Partitioning of N-Ethylmaleimide-Sensitive Fusion (NSF) Protein Function in Drosophila melanogaster: dNSF1 Is Required in the Nervous System, and dNSF2 Is Required in Mesoderm

Corresponding author: Leo Pallanck, Department of Genetics, University of Washington, Health Sciences J113, Box 357360, Seattle, WA 98195., pallanck{at}genetics.washington.edu (E-mail)

Communicating editor: R. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The N-ethylmaleimide-sensitive fusion protein (NSF) promotes the fusion of secretory vesicles with target membranes in both regulated and constitutive secretion. While it is thought that a single NSF may perform this function in many eukaryotes, previous work has shown that the Drosophila genome contains two distinct NSF genes, dNSF1 and dNSF2, raising the possibility that each plays a specific secretory role. To explore this possibility, we generated mutations in the dNSF2 gene and used these and novel dNSF1 loss-of-function mutations to analyze the temporal and spatial requirements and the degree of functional redundancy between dNSF1 and dNSF2. Results of this analysis indicate that dNSF1 function is required in the nervous system beginning at the adult stage of development and that dNSF2 function is required in mesoderm beginning at the first instar larval stage of development. Additional evidence suggests that dNSF1 and dNSF2 may play redundant roles during embryonic development and in the larval nervous system. Ectopic expression studies demonstrate that the dNSF1 and dNSF2 gene products can functionally substitute for one another. These results indicate that the Drosophila NSF proteins exhibit similar functional properties, but have evolved distinct tissue-specific roles.

VESICLE trafficking in the constitutive and regulated secretory pathways requires a set of core polypeptides that mediate interactions between the transport vesicle and the target membrane. Among these core polypeptides are vesicular membrane proteins known as v-SNAREs, target membrane proteins known as t-SNAREs, and a soluble ATPase known as the N-ethylmaleimide-sensitive fusion protein (NSF). Although the precise functional roles of these components in vesicle exocytosis remain unclear, a large body of work has demonstrated that v-SNAREs and t-SNAREs can assemble to form a complex and that NSF can catalyze disassembly of this complex by coupling SNARE complex disassembly to ATP hydrolysis (JAHN and SUDHOF 1999 ; KLENCHIN and MARTIN 2000 ; LIN and SCHELLER 2000 ; WICKNER and HAAS 2000 ). These observations, together with other work on NSF and SNARE function, suggest several models by which these components may contribute to vesicle exocytosis. One possibility is that NSF is recruited to a SNARE complex anchoring a secretory vesicle to its target membrane and promotes membrane fusion by catalyzing complete or partial disassembly of the SNARE complex. Another model consistent with available data on NSF and SNARE function is that NSF is required after vesicle fusion to disassemble a SNARE complex formed prior to or during membrane fusion, thereby reactivating SNAREs for another round of exocytosis. Yet another possibility is that NSF plays both prefusion and postfusion roles in SNARE complex metabolism.

More recent work suggests the possibility of novel functional roles for NSF. MULLER et al. 1999 have shown that a mutationally altered version of NSF that apparently lacks both ATPase and SNARE complex disassembly activity is capable of promoting membrane fusion in an in vitro Golgi reassembly assay, suggesting, among other possibilities, a role for NSF in SNARE activation prior to membrane fusion. Another recent series of studies has shown that NSF may regulate the function or subcellular distribution of the AMPA class of postsynaptic ionotropic glutamate neurotransmitter receptors in hippocampal neurons (NISHIMUNE et al. 1998 ; OSTEN et al. 1998 ; SONG et al. 1998 ; LUSCHER et al. 1999 ; LUTHI et al. 1999 ; NOEL et al. 1999 ). This function of NSF appears distinct from its more familiar role in vesicle trafficking, as it requires direct binding of NSF to the glutamate receptor (NISHIMUNE et al. 1998 ; OSTEN et al. 1998 ; SONG et al. 1998 ). Several models have been proposed to explain these results: NSF may regulate the insertion of glutamate receptors into the plasma membrane, anchor the receptors to prevent lateral diffusion away from release sites, or act as a chaperone to regulate receptor conformation or interactions with other proteins (NISHIMUNE et al. 1998 ; OSTEN et al. 1998 ; SONG et al. 1998 ; LUSCHER et al. 1999 ; LUTHI et al. 1999 ; NOEL et al. 1999 ).

To elucidate the cellular functions of NSF, we have chosen to use a genetic approach in the fruit fly, Drosophila melanogaster. Previous work has shown that there are two Drosophila NSF genes encoding polypeptides with 84% amino acid identity to one another, designated dNSF1 and dNSF2 (BOULIANNE and TRIMBLE 1995 ; PALLANCK et al. 1995B ). dNSF1 is expressed at high levels in adult flies and at relatively lower levels earlier in development, while dNSF2 is expressed at similar levels during the larval and adult stages of development (PALLANCK et al. 1995B ). Temperature-sensitive alleles of dNSF1, termed comatose (here designated dNSF1^comt), were identified in a classical genetic screen for mutations causing paralysis at elevated temperature (SIDDIQI and BENZER 1976 ). Electrophysiological and ultrastructural studies of dNSF1^comt mutants have demonstrated that dNSF1 plays a priming role in preparing docked synaptic vesicles for fast calcium-triggered fusion (KAWASAKI et al. 1998 ). Further, biochemical characterization of dNSF1^comt mutants indicates that the dNSF1 priming role involves disassembly or rearrangement of a neural SNARE complex at the plasma membrane (TOLAR and PALLANCK 1998 ).

While studies of the dNSF1^comt mutants have told us much about the role that dNSF1 plays in neurotransmitter release, several features of dNSF1 function remain unexplored. For example, while it is well established that dNSF1 plays an important presynaptic role in the adult nervous system, it is not known whether it plays a similar functional role prior to the adult stage of development. Further, it remains unknown whether dNSF1 functions in a postsynaptic capacity to regulate glutamate receptor function or whether dNSF1 plays secretory roles outside of the nervous system. It also remains completely unclear what function is provided by the closely related dNSF2 gene. To investigate these issues, we generated mutations in dNSF2 and subjected dNSF2 mutants, along with novel loss-of-function dNSF1 mutants, to molecular and genetic analysis. Results of these analyses show that dNSF2 function is required for viability beginning at the first instar larval stage of development, while dNSF1 function is required for viability at the adult stage of development. Transgenic rescue experiments using the GAL4 system (BRAND and PERRIMON 1993 ) indicate that the primary function of dNSF1 resides in the nervous system, while the primary function of dNSF2 resides in mesoderm. These results, coupled with other work on NSF function, suggest the possibility that dNSF1 functions primarily in a presynaptic capacity to regulate neurotransmitter release and that dNSF2 may function in a postsynaptic capacity, perhaps in the regulation of glutamate receptor activity or synaptic maturation. Availability of dNSF1 and dNSF2 loss-of-function alleles will provide a foundation for directly testing these hypotheses.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Generation of dNSF2 mutations:
Identification of chromosomes with deficiencies that remove the dNSF2 gene was carried out using in situ hybridization to Drosophila polytene chromosomes as previously described (PALLANCK et al. 1995B ). To generate recessive lethal dNSF2 mutations, male flies isogenic for a third chromosome bearing the recessive marker scarlet (st) were fed 25 mM ethylmethanesulfonate (EMS) and mated to TM3/TM6B females. Balanced offspring from this cross, each of which bears a unique mutagenized third chromosome, were then individually mated to Df(3R)urd/TM6C flies. Because the Df(3R)urd chromosome also bears the st marker, an absence of scarlet-eyed progeny from this cross indicates that a recessive lethal mutation in the Df(3R)urd deletion interval resides on the mutagenized chromosome. Stocks bearing recessive lethal mutations in the Df(3R)urd region were established by recovering these mutations in trans to the TM6C balancer chromosome. To place these mutations into complementation groups, all of the mutants recovered in this screen were mated to one another. Pairs of mutants that failed to produce scarlet-eyed progeny when mated were considered to be in the same complementation group.

