The Maintenance of Synaptic Homeostasis at the Drosophila Neuromuscular Junction Is Reversible and Sensitive to High Temperature

Homeostatic potentiation using a dominant-negative glutamate receptor subunit

A prior study described a transgene encoding a dominant-negative GluRIIA subunit, UAS-GluRIIA^M614R (herein termed GluRIIA^M/R or dominant-negative; DiAntonio et al. 1999). The GluRIIA M614R amino acid substitution resides in the ion conduction pore of the GluRIIA subunit, and it cripples channel function (DiAntonio et al. 1999). Transgenic expression of UAS-GluRIIA^M/R in muscles renders a strong homeostatic challenge (markedly diminished quantal size) and an equally strong compensatory response (increase in quantal content). The prior report demonstrated that evoked amplitudes remain normal as a result of this compensation, presently termed presynaptic homeostatic potentiation (PHP; DiAntonio et al. 1999).

We acquired the dominant-negative transgenic line to test homeostatic reversibility. First, we replicated the published experiments, raising fruit fly larvae at temperatures compatible with the TARGET system that we planned to use to test reversibility (McGuire et al. 2003). We tested 25°C (a common culturing temperature) and 29°C (a temperature at which the GAL80^TS protein ceases to inhibit GAL4). We drove UAS-GluRIIA^M/R expression with MHC-Gal4, which turns on in first-instar larval muscles (Schuster et al. 1996a, b). We recorded from NMJs of WT, GAL4 driver control, and dominant-negative wandering third instar larvae.

For animals reared at 25°C, WT NMJ electrophysiology was robust and consistent with values canonically reported for WT at this synapse (Fig. 1A, C, D; Table 1 for raw electrophysiological data). For heterozygous MHC-Gal4/+ driver control NMJs (“MHC-Gal4” on figures), we noted a small decrease in quantal size compared to WT (Figs. 1A, C, D). This quantal size decrease was consistent across the entire distribution of the spontaneous miniature excitatory postsynaptic potential (mEPSP) sample (Figs. 2A, B). Driver controls had a fully offsetting increase in QC, keeping evoked potentials robust (Figs. 1A, C; Table 1).

Muscle-specific expression of the UAS-GluRIIA^M/R dominant-negative transgene at 25°C caused a stark diminishment of quantal size and frequency compared to controls, evident from a small average mEPSP amplitude and decreased mEPSP amplitudes across the entire distribution (Figs. 1A, C, and 2A, B; Table 1). Despite decreased quantal size, there was no diminishment in the average evoked excitatory postsynaptic potential (EPSP) amplitude for dominant-negative NMJs (Figs. 1A, C, D; Table 1)—again, because of a fully offsetting homeostatic increase in QC (Figs. 1A, C; Table 1). In sum, our recordings at 25°C agreed with the prior finding for the dominant-negative transgene (DiAntonio et al. 1999): full PHP.

For animals raised at 29°C, we garnered similar results as at 25°C, noting a few differences. At 29°C, WT NMJs had slightly smaller mEPSP quantal amplitude and frequency than 25°C WT NMJs (Fig. 1A–D; Table 1). This finding was consistent with a prior report measuring NMJ physiology at elevated temperatures (Ueda and Wu, 2015). WT EPSP amplitudes and QC at 29°C were nevertheless robust, similar to the values garnered at 25°C (Fig. 1A–D; Table 1). MHC-Gal4/+ driver controls at 29°C showed slightly diminished average mEPSP amplitudes, a diminishment that persisted across the entire distribution of events (Figs. 1B, C, and 2C, D). MHC-Gal4/+ driver controls at 29°C maintained normal evoked potentials because of homeostatically increased QC (Fig. 1B–D).

By contrast, UAS-GluRIIA^M/R-expressing NMJs from 29°C showed a profound decrease in mEPSP size, a partially compensatory increase in QC compared to WT, and markedly reduced quantal frequency compared to WT (Figs. 1B, C, and 2C, D). Although QC was significantly increased in dominant-negative NMJs compared to WT, it did not increase enough to bring average dominant-negative EPSP amplitudes at 29°C fully back to WT or MHC-Gal4/+ control levels (Fig. 1B–D).

