Axon Guidance in the Mouse Optic Chiasm: Retinal Neurite Inhibition

Developmental Biology 221, 132–147 (2000) doi:10.1006/dbio.2000.9660, available online at http://www.idealibrary.com on

View metadata, citation and similar papers at core.ac.uk brought to you by CORE Axon Guidance in the Mouse Optic Chiasm: provided by Elsevier - Publisher Connector Retinal Neurite Inhibition by Ephrin “A”-Expressing Hypothalamic Cells in Vitro

Riva C. Marcus,* Glennis A. Matthews,* Nicholas W. Gale,† George D. Yancopoulos,† and Carol A. Mason* *Departments of Pathology and Anatomy and Cell Biology, Center for Neurobiology and Behavior, Columbia University College of Physicians and Surgeons, 630 West 168th Street, New York, New York 10032; and †Regeneron Pharmaceuticals, 777 Old Saw Mill River Road, Tarrytown, New York 10591

In the mammalian visual system, retinal axons undergo temporal and spatial rearrangements as they project bilaterally to targets on the brain. Retinal axons cross the neuraxis to form the optic chiasm on the hypothalamus in a position deﬁned by overlapping domains of regulatory gene expression. However, the downstream molecules that direct these processes remain largely unknown. Here we use a novel in vitro paradigm to study possible roles of the Eph family of receptor tyrosine kinases in chiasm formation. In vivo, Eph receptors and their ligands distribute in complex patterns in the retina and hypothalamus. In vitro, retinal axons are inhibited by reaggregates of isolated hypothalamic, but not dorsal diencephalic or cerebellar cells. Furthermore, temporal retinal neurites are more inhibited than nasal neurites by hypothalamic cells. Addition of soluble EphA5-Fc to block Eph “A” subclass interactions decreases both the inhibition and the differential response of retinal neurites by hypothalamic reaggregates. These data show that isolated hypothalamic cells elicit speciﬁc, position-dependent inhibitory responses from retinal neurites in culture. Moreover, these responses are mediated, in part, by Eph interactions. Together with the in vivo distributions, these data suggest possible roles for Eph family members in directing retinal axon growth and/or reorganization during optic chiasm formation. © 2000 Academic Press Key Words: Ephs; ephrins; retinal axon guidance; optic chiasm; hypothalamus; nasal retina; temporal retina.

INTRODUCTION cells mediate general retinal axon patterning and, by selective inhibition, axon divergence (Marcus and Mason, 1995; Mason et al., 1996). During development of the visual pathways, the optic In vitro studies have implicated both contact-mediated chiasm represents an important intermediate target where (Sretavan and Reichardt, 1993; Wang et al., 1995; Wizen- retinal axons undergo several temporal and spatial rear- mann et al., 1993; Godement et al., 1994) and diffusible rangements (Guillery et al., 1995; Colello and Coleman, 1997; Chan and Chung, 1999). In addition, in mammals, (Wang et al., 1996) inhibitory cues in directing retinal axon retinal axons from each eye diverge in the optic chiasm to growth and divergence in the chiasm. Furthermore, the project to targets on both sides of the brain. The molecular coincidence between commissure formation and regulatory and cellular cues that both position the optic chiasm on the protein expression domains has been suggested to form hypothalamus and direct retinal axons have been areas of inhibitory zones that discourage growth cone exploration intense investigation. The chiasm forms on the hypothala- into inappropriate regions (Wilson et al., 1997). However, mus in a position deﬁned by overlapping zones of regulatory the identity of the downstream effector molecules respon- protein expression domains (Marcus et al., 1999). The sible for guiding commissure formation, including the optic region of the developing chiasm is further deﬁned by a zone chiasm, remains largely unknown. occupied by a palisade of radial glia and an early differenti- Recently, a number of different molecular families have ating population of neurons (CD44/SSEA neurons; reviewed been implicated in presenting inhibitory cues to growing in Mason and Sretavan, 1997). We hypothesize that these axons. These include chondroitin sulfate proteoglycans,

netrins, collapsin/semaphorins, Robo-related immuno- Antibody Fusion Protein Binding globulin proteins and their slit ligands, and the Eph family The distributions of the ephrin-A and ephrin-B subclasses were of receptor tyrosine kinases (reviewed in Tessier-Lavigne visualized by receptor–antibody fusion protein binding on 15-␮m- and Goodman, 1996; Varela-Echaverria and Guthrie, 1997; thick unfixed frozen sections and cellular reaggregates, as described Stoeckli and Landmesser, 1998; Guthrie, 1999). To date, at previously (Gale et al., 1996; Marcus et al., 1996a). Receptor– least 14 distinct Eph family members have been identified, antibody fusion proteins consisting of the extracellular portions of and these receptors are divided into two subfamilies based the Eph receptors EphA2 (Eck), EphA5 (Ehk1, Bsk), EphB1 (Elk, on their ability to bind either glycosyl phosphotidylinositol Cek6), and EphB2 (Cek5, Nuk) and the ephrins ephrin-A1 (B61) and (GPI)-anchored (“A” subfamily) or transmembrane (“B” ephrin-B1 (Elk-L) fused to the human IgG1 Fc domain were used ␮ subfamily) ligands (Davis et al., 1994; Gale et al., 1996; Eph (Davis et al., 1994; Gale et al., 1996). Briefly, 15- m-thick sections, Nomenclature Committee, 1997). Repulsive interactions collected on subbed slides, were blocked for 30 min in a solution containing 10% normal goat serum, 2% BSA, 0.02% Na azide between Eph-related receptors and their ligands, the (block). Slides were then incubated in 2 ␮g/ml of receptor or ligand ephrins, have been implicated in several aspects of neural fusion proteins in 0.5ϫ block for 1 h. The tissue was then fixed in development, including the development of topographic fresh 4% paraformaldehyde, washed in PBS, heat treated at 70°C for maps, axonal pathway selection, targeted cell migration, 1 h to destroy endogenous phosphatase activity, and incubated for and the establishment of regional pattern (reviewed in 1 h in goat anti-human IgG alkaline phosphatase-conjugated sec- Flanagan and Vanderhaeghen, 1998; Holder and Klein, 1999; ondary antibody (1:1000 in 0.5ϫ block; Promega). Sections were O’Leary and Wilkinson, 1999). color developed in a solution of NBT and BCIP, followed by fixation In order to identify molecules involved in directing reti- in 4% paraformaldehyde, and mounted. nal axon growth through the developing optic chiasm, we Living cultures were incubated either overnight or for 30 min in ␮ ␮ developed an in vitro paradigm to test for inhibitory inter- 50 g/ml of EphA2-Fc or EphA5-Fc, or 12 g/ml of an Fc control. The concentration of the receptor–antibody fusion proteins used actions between retinal axons and cells isolated from the was that previously determined to be effective in binding available hypothalamus upon which the chiasm forms. This assay ligand (Gale et al., unpublished results). The Fc control concentra- revealed that retinal neurites are specifically inhibited by tion was matched to the protein concentration of the Fc tag used in hypothalamic cells, and that neurites from temporal retina the receptor–antibody fusion proteins. Cultures were washed with are more inhibited than neurites from nasal retina. We next calcium- and magnesium-free PBS, fixed in 4% paraformaldehyde, determined that members of both the “A” and the “B” and processed as for the tissue sections outlined above. subfamilies are present along the retinofugal pathway in patterns that suggest a role in its development. Members of In Situ Hybridization the “A” subfamily of receptors and ligands occupy adjacent in situ domains in the developing hypothalamus, and we further Fifteen micrometer cryostat sections prepared for hybridization using probes for ephrin-A3, -A4, -A5, and -B2 for the studied possible roles of this subfamily by analyzing the analysis in Marcus et al. (1996a), were examined for expression in effects of blocking endogenous receptor–ligand interactions the hypothalamus. in vitro. These experiments demonstrate that endogenous “A” subclass ephrins present on hypothalamic cells con- Culture Methods tribute to both the inhibition and the differential response of retinal neurites by chiasmatic reaggregates, suggesting a Retinal explants. Retinal explants from E14 mouse embryos role for local Eph receptor–ligand interactions in directing were collected in ice-cold DMEM/F12 medium with 15 mM Hepes retinal axon growth during optic chiasm formation. More- buffer (Gibco). For orientation, a small cut was made prior to over, our results, in conjunction with previous reports on dissection, at the ventral point of the diamond-shaped opening of possible roles of Eph family members in the developing the eye cup formed by the pigmented epithelium. Retinae were dissected free from the associated pigmented epithelium, lens, visual system, suggest that specific receptor–ligand interac- vitreous, and vasculature, and a 250-␮m-wide strip of retina incor- tions underlie distinct aspects of visual system pathway porating the periphery of either temporal (DT and VT) or nasal development and organization. retina (DN and VN) was cut. Retinae were plated ganglion cell side down on glass coverslips prepared with polylysine (100 ␮g/ml, Specialty Media) and laminin (20 ␮g/ml, Sigma) and covered with MATERIALS AND METHODS 80–100 ␮l of serum-free medium (SFM: DMEM/F12 supplemented with 1% BSA, 20 units/ml penicillin/streptomycin, and Sigma Experiments were performed on C57Bl/6J mice and Sprague– I-1884 (5 mg/ml insulin, 5 mg/ml transferrin, 5 ␮g/ml sodium Dawley rats obtained from timed-pregnancy breeding colonies. selenite)) containing 0.4% methyl cellulose (Sigma), as described Time of conception was considered midnight before the day a plug previously (Marcus et al., 1996b; Wang et al., 1995). Treated was found and noon the following day 0.5 (0.5). Pregnant mice cultures had 50 ␮g/ml of either EphA2-Fc or EphA5-Fc or 12 ␮g/ml containing E14 embryos were anesthetized with a mixture of of the Fc control diluted directly in SFM-containing methyl cellu- ketamine and xylazine, and the embryos were removed one at a lose. time by cesarean section. Pregnant rats containing E16 embryos Cellular reaggregates. While initial experiments were per-

were sacrificed by CO2 asphyxiation and the embryos removed and formed on retinae and cells isolated from mice, it was difficult to kept on ice until use. P0 and P5 rat pups were sacrificed by distinguish retinal neurites from neurites emanating from the test decapitation. cellular reaggregates. The data reported here derive from cocultures

