Molecular insights into the axon guidance molecules Sidestep and Beaten path

The transmembrane protein Sidestep (Side) functions as a substrate-bound attractant for motor axons in Drosophila. Outgrowing motor axons recognize Side via Beaten path Ia (Beat) and migrate along Side-expressing tissues. Here, we report a structure-function analysis of these guidance molecules using a variety of mutant lines and transgenic constructs. Investigation of Side mutants shows that the exchange of a single amino acid (L241H) in the second immunoglobulin domain disturbs Side function and subcellular localization. Overexpression of Side and Beat deletion constructs in S2 cells and muscles demonstrate that the first Ig domains of both proteins are necessary for their interaction. Furthermore, subcellular distributions of several Beat constructs identify functional domains and suggest a potential posttranslational processing step in ER compartments. In fact, fusing full-length Beat at both the N- and C-terminus with GFP and mCherry, respectively, shows that the N-terminal domain is transported to the plasma membrane and exposed on the cell surface, while the C-terminal domain accumulated in the nucleus. Taken together, these results give insights into the interaction of Side and Beat and imply that Beat might be subject to proteolytic cleavage during maturation.

(A) Western blot of anti-GFP immunoprecipitates developed with anti-Side antibodies. For each lane, the indicated GFP-tagged Beat constructs were expressed in body wall muscles. The resulting lysates were combined with body wall lysates of Side-Cherry-expressing larvae. Beat proteins were immunoprecipitated using anti-GFP antibodies, separated by SDS-PAGE and transferred to PVDF membranes. Blots were probed with anti-Side antibodies. Beat-GFP, BeatnewTM-GFP and Beat_1-322-GFP but not BeatDIg1-GFP interacted directly or indirectly with Side-Cherry. Side-Cherry migrates at 150 kDa, whereas the lower band is believed to represent a degradation product. n = 2 independent experiments.
(B) Expression levels of various Beat proteins fused to GFP in larval muscles. Western blot of lysates from body wall preparations of dissected w 1118 larvae expressing the indicated Beat constructs in muscles using Mef2-Gal4. The blot was developed with anti-GFP antibodies (12s exposure time). GFP-Beat-Cherry was expressed at very high levels and showed a prominent band at 65 kDa. The remaining four fusion proteins ran at the expected sizes. Beat-GFP and BeatDIg1-GFP were notoriously difficult to detect in Western blots in our hands but were detectable at longer exposures (93s and 275s). w 1118 larvae showed some background signals. The same membrane was re-probed with anti-a-Tubulin antibodies as a loading control (20s exposure time). n = 3 independent experiments.

Supplementary Figure S2
Surface areas of cell aggregates formed in S2 cell-cell interactions assays.
S2 cells were transfected with the indicated Cherry-tagged Side constructs and then mixed with S2 cells expressing the indicated GFP-tagged Beat constructs. Mixed cell suspensions were incubated on a rocking platform to allow for cell-cell interactions. Side-Cherry induced large aggregates with Beat-GFP and BeatnewTM transfected cells but not with Beat_1-322-GFP transfected cells. The surface area of randomly selected aggregates was determined using Fiji is just ImageJ software. Two-tailed Mann-Whitney U test, n = 30, ns, non-significant, *** p≤0.001.

Supplementary Figure S3
BeatnewTM-GFP co-localizes with the ER marker GRP78 (also called BIP, Binding Immunoglobulin Protein) in muscles. BeatDTM-GFP are not affected by the nanobody, as the GFP tag is likely hidden in extracellular compartments. (D') BeatnewTM-GFP is strongly downregulated. (E') Expression levels and localization of GFP-Beat-Cherry is not affected as GFP is not exposed in the cytoplasm.
(A''-E'') Schematic models of the effects of the deGradFP system on various Beat constructs. Only fusion proteins with cytoplasmic GFP-tags are ubiquitinated (orange ellipses) and degraded or relocalized (green circles). Scale bar: 100 µm.

Supplementary Figure S5
Expression of a functional beat cDNA in non-neuronal tissues does not rescue beat mutant innervation phenotypes.
(A-F) Confocal images acquired through the transparent cuticle of third instar larvae showing dorsal, lateral and ventral body wall regions of the same abdominal hemisegment. Muscles and NMJs are stained with ShGFP (green), which is expressed in all genetic backgrounds. While expression of a fulllength beat cDNA in glia (B, Repo-Gal4), hemocytes (C, Serpent-Gal4) or fat body (D, Pumpless-Gal4) did not rescue beat mutant innervation phenotypes, expression in all neurons (Elav-Gal4) fully restored the innervation pattern (E). Splitting Beat in half by co-expression of an N-and C-terminal fragment of Beat in all neurons, however, did not rescue (F). Arrows mark correctly innervated muscles M9 (dorsal), M24 (lateral), M14 (ventral). Arrowheads mark the respective non-innervated fibers.
Other muscles lack NMJs as well but are not labelled. Scale bars: 20 µm.

Supplementary Figure S6
Quantification of the rescue ability of untagged Beat in various tissues.