Report

Anterior-Posterior Guidance of Commissural Axons by Wnt-Frizzled Signaling

See allHide authors and affiliations

Science  12 Dec 2003:
Vol. 302, Issue 5652, pp. 1984-1988
DOI: 10.1126/science.1089610

Abstract

Commissural neurons in the mammalian dorsal spinal cord send axons ventrally toward the floor plate, where they cross the midline and turn anteriorly toward the brain; a gradient of chemoattractant(s) inside the spinal cord controls this turning. In rodents, several Wnt proteins stimulate the extension of commissural axons after midline crossing (postcrossing). We found that Wnt4 messenger RNA is expressed in a decreasing anterior-to-posterior gradient in the floor plate, and that a directed source of Wnt4 protein attracted postcrossing commissural axons. Commissural axons in mice lacking the Wnt receptor Frizzled3 displayed anterior-posterior guidance defects after midline crossing. Thus, Wnt-Frizzled signaling guides commissural axons along the anterior-posterior axis of the spinal cord.

Axonal connections are patterned along the anterior-posterior (A-P) and dorsal-ventral (D-V) neuraxes. Guidance molecules that play essential roles in the D-V guidance of axons have been identified, whereas the nature of the A-P guidance cues has remained an enigma (1, 2). The dorsal spinal cord commissural neurons form several ascending somatosensory pathways. During embryonic development, they project axons to the ventral midline (Fig. 1A). At the floor plate, commissural axons cross the midline, enter the contralateral side of the spinal cord, and make a sharp anterior turn toward the brain (Fig. 1B) (3). The initial ventral growth of the commissural axons is directed by a collaboration of two chemoattractants, netrin-1 (46) and Sonic hedgehog (Shh) (7), and chemorepellents of the bone morphogenetic protein (BMP) family (8). As the axons cross the midline, they lose responsiveness to these chemoattractants (9) but gain responsiveness to several chemorepellents, including Slit and semaphorin proteins (10), which guide axons from the D-V axis into the A-P axis (10).

Fig. 1.

Chemoattractant(s) guides commissural axons along the A-P axis. (A) Transverse section of E13 rat spinal cord showing the D-V trajectory (solid red line) and A-P trajectory (dashed red line) of commissural axons. (B) “Open-book” view of E13 rat spinal cord showing midline crossing and anterior turning. The green subpopulation of commissural axons projects anteriorly along a medial pathway, close to the floor plate (ventral funiculus). The red subpopulation projects along the floor plate initially but gradually fans out to occupy more lateral positions (lateral funiculus). Both populations project anteriorly immediately after midline crossing. (C) Agradient of diffusible guidance cue(s) might be disrupted when explants are cut shorter, causing misrouting of commissural axons along the A-P axis. (D) Agradient of nondiffusible guidance cue(s) will not be affected when the explants are cut shorter, and the axons should still project anteriorly. (E) In longer explants (3 mm), all postcrossing commissural axons projected anteriorly after midline crossing (fp, floor plate) after 16 to 20 hours of culture. Commissural axons were labeled by lipophilic DiI injection into the dorsal side of the explants by iontophoresis. (F) In shorter explants (0.5 mm), axons frequently stalled after midline crossing. (G) In shorter explants (0.5 mm), axons often projected randomly along the A-P axis. (H) Quantification of data. Anterior turn indicates normal projection. Knotting/stalling and random A-P turns are abnormal behaviors observed in shorter explants. n = number of explants. (I) Along explant (3 mm) was injected with DiI close to the anterior end, in the middle, and at the posterior end. The axons at anterior ends are almost always misrouted (∼90%). (J) Quantification of the open-book assays with anterior, middle, and posterior injections. n = number of injection sites. Scale bars, 100 μm.

To determine why axons turn in an anterior direction, we studied the turning of commissural axons after midline crossing in “open-book” spinal cord explants in culture. We observed that when a segment of E13 (embryonic day 13) rat spinal cord was cultured in collagen gel, commissural axons projected ventrally cross the midline and turned anteriorly within the explant (Fig. 1E). We reasoned that if A-P guidance is controlled by a diffusible gradient of either an attractant(s) or a repellent(s), then cutting these open-book explants shorter might eliminate the gradient and lead to abnormal pathfinding along the A-P axis (Fig. 1C); if A-P guidance is controlled by a nondiffusible cue(s), commissural axons will still have the normal anterior turn in shorter explants, because the gradient will be maintained (Fig. 1D).

