Implications of the Sap47 null mutation for synapsin phosphorylation, longevity, climbing proficiency and behavioural plasticity in adult Drosophila

The Sap47 gene was identified in Drosophila by using a monoclonal antibody from the Würzburg hybridoma library (Reichmuth et al., 1995; Hofbauer et al., 2009). Sap47 represents a founding member of the superfamily of proteins containing a BSD domain, which is observed in BTF2-like transcription factors, Sap47 homologues, and DOS2-like proteins (Doerks et al., 2002). In third instar larvae and adults, Sap47 is concentrated in synaptic neuropil and in synaptic boutons, where the protein is mainly found in close proximity to synaptic vesicles (Reichmuth et al., 1995; Saumweber et al., 2011). However, Sap47 mRNA in embryos is detected prior to synaptogenesis (Flybase; Gramates et al., 2017), indicating that Sap47 is not an exclusively synapse-specific protein. Drosophila Sap47 null mutants are viable and fertile (Funk et al., 2004) but third instar larvae show a ∼50% reduction in the ability to learn and/or remember the association of an odorant with a rewarding tastant (Saumweber et al., 2011; Kleber et al., 2015). Basic synaptic transmission is normal in Sap47 null mutants, but voltage clamp recordings at larval neuromuscular junctions revealed enhanced synaptic depression during high frequency stimulation, indicating a defect in short-term synaptic plasticity (Saumweber et al., 2011).

The human orthologue of Sap47, termed SYAP1, was found to be differentially regulated by tamoxifen in breast cancer cells (Al-Dhaheri et al., 2006) and SYAP1 mRNA is detected in most human tissues (Chang et al., 2001). Syap1 interacts with the telomere component POT1 (Lee et al., 2011) and associates with proteasome subunits (Havugimana et al., 2012). Recently, Syap1 (also known as BSTA or BSD domain-containing signal transducer and Akt interactor) was shown to play an essential role in adipocyte differentiation that required the suppression of the FoxC2 transcription factor gene. It was observed that Syap1 enhances phosphorylation of protein kinase B1 (PKB1/Akt1) at Ser473 in certain mammalian cell lines after growth factor stimulation. In addition, the BSD domain was demonstrated to be essential for the interaction of Syap1 and Akt1, which in turn appears to depend on mTORC2-mediated phosphorylation of Syap1 (Yao et al., 2013). In mice, Syap1 has been detected in all tested regions of the nervous system and also in non-neural tissues such as muscle or liver (Schmitt et al., 2016). In cultured primary motoneurons, Syap1 is concentrated in the perinuclear region near the Golgi complex and is also found in axons and growth cones. In these cells, however, Syap1 knockdown or knockout does not reduce total Akt phosphorylation (Schmitt et al., 2016). In the postnatal mouse brain, Syap1 is widely distributed in synaptic neuropil with high concentrations in glutamatergic synaptic regions, and is also detected in perinuclear structures close to the Golgi of specific neuronal somata. Similarly to Sap47 null mutant flies, Syap1 knockout mice are viable and fertile (Schmitt et al., 2016) but show distinct deficiencies in motor behaviour (von Collenberg et al., 2019).

Synapsins constitute a family of conserved vesicle-associated synaptic phospho-proteins (Greengard et al., 1993). They are involved in synaptic vesicle clustering and appear to regulate the reserve pool of synaptic vesicles in a phosphorylation-dependent manner. Vertebrate synapsins are substrates for various kinases, including PKA, CaMKI, CaMKII, CaMKIV, MAPK, Src and Cdk1/5, which regulate binding of actin and synaptic vesicles and neurotransmitter release (reviewed by Cesca et al., 2010). In particular, impairment of synapsin function modifies synaptic plasticity as revealed by paired-pulse facilitation as well as post-tetanic and long-term potentiation. Mutations in the synapsin genes of humans are associated with epilepsy, aggressive behaviour, learning problems and autism (Garcia et al., 2004; Fassio et al., 2011; Corradi et al., 2014). Indeed, SYN1 mutations are considered to be one cause of X-chromosome-linked intellectual disability (Ropers and Hamel, 2005).

The Drosophila synapsin gene (Syn) was identified as the first invertebrate member of the synapsin protein family (Klagges et al., 1996). At larval neuromuscular junctions (NMJs) synapsin directly interacts with the endocytic scaffolding protein Dap160. The resulting functional complex contains epidermal growth factor receptor substrate 15 (Eps15) and is essential for the correct localization of synapsin and the re-clustering of synaptic vesicles in active NMJs (Akbergenova and Bykhovskaia, 2007, 2010; Koh et al., 2007; Pechstein and Shupliakov, 2010; Denker and Rizzoli, 2010; Winther et al., 2015). Syn null mutants show defects in behaviour, including impaired olfactory habituation (Sadanandappa et al., 2013) and associative learning and memory (Godenschwege et al., 2004; Michels et al., 2005, 2011; Knapek et al., 2010; Diegelmann et al., 2013; Kleber et al., 2015; Walkinshaw et al., 2015; Niewalda et al., 2015). These defects can be ‘rescued’ by transgenic expression of wild-type synapsin but not by synapsin with mutated phosphorylation sites. Thus the tissue distribution of phosphorylated synapsin in the adult brain is of particular interest. Here, we use an antiserum specific for synapsin phosphorylated at Ser6 to report this distribution for the first time.

Considering that knockout of either synapsin or Sap47, or both, has similar effects on associative learning and that the effects are not additive in the double mutant (Saumweber et al., 2011; Kleber et al., 2015), we wondered if this applies to other phenotypes as well. Within this context, we describe previously uncharacterized behavioural traits of adult Sap47 and Syn null mutants and investigate the interaction of the two synaptic proteins by molecular, electrophysiological and genetic techniques. We present evidence that synapsin and Sap47 functionally cooperate differentially in different behavioural settings but apparently do not interact directly.

