B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ l oca te /b ra in res

Research Report

High-resolution neuroanatomical tract-tracing for the analysisof striatal microcircuits

Pascal Salina, María Castleb, Philippe Kachidiana, Pedro Barroso-Chineab, Iciar P. Lópezb,Alberto J. Ricob, Lydia Kerkerian-Le Goffa, Patrice Coulonc, José L. Lanciegob,⁎aDevelopmental Biology Institute of Marseille-Luminy, UMR 6216 CNRS-Université de la Méditerranée, Marseille, FrancebBasal Ganglia Neuromorphology Laboratory, Neurosciences Division, Center for Applied Medical Research (CIMA and CIBERNED),University of Navarra Medical College, Pamplona, SpaincLab. Plasticité et Physio-Pathologie de la Motricité, UMR 6196 CNRS-Université de la Méditerranée, Marseille, France

A R T I C L E I N F O

⁎ Corresponding author. Basal Ganglia NeuromPio XII Avenue No 55, 31008 Pamplona, Spain

E-mail address: jlanciego@unav.es (J.L. La

0006-8993/$ – see front matter © 2008 Elsevidoi:10.1016/j.brainres.2008.05.011

A B S T R A C T

Article history:Accepted 2 May 2008Available online 16 May 2008

Althoughcurrently available retrograde tracers areuseful tools for identifying striatal projectionneurons, transported tracers often remained restricted within the neuronal somata and thethickest, main dendrites. Indeed, thin dendrites located far away from the cell soma as well aspost-synaptic elements such as dendritic spines cannot be labeled unless performingintracellular injections. In this regard, the subsequent use of anterograde tracers for thelabeling of striatal afferents often failed to unequivocally elucidate whether a given afferentmakes true contacts with striatal projections neurons. Here we show that such a technicalconstraint can now be circumvented by retrograde tracing using rabies virus (RV).Immunofluorescence detection with a monoclonal antibody directed against the viralphosphoprotein resulted in a consistent Golgi-like labeling of striatal projection neurons,allowing clear visualization of small-size elements such as thin dendrites as well as dendriticspines. The combination of this retrograde tracing together with dual anterograde tracing ofcortical and thalamic afferents has proven to be a useful tool for ascertaining striatalmicrocircuits. Indeed, by taking advantage of the trans-synaptic spread of RV, differentsubpopulations of local-circuit neurons modulating striatal efferent neurons can also beidentified. At the striatal level, structures displaying labelingwere visualized under the confocallaser-scanning microscope at high resolution. Once acquired, confocal stacks of images werefirstlydeconvolutedandthenprocessed through3D-volumerendering inorder tounequivocallyidentify true contacts betweenpre-synaptic elements (axon terminals from cortical or thalamicsources) and post-synaptic elements (projection neurons and/or interneurons labeledwith RV).

© 2008 Elsevier B.V. All rights reserved.

Keywords:Rabies virusAnterograde tracingRetrograde tracingTrans-synaptic tracingBasal gangliaStriatumConfocal microscopy

1. Introduction

Charting the complex connections of the central nervoussystem is fundamental to understand how brain networks

orphology Laboratory, Ne. Fax: +34 948 194715.nciego).

er B.V. All rights reserved

functions. Remarkable advanceshavebeenmade in this field inrecent decades due to the expansion of neural tracingparadigms. Anterograde tracers such as Phaseolus vulgaris-leucoagglutinin (PHA-L; Gerfen and Sawchenko, 1984) and

urosciences Division, Center for Applied Medical Research (CIMA),

.

50 B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

biotinylated dextran amine (BDA; Veenman et al., 1992) providea proper elucidation of the pre-synaptic element. However, theaccurate visualization of the post-synaptic element remainedelusive by means of retrograde tract-tracing methods. Wheninjected in a given brain area, retrograde tracers are takenupbyaxon terminals innervating this area and retrogradely trans-ported to the parent cell body (for a review, see Lanciego andWouterlood, 2006). When relying on “classical” retrogradetracers, the main caveat is that the transported tracer remainsrestricted to the cell body and to the thickest, main dendrites,without significant labeling of neuronal processes located faraway from the parent cell body such as thin dendrites anddendritic spines. This is also a common drawback when usingretrograde tracers with a different nature such as for exampleseveral strains of trans-synaptically transported viruses (Geer-ling et al., 2006), and as such, it represents amajor limitation forthe detailed analysis of anatomical interactions.

In summary, there is a clear need for a tracer that meetsseveral demands, including: (i) to allow complete visualizationof the labeled neurons, including the full dendritic tree as wellas small structures such as dendritic spines, (ii) to betransported exclusively in the retrograde direction, thereforeallowing the accurate identification of the brain areas targetedby labeled neurons and (iii) to be compatible with otherexisting tools for tract-tracing. When searching the technicalarsenal currently available, it becomes evident that rabiesvirus (RV; Challenge Virus Standard strain) does approach theideal. When compared to other commonly used trans-synap-tically transported viruses such as the herpes viruses, rabiesvirus exhibit higher efficiency for retrograde tracing and lowcytopathicity within infected cells (Ugolini, 1995). Further-more, the lack of viral transport through fibers of passage alsorepresents an additional advantage (Ugolini, 1995; Kelly andStrick, 2000; Nassi and Callaway, 2007). Indeed, the trans-synaptic spread of the virus is a useful strategy for the analysisof neuronalmicrocircuits, and this specially holds true in brainareas such as the striatum, in which interneurons play keyroles in modulating the activity of projection neurons (Kawa-

