Active Zone Maturation Controls Presynaptic Output and Release Mode and Is Regulated by Neuronal Activity

A genetic toolkit for time-stamping synapse age

Prior quantal imaging studies at Drosophila NMJs revealed a ∼50-fold heterogeneity in evoked P_r across the hundreds of AZs formed by a single MN (Peled and Isacoff, 2011; Melom et al., 2013; Newman et al., 2017, 2022; Akbergenova et al., 2018; Jetti et al., 2023; Medeiros et al., 2024). One model for this heterogeneity predicts each AZ is fated to have a distinct output strength that is maintained over the life of that individual AZ. This could arise from AZ components being delivered in variable abundance and/or composition that are deposited together at forming synapses or through the presence of unique components that direct future output to a set P_r. An alternative model is that AZs are initially assembled with poor release capacity and mature over development to a higher P_r state, with heterogeneity reflecting distinct stages of age-dependent maturation across the AZ population. Serial imaging of NMJ growth in briefly anesthetized Drosophila larvae provided support for the second model, suggesting newly formed AZs are weaker than their more mature counterparts (Akbergenova et al., 2018). However, the rapid rate of AZ addition during larval growth makes it difficult to follow defined AZs across multiple imaging sessions.

To more precisely examine how synapse age contributes to presynaptic output at individual AZs, we developed a new toolkit for time-stamping synapses. mMaple is a genetically encoded fluorophore that undergoes complete green-to-red PC upon illumination with 405 nm light (McEvoy et al., 2012). By attaching mMaple to the core postsynaptic glutamate receptor subunit GluRIIE in transgenic animals, all PSDs opposed to presynaptic AZs at the NMJ can be reliably labeled (Fig. 1A). Brief 15 s illumination at 405 nm through the cuticle of intact first instar larvae was sufficient to generate complete transition of the entire pool of GluRIIE^Maple from green to red at M4 NMJs (Fig. 1B). Serial imaging of NMJs identified a small pool of nonsynaptic red⁺ GluRIIE^Maple that continued to incorporate into PSDs during the first day after PC, resulting in a mild increase in the number of red⁺ GluRIIE-positive PSDs (Fig. 1C). However, 48 h post PC, all newly formed PSDs contained only non-PC green⁺ GluRIIE^Maple (hereafter referred to as GluR^New; Fig. 1C,D). Quantification of the number of PC red⁺ GluRIIE^Maple-positive PSDs (hereafter referred to as GluR^Old) revealed no change between 24 and 48 h (Fig. 1D), indicating the entire pool of previously PC GluRIIE^Maple had inserted into PSDs within a day and did not undergo lateral movement into newly formed synapses. Although no new GluR^Old PSDs were observed, NMJs continued to grow with the number of GluR^New PSDs increasing as expected (Fig. 1C,E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

GluRIIE-mMaple PC to timestamp synapse age. A, Schematic showing PC of GluRIIE^Maple from green to red. All existing PSDs at the time of PC contain red⁺ GluRIIE (PSD^Old), while newer PSDs formed after PC contain green⁺/red⁻ GluRIIE (PSD^New). B, Complete PC of larval NMJ PSDs containing GluRIIE^Maple from green⁺ to red⁺ after 15 s transcuticular exposure to 405 nm light. C, Serial imaging of the same M4 NMJ following brief anesthetization at three time points (0, 24, 48 h). PC was performed at timepoint 0 in the first instar stage. Although new PSDs incorporated nonsynaptic red⁺ GluRIIE during the first day post-PC (middle panel), only non-PC green⁺/red⁻ GluRIIE was present at PSDs that formed after 24 h (bottom panel). D, The percentage of red⁺ GluRIIE PSDs remains unchanged between 24 and 48 h PC, indicating new PSDs formed after 24 h contain only newly synthesized GluR^New (24 h, 100%; n = 15 NMJs in 5 larvae; 48 h, 101.4% ± 0.85; n = 15 in 5 larvae; p = 0.11). E, Representative single bouton images of PSDs containing GluR^New and GluR^Old at 24 and 144 h post-PC. F, Red⁺ GluR^Old fluorescent intensity at larval NMJ PSDs is stable and does not undergo significant decay between 24 and 144 h post-PC (Day 1, 9,070 ± 97.3 average RFU; n = 1,198 PSDs in 4 larvae; Day 6, 8,910 ± 162.2; n = 858 PSDs in 4 larvae; p = 0.3725). Statistical significance determined with Student's t test, ns, not significant. Raw values and statistical details are provided in Data S1.

To determine the extent of GluR turnover, average fluorescent intensity of red⁺ GluR^Old PSDs was compared at Day 1 and Day 6 post-PC. Any decay in fluorescence of GluR^Old likely reflects turnover, as red-shifted mMaple is highly stable and cannot spontaneously transition back to the green state. No significant decrease in GluR^Old fluorescence was observed over this 5 d interval (Fig. 1E,F; Data S1). These data indicate GluR clusters at Drosophila NMJs are highly stable and do not undergo robust turnover during larval development, consistent with prior imaging of GluRs (Rasse et al., 2005). The absence of GluR^Old at newly formed PSDs later in development also indicates GluRs do not undergo lateral movement between PSDs, similar to observations using mMaple-tagged Cac (the sole Drosophila Ca_v2 Ca²⁺ channel) that is retained within individual presynaptic AZs (Cunningham et al., 2022). Together, these data establish GluRIIE^Maple as a tool for time-stamping synapses.

