Clarke s column neurons as the focus of a corticospinal corollary circuit

Transcription

1 lrke s column neurons s the focus of corticospinl corollry circuit Adm W Hntmn & Thoms M Jessell Proprioceptive sensory signls inform the NS of the consequences of motor cts, ut effective motor plnning involves internl neurl systems cple of nticipting ctul sensory feedck. Just where nd how predictive systems exert their influence remins poorly understood. We explored the possiility tht spinocereellr neurons tht convey proprioceptive sensory informtion lso integrte informtion from corticl commnd systems. Anlysis of the circuitry nd physiology of identified dorsl spinocereellr trct neurons in mouse spinl cord reveled distinct popultions of lrke s column neurons tht received direct excittory nd/or indirect inhiitory inputs from descending corticospinl xons. The convergence of these descending inhiitory nd excittory inputs to lrke s column neurons estlished locl spinl circuits with the cpcity to mrk or modulte incoming proprioceptive input. Together, our genetic, ntomicl nd physiologicl results indicte tht lrke s column spinocereellr neurons nuclete locl spinl corollry circuits tht re relevnt to motor plnning nd evlution. The coordintion of movement depends on proprioceptive sensory signls tht convey the stte of muscle ctivity nd ody position to motor commnd centers in the NS 1. Proprioceptive fferent input from the lims is relyed to corticl motor centers long diverse scending rely pthwys, the most prominent of which engge the spinocereellr system 2 4. In mmmls, there re dozen or so distinct clsses of spinocereellr neurons 5, ech ssigned to discrete spects of somtosensory nd motor processing 4,6. Proprioceptive sensory signling from the hindlim is relyed primrily y set of dorsl spinocereellr () trct neurons tht occupy discrete thorcic nd lumr nucleus known s lrke s column 7,8. Although the sensory rely properties of lrke s column hve long een pprecited 9, the possiility tht this set of spinocereellr neurons hs dditionl integrtive functions in spinl sensory processing hs not een explored in detil. Proprioceptive sensory informtion of peripherl origin provides one crucil conduit to the motor system, ut there is now emerging evidence tht motor systems hve the dditionl cpcity to generte internl predictions of the sensory consequences of motor cts 10,11. Predictive signls hve een suggested to provide corticl motor centers with rpid updtes out plnned ctions, helping to ovite delys incurred when proprioceptive feedck is ctivted from the periphery 12,13. orticlly derived predictions my lso e used to negte the sensory consequences of self-generted movements, finetuning the motor system to unnticipted sensory events 14. Such predictions re thought to e generted y ctivting corollry pthwys tht mp directly onto sensory processing strems 10,11. The cereellum nd cererl cortex hve trditionlly een invoked s sites for the convergence of corticl corollry nd sensory feedck pthwys involved in internl motor predictions 15,16, lthough the locl circuitry underlying convergence in these regions is oscure. The likely existence of distriuted sites for suprspinl convergence of motor corollry nd sensory feedck pthwys prompted us to consider whether corollry convergence might lso occur t erlier steps in the proprioceptive processing pthwy, perhps even in the spinl cord. Prior studies of cutneous sensory coding re consistent with the notion tht corollry pthwys intersect t erly steps of spinl sensory processing 17. In ddition, physiologicl studies in the ct found tht trct neurons cn e ctivted y stimultion of descending corticl trcts 18,19. These studies reveled tht stimultion of descending corticospinl fiers cn result in trnsient excittion followed y prolonged inhiition of trct neurons 18,19, lthough the prticulr suclss of dorsl spinocereellr neurons ws not determined. The issue of whether nd how the output of lrke s column neurons is shped y corticl input therefore remins uncler. To explore the possiility of spinl focus for convergent corticl corollry nd proprioceptive sensory feedck pthwys, we used comintion of moleculr genetic, ntomicl nd physiologicl pproches to mp the circuitry nd integrtive functions of identified lrke s column neurons in the mouse spinl cord. We found tht rely neurons in lrke s column integrte corticl excittory nd inhiitory inputs in mnner cple of mimicking nd suppressing proprioceptive feedck. ollectively, our dt indicte tht lrke s column neurons nuclete spinl corollry dischrge circuit of potentil relevnce to motor plnning nd its evlution. RESULTS neurons mrked ntomiclly nd geneticlly To mp the orgniztion of lrke s column neurons in mice, we injected fluorogold or choler toxin B suunit (TB) into the cereellum of postntl dy 5 7 (P5 7) mice nd monitored the position of retrogrdely leled neurons t cervicl, thorcic nd lumr levels of Howrd Hughes Medicl Institute, Kvli Institute for Brin Science, Deprtments of Neuroscience nd Biochemistry nd Moleculr Biophysics, olumi University, New York, New York, USA. orrespondence should e ddressed to T.M.J. (tmj1@columi.edu). Received 25 June; ccepted 13 August; pulished online 12 Septemer 2010; doi: /nn.2637 nture NEUROSIENE VOLUME 13 NUMBER 10 OTOBER