Fig. 1 – Injection sites and Golgi-like labeling obtainedwith retrogshowing the deposit of rabies virus (RV) in GP. Epifluorescent illuanterograde tracer Phaseolus vulgaris-leucoagglutinin (PHA-L) at thEpifluorescent illumination. For reference purposes, the positionsubthalamic nucleus (STN) are delineated. Scale bar=1 mm. (C) D(BDA) in the cortical area M1. Injected BDA was visualized by anScale bar=1 mm. (D and E) Comparison between two different wthe retrograde tracer Fluoro-Gold (D) or Rabies Virus (E). Fluoro-Gprojection neurons that innervate a given brain area to which thFluoro-Gold (as well as any other of the retrograde tracers availaparental cell body and the thickest dendrites. By contrast, rabiesdirected against viral genome-associated proteins, our protocol ina form of the viral phosphoprotein present in the whole cytoplasdendritic tree can be visualized by immunofluorescence and evelabeled (as shown for striato-pallidal neurons in panels F and G). Tpost-synaptic elements of a given population of projection neurois 60 µm for D and E and 120 µm for F and G. (H–I) Any other typeexhibited Golgi-like labeling, such as STN neurons projecting toinitially described by Kincaid et al., 1991), as well as neurons wiinnervating the SNr (panels J and K, respectively). Scale bar for p

guchi et al., 1995). More recently, a deletion-mutant version ofRV lacking the glycoprotein responsible for viral neuronaltropism became available to study the morphology andphysiology of projection neurons (Wickersham et al., 2007a).The virus incorporates the glycoprotein in its envelope using atranscomplementation strategy, so that it can infect neuronsprojecting to the injection site but it cannot spread beyond theinfected cells. Furthermore, a trans-synaptic tracer based onRV able to cross only one synapse beyond the primary infectedcells has been designed for in vitro studies in brain slices. Thisprovides an elegant technique to label neurons that project to asingle, genetically targeted synapse (Wickershamet al., 2007b).To date, RV has been essentially used to trace multi-synapticcircuits and it is generally detectedwith antibodies against theRVnucleoprotein, which provides labeling restricted to the cellbody and the proximal dendrites.

Here, we designed a protocol in which RV is used as a tool tostudy morphology and connectivity of neurons identified fromtheir projection site (e.g., first-order neurons). Viral detection isaccomplished by relying on a mouse monoclonal antibodyagainst a viral phosphoprotein present in the whole cytoplasmof the RV-infected neuron (31G10 isolated during the fusionexperiment described by Raux et al., 1997), therefore enablingthe unambiguous characterization of the post-synaptic ele-ment. The conducted procedure for viral detection is fullycompatible with existing tools for anterograde tract-tracingneurons, which represents its main added value. It is combinedin thepresent studywithanterogradeaxonal tracingwithPHA-Lan BDA followed by multiple fluorescence labeling. By thiscombination, the inputs to a specific retrogradely-labeledpopulation of efferent neurons and hence connected interneur-ons (first-order and second-order neurons, respectively) can beanalyzing qualitatively and quantitatively under the confocalmicroscope. Structures displaying labeling were visualizedunder the confocal laser-scanningmicroscope. The subsequentuse of powerful post-acquisition image software resulted in theunequivocal visualization of the pre- and the post-synapticelements with an unprecedented level of detail.

rade tracing using rabies virus. (A) Low-magnification picturemination. Scale bar=1 mm. (B) Iontophoretic delivery of thee level of the parafascicular nucleus of the thalamus (PF).of the fasciculus retroflexus (fr) and the boundaries of theelivery of the anterograde tracer biotinylated dextran amineHRP-coupled streptavidin followed by a DAB-Ni precipitate.ays to visualize striatopallidal medium-spiny neurons usingold is currently the first choice retrograde tracer to identifye tracer is stereotaxically delivered. The main caveat of usingble so far) is that the transported tracer is restricted to thevirus offers better performance. Instead of using antibodiesvolves the use of a primarymonoclonal antibody that binds tom of the infected neuron. With this strategy, the entiren the thinnest dendrites and dendritic spines are clearlyo the best of our knowledge, this is the first time in which the

ns are revealed at this unprecedented level of detail. Scale barof projection neuron innervating the area of RV deposit alsoGP (H), thalamo-pallidal projecting neurons (I; a projectionthin the raphe nuclei and the pedunculopontine nucleusanels H–K=40 µm.

51B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

This approach is applied to study microcircuits in thestriatum which comprises neuronal populations of interneur-ons and two main populations of projection neurons. Thestriatum receives two major excitatory glutamatergic inputsoriginating from the cerebral cortex and from the caudalintralaminar thalamic nuclei (notably the parafascicularnucleus — Pf). Elucidating the fine connectivity betweenthese systems and their remodeling in pathophysiologicalconditions, is of primordial importance in the field of basalganglia related motor control and disorders.

2. Results

2.1. Striatal afferents and striatofugal neurons

The delivery of RV into either the substantia nigra pars reticulata(SNr) or the globus pallidus (GP) labeled a substantial number of

neurons within the dorsolateral striatum. Up to several hun-dreds of neurons are clearly visible in almost any single sectioncomprising striatal levels. All neurons displayed a robustlabeling in a Golgi-like fashion. The entire dendritic tree becamenoticeable, together with small structures such as dendriticspines. Nicely stained neurons were also found within otherbrain areas innervating GP and/or SNr such as the subthalamicnucleus, the raphe nuclei and the pedunculopontine nucleus(Fig. 1), albeit in this case labeled neurons are characterized bysmooth, and spiny dendrites. When compared with the kind oflabeling onemay expect when using Fluoro-Gold for identifyingstriatopallidal neurons, the use of RV clearly afforded a superiorperformance in a similar paradigm (Fig. 1).

As expected, the deposit of the tracer BDA into the primarymotor area of the cortex (M1) resulted in massive anterogradelabeling within both the ipsilateral and the contralateralstriatum. Dense terminal fields of corticostriatal afferentswere clearly visible throughout dorsolateral striatal territories.

