Rods in Daylight Act As Relay Cells for Cone-Driven Horizontal Cell&Ndash

Home , Rod cell

© 2014 Nature America, Inc. All rights reserved. one another one with gap another junctions described been also have pathways Alternative cells. bipolar pathways: mary from rods to rod bipolar cells and from cones to cone pri by two retina inner the enters photoreceptors from Information levels. light high and cones in and levels light low at rods in zation organi field receptive center-surround antagonistic in results cells spread signal lateral mediate to shown been have cells horizontal between synapses electrical axo-axonic and dendro-dendritic Furthermore, tive feedback synapses with its corresponding photoreceptor class and that each of these horizontal cell compartments forms nega local Correspondence should Correspondence be addressed to B.R. ( Université Pierre et Marie Curie, Université Denis Diderot, Paris, France. 11 Paris, ViewMaintain, des France. Département Quinze-Vingts, Hospitalo-Universitaire de la Vision, Paris, France. USA. Munich, University,Ludwig-Maximilians Munich, Germany. British Columbia, Canada. 1 dendrites cell horizontal contact cones while terminals axon cell cell of horizontal long a type with axon-bearing the only one is there mouse In without. one cells, and axon horizontal of types two are there mammals photoreceptors both to inhibition feedback provide which cells, horizontal the retina, outer the of type with cell neuronal interact third the photoreceptors two These daytime. of characteristic intensities, light higher at operate cones while night-time, of teristic tions and cones, which respond to light directly with graded hyperpolariza rods photoreceptors, of types two on based is vision Image-forming surround inhibition. cells. Our results show that the retinal circuitry repurposes rods, when they are not directly sensing light, to relay cone-driven reversibly inactivated. Rod is depolarization conveyed to the inner retina via postsynaptic circuit elements, namely the rod bipolar that of cones. Rod responses at high light levels are abolished in mice with cones nonfunctional and when horizontal cells are activate their surrounds, rods depolarize. Rod increases depolarization with stimulus size, and its action spectrum matches rods act as relay cells for cone-driven horizontal surround cell–mediated inhibition. In response to large, bright stimuli that fields of both exhibit photoreceptors antagonistic center-surround Here organization. we show that at bright light levels, mouse Rods hyperpolarizations. operate in the lower range of light intensities while cones operate at brighter intensities. The receptive Vertebrate vision relies on two types of rods photoreceptors, and cones, which signal increments in light intensity with graded & Botond Roska Martin Biel Tamas Szikra horizontal cell–mediated surround inhibition Rods in daylight act as relay cells for cone-driven nature nature Received 1 September; accepted 29 September; published online 26 October 2014; Neural Neural Circuit Friedrich Laboratories, Miescher Institute for Biomedical Research, Basel, Switzerland. Department Department of Physics, École Normale Supérieure, Paris, France. Rods and cones are also directly connected both to themselves and to 6 1 Université Université Pierre et Marie Universités, Institut Curie-Sorbonne de la Vision, Paris, France. . Rods respond in the lower range of light intensities, charac intensities, light of range lower the in respond Rods . 1 3 . The neuronal circuit formed by rods, cones and horizontal horizontal and cones by rods, formed circuit neuronal . The NEUR OSCI 4 , , Gautam Awatramani 1 3 , , Stuart Trenholm . It has been proposed that rods contact horizontal horizontal contact rods that proposed . It been has EN 1 C 3 8 Univ E Centre Centre national de la recherche Institut scientifique, de la Vision, Paris, France. advance online publication online advance ersity ersity of Basel, Basel, Switzerland. 14 , 1 5 . . As a result of coupling, cone-rod [email protected] 1 , 2 , , Antonia Drinnenberg 2 , , Damon A Clark 5 Department Department of Molecular, Cellular and Biology,Developmental Yale University, New Haven, Connecticut, ). 4 12 Department Department of for Pharmacy–Center Drug Research, Center for Integrated Protein Science

2 Laboratoire Laboratoire de Physique Statistique, Centre National de la Recherche Scientifique, . In most most In . 2 , 13 1 4 Present Present address: Research Neuro-Electronics Flanders Imec, Leuven, Belgium. . 6– 5 1 , , José-Alain Sahel 4 2 , 5 - - - - - .

1 tion to downstream visual circuits. Like rods, cones may also display display may also cones rods, Like circuits. visual to downstream tion inhibi surround cell–mediated horizontal convey rods cone-driven, before proposed been has as retina, inner the pathway provides cones with a ‘centersecond conserving pathway’ to copy cones and In sign- this different. rods functionally the case, first times circadian at different as such conditions, or different under scales or temporal spatial may at act they different cones and for between rods are crosstalk not mutually exclusive since notion that has recently been challenged posed that the long axons of horizontal cells do not conduct signals However,synapses. via sign-inverting it has depolarization probeen rod drives then and spreads terminals to axon dendrites cell signal horizontal from this and dendrites, cell horizontal drive cones cells: horizontal via inhibition feedforward involves circuit This response. depolarizing a exhibit to rods causing polarity, response in change a arrive potentially at rods via the horizontal inhibitory cells, to leading could cone signal Alternatively, mediated light the bright responses). light hyperpolarizing is, (that rods in responses sign-conserving in resulting via gap changes in drive junctions, rod potential membrane For rod this indirect response, cones would act as photoreceptors and are at photoresponses their saturation. when even at levels, light high indirectly, light, to respond will rods mammalian that expected is it doi:10.1038/nn.385 10 , 3 Fondation Ophtalmologique Fondation Adolphe Ophtalmologique de Rothschild, Paris, France. , , Josephine Jüttner The functional consequences of these two crosstalk pathways are are pathways crosstalk two these of consequences functional The 7 Institut Institut national de la santé et de la recherche médicale, Institut 2 Department Department of Biology, University of Victoria, Victoria, 6 – 2 10 9 Centre Centre Hospitalier National d’Ophtalmologie , , Rava da Azeredo Silveira 1 , , Zoltan Raics 1 2 . . These two possible circuits 1 , , Karl Farrow 1 1 4 6 . In the second case case second the In . . t r a 11 , 12

C I

, e l 13 , 9

, , a s - - © 2014 Nature America, Inc. All rights reserved. diameter spot. Normalization was performed as in stimulus intensity. The stimulus was an 800- of rods magnitudes and cones as a of function from was replotted rod The response wholemount in wholemount. in and slice intensities light at spots different to 800- of rod and photoreceptors cone responses light Bottom, (right). wholemount to a and in rods points body) cell arrow cone in (middle; cones slice fluorescent bodies), cell to rod points arrow in (left; rods slice showing ( hyperpolarization. in as shown that such series in an experimental depolarization ( responses. depolarizing represent values positive while responses hyperpolarizing represent values Negative and sizes. spot levels light different across hyperpolarization normalized to the maximum average ( and with different spot sizes. Responses were wholemount rods at two background intensities ( of stimulation. the the light timing indicate are rows ~1 in log apart intensity.unit bars Gray Successive intensities. background different throughout was constant kept contrast stimulus The retina. a from rod in wholemount intensities and stimulus background at different measured were sizes by of spots different elicited ( retina. in wholemount levels at light spots high large 1 Figure technique. clamp patch perforated the using in mouse retina, wholemount rod bodies cell single from responses voltage recorded We Rod depolarization at high light levels RESULTS circuits. visual downstream to horizontal cell–mediated surround inhibition mouse intensities, rods act as relay conveying cells cone-driven, light of range sensing photon- their of depolarization. outside consequence, cone a As cause to background levels low at light hyperpolarization rod and depolarization rod cause to levels light background high at hyperpolarization cone ing allow directions, both in information convey axons cell horizontal mouse that suggest results These depolarization. with respond also cones spots, large with At stimulated when cells. levels light bipolar background low rod via retina inner the to spreads not Rod depolarization are inactivated. reversibly cones are cells when horizontal absent when or is functional and photoresponses cone as trum spec action same the has depolarization Rod depolarization. with respond rods spots, large with back stimulated high when at levels light that, ground show We retina. mouse wholemount in cones understood. well not are retinas mammalian these of crosstalk pathways, in different magnitude background light conditions, in intact relative the and presence The cells. horizontal via or junctions gap pathway—via the on depending sign-inverting, or sign-conserving either be would response the of polarity the and rods be would response light the of origin the Here light. to directly indirect t r a Error bars, s.e.m.; b

) Normalized response magnitudes in Here we recorded light-evoked signals from single rods and and rods single from signals light-evoked recorded we Here a

C I 1 ) Light responses to contrast increments increments to ) contrast responses Light Rods respond with depolarization to depolarization with respond Rods 5 responses at low light levels at which they do not respond respond not do they which at levels light low at responses e l a , normalized to the largest to the largest , normalized s n a d , number of recorded cells. . . ( – f) g Top, images example – i ) Normalized response ) response Normalized c ) The largest ) The largest n = 50–100)

b .

