Developmental arrest of Drosophila larvae elicits presynaptic depression and enables prolonged studies of neurodegeneration

Synapses are confronted with extensive challenges during development, maturation and aging, yet maintain stable information exchange. The dynamic and at times massive changes in synapse growth, pruning and remodeling, coupled with intrinsic adjustments in neuronal excitability, can lead to unstable physiological activity. The resulting imbalances in excitation and inhibition would propagate within neural circuits to undermine network stability. To adapt to such challenges, synapses are endowed with the capacity to homeostatically adjust neurotransmission while still permitting the flexibility necessary for Hebbian forms of plasticity (Pozo and Goda, 2010; Turrigiano, 2012). The homeostatic control of neural activity operates throughout the organismal lifespan to balance the tension between stability and flexibility, and is thought to break down in neurological and psychiatric diseases (Li et al., 2019; Nelson and Valakh, 2015; Wondolowski and Dickman, 2013). Although it is clear synapses have the capacity to express both Hebbian and homeostatic forms of plasticity, how these processes are integrated and balanced, particularly during development and aging, remain enigmatic.

The Drosophila larval neuromuscular junction (NMJ) is an accessible and versatile model for studying synaptic function, plasticity and disease. This model glutamatergic synapse has enabled fundamental insights into synaptic growth, transmission, homeostatic plasticity and injury (Frank et al., 2020; Menon et al., 2013). However, studies in this system are limited by the relatively short larval period of 3-4 days before pupariation, when NMJ accessibility is lost. This short temporal window limits the use of the third instar larval NMJ as a model for interrogating dynamic processes over chronic timescales. Intriguingly, recent studies on the signaling cascades in Drosophila that control the transition from third instar to the pupal stage have revealed attractive targets for extending the duration of the third instar stage (Gibbens et al., 2011; Rewitz et al., 2009; Walkiewicz and Stern, 2009).

Developmental progression in Drosophila larvae is coordinated through two semi-redundant signaling pathways via Torso and insulin-like receptors that ultimately lead to ecdysone synthesis and release from the prothoracic gland (PG) to drive the transition from the larval stage to pupation (Rewitz et al., 2009; Walkiewicz and Stern, 2009; Yamanaka et al., 2013). A previous study reduced signaling through one arm of this pathway to extend the third instar stage from 5 to 9 days, from which the important observation that NMJs continue to grow and function throughout this period was made (Miller et al., 2012). More recent work has demonstrated that loss of key transcription factors in the PG, including Smox (also known as dSMAD2), can disrupt both signaling pathways to fully arrest larval development and prevent the transition to pupal stages (Gibbens et al., 2011; Ohhara et al., 2017). Remarkably, these arrested third instars (ATI) remain in the larval stage until death. The development of ATI larvae now provides an opportunity to characterize synaptic growth, function and plasticity in a system of terminally persistent expansion beyond normal physiological ranges and has the potential to reveal new insights into processes such as neurodegeneration.

Here, we have developed an optimized approach to arrest Drosophila larvae at third instar stages to characterize NMJ growth, function and plasticity. We find that ATI larvae continue to grow and survive for up to 35 days, during which NMJs exhibit exuberant expansion in both pre- and postsynaptic compartments. Interestingly, although this growth should enhance synaptic strength, no significant change is observed compared with baseline values. Instead, a potent reduction in presynaptic neurotransmitter release maintains stable synaptic strength across the life of an ATI larva. Finally, the ATI larvae enabled new insights into the progression of neurodegeneration in stathmin mutants. Together, arresting larval development now provides a powerful foundation to probe the mechanisms of synaptic growth, function, homeostatic plasticity and neurodegeneration at a model glutamatergic synapse in a genetically tractable system.

