Stimulation With N-Methyl-D-Aspartateor Kainic Acid Increases Cyclic

Guanosine Monophosphate-LikeImmunoreactivity in Turtle Retina:

Involvement of Nitric Oxide Synthase

TODD A. BLUTE, JENNIFER DE GRENIER, AND WILLIAM D. ELDRED*Department of Biology, Boston University, Boston, Massachusetts 02215

ABSTRACTIn brain and retina, stimulation with excitatory amino acids (EAA) can generate nitric

oxide (NO) and increase levels of cyclic guanosine monophosphate (cGMP). Because nitricoxide synthase (NOS) has been found in retinas of all species examined to date, an NOsignal-transduction pathway is likely to be present in all retinas. We tested the hypothesisthat stimulation of ionotropic glutamate receptors in turtle retina would result in increasesin cGMP through an NOS/NO/cGMP pathway. Following in vitro incubations of turtle eyecups with the glutamate receptor agonists, N-methyl-D-aspartate (NMDA) or kainic acid(KA), we quantified the increases in cGMP-like immunoreactivity (cGMP-LI) by using enzyme-linked immunosorbant assay (ELISA) and localized the increased cGMP-LI by using anantibody against cGMP. Stimulation with NMDA or KA increased cGMP-LI in bipolar andamacrine cells as well as in some somata in the ganglion cell layer. Either KA or NMDAproduced statistically significant increases in total retinal cGMP-LI by ELISA. To test theinvolvement of NO, we used the NOS inhibitors 7-nitroindazole and L-nitroarginine. Bothinhibitors blocked virtually all of the KA- or NMDA-stimulated increases in cGMP-LI. Theseresults indicate that activation of ionotropic glutamate receptors can increase cGMP in selectretinal neurons. Differences between the agonist-evoked increases of retinal cGMP-LI suggestthat there can be specificity in the activation of the NOS/NO/cGMP signal-transductionpathway by glutamate. This suggests that, in addition to short-term electrical changes,activation of ionotropic glutamate receptors also may produce longer term modulatory ormetabolic effects involving NO/cGMP. J. Comp. Neurol. 404:75–85, 1999.r 1999 Wiley-Liss, Inc.

Indexing terms: amacrine cells; inner plexiform layer; soluble guanylate cyclase;

immunocytochemistry; enzyme-linked immunosorbant assay

The second messenger, cyclic guanosine monophosphate(cGMP), is involved in many aspects of retinal processing,beginning with gating channels in photoreceptor transduc-tion (Yau and Baylor, 1989). Similar cGMP-gated channelshave been found in ganglion cells (Ahmad et al., 1994) andbipolar cells (Nawy and Jahr, 1991; Shiells and Falk, 1992)as well as in Muller cells (Kusaka et al., 1996). cGMP alsocan modulate gap junctions between horizontal cells(DeVries and Schwartz, 1989; Miyachi and Murakami,1991) and between amacrine cells and bipolar cells (Millsand Massey, 1995). Because cGMP is involved in manydiverse roles in retinal function, it will be important todetermine the synaptic mechanisms that can modulate thelevels of cGMP in specific retinal neurons.

In the retina, cGMP can be synthesized by the followingthree classes of guanylate cyclase (GC): retina-specific,membrane-bound guanylate cyclases involved in photo-transduction (Ret GC: Shyjan et al., 1992; Ret GC 2: Loweet al., 1995); at least five different ligand-activated, particu-late GCs (pGCs; Chrisman et al., 1992; Dizhoor et al.,1994; Duda et al., 1993; Kutty et al., 1992; Lowe et al.,1995); and soluble guanylate cyclase (sGC) activated by

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

Boston University, 5 Cummington Street, Boston, MA 02215.Received 20 March 1998; Revised 6 August 1998; Accepted 1 October

1998

THE JOURNAL OF COMPARATIVE NEUROLOGY 404:75–85 (1999)

r 1999 WILEY-LISS, INC.

nitric oxide (NO; Koch et al., 1994). sGC and pGC havebeen localized by using in situ hybridization in all regionsof the retina; however, sGC was concentrated in the innerretina, whereas pGC was found at higher levels in theouter retina (Ahmad and Barnstable, 1993). In the turtleretina, sGC has been localized functionally in numerousamacrine or bipolar cell somata in the inner nuclear layer(INL) and the outer nuclear layer (ONL), in many somatain the ganglion cell layer (GCL), and in many processes inthe inner plexiform layer (IPL; Blute et al., 1998). Nitricoxide synthase (NOS) and its related nicotinamide ad-enine dinucleotide phosphate (NADPH)-diaphorase activ-ity have been localized in three amacrine cell types, insomata in the INL and GCL, and in the ellipsoids of innersegments of photoreceptors in the turtle retina (Blute etal., 1997).

In the brain, stimulation with kainic acid (KA) orN-methyl-D-aspartate (NMDA) increases intracellular lev-els of calcium (Garthwaite, 1991); and these increases canactivate NOS (Okada, 1992). Excitotoxic studies in therabbit retina indicate that NADPH-positive amacrine cellshave NMDA and KA receptors (Sagar, 1990). In the chickretina, glutamate agonists activate a calcium/calmodulin-dependent NOS to produce NO through an increase inintracellular calcium (Zeevalk and Nicklas, 1994). Thespecific retinal cell types that have increased levels ofcGMP in response to this excitatory amino acid (EAA)-stimulated production of NO have not been determined.

This study examines the effects of the glutamate recep-tor agonists, NMDA and KA, on cGMP levels in the turtleretina (Pseudemys scripta elegans). The involvement of theNOS/NO/cGMP pathway was tested by using two differentNOS inhibitors. Our results indicate that stimulation withNMDA and KA increased cGMP-like immunoreactivity(cGMP-LI) in several sGC-containing cell types and thatthere were significant regional differences. The differentresponses to these agonists suggest that there are regionaldifferences in the NO/cGMP signal-transduction pathwayin the turtle retina. The results seen with NOS inhibitorsindicate that the increased cGMP-LI in response to KA orNMDA was mediated by NO and that activation of iono-tropic glutamate receptors can lead to the activation ofNOS and subsequent increases in cGMP in select retinalneurons. Therefore, activation of ionotropic glutamatereceptors also may produce modulatory or long-term bio-chemical changes based on NO/cGMP in addition to theknown electrophysiological effects. By working throughNO, these biochemical or modulatory changes may occurin cells that may not even have the activated glutamatereceptors.

MATERIALS AND METHODS

Reagents

All reagents were from Sigma Chemical Company (St.Louis, MO) with the following exceptions: The Vector Eliteavidin biotin complex (ABC) reagents were from VectorLaboratories, Inc. (Burlingame, CA), and the Immunopureymetal-enhanced diaminobenzidine (DAB) substrate wasobtained from Pierce Chemical Company (Rockford, IL).