2 ereellum:fg Gdnf mrna c Gdnf::LcZ FG LcZ d Gdnf::LcZ T L e f g h ereellum:fg:biocytin ereellum:fg:biocytin Gdnf::reERT2, Tu::IsI-mGFP Gdnf::reERT2, Tu::IsI-mGFP R M L R M L R i R III II I vglut1 VIII Gdnf::reERT2, Tu::IsI-mGFP j l M L ortex ereellum Figure 1 Antomic nd genetic chrcteriztion of postntl mouse neurons. () Loction of thorcic spinl neurons retrogrdely leled from cereellum with fluorogold (FG). lrke s column is mrked y the rrowhed. Inset, high-mgnifiction of fluorogold in lrke s column. () Gdnf mrna expression in thorcic spinl cord (inset, Gdnf expression in lrke s column of inset in ). A smll group of retrogrdely leled, Gdnf-expressing neurons were lso found in the deep dorsl horn. (c) β-glctosidse ctivity in Gdnf<LcZ spinl cord (inset, coleling of β-glctosidse nd retrogrde fluorogold lel from the cereellum in lrke s column). (d) Top, Gdnf neurons long the rostrl-cudl xis (, cervicl; T, thorcic; L, cudl lumr). Lower, summry of neuron distriution. We found high density of neurons in cudl thorcic levels, low density in rostrl thorcic nd neurons were sent in cervicl, cudl lumr nd scrl (dt not shown) levels. (e,f) Morphology of iocytin-filled, fluorogold-positive neuron in trnsverse section (inset, low-mgnifiction imge, e) nd in sgittl section (rrow indictes xon, f). Scle rs represent 50 μm (f) nd 25 μm (e). (g,h) Morphology of Gdnf<reERT2, Tu<lsl-mGFP positive neurons (rrow indictes xon) in trnsverse section (g) nd in sgittl section (rrow indictes xon; inset, horizontl spinl section of dorsl lterl funiculus, h). Scle rs represent 25 μm. (i) Gdnf<reERT2, Tu<lsl mgfp positive xons in sgittl section of cereellum. (j) Mgnified imge of Gdnf<reERT2, Tu<lsl mgfp positive xons of loule VIII of i. (k) Gdnf<reERT2, Tu<lsl mgfp positive xons in coronl section of cereellum. Scle rs represent 50μm. (l) vglut1 expression in Gdnf<reERT2, Tu<lsl mgfp positive xons. Scle r represents 1 μm. (m) Summry of ntomy (DRG, dorsl root gngli). k m DRG the spinl cord (Fig. 1). At cudl thorcic nd rostrl lumr levels, retrogrdely leled neurons were concentrted in the medil region of dorsl horn lmin VII (Fig. 1), the loction of lrke s column in other mmmls 20. In the trnsverse plne, retrogrdely leled neurons were rryed in n nnulr pttern tht surrounded neuronsprse neuropil core (Fig. 1 nd Supplementry Fig. 1). To define the dendritic orgniztion of this set of neurons, we injected iocytin into individul fluorogold-leled neurons. Regrdless of their clock position in the nnulus, the iocytin-filled dendrites of neurons were oriented towrd the core of lrke s column (Fig. 1e,f). To mp the trjectory nd circuitry of the entire popultion of lrke s column neurons, we serched for genetic mrkers tht delinete this suset of spinocereellr projection neurons. We found tht lrke s column neurons could e defined y selective expression of the glil derived neurotrophic fctor (Gdnf) gene 21 (Fig. 1 d). Endogenous Gdnf mrna, s well LcZ expression in Gdnf<LcZ trnsgenic mouse line 22, ws detected in identified lrke s column neurons, mrked retrogrdely y cereellr fluorogold or TB injection (Fig. 1 c). To trce the projections of Gdnf-expressing neurons, we nlyzed green fluorescent protein (GFP) expression in mice generted y Gdnf<creERT2 trnsgenic driver nd myristoylted (m)gfp driven y Tu with loxp-flnked STOP codon (Tu<lsl-mGFP) 23 reporter crosses. After tmoxifen exposure, the xons of GFP-leled lrke s column neurons were detected in the ipsilterl dorsolterl funiculus (Fig. 1g,h) nd inferior cereellr peduncle nd terminted s mossy fiers in the grnulr lyer of the cereellum (Fig. 1i m). These GFP-leled xons were segregted into prsgittl stripes in cereellr loules I, II, III 1234 VOLUME 13 NUMBER 10 OTOBER 2010 nture NEUROSIENE

3 Pvl::GFP Gdnf::LcZ Figure 2 Antomy of proprioceptive inputs to neurons. () Lowmgnifiction imge of thorcic spinl cord including Pvl<GFP expressing proprioceptor terminls nd Gdnf<LcZ neurons. Scle r represents 50 μm. () Pvl<GFP expressing proprioceptor terminls on n individul Gdnf<LcZ neuron (top right, high-mgnifiction imge of rcketed re; ottom right, Pvl nd vglut1 expression of rcketed re). Scle r represents 5 μm. Pvl::GFP Gdnf::LcZ vglut1 nd VIII (Fig. 1k), consistent with the known orgniztion of projections 24. As with iocytin injection, GFP-leled dendrites of Gdnf-expressing neurons were confined to the core of lrke s column (Fig. 1g,h nd Supplementry Fig. 1). Thus, Gdnf expression defines lrke s column neurons, distinguishing them from other dorsl nd ventrl spinocereellr projection neurons. neurons integrte proprioceptive nd descending input We next used ntomicl nd physiologicl methods to determine whether lrke s column neurons receive input from corticl descending s well s proprioceptive sensory pthwys. A prvlumin (Pvl)<GFP cteril rtificil chromosome (BA) trnsgenic line 25 ws used to mp the intrspinl projections of proprioceptive xons 26. Figure 3 Physiology of proprioceptive nd corticl inputs to neurons. () Ten dorsl root evoked (lue rrow) excittory postsynptic currents (EPSs) in neuron (inset, expnded time scle round response onset). () Single dorsl root evoked excittory postsynptic potentil (EPSP) in neuron; dorsl root evoked ction potentils were detected in 26 of 31 neurons. A schemtic of sensory (S, lue) input to neurons is shown. (c) Ten dorsl column evoked (green rrow) EPSs in neuron (inset, expnded time scle round response onset). (d) Single dorsl column evoked EPSP in neuron. Dorsl column evoked ction potentils were generted in 14 of 20 neurons. A schemtic of corticl (, green) input to neurons is shown. (e) Hemisected spinl cord