THE JOURNAL OF COMPARATIVE NEUROLOGY 368~295-303 ( 1996)

Topographic Organization in the Retinocollicular Pathway of the Fetal Cat Demonstrated by

Retrograde Labeling of Ganglion Cells

LEO M. CHALUPA, CARA J. SNIDER, AND MICHAEL A. KIRBY Section of Neurobiology, Physiology, and Behavior, Department of Psychology and

Center for Neuroscience, University of California, Davis, California 95616 (L.M.C., C.J.S.); Department of Pediatrics, Loma Linda University, Loma Linda, California 92350 (M.A.K.)

ABSTRACT The topographic organization of the developing retinocollicular pathway was assessed by

making focal deposits of a retrograde tracer (usually rhodamine latex beads) into the superficial layers of the superior colliculus of fetal cats at known gestational ages. Subsequently, the distributions of labeled cells in the contralateral and ipsilateral retinas were examined. At all stages of development, a high density of labeled cells was found in a delimited area (core region) of both retinas. The locations of the retinal regions containing the high density of labeled cells varied with the locus of the tracer deposit in the superior colliculus in a manner consistent with the topographic organization of the mature cat’s retinocollicular pathway. Additionally, some labeled ganglion cells, considered to be ectopic, were found to be scattered throughout the contralateral and ipsilateral fetal retinas. Such ectopic cells were few in number throughout prenatal development. For every 100 cells projecting to the appropriate region of the colliculus, we estimate that less than one ganglion cell makes a gross projection error. The incidence of ectopic cells did not differ between the contralateral and ipsilateral retina, even though the overall density of crossed labeled cells was always greater than that of uncrossed labeled cells. In the youngest fetal animals, tracer deposits into the caudal portion of the superior colliculus resulted in a core region of labeled cells in the contralateral nasal retina as well as in the nasal ipsilateral retina. Such uncrossed nasal cells, not seen in more mature animals, appear to innervate the appropriate topographic location of the superior colliculus, but on the wrong side of the brain. Most likely, these uncrossed nasal ganglion cells contribute to the widespread distribution of the ipsilateral retinocollicular pathway observed in fetal cats after intraocular injections of anterograde tracers (Williams and Chalupa, 1982). Collectively, our findings demonstrate that the developing retinocollicular pathway of the fetal cat is characterized by a remarkable degree of topographic precision. o 1996 WiIey-Liss, Inc.

Indexing terms: retinotopy, fetal development, superior colliculus

At maturity, the projections of retinal ganglion cells are characterized by an exquisite specificity, with adjacent cells innervating neighboring retinorecipient neurons. How such retinotopic order is established is unknown, and this prob- lem continues to pose a formidable challenge to developmen- tal neurobiologists. Studies on a variety of species have shown, however, that retinal projections are generally less precise early in development than at maturity (for review, see Udin and Fawcett, 1988).

In the retinotectal pathway of the fish and frog, topo- graphic order has been shown to be established by the initial ingrowth and arborization of retinal fibers into appropriate regions of the target (Holt and Harris, 1983;

Sakaguchi and Murphy, 1985; Stuermer, 1988). Develop- mental refinement in this system results primarily from the unequal growth patterns of the tectum relative to that of retinal arbors (Sakaguchi and Murphy, 1985; Fujisawa, 1987; Stuermer, 1988). In some lower vertebrates, such disparate growth occurs throughout the lifetime of the animal, requiring a continued shifting of retinotectal synap- tic contacts (Easter and Stuermer, 1984; Raymond, 1986).

Accepted November 17,1995. Address reprint requests to Leo M. Chalupa, Section of Neurobiology,

Physiology, and Behavior, Division of Biological Sciences, University of California, Davis, CA 95616.

O 1996 WILEY-LISS, INC.

296

Owing to formidable technical impediments, investiga- tion of retinotopic order in developing mammals has lagged behind that of non-mammalian species. I t has been com- monly assumed, however, that the early projections of the mammalian retina are largely diffuse. This was first sug- gested by the results of studies describing the widespread distribution of early retinal projections after intraocular injections of various tracers (Rakic, 1976; Land and Lund, 1979; Williams and Chalupa, 1982; Shatz, 1983). The imprecise nature of early retinal projections in mammals has been further underscored by the recent work of O’Leary and colleagues. These investigators utilized retrograde (O’Leary et al., 1986) as well as anterograde (Simon and O’Leary, 1990, 1992) tracing methods to study the crossed retinocollicular pathway of the developing rat. Although “some bias for the correct region of the superior colliculus” was observed, the early retinocollicular projection was characterized as “very diffuse” in that ingrowing axons were found to mistarget widely (Simon and O’Leary, 1990). It has also been demonstrated that activity-mediated events play a crucial role in the elimination of gross targeting errors in the rodent’s retinocollicular pathway (O’Leary et al., 1986; Thompson and Holt, 1989).

