Collapsin-1 is a member of the semaphorin family of signaling molecules that acts as a repellent for growing spinal sensory axons. We have constructed a chimeric collapsin-1/alkaline phosphatase probe to visualize putative collapsin-1 receptors in vitro andin situ. As predicted by the activity profile of collapsin-1, the probe binds spinal sensory tracts, ventral spinal roots, and the sympathetic chain but does not bind retinal axons. In addition, we find that the probe binds sensory axons arising from the olfactory epithelium and some, but not all, cranial sensory nerves. As predicted by these binding studies, in vitro assays demonstrate that primary olfactory sensory, trigeminal, and jugular ganglion growth cones collapse in the presence of soluble collapsin-1. Comparing the expression pattern of collapsin-1 with the trajectories of collapsin-1 responsive axons suggests that in both the spinal cord and the olfactory bulb, collapsin-1 prevents premature entry of sensory axons into their target and helps determine the final location of sensory terminations.
- olfactory bulb
- chick hindbrain
- axon guidance
- alkaline phosphatase fusion protein
- collapsin-1 receptors
Growth cones at the tips of extending axons must be able to detect and respond appropriately to attractive and repellent guidance cues in their immediate environment. It is likely that many of these guidance cues are signaling molecules that bind and activate specific cell surface receptors. The trajectory of a growth cone will depend on the repertoire of receptors expressed on its surface and on the effects that activation of these receptors have on its motile apparatus. Notable examples of receptors important for growth cone guidance now include unc-40 or deleted in colorectal cancer (Hedgecock et al., 1990; Serafini et al., 1994; Chan et al., 1996; Keino-Masu et al., 1996), the Eph family of receptors (Cheng et al., 1995; Henkemeyer et al., 1996; Nakamoto et al., 1996; Zhang et al., 1996), and receptor tyrosine phosphatases (Desai et al., 1996;Krueger et al., 1996).
Here we examine the embryonic distribution of binding sites, and therefore potential receptors, for the sensory axon repellent collapsin-1. Collapsin-1 is a member of a large family of signaling proteins, the semaphorins, that are expressed in specific patterns within the developing embryo (Kolodkin et al., 1993; Luo et al., 1995;Püschel et al., 1995; Adams et al., 1996). Collapsin-1 is the chick homolog of mouse semaphorin-D and human semaphorin-III. It is a secreted glycoprotein that has been shown to inhibit the motility of dorsal root ganglion (DRG) growth cones in vitro (Luo et al., 1993) and to act as a repellent of DRG axons in collagen-stabilized cultures (Messersmith et al., 1995; Püschel et al., 1995). Sensory growth cones respond to concentrations of soluble collapsin-1 on the order of 10 pm and avoid small beads to which native collapsin-1 is covalently attached (Luo et al., 1993; Fan and Raper, 1995). These findings suggest that collapsin-1 is a growth cone guidance cue that acts as a repellent and that its effects are mediated by a high-affinity cell surface receptor.
Collapsin-1 inhibits the motility of DRG, sympathetic, ciliary, and spinal motor neurons (Shepherd et al., 1996; Koppel et al., 1997) (H. Kobayashi, unpublished observations). These effects are specific, because the motilities of retinal ganglion cell and olfactory mitral cell growth cones are not inhibited by collapsin-1. Collapsin-1 mRNAs are expressed in specific, highly localized regions of the developing embryo, including the ventral spinal cord, dermamyotome, clusters of cells in the brainstem, and the olfactory bulb (Giger et al., 1996;Shepherd et al., 1996). The widespread but selective distribution of collapsin-1 mRNA in the developing embryo and the ability of several different cell types to respond to collapsin-1 indicate that it is likely to play a role in a variety of axon guidance decisionsin vivo.
A systematic search for axonal guidance decisions affected by collapsin-1 would begin by identifying all the neuron types that are responsive to collapsin-1. Their axon trajectories would then be compared with the expression pattern of collapsin-1 to identify locations where collapsin-1 is likely to act as a guidance cue. We have constructed a chimeric collapsin-1/alkaline phosphatase (AP-collapsin-1) probe that should bind and thereby allow the visualization of collapsin-1 receptors (Flanagan and Leder, 1990). We have used this probe to identify axon tracts likely to be responsive to collapsin-1. In selected cases we have confirmed that tracts that bind the probe are indeed responsive to collapsin-1 in an in vitro growth cone collapse assay. A comparison of this information with the known distribution of collapsin-1 in the embryo suggests specific instances in which this signaling molecule may play a role in growth cone guidance in vivo. This approach could provide a general method to determine systematically candidate functions for other collapsin and semaphorin family members.
