Alternative splicing of Drosophila Dscam generates axon guidance receptors that exhibit isoform-specific homophilic binding

Abstract

Dscam is an immunoglobulin (Ig) superfamily protein required for the formation of neuronal connections in Drosophila. Through alternative splicing, Dscam potentially gives rise to 19,008 different extracellular domains linked to one of two alternative transmembrane segments, resulting in 38,016 isoforms. All isoforms share the same domain structure but contain variable amino acid sequences within three Ig domains in the extracellular region. We demonstrate that different isoforms exhibit different binding specificity. Each isoform binds to itself but does not bind or binds poorly to other isoforms. The amino acid sequences of all three variable Ig domains determine binding specificity. Even closely related isoforms sharing nearly identical amino acid sequences exhibit isoform-specific binding. We propose that this preferential homophilic binding specificity regulates interactions between cells and contributes to the formation of complex patterns of neuronal connections.

Figure 1. Alternative Splicing of Dscam Potentially Generates 38,016 Isoforms

(A) Schematic representation of Dscam gene, mRNA, and protein. The Dscam protein contains both constant and variable domains. The variable domains are encoded by alternative exons. Each block of alternative exons is indicated by a different color. A transcript contains only one alternative exon from each block. The Dscam gene encodes 12 alternative exons for the N-terminal half of Ig2 (red), 48 alternative exons for the N-terminal half of Ig3 (blue), and 33 alternative exons for Ig7 (green). There are two alternative transmembrane domains (yellow). (B) Schematic representation of Dscam proteins used in this study and explanation of isoform nomenclature. Immunoglobulin domain (Ig), horseshoe; fibronectin type III repeat (FNIII), black rectangle; transmembrane domain (TM), vertical rectangle. The NH₂ and COOH termini are indicated. Individual isoforms are denoted by the combination of alternative variable Ig domains. For instance, the isoform comprising Ig2 alternative 7, Ig3 alternative 27, and Ig7 alternative 25 is designated Dscam^7.27.25 or simply as 7.27.25. In this study, both full-length proteins and fragments of the extracellular domain were used. Full-length Dscam proteins expressed in COS cells and transgenic flies did not contain an epitope tag. Full-length proteins expressed in Drosophila S2 cells contained an internal FLAG epitope tag in the C-terminal cytoplasmic tail (denoted by a flag icon). For simplicity, the alternative transmembrane domain used in full-length isoforms is not indicated. Purified Dscam proteins containing different regions of the extracellular domain contained either a 6xHis tag (schematic not shown) or the Fc region of human IgG at the C terminus. Fc tags contain the hinge region, which dimerizes the Dscam proteins. 6xHis tags do not dimerize the Dscam proteins.

Figure 2. Dscam Promotes Interactions between Cells In Vivo

To assess whether Dscam can mediate interactions between cells in vivo, a transgene expressing a single isoform of Dscam was expressed in both midline cells and a subclass of interneurons using the Gal4/UAS system (see Experimental Procedures). (A) Schematic representation of the embryonic ventral nerve cord (VNC, gray) at three stages of development showing the spatiotemporal pattern of expression of two Gal4 driver lines used in this assay. The midline cell driver (M; green) is expressed before the interneurons (IN; red) cross the midline. During early stage 13, anterior commissure (AC) axons begin crossing the midline. The posterior commissure (PC) axons develop later and have crossed the midline by late stage 14. By this stage, PC interneuron cell bodies form a tight cluster. Embryos were analyzed at stage 16. Anterior is to the top. (B) Confocal section of a region of the VNC in a control embryo lacking the Dscam transgene. Interneurons (arrows) are labeled in red. (C) Confocal section of an embryo expressing transgenic Dscam in midline cells. Midline transgenic Dscam expression does not affect the trajectory of the interneurons (arrows). (D) Endogenous Dscam (green) is expressed throughout the VNC. (E) Confocal projection of an embryo expressing the Dscam transgene in midline cells (arrows). Strong midline staining is seen in addition to endogenous Dscam. (F) Confocal section of a control embryo lacking the Dscam transgene. Interneuron cell bodies form tight clusters, and axons cross the midline. (G–I) Confocal sections of three different embryos expressing the Dscam transgene only in interneurons (red). About 50% of PC axons show midline crossing phenotypes (arrows). In addition, cell bodies often fail to form tight clusters (arrowheads). (J) Confocal section of a control embryo lacking the Dscam transgene stained with midline (green) and interneuron (red) markers. (K–M) Three examples of embryos expressing the Dscam transgene in both midline cells and interneurons. PC axons rarely cross the midline, and processes remain on the ipsilateral side (arrows). Cell body phenotypes are also observed (arrowheads). (N) Quantification of PC axon midline crossing. The x axis indicates whether transgenic Dscam was expressed in midline cells (M), interneurons (IN), or both. Only 3.5% of PC axons from embryos expressing the Dscam transgene in both cell types crossed the midline. n = number of segments scored.

