Transient, High Levels of SNAP-25Expression in Cholinergic Amacrine Cells

During Postnatal Developmentof the Mammalian Retina

M.H. WEST GREENLEE,1,2,3 S.K. FINLEY,1 M.C. WILSON,5 C.D. JACOBSON,1,2,4

AND D.S. SAKAGUCHI1,2,3*1Department of Zoology and Genetics, Iowa State University, Ames, Iowa 50011

2Neuroscience Program, Iowa State University, Ames, Iowa 500113Signal Transduction Training Group, Iowa State University, Ames, Iowa 500114Department of Veterinary Anatomy, Iowa State University, Ames, Iowa 50011

5Department of Biochemistry, University of New Mexico School of Medicine,Albuquerque, New Mexico 87131

ABSTRACTIn the present study, we have examined the development of cholinergic amacrine cells in

the retina of the Brazilian opossum, Monodelphis domestica. An antibody directed againstcholine acetyltransferase (ChAT) revealed that ChAT-like immunoreactivity (ChAT-IR) wasfirst observed at 15 days postnatal (15PN). By 25PN, ChAT-IR identified two matchingpopulations of amacrine cells in the inner nuclear and ganglion cell layer. Bromodeoxyuridinebirth-dating analysis coupled with immunolabeling with the anti-ChAT antibody revealedthat the cholinergic amacrine cells are born postnatally, between 2PN and 15PN. In addition,we have examined the differentiation of the cholinergic amacrine cells by using an antibodydirected against a presynaptic terminal-associated protein, synaptosomal-associated proteinof 25 kDa (SNAP-25). Double-labeling analysis revealed that relatively high levels ofSNAP-25-IR were selectively present in cholinergic amacrine cells prior to eye opening.However, in the mature retina, high levels of SNAP-25-IR were no longer observed in theChAT-IR amacrine cells. These results reveal a distinct period in development, prior to eyeopening, when high levels of SNAP-25-IR are selectively expressed in cholinergic amacrinecells. The specificity and time course of the high levels of SNAP-25 in cholinergic amacrinecells may be critical in mediating the transient properties of these cells during visual systemdevelopment. J. Comp. Neurol. 394:374–385, 1998. r 1998 Wiley-Liss, Inc.

Indexing terms: Monodelphis domestica; starburst amacrine cells; retinal development

Cholinergic amacrine cells have been identified in allmammalian retinas examined thus far (Masland et al.,1984b; Marc, 1986; Wassle et al., 1987) and are thereforelikely to play important conserved functional roles fromspecies to species. Cholinergic amacrine cells, also knownas starburst amacrine cells because of their distinctivemorphology, consist of two mirror-symmetric subpopula-tions, with their somata located in the inner nuclear layer(INL) and ganglion cell layer (GCL) and their processesramifying principally in sublaminae 2 and 4 of the innerplexiform layer (IPL), respectively (Famiglietti, 1983;Karten and Brecha, 1983; Marc, 1986). In the matureretina, cholinergic amacrine cells may be involved in theproduction of directional selectivity and may also have animportant role in regulating the responsiveness of the

ganglion cells (Ariel and Daw, 1982; Masland et al.,1984a). In addition to their function in the mature retina,

Grant sponsor: National Science Foundation; Grant number: IBN-9311054; Grant sponsor: Whitehall Foundation; Grant number: M93-25;Grant sponsor: Iowa State University Biotechnology Council; Grant num-ber: 102-47-45; Grant sponsor: The Carver Trust; Grant number: 701-17-52;Grant sponsor: The Sigma Xi Research Foundation; Grant sponsor: HowardHughes Medical Institute Education Initiative; Grant sponsor: Hatch Actand State of Iowa Funds.

S.K. Finley’s current address: Department of Neurobiology and Behavior,State University of New York at Stonybrook, Stonybrook, NY 11794.

*Correspondence to: D.S. Sakaguchi, Department of Zoology and Genet-ics, 339 Science II, Iowa State University, Ames, IA 50011.E-mail: dssakagu@iastate.edu

Received 8 September 1997; Revised 23 December 1997; Accepted 29December 1997

THE JOURNAL OF COMPARATIVE NEUROLOGY 394:374–385 (1998)

r 1998 WILEY-LISS, INC.

cholinergic amacrine cells appear to have a distinct devel-opmental role. Feller et al. (1996) recently demonstratedthe requirement of cholinergic synaptic transmission inthe propagation of spontaneous retinal waves. Becausestarburst amacrine cells are the only cholinergic neuronsin the developing mammalian retina, they appear to be thecritical cellular element required for wave production(Feller et al., 1996). The spontaneous, patterned activity,generated before visual stimulation, is likely to play anessential role during the development and differentiationof synaptic connectivity within the visual pathway (re-viewed in Goodman and Shatz, 1993; Shatz, 1996).

This period of spontaneous, correlated activity prior toeye opening corresponds to an exceptionally dynamicepoch in visual system development involving processoutgrowth and synaptogenesis. In the developing mamma-lian retina, the expression of specific presynaptic terminal-associated proteins appears to be developmentally regu-lated (Catsicas et al., 1992; West Greenlee et al., 1996;Alexiades and Cepko, 1997) and therefore may be criticalfor synaptogenesis. In particular, synaptosomal-associ-ated protein of 25 kDa (SNAP-25), is predominantlyassociated with neurons and has multiple functions includ-ing neurotransmitter release (reviewed in Bark and Wil-son, 1994; Sudhof, 1995; Calakos and Scheller, 1996),neurite outgrowth, differentiation, and synaptogenesis(Osen-Sand et al., 1993, 1996; Catsicas et al., 1994). Thus,the regulated spatial and temporal expression of SNAP-25in the developing retina may play an integral role in theestablishment of appropriate retinal circuitry.

We have examined the genesis and differentiation ofcholinergic amacrine cells in relation to the differentialexpression of SNAP-25-like immunoreactivity (IR) in thedeveloping retina of the Brazilian opossum, Monodelphisdomestica. From the onset of detectable choline acetyltrans-ferase immunoreactivity (ChAT-IR) at 15 days postnatal(15PN), these cells displayed relatively high levels ofSNAP-25-IR in their soma and processes relative to theSNAP-25-IR in surrounding retinal tissue. This relativelyhigh level of SNAP-25 expression was transient, peakedaround 25PN, and decreased to a lower level by the time ofeye opening, around 35PN. This is the first study to reportthe transient and specific localization of relatively highlevels of SNAP-25-IR within cholinergic amacrine cellsduring postnatal development prior to eye opening. Thespecific spatial and temporal localization within ChAT-IRamacrine cells suggests that this regulated expression ofSNAP-25 may be critical for developmental plasticity inthe mammalian visual system.

