karten-et-al-1997.pdf

Two Distinct Populations of TectalNeurons Have Unique Connections Within

the Retinotectorotundal Pathwayof the Pigeon (Columba livia)

HARVEY J. KARTEN,1* KEVIN COX,1 AND JORGE MPODOZIS21University of California, San Diego, La Jolla, California 92093-0608

2University of Chile, Santiago, Chile

ABSTRACTThe tectofugal pathway is a massive ascending polysynaptic pathway from the tectum to

the thalamus and then to the telencephalon. In birds, the initial component of this pathway isknown as the tectorotundal pathway; in mammals, it is known as the tectopulvinar pathway.The avian tectorotundal pathway is highly developed; thus, it provides a particularlyappropriate model for exploring the fundamental properties of this system in all amniotes. Tofurther define the connectivity of the tectorotundal projections of the tectofugal pathway, weinjected cholera toxin B fragment into various rotundal divisions, the tectobulbar projection,and the ventral supraoptic decussation of the pigeon. We found intense bilateral retrogradelabeling of neurons that stratified within layer 13 and, in certain cases, granular staining inlayer 5b of the optic tectum. Based on these results, we propose that there are two distincttypes of layer 13 neurons that project to the rotundus: 1) type I neurons, which are found in theouter sublamina of layer 13 (closer to layer 12) and which project to the anterior and centralisrotundal divisions, and 2) type II neurons, which are found in the inner sublamina of layer 13(closer to layer 14) and which project to the posterior and triangularis rotundal divisions. Onlythe labeling of type I neurons produced the granular dendritic staining in layer 5b. Anadditional type of tectal neuron was also found that projected to the tectobulbar system. Wethen injected Phaseolus vulgaris-leucoagglutinin in the optic tract and found that the retinalaxons terminating within tectal layer 5b formed narrow radial arbors (710 m in diameter)that were confined to layer 5b. Based on these results, we propose that these axons are derivedfrom a population of small retinal ganglion cells (4.56.0 m in diameter) that terminate onthe distal dendrites of type I neurons.

This study strongly indicated the presence of a major bilateral oligosynaptic retinotectoro-tundal pathway arising from small retinal ganglion cells projecting to the rotundus with onlya single intervening tectal neuron, the proposed type I neuron. We suggest that a similarorganization of retinotectopulvinar connections exist in reptiles and in many mammals. J.Comp. Neurol. 387:449465, 1997. r 1997 Wiley-Liss, Inc.

Indexing terms: tectofugal pathway; visual system; birds; nucleus rotundus; retinal ganglion cells

The tectofugal pathway, which is a massive ascendingpolysynaptic pathway to the telencephalon, is first de-scribed in birds by Karten and Revzin (1966). A similarpathway is described by Diamond in tree shrews, insectivo-rous rodents, and primates (Diamond, 1973). Subsequentstudies by Hall and Ebner (1970) and Pritz (1975) inreptiles as well as by Northcutt and Butler (1980) inteleost fishes indicate that this major pathway is anancient route for conducting visual information into thevertebrate telencephalon. The highly developed aviantectofugal pathway provides a particularly appropriate

model for exploring the fundamental properties of thissystem in all amniotes. The central origin of this pathwayand the main retinorecipient structure in the brain of mostnonmammalian vertebrates is the optic tectum.

Grant sponsor: Fondecyt; Grant number: 1081; Grant sponsor: NEI;Grant number: EY-06890; Grant sponsor: NINDS; Grant number: NS 24560.

*Correspondence to: Dr. Harvey J. Karten, Department of Neurosciences-0608, University of California, San Diego, La Jolla, CA 92093-0608. E-mail:[email protected]

Received 3 March 1997; Revised 29 May 1997; Accepted 4 June 1997

THE JOURNAL OF COMPARATIVE NEUROLOGY 387:449465 (1997)

r 1997 WILEY-LISS, INC.

In birds, the optic tectum is a complex, topographicallyorganized structure that receives approximately 90% ofthe retinal ganglion cell projections from the contralateralretina. The retinotectal projection is formed by the axonsof various subpopulations of retinal ganglion cells, distin-guishable by conduction velocities (Mpodozis et al., 1995),morphology, neurotransmitters, and neural-related molecu-lar contents (Karten et al., 1990). These retinal axonsarborize at five different levels, in layers 2, 3, 4, 5, and 7, ofthe retinorecipient zone (Ramon y Cajal, 1911). Based onthe morphology of single axonal arborizations, Ramon yCajal (1911) suggested that each of the retinorecipientlayers contains terminals of a single specific type ofganglion cell, a suggestion that has been partially con-firmed by modern studies (Lettvin et al., 1959; Ehrlich etal., 1987; Britto et al., 1994). In addition to the retinalinputs, the tectum also receives massive inputs from otherstructures located in the forebrain, the thalamus, thepretectal region, and the midbrain, each exhibiting adefinite pattern of arborization within specific tectal layers(Ramon y Cajal, 1911; Brecha, 1978).

Efferent tectal projections arise from neurons in layers914 and target such structures as the isthmic nuclei, thelateral pons, some of the nuclei of the dorsal and ventralpretectal complex, and various visual nuclei of the thala-mus (Ramon y Cajal, 1911), including the nucleus rotun-dus. Various studies have demonstrated that the tectoro-tundal projections are bilateral, are much denseripsilaterally than contralaterally (Hunt and Kunzle, 1976),and have no obvious topographical organization (Benowitzand Karten, 1976). Benowitz and Karten also demon-strated that these projections arise from neurons locatedin layer 13, the stratum griseum centrale (SGC), of theoptic tectum.

The SGC is a distinctive, multistratified structure formedby neurons of roughly homogeneous size (ca. 20 m cellbody diameter). Dendrites of SGC neurons extend broadlyto more superficial tectal layers and have been seen toreach the retinorecipient layer 5b (Ramon y Cajal, 1911;Hunt and Kunzle, 1976). Physiological studies using intra-cellular recordings (Hardy et al., 1984, 1985) have shownthat some SGC neurons receive a monosynaptic input,whereas others receive a polysynaptic input from theretina. The axons of SGC neurons run ventrally to form the

tectothalamic tract and decussate at the level of the dorsalsupraoptic decussation. These axons also provide ipsilat-eral inputs to the intrinsic nuclei of the tectothalamic tract(Tombol et al., 1994; Mpodozis et al., 1996). Furthermore,the SGC is the sole tectal source of inputs to the rotundus(Mpodozis et al., 1996).

In addition to the projections from the SGC, the rotun-dus, which is a large, discrete group of neurons located inthe thalamus of the pigeon, also receives g-aminobutyricacid (GABA) inputs from both the intrinsic nuclei of thetectothalamic tract and the thalamic reticular formation(Benowitz and Karten, 1976; Mpodozis et al., 1996). Therotundus projects exclusively to the ectostriatum of theipsilateral telencephalon (Karten and Hodos, 1970). In ourprevious report (Mpodozis et al., 1996), we outline asimplified scheme of four divisions within the rotundus:anterior, centralis, posterior, and triangularis (Fig. 1). Theanterior and centralis divisions can be subdivided further(Benowitz and Karten, 1976; Martinez-de-la-Torre et al.,1990; Mpodozis and Karten, unpublished observations).For the purpose of the present study, however, we willconcentrate only on the four main divisions for our analy-sis of the tectofugal pathway.

