Axons Guided by Insulin Receptor in Drosophila Visual System

See allHide authors and affiliations

Science  18 Apr 2003:
Vol. 300, Issue 5618, pp. 502-505
DOI: 10.1126/science.1081203


Insulin receptors are abundant in the central nervous system, but their roles remain elusive. Here we show that the insulin receptor functions in axon guidance. The Drosophila insulin receptor (DInR) is required for photoreceptor-cell (R-cell) axons to find their way from the retina to the brain during development of the visual system. DInR functions as a guidance receptor for the adapter protein Dock/Nck. This function is independent of Chico, the Drosophila insulin receptor substrate (IRS) homolog.

Insulin receptors in the central nervous system have been implicated in control of food uptake, learning, and memory, and pathophysiologies such as Alzheimer's disease (16). Drosophila harbor one receptor tyrosine kinase of the insulin receptor family (79), which avoids genetic redundancy in mammals that have three members of the insulin receptor family (10). DInR is expressed ubiquitously throughout the fly life cycle and is required for viability (11, 12), longevity, and female fertility (13, 14). Some combinations of hypomorphic alleles support survival, producing animals that are developmentally delayed and smaller than wild type (11, 12, 15). The growth-related phenotypes are likely mediated through Chico, a Drosophila IRS-like protein (16).

To identify additional downstream signaling partners, we used the DInR intracellular domain as bait in a yeast two-hybrid screen and identified Dreadlocks (Dock, Fig. 1A). Dock, a homolog of mammalian Nck, is an adaptor protein composed of one SH2 and three SH3 domains (1719). Interactions between DInR and Dock depend on DInR's C-terminal tail (11, 20), which contains tyrosine phosphorylation sites and proline-rich sequences and is thought to mimic some functions of insulin receptor substrates (IRSs) (21, 22). A kinase-inactive form of DInR [in which Lys1358 is replaced by Ala (K1358A)] did not interact with Dock. DInR-K1358A was expressed at levels comparable to that of the wild type but was not detectably autophosphorylated (Fig. 1A, inset). Additional yeast two-hybrid assays indicated that DInR interacts with both the SH2 and SH3 domains of Dock (fig. S1). The absolute requirement for DInR autophosphorylation likely reflects both ligand-induced phosphotyrosine interaction(s) with Dock's SH2 domain and autophosphorylation-induced conformational change that allows the C-terminal tail to bind Dock's SH3 domains. This dual interaction is consistent with the finding that Dock's SH2 and SH3 domains can partially substitute for each other to support R-cell axon guidance (19).

Fig. 1.

DInR and Dock form a stable complex and colocalize in vivo. (A) DInR and Dock interacted in yeast two-hybrid assays. Interactions were quantified by using standard β-galactosidase units (33). Vectors were pACT-Dock and pAS2.1-DInR, or derivatives, as indicated. (Inset) Wild type, but not DInR-K1358A, is tyrosine-autophosphorylated in yeast. Western blots of yeast extracts expressing pAS2.1-DInR intracellular domain (WT) or pAS2.1-DInR-K1358A (K-A) were probed with DInR-specific antibody (left panel) or phosphotyrosine antibodies (right panel). (B) DInR is enriched in R-cell projections. Immunostaining of third-instar larval eye-brain complexes. (i) DInR-specific antibody (green); (ii) MAb24B10 (red) staining R-cell projections; (iii) overlay. (C) DInR and Dock form a stable complex in vivo. (i) Wild-type adult fly lysate immunoprecipitated with unrelated immunoglobulin G (IgG); Nck-specific antibody; or protein A–Sepharose (PAS). DInR-specific antibody in a Western blot revealed full-length receptor (∼170 kD) and cleaved C-terminal tail (∼60 kD) in Dock immunoprecipitates. (ii) Wild-type adult fly lysate precipitated with preimmune serum; DInR-specific antibody; or PAS. Nck-specific antibody on Western blot revealed Dock (∼47 kD) in DInR immunoprecipitates.

Dock is required for R-cell axon guidance and is expressed in the neuropils of the lamina and medulla where R-cell growth cones terminate (18, 19). Although we detected DInR ubiquitously, it was markedly enriched in R-cell axon projections and growth cones of third-instar larval eye-brain complexes (Fig. 1B), as is Dock (18, 23), although the relative enrichment in R-cell projections is more pronounced for DInR than Dock.

