Dendritic Field Development of RetinalGanglion Cells in the Cat FollowingNeonatal Damage to Visual Cortex:

Evidence for Cell ClassSpecific Interactions

A.J. WEBER,1* R.E. KALIL,2,4 AND L.R. STANFORD3,4

1Department of Anatomy, Michigan State University, East Lansing, Michigan 488242Department of Ophthalmology and Visual Sciences, University of Wisconsin,

Madison, Wisconsin 537063Department of Comparative Biosciences and the Waisman Center on Mental Retardation,

University of Wisconsin, Madison, Wisconsin 537064Center for Neuroscience, University of Wisconsin, Madison, Wisconsin 53706

ABSTRACTA well-known feature of the mammalian retina is the inverse relation that exists in

central and peripheral retina between the density of retinal ganglion cells and their dendriticfield sizes. Functionally, this inverse relation is thought to represent a means by which retinalcoverage is maintained, despite significant changes in ganglion cell density. While it isgenerally agreed that the dendritic fields of mature retinal ganglion cells reflect, in part,competitive interactions that occur during development, the issue of whether these interac-tions are cell class specific remains unclear. In order to examine this question, we usedintracellular staining techniques and an in vitro, living retina preparation to compare thesoma and dendritic field sizes of alpha and beta ganglion cells from normal retinae with thoseof cells located in matched areas of retinae in which the density of beta ganglion cells had beenreduced selectively by neonatal removal of visual cortex areas 17, 18, and 19. Our intracellulardata show that while an early, selective, reduction in beta cell density has little or no effect onthe cell body and dendritic field sizes of mature alpha cells, it results in a 13% increase in themean soma area and an 83% increase in the mean dendritic field area of surviving beta cells.This differential effect suggests that the soma and dendritic field sizes of alpha and beta ganglioncells in the mature cat retina result primarily from competitive interactions during developmentthat are cell class specific. J. Comp. Neurol. 390:470–480, 1998. r 1998 Wiley-Liss, Inc.

Indexing terms: in vitro; intracellular; Lucifer Yellow; alpha; beta

Much of our current understanding of the vertebratevisual system is derived from studies of the central visualpathway in the cat. Since this pathway is highly organizedand the structure, function, and connectivity of its constitu-ent neurons are well understood, the feline visual systemoften has served as a model for studying neural develop-ment and plasticity (see Lennie, 1980; Sherman, 1985;Sherman and Spear, 1982 for reviews). In particular,studies of the developing retina have shown that retinalganglion cells, which represent the final stage of visualprocessing within the eye, are highly immature at birth,and that the somata and dendritic arbors of these neuronsundergo extensive growth and remodeling postnatally(Mastronarde et al., 1984; Maslim et al., 1986; Dann et al.,

1988; Ramoa et al., 1988). The result of these developmen-tal changes, which occur along a central-peripheral gradi-ent, is a mature retina in which it is possible to identify atleast three classes of retinal ganglion cells, based ondifferences in soma size and dendritic field architecture(alpha, beta, and gamma; Boycott and Wassle, 1974). Ofthese three classes of retinal neurons, the alpha and beta

Grant sponsor: NIH; Grant numbers: EY01331 and EY04977.*Correspondence to: Dr. Arthur J. Weber, Department of Anatomy, B-512

West Fee Hall, Michigan State University, East Lansing, MI 48824.E-mail: weberar@com.msu.edu

Received 29 May 1997; Revised 25 August 1997; Accepted 25 August 1997

THE JOURNAL OF COMPARATIVE NEUROLOGY 390:470–480 (1998)

r 1998 WILEY-LISS, INC.

cells have been described most completely. In brief, alphacells represent about 5% of the ganglion cells in the catretina. They have the largest somata, and they containlarge, radially-oriented, dendritic arbors that commonlyoriginate from five to six primary dendrites. These neuronshave been shown to be the anatomical correlates of retinalY-cells (Cleland et al., 1975; Peichl and Wassle, 1981;Saito, 1983; Fukuda et al., 1984; Stanford and Sherman,1984). Beta ganglion cells comprise about 55% of theganglion cells in the cat retina. They have medium-sizedsomata, and their small to medium-sized dendritic treesoften originate from a single primary dendrite that thengives rise to a compact, bushy dendritic arbor. Theseneurons have been shown by intracellular injection ofphysiologically identified retinal ganglion cells to be theanatomical correlates of retinal X-cells (Saito, 1983; Fukudaet al., 1984; Stanford and Sherman, 1984; Stanford, 1987).

Studies that have compared the morphologies of alphaand beta cells in different regions of the cat retina haveshown that the somata and dendritic fields of both celltypes increase in size with distance from the area centralis(Boycott and Wassle, 1974; Stanford, 1987; Dann et al.,1988). Since retinal ganglion cell density in general de-creases as a function of retinal eccentricity, it is commonlyagreed that the larger dendritic fields of alpha and betacells in more peripheral retina reflect a compensatorymechanism whereby each class of ganglion cell maintainscomplete retinal coverage (see Peichl, 1991). While factorsintrinsic to retinal ganglion cells have been thought to playa role in the development of retinal ganglion cell dendriticstructure (Montague and Friedlander, 1991; Caserta et al.,1995), there is considerable evidence that competitiveinteractions among the dendritic processes of neighboringganglion cells also are involved. Studies in the cat (Eysel etal., 1985; Leventhal et al., 1988; Ault et al., 1993; Deplanoet al., 1994), rat (Linden and Perry, 1982; Perry andLinden, 1982; Perry and Maffei, 1988; Linden, 1993), andmonkey (Leventhal et al., 1989) have demonstrated thatneighboring ganglion cell dendrites will grow into regionsof the retina experimentally depleted of retina ganglioncells. Similarly, increasing retinal ganglion cell density byprenatal monocular eye enucleation results in ganglioncells with smaller than normal somata and dendritic fields(Kirby and Chalupa, 1986).

An important issue relevant to all of the above studies iswhether the factors influencing dendritic field develop-ment are cell class specific. For example, during develop-ment, do the dendritic processes of alpha cells compete foravailable space only with the dendrites of other alpha cells,or do they also exert an influence on neighboring betacells? Most previous studies of retinal ganglion cell devel-opment have not been able to address this question of classspecific interactions because the experimental approachesused to alter ganglion cell density (e.g., direct retinallesions, lesions of retinal targets, and eye enucleation) didnot affect a specific cell class selectively. However, neonataldamage to visual cortex results in the transynaptic, retro-grade loss of ganglion cells in the retina, and this lossappears to be restricted to beta cells (Kalil, 1984, 1990;Pearson et al., 1981; Payne et al., 1984; but see also Rowe,1990).

