The retinae of vertebrates project in a topographic manner to several visual centers of the brain. The formation of these projections could depend on the existence of position-specific properties of retinal and target cells. In this study, we have tested the in vitro growth of mouse retinal fibers on membranes derived from various regions of the embryonic superior colliculus, a main target of the retina in this species. Fibers had the choice of elongating on membranes taken from either the anterior or the posterior half of the superior colliculus. Fibers from temporal areas of the retina prefer to elongate on anterior collicular membranes, while fibers from nasal areas do not show a preference. These phenomena are observed with membranes from embryonic (E15–E18) or young postnatal mice. In interspecies cultures where mouse retinal fibers had to grow on chick tectal membranes, or vice versa, the same preference for anterior tectal or collicular membranes in growth of temporal retinal fibers is observed, suggesting some similarities in the cues used in both species.

Guidance of growing axons to appropriate termination sites in their targets could involve a variety of mechanisms. Current evidence suggests that while local synaptic activity-dependent processes could be involved in the refinement of connections (see Purves & Lichtman, 1985 for review), spatial (chemical) markers, carried on the surface of axons and cells in target areas (Sperry, 1963), could be involved in establishing the polarity and a crude ordering of topographic projections at early stages (Fraser, 1980; Gierer, 1981, 1987; Bonhoeffer & Gierer, 1984). This hypothesis is supported by the results of in vivo experiments in lower vertebrates (Harris, 1984; Holt, 1984), and by a series of in vitro experiments on the retinotectal projections in chick. Particularly relevant are experiments showing that axons from temporal sectors of the chick retina can recognize and respond to graded distributions of cellular cues along the anterior-posterior axis of the optic tectum (Bonhoeffer & Huf, 1982). As such cellular components are associated with membranes, have their distribution with the expected polarity in the target area and are present at the relevant developmental stages in the tectum, they could be involved in the ordering of the retinotectal projection in vivo (Walter et al. 1987a,b).

To characterize further these position-related properties, and to study their possible relevance for development of visual projections, we were interested to know if similar properties can be observed in species other than chick and whether similar cellular components are used in different species. It is also of interest to establish whether such in vitro properties bear any resemblance to in vivo properties that determine the detailed patterning of retinotectal projections, which varies slightly among species. In this study, we have examined these points by testing whether retinal fibers from mouse embryos can recognize such position-specific components by giving them the choice to grow on membranes from different regions of the superior colliculus -a major target for these fibers in mammals and a homolog of the avian optic tectum. The results are analogous to those with the chick visual system, in that, when given the choice, fibers from the temporal aspect of the mouse retina prefer to elongate on substrata consisting of cell membranes prepared from anterior, rather than from posterior aspects of the superior colliculus. Furthermore, the same specific choice of growth is made by mouse retinal fibers when cultured on chick tectal cell membranes, and vice versa. Thus, the position-specific retinal and tectal cell components involved in these choices seem to be evolutionarily conserved. An abstract summarizing this work has appeared elsewhere (Godement & Bonhoeffer, 1988).

Materials

Mouse embryos and postnatal mice (all of the C57B1/6 strain) were obtained from a breeding colony maintained in the laboratory. Pregnancies were checked daily, with embryonic day zero (EO) corresponding to the presence of a vaginal plug. White Leghorn chick embryos were obtained from a local dealer. Culture medium was Ham’s F12 supplemented with glutamine (2 DIM), penicillin and streptomycin (10 units ml-1), fetal calf serum (10%), and chicken serum (2%). The medium contained methylcellulose (0·4%), except when specifically mentioned.

Sources of chemicals: F12 medium, fetal calf serum, chick serum (Gibco); neuraminidase inhibitor (2,3-dehydro-2-deoxy-N-acetyl neuraminic acid), aprotinin, leupeptin, pepstatin, rhodamine-B-isothiocyanate, poly-L-lysine (A/r 380000), methylcellulose (Sigma); laminin from EHS mouse tumor (BRL).