preprtion. Lumr (L4 L6) dorsl roots (lue) were stimulted with suction electrode (SE) nd cervicl dorsl column (green) ws stimulted with concentric ipolr electrode c e BE ervicl Thorcic At cudl thorcic nd rostrl lumr levels of the spinl cord, proprioceptive xonl projections were concentrted in the vicinity of lrke s column neurons (Fig. 2). We found tht ll fluorogold-, TBor Gdnf-leled neurons were contcted y Pvl-positive xons leled in Pvl<GFP mice nd tht tht ll Pvl-positive outons on neurons coexpressed vesiculr glutmte trnsporter 1 (vglut1), confirming their sensory synptic sttus (Fig. 2). The vst mjority of GFP nd vglut1 doule-positive outon contcts were locted on the dendrites of neurons, in the core of lrke s column (Fig. 2). In ddition, more thn 95% of sensory contcts with lrke s column neurons, defined y trnsgnglionic trnsport nd ccumultion of TB fter hindlim injection, coexpressed Pvl nd vglut1 (Supplementry Figs. 2 5), indicting 27 tht lmost ll direct sensory inputs to neurons derive from proprioceptive fferents. To determine the functionl consequences of proprioceptive input, we recorded from identified neurons in hemisected P8 15 mouse spinl cord preprtions (Fig. 3). Ptch-clmp recordings from retrogrde fluorogold-leled neurons t the T6 to L2 levels reveled tht 56% (47 of 84) of neurons exhiited excittory responses to low-threshold stimultion of L4 or L5 dorsl roots (Fig. 3). Excittory synptic potentils recorded from individul neurons hd short ltencies nd low vrition in onset time (coefficient of vrition < 0.05; Fig. 3), n indiction of their monosynptic origin. Moreover, dorsl root stimultion elicited ction potentils in 84% of responsive neurons (Fig. 3). As our ntomicl results indicted tht ll neurons re contcted y proprioceptive sensory terminls, we presume tht the nonresponsive popultion do in fct receive sensory input from fferent xons tht enter the spinl cord through other, unstimulted, lumoscrl roots. (BE). (f) EPSPs of neurons recorded fter dorsl root nd dorsl column stimultion (inset, reduced interstimulus intervl etween dorsl root nd dorsl column stimultion). A schemtic of convergence of corticl nd sensory inputs on neurons is shown. 100 pa 10 ms 50 pa 0.5 ms SE Lumr 65 d 65 f ms 20 ms S S nture NEUROSIENE VOLUME 13 NUMBER 10 OTOBER

4 Emx1::GFP Gdnf::LcZ Figure 4 Antomy of corticl inputs to neurons. () Lowmgnifiction imge of thorcic spinl cord including Emx1<GFP expressing corticospinl terminls nd Gdnf<LcZ neurons. Scle r represents 50 μm. () Emx1<GFP corticospinl terminls on n individul Gdnf<LcZ neuron (top right, high-mgnifiction imge of rcketed re; ottom right, GFP nd vglut1 expression of rcketed re). Scle r represents 5 μm. (c) Orgniztion of Emx1<GFP corticospinl terminls with respect to loction of Gdnf<LcZ neurons (top right, mgnified imge of * rcketed re; ottom right, mgnified imge of ** rcketed re). Scle r represents 25 μm. c Emx1::GFP Gdnf::LcZ vglut1 Emx1::GFP Gdnf::LcZ To determine whether neurons lso receive input from corticospinl xons, we first nlyzed the spinl cord of Emx1<GFP BA trnsgenic mice in which xons of corticl origin re selectively mrked 28 (Fig. 4). At thorcic nd rostrl lumr spinl levels, GFPpositive xons were restricted to the ventrl-most spect of the dorsl funiculus (Fig. 4) nd gve rise to collterl projections tht terminted in the vicinity of lrke s column (Fig. 4). In Emx1<GFP mice, GFP-positive terminls were oserved in contct with the cell odies of out hlf of ll leled neurons nd mny of these GFP-positive terminls expressed protein kinse γ (PKγ), mrker selective for corticospinl xons 29 (Fig. 4 nd Supplementry Fig. 1). In the nnulus, we detected mrked symmetry in the pttern of corticospinl input; neurons tht occupied dorsomedil position in the nnulus received 12-fold greter density of GFP nd vglut1 doule-positive terminl contcts thn did neurons in ventrolterl position, which were contcted y few if ny corticospinl xons (Fig. 4,c nd Supplementry Fig. 1). These oservtions * * ** ** provide ntomicl evidence tht suset of lrke s column neurons in receipt of proprioceptive sensory input re lso contcted y the terminls of corticospinl neurons. To ssess the function of these corticospinl inputs, we nlyzed the response of fluorogold-leled lrke s column neurons to focl stimultion of corticospinl xons present in the ventrl-most spect of the dorsl columns, t cervicl levels. A series of control ntomicl nd physiologicl experiments indicted tht neuronl responses to dorsl column stimultion reflect input from corticospinl rther thn sensory or spinl neurons (see Supplementry Results nd Supplementry Figs. 2 5). Dorsl column stimultion elicited monosynptic excittory synptic responses in ~65% (54 of 84) of neurons (Fig. 3c). Dorsl column evoked excittory input ws sufficient to trigger ction potentils in ~70% of this set of neurons (Fig. 3d), n indiction of the efficcy of corticospinl excittory drive to these neurons. Moreover, dorsl column evoked excittory synptic responses in neurons were locked y comined exposure to AMPA nd NMDA receptor ntgonists (6-cyno-7-nitroquinoxline-2,3-dione (NQX) nd d( )-2-mino- 5-phosphonovleric cid, respectively, dt not shown), estlishing their glutmtergic sttus. We lso exmined directly whether, s implied y the summed incidence of corticospinl nd proprioceptive synptic potentils, individul lrke s column neurons receive convergent input from these two neuronl clsses. Anlysis of responses reveled