Exp Brain Res (2005) 161: 47–61DOI 10.1007/s00221-004-2045-4

RESEARCH ARTICLE

J. E. Brown . J. P. Card . B. J. Yates

Polysynaptic pathways from the vestibular nuclei to the lateralmammillary nucleus of the rat: substrates for vestibular inputto head direction cells

Received: 30 January 2004 / Accepted: 25 June 2004 / Published online: 3 November 2004# Springer-Verlag 2004

Abstract The activity of some neurons in the lateralmammillary nucleus (LMN) of the rat corresponds withthe animal’s current head direction (HD). HD cells havebeen studied extensively but the circuitry responsible forthe generation and maintenance of the HD signal has notbeen established. The present study tested the hypothesisthat a polysynaptic pathway connects the vestibular nucleiwith the LMN via one or more relay nuclei. This circuitrycould provide a substrate for the integration of sensoryinput necessary for HD cell activity. This hypothesis isbased upon the prior demonstration that labyrinthectomyabolishes HD selectivity in thalamic neurons. Viraltransneuronal tracing with pseudorabies virus (PRV) wasused to test this hypothesis. We injected recombinants ofPRV into the LMN and surrounding nuclei of adult malerats and defined the patterns of retrograde transneuronalinfection at survival intervals of 60 and 72 h. Infectedmedial vestibular neurons (MVN) were only observed atthe longest postinoculation interval in animals in whichthe injection site was localized largely to the LMN. Robust

infection of the dorsal tegmental nucleus (DTN) andnucleus prepositus hypoglossi (PH) in these cases, but notin controls, at both survival intervals identified thesenuclei as potential relays of vestibular input to the LMN.These data are consistent with the conclusion thatvestibular information that contributes to the LMN HDcell activity is relayed to this caudal hypothalamic cellgroup via a polysynaptic brainstem circuit.

Keywords Pseudorabies virus . Navigation . Spatialcognition . Hypothalamus . Tegmentum . Orientation

Introduction

Neural activity in several areas of the rat brain correspondswith the animal’s current head direction. “Head direction(HD) cells” have been studied primarily in the post-subiculum (PoS) (Taube et al. 1990), anterior dorsalthalamus (ADN) (Taube 1995), and lateral mammillarynucleus (LMN) (Blair et al. 1998; Stackman and Taube1998). Neurons exhibiting direction-related activity havealso been observed in the lateral dorsal thalamic nucleus(Mizumori and Williams 1993), dorsal striatum (Weiner1993), and retrosplenial cortex (Chen et al. 1994). Theactivity of rat HD cells varies as a function of the animal’shead direction in the horizontal plane and is independentof the rat’s location in the environment as well as ongoingbehavior.

Extensive characterization of head direction cell activityhas been performed (see Sharp et al. 2001a and Taube1998 for review) and evidence suggests that that thesecells participate in navigation and spatial cognition(Muller et al. 1996). The circuitry responsible for theoverall generation and maintenance of the HD signal hasyet to be determined, although some relationships betweenareas containing HD cells have been characterized (seeTaube 1998 for review). For example, the HD signal in theLMN is thought to play a role in generating HD cellactivity in the ADN. Blair and colleagues (1999) foundthat bilateral lesions of the LMN resulted in the loss of the

J. E. Brown . J. P. Card . B. J. YatesDepartment of Neuroscience, University of Pittsburgh,Pittsburgh, PA 15260, USA

J. E. BrownCenter for the Neural Basis of Cognition, University ofPittsburgh,Pittsburgh, PA 15260, USA

B. J. Yates (*)Department of Otolaryngology, Eye and Ear Institute, Room106, University of Pittsburgh,203 Lothrop Street,Pittsburgh, PA 15213, USAe-mail: byates@pitt.eduTel.: +1-412-6479614Fax: +1-412-6470108

Present address:J. E. BrownDepartment of Psychological and Brain Sciences, DartmouthCollege,6207 Moore Hall,Hanover, NH 03755, USA

directional specificity of ADN HD cells. This resultindicated that the LMN HD signal is necessary for ADNHD activity. Additionally, cells corresponding to theanimal’s angular head velocity (AHV) have been foundin the LMN (Blair et al. 1998; Stackman and Taube 1998)and are likely to be involved in pathways supporting aninternal representation of head direction.

Several sensory signals are thought to participate inshaping the firing patterns of HD cells, including vestib-ular, visual, and proprioceptive information (Brown et al.2002). A recent lesion study by Stackman and Taube(1997) demonstrated that elimination of labyrinthineinputs abolishes the directional sensitivity of HD cells inthe ADN. Given that the vestibular complex is a first-orderobligate synapse for labyrinthine input, these resultssuggest that the vestibular system plays an essential rolein generating HD cell activity. Furthermore, a relativelydirect polysynaptic pathway via the dorsal tegmentalnucleus of Gudden (DTN) has been proposed that couldrelay vestibular signals to the LMN (Bassett and Taube2001; Brown et al. 2002; Sharp et al. 2001b). Thisproposed pathway was developed from anatomical workusing monosynaptic tracers showing that regions of thevestibular nuclei project to DTN, and that some DTNneurons project to LMN (Allen and Hopkins 1989;Hayakawa and Zyo 1985; Liu et al. 1984; Shibata1987). However, there currently is no direct evidence toshow that vestibular nucleus neurons project to tegmentalneurons that in turn project to areas containing a largenumber of HD cells. Such evidence can only be gatheredwith the use of a transneuronal tracer that is transmittedsequentially across synaptically linked cells that comprisea neural circuit.

The present study tested the hypothesis that a polysyn-aptic pathway connects the vestibular nuclei with theLMN via one or more relay nuclei. This hypothesis wastested using retrograde transneuronal viral replication andtransport to define the synaptology of neural circuitsimpinging upon the LMN.

Methods and materials

All procedures in this study conformed to the NationalInstitutes of Health Guide for the Care and Use ofLaboratory Animals and were approved by the Universityof Pittsburgh’s Institutional Animal Care and UseCommittee. Fifty-two adult male Sprague-Dawley ratswere used in these experiments. They were given food andwater ad libitum and were kept in paired housing on a 12-h/12-h light/dark cycle. The characteristics of the tworecombinants of the Bartha strain of pseudorabies virus(PRV) employed in this study, PRV-BaBlu and PRV-152,have been published elsewhere (Billig et al. 2000). Bothviruses were the generous gift of Dr. Lynn Enquist(Princeton University, N.J., USA). PRV-BaBlu expressesβ-galactosidase (β-gal), and PRV-152 expresses enhancedgreen fluorescent protein (EGFP), under the gG andcytomegalovirus immediate early gene promoters, respec-

tively. Both recombinants were grown in pig kidney(PK15) cells and were adjusted to a final concentration of1×108 plaque-forming units/ml.

Two sets of experiments were performed. Initially, PRV-BaBlu was injected into the LMN of 12 rats in order tooptimize the volume and placement of injections and toestablish the time course of the progression of infectionproduced by each recombinant. Post-injection survivaltimes of 50, 60, and 72 h were found to result in theinfection of first-order, and putative second- and third-order neurons, respectively. Once these data weregathered, the synaptic organization of circuits projectingto the LMN was characterized in 40 animals by injectingPRV-BaBlu into the left LMN, and PRV-152 into the rightLMN, of each animal. Because PRV-BaBlu and PRV-152express unique reporters, this paradigm allowed us toobtain two data sets from each case. This was possiblebecause of the unique reporters that could be localizedimmunocytochemically in separate sets of sections (seebelow for more detail).