52 B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

The same also holds true for the injection of PHA-L into theparafascicular thalamic nucleus (PF), albeit in that casethalamostriatal afferents only innervate the ipsilateral stria-tum.When injectedwithin dorsal areas of PF, striatal afferentsfrom thalamic sources were distributed within dorsolateralterritories of the striatum, therefore showing a high degree ofconvergence with striatal afferents arising from the M1cortical area. A great morphological detail was noticed inboth the BDA- and the PHA-L-labeled axon terminals, in whichaxonal varicosities and terminal boutons were clearly visible.

In all cases, a large number of RV-labeled striatofugalneurons were clearly noticed intermingled within the term-inal fields of corticostriatal and thalamostriatal projections(Fig. 2). Several examples of RV-containing neurons locatedinside overlapping areas for cortical and thalamic terminal

fields were randomly selected for high-resolution confocallaser-scanning and further processed through deconvolutionfollowed by volumetric rendering. Once reconstructed, severalpotential contacts between anterogradely-labeled terminalsand retrogradely-labeled neurons became evident. Overall andin keeping with the existing knowledge, corticostriatal axonspreferentially innervate dendritic spines of striatofugal neu-rons, whereas thalamostriatal terminals make contacts withdendritic shafts of the medium-sized spiny striatal neurons(MSNs) and, to a lesser degree, with dendritic spines. Regard-less the identity of the post-synaptic structure targeted bystriatal afferents from cortical or thalamic origin, it is worthnoting that most of the presumptive contacts were noticed ondendritic shafts or spines located far away from the cell soma(Fig. 2).

Fig. 3 – Trans-synaptic tracing with RV to identify striatal interneurons innervating striatal projection neurons. (A and E):thalamostriatal afferents, labeled with PHA-L. (B and F): ChAT+ striatal interneurons. (C and G): Labeling with RV, includingretrograde labeling of the first-order neurons (striatonigral-projecting neurons) and trans-synaptic tracing of a subpopulation ofChAT+ neurons (asterisks; second-order neurons). (D and H): merged channels. Only the subpopulation of cholinergic neuronsconnected to the RV-infected striatofugal neuronal population was labeled with the virus. Cholinergic interneurons notassociated with infected striatofugal neurons are only recognized by the anti-ChAT antibody (arrowheads). The labeling oflocal-circuit neurons is weak, since we used only a short period of RV tracing in an attempt to minimize second-order infectionof the striatal input systems. Despite the weak labeling of interneurons, this method produced an accurate anatomicalassessment of striatal microcircuits. Indeed, for the first time it will be possible to ascertain whether one thalamostriatalafferent innervates either a given striatofugal neuron or an interneuron that is in turn linked to the projection neuron (or both).Scale bar=100 µm in all panels.

53B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

2.2. Trans-synaptic labeling: Interneurons

Besides labeling a large number of striatal projection neurons(first-order neurons), the presence of a moderate number of

Fig. 2 – The coupling of retrograde rabies virus tracing of striatothalamic afferents to study striatal connectivity. (A) Flow chart suexamples illustrating the results of the post-acquisition procedurby 3D-volumetric rendering). Once deconvoluted and 3D-renderedwithout significantly reducing the resolution. The final magnificpre- and post-synaptic elements. Indeed, the images obtained caexists (i.e.: without any tissue interposed in between). Accordingafferents make contacts with dendritic spines of striatal projectiopost-synaptic targets for thalamic axons arising from the caudalwhether or not glutamatergic cortical and thalamic projections toprojection neurons. Here we provide the first evidence for converneuron. Panels G-I illustrate the arrival of cortical inputs (color-coone thalamic afferent (blue terminal) contacts the dendritic shaft oprojection illustrated in panel G, it is somewhat unclearwhether tshaft or with the neck of the dendritic spine located to the right.rotated through any angle to clearly demonstrate that there is nodendritic shaft located below. In contrast, it is apparent that therneck of the dendritic spine (as shown in the projections illustratepanels C-F, 7 µm in panel G. Color codes: Green (striatopallidal neBlue (thalamostriatal afferents, labeled with PHA-L).

faintly-labeled striatal neurons with morphologies other thanwhat typically accounts for MSNs was consistently found in allcases. Theseneurons (second-order neurons) are characterizedby awide range of body sizes (fromsmall-sized to large somata)

fugal neurons with dual anterograde tracing of cortical andmmarizing the conducted experimental design. (B–J) Severales carried out here on confocal stacks (deconvolution followed, the images can bemagnified by several orders ofmagnitudeation is sufficient to resolve individual contacts betweenn be moved in the x-y-z axis to make sure that a ‘true’ contactto the current opinion, our data show that corticostriataln neurons, whereas dendritic shafts are the mainintralaminar nuclei. At present, there is a strong debate as tothe striatum converge onto a specific subtype of striatal

ging cortical and thalamic inputs onto a single striatopallidalded in red) to dendritic spines at higher magnification, whilef the same striatopallidal-projecting neuron. By relying on thehe blue terminal (labeled #4)makes contact with the dendriticTaking advantage of the 3D reconstruction, the image can betissue interposed between the blue contact (#4) and the

e is no contact between the thalamostriatal terminal and thed in panels H and I). Scale bar=50 µm for panel B, 13.5 µm inuron, RV-labeled), Red (corticostriatal afferents, BDA-labeled),