MATERIALS AND METHODS
Materials. The following materials were obtained from the indicated sources: nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophin-4 (NT4) from Alamone Labs (Jerusalem, Israel); progesterone, nitroblue tetrazolium (NBT), and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) from Sigma (St. Louis, MO); insulin, transferrin, and putrescine from Collaborative Research (Bedford, MA); fetal calf serum (FCS) and laminin from Life Technologies (Grand Island, NY); nitrocellulose paper from Schleicher & Schuell (Keene, NH); slot blot apparatus, Hoefer PR-648, from Hoefer Pharmacia Biotech (San Francisco, CA); EX CELL 405 from JRH Biosciences (Lenexa, KS); 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) from Molecular Probes (Eugene, OR); Dispase II and alkaline phosphatase-conjugated anti-digoxigenin (DIG) antibodies from Boehringer Mannheim (Mannheim, Germany); Cy3-conjugated antibody from Jackson ImmunoResearch (West Grove, PA); Tissue-Tek OCT compound from Sakura (Tokyo, Japan); and superfrost slides from Fisher Scientific (Pittsburgh, PA).
Construction and expression of alkaline phosphatase fusion proteins. Plasmids containing alkaline phosphatase (AP) were the generous gift of Dr. John Flanagan. One contained AP with a stop codon (C-terminal fusion); the other contained AP with a leading signal sequence (N-terminal fusion). To produce Col-1–AP, we cut out the AP sequence from the vector and inserted this sequence after the C-terminal end of the coding sequence for collapsin-1. The amino acid sequence at the junction between collapsin-1 and AP was PRSV-(LE)-IIPV, with PRSV in collapsin-1. The two amino acids in parenthesis served as a linker. To produce AP–Col-1, we replaced the signal sequence of collapsin-1 with the AP signal and coding sequences. The amino acid sequence at the junction between AP and collapsin-1 was SGRS-(GS)-KNNV, with KNNV in collapsin-1. Recombinant protein was produced using the expression plasmid pAG3, derived from a pcDNA3 backbone (Invitrogen, San Diego, CA) with a cytomegalovirus enhancer, a chick β-actin promoter, a rabbit β-globin splice site, a bovine growth hormone polyadenylation site, and a hygromycin resistance gene (Miyazaki et al., 1989).
These two constructs, Col-1–AP and AP–Col-1, were transfected into transformed human kidney epithelial cells (293T) using calcium phosphate precipitation. Chloroquine (1:1000 of 25 mm stock solution) was added to the cells at the time of transfection. After a 4–5 hr incubation at 37°C, the cells were washed and cultured in DMEM with 10% heat-inactivated FBS. Conditioned media containing the secreted fusion proteins were collected ∼20 hr later.
Determining AP-collapsin-1 concentrations. To determine the relative amounts of AP-collapsin-1 fusion proteins produced in each transfection, we blotted a dilution series of culture supernatants onto nitrocellulose paper with a Hoefer slot blot apparatus and reacted the series for AP activity. Supernatants were diluted with DMEM containing 10% heat-inactivated FCS. The nitrocellulose paper was rinsed in PBS and heat-inactivated at 65°C for 3 hr in PBS. The paper was then reacted with an AP reaction mixture containing 100 mm Tris, pH 9.5, 100 mm NaCl, 5 mmMgCl2, 0.33 mg/ml NBT, and 0.17 mg/ml BCIP. To compare the relative amounts of AP–collapsin-1 and standardized unlabeled collapsin-1 of known potency, we probed slot blots with monoclonal antibody (mAb) E7 anti-collapsin-1 and detected them with a HRP-conjugated secondary antibody (Jackson ImmunoResearch, West Grove, PA). Estimates of relative collapsin-1 and AP–collapsin-1 concentrations were accurate to approximately a factor of two when the relative intensities of the dilution series were matched by eye.
Col-1–AP binding to cultures. DRGs from E7 chick embryos or retinae from E6 embryos were cultured in 0.5 ml of medium on glass coverslips coated with laminin as described under Cell culture. One to 10 μl of 293T cell culture supernatant containing Col-1–AP were then added to each explant culture and incubated for 1 hr at 37°C. The culture was fixed by the addition of 4% paraformaldehyde, 10% sucrose, and PBS. After 30 min, the culture was washed with PBS several times, the endogenous alkaline phosphatase was heat-inactivated at 65°C for 2 hr, and the culture was processed for AP reaction. For experiments in which unlabeled collapsin-1 was used to compete with Col-1–AP binding, 10 μl of Col-1–AP containing supernatant was mixed with 10 μl (containing a 20–200-fold greater concentration) of unlabeled collapsin-1 produced in a baculovirus high-expression system. The mixture was added to explant cultures and processed as described above.
Quantitative estimation of AP-Col-1 binding to cultured sympathetic cells. Approximately 5 × 104dissociated sympathetic neurons were plated into each well in a 48 well plate and cultured overnight. Then AP-Col-1 was added to each well at a series of protein concentrations ranging from 8 pm to 0.4 nm, and the cultures were incubated at 37°C for 1 hr. The cultures were then fixed in 4% paraformaldehyde for 30 min, washed three times with a buffer of 0.15 m NaCl and 20 mm HEPES, pH 7.4, and incubated at 65°C for 2.5 hr to destroy the endogenous alkaline phosphatase activity. The AP–Col-1 binding activity was determined colorimetrically by measuring optical density 415 after incubation in 200 μl of 1 mdiethanolamine, pH 9.8, 0.5 mm MgCl2, 10 mm l-homoarginine, 0.5 mg/ml BSA, and 12 mm p-nitrophenyl phosphate for 6–10 hr. Nonspecific binding was determined through the addition of a 100-fold excess of recombinant collapsin-1 with AP-Col-1 and was found to be <5% of the total binding activity. The binding data were analyzed either by the Prism II ligand analysis program or by Sigma plot.