Synaptic output and release mode are dependent on AZ age and maturation state

To examine the effect of age-dependent synapse maturation on presynaptic output, we performed quantal imaging of single SV release events at individual AZs using postsynaptically expressed membrane-tethered myrGCaMP7s (Movie 1) as previously described (Akbergenova et al., 2018; Sauvola et al., 2021; Jetti et al., 2023). Animals expressing myrGCaMP7s and GluRIIE^Maple were PC during the early second instar larval stage, and quantal imaging was performed 4 d later to generate AZ P_r maps. The myrGCaMP7s signal is much dimmer than GluRIIE^Maple at rest and does not interfere with identification of GluR^New PSDs. Similarly, the dramatic increase in myrGCaMP7s fluorescence following SV fusion and subsequent postsynaptic opening of GluRs allows simultaneous imaging of synaptic activity on top of baseline GluRIIE^Maple. Older PSDs had more GluR^Old and brighter red fluorescence compared with newer PSDs that accumulated only GluR^New (Fig. 2A). Total GluR^Old intensity at individual PSDs was highly correlated with P_r (Pearson's r value = 0.56; Fig. 2A,B), indicating evoked output of the corresponding AZ was strongly dependent on synapse age. Comparison of all sites that contained any amount of GluR^Old with newer synapses that contained only GluR^New revealed a ∼2.5-fold difference in P_r (Fig. 2C). These data indicate older AZs establish and maintain higher evoked P_r across larval development.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Enhanced evoked synaptic output at older PSDs. A, Representative larval M4 NMJ imaged 96 h after PC with GluR^Old PSDs individually numbered (left panel). The corresponding evoked P_r heatmap following quantal imaging with myrGCaMP7s is shown on the right, revealing AZs opposed to the oldest PSDs are among the strongest release sites. B, Correlation between PSD age, determined by GluR^Old sum fluorescent intensity of individual PSDs, and evoked P_r (Pearson's r value = 0.56). GluR^Old sum intensity was normalized for each NMJ across animals, with the max sum fluorescence PSD signal set to 1.0 for each corresponding NMJ. C, Average evoked P_r for AZs opposed to older PSDs containing any red⁺ GluR^Old compared with younger PSDs with only green⁺/red⁻ GluR^New (red⁺ PSDs, 0.193 ± 0.013; n = 174 PSDs from 5 larvae; red⁻ PSDs, 0.081 ± 0.005; n = 378 PSDs from 5 larvae; p < 0.0001). D, Representative larval M4 NMJ imaged 48 h after PC with GluR^Old PSDs individually numbered (left panel). The corresponding evoked P_r heatmap following quantal imaging with myrGCaMP7s is shown on the right. E, Correlation between PSD age determined by red⁺ GluR^Old fluorescent intensity and evoked P_r (Pearson's r value = 0.66). F, Average evoked P_r for AZs opposed to older PSDs containing any red⁺ GluR^Old compared with younger PSDs with only green⁺/red⁻ GluR^New (red⁺ PSDs, 0.174 ± 0.01; n = 179 PSDs from 5 larvae; red⁻ PSDs, 0.0397 ± 0.004; n = 192 PSDs from 5 larvae; p < 0.0001). G, Representative images of NMJ growth at M4 (left) and M6/7 (right) 4 d after PC of GluRIIE^Maple first instar larvae. Note the addition of terminal boutons lacking older red⁺ PSDs (arrow) at M4 versus the internal addition of new boutons lacking red⁺ PSDs at M6 (white outline and arrows). H, Quantification of the percent of GluR^Old PSDs to all PSDs at the terminal bouton versus the second bouton from the end. Older PSDs are enriched in the terminal bouton at M6/7 NMJs (terminal bouton GluR^Old positive PSDs as a percent of all PSDs, M6/7, 34.3% ± 3.4; n = 15 NMJs from 5 larvae; M4, 14.2% ± 2.7; n = 29 NMJs from 5 larvae; second bouton from end, M6/7, 29.7% ± 3.7; n = 14 NMJs from 5 larvae; M4, 38.6% ± 3.0; n = 26 NMJs from 5 larvae). Statistical significance determined with Student's t test, ns, not significant. Asterisks denote the following p value: ***p ≤ 0.001. Raw values and statistical details are provided in Data S1.

Quantal imaging after 4 d of PC does not allow precise determination of the output of newer release sites, as GluR^New PSDs are indistinguishable whether they formed during Day 1 or Day 4 after PC. To more precisely examine AZ age and release strength, we performed quantal imaging over a shorter 2 d interval after PC of early second instar larvae (Fig. 2D). A stronger correlation was observed for evoked P_r at sites containing GluR^Old versus newly formed PSDs that were <1 d old and contained only GluR^New (Pearson's r value = 0.66; Fig. 2E). Indeed, a more than fourfold difference in P_r between older synapses containing any GluR^Old signal was observed compared with those without any (Fig. 2F). Together, these data indicate synapse age and the strength of evoked release at individual AZs are interlinked. Given these variables are not perfectly correlated, other factors are also likely to contribute to presynaptic strength.

Prior studies identified heterogeneity in synaptic transmission strength along some Drosophila larval motor axons, with terminal boutons showing enhanced Ca²⁺ influx and release (Guerrero et al., 2005). This graded transmission was not uniform across all MNs, as M6/7 MNs displayed enhanced output at terminal boutons while M4 MNs lacked this effect (Lnenicka et al., 2006). Molecular mechanisms explaining enhanced release from terminal boutons and differences across MN types remain unclear. Given NMJ expansion can result in individual release sites changing their relative position during larval development, we hypothesized distinct patterns of bouton addition during synaptic growth might contribute to the observed release heterogeneity. To compare patterns of NMJ growth at M6/7 versus M4, the relative abundance of GluR^Old PSDs was quantified at individual boutons from proximal to distal after several days of growth post-PC (Fig. 2G,H). Axon terminals of M6/7 MNs commonly grew by “stretching” and adding new internal boutons such that terminal boutons contained more GluR^Old PSDs. In contrast, M4 MNs often grew by budding new branches from the end of the axon, resulting in terminal boutons lacking or having fewer older PSDs. These distinct patterns of synaptic growth and AZ addition, along with the 2.5-fold enhancement of mature AZs at terminal boutons of M6/7 NMJs, provide an underlying mechanism to explain differences in graded transmission between these two MN subtypes.

In addition to evoked release, single SVs also fuse through an action potential–independent mechanism to generate spontaneous “mini” events. We previously determined that ∼10% of the AZ population at mature third instar larval NMJs display only spontaneous release and lack evoked fusion (Akbergenova et al., 2018). Given the role of maturation for evoked P_r, spontaneous-only AZs might reflect an early developmental stage prior to accumulation of Cac channels required to support evoked release. To determine if spontaneous fusion could occur at AZs lacking Cac, quantification of evoked P_r and spontaneous release rate was performed in larvae expressing endogenously CRISPR-tagged Cac^TagRFP (Gratz et al., 2019). Comparison of heatmaps for the spontaneous release rate and evoked P_r revealed a population of AZs capable of spontaneous fusion but lacking evoked release and detectable Cac (Fig. 3A). Although these data suggest spontaneous-only sites represent an immature state of AZ development, it is unclear if spontaneous release rate is enhanced during AZ maturation as observed for evoked P_r. To correlate synapse age and spontaneous release efficacy, we performed quantal imaging 4 d following PC in GluRIIE^Maple larvae. Although a positive correlation between spontaneous release rate and PSD age was observed (Pearson's r value = 0.31; Fig. 3B,C), the correlation was significantly less than for evoked P_r (r = 0.56; Fig. 2B). Overall, AZs opposed to sites that contained any GluR^Old had a ∼2-fold increase in spontaneous release rate compared with sites containing only GluR^New (Fig. 3D). Similarly, quantifying spontaneous release rate in larvae expressing endogenously CRISPR-tagged Cac^RFP revealed that Cac^RFP+ AZs displayed a 2.2-fold increase in spontaneous release rate over AZs lacking Cac (Fig. 3E). We conclude that spontaneous-only AZs largely represent an early developmental state and that evoked and to a lesser extent spontaneous release become more efficient as AZs accumulate scaffolding proteins and VGCCs during maturation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

AZs lacking Cac can support spontaneous fusion. A, Representative third instar M4 NMJ from a larva expressing GluRIIB-GFP and Cac^RFP. Several PSDs lacking opposed Cac^RFP+ signal are individually numbered. Cac^RFP signal is shown in Panel 2, with spontaneous activity (release events per minute) presented as a heat map at AZs opposed to GluRIIB⁺ PSDs in Panel 3. The corresponding evoked P_r heatmap is shown on the right, highlighting lack of evoked release at Cac⁻ AZs. AZs lacking Cac can display low (Sites 1 and 5), moderate (Site 3), or high (Sites 2 and 4) rates of spontaneous release with no corresponding evoked output. B, Representative third instar larval M4 NMJ imaged 96 h after PC to identify GluR^Old PSDs (left panel). The corresponding heatmap for spontaneous activity (release events per minute) is shown on the right. C, Correlation between PSD age determined by red⁺ GluR^Old fluorescent intensity and spontaneous release rate (Pearson's r value = 0.31). D, Spontaneous release rate for AZs opposed to older PSDs containing any red⁺ GluR compared with younger PSDs with only green⁺/red⁻ GluR (red⁺ synapses, 0.841 ± 0.1 events per min; n = 80 AZs from 4 larvae; red⁻ synapses, 0.398 ± 0.04; n = 157 AZs from 4 larvae; p < 0.0001). E, Spontaneous release rate for AZs containing Cac^RFP compared with those without (Cac^RFP+, 1.11 ± 0.05 events per min; n = 366 AZs from 5 larvae; Cac^RFP−, 0.497 ± 0.07; n = 55 AZs from 5 larvae; p < 0.0001). Statistical significance determined with Student's t test; ns, not significant. Asterisks denote the following p values: ***p ≤ 0.001 and ****p ≤ 0.0001. Raw values and statistical details are provided in Data S1.