Transgenic Drosophila lines:
The dNSF1 transgene consisting of a dNSF1 cDNA under control of an hsp70 heat-shock promoter has been described (PALLANCK et al. 1995A ). A dNSF2 transgenic construct consisting of a full-length dNSF2 cDNA under control of the hsp70 promoter was generated by replacing the dNSF1 cDNA in the dNSF1 hsp70 transformation construct with the dNSF2 dN2.14 cDNA (PALLANCK et al. 1995B ). UAS-dNSF1 and UAS-dNSF2 transgenes were created by cloning the full-length dNSF1 dN20 cDNA (ORDWAY et al. 1994 ) and dNSF2 dN2.14 cDNA into the pUAST transformation vector (BRAND and PERRIMON 1993 ). All transgenes were introduced into flies using standard germline transformation methods.

A heat-shock-inducible dNSF2 transgenic line was used to rescue candidate dNSF2 mutants recovered in the Df(3R)urd interval by crossing the candidate dNSF2 mutants to a stock bearing the dNSF2 transgene in trans to an SM5 balancer chromosome and the Df(3R)urd chromosome in trans to the TM3 balancer chromosome and maintaining the offspring at 25°. dNSF2 mutants were identified as those stocks producing non-SM5, non-TM3 adult flies from this cross. To rescue dNSF2 mutants with the heat-shock dNSF1 transgenic line, flies were subjected to heat shocks of 38° for 1 hr each day throughout development. To test for rescue of the dNSF1^comt temperature-sensitive paralytic phenotype, adult male flies bearing a dNSF1^comt mutation, a UAS-dNSF1 or UAS-dNSF2 transgenic line, and the relevant GAL4 line were aged 3–6 days and examined in groups of 10 as described (PALLANCK et al. 1995A ). At least three groups of 10 flies each were tested for each experiment, and the standard deviation was determined for the three trials.

Sequencing of dNSF1 and dNSF2 mutations:
To generate flies suitable for isolating the dNSF1 mutations for sequence analysis, females bearing the dNSF1^clp and dNSF1^3-13-3 alleles in trans to the FM7 balancer chromosome were crossed to males bearing a heat-shock-inducible dNSF1 transgene. Offspring from this cross were subjected to daily heat shocks of 1 hr at 38° to allow survival of hemizygous dNSF1^clp and dNSF1^3-13-3 adult males. Genomic DNA was extracted from the hemizygous dNSF1 mutants and subjected to polymerase chain reaction using primers flanking the dNSF1 open reading frame. PCR products corresponding to the dNSF1 mutant alleles were separated from the smaller PCR product derived from the dNSF1 cDNA transgene using agarose gel electrophoresis. The dNSF1^clp and dNSF1^3-13-3 PCR products were recovered from the agarose gel and subjected to dideoxy sequencing using a combination of primers that allow full coverage of the entire dNSF1 open reading frame.

A similar strategy was used to sequence the dNSF2 mutations. Crosses were carried out to generate flies bearing the dNSF2 mutations in trans to the Df(3R)urd chromosome in a background of transgenic dNSF2 expression. dNSF2 sequences were amplified using PCR primers flanking the dNSF2 open reading frame, and the resulting products were fractionated on agarose gels to separate the dNSF2 mutant alleles from the smaller dNSF2 transgenic product. The relevant PCR products were recovered from the agarose gel and subjected to dideoxy sequencing using primers that allow full coverage of the dNSF2 open reading frame.

Germline clone analysis:
To generate female germlines homozygous for dNSF1 or dNSF2 mutations, the FLP/FRT, ovo^D1 system was used. Recombinant chromosomes bearing the dNSF1^clp or dNSF1^3-13-3 mutations as well as a proximal FRT101 element were kindly provided by Rick Ordway, and a chromosome bearing the dNSF2⁵⁵ mutation with a proximal FRT82B element was generated. dNSF1 germline clones were generated by mating females bearing the dNSF1^clp FRT101 or dNSF1^3-13-3 FRT101 chromosomes in trans to the FM7 balancer chromosome to FRT101 ovo^D1; hsFLP/hsFLP males. Offspring from this cross were heat shocked at 38° for 2 hr each day for two successive days beginning at the first instar larval stage of development to induce expression of the FLP recombinase from the heat-shock promoter. Because both heterozygous and homozygous ovo^D1-bearing germline clones degenerate (CHOU et al. 1993 ), the only oocytes produced in female offspring from this cross must be dNSF1^clp or dNSF1^3-13-3 homozygotes produced by FLP recombinase-induced mitotic recombination at the FRT sites (XU and RUBIN 1993 ). Females bearing dNSF1^clp or dNSF1^3-13-3 germline clones were then mated to males bearing an FM7 balancer chromosome marked with green fluorescent protein (GFP). Non-GFP embryos corresponding to hemizygous dNSF1^clp or dNSF1^3-13-3 males were identified using fluorescence microscopy, transferred to fresh medium, and monitored daily until lethality was observed.

dNSF2 germline clones were generated by mating females bearing the dNSF2⁵⁵ FRT82B chromosome in trans to a TM3 balancer chromosome to hsFLP; FRT82B ovo^D1 males. Offspring from this cross were heat shocked to induce expression of the FLP recombinase as described above, and female progeny were mated to males bearing the dNSF2 deletion chromosome, Df(3R)urd, in trans to a TM3 balancer chromosome marked with GFP. dNSF2⁵⁵/Df(3R)urd embryos were identified by their lack of GFP fluorescence and monitored daily until lethality was observed.

Preparation and analysis of SNARE complexes:
Flies used for examination of SNARE complexes were aged 3–6 days. Samples for SNARE complex analysis were prepared and subjected to electrophoresis and electroblotting onto nitrocellulose membranes as described (TOLAR and PALLANCK 1998 ). Blots were incubated with an affinity-purified polyclonal antiserum that recognizes the cytoplasmic domain of n-synaptobrevin (VAN DE GOOR et al. 1995 ). After staining blots with antiserum, bands were detected by enhanced chemiluminescence (Amersham, Arlington Heights, IL). Quantitation of n-synaptobrevin and SNARE complexes was performed using a Digital Science Image Station 440CF (Kodak, Rochester, NY) and 1D Image Analysis software (Kodak).

