in BL buffer (50 mM HEPES 05 mMEDTA 036 mM phenylacetic acid and 007 mM oxalic acid) andD-Luciferin (046 mM Molecular Probes) was added to measure fireflyluciferase activity This was followed by adding coelenterazine-h (3 mMPromega) to measure renilla luciferase activity A Varioskan flash platereader (Thermo Scientific) was used to measure luminescence
Seizure behaviour testTwenty virgin females of parabss Cha-Gal4(19B) were mated with five malesof UAS-pumRNAi UAS-pum or WT Because parabss is on the Xchromosome and heterozygous parabss+ females show significantlyreduced recovery time we used parabssY male F1 progeny for behaviouralscreening For adult seizure determination male flies (3 days old) were testedat least one day after collection to ensure total recovery fromCO2-anaesthesiaTen flies were transferred to an empty plastic fly vial and left to recover for30 min before a mechanical shock induced by vortexing the vial at maximumspeed for 10 s Recovery time (RT) was calculated from the average timetaken for all 10 flies to recover from paralysis to standing (to produce a singlevalue) At least three replicates (of 10 flies per vial) were performed for eachcondition tested and the recovery time averaged across the three vialsAvobenzone was fed to young adult male flies (parabssY) within 8 h ofeclosion Groups of 10 flies were placed in an empty vial and exposed to drug-soaked filter paper Drug was first mixed with a sucrose solution (5) toproduce a final concentration of 04 mgml (16 DMSO) Filter papersoaked in this solution was added to vials and left for 24 h before testing
To measure seizure in larvae an electroshock assay was performed aspreviously described (Marley and Baines 2011) Briefly L3 male larvae(parabssY) were transferred to a plastic dish after washing to remove foodresidue and gently dried using paper tissue Once normal crawling behaviourresumed a conductive probe composed of two tungsten wires (01 mmdiameter sim1-2 mm apart) was positioned over the approximate position ofthe CNS on the anterior-dorsal cuticle of the animal A 30 VDCpulse for 3 sgenerated by a Grass S88 stimulator (Grass instruments RI USA) wasapplied In response to the electric stimulus we observed a transitory paralysisin which larvae tonically contracted and occasionally exhibited spasms Thetime to resumption of normal crawling behaviour was measured as RT Fordrug-feeding studies larvae were raised on food containing avobenzone(PHR1073 Sigma) in 08 DMSO until reaching L3
ElectrophysiologyWhole-cell voltage-clamp recordings were performed on aCC motoneuronsat L3 as previously described (Marley and Baines 2011) Leak currentswere subtracted on-line (P4) The same stimulation protocol was appliedthree times to each neuron and the recordings averaged Current amplitudes
were normalised for cell capacitance determined by integrating the area(1 ms time range) under the capacity transients elicited by stepping the cellfrom minus60 to minus90 mV for 30 ms Cells exhibiting no measurable INaP(resulting from excessive resurgent INa) were not included in the quantitativeanalysis
To evaluate the effect of pum manipulation on INa virgin females ofparabss RRa-Gal4 were crossed with UAS-pumRNAi UAS-pum or WTmales Only parabssY male F1 progeny was recorded at L3 To investigateavobenzone action parabss RRa-Gal4 larvae were raised on foodcontaining 08 DMSO or avobenzone at different concentrations (0102 and 04 mgml) until reaching L3 Acute drug treatment was performedby bath-applying avobenzone to the external saline (05 DMSO) INa wasrecorded from parabss RRa-Gal4 aCC motoneurons before and 1 min afterbath application Controls were exposed to DMSO alone
StatisticsStatistical significance between group means was assessed using either aStudentrsquos t-test (where a single experimental group is compared with asingle control group) or ANOVA followed by the Bonferronirsquos post hoc test(multiple experimental groups) The Chi-square test was used for statisticalanalysis of categorized data Data shown is meanplusmnsd