The diminished mEPSP frequency for dominant-negative-expressing NMJs (Fig. 1C) is an interesting phenotype; it mirrors numerous cases at the Drosophila NMJ in which there is greatly diminished glutamate receptor subunit expression or function (Featherstone et al. 2005; Daniels et al. 2006; Kim et al. 2012, 2015; Brusich et al. 2015; Ramos et al. 2015). Given the very low observed mEPSP frequencies for dominant-negative NMJs, one potential concern was a severe underestimation of QC in these recordings due to unobserved mEPSP events that could have been lost in the noise of the recordings. In theory, such lost events could confound analyses of homeostatic compensation. We calculated the average noise for mEPSP recordings from every genotype, as well as an mEPSP size/mEPSP noise ratio (Table 1). Noise varied slightly from genotype to genotype, but it was vanishingly small for each one (∼0.02 mV/mEPSP for each; Table 1). This small noise yielded very large mEPSP/noise ratios for all genotypes (Table 1). Those ratios did vary because of different average quantal sizes between the genotypes; as a result, the ratio was a little smaller for dominant-negative NMJs than for WT controls (Table 1). Importantly, however, dominant-negative GluRIIA^M/R-expressing NMJs showed similar values by these noise measures at 25°C versus 29°C (Table 1). This indicated that recording noise was not a factor in determining full versus partial homeostatic compensation between the 25°C and 29°C conditions.

Early third instar larvae express homeostatic plasticity

Our initial experiments showed that it is possible to observe increases in QC for UAS-GluRIIA^M/R-expressing animals compared to WT, when raising flies at either 25°C or 29°C. We sought to test homeostatic reversibility. The TARGET system augments GAL4/UAS expression, adding a ubiquitously expressed Tubulin_P-Gal80^TS GAL4 inhibitor transgene (McGuire et al. 2003). We added this GAL80^TS expression feature to the MHC-Gal4 ≫ UAS-GluRIIA^M/R genetic background. The temperature sensitivity of the GAL80^TS protein permits tight control over when GAL4-responsive transgenes are expressed (McGuire et al. 2003). We predicted that the dominant-negative transgene should be repressed by GAL80^TS at low temperatures; conversely, UAS-GluRIIA^M/R should be actively expressed when GAL80^TS is inactive (∼29°C or higher; McGuire et al. 2003). To study the reversibility of PHP, animals could be reared at high temperature and then swapped to a lower temperature at an appropriate developmental time point.

An ideal temperature swap point would be late enough to detect PHP but early enough to allow recovery time after cessation of the dominant-negative UAS-GluRIIA^M/R expression. We crossed UAS-GluRIIA^M/R; Tub_P-Gal80^TS × MHC-Gal4 stocks. Mated animals were transferred to 25°C or 29°C for egg laying and subsequent larval development. We selected early third instar progeny for electrophysiological recording. At temperature ranges of 25°C–29°C, early third instar larvae develop roughly 48–60 h after egg-laying. We staged these small animals unambiguously by examining their posterior spiracles for an orange-colored tip.

For animals raised entirely at 25°C, early third instar NMJ mEPSP size was large—significantly larger than one would observe for third instar larvae (Fig. 3A, B; Table 1; compare with Fig. 1). This was expected because small muscles have a significantly greater input resistance and enhanced quantal size (Lnenicka and Mellon, 1983a, b; Davis and Bezprozvanny, 2001; Table 1). Nevertheless, early third instar larval NMJs raised at 25°C showed normal evoked amplitudes, consistent with a stable level of evoked muscle excitation throughout development (Fig. 3A, B; Table 1; compare with Fig. 1).

We expected that genotypically equivalent early third instars raised entirely at 29°C would express the dominant-negative transgene. NMJs from these 29°C animals showed sharply reduced average mEPSP amplitude and frequency compared to their stage- and size-matched counterparts raised at 25°C (Fig. 3A, B; Table 1), and this mEPSP amplitude decrease persisted throughout the distribution of events (Fig. 3C, D). There was a robust increase in quantal content at 29°C, resulting in EPSP amplitudes that were nearly normal, but not quite at the same level as at 25°C. Therefore, as with the earlier experiments, PHP for dominant-negative early third instar NMJs raised exclusively at 29°C was present, but not full (Fig. 3A, B; Table 1). Again, as a quality-check control, absolute mEPSP noise levels were small for both conditions (Table 1), and the mEPSP size/mEPSP noise ratios were similar to values obtained for the prior experiment (Table 1).