Ault et al. (1993) used this selective degeneration tostudy ganglion cells in the retinae of cats that receivedlarge unilateral lesions of visual cortex (areas 17, 18, and19) and the lateral suprasylvian visual area (PMLS) at

birth. By filling retinal ganglion cells retrogradely withhorseradish peroxidase (HRP), these authors reportedthat selectively reducing the density of beta cells in thefeline retina early in development results in an increase inthe soma size of alpha cells located in central retina and adecrease in the soma size of alpha cells located in periph-eral retina. In agreement with the retinal lesion studies,they also found that the dendrites of alpha cells locatedadjacent to the hemiretina with reduced beta cell densitytended to orient toward the region of lower cell density.Ault et al. (1993) interpreted these results to indicate thatthe development of retinal alpha cells was influenced byboth cell class specific interactions and those that were notcell class specific. However, because the retrograde label-ing technique often resulted in the incomplete labeling ofdistal dendritic processes, Ault et al. (1993) were unable tocompare quantitatively the dendritic field sizes of alphacells in the normal hemiretina with those in the hemiretinawith reduced numbers of beta cells. Further, their studydid not examine the morphologies of those beta cellslocated in the affected region of the retina that survivedthe early visual cortex lesion.

For some time, we too have been interested in thedevelopmental changes that occur as a result of theselective loss of beta cells in the neonatal cat retina (Kalil,1984, 1990; Kalil and Behan, 1987; Weber et al., 1989). Inthe present study, we examined the effect that a cell classspecific reduction in ganglion cell density has on the somaand dendritic field development of both alpha and betacells. In order to avoid the limitations often encountered inlabeling retinal ganglion cells and their dendritic arborsretrogradely with HRP, namely, the incomplete filling ofdendrites and the overlapping of dendritic arbors of neigh-boring cells, we labeled single alpha and beta cells innormal adult cats and in adult cats that received unilat-eral lesions of visual cortex (areas 17, 18, and 19, but notPMLS) at birth by injecting them directly with the fluores-cent dye Lucifer Yellow CH in a living, isolated retinapreparation.

In contrast with Ault et al. (1993), we find that neitherthe cell body nor dendritic field sizes of alpha cells in catswith neonatal lesions of visual cortex are significantlydifferent from normal. The cell bodies and dendritic arborsof surviving beta cells, however, are 13% and 83% largerthan normal, respectively. These data suggest that thefinal dendritic field dimensions of ganglion cells in the catretina are influenced strongly by cell class specific interac-tions during development.

MATERIALS AND METHODS

Fourteen adult cats were used in these studies. Four ofthese animals were reared normally and ten receivedunilateral lesions of visual cortex (areas 17, 18, and 19, butnot PMLS) within 36 hours of birth (referred to here asKVC cats). Using sterile techniques and methoxyfluraneanesthesia, a small bone flap was removed from the skullof each kitten, exposing the lateral gyrus near the represen-tation of the central visual field in cortical areas 17 and 18.The lateral, postlateral, posterior suprasylvian, and sple-nial cortices then were aspirated to produce extensivedamage to areas 17, 18, and 19 (Otsuka and Hassler, 1962;Tusa et al., 1978). The cavity created by the lesion thenwas packed with a sterile gelatin sponge, the bone flap wasreplaced, and the scalp incision was closed with wound

DENDRITIC COMPETITION IN THE NEONATAL CAT RETINA 471

clips. The kittens were placed on a heating pad duringrecovery from anesthesia, and then returned to theirmothers until weaned at 8 weeks of age. Intracellularinjections of retinal ganglion cells were made when theanimals were at least 1 year of age. All procedures wereapproved by the Animal Use and Care Committee of theUniversity of Wisconsin-Madison, and all conformed toNIH guidelines.

Intracellular injection procedures

Intracellular labeling of single ganglion cells in theretinae of normal and KVC cats was done in isolatedpreparations of living retinae. Each animal first wasanesthetized with an intramuscular injection of ketamineHCl (10 mg/kg) and then was anesthetized deeply by anintravenous injection of pentobarbital sodium. The eyeswere removed quickly from the animal when the depth ofanesthesia was such that a gentle touch to the cornea andeyelid no longer produced a blink reflex, and euthanasiawas induced by an overdose of pentobarbital sodium.

Immediately upon enucleation, the anterior segment ofeach eye was removed with a pair of fine scissors, thevitreous was extracted, and the eyecup was placed in asolution of artificial cerebrospinal fluid (aCSF) at pH 7.4(Saito, 1983) saturated with a mixture of 95% O2 and 5%CO2 at room temperature. Any remnant of the vitreousbody was removed using a glass pipette and gentle suction.The retina then was isolated by teasing it away from thechoroid and sclera using a pair of forceps and a fine-tippedbrush. Care was taken to avoid mechanical stress on theretina. After removal from the eyecup, the retina wasflattened by making 3–4 radial relief cuts, and it wasplaced, ganglion cell layer up, in a plexiglass chamber thatalso was perfused with oxygenated aCSF (3–6 cc/minute).The retina was held submersed in the chamber by nylonnetting. The elasticity of the nylon permitted the retina tobe held securely in place without undue pressure beingplaced on its surface. The chamber holding the retina thenwas mounted on the stage of an upright microscopeequipped with differential interference contrast and epi-fluorescence optics. A neutral density filter was used tocontrol the intensity of the light from the mercury vaporsource. Single ganglion cells were observed with a 403water immersion objective (N.A. 0.55) with a workingdistance of 1.6 mm.

Intracellular injections of single retinal ganglion cellswere made by using glass microelectrodes and a four-axishydraulic micromanipulator attached to the microscopestage. Glass micropipettes were filled with a solution of 3%Lucifer Yellow CH (Sigma Chemical Co., St. Louis, MO:L0259) in 0.1 M LiCl (pH 7.6), and beveled to a tipresistance of 80–100 Mohms. For injection, single retinalganglion cells were first positioned in the center of themicroscope field and the dye-filled electrode was loweredinto position adjacent to the cell body. Each cell then wasimpaled by slowly advancing the electrode along its axisuntil the tip of the electrode was seen to penetrate the cell’smembrane. Complete filling of the soma, dendritic tree,and from 1–2 mm of the cell’s intraretinal axon segmentwas achieved by passing a constant, 2–8 nA negativecurrent through the pipette. Typically, 2–3 minutes weresufficient to label completely the dendritic trees of alphacells, and approximately 1–2 minutes were required forbeta cells. The progress of each intracellular injection wasmonitored visually, and care was taken to minimize that

amount of time that individual ganglion cells were exposedto the fluorescent excitation. After the last cell had beeninjected, the retina was removed from the injection cham-ber and immersion-fixed for 3–4 hours in 4.0% paraformal-dehyde in 0.1 M sodium phosphate buffer (pH 7.4). Eachretina then was rinsed with 0.1 M sodium phosphatebuffer, whole-mounted on a gelatinized glass slide, dehy-drated and defatted in a series of graded alcohols andxylenes, and coverslipped.

