BIOLOGICAL OVERVIEW

Proper synaptic development is critical for establishing all aspects of neural function including learning, memory, and locomotion. wishful thinking (wit) gene is the Drosophila homolog of the vertebrate BMP type II receptor. Mutations in wit result in pharate lethality that can be rescued by expression of a wit transgene in motor neurons but not in muscles. Mutant larvae exhibit small synapses, severe defects in evoked junctional potentials, a lower frequency of spontaneous vesicle release, and an alteration in the ultrastructure of synaptic active zones. wit NMJs have decreased levels of the synaptic cell adhesion molecule Fasciclin II, and synaptic membrane detachment at active zones. Wit is expressed by a subset of neurons, including motoneurons. The neuromuscular junction (NMJ) phenotype is specifically rescued by transgenic expression of Wit only in motoneurons. Thus, Wit appears to function as a presynaptic receptor that regulates synaptic size at the Drosophila NMJ (Marqués, 2002 and Aberle, 2002).

All three BMP ligands, Decapentaplegic, Gbb and Screw, appear to share a common set of receptors that include the type II receptor Punt and the type I receptors Thick veins and Saxophone. Activation of these receptors leads to phosphorylation and nuclear translocation of Mad, the Drosophila homolog of Smad1, in a complex with Medea, the Drosophila Smad4 homolog. In contrast, the activin type I receptor Baboon signals in conjunction with Punt through dSmad2 in response to an as yet uncharacterized ligand (Marqués, 2002 and Aberle, 2002 and references therein).

Based on sequence information, the only principal TGF-ß type signal transduction component remaining to be described in Drosophila is predicted to code for a type II receptor. This receptor, Wishful thinking, is most closely related to the vertebrate BMPRII and Müllerian Inhibitory Substance (MIS) type II receptors (Baarends, 1994; Kawabata, 1995; Liu, 1995), both of which are characterized by presence of a large carboxy-terminal extension distal to the kinase domain that is not found in the ActRII and TGF-ß type II receptors. BMPRII is the only vertebrate type II receptor that specifically binds BMP type ligands and not other members of the TGF-ß family. Knockout of this receptor in mice results in embryonic lethality, suggesting that it is the type II receptor that functions with ALK-3 to convey the BMP-4 signal during gastrulation (Beppu, 2000). Studies in Xenopus indicate that BMPRII is required for mesodermal patterning and suppression of neural fate in the ectoderm (Frisch, 1998). In humans, mutations in BMPRII cause primary pulmonary hypertension (PPH) (Deng, 2000; Lane, 2000; Thomson, 2000).

In mice, BMPRII is expressed in a dynamic pattern in many tissues of all three germ layers during embryonic development. In the nervous system, prominent expression is seen in the hippocampus, the cerebellum, motoneurons, and sensory neurons of the dorsal root ganglia (Charytoniuk, 2000). Due to a widespread expression and requirement during early developmental stages, BMPRII-deficient mice die shortly after implantation, thus far precluding the study of the function of BMPRII in vivo during synapse development (Beppu, 2000). A regulatory function for vertebrate BMPs in dendritic growth has been suggested by in vitro results on cultured sympathetic neurons. Exposure of these neurons to recombinant BMP7 rapidly induces the upregulation of dendritic marker proteins and the outgrowth of new dendrites (Lein, 1995). BMP2, BMP6, and Drosophila Gbb (60A) have been shown to stimulate similar growth effects (Guo, 1998). The dendritic growth-promoting activity of BMP7 is also effective on cultured hippocampal neurons. Because BMP7 and other BMPs are expressed in the hippocampus, it has been suggested that they might play a developmental role in dendritic growth and synapse formation (Withers, 2000).

