Distinct functions of a cGMP-dependent protein kinase in nerve terminal growth and synaptic vesicle cycling

Neurotransmission is essential for proper nervous system function. Chemical neurotransmission involves the fusion of synaptic vesicles (SVs) with the presynaptic plasma membrane and the release of neurotransmitter molecules (exocytosis). Subsequently, SV components are retrieved from the plasma membrane and SVs are re-formed (endocytosis). Maintaining a balance between SV exocytosis and endocytosis requires a tight coupling of these two processes, which is critical for sustained synaptic transmission (Haucke et al., 2011; Wu et al., 2014; Soykan et al., 2016). However, the mechanisms underlying this coupling have largely remained elusive. Presynaptic Ca²⁺ (Hosoi et al., 2009), and Synaptotagmin (Syt) (Poskanzer et al., 2003; Yao et al., 2012) and SNARE (Deák et al., 2004; Xu et al., 2013; Zhang et al., 2013) proteins are thought to be involved in coupling SV exocytosis and endocytosis. Many presynaptic proteins that regulate SV exocytosis and endocytosis undergo phosphorylation/dephosphorylation (Turner et al., 1999; Clayton et al., 2007; Kohansal-Nodehi et al., 2016). Changes in phosphorylation state can alter the affinity of proteins for some of their binding partners (Clayton et al., 2009; Cho et al., 2015; Geng et al., 2016), as well as alter the probability of ion channels opening (Huang and Zamponi, 2017). Kinases are thus compelling candidates for involvement in the coupling of SV exocytosis and endocytosis.

Kinases are multifaceted by nature and regulate a wide variety of cellular processes (Jordán-Álvarez et al., 2012; Chen et al., 2014; Piccioli and Littleton, 2014; Zhao et al., 2015; Spring et al., 2016; Krill and Dawson-Scully, 2016; Cantarutti et al., 2018). In the nervous system, a single kinase can affect multiple developmental and physiological processes. Consequently, pharmacological and genetic approaches have been unable to determine whether a given kinase affects the coupling of SV exocytosis and endocytosis or only one with concomitant effects on the other. To understand their role in synaptic function it is critical to be able to inactivate kinases with high spatiotemporal resolution.

Optogenetic approaches have recently been used to acutely activate kinases (Krishnamurthy et al., 2016; O'Banion et al., 2018). However, acute inactivation of an endogenously expressed kinase has not been achieved to date. Fluorescein-assisted light inactivation (FALI) has been used to acutely inactivate proteins at the Drosophila larval neuromuscular junction (NMJ) within minutes, and bypass any lethality and developmental effects of these proteins (Marek and Davis, 2002; Kasprowicz et al., 2008, 2014; Heerssen et al., 2008; Vanlandingham et al., 2014). Furthermore, the Drosophila dynamin [shibire (shi)] mutant, shibire-ts1, can be used to uncouple SV exocytosis and endocytosis; this mutant undergoes reversible, temperature-sensitive endocytic blockade and can be used to trap SVs on the plasma membrane following SV fusion (Koenig and Ikeda, 1989; Ramaswami et al., 1994). FALI has been used in conjunction with SV trapping to test the function of Syt1 in SV endocytosis following normal stimulus-dependent exocytosis (Poskanzer et al., 2003).

Here, we use this approach to investigate the functions of a cGMP-dependent protein kinase (PKG) at the synapse. PKGs are good candidates to be involved in coupling SV exocytosis and endocytosis because they are thought to affect both processes (Yawo, 1999; Luo et al., 2012; Eguchi et al., 2012; Taoufiq et al., 2013; Collado-Alsina et al., 2014). However, these two processes are intricately linked at the synapse, making it possible that the reported effects of a PKG on one process could be due to concomitant effects on the other. In addition, PKGs have also been reported to have a developmental role that involves regulating axon guidance and growth (Peng et al., 2016; Sild et al., 2016). Whether the effects of a PKG on SV exocytosis and endocytosis are interrelated or distinct from these developmental effects has not been determined. Furthermore, presynaptic, postsynaptic and glial PKGs may all contribute to these synaptic effects. Thus, a PKG at the synapse may have several distinct temporal and spatial functions. To investigate this, we tagged the Drosophila foraging (for) gene, which encodes a PKG orthologue of the mammalian Prkg1 gene (Osborne et al., 1997), for 4′,5′-bis[1,3,2-dithioarsolan-2-yl] fluorescein (FlAsH)-mediated FALI (FlAsH-FALI), and show that FOR does have multiple separable functions at the synapse.