Tectal membrane preparation

Pregnant mice were anaesthetized with ether and embryos were aseptically removed and transferred to cold F12 medium. Tissue was taken from 12–18 embryos for each experiment, depending on their age. All steps were performed at 4°C. The mesencephalic roofs of the brains of embryos were dissected out in the same medium and freed of meninges. Tissue corresponding to the inferior colliculus, or just lateral to the superior colliculus, was cut out and discarded. The superior colliculi were bisected (i.e. typically the rostral and caudal halves; in the rest of the paper, these will be referred to as anterior and posterior tectum). These were pooled separately in 1ml of homogenizing buffer (10mm-Tris-HCl, pH 7·4; 1·5mm-CaC12; 1 mm-spermidine) containing inhibitors (50 μM-neuraminidase inhibitor, 200i.u.ml-1 aprotinin, 50 μM-leupeptin, 2 μM-pepstatin) and washed in the same amount of buffer. The tissue was homogenized by slowly passing three times through a syringe needle (400 μm inner diameter, 40 mm long). Each resulting suspension, adjusted to 500μl, was loaded onto a sucrose step gradient (150 μl 50% sucrose, w/w, overlaid by 50 μl 5% sucrose, w/w) and centrifuged at 50000g for 10 minutes, using a Beckmann SW 5OL rotor. The material at the interface between the sucrose solutions, which consists of membrane fragments, was collected and diluted with 1ml of PBS containing inhibitors (same concentrations as for the homogenizing buffer). The membranes were pelleted in a table-top centrifuge and resuspended in 850 μl of the same buffer. The protein concentrations of each membrane suspension were then normalized after measuring the optical density of 15-fold dilutions in 2 % SDS at 220 nm. Concentrations corresponding to an optical density of 0·2 were used; this represented 160–180 μg protein ml-1, as measured by comparison with BSA samples using a modified Bradford method (Simpson & Sonne, 1982).

Similar procedures were used to prepare membrane material from postnatal mice. In the case of P24 mice, only the superficial layers of the superior colliculi were dissected out (excluding part of the stratum opticum and deeper layers).

Chick tectal membranes were taken from E9 embryos and were prepared according to similar methods, except that the tecta were divided in three equal-sized parts, and only the most anterior and posterior parts were used (Walter et al. 1987b).

Preparation of striped membrane carpets

These were prepared according to procedures detailed elsewhere (Walter et al. 1987b). In brief, FITC-labeled microspheres (20 or 40 μl ml-1 of a 1/50 dilution made from the stock suspension; Covaspheres MX, 0 · 3 μ m diameter, Covalent Technology Corp.) were added to one of the two membrane suspensions to make one set of membrane lanes visible by microscopy. The membranes of each type were then sucked onto Nuclepore polycarbonate filters (0·1 μm pore size, Nuclepore Corp.) using a two-step procedure. The first set of membrane lanes is made by applying suction (0·03 bars, 120 s) to a filter overlaid with the desired membrane suspension (150 μl) via a series of about forty parallel microchannels (90 μm wide, 10 mm long, and spaced 90/mi apart) that are encased in a silicon matrix. The second set of lanes is then made by sucking membranes into the empty lanes (i.e. where the pores of the filter are not plugged) in between the first set of lanes, using a fine nylon mesh to apply suction (0-03 bars, 90s). The filters were then washed with PBS and stored on humidified agar plates at 4°C until used (1–3 h later). The lanes prepared first always contain in the end a few more membrane vesicles than the second lanes (Walter et al. 1987b). This asymmetry between the first and the second set of lanes could not be avoided. Therefore in order to exclude artefacts due to this asymmetry many experiments were done with both types of stripe carpets. Those which contained anterior membrane vesicles in the first set of lanes, and posterior in the second set of stripes, and those which had been prepared in the reverse sequence.

Retinal explants

Retinae from chick embryos (E6) were prepared as described previously (Walter et al. 1987b). Retinae from mice embryos were dissected in their entirety into F12 medium. As a landmark for orienting the retina, the optic fissure was used, or a small stitch was made, usually at the dorsal pole of the retina. The retinae were stained in RJTC solution (10 μg ml-1) in F12 medium at 37°C for 15min, and washed in culture medium (without added methocell) at 37 °C in 4% CO2 for 1 – 2 h. They were then flat mounted onto a black (untreated) nitrocellulose filter, oriented, and cut into 300 μm wide strips using a McIlwain type tissue chopper. Retinae from embryos as young as E12 could be prepared, but for these experiments retinae from E14 and E15 embryos were mostly used, from which 6-8 such strips could be made.