MATERIALS AND METHODS

Animals

Brazilian opossums were obtained from a colony main-tained at Iowa State University (Ames, IA). The animalsused to maintain the breeding colony were obtained fromthe Southwest Foundation for Research and Education(San Antonio, TX). Laboratory procedures were carried outin accordance with guidelines and had the approval of theIowa State University Committee on Animal Care. Theanimals were maintained in a constant-temperature envi-ronment (26°C) on a 14:10-hour light:dark cycle with food(Reproduction Fox Chow, Milk Specialties Products, Madi-son, WI) and water ad libitum. The Brazilian opossum hasa gestation period of approximately 14.5 days. The date of

birth was designated as postnatal day 1 (1PN) and eyeopening occurs at approximately 35PN. Pups are weanedfrom the mother at 60PN (Kuehl-Kovarik et al., 1995).

Tissue preparation

We have examined retinal development in opossumpups at postnatal days 1, 5, 10, 15, 20, 25, 30, and 35 and inadult opossums (1.5–2 years). Pups between 1PN and15PN were anesthetized by hypothermia. The animalswere decapitated, and their heads were fixed by immersionin 4% paraformaldehyde in 0.1M PO4 buffer (pH 7.5) for 48hours. Postnatal day 20, 25, 30, and 35 pups were deeplyanesthetized with ether and perfused transcardially with4% paraformaldehyde. The heads were then postfixed for48 hours. Adult opossums were anesthetized, perfusedtranscardially with 4% paraformaldehyde, and enucle-ated, and eyes were subsequently postfixed for 48 hours.Tissue was cryoprotected in a 30% sucrose solution in0.1 M PO4 buffer (pH 7.4), embedded in OCT freezingmedium (Baxter Scientific, McGaw Park, IL), and sec-tioned coronally at 20 µm on a cryostat. Sections werethaw-mounted onto Superfrost microscope slides (FisherScientific, Pittsburgh, PA) and stored at -20°C until processed.

Immunohistochemistry

Identification of cholinergic neurons in the Monodelphisretina was carried out by using immunohistochemicalprocedures with an antibody directed against ChAT, thesynthetic enzyme for acetylcholine (Tucek, 1976). Previousstudies have demonstrated that ChAT-IR neurons in themammalian retina have the same morphological character-istics as the cholinergic starburst amacrine cells (Famigli-etti, 1983; Voigt, 1986). Starburst amacrine cells are theonly cholinergic cells that have been described in themammalian retina, and, thus, ChAT-IR and its character-istic staining pattern have been used as a marker forstarburst amacrine cells (Hutchins and Hollyfield, 1987b;Dann, 1989; Feller et al., 1996).

Bromodeoxyuridine birth-dating analysis of cholin-

ergic amacrine cells. Procedures using bromodeoxyuri-dine (BrdU) to identify newly generated cells were per-formed as described by Iqbal et al. (1995). Female opossumswith pups at various postnatal ages (1, 2, 3, 4, 5, 10, 12, 15,20, and 25PN) were anesthetized with 2% isoflurane. Theopossum pups were then injected subcutaneously alongthe dorsal midline with 20 µl of BrdU solution (1 µg/ml;Sigma Chemical Co., St. Louis, MO) in 0.15 M NaCl whilestill attached to their anesthetized mothers. The pupswere allowed to survive until 30PN or 60PN, at which timethey were anesthetized and prepared for immunohisto-chemical analysis as described above. The tissue sectionswere rinsed with 0.5 M potassium phosphate bufferedsaline (KPBS; 0.15 M NaCl, 0.034 M K2HPO4, 0.017 MKH2PO4, pH 7.4) and pretreated with 0.06% trypsin (Bo-vine type III; Sigma) and 5.4 3 10-3M CaCl2 in KPBS for 30minutes at 37°C. After washing for 10 minutes with KPBS,the tissue sections were treated with 0.1 N HCl (ice cold)for 10 minutes followed by incubation in 2 N HCl at 37°Cfor 30 minutes. The tissue sections were then neutralizedin basic 0.5 M KPBS (pH 8.5) and subsequently processedfor routine immunohistochemistry.

Slide-mounted tissue sections were rinsed in KPBS, andendogenous peroxidase activity was eliminated by a 30-minute incubation in 0.3% hydrogen peroxide solution inKPBS. The sections were then incubated for 2 hours in

SNAP-25 IN CHOLINERGIC AMACRINE CELLS 375

blocking solution with normal horse serum (Vector Labora-tories Inc., Burlingame, CA) and incubated overnight atroom temperature (24°C) in anti-BrdU antibody (DAKOCorporation, Carpinteria, CA). On the following day, tissuesections were rinsed in KPBS with 0.2% Triton X-100 andincubated in biotinylated secondary antibody (Vector) for 2hours at room temperature, rinsed, and incubated inhorseradish peroxidase-avidin-biotin complex for 1 hour atroom temperature (Vector Elite ABC Kit; 1:200). To visual-ize the BrdU-IR nuclei, the tissue was reacted with asubstrate of 0.4% 38,38-diaminobenzidine tetrahydrochlo-ride (DAB; Sigma), 0.16 M nickel sulfate (Sigma), and0.05% hydrogen peroxide (Fisher) in 0.1 M sodium acetate(Fisher). The reaction was terminated in successive rinsesof 0.15 M NaCl solution. To determine which BrdU-IR cellswere cholinergic amacrine cells, sections were processedfor ChAT immunohistochemistry as described above, ex-cept that after incubation in secondary antibody, tissueswere incubated in ABC complex (1:600) and nickel sulfatewas omitted from the DAB solution. The resulting ChAT-IRappeared as a brown reaction product and could easily bedistinguished from the dark-purple BrdU-IR. Sectionswere then dehydrated through a graded ethanol series,cleared with Americlear (Baxter), and coverslipped withAccumount (Baxter).

Negative controls were run in parallel during all immu-nohistological processing by the omission of the primary orsecondary antibodies. No antibody staining was observedin the controls.

ChAT/SNAP-25 double-label immunohistochemistry.