Surgical procedures

Animals were housed, injected with PRV, and euthanizedwithin a biosafety level 2 (BSL-2) facility. Animals wereacclimated to the facility for at least 1 week prior tosurgery, which was performed using aseptic techniques.Rats were deeply anesthetized with 1–2% isoflurane, andfixed in a stereotaxic frame. The scalp was incised andretracted, and small craniotomies were made over the leftand right LMN using a Dremel drill. An injection of PRVwith a volume of 50–100 nl was made into the left andright LMN using a 1-µl Hamilton syringe equipped with a32-gauge beveled-tip needle. Injections were placed at thefollowing stereotaxic coordinates: AP −4.65, ML ±1.10,DV −9.45 (Paxinos and Watson 1998), with the opening ofthe beveled needle oriented toward the LMN. Theinjections were made at the rate of 10 nl/min, and thesyringe was left in place for 10 min following injection toensure that virus would be less likely to move up theinjection tract following removal of the needle. Followingthe removal of the needle, each craniotomy was pluggedwith bone wax, the scalp was sutured, and rats werereturned to their home cages to recover from anesthesia.Analgesia was provided by 3 mg/kg intramuscularinjections of ketoprofen at 12-h intervals after surgery.Following the designated survival times, rats were deeplyanesthetized with 50 mg/kg intraperitoneal injections ofsodium pentobarbital and perfused transcardially with 0.5 lof 9% saline followed by 1 l of 4% paraformaldehyde-lysine-periodate (PLP) fixative (McLean and Nakane1974). The brains were removed, postfixed for 2–4 h inPLP, and cryoprotected for 2 days in 20% phosphatebuffered sucrose. Postfixation and cryoprotection weredone at 4°C.

48

Tissue processing and immunohistochemicalprocedures

Brains were sectioned at 35 µm in the coronal plane usinga freezing microtome, and sections were collectedsequentially in six wells of cryopreservant (Watson et al.1986). Cryoprotected sections were stored at −20°C untilthey were processed for immunohistochemical localizationof infected neurons. Infected neurons were identified usinga polyclonal antiserum generated in rabbit against acetone-inactivated PRV (Card et al. 1990) that identifies allrecombinants, or with antibodies specific for the uniqueprotein reporters of the two recombinants. These includeda mouse monoclonal antibody that recognized β-gal(1:1500; Sigma Chemical, St. Louis, Mo., USA) expressedby PRV-BaBlu and a rabbit polyclonal antibody thatrecognized EGFP (1:1000; Molecular Probes, Eugene,Ore., USA) expressed by PRV-152. These antigens werelocalized in alternate sections using the avidin-biotinmodification of the peroxidase-antiperoxidase procedure(Hsu et al. 1981), affinity purified secondary antibodies(Jackson ImmunoResearch Labs, West Grove, Pa., USA),and Vectastain reagents (Vector Laboratories, Burlingame,Calif., USA). Following immunohistochemical proces-sing, sections were mounted on gelatin-coated slides,dehydrated, cleared and coverslipped using Cytoseal 60(VWR Scientific, West Chester, Pa., USA).

Tissue analysis

Processed tissue sections were examined and photo-graphed with a Zeiss Axioplan photomicroscope. Imageswere digitized using a Hamamatsu digital camera(Hamamatsu Photonics, Hamamatsu, Japan) and a Sim-ple-32 PCI image analysis system (Compix, LakeOswego, Ore., USA). Mapping of the distribution ofinfected neurons in sections at a frequency of 210 µmthrough the rostrocaudal extent of the forebrain andbrainstem was accomplished using Stereo Investigatorsoftware (MicroBrightField, Williston, Vt., USA). In orderto test our hypothesis, this study focused primarily onbrainstem projections to the mammillary complex.

There are no documented direct projections from thevestibular nuclei to LMN (Matesz et al. 2002; Shibata1987). We analyzed the patterns of infection throughoutthe entire brainstem, but focused upon areas that receivevestibular input and have previously been shown toprovide projections to the mammillary complex. We choseto focus on labeling in DTN, laterodorsal tegmentalnucleus (LDTN), and ventral tegmental nucleus (VTN)because of their documented topographically organizedprojections to LMN, SUM, and MMN, respectively(Shibata 1987). DTN was considered as a potential relaynucleus for conveying vestibular information to LMNbecause it receives input from the vestibular nuclei (Liu etal. 1984). Additionally, two studies have demonstrated thatsome DTN neurons code for angular head velocity(Bassett and Taube 2001; Sharp et al. 2001b), raising the

possibility that these neurons participate in the transfor-mation of signals reflecting angular acceleration of thehead to those reflecting head direction. These studies alsofound that that a small percentage of cells in the DTN were“classic” HD cells. Along with the anatomical connectiv-ity of DTN and LMN, these findings imply a functionalconnection between the two areas.

The supragenual nucleus (SUG), nucleus prepositushypoglossi (PH), and areas in the medullary reticularformation that receive vestibular input were also subjectsof our attention. The SUG and PH project to DTN(Hayakawa and Zyo 1985; Liu et al. 1984) and are thoughtto be a component of oculomotor pathways (Korp et al.1989). Some areas of the medullary reticular formationreceive vestibular input (Belknap and McCrea 1988;Iwasaki et al. 1999) and project to DTN (Hayakawa andZyo 1985; Liu et al. 1984). Thus, they also representpotential relays for processed vestibular sensory informa-tion to contribute to HD cell activity. Finally, the vestibularnuclei were specifically analyzed because the goal of thisstudy was to determine whether inputs from the vestibularnuclei are conveyed polysynaptically to LMN.

Results

This experiment was designed to test the hypothesis that apolysynaptic pathway connects the vestibular nuclei withthe LMN via one or more relay nuclei. The results from 27injections are presented in this analysis. Of the 11injections that included the LMN, 6 were from the 60-hsurvival time and 5 were from the 72-h survival time.Sixteen cases served as controls. In these cases, theinjection was principally in an area adjacent to LMN. Tencontrol cases were from the 60-h survival time and sixwere from the 72-h survival time. The injections notdescribed here were unusable in the present analysisbecause they were located outside of the target areas orthey failed to produce a productive infection. For example,any injection where the needle broke through the bottomof the brain was not used because virus spread into thesubarachnoid space.

Several criteria were used to identify the zone of viraluptake that led to productive infection of neurons.Injection sites were first identified by locating the end ofthe cannula tract in the tissue. In all cases, at least two binsof tissue were processed to obtain an accurate localizationof the cannula. Axon terminals have the highest affinityfor alpha herpesviruses (Marchand and Schwab 1986;Vahlne et al. 1980). Thus, virus is often taken up byterminals and transported retrogradely from the injectionsite. Because of this, the extent of infection around thecannula tip is not a reliable determinant of viral diffusionand uptake. Previous work has determined that the zone ofvirus uptake that leads to viral replication after an injectionof 100 nl (1×108 pfu/ml) of PRV delivered at 20 nl/min iswithin an approximate 500 µm radius of the cannula tip(Card et al. 1999; Jasmin et al. 1997). Therefore, injectionsites in this study were defined by the end of the cannula

49

tract plus a 500 µm radius surrounding that point. Thezone of uptake was biased toward the LMN through theuse and positioning of a needle with a beveled tip, so thisis a conservative estimate of viral spread. Due to the smallsize and depth of the injection targets, most injectionswere not isolated to one nucleus (such as LMN), butusually extended to two or more nuclei (such as LMN plusSUM; see Table 1).

PRV transport following injection into LMN

Data were analyzed from 11 cases in which PRV wasinjected into the LMN. These animals were killed either 60h (n=6) or 72 h (n=5) post-injection. Numerous neurons inmultiple locations were infected following these injec-tions, which reflects the extensive projections thatterminate in the mammillary complex and caudal lateralhypothalamus. Identification of infected nuclei was basedon standard atlases of the rat brain (Paxinos and Watson1998; Swanson 1998), although only labeling in areas thatare most relevant to the current study will be emphasizedhere and are presented in Table 1. The estimated zone ofPRV uptake in all of the cases was focused upon the LMN,but included adjacent nuclei to varying degrees. Forexample, the injection site in case 3 was centered in SUMbut included LMN, medial mammillary nucleus (MMN),and ventral tuberomammillary nucleus (TMv). Conver-sely, the injection site in case 4 was centered in LMN butincluded TMv, and lateral MMN. In all cases, extensiveretrograde infection of neurons that projected to the

injection site was observed. The extent of infection at eachinjection site varied according to the nuclei involved, withthe most extensive infection observed when the injectioninvolved nuclei with local circuit connections.