Col-1–AP binding to sections. Chick embryos were decapitated, embedded in Tissue-Tek OCT compound, and frozen in a dry ice–acetone bath. Frozen sections were cut with a cryostat and collected on Superfrost Plus slides. Care was taken not to allow the sections to dry out. After being picked up, they were allowed to attach to the slides in a moist chamber for 2–3 min. They were then immediately frozen by putting the slides onto a precooled metal stage in the cryostat. Sections were post-fixed within 1 hr of being cut with precooled methanol at −20°C for 7–10 min. They were then washed with PBS for 5 min twice and blocked with PBS containing 10% FBS for 15 min at 20°C. Sections were incubated with diluted 293T cell culture supernatants containing Col-1–AP at 20°C for 1 hr. Depending on the efficiency of transfection, 293T cell culture supernatants contained 2000–5000 collapsing units (CU)/ml, and dilutions between 1:20 and 1:100 were used. Sections were rinsed with PBS; fixed with 60% acetone, 3% paraformaldehyde, and 20 mm HEPES, pH 7.0, for 3 min; and then washed with PBS several times. Sections were then incubated at 65°C for 3 hr to inactivate endogenous alkaline phosphatases. Sections were then processed for AP in 100 mmTris, pH 9.5, 100 mm NaCl, 5 mmMgCl2, 0.33 mg/ml NBT, and 0.17 mg/ml BCIP at 20°C overnight. For experiments in which unlabeled collapsin-1 competed with Col-1–AP binding, collapsin-1 obtained from either a baculovirus expression system or 293T cells was used. Sections were prepared in the same way as described above and first preincubated with unlabeled collapsin-1 for 15 min and then incubated with the mixture of the probe and unlabeled collapsin-1 for 1 hr and processed the same way.
When double-stained for neurofilament and AP-Col-1, sections were prepared in the same way but were first blocked with PBS containing 10% FBS and then were incubated with a mixture of hybridoma culture supernatant (4H6, anti-neurofilament) and AP–Col-1. The sections were then fixed, washed, heat-inactivated, and incubated with Cy3-conjugated secondary antibody for 2 hr. Sections were then fixed again, washed with PBS, and processed for AP.
Reconstruction. Images of tissue sections were taken through a television camera (Hamamatsu; C2400) attached to an inverted Zeiss microscope. Images were then projected onto a television monitor screen (Sony, Tokyo, Japan; PVM-122) and traced onto transparent sheets. These tracings were scanned into Adobe photoshop. Schematics were based on reconstructions generated by this method.
In situ hybridization. In situhybridizations were performed as described previously with only minor variations (Shepherd et al., 1996). Sections were cut at 30 μm, and prehybridizations were performed for 1 hr at 65°C. Alkaline phosphatase-conjugated anti-DIG antibodies were used at a 1:2500 dilution. The AP reaction was performed as described under Col-1–AP binding to sections.
Cell culture. DRGs were dissected from E6 or E7 chick embryos and retinas from E6 embryos. Explants were plated onto 10 mm coverslips coated with laminin (40 μg/ml) and cultured essentially as described by Luo et al. (1993) for 18–24 hr. Olfactory epithelia were isolated from stage 27–29 chick embryos. Tissues containing the nasal sac were dissected out and incubated in Hank’s buffer containing 2 mg/ml Dispase II for 30–60 min at 20°C. The olfactory epithelium was isolated with the aid of 30 gauge needles. Care was taken to obtain the whole epithelium with the olfactory nerve attached. The portion of the olfactory epithelium near the olfactory nerve was explanted. Explants were cultured on laminin for 2 d in the presence of 16 μm AraC to suppress the proliferation of non-neuronal cells. Trigeminal (Vth), vestibular (VIIIth), and jugular (proximal Xth) ganglia were dissected from E6 (stage 29–30) chick embryos. Each ganglion was halved, plated onto laminin-coated glass, and cultured for 20 hr. The ophthalmic lobe of the trigeminal ganglion was used for all experiments in the present study. The culture medium used was F-12 (Life Technologies) supplemented with 200 μg/ml bovine pituitary extract (Tsao et al., 1982) dialyzed overnight against F-12 medium, 14 mm NaHCO3, 2 mm glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 6 mg/ml glucose, 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml selenious acid, 100 μm putrescine, 20 nm progesterone, and 20 ng/ml 7 S NGF. VIIIth vestibular ganglia were cultured in the same medium supplemented with 10 ng/ml NT-4, 10 ng/ml BDNF, and 8 μm AraC; and olfactory epithelium were in the same medium plus 16 μm AraC. For the quantitative binding assays to sympathetic neurons, sympathetic ganglionic chains were dissected from E8 chick embryos, incubated for 15 min in Hank’s buffer containing 0.05% trypsin and 0.5 mm EDTA at 37°C, and dissociated into single cells by trituration. Approximately 5 × 104 cells were plated into each polylysine and laminin-coated well in a 48 well plate and were cultured overnight.