To further define the functional organization of thetectofugal pathway, we began a series of studies of theanatomical organization of this pathway by using moresensitive techniques than were available previously, includ-ing use of the tracers Phaseolus vulgaris-leucoagglutinin(PHA-L) and cholera toxin B subunit (CTb). In our previ-ous report (Mpodozis et al., 1996), we demonstrated thatthe tectorotundal projection is composed of two differentpathways: one direct, bilateral, and presumably excitatoryand the other indirect, ipsilateral, and presumably inhibi-tory. The direct pathway arises from tectal neurons of

Abbreviations

Ce nucleus rotundus, pars centralisDa nucleus rotundus, pars dorsalis anteriorGLv nucleus geniculatus lateralis ventralisIPS nucleus interstitiopretectosubpretectalisnBOR nucleus of the basal optic rootnPT nucleus pretectalisOv nucleus ovoidalisPost nucleus rotundus, pars posteriorPV nucleus posteroventralis thalamiRt nucleus rotundusSAC stratum album centraleSP nucleus subpretectalisSPcd nucleus subpretectalis, pars caudalisSpL nucleus spiriformis lateralisStOp stratum opticumTeO tectum opticumTIO isthmooptic tractTr nucleus rotundus, pars triangularisVSOD ventral supraoptic decussation

Fig. 1. Divisions of the rotundus. The four major divisions ofrotundus, dorsalis anterior (Da), centralis (Ce), triangularis (Tr), andposterior (Post), are shown in transverse (left) and sagittal (right)views. Drawings are based on Giemsa-stained sections as well as insections immunoreacted for acetylcholinesterase, parvalbumin, andcalcium-binding protein. Numbers below each drawing indicate theapproximate stereotaxic level (Karten and Hodos, 1967) of the corre-sponding section. For abbreviations, see list. Scale bar 5 3 mm.

450 H.J. KARTEN ET AL.

layer 13 and provides inputs to all divisions of the rotun-dus, both ipsilaterally and contralaterally. The indirectpathway arises from collaterals of the layer 13 neurons enroute to the rotundus. These collaterals end ipsilaterally inthe intrinsic GABAergic nuclei of the tectothalamic tract.The neurons of these nuclei, in turn, terminate in distinctdivisions of the ipsilateral rotundus, such that differentdivisions of the rotundus receive inputs from differentcomponents of the intrinsic nuclear complex of the tectotha-lamic tract.

The present study demonstrates that these tectorotun-dal projections arise from what we propose are two distinctclasses of layer 13 tectal neurons. Each class of layer 13neuron is the origin of a direct and an indirect projection todifferent divisions of the rotundus. Type I neurons, whichlie within the outer sublamina of layer 13 (the sublaminacloser to layer 12), project bilaterally to the anterior andcentralis divisions of the rotundus and are associated withdendritic staining in layer 5b. Type II neurons lie withinthe inner sublamina of layer 13 (nearer layer 14), projectbilaterally to the posterior and triangularis divisions of therotundus, and lack any associated dendritic staining inlayer 5b.

Our study also provides a preliminary characterizationof a specific category of retinal ganglion cells that termi-nate in layer 5b and their morphological pattern of axonaltermination. These studies indicated that there is a farmore direct impact of retinal afferentation on rotundalneurons than was appreciated previously. Some prelimi-nary results from this study have appeared in abstractform (Karten et al., 1993).

MATERIALS AND METHODSWhite Carneaux pigeons of either sex were used in this

study and were obtained from either the Palmetto PigeonPlant (Sumter, SC), or Bowman Gray School of Medicine(Winston Salem, NC). All surgical procedures performedon these animals were approved by the Animal CareCommittee of the University of California, San Diego, andconformed to the guidelines of the National Institutes ofHealth on Use of Experimental Animals in Research.

Pigeons were anesthetized with a combination of ket-amine (40 mg/kg) and xylazine (12 mg/kg) injected intra-muscularly. A single dose usually proved satisfactory forthe duration of the surgical procedure. If necessary, asupplementary injection of anesthetic was administered.

Once the animal was anesthetized, the feathers over theskull were removed, and the scalp was washed with anantiseptic solution. Animals were placed in a stereotaxichead holder, the skull was exposed, and a hole was madethrough the skull above the approximate location of theproposed injection. Tracer injections into various divisionsof the rotundus, the tectobulbar projection, the ventralsupraoptic decussation, and the optic tract were achievedby using the stereotaxic coordinates of Karten and Hodos(1967). Intraocular injections as well as those into the optictectum were made under direct visual control. To enhancethe accuracy of some of the placements of the tracers,electrophysiological responses were monitored. Either a1% solution (by weight) of CTb (List Laboratories, Camp-bell, CA) was injected by using manual direct pressure or aPicoSpritzer II pressure system (General Valve, Fairfield,

NJ), or a 2.5% solution of PHA-L (Vector Laboratories,Burlingame, CA) was deposited iontophoretically by usinga constant-current device (Midgard Electronics, Cam-bridge, MA; Gerfen and Sawchenko, 1983; for a detailedreview of these procedures, see Mpodozis et al., 1996).

In 410 days, depending on the tracer employed, theanimals were anesthetized with an overdose of ketamineand xylazine and were perfused through the aorta with500800 ml of a 0.9% saline solution, followed by 1,000 mlof an ice-cold solution of 4% paraformaldehyde in 0.1 Mphosphate buffer (PB), employing a peristaltic perfusionpump to maintain constant perfusion pressure. Followingthe perfusion, the brains were removed from the skull andpostfixed for 614 hours in the paraformaldehyde solutionand were then transferred to a 30% sucrose solution madein PB for cryoprotection before cutting on a sliding micro-tome. The brains were cut into 30-m-thick sections ineither the horizontal, sagittal, or coronal plane.

Following a 30-minute wash in phosphate-buffered sa-line, the tissue was incubated in either goat anti-CTb(1:15,000; List Laboratories) or rabbit anti-PHA-L (1:10,000; Vector Laboratories), depending on the tracerinjected. The tissue was then processed by using theavidin-biotin-peroxidase method (ABC Elite kit; VectorLaboratories; for a more detailed description of theseimmunohistochemical procedures, see Mpodozis et al.,1996). The tissue was mounted on gelatinized slides andallowed to dry. The tissue was then either stained with0.04% osmium tetroxide for 1020 seconds or counter-stained with Giemsa stain (Iniguez et al., 1985), dehy-drated, cleared, and coverslipped by using Permount.

Data analysis was performed by using a matched seriesof sections immunoreacted against CTb. One series wasosmicated, and the other series was counterstained withGiemsa. The two series were compared, and selectedsections from each were drawn by using a macroprojector.The distribution of axonal processes was evaluated fre-quently by using darkfield microscopy. Labeled neuronsand axons were plotted on these drawings, and the resul-tant drawings were entered directly into a Mac IIfx datafile using a Hewlett-Packard desk-top scanner. The scanneddrawing was traced by using Canvas (Deneba Corp.,Miami, FL). Additional sections were charted by usingNeurolucida, a computer-controlled, motorized stage andsoftware system (MicroBrightField Inc., Colchester, VT).

Video images of neurons with nonoverlapping fields ofadjacent neuronal bodies were captured by using a Dage-MTI 72S monochrome CCD camera (Dage-MTI, Inc., Michi-gan City, IN), digitized by using a Scion LG-3 (Scion Corp.,Frederick, MD), and displayed on a color monitor. Thedimensions of labeled neurons in layer 13 were measuredby using the processing program by Wayne Rasband, NIHImage, on a Macintosh PowerPC.

Photographs for figures are digital images that werecaptured by using a Leaf Lumina scanning CCD camera(Leaf System Inc., Southborough, MA) mounted on aNikon Microphot FX photomicroscope and collected di-rectly into Adobe Photoshop (Adobe Systems Inc., Moun-tain View, CA) by using a Macintosh PowerPC. Imageswere collected as RGB files, converted to gray scale im-ages, and brightness and contrast were adjusted to providethe resultant final printed images. No additional digitalfiltration or image modification was performed. Final

RETINOTECTAL INPUTS TO ROTUNDUS 451

prints were prepared by using a Kodak (Rochester, NY)XLS 8600 PS printer.

RESULTSTectal afferents to rotundus

To study the tectorotundal projections, CTb was injectedinto various divisions of the rotundus of ten pigeons:anterior (two cases), centralis (dorsalis and ventralis; twocases), posterior (one case), and triangularis (one case).The remaining cases involved injections into more thanone division: dorsal anterior and centralis (one case),centralis and posterior (one case), posterior and triangula-ris (one case), and posterior and caudal surround (onecase). Tracers were also injected into the tectobulbar

projection (two cases), the ventral supraoptic decussation(one case), the optic tract (three cases), the optic tectum(four cases), and the eye (two cases). Some representativeinjection sites are shown in Figure 2.