Slide-mounted tissue sections were rinsed in 0.5 M KPBS(pH 7.4), incubated for 2 hours in blocking solution (KPBSwith 1% bovine serum albumin [BSA; Sigma], 0.4% TritonX-100 [Fisher], and 1.5% normal donkey serum [NDS;Jackson ImmunoResearch, West Grove, PA]) and incu-bated overnight at 24°C in anti-ChAT antibody (ChemiconInternational, Temecula, CA). On the following day, tissuesections were rinsed in KPBS with 0.2% Triton X-100 andincubated in biotinylated secondary antibody (Jackson). Tovisualize the antibody staining pattern, sections were thenincubated in Texas-Red-conjugated avidin D (1:1,500; Vec-tor). After thorough rinsing, tissue sections were incubatedin blocking solution (PBS: 14 mM NaCl, 2.7 mM KCL, 5.37mM Na2HPO4, 1.76 mM KH2PO4, pH 7.4; with 5% goatserum, 0.4% BSA, and 0.2% Triton X-100) for 30 minutesand incubated in diluted anti-SNAP-25 antibody at 4°Covernight. After washes in PBS with 0.25% Triton X-100,SNAP-25-IR was visualized by incubation with a fluores-cein isothiocyanate (FITC)-conjugated secondary antibody(Fisher). After subsequent rinsing, slides were cover-slipped with Vectashield fluorescence mounting medium(Vector).

Antibodies. The anti-ChAT antibody (produced in goat;Chemicon) was diluted at 1:200. Bromodeoxyuridine wasdetected by using an anti-BrdU antibody (mouse monoclo-nal; DAKO) and was diluted 1:200. Anti-SNAP-25 rabbitantiserum was raised against a synthetic peptide corre-sponding to the carboxyl-12 residues of the mouse protein(Oyler et al., 1989) and was diluted at 1:200. Secondaryantibodies were diluted in KPBS with 1.5% NBS, 1% BSA,0.02% Triton X-100 (biotinylated donkey-anti-goat [1:200;Jackson], biotinylated horse anti-mouse [1:200; Vector]) orin PBS with 5% goat serum, 0.4% BSA, and 0.2% TritonX-100 (FITC-goat-anti-rabbit 1:200; Fisher). The specific-

ity of the SNAP-25 antisera has been previously examinedin immunoblot analysis, with the antibody detecting aspecific band with an apparent molecular weight of approxi-mately 25 kDa (Swanson et al., 1996).

Analysis of tissue sections

In this analysis, two to six eyes from different animalswere examined at each time point. To determine theaverage number of labeled cells, three sections from eachretina were examined, and the data from all the retinas ofa particular age were pooled. Within each retina, onesection was at a position hemisecting the eye along thenasal-to-temporal axis, and the other two sections wereapproximately 240 µm on each side. For the ChAT/SNAP-25 colocalization analysis, all ChAT-IR cell bodieswere counted. SNAP-25-IR cell bodies in the GCL andinner one-half of the INL were counted as highly immuno-reactive if the intensity of the label outlining the soma wascomparable to that observed in sublamina 2 or 4 of the IPLand thus significantly brighter than the lower levels ofimmunoreactivity in the surrounding tissue. For the cholin-ergic amacrine cell birth-dating analysis, darkly labeledBrdU-IR, ChAT-IR, and double-labeled (BrdU/ChAT-IR)cells were counted in each section within the INL and GCL.

Tissue sections were examined with a Nikon MicrophotFXA photomicroscope equipped with epifluorescence andphotographed with Kodak Ektachrome 400 Elite colorslide film or Fujicolor 400 color print film. Images werealso captured by using a Kodak Megaplus Camera (Model1.4) connected to a Perceptics MegaGrabber framegrabberin a Macintosh 8100/80AV computer (Apple Computer,Cupertino, CA) with NIH Image 1.55VDM software (WayneRasband, National Institutes of Health, Bethesda, MD;obtained at FTP site zippy.nimh.nih.gov).

Density plot profiles were prepared by using the NIHImage software. For this analysis, a profile line plot wasobtained from digitized images of immunostained tissuesections in Figures 3 and 6. Each profile line plot corre-sponded to a thin strip of 37I917 pixels (5I124.5 µm; inFig. 4) or 35I596 pixels (5I86 µm; in Fig. 7) marked byarrows in Figure 3 and arrowheads in Figure 6. Thismethod of analysis provides information about relativeintensities of staining. Although this approach does notprovide evidence for differences in absolute levels of expres-sion of the proteins of interest, this strategy allows acomparison of the relative levels of expression within thedifferent layers and sublamina of the retina (Catsicas etal., 1992). Figures were prepared by using Adobe Photo-shop Version 3.0 and Macromedia Freehand Version 5.5 forthe Macintosh. Outputs were generated on a Tektronixphaser continuous tone color printer.

RESULTS

In the retina of the Brazilian opossum, an antibodydirected against ChAT produces a distinct pattern ofChAT-IR that displays two subpopulations of cells, aconventional population in the INL and a displaced popula-tion in the GCL, with processes principally in sublaminae2 and 4 of the IPL, respectively (Fig. 1A). Inspection offlat-mounted retinas revealed that these two subpopula-tions of cholinergic amacrine cells are distributed in arelatively uniform pattern across the retina (Fig. 1B).These patterns of ChAT-IR are similar to the distributionof cholinergic starburst amacrine cells that have been

376 M.H. WEST GREENLEE ET AL.

described in all mammalian retinas examined thus far(Marc, 1986; Pourcho and Osman, 1986; Voigt, 1986;Hutchins and Hollyfield, 1987a; Reese et al., 1994; Felleret al., 1996).

Neurogenesis of cholinergic amacrine cellsoccurs during postnatal development

To characterize neurogenesis of the cholinergic ama-crine cells, we have carried out a BrdU birth-datinganalysis. The birth dates of the cholinergic amacrine cellswere determined by double-label BrdU/ChAT immunohis-tochemistry. Choline acetyltransferase-IR that was ob-served in cells that also exhibited BrdU-IR were likely tohave undergone their terminal division on the day of theBrdU injection. Although neurogenesis in the Monodelphisretina begins 1–2 days prior to birth (Finley, Jensen, Iqbal,Jacobson, and Sakaguchi, unpublished observations), nodouble-labeled cells were observed until 3PN, as illus-trated in Figure 2. The genesis of ChAT-IR neuronsreached a peak at around 5PN (Fig. 2), after whichproduction of ChAT-IR cells decreased until no double-labeled cells were found following BrdU injections on day15, 20, or 25 (Fig. 2). These results indicate that thecholinergic amacrine cells in the retina of the Brazilianopossum are born postnatally, between the ages of 2PNand 15PN. Furthermore, we observe an approximate peakin the genesis of these neurons at around 5PN.