Partial homeostatic potentiation is reversible

At 29°C, PHP was not full, but QC increases versus controls were robust, still making it possible to test homeostatic reversibility. We generated additional MHC-Gal4 ≫ UAS-GluRIIA^M/R larvae with the Tub_P-GAL80^TS transgene. For reversibility experiments, we chose 21°C as a permissive GAL80^TS shift temperature because 21°C permitted multiple electrophysiological time point measurements over a long recovery window (Fig. 4A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Presynaptic homeostatic potentiation is reversible. A, Diagram of a temperature swap paradigm. Mated animals laid eggs for a small window of 6–8 h. One set of larvae was reared entirely at 21° from egg laying to electrophysiological recording. A second set was raised entirely at 29°C. To test for reversibility of homeostatic potentiation, additional sets were reared initially at 29°C and then swapped to 21°C. Animals were allowed to recover for 1, 2, or 3 d before recording. B, Expression of the dominant-negative transgene throughout life (29°C) caused a dramatic decrease in mEPSP frequency (***p < 0.001, one-way ANOVA with Tukey’s post hoc). Frequency remained low after 1 d of recovery at 21°C, compared to baseline 21°C controls, but by 2 d of recovery, the frequency was significantly increased compared to 29°C (***p < 0.001), but still decreased compared to 21°C controls (**p < 0.01). By 3 d of recovery, mEPSP frequency was no different from animals raised entirely at 21°C. C, Normalized electrophysiological data for mEPSP amplitude, EPSP amplitude, and QC for the NMJs of animals raised as described in A. Longer recovery periods yielded electrophysiology that closely approximated the control 21°C rearing condition (all statistics: *p < 0.05, **p < 0.01, ***p < 0.001 compared to 21°C control by one-way ANOVA with Tukey’s post hoc). D, Representative electrophysiological traces for 21°C and 29°C. E, Traces for the three recovery conditions. Scale bars for EPSPs (and mEPSPs) are 5 mV (1 mV) and 50 ms (1000 ms) for both D and E.

Control (no PHP) animals raised entirely at 21°C took ∼120 h after egg laying to reach the wandering third instar stage, whereas control (PHP) animals raised entirely at 29°C took ∼96 h to reach the same stage (Fig. 4A). A 1-d recovery condition was used, exposing animals to the UAS-GluRIIA^M/R challenge for ∼72 h (29°C) and allowing them to recover at 21°C for 1 d before recording. We also tested 2- and 3-d recovery conditions, rearing larvae at 29°C for ∼49–50 h and then swapping them to 21°C until they reached wandering third instar stage. Because of the length of the egg lays and small variations in developmental time, some animals reached the wandering third instar recording stage after ∼2 d at 21°C, and others took closer to 3 d (Fig. 4A).

Control MHC-Gal4 ≫ UAS-GluRIIA^M/R larvae with ubiquitously expressed Tub_p-Gal80^TS and raised at 21°C showed neurotransmission indistinguishable from WT control animals reared at 25°C (compare Figs. 1 and 4; Table 1). By contrast, genetically identical control animals raised at 29°C throughout life showed electrophysiological phenotypes similar to dominant-negative MHC-Gal4 ≫ UAS-GluRIIA^M/R animals raised at 29°C (compare Figs. 1 and 4; Table 1). Compared to counterparts raised at 21°C, animals raised at 29°C showed reduced mEPSP frequency, reduced mEPSP size, slightly below-normal EPSP amplitudes, and increased QC, indicating robust PHP (Fig. 4B, C; Table 1).

Recovery conditions showed physiologic signatures that corresponded with how much time was spent at 21°C. Compared to the 29°C condition, the 1-d 21°C recovery condition showed no significant changes in physiologic properties and no reversal of PHP (Fig. 4B–E). By contrast, the 2-d recovery condition showed intermediate physiology. Compared to constant exposure to 21°C, the 2-d recovery condition still had diminished mEPSP amplitude and frequency, but not nearly as diminished as constant exposure to 29°C (Fig. 4B–E). Interestingly, the 2-d recovery EPSP amplitudes revealed fully restored levels of excitation, due to an offsetting QC level (Fig. 4C–E). The 3-d recovery condition showed electrophysiology that was not significantly different from the constant 21°C condition (Fig. 4B–E), indicating a full reversal of PHP.

