Unraveling Synaptic GCaMP Signals: Differential Excitability and Clearance Mechanisms Underlying Distinct Ca²⁺ Dynamics in Tonic and Phasic Excitatory, and Aminergic Modulatory Motor Terminals in Drosophila

Distinct frequency responses of GCaMP Ca²⁺ signals in glutamatergic type Ib, Is, and aminergic type II synapses

We performed a comprehensive Ca²⁺ imaging analysis by targeted expression of GCaMP indicators in the larval NMJ and found striking differences in the ionotropic glutamatergic type Ib (tonic) and type Is (phasic), and metabotropic octopaminergic type II synapses. Most Ca²⁺ imaging studies have used abdominal muscles 6 and 7 (Macleod et al., 2002; Reiff et al., 2005; Lnenicka et al., 2006), which are innervated by types Ib and Is, but not by type II, motor axon terminals. We instead chose muscles 12 and 13 as the primary regions of interest. These two muscles are innervated by all three types of synapses, which are individually identifiable by distinct synaptic bouton sizes and terminal branches (Fig. 1A1), as shown previously in immunostaining studies (Johansen et al., 1989; Budnik et al., 1990; Kurdyak et al., 1994; Monastirioti et al., 1995; Zhong and Wu, 2004; Koon et al., 2011). Type Ib, Is, and II axonal terminals each represent the projection from a different motor neuron. Notably, the phasic type Is synaptic terminals in muscles 6, 7, 12, and 13 are derived from separate axonal branches of a single motor neuron (MNSNb/D-Is; Hoang and Chiba, 2001; see also Lnenicka and Keshishian, 2000), whereas muscles 12 and 13 share the same type II motor neuron input (MNSNb/D-II; Hoang and Chiba, 2001; see also Schmid et al., 1999). However, the tonic type Ib motor terminals in muscle 12 and 13 are separately innervated by different motor neurons (MN12-Ib and MN13-Ib; Hoang and Chiba, 2001).

The motor patterns found in Drosophila larval NMJs usually consist of trains of high-frequency repetitive action potentials (Budnik et al., 1990; Cattaert and Birman, 2001; Fox et al., 2006; Chouhan et al., 2010). To explore how axonal firing frequency determines the GCaMP Ca²⁺ signals in different synaptic boutons, we applied 2-s trains of repetitive stimuli to the motor axons at increasing frequencies (2, 10, 20, and 40 Hz), which are within the normal range of firing rates of type Ib and Is motor axons (Cattaert and Birman, 2001; Chouhan et al., 2010).

We used HL3.1 saline (Feng et al., 2004; Ueda and Wu, 2006) to optimize the expression of the well characterized excitability mutant phenotypes while retaining a desirable property of the HL3 saline (Stewart et al., 1994) of promoting the longevity of the NMJ preparation. Except for some experiments examining Ca²⁺ dependence, the majority of experiments were conducted with saline containing 0.1 mM Ca²⁺, which not only could enhance the hyperexcitable mutant phenotypes but also effectively suppressed muscle contraction during imaging (cf. Ueda and Wu, 2009a,b; Ueda and Wu, 2015). This approach also circumvented the use of high-concentration glutamate for postsynaptic receptor desensitization to suppress muscle contraction during imaging (Macleod et al., 2002, 2004, 2006; Lnenicka et al., 2006) and thus minimized potential complications from activating presynaptic metabotropic glutamate receptors (Zhang et al., 1999).

At low frequencies of nerve stimulation (e.g., 2 Hz), we did not detect any significant GCaMP Ca²⁺ signals (ΔF/F) in WT larvae (Fig. 1B1). With 10-Hz stimulation, only type II synapses displayed GCaMP signals (see the pseudo-color map in Fig. 1A2, and example traces with bar graphs in Fig. 1B2 ). GCaMP signals appeared in type I synapses only at higher stimulation frequencies, above 20 Hz for type Is synapses (Fig. 1A3,B3 ) and beyond 40 Hz for type Ib synapses (Fig. 1A4,B4 ).

The distinct frequency responses of GCaMP Ca²⁺ signals in type Ib, Is, and II boutons described above apparently reflect intrinsic differences in synaptic properties and are also evident for NMJs in other ventral and dorsal muscles examined. As demonstrated in Figure 1A, despite the large variations in size and fluorescence intensity (Fig. 1A1), the boutons along the same terminal branch behaved relatively uniformly (Fig. 1A2–A4 ) and consistent in the general wave form (see representative traces in Fig. 1B1–B4), so that clear distinction in the threshold of frequency responses could be established among each type II, Is, and Ib motor terminals. More detailed morphometric analyses of Ca²⁺ fluorescent signals demonstrate that the characteristic frequency responses of type Ib, Is, and II synapses were independent of and could not be attributed to differences in basal GCaMP expression levels or synaptic bouton sizes (Xing and Wu, 2016).

We found that the distinction in frequency responses among type Ib, Is, and II synapses remained at higher Ca²⁺ concentrations (with 7 mM glutamate in HL3.1 to suppress muscle contraction; Macleod et al., 2004), with the same sequence of responsiveness to various stimulation frequencies. As shown in Figure 2, at 0.5 mM external Ca²⁺, type Ib synapses remained to be the least responsive and type II the most responsive (20 Hz for type Ib and 2 Hz for type II; Fig. 2B,D), despite the overall enhanced fluorescent signals compared to 0.1 mM Ca²⁺ (Fig. 1B1–B4 ). Consistently, the data also indicate that saturation levels of GCaMP signals were reached at different frequencies for three types of boutons (at 0.5 mM Ca²⁺, 10 Hz for type II, ∼40 Hz for type Is, and well above 40 Hz for type Ib; Fig. 2B,D).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Frequency responses of GCaMP1.3 Ca²⁺ signals at different extracellular Ca²⁺ concentrations. A, B, Example ΔF/F traces for type Ib, Is, and II synaptic boutons stimulated at 2, 10, 20, and 40 Hz in 0.1 and 0.5 mM Ca²⁺ saline. C, D, Summary line plots showing the frequency responses of averaged max ΔF/F for type Ib, Is, and II synapses at 0.1 mM Ca²⁺ (C; 8–11 NMJs, same set of data as in Fig. 1 replotted for comparison) and 0.5 mM Ca²⁺ (D; four NMJs). In this and Figures 4, 10, 13, for each NMJ, one synaptic terminal was measured for each type of motor synapses and readings of all boutons within the same terminal were averaged. See Table 1 for total bouton numbers and corresponding NMJ numbers. For clarity, some data points are displaced horizontally and statistically significant differences against type Ib synapses are indicated at each stimulus frequency. In this figure and Figures 4, 10, 13, 14, error bars indicate SEM. Asterisks denote levels of statistically significant differences (*p < 0.05; **p < 0.01; ***p < 0.001), based on KW tests and Bonferroni correction.