In vitro incubations

Adult turtles (Pseudemys scripta elegans) were main-tained on a 12-hour light/12-hour dark cycle. Near themiddle of the light cycle, they were decapitated by using a

method approved by the Boston University Charles RiverCampus Institutional Animal Care and Use Committee.The enucleated eyes were hemisected along the verticalmidline perpendicular to the visual streak. The isolatedturtle eyecups were preincubated for 15 minutes in anaerated, normal Ringer’s solution (110 mM NaCl, 2.5 mMKCl, 3 mM CaCl2, 2 mM MgCl2, 10 mM glucose, and 5 mMHEPES), pH 7.4, at room temperature with 1 mM 3-isobu-tyl-1-methylxanthine (IBMX; a nonspecific phosphodiester-ase inhibitor). Following the preincubation in IBMX, theeye cups were transferred to fresh IBMX-Ringer’s for 30minutes with or without the following drugs: KA (10 µM),NMDA (100 µM), or one of the nNOS inhibitors, 7-nitroin-dazole (7-NI; 10–100 µM) or N-nitro-L-arginine (L-NNA;1 mM). NOS inhibitors were included at the same concen-tration in the preincubation solutions, as appropriate. Wealso tested the following control treatments: no agonist(only IBMX-Ringer’s), and control Ringer’s (no agonist orIBMX). All incubations were done in normal room light.

Tissue preparation

Following the pharmacological incubations, the eye cupswere either fixed in 4% paraformaldehyde with 5% sucrosein 0.1 M phosphate buffer, pH 7.4 (PB), for 90–180minutes, or the retinas were isolated for enzyme-linkedimmunosorbant assay (ELISA). Following fixation, the eyecups were infiltrated gradually with increasing sucroseconcentrations and ultimately embedded in two parts 20%sucrose to one part optimum cutting temperature embed-ding compound (Barthel and Raymond, 1990). The frozenblocks were cut by using a cryostat into 14-µm-thick crosssections that were thaw mounted onto slides coated withchrome gelatin.

ELISA assays

Following stimulation, isolated retinas were homog-enized in cold 50 mM PB, pH 6.5, with 10% trichloroaceticacid (TCA). Homogenates were centrifuged at 15,000 RPMfor 10 minutes, and the supernatants were removed. TheTCA was extracted from the supernatant with 5 3 5volumes of water saturated ether. The supernatant sampleswere finally heated to 70°C to evolve any remaining ether.Samples were then frozen on dry ice and stored at 280°Cuntil assayed. The purified supernatants (diluted 1:50 or1:100) were analyzed in duplicate by using a commercialcGMP ELISA assay (Cayman Chemical Company, AnnArbor, MI). The protein concentrations were determinedby using a 96-well Bradford microplate assay (Redinbaughand Campbell, 1985).

Immunocytochemical staining

All antisera solutions were diluted in 0.3% Triton X-100in 0.1 M PB, pH 7.4 (PBtx). To control for nonspecificlabeling, sections were pretreated with 5% normal goatserum for 60 minutes at 22°C in a moist, humid chamber.The sections were then incubated overnight in a 1:2,500dilution of rabbit antiserum raised against cGMP. Theproduction, characterization, and specificity of this anti-body have been described previously (de Vente and Stein-busch, 1992). The sections were subsequently incubated byusing standard immunocytochemical techniques employ-ing Vector Elite ABC reagents diluted 1:1,000 in PBtx(Ellis and Halliday, 1992), and the labeled cells werevisualized by using Immunopure metal-enhanced DABsubstrate. The enzyme reaction was monitored visually

76 T.A. BLUTE ET AL.

and was terminated with several washes in PB; then, thesections were mounted by using Crystal Mount (BiømedaCorp., Foster City, CA). For a control for nonspecificcGMP-LI, the cGMP primary antibody was omitted, or thestaining was done by using cGMP primary antibody thatwas blocked previously with 1025 M cGMP.

Quantative image analysisof immunoreactivity

Images of retinal sections were analyzed using a CCDvideo camera and Image-Pro Plus image-analysis software(Media Cybernetics, Silver Spring, MD) to quantify therelative location of the immunocytochemical labeling inthe IPL. The vertical averaging function was used toaverage levels of DAB reaction product from adjacentregions of the same strata of the retina. This had the effectof increasing signal-to-noise ratios and eliminating anymeasurement errors due to small regional differences.Relative locations within the IPL are described such thatthe border between the INL and the IPL is S0, the middleof the IPL is S50, and the border between the IPL and theGCL is S100. The INL has been divided into three tiers ofcell bodies so that tier 1 is adjacent to the INL/IPL border,tier 2 is the central INL, and tier 3 is adjacent to the outerplexiform layer OPL/INL border.

RESULTS

Activation of glutamate receptors with either of theagonists, KA or NMDA, significantly increased cGMP-LIin the turtle retina. There were clear regional differencesbetween the agonists in terms of the labeled cell types, thelabeling intensity of the cells, and the densities of labeledcells. The primary regional distinctions were seen betweenthe visual streak, the peripheral retina above and belowthe streak, and near the ora serrata. Specific bipolar andamacrine cell types were seen in retinas stimulated withNMDA or KA. In addition, with either KA or NMDA, therewere large somata in the GCL with faint cGMP-LI.

All cells with elevated cGMP-LI in response to KA orNMDA resembled previously described sGC-containingcells and will be referred to by using the previouslypublished nomenclature for these cells (Blute et al., 1998).The classification of the sGC-containing amacrine andbipolar cell types is based on the relative locations of theircell somata within the INL or ONL, their soma shape anddendritic branching pattern, and the stratification of theirdendrites in the IPL or OPL. Figure 1A is a schematicrepresentation of all of the sGC-containing neurons in theinner retina of the turtle (Blute et al., 1998). Following thisnomenclature, all of the cells are referred to by thedesignation sGC for soluble guanylate cyclase, followed bya letter for amacrine cell (A) or bipolar cell (B) and anumber. If there is no A or B, then the soma of the cell islocated in the GCL. The number indicates the relativesensitivity of each cell type to an NO donor, as establishedby using dose-response studies (Blute et al., 1998), withtype 1 as the most sensitive and 14 as the least sensitive.This same number also reflects the relative frequency ofeach cell type. For example, numerous sGC A1 cells areseen at all concentrations of NO donor, whereas only smallnumbers of sGC 14 cells are seen when retinas arestimulated with the highest concentrations of NO donor.With the exception of extreme peripheral regions near the

ora serrata, the characteristic appearance and the den-dritic stratifications of these neurons were maintaineduniformly throughout the retina. Figure 1B,C schemati-cally summarizes the pattern of cGMP-LI observed inresponse to NMDA and KA stimulation, respectively. Itwas shown previously (Blute et al., 1998) in control retinasthat, in the absence of stimulation or when primaryantibody blocked with cGMP was used, no appreciablecGMP-LI was visible anywhere in the retina. The undetect-able basal levels of cGMP-LI in unstimulated controlretinas were not significantly increased by light or darkadaptation.