Collapse assay. The procedure for the collapse assay was essentially the same as that described previously (Raper and Kapfhammer, 1990). Briefly, 10 μl aliquots of diluted recombinant collapsin-1 were added to 500 μl of culture medium. The added material was gently mixed into the culture medium, and the cultures were incubated at 37°C in 5% CO2 for 1 hr. Cultures were then fixed for 1 hr by gently adding 4% paraformaldehyde in PBS containing 10% sucrose. Growth cones without lamellipodia or filopodia were scored as collapsed.
Large-scale production of collapsin-1 for competition experiments. Hi5 insect cells (Invitrogen) were grown in a spinner flask as a suspension culture. Cells were maintained at a density between 105 and 106 cells/ml in EX CELL 405 at 27°C. The spinners rotated at 100–110 rpm. Hi5 cells were infected with baculovirus carrying a recombinant collapsin-1–myc construct (Shepherd et al., 1997) at a density of 106 cells/ml. After 48 hr, the supernatant was harvested by gently pelleting the cells. The supernatant was centrifuged at 100,000 × g for 1 hr. The resulting pellet was solubilized in 50 mm Tris, pH 7.4, 0.1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), and 1 m NaCl (buffer A) and was centrifuged again at 100,000 × g for 1 hr. The supernatant was collected and mixed with 50 mm Tris, pH 7.4, 0.1% CHAPS, and 0.1m NaCl (buffer B) at the ratio of 1:5 and was loaded on an S-Sepharose column. The column was washed with buffer B, and pure collapsin-1 was eluted with buffer A.
Collapsin-1 alkaline phosphatase fusion construct
Recombinant fusion proteins consisting of human placental alkaline phosphatase and the full coding sequence of collapsin-1 were produced in transiently transfected human 293T cells. Two different chimeras were made (Fig. 1 A), one with alkaline phosphatase fused to the N-terminal end of collapsin-1 behind a signal sequence (AP–Col-1) and another with the enzyme fused to the C-terminal end (Col-1–AP). To verify that collapsin-1 function is not compromised by the addition of alkaline phosphatase, we tested these fusion proteins for sensory growth cone collapsing activity in an in vitro assay. The specific collapsing activities of both the AP–Col-1 and Col-1–AP fusion proteins are as high or higher than that of collapsin-1 itself (Fig.1 B).
Visualization of collapsin-1 binding on cultured explants
Collapsin-1 is known to induce the collapse of DRG but not of retinal growth cones in vitro (Luo et al., 1993). A plausible inference is that DRG growth cones will bind collapsin-1, whereas retinal axons may not. Axons extending from E7 DRG explants were incubated with Col-1–AP for 1 hr at 37°C and fixed. The presence of alkaline phospatase reaction product demonstrates the binding of collapsin-1 to DRG axons (Fig.2 B). Not only axons, but also collapsed growth cones, are intensely labeled with Col-1–AP (Fig. 2 D). No binding is observed in the explants themselves or in migrating non-neuronal cells. We compared the collapsing activity and intensity of the AP signal at various concentrations of the Col-1–AP probe. If a CU is defined as the concentration of collapsin-1 required to collapse 50% of DRG growth cones, the binding of Col-1–AP to DRG axons is first detected at 5 CU, although staining intensity at this concentration is weak (data not shown). Fifty collapsing units, or ∼100 ng/ml, give a strong AP signal. To test the specificity of the Col-1–AP signal, we tested whether unlabeled collapsin-1 could compete with Col-1–AP binding. No Col-1–AP binding is detected in DRG cultures simultaneously exposed to a 200-fold excess (∼20 μg/ml) of unlabeled collapsin-1 (Fig.2 E). Even a 20-fold excess of unlabeled collapsin-1 reduces the intensity of the staining significantly (data not shown). The specificity of Col-1–AP binding was confirmed further by testing its ability to bind retinal growth cones. No detectable AP reaction product is observed on retinal ganglion cell axons or growth cones incubated in concentrations of Col-1–AP that give strong DRG labeling (Fig. 2 F). These results demonstrate that Col-1–AP specifically binds collapsin-1 sensitive axons.
The binding affinity of collapsin-1 to its putative receptors was estimated by examining the binding to dissociated neurons from E8 sympathetic ganglia. Sympathetic ganglia were chosen for this study because they are responsive to collapsin-1, are almost free from non-neuronal cells, and contain a relatively homogeneous population of neurons. Binding studies (like those just described for sensory neurons) indicate that AP–Col-1 stains the surface of sympathetic axons and cell bodies but not the surface of non-neuronal cells. A 100-fold excess of unlabeled recombinant collapsin-1 prevents AP–Col-1 from binding, demonstrating the specificity of the binding interaction (data not shown). AP–Col-1 binds to dissociated sympathetic neurons in a nearly saturable manner with a K d of ∼800 pm (Fig. 1 C). More-detailed analysis at lower concentrations of the ligand shows that the binding can be fitted well with a two site ligand–receptor binding model (Fig.1 D). A nonlinear regression analysis of the binding data suggests the presence of separate high and low affinity binding sites. The K d for the high affinity site is estimated to be ∼30 pm. A Scatchard analysis of the data is consistent with the presence of a high affinity binding site (Fig.1 D, inset). In six out of seven cases, Scatchard analysis showed the presence of two binding sites. The number of high affinity sites per sympathetic neuron is estimated to be on the order of ∼104 sites/cell.
