BIOLOGICAL OVERVIEW

highwire (hiw) encodes a large protein (5233 amino acids) that controls synaptic growth at the larval neuromuscular junction (NMJ). The increase in synaptic structure in hiw mutants is dramatic: NMJ growth is exuberant. Hiw protein is localized at periactive zones which surround active zones at presynaptic terminals, the same region where Fasciclin II is localized. Hiw appears to function as a negative regulator of synaptic growth (Wan, 2000).

The most accessible synapse in Drosophila for a genetic approach to uncover mutants with changes in synaptic rate and extent of growth is the NMJ. In contrast to the fairly static NMJ in mammals, the Drosophila NMJ is a dynamic structure, growing new boutons and sprouting new branches during larval development. The Drosophila NMJ shares several important features with central excitatory synapses in the vertebrate brain. The Drosophila NMJ is a glutamatergic synapse and has ionotropic glutamate receptors homologous to those of humans. This synapse does not have a conspicuous basal lamina separating the two sides, as does the vertebrate NMJ, but rather appears by ultrastructural analysis to consist of close membrane apposition of pre- and post-synaptic cells. This synapse is organized into a series of boutons and branches that can be added or eliminated during development and plastic changes (Wan, 2000 and references therein).

As the Drosophila larva develops from first to third instar over a period of several days, the surface area of the postsynaptic muscle increases as much as 100-fold. During this developmental period, there is a concomitant growth of the presynaptic nerve terminal, resulting in up to a 10-fold increase in the number of boutons and a 10-fold increase in the number of active zones per bouton. A number of activity-dependent and activity-independent mechanisms appear to control the dramatic larval growth of this synapse. Long-term activity-dependent synaptic plasticity was first demonstrated at this synapse by genetic manipulations that alter neuronal activity and cAMP concentrations. K+ channel mutants, such as ether a go-go (eag) and Shaker (Sh), increase neuronal activity, and synaptic structure and branching. In addition, dunce (dnc) mutants, deficient in cAMP phosphodiesterase, increase synaptic structure. This activity-dependent synaptic growth is controlled in part by the regulation of the levels of the synaptic cell adhesion molecule (CAM) Fasciclin II (Fas II). Genetic analysis has shown that this downregulation of Fas II is both necessary and sufficient for structural plasticity at this synapse: FasII mutants that decrease Fas II levels by ~50% lead to sprouting, similar to eag and dnc, while transgenes that maintain synaptic Fas II levels suppress sprouting in eag, Sh and dnc (Wan, 2000).

A role for CAMs in the control of synaptic growth is not unique to Drosophila. In fact, CAMs of both the immunoglobulin and cadherin families are thought to play important roles in the regulation of synaptic growth in organisms as diverse as snails and mammals. Although CAMs have been implicated in the control of synaptic growth, little is known about the molecular machinery that carries out this growth process. This lack of knowledge led to a genetic approach to identify genes and genetic pathways controlling synaptic growth. Advantage was taken of a behavioral screen of the X chromosome conducted to detect walking mutants. A secondary anatomical screen was undertaken of the larval NMJ of this preselected collection of walking mutants. This screen led to the identification of the highwire gene. In hiw mutants, the specificity of motor axon pathfinding and synapse formation appears normal. However, NMJ synapses grow exuberantly and are greatly expanded in both the number of boutons and the extent and length of branches. These synapses appear normal ultrastructurally but have reduced quantal content physiologically. hiw encodes a large protein found at presynaptic terminals. Within presynaptic terminals, Hiw is localized to the periactive zone surrounding active zones; Fasciclin II (Fas II), which also controls synaptic growth, is found at the same location (Wan, 2000).

The hiw transcript encodes a protein that contains a RING-H2 finger, a domain identified in a large family of E3 ubiquitin ligases. Hiw, a potential synaptic E3 ligase, has been identified as a lethal enhancer of neuronal overexpression of fat facets, coding for a deubiquitinating protease. Overexpression of Fat facets leads to a profound disruption of synaptic growth control; there is a large increase in the number of synaptic boutons, an elaboration of the synaptic branching pattern, and a disruption of synaptic function. Genetic interactions between fat facets and highwire, a negative regulator of synaptic growth that has structural homology to a family of ubiquitin ligases, suggest that synaptic development may be controlled by the balance between positive and negative regulators of ubiquitination (Diantonio, 2001).