Fly strains:
The GAL4 driver lines used include the following: w- P{[w+] elav^C155-GAL4} (ROBINOW and WHITE 1988 ), w-; P{[w+] 24B-GAL4} (BRAND and PERRIMON 1993 ), P{[w+] BG57-GAL4} (BUDNIK et al. 1996 ), P{[ry+] MHC-82-GAL4} (SCHUSTER et al. 1996 ), and w-P{[w+]Dmef2-GAL4} (RANGANAYAKULU et al. 1996 ). Stocks bearing the GFP-marked TM3 and FM7 balancer chromosomes, as well as stocks necessary for the production of germline clones, are available from the Bloomington stock center. The specific stocks used in this work are as follows: (1) w-; TM3 P{[w+]ActGFP} Ser/Sb; (2) FM7i P{[w+]ActGFP}/C(1)DX; (3) y1 f1 w-; P{[ry+]neoFRT}82B P[w+]87E, which was used to generate the dNSF2⁵⁵ FRT82B recombinant chromosome; (4) w-; P{[ry+]neoFRT}82B P{[w+] ovo^D1-18}3R1P{[w+]ovo^D1-18}3R2/st betaTub85DD ss es/TM3, which was used to generate dNSF2⁵⁵ germline clones; (5) y w- v P{[w+]FRT101}, which was used to generate the FRT101 dNSF1^clp and FRT101 dNSF1^3-13-3 recombinant chromosomes; and (6) w-ovo^D1 v24 P{[w+] FRT101}/C(1)DX; hsFLP/hsFLP, which was used to generate dNSF1^clp and dNSF1^3-13-3 germline clones.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Generation of dNSF2 mutations:
The Drosophila dNSF2 gene has been previously localized to polytene region 87F14-15 on the right arm of chromosome 3 (PALLANCK et al. 1995B ). To generate mutations in this gene, chromosomes bearing deficiencies in this region were subjected to in situ hybridization using a dNSF2 cDNA probe to identify deficiencies that remove the dNSF2 gene. One of these chromosomes, Df(3R)urd, failed to produce a hybridization signal in this analysis, indicating that the dNSF2 gene lies within the boundaries of the Df(3R)urd deficiency (data not shown). This chromosome was used in an F₂ genetic screen to identify recessive lethal mutations in the deletion interval. From a screen of 7000 EMS-mutagenized third chromosomes, 18 mutations were recovered that were lethal in trans to the Df(3R)urd chromosome. Crosses conducted with these 18 mutants indicated that they represent six complementation groups, each containing between one and seven alleles.

To determine which of these complementation groups represent dNSF2, a transposable P-element construct containing a dNSF2 cDNA under the control of a heat-shock-inducible promoter was introduced into the Drosophila germline and tested for its ability to rescue the lethality associated with the 18 candidate dNSF2 mutations. The results of this analysis indicate that the dNSF2 gene is represented by eight recessive lethal alleles. Six of the seven alleles from one complementation group (designated dNSF2¹¹, dNSF2¹⁵, dNSF2⁴², dNSF2⁵⁵, dNSF2¹⁰², and dNSF2¹⁰³) were efficiently rescued from lethality by the dNSF2 transgene, and all of the rescued flies appeared normal and were fertile in crosses. The one allele from this complementation group that could not be rescued by the dNSF2 transgene (dNSF2⁶⁶) also failed to complement a mutation in a distinct complementation group, suggesting that the dNSF2⁶⁶ mutation may affect more than one gene. The dNSF2 transgene also rescued a mutation that complemented all of the other mutations in the Df(3R)urd interval (designated dNSF2²¹), indicating that this mutation may represent a hypomorphic dNSF2 allele.

Characterization of dNSF2 mutants:
In an effort to rank the different dNSF2 alleles in terms of their severity, the dNSF2 mutants were subjected to lethal phase analysis, and the molecular nature of these alleles was determined through direct sequencing of genomic DNA. All of the dNSF2 mutants appear to develop normally as embryos, are able to hatch from the egg case, and show normal larval locomotion as hemizygotes with the Df(3R)urd chromosome. However, the most severe dNSF2 mutants ceased moving and died during the mid-first instar larval stage of development (Table 1). Two of the most severe alleles, dNSF2¹⁵ and dNSF2⁵⁵, contain premature stop codons in the dNSF2 coding sequence and likely represent null alleles of the dNSF2 gene (Fig 1). Another of the most severe alleles, dNSF2¹⁰³, results in the replacement of the final 314 codons of the dNSF2 gene with 43 novel codons followed by a premature translation termination signal (Fig 1). The remaining alleles were found to be missense mutations or were devoid of mutations in the coding sequence, and most were found to cause death later in development, suggesting that many of these mutations are not null alleles of dNSF2 (Fig 1; Table 1).

View larger version (10K): In this window In a new window Download PPT slide   Figure 1. The amino acid changes caused by six dNSF2 and two dNSF1 recessive lethal mutations are shown above and below the NSF domain structure, respectively (TAGAYA et al. 1993 ). Numbers refer to the amino acid position in the protein sequence. The dNSF2103 mutation is a 334-nucleotide deletion that completely removes a 240-nucleotide intron, as well as flanking 5' and 3' exon sequences. This deletion fuses the first two bases of dNSF2 codon 438 to the first base of codon 470. Theoretical translation of the dNSF2103 allele predicts a severely truncated polypeptide with 43 novel amino acids following threonine 438 before terminating at an out-of-frame stop codon. The dNSF1clp mutation is a one-nucleotide deletion of the first position of codon 709, which upon translation should result in replacement of the 37 carboxy-terminal dNSF1 amino acids with 18 novel amino acids. The dNSF221 allele was also subjected to sequence analysis, but no mutation was identified in the coding sequence, suggesting that the dNSF221 mutation may affect dNSF2 transcription or splicing. Hatched box, N-domain; shaded box, ATPase domain 1; open box, ATPase domain 2.

To investigate whether the lethal phase of the dNSF2 mutants is influenced by a possible maternal deposition of dNSF2 protein, the FLP/FRT, ovo^D1 system (CHOU and PERRIMON 1996 ) was used to create female flies with germlines homozygous for a putative null allele of dNSF2, dNSF2⁵⁵. These females were crossed to males bearing a Df(3R)urd chromosome in trans to a balancer chromosome marked with GFP, and the lethal phase of the non-GFP offspring of this cross, which are devoid of germline and zygotic expression of dNSF2, was monitored. Elimination of maternal and zygotic dNSF2 expression was found to result in a first instar larval lethal phase, indicating that maternal contribution of dNSF2 protein does not affect the dNSF2 lethal phase (Table 1).

Characterization of dNSF1 mutants:
To compare and contrast dNSF2 and dNSF1 function, severe loss-of-function alleles of dNSF1 were obtained and characterized. The recessive lethal alleles of dNSF1 chosen for this analysis, dNSF1^clp (kindly provided by K. S. Krishnan) and dNSF1^3-13-3 (kindly provided by S. Titus and B. Ganetzky), were obtained from noncomplementation screens using the conditional comatose alleles of the dNSF1 gene. Direct sequencing of genomic DNA corresponding to the dNSF1^clp allele revealed a one-nucleotide deletion in the carboxy-terminal coding region of the dNSF1 gene, which, if translated, would produce a dNSF1 polypeptide with 18 novel amino acids replacing the normal 37 carboxy-terminal amino acids (Fig 1). The dNSF1^3-13-3 allele contains a missense mutation that results in an E to K substitution at amino acid position 606 of the second ATPase domain (Fig 1). While the nature of the molecular lesions associated with these two dNSF1 loss-of-function alleles leaves open the possibility for residual dNSF1 function, Western blot analysis of protein extracts obtained from dNSF1^clp mutants indicates that this allele fails to produce detectable dNSF1 protein (data not shown). Thus, the dNSF1^clp mutation may represent a null allele of this gene.

Lethal phase analysis conducted with males hemizygous for the dNSF1^clp and dNSF1^3-13-3 alleles indicates that these mutations result in pharate adult lethality (Table 1). In many cases the flies bearing these mutations manage to protrude their heads out of the pupal case, but they are rarely able to eclose completely. When dissected from the pupal case, the dNSF1^clp and dNSF1^3-13-3 mutants appear morphologically normal but are incapable of coordinated movement and die within several days. The lethal phase of these mutants cannot be explained by secondary mutations residing on the X chromosomes bearing the dNSF1^clp and dNSF1^3-13-3 mutations, as males bearing these alleles can be fully rescued by ectopic expression of dNSF1 protein from a transgene (see below).