Two lines of Canton S wild type (CS^NF and CS^V, denoted as WT_CS1 and WT_CS2) maintained separately for more than 10 years were used. The Sap47^156CS and Syn^97CS mutants denoted as Sap^−/− and Syn^−/−, respectively, are P-element jump-out null mutant lines brought into CS^NF background by at least 12 generations of back-crossing and have been described in detail previously (Saumweber et al., 2011; Funk et al., 2004; Godenschwege et al., 2004; Michels et al., 2005, 2011; Sadanandappa et al., 2013). The Syn⁷⁹ mutant (denoted as Syn^−/−79) represents an independent synapsin null allele (Godenschwege et al., 2004). Sap47^156w, Sap47^201w, Sap47^208w and Syn^79w are non-Cantonized independent jump-out null mutants of the respective P-element lines (Funk et al., 2004; Godenschwege et al., 2004). Note that the epitopes of the monoclonal antibodies against synapsin (Synorf1) and Sap47 (nc46) used here are located in the N-terminal domains common to all known isoforms (Godenschwege et al., 2004; Funk et al., 2004; Saumweber et al., 2011) verifying that all isoforms are eliminated by the respective mutations. The nervous system-specific rescue of the Sap47¹⁵⁶ mutant was accomplished by using the F₁ generation of elav-Gal4;;Sap47¹⁵⁶ crossed to w;;UAS-Sap47-RF,Sap47¹⁵⁶ as described by Saumweber et al. (2011). Sap47-RF represents the most strongly expressed nervous system specific splice variant of the Sap47 gene (FlyAtlas2; Leader et al., 2018). The Sap-Syn double mutants were generated as follows: from individual recombinant chromosomes obtained by mating F₁ female offspring of a Sap47^156CS×Syn^97CS cross to TM3/TM6 double balancer males, stocks were established and balanced flies were screened by PCR for deficiency of both genes. Four independent recombinants were isolated, which all were homozygous lethal but viable as transheterozygotes, indicating that all four recombinant chromosomes contained different third site lethal mutations. The transheterozygote stocks are designated Sap-Syn^NS17, Sap-Syn^NS62, Sap-Syn^NF and Sap-Syn^V. From one recombinant chromosome, Sap-Syn^V, the third site lethality was removed by outcrossing to WT CS^V flies for six generations, using single-fly PCR to identify recombinant chromosomes, thereby generating the homozygous viable recombinants Sap-Syn^V1, Sap-Syn^V2, Sap-Syn^V3. The lack of transcripts was verified by RT-PCR (Fig. S1). No significant differences among the three double-mutant lines were detected in any of the tests employed in the present study. No significant change of Syn transcript in Sap47¹⁵⁶ or of Sap transcript in Syn⁹⁷ mutants were observed (Fig. S1). The mutations were also verified by western blots using antibodies against synapsin and Sap47, which demonstrated the absence of all isoforms of both proteins (Fig. S2). The Syn(S6A,S533A) line, kindly provided by Birgit Michels (Michels et al., 2011), expresses a synapsin cDNA with point mutations replacing Ser6 and Ser533 with Ala under the control of elav-Gal4 in the Syn null background (Michels et al., 2011). Ser6 mutation prevents the phosphorylation of Ser6, a site specifically recognized by anti-PSyn(S6).

Generation of the Latrophilin (Cirl) mutant [w¹¹¹⁸; Cirl^ko attP^Cirl loxP-w_⁺-loxP;; (Cirl^ko w⁺)] and rescue [w¹¹¹⁸; Cirl^ko {w^+mC=pTL370[Cirl]}attP^Cirl loxP;; (Cirl^rescue w⁺)] strains used in this study have been described previously (Scholz et al., 2015). The w;UAS-Aβ42;+ transgenic line was obtained from the Bloomington Stock Center (no. 33769). w;gmr-Gal4;+, elav-Gal4;+;+, elav-Gal4;+;Sap47¹⁵⁶, and the double balancer w;Sco/CyO;TM6,Tb,Hu/MKRS,Sb were kindly provided by Burkhard Poeck, Stephan Sigrist, Birgit Michels, and Thomas Raabe, respectively. Using the double balancer, homozygous w;UAS-Aβ42;Sap47¹⁵⁶ and w;gmr-Gal4;Sap47¹⁵⁶ recombinants were generated. w;UAS-Aβ42;Sap47¹⁵⁶ was crossed with either w;gmr-Gal4;+ or elav-Gal4;+;+ or w;gmr-Gal4;Sap47¹⁵⁶ or elav-Gal4;+;Sap47¹⁵⁶ driver lines to generate the flies expressing Aβ42 either in the retina (gmr-Gal4) or in the nervous system (elav-Gal4) in Sap47¹⁵⁶/+ heterozygous or Sap47¹⁵⁶ homozygous background, respectively.

The Sap47^rescue line (elav-Gal4/+;UAS-Sap47-RF/+;Sap47¹⁵⁶/Sap47¹⁵⁶) was generated by crossing elav-Gal4;;Sap47¹⁵⁶ with UAS-Sap47-RF;Sap47¹⁵⁶ described previously (Saumweber et al., 2011). The Sap47 deficiency in the parental lines and Sap47 expression in the rescue line were verified by western blotting.

The lines expressing Gal4 in different sets of ring neurons of the central complex, c819-Gal4, 189y-Gal4, c232-Gal4 (Renn et al., 1999), EB1-Gal4 (Spindler and Hartenstein, 2010) and ftz-ng-Gal4/TM3,Sb (Kuntz et al., 2012) were obtained from Burkhard Poeck and crossed with a line with 10xUAS-GFP constructs on the X chromosome, which was kindly provided by Christian Wegener.

Monoclonal antibodies from the Würzburg hybridoma collection (Hofbauer et al., 2009) were used against the following proteins: Sap47 (nc46; 1:200 for WB, 1:50 for IHC), Syn [Synorf1 (3C11) 1:100 for WB, 1:20 for IHC], bruchpilot (nc82; 1:100), CSP (ab49; 1:200 for WB, 1:50 for IHC). These antibodies are available from DSHB. Anti-PSyn(S6) rabbit serum against synapsin phosphorylated at Ser6 has been described previously (Sadanandappa et al., 2013) and was used at 1:200 (WB) or 1:500 (IHC). The specificity of binding of the PSyn(S6) antiserum to synapsin phosphorylated at Ser6 is demonstrated by the lack of signals in transgenic flies expressing synapsin with an S6A mutation in a synapsin null background, although the anti-Syn antibody Synorf1 proves the expression of the mutated synapsin [Fig. S2B, top, lane Syn(S6)]. Specificity of nc46 and Synorf1 antibodies has been demonstrated previously (Saumweber et al., 2011; Funk et al., 2004; Klagges et al., 1996; Godenschwege et al., 2004; Michels et al., 2005, 2011) and was verified here again by western blotting as no signals were obtained with Syn or Sap47 antibodies for the respective mutants (Fig. S2A,B). Cy3-conjugated anti-HRP (Jackson ImmunoResearch) stains Drosophila neuronal membranes via a cross-reaction with Drosophila Na-K-ATPase and was used at 1:250. Secondary antibodies, goat anti-rabbit peroxidase coupled (WB: 1:1000) and goat anti-mouse peroxidase coupled (WB: 1:1000) and fluorophore-coupled antibodies (IHC 1:500) were provided by Jackson Laboratory, Farmington, CT, USA.