We further analyzed the aggregate data from the reversibility experiment. We wished to analyze reversal parameters related to recovery time. We plotted mEPSP (Fig. 5A), EPSP (Fig. 5B) and QC (Fig. 5C) values versus the number of hours animals spent at the recovery temperature, with a specific recovery time value for each NMJ, time-locked to the egg laying period and the recording time after recovery/development at 21°C. Each plot showed hallmarks of PHP reversal: mEPSP and EPSP amplitudes positively correlated with time at 21°C (Fig. 5A, B), and QC values inversely correlated with time at 21°C (Fig. 5C). Consistent with the observation that PHP was present (but not full) at 29°C, individual data points from NMJs of animals raised entirely at 29°C showed a wide variability of QC values (Fig. 5C). Prior studies of homeostatic plasticity at the Drosophila NMJ have shown that by analyzing hundreds of individual recording values, QC inversely scales with quantal size across genotypes, and as a result, evoked excitation levels remain stable (Frank et al. 2006; Davis and Müller, 2015; Gaviño et al. 2015). For our temperature swap experiments—this time conducting a comparison within a single genotype—this also proved to be the case (Fig. 5D). Finally, we checked whether there was a leading component to the recovery, testing whether mEPSP and EPSP values recovered at similar or different rates. We coplotted each parameter’s recovery curve relative to the 21°C control values and fitted the nonlinear curves with quadratic functions. Both parameters appear to recover at similar rates, with the EPSP values reaching the 21°C baseline level first (Fig. 5E).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Strong correlations between recovery time and classic homeostatic parameters. Electrophysiological parameter data plotted from the reversibility experiments detailed in Fig. 4. Animals reared for “0 hours” at 21°C are the 29°C condition; animals in the 21°C condition were reared at that temperature for ∼120 h. Each data point is time-locked to the times of the relevant egg-laying period and the recording. A, mEPSP amplitude plotted as a function of hours raised at 21°C. The data show a strong positive correlation between the two parameters (R ² = 0.71; nonlinear fit as a quadratic equation). B, EPSP amplitude plotted as a function of hours raised at 21°C. The data show a positive correlation between the two values (R ² = 0.28; nonlinear fit as a quadratic equation). C, QC plotted as a function of hours raised at 21°C. The data show a negative correlation between the two values (R ² = 0.34; nonlinear fit as a quadratic equation). D, Plots of QC versus mEPSP amplitude for each recording. Consistent with prior reports examining PHP at the NMJ, there was a negative correlation between the two values (R ² = 0.47; nonlinear fit as a quadratic equation). E, A comparison of quadratic fits for the mEPSP recovery curve (y = –1.835x ² + 20.38x + 45.41) versus EPSP recovery curve (y = –2.648x ² + 18.93x + 73.18). Both parameters are plotted as a percentage of the 21°C control values, as a function of days (= hours/24) reared at 21°C.

Full homeostatic potentiation is also reversible

Homeostatic potentiation was robust, yet partial, at the 29°C condition. We hoped to test a condition that was both compatible with full PHP and the reversibility assay. We attempted a new temperature swap, changing parameters. We reasoned that by making the environmental challenge to the NMJ less harsh—by using a lower temperature, putting forth a less severe level glutamate receptor loss, or both—we would be able to generate an ideal full PHP reversibility test condition. For one change, we lowered the restrictive GAL80^TS-impairing temperature from 29°C to 28.5°C. Other studies using the TARGET system in Drosophila have reported that 28.5°C is somewhat effective at impairing GAL80^TS function (Staley and Irvine, 2010; Corrigan et al. 2014; Redhai et al. 2016). Second, we replaced the MHC-Gal4 muscle driver with the BG57-Gal4 muscle driver (Budnik et al. 1996). Finally, we tried 25°C as the permissive condition.