d a Spot diameter intensity: intensity: R* Stimulus Stimulus 1,090 0.26 10,700 110 R* 1.5 s 1,090 ( 13 –1 µ 0.26 Rod photoreceptor 110 m) 1.5 s 13 –1 : 25 in slice 10 50 - - - 0.05 mV µ 2 m s rods coupling between in was due likely response increase to the electrical initial The rods. of organization field receptive center-surround cal classi the displaying size, spot increasing with decreased then and increased and els. The was its initially sustained, response magnitude contrast increments was at hyperpolarizing low background light lev ( decades six intensities spanning light background different at but contrast fixed a at s, 2 maximum their to response. relative hyperpolarizing responses their using compared were rods different with experiments Therefore, amplitudes. response real represent not did likely most amplitudes response ratio, resistance were responses recorded the ( small Since illumination. infrared under bodies cell rod individual visualize could we side this from of because decrease the components. cascade We preclude approached rods from the ganglion cell side phototransduction the to of washout to due clamp time over photoresponse patch perforated used We 100 photoreceptor inslice e We presented retinas with white spots of light of different sizes for sizes of different of light spots white Wewith retinas presented 14 , 18 200 Supplementary Fig. 1a Fig. Supplementary Cone , 1 9 advance online publication online advance and the decrease can be attributed to feedback inhibition inhibition feedback to attributed be can decrease the and 10 4 mV 400 2 µ s m 800 f Rod photoreceptor in wholemount Fig. 1 Fig. 1,600 0.01 mV a 2 2 5 0.05 mV ). As expected As ). ), likely owing to small seal-to-access seal-to-access small to owing likely ),

s s µ m b h g Normalized voltage (%) i Normalized voltage (%) Normalized voltage (%) Normalized voltage (%) –100 100 150 –50 –150 –100 –150 –100 –150 –100 50 –50 100 150 –50 100 150 –50 100 150 0 50 50 50 c 0 0 0

Normalized 0 10 10 10 nature nature

voltage (%) n –2 –2 –2 1 Cone photoreceptor Rod photoreceptorin –100 = Rod photoreceptor 7 100 200 300 400 , the rod response to to response rod the , 8 10 10 10 Spot size 0 wholemount 0 0 0 in slice in slice n 10 = 10 10 800 NEUR 1,090 2 2 8 2 1.5 intensity: Stimulus ( µ 10 10 10 m) 1,200 R* OSCI 4 4 4 R* s s –1 R* R* R* –1 n n n 1,600 = = = EN s s s –1 –1 8 –1 5 6 C E - -

© 2014 Nature America, Inc. All rights reserved. Next we investigated the mechanism driving the depolarizing depolarizing the is mediated by rod via feedback horizontal cells rods because were close depolarization this driving that unlikely is It rods. mechanism of response surround the investigated we Next depolarization rod for required are photoresponses Cone noted. otherwise unless retinas wholemount in out hyperpolarizing exclusively were ( responses the level: light any at responses any depolarizing detect not did we preparations, slice inal reversed was ( contrast the of polarity the if hyperpolarization into with monotonically the contrast increased of the stimulus and turned response ( hyperpolarizing maximum the as large as at least was response izing ( depolarization large a caused strong. spots large remained with stimulation response Notably, surround the spots for extended However, spatially saturation. rod con with levels, sistent light low at that to compared smaller became spots small to response the cells horizontal from * cells recorded s.e.m.; bars, Error light. 405-nm and 520- with stimulated retina wholemount ( retina. ventral the in depolarization rod and hyperpolarization cone ( wavelengths. different at light full-field with stimulation under retina the ( a cone and ( the recordings. out to in next are red) shown (crossed connection synaptic or element circuit blocked the and electrode) the by to (pointed cell recorded the cell), horizontal and cone (rod, circuit retinal outer the of elements the showing diagrams t ( depolarization values positive hyperpolarization, s R* 0.26 the by evoked hyperpolarization maximum the In diameter. 800 of spot contrast a positive was stimulus the in and type wild in intensities stimulus and background different ( depolarization. rod 2 Figure nature nature -test). -test). In this and subsequent figures, schematic P Supplementary Fig. 1c Fig. 1a Fig. Fig. 1d a = 0.037, = 0.037, ,

– b NEUR Cone light responses are necessary for for necessary are responses light Cone −1 i and ). All the measurements described henceforth were carried were carried henceforth described ). All the measurements Fig. Fig. 1 stimulus. Negative values indicate indicate values Negative stimulus. Horizontal cell e t c 1,090 R* 1,090 R* (6) = 2.204, Student’s unpaired unpaired Student’s = 2.204, (6) ) recorded in the ventral part of part ventral the in ) recorded , responses were normalized to normalized were , responses Control NBQX OSCI Supplementary Supplementary Fig. 1b P d c < 0.05. , e ). ). Rod depolarization at high background light levels ) Light responses of a rod ) ( responses Light a s s Cnga3 1 mV EN –1 –1 – 2 g c s ) Response of a rod in a rod of ) Response ) Rod responses at two two at responses ) Rod 6 C . At high light levels levels light high At . E , f d

−/− ) Action spectrum of spectrum ) Action ). When rod responses were recorded in ret advance online publication online advance mouse line. The The line. mouse b n , number of , number Rod photoreceptor ); ); the magnitude of the depolar 1,090 R* 1,090 R* µ Contro NBQX d m

) -

l s s –1 –1 Wavelength (nm): d a Horizontal cell 0.05 mV Rod 2 s Cone Rod photoreceptor Stimulus intensity: d c

430 405 370 Normalized voltage (%) Normalized voltage (%) 550 520 490 455 1,090 R* 0.26 R* –100 –120 –100 100 200 –80 –60 –40 –20 Control 0 0 s s 0.03 mV 2 Control Control –1 - - –1 s

( spots large with stimulated when even levels, light background high these mouse lines, rods did not respond to light with depolarization at mice, whose retinas lack cone functional photoresponses from rods in recorded both we response, depolarizing rod the of prerequisite a is photoresponse ( slice in recorded response cone of onset the to corresponded retinas wholemount in transient light level at which the hyperpolarizing response of rods became more to saturation as shown by their small center response. Importantly, the of rods, peaks of atrods, peaks 500 nm (ref. (ref. nm 355 at absorption peak with photopigment ultraviolet express dominantly In of cones pre part retina, the the ventral mouse hyperpolarization. trum of rod should matchdepolarization of the action spectrum cone depolarization. rod for necessary are photoresponses 2 Rod photoreceptor 0.1 mV Fig. 2a Fig. n n s Horizontal cell = ** If rod depolarization is caused by cone activity, the action spec action the activity, cone by caused is depolarization rod If = b * 6 5 NBQ NBQX e – W X c Rod photoreceptor Stimulus intensity: and and 1,090 R* 0.26 R* Washout Cnga3 ashout 2 Supplementary Fig. 2a Fig. Supplementary 3 ). The absorption of rhodopsin, the photopigment photopigment the rhodopsin, of absorption The ). –/ s 2 s 1 mV – –1 –1 s 0.1 mV 2 s ** s.e.m.; U and ( depolarization values positive hyperpolarization, indicate values Negative respectively. depolarization, maximum and hyperpolarization maximum the to normalized s R* 1,090 at 800 of spot contrast a positive ( rod and ( depolarization. rod abolishes cells horizontal 3 Figure clamp in the ventral part of at part clamp the retina in high the ventral rods using infrared unlabeled from and clamp patch perforated cally labeled cones using two-photon targeted a = 0, Mann-Whitney Mann-Whitney = 0, f – g P d < 0.01. Fig. 1d Fig. Cnga3 ) The effect of NBQX on horizontal cell ( cell horizontal on NBQX of effect ) The P

Normalized and c intensity: Stimulus 2 = 0.005 for for = 0.005 Normalized voltage (%)

3 corrected response (%) –150 n 100 –50 250 150 350 450 550 650 ). ). We therefore recorded from geneti

, number of recorded cells. * cells. recorded of , number b 20 40 60 80 50 Blocking photoreceptor input to input photoreceptor Blocking 0 , Wavelength (nm): d −/− – ) responses. The stimulus was was stimulus The ) responses. i Rod photoreceptor −1 ). To examine whether the cone cone the To). whether examine (ref. 0.26 400 . In . In , Control n b Wavelength (nm) c = ). This indicates that cone cone that indicates This ). 2 and and c

5 0 targeted perforated patch targeted perforated and and 1,090 U ) ) and 450 -test). Error bars, bars, Error -test). d 520 , respectively, , respectively, d , responses were were , responses t r a Gnat2 Rod Cone Rod photoreceptor 0.26 500 P * µ = 0.012 = 0.012 n Cnga3 m diameter m diameter = n n = = 2 cpfl3 4 2 405 550 3 3 . . In both of C I 1,090 –/– P < 0.05, < 0.05, (ref.

2 0.5 mV

e l

R* s s a 2 –1 , 1 c s ) - - - )

© 2014 Nature America, Inc. All rights reserved. cells will abolish rod depolarization. To test this, we developed a developed we this, test To depolarization. rod horizontal abolish silencing will inhibi that cells expect feedforward we rods, of and conduits cones as between tion acting are cells horizontal If depolarization rod mediate cells Horizontal that depolarization. rod drive photosensors the are cones that notion the to support further ( them depolarized light blue whereas rods, hyperpolarized light green rods: in polarity opposite directly, with illumination green and blue light caused responses with light to respond rods and cones both when levels, light mesopic At that wavelength shortest we were able to use for ( stimulation the nm, 370 at peaked both and closely, matched depolarization rod ( ity polar opposite having therefore depolarization, with rods and tion hyperpolariza with cones responded Atwavelengths, stimulation all with wavelengths ranging from 370 to 550 nm. the retina with full-field monochromatic light stimulated we while levels light background ** cells. recorded U t (for depolarization values positive hyperpolarization, represent values Negative flash. switch the before flash test the to response the of magnitude the to normalized were Responses flash. switch the after 5 min and flash switch the s after 10 flash, switch the before flash ( traces). (bottom flash switch after 5 min least at flash test and traces) (middle flash switch s after 10 flash test traces), (top flash test to ( cells horizontal expressing ( flash. switch the s after 10 blocked is arrow) (black flash test the to response The solution. control in arrow) black flash, (switch ms 50 for light blue intensity high with illuminated cells horizontal bi-ChR2-expressing from recorded ( cells. horizontal in bi-ChR2 activate not but depolarization rod evoke to chosen was bar) (gray flash test from replotted retinas, control in responses rod shows trace Black bi-ChR2. by mediated are responses light depolarizing These NBQX. of presence the in level) light each at contrast positive (same levels 800- an to responses cell horizontal bi-ChR2-expressing ( laser. photon two- by visualized is retina wholemount in cells horizontal AAV-infected from recording ( cells. horizontal marks which antibody, a calbindin AAV with using transduction ( tdTomato ( expressing AAV. cells conditional Horizontal the with injected were cells tdTomato horizontal in and Cre expressing Mice (AAV) construct. virus adeno-associated EGFP-expressing and ( photoreceptors. rod in responses light depolarizing abolishes cells 4 Figure t r a ( line mouse i e -test; for for -test; n