Sequential addition and accumulation of AZ proteins at nascent synapses

Although a number of transport mechanisms are implicated in material delivery to synapses (Goldstein et al., 2008; Rizalar et al., 2021; Petzoldt, 2023), it is unclear if most AZ components are cotrafficked and delivered as preassembled modules. At Drosophila NMJs, a sequential process of AZ assembly has been observed, with Liprin-α and Syd1 arriving early during AZ formation (Owald and Sigrist, 2009). To quantify the pattern of accumulation for a broad range of AZ proteins, we analyzed fluorescently tagged lines expressing Liprin-α, Syd1, BRP, RIM, RBP, Cac, Unc13B, and Unc13A at early second instar larval NMJs when AZ addition is robust. The presence of these proteins across the AZ population was correlated with the distribution of endogenously CRISPR-tagged Cac channels (Cac^GFP or Cac^RFP) and endogenous RBP, a late AZ scaffold, detected by antibody staining (Fig. 4A). Liprin-α was the earliest arriving AZ protein detected using this colocalization assay, with 34.1% of Liprin-α⁺ puncta lacking RBP and 43.4% lacking Cac. To determine if Liprin-α⁺ puncta were newly formed AZs or accumulations of the protein outside of release sites, we quantified the presence of co-opposed postsynaptic GluRIIB⁺ receptor fields. Of the Liprin-α puncta, 93.3% were opposed to GluRIIB⁺ PSDs (Fig. 4B), indicating most Liprin-α⁺ clusters detected using this method represent AZs in various states of maturation. RIM and Syd1 also appeared early at forming AZs, with 9.7% of RIM⁺ AZs lacking RBP and 24.3% lacking Cac, while 9.6% of Syd1⁺ AZs lacked RBP and 18.1% lacked Cac (Fig. 4A). In contrast, BRP arrived later, with only 1.6% of BRP⁺ AZs lacking RBP and 11% lacking Cac, similar to prior observations (Fouquet et al., 2009). To assay the arrival and distribution of the two Unc13 spicing isoforms (Böhme et al., 2016), CRISPR was used to endogenously tag Unc13B with mClover and Unc13A with mRuby. Unc13B was an early arriving AZ protein as previously observed (Böhme et al., 2016), with 10.8% of Unc13B⁺ AZs lacking RBP and 19.6% lacking Cac. Unc13A followed a similar pattern to the later-arriving scaffolds BRP and RBP, with only 1.7% of Unc13A⁺ AZs lacking RBP and 9% lacking Cac (Fig. 4A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Time course for protein accumulation during AZ maturation. A, Representative second instar M4 synaptic boutons with an analysis of the distribution of fluorescent puncta for the indicated AZ proteins (left panel A) in relation to immunostained RBP (middle panel B) and endogenous CRISPR GFP-tagged or RFP-tagged Cac (right panel C). The merged image shows labeling for Cac and the AZ protein indicated in panel A, with representative AZs lacking Cac indicated by white arrows. The right panels show the percentage of individual puncta for each AZ protein lacking RBP or Cac (top values) versus RBP or Cac puncta lacking the indicated AZ protein (bottom values). N = 7 NMJs from two larvae for each correlation. B, Representative image of a second instar M4 synaptic bouton showing Liprin-α puncta compared with the distribution of GluRIIB. Most puncta overlap, although a few are positive for Liprin-α only (white arrows). C, Serial imaging of RBP and Cac appearance at second instar M4 AZs over three time points indicate RBP accumulates before Cac arrival. The appearance of new RBP or Cac accumulations within 6 h intervals was scored (n = 10 NMJs from 3 larvae), revealing 62% of newly formed puncta are only RBP⁺ and 38% are RBP⁺ and Cac⁺. No case of Cac at newly formed AZs lacking RBP was observed. Raw values and statistical details are provided in Data S1.

Although differential accumulation of proteins observed in fixed imaging suggests a sequential AZ maturation process, it doesn't provide a time course for AZ assembly. To examine AZ maturation over time, serial imaging of briefly anesthetized larvae expressing fluorescently tagged RBP and Cac was performed at 0, 6, and 12 h time points (Fig. 4C). The appearance of new RBP⁺ AZ clusters and their subsequent accumulation of Cac was scored across the three imaging sessions. All RBP⁺ AZs lacking Cac at timepoint 0 had accumulated Ca²⁺ channels by the 12 h timepoint. As such, Cac must arrive within an interval shorter than 12 h. We next focused on comparing Cac accumulation at RBP⁺ AZs within a shorter 6 h interval. If a RBP⁺ AZ was already present, Cac was observed in 100% of cases after 6 h. For 62% of cases where new RBP clusters appeared during a 6 h imaging window, Cac was not detected at these AZs. Newly formed RBP⁺ clusters that also contained Cac appeared 38% of the time, suggesting ∼4 h is required for Cac channels to arrive after RBP accumulation. Consistent with other assays measuring AZ protein accumulation at Drosophila NMJs (Rasse et al., 2005; Graf et al., 2009; Owald et al., 2012; Özel et al., 2019; Petzoldt et al., 2020), our data support a sequential addition process where Liprin-α arrives first, followed by a cohort of early scaffolds (Syd1, RIM, and Unc13B). BRP, RBP and Unc13A arrive at AZs later and around the same time, with the VGCC Cac following several hours after late scaffolds accumulate.

Early AZ scaffold levels correlate poorly with evoked release, while late AZ scaffold abundance predicts presynaptic release output

Given sequential addition of proteins to maturing AZs, how their abundance correlates with synapse function was assayed. Presynaptic release strength has been previously correlated with AZ abundance of Ca²⁺ channels (Sheng et al., 2012; Akbergenova et al., 2018; Gratz et al., 2019; Medeiros et al., 2024), while late AZ scaffolding proteins like RBP and BRP facilitate clustering of Cac at Drosophila AZs (Kittel et al., 2006; Fouquet et al., 2009; Liu et al., 2011) and enhance release (Peled et al., 2014; Muhammad et al., 2015; Reddy-Alla et al., 2017; Akbergenova et al., 2018; Newman et al., 2022). Whether the abundance of early synaptic scaffolds regulates release output is unclear. To examine early versus late scaffold abundance and presynaptic output, fluorescently tagged lines expressing Liprin-α, Syd1, RBP and Unc13A were assayed for Cac levels and evoked P_r. AZ Cac levels correlated weakly with the abundance of the early scaffolding proteins Liprin-α (Pearson's r value = 0.19) and Syd1 (Pearson's r value = 0.26, Fig. 5A–C). In contrast, AZ abundance of the late scaffolding proteins RBP (Pearson's r value = 0.51) and Unc13A (Pearson's r value = 0.66) were strongly correlated with Cac levels (Fig. 5A,D,E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