Sampling and analysis of ganglion cellsin normal and KVC retinae

Because of the partial crossing of ganglion cell axons inthe optic chiasm of the cat, a unilateral lesion of visualcortex results in a decrease in beta ganglion cell density inthe temporal hemiretina of the eye ipsilateral to thecortical lesion and the nasal hemiretina of the eye contra-lateral to the damaged cortex. In order to ensure that all ofthe ganglion cells injected in KVC cats in our sample werelocated within the affected hemiretina of each eye andprojected to the damaged side of the brain, we only injectedneurons that were located at least 2 mm, but not morethan 5 mm, from the area centralis. Intracellular injec-tions in normal retinae were made at comparable retinaleccentricities. The location of the area centralis in eachretina was identified visually by retinal blood vessel patternsand by measuring the distance from the optic disc.

Quantitative analyses of the morphological dimensionsof ganglion cells in the retinae of normal and KVC animalswere accomplished with an Image-1 image analysis sys-tem (Universal Imaging, Corp., West Chester, PA). Indi-vidual images were captured with a high-resolution videocamera and averaged over 16 frames before being digitizedand stored for subsequent analysis. Measurements of cellsoma and dendritic field areas were made by using themorphometric software included with the Image-1 system.Cell soma area was determined by tracing the outline ofeach injected cell body, and the sizes of dendritic fieldswere determined by measuring the area that was enclosedby connecting the distal tips of each neuron’s dendriticarbor. Size distributions were compiled for injected alphaand beta cell somas and dendritic arbors in normal andKVC cats. Differences in the distributions of alpha andbeta cell soma and dendritic field sizes were analyzed withthe Kolmolgorov-Smirnov (K-S) cumulative frequency test,and differences in the mean sizes of alpha and beta cellsoma and dendritic fields were evaluated with a two-tailedt-test.

RESULTS

Two hundred thirty-six intracellularly-labeled retinalganglion cells, judged by visual inspection to be completelyfilled, were analyzed in this study; 162 of these neurons, 71alpha cells and 91 beta cells, were recovered from theretinae of four normal adult cats, while the remaining 75ganglion cells, 48 alpha cells, and 27 beta cells wereobtained from the eyes of ten adult cats that had received aunilateral lesion of visual cortex soon after birth.

The photomicrographs presented in Figure 1 show ex-amples of typical alpha and beta ganglion cells from theretinae of normal and KVC cats that were injected intracel-lularly with Lucifer Yellow CH and digitized with theimage analysis system. Alpha ganglion cells in both nor-mal (Fig. 1A) and KVC cats (Fig. 1B) display the morpho-

472 A.J. WEBER ET AL.

logical characteristics generally accepted as definitive ofthis retinal cell class (i.e., they had large somata, and long,thick, radially oriented dendrites that branched infre-quently and tapered gradually with increasing distancefrom the cell soma). Beta ganglion cells in both normal(Fig. 1C) and KVC (Fig. 1D) cats were distinguishedreadily from alpha cells by their medium-sized somata andthin dendrites that form a compact arbor.

Ganglion cell soma size in normaland KVC cats

Alpha ganglion cells. The histograms in Figure 2Aillustrate the distribution of alpha cell somata cross-

sectional areas in normal (upper histogram) and KVC(lower histogram) cats. While all of the alpha cells injectedin normal cats had soma sizes that were between 200 µm2

and 2,000 µm2, approximately 70% of the alpha cellsstudied fell within the much narrower range of 800 µm2 to1,400 µm2. The distribution of soma sizes of alpha cells inKVC cats was nearly identical to that in normal animals,and was indistinguishable from it statistically (P . 0.1,K-S test). As expected from the similarities of these cellsize distributions, the mean sizes of alpha ganglion cells innormal and KVC cats (Fig. 2B) are not different from eachother (1,168 µm2 6 302 S.D. vs. 1,145 µm2 6 414 S.D.; P .0.2, two-tailed t-test).

Fig. 1. Photomicrographs of intracellularly labeled alpha (A,B)and beta (C,D) ganglion cells from comparable regions of the temporalhemiretina of a normal adult cat (A, C) and an adult cat that had areas

17, 18, and 19 of visual cortex removed at birth (KVC cat, B,D). Noticethat neonatal removal of visual cortex does not affect the ability toclassify these neurons morphologically. Scale bars 5 20 µm.

DENDRITIC COMPETITION IN THE NEONATAL CAT RETINA 473

Fig. 2. Soma size comparisons for alpha (A,B) and beta (C,D)ganglion cells labeled intracellularly in normal and KVC cats. A and Ccompare the size distributions for each class of neurons, while themean cross-sectional area comparisons are presented in B and D. Note

that for the region of retina sampled (see text), neonatal removal ofvisual cortex has little influence on the soma sizes of either alpha orbeta ganglion cells.

474 A.J. WEBER ET AL.

Beta ganglion cells. Most of the beta cell somas innormal cats (Fig. 2C, upper) occupy the same size range asthose in KVC cats (Fig. 2C, lower), and the majority of cellsin each group of cats have cross-sectional areas that rangefrom 200 to 800 µm2. However, in proportion to the totalnumber of cells, relatively more large beta cells, 600 µm2 orgreater, were found in the retinae of KVC cats than innormals. This results in a modest skewing in the distribu-tion of beta cell soma sizes, but the size distribution is notstatistically different from that of beta cells in normal cats(P . 0.1, K-S test). Similarly, while the mean size of betacells in KVC cats is 13% greater than that of beta cells innormal cats (Fig. 2D), the mean sizes of beta ganglion cellsin normal and KVC cats are not significantly different(448µm2

6 157 S.D. vs. 505 µm2 6 240 S.D.; P . 0.1,two-tailed t-test).

Ganglion cell dendritic field sizesin normal and KVC cats

Alpha ganglion cells. A comparison of the sizes of thedendritic fields of alpha cells in normal and KVC catsreveals that they are distributed very similarly (Fig. 3A).In both groups, alpha ganglion cells can be found withdendritic fields as small as 20,000 µm2 or as large as180,000 µm2. While the mode of the distribution of alphacell dendritic field sizes in normal cats is shifted slightlytoward smaller sizes than is that of the distribution forKVC cats, the two distributions are not statistically signifi-cant (P . 0.1, K-S test), nor are the mean sizes (Fig. 3B) ofalpha cell dendritic fields in normal and KVC cats different(82,733 µm2 6 39,551 S.D. vs. 80,418 µm2 6 29,775 S.D.;P . 0.2, two-tailed t-test).