Figure 3. Dscam Binds to Itself In Vitro

Three assays were used to assess Dscam binding to itself. (A) Dscam promotes bead aggregation. Fluorescent beads were decorated with the extracellular domain of a Dscam isoform tagged with Fc, Dscam^7.27.25-Fc (blue), or Dscam^1.30.30-Fc (red). Beads were incubated with agitation, and a time course of aggregation over 30 min was monitored by FACS analysis. Aggregation was not observed for beads decorated with either Fc (green) or Fc fused to DCC (DCC-Fc, yellow), another Ig/FNIII-containing protein. Each time point represents the mean and SD of two independent binding experiments. Two different preparations of each protein were used. (B) Confocal images of beads taken from 30 min time point in (A). Dscam-Fc coated beads form aggregates, while control DCC-Fc and Fc coated beads remained largely monomeric. (C) (Left panel) Coomassie-stained gel of purified Fc proteins used in binding assays. (Right panel) Control beads (Fc and DCC-Fc) and Dscam-Fc beads contained similar levels of protein as assessed by immunoblotting using an anti-Fc antibody following the aggregation assay. (D) Dscam beads bind to cells expressing Dscam. Red fluorescent beads decorated with the indicated isoforms of Dscam-Fc were incubated with COS cells (green) expressing the same Dscam isoform or with control cells. Dscam-Fc beads bound to Dscam-expressing cells but not to control cells. (E) Dscam-Fc proteins bind to full-length Dscam in cell extracts. Dscam-Fc or Fc alone was incubated with extracts of Drosophila S2 cells transfected with full-length FLAG-tagged Dscam. Protein G Sepharose beads were used to bind Fc tags and pull down associated proteins. Dscam-Fc proteins pulled down Dscam-FLAG, but Fc control proteins did not. The same amount of Dscam-Fc or Fc protein (1 μg) was added to each pull-down experiment. The level of FLAG-tagged protein in each extract is shown in the lower panel. The flag icon denotes full-length FLAG-tagged Dscam. For assays used in (A), (D), and (E), see Experimental Procedures.