Phenotypic analysis of hiw has focused on the glutamatergic NMJ, but since all neurons appear to express both HIW mRNA and Hiw protein, it is reasonable to suggest that hiw functions at most, if not all, synapses. The original ethyl methanesulfonate (EMS) allele was isolated in a behavioral screen that selected for viable mutations. Subsequently 12 additional EMS alleles were isolated. All 13 EMS alleles and 3 P element alleles are viable, either alone or in trans over one another, or over either of two deficiencies that remove the gene. Several of these alleles are likely to be true nulls based on their phenotype over deficiencies and the presence of premature stop codons. Thus, the hiw mutant phenotype -- viable flies with dramatic synaptic overgrowth -- is likely to represent the true null phenotype (Wan, 2000).

In hiw mutants, motor axon pathfinding and target recognition appear normal. Synapses initially form at their normal locations on the appropriate muscles and induce appropriate postsynaptic glutamate receptor clustering. These synapses normally grow and sprout during larval development. But in hiw mutants, they are greatly expanded in both the number of boutons, and the extent and length of branches. The expanded synapses in hiw mutants have smaller-than-normal boutons but appear normal ultrastructurally in terms of active zones, T bars, clusters of synaptic vesicles, and folds of the subsynaptic reticulum. Occasionally, the overgrowth is so extensive that motor axons wander over to adjacent muscles which they would not innervate normally. This aspect of the hiw phenotype is similar to what is seen when the CAM Fas II is overexpressed on target muscles (Wan, 2000).

Physiologically, hiw synapses are functional, as is evident from the viable nature of the null mutants and the fact that they are capable of many motor behaviors. However, although greatly expanded, these synapses are not stronger, as one might have expected, but rather have a reduced quantal content. One possibility is that the reduction in quantal content is itself the primary defect and that this physiological defect leads to the increased growth in an attempt to compensate and return the strength back to normal. But this is unlikely, since mutant conditions that lead to an even greater reduction in quantal content (e.g., mutations in synaptic vesicle protein synaptotagmin) do not lead to presynaptic sprouting at this synapse. Thus, the overgrowth of the synapse appears to be a primary defect. The data suggest that the physiological defect is either secondary to the overgrowth or an independent primary defect (Wan, 2000).

Within presynaptic terminals, Hiw is localized to areas surrounding active zones: the periactive zones. Hiw is conspicuously absent from active zones. At the immuno-EM level, Hiw is associated with the plasma membrane and with vesicles near the membrane but outside the active zone. Fas II, a CAM that also controls synaptic growth, is found at the same location as Hiw at the periactive zones (Wan, 2000).

These results suggest that Hiw functions largely on the presynaptic side of the synapse. Certainly, at the NMJ, high levels of HIW mRNA are observed in motor neurons and are undetectable levels in muscles. Transgenic rescue experiments of this gene could not be performed due to the enormous size of the ORF and the complexity of the genomic region containing the hiw transcript. Nevertheless, a partial hiw cDNA transgene was used to generate a dominant-negative phenotype in which the NMJ overgrows, just as in hiw loss-of-function mutants. This dominant-negative phenotype is only observed when the Hiw fragment is expressed on the presynaptic rather than postsynaptic side of this synapse, further suggesting that Hiw protein normally functions on the presynaptic side (Wan, 2000).

rpm-1, the homologous gene in the nematode C. elegans, also seems to function cell autonomously in presynaptic neurons. The expression of RPM-1 on the presynaptic side of the synapse completely rescues function, and rpm-1 is required only in presynaptic neurons in genetic mosaic analysis (Schaefer, 2000; Zhen, 2000). Thus, taken together with the nematode data, it has been concluded that Hiw regulates synaptic growth by functioning on the presynaptic side of the synapse (Wan, 2000).

Phenotypic analysis was focused on the NMJ, where the branching pattern of individual presynaptic terminal arbors could be observed, the number of boutons could be counted, confocal microscopy could be used to image subcellular localization of components, and conventional and immuno-EM could be used to analyze identified synapses. These techniques are much more difficult to apply to identified terminal arbors and synapses in the CNS. Single cell probes that would allow this kind of quantitative analysis in the fruit fly CNS are not available (Wan, 2000).

Nevertheless, it is speculated that Hiw might function at many synapses, and not just the NMJ. (1) HIW mRNA is expressed by all CNS and PNS neurons. Interestingly, toward the end of embryogenesis, HIW mRNA is detected in the synaptic neuropil in the CNS, suggesting that HIW mRNA may be actively transported to near CNS synapses. HIW mRNA is not observed in motor or sensory axons in the periphery. HIW mRNA is not observed in the nerve roots in the CNS. Thus, it is concluded that HIW mRNA is localized in the areas where the synaptic neuropil forms. Whether HIW mRNA is expressed specifically at synapses awaits future investigation (Wan, 2000).

