Abstract
Pre-mRNA alternative splicing is an important mechanism for the generation of synaptic protein diversity, but few factors governing this process have been identified. From a screen for Drosophila mutants with aberrant synaptic development, we identified beag, a mutant with fewer synaptic boutons and decreased neurotransmitter release. Beag encodes a spliceosomal protein similar to splicing factors in humans and Caenorhabditis elegans. We find that both beag mutants and mutants of an interacting gene dsmu1 have changes in the synaptic levels of specific splice isoforms of Fasciclin II (FasII), the Drosophila ortholog of neural cell adhesion molecule. We show that restoration of one splice isoform of FasII can rescue synaptic morphology in beag mutants while expression of other isoforms cannot. We further demonstrate that this FasII isoform has unique functions in synaptic development independent of transsynaptic adhesion. beag and dsmu1 mutants demonstrate an essential role for these previously uncharacterized splicing factors in the regulation of synapse development and function.
Introduction
Synapses are exquisitely specialized cell–cell contacts characterized by dense multiprotein complexes surrounding the tightly apposed presynaptic and postsynaptic membranes. Bridging this cleft are synaptic adhesion molecules (SAMs), which provide adhesive structural stability and act as transsynaptic signaling proteins (Dalva et al., 2007). Many SAM genes have extensive alternative splice isoform diversity, with some such as neurexins capable of producing hundreds of different protein splice isoforms (Ullrich et al., 1995).
The neural cell adhesion molecule (NCAM) is an extensively studied homophilic synaptic adhesion molecule (Hartz and Ronn, 2010; Kristiansen and Hortsch, 2010; Muller et al., 2010) that can modulate signaling through its intracellular domain (Ditlevsen and Kolkova, 2010). Fasciclin II (FasII), the Drosophila ortholog of NCAM, is essential for normal synaptic development and growth (Packard et al., 2003; Kristiansen and Hortsch, 2010). In fasII-null mutants initial neuromuscular junction (NMJ) synaptic contacts form but later retract (Schuster et al., 1996a), while in partial loss-of-function mutants, synaptic terminal size is reduced (Stewart et al., 1996). Regulation of synaptic FasII and NCAM is also important for synaptic plasticity (Muller et al., 1996; Schuster et al., 1996b; Cremer et al., 1997; Koh et al., 1999; Sigrist et al., 2003).
Both NCAM and FasII pre-mRNAs undergo alternative splicing to generate multiple protein isoforms (Cunningham et al., 1987). NCAM has three major splice isoforms, NCAM 120, 140, and 180, which have identical extracellular domains. NCAM 120 is attached to the membrane via a glycophosphatidylinositol (GPI) linkage, while NCAM 140 and 180 are transmembrane proteins. The expression pattern and intracellular binding partners differ between NCAM isoforms (Pollerberg et al., 1985, 1986; Persohn et al., 1989). Similarly, Drosophila FasII undergoes alternative splicing to create two transmembrane isoforms and two additional isoforms, one of which may be GPI-linked (Grenningloh et al., 1991; Lin et al., 1994). The distinct functional requirement for these isoforms in synapse development is unclear. Similar to most SAM genes, the factors that regulate alternative splicing of NCAM and FasII are unknown. Proteomic analysis of spliceosomes in vertebrates and Drosophila has identified many proteins with the potential to influence splicing; however, few splicing factors that regulate neuronal protein diversity have been characterized (Neubauer et al., 1998; Zhou et al., 2002; Herold et al., 2009).
Here we identify novel mutants in two genes, beag and dsmu1, that regulate Drosophila synaptic morphology and function. The proteins encoded by these genes are components of the Drosophila spliceosome (Herold et al., 2009) and are related to human and Caenorhabditis elegans spliceosomal proteins (Neubauer et al., 1998; Assier et al., 1999). We show that the aberrant synaptic morphology of beag mutants is due to a change in the alternative splice isoform distribution of FasII. We demonstrate that only transmembrane isoforms of FasII are required for normal synapse development and show that restoration of one specific transmembrane isoform can rescue beag mutant synaptic morphological defects. Our data reveal that Beag and Dsmu1 are essential alternative splicing factors required for normal synaptic development and the regulation of FasII.
Materials and Methods
Drosophila stocks
The following Gal4 lines were used: OK6-Gal4, G14-Gal4 (Aberle et al., 2002), C155-Gal4 (Lin and Goodman, 1994), OK371-Gal4 (Mahr and Aberle, 2006), OK319-Gal4 (B.J. Choi and B.D. McCabe, unpublished observations), D42-Gal4 (Brand and Perrimon, 1993), and Da-Gal4 (Wodarz et al., 1995). We use the following nomenclature: FasII-A-PEST+ (Lin and Goodman, 1994; FasII-PA Flybase), FasII-TM-A-PEST− (Lin and Goodman, 1994), FasII-C (FasII-PC Flybase; Grenningloh et al., 1991), and FasII-B (FasII-RB Flybase). The following FasII lines were used: UAS-FasII-A-PEST+ (Schuster et al., 1996a), UAS-FasII-A-PEST− (Lin and Goodman, 1994), UAS-FasII-C (Lin and Goodman, 1994; D. Lin and C.S. Goodman, unpublished observations), UAS-FasII-A-PEST+AAE (Zito et al., 1997; K. Zito and C.S. Goodman, unpublished observations), UAS-CD8-FasII-A-PEST+Intra (Zito et al., 1997), fasIIe76, fasIIeB112 (Grenningloh et al., 1991), Df(1)BSC869 (K. Cook, S, Christensen, personal communication to Flybase), UAS-FasIIRNAi-total (Dietzl et al., 2007; Vienna Drosophila RNAi Centre, #v103807),UAS-FasIIRNAi-A, UAS-RNAiResistantFasII-A-PEST+, and UAS-RNAiResistantFasII-A-PEST− (construction described below). To confirm the sequence of the UAS-FasII-C transgene, genomic DNA was isolated from the transgenic flies and primers specific to the pUAST vector (UAST-FWD GAGCGCCGGAGTATAAATAGAGG and UAST-REV CTCCCATTCATCAGTTCCATAGGT) were used to PCR-amplify the transgene. The sequence of this PCR product corresponded to the entire coding sequence of FasII-C as in Flybase, with the exception of several single base pair substitutions that do not alter the amino acid sequence. The following beag lines were used: Df(3R)Exel6151 (Thibault et al., 2004), P[EP]CG18005EP3260 (Rørth, 1996), beag1, UAS-EYFPBeag, UAS-Beag, and Genomic beag (construction described below). The following dsmu1 lines were used: Df(3R)Exel6182 (Thibault et al., 2004), Pbac[WH]CG5451f03090 (Thibault et al., 2004), UAS-myc-tagRFP2-Dsmu1,and UAS-Dsmu1 (construction described below). Animals of either sex were analyzed for mutants or constructs on autosomes. For hemizygous or heterozygous mutants on the X chromosome, males or females were selected as appropriate. The number of animals analyzed per genotype is detailed in Table 1.
Construction of transgenes
UAS-EYFPBeag, UAS-Beag.
