Mechanisms
of
vision
in
the
fruit
fly
Lucia
de
Andres-Bragado
and
Simon
G
Sprecher
Vision isessential to maximizetheefficiency of dailytasks suchas
feeding,avoidingpredatorsorfindingmatingpartners.An
advantageousmodelisDrosophilamelanogaster,sinceitoffers
toolsthatallowgeneticandneuronalmanipulationwithhigh
spatialandtemporalresolution,whichcanbecombinedwith
behavioral,anatomicalandphysiologicalassays.Recent
advanceshaveexpandedourknowledgeontheneuralcircuitry
underlyingsuchimportantbehaviorsascolorvision(roleof
reciprocalinhibitiontoenhancecolorsignalatthelevelofthe
ommatidia);motionvision(motion-detectionneuronesreceive
both excitatory and inhibitoryinput), and sensoryprocessing (role
ofthecentralcomplexinspatialnavigation,andinorchestrating
theinformationfromothersensesandtheinnerstate).Research
onsynergiesbetweenpathwaysisshapingthefield.
Address
DepartmentofBiology,UniversityofFribourg,Fribourg,Switzerland
Correspondingauthor:Sprecher,SimonG(simon.sprecher@unifr.ch)
Introduction
The importance of vision for survival cannot be over-stated for a vast majority of animals. Therefore, many specieshavedeveloped extremelycomplex andprecise visualsystems,andtheyallocateafairamountofenergy building them and maintaining them. Light detection impliescollectingdataaboutitsphysicalattributessuch as intensity, wavelength or polarization. Moreover, this information must be later decoded in the brain before motordecisionsaremade.Suchprocessingofthephysical input is necessary to discriminate patterns, shades, changes of illumination, and motion, either related to movingobjectsorto theanimal’sown displacement. Visualperception starts in theeye, and thereare many differentkindsofeyeswithawiderangeofpurposesand sensitivities;suchas eyecups,eyespotsorin thecaseof insects,thecompoundeye.Drosophilacompoundeyesare typically composed of around 750 smaller units called
ommatidia(Figure 1a,b).Each of theseommatidia has both a simple lens and several photoreceptors, with specialized membrane structures (rhabdomeres) in which light-sensing Rhodopsins are located. The Rhodopsinproteinsareorganizedwithinarhabdomeric structure and are predominantly oriented along one axis, thereby making insect photoreceptor cells inherently polarization-sensitive [1]. Light comes in andstimulates the retinal molecule thatis embedded within Rhodopsin, triggering the isomerization of retinal.Thisconformationalswitchchangestheaffinity of Rhodopsin from guanosine diphosphate (GDP) to Guanosine-5’-triphosphate(GTP),whichsubsequently triggers the downstream phototransduction cascade [2,3] (Figure1b,c).
The visual information is then further processed, in particular in a specialized region of the insect brain: the optic lobes. One optic lobe can be functionally divided into three neuropiles: the lamina, the medulla andthelobulacomplex(Figure1d),whichdependingon thespecies are either anatomically combined or subdi-vided.Theanatomicalreferencemapdoneby[4]enables acomparison betweendifferentinsectspecies.
Inthisreview,wewillfocusonthelatestresearchdonein threekeytopicsofDrosophilavision:colorvision,motion vision and multisensory integration. These three fields have benefited from the genetic tools available for Drosophilaandfromrecentconnectomicsandfunctional studiesto elucidate thefunction ataneuronaland ata circuit level. We will briefly compare the state of these fields in insects, mainly focusing on Drosophila melanogaster.
Decoding
chromatic
information:
from
the
eye
to
the
brain
Color-vision is important for survival as it confers rich informationabouttheexternalworld,suchasthetimeof thedayorthetypeoffoodandmatingpartners.Whatis commonlyknownascolor-visioncanmakereferenceboth to ‘true color vision’, where chromatic information is sensed independently of the intensity; and to ‘wavelength-specificbehavior’,which isabehavior that is dependenton the intensity within each wavelength. Theprocessingofthischromaticinformationreliesboth on physical structures in the eyes and in underlying neuronalcircuitstodecodethisinformationinthebrain. Attheleveloftheretina,colorvisiondependsonseveral photoreceptorproperties,suchastheirspectralsensitivity [5–7]ortheirspatialdistribution[8].InDrosophila,there
http://doc.rero.ch
Published in "Current Opinion in Insect Science 36(): 25–32, 2019"
which should be cited to refer to this work.
are several types of photoreceptor cells: six outer ones (R1–R6), which project their axons to the first visual neuropile – the lamina, and two inner ones (R7 and R8), which extend their axons to the second visual
neuropile – the medulla (Figures 1d, 2 a). The outer R1–R6photoreceptorshaveabroadbandspectraltuning, whiletheinnerR7andR8photoreceptorshaveanarrower one [7]. Even though R1–R6 have traditionally been
Figure1
Cornea
R8
R7
R1 - R6
hv Ca2+ TRP GDP GTPPLC
β Gqα GDP GTP Na+ β γ PIP2 DAG InsP3R
(c) M* TRPL Gqαβ γCentral complex
Optic lobe
Ommatidia (a) Re La Me Lo LoPPB
FB
EB
No
(e)Re
R1 - R6 R7 R8 Dm Tm L1 (d)La
Dm Dm L1Lo
LoP
R7 R7R8R8Lo
Tm TmMe
R1 R1Current Opinion in Insect Science
AnatomicalstructuresinvolvedinvisualsensoryprocessingintheDrosophilamelanogasterbrain.