TABLE 1

Culture condition

Cells: Chiasm. D. dien. Mixed cb Granule c. Chiasm. Chiasm. Chiasm. Treatment: None None None None Fc control EPhA5-Fc EphA2-Fc

Avoidance 58.6 (85) 10.0 (5) 10.0 (9) 5.6 (1) 39.4 (37) 28.8 (36) 25.0 (10) Mixed inhib 28.3 (41) 18.0 (9) 5.6 (5) 5.6 (1) 33.0 (31) 16.0 (20) 42.5 (17) Total inhib 86.9 (126) 28.0 (14) 15.6 (14) 11.1 (2) 72.3 (68) 44.8 (56) 67.5 (27) Permissive 13.1 (19) 72.0 (36) 84.4 (76) 88.9 (16) 27.7 (26) 55.2 (69) 32.5 (13) Ambiguous (3) (6) (0) (1) (3) (5) (0) Total n (148) (56) (90) (19) (97) (130) (40) No. explants (37) (17) (28) (5) (23) (36) (14)

Note. Data represent the percentage of the total number of encounters (minus those scored as ambiguous) for each culture condition. The raw numbers are given in parentheses. Total inhib, avoidance ϩ mixed inhib categories; chiasm., chiasmatic reaggregates; d. dien., dorsal diencephalic reaggregates; mixed cb, mixed cerebellar reaggregates; granule c., puriﬁed cerebellar granule cell reaggregates.

of mouse retina with rat cellular reaggregates; both combinations neurite lengths from explants cultured for 24 h alone or in the gave qualitatively similar results. Four sources of rat cells were presence of 50 ␮g/ml EphA2-Fc or EphA5-Fc or 12 ␮g/ml of the Fc used: late E16 chiasm (including surrounding hypothalamic cells) control. In our analysis “retinal neurites” include processes ema- and dorsal diencephalon, and P4–P5 whole cerebella and isolated nating singly or as bundles from the retinal explants. cerebellar granule cells. E16 brains were dissected free from the For coculture analysis, cultures were blinded to both treatment skull and the region of the developing optic chiasm and hypothala- condition and retinal origin (i.e., nasal or temporal retinal strip mus isolated, as described previously (Wang et al., 1995; Marcus et from either the right or left eye), and the behaviors of retinal al., 1996b). To obtain dorsal diencephalic cells, the cortex was neurites encountering cellular reaggregates were scored by two removed and an approximately 400-␮m-diameter piece of dien- independent observers (Fig. 1). Positive behaviors included those in cephalon just lateral to the dorsal midline was isolated. Tissue which a majority of the retinal neurites appeared to grow over the chunks were collected in DMEM/F12 medium and dissociated reaggregates and their neurites (Fig. 1.3) or those in which retinal using a modification of previously published methods (Alder et al., axons appeared to freely grow onto the cellular portion of the 1996; Wang et al., 1995). Briefly, tissue was sequentially incubated reaggregates (Fig. 1.4). Negative behaviors included an avoidance in medium containing 0.08% trypsin, 0.25% trypsin plus 0.1% category in which retinal neurites stopped short or clearly turned collagenase, and 0.05 mg/ml DNase containing 1 ml of 0.05 mg/ml away from the reaggregates (Fig. 1.1) and a mixed category in which trypsin inhibitor, for 15 min each at 35.5°C. The tissue was less than 50% of the retinal neurites grew freely over the reaggre- dissociated using fire-polished pipets and passed over a 33-␮m gates (Fig. 1.2). An ambiguous category included encounters which Nytex filter. Dissociated whole cerebella and cerebellar granule did not clearly fall in any of the categories or those in which the cells were obtained according to previously published methods two independent observers did not agree (Tables 1 and 2). (Gao et al., 1991; Hatten et al., 1998). Possible effects of the different treatment conditions on neurites Dissociated cells were resuspended at 2–3 ϫ 106 cells/ml in from the four retinal quadrants were analyzed by dividing each medium containing 10% horse serum. Three-hundred to 500 ␮lof retinal explant in thirds and only including those encounters with the suspended cells was added to uncoated Nunc Lab-Tek (Nape- retinal neurites arising from the end thirds of each retinal explant. ville, IL) 7-mm culture wells and incubated overnight at 35.5°C. Only the retinal neurites in the outer thirds were analyzed because The next day, cellular reaggregates were resuspended in a large the middle third was thought to contain a mixture of the two volume of SFM and maintained at 35.5°C for at least 3 h prior to populations. use. Cellular reaggregates (approximately 50–80 ␮m in diameter) A total of 580 encounters between retinal neurites and cellular were transferred in less than 5 ␮l of SFM to coverslips containing reaggregates, from 161 retinal explants, were included in the retinal explants and positioned using fine forceps (see Gao et al., analysis. Statistical significance was evaluated using the ␹2 test. 1991; Hatten et al., 1998). Cocultures were maintained at 35.5°C Figures were assembled with Adobe Photoshop. For some images, for 20–24 h and then fixed for 30 min in 4% paraformaldehyde. brightness or contrast levels were globally altered in order to To unambiguously identify neurites from mouse retinal explants enhance clarity. from cellular processes arising from rat cellular reaggregates, mouse retinal neurites were labeled with a mouse-specific antibody, M6, according to previously published methods (Baird et al., 1992; Lagenaur et al., 1992). Labeled cultures were coverslipped RESULTS with Gelmount (Biomeda Corp.) and photographed on a Zeiss Axiophot microscope. Cellular Reaggregate Cultures As a first step in elucidating the molecules which direct Analysis axon growth through the hypothalamus, we developed an in Possible effects of the receptor–antibody fusion proteins on vitro paradigm to readily assay interactions between retinal retinal neurite outgrowth were evaluated by comparing retinal axons and molecules expressed by endogenous hypotha-

TABLE 2

DT VT DN VN Temporala Nasala

Untreated Avoidance 62.5 (25) 64.5 (20) 36.4 (8) 40.0 (6) 62.2 (56) 40.8 (20) Mixed inhibb 32.5 (13) 29.0 (9) 40.9 (9) 20.0 (3) 32.2 (29) 30.6 (15) Total inhibc 95.0 (38) 93.5 (29) 77.3 (17) 60.0 (9) 94.4 (85) 71.4 (35) Permissived 5.0 (2) 6.5 (2) 22.7 (5) 40.0 (6) 5.6 (5) 28.6 (14) Ambiguous (0) (1) (0) (0) (1) (0) Total n (40) (32) (22) (15) (90) (49) Fc control Avoidance 64.3 (9) 50.0 (7) 31.6 (6) 27.3 (3) 63.2 (24) 25.5 (13) Mixed inhibb 21.4 (3) 35.7 (5) 36.8 (7) 36.4 (4) 26.3 (10) 35.3 (18) Total inhibc 85.7 (12) 85.7 (12) 68.4 (13) 63.6 (7) 89.5 (34) 60.8 (31) Permissived 14.3 (2) 14.3 (2) 31.6 (6) 36.4 (4) 10.5 (4) 39.2 (20) Ambiguous (0) (0) (1) (1) (1) (2) Total n (14) (14) (20) (12) (38) (51) EphA5-Fc Avoidance 36.5 (8) 35.3 (6) 5.6 (1) 46.7 (7) 32.3 (20) 18.8 (13) Mixed inhibb 22.7 (5) 17.6 (3) 33.3 (6) 6.7 (1) 19.4 (12) 16.1 (10) Total inhibc 59.1 (13) 52.9 (9) 38.9 (7) 53.4 (8) 51.6 (32) 37.1 (23) Permissived 40.9 (9) 47.1 (8) 61.1 (11) 46.7 (7) 48.4 (30) 62.9 (39) Ambiguous (2) (0) (2) (0) (3) (1) Total n (24) (17) (20) (15) (62) (62) Dorsal di Avoidance 0.0 (0) 28.6 (2) 14.3 (1) 20.0 (1) 13.0 (3) 8.7 (2) Mixed inhibb 25.0 (2) 0.0 (0) 28.6 (2) 0.0 (0) 17.4 (4) 17.4 (4) Total inhibc 25.0 (2) 28.6 (2) 42.9 (3) 20.0 (1) 30.4 (7) 26.1 (6) Permissived 75.0 (6) 71.4 (5) 57.1 (4) 80.0 (4) 69.6 (16) 73.9 (17) Ambiguous (2) (1) (0) (1) (4) (1) Total n (10) (8) (7) (6) (23) (23)