The Beag open reading frame (ORF) was amplified from the LD21347 cDNA clone (Stapleton et al., 2002; Berkeley Drosophila Genome Project, BDGP) by PCR and subcloned into a pUAST vector with or without a sequence encoding EYFP (B,D.M., unpublished observations). Transgenic flies with insertions on chromosome II were generated by standard P-element techniques.
Genomic beag.
The beag genomic rescue construct was generated by PCR amplification of the beag gene from genomic DNA using a forward primer 1 kb upstream of the transcription start site and a reverse primer 0.5 kb downstream of the 3′UTR (primers beag-GR-F CACCAGACCGAAAGTTTCCGCAGCA and beag-GR-R TTTGGATCCCCCGCGAAGGTAATTACATTT). A 3× Flag tag was inserted immediately upstream of the beag start codon, and a stop codon was inserted at the 40th base pair of the ORF of the upstream gene ada to avoid ada overexpression artifacts. This sequence was subcloned into the pBID vector (J. Wang, E.S. Beck, and B.D. McCabe, unpublished observations). Transgenic flies with insertions on chromosome II at the attP51D landing site were generated using Phi3C1 transgenesis (Groth et al., 2004).
UAS-myc-tagRFP-Dsmu1, UAS-Dsmu1.
The Dsmu1 ORF was amplified from the LD41216 cDNA clone ((Rubin et al., 2000; BDGP) by PCR (primers Smu1-F CACCATGTCCATAGAAATCGAATCA and Smu1-R AAACTCGAGCCACCAACTAAAACTAACTAG), TOPO cloned into pENTR, and then Gateway cloned into pBID-UASC-GW and pBID-UASC-MRG-GW (J. Wang, E.S. Beck, and B.D. McCabe, unpublished observations). Transgenic flies with insertions on chromosome II at the attP40 landing site were generated using Phi3C1 transgenesis (Groth et al., 2004).
UAS-FasIIRNAi-A.
UAS-FasIIRNAi-A was generated by PCR amplification of the sequence corresponding to FasII exon 12 from cDNA made from larval brains (primers FasII-RNAi-A-F, CACCCGTCATCCAAGTGGCTGAGCG and FasII-RNAi-A-R, GCTTGGCCTCGTCGTCGATTT). The PCR product was TOPO cloned into pENTR and then Gateway cloned into pBID-UAS-GGi (J. Wang, E.S. Beck, and B.D. McCabe, unpublished observations) such that each product was inserted twice, in opposite directions, immediately downstream of a UAS sequence and with a ftz intron sequence in between the two inserts, allowing the formation of a hairpin RNA when transcribed. Transgenic flies with insertions in the attP2 on chromosome III site were generated using Phi3C1 transgenesis (Groth et al., 2004).
UAS-RNAiResistantFasII-A-PEST+, UAS-RNAiResistantFasII-A-PEST−.
A modified version of the final 101 bp of the FasII-A-PEST+ and A-PEST− ORF were synthesized and inserted in the pUC57 vector (GenScript) such that the last base of every codon was altered to maintain the amino acid sequence while inhibiting targeting by FasIIRNAi-A.
The sequence of this synthesized fragment was as follows: GGTACCCCCTTAAGGACGCCTACGGGAAGTATAAAGCAAAATTCCACGATAGAGTTTGATGGCCGTTTTGTACATTCCCGGAGCGGAGAAATTATAGGCAAAAACTCTGCAGTCTAATCTAGA.
The rest of the FasII-A-PEST+ and FasII-A-PEST− coding sequences were amplified individually from cDNA made from larval brains (primers FasIIRNAiRes-F CACCGAATTCAACATGGGTGAATTGCCGCCAAAT, FasIIRNAiRes-F-PEST+R AAGCTTCTAGAGCGTCCTTAAGGGCTCCTTTTCGTCCAATGGCGTGC, and FasIIRNAiRes-F-PEST−R AAGCTTCTAGAGCGTCCTTAAGGGCTCCTTTTCGTCCCTGCCCAGC) and cloned into the pCMV-tag5a vector using EcoRI and HindIII restriction sites included in the primers. The mutated 3′ sequence was subcloned into a plasmid containing the rest of the coding sequence using engineered AflII and XbaI restriction sites, and then the entire FasII coding sequence was subcloned into the pBID-UASC vector using introduced EcoRI and HindIII restriction sites (J. Wang, E.S. Beck, and B.D. McCabe, unpublished observations). Transgenic flies with insertions on chromosome II at the attp40 landing site were generated using Phi3C1 transgenesis (Groth et al., 2004).
Immunohistochemistry
Wandering third instar larvae were collected, dissected, and stained as previously described (Brent et al., 2009a,b). Primary antibodies used were mouse anti-Bruchpilot (Wagh et al., 2006)(1:500, NC82; Developmental Studies Hybridoma Bank, DSHB), mouse anti-cysteine string protein (CSP; Zinsmaier et al., 1990)(1:200; DSHB), rabbit anti-DAP160 (1:100; Marie et al., 2004), mouse anti-DLG (Parnas et al., 2001)(1:200; from C.S. Goodman), rat anti-elav (1:500; DSHB), mouse anti-FasII TM (mAb 1D4, 1:900; DSHB, from C.S. Goodman), mouse anti-FasII total (mAb 34B3, 1:20; DSHB), mouse anti-Flag (1:500; Sigma), chicken anti-GFP (1:1000; Abcam), mouse anti-Map1B (1:100; Hummel et al., 2000), mouse anti-myc (1:400; DSHB), mouse anti-neuroglian (Hortsch et al., 1990)(1:100; DSHB), guinea pig anti-phospho-Mad (1:1000; gift from E. Laufer), mouse anti-synaptotagmin (1:1000, mAb 3H2 2D7; DSHB), and Cy5-conjugated goat anti-horseradish peroxidase (HRP) (1:400; Jackson ImmunoResearch). Conjugated secondary antibodies were used (1:1000, Jackson ImmunoLabs; 1:2000, Invitrogen). Larval preparations were imaged on a Zeiss LSM 510 Confocal microscope. Relative immunohistochemistry intensity measurements were made using these images. The average intensity of staining of the antibody of interest was calculated within the synaptic area colabeled by anti-HRP, which labels neuronal membranes. Synaptic boutons at muscle 4 of segment A3 were used for intensity measurements and calculations were done using MetaMorph software (Molecular Devices).
Morphological analysis of NMJs
Wandering third instar larvae were stained with antibodies against CSP and HRP (described above). All morphological analysis was done at muscle 4 of segment A3. Type Ib and Is boutons were counted using a 40× objective on a Zeiss Axio Imager.Z1 microscope. Muscle area was measured using a 20× objective micrometer. Raw bouton and muscle surface area for all genotypes analyzed are in Table 1. Synaptic area measurements were made with MetaMorph software on images obtained by confocal microscopy. Total presynaptic area was determined by calculating the area of all type Ib boutons on a single muscle 4, measured by the area stained by CSP. Total area was divided by bouton number to determine average bouton area. At least 35 synapses were analyzed per genotype.
