Control of timing, rte nd ursts of hippompl ple ells y dendriti nd somti inhiition Séstien Royer,2, Boris V Zemelmn,5, Attil Losonzy,3, Jinhyun Kim,2, Frnes Chne, Jeffrey C Mgee & György Buzsáki,4 22 Nture Ameri, In. All rights reserved. A onsortium of inhiitory neurons ontrol the firing ptterns of pyrmidl ells, ut their speifi roles in the ehving niml re lrgely unknown. We performed simultneous physiologil reordings nd optogeneti silening of either perisomti (prvlumin () expressing) or dendrite-trgeting (somtosttin () expressing) interneurons in hippompl re CA of hed-fixed mie tively moving tredmill elt rih with visul-ttile stimuli. Silening of either or interneurons inresed the firing rtes of pyrmidl ells seletively in their ple fields, with nd interneurons hving their lrgest effet during the rising nd deying prts of the ple field, respetively. interneuron silening powerfully inresed urst firing without ltering the thet phse of spikes. In ontrst, interneuron silening hd no effet on urst firing, ut insted shifted the spikes thet phse towrd the trough of thet. These findings indite tht perisomti nd dendriti inhiition hve distint roles in ontrolling the rte, urst nd timing of hippompl pyrmidl ells. Computtion in the rin relies on dynmi intertions etween exittory nd inhiitory iruits 4. Appropritely timed inhiition exerted on speifi somtodendriti omprtments of prinipl ells is needed not only to lne exittion, ut lso for the seletive filtering of synpti exittion, timing of spike output, gin ontrol, governing urst firing nd synpti plstiity, nd, t the network level, oordintion of ell ssemlies through mintenne of osilltions nd synhrony,3,5. A lrge diversity of inhiitory interneurons re ville for suh tsks, ut the division of lor mong them nd their speifi ontriutions re poorly understood. In the hippompus, interneurons re known to e involved in severl network tivity ptterns nd rhythms tht support hippompl funtions, suh s sptil nvigtion nd the storge nd rell of episodi informtion,2. During explortion, ple ell tivity is the hrteristi network pttern of the hippompus. Ple ells enode informtion through severl mens suh s firing rte, pttern of dishrge nd spike timing. Inreses in firing rte our when the niml visits their respetive ple field. In the ple field, pyrmidl neurons lternte etween two distint modes of dishrge: single spikes nd ursts of spikes 3. Finlly, spikes re temporlly orgnized in thet osilltions, suh tht neurons with ple fields loted ehind the niml in its running pth dishrge on the desending phse of thet nd neurons with ple fields loted in front of the niml dishrge on the sending phse 4 2. The mehnisms proposed for the ontrol of spike rte, pttern nd timing of CA ple ells rnge from omplete inheritne from upstrem neurons to dominnt role for somti or dendriti inhiition 2,4,2 29. As hnges in rte nd thet phse tend to our together s the niml rosses the ple fields, severl models infer tht oth spets originte from relted mehnisms 2 23, wheres other findings suggest independent implementtions 8,3. Models supporting either of these views involve inhiitory mehnisms, medited y either perisomti interneurons or dendrite-trgeting interneurons 2 29. Reent work hs egun to explore the firing pttern orreltes of ntomilly identified interneuron types during thet osilltions nd other network ptterns in nesthetized nimls 2,5. However, full understnding of their domin-speifi funtions requires the seletive mnipultion of eh neuron type in pproprite ehviorl ontexts. To ddress the ontriution of perisomti nd dendritetrgeting inhiitory interneurons to the firing ptterns of pyrmidl ells, we used omintion of lrge-sle unit reording, optogeneti methods 3 nd forging ehvior on tredmill pprtus, whih enled us to seletively nd folly silene either prvlumin ()- or somtosttin ()-expressing interneurons during ple ell tivity. Our findings show distint roles for perisomti nd dendriti inhiition in ontrolling rte, urst nd thet yle timing of ple ell tivity. RESULTS Ple ell sequenes during sttionry tredmill running To generte sequentil tivtion of hippompl neurons, we trined mie to run hed-fixed 26,32 on tredmill equipped with.7-m-long elt rih in visul nd ttile ues (Fig., Supplementry Fig. nd Supplementry Movie ). The elt ws moved tively y the mouse nd provided ontinuously hnging lol ues while environmentl ues remined sttionry. A wter-surose rewrd ws delivered on Howrd Hughes Medil Institute, Jneli Frm Reserh Cmpus, Ashurn, Virgini, USA. 2 Center for Funtionl Connetomis, Kore Institute of Siene nd Tehnology, Seoul, Repuli of Kore. 3 Deprtment of Neurosiene, Columi University, New York, New York, USA. 4 Center for Moleulr nd Behviorl Neurosiene, Rutgers, The Stte University of New Jersey, Newrk, New Jersey, USA. 5 Present ddress: Center for Lerning nd Memory, University of Texs t Austin, Austin, Texs, USA. Correspondene should e ddressed to J.C.M. (mgeej@jneli.hhmi.org) or G.B. (gyorgy.uzski@nyum.org). Reeived 4 Jnury; epted 27 Ferury; pulished online 25 Mrh 22; orreted fter print 5 June 22; doi:.38/nn.377 NATURE NEUROSCIENCE VOLUME 5 NUMBER 5 MAY 22 769
22 Nture Ameri, In. All rights reserved. Figure Self movement ontrolled lol stimuli generte hippompl ell sequenes. () Experimentl setup for optil proe reording in hed-fixed mie during tredmill running. Visul nd ttile stimuli of the elt re illustrted. Three refletor ues pled t equl intervls (lue, red nd ornge olors re only for illustrtion purposes) triggered photoem sensors for the ontrol of wter rewrd delivery nd lser stimultion. () Running speed s funtion of elt position during the first nd lst dys of trining in mouse. Eh line is single tril. Surose-wter rewrd ws delivered on every tril t position. () Averge speed etween the 57- nd 3-m positions nd numer of trils ompleted fter 3 min (men s.d. of three mie). (d) Wide-nd ( to 5 k) reording of eletril tivity on single shnk of the optil proe. Color tiks indite spikes of isolted CA pyrmidl ells. (e) Ple field of single neuron. Men firing rte, single trils nd spike thet phse s funtion of elt position. (f) Ativity of 55 ple ells reorded simultneously. Eh row is one neuron normlized firing rte. Suessive trils re ontented (only trils 9 3 re shown) nd the ells re ordered ording to their ple field position. Right, verge of 3 trils. every tril (omplete elt rottion) t the sme position of the elt. Both the running speed nd the numer of trils inresed over trining dys, rehing n symptoti level y pproximtely dy 6 (Fig.,). After pproximtely 2 weeks of trining, we performed reordings from the CA hippompl pyrmidl lyer using n eight-shnk silion proe. The lol field potentil (LFP) showed prominent