The nucleus rotundus lies in a region of the thalamussurrounded by other cell groups that are also in directreceipt of tectal projections, including the nucleus opticusprincipalis thalami (OPT; the avian homologue of themammalian nucleus geniculatus lateralis pars dorsalis orGLd), which is located immediately dorsal to the rotundus;the nucleus ventrolateralis thalami (VLT), which is locatedon the ventromedial aspect of the rotundus; the intergenicu-late leaflet (IGL), which is located on the rostral andlateral aspects of the rotundus; and the nucleus genicula-tus lateralis ventralis (GLv), which is located ventral to

Fig. 2. Pattern of cholera toxin B subunit (CTb) labeling observedin the tectum/pretectum as a result of CTb injections into the centralis(top) and triangularis (middle) rotundal divisions and the ventralsupraoptic decussation (VSOD; bottom). Injection sites are shown onleft, and the resulting neuronal labeling is shown on the right.Drawings were made from CTb-treated, Giemsa-counterstained sec-tions. Approximate anterior-posterior stereotaxic levels (Karten andHodos, 1967) are indicated below each drawing. Solid black areasrepresent the actual deposit of the tracer, black dots mark the locationof labeled somata, and hatched and solid gray areas indicate axonal or

dendritic labeling, respectively. Note that the centralis and VSODcases show a bilaterally dense plexus of axonal processes, which coverthe nucleus subpretectalis/nucleus interstitiopretectosubpretectalis(SP/IPS) and progress to surround the pretectalis. By contrast, theCTb injections into the triangularis resulted in no labeling of thesepretectal nuclei. All cases resulted in bilateral labeling of neurons oftectal layer 13, but only the centralis and VSOD cases showedbilateral labeling of processes within tectal layer 5b. For otherabbreviations, see list. Scale bars 5 3 mm.


the rotundus. Therefore, we made CTb injections into eachof these regions. We concluded that the tectal neuronsprojecting to the rotundus did not project to any of theseregions. The laminar origins of tectal projections to each ofthese cell groups will be described in a subsequent report.

Following injections of CTb into either the anteriordivision or the centralis division of the rotundus, there wasextensive bilateral retrograde CTb labeling of tectal neu-rons only within the outer sublamina of layer 13. Con-versely, when the tracer was placed in either the posterioror triangularis divisions of the rotundus, retrogradely CTblabeled neurons were confined to the inner sublamina oflayer 13. Despite the regional variation in the relativethickness of each of these sublaminae (see below), there

was no substantial overlap of the distribution of the layer13 neurons projecting to the anterior or centralis divi-sions vs. the posterior or triangularis divisions of therotundus (Figs. 3, 4). Labeled neurons in the outer sub-lamina of layer 13 were predominantly round or polygonalin shape (18 m mean major axis, 13 m mean minoraxis), with their major axis running radially to the tectalsurface. CTb-labeled neurons in the inner sublamina oflayer 13 were mostly of fusiform shape (19 m mean ma-jor axis, 9 m mean minor axis), with their major axisrunning parallel to the tectal surface. In all of the experi-mental cases, the number of ipsilateral vs. contralateralCTb-labeled neurons in layer 13 was approximately 2:1(Fig. 3).

Fig. 3. Retrograde bilateral labeling of the optic tectum followingunilateral injections of cholera toxin B subunit (CTb) into the centralis(A,B) or the triangularis (C,D) divisions of the rotundus. Digitalimages of the dorsolateral aspect of the optic tectum were taken at thelevel of the commissura tectalis of CTb-labeled, Giemsa-counter-stained transverse sections. Compare the location of the retrogradelylabeled neurons: CTb injections into the centralis (A,B) labeledneurons that were restricted to the outer sublamina of layer 13,whereas CTb injections into the triangularis (C,D) labeled neurons

that were mostly confined to the inner sublamina of layer 13. Anapproximate 2:1 ratio of ipsilateral:contralateral labeled neurons canbe noted in both instances. In the centralis injection, a fine granulardendritic staining was found in the full extent of layer 5b. Suchstaining in the ipsilateral side is more dense than in the contralateralside. There is no evidence of CTb labeling of layer 5 following CTbinjections into the triangularis. The injection sites of each case areshown in Figure 2. For abbreviations, see list. Scale bar 5 180 m.


Dendritic immunoreactivity in layer 5bof the optic tectum

Intense dendritic staining of the processes of layer 13neurons labeled following rotundal CTb injections wasseen extending superficially to form a dense matrix inlayers 912 (Figs. 3, 5). This was associated with thepresence of retrograde CTb labeling of the inner and outersublamina of layer 13. However, a band of granularmaterial was seen in layer 5b of the optic tectum onlyfollowing injections confined to the anterior or centralisdivisions of the rotundus (Figs. 25). Although it wasnotably evident on the ipsilateral side, a somewhat lighterpattern of staining within layer 5b was also observed inthe contralateral optic tectum (Fig. 3). In clear contrast, aCTb injection within the posterior or triangularis rotundusresulted in dense CTb labeling of the inner sublamina oflayer 13 but failed to produce evidence of such granularstaining in layer 5b (Figs. 2, 3).

Thus, the granular staining was associated only withretrogradely CTb-labeled neurons in the outer sublaminaof layer 13. This did not represent a projection of therotundal neurons to the tectum. This study repeatedlyindicated that the rotundal neurons projected only to the

ipsilateral telencephalon and not to the optic tectum.Similarly, injections of tracers in the various control sites,including OPT/GLd, GLv, VLT, and IGL, did not result inprojections to layer 5b.

This was further supported by our finding that injec-tions in the ventral supraoptic decussation resulted in anequal, bilateral CTb labeling of many neurons in both theinner and outer sublamina of layer 13 as well as an equallydense, bilateral, granular staining in layer 5b (Fig. 2). Wefound no evidence of retrograde CTb labeling of neurons inthe rotundus or in the intrinsic nuclei of the tectothalamictract. Thus, the granular staining in layer 5b is presumedto represent the distal dendrites of layer 13 neurons, asoriginally described by Ramon (1899), Ramon y Cajal(1911), and Hunt and Kunzle (1976).

Regional variation of layer 5b and layer 13The pattern of CTb labeling found in layer 5b following

injections of CTb into the anterior or centralis rotundusshowed a clear regional variation (Fig. 5). The overallwidth of layer 5b in the ventral tectum was threefoldthicker than in the dorsal tectum (ca. 150 m vs. 50 m).

Fig. 4. Retrogradely cholera toxin B subunit (CTb)-labeled neu-rons in the optic tectum following unilateral injections of CTb into thecentralis (A; injection site shown in Fig. 2) and triangularis (B)divisions of the rotundus (injection site shown in Fig. 2) and into thetectobulbar projection (C). In each case, digital images of the lateralaspect of the optic tectum were taken at the level of the commissuratectalis of CTb-labeled, Giemsa-counterstained sections. Note thepolygonal shape and the radial orientation of the CTb-labeled neuronsin the centralis case (A) and compare it with the fusiform shape and

tangential orientation of the CTb-labeled neurons in the triangulariscase. In the centralis and triangularis cases, the CTb-labeled neuronsstratified within layer 13 in the outer or the inner sublamina,respectively. In the tectobulbar case (C), most of the CTb-labeledneurons were found in layer 14 and the underlying tegmentum. Notethat tectobulbar-projecting neurons are smaller than the tectorotundal-projecting neurons. Also, tectobulbar-projecting neurons did not forma matrix of dendrites that projected superficially. Scale bar 5 100 m.


The transition point between the thinnest and thickesttectal regions was very abrupt and was located at thelateral margin of the tectum. This distribution matchedexactly the pattern of retinal terminals in layer 5b that havebeen described previously (Gamlin et al., 1996) and that weobserved following intraocular injections of CTb (Fig. 6).