AcknowledgementsThe authors thank Miaomiao He Yuen Ngan Fan Nikki Leek and Ping Wang fortechnical support
Competing interestsThe authors declare no competing or financial interests
Author contributionsW-HL and RAB designed research W-HL and CNGG performed researchW-HL andCNGG analyzed dataW-HL CNGG and RAB wrote the paper
FundingThis work was supported by funding to RAB from the Biotechnology and BiologicalSciences Research Council (BBJ0050021 and BBL0276901) We are grateful toMedical ResearchCouncil Technology (MRCT) for provision of the drug libraryWorkon this project benefited from the Manchester Fly Facility established through fundsfrom the University of Manchester and the Wellcome Trust (087742Z08Z)
Data availabilityRNA-seq raw data is deposited in Harvard Dataverse and is available at doi107910DVN1N7EIG
Supplementary informationSupplementary information available online athttpdmmbiologistsorglookupdoi101242dmm027045supplemental
ReferencesBaines R A (2005) Neuronal homeostasis through translational control Mol
Neurobiol 32 113-121Callaghan D A and Schwark W S (1980) Pharmacological modification of
amygdaloid-kindled seizures Neuropharmacology 19 1131-1136Davis G W (2013) Homeostatic signaling and the stabilization of neural function
Neuron 80 718-728Driscoll H E Muraro N I He M and Baines R A (2013) Pumilio-2 regulates
translation of nav16 to mediate homeostasis of membrane excitabilityJ Neurosci 33 9644-9654
Escayg A and Goldin A L (2010) Sodium channel SCN1A and epilepsymutations and mechanisms Epilepsia 51 1650-1658
Fiore R Khudayberdiev S Christensen M Siegel G Flavell S W Kim T-K Greenberg M E and Schratt G (2009) Mef2-mediated transcription of themiR379-410 cluster regulates activity-dependent dendritogenesis by fine-tuningPumilio2 protein levels EMBO J 28 697-710
Ganetzky B and Wu C F (1982) Indirect suppression involving behavioralmutants with altered nerve excitability in Drosophila melanogasterGenetics 100597-614
Gerber A P Luschnig S Krasnow M A Brown P O and Herschlag D(2006) Genome-wide identification of mRNAs associated with the translationalregulator PUMILIO in Drosophila melanogaster Proc Natl Acad Sci USA 1034487-4492
149
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
Disea
seModelsampMechan
isms
Glasscock E Singhania A and Tanouye M A (2005) The mei-P26 geneencodes a RING finger B-box coiled-coil-NHL protein that regulates seizuresusceptibility in Drosophilia Genetics 170 1677-1689
Grieco T M Malhotra J D Chen C Isom L L and Raman I M (2005)Open-channel block by the cytoplasmic tail of sodium channel beta4 as amechanism for resurgent sodium current Neuron 45 233-244
Gupta Y K Lee T H Edwards T A Escalante C R Kadyrova L YWharton R P and Aggarwal A K (2009) Co-occupancy of two Pumiliomolecules on a single hunchback NRE RNA 15 1029-1035
Huang D W Sherman B T and Lempicki R A (2009a) Bioinformaticsenrichment tools paths toward the comprehensive functional analysis of largegene lists Nucleic Acids Res 37 1-13
Huang D W Sherman B T and Lempicki R A (2009b) Systematic andintegrative analysis of large gene lists using DAVID bioinformatics resources NatProtoc 4 44-57
Klitgaard H Matagne A Nicolas J-M Gillard M Lamberty Y De Ryck MKaminski R M Leclercq K Niespodziany I Wolff C et al (2016)Brivaracetam rationale for discovery and preclinical profile of a selective SV2Aligand for epilepsy treatment Epilepsia 57 538-548
Lasarge C L and Danzer S C (2014) Mechanisms regulating neuronalexcitability and seizure development following mTOR pathway hyperactivationFront Mol Neurosci 7 18
Lin W-H Gunay C Marley R Prinz A A and Baines R A (2012) Activity-dependent alternative splicing increases persistent sodium current and promotesseizure J Neurosci 32 7267-7277
Lin W-H He M and Baines R A (2015) Seizure suppression throughmanipulating splicing of a voltage-gated sodium channel Brain 138 891-901
Marley R and Baines R A (2011) Increased persistent Na+ current contributesto seizure in the slamdance bang-sensitive Drosophila mutant J Neurophysiol106 18-29