Although the cat has been often used in visual develop- ment studies, remarkably little is known about the topo- graphic organization of early retinal projections in carni- vores. The available evidence, although limited, seems to indicate a substantial degree of topographic order in the developing carnivore retinogeniculate pathway. For in- stance, small deposits of horseradish peroxidase (HRP) into the optic tract of fetal cats has been found to produce a well-localized pattern of labeled retinogeniculate terminals, even at a time when the projections of the two eyes are overlapped (Sretavan and Shatz, 1987). Furthermore, Jef- fery (1985) has shown that after small retinal lesions in the newborn ferret, intraocular injections of HRP resulted in a sharply defined non-labeled region in the contralateral lateral geniculate nucleus, corresponding to the representa- tion of the artificial scotoma. This approach, however, is not suitable for resolving the incidence or magnitude of target- ing errors.

In the present study, we describe the results of experi- ments which relied on small deposits of retrograde tracers to assess the degree of topographic order in the retinocollicu- lar pathway of the fetal cat. In particular, we were inter- ested in examining the prenatal period during which the projections from the two eyes to the superior colliculus are largely intermingled (Williams and Chalupa, 1982). Ab- stracts summarizing some of the findings have been pub- lished (Ostrach et al., 1986; Snider and Chalupa, 1995).

L.M. CHALUPA ET AL.

MATERIALS AND METHODS A total of 19 fetal animals ranging in age from embryonic

day (E) 37 to E59, four postnatal cats, and one adult animal were used in this study (Table 1). All animals were obtained from our breeding colony. To time gestational age, estrous females were placed with a tomcat for a period of 8-12 hours. Gestation is 65 days ( + 2 days), and the first 24 hours after birth is designated as postnatal day 0.

Surgical procedure and retrograde labeling The in utero surgical procedures have been described in

detail in previous publications from this laboratory (e.g., Williams and Chalupa, 1982). Here we emphasize that all surgeries were carried out under sterile conditions in

TABLE 1. Experimental Conditions’

Age Tracer No. of animals

E38 E39 E40 E40 E42 E45 E47 E51 E59 P3 P3 P4 P8 Adult

WGA-HRP RLB RLB DY RLB WGA-HRP RLB DY HRP RLB WGA-HRP RLB RLB WGA-HRP

2 4 4 1 2 2 1 1 1 1 1 1 1 1

‘WGA-HRP, wheatgerm agglutinin-horseradish peroxidase; RLB, rhodamine latex bead; DY, diamidino yellow.

accordance with a protocol approved by the campus animal use and care committee.

Anesthesia was induced in timed-pregnant animals with Halothane, and maintained during surgery with 2% Halo- thane in a mixture of 50% nitrous oxide and 50% oxygen. Following a midline abdominal incision, the fetal head was exposed through a small cut over the non-placental portion of the uterus. In fetal animals the cortex does not cover the superior colliculus, so it was feasible to directly visualize the dorsal surface of this midbrain structure by retracting a bone flap over the posterior portion of the skull. A single deposit of a retrograde tracer (wheat germ agglutinin- horseradish peroxidase [WGA-HRP], rhodamine latex beads [RLB], or diamidino yellow [DY]) was made into the colliculus under visual control by using a glass micropipette with a tip diameter of 30-50 pm. Typically one fetus was injected in each uterine horn. All incisions were then closed, and the animal was allowed to recover from anesthesia. After allowing 24-48 hours for transport of the tracer, the fetal animals were removed by cesarian section, adminis- tered a lethal dose of barbiturate, and perfused transcardi- ally with 0.9% saline followed by appropriate fixatives. For peroxidase processing the fixative was 1% paraformalde- hydei2% glutaraldehyde, and in the animals injected with the fluorescent tracers 10% formalin was used. All solutions were buffered in 0.1 M phosphate buffer.

The postnatal animals were anesthetized with a combina- tion of ace promazine (0.1 mgikg) and ketamine (20 mgikg) or with Telazol (15 mgikg). Following placement of the animal in a stereotaxic instrument, an opening was made in the skull, and the cortex was aspirated to reveal the dorsal surface of the superior colliculus. A single deposit of tracer (WGA-HRP, RLB, or DY) was made into the superficial layers of the superior colliculus through a glass micropi- pette with a tip diameter of 30-50 Fm. Iontophoresis was used in the case of WGA-HRP, and pressure for the other tracers. After a 10- to 48-hour survival period the animal was overdosed with barbiturate and perfused transcardially as described above.

Histological processing The retinas were dissected immediately after perfusion.