Time-lapse live imaging
NMJs were visualized using MHC::hCD8-GFP-Shaker protein in control and beag mutant larvae (Zito et al., 1999). Animals were anesthetized by ∼15 min exposure to a vapor mixture of 35% methyl salicylate and 16% menthol (Haw Par Healthcare) at the second instar and again 48 h later at the third instar. For imaging, larvae were placed on a slide with 70% glycerol and a coverslip. Imaging time was limited to <30 min after which animals were washed gently with PBS, allowed to recover, and returned to the food media. Mutants and corresponding controls were imaged on the same day in a random order to minimize handling variability. Only images from animals that survived the entire 2 d imaging procedure were included in analysis. Images were collected using a Zeiss Z1 Apotome system using a 63× lens. To count bouton addition in live images only new distal boutons, which can be reliably recognized by live imaging, were used for analysis.
Active zone quantification
To determine the number of active zones per NMJ, wild-type and beag1/Df larvae were dissected and stained as described above with anti-Bruchpilot. The number of discrete anti-Bruchpilot-stained active zones per NMJ were counted using an E600 Nikon epifluorescence microscope with a Plan Apo × 100/1.4 NA objective. Counting was done for the NMJ synaptic terminal on muscle 4 of segment A3.
Electrophysiology
Intracellular recording from muscle 6, segment A3 was performed as previously described (Imlach and McCabe, 2009). Third instar larvae were dissected in HL3 (Stewart et al., 1994) and recordings were performed in HL3 containing 1 mm Ca2+ (except where noted). Data were only analyzed when the resting membrane potential was <−55 mV. Evoked junctional potential (EJP) amplitude was analyzed using Clampex v 8.2.0.235 software (Molecular Devices). Miniature excitatory junctional potential (mEJP) amplitude and frequency were analyzed using Mini analysis software (Synaptosoft, v 6.0.3). Recordings were done from a minimum of 10 muscles per genotype.
In situ hybridization
In situ hybridization of Drosophila embryos was performed as previously described (Kosman et al., 2004), with the following modifications. After incubation in anti-Dig-AP Fab fragments antibody (1:500; Roche), in situs were developed for 10–30 min using the BCIP/NBT Alkaline Phosphatase Substrate Kit IV (Vector Laboratories). Antisense and sense beag RNA probes were transcribed in vitro from PCR products amplified from the LD21347 cDNA clone (Stapleton et al., 2002; BDGP). The entire beag ORF was amplified using the 5′ primer TAATACGACTCACTATAGGGAGAACGCGCTACAATTAACATAAC and the 3′ primer GCAGATCTGATATCATCGCCACT for the antisense probe and the 5′ primer ACGCGGCTACAATTAATACATAACC and the 3′ primer AGCCGATTCATTAATGCAGGT for the sense (negative control) probe.
Quantitative real-time PCR
RNA was isolated from third instar larval brains using TRIzol and isopropanol precipitation. RNA samples were DNase treated (TURBO DNA-free Kit; Applied Biosystems) and reverse transcribed (SuperScript III First-Strand Synthesis System for RT-PCR; Invitrogen). Quantitative PCR was performed using an Eppendorf Mastercycler ep realplex Thermal Cycler.
The Primer 3 web site (http://fokker.wi.mit.edu/primer3/input.htm) was used to design primers for qPCR that spanned exon–exon junctions to avoid amplification of genomic DNA. The following primers pairs were used:
RP49: F-CATCCGCCCAGCATACAG, R-CCATTTGTGCGACAGCTTAG.
FasII-A: F-TACTGTCCGGGCGTTAAGAT, R-ACGTCAATTCCTCGTGTCGT, FasII-C: F-TACTGTCCGGGCGTTAAGAT, R-GAATCGGACTCACCTCGTGT, FasII total: F-CAACCAGGTGGGATTAGGAA, R-TAACGCCCGGACAGTATTTG.
qRT- PCRs was performed at least three times for each of at least three biological replicates per genotype. ddCt was calculated for each FasII primer set, using rp49 as a housekeeping gene for normalization.
Semiquantitative PCR
cDNA isolated from larval brains of wild-type and beag mutant larvae was amplified with a 5′ primer in the last common FasII exon and a 3′ primer in the shared 3′ UTR of FasII-A-PEST+ and FasII-A-PEST−. This PCR yielded a slower migrating band corresponding to FasII-A-PEST+ and a faster migrating band corresponding to FasII-A-PEST− by agarose gel electrophoresis. Gels were photographed and the intensity of the FasII-A-PEST+ and PEST− bands and the background intensity were measured from the same image using ImageJ. The background intensity was subtracted from the band intensities and the ratio of the intensity of the two bands was calculated. Four biological replicates of wild-type and beag cDNA were used and for each set the PCR was done one to three times, leading to a final N of 7.
Western blotting
Samples for Western blots were prepared by putting 5–10 whole larvae, dissected larval brains, or dissected larval body walls directly into sample buffer.
The primary antibodies used were as follows: mouse anti-elav (1:500, mAb 9F8A9; DSHB), mouse anti-FasII TM (mAb 1D4, 1:900; DSHB, from C.S. Goodman), and mouse anti-FasII total (mAb 34B3, 1:20; DSHB). Peroxidase-conjugated goat anti-rabbit (1:5000; Jackson ImmunoLabs) was used as a secondary antibody.
Statistical analysis
For all comparisons, statistical significance was calculated using ANOVA.
Results
Mutations in the gene beag lead to a decrease in NMJ synaptic bouton number and increased bouton size
During larval development, Drosophila NMJs grow by the iterative addition of synaptic boutons (Zito et al., 1999; Prokop, 2006). On average these synaptic boutons have a diameter of ∼3 μm and bouton diameters rarely exceed 6 μm (Johansen et al., 1989). When normalized to muscle surface area, the number of synaptic boutons at each NMJ terminal is also highly stereotyped (Schuster et al., 1996a). From a forward genetic screen for mutants with aberrant synapse development (McCabe et al., 2004), we identified a novel mutant beag (pronounced “be-yug,” Gaelic for small or petite). beag mutants were adult semilethal with 39% of the expected number of animals surviving to adulthood. beag mutant larvae had decreased numbers of NMJ synaptic boutons compared with wild type, but a significant increase in the area of the boutons present (Fig. 1A,B). To characterize beag mutant NMJs, we used labeling with an antibody against the synaptic vesicle protein CSP (Zinsmaier et al., 1994) to count the number of synaptic boutons as well as measure the total presynaptic bouton area. We divided the total presynaptic bouton area by the number of synaptic boutons to measure the average bouton area. beag mutants had a 39% decrease (p < 0.001) in the number of synaptic boutons while the average bouton area was dramatically increased to 180% of wild type (p < 0.001) (Fig. 1C,D, Table 1). Thus the total presynaptic area was increased by 21% (p < 0.01) compared with wild type (Fig. 1E). This result suggested to us that the process of bouton addition might be defective beag mutants. To directly examine this process, we imaged NMJ terminals by time-lapse live imaging through the transparent cuticle of whole animals in wild type and beag strains that expressed CD8-GFP-Sh, a marker of postsynaptic NMJ membranes (Zito et al., 1999). We anesthetized animals at the second larval instar, imaged their NMJ terminals, returned them to the food media, and then imaged the terminal again 48 h later at the third larval instar. To measure new bouton formation, we restricted our analysis to the addition of distal boutons, which can be reliably detected by live imaging. After 48 h, 51.4% of wild-type NMJ terminals had new boutons added to the distal end of NMJ terminals (Fig. 1F). In contrast, only 6.7% (p < 0.001) of beag mutant NMJs had new distal boutons added after 48 h (Fig. 1G). These results indicated that beag mutants had a failure of new bouton addition coupled with aberrant morphological expansion of existing boutons.