Mee C J Pym E C Moffat K G and Baines R A (2004) Regulation ofneuronal excitability through pumilio-dependent control of a sodium channelgene J Neurosci 24 8695-8703
Meisler M H and Kearney J A (2005) Sodium channel mutations in epilepsyand other neurological disorders J Clin Invest 115 2010-2017
Muraro N I and Baines R A (2008) Drosophila melanogaster in the study ofepilepsy SEB Exp Biol Ser 60 141-160
Muraro N I Weston A J Gerber A P Luschnig S Moffat K G andBaines R A (2008) Pumilio binds para mRNA and requires Nanos and Brat toregulate sodium current in Drosophila motoneurons J Neurosci 28 2099-2109
Parker L Padilla M Du Y Dong K and Tanouye M A (2011) Drosophila asamodel for epilepsy bss is a gain-of-functionmutation in the para sodium channelgene that leads to seizures Genetics 187 523-534
Pavlidis P Ramaswami M and Tanouye M A (1994) The Drosophila easilyshocked gene a mutation in a phospholipid synthetic pathway causes seizureneuronal failure and paralysis Cell 79 23-33
Reenan R A Hanrahan C J and Ganetzky B (2000) The mle(napts) RNAhelicase mutation in Drosophila results in a splicing catastrophe of the para Na+channel transcript in a region of RNA editing Neuron 25 139-149
Robinson M D McCarthy D J and Smyth G K (2010) edgeR a Bioconductorpackage for differential expression analysis of digital gene expression dataBioinformatics 26 139-140
Rundfeldt C Honack D and Loscher W (1990) Phenytoin potently increasesthe threshold for focal seizures in amygdala-kindled rats Neuropharmacology 29845-851
Russo E Citraro R Donato G Camastra C Iuliano R Cuzzocrea SConstanti A and De Sarro G (2013) mTOR inhibition modulatesepileptogenesis seizures and depressive behavior in a genetic rat model ofabsence epilepsy Neuropharmacology 69 25-36
Salvaterra P M and Kitamoto T (2001) Drosophila cholinergic neurons andprocesses visualized with Gal4UAS-GFP Brain Res 1 73-82
Schweers B A Walters K J and Stern M (2002) The Drosophilamelanogaster translational repressor pumilio regulates neuronal excitabilityGenetics 161 1177-1185
Shiotani T Nakamoto Y Watabe S Yoshii M and Nabeshima T (2000)Anticonvulsant actions of nefiracetam on epileptic EL mice and their relation toperipheral-type benzodiazepine receptors Brain Res 859 255-261
Siemen H Colas D Heller H C Brustle O and Pera R A R (2011) Pumilio-2 function in the mouse nervous system PLoS ONE 6 e25932
Song J and Tanouye M A (2008) From bench to drug human seizure modelingusing Drosophila Prog Neurobiol 84 182-191
Song J Parker L Hormozi L and Tanouye M A (2008) DNA topoisomerase Iinhibitors ameliorate seizure-like behaviors and paralysis in a Drosophila model ofepilepsy Neuroscience 156 722-728
Stafstrom C E (2007) Persistent sodium current and its role in epilepsy EpilepsyCurr 7 15-22
Turrigiano G (2012) Homeostatic synaptic plasticity local and globalmechanisms for stabilizing neuronal function Cold Spring Harb Perspect Biol4 a005736
Vessey J P Schoderboeck L Gingl E Luzi E Riefler J Di Leva F KarraD Thomas S Kiebler M A and Macchi P (2010) Mammalian Pumilio 2regulates dendrite morphogenesis and synaptic function Proc Natl Acad SciUSA 107 3222-3227
Weston A J and Baines R A (2007) Translational regulation of neuronalelectrical properties Invert Neurosci 7 75-86
Wharton R P Sonoda J Lee T Patterson M and Murata Y (1998) ThePumilio RNA-binding domain is also a translational regulatorMol Cell 1 863-872
Wu X-L Huang H Huang Y-Y Yuan J-X Zhou X and Chen Y-M (2015)Reduced Pumilio-2 expression in patients with temporal lobe epilepsy and in thelithium-pilocarpine induced epilepsy rat model Epilepsy Behav 50 31-39
Yasuyama K and Salvaterra P M (1999) Localization of cholineacetyltransferase-expressing neurons in Drosophila nervous system MicroscRes Tech 45 65-79
Zhang H Tan J Reynolds E Kuebler D Faulhaber S and Tanouye M(2002) The Drosophila slamdance gene a mutation in an aminopeptidase cancause seizure paralysis and neuronal failure Genetics 162 1283-1299
150
RESEARCH ARTICLE Disease Models amp Mechanisms (2017) 10 141-150 doi101242dmm027045