60-h survival time. All six cases in the 60-h survival timegroup exhibited infected cells in the DTN and laterodorsaltegmental nucleus (LDTN), which are known to havemonosynaptic projections to LMN and SUM, respectively(Hayakawa et al. 1993; Shibata 1987). DTN labeling wasalways more prominent ipsilateral to the injection site,while less extensive DTN labeling was observed contra-laterally. Labeling in LDTN followed the same trend,except in cases 4 and 6 where the infected cells wereconcentrated on the contralateral side. Figure 1 illustratesthe labeling patterns in an animal from the 60-h survivaltime group (case 5). This injection site was centered in theLMN (Fig. 1a) and light labeling was observed in LMN,SUM, and TMv. The resultant infection in DTN (Fig. 1b)was restricted to the ipsilateral side, whereas the infectedcells in LDTN were observed bilaterally with a higherconcentration on the ipsilateral side. Injection sites in cases2, 3, and 4 included the MMN, and in all three caseslabeled cells were observed in the ventral tegmentalnucleus (VTN). The VTN projects monosynaptically tothe MMN (Allen and Hopkins 1989) and is one of thetopographically organized projections from the tegmentumto the mammillary complex.

Infected cells were observed in the SUG in four of thesix cases in this group. These neurons were usually locatedbilaterally with a concentration on the ipsilateral side.

Fig. 1 Photomicrographs of aninjection site (a) and retrograd-ely infected brainstem nuclei(b–d) in a 60-h survival timeanimal (case 5). This injectionsite included the lateral mam-millary nucleus (LMN), supra-mammillary nucleus (SUM), andtuberomammillary nucleus parsventralis (TMv). a Light labelingin the left LMN and TMv;heavier labeling in SUM. b Thelaterodorsal tegmental nucleus(LDTN) was heavily labeledbilaterally along with lighterlabeling of the dorsal tegmentalnucleus (DTN) ipsilateral to theinjection site. c Light bilaterallabeling in the supragenual nu-cleus (SUG) dorsal to the genuof the facial nerve (VII). d Nolabeling was found in the pre-positus hypoglossal nuclei (PH)or medial vestibular nuclei(MVN). Marker bars=500 μm.

50

Figure 1c illustrates light SUG labeling in case 5. TheSUG is adjacent and directly anterior to the rostral extentof Prh. In cases where both nuclei were labeled, it wasnoted that there was a distinct space lacking infected cellsbetween the caudal SUG and rostral PH. PH neuronsconfined to the rostral half of the nucleus were labeled incases 2 and 3 only. No infected cells were found in thevestibular nuclei 60 h subsequent to LMN injections.Figure 1d shows that PH and MVN in case 5 were devoidof infection.

In some animals, infected neurons were also located inthe periaqueductal gray, interpeduncular nucleus, parabra-chial nucleus, locus coeruleus and reticular formation.Regions of the reticular formation, such as the mesence-phalic reticular nucleus, paragigantocellular reticular nu-cleus, gigantocellular reticular nucleus, and pontine retic-ular nucleus, exhibited more prominent labeling whenheavier infections were observed near the injection sites.For example, case 3 exhibited the most infected cells

around the injection site, as well as the largest number ofinfected cells in other regions of the brain. Conversely, thelowest number of infected cells was observed near theinjection site and elsewhere in the brain in cases 4 and 6.

72-h survival time. The locations of infected neuronsobserved in the five 72-h cases were similar to those foundin the 60-h survival time group, although labelingextended into more caudal nuclei in the longer survivaltime group. All five of these cases exhibited some labelingin the LMN as well as one or more surrounding nuclei.Figure 2 illustrates labeling in case 9, where an injectionwas placed deep in the mammillary complex between theLMN and the MMN pars lateralis (Fig. 2a). Additionallabeling near the injection site was located in TMv.

In all cases, the DTN and LDTN were found to containinfected cells bilaterally, with a concentration on the sideipsilateral to the injection site. Similar to the 60-h group,the LDTN labeling was generally heavier than the DTN

Fig. 2 Photomicrographs of aninjection site (a) and retrograd-ely infected brainstem nuclei(b–f) in a 72-h survival timeanimal (case 9). a Injection siteincluded lateral mammillarynucleus (LMN), medial mam-millary nucleus pars lateralis(MMNpl), supramammillary nu-cleus (SUM), and tuberomam-millary nucleus pars ventralis(TMv). The cannula tract can beseen descending into the mam-millary complex, and the tissuewas torn during processing. bTegmental level labeling locatedbilaterally in both the DTN andLDTN, with primarily ipsilaterallabeling in the superior centralnucleus raphe (CS). c Heavylabeling of the contralateralSUG. d Extensive labeling ofipsilateral PH, light labeling ofcontralateral PH, and equivalentbilateral labeling of MVN.Higher power photomicrographsfrom panel d of the PH (e) andMVN (f) are illustrated. Markerbars in a–d=500 μm, e=125μm, f=62.5 μm.

51

Tab

le1

Locations

ofneuron

sinfected

with

PRV

injected

into

thelateralmam

millary

nucleus(LMN)or

surrou

ndingnu

clei,follo

wingasurvival

timeof

60or

72h.

Labelingon

the

side

ipsilateral(ipsi)

and

contralateral(con

tra)

tothe

injection

site

isdesign

ated

separately.X

1–10

cells,X

X11–20cells,X

XX>20

cells,M

MNmedialmam

millarynu

cleus,

PMN

prem

ammillary

nucleus,

SUM

supram

ammillary

nucleus,

TMvtuberomam

millary

nucleus,DTNdo

rsal

tegm

entalnu

cleus,LDTNlaterodo

rsal

tegm

entalnu

cleus,VTNventral

tegm

entalnu

cleus,

MRF

medialreticular

form

ation,

SUG

supragenualnu

cleus,

PrH

prepositu

shy

poglossi,MVN

medialvestibular

nucleus

Case

Nucleiinclud

edin

injectionsite

Hindb

rain

labelin

g

LMN

MMN

PMN

SUM

TMv

DTN

LDTN

VTN

MRF

SUG

PrH

MVN

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

ipsi

contra

60-h

survival,injectioninclud

edLMN

Case1

XXX

XX

XX

XX

XX

XX

XX

XX

Case2

XX

XX

XX

XX

XX

XX

XXX

XX

Case3

XXX

XX

XXX

XXX

XX

XX

XXX

XXX

XX

XX

XX

XX

XX

XX

Case4

XXX

XX

XX

Case5

XX

XX

XX

XXX

XXX

XX

Case6

XXX

XX

XX

X72

-hsurvival,injectioninclud

edLMN

Case7

XX

XXX

XXX

XXX

XXX

XXX

XX

XXX

XX

XXX

XXX

XX

XX

XX

Case8

XX

XXX

XX

XXX

XXX

XXX

XX

XXX

XXX

XXX

XXX

XX

XX

XX

Case9

XXX

XXXX

XXX

XXX

XXX

XXXX

XXX

XXX

XXX

XX

XXXX

XXX

XX

XXXX

XX

XX

XCase10

XX

XX

XX

XX

XX

XX

XX

Case11

XXX

XX

XX

XX

XX

XX

XX

X60

-hsurvival,injectionou

tsideLMN

Case12

XXX

XXX

XXX

XXX

XX

XX

Case13

XXX

Case14

XX

XX

XX

XX

XX

XX

XCase15

XX

XX

Case16

XX

XCase17

XX

XX

XCase18

XX

XX

Case19

XXX

XCase20

XXX

XX

Case21

XX

XX

XX

XX

72-h

survival,injectionou

tsideLMN

Case22

XX

XX

XX

XX

XX

XCase23

XX

XX

XX

XCase24

XX

XX

XX

XX

XXX

XXXX

XXX

XX

Case25

XX

XX

Case26

XX

XX

XX

XCase27

XX

XX

X

52

labeling, corresponding to the extent of labeling found inthe SUM and LMN, respectively. Figure 2b provides anexample of bilateral labeling in both the DTN and LDTNfrom case 9. In this animal, infected cells in the MMNwere confined to the side of the injection.