The relative differences in GCaMP signal frequency responses remained unaltered for the various GCaMP indicators with different sensitivity and subcellular localization. Importantly, as Figure 3 shows, similar distinctions in frequency responses were obtained among the three synaptic types with myrGCaMP5 (Melom and Littleton, 2013; dissociation constant K_d = 447 nM, Akerboom et al., 2012), GCaMP6m (K_d = 167 nM; Chen et al., 2013), and GCaMP1.3 (K_d = 234 nM; Nakai et al., 2001), despite their differences in affinity and sensitivity. Additional indicator GCaMP6f and the Gal4 driver c386Y produced the same frequency responses. In particular, MyrGCaMP5, which is localized to plasma membrane (Melom and Littleton, 2013), still produced the same frequency response differences, suggesting that surface-volume ratio differences among type Ib, Is, and II synaptic boutons did not significantly contribute to their distinct GCaMP signal frequency dependencies, which was consistent with previous report on mammalian neurons (Mao et al., 2008).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Signals generated by GCaMP indicators of different sensitivity and subcellular localization in response to 10-, 20-, and 40-Hz stimulation. A–C, Example ΔF/F traces for type Ib, Is, and II synaptic boutons recorded with different indicators, c164/GC6m, c164/myrGC5, and c164-GC1 (standing for genotypes: c164-Gal4/UAS-GCaMP6m, c164-Gal4/+; UAS-myrGCaMP5/+ and +/Y; c164-GCaMP1.3), in 0.1 mM Ca²⁺ saline at the designated stimulation frequencies. Consistently, GCaMP signals of type II synapses appeared in the lowest and type Ib the highest stimulus frequency ranges, regardless of the indicator types.

Differential excitability control of residual Ca²⁺ dynamics by K⁺ channels in type Ib, Is, and II synapses

Previous work has shown that neuronal hyperexcitability caused by mutations or drug blockage of the various K⁺ channels could lead to greatly enhanced transmitter release often associated with different patterns of repetitive firing of axonal action potentials at Drosophila larval NMJs (Ganetzky and Wu, 1982a, 1983, 1985; Ueda and Wu, 2006). Since GCaMP Ca²⁺ imaging enabled detection at a subcellular resolution to indicate variations in excitability properties among the three categories of synapses, we examined the effects of altering specific K⁺ channels pharmacologically. The differentiation power was further enhanced by systematic analyses of mutants of identified K⁺ channels (Table 1).

TEA and 4-AP are two well-characterized, commonly used K⁺ channel blockers known to cause neuronal hyperexcitability in Drosophila NMJs (Jan et al., 1977; Ganetzky and Wu, 1983, 1985; Fox et al., 2005; Ueda and Wu, 2006), as well as other species (Thesleff, 1980; Hille, 2001). Both TEA (10 mM) and 4-AP (200 μM) treatments shifted the frequency responses so that substantial GCaMP Ca²⁺ signals appeared at lower stimulus frequencies (Table 1, compare WT with 4-AP and TEA rows). However, as shown in Figure 4, striking preferential effects were apparent, with TEA strongly enhancing GCaMP signals in type II (at frequencies as low as 2 Hz; Fig. 4A,B) and 4-AP greatly increasing that in type Is synapses (10 Hz; Fig. 4C,D). For type Ib synapses, the effects of TEA and 4-AP were milder and only evident at high stimulation frequencies (above 40 Hz; Table 1). Notably, Sh mutations (Sh^M, null allele, Sh¹²⁰, point mutation) and 4-AP closely resemble in their differential effects on type Ib, Is, and II synapses (Fig. 4; Table 1), consistent with the fact that 4-AP specifically blocks the Sh K_V1 channels that mediate fast-inactivating transient K⁺ current I_A in muscle (Salkoff and Wyman, 1983; Wu and Haugland, 1985). Unlike 4-AP, TEA has a different and wider spectrum of action on several types of K⁺ channels at tens of millimolar concentrations (Koketsu, 1958; MacKinnon and Yellen, 1990; Hille, 2001). The contrasting preferential effects of 4-AP on type Is and TEA on type II raise the possibility of different K⁺ channel compositions in type Ib, Is, and II synapses.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Genetic and pharmacological dissection of the distinct excitability of type Ib, Is, and II synapses. Representative ΔF/F traces under 2- and 10-Hz stimulations (A, C), comparing WT control, WT with TEA (10 mM) and 4-AP (200 μM) treatments, Sh^M mutant and double mutant eag¹ Sh¹²⁰. Note the preferential effects of TEA on type II, and 4-AP, Sh^M and eag¹ Sh¹²⁰ on type Is synapses. B, D, Bar charts for sample statistics corresponding to the traces in A, C, with NMJ numbers and SEMs indicated. KW tests with Bonferroni correction were performed; *p < 0.05, **p < 0.01, ***p < 0.001.