Layer 13 also showed a clear ventrodorsal differentia-tion. The overall width of layer 13 in the ventral tectum

was 1.31.6 times thicker than the dorsal tectum, depend-ing on the rostrocaudal level of the analyzed section. Thisvariation matched the differential distribution of the rotun-dal-projecting neurons located in the outer sublamina oflayer 13 (Fig. 7). Following injections of CTb into theanterior or centralis divisions of the rotundus, CTb-labeledneurons were restricted to the outer lamina of layer 13, butCTb-labeled neurons in the ventral tectum were notably

Fig. 5. Regional variation of the ipsilateral layer 5b labeledfollowing a unilateral cholera toxin B subunit (CTb) injection into thecentralis division. Digital images of the dorsal (A), dorsolateral (B),and ventral (C) aspects of the optic tectum were taken at the level ofthe oculomotor nuclei of CTb-labeled, Giemsa-counterstained sections.In each aspect, the density of the CTb labeling in layer 5b was similar,

but the width of the labeled zone diminished abruptly from the ventralto the dorsal tectum. Tectal layers located superficially to layer 5showed no evidence of CTb labeling. Note that the CTb-labeled neurons inthe outer sublamina of layer 13 form a dense plexus of dendriticprocesses that extend superficially. Note also the higher density ofCTb-labeled neurons in the ventral tectum. Scale bar 5 100 m.


more abundant and densely packed than those found inthe dorsal tectum (Fig. 7). The outer sublamina of layer 13in the ventral tectum was 2.02.5 times thicker than in thedorsal tectum, depending on the rostrocaudal level of thesection. The inner sublamina of layer 13 showed anopposite but less marked dorsoventral polarity, with, atmost, the dorsal tectum 1.4 times wider than the ventraltectum. CTb-labeled neurons in layer 13 following injec-

tions of CTb into the posterior or triangularis rotunduswere distributed along the entire extent of the innersublamina of layer 13 and showed a relatively higherdensity in the dorsal tectum.

At the level of the oculomotor nuclei, layer 13 of thedorsal tectum had a mean width of 280 m, with an outersublamina width of 110 m and an inner sublamina widthof 170 m. In the ventral tectum, the mean width was 400

Fig. 6. Regional variation of the contralateral retinorecipientlayers of the optic tectum labeled following an intraocular injection ofcholera toxin B subunit (CTb). Digital images of the dorsal (A),dorsolateral (B), and ventral (C) aspects of the optic tectum weretaken at the level of the commissura tectalis of CTb-labeled, Giemsa-counterstained sections. Note that the width of layer 5b diminishes

abruptly from the ventral to the dorsal tecum, whereas the width oflayers 2, 3, and 4 show an opposite variation. The width of layer 7remains constant in all tectal regions. Note also that, in the ventraltectum, layer 5b occupies the major proportion of the retinorecipientzone. StOp, stratum opticum. Scale bar 5 50 m.


m, with outer and inner layer 13 sublamina widths of 260m and 140 m, respectively. We concluded from theseobservations that the overall distribution of tectal neuronsprojecting to anterior and central rotundus (the proposedtype I neurons) exhibited a close correlation with thedistribution of layer 5b retinal terminals.

Tectal neurons projecting to the anterioror central rotundus also collateralize

on the nucleus pretectalisThe nucleus pretectalis is found in the lateral pretectal

region, lying on the dorsomedial margin of the rostral optictectum. The pretectalis contains a compact central regionof neurons and a surrounding cell-free region (Gamlin etal., 1996). The pretectalis projects to the ipsilateral andcontralateral tectum, with a dense field of terminals inlayer 5b. Injections of CTb into the anterior or centralisrotundus resulted in axonal projections to the outer shellof the pretectalis (Figs. 2, 8). A similar but less denseprojection was also found in the outer shell of the contralat-eral pretectalis. In contrast, injections of CTb into theposterior or triangularis rotundal divisions resulted in noevidence of connections to the pretectalis (Fig. 2, 8). Therewas no indication of any retrograde labeling of neuronswithin the pretectalis. Injections of CTb directly into the

decussating axons of layer 13 neurons at the level ofventral supraoptic decussation resulted in equally densebilateral axonal staining in the outer shell of the pretecta-lis (Fig. 2). There was no indication of retrograde labelingof rotundal neurons. We therefore concluded that theaxonal projections of the proposed type I neurons to thedorsal anterior or centralis rotundus also collateralized onthe outer shell of the pretectalis.

Tectorotundal neurons are distinctfrom tectobulbar neurons

In the present study, we were readily able to distinguishsubpopulations of tectal neurons projecting to the rotun-dus from those projecting downstream to the bulbar reticu-lar formation and the tectobulbar projection. FollowingCTb injections into the tectobulbar projection (predorsalbundle) at the level of the pontine nuclei, retrogradelyCTb-labeled tectal neurons were occasionally found at themargin between layers 13 and 14 but were more generallyfound in layer 14 and the underlying tegmentum (Fig. 4).These neurons were polygonal in shape and were consis-tently smaller than the neurons projecting to the rotundus(14 m mean major axis). Furthermore, these neurons didnot form a dense matrix of dendritic arborizations extend-ing superficially and only rarely showed evidence of den-

Fig. 7. Regional variation of the cholera toxin B subunit (CTb)labeling in the ipsilateral tectal layer 13 following a unilateral CTbinjection into the centralis division of the rotundus. Digital images ofthe dorsal (A), dorsolateral (B), and ventral (C) aspects of the optictectum were taken at the level of the oculomotor nuclei of CTb-labeled,Giemsa-counterstained sections. In each aspect, CTb-labeled neurons

were confined to the outer sublamina of layer 13. The width of thislamina as well as the number of CTb labeled neurons it containeddiminished abruptly from the dorsal to the ventral tectum. Note thepolygonal shape of most of the neurons and that their dendriticprocesses extend only to more superficial levels. Scale bar 5 35 m.


Fig. 8. Retrograde labeling of the nucleus pretectalis (nPT) follow-ing unilateral cholera toxin B subunit (CTb) injections into thecentralis or the triangularis divisions of the rotundus. Digital imageswere taken from CTb-labeled, Giemsa-counterstained transverse sec-tions. The centralis case shows a bilateral bundle of CTb-labeled axonsthat surround and extend into the pretectalis. The density of such CTb

labeling in the ipsilateral side (A) is higher than in the contralateralside (B). In contrast, the CTb injection into the triangularis resulted inno label of axonal processes either inside or around the pretectalis(C,D). Neurons of the pretectalis were not retrogradely CTb labeled inany of these cases. Injection sites of each of these cases are shown inFigure 2. SpL, nucleus spiriformis lateralis. Scale bar 5 300 m.


dritic branches in the retinorecipient layers of the tectum.There was no evidence of axonal projections of theseneurons to the rotundus, the intrinsic nuclei of the tectotha-lamic tract, or the pretectalis. Further support of thisinterpretation was derived from our analysis of the CTbinjections into the ventral supraoptic decussation. Injec-tions into the ventral supraoptic decussation resulted inretrograde CTb labeling of the neurons of layer 13 project-ing to the rotundus as well as the labeling of the axonalprojections of these neurons, as described above. In thisinstance, however, there was no indication of collateralaxonal projections to the tectobulbar pathway or to theipsilateral tectoreticular pathway. On the basis of this setof observations, we concluded that the output of the tectumto the rotundus is independent of that to the brainstem.

Retinal inputs to layer 5bThe observation of the many dendritic arborizations of

the proposed layer 13 type I neurons in tectal layer 5bprompted us to investigate the nature of the retinal inputsto layer 5b. Therefore, we 1) injected the anterogradetracer PHA-L into the optic tract to examine the morphol-ogy of retinal axonal terminals in layer 5b and 2) injectedCTb into either the dorsal or the ventral tectum tocompare the features of the resulting retrogradely CTb-labeled retinal ganglion cell populations. The rationale forthis latter experiment was that the marked increase in thethickness of layer 5b in the ventral tectum, accompaniedby a corresponding reduction in the thickness of retinore-cipient layers 2, 3, and 4 (Fig. 6), indicated that the ratio ofretinal ganglion cells projecting to layer 5b vs. retinalganglion cells projecting to the other layers was markedlyincreased in the portion of the retina projecting to theventral tectum.