Labeling in the SUG was observed in four of the fivecases. This labeling was always bilateral, although therelative number of infected SUG neurons ipsilateral andcontralateral to the injection site varied between animals.Figure 2c shows SUG labeling that was much heavier onthe contralateral side. Infected cells were found in the PHof all five cases, and an example of heavy PH labelingconcentrated on the ipsilateral side is illustrated in Fig. 2d.A higher magnification photomicrograph (Fig. 2e) demon-strates the density of PH infection in this case. The PHinfections in all five cases were limited to the rostral halfof the nucleus. This pattern of labeling corresponds to thePH infections observed in the 60-h survival time group,although the density of labeling was higher in the 72-hsurvival time cases.

The MVN was found to contain cells that were infectedby retrograde transneuronal transport of PRV in four of thefive 72-h cases where the injection site was localizedlargely to the LMN. This labeling was found to be bilateralin three of the four cases, with a slightly higher number ofinfected cells on the ipsilateral side. Figure 2d shows theMVN labeling found in case 9, which is shown at highermagnification in Fig. 2f. The four cases containinginfected MVN cells exhibited labeling throughout therostrocaudal extent of MVN, although there was aconcentration of infected cells in the rostral half of thenucleus. Labeled cells were not observed in the othervestibular nuclei in any of the cases. Case 11 did notexhibit any infected cells in the MVN, although the patternof infection throughout the rest of the brain was consistentwith the other 72-h survival time cases with injections thatincluded LMN. The number of infected cells in the MVNwas higher in those cases where the LMN and DTNinfections were most extensive (cases 7, 8, 9). Figure 3illustrates the locations of infected cells at several levels ofthe brain from case 9, starting at the level of the injection

Fig. 3 The distribution ofinfected neurons from the caseshown in Fig. 2 (case 9) areillustrated. Each dot representsan individual neuron. Theboundaries of the PAG areindicated to provide an anatom-ical point of reference. Sectionsare arranged from rostral (a) tocaudal (h), with the coronalplanes relative to Bregma notedat the lower right of each sec-tion. The boxed areas in a, e, g,and h define regions illustratedin Fig. 2a, 2b, 2c, and 2d,respectively.

53

site and continuing caudal to the MVN. Both the largenumber and wide distribution of infected neurons resultingfrom the retrograde transynaptic transport of PRV over 72h are evident in this figure.

As also observed in the 60-h survival time cases,infected neurons were noted in some animals in theperiaqueductal gray, interpeduncular nucleus, parabrachialnucleus, locus coeruleus, reticular formation, and otherareas. Labeling in areas of the reticular formation, such asthe mesencephalic reticular nucleus, paragigantocellularreticular nucleus, gigantocellular reticular nucleus, andpontine reticular nucleus, was more prominent when alarge number of infected cells was present at the injectionsites.

Forebrain labeling was also quantified in both the 60-hand 72-h cases and was found to be consistent with thepattern that would be predicted based on relevant studies(Alonso and Kohler 1984; Risold and Swanson 1996,1997). These studies demonstrated topographically orga-nized bisynaptic pathways between the hippocampus andhypothalamus that are relayed through the lateral septum.The analysis of forebrain data was beyond the scope of thepresent study and will be presented in a separatemanuscript.

Control injections: PRV transport following injectioninto areas surrounding LMN

Control data were analyzed from 16 cases where PRV wasplaced into areas surrounding, but not including, the LMN.

The goal of this analysis was to determine whether thebrainstem labeling observed in the LMN-injected animalswas due to PRV transport from the LMN or from adjacentnuclei. These animals were killed 60 h (n=10) or 72 h(n=6) following injections.

Figure 4 illustrates representative 60-h labeling fromcase 12, where the injection site was centered in TMv andincluded SUM and lateral hypothalamic area (Fig. 4a). TheTMv runs along the ventral boundary of the mammillarycomplex, as evidenced by the labeling shown in this case.No labeling was observed in the LMN in any sections ofthis brain, and the locations of infected neurons caudal tothe injection site were different from those in cases wherethe injection site included LMN (see Table 1). Figure 5illustrates a 72-h control case (case 24), where labelingaround the injection site was observed in the TMv, SUM,lateral hypothalamic area, ventral tegmental area (VTA),and substantia nigra pars compacta (SNc), but not theLMN. The extensive infection in VTA and SNc lateral tothe injection site presumably resulted from the intercon-nectivity within the nuclei and extensive axonal termina-tions near the injection site, similar to the TMv infectionshown in Fig. 4a. Figure 6 illustrates the locations ofinfected cells from case 24 at several levels of the brain.

In general, the labeling of DTN in the control casesdiffered dramatically from the DTN labeling observed inthe LMN-injected cases. Infected DTN cells in the controlgroup were observed only in two of the ten control casesin the 60-h survival time group (17 and 20) and in none ofthe cases in the 72-h survival group. These two controlcases exhibited very light, ipsilateral DTN infection. This

Fig. 4 Photomicrographs il-lustrating a control injection site(a) and the associated retrogradeinfection of brainstem nuclei (b–d) in case 12 with a 60-hsurvival time. a This injectionsite was focused within thesupramammillary nucleus(SUM) and tuberomammillarynucleus pars ventralis (TMv) andonly slightly impinged upon thelateral mammillary nucleus(LMN). Labeling in TMv ex-tends lateral and ventral to theLMN without including it. bBilateral labeling was observedin LDTN and no infected neu-rons were present in the DTN. cVery light bilateral labeling ispresent in the SUG. d Nolabeling was found in PH orMVN. Marker bars=500 μm.

54

differed from the LMN-injected animals where DTNlabeling was found in all of the cases, the number ofinfected cells was much higher, and the infected cells werefound bilaterally in 9 of the 11 rats. The LDTN wasconsistently infected in the both the 60-h and 72-h controlcases where the injection sites included SUM. This patternwas similar to the labeling observed in the LMN injectioncases that additionally included SUM. LDTN labeling wasusually bilateral with a greater number of cells found onthe ipsilateral side. Figure 4b demonstrates the bilateralinfection in LDTN from case 12 as well as the lack oflabeled cells in the DTN. Infected cells in the VTN wereobserved in two control cases (21 and 26); the injectionsites in both animals included the MMN. Figure 5billustrates the pattern of infection at the level of DTN andLDTN following a 72-h survival time.

The SUG was labeled in two cases from the 60-hcontrol group (cases 12 and 14). Both displayed bilateralSUG labeling with higher numbers of infected cells on theipsilateral side. There was no clear pattern of infectionaround the injection sites of these cases that set them apartfrom the rest of the control group. Figure 4c illustrates thelight labeling in the SUG from case 12. Infected cells inthe SUG of the 72-h control group were observed in fourof the six cases, also with the heaviest labeling found onthe ipsilateral side. Figure 5c illustrates the labeling inSUG from case 24, which is representative of the SUGlabeling found in the other three cases.