The Drosophila larval NMJ is a well-established preparation for studying nerve terminal growth and SV cycling (Harris and Littleton, 2015). To determine whether FOR was expressed at the larval NMJ, we used recombineering to generate a bacterial artificial chromosome (BAC) containing the entire for locus with a flag-4c tag on the C-terminus (Fig. 1A). We expressed this BAC in a for null background. The C-terminus was tagged because it is common to all FOR isoforms (Allen et al., 2017; Fig. 1A). Western blot analysis showed that +;for⁰;{for^flag-4c} larvae had similar levels of FOR to wild-type larvae (Fig. 1B,C). To visualize FOR expression at the NMJ of muscle fibres 6 and 7, mid-third-instar larvae expressing FOR-flag-4c under endogenous for promoter control were stained with an anti-flag antibody. These muscles are innervated by two types of axons that result in type 1b and type 1s boutons, which differ in both their morphological and physiological properties (Kurdyak et al., 1994). We found FOR enriched in both 1b and 1s boutons (Fig. 1D,F) and colocalized with the presynaptic cytoplasmic protein Complexin (Cpx) (Fig. 1G) and the plasma membrane marker horseradish peroxidase (HRP) (Fig. 1H). No obvious signal was observed in muscles (Fig. 1D). As expected, control larvae without the flag tag showed no anti-flag staining (Fig. 1E). FOR was also present in glia (Fig. 1D), and this signal overlapped with GFP driven by a glial driver, 46F-GAL4 (Brink et al., 2012; Fig. 1I). To further confirm that FOR is expressed in glia at the larval NMJ we examined several promoter-driven for-GAL4 drivers (Allen et al., 2018). Four GAL4 lines were previously generated by cloning regions surrounding the four transcription start sites of the for gene (Allen et al., 2018). We found that one expressed strongly in glia at the NMJ (Fig. S1A,B) and this was confirmed by co-staining for the glial marker Gliotactin (Gli) (Fig. S1C). No presynaptic bouton expression was seen with any of the promoter-driven for-GAL4 drivers. This is likely because the cloned regions used for the promoter-driven for-GAL4 drivers likely do not encapsulate all of the regulatory elements from the for locus, as together they covered less than half of the gene. Collectively, these data show that the Drosophila NMJ is an ideal preparation for dissecting the role(s) of FOR in nerve terminal growth and SV cycling, as FOR is found in presynaptic boutons and glia.

To investigate the role of FOR in nerve terminal growth, we stained NMJs from muscle fibres 6 and 7 of abdominal segment 3 from mid-third-instar larvae with anti-HRP (a neuronal membrane marker) and anti-Bruchpilot (BRP; an active zone marker; Wagh et al., 2006) antibodies. The number of 1b and 1s boutons was significantly increased in the for⁰ null mutant in comparison to control and genomic rescue larvae (Fig. 2A,B). The number of active zones per bouton, estimated by counting the number of BRP spots from a sample of 1b boutons (∼5μm in diameter) and 1s boutons (∼2μm in diameter), was not different in the for⁰ null mutant in comparison to control and genomic rescue larvae (Fig. 2C,D). We conclude from these data that FOR negatively regulates nerve terminal growth at the NMJ, but does not affect the number of active zones per bouton.

To determine whether presynaptic, postsynaptic (muscle) or glial FOR was required for nerve terminal growth, we used the GAL4/UAS system (Brand and Perrimon, 1993) to perform tissue-specific rescues. To accomplish this, we generated transgenic flies to express FOR with a flag-4c tag on the C-terminus (UAS-for-flag-4c) in selective tissues. When UAS-for-flag-4c was driven by the ubiquitous GAL4 driver, da-GAL4, FOR expression was increased (Fig. S2A). Using tissue-specific GAL4s, we selectively drove FOR expression presynaptically (n-syb-GAL4; Fig. S2B,C), postsynaptically (24B-GAL4; Fig. S2D) or in glia (Repo-GAL4; Fig. S2E) at the larval NMJ. We defined a rescue as being not significantly different from wild type, but significantly different from larvae that express the GAL4 and larvae that express the UAS transgene in a for⁰ null mutant background. Expression of FOR in glia in a for⁰ null mutant background was sufficient to rescue the increased nerve terminal growth seen in for⁰ null mutants (Fig. 2E,F). In contrast, presynaptic (Fig. 2G) or postsynaptic (Fig. 2H) expression of FOR did not rescue the increased nerve terminal growth seen in for⁰ null mutants. Together, these data show that FOR acts in glia to negatively regulate nerve terminal growth at the NMJ.

Glia at the Drosophila NMJ are known to release Maverick (Mav), a TGF-β ligand that regulates nerve terminal growth (Fuentes-Medel et al., 2012). We hypothesized that FOR functions through Mav to regulate nerve terminal growth. We knocked down glial Mav in a for⁰ null mutant background and found that this rescued the increased nerve terminal growth seen in for⁰ null mutants (Fig. 3A,B). Mav is thought to control the release of the Drosophila TGF-β/BMP homologue, Glass bottom boat (Gbb), from muscles, which fine-tunes the ability of motor neurons to extend new synaptic boutons (Fuentes-Medel et al., 2012). We hypothesized that if FOR was acting through Mav, then knocking down muscle Gbb would also rescue the increased nerve terminal growth seen in for⁰ null mutants. As we predicted, knocking down Gbb in muscles rescued the increased nerve terminal growth seen in for⁰ null mutants (Fig. 3C,D). We also reduced expression of Gbb in glia and found that it did not rescue the increased nerve terminal growth seen in for⁰ null mutants (Fig. S3A,B). Thus, FOR acts in glia to negatively regulate nerve terminal growth, likely through a pathway that involves glial Mav and muscle Gbb.