thet osilltions nd thet phse-modulted gmm osilltions during running (Fig. d) 33. Single pyrmidl neurons fired lusters of spikes t speifi lotions of the elt, whih repeted relily over suessive -enphr-gfp 4 µm ori pyr rd -enphr-gfp ori pyr rd Im Spots Speed (m s ) f Cell numer 3 45 Optil proe Tredmill 7 Position (m) 55 9 6 2 2 3 s Dy 4 Dy Speed (m s ) Numer of trils Spines 35 Velro d 4 Dy Units Lol field mv. s 95 5 5 2 25 Tril numer Figure 2 Light-ssisted silening of nd neurons. () Speifi hlorhodopsin expression in hippompl interneurons. Top, low-mgnifition imges of hlorhodopsin-gfp in nd interneurons following virl injetion. Bottom, higher mgnifition of the oxed res. Ori, strtum oriens; pyr, pyrmidl lyer; rd, strtum rditum; lm, strtum lunosummoleulre. () Peristimulus histogrms of simultneously monitored neuronl responses to -s-long light pulses (5 pulses, 7-s interpulse intervl), ligned to their respetive reording shnks in -Cre mouse. () Higher mgnifition of the oxed peristimulus histogrms in. (d) Perent derese nd inrese of firing rtes of ells (lk, 69 ells) nd other non- neurons (white, 722 ells), respetively, shown seprtely for eh shnk (men s.d. of ll neurons d Perentge hnge trils (Fig. e). In these firing fields, the thet phse of spikes shifted forwrd (phse preession ) (Fig. e). Individul pyrmidl ells fired t different positions on the self-ontrolled tredmill so tht the firing fields of the popultion evenly overed the entire elt (Fig. f). These findings indite tht tive movement generted hnges of lol ues re suffiient to generte sequenes of hippompl pyrmidl ell tivity with mny of the physiologil hrteristis of ple ells,34. For this reson, we shll refer to these ells nd their tivity s ple ells nd ple fields. 5 5 n = e Rewrd Lser interneuron numer Pyrmidl ell numer 3 56 f Tril numer Tril numer e Tril numer phse 5 4 2 27 7 Position (m) m Light Light Control Control 8 8 82 83 84 85 86 87 88 89 5 Control Tril numer Light 7 5 7 7 7 Position (m) reorded from sessions in 5 mie). Note tht the lrgest effet ourred on the shnks rrying optil fiers. On shnks loted further wy from the stimulted volume, the mgnitude of hnges ws deresed nd the numer of detetle neurons ws redued (n = nd on the seventh nd eighth shnks, respetively). (e) Normlized tivity of simultneously reorded interneurons nd pyrmidl ells during lternte light nd lightoff (ontrol) trils. (f) Exmple interneuron nd pyrmidl ell firing ptterns. Eh line is single tril. The olor sle is the sme s tht used in e. Right, session verge firing rtes of the two ells (verges of the 5 trils). The lue line indites ontrol trils nd the red line indites light trils. Mgent dots indite signifint rte differenes in different pixels (P <.5, rnk test,, shuffles, see Online Methods). 24 29 Averge 77 VOLUME 5 NUMBER 5 MAY 22 NATURE NEUROSCIENCE
Figure 3 Firing tivity of different ell types during SWRs nd thet osilltions. () Left, popultion verge of peri-ripple histogrms for (n = 23), (n = 8), like (n = 89), non-ursting (n = 8) nd pyrmidl ells (n = 77). Right, distriution of ripple modultion index in the five ell groups. Higher positive indexes indite lrger inreses in firing rte during ripples, wheres lrger negtive indexes indite lrger dereses in firing rte during ripples. The imodl ehvior of neurons indites tht they represent mixture of oriens-lunosum moleulre (ripple suppressed) nd istrtified (ripple tivted) interneurons 2. () Left, popultion verge of thet phse histogrms 4 2 Ripple-triggered verge like Non-ursting Pyrmidl 2 2 ms Firing modultion for (n = 3 ells, 3,369 spikes), (n = 9 ells, 7,78 spikes), like (n = 98 ells, 6,66 spikes), non-ursting (n = 56 ells, 6,36 spikes) nd pyrmidl ells (n = 356 ells, 38,623 spikes) during ontrol trils. Right, distriution of men thet phses. Cell ount 2 2 like 2 Pyrmidl Index Normlized rte phse histogrm 2 4 phse Cell ount 6 33 3 47 Men phse distriution 2 4 phse 22 Nture Ameri, In. All rights reserved. Silening of som- nd dendrite-trgeting interneurons To trnsiently silene interneurons mediting perisomti nd dendriti inhiition 6, we virlly expressed (deno-ssoited virus, AAV) Cre reominse dependent onstrut ontining the NpHReGFP fusion protein (hlorhodopsin) in the CA region of the dorsl hippompus of mie expressing Cre reominse under the ontrol of the promoter for either the prvlumin gene (-Cre mie) or the somtosttin gene (-Cre mie) 3. Histologil exmintion nd in vitro experiments showed tht lrge frtion of -ontining nd -ontining ells in the dorsl CA region expressed high levels of hlorhodopsin (n = 3 -Cre mie, n = 2 -Cre mie; Fig. 2 nd Supplementry Figs. 2 4). The si iophysil properties of the virus-infeted nd non-infeted neurons from -Cre nd Som-Cre mie, respetively, were similr nd the quntity of interneurons ws unltered y hlorhodopsin expression (Supplementry Figs. 2 4). To llow for simultneous reording of single-unit tivity nd fol intivtion of hlorhodopsin-expressing neurons, we equipped four djent shnks of the silion proe with mirometer-sle light guides (Supplementry Fig.,) 3. For testing lol physiologil hnges without ffeting the ptterns of tivity in the entire hippompus, fol light stimultion onfined to the volume of neurons monitored y the reording eletrodes is desired. To test the volume of tissue ffeted y the light stimultion, we delivered -s pulses of light (56 nm, ~ mw t the tip of the shnks) 5 times every 7 s in reording sessions from five -Cre mie (Fig. 2 d). The mgnitude of oth suppression nd enhnement of spiking deresed with distne from the illuminted shnks, inditing tht the light stimultion ws reltively fol 3 (Fig. 2,d). To exmine the onsequenes of interneuron silening on ple ell tivity, we delivered light on lternte trils (Fig. 2e,f) during the sme 57-m segment of the elt (see Fig. ), tht is, in the stedy run portion of the trvel (Fig. ). To ensure tht the nlysis of ple ell tivity is performed during stedy running, we inluded in the nlyses only trils in whih the mie rn this segment in <3 s. The numer of light off (ontrol) nd light on trils meeting this riterion rnged from 8 to 75 (45.6 6.6, men s.d.) nd from 8 to 79 (45.9 7.9), respetively. We isolted totl of,678 ells from four -Cre mie (n = 852 neurons in 8 sessions) nd six -Cre mie (n = 826 neurons in 9 sessions). Identifition of puttive neuron types Prior to nlyzing the effet of interneuron silening on ple ells, we lssified the reorded neurons into puttive ell types (Figs. 3 nd 4 Figure 4 Light-indued hnges in firing ptterns. () Normlized tivity of interneurons, non- interneurons (-like interneurons nd non-ursting ells) nd pyrmidl ells during ontrol (left) nd light (right) trils in -Cre mie (Supplementry