The anterior chamber of each eye was opened to remove the lens and vitreous body. In the case of the WGA-HRP deposits, the dissected retinas were reacted for peroxidase by using the metal-intensified pyrocatechol-phenylenedi- amine protocol described by Hanker et al. (1977). Retinas were flattened onto a gelatinized slide under a polycarbon- ate membrane filter paper with a weighted coverslip and

TOPOGRAPHY IN RETINOCOLLICULAR PATHWAY 297

fixed overnight in 2% cacodylate-buffered glutaraldehyde. Before coverslipping, the tissue was dehydrated in dimethyl sulfoxide (DMSO). For the fluorescent tracers the retinas were wholemounted on gelatinized slides and coverslipped with a mixture of 50% glycerol/50% phosphate buffer.

The brains were immersed in 10% sucrose for cryoprotec- tion and embedded in either gelatin (HRP deposits) or in egg yolk (fluorescent deposits). Serial sections were cut frozen in the coronal plane using a sliding microtome. To visualize the WGA-HRP deposits, the tissue was reacted for peroxidase following the tetramethyl benzadiene (TMB) protocol of Mesulam (19781, and alternate sections were stained with either thionin or neutral red.

Reconstruction of deposit sites Prior to sectioning a brain, photographs were taken of

the midbrain dorsal surface. Subsequently, the deposit site was reconstructed from a series of camera lucida drawings of sections through the superior colliculus containing the tracer. The locus and extent of the deposit was then indicated on a dorsal view drawing of the midbrain.

Analysis of labeled retinal ganglion cells Each retina was mapped initially at low-power using

stage coordinates that corresponded to a drawing of the flatmount. In all cases with a focal deposit in the superficial layers of the superior colliculus, there was a well-defined area with a high density of labeled cells (the core region). Cell densities were determined by counting all labeled cells in adjacent grids (158 Fm2) using a 4 0 ~ objective. From such counts isodensity lines were constructed in two- or three-dimensional planes (Surfer Software). This defined the core region of labeled cells. Outside the core region the position of every labeled cell was indicated on the flatmount drawing. Counts of the total number of labeled cells outside the core area provided an estimate of the number of ganglion cells in topographically inappropriate regions of the retina.

RESULTS Irrespective of the age of the fetal animal or the type of

tracer employed, focal deposits within the superficial layers of the superior colliculus resulted in a high density of labeled cells in a delimited region of the retina. The RLB were used in preference to DY and WGA-HRP, because this tracer yielded the smallest deposit site as well as a high density of well-labeled retinal ganglion cells. An example of an RLB deposit in an E40 animal is illustrated in Figure 1A,B. At this age the ipsilateral projection is still wide- spread, and there is almost complete overlap in the projec- tion patterns of the two eyes (Williams and Chalupa, 1982). Note that the entire midbrain can be visualized at this age (Fig. lA), although the precise boundary between the superior and inferior colliculus cannot be discerned in this dorsal surface view. The RLB deposit site is clearly evident in the right superior colliculus (Fig. 1A). In coronal sections this appeared as a punctate site confined to the superficial layers of the colliculus (Fig. 1B). Figure 1C-E shows three photomicrographs from the contralateral retina of this animal, which are examples of RLB-labeled ganglion cells. Labeled cells within the center of the core region may be seen in Figure 1C. The density of labeled cells decreased precipitously toward the border of the core region (Fig. 1D). The location of the core region, containing the high density

of labeled cells, within the contralateral and ipsilateral retinas reflected the locus of the tracer deposit within the superior colliculus in a manner consistent with the topo- graphic organization of the mature retinocollicular path- way (described below). Outside of the core regions only a few labeled cells were observed (Fig. IE). We considered such neurons to be ectopic, and these appeared to be distributed randomly across the surface of the retina.

The variation in the density of labeled cells within the core region may be better seen in Figure 2, which illustrates the isodensity profiles of labeled cells in the contralateral and ipsilateral retinas of two animals (E38 and E45). In these representative cases, as well as in the overall sample, the number of labeled cells in the contralateral core area is greater than the ipsilateral core area. This reflects the higher peak density of labeled cells as well as the greater extent of the core region within the contralateral retina. This difference in the labeling pattern between the contra- lateral and ipsilateral retina was less pronounced at youn- ger ages. In other words, the number of labeled cells in each retina was more comparable in younger than in older fetal retinas, although this point is not obvious in the distribu- tions shown in Figure 2. This finding is illustrated in Figure 3, which depicts the ratios of the peak number of ipsilateral and contralateral labeled cells at different developmental ages. It should be stressed that the foregoing comparisons are all based on core areas located within the contralateral nasal retina and the ipsilateral temporal retina.