Fig. 2. Cholinergic amacrine cells are born between postnatal days(PN) 2 and 15. The vertical axis represents the average number ofbromodeoxyuridine (BrdU) / choline acetyltransferase (ChAT) double-immunoreactive cells per section. The horizontal axis represents daysof postnatal development. Bromodeoxyuridine injections were carriedout on postnatal days 1, 2, 3, 4, 5, 10, 12, 15, 20, and 25. The firstChAT-IR amacrine cells were born around 3PN, with an approximatepeak at around 5PN. The number of ChAT-IR cells born on later daysdecreased until 15PN, when no double-labeled cells were observed.Error bars represent S.E.M. Sample sizes for each age are as follows:1PN (n 5 9 retinal sections from 3 eyes, 9/3), 2PN (n 5 9/3), 3PN(n 5 6/2), 4PN (n 5 6/2), 5PN (n 5 9/3), 10PN (n 5 6/2), 12PN (n 5 9/3),15PN (n 5 6/2), 20PN (n 5 9/3), 25PN (n 5 9/3).

Fig. 1. Choline acetyltransferase-like immunoreactivity (ChAT-IR)in an adult retina. A: An antibody directed against ChAT identifies twosubclasses of cholinergic amacrine cells: A conventional population inthe inner nuclear layer (INL) and a displaced population in theganglion cell layer (GCL), with processes in sublaminae 2 and 4 of the

inner plexiform layer (IPL), respectively. B: A flat-mounted retinaprocessed for anti-ChAT immunohistochemistry revealed that thecholinergic amacrine cells were distributed in a relatively uniformpattern across the retina. Numbers in A represent sublamina of theIPL. Scale bars 5 20 µm.

SNAP-25 IN CHOLINERGIC AMACRINE CELLS 377

Fig. 3. Choline acetyltransferase immunoreactivity (ChAT-IR) andhigh levels of synaptosomal-associated protein of 25 kDa immunoreactivity(SNAP-25-IR) colocalize to the same cells. A–C: Fluorescence photomicro-graphs from a 25PN retina double labeled with antibodies against ChAT(A) and SNAP-25 (B). A fluorescence double-exposure photomicrographreveals colocalization of the two immunoreactivities in cell bodies in the

inner nuclear layer (INL) and ganglion cell layer (GCL) and their processesin sublaminae 2 and 4 of the inner plexiform layer (IPL; C). D: Differentialinterference contrast image of the retinal section in A–C. The arrows in Aand B mark the region of retina that was analyzed by using density plotprofiles, illustrated in Figure 4. RPE, retinal pigment epithelium; OPL,outer plexiform layer. Scale bar 5 50 µm.

378 M.H. WEST GREENLEE ET AL.

Cholinergic amacrine cells exhibit high levelsof SNAP-25 prior to eye opening

To gain a better understanding of retinal developmentand differentiation, we have used immunohistochemistryto characterize the spatial and temporal distribution ofpresynaptic terminal-associated proteins during postnataldevelopment (West Greenlee et al., 1996). An antiserumdirected against SNAP-25 produced an intense labelingpattern remarkably similar to the distribution of theChAT-IR amacrine cells in the 25PN retina. To investigatethe possibility that these ChAT-IR amacrine cells werespecifically and selectively expressing relatively high lev-els of SNAP-25, we have carried out a double-label immu-nohistochemical analysis. Figure 3A illustrates the pat-tern of ChAT-IR observed in a 25PN Monodelphis retina.Immunohistochemical analysis with the anti-SNAP-25antiserum revealed relatively high levels of immunoreac-tivity outlining a subset of somata within the INL andGCL (Fig. 3B). The annular appearance of immunoreactiv-ity provided evidence that the protein was associated withthe somal membrane. Within the IPL, the immunoreactiv-ity for SNAP-25 was very distinct, exhibiting two discretebands of intense immunoreactivity in sublaminae 2 and 4.In contrast, relatively low levels of immunoreactivity werepresent in IPL sublaminae 1, 3, and 5 (Fig. 3B). Double-labeling immunofluorescence revealed a selective localiza-tion of relatively high levels of SNAP-25-IR in the ChAT-IRsomata (Fig. 3A,B). Choline acetyltransferase immunore-activities and SNAP-25-IRs were selectively colocalized incell bodies and within the processes ramifying withinsublaminae 2 and 4 of the IPL (Fig. 3C). Densitometricanalysis clearly illustrates the relative intensities ofChAT-IR and SNAP-25-IR within a 25PN retina (Fig.4A,B). Although SNAP-25-IR was also observed in theouter one-half of the retina, the relative levels of immuno-reactivity were considerably less than those observedalong the inner region of the retina (Figs. 3B, 4B). Withinthe IPL, the two peaks of highest relative intensitiescorrespond to sublaminae 2 and 4 (dashed lines in Fig. 4),and the peaks in the GCL correspond to the soma of thedisplaced amacrine cell (dotted lines in Fig. 4) illustratedin Figure 3A,B (arrow). This comparative densitometricanalysis from the same location of retina (arrows in Fig. 3)further supports and illustrates the colocalization of thetwo immunoreactivities.

To characterize further the relationship of the distribu-tion of ChAT-IRs and SNAP-25-IRs, we examined theircolocalization along the dorsal-to-ventral and nasal-to-temporal axes of the retina. Figure 5 illustrates ChAT-IRsand SNAP-25-IRs in representative images from a sectionof a double-labeled 25PN retina progressing from a moredorsal (Fig. 5A,B) to a more ventral (Fig. 5K,L) retinalposition. The colocalization of the two immunoreactivitieswas clearly evident throughout the dorsal-to-ventral ex-tent of the 25PN retina (Fig. 5). In a similar fashion,analysis of sections from more nasal and temporal retinalpositions also revealed colocalization of ChAT-IRs andSNAP-25-IRs (data not shown).

Transient colocalization of SNAP-25in ChAT-IR amacrine cells

To examine the time course and specificity of theSNAP-25 expression in cholinergic amacrine cells, weperformed a double-labeling immunohistochemical analy-

sis on neonatal and adult opossum retinal tissue. Cholineacetyltransferase immunoreactivity was first observed inthe 15PN Monodelphis retina (Fig. 6A). At this age,ChAT-IR was diffuse throughout the inner portion of theretina, with a few brightly labeled cell bodies in thedeveloping INL and GCL (Fig. 6A). Although SNAP-25-IRwas detected as early as 5PN, the immunoreactivity atthat age was very diffuse (West Greenlee et al., 1996).However, at 15PN, SNAP-25-IR resembled the pattern ofreactivity observed with the anti-ChAT antibody (Fig. 6B).The colocalization of their immunoreactivities was evidentin somata located along the inner one-third of the retina.In addition, diffuse immunoreactivity was observedthroughout the nascent IPL with both antibodies (Fig.6A,B). At 20PN, ChAT-IRs and SNAP-25-IRs within theINL were more distinct, and ChAT-IR processes began tosegregate to sublaminae 2 and 4 of the developing IPL(data not shown). At this age, the average number ofsomata immunoreactive for SNAP-25 and ChAT weresimilar, and the vast majority (90%) of SNAP-25-IR so-mata also exhibited ChAT-IR (Fig. 7), revealing theirspecific colocalization at this period of development.