Figs. 5 8). Eh line is the verge of ontrol or light trils for single neuron. The ells re ordered ording to the position of their ple field. Vertil yellow lines delimit the light-stimulted segment of the elt. White lines indite group verge. () Popultion verges of firing rtes for eh neuron group during ontrol (lue) nd light (red) trils in -Cre mie. Numers in eh pnel indite the numer of ells inluded in eh verge. Pyrmidl neurons whose ple fields ompletely overed the stimulted re (in) or tht hd no overlp with the stimulted re interneuron numer Non- interneuron numer Pyrmidl ell numer 3 35 Control (out) re shown seprtely. Mgent dots indite signifint rte differenes in different pixels (P <.5, two-til pired t test). () Dt re presented s in for neurons in -Cre mie. (d) Top, inning of the ple fields. Bottom, perent rte inrese in eh in (men s.e.m.) in -Cre nd -Cre mie. The perent rte inrese ws first omputed for individul neuron in eh ple field in overlpping with the stimulted re of the elt nd then the verge inrese of the neuron popultion ws omputed for eh in (*P <.5, **P <.5, # P <.5, unpired t test). Light silening silening like Nonursting Nonursting Pyrmidl Out In n = 3 256 7 7 7 7 Position (m) Position (m) Position (m) Normlized rte 2 86 23 34 5 53 42 27 9 d Perentge hnge 8% 4% 3 2 3 4 5 Bin numer # # # # * * * ** ** 2 3 4 5 Bin numer NATURE NEUROSCIENCE VOLUME 5 NUMBER 5 MAY 22 77
22 Nture Ameri, In. All rights reserved. Proility (%) LFP Units 4 4 Burst * * # ##* Single spikes Single * Burst nd Supplementry Figs. 5 7). Hlorhodopsin-expressing nd interneurons were signifintly suppressed y light in most pixels of the stimulted segment (P <.5; Fig. 4 nd Supplementry Fig. 5). An verge of 4.6 (rnge, 2 to 7) neurons were identified per session. Only nine neurons were identified y light, likely euse the reordings were mde mostly in the pyrmidl lyer, wheres most ell odies re loted in the strtum oriens. We lssified the remining neurons into three groups y their spike-urst mgnitude nd refrtory period (Supplementry Fig. 5): -like neurons tht hd similr physiologil hrteristis s identified neurons (Supplementry Figs. 5 7), non-ursting neurons (puttive other interneurons) nd pyrmidl neurons. Pyrmidl neurons inluded virtully ll ells tht were identified s presynpti exittory ells (Supplementry Fig. 6) nd hd defined ple fields (Fig. 4 nd Supplementry Fig. 7). nd -like neurons hd short refrtory periods, high firing rtes nd fired on most prts of the elt (Supplementry Figs. 5 7). In ddition, they ounted for most of the neurons exited y the puttive pyrmidl ells (4% of neurons were monosynptilly exited, versus 4% of non-ursting nd.5% of pyrmidl ells; Supplementry Fig. 6). nd nonursting neurons showed heterogeneous firing ptterns in terms of firing rtes, refrtoriness nd ursting (Supplementry Figs. 5 7). Given tht histologilly identified interneurons were previously tegorized y thet phse preferene nd shrp wve ripples 2,5,7 (SWRs), we lso exmined these physiologil fetures. During periods of immoility, thet osilltions disppered nd were repled y SWRs. For three reording sessions from one -Cre mouse nd eight sessions from six mie, the reordings were extended to inlude long periods of immoility with SWR events. We deteted 29 933 SWRs per session y pplying threshold to the ndpss-filtered (2 7 ) LFP tre nd orreting y visul inspetion. The mjority of nd -like interneurons formed reltively homogeneous group tht fired roustly nd mximlly during SWRs, s in the rt 5,7 (Fig. 3 nd Supplementry Fig. 8). In ontrst, nd non-ursting 5 5 2 25 2 3 4 5 Interspike intervl (ms) Inrese (%) 3 4 Single * # ms # Burst length Figure 5 interneurons ontrol spike ursts in pyrmidl ells. () LFP nd unit tivity of one pyrmidl ell exmple, illustrting single spikes nd urst dishrge. () Averge (men ± s.e.m.) proility distriution of interspike intervls of pyrmidl neurons. Control (lue) nd light (red) histogrms inlude 2,54 nd 23,557 spikes for -Cre mie (23 ells) nd 6,336 nd 9,68 spikes for -Cre mie (98 ells) (omprison etween ontrol nd light, *P <.5, # P <.5; two-til pired t test). () Reltive inrese (men ± s.e.m.) of ourrene for different urst lengths (omprison etween -Cre nd -Cre mie, *P <.5, # P <.5; unpired t test). neurons showed heterogeneous firing ptterns, either eing suppressed or exited during SWRs. This imodlity my reflet the presene of two supopultions with distint SWR preferene (oriens-lunosum moleulre nd istrtified interneurons) 2. Individul pyrmidl ells were either strongly reruited or strongly suppressed during SWRs, with minority tht were not influened y SWRs t ll 7. Most of the nd -like neurons fired preferentilly efore the trough of the thet wve, wheres ll of the neurons were phseloked to the sending phse of the yle (Fig. 3) 2. Memers of the non-ursting group fired t vrious phses, ut voided the pek of thet. Pyrmidl ells hd the widest rnge of phse preferene 7. As these oservtions were mde on limited dtsets, pproprite popultion pttern lssifition will require more extensive studies. Interleved gin ontrol of firing rtes We first exmined the effet of light stimultion on the firing rte of pyrmidl ells. To ompre the effet of light stimultion inside nd outside of the ple fields, we seprtely omputed the verge firing fields of neurons depending on whether their ple fields were inside or outside of the stimulted segment. For oth nd interneuron silening, we oserved signifint inreses in firing rte only in the ple fields (P <.5; Fig. 4,), suggesting tht the sene of firing outside of the ple field is minly result of lk of exittory drive rther thn n enhned inhiition. Next, we exmined the effet of light on different prts of the ple fields. We divided eh field into five sptil ins nd omputed the perent hnge in firing rte etween ontrol nd light trils for eh in loted in the stimulted segment of the elt. Only ins ontining t lest 2 spikes during ontrol trils were onsidered to ensure relile sttistil signifine of the perent hnge estimte. We then omputed the verge perent hnge ross the neuron popultion for eh in (Fig. 4d). We found tht the effet of nd neuron silening exhiited opposite trends, s the influene of interneurons ws progressively redued, wheres tht of interneurons grdully inresed in the ple fields. Together, these dt indite db LFP Filtered LFP (5 ) 8 6 db 2 6 mv s 2 2 Control Light 4 8 2 db.5 mv ms 5 6 Figure 6 Thet osilltions re not ffeted y fol disinhiition. () Wide-nd LFP segment during light stimultion (green line) nd thet nd filtered tre (ottom). Right, three thet yles during lights off nd light on onditions. () Power spetr lulted seprtely for ontrol nd light trils of representtive reording session in -Cre mouse. Inset, zoom on thet nd. () Averge power spetr of ll sessions from -Cre mie. Dsh lines indite stndrd error (n = 8 sessions, P >.5, two-til pired t test). 9 772 VOLUME 5 NUMBER 5 MAY 22 NATURE NEUROSCIENCE