(2) Hiw protein, as detected with anti-Hie antibodies, is localized in the periphery at synaptic terminals of the NMJ, and in the CNS throughout the synaptic neuropil. Since no axonal staining is observed with the antibody, it is thought the staining is consistent with Hiw being localized at synapses (Wan, 2000).

(3) Although the phenotypic characterization of hiw mutants in Drosophila focuses on the glutamatergic NMJ, mutations in the homologous gene in the nematode give rise to altered synaptic structure at GABAergic synapses (Zhen, 2000) and glutamatergic and cholinergic synapses (Schaefer, 2000). Thus, Hiw protein functions at a variety of synapses in the nematode and at the glutamatergic NMJ in Drosophila, and it is expressed throughout the synaptic neuropil in Drosophila (Wan, 2000).

Why do hiw mutants have expanded synapses in Drosophila? One possibility is that synapses never properly form in hiw mutants and that the Hiw protein controls synapse formation. In this respect, the exuberant terminal motor axon branching across muscles observed in hiw mutants resembles the expanded nerve branching across muscles seen in agrin or MuSK knockout mice. But in these mutant mice, synapses do not properly form, and postsynaptic receptors do not properly cluster. These NMJs do not function, and thus these mouse mutants are lethal. In contrast, in hiw mutants, synapses form, active zones appear normal ultrastructurally, and postsynaptic receptors cluster. hiw mutant flies are viable. Thus, Hiw is not necessary for synapse formation per se (Wan, 2000).

A second and more likely possibility is that Hiw functions in a mechanism that regulates either synaptic stabilization, synaptic growth, or both. The major phenotype is synaptic overgrowth. It is difficult to distinguish between a model in which the primary function of Hiw is for synaptic stabilization versus one in which the primary function is for synaptic growth. Moreover, these two processes are likely to be intimately linked (Wan, 2000).

At the Drosophila NMJ synapse, the CAM Fas II controls synaptic stabilization. In its complete absence, the synapse retracts. But Fas II levels control synaptic growth, and these levels of Fas II are regulated by neural activity. A 50% decrease in Fas II leads to an increase in presynaptic sprouting and growth. These data suggest that while adhesion is required for synaptic stabilization, the modulation of the levels of adhesion regulate synaptic growth. Too much adhesion stabilizes the synapse and inhibits growth; a reduction of adhesion destabilizes the synapse and permits growth. It is at this interface between synaptic stabilization and the regulation of synaptic growth that Hiw appears to function (Wan, 2000 and references therein).

Mutations in other genes have been identified that lead to an increase in synaptic sprouting and growth at the Drosophila NMJ, including K⁺ channel mutants, such as eag and Shaker, and mutants in the cAMP phosphodiesterase dunce. Although these genetic conditions lead to synaptic sprouting, none of them are as dramatic as is hiw. No other known mutation has such a tremendous overgrowth phenotype at this synapse as does hiw. Taken together, all of these considerations have lead to the conclusion that Hiw normally functions to control synaptic stabilization and growth (Wan, 2000).

In certain respects, Hiw protein is reminiscent of two other non-Drosophila large presynaptic proteins, Bassoon and Piccolo. Piccolo is identical to a previously described protein, Aczonin. Both Bassoon and Piccolo are expressed presynaptically, both are cytoplasmic, both are enormous (>420 kDa), both contain zinc fingers (of the kind associated with protein-protein interactions), and both contain coiled-coil domains. But these two proteins also have striking differences from Hiw. Whereas Bassoon and Piccolo are expressed at presynaptic active zones around the clustered synaptic vesicles and appear to function as part of a presynaptic cytomatrix involved in the movement and clustering of synaptic vesicles, Hiw is expressed at periactive zones. In addition, there do not appear to be orthologs of Bassoon and Piccolo in the fly genome. Thus, Hiw appears to define a new function and, perhaps, a new kind of presynaptic cytomatrix that is localized to the periactive zone and functions in the control of synaptic stabilization and growth (Wan, 2000).

On the presynaptic side of a synapse, most studies have focused on the active zone and its components. The active zone contains the vesicle docking and release machinery, the clustered synaptic vesicles, the machinery to move and restrain vesicles, and the modulators and regulators of transmitter release. Synaptic function is associated with the active zone. But the active zone does not contain all functions associated with the synapse (Wan, 2000).