54 B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

with long and smooth dendrites, which is a characteristicfeature of interneurons vsMSNs. Immunodetection ofmarkerscharacterizing themajor typesof striatal interneurons (choline-acetyl-transferase, parvalbumin, calretinin and nitric oxidesynthase; for a review, see Kawaguchi et al., 1995) confirmedthat all these neurons with weak RV labeling are interneurons.Examples of this kind are given in Fig. 3, presenting cholinergicneurons for illustrative purposes. Labeling of the second-orderneurons by means of trans-synaptic tracing has proven to be auseful tool for the analysis of brain microcircuits in whichinterneurons play a keymodulator role such as, for example, atthe striatal level. As depicted in Fig. 3, colocalization phenom-ena between viral proteins and ChATwere only found within asubpopulation of cholinergic neurons, and a substantialnumber of ChAT+ interneurons remained unlabeled with RV.In other words, only the fraction of ChAT+ interneuronsconnected with striatal neurons projecting to a given targetbecame infected with the virus by means of trans-synaptictracing. Since it is well known that both the projection neuronsas well as the cholinergic interneurons are innervated bythalamic projections, this procedure enables for the first timethe unambiguous visualization of the “three corners” of thisstriatalmicrocircuit, composedby thalamostriatalaxons reach-ing cholinergic neurons innervating striatofugal neurons,which in turn were also directly approached by the samesource of thalamic afferents (Fig. 3).

3. Discussion

Here, we have designed a procedure using RV to obtain Golgi-like retrograde labeling of projection neurons. When com-bined with anterograde tracing, the presence of potentialcontacts between the pre- and post-synaptic elements can beelucidated in a single histological section at an unprecedentedlevel of detail. The method is based on multiple immuno-fluorescence detection followed by deconvolution and volu-metric rendering of the confocal images acquired.

3.1. Alternative protocols enabling the visualization of thecompletemorphology of projection neurons in vivo. Advantagesand limitations

Several approaches addressing the complete visualization ofprojection neurons have been made available, comprising anumber of different strategies each one with their ownstrengths and pitfalls. Among others, neuronal stains basedon Golgi impregnations seem to be a natural choice. In thehands of experienced researchers, the Golgi method can affordexquisite and detailed visualization of the complete dendriticarborizations of striatal MSNs in a wide range of different brainspecimens, also including human brain samples obtained fromnecropsies (Stephens et al., 2005). Although aesthetically verypleasant, even the best Golgi impregnations cannot be used forthe accurate identification of striatonigral- and striatopallidal-projecting neurons, since both subtypes of striatal MSNs arebasically identical from amorphological point of view. Further-more, Golgi-like retrograde labeling of striatofugal neurons hasbeen reported after the delivery of BDA in basal ganglia outputnuclei (Rajakumar et al., 1993; Sidibé and Smith, 1996).

A different strategy aimed at circumventing these limita-tions of the Golgi method consists on performing intracellularinjections with a fluorescent dye (Day et al., 2006) followed bythe use of antibodies against the dopamine receptors 1 and 2,sinceMSNs innervating the SNr contain D1 receptors, whereasMSNs projecting to the GP are known to express D2 receptors(Gerfen et al., 1990). However, recent data demonstrated thatthe sharp D1-D2 segregation in striatal MSNs couldn't longerbe maintained (Meador-Woodruff et al., 1991; Weiner et al.,1991; Surmeier et al., 1992, 1993; Ariano et al., 1993), in keepingwith existing data showing that the striatofugal system ishighly collateralized (Kawaguchi et al., 1991; Parent et al., 1995;2000; Wu et al., 2000; Castle et al., 2005).

In order to achieve an unambiguous characterization ofprojection neurons, a feasible choice has been introduced fewyears ago by Buhl and Lubke (1989), comprising intracellularinjections of Lucifer Yellow (LY) into neurons retrogradelylabeledwith Fast Blue. Since LY-labeledneurons can be furtherdetected using antisera against LY, this technique can becombinedwith other commonlyused anterograde tracers suchas BDA and PHA-L (Wouterlood et al., 1990, 1992; Shi andCassell, 1993; Jorritsma-Byham et al., 1994). Although techni-cally very demanding, the combination of intracellular stain-ing of retrogradely-labeled neurons with anterograde tract-tracing techniques has settled themethodological standard forthe accurate visualization of pre- and post-synaptic elementsunder light microscopy over many years. Two main caveatshave impeded the implementation of this protocol for routineanalysis of brain microcircuits: firstly, only a reduced numberof neurons can be intracellularly injected per histologicalsection and secondly, since the anterogradely-labeled term-inal fields are not visible by the time in which the intracellularinjection is performed, to obtain a number of LY-filled cellsreceiving a substantial number of pre-synaptic contacts oftenis only a matter of chance.

When compared to existing techniques, the protocolpresented here has several advantages since it allows thevisualization of the complete morphology of projection neu-rons in a Golgi-like fashion and at the same time this procedurecan easily be combined with other already existing tools foridentifying the pre-synaptic element by means of anterogradetract-tracing. The need for a two-step surgical procedure fortracer delivery (shorter survival times are required for RVtracing than for PHA-L and BDA transport) and the specificsafety issues that must be taken into consideration whendealing with this viral strain (Kelly and Strick, 2000) arebalanced by the remarkable results obtained.

3.2. Theadded value of trans-synaptic tracing for identifyinglocal-circuit neurons

As in many other brain areas, interneurons play key roles inmodulating the activity of projection neurons within thestriatum (Kawaguchi et al., 1995; Koos and Tepper, 1999).When it comes to study the architecture of brainmicrocircuits,it has been a long desire to unequivocally ascertain theidentity of a given interneuron that modulates projectionneurons innervating a particular target. In this regard, thetrans-synaptic spread of RV is a good choice that fulfills theseneeds, since only the interneurons directly linked to RV-

55B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

retrogradely-labeled projection neurons became infected withthe virus, whereas interneurons innervating striatofugalneurons projecting to brain areas other than the one injectedwith RV remain unlabeled. Indeed, it is well known that mostof the different types of striatal interneurons are contacted byafferents from cortical or thalamic sources (Lapper and Bolam,1992; Lapper et al., 1992; Bennet and Bolam, 1994; Dimovaet al., 1993; Kawaguchi et al., 1995; Kachidian et al., 1996;Rudkin and Sadikot, 1999; Sidibé and Smith, 1999; Gonzaloet al., 2001). In this regard, the protocol presented here enablesfor the very first time the visualization of the entire circuit,consisting of one afferent system contacting a particular kindof interneuron, which in turn is linked to a projection neuron.In other words, this technique allows the accurate identifica-tion of both the pre- and post-synaptic elements for a giveninterneuron.