8% Men phse phse histogrm phse Light Control 4 2 4% 2 3 4 5 Bin numer phse histogrm Non-ursting like.5 Normlized rte 7 Bin 4 2 4 Position (m) phse silening silening 22 Nture Ameri, In. All rights reserved. Normlized rte Bin numer.5 2 3 4 5 /2 m Phse shift 7 /2.4.3 P =.2 Figure 7 Within-thet timing of spikes is regulted y interneurons. () Left, ple field of n exmple pyrmidl ell from -Cre mouse nd spike thet phse s funtion of elt position during ontrol (lue) nd light (red) trils. Right, thet phse histogrm of spikes in in 4 of the ple field (ontrol, n = 6 spikes, 8 trils; light, n = 69 spikes, 8 trils). The rrow indites the diretion of phse shift during disinhiition. () Comprison of rte nd thet phse shift of spikes in -Cre nd -Cre mie, shown seprtely for rte-ffeted (upwrd rrow) nd non-ffeted sugroups (O). Top, sugroup men firing rtes during ontrol nd light trils (, 78 non-ffeted ells, 69 rte-ffeted ells;, 59 non-ffeted ells, 5 rteffeted ells). Mgent dots indite signifint rte differenes in different pixels (P <.5, two-til pired t test). Bottom, sugroup men phse shifts (±s.e.m.). P vlues for ins with signifint phse shifts re indited (see Online Methods). () Men thet phses ( s.e.m.) during ontrol nd light trils for (n = 3), (n = 9), -like (n = 2 nd 86 in -Cre nd -Cre mie respetively), non-ursting (n = 23 nd 33 in -Cre nd -Cre mie, respetively) nd pyrmidl ells (from -Cre mie, sme ell numers s in ). Ornge shding indites dt from -Cre mie. Pyrmidl neurons in -Cre mie with no rte hnge to light (O) nd with signifint inreses (upwrd rrow) to light re shown seprtely. Note the lte-to-erly phse dvnement of spikes ross the ins of the field nd the signifint phse shifts in the rte ffeted group. Right, verge thet phse histogrms for nd interneurons nd in ins of ple field for the rte ffeted pyrmidl ells (ontrol, n = 5,872 spikes; light, 8,276 spikes; see lso Supplementry Fig. 9). Pyrmidl ells Bin 2 3 4 5 2 2 4 phse phse tht the two lsses of interneurons do not ontriute to the ell shpe pperne of ple fields, ut insted oordinte to produe n interleved gin ontrol on firing tivity in the ple fields. Control of urst firing y dendriti inhiition Hippompl pyrmidl ells fire oth single spikes nd spike ursts 3 (Fig. 5), nd these different firing modes hve somewht different thet phse preferene 35. Previous intrellulr work hs suggested tht genertion of spike ursts involves either dendriti 36 38 or somti 39 mehnisms. In vitro experiments mnipulting dendriti inhiition fvor dendrites s the soure of spike ursts 37,4. We defined spike ursts in pyrmidl ells s events with more thn two spikes with <9-ms interspike intervls 3,4. Silening interneurons during tredmill running gretly inresed the proility of pyrmidl ell urst firing, prtiulrly promoting the dishrge of longer ursts (perent inrese rnging from 45% for spike doulets to 28% for ursts of 5 spikes; Fig. 5, nd Supplementry Fig. 9). In ontrst, silening interneurons produed muh more modest effet on spike ursts, exept for spike doulets (3% inrese). Thus, these results indite tht dendriti inhiition y interneurons, ut not somti inhiition, re ritil for ontrolling spike urst firing during tive explortion, extending previous in vitro oservtions 37,4. Control of spike thet phse y somti inhiition Given tht the men firing rtes of ple ells were elevted to similr degree y oth perisomti nd dendriti disinhiition, we sked whether these two forms of inhiition hve similr effets on ple ell spike timing. We first investigted the effets of light stimultion on the LFP. The light stimultion did not ffet the LFP thet (5 ) tivity (Fig. 6), whih ws likely result of volume ondution of the field from surrounding unffeted res nd/or the fol perturtion of inhiition. Next, we determined the spike-thet phse reltionship of single neurons using two omplementry mesures: the spike phse histogrm nd the men thet phse of the spikes 42. For pyrmidl ells, we lulted these mesures for eh ple field in seprtely (Fig. 7). Ple ells whose firing rtes were inresed signifintly y light (ffeted group) nd pyrmidl neurons tht were not ffeted (unffeted group) were nlyzed independently (Fig. 7). We found no signifint effet on spike phse during the silening of NATURE NEUROSCIENCE VOLUME 5 NUMBER 5 MAY 22 773
22 Nture Ameri, In. All rights reserved. interneurons (P >.5). In ontrst, we oserved signifint phse shift in severl ins of the ffeted group during the silening of interneurons. In these ells, light stimultion hevily ltered the spike thet-phse reltionship profile s it indued forwrd phse shift t the eginning of the field nd kwrd shift in the enter nd outound regions (Fig. 7 nd Supplementry Fig. ). Together, perisomti disinhiition used the spikes to e gthered towrd the trough of thet, reduing the phse differene etween the onset nd the pek of the ple fields y more thn hlf (from 244 to 7 ; Fig. 7). These dt indite tht perisomti inhiition is fr more effetive t regulting spike timing thn dendriti inhiition, even though their silening produed the sme level of disinhiition. DISCUSSION We found tht the rte, timing nd urst firing of pyrmidl neuron tivity re ontrolled oopertively y interneuron medited perisomti nd interneuron medited dendriti inhiitory mehnisms. Both types of inhiition influened the dishrge rte inside of, ut not outside of, the ple fields, inditing tht neither somti nor dendriti () inhiition is ritil for djusting the sptil sle of ple field. In the ple field, the strength of somti nd dendriti inhiition evolved in n interleved mnner, with the strongest interneuron nd interneuron medited inhiitory effets eing present t the eginning nd end of the ple field, respetively. This pttern fits in vitro dt reporting grdul shift of inhiition from somti to dendriti sites when CA pyrmidl ells re repetitively tivted t thet frequeny 43, nd my e explined y short-term plstiity t the exittory nd/or inhiitory synpses 43,44. The omplementry ontrol of nd inhiition of ple ell firing n lso ount for the in vivo oservtion tht spike phse vriility in the deying prt of ple field is onsiderly lrger thn tht in the rising prt 6,7. Dendriti inhiition ws most effetive t ontrolling the urst dishrge of pyrmidl ells, nd this effet likely reflets n enhned ility of dendriti inhiition to regulte dendriti plteu potentil genertion 36 38. As urst firing is prtiulrly effetive t promoting synpti plstiity nd downstrem exittion 38,45, this finding suggests tht dendriti inhiition is ritil for