Figure 9 shows a representative selection of PHA-L-labeled retinal terminals arborizing in layer 5b in theventral optic tectum following injections of PHA-L into theoptic tract. These retinal terminals formed finely branch-ing, radial processes that were confined to the radialextent of layer 5b. The total length of these terminal arborswas ca. 125 m, and their width ranged from 7 m to 15m. The individual branches were extremely fine, wereoften less than 1 m, and showed several terminal bou-tonal expansions of 2.02.2 m in diameter. Retinal axonsarborizing in layer 5b did not form branches or terminalprocesses in the superficial retinorecipient layers. Further-more, the retinal terminals in tectal layers 3, 4, and 7 wereof distinctively different morphology and were readilydistinguishable from the terminals in layer 5b (Fig. 9).Therefore, we concluded that the retinal ganglion cellsarborizing in layer 5b did not have endings in other tectallayers.

Injections of CTb restricted to the ventral optic tectumresulted in a well-defined cluster of retrogradely labeledretinal ganglion cells that were confined mainly to thedorsotemporal quadrant of the retina, or red field (Fig. 10).The resulting CTb-labeled cell population was composedalmost exclusively of extremely small, round cells withsomal diameters of 4.56.0 m. These cells were denselypacked and were placed one on top of another in amultilayered array. The point of origin of the dendriticarborizations of these cells was not evident. Larger (812m in diameter), CTb-labeled retinal ganglion cells werealso found, but they were rare. We estimated that the totalnumber of these small cells in the red field was about 1.5

million, i.e., 65% of the total retinal population of ganglioncells.

In contrast, CTb injections restricted to the dorsolateralaspect of the optic tectum resulted in the retrogradelabeling of a more heterogeneous and remarkably lessdense population of ganglion cells located at the inferiornasal quadrant of the retina (Fig. 10). The CTb-labeledcells were mainly polygonal, and most of them clearlyexhibited dendritic branches. The somal size of these cellsranged from 5 m to 15 m, but the greater proportion was710 m in diameter (ca. 75%). The small, round cells werealso present but in a lesser proportion (ca. 20%). Thesecells were evenly distributed and did not show a multilay-ered array.

A remarkable feature of the pattern of CTb labeling ofthe retina following the ventral tectum CTb injections wasfound in the inner plexiform layer. The inner plexiformlayer showed a dense, confluent pattern of staining forCTb. However, at the margin of the region containingretrogradely labeled neurons in the retina, the pattern ofstaining appeared increasingly patchy. This patchy appear-ance was made up of circular domains of roughly homoge-neous size, ranging from 25 m to 40 m in diameter (Fig.11). We interpret this patch as representing the individualdendritic domains of the small retinal ganglion cells.Similarly sized patches were also present, even withinregions of very dense staining, presumably reflectingoverlapping dendritic domains of these ganglion cells.From these observations, we concluded that tectal layer 5breceives retinal input mainly, if not exclusively, from theseextremely small retinal ganglion cells that possess smalldendritic fields in the retina and narrow radial axonalarbors in the tectum.

DISCUSSIONThe present study provided new insights into some of

the major features of the tectorotundal pathways of birds.Previous studies (Karten and Revzin, 1966; Revzin andKarten, 1967) have reported that layer 13 neurons are themajor source of tectal projections to the nucleus rotundus.These tectal projections, Benowitz and Karten (1976)suggested, project to the different divisions of the rotundusand derive from different sublaminae of layer 13. By usingCTb, which is a much more sensitive tracer than thepreviously used horseradish peroxidase, the present studydemonstrates that this differential sublaminar origin isfar more prominent than Benowitz and Karten indicated.

Specifically, our study showed 1) the existence of twodiscrete subpopulations of tectorotundal projecting neu-rons with different afferent and efferent relationships, 2)the nature of the retinal ganglion cells that provide amajor (presumably monosynaptic) input to one of thesesubpopulations, and 3) the pattern of direct and indirectprojections of these two subpopulations to different divi-sions of the rotundus.

Two subpopulations oftectorotundal-projecting neurons

One of the most prominent findings of the present studywas the intense pattern of granular staining found bilater-ally in layer 5b of the optic tectum following injections ofCTb into the anterior or centralis divisions of the rotun-dus. This staining was not observed in those cases with


injections into either the posterior or triangularis divi-sions.

The intense layer 5b staining found on the contralateralside appeared to be proportional to the number of contralat-eral CTb-labeled layer 13 neurons. Furthermore, injec-tions into the ventral supraoptic decussation that resultedin retrogradely labeled neurons in layer 13 of the tectumalso resulted in a similar pattern of layer 5b staining,

suggesting that this staining represented labeled pro-cesses of the layer 13 neurons. We found no indication thatthis staining was due to projections from the nucleusrotundus or from any other nearby nuclei that mightbecome involved by the spread of tracer.

Ramon y Cajal (1911) repeatedly illustrated the externalarbors of the neurons of tectal layer 13 and drew them toindicate a unique expansion as the dendrites entered the

Fig. 9. Representative retinal axonal terminals in the optic tectumlabeled following a Phaseolus vulgaris-leucoagglutinin (PHA-L) injec-tion into the optic chiasm. Digital images were taken from PHA-L-labeled transverse sections. Note that the terminals in layer 5 (L.5)

form fine branchings that are confined to the extent of the layer andthat are oriented radially to the tectal surface. In comparison,terminals in layers 2 (L.2), 4 (L.4), and 7 (L.7) are more coarse and runparallel to the tectal surface. Scale bars 5 30 m.


region of layer 5. However, he did not specifically label thislamina as layer 5. Hunt and Kunzle (1976) suggested thatthe dendrites of layer 13 (SGC) are a major component ofthe neuropil of layer 5 (designated as layer Id in thenomenclature of Cowan et al., 1961), as reflected in thedense autoradiographic labeling of layer 5 following injec-tions of radioactive tracers into layer 13. However, thelimit of deposit of the injected isotope could not be assessedaccurately. Hunt and Kunzle then compared their radioac-tive amino acid uptake and transport pattern to themorphology of layer 13 neurons observed in their Golgi-stained material. They demonstrated dendritic arbors oflayer 13 neurons in layer 5, although the density of sucharbors was not evident in their Golgi studies.

In our CTb-labeled preparations, the granular stainingin layer 5b appeared to lie in a radial orientation, but theconnecting processes were so fine in caliber that theirmorphology could not be discerned readily. However, basedboth on the specific morphological characteristics of theneurons and on the studies of Ramon (1899), Ramon yCajal (1911), and Hunt and Kunzle (1976), we concludedthat the granular staining was contained in distal dendriticterminals of neurons in the outer sublamina of layer 13.

Thus, we propose the presence of two major morphologi-cal types of neurons within layer 13 of the avian optictectum: 1) type I neurons, which lie in the outer sublaminaof layer 13, project to the anterior and centralis divisions ofthe rotundus, and possess extensive, fine dendritic arborsin layer 5b of the optic tectum; and 2) type II neurons,which lie in the inner sublamina of layer 13 and project tothe posterior and triangularis divisions of the rotundusbut do not show dendritic arbor staining in layer 5b.

Fig. 11. Contralateral inner plexiform layer of the retina labeledfollowing a cholera toxin B subunit (CTb) injection into the ventraltectum. This darkfield digital image was taken from a CTb-labeledhorizontal section of the retina that passed through the inner plexi-form layer. Note the circular patches of CTb-labeled processes (ar-rows). Scale bar 5 40 m.

Fig. 10. Retinal ganglion cells labeled following cholera toxin Bsubunit (CTb) injections into the ventral or the dorsal contralateraloptic tectum. Digital images were taken at the zone of highest densityof CTb-labeled cells in horizontal sections of the retina. Note the smallsize, homogeneous shape, and high density of the ganglion cell

population labeled following the ventral tectum injection (A). Comparethis with the lower density and heterogeneity of the CTb-labeledganglion cell population following the CTb injection into the dorsaltectum (B). Scale bar 5 15 m.


Retinal inputs to layer 5bThe pattern of retinal terminals within different layers

of the superficial laminae of the optic tectum (layers 2, 3, 4,5, and 7) have been examined by many authors over thepast 100 years, including Ramon (1899), Ramon y Cajal(1911), LaVail and Cowan (1971), and Stone and Freeman(1971). The most notable feature is the differing morphol-ogy of retinal terminals found in each layer of the tectum.This is clearly reflected in the illustrations of Ramon yCajal (1911), who also suggested that these different typesof axonal terminals arise from different retinal ganglioncell types.