There were no infected cells in the PH, MVN, or othervestibular nuclei of any of the control cases. Figure 4dillustrates the lack of labeling in rostral PH, as well as the

absence of cells in rostral MVN in case 12. Figure 5dillustrates a similar absence of infected cells in both the PHand MVN. Infected cells observed in the reticularformation of the control cases were limited to a sparse,ipsilateral distribution in the pontine reticular formation ofcases 15 and 16, and the paragigantocellular reticularnucleus of case 22.

Discussion

Previous electrophysiological observations have demon-strated that direction-sensitive firing of HD cells in theADN is reduced or eliminated following chemical laby-rinthectomy in rats (Stackman and Taube 1997). BecauseADN HD activity is dependent on an intact LMN (Blair etal. 1999), these observations raise the possibility thatvestibular input to the LMN participates in shaping thedirectional activity of HD or AHV cells located in thisnucleus. The absence of direct projections between thevestibular complex and the LMN indicates that thesevestibular influences must be polysynaptic, but the identityand organization of projection pathways conveying laby-rinthine information to LMN is not known. The presentstudy used viral transneuronal tracing to provide insightinto the pathways through which the vestibular systemmodulates the activity of cells in LMN that are related tothe HD signal. Specifically, the data support the conclu-sion that vestibular information reaches the LMN via apolysynaptic circuitry that includes the DTN and otherpotential relays. Collectively, these data provide novel

Fig. 5 Photomicrographs of acontrol injection site (a) andassociated retrograde infectionof brainstem (b–d) in an animalwith a 72-h survival time (case24) are illustrated. a The injec-tion included the SUM, TMv,lateral hypothalamic area, ven-tral tegmental area, and sub-stantia nigra pars compacta. bBilateral labeling was apparentin LDTN but not DTN. c Lightbilateral labeling in SUG, aswell as locus coeruleus andnucleus raphe magnus are illu-strated. d No labeled cells werefound in PH or MVN. Markerbars=500 μm.

55

insights into the brainstem networks that contribute tospatial cognition and that may be relevant to an organism’sability to navigate effectively within its environment.

Technical considerations

The factors that influence the uptake and transport of PRVthrough neural circuitry following intracerebral injectionare important considerations in evaluating the conclusionspresented in this study. Prior work has demonstrated thatthe affinity of alpha herpesvirus envelope glycoproteinsfor neuronal profiles and extracellular matrix moleculesrestrict the diffusion of PRV injected intraparenchymally(see Card et al. 1999 for review). As noted earlier, detailedstudies indicate that the volume and concentration of PRVused in the present analysis is taken up within a 500 µmradius of injection (see Card 2001 for review). Never-

theless, it is clear that the architecture of the region ofinjection exerts an important influence upon the zone ofvirus uptake. Important in this regard is the demonstrationthat axon terminals exhibit the highest affinity for PRV(Marchand and Schwab 1986). Thus, the results of thepresent study must be viewed within the context of thearchitecture of the LMN and neighboring cell groups.

Previous studies report that the LMN and MMN containfew local circuit neurons and that the neuropil is composedprincipally of afferents from other regions (Allen andHopkins 1988, 1989; Gonzalo Ruiz et al. 1993; Hayakawaand Zyo 1992; Takeuchi et al. 1985). In contrast, localcircuit connections provide a more prominent componentof the supramammillary nucleus neuropil (Hayakawa et al.1994). These observations are parsimonious with thesparse infection of LMN neurons observed in cases inwhich the injection site was centered within LMN and theprevalence of infected SUM neurons in cases where the

Fig. 6 The distribution ofinfected neurons from the caseshown in Fig. 5 (case 24) areillustrated. Each dot representsan individual neuron. Theboundaries of the PAG areindicated to provide an anatom-ical point of reference. Sectionsare arranged from rostral (a) tocaudal (h), with the coronalplanes relative to Bregma notedat the lower right of each sec-tion. The boxed areas in a, e, g,and h define regions illustratedin Fig. 5a, 5b, 5c, and 5d,respectively.

56

injection involved this nucleus. The DTN provides aprominent afferent projection to the LMN (Allen andHopkins 1989; Hayakawa and Zyo 1992; Shibata 1987;Wirtshafter and Stratford 1993) and this pathway is knownto be reciprocal (Hayakawa and Zyo 1990; Groenewegenand Van Dijk 1984; Liu et al. 1984; Shibata 1987).Therefore, it is possible that PRV injected into the LMNproduced a retrograde infection of DTN neurons withsubsequent transneuronal infection of LMN neuronsprojecting to the DTN. This interpretation is consistentwith the more robust infection of LMN neurons in the 72-hsurvival times. It is also consistent with the kinetics ofviral replication and the axonal architecture (e.g., length ofaxons) that make the post-injection survival intervalsincorporated in this analysis suitable for the analysis ofthese brainstem circuits.

The conservative approach employed in our analysisidentified cases in which the injection was centered in theLMN and the orientation of the beveled needle biased theflow of virus into this subdivision of the mammillarycomplex. Nevertheless, we cannot exclude the possibilitythat PRV was taken up in neighboring cell groups in thesecases. Comparative analysis of the retrograde infectionproduced by injection of virus into neighboring cell groupsin control cases allowed us to address this issue. Such acomparison demonstrated that cases in which the injectionwas centered in the LMN always produced a retrogradeinfection of DTN whereas injections of the neighboringSUM rarely produced DTN infection but reliably infectedthe LDTN. Retrograde infection of neurons in the MVNwas present in cases with a robust infection of the DTNand did not occur in cases in which there was robustinfection of LDTN with little or no infection of DTN.These data demonstrate consistent patterns of infectionthat correlate with the location of the injection site in thecaudal hypothalamus as well as with literature demonstrat-ing differential projections of the DTN and LDTN to theLMN and SUM, respectively (Allen and Hopkins 1989;Hayakawa and Zyo 1992; Shibata 1987). When consid-ered with literature demonstrating the specificity ofretrograde transport of PRV-Bartha through synapticallylinked neurons (Pickard et al. 2002), these data stronglysupport the conclusion that the analytical approachemployed in this analysis provides a reliable means ofdefining cell groups that contribute to the polysynapticrelay of vestibular information to the LMN.

The organization of polysynaptic pathways linkingLMN and MVN

Potential relay nuclei that carry vestibular information tothe LMN can be identified by combining the patterns oflabeling observed in the present study with currentliterature describing the functional roles of areas contain-ing infected cells.

There are several tegmental nuclei that send ascendingprojections to the mammillary complex via topographi-cally organized, parallel pathways, including the projec-

tion from the DTN to the LMN mentioned previously(Allen and Hopkins 1989; Hayakawa and Zyo 1992;Wirtshafter and Stratford 1993). Our data are mostconsistent with the DTN serving as a relay of vestibularinformation to the LMN. The main LDTN projection tothe mammillary complex terminates in the SUM (Shibata1987). However, Hayakawa and Zyo (1992) providedevidence for LDTN projections to LMN followinginjections of an anterograde tracer (WGA-HRP) into theLDTN. The LMN has been found to contain a high densityof cholinergic terminals (Ruggiero et al. 1990), and theLDTN is also known to contain cholinergic neurons(Gonzalo-Ruiz et al. 1999). This suggests that the LDTNprojects to the LMN, and therefore the LDTN labeling inthe present study could partly be due to retrogradetransport from the LMN. Additionally, Groenewegen andVan Dijk (1984) showed that the DTN are reciprocallyconnected. It is therefore likely that the bilateral DTNlabeling observed in the present data was due to PRVtransport between the two DTN. Nonetheless, such aninterpretation of the current data does not detract from theargument that the present results, when combined withfunctional data discussed in the next paragraph, supportthe DTN as the probable relay in the polysynaptic circuitfrom MVN to LMN. The MMN receives input from theVTN of Gudden and the superior central nucleus(Hayakawa and Zyo 1991). LDTN and VTN labelingwas observed when the SUM and the MMN, but not theLMN, were included in the injection sites, and aretherefore less likely be participate as relay nuclei in thepolysynaptic circuit from MVN to LMN.