Note. Data represent the percentage of the total number of encounters (minus those scored as ambiguous). In parentheses are the raw numbers for each condition. DT (dorsotemporal), VT (ventrotemporal), DN (dorsonasal), and VN (ventronasal) are data from the end one-thirds of temporal and nasal explants, respectively (see Materials and Methods for details). Uncrossed axons arise almost exclusively from VT retina. a “Temporal” and “nasal” represent data from entire temporal and nasal retinal explants. b Mixed inhib (mixed inhibitory), Ͻ50% of retinal neurites grew over reaggregates. c Total inhib (total inhibitory) ϭ avoidance ϩ mixed inhib. d Permissive, retinal neurites freely crossed or grew onto reaggregates.

lamic cells (see Marcus et al., 1999, for a detailed explana- retina (dorsoventral (DV), dorsonasal (DN), and ventronasal tion of the anatomical descriptors used in this paper). (VN) retina). Moreover, uncrossed axons avoided groupings of glia and neurons, while crossed axons extended on them (Wang et al., 1995). Retinal Axons Are Speciﬁcally Inhibited by While these studies uncovered several interesting inter- Cellular Reaggregates from the Chiasmatic Region actions between retinal axons and dissociated hypotha- of the Hypothalamus lamic cells, encounters of retinal axons with groupings of Previous studies from our lab revealed that all retinal cells were rare and highly variable. In order to more rigor- neurites are inhibited by dissociated cells from the chias- ously evaluate inhibitory interactions between retinal neu- matic region (Wang et al., 1995; Marcus et al., 1996b). rites and clusters of cells from the chiasmatic region, Furthermore, contralaterally (“crossed”) and ipsilaterally dissociated hypothalamic cells were reaggregated overnight (“uncrossed”) projecting retinal axons responded differen- in serum-containing medium, and then individual reaggre- tially to dissociated chiasm cells. Speciﬁcally, in cocultures gates of uniform size were plated at a distance from the of retinal explants with cells dissociated from the region of retinal explants (Fig. 1). Four different sources of cells were the developing chiasm, uncrossed axons, which primarily used to make cellular reaggregates: dissociated cells iso- arise from the ventrotemporal (VT) retina, grew fewer and lated from the chiasmatic region (“chiasmatic shorter neurites than crossed axons from the rest of the reaggregates”—which include developing hypothalamic

FIG. 1. Schematic diagram of reaggregate cultures. Cellular reaggregates from four different cellular sources were cocultured for 24–28 h with a strip of nasal or temporal retina. Following ﬁxation, cultures were blinded, and the behaviors of retinal neurites to individual reaggregates classiﬁed by two independent observers according to the following criteria: 1(Ϫ), “avoidance,” retinal neurites clearly avoid cellular reaggregates; 2(Ϫ), “mixed inhibitory,”- retinal neurites demonstrate a mixed response of growth over and growth around the cellular reaggregates; 3(ϩ), “growth over,” retinal neurites continue along their original trajectories when encountering cellular reaggregates; 4(ϩ), “growth on,” retinal neurites grow onto cell bodies of cellular reaggregates. Reaggregates from categories 1 and 2 were considered to present negative or inhibitory cues for retinal axon growth, whereas those from categories 3 and 4 were considered to present positive or neutral cues to retinal axons.

neurons and glia), dorsal diencephalon, whole cerebellum, Over 85% of encounters between retinal neurites (from or purified cerebellar granule cells. Retinal neurites were both nasal and temporal strips) and chiasmatic reaggregates unambiguously identified from processes arising from the were considered inhibitory (Table 1; Fig. 3). Of these, the cellular reaggregates by coculturing mouse retinal explants majority (Ͼ67%) were classified as demonstrating avoid- with reaggregates of rat cells and visualizing the retinal ance vs a mixed response. The inhibition of retinal neurites neurites with the mouse neuron-specific antibody, M6, at by chiasmatic reaggregates was significantly greater than the conclusion of the culture period. that observed with reaggregates of dorsal diencephalic Numerous retinal neurites extend radially out from the (28%), mixed cerebellum (16%), or purified granule cells cut edge of retinal explants grown on a planar substrate (11%) (P Ӷ 0.001, Table 1). The majority of retinal neurites coated with polylysine and laminin alone (e.g., see Fig. 2 in grew freely over reaggregates of dorsal diencephalic cells Wang et al., 1995). Neurites emerged from retinal explants (Figs. 2C, 2D and 3), whereas retinal growth cones were singly and as bundles, and both were analyzed. Retinal frequently found on top of cellular reaggregates composed neurites encountering reaggregates of cells from the four of mixed cerebellar or purified cerebellar granule cells (Figs. cellular sources exhibited a range of behaviors. In some 2E, 2F and 3). Thus, retinal axons can detect and respond cases, retinal neurites continued along their original trajec- differentially to cues presented by cellular reaggregates tories when encountering processes from the cellular re- from different sources. Moreover, retinal neurites are spe- aggregates (Figs. 1.3 and 2D). In other cases, retinal growth cifically inhibited by reaggregates of cells from the hypo- cones grew directly onto the cell bodies within the reaggre- thalamus on which the chiasm forms. gates (Figs. 1.4 and 2F). Reaggregates which retinal neurites grew directly onto, or which the majority of retinal axons Temporal and Nasal Retinal Axons Are grew over, were considered to present either positive or Differentially Affected by Chiasmatic neutral cues to retinal axon growth. These behaviors were Reaggregates in contrast to cases in which retinal neurites clearly avoided (Fig. 1.1 and 2B) or demonstrated a mixed response Both inhibitory and permissive cues associated with a of growth over and growth around (Figs. 1.2 and 2A) the zone centered around the chiasmatic midline are impli- cellular reaggregates. In cases showing a mixed response, cated in directing retinal axon divergence (Godement et al., retinal neurites growing through or near the reaggregates 1994; Sretavan et al., 1994; Marcus and Mason, 1995; Wang frequently appeared to form a small number of tight, thick et al., 1995). In addition, a diffusible signal(s) from the fascicles, reminiscent of the response of retinal neurites chiasmatic region suppresses the growth of all optic axons, grown on dissociated chiasm cells (Wang et al., 1995). regardless of their retinal origin (Wang et al., 1996). To Reaggregates that elicited the latter types of behaviors were determine whether crossed and uncrossed axons were dif- considered to present negative or inhibitory cues for retinal ferentially or equally affected by the inhibition uncovered axon growth. in the in vitro assay described above, we compared the

FIG. 2. Retinal neurite responses to cellular reaggregates. Low (A, C, E) and high (B, D, F) power photographs of retinal explants cocultured with chiasmatic (A, B), dorsal diencephalic (C, D), or mixed cerebellar (E, F, arrow) reaggregates. Retinal neurites avoid or demonstrate a mixed inhibitory response to chiasmatic reaggregates. In contrast, retinal neurites grow over or onto cellular reaggregates composed of dorsal diencephalic or mixed cerebellar cells.

behaviors of uncrossed retinal neurites arising from VT temporal vs nasal distribution. Over 90% of encounters retina with crossed neurites arising from DT, DN, and VN between chiasmatic reaggregates and temporal (both dorsal retina (see Materials and Methods). and ventral) retinal neurites were inhibitory, compared to Comparison of the behaviors of retinal neurites from the about 70% of encounters with nasal retinal neurites (P Ӷ four retinal quadrants revealed that the degree of retinal 0.001). Several explanations can account for the difference neurite inhibition was not equal across the retina (Table 2; between these results and our former study. First, in our Fig. 4). These differences did not correspond to an uncrossed former study, retinal neurites growing among dissociated (ventrotemporal) vs a crossed retinal origin as in our previ- chiasm cells may be primed to respond differentially to ous in vitro model (Wang et al., 1995), but rather reﬂected a larger clusters of chiasmatic cells. Alternatively, in the

mentary domains in the region of the developing optic chiasm (Fig. 5). Both EphA2-Fc and EphA5-Fc demonstrated strong labeling just ventral to the optic chiasm (Fig. 5a, C, F, and I). This region overlaps with previously described territories of gene expression and an early differentiating population of neurons implicated in guiding retinal axon growth through the hypothalamus (Fig. 5b, eЈ) (Marcus and Mason, 1995; Marcus et al., 1999). In contrast, ephrin-A1-Fc binding revealed the presence of EphA receptors within the region of the optic chiasm, in a pattern reminiscent of a previously described radial glial palisade that straddles the chiasmatic midline (Fig. 5a, B, E, and H; and Fig. 5b, eЈ and fЈ; see also Fig. 2 in Marcus et al., 1995). The complementary distributions of the “A” subclass receptors and ligands were already present at E11.5– E12.5, prior to the first retinal axon ingrowth into the hypothalamus (Fig. 5a, B and C), indicating binding sites FIG. 3. Retinal neurites are specifically inhibited by chiasmatic independent of those present on the retinal axons them- reaggregates. Retinal neurite encounters with reaggregates from selves. Little or no binding was detectable in the optic each of the four cellular sources were scored, and the percentage of stalks. The localization of members of ephrin-A ligands in encounters between retinal neurites and individual cellular re- the region of the developing hypothalamus is in agreement aggregates that demonstrated an inhibitory response was calcu- with previous reports in both mouse (Donoghue et al., 1996; lated. Zhang et al., 1996) and zebrafish (Brennan et al., 1997; Macdonald et al., 1997). Ephrin-B1-Fc bound two domains in the developing hy- present study, inclusion of serum in the medium (necessary for the formation of cellular reaggregates) may have altered the expression of cell surface molecules required for elicit- ing differential behaviors of crossed and uncrossed retinal neurites. Nevertheless, the present assay provides a highly reproducible and easily quantifiable method for measuring position-dependent retinal neurite responses to chiasmatic cells. Because the cultures are grown on a planar substrate in a large volume of medium, these responses most likely represent contact-mediated, or short-range, diffusible cues.