The beag point mutation (beag1) was mapped by complementation to the deficiency Df(3R)Exel6151, covering the cytological region 85C3–85C11 (Thibault et al., 2004). Sequencing of genes covered by this deficiency led to the identification of a point mutation in the previously uncharacterized gene CG18005, which we renamed beag. This mutation introduces a premature stop codon at amino acid 304 of the predicted 557 aa protein. We also identified a P element, EP3260, inserted at the 94th nucleotide of the beag ORF as an additional beag allele (Rørth, 1996). beag1/Df(3R)Exel6151 and beag1/beagEP3260 mutants recapitulated the synaptic structural phenotypes of beag1 homozygotes (Fig. 1C), suggesting that the beag1 mutation is a strong loss-of-function allele of CG18005. For all subsequent experiments, except where noted, we used the beag1/Df(3R)Exel6151 mutant combination.
Beag is a broadly expressed nuclear protein
In situ hybridization of embryos with RNA probes against the beag transcript showed broad expression of beag message, including expression in the CNS (data not shown). We generated an epitope-tagged beag transgene “genomic Beag” that includes the entire beag gene, with 1 kb of DNA upstream of the transcription start site and 0.5 kb of DNA downstream of the beag 3′UTR. This construct, which rescued beag mutant synapse morphology defects and beag mutant viability (Fig. 1C–E, data not shown), revealed that Beag protein is expressed in many tissues in the larva. Beag protein was localized to the nuclei of neurons, where it colocalized with the neural nuclear protein Elav (Fig. 1H,I). The nuclear localization of Beag was confirmed at higher resolution with an EYFP-tagged, Gal4-driven beag cDNA transgene (data not shown). Genomic Beag expression was also observed in the nuclei of non-neural cells, including muscles (data not shown).
Beag encodes a spliceosomal protein
Sequence analysis of the Beag protein revealed that it was most similar to RED protein in humans (Assier et al., 1999) and SMU-2 protein in C. elegans (Spartz et al., 2004). RED is a ubiquitously expressed nuclear protein that was purified from human spliceosomes. Recently, Beag was also found to be a constituent of Drosophila spliceosomes (Herold et al., 2009). The C. elegans homolog of Beag, SMU-2, which is also expressed in the nuclei of all cells (Spartz et al., 2004), was first identified as a regulator of unc-52 alternative splicing. smu-2 loss-of-function mutants can suppress lethal point mutations in unc-52 by altering the alternative splicing of an UNC-52 exon. Alone, mutants of smu-2 have mild phenotypes including decreased mobility, growth, and brood size; however, no neuronal phenotype has been reported (Lundquist and Herman, 1994; Spartz et al., 2004). Like RED and SMU-2, Beag does not have a recognizable RNA binding motif. These proteins may influence alternative splicing through interaction with other RNA binding proteins (Black, 2003). The presence of Beag in the Drosophila spliceosome and its homology to SMU-2, which has a demonstrated role in alternative splicing, suggested that Beag might regulate synaptic development through the regulation of pre-mRNA splicing.
Dsmu1 and Beag have similar synaptic phenotypes and function in the same genetic pathway
In the same study in which C. elegans smu-2 was identified, a second splicing regulator, smu-1, was also isolated. smu-1 and smu-2 mutants have similar phenotypes, both alter unc-52 alternative splicing, and SMU-1 and SMU-2 proteins physically interact (Lundquist and Herman, 1994; Spartz et al., 2004). Like SMU-2, SMU-1 protein is located in the nucleus and is ubiquitously expressed (Spike et al., 2001). The human protein most similar to SMU-1 is known as fSAP-57 or SMU1 and is enriched in the brain (Di Benedetto et al., 2001; Spike et al., 2001). This protein was purified from human spliceosomes (Zhou et al., 2002) and has also been shown to regulate alternative splicing (Sugaya et al., 2006). We hypothesized that Beag might regulate synaptic development as part of a conserved complex with SMU1, and we sought to identify the Drosophila homolog of this protein. We examined the Drosophila genome and determined that the protein most similar to C. elegans SMU-1 and human SMU1 was encoded by the previously uncharacterized gene CG5451. Similar to Beag, the product of this gene is also a component of Drosophila spliceosomes (Herold et al., 2009) and we renamed CG5451 dsmu1.
We identified a piggyBac transposon, pBacf03090 (Thibault et al., 2004), inserted between the first two protein-coding exons in the dsmu1 gene. dsmu1pBacf03090/Df(3R)Exel6182 mutants had a comparable synaptic phenotype to beag mutants (Fig. 2B,C). dsmu1 mutant NMJs had a 30% (p < 0.001) decrease in synaptic bouton number (Fig. 2E) and a 77% (p < 0.001) increase in average synaptic bouton area (Fig. 2F), leading to a 29% increase in total synaptic area (p < 0.05) (Fig. 2G). To determine the subcellular localization of Dsmu1, we generated an epitope-tagged, Gal4/UAS-driven Dsmu1 cDNA transgene. Expression of this transgene in a subset of motor neurons with OK319-Gal4 demonstrated that Dsmu1, like Beag, colocalized with Elav in the nucleus; however, we also observed some Dsmu1 protein in the cytoplasm (Fig. 2H,I). Mammalian SMU1 protein has also been found to localize both to the nucleus and cytoplasm (Di Benedetto et al., 2001).
To test if beag and dsmu1 function in the same genetic pathway, we first looked for a genetic interaction between beag and dsmu1 heterozygotes. beag, dsmu1 trans-heterozygotes had an 11% decrease (p < 0.05) in bouton number compared with dsmu1 heterozygotes and a 14% decrease (p < 0.05) compared with beag heterozygotes (Fig. 2E), suggesting the two genes function together. To test this further we generated beag, dsmu1 double mutants. The decrease in the synaptic bouton number of these mutants was not significantly different from that observed in single mutants of beag or dsmu1 (Fig. 2D,E).Together, these data demonstrate that both genes are required for normal synaptic development in Drosophila and suggest that, similar to their homologs in C. elegans, these two genes function together.