Beta ganglion cells. The size distributions of thedendritic arbors of beta ganglion cells in normal and KVCcats are shown in Figure 3C. As noted by many others, thearbors of beta cell dendrites in normal cats (upper histo-gram) are very compact in comparison with that of normalalpha cells (cf. Fig. 1A). Of the 91 beta cells injected innormal cats, nearly 85% had dendritic arbors within therange 10,000–25,000 µm2, confirming the impression ob-tained visually that one of the distinguishing features ofbeta cells as a class is that they are very similar to eachother with respect to the sizes of their dendritic arbors.The distribution of dendritic arbor sizes among beta cellsin KVC cats differs from that of beta cells in normal cats.First, the distribution of dendritic arbor sizes is broader inKVC beta cells than in normal beta cells and, second, theentire distribution is shifted toward larger arbor sizes.Indeed, more than 60% of the beta cells in KVC cats haddendritic arbors that were larger than all but 1.0% of thebeta cells in normal cats. As a result of these dissimilari-ties, the size distributions of beta cell dendritic arbors innormal and KVC cats are statistically different (P , 0.001,K-S test).

The difference in arbor size is seen most clearly bycomparing the mean sizes of the dendritic fields of betacells that were injected in normal cats with those injectedin KVC cats. The dendritic fields of KVC beta cells are, onaverage, approximately 83% larger than those of beta cellsin normal cats (Fig. 3D). This difference also is significant(29,257 µm2 6 9,173 S.D. vs. 15,954 µm2 6 5,468 S.D.; P ,0.001, two-tailed t-test).

DISCUSSION

The goal of these studies was to examine whether thecompetitive interactions thought to underlie the morpho-logical development of alpha and beta ganglion cells in thefeline retina are cell class specific or not. To accomplishthis, we compared the soma and dendritic field sizes ofintracellularly labeled alpha and beta cells in the retinaeof normal adult cats with ganglion cells from adult catsthat had received unilateral lesions of visual cortex (areas17, 18, and 19) soon after birth (KVC cats). Since neonataldamage to visual cortex produces a rapid1 and highlyselective decrease in the density of beta cells (Kalil, 1990),this experimental paradigm provides a unique means forstudying cell class specific interactions among developingretinal ganglion neurons. In brief, if alpha and beta cellscompete with each other for dendritic space during devel-opment, then placing alpha cells at a competitive advan-tage, by means of an early visual cortex lesion, shouldresult in mature alpha cells with dendritic fields that arelarger than normal. However, if final dendritic field size isregulated by competitive interactions among ganglioncells only of the same class, then beta, but not alpha, cellsin the retinae of KVC cats should exhibit changes indendritic field size. While the dendritic fields of alpha cellsin the affected hemiretinae of KVC cats remained normalin size, the dendritic arbors of beta cells that survived thecortical lesion were found to be significantly larger thannormal. Thus, the results of the present study support thehypothesis that ganglion cells in the cat retina achievetheir final size primarily through competitive interactionsthat are cell class specific.

Retinal ganglion cell developmentin the normal cat

Before comparing the results presented here with thoseof other studies that have investigated cell-cell interac-tions and retinal ganglion cell development, it is usefulfirst to describe briefly the developmental state of the catretina at birth, when visual cortex was removed. Previousstudies have shown that ganglion cell production in the catbegins around embryonic day 21 and is complete byembryonic day 35 (Walsh et al., 1983; Walsh and Polley,1985; gestation in the cat is 65 days). By embryonic day 45the primitive ganglion cell layer, which consists primarilyof a uniform layer of ganglion cells and putative ganglioncell precursors, is well established, and the axons of someganglion cells have reached target nuclei in the thalamusand midbrain (Stone et al., 1982; Williams and Chalupa,1982; Rapaport and Stone, 1983; Shatz, 1983; Robinson,1987). Over the next 2–3 weeks there is approximately a3-fold increase in the area of the retina, and regionaldifferences in ganglion cell density between central andperipheral retina become apparent. In addition, someganglion cells, particularly those located near centralretina, begin to display morphological features characteris-tic of ganglion cells in the mature retina (e.g., alpha, beta,and gamma) (Stone et al., 1982; Rapaport and Stone, 1983;Maslim et al., 1986; Robinson, 1987; Ramoa et al., 1988).

1An examination of retinal ganglion cell somas in whole mounts fromKVC cats suggests that ganglion cells begin to degenerate during thesecond postoperative week (R.E. Kalil, unpublished observations).

DENDRITIC COMPETITION IN THE NEONATAL CAT RETINA 475

Fig. 3. Dendritic field size comparisons for alpha (A,B) and beta(C,D) ganglion cells in normal and KVC cats. Similar to Figure 2, Aand C compare the distribution of dendritic field sizes for each class of

ganglion cell while the mean areal differences are presented in B andD. Note the significant increase in the mean dendritic field sizes ofbeta, but not alpha, ganglion cells.

476 A.J. WEBER ET AL.

Despite the considerable development that occurs prena-tally, the retina of the newborn kitten is highly immaturerelative to that of the adult cat. Between birth andadulthood, the area of the retina again triples in size, andbecause this growth is not uniform (Stone et al., 1982;Mastronarde et al., 1984), the gradient in ganglion celldensity from central to peripheral retina increases mark-edly. In addition to an increase in the overall size of theretina, the birth of the kitten also represents a time ofsignificant growth and remodeling for retinal ganglioncells and their dendritic arbors. In brief, these changesinclude the pruning of exuberant, spine-like appendages, areduction in dendritic branch points, simplification ofdendritic branching patterns, completion of synaptogen-esis, and the establishment of adult soma and dendriticfield dimensions (Rusoff and Dubin, 1978; Morrison, 1982;Stone et al., 1982; Rapaport and Stone, 1983; Mastronardeet al., 1984; Maslim et al., 1986; Ramoa et al., 1988; Dannet al., 1988).

Retinal development following neonatalremoval of visual cortex

As noted previously, the result of the developmentalevents described above is a mature retina in which there isa closely matched, inverse relation between retinal gan-glion cell density and dendritic field size as one moves fromcentral to peripheral retina (Boycott and Wassle, 1974;Peichl and Wassle, 1979; Wassle et al., 1981a,b; Stanford,1987; Peichl, 1991). In simplest terms, it appears thatduring development the various classes of ganglion cellsare distributed differentially across the retinal surface andtheir dendritic fields then grow to fill the space available,thus ensuring a relatively uniform level of retinal cover-age. Since no new ganglion cells are added to the cat retinapostnatally, it has been suggested that formation of thecentral-peripheral gradient in ganglion cell density in-volves the redistribution of existing neurons, either throughdifferential growth of the retina (Mastronarde et al., 1984),differential maturation of ganglion cell precursors (Stoneet al., 1982), or a combination of both.