3.3. On the use of high-resolution confocal scanning forascertaining synaptic contacts. Is there still a need for electronmicroscopy?

The recent availability of powerful computer-assisted techni-ques for image post-acquisition processing has boosted theanalysis of data coming from neuroanatomical studies to anunprecedented level of detail. Computer-processing softwaresuch as deconvolution and volumetric or surface renderingsare especially well suited for improving the images taken aftermultifluorescence confocal laser scanning, therefore pushingforward the resolution limit that typically accounts for opticalmicroscopes. At present, only the electron microscope (EM)affords a magnification high enough to properly visualize thepre- and post-synaptic densities at the synaptic junction andtherefore EM is the ultimate method of choice for thevalidation of whether a presumptive contact between pre-and post-synaptic elements really accounts for either type I ortype II synapses. Although technically demanding, double-labeling EM allows for a clear assessment of excitatory andinhibitory interactions between neurons (see Sesack et al.,2006, for review purposes). However, high-resolution confocallaser scanning followed by deconvolution and volumetricrendering is a procedure that properly fills the gap betweentraditional optical microscopy and EM. Indeed, it has beencalculated that the combination of confocal laser scanningfollowed by post-acquisition image processing softwareenhances the resolution limit by a factor of 1.4 relative tolight microscopy (Pawley, 1995; Inoué, 1995). It is worth notingthat in order that a presumptive contact can be eligible as atrue contact, no optically empty space should be visiblebetween the pre- and the post-synaptic elements at anyangle of inspection in the three dimensional reconstruction(Shi and Cassell, 1993;Wouterlood et al., 2002). Further elegantrefinements of this concept have incorporated the indirectvisualization of the synaptic interface by conducting theimmunofluorescent detection of proteins located at the post-synaptic density such as ProSAP2/Shank3. In cases in whichtrue synapses exist, post-synaptic density-associated scaf-folding proteins appeared ‘sandwiched’ between the pre- andthe post-synaptic elements and therefore pinpointing thelocalization of synapses in a very accurate way (Wouterloodet al., 2003).

4. Experimental procedures

MaleWistar rats, with a body weight ranging from 240 to 280 gwere used in this study. The animals were handled at all timesaccording to the European Council Directive 86/609/EEC. Allexperimental procedures were approved by the EthicalCommittee for Animal Testing of the University of Navarra(Ref: 010-06). All stages of experiments which require the useof RV were achieved in a biosafety level 2 laboratory and thepersonnel involved in those experiments has been previouslyvaccinated.

The virus used in the experimentswas the laboratory strainChallenge Virus Standard (CVS-11). This strain is commonlyused in the tracing experiments performedwith RV, especiallywhen injected centrally (Kelly and Strick, 2003; Miyachi et al.,2005; Nassi and Callaway, 2006).

4.1. Surgical procedure (I): Anterograde tracer delivery

Animals were deeply anesthetized with an intraperitonealinjection of equithesin (4 ml/Kg). The rats were then placed ina stereotaxic frame (David Kopf, Tujunga, CA). The antero-grade tracers BDA (Molecular Probes, Leiden, The Netherlands)and PHA-L (Vector Labs, Burlingame, CA) were injected in asingle surgical session. BDA was pressure-delivered through aHamilton syringe in the left primary motor area M1 as a 10%solution in 0.01 M phosphate buffer (PB), pH 7.25 (injectedvolume of 0.3 µL). Next, PHA-L was iontophoretically injectedto the left parafascicular nucleus as a 2.5% solution in 0.1 M PB,pH 7.3, using a glass micropipette (inner tip diameter 20–30 µm) and a positive-pulsed direct current (5 µA, 7 s on/off).The tracer was injected for 20 min and once completed themicropipette was left in place for 5 min before withdrawal.

4.2. Surgical procedure (II): Delivery of rabies virus

Eight days post-delivery of the anterograde tracers, theanimals were reanesthetized as described above and placedagain in a stereotaxic frame. RV was then pressure-injectedthrough a Hamilton syringe within either the GP or the SNr, asa cell culture supernatant in minimal essential mediumtitrating 4×107 plaque forming units/ml. The final volumewas 200 nL.

4.3. Tissue processing

Ten days post-injection of PHA-L and BDA (40–42 h after viraldelivery), animals were anesthetized with an overdose ofequithesin and then transcardially perfused with a salineRinger's solution followed by 500mL of a cold fixative solutioncontaining 4% paraformaldehyde in 0.125 M PB, pH 7.4. Afterperfusion, the skull was opened and the brain removed andstored in a cryoprotective solution containing 20% glycerinand 2% dimethylsulphoxide in 0.125 M PB, pH 7.4 (Roseneet al., 1986). Frozen coronal microtome sections (40 µm-thick)were obtained and collected in 0.125 M PB, pH 7.4, in 10 seriesof adjacent sections. One series of sections were used for thetriple fluorescent neuroanatomical tracing combining BDA,PHA-L and RV. In four additional series of sections, the PHA-Lprotocol was replaced by the immunofluorescent detection of

56 B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

striatal interneurons (cholinergic neurons, parvalbumin-posi-tive neurons, neurons expressing nitric oxide syntase, as wellas calretinin-containing neurons), and then combined withthe visualization of transported BDA and rabies virus. Theremaining five series of sections were stored at −80 °C untilused for further histological processing.