(a)Generalvisionofaflybrain.Lightgoesinthroughtheretina(Re),whereitexcitesthephotoreceptorcells.Thesephotoreceptorcellstarget
differentstructuresintheopticlobe:thelamina(La),medulla(Me)andlobulacomplex,whichisinturncomposedofthelobula(Lo)andthelobula
plate(LoP).Thecentralcomplexisastructurespecializedinsensoryintegrationthatislocatedinthemiddleofthebrain.(b)Enlargedviewofan
ommatidiumcontainingthecorneaandphotoreceptorsR1–R8.TheR1–R6photoreceptorsarelocatedintheouterpartoftheommatidia,whileR7
andR8arelocatedinthemiddle,R7beingontopofR8.Lightgoesinthroughthelens(right)andenterstherhabdomeresofthedifferent
photoreceptorstostimulaterhodopsin.(c)Drosophilarhabdomere,wherethekeyplayersinvolvedinthephototransductioncascadewithinthe
rhabdomeremembraneareshown.Afterlightenterstheretina,aphotontriggerstheisomerizationofretinalwithintherhodopsinmolecule,which
leadstotheconformationalchangeofrhodopsin(R)tometarhodopsin(M*).ThistriggerstheactivationofaheterotrimericG-protein(Gaq)that
activatesthephospholipaseCb(PLCb)andhydrolysesPIP2intoInsP3andthemembrane-boundDAG.Thisultimatelyleadstotheopeningof
TRPandTRPLchannels,whichleadstotheentranceofCa2+ionsandNa+ions.(d)EnlargedviewoftheDrosophilavisualsystemincludingthe
retina(Re)andfourneuropileswithintheopticlobe.Theouterphotoreceptors(R1–R6)projecttothelamina,whiletheinnerphotoreceptors(R7
andR8)projecttothemedulla.Differenttypesofneuronesareshown:theLaminaneurons(Lm)aredirectlydownstreamofR1–R6
photoreceptors.Thetransmedulla(Tm)neuronsspanthemedullaandprojecttothelobula,whilethedistalmedulla(Dm)neuronsprojecttodistal
layersofthemedulla.(e)Enlargedviewofthecentralcomplex.Itiscomposedoftheprotocerebralbridge(PB),thefan-shapedbody(FB),the
ellipsoidbody(EB),andthenoduli(NO).
associatedwithmotion-detection,recentworkhasshown thattheyalsoplayaroleincolor-detection[9].Incontrast, theinnerR7andR8photoreceptorsaresufficientforcolor vision,althoughtheyhaverecently beenshown tohave someroleinthemotion-detectionpathwayas well[10].
This shows increasing evidence that there is crosstalk betweenbothcircuits.
TherearedifferenttypesofR7andR8photoreceptors:in thedorsalrimareawecanfindafewommatidiathatare
Figure2 “pale” ommatidia
-R8p ON /R7pOFF R8pOFF/R7pON R8y ON /R7yOFFR8yOFF/R7yON
Sensitivity Sensitivity 400 500 600 400 500 600 700 λ (nm) λ (nm) Hassenstein-Reichardt Barlow-Levick Gruntmann et. al 2018
x
:
PD ND PD ND τ τ T4 neurones Retinal ganglion cellsPD ND
B
τ τ spatial filters temporal filters Beetle -(c) (b) “yellow” ommatidiaRe
La
Me
LoP
Lo
Colour
Motion
T4
T5
R7
R8
R1-6 ONL1
L2
L3
Dm8 Tm5c Tm9 OFF OFF OFF Tm3 Mi1 (a)Current Opinion in Insect Science
ColorvisionandmotionvisionaretwonearlyindependentpathwaysinDrosophilavision.
