localization of the origin of retinal efferents in the turtle brain and the involvement of nitric...

Localization of the Origin of RetinalEfferents in the Turtle Brain and theInvolvement of Nitric Oxide Synthase

SILKE HAVERKAMP AND WILLIAM D. ELDRED*Department of Biology, Boston University, Boston, Massachusetts 02215

ABSTRACTPrevious studies have used selective neurochemical markers or retrograde tracers to

localize the cells in the brain giving rise to efferents to the turtle retina. Because of the relativeselectivity of the neurochemical markers or the lack of sensitivity of the previously employedretrograde tracers, these studies did not locate all the efferent cell bodies, or they could notdescribe the anatomy of the efferent cells. In the present study, cholera toxin B was used as ahighly sensitive retrograde tracer to investigate the distribution, number, and morphology ofthe retinal efferent or centrifugal cell system in turtle brain. Previous studies of the turtleretina have indicated that nitric oxide synthase may be found in some retinal efferents.Therefore, we also did colocalization studies of the retrograde tracer with reduced nicotin-amide adenine dinucleotide phosphate (NADPH)-diaphorase histochemistry to investigatenitric oxide as a possible transmitter used by efferent fibers and to localize these NADPH-diaphorase-positive efferent cell bodies in the turtle brain. We found that each eye receivedprojections from approximately 40 efferent cell bodies that were located primarily in thecontralateral midbrain. The majority of efferent cell bodies were centered in the isthmictegmentum; other efferent cells extended more rostrally into the substantia nigra, and someefferent cells extended more caudally into the nucleus raphes superior. The double-labelresults showed that 30% of the cholera toxin B-like immunoreactive cells were also positive forNADPH-diaphorase. The location of these double-labeled cells around the locus coeruleuscorresponded to the NADPH-diaphorase-positive efferent cells in the avian isthmo-optic field.The localization of NADPH-diaphorase in these efferents indicated that they may use nitricoxide to modulate retinal function. J. Comp. Neurol. 393:185–195, 1998. r 1998 Wiley-Liss, Inc.

Indexing terms: retina; centrifugal; NADPH-diaphorase; chlorea toxin B: double-label

The existence of efferent innervation from the brain tothe retina has been demonstrated in many vertebratespecies including the turtle (Reperant et al., 1989; Uchi-yama, 1989). Although many studies have investigated theanatomy and physiology of these projections, the func-tional role of the efferent system is still unclear. The avianefferent projection is highly developed and has been exten-sively studied. It arises from the isthmo-optic nucleus(ION), located in the caudal mesencephalon, and from agroup of neurons called the ectopic isthmo-optic cells,scattered around the ION (Cowan, 1970; Hayes and Web-ster, 1981). Electrical stimulation of the avian ION altersthe receptive field properties of retinal ganglion cells(Miles, 1972). Similar results are found in the turtle, inwhich stimulation of efferent fibers produces excitatoryeffects on amacrine and ganglion cells (Cervetto et al.,1976; Marchiafava, 1976).

In the most complete anatomical study of efferents in theturtle to date, Schnyder and Kunzle (1983) calculate a

maximum of 40 cells in the mesencephalic reticular area ofthe turtle brain project to each eye; however, the quality oftheir labeling did not allow detailed study of the cellularmorphology of these neurons. By using immunohistochemi-cal methods in the turtle, somata in the brain that projectto the retina have been shown to contain enkephalin(Weiler, 1985) or serotonin (Schutte and Weiler, 1988), andaspartate-positive efferent fibers have been shown to leavethe optic nerve and arborize in the retina (Yaqub andEldred, 1991). Additionally, reduced nicotinamide adeninedinucleotide phosphate (NADPH)-diaphorase and nitric

Grant sponsor: National Eye Institute; Grant number: EY04785-14.*Correspondence to: Dr. William D. Eldred, Department of Biology,

Boston University, 5 Cummington St., Boston, MA 002215.E-mail: [email protected]

Received 16 July 1997; Revised 7 November 1997; Accepted 10 November1997

THE JOURNAL OF COMPARATIVE NEUROLOGY 393:185–195 (1998)

r 1998 WILEY-LISS, INC.

oxide synthase (NOS)-positive efferent fibers have beenreported in the turtle retina (Blute et al., 1997), but theorigin of these fibers in the brain was not determined. Inthe chicken, retinal efferents are also positive for NADPH-diaphorase staining, and nitric oxide (NO) is postulated tobe involved in efferent function (Morgan et al., 1994).

The aims of the present study were the following: 1) todetermine the distribution of efferent cells in the turtlebrain by using intraocular injections of the highly sensi-tive tracer cholera toxin B (CTB; Reiner et al., 1996); and2) to localize NADPH-diaphorase activity in these CTB-like immunoreactive (CTB-LI) efferent cells. Establishingthe localization of the NADPH-diaphorase-containing effer-ent cell bodies in the turtle brain will permit future studiesto examine the effects of the selective activation of theseNADPH-diaphorase-positive efferent projections on reti-nal function.