To arrest larval development at the third instar stage, we targeted genes that could either disrupt both Torso and insulin signaling pathways or broadly inhibit the synthesis of ecdysone synthesis in the PG, processes ultimately necessary for the transition to pupal stages (Fig. 1A; Gibbens et al., 2011; Ohhara et al., 2017). We reasoned that, if we could prevent the release of ecdysone from the PG by knocking down a key transcript(s), pupation would be delayed indefinitely (Yamanaka et al., 2013). Importantly, knock down of gene expression was restricted to the PG by expression with phm-Gal4 (Ono et al., 2006), minimizing secondary effects from possible functions in other tissues. We screened several lines described by other investigators that prevent larval transitions to pupariation (Danielsen et al., 2016) and found that a particular RNAi line targeting smox was the most effective, reliably preventing pupation in nearly all animals (Fig. 1A,B). smox encodes a transcription factor that is required for expression of both torso and insulin receptor genes in the PG (Gibbens et al., 2011), as well as having functions in a variety of other tissues (Brummel et al., 1999; Hevia et al., 2017). These persist as ATI larvae and live up to 35 days after egg lay (AEL). Typical wild-type larvae spend ∼3 days in the third instar stage before pupation and metamorphosis, living beyond 60 days AEL as adults (Rewitz et al., 2009). For the first 5 days of development, ATI larvae appear to be largely unchanged compared with wild type, but they fail to progress to become ‘wandering’ third instars. Rather, they continue to feed and gain body mass, peaking at ∼17 days AEL (ATI.17) and then gradually losing body mass until dying soon after 33 days AEL (ATI.33) (Fig. 1B). For further experiments, we compared wild-type larvae at 5 days AEL (WT.5) to ATI larvae at varying time points, including 5 days AEL (ATI.5), a time point similar to wild type; ATI.17, a time corresponding to peak body mass; and ATI.33, a time near the terminal stage of the ATI lifespan.

To investigate NMJs across the ATI lifespan, we first characterized muscle size and passive electrical properties of the muscle. We observed a progressive gain in muscle size across the ATI lifespan, with muscle surface area increasing by over 50%, peaking at ATI.17 and then decreasing to ATI.33 (Fig. 1C,D). Consistent with this substantial increase in muscle size, electrophysiological recordings of NMJs across the ATI lifespan revealed a massive decrease in input resistance, peaking ∼ATI.17 (Fig. 1D). Remarkably, despite these changes in muscle size, synaptic strength [measured by excitatory postsynaptic potential (EPSP) amplitude] remains constant across ATI NMJs (Fig. 1E,F). Thus, as larvae grow and decline through an ATI lifespan, synaptic strength at the NMJ remains constant.

Clearly, ATI NMJs maintain synaptic strength despite the substantial increase in muscle size that progresses through arrested larval development. In principle, modulations to the number of presynaptic release sites (N), the probability of release at each individual release site (P_r), and/or the postsynaptic response to glutamate release from single synaptic vesicles (quantal size, Q) could stabilize synaptic strength at these NMJs (Dittman and Ryan, 2019). We first assessed synaptic growth to determine whether the number of presynaptic release sites increases in proportion to the muscle surface area. During the conventional 3-4 day period of larval development, there is a 100-fold expansion in the NMJ, with changes to the passive electrical properties of the muscle and a concomitant growth of pre- and postsynaptic compartments (Atwood et al., 1993; Menon et al., 2013; Schuster et al., 1996). These changes are thought to scale NMJ function in parallel with growth and maintain sufficient depolarization for muscle contraction (Davis and Goodman, 1998). However, the progressive increase in muscle size at ATI NMJs poses a further challenge, in which synapses may need to expand to compensate for overgrowth. We therefore considered whether adaptive changes in the growth of motor terminals and/or number of synapses served to stabilize synaptic strength (EPSP amplitude). Using immunostaining, we instead found a progressive enhancement in the neuronal membrane surface area and in the number of boutons per NMJ throughout the ATI lifespan (Fig. 2A-D). In fact, the bouton to muscle area ratio even overshoots the scaling that is normally observed at conventional development between first and third instar larval stages [WT.5: 40 boutons/40,000 µm² ratio (Schuster et al., 1996); ATI.17: 100 boutons/75,000 µm² ratio; Table S2]. Hence, motor neuron terminals grow in excess to muscle growth.

As NMJ boutons expand across ATI stages, we considered the possibility of a compensatory reduction in the anatomical number of release sites. There is precedence for a reduction in the density of active zones (AZs), independent of NMJ growth, to maintain synaptic strength (Goel et al., 2019b; Graf et al., 2009). To identify individual presynaptic release sites, we immunostained NMJs with an antibody against Bruchpilot (BRP), a central scaffolding protein that constitutes the ‘T-bar’ structure at AZs in Drosophila (Kittel et al., 2006). As ∼96% of release sites are labeled by BRP at the fly NMJ (Akbergenova et al., 2018; Gratz et al., 2019; Wagh et al., 2006), we defined an anatomical release site as a BRP punctum and quantified these structures across the ATI lifespan. Interestingly, we found no significant changes in BRP puncta density across ATI stages, with total BRP puncta number per NMJ increasing in proportion to neuronal membrane area and bouton number (Fig. 2A-D; Table S2). Finally, although the number of BRP puncta increased, the size and fluorescence intensity of these puncta can be reduced at NMJs to compensate for synaptic overgrowth, reducing P_r at individual release sites and maintaining overall synaptic strength (Goel et al., 2019a; 2019b). However, although BRP number at ATI NMJs increases to over threefold that of wild-type NMJs, no compensatory reduction in size and/or intensity of BRP puncta was observed (Fig. 2A, lower panel; Table S2). Indeed, BRP puncta intensity was significantly increased compared with WT.5 levels (Table S2), which may reflect the age-dependent increase in size and intensity and active zones documented at the fly NMJ (Akbergenova et al., 2018). Thus, this anatomical analysis reveals an increase in N at ATI NMJs, implying that other adaptations compensate for excessive growth in ATI larvae.