The abundance of early AZ scaffolds poorly predicts Cac accumulation and P_r. A, Representative images of third instar M4 synaptic boutons expressing endogenous CRISPR-tagged Cac^GFP (top row) and the indicated AZ scaffolding protein (bottom row). One of the brightest Cac⁺ puncta in each bouton is highlighted (white arrows) to compare abundance with the indicated scaffolding protein. B–E, Correlation between AZ abundance (fluorescent intensity) of Cac and Liprin-α (B, n = 1,422 AZs from 6 NMJs from 6 larvae), Syd1 (C, n = 1,579 AZs from 6 NMJs from 6 larvae), RBP (D, n = 996 AZs from 6 NMJs from 6 larvae), and Unc13A (E, n = 984 AZs from 6 NMJs from 6 larvae) at third instar M4 NMJs. F–J, Correlation between evoked P_r and AZ abundance (fluorescent intensity) of Liprin-α (F, n = 913 AZs from 6 NMJs from 6 larvae), Syd 1 (G, n = 525 AZs from 6 NMJs from 6 larvae), RBP (H, n = 407 AZs from 6 NMJs from 6 larvae), Unc13A (I, n = 462 AZs from 8 NMJs from 6 larvae), and Cac (J, n = 486 AZs from 7 NMJs from 6 larvae) at third instar M4 NMJs. Raw values and statistical details are provided in Data S1.

To assay AZ protein abundance and evoked release strength, functional imaging was performed in larvae expressing LexOP-myrGCaMP7s in muscles with Mef2-LexA and a UAS-fluorescently tagged AZ protein in MNs using elav-GAL4. As suggested by their low correlation with Cac abundance, AZ levels of Liprin-α (Pearson's r value = 0.13, Fig. 5F) and Syd1 (Pearson's r value = 0.12, Fig. 5G) poorly predicted evoked P_r compared with the AZ abundance of RBP (Pearson's r value = 0.37, Fig. 5H) and Unc13A (Pearson's r value = 0.52, Fig. 5I). Even though late scaffolds correlated with evoked output, their abundance was slightly less predictive of P_r that Cac AZ levels (Pearson's r value = 0.54, Fig. 5J). Although prior data indicate early AZ scaffolds regulate AZ seeding and synapse formation, our findings suggest their absolute abundance at mature release sites is less critical for Cac levels and P_r than late-arriving AZ proteins like RBP and Unc13A that play a more direct role in SV and Ca²⁺ channel accumulation.

Reduced synaptic activity increases AZ size via a Rab3-independent mechanism

Alterations in synaptic output can occur through both Hebbian and homeostatic mechanisms in response to changes in neuronal activity. At Drosophila NMJs, the segregation of GluRs within a PSD (stronger GluRIIA⁺-containing receptors at the center vs weaker GluRIIB⁺ at the periphery) is enhanced during early stages of synapse formation following elevated neuronal activity (Akbergenova et al., 2018). How changes in neuronal activity influence presynaptic AZ maturation and structure is less clear, though recent studies suggest AZ remodeling can contribute to homeostatic plasticity (Frank et al., 2020). To assay changes in AZ formation and maturation in response to altered neuronal activity, AZ protein area and accumulation were examined following manipulations that decrease synaptic output. Synaptotagmin 1 (Syt1) functions as the Ca²⁺ sensor for synchronous SV fusion and syt1^−/− null mutants display a severe reduction in evoked release (Yoshihara and Littleton, 2002). Pan-neuronal expression of a dominant-negative Syt1 transgene that blocks Ca²⁺ binding by the C2B domain (UAS-Syt1C2B^D416N,D418N, hereafter as Syt1^DN) causes lethality and reduces evoked release to a greater extent than observed in syt1^−/− null mutants (Guan et al., 2017). Similarly, pan-neuronal expression of tetanus toxin light chain (hereafter referred to as UAS-TeNT) cleaves the v-SNARE n-Synaptobrevin (nSyb), causing lethality and abolishing evoked release (Sweeney et al., 1995). To characterize changes in AZ development following disruption to synaptic transmission, single neuron manipulations of synaptic output were performed using a MN1-Ib specific Gal4 driver that expresses in one of the ∼35 larval MNs that uniquely innervates M1 (Aponte-Santiago et al., 2020; Jetti et al., 2023). This approach bypasses confounding effects of whole-animal activity manipulations by restricting synaptic dysfunction to a single MN in each larval hemi-segment. TeNT or Syt1^DN were expressed in MN1-Ib and the distribution of BRP, RBP and Cac was analyzed and compared with control NMJs. Analysis of Cac AZ area revealed a 47% enlargement following TeNT expression and a 21% increase following Syt1^DN expression (Fig. 6A,B). In addition, AZ accumulation of BRP, RBP and Cac were enhanced at MN1-Ib NMJs expressing either TeNT or Syt1^DN (Fig. 6C–E). These data indicate Cac AZ size and the abundance of late AZ scaffolds and Cac channels are enhanced following a reduction in presynaptic release.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Presynaptic silencing results in AZ enlargement via a Rab3-independent mechanism. A, Representative images of Cac, BRP, and RBP immunolabeling within synaptic boutons of third instar M1 NMJs of controls or following expression of TeNT (middle panel) or Syt1^DN (right panel) with the MN1-Ib-Gal4 driver. B, Quantification of AZ Cac area in the three genotypes (Ctrl, n = 18 NMJs from 5 larvae; TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; Syt1^DN; n = 19 NMJs from 5 larvae; p < 0.0001). C–E, Quantification of AZ sum Cac fluorescence (C, Ctrl, n = 18 NMJs from 5 larvae; TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; Syt1^DN, n = 19 NMJs from 5 larvae; p < 0.0001), BRP (D, Ctrl, n = 18 NMJs from 5 larvae; TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; Syt1^DN, n = 19 NMJs from 5 larvae; p < 0.0001) and RBP (E, Ctrl, n = 18 NMJs from 5 larvae; TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; Syt1^DN, n = 19 NMJs from 5 larvae; p < 0.0001) across the three genotypes. F, Representative images of Cac, BRP, and RBP immunolabeling within synaptic boutons of third instar M1 NMJs of controls (left panel) or following expression of rab3 RNAi alone (second panel) or coexpressed with TeNT (third panel) or Syt1^DN (right panel) with the MN1-Ib-Gal4 driver. G, Quantification of AZ Cac area in the four genotypes (Ctrl, n = 19 NMJs from 5 larvae; rab3 RNAi, n = 19 NMJs from 5 larvae; p = 0.0002; rab3 RNAi + TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; rab3 RNAi + Syt1^DN, n = 19 NMJs from 5 larvae; p < 0.0001). H–J, Quantification of AZ sum fluorescence for Cac (H, Ctrl, n = 19 NMJs from 5 larvae; rab3 RNAi, n = 19 NMJs from 5 larvae, p = 0.019; rab3 RNAi + TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; rab3 RNAi + Syt1^DN, n = 19 NMJs from 5 larvae; p < 0.0001), BRP (I, Ctrl, n = 19 NMJs from 5 larvae; rab3 RNAi, n = 19 NMJs from 5 larvae; p = 0.0009; rab3 RNAi + TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; rab3 RNAi + Syt1^DN, n = 19 NMJs from 5 larvae; p < 0.0001), and RBP (J, Ctrl, n = 19 NMJs from 5 larvae; rab3 RNAi, n = 19 NMJs from 5 larvae; p = 0.0009; rab3 RNAi + TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; rab3 RNAi + Syt1^DN, n = 19 NMJs from 5 larvae; p < 0.0001) across the four genotypes. K, Representative images of Cac and HRP immunolabeling within synaptic boutons of M1 NMJs of controls or following expression of rab3 RNAi, Syt1^DN or TeNT alone, or rab3 RNAi coexpressed with TeNT or Syt1^DN, with the MN1-Ib-Gal4 driver. L, Quantification of Cac⁺ AZ number normalized to NMJ HRP area across the six genotypes (Ctrl, n = 22 NMJs from 5 larvae; rab3 RNAi, n = 11 NMJs from 4 larvae, p < 0.0001; TeNT, n = 21 NMJs from 5 larvae; p < 0.0001; Syt1^DN, n = 19 NMJs from 5 larvae; p < 0.0001; rab3 RNAi + TeNT, n = 15 NMJs from 4 larvae; p < 0.0001; rab3 RNAi + Syt1^DN, n = 24 NMJs from 5 larvae; p < 0.0001). Statistical significance determined with one-way ANOVA followed by Tukey's multiple-comparison test. Asterisks denote the following p values: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. Raw values and statistical details are provided in Data S1.