TGF-ß family members may also have a regulatory function in the development of the vertebrate NMJ. A ligand of the TGF-ß superfamily, Myostatin (GDF-8), is expressed exclusively in developing and adult somatic muscles and negatively regulates muscle growth. Mice mutant for the myostatin gene are significantly larger than their littermates due to a 2- to 3-fold increase in their skeletal muscle size and mass (McPherron, 1997). In addition, TGF-ß2 is thought to act as a target-derived neurotrophic factor for motoneurons because it is localized at the postsynaptic side of the NMJ, and its receptors are expressed in motoneurons (Jiang, 2000). These results show that ligands and receptors of the TGF- superfamily are present in both muscles and motoneurons and suggest that they might function in regulating the growth of the neuromuscular synapses. However, a functional requirement for BMP signaling in the growth of any synapse in vivo has not yet been demonstrated (Marqués, 2002 and Aberle, 2002 and references therein).

In Drosophila embryos and third instar larvae, wit is heavily expressed in a subset of neuronal cells. Consistent with a role for Wit as a BMP type II receptor that modulates neuronal function, wit null mutant embryos are found show a specific loss of phosphorylated-Mad (P-Mad) staining within motor neurons but not in other tissues. Using various Gal4 drivers, it has been demonstrated that Wit expression in motor neurons is sufficient to recover viability. Electrophysiological analysis of the neuromuscular junction (NMJ) synapses in third instar larvae demonstrates that wit mutant animals have severely reduced evoked junctional potentials and display a high rate of failures at low extracellular [Ca2+]. The frequency of spontaneous vesicle release is significantly reduced as are synapse size and bouton number. In contrast, quantal size appears relatively normal. Ultrastructural analysis of mutant boutons reveals alterations in the morphology of the active zones, suggesting that reduced quantal transmission results from a presynaptic active zone defect. These results highlight a role for BMP signaling in regulating Drosophila synapse assembly and/or maintenance. Furthermore, they raise the possibility that, like TGF-ß, BMP signaling may regulate additional aspects of synaptic plasticity such as those associated with long-term facilitation and memory (Marqués, 2002 and Aberle, 2002).

Defects in synaptic structure in wit mutants suggest that BMP signaling plays a key role in regulating presynaptic morphology and function at the Drosophila neuromuscular junction. Synapse formation and maturation are relatively late steps in neural development and are preceded by neural induction, neurogenesis, neural migration, and axonal pathfinding. Previously, BMPs and their cognate receptors have been shown to be widely expressed in the developing and mature vertebrate nervous system (Lorentzon, 1996; Soderstrom, 1996), and numerous lines of evidence have implicated them in controlling several different aspects of neural development and function, including neurulation, morphogenesis, lineage decisions, and cellular maturation. While wit is expressed in a subset of neurons beginning at stage 12, wit is probably not required during early neurogenesis since no defect in patterning or cell fate changes could be detected during the development of the embryonic CNS, and wit mutants could be rescued with either the elav-Gal4 or nervana-Gal4 drivers, the expression of these drivers being restricted to mature neurons after most differentiation has taken place (Marqués, 2002).

The results reveal a role for BMP type II receptors in regulating the assembly or maintenance of presynaptic active zones and synaptic transmission. In wit mutants, transmitter release is severely reduced. While this exocytotic defect could be explained by a potential endocytotic abnormality, wit mutant boutons do not show depletion of synaptic vesicles. Further, quantal size remains largely unaffected by the mutation, suggesting that postsynaptic receptors are normal. These results are consistent with the observation that presynaptic expression of Wit can completely rescue lethality and also significantly rescues both the frequency of spontaneous release and the EJC amplitude. Finally, ultrastructural studies indicate that defects in active zones may underlie the severe reduction in transmitter release (Marqués, 2002).

wit mutants exhibit a specific loss of P-Mad accumulation in a subset of embryonic neurons, including most motoneurons, implying that Wit mediates a BMP type signal. This suggests that Sax, Tkv, or both are likely to be the type I partners for Wit function. Consistent with this view, tkv, sax, and mad mutants show NMJ phenotypes similar to wit mutants and combinations of activated Sax and Tkv receptors, as well as chimeric combinations of Tkv and Wit, can rescue many aspects of the wit phenotype (McCabe, unpublished data; J. Rawson and S. Selleck, personal communication, cited in Marqués, 2002). At this point, the possibility that Wit might also participate with Babo in mediating a dSmad2 type signal or may participate in a type I receptor independent mechanism cannot be excluded (Marqués, 2002).