tht synptic responses to c Slc65 Gd1 Gd2 Gd2::GFP PKγ vglut1 d Emx1::re, Tu::IsI-mGFP Gd1 ereellum:fg:biocytin Figure 5 Antomy of corticlly evoked inhiition of neurons. () In situ hyridiztion of Slc65, Gd1 nd Gd2 proes in thorcic spinl cord (inset, mgnified imge of oxed re). () Distriution of Emx1<cre, Tu<lsl mgfp positive corticospinl terminls 39 in thorcic spinl cord. (c) Apposition of PKγ nd vglut1 doule-positive corticospinl terminls nd Gd2<GFP neurons (inset, mgnified imge of oxed re). Scle r represents 5 μm. (d) Gd1-positive inhiitory inputs on iocytin-filled neuron. Scle r represents 50 μm. (e,f) GABAergic (Gd1, e) nd glycinergic (GlyT2 nd VIAAT, f) inhiitory inputs on Gdnf<reERT2, Tu<lsl mgfp positive neurons. (g) Gd2-positive inhiitory terminls on Pvl<GFP proprioceptive terminls in lrke s column. Scle rs represent 1 μm. Gd1 Gdnf::reERT2, Tu::IsI-mGFP GlyT2 Gdnf::reERT2, Tu::IsI-mGFP VIAAT e f g Pvl::GFP Gdnf::LcZ Gd VOLUME 13 NUMBER 10 OTOBER 2010 nture neurosiene

5 Figure 6 Physiology of corticlly evoked inhiition of neurons. () Left, schemtic of corticl excittion (green) of n inhiitory input (gry) to neurons. Right, 10 dorsl column evoked IPSPs in neuron. () Phrmcologicl ssessment of the effects of icuculline (ic, GABA A -receptor ntgonist, 8 μm) nd strychnine (str, glycine-receptor ntgonist, 10 μm) on dorsl column evoked IPSPs in neuron (ech trce is verge of ten trils). (c) Phrmcologicl ssessment of the effects of NQX (AMPA-receptor ntgonist, 10 μm) on dorsl column evoked IPSs of neuron (ech trce is verge of ten trils). (d) Left, schemtic of corticl excittion (green) nd inhiition (gry) of neuron. Right, ten dorsl column evoked EPSs nd IPSs in neuron (inset, reversl potentil of inhiitory component). dul dorsl column nd dorsl root stimultion were detected in ~55% of smpled neurons, independent of input order or interstimulus intervl (Fig. 3f). Together, these results indicte tht suset of neurons, presumly those in dorsomedil nnulr position, integrte proprioceptive sensory nd corticospinl inputs. orticlly evoked inhiitory input to neurons The results of our ntomicl nd physiologicl nlyses lso provide evidence for corticlly evoked inhiitory inputs to lrke s column neurons (Fig. 5). Fluorogold-leled, nd Gdnf-expressing, neurons were contcted y glutmte decroxylse 67 (GAD67)-positive GABAergic (Fig. 5d,e) nd glycine trnsporter 2 (GlyT2) nd vesiculr inhiitory mino cid trnsporter (VIAAT) doule-positive glycinergic outons (Fig. 5f). Inhiitory synptic inputs were detected on neurons t ll nnulr positions. The source of these inhiitory outons remins uncler, ut we did oserve high density of neurons expressing Gd2 (GAD65, GABAergic), Gd1 (GAD67, GABAergic) nd Slc65 (GlyT2, glycinergic) in the regions of the intermedite spinl cord surrounding lrke s column (Fig. 5). Furthermore, we found tht mny peri- GABAergic inhiitory neurons, mrked y GFP expression in Gd2<GFP trnsgenic mice 30, received dense innervtion from vglut1 nd PKγ doule-positive corticospinl terminls (Fig. 5,c). onsistent with these ntomicl findings, we found tht dorsl column stimultion evoked long-lsting inhiitory postsynptic potentils (IPSP; Fig. 6) in mny fluorogold-leled neurons (decy c d ontrol Bic + Str Wsh ontrol NQX Wsh pa 25 ms 5 50 ms 2 50 ms 10 pa 25 ms 50 pa 50 ms time constnt 92 ± 39 ms; men ± s.d.; n = 8 neurons; Fig. 6). Overll, 62% of lrke s column neurons exhiited inhiitory synptic responses to dorsl column stimultion, with out hlf of these exhiiting inhiitory postsynptic currents (IPSs) lone nd hlf showing monosynptic excittory input efore inhiition (Fig. 6,d). IPSPs recorded from neurons were locked y exposure to oth icuculline nd strychnine (Fig. 6), ut were only prtilly inhiited fter ppliction of ech ntgonist lone, indicting tht oth GABAergic nd glycinergic inhiitory inputs were involved. In ddition, we found tht dorsl column stimultion evoked IPSPs were locked y the AMPA receptor ntgonist NQX (Fig. 6c), indicting tht the inhiitory interneurons tht ct on neurons re themselves ctivted y corticospinl input. S S c S 300 ms 300 ms ms ms ms 63 Bic+ Str 300 ms ms Figure 7 orticl inhiition of sensory-evoked responses in neurons. Three modes of dorsl column inhiition of dorsl root responses in neurons. () The first trce is dorsl column evoked IPSP in neuron (inset, longer time scle to show durtion of inhiition). The second trce is dorsl root evoked EPSPs in neuron. The third trce is dorsl root-evoked EPSP preceded y dorsl column evoked IPSP. The finl trce is dorsl root input preceded y dorsl column input t different time intervls. () The first trce is dorsl column evoked EPSP nd IPSP in neuron (inset, longer time scle to show durtion of inhiition). The second trce is dorsl root evoked EPSP in neuron. The third trce is dorsl root evoked EPSP preceded y dorsl column evoked EPSP nd IPSP (inset, longer interstimulus intervl etween dorsl root nd dorsl column stimultion). (c) The first trce is dorsl column evoked EPSP in neuron. The second trce is middle, dorsl root evoked EPSP in neuron. The third trce is dorsl root evoked EPSP preceded y dorsl column evoked EPSP (inset, dorsl root input preceded y dorsl column input in the presence of icuculline nd strychnine). All trces represent single trils. No IPSPs were detected in this cell, even t depolrized holding potentils. These findings re suggestive of presynptic inhiitory mechnism, which could lso occur in neurons exhiiting corticlly evoked postsynptic inhiition. nture NEUROSIENE VOLUME 13 NUMBER 10 OTOBER