Figure 4 shows the contralateral and ipsilateral core regions in animals of different ages resulting from place- ments of tracer deposits into different loci of the superior colliculus. In these, and all other cases, the location of the core region of labeled cells reflected the locus of the tracer placements within the superior colliculus. Moreover, with the exception of the labeling seen in the ipsilateral retina after tracer deposits into the caudal portion of the SC (to be described below), the loci of the core regions reflected the topographic order of the mature retinocollicular pathway. Tracer deposits into the medial aspect of the colliculus resulted in labeled regions in the inferior retina; those into the lateral colliculus produced labeling in the superior portion of the retina. In other cases, deposits into the rostra1 portion of the superior colliculus, representing the central portion of the visual field at maturity, yielded core regions which clustered within and around the presumed area centralis.

To assess the incidence of fetal ganglion cells with gross misprojection errors, counts were made in representative cases of every labeled cell outside the core region. These results are summarized in Table 2, which indicates the total number of ectopic cells in the contralateral and ipsilateral retinas of animals at different fetal ages, and the peak cell count (in terms of mm2) in the topographically appropriate core regions. As stated previously, the ectopic cells appeared to be randomly distributed throughout the retina, and in all cases, the number of such misprojecting cells was relatively small when compared to the peak number of labeled cells. This was the case even in the youngest fetal animals studied, those injected at E38. Furthermore, we observed no significant difference in the incidence of crossed and uncrossed ectopic cells.

Following intraocular injections of anterograde tracers in fetal cats younger than E50, Williams and Chalupa (1982) showed that the superficial layers of the entire ipsilateral superior colliculus is labeled, with exception of the most

3

TOPOGRAPHY IN RETINOCOLLICULAR PATHWAY 299

By E58 the overall pattern of the retinocollicular projec- tion appeared similar to that seen in the adult cat (Williams and Chalupa, 1982). By this age the ipsilateral projection had withdrawn from the caudal aspect of the colliculus. A deposit of WGA-HRP into the caudal colliculus of an E58 fetus yielded labeled cells only in the contralateral retina (data not illustrated). As expected, the core region was in the peripheral nasal retina, with a scattering of crossed ectopic cells. No labeled neurons were seen in the ipsilateral retina.

caudal pole (see their Figs. 3,4). In contrast in older fetuses, as in the mature cat, the ipsilateral retinocollicular projec- tion is confined to the binocular segment in the rostral half of the colliculus. In the adult cat, the uncrossed projection arises from ganglion cells situated in the temporal half of the retina. In the present study, we were particularly interested in determining the regional distribution of the ganglion cells in the ipsilateral retina projecting to the caudal portion of the colliculus in the younger fetal animals. Before the present study was initiated, three different outcomes could be envisaged. One possibility was that the exuberant ispilateral projection is primarily comprised of widely misprojecting cells, similar to the early pattern described by O’Leary’s laboratory in their work on the crossed retinocollicular pathway of the developing rat (O’Leary et al., 1986; Simon and O’Leary, 1990, 1992). Another possible outcome was that the uncrossed projec- tion to the caudal colliculus stems from ganglion cells largely restricted to the temporal hemiretina. This would be the case if the arbors of these neurons ramified widely to innervate both the appropriate rostral portion of the collicu- lus as well as the caudal aspect. The third possible outcome is that in the early fetal cat the nasal hemiretina maps onto the caudal portion of the colliculus.

In our material there were four cases in which a focal deposit was made into the caudal aspect of the superior colliculus: two cases at E38, one at E42, and one at E58. The ipsilateral retinocollicular projection is widespread between E38 and E46, but by E58 it is already restricted to the rostral binocular portion of the colliculus. Thus, the three younger fetuses provided an opportunity to examine the merits of the outcomes considered above. In each of the younger fetuses, a core region of labeled cells was seen in the nasal portion of both the ipsilateral and contralateral retina. This finding is illustrated for the two E38 cases in Figure 5A,B. Note the variation in the location of the core regions between the two animals as a function of tracer deposit in the superior colliculus. In Figure 5A the retinal label is in the superior retina, and the tracer deposit is in the lateral aspect of the caudal colliculus. The core regions within the inferior retinas reflect the more medial tracer deposit (Fig. 5B) in the other E38 animal. In these early fetal animals, the density of labeled cells in the nasal ipsilateral retina was much lower than in the contralateral retina, and also substantially lower than that found in the temporal ipsilateral retina after more rostral deposits. For instance, cell counts obtained in the case shown in Figure 5A reached a peak density of 35 cells/mm2 in the nasal ipsilateral retina as compared to a peak count of 626 cells/mm2 in the contralateral nasal retina.