We next assessed the role of FOR in neurotransmitter release by recording the compound excitatory junction potential (EJP) generated by tonic-like type 1b and phasic-like type 1s boutons from segment 3 of muscle fibre 6 in mid-third-instar larvae in HL6 saline (0.5 mM Ca²⁺) in response to low-frequency stimulation (0.05 Hz). We found a significant increase in EJP amplitude in for⁰ null mutants compared to control and genomic rescue larvae (Fig. 4A,B), showing that FOR affects evoked neurotransmitter release. There was no significant difference in miniature excitatory junction potential (mEJP) amplitude (Fig. 4A,C). There was a mild increase in mEJP frequency; however, this increase was not significantly different in comparisons to controls (Fig. 4A,D). Thus, FOR does not appear to affect spontaneous neurotransmitter release. At higher external Ca²⁺ concentrations (1 mM Ca²⁺), the EJP amplitude in for⁰ null mutants was significantly larger compared to control and genomic rescue larvae (Fig. 4H,I). However, the effects were more severe at lower Ca²⁺ concentrations (0.5 mM Ca²⁺), demonstrating a Ca²⁺ dependence of the effects of FOR on evoked neurotransmitter release.

The enhancement of evoked neurotransmitter release in the for⁰ null mutant was rescued by presynaptic expression of FOR (Fig. 4E). In contrast, neither postsynaptic (Fig. 4F) nor glial (Fig. 4G) expression of FOR rescued the increased neurotransmitter release seen in the for⁰ null mutant. These results demonstrate a presynaptic role for FOR in neurotransmission. Specifically, presynaptic FOR negatively regulates evoked neurotransmitter release, but does not appear to regulate spontaneous neurotransmitter release.

PKG has been shown to regulate intracellular Ca²⁺ in neuronal and non-neuronal tissues (Ruth et al., 1993; Meriney et al., 1994; Schlossmann et al., 2000). We hypothesized that the increased neurotransmitter release seen in for⁰ null mutants in response to low-frequency stimulation may be due to changes in presynaptic Ca²⁺ levels. To test this hypothesis, we assayed presynaptic Ca²⁺ signals evoked by single action potentials at high temporal resolution using Oregon Green 488 BAPTA-1 conjugated to 10-kDa Dextran (OGB-1). This approach is sensitive enough to detect changes in presynaptic Ca²⁺ in response to single pulses (Macleod et al., 2002). Nerve terminals were forward filled with the indicator and a single pulse was delivered to the motor axons. Signals from single type 1b boutons at the ends of nerve branches were analysed. Absolute fluorescence prior to stimulation was not significantly different between genotypes [control, 296.60±35.59 arbitrary units (a.u.); for⁰ null, 288.89±28.84 a.u.; genomic rescue, 292.29±80.08 a.u.]. We found that peak amplitudes of Ca²⁺ signals in response to single pulses were significantly enhanced by 40% in the for⁰ null mutant compared to the control and the genomic rescue (Fig. 4J,K). This increase was large enough to account for the increased neurotransmitter release seen in for⁰ null mutants (see Discussion). Overall, FOR inhibits evoked neurotransmitter release by negatively regulating both nerve terminal growth and presynaptic Ca²⁺ levels during low-frequency stimulation.

We tested the ability of for⁰ null mutants to maintain neurotransmitter release during high-frequency stimulation. There were no significant differences in the initial EJP amplitudes between genotypes in saline with 2 mM Ca²⁺ (data not shown). EJP amplitudes in for⁰ null mutants displayed a significant decline when stimulated at 10 Hz for 5 min in 2 mM Ca²⁺ in comparison to control and rescue larvae (Fig. 5A,B). We hypothesized that the underlying mechanism for the depression in neurotransmitter release seen in for⁰ null mutants during high-frequency stimulation could be a defect in SV recycling. We examined the role of for in SV recycling using our mutant and rescue genotypes. To accomplish this, we used the lipophilic styryl dye FM1-43 (Betz and Bewick, 1992) to assess SV cycling, measuring FM1-43 uptake by stimulating preparations with high K⁺ saline for 2 min in the presence of FM1-43. We found a significant impairment in FM1-43 uptake in for⁰ null mutants in comparison to control and genomic rescue larvae (Fig. 5C,D). To determine whether this impairment was due to a defect in SV exocytosis, we next measured FM1-43 unloading by applying high K⁺ saline for 2 min and calculating the released fraction. We found that for⁰ null mutants released a similar fraction of FM1-43 in comparison to control and genomic rescue lines (Fig. 5C,E), demonstrating that recycled SVs could be released. Consequently, the defects in FM1-43 uptake are likely attributable to an impairment in SV endocytosis.

It is unknown whether presynaptic, postsynaptic or glial PKG regulates SV endocytosis. We used tissue-specific rescues to determine whether FOR was required for SV endocytosis presynaptically, postsynaptically (muscle) or in glia. Presynaptic expression of FOR rescued the reduction in FM1-43 uptake in for⁰ null mutant backgrounds (Fig. 5F,G), whereas postsynaptic or glial expression of FOR did not rescue the impaired FM1-43 uptake of for⁰ null mutants (Fig. 5I,K). The released fraction was similar between all genotypes (Fig. 5F,H,J,L). From these data, we conclude that presynaptic FOR is required for SV endocytosis.