6 The microcircuitry of lrke s column led us to exmine the influence of corticlly ctivted inhiitory inputs on neuronl responses to sensory stimultion. In the popultion of neurons tht received exclusively inhiitory input, dorsl root evoked ction potentils could e suppressed for ~100 ms following single conditioning stimulus pplied to the dorsl columns (7 of 7 neurons exmined; Fig. 7). Similrly, for the set of neurons tht exhiited dul excittory nd inhiitory synptic responses, single stimulus pplied to the dorsl column impired the ility of sensory stimultion to elicit ction potentils (8 of 8 neurons; Fig. 7). Thus, descending corticospinl xons cn lso suppress neuronl responses to proprioceptive sensory input through the recruitment of locl inhiitory interneurons. orticospinl nd proprioceptive sensory inputs to neurons exhiited n dditionl inhiitory interction. In smll suset of neurons tht received direct sensory nd corticl excittory inputs in the sence of ccompnying IPSPs, dorsl column stimultion reduced the proility of dorsl root evoked ction potentils y ~80% (Fig. 7c). This dorsl column evoked suppression of sensory responses ws locked y exposure to icuculline nd strychnine, despite the lck of detectle IPSPs in these neurons (Fig. 7c). This IPS-independent inhiitory influence ws evident for up to 1,000 ms, durtion tht ws consistently longer thn tht evoked y postsynptic inhiition of neurons. One potentil sustrte for this sustined inhiitory influence is the corticl recruitment of interneurons tht exert presynptic inhiitory control of proprioceptive sensory fferent input 7,17,31. In support of this possiility, we found tht the GFP nd vglut1 doule-positive proprioceptive sensory terminls detected on neurons in Pvl<GFP mice were frequently contcted y GABAergic outons tht coexpressed Gd2 nd Gd1 (Fig. 5g), cytochemicl hllmrk of presynptic inhiitory terminls on proprioceptive sensory fferents in the ventrl region of mouse spinl cord 32. Thus, ctivtion of corticospinl pthwys ppers to elicit oth pre- nd postsynptic inhiition of proprioceptive input to lrke s column neurons. DISUSSION Our findings suggest tht lrke s column neurons represent focl trget for the convergence of descending corticl nd sensory fferent pthwys, nucleting spinl microcircuits with the potentil to predict nd modulte proprioceptive feedck signls. Prior physiologicl studies hve provided evidence for oth rpid excittion nd prolonged inhiition of neurons y descending corticospinl projections 18,19. Our findings re consistent with these erlier in vivo studies, lthough it is worth emphsizing tht these nlyses did not estlish the direct nture of corticlly evoked excittion. Nor were lrke s column neurons distinguished from other clsses of neurons. In fct, the most complete physiologicl description of neuronl ctivity nd input chrcteristics ws otined t cudl lumr segmentl levels, region devoid of lrke s column neurons 18,19. At these cudl levels, neurons re known to receive prominent direct input from cutneous sensory fferents 6, which contrsts with the nerly pure proprioceptive sensory origin of direct input to lrke s column neurons. Thus, our results provide cler evidence for corticl regultion of the ctivity nd output of identified lrke s column neurons. When tken together with previous findings 18,19, our results suggest common strtegy for descending corticl control of distinct suclsses of neurons. Our findings lso suggest tht there re distinct clsses of lrke s column neurons. Neurons positioned in the dorsomedil sector of the lrke s column nnulus cn e distinguished from their ventrolterl counterprts y synptic outon contcts from corticospinl xons. We presume tht these neurons represent the set of physiologiclly defined neurons tht exhiit excittory responses to dorsl column stimultion. In ddition, neurons throughout the nnulus receive dense input from GABAergic nd glycinergic inhiitory interneurons, providing plusile ntomicl sustrte for the prominent inhiitory responses elicited in most neurons y dorsl column stimultion. Nevertheless, some neurons exhiited exclusively excittory responses to dorsl column stimultion, rising the possiility tht the inhiitory outons tht contct some, presumly dorsomedil, neurons re derived from interneurons tht lck corticospinl input. The existence of three distinct synptic rrngements on neurons suggests tht multiple chnnels for proprioceptive processing operte in lrke s column (Supplementry Fig. 6). Our ntomicl nlysis of excittory synptic terminls on lrke s column neurons reveled vst predominnce of vglut1- over vglut2-leled outons nd Pvl ws expressed y essentilly ll of the vglut1-positive outons. This synptic phenotype implies tht virtully ll sensory inputs to this set of neurons re derived from proprioceptors, n unnticipted finding when considered from the perspective of physiologicl studies rguing tht neurons serve s sensory monitors of lim position through the convergence of diverse cutneous nd proprioceptive sensory signls 4. One wy of reconciling our findings with prior oservtions is to invoke the ide tht the influence of cutneous sensory input on the integrtive sensory properties of lrke s column neurons is chieved indirectly vi intermediry interneuronl pthwys. The functionl circuitry of neurons uncovered through our physiologicl nlysis of dorsl column inputs suggests tht lrke