One of the most striking demonstrations of extreme hyperexcitability in Drosophila NMJs can be seen in eag Sh double mutants, in which axon bundles display high-frequency (∼100 Hz) repetitive firing of supernumerary action potentials triggered by just one nerve stimulus (Ganetzky and Wu, 1982a, 1983, 1985; Wu et al., 1983). However, it has not been resolved how type Is and Ib terminals each contributes to this extreme phenotype. As shown in Figure 4, in the double-mutant eag¹ Sh¹²⁰, type Is was clearly the most drastically affected, since robust single stimulus-evoked GCaMP signals could be observed only in type Is, but not Ib, synapses (Fig. 4A, 2 Hz), and could not be seen in either Sh or eag single mutants (Table 1). Enhanced GCaMP signals in type Ib were seen only when stimulated beyond 40 Hz (Table 1, first column) while type II synapses only occasionally displayed single stimulus-evoked responses (see further details below). Taken together, these observations demonstrate type Is motor axon activities as the source of the supernumerary repetitive axonal action potentials in eag Sh double mutants and support the notion of different combinations of K⁺ channel subtypes in type Ib, Is, and II synapses.

To extend the above findings based on GCaMP1.3 measurements, we performed a separate set of experiments with GCaMP6m, an improved version of GCaMP with increased sensitivity (compare Fig. 3), for a more systematic analysis of the distinctions in K⁺ channel functioning among type Ib, Is, and II synapses (Figs. 5, 6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Distinct effects of various genetic and pharmacological perturbations of K⁺ channels on type Ib, Is, and II synapses at different stimulus frequencies. Note that these perturbations preferentially lowered the frequency threshold for GCaMP signals in type Ib, Is, and II synapses. A, +; +; nSyb-GCaMP6m larvae used as WT control (0.1 mM Ca²⁺, top-row traces), examined by applying 4-AP (200 µM), quinidine (abbreviated as Q, 20 µM), and TEA (10 mM). Black trace, 2-Hz stimulation; blue, 10 Hz; red, 20 Hz. Note the drastic frequency response change in type II synapses (2 Hz, Q and TEA), in type Is (10 Hz, 4-AP), and in type Ib (20 Hz, 4-AP). B, Genetic perturbation of GCaMP signals examined with Sh^M/Y;; nSyb-GCaMP6m/+, eag¹/Y;+; nSyb-GCaMP6m/+, and UAS-ShabRNAi/+; nSyb-GCaMP6m/+. Three to five larvae for each condition were tested and consistent results were obtained.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Extreme excitability indicated by single pulse-evoked giant GCaMP signals: double and triple insults of combined mutation and drug effects. GCaMP6m signals (ΔF/F) in response to 2-Hz (black) and10-Hz (blue) stimulation from larvae carrying the nSyb-GCaMP6m construct (0.1 mM Ca²⁺). nSyb-GCaMP6m, used as WT control (0.1 mM Ca²⁺), examined by coapplication A, Double insults of combined effects of 4-AP (200 µM) + quinidine (Q, 20 µM), Sh^M + quinidine (20 µM), ShabRNAi + 4-AP (200 µM), and 4-AP (200 µM) + TEA (10 mM). Note the robust GCaMP signals in type II and Is synapses evoked by individual stimuli (2 Hz) and the preferential response of type Ib synapses to combined insults of 4-AP + Q or ShabRNAi (10 Hz). B, Triple insults of K⁺ channel perturbations with Sh^M + Q (20 µM) + TEA (10 mM), or ShabRNAi + 4-AP (200 µM) + TEA (10 mM). Note the single pulse-evoked giant GCaMP signals in type Ib synapses (2 Hz), indicating a state near the ceiling effect of hyperexcitability. Also note that hyperexcitability resulted in interrupted, aborted or rebounding GCaMP signals in type II synapses (2 and 10 Hz) and rebounds of type Is responses (10 Hz). Based on observations from three to five larvae for each condition.

For the ease of comparisons, representative traces of GCaMP6 responses to 2-, 10-, and 20-Hz stimulation are superimposed for each genotype and condition. For the cases of altered type II synapses, mutations of eag channels, as well as ShabRNAi knockdown or quinidine (abbreviated as Q in all figures) inhibition of Shab K_V2 channels (Singh and Wu, 1989; Wu et al., 1989; Singh and Singh, 1999; Ueda and Wu, 2006), led to clearly discernable GCaMP responses to individual nerve stimuli delivered at 2 Hz in HL3.1 saline containing 0.1 mM Ca²⁺ (Fig. 5; 2-Hz traces). Even more robust effects were obtained with TEA treatment, which has a broad-spectrum action and can act on Shab and eag channels (Pak et al., 1991; Brüggemann et al., 1993). However, type II synaptic GCaMP signals were less affected by Sh mutations or 4-AP treatment, consistent with the GCaMP1.3 results (Fig. 4; Table 1).

In comparison, type Is synapses were most strongly influenced by manipulations of Sh channels, with some small GCaMP responses to individual stimuli in the 2-Hz stimulus train detectable in Sh or 4AP-treated WT NMJs (Fig. 5, middle column). For type Ib synapses, however, none of the above genetic or pharmacological manipulations was effective in producing GCaMP responses with stimulus frequencies below 20 Hz (Fig. 5, middle column). These GCaMP6m observations confirm the conclusions drawn from the experiments based on GCaMP1 (Fig. 4; Table 1) regarding the differential excitability control of by K⁺ channels in type Ib, Is, and II synapses. Taken together, it could be seen that type II synapses were most sensitive to manipulations of eag and Shab channels, whereas type Is synapses appeared to be particularly vulnerable to disruptions of Sh channels, and type Ib synapses were most resistant to K⁺ channel perturbations.

Extreme hyperexcitability and motor terminal repetitive firing

Apparently, a ceiling effect of hyperexcitability is indicated by the robust responses to individual stimuli (delivered in the 2-Hz nerve stimulation), as seen in TEA-treated type II synapses (Fig. 5, right column). We found that type Is synapses could also reach this ceiling effect at 0.1 mM Ca²⁺ but only after double insults of the combined perturbations of Sh with either eag or Shab channels. Similar to the GCaMP1 results of eag Sh double mutants (Fig. 4), greatly enhanced GCaMP6m signals in response to individual nerve stimuli could be seen in type Is synapses by the joint effects of 4AP (or Sh mutations) with either quinidine (or ShabRNAi) or TEA (Fig. 6A, middle column, 2-Hz traces).