Effects on synaptic transmission and nerve terminal growth have been shown in some cases to be interrelated (Budnik et al., 1990; Stewart et al., 1996; Davis and Goodman, 1998). However, these processes are not always tightly linked (Romero-Pozuelo et al., 2007, 2014; Dason et al., 2009). We next tested whether the increased neurotransmitter release in the for⁰ null mutant (Fig. 4A,B) could occur independently of changes in nerve terminal growth. To separate the effect of FOR on nerve terminal growth from its effect on synaptic transmission, we acutely inactivated FOR using FlAsH-FALI (Marek and Davis, 2002). This technique is based on the use of a tetracysteine epitope tag (4c) that binds to membrane-permeable FlAsH-EDT₂. When excited with light of the appropriate wavelength, FlAsH-EDT₂ bound to the 4c tag inactivates the tagged protein with a half-maximal distance of ∼40 Å through the generation of reactive oxygen species (Beck et al., 2002). FlAsH-FALI has been shown to be very selective in inactivating only the tagged protein. For example, photoinactivation of dynamin or its binding partner EndophilinA results in different endocytic defects, thereby demonstrating the specificity of FlAsH-FALI (Kasprowicz et al., 2014). When we expressed FOR-flag-4c, under the control of endogenous promoters in a for⁰ mutant background, we rescued both the increased nerve terminal growth (Fig. 6A,B) and increased synaptic transmission (Fig. 6C,D) previously seen in the for⁰ null mutant (Figs 2B and 4B). Thus, FOR-flag-4c recapitulates FOR function for both nerve terminal growth and synaptic transmission. We next investigated whether acute photoinactivation of FOR affected synaptic transmission. We incubated NMJs with FlAsH-EDT₂, then washed unbound FlAsH-EDT₂ and illuminated with epifluorescent light filtered at 488 nm (see Materials and Methods for details). In controls, incubation with FlAsH-EDT₂ and illumination with light had no effect on EJP amplitude (Fig. 6C,D). However, the amplitude of evoked EJPs in response to low-frequency stimulation (0.05 Hz) of larvae expressing FOR-flag-4c in for⁰ null mutants was significantly enhanced following light inactivation compared to controls (Fig. 6C,D). The increase in synaptic transmission occurs in a timeframe that is too fast for changes in nerve terminal growth to occur (Ataman et al., 2008). This demonstrates that the effect of FOR on synaptic transmission can occur without its effect on nerve terminal growth.

Increased neurotransmitter release can lead to depletion of the readily releasable pool of SVs (Zucker and Regehr, 2002) and this could contribute to the increased depression seen in for⁰ null mutants during high-frequency stimulation (Fig. 5A,B). Therefore, we next investigated whether the effect of FOR on SV recycling was separate from its effect on neurotransmitter release. In order to test the function of FOR during endocytosis without affecting FOR during exocytosis, we used a shi temperature-sensitive dynamin mutant to deplete the entire SV pool and thereby uncouple exocytosis and endocytosis. Prolonged stimulation of shi mutant NMJs at nonpermissive temperature (30°C) blocks SV recycling, which leads to loss of SVs (Macleod et al., 2004; Dason et al., 2010) and neurotransmitter release (Delgado et al., 2000; Dason et al., 2010). This block is reversible and SVs reform when returned to permissive temperatures (22°C). This block does not occur at wild-type NMJs (Dason et al., 2010). We recorded EJPs in response to 0.05 Hz stimulation at 22°C to assess synaptic transmission prior to SV depletion, followed by a stimulus train (10 Hz, 12 min) at 30°C designed to deplete the terminals of SVs. We then assessed recovery of synaptic transmission by recording EJPs in response to 0.05 Hz stimulation at 22°C for the subsequent 20 min (Fig. S4A,B). Consistent with previous studies, we found that synaptic transmission in shi mutants was progressively impaired at the restrictive temperature (30°C), but recovered after return to the permissive temperature (Delgado et al., 2000; Dason et al., 2010). When we expressed FOR-flag-4c in a shi, for⁰ double-mutant background, there was a gradual loss of EJP amplitude during stimulation at the restrictive temperature and recovery of synaptic transmission after return to the permissive temperature (Fig. S4A,B). Recovery of synaptic transmission was similar between shi mutants with wild-type FOR and shi mutants that expressed FOR-flag-4c in a for⁰ mutant background, demonstrating that FOR-flag-4c can recapitulate FOR function during SV recycling.

We then modified our shi protocol to use FlAsH-FALI to test the effects of acutely photoinactivating FOR after normal stimulus-dependent exocytosis (Fig. 7A). We recorded EJPs in response to 0.05 Hz stimulation at 22°C to assess synaptic transmission prior to SV depletion, followed by a stimulus train (10 Hz, 12 min) at 30°C designed to deplete the terminals of SVs. We kept both control (shi;+;+) and experimental (shi;for⁰;{for^flag-4c}) NMJs at 30°C for an additional 18 min during our FlAsH-FALI protocol (9 min incubation with FlAsH-EDT₂, followed by 4 min wash with BAL buffer to remove unbound FlAsH, and then illumination with epifluorescent light filtered at 488 nm for 5 min to photoinactivate FOR). Recovery of synaptic transmission was then assessed by recording EJPs in response to 0.05 Hz stimulation at 22°C. Recovery of synaptic transmission in control larvae was similar to recovery in the absence of FlAsH-EDT₂ (Fig. 7B,C), whereas acute photoinactivation of FOR resulted in a significant reduction in the recovery of synaptic transmission after return to the permissive temperature compared to controls (Fig. 7B,C). This demonstrates that acute inactivation of FOR reduces the recovery of synaptic transmission, likely by impairing SV endocytosis.