Given the substantial increase in AZ number and intensity but stable synaptic strength, we next considered the possibility that a reduction in Q may have offset the observed presynaptic overgrowth at ATI NMJs. For example, a reduction in the abundance, composition and/or function of postsynaptic glutamate receptors (GluRs) may have occurred at ATI NMJs. At the fly NMJ, two receptor subtypes containing either GluRIIA or GluRIIB subunits form complexes with the essential GluRIIC, GluRIID and GluRIIE subunits to mediate the postsynaptic currents driving neurotransmission (Qin et al., 2005). GluRIIA-containing receptors mediate larger current amplitudes and slower decay kinetics compared with the GluRIIB-containing receptor counterparts (Han et al., 2015; Petersen et al., 1997). We examined the postsynaptic GluRs using antibodies that specifically recognize the GluRIIA or GluRIIB subunits, as well as the common GluRIID subunit (Fig. 3A). Consistent with presynaptic overgrowth, total GluR puncta numbers per NMJ mirrored the increase in presynaptic AZ number (Fig. 3B). Similarly, we observed a significant increase in the abundance of all GluR subunits assessed at ATI NMJs revealed by enhanced fluorescence intensity (Fig. 3C). Together, this demonstrates that postsynaptic receptor fields progressively expand in number and abundance, mirroring the accumulation in presynaptic structures across the ATI lifespan.

We next considered whether an apparent reduction in GluR functionality compensated for the expansion of glutamate receptor fields at ATI NMJs. We determined GluR functionality by electrophysiologically recording miniature events at ATI NMJs. Consistent with the increased fluorescence intensity of all subunits, we observed an ∼50% increase in miniature (m) EPSP amplitude compared with wild-type levels in ATI.17 larvae, an enhancement that persisted through ATI.33 (Fig. 3D,E). Consistent with increased presynaptic growth, we also observed an increase in mEPSP frequency (Fig. 3F). Along with the imaging data suggesting there is no change in GluR subunit stoichiometry, we observed no difference in mEPSP decay time constants (Fig. 3G). Thus, both imaging and electrophysiological data indicate that postsynaptic GluR abundance increases concurrently with presynaptic overgrowth at ATI NMJs.

In principle, it is possible that a change in the size of synaptic vesicles at ATI NMJs could also contribute to the enhanced quantal size observed at ATI NMJs. We therefore quantified synaptic vesicle diameter at NMJs of WT and ATI using electron microscopy (Fig. 4A). We observed no significant difference in the average vesicle diameter at NMJs of ATI compared with wild type (Fig. 4B). Therefore, the enhanced mEPSP amplitudes of ATI NMJs are due to an increase in postsynaptic GluR abundance, and yet synaptic strength remains stable across the ATI lifespan. This suggests a homeostatic adjustment in P_r is elicited at ATI NMJs that counteracts both the increased N and enhanced Q. It is unlikely that a retrograde signal from the muscle served to induce this reduction in P_r, as increased expression of postsynaptic GluRs at the fly NMJ similarly enhances mEPSP amplitude but does not alter presynaptic neurotransmitter release (Li et al., 2018c; Petersen et al., 1997). Thus, a reduction in P_r of sufficient magnitude must be induced at ATI NMJs to fully counteract the increase in N and Q to maintain stable synaptic strength.