While Mastronarde et al. (1984) have suggested thatdifferential postnatal growth of the retina and the passivestretching of retinal ganglion cell dendrites contribute tothe establishment of final dendritic arbor size in the cat,these authors point out that this simple mechanism aloneis not sufficient, and that active growth and expansion ofthe dendritic arbor also must be involved. As noted above,there is considerable evidence for such growth, not onlyfrom studies of dendritic field development in the normalretina (Rusoff and Dubin, 1978; Mastronarde et al., 1984;Maslim et al., 1986; Ramoa et al., 1988; Dann et al., 1988),but also from studies of retinae in which normal ganglioncell density has been altered experimentally (Linden andPerry, 1982; Perry and Linden, 1982; Rapaport and Stone,1983; Eysel et al., 1985; Kirby and Chalupa, 1986; Leven-thal et al., 1988; Perry and Maffei, 1988; Leventhal et al.,1989; Linden, 1993; Ault et al., 1993; Deplano et al., 1994).Of these studies, however, only those of Ault et al. (1993)and Deplano et al. (1994) have attempted to addressspecifically whether the development of retinal ganglioncells in the cat involve competitive interactions that arecell class specific. Since this also was the central issue ofthe present study, we have chosen here to focus primarilyon comparing our findings with the results of these twostudies.

One of the major differences between our study and thatof Deplano et al. (1994) is the means by which retinalganglion cell density was reduced in the experimental eye.While we made lesions of visual cortex in newborn cats toreduce the number of retinal beta cells selectively (Kalil,1984, 1990; Pearson et al., 1981; Payne et al., 1984; but seealso Rowe, 1990), Deplano et al. (1994) used a needle tomake a small lesion in the retina of a newborn kitten.Although this approach produces a reduction in ganglioncell density, it also results in damage to preganglionic cellsin the retina and produces an indirect loss of retinalganglion cells whose axons pass through the lesion area.

In order to examine the effect that the retinal lesion hadon surrounding ganglion cells, Deplano et al. (1994) la-beled the dendritic fields of these neurons retrogradelywith horseradish peroxidase (HRP). In agreement withprevious retinal lesion studies, they found that most of theganglion cells located along the border of the lesion havedendrites that are directed toward the zone of cell deple-tion, and that the magnitude of this effect decreases withincreased distance from the center of the lesion (Perry andLinden, 1982; Eysel et al., 1985; Perry and Maffei, 1988).Based on quantitative comparisons of dendritic field orien-tation and the spatial relations of alpha and beta cells,Deplano et al. (1994) conclude that the elongation of a betacell’s dendritic field toward the damaged area is influencedby three factors: (a) distance from the cell-depleted area,(b) proximity of the cell to neighboring beta cells, and (c)the presence or absence of alpha cells located between thebeta cell and the lesion site. While the authors cite theinvolvement of alpha cells as evidence that dendriticinteractions occur between developing ganglion cells ofdifferent classes, it is important to note that the dendriticfields of the beta cells described by Deplano et al. (1994)seem to vary in orientation independently of their locationrelative to the lesion or to neighboring ganglion cells (seetheir Fig. 2). Indeed, some beta cells located immediatelyadjacent to the lesion have dendritic fields that appear toshow no orientation preference. Given this cell to cellvariability in dendritic field orientation, it is difficult todetermine for many of the beta cells studied by Deplano etal. (1994) whether the presence or absence of an interposedalpha cell played a primary role in influencing theirdendritic field. Moreover, if the development of dendriticfields of alpha and beta cells are influenced by competitiveinteractions that are not cell class specific, then beta cellsshould influence the development of neighboring alphacells. However, the data of Deplano et al. (1994) suggestthat this does not happen. In nearly every instance wherea beta cell is located between the lesion site and an alphacell, the dendrites of the alpha cell pass freely through thedendritic field of the intervening beta cell. That thisphenomenon occurs suggests strongly that the competitiveinteractions obtained in the retinal studies of Deplano etal. (1994) are cell class specific.

One possible explanation for this apparent lack ofcompetitive reciprocity between alpha and beta cells isthat, following neonatal damage to the retina, the primaryfactor influencing retinal ganglion cell development is notdendritic competition, but rather the presence of trophicsubstances released from cells located within the affectedarea (see Perry and Maffei, 1988). Recent studies haveidentified a wide variety of trophic factors in the develop-ing vertebrate retina, and many of these factors also havebeen shown to be upregulated following injury to either the

DENDRITIC COMPETITION IN THE NEONATAL CAT RETINA 477

retina or the optic nerve (Johnson et al., 1986; Sievers etal., 1987; Thanos et al., 1989; Park and Hollenberg, 1989;Noji et al., 1990; LaVail et al., 1991; Bost et al., 1992; Meyet al., 1993; Mansour-Robaey et al., 1994; Kostyk et al.,1994; Wen et al., 1995; Perez and Caminos, 1995; Peinado-Ramon et al., 1996). Further, if alpha cells are moreefficient at accumulating such factors, it could explainsome of the variability in the dendritic fields of neighbor-ing beta cells, as well as the apparent failure of alpha celldendrites to respect the dendritic processes of interposedbeta cells. Additional support for a trophic influence comesfrom Perry and Maffei’s (1988) finding that reducingganglion cell density, and therefore presumably dendriticcompetition, prior to the retinal lesion did not attenuatethe abnormal dendritic field biases of ganglion cells sur-rounding the region of retinal damage. Thus, while it ispossible that the results of Deplano et al. (1994) provideevidence for cell class independent interactions, it isimportant to keep in mind that a lesion of the retina mayhave important, nonspecific consequences for the globalenvironment in which surviving retinal ganglion cellsdevelop. By far the greatest difference between the studyof Deplano et al. (1994) and the present investigation is thespecificity of the lesion. In the present study there was aselective reduction in the number of beta cells, while alphacells suffered no primary effect from the experimentalmanipulation. That alpha cells also appear to display nosecondary effects provides strong evidence that alpha cellsmaintain normal competitive interactions among them-selves in KVC retinae.