In contrast, type Ib synapses remained largely nonresponsive at 2 Hz during the above double perturbations (Fig. 6A, left column). It required at least 10 Hz for type Ib synapses to exhibit clearly detectable GCaMP signals even with both Sh and Shab channels disrupted (Fig. 6A; for single perturbation results, see Table 1; Fig. 5). However, type Ib synapses could display the hallmark of hyperexcitability if given the condition of triple insults with mutations or blockers of K⁺ channels. With triple insults, i.e., combinations of TEA and 4AP (or Sh) plus quinidine (or ShabRNAi), type Ib synapses exhibited single pulse-evoked robust GCaMP signals in 0.1 mM Ca²⁺ saline (Fig. 6B, left column, 2-Hz traces), directly comparable to the ceiling level of hyperexcitability found in type Is and type II synapses (Figs. 5, 6A ). Therefore, compared to type II and Is synapses, type Ib synapses appeared to possess a greater K⁺ channel-mediated repolarization capacity and were hence more resilient to disruptions in different types of K⁺ channels.

To understand how the modified GCaMP signals were correlated with the amplitude and frequency of the synaptic transmission events, we performed simultaneous focal recording in conjunction with GCaMP imaging on the same synaptic boutons. efEJPs follow the faster time course of excitatory junctional currents (EJCs) as compared to EJPs and their amplitudes are linearly proportional to the local EJCs generated by the synaptic release in region under the electrode tip (Fatt and Katz, 1952; Sakmann and Neher, 1984). This approach enabled a direct correlation of postsynaptic response induced by transmitter release with the presynaptic GCaMP signals triggered by nerve stimulation. At low external Ca²⁺ (0.1 mM), both type Ib and Is produced only small efEJP responses at a level of quantal fluctuation in WT larvae and there were no detectable signals from GCaMP1.3 and GCaMP6m indicators (Figs. 7, 8). However, we found both GCaMP1.3 and GCaMP6m indicators were unable to produce detectable signals even when efEJPs were clearly discernable. For example, type Ib and Is boutons in Sh larvae (Fig. 7), or in WT following 4-AP and quinidine treatment (Fig. 8), produced greatly enhanced efEJPs that were not registered by either GCaMP1.3 or GCaMP6m.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Failure to detect presynaptic GCaMP signals when synaptic transmission events were registered by efEJPs. A, Traces of GCaMP1.3 ΔF/F (blue) and efEJPs (black) at 2-Hz stimulation collected from the same boutons of type Ib and Is motor terminals of WT control (c164-GCaMP1.3) and Sh¹²⁰/Y; c164-GCaMP1.3. Each stimulus is indicated by a gray-scaled dot. B, Expanded traces of efEJPs in response to the sequential stimuli that are shown in A. Note that no detectable GCaMP signals (A) despite significant transmissions (B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Correlation of giant GCaMP signals with repetitive efEJPs evoked by individual stimuli. A, ΔF/F signals of the GCaMP6m indicator correlated with simultaneous focal recording of efEJPs from the same boutons in type Ib and Is motor terminals (2-Hz stimulation in 0.1 mM Ca²⁺ HL3.1 saline). K⁺ channel blockade with 4-AP (200 µM) and quinidine (Q, 20 µM) in nSyb-GCaMP6m larvae led to giant GCaMP signals in type Is, but not Ib, synapses. Expanded efEJP traces (20 and 100 times, shaded time windows) in B, C indicate supernumerary transmitter releases (arrowheads) following each nerve stimulus (arrow).

A notable exception was during the robust 2-Hz peaks of GCaMP6 signals (the hallmark hyperexcitability ceiling effect) in type Is synapses at low external Ca²⁺ (0.1 mM) after double insults of K⁺ channels, such as application of 4-AP plus quinidine, where giant efEJPs occurred in correlation with each GCaMP ΔF/F peak (Fig. 8, right column; compare Fig. 6, middle column,). A closer examination of these highly nonlinear events revealed that these efEJPs consisted of massive discharge of multiple release of transmitters (expanded type Is efEJP traces in Fig. 8; and also Fig. 9 for eag Sh double mutant effect). Further investigation with triple recordings to correlate axonal action potentials with efEJP and GCaMP signals demonstrated that these massive efEJP discharges underlying individual GCaMP ΔF/F peak were evoked by high-frequency supernumerary firing of motor axons (∼100 Hz; Fig. 9).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Electrophysiological basis of single pulse-evoked giant GCaMP signals: repetitive firing due to extreme hyperexcitability of the motor axon. GCaMP signals (ΔF/F traces) of type Ib (A1) and Is (B1, C1) synapses from three eag¹ Sh¹²⁰/Y; c164-GCaMP1.3 larvae at 2-Hz stimulation in 0.1 mM Ca²⁺ HL3.1 saline. Corresponding simultaneous records of efEJPs (A2, B2) and nerve action potentials (C2) are displayed in black with the associated ΔF/F traces in blue (expanded from A1, B1, C1 with individual nerve stimuli marked by arrows). One of the two shaded areas in A2, B2, C2 are further expanded to show efEJPs with multiple peaks (arrowheads), indicative of repetitive nerve action potentials, which were more robust in type Is than Ib synapses (A3, B3). En passant nerve recording revealed motor axon high-frequency repetitive firing (over 100 Hz; C3).

It is important to note that in all of the experiments reported here, the segmental nerve was severed from the ventral ganglia. Therefore, the repetitive action potentials were generated within the axon and were not driven by CNS activities mediated by the motor neurons. Thus, these hallmark, ceiling effects of GCaMP signals reflect axonal membrane hyperexcitability and are consistent with previous reports of multiple firing of motor axons associated with prolonged, giant EJPs at low external Ca²⁺ levels found in eag Sh (Ganetzky and Wu, 1982a) and Sh Shab (Ueda and Wu, 2006) double mutants.