s column serves n integrtive role well eyond tht of simple sensory rely nucleus. The existence of strong excittory corticl inputs to suset of neurons provides potentil intrspinl pthwy for the trnsfer of descending corticl commnds onto sensory rely system destined for the cereellum (Supplementry Fig. 6). These corticlly derived signls pper to e well suited to deliver predictions of the sensory consequences of motor cts, nticipting peripherlly derived sensory feedck. The corticlly evoked inhiitory responses detected in neurons typiclly persist for 100 ms or more, period tht spns the temporl dely incurred through peripherl routing of proprioceptive sensory feedck 33. These findings rgue in fvor of n independent function for descending corticl commnds in suppressing or modulting the effects of ctivtion of neurons y proprioceptive sensory feedck (Supplementry Fig. 6). The origin of dorsl column inputs to lrke s column neurons, whether from motor nd/or somtosensory corticl res, remins unknown. The functions invoked for lrke s column neurons in the regultion of proprioceptive sensory processing shre certin fetures with the integrtive properties of ventrl spinocereellr (vs) trct pthwys. Physiologicl studies hve shown tht vs trct neurons receive inputs from sensory, locl nd descending xons tht mirror those converging on spinl motor neurons 34,35. One mjor role of the vs trct therefore ppers to e to rely corollry copy of inputs to spinl motor neurons directly to suprspinl processing centers 36. Our findings imply tht neurons serve similr corollry function, with the notle difference eing tht informtion conveyed vi the lrke s column pthwy reports on nticipted proprioceptive input, rther thn on imminent motor output. In turn, these considertions pose the downstrem prolem of how the cereellum integrtes inputs from multiple spinocereellr signling strems. In recent genetic trcing studies (dt not shown), we found tht the terminls of nd vs xons converge on the sme cereellr foli nd were frequently found in proximity to the sme grnule neuron, suggesting tht cereellr processing involves the convergence of spinl inputs onto common grnule cell trget VOLUME 13 NUMBER 10 OTOBER 2010 nture neurosiene

7 Notly, the intrspinl circuitry for corticlly medited inhiition of signling exhiits n orgniztion tht conforms to tht of n insect corollry dischrge circuit employed in the cncelltion of selfgenerted uditory stimuli 37. Mmmlin nd rthropod circuits oth rely on the ctivtion of interneurons tht exert presynptic inhiition of sensory fferent input nd postsynptic inhiition of primry centrl rely neuron. The dditionl predictive role of neurons tht we found my e more kin to the corollry ctivities oserved the sensory processing centers of higher mmmls 12,38. In the mouse, the moleculr delinetion of neurons opens the wy for future genetic mnipultion of neuronl elements in this spinl corollry circuit. Trditionlly, corollry dischrge circuits involved in motor plnning hve een ssigned to pontocereellr nd intrcorticl pthwys 15,16, rising the further question of the merits of constructing intrspinl circuits with similr design fetures (Supplementry Fig. 6). A spinlly focused corollry circuit will inevitly incur slightly greter temporl delys thn its suprspinl counterprts. But, y wy of compenstion, it ffords descending corticl systems direct ccess to selective sensory chnnel tht, in principle, permits erly nd effective nticiption or cncelltion of the proprioceptive consequences of movement. Methods Methods nd ny ssocited references re ville in the online version of the pper t Note: Supplementry informtion is ville on the Nture Neuroscience wesite. Acknowledgments We thnk B. Hn nd K. Mio for technicl ssistnce, J. de Nooij nd G. Sürmeli for help in retrogrde leling of proprioceptive fferents, J. Kirklnd nd M. Mendelsohn for niml cre, nd K. McArthur nd I. Schieren for help in prepring the mnuscript. We re grteful to S. Arer, N. Asi,. erin, Z.J. Hung, J. Milrndt, J. Snes, G. Szo nd MMR-GENSAT for mouse lines, nd especilly to F. ostntini for permission to use unpulished Gdnf<reERT2 mice. We thnk R. Axel, M. hurchlnd, K. Frnks, S. Grillner, J. Krkuer,. Mill, S. Polik, K. Ritol, D. Stettler nd D. Wolpert for discussion nd/or comments on the mnuscript. A.W.H. ws supported y the Howrd Hughes Medicl Institute nd the Roert Leet nd lr Guthrie Ptterson Trust Fellowship. T.M.J. ws supported y grnts from the Ntionl Institute of Neurologicl Disorders nd Stroke, the Wellcome Trust, the G. Hrold nd Leil Y. Mthers Foundtion, Project A.L.S. nd is n investigtor of the Howrd Hughes Medicl Institute. AUTHOR ONTRIBUTIONS A.W.H. nd T.M.J. conceived of the project nd plnned the experiments. A.W.H. performed the experiments. A.W.H. nd T.M.J. nlyzed the dt nd wrote the mnuscript. OMPETING FINANIAL INTERESTS The uthors declre no competing finncil interests. Pulished online t Reprints nd permissions informtion is ville online t reprintsndpermissions/. 1. Eccles, J.., Oscrsson, O. & Willis, W.D. Synptic ction of group I nd II fferent fires of muscle on the cells of the dorsl spinocereellr trct. J. Physiol. (Lond.) 158, (1961). 2. Lunderg, A. Ascending spinl hindlim pthwys in the ct. Prog. Brin Res. 12, (1964). 3. Oscrsson, O. Functionl orgniztion of the spino- nd cuneocereellr trcts. Physiol. Rev. 45, (1965). 4. Bosco, G. & Poppele, R.E. Proprioception from spinocereellr perspective. Physiol. Rev. 81, (2001). 