Eph Receptors and Their Ligands Are Distributed along the Retinofugal Pathway Recently, a number of molecular families with inhibitory properties have been uncovered. We investigated whether one of these, the Eph family of receptor tyrosine kinases and their ligands, the ephrins, might contribute to the inhibitory response of retinal axons to chiasmatic reaggregates, by FIG. 4. Neurites from temporal and nasal retina respond differen- determining the distributions of Eph family members in the tially to chiasmatic reaggregates. Graph of the percentage of mouse during the period of optic chiasm formation. inhibitory encounters between retinal neurites and chiasmatic In a previous study, we used receptor– and ligand– reaggregates by retinal quadrant. Data are presented for both antibody fusion proteins to localize their corresponding untreated cultures and cultures grown in the presence of soluble Fc ligands and receptors in the developing mouse retina (Mar- protein. Numbers above bars represent total number of encoun- cus et al., 1996a). These studies demonstrated that recep- tered scored for each condition. Retinal neurites from both temporal quadrants (dorsotemporal, DT; ventrotemporal, VT) are inhib- tors and ligands from either the “A” and the “B” speciﬁcity ited more than retinal neurites from the nasal quadrants subclasses distribute in opposing gradients along the tem- (dorsonasal, DN; ventronasal, VN). Crossed axons from DT retina poronasal or dorsoventral axes by E10.5–E11. Analysis of and uncrossed retinal axons from VT retina are affected equally antibody fusion protein binding in the hypothalamus dur- indicating that the inhibitory signals associated with chiasmatic ing the period of retinal axon growth revealed that “A” reaggregates do not differentially affect crossed and uncrossed subclass receptors and ligands occupy adjacent, comple- retinal axons.

FIG. 5. Eph-related receptors and ephrins in the retinofugal pathway. (a) Ephrin-B1-Fc (A, D, G), ephrin-A1-Fc (B, E, H), and Eph-A5-Fc (C, F, I) binding in the hypothalamus of E11.5 (A–C), E12.5 (D–F), and E14.5 (G–I) mice. Labels in the vertical axis indicate the detected proteins. Planes of sections in A–F and I are “horizontal” and G and H are “frontal,” as described in Marcus et al. (1995). “A” subclass receptors and ligands distribute in complementary domains (e.g., E, F). Both “A” and “B” subclass receptors colocalize within the midline region occupied by the previously described radial glial palisade (G, H) (see Marcus et al., 1995). “A” subclass ligands as well as “B” subclass receptors are distributed in the region caudal to the chiasm (e.g., D, F, I). Dotted line in I demarcates rostral border of the chiasm. (b) Nissl-stained sections of E14.5 brain in horizontal (aЈ) and frontal (cЈ) planes, as in Marcus et al. (1995). (bЈ and dЈ) Enlargements of aЈ and cЈ, showing in situ hybridization. (bЈ). As in a, C, F, and I, ephrin-A5 is expressed broadly in the hypothalamus ventral and caudal to the chiasm (compare to a, F and I). (dЈ). Ephrin-B2 expression is prominent in cells ﬂanking the midline. (eЈ,fЈ). Schematic horizontal (eЈ) and frontal (fЈ) views of chiasmatic region indicating midline radial glial palisade (dots), early differentiating neurons (hatched area), and trajectory of uncrossed (from ventrotemporal (vt) retina) and crossed retinal axons. Bar (in a, I): a, A–C, 200 ␮m; D–F, 100 ␮m; G–I, 50 ␮m; b, aЈ, 150␮m; bЈ and dЈ, 100 ␮m; cЈ, 250 ␮m.

pothalamus. The ﬁrst overlapped with the distribution of speciﬁcity subclasses colocalize with the chiasmatic glial the ephrin-A1-Fc binding straddling the chiasmatic mid- palisade (Fig. 5a, G and H; Fig. 5b, fЈ). The second domain line, suggesting that receptors of both the “A” and “B” (Fig. 5a, A and D) overlapped in part with the region labeled

Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved. 140 Marcus et al. by the EphA5-Fc antibody fusion protein and contains the control for possible inhibitory-releasing effects of the Fc tag, early born population of CD44/SSEA neurons (Fig. 5b, eЈ) cocultures were incubated in the presence of soluble Fc (Marcus et al., 1999; Mason and Sretavan, 1997). In con- protein added in a concentration that matched that of the Fc trast, no labeling was observed with either an EphB1– or tag present on the antibody fusion proteins (Figs. 6A and EphB2–antibody fusion protein. 6B). Previously we used probes specific for ephrin-A3, -A4, Addition of EphA5-Fc, EphA2-Fc, or the Fc control did -A5, and -B2 to determine the specific identities of retinal not noticeably affect the numbers or lengths of retinal ligands revealed by receptor–antibody fusion binding (Mar- neurites grown on polylysine/laminin alone (neurite cus et al., 1996a). Inspection of those data revealed high lengths: untreated, 1013 Ϯ 196 ␮m, n ϭ 12; EphA5-Fc levels of ephrin-A5 in the hypothalamus (Fig. 5b, aЈ and bЈ). treated, 1013 Ϯ 186 ␮m, n ϭ 10; EphA2-Fc treated, 955 Ϯ Interestingly, in contrast to the lack of binding with either 182 ␮m, n ϭ 6). In contrast, addition of EphA5-Fc, but not an EphB1-Fc or EphB2-Fc receptor antibody–fusion protein, EphA2-Fc, significantly decreased the frequency of inhibi- hybridization with an ephrin-B2 probe revealed expression tory responses of neurites from all poles of the retina to flanking the midline of the developing chiasm (Fig. 5b, cЈ, chiasmatic reaggregates (P Ͻ 0.001, Table 1). Fc addition to dЈ, and fЈ; Nakagawa et al., 2000). Ephrin-B2 labeling is the cocultures resulted in a small, but significant, decrease concentrated in the region overlapping the radial glial in the number of inhibitory responses of retinal neurites palisade, coincident with the expression of EphA and EphB (P Ͻ 0.001, Table 2). However, this decrease could not receptors revealed by antibody fusion protein binding. The account for the reduced inhibition observed in the presence overlapping expression of “B” subclass receptors and li- of EphA5-Fc (P Ͻ 0.001 vs Fc control). Thus, signaling by gands may explain the discrepancy between the ephrin-B ephrins contributes to the inhibition of retinal neurites by expression revealed by binding studies and in situ hybrid- chiasmatic reaggregates. ization, as ligands may be masked by endogenous receptors We next investigated whether “A” subclass ephrins con- in sites where overlaps occur (Sobiezczuk and Wilkinson, tribute to the differential response of temporal vs nasal 1999). Further in situ hybridization studies will be neces- neurites to untreated chiasmatic reaggregates by determin- sary to precisely determine the distributions of individual ing the frequency of inhibitory responses from entire tem- receptors and ligands present in the hypothalamus. poral or nasal retinal strips in the presence of EphA5-Fc (see Materials and Methods). Addition of EphA5-Fc decreased the inhibition of temporal and nasal neurites by 42.8 and “A” Subclass Ephrins Contribute to Retinal 34.3%, respectively, when compared to untreated controls Neurite Inhibition by Chiasmatic Reaggregates (Table 2, Figs. 4 and 7). If signaling via Eph receptors is Several studies suggest that growth cone extension is responsible for the greater inhibition of temporal vs nasal regulated by domains of regulatory genes expressed in the neurites by chiasmatic reaggregates, blocking this signal early neuroepithelium (see Wilson et al., 1997; Marcus et would cause nasal and temporal neurites to respond equally al., 1999). In particular, many commissures appear to form to the cellular reaggregates. In support of this hypothesis, in locations predicted by different expression domains. when the data from Fig. 7 are expressed as the difference in However, it remains unclear which, if any, of the guidance the percentage of inhibition of temporal vs nasal retinal molecules lying upstream of these regulatory proteins in- neurites, the disparity in the amount of inhibition between fluence growth cone extension. “A” subclass ephrins are temporal and nasal neurites (23%) was reduced in the expressed along the ventral aspect of the developing chi- presence of EphA5-Fc (14.5%) (Fig. 8). asm. This region coincides with previously described terri- Although addition of EphA5-Fc reduced the inhibition tories of regulatory gene expression (Marcus et al., 1999), seen by all four retinal quadrants, it did not differentially prompting us to investigate whether endogenous hypotha- affect uncrossed retinal axons from VT retina vs crossed lamic ephrins can affect retinal neurite extension. axons arising from DT retina (Table 2). Furthermore, com- First we determined whether “A” class ephrins continued parison of EphA5-Fc-treated cultures with control cultures to be expressed in the cultures. Both EphA2-Fc and containing dorsal diencephalic reaggregates indicates that EphA5-Fc bound to chiasmatic reaggregates in culture (Figs. the addition of EphA5-Fc does not totally block the inhibi- 6E–6H). Little or no binding was detected on the retinal tory effects of chiasmatic reaggregates on temporal or nasal neurites (data not shown). In vivo, “A” subclass receptors neurites (Fig. 7). Several factors could account for why this and ligands demonstrate complementary temporal–nasal difference was not totally abolished. First, the addition of distributions (Marcus et al., 1996a). The apparent lack of exogenous EphA5-Fc may have incompletely blocked avail- retinal binding could reflect either a low number of ligands able ligand on the hypothalamic cells or the retinal axons on the retinal axons or that existing ligands are masked by themselves. Second, despite demonstrations of promiscu- binding to endogenous receptors on the same or adjacent ous binding between receptors and ligands of a given retinal axons. Next we added either EphA2– or EphA5– subclass in vitro, it is unclear whether such promiscuous receptor–antibody fusion proteins to disrupt potential binding occurs in vivo. Consistent with this idea, we found receptor–ligand interactions in cocultures of retinal ex- that EphA2-Fc was much less effective than EphA5-Fc in plants with chiasmatic reaggregates (Figs. 6C and 6D). To blocking inhibitory interactions between retinal neurites