One other key issue that remains to be resolved is the identity and source of the ligand or ligands that bind to Wit. Potential ligands could be synthesized in the neurons themselves and provide an autocrine or paracrine type signal for presynaptic differentiation, or they could be secreted from the muscles and provide a retrograde signal. In Drosophila, motoneurons can differentiate and form active synapses in the complete absence of target muscles. This suggests that intrinsic cues guide most of the basic processes that regulate NMJ synaptogenesis. Drosophila dActivin ß_B (Myoglianin ) is expressed exclusively in the embryonic and larval CNS (Lo, 1999 and T.E.H. and M.B.O., unpublished data cited in Marqués, 2002) and therefore might provide an intrinsic cue that helps regulate synapse formation, maintenance, and function. However there is no evidence as yet that Wit participates in an activin-like pathway (Marqués, 2002).

Although the initial formation of synapses in embryos might be driven by intrinsic cues, during larval stages the muscle volume increases over 150-fold and the synapse must grow in a regulated manner to maintain proper synaptic strength. Since in wit mutants, synapse size is close to normal at the end of embryogenesis but is abnormal soon after, it is inferred that wit is required for synaptic growth. Whether it is required continuously during all larval stages or only at a defined time during late embryogenesis and the early 1^st instar to set in motion a particular developmental program is difficult to determine. No P-Mad accumulation is found after the first instar stage. However it is possible that a lower level of BMP signaling is required for synapse growth throughout the larval stages (Marqués, 2002).

Since Wit function is required in the presynaptic cell, it is speculated that Wit could transduce a retrograde signal from the muscle to the nerve cell that provides a means of coordinating synapse growth with muscle growth. Evidence supporting a retrograde signal that modulates presynaptic transmitter release at the NMJ in Drosophila has come from experiments in which glutamate receptor levels were manipulated in postsynaptic cells. In vertebrates, recent evidence indicates that Wnt7a is a target-derived synaptogenic signal at the mossy fiber-granule cell synapse. While no direct evidence linking BMP signaling to synaptogenesis has been described previously, it is interesting to note that TGF-ß2 has been found to be associated with the subsynaptic muscle nuclei of mature rat neuromuscular junctions, and that TGF-ß receptors are transported in an anteriograde fashion and inserted in the nerve terminal (Jiang, 2000; McLennan, 1994). Tkv is specifically localized to the NMJ synapse, suggesting that a similar mechanism could operate in Drosophila (McCabe, unpublished data cited in Marqués, 2002). Candidate ligands for Wit would then be TGF-ßs expressed in the muscle (Marqués, 2002).

The targets of wit signaling are also unknown. The primary defect could be in electrical activity. It is well known that electrical activity influences synapse plasticity and produces changes in both bouton number and morphology. It is interesting to note in this regard that target-derived TGF-ß1 has been observed to stimulate the functional expression of Ca⁺²-activated K⁺ channels in developing chick ciliary ganglion. In this case, however, the effect appears to be mediated posttranslationally and therefore does not involve transcriptional responses by Mad type factors (Cameron, 1998, 1999, Marqués, 2002).