5. Mtsushit, M. & Hosoy, Y. ells of origin of the spinocereellr trct in the rt, studied with the method of retrogrde trnsport of horserdish peroxidse. Brin Res. 173, (1979). 6. Edgley, S.A. & Jnkowsk, E. Informtion processed y dorsl horn spinocereellr trct neurones in the ct. J. Physiol. (Lond.) 397, (1988). 7. Wlmsley, B. entrl synptic trnsmission: studies t the connection etween primry fferent fires nd dorsl spinocereellr trct (DST) neurones in lrke s column of the spinl cord. Prog. Neuroiol. 36, (1991). 8. Hongo, T. et l. Trjectory of group I nd I fiers from the hind-lim muscles t the L3 nd L4 segments of the spinl cord of the ct. J. omp. Neurol. 262, (1987). 9. Mnn, M.D. lrke s column nd the dorsl spinocereellr trct: review. Brin Behv. Evol. 7, (1973). 10. Wolpert, D.M., Ghhrmni, Z. & Jordn, M.I. An internl model for sensorimotor integrtion. Science 269, (1995). 11. Wolpert, D.M. & Mill, R.. Forwrd models for physiologicl motor control. Neurl Netw. 9, (1996). 12. Sommer, M.A. & Wurtz, R.H. A pthwy in primte rin for internl monitoring of movements. Science 296, (2002). 13. Jennerod, M. Motor ognition: Wht Actions Tell the Self 1 21 (Oxford University Press, Oxford, 2006). 14. Roy, J.E. & ullen, K.E. Selective processing of vestiulr refference during selfgenerted hed motion. J. Neurosci. 21, (2001). 15. Mulliken, G.H., Musllm, S. & Andersen, R.A. Forwrd estimtion of movement stte in posterior prietl cortex. Proc. Ntl. Acd. Sci. USA 105, (2008). 16. Ener, T.J. & Pslr, S. ereellum predicts the future motor stte. ereellum 7, (2008). 17. Seki, K., Perlmutter, S.I. & Fetz, E.E. Sensory input to primte spinl cord is presynpticlly inhiited during voluntry movement. Nt. Neurosci. 6, (2003). 18. Hongo, T. & Okd, Y. orticlly evoked pre- nd postsynptic inhiition of impulse trnsmission to the dorsl spinocereellr trct. Exp. Brin Res. 3, (1967). 19. Hongo, T., Okd, Y. & Sto, M. orticofugl influences on trnsmission to the dorsl spinocereellr trct from hindlim primry fferents. Exp. Brin Res. 3, (1967). 20. Mott, F. Microscopicl exmintion of lrke s column in mn, the monkey nd the dog. J. Ant. Physiol. 22, (1888). 21. Nosrt,.A. et l. ellulr expression of GDNF mrna suggests multiple functions inside nd outside the nervous system. ell Tissue Res. 286, (1996). 22. Moore, M.W. et l. Renl nd neuronl normlities in mice lcking GDNF. Nture 382, (1996). 23. Hippenmeyer, S. et l. A developmentl switch in the response of DRG neurons to ETS trnscription fctor signling. PLoS Biol. 3, e159 (2005). 24. Mtsushit, M. & Go, X. Projections from the thorcic cord to the cereellr nuclei in the rt, studied y nterogrde xonl trcing. J. omp. Neurol. 386, (1997). 25. Dumitriu, D., ossrt, R., Hung, J. & Yuste, R. orreltion etween xonl morphologies nd synptic input kinetics of interneurons from mouse visul cortex. ere. ortex 17, (2007). 26. Yoshid, Y., Hn, B., Mendelsohn, M. & Jessell, T.M. PlexinA1 signling directs the segregtion of proprioceptive sensory xons in the developing spinl cord. Neuron 52, (2006). 27. Alvrez, F.J., Villl, R.M., Zerd, R. & Schneider, S.P. Vesiculr glutmte trnsporters in the spinl cord, with specil reference to sensory primry fferent synpses. J. omp. Neurol. 472, (2004). 28. Breyre, F.M., Kerschensteiner, M., Misgeld, T. & Snes, J.R. Trnsgenic leling of the corticospinl trct for monitoring xonl responses to spinl cord injury. Nt. Med. 11, (2005). 29. Finger, J.H. et l. The netrin 1 receptors Unc5h3 nd Dcc re necessry t multiple choice points for the guidnce of corticospinl trct xons. J. Neurosci. 22, (2002). 30. López-Bendito, G. et l. Preferentil origin nd lyer destintion of GAD65-GFP corticl interneurons. ere. ortex 14, (2004). 31. Lomelí, J., Quevedo, J., Linres, P. & Rudomin, P. Locl control of informtion flow in segmentl nd scending collterls of single fferents. Nture 395, 6004 (1998). 32. Betley, J.N. et l. Stringent specificity in the construction of GABAergic presynptic inhiitory circuit. ell 139, (2009). 33. Prochzk, A., Gillrd, D. & Bennett, D.J. Implictions of positive feedck in the control of movement. J. Neurophysiol. 77, (1997). 34. Arshvsky, Y.I., Berkinlit, M.B., Fukson, O.I., Gelfnd, I.M. & Orlovsky, G.N. Origin of modultion in neurones of the ventrl spinocereellr trct during locomotion. Brin Res. 43, (1972). 35. Arshvsky, Y.I., Gelfnd, I.M., Orlovsky, G.N., Pvlov, G.A. & Popov, L.B. Origin of signls conveyed y the ventrl spino-cereellr trct nd spino-reticulo-cereellr pthwy. Exp. Brin Res. 54, (1984). 36. Lunderg, A. Function of the ventrl spinocereellr trct. A new hypothesis. Exp. Brin Res. 12, (1971). 37. Poulet, J.F. & Hedwig, B. The cellulr sis of corollry dischrge. Science 311, (2006). 38. Blkemore, S.J. & Sirigu, A. Action prediction in the cereellum nd in the prietl loe. Exp. Brin Res. 153, (2003). 39. Gorski, J.A. et l. orticl excittory neurons nd gli, ut not GABAergic neurons, re produced in the Emx1-expressing linege. J. Neurosci. 22, (2002). nture NEUROSIENE VOLUME 13 NUMBER 10 OTOBER

