Function of an Axonal Chemoattractant Modulated by Metalloprotease Activity

See allHide authors and affiliations

Science  25 Aug 2000:
Vol. 289, Issue 5483, pp. 1365-1367
DOI: 10.1126/science.289.5483.1365


The axonal chemoattractant netrin-1 guides spinal commissural axons by activating its receptor DCC (Deleted in Colorectal Cancer). We have found that chemical inhibitors of metalloproteases potentiate netrin-mediated axon outgrowth in vitro. We have also found that DCC is a substrate for metalloprotease-dependent ectodomain shedding, and that the inhibitors block proteolytic processing of DCC and cause an increase in DCC protein levels on axons within spinal cord explants. Thus, potentiation of netrin activity by inhibitors may result from stabilization of DCC on the axons, and proteolytic activity may regulate axon migration by controlling the number of functional extracellular axon guidance receptors.

There is evidence that proteolytic activity present on neuronal growth cones may regulate their migratory activity (1). This possibility is supported by the presence of axonal stalling defects in flies with mutations in the kuzbanian gene, which encodes a member of the ADAM class of metalloproteases (2). ADAM class metalloproteases are also implicated in other biological processes, including fertilization (3), lateral inhibition during neurogenesis (4), and protein ectodomain shedding of a variety of ligands and receptors [(5–7), reviewed in (5)].

The netrin-1 protein and its receptor, DCC, are required for commissural axon outgrowth in vitro and for the proper guidance of spinal commissural neurons in vivo (8–12). We previously characterized an unidentified proteinaceous activity, termed netrin-synergizing activity (NSA), that potentiates the axon outgrowth–promoting effects of netrin on rat E11 (embryonic day 11) dorsal spinal cord explants (7, 13). In a screen of several dozen known molecules (including many axon guidance molecules), none could potentiate netrin activity (13). Upon further screening, we found that netrin-1 activity was potentiated by IC-3, a specific hydroxamate inhibitor of metalloproteases (6,14) (Fig. 1). In the absence of any factors, or in the presence of a low concentration of netrin-1 (∼50 ng/ml), very little outgrowth was observed from these explants (Fig. 1, A and B). Robust outgrowth was observed from explants grown in the presence of both 40 μM IC-3 and netrin-1 (∼50 ng/ml) (Fig. 1D), which was greater than that observed with a high concentration of netrin-1 (1 μg/ml) alone (Fig. 1E). In contrast, much less outgrowth was observed from explants grown in the presence of 40 μM IC-3 alone (Fig. 1C); the fact that any outgrowth was observed may reflect a potentiation of low levels of endogenous netrin-1 present in dorsal regions of the spinal cord (9). The responsive axons in these assays are commissural, as assessed by expression of the commissural axon markers TAG-1 and DCC (15). The degree of outgrowth observed in the presence of both IC-3 and netrin-1 (thick and long axon fascicles emerging primarily but not exclusively from the ventral cut edge of the explants) was similar to that observed with netrin-1 and NSA (8,13). Quantification of axonal outgrowth (estimated as total axonal length) with IC-3 alone or IC-3 plus netrin-1 (Fig. 1F) revealed peak activity of IC-3 at 80 μM, and only slight activity at 20 μM.

Figure 1

The chemical metalloprotease inhibitor IC-3 potentiates netrin-mediated axon outgrowth from E11 rat dorsal spinal cord explants (23). (A) Explant cultured alone. (B) In the presence of netrin-1 (∼50 ng/ml), there is little outgrowth, as in (A). (C) Some outgrowth is observed in the presence of 40 μM IC-3. (D) Robust outgrowth is observed in the presence of both 40 μM IC-3 and netrin-1 (∼50 ng/ml). (E) A high concentration of netrin-1 (1 μg/ml) also elicits robust outgrowth. In all panels, dorsal is top and the ventral cut edge of the explant is at the bottom; scale bar, 200 μm. (F) E11 dorsal spinal cord explants were cultured for 36 to 40 hours in the presence (diamonds) or absence (squares) of netrin-1 (∼50 ng/ml) and increasing concentrations of IC-3. The outgrowth under each condition was quantified by measuring the total length of all the axon fascicles emerging from each explant [a useful measure of outgrowth in response to netrin-1 (8)]. The data are from three independent experiments ± SEM. Where no error bars are visible, the symbol is larger than the error bars.