To test whether DInR and Dock form a stable complex in vivo, we prepared lysates from wild-type adult flies and immunoprecipitated them with an antibody against human Nck that recognizes Dock (24), or with DInR-specific antibody. When we used DInR antibodies in a Western blot of immunoprecipitates from the addition of Nck antibodies (Fig. 1C, i), we found an ∼170-kD species (full-length DInR) and an ∼60-kD species [proteolytically cleaved C-terminal tail (11)], which were absent from controls. Dock (∼47 kD) was detected in immunoprecipitates from DInR-specific antibody but not controls (Fig. 1C, ii). We substantiated this finding using lysates from adult heads (AH) and third-instar larval eye-brain complexes (EB), which also verified the absence of DInR or Dock immunoreactivity from dock mutant animal immunoprecipitates (fig. S2). Together, these results demonstrate that Dock and DInR are associated in a stable complex in wild-type animals.

To determine whether DInR affects axon targeting in the visual system, we used the eyFLP-FRT system (25) to generate homozygous dinr mutant tissue in a heterozygous background. In dinr+/+ control animals (Fig. 2, A and B), as in wild type, axon projections from each ommatidium exited the eye disc in a bundle through the optic stalk. Axons spread in a stereotypic fashion in the optic lobe (2628). Projections from R1 to R6 cells terminated in the lamina. Axons from R7 and R8 extended through the lamina, and growth cones expanded at their termination points in the medulla. For dinr mosaic animals (Fig. 2, C to F), axons projected normally through the optic stalk but failed to target properly in the optic lobe. Gaps in the R-cell array were evident, and fibers crossed, indications of premature termination. The lamina neuropil was uneven with gaps and abnormally densely packed regions. In the medulla, projections failed to form the staggered array seen in wild type, and growth cones terminated in thick blunt ends. All of these phenotypes were indistinguishable from defects observed in dock mutant animals (Fig. 2, G and H) (18) or dock mosaics generated with eyFLP (Fig. 2I). We found that homozygous mutant clones were generally larger for dock than dinr, such that defects in dock mosaics appeared more severe. However, in rare cases, large dinr–/– mutant clones were recovered and, in these cases, phenotypes were equally strong (Fig. 2J).

Fig. 2.

DInR is required for generation of a retinotopic map in Drosophila. Eye-brain complexes of third-instar larvae stained with MAb24B10. See table S1 for quantification. (A to F; I to K) Mitotic clones generated with eyFLP (25). The pairs (A, B), (C, D), and (E, F) show two different animals each for the indicated genotype, at lower and higher magnification. (A and B) Mitotic clones in an FRT82B, dinr+/+ background as control. Retinotopic map was indistinguishable from the map of wild type. Growth cones from R1 to R6 terminated in the lamina (la). Axons from R7 and R8projected into medulla (med), where growth cones expand. (C and D) dinrex15 mosaics. Axons crossed and terminated prematurely (arrow), causing gaps (arrowheads) and densely packed regions in the lamina. Axons projected into the medulla failed to expand, resulting in a blunt-ended morphology (D, med). (E and F) dinr353 mosaics. Phenotypes were similar to but milder than for dinrex15. (G and H) dockP2 mutant animal. Note crossing of axon projections (arrows), gaps in the lamina (arrowheads), and blunt ended termination points in the medulla (med). (I) dockp1 mosaic. Gaps (arrowheads) and clumps of axons (double arrowhead) are indicated. (J) dinrex15 mosaic exhibiting a more severe phenotype: large gaps (arrowheads) and clumps of axon bundles (double arrowhead) are seen, similar to (I). Axon projections were indistinguishable from control in (K) chico1 mosaic and (L) chico1 animal. (M to P) Surviving dinr hypomorphs displayed axon guidance defects. (M) dinr273/273 larva. Gaps (arrowhead) and clumps of axons (double arrowhead) at lamina layer indicated. (N) dinr273/353 larva. Multiple gaps (arrowheads), fiber crossing (arrow) and clumps of axons (double arrowhead) in the lamina. Growth cones were unexpanded in the medulla. (O and P) dinr273/353 severe phenotype. Axons formed clumps and large gaps. Growth cones failed to expand in the medulla.