We generated new sets of larvae for this swap experiment: BG57-Gal4 ≫ UAS-GluRIIA^M/R larvae with the TubP-Gal80^TS transgene. As expected, animals raised at 25°C throughout life (GAL80^TS on) developed NMJs with electrophysiological properties similar to other control conditions already reported (Fig. 6A, C; Table 1). Also as expected, animals raised at 28.5°C throughout life (GAL80^TS impaired) had significantly diminished NMJ mEPSP amplitudes (Fig. 6A; Table 1). Of note, the 28.5°C NMJs had markedly diminished mEPSP frequency (Fig. 6B), which is an indication of successful expression of the dominant-negative GluRIIA^M/R transgene (Fig. 1). NMJs from those 28.5°C animals also showed completely normal EPSP amplitudes because of a full, offsetting homeostatic increase in QC (Fig. 6A, C; Table 1). Finally, animals raised at 28.5°C until early third instar and then swapped to 25°C for the final 2 d of larval development showed NMJ electrophysiology indistinguishable from that of animals raised at 25°C throughout life, indicating a complete reversal of PHP (Fig. 6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Full homeostatic compensation is reversible. A, Dominant-negative animals with the Gal80^TS transgene raised at 28.5°C showed a significant decrease in mEPSP amplitude (*p < 0.05, compared to 25°C condition; one-way ANOVA with Tukey’s post hoc), accompanied by a fully offsetting increase in QC (**p < 0.01). Reversibility of homeostatic potentiation was demonstrated by initially rearing w; GluRIIA^M/R/+; BG57-Gal4/Gal80^TS animals at 28.5°C for 2 d and swapping them to 25°C for 2 d. mEPSP amplitude, EPSP amplitude, and QC returned to control (25°C throughout life) levels. EPSP amplitudes were virtually identical across all of the conditions. B, Dominant-negative animals raised at 28.5°C throughout life showed a significant decrease in mEPSP frequency (*p < 0.05, one-way ANOVA with Tukey’s post hoc). In contrast, the mEPSP frequency in animals after 2 d of recovery was not significantly different from that of the animals reared at 25°C. C, Representative electrophysiological traces. Scale bars for EPSPs (and mEPSPs) are 5 mV (1 mV) and 50 ms (1000 ms).

Homeostatic potentiation induced by glutamate receptor function loss is impaired at high temperatures

UAS-GluRIIA^M/R transgene-induced homeostatic potentiation was robust, yet partial, at the 29°C condition. We wondered whether PHP might specifically be impaired at the NMJ when flies are raised at high temperatures. An alternative idea is that a functional limit of the homeostatic signaling system could be due to a dual combination of factors challenging synapse function, such as high temperature combined with glutamate receptor function loss.