Evidence is presented supporting the idea that the zone outside of the active zone -- the periactive zone -- is an important area that contains the regulatory machinery for synaptic stabilization and growth. The CAM Fas II and the guanine nucleotide exchange factor (GEF) SIF are both localized in the periactive zone. Hiw protein is also localized to the periactive zone and the localization of Fas II has been confirmed. Double labeling experiments with Hiw and Fas II could not be performed because both antibodies were generated in mice. Nevertheless, it is believe that these two proteins are colocalized to the periactive zone based on their similar localization patterns, each complementary to that of Pak. Both Fas II and Hiw regulate synaptic growth. It is proposed that this region that surrounds active zones contains the machinery for a distinct set of functions, including the long-term regulation of synaptic growth and sprouting (Wan, 2000).

It is interesting to note that the structural domains predicted by the primary sequence of Hiw have been shown to mediate protein-protein interactions as well as to regulate signaling pathways. In RCC1, the seven tandem repeats form a seven-bladed propeller structure, as determined by X-ray crystallography, and are important for RCC1's function as a GEF for the nuclear small GTP binding protein Ran. A similar RCC1-like domain in a human protein p532 (HERC1) has also been demonstrated to stimulate guanine nucleotide exchange on other GTPases, such as ARF1, Rab3A, and Rab5 (see Drosophila Rab5). A second RCC1-like domain in p532 binds to clathrin and forms a cytosolic ternary complex with clathrin and Hsp70. The RCC1-like domain in Hiw might act as a GEF for an unidentified GTPase in a signaling pathway; alternatively, it might serve as a protein-protein interaction domain (Wan, 2000).

The region of the last 430 amino acids at the C terminus of Hiw is cysteine rich and is predicted to contain multiple zinc fingers, two of which can be categorized as B-box and RING finger motifs. Both the RING and the B-box motifs are zinc binding domains shown to participate in the formation of protein complexes. Recent studies also suggest a new function for the RING finger motif as an E2-dependent ubiquitin ligase. These structural motifs in Hiw may mediate protein-protein interactions and/or regulate levels of proteins via the ubiquitin pathway at periactive zones to control synaptic growth (Wan, 2000).

A region in Hiw has been found that behaves in a dominant-negative manner when expressed panneurally. It is interesting that this region does not contain either the RCC1-like domain or the zinc finger cluster. Instead, it contains the coiled-coil domain and the putative Myc binding domain. It is possible that this region in Hiw is involved in interactions between Hiw and other proteins and that the overexpression of this region interferes with such interaction and thus leads to a dominant-negative phenotype (Wan, 2000).

The hiw gene is highly conserved across phyla. There appears to be one gene in the fruit fly Drosophila (hiw), one complete gene in the nematode C. elegans (rpm-1; there is a gene duplication in C. elegans, producing a second minigene next to rpm-1 that expresses only the first 2000 amino acids of RPM-1), and one known gene in human (PAM). The function of Hiw at the synapse is also conserved between the fruit fly and the worm (Wan, 2000).

hiw mutations in the fly lead to synaptic sprouting and overgrowth, increasing the number of boutons and the complexity of presynaptic branches. No defect is seen in axon pathfinding, target selection, or the initial stages of synapse formation. rpm-1 mutations in the nematode (Zhen, 2000) lead to variable phenotypes in different types of NMJ synapses and, most noticeably, a larger presynaptic terminal with multiple active zones. EM analysis shows that the majority of the mutant GABAergic NMJs contain more than two individual presynaptic active zones within the same presynaptic terminal, while a smaller portion have fewer synaptic vesicles and instead are filled with electron-dense debris-like material. Mutant cholinergic NMJs appear to have a single larger active zone with increased number and density of synaptic vesicles. The overall axon morphology of the DD, VD, and other ventral cord motor neurons appears normal, and cholinergic synaptic transmission also appears unaffected. However, certain pre- and post-synaptic components (synaptotagmin SNT-1 and GABA receptor UNC-49) appear to be mislocalized in the mutant. In contrast, rpm-1 mutant mechanosensory neurons fail to accumulate synaptic vesicles, retract certain synaptic branches, and extend ectopic processes in the neuropil (Schaefer, 2000; Wan, 2000).