Fly heads were homogenized in Laemmli sample buffer, proteins were separated by standard SDS-PAGE, transferred to a nitrocellulose membrane by wet blotting (Bio-Rad Laboratories, München, Germany), and detected after incubation with specific primary and HRP-coupled secondary antibodies by chemiluminescence (ECL, GE Healthcare Europe, Freiburg, Germany). For comparison of native versus de-phosphorylated synapsin, fly heads were homogenized in buffer A (25 mmol l⁻¹ Tris-HCl, 150 mmol l⁻¹ NaCl, 10% glycerol, 0.1% NP40, pH 7.7), to only one of two aliquots (each containing 1 head equivalent) 40 U of shrimp alkaline phosphatase (Promega, Madison WI, USA) were added (AP+), both aliquots (AP+ and AP−) were incubated for 15 min at 37°C (to activate the enzyme), the reaction was stopped with 2× Laemmli buffer, and the samples were loaded on a standard SDS gel. For separation of protein complexes under non-denaturing conditions, we used blue native polyacrylamide gel electrophoresis (Wittig and Schägger, 2008) with the NativePAGE buffer system (Invitrogen, Karlsruhe, Germany) following the manufacturer's instructions. Single lanes were cut from the blue native gel and subjected to SDS PAGE using a single well 4–12% Bis–tris gel and the associated buffer system (NuPAGE, Invitrogen) according to the manufacturer's instructions. Proteins were transferred to a nitrocellulose membrane by wet blotting and detected by ECL as described above.

For immunohistochemistry of whole-mount fly brains, we followed the procedure described by Wu and Luo (2006). Preparations of fly heads for cryo-sectioning requires removal of proboscis and air sacks below the brain, fixation for 3 h in fresh 4% buffered paraformaldehyde, freeze protection in 25% sucrose in Ringer’s solution (128 mmol l⁻¹ NaCl, 4.7 mmol l⁻¹ KCl, 1.7 mmol l⁻¹ CaCl, 0.7 mmol l⁻¹ Na₂HPO₄, 0.35 mol l⁻¹ KH₂PO₄, pH 7.4) overnight, and sectioning at 30 µm thickness using a cryostat microtome. Staining for fluorescence microscopy has been described previously (Halder et al., 2011; Blanco-Redondo et al., 2013). Image stacks were taken at 2 µm z-steps using an Olympus Fluoview FV1000 confocal microscope equipped with an Olympus UPLSAPO 20× (air) objective. Whenever the WT image of a CLSM (confocal laser-scanning microscopy) scan was contrast enhanced (40–50%) to compensate for staining variability due to different z-positions within the 30 µm section the corresponding mutant image was contrast enhanced by the same value. No signals are obtained at the same parameter settings when the first antibody is omitted. Immunohistochemistry of larval neuromuscular junctions was performed as previously described (Schmid and Sigrist, 2008).

For 2D electrophoresis, iso-electric focusing by the ZOOM IPGRunner system was used (Invitrogen, Carlsbad, CA, USA). 100 fly heads were separated from bodies and appendages by freezing in liquid nitrogen and vigorous shaking using two successive sieves. The heads were homogenized in 100 µl of the sample homogenizing mix (ZOOM 2D Protein Solubilizer 1) and protease inhibitors. The homogenate was then centrifuged twice and the post-nuclear supernatant was incubated with 1 µl N,N-dimethylacetamide (DMA; Sigma) for 15 min. Protein alkylation was stopped with 2 µl of 2 mol l⁻¹ dithiothreitol (DTT; Sigma). 25 µl of this homogenate were added to the rehydration mixture as recommended by the manufacturer. Thereafter, the sample was loaded into the sample loading well (ZOOM IPG Runner Cassette). The ZOOM Strip (pH 3–10) was inserted into the cassette and loaded and sealed, the ZOOM IPG Runner Cassette was left at 18°C overnight. The next day, the cassette was placed in the ZOOM IPG Core and slid into the Mini-Cell Chamber of the IPG Runner and locked. The proteins went through five voltage steps before being placed into the single-well 4–12% Bis-Tris NuPAGE 2D gel and being treated as a regular western blot as described above.

For quantification of transcript levels in the mutants by real-time PCR total RNA from fly heads was isolated using the RNeasy mini kit (QIAGEN, Hilden, Germany, following the manufacturer's instructions) and reverse transcribed. The Rotor-Gene Q SYBR Green system (QIAGEN, Hilden, Germany) was applied to 100 ng of purified cDNA following the manufacturer's instructions. Primer sequences used were: Synapsin, for 5′-CTTAACGTTCATCGGCCATT-3′ and rev 5′-AGGGGTTCGCTTCGTTACTA-3′; Sap47, for 5′-TAAAAGTTGGAGAGCCAGGA-3′ and rev 5′-GGTGGCTTCGGATACTAATG-3′; Cirl, for 5′-GGATGATGCTCATGGATTG-3′ and rev 5′-AAAGCCCCGTAGTCAAGAG-3′.

Two-electrode voltage clamp (TEVC) recordings (Axoclamp 900 A amplifier, Molecular Devices) were performed on male third instar Drosophila larvae, muscle 6 in segments A2 and A3 at room temperature using intracellular electrodes (resistance 10–20 MΩ), filled with 3 mol l⁻¹ KCl, as described previously (Ljaschenko et al., 2013). Composition of extracellular haemolymph-like solution (HL-3; Stewart et al., 1994) was as follows (in mmol l⁻¹): NaCl 70, KCl 5, MgCl₂ 20, NaHCO₃ 10, trehalose 5, sucrose 115, Hepes 5 and CaCl₂ 1, pH adjusted to 7.2. For analyses, only cells with an initial membrane resistance ≥4 MΩ and membrane potential of at least −50 mV were accepted. Cells were held at −60 mV, except for minis (miniature excitatory postsynaptic currents), where holding potential was adjusted to −80 mV (for 90 s). To elicit action potential-evoked excitatory postsynaptic currents (eEPSCs), brief (300 µs) pulses were applied via a suction electrode at typically 10–20 V. Paired-pulse recordings were performed with increasing inter-stimulus intervals of (in ms): 10, 30, 100, 300 and 1000 with 10 s rest in between the recordings. For low frequency stimulation and paired-pulse recordings, 10 single traces were averaged per cell and inter-pulse interval. The amplitude of the second response in 10 ms inter-pulse recordings was measured from the peak to the point of interception with the extrapolated first response. Recordings were sampled at 10 kHz and low-pass filtered at 1 kHz. Analyses were carried out using Clampfit 10.5 (Molecular Devices) and Sigmaplot 12.5 (Systat Software). Data are presented as means±s.e.m. and statistics employed the non-parametric rank-sum test (versus controls). For electrophysiological recordings and analyses, all genotypes were blinded.