(a)DiagramofthetwoparallelpathwaysinDrosophilavisionfromtheretina(Re)totheopticlobe.TheinnerR7andR8photoreceptorsprojectto
themedulla(Me)andaremostlyinvolvedincolorvision.BoththedistalmedulaneuronDm8andthetransmedullaneuronTm5aredownstreamof
theR7pathway.R1–R6photoreceptorsprojecttothelamina(La)andaremostlyinvolvedinmotiondetection.ThelaminaneuronesL1–L3receive
informationfromR1-R6andthensynapsetothetransmedullaneuroneslikeTm9,whichinturnsynapsetothemotion-sensitiveT4andT5
neuronesinthelobula(Lo)andlobulaplate(LoP).(b)Classicalmotion-detectionmodelsincludedetectorsthatarespatiallyseparatedandthat
makeuseoftemporalfilters(t)whichultimatelyleadeithertothesignalbeingamplifiedinthepreferreddirection(PD,Hassenstein–Reichardt
model)orsuppressedinthenulldirection(ND,Barlow–Levick).TheHassenstein–Reichardtmodelwasformulatedbasedondatafromthe
Chlorophanusbeetle,whiletheBarlow–Levickmodelwasbasedonexperimentaldatafromrabbitretinalganglioncells.Morerecentmodels,like
theoneproposedbyGruntmanetal.[21]forDrosophilaT4neurones,incorporateexperimentalevidencethattheyhavebothexcitatoryand
inhibitoryinputthatarespatiallyseparatedandalsotakesintoaccountsomebiophysicalnonlinearity(b).Themodelproposesbothspatialand
temporalfilters,whichultimatelyleadtoasuppressionofthenulldirection.(c)‘pale’and‘yellow’ommatidiadifferintherhodopsinexpressionin
R7andR8photoreceptorsandthereforeintheirspectralpreference.RecentevidencefromRef.[13]hasshownthatwithin‘pale’or‘yellow’
ommatidia,R7andR8inhibiteachother,whichcanbeseenintheirspectraltuningwhenshownacompositelightstimulus.
responsible for polarization vision, although the most common ones are distributed throughout the eye and arecalled‘yellow’or ‘pale’dependingonwhich pairof rhodopsinsareexpressedin R7andR8.
Truecolorvisioncannotbeexecutedbyasingle photo-receptor onits own and requires complexfurther com-parisons as can be simple subtractions. These compar-isons have been shown to take place in the synaptic terminalsofR7andR8intheflymedullathroughmutual inhibition[11,12].Recentworkusingtwo-photoncalcium imagingwithgeneticallyencodedcalciumindicators,has shown that the signal from R7 and R8 undergoes a comparison within one ommatidium. The mechanism usedforthiscomparison isthroughreciprocalinhibition atthepresynapticterminal[13](Figure2c).
However,colorvisionisnotonlyrelatedtothedetection of photons of a given wavelength:the chromatic visual inputmust befurtherdecodedin thebrain.The direct synaptictargetsofR7andR8areneuronespresentinthe medulla: either transmedulla (Tm), or distal medulla (Dm) amacrine neurones [11]. The determination of thedownstreamsynapticpartnersofR7alongwith func-tionalstudieshaveshownthatthereisasingle pathway mediatingtheinnateUVpreference,whichinvolvesboth Dm8andTm5c[11,14](Figure 2a).Thiscontrastswith thegreat redundancy found in the color-discrimination pathwayforlongerwavelengths,suchasgreenandblue. Fourtransmedulla neurones (Tm5a-c and Tm20) have been shown to be involved in this behavior but the inactivation of any single one of these neurones is not enoughtopreventcolor-learning [15].
Morerecentstudieshavefocusedinthereconstructionof visualprojectionneuronesconnectingthelobulawiththe centralbrain.TheuseofamodifiedGFPreconstruction across synaptic partners (GRASP) [16] has allowed the characterization of the downstream targets of the chro-matic Tm neurones in the lobula [17]. Furthermore, some of these visual projection neurones (VPNs), also calledlobulacolumnarcells,canalreadylinkvisualinput to specific behavioral actions, such as taking off before flyingor walkingbackwards[18].
Motion-vision:
the
update
of
a
paradigm
Aclassicaspectofinsectvisionwithgreathistoric signifi-cance is motion-detection. To avoid predators and to captureprey,insectsneedtodetermineboththeirflying directionw.r.tfixedreferencesandtheirrelativepositionto other movingindividuals.However, motiondetection is not trivial:itcannot onlyrely onsignal detectionatthe photoreceptorlevel,butitrequirescomplexcomputations inthebrainitself.Therefore,theinterestinmodels mim-ickingneural computations has never stopped growing. Classical examples wouldbe theHassenstein-Reichardt correlator [19], which was developed from behavioralstudiesonthebeetleChlorophanus,ortheBarlow-Lewick extension[20],whereaninhibitoryorexcitatorysignalis generatedaccordingtothedirectionofmotionandthese signals are subsequently compared to detect motion (Figure 2b, for reviews check [24–27]). More recently, Gruntmanet al.havedocumentedthe roleofT4onthe detectionoftheoffsetsbetweenexcitatoryandinhibitory inputs[21],andBorstetal.havemappedtheneuralcircuit intotheONandOFFpathways[22].Furtherresearchis needed to clarify a few elements, like therole of non-linearity,whichis presentin theHassenstein–Reichardt correlator,andhasbeenchallengedbyGruntmanetal.,who claimthatnon-linearityisnotneeded.Anotherexample wouldbegroupsonlyobservingnull-directionsuppression, butnopreferreddirectionenhancement,whileBorstetal. arguethatbothexist.Clearly,possessingsuchmodelshas value in itself, and the interest to develop them is not reducedtoadultDrosophila.Anexampleofthisis therecent successful application of a generalized Markov chainto modelthetaxisofDrosophilalarvae[23].Approachesbased onacostfunctionandaMarkovian-likeprocessweighting theinvolveddecision-makinglandscapeopenupa quanti-tativewayofmeasuringthebalancebetweenstochasticand drivenprocessesoperatinginthebrain(chanceand neces-sity), which are ingrained after a robust evolutionary process.