In contrast with the retinal lesion studies of Deplano etal. (1994), Ault et al. (1993) used a method similar to thatused in the present study, namely, neonatal removal ofvisual cortex, to create a competitive imbalance betweendeveloping alpha and beta ganglion cells. Based on compari-sons of the soma sizes and retinal distributions of retro-gradely-labeled alpha cells in normal and beta cell de-pleted regions of the retina, Ault et al. (1993) concludedthat retinal ganglion cell development in the cat involvesinteractions that are cell class specific and not cell classspecific. As evidence for cell class specific interactions, Aultet al. (1993) cite data showing a normal density anddistribution of alpha cells in the affected hemiretina, aresult that is in agreement with previous anatomical(Pearson et al., 1981; Kalil, 1984, 1990; Payne et al., 1984;Rowe, 1990) and physiological (Tong et al., 1982) studies.The conclusion by these authors that interactions that arenot cell class specific also are involved in retinal ganglioncell development is based primarily on cell size measure-ments showing that alpha cells located in the beta celldepleted region of the area centralis are larger thannormal. However, it is important to point out that, asshown in Figure 2 of Ault et al. (1993), this effect isrestricted only to those alpha cells located within 1 mm ofthe area centralis. By 2 mm from the area centralis theeffect is lost, and alpha cells located at greater retinaleccentricities in the affected hemiretina appear to besmaller than normal.

We did not study alpha cells near central retina becausewe wanted to avoid cells in the affected hemiretina thatmight project to the undamaged side of the brain (Stone,1966). Nevertheless, our finding that neonatal damage tovisual cortex has no significant effect on either the soma ordendritic field sizes of alpha cells located outside the mostcentral area of the retina is consistent with the results of

Ault et al. (1993) and is supported by the work of Rowe(1990). With respect to the latter study, it is important tonote that unlike Ault et al. (1993), Rowe (1990) found nosignificant change in the soma sizes of alpha cells locatedin either central or peripheral temporal retina in cats thathad lesions of visual cortex at birth.

One factor that might be important in explaining thedifferent results of these studies is a difference in theextent of the cortical lesions. While our lesions and those ofRowe (1990) were restricted to areas 17, 18, and 19 ofvisual cortex, those of Ault et al. (1993) also included theadjacent lateral suprasylvian area (LS). In the normaladult cat, this cortical area receives retinal input fromalpha cells by way of the medial intralaminar nucleus, andfrom gamma cells through a pathway that involves theC-layers of the lateral geniculate nucleus (LGN) andgeniculate wing. The lateral suprasylvian area does notappear to receive input from retinal beta cells. Instead,these neurons project mainly to medium-sized relay cellsin the A-layers of the LGN, which in turn project to corticalarea 17 (see Sherman, 1985 for review). Following removalof visual cortex at birth, there is a general reorganizationof the retino-geniculo-cortical pathway that involves anenhancement of the original projection to LS from smallgamma cells in the C-layers of the LGN, and the additionof a projection from large, presumably alpha, relay cellslocated in both the A- and C-layers of the nucleus (seeKalil, 1984 for review). In addition, more recent work byPayne et al. (1993) has indicated that neonatal removal ofvisual cortex can lead to the formation of a novel projectionfrom the retina to the lateral posterior nucleus (LP), whichalso has been shown to project to LS (Symonds et al., 1981;Raczkowski and Rosenquist, 1983; Kalil et al., 1991).

Taken together, these data suggest strongly the possibil-ity that the larger cortical lesions of Ault et al. (1993)might have resulted in a more extensive reorganization ofthe retino-geniculo-cortical pathway in their animals com-pared with ours. Indeed, it is interesting to note that,although Ault et al. (1993) report no significant differencesin the number and distribution of alpha cells, they also donot indicate the retrograde labeling of any surviving betacells in their experimental animals. Since several studieshave shown that neonatal removal of areas 17, 18, and 19results in a significant reduction (65–90%), but not thecomplete loss, of these neurons (Tong et al., 1982; Kalil,1984; Payne et al., 1984; Rowe, 1990), it is possible thatone consequence of the larger lesions used by Ault et al.(1993) was a more complete removal of retinal beta cells.Of particular significance then is whether this additionalcell loss might have involved primarily beta cells in thearea centralis, the region where Ault et al. (1993) found thegreatest increase in alpha cell soma size. The work of Kalil(1984) and Payne et al. (1993) indicate that followinglesions of areas 17, 18, and 19, the novel and enhancedprojections to LS and LP are to regions of those nuclei thatrepresent central retina. In addition, Payne et al. (1992)note that following lesions of areas 17 and 18 at differentpostnatal ages, the first beta cells to become resistant tothe cortical lesion are those located in the area centralis.Based on these findings, and the work of Rapaport andStone (1983) showing that complete section of the opticnerve also results in a general increase in ganglion cellsoma sizes, but that the effect is restricted to central retinawhere ganglion cell density is greatest, the possibilityexists that the increased soma sizes of alpha cells found by

478 A.J. WEBER ET AL.

Ault et al. (1993) represents a non-specific response to ageneral ‘‘decrowding’’ of central retina neurons, and notcompetitive interactions among specific classes of ganglioncells.

Finally, it is important to keep in mind that the competi-tion that is believed to occur during retinal ganglion celldevelopment is among dendrites for synaptic space. Owingto the incomplete staining of the dendritic arbors of thealpha cells examined by Ault et al. (1990), it is difficult toevaluate whether the increase in ganglion cell soma sizesthey reported also are accompanied by increases in den-dritic field size. While the staining of single ganglion cellsby intracellular injection is limited by the number ofganglion cells that can be stained in each eye, this tech-nique has the distinct advantage over retrograde labelingmethods of providing detailed information concerning boththe soma and dendritic field morphologies of individualneurons. In addition, because dendrites are stained com-pletely, it removes the ambiguity of attempting to classifyganglion cells based on soma size alone, a particularlydifficult task in the central retina.

These considerations aside, the most significant differ-ence between the present study and that of Ault et al.(1993) is in the area of the retina in which retinal ganglioncells were analyzed. We did not inject cells near the areacentralis. However, when comparisons are made at similardistances from the area centralis, our results agree wellwith those obtained by Ault et al. (1993). Whether thereare competitive interactions in central retina that are notcell class specific remains unclear because the two previ-ous investigations that compared retinal ganglion celldimensions in the area centralis reported conflicting re-sults (Rowe, 1990; Ault et al., 1993). We believe, however,that the results presented here provide strong evidencethat for many retinal ganglion cells the final sizes of theirdendritic arbors are determined by competitive interac-tions between cells of the same class.

ACKNOWLEDGMENTS

This work was supported by NIH grants EY01331(R.E.K.) and EY04977 (L.R.S.).

LITERATURE CITED

Ault, S.J., K.G. Thompson, Y. Zhou, and A.G. Leventhal (1993) Selectivedepletion of beta cells affects the development of alpha cells in catretina. Vis. Neurosci. 10:237–245.