Fig. 1. Rhodamine latex bead (RLB) deposit into the superior colliculus of an embryonic day 40 cat. A A photograph of the dorsal surface of the fetal cat brain showing the exposed midbrain and the deposit site in the contralateral superior colliculus. B: A 50-pm coronal section through the approximate center of the deposit site. Note the punctate nature of the deposit after the 48-hour survival period. C: Photomicrographs of ganglion cells in the contralateral retina labeled with RLB in portion of the core region containing the highest density of labeled cells. D: A lower density of cells labeled with RLB at the periphery of the core region. Note the sharp boundary between the region of the retina containing labeled cells and that lacking such neurons. E: Ganglion cells in the contralateral retina considered to he ectopic. Such labeled neurons were infrequent and appeared to be dispersed randomly across the retinal surface. Scale bar = 5 mm in A, 0.5 mm in B, 100 pm in C and E, 50 km in D.

DISCUSSION Because photoreceptors are not generated until relatively

late during ontogeny, the retinotopic organization of the early developing visual system can only be assessed by anatomical tracing methods. In the present study we relied on focal deposits of tracers into the superior colliculus to assess the topographic organization of the retinocollicular pathway in the fetal cat. Before discussing the implications of the findings, it is important to consider the limitations of this approach. One shortcoming is that it is unknown how the size of the reconstructed deposit site relates to the virtual uptake zone of the tracer. Even with tracer deposits of identical size and locus, it is likely that uptake of the tracer will differ from one case to another. For this reason it is not feasible to meaningfully relate the extent of the retinal region containing labeled cells with the size of the tracer deposits into the target structure. I t is also unknown how much of the axonal arbor needs to be exposed to the tracer to obtain detectable levels of label in projection neurons. Still another concern is the fiber of passage problem, which refers to the well-recognized possibility that axons coursing through a deposit site could result in retrogradely labeled cells. Such technical limitations have been discussed in detail elsewhere (e.g., Yhip and Kirby, 1990).

These technical considerations notwithstanding, several conclusions can be derived from the present results. The retinocollicular pathway of the cat was found to exhibit a remarkable degree of order throughout prenatal develop- ment. At all fetal ages, all but a few ganglion cells were found to project from a clearly delimited region of the retina to a specific region of the superior colliculus. This was observed for both the crossed and uncrossed pathway. It is also the case that during prenatal development small numbers of labeled cells were found to be distributed throughout the retina. We considered such neurons to be ectopic because equivalent injections did not produce label- ing of dispersed retinal ganglion cells in the postnatal animals. It is possible, however, that prenatal retinocollicu- lar fibers passing through the deposit site incorporate tracer more avidly than mature axons. If this were the case then our low estimate of misprojecting retinocollicular cells would be even lower.

Our findings also indicate that in one important respect the organization of the early uncrossed projection differs from that of the crossed projection. At maturity, uncrossed retinal fibers only innervate the rostral binocular portion of the cat’s superior colliculus (Graybiel, 1975; Harting and Guillery, 1976). In contrast, during early fetal develop- ment, the ipsilateral projection is widespread, innervating virtually the entire rostro-caudal extent of this midbrain structure (Williams and Chalupa, 1982). We were particu- larly interested in establishing the distribution of the

E38

Computer reconstructions of the distribution of labeled cells in an E38 and E45 fetal cat following placement of discrete deposits into the superior colliculus (SC). In these two representative cases, as well as all other animals studied, a single topographically appropriate area of the retina was found to contain a high concentration of labeled cells. Additionally, a few “ectopic” cells were found outside of this retinal location, although these constituted a minor percentage of the total labeled population. The distribution of labeled cells in each retina is

represented in three-dimensional plots where the Z-axis denotes cell density (cells/mm2). Isodensity plots (inset) indicate orientation of the retina and denote the location of individual labeled cells outside of the topographically appropriate area. The location of the deposit site in the SC of each animal is indicated by a solid circle. D, dorsal; V, ventral; T, temporal; N, nasal; L, left; R, right; r, rostral; small diamond within retinal wholemount, optic disk.

TOPOGRAPHY IN RETINOCOLLICULAR PATHWAY

1.0 -

.9 -

.8 - 2 . 7 - p: $ .6 -

U 5 .5 -

2 .4 - .3 - .2 -

.1 -

CI

301

a

I I I I I I I I I/ I

35 40 45 50 55 60 65' 70 adult

Gestational Age

Fig. 3. The ratios of the peak densities of labeled cells in the ipsilateralicontralateral retinas at different stages of development. Note that in all cases the peak density of labeled cells was greater in the contralateral retina, but this difference was less pronounced during early fetal development than in more mature animals. *denotes day of birth

retinal cells which early in development project to what will become the monocular segment in the caudal half of the superior colliculus. Although our material is based on only three successful early caudal injections, the results were unambiguous and consistent across animals. In each case the caudal deposit resulted in labeled ganglion cells concen- trated within the nasal retina of the ipsilateral eye. At maturity the caudal superior colliculus receives input from cells in the nasal retina, but only from the contralateral eye. This means that during early development a contingent of retinal ganglion cells in the nasal ipsilateral retina inner- vates the topographically appropriate region of the superior colliculus, but on the wrong side of the brain. Thus, innervation of the caudal portion of the superior colliculus (by the uncrossed ganglion cells situated in the nasal retina) appears to reflect a decussation error at the optic chiasm. Upon reaching the ipsilateral superior colliculus, the axons of these neurons appear capable of responding to target- derived cues to innervate the topographically appropriate region of the target.