A possible maternal contribution of dNSF1 protein does not appear to account for the survival of the most severe dNSF1 mutants to the pharate adult stage of development. The FLP/FRT, ovo^D1 system (CHOU and PERRIMON 1992 ) was used to create female flies bearing a germline homozygous for the dNSF1^clp or dNSF1^3-13-3 mutations, and these females were crossed to males bearing an X chromosome balancer marked with GFP. Non-GFP male offspring from these crosses, hemizygous for one of the two dNSF1 mutant alleles, also die as pharate adults (Table 1), indicating that wild-type dNSF1 protein is not required prior to the adult stage of development. Further support for an adult-specific requirement of dNSF1 protein for viability is provided by lethal phase analysis conducted with the temperature-sensitive comatose alleles of dNSF1. When dNSF1^{comt st17} and dNSF1^{comt tp7} mutants are shifted to the restrictive temperature of 29° within the first 24 hr of development, these mutants exhibit a lethal phase identical to that of the dNSF1^clp and dNSF1^3-13-3 mutants (Table 1). Rare adult dNSF1^{comt st17} and dNSF1^{comt tp7} flies that manage to eclose at 29° exhibit the same uncoordinated movement as the dNSF1^clp and dNSF1^3-13-3 mutants dissected from the pupal case and also die within several days if maintained at 29°. Control experiments carried out with wild-type flies confirmed that these phenotypes are specific to the dNSF1^comt mutants.

Although hemizygous male larvae bearing these dNSF1 alleles are viable, they exhibit several phenotypes. For example, when rolled onto their dorsal surface, the larvae take far longer to reorient to their ventral surface than do wild-type larvae (data not shown). In addition, dNSF1^clp and dNSF1^3-13-3 hemizygous male larvae are often found protruding vertically from the surface of the medium for long periods of time and often burrow deeply into the medium and die, possibly as a result of suffocation. dNSF1^clp and dNSF1^3-13-3 hemizygous male larvae also accumulate gas in their intestinal tract, which may result from a defect in secretion of digestive enzymes. These results indicate that the dNSF1 gene provides a nonessential function during the larval stage of development.

Tissue requirements of dNSF1 and dNSF2:
To identify tissues requiring dNSF1 and dNSF2 function, the GAL4/UAS system (BRAND and PERRIMON 1993 ) was used to drive ectopic expression of these proteins in specific locations in a dNSF1 or dNSF2 mutant background. Previous work has shown that the dNSF1 gene product plays an important role in the regulated release of chemical neurotransmitters in the adult Drosophila nervous system (SIDDIQI and BENZER 1976 ; KAWASAKI et al. 1998 ; LITTLETON et al. 1998 ; TOLAR and PALLANCK 1998 ; KAWASAKI and ORDWAY 1999 ). Thus, to investigate whether dNSF1 function is required solely in neural tissue, the neuron-specific elav-GAL4 element was used to test whether expression of dNSF1 protein in the nervous system could rescue the phenotypes associated with mutations in the dNSF1 gene (Fig 2A). Two independently generated UAS-dNSF1 transgenic lines (designated 21A and 73C) containing a dNSF1 cDNA downstream of multiple UAS elements were tested in combination with the dNSF1 lethal alleles, dNSF1^clp and dNSF1^3-13-3, and the temperature-sensitive paralytic dNSF1^{comt st17} and dNSF1^{comt tp7} alleles. All crosses involving the dNSF1 lethal alleles produced behaviorally and morphologically normal non-FM7 (rescued) male progeny, indicating that expression of dNSF1 protein in the nervous system is sufficient to rescue the lethality of the dNSF1^clp and dNSF1^3-13-3 mutations (Fig 2B). Furthermore, neural expression of wild-type dNSF1 protein greatly alleviated the temperature-sensitive paralytic phenotype induced by the dNSF1^{comt st17} or dNSF1^{comt tp7} mutations (Fig 2C). In several cases, the rescued flies are indistinguishable from wild-type flies; wild-type Canton-S flies cease locomotion after roughly 80 min at 38° (data not shown). These results indicate that the dNSF1 gene plays an essential role in the nervous system and that nervous system dysfunction underlies the temperature-sensitive paralytic phenotype of the dNSF1^comt mutants.

View larger version (38K): In this window In a new window Download PPT slide   Figure 2. Ectopic expression of dNSF1 in the nervous system rescues the phenotypes of dNSF1 mutations. (A) The crosses used for the rescue experiments. Females bearing an X chromosome with a dNSF1 mutation (dNSF1clp, dNSF13-13-3, dNSF1comt st17, or dNSF1comt tp7) and the elav-GAL4 element in trans to an FM7 balancer chromosome were crossed to transgenic males homozygous for a UAS-dNSF1 transgene on the second chromosome. Male progeny from this cross that lack the FM7 balancer chromosome are hemizygous for the dNSF1 mutation. (B) Rescue of dNSF1 lethal alleles. dNSF1clp and dNSF13-13-3 were tested in combination with two UAS-dNSF1 transgenic lines, designated 21A (hatched bars) and 73C (shaded bars). Percentage rescue is defined as the percentage of non-FM7 (rescued) male progeny with respect to the total number of female progeny, divided by 0.5, the expected ratio if flies were fully rescued to Mendelian proportions. A minimum of 250 progeny were scored from each cross. (C) Rescue of dNSF1 temperature-sensitive alleles. Time to paralysis represents the time required for one-half of the hemizygous dNSF1 mutants to become paralyzed at 38°. Solid bars indicate the time to paralysis for flies carrying a dNSF1 mutation and elav-GAL4 element, but no UAS-dNSF1 transgene. No rescue of the dNSF1 recessive lethal or temperature-sensitive paralytic mutations was observed in flies segregating a UAS-dNSF1 transgene in the absence of the elav-GAL4 element.

To examine whether similar neural expression of dNSF2 protein could rescue the lethality associated with dNSF2 mutations, UAS-dNSF2 transgenic lines were generated and tested for their ability to rescue dNSF2 mutants using the elav-GAL4 driver (Fig 3A). Three of the most severe dNSF2 alleles (dNSF2¹⁵, dNSF2⁵⁵, and dNSF2¹⁰³) and two independent UAS-dNSF2 lines (designated 2A and 23A) were tested in this analysis. Crosses involving dNSF2¹⁵ and dNSF2¹⁰³ alleles produced no rescued adult offspring, and those involving the dNSF2⁵⁵ mutation produced only three apparently rescued adult offspring among the >750 adults scored from this cross (Fig 3C, solid bar). Further analysis of dNSF2 mutants expressing dNSF2 in the nervous system indicates that this manipulation results primarily in first and second instar larval lethality, with a small percentage of offspring surviving to later stages of development. These results suggest that dNSF2 plays some role in the nervous system. However, neural expression of dNSF2 alone is insufficient for efficient rescue of the dNSF2 mutant phenotype.