To be able to compute motion-vision, it isessential to havedetectorswithreceptivefieldsthatareseparatedin space, whose input can be temporally compared. As a consequence, downstream of the detectors, flies have direction-selective neurones: cells that respond differ-ently tovisualcuesmovingin oppositedirections. InDrosophila,thefirstrealdirection-selectivecellsalong thepathwaywerefoundalongtimeago:TheT4andT5 neuronesdifferentiallyrespondeithertolightincrements (T4,ON pathway)or lightdecrements(T5, OFF path-way) in a direction-selective manner [28,29] and are crucial components for motion detection [28]. Further-more, bothT4 and T5 have fourdifferent subtypes of neurones (T4a-d and T5a-d), which are all columnar neuronesandextendtheirdendritesoverafewcolumns. Each one of these subtypes of neurones is tuned for detecting movementin one thefourcardinaldirections andconsequentlyprojectstooneofthefourlayersofthe lobula plate[28,30].T4 andT5 are postsynapticto the medulla neurons, which have been extensively studied bothattheanatomicallevelandatthephysiologicallevel [26–30].Thesemedulla neurons arepostsynaptic toL1 andL2neuronsinthelamina,whichinturnreceivethe visualinputfromR1-R6photoreceptors[31](Figure2a). However, T4 and T5 are the first neurones in this network thatshowdirection-selectivity[28].
RecentworkfocusedonthesynapticinputstoT4andT5 hasshownthattheyreceivebothexcitatoryandinhibitory
inputs [32,33,34]. Furthermore, anatomical efforts to map these neurones using novel high-precision EM andcalcium imaging assays,have shown thatthere isa spatial separation betweenthe two components, which canbeproducedbyspatialortemporaldelaysofinputs alongspatiallyorientedT4dendriticbranches[34,35]. Most recently,Gruntman etal. [21]have usedinvivo whole-cellrecordingsofT4tofindthatdirectional selec-tivityderivesfromthecombinationofspatiallyoffsetfast excitatoryandslow inhibitoryinputs.Such aconclusion has been reinforced by simulations that prove how a significantpartofthedirectionalinformationdrawnfrom the T4 dendrites is related to different input signals acquired over spatially separated regions of the flies’ eye.Apartfromthespatialseparation,bothtypesofinput differintheirtemporalprofile: excitationisafastevent whileinhibitionisslowerone[21].Theserecent find-ingshavechallengedtheclassicalmotion-detection mod-els,andleadtotheformulationofnewerandintegrative ones,whicharebasedonpassiveconductancebiophysical properties.Thishaspromptedmodelsthatdescribethe physiologicalproperties of theT4neurones more accu-rately than before byexploiting theidea of combining excitation and inhibition both temporally and spatially [21,22](Figure2b).
Sensory
integration
Oncethevisualinputisperceivedandprocessedinthe earlyprocessingcentres,suchastheretinaandtheoptic lobe, the signal is transmitted to higher brain regions through the visual projection neurones (VPNs). The visualinputmustbeintegratedwithothersensory infor-mationaswellaswiththeindividual’sinnerstatebefore the appropriate response is coordinated. In arthropods, thestructurethathasbeenlinkedtothisprocessiscalled thecentral complex.
The central complex isa highly conservedstructure in insects, which is organized in modular neuropiles (For review, see Ref. [36]).These neuropiles are called the ellipsoidbody(EB),thefan-shapedbody(FB),thenoduli (NO)andtheprotocerebralbridge(PB)[4].Thefunction oftheseneuropilesseemstobelargelyconservedamong differentinsect speciesand it hasrecently beenshown that some of these neuropiles come from homologous developmentallineages in fliesandgrasshoppers [37]. The central complex has long been known to be involvedindiversefunctionssuchasmemoryandmotor control. Recently, there have been a lot of studies analyzingitsroleinprocessingvisualinformation.Most ofthestudieshinttowardstheimportanceofthecentral complexinspatialorientationthroughdifferent mecha-nisms:bytheuseoflandmarkdetection[38,39],orusing theso-called‘compasscells’tocalculatethepositionof the sun [40], or the animal’s heading, as encoded by head-directioncells [39,41,42].
Moreover,work by Omotoet al. hasstarted elucidating the way in which the central complex receives visual input,andthelineagesfromwhichtheseneuronescome[
43]. It had also been previously hypothesized that the visualsystemmustinputintotheellipsoidbodyneurones astheseneuronesaresensitivetovisualinput[38].The inputofvisualneuronesintothecentralcomplexhasalso been analysedin other insects,also suggesting thatthe centralcomplexmighthavearoleinefficientnavigation aidedbythedetectionofpolarized lightintheUV-rich sky[44,45].