4.4. Histology

For illustrative purposes, a detailed description is providedbelow dealing with the simultaneous visualization of trans-ported BDA, PHA-L and RV. Only minor adjustments mainlyrelated to the selection of antibodies are required whencombining the visualization of a given interneuron togetherwith the detection of BDA and RV.

The sections were first incubated in a cocktail solution ofprimary antisera containing 1:10,000mouse anti-RV phospho-protein monoclonal antibody and 1: 2,000 rabbit anti-PHA-L(Vector Labs) for 60 h at 4 °C. Next, the sections were exposedto a cocktail solution comprising 1:200 donkey anti-mouse IgGcoupled to Alexa®488 (Molecular Probes) and 1:200 donkeyanti-rabbit IgG labeled with Alexa®555 (Molecular Probes) for2 h at room temperature. Finally, sections were incubated inAlexa®633-conjugated streptavidin (1:100, Molecular Probes)for 90 min at room temperature.

Sections were mounted on glass slides using a 2% solutionof gelatin (Merck, Darmstadt, Germany) in 0.05 M Tris–HCl pH7.6, dried at room temperature in the dark, rapidly dehydratedin toluene, and coverslipped with Entellan (Merck). All anti-sera used in this procedure were diluted in 0.05 M Tris-buffered-saline pH 8, with 0.5% Triton X-100 (Sigma, St. Louis,MO; TBS-Tx). Extensive washing with 0.05 M TBS-Tx pH 8 wascarried out throughout the procedure. Several rinses with0.05 M Tris–HCl pH 7.6 were conducted prior to mounting thesections in gelatin.

4.5. Confocal visualization and post-acquisition processing

Sections were inspected under a Zeiss confocal microscope(LSM 510 Meta). The sections were first examined using low-magnification lenses and photomicrographs were taken fromthe areas of interest at high magnification (X63 oil-immersionPlan-Apochromat objective lens, NA 1.4) at a resolution of1024×1024 pixels. The emission from the argon laser at 488 nmwas filtered with a bandpass of 505–530 nm, whereas theemission following the excitation with the helium-neon laserat 543 nm was filtered though a bandpass of 560–615 nm.Finally, a longpass filter of 650 nm was used for the visualiza-tion of the emission arising from the helium-neon laser at633 nm.

Post-acquisition management of the resulting confocalstacks is a two-step process, firstly comprising the deconvolu-tionof theacquired images, and then followedby3D-volumetricrendering. Deconvolution is the process by which an imageacquired with a light microscope is restored using an imageformation model where a convolution is involved. This statis-tical approach corrects the amount of blur inherent within anyimage taken by an optical instrument, in an attempt to restorethe true appearance of the original image with the highestdegree of confidence, and has proven to be a particularly useful

tool for improving the appearance of confocal stacks. Theimages have been restoredusing the deconvolution software bySVI (Scientific Volume Imaging, The Netherlands), CMLE with200 iterations, 0.001% quality threshold and a theoretical PSF(point spread function). Deconvoluted images were thenprocessed through the so-called “volume rendering”, a kind of3D rendering resulting in a simulated 3D image of thefluorescence of all voxels within the entire Z series. A detailedexplanation on the rationale behind these procedures can befound elsewhere (Wouterlood, 2006).

Acknowledgments

We thank Ms. Elvira Roda for her expert technical assistance.Supported by Ministerio de Educación y Ciencia (BFU2006-06744), CIBERNED (CB06/05/0006), Fondo de InvestigacionesSanitarias (PI050137), Departamento de Educación del GobiernodeNavarra (18/2005), Fundación de InvestigaciónMédicaMutuaMadrileña, the UTE-project/FIMA, the CNRS (Centre National dela Recherche Scientifique), Université de laMéditerranée and bytheAgenceNationale pour la Recherche (ANR-05-NEUR-013-03).

R E F E R E N C E S

Ariano,M.A., Fisher, R.S., Smyk-Randall, E., Sibley, D.R., Levine,M.S.,1993.D2dopamine receptor distribution in the rodentCNSusinganti-peptide antisera. Brain Res. 609, 71–80.

Bennet, B.D., Bolam, J.P., 1994. Synaptic input and output ofparvalbumin immunoreactive neurons in the neostriatum ofthe rat. Neuroscience 62, 707–719.

Buhl, E.H., Lubke, J., 1989. Intracellular lucifer yellow injection infixed brain slices combined with retrograde tracing, light andelectron microscope. Neuroscience 28, 3–16.

Castle, M., Aymerich, M.S., Sánchez-Escobar, C., Gonzalo, N.,Obeso, J.A., Lanciego, J.L., 2005. Thalamic innervation of thedirect and indirect basal ganglia pathways in the rat: ipsi- andcontralateral projections. J. Comp. Neurol. 483, 143–153.

Day, M., Wang, Z., Ding, J., An, X., Ingham, C.A., Shering, A.F.,Wokosin, D., Ilijic, E., Sun, Z., Sampson, A.R., Mugnaini, E.,Deutch, A.Y., Sesack, S., Arbuthnott, G.W., Surmeier, D.J., 2006.Selective elimination of glutamatergic synapses onstriatopallidal neurons in Parkinson disease models. Nat.Neurosci. 9, 251–259.

Dimova, R., Vuillet, J., Nieoullon, A., Kerkerian-Le Goff, L., 1993.Ultrastructural features of the cholineacetyltransferase-containing neurons and relationships withnigral dopaminergic and cortical afferent pathways in the ratstriatum. Neuroscience 53, 1059–1071.