MATERIALS AND METHODS

Tissue preparation

The present study examined red-eared turtles, Pseud-emys scripta elegans, carapace length 15–25 cm main-tained on a 12/12 hour light/dark cycle. The animal useand care protocols used in this study were approved by theBoston University Charles River Campus InstitutionalAnimal Care and Use Committee. Following anesthesiawith ketamine (0.66 ml/kg) and xylazine (0.33 ml/kg), eachturtle was injected intraocularly with 5 µl of 1% CTB inone eye by using a Hamilton syringe and allowed tosurvive for 72–96 hours. The CTB (List Biological Labora-tories, Campbell, CA) was diluted in distilled water contain-ing 2% dimethyl sulfoxide (DMSO) or in 0.05 M phosphatebuffer (PB) at pH 7.4 with 10% DMSO. Following thesesurvival times, the turtles were killed by intracoelomicinjection of 100 mg/kg of sodium pentobarbital. Immedi-ately after death, the turtles were perfused transcardiallywith 0.9% NaCl and then with 4% paraformaldehyde in 0.1M PB at pH 7.4. For each turtle, the brain was isolated,postfixed for 2 hours in the same fixative, graduallyinfiltrated with increasing sucrose concentrations, andsoaked in 30% sucrose in PB overnight at 4°C. The brainwas then frozen and serially sectioned into 50 µm thickcross sections by using a sliding microtome with a freezingstage; the sections were subsequently processed for immu-nohistochemical visualization of the CTB.

The numbers of efferent cells were quantified in detailfor eight animals by counting all the labeled cells incomplete sets of serial sections taken through the entireturtle brain. The selective depth of focus obtained using a603 oil immersion lens and the 50 µm thickness of thesections allowed us to examine carefully the somata ofeach of the labeled cells. Through detailed analysis ofadjacent serial sections we were able to eliminate thepossibility that any labeled cells were also counted in theadjacent serial section. A combination of a careful analysisof all of the serial sections, the relatively sparse spacing ofthe cells, and the unique dendritic arborizations of thecells prevented any ambiguous cell counts.

Immunohistochemical processing

Immunohistochemical labeling was done by using stan-dard methods and the following sequence of reagents: 1)5% normal donkey serum in 0.3% Triton X-100/PB (PBTX)for 1 hour at room temperature; 2) a goat-derived primaryanti-CTB antibody (List Biological Laboratories), diluted1:10,000 in PBTX overnight at 4°C; 3) a donkey-derivedbiotinylated antigoat-IgG (Jackson ImmunoResearch Labo-ratories, West Grove, PA) diluted 1:200 or 1:1,000 in PBTXfor at least 2 hours at room temperature; and 4) VectorElite ABC reagents (Vector, Burlingame, CA) diluted1:1,000 in PBTX for 2 hours at room temperature orovernight at 4°C. The labeling was visualized by usingImmunopurey metal-enhanced diaminobenzidine (DAB)substrate (Pierce, Rockford, IL). Between each of thesereagent steps, the tissue was washed 3 3 10 minutes withPB. Following immunohistochemical processing, the sec-tions were mounted on slides coated with chrome andgelatin and then coverslipped.

Colocalization studies

In studies colocalizing NADPH-diaphorase enzyme activ-ity and CTB, the fixation time of the brains was reduced to45 minutes. The double labeling was examined by usingeither peroxidase or fluorescein isothiocyanate (FITC)histochemistry to localize CTB-positive cells in a total ofseven animals. The peroxidase-labeled cells were preparedas previously described. To produce the FITC-labeled cells,the primary anti-CTB antibody was used at 1:500; thesecondary antibody was an FITC-conjugated donkey-antigoat IgG (Jackson ImmunoResearch Laboratories),used at 1:100. After a 2 hour incubation with the secondary

Abbreviations

Cb cerebellumCbL nucleus cerebellaris lateralisCbM nucleus cerebellaris medialisdIV decussatio nervi trochlearisFLM fasciculus longitudinalis medialisFRL formatio reticularis lateralis mesencephaliGC griseum centraleGCL granule cell layerImc nucleus isthmi pars magnocellularisImr nucleus isthmi magnocellularis pars rostralisIP nucleus interpeduncularisIpc nucleus isthmi pars parvocellularisLa nucleus laminaris of the torus semicircularisLoC locus coeruleusLL lateral lemniscusML molecular layerMV nucleus mesencephalicus nervi trigeminimV nucleus motorius nervi trigemininIII nucleus nervi oculomotorius

nIV nucleus nervi trochlearisnPM nucleus profundus mesencephaliPb nucleus parabrachialisPrV nucleus princeps nervi trigeminiPV peduncularis ventralis fasciculi prosencephali lateralisRas nucleus raphes superiorRis nucleus reticularis isthmiRs nucleus reticularis superiorSGC stratum griseum centraleSGF stratum griseum et fibrosum superficialeSGP stratum griseum profundumSN substantia nigraSO stratum opticumTeO tectum opticumTSC torus semicircularisTT tractus tectothalamicusVeL nucleus vestibularis lateralisVeS nucleus vestibularis superior

186 S. HAVERCAMP AND W.D. ELDRED

antibody at room temperature, the sections were washed,mounted on slides and coverslipped with Vectashieldy(Vector Laboratories). Labeled cells were photographed byusing an Olympus Vannox microscope equipped with epi-fluorescence illumination. To establish landmarks for thecolocalizations with NADPH-diaphorase, we used a charge-coupled device (CCD) camera (Javelin, Los Angeles, CA) incombination with Image Proy image analysis software(Media Cybernetics, Silver Springs, MD) to examine andrecord both fluorescent and brightfield images of the cellsand other landmarks.