For each genotype (CS^NF, CS^V, Syn⁹⁷, Sap¹⁵⁶, Sap-Syn^V3, Sap47^rescue) five vials of 10 males each were kept at 25°C. Genotypes were encoded for fully blind evaluation. Twice a week the flies were transferred to fresh food vials, and the numbers of dead flies were counted. The Kolmogorov–Smirnov test was used to determine statistical significance of the differences between genotypes. This test is not based on standard errors of individual age points. All error bars but one per genotype have therefore been omitted for clarity.

Rapid induced negative geotaxis (Liu et al., 2015; Cao et al., 2017; Manjila and Hasan, 2018) was used as a simple test for sensory-motor performance. Experiments were done in the late afternoon. Adult male and female flies were assayed separately, but since no consistent differences between the genders were noted, the data were pooled. For each of the genotypes 6 vials of 10 flies each were kept at 25°C and transferred to fresh vials twice a week. Prior to the climbing experiment, flies of the same age were transferred to empty vials (2.5 cm diameter; maximum climbing height, 7.5 cm) for 30 min for acclimation. In a box holding the vials with the different genotypes, the flies were tapped to the bottom and allowed to climb upward for 4 s before a photograph of the assembly was taken. After another 10 s, the procedure of tapping, waiting and image capture was repeated, for a total of 10–12 times. The position of the genotypes in the box was randomized and encoded for fully blind evaluation. Photos of the assembly were analysed using the threshold and particle analysis features of ImageJ and a custom written Basic subroutine to score the flies' positions. For each age of the flies, the mean of the fly positions in each vial was scored and averaged over the 10–12 repeats for each genotype. Normalization of each experiment to the response of CS did not reduce standard deviations of the mutant data, indicating that technical variations such as differences in tapping or environmental effects like time of day, temperature, weather conditions, etc. did not systematically influence the results. The climbing success of the mutants relative to WT was averaged over all ages. The Student's t-test with Bonferroni correction (six comparisons) was used to determine statistical significance of the differences between the four genotypes.

This assay was performed as described in detail previously (Agrawal et al., 2013; Manjila and Hasan, 2018). Briefly, flies were anesthetized for 10–15 min on ice and then attached with nail polish between head and thorax to a thin wire. Flight was initiated by a gentle air puff and the time to a spontaneous termination of flight was recorded. The Student's t-test was used to determine statistical significance of the differences between genotypes.

The locomotor activity of single male flies was recorded using the TriKinetics DAM2 System (TriKinetics, Waltham, MA, USA) at 20±0.5°C and 60±1% relative humidity, as described previously (Gmeiner et al., 2013). For the first 6 days, flies were recorded under 12 h:12 h light:dark (LD) cycles and subsequently for 13 subsequent days in constant darkness (DD). Illumination during the light phase was provided by white LEDs (Lumitronix LED-Rechnik, Jungingen, Germany) and set to 47.6 μW cm⁻². Actograms of individual flies were plotted and the free-running period and power was calculated by Sokolov–Bushell periodogram analysis using the program ActogramJ (Schmid et al., 2013).

For testing the plasticity of the circadian clock, flies were reared under different photoperiods (LD 8:16, LD 12:12 and LD 16:8) and then recorded for 14 days in DD. Sleep analysis was only performed during LD 12:12 cycles. Sleep was defined as a period of inactivity longer than 5 min (Hendricks et al., 2000; Shaw et al., 2000). Calculations of total sleep were performed for each hour of the day using a macro written in Microsoft Excel 2007 (courtesy of Dr Taishi Yoshii).

For 12 h sleep deprivation, the glass tubes with the flies were taken out from TriKinetics system at lights-off and placed on a programmable rotator (Multi Bio RS-24, BioSan). The programme was set to rotate at 18 rpm 4 times in one direction and then 4 times in the opposite direction. Each directional shift was separated by three 5 deg vibro-rotations. After sleep deprivation, flies were re-inserted into the Trikinetics system and recorded for a further 3 days. All procedures were carried out under LD 12:12 at 20°C.

In order to obtain information on the molecular function of Sap47 we attempted to identify its interaction partners. After immunoprecipitation, we failed to detect any signals in silver-stained gels that were not also detected by our anti-Sap47 antibody (nc46). We also systematically compared western blots of WT and Sap^−/− head homogenates probed with antibodies against various different synaptic proteins. In this context, we discovered a qualitatively altered synapsin signal of the monoclonal antibody Synorf1 in all three independent Sap47 null mutants, Sap47¹⁵⁶, Sap47²⁰¹ and Sap47²⁰⁸, irrespective of their genetic background [Canton-S (CS) or w¹¹¹⁸ (w)] (Fig. 1A). In Sap47 null mutants, an additional synapsin-specific band at slightly higher apparent molecular weight could be discerned (arrow in Fig. 1A, AP−). Interestingly, this additional signal could be eliminated by treating the head homogenate with alkaline phosphatase (Fig. 1A, AP+), hinting at a hyper-phosphorylation of synapsin. However, mass spectrometric analysis of synapsin peptides after liquid chromatography (LC-MS/MS) identified only a single phospho-serine in adult Sap^−/− mutants that was not phosphorylated in WT (S574 in Fig. S3), at least within those regions of the protein for which the present methods had sufficient coverage. Indeed, coverage of synapsin phosphorylation by LC-MS/MS has repeatedly been reported to be partial (Kleber et al., 2015; Niewalda et al., 2015; Nuwal et al., 2011). We were therefore not discouraged from investigating Ser6 phosphorylation, which was not covered by the present LC-MS/MS approach. Using a custom-made rabbit antiserum [anti-PSyn(S6)] (Sadanandappa et al., 2013) against an N-terminal synapsin peptide containing phosphorylated Ser6, we confirmed by western blotting that Ser6 phosphorylation in Sap^−/− mutants was reduced compared with WT levels (Fig. S2A,C). Comparison of western blot signals of anti-PSyn(S6) antiserum and the pan-Syn antibody Synorf1 revealed that the read-through synapsin isoforms of 143 kDa (Klagges et al., 1996) were more efficiently phosphorylated at Ser6 compared with the short isoforms of 70–80 kDa (Fig. S2D).