Geerling, J.C., Mettenleiter, T.C., Loewy, A.D., 2006. Viral tracersfor the analysis of neural circuits. In: Zaborszky, L.,Wouterlood, F.G., Lanciego, J.L. (Eds.), NeuroanatomicalTract-Tracing 3: Molecules, Neurons, Systems. Springer, NewYork, pp. 263–303.

Gerfen, C.R., Sawchenko, P.E., 1984. An anterogradeneuroanatomical tracing method that shows the detailedmorphology of neurons, their axons and terminals:immunohistochemical localization of an axonally transportedplant lectin, Phaseolus vulgaris leucoagglutinin (PHA-L). BrainRes. 290, 219–238.

Gerfen, C.R., Engber, T.M., Mahan, L.C., Susel, Z., Chase, T.N.,Monsma Jr., F.J., Sibley, D.R., 1990. D1 and D2 dopamine

57B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

receptor-regulated gene expression of striatonigral andstriatopallidal neurons. Science 250, 1429–1432.

Gonzalo, N., Moreno, A., Erdozain, M.A., García, P., Vázquez, A.,Castle, M., Lanciego, J.L., 2001. A sequential protocol combiningdual neuroanatomical tract-tracing with the visualization oflocal circuit neurons within the striatum. J. Neurosci. Methods111, 59–66.

Inoué, S., 1995. Foundations of confocal scanned imaging in lightmicroscopy, In: Pawley, J.B. (Ed.), Handbook of BiologicalConfocal Microscopy, 2nd edn. Plenum, New York, pp. 1–17.

Jorritsma-Byham, B., Witter, M.P., Wouterlood, F.G., 1994.Combined anterograde tracing with biotinylated dextranamine, retrograde tracing with fast blue and intracellular fillingof neurons with lucifer yellow: an electron microscopicmethod. J. Neurosci. Methods 52, 153–160.

Kachidian, P., Vuillet, S.R., Nieullon, A., Lafaille, G., Kerkerian-LeGoff, L., 1996. Striatal neuropeptide Y neurons are not a targetfor thalamic afferent fibres. NeuroReport 7, 1665–1669.

Kawaguchi, Y., Wilson, C.J., Emson, P.C., 1991. Projection subtypesof rat neostriatal matrix cells revealed by intracellular injectionof biocytin. J. Neurosci. 10, 3421–3438.

Kawaguchi, Y., Wilson, C.J., Augood, S.J., Emson, P.C., 1995. Striatalinterneurons: chemical, physiological and morphologicalcharacterization. Trends Neurosci. 18, 527–535.

Kelly, R.M., Strick, P.L., 2000. Rabies as a transneuronal tracer ofcircuits in the central nervous system. J. Neurosci. Methods103, 63–71.

Kelly, R.M., Strick, P.L., 2003. Cerebellar loops with motor cortexand prefrontal cortex of a nonhuman primate. J. Neurosci. 23,8432–8444.

Kincaid, A.E., Penney Jr., J.B., Young, A.B., Newmann, S.W., 1991.The globus pallidus receives a projection from theparafascicular nucleus in the rat. Brain Res. 553, 18–26.

Koos, T., Tepper, J.M., 1999. Inhibitory control of neostriatalprojection neurons by GABAergic interneurons. Nat. Neurosci.2, 467–472.

Lanciego, J.L., Wouterlood, F.G., 2006. Multiple neuroanatomicaltract-tracing: approaches for multiple tract-tracing. In:Zaborszky, L., Wouterlood, F.G., Lanciego, J.L. (Eds.),Neuroanatomical Tract-Tracing 3: Molecules, Neurons,Systems. Springer, New York, pp. 336–365.

Lapper, S.R., Bolam, J.P., 1992. Input from the frontal cortexand the parafascicular nucleus to cholinergic interneuronsin the dorsal striatum of the rat. Neuroscience 51,533–545.

Lapper, S.R., Smith, Y., Sadikot, A.F., Parent, A., Bolam, J.P.,1992. Cortical input to parvalbumin-immunoreactiveneurons in the putamen of the squirrel monkey. Brain Res.580, 215–224.

Meador-Woodruff, J.H.,Mansour,A., Healy,D.J., Kuehn, R., Zhou,Q.Y.,Bunzow, J.R., Akil, H., Civelli, O.,Watson Jr., S.J., 1991. Comparisonof the distribution of D1 and D2 dopamine receptormRNAs in ratbrain. Neuropsychopharmacology 5, 231–242.

Miyachi, S., Lu, X., Inoue, S., Iwasaki, T., Koike, S., Nambu, A.,Takada, M., 2005. Organization of multisynaptic inputs fromprefrontal cortex to primary motor cortex as revealed byretrograde transneuronal transport of rabies virus. J. Neurosci.25, 2547–2556.

Nassi, J.J., Callaway, E.M., 2006. Multiple circuits relaying primateparallel visual pathways to the middle temporal area.J. Neurosci. 26, 12789–12798.

Nassi, J.J., Callaway, E.M., 2007. Specialized circuits from primaryvisual cortex to V2 and area MT. Neuron 55, 799–808.

Parent, A., Charara, A., Pinault, D., 1995. Single striatofugal axonsarborizing in both pallidal segments and in the substantianigra in primates. Brain Res. 698, 280–284.

Parent, A., Sato, F., Wu, Y., Gauthier, J., Lévesque, M., Parent, M.,2000. Organization of the basal ganglia: the importance ofaxonal collateralization. Trends Neurosci. 23, S20–S27.

Pawley, J.B., 1995. Fundamental limits in confocal microscopy, In:Pawley, J.B. (Ed.), Handbook of Biological Confocal Microscopy,2nd edn. Plenum, New York, pp. 19–37.