NADPH-diaphorase histochemistry

Following photography of the FITC- or peroxidase-labeled CTB-LI cells in the brain, the coverslips wereremoved and the slides were washed overnight in PB. Thenext day, the sections were incubated in a solution of1 mg/ml NADPH with 0.1 mg/ml of nitroblue tetrazolium(NBT) in PBTX for 1–2 hours at 37°C in dark. Thehistochemical reaction was periodically monitored visuallyand the reaction was stopped by using 3 3 10 minutewashes in PB and a final 5 minutes in distilled H2O. Theslides were then coverslipped with glycerin, and theNADPH-diaphorase staining was compared with the pho-tographs and digital images of the retrogradely labeledCTB-LI cells. The terminology for the regions and nuclei ofthe turtle midbrain is based on a previous study (Powersand Reiner, 1993).

RESULTS

Distribution of efferent cellsin the turtle brain

In all the successful intraocular injections of CTB, theanterograde transport of the tracer heavily labeled all thebrain areas known to receive primary visual input (Reineret al., 1996). There was no indication of any transsynaptictransport of the CTB. In addition, there were clear retro-gradely labeled neurons in the caudal mesencephalon. Thenumbers of CTB-LI efferent cell bodies we found in theturtle brain in eight different experiments are summa-rized in Table 1. In each case, approximately 40 efferentcell bodies were found to project to each eye. The majorityof these cells were located in the contralateral half of thebrain (32–58 cells), with many fewer cells located in theipsilateral half of the brain (0–4 cells).

Drawings of transverse sections through the midbrainand rostral rhombencephalon of the turtle brain illustrat-ing the distribution of CTB-LI cell bodies are shown inFigure 1. The efferent neurons were located in a largeregion primarily within the caudal mesencephalon, butthey also extended into the rostral rhombencephalon. The

rostral-caudal extension of the labeled cells was about 4mm in animals with a carapace size of about 20 cm. Most ofthe cells were found in the isthmic tegmentum around thesubstantia nigra (SN) and in the more caudal isthmicregion between the locus coeruleus (LoC), the laterallemniscus (LL), and the nucleus reticularis isthmi (Ris).Additional cells were observed in the nucleus raphessuperior (Ras). In some cases, we also found labeled cells inthe formatio reticularis lateralis mesencephali (FRL), inthe nucleus isthmi pars parvocellularis (Ipc), and in therostral portion of the hindbrain in the nucleus reticularissuperior (Rs).

Anatomy of the efferent cells

The efferent cells were generally multipolar or bipolar inshape, with cell somata ranging in diameter from 10 to 20µm (Figs. 2–6). Cells with either soma shape usually hadlong, smooth processes that branched sparsely, formingrelatively simple dendritic arborizations. Some dendritescould be followed for at least 500 µm in a single brainsection. The dendrites of these neurons had no dendriticspines and no swellings resembling synaptic boutons. Themultipolar shaped cells had between three and five pri-mary dendrites and a soma 15–20 µm in diameter. Some ofthe primary processes of the multipolar cells were quitestout; however, these processes quickly branched to formmore delicate secondary processes that tapered only slightlyuntil they terminated. The bipolar or fusiform cells hadtwo primary dendrites and usually a smaller cell soma10–15 µm in diameter. In some cases one of these bipolarprimary processes was stouter than the other; with thestouter process forming secondary dendrites further fromthe soma than did the thinner primary process. Themultipolar- and bipolar-shaped somata were found inter-spersed with one another.

Localization of NADPH-diaphorasein this region of turtle brain

The detailed localization of both NADPH-diaphoraseand NOS immunoreactivity in the turtle brain have beenpreviously described (Bruning et al., 1994), and our pres-ent results were similar. Many strongly NADPH-diapho-rase-positive neurons were scattered in the LoC, and someweakly NADPH-diaphorase-positive cells and fibers werescattered in the caudal part of LL (Fig. 9). In a more rostralsection (Fig. 7), the NADPH-diaphorase enzyme activity ofthe LL was much stronger, and many smaller neuronswere scattered in the area of the Ris. The Ipc was also filledby an accumulation of intensely stained fibers. However, inthe nucleus isthmi pars magnocellularis (Imc), in thefasciculus longitudinalis medialis (FLM), and in the Ras,no NADPH-staining was observed. The section in Figure 7is comparable with the transverse section P 3.2 in Figure 1.