Thecentralcomplexisastructurewherewecanseehow theuseofgenetictechniquesavailableforDrosophila[46] combinedwiththeefforttohaveamorphologicalmapof thestructure with asingle-cell level resolution [47–49] havepavedthewayforfunctionalpredictions.For exam-ple,Turner-Evansetal.havemadeuseoftheanatomical dataavailablefortwooftheCentralComplexneuropiles (EBandPB)to proposeaconceptualmodelof howthe brainwouldbeabletokeeptrackofcompass-like infor-mation. For such a model, they had to make several hypotheses: Firstly, that a group of neurones called P-EN would encode the animal’s angular velocity. Sec-ondly, that the activity of thesubpopulations of P-EN neuronesonbothsidesof thePBwouldco-localizeina ‘bump’,whichcorrespondstoagroupofneuronesthatare activeatthesametime,andthatitwouldencodeforthe fly’srotationalvelocity andheading. Lastly,that the P-ENgroup of neuroneswouldupdateasecond groupof neurones, the so-called ‘compass neurones’, which encodethefly’sheadingbothwhentheflyturnsin the darkness and when there are visual cues. Using the genetictoolsavailableforD.melanogaster,theywereable to record the synaptic activity of these neurones using calciumimagingandelectrophysiologyandconfirmthese assumptions. Furthermore, they recorded the synaptic activityofboth populationsof neuronessimultaneously and were able to complete the model and add the requirementof athird additional component: recurrent inhibition[42].Despiteotherspeciesnothaving sucha vastnumberofgenetictools,abodyofworkonthecentral complex is also being generated due to physiological recordingsandanatomical mappings[50].
Apart from the central complex, there are other more directcircuitsthatareinchargeofvisualsensory integra-tion.Thisisthecaseofsomedescendingneuronesthat aredownstreamoflarge-fieldvisualinterneuronsandthat synapse directly into motor centres, consequently con-trollingflightbehavior[51]orthelooming-sensitive neu-ronesthatdirectlysynapseontothegiantfibredrivingan escaperesponse[52,53].
Another form of sensory integration is to relate the sensory input to the animal’s inner state. The central complexhasbeen shownto beinvolvedin this typeof
sensoryintegrationforalongtimebutnewerworkshows thatotherstructures,suchastheopticlobes,areinvolved inthebehavioralmodulationofvisualprocessing[54–57]. Morerecentworkhasalsoshownthatthebehavioralstate (eitherwalkingortetheredflyingflies)playsaroleinthe activityofneuronesinthefan-shapedbody:These neu-rones are only responsive to visual stimuli in tethered flyingflies[58,59].Interestingly,Zachariasetal.havealso shownthatdifferentbehavioralstates(suchasthespeed of walking) also influence quick responses such as the onesprocessedbythelooming-sensitiveneurones[60].
Discussion
Visualprocessing inDrosophilaisdividedinto two par-allel pathways within the first neuropiles of the optic lobe: motion vision and color vision. In turn, motion visioniscomputedintwoparallelpathwaysaswell:one thatisspecializedindetectionoflightincrements(ON pathway)andanotheronethatisspecializedindetecting lightdecrements (OFF path-way). This parallelization of vision tasks presumably confers a high efficiency. After this first decoding of visual information, it is subsequentlypooledinthecentralcomplexwithinputs fromothersenses and with theflies’ innerstate in the centralbrain.
The availability of a sophisticated toolbox and thefact thatinsectspresentrobustandstereotypicbehaviorsthat can be easily quantified renders the topic of vision in insects an informative research venue. It is possible to combinebehavioraldatawithanatomicaldatafrom Elec-tronMicroscopy (EM)andwithphysiological datafrom calciumimagingorelectrophysiologyexperiments, yield-ing understanding of the process of vision at different levels.Eachof thesetechniqueshasitsownadvantages anddrawbacks:EMdoesnotgiveinformationaboutthe functionality of synapses, but it can provide a general understandingof thecircuit,calcium imagingis techni-callyeasiertoperformthanelectrophysiology,butitlacks temporal resolution and the possibility of measuring inhibition. Finally,genome-editing techniques, such as CRISPR–Cas9[61],havewidenedthepossibilityto per-form genetical manipulations in species that were not originallyestablishedas geneticmodels.
Inconclusion,acombinationofsophisticatedbehavioral assays[62] andfunctional recording techniquessuchas electrophysiologyorcalciumimagingeveninwalkingor tetheredflyingflies [63],hashelped to establishcausal linksbetweenvisualinputand behavioraloutput anda clearcorrelationbetweenanatomicalstructuresand func-tionalunits.Bridgingallthatinformationtogether, mech-anisticinsightfromtheprocessingofvisualinformationat differentlevelshasbeen obtained:fromgenes,to mole-cules, to individual neurones and even to neuronal circuits.
Conflict
of
interest
statement
Nothingdeclared.Acknowledgements
WethankallmembersofProf.SimonSprecher’slabforfruitfuldiscussions
andcomments.ThisworkhasbeensupportedbytheSystemsXinitiative.