Raux, H., Iseni, F., Lafay, F., Blondel, D., 1997. Mapping ofmonoclonal antibody epitopes of the rabies virus P protein.J. Gen. Virol. 78, 119–124.

Rajakumar, N., Elisevich, K., Flumerfelt, B.A., 1993. Biotinylateddextran: a versatile anterograde and retrograde neuronaltracer. Brain Res. 607, 47–53.

Rosene, D.L., Roy, N.J., Davis, B.J., 1986. A cryoprotection methodthat facilitates cutting frozen sections of whole monkey brainfor histological and histochemical processing without freezingartifact. J. Histochem. Cytochem. 34, 1301–1316.

Rudkin, T.M., Sadikot, A.F., 1999. Thalamic input toparvalbumin-immunoreactive GABAergic interneurons:organization in normal striatum and effect of neonataldecortication. Neuroscience 88, 1165–1175.

Shi, C., Cassell, M.D., 1993. Combination of intracellular staining ofretrogradely labeled neurons and anterograde fluorescenttracing: use of the confocal laser scanning microscope.J. Neurosci. Methods 47, 23–31.

Sesack, S.R., Miner, L.A.H., Omelchenko, N., 2006. Preembeddingimmunoelectron microscopy: Applications for studiesof the nervous system. In: Zaborszky, L., Wouterlood, F.G.,Lanciego, J.L. (Eds.), Neuroanatomical Tract-Tracing3:Molecules,Neurons, Systems. Springer, New York, pp. 6–71.

Sidibé, M., Smith, Y., 1996. Differential synaptic innervation ofstriatofugal neurons projecting to the internal or externalsegments of the globus pallidus by thalamic afferents in thesquirrel monkey. J. Comp. Neurol. 365, 445–465.

Sidibé, M., Smith, Y., 1999. Thalamic inputs to striatalinterneurons in monkeys: synaptic organization andco-localization of calcium binding proteins. Neuroscience 89,1189–1208.

Stephens, B., Mueller, A.J., Shering, A.F., Hood, S.H., Taggart, P.,Arbuthnott, G.W., Bell, J.E., Kilford, L., Kingsbury, A.E., Daniel,S.E., Ingham, C.A., 2005. Evidence of a breakdown ofcorticostriatal connections in Parkinson's disease.Neuroscience 132, 741–754.

Surmeier, D.J., Eberwine, J., Wilson, C.J., Cao, Y., Stefani, A., Kitai,S.T., 1992. Dopamine receptor subtypes colocalize in ratstriatonigral neurons. Proc. Natl. Acad. Sci. U. S. A. 89,10178–10182.

Surmeier, D.J., Reiner, A., Levine, M.S., Ariano, M.A., 1993. Areneostriatal dopamine receptors co-localized? Trends Neurosci.16, 299–305.

Ugolini, G., 1995. Specificity of rabies virus as a transneuronaltracer of motor networks: transfer from hypoglossalmotorneurons to connected second-order and higher ordercentral nervous system cell groups. J. Comp. Neurol. 356,457–480.

Veenman, C.L., Reiner, A., Honig, M.G., 1992. Biotinylated dextranamine as an anterograde tracer for single- and double-labelstudies. J. Neurosci. Methods 41, 239–254.

Weiner, D.M., Levey, A.I., Sunahara, R.K., Niznik, H.B., O'Dowd, B.F.,Seeman, P., Brann, M.R., 1991. D1 and D2 dopamine receptormRNA in rat brain. Proc. Natl. Acad. Sci. U. S. A. 88,1859–1863.

Wickersham, I.R., Finke, S., Conzelmann, K.K., Callaway, E.M.,2007a. Retrograde neuronal tracing with a deletion-mutantrabies virus. Nature Meth. 4, 47–49.

Wickersham, I.R., Lyon, D.C., Barnard, R.J.O., Mori, T., Finke, S.,Conzelmann, K.K., Young, J.A.T., Callaway, E.M., 2007b.Monosynaptic restriction of trans-synaptic tracing from single,genetically targeted neurons. Neuron 53, 639–647.

Wouterlood, F.G., 2006. Combined fluorescence methods todetermine synapses in the light microscope: multilabelconfocal laser scanning microscopy. In: Zaborszky, L.,Wouterlood, F.G., Lanciego, J.L. (Eds.), Neuroanatomical

58 B R A I N R E S E A R C H 1 2 2 1 ( 2 0 0 8 ) 4 9 – 5 8

Tract-Tracing 3: Molecules, Neurons, Systems. Springer, NewYork, pp. 394–435.

Wouterlood, F.G., Jorritsma-Byham, B., Goede, P.H., 1990.Combination of anterograde tracing with Phaseolusvulgaris-leucoagglutinin, retrograde fluorescent tracing andfixed-slice intracellular injection of lucifer yellow. J. Neurosci.Methods 33, 207–217.

Wouterlood, F.G., Goede, P.H., Arts, M.P., Groenewegen, H.J., 1992.Simultaneous characterization of efferent and afferentconnectivity, neuroactive substances, and morphology ofneurons. J. Histochem. Cytochem. 40, 457–465.

Wouterlood, F.G., Vinkenoog, M., van den Oever, M., 2002. Tracingtools to resolve neural circuits. Network 13, 327–342.

Wouterlood, F.G., Böckers, T., Witter, M.P., 2003. Synaptic contactsbetween identified neurons visualized in the confocallaserscanning microscope. Neuroanatomical tracing combinedwith immunofluorescence detection of post-synaptic densityproteins and target neuron-markers. J. Neurosci. Methods 128,129–142.

Wu, Y., Richard, S., Parent, A., 2000. The organization of the striataloutput system: a single-cell juxtacellular labeling study in therat. Neurosci. Res. 38, 49–62.