Bost, L.M., A.E. Aotaki-Keen, and L.M. Hjelmeland (1992) Coexpression ofFGF-5 and bFGF by the retinal pigment epithelium in vitro. Exp. EyeRes. 55:727–734.

Boycott, B.B., and H. Wassle (1974) The morphological types of ganglioncells of the domestic cat’s retina. J. Physiol. (London) 240:397–419.

Caserta, F., W.D. Eldred, R.E. Hausman, L.R. Stanford, and H.E. Stanley(1995) Determination of fractal dimension of physiologically character-ized neurons in two and three dimensions. J. Neurosci. Methods56:133–144.

Cleland, B.G., W.R. Levick, and H. Wassle (1975) Physiological identifica-tion of a morphological class of cat retinal ganglion cells. J. Physiol.(London) 248:151–171.

Dann, J.F., E.H. Buhl, and L. Peichl (1988) Postnatal dendritic maturationof alpha and beta ganglion cells in cat retina. J. Neurosci. 8:1485–1499.

Deplano, S., G.M. Ratto, and S. Bisti (1994) Interplay between the dendritictrees of alpha and beta ganglion cells during the development of the catretina. J. Comp. Neurol. 342:152–160.

Eysel, U.T., L. Peichl, and H. Wassle (1985) Dendritic plasticity in the earlypostnatal feline retina: Quantitative characteristics and sensitive pe-riod. J. Comp. Neurol. 242:134–145.

Fukuda, Y., C.-H. Hsiao, M. Watanabe, and H. Ito (1984) Morphologicalcorrelates of physiologically identified Y-, X-, and W-cells in cat retina.J. Neurophysiol. 52:999–1013.

Johnson, J.E., Y-A. Barde, M. Schwab, and H. Thoenen (1986) Brainderived neurotrophic factor supports the survival of cultured rat retinalganglion cells. J. Neurosci. 6:3031–3038.

Kalil, R.E. (1984) Removal of visual cortex in the cat: Effects on themorphological development of the retino-geniculo-cortical pathway. InJ. Stone, B. Dreher and D.H. Rapaport (eds): Development of VisualPathways in Mammals. New York: Alan R. Liss, pp. 257–274.

Kalil, R.E. (1990) Reorganization of retinogeniculate connections in the catfollowing damage to the visual cortex. In A. Bjorkland, A.J. Aguayo, andD. Ottoson (eds): Brain Repair. London: Macmillan Press, pp. 285–307.

Kalil, R.E., and M. Behan (1987) Synaptic reorganization in the dorsallateral geniculate nucleus following damage to visual cortex in newbornor adult cats. J. Comp. Neurol. 257:216–236.

Kalil, R.E., L. Tong, and P.D. Spear (1991) Thalamic projections to thelateral suprasylvian visual area in cats with neonatal or adult visualcortex damage. J. Comp. Neurol. 314:512–525.

Kirby, M.A., and L.M. Chalupa (1986) Retinal crowding alters the morphol-ogy of alpha ganglion cells. J. Comp. Neurol. 251:532–541.

Kostyk, S.K., P.A. D’Amore, I.M. Herman, and J.A. Wagner (1994) Opticnerve injury alters basic fibroblast growth factor localization in theretina and optic tract. J. Neurosci. 14:1441–1449.

LaVail, M.M., E.G. Faktorovich, J.M. Hepler, K.L. Person, D. Yasumura,M.T. Matthes and R.H. Steinberg (1991) Basic fibroblast growth factorprotects photoreceptors from light-induced degeneration in albino rats.Ann. N.Y. Acad. Sci. December 20, 1991.

Lennie, P. (1980) Parallel visual pathways: A review. Vis. Res. 20:561–594.Leventhal, A.G., S.J. Ault, D.J. Vitek, and T. Shou (1989) Extrinsic

determinants of retinal ganglion cell development in primates. J. Comp.Neurol. 286:170–189.

Leventhal, A.G., J.D. Schall, and S.J. Ault (1988) Extrinsic determinants ofretinal ganglion cell structure in the cat. J. Neurosci. 8:2028–2038.

Linden, R. (1993) Dendritic competition in the developing retina: Ganglioncell density gradients and laterally displaced dendrites. Vis. Neurosci.10:313–324.

Linden, R., and V.H. Perry (1982) Ganglion cell death within the developingretina: A regulatory role for retinal dendrites? Neuroscience 7:2813–2827.

Mansour-Robaey, S., D.B. Clarke, Y.-C. Wang, G.M. Bray, and A.J. Aguayo(1994) Effects of ocular injury and administration of brain-derivedneurotrophic factor on survival and regrowth of axotomized retinalganglion ells. Proc. Natl. Acad. Sci. U.S.A. 91:1632–1636.

Maslim, J., M. Webster, and J. Stone (1986) Stages in the structuraldifferentiation of retinal ganglion cells. J. Comp. Neurol. 254:382–402.

Mastronaarde D.N., M.A. Thiebeault, and M.W. Dubin (1984) Non-uniformpostnatal growth of the cat retina. J. Comp. Neurol. 228:598–608.

Mey, J., and S. Thanos (1993) Intravitreal injections of neurotrophic factorssupport the survival of axotomized retinal ganglion cells in adult rats invivo. Brain Res. 602:304–317.

Montague, P.R., and M.J. Friedlander (1991) Morphogenesis and territorialcoverage by isolated mammalian retinal ganglion cells. J. Neurosci.11:1440–1457.

Morrison, J.D. (1982) Postnatal development of the area centralis of thekitten retina: An electron microscopic study. J. Anat. 135:255–271.

Noji, S., T. Matsuo, E. Koyama, T. Yamaai, T. Nohno, N. Matsuo, and S.Taniguchi (1990) Expression pattern of acidic and basic fibroblastgrowth factor genes in adult rat eyes. Biochem. Biophys. Res. Comm.168:343–349.

Otsuka, R., and R. Hassler (1962) Uber aufbau and gliederrung dercorticalen sehsphare bei der katze. Arch. Psychiaatr. Z. Ges. Neurol.203:212–234.

Park, C.M., and M.J. Hollenberg (1989) Basic fibroblast growth factorinduces retinal regeneration in vivo. Dev. Biol. 134:201–205.

Payne, B.R., H.A. Foley, and S.G. Lomber (1993) Visual cortex damage-induced growth of retinal axons in the lateral posterior nucleus of thecat. Visual Neurosci. 10:747–752.

Payne, B.R., H.E. Pearson, and P. Cornwell (1984) Transneuronal degenera-tion of beta retinal ganglion cells in the cat. Proc. R. Soc. 222:15–32.