Colocalization study of CTB-LI cellswith NADPH-diaphorase histochemistry

The double labeling was successfully examined usingboth FITC and peroxidase immunohistochemistry to local-ize the CTB-like immunoreactivity. Figures 11–16 showexamples of FITC-labeled efferent cells that were alsoNADPH-diaphorase positive. The left column shows theFITC-labeled CTB-LI cells and the right column demon-strates the same cells after NADPH-diaphorase histochem-istry. The cell shown in Figures 11 and 12 was found at theborder of the most caudal part of the LoC. The precise

TABLE 1. Numbers of Cholera Toxin B-like Immunoreactive EfferentCells Found in the Turtle Retina

Animal no. Contralateral brain Ipsilateral brain Total/eye

1 37 2 392 36 1 373 41 1 424 58 2 605 32 0 326 38 3 417 34 1 358 38 4 42Mean 6 SD 39 6 8 1.8 6 1.3 41 6 8

EFFERENTS TO THE TURTLE RETINA 187

Fig. 1. Drawings of transverse sections through the midbrain and rostral rhombencephalon of theturtle illustrating the distribution of cholera toxin B-like immunoreactive (CTB-LI) cell bodies (asterisks).The numbers below the sections indicate the stereotaxic level of the sections (Powers and Reiner, 1993).For abbreviations, see list.


Figs. 2–6. Fig. 2: Low-magnification photomicrograph showingscattered CTB-LI cells. Figs. 3–5: Individual CTB-LI cells, which areeither bipolar (Figs. 4 and 5) or multipolar (Fig. 3) in shape, that giverise to long smoothly tapering processes. Fig. 6: Camera lucidadrawings of two efferent cell bodies, a typical bipolar (left) and

multipolar (right) cell labeled with CTB. These cells have relativelysimple dendritic arborizations with two (left) or three (right) primarydendrites. The cell in Figure 5 is the same cell in the left camera lucidadrawing in Figure 6. Scale bars 5 50 µm in Figure 2, 25 µm in Figures3–5, 20 µm in Figure 6.


location of this specific cell in relation to other NADPH-diaphorase-positive cells in this region is shown in Figures9 and 10 (arrows). The cell shown in Figures 13 and 14 wasalso found in this same region. Figure 15 shows twoFITC-labeled cells. The double labeling of CTB-like immu-noreactivity and NADPH-diaphorase histochemistryshowed that only one cell was NADPH-diaphorase positive(large arrows, Figs. 15, 16); the other cell was NADPH-diaphorase negative (small arrows, Figs. 15, 16). Thesecells were found in a brain section comparable with thesection shown in Figure 7 and were located in the isthmicregion adjacent to the LoC.

Examples of cells double-labeled using a combination ofperoxidase-labeled CTB immunohistochemistry andNADPH-diaphorase histochemistry are shown in Figures17–20. All six efferent cell bodies shown in Figures 17 and

19 were found in brain sections taken through the isthmiccell field in or around the LoC (Fig. 8). Five of these CTB-LIcells were NADPH-diaphorase positive (large arrows, Figs.18, 20), whereas one was not double-labeled (small arrows,Figs. 19, 20). In all cases, the dark blue formazan reactionproduct of the NADPH-diaphorase histochemistry in thedouble-labeled cells was easily distinguished from thelighter, punctate brown reaction product of DAB in cellsthat were not double-labeled. The DAB reaction productwas primarily confined to the cell bodies, while the forma-zan NADPH-diaphorase reaction product extended muchfurther into the primary dendrites of the labeled cells.

In summary, 30% of the CTB-LI efferent cells weredouble-labeled. These NADPH-diaphorase-positive cellswere localized in the isthmic region, in or adjacent to theLoC (Figs. 7, 9 and Fig. 1, P 3.2). In many cases, it was not

Figs. 7–10. Localization of NADPH-diaphorase in the isthmicregion of the turtle midbrain. The section in Figure 7 corresponds to P3.2 in Figure 1, and the section in Figure 9 corresponds to P 3.5 (notshown in Fig. 1). Figs. 8 and 10: More details of the NADPH-diaphorase-positive stained cells scattered in the locus coeruleus (LoC). The

arrows in Figures 7 and 8 mark one of the same CTB-LI cells shown athigher magnification in Figures 19 and 20. The same CTB-LI cellmarked by an arrow in Figures 9 and 10 is also shown at highermagnification in Figures 11 and 12. For abbreviations see list. Scalebars 5 250 µm in Figures 7 and 9, 100 µm in Figures 8 and 10.


Figs. 11–16. Examples of fluorescein isothiocyanate (FITC)-labeled CTB-LI/NADPH-diaphorase double-labeled cells. The leftcolumn shows the FITC-labeled CTB-LI cells, and the right columndemonstrates the same cells after NADPH-diaphorase histochemistry.The same cell shown in Figures 11 and 12 (arrows), which was found atthe border of the most caudal part of the LoC, is also shown in Figures9 and 10. The cells shown in Figures 13 and 14 were also found in this

same region of the brain. The two FITC-labeled CTB-LI cells in Figure15 (arrows) were found in a section corresponding to P 3.2 in Figure 1,and these cells were also located adjacent to the LoC. The doublelabeling with NADPH-diaphorase histochemistry and CTB immuno-histochemistry indicated that only one of these cells was NADPH-diaphorase-positive (large arrow, Figs. 15 and 16); the other cell hadonly CTB-LI (small arrow, Figs. 15 and 16). Scale bars 5 25 µm.


possible to determine whether these efferent cells wereintermixed with cells of the LoC or were just adjacent tothe LoC. Many of the nonefferent NADPH-diaphorase-positive cells in the LoC appeared to have very similarmorphology to the NADPH-diaphorase-positive/CTB-LIefferent cells.