Amacrine cell types

NMDA stimulation. All amacrine cell types with in-creased cGMP-LI in response to stimulation with 100 µMNMDA had monostratified dendritic arborizations in theIPL. sGC A1 amacrine cells had pyriform-shaped somatalocated in the first tier of the INL that gave rise to thin,primary dendritic processes that arborized in S15–S35 ofthe IPL (Fig. 2A). The second identifiable amacrine celltype, sGC A2, had a large round soma that was found inthe first tier of the INL and a single, thick (2–3 µm indiameter) primary process that arborized at approxi-mately S20 of the IPL (Fig. 2B). sGC A3 amacrine cells hadmore rounded, pyriform-shaped somata located in the firsttier of the INL. The somata of the sGC A3 cells tapered toform a thick primary dendrite that branched in S50 beforearborizing in S60 of the IPL (Fig. 2C). In the GCL, therewere small, pyriform somata (Fig. 2D) that resembled sGC4 displaced amacrine cells. The exact stratification of thesedisplaced cells could not always be resolved; however, thesomata of these cells tapered into primary processes thatprojected toward S60 of the IPL. Finally, there were smallnumbers of sGC A11 amacrine cells with flattened somatalocated in the first tier of the INL that gave rise to twoprimary processes that arborized in S20/S30 of the IPL(Fig. 2A).

KA stimulation. Activation of glutamate receptors by10 µM KA increased cGMP-LI in four amacrine cell typesthroughout the retina. Three of these four amacrine celltypes were the same sGC A1, sGC A2 (Fig. 3A), and sGC A3(Fig. 3B) amacrine cell types described above. There werealso occasional somata resembling sGC 4 (Fig. 3C) ama-crine cells. KA stimulation never increased cGMP-LI insGC A11 amacrine cells.

Bipolar cell types

Stimulation with either NMDA (Fig. 2A–C) or KA (Fig.3A–C) increased cGMP-LI both in normally placed bipolarcells and in bipolar cells with somata displaced to theONL. Although the somata of these bipolar cells were allwell labeled, their processes were weakly immunoreactive,and the arborizations of these cells usually could not betraced in the OPL or the IPL. The location and shape of thebipolar somata were consistent with the three sGC-containing bipolar cell types in the turtle retina (Blute etal., 1998). The lack of arborizations made classification ofmost of these bipolar cells impossible. However, there weresome displaced bipolar cells in KA-stimulated retinas thatclearly resembled sGC B2 bipolar cells (Fig. 3D).

Regional variations

Regional variations were found to be most significantwithin, above, and below the visual streak. The following

NMDA- AND KA-STIMULATED cGMP IN RETINA 77

descriptions will refer to regions closer to the ora serrata ofthe retina as peripheral. The regional differences betweenNMDA- and KA-stimulated cGMP-LI are summarized inTable 1 to allow for direct comparison.

NMDA stimulation. With NMDA stimulation, therewas increased cGMP-LI in both normally located anddisplaced bipolar cell somata. In the visual streak, the cellbodies of these bipolar cells showed intense cGMP-LI, andthe proximal processes were distinguishable (Fig. 4A).

Below the visual streak, the overall intensity of immunore-activity in the IPL increased toward the inferior oraserrata. In the inferior retina and near the inferior oraserrata, there were robustly labeled, displaced bipolarcells, most likely sGC B2 bipolar cells. However, the regionabove the visual streak in the NMDA-stimulated retinasconsistently had fewer labeled bipolar cell somata, and thecGMP-LI in these bipolar cells was less intense than inother retinal regions.

Fig. 1. Schematic summary of cells with increased cyclic guanosinemonophosphate-like immunoreactivity (cGMP-LI) in response to anitric oxide (NO) donor, 400 µM S-nitroso-N-acetylpenicillamine(SNAP; Blute et al., 1998; A). The numbers correlate with the relativesensitivity of the cells to NO donor, from 1 (the most sensitive) through14 (the least sensitive). The numbers also represent the relativefrequency of that cell type. All of these cells are referred to by the

designation for soluble guanylate cyclase (sGC) followed by a letter foramacrine cells (A) or bipolar cells (B) and a number. If there is noletter, then the soma of the cell is located in the ganglion cell layer(GCL). Schematic summaries of the sGC cell types with increasedcGMP-LI in response to 100 µM N-methyl-D-aspartate (NMDA; B) or10 µM kainic acid (KA; C). ONL, outer nuclear layer; OPL, outerplexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer.

78 T.A. BLUTE ET AL.

In the IPL, cGMP-LI was present in two or three specificbands, depending on the region of the retina. In the visualstreak, there was one distinct band at S15–S20 and asecond band at S30–S35 (Fig. 4A). In regions outside of thevisual streak, these two bands merged to form one diffuseband that extended from S15 to S35 (Fig. 2A,C,D). Theintensity of cGMP-LI in S15–S35 of the IPL was reducedabove the visual streak. A second discrete band of cGMP-LIwas located near S60 of the IPL (Fig. 2B,C). In the visualstreak, the intensity of cGMP-LI in S15–S35 was greaterthan in S60 (Fig. 4A). Above the streak, the S60 band wasfaint and sparse (Fig. 2D) due to the smaller numbers oflabeled sGC A3 amacrine cells, putative sGC 4 amacrine

cells, and sGC B2 bipolar cells that form this band. At theextreme peripheral margins of the retina approaching theora serrata, cGMP-LI was seen as two bands (S15–35 andS60) in the inferior retina and as only one band (centeredon S20) in the superior retina.