FIG. 6. Retinal neurites are inhibited by endogenous ephrins on chiasmatic reaggregates. Low (A, C) and high power (B, D) photographs of retinal explants cocultured with chiasmatic reaggregates in the presence of either soluble Fc (A, C) or EphA5-Fc (B, D). Retinal neurite inhibition by chiasmatic reaggregates is reduced in the presence of EphA5-Fc. EphA5-Fc antibody fusion protein bound to chiasmatic reaggregates in culture (E, F), indicating expression of EphA5 on these reaggregates, whereas Fc alone did not (G, H).

and chiasmatic reaggregates, even though both EphA2-Fc TOPAP (Savitt et al., 1995). Nevertheless, the reduction in and EphA5-Fc were equally effective in localizing ephrin the number of inhibitory responses of retinal neurites to distributions in the hypothalamus (see also Krull et al., chiasmatic reaggregates is consistent with the addition of 1997). Third, other spatially restricted molecules could also soluble EphA5-Fc in blocking inhibitory (or repulsive) contribute to the different behaviors of temporal and nasal receptor–ligand interactions between endogenous retinal neurites in response to chiasmatic reaggregates. A growing axons and hypothalamic cells. list of molecules with distinct temporal–nasal and dorsal– ventral distributions have been isolated (reviewed in Kap- rielian and Patterson, 1994). Molecules with distinct DISCUSSION temporal–nasal distributions include Eph-related receptors and ephrins (Cheng et al., 1995; Marcus et al., 1996a), the Previous studies have implicated inhibitory signals in winged helix transcription factors BF-1 and BF-2 (Hatini et directing retinal axon growth and divergence through the al, 1994; Huh et al., 1999), the homeobox containing gene hypothalamus during optic chiasm formation. Here we SOHO-1 (Deitcher et al., 1994), and the cell surface protein developed a novel in vitro assay to show that retinal

cated Sek1 receptor in zebrafish and Xenopus embryos results in abnormal hindbrain segmentation, possibly through restricting the movement of cells between adjacent rhombomeres (Xu et al., 1995; Mellitzer et al., 1999). Likewise, repulsive receptor–ligand interactions appear to be necessary for restricting neural crest cell migration in the trunk and branchial arches (Krull et al., 1997; Smith et al., 1997). The recent identification of Krox-20 as a direct transcriptional activator of EphA4 suggests one mechanism by which the identity and movement of cells are coupled to produce sharply restricted segmental domains (Theil et al., 1998). Studies of regulatory gene expression patterns suggest that the forebrain, like the hindbrain and spinal cord, consists of a series of transverse and longitudinally orga- nized domains (Figdor and Stern, 1993; Puelles and Ruben- FIG. 7. Addition of EphA5-Fc releases, in part, the inhibition of retinal neurite inhibition by chiasmatic reaggregates. EphA5-Fc stein, 1993). Furthermore, commissures frequently form at addition significantly decreases the amount of retinal neurite locations predicted by domains of regulatory gene expres- inhibition by chiasmatic reaggregates compared to untreated or sion (see Wilson et al., 1997). Interestingly, the complemen- Fc-treated cocultures. EphA5-Fc-treated retinal neurites are still tary domains of “A” subclass receptors and ligands we more inhibited by chiasmatic reaggregates than dorsal diencephalic observed in the hypothalamus correspond to regulatory reaggregates, suggesting the presence of additional inhibitory cues gene expression domains implicated in patterning retinal on the hypothalamic cells. Numbers above bars represent total axon growth through this region (Marcus et al., 1999) and, number of encountered scored for each condition. as such, might be downstream targets of these genes. Related to these findings, expression of a truncated Sek1 receptor in the zebrafish embryo results in the misspecifi- cation of forebrain structures, including the eyes and the neurites are specifically inhibited by signals associated with ventral forebrain (Xu et al., 1996). Together, these results reaggregates of chiasmatic cells and that temporal retinal suggest that Eph family members may play an early devel- neurites are more inhibited than nasal ones. Using a com- opmental role in patterning the hypothalamus on which the bination of receptor–antibody fusion protein binding and in optic chiasm forms. situ hybridization we demonstrate that both “A” and “B” In addition to the complementary patterns of “A” sub- subclass receptors and ligands occupy the hypothalamus in patterns that implicate them in directing retinal axon growth. “A” subclass ligands are expressed ventral to the developing chiasm, and we demonstrate that disrupting signaling between “A” subclass receptors and ligands contributes both to the inhibition and the differential response of retinal neurites to chiasmatic cells. Eph family interactions have been implicated in several developmental processes including the establishment of regional pattern, axon guidance, and the encoding of positional labels. Below we discuss how our results implicate this family in similar roles during optic chiasm development.

Eph Family Members and the Establishment of Regional Pattern Using antibody fusion proteins we demonstrate that “A” subclass receptors and ligands occupy adjacent, complementary domains in the hypothalamus. Eph-related recep- FIG. 8. Ephrins contribute to the differential response of temporal tors and their ligands are expressed in adjacent domains and nasal neurites to chiasmatic reaggregates. Graph comparing the throughout the mouse embryo (Flenniken et al., 1996; Gale percentage difference in the inhibition of temporal and nasal retinal et al., 1996), suggesting that signaling at the interface neurites in control and EphA5-Fc-treated cultures. In the presence between adjacent domains contributes to the segmental of EphA5-Fc the difference in the response of temporal and nasal organization of the developing embryo (Holder and Klein, neurites to chiasmatic reaggregates is reduced to roughly half that 1999). Consistent with such a role, expression of a trun- seen in untreated or Fc-treated cocultures.

Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved. Ephs and Ephrins in the Mouse Optic Chiasm 143 class receptors and ligands in the hypothalamus, we de- Thus the addition of exogenous EphA5-Fc clearly bound to tected a subregion of EphB expression within the ephrin-A- ligand on the chiasmatic reaggregates, although we cannot expressing domain and a region of overlapping expression of rule out additional effects of the EphA5-Fc treatment on the “A” and “B” receptors with ephrin-B2 centered on the retinal axons themselves. chiasmatic midline. Thus, receptors and ligands are not In agreement with the present findings, analyses of ax- restricted to mutually exclusive domains. Sites of overlap- onal defects in EphA8 mutant mice suggest how ephrin-rich ping expression have also been detected by in situ hybrid- territories may exclude axons, thereby forcing their growth ization in somites and the branchial arches (Wang and across the midline (Frisen and Barbacid, 1997; Park et al., Anderson, 1997). The visualization of ephrin-B2 by in situ 1997). In normal mice, EphA8-expressing neurons in the hybridization, but not by receptor–antibody fusion protein superior colliculus send axons toward the inferior collicu- binding, is consistent with masking of ligand by endoge- lus that expresses high levels of ephrin-A2 and ephrin-A5. nous receptors in sites where overlaps occur (see also Sefton At the border between the superior and inferior colliculi, et al., 1997; Sobiezczuk and Wilkinson, 1999). EphA8-expressing axons turn and cross the midline. In mutant animals, neurons that normally express EphA8 receptors cannot respond to ephrins present in the inferior Eph Family Interactions and Axon Guidance colliculus and instead project their axons ipsilaterally to- Eph-related receptors and their ligands have also been ward the spinal cord. implicated as axon guidance molecules in several systems Targeted disruption of EphB2 also leads to defects in including the topographic mapping of retinotectal (Cheng et commissure formation (Henkemeyer et al., 1996). In mu- al., 1995; Drescher et al., 1995), retinogeniculate (Feldheim tant mice, axons forming the posterior part of the anterior et al., 1998), and hippocamposeptal (Gao et al., 1996; Zhang commissure do not cross the midline, but rather penetrate et al., 1996) projections; axon fasciculation (Winslow et al., into the ventral forebrain. Interestingly, the misdirected 1995); the development of cortical (Castellani et al., 1998) axons express ephrin-B ligands, suggesting that ephrin-B and hippocampal (Stein et al., 1999) circuits; guidance of ligands on axons in the anterior commissure transduce an sensory and motor axons (Gao et al., 1998; Ohta et al., 1997; inhibitory signal in response to receptors along their path- Wang and Anderson, 1997); and the establishment of tracts way. The colocalization of Eph receptors with the region of across the midline (Henkemeyer et al., 1996; Orioli et al., the radial glial palisade in the hypothalamus suggests that 1996; Park et al., 1997). Here we demonstrate that Eph- similar interactions may act to guide ephrin-expressing related receptors and ligands are present both in the retina retinal axons. and in the hypothalamus upon which the chiasm forms and Addition of soluble EphA5-Fc did not reduce the inhibi- that “A” subclass interactions contribute to the inhibition tion of retinal neurites by chiasmatic reaggregates to con- of retinal neurite growth by hypothalamic cells in vitro. trol levels, suggesting that other inhibitory cues are still Addition of EphA5-Fc to our cultures could, therefore, exert present in the cocultures. One candidate is CD44, a cell its effects at the hypothalamic or retinal levels. surface molecule present in the hematopoietic system and In vivo, “A” subclass ephrins occupy a domain just on a population of early differentiating hypothalamic neu- ventral to the optic chiasm, where they may restrict retinal rons (reviewed in Mason and Sretavan, 1997). CD44 appears axons from growing into inappropriate regions of the hypo- to inhibit retinal axon growth in vitro (Sretavan et al., thalamus. Consistent with such a role, ephrin-A5, which is 1994). In addition, the hypothalamus secretes an inhibitory known to be expressed in the hypothalamus (Fig. 5b and see diffusible cue (Wang et al., 1996; Tuttle et al., 1998) that we Donoghue et al., 1996; Zhang et al., 1996), is implicated in propose may function as a general guidance mechanism in inhibiting retinal axon growth into the inferior colliculus in intermediate targets to prime growth cones to perceive both zebrafish (Brennan et al., 1997) and mice (Frisen et al., other, more specific cues (Wang et al., 1996). This diffusible 1998) and to exclude limbic thalamic afferents from inner- cue is unlikely to involve signaling via Eph-related recep- vating sensorimotor cortex (Gao et al., 1998). Ligands are tors since receptors are not activated by ephrins presented also present in a high nasal to low temporal gradient in the in soluble form (Davis et al., 1994). retina, and recent experiments in chick demonstrate that altering the retinal expression of “A” subclass ephrins The Eph Family and the Encoding of Positional modulates retinal ganglion cell responses to ephrins ex- Labels pressed as a growth substrate or in the tectum (Hornberger et al., 1999). In these experiments, retinal axons were more In addition to roles in patterning and axon guidance, responsive to the repellent action of ephrin-A5 when “A” Eph-related receptors are implicated in encoding positional subclass ephrins were removed from the retinal axons. information, including the proper mapping of Eph- Addition of EphA5-Fc in our cultures would likewise be expressing retinal axons in their targets (reviewed in Flana- expected to lower the amount of ligand available and gan and Vanderhaeghen, 1998; O’Leary and Wilkinson, increase the pool of free retinal receptors and sensitivity to 1999). Prior to reaching their targets, however, retinal axons external cues. However, we observed that the inhibition of undergo several spatial and temporal rearrangements in the retinal axons in our cultures was lessened, not enhanced. region of the optic chiasm (reviewed in Guillery et al.,

1995). Thus, proper position-dependent sorting of retinal plicating these interactions in the position-dependent sort- axons at the chiasm is essential for the formation of ing of retinal axons along their path. topographically precise projections. In the mammalian optic chiasm, retinal axons sort out in a position-dependent manner to project to targets on both SUMMARY the ipsilateral and the contralateral sides of the brain (Chan and Chung, 1999). Previously we hypothesized that the Recently isolated mutations in zebrafish indicate that decision to cross or not cross the midline may be mediated axon patterning in the visual system is regulated by genes by a threshold response of uncrossed axons from ventrotem- that act in different parts of the optic pathway (Baier et al., poral retina to inhibitory cues associated with the chias- 1996; Karlstrom et al., 1996; Trowe et al., 1996). Eph- matic midline (Mason et al., 1996). Eph family receptors related receptors and ephrins are distributed throughout the and their ligands from each of the two specificity subclasses retinofugal pathway, making them prime candidates for distribute in opposing gradients in the developing mouse regulating the topographic organization of the visual sys- retina (Marcus et al., 1996a). The orientation of these tem. Eph family members are also implicated in the map- gradients is such that the highest concentration of receptors ping of retinal axons on the tectum (reviewed in Drescher et from both the “A” and “B” subclasses occupies the ventro- al., 1997; Flanagan and Vanderhaeghen, 1998) and the temporal portion of the retina (see also Connor et al., 1998), lateral geniculate nucleus (Feldheim et al., 1998), in in- raising the question of whether signaling through these traretinal (Marcus et al., 1996a; Braisted et al., 1997; Holash receptors might contribute to the sorting of retinal axons at et al., 1997; Sefton et al., 1997; Hornberger et al., 1999) and et al., et al., the optic chiasm. In our experiments, crossed and un- intratectal (Braisted 1997; Connor 1998) organization, and in the development of thalamocortical projec- crossed axons from DT and VT retina were equally affected tions (Gao et al., 1998). Here we suggest that Eph family by addition of EphA5-Fc (Table 2), suggesting that Eph “A” interactions help define a developmental domain that di- family interactions alone are insufficient to direct the rects retinal axon growth during chiasm formation by divergence of ipsilaterally and contralaterally projecting restricting axons from entering inappropriate regions of the retinal axons in the chiasm. In contrast, ectopic expression brain and/or contributing to position-related resorting of of ephrin “A” ligands leads to an increased ipsilateral optic fibers. These findings suggest that specific combina- projection in chick (Dutting et al., 1999). The reason for this tions of Eph-related receptors and ephrins may act at increase is unclear. As shown previously, retinal axons multiple steps along a given pathway. avoid ectopically expressed ephrins in the tectum (Naka- moto et al., 1996). Retinal axons may likewise avoid ectopic patches of ephrins in the chiasm, resulting in the growth of ACKNOWLEDGMENTS retinal axons toward the ipsilateral optic tract. Alterna- tively the modulation of retinal receptors by ectopically We thank Drs. Pierre Godement, Mary Morrison, and Lynda expressed ephrins in the retina may lead to errors in the Erskine for their help during the experiments and comments on the position-dependent responses of retinal axons to chiasmatic manuscript; Drs. Christine Holt and Shinichi Nakagawa for allow- cues (see below). Ephrin-B2 colocalizes with the radial glial ing us to cite their unpublished results and for fruitful conversa- palisade within which the chiasm forms, suggesting a tions; Ms. Debra Compton for performing the in situ hybridiza- possible role for the Eph “B” subfamily in retinal axon tions; and Richard Blazeski for expert assistance with the figures. sorting at the chiasm (this study and Nakagawa et al., This work was supported by NIH Grants NS 27615 and PO 30532 (C.A.M.) and NRSA Fellowship F32 EY06510 (R.C.M). 2000). Further study will be necessary to test this hypothesis. Rather than ventrotemporal, or uncrossed, axons display- REFERENCES ing selective inhibition by the chiasmatic cells, temporal and nasal axons were differentially inhibited by chiasmatic Alder, J., Cho, N. K., and Hatten, M. E. (1996). Embryonic precursor reaggregates in culture. This differential response was par- cells from the rhombic lip are specified to a cerebellar granule tially blocked by the addition of EphA5-Fc, consistent with neuron identity. Neuron 17, 389–399. a role for “A” family interactions in these position- Baier, H., Klostermann, S., Trowe, T., Karlstrom, R. O., Nusslein- dependent responses. Recent analyses using double fluores- Volhard, C., and Bonhoeffer, F. (1996). Genetic dissection of the cent dye labeling indicate retinal axons sort out in a retinotectal projection. Development 123, 415–425. position-dependent manner as they approach, confront the Baird, D. H., Hatten, M. E., and Mason, C. A. (1992). Cerebellar midline, and enter the optic tracts en route to their bilateral target neurons provide a stop signal for afferent neurite extension in vitro. J. Neurosci. 12, 619–634. targets (Chan and Chung, 1999). These findings suggest that Braisted, J. E., McLaughlin, T., Wang, H. U., Friedman, G. C., local cues in the hypothalamus contribute to retinal axon Anderson, D. J., and O’Leary, D. D. M. (1997). Graded and reorganization. Here we demonstrate that Eph family inter- lamina-specific distributions of ligands of EphB receptor tyrosine actions contribute to the differential responses of temporal kinases in the developing retinotectal system. Dev. Biol. 191, and nasal neurites to chiasmatic reaggregates, thereby im- 14–28.