Electrophysiological recording data describing thefunctional activity of DTN neurons provides further strongevidence regarding their role as potential elements in thecircuit relaying vestibular signals to the LMN. Studies byboth Bassett and Taube (2001), and Sharp and colleagues(2001b), found that DTN activity in freely behaving ratswas related to their angular head velocity, which is a signalthat could be used to generate HD cell activity.Furthermore, the fact that a small percentage of HD cellswere located in DTN (Basset and Taube 2001; Sharp et al.2001b) indicates that the DTN could be performing anintegrative function regarding vestibular signals relevantto the HD system. Based on these data, in combinationwith the results of the present study, it is reasonable toconclude that the DTN participates in the relay ofvestibular information to the LMN.

It is possible that relay nuclei besides the DTNparticipate in the circuit relaying vestibular informationto HD cells in the LMN. In addition to the tegmentalnuclei, the following regions exhibited relatively heavylabeling following PRV injections that included LMN: PH,SUG, reticular formation, interpeduncular nucleus (IPN),periaqueductal gray, parabrachial nucleus, and locuscoeruleus. Whether or not these areas can be consideredto be relay nuclei in the circuit of interest can beascertained from the patterns of infection observed in thepresent study, plus their known functionality and/orconnectivity with LMN, DTN, and/or MVN.

57

Infected neurons were observed in the periaqueductalgray following both injections that included LMN andsome control cases. The periaqueductal gray projects tothe mammillary complex (Shibata 1987), but projectionsto DTN appear to be limited (Hayakawa and Zyo 1985;Liu et al. 1984). Furthermore, this region has not beendocumented to receive input from the MVN (Beitz et al.1986; Marchand and Hagino 1983). The parabrachialnuclei were heavily infected in several cases in this studywhere the injection site included LMN, but were alsolabeled in some control cases. Although the parabrachialnuclei receive vestibular inputs (Balaban et al. 2002), theydo not project to the DTN (Hayakawa and Zyo 1985; Liuet al. 1984) or LMN, and are therefore unlikely to becomponents of the circuit conveying vestibular signals toHD cells. The locus coeruleus is known to receive inputfrom PH (Ennis and Aston-Jones 1989) and the vestibularnuclei (Luppi et al. 1995), but does not project to DTN(Hayakawa and Zyo 1985; Liu et al. 1984). The IPNprojects to both the DTN and LDTN (Cornwall et al. 1990;Hayakawa and Zyo 1985; Liu et al. 1984). In the presentstudy, IPN neurons were labeled in both LMN-injectedcases and control cases; IPN cells were also infected inanimals where the LDTN, but not the DTN, was found tocontain labeling. In addition, IPN has not been demon-strated to receive vestibular inputs (Marchand et al. 1980;Shibata et al. 1986). Thus, it seems unlikely that theperiaqueductal gray, parabrachial nucleus, locus coeruleus,or IPN play a primary role in relaying vestibular signals toDTN neurons that in turn project to LMN. The nucleusincertus contained infected neurons in some of our animalsand is known to provide ascending projections to thenumerous areas including the LMN (Goto et al. 2001;Olucha-Bordonau et al. 2003). The nucleus incertusreceives input from other areas labeled transynapticallyin this study including the IPN, PAG, and superior centralnucleus. However, it is unlikely that this nucleus plays arole in transmitting vestibular information as it does notreceive input from the vestibular nuclei or dorsal tegmen-tum (Goto et al. 2001).

In contrast, PH both sends projections to the ipsilateralDTN (Hayakawa and Zyo 1985; Liu et al. 1984) andreceives afferents from vestibular nuclei (Iwasaki et al.1999), but does not project to LMN (Shibata 1987).Furthermore, PH was heavily infected following injectionsof PRV into LMN, but not adjacent nuclei. It is thereforereasonable to consider PH as a potential relay forascending vestibular signals to LMN neurons, as Bassettand Taube (2001) have predicted. Figure 7 illustrates thepossible position of PH in the pathway, as it receivesvestibular signals from MVN and also projects to DTN.However, PH receives both vestibular and optokineticinformation (Lannou et al. 1984), as well as oculomotorsignals (Cannon and Robinson 1987; McCrea and Baker1985); the rostral portion of the nucleus, which containedthe neurons infected following injections of PRV intoLMN in this study, is known to be a component ofpathways that mediate horizontal eye movements (McCrea1988). Thus, it is alternatively possible that PH neurons

that provide inputs to DTN integrate signals unrelated tothe downstream generation of a head direction signal; thefunctional role of these neurons in the circuit is yet to bedetermined.

The mesencephalic reticular nucleus, pontine reticularformation, and gigantocellular reticular nucleus were oftenheavily labeled in cases where the injection site wascentered in the LMN. Portions of the medial reticularformation receive substantial vestibular inputs (Matesz etal. 2002; Zelman et al. 1984), and their previouslyreported connectivity with DTN (Liu et al. 1984) issimilar to that found in the present study. However, therewas no clear pattern of infection in the reticular formationthat resulted from our injections. Thus, some parts of thereticular formation are potential components of thepathway relaying vestibular signals to LMN, althoughfurther experiments will be necessary to test thispossibility.

The SUG has often been classified as one of theperihypoglossal nuclei or a rostral extension of PH (Brodal1983; Korp et al. 1989). As a result, its synaptology andfunctionality are often unclear in the literature. Thisnucleus is known to project to the contralateral DTN(Hayakawa and Zyo 1985; Liu et al. 1984). Although aprevious tracing study did not demonstrate the presence ofefferents from the MVN to SUG in the rat (Matesz et al.2002), further work is necessary to fully consider oreliminate SUG as a potential relay nucleus for ascendingvestibular signals to HD cells. Figure 7 reflects thecontribution that areas of the reticular formation and SUGmay play in relaying labyrinthine inputs to LMN.

The only vestibular nucleus that contained infected cellsafter PRV injection into LMN was the MVN, which waslabeled in four of the five cases following a 72-h survivaltime. It is possible that the temporal progression of PRV

Fig. 7 Schematic representation of postulated connections betweennuclei that contribute to the HD signal (in bold) based on thelabeling in the present study. Other areas that may potentially relayvestibular information to the HD pathway include the reticularformation and supragenual nucleus. Connections between tegmentalnuclei and mammillary nuclei are shown to provide an illustration ofthe topographical organization of projections between those regions.

58

replication and transport from LMN in case 11 wasinsufficient to generate infection of MVN neurons.Nonetheless, our results substantiate the hypothesis thatthe vestibular system projects to the LMN via a relativelydirect polysynaptic pathway that, based on the currentdata, most likely includes the DTN as a primary relaynucleus. Previous studies have demonstrated directprojections from MVN to DTN (Hayakawa and Zyo1985; Liu et al. 1984). The present study also confirmsthat the LMN does not receive direct input from vestibularnuclei (Shibata 1987). Therefore, our results are consistentwith the scheme illustrated in Fig. 7, where vestibularsignals from the MVN are being relayed through the DTNto the LMN. It remains to be determined, however,whether direct projections from MVN to DTN play thepredominant role in relaying vestibular signals to HDcells, or whether connections that involve PH, the reticularformation, or perhaps SUG are functionally important aswell.