By 2D electrophoresis, approximately 20 synapsin isoforms of 70–80 kDa were separated that are recognized by the monoclonal antibody Synorf1 (Fig. 1B). Of these, the isoforms of the lower row tentatively labelled 17 to 21 in Fig. 1B (arrow in Fig. 1D) were not phosphorylated at Ser6 while isoforms of the middle and upper rows were phosphorylated at this amino acid (Fig. 1C).

Repeated attempts to co-immunoprecipitate synapsin with the Sap47 antibody or Sap47 with the Synorf1 antibody failed to show any sign of direct binding of the two proteins (not shown). We therefore tested if the two proteins migrate in a non-denaturing gel in the same protein complex. By subjecting the protein complexes of WT fly heads homogenized in non-denaturing buffer first to non-denaturing electrophoresis (horizontal dimension in Fig. 1E) and then to SDS-PAGE (vertical dimension in Fig. 1E), we show that synapsin isoforms are found in protein complexes of >450 kDa, whereas Sap47 co-migrated with complexes of <450 kDa (Fig. 1E). Thus, it seems unlikely that the two proteins in the brain tissue interact directly as components of the same macromolecular complex.

Syap1, the mammalian homologue of Sap47, has been implicated in modulating the target of rapamycin (TOR) pathway (Yao et al., 2013) which plays a central role in the regulation of lifespan (de Cabo et al., 2014). We therefore wondered if mutation of the Sap47 gene influences Drosophila lifespan. Considering the similar effects of knockout of Sap47 or synapsin on associative learning and the clear influence of the Sap47 mutation on synapsin isoforms, we also asked whether the roles of the two genes for longevity are likewise similar and whether possible defects add up in double mutants. Similarly to the single-mutation lines, three viable Sap-Syn double-mutant lines generated by recombination (see Materials and Methods) showed no obvious phenotype. Quantitative comparison with WT, however, revealed clear deficiencies. The results of the quantification of survival of Sap^−/− and Syn^−/−, as well as Sap^−/−, Syn^−/− flies are shown in Fig. 2A. The Sap47 gene had a clear detrimental influence on lifespan. Under our experimental conditions, we found no significant difference in survival times between the Syn^−/− line and two Canton-S WT strains, or between Sap^−/− and Sap^−/−, Syn^−/− flies, but the latter two lines showed about a 38% shorter life time (50% survival score) compared with the WT (Fig. 2A). Pan-neural expression of the most abundant brain-specific Sap47 mRNA (RF splice variant, see Discussion) using the UAS-Gal4 system with elav-Gal4 as driver rescued the short lifespan of the Sap^−/− mutant (Fig. S4A).

Compromised motor performance is a hallmark of many neurodegenerative diseases. We wondered how the loss of two proteins abundantly expressed in most or all presynaptic terminals affect motor performance with and without induction of neurodegeneration. Climbing, a simple test for motor performance, makes use of a negative geotactic reflex that can be elicited by physical disturbance (see Materials and Methods). Mean climbing proficiency was reduced by 27.6% in Syn^−/− (P=0.011), by 49.9% in Sap^−/− (P=0.0016) and by 66.5% in Sap^−/−, Syn^−/− double mutants (average of the 3 Sap-Syn lines Sap-Syn^NS62, Sap-Syn^V2, Sap-Syn^V3; P=0.000013) compared with WT-CS animals. With increasing age, climbing proficiency deteriorated at roughly similar rates in all four genotypes (Fig. 2B). The climbing defect of the Sap^−/− mutant was rescued by the pan-neural expression of the Sap47 mRNA RF splice variant (Fig. S4B). We also tested the impairment of motor function in Sap^−/− mutants in a flight duration test (Agrawal et al., 2013; Manjila and Hasan, 2018). Sap^−/− knockout flies terminated air-puff-induced flight 2.45 min earlier than WT control flies (Fig. S4C, P=0.00104).

Age-related decline in climbing proficiency can be accelerated by induction of neurodegeneration via pan-neural (elav-Gal4) expression of human Aβ42 peptide (Liu et al., 2015; Fernendez-Funez et al., 2015; Ling et al., 2009; Ping et al., 2015; Rogers et al., 2012; Sofola et al., 2010; Wang et al., 2015; Costa et al., 2011). We investigated whether the Sap^−/− deficiency had any influence on this effect (Fig. 2C). As controls, we used the driver gmr-Gal4 to obtain flies expressing Aβ42 only in the retina, which leads to a ‘rough eye’ phenotype but does not impair brain function (Costa et al., 2011; Prüßing et al., 2013; Wang et al., 2015; Barucker et al., 2015). Similarly to the data shown in Fig. 2A, in these controls, homozygous Sap^−/− deficiency significantly reduced climbing proficiency by 20% (P=0.0015) compared with heterozygous deficiency. Surprisingly, when human Aβ42 was expressed pan-neurally using the elav-Gal4 driver, no such difference between homozygous versus heterozygous Sap^−/− deficiency was observed (P=0.132) (Fig. 2C) (see Discussion). Importantly, the rate of age-related decline in motor performance by Aβ42 expression was also not modified by the presence or absence of Sap47 (Fig. 2C), suggesting that the molecular mechanisms by which the Aβ42 transgene and the Sap47 gene cause the age-related climbing defect are not independent. Aβ42 expression appears to block the deleterious effect of Sap47 deficiency on climbing.

Sap47 mRNA was found to be enriched in the ventral lateral clock neurons (LN_vs) and showed different levels of accumulation during the day and night (Kula-Eversole et al., 2010), suggesting that it may be linked with the circadian clock or sleep. To test the influence of Sap47 on rhythms and sleep, we monitored the locomotor activity of Sap^−/−, Syn^−/− or Syn-Sap double mutants under light–dark cycles and constant darkness. Furthermore, we monitored the plasticity of the circadian period after rearing the flies under different day lengths (LD 8:16, LD 12:12, LD 16:8). A previous study had shown that flies reared under short days (LD 8:16) exhibit a significantly longer period length under following constant darkness (DD) conditions compared with flies reared under extended daylight (LD 16:8) (Tomioka et al., 1997). To test a possible effect of Sap47 and Syn on sleep plasticity, we recorded the sleep rebound of the mutants after a 12-h night of sleep deprivation, which is known to cause significantly increased sleep during the subsequent day in WT flies (Hendricks et al., 2000; Shaw et al., 2000).