We next examined how changes in AZ size due to activity silencing compared with those previously identified in Drosophila rab3 mutants. Loss of Rab3 cause a striking defect in AZ material accumulation, with Cac channels and late scaffolds hyperaccumulating at a subset of AZs, while other AZs are devoid of these components and only contain early arriving AZ proteins (Graf et al., 2009; Goel et al., 2019). To determine if the increase in AZ size following activity reduction requires the Rab3 pathway, UAS-rab3 RNAi constructs were coexpressed in MN1-Ib with TeNT or Syt1^DN. Expression of rab3 RNAi alone induced synaptic defects at M1 NMJs similar to rab3 mutants, including increased Cac AZ area (Fig. 6F,G), enhanced BRP, RBP and Cac levels at these AZs (Fig. 6H–J), and decreased BRP⁺ AZ density (Fig. 6L). Expression of TeNT or Syt1^DN together with rab3 RNAi led to a synergistic enlargement of Cac AZ area (Fig. 6F,G) and even higher levels of BRP, RBP and Cac accumulation compared with controls or expression of TeNT, Syt1^DN or rab3 RNAi alone (Fig. 6H–J). Beyond the increase in Cac AZ size and protein content, expression of TeNT or Syt1^DN at M1 NMJs also resulted in a reduction in Cac⁺ AZ number (Fig. 6L), indicating neuronal silencing decreases AZ seeding. Co-expression of rab3 RNAi with TeNT or Syt1^DN resulted in a synergistic decrease in AZ density as well (Fig. 6L).

To reduce synaptic activity independent of a blockage in SV fusion by TeNT or Syt1^DN, UAS-RNAi knockdown of the Para voltage-gated Na⁺ channel was performed using the MN1-Ib-Gal4 driver. Like TeNT and Syt1^DN, blocking action-potential generation in MN1-Ib increased AZ Cac area and Cac accumulation (Fig. 7A–C). These effects were also enhanced by co-expression of rab3 RNAi. Given loss of Para disrupts action-potential–triggered evoked release and not spontaneous fusion, the effects of activity reduction on AZ maturation and size are likely downstream of signaling pathways associated with evoked output. Consistent with loss of evoked release being the major driver for AZ size expansion, the Syt1^DN transgene dramatically reduces evoked output while increasing the spontaneous release rate (Guan et al., 2017). Together, these data indicate synaptic inactivity and loss of Rab3 trigger AZ enlargement via distinct mechanisms, with neither pathway alone reaching an upper limit for these processes. We hypothesize neurotransmitter release at individual AZs activates a feedback loop to regulate synapse size and accumulation of synaptic components. At later stages when P_r is established, a negative feedback signal arrests further AZ enlargement and material accumulation. When synaptic activity is reduced, negative feedback is suppressed, and AZs continue to accumulate material and increase in size.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Presynaptic knockdown of the Para sodium channel mimics effects of blocking SV fusion on AZ enlargement. A, Representative images of Cac immunolabeling within synaptic boutons of third instar M1 NMJs of controls (left panel) or following expression of para RNAi (second panel), rab3 RNAi (third panel), or para RNAi + rab3 RNAi (right panel) with the MN1-Ib-Gal4 driver. RNAi-mediated suppression of para also triggers AZ enlargement that is enhanced following Rab3 knockdown. Scale bar, 1 µm. B, Quantification of the M1 AZ Cac area in the four genotypes (Ctrl, n = 15 NMJs from 4 larvae; para RNAi, n = 14 NMJs from 4 larvae; p < 0.0001; rab3 RNAi, n = 15 NMJs from 4 larvae; p < 0.0001; para RNAi + rab3 RNAi, n = 14 NMJs from 4 larvae; p < 0.0001). C, Quantification of M1 AZ Cac sum fluorescence in the four genotypes (Ctrl, n = 15 NMJs from 4 larvae; para RNAi, n = 14 NMJs from 4 larvae; p < 0.0001; rab3 RNAi, n = 15 NMJs from 4 larvae; p = 0.0002; para RNAi + rab3 RNAi, n = 14 NMJs from 4 larvae; p < 0.0001). Statistical significance determined with Student's t test; asterisks denote the following p value: ***p ≤ 0.001. Raw values and statistical details are provided in Data S1.