Although alterations in electrical activity per se could be responsible for some of the morphological and functional defects observed, it is more likely that wit mutants alter the expression of other targets. One gene whose expression level is known to influence synapse morphology is the adhesion molecule Fas II. Fas II is expressed both pre- and post-synaptically and is required in both the muscle and neuron for proper synapse development. Hypomorphic mutants of Fas II exhibit either an increase or a decrease in bouton number at the third larval instar stage depending on the level of the residual Fas II protein, and Fas II is downregulated in wit mutants (Aberle, 2002). However even if there is a direct effect of wit signaling on Fas II expression, Fas II is not likely to be the only Wit target since synaptic efficacy is not altered in fas II mutants, whereas in wit mutants there is a severe reduction in synaptic strength. Thus, if wit mutants do affect Fas II expression, then they must also alter the homeostatic compensation mechanism that normally acts to maintain constant synaptic strength despite differences in pre- and post-synaptic levels of Fas II protein in various combinations of fas II alleles (Marqués, 2002).

Other possible targets for wit signaling include components of the exocytotic machinery that controls vesicle targeting, docking, fusion, or release. Certain allelic combinations of mutations in the rop gene, for example, cause severe reductions in both EJPs and the frequency of spontaneous fusion events similar to what has been found for wit mutants. Rop codes for a member of the Sec1 family of proteins thought to regulate secretion by modulating syntaxin, SNAP-25, and synaptobrevin complex function. Unlike wit mutants however, mutations in rop do not affect overall synapse morphology (Marqués, 2002).

Perhaps the most interesting candidate targets for Wit signaling are those genes whose products reside in or contribute to active zone assembly. Very little is known about the composition or mechanism of assembly of this specialized structure. On the postsynaptic side, glutamate receptors are highly concentrated at active zone sites. However, it is unlikely that receptor clustering is the primary defect because the amplitude distribution of spontaneous vesicle releases is normal (Marqués, 2002).

The finding that a BMP pathway modulates synaptic structure and function at the Drosophila NMJ is particularly intriguing in light of other recent reports implicating TGF-ß type components in modulating neuronal plasticity (Chin, 1999; Zhang, 1997). Equally appealing in terms of a potential BMP connection is the observation that long-term sensitization training in Aplysia induces expression of a Tld/BMP-1-like product (Liu, 1997). Tld/BMP-1 type proteins encode metalloproteases that cleave the BMP inhibitors Sog and Chordin, respectively. Thus, the complex regulatory circuit that modulates the activity of BMP type ligands during early embryonic development might also govern the activities of these ligands during neuronal development and may play specific roles in regulating synaptic plasticity associated with long-term learning and memory (Marqués, 2002).

GENE STRUCTURE

cDNA clone length - 4034

Bases in 5' UTR - 208

Exons - 6

Bases in 3' UTR - 1113

PROTEIN STRUCTURE

Amino Acids - 903

Structural Domains

Conceptual translation of the wishful thinking ORF reveals a protein with a signal peptide, a single membrane-spanning domain, and a Ser-Thr kinase domain, containing an ATP binding site and a highly conserved coiled-coil/leucine zipper motif, followed by a carboxy-terminal tail of 378 aa with no distinctive features other than an abundance of Ser and Thr residues. The overall domain structure, including the presence of the carboxy-terminal tail, is a characteristic feature of the BMPRII and MIS type II receptors (Baarends, 1994; Kawabata, 1995; Liu, 1995). Phylogenetic analysis of the kinase domain clearly places Wit as a close homolog of the vertebrate BMPR-II receptor. However, despite the similarity within the kinase domain, it is noted that the C-terminal tail of Wit shows very little identity to the tails of either BMPRII or the MISR other than a short stretch of 7 aa shared with BMPRII within the C-terminal region of the tail. At the amino acid level, the overall identity of Wit to human BMPRII is 30% with the strongest homology in the kinase domain (Marqués, 2002 and Aberle, 2002).

In vertebrates, two alternatively spliced isoforms of the BMP type II receptor have been reported (Kawabata, 1995). The shorter version ends 28 aa after the kinase domain and does not contain the carboxy-terminal extension. By developmental Northern analysis a single 4.5 kb transcript was detected at all stages, suggesting that only one isoform of wit is expressed in Drosophila (Marqués, 2002).

date revised: 10 March 2002

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.