We systematically cultured open-book explants of different A-P lengths (3 mm to 0.5 mm) and found that at 0.5 mm, abnormal pathfinding of postcrossing commissural axons was observed, including knotting and stalling (Fig. 1F) and randomized turning along the A-P axis (Fig. 1G). In contrast, all axons turned anteriorly in 3-mm explants (Fig. 1E). In both short and long explants, commissural axon pathfinding from the dorsal spinal cord to the floor plate was normal. Only 18% of the injection sites in the short explants showed normal anterior turning, presumably due to the loss of guidance information; this finding suggests that the guidance cue(s) are diffusible.

To address whether the A-P guidance cue is attractive or repulsive, we injected 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) focally into the dorsal spinal cord close to the anterior end, in the middle, and close to the posterior end (Fig. 1I) of long explants (3 mm or longer). Axons in the middle and close to the posterior end of the explants always projected anteriorly, whereas axons close to the anterior end almost always made mistakes, either stalling or projecting randomly along the A-P axis after midline crossing, similar to those in the short explants (0.5 mm). These results indicate that a gradient of an attractive cue(s) plays a role in the anterior turn of the postcrossing commissural axons (fig. S1). We further determined that a cut in the open-book explants did not prevent axons from projecting rostrally (fig. S2), indicating that the A-P gradient of the guidance cue(s) is preserved in such a preparation and a cut (damage) to the spinal cord itself does not produce a cue(s) to repel postcrossing commissural axons.

To identify the anterior turning signal, we tested candidate molecules by expressing them in COS cell aggregates positioned next to rat spinal cord explants of postcrossing commissural axons in collagen gels (10). Candidate molecules found in the limb bud were considered, because embryonic limb bud tissue stimulates extension of commissural axons only after midline crossing (11): hepatocyte growth factor (HGF), fibroblast growth factors 4 and 8 (FGF4 and FGF8), BMP4, BMP7, Shh, and Wnt1, Wnt4, Wnt5a, Wnt 6, and Wnt7b. Of these factors, only Wnt1 (11), Wnt4, Wnt5a, Wnt6 (11), and Wnt7b stimulated the extension of the postcrossing commissural axons (Fig. 2, B to D). HGF, the FGFs, the BMPs, and Shh did not show stimulation of postcrossing commissural axon growth in our assay (11). None of these Wnts affected the outgrowth of precrossing commissural axons (fig. S3).

Fig. 2.

Wnt proteins stimulate the extension of postcrossing commissural axons and expression of Wnts in the developing spinal cord. (A to D) Wnt proteins stimulated the extension of postcrossing commissural axons. (E) Quantification of postcrossing commissural axon extension stimulated by Wnts as described (10). (F) The growth cones of postcrossing commissural axons were enlarged and more complex in the presence of Wnt4. Large arrowheads indicate large and complex growth cones; small arrowheads indicate small and simple growth cones. Growth cones were stained with a rabbit antibody to GAP-43 and visualized in three dimensions by confocal microscopy. Addition of purified sFRP2 protein (0.2 μg/ml) reversed Wnt4-induced increase of growth cone size and complexity within 1 hour. Scale bar, 10 μm. (G) Quantification of growth cone morphology. Green bars indicate the percentage of large and complex growth cones (diameter ∼10 μm). Orange bars indicate small and simple growth cones (diameter ≤5 μm). (H to K) In situ hybridization of E11.5 mouse transverse sections (scale bar, 100 μm). (L) Nine serial sections along the A-P axis at 400-μm intervals probed by Wnt4. Arrows indicate floor plate. Scale bar, 100 μm. (M) Nine serial sections along the A-P axis at 400-μm intervals probed by netrin-1. These sections are the corresponding adjacent sections to those shown in (L). Arrows indicate floor plate. Scale bar, 100 μm. (N) Quantification of signal intensity for Wnt4. (O) Quantification of signal intensity for netrin-1. OD was determined on the basis of signal density using NIH ImageJ (http://rsb.info.nih.gov/ij/index.html).

