BIOLOGICAL OVERVIEW

Before a closer look at Plexin A, the subject of this overview, it might be useful to look at the Plexin family of proteins in terms of its relationship to two other protein families: Semaphorins and Neuropilins.

To date, searching the genomes of the nematode C. elegansand the fruit fly Drosophila has not revealed homologs for Neuropilins, the receptors (or receptor subunits) for class III Semaphorins (He, 1997 and Kolodkin, 1997). Nematodes and fruit flies also appear to lack class III Semaphorins; worms have classes I and II, while flies have classes I, II, and V. Both these invertebrate genomes also appear to lack MET tyrosine kinase receptor homologs (serving as receptors for semaphorins in vertebrates), but they do contain genes encoding Plexins. With the lack of Neuropilins and MET tyrosine kinase receptors in worms and flies, it is suggested that Plexins are the oldest Semaphorin-binding proteins. It may be that during chordate evolution, a new class of Semaphorins (type III) evolved, and with them, a new class of Semaphorin receptors (the Neuropilins) (Winberg, 1998 and references therein).

What proteins serve as receptors for Semaphorins? Neuropilin-1, originally discovered in Xenopus, is expressed by specific layers in the optic tectum. Though it can mediate homotypic adhesion, recent studies have revealed that Neuropilin-1 also binds with high affinity many class III Semaphorins, including Sema III, Sema E, and Sema IV. Clues as to the nature of additional Semaphorin receptors come from examination of viral Semaphorins. The genomes of several viruses, including the poxviruses vaccinia and variola (Kolodkin, 1993) and a herpesvirus (Ensser, 1995), harbor open reading frames encoding secreted Semaphorins that might modulate immune function. One such viral Semaphorin, vaccinia A39R, was used to affinity purify a novel cellular receptor from a human B cell line (Comeau, 1998). This virus-encoded Semaphorin protein receptor (VESPR) is a novel transmembrane protein that is a divergent member of the Plexin family. VESPR binds A39R with high affinity and herpesvirus AHV Sema with lower affinity. Human Sema K1 is a class VII Semaphorin that may be the cellular counterpart of AHV Sema; Sema K1 does not bind Neuropilin (Winberg, 1998 and references therein).

Like Neuropilins, Plexins were identified through a screen for antigens expressed in the optic tectum (Takagi, 1987). Xenopus Plexin-1 has been shown to contain an extracellular domain related to the c-met protooncogene and to mediate calcium-dependent homophilic binding (Ohta, 1995). Neuropilin-1 and Plexin-1 are expressed on different groups of olfactory axons that independently segregate, leading to the speculation that these two proteins help organize axon projections (Satoda, 1995). Plexin-1 was subsequently cloned in mice and is the prototype of a family of proteins (Kameyama, 1996). Plexins in human were cloned independently, based on their homology to the ectodomain of the oncogene MET (Maestrini, 1996). Called the SEX family of genes, several members (SEX, OCT, and NOV) are expressed predominantly in the brain, giving rise to the suggestion that they represent novel neuronal receptors for unknown ligands (Winberg, 1998 and references therein)

Given that a divergent member of the Plexin/SEX family is a receptor for the viral Semaphorin A39R and that most Plexins are expressed in the developing nervous system, Plexins have been hypothesized to serve as neuronal Semaphorin receptors. It was thought that they could function to control axon guidance. To investigate this possibility, the genes encoding two Plexins in Drosophila were cloned and characterized. Genetic and biochemical evidence support the idea that Plexin A (PlexA) in Drosophila is a receptor for the transmembrane class I Semaphorins Sema 1a and 1b. PlexA has been shown to control important aspects of axon guidance. Because there are two class I Semaphorins and two class II Semaphorins in Drosophila, the original Sema I has been renamed Sema 1a; the new class I Semaphorin is Sema 1b; the original Sema II is now Sema 2a, and the new class II Semaphorin is Sema 2b. Previous studies (Yu, 1998) have shown that mutations in Sema1a display specific defects in motor and CNS axon guidance. Mutations in PlexA display the same phenotypes and the two loci interact genetically in a fashion consistent with their functioning in the same signaling pathway. Along with selective binding results, these data show that Plexins do indeed function as neuronal Semaphorin receptors and that Plexins control important aspects of axon guidance (Winberg, 1998).