Distinct effects on AZ seeding in rab3 mutants compared with activity silencing

To examine how synaptic inactivity decreases AZ number, we used CRISPR to endogenously tag Unc13B and Unc13A splice variants in the same transgenic line to directly visualize early and late AZ scaffold distribution in vivo (Fig. 8A). The density of AZs containing the late scaffold protein Unc13A was reduced at third instar M1 NMJs in rab3 mutants (2.4-fold) and larvae expressing either TeNT (1.9-fold) or Syt1^DN (1.8-fold) with the MN1-Ib driver (Fig. 8A,C). In contrast, the density of AZs containing the early scaffold Unc13B was similar between controls and rab3 mutants, while expression of TeNT (1.8-fold) or Syt1^DN (1.8-fold) reduced Unc13B⁺ AZs to a similar extent to Unc13A⁺ sites (Fig. 8A–D). We conclude that unlike Rab3 that controls the distribution of late scaffolds across the AZ population, synaptic inactivity results in a global reduction in AZ seeding that affects early (Unc13B) and late (Unc13A, RBP, BRP, and Cac) AZ components.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Presynaptic silencing reduces seeding of both early and late AZ scaffolds. A, Representative images of Unc13B-mClover and Unc13A-mRuby localization within synaptic boutons of third instar M1 NMJs of controls (top panel) or following expression of rab3 RNAi (second panel), TeNT (third panel), or Syt1^DN (bottom panel) with the MN1-Ib-Gal4 driver. The merged image is shown on the right. Scale bar, 1 µm. B, Quantification of Unc13B⁺ AZ number (per 10 mm²) across the four genotypes (Ctrl, n = 19 NMJs from 5 larvae; rab3 RNAi, n = 28 NMJs from 6 larvae; p = 0.99; TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; Syt1^DN, n = 12 NMJs from 4 larvae; p < 0.0001). C, Quantification of the Unc13A⁺ AZ number (per 10 mm²) across the four genotypes (Ctrl, n = 19 NMJs from 5 larvae; rab3 RNAi, n = 28 NMJs from 6 larvae; p < 0.0001; TeNT, n = 19 NMJs from 5 larvae; p < 0.0001; Syt1^DN, n = 12 NMJs from 4 larvae; p < 0.0001). D, Quantification of the Unc13A⁺/Unc13B⁺ AZ ratio across the four genotypes (Ctrl, n = 19 NMJs from 5 larvae; rab3 RNAi, n = 28 NMJs from 6 larvae; p < 0.0001; TeNT, n = 19 NMJs from 5 larvae; p = 0.07; Syt1^DN, n = 12 NMJs from 4 larvae; p = 0.57). E, Representative images of third instar M4 boutons immunostained for BRP (cyan) after PC of GluRIIE^Maple at the first or second instar stage in rab3 mutants or the first instar stage in controls. BRP⁺ AZs were rarely observed at newly formed green⁺/red⁻ GluR^New PSDs (green arrowheads) compared with older red⁺ PSDs (red arrowheads) after PC at the first instar stage in rab3 mutants. When PC was performed in the second instar stage in rab3 mutants, only older red⁺ PSDs were opposed to AZs that were BRP⁺. AZs opposed to newly formed green⁺/red⁻ PSDs were BRP⁻. Controls showed a normal pattern of synapse maturation, with AZs opposed to older red⁺ PSDs containing BRP and 89% of newly formed green⁺/red⁻ PSDs opposed to AZs being BRP⁺. F, Quantification of BRP⁺ AZs opposed to newly formed green⁺/red⁻ PSDs after PC in the first instar stage in controls and rab3 mutants (Ctrl, 86.6 ± 1.3%; n = 18 NMJs from 3 larvae; rab3, 8.5 ± 2.0%; n = 11 NMJs from 3 larvae; p > 0.001) and after PC in the second instar stage in rab3 mutants (0.9 ± 0.3%; n = 14 NMJs from 3 larvae). G, Relative P_r frequency distribution across the AZ population for rab3 and control M4 NMJs reveals a right-shifted and more homogenous distribution in rab3 mutants. Statistical significance determined with one-way ANOVA followed by Tukey's multiple-comparison test. Asterisks denote the following p values: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; and ****p ≤ 0.0001. Raw values and statistical details are provided in Data S1.

One hypothesis for the rab3 phenotype is that the earliest forming AZs require a distinct maturation mechanism coupled to axonal pathfinding and target recognition that is Rab3-independent. Addition of late scaffolds and Cac to AZs formed during subsequent NMJ expansion would then require Rab3 function. This model predicts only AZs formed in the embryo and first instar stage would contain late AZ material in rab3 mutants and that evoked P_r heterogeneity would be reduced given mature AZs would be more age-matched than controls. To test this hypothesis, GluRIIE^Maple was expressed in controls and rab3 mutants, and PC was performed at different larval stages. Larvae were subsequently immunostained to determine the presence of BRP at third instar AZs (Fig. 8E,F). Control NMJs PC at the first instar stage and assayed at the third instar stage had 87% of GluR^New PSDs opposed by BRP-containing AZs (Fig. 8F), consistent with a delay in late scaffold accumulation at some newly formed synapses. In contrast, 99.9% of GluR^Old PSDs were opposed by BRP-containing AZs (Fig. 8F), indicating all mature release sites have accumulated BRP. In the case of rab3 mutants, if BRP⁺ AZs were only opposed to GluR^Old PSDs, it would argue that BRP cannot accumulate at AZs formed later in development in rab3 mutants. Consistent with this model, BRP colocalized with just 8.5% of opposed PSDs that only contained GluR^New in rab3 mutants when PC was performed at the first instar stage, compared with 87% in controls (Fig. 8E,F). Larvae PC at the second instar stage to ensure labeling of only the later wave of new synapses revealed an even lower percentage (0.9%) of BRP⁺ AZs opposed to GluR^New PSDs in rab3 mutants. Consistent with AZ age and material accumulation correlating with evoked P_r, quantal imaging in rab3 mutants demonstrated P_r distribution was more homogenous and shifted toward higher P_r than controls (Fig. 8G). These data indicate early formed AZs contain late scaffolds and Cac channels in rab3 mutants, with AZs formed later in development requiring Rab3 to accumulate these proteins. The preferential effect on late AZ proteins in rab3 mutants contrasts with activity silencing, which reduces the number of AZs containing both early and late scaffolds.

Synaptic inactivity increases T-bar area while loss of Rab3 causes multiple T-bars per AZ

Transmission electron microscopy (TEM) of Drosophila synapses demonstrated presynaptic AZs are highlighted by a central electron-dense body referred to as the T-bar (Atwood et al., 1993). Most AZs contain a single T-bar composed of scaffolding proteins like BRP (Fouquet et al., 2009; Wichmann and Sigrist, 2010), suggesting a modular organization for these release sites. To compare how loss of Rab3 and synaptic inactivation alter the ultrastructure of individual AZs, TEM was performed at larval M1 NMJs from controls or lines expressing TeNT, rab3 RNAi, or both using the MN1-Ib driver (Fig. 9A). Silencing synaptic activity resulted in a 39% increase in the size of individual T-bars (Fig. 9B, measured as length of the top T-bar platform) but did not cause increased incorporation of T-bars at single AZs in MN1-Ib NMJs (Fig. 9C). Loss of Rab3 resulted in a distinct phenotype, with no change in the individual T-bar size but a ∼1.5-fold increase in the number of T-bars per AZ as previously reported (Graf et al., 2009). Reducing activity together with Rab3 depletion resulted in more severe AZ disorganization, with multiple enlarged T-bars per release site (Fig. 9A–C). In contrast to the increase in T-bar size or number, the length of the electron-dense synaptic cleft was largely unaffected, with only a mild enhancement following TeNT expression in rab3 knockdown larvae (Fig. 9D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Presynaptic silencing increases T-bar area, while rab3 mutants contain multiple T-bars per AZ. A, Representative EM images of synaptic boutons at third instar M1 NMJs of controls (top panel) or following expression of TeNT (second panel), rab3 RNAi (third panel), or rab3 RNAi + TeNT (bottom panel) with the MN1-Ib-Gal4 driver. Scale bar, 200 nm. B, Quantification of T-bar size (length of the top T-bar platform) across the four genotypes (Ctrl, n = 25 T-bars from 3 larvae; TeNT, n = 30 T-bars from 3 larvae; p < 0.0001; rab3 RNAi, n = 39 T-bars from 3 larvae; p = 0.88; rab3 RNAi + TeNT, n = 33 T-bars from 3 larvae; p = 0.0007). C, Quantification of T-bar number per AZ across the four genotypes (Ctrl, n = 25 AZs from 3 larvae; TeNT, n = 29 AZs from 3 larvae; p = 0.74; rab3 RNAi, n = 28 AZs from 3 larvae; p = 0.04; rab3 RNAi + TeNT, n = 21 AZs from 3 larvae; p < 0.0001). D, Quantification of synapse length (measurement of synaptic cleft electron-dense material length) across the four genotypes (Ctrl, n = 26 synapses from 3 larvae; TeNT, n = 37 synapses from 3 larvae; p = 0.90; rab3 RNAi, n = 45 synapses from 3 larvae; p = 0.09; rab3 RNAi + TeNT, n = 27 synapses from 3 larvae; p = 0.001). E, Representative confocal and MINFLUX images of synaptic boutons at third instar M1 NMJs of controls (top panel) or following expression of TeNT (second panel), rab3 RNAi (third panel), or rab3 RNAi + TeNT (bottom panel) with the MN1-Ib-Gal4 driver. The red dashed line denotes a representative BRP cluster diameter measurement. Scale bar, 500 nm. F, Representative SIM images of synaptic boutons at third instar M1 NMJs of controls (top panel) or following expression of TeNT (second panel), rab3 RNAi (third panel), or rab3 RNAi + TeNT (bottom panel) with the MN1-Ib-Gal4 driver. The blue dashed line denotes a representative BRP ring diameter measurement. Scale bar, 1 mm. G, Quantification of BRP cluster diameter measured with MINFLUX across the four genotypes (Ctrl, n = 49 AZs from 3 larvae; TeNT, n = 53 AZs from 3 larvae; p = 0.0001; rab3 RNAi, n = 65 AZs from 3 larvae; p < 0.0001; rab3 RNAi + TeNT, n = 28 AZs from 3 larvae; p < 0.0001). H, Quantification of BRP ring diameter measured with SIM across the four genotypes (Ctrl, n = 45 AZs from 3 larvae; TeNT, n = 59 AZs from 3 larvae; p < 0.0001; rab3 RNAi, n = 24 AZs from 3 larvae; p = 0.71; rab3 RNAi + TeNT, n = 95 AZs from 3 larvae; p < 0.0001). Statistical significance determined with one-way ANOVA followed by Tukey's multiple-comparison test. Asterisks denote the following p values: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. Raw values and statistical details are provided in Data S1.