ATI larvae exhibit exuberant synaptic growth, with accumulations of both pre- and postsynaptic components, resulting in an increased N and Q, factors that should enhance synaptic strength. However, EPSP amplitudes remain stable across the ATI lifespan, implying that P_r must be substantially and precisely diminished to compensate. To further test this idea, we calculated quantal content (the number of synaptic vesicles released per stimulus) and found a substantial reduction at ATI NMJs (Fig. 5A). Next, we assessed presynaptic function independently of mEPSP amplitude by performing failure analysis, where repeated stimulations in low extracellular Ca²⁺ (0.15 mM) fail to elicit a response in ∼50% of trials in wild type. At ATI NMJs, the failure rate was markedly increased (Fig. 5B), consistent with reduced quantal content. Finally, we assayed paired-pulse ratios to gauge P_r. At low extracellular Ca²⁺ (0.3 mM), paired-pulse facilitation (PPF) is observed at wild-type NMJs, whereas paired-pulse depression (PPD) is found in elevated Ca²⁺ (1.5 mM) (Böhme et al., 2016; Li et al., 2018c). In ATI.17 and ATI.33 NMJs, PPF was significantly increased, whereas PPD was reduced, consistent with reduced P_r relative to wild type (Fig. 5C,D). It is interesting to note that a similar phenomenon has been observed at the Drosophila NMJ in the context of typical larval development, referred to as presynaptic homeostatic depression (PHD). Here, mEPSP size is enhanced while quantal content is reduced to maintain normal EPSP amplitudes (Daniels et al., 2004; Gaviño et al., 2015; Li et al., 2018c). Although it is not clear that the mechanism of depression is shared between later ATI time points and PHD, we can posit that a homeostatic reduction in presynaptic release probability compensates for increased quantal size to maintain synaptic strength across the ATI life span.

It has previously been shown that NMJs expressing PHD can also express other forms of homeostatic plasticity, including a process referred to as presynaptic homeostatic potentiation (PHP) (Gaviño et al., 2015; Goel et al., 2019a; Li et al., 2018b). To induce PHP, we applied the postsynaptic GluR antagonist philanthotoxin-343 (PhTx) (Frank et al., 2006). Incubation for 10 min in PhTx reduces mEPSP amplitude in both wild-type and ATI NMJs, as expected (Fig. S1). In turn, EPSP amplitude is maintained at baseline levels due to a retrograde, homeostatic increase in presynaptic neurotransmitter release in wild type (Goel et al., 2017). Similarly, PHP is robustly expressed across ATI NMJs (Fig. S1). Thus, like PHD and other forms of homeostatic plasticity studied at the Drosophila NMJ, the presynaptic inhibition observed at ATI NMJs can be balanced with acute GluR challenge to express PHP and maintain stable synaptic strength.

In our final set of experiments, we considered whether ATI larvae could be used as models for aging and/or neurodegeneration. We hypothesized that NMJs in ATI larvae were unlikely to exhibit classical hallmarks of aging synapses. Although muscle integrity appears to degrade in ATI.33 compared with earlier time points (Fig. 1), synaptic growth (Fig. 2), GluR receptor fields (Fig. 3) and presynaptic function (Fig. 5), all appear similar in ATI.33 relative to earlier time points. Indeed, although reductions in synaptic components and neurotransmission have been observed at aging mammalian NMJs (Li et al., 2018a; Taetzsch and Valdez, 2018), NMJ structure and function remain surprisingly robust in ATI larvae nearing death, with no apparent defects in synaptic function or even PHP plasticity. One additional canonical indicator of aging reported at mammalian NMJs includes presynaptic retractions and fragmentation (Li et al., 2018a; Taetzsch and Valdez, 2018). We therefore assessed synaptic retractions across the ATI lifespan using an established ‘footprint’ assay, in which a postsynaptic marker is observed to persist without a corresponding presynaptic marker (Eaton et al., 2002; Graf et al., 2011; Perry et al., 2017). However, ATI NMJs, including ATI.33, showed surprisingly stable synapses, with no apparent increases in footprints compared with earlier time points (Fig. 6A,B). Together, these results indicate that NMJ structure, function and integrity remain surprisingly robust across all stages of ATI larvae, even at terminal periods, and are therefore unlikely to serve as a compelling model for age-related synaptic decline.