DISCUSSION

Efferent projections from the turtle brain

In previous studies, although efferent cell bodies arefound in both the ipsilateral and contralateral sides of theturtle brain, the quality of labeling did not allow detailedstudy of the cellular morphology of these neurons (Schnyderand Kunzle, 1983; Weiler, 1985; Schutte and Weiler, 1988).In the present study, our use of CTB as a highly sensitive

retrograde tracer has provided new details on the num-bers, locations, and cellular morphology of these efferentneurons. Our results indicated that each eye can receiveprojections from approximately 40 efferent cells, the major-ity of which were located in the contralateral midbrain.This number correlates closely with the approximately 40efferent somata that Schnyder and Kunzle (1983) calculateto project to each eye of the turtle.

It is interesting to note that although similar smallnumbers of ipsilaterally located efferent somata have beendescribed in both our study and previous studies (Schnyderand Kunzle, 1983; Weiler, 1985; Schutte and Weiler, 1988),we find many more contralaterally located somata thanare found in some of the previous studies. The consistencyin the small numbers of ipsilateral cells suggests thatthere are in fact relatively few ipsilateral projections.

Figs. 17–20. Examples of peroxidase-labeled CTB-LI/NADPH-diaphorase double-labeled cells. All six of the efferent cell bodiesshown in Figures 17 and 19 (arrows) were found in the brain regionmarked in Figures 7 and 8 by an arrow. Five of these CTB-LI cells were

NADPH-diaphorase positive (large arrows, Figs. 18 and 20), and onewas only strongly CTB-LI (small arrow, Figs. 19 and 20). Scale bars 525 µm in Figures 17 and 19.


Perhaps the efferent projections have high concentrationsof sialidase stable monosialosyl gangliosides that bindCTB (van Heyningen, 1974) on their arborizations in theretina, or perhaps the arborizations of the contralateralcells are very extensive in the retina. This specific uptakemechanism may explain the increased sensitivity andlarge numbers of contralateral cells we describe in compari-son with some previous studies.

Additionally, we found a much greater rostral-caudalextension of labeled cells within the midbrain (about 4mm), in comparison with previous studies, in which nuclearyellow was used as a neuronal tracer (Weiler, 1985;Schutte and Weiler, 1988). In these previous studiesemploying nuclear yellow, the efferent cell bodies were alllocalized in the caudal mesencephalon, basal to the nucleiisthmi, between the LL and the LoC, with a rostral-caudalextent of only 600–800 µm. In our experiments using CTBas a tracer, labeled cells were also found more caudally inthe Ras and more rostrally around the SN.

It is unlikely that these additional new localizations ofefferent somata we describe in our study are due totranssynaptic labeling, because CTB has been reportednot to produce any transsynaptic labeling even at longsurvival times (Mikkelsen, 1992). More importantly, in aprevious study examining the retinal projections intoturtle brain (although these authors used the same CTBprotocol we used in the present study), there was noevidence of transsynaptic transport of CTB (Reiner et al.,1996). In both the present study and the previous study ofthe central retinal projections, the increased sensitivity ofthe CTB technique provided greater anatomical detail andmore extensive anatomical localizations.

NADPH-diaphorase-positive efferentsin turtle: Possible function of NO

Despite the anatomical evidence for efferent projectionsto the retina, our knowledge of their function in the turtleis poor. Electrophysiological investigations of the efferentsystem in turtle retina (Cervetto et al., 1976; Marchiafava,1976) show that electrical stimulation of the optic nerveproduces action potentials and late synaptic potentials inganglion cells, as well as large excitatory postsynapticpotentials in amacrine cells. However, it is not clearwhether these responses reflect the orthodromic activationof the efferent projections and/or the antidromic activationof the ganglion cells themselves. Weiler (1985) reportsenkephalin immunoreactivity in about a third of thenuclear yellow-labeled isthmic efferent neurons from themesencephalon in the turtle, and Schutte and Weiler(1988) find one serotonin-containing efferent neuron origi-nating from the same brain region. Furthermore, four toten aspartate-containing efferent nerve fibers (Yaqub andEldred, 1991) and seven to ten NADPH-diaphorase/NOS-positive efferents (Blute et al., 1997) are described in theturtle retina; in both cases the origin of these fibers in thebrain was not studied.