The lamination pattern of the 25PN retina was moredistinct than at earlier ages. Choline acetyltransferase-immunoreactive and SNAP-25-IR cell bodies were re-stricted to the INL and GCL, and there was a furtherrestriction of their processes to sublaminae 2 and 4 of theIPL (Fig. 6C,D). At this age, there was a concomitantincrease in the average numbers of ChAT-IR and SNAP-25-IR cell bodies per section, with approximately 95% ofSNAP-25-IR somata co-immunoreactive with the ChATantibody (Fig. 7). At 30PN, the retina displayed a markedincrease in the average number of ChAT-IR cells persection (Fig. 7). Although the average number of SNAP-25-IR somata decreased, almost all (95.4%) were doublelabeled with the ChAT antibody (Fig. 7). In the 35PNretina, the average number of ChAT-IR somata continuedto increase (Fig. 7), and the immunoreactivity observed insublaminae 2 and 4 of the IPL was even more distinct

Fig. 4. Density plot profiles of choline acetyltransferase immunore-activity (ChAT-IR; A) and synaptosomal-associated protein of 25 kDaimmunoreactivity (SNAP-25-IR; B) taken from the images in Figure3A,B at the locations marked by the arrows. The density plot profilespermit a semiquantitative evaluation of immunoreactivity within aretinal section (Catsicas et al., 1992). ChAT-IR was colocalized withhigh levels of SNAP-25-IR in the inner plexiform layer (IPL; dashedlines) and in a cell body in the ganglion cell layer (GCL; dotted lines).In addition, lower levels of SNAP-25-IR were present in the outer halfof the retina. Baseline levels were comparable to background fluores-cence observed in unlabeled and secondary antibody controls.

SNAP-25 IN CHOLINERGIC AMACRINE CELLS 379

(Fig. 6E). In contrast, the average number of stronglylabeled SNAP-25-IR somata decreased (Fig. 7), althoughthe intense immunoreactivity was still evident in sublami-nae 2 and 4 of the IPL (Fig. 6F). Despite the decrease in the

average number of SNAP-25-IR somata, 91% were never-theless ChAT-IR (Fig. 7).

In the adult retina, although the average number ofChAT-IR cells per section decreased (Fig. 7), ChAT-IR

Fig. 5. Choline acetyltransferase immunoreactivity (ChAT-IR) andsynaptosomal-associated protein of 25 kDa immunoreactivity (SNAP-25-IR) are colocalized in amacrine cells throughout the retina. Imagesfrom a 25PN retinal section starting from dorsal retina (A,B) to

ventral retina (K,L). ChAT-IR (A,C,E,G,I,K) and SNAP-25-IR(B,D,F,H,J,L) are extensively colocalized throughout the dorso-ventral extent of the retina. Scale bar 5 30 µm.

380 M.H. WEST GREENLEE ET AL.

Fig. 6. Developmental distribution of choline acetyltransferaseimmunoreactivity (ChAT-IR) and synaptosomal-associated protein of25 kDa immunoreactivity (SNAP-25-IR). ChAT-IR was first detectedin the 15PN retina (A) shortly after high levels of SNAP-25-IR werefirst observed (B). In the 25PN retina, ChAT-IR (C) and SNAP-25-IR(D) were present in the same cells and processes in the inner plexiformlayer (IPL). At 35PN, the distinctive pattern of ChAT-IR (E) in cellbodies and processes was still evident, whereas high levels of SNAP-

25-IR (F) in cell bodies were greatly diminished, although the bilami-nar immunoreactivity in the IPL was still present. In the adult retina,ChAT-IR displayed its characteristic pattern (G), whereas SNAP-25-IR was now diffuse throughout the IPL and essentially absent fromcell bodies (H). Arrowheads mark the region of retina that wasanalyzed by using density plot profiles illustrated in Figure 8. Scalebar 5 20 µm.

exhibited the same basic pattern as was observed in the35PN retina (Fig. 6G). The decrease in the average num-ber of ChAT-IR cells per retinal section may be due to celldeath or to a change in the distribution of ChAT-IR cells asthe retina continued to grow. In contrast, SNAP-25-IR inthe adult retina differed significantly from the pattern ofreactivity that was observed at earlier ages (Fig. 6H). Veryfew somata were immunoreactive for SNAP-25 (Fig. 7),and, furthermore, SNAP-25-IR was diffuse throughout theIPL, lacking any obvious laminar pattern of labeling (Fig.6H). These changing patterns of immunoreactivity ob-served in the IPL with the anti-ChAT and anti-SNAP-25antibodies are illustrated with density plot profiles (Fig.8). This densitometric analysis carried out on the imagesin shown in Figure 6 (Fig. 6A–H; at the location markedwith an arrowhead) clearly illustrates the dynamic changesin the characteristic patterns of ChAT-IRs and SNAP-25-IRs in sublaminae 2 and 4 of the IPL (Fig. 8). Cholineacetyltransferase immunoreactivity within the IPL wasinitially diffuse (Figs. 6A, 8A) and gradually becamerestricted to IPL sublaminae 2 and 4 (Figs. 6C,E,G,8C,E,G). In a similar fashion, SNAP-25-IR within the IPLwas initially diffuse (Figs. 6B, 8B) and became restrictedpredominantly to sublaminae 2 and 4 at 25PN and 35PN(Figs. 6D,F, 8D,F). In contrast, in the adult, SNAP-25-IRwas expressed in a diffuse pattern throughout the IPL(Figs. 6H, 8H). These results revealed a specific andtransient colocalization of high levels of SNAP-25-IR inChAT-IR amacrine cells. This discrete colocalization oc-curs during a dynamic period of retinal developmentduring which cholinergic amacrine cells are likely to play afundamental role in guiding the establishment of appropri-ate circuitry within the visual system.

DISCUSSION

In the present study, we have examined the develop-ment of cholinergic amacrine cells in the mammalianretina. Bromodeoxyuridine birth-dating analysis revealedthat ChAT-IR amacrine cells in the Brazilian opossum areborn between 2PN and 15PN, with the greatest proportionof cells being generated at around 5PN. By using theanti-ChAT antibody and an antibody directed againstSNAP-25, we have examined the spatial and temporalrelationship of their patterns of expression. Our resultsrevealed a specific and transient expression of relativelyhigh levels of SNAP-25-IR in cholinergic amacrine cellsduring postnatal retinal development. This period in devel-opment began around 15PN, with the appearance ofChAT-IR within the same cells. In the 25PN retina, thegreatest proportion of ChAT-IR amacrine cells displayedhigh levels of SNAP-25-IR. By 35PN, the approximate ageof eye opening, most ChAT-IR amacrine cells no longerexhibited the intense SNAP-25-IR in their cell bodies,although prominent immunoreactivity was still observedin processes ramifying in sublaminae 2 and 4 of the IPL. Inthe adult retina, high levels of SNAP-25-IR were no longerobserved within ChAT-IR amacrine cells; rather, there wasdiffuse SNAP-25-IR throughout the IPL.