To examine T-bar morphology at the nanoscale, MINFLUX imaging (Schmidt et al., 2021) that has a resolution comparable to EM was used, allowing visualization of filamentous BRP clusters at AZs (Fig. 9E). This single-molecule localization technology determines initially unknown fluorophore positions by relating it to the known zero-intensity position of a donut-shaped excitation beam. The diameter of BRP clusters increased following TeNT or Rab3 RNAi expression and was further enlarged when TeNT and Rab3 RNAi were coexpressed (Fig. 9E,G). Structural illumination microscopy (SIM) provided an optimal approach to image individual BRP AZ rings in larvae (Fig. 9F). Expression of TeNT resulted in a 53% enlargement in BRP ring diameter, while Rab3 reduction did not alter ring diameter but displayed multiple BRP rings at single AZs (Fig. 9H) as observed with TEM. When TeNT and Rab3 RNAi were combined, larger BRP rings were present as clusters within individual AZs (Fig. 9F,H). These data indicate activity reduction increases the AZ size via a morphologically distinct process compared with loss of Rab3.

AZ enlargement downstream of neuronal silencing is accompanied by changes in BRP turnover and requires postsynaptic GluRIIA-mediated signaling

One hypothesis for enhanced AZ material accumulation and increased T-bar size is that silenced neurons increase transcription or translation of key AZ proteins to compensate for reduced synaptic output, resulting in increased transport and delivery of these proteins across development. Alternatively, the turnover of key AZ proteins could be reduced, leading to increased abundance and accumulation at release sites. To assay these potential mechanisms, an endogenously tagged photoconvertible BRP strain was generated by inserting mEosEM (Paez-Segala et al., 2015) into a previously described mimic site (Nagarkar-Jaiswal et al., 2015) within the Brp locus using CRISPR. mEosEM-tagged BRP localized to AZs as expected and could be completely and irreversibly photoconverted from green (hereafter referred to as BRP^New) to red (hereafter referred to as BRP^Old) using UV illumination through the cuticle (Fig. 10A) as previously described for GluR^Maple. To determine if activity reduction altered BRP delivery or turnover, we expressed UAS-TeNT with MN1-Ib-GAL4 in the mEosEM-tagged BRP line. Following PC of silenced M1 NMJs and unaffected M9 NMJs in the same animal, BRP turnover was determined by comparing the rate of decay of red⁺ BRP^Old signal between the two NMJs 1 d after PC at 25°C (Fig. 10B,C). At control M9 NMJs, BRP^Old signal decreased by 27.1%, indicating a BRP half-life (T_1/2) of 2.56 d. In contrast, the BRP^Old signal in M1-silenced NMJs in the same larvae decreased only 17.3%, suggesting a BRP T_1/2 of 4.0 d and a ∼56% reduction in turnover within silenced synapses (Fig. 10B,C). To quantify BRP accumulation at AZs, we assayed the amount of newly delivered green⁺ BRP^New signal 24 h following PC (Fig. 10A,D,E). Silenced M1 NMJs displayed an increase in the sum intensity of BRP^New signal per AZ compared with unaffected M9 NMJs in the same larvae, consistent with expansion of the AZ area and more “slots” available for BRP seeding. However, the mean intensity of BRP^New per AZ was not affected, indicating BRP is spread over a larger AZ area. Given BRP^New will likely experience the same enhanced turnover rate during silencing, along with reduced AZ number under these conditions, this approach is poorly suited to precisely quantify BRP delivery rates. In conclusion, we find that neuronal silencing decreases BRP turnover, suggesting this mechanism is likely to contribute to enhanced BRP levels at AZs and the observed expansion in T-bar size.