Payne, B.R., H.E. Pearson, and P. Cornwell (1992) Survival and death ofretinal ganglion cells in the cat. Invest. Ophthalmol. Vis. Sci. 33:1132.

Pearson, H.E., D.L. Labar, B.R. Payne, P. Cornwell, and N. Aggarwal (1981)Transneuronal retrograde degeneration in the cat retina followingneonatal ablation of visual cortex. Brain Res. 212:470–475.

DENDRITIC COMPETITION IN THE NEONATAL CAT RETINA 479

Peichl, L. (1991) Alpha ganglion cells in mammalian retinae: Commonproperties, species differences, and some comments on other ganglioncells. Vis. Neurosci. 7:155–169.

Peichl, L., and H. Wassle (1981) Morphological identification of on- andoff-centre brisk transient (Y) cells in the cat retina. Proc. R. Soc.212:139–156.

Peinado-Ram(n, P., M. Salvador, M.P. Villegas-P(rez, and M. Vidal-Sanz(1996) Effects of axotomy and intraocular administration of NT-4, NT-3,and brain-derived neurotrophic factor on the survival of adult ratretinal ganglion cells. Invest. Ophthalmol. Vis. Sci. 37:489–500.

Perez, M.-T.R., and E. Caminos (1995) Expression of brain-derived neuroto-phic factor and of its function receptor in neonatal and adult rat retina.Neurosci. Lett. 183:96–99.

Perry, V.H., and R. Linden (1982) Evidence for dendritic competition in thedeveloping retina. Nature 297:683–685.

Perry, V.H., and L. Maffei (1988) Dendritic competition: Competition forwhat? Dev. Brain Res. 41:195–200.

Raczkowski, D., and A.C. Rosenquist (1983) Connections of the multiplevisual cortical areas with the lateral posterior-pulvinar complex andadjacent thalamic nuclei in the cat. J. Neurosci. 3:1912–1942.

Ramoa, A.S., G. Campbell, and C. Shatz (1988) Dendritic growth andremodeling of cat retinal ganglion cells during fetal and postnataldevelopment. J. Neurosci. 8:4239–4261.

Rapaport, D.H., and J. Stone (1983a) Time course of morphologicaldifferentiation of cat retinal ganglion cells: Influences on soma size. J.Comp. Neurol. 221:273–279.

Robinson, S.R. (1987) Ontogeny of the area centralis in the cat. J. Comp.Neurol. 255:50–67.

Rowe, M.H. (1990) Evidence for degeneration of retinal W cells followingearly visual cortical removal in cats. Brain, Behav. Evol. 35:253–267.

Rusoff, A.C., and M.W. Dubin (1978) Kitten ganglion cells: Dendritic fieldsize at 3 weeks of age and correlation with receptive field size. Invest.Ophthalmol. Vis. Sci. 17:819–821.

Saito, H. (1983) Morphology and physiologically identified X-, Y-and W-typeretinal ganglion cells of the cat. J. Comp. Neurol. 221:279–288.

Shatz, C.J. (1983) The prenatal development of the cat’s retinogeniculatepathway. J. Neurosci. 3:482–499.

Sherman, S.M. (1985) Functional organization of the W-, X-, and Y-cellpathways in the cat: A review and hypothesis. Prog. Psychobiol. Physiol.Psychol. 11:233–314.

Sherman, S.M., and P.D. Spear (1982) Organization of visual pathways innormal and visually deprived cats. Physiol. Rev. 62:738–855.

Sievers, J., B. Hausmann, K. Unsicker, and M. Berry (1987) Fibroblastgrowth factors promote the survival of adult retinal ganglion cells aftertransection of the optic nerve. Neurosci. Lett. 76:157–162.

Stanford, L.R., and S.M. Sherman (1984) Structure/function relationshipsof retinal ganglion cells in the cat. Brain Res. 297:381–386.

Stanford, L.R. (1987) X-cells in the cat retina: Relationships between themorphology and physiology of a class of cat retinal ganglion cells. J.Neurophysiol. 58:940–964.

Stone, J. (1966) The naso-temporal division of the cat’s retina. J. Comp.Neurol. 136:585–600.

Stone, J., D.H. Rapaport, R.W. Williams, and L. Chalupa (1982) Uniformityof cell distribution in the ganglion cell layer of prenatal cat retina:Implications for mechanisms of retinal development. Dev. Brain Res.2:231–242.

Symonds, L.L., A.C. Rosenquist, S.B. Edwards, and L.A. Palmer (1981)Projections of the pulvinar-lateral posterior complex to visual corticalareas in the cat. Neuroscience 6:1995–2020.

Thanos, S., M. Bahr, Y.-H. Barde, and J. Vanselow (1989) Survival andelongation of adult rat retinal ganglion cells: In vitro effect of lesionedsciatic nerve and brain derived neurotrophic factor. Eur. J. Neurosci.1:9–26.

Tong, L., P.D. Spear, R.E. Kalil, and E.C. Callahan (1982) Loss of retinalX-cells in cats with neonatal or adult visual cortex damage. Science.217:72–75.

Tusa, R.J., L.A. Palmer, and A.C. Rosenquist (1978) The retinotopicorganization of area 17 (striate cortex) in the cat. J. Comp. Neurol.177:213–236.

Walsh, C., E.H. Polley, T.L. Hickey, and R.W. Guillery (1983) Generation ofcat retinal ganglion cells in relation to central pathways. Nature302:611–614.

Walsh, C., and E.H. Polley (1985) The topography of ganglion cell produc-tion in the cat’s retina. J. Neurosci. 5:741–750.

Wassle, H., L. Peichl, and B.B. Boycott (1981a) Morphology and topographyof on- and off-alpha cells in the cat retina. Proc. R. Soc. Lond. B.212:157–175.

Wassle, H., B.B. Boycott, and R.-B. Illing (1981b) Morphology and mosaic ofon- and off-beta cells in the cat retina and some functional consider-ations. Proc. R. Soc. Lond. B. 212:177–195.

Weber, A.J., R.E. Kalil, and L.R. Stanford (1989) Morphology of single,physiologically identified retinogeniculate Y-cell axons in the cat follow-ing damage to visual cortex at birth. J.Comp. Neurol. 282:446–455.

Wen R., Y. Song, T. Cheng, M.T. Matthes, D. Yasumura, M.M. LaVail, andR.H. Steinberg (1995) Injury-induced upregulation of bFGF and CNTFmRNAS in the rat retina. J. Neurosci. 15:7377–7385.

Williams, R.W., and L.M. Chalupa (1982) Prenatal development of retinocol-licular projections in the Cat: An anterograde tracer transport study. J.Neurosci. 2:604–622.

480 A.J. WEBER ET AL.