8 ONLINE METHODS Retrogrde neuronl leling. Retrogrde leling of spinl cord neurons projecting to the cereellum ws performed s descried 40 using oth fluorogold (Fluorochrome) nd TB (Vector Lortories) trcers with minimum of 3 d of trnsport. Mice (P5 7) were sujected to hypothermic nesthesi, the skin covering the skull ws clened, 500 nl of retrogrde lel ws delivered to the cereellum using PB600 repeting dispenser (Hmilton) with 26 guge needle, the needle ws slowly retrcted, the surfce of the hed reclened, nd the niml ws wrmed nd returned to its cge. The needle ws mnully guided into the cereellum nd the injection site ws evluted y exmining residul trcer t the injection site nd y the pttern of retrogrde leling. In situ hyridiztion histochemistry. Proes for in situ hyridiztion detection of Gdnf, Gd2, Gd1 nd Slc65 mrna were generted using PR nd TOPO cloning (Invitrogen). In situ hyridiztion histochemistry ws performed on cryostt sections (20 μm). For dul-lel in situ hyridiztion histochemistry nd retrogrde leling, spinl cord sections were cut, ech section ws registered, fluorogold-leling ws photogrphed t two different mgnifictions, sections were hyridized with Gdnf proe, the resulting hyridiztion signl ws photogrphed nd imges of fluorogold nd hyridiztion lel were superimposed. -glctosidse stining. β-glctosidse ctivity ws visulized using stndrd X-Gl stining protocol. Gdnf<LcZ mice were killed y perfusion with 4% prformldehyde (wt/vol) nd post-fixed for mximum of 3 h. Spinl cord tissue ws hemisected or trnsversely sectioned (100 μm) with virtome. Biocytin-leling nd electrophysiology. Biocytin-leling nd electrophysiology were performed s descried 41. Fluorogold ws injected 3 or more dys efore iocytin leling nd electrophysiology. The mice were decpitted, the verterl column ws removed nd lminectomy ws performed. After removl of the dur, lumr dorsl roots were cut proximl to the dorsl root gngli nd the spinl cord ws cut long the ventrl midline to the dorsl columns. The spinl cord ws removed from the verterl column, incuted in rtificil cererospinl fluid (125 mm Nl, 26 mm NHO 3, 1.25 mm NH 2 PO 4, 2.5 mm Kl, 2 mm l 2, 1 mm Mgl 2, 0.4 mm scoric cid, 2 mm pyruvte nd 26 mm d-glucose, nd equilirted with 95% O 2 nd 5% O 2 ) for 1 h nd then positioned in chmer (Wrner Instruments). A lumr dorsl root (L4 to L6) ws drwn into suction electrode nd concentric ipolr electrode (FH) ws positioned with micromnipultor (MP-285, Sutter Instruments) on the ventrl spect of the dorsl column in cervicl spinl cord. To selectively stimulte corticl xons, we mpped the corticospinl trct using GFP fluorescence in hemisected preprtions from Emx1<GFP trnsgenic mice. This position ws identified in non GFP-expressing tissue using infrred-differentil interference contrst (IR-DI) imging. Electrophysiologicl experiments were performed on hemisected spinl cord preprtions of P10 15 mice t Fluorogold-leled neurons, reveled y fluorescence nd IR-DI imging, were selected from T6 L2 spinl levels. This ensured minimum six segment seprtion etween the dorsl column stimultion electrode nd the recording site. Whole-cell, ptch-clmp recordings (internl pipette contents: 130 mm potssium gluconte, 5 mm Nl, 1 mm l 2, 1 mm Mgl 2, 10 mm HEPES nd 4 mm N 2 ATP) were otined from fluorogold-leled neurons with resting memrne potentils less thn 50. Evoked responses were ssyed (Multilmp 700B, MDS) from grded stimultion (S88 Grss stimultor; Astro-Med) through the suction electrode (dorsl roots, sensory xons) or the concentric ipolr electrode (dorsl column, corticospinl trct). Phrmcologicl experiments were performed y dding drugs (Sigm Aldrich) to the superfuste. For iocytin-leling, spinl tissue contining iocytin-filled neurons were fixed for minimum of 12 h nd processed for immunohistochemistry. Tmoxifen injection. To induce re recomintion nd mgfp expression in neurons, we injected tmoxifen (0.5 to 1 mg, Sigm Aldrich) intrperitonelly in ~5 g, P5 7 mice nd hrvested the tissue 3 7 d fter injection. Tmoxifen (10 mg ml 1, Sigm Aldrich) ws dissolved in sesme oil (Sigm Aldrich) nd ethnol, t 37. Immunohistochemistry. Immunohistochemistry ws performed on cryostt nd virtome sections through sequentil exposure to primry ntiodies nd fluorophore-conjugted secondry ntiodies (Jckson Immunoreserch). Virtome sections ( μm) were incuted in primry ntiodies for 3 d t 4 nd secondry ntiodies for 1 d t 4. onfocl microscopy. onfocl imges were tken with LSM 710 microscope (rl Zeiss) with motorized stge to fcilitte the cretion of montge imges. onfigurtions for high-mgnifiction imges were set for thick virtome sections of cereellum. Stge xy coordintes covering the entire specimen were determined nd the Zeiss Multi-time mcro ws used to ssemle montge high-resolution imges. Mice. We used P10 21 Pvl<GFP nd Emx1<GFP mice to mp proprioceptive (Pvl) nd corticl (Emx1) inputs to retrogrdely or geneticlly leled neurons. We lso used Gdnf<LcZ mice 22, Gdnf<reERT2 ( generous gift from F. ostntini, olumi University), Gd2<GFP 30, Pvl<GFP 25, Emx1<GFP (MMRR-GENSAT), Emx1<cre 39, Pvl<re nd Tu<lsl-mGFP 23. Procedures performed in this study were pproved y the Institutionl Animl re nd Use ommittee of olumi University. Antiodies. We used got nd chick ntiody to β-glctosidse from Biogenesis nd Acm, rit nd chick ntiody to GFP from Invitrogen nd Aves L, rit nd guine pig ntiody to vglut1 from Synptic Systems GmH nd ref. 32, mouse nd rit ntiody to Gd1 from hemicon nd ref. 32, guine pig ntiody to GlyT2 from hemicon, rit ntiody to VGAT from ref. 32, nd rt nd mouse ntiody to Gd2 from ref Arsénio Nunes, M.L. & Sotelo,. Development of the spinocereellr system in the postntl rt. J. omp. Neurol. 237, (1985). 41. Hntmn, A.W., vn den Pol, A.N. & Perl, E.R. Morphologicl nd physiologicl fetures of set of spinl sustnti geltinos neurons defined y green fluorescent protein expression. J. Neurosci. 24, (2004). nture NEUROSIENE doi: /nn.2637