To determine whether the impairment in recovery of synaptic transmission following the acute inactivation of FOR was due to reduced SV endocytosis, we assessed SV recycling using the lipophilic styryl dye FM4-64 (Fig. 7D). We used FM4-64 because the excitation and emission spectra of FM1-43 overlap with those of FlAsH-EDT₂ more extensively than those of FM4-64. Control (shi;+;+) and experimental (shi;for⁰;{for^flag-4c}) NMJs were stimulated with high K⁺ saline for 10 min at 30°C to trap SVs, followed by FlAsH-FALI at 30°C (9 min incubation with FlAsH-EDT₂, 4 min wash with BAL buffer to remove unbound FlAsH, and then illumination with epifluorescent light filtered at 488 nm for 5 min to photoinactivate FOR). SV endocytosis was then assayed by incubating preparations with FM4-64 for 20 min at 22°C, followed by a 10-min wash to remove extracellular FM4-64. In the absence of FlAsH-FALI, we found no significant difference in the amount of FM4-64 uptake between control (shi;+;+) and experimental (shi;for⁰;for^flag-4c) NMJs (Fig. S4C,D). FM4-64 uptake in control NMJs subjected to the FlAsH-FALI protocol was similar to FM4-64 uptake in the absence of FlAsH-EDT₂ (Fig. 7E,F). In contrast, acute photoinactivation of FOR resulted in a significant reduction in FM4-64 uptake (Fig. 7E,F). The released fraction of FM4-64 following high K⁺ stimulation was similar for all genotypes (Fig. 7G; Fig. S4E), demonstrating that recycled SVs could be released. Overall, the FM4-64 data are consistent with our electrophysiology data and clearly demonstrate that acute inactivation of FOR, following normal-stimulus dependent exocytosis, results in impaired recovery of synaptic transmission due to reduced SV endocytosis.

Our study is the first to experimentally separate the exocytic and endocytic functions of a PKG in neurotransmission. We show that, during periods of low synaptic activity, presynaptic FOR negatively regulates neurotransmitter release by inhibiting presynaptic Ca²⁺ levels. During periods of high synaptic activity, presynaptic FOR maintains neurotransmission by facilitating SV recycling. In addition, we show that glial FOR negatively regulates nerve terminal growth through Mav and Gbb signalling, and this effect is likely distinct from its effects on neurotransmitter release and SV recycling.

We found that for⁰ null mutants have increased nerve terminal growth. PKG has previously been shown to regulate axon guidance and synaptogenesis (Peng et al., 2016; Sild et al., 2016). Expression of a dominant-negative form of PKG in Xenopus radial glia impairs glial motility and synaptogenesis (Sild et al., 2016). Here, we show that glial FOR negatively regulates nerve terminal growth, whereas presynaptic and postsynaptic FOR do not regulate nerve terminal growth. Drosophila glia regulate nerve terminal growth by releasing the TGF-β ligand, Mav, to modulate TGF-β/BMP retrograde signalling, which leads to the addition of new synaptic boutons (Fuentes-Medel et al., 2012). Glial Mav regulates nerve terminal growth through muscle-derived Gbb, the Drosophila TGF-β/BMP homologue (Fuentes-Medel et al., 2012). We found that glial-specific knockdown of Mav or muscle-specific knockdown of Gbb rescues the nerve terminal overgrowth phenotype of for⁰ null mutants. This suggests that FOR acts through Mav and Gbb to negatively regulate nerve terminal growth, although it remains possible that FOR functions in a parallel pathway to Mav and Gbb. Interestingly, FOR also regulates TGF-β/BMP signalling in the developing wing of Drosophila (Schleede and Blair, 2015). PKG regulates intracellular Ca²⁺ levels in rat glial cell cultures (Willmott et al., 2000). Thus, it is tempting to speculate that FOR can affect Mav secretion by regulating Ca²⁺ levels in glia. However, currently, little is known about the mechanisms underlying Mav secretion from glia.

PKG has been implicated in regulating neurotransmitter release (Yawo, 1999; Wei et al., 2002; Luo et al., 2012). However, these studies have been contradictory, suggesting both a stimulatory and inhibitory role. Reconciling these findings, we show that FOR plays an inhibitory role in neurotransmitter release during low-frequency stimulation and a facilitatory role in SV recycling that can maintain sustained neurotransmitter release during high-frequency stimulation. Interestingly, the frequency of spontaneous neurotransmitter release is not altered in for⁰ null mutants (Fig. 4D), despite for⁰ null mutants having more synaptic boutons, and thus more active zones, per NMJ. Release probability for evoked and spontaneous neurotransmitter release at individual active zones at the Drosophila larval NMJ are not correlated, suggesting that they are independently regulated (Melom et al., 2013). Thus, FOR may be having a selective effect on active zones involved in evoked neurotransmitter release. Alternatively, some kind of homeostasis (Stewart et al., 1996; Davis and Goodman, 1998) may be occurring in the for⁰ null mutant to compensate for the increased nerve terminal growth.