) A patch pipette filled with Alexa 594 594 Alexa with filled pipette ) A patch -test). Error bars, s.e.m.; s.e.m.; bars, Error -test). – , o h ) ) n ) Quantification of horizontal cell cell horizontal of ) Quantification Fig. 2d Fig. Quantification of the responses to the test test the to responses the of Quantification   : : P l , < 0.001, < 0.001,

m C I Reversible inactivation of horizontal horizontal of inactivation Reversible o ) Voltage responses of bi-ChR2- of responses ) Voltage µ b : : ) and EGFP ( EGFP ) and m light stimulus at different light light different at stimulus m light P , e e l Gja10-Cre = 0.005, = 0.005, j ). The action spectrum of cone hyperpolarization and and hyperpolarization cone of spectrum action The ). Figure 1 Figure ) Blue trace shows normalized normalized shows trace ) Blue k P s ) Sustained depolarization depolarization ) Sustained t < 0.01, *** < 0.01, (3) = 131, Student’s paired paired Student’s = 131, (3) a ) Conditional bi-ChR2- bi-ChR2- ) Conditional i . The intensity of the the of intensity . The U c ) that expressed Cre recombinase Cre expressed that ) ) are colabeled ( colabeled ) are = 0, Mann-Whitney Mann-Whitney = 0, n l , number of , number ) and rods ( rods ) and P < 0.001. Fig. 2 Fig. m g ). These experiments provide provide experiments These ). ) d ). ).

a IT l i e R EGFP EF1a loxP Calbindin AAV construct 2 EGFP

4 /

and a red red a and after switchflash lox2722 Alexa 594 Horizontal cell Fig. Test flash Test flash Test flash 2A bi-ChR 30

20 2 WPR f µ µ ). ). poly(A 2 m - - - m

2 mV 2 E s f IT zontal cells ( cells zontal in hori bi-ChR2 expressing and in rods cells retinas from horizontal ( efficient (Online GFP with cells horizontal and Methods the labeled also viruses these horizontal cells using conditional adeno-associated viruses ( viruses adeno-associated using conditional cells horizontal ( reversibly recording cells horizontal while light inactivated from responses we rods depolarization, rod drive consequently, and, rods to depolarization. rod for required is transmission ( ionotropic detected the responses were rod using depolarizing no blocked NBQX, cells, blocker receptor horizontal glutamate pharmacologically to we inputs When photoreceptor cells. horizontal from record to labeled clamp patch laser-targeted two-photon We used cells. then horizontal in specifically tdTomato, marker, fluorescent ) j Fig. Fig. Normalized voltage (%) R –150 –100 To provide causal evidence that horizontal cells convey cone signals –50 100 150 b m 50 0 Virus EGFP 4 10 ). We delivered a bistable channelrhodopsin (bi-ChR2) channelrhodopsin bistable a delivered We ). Mouse tdTomato –2 Fig. 4e Fig. advance online publication online advance 10 n Rod = Fig. 4 Fig. 0 8 Fig. 4b Fig. 30 Test flash 10 Rod photoreceptor after switchflash 2 µ – i. 3a Fig. m 20 h Test flash Test flash i Test flash 10 ). and and µ 4 g m – n Horizontal cell = d R* ). Horizontal cell labeling was selective and and selective was labeling cell Horizontal ). 5 c Supplementary Fig. 3 Fig. Supplementary s and virusEGFP – –1 d Calbindin 0.05 mV ). Therefore, glutamatergic synaptic synaptic glutamatergic Therefore, ). 2 s Virus EGFP k Switch flash n o 30

Normalized voltage (%) Test flash

Normalized voltage (%) Rod photoreceptor –100 –100 µ Horizontal cell 100 200 –40 –80 –60 –20 4 m 4 mV 20 s 0 0

h µ m Test flash Test flash nature nature (1)-neurons also

expressing (2), % d 100 ). We then recorded recorded Wethen ). 20 40 60 80 n n 0 ** = tdTomato /EGFP = switch flash switch flash Test flash Test flash 6 *** 4 NEUR 91% after after (1) Calbindin (2) GFP 96% OSCI (1) GFP Test flash Test flash Fig. (2) Calbindin retinas 20 n 4% = EN (1) GFP 2

µ 4 5 4 m

to to (2) No Calbindin a C ); ); E -

© 2014 Nature America, Inc. All rights reserved. larization significantly, a HEPES-independent rod depolarization depolarization rod HEPES-independent a significantly, larization elements circuit the of some influence transmitters inhibitory these that suggests which waveform, response rod the affect did however, antagonists, glycine ( strychnine and picrotoxin by blocked were receptors glycine and GABA when suppressed not was response rod depolarizing of the amplitude the hypothesis, this with buffer pH a HEPES, by blocked be feedback to thought this is However, glycine. or GABA neurotransmitters tory inhibi the use not does that mechanism a via photoreceptors with ways understood well not is ceptors regime. intensity light in higher the observed rod responses depolarizing inverted, the causes activity cell horizontal cone-driven that conclude we ments, experi these From flash. switch the by desensitized not was retina the that showing a control, as served also coupling. hyperpolarization rod The cone-rod through rods reaches that activity cone reflect could hyperpolarization the Alternatively, saturated. completely not were that and it to coupled electrically were that rods the of and rod recorded of the stimulation light direct the reflected likely larization ( response by replaced hyperpolarizing small was a and disappeared flash) test the (to response izing depolar the flash, of switch the presentation the However, following found in control retinas, the we first test flash evoked rod depolarization. what to Similarly protocol. same the using rods from measured ( again once flash test the to responded and state ( the switch horizontal flash, cells did not respond to a second test flash of presentation after Tenseconds NBQX. by blocked or intact were inputs cells photoreceptor to horizontal whether ing the flash, switch ( seconds of tens for cells horizontal depolarized flash switch the contrast, In cells. horizontal in response hyperpolarizing a evoked flash test The rods. in depolarization the quently, block conse and, cells horizontal on input cone the of effect the suppress would turn, in shunting, this cells; would horizontal shunt and flash switch depolarize the by bi-ChR2 of activation the that expected We reproducible. was it ensure to series flash the repeated Wethen to return to again. flash test the cells to responses the measured and state, native horizontal their allowing time, of waited period we longer Finally, a again. for flash test the to responses the measured we with bi-ChR2 the activated switch and, flash seconds, after several Next flash. test the to cells horizontal and rods of responses voltage ( bi-ChR2 activate to enough bright was flash switch the contrast, In depolarization. rod and larization hyperpo cell horizontal and cone bi-ChR2, evoke to enough bright activate was it but to enough intense not was flash test The flash’). ‘switch (the ms 50 for lasted that flash blue high-intensity brief, a or flash’) ‘test (the s 2 for lasted that flash white a retina: the to stimuli ( s.e.m.; s R* 1,090 800- the by evoked depolarization maximum the to normalized ( 5 Figure nature nature ( persisted a Fig. 4l Fig. Supplementary Fig. 4 Fig. Supplementary ) The effect of HEPES on the depolarizing rod responses. ( responses. rod depolarizing the on HEPES of effect ) The The molecular nature of the horizontal cell feedback to photore to feedback cell horizontal the of nature molecular The We first applied this protocol while recording from horizontal cells. logic following the to according constructed was experiment This 27– n ,

3 n , number of recorded cells. * cells. recorded of , number NEUR 1 The influence of HEPES on the depolarizing rod response. response. rod depolarizing the on HEPES of influence The ). After 5 min, horizontal cells had returned to their native native their to returned had cells horizontal min, 5 After ). . It has been suggested that horizontal cells communicate communicate cells horizontal that suggested been has It . Fig. 4k Fig. −1 Fig. 5a Fig. intensity ( intensity OSCI , l , b ): we observed a sustained depolarization follow depolarization sustained a observed we ): EN ). 3 2 . While HEPES decreased the mean rod depo rod mean the decreased HEPES While . C P = 0.031, Wilcoxon signed-rank test). Error bars, bars, Error test). signed-rank Wilcoxon = 0.031, E ). A projector provided two kinds of light light of kinds two provided projector A ). advance online publication online advance 13 Supplementary Fig. 5 Fig. Supplementary , 2 6 P and may combine several path several combine may and Fig. 4m Fig. < 0.05. Fig. 4 Fig. j , ). First we recorded the the recorded we First ). o ). This small hyperpo small This ). Fig. 4l Fig. 13 , 2 7 . In agreement agreement In . ). GABA and and GABA ). b , n ) Responses ) Responses µ ). We ). then