Thus, some aspects of the hiw mutant phenotype are consistent with the rpm-1 mutant phenotypes, while other aspects of the phenotypes are different. Synapses in both the nematode and the fruit fly contain presynaptic active zones with similar components. But the anatomy of synaptic contacts in the two organisms is quite distinct. In the nematode, synapses tend to be en passant contacts of one axon with another, with synaptic contacts being made along the axon shaft. In contrast, in the fruit fly, axons arborize into presynaptic terminals, and synaptic contacts are organized into boutons along the branches of the terminal arbor. At the NMJ, these terminal arbors and synaptic boutons undergo a dramatic growth process as boutons expand, new boutons are added, and new branches sprout and grow. In many respects, this synaptic organization in the fruit fly is more reminiscent of synapses in vertebrates. A dramatic overgrowth of the terminal arbor and increase in the number of boutons is observed in the Drosophila hiw mutant, while the rpm-1 mutant reveals defects in the spacing and organization of active zones (and more occasionally, retraction or expansion of synaptic branches). These differences may reflect the increased complexity of synaptic structure in the fruit fly compared with the nematode (Wan, 2000).

Finally, how can the difference between what is reported for Hiw in Drosophila be reconciled with the initial observation of the expression and potential function of Pam, the Hiw homolog in human? Pam was discovered as a putative Myc binding protein that is expressed in the nucleus of tissue culture cells (Guo, 1998). No evidence has been found for nuclear localization of Hiw in Drosophila. Moreover, Hiw lacks part of the region that has been identified as the Myc binding domain. No evidence was found for genetic interactions in Drosophila between hiw and dmyc; moreover, dmyc mutants do not show a synaptic phenotype in Drosophila. At present, nothing is known about Pam function in mammals. It is also possible that additional Hiw homologs may exist in mammals. Thus, whether Hiw homologs in vertebrates function only as presynaptic regulators of synaptic stabilization and growth or also have taken on additional functions remains to be determined (Wan, 2000).

PROTEIN STRUCTURE

Amino Acids - 5233

Structural Domains

The full-length cDNA is 16,179 bp in length and encodes an enormous protein of 5233 amino acids. The predicted molecular weight is 566 kDa. There are several conspicuous protein motifs present in Hiw. These motifs include at the N terminus, a region of seven tandem repeats similar to those first described in a protein known as RCC1, for regulator of chromosome condensation. The C terminus contains a cysteine-rich domain that may have multiple zinc finger motifs, two of which can be clearly defined as a RING-type and a B-box-type zinc finger. Toward the middle of the protein are two direct repeats of 90 amino acids (PHR repeats) and a coiled-coil domain. Neither signal peptides nor transmembrane domains were observed. Thus, it is inferred that Hiw is a putative cytoplasmic protein. Hiw is homologous to a human protein called Pam, for protein associated with Myc (Guo, 1998), as well as to a C. elegans protein shown to be encoded by the rpm-1 gene (Schaefer, 2000; Zhen, 2000; Wan, 2000)

Pam was identified in an expression screen for proteins that bind to Myc, the product of c-MYC protooncogene. Pam is located in the nucleus and associated with Myc in tissue culture cells. rpm-1, the C. elegans homolog of hiw, was identified in two different genetic screens for synaptic mutants. In one screen (Zhen, 2000), mutants were recovered that showed abnormal distribution of a presynaptic marker at synapses from GABAergic motor neurons. EM analysis reveals that rpm-1 mutant GABAergic and cholinergic synapses have abnormal ultrastructure, such as a single presynaptic terminal containing multiple active zones. In the other screen (Schaefer, 2000), rpm-1 mutants were recovered that showed abnormal distribution of a presynaptic marker at synapses from posterior lateral mechanosensory neurons. These are most likely glutamatergic synapses. This study subsequently examined cholinergic motor neurons and found similar synaptic defects. In addition, the axonal branching pattern of the mechanosensory neurons was found to be abnormal in the mutants (Wan, 2000).

The homology among these three proteins (Drosophila HIW, C. elegans RPM-1, and human Pam) is distributed throughout the entire ORF. All three proteins contain the RCC1-like domain, PHR repeats (named for Pam/HIW/RPM-1), and the zinc finger cluster. Sequences outside the obvious structural domains also share similarities. From an evolutionary perspective, Hiw is more closely related to Pam than is RPM-1. The most conserved regions are the C-terminal zinc finger cluster domain of 430 amino acids as well as the internal PHR repeats. Hiw is the largest of the three proteins: 593 amino acids larger than Pam and 1468 amino acids larger than RPM-1. These extra sequences are distributed throughout the protein. Sequences unique to Hiw have similarity to coiled-coil domains. One feature common to only Hiw and RPM-1 is that they both partially lack a 300 amino acid region in Pam that has been shown to be required for binding to Myc (Wan, 2000).

date revised: 30 August 2001

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.