seleting the CA inputs to e strengthened nd the CA outputs to e most effetively trnsmitted to downstrem res. Thus, dendriti inhiition is potentilly n importnt omponent of the input omprison type of omputtion hypothesized to e performed in re CA (refs. 38,46). Notly, suppression of dendriti inhiition did not ffet the spike phse preferene of pyrmidl ell firing even though interneurons showed strong thet-phse firing preferene. On the other hnd, lol suppression of perisomti inhiition resulted in firing rte elevtions tht o-ourred with shifts of spike phse preferene. Thus, spike rte nd thet phse were either oupled or independently ontrolled depending on the somtodendriti origin of disinhiition. It therefore ppers tht perisomti inhiition possesses unique ility to regulte spike phse preferene nd it will e importnt to determine in the future the ext ellulr/network level mehnisms involved. At this point, however, the dt re onsistent with the following sheme. As interneurons provide thet phse-loked perisomti inhiition tht integrtes with depolrizing exittory input from the dendrites 2,24,26, silening of perisomti inhiition inreses the level of som/xonl depolriztion ner the trough of thet, dvning spike phse t the strt of the field nd delying spike phse t lter lotions. The spike phse preession tht remins fter interneuron silening ould e the result of n inomplete silening or, more likely, indites tht dditionl mehnisms re involved. These inlude phse dvne of inoming exittion nd/or influene from other types of interneurons,24. In this sheme, under norml onditions, dendriti inhiition dereses the mount of exittory depolrizing urrent rehing the som/xon y removing exittion t the site of input nd, perhps more importntly, y limiting regenertive dendriti tivity 4. During dendriti disinhiition, the inrese in firing rte is, in prt, result of the onversion of single spikes into ursts of multiple tion potentils. However, the timing of the dditionl spikes is mintined, s the perisomti inhiition responsile for regulting spike phse is still intt. Previous work hs shown tht perisomti nd dendriti inhiitory ells re differentilly trgeted y suortil neuromodultors nd other extrhippompl fferents 3,33, nd this ws suggested to provide flexile sustrtes for ontext- nd ehviorl stte dependent reonfigurtion of the hippompl network during lerning 47. Our results extend this knowledge y demonstrting tht dendriti or perisomti inhiitory iruits differentilly regulte hippompl output modes ritil for sptil nvigtion nd memory mehnisms 48. Given tht these fetures re fundmentl elements of ortil networks, it is likely tht the distint mehnisms of rte, urst nd phse ontrol desried here re lso in ple in other ortil strutures. METHODS Methods nd ny ssoited referenes re ville in the online version of the pper t http://www.nture.om/ntureneurosiene/. Note: Supplementry informtion is ville on the Nture Neurosiene wesite. ACKNOWLEDGMENTS We thnk G.F. Turi, T. Adelmn nd T. Thnik for tehnil ontriutions, nd D. Huer for useful dvie. This work ws supported y the Howrd Hughes Medil Institute, the US Ntionl Institutes of Helth (NS34994, MH5467), the Jmes S. MDonnell Foundtion nd the World Clss Institute (WCI) Progrm of the Ntionl Reserh Foundtion of Kore (NRF) funded y the Ministry of Edution, Siene nd Tehnology of Kore (NRF grnt numer: WCI 29-3). A.L. ws supported y the Kvli nd Serle Foundtions. AUTHOR CONTRIBUTIONS The experiments were designed y S.R., J.C.M. nd G.B. S.R. developed the tredmill nd optil proe nd performed the in vivo experiments nd nlysis. B.V.Z. nd J.K. generted the Cre-dependent viruses. B.V.Z. generted the -Cre mie. A.L. nd F.C. performed the in vitro physiologil experiments. A.L. performed the histologil experiments. S.R., J.C.M. nd G.B. wrote the mnusript. COMPETING FINANCIAL INTERESTS The uthors delre no ompeting finnil interests. Pulished online t http://www.nture.om/ntureneurosiene/. Reprints nd permissions informtion is ville online t http://www.nture.om/ reprints/index.html.. Freund, T.F. & Buzsáki, G. Interneurons of the hippompus. Hippompus 6, 347 47 (996). 2. Kluserger, T. & Somogyi, P. Neuronl diversity nd temporl dynmis: the unity of hippompl iruit opertions. Siene 32, 53 57 (28). 3. Mrkrm, H. et l. Interneurons of the neoortil inhiitory system. Nt. Rev. Neurosi. 5, 793 87 (24). 4. Isson, J.S. & Snzini, M. How inhiition shpes ortil tivity. Neuron 72, 23 243 (2). 5. Kluserger, T. et l. Brin stte nd ell type speifi firing of hippompl interneurons in vivo. Nture 42, 844 848 (23). 6. Crdin, J.A. et l. Driving fst-spiking ells indues gmm rhythm nd ontrols sensory responses. Nture 459, 663 667 (29). 7. Csisvri, J., Hirse, H., Czurkó, A., Mmiy, A. & Buzsáki, G. Osilltory oupling of hippompl pyrmidl ells nd interneurons in the ehving Rt. J. Neurosi. 9, 274 287 (999). 8. Runyn, C.A. et l. Response fetures of prvlumin-expressing interneurons suggest preise roles for sutypes of inhiition in visul ortex. Neuron 67, 847 857 (2). 774 VOLUME 5 NUMBER 5 MAY 22 NATURE NEUROSCIENCE
22 Nture Ameri, In. All rights reserved. 9. Sillito, A.M. The ontriution of inhiitory mehnisms to the reeptive field properties of neurones in the strite ortex of the t. J. Physiol. (Lond.) 25, 35 329 (975).. Wehr, M. & Zdor, A.M. Blned inhiition underlies tuning nd shrpens spike timing in uditory ortex. Nture 426, 442 446 (23).. O Keefe, J. Hippompl neurophysiology in the ehving niml. in The Hippompus Book (eds. Andersen, P., Morris, R.G.M., Amrl, D.G., Bliss, T.V.P. & O Keefe, J.) 475 548 (Oxford, Oxford Neurosiene, 26). 2. MNughton, B.L., Bttgli, F.P., Jensen, O., Moser, E.I. & Moser, M.B. Pth integrtion nd the neurl sis of the ognitive mp. Nt. Rev. Neurosi. 7, 663 678 (26). 3. Rnk, J.B. Jr. Studies on single neurons in dorsl hippompl formtion nd septum in unrestrined rts. I. Behviorl orreltes nd firing repertoires. Exp. Neurol. 4, 46 53 (973). 4. O Keefe, J. & Ree, M.L. Phse reltionship etween hippompl ple units nd the EEG thet rhythm. Hippompus 3, 37 33 (993). 5. Jensen, O. & Lismn, J.E. Hippompl CA3 region predits memory sequenes: ounting for the phse preession of ple ells. Lern. Mem. 3, 279 287 (996). 6. Skggs, W.E., MNughton, B.L., Wilson, M.A. & Brnes, C.A. Thet phse preession in hippompl neuronl popultions nd the ompression of temporl sequenes. Hippompus 6, 49 72 (996). 7. Drgoi, G. & Buzsáki, G. Temporl enoding of ple sequenes y hippompl ell ssemlies. Neuron 5, 45 57 (26). 8. Huxter, J., Burgess, N. & O Keefe, J. Independent rte nd temporl oding in hippompl pyrmidl ells. Nture 425, 828 832 (23). 9. Huxter, J.R., Senior, T.J., Allen, K. & Csisvri, J. Thet phse speifi odes for two-dimensionl position, trjetory nd heding in the hippompus. Nt. Neurosi., 587 594 (28). 