Depending on the experiment, two or four retina strips were explanted on each striped membrane carpet. First, a small quantity (circa 50 μl) of culture medium was applied to a filter containing the membrane stripes and it was placed in a 35 mm diameter tissue culture dish (Greiner). Then, the strips holding the retinal explants were laid onto the filter, metal weights were added to hold them in place, and culture medium (2·5 ml) was added together with a half tectum from an embryo of the same age as the retinae. The cultures were incubated for 48 h at 37 °C in a humidified atmosphere containing 4% CO2, then they were fixed (4% paraformaldehyde, 0 · 33 m-sucrose, in pH 7 · 4 PBS), rinsed and dried.

As concluded from initial experiments, some steps during the preparation of explants of mouse retinae were critical. The nitrocellulose filters supporting the explants must not be treated with concanavalin A. Although the retinae would stick better to the filters, ConA treatment interferes with fiber outgrowth in the mouse system. The strips holding the retinal tissue were kept moist, rather than immersed in medium in order to avoid detachment of the explants. To obtain adequate axonal outgrowth the retinal tissue must not be pressed against the Nuclepore filters supporting the membrane carpets. To this effect, a spacer (dialysis membrane, type 20/32, 90 μm thickness, Visking) was used to keep the filter strip that carries the explant at a distance from the striped membrane carpets.

Other procedures

Membranes from other areas of the brain (diencephalon, neocortex) were prepared using methods similar to those for tectal membranes. In some cases, membrane carpets were made using membranes from only one source. For this, the first step in the procedure for making stripes was omitted, and membranes were uniformly applied by aspirating them onto the filter with the nylon mesh (see above). In initial experiments, aimed at testing some general outgrowth capabilities of mouse retinal fibers, laminin was used as a substratum on glass coverslips. The coverslips were first incubated with high molecular weight poly-L-lysine (200 μg ml-1 in F12 medium; overnight at 37°C in 4 % CO2), washed, thoroughly dried and incubated with laminin (20 μg ml-1; 3h at 37°C in 4% CO2). Following this, they were washed twice in F12 medium and used for cultivating the retinae. Laminin was also applied to uncoated coverslips, but extremely high concentrations were needed to obtain outgrowth from the explants.

Observations were carried out using an epifluorescence microscope equipped with filter sets for Rhodamine (to observe fiber outgrowth) and for FTTC fluorescence (to visualize lanes of membranes). Photographs were made using Ilford XPI 400 film.

Outgrowth of fibers from mouse retinae: general observations

Mouse retinal fibers could grow on pure laminin, but extremely high (1 mg ml-1) concentrations of the molecule were needed to coat the coverslips (see also Smalheiser et al. 1984). When laminin (20 μg ml-1) was applied to coverslips that had been first coated with poly-L-lysine and dried, outgrowth was very profuse.

Fibers from embryonic mouse retina can grow very densely on cell membranes taken from a variety of sources: tectum, diencephalon, or neocortex from newborn mice. On membranes from the optic tectum, the axonal front reached a distance of 0·8 – 1·2 mm after being 40–48 h in culture.

Outgrowth was excellent from retinae of embryos aged E12 – E15, and was less extensive from retinae of older embryos. In experiments where dorsal or ventral retinal sectors were cultured (on laminin or on cell membranes from one source), they both extended profuse neurites, although with young (E12, E13) retinae denser outgrowth occurred from the dorsal retina, the reason for which is unknown.

Growth of mouse retinal fibers using cell membranes from the mouse optic tectum

As shown by applying only one type of membranes to the supporting filter, and as reported for chick (Walter et al. 1987b), both anterior and posterior mouse tectal cell membranes were equally supportive of axonal growth from nasal or temporal retina (data not shown).

When retinal fibers were challenged to grow on alternate lanes of anterior and posterior tectal cell membranes, temporal fibers chose to grow on the lanes containing anterior membranes (Fig. 1B,D). This was the case independent of the order of the two-step sequence by which membranes from either area had been laid onto the filter. However, when anterior membranes had been applied first, the retinal fibers appeared to colonize these lanes more extensively than in the reverse case, which is different from the chick system (J. Walter, personal communication).

Fig. 1.