The intense immunoreactivity for SNAP-25 during reti-nal development was highly specific to cholinergic ama-crine cells. A small percentage (,5–10%) of the SNAP-25-IR somata lacked any obvious ChAT-IR, which may bedue to the temporal difference in expression of SNAP-25 orto our ability to detect higher levels of the protein becauselow levels of SNAP-25-IR appear to precede detectableChAT-IR during retinal development. Alternatively, theSNAP-25-IR neurons that lacked ChAT-IR may representan additional, less populous, subclass of amacrine cellsthat also exhibits this differential expression of SNAP-25-IR during retinal development. Although the relativelyhigh levels of SNAP-25-IR were observed within the innerretina, at younger ages SNAP-25-IR was also observed inthe outer half of the retina, in the region of differentiatingphotoreceptors. The significance of the lower levels ofSNAP-25 expression in the outer retina has yet to bedetermined and may indicate a functional role for SNAP-25during the differentiation of retinal cell types.

The functional requirement of SNAP-25 during transmit-ter release has been determined largely by studies usingbotulinum toxins (reviewed in Niemann et al., 1994).Other studies have defined additional roles for the proteinduring development (Osen-Sand et al., 1993, 1996; Bark etal., 1995) and possibly regeneration (Boschert et al., 1996).The spatial and temporal pattern of SNAP-25-IR in cholin-ergic amacrine cells, whose role during retinal develop-ment is presently under intense investigation (Feller et al.,1996; Sernagor and Grzywacz, 1996; Zhou and Fain, 1996),may afford a unique opportunity to characterize furtherthe function of SNAP-25 in vivo.

Long before the onset of visual function, spontaneousactivity is present in the developing vertebrate retina(Galli and Maffei, 1988; Meister et al., 1991; Wong et al.,1993; Feller et al., 1996). Blocking spontaneous activityprevents the normal segregation of ganglion cell axons intotheir appropriate eye-specific layers within the lateralgeniculate nucleus (Shatz and Stryker, 1988). This corre-lated bursting activity is developmentally regulated andsubsides just prior to eye opening (Wong et al., 1993).

Fig. 7. Transient localization of high levels of synaptosomal-associated protein of 25 kDa immunoreactivity (SNAP-25-IR) withincholine acetyltransferase immunoreactive (ChAT-IR) amacrine cellsduring development. The graph represents the average number ofChAT-IR cell bodies and the average number of highly SNAP-25-IRcell bodies per section of retina from each age examined (see Materialsand Methods). The horizontal axis represents the ages examined, andthe vertical axis represents the average number of highly SNAP-25-IRor ChAT-IR cell bodies per section. Arrowheads pointing to darkhorizontal lines represent the average number of double-labeled cellsper section. Error bars represent S.E.M. Sample sizes for each age areas follows: 20PN (n 5 12 sections from 4 eyes, 12/4), 25PN (n 5 17/6),30PN (n 5 15/5), 35PN (n 5 9/3), adult (n 5 6/2).

382 M.H. WEST GREENLEE ET AL.

Retinal activity prior to visual stimulation is importantnot only for refinement of the visual projection but also forthe establishment of appropriate circuitry within theretina itself. In the developing turtle retina, blockingcholinergic-dependent spontaneous retinal activity re-sulted in abnormally large retinal ganglion cell receptivefields (Sernagor and Grzywacz, 1996). The propagationand modulation of spontaneous retinal activity may befacilitated by several unique characteristics of cholinergicamacrine cells. Their relatively large arborizations mayfacilitate communication by a single amacrine cell withmany ganglion cells and other amacrine cells, via bothchemical and electrical synapses (Penn et al., 1994). Inaddition to acetylcholine, starburst amacrine cells releasegamma-aminobutyric acid, an inhibitory neurotransmit-ter (O’Malley et al., 1992). The corelease of an excitatoryand inhibitory neurotransmitter may contribute to thedifferential bursting activity that develops between on andoff ganglion cells as the retina continues to mature (Wongand Oakley, 1996). Furthermore, it is likely that themembrane excitability of the cholinergic amacrine cellschanges during development. Zhou and Fain (1996) showedin the rabbit retina that displaced starburst amacrine cellsundergo a dramatic transition from spiking to nonspikingjust after eye opening. Together, these results suggest thatstarburst amacrine cells may play an important role in thedevelopment of synaptic circuitry in the mammalian vi-sual system.

The cholinergic-dependent spontaneous activity ob-served during the early development of the visual system

may in part be regulated by differential expression ofproteins that constitute the synaptic machinery necessaryfor regulated neurotransmitter release. One protein inparticular, SNAP-25, appears to have two divergent rolesin vesicle fusion events (Bark et al., 1995). SNAP-25 iscrucial during regulated exocytosis of neurotransmitter(Bark and Wilson, 1994; Sudhof, 1995; Calakos andScheller, 1996; Mehta et al., 1996) and may likely beinvolved in other vesicle fusion events during neuriteextension (Osen-Sand et al., 1993, 1996; Catsicas et al.,1994). The transient and specific localization of high levelsof SNAP-25 within ChAT-IR amacrine cells suggests thatit may be involved in regulating vesicular fusion eventsduring the dynamic period of retinal development prior toeye opening. The relatively high levels of SNAP-25 mayrepresent a selective enhancement of synaptic machineryin these neurons. Increased abundance of SNAP-25 mayserve to facilitate the docking of a large number of synapticvesicles with the presynaptic terminal membrane, therebyincreasing the likelihood of transmitter release in re-sponse to an excitatory event. The role of SNAP-25 duringdocking of a synaptic vesicle involves its intimate associa-tion with other SNARE proteins (Bark and Wilson, 1994).In the Monodelphis retina, immunoreactivity for theSNARE protein Syntaxin is present in the developing IPLas early as 10PN and persists throughout development(Warme, West Greenlee, and Sakaguchi, unpublished ob-servations). Furthermore, the synaptic terminal-associ-ated proteins Synaptotagmin, Rab3A, and Synaptophysinare present throughout the developing IPL from 15PN into

Fig. 8. Densitometric analysis illustrating the changing levels ofcholine acetyltransferase immunoreactivity (ChAT-IR; A,C,E,G) andsynaptosomal-associated protein of 25 kDa immunoreactivity (SNAP-25-IR; B,D,F,H) within the inner plexiform layer (IPL) of the develop-ing Monodelphis retina. These density plot profiles were taken fromthe images in Figure 6, at the location marked with arrowheads.ChAT-IR became more restricted to discrete sublamina within the IPL.In contrast, SNAP-25-IR within the IPL was initially diffuse (15PN)

and became more restricted to sublaminae 2 and 4 at 25PN and 35PN.In the adult retina, SNAP-25 once again displayed a diffuse pattern ofimmunoreactivity. The solid line indicates the approximate region ofthe IPL at each age. Baseline levels were comparable to backgroundfluorescence observed in unlabeled and secondary antibody controls.In each panel, plots from left to right represent outer to inner retina.Scale bars 5 5 µm.