What about the neuropilins? It has been observed that the main semaphorin-binding domain of neuropilins (CUB domain) is not required for the interaction with plexins, as indicated by the association of the relevant neuropilin-2 deletion construct with plexin-B1. Thus the existence of a ternary complex is envisioned, where neuropilins use two distinct protein modules to form a bridge between the sema domain of semaphorins and the sema domain of plexins. Taken together, these findings raise the possibility that plexins are the long-sought functional partners of neuropilins required for transducing signals mediated by class 3 semaphorins. In flies, which lack both neuropilins and class 3 semaphorins, it is notable that Drosophila PlexA appears sufficient as a functional receptor for Sema-1a, a transmembrane class 1 semaphorin (Winberg, 1998). In further support of the hypothesis that plexins are functional coreceptors for secreted semaphorins, it has been shown that a truncated human plexin-A1 construct expressed in Xenopus spinal neurons abolishes repulsive responses to Sema3A without markedly affecting attractive responses to netrin-1. Similarly, expression of a dominant-negative plexin-A1 in sensory neurons blocks Sema3A-induced growth cone collapse, as reported independently by Takahashi (1999). Therefore, neuropilins serve along with plexins as co-receptors and receptors for Semaphorins (Tamagnone, 1999).

Sema 1a is expressed by neurons and is required for appropriate defasciculation. Loss-of-function analysis for this gene does not indicate whether Sema 1a functions as a ligand or as a receptor. However, misexpressing Sema 1a on muscles repels motor axons, demonstrating that Sema 1a is able to act as a target-derived repulsion cue (Yu, 1998). If PlexA is the receptor for Sema 1a, then reducing PlexA expression levels in the presence of ectopic Sema 1a ligand should suppress the severity of the gain-of-function repulsion phenotypes. Two different GAL4 enhancer trap lines were used to misexpress Sema 1a on muscle subsets. The first, H94-GAL4, is highly expressed by muscle fibers 6 and 13, and moderately by muscle 12 (genetic rescue data suggest that it is also expressed by some motor neurons at a very low level, although this has never been directly visualized. Using H94-GAL4 to drive UAS-Sema1a in these muscles disrupts their innervation by ISNb axons, with the strongest effect seen at muscle 13. A second line, F63-GAL4, was used to drive UAS-Sema1a specifically in muscles 6 and 7, thereby inhibiting innervation at the muscle 6/7 cleft (F63 is not expressed by motor neurons). For both GAL4 lines, the inhibitory effect of muscle-derived Sema 1a is suppressed in PlexA Df heterozygotes. This dominant suppression of the Sema1a gain-of-function suggests that neuronally expressed PlexA acts downstream of Sema 1a in mediating repulsion (Winberg, 1998).

Ectopic expression also provided a means to test for interactions between PlexA and Sema 1b, another Drosophila protein similar to Sema 1a (Yu, 1997). H94-GAL4 and F63-GAL4 were used to test whether Sema 1b can also act as a muscle-derived repellent. Sema 1b was found to be as capable as equally capable Sema 1a in repelling motor axons. Likewise, reducing the gene dose of PlexA suppresses to a similar extent the guidance defects caused by misexpression of Sema 1b. Based on the similarity of structure and sequence, as well as parallel gain-of-function phenotypes and suppression, it is proposed that Sema 1b may serve as an additional ligand for PlexA (Winberg, 1998).