View larger version (41K): In this window In a new window Download PPT slide   Figure 3. Ectopic expression of dNSF2 in mesoderm is sufficient to rescue the lethality of dNSF2 mutations. (A) Rescue crosses involving the nerve-specific elav-GAL4 element. Females homozygous for an elav-GAL4 element on the X chromosome and a dNSF2 mutation (dNSF215, dNSF255, or dNSF2103) in trans to the TM3 balancer chromosome were crossed to males bearing a UAS-dNSF2 transgene on the second chromosome and a Df(3R)urd chromosome in trans to TM3. Progeny lacking the TM3 balancer chromosome represent rescue of the dNSF2 mutation. (B) Rescue crosses involving the mesoderm-specific 24B-GAL4 element in combination with elav-GAL4. Males hemizygous for an elav-GAL4 element and bearing a dNSF2 mutation in trans to TM3 were crossed to females homozygous for a UAS-dNSF2 element on the second chromosome and carrying a recombinant third chromosome consisting of the Df(3R)urd deletion and the 24B-GAL4 element in trans to the TM3 balancer chromosome. Male progeny from this cross that lack the TM3 balancer chromosome contain only the 24B-GAL4 element and thus represent dNSF2 mutants rescued by expression of dNSF2 in mesoderm. Female progeny from this cross that lack the TM3 balancer chromosome contain both the 24B-GAL4 and the elav-GAL4 elements and thus represent dNSF2 mutants rescued by expression of dNSF2 in both the nervous system and mesoderm. (C) Rescue of the dNSF2 lethal alleles. Two UAS-dNSF2 transgenic lines (2A and 23A) were tested in combination with three dNSF2 alleles. Percentage rescue of the dNSF2 mutations mediated by nervous system expression of dNSF2 protein (solid bar), by expression of dNSF2 protein in mesoderm only (hatched bars), or by expression of dNSF2 protein in both the nervous system and mesoderm (shaded bars) is shown. Percentage rescue was calculated by dividing the percentage of non-TM3 progeny by 0.17, the expected ratio if flies were fully rescued to expected Mendelian proportions. A minimum of 350 flies were counted for each cross. No rescue of the dNSF2 mutants was observed in flies segregating a UAS-dNSF2 transgene in the absence of GAL4 elements.

Recent studies demonstrating a direct involvement of the vertebrate NSF in glutamate receptor function (NISHIMUNE et al. 1998 ; OSTEN et al. 1998 ; SONG et al. 1998 ; LUSCHER et al. 1999 ; LUTHI et al. 1999 ; NOEL et al. 1999 ), coupled with the fact that neuromuscular transmission in Drosophila is mediated by glutamatergic transmission (JAN and JAN 1976 ; PETERSEN et al. 1997 ; DIANTONIO et al. 1999 ), led us to examine whether dNSF2 function is required in muscle. To test this hypothesis, the mesoderm-specific 24B-GAL4 line, which drives expression of GAL4 protein specifically in the tissues that give rise to somatic and visceral muscles, was used in conjunction with the UAS-dNSF2 transgenic lines in an attempt to rescue the dNSF2 mutant phenotype (Fig 3B). In addition, the possibility that dNSF2 function is required in both the nervous system and the musculature was tested by setting up crosses in which males segregating an elav-GAL4 element as well as one of three dNSF2 alleles were mated to females segregating the Df(3R)urd chromosome, a 24B-GAL4 element, and a UAS-dNSF2 transgene (Fig 3B). These crosses produced behaviorally and morphologically normal rescued progeny that fell into two classes: (1) rescued males, which express dNSF2 protein only in mesoderm (Fig 3C, hatched bars); and (2) rescued females, which express dNSF2 protein simultaneously in the nervous system and mesoderm (Fig 3C, shaded bars). These results indicate that mesoderm expression of dNSF2 is sufficient to produce viable adults. However, the extent of rescue was higher in flies expressing dNSF2 protein in both the nervous system and mesoderm than that in flies expressing dNSF2 protein only in mesoderm, indicating that dNSF2 gene function is required primarily in mesoderm, but that dNSF2 also plays a secondary role in the nervous system.

To confirm these findings, several additional mesoderm-specific GAL4 lines were tested for their ability to mediate rescue of the recessive lethal dNSF2 mutant phenotype. These GAL4 lines were also used to test whether mesoderm expression of dNSF1 is capable of rescuing the dNSF1 mutant phenotypes. For this analysis, only the dNSF1^clp and dNSF2¹⁵ alleles were used in conjunction with the UAS-dNSF1 21A and UAS-dNSF2 2A transgenes, respectively. All of the mesoderm-specific GAL4 lines tested mediated rescue of the dNSF2 mutation to at least 40% of expected Mendelian levels, while none of those tested produced any rescue of the dNSF1 mutation (Table 2). These results verify that the primary function of dNSF2 resides in mesoderm and indicate that the dNSF1 mutant phenotype can be rescued only by nervous system expression of dNSF1 protein.

Intergenic complementation analysis:
To investigate the degree of functional overlap between the two Drosophila NSF proteins, the ability of dNSF1 and dNSF2 to substitute for one another when ectopically expressed was examined using the GAL4/UAS system. Because dNSF1 gene function is required in the nervous system, we sought to test whether a UAS-dNSF2 transgene, expressed in neural tissue under the direction of elav-GAL4, could rescue the phenotypes associated with mutations in the dNSF1 gene. For these crosses, females bearing an elav-GAL4 element and a dNSF1 mutation were crossed to males homozygous for a UAS-dNSF2 transgene (Fig 4A). Results of these crosses indicate that dNSF2 expression in the nervous system is sufficient to rescue the lethality of the dNSF1^clp and dNSF1^3-13-3 mutations (Fig 4B). Similarly, expression of dNSF2 in the nervous system rescued the temperature-sensitive paralytic phenotype of the dNSF1^{comt st17} and dNSF1^{comt tp7} mutants. The extent of rescue of the dNSF1 mutations by dNSF2 is comparable to that observed when these mutations are rescued by neural expression of dNSF1 protein.

View larger version (40K): In this window In a new window Download PPT slide   Figure 4. Neural expression of dNSF2 rescues the dNSF1 mutant phenotypes. (A) The crosses used for the rescue experiments. Females carrying a dNSF1 mutation and an elav-GAL4 element were crossed to transgenic males homozygous for a UAS-dNSF2 transgene on the third chromosome. Male progeny lacking the FM7 balancer are hemizygous for the dNSF1 mutation. (B) Rescue of dNSF1 lethal alleles. dNSF1clp and dNSF13-13-3 were tested in combination with the UAS-dNSF2 transgenic lines, 2A (hatched bars) and 23A (shaded bars). Percentage rescue was defined as the percentage of non-FM7 (rescued) males to the total number of females, divided by 0.5, the expected ratio if flies were fully rescued to Mendelian proportions. A minimum of 350 flies were counted for each cross. (C) Rescue of dNSF1 temperature-sensitive alleles. dNSF1comt st17 and dNSF1comt tp7 were tested in combination with the two UAS-dNSF2 transgenic lines, 2A (hatched bars) and 23A (shaded bars). Time to paralysis represents the time required for one-half of the hemizygous dNSF1 mutants to become paralyzed at 38°. Solid bars designate paralytic profiles of the dNSF1comt mutants in the absence of ectopic dNSF2 expression. No rescue of the dNSF1 recessive lethal or temperature-sensitive paralytic alleles was observed in flies segregating the UAS-dNSF2 transgene in the absence of the elav-GAL4 element.