Although NMJs remain structurally intact and stable across the lifespan of the ATI larvae, this manipulation does enable a substantially longer timescale compared with the typical 5 days of larval development to investigate insults that contribute to neuronal degeneration. We chose to characterize NMJ growth and stability in stathmin mutants extended through the ATI manipulation. Stathmin is a tubulin-associated factor involved in maintaining the integrity of the axonal cytoskeleton (Duncan et al., 2013; Lachkar et al., 2010). The mammalian homolog of Drosophila stathmin (SCG10; also known as STMN2) is highly conserved and is thought to function as a surveillance factor for axon damage and degenerative signaling (Shin et al., 2012). In Drosophila, loss of stathmin leads to a marked increase in NMJ footprints, with more posterior segments showing increased severity relative to more anterior segments (Graf et al., 2011). Surprisingly, stathmin mutants are still able to pupate and develop into adults. However, stathmin mutants extended in larval stages by the ATI manipulation die shortly after 21 days AEL. We therefore sought to use the ATI system to determine the impact of a prolonged phase of axonal instability in stathmin mutants. Indeed, NMJs exhibit increased footprints in stathmin mutants extended to ATI.13 time points (stai.13) when compared with stai.5 controls in both frequency (Fig. 6C,D) and severity (Fig. 6E), with the most severe retractions observed in posterior abdominal segments (A3-A5). Finally, we tested whether NMJ growth increased in ATI-extended stathmin NMJs, as it does in wild type. Although control ATI synapses grow in bouton and BRP puncta number between 5 and 13 days AEL, stathmin NMJs fail to consistently expand (Fig. 6F-H). These experiments highlight the potential of the ATI system to be a useful tool for defining the progression of neurodegeneration at the Drosophila NMJ, which is otherwise limited to short larval stages.

By arresting further maturation at third instar Drosophila larvae, we have been able to accomplish a detailed study of NMJ structure, function and plasticity over much longer timescales than previously possible. This ATI larval system has revealed how the NMJ maintains stable transmission over a vastly extended developmental timescale, in which persistent overgrowth in both pre- and postsynaptic compartments is offset through a potent and homeostatic reduction in neurotransmitter release. Hence, this study not only provides evidence for a potentially novel homeostatic signaling system that balances release probability with synaptic overgrowth, but now extends the temporal window to enable the characterization of a variety of processes, including neurodegeneration, at a powerful model synapse.

As described by Miller et al. (2012), NMJs in third instar larvae that have been developmentally arrested for at least a week beyond the normal time of pupariation continue to grow and add new boutons. Here, we extend this observation to larvae arrested at the third instar for over 30 days, further demonstrating that mechanisms do not exist to suppress or negatively regulate growth when developmental timing is artificially extended. During normal larval growth from first to third instar, the body wall muscles undergo rapid and immense expansion, growing nearly 100-fold in surface area within a few days (Menon et al., 2013). Presynaptic terminals grow and add new boutons in parallel with muscle growth, presumably to maintain stable NMJ strength. In effect, sufficient levels of muscle excitation is sustained through a coordinated increase in all three parameters controlling synaptic physiology: N (number of release sites), P_r (release probability at each site), and Q (quantal size) (Dittman and Ryan, 2019). Hence, during typical stages of larval development, increasing muscle growth requires a concomitant elaboration in NMJs, implying robust signaling systems exist to ensure synaptic size, structure, and function expand in a coordinated manner. This tight structural coupling between muscle fiber and NMJ growth is also observed in mammals and is thought to be a primary mechanism for maintaining NMJ strength during post-developmental muscle growth or wasting (Sanes and Lichtman, 1999, 2001). However, when the normal developmental program is made to continue without terminating in pupariation, NMJ growth continues apparently unchecked, posing a potential challenge of hyperexcitation. There is emerging evidence that when NMJ growth is genetically perturbed, a redistribution of AZ material or adaptations in synapse morphogenesis or postsynaptic neurotransmitter receptors can maintain stable synaptic strength (Goel et al., 2019a,b; Graf et al., 2009). In the case of NMJ overgrowth in endophilin A mutants, a homeostatic scaling in AZ size compensates for increased number to lower release probability and maintain stable synaptic strength (Goel et al., 2019a). However, NMJs in ATI larvae do not appear to use such strategies. Rather, a latent form of adaptive plasticity is revealed at ATI NMJs that is sufficiently potent and precise to inhibit neurotransmitter release probability and compensate for the overgrowth of both pre- and postsynaptic compartments.