Growth patterns of nasal (A,C) and temporal (B,D) mouse retinal fibers on anterior and posterior tectal cell membranes from mouse. A,B show fibers from an E14 retina on membranes from E16 embryos. C,D show fibers from E15 retina on membranes from E18 embryos. In this and subsequent figures, the location of lanes containing membranes of each kind, a (anterior tectal membranes) or p (posterior), is indicated below each picture together with the appearance of FITC-labeled microspheres used to locate each set of lanes. Anterior membranes were mixed with the microspheres and were applied first (A,B) or second (C,D). Temporal fibers grow along lanes with anterior membranes, nasal grow throughout. Scale, 100 μm.

Fig. 1.

Growth patterns of nasal (A,C) and temporal (B,D) mouse retinal fibers on anterior and posterior tectal cell membranes from mouse. A,B show fibers from an E14 retina on membranes from E16 embryos. C,D show fibers from E15 retina on membranes from E18 embryos. In this and subsequent figures, the location of lanes containing membranes of each kind, a (anterior tectal membranes) or p (posterior), is indicated below each picture together with the appearance of FITC-labeled microspheres used to locate each set of lanes. Anterior membranes were mixed with the microspheres and were applied first (A,B) or second (C,D). Temporal fibers grow along lanes with anterior membranes, nasal grow throughout. Scale, 100 μm.

Nasal fibers did not show any preference for one type of membranes over the other (Fig. 1A,C). They either grew indiscriminately, or with a slight bias for the lanes that had been prepared first, independently of the type of membranes (i.e. anterior or posterior) used for the first lanes.

The difference in growth of nasal and temporal fibers from each source could be detected only when the retinae used for explants had been cut parallel to the dorsoventral axis, such that the retina strips contained tissue either temporal or nasal to the optic disk from both dorsal and ventral retinal sectors. When explants were cut perpendicular to this orientation, such that they contained tissue from both temporal and nasal sectors, either dorsal or ventral to the optic disk, the results were variable. One explanation for this is that, due to the relatively small width of the retinae, some extensive mixing of fibers from nasal and temporal areas could occur within the explants, which could preclude observing each fiber population separately.

The preference of temporal fibers for anterior membranes was observed with tectal cell membrane preparations from embryos of the ages between E15 and E19, which are the stages when retinal fibers attain and then grow into the tectal plate (Edwards et al. 1986; Godement et al. 1984, 1987). Experiments with membranes from postnatal animals were also performed. A preference of temporal fibers was observed for anterior tectal membranes from P0 – P3 mice, but not from P24 or P6 mice (or a very weak choice in the last case).

In one set of experiments, dorsal and ventral retinal fibers were forced to choose between membranes from medial or lateral aspects of the optic tectum. These did not reveal any preference for growth of retinal fibers on one or the other type of membranes.

Partition of the retina

It was of interest to determine more precisely the retinal origins of fibers with the two different properties concerning choice behavior. In mice as in other mammals (c.f. in chick), the retina is partitioned into two adjacent areas of unequal size, one, extending through the nasal retina and part of temporal retina, which gives rise to projections mostly to the contralateral side of the brain, and the other, near the temporal retinal edge, with an early-developing projection to the ipsilateral side of the brain, and later to both sides.

In all experiments, fibers with positive choice behavior were typically obtained from explants taken from any location within the temporal retinal half, of retinae taken from E14, E15, or even E16 embryos. A more detailed analysis was done, comparing outgrowth from explant strips of various widths (200 – 300 μm) taken at various distances from the temporal pole, and laid out in sequence on the supporting filters. Fiber outgrowth in such explants taken from any location within 600 μm of the temporal retinal edge was indistinguishable and the great majority of fibers elongated preferentially on the anterior tectal cell membranes (Fig. 2A,B). Thus, fibers with positive choice ability seem to originate from a wider territory than just the region of retina giving rise to the normal uncrossed projection, which in embryonic mice of slightly older (E16) embryos can be determined to be located near the very edge of the retina (P. Godement, unpublished findings; see Fig. 2, inset). Most fibers growing out of explants consisting of either the dorsal or ventral temporal quadrants also prefered to grow on the lanes with anterior tectal cell membranes. However, outgrowth in explant strips close to the optic disk -but still temporal to it -gave rise to two types of fibers: those that crossed the lanes freely, and those that grew only on the lanes with anterior membranes (Fig. 2C).