Visualization of collapsin-1 binding in sections
We next turned our attention to the distribution of collapsin-1 binding in the developing nervous system of the chick. Fresh-frozen sectioned embryos were lightly post-fixed in cold methanol and reacted with either Col-1–AP or AP–Col-1, and the alkaline phosphatase was reacted to form an insoluble product that accumulated where the probe was bound. Both probes revealed the same distribution of collapsin-1 binding on sections. Most figures in this paper were stained with AP–Col-1.
As would be predicted by their responsiveness to collapsin-1 and their ability to bind Col-1–AP in culture, sensory afferents in the spinal cord are heavily labeled with the probe. In sections of stage 26 embryos, the nascent dorsal sensory columns are well labeled (Fig.3 A). Decreased labeling is obtained when a 50-fold excess of unlabeled collapsin-1 competes with the AP–Col-1 probe, and no labeling is obtained when a 200-fold excess of unlabeled collapsin-1 is used (Fig. 3 B). Not only the central projections but also the peripheral sensory projections bind AP–Col-1. The sensory nerve just distal to the DRG binds AP–Col-1 before it joins the ventral roots (Fig. 3 A,C). Distal sensory projections may account for AP–Col-1 binding in the peripheral nerves; however, it is possible that binding to motor axons contributes as well. The growth cones of putative ventral spinal cord motor neurons have been shown to collapse in response to collapsin-1 (Shepherd et al., 1996), and the ventral roots of stage 21 embryos are clearly labeled with the probe (Fig. 3 D). Much weaker labeling is sometimes apparent in the motor columns and ventral roots of older embryos (Fig. 3 A). Collapsin-1 has been shown previously to induce the collapse of cultured sympathetic growth cones. It is therefore not surprising that dissociated sympathetic neurons and axons bind AP-Col-1 and that AP–Col-1 labels sympathetic ganglia in situ (Fig. 3 D).
Centrally projecting sensory axons continue to bind AP–Col-1 as development proceeds. The binding of AP–Col-1 to sensory axons is compared to the distribution of collapsin-1 expressing cells at stage 30 in Figure 4, A andB. Collapsin-1 expression is concentrated in the ventral and medial portions of the gray matter, whereas the collapsin-1 sensitive sensory axons are concentrated in the dorsal and lateral portion of the white matter. This same complementariness is observed later in development. At stage 38, sensory afferents that bind AP–Col-1 fill the dorsal horn (Fig. 4 C) but do not enter the ventral gray matter where collapsin-1 expression is high (Fig.4 D). Interestingly, the most medial portion of the dorsal columns that contains a greater proportion of muscle afferents as compared with cutaneous afferents is less heavily labeled by AP–Col-1 than are the lateral dorsal columns.
The binding of AP–Col-1 to spinal sensory, motor, and sympathetic axons matches their known sensitivity to collapsin-1 in vitro. However, the real utility of the alkaline phosphatase fusion construct is the identification of new axon tracts likely to be sensitive to collapsin-1. We next examined the hindbrain to determine what neuronal types are likely to be collapsin-1 responsive.
AP–Col-1 binding is first detected in trigeminal axons at stage 18. The binding pattern in and around the trigeminal ganglion at stage 25 is shown in Figure 5 A. Figure 6 A is a schematic diagram of AP–Col-1 binding reconstructed from serial sections. All three sensory nerves originating in the trigeminal ganglion, the ophthalmic, the maxillary, and the mandibular nerves, are intensely stained with AP–Col-1. The trigeminal root entering the brainstem as well as ascending and descending trigeminal axon tracts in the brainstem are strongly labeled. No AP–Col-1 binding is detected in the trigeminal ganglion itself. Antibodies raised against neurofilaments visualize fibers of passage through the trigeminal ganglion, axon tracts in the brainstem, and the peripheral nerve branches (Fig. 5 B). Axons coursing through the ganglion that are not stained with AP–Col-1 are likely to originate within the hindbrain from the motor and mesencephalic sensory nuclei of V. Figures5 C and 6 B show the position of collapsin-1 expression within the hindbrain. Collapsin-1 is strongly expressed at the lateral border and more weakly expressed at the dorsal and medial margins of the descending trigeminal tract. Surprisingly, collapsin-1 is also expressed in and around the nerve V root where it enters the brainstem.