Brennan, C., Monschau, B., Lindberg, R., Guthrie, B., Drescher, U., Frisen, J., and Barbacid, M. (1997). Genetic analysis of the role of Bonhoeffer, F., and Holder, N. (1997). Two Eph receptor tyrosine Eph receptors in the development of the mammalian nervous kinase ligands control axon growth and may be involved in the system. Cell Tissue Res. 290, 209–215. creation of the retinotectal map in the zebrafish. Development Frisen, J., Yates, P. A., McLaughlin, T., Friedman, G. C., O’Leary, 124, 655–664. D. D. M., and Barbacid, M. (1998). Ephrin-A5 (AL-1/RAGS) is Castellani, V., Yue, Y., Gao, P.-P., Zhou, R., and Bolz, J. (1998). essential for proper retinal axon guidance and topographic map- Dual action of a ligand for Eph receptor tyrosine kinases on ping in the mammalian visual system. Neuron 20, 235–243. specific populations of axons during the development of cortical Gale, N. W., Holland, S. J., Valenzuela, D. M., Flenniken, A., Pan, circuits. J. Neurosci. 18, 4663–4672. L., Henkemeyer, M., Strebhardt, K., Hirai, H., Wilkinson, D. G., Chan, S. O., and Chung, K. Y. (1999). Changes in axon arrangement Pawson, T., Davis, S., and Yancopoulos, G. D. (1996). Eph in the retinofugal pathway of mouse embryos: Confocal micros- receptors and ligands comprise two major specificity subclasses, copy study using single- and double-dye label. J. Comp. Neurol. and are reciprocally compartmentalized during embryogenesis. 406, 251–262. Neuron 17, 9–19. Cheng, H. J., Nakamoto, M., Bergemann, A. D., and Flanagan, J. G. Gao, P. P., Yue, Y., Zhang, J.-H., Cerretti, D. P., Levitt, P., and (1995). Complementary gradients in expression and binding of Zhou, R. (1998). Regulation of thalamic neurite outgrowth by the ELF-1 and Mek4 in development of the topographic retinotectal Eph ligand ephrin-A5: Implications in the development of projection map. Cell 82, 371–381. thalamocortical projections. Proc. Natl. Acad. Sci. USA 95, Colello, S. J., and Coleman, L.-A. (1997). Changing course of 5329–5334. growing axons in the optic chiasm of mouse. J. Comp. Neurol. Gao, P. P., Zhang, J. H., Yokoyama, M., Racey, B., Dreyfus, C. F., 379, 495–514. Black, I. B., and Zhou, R. (1996). Regulation of topographic Connor, R. J., Menzel, P., and Pasquale, E. B. (1998). Expression and projection in the brain: Elf-1 in the hippocamposeptal system. tyrosine phosphorylation of Eph receptors suggest multiple Proc. Natl. Acad. Sci. USA 93, 11161–11166. mechanisms in patterning of the visual system. Dev. Biol. 193, Gao, W.-Q., Heintz, N., and Hatten, M. E. (1991). Cerebellar 21–35. granule cell neurogenesis is regulated by cell-cell interactions in Davis, S., Gale, N. W., Aldrich, T. H., Maisonpierre, P. C., Lhotak, vitro. Neuron 6, 705–715. V., Pawson, T., Goldfarb, M., and Yancopoulos, G. D. (1994). Godement, P., Wang, L.-C., and Mason, C. A. (1994). Retinal axon Ligands for EPH-related receptor tyrosine kinases that require divergence in the optic chiasm: Dynamics of growth cone behav- membrane attachment or clustering for activity. Science 266, ior at the midline. J. Neurosci. 14, 7024–7039. 816–819. Guillery, R. W., Mason, C. A., and Taylor, J. S. H. (1995). Develop- Deitcher, D., Fekete, D., and Cepko, C. (1994). Asymmetric expres- mental determinants at the mammalian optic chiasm. J. Neuro- sion of a novel homeobox gene in vertebrate sensory organs. sci. 15, 4727–4737. J. Neurosci. 14, 486–498. Guthrie, S. (1999). Axon guidance: Starting and stopping with Slit. Donoghue, M. J., Lewis, R. M., Merlie, J. P., and Sanes, J. R. (1996). Curr. Biol. 9, R432–R435. The Eph kinase ligand AL-1 is expressed by rostral muscles and Hatten, M. E., Gao, W.-Q., Morrison, M. E., and Mason, C. A. inhibits outgrowth from caudal neurons. Mol. Cell. Neurosci. 8, (1998). The cerebellum: Purification and co-culture of identified 185–198. cell populations. In “Culturing Nerve Cells” (G. Banker and K. Drescher, U., Bonhoeffer, F., and Muller, B. K. (1997). The Eph Goslin, Eds.), pp. 419–459. MIT Press, Cambridge, MA. family in retinal axon guidance. Curr. Opin. Neurobiol. 7, 75–80. Hatini, V., Tao, W., and Lai, E. (1994) Expression of winged helix Drescher, U., Kremoser, C., Handwerker, C., Loschinger, J., Noda, genes, BF-1 and BF-2, define adjacent domains within the devel- M., and Bonhoeffer, F. (1995). In vitro guidance of retinal ganglion oping forebrain and retina. J. Neurobiol. 25, 1293–1309. cell axons by RAGS, a 25 kDa tectal protein related to ligands for Henkemeyer, M., Orioli, D., Henderson, J. T., Saxton, T. M., Roder, Eph receptor tyrosine kinases. Cell 82, 359–370. J., Pawson, T., and Klein, R. (1996). Nuk controls pathfinding of Dutting, D., Handwerker, C., and Drescher, U. (1999). Topographic commissural axons in the mammalian central nervous system. targeting and pathfinding errors of retinal axons following over- Cell 86, 35–46. expression of ephrinA ligands on retinal ganglion cell axons. Dev. Holash, J. A., Soans, C., Chong, L. D., Shao, H., Dixit, B. M., and Biol. 216, 297–311. Pasquale, E. B. (1997). Reciprocal expression of the Eph receptor Eph Nomenclature Committee (1997). Unified nomenclature for Cek5 and its ligand(s) in the early retina. Dev. Biol. 182, Eph family receptors and their ligands, the ephrins. Cell 90, 403. 256–269. Feldheim, D. A., Vanderhaeghen, P., Hansen, M. J., Frisen, J., Lu, Holder, N., and Klein, R. (1999). Eph receptors and ephrins: Q., Barbacid, M., and Flanagan, J. G. (1998). Topographic guid- Effectors of morphogenesis. Development 126, 2033–2044. ance labels in a sensory projection to the forebrain. Neuron 21, Hornberger, M. R., Dutting, D., Ciossek, T., Yamada, T., Hand- 1303–1313. werker, C., Lang, S., Weth, F., Huf, J., Webel, R., Logan, C., Figdor, M., and Stern, C. D. (1993). Segmental organization of Tanaka, H., and Dresher, U. (1999). Modulation of EphA receptor embryonic diencephalon. Nature 363, 630–633. function by coexpressed EprhinA ligands on retinal ganglion cell Flanagan, J. G., and Vanderhaeghen, P. (1998). The ephrins and Eph axons. Neuron 22, 731–742. receptors in neural development. Annu. Rev. Neurosci. 21, Huh, S., Hatini, V., Marcus, R. C., Li, S. C., and Lai, E. (1999). 309–345. Dorsal–ventral patterning defects in the eye of BF-1-deficient Flenniken, A. M., Gale, N. W., Yancopoulos, G. D., and Wilkinson, mice associated with a restricted loss of shh expression. Dev. D. G. (1996). Distinct and overlapping expression patterns of Biol. 211, 53–63. ligands for Eph-related receptor tyrosine kinases during mouse Kaprielian, Z., and Patterson, P. H. (1994). The molecular basis of embryogenesis. Dev. Biol. 179, 382–401. retinotectal topography. Bioessays 16, 1–11.