The presynaptic inhibition of neurotransmitter release that maintains synaptic strength at ATI NMJs is a potentially novel phenomenon of homeostatic plasticity. This form of presynaptic depression appears to be an entirely functional change that reduces release probability, without any apparent adaptations to AZ number, intensity or synaptic structure. Electrophysiologically, the presynaptic inhibition demonstrated at ATI NMJs resembles PHD, a form of homeostatic plasticity characterized at the Drosophila NMJ in which excess glutamate release induces a compensatory reduction in release probability that maintains stable synaptic strength (Daniels et al., 2004; Gaviño et al., 2015; Li et al., 2018c). As in PHD, the presynaptic inhibition at ATI NMJs is not reflected in changes to the AZs or synaptic structure (Goel et al., 2019a; Gratz et al., 2019; Li et al., 2018c). However, the only mechanisms known to be capable of inducing PHD require enhanced synaptic vesicle size that results from endocytosis mutants or overexpression of the vesicular glutamate transporter (Daniels et al., 2004; Dickman et al., 2005; Goel et al., 2019a; Verstreken et al., 2002). We found no evidence for changes in synaptic vesicle size at ATI NMJs, as the enhanced postsynaptic glutamate receptor levels observed are sufficient to explain the increased quantal size (Fig. 3) and direct measurement of vesicle size by EM revealed no significant change from wild type (Fig. 4). Hence, if the homeostatic depression observed at ATI NMJs is ultimately the same plasticity mechanism as PHD, then this would be the first condition that does not require enlarged synaptic vesicle size. In this case, perhaps excess global glutamate release from increased release sites at ATI NMJs induces the same homeostatic plasticity that increased glutamate released from individual synaptic vesicles does. This would be consistent with a ‘glutamate homeostat’, responding to excess presynaptic glutamate release, necessary to induce and express PHD (Li et al., 2018c). Alternatively, the presynaptic inhibition triggered at ATI NMJs could be a novel form of PHD, which is induced in response to synaptic overgrowth. Interestingly, although increased postsynaptic glutamate receptor abundance leads to enhanced mEPSP amplitude, no adaptive change in presynaptic function results, which leads to a concomitant increase in synaptic strength (DiAntonio et al., 1999; Li et al., 2018c). One possibility is that a coordinated increase in both pre- and postsynaptic compartments may be necessary to induce the presynaptic inhibition observed at ATI NMJs. The ATI model provides a unique opportunity to interrogate the interplay between developmental growth, adaptive presynaptic inhibition and other homeostatic signaling systems.

Extending the larval stage through the ATI manipulation will circumvent limitations of the brief time window provided by the standard developmental program. Although the ATI model does not appear to exhibit the features described at aging mammalian NMJs (Li et al., 2018a; Taetzsch and Valdez, 2018), we have demonstrated its potential for modeling neurodegenerative conditions by showing the extent of synaptic destabilization caused by loss of stathmin that was not fully apparent when restricted to the normal short developmental period in Drosophila larvae (Graf et al., 2011). By examining stathmin mutant phenotypes in ATI-extended larvae, we were able to observe progressive, time-dependent retractions of presynaptic terminals and gain further insight into the role of stathmin in normal NMJ growth and stability. Consistent with the role of stathmin in flies, the mammalian homolog SCG10 is thought to be part of an axonal injury surveillance system, where it accumulates after injury and is involved in regenerative signaling (Shin et al., 2012). More generally, previous studies of degenerative disease models in the larval system have been limited by the brief timespan. For example, one important ALS disease model in flies involves overexpression of repetitive RNAs and peptides derived from the human C9orf72 gene (Mizielinska et al., 2014; Xu et al., 2013). However, although a variety of progressive and degenerative phenotypes are observed in photoreceptors of adult flies, only the most toxic transgenes are capable of inducing substantial neurodegeneration at the larval NMJ (Perry et al., 2017), likely owing to the limited time frame of typical larval development. The longer timescale enabled by the ATI model therefore provides new opportunities to study progressive phenotypes during neuronal injury, stress and neurodegeneration in addition to the plasticity discussed above in a rapid and genetically tractable system. Indeed, fly models of neurodegenerative conditions such as ALS, Huntington's, Parkinson's and Alzheimer's diseases (McGurk et al., 2015) can benefit from the high-resolution imaging and electrophysiological approaches established at the larval NMJ. The powerful combination of established genetic tools, including binary expression systems (Gal4/UAS, LexA, QF systems; Venken et al., 2011) and emerging CRISPR/Cas9 manipulations (Bier et al., 2018) with the ATI model provides an exciting foundation to gain new insights into synaptic growth, structure, function, plasticity, injury and neurodegeneration over long time frames using the glutamatergic NMJ as a model.