Beag and Dsmu1 function in neurons to regulate NMJ growth
To determine in which cells Beag and Dsmu1 are required for normal NMJ synapse development, we constructed transgenes encoding the beag or dsmu1 ORF under Gal4/UAS control (Brand and Perrimon, 1993). Restoration of Beag or Dsmu1 expression with the muscle-specific driver G14-Gal4 (Aberle et al., 2002) did not rescue the synaptic bouton number of the respective mutants (data not shown). In contrast, when driven with the motor neuron driver OK6-Gal4 (Aberle et al., 2002) (Fig. 3C,H) or the pan-neural driver C155-Gal4 (Lin and Goodman, 1994) (Fig. 3F,H), UAS-Beag and UAS-Dsmu1 fully rescued the decreased bouton number observed in beag and dsmu1 mutants, respectively. Neural expression of epitope-tagged UAS-Beag and UAS-Dsmu1 also fully rescued the decreased bouton number in beag and dsmu1 mutants, respectively (data not shown). Neural overexpression of Beag or Dsmu1 in a wild-type background had no effect on the synaptic bouton number or area (Fig. 3H). Expression of transgenic Beag in all neural cells with C155-Gal4 or solely in glutamatergic neurons (which include motor neurons) with OK371-Gal4 (Mahr and Aberle, 2006) also rescued beag mutant adult viability (data not shown). Therefore, both Beag and Dsmu1 are required in neural cells for normal NMJ development, and even though Beag is broadly expressed, it is predominantly required in neurons for adult viability.
We also used these transgenes to further probe the genetic interactions between beag and dsmu1. Neural overexpression of Dsmu1 with C155-Gal4 did not alter the bouton number of beag mutants (Fig. 3D,H). In contrast, neural overexpression of Beag resulted in a 28% (p < 0.01) increase in bouton number of dsmu1 mutants (Fig. 3G,H). These results confirm that beag and dsmu1 are components of a common genetic pathway and indicate that beag functions downstream of dsmu1.
Beag and Dsmu1 are required for normal neurotransmitter release at the NMJ
To determine the functional consequences of beag and dsmu1 mutations, we examined neurotransmitter release properties at the larval NMJ. Beag1/EP3260 mutants had a 47% decrease (p < 0.001) in EJP amplitude compared with wild type (Fig. 4A–C). Similarly, dsmu1 mutants had a 43% decrease (p < 0.001) in EJP amplitude compared with wild type (Fig. 4C). mEJP amplitude and frequency were unchanged in beag1/EP3260 mutants leading to a 34% (p < 0.001) decrease in quantal content compared with wild type (Fig. 4A,B,D–F). The decrease in quantal content observed in beag mutants suggested that the defect in EJP amplitude in these mutants was presynaptic in origin. This was confirmed when we fully rescued the reduction in EJP amplitude and quantal content in beag mutants by expressing UAS-Beag in motor neurons with OK6-Gal4 (Fig. 4C–F). To determine whether the decrease in EJP amplitude observed in beag mutants could be due to a reduced number of neurotransmitter release sites, we counted the number of active zones labeled by Bruchpilot (Kittel et al., 2006; Wagh et al., 2006) in wild-type and beag mutant terminals. We found that beag mutant NMJ terminals had no difference in the number of active zones labeled with NC82 compared with wild type (Fig. 4G–I). In addition, measurement of EJP amplitude in wild-type and beag mutant larvae at a range of extracellular calcium concentrations revealed that at high calcium concentrations the difference in EJP amplitude was exaggerated in beag mutants. This is consistent with a defect in calcium regulation of neurotransmitter release (data not shown). Our results therefore suggest that both Beag and Dsmu1 are required for normal neurotransmitter release as well as for morphological development of the NMJ.
Synaptic FasII expression is altered in an isoform-specific manner in beag mutants
We next sought to determine whether altered splicing of specific genes could explain the observed beag NMJ phenotypes. Both the C. elegans and mammalian homologs of Beag have been shown to regulate alternative splicing of homologs of perlecan (encoded by unc-52 in C. elegans and trol in Drosophila) (Lundquist and Herman, 1994; Park et al., 2003; Spartz et al., 2004; Sugaya et al., 2006). We investigated if Beag could regulate NMJ development and function via modulation of perlecan; however, neither immunohistochemical nor genetic data supported an interaction between perlecan and Beag in the regulation of NMJ synaptic development (data not shown). Therefore, to identify synaptic molecules that could be regulated by beag and dsmu1, we used antibodies against protein components of the presynaptic terminal to look for changes in expression level or localization in beag mutants. For the majority of antibodies we tested, including antibodies against CSP (Zinsmaier et al., 1990), Bruchpilot (Wagh et al., 2006), DAP160 (Marie et al., 2004), Dlg (Parnas et al., 2001), Map-1B (Fujita et al., 1982), Neuroglian (Hortsch et al., 1990), Phospho-Mad (McCabe et al., 2004), Synaptotagmin (Dubuque et al., 2001), VAP (Tsuda et al., 2008), and Wallenda (Collins et al., 2006), we did not observe any changes in expression level or localization. However, we did note a significant decrease in levels of synaptic FasII in beag mutants compared with wild type using the isoform-specific anti-FasII monoclonal antibody 1D4 (Fig. 5A,B,D) (Van Vactor et al., 1993).
The fasII gene generates four known isoforms, FasII-A-PEST+, FasII-A-PEST−, FasII-B, and FasII-C, generated by alternative splicing of several exons at the 3′ end of the fasII gene (Fig. 5A; see Materials and Methods for other names that have been used to refer to these isoforms) (Grenningloh et al., 1991; Lin and Goodman, 1994). All four isoforms are identical in the 737 aa at the N terminal, which forms the majority of the extracellular domain of the protein. FasII-A-PEST+ and FasII-A-PEST− encode transmembrane proteins that have identical extracellular and intracellular domains with the exception of 29 aa encoded by a single exon that is included in FasII-A-PEST+ but excluded in FasII-A-PEST−. The FasII-B isoform is poorly characterized. The FasII-C isoform has no cytoplasmic domain and has been suggested to attach to the membrane through a GPI linkage (Grenningloh et al., 1991). The functional differences between these four isoforms have not been systematically studied.
The anti-FasII 1D4 monoclonal antibody recognizes the intracellular domains of the FasII-A-PEST+ and FasII-A-PEST− isoforms, but does not recognize FasII-C and likely does not recognize FasII-B (Van Vactor et al., 1993; Lin and Goodman, 1994; Schuster et al., 1996a). To determine the levels of synaptic and axonal FasII we used confocal microscopy and MetaMorph software. We measured anti-FasII staining intensity within the area that was costained with anti-HRP antibody, a specific marker of insect neuronal membranes (Jan and Jan, 1982). Intensity of NMJ staining with 1D4 was reduced by 26% in beag mutants compared with wild type (p < 0.001) and the intensity of motor neuron axon staining was decreased by 12% (p < 0.05) (Fig. 5B,D). We did not, however, observe any change in the distribution of FasII within NMJ synaptic boutons in beag mutants (data not shown). The reduction in synaptic 1D4 staining in beag mutants was fully rescued by the Beag genomic transgene (Fig. 5D). Similar to beag mutants, the intensity of synaptic 1D4 staining was decreased in dsmu1 mutants by 42% compared with wild type (p < 0.001) and motor neuron axon staining was reduced by 13% (p < 0.01) (Fig. 5D). In contrast, when we used a different anti-FasII monoclonal antibody, 34B3, which recognizes the extracellular domain of FasII common to all four isoforms (Grenningloh et al., 1991), we observed no difference in synaptic staining intensity in beag or dsmu1 mutants compared with controls (Fig. 5C,D). We did, however, observe a 19% increase (p < 0.001) in 34B3 staining in the motor neuron axons of beag mutants (Fig. 5C). The alterations in FasII levels appear to be motor neuron specific, as we did not observe changes in FasII protein levels in other neurons in the CNS (data not shown). Together, our results reveal that levels of transmembrane FasII are reduced at the NMJ and in motor neuron axons in beag and dsmu1 mutants, whereas axonal levels of other isoforms of FasII are increased.