In the present study, we found that retinal terminals inlayer 5b had a notably different morphology than theaxons ending in layers 4 and 7 (Fig. 9). The retinalterminals in layers 4 and 7 were more coarse than thosefound in layer 5b and generally ran parallel to the surfaceof the tectum. In contrast, layer 5b terminals formed fine,radial processes with numerous small terminal expan-sions. In our PHA-L-labeled material, the axons that gaverise to these layer 5b terminals did not show branchingprocesses in any other tectal layer. This strongly suggestedthat the retinal ganglion cells that arborized in layer 5bdid not have synaptic endings in other tectal layers.

A striking feature of layer 5b is the marked variation inits thickness in different regions of the optic tectum(Reiner et al., 1982b; Gamlin et al., 1996). Layer 5b, thearea of termination of the retinal ganglion cells from thered field (ventral tectum), was as much as three timesthicker than the terminal area of the yellow field (dorsaltectum). Layers 2, 3, and 4 showed a complementaryopposite variation. We also found that retinal terminalsspan the full thickness of layer 5b in all areas, but thewidth of the arborizations varies markedly depending onthe area, with the yellow field area three times wider thanthe red field area (Karten and Mpodozis, unpublishedobservations). Thus, the overall density of the retinalendings per square m of horizontal plane in layer 5b ofthe ventral tectum was much greater than in layer 5b ofthe dorsal tectum.

This correlates directly with the high density of retinalganglion cells labeled following our CTb injections into theventral tectum (Fig. 10). The diameter of the majority ofthese labeled cells ranged from 4.5 m to 6.0 m (with onlyan occasional cell up to 12 m) and represented a popula-tion of the very smallest ganglion cells in the pigeon retina.The small size of these cells was not a consequence of theirrelative retinal eccentricity, because they showed similarsizes across the entire extent of the red field, from pericen-tral to peripheral regions of the red field. In comparison,the retinal ganglion cells labeled following injections intothe dorsal tectum formed a remarkably less dense, moreheterogeneous population, with cell diameters rangingfrom 5.0 m to 15.0 m. In this population, the small,round cells were also present, but only as a minor fraction.

Thus, based on our results, we propose that 1) in thepigeon, the majority of the retinal ganglion cells terminat-ing in layer 5b was composed of small, round cells withsomal diameters of 4.56.0 m and dendritic fields from25.0 m to 40.0 m in diameter; 2) retinal ganglion cellsthat terminated in layer 5b were distributed across theentire retina and clearly showed a peak in density in thered field area; 3) the tectal projections of these cells endedexclusively in layer 5b, and the larger retinal ganglion

cells in this region terminate in layers 3, 4, and 7; and 4)the small cells were the main, if not the only, source ofretinal afferents to layer 5b. We designated these retinalganglion cells as w-5b, reflecting their small size andtheir tectal target.

Oligo- and polysynapticretinotectorotundal pathways

Our results further support the original suggestion ofRamon (1899) and the recent suggestions and observationsof Hardy et al. (1984, 1985) and Catsicas et al. (1992) thatthe neurons of layer 13 are in direct receipt of retinalinput. The specific dendritic arborizations of type I neu-rons in layer 5b suggest that these neurons may receive amonosynaptic input from the terminals of the w-5b gan-glion cells. However, in view of the complexity of inputs tolayer 5b (see below), confirmation of such an input requiresdirect electron microscopic evidence.

The present study revealed that the anterior and centra-lis divisions of the rotundus may receive a bisynaptic inputfrom the retina, a much more direct and tightly linkedinput than was previously appreciated. A striking featureof this presumptive oligosynaptic pathway was that itappeared to arise from the smallest retinal ganglion cells,the w-5bs, which, in turn, converged on the large type Ineurons of layer 13. This can account for both the largereceptive fields and the fine sensitivity to the motion ofsmall objects that the neurons of the tectofugal pathwayhave been reported to have at the tectal and rotundallevels (Revzin, 1979; Wang et al., 1993; Letvin, personalcommunication).

The polysynaptic retinotectorotundal pathway and thenature of the retinal inputs to the type II neurons were lessclearly defined. Type II neurons showed a dense pattern ofdendritic arborizations within the deep layers of thetectum (layers 1013), but we found no evidence of theextension of these processes to the retinorecipient layers.We suggest, then, that type II neurons may receive polysyn-aptic inputs from the retina, derived from some interven-ing retinorecipient tectal neurons. Our suggestion is consis-tent with the electrophysiological studies of Hardy et al.(1984, 1985) that demonstrate a population of layer 13neurons in receipt of polysynaptic inputs from the retina.

Thus, the anterior and centralis divisions of the rotun-dus presumably receive bisynaptic inputs from the retina,whereas the posterior and triangularis divisions appear toreceive polysynaptic inputs. This difference, in turn, maybe related to the different visual sensory properties thatthe divisions of rotundus exhibit (Revzin, 1979, Grandaand Yazulla, 1971; Wang et al., 1993).

Projections of type I neuronsto the pretectum

This study also found the presence of a tectal projectionto the nucleus pretectalis of the dorsal pretectal region.The pretectalis is a nucleus that projects heavily to boththe ipsilateral and the contralateral optic tectae, terminat-ing specifically within layer 5b (Gamlin et al., 1996).Neurons of the pretectalis are rich in both neuropeptide Y(NPY)- (Gamlin et al., 1996) and GABA-like (Veenman andReiner, 1994) immunoreactivity. The layer 13 projection tothe pretectalis was apparently due solely to the axoncollaterals of type I neurons. Type II neurons did notappear to project to the pretectalis. The implications of thisdifference were most interesting, because the type I neu-


rons possessed dense dendritic arbors within layer 5b.This may suggest that the type I neurons are involved in aunique feedback loop whereby they send presumablyexcitatory, glutamatergic inputs to the pretectalis (Dyeand Karten, 1993), which, in turn, send bilateral inhibi-tory projections to layer 5b. The exact synaptic target ofthe pretectalis projections, however, is still unknown. Dothe pretectalis axons with NPY- and GABA-like immunore-activity terminate on the type I dendrites, the retinalterminals, other afferent axons (see below), or dendrites ofother tectal neurons within layer 5b?

Summary of the tectorotundal connectionsCombining the data from previous reports and our

present study (Dye and Karten, 1993; Gamlin et al., 1996;Mpodozis et al., 1996), we find that the neurons of layer 13of the tectum influence a number of pathways that affectrotundal function. 1) Type I neurons project bilaterally tothe anterior and centralis divisions of the rotundus. 2)Type II neurons project bilaterally to the posterior andtriangularis divisions. 3) Both type I and type II neuronsproject to the ipsilateral intrinsic nuclei of the tectotha-lamic tract, with collaterals of type I neurons ending in thenucleus posteroventralis thalami (PV), the nucleus subpre-tectalis (SP), and the nucleus interstitiopretectosubpretec-talis (IPS) and with collaterals of type II neurons ending inthe PV and the SP, pars caudalis (SPcd). 4) The GABAergicneurons of the SP and IPS terminate in the anterior andcentralis divisions of the ipsilateral rotundus. Neurons ofthe SPcd terminate in the posterior and triangularisdivisions of the ipsilateral rotundus. 5) Type I neurons alsoproject to the nucleus pretectalis, terminating mainlyipsilaterally. The pretectalis neurons with NPY- and GABA-like immunoreactivity project bilaterally to layer 5b of thetectum, the tectal layer in which w-5b retinal terminalsand type I dendrites are found in close, possibly synaptic,association. This closed loop potentially provides a pro-found inhibitory effect on throughput processing of thetectum on the rotundus. 6) Both types of layer 13 neuronspresumably operate through a glutamatergic, excitatorypathway.