m spot at m spot

------of the response depend on the stimulus history. Specifically, in each each in Specifically, history. stimulus the on depend response the of represents an adaptive transfer in which the magnitude and the shape tions with similar forms (Online Methods). Each of these three blocks the in ‘blocks’ three model, for the rod, cone and model: horizontal cell, are circuit described by simple our equa for block building a as cell–rod circuit ( cone–horizontal the of model phenomenological a Weconstructed depolarization rod of model Phenomenological et e netgtd hte te eoaiig o rsos at response rod retina. inner the of function depolarizing the affects levels light the background high whether investigated we Next retina inner the to propagates depolarization Rod levels. light at low extent, to though ata cones light levels, depolarize, smaller background high ( preparation sive in slice at light those hyperpolarization levels at which cones respon became ( slice in cones in response ( slice in and wholemount in rods in hyperpolarization light evoked low that levels those at depolarization slow and small a with responded retina with large, positive-contrast spots. Cones in wholemount retina the stimulating while decades, six spanning intensities, light ground directly? to light respond not do cones when levels at light low cones to directly. respond light conversely,Do rods, drive do not rods when levels, at light high depolarization rod drive Cones levels light low at depolarization Cone ( intensities light of background decades several across responses of voltage dynamics spite of its and simplicity, the the captures magnitude both the model through the horizontal cell block, for which parameters were fitted. In blocks two the coupled we parameters, cone and rod these refitting capture the behavior of rods and cones in slice ( ( procedure fitting the simplifies which responses, the of values steady-state of the by fixed some are parameters the equations; cone and rod the of parameters fitted first We same the that model can be also used to rods describe in but slice and in wholemount retina. parameters of number small the and lation block rod Methods). the (Online to appropriately coupled is latter the and block, cell horizontal the to input the as serves output cone the Finally, block). input (or two other filtered versions of the input, in the case of the rod of the version filtered way by another in a nonlinear modulated then is quantity input the block and is a with resulting the convolved filter Stimulus intensity: a To we from cones at recorded question, back answer this different formu of the simplicity the only not is approach of this merit The Figs. 1d Figs. – i i Fig. Fig. 6 Fig. 6 Fig. and 1,090 R* Washout HEPES Control 7a a c ). ). We used an earlier model of cone response ). – Fig. 1e Fig. s c –1 ). The same stimulus evoked little or no no or little evoked stimulus same The ). Fig. 1e Fig. 0.08 mV 2 s , h ). Therefore, while rods depolarize depolarize rods while ). Therefore, , h b

). The depolarization turned to to turned depolarization The ). Normalized voltage (%) 120 100 20 40 60 80 Fig. 6 Fig. 0 Control b ). The resulting blocks blocks ). The resulting Fig. Fig. 6 n t r a = *

depolarization in depolarization HEPES 6 c ). ). Next, without C I Washout e l s 3 3 - - - -

© 2014 Nature America, Inc. All rights reserved. that isolated signals induced by rods via rod bipolar cells and and cells bipolar rod via rods by induced signals isolated that the retina, the ganglion cells, we an developed approach experimental retina. inner the to conveyed is levels light high sur at depolarization rod to leading cell–mediated to inhibition round horizontal cells cone-driven, the that horizontal conclude We from inhibition cells. of bipolar rod response the blocked direct APB since cells bipolar rod from levels result light high at cannot response hyperpolarizing cell bipolar rod The ( cells horizontal and photoreceptors between munication dihydrobenzo[ rods and rod or between bipolar cells, NBQX (6-nitro-2,3-dioxo-1,4- ( APB either acid), a amino-4-phosphonobutyric drug that blocks by communication blocked were responses hyperpolarizing ( responses hyperpolarizing displayed cells bipolar rod the spots, large to responses depolarizing exhibited rods when However, at changed responses. levels: izing light behavior high their depolar usual the with of light spots to large responded cells bipolar labeled GFP were cells these which in line mouse a in intensities, background different at cells bipolar we from of rod recorded cells, rod the response bipolar can influence rods of depolarization the whether To elucidate hyperpolarize. rods when depolarization cell bipolar rod yield which synapses, inverting cells is bipolar retina rod via inner the to transmitted is activity rod which in way One from re-plotted recordings trace; black fit, Model column). (right wholemount and columns) middle and (left slice in intensities light of decades five response, (steady-state magnitude response the captures model quantitative This here. schematized equations the for Methods Online See properties. processing its affecting thereby line), (green block rod the to coupled is block cell horizontal the of output The curves). (red filter exponential an with a convolution arrows), (red processing of stage next the dynamically modulate filters secondary The fit. the in used timescales different the illustrate filters these of widths relative the curves); (blue filter(s) secondary rod) (for two or cell) horizontal and cone (for one and curves) (black filter a primary with convolutions of consists block Each arrow). (green output rod to arrows) (yellow input light from transformation the ( 6 Figure t r a a

c a ) Block model of the cone–horizontal cell–rod circuit, which describes describes which circuit, cell–rod cone–horizontal the of model ) Block Normalized voltage (%) in cells output the affect can depolarization rod whether test To –150 –100 Figure 1d Figure –50 100 150 50 0

C I Computational modeling of the cone–horizontal cell–rod circuit. circuit. cell–rod cone–horizontal the of modeling Computational 10 b ) and the dynamics ( dynamics the ) and –2 e l – 10 f f , purple. , purple. qioaie7sloaie, hc bok com blocks which ]quinoxaline-7-sulfonamide), 0 s 10 2 . Rod bipolar cells receive rod signals via sign- via signals rod receive cells bipolar Rod . 2 10 10 4 µ m n c = ) of rod and cone light responses across across responses light cone and rod ) of R* 3 s –1 3 b 4 ( Fig. 8 Fig. intensity: Stimulus 10,700 1,090 0.26 R* 110 a 1.5 13 ). At low light levels, levels, light low At ). s –1 Fig. 8 Fig. b Fig. 8 Fig. l ). These These ). 0.1 mV -(+)-2- 2 s c ). ). - - -

s.e.m.; s.e.m.; bars, Error intensity. stimulus of a function as cones of magnitudes ( levels. light high at hyperpolarization and levels light low at depolarization produced Cones intensities. light different at ( image. a two-photon on bodies cell indicates arrow The retina. mouse ( retina. wholemount 7 Figure retinal inner of surround antagonistic the retina, light-adapted the in Therefore, cells. bipolar rod in hyperpolarization and rods in tion as depolariza is measured inhibition surround The retina. inner the to inhibition surround cell–mediated horizontal cone-driven, relay rods conditions, light bright in that is work this in finding key The DISCUSSION the to relayed is retina. the of cells output depolarization rod of influence the that suggests ( raised was intensity light background the when polarity reversed spots, large with stimulation to response in input in cells, PV-5 ganglion inhibitory the cells, As in bipolar rod activity cell bipolar cone-driven from decoupled is pathway this are which cells, themselves stimulated by rod amacrine bipolar cells; in the AII from received is cells PV-5 to input (ref. gene junction gap the lack that mice in PV-5 type, cell cells ganglion marked pathway in a tory genetically inhibi characterized previously a We used cones. by induced those b c b a ) Responses of a cone to a positive-contrast, 800- a positive-contrast, to a cone of ) Responses intensity: Stimulus

3 Response magnitude R* 6 1,090 –0.5 n 0.26 ), as an indicator of rod bipolar-driven activity. Inhibitory Inhibitory activity. bipolar-driven rod of indicator an as ),

0.5 1.0 1.5 110 , number of recorded cells. recorded of , number 1.5 s Cone responses at different background intensities in intensities background different at responses Cone 13 0 –1 advance online publication online advance 10 Rod photoreceptor –2 10 in slice 0 10 2 a 10 ) Cones were targeted in the the in targeted were ) Cones 4 Cone block Rod block –0.5 0.5 1.0 1.5 0 10 Cone photoreceptor –2 10 in slice 0 10 2 10 Fig. 8d Fig. 4 Gjd2 nature nature c –0.5 ) Normalized response response ) Normalized 0.5 1.0 1.5 Chrnb4- , also known as as known also , Horizontal cellblock µ , 0 Cx36 e m-diameter spot spot m-diameter 10 ). This experiment experiment This ). Rod photoreceptor –2 in wholemount 10 NEUR −/− GFP GFP 0 10 background 2 OSCI 10

R* Cx36 EN 2 s s –1 C 3 3 5 5 E - - . ,

© 2014 Nature America, Inc. All rights reserved. the rod the pathway is broader, as and are rods by coupled gap junctions, via surround the properties: functional have different likely pathway cell pathway. bipolar The cone surround relayed by the rod pathway cone-cone and the cone the via arrives that pathway portion the rod with the merges via arrives that surround cone the of portion the cells with bipolar cone OFF synapses chemical inhibitory sign-inverting and cells cone bipolar ON with synapses electric sign-conserving make which cells, amacrine AII the via cells, bipolar OFF cone and ON cone of minals ter the at system cone the to returns likely and cells bipolar rod via rods. many lift can cone by one controlled seesaw the cells, of Because the coupling extensive electric of both rods and horizontal sign of and, cone because the light response), a (inducing hyperpolarizing down end cone the pushes increment a light rods, saturates that tion ( ‘seesaw’ like a acts circuit This other. each to coupled also are rods addition, In cells. horizontal neighboring of compartments same the to pled cou electrically are cells horizontal of dendrites the axon and the terminals both because component spatial strong a has circuit The pathway. inhibitory feedforward a is This synapse. sign-inverting a induces horizontal rod through cell hyperpolarization depolarization where axons, long their via cells horizontal of terminals axon the to that contact cones dendritic hyperpolarization hyperpolarize; spreads cells of horizontal dendrites the in to turn, light; in response polarize in cones hyper the rod pathway: inhibition surround to cone-driven leads of events sequence following the that suggest experiments Our circuit Seesaw work. present the in possibility this investigated we have not though retina, inner to the spread also potentially could pathway cone the in surround rod This light. to directly responding are rods only when intensities, low at stimulation to extended spatially depolarization exhibit they because levels light low at surround by the rod pathway. Our cone that rod suggest recordings cones carry a second one, described here, that is mediated one that is mediated by the cone components: pathway and two have to predicted is cells recorded cells. * paired responses in two different light levels. ( an 800- of presentation the during recorded cell PV5 ( application. drug before responses hyperpolarizing of magnitude U NBQX, and ABP both s R* 1,090 at responses hyperpolarizing the on NBQX and APB of ( levels. light different two at spot 800- an to cell ( 594. Alexa with filled electrode patch a by targeted cell bipolar rod a indicates arrow The retina. wholemount in (arrowhead) cells bipolar rod GFP-labeled target to used was line mouse ( retina. inner the to and cells bipolar rod to carried are 8 Figure nature nature sustained. more likely -test). Values were normalized to the the to normalized were Values -test). Rod depolarization carrying cone surround enters the inner retina enters retina cone the inner surround carrying depolarization Rod - inverting horizontal cell, the rod-end swings up (depolarizes). (depolarizes). up swings rod-end the cell, horizontal inverting b t -test). Error bars, s.e.m.; ) Voltage responses in a rod bipolar bipolar rod a in responses Voltage ) µ

m-diameter, positive-contrast spot at NEUR Depolarizing rod responses responses rod Depolarizing d Supplementary Fig. 6 Fig. Supplementary a ( ) The The ) µ P OSCI P m-diameter, positive-contrast positive-contrast m-diameter, = 0.04, < 0.05, ** d ) Inhibitory currents to a to currents Inhibitory ) Pcp2-GFP EN U = 0, Mann-Whitney Mann-Whitney 0, = C −1 2 t e E . At the axon terminals of cone bipolar cells, cells, bipolar cone of terminals axon Atthe . (2) = 3.514, Student’s ) Quantification of