To determine whether the alteration in FasII protein isoform levels was paralleled by an isoform-specific change in FasII mRNA levels, we performed quantitative real-time PCR (qRT-PCR) from cDNA isolated from wild-type or beag larval brains. We found that the mRNA level of the combined FasII-A-PEST+ and FasII-A-PEST− isoforms was decreased in beag mutants by 32% compared with wild type (SE = 9) (Fig. 5E). In contrast, the FasII-C mRNA level was increased by 17% compared with wild type by qRT-PCR (SE = 12) (Fig. 5E). Total FasII mRNA levels were not changed in beag mutants compared with wild type (Fig. 5E). To determine whether the levels of individual FasII-A-PEST+ and PEST− isoforms were also altered in beag mutants we measured their relative levels by semiquantitative PCR (Fig. 5F,G). In wild-type animals we found that message for FasII-A-PEST+ was more abundant than message for FasII-A-PEST−. In beag mutants, we found that this ratio was altered with a relative decrease (−42%, p < 0.05) in the ratio of FasII-A-PEST+ to PEST− message. Therefore both immunohistochemistry and mRNA transcript analysis are consistent with a selective reduction of FasII-A-PEST+ in beag mutants.
fasII and beag function in the same genetic pathway to regulate NMJ growth
fasII loss-of-function mutants have previously been reported to have a decrease in bouton number similar to what we observe in beag mutants (Schuster et al., 1996a). To confirm this result and to determine whether fasII and beag function in the same genetic pathway, we examined the NMJ phenotype of three fasII alleles. fasIIe76 is a hypomorphic allele that produces ∼10% of the wild-type level of FasII protein (Grenningloh et al., 1991). fasIIeB112 is a protein-null allele (Grenningloh et al., 1991), and Df(1)BSC869 is a deficiency that fully removes the fasII locus in addition to other genes. fasIIe76/e76 (−50%, p < 0.001), fasIIe76/eB112 (−50%, p < 0.001), and fasIIe76/Df (−59%, p < 0.001) mutant combinations all had fewer synaptic boutons than controls (Fig. 6F,H). In contrast, heterozygotes of any of these alleles had no change in bouton number compared with controls (Fig. 6C,H), even though synaptic FasII levels were reduced (data not shown).
We then tested whether beag and fasII function in the same genetic pathway. Like fasII heterozygotes, beag heterozygotes had no decrease in bouton number compared with controls (Fig. 6B,H). However, animals heterozygous for both fasIIe76 and beag1 had a 42% decrease in synaptic bouton number (p < 0.001) compared with wild type (Fig. 6D,H). This decrease was not significantly different from the decrease in bouton number observed in beag1/Df larvae or in fasIIe76/e76 larvae (Fig. 6E,F,H). Furthermore, combining beag1/Df mutants with heterozygous or homozygous fasIIe76 did not further decrease bouton number compared with beag mutants alone (Fig. 6E–H). These data suggest a strong genetic interaction between fasII and beag.
The transmembrane isoforms of FasII are necessary and sufficient for normal NMJ growth
Our data suggested that the beag phenotype was linked to misregulation of FasII isoforms. We therefore set out to determine which isoforms of FasII are required for normal NMJ growth. To do this we used an RNAi transgene that targets all isoforms of FasII (FasIIRNAi-Total; Dietzl et al., 2007) and we also generated an RNAi transgene that targets only the transmembrane FasII isoforms (FasIIRNAi-A). Ubiquitous expression of either of these RNAi constructs with Da-G4 caused a large decrease in FasII protein levels both at the NMJ and in the brain (data not shown). Knockdown of all FasII isoforms with FasIIRNAi-Total resulted in a 32% (p < 0.001) decrease in synaptic bouton number compared with controls (Fig. 7B,K). Knockdown of the transmembrane isoforms of FasII with FasIIRNAi-A resulted in a 20% (p < 0.05) decrease in bouton number (Fig. 7C,K), demonstrating that the transmembrane isoforms of FasII are required for normal NMJ growth. To determine whether FasII-A-PEST+, FasII-A-PEST−, or both were required for normal NMJ growth, we generated RNAi-resistant FasII-A-PEST+ and PEST− transgenes. In these transgenes, the portion of the wild-type cDNA targeted by FasIIRNAi-A was replaced with a synthetic sequence in which codon usage was changed such that the amino acid sequence was maintained but the mRNA would not be degraded by FasIIRNAi-A. Ubiquitous expression of RNAi-resistant FasII-A-PEST+ or RNAi-resistant FasII-A-PEST− fully rescued the reduction in bouton number induced by FasIIRNAi-A expression (Fig. 7E,F,K). In contrast, expression of FasII-C did not lead to a significant increase in bouton number in larvae in which the transmembrane FasII isoforms were knocked down (Fig. 7D,K). These results indicate that transmembrane FasII is required for normal NMJ growth.
We next tested which isoforms of FasII were sufficient to rescue NMJ growth in fasII mutants. Using the 34B3 antibody, which detects all FasII isoforms, we determined that each transgene expressed comparable levels of protein (data not shown). We observed increased levels of FasII in neuronal soma, axons, and NMJ synapses when FasII-A-PEST+ and FasII-A-PEST− were overexpressed. In contrast, while overexpression of FasII-C led to a similar increase in neuronal soma FasII levels, there was a smaller increase in synaptic FasII levels at the NMJ compared with overexpression of FasII-A-PEST+ or FasII-A-PEST− (data not shown). Expression of FasII-C in motor neurons of fasIIe76/Df mutants with OK6-Gal4 did not rescue bouton number (Fig. 7H,K). In contrast, expression of FasII-A-PEST+ or FasII-A-PEST− fully rescued bouton number in fasII mutants to wild-type levels (Fig. 7I–K). Therefore, both RNAi and rescue experiments demonstrate that the transmembrane isoforms of FasII are essential for normal NMJ development, in contrast to the FasII-C isoform.