In our present study, we examined the role of NOS inthese efferents by double-label studies to look for NADPH-diaphorase enzyme activity in the CTB-LI cells we foundin the midbrain. Our colocalization study showed that 30%of the labeled efferent cells had this enzyme activity. Thesedouble-labeled cells were localized in the isthmic cell fieldnear the locus coeruleus and nucleus reticularis isthmi.Although in many cases NADPH-diaphorase activity hasoften been equated with the presence of neuronal NOS,

there is substantial evidence that not all NADPH-diaphorase activity is attributable to NOS (Belai et al.,1992; Matsumoto et al., 1993; Tracey et al., 1993; Murphy,1994). However, in the present case, Bruning et al. (1994)investigated the distribution pattern of NOS in these sameregions of the turtle brain and found that such cells haveexcellent colocalization of both NADPH-diaphorase histo-chemistry and NOS immunohistochemistry. Furthermore,the putative efferents with NOS-like immunoreactivity inthe turtle retina also show colocalization of NADPH-diaphorase enzyme activity (Blute et al., 1997). It istherefore likely that in addition to the peptidergic, aspar-tate-containing, and serotonergic efferent projections tothe retina in turtle mentioned above, an additional effer-ent system exists, one that uses NO as a transmitter ormodulator. Because of the number of CTB-containingefferent somata in comparison with the relatively smallnumber of efferent cells containing identified neuroactivesubstances, it is likely that other neurotransmitters orneuromodulators may also play a role in the efferent cells.

The fact that the NADPH-diaphorase/NOS-LI efferentfibers cover large areas of the retina suggests that theymay play a relatively diffuse modulatory role (Blute et al.,1997). It is possible that the NO released from theseefferent terminals may be responsible for the previouslyreported effects of the orthodromic activation of theseefferents (Cervetto et al., 1976; Marchiafava, 1976). How-ever, it is also possible that some of these stimulationeffects may be due to NO released from the anterodromicactivation of ganglion cells, because many ganglion cells inturtle retina also contain NOS (Blute et al., 1997). Ineither case, perhaps the NO activates soluble guanylylcyclase in ganglion cells and modulates the conductivity ofthe cGMP-gated channels that have been demonstrated inganglion cells (Ahmad et al., 1994). It is also possible thatthe NO released may modulate gap junctions betweenretinal amacrine cells (Mills and Massey, 1995).

Comparison with the avian isthmo-opticefferent projections

In the well-developed isthmo-optic efferent projection ofbirds, two types of efferent cells can be distinguished. Thefirst are the cells of the isthmo-optic nucleus or ION cells(8,000–10,000 cells in ground-feeding birds; Cowan, 1970)that terminate in special synaptic endings on amacrinecells and displaced ganglion cells in the retina (Maturanaand Frenk, 1965). The second are the ectopic efferent cellsomata, which are far fewer in number than the ION cellsand are scattered ventrally to the ION (Hayes and Web-ster, 1981). The axons of the ectopic cells give widespreadarborizations in the inner plexiform layer (Fritzsch et al.,1990). In the turtle, the aspartate-containing efferentfibers described by Yaqub and Eldred (1991) and theNOS-containing efferent fibers (Blute et al., 1997) arborizeextensively through many millimeters of retina and aremore similar to the widespread retinal arborizations of theectopic cells of birds. Also, the morphology of the ectopiccell somata in owls (Weidner et al., 1987) is similar to theCTB-LI cells in turtles (Figs. 2–6). In both owls andturtles, the cells are multipolar or fusiform in shape, witha soma size of 15–20 µm in diameter. In comparison, theION cells are smaller and have more rounded somata.Thus, the size and shapes of the ectopic cells, the smalleroverall numbers of cells in comparison with ION cells, andtheir arborization patterns in the retina suggest that the


cells in the turtle more closely resemble the ectopic cells inthe bird.

A comparison of previous neurochemical studies of theavian and turtle midbrain can be used to examine thehomologies in the efferent projections of these two species.In both birds and turtles, the Ras is shown to haveserotonin-containing cells (avian: Dube and Parent, 1981;turtle: Ueda et al., 1983) but not NADPH-diaphorase-positive cells (Bruning, 1993; Bruning et al., 1994). It ispossible that the CTB-LI cells found in the Ras in ourstudy may use serotonin as a transmitter and may repre-sent the serotonergic efferent system described by Schutteand Weiler (1988).

The distribution of choline acetyltransferase (ChAT) inthe avian and turtle isthmus is also informative. Theprecise localization of ChAT-immunoreactive neurons inand around the ION of birds is still unresolved. Someauthors report ChAT-LI cells in the ION and in numerousstrongly labeled cells surrounding the ION, with the lattercells potentially representing the ectopic isthmo-optic neu-rons (Bagnoli et al., 1992). By contrast, other authors findonly extremely lightly labeled cells in the ION, as well assome cells with ChAT-LI surrounding the ION (Medinaand Reiner, 1994). Additionally, a distinct group of largecholinergic neurons is present in the pedunculopontinetegmental nucleus (PPN) in birds, and ChAT-immunoreac-tive cells are also present in the laterodorsal tegmentalnucleus, partially intermingled with the noradrenergicneurons of the LoC (Medina and Reiner, 1994).