Differential effects of Na⁺ channel mutations on GCaMP signals and nerve conduction failure in type II motor axons

The above observations reveal distinct repolarization mechanisms of the three synaptic terminals, presumably reflecting differential expression of K⁺ channel genes. We found similarly disparate sensitivities to Na⁺ channel mutations among the three types of synapses, although in Drosophila para is the only known gene that encodes voltage-gated Na⁺ channel (Na_V1), which regenerates a large number of splice variants (Loughney et al., 1989; Thackeray and Ganetzky, 1994; Olson et al., 2008; Lin et al., 2012). To investigate the role of the Na⁺ channel, we took advantage of two well-characterized para alleles that confer drastic but opposite effects on the membrane depolarization mechanism. The hypoexcitable, temperature-sensitive allele para^ts1 is thought to decrease the synthesis of functional Na⁺ channels (Thackeray and Ganetzky, 1994), resulting in increased axonal refractory period and higher sensitivity to Na⁺ channel-specific toxins (Wu and Ganetzky, 1980; Suzuki and Wu, 1984), and becomes paralyzed at high temperature due to axonal conduction failure (Suzuki et al., 1971; Siddiqi and Benzer, 1976; Ganetzky et al., 1986). In contrast, the hyperexcitable, bang-sensitive allele bss1 displays seizure behaviors on mechanical stress (Ganetzky and Wu, 1982b; Burg and Wu, 2012), which is associated with seizure-like nerve spike discharges in central circuits (Lee and Wu, 2006; Parker et al., 2011) and increased synaptic transmission with axonal multiple firing at the larval NMJ (Ganetzky and Wu, 1982b; Giachello and Baines, 2015).

The GCaMP1 imaging demonstrated that like Sh K⁺ channel mutations, the hyperexcitable para allele bss1 had a strong preferential effect on type Is over Ib synapses (Fig. 10A, 10 and 20 Hz; Table 1). Notably, GCaMP signals were significantly enhanced in type Is boutons on stimulation at 10 and 20 Hz, whereas little effect on type Ib synapses was detectable up to 40 Hz. This preferential effect again raises the possibility of differential expression or posttranslational modifications of Na⁺ channel slice isoforms (Thackeray and Ganetzky, 1994; Olson et al., 2008; Lin et al., 2012). However, the bss effect on type Ib could be revealed once combined with Sh¹²⁰ since the double mutant produced much larger response than either single mutant alone (Table 1). On the contrary, in both type Ib and Is synapses, the hypoexcitable temperature-sensitive para allele ts1 produced only small, but statistically significant, reductions in the GCaMP signals at room temperature (Table 1; Fig. 10A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

Differential effects of Na⁺ channel mutations on the Ca²⁺ dynamics of type Ib, Is, and II synapses. GCaMP Ca²⁺ signals of type Ib and Is (A) and II (B) NMJs from a hypoexcitable allele, para^ts1 (para) and a hyperexcitable allele, para^bss1 (bss) are compared, with representative ΔF/F traces at specified frequencies shown in left panels and the corresponding frequency dependence of max ΔF/F in right panels (mean and SEM, of 6–11 NMJs for each genotype, compare Table 1). Note that type II synaptic boutons in bss frequently displayed “intermittent” or “aborted” GCaMP signals on repetitive stimulation, lowering the average max ΔF/F above 10 Hz (see Fig. 11 for details). KW ANOVA with Bonferroni corrections were performed (*p < 0.05, **p < 0.01).

In comparison, type II synapses in both ts1 and bss1 alleles produced more striking, clear-cut alterations. The hypoexcitable ts1 mutation results in significantly smaller GCaMP signals in type II synapses even at room temperature (10 and 20 Hz; Fig. 10B). Notably, a significant fraction of the stimulation trials produced no response at all (20-Hz sample trace in Fig. 10B), indicating axonal conduction failure. This lowered safety margin for action potential propagation in the type II motor axons could be explained by the higher longitudinal internal resistance for axons with smaller diameters (Aidley, 1998), making them more vulnerable to the effect of reduced density or modified function in Na⁺ channel that determine the depolarizing Na⁺ current across the plasma membrane. Such axonal conduction failure was not observed in the larger type Ib or Is motor terminals in para^ts1 NMJs. For the hyperexcitable allele bss1, a striking nonlinear frequency response was illustrated by the GCaMP signals in type II synapses. Greatly enhanced GCaMP signals were seen at 2 Hz, with discernable responses to individual stimuli (compare type II, 2-Hz traces in Figs. 5, 10B). However, at increased stimulus frequencies a suppression of GCaMP responses was seen, concurrently with a novel, unexpected pattern of interrupted or aborted GCaMP signals. These events occurred in the rising phase of GCaMP signals during repetitive stimulation (Fig. 11), resulting in a net decrease in the average Max ΔF/F in bss1 (Fig. 10B, plot, 20–40 Hz). Boutons of the same type II synaptic terminal in the same NMJ behaved uniformly in the interrupted or aborted pattern, with the timing and kinetics entirely the same. Such seemingly puzzling GCaMP responses suggest intermittent axonal AP conduction failure at increasing stimulus frequencies.

• Download figure
• Open in new tab
• Download powerpoint

Figure 11.

Variability in the wave form of GCaMP signals in type II synaptic terminals at high frequencies of stimulation. Representative regular (normal) and irregular (aborted, intermittent, and no response) ΔF/F traces at 20 Hz from WT, eag, eag Sh, bss, and para type II boutons (with c164-GCaMP1.3) are shown. For each genotype, the numbers of NMJs exhibiting each wave form are indicated together with the total sample size (numbers of NMJs examined). N.F., not found. Note that both hypoexcitable (para^ts1) and hyperexcitable (others) mutations increased the occurrence of irregular waveforms, indicating axonal conduction failure in type II motor terminals (see text). Similar response waveforms were also revealed by GCaMP6m indicator (e.g., for eag¹/Y;; nSyb-GCaMP6m/+; normal: 0/6, aborted: 2/6, intermittent: 3/6, no response: 1/6).

It has been shown that in the oocyte expression system, para^bss1 Na⁺ channels exhibit slower recovery from inactivation (Parker et al., 2011), which could lengthen action potential refractory period. This would make small-diameter type II axons particularly prone to action potential failure at high stimulation frequencies.