FasII-A-PEST+, but not other FasII isoforms, rescues beag NMJ morphological defects
Transmembrane isoforms of FasII are reduced in beag mutants and we have shown that these isoforms are essential for normal NMJ development. We therefore attempted to rescue beag mutants by expressing cDNA transgenes (which lack introns) encoding FasII-A-PEST+, FasII-A-PEST−, or FasII-C. Neural overexpression of FasII-A-PEST+ or FasII-C had no effect on NMJ synaptic bouton number (Fig. 8G). However, overexpression of FasII-A-PEST− did induce a significant increase in synaptic bouton number (Fig. 8G). We then tested the ability of these transgenes to rescue beag mutants. Neither FasII-A-PEST− nor FasII-C expression altered beag mutant NMJ morphology (Fig. 8C,D,G). In contrast, expression of FasII-A-PEST+ fully rescued beag mutant synaptic bouton number and average synaptic bouton area to wild-type levels (Fig. 8E,G,H). These rescue experiments demonstrate a difference between FasII-A-PEST+ and FasII-A-PEST− function at the NMJ. In addition, these results, combined with FasII protein and mRNA measurements, suggest that a specific reduction in the FasII-A-PEST+ isoform is sufficient to explain the synaptic morphological defects in beag mutants. We also found that neural expression of FasII-A-PEST+ partially rescued dsmu1 mutant synaptic bouton number by 15% (p < 0.05) providing evidence that the decrease in bouton number in dsmu1 mutants is also at least in part due to a reduction in FasII-A-PEST+ and further supporting a common role for Beag and Dsmu1 in the regulation of NMJ morphology through modulation of FasII.
While synaptic morphological development can be rescued by restoration of FasII-A-PEST+ in beag mutants, we wished to determine whether neurotransmitter release properties were also restored. We measured EJP amplitude in beag mutants expressing transgenic FasII-A-PEST+ and found that the defect in evoked neurotransmitter release was not rescued. In addition, beag mutant viability was not rescued by neural expression of FasII-A-PEST+ (data not shown). This result is consistent with previous studies showing that fasII can regulate synaptic structure without altering neurotransmitter release (Stewart et al., 1996).
The intracellular domain of FasII-A-PEST+ is sufficient to rescue beag mutant synaptic morphology
Our data suggest that the intracellular domain of FasII, which is included in FasII-A-PEST+ and FasII-A-PEST−, but not FasII-C, is important for FasII's synaptic function. We therefore examined if the extracellular domain of FasII was required to rescue beag mutants. To test this hypothesis we used a chimeric transgene in which the extracellular and transmembrane domains of FasII-A-PEST+ were replaced with the extracellular and transmembrane domains of the human T-lymphocyte protein CD8 (Littman et al., 1985) fused in frame to the entire FasII-A-PEST+ intracellular region (CD8-FasII-A-PEST+Intra) (Zito et al., 1997). When we drove expression of this construct in motor neurons, we saw a dramatic increase in synaptic FasII immunoreactivity using the 1D4 antibody and observed that CD8 was targeted to the synapse, demonstrating that the intracellular domain of FasII-A-PEST+ is sufficient for presynaptic localization (data not shown). In wild-type larvae, expression of CD8-FasII-A-PEST+Intra did not affect synaptic bouton number (Fig. 8G), similar to expression of full-length FasII-A-PEST+. We then expressed this construct in the beag mutants. Similar to full-length FasII-A-PEST+, CD8-FasII-A-PEST+Intra fully rescued the decrease in synaptic bouton number in beag mutants to wild-type levels (Fig. 8F,G). This result demonstrates a specific requirement for the intracellular domain of transmembrane FasII in the rescue of synaptic morphology in beag mutants.
To further dissect the contribution of individual segments of the FasII-A-PEST+ intracellular domain we expressed a version of FasII-A-PEST+ in which the last three amino acids, which encode a PDZ-interaction domain, were altered from S-A-V to A-A-E (FasII-A-PEST+AAE). To determine the localization of this FasII transgene, we expressed it in the motor neurons of fasIIe76/y mutants, which have no detectible synaptic or axonal FasII staining (data not shown). When FasII-A-PEST+AAE was expressed in these mutants, FasII staining was observed in both motor neuron axons and at the NMJ (data not shown). Thus, when overexpressed, the PDZ-interaction domain of FasII is not absolutely required for presynaptic localization of FasII. To determine whether the PDZ-interaction domain of FasII was required for rescue of beag mutant NMJ morphology, we then expressed FasII-A-PEST+AAE− in motor neurons of wild-type and beag mutants. Expression in wild-type animals led to an increase in bouton number, while expression in beag mutants restored bouton numbers to levels not significantly different from wild type (Fig. 8G). These results indicate that the PDZ-interaction domain is not essential for the rescue of synaptic morphology in beag mutants by FasII.
Discussion
Regulation of the splice isoform diversity of synaptic adhesion molecules is essential for differential synaptic localization and function (Lisé and El-Husseini, 2006; Dalva et al., 2007). We have identified Beag as an alternative splicing factor that regulates specific isoforms of FasII, the Drosophila ortholog of NCAM. We establish that Beag works together with Dsmu1 in a neuronal genetic pathway required for NMJ synaptic terminal growth and neurotransmitter release. We show genetic interactions between fasII and beag and find a specific reduction of both mRNA and protein levels of transmembrane isoforms of FasII in beag mutants and similar protein changes in dsmu1 mutants. Through analysis of the requirements for individual FasII splice isoforms for normal NMJ growth, we find that the transmembrane isoforms of FasII are both necessary and sufficient for this process. We show that restoration of one of these transmembrane isoforms, FasII-A-PEST+, can completely rescue the synaptic structural defects of beag mutants while other FasII isoforms cannot, consistent with a relative reduction of FasII-A-PEST+ transmembrane isoform mRNA levels in beag mutants. Finally, rescue of beag mutants by the FasII-A-PEST+ intracellular region alone reveals a function in synapse development for this isoform independent of transsynaptic adhesion. Our data establish that Beag and Dsmu1 govern synaptic morphological development through the regulation of the alternative splicing of FasII.
Beag and Dsmu1 function in the nervous system to regulate alternative splicing
Beag, like SMU-2 and RED (Assier et al., 1999; Spartz et al., 2004), is expressed in many tissues. However, our rescue data demonstrate that Beag is required only in motor neurons for normal adult Drosophila viability, NMJ morphology, and neurotransmitter release. Neuronal sensitivity to the loss of ubiquitously expressed proteins has been demonstrated for other RNA regulatory proteins, notably the RNA processing factor Survival of Motor Neurons, the gene disrupted in spinal muscular atrophy (Monani, 2005). Potentially, Beag and Dsmu1 could interact with neuronal-specific proteins to regulate synapse development and function. Unlike most characterized alternative splicing factors, Beag and Dsmu1 proteins lack canonical RNA binding domains (although Dsmu1 does contain a WD domain which in Gemin 5 can bind RNA) (Lau et al., 2009) limiting biochemical techniques to identify pre-mRNA targets (Ule et al., 2005). It therefore seems likely that Beag and Dsmu1 exert their effects on pre-mRNA splicing through interactions with other RNA binding proteins similar to mechanisms described for splicing regulators such as Drosophila transformer and crooked neck (Tian and Maniatis, 1993; Edenfeld et al., 2006). Genetic model organisms are particularly useful for the identification and characterization of these types of factors. While we show that a change in FasII isoform distribution can explain the synaptic morphology defects in beag mutants, the neurotransmitter release and animal viability phenotypes of these mutants are not rescued by FasII expression. We hypothesize therefore that similar to other regulators of pre-mRNA splicing, Beag and Dsmu1 alter the alternative splicing of multiple neuronal genes. It is also seems likely given their broad expression pattern that they may regulate some splicing events in non-neuronal tissues.