We found that the basic activity/sleep patterns of WT and mutant flies were very similar (Fig. 3A). Under LD conditions, all flies showed locomotor activity in the morning and evening that was interrupted by sleep during the middle of the day (the siesta) and at night (a two-way ANOVA revealed no influence of the genotype on day time and night time sleep; F_3,198=1.188; P=0.757). Sleep during the siesta appeared slightly more pronounced in the mutants than in the WT (Fig. 3A,C,D), but this difference was not significant in a Bonferroni post hoc test (P=1.000). Under constant darkness (DD), all flies showed significant circadian rhythms with a strain-specific period length (Fig. 3B). Rearing the flies under short and long days significantly changed period length and power in the WT strains, but not in the mutants (Fig. 3B). In WT flies, period was significantly shorter (P<0.001) and power lower (P<0.001) in flies that were reared under long days as compared with those reared under short days. These differences were absent in the mutants (P=1.000), suggesting that they lack the WT-like plasticity in circadian behaviour. After rescuing Sap47 in all neurons, the WT plasticity in period (P<0.001) and power (P=0.008) was restored (Fig. 3B).

After sleep deprivation, Sap^−/− and Syn^−/−, as well as Sap-Syn double mutants lacked the WT-like sleep rebound (Fig. 3C,D). Whereas WT flies slept significantly more during the entire following day (P<0.001), the mutants did so only during the first 2 h of the day (arrows in Fig. 3C). When the amount of sleep in the mutants before and after sleep deprivation was compared over the entire following day, no significant differences were found (P=1.000).

To study the influence of Sap47 and synapsin on neurotransmission, two-electrode voltage clamp (TEVC) recordings were performed from larval neuromuscular junctions (NMJs; Fig. 4). Spontaneous presynaptic neurotransmitter release can be assessed by measuring miniature postsynaptic currents (minis) elicited by individual vesicle fusions. Mini frequencies were subtly but significantly increased in both single mutants and nearly tripled in the double mutant. Mini amplitudes were reduced in Sap^−/− larvae (Sap47¹⁵⁶), indicating a slight decrease in transmitter content of synaptic vesicles and/or postsynaptic sensitivity compared with WT CS (CS^NF) (Fig. 4A,B). Next, we investigated eEPSCs. While basal eEPSC amplitudes (0.2 Hz stimulation) were comparable in all genotypes (Fig. 4C), we observed diminished paired-pulse facilitation in single and double mutants upon closely spaced stimulation (Fig. 4D,E).

Alerted by a possible genetic interaction of Syn and Cirl by microarray and qPCR (Fig. S1) (Nuwal, 2010), we investigated synapsin levels in synaptic boutons of Cirl null mutants. Quantification of anti-synapsin immunofluorescence in larval nerve muscle preparations of Cirl null mutants indicates a significant reduction of synapsin content in type Ib synaptic boutons, a defect that was rescued to WT values upon re-constitution of the Cirl locus (Fig. 5).

The vertebrate Sap47 orthologue Syap1 is localized close to the Golgi complex of neuronal perikarya in addition to its concentration in neuropil (Schmitt et al., 2016). We therefore repeated immunostainings of frozen Drosophila brain sections using a Sap47 antibody (nc46) and observed that, in addition to the prominent localization of Sap47 in synaptic neuropil (Reichmuth et al., 1995; Funk et al., 2004), there is also significant labelling of axonal fibre tracts and a subset of perikarya in the cellular cortex of the brain (Fig. 6A,C). In contrast, the active zone protein bruchpilot (BRP) (Hofbauer et al., 2009; Wagh et al., 2006; Kittel et al., 2006; Wichmann and Sigrist, 2010) is exclusively found in synaptic neuropil (Fig. 6B).

Intrigued by the importance of synapsin phosphorylation for behavioural plasticity, we investigated the distribution of total synapsin and synapsin phosphorylated at Ser6 in the adult Drosophila brain by performing immunohistochemical experiments. Fig. 7, Fig. S4D and Fig. S5A,B show whole-mount brains and frozen horizontal head sections of white-eyed w;;Syn^+/+ (w¹¹¹⁸) and w;;Syn^−/− null mutant flies stained with the monoclonal antibody Synorf1 (3C11) and the anti-PSyn(S6) antiserum described above. All flies were in the w¹¹¹⁸ background to avoid autofluorescence signals from eye pigments diffusing into the brain during preparation and fixation. As reported previously (Godenschwege et al., 2004), and shown here by comparison with the synapsin knockout (Figs S4D and S5A,B), synapsin distribution in the WT is ubiquitous throughout the neuropil. The PSyn(S6) antiserum, on the other hand, shows significant specific immunoreactivity in the entire central brain neuropil (whole-mount brain preparation in Fig. 7A,B, cryo-sections in Figs S4D and S5A,B), as well as very strong specific staining of various unidentified synapses, including a ring of synapses in the ellipsoid body and sets of apparently large synapses at the rim of the lateral triangle (Fig. 7B and cryosection in C,D). Comparison of Fig. S5A,B left and middle columns indicates that the strong PSyn(S6) staining is indeed due to enhanced phosphorylation rather than stronger synapsin expression. We conclude that synapsin is partially phosphorylated at Ser6 in most synapses of the central brain but strongly hyper-phosphorylated at this amino acid in a subset of central complex synapses and several other structures that show enhanced PSyn(S6) staining. No qualitative difference in the distribution of synapsin and Ser6-phosphorylated synapsin between Sap^−/− mutants and WT was detected in six experiments by staining cryo-sections of w;;Syn^+/+, w;;Sap47²⁰⁸ and w;;Syn^−/− flies on the same microscope slides with Syn and PSyn(S6) antibodies. In the visual system, phosphorylation of synapsin at Ser6 is apparently low; however, detailed analysis is not feasible because, as revealed by staining in the Syn^−/− mutant, the PSyn(S6) antiserum crossreacts with an unknown antigen widely distributed in several synaptic layers of the visual system (Fig. S4D, middle column). This antigen could possibly also be responsible for the weak PSyn(S6) western signal in the Syn^−/− mutant lane of Fig. S2D marked by an asterisk.