It is important to point out that a similarly phenomenon of interrupted GCaMP signals in type II motor terminals was also found in other hyperexcitable K⁺ channel mutants, such as eag and eag Sh (Fig. 11; see also Fig. 4, eag¹ Sh¹²⁰ type II, 10 Hz). This apparently counter-intuitive observation in hyperexcitable K⁺ channel mutants could also be associated with increased action potential refractory period (cf. para^bss1), which made the axons nonresponsive to the upcoming stimuli after excessive firing activities. It is known that in eag¹ Sh¹²⁰ a longer refractory period (Engel and Wu, 1992) results from excessive inactivation of Na⁺ channels following over excitation due to weakened repolarizing K⁺ currents. As shown in Figures 5, 6, different mutational and drug manipulations of K⁺ channels could lead to irregular waveforms of GCaMP6 signals, most pronouncedly in type II synapses. Intermittent, delayed, aborted, or sometimes lost accumulation of the GCaMP signal occurred frequently (Fig. 5, type II 10 and 20 Hz). Following more drastic manipulations (double or triple insults), these irregularities became more extreme, occurring even at 2-Hz stimulation (Fig. 6). Notably, these extreme hyperexcitability conditions also caused GCaMP signal rebound after stimulus cessation, which was encountered more frequently in type II and occasionally in type Is synaptic terminals (Figs. 5, 6).

Differential effects on rise and decay kinetics of alterations in Ca²⁺ influx and clearance

The above irregular waveforms of GCaMP signals reflect either high-frequency axonal repetitive firing or intermittent axonal conduction failure. However, in most cases GCaMP signals registered gradual residual Ca²⁺accumulation and followed a time course of smooth rise during repetitive stimulation and an approximately exponential decay at the cessation of stimulus trains.

In principle, the GCaMP Ca²⁺ signal reflects an integrated dynamic process of Ca²⁺ influx and subsequent clearance of intracellular free Ca²⁺. Thus, analysis of the rise and decay kinetics of GCaMP signals may yield further insight into how alterations in membrane excitability can change the time course of intracellular Ca²⁺ accumulation. To visualize the overall GCaMP Ca²⁺ signal rise and decay kinetics in response to repetitive stimulation, ΔF/F were normalized and overlaid to contrast for changes caused by channel blockers and mutations (Fig. 12; Tables 2, 3). As expected, we found that the rise phase of GCaMP signals was accelerated by acute application of higher external Ca²⁺ concentrations (Fig. 12A), increased stimulus frequencies (Fig. 12B), K⁺ channel blockers (Fig. 12C), and mutations that increase membrane excitability (Fig. 12D). All of these manipulations, including 4-AP and TEA treatments as well as Sh and bss1 mutations, resulted in faster GCaMP signals with shortened peaking or plateauing time. All these changes in rise kinetics were indicated by shortened half-rise time, as quantified and summarized in Table 2. Note that the preferential effects of individual channel blockers and mutations on type Ib, Is, and II synapses are again illustrated in the extents of rise time acceleration (Fig. 12). Single pulse-evoked GCaMP signals showed even shorter peaking time (Figs. 5, 6), consistent with the high-frequency repetitive firing of action potentials triggered by each stimulus pulse in these cases of hyperexcitability (Fig. 9).

• Download figure
• Open in new tab
• Download powerpoint

Figure 12.

Kinetic analysis of presynaptic GCaMP signals with different manipulations enhancing Ca²⁺ influx. Comparison of original (upper) and normalized (lower) ΔF/F traces collected from individual boutons of (A) WT with 0.1 mM Ca²⁺ (black) versus 0.5 mM Ca²⁺ (blue), (B) WT with 0.1 mM Ca²⁺ under the designated stimulation frequency (black, 40 Hz for Ib, 20 Hz for Is, and 10 Hz for II) versus 80-Hz stimulation (blue), and (C) WT control (black) versus 4-AP (200 µM, red) and TEA (10 mM, blue) treatments, 0.1 mM Ca²⁺. D, WT control (black) versus Sh^M (Sh, red), and bss1 (bss, blue). Type Ib, Is, and II synapses, stimulated at 40, 20, and 10 Hz, respectively, in 0.1 mM Ca²⁺. Note clear alterations in the rise kinetics with minimal effects in the decay kinetics indicated by superimposing normalized traces. See Tables 2, 3 for ensemble data of rise and decay kinetics.

Remarkably, when compared to the half-rise time, the half-decay time of GCaMP signals was much less affected by enhanced membrane excitability (4-AP and TEA in Fig. 12C; Sh and para^bss1 in Fig. 12D; compare Tables 2, 3 for additional genotypes and conditions). This is true also for a higher external Ca²⁺ concentration (0.5 mM; Fig. 12A; Tables 2, 3). These observations apparently reflect an abundant reserve of Ca²⁺ clearance capacities that could still handle the increased influx caused by drug treatments or mutations in our experiments.

However, it should be noted that in extreme hyperexcitability conditions, such as the combined effects of 4-AP, quinidine, and TEA, long-lasting repetitive firing of action potentials could occur after the end of stimulation (Ganetzky and Wu, 1982a; Ueda and Wu, 2006), which extended the decay time course of GCaMP signals in a manner of “rebound” or oscillation (mostly in type II and sometimes in type Is synapses, see examples in Figs. 5, 6).

In addition to manipulating the factors that regulate Ca²⁺ influx, we also examined known mechanisms affecting Ca²⁺ clearance for indications of their differential actions on type Ib, Is, and II synapses. As previously shown for type Ib synapses, a major Ca²⁺ extrusion mechanism in Drosophila NMJ is PMCA (Lnenicka et al., 2006; Klose et al., 2009). The H⁺/Ca²⁺ exchange pump PMCA relies on inward transport of proton counter-ions for Ca²⁺export, and can thus be inhibited by high external pH (Niggli and Sigel, 2008). High-pH inhibition of PMCA is effective and has been employed to manipulate Ca²⁺ transients in Drosophila NMJ synapses (Lnenicka et al., 2006; Klose et al., 2009; Caldwell et al., 2013), squid giant axon (Dipolo and Beaugé, 1982), and mammalian neurons and pancreatic cells (Benham et al., 1992; Chen et al., 2003).