The effect of IC-3 is likely to be due to its metalloprotease inhibitor activity because this compound has not been found to have any effect other than metalloprotease inhibition in a wide variety of cellular assays (including protein phosphorylation and cell viability) or in mice (14). To confirm this, we also compared the actions of a chemically distinct hydroxamate metalloprotease inhibitor, GM6001, and an inactive control isomer of this compound (16). As expected, GM6001 but not the control isomer showed a similar potentiation of netrin-1 activity to IC-3 (15), consistent with the potentiation of netrin-1 being due to inhibition of metalloprotease activity. Members of another class of inhibitors that function through chelation of the divalent cations required for metalloprotease function (including 1,10-phenanthroline and EDTA) could not be tested in these assays because they were toxic to the explants (15).

Dorsal spinal cord explants grown in the presence of IC-3 exhibited much brighter staining for DCC than did control axons grown in the absence of any factors (compare Fig. 2, C and D). This effect was specific for DCC, because explants grown with or without IC-3 and stained for TAG-1 exhibited no clear difference in staining intensity except at the ventral cut edge of the explant (Fig. 2, A and B). Quantification of fluorescence intensity across the dorsoventral axis of DCC-stained explants revealed an increase over almost the entire dorsoventral length of the explants (Fig. 2E).

Figure 2

IC-3 increases DCC but not TAG-1 protein levels in E11 rat dorsal spinal cord explants (24). (A) Explant cultured without factors and stained with TAG-1 mAb 4D7. (B) Explant cultured with 40 μM IC-3 and stained with 4D7. Although the explant shown may appear slightly brighter than that in (A), this was not observed consistently; the difference shown is no greater than that seen between multiple explants cultured under the same conditions. The only reproducibly noticeable difference in TAG-1–stained explants is the region of brighter staining at the ventral cut edge of explants grown in the presence of IC-3 [arrow in (B)]. This may reflect more complete and/or faster growth of commissural axons to the ventral cut edge within these explants. (C) Control explant stained with DCC mAb AF5. (D) Explant cultured with 40 μM IC-3 and stained with DCC mAb. Explants are oriented as in Fig. 1; scale bar, 200 μm. (E) Quantification of the fluorescence intensity of DCC mAb staining along the dorsoventral axis of a control explant [line along which intensity is measured is shown in (C)] and an explant cultured in the presence of 40 μM IC-3 [line along which intensity is measured is shown in (D)]. The explants shown here represent the average intensity difference within multiple pairs of observed explants.

Because IC-3 increased DCC staining levels within dorsal spinal cord explants, we tested whether DCC might be a substrate for metalloprotease-dependent ectodomain shedding (17). Transfected CHO cells expressing DCC were metabolically labeled and then chased with medium that included or omitted IC-3. Cell extracts and supernatants were analyzed by immunoprecipitation with a monoclonal antibody to the extracellular domain of DCC (DCC mAb). A 160-kD protein was immunoprecipitated from extracts of DCC-transfected CHO cells but not from control cells (Fig. 3A). In addition to the presence of full-length DCC in cell extracts, a protein of ∼130 kD, the predicted size of the entire DCC ectodomain, was also immunoprecipitated from the supernatant of DCC cells (Fig. 3B). This protein was absent from control transfected cells and was the same size as an immunoprecipitated protein from transfected cells expressing a soluble form of the complete DCC ectodomain (Fig. 3B). The metalloprotease inhibitors IC-3, GM6001, and 1,10-phenanthroline all abolished the presence of the ∼130-kD protein from supernatants of cells transfected with full-length DCC, whereas the control isomer of GM6001 did not (Fig. 3B; only the IC-3 result is shown here) (15); these findings suggest that the protein resulted from a metalloprotease-dependent cleavage near the transmembrane region of DCC. We also investigated whether DCC could be shed from primary cultures of dissociated rat dorsal spinal cord (18). A protein corresponding to full-length DCC was visible in cell extracts that had been immunoprecipitated with DCC mAb, but not in extracts that were mock-immunoprecipitated with only the secondary antibody (Fig. 3C). A ∼130-kD protein corresponding to the DCC ectodomain could be immunoprecipitated from the chase medium supernatant only if DCC mAb was included (Fig. 3D). The presence of IC-3 or GM6001 in the chase medium abolished the presence of this protein (1,10-phenanthroline could not be tested because it induced neural cell detachment). These results indicate that dissociated commissural neurons can shed, in a metalloprotease-dependent manner, the ectodomain of full-length DCC.