Facial, acoustic, and vestibular ganglia
AP–Col-1 labeling in the vicinity of the fused VIIth (facial) and VIIIth (vestibuloacoustic) ganglia at stage 25 is shown in Figure 5,E and F. A schematic diagram of AP–Col-1 binding is shown in Figure 6 A. Weak labeling is detected in the most distal margin of the proximal VIIth ganglion (Fig.5 E, arrow). This axonal staining can be followed into the nerve joining the proximal to the distal VIIth (geniculate) ganglion (Fig. 6 A). The distal facial nerve also binds the AP–Col-1 probe (Fig. 6 A). There are no detectable AP–Col-1-labeled axons in either the acoustic or the vestibular portions of the VIIIth ganglion. In contrast, collapsin-1 expression is high in both portions of the VIIIth ganglion (Fig.5 G). Collapsin-1 is not expressed in the VIIth ganglion. In the hindbrain, AP–Col-1 labeling is detected in a tract just lateral to the descending spinotrigeminal tract (Fig. 5 E,arrowhead). As seen in Figure 6 A, this tract courses just lateral to the entry point for VIIth and VIIIth ganglion axons. Labeling the central projections of VIIth and VIIIth sensory axons by applying DiI to the nerve root indicates that the AP–Col-1-labeled tract corresponds with ascending and descending sensory axons originating in the VIIth and/or VIIIth ganglion (Fig.5 J). Because AP–Col-1 labeling in the periphery is only associated with nerve VII fibers, we attribute the lateral AP–Col-1 positive projection in the brainstem to VIIth ganglion sensory fibers. Figures 5 D and 6 B show the position of collapsin-1 expression in the hindbrain near the VIIth and VIIIth nerve root. Collapsin-1 is expressed at the lateral border of the facial sensory tract and between the spinal trigeminal and the facial tracts.
Glossopharyngeal and vagus nerves
AP–Col-1 labeling in the vicinity of the IXth (glossopharyngeal) and Xth (vagus) nerves at stage 25 is shown in Figure 5, H and I. A schematic diagram of AP–Col-1 binding is shown in Figure 6 A. Strong labeling is seen in the Xth nerve, but no labeling is detected in the IXth nerve or in the distal IXth (petrosal) ganglion. However, the proximal IXth ganglion (superior) is labeled with AP–Col-1 (Fig.6 A).
One dramatic example of the identification of a new axon tract likely to be sensitive to collapsin-1 is the labeling of axons originating in the olfactory sensory epithelium. The olfactory nerve is strongly stained with AP–Col-1 at stages 23–25 (Fig.7 A). At this stage of development, the olfactory nerve spans the distance between the olfactory epithelium and the telencephalic vesicle. The olfactory bulb has not yet formed, and the sensory axons do not penetrate into the CNS. AP–Col-1 binding is detected along the entire length of the nerve, from the olfactory epithelium to the end of the nerve at the surface of the telencephalon (Fig. 7 C). Collapsin-1 is expressed in the olfactory epithelium (Fig. 7 B) and in the most superficial layers of the telencephalon (Fig. 7 D). Between stages 25 and 30, olfactory sensory axon endings appear to accumulate on the surface of the telencephalon (Fig. 7 E), and collapsin-1 continues to be expressed superficially in the telencephalon (Fig. 7 F). Olfactory sensory axons have invaded the nascent olfactory bulb by stage 38 and terminate in a superficial layer. They continue to bind AP–Col-1 (Fig.7 G). At the same time, collapsin-1 is expressed in what is now an intermediate cell layer between the ventricular zone and the terminating sensory axons (Fig. 7 H).
Responsiveness of cranial sensory and olfactory growth cones to collapsin-1
To test whether AP–Col-1 binding predicts collapsin-1 responsiveness, we conducted collapse assays on selected axons found in this study to bind AP–Col-1 in situ. Growth cones of axons growing from explanted stage 27–29 olfactory epithelia collapse in response to concentrations of recombinant collapsin-1 comparable with those that affect spinal sensory growth cones (Figs.8 A,B,9). The same is true for the growth cones of explanted stage 29–30 trigeminal (Vth) ganglia (Figs.8 C,D, 9). Growth cones extending from the geniculate (distal VIIth) and the superior IXth ganglia are found to collapse in response to collapsin-1 (data not shown). Growth cones extending from stage 29–30 jugular (Xth) ganglia also collapse in response to collapsin-1 (Figs. 8 G,H, 9). Axons from the VIIIth (vestibuloacoustic) ganglion are not labeled with AP–Col-1 on sections. In agreement with this observation, growth cones of axons originating in explanted stage 29–30 VIIIth vestibular ganglia do not collapse in response to collapsin-1 (Figs. 8 E,F, 9). This is true even when they are exposed to a concentration of collapsin-1 10-fold greater than the concentration that induces 50% collapse of spinal sensory growth cones. All of these activity profiles are as predicted by the pattern of AP–Col-1 binding observed in sections.
The objective of this study was to systematically identify axons in the developing chicken nervous system likely to be responsive to the signaling protein collapsin-1. We constructed alkaline phosphatase–collapsin-1 fusion proteins that should bind collapsin-1 receptors and thereby reveal their distribution in tissue sections.