The increased neurotransmitter release seen in the for⁰ null mutant is likely due to both increased nerve terminal growth and increased presynaptic Ca²⁺ entry. PKG has been reported to affect Ca²⁺ levels and currents in several tissues (Ruth et al., 1993; Meriney et al., 1994; Schlossmann et al., 2000). However, studies examining the effects of PKG on neurotransmitter release did not correlate these changes with changes in presynaptic Ca²⁺ signals (Gray et al., 1999; Yawo, 1999; Luo et al., 2012). We found that the for⁰ null mutant had significantly increased presynaptic Ca²⁺ signals evoked by single action potentials. Neurotransmitter release varies as a fourth-order function of Ca²⁺ (Dodge and Rahamimoff, 1967; Augustine and Charlton, 1986). Assuming a fourth-power relationship, the increased amplitude of Ca²⁺ signals observed in the absence of FOR was sufficient to account for the twofold increase in neurotransmitter release.

The most plausible mechanistic explanations for the increased presynaptic Ca²⁺ levels are direct or indirect regulation of Ca²⁺ channels by FOR. A recent study reported that PKG downregulates L-type Ca²⁺ channel activity in HEK-293 cells through phosphorylation of two serine residues in the intracellular loop of the pore-forming Ca_vα1 subunit (Sandoval et al., 2017). Ca²⁺ channels found in presynaptic boutons at the Drosophila larval NMJ are likely N- or P/Q-type (Rieckhof et al., 2003; Macleod et al., 2006). PKG agonists inhibit N-type Ca²⁺ channel currents in dorsal root ganglion neurons (Yoshimura et al., 2001); thus, it remains possible that a similar type of regulation by PKG occurs for N- or P/Q-type channels. Alternatively, the effects of FOR on presynaptic Ca²⁺ could be a consequence of the modulation of K⁺ channels. PKG phosphorylates Ca²⁺-activated K⁺ channels (Alioua et al., 1998; Fukao et al., 1999; Zhou et al., 2001; Ferreira et al., 2015) and PKG inhibitors reduce Ca²⁺-activated K⁺ channel activity (Klyachko et al., 2001). Interestingly, cultured Drosophila neurons of an allelic variant of for with higher PKG activity displayed increased voltage-dependent K⁺ currents (Renger et al., 1999). A reduction in K⁺ currents in for⁰ null mutants could increase action potential duration, leading to increased presynaptic Ca²⁺ entry and increased neurotransmitter release.

Neurotransmitter release was significantly reduced in for⁰ null mutants during high-frequency stimulation (Fig. 5B). The inability of for⁰ null mutants to maintain normal levels of neurotransmitter release during high-frequency stimulation is mechanistically consistent with a defect in SV recycling. Pharmacological inhibition (Petrov et al., 2008; Eguchi et al., 2012; Taoufiq et al., 2013) and RNA interference (RNAi) knockdown of PKG (Collado-Alsina et al., 2014) reduces SV endocytosis. However, these studies did not experimentally separate exocytic and endocytic functions of PKG. Here, we used FlAsH-FALI in conjunction with a temperature-sensitive shi mutant to tease apart the role of FOR in SV exocytosis and endocytosis by dissociating these two processes. We used FlAsH-FALI to photoinactivate FOR following SV exocytosis but prior to SV endocytosis, and then assayed SV recycling using electrophysiology and FM4-64 imaging. We depleted the entire pool of SVs and then tested the role of FOR in SV endocytosis. By temporally uncoupling SV exocytosis and endocytosis, and then acutely inactivating FOR, we show that FOR is necessary for efficient endocytosis of SVs that have undergone exocytosis using a functional FOR protein. This approach can be modified to study the exocytic and endocytic functions of other kinases in mammalian preparations by using FALI in conjunction with dynamin inhibitors, such as dynasore (Macia et al., 2006). In addition, we found that presynaptic, but not postsynaptic or glial, FOR was required for SV recycling. Contributions of presynaptic, postsynaptic and glial PKG were not addressed in previous studies. PKG is thought to function at a step between the upregulation of cGMP and phosphatidylinositol 4,5-bisphosphate (PIP₂) levels through a Rho kinase (Taoufiq et al., 2013). Thus, the effects of FOR on SV endocytosis may be through PIP₂ signalling and the recruitment of AP-2 and Clathrin proteins to sites of endocytosis.

Kinases have multiple functions and are often thought of as pleiotropic. Kinases can phosphorylate multiple targets, be expressed in different cells, have different subcellular expression patterns and be temporally activated at different times, all of which are means of pleiotropy. FOR has been implicated in behavioural pleiotropy (Allen et al., 2017; Anreiter et al., 2017; Allen et al., 2018; Anreiter and Sokolowski, 2018). Here, we show distinct spatial and temporal requirements for FOR at the synapse. Glial FOR is required for the developmental effects of FOR on nerve terminal growth, whereas presynaptic FOR is required for the acute effects of FOR on neurotransmitter release and SV recycling. Thus, the developmental effects of FOR are likely independent from its effects on SV exocytosis and endocytosis. Furthermore, we show that the effects of FOR on neurotransmitter release and SV recycling can be temporally separated. We conclude that FOR plays critical and distinct roles in regulating nerve terminal growth and coupling SV exocytosis and endocytosis. Our work highlights the multifaceted nature of kinases and demonstrates the importance of temporally and spatially targeting them when studying their functions.