Figure 4. Dscam Binding Maps to a Region Containing the Variable Ig Domains

A C-terminal deletion series of the extracellular domain was tested for bead aggregation and binding to full-length Dscam. (A) Schematic representation of Dscam-Fc deletions used and summary of binding data. Red, blue, and green lines indicate variable regions in Ig2, Ig3, and Ig7, respectively. EC indicates extracellular domain. The number following EC denotes the number of domains (i.e., Ig [horseshoe] and FNIII [rectangle]) in each protein. The icon at the C terminus symbolizes the Fc tag. (B) Beads containing the eight N-terminal Ig domains promote aggregation. Fluorescent beads were decorated with different Dscam-Fc deletions, and a time course of bead aggregation over 30 min was monitored by FACS analysis. Proteins containing the N-terminal eight Ig domains (red) show binding indistinguishable from the entire extracellular domain. Proteins containing six or fewer N-terminal Ig domains (blue) did not support bead aggregation. Beads were decorated using equimolar amounts of protein. Each time point represents the mean and SD of two independent binding experiments. Two different preparations of each protein were used. (C) Confocal images of beads containing the proteins indicated at 30 min time point in (B). Beads coated with proteins containing the N-terminal eight Ig domains form aggregates similar in size to beads containing the entire extracellular domain, while beads decorated with proteins containing six Ig domains remained monomeric. (D) (Left panel) Coomassie-stained gel of purified Fc-tagged proteins used in binding assays. (Right panel) Immunoblots show that beads used in the aggregation assay were coated with equivalent levels of the indicated proteins. (E) Beads containing the eight N-terminal Ig domains bind to Dscam-expressing cells. Beads coated with different Dscam-Fc deletions were incubated with Dscam-transfected COS cells. Beads containing eight or more N-terminal Ig domains bound, whereas beads containing six N-terminal Ig domains did not. (F) Dscam-Fc proteins containing the N-terminal eight Ig domains bind to full-length Dscam in cell extracts. Dscam-Fc proteins were incubated with extracts of S2 cells transfected with full-length FLAG-tagged Dscam, and protein complexes were pulled down using protein G Sepharose beads. The ability of Dscam deletions to pull down full-length Dscam-FLAG was assessed by immunoblotting with anti-FLAG antibody. Proteins containing the eight N-terminal Ig domains bound to full-length Dscam, whereas proteins with six or fewer N-terminal Ig domains did not. The flag icon denotes full-length FLAG-tagged Dscam. The level of FLAG-tagged protein in each pull-down is shown in the lower panel.

Figure 5. Dscam Exhibits Isoform-Specific Binding

Two Dscam isoforms were tested for binding to each other. (A) Schematic representation of the two isoforms used in this experiment showing amino acid sequence identity between variable Ig domains. The percent identity is indicated. The icon at the C terminus indicates the Fc tag. (B) Beads decorated with one isoform of Dscam-Fc bind only to cells expressing the same isoform. Dscam^7.27.25-Fc and Dscam^1.30.30-Fc beads (red) were incubated with transfected COS cells and neurons from transgenic flies (green) overexpressing a single Dscam isoform as indicated. (C) Dscam-Fc proteins bind only to the same isoform of full-length Dscam in cell extracts. Dscam^7.27.25-Fc and Dscam^1.30.30-Fc were incubated with extracts of transfected S2 cells expressing full-length FLAG-tagged versions of these isoforms. The Fc-tagged proteins were captured by protein G Sepharose, and their association with FLAG-tagged Dscam proteins was assessed by immunoblotting with anti-FLAG antibody. The same amount of Dscam-Fc protein (1 μg) was added to each pull-down experiment. The flag icon denotes full-length FLAG-tagged Dscam. The level of FLAG-tagged protein in each extract is shown in the right panel.

Figure 6. Competition Assays Reveal Isoform-Specific Binding

Aggregation of Dscam EC16-Fc-coated beads was monitored in the presence of increasing amounts of soluble competitor, Dscam EC8-6xHis protein, of the same or a different isoform, as indicated. Beads were incubated with agitation, and a time course of aggregation over 45 min was monitored by FACS analysis. Bead aggregation for both Dscam isoforms was inhibited by the presence of the same competing isoform in a dose-dependent manner. Inhibition of bead aggregation was first detected between 30 and 100 μg/ml competitor and was complete by 3 mg/ml competitor. Bead aggregation for both Dscam isoforms was unaffected by competition with up to 3 mg/ml of the different isoform. Each time point represents the mean and SD of two independent binding experiments. Two different preparations of each Dscam-Fc protein were used.