Given the model that PlexA mediates repulsive guidance and thus drives defasciculation, it was asked whether removal of the major motor axon cell adhesion molecule, Fasciclin II, would genetically suppress PlexA mutant phenotypes. In agreement with the model, all of the ISNb motor axon PlexA phenotypes (but not the SNa mutant phenotypes) and the CNS PlexA phenotypes are partially suppressed when one copy of the FasII gene is removed (Winberg, 1998).

Defects arising from overexpression of PlexA were also examined. Driving high levels of UAS-PlexA in all neurons using elav-GAL4 leads to axon guidance phenotypes in all parts of the motor projection and also within the CNS. In most cases, PlexA overexpression results in phenotypes that are the opposite of those seen in the PlexA loss-of-function. For example, in 66% of segments, PlexA overexpression caused the dorsal SNa to split prematurely into multiple projections. Similarly, the dorsal extension of the ISN defasciculated inappropriately in 25% of segments. The transverse nerve, which made exuberant contacts onto ventral muscles in the deficiency embryos, stalled in 26% of segments with PlexA overexpression, resulting in the two halves of the nerve failing to meet. These defects are readily interpreted as resulting from increased sensitivity to repulsive cues (Winberg, 1998).

Genetic analysis provides strong evidence that PlexA is a necessary component of the Sema 1a signaling pathway and is likely to function as a Sema 1a receptor. Moreover, it suggests that PlexA may have additional ligands, including Sema 1b. These possibilities were tested in a heterologous expression system using alkaline phosphatase (AP) fusion proteins in binding assays on membranes from COS cells transiently transfected with a full-length PlexA construct. Both AP-Sema 1a and AP-Sema 1b bind to PlexA-expressing membranes at significantly higher levels than to mock-transfected membranes or to membranes containing another Drosophila repulsive axon guidance receptor, Robo 1. No specific binding occurs between PlexA and other Semaphorins, including Sema III-AP, AP-Sema E, and AP-Sema B, nor between PlexA and AP-Beaten path. It was also asked whether Sema-Plexin binding, like Plexin homophilic interaction in Xenopus, requires the presence of divalent cations. Binding is eliminated by the inclusion of chelating agent in the ligand supernatant, or by the omission of divalent cations from the wash buffer. Binding is preserved in the presence of either Mg2+ or Ca2+. In contrast, binding of Semaphorins to Neuropilins does not appear to require calcium or magnesium (Winberg, 1998).

Comeau (1998) noted that VESPR and other Plexins share a ~100 amino acid region of homology with Semaphorins near the C terminus of the Sema domain. Upon further sequence analysis, Plexins and their relatives, the MET-related tyrosine kinase receptors (including MET, RON, and SEA), all are found to contain a complete ~500 amino acid Sema domain near the N terminus of their ectodomains. Thus, Plexins may be considered as large transmembrane Semaphorins. The CLUSTAL algorithm was used to examine the relationships among the Sema domains of Plexins, Semaphorins, and MET-related receptors. Each of the three groups of proteins clusters separately. The structure of the tree suggests a phylogenetic model in which Plexins may have been the ancestral molecules from which Semaphorins and MET-related receptors evolved; from this perspective, Semaphorins may be considered as specialized Plexins (Winberg, 1998).

The data are consistent with a model in which Semaphorins are ligands and Plexins are receptors mediating repulsive axon guidance. Such a model may capture only part of the complexity of how these molecules function. The finding that Plexins can bind Semaphorins, combined with previous studies showing that Plexins can bind Plexins (Ohta, 1995), raises questions as to whether some Semaphorins might bind one another and whether class I Semaphorins might also function as receptors. In the immune system, CD100, a class IV Semaphorin, promotes B cell aggregation and also plays a role in T cell activation (Hall, 1996 and Herold, 1996). Some of its functions are consistent with CD100 functioning as a receptor as well as a ligand (Winberg, 1998).