Functional implications of ascending vestibularprojections to the LMN for the generation of the HDsignal

One of the inputs to MVN is from the horizontalsemicircular canals, which generate angular head acceler-ation signals; many cells in MVN also receive convergentcanal and otolith input (Schor et al. 1998). Although HDcells in the LMN code for head direction, there is evidencethat AHV cells are also located in the LMN as well (Blairet al. 1998; Stackman and Taube 1998). For vestibularinformation to be used in generating the HD signal, theangular head acceleration signals would have to beintegrated twice over time to produce a position signal.Such neural integration could be accomplished overseveral steps that have been discussed previously,including routing the signal through cells that extract anAHV signal (Bassett and Taube 2001; Brown et al. 2002;Sharp et al. 2001b). Similar neural integration takes placein computing eye position signals for the vestibulo-ocular(VOR) reflex (Baker 1977). The relationship betweenLMN AHV and HD cells is unclear, and the results of thisstudy cannot indicate whether AHV or HD cells arereceiving input from the polysynaptic pathway from theMVN. It is likely that this input influences both sets ofcells in support of the HD signal.

Furthermore, it is probable that there is additional inputor processing in the HD pathway because the HD signal inLMN is not a mere “head position” signal. HD cells codefor head direction in the vertical plane, regardless of therelationship of the rat’s head with its trunk (Taube 1995).In addition, Stackman et al. (2000) and Kim et al. (2003)have provided evidence that, under some conditions, ADNHD cell activity remains stable in the vertical plane.Therefore, it is probable that other vestibular information,such as otolith signals, could play an important role in themaintenance of the HD signal. The nature of HD cellactivity in a true three-dimensional environment is yet to

be determined. Finally, vestibular neurons that respond tohead rotation also respond to visual field rotation (Henn etal. 1974), and the role that optokinetic stimuli play in thegeneration of HD activity, if any, is unclear.

Summary

The present data indicate that the MVN projects to theLMN via a polysynaptic pathway that involves the DTNand perhaps other nuclei. The anatomical and functionalrelationship of the HD signal with vestibular, visual, andoptokinetic information appears to be quite complex.Additional work must be performed to differentiate theindividual contributions of these types of information tothe generation and maintenance of the HD signal.

Acknowledgements The authors thank Lucy Cotter, Jen-ShewYen, Rebecca Edelmeyer, Katie Wilkinson, Andrew Maurer andBrian Sadacca for valuable technical assistance. This work wassupported by grant R21-DC006049 from the National Institutes ofHealth.

References

Allen GV, Hopkins DA (1988) Mammillary body in the rat: acytoarchitectonic, Golgi, and ultrastructural study. J CompNeurol 275:39–64

Allen GV, Hopkins DA (1989) Mammillary body in the rat:topography and synaptology of projections from the subicularcomplex, prefrontal cortex and midbrain tegmentum. J CompNeurol 286:311–336

Alonso A, Kohler C (1984) A study of reciprocal connectionsbetween the septum and entorhinal area using anterograde andretrograde axonal transport methods in the rat brain. J CompNeurol 225:327–343

Baker R (1977) The nucleus prepositus hypoglossi. In: Brooks BA,Bajandas FJ (eds) Eye movements. Plenum, New York, pp145–178

Balaban CD, McGee DM, Zhou J, Scudder CA (2002) Responses ofprimate caudal parabrachial nucleus and Kolliker-fuse nucleusneurons to whole body rotation. J Neurophysiol 88:3175–3193

Bassett JP, Taube JS (2001) Neural correlates for angular headvelocity in the rat dorsal tegmental nucleus. J Neurosci21:5740–5751

Beitz AJ, Clements HR, Mullett MA, Ecklund LJ (1986) Differen-tial origin of brainstem serotonergic projections to the midbrainperiaqueductal gray and superior colliculus of the rat. J CompNeurol 250:498–509

Belknap DB, McCrea RA (1988) Anatomical connections of theprepositus and abducens nuclei in the squirrel monkey. J CompNeurol 268:12–38

Billig I, Foris JM, Enquist LW, Card JP, Yates BJ (2000) Definitionof neuronal circuitry controlling the activity of phrenic andabdominal motoneurons in the ferret using recombinant strainsof pseudorabies virus. J Neurosci 20:7446–7454

Blair HT, Cho J, Sharp PE (1998) Role of the lateral mammillarynucleus in the rat head direction circuit: a combined single unitrecording and lesion study. Neuron 21:1387–1397

Blair HT, Cho J, Sharp PE (1999) The anterior thalamic head-direction signal is abolished by bilateral but not unilaterallesions of the lateral mammillary nucleus. J Neurosci 19:6673–3383

Brodal A (1983) The perihypoglossal nuclei in the macaque monkeyand the chimpanzee. J Comp Neurol 218:257–269

59

Brown JE, Yates BJ, Taube JS (2002) Does the vestibular systemcontribute to head direction cell activity in the rat? PhysiolBehav 77:743–748

Cannon SC, Robinson DA (1987) Loss of the neural integrator ofthe oculomotor system from brain stem lesions in monkey. JNeurophysiol 57:1383–1409

Card JP (2001) Pseudorabies virus neuroinvasiveness: a windowinto the functional organization of the brain. Adv Virus Res56:39–71

Card JP, Enquist LW, Moore RY (1999) Neuroinvasiveness ofpseudorabies virus injected intracerebrally is dependent on viralconcentration and terminal field density. J Comp Neurol407:438–452

Card JP, Rinaman L, Schwaber JS, Miselis RR, Whealy ME,Robbins AK, Enquist LW (1990) Neurotropic properties ofpseudorabies virus: uptake and transneuronal passage in the ratcentral nervous system. J Neurosci 10:1974–1994

Chen LL, Lin LH, Barnes CA, McNaughton BL (1994) Head-direction cells in the rat posterior cortex. I. Anatomicaldistribution and behavioral modulation. Exp Brain Res101:24–34

Cornwall J, Cooper JD, Phillipson OT (1990) Afferent and efferentconnections of the laterodorsal tegmental nucleus in the rat.Brain Res Bull 25:271–284

Ennis M, Aston-Jones G (1989) Potent inhibitory input to locuscoeruleus from the nucleus prepositus hypoglossi. Brain ResBull 22:793–803

Gonzalo-Ruiz A, Sanz-Anquela JM, Spencer RF (1993) Immuno-histochemical localization of GABA in the mammillary com-plex of the rat. Neuroscience 54:143–156

Gonzalo-Ruiz A, Morte L, Flecha JM, Sanz JM (1999) Neurotrans-mitter characteristic of neurons projecting to the supramammil-lary nucleus of the rat. Anat Embryol 200:377–392

Goto Y, Swanson LW, Canteras NS (2001) Connections of thenucleus incertus. J Comp Neurol 438:86–122

Groenewegen HJ, Van Dijk CA (1984) Efferent connections of thedorsal tegmental region in the rat, studied by means ofanterograde transport of the lectin Phaseolus vulgaris-leucoag-glutinin (PHA-L). Brain Res 304:367–371

Hayakawa T, Ito H, Zyo K (1993) Neuroanatomical study of afferentprojections to the supramammillary nucleus of the rat. AnatEmbryol (Berl) 188:139–148

Hayakawa T, Ito H, Zyo K (1994) Fine structure of thesupramammillary nucleus of the rat: analysis of the ultrastruc-tural character of dopaminergic neurons. J Comp Neurol346:127–136

Hayakawa T, Zyo K (1985) Afferent connections of Gudden’stegmental nuclei in the rabbit. J Comp Neurol 235:169–181

Hayakawa T, Zyo K (1990) Fine structure of the lateral mammillaryprojection of the dorsal tegmental nucleus of Gudden in the rat.J Comp Neurol 298:224–236

Hayakawa T, Zyo K (1991) Quantitative and ultrastructural study ofascending projections to the medial mammillary nucleus in therat. Anat Embryol (Berl) 184:611–622

Hayakawa T, Zyo K (1992) Ultrastructural study of ascendingprojections to the lateral mammillary nucleus of the rat. AnatEmbryol (Berl) 185:547–557

Henn V, Young LR, Finley C (1974) Vestibular nucleus units in alertmonkeys are also influenced by moving visual fields. Brain Res71:144–149