( P P advance online publication online advance = 0.005, for for 0.005, = < 0.01. transgenic transgenic n , number of c ) The effect effect The )

): under steady bright illumina bright steady under ):

a c Normalized voltage (%) –120 –100 –80 –60 –40 –20 20 40 60 0

Control ** n = APB - - - - 6 Rod bipolarcell tions, depending on the light conditions. First, in night light, rods rods func light, night in distinct First, two conditions. light least the on at depending tions, fulfill rods that indicate findings Our functions distinct two have Rods to light directly responding rods from arises to likely that responses a large stimuli of part the fast hyperpolarizing lacking cone responses ( R* s (1,090 levels light same the at However, coupling. cone-rod by explained partially be could This fast the larger hyperpolarization depolarization. sustained occasionally recorded a small hyperpolarizing transient in rods before retinal slices hyperpolarization when rods are acting as photoreceptors as shown in cone increase would coupling Similarly, cone-rod photoreceptors. as recorded in light hyperpolarization regimes at which cones are acting to rods is weak: for these stimuli, cone-rod coupling would increase rod small stimuli the suggested feedforward inhibitory pathway from cones the recorded voltage responses in our study in various ways. First, with retina inner the to activity rod transmitting for mice in coupling cone-rod emphasized the importance of the rod-rod bipolar pathway relative to experi our ments, in which were pathway performed during this daytime. A of recent report weakness has also relative the explain daytime could in weaker and night at stronger is it circa rhythm; the by dian modulated is coupling cone-rod has It that rods. suggested on been cells horizontal of action direct the with competes retina. the of part same the in tion stimula to full-field response in rods in depolarization and cones in recordings argue against this model, as we recorded hyperpolarization spread to coupling, causing rods Our via to electrical depolarize. may depolarization this and cones, in depolarization evoke may lus bright stimu is cells from strong, a horizontal tive feedback full-field rods and cones between coupling cells electrical with together horizontal and cones between feedback negative involving model, circuit by an alternative explained be depolarization rod Can coupling cone-rod of Contribution b Washout Stimulus intensity: Cone-rod coupling counteracts rod depolarization and therefore therefore and depolarization rod counteracts coupling Cone-rod 1,090 R* 0.26 R* ** 3 7 . Nevertheless, cone-rod coupling could have contributed to to contributed have could coupling cone-rod Nevertheless, . NBQX s s 1 –1 –1 5 Washout . Second, using large stimuli, even at high light levels, we −1 Gnat2 ) we also recorded hyperpolarization in mice hyperpolarization ) recorded we also 1 mV 2 e s cpfl3 Stimulus intensity: and d Stimulus intensity Ganglion cell

Current (pA) 1,090 R* Cnga3 –40 –60 –20 1.5 R* 20 40 60 80 0 3 −/− 8 s s –1 –1 . n 1.5 = mice). Therefore, it is Ganglion cell 3 t r a * 1,090 1 C I 4 ? If ? nega 60 pA 2 1 s 6 R* e l . This This . s –1 1 s 3 ------, © 2014 Nature America, Inc. All rights reserved. 7. 6. 5. 4. 3. at online available is information permissions and Reprints The authors declare no competing interests.financial model and wrote the paper. experiments and wrote the paper. B.R. experiments, designed implemented the opponent rod pathway. developed the R.A.d.S. quantitative model, designed performed simulations. J.-A.S. suggested the experiments with the spectrally experiments. D.A.C. developed and implemented the quantitative model and recorded from ganglion cells. M.B. provided the the stimulation and recording software. K.F. built the two-photon microscope and J.J. anddesigned made the AAV vectors and did AAV injections. developed Z.R. cells. A.D. did AAV injections and quantified the horizontal cell AAV infections. microscope and wrote the paper. S.T. recorded from horizontal and rod bipolar and rod bipolar cells, did injections, experiments, designed built the two-photon T.S. performed all experiments with rods and cones, recorded from horizontal UMR 8550 to R.A.S. and a Sinergia grant and Centre National de la Recherche through Scientifique TREATRUSH,SEEBETTER, OPTONEURO and 3X3D Imaging grants to B.R.; in Research Molecular Systems Sinergia Engineering, and European Union European Research Council, Swiss-Hungarian National Centres of Competence fellowship to A.D.; Foundation,Gebert-Ruf Swiss National Foundation,Science Human Frontier ProgramScience fellowship to S.T.; IngelheimBoehringer Fonds for commenting on the manuscript. We acknowledge the following grants: a AAV production and P. King, S. andOakeley members of the Roska laboratory We thank C. Cepko, L. Vandenberghe and S. Rompani for help with high-yield online version of the pape Note: Any Supplementary Information and Source Data files are available in the the pa the of in version available are references associated any and Methods M understood. be to remains information opponent spectrally this use centers visual higher and retina the How mation. infor opponent spectrally carry rods rods, of spectrum absorption the from different substantially is rods to inhibition cell horizontal drive that cones the of spectrum absorption the where and sensors different in purposes conditions light different for used are neurons retinal same the efficient: therefore is retina The inhibition. surround cell zontal hori for cells relay as act rods intensity, light in changes to respond directly cones only which at levels light high at Second, brain. the to information visual monochromatic convey and photons capture t r a 2. 1. reprints/index.htm COM AU Acknowledgment

ethods Notably, at light levels at which rods and cones both act as photo as act both cones and rods which at levels Notably,light at T Baylor, D.A., Fuortes, M.G. & O’Bryan, P.M. Receptive fields of cones in the retina the in cones of P.M.fields O’Bryan, Receptive & M.G. Fuortes, D.A., Baylor, currents calcium regulates feedback cell Horizontal W.B. Thoreson, & N. Babai, electron retina: primate the of layer plexiform outer the of Organization H. Kolb, retina the in photoreceptors and cells horizontal between connections The H. Kolb, rodent in cell horizontal of types Morphological J. González-Soriano, & L. Peichl, äse H Prle poesn i te amla retina. mammalian the in processing Parallel H. Wässle, variations. and motifs Phototransduction R.C. Hardie, & K.-W. Yau, of the turtle. the of mouse and salamander of retina. photoreceptors rod in levels calcium intracellular and (1970). cells. Golgi-impregnated of microscopy (1974). preparations. Golgi of microscopy electron cat: the of pig. guinea and (1994). gerbil, 501–517 mouse, rat, of comparison a retinae: (2004). 747–757 (2009). 246–264 H P O ET R R C I I J. Physiol. J. CONT NG e l FI J. Physiol. (Lond.) Physiol. J. N l RIBU . 35 A s pe ( , NC 39 Lond. r . , T r 4 IA . s 0 I . ON ) L 587 I NTE S , 2353–2364 (2009). 2353–2364 ,

214 R E S , 265–294 (1971). 265–294 , T Phil. Trans.Phil. B Lond. Soc. R. S Cnga3 −/− J. Comp. Neurol. Comp. J. mouse. G.A. designed http://www.nature.c a. e. Neurosci. Rev. Nat. i. Neurosci. Vis.

258

155 , 261–283 , Cell onli , 1–14 ,

139

om/ 11

ne ne 5 - - - , , ,

20. 19. 18. 17. 16. 15. 14. 13. 12. 11. 10. 9. 8. 22. 21. 40. 39. 38. 37. 36. 35. 34. 33. 32. 31. 30. 29. 28. 27. 26. 25. 24. 23.