Collectively, our observations provide evidence that in- growing retinal s o n s may mistarget in two different ways: by terminating in a wrong location on the correct side of the brain, or by terminating in the correct location on the wrong side. It seems reasonable to surmise that these two types of errors reflect completely different causes related in one case to guidance mechanisms expressed at the optic chiasm and in the other to target-derived cues.

The cells in the nasal retina of the developing cat that give rise to the uncrossed projection are not very great in number, about 15% of the overall ganglion cell population (Lia et al., 1987). Such neurons are probably eliminated during the normal course of development (Williams et al., 1986). Another possibility is that a small contingent of nasal cells project bilaterally early in development, and the loss of the uncrossed nasal cells reflects the withdrawal of supernumerary axonal branches. There is evidence for such

E 38 60

E 44

ipsi contra -

00

Fig. 4. Schematic maps showing the locus of tracer deposit in the superior colliculus (denoted by the dot) and the resulting regions in the contralateral and ipsilateral retinas containing the high density of labeled cells (shaded areas). The embryonic age denotes the day after conception when the tracer deposit was made. In three of the cases the survival period was 48 hours, whereas in the E44 animal it was 24 hours. The black dots in the retinas denote the optic disks. Note that the increase in the size of the retinas at the progressively older ages is scaled to the reflect the actual expansion of the retinal surface with maturation. The orientation of the drawings is indicated in the uppermost schematic. r, rostral; T, temporal; N, nasal.

302 L.M. CHALUPA ET AL.

The results of early eye removal studies have also shown that the elimination of the nasal ganglion cells with un- crossed projections involves binocular interactions because many ganglion cells in the nasal retina, which project ipsilaterally, remain in early monocularly enucleated ani- mals (e.g., Insausti et al., 1984; Thompson et al., 1993). Interestingly, in an electrophysiological mapping study of the superior colliculus in adult hamsters, monocularly enucleated at the day of birth, Thompson (1979) found that the visual map in the ipsilateral caudal colliculus exhibited polarities appropriate for the contralateral projection. More recently, Steineke et al. (1992) have reported the presence of cells in the caudal portion of the superior colliculus with visual receptive fields in the ipsilateral hemifield of adult cats monocularly enucleated before birth. Such aberrant receptive fields were found to be intermingled with contra- laterally responding cells with a normal visuotopic organiza- tion. These electrophysiological observations are entirely in accord with the results of the present study, if one assumes that the early eye removal stabilizes in part the normally transient uncrossed retinocollicular pathway stemming from the nasal hemiretina.

The results of the present study appear to differ from the observations of O’Leary et al. (1986) who also utilized retrograde tracing methods to examine the crossed retino- collicular pathway of the developing rat. After focal deposits of a fluorescent dye (fast blue) into the caudal pole of the superior colliculus of the newborn rat, these investigators found a large number of labeled ganglion cells scattered throughout the contralateral retina, with the topographi- cally appropriate peripheral nasal retina containing the highest density of labeled cells. To estimate the degree of precision in the developing retinocollicular projection, O’Leary and colleagues devised a percentage error measure denoting the number of misprojecting cells for every 100 retinal ganglion cells making the appropriate projection. This is calculated by comparing the peak density of labeled cells in the topographically appropriate retinal region with the density found in the equivalent region in the temporal retina. On average in the newborn rat for every 100 cells projecting to the topographically appropriate portion of the superior colliculus, about 14 cells were estimated to make gross topographic errors. Equivalent calculations from our data yield an error score of 0.2 misprojecting ganglion cells per 100 cells that comprise the retinocollicular pathway of the fetal cat. The difference in the degree of specificity exhibited between the two species can be appreciated by comparing Figure 1E of the present study with Figure 1B of O’Leary et al. (1986); both figures provide photomicro- graphs illustrating the incidence of misprojecting retinocol- licular cells. More recently, Simon and O’Leary (1992) have shown that while retinal axons show no obvious preference for the rostra1 or caudal portions of the superior colliculus, retinocollicular specificity is established by topographically ordered arborizations.