SNAP-25 IN CHOLINERGIC AMACRINE CELLS 383

adulthood (West Greenlee et al., 1996). Thus, the synapticmachinery seems to be present in the IPL such that highlevels of SNAP-25 in ChAT-IR cells may be acting duringrelease of neurotransmitter.

Alternatively, the high levels of SNAP-25 expression inChAT-IR amacrine cells may also be involved in mediatingdistinct vesicle-membrane fusion events during neuriteextension. Synaptosomal associated protein-25 proteinhas been shown to be necessary for neurite outgrowthduring development (Osen-Sand et al., 1993, 1996). Thesestudies implicate a role for SNAP-25 in the docking andfusion of plasmallemma precursor vesicles at the growingneurite tip (Osen-Sand et al., 1996). The distinction be-tween two divergent vesicle fusion events leading toneurotransmission and neurite extension may be medi-ated by different isoforms of SNAP-25 (i.e., SNAP-25a andSNAP-25b; Bark et al., 1995). These isoforms appear to bedevelopmentally and spatially regulated. SNAP-25a ispresent predominantly during development and regenera-tion (Bark et al., 1995; Boschert et al., 1996), whereasSNAP-25b expression appears to coincide with synapticmaturation (Bark et al., 1995). Due to the differentiallocalization and presumed functions of the two SNAP-25isoforms, studies to identify which isoform(s) constitutesthe high levels in ChAT-IR amacrine cells may ultimatelylead to a better understanding of their functional signifi-cance. Furthermore, high levels of SNAP-25 may be in-volved during the establishment of synaptic circuitry inChAT-IR amacrine cells. Studies by Osen-Sand et al.(1996) have demonstrated that cleavage of SNAP-25 withbotulinum toxin inhibit synapse formation in developingcentral neurons. Thus, the period of high SNAP-25 expres-sion in ChAT-IR neurons may represent a period of exten-sive synaptogenesis and/or synaptic plasticity for thesecells.

In addition to the Monodelphis retina, we have recentlyfound that relatively high levels of SNAP-25-IR are specifi-cally localized to ChAT-IR cells in the postnatal rat retinaprior to eye opening (6PN; West Greenlee and Sakaguchi,unpublished observations). These results provide evidencethat high levels of SNAP-25 in ChAT-IR amacrine cells arelikely to be a ubiquitous phenomenon required for choliner-gic-dependent processes during mammalian retinal devel-opment.

It is likely that cholinergic amacrine cells mediateevents crucial for the establishment of proper retinalcircuitry and therefore would influence further develop-ment of the mammalian visual system. The specificity andtime course of the expression of high levels of SNAP-25-IRin cholinergic amacrine cells provides evidence that thispresynaptic terminal-associated protein may play an inte-gral role in mediating the transient function provided bythese amacrine cells during this critical period of develop-ment. Further studies should reveal whether additionalaspects of membrane trafficking and vesicle fusion, facili-tated by SNAP-25, underlie the specific needs for thedevelopmental plasticity of the nervous system.

ACKNOWLEDGMENTS

We thank Teresa Gray for assistance during initialstudies characterizing development of ChAT-IR. We alsothank Drs. Mary Helen Greer and Larry Jackson forproviding equipment and assistance with the anesthesiaprocedures. S.K.F. was funded in part by a summer

internship from the Howard Hughes Medical InstituteEducation Initiative. This manuscript, designated by IowaState University as J-17514 of the Iowa Agriculture andHome Economics Experiment Station, Ames, Iowa, projectnumber 3205, was supported by Hatch Act and State ofIowa Funds.

LITERATURE CITED

Alexiades, M.R. and C.L. Cepko (1997) Subsets of retinal progenitorsdisplay temporally regulated and distinct biases in the fates of theirprogeny. Development 124:1119–1131.

Ariel, M. and N.W. Daw (1982) Effects of cholinergic drugs on receptive fieldproperties of rabbit retinal ganglion cells. J. Physiol. (Lond.) 324:135–160.

Bark, I.C. and M.C. Wilson (1994) Regulated vesicular fusion in neurons:Snapping together the details. Proc. Natl. Acad. Sci. USA 91:4621–4624.

Bark, I.C., K.M. Hahn, A.E. Ryabinin, and M.C. Wilson (1995) Differentialexpression of SNAP-25 protein isoforms during divergent vesicle fusionevents of neural development. Proc. Natl. Acad. Sci. USA 92:1510–1514.

Boschert, U., C. O’Shaughnessy, R. Dickinson, M. Tessari, C. Bendotti, S.Catsicas, and E.M. Pich (1996) Developmental and plasticity-relateddifferential expression of two SNAP-25 isoforms in the rat brain. J.Comp. Neurol. 367:177–193.

Calakos, N. and R.H. Scheller (1996) Synaptic vesicle biogenesis, docking,and fusion: A molecular description. Physiol. Rev. 76:1–29.

Catsicas, S., M. Catsicas, K.T. Keyser, H.J. Karten, M.C. Wilson, and R.J.Milner (1992) Differential expression of the presynaptic protein SNAP-25in mammalian retina. J. Neurosci. Res. 33:1–9.

Catsicas, S., G. Grenningloh, and E.M. Pich (1994) Nerve–terminal pro-teins: To fuse to learn. Trends Neurosci. 17:368–373.

Dann, J.F. (1989) Cholinergic amacrine cells in the developing cat retina. J.Comp. Neurol. 289:143–155.

Famiglietti, E.V. (1983) On and off pathways through amacrine cells inmammalian retina: The synaptic connections of starburst amacrinecells. Vision Res. 23:1265–1279.

Feller, M.B., D.P. Wellis, D. Stellwagen, F.S. Werblin, and C.J. Shatz (1996)Requirement for cholinergic synaptic transmission in the propagationof spontaneous retinal waves. Science 272:1182–1187.