Tectorotundal vs. tectobulbar pathwaysA previous study of the tectal efferents reported that

individual neurons of layer 13 as well as neurons fromdeeper layers are the source of the descending projectionsto the tectobulbar system of the brainstem (Reiner andKarten, 1982). This could imply a direct output of thoseneurons to both premotor targets and the tectorotundoecto-striatal system. The present study clearly revealed thatlayer 13 contained at least three different populations ofneurons: type I and II neurons that projected to therotundus and an additional type of neuron that projectedto the tectobulbar system. Neurons of this additional typewere found occasionally at the border of layers 13 and 14(where they were found intermingled with type I or IIneurons), although they were located mainly in layer 14and the underlying tegmentum. Some of these neurons,however, were found to have dendrites that extended intothe retinorecipient layers of the tectum. These neuronswere smaller than those that projected to the rotundus. Afuture study will address the issue of the identity of tectalneurons that contribute to the tectopontine projections tothe lateral pontine nucleus.

Nonvisual afferents to layer 13The nature of the inputs to the layer 13 neurons will

obviously determine the type of information conductedrostrally within the tectofugal visual pathway. A variety ofprojections to the layer 13 neurons have been reported,including those directly from the telencephalon (Edingeret al., 1903; Zeier and Karten, 1971; Karten et al., 1973;Miceli and Reperant, 1983), indirectly from the paleostria-tum primitivum of the telencephalon via the nucleusspiriformis lateralis (Reiner et al., 1982a), possible inputsof the nucleus semiluminaris, inputs from the locus coer-uleus, possible inputs from raphe neurons (Brecha, 1978),auditory inputs via the inferior colliculus external nucleus(Knudsen and Knudsen, 1983), possible somatosensoryinputs (Cotter, 1976), and inputs from the reticular forma-tion (Brecha, 1978). We have been unable to find anyevidence of projections to layer 13 from the contralateraloptic tectum.

The demonstration that layer 13 consists of a clear innerand outer sublamina may prompt reconsideration of thedifferent inputs to each of these regions. This clarificationis required to determine where the auditory and somatosen-sory inputs actually terminate (Knudsen and Knudsen,1983; Stein and Meredith, 1993). Thus, there may well be aconvergent polymodal input to the rotundus. However,there may be intrinsic segregation of connections byindependent pathways within and without the tectum.

Diversity of inputs to layer 5bThe present study further revealed the complex nature

of inputs to layer 5b of the avian optic tectum. Layer 5b isin receipt of 1) massive retinal glutamatergic (Dye andKarten, 1996) inputs, presumably from the small w-5bneurons; 2) rich dendritic arbors of type I neurons; 3) densebilateral GABA/NPY inputs from the nucleus pretectalis(Gamlin et al., 1996); 4) massive inputs from the nucleusisthmi parvocellularis (Ramon, 1899) that were foundsubsequently to be cholinergic (Medina and Reiner, 1994);and 5) distinctive enkephalin-immunoreactive radial pro-cesses, possibly arising from the enkephalin-immunoreac-tive neurons of tectal layer 10 (Reiner et al., 1982b). Layer5b is also prominently calbindin immunoreactive, which isderived from a distinct population of horizontal neuronslying within layer 5b (Mpodozis and Karten, unpublishedobservations). However, in the absence of details regardingthe specific synaptic relationships of the terminals of theretina, the nucleus pretectalis, the nucleus isthmi parvocel-lularis, and the type I dendrites, we cannot assess theconsequences of these interactions upon the nature of theinformation being conveyed to the nucleus rotundus(Fig. 12).

Retinotectorotundal topographyDespite the seeming lack of a precise retinal topography

within the tectorotundal system, we have confirmed andextended the existence of a regional topographic organiza-tion in this system, i.e., neurons located in differentsublaminae of the tectum projected to different divisions ofthe rotundus. But another notable feature emerged inthese studiesthese tectal sublaminae showed regionalvariation. The number of type I neurons varied markedlybetween the dorsal and ventral tectum in a way thatfollows the regionalization of the retinal inputs to layer 5b.The number of type II neurons, however, was relatively


similar in all tectal areas, although the numbers weresomewhat reduced in the ventral tectum. The conse-quences of this would be manifest in the predominance of ared field representation in the anterior and centralisdivisions of rotundus, because they were in specific receiptof type I tectal neurons. In contrast, the posterior andtriangularis divisions of rotundus that received theirinputs from the type II neurons should possess a moreuniform representation of all portions of the visual field.

However, the absence of a correspondingly precise reti-nal topography in the tectorotundal system continues topresent a problem: How does this system deal with theprecise localization of stimuli within the global receptivefield? We hope to address this issue in our subsequentstudies, which will describe in more detail the morphologyof layer 13 neurons as well as the topography of theretinotectorotundal connectivity.

ACKNOWLEDGMENTSThe authors gratefully acknowledge the excellent techni-

cal assistance of Agnieszka Brzozowska-Prechtl. We alsothank Koichi Naruishi and Ginger Phelps for their assis-tance with the figures and Ellie Watelet for her administra-tive assistant assistance. This work was supported by

grants Fondecyt 1081 to J.M. and NEI EY-06890 andNINDS NS 24560 to H.J.K.

LITERATURE CITEDBenowitz, L.I., and H.J. Karten (1976) Organization of the tectofugal visual

pathway in the pigeon: A retrograde transport study. J. Comp. Neurol.167:503520.

Brecha, N. (1978) Organization of the avian optic tectum: Afferent, projec-tion systems (Ph.D. dissertation). State University of New York atStony Brook.

Britto, L.R.G., A.S. Torrao, D.E. Hamassaki-Britto, J. Mpodozis, K.T.Keyser, J.M. Lindstrom, and H.J. Karten (1994) Effects of retinallesions upon the distribution of nicotinic acetylcholine receptor sub-units in the chick visual system. J. Comp. Neurol. 350:473484.

Catsicas, M., Y. Pequignot, and P.G. Clarke (1992) Rapid onset of neuronaldeath induced by blockade of either axoplasmic transport or actionpotentials in afferent fibers during brain development. J. Neurosci.12:46424650.

Cotter, J.R. (1976) Visual and nonvisual units recorded from the optictectum of Gallus domesticus. Brain Behav. Evol. 13:121.

Cowan, W.M., L. Adamson, and T.P.S. Powell (1961) An experimental studyof the avian visual system. J. Anat. (London) 95:545563.

Diamond, I.T. (1973) The evolution of the tectal-pulvinar system inmammals: Structural and behavioral studies of the visual system.Symp. Zool. Soc. London 33:205333.

Dye, J.C., and H.J. Karten (1993) An in vitro slice preparation of tectofugalvisual pathways in the chick. Soc. Neurosci. Abstr. 19:767.

Dye, J.C., and H.J. Karten (1996) An in vitro study of retinotectaltransmission in the chick: Role of glutamate and GABA in evoked fieldpotentials. Vis. Neurosci. 13:747758.

Edinger, L., A. Wallenberg, and G.M. Holmes (1903) Untersuchungen uberdie vergleichende Anatomie des Gehirns. 5. Das Vorderhirn der Vogel.Abhandl. d. Sneckenb. nat. Gesellsch. Frankfurt am Main 20:343426.

Ehrlich, D., K.T. Keyser, and H.J. Karten (1987) Distribution of substanceP-like immunoreactive retinal ganglion cells and their pattern oftermination in the optic tectum of chick (Gallus gallus). J. Comp.Neurol. 266:220233.

Gamlin, P.D.R., A. Reiner, K.T. Keyser, N. Brecha, and H.J. Karten (1996)Projection of nucleus pretectalis to a retinorecipient tectal layer in thepigeon (Columba livia). J. Comp. Neurol. 368:424438.

Gerfen, C.R., and P.E. Sawchenko (1983) An anterograde neuroanatomicaltracing method that shows the detailed morphology of neurons, theiraxons and terminals: Immunohistochemical localization of an axonallytransported plant lectin Phaseolus vulgaris leucoagglutinin. Brain Res.290:219238.

Granda, A.M., and S. Yazulla (1971) The spectral sensitivity of single unitsin the nucleus rotundus of pigeon, Columba livia. J. Gen. Physiol.57:363384.