Fig. 2.

Growth of retinal fibers from an E15 mouse retina on E18 mouse tectal membranes. Strips of retinal tissue from various excentricities along the temporonasal axis were explanted sequentially on the striped membrane carpet (see diagram in inset, which shows outlines of retina, approximate location of cells with uncrossed axons,-ways of cutting the retina and explanting it). Anterior membranes were marked with F1TC beads and applied second. Growth from stripes A and B is predominantly on anterior tectal cell membranes, while growth from stripe C is mixed, with most fibers growing throughout but a few collecting on lanes with anterior membranes (for each strip, A – C, growth from each side of the retina explant, corresponding to the dark bar, is shown). Fibers from D, in the nasal retina, grow throughout. Scale, 100 μm.

Fig. 2.

Growth of retinal fibers from an E15 mouse retina on E18 mouse tectal membranes. Strips of retinal tissue from various excentricities along the temporonasal axis were explanted sequentially on the striped membrane carpet (see diagram in inset, which shows outlines of retina, approximate location of cells with uncrossed axons,-ways of cutting the retina and explanting it). Anterior membranes were marked with F1TC beads and applied second. Growth from stripes A and B is predominantly on anterior tectal cell membranes, while growth from stripe C is mixed, with most fibers growing throughout but a few collecting on lanes with anterior membranes (for each strip, A – C, growth from each side of the retina explant, corresponding to the dark bar, is shown). Fibers from D, in the nasal retina, grow throughout. Scale, 100 μm.

Interspecies cultures

Cross-species experiments, in which retinal fibers from mouse (or chick) had to grow on lanes of anterior and posterior tectal membranes from chick (or mouse), showed that the preferences in growth of temporal fibers observed for each species also extended across species.

In experiments where chick retinal fibers grew on mouse tectal membranes (Fig. 3), outgrowth was extensive and, while fibers from the nasal parts of the explants grew equally well on both sets of membranes, those on the temporal side (roughly the temporal half) grew almost exclusively on the lanes containing anterior membranes. This was observed with tectal membrane preparations from embryonic (E16) or postnatal (P2) mice.

Fig. 3.

Growth of retinal fibers from nasal (A,C) and temporal (B,D) sectors of E6 chick retina on anterior and posterior tectal cell membranes from E16 (A,B) or postnatal day 2 (C,D) mouse tectal membranes. Anterior membranes are marked with F1TC beads and were applied first. Scale, 100 μm.

Fig. 3.

Growth of retinal fibers from nasal (A,C) and temporal (B,D) sectors of E6 chick retina on anterior and posterior tectal cell membranes from E16 (A,B) or postnatal day 2 (C,D) mouse tectal membranes. Anterior membranes are marked with F1TC beads and were applied first. Scale, 100 μm.

In the converse experiments in which mouse retinal fibers were cultivated on E9 chick tectal membranes (Fig. 4), outgrowth of the mouse retinal explants was less than on mouse membranes. Both nasal and temporal fibers grew well on substrata consisting of only anterior or posterior tectal membranes, although on the latter they tended to grow more fasciculated. When given the choice between anterior and posterior chick tectal membranes, it was clear that nasal fibers grew as well on both sets, while temporal fibers strongly preferred to grow on the lanes with anterior membranes.

Fig. 4.

Growth of retinal fibers from nasal (A,C) and temporal (B,D) sectors of E14 mouse retina on anterior and posterior E9 chick tectal membranes. Anterior membranes were marked with FITC beads and were applied first (A,B) or second (C,D). Scale, 100 μm.

Fig. 4.

Growth of retinal fibers from nasal (A,C) and temporal (B,D) sectors of E14 mouse retina on anterior and posterior E9 chick tectal membranes. Anterior membranes were marked with FITC beads and were applied first (A,B) or second (C,D). Scale, 100 μm.