To test this idea further, we drove the dominant-negative transgene in the muscle, setting up crosses to generate both MHC-Gal4 ≫ UAS-GluRIIA^M/R and BG57-Gal4 ≫ UAS-GluRIIA^M/R animals, as well as driver-specific controls. For the driver controls, quantal size was somewhat diminished at 30°C (Fig. 7A–D; Table 1). This was consistent with the idea that quantal size is generally diminished at very high temperatures (Ueda and Wu, 2015), although taken together with our prior results (Fig. 1), it also seemed likely that simply driving GAL4 protein in the muscle can cause this phenotype. Despite diminished quantal size, evoked EPSP amplitudes were still robust for 30°C driver controls because of homeostatic increases in QC (Fig. 7A–D; Table 1). For the dominant-negative NMJs, there was a significant reduction in mEPSP size beyond what was measured for the driver controls. Moreover, evoked amplitudes were weak, diminished significantly versus driver controls because of no significant increase in QC versus WT controls (Fig. 7A–D; Table 1). These data indicated that signaling processes that maintain PHP in response to the dominant-negative transgene may break down at high temperatures.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Presynaptic homeostatic potentiation can be impaired at high temperatures. A, WT (w¹¹¹⁸), driver controls (w;; MHC-Gal4/+), and dominant-negative NMJs (w; UAS-GluRIIA^M/R/+; MHC-Gal4/+) were compared at 30°C. For driver controls, the mEPSP amplitude was significantly decreased compared to WT (***p < 0.001, one-way ANOVA with Tukey’s post hoc), but the EPSP amplitude was only slightly smaller than normal (*p < 0.05 versus WT) because of a homeostatic increase in QC (***p < 0.001 versus WT). For dominant-negative animals, the mEPSP amplitude was markedly decreased (***p < 0.001 versus WT; ##p < 0.01 versus driver controls, one-way ANOVA with Tukey’s post hoc). A dramatic decrease in EPSP amplitude for the dominant-negative NMJs also occurred (***p < 0.001 versus WT) because there was no significant change in QC (p = 0.58 versus WT), indicating a block of PHP. B, Representative electrophysiological traces. Scale bars for EPSPs (and mEPSPs) are 5 mV (1 mV) and 50 ms (1000 ms). C, WT, driver controls (w;; BG57-Gal4/+), and dominant-negative animals (w; UAS-GluRIIA^M/R/+; BG57-Gal4/+) were compared at 30°C. For driver controls, the mEPSP amplitude was significantly decreased compared to WT (***p < 0.001, one-way ANOVA with Tukey’s post hoc), but the EPSP was normal because of a homeostatic increase in QC (***p < 0.001 versus WT). In contrast, for dominant-negative animals, the mEPSP amplitude was markedly decreased (***p < 0.001 versus WT; #p < 0.01 versus driver controls, one-way ANOVA with Tukey’s post hoc). A dramatic decrease in EPSP amplitude for the dominant-negative NMJs also occurred (***p < 0.001 versus WT) because there was no significant change in QC (p = 0.71), indicating a block of PHP. D, Representative electrophysiological traces. Scale bars as in B. E, GluRIIA^SP16 animals and WT controls were raised at 30°C. mEPSPs were significantly decreased for GluRIIA mutants (***p < 0.001, Student’s t test). EPSP values were also decreased (***p < 0.001) but not to the same magnitude as mEPSP amplitudes because the QC was significantly increased (**p < 0.01). F, Representative electrophysiological traces. Scale bars as in B. G, GluRIIA^SP16 animals and WT controls were raised at 31°C. mEPSP and EPSP amplitude were both significantly decreased in GluRIIA^SP16 animals (***p < 0.001, Student’s t test) and QC was increased (**p < 0.01). H, Representative electrophysiological traces. Scale bars as in B. I, The same genotypes were raised at 32°C. GluRIIA^SP16 animals showed a decrease in mEPSP and EPSP amplitude (**p < 0.01, Student’s t test), and they failed to show increase in QC (p = 0.17). J, Representative electrophysiological traces. Scale bars as in B.

We were uncertain whether this high-temperature failure in PHP was due to the expression levels of the dominant-negative glutamate receptor subunit at 30°C, the high temperature, or a combination of both. High temperature alone seemed unlikely, because the 30°C driver controls also had diminished quantal size accompanied by robust PHP. Glutamate receptor function loss alone also seemed unlikely, because quantal size for the dominant-negative NMJs was not starkly diminished at 30°C compared to earlier experiments run at 25°C and 29°C (Fig. 1; Table 1).

To extend our inquiry, we examined an additional homeostatic challenge to NMJ function. We raised WT and GluRIIA^SP16 null deletion flies (Petersen et al. 1997) at 30°C, 31°C, and 32°C. WT NMJs showed relatively normal physiology at 30° and 31°C (Fig. 7E–H; Table 1). Interestingly, NMJs from GluRIIA^SP16 deletion animals raised at 30° and 31°C showed significant PHP, with reduced mEPSP amplitudes and robust increases in QC compared to WT (Fig. 7A, B; Table 1). However, we noted that GluRIIA^SP16 EPSP amplitudes were reduced compared to WT, meaning that the PHP was only partial (Table 1). These GluRIIA^SP16 QC data (Fig. 7E, G) and the 30°C muscle-driver control QC data (Fig. 7A, C) argue against the notion that high temperature alone completely impairs PHP. Temperature could still be a contributing factor.

We next raised WT and GluRIIA^SP16 null animals at 32°C. This extreme temperature was suboptimal for animal health. WT animals, while able to reach the third instar stage, had poor muscle health, as measured by low input resistance and low mEPSP amplitude compared to conditions of normal rearing temperatures (Table 1). Nevertheless, 32°C WT NMJs had evoked events that were nearly normal in size (Fig. 7I, J; Table 1). Homozygous GluRIIA^SP16 larvae that survived to the wandering third instar stage at 32°C were rare. For the 32°C GluRIIA^SP16 homozygotes that did survive, the NMJs showed a further decrease in mEPSP amplitude and no PHP expression (Fig. 7I, J; Table 1). Taken together, the electrophysiological data are consistent with the idea that a dual challenge—glutamate receptor function loss plus extreme high temperature—abolishes PHP.