PROTEIN STRUCTURE

Amino Acids - 1945

Structural Domains

Candidate partial cDNAs for Drosophila Plexins were found in a screen for transmembrane and secreted cDNAs (Kopczynski, 1996; Kopczynski, 1998). Additional cDNA clones were identified in a public database of expressed sequence tags. These clones represent transcripts from two genes that have been named PlexA and PlexB. Probes derived from these cDNAs were used to isolate additional clones that span the complete open reading frames (ORF). Conceptual translation predicts that each is a large type 1 membrane glycoprotein; the PlexA ORF is 1945 amino acids long with 16 potential N-linked glycosylation sites, and the Plex B ORF is 2051 amino acids with 15 sites. These two fly proteins are similar to previously described Plexins (Ohta, 1995; Kameyama, 1996; Maestrini, 1996) with homology extending the length of each protein. The extracellular domains include the motif of three cysteine-rich repeats (called MET-related sequences [MRS]) (Maestrini, 1996) in the middle of the ectodomain. Plexins and Met-like receptors contain short cysteine-rich motifs, termed 'Met-related sequences' (MRS), whose minimal consensus is: C-X(5-6)-C-X(2)-C-X(6-8)-C-X(2)-C-X(3-5)-C (Maestrini, 1996; Tamagnone, 1997). The proteins of the Met family contain a single MRS (in their receptor beta chains), whereas in plexin family members, there are three repeated MRS motifs. The membrane-proximal regions of the Plexin ectodomains contain another type of repeat that is rich in glycine and proline residues (G-P repeats). Three G-P repeats are spaced ~50 amino acids apart; the intervening sequences are not notably similar except that each contains two conserved cysteines. All of these features of the ectodomains are shared among Drosophila and vertebrate Plexins. They are also shared with MET-like tyrosine kinase receptors and VESPR, except that in these cases, there are different numbers of repeats (Winberg, 1998).

The intracellular region of vertebrate Plexins has been divided into two blocks of strong homology, separated by a variable linker (Maestrini, 1996). The cytoplasmic domain of plexins contains an ~600 amino acid domain that has been termed the SP domain ('Sex and Plexins') that is highly conserved within the family (57%-97% similarity) and in evolution (over 50% similarity between invertebrates and humans). The SP domain does not share homology with any known protein. It includes a number of potential tyrosine phosphorylation sites but lacks the typical motifs of catalytic tyrosine kinases. Interestingly, the predicted secondary structure of the SP domain includes long, conserved alpha helices typically found in protein-protein interaction modules. Furthermore, there are several dihydrophobic amino acid motifs (such as LL or LI) known to mediate the internalization and downregulation of transmembrane receptors (Tamagnone, 1999). Both Drosophila proteins conform to this pattern; indeed, the area of highest amino acid identity between the fruit fly and vertebrate Plexins is within these intracellular domains. A region of ~100 amino acids overlapping with the first of the cysteine (MRS) repeats has been shown (Comeau, 1998) to resemble sequences at the C terminus of the Sema domain found in all Semaphorins (Kolodkin, 1993). Upon further examination it is found that the ectodomains of all Plexins contain a complete ~500 amino acid Sema domain, with 14 of the 16 conserved cysteines and many short peptide sequences that are characteristic of the Semaphorin family. The MET-related receptors also contain complete (albeit divergent) Sema domains (Winberg, 1998).

The Sema domains of Semaphorins typically share 25%-30% amino acid identity between pairs of Semaphorins. The Sema domains of Plexins share 25%-30% amino acid identity between pairs of Plexins. Between Plexins and Semaphorins, the Sema domains share only 15%-20% amino acid identity. The Sema domains of MET receptors share only about 10% amino acid identity with Semaphorins and 15%-20% identity with Plexins. Examination of phylogenetic relationships among the Sema domains of Plexins, Semaphorins, and MET-related receptors shows that each of the three groups of proteins clusters separately (Winberg, 1998).

date revised: 10 January 2000

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.