References
and
recommended
reading
Papersofparticularinterest,publishedwithintheperiodofreview,
havebeenhighlightedas:
ofspecialinterest
ofoutstandinginterest
1. IsraelachviliJN,WilsonM:Absorptioncharacteristicsof orientedphotopigmentsinmicrovilli.BiolCybern1976,21:9-15.
2. MontellC:Drosophilavisualtransduction.TrendsNeurosci 2012,35:356-363.
3. MontellC:VisualtransductioninDrosophila.AnnuRevCellDev Biol1999,15:231-268.
4. ItoK,ShinomiyaK,ItoM,ArmstrongJD,BoyanG,HartensteinV, HarzschS,HeisenbergM,HombergU,JenettAetal.:A systematicnomenclaturefortheinsectbrain.Neuron2014, 81:755-765.
5. HuangS-C,ChiouT-H,MarshallJ,ReinhardJ:Spectral sensitivitiesandcolorsignalsinapolymorphicdamselfly. PLoSOne2014,9:e87972.
6. MenzelR,BlakersM:Colourreceptorsinthebeeeye– morphologyandspectralsensitivity.JCompPhysiol1976, 108:11-133.
7. SalcedoE,HuberA,HenrichS,ChadwellLV,ChouW-H, PaulsenR,BrittSG:Blue-andgreen-absorbingvisualpigments ofDrosophila:ectopicexpressionandphysiological characterizationoftheR8photoreceptorcell-specificRh5and Rh6rhodopsins.JNeurosci1999,19:10716-10726.
8. WernetMF,PerryMW,DesplanC:Theevolutionarydiversityof insectretinalmosaics:commondesignprinciplesand emergingmolecularlogic.TrendsGenet2015,31:316-328.
9. SchnaitmannC,GarbersC,WachtlerT,TanimotoH:Color discriminationwithbroadbandphotoreceptors.CurrBiol2013, 23:2375-2382.
10. WardillTJ,ListO,LiX,DongreS,McCullochM,TingC-Y, O’KaneCJ,TangS,LeeC-H,HardieRCetal.:Multiplespectral inputsimprovemotiondiscriminationintheDrosophilavisual system.Science2012,336:925-931.
11. GaoS,TakemuraS,TingC,HuangS,LuZ,LuanH,RisterJ, ThumAS,YangM,HongSetal.:Theneuralsubstrateofspectral preferenceinDrosophila.Neuron2008,2:328-342.
12. TakemuraS-Y,LuZ,MeinertzhagenIA:Synapticcircuitsofthe Drosophilaopticlobe:theinputterminalstothemedulla.J CompNeurol2008,509:493-513.
13.
SchnaitmannGriesbeckO,C,ReiffHaikalaDF:ColorV,AbrahamprocessingE,OberhauserintheearlyV,Thestrupvisual T, systemofDrosophila.Cell2018,172:318-330.e18.
Thefirsteverevidenceofchromaticprocessingthroughlateralinhibition
attheleveloftheommatidiahasbeenreportedusingcalciumimagingby
recordingtheresponseof‘pale’and‘yellow’ommatidia.
14. KaruppuduraiT,LinT-Z,TingC-Y,PursleyR,MelnatturKV, DiaoF,WhiteBH,MacphersonLJ,GallioM,PohidaT:A hard-wiredglutamatergiccircuitpoolsandrelaysUVsignalsto mediatespectralpreferenceinDrosophila.Neuron2014, 81:603-615.
15. MelnatturKV,PursleyR,LinT-Y,TingC-Y,SmithPD,PohidaT, LeeC-H:Multipleredundantmedullaprojectionneurons mediatecolorvisioninDrosophila.JNeurogenet2014, 28:374-388.
16. MacphersonLJ,ZaharievaEE,KearneyPJ,AlpertMH,LinT-Y, TuranZ,LeeC-H,GallioM:Dynamiclabellingofneural connectionsinmultiplecoloursbytrans-synaptic
fluorescencecomplementation.NatCommun2015,6:10024.
17.
LeeLinT-Y,C-H:LuoMappingJ,ShinomiyachromaticK,TingpathwaysC-Y,LuintheZ,MeinertzhagenDrosophilavisualIA, system.JCompNeurol2016,524:213-227.
The combination of a modified GFP reconstruction across synaptic
partners(GRASP)andamethodtolabelbothmitochondriaanddendrites
makespossibleananatomicalreconstructionoftheneurons
postsynap-tictothechromaticTmneuronsinthelobula.Theuseoflightimagingand
electronmicroscopyandthedouble-labellingmethodprovidesaclear
anatomicalviewof clear playersinDrosophila chromatic vision.This
methodcanbefurtherusedforotherbigandcomplexneurones.
18.
WuCardM,GM,NernRubinA,WilliamsonGM:VisualWR,projectionMorimotoneuronsMM,ReiserintheMB, Drosophilalobulalinkfeaturedetectiontodistinctbehavioral programs.eLife2016,5:e21022.
Individuallobular columnarneurones(LC) have been optogenetically
activated,andspecificbehaviorsrelatedtothemhavebeentraced.This
showshowtheseneuronesaretheoutputchannelsofdifferentsensory
perceptionmodalities,suchasvision,andhowtheyaredirectly
respon-sibleforveryspecificbehaviors.
19. HassensteinB,ReichardtW:SystemtheoretischeAnalyseder Zeit-,ReihenfolgenundVorzeichenauswertungbeider BewegungsperzeptiondesRusselkjeffersChlorophanus. Naturforschg1956,11b:513-524.