i PH, ewi, . Ln, .. Shaf JL Gpjntoa culn of M. Biel, coupling Gap-junctional J.L. Schnapf, & J.H. Long, J., Verweij, P.H., Li, vision night for P.Microcircuits Sterling, & M. Ueda, K., Tsukamoto,Y.,Morigiwa, Yau, K.W. Phototransduction mechanism in retinal rods and cones. The Friedenwald controls retina the in clock circadian The S.C. Mangel, & Y. Cao, C., Ribelayga, junctions gap through signal rods Mouse L. Cangiano, & C. Gargini, S., Asteriti, Bloomfield, S.A. & Völgyi, B. The diverse functional roles and regulation of neuronal retina. outer the in interactions Lateral S.C. W.B.Mangel, Thoreson, & Trümpler,J. rod to cells horizontal from T.M.Feedback Bartoletti, & N. W.B.,Babai, Thoreson, retina. rabbit the in cells horizontal to input cone and Rod Massey,F.S.C. & Pan, with retina cat P.in Gouras, cells & Horizontal H. Kolb, Litzow,A., von R., Nelson, Detwiler,C.M., Davenport, P.B. horizontal on buffering Dacey,pH & of Effects D.M. Altimus, C.M. Altimus, B. Chang, Farrow,K. mechanisms circuit and synaptic The F. Rieke, & G.W. Schwartz, W.N., Grimes, F.Naarendorp, Ke, J.-B. Deans, M.R., Volgyi, B., Goodenough, D.A., Bloomfield, S.A. & Paul, D.L. Connexin36 T.A.Münch, Dynamical R. cells Silveira, Purkinje of Purification M. Yuzaki, & da J.I. Morgan, D.S., Rice, Tomomura, Azeredo M., & M. Meister, R., Benichou, D.A., Clark, in channels Calcium S. Barnes, & N.C. Brecha, X., Sun, A.A., Hirano, X., Liu, sensor pH optogenetic an Imaging R.H. Kramer, & L.C. Holzhausen, T.-M.,Wang, Kamermans, M. mediate cleft synaptic invaginating the in changes pH A. Kaneko, & H. Hirasawa, Guo, C., Hirano, A.A., Stella, S.L. Jr., Bitzer, M. & Brecha, N.C. Guinea pig horizontal the J.P. Vessey, in communication Ephaptic M. Kamermans, & L.J. Klaassen, R., Vroman, Bi-stable K. Deisseroth, & P. Hegemann, L.A., Gunaydin, O., Yizhar, A., Berndt, S. Siegert, and UV-Jr. E.N. P. Pugh, & Valentini, M.E., Pennesi, B., Falsini, A.L., Lyubarsky, 3552–3562 (2012). 3552–3562 detection. visual on effect its and photoreceptors rod mammalian retina. mouse in Lecture. coupling. rod-cone cones. with retina. the in junctions gap Res. Eye retina. mouse the retina. vertebrate in photoreceptors Neurol. Comp. J. proton systems. dendritic independent the for evidence retina: primate in formation. responses surround of hypothesis light cell ganglion and crmtpi de o mtto i Gnat2. in (2006). 5017–5021 mutation a to due achromatopsia CNG3. channel gated and visual perception at cone threshold. cone at perception visual and retina. the in encoding spatial (2014). in change a underlying vision. cone mouse of sensitivity synapses. cell Neuron is essential for transmission of rod-mediated visual signals in the mammalian retina. circuit. neural green thatexpress mice transgenic protein. fluorescent from sorting cell fluorescence-activated by photoreceptors. in adaptation an through mechanism. (2013). 3309–3324 photoreceptors pH-sensitive of and inhibition GABA- feedback unconventional regulate cells horizontal rat retina. the (2014). in 262–268 inhibition lateral mediate protons that reveals 292 Physiol. Gen. J. feedback from horizontal cells to cone photoreceptors by modulating Ca2+ channels. transporter. cells express GABA, the GABA-synthesizing enzyme GAD 65, and the GABA vesicular synapse. photoreceptor cone (2005). the at channels retina. vertebrate switches. state neural Neurosci. Nat. possible a mouse: the of photopigments. (1999). cone of responses coexpression for retinal phenotype cone-driven midwave-sensitive intensities. light of range wide , 1178–1180 (2001). 1178–1180 ,

36 et al. et advance online publication online advance Invest. Ophthalmol. Vis. Sci. Vis. Ophthalmol. Invest. et al. 31 et al. et , 703–712 (2002). 703–712 , et al. et t al. et J. Comp. Neurol. Comp. J. Elife , 407–441 (2012). 407–441 , et al. et et al. et t al. et Selective loss of cone function in mice lacking the cyclic nucleotide- cyclic the lacking mice in function cone of loss Selective Adaptation to background light enables contrast coding at rod bipolar

et al. et Neuron Nat. Neurosci. Nat. 15 et al. et Ambient illumination toggles a neuronal circuit switch in the retina the in switch circuit neuronal a toggles illumination Ambient

et al. Transcriptional code and disease map for adult retinal cell types. cell retinal adult for map disease and Transcriptionalcode 122

oe htrcpo fnto ls-, nvl os mdl of model mouse novel a loss-3, function photoreceptor Cone J. Neurosci. J. 3 Approach sensitivity in the retina processed by a multifunctional a by processed retina the in sensitivity Approach Rod and cone contributions to horizontal cell light responses in responses light cell horizontal to contributions cone and Rod Front. Hum. Neurosci. Hum. Front. 500 , 487–495, S1–S2 (2012). S1–S2 487–495, , J. Neurosci. J. rtnmdae febc ihbto o peyatc calcium presynaptic of inhibition feedback Proton-mediated , e01386 (2014). e01386 , Neuron Rod photoreceptors drive circadian photoentrainment across a across photoentrainment circadian drive photoreceptors Rod Eur. J. Neurosci. J. Eur. Dark light, rod saturation, and the absolute and incremental and absolute the and saturation, rod light, Dark , 657–671 (2003). 657–671 , Hemichannel-mediated inhibition in the outer retina.

, 815–831 (2007). 815–831 , Proc. Natl. Acad. Sci. USA Sci. Acad. Natl. Proc. Nat. Neurosci. Nat. 81 , 388–401 (2014). 388–401 ,

59 Nat. Rev. Neurosci. Rev. Nat.

518 21 PLoS Comput. Biol. Comput. PLoS , 790–801 (2008). 790–801 , 12

28 Nat. Neurosci. Nat. Science , 8616–8623 (2001). 8616–8623 , , 1308–1316 (2009). 1308–1316 , J. Neurosci. J. , 6818–6825 (2008). 6818–6825 , , 1647–1669 (2010). 1647–1669 , J. Neurosci. J. J. Neurosci. J.

14 12 35

, 57–63 (2001). 57–63 , , 229–234 (2009). 229–234 , 189 7 , 9–32 (1994). 9–32 , Neuron , 612 (2013). 612 , , 137–139 (1975). 137–139 ,

13 28 net Ohhlo. i. Sci. Vis. Ophthalmol. Invest.

10 28

, 12495–12507 (2010). 12495–12507 , 78 , 1107–1112 (2010). 1107–1112 ,

, 456–464 (2008). 456–464 , 96 . Neurosci. J. 9 nature nature , 495–506 (2009). 495–506 , , 5691–5695 (2008). 5691–5695 , , 325–338 (2013). 325–338 , , e1003289 (2013). e1003289 , , 7553–7557 (1999). 7553–7557 , . Neurosci. J. . Physiol. J. Neuron NEUR a. Neurosci. Nat.

25 J. Neurosci. J.

4108–4117 , 82 19 ( OSCI Lond. Prog. Retin. Prog. 460–473 , , 442–455 442–455 , Science ) EN 591

32 17 47 C E , , , ,

© 2014 Nature America, Inc. All rights reserved. lowing antibodies were used: primary: rabbit anti-GFP (1:200, Invitrogen, Invitrogen, (1:200, anti-GFP rabbit primary: fol used: the were cells, antibodies horizontal lowing antifade infected Gold of number Prolong the using quantify To mounted were (Invitrogen). retinas the PBS, in steps ing 10 Diagnostics, Roche dihydrochloride, (4,6-diamidine-2-phenylindole DAPI incubated for 5–7 d. Secondary antibodies were incubated for 2 h together with were antibodies Primary PBS. in Triton X-100 0.5% BSA, and azide 1% sodium 0.02% NDS, 3% in applied were antibodies secondary and Primary 7.4). pH TritonPBS, in 0.5% X-100 (BSA), albumin serum bovine 1% Chemicon), (NDS; serum donkey normal (10% solution a in blocking min for 30–60 ated were performed at room temperature. After washing in PBS, retinas procedures were following of the All incub sucrose. 30% with cryoprotection after cycles freeze-thaw three with treated were retinas The °C. 4 at d 1 KCl, least at for PBS 2.7 NaCl, 137 mM: (in Na 4.3 (PBS) saline phosphate-buffered in dehyde Immunohistochemistry. animals. injected with experiments performing before d 14 of time incubation minimum a was There injections. for used were P30 than older Mice Instruments). (World Precision micromanipulator a by held 5 blunt a using space subretinal the into slowly near the the sclera lens. this hole, Through 2 in needle a with 27-gauge was made sharp incision A small Attane isoflurane. AAVadministration. Subretinal 10 × at 1.84 measured were titers and K, proteinase using AAV the of particles denaturation lowing fol TCCGTGGTGTTGTCG) probe: CCAAGGAAAGGACGATGATTTC; primer: reverse GGCTGTTGGGCACTGACAA; primer: (forward RT-PCR TaqMan using quantified was DNA Encapsidated Optiprep). (Sigma, dient Harvard Cepko, gra iodixanol C. discontinuous a using by purified Vectors were School). provided Medical (both plasmid genes pHGTI-Adeno1 adenoviral the helper and harboring 8, serotype for Cap and Rep2 encoding of repeats AAV2, terminal the internal gene between the AAV-helper plasmid trans the containing plasmid a with 23966) no. (Polysciences, ethylenimine described production. AAV ligation a and performed. was AscI, and NheI using digested and Inc. DNA2.0 by thesized 3 and (GCCACC) sequence 5 modified was Bi-ChR2-2A- EGFP AscI. and NheI using University) Stanford Deisseroth, K. by (provided pAAV-EF1a-double floxed-hChR2(H134R)-EYFP-WPRE-hGHpA pAAV-EF1a-double floxed-bi-ChR2-2A-EGFP-WPRE-hGHpA, we linearized AAV plasmids. Thy1 ( USA Cambridge, University, Harvard Sanes, J. vided by S. Arber, Friedrich Miescher Institute, Basel, Switzerland ( Switzerland Basel, Institute, Miescher by S. Arber, Friedrich vided mice. and GFP #004690). from (B6:FVB-Tg(Pcp2-EGFP)2Yuza, Jackson Laboratory GFP with labeled are cells bipolar rod which in mice, GFP Cre is only expressed in horizontal cells obtained from MMRRC (Tg(Chrnb4-EGFP)1Gsat). Chrnb4 #006795). (B6.Cg-Gnat2cpfl3/Boc, Laboratory Jackson from obtained were M. Biel, Ludwig-Maximilians University, Munich, Germany Ltd. RCC from obtained were type, wild the The ages of the animals varied from 30 to 90 d. C57BL/6 mice, as which served cycle. light-dark 12-h a under maintained were animals and cage per housed were mice Three of Basel-Stadt. Canton of the Department Veterinary by the Guidelines and were approved 86/609/EEC) on the Care and UseAnimals, of Communities Laboratory (European guidelines ethical standard with ance Animals. ONLINE METHODS doi: µ 10.1038/nn.3852 g ml g Stp-EYFP Pvalb 2 HPO -GFP mice, in which cones are selectively labeled with GFP with labeled selectively are cones which in mice, -GFP −1 All animal experiments and procedures were performed in accord were performed and procedures experiments animal All Pcp2- 4 2 ), which was used to visualize cell nuclei. After three final wash final three After nuclei. cell visualize to used was which ), Cre by triple transfection of HEK 293T cells using branched poly branched using cells 293T HEK of transfection triple by ) and D. Paul, Harvard Medical School, Boston, USA ( USA Boston, School, Medical D. and Harvard ) Paul, 4 , 1.47 KH 1.47 , × GFP mice were backcrossed at least five times with C57BL/6 C57BL/6 with times five least at backcrossed were mice GFP We refer to the C128S ChR2 mutant ChR2 We C128S the to refer Thy Serotype 8 recombinant AAVs were made as previously previously as made were AAVs recombinant 8 Serotype Stp-EYFP 12 ′ 2 with an NheI site followed by an optimized Kozak Kozak optimized an by followed site NheI an with genomic copies per ml. per copies genomic PO Retinas were fixed for 30 min in 4% paraformal 4% in min 30 for fixed were Retinas × 4 ′ , pH 7.4) at room temperature and washed with with washed and temperature room at 7.4) pH , with an AscI site. The DNA fragment was syn was fragment DNA The site. AscI an with Cx36 The animals were anesthetized with 2.5% 2.5% with anesthetized were animals The −/− mice 2 4 , , were developed in-house. The 3 5 µ were made by crossing mice pro Cnga3 l l of AAV were injected particles µ l Hamilton syringe that was was that syringe Hamilton l Thy −/− Gja10 2 5 mice were provided by provided were mice Stp-EYFP as bi-ChR2. To bi-ChR2. obtain as -Cre mice, in which 2 3 0 4 ; ; , also known as as known also , , were obtained obtained were , Gnat2 Cx36 Pvalb cpfl3 Chrnb4 4 1 , were were , Pcp2- mice −/− Cre ). ), ), ------