DInR is thought to influence cell size through interactions with Chico (15, 16). To test whether DInR-associated axon guidance phenotypes depend on Chico, we examined mitotic chico1 clones generated with eyFLP, and homozygous chico1 mutant animals (Fig. 2, K and L). R-cell projection patterns were indistinguishable from wild type in both cases. Thus, effects of DInR on axon pathfinding are independent of Chico. Further, the absence of guidance defects in chico1 mosaic animals, where clones were small, demonstrates that dinr-associated guidance defects are not the result of small R-cells sending projections into a wild-type brain.

To examine axon guidance in whole mutant animals, we isolated eye-brain complexes from surviving dinr273/273 and dinr353/273 third-instar larvae (Fig. 2, M to P). For dinr273/273, we observed mild axon guidance defects, including gaps and abnormally high density (“clumping”) of axons in the lamina. For surviving transheterozygous dinr353/273, more dramatic defects were noted: gaps in the lamina, crossing of axon fibers, and failure of growth cone expansion in the medulla. Forty percent of these animals displayed severe defects similar to those originally reported for dock: Axons showed no organized pattern of migration through the brain and appeared to terminate at arbitrary positions; clumping and large gaps were observed, and growth cone expansion in the medulla did not occur.

Axon guidance defects were also observed in dinr mosaic adult brains. In wild-type fruit flies, growth cones from R8 and R7 projections terminate in even rows (Fig. 3A, dinr+/+ control). Similar patterns were observed for chico1 mosaics (Fig. 3B). Mosaic dock animals exhibit varying defects including fiber crossing, uninnervated regions, and gaps in the R7 layer (Fig. 3C). Similar defects were observed for animals carrying dinrex15 mitotic clones (Fig. 3D).

Fig. 3.

dinr mutant R-cells exhibit axon-targeting defects. (A to D) DInR is required for axon targeting in adult brain. R-cell projections in brains of newly hatched adult heads stained with MAb24B10. Mitotic clones were generated with eyFLP.(A) FRT82B, dinr+/+ control. Multiple mitotic clones were observed in the eye disc. R7 and R8growth cones terminated in even rows in the medulla. (B) chico1 mosaic. Axon projection patterns similar to those in (A). (C) dockp1 mosaic. Multiple mitotic clones were observed. Gaps were evident (arrow). (D) dinrex15 mosaic. Only small clones were recovered, which revealed that projections were normal in most of the brain. Boxed area shows defects associated with dinrex15/ex15 clones: gaps in R7 layer (arrows) and fibers crossed (arrowhead). (E to J) DInR mutant R-cells show guidance defects. Third-instar larval (E to G) or adult (H to J) brains stained with DInR-specific antibody (green) and MAb24B10 (red). (E to G) dinr353 mosaic; medulla shown. Where DInR was strongly expressed, growth cones expanded. In regions with decreased DInR levels (brackets), growth cones did not expand. (H to J) dinrex15 mosaic. R7 and R8targeted to the correct layers where DInR was expressed; both layers were disrupted where DInR was not expressed. (K to P) The MARCM system to identify dinr mutant R-cells. Homozygous mutant R-cells were positively labeled by an antibody against green fluorescent protein (GFP) (green), and projection patterns were assessed with MAb24B10 (red). See table S2 for quantification. (K) FRT82Bdinr+/+ control. Multiple clones were observed. R7 and R8targeted correctly in the medulla. (L, M, N, and O) dinrex15 mosaic. In most cases, only small, single clones were observed. (M) GFP staining identifying mutant axon (N) overlay. (L) An example where multiple clones where recovered. (P) dinr353 mosaic. Defects were similar to dinrex15, but most mutant R7 cells targeted correctly. Gaps in R7 layer and crossing of mutant fibers were observed.