In a first attempt to identify the cells whose synapses contain synapsin hyper-phosphorylated at Ser6 in the synaptic rings of the ellipsoid body, we used five Gal4 lines to selectively express green fluorescent protein (GFP) in specific subsets of so-called ‘ring’ neurons. In Fig. 7C,D (EB1-Gal4) and Fig. S6, we compare the Gal4-driven GFP expression in ellipsoid body ring neurons using the lines c819-Gal4 (R2/R4; Renn et al., 1999), 189y-Gal4 (R3; Renn et al., 1999; Kuntz et al., 2012), c232-Gal4 (R3/R4d; Renn et al., 1999; Thran et al., 2013), EB1-Gal4 (R2; Spindler and Hartenstein, 2010) and ftz-ng-Gal4 (R4; Kuntz et al., 2012) with the distribution of endogenous synapsin hyper-phosphorylated at Ser6. None of these Gal4 lines showed GFP expression colocalizing with PSyn(S6) staining, as revealed by careful inspection of individual optical sections with 40× oil optics and confocal microscopy.

The two genes under study here encode several protein isoforms. Alternative splicing generates at least 10 Sap47 transcripts (Flybase; Gramates et al., 2017), while in western blots nine separate bands have been tentatively identified (Funk et al., 2004) of which the 47 kDa isoform is by far the most abundant. For synapsin, four transcripts (Flybase; Gramates et al., 2017) and five isoforms have been described, of which three in SDS gels run at 70, 74 and 80 kDa, while two generated by UAG read-through run at 143 kDa (Klagges et al., 1996). Information on the N-terminus of Drosophila synapsin has been ambiguous. The sequence XKRGFSSGDL identified by Edman degradation of synapsin purified from adult heads has significant homology with the vertebrate synapsin N-terminal A domain (Godenschwege et al., 2004). However, sequence comparisons with various invertebrate synapsin genes demonstrate that 16 amino acid codons upstream of this sequence are highly conserved, suggesting an alternative translation start which results in the N-terminal sequence: LNFSSFKSSFTSNVNFLKRGFSSGDL. As a result of this N-terminal ambiguity, amino acid numbering may differ by 16 amino acids, e.g. Ser6 in focus here is alternatively referred to as Ser22 in some publications (Kleber et al., 2015; Niewalda et al., 2015) and, for easy comparison, in the sequences of Fig. S3.

In any event, the present hypothesis on the role of synapsin in olfactory associative short-term learning assumes that the stimulation of protein kinase A (PKA) leads to phosphorylation of synapsin at one or multiple sites. This is thought to result in a modification of neurotransmitter release (Heisenberg, 2003; Diegelmann et al., 2013; Menzel, 2014; Hige et al., 2015; Gerber and Aso, 2017; Saumweber et al., 2018). This modified transmitter release is assumed to be the basis for learned behaviour. There are two candidate sites (RRXS) for PKA-dependent phosphorylation of synapsin, the conserved Ser6 in the N-terminal A domain and the non-conserved Ser533 between domains C and E. The PKA target site in domain A is removed in most synapsins of larval and adult Drosophila by RNA editing, resulting in RXXS (Diegelmann et al., 2006). The ectopic expression of such an edited cDNA in the mushroom body rescues the learning defect of the null mutant (larvae: Michels et al., 2011; adults: Niewalda et al., 2015). This rescue is prevented if synapsin with mutated RRXS sites (S6A and S533A mutations) is expressed (Michels et al., 2011). Ser→Ala mutation prevents phosphorylation at these sites. Also, olfactory short-term habituation, a simple form of behavioural plasticity, critically depends on the expression of synapsin that can be phosphorylated at Ser6 and/or Ser533 in GABA-ergic local interneurons. In this experimental paradigm, activity of calcium and calmodulin-dependent kinase II (CaMKII) is required, which may also target these sites irrespective of editing (target consensus RXXS) (Sadanandappa et al., 2013). Whether Ser6 or Ser533, or both, are essential for behavioural plasticity in Drosophila remains to be elucidated. It is also not known if PKA can activate, directly or indirectly, CaMKII and in this way phosphorylate Ser6 and/or Ser533 of synapsin irrespective of RNA editing. Thus additional experiments are required to clarify the role of PKA and CaMKII and the detailed molecular modifications of synapsin in behavioural plasticity. For vertebrates, it has recently been shown that in an aqueous environment synapsin can form a distinct liquid phase that rapidly disassembles upon phosphorylation by CaMKII (Milovanovic et al., 2018). It remains to be investigated whether this process is relevant for synapsin function in the behavioural plasticity discussed here. No information is presently available on the function of the two large synapsin isoforms. Of note, these large isoforms are more efficiently phosphorylated at Ser6 than the small isoforms (Fig. S2D).

In qualitative western blots, in addition to the synapsin signals seen in the WT, a band at a slightly larger apparent molecular weight is reliably observed in three independent Sap47 null mutant alleles. These additional western signals were not present when the homogenate is treated with alkaline phosphatase, indicating that the shifted signals may represent phosphorylated synapsin isoforms (Fig. 1A). Quantification of synapsin signals in non-saturated western blots does not indicate an upregulation of total synapsin expression in Sap^−/− mutants. Mass spectrometric analysis of synapsins from adult WT heads showed that at least 32 amino acids of Drosophila synapsins may be phosphorylated (Nuwal et al., 2011; Niewalda et al., 2015). Comparison of synapsin phosphorylation in WT and Sap^−/− mutant larvae revealed a complex pattern of hyper- and hypo-phosphorylation in the mutant at 23 sites (Kleber et al., 2015). Here, we observe the phosphorylation of synapsin Ser574 in adult Sap^−/− mutants, which has not been detected in the WT (Fig. S3, amino acid numbering according to upstream N-terminus). Interestingly, synapsin phosphorylation at Ser6 has been found only once in MS studies (Niewalda et al., 2015), but synapsin phosphorylated at this site is readily detected in western blots using the PSyn(S6) antiserum, which is specific for phosphorylated Ser6 of synapsin (Fig. S2); this signal is reduced in Sap^−/− mutants (Fig. S2A,C,D). Given that a direct binding of both proteins seems unlikely (Fig. 1E), Sap47 might modulate synapsin phosphorylation by regulating site-specific kinases or phosphatases to prevent hyper-phosphorylation at some sites while enhancing phosphorylation of Ser6. This hypothesis would be compatible with the available data but requires further experimental support.

In order to test the possibility that Sap47 has an effect on lifespan and to determine to what extent the apparent interdependence of Sap47 and synapsin influences viability, we compared survival time of the two single mutants and the double knockout flies. Our results suggest that the decline in viability of Sap^−/− mutants is independent of synapsin (Fig. 2A). This is in stark contrast to associative learning, which depends equally on both proteins.