Previous work conducted with the dNSF1^comt alleles has demonstrated that a neural SNARE complex composed of a v-SNARE, n-synaptobrevin, and two t-SNAREs, SNAP-25 and syntaxin, accumulates at the dNSF1^comt restrictive temperature (TOLAR and PALLANCK 1998 ). This biochemical phenotype is largely rescued by expression of wild-type dNSF1 protein from a transgene (TOLAR and PALLANCK 1998 ). To examine whether dNSF2-mediated rescue of the dNSF1^comt temperature-sensitive paralytic phenotype is paralleled by rescue of the biochemical defect in these mutants, we compared SNARE complex levels in wild-type flies, dNSF1^comt mutants, and dNSF1^comt mutants expressing either dNSF1 or dNSF2 in the nervous system at restrictive temperature. Crosses for this analysis were performed as shown in Fig 2A and Fig 4A. Western blot analysis of a protein lysate from wild-type fly heads using antiserum to the v-SNARE n-synaptobrevin revealed a 23-kD band representing monomeric n-synaptobrevin and a 73-kD band representing the SNARE complex (Fig 5). As demonstrated previously, SNARE complex levels were dramatically elevated with respect to monomeric n-synaptobrevin in dNSF1^comt mutants at the restrictive temperature as compared to wild-type flies (LITTLETON et al. 1998 ; TOLAR and PALLANCK 1998 ). Ectopic neural expression of either dNSF1 or dNSF2 in the dNSF1^comt mutants dramatically alleviates the SNARE complex accumulation phenotype. All of the UAS-dNSF1 and UAS-dNSF2 lines reduce the ratio of SNARE complex to n-synaptobrevin monomer to near, or even below, wild-type levels. Those transgenic lines that reduce the SNARE complex accumulation phenotype most also tended to confer more efficient rescue of the paralytic phenotype of the dNSF1^comt mutants, indicating that these two phenotypes are directly related. These results show that when expressed in the nervous system, dNSF2 can catalyze disassembly of neuronal SNARE complexes and rescue the dNSF1^comt temperature-sensitive paralytic phenotype, demonstrating that dNSF2 protein can effectively substitute for dNSF1 protein.

View larger version (42K): In this window In a new window Download PPT slide   Figure 5. Neural expression of dNSF1 or dNSF2 rescues SNARE complex accumulation in dNSF1comt mutants. Protein extracts for Western blot analysis were prepared from wild-type (CS), dNSF1comt st17, and dNSF1comt tp7 flies carrying UAS-dNSF1 (21A or 73C) or UAS-dNSF2 (2A or 23A) transgenes driven by the elav-GAL4 promoter, following a 20-min exposure to restrictive temperature (38°). Western blot analysis was performed using a polyclonal antibody to n-synaptobrevin (VAN DE GOOR et al. 1995 ). SNARE complex/synaptobrevin monomer ratios, normalized to wild type, are shown below each lane.

To investigate whether ectopic expression of dNSF1 in mesoderm can rescue the dNSF2 mutant phenotype, we tested whether a UAS-dNSF1 transgene expressed using 24B-GAL4 could rescue the lethality associated with the dNSF2¹⁵, dNSF2⁵⁵, and dNSF2¹⁰³ mutations. None of these crosses produced rescued progeny from among the >350 flies counted per cross, indicating that dNSF1 expression in mesoderm is insufficient to rescue the dNSF2 mutations. As mesoderm expression of dNSF2 was only partially capable of rescuing lethality of the dNSF2 mutations, we investigated whether more widespread and possibly higher levels of dNSF1 expression could rescue the dNSF2 mutant phenotype. To perform these experiments, we made use of a heat-shock-inducible dNSF1 transgene. The heat-shock dNSF1 transgene was introduced into a dNSF2 mutant background by mating flies bearing this transgene and the Df(3R)urd chromosome to flies bearing one of the three dNSF2 alleles and subjecting the progeny to daily heat shocks throughout development (Fig 6A). In contrast to the lack of rescue of dNSF2 mutations observed when dNSF1 expression is restricted to mesoderm, significant numbers of rescued progeny were observed using an extensive heat-shock regimen to drive ubiquitous expression of dNSF1 protein, demonstrating that dNSF1 protein is capable of rescuing mutations in dNSF2 (Fig 6B). Together these results show that the dNSF1 and dNSF2 proteins are capable of substituting for one another under defined conditions of expression.

View larger version (25K): In this window In a new window Download PPT slide   Figure 6. Ectopic expression of dNSF1 mediated by a heat-shock-inducible promoter rescues the recessive lethal dNSF2 mutant phenotype. (A) The crosses used for the rescue experiments. Females homozygous for a heat-shock-dNSF1 transgene (HS-dNSF1) on the X chromosome and carrying the Df(3R)urd chromosome in trans to the TM3 balancer chromosome were crossed to males carrying a dNSF2 mutation in trans to the TM6C balancer chromosome. Offspring were heat shocked at 38° for 60 min daily throughout development. Progeny lacking both the TM3 and TM6C chromosomes represent rescue of the dNSF2 mutations by dNSF1 expression. (B) Rescue of the dNSF2 mutations. Percentage rescue was defined as the percentage of non-TM3, non-TM6C progeny divided by 0.33, the expected ratio if flies were fully rescued to expected Mendelian proportions. No rescued progeny were observed in the absence of heat shock. A minimum of 200 flies were counted for each cross.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In an effort to explore the functions of the two NSF genes in Drosophila, we generated mutations in the dNSF2 gene and used novel recessive lethal alleles of dNSF1 to examine the temporal and spatial requirements and degree of functional overlap between these two genes. Our results show that both Drosophila NSF genes are required for viability. Lethal phase and germline clone analyses of these mutants show that dNSF1 function is required for viability at the adult stage of development, while the requirement for dNSF2 begins at the first instar larval stage. Ectopic expression studies demonstrate that the essential function of dNSF1 resides in the nervous system, while the essential function of dNSF2 resides in mesoderm. Despite their differing temporal and spatial requirements, ectopic expression studies show that although they do not normally do so, the dNSF1 and dNSF2 proteins are capable of substituting for one another, indicating that they have overlapping functional properties.

The finding that expression of dNSF1 in the nervous system rescues the recessive lethal and temperature-sensitive paralytic dNSF1 mutant phenotypes is consistent with the well-characterized functional role of dNSF1 in neurotransmission (SIDDIQI and BENZER 1976 ; KAWASAKI et al. 1998 ; LITTLETON et al. 1998 ; TOLAR and PALLANCK 1998 ; KAWASAKI and ORDWAY 1999 ). Further experiments demonstrating that dNSF1 function is required only in adults are also consistent with the finding that this gene is expressed at highest levels at this stage of development (PALLANCK et al. 1995B ). Although these observations indicate that dNSF1 is dispensable in the embryonic and larval nervous systems, the dNSF1 mutants exhibit a variety of larval behavioral phenotypes, including locomotion defects and temperature-sensitive paralysis (SIDDIQI and BENZER 1976 ), suggesting that dNSF1 plays nonessential functional roles in the larval nervous system. dNSF1 mutant larvae also display features of intestinal system dysfunction, suggesting that this gene may also provide nonneuronal functions.

In contrast to results obtained in studies of the dNSF1 mutations, ectopic expression of dNSF2 protein in the nervous system was insufficient for significant rescue of the recessive lethal dNSF2 mutant phenotype. Instead, rescue of the dNSF2 mutants was produced by ectopic expression of dNSF2 protein in mesoderm. These results are supported by previous work establishing that dNSF2 is expressed in this tissue (BOULIANNE and TRIMBLE 1995 ). Mesoderm is the precursor to several larval and adult tissues in Drosophila, principally visceral and somatic muscle (BATE 1993 ), suggesting that dNSF2 may play an essential functional role in muscle. The fact that dNSF2 mutant larvae initially display normal locomotion suggests that muscle differentiation and early events in neuromuscular junction formation are not adversely affected by the dNSF2 mutations. A model more consistent with the dNSF2 lethal phase is that the dNSF2 mutants are defective in the synaptic growth and maturation that occurs during larval development, a process that requires proper trafficking of muscle-specific membrane proteins and secretion of a retrograde signal that acts to regulate presynaptic activity (PETERSEN et al. 1997 ; DAVIS et al. 1998 ; DIANTONIO et al. 1999 ; GRAMATES and BUDNIK 1999 ).