To determine whether axon guidance defects occur in dinr mutant R-cells, we made use of the fact that immunoreactivity to DInR is reduced in dinr353/353 and undetectable in dinrex15/ex15 mutant cells (fig. S3) to identify dinr–/– mutant cells in mosaic eye-brain complexes. If guidance defects occur in dinr–/– cells, these should be detected with the monoclonal antibody (MAb) MAb24B10, but should show little or no immunoreactivity to DInR-specific antibody. In eye-brain complexes of dinr353 mosaic larvae, regions of the optic lobe innervated by mutant R-cells were identifiable by decreased levels of DInR-specific immunoreactivity (Fig. 3E, bracket). Axon guidance defects were observed here; mutant R-cells failed to expand at their termination points in the medulla. In contrast, neighboring R-cells showed a normal expanded morphology (Fig. 3, F and G). This was also evident in preparations from late pupal eye-brain complexes (Fig. 3, H to J). In dinrex15 mutant patches, an abnormal pattern of R-cell projections was observed (Fig. 3, I and J, bracketed area). Gaps in the array occurred and R-cell projections failed to reach their targets. These defects occurred in dinrex15 mutant R-cells, as evidenced by the absence of DInR-specific immunoreactivity (Fig. 3H, bracketed area), but were not evident in neighboring wild-type cells.

The MARCM technique (29, 30) was used to positively label dinr mutant R-cells. For dinr+/+ control clones (Fig. 3K), projection patterns were indistinguishable from those of wild type. Homozygous dinrex15 R-cell projections (Fig. 3, L to O) displayed the defects in axon targeting described above: gaps in the R7 layer and crossed fibers. Similar defects were observed in dinr353 mutant R-cells, although defects were less severe (Fig. 3P).

Guidance defects were not a secondary consequence of aberrant R-cell fate determination or differentiation. Ommatidia were small but had normal arrangement and morphology; R-cell and glial-cell differentiation occurred normally, and proliferation and differentiation in the optic lobe were unaffected (fig. S4).

To test genetic interactions between dinr and dock, eye-brain complexes from adult +/dock; +/dinrex15 animals were examined (Fig. 4, F and G). Axon guidance phenotypes resembled those seen for dockP1 mosaics (Fig. 4E): uninnervated areas were observed, and gaps in the R7 layer were evident. Similarly, guidance defects were observed in +/dockP1; +/dinrex15 and +/dockP2; +/dinrex15 larvae (Fig. 4, J to L). dockP1 and dockP2 were independently induced, providing a control for genetic background effects. No obvious defects were observed in +/chico1; +/dinrex15 animals (Fig. 4H). These observations support the notion that DInR signals through Dock to regulate axon targeting in the visual system.

Fig. 4.

DInR and Dock interact to regulate axon guidance. R-cell projections assessed with MAb24B10 staining. See table S3 for quantification. (A) Wild type. Axons projected into the medulla forming an even array. Axon projections in the presence of one copy of (B) dockp1, (C) dinrex15,or(D) chico1, were indistinguishable from wild-type fruit flies. (E) dockp1 mosaic animal demonstrates guidance phenotypes including gaps in the R7 array (arrows). (F) +/dockp1; +/dinrex15 animals showed defects: gaps in R7 terminal array (arrows) and uninnervated areas (arrowheads). (G) +/dockp2; +/dinrex15 phenotypes, as in (F). (H) +/chico1; +/dinrex15 animals displayed normal projection patterns. (I to L) Axon projections as seen in larvae; genotypes are indicated.

This study demonstrates that DInR, an IR-family member, is necessary for axon guidance and targeting in the Drosophila visual system. Animals carrying dinr mutations elaborate abnormal retinotopic connections, characterized by gaps and crossing of axons in the lamina, clumping of axon bundles and failure of growth cone expansion in the medulla. DInR interacts directly with Dock, in vitro and in vivo, to form a stable complex. The interaction requires DInR kinase activity and the C-terminal tail that contains SH2 domain–binding sites; these requirements suggest that DInR recruits Dock in response to an external signal that stimulates autophosphorylation. Thus, DInR functions as a guidance receptor upstream of Dock in the visual system. Dscam is a partner of Dock in the embryonic central nervous system and Bolwig's nerve (31) but not in R-cell axon guidance (32). Thus, different cell surface receptors appear to activate the Dock pathway at different developmental stages. We speculate that vertebrate IR action involving learning, memory, and eating behavior involves changes in neuronal targeting that use pathways similar to that uncovered here for the Drosophila IR.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Tables S1 to S3


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article