Galli, L. and L. Maffei (1988) Spontaneous impulse activity of rat retinalganglion cells in prenatal life. Science 242:90–91.

Goodman, C.S. and C.J. Shatz (1993) Developmental mechanisms thatgenerate precise patterns of neuronal connectivity. Cell Suppl. 72:77–98.

Hutchins, J.B. and J.G. Hollyfield (1987a) Acetylcholinesterase in thehuman retina. Brain Res. 400:300–311.

Hutchins, J.B. and J.G. Hollyfield (1987b) Cholinergic neurons in thehuman retina. Exp. Eye Res. 44:363–375.

Iqbal, J., J.K. Elmquist, L.R. Ross, M.R. Ackermann, and C.D. Jacobson(1995) Postnatal neurogenesis of the hypothalamic paraventricular andsupraoptic nuclei in the Brazilian opossum brain. Brain Res. Dev. BrainRes. 85:151–160.

Karten, H.J. and N. Brecha (1983) Localization of neuroactive substancesin the vertebrate retina: Evidence for lamination in the inner plexiformlayer. Vision Res. 23:1197–1205.

Kuehl-Kovarik, M.C., D.S. Sakaguchi, J. Iqbal, I. Sonea, and C.D. Jacobson(1995) The gray short-tailed opossum: A novel model for mammaliandevelopment. Lab. Animal 24:24–29.

Marc, R.E. (1986) Neurochemical stratification in the inner plexiform layerof the vertebrate retina. Vision Res. 26:223–238.

Masland, R.H., J.W. Mills, and C. Cassidy (1984a) The functions ofacetylcholine in the rabbit retina. Proc. R. Soc. Lond. B Biol. Sci.223:121–139.

Masland, R.H., J.W. Mills, and S.A. Hayden (1984b) Acetylcholine-synthesizing amacrine cells: Identification and selective staining byusing radioautography and fluorescent markers. Proc. R. Soc. Lond. BBiol. Sci. 223:79–100.

Mehta, P.P., E. Battenberg, and M.C. Wilson (1996) SNAP-25 and synapto-tagmin involvement in the final Ca21 dependent triggering of neurotrans-mitter release. Proc. Natl. Acad. Sci. USA 93:10471–10476.

Meister, M., R.O.L. Wong, D.A. Baylor, and C.J. Shatz (1991) Synchronousbursts of action potentials in ganglion cells of the developing mamma-lian retina. Science 252:939–943.

384 M.H. WEST GREENLEE ET AL.

Niemann, H., J. Blasi, and R. Jahn (1994) Clostridial neurotoxins: newtools for dissecting exocytosis. Trends Cell Biol. 4:179–185.

O’Malley, D.M., J.H. Sandell, and R.H. Masland (1992) Co-release ofacetylcholine and GABA by the starburst amacrine cells. J. Neurosci.12:1394–1408.

Osen-Sand, A., M. Catsicas, J.K. Staple, K.A. Jones, G. Ayala, J. Knowles,G. Grenningloh, and S. Catsicas (1993) Inhibition of axonal growth bySNAP-25 antisense oligonucleotides in vitro and in vivo [see com-ments]. Nature 364:445–448.

Osen-Sand, A., J.K. Staple, E. Naldi, G. Schiavo, O. Rossetto, S. Petitpierre,A. Malgaroli, C. Montecucco, and S. Catsicas (1996) Common anddistinct fusion proteins in axonal growth and transmitter release. J.Comp. Neurol. 367:222–234.

Oyler, G.A., G.A. Higgins, R.A. Hart, E. Battenberg, M. Billingsley, F.E.Bloom, and M.C. Wilson (1989) The identification of a novel synapto-somal-associated protein, SNAP-25, differentially expressed by neuro-nal subpopulations. J. Cell Biol. 109:3039–3052.

Penn, A.A., R.O. Wong, and C.J. Shatz (1994) Neuronal coupling in thedeveloping mammalian retina. J. Neurosci. 14:3805–3815.

Pourcho, R.G. and K. Osman (1986) Cytochemical identification of choliner-gic amacrine cells in cat retina. J. Comp. Neurol. 247:497–504.

Reese, B.E., W.F. Thompson, and J.D. Peduzzi (1994) Birthdates of neuronsin the retinal ganglion cell layer of the ferret. J. Comp. Neurol.341:464–475.

Sernagor, E. and N.M. Grzywacz (1996) Influence of spontaneous activityand visual experience on developing retinal receptive fields. Curr. Biol.6:1503–1508.

Shatz, C.J. (1996) Emergence of order in visual system development. Proc.Natl. Acad. Sci. USA 93:602–608.

Shatz, C.J. and M.P. Stryker (1988) Prenatal tetrodotoxin infusion blockssegregation of retinogeniculate afferents. Science 242:87–89.

Sudhof, T.C. (1995) The synaptic vesicle cycle: A cascade of protein–proteininteractions. Nature 375:645–653.

Swanson, J.J., M.C. Kuehl-Kovarik, M.C. Wilson, J.K. Elmquist, and C.D.Jacobson (1996) Characterization and ontogeny of synapse-associatedproteins in the developing facial and hypoglossal motor nuclei of thebrazilian opossum. J. Comp. Neurol. 368:270–284.

Tucek, S. (1976) Supply of acetylcoenzyme A and choline acetyltransferasefor acetylcholine synthesis in cholinergic nerve endings. Act. Nerv.Super. (Praha) 18:109–110.

Voigt, T. (1986) Cholinergic amacrine cells in the rat retina. J. Comp.Neurol. 248:19–35.

Wassle, H., M.H. Chun, and F. Muller (1987) Amacrine cells in the ganglioncell layer of the cat retina. J. Comp. Neurol. 265:391–408.

West Greenlee, M.H., J.J. Swanson, J.J. Simon, J.K. Elmquist, C.D.Jacobson, and D.S. Sakaguchi (1996) Postnatal development and thedifferential expression of presynaptic terminal-associated proteins inthe developing retina of the Brazilian opossum, Monodelphis domestica.Dev. Brain. Res. 96:159–172.

Wong, R.O. and D.M. Oakley (1996) Changing patterns of spontaneousbursting activity of on and off retinal ganglion cells during develop-ment. Neuron 16:1087–1095.

Wong, R.O., M. Meister, and C.J. Shatz (1993) Transient period of corre-lated bursting activity during development of the mammalian retina.Neuron 11:923–938.

Zhou, Z.J. and G.L. Fain (1996) Starburst amacrine cells change fromspiking to nonspiking neurons during retinal development. Proc. Natl.Acad. Sci. USA 93:8057–8062.

SNAP-25 IN CHOLINERGIC AMACRINE CELLS 385