Fly stocks were grown in uncrowded conditions at 25°C on molasses-based fly medium (Anreiter et al., 2016). Mid-third-instar larvae were used for all experiments. A sitter strain (+;+;+) was used as a wild-type control and all lines were backcrossed to it (Allen et al., 2017). The for⁰ null mutant contains a 35 kb deletion that removes the entire for locus (Allen et al., 2017). {for^wt} is a genomic rescue inserted on the third chromosome that contains a BAC with the entire for locus (Allen et al., 2017). {for^flag-4c} is a genomic rescue generated in this study that contains a BAC with the entire for locus with a flag tag and a tetracysteine (4c) tag on the C-terminus. The GAL4/UAS system (Brand and Perrimon, 1993) was used for tissue-specific expression of transgenes. n-syb-GAL4 was used to drive expression in neurons (Verstreken et al., 2009). 24B-GAL4 was used to drive expression in muscles (Sweeney et al., 1995). Repo-GAL4 or 46F-GAL4 was used to drive expression in glia (Sepp et al., 2001; Brink et al., 2012). da-GAL4 was used as a ubiquitous driver (Wodarz et al., 1995). for^pr3-GAL4 was used to drive expression of UAS-mCD8-GFP in a subset of cells that express for (Allen et al., 2018). UAS-mav-RNAi was used to knock down mav (Fuentes-Medel et al., 2012). UAS-gbb-RNAi was used to knock down gbb (Tian and Jiang, 2014). UAS-for-flag-4c was generated in this study and used to express FOR with a flag-4c tag on the C-terminus in selective tissues. The temperature-sensitive shi mutant (Grigliatti et al., 1973) was used to reversibly block SV recycling.

Recombineering-mediated tagging was performed as previously described (Venken et al., 2008). A modified P(acman) vector containing a full-length 35 kb genomic fragment of the for gene was used (Allen et al., 2017). A PCR fragment coding for flag and a tetracysteine cassette (4c) was amplified from a PL-452 C-flag-4c plasmid (Addgene plasmid #19180; Venken et al., 2008) using the forward primer 5′-ATACAACAAACTTTGATGACTATCCTCCCGATCCTGAGGGTCCGCCGCCAGATGATGTCACTGGATGGGACAAGGACTTCGGTCCCTCGAGGGGATCCATGG-3′ and the reverse primer 5′-TGGCTTTCATATCCTCCTGGTTTTCTGCAGCAGAAGCGTTTAGAGAGCATCGTCTAGGAAACGGGTTCTGATTCTCCTCATTAGGGCTCCATGCAGCAG-3′. The fragment was inserted onto the C-terminal end of for using recombineering (Venken et al., 2008) and a galk selection method (Warming et al., 2005). The BAC was incorporated into the fly's genome using φC31 integration into the attP2 landing site on the third chromosome. Transgenesis was performed by BestGene Inc. (Chino Hills, CA).

The open reading frame (ORF) of for P1 was amplified from a pUASTattB-P1 vector with the forward primer 5′-GCGGCCGCATGCGTTTCTGCTTTGATCG-3′ and the reverse primer 5′-CCATGGATCCCCTCGAGGGACCGAAGTCCTTGTCCCATCCAGTGAC-3′. A PCR fragment coding for flag and a tetracysteine cassette (4c) was amplified from a PL-452 C-flag-4c plasmid (Addgene plasmid #19180; Venken et al., 2008) using the forward primer 5′-GGTCCCTCGAGGGGATCCATGG-3′ and the reverse primer 5′- TATGCGGCCGCTTAGGGCTCCATGCAGCAG-3′. A fusion PCR protocol was used to generate a for-flag-4c construct (Hobert, 2002). for-flag-4c was then cloned into the pUASTattB vector (Bischof et al., 2007). The pUASTattB vector was incorporated into the fly's genome using φC31 integration into the attP2 landing site on the third chromosome. Transgenesis was performed by BestGene Inc. (Chino Hills, CA).

Western blots were performed as previously described (Allen et al., 2017). Briefly, samples were resolved by SDS-PAGE and blots were incubated with a guinea pig polyclonal anti-FOR antibody (1:5000) (Belay et al., 2007) or a mouse monoclonal anti-actin antibody (MAB1501, Sigma-Aldrich; 1:20,000) for 1 h at room temperature. The following day, the membrane was washed, incubated with the secondary antibody for 1 h at room temperature, washed again, and immunoreactivity was then detected using GE Healthcare Amersham ECL Prime Detection Reagent.

Immunohistochemistry was performed as previously described (Dason et al., 2014). Briefly, fixed preparations were incubated overnight at 4°C with fluorescein isothiocyanate (FITC)-conjugated anti-HRP (1:800; Jackson ImmunoResearch, USA) to visualize neurons, mouse monoclonal anti-BRP (1:100; Developmental Studies Hybridoma Bank, Iowa City, IA; Wagh et al., 2006) to visualize active zones, rabbit anti-Cpx (Huntwork and Littleton, 2007), mouse anti-Gli (Schulte et al., 2003), polyclonal rabbit anti-DSYT2 (also known as Syt1) (1:800; Littleton et al., 1993), mouse monoclonal anti-GFP (1:500; 3E6, Thermo Fisher Scientific), chicken monoclonal anti-GFP (1:500; AA10262, Thermo Fisher Scientific) or mouse monoclonal anti-Flag M2 (1:500; F1804, Sigma-Aldrich; Hampel et al., 2011). All primary antibodies were diluted in blocking solution. Preparations were mounted in Permafluor (Immunon, Pittsburgh, PA) on a glass slide with a cover slip and viewed under a TCS SP5 confocal laser-scanning microscope (Leica, Heidelberg, Germany) with a 63× oil-immersion objective (1.4 NA).