The organization of the developing cat’s retinocollicular pathway also appears to differ from what has been de- scribed recently by Marotte (1993) in the marsupial mam- mal, the wallaby. Relying mainly on retrograde transport of WGA-HRP, this investigator showed a high degree of precision in the crossed retinocollicular pathway of the developing wallaby. In this respect, the present findings in the cat appear similar. At the same time, the small number of inappropriately projecting cells in the wallaby retina

TABLE 2. Ectopic Cell Counts

Age E37 E37 E38 E39 E40 E45 E58 E63

Tracer

Beadr Beads Beads Beads Beads DY WGA-HRP Beadv

Ipsilateral Contralateral

Peak Ectopic Peak Ectopic

1,363 36 1,934 37 1,480 66 1,636 63

-35 17 626 55 884 12 1,666 43 874 55 1,096 62

1,616 35 5,050 96 *O 0 1,449 40

313 39 1,040 99

‘denotes caudal injection

A r

00

ipsi

B 00 contra

Fig. 5. A, B: The location of the retinal core regions, containing the highest density of labeled cells, in two E38 animals after tracer deposit into the caudal portion of the superior colliculus. Note that in both cases labeled cells were concentrated within both the ipsilateral and contralateral nasal retina. The medial-lateral variation in the place- ment of the tracer in the colliculus of the two animals is reflected in a superior-inferior difference in the location of the core retinal regions. The orientation of the retinas and the superior colliculi is as in Figure 4.

bilaterally projecting ganglion cells, but such neurons ap- peared to be very rare at all stages of development (Lia et al., 1987).

The fact that ganglion cells in the nasal retina contribute to the widespread ipsilateral retinocollicular projection seen during early development has been documented in numerous previous studies (e.g., Land and Lund, 1979; Sretavan, 1990; Colello and Guillery, 1990; Thompson and Morgan, 1993). The present results, however, provide the first demonstration that the uncrossed nasal projection innervates the topographically appropriate region of the ipsilateral colliculus.

TOPOGRAPHY IN RETINOCOLLICULAR PATHWAY 303

Marotte, L.R. (1993) Location of retinal ganglion cells contributing to early imprecision in the retinotopic order of the developing projection to the superior colliculus of the wallaby (Macropus eugenii). J. Comp. Neurol. 331; 1-13.

Mesulam, M.M. (1978) Tetramethyl benzadine for horseradish peroxidase neurohistochemistry: a non-carcinogenic blue reaction-product with superior sensitivity for visualizing neural afferents and efferents. J. Histochem. Cytochem. 26,106-117.

O’Leary, D.D.M., J.W. Fawcett, and W.M. Cowan (1986) Topographic targeting errors in the retinocollicular projection and their elimination by selective ganglion cell death. J. Neurosci. 6r3692-3705.

Ostrach, L.H., M.A. Kirby, and L.M. Chalupa (1986) Topographic organiza- tion of retinocollicular projections in the fetal cat. Soc. Neurosci. Abstr. 12:119.

Rakic, P. (1976) Genesis of the dorsal lateral geniculate nucleus in the rhesus monkey: site and time oforigin, kinetics of proliferation, routes of migration and pattern of distribution of neurons. J. Comp. Neurol. I76:23-52.

Raymond, P.A. (1986) Movement of retinal terminals in goldfish tectum predicted by analysis of neuronal proliferation. J. Neurosci. 6:2479- 2488.

Sakaguchi, D.S., and R.K. Murphy (1985) Map formation in the developing Xenopus retinotectal system: an examination of ganglion cell terminal arborizations. J. Neurosci. 5,3228-3245.

Shatz, C.J. (1983) The prenatal development of the cat’s retinogeniculate pathway. J. Neurosci. 3t482499.

Simon, D.K., and D.D.M. O’Leary (1990) Limited topographic specificity in the targeting and branching of mammalian retinal axons. Dev. Biol. 137:125-134.

Simon, D.K., and D.D.M. O’Leary (1992) Responses of retinal axons in uiuo and in uitro to position-encoding molecules in the embryonic superior colliculus. Neuron 9:977-989.

Snider, C.J., and L.M. Chalupa (1993) Specificity of the retinocollicular pathway in developing cat and ferret. SOC. Neurosci. Abstr. 19t54.

Snider, C.J., and L.M. Chalupa (1995) Factors underlying restriction of the early uncrossed retinocollicular projection. SOC. Neurosci. Abstr. 21 :818.

Sretavan, D.W. (1990) Specific routing of retinal ganglion cell axons at the mammalian optic chiasm during embryonic development. J. Neurosci. lOt1995-2007.

Sretavan, D.W., and C.J. Shatz (1987) Axon trajectories and pattern of terminal arborization during the prenatal development of the cat’s retinogeniculate pathway. J. Comp. Neurol. 255,386400.