20. BarlowHB,LevickWR:Themechanismofdirectionally selectiveunitsinrabbit’sretina.JPhysiol1965,178:477-504.
21.
GruntmanexcitationE,andRomainoffset,S,delayedReiserM:inhibitionSimplecomputesintegrationdirectionaloffast selectivityinDrosophila.NatNeurosci2018,21:250-257.
Ademonstrationthatdirection-selectivityarisesbothfromthespatialand
temporalintegrationofexcitationandinhibitoryinputbyusinginvivo
electrophysiologicalrecordingsofT4neurons.Becauseofthetemporal
resolutionofelectrophysiologicalrecordings,theauthorschallengethe
currentworkingmodelsandprovideabetterinsightintotheT4
physiol-ogyanddirection-selectivity.
22. BorstA:Abiophysicalmechanismforpreferreddirection enhancementinflymotionvision.PLoSComputBiol2018,14:1-15.
23. deAndres-BragadoL,MazzaC,SennW,SprecherSG:Statistical modellingofnavigationaldecisionsbasedonintensityversus directionalityinDrosophilalarvalphototaxis.SciRep2018, 8:11272.
24. MaussAS,VlasitsA,BorstA,FellerM:Visualcircuitsfor directionselectivity.AnnuRevNeurosci2017,40:211-230.
25. BorstA:Insearchoftheholygrailofflymotionvision.EurJ Neurosci2014,40:3285-3293.
26. SiliesM,GohlDM,ClandininTR:Motion-detectingcircuitsin flies:comingintoview.AnnuRevNeurosci2014,37:307-327.
27. BorstA,HaagJ,ReiffDF:Flymotionvision.AnnuRevNeurosci 2010,33:49-70.
28. MaisakMS,HaagJ,AmmerG,SerbeE,MeieM,LeonhardtA, SchillingT,BahlA,RubinGM,NernAetal.:Adirectionaltuning mapofDrosophilaelementarymotiondetectors.Nature2013, 500:212-216.
29. SchnellB,RaghuSV,NernA,BorstA:Columnarcellsnecessary formotionresponsesofwide-fieldvisualinterneuronsin Drosophila.JCompPhysiolA2012,198:389-395.
30. TakemuraS-Y,BhariokeA,LuZ,NernA,VitaladevuniS,RivlinPK, KatzWT,OlbrisDJ,PlazaSM,WinstonPetal.:Avisualmotion detectioncircuitsuggestedbyDrosophilaconnectomics. Nature2013,500:175-181.
31. BausenweinB,DittrichAPM,FischbachK-F:Theopticlobeof Drosophilamelanogaster.CellTissueRes1992,267:17-28.
32. HaagJ,ArenzA,SerbeE,GabbianiF,BorstA:Complementary mechanismscreatedirectionselectivityinthefly.eLife2016,5: e17421.
33. LeongJCS,EschJJ,PooleB,GanguliS,ClandininTR:Direction selectivityinDrosophilaemergesfrompreferred-direction enhancementandnull-directionsuppression.JNeurosci2016, 36:8078-8092.
34.
StrotherRubinGM,JA,ReiserWuS-T,MB:WongTheemergenceAM,NernA,ofRogersdirectionalEM,LeselectivityJQ, inthevisualmotionpathwayofDrosophila.Neuron2017, 94:168-182.e10.
Theauthorsshowthatmotiondirection-selectivityfirstemergesinthe
dendritesoftheT4neurons.Thisisshownbygeneticallymanipulating
inputneuronstoT4andrecordingthemusingcalciumimaging.These
resultshinttowardsT4neuronesusingcomponentsfrombothclassic
motiondetectionmodels.
35.
TakemuraMeinertzhagenS,NernIA:TheA,ChklovskiicomprehensiveDB,SchefferconnectomeLK,RubinofaGM,neural substratefor‘ON’motiondetectioninDrosophila.eLife2017, 6:e24394.
UsinganewEMtechnique,Takemuraetal.refinetheanatomicalmapof
theneuronalinputstoT4andshowthatthereisaspatialseparation
betweentheexcitatoryandinhibitoryinput.
36. PfeifferK,HombergU:Organizationandfunctionalrolesofthe centralcomplexintheinsectbrain.AnnuRevEntomol2014, 59:165-184.
37. BoyanG,LiuY,KhalsaSK,HartensteinV:Aconservedplanfor wiringupthefan-shapedbodyinthegrasshopperand Drosophila.DevGenesEvol2017,227:253-269.
38. SeelingJD,JayaramanV:Featuredetectionandorientationtuning intheDrosophilacentralcomplex.Nature2013,503:186-266.
39. SeelingJD,JayaramanV:Neuraldynamicsforlandmark orientationandangularpathintegration.Nature2015, 17551:186-191.
40. GiraldoYM,LeitchKJ,RosIG,WarrenTL,WeirPT,DickinsonMH: SunnavigationrequirescompassneuronsinDrosophila.Curr Biol2018,28:2845-2852.e4.