Alternatively, another possible role for dNSF2 in muscle is suggested by recent work demonstrating that postsynaptic glutamate receptor function or subcellular localization is regulated by NSF in the vertebrate nervous system (NISHIMUNE et al. 1998 ; OSTEN et al. 1998 ; SONG et al. 1998 ; LUSCHER et al. 1999 ; LUTHI et al. 1999 ; NOEL et al. 1999 ). Glutamatergic neurotransmitter receptors reside in the Drosophila larval body wall muscles in opposition to neurotransmitter release sites and are essential for fast synaptic transmission at the larval neuromuscular junction (JAN and JAN 1976 ; SCHUSTER et al. 1991 ; PETERSEN et al. 1997 ; DIANTONIO et al. 1999 ). Thus, similar interactions between dNSF2 and Drosophila glutamate receptors in muscle could explain the mesoderm-specific role of dNSF2. Although the vertebrate glutamate receptor sequences that mediate interactions with NSF are poorly conserved in the Drosophila glutamate receptors (SCHUSTER et al. 1991 ; PETERSEN et al. 1997 ; NISHIMUNE et al. 1998 ; OSTEN et al. 1998 ; SONG et al. 1998 ), the dNSF2 protein may have evolved compensatory changes to allow maintenance of this interaction. Another possibility is that dNSF2 interacts with other muscle proteins involved in synapse development or function that contain sequences similar to the NSF-binding sequence present in mammalian glutamate receptors. Further experiments are currently in progress to define the precise functional role of dNSF2 in muscle.

In addition to the role of dNSF2 in mesoderm, the increased efficiency of rescue of dNSF2 mutations mediated by simultaneous expression of dNSF2 in the nervous system and mesoderm indicates that dNSF2 also functions in the nervous system. This is also supported by the finding that expression of dNSF2 in the nervous system alone can extend the lethal phase of the dNSF2 mutants. It is possible that dNSF2 participates in constitutive vesicle trafficking in the nervous system, and/or that dNSF2 collaborates with dNSF1 in presynaptic neurotransmitter release mechanisms, as suggested by experiments showing that dNSF2 can rescue the dNSF1 mutations and can participate in neural SNARE complex disassembly. Alternatively, dNSF2 may be involved in regulating postsynaptic membrane insertion or localization of glutamate receptors in the central nervous system. Future electrophysiological studies and glutamate receptor localization experiments involving the dNSF2 mutants should resolve these possibilities.

A somewhat surprising result emerging from our studies is that neither dNSF1 nor dNSF2 is singly required for embryonic development and early larval locomotion. This observation could be accounted for in three ways: dNSF1 and dNSF2 may be redundant at these stages of development, a distantly related Drosophila NSF homolog may promote secretion in embryos, or there may be no requirement for NSF-like activity in developing embryos. The last possibility appears highly unlikely, since several other secretory components that interact either directly or indirectly with NSF have been shown to be required for embryonic development (HARRISON et al. 1994 ; SCHULZE et al. 1995 ; BURGESS et al. 1997 ). Although the second possibility is difficult to rule out, the finding that both dNSF1 and dNSF2 are expressed in embryos favors the idea that these two Drosophila NSF genes perform redundant functions at this stage of development (ORDWAY et al. 1994 ; BOULIANNE and TRIMBLE 1995 ; PALLANCK et al. 1995B ). Lack of an essential role for either gene in the larval nervous system argues for similar redundancy of dNSF1 and dNSF2 function in this tissue.

Although previous work has shown that ectopic expression of dNSF1 protein from a heat-shock-inducible transgene can rescue the neural SNARE complex accumulation phenotype of the dNSF1^comt mutants to near wild-type levels (TOLAR and PALLANCK 1998 ), several of the transgenic lines used in the current study were found to reduce SNARE complex levels to below those found in wild-type flies. In particular, the UAS-dNSF2 23A transgenic line reduced the SNARE complex to synaptobrevin monomer ratio of the dNSF1^{comt tp7} mutants by fivefold with respect to the wild-type ratio. This finding bears on the functional role of NSF. According to a recent model, the SDS-resistant neural SNARE complex is hypothesized to define a docked, readily releasable synaptic vesicle (LONART and SUDHOF 2000 ). If this model is correct, it predicts a sharp reduction in the readily releasable synaptic vesicle pool in flies expressing dNSF2 protein from the UAS-dNSF2 23A transgene, and this severe reduction in the readily releasable synaptic vesicle pool might well be expected to produce correspondingly severe behavioral manifestations. However, flies of this genotype are viable and exhibit no unusual behavioral phenotypes, suggesting that the readily releasable synaptic vesicle pool is not severely depleted. These results more readily fit a role for NSF in the disassembly of a SNARE complex formed transiently during synaptic vesicle fusion or, alternatively, a role for NSF in catalyzing partial or complete disassembly of a trans SNARE complex to activate synaptic vesicles for calcium-triggered fusion.

It is notable that despite their differing spatial requirements, the two Drosophila NSF proteins are capable of substituting for one another when expressed ectopically. Specifically, dNSF2 can substitute for dNSF1 when expressed in the nervous system, and dNSF1 can substitute for dNSF2 when expressed ubiquitously from a heat-shock promoter. Although it remains unclear why expression of dNSF1 in mesoderm alone fails to rescue the dNSF2 mutants, a likely possibility is that dNSF1 does not efficiently substitute for dNSF2. Our experiments involving dNSF2 transgenes indicate that optimal rescue of dNSF2 mutants requires expression in both mesoderm and neural tissue. The fact that dNSF1 can rescue dNSF2 mutants only when expressed from a heat-shock promoter may reflect the broader expression conferred by this system or possibly higher levels of expression resulting from the extensive heat-shock regimen employed (see MATERIALS AND METHODS). Additional experiments using heat-shock promoter-driven constructs, suggest that dNSF2 does not substitute efficiently for dNSF1 (R. ORDWAY, personal communication; our data not shown). These cross-rescue barriers, preventing efficient substitution of one NSF for the other, may involve subtle differences in the functional properties of the proteins, differential stabilities of the two proteins in the various tissues examined, sequence elements in the dNSF1 and dNSF2 mRNAs responsible for proper mRNA metabolism, or any combination of these three possibilities. Recent studies suggest that untranslated sequence elements in the dNSF1 and dNSF2 mRNAs are at least one important determinant preventing efficient cross-rescue (R. ORDWAY, personal communication).

The reason that Drosophila requires two closely related NSF genes remains unclear. Searches of the Caenorhabditis elegans and human genome databases reveal the presence of only one NSF gene in each of these two organisms, indicating that a single NSF gene is sufficient to support constitutive and regulated secretion, and possibly postsynaptic glutamate receptor regulation in these eukaryotes. However, this finding does not rule out the possible presence of multiple NSF proteins resulting from alternative splicing of the single NSF gene in these organisms. Evidence in support of this possibility is provided by Northern blot analysis of the rat NSF gene (PUSCHEL et al. 1994 ). Thus, the presence of multiple NSF genes in Drosophila may simply reflect the mechanism by which evolution has generated multiple NSF proteins with subtle functional differences in this organism. Partitioning of NSF function between two different NSF genes in Drosophila may make this organism particularly useful for teasing apart the distinct cellular roles of NSF.