There were clear regional differences in the cells withNMDA-stimulated increases in cGMP-LI in terms of thelabeled cell types and their intensity of cGMP-LI. Belowthe visual streak, the processes of amacrine cells weredefined clearly and had intense cGMP-LI (Fig. 2B). sGC A1and sGC A2 amacrine cell types were found throughout theretina, except in the visual streak, where only sGC A1 cellshad cGMP-LI (Fig. 4A). Amacrine cell type sGC A3 was

Fig. 2. Cross sections of retina stimulated with 100 µM NMDA andstained for cGMP-LI. In this figure, horizontal arrowheads on the leftin A–D demarcate the border of the IPL. A: sGC A1 (horizontal arrow)and sGC A11 (vertical arrow) amacrine cells with cGMP-LI in theinferior retina. sGC A11 cells were found only in the inferior retina andwere somewhat rare. The sGC A11 cells gave rise to several stoutprimary processes that left the somata laterally and then arborized inthe S15–S35 region of the IPL. Several somata with varying levels ofcGMP-LI were visible in the GCL (asterisk). Displaced bipolar cellsomata (vertical arrowheads) exhibited increased cGMP-LI. B: In thiscross section of inferior retina, sGC A2 (horizontal arrow) and sGC A3(vertical arrow) amacrine cells with elevated cGMP-LI were seen toarborize in S15–S35 (asterisk) and S60 (asterisks), respectively. Inthis section, the subdivisions in the S15–S35 band of immunoreactiv-ity were visible immediately below the sGC A2 soma. Displaced bipolarcell somata (horizontal arrowheads) exhibited increased cGMP-LI,

and labeled lateral processes of displaced bipolar cells were observedin the OPL (vertical arrowhead). C: An sGC A3 amacrine cell (horizon-tal arrow) with cGMP-LI located in the inferior peripheral retina.These monostratified cells had a single primary process that descended toS50 in the IPL and then branched before they ramified in S60 (asterisks).Several displaced bipolar cells (vertical arrowheads) and a normallyplaced bipolar cell (horizontal arrowhead) had labeled somata. D: Sectionof superior retina. In the GCL, a strongly immunoreactive soma (horizon-tal arrow) extended an apical process into the IPL approaching S60.Based on the morphology of the soma and its location in the GCL,combined with the apparent arborization in S60 of the IPL, this cellwas classified as a putative sGC 4. In this section, the S15–S35 stratawas quite diffuse, and the cGMP-LI in S60 was discontinuous andweak except in the vicinity of these putative sGC 4 cells. Also presentin the GCL was a weakly labeled ganglion cell (vertical arrow) with anaxon. Abbreviations as defined in Fig. 1. Scale bars 5 25 µm.

NMDA- AND KA-STIMULATED cGMP IN RETINA 79

observed most frequently in the inferior retina. In thesuperior retina, sGC A3 amacrine cells were scarce, andthe S60 stratum was discontinuous and vanished near thesuperior ora serrata. The rare amacrine cell type sGC A11also was found only below the visual streak, near the opticnerve head (Fig. 2A).

KA stimulation. The relative number of all labeledamacrine cell types was greater below than above thevisual streak. However, KAstimulation increased cGMP-LImore uniformly throughout the retina than NMDA stimu-lation. In the superior part of NMDA-stimulated retinas,the amacrine cells had relatively weak cGMP-LI, whereasthe amacrine cells in KA-stimulated retinas had similarlevels of strong cGMP-LI in both the superior retina andthe inferior retina. Compared with NMDA-stimulatedretinas, the visual streak of KA-stimulated retinas had

TABLE 1. Regional Differences in the Relative Levels of Cyclic GuanosineMonophosphate-Like Immunoreactivity in Specific Retinal Cell Types

in Response to Stimulation With N-Methyl-D-Aspartate or Kainic Acid1

Visualstreak

Inferiorretina

Inferior oraserrata

Superiorretina

Superior oraserrata

NMDAsGC A1 111 111 111 11 11sGC A2 2 111 111 11 11sGC A3 2 111 111 11 11sGC A4 like 2 11 11 1/2 1/2sGC A11 2 11 2 2 2sGC B somata 111 11 11 1/2 1/2

KAsGC A1 1 11 11 11 11sGC A2 2 11 11 11 11sGC A3 2 11 11 11 11sGC A4 like 2 11 11 1/2 1/2sGC B somata 1 11 11 1 1

1NMDA, N-methyl-D-aspartate; KA, kainic acid; sGC, soluble guanylate cyclase. 111,Strong; 11, moderate; 1, weak; 2, none present.

Fig. 3. cGMP-LI in cross sections of 10 µM KA-stimulated retinas.In contrast to NMDA, with which there was more cGMP-LI in theinferior retina, KA stimulated cGMP-LI more uniformly throughoutthe retina. The exception to this uniformity was the common lack ofcGMP-LI in S60 of the IPL. In this figure, horizontal arrowheads atthe left in A–D demarcate the border of the IPL. A: Near the centralretina, monostratified sGC A1 (horizontal arrow) and sGC A2 (verticalarrow) amacrine cells and displaced bipolar cell somata (verticalarrowhead) demonstrated cGMP-LI. B: In response to KA stimulation,cGMP-LI was increased in S60 of the IPL near sGC A3 amacrine cells(vertical arrow). sGC A3 cells had a single primary process thatdescended to S50 before dividing and arborizing in S60 (asterisk).

C: In the GCL, there were putative sGC 4 displaced amacrine cells(horizontal arrow) with an apical process that apparently projected toS60 of the IPL. An sGC A2 amacrine cell (vertical arrow) is stronglylabeled. Displaced (vertical arrowhead) bipolar cell somata exhibitedincreased cGMP-LI. D: In the region below the visual streak approach-ing the ora serrata, strong cGMP-LI was observed in many bipolar cellsomata. Both displaced (vertical arrowheads) and normal (horizontalarrowheads) bipolar cell somata, as well as lateral processes in theOPL (vertical arrows), had cGMP-LI. A putative sGC 4 amacrine cell(horizontal arrow) in the GCL had elevated cGMP-LI. Abbreviationsas defined in Fig. 1. Scale bars 5 25 µm.

80 T.A. BLUTE ET AL.

little cGMP-LI, with only some lightly stained bipolar cellsand an occasional sGC A1 amacrine cell with weaklylabeled primary processes (Fig. 4B). sGC A2 amacrine celltypes were found throughout the retina, except in thevisual streak. sGC A3 amacrine cells were less commonthan the sGC A1 and sGC A2 cell types, but they were seenthroughout the retina outside of the visual streak, includ-ing the midperiphery of the superior retina. Both normaland displaced bipolar cells were labeled throughout theretina, but there were greater numbers and more intenselylabeled bipolar cells in the peripheral margins of theretina.

The visual streak showed only faint staining in S15–S25of the IPL (Fig. 4B). The cGMP-LI in S60 of the IPL wassparse in the streak. Although there was some cGMP-LI inS60 of the IPL in the superior retina, it was faint anddisappeared completely before reaching the ora serrata.However, in the inferior retina, cGMP-LI in S60 of the IPL

increased in intensity toward the ora serrata (Fig. 3D).Overall, the KA-stimulated increases of cGMP-LI in theIPL at the extreme peripheral margins of the retinademonstrated the same pattern as NMDA stimulation(inferior, S15–S35 and S60; superior, S15–S35).