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

Changes in BRP turnover and role of GluRIIA in AZ enlargement following presynaptic silencing. A, Representative images of mEosEM-tagged BRP before (green BRP^New) and after (red BRP^Old) PC of M9 (control) and M1 (TeNT) NMJs in larvae expressing UAS-TeNT with the MN1-Ib-Gal4 driver. B, BRP^Old sum fluorescence per AZ puncta was compared before and 24 h after PC at M9 (control) and M1 (TeNT) NMJs (Ctrl 0 Hr, 121,279 ± 5,214; n = 28 NMJs from 7 larvae; Ctrl 24 Hr, 88,464 ± 4,428; n = 32 NMJs from 14 larvae; p < 0.0001; TeNT 0 Hr, 187,774 ± 7,831; n = 28 NMJs from 7 larvae; TeNT 24 Hr, 155,278 ± 6,958; n = 32 NMJs from 14 larvae; p = 0.0029). C, Decay in BRP^Old as a percentage of the signal detected immediately after PC (Ctrl, 72.94% ± 3.65; n = 32 NMJs from 14 larvae; TeNT, 82.69 ± 3.71; n = 32 NMJs from 14 larvae; p = 0.0034). D, Quantification of mean BRP^New fluorescence 24 h post-PC (normalized by pixel area of each AZ; Ctrl, 6,315 ± 613.4; n = 32 NMJs from 14 larvae; TeNT, 6,795 ± 533.8; n = 32 NMJs from 14 larvae; p = 0.0647). E, Quantification of sum BRP^New fluorescence 24 h post-PC (Ctrl, 84,669 ± 7,899; n = 32 NMJs from 14 larvae; TeNT, 110,841 ± 8,890; n = 32 NMJs from 14 larvae; p < 0.0001). F, Representative images showing BRP-NEMOm fluorescence in response to Ca²⁺ influx during 10 Hz stimulation in M9 (control) and M1 (TeNT) NMJs in larvae expressing UAS-TeNT with the MN1-Ib-Gal4 driver. Basal fluorescence of BRP-NEMOm without stimulation is undistinguishable from background (data not shown). G, Quantification of sum BRP-NEMOm fluorescence per AZ (Ctrl, 229,046 ± 15,421; n = 14 NMJs from 7 larvae; TeNT, 437,306 ± 27,187; n = 14 NMJs from 7 larvae; p < 0.0001). H, Quantification of mean BRP-NEMOm fluorescence per AZ (Ctrl, 9,784 ± 241.9; n = 14 NMJs from 7 larvae; TeNT, 11,362 ± 471.1; n = 14 NMJs from 7 larvae; p = 0.0044). Statistical significance determined with paired t test. I, Representative images of Cac (top panel) and GluRIIA (bottom panel) immunolabeling at synaptic boutons of third instar M9 and M1 NMJs of controls, GluRIIA^−/− null mutants or following expression of Syt1^DN or rab3 RNAi with the MN1-Ib-Gal4 driver in control GluRIIA^+/− heterozygotes or GluRIIA^−/− nulls. J, Quantification of the M1/M9 ratio for AZ Cac area across the six genotypes (Ctrl, n = 14 NMJs from 4 larvae; GluRIIA^+/− + Syt1^DN, n = 14 NMJs from 4 larvae; p < 0.0001; GluRIIA^−/− + Syt1^DN, n = 26 NMJs from 6 larvae; p = 0.84; GluRIIA^−/−, n = 14 NMJs from 4 larvae; p = 0.83; GluRIIA^+/− + rab3 RNAi, n = 13 NMJs from 4 larvae; p = 0.0002; GluRIIA^−/− + rab3 RNAi, n = 14 NMJs from 4 larvae; p < 0.0001). K, Quantification of AZ Cac area at control M1 and M9 NMJs and manipulated M1 NMJs across the five genotypes (Ctrl M9, n = 14 NMJs from 4 larvae; Ctrl M1, n = 14 NMJs from 4 larvae; GluRIIA^+/− + Syt1^DN, n = 14 NMJs from 6 larvae; p < 0.0001; GluRIIA^−/− + Syt1^DN, n = 26 NMJs from 4 larvae; p = 0.30; GluRIIA^−/−, n = 14 NMJs from 4 larvae; p = 0.98; GluRIIA^+/− + rab3 RNAi, n = 13 NMJs from 4 larvae; p < 0.0001; GluRIIA^−/− + rab3 RNAi, n = 14 NMJs from 4 larvae; p < 0.0001). L, Quantification of M1 and M9 Cac⁺ AZ number in controls or GluRIIA^−/− null mutants with or without expression of Syt1^DN using the MN1-Ib-Gal4 driver (Ctrl M1, n = 14 NMJs from 4 larvae; Ctrl M9, n = 15 NMJs from 4 larvae; p = 0.99; M1 in MN1-Ib-Gal4 > Syt1^DN, n = 20 NMJs from 5 larvae; p = 0.0003; M9 in MN1-Ib-Gal4 > Syt1^DN, n = 19 NMJs from 5 larvae; p = 0.7; M1 in GluRIIA^−/−, n = 13 NMJs from 4 larvae; p = 0.99; M9 in GluRIIA^−/−, n = 13 NMJs from 4 larvae; p = 0.99; M1 in GluRIIA^−/− + MN1-Ib-Gal4 > Syt1^DN, n = 17 NMJs from 4 larvae; p < 0.0001; M9 in GluRIIA^−/− + MN1-Ib-Gal4 > Syt1^DN, n = 16 NMJs from 4 larvae; p = 0.90). Statistical significance determined with one-way ANOVA followed by Tukey's multiple-comparison test. Asterisks denote the following p values: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001, and ****p ≤ 0.0001. Raw values and statistical details are provided in Data S1.

The changes to AZs in silenced neurons suggest a compensation mechanism to increase synaptic output from individual release sites. Given SV fusion is blocked in silenced neurons expressing TeNT and Syt1^DN, any compensation would fail to generate enhanced evoked output that could be measured electrophysiologically in the postsynaptic cell. However, these manipulations should not disrupt presynaptic Ca²⁺ influx from Cac channels given they target the SV fusion machinery directly. To determine if larger AZs in silenced neurons displayed enhanced function, we compared presynaptic Ca²⁺ influx between silenced and unaffected NMJs in the same animal. Transgenic lines containing the NEMOm Ca²⁺ indicator (Li et al., 2023) attached to the endogenous Brp locus at the N-terminus of the protein that resides near Cac channels were generated using CRISPR. Simulation of MNs at 10 Hz allowed robust detection of presynaptic Ca²⁺ influx that remained confined to single AZs (Movie 2; Fig. 10F). Consistent with enhanced function and increased Cac channels at larger AZs in silenced MNs, TeNT expression in MN1-Ib resulted in a 90.9% increase in presynaptic Ca²⁺ influx per AZ compared with control M9 NMJs in the same animal (Fig. 10G,H).

Changes to AZ structure have also been observed during presynaptic homeostatic plasticity (PHP) that contributes to enhanced SV release that helps maintain normal levels of synaptic output when postsynaptic GluR function is reduced (Weyhersmüller et al., 2011; Böhme et al., 2019; Mrestani et al., 2021). To examine if reduced presynaptic output may utilize a similar pathway as PHP downstream of loss or blockage of GluRIIA, we expressed Syt1^DN in MN1-Ib in GluRIIA null mutants (Han et al., 2023). In controls, the Cac AZ area was indistinguishable at M1 and neighboring M9 NMJs, resulting in a M1/M9 AZ area ratio of ∼1 (Fig. 10I,J). While expression of Syt1^DN in MN1-Ib triggered AZ enlargement at M1 NMJs, resulting in a larger M1/M9 AZ area in heterozygous GluRIIA^+/− controls, expression in GluRIIA homozygous mutants failed to induce an enlargement in the AZ Cac area (Fig. 10J,K). In contrast, loss of GluRIIA did not prevent AZ size expansion in rab3 mutants (Fig. 10J,K), consistent with distinct mechanisms mediating AZ enlargement downstream of Rab3. Although GluRIIA is necessary for AZ enlargement following reduced presynaptic output, it was not required for the corresponding reduction in AZ number induced by Syt1^DN expression (Fig. 10L). These data suggest increased AZ size is not strictly due to enhanced delivery of material over a reduced number of AZs following neuronal silencing. To determine if loss of GluRIIA alone resulted in an increase in AZ area that would preclude further expansion following Syt1^DN expression, AZ Cac area was compared between controls and GluRIIA null mutants. No increase in AZ Cac area was observed in GluRIIA mutants alone (Fig. 10J,K). Together, these data indicate postsynaptic GluRIIA-containing receptors are part of the mechanism for transducing reduced evoked release into retrograde signals that support increased presynaptic AZ size.