Drosophila stocks were raised at 25°C on standard molasses food. The w¹¹¹⁸ strain is used as the wild-type control unless otherwise noted, as this is the genetic background of the genetic mutants used in this study. ATI larvae were generated by crossing phm-GAL4 (Gibbens et al., 2011) to UAS-smox-RNAi [Bloomington Drosophila Stock Center (BDSC), 41670]. stathmin mutations were introduced into the ATI background (stai allele: BDSC, 16165). All experiments were performed on third instar larvae or ATI larvae of both sexes and at various time points. See Table S8 for a complete list of all stocks and reagents used in this study.

Third instar male or female larvae were dissected in ice cold Ca²⁺-free HL-3 and fixed in Bouin's fixative for 5 min as previously described (Kikuma et al., 2017). Briefly, larvae were washed with PBS containing 0.1% Triton X-100 (PBST) for 30 min, blocked for 1 h with 5% normal donkey serum in PBST, and incubated overnight in primary antibodies at 4°C followed by washes and incubation in secondary antibodies. Samples were mounted in VectaShield (Vector Laboratories). The following antibodies were used: mouse anti-Bruchpilot [nc82; 1:100; Developmental Studies Hybridoma Bank (DSHB)]; rabbit anti-DLG (1:10,000; Pielage et al., 2006); guinea pig anti-vGlut (1:2000; Chen et al., 2017); mouse anti-GluRIIA (8B4D2; 1:100; DSHB); affinity purified rabbit anti-GluRIIB (1:1000; Goel and Dickman, 2018); guinea pig anti-GluRIID (1:1000; Kikuma et al., 2019). Donkey anti-mouse, anti-guinea pig and anti-rabbit Alexa Fluor 488-, Cyanine 3 (Cy3)- and DyLight 405-conjugated secondary antibodies (Jackson ImmunoResearch) were used at 1:400. Alexa Fluor 647-conjugated goat anti-HRP (Jackson ImmunoResearch) was used at 1:200. Tetramethylrhodamine (TRITC)-conjugated phalloidin (Thermo Fisher Scientific) was used at 1:1000. All antibody information is summarized in Table S8.

Samples were imaged using a Nikon A1R resonant scanning confocal microscope equipped with NIS Elements software and a 100× APO 1.4NA oil immersion objective using separate channels with four laser lines (405, 488, 561 and 637 nm). For fluorescence quantifications of BRP intensity levels, z-stacks were obtained using identical settings for all genotypes with z-axis spacing 0.5 µm within an experiment and optimized for detection without saturation of the signal as previously described (Perry et al., 2017). Boutons were counted using vGlut- and HRP-stained Ib NMJ terminals on muscle (M) 4 of segment A2-A4, considering each vGlut punctum to be a bouton. The general analysis toolkit in the NIS Elements software was used for image analysis as previously described (Kikuma et al., 2017). Neuronal surface area was calculated by creating a mask around the HRP channel that labels the neuronal membrane. BRP puncta number, area and total BRP intensity per NMJ were quantified using a bright-spot detection method and filters to binary layers on the BRP-labeled 488 channel in a manner similar to that previously described (Goel et al., 2019b). GluRIIA, GluRIIB and GluRIID puncta intensities were quantified by measuring the total sum intensity of each individual GluR punctum and these values were then averaged per NMJ to get one sample measurement (n). For NMJ retraction analysis, footprints were scored by eye as reported in Eaton et al. (2002) on M6/7 segments A2-A5. Anti-DLG was used as a postsynaptic marker and either anti-vGlut or anti-BRP for a presynaptic marker (wild-type controls yielded similar retraction scores for either presynaptic marker).

EM analysis was performed as previously described (Russo et al., 2019) with minor modifications. Wandering third instar larvae were dissected in Ca²⁺-free HL-3 and then fixed in 4% paraformaldehyde prepared by mixing 16% paraformaldehyde (aqueous, EMS, 15700) in PBS at 4°C. The larval pelts were stored in this fixative at 4°C. On the day of preparation, larvae were transferred to 1% glutaraldehyde in distilled water for 1 h at room temperature, then washed 3× for 5 min in distilled water. The larval pelts were then placed in 1% osmium tetroxide in distilled water for 90 min at room temperature and in ferracyanide-reduced osmium tetraoxide for 90 min. After washing 3× in water, larvae were dehydrated in a graded acetone series, infiltrated with 50% uncatalyzed Spurr's resin in acetone for 24 h, followed by two 24 h changes of 100% uncatalyzed Spurr's and polymerization in freshly catalyzed Spurr's for 24 h at 60°C. Then 60 nm sections were collected on Formvar-coated 2×1 mm slot grids, stained with 2% uranyl acetate and Reynolds lead citrate and imaged by transmission EM at 80 kV in a Zeiss EM10 equipped with an Erlangshen CCD camera (Gatan). Images were analyzed blind to genotype using measurement tools in the ImageJ software. The outer diameter of each synaptic vesicle within a 300 nm radius from the center of an active zone within a bouton was quantified for all genotypes.