	ACKNOWLEDGMENTS

The authors express their gratitude to the following individuals: Carrie Jones and Robert Kreber for their assistance in the genetic screen to generate dNSF2 mutants; K. S. Krishnan and Barry Ganetzky for providing dNSF1 loss-of-function mutations; Steve Titus for assistance in sequencing of dNSF1 and dNSF2 alleles; Hugo Bellen and Mingli Zhao for generating UAS-dNSF2 transgenic lines, providing GAL4 lines, and providing insight into the tissues requiring dNSF2 function; Regis Kelly for providing antiserum to n-synaptobrevin; Richard Ordway for providing dNSF1 FRT101 recombinant chromosomes and GAL4 lines and for communication of unpublished results; and Liqun Luo and Meg Winberg for providing GAL4 lines. We also thank all members of the Pallanck lab for critical reading of the manuscript. Finally, special thanks to Barry Ganetzky, in whose laboratory some of this work was performed. This work was supported by a University of Washington Royalty Research Fund Grant and a National Science Foundation CAREER award.

Manuscript received October 19, 2000; Accepted for publication January 26, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BATE, M., 1993 The mesoderm and its derivatives, pp. 1013–1090 in The Development of Drosophila melanogaster, edited by M. BATE and A. M. ARIAS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

BOULIANNE, G. L. and W. S. TRIMBLE, 1995  Identification of a second homolog of N-ethylmaleimide-sensitive fusion protein that is expressed in the nervous system and secretory tissues of Drosophila. Proc. Natl. Acad. Sci. USA 92:7095-7099[Abstract/Free Full Text].

BRAND, A. H. and N. PERRIMON, 1993  Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118:401-415[Abstract].

BUDNIK, V., Y. H. KOH, B. GUAN, B. HARTMANN, and C. HOUGH et al., 1996  Regulation of synapse structure and function by the Drosophila tumor suppressor gene dlg. Neuron 17:627-640[Medline].

BURGESS, R. W., D. L. DEITCHER, and T. L. SCHWARZ, 1997  The synaptic protein syntaxin1 is required for cellularization of Drosophila embryos. J. Cell Biol. 138:861-875[Abstract/Free Full Text].

CHOU, T. B. and N. PERRIMON, 1992  Use of a yeast site-specific recombinase to produce female germline chimeras in Drosophila. Genetics 131:643-653[Abstract].

CHOU, T. B. and N. PERRIMON, 1996  The autosomal FLP-DFS technique for generating germline mosaics in Drosophila melanogaster. Genetics 144:1673-1679[Abstract].

CHOU, T. B., E. NOLL, and N. PERRIMON, 1993  Autosomal P[ovoD1] dominant female-sterile insertions in Drosophila and their use in generating germ-line chimeras. Development 119:1359-1369[Abstract].

DAVIS, G. W., A. DIANTONIO, S. A. PETERSEN, and C. S. GOODMAN, 1998  Postsynaptic PKA controls quantal size and reveals a retrograde signal that regulates presynaptic transmitter release in Drosophila. Neuron 20:305-315[Medline].

DIANTONIO, A., S. A. PETERSEN, M. HECKMANN, and C. S. GOODMAN, 1999  Glutamate receptor expression regulates quantal size and quantal content at the Drosophila neuromuscular junction. J. Neurosci. 19:3023-3032[Abstract/Free Full Text].

GRAMATES, L. S. and V. BUDNIK, 1999  Assembly and maturation of the Drosophila larval neuromuscular junction. Int. Rev. Neurobiol. 43:93-117[Medline].

HARRISON, S. D., K. BROADIE, J. VAN DE GOOR, and G. M. RUBIN, 1994  Mutations in the Drosophila Rop gene suggest a function in general secretion and synaptic transmission. Neuron 13:555-566[Medline].

JAHN, R. and T. C. SUDHOF, 1999  Membrane fusion and exocytosis. Annu. Rev. Biochem. 68:863-911[Medline].

JAN, L. Y. and Y. N. JAN, 1976  L-glutamate as an excitatory transmitter at the Drosophila larval neuromuscular junction. J. Physiol. 262:215-236[Abstract/Free Full Text].

KAWASAKI, F. and R. W. ORDWAY, 1999  The Drosophila NSF protein, dNSF1, plays a similar role at neuromuscular and some central synapses. J. Neurophysiol. 82:123-130[Abstract/Free Full Text].

KAWASAKI, F., A. M. MATTIUZ, and R. W. ORDWAY, 1998  Synaptic physiology and ultrastructure in comatose mutants define an in vivo role for NSF in neurotransmitter release. J. Neurosci. 18:10241-10249[Abstract/Free Full Text].

KLENCHIN, V. A. and T. F. MARTIN, 2000  Priming in exocytosis: attaining fusion-competence after vesicle docking. Biochimie 82:399-407[Medline].

LIN, R. C. and R. H. SCHELLER, 2000  Mechanisms of synaptic vesicle exocytosis. Annu. Rev. Cell Dev. Biol. 16:19-49[Medline].

LITTLETON, J. T., E. R. CHAPMAN, R. KREBER, M. B. GARMENT, and S. D. CARLSON et al., 1998  Temperature-sensitive paralytic mutations demonstrate that synaptic exocytosis requires SNARE complex assembly and disassembly. Neuron 21:401-413[Medline].

LONART, G. and T. C. SUDHOF, 2000  Assembly of SNARE core complexes prior to neurotransmitter release sets the readily releasable pool of synaptic vesicles. J. Biol. Chem. 275:27703-27707[Abstract/Free Full Text].

LUSCHER, C., H. XIA, E. C. BEATTIE, R. C. CARROLL, and M. VON ZASTROW et al., 1999  Role of AMPA receptor cycling in synaptic transmission and plasticity. Neuron 24:649-658[Medline].

LUTHI, A., R. CHITTAJALLU, F. DUPRAT, M. J. PALMER, and T. A. BENKE et al., 1999  Hippocampal LTD expression involves a pool of AMPARs regulated by the NSF-GluR2 interaction. Neuron 24:389-399[Medline].

MULLER, J. M., C. RABOUILLE, R. NEWMAN, J. SHORTER, and P. FREEMONT et al., 1999  An NSF function distinct from ATPase-dependent SNARE disassembly is essential for Golgi membrane fusion. Nat. Cell Biol. 1:335-340[Medline].

NISHIMUNE, A., J. T. ISAAC, E. MOLNAR, J. NOEL, and S. R. NASH et al., 1998  NSF binding to GluR2 regulates synaptic transmission. Neuron 21:87-97[Medline].

NOEL, J., G. S. RALPH, L. PICKARD, J. WILLIAMS, and E. MOLNAR et al., 1999  Surface expression of AMPA receptors in hippocampal neurons is regulated by an NSF-dependent mechanism. Neuron 23:365-376[Medline].

ORDWAY, R. W., L. PALLANCK, and B. GANETZKY, 1994  Neurally expressed Drosophila genes encoding homologs of the NSF and SNAP secretory proteins. Proc. Natl. Acad. Sci. USA 91:5715-5719[Abstract/Free Full Text].

OSTEN, P., S. SRIVASTAVA, G. J. INMAN, F. S. VILIM, and L. KHATRI et al., 1998  The AMPA receptor GluR2 C terminus can mediate a reversible, ATP-dependent interaction with NSF and alpha- and beta-SNAPs. Neuron 21:99-110[Medline].

PALLANCK, L., R. W. ORDWAY, and B. G