mn--hshnbtrc cd bokn mtbtoi guaae recep tors), 100 glutamate metabotropic blocking acid, amino-4-phosphonobutyric Ringer’s medium (in mM: 110 NaCl, 2.5 KCl, 1 CaCl in isolated were Retinas for 2 recording. h before dark-adapted were Animals slices. retina from and retinas wholemount from taken were types cell retinal T cells. calbindin-positive of number total by the EGFP and calbindin for both positive of number cells the by dividing rate calculated was infection EGFP.The virus-driven the expressing cells calbindin-positive counting by determined was cells horizontal of infected number the regions, retinal same AAV-transduced of regions retinal infected in highly cells calbindin-positive the counting by of number determined was total cells The horizontal cells. horizontal mark to used was Calbindin counts. automatic to false correct were inspected stacks matic manually counting, cell were Cells microscope. counted automatically using confocal the spot function of Imaris the (Bitplane). After auto with scanned was animal per regions retinal infected layer of layer and highly at nuclear outer plexiform three least cells. horizontal infected of Quantification used. was scope C A31571). Invitrogen, (1:200, 647 Fluor Alexa with conjugated IgG anti-mouse donkey A21206), Invitrogen, Fluor (1:200, 488 Alexa IgG with conjugated anti-rabbit A11122), mouse anti-calbindin (1:1,000, Swant, Code 300); donkey secondary: within the cell increased rod depolarization during the experiments. experiments. the during depolarization rod increased concentration cell chloride in the change potential within a that unlikely therefore is It where we others measured it at in the end of and ( series the experimental seal a obtaining after directly it measured we which in experiments in depolarization rod of magnitude the in difference a find not We did Ringer’s medium. the in and pipette the in chloride of concentration the by set was mV) (−60 currents chloride of potential reversal the chloride, with amphotericin B (70 supplemented was solution intracellular the recordings, patch-clamp forated were recorded using perforated patch clamp in current-clamp mode bipolar cells were recorded in current-clamp whole-cell mode. Photoreceptors 0.5 CaCl 10 KCl, 1 gluconate, MgCl potassium 115 mM): (in with were filled pipettes (BF100-50-10, glass M to 8–11 pulled borosilicate Instruments) Sutter from made were electrodes Patch Hz. 50 at filtered low-pass and kHz 1 at digitized were Signals Devices). (Molecular retina. the in deep cells of visualization better enable to filtered digitally was image infrared The retina. the in cells labeled target to used was image photon infrared-two fused A scanning. electrode two-photon recording during the and retina the visualize to used were Instruments) (ref. nm 920 to and wholemount” below) using a two-photon microscope ( Sigma-Aldrich. from HEPES and CPP, strychnine picrotoxin, NBQX, from APB Calbiochem; was obtained acid). piperazineethanesulfonic strychnine (blocking glycine receptors), 20 mM HEPES (4-(2-hydroxyethyl)-1- 10 receptors), kainate and AMPA blocking sulfonamide, 10 tions: a using (VC-6, Warnersystem valve concentra in the following Instruments) at 36 °C, bubbled with 5% CO in Ringer’s were superfused and medium tissues Wholemount slice mination. slices were prepared using a tissue chopper (Stoelting Co.) under illu infrared 200- window. a without but paper filter of type same the on down middle. For retinal slice preparations, retinas were mounted ganglion-cell-side the in window mm 2 × ~2 a with Millipore) (MF-membrane, paper filter on up ganglion-cell-side mounted were retinas Wholemount goggles. infrared NaHCO 22 wo-photon targeted patch clamp recordings. clamp patch targeted wo-photon noa microscopy. onfocal Whole-cell recordings were made using an Axon Multiclamp 700B amplifier in cells “Targeting also (see recording for targeted were cells Fluorescent 8 ) equipped with a Mai Tai HP two-photon laser (Spectra Physics) tuned tuned Physics) a Mai with (Spectra Tai ) laser equipped HP two-photon

with a Plan-Apochromat ×40 oil immersion objective lens (NA 1.2) 1.2) (NA lens objective immersion oil ×40 Plan-Apochromat a with 2 , , 1.5 EGTA, 10 HEPES, 4 ATP-Na µ µ M NBQX (6-nitro-2,3-dioxo-1,4-dihydrobenzo[ NBQX M M M picrotoxin (blocking GABA 3 , bubbled with 5% CO 5% with , bubbled 4 0 ). Infrared illumination and a CCD camera (Diagnostic (Diagnostic camera CCD a and illumination Infrared ). Zis S 70 ae-cnig ofcl micro confocal laser-scanning 700 LSM Zeiss A µ g/ml, g/ml, Sigma). Since amphotericin B is permeable for 2 /95% O 2 /95% O /95% 2 Ω . Pharmacological agents were delivered resistance. For voltage recording, the the recording, For voltage resistance. A 2 and GABA 2 at pH 7.2. Horizontal cells and rod ) using infrared illumination and illumination infrared ) using The entire depth of the inner inner the of depth entire The Recordings from different different from Recordings nature nature 2 , , 1.6 MgCl Supplementary Fig. 1a Supplementary Gja10- C Supplementary Figs. Supplementary 7 receptors) and 10 µ NEUR M APB ( APB M Cre mice. In the the In mice. Cre f ]quinoxaline-7- 2 , , 10 OSCI 4 d 3 µ . . For per -glucose, -glucose, l m-thick m-thick -(+)-2- EN µ C M 2 ). ). E - - - - - ,