The specificity and utility of these probes is demonstrated by several observations. AP–Col-1 and Col-1–AP are both at least as active as collapsin-1 in an in vitro collapse assay using DRG growth cones, demonstrating their ability to bind the receptor with high affinity. Binding studies of AP–Col-1 on whole cultured sympathetic neurons indicate the presence of a high affinity binding site. This is the predicted result, because sympathetic growth cones collapse in response to concentrations of collapsin-1 of ∼30 pm. We infer that the activation of a high affinity collapsin-1 receptor is likely to initiate sympathetic growth cone collapse.
The ability of AP-collapsin-1 constructs to bind a specific collapsin-1 receptor is supported further by their ability to specifically label appropriate axons in culture. Cultured collapsin-1 responsive DRG axons bind the Col-1–AP probe. Binding of the Col-1-AP probe can be detected with concentrations fivefold greater than those required to obtain 50% collapse, although concentrations 10-fold greater yield intense staining. Binding of the Col-1–AP probe is competed off with excess unlabeled probe, demonstrating that the probe and collapsin-1 compete for a limited number of specific sites on the cell surface. Cultured retinal axons do not bind detectable Col-1–AP at concentrations that give strong labeling of DRG axons. This result is consistent with the previous observation that retinal axons do not collapse in response to collapsin-1 (Luo et al., 1993). The specificity of the probe is demonstrated further by their specific patterns of binding in sectioned embryos. Only a small number of axon tracts are labeled in sections probed with AP–Col-1 or Col-1–AP. Tracts arising from neurons known to be collapsin-1 responsive are labeled, and those from neurons known to be unresponsive in collapsin-1 in vitro assays are not labeled. This labeling can be competed off with excess unlabeled collapsin-1, again demonstrating that the binding sites are specific and saturable.
The most important test of the utility of the AP-collapsin-1 probes is their ability to identify new collapsin-1 responsive axons. The binding of AP–Col-1 and Col-1–AP to primary olfactory sensory axons, as well as to axons extending from the trigeminal, facial, ciliary, and jugular ganglia, implies that these axons possess collapsin-1 receptors and are likely to be responsive to collapsin-1. Growth cones originating from these explanted neural tissues were subsequently all found to collapse in response to low concentrations of collapsin-1. In contrast, sensory growth cones extending from the VIIIth ganglion do not collapse in response to collapsin-1, as expected by their failure to bind AP–Col-1 in sections.
These findings demonstrate that chimeric alkaline phosphatase–collapsin-1 probes detect a specifically localized collapsin-1 binding component the presence of which on axons correlates with collapsin-1 responsiveness. It is likely that these probes visualize a collapsin-1 receptor. Our estimation of affinity constants for the strength of binding between collapsin-1 and its receptor suggests the presence of high affinity receptor on sympathetic axons with a K d similar to the 30 pm of collapsin-1 that gives a half maximal collapse response, along with lower affinity sites of ∼800 pm. Presumably, it is this high affinity binding site we see in sections, although we cannot exclude the possibility that we are visualizing a lower affinity binding site that colocalizes with a higher affinity collapsin-1 receptor.
The original rationale for this study was to correlate the distribution of collapsin-1 receptors with the expression pattern of collapsin-1. Such a comparison should suggest possible biological functions for collapsin-1 signaling in vivo. We now turn our attention to the relative distributions of collapsin-1 and putative collapsin-1 receptors in the spinal cord, brainstem, and olfactory system.
Collapsin-1 may play an important role in the organization of DRG sensory afferents in the spinal cord. It is expressed at high levels in the ventral cord and at lower levels in the dorsal gray matter of the cord at stage 23 when sensory afferents first enter the dorsal roots (Shepherd et al., 1996). Sensory axons extend within the developing dorsal columns for the next 2 d, but their afferent branches do not invade the dorsal gray matter until stage 28–29 (Davis et al., 1989). Their entry correlates with a concomitant loss of collapsin-1 expression in the nascent dorsal horn (Shepherd et al., 1997). All growth cones extending in vitro from E7 DRG ganglia appear to collapse in response to collapsin-1 (Luo et al., 1993; Püschel et al., 1996; Sharma et al., 1996; Shepherd et al., 1997). It is therefore possible that the loss of collapsin-1 expressing cells in the dorsal gray matter around stage 29 permits collapsin-1 sensitive sensory afferents to enter the spinal gray matter. Consistent with this hypothesis is the presence of strong AP–Col-1 binding on axons within the dorsal columns throughout the time course of these events.
DRG afferents can be classified by their ultimate destinations in the spinal cord. Most afferents terminate in various laminae within the dorsal horn and never extend into the ventral cord. A minority of afferents, including the group Ia stretch receptors, ultimately extend into and synapse within the ventral cord (Brown, 1981; Willis and Coggeshall, 1991). Collapsin-1 expression is gradually confined to the ventral cord between stages 30 and 36. Dorsally terminating afferents remain collapsin-1 sensitive and are confined to the dorsal cord by ventrally expressed collapsin-1, whereas ventrally terminating afferents become collapsin-1 insensitive and are thereby able to invade the ventral cord (Messersmith et al., 1995; Behar et al., 1996;Püschel et al., 1996; Shepherd et al., 1997). In agreement with this, AP–Col-1 only labels the dorsal gray matter. Interestingly, by stage 38 the most medial portion of the dorsal white matter binds significantly less AP–Col-1 than does the more lateral dorsal white matter (Fig. 4 C). This medial portion of the white matter is enriched in TrkC-positive, ventrally invading afferents (Oakley et al., 1997). TrkA-positive dorsally terminating afferents are predominant in the more lateral white matter where strong AP–Col-1 binding is evident. These results suggest that ventrally invading afferents become collapsin-1 insensitive because they lose the collapsin-1 receptor. The cell- and time-specific downregulation of collapsin-1 receptor expression may control which sensory afferents invade ventrally and at what developmental time they begin their invasion.