Steineke, T.C., Q.S. Fischer, P.D. Wilson, and M.A. Kirby (1992) Reorganiza- tion of retinocollicular topography following fetal enucleation in the cat. Soc. Neurosci. Abstr. 18:1312.

Stuermer, C.A.O. (1988) Retinotopic organization of the developing retino- tectal projection in the zebrafish embryo. J. Neurosci. 8:45134530.

Thompson, I.D. (1979) Changes in the uncrossed retinotectal projection after removal of the other eye at birth. Nature 279:63.

Thompson, I.D., and C. Holt (1989) Effects of intraocular tetrodotoxin on the development of the retinocollicular pathway in the Syrian hamster. J. Comp. Neurol. 282;371-388.

Thompson, I.D., and J.E. Morgan (1993) The development of retinal ganglion cell decussation patterns in postnatal pigmented and albino ferrets. Eur. J. Neurosci. 5:341-356.

Thompson, I.D., J.E. Morgan, and Z. Henderson (19931 The effects of monocular enucleation on ganglion cell number and terminal distribu- tion in the ferret’s retinal pathway. European J. Neurosci. 51357-367.

Udin, S.B., and J.W. Fawcett (1988) Formation of topographic maps. Ann. Rev. Neurosci. 11:289-327.

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.

Williams, R.W., M.J. Bastiani, B.L. Lia, and L.M. Chalupa (19861 Growth cones, dying axons and developmental fluctuations in the fiber popula- tion of the cat’s optic nerve. J. Comp. Neurol. 246:32-69.

Yhip, J.P.A., and M.A. Kirby (1990) Topographic organization of the retinocollicular projection in the neonatal rat. Vis. Neurosci. 4:313-329.

were always found to originate from regions adjacent to the retinotopically appropriate area. In contrast, the relatively few number of cells we considered ectopic in the cat were found to be scattered throughout the retina.

As stated earlier, there are several technical concerns in using retrograde tracing methods. Although we did not observe any obvious differences with the three various retrograde tracers employed, it is possible that some of the differences discussed above among mammalian species could be due to procedural variations among laboratories. This point is emphasized by the different estimates of developmental specificity provided by studies of the develop- ing rat’s retinocollicular pathway. Thus, whereas O’Leary et al. (1986) reported a high degree of imprecision in this system (summarized above), a more recent study by Yhip and Kirby (1990) found substantially fewer misprojecting cells in the retinocollicular projection of the same species. For these reasons, it is important to further examine the topographic specificity of the developing carnivore retinocol- licular pathway using anterograde tracings methods. The results of such anterograde tracing experiments on the carnivore retinocollicular pathway (Snider and Chalupa, 1993, 1995, and unpublished) are entirely consistent with the degree of order observed in the present study.

ACKNOWLEDGMENTS We thank Drs. Louis Ostrach and Barry Lia for their

participation in some of the experiments. This study was supported by NIH grant EY03391.

LITERATURE CITED Colello, R.J., and R.W. Guillery (1990) The early development of retinal

ganglion cells with uncrossed axons in the mouse: retinal position and axonal course. Development 108,515-523.

Easter, S.S., Jr., and C.A.O. Stuermer (1984) An evaluation of the hypoth- esis of shifting terminals in goldfish optic tectum. J. Neurosci. 4: 1052-1063.

Fujisawa, H. (1987) Mode of growth of retinal axons within the tectum of Xenopus tadpoles, and implications in the ordered neuronal connections between the retina and the tectum. J. Comp. Neurol. 260,127-139.

Graybiel, A.M. (19751 Anatomical organization of retinotectal afferents in the cat: an autoradiographic study. Brain Res. 96:l-23.

Hanker, J.S., P.E. Yates, C.B. Metz, and A. Rustioni (19771 A new specific sensitive and non-carcinogenic reagent for the demonstration of horse- radish peroxidase. Histochem. J. 9: 789-792.

Harting, J.K., and R.W. Guillery (1976) Organization of retinocollicular pathway in the cat. J. Comp. Neurol. 166:133-144.

Holt, C.E., and W.A. Harris (1983) Order in the initial retinotectal map in Xenopus: a new technique for labeling growth nerve fibers. Nature 301:150-152.

Insausti, R., C. Blakemore, and W.M. Cowan (1984) Ganglion cell death during development of ipsilateral retino-collicular projection in golden hamster. Nature 308:362-365.

Jeffrey, G. (1985) Retinotopic order appears before ocular separation in developing visual pathways. Nature 313:575-576.

Land, P.W., and R.D. Lund (1979) Development of the rat’s uncrossed retinotectal pathway and its relation to plasticity studies. Science 205:698-700.

Lia, B., M.A. Kirby, and L.M. Chalupa (1987) Decussation of retinal ganglion cell projections during prenatal development of the cat. Soc. Neurosci. Abstr. 13:1690.