Involvement of NOS

None of the faint cGMP-LI in large somata in the GCLwas blocked with either NOS inhibitor. This cGMP-LIlikely is the result of other cGMP-related signal-transduc-tion pathways that impinge on ganglion cells in responseto glutamate receptor stimulation. The inclusion of 10 µM7-NI did not attenuate any of the agonist-specific increasesin cGMP-LI. In contrast, in the presence of 100 µM 7-NI or1 mM L-NNA, there was no significant NMDA- or KA-stimulated cGMP-LI (Fig. 4C and Fig. 4D, respectively) inthe IPL, and the visual streak had no immunoreactivebipolar or amacrine cells. However, in NMDA-stimulated

Fig. 4. In this figure, horizontal arrowheads at the left in A–Ddemarcate the border of the IPL. A: The visual streak region of a turtleretina stimulated with NMDA. There were numerous bipolar cellswith both normal (horizontal arrowheads) and displaced (verticalarrowheads) somata with elevated cGMP-LI. In this region of theretina, only sGC A1 amacrine cells were labeled (arrows). Threediscreet bands of cGMP-LI were visible in S15–S25, S30–S35, and S60of the IPL (asterisks). Outside of the visual streak, the S15–S25 andS30–S35 bands often were fused. B: The visual streak region of aturtle retina stimulated with KA. Like with NMDA, both normal and

displaced bipolar cell somata had elevated cGMP-LI. Similarly, onlysGC A1 amacrine cells had increased cGMP-LI (arrow). Two bands ofimmunoreactivity were present at S15–S25 and S30–S35 in the IPL(asterisks). C: The visual streak region (asterisk) stimulated withNMDA in the presence of 100 µM 7-nitroindazole. There was nosignificant cGMP-LI. Similar results were obtained with N-nitro-L-arginine (L-NNA). D: In the visual streak region (asterisk) stimulatedwith KA in the presence of 100 µM 7-nitroindazole, there was also nosignificant cGMP-LI. Similar results were obtained with L-NNA.Abbreviations as defined in Fig. 1. Scale bars 5 25 µm.

NMDA- AND KA-STIMULATED cGMP IN RETINA 81

retinas with 100 µM 7-NI, occasional bipolar cell somatashowed slightly increased cGMP-LI in the superior andinferior retina adjacent to the ora serrata. Similarly,KA-stimulated retinas in the presence of 100 µM 7-NI hadfaint cGMP-LI in isolated bipolar cell somata in theinferior retina adjacent to the ora serrata.

ELISA assay of cGMP-LI

Stimulation of turtle eye cups with either NMDA or KAresulted in statistically significant increases in the totalretinal levels of cGMP-LI. All incubations were conductedin normal room light. Control eye cups incubated in IBMXcontaining Ringer’s solution had basal concentrations ofcGMP-LI of 12.3 6 1.1 pM cGMP per mg of protein 6S.E.M (n 5 6). NMDA-stimulated retinas contained 24.2 63.8 pM cGMP per mg of protein 6 S.E.M. (n 5 6). Inresponse to KA, the mean concentration of cGMP was19.4 6 1.8 pM cGMP per mg of protein 6 S.E.M. (n 5 6).

DISCUSSION

Stimulation of ionotropic glutamate receptors with KAor NMDA significantly increased cGMP-LI in select turtleretinal neurons. These increases were seen in a subset ofneurons that have been shown previously to contain sGC.The use of two different NOS inhibitors confirmed thatthese agonist-induced increases in cGMP-LI were due toNO stimulation of sGC. Moreover, the cell types exhibitingincreased cGMP-LI in response to KA or NMDA corre-sponded with the cell types that are most sensitive to NO(with the interesting exception of sGC A11). This indicatesthat ionotropic glutamate receptors have the ability toactivate a subset of sGC-containing cells, and it demon-strates that NO signaling is not an ‘‘all or none’’ responsein terms of activation of sGC. The increases we saw in thetotal retinal cGMP, as determined by ELISA, were consis-tent with the increases in cGMP-LI we observed immuno-cytochemically.

Differences in cGMP-LI with NMDA vs. KA

Regional differences. Although many of the same celltypes had increased cGMP-LI in response to both agonists,there were distinct regional differences seen in the re-sponse to KA vs. NMDA. There were five amacrine celltypes with cGMP-LI in the NMDA-stimulated retinas andonly four amacrine cell types in the KA-stimulated retinas.Both agonists produced cGMP-LI in S15–S35 in the visualstreak, although, in the KA-stimulated retinas, the overalllevels of cGMP-LI were less than seen with NMDA.Stimulation with KA increased cGMP-LI in amacrine cellsthat were distributed relatively uniformly throughout theretina. In the NMDA-stimulated retinas, there were dra-matically fewer bipolar cells with cGMP-LI above thevisual streak compared with those below the streak. Therewere also fewer sGC A3 amacrine cells with cGMP-LIabove the streak when stimulated with NMDA.

These regional differences could occur at multiple levels.The first level could be a regional difference in the distribu-tion of NMDA or KA receptors. Although no immunocyto-chemical studies have been done in turtle retina, both KAglutamate receptor (GluR) subunits GluR6/7 and KA2(Brandstatter et al., 1997) and NMDA receptor subunitNR2A (Hartveit et al., 1994) have been localized immuno-cytochemically in mammalian retina. In these studies,these KA and NMDA receptor subunits were found in both

the ON region and the OFF region of the IPL. Thus,activation of either receptor type could potentially stimu-late the NOS amacrine cells to release NO in all strata ofthe IPL. However, neither of these studies referred to anyregional specializations in the localization of either KA orNMDA receptor subunits.

Previous studies in the turtle retina have determinedthat stimulation with NMDA or KA can modulate levels ofseveral neurotransmitters or neuropeptides (Yaqub andEldred, 1993) or transcription factors (Yaqub et al., 1995),and receptor binding studies have localized NMDA recep-tors on some somata in the INL (Cooper et al., 1990).However, none of these studies describe any regionalspecializations of NMDA or KA receptors in the turtleretina. However, the possibility still exists that a regionalspecialization in glutamate receptor subtypes may explainwhy NMDA increases cGMP-LI intensely in bipolar cells inand below the visual streak but weakly above the visualstreak or why KA-stimulated retinas had uniform cGMP-LIin bipolar cells throughout the retina. Similarly, the differ-ences seen above and below the visual streak in the celldensity and the intensity of cGMP-LI in the sGC A1 andsGC A2 amacrine cells in the NMDA-stimulated retinasalso suggest that there might be regional differences in thetypes of glutamate receptors on the NOS-containing cells,most likely the NOS amacrine cells. The idea that NOS-containing amacrine cells have different receptor types issupported by the excitotoxicity study of Sagar (1990),which indicated that two NADPH-diaphorase positiveamacrine cell types in the rabbit retina have differentexcitotoxic responses to NMDA vs. KA.