Following an established protocol, we compared Ca²⁺ extrusion efficacy among the three types of synapses in low and high pH saline (pH 7.2 vs 8.8; cf. Lnenicka et al., 2006). The results showed that acutely increased extracellular pH not only slowed down the decay kinetics as expected (Fig. 13A,B), but could also accelerate the rise and increase the amplitude of GCaMP signals significantly under some conditions (Fig. 13C–E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 13.

Effects of PMCA inhibition at high pH on Ca²⁺ dynamics of type Ib, Is, and II synapses. A, ΔF/F traces of WT larvae in control pH (7.2, black) and high pH (8.8, blue) saline. Overlapping traces from the same individual boutons illustrate the consequences of suppressing plasma membrane Ca²⁺ extrusion via PMCA at a lower (subplateauing, upper row) and a higher (plateauing, lower row) stimulation frequencies. B–D, Averaged half-decay time (T_1/2Decay), half-rise time (T_1/2Rise), and peak GCaMP signals (max ΔF/F), respectively, at different stimulation frequencies for type Ib, Is, and II synapses (five to six NMJs from different larvae; N.A., not applicable due to small signals unsuitable for reliable kinetic determination). Statistically significant differences between high and low pH are indicated (t tests; *p < 0.05, **p < 0.01, ***p < 0.001). E, Relative increment in GCaMP signal amplitude determined as the ratio of max ΔF/F between pH 8.8 and 7.2 measurements. One-way ANOVA and Fisher’s LSD tests, a, p < 0.001, significantly different from group c; b, p < 0.05, significantly different from group c. The dashed line indicates the ratio of 1. Note that PMCA inhibition most effectively increased max ΔF/F around the threshold frequencies (D, E; i.e., 40, 20, and 10 Hz), coupled with lengthening of decay time above the plateauing frequencies (B; i.e., 80, 40, and 10 Hz) for Ib, Is, and II synapses, respectively.

In pH 7.2 saline, the half-decay time of GCaMP1 signals was ∼0.5 s for both type Ib and Is and ∼0.7–0.8 s for type II synapses at both 0.1 and 0.5 mM Ca²⁺ (Fig. 13B; Table 3; other GCaMP indicators reported similar decay times, compare Fig. 3; data not shown). Immediately on pH increase to 8.8, the half-decay time in both type Ib and Is synapses increased to 0.6–0.7 s (Fig. 13B, 40–80 Hz) while a more pronounced increase to ∼1.1 s was seen in type II synapses (Fig. 13B, 10–40 Hz).

Compared to retarded decay kinetics, high-pH treatment rendered milder effects, in terms of accelerated rise and enhanced peak amplitude, on GCaMP signals in all three types of synapses. Acceleration of the rise phase was detected somewhere below the “saturation frequency” where GCaMP signals approached the highest attainable amplitude (Fig. 13C, 80 and 40 Hz for type Ib and Is; compare Figs. 2C, 13D, Table 1). It is interesting to note that these were similar to the frequencies at which the lengthening in decay time became evident for type Ib and Is synapses (80 and 40 Hz; Fig. 13B). In contrast, significant enhancement of amplitude (max ΔF/F) was found around the “threshold frequencies”, where GCaMP signals became readily detectable (Fig. 13C, 40, 20, and 10 Hz for type Ib, Is, and II synapses, respectively), and the relative enhancement (fold change) was greatest in type Ib synapses (Fig. 13D,E, at 40 Hz).

To investigate the extent of involvement of PMCA in synaptic Ca²⁺ clearance, we further increased saline pH to 9.8 and found that pH 9.8 led to greater increases in the decay time of GCaMP signals in all three synapses (>1 s for type I synapses) and shifted GCaMP signals to lower frequency ranges (Ib: 20 Hz, Is: 10 Hz). However, under this condition, the GCaMP signal rapidly ran down within about ten minutes, precluding reliable quantitative determination of the lengthening in half-decay time.

Prolonged mitochondrial inhibition and energy-dependent Ca²⁺ clearance in type Ib, Is, and II synapses

Exporting Ca²⁺ out of cell or into intracellular organelles against its concentration gradient requires energy expenditure, e.g., ATP hydrolyzis. Mitochondria are critical for ATP synthesis to power active Ca²⁺ clearance mechanisms such as PMCA, a Ca²⁺-ATPase (Zenisek and Matthews, 2000; Shutov et al., 2013). A maintained proton gradient across mitochondrial inner membrane is required for these ATP-dependent Ca²⁺ clearance mechanisms, as well as direct mitochondrial sequestration of intracellular free Ca²⁺ (Tang and Zucker, 1997; David et al., 1998; Suzuki et al., 2002).

The proton ionophore DNP is known to reversibly uncouple mitochondrial proton gradient from oxidative phosphorylation (Higgins and Rogers, 1976; Nguyen et al., 1997; Buckler and Vaughan-Jones, 1998; Petrenko et al., 2010). We performed time-lapse measurements of GCaMP signals during the 1-hour DNP treatment. Subsequent to the basal measurements (control, or “0” min; Fig. 14), the preparation was incubated in 0.2 mM DNP saline (0.1 mM Ca²⁺ in HL3.1). Among the three synaptic types, type II synapses first showed drastically lengthened decay time course (20-min incubation; Fig. 14A–D, dark blue traces and bars) in correlation with its slowest basal and high pH decay times among the three (Fig. 13B; Table 3). On further incubation (up to 60 min), drastically prolonged decay times of GCaMP signals were also observed in type Ib and Is synapses, while type II synapse stopped responding to stimulation by this time (Fig. 14A–D, light blue traces and bars). However, further continuous DNP incubation eventually led to total loss of GCaMP responses in all synapses. It is important to note that the lengthened decay time course of individual synaptic terminals was partially reversible, if DNP washout was done before GCaMP response loss. Irreversible loss of GCaMP responses of individual synaptic terminals occurred first in some type II synapses (starting around 20 min into incubation) and then in a smaller portion of type Ib and Is synapses (∼60 min). Compared to the dramatic alterations in decay kinetics, the rise kinetics, i.e., half-rise time, were only mildly affected by DNP treatment (Fig. 14, compare D, E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 14.