Recognition of position-related properties by temporal retinal fibers in vitro

These experiments show that, in mouse, as in chick, retinal fibers from temporal and nasal sectors of the retina strongly differ in their ability to choose between, and elongate on, membranes from anterior versus posterior aspects of their target. It should be stressed that, in both species, anterior tectum is normally innervated by fibers from temporal retina, and thus the choices made in vitro by temporal fibers are appropriate to this feature of the in vivo situation in each case. Our experiments also suggest that this difference between anterior and posterior tectal membranes seems to be abolished during the first postnatal week in mouse, the period after which a rough topographic map of the retina has been formed (Schneider & Jhaveri, 1984; O’Leary et al. 1986). A similar phenomenon occurs also in chick (Walter et al. 1987b).

It is highly unlikely that these results reflect a maturational effect, i.e. differences in the timing of outgrowth of fibers from areas temporal or nasal to the median axis in the mouse retina are slight, if at all present (unpublished observations). In this study, we additionally observed similar results for fibers from retinal explants coming from embryos aged E14 or E15 (and even E16 in a few experiments), ages which encompass major phases of axon outgrowth in vivo. Though slight anterior-posterior gradients in the timing of development occur in mammals (DeLong & Sidman, 1962; Altman & Bayer, 1981), it is unlikely that they could be the source of the choices made by temporal fibers: the choices occur with tecta from widely different ages -E16 to around postnatal day 4.

The membranes used for the present experiments came from embryos in which retinal fibers have already reached the superior colliculus. Few fibers, however, are present at E15, the youngest age that we used, and for which a decision of temporal fibers to grow on anterior membranes was seen. In chick, it has been shown that the anteroposterior gradient does not depend on previous innervation by retinal fibers (Bonhoeffer & Huf, 1982; Walter et al. 1987b).

Retinal fibers from one species can use cues on cells from another species

Our experiments also show that, at least with mouse and chick, some degree of similarity in the cues used by retinal fibers must exist: temporal mouse retinal fibers can choose between anterior and posterior tectal membranes, whether they come from mouse or chick, and vice versa. Thus, both in the retina and in the tectum, positional differences exist which must be rather similar in nature and overall distribution in both species. Experiments in the regenerating retinotectal projection in adult goldfish (Vielmetter & Stuermer, 1988, 1989) suggest that, in this system also, temporal but not nasal fibers grow preferentially on anterior tectal membranes from the same species.

The basis for this selective behavior of temporal fibers could a priori be the result of a number of mechanisms: preferential adhesion (Letourneau, 1975), selective repulsion or inhibition (Kapfhammer & Raper, 1987; Walter et al. 1987u), or any mechanism that influences the guidance and motility of the growth cone cytoskeleton. In chick, the preferential growth of temporal fibers on anterior tectal membranes has been shown to be due to the presence of an ‘inhibitory’ or ‘deflecting’ component in the membranes from posterior tectum (Walter et al. 1987a). The same component could be present in mouse tectum.

The existence of similar position-related properties of retinal and tectal cells in mouse and chick, which moreover seem to be due to evolutionarily conserved components, suggests that they could have a basic role in the development of the retinotectal projection. This together with the fact that, in both species, the anterior-posterior tectal gradient is congruent with the polarity of the retinotectal map and is expressed during its development, suggests that these properties could be involved in the ordering of the retinotectal projection in vivo.

Correlations with development in vivo

The choices made by growing temporal retinal fibers in vitro have some correlates in the results of experiments done in vivo. In particular, grafting experiments (Thanos & Diitting, 1987) suggest that chick temporal retinal fibers can in vivo recognize and use a gradient along the anterior-posterior axis of the chick tectum. In mouse uncrossed fibers, mainly from a small sector of the temporal retina, can be made to innervate the visual centers without competing with crossed fibers, by eliminating the latter at an early stage. They then have their heaviest projection to appropriate aspects of the tectum and dorsal geniculate nucleus (Godement et al. 1980, 1981). As suggested by these and previous experiments with tectal membranes in vitro, a basis for this observation could be that their growth is influenced by spatial gradients in the visual centers.