Figure 7. Each Variable Ig Domain Contributes to Binding Specificity

To investigate which variable Ig domains mediate binding specificity, a series of swapping experiments was conducted. (A) Variable domain Ig2 contributes to isoform-specific binding. (Top panel) Schematic representation of isoforms tested for binding. These isoforms have different N-terminal halves of Ig2 (50% amino acid sequence identity) but are otherwise identical. The icon at the C terminus indicates the Fc tag. (Middle panel) Red fluorescent beads decorated with Dscam-Fc proteins were incubated with transfected COS cells (green) expressing full-length versions of the same or a different Dscam isoform, as indicated. Identical isoforms bound, while isoforms differing in Ig2 did not bind to each other in this assay. (Bottom panel) Dscam-Fc proteins were incubated with extracts of S2 cells expressing the same or a different isoform of full-length FLAG-tagged Dscam. The association of full-length Dscam-FLAG with Dscam-Fc in protein G Sepharose pull-downs was assessed by immunoblotting with anti-FLAG antibody. The same amount of Dscam-Fc protein (1 μg) was added to each pull-down experiment. Only identical Dscam isoforms bound to each other in these assays. (B and C) Variable domains Ig3 and Ig7 contribute to isoform-specific binding. Isoforms differing in Ig3 (B) and isoforms differing in Ig7 (C) were analyzed as described in (A). Identical isoforms bound to each other but did not bind to isoforms differing in either Ig3 or Ig7. The gray arrowheads indicate the variable Ig domain swapped in each experiment. The flag icon denotes full-length FLAG-tagged Dscam.

Figure 8. Closely Related Isoforms Differing by 7–12 Amino Acids Exhibit Preferential Isoform-Specific Binding

Binding specificity of a series of nearly identical isoforms was examined. (A) Schematic representation of isoforms tested for binding. All share the same Ig2 and Ig3 domains but differ in Ig7. The alternative Ig7 domains tested are highly related. Ig7.25 and Ig7.26 share the highest sequence identity of any two Dscam isoforms differing in only seven amino acids. The icon at the C terminus indicates the Fc tag. (B) Beads (red) decorated with Dscam-Fc isoforms, as indicated, bind to COS cells (green) expressing the same isoform. (C) Beads with isoforms differing from Dscam^7.27.25 show reduced or no binding to COS cells expressing Dscam^7.27.25. (D) Quantification of bead particle binding on COS cells. A bead particle may consist of one bead or an aggregate of them. Error bars depict SEM. n, number of cells counted. (E) (Top panel) Each Dscam-Fc isoform binds to full-length versions of the same isoform tagged with FLAG in pull-downs from extracts of transfected S2 cells. (Bottom panel) Dscam-Fc isoforms differing from Dscam^7.27.25-Fc do not bind to FLAG-tagged Dscam^7.27.25. Equivalent levels of Dscam-Fc protein (1 μg) and extract containing Dscam^7.27.25-FLAG were added to each pull-down. Arrowheads indicate variable Ig7 domains. Flag icon denotes full-length FLAG-tagged Dscam.

Figure 9. A Model for Dscam-Mediated Interactions

(A) Schematic representation of homophilic binding mediated by the three variable Ig domains. The eight N-terminal Ig domains (circles) are shown. Constant Ig domains are gray, and variable Ig domains are in color. The remainder of the protein is represented by a gray rectangle. Isoforms sharing identical variable Ig domains (represented by matching colors) bind to each other, while isoforms differing in only one variable Ig domain (gray arrowheads) do not. We propose that each variable Ig domain binds to the same variable Ig domain in an opposing molecule. As isoforms sharing any two identical variable Ig domains do not bind or exhibit reduced binding relative to binding between identical isoforms, it is likely that the binding of all three variable Ig domains is required to stabilize otherwise weak interactions between individual variable Ig domains. (B) Schematic representation of Dscam-mediated interactions between neurites. We propose that differences in levels of Dscam signaling influence the nature of the interactions between neurites. High levels of signaling between neurites expressing the same array of Dscam isoforms result in contact-dependent repulsion (left panel), while low levels of signaling between neurites expressing some of the same isoforms or isoforms that bind weakly to each other result in adhesion (middle panel). Some neurites may express isoforms of Dscam that do not interact (right panel). Dscam-mediated contact-dependent repulsion and adhesion may regulate interactions between neurites during axon guidance, targeting, or synapse formation.