Drastic effects of prolonged DNP treatment on Ca²⁺ dynamics in all three types of synapses. A–C, Overlaid representative GCaMP Ca²⁺ signal ΔF/F traces under 10-, 20-, and 40-Hz stimulation, after 0-, 20-, and 60-min exposure to DNP (black, blue, and light blue, respectively). Note that type II synapses (right column) completely lost responses after a 60-min exposure (light blue), and within a 20-min exposure, about half of type II synapses had already lost responses (data not shown). D–F, Summary bar charts for the half-decay time, half-rise time, and maximum ΔF/F of GCaMP Ca²⁺ signal under 10, 20, and 40 Hz of repetitive stimulation. N.A., not applicable because the corresponding GCaMP ΔF/F traces were too small or simply nonresponsive to stimulation, thus excluded from kinetic determination (e.g., type II after a 60-min DNP treatment, and type Ib and Is at low stimulation frequencies). Number of boutons (sampled from six different larvae) are indicated. KW tests with Bonferroni corrections were performed among each frequency group; * and +p < 0.05, **p < 0.01, *** and +++p < 0.001. For type II synapses after a 20-min DNP treatment, + and +++ denote statistical differences with the dwindling sample sizes due to loss of responses in some type II NMJs after DNP treatment.

DNP incubation greatly enhanced GCaMP signals of both type Ib and Is synapses so that significantly greater GCaMP signals were observed even at low (10 Hz) stimulation frequencies (Fig. 14A,F, left and middle columns). Such a shift of threshold frequency was most evident in type Ib synapses (from 40 to 10 Hz; Fig. 14F). Notably, DNP effects were further promoted with hyperactivity. It was found that hyperexcitable eag Sh accelerated the onset of characteristic DNP effect in type Ib and Is synapses (slower decay and enhanced amplitude) to within 25 min, compared to >40 min in WT. Conceivably, higher activity levels in eag Sh could drain the ATP reserve at a faster rate.

A possible effect of long-term proton ionophore (e.g., DNP) treatment is depolarization of the plasma membrane potential (Hodgkin and Keynes, 1955) and thus increased Ca²⁺ influx, as prolonged mitochondrial inhibition leads to diminished ATP supply for Na⁺/K⁺ ATPase, which maintains the resting membrane potential. However, this effect may not have reached a significant level within the time frame of our experiments, i.e., the repetitive firing as demonstrated by correlating focal recording with GCaMP imaging (compare Figs. 8, 9). Within 1 hr, DNP treatments induced enhanced GCaMP signals with prolonged decay time course, but this was seldom accompanied by supernumerary efEJPs or significant elevation of baseline GCaMP fluorescence (except for 1 out of >10 larvae, excluded from analysis). To obtain a different line of evidence for this striking mitochondrial inhibition effect, we used azide, which inhibits the complex IV in the electron transport chain (Fei et al., 2000). Long-term incubation in azide (1 mM, 60–90 min) resulted in similarly drastic slow decay kinetics of GCaMP6m signals (half-decay time >3 s in all three synapses) and shift in threshold frequency (Ib: from 40 to 20 Hz, Is: from 20 to 10 Hz).

All these lines of evidence indicate that intact mitochondrial metabolism is critical for presynaptic Ca²⁺ clearance, which differs in capacity in type Ib, Is, and II synapses. We therefore examined the mitochondrial density in these synapses by staining the NMJ with TMRM, a fluorescent indicator sequestered into mitochondria by the electrochemical proton gradient (Scaduto and Grotyohann, 1999). By overlaying the mitochondrial staining with GCaMP fluorescence, it was evident that type Ib synapses were most enriched with abundant mitochondria, with 94% of boutons (22 NMJs, four WT larvae, SD = 11%) housing one or more mitochondria (Fig. 15B). type Is synaptic boutons also contained a good amount of strongly-stained mitochondria (71 ± 39% of boutons, 21 NMJs, four larvae; Fig. 15B,C), although less than type Ib (p < 0.05, t tests with Bonferroni correction). However, type II synaptic boutons rarely contained visible mitochondrial staining (16 ± 21%, 18 NMJs, four larvae; Fig. 15C), which suggests either significantly less mitochondria availability or lower membrane potential to render TMRM staining of existing mitochondria, or both, than type Ib and Is (p < 0.001, t tests with Bonferroni correction). The lower density of active mitochondria in type II synapses correlates well with its slower Ca²⁺ clearance rate (Fig. 13B; Table 3) and higher vulnerability to DNP inhibition (Fig. 14).

• Download figure
• Open in new tab
• Download powerpoint

Figure 15.

Differential mitochondrial density in type Ib, Is, and II synapses. Columns 1 and 2 are epifluorescent NMJ images obtained with GCaMP1 baseline fluorescence (cyan) and TMRM mitochondrial staining (red), respectively. Column 3 is the merge of columns 1 and 2. The images of the cyan and red channels were separately thresholded before merging to reveal the bouton contour (cyan) and the most strongly stained spots (red). Content in non-bouton, non-axon area was erased in column 3 for clearer demonstration of overlapping. Row A is representative images of a muscle 12 (bottom, M12 in A3) and 13 (top, M13) NMJs from a WT larva (+/Y; c164-GCaMP1.3). The border between M12 and M13 is faintly visible in A2. Rows B, C are enlarged views of the broken and continuous boxed regions in A, respectively. In Column 1, type Ib, Is, and II boutons are designated with filled arrowheads, open arrowheads, and open circles, respectively. In Column 2, asterisks indicate some areas of stronger TMRM mitochondrial staining. Note that the weak mitochondrial staining in type II synapses was well below the intensity level of type Is synapses. Statistics are presented in Results.