2. Geisler, C. et l. Temporl delys mong ple ells determine the frequeny of popultion thet osilltions in the hippompus. Pro. Ntl. Ad. Si. USA 7, 7957 7962 (2). 2. Kmondi, A., Asády, L., Wng, X.J. & Buzsáki, G. Thet osilltions in somt nd dendrites of hippompl pyrmidl ells in vivo: tivity-dependent phsepreession of tion potentils. Hippompus 8, 244 26 (998). 22. Meht, M.R., Lee, A.K. & Wilson, M.A. Role of experiene nd osilltions in trnsforming rte ode into temporl ode. Nture 47, 74 746 (22). 23. Hrris, K.D. et l. Spike trin dynmis predits thet-relted phse preession in hippompl pyrmidl ells. Nture 47, 738 74 (22). 24. Losonzy, A., Zemelmn, B.V., Vziri, A. & Mgee, J.C. Network mehnisms of thet relted neuronl tivity in hippompl CA pyrmidl neurons. Nt. Neurosi. 3, 967 972 (2). 25. Mgee, J.C. Dendriti mehnisms of phse preession in hippompl CA pyrmidl neurons. J. Neurophysiol. 86, 528 532 (2). 26. Hrvey, C.D., Collmn, F., Domek, D.A. & Tnk, D.W. Intrellulr dynmis of hippompl ple ells during virtul nvigtion. Nture 46, 94 946 (29). 27. Hngy, B., Li, Y., Muller, R.U. & Czurkó, A. Complementry sptil firing in ple ell-interneuron pirs. J. Physiol. (Lond.) 588, 465 475 (2). 28. Geisler, C., Roe, D., Zugro, M., Sirot, A. & Buzsáki, G. Hippompl ple ell ssemlies re speed-ontrolled osilltors. Pro. Ntl. Ad. Si. USA 4, 849 854 (27). 29. Murer, A.P., Cowen, S.L., Burke, S.N., Brnes, C.A. & MNughton, B.L. Phse preession in hippompl interneurons showing strong funtionl oupling to individul pyrmidl ells. J. Neurosi. 26, 3485 3492 (26). 3. Hirse, H., Czurkó, A., Csisvri, J. & Buzsáki, G. Firing rte nd thet-phse oding y hippompl pyrmidl neurons during spe lmping. Eur. J. Neurosi., 4373 438 (999). 3. Royer, S. et l. Multi-rry silion proes with integrted optil fiers: light-ssisted perturtion nd reording of lol neurl iruits in the ehving niml. Eur. J. Neurosi. 3, 2279 229 (2). 32. O Connor, D.H., Huer, D. & Svood, K. Reverse engineering the mouse rin. Nture 46, 923 929 (29). 33. Buzsáki, G. et l. Hippompl network ptterns of tivity in the mouse. Neurosiene 6, 2 2 (23). 34. Olton, D.S., Beker, J.T. & Hndelmnn, G.E. Hippompus, spe nd memory. Behv. Brin Si. 2, 33 322 (979). 35. Mizuseki, K., Sirot, A., Pstlkov, E. & Buzsáki, G. Thet osilltions provide temporl windows for lol iruit omputtion in the entorhinl-hippompl loop. Neuron 64, 267 28 (29). 36. Buzsáki, G., Penttonen, M., Nádsdy, Z. & Brgin, A. Pttern nd inhiitiondependent invsion of pyrmidl ell dendrites y fst spikes in the hippompus in vivo. Pro. Ntl. Ad. Si. USA 93, 992 9925 (996). 37. Miles, R., Tóth, K., Gulyás, A.I., Hájos, N. & Freund, T.F. Differenes etween somti nd dendriti inhiition in the hippompus. Neuron 6, 85 823 (996). 38. Tkhshi, H. & Mgee, J.C. Pthwy intertions nd synpti plstiity in the dendriti tuft regions of CA pyrmidl neurons. Neuron 62, 2 (29). 39. Azouz, R., Jensen, M.S. & Yri, Y. Ioni sis of spike fter-depolriztion nd urst genertion in dult rt hippompl CA pyrmidl ells. J. Physiol. (Lond.) 492, 2 223 (996). 4. Lovett-Brron, M. et l. Regultion of neuronl input output trnsformtion y tunle dendriti inhiition. Nt. Neurosi. 5, 423 43 (22). 4. Hrris, K.D., Hirse, H., Leinekugel, X., Henze, D.A. & Buzsáki, G. Temporl intertion etween single spikes nd omplex spike ursts in hippompl pyrmidl ells. Neuron 32, 4 49 (2). 42. Royer, S., Sirot, A., Ptel, J. & Buzsáki, G. Distint representtions nd thet dynmis in dorsl nd ventrl hippompus. J. Neurosi. 3, 777 787 (2). 43. Pouille, F. & Snzini, M. Routing of spike series y dynmi iruits in the hippompus. Nture 429, 77 723 (24). 44. Kullmnn, D.M. & Lms, K.P. Long-term synpti plstiity in hippompl interneurons. Nt. Rev. Neurosi. 8, 687 699 (27). 45. Lismn, J.E. Bursts s unit of neurl informtion: mking unrelile synpses relile. Trends Neurosi. 2, 38 43 (997). 46. Vinogrdov, O.S. Hippompus s omprtor: role of the two input nd two output systems of the hippompus in seletion nd registrtion of informtion. Hippompus, 578 598 (2). 47. Freund, T.F. Interneuron Diversity series: rhythm nd mood in perisomti inhiition. Trends Neurosi. 26, 489 495 (23). 48. Dupret, D., O Neill, J., Pleydell-Bouverie, B. & Csisvri, J. The reorgniztion nd retivtion of hippompl mps predit sptil memory performne. Nt. Neurosi. 3, 995 2 (2). NATURE NEUROSCIENCE VOLUME 5 NUMBER 5 MAY 22 775
22 Nture Ameri, In. All rights reserved. ONLINE METHODS All experiments were onduted in ordne with institutionl regultions (Institutionl Animl Cre nd Use Committees of the Howrd Hughes Medil Institute, Jneli Frm Reserh Center nd Rutgers University). Virus preprtion nd injetion. The NpHReGFP fusion protein ws loned into n deno-ssoited virl ssette ontining the mouse synpsin promoter, woodhuk post-trnsriptionl regultory element (WPRE), SV4 polydenyltion sequene, nd two inverted terminl repets. raav-flex-rev- NpHReGFP 49 ws ssemled using modified helper-free system (Strtgene) s serotype 2/7 (rep/p genes) AAV, nd hrvested nd purified over sequentil esium hloride grdients s previously desried 5. The virus ws injeted into the dorsl hippompl CA region of Pvlre 5 (-Cre) trnsgeni mie (hyrid C57BL/6-29/SV, 3 5 weeks old, Jkson Lortory) t three sites: 2.2, 2.4 nd 2.7 mm posterior from regm, nd 2. mm from midline. We injeted 2 nl of virus every 5 m from.55 mm to.95 mm elow pi. The pipette ws held t.95 mm for 3 min efore eing ompletely retrted from the rin. Sst-re 4 mie (-Cre, hyrid C57BL/6-29/SV, 4 7 weeks old) were injeted t one site 2. mm posterior from regm nd.6 mm from the midline, with 4 5 nl of virus injeted.5 nd.5 mm elow the pi. In vitro eletrophysiology. -Cre mie were injeted with Cre-dependent enphr-sfgfp (n = 8) or with Cre-dependent tdtomto AAV (n = 5) into the CA region of the dorsl hippompus. -Cre mie were injeted with Cre-dependent enphr-sfgfp (n = 6) or rossed with tdtomto Cre reporter mie (B6;29S6-Gt(ROSA)26Sortm9(CAG-tdTomto)e/J from Jkson Lortory). -Cre Cre reporter mie (n = 7) were injeted with raav- FLEX-ChR2-sfGFP. The nimls were deeply nesthetized nd depitted 2 5 weeks fter the injetion nd oronl slies (35 4 m) from the dorsl prt of the hippompus were prepred. Cells were visulized using epifluoresene mirosopes equipped with Dodt grdient ontrst or differentil interferene ontrst (DIC) under infrred illumintion nd wter immersion lens (4 with.8 NA). Experiments were performed t physiologil temperture (32 35 C) in rtifiil ererospinl fluid ontining 25 mm NCl, 2.5 3 mm KCl, 25 mm NHCO 3,.25 mm NH 2 PO 4,.3 2 mm CCl 2, mm MgCl 2, mm sorte, 3 mm pyruvte nd 2 25 mm gluose, nd sturted with 95% O 2 nd 5% CO 2. Whole-ell urrent-lmp