The second possibility is that there are regional special-izations in the localization of NOS in the turtle retina. Todate, studies have examined only the localization of nNOSin turtle. nNOS is found in three amacrine cell types thatarborize in both the ON and the OFF layers of the IPL, inphotoreceptors, and in somata in the GCL (Blute et al.,1997). There were no obvious regional specializations inthe distribution of these nNOS-positive neurons in theturtle retina (Blute et al., 1997). At the concentrations ofnNOS inhibitor we used, we cannot eliminate the potentialinvolvement of endothelial NOS (eNOS) in the increases incGMP-LI that we see in response to NMDA or KA. Nothingis known about the localization of eNOS in turtle retina.Therefore, there is no positive support that regional differ-ences in the localization of NOS can explain our results.

The next possible explanation could be regional differ-ences in the localization of sGC or its sensitivity foractivation by NO. A recent study (Blute et al., 1998) haslocalized sGC functionally in the turtle retina. Because thesGC was localized functionally and not by using immunocy-tochemistry, it is not possible to discriminate regionaldifferences in enzyme levels from regional differences inthe activation of sGC by NO. However, these studies didindicate that low doses of NO do stimulate higher levels ofcGMP-LI in the peripheral retina vs. the central retina.This peripheral bias is similar to the higher levels ofEAA-stimulated cGMP-LI we see in peripheral retina.Thus, it is possible that regional differences in sGC itself orin the activation of sGC by NO may explain some of ourregional results. However, other factors also must beinvolved, because there were no dramatic differences inthe activation of sGC above vs. below the visual streak(Blute et al., 1998).

82 T.A. BLUTE ET AL.

Finally, it is possible that there may be regional differ-ences in phosphodiesterase (PDE) activity. The prelimi-nary results of Eldred et al. (1998) indicate that there aresome differences in the function of type I Ca21/calmodulin-dependent and type V cGMP-specific PDE above vs. belowthe visual streak in the turtle retina. In total, all of theseresults raise the possibility that a combination of topo-graphic specializations in receptor type, sGC, and PDEmay be responsible for the regional differences we see inNMDA- and KA-stimulated cGMP-LI.

The existence of regional specializations in many differ-ent species indicates that they are likely to play animportant role in visual processing. In the cat, retinalganglion cells located in the periphery of the retina havereceptors for both g-aminobutyric acid (GABA) and gly-cine, whereas those in the area centralis have receptors foronly one or the other of these transmitters, but not for both(Priest et al., 1985). In the turtle retina, there is evidencefor regional differences in synaptic connectivity. For in-stance, there are two distinct classes of amacrine cellscontaining corticotropin-releasing factor-like immunoreac-tivity (CRF-LI): one is found only in the visual streak, andthe other class is found only below the visual streak, withno cells with CRF-LI above the visual streak (Williamsonand Eldred, 1989). The cells with CRF-LI in the visualstreak have different synaptic connectivity than the cellsbelow the visual streak (Williamson and Eldred, 1991). Itis possible that the relatively higher levels of NMDA- orKA-stimulated cGMP-LI below vs. above the visual streakmay reflect a regional difference in visual processing.Perhaps there are specializations in the inferior retina toenhance the detection of motion to defend the turtles fromlarger predators or to detect potential prey in the visualspace above.

Pharmacologic differences. The original classifica-tions of the sGC cells and their sensitivity to NO (Blute etal., 1998) were determined by using different concentra-tions of NO donor in the presence of PDE inhibitors. In thisprevious study, the sGC A11 amacrine cell type was not oneof the most sensitive cell types to NO. In the present study,in addition to the more NO-sensitive amacrine cell types,the sGC A11 amacrine cell type was observed with NMDAstimulation but not with KA stimulation. This indicatesthat NMDA produces some additional physiological effectthat results in increased cGMP-LI in the sGC A11 cell typein the presence of low levels of NO. For example, normallyactive PDE may be inhibited, thereby allowing the level ofcGMP to be preserved at a detectable level. Alternatively,sGC A11 cells may be positioned ideally to receive NO fromNMDA receptor-containing NOS cells. nNOS has beenshown to be present at discreet pre- and postsynapticlocations (Haverkamp and Eldred, 1998). It is possiblethat close proximity between the source of NO from theprocesses of the NMDA-activated, NOS-containing cellsand the sGC A11 cell processes may create a microenviron-ment with a high enough concentration of NO to activatethe sGC in the sGC A11 cells. The presence of cGMP-LI insGC A11 amacrine cells may be a combination of bothfactors.

It is also possible that some of these differences seenwith NMDA vs. KA may be due to differences in the level ofactivation of calcium/calmodulin-dependent NOS (Bredtand Snyder, 1990). Theoretically, the voltage-dependentblockage of NMDA receptors by magnesium could reducethe effects of NMDA stimulation (Hollman and Heine-

mann, 1994; Schoepfer et al., 1994). However, previousstudies in turtle retina have found that stimulation withNMDA can dramatically modulate levels of several neuro-transmitters and neuropeptides (Yaqub and Eldred, 1993)or transcription factors (Yaqub et al., 1995) throughout theretina. Furthermore, the precise NMDA receptor subunitsthat are present in turtle retina are not known, and someNMDA receptors containing NMDAR2C subunits are rela-tively insensitive to magnesium blockade (Hollman andHeinemann, 1994). It is also possible that the NMDA-stimulated cells already may be depolarized enough toremove any potential voltage sensitive blockade of NMDAstimulation.

Although the difference was not statistically significant,there was more cGMP produced with NMDA stimulationthan with KA stimulation, as measured by ELISA. Thismay be related to the fact that activated NMDA channelsflux more calcium than KA channels (MacDermott et al.,1986; Hollman and Heinemann, 1994). When the higherbiochemical levels of cGMP-LI are considered along withthe smaller overall numbers of cells activated, this wouldsuggest that NMDA stimulation can raise intracellularlevels of cGMP higher than KA stimulation.