In mammals, Syap1 is strongly expressed in various tissues such as muscles and liver (Schmitt et al., 2016). RNA-seq data from FlyAtlas2 (Leader et al., 2018) show that in Drosophila five Sap47 mRNA splice variants (RE, RF, RG, RI, FJ) are specifically expressed in the adult nervous system, whereas one transcript (RA) is found in all other analysed tissues, albeit at approximately 10-fold lower abundance. At the protein level, Sap47 in Drosophila has so far been described only in the nervous system. Repeated earlier attempts to detect differences in Sap47 staining between WT and Sap47¹⁵⁶ null mutants outside the nervous system have failed. This could have been caused by difficulties identifying weak immunohistochemical signals on top of background due to unspecific antibody binding or might point to inefficient mRNA translation outside the nervous system. We report here that transgenic pan-neural expression of the most abundant brain-specific Sap47 splice variant (RF) in the null mutant rescues the viability and climbing deficiencies of the mutant (Fig. S4A,B). This demonstrates that the expression of Sap47 outside the nervous system is not required for normal lifespan and climbing ability. The functional relevance of expression of the Sap47 RA splice variant may thus be subtle and needs to be investigated in a separate study, which could then also clarify how to interpret the striking increase in lifespan of the rescued flies compared with the WT (Fig. S4A).

Age-related decline of motor performance is a standard test in Drosophila models of neurodegenerative disease (Pendleton et al., 2002; Barone and Bohmann, 2013; Long et al., 2014). We wanted to determine how knockout of synapsin or Sap47, or both, affects amount and rate of age-related decline of climbing proficiency in rapid induced negative geotaxis. Our results suggest that synapsin and Sap47 control climbing through independent pathways such that their defects are compounded in the double mutant (Fig. 2B; see also Fig. S4B for a neuronal rescue of the Sap47 phenotype in climbing). Of note, only the amount, not the age-related rate of decline of climbing proficiency was different in the three genotypes, suggesting that the mutations affect the efficiency of climbing but not the age-related decline in motor function. This interpretation is supported by the observation that the accelerated age-related decline in climbing induced by transgenic pan-neural expression of human Aβ42 peptide is not further enhanced by homozygous Sap47 deficiency (Fig. 2C). The suppression of the climbing deficiency of Sap^−/− mutants in flies expressing Aβ42 (compare violet with red curves in Fig. 2C) could indicate an interesting functional interaction of Sap47 and Aβ42.

In experiments comparing genome-wide gene expression in a subset of adult head clock neurons – the ventral lateral neurons (LN_vs) identified by expression of pigment dispersing factor (PDF) – with expression in randomly selected adult head neurons, Sap47 mRNA was found to be enriched at lights on (ZT0) but not at lights off (ZT12) during a 24 h light–dark cycle (Kula-Eversole et al., 2010). Synapsin concentration in the adult brain, on the other hand, was shown to increase during sleep deprivation (Gilestro et al., 2009). We therefore wondered if knockout of Sap47 and/or synapsin modified circadian activity and/or sleep behaviour. We found that the typical locomotor activity pattern with a morning and an evening peak interrupted by a siesta and a pronounced night-time sleep was also displayed by the mutants (Fig. 3A,C). However, after rearing flies under LD 8:16 conditions (short days, long nights) only WT flies showed the typical extended period and increased amplitude of activity cycles under DD conditions (Fig. 3B), indicating that Sap47 and synapsin may be required for plastic adaptations of the free-running activity cycle. Similarly, sleep rebound requiring a ‘memory’ of sleep deprivation is reduced in the mutants (Fig. 3C,D). These observations of impaired behavioural plasticity are in line with earlier reports on defects in olfactory short-term habituation of Syn^−/− mutants and in olfactory associative conditioning in Sap^−/− and Syn^−/− mutants referred to above.

Previous work on Sap47 and synapsin single mutants showed normal basal but impaired evoked, high-frequency synaptic transmission (Godenschwege et al., 2004; Saumweber et al., 2011; Akbergenova and Bykhovskaia, 2010). Our electrophysiological recordings of mutant NMJs show that evoked transmitter release is indeed altered upon closely spaced paired stimulation. At short intervals, paired-pulse facilitation is reduced non-additively, i.e. to a similar extent in both single and double mutants (Fig. 4E); given that paired-pulse facilitation is almost absent in the single mutants, baseline effects need to be considered for this measure of evoked release. However, the almost threefold increase in mini frequency in Sap-Syn double mutants (Fig. 4B) could result from an additive effect of the individual mutations. Note that a similar additivity, which would be inconsistent with an obligatory cooperative function of synapsin and Sap47, is observed in climbing deficiency. This raises the question as to whether there could be a functional link between increased mini frequencies, which are indicative of increased spontaneous release, and climbing defects.

The reduced synapsin levels in larval neuro-muscular boutons of Cirl mutants (Fig. 5) are difficult to interpret. Cirl codes for an adhesion-type G-protein-coupled receptor known for its interaction with α-Latrotoxin, which causes massive neurotransmitter release from Drosophila NMJs (Umbach et al., 1998). Drosophila Latrophilin/Cirl has recently been shown to modulate ionotropic receptor currents in mechanosensory neurons (Scholz et al., 2017) but is likely to have additional functions in the adult brain as its depletion by pan-neuronal knockdown induces hyperactivity and reduced sleep (van der Voet et al., 2016). Additional experiments are required to explain the present observations of reduced Cirl mRNA in synapsin mutants (‘rescued’ by the Sap^−/− mutation, Fig. S1) and the effect of Cirl knockout on synapsin levels.

The distribution of synapsin phosphorylated at Ser6 is highly intriguing (Fig. 4D and Fig. S5A,B). It now needs to be investigated to which cells the synapses that contain the extremely high concentrations of synapsin phosphorylated at Ser6 belong (Fig. 7C,D and Fig. S5A,B). Initial attempts to answer this question for the central complex using transgenic flies expressing GFP in subsets of ring neurons (Fig. S6) were inconclusive.

Although the current study provides the most detailed account of Sap47 expression and function to date, the need for further research is obvious. The present work may serve as guide for such continued efforts. In particular, it shows that Sap47 is found not only at synaptic boutons of larval motoneurons and in most neuropil regions of larval and adult nervous system, but it is also detected in axons and selectively in subsets of neuronal cell bodies. Lack of Sap47 leads to premature death, climbing defects, impaired behavioural plasticity, reductions in paired-pulse facilitation and causes modified phosphorylation of synapsin. In particular, the latter data suggest an integrated mode of action of these two highly abundant evolutionarily conserved neuronal proteins.

Stocks obtained from the Bloomington Drosophila Stock Center (NIH P40OD018537) were used in this study.