reordings from interneurons nd pyrmidl neurons, loted in the injeted re, were performed using pth-lmp mplifier in the tive ridge mode, filtered t 3 k nd digitized t 5 k. Reording eletrodes were filled with n internl solution ontining either 2 mm potssium gluonte, 6 mm KCl, mm HEPES, 4 mm NCl, 8 mm Mg 2 ATP,.6 mm Tris 2 GTP, 28 mm phosphoretine nd. mm Alex 594, ph = 7.25, or 34 mm potssium gluonte, 6 mm KCl, mm HEPES, 4 mm NCl, 4 mm Mg 2 ATP,.3 mm Mg 2 GTP, 4 mm phosphoretine nd. mm Alex 594, ph = 7.25. Interneurons expressing enphr-sfgfp or tdtomto were identified under two-photon mirosopy nd were susequently pthed under Dodt grdient ontrst or DIC. For enphr tivtion, yellow light (EXFO Exit 2 r lmp, 582/2 nm exittion nd pss filter,.5 mw t prep) ws delivered using full-field epi-fluoresene pth of the mirosope. o-loliztion, we identified individul NpHR-sfGFP leled interneurons nd susequently sored for immunoretivity. To void ross-tlks etween smples, we olleted sns from eh smple in multiple trk mode. Preprtion for in vivo experiment. During first surgery under isofluorne nesthesi, five mle -Cre nd six -Cre mie were injeted with Credependent enphr-sfgfp AAV in the dorsl hippompus (left hemisphere) s desried ove. To prepre mie for hroni reordings in the tredmill, two smll wth-srews were driven into the one ove the ereellum to serve s referene nd ground eletrodes. A ustom-frited pltinum hed-plte with window opening ove the left hippompus ws emented to the skull with dentl ryli. Behviorl trining. After post-surgery reovery period of 7 d, the mie were wter restrited to ml d of wter nd trined for 2 weeks (4-min sessions per dy) to run on the tredmill with their hed fixed (the hed plte lmped to holder devie 32 ). Surose-in-wter (%) rewrds were delivered every tril t the sme position of the elt vi liking port (Supplementry Movie ). After ehviorl lerning rehed n symptote, the nimls ompleted 64 to 24 trils in the first 3 min of the sessions. The quntity of surose-in-wter drnk in the tredmill ws mesured fter eh session, nd dditionl wter ws provided in order tht the mie drnk totl mount of ml d. Reording proedures. We previously desried the preprtion of mirometersle fiers nd onstrution of silion optrodes 3. Reordings were performed 3 6 weeks fter the virus injetion. On reording dys, the mie were initilly nesthetized with isofluorne nd their heds were fixed in the tredmill. On the first dy of reording, the hole used for the virus injetion ws enlrged nd the dur ws removed (on susequent dys, the hole ws simply lened with sline). The optil proe ws fixed to miromnipultor nd lowered into the rin. The hole ws then seled with liquid gr (.5%) pplied t ner ody temperture. Aluminum foil ws folded round the entire optil proe ssemly, whih oth served s Frdy ge nd prevented the mie from seeing the light emitted y the optil fiers. After rehing the CA pyrmidl lyer with the proe, mie were llowed to reover ompletely from the nesthesi. Reording sessions typilly lsted for 9 min, during whih the niml s ehvior lternted etween periods of running nd immoility. After eh reording session, the proe ws removed nd the hole ws filled with mixture of one wx nd prffin oil, nd overed with silion selnt (WPI, Kwik-sil). Eh mouse ws reorded for mximum of three sessions (one session per dy). Light stimultion. A DPSS lser (56 nm, mw, Crystlser) ontrolled y trnsistor-trnsistor logi ws used for hlorhodopsin tivtion. To djust the intensity of the lser, we pled neutrl density filter wheel in front of the em. An optil proe with four optil fiers ws used (Supplementry Fig. ), thus, the lser em ws first split with em splitters (ThorLs #CM-BS) nd diverted y refleting mirrors (ThorLs #CM-P) into four seprte fier ports (ThorLs #PAF-X-7-A). Long single-mode optil fiers onneted the fier ports to the proe fiers. Antomy. -Cre mie were injeted unilterlly in the CA region of the dorsl hippompus with Cre-dependent enphr-sfgfp AAV. Animls were nesthetized nd perfused trnsrdilly with 4% prformldehyde in phosphte uffer, 4, 23 nd 32 dys fter virl infetion (n = 2 nimls in eh ge group). We ut 6- m-thik oronl setions using virtome. Setions were loked in 5% norml got serum (NGS, vol/vol) nd.% Triton X- (vol/vol). Slies were then inuted for 48 h t 4 C in 5% NGS,.% Triton X- nd primry ntiody (R- - polylonl, :3). After Tris-uffered sline wshes, setions were then inuted with Alex Fluor(r) 633 onjugted donkey ntiody to rit (:2). After wshes, the setions were mounted nd overslipped on slides in Vetshield mounting medium. Confol stk imges (2 3 slies, - m optil thikness) from the CA region of the injeted nd the ontrlterl ontrol hippompi were quired (2 ojetive,.5 NA). Numer of enphr-sfgfp nd immunopositive ell odies ws ounted on stk imges. To quntify Behvior ontrol nd dt quisition. The vlve for rewrd delivery, the light em sensors nd the lser ontrol iruit were onneted to omputer ord (Ntionl Instruments #NI PCI-622) nd ontrolled y ustom-mde LView (Ntionl Instruments) nd Python progrms. Neurophysiologil signls were quired ontinuously t 32.552 k on 28-hnnel DigiLynx system (Neurlynx). The widend signls were digitlly high-pss filtered (.8 5 k) offline for spike detetion or low-pss filtered ( 5 ) nd down smpled to.252 k for LFPs. Spike sorting ws performed semi-utomtilly, using KlustKwik (http://osiris.rutgers.edu/frontmid/indexmid.html), followed y mnul djustment of the lusters 52. Additionl dt nlysis ws done using ustom Mtl routines. Single neuron firing field implementtion. The length of the elt ws divided into pixels. For eh pixel, the numer of spikes dishrged in the pixel ws divided y the time the niml spent in the pixel to generte vetor of firing NATURE NEUROSCIENCE doi:.38/nn.377