© 2014 Nature America, Inc. All rights reserved. (SC10, Thorlabs). The optic fiber of the Polychrome V was plugged into the the into plugged was V Polychrome the of fiber optic The Thorlabs). (SC10, shutter V computer-controlled a Polychrome with a combined used source we light Photonics) retina, (Till the of part ventral the in responses cone and rod of spectrum the recording For source. infrared an as served Corp.) in jector combination with an nm, (750/50 filter Chroma infrared Technology Technology Chroma Corp.). A nm, fast shutter was (470/40 used to gate filter the light path. band-pass The DLP a pro with filtered was from projector light DLP the the Here flash. blue full-field, ms, 50 a was flash switch The ing. filter spectral without light white as projector DLP the from came flash test 1,600 800, 400, 200, 100, 50, (25, presented were diameters different of spots receptors, in ment shown constant across all The light kept levels. contrast was (30, 50, varied 100, 200) for the experi was this and 1,000, was stimulation light the of background) Thorlabs). (FW102, wheels filter two into built were which and Thorlabs) (SC10, was by modulated and density filters, neutral band-pass 0.5 10 × 1.03 to 0.006 of from area collecting a 0.2 and using rods photoreceptors the of sensitivity (photoisomeriza rate s R* second, per rod photoisomerization per tions The Optics). Ocean UV-VIS; the on retina the of stage, with a (USB4000- and power meter (S130VC; Thorlabs) a position spectrometer the at plane, focal the at measured were light was (V-339, projector 229 the Hz by 75 produced of power rate maximum The refresh a Corp.). Vision with PLUS (DLP) processor light digital a with equipped projector a by generated patterns light with stimulated were cells L body cell the of size and shape by the and Thy the in identified were cells ganglion in PV5 EYFP type. cell or GFP respective the expressed that mice in imaging infrared combined and using two-photon performed was targeting cell ganglion and cell horizontal are There no layer). layer.in cones this outer the plexiform from rod bipolar,Cone, (measured layer photoreceptor second the in rods patching and layer by light into the pipette lowering the photoreceptor infrared using was performed targeting Rod contrast). infrared high give and shaped, regularly more and smaller are bodies cell (rod bodies cell to the of owing appearance in change detectable the was layer photoreceptor the to layer nuclear inner the from transition layer.the Similarly, nuclear inner the in bodies cell emerging the to owing detectable was layer nuclear inner the to layer the plexiform inner from transition The layer. photoreceptor the toward was diagonally and moved layer plexiform inner the through retina, outer advanced was the pipette in patch cells the patching For pipette. the inside pressure positive pipette new with area clean a the toward lowered and solution cleaning, intracellular with After filled was solution. Ringer’s mouse patch with a with filled light pipette infrared under membrane limiting inner the of cleaned T recordings. individual 3–20 averaged we cells, other in recordings; from rods are traces The response presented averages individual from 50–100 Inc.). (MathWorks Matlab in analyzed and Instruments) (National Labview on based software using made were Recordings correction. voltage for used we the at cells clamped 0 mV. and was measured potential The junction liquid 7.5 at mM chloride neurobiotin pH 7.2. To input measure inhibitory currents, ATP-Na 1 MgSO mM to M 5–8 pulled electrodes retina. Current recordings were made voltage-clamp in mode, with whole-cell of the axis dorsal-ventral to the relative positions at random text, the in cated indi unless disc, optic the from mm 0.5–1 made were recordings The rods. in of responses the polarity the changed in pipette the concentration chloride lated with large spots at it light levels, background high is not that possible the since when rod Furthermore, stimu bipolar cells displayed hyperpolarization nature nature illumination. full-field provided and microscope the of part epifluorescence argeting cells in wholemount. in cells argeting gt stimulation. ight For most experiments, the contrast (intensity of foreground/intensity of of foreground/intensity of (intensity contrast the experiments, most For Stp-EYFP ± 35 mW/cm 2 , 0.5 mM GTP-Na mM 0.5 , NEUR µ × m). For bi-ChR2 stimulation, we used two types of stimuli. The The stimuli. of types two used we stimulation, bi-ChR2 For m). 4 Cx36 µ , 7.8 × 10 , 7.8 m Supplementary Figure 1 Figure Supplementary OSCI 2 2 (mean for cones for −/− Photoreceptors and bi-ChR2 expressing horizontal horizontal expressing bi-ChR2 and Photoreceptors retina by of the the cell dendrites ganglion stratification EN −3 7 R* s R* C mM CaCl mM ± s.e.m.). The intensity and spectrum of the projected E Ω 3 2 , 5 mM lidocaine lidocaine mM 5 , resistance and filled with 112.5 mM CsCH 112.5 with and filled resistance −1 3 . Light intensity covered 10 log units, ranging ranging units, log 10 covered intensity Light . −1 . The light path was intersected with a shutter shutter a with intersected was path light The . ) was computed on the basis of the spectral spectral the of basis the on computed was ) A small area above the ganglion cells was was cells ganglion the above area small A 2 , 0.5 mM BAPTA, 10 mM HEPES, 4 mM BAPTA, mM , 0.5 HEPES, mM 10 . . Toin photo responses light record 3 5 . N -ethyl bromide (QX314-Br), (QX314-Br), bromide -ethyl Pvalb µ m 3 2 Cre SO for for × 3 ------ ,

t yields good fits; the second filter has a similar form but contains a delayed delayed a contains but form similar a has filter second the fits; good yields where as convolutions, linear functions, time-varying and respectively, potentials, membrane where as calculated cones of description menological M Matlab in written Python. or Instruments) (National software Labview (MathWorks Inc.), using generated were sequences stimulus The took the values values the took in shown traces were fit by eters hand to experimental five the set we that so the rod, we that find data is of well fit in anythe absence nonlinearity, delayed and the filter where reads slice in response rod of model The stimulus. to for allow scale time of off the the slower after turning tapering the response where filters. our all for forms these 0 where as component, slice by an interaction term, term, interaction by an slice It cell). is the horizontal computed simply in the by rod model supplementing r s/R*, rh C y odel. Here we also describe the cone response, response, cone the describe also we Here For the response of both cone in slice and rod in slice, the model param model the slice, in rod and slice in cone both of response the For additional an introduce we accurately, slice in responses rod Todescribe Finally, we model the response of the rod in the wholemount preparation, preparation, wholemount the in rod the of response the model we Finally, = 43 ms, ms, 43 = (where the the (where t I y r R r We model the outer retina response in an extension of the pheno the of extension an in response retina outer the model We ( c = = t = 600 ms, ms, 600 = ( ≤ ) is the light intensity. It was found that a filter with the simple form simple the with intensity. a filter that Itlight found the ) is was t

) and and ) γ V

≤ t K t ( t γ 1 ensures proper normalization. Throughout, we shall adopt adopt shall we Throughout, normalization. proper ensures 1 t C z H = 1 in in 1 = w r ) – – ) R R z = 450 ms, ms, 450 = α in the subscript is a reminder of the cone feedback through through feedback cone the of reminder a is subscript the in ( dr c t K c ( ) dt t t RH = –8.8 × 10 × –8.8 = V t has the same form as the filter a t z ) are calculated with the filters filters the with calculated are ) r R rest r R C y ( t K dr = 17 ms, ms, 17 = ) dt dr = Z t w , , R dt a t R = y t z t y C V R a ( r ( ( ( g = t dr dt ( ( R R ) = ) and and ) γ t ) ) . t t r t K ) and and ) t y ) a = 0.64, 0.64, = e t h = − = y = − = y , that provides inhibitory feedback inhibitory provides that , R R C C − − ( ( −6 t y t − ∫ ∫ t t • − • t y ∫ t ) • w ) z R t t y s/R*, s/R*, dt dt ( 3 + − / dt = ( y ( V = 260 ms, ms, 260 = t 3 t ( ), are related to the light input through through input light the to related are ), ′ ′ . There the photoreceptor response is is response photoreceptor the There . 1 ) t K rest   t K y ) t t ′ α 1 ) t K 1 e t + + − + − y z r y + − + + w R are the instantaneous and resting resting and instantaneous the are = –3.9 × 10 × –3.9 = ) ( d α   ) ( ] [ β ] [ 1 d 1 ) ( ) ( b 1 − R R and and c 1 − R R t R R t w = 1.2 × 10 × 1.2 = − r / t w y I t I t b t z c t b b ′ g I t ′ , according to this model this to according , t z y C C δ ′ ( R R ( β ( ( r t z t t t z ) ( ) ( ) = 2.9 × 10 × 2.9 = are numerical factors. The The factors. numerical are e t t ) ) t ) ) ( z ′ t K   t ′ ( t K r ( ′ ) ( y r r − −3 ) C ) RH y ( t   r s/R*, s/R*, / ( z r −5 C ) t R above. In the of case ) z doi: + s/R*, s/R*, and and −5 H 10.1038/nn.3852 β s/R*. Figure Figure t K r = 3.9 × 10 × 3.9 = t z C r C ( = 24 ms, ms, 24 = ) . 6 . They . They (1) (9) (8) (7) (6) (5) (4) (3) (2) −3 -

© 2014 Nature America, Inc. All rights reserved. response, response, values the took They data. wholemount to fit the in slice data. Thus, only the three horizontal cell parameters were allowed to vary from are in model ones the extracted parameters other all (9): in equation appropriately. Note that the subscript dynamics rod slow the affects it that furthermore, and, intensity light of ades dec several over effective be can range, dynamic a narrow has which output, of properties adaptation temporal and control input as serving output cone feedback, cell The horizontal doi: In equation (9), (9), equation In 10.1038/nn.3852 β h = –2.8 per unit of cone response, response, cone of unit per –2.8 = r h a t enters in such a way as to be modulated by the gain gain the by modulated be to as way a such in enters r H dr dt H r h − = , is calculated in the same formalism, with the with in formalism, same the , is calculated C H r r H appears in the newly added term only, ( 1 + b t r C H r rh H = 300 ms. 300 = ; this ensures that the cone cone the that ensures this ; ) r α H h = 80 per unit of cone cone of unit per 80 = (10) - 43. 42. 41. type). wild versus (knockout allocation group animal for blinded not were experiments The performed. was zation randomi data no and study the from excluded were animals or samples No for * for *** by indicated is Significance (s.e.m.). analysis. Statistical A A

Rae, J., Cooper, K., Gates, P. & Watsky, M. Low access resistance perforated patch perforated resistance P.access Gates, Watsky,Cooper,Low K., & J., M. Rae, Grieger,V.W.Choi, J.C., adeno- of characterization and Production R.J. Samulski, & S. Siegert, recordings using amphotericin B. amphotericin using recordings vectors. viral associated (2009). 1197–1204 Supplementary Supplementary P < 0.001. 0.001. < t al. et n eei ades ok o rtnl el types. cell retinal for book address Genetic shown on figures refers to the number of recorded cells. cells. recorded of number the to refers figures on shown Bar graphs indicate mean and standard error of mean mean of error standard and mean indicate graphs Bar M ethods ethods Nat. Protoc. Nat. C hecklist J. Neurosci. Methods Neurosci. J.

1 , 1412–1428 (2006). 1412–1428 , is available. is P < 0.05, ** for for ** 0.05, < nature nature

37 , 15–26 (1991). 15–26 , NEUR a. Neurosci. Nat. P < 0.01 and and 0.01 < OSCI EN

12 C E - ,