In chick, a sharp line going vertically near the center of the retina appears to bisect the retina into two areas where fibers have different choice behaviors on tectal membrane vesicles (Walter et al. 1987b). This may be related to the fact that fibers from both halves of the retina project to a different hemitectum (Crossland et al. 1974). The only sharp partition that is known to exist in the retina of mammals is that between areas of retina containing ganglion cells with projections to the ipsilateral side of the brain, and areas which do not contain such cells. For mouse, at early stages (prior to about E14 or E15; Drâger, 1985; P. Godement, unpublished findings), nearly all ganglion cells within a small, temporal sector of the retina project to the ipsilateral side of the brain, while in the rest of the retina most cells project contralaterally. Slightly later, cells with a contralateral projection develop in this temporal sector, which in the end outnumber the early ipsilaterally-projecting population of ganglion cells (Drâger, 1985). While the choice abilities of fibers from the two ganglion cell types populating the temporal crescent (i.e. ipsilaterally and contralaterally projecting) were not distinguished in these experiments, the behaviors of fibers on membrane vesicles did not change abruptly, going from one to the other sector of the mouse retina. This is because we found that fibers from temporal and midtemporal aspects of the retina, that is a larger area than that which contains early ganglion cells with ipsilateral projections, have similar preferences for anterior tectal membranes. On the other hand, unlike in chick, a substantial proportion of fibers from central aspects of the temporal hemiretina did not choose between both membrane vesicles. An abrupt transition between the two types of fiber behaviors observed in vitro may occur on either side of a line going through the temporal retina rather than through the center of the retina. There are other alternatives, for example, there could be a smooth rather than abrupt transition between the two types of behaviors across the mouse retina.

Interpretation of characteristics of the retinal partition is delicate in these in vitro observations of fiber growth, which may relate to features other than topography of the visual projections. Nevertheless, further analysis comparing properties of fibers from various retinal sectors in mice or in other mammals with better developed binocular vision might be instructive in that it could reveal possible links between position-related properties of retinal cells observed in vitro and their natural patterns of central terminations.

We thank Anita Goverdhan-Loebbert, Julita Huf, Stephan Kröger, Jost Vielmetter and Jochen Walter for helpful suggestions during this work. We thank Dr Timothy Allsopp for reading the manuscript. We are grateful to Fritz Hegel for taking excellent care of the mouse colony.