Involvement of NOS

In the presence of either NOS inhibitor, 7-NI or L-NNA,there were no significant increases in cGMP-LI in responseto either KA or NMDA. These results strongly support thefinding that both KA and NMDA can stimulate NO releasein the retina, as they have been shown to do in the brain(Garthwaite, 1991; Okada, 1992). Our results also areconsistent with the presence of NMDA and KA receptorson NADPH-positive amacrine cells in the rabbit retina(Sagar, 1990) and with a biochemical study of embryonicchick retina (Zeevalk and Nicklas, 1994) that found thateither NMDA or KA stimulation can increase cGMP levelstwo- to threefold over basal levels. Moreover, in chickretina, these increases in cGMP are inhibited by the NOSinhibitor L-NNA and are blocked by hemoglobin, both ofwhich support the involvement of NO (Zeevalk and Nick-las, 1994). In an earlier study of EAA receptor activation inchick retina (Anand et al., 1985), either NMDA or KAsignificantly increased cGMP levels (Table 2) to valuesthat were closely comparable to the changes in cGMPlevels we found in turtle.

One consideration when using NOS inhibitors in intactretinal preparations is that, in biochemical studies ofretinal homogenates, micromolar concentrations are suffi-cient to totally block NOS activity. In intact retina, concen-trations 20–3000 times higher are needed to block NOSactivity (Wellard et al., 1995). Incomplete inhibition of

TABLE 2. Changes in Total Retinal Cyclic Guanosine Monophosphatein Response to Stimulation With N-Methyl-D-Aspartate or Kainic Acid1

StimulantChick retina

(Anand et al., 1985)Turtle retina

(mean 6 S.E.M.)

Control 6 pmol/mg 12 pmol/mg 6 1 (n 5 6)NMDA

100 µM 22 pmol/mg 24 pmol/mg 6 3 (n 5 6)*300 µM 32 pmol/mg

KA10 µM 19 pmol/mg 6 2 (n 5 6)*

100 µM 42 pmol/mg

1NMDA, N-methyl-D-aspartate; KA, kainic acid.*Different from control, at P , 0.05 (analysis of variance with Student-Neuman-Keulsmultiple-comparison, post-hoc test).

NMDA- AND KA-STIMULATED cGMP IN RETINA 83

NOS is supported by the lack of attenuation of cGMP-LI inresponse to NMDA stimulation by 10 µM 7-NI, whereas100 µM 7-NI or 1 mM L-NNA eliminated almost all of theincreases in cGMP-LI in response to both glutamateagonists.

Physiological implications

It has been shown that, in the turtle retina, the stratifi-cation of the IPL obeys the functional division of ON(S40–S100) and OFF (S0–S40) sublamina (Ammermullerand Kolb, 1995.) It is tempting to speculate that thestratification of the sGC neurons with NMDA- or KA-stimulated increases in cGMP-LI may relate to activationof the ON or OFF pathways in the retina. However, we findthat light and dark adaptation has little effect on immuno-cytochemically detected cGMP-LI, and a previous studywas unable to detect changes in cGMP-LI in photorecep-tors or many ON-bipolar cells (Blute et al., 1998). Thiswould suggest that the levels of cGMP-LI that change inresponse to light or dark are below the levels of immunocy-tochemical detection (Blute et al., 1998) and that thechanges we see are due to ionotropic glutamate receptorsactivating a distinct NO/cGMP signal-transduction path-way.

This still leaves the question regarding the stratificationof the cGMP-LI that we find in the IPL. Presumably, KA orNMDA are activating nNOS, probably in nNOS-contain-ing amacrine cells, and the NO produced is activating sGCto increase cGMP-LI. In the turtle retina, nNOS-LI isfound in amacrine cells that arborize in both the ON andOFF regions of the IPL; therefore, NO could be released inboth of these regions as well. With the exception of the sGCA11 cells, as described above, the cells with NMDA- orKA-stimulated increases in cGMP-LI are the sGC cellsthat are most sensitive NO. Thus, it appears that thestratification of cGMP-LI we see is not a reflection of theselective activation of the ON or OFF pathways in the IPLbut, rather, the relative sensitivity of particular sGC cellsto NO. This idea is supported by the fact that, even thoughNMDA or KA increase cGMP-LI in a limited subset ofsGC-containing neurons, the stimulated cells arborize inboth the ON and OFF sublamina of the turtle retina.

In conclusion, the stimulation with either NMDA or KAincreases cGMP-LI in specific retinal cell types withdistinct regional variations. The blocking of these in-creases in cGMP-LI by NOS inhibitors indicates theinvolvement of the NOS/NO/cGMP signal-transductionpathway. These results indicate that activation of iono-tropic glutamate receptors can lead to the activation ofcGMP second-messenger systems in a wide variety ofretinal neurons. Our results also indicate that, by workingthrough NO, activation of glutamate receptors can in-crease levels of cGMP in cells that may or may not evenhave glutamate receptors.

ACKNOWLEDGMENTS

We wish to thank Felicitas B. Eldred for her excellenttechnical assistance.

LITERATURE CITED

Ahmad I, Barnstable CJ. 1993. Differential laminar expression of particu-late and soluble guanylate cyclase genes in rat retina. Exp Eye Res56:51–62.

Ahmad I, Leinders-Zufall T, Kocsis JD, Shepherd GM, Zufall F, BarnstableCJ. 1994. Retinal ganglion cells express a cGMP-gated cation conduc-tance activatable by nitric oxide donors. Neuron 12:155–165.

Ammermuller J, Kolb H. 1995. The organization of the turtle inner retina.I. ON- and OFF-center pathways. J Comp Neurol 358:1–34.

Anand H, Roberts PJ, Lopez-Colome AM. 1985. Excitatory amino acids inthe chick retina: possible involvement of cyclic guanosine monophos-phate. Neurosci Lett 58:31–36.

Barthel LK, Raymond PA. 1990. Improved method for obtaining 3 µmcryosections for immunocytochemistry. J Histochem Cytochem 38:1383–1388.

Blute TA, Mayer B, Eldred WD. 1997. Immunocytochemical and histochemi-cal localization of nitric oxide synthase in the turtle retina. Vis Neurosci14:717–729.

Blute TA, Velasco P, Eldred WD. 1998. Functional localization of solubleguanylate cyclase in the turtle retina: modulation of cGMP by nitricoxide donors. Vis Neurosci 15:485–498.

Brandstatter JH, Koulen P, Wassle H. 1997. Selective synaptic distributionof kainate receptor subunits in the two plexiform layers of the ratretina. J Neurosci 17:9298–9307.

Bredt DS, Snyder SH. 1990. Isolation of nitric oxide synthase, a calmodulin-requiring enzyme. Proc Natl Acad Sci USA 87:682–685.

Chrisman TD, Schulz S, Garbers DL. 1992. Guanylyl cyclases: ligands andfunctions. Cold Spring Harbor Symp Quant Biol 57:155–161.