In the presence of Wnt4, most growth cones of the postcrossing commissural axons (79%) became enlarged and complex (large arrowheads, Fig. 2, F and G). Recombinant secreted Frizzled-related proteins (sFRPs) are soluble proteins that can block the interaction of Wnts with their receptors, the Frizzleds (12); addition of sFRP2 reduced these effects within 1 hour (Fig. 2F). The percentage of large and complex growth cones was reduced to 50% after 1 hour, 37% after 2 hours, and 30% after 3 hours. In control cultures, 78% growth cones were small, regardless of whether sFRP2 protein was added (Fig. 2G). Although diffusion of molecules in collagen gel matrix is likely reduced compared to that in culture medium, the blocking effect of sFRP2 was still observed within 1 hour, suggesting that Wnt proteins likely stimulate postcrossing commissural axon growth by directly and acutely acting on the growth cones.

We examined the expression pattern of Wnts by in situ hybridization in developing mouse embryos during the stages when commissural axons are turning anteriorly. At E11.5 (equivalent to E13 in the rat), Wnt4, Wnt7b, and Wnt5a (11) were found expressed in the areas where the postcrossing axons turn anteriorly. Wnt4 was found specifically enriched in the floor plate and the ventricular zone (Fig. 2H), exhibiting a decreasing A-P gradient along the entire length of the floor plate from E10.5 (11) to E11.5 (Fig. 2L) and E13.5 (fig. S4). Netrin-1 did not display a decreasing A-P gradient (Fig. 2M). Wnt7b was found expressed on the two lateral margins of the floor plate, where anterior turning of postcrossing commissural axons occurs (Fig. 2I). Wnt5a was widely expressed in the spinal cord and was particularly abundant in the floor plate and the ventral areas of the spinal cord next to the lateral funiculus (11). Therefore, several Wnts are expressed in the correct place during development, with Wnt4 (from E10.5 to E13.5) displaying an A-P decreasing gradient in the floor plate. A similar Wnt4b gradient in the floor plate was found in zebrafish embryos at similar stages (13).

The Wnt receptor frizzled genes fz3, fz8, and fz9 were found expressed in the spinal cord from E9.5 to E13.5 during the time when commissural axons are making the anterior turn. fz3 is expressed in the area where commissural neuron cell bodies are located (arrowheads, Fig. 2J) and in the ventral and lateral margins of the spinal cord where postcrossing commissural axons project longitudinally (arrows, Fig. 2J).

To test whether Wnts are required for the proper anterior turning of the postcrossing commissural axons, we used sFRPs to block Wnt function in the open-book explants. sFRP-expressing COS cells were embedded in the bottom layer of collagen gel, and long open-book spinal cord explants were placed on top and embedded in the top collagen gel. The efficacy of this experimental system was first tested using netrin-1–expressing COS cells included in the bottom collagen gel, which stimulated outgrowth of precrossing commissural axons from explants placed in top collagen gel (fig. S5). In the presence of any of the three sFRPs (sFRP1, sFRP2, and sFRP3) or a mixture of all three sFRPs, anterior turning of commissural axons after midline crossing was severely impaired (Fig. 3, B to D). Axons either stalled or turned randomly along the A-P axis, displaying behaviors similar to those observed in the short explants and the anterior injection sites in the long explants (Fig. 1, F, G, and I). In contrast, in the presence of the control COS cells, all commissural axons turned anteriorly after midline crossing (Fig. 3A). No abnormal pathfinding behavior was observed in the precrossing segment of the commissural axons, which suggests that the Wnt signaling pathway is not required for the D-V projection of the precrossing commissural axons.

Fig. 3.

Wnts are required for proper anterior turning of postcrossing commissural axons in open-book explants, and Wnt4 can act as an instructive axon attractant. (A) When COS cells transfected with vector only were resuspended and embedded in the bottom collagen gel, postcrossing commissural axons displayed normal anterior turning in the open-book explants. (B to D) When sFRP-expressing COS cells were embedded in collagen gel, commissural axons displayed severe defects in the A-P axis, including knotting/stalling (asterisks) and A-P random turns (double-headed arrows). (E) Quantification (as in Fig. 1) of effects of sFRP1, 2, and 3, alone or combined. n = number of injection sites. (F) Vector-only COS cells did not rescue guidance defects. (G) Wnt4-transfected COS cells attracted postcrossing commissural axons toward them when positioned on the anterior end of the explants, rescuing the A-P pathfinding defects. (H) Vector-only COS cells did not reorient commissural axons after midline crossing. (I) COS cells expressing Wnt4 reoriented the postcrossing commissural axons to project posteriorly when positioned on the posterior side of the explants. (J) Quantification (as in Fig. 1) of Wnt4 rescue experiments. (K) Quantification of the Wnt4 reorientation experiments. n = number of injection sites. Light blue bars indicate the percentage of the injection sites whereby all axons turned posteriorly. Scale bars, 100 μm.