Hsu SM, Raine L, Fanger H (1981) The use of antiavidin antibodyand avidin-biotin-peroxidase complex in immunoperoxidasetechniques. Am J Clin Pathol 75:816–821

Iwasaki H, Kani K, Maeda T (1999) Neural connections of thepontine reticular formation, which connects reciprocally withthe nucleus prepositus hypoglossi in the rat. Neuroscience93:195–208

Jasmin L, Burkey AR, Card JP, Basbaum A (1997) Transneuronallabeling of a nociceptive pathway, the spinal-(trigemino-)parabrachio-amygdalaoid, in the rat. J Neurosci 17:3751–3765

Kim SY, Frohardt RJ, Taube JS (2003) Head direction cells shiftreference frames in the vertical plane. Program No. 519.19,2003 Abstract viewer/itinerary planner. Society for Neurosci-ence, Washington, DC

Korp BG, Blanks RHI, Torigoe Y (1989) Projections of the nucleusof the optic tract to the nucleus reticularis tegmenti pontis andthe prepositus hypoglossi nucleus in the pigmented rat asdemonstrated by anterograde and retrograde transport methods.Vis Neurosci 2:275–286

Lannou J, Cazin L, Precht W, Le Taillanter M (1984) Responses ofprepositus hypoglossi neurons to optokinetic and vestibularstimulations in the rat. Brain Res 301:39–45

Liu R, Chang L, Wickern G (1984) The dorsal tegmental nucleus: anaxoplasmic transport study. Brains Res 310:123–132

Luppi PH, Aston-Jones G, Akaoka H, Chouvet G, Jouvet M (1995)Afferent projections to the rat locus coeruleus demonstrated byretrograde and anterograde tracing with cholera-toxin b subunitand phaseolus vulgaris leucoagglutinin. Neuroscience 65:119–160

Marchand ER, Riley JN, Moore RY (1980) Interpeduncular nucleusafferents in the rat. Brain Res 193:339–352

Marchand JE, Hagino (1983) Afferents to the periaqueductal gray inthe rat. A horseradish peroxidase study. Neuroscience 9:95–106

Marchand CF, Schwab ME (1986) Binding, uptake and retrogradeaxonal transport of herpes virus suis in sympathetic neurons.Brain Res 383:262–270

Matesz C, Bacskai T, Nagy E, Halasi G, Kulik A (2002) Efferentconnections of the vestibular nuclei in the rat: a neuromorpho-logical study using PHA-L. Brain Res Bull 57:313–315

McCrea RA (1988) Neuroanatomy of the oculomotor system. Thenucleus prepositus. Rev Oculomot Res 2:203–223.

McCrea RA, Baker R (1985) Anatomical connections of the nucleusprepositus in the cat. J Comp Neurol 237:377–407

McLean IW, Nakane PK (1974) Periodate-lysine-paraformaldehydefixative. A new fixative for immunoelectron microscopy. JHistochem Cytochem 22:1077–1083

Mizumori SJY, Williams JD (1993) Directionally selective mne-monic properties of neurons in the lateral dorsal thalamicnucleus of rats. J Neurosci 13:4015–4028

Muller RU, Ranck JB, Taube JS (1996) Head direction cells:properties and functional significance. Curr Opin Neurobiol6:196–206

Olucha-Bordonau FE, Teruel V, Barcia-Gonzalez J, Ruiz-Torner A,Valverde-Navarro AA, Martinez-Soriano F (2003) Cytoarchi-tecture and efferent projections of the nucleus incertus of therat. J Comp Neurol 464:62–97

Paxinos G, Watson C (1998) The rat brain in stereotaxic coordinates.Academic Press, San Diego

Pickard GE, Smeraski CA, Tomlinson CC, Banfield BW, KaufmanJ, Wilcox CL, Enquist LW, Sollars PJ (2002) Intravitrealinjection of the attenuated pseudorabies virus PRV Bartharesults in infection of the hamster suprachiasmatic nucleus onlyby retrograde transsynaptic transport via autonomic circuits. JNeurosci 22:2701–2710

Risold PY, Swanson LW (1996) Structural evidence for functionaldomains in the rat hippocampus. Science 272:1484–1486

Risold PY, Swanson LW (1997) Connections of the rat lateral septalcomplex. Brain Res Brain Res Rev 24:91–113

Ruggiero DA, Giuliano R, Anwar M, Stornetta R, Reis DJ (1990)Anatomical substrates of cholinergic-autonomic regulation inthe rat. J Comp Neurol 292:1–53

Schor RH, Steinbacher BC, Yates BJ (1998). Horizontal linear andangular responses of neurons in the medial vestibular nucleusof the decerebrate cat. J Vestib Res 8:107–116

Sharp PE, Blair HT, Cho J (2001a) The anatomical and computa-tional basis of the rat head-direction signal. Trends Neurosci24:289–294

Sharp PE, Tinkelman A, Cho J (2001b) Angular velocity and headdirection signals recorded from the dorsal tegmental nucleus ofGudden in the rat: implications for path integration in the headdirection cell circuit. Behav Neurosci 115:571–588

60

Shibata H (1987) Ascending projections to the mammillary nuclei inthe rat: a study using retrograde and anterograde transport ofwheat germ agglutinin conjugated to horseradish peroxidase. JComp Neurol 264:205–215

Shibata H, Suzuki T, Matsushita M (1986) Afferent projections tothe interpeduncular nucleus in the rat, as studied by retrogradeand anterograde transport of wheat germ agglutinin conjugatedto horseradish peroxidase. J Comp Neurol 248:272–284

Stackman RW, Taube JS (1997) Firing properties of head directioncells in the rat anterior thalamic nucleus: dependence onvestibular input. J Neurosci 17:4349–4358

Stackman RW, Taube JS (1998) Firing properties of rat lateralmammillary single units: head direction, head pitch, andangular head velocity. J Neurosci 18:9020–9037

Stackman RW, Tullman ML, Taube JS (2000) Maintenance of rathead direction cell firing during locomotion in the verticalplane. J Neurophysiol 83:393–405

Swanson LW (1998) Brain maps: structure of the rat brain. Elsevier,Amsterdam

Takeuchi Y, Allen GV, Hopkins DA (1985) Transnuclear transportand axon collateral projections of the mamillary nuclei in therat. Brain Res Bull 14:453–468

Taube JS (1995) Head-direction cells recorded in the anteriorthalamic nuclei of freely moving rats. J Neurosci 15:70–86

Taube JS (1998) Head direction cells and the neurophysiologicalbasis for a sense of direction. Prog Neurobiol 55:225–256

Taube JS, Muller RU, Ranck JB (1990) Head-direction cellsrecorded from the postsubiculum in freely moving rats. I.Description and quantitative analysis. J Neurosci 10:420–435

Vahlne A, Svennerholm B, Sandber M, Hamberger A, Lycke E(1980) Differences in attachment between herpes simplex type1 and type 2 viruses to neurons and glial cells. Infect Immun28:675–680

Watson RE, Wiegand ST, Clough RW, Hoffman GE (1986) Use ofcryoprotectant to maintain long-term peptide immunoreactivityand tissue morphology. Peptides 7:155–159

Weiner SI (1993) Spatial and behavioral correlates of striatalneurons in rats performing a self-initiated navigational task. JNeurosci 13:3802–3817

Wirtshafter D, Stratford TR (1993) Evidence for GABAergicprojections from the tegmental nuclei of Gudden to themammillary body in the rat. Brain Res 630:188–194

Zelman FP, Behbehani MM, Beckstead RM (1984) Ascending anddescending projections from nucleus reticularis magnocellularisand nucleus reticularis gigantocellularis: an autoradiographicand horseradish peroxidase study in the rat. Brain Res 292:207–220

61