Karlstrom, R. O., Trowe, T., Klostermann, S., Baier, H., Brand, M., Orioli, D., Henkemeyer, M., Lemke, G., Klein, R., and Pawson, T. Crawford, A. D., Grunewald, B., Haffter, P., Hoffmann, H., (1996). Sek4 and Nuk receptors cooperate in guidance of com- Meyer, S. U., Muller, B. K., Richter, S., van Eeden, F. J. M., missural axons and in palate formation. EMBO J. 15, 6035–6049. Nusslein-Volhard, C., and Bonhoeffer, F. (1996). Zebrafish muta- Park, S., Frisen, J., and Barbacid, M. (1997). Aberrant axonal tions affecting retinotectal pathfinding. Development 123, 427– projections in mice lacking EphA8 (Eek) tyrosine protein kinase 438. receptors. EMBO J. 16, 3106–3114. Krull, C. E., Lansford, R., Gale, N. W., Collazo, A., Marcelle, C., Puelles, L., and Rubenstein, J. L. R. (1993). Expression patterns of Yancopoulos, G. D., Fraser, S. E., and Bronner-Fraser, M. (1997). homeobox and other putative regulatory genes in the embryonic Interactions of Eph-related receptors and ligands confer rostro- mouse forebrain suggest a neuromeric organization. TINS 16, caudal pattern to trunk neural crest migration. Curr. Biol. 7, 472–479. 571–580. Savitt, J. M., Trisler, D., and Hilt, D. C. (1995). Molecular cloning Lagenaur, C., Kunemund, V., Fischer, G., Fushiki, S., and of TOPAP: A topographically graded protein in the developing Schachner, M. (1992). Monoclonal M6 antibody interferes with chick visual system. Neuron 14, 253–261. neurite extension of cultured neurons. J. Neurobiol. 23, 71–88. Sefton, M., Araujo, M., and Nieto, M. A. (1997). Novel expression Macdonald, R., Scholes, J., Stra¨hle, U., Brennan, C., Holder, N., gradients of Eph-like receptor tyrosine kinases in the developing Brand, M., and Wilson, S. W. (1997). The Pax protein Noi is chick retina. Dev. Biol. 188, 363–368. required for commissural axon pathway formation in the rostral Smith, A., Robinson, V., Patel, K., and Wilkinson, D. G. (1997). The forebrain. Development 124, 2397–2408. EphA4 and EphB1 receptor tyrosine kinase and ephrin-B2 ligand Marcus, R. C., Blazeski, R., Godement, P., and Mason, C. A. (1995). regulate targeted migration of branchial neural crest cells. Curr. Retinal axon divergence in the optic chiasm: Uncrossed axons Biol. 7, 561–570. diverge from crossed axons within a midline glial specialization. Sobieszczuk, D. F., and Wilkinson, D. G. (1999). Masking of Eph J. Neurosci. 15, 3716–3729. receptors and ephrins. Curr. Biol. 9, R469–R470. Marcus, R. C., Gale, N. W., Morrison, M. E., Mason, C. A., and Sretavan, D. W., Feng, L., Pure, E., and Reichardt, L. F. (1994). Yancopoulos, G. D. (1996a). Eph family receptors and their Embryonic neurons of the developing optic chiasm express L1 ligands distribute in opposing gradients in the developing mouse and CD44, cell surface molecules with opposing effects on retina. Dev. Biol. 180, 786–789. retinal axon growth. Neuron 12, 957–975. Marcus, R. C., Wang, L.-C., and Mason, C. A. (1996b). Retinal axon Sretavan, D. W., and Reichardt, L. F. (1993). Time-lapse video divergence in the optic chiasm: Midline cells are unaffected by analysis of retinal ganglion cell axon pathfinding at the mamma- the albino mutation. Development 122, 859–868. lian optic chiasm: Growth cone guidance using intrinsic chiasm Marcus, R. C., and Mason, C. A. (1995). The first retinal axon cues. Neuron 10, 761–777. growth in the mouse optic chiasm: Axon patterning and the Stein, E., Savaskan, N. E., Ninnemann, O., Nitsch, R., Zhou, R., cellular environment. J. Neurosci. 15, 6389–6402. and Skutella, T. (1999). A role for the Eph Ligand ephrin-A3 in Marcus, R. C., Shimamura, K., Sretavan, D., Lai, E., Rubenstein, entorhinal–hippocampal axon targeting. J. Neurosci. 19, 8885– J. L. R., and Mason, C. A. (1999). Domains of regulatory gene 8893. expression and the developing optic chiasm: Correspondence Stoeckli, E. T., and Landmesser, L. T. (1998). Axon guidance at with retinal axon paths and candidate signaling cells. J. Comp. choice points. Curr. Opin. Neurobiol. 8, 73–39. Neurol. 403, 346–358. Tessier-Lavigne, M., and Goodman, C. S. (1996). The molecular Mason, C. A., Marcus, R. C., and Wang, L.-C. (1996). Retinal axon Science divergence in the optic chiasm: Growth cone behaviors and biology of axon guidance. 274, 1123–1133. signaling cells. In “Neural Development and Plasticity” (R. R. Theil, T., Frain, M., Gilardi Hebenstreit, P., Flenniken, A., Char- Mize and R. S. Erzurumlu, Eds.), Prog. Brain Res., Vol. 108, pp. nay, P., and Wilkinson, D. G. (1998). Segmental expression of the 95–107. Elsevier, New York. EphA4 (Sek-1) receptor tyrosine kinase in the hindbrain is under Mason, C. A., and Sretavan, D. W. (1997). Glia, neurons, and axon direct transcriptional control of Krox-20. Development 125, pathfinding during optic chiasm development. Curr. Opin. Neu- 443–452. robiol. 7, 647–653. Trowe, T., Klostermann, S., Baier, H., Granato, M., Crawford, Mellitzer, G., Xu, Q., and Wilkinson, D. G. (1999). Eph receptors A. D., Grunewald, B., Karlstrom, R. O., Meyer, S. U., Muller, and ephrins restrict cell intermingling and communication. B. K., Richter, S., Nusslein-Volhard, C., and Bonhoeffer, F. (1996). Nature 400, 77–81. Mutations disrupting the ordering and topographic mapping of Nakagawa, S-i., Brennan, C., Johnson, K., Shewan, D., Harris, axons in the retinotectal projection of the zebrafish, Danio rerio. W. A., and Holt, C. E. (2000). Ephrin-B regulates the ipsilateral Development 123, 439–450. routing of retinal axons at the optic chiasm. Neuron, in press. Tuttle, R., Braisted, J. E., Richards, L. J., and O’Leary, D. D. M. Nakamoto, M., Cheng, H. -J., Friedman, G. C., McLaughlin, T., (1998). Retinal axon guidance by region-specific cues in dien- Hansen, M. J., Yoon, C. H., O’Leary, D. D. M., and Flanagan, J. G. cephalon. Development 125, 791–801. (1996). Topographically specific effects of ELF-1 on retinal axon Varela-Echaverria, A., and Guthrie, S. (1997). Molecules making guidance in vitro and retinal axon mapping in vivo. Cell 86, waves in axon guidance. Genes Dev. 11, 545–557. 755–766. Wang, H. U., and Anderson, D. J. (1997). Eph family transmem- O’Leary, D. D. M., and Wilkinson, D. G. (1999). Eph receptors and brane ligands can mediate repulsive guidance of trunk neural ephrins in neural development. Curr. Opin. Neurobiol. 9, 65–73. crest migration and motor axon outgrowth. Neuron 18, 383–396. Ohta, K., Iwamasa, H., Drescher, U., Terasaki, H., and Tanaka, H. Wang, L.-C., Dani, J., Godement, P., Marcus, R. C., and Mason, (1997). The inhibitory effect on neurite outgrowth of motoneu- C. A. (1995). Crossed and uncrossed retinal axons respond rons exerted by the ligands ELF-1 and RAGS. Mech. Dev. 64, differently to cells of the optic chiasm midline in vitro. Neuron 127–135. 15, 1349–1364.

Wang, L.-C., Rachel, R. A., Marcus, R. C., and Mason, C. A. (1996). Xu, Q., Alldus, G., Holder, N., and Wilkinson, D. G. (1995). Chemosuppression of retinal axon growth by the mouse optic Expression of truncated Sek-1 receptor tyrosine kinase dis- chiasm. Neuron 17, 849–862. rupts the segmental restriction of gene expression in the Wilson, S. W., Brennan, C., MacDonald, R., Brand, M., and Holder, Xenopus and zebrafish hindbrain. Development 121, 4005– N. (1997). Analysis of axon tract formation in the zebrafish brain: 4016. The role of territories of gene expression and their boundaries. Xu, Q., Alldus, G., Macdonald, R., Wilkinson, D. G., and Holder, N. Cell Tissue Res. 290, 189–196. (1996). Function of the Eph-related kinase rtk1 in paterning of the Winslow, J. W., Moran, P., Valverde, J., Shih, A., Yuan, J. Q., Wong, zebrafish forebrain. Nature 381, 319–322. S. C., Tsai, S. P., Goddard, A., Henzel, W. J., Hefti, F., Beck, K. D., Zhang, J.-H., Cerretti, D. P., Yu, T., Flanagan, J. G., and Zhou, R. and Caras, I. W. (1995). Cloning of AL-1, a ligand for an Eph- (1996). Detection of ligands in regions anatomically connected to related tyrosine kinase receptor involved in axon bundle forma- neurons expressing the Eph receptor Bsk: Potential roles in tion. Neuron 14, 973–981. neuron-target interaction. J. Neurosci. 16, 7182–7192. Wizenmann, A., Thanos, S., Boxberg, Y. V., and Bonhoeffer, F. (1993). Differential reaction of crossing and non-crossing rat Received for publication July 30, 1999 retinal axons on cell membrane preparations from the chiasm Revised February 4, 2000 midline: An in vitro study. Development 117, 725–735. Accepted February 4, 2000