To further test the role of Wnts in A-P guidance, we examined whether applying an ectopic anterior source of Wnt protein(s) could rescue the anterior turn of commissural axons after midline crossing. Wnt4-expressing COS cell aggregates that were placed anterior to the short explants attracted postcrossing commissural axons and rescued A-P guidance defects (Fig. 3G). COS cells transfected with vector only had no effect (Fig. 3F). Furthermore, Wnt4-expressing COS cells placed posteriorly to the short explants redirected axons to turn posteriorly (Fig. 3I), which suggests that Wnt4 is an instructive cue rather than a permissive cue. In both experimental conditions, postcrossing commissural axons projected along a restricted trajectory between the floor plate and the ventral spinal cord, even though the Wnt4 source was presented in a wider area because of the size of the COS cell aggregates. This suggests that other guidance cues shape the trajectory and keep commissural axons within the narrow corridor immediately after midline crossing. For instance, the Slit proteins and a subset of secreted semaphorins play a role in “squeezing” postcrossing commissural axons between the floor plate and the ventral spinal cord as they exit the midline (10).

Midline pathfinding behavior of commissural axons in fz3 knockout embryos (14) was examined by DiI labeling and immunohistochemistry with a monoclonal antibody (mAb) to TAG-1, a commissural axonal marker present only in the precrossing and the mid-line-crossing segments of commissural axons. D-V projection of precrossing commissural axons was normal (Fig. 4B) compared to wild-type control (Fig. 4A), but postcrossing commissural axons projected randomly along the A-P axis after midline crossing with 100% penetrance along the length of the spinal cord (Fig. 4, D and E); this finding suggests that the Wnt-Frizzled pathway is specifically required for A-P axon guidance after midline crossing in vivo. No spinal cord patterning defects were observed in the fz3 knockout mice at this stage of development, as assessed by markers such as Nkx2.2, HNF-3β, Lim2, and Isl1 (14). LRP6 is a coreceptor of Frizzleds for the canonical Wnt–β-catenin signaling pathway, required in patterning and cell fate determination. Both D-V and A-P pathfinding of commissural axons were normal in LRP6–/– embryos, although patterning defects such as spina bifida, truncation of axial skeleton, and loss of distal limb structures were observed in these mice (15), which suggests that LRP6 is not involved in the differentiation, D-V pathfinding, or A-P guidance decision of commissural axons at the midline (11).

Fig. 4.

Frizzled3 is specifically required for A-P guidance of postcrossing commissural axons. (A) D-V trajectory of precrossing commissural axons in a wild-type embryo, as revealed by TAG-1 mAb. (B) Normal D-V projection of precrossing commissural axons in fz3 knockout embryos. (C) Commissural axons crossed the midline and turned anteriorly after midline crossing in an E11.5 wild-type littermate embryo. (D) Commissural axons reached the midline and crossed to the contralateral spinal cord but projected randomly along the A-P axis in an E11.5 fz3 mutant embryo. These axons are oriented randomly immediately after midline crossing. (E) A-P guidance defects of postcrossing commissural axons in an E11.5 fz3 mutant embryo slightly older than the one shown in (D). These axons extended further than those in (D). (F) Quantification of the postcrossing A-P guidance defects in fz3 knockout mice. Four litters of fz3 knockout mice were analyzed (three litters were analyzed in blinded experiments). A total of seven mutant embryos were analyzed. n = number of injection sites. Scale bars, 100 μm.

The nature of the molecular guidance cues controlling A-P pathfinding has long remained a mystery. Our results show that a Wnt-Frizzled pathway controls the anterior turning of spinal cord commissural axons after midline crossing via an attractive mechanism. A recent study suggested that a Wnt5-Derailed pathway mediates axonal repulsion in regulating the pathway selection of a subset of axons before entering the fly midline (16). We found that the mouse Derailed homolog Ryk (a receptor tyrosine kinase–like molecule) is not expressed in the spinal cord commissural neurons (11), providing evidence that Wnt proteins can act as axonal attractants via a signaling pathway involving Frizzled3 but independent of Ryk/Derailed.

Supporting Online Material

www.sciencemag.org/cgi/content/full/302/5652/1984/DC1

Materials and Methods

Figs. S1 to S5

References and Notes

View Abstract

Navigate This Article