Comparable cholinergic cell groups have also been de-scribed in the lateral isthmic tegmentum in turtles. There,the ChAT-immunoreactive cells are partly intermixed withthe catecholaminergic cells of the LoC (Powers and Reiner,1993) and are located at the same place as the NADPH-positive efferent cells we found in this study. This findingsuggests that some of the efferent cells, regardless ofwhether they are NADPH-diaphorase positive or not,might use acetylcholine as a transmitter in turtle. Atriple-label study of CTB-LI efferent cells, ChAT-immuno-histochemistry, and NADPH-diaphorase histochemistrycould test this hypothesis. A similar study would beinteresting in birds, in which the ectopic cells are goodcandidates for the cholinergic- (Bagnoli et al., 1992) andNADPH-diaphorase-positive cells (Bruning, 1993) scat-tered around the ION; direct proof is lacking.

In the isthmus of birds, there are distinct cholinergiccells in the PPN (Powers and Reiner, 1993), distinctnoradrenergic cells in the LoC (Reiner et al., 1994), anddistinct NADPH-diaphorase-positive cells in the ION (Mor-gan et al., 1994). These same cell groups are present inturtles and can be recognized histochemically, but theirborders are more indistinct and overlap. This findingsuggests that the turtle ‘‘isthmo-optic nucleus’’ is in theright general place but that its cytoarchitecture is notnearly as well defined as that of the bird ION. The locationof the efferent cell bodies in the turtle brain providesevidence that both birds and reptiles had a commonancestor with a primitive efferent system. This efferentsystem has remained relatively primitive in turtles,whereas it has become more highly developed in birds. Theefferent cells in this isthmic region in turtles could corre-spond to the well-developed ION in birds, or to the ectopiccells in birds, or to both. The isthmic region and the retinalefferent system in crocodiles (Ferguson et al., 1978) isapparently better developed than that of turtles, but not as

well organized as that of birds. This indicates a gradualincrease in the level of organization as one moves fromturtles to crocodiles to birds.

Morgan et al. (1994) suggested that the efferent projec-tion to the retina in birds uses NO as a messenger ortransmitter, in addition to a more conventional but as yetunidentified transmitter, possibly acetylcholine. Our re-sults suggest that in addition to serotonin (Schutte andWeiler, 1988), enkephalin (Weiler, 1985), aspartate (Yaquband Eldred, 1991), or possibly acetylcholine, 30% of theefferent cells may use NO as a modulator or transmitter inturtle. Future studies will be required to clarify thedifferent transmitters and their functions in the efferentsystems in turtles and birds. The precise localization of theNADPH-diaphorase-containing efferent cell bodies in thebrain in this study will provide the foundation for futurestudies examining the effects of the selective activation ofthese efferents on the retina.

ACKNOWLEDGMENTS

We thank Felicitas B. Eldred for her excellent technicalassistance, Dr. Danru Zhang for her help with the initialretrograde labeling studies, and particularly Dr. AntonReiner for his invaluable discussions on the comparativeneurochemistry of the avian and turtle brains and hiscritical reading of the manuscript. W.D.E. was the recipi-ent of a grant from NEI.

LITERATURE CITED

Ahmad, I., T. Leinders-Zufall, J.D. Kocsis, G.M. Shepherd, F. Zufall, andC.J. Barnstable (1994) Retinal ganglion cells express a cGMP-gatedcation conductance activatable by nitric oxide donors. Neuron 12:155–165.

Bagnoli, P., G. Fontanesi, R. Alesci, and J.T. Erichsen (1992) Distribution ofneuropeptide Y, substance P, and choline acetyltransferase in thedeveloping visual system of the pigeon and effects of unilateral retinaremoval. J. Comp. Neurol. 318:392–414.

Belai, A., H.H.H.W. Schmidt, C.H.V. Hoyle, C.J.S. Hassall, M.J. Saffrey, J.Moss, U. Forstermann, F. Murad, and G. Burnstock (1992) Colocaliza-tion of nitric oxide synthase and NADPH-diaphorase in the myentericplexus of the rat gut. Neurosci. Lett. 143:60–64.

Blute, T.A., B. Mayer, and W.D. Eldred (1997) Immunocytochemical andhistochemical localization of nitric oxide synthase in the turtle retina.Vis. Neurosci. 14:717–729.

Bruning, G. (1993) Localization of NADPH-diaphorase in the brain of thechicken. J. Comp. Neurol. 334:192–208.

Bruning, G., S. Wiese, and B. Mayer (1994) Nitric oxide synthase in thebrain of the turtle Pseudemys scripta elegans. J. Comp. Neurol.348:183–206.

Cervetto, L., P.L. Marchiafava, and E. Pasino (1976) Influence of retinalfibres on responsiveness of ganglion cells to light. Nature 260:56–57.

Cowan, W.M. (1970) Centrifugal fibers to the avian retina. Br. Med. Bull.26:112–118.

Dube, L., and A. Parent (1981) The monoamine-containing neurons inavian brain: I. A study of the brain stem of the chicken (Gallusdomesticus) by means of fluorescence and acetylcholinesterase histo-chemistry. J. Comp. Neurol. 196:695–708.