All electrophysiological and imaging experiments were conducted in the haemolymph-like solution HL6 (Macleod et al., 2002). CaCl₂ concentrations are indicated for each experiment.

Intracellular recordings were performed as previously described (Dason et al., 2009). Briefly, the ventral longitudinal muscle fibre 6 (abdominal segment 3) of dissected larvae was impaled with a sharp glass electrode filled with 3 M KCl (∼40 MΩ) to measure spontaneously occurring mEJPs and stimulus-evoked EJPs. All preparations tested had a resting membrane potential of −60 mV to −65 mV. Cut segmental nerves were stimulated at either 0.05 Hz or 10 Hz using a suction electrode. Electrical signals were recorded using the MacLab/4S data acquisition system (ADInstruments).

Motor neurons were forward filled with the Ca²⁺ indicator OGB-1 as previously described (Macleod et al., 2002). The relatively fast Ca²⁺-binding and -unbinding kinetics of OGB-1 allowed us to measure Ca²⁺ signals in response to single pulses. All experiments with OGB-1 were performed on a Leica TCS SP5 confocal laser-scanning microscope with a 63× water-dipping objective (0.9 NA). Cut segmental nerves were stimulated with single pulses using a suction electrode. A line-scanning method (4 ms per scan) was used to follow Ca²⁺ responses to single pulses in a single 1b bouton with high temporal resolution. Fluorescence (F) was reported using the equation ΔF/F_rest=(F_stim– F_rest)/F_rest.

FM1-43 experiments were performed using a Leica TCS SP5 confocal laser-scanning microscope with a 63× water-dipping objective (0.9 NA) as previously described (Dason et al., 2010). Briefly, FM1-43 loading and unloading was induced by high K⁺ depolarization using the following high K⁺ saline: 25 mM NaCl, 90 mM KCl, 10 mM NaHCO₃, 5 mM HEPES, 30 mM sucrose, 5 mM trehalose, 10 mM MgCl₂, 2 mM CaCl₂, pH 7.2 (Verstreken et al., 2008). Preparations were loaded with FM1-43 (Invitrogen) by high K⁺ depolarization for 2 min. Preparations were extensively washed for 5 min in Ca²⁺-free HL6. To reduce background fluorescence from extracellular FM1-43, 75 µM Advasep-7 was included for the first 1 min of the wash (Kay et al., 1999). After an image of FM1-43 uptake was taken, high K⁺ depolarization for 2 min was used to induce exocytosis. Another image was then taken to document FM1-43 unloading. Fluorescence was measured and averaged from three 1b boutons per image. The released fraction was calculated using the following formula: (fluorescence of load–fluorescence of unload)/fluorescence of load.

FM4-64 experiments were performed using a Leica TCS SP5 confocal laser-scanning microscope with a 63× water-dipping objective (0.9 NA) as previously described (Dason et al., 2010). Briefly, preparations were stimulated with the following high K⁺ saline at 30°C to trap SVs: 25 mM NaCl, 90 mM KCl, 10 mM NaHCO₃, 5 mM HEPES, 30 mM sucrose, 5 mM trehalose, 10 mM MgCl₂, 2 mM CaCl₂, pH 7.2 (Verstreken et al., 2008). Preparations were then loaded with 10μM FM4-64 (Invitrogen) for 20 min at 22°C. Preparations were extensively washed for 10 min in Ca²⁺-free HL6. To reduce background fluorescence from extracellular FM4-64, 75 µM Advasep-7 was included for the first 2 min of the wash (Kay et al., 1999). After an image of FM4-64 uptake was taken, high K⁺ depolarization for 10 min was used to induce exocytosis. Another image was then taken to document FM4-64 unloading. The released fraction was calculated using the following formula: (fluorescence of load–fluorescence of unload)/fluorescence of load.

FlAsH-FALI experiments were performed as previously described (Habets and Verstreken, 2011) using a TC-FlAsH™ II In-Cell Tetracysteine Tag Detection Kit (Green Fluorescence; Thermo Fisher Scientific). Briefly, dissected third-instar larvae were incubated with 1 µM FlAsH-EDT₂ in HL6 for 9 min and then washed with BAL buffer for 4 min to remove unbound FlAsH. Muscle fibres 6 and 7 were illuminated with epifluorescent light filtered at 488 nm for 5 min to photoinactivate FOR.

SigmaPlot (version 11.0; Systat Software) was used for statistical analysis. Sample sizes were chosen based on what is appropriate and accepted in this field of research. Unpaired t-tests were used for comparing datasets of two groups and one-way ANOVA tests (with a Holm–Sidak post hoc test) were used for comparing datasets of more than two groups. Error bars in all figures represent ±s.e.m.

We thank Drs Vanessa Auld, Vivian Budnik and Troy Littleton for fly stocks and antibodies, and Drs Harold Atwood, Joel Levine and Ina Anreiter for comments on an early draft of the manuscript.