Hall, W.C. and F.F. Ebner (1970) Thalamotelencephalic projections in theturtle (Pseudemys scripta). J. Comp. Neurol. 140:101122.

Hardy, O., Leresche, N., and D. Jassik-Gerschenfeld (1984) Postsynapticpotentials in neurons of the pigeons optic tectum in response to afferentstimulation from the retina and other visual structures: An intracellu-lar study. Brain Res. 311:6574.

Hardy, O., Leresche, N., and D. Jassik-Gerschenfeld (1985) Morphology andlaminar distribution of electrophysiologically identified cells in thepigeons optic tectum: An intracellular study. J. Comp. Neurol. 162:447466.

Hunt, S.P., and H. Kunzle (1976) Observations on the projections andintrinsic organization of the pigeon optic tectum: An autoradiographicstudy based on anterograde and retrograde, axonal and dendritic flow.J. Comp. Neurol. 170:153172.

Iniguez, C., M.J. Gayoso, and J. Carreres (1985) A versatile and simplemethod for staining nervous tissue using Giemsa dye. J. Neurosci.Methods 13:7786.

Karten, H.J., and W. Hodos (1967) A Stereotaxic Atlas of the Brain of thePigeon (Columba livia). Baltimore, MD: Johns Hopkins Press.

Karten, H.J., and W. Hodos (1970) Telencephalic projections of the nucleusrotundus in the pigeon (Columba livia). J. Comp. Neurol. 140:3552.

Karten, H.J., and A.M. Revzin (1966) The afferent connections of thenucleus rotundus in the pigeon. Brain Res. 2:368377.

Karten, H.J., W. Hodos, W.J.H. Nauta, and A.M. Revzin (1973) Neuralconnections of the visual wulst of the avian telencephalon. Experimen-

Fig. 12. Schematic summary of the two main pathways within theretinotectorotundal pathway. Only the ipsilateral connections areshown. Dendrites of type I tectal neurons (top) arborize within layer5b of the tectum and are presumably in receipt of monosynapticterminals of w-5b ganglion cells. Axons of type I neurons projectbilaterally to the dorsal anterior and centralis divisions of the rotun-dus. Dendrites of type II neurons (bottom) do not extend to theretinorecipient layers of the tectum. They receive polysynaptic inputsfrom retinal ganglion cells. Axons of type II neurons project bilaterallyto the posterior and triangularis divisions of the rotundus. In additionto the projections directly to the rotundus, both type I and type IItectal neurons also send collaterals to the g-aminobutyric acid (GAB-A)ergic intrinsic nuclei of the tectothalamic tract. Type I neuronsarborize in the SP and the IPS. Type II neurons arborize in the nucleussubpretectalis, pars caudalis (SPcd). The intrinsic nuclei project to therotundus, as described in Mpodozis et al. (1996). Type I neurons alsoterminate in the pretectalis. The pretectalis projects back to layer 5bbilaterally. Tri, nucleus rotundus, pars triangularis; for other abbrevia-tions, see list.


tal studies in the pigeon (Columba livia) and owl (Speotyto cunicularia).J. Comp. Neurol. 150:253278.

Karten, H.J., K.T. Keyser, and N.C. Brecha (1990) Biochemical andmorphological heterogeneity of retinal ganglion cells. In B. Cohen and I.Bodis-Wollner (eds): Vision and the Brain: The Organization of theCentral Visual System. New York: Raven Press, pp. 1933.

Karten, H.J., K. Cox, J. Mpodozis, H.-J. Bischof, and T. Shimizu (1993)Little cells ending on big cells: An oligosynaptic retino-tecto-pulvinarsystem in pigeon. Soc. Neurosci. Abstr. 19:969.

Knudsen, E.I., and P.F. Knudsen (1983) Space-mapped auditory projectionsfrom the inferior colliculus to the tectum in the barn owl (Tyto alba). J.Comp. Neurol. 218:187196.

LaVail, J.H., and W. M. Cowan (1971) The development of the chick optictectum. I. Normal morphology and cytoarchitectonic development.Brain Res. 28:391419.

Lettvin, J.Y., H.R. Maturana, W.S. McCulloch, and W.H. Pitts (1959) Whatthe frogs eye tells the frogs brain. Proc. Inst. Radio Eng. N.Y.47:19401951.

Martinez-de-la-Torre, M., S. Martinez, and L. Puelles (1990) Acetylcholines-terase-histochemical differential staining of subdivisions within thenucleus rotundus in the chick. Anat. Embryol. 181:129135.

Medina, L., and A. Reiner (1994) Distribution of choline acetyltransferaseimmunoreactivity in the pigeon brain. J. Comp. Neurol. 342:497537.

Miceli, D., and J. Reperant (1983) Hyperstriatal-tectal projections in thepigeon (Columba livia) as demonstrated by the retrograde double-labelfluorescence technique. Brain Res. 276:147153.

Mpodozis, J., Letelier, J.C., Concha, M.L., and Maturana, H. (1995)Conduction velocity groups in the retino-tectal and in the retino-thalamic visual pathways of the pigeon (Columba livia). Int. J. Neuro-sci. 81:123136.

Mpodozis, J., Cox, K., Shimizu, T., Bischof, H.-J., Woodson, W., and Karten,H.J. (1996) GABAergic inputs to the nucleus rotundus (pulvinarinferior) of the pigeon (Columba livia). J. Comp. Neurol. 374:204222

Northcutt, R.G., and A.B. Butler (1980) Projections of the optic tectum inthe longnose gar, Lepisosteus osseus. Brain Res. 190:333346.

Pritz, M.B. (1975) Anatomical identification of a telencephalic visual areain crocodiles: Ascending connections of nucleus rotundus in Caimancrocodilus. J. Comp. Neurol. 164:32338.

Ramon, P. (1899) Adiciones a nuestros trabajos sobre los centros opticos delas aves. Rev. Trimestr. Micrograf. 4:7786.

Ramon y Cajal, S. (1911) Histologie de Systeme Nerveux de lHomme et desVertebres. Paris: Maloine.

Reiner, A., and H.J. Karten (1982) Laminar distribution of the cells of originof the descending tectofugal pathways in the pigeon (Columba livia). J.Comp. Neurol. 204:16587.

Reiner, A., N.C. Brecha, and H.J. Karten (1982a) Basal ganglia pathways tothe tectum: The afferent and efferent connections of the lateral spiri-form nucleus of pigeon. J. Comp. Neurol. 208:1636.

Reiner A., H.J. Karten, and N.C. Brecha (1982b) Enkephalin-mediatedbasal ganglia influences over the optic tectum: Immunohistochemistryof the tectum and the lateral spiriform nucleus in pigeon. J. Comp.Neurol. 208:3753.

Revzin, A.M. (1979) Functional localization in the nucleus rotundus. InA.M. Granda and J.H. Maxwell (eds): Neural Mechanisms of Behaviorin The Pigeon. New York: Plenum Press, pp. 165175.

Revzin, A.M., and H.J. Karten (1967) Rostral projections of the optic tectumand the nucleus rotundus in the pigeon. Brain Res. 3:264276.

Stein, B.E., and M.A. Meredith (1993) The Merging of The Senses.Cambridge: MIT Press.

Stone, J., and J.A. Freeman (1971) Synaptic organization of the pigeonsoptic tectum: A Golgi and current source-density analysis. Brain Res.27:203221.

Tombol, T., G. Egedi, and A. Nemeth (1994) Phaseolus vulgaris lectinlabeled and GABA immunogold stained terminals in nucleus rotundus.An EM study. J. Hirnforsch 35:233252.

Veenman, C.L., and A. Reiner (1994) The distribution of GABA-containingperikarya, fibers, and terminals in the forebrain and midbrain ofpigeons, with particular reference to the basal ganglia and its projectiontargets. J. Comp. Neurol. 339:209250.

Wang, Y.C., S. Jiang, and B.J. Frost (1993) Visual processing in pigeonnucleus rotundus: Luminance, color, motion, and looming subdivisions.Vis. Neurosci. 10:2130.

Zeier, H., and H.J. Karten (1971) The archistriatum of the pigeon:Organization of afferent and efferent connections. Brain Res. 31:313326.