Ferguson, J.L., P.J. Mulvanny, and S.E. Brauth (1978) Distribution ofneurons projecting to the retina of Caiman crocodilus. Brain Behav.Evol. 15:294–306.

Fritzsch, B., M.-D. Crapon de Caprona, and P.G.H. Clarke (1990) Develop-ment of two morphological types of retinopetal fibers in chick embryos,as shown by the diffusion along axons of a carbocyanine dye in the fixedretina. J. Comp. Neurol. 300:405–421.

Hayes, B.P., and K.E. Webster (1981) Neurons situated outside theisthmo-optic nucleus and projecting to the eye in adult birds. Neurosci.Lett. 26:107–112.

Marchiafava, P.L. (1976) Centrifugal actions on amacrine and ganglioncells in the retina of the turtle. J. Physiol. 255:137–155.


Matsumoto, T., M. Nakane, J.S. Pollock, J.E. Kuk, and U. Forstermann(1993) A correlation between soluble brain nitric oxide synthase andNADPH-diaphorase activity is only seen after exposure of the tissue tofixative. Neurosci. Lett. 155:61–64.

Maturana, H.R., and S. Frenk (1965) Synaptic connections of the centrifu-gal fibers of the pigeon retina. Science 150:359–362.

Medina, L., and A. Reiner (1994) Distribution of choline acetyltransferaseimmunoreactivity in the pigeon brain. J. Comp. Neurol. 342:497–537.

Mikkelsen, J.D. (1992) Visualization of efferent retinal projections byimmunohistochemical identification of cholera toxin subunit B. BrainRes. Bull. 28:619–623.

Miles, F.A. (1972) Centrifugal control of the avian retina, III: Effects ofelectrical stimulation of the isthmo-optic tract on the receptive-fieldproperties of retinal ganglion cells. Brain Res. 48:115–129.

Mills, S.L., and S.C. Massey (1995) Differential properties of two gapjunctional pathways made by AII amacrine cells. Nature 377:734–737.

Morgan, I.G., P. Miethke, and Z.K. Li (1994) Is nitric oxide a transmitter ofthe centrifugal projection to the avian retina? Neurosci. Lett. 168:5–7.

Murphy, S. (1994) Histochemical localization of nitric oxide synthase in theCNS. Reply. Trends Neurosci. 17:106.

Powers, A.S., and A. Reiner (1993) The distribution of cholinergic neuronsin the central nervous system of turtles. Brain Behav. Evol. 41:326–345.

Reiner, A., E.J. Karle, K.D. Anderson, and L. Medina (1994) Catecholamin-ergic perikarya and fibers in the avian nervous system. In W.J.A.J.Sweets, and A. Reiner (eds): Phylogeny and Development of Catechol-amine Systems in the CNS of Vertebrates. Cambridge: CambridgeUniversity Press, pp. 135–181.

Reiner, A., D.R. Zhang, and W.D. Eldred (1996) Use of the sensitiveanterograde tracer cholera toxin fragment B reveals new details of thecentral retinal projections in turtles. Brain Behav. Evol. 48:307–337.

Reperant, J., D. Miceli, N.P. Vesselkin, and S. Molotchnikoff (1989) Thecentrifugal visual system of vertebrates: A century-old search reviewed.Int. Rev. Cytol. 118:115–172.

Schnyder, H., and H. Kunzle (1983) The retinopetal system in the turtlePseudemys scripta elegans. Cell Tissue Res. 234:219–224.

Schutte, M., and R. Weiler (1988) Mesencephalic innervation of the turtleretina by a single serotonin-containing neuron. Neurosci. Lett. 91:289–294.

Tracey, W.R., M. Nakane, J.S. Pollock, and U. Forstermann (1993) Nitricoxide synthases in neuronal cells, macrophages and endothelium areNADPH diaphorases, but represent only a fraction of total cellularNADPH diaphorase activity. Biochem. Biophys. Res. Commun. 195:1035–1040.

Uchiyama, H. (1989) Centrifugal pathways to the retina: Influence of theoptic tectum. Vis. Neurosci. 3:183–206.

Ueda, S., Y. Takeuchi, and Y. Sano (1983) Immunohistochemical demonstra-tion of serotonin neurons in the central nervous system of the turtle(Clemmys japonica). Anat. Embryol. 168:1–19.

van Heyningen, W.E. (1974) Gangliosides as membrane receptors fortetanus toxin, cholera toxin and serotonin. Nature 249:415–417.

Weidner, C., J. Reperant, A.-M. Desroches, D. Miceli, and N.P. Vesselkin(1987) Nuclear origin of the centrifugal pathway in birds of prey. BrainRes. 436:153–160.

Weiler, R. (1985) Mesencephalic pathway to the retina exhibits enkephalin-like immunoreactivity. Neurosci. Lett. 55:11–16.

Yaqub, A., and W.D. Eldred (1991) Localization of aspartate-like immunore-activity in the retina of the turtle (Pseudemys scripta). J. Comp. Neurol.312:584–598.


localization of the origin of retinal efferents in the turtle brain and the involvement of nitric...

Documents