Altman
,
J.
&
Bayer
,
S. A.
(
1981
).
Time of origin of neurons of the rat superior colliculus in relation to other components of the visual and visuomotor pathways
.
Expl Brain Res
.
42
,
424
434
.
Bonhoeffer
,
F.
&
Gierer
,
A.
(
1984
).
How do retinal axons find their targets on the tectum?
Trends Neurosci
.
1
,
378
381
.
Bonhoeffer
,
F.
&
Huf
,
J.
(
1982
).
In vitro experiments on axon guidance demonstrating an antero-posterior gradient on the tectum
.
EM BO J
.
1
,
427
431
.
Crossland
,
W. J.
,
Cowan
,
W. M.
,
Rogers
,
L. A.
&
Kelly
,
J. P.
(
1974
).
Specification of the retinotectal projection in the chick
.
J. comp. Neurol
.
155
,
127
164
.
Delong
,
G. R.
&
Sidman
,
R. L.
(
1962
).
Effects of eye removal at birth on histogenesis of mouse superior colliculus. An autoradiographic analysis with tritiated thymidine
.
J. comp. Neurol
.
118
,
205
221
.
Drager
,
U. C.
(
1985
).
Birthdates of retinal ganglion cells giving rise to the crossed and uncrossed optic projections in the mouse
.
Proc. R. Soc. Lond. B
224
,
57
77
.
Edwards
,
M. A.
,
Schneider
,
G. E.
,
Caviness
,
V. S.
, Jr
(
1986
).
Development of the crossed retinocollicular projection in the mouse
.
J. comp. Neurol
.
248
,
410
421
.
Fraser
,
S. E.
(
1980
).
A differential adhesion approach to the patterning of nerve connections
.
Devi Biol
.
79
,
453
464
.
Gierer
,
A.
(
1981
).
Development of projections between areas of the nervous system
.
Biol. Cybernetics
42
,
69
78
.
Gierer
,
A.
(
1987
).
Directional cues for growing axons forming the retinotectal projections
.
Development
101
,
479
489
.
Godement
,
P.
&
Bonhoeffer
,
F.
(
1988
).
Naso-temporal differences in growth of retinal fibers from mouse and chick on tectal cell membranes
.
Eur. J. Neurosci. Suppl
.,
337
.
Godement
,
P.
,
Saillour
,
P.
&
Imbert
,
M.
(
1980
).
The ipsilateral optic pathway to the dorsal lateral geniculate nucleus and superior colliculus in mice with prenatal or postnatal loss of one eye
.
J. comp. Neurol
.
190
,
611
626
.
Godement
,
P.
,
Salaün
,
J.
&
Imbert
,
M.
(
1984
).
Prenatal and postnatal development of retinogeniculate and retinocollicular projections in the mouse
.
J. comp. Neurol
.
230
,
552
575
.
Godement
,
P.
,
Salaün
,
J.
&
Saillour
,
P.
(
1981
).
Absence de projections ipsilatérales après destruction très précoce in utero d’un oeil chez la souris. C
.
r. hebd. Séanc. Acad. Sci. D (Paris)
293
,
625
630
.
Godement
,
P.
,
Vanselow
,
J.
,
Thanos
,
S.
&
Bonhoeffer
,
F.
(
1987
).
A study in developing visual systems with a new method of staining neurones and their processes in fixed tissue
.
Development
101
,
697
713
.
Harris
,
W. A.
(
1984
).
Axonal pathfinding in the absence of normal pathways and impulse activity
.
J. Neurosci
.
4
,
1153
1162
.
Holt
,
C. E.
(
1984
).
Does timing of axon outgrowth influence initial retinotectal topography in Xenopus?
J. Neurosci
.
4
,
1130
1152
.
Kapfhammer
,
J.
&
Raper
,
J. A.
(
1987
).
Collapse of growth cone structure on contact with specific neurites in culture
.
J. Neurosci
.
7
,
201
212
.
Letourneau
,
P. C.
(
1975
).
Cell-to-substratum adhesion and guidance of axonal elongation
.
Devi Biol
.
44
,
92
101
.
O’Leary
,
D. D. M.
,
Fawcett
,
J. A.
&
Cowan
,
M. A.
(
1986
).
Topographic targeting errors in the retinocollicular projection and their elimination by selective ganglion cell death
.
J. Neurosci
.
6
,
3692
3705
.
Purves
,
D.
&
Lichtman
,
J. W.
(
1985
).
Principles of Neural Development
.
Sinauer, Sutherland
.
Schneider
,
G. E.
&
Jhaveri
,
S.
(
1984
).
Rapid postnatal establishment of topography in the hamster retinotectal projection
.
Soc. Neurosci. Abstr
.
10
,
467
.
Simpson
,
I. A.
&
Sonne
,
O.
(
1982
).
A simple, rapid, and sensitive method for measuring protein concentration in subcellular membrane fractions prepared by sucrose density ultracentrifugation
.
Analyt. Biochem
.
119
,
424
427
.
Smalheiser
,
N. R.
,
Crain
,
S. M.
&
Reid
,
L. M.
(
1984
).
Laminin as a substrate for retinal axons in vitro
.
Dev. Brain Res
.
12
,
136
140
.
Sperry
,
R. W.
(
1963
).
Chemoaffinity in the orderly growth of nerve fiber patterns and connections
.
Proc. natn. Acad. Sci. U.S.A
.
50
,
703
710
.
Thanos
,
S.
&
Dotting
,
D.
(
1987
).
Outgrowth and directional specificity of fibers from embryonic retinal transplants in the chick optic tectum
.
Dev. Brain Res
.
32
,
161
179
.
Vielmetter
,
J.
&
Stuermer
,
C. A. O.
(
1988
).
Goldfish retinal axons respond to position-specific properties of tectal cell membranes in-vitro
.
Eur. J. Neurosci. Suppl
.,
337
.
Vielmetter
,
J.
&
Stuermer
,
C. A. O.
(
1989
).
Goldfish retinal axons respond to position-specific properties of tectal membranes in vitro
.
Neuron, in the press
.
Walter
,
J.
,
Henke-Fahle
,
S.
&
Bonhoeffer
,
F.
(
1987a
).
Avoidance of posterior tectal membranes by temporal retinal axons
.
Development
101
,
909
913
.
Walter
,
J.
,
Kern-Veits
,
B.
,
Huf
,
J.
,
Stolze
,
B.
&
Bonhoeffer
,
F.
(
1987b
).
Recognition of position-specific properties of tectal cell membranes by retinal axons in vitro
.
Development
101
,
685
696
.