Figure 3

DCC is a substrate for metalloprotease-dependent ectodomain shedding from transfected CHO cells and dissociated dorsal spinal cord neurons (17, 18). (A) DCC immunoprecipitations from cell extracts of metabolically labeled CHO cells chased in the presence or absence of IC-3 (indicated above each lane) and previously transfected with one of three plasmids: control vector (C), an expression plasmid encoding full-length DCC (FL), or an expression vector encoding a DCC mutant molecule that lacks the intracellular and transmembrane domains and produces the entire DCC ectodomain as a secreted protein (E). Control CHO cell extracts contained no DCC-immunoprecipitable proteins, whereas extracts from FL transfected cells contained a protein of 160 kD, the predicted size of full-length DCC. The secreted extracellular domain can be seen as a faint band within cell extracts at ∼130 kD. (Molecular mass markers are indicated to the left of each autoradiogram; the upper marker in each case is 200 kD and the lower marker is 116 kD.) (B) DCC immunoprecipitations from supernatants of the metabolically labeled CHO cells whose cell extracts are shown in (A). Control supernatants contained no DCC-immunoprecipitable protein, whereas FL transfected CHO cells contained a protein of ∼130 kD that was the same size as the constitutively expressed DCC extracellular domain. The presence of the extracellular DCC fragment in the supernatant was abolished by inclusion of IC-3 in the chase medium. (C) DCC or mock immunoprecipitation from cell extracts of dissociated E13 commissural neurons chased in the presence or absence of IC-3. A faint band corresponding to full-length DCC (arrow) can be seen in the presence of DCC mAb (α-DCC). The band above the 116-kD marker is nonspecific and is present in all immunoprecipitations from extracts of dissociated dorsal spinal cord cells. (D) DCC or mock immunoprecipitations from supernatants of the dissociated E13 commissural neurons chased in the presence or absence of IC-3 in (C). The ∼130-kD protein observed in the presence of DCC mAb and absence of IC-3 corresponds to an ectodomain fragment of DCC. Close examination of this band on this and other gels reveals it to be a doublet, suggesting that there is more than one cleavage of DCC from these neurons. Chase media containing IC-3 and/or mock immunoprecipitations do not contain a DCC-immunoprecipitable protein. Note that the percentage of DCC shed from CHO cells during the 1-hour chase was far less than that shed from neurons, because an increase in full-length DCC in extracts of cells exposed to IC-3 was only occasionally detected in the former (A) but was readily detected in the latter (C).

Although the axon outgrowth potentiation observed with metalloprotease inhibitors is similar to that observed with NSA (8, 13), it is not clear whether NSA acts by the same mechanism, because it failed to increase DCC expression within E11 dorsal spinal cord explants and failed to protect DCC from metalloprotease-dependent ectodomain shedding (15). It is possible that NSA acts in a mechanistically distinct manner that does not involve metalloprotease inhibition. Alternatively, it is possible that both NSA and IC-3 act as metalloprotease inhibitors, but that NSA (and perhaps also IC-3) inhibits degradation of some extracellular component other than DCC that is necessary for commissural axon guidance.

Our findings indicate that dorsal spinal cord explants display a metalloprotease activity that mediates (directly or indirectly) the proteolytic degradation of the netrin receptor DCC to a presumably nonfunctional form, and that the inhibition of this metalloprotease activity leads to enhanced responsiveness to soluble netrin-1. This functional effect may result from inhibition of DCC cleavage, although we cannot exclude the possibility that it may also result partly or entirely from inhibition of cleavage of other substrates. Nonetheless, our results imply that the balance of metalloprotease activity and metalloprotease inhibitory activity within the dorsal spinal cord or along the trajectory of commissural neurons may be an important regulator of commissural axon guidance. Interestingly, it was recently shown in Caenorhabditis elegans that the MIG-17 metalloprotease, a member of the ADAM family, is required for proper migration of distal tip cells (DTCs), and a possible role for MIG-17 in regulating the function of the netrin UNC-6 was suggested by interactions between mig-17 and unc-6 mutant alleles in DTC migration (19). Thus, metalloprotease regulation of netrin and/or netrin receptor function may be phylogenetically conserved and involved in both axon guidance and cell migration.

The ability of axons to alter their responses to guidance cues as they progress through the environment is known to be essential for their progression from one intermediate target to the next (20,21), and there is evidence that altered axonal responsiveness may result from increased or decreased insertion of receptor protein in the axonal membrane (21) or by alteration of downstream signaling pathways through activation of modulatory second messenger systems (22). The regulated shedding of axon guidance receptor ectodomains provides an additional mechanism through which axonal responsiveness to guidance cues can be regulated to help shape the precise trajectories of axons that are essential to the proper wiring of the nervous system.

  • * To whom correspondence should be addressed. E-mail: marctl{at}


View Abstract

Navigate This Article