Sensory axons extending from ganglia contributing to the Vth (trigeminal), VIIth (facial), and Xth (vagus) cranial nerves bind AP–Col-1 and collapse in response to collapsin-1. This collapsin-1 responsiveness could play a role in the guidance of these axons in the periphery, as proposed previously for the peripheral axons of DRG sensory axons (Wright et al., 1995). The finding that collapsin-1 responsive sensory axons abut collapsin-1-expressing cells in the brainstem suggests that collapsin-1 may help to define the medial and lateral limits of these central sensory tracts. Collapsin-1 expression between the central spinal trigeminal and VIIth sensory tracts may help keep them separate.
Fused VIIth and VIIIth ganglia
The VIIth and VIIIth nerves share the same entry point into the hindbrain. In the chick, there are two facial sensory ganglia, a distal (geniculate) ganglion and a more proximal ganglion that is fused with the VIIIth (vestibular) ganglion (Yamamoto and Schwarting, 1991). Axons extending from these ganglia enter the hindbrain through the fused VII and VIII nerve root. Centrally projecting facial sensory axons must therefore navigate through a choice point where they choose to enter the VII and VIII root and not the VIIIth ganglion. Collapsin-1 is expressed in both the vestibular and acoustic parts of the VIIIth ganglion. AP–Col-1 binds to VIIth nerve axons, and growth cones extending from explanted VIIth ganglia respond to collapsin-1. This complementary pattern of collapsin-1 expression and collapsin-1 receptor distribution suggests that collapsin-1 in the VIIIth ganglion denies entry to collapsin-1 responsive VIIth nerve axons growing toward the brainstem.
Just as spinal sensory afferents observe a waiting period before they enter the gray matter of the spinal cord, axons in the olfactory nerve wait on the surface of the telencephalon before entering the developing olfactory bulb. The time course of olfactory nerve innervation of the bulb has been extensively studied in rodents (Doucette, 1989; Marin-Padilla and Amieva, 1989; Santacana et al., 1992). Olfactory sensory epithelial axons grow through the mesenchyme between the epithelium and the telencephalon, turn anteriorly once they reach the telencephalon, and stop where the olfactory bulb will later differentiate. Although a small number of pioneer olfactory axons transiently penetrate into the telencephalon, the vast majority of olfactory axons accumulate for days outside the CNS without entering (Valverde et al., 1992; Gong and Shipley, 1995). These fibers only enter the CNS as the telencephalic vesicle begins to evaginate when the olfactory bulb starts to form. Olfactory nerve axons ultimately terminate in glomeruli located within a superficial layer of the bulb.
Although much less information is available about these events in the developing chick, our observations are consistent with this general scheme (Fig. 10). Olfactory nerve axons have arrived at the telencephalic surface by stage 25 (Fig. 10,St. 25) but have not invaded the CNS even by stage 31 (Fig.10, St. 31). Interestingly, throughout this time period, collapsin-1 is expressed within the most superficial layers of the telencephalic vesicle (for similar observations in the rat, see Giger et al., 1996). Our demonstration that AP–Col-1 binds to olfactory sensory axons and that collapsin-1 collapses olfactory growth conesin vitro suggests that collapsin-1 prevents olfactory axon entrance into the telencephalon during this period. By stage 38, olfactory axons have invaded the nascent olfactory bulb and now occupy its outermost layer (Fig. 10, St. 38). At this stage of development, collapsin-1 is expressed in an intermediate layer just beneath the olfactory axon terminations. We therefore predict that sensory axons are confined to superficial layers of the olfactory bulb by the expression of collapsin-1 in intermediate layers.
Both in the dorsal horn of the spinal cord and in the olfactory bulb, collapsin-1 receptive sensory axons are excluded for several days from immediately adjacent areas that express collapsin-1. In both cases, the final pattern of sensory axon termination is complementary to the distribution of collapsin-1 expression. Thus, both the timing of target penetration and the final spatial localization of terminations may be controlled by the timing and localization of collapsin-1 expression.
This work was supported by grants from the National Institutes of Health and the McKnight Foundation. Hiroaki Kobayashi is supported by a fellowship from the Japanese Society for the Promotion of Science. We thank Dr. John Flanagan for plasmids to generate AP fusion protein and Dr. Steven Scherer for his kind advice on preparing frozen sections.
Correspondence should be addressed to Dr. Hiroaki Kobayashi, 105 Johnson Pavilion, Department of Neuroscience, University of Pennsylvania School of Medicine, 3600 Hamilton Walk, Philadelphia, PA 19104.