22 Nture Ameri, In. All rights reserved. rtes. The firing rte vetor ws smoothed y onvolving it with Gussin funtion (6 pixels ( m) hlf-width). Signifine of hnges in firing rte for single neuron. The rte differene ( f) etween ontrol nd light trils ws lulted for eh pixel. The sme mesure ws repeted, times for different shuffles of ontrol nd light trils ( f s: ). The P vlue in eh pixel ws determined from the rnk of f in the f s: distriution. For given pixel, if less thn 5 vlues from f s: were lower (higher) thn f, tht is, the P vlue ws less thn.5 (5/,), then the firing rte derese (inrese) ws deemed signifint. Burst proility nd urst index. Bursts were defined s groups of spikes with interspike intervls <9 ms. Threshold vlues of 6 nd 2 ms were lso tested. In first nlysis, the numer of ursts N B nd single spikes N S ourring in the stimulted portion of the elt ws ounted for eh ells (ontrol nd light trils seprtely). The urst proility ws lulted using the following eqution: Burst proility = N B /(N B + N S ). The hnge in urst proility (Supplementry Fig. 9) ws the differene etween the urst proility during light nd ontrol trils. In seond nlysis, the numer of ursts omposed of two, three, four nd more thn four spikes were ounted seprtely. Either the perent hnge (N light N ontrol )/N ontrol (Fig. 5 nd Supplementry Fig. 9) or the proility (Supplementry Fig. 9d) of eh urst length ws mesured. We lso defined urst index tht ptured the propensity of neurons to dishrge in ursts. This mesure ws used for ell-type identifition (see elow). The mplitude of the urst ws estimted from the spike uto-orrelogrm (-ms in size) y sutrting the men vlue etween 4 nd 5 ms (seline) from the pek mesured etween nd ms. Positive urst mplitudes were normlized to the pek nd negtive mplitudes were normlized to the seline to otin indexes rnging from to. Binning of ple fields. The ontrol trils (no-light ondition) were used to determine five ins of the ple field. For eh ple field, we deteted the pixel of the eginning, 4% ross-points, 8% ross-points nd end (Fig. 3). The estimtion of eginning nd end of ple fields ws not lwys equivol s result of the osionl presene of multiple ple fields. To orret for multiple ple fields, we first smoothed the firing field vetor y verging over moving windows of size equl to the ple field hlf-width t 66% of pek mplitude. The first lol (smoothed) minim on oth sides of the ple field pek were deteted on the smoothed pixels. Next, the rel minim were identified on the unsmoothed pixels, in the viinity of the smoothed minim. The ins were defined s the intervls etween these onseutively deteted pixel positions, s shown in Figure 3. Only ins in the stimulted segment of the elt were used in the nlysis. When in overlpped prtilly with the stimulted segment, the in ws trunted to the overlpping frtion only. The sme in positions were used for the light trils. Lol field power spetrum. The LFP epohs during the stimulted segment of the elt were seleted, nd the power spetrum over to ws omputed for eh tril. Power spetr of ontrol nd light trils were verged seprtely. The signifine of differenes etween ontrol nd light trils ws estimted the sme wy s for the firing fields. The differene ( pwr) etween ontrol nd light trils ws lulted for eh frequeny. The sme mesure ws repeted, times for different shuffles of ontrol nd light trils ( pwr s: ). The P vlue for eh frequeny ws determined from the rnk of pwr in the pwr s: distriution. For given frequeny, if less thn 5 vlues from pwr s: were lower (higher) thn pwr, the P vlue ws less thn.5 (5/,) nd the power derese (inrese) ws deemed signifint y the light stimultion. None of the eight reording sessions in the -Cre mie showed signifint power differene in the thet nd (5 5 ), nd only one of the nine sessions in -Cre mie showed signifint inrese of thet on few hnnels. A likely explntion for the lk of differene is volume ondution of thet signl from the surrounded, non-ffeted, tissue 53. Thet phse of spikes. The LFP ws nd-pss filtered in the thet rnge (5 ). A vetor of instntneous phse ws derived using the Hilert trnsform. The thet phse of eh spike ws interpolted from the vetor of instntneous phse. Thet phse histogrms. The thet yle ws divided into 6 equl phse ins. The phse histogrm of the spikes ws omputed for eh individul neuron. Eh phse histogrm ws smoothed y onvolving Gussin funtion (four ins hlf width). For the popultion verge in Figure 4, histogrms of eh neuron, for oth ontrol nd light trils, were first normlized to the pek vlue of the ontrol trils. For pyrmidl neurons, the thet phse histogrms were omputed in eh in of ple field. For other ell types, ll spikes during the stimulted segment of the elt were used. Men thet phse. For the determintion of phse preession of spikes 4, we exmined the men phses of ple field ins for two resons. First, light stimultion ws delivered only in the middle segment of the elt (57 to 3 m) where the running speed ws most onstnt. Although spikes of mny neurons ould hve een ffeted y the light stimultion in this segment, severl of them hd only prt of their ple fields represented mking diffiult the use of stndrd phse slope methods 4,6,7,22,23,35. Seond, our preliminry nlysis of phse preession showed tht despite the relile rte-position reltionship, spike phse-position reltionship ws muh noisier in the mouse ompred to rt (onfirming previous oservtions y Mynk Meht, personl ommunition). The inning method provided lerer phse vlues ompred with the trditionl disply of ll spikes (Fig. e). To lulte the men thet phse of single neurons, eh spike ws expressed s unitry vetor in polr oordintes with diretion its thet phse, nd ll spike vetors were summed. The men thet phse of ell ws the ngle of the resultnt vetor sum. In mthemtil terms (eqution 2.9 from ref. 54), Let os i nd i with s the thet phse nd i s the spike index. If nd If If nd sin i rtn( / ) rtn( / ) rtn( / ) 2 where is the men thet phse. After the lultion of eh neuron s men thet phse, either the popultion distriution (Fig. 3) or the popultion men (Fig. 3 nd 7) were omputed. The sme method s desried ove ws used to ompute the popultion men, using eh neuron s men thet phse s unitry vetors. The irulr stndrd errors were lulted (eqution 4.2 of ref. 54). 2 Letd ( q)/( 2p ) where q os( 2( i ))/ n i nd P os( i )/ n i then the irulr stndrd error (SE) is SE d n () (2) (3) To test the signifine of thet phse hnges used y the light stimultion, we tested the hypothesis of ommon mens etween the ontrol nd light trils phses (equtions 8.4 nd 5.3.4 from ref. 54). Neuron type identifition nd hrteriztion. Severl prmeters were mesured to identify nd hrterize the different neuronl types (Supplementry Figs. 5 8). nd interneurons ould e identified y the roust suppression of their firing rtes y the light stimultion (Supplementry Fig. 5,). The physiologil hrteristis tht were found to most effetively differentite interneurons from other ell types were the ell s refrtory period nd ursting property (Supplementry Fig. 5,d). Three reltively distint lusters were tenttively ssigned to pyrmidl ells, interneurons nd non-usting neurons (Supplementry Fig. 5d). The interneuron group inluded the light-suppressed interneurons nd other unffeted ells ( like), inluding presumly interneurons tht were not infeted y the AAV virus. The seprtion of these three groups ws further supported y informtion otined from monosynpti intertions etween neuron pirs 55. Neuron pirs with ross-orrelogrms showing shrp nd signifintly lrge ins in monosynpti lteny (<3 ms) were identified 56. Referene (puttive presynpti) neurons were doi:.38/nn.377 NATURE NEUROSCIENCE