41. GreenJ,AdachiA,ShahKK,HirokawaJD,MaganiPS,MaimonG: Aneuralcircuitarchitectureforangularintegrationin Drosophila.Nature2017,546:101-106.
42. Turner-EvansD,WegenerS,RouaultH,FranconvilleR,WolffT, SeeligJD,DruckmannS,JayaramanV:Angularvelocity integrationinaflyheadingcircuit.eLife2017,6:e23496.
43. OmotoJJ,KelesMF,NguyenBM,BolanosC,LovickJK,FryeMA, HartensteinV:Virtualrealityforfreelymovinganimals.CurrBiol 2017,27:1098-1110.
44. elJundiB,WarrantEJ,ByrneMJ,KhaldyL,BairdE,SmolkaJ, DackeM:Neuralcodingunderlyingthecuepreferencefor celestial orientation. Proc Natl Acad Sci U S A 2015, 112: 11395-11400.
45. elJundiB,WarrantEJ,PfeifferK,DackeM:Neuroarchitectureof thedungbeetlecentralcomplex.JCompNeurol2018, 526:2612-2630.
46. NernA,PfeifferBD,RubinGM:Optimizedtoolsformulticolor stochasticlabelingrevealdiversestereotypedcell
arrangementsintheflyvisualsystem.ProcNatlAcadSciUSA 2015,112:E2967-E2976.
47. WolffT,IyerNA,RubinGM:Neuroarchitectureand neuroanatomyoftheDrosophilacentralcomplex:a GAL4-baseddissectionofprotocerebralbridgeneuronsand circuits.JCompNeurol2015,523:997-1037.
48. WolffT,RubinGM:NeuroarchitectureoftheDrosophilacentral complex:acatalogofnodulusandasymmetricalbody neuronsandarevisionoftheprotocerebralbridgecatalog.J CompNeurol2018,526:2585-2611.
49. FranconvilleR,BeronC,JayaramanV:Buildingafunctional connectomeoftheDrosophilacentralcomplex.eLife2018,7: e37017.
50. StoneT,WebbB,AddenA,WeddigNB,HonkanenA,TemplinR, WcisloW,ScimecaL,WarrantE,HeinzeS:Ananatomically
constrainedmodelforpathintegrationinthebeebrain.Curr Biol2017,27:3060-3085.
51. SuverMP,HudaA,IwasakiN,SafarikS,DickinsonMH:Anarray ofdescendingvisualinterneuronsencodingself-motionin Drosophila.JNeurosci2016,36:11768-11780.
52. vonReynCR,NernA,WilliamsonWR,BreadsP,WuM,NamikiS, CardGM:Featureintegrationdrivesprobabilisticbehaviorinthe Drosophilaescaperesponse.Neuron2017,94:1190-1204.e6.
53. AcheJM,PolskyJ,AlghailaniS,ParekhR,BreadsP,PeekMY, BockDD,vonReynCR,CardGM:Neuralbasisforloomingsize andvelocityencodingintheDrosophilagiantfiberescape pathway.CurrBiol2019,29:1073-1081.e4.
54. MaimonG,StrawAD,DickinsonMH:Activeflightincreasesthe gainofvisualmotionprocessinginDrosophila.NatNeurosci 2010,13:393-399.
55. ChiappeME,SeeligJD,ReiserMB,JayaramanV:Walking modulatesspeedsensitivityinDrosophilamotionvision.Curr Biol2010,20:1470-1475.
56. FujiwaraT,CruzTL,BohnslavJP,ChiappeME:Afaithfulinternal representationofwalkingmovementsintheDrosophilavisual system.NatNeurosci2017,20:72-81.
57. KimAJ,FenkLM,LyuC,MaimonG:Quantitativepredictions orchestrate visual signaling in Drosophila. Cell 2017, 168: 280-294.e12.
58. WeirPT,SchnellB,DickinsonMH:Centralcomplexneurons exhibitbehaviorallygatedresponsestovisualmotionin Drosophila.JNeurophysiol2014,111:62-71.
59. WeirPT,DickinsonMH:Functionaldivisionsforvisual processinginthecentralbrainofflyingDrosophila.ProcNatl AcadSciUSA2015,112:E5523-E5532.
60. ZacariasR,NamikiS,CardGM,VasconcelosML,MoitaMA: Speeddependentdescendingcontroloffreezingbehaviorin Drosophilamelanogaster.NatCommun2018:9.
61. JinekM,EastA,ChengA,LinS,MaE,DoudnaJ: RNA-programmedgenomeeditinginhumancells.eLife2013,2:1-9.
62. StowersJR,HofbauerM,BastienR,GriessnerJ,HigginsP, FarooquiS,FischerRM,NowikovskyK,HaubensakW,CouzinID etal.:Virtualrealityforfreelymovinganimals.NatMethods 2017,14:995-1002.
63. SeelingJD,ChiappeME,LottGK,DuttaA,OsborneJE,ReiserMB, JayaramanV:Two-photoncalciumimagingfromhead-fixed Drosophiladuringoptomotorwalkingbehavior.NatMethods 2010,7:530-540.