All dissections and recordings were performed in modified HL-3 saline (Stewart et al., 1994; Dickman et al., 2005; Kiragasi et al., 2017) containing 70 mM NaCl, 5 mM KCl, 10 mM MgCl₂, 10 mM NaHCO₃, 115 mM sucrose, 5 mM Trehalose, 5 mM HEPES and 0.4 mM CaCl₂ (pH 7.2). NMJ sharp electrode (electrode resistance of 10-30 MΩ) recordings were performed on M6 and M7 of abdominal segments A2 and A3 in wandering third instar larvae as previously described (Goel et al., 2019a). Recordings were performed on an Olympus BX61 WI microscope using a 40×/0.80 water-dipping objective, and acquired using an Axoclamp 900A amplifier, Digidata 1440A acquisition system and pClamp 10.5 software (Molecular Devices). Electrophysiological sweeps were digitized at 10 kHz and filtered at 1 kHz. Data were analyzed using Clampfit (Molecular Devices), MiniAnalysis (Synaptosoft) and Excel (Microsoft) software.

mEPSPs were recorded in the absence of any stimulation and cut motor axons were stimulated to elicit EPSPs. Average mEPSP, EPSP and quantal content were calculated for each genotype by dividing EPSP amplitude by mEPSP amplitude. Muscle input resistance (R_in) and resting membrane potential (V_rest) were monitored during each experiment. Recordings were rejected if the V_rest was above −60 mV, if the R_in was less than 5 MΩ, or if either measurement deviated by more than 10% during the course of the experiment. Larvae were incubated with or without philanthotoxin-433 (PhTx; Sigma-Aldrich; 20 μM) and resuspended in HL-3 for 10 min, as previously described (Frank et al., 2006; Dickman and Davis, 2009).

Failure analysis was performed in HL-3 solution containing 0.15 mM CaCl₂, which resulted in failures in about half of the stimulated responses in wild-type larvae. A total of 40 trials (stimulations) were performed at each NMJ in all genotypes. Failure rate was obtained by dividing the total number of failures by the total number of trials (40). Paired-pulse recordings were performed at a Ca²⁺ concentration of 0.3 mM to assay PPF and 1.5 mM for PPD. Following the first stimulation, a second excitatory postsynaptic current (EPSC) was evoked at an interstimulus interval of 16.67 ms (60 Hz). Paired-pulse ratios were calculated as the EPSC amplitude of the second response divided by the first response (EPSC2/EPSC1).

For electrophysiological and immunostaining experiments, each NMJ terminal (M6 for physiology, M4 for immunostaining analyses) is considered an n of 1, as each presynaptic motor neuron terminal is confined to its own muscle hemi-segment. For these experiments, M4 or M6 were analyzed from hemi-segments A2-A4 from each larva, typically two NMJs per animal per experiment. To control for variability between larvae within a genotype, NMJs were analyzed from at least five individual larvae. See Tables S1-S7 for additional details.

Statistical analysis was performed using GraphPad Prism (version 7.0) or Microsoft Excel software (version 16.22). Data were assessed for normality using a D'Agostino-Pearson omnibus normality test, which determined that the assumption of normality of the sample distribution was not violated. Normally distributed data were analyzed for statistical significance using a Student's t-test (pairwise comparison) or an analysis of variance (ANOVA) and Tukey's test for multiple comparisons. Data were then compared using either a one-way ANOVA and tested for significance using a Tukey's multiple comparison test or using an unpaired 2-tailed Student's t-test with Welch's correction. All data are presented as mean±s.e.m. with varying levels of significance assessed as *P≤0.05, **P≤0.01, ***P≤0.001 (N.S., not significant). See Tables S1-S7 for additional statistical details and values.

We thank Naoki Yamanaka (University of California, Riverside, CA, USA) and Mike O'Connor (University of Minnesota, MN, USA) who sent us other RNAi lines to test for the ATI model. We also thank the Bloomington Drosophila Stock Center and Developmental Studies Hybridoma Bank for additional stocks and reagents (National Institutes of Health grant P40OD018537).

The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.186312.reviewer-comments.pdf