Dual-challenged NMJs show diminished growth at high temperatures

We asked whether synaptic growth defects might correlate with PHP deficits when raising glutamate receptor–compromised animals at high temperatures. We took larvae raised in a subset of the high temperature conditions described (Fig. 7) and immunostained NMJs with anti-Synapsin (Syn, presynaptic vesicles) and anti-Discs Large (Dlg, postsynaptic density) antibodies to assess NMJ development by immunofluorescence. We quantified synaptic bouton elaboration. For dominant-negative glutamate receptor expression under MHC-Gal4 control, there was a sharp decrease in NMJ growth at 30°C compared to the 29°C condition (Fig. 8A–C). By contrast, BG57-Gal4–driven dominant-negative expression elicited sharply diminished growth at both 29°C and 30°C (Fig. 8A, B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

NMJ undergrowth correlates with glutamate receptor function loss plus high temperature. A, w; UAS-GluRIIA^M/R/+; MHC-Gal4/+ and w; UAS-GluRIIA^M/R/+; BG57-Gal4/+ animals were raised at 29°C or 30°C and costained with anti-DLG (red) and anti-Synapsin antibodies (green) to visualize synaptic boutons. Boutons were counted based on postsynaptic DLG staining and double checked for presynaptic Synapsin. At abdominal segment A2 muscle 6/7, w; UAS-GluRIIA^M/R/+; MHC-Gal4/+ animals raised at 30°C showed a significant decrease in bouton number compared to genetically identical counterparts raised at 29°C (***p < 0.001, Student’s t test). w; UAS-GluRIIA^M/R/+; BG57-Gal4/+ animals raised at 30°C showed only a slight decrease in bouton number compared to counterparts at 29°C (*p < 0.05), although in this case, the 29°C NMJs already showed stunted growth. B, Boutons were also quantified at synapse A3 6/7. w; UAS-GluRIIA^M/R/+; MHC-Gal4/+ animals raised at 30°C showed a significant decrease in bouton number compared to counterparts raised at 29°C (***p < 0.001). There was no significant difference between 29°C and 30°C for w; UAS-GluRIIA^M/R/+; BG57-Gal4/+ animals; again, both conditions for this genotype showed severely stunted NMJ elaboration. C, Representative NMJ images of synapse A2 6/7 for w; GluRIIA^M/R/+; MHC-Gal4/+ animals raised at 29°C and 30°C. Scale bar, 10 μm. D, WT and GluRIIA^SP16 animals were raised at 21°C, 29°C, or 30°C and stained with anti-DLG (red) and anti-Synapsin (green). At 29°C, GluRIIA^SP16 animals showed a significant decrease in bouton number compared to 29°C WT controls (***p < 0.001, one-way ANOVA with Tukey’s post hoc). Additionally, WT NMJs raised at 30°C showed decreased growth compared to WT counterparts raised at 21°C (###p < 0.001). E, DLG boutons were quantified at abdominal segment 3 for the same genotypes. GluRIIA^SP16 NMJs at 29°C showed a decrease in bouton number compared to WT at the same temperature (*p < 0.05). Additionally, WT NMJs raised at 29°C (##p < 0.01) and 30°C (###p < 0.001) showed decreased growth compared to the 21°C condition. F, Representative images of synapse A2 6/7 WT and GluRIIA^SP16 NMJs at 30°C, showing few boutons and a disorganized anti-Synapsin pattern. Scale bar, 10 μm.

For WT and GluRIIA^SP16, we collected NMJ growth data over a range of temperatures: 21°, 29°, and 30°C. At 21°C, growth was similar between the two genotypes (Fig. 8D, E). At 29°C, an impairment of NMJ growth in GluRIIA^SP16 nulls was apparent, whereas WT NMJ growth was still relatively normal (Fig. 8D, E). At 30°C, both WT and GluRIIA^SP16 nulls showed diminished overall growth (Fig. 8D–F), with GluRIIA^SP16 growth numerically (but not statistically) diminished. At 30°C for these genotypes, the anti-Syn staining had a tendency to appear disorganized, spreading out over a larger area in the presynaptic bouton (Fig. 8F). Collectively, these data are consistent with the idea that at high temperatures under conditions of glutamate receptor loss, PHP failure correlates with NMJ growth and development defects. It is uncertain whether these growth parameters are causally related to the limit in homeostatic signaling.