Regulation of FasII alternative splicing by Beag
Our data suggest that in beag mutants FasII alternative splicing is altered, resulting in a decrease in the level of the transmembrane FasII isoforms coupled with a change in the relative level of FasII-A-PEST+ mRNA compared with FasII-A-PEST- mRNA. Splicing of Drosophila FasII transcripts generates at least four protein isoforms with identical extracellular domains but different forms of membrane attachment and intracellular domains (Grenningloh et al., 1991). This splicing pattern is conserved in both invertebrate and mammalian FasII homologs. In Manduca sexta, GPI-linked and transmembrane FasII isoforms are expressed by different cells and may have unique roles in cell migration, neurite outgrowth, and synapse formation (Wright et al., 1999, 2001; Knittel et al., 2001). In Aplysia, apCAM isoform expression also varies by cell type, and presynaptic and postsynaptic compartments can contain different apCAM isoforms (Mayford et al., 1992; Schacher et al., 2000). In mammals, the NCAM transmembrane isoform with the longest cytoplasmic domain, NCAM 180, is upregulated over the course of development and predominantly localizes to mature neuronal synapses, while the transmembrane isoform with a shorter cytoplasmic domain, NCAM 140, is localized to growing axons and glia in addition to synapses (Cunningham et al., 1987; Barbas et al., 1988). In contrast, the GPI-linked isoform, NCAM 120, is expressed mainly in glia (Pollerberg et al., 1985, 1986; Persohn et al., 1989). Depolarization of cultured hippocampal neurons causes increased skipping of the NCAM 180-specific exon (Schor et al., 2009) and neural activity can also regulate the relative levels of apCAM isoforms (Schacher et al., 2000). Given the general conservation of splice isoform structure between these genes and FasII, it is possible that homologs of Beag and Dsmu1 could contribute to this regulation.
FasII at the synapse
Although important roles of FasII in synapse growth and plasticity are well established (Packard et al., 2003; Kristiansen and Hortsch, 2010), the expression and function of individual Drosophila FasII splice isoforms had not previously been systematically analyzed. In this study, we demonstrate that the transmembrane isoforms of FasII are essential for synaptic development. In contrast, FasII-C cannot rescue synaptic development in fasII mutants. These results are consistent with prior results showing that expression of transgenic FasII-A-PEST+, in contrast to FasII-C, can alter synapse function in the CNS (Baines et al., 2002). Previous studies have also shown that reducing levels of all FasII isoforms by 90–100% causes a large decrease in NMJ bouton number, which we confirm here (Schuster et al., 1996a, Ashley et al., 2005). However studies of more modest reductions of fasII in heterozygote mutants have been inconsistent (Schuster et al., 1996a, Ashley et al., 2005). Here we find that the bouton number is unchanged compared with wild type in larvae heterozygous for either of two fasII alleles or for a deficiency that removes the entire fasII gene. We have confirmed that these fasII manipulations have a ∼50% decrease in synaptic FasII levels. The contrast between fasII heterozygotes and beag mutants suggests that a change in the relative FasII isoform levels and the presynaptic to postsynaptic FasII ratio, rather than a uniform reduction in FasII levels, can produce aberrant synaptic morphology.
The importance for normal NMJ development of the relative levels and presynaptic and postsynaptic distribution of individual FasII isoforms is supported by several lines of evidence. Simultaneous presynaptic and postsynaptic overexpression of FasII-A-PEST+ induces synaptic overgrowth (Ashley et al., 2005), while in contrast we find that solely presynaptic overexpression of FasII-A-PEST+ does not alter NMJ morphology. In addition, we show that neuronal overexpression of FasII-A-PEST− can induce synaptic overgrowth while overexpression of FasII-C, like FasII-A-PEST+, does not, revealing unique effects of each isoform on synapse growth. Furthermore, expression of FasII-A-PEST+ but not FasII-A-PEST− can rescue the beag NMJ morphology defects. These data suggest that these two isoforms have distinct activities and that a disruption in the presynaptic balance of transmembrane FasII isoforms at the NMJ in beag mutants leads to a decrease in bouton number, even while overall total synaptic FasII levels are maintained.
Our results also show that the intracellular domain of FasII-A-PEST+ alone is sufficient to rescue beag mutant synapse morphology defects, revealing an important function for this domain that is independent of transynaptic adhesion. It is noteworthy that differences have been described in the downstream signaling of transmembrane isoforms of mammalian NCAM. For example, the src family kinase Fyn interacts with NCAM 140, but not 180, and this interaction is essential for neurite outgrowth (Beggs et al., 1994, 1997). In contrast, the scaffolding protein Spectrin has a higher affinity for NCAM 180 than 140, which is required for NCAM-induced recruitment of synaptic proteins to the postsynaptic density (Pollerberg et al., 1986, 1987; Leshchyns'ka et al., 2003). Little is known about the cytoplasmic signaling functions of either of the Drosophila FasII transmembrane isoforms. Comparison of NCAM 180 and 140 with FasII-A-PEST+ and PEST− shows little homology between their intracellular domains. The full cytoplasmic domain of FasII-A-PEST+ is sufficient for postsynaptic localization at the NMJ (Zito et al., 1997), and we show that this domain is also sufficient for NMJ presynaptic localization. The cytoplasmic domains of both FasII-A isoforms contain a C-terminal PDZ-binding sequence. We find this domain is not absolutely required for presynaptic localization of FasII-A-PEST+ and is also not required for rescue of beag mutants by FasII-A-PEST+. The only difference between FasII-A-PEST+ and FasII-A-PEST− is one exon encoding 29 aa within the intracellular region. These 29 aa contain a PEST sequence (rich in proline, glutamic acid, serine, and threonine residues) that could preferentially target FasII-A-PEST+ for proteasomal degradation (Rechsteiner, 1988). It is difficult, however, to relate enhanced degradation to the specific synaptic activity of FasII-A-PEST+ in beag mutants. Nonetheless, analysis of beag and dsmu1 mutants has revealed unique roles for transmembrane FasII isoforms at the synapse and an important future goal will be to determine the nature of these distinctions.
Footnotes
E.S.B. was funded by the National Institutes of Health (NIH) GM07367 and B.D.M. was funded by NIH NS075572. We would like to thank James Manley and Rachel Kraut for advice and members of the McCabe lab for helpful discussions. We thank Chris Henderson, Wesley Grueber, and Jane Dodd for careful review of the manuscript. We are very grateful to Hermann Aberle, Karen Zito, Corey Goodman, Ji-Wu Wang, the Developmental Studies Hybridoma Bank, and the Bloomington stock center for reagents.
The authors declare no competing financial interests.
- Correspondence should be addressed to Brian D. McCabe, Center for Motor Neuron Biology and Disease, Department of Pathology and Cell Biology and Department of Neuroscience, Columbia University Medical Center, 630 W. 168th Street, New York, New York 10032. brian{at}mccabelab.org