Cooper EJ, Millar TJ, Anderton PJ. 1990. Distribution of [3H]MK-801binding in the turtle retina: an autoradiographical study. Neurosci Lett112:127–132.

de Vente J, Steinbusch HWM. 1992. On the stimulation of soluble andparticulate guanylate cyclase in the rat brain and the involvement ofnitric oxide as studied by cGMP immunocytochemistry. Acta Histochem92:13–38.

DeVries SH, Schwartz EA. 1989. Modulation of an electrical synapsebetween solitary pairs of catfish horizontal cells by dopamine andsecond messengers. J Physiol 414:351–375.

Dizhoor AM, Lowe DG, Olshevskaya EV, Laura RP, Hurley JB. 1994. Thehuman photoreceptor membrane guanylyl cyclase, retGC, is present inouter segments and is regulated by calcium and a soluble activator.Neuron 12:1345–1352.

Duda T, Goraczniak RM, Sitaramayya A, Sharma RK. 1993. Cloning andexpression of an ATP-regulated human retina C-type natriuretic factorreceptor guanylate cyclase. Biochemistry 32:1391–1395.

Eldred WD, Blute TA, Mikaelian S. 1998. Functional localization of type ICa21/calmodulin-dependent and type V cGMP-specific phosphodiester-ase in the turtle retina. Invest Ophthalmol Vis Sci 39(Suppl):986.

Ellis J, Halliday G. 1992. A comparative study of avidin-biotin-peroxidasecomplexes for the immunohistochemical detection of antigens in neuraltissue. Biotech Histochem 67:367–371.

Garthwaite J. 1991. Glutamate, nitric oxide and cell-cell signaling in thenervous system. TINS 14:60–67.

Hartveit E, Brandstatter JH, Sassoe-Pognetto M, Laurie DJ, Seeburg PH,Wassle H. 1994. Localization and developmental expression of theNMDA receptor subunit NR2A in the mammalian retina. J CompNeurol 348:570–582.

Haverkamp S, Eldred WD. 1998. Localization of nNOS in photoreceptor,bipolar and horizontal cells in turtle and rat retinas. Neuroreport9:2231–2235.

Hollman M, Heinemann S. 1994. Cloned glutamate receptors. Annu RevNeurosci 17:31–108.

Koch K-W, Lambrecht H-G, Haberecht M, Redburn D, Schmidt HHHW.1994. Functional coupling of a Ca21/calmodulin-dependent nitric oxidesynthase and a soluble guanylyl cyclase in vertebrate photoreceptorcells. EMBO J 13:3312–3320.

Kusaka S, Dabin I, Barnstable CJ, Puro DG. 1996. cGMP-mediated effectson the physiology of bovine and human retinal Muller (glial) cells. JPhysiol (London) 497:813–824.

Kutty RK, Fletcher RT, Chader GJ, Krishna G. 1992. Expression ofguanylate cyclase-A mRNA in the rat retina: detection using polymer-ase chain reaction. Biochem Biophys Res Commun 182:851–857.

Lowe DG, Dizhoor AM, Liu K, Gu Q, Spencer M, Laura R, Lu L, Hurley JB.1995. Cloning and expression of a second photoreceptor-specific mem-brane retina guanylyl cyclase (RetGC), RetGC-2. Proc Natl Acad SciUSA 92:5535–5539.

MacDermott AB, Mayer ML, Westbrook GL, Smith SJ, Barker JL. 1986.NMDA-receptor activation increases cytoplasmic calcium concentra-tion in cultured spinal cord neurones. Nature 321:519–522.

84 T.A. BLUTE ET AL.

Mills SL, Massey SC. 1995. Differential properties of two gap junctionalpathways made by AII amacrine cells. Nature 377:734–737.

Miyachi E-I, Murakami M. 1991. Synaptic inputs to turtle horizontal cellsanalyzed after blocking of gap junctions by intracellular injections ofcyclic nucleotides. Vis Res 31:631–635.

Nawy S, Jahr CE. 1991. cGMP-gated conductance in retinal bipolar cells issuppressed by the photoreceptor transmitter. Neuron 7:677–683.

Okada D. 1992. Two pathways of cyclic GMP production through glutamatereceptor-mediated nitric oxide synthesis. J Neurochem 59:1203–1210.

Priest TD, Robbins J, Ikeda H. 1985. The action of inhibitory neurotransmit-ters, gamma-aminobutyric acid and glycine may distinguish betweenthe area centralis and the peripheral retina in cats. Vis Res 25:1761–1770.

Redinbaugh MG, Campbell WH. 1985. Adaptation of the dye-bindingprotein assay to microtiter plates. Anal Biochem 147:144–147.

Sagar SM. 1990. NADPH-diaphorase reactive neurons of the rabbit retina:differential sensitivity to excitotoxins and unusual morphological fea-tures. J Comp Neurol 300:309–319.

Schoepfer R, Monyer H, Sommer B, Wisden W, Sprengel R, Kuner T, LomeliH, Herb A, Kohler M, Burnashev N, Gunther W, Ruppersberg P,Seeburg P. 1994. Molecular biology of glutamate receptors. ProgrNeurobiol 42:353–357.

Shiells RA, Falk G. 1992. Properties of the cGMP-activated channel ofretinal on-bipolar cells. Proc R Soc London [Biol] 247:21–25.

Shyjan AW, de Sauvage FJ, Gillett NA, Goeddel DV, Lowe DG. 1992.Molecular cloning of a retina-specific membrane guanylyl cyclase.Neuron 9:727–737.

Wellard JW, Miethke P, Morgan IG. 1995. Neural barriers affect the actionof nitric oxide synthase inhibitors in the intact chicken retina. NeurosciLett 201:17–20.

Williamson DE, Eldred WD. 1989. Amacrine and ganglion cells withcorticotropin-releasing-factor-like immunoreactivity in the turtle retina.J Comp Neurol 280:424–435.

Williamson DE, Eldred WD. 1991. Synaptic organization of two types ofamacrine cells with CRF-like immunoreactivity in the turtle retina. VisNeurosci 6:257–269.

Yau KW, Baylor DA. 1989. Cyclic GMP-activated conductance of retinalphotoreceptor cells. Annu Rev Neurosci 12:289–327.

Yaqub A, Eldred WD. 1993. Effects of excitatory amino acids on immunocy-tochemically identified populations of neurons in turtle retina. JNeurocytol 22:644–662.

Yaqub A, Guimaraes M, Eldred WD. 1995. Neurotransmitter modulation ofthe expression of c-fos- and c-jun-like proteins in the turtle retina. JComp Neurol 354:481–500.

Zeevalk GD, Nicklas WJ. 1994. Nitric oxide in retina: relation to excitatoryamino acids and excitotoxicity. Exp Eye Res 58:343–350.

NMDA- AND KA-STIMULATED cGMP IN RETINA 85