Synaptic homeostasis at the Drosophila neuromuscular junction is a reversible signaling process that is sensitive to high temperature

Homeostasis is a vital mode of biological self-regulation. The hallmarks of homeostasis for any biological system are a baseline set point of physiological activity, detection of unacceptable deviations from the set point, and effective corrective measures to counteract deviations. Homeostatic synaptic plasticity (HSP) is a form of neuroplasticity in which neurons and circuits resist environmental perturbations in order to maintain appropriate levels of activity. One assumption is that if an environmental perturbation triggers homeostatic corrective changes in neuronal properties, those corrective measures should be reversed upon removal of the perturbation. We test the reversibility and limits of HSP at a well-studied model synapse, the Drosophila melanogaster neuromuscular junction (NMJ). At the Drosophila NMJ, impairment of glutamate receptors causes a decrease in quantal size, which is offset by a corrective, homeostatic increase in the number of vesicles released per evoked presynaptic stimulus, or quantal content. This process has been termed presynaptic homeostatic potentiation (PHP). Taking advantage of a GAL4/GAL80TS/UAS expression system, we triggered PHP by expressing a dominant-negative glutamate receptor subunit at the NMJ. We then reversed PHP by halting expression of the dominant-negative receptor. Our data show that PHP is fully reversible over a time course of 48-72 hours after the dominant-negative glutamate receptor stops being genetically expressed. Additionally, we found that the PHP response triggered by the dominant-negative subunit was ablated at high temperatures. Our data show that the long-term maintenance of PHP at the Drosophila NMJ is a reversible regulatory process that is sensitive to temperature. SIGNIFICANCE STATEMENT Biological homeostatic systems must upregulate or downregulate cellular parameters in order to maintain appropriate set points of physiological activity. Homeostasis is a well-documented mode of regulation in metazoan nervous systems. True homeostatic control should be a reversible process – but due to technical difficulties of presenting and removing functional challenges to living synapses, the reversibility of homeostatic forms of synapse regulation has not been rigorously examined in vivo over extended periods of developmental time. Here we formally demonstrate that homeostatic regulation of Drosophila melanogaster neuromuscular synapse function is reversible and temperature-labile. This is significant because developing methods to study how homeostatic regulatory systems are turned on and off could lead to fundamental new insights about control of synaptic output.

10 inant-negative transgene should be repressed by GAL80 TS at low, permissive temperatures; and 2 3 1 conversely, UAS-GluRIIA M/R should be actively expressed when GAL80 TS is inactive (~29ºC or 2 3 2 higher) (McGuire et al., 2003). In order to study the reversibility of PHP, animals could be reared 2 3 3 at high temperature and then swapped to a lower temperature at an appropriate developmental 2 3 4 time point.

3 5
We needed to identify a suitable developmental stage for temperature swaps. An ideal 2 3 6 swap point would be late enough to detect PHP, but early enough to allow recovery time after Tub P -Gal80 TS x MHC-Gal4 stocks. Mated animals were transferred to 25ºC or 29ºC for egg lay- ing and subsequent larval development. We selected early third instar progeny for electrophysi-2 4 0 ological recording. At temperature ranges of 25-29ºC, early third instar larvae emerge roughly 2 4 1 48-60 hours after an egg-laying period. We staged small animals unambiguously by examining 2 4 2 their posterior spiracles for an orange-colored tip. For animals raised entirely at 25ºC, early third instar NMJ mEPSP size was large -sig-2 4 4 nificantly larger than one would observe for third instar larvae (Figs 2A We predicted genotypically equivalent early third instars raised entirely at 29ºC would 2 5 0 express the dominant-negative transgene. As expected, NMJs from these animals showed 2 5 1 sharply reduced mEPSP amplitude and frequency compared to their stage-and size-matched  Table 1). There was a robust increase in quantal 2 5 3 content at 29ºC, resulting in EPSP amplitudes that were nearly normal, but not quite at the 2 5 4 same level as at 25ºC. As with the earlier experiments, presynaptic homeostatic potentiation 2 5 5 11 (PHP) for early third instar NMJs raised exclusively at 29ºC was present, but not perfect (Figs. At 29ºC, PHP was not perfect, but QC increases versus controls were robust, making it 2 6 0 possible to test reversibility. We generated additional MHC-Gal4 >> UAS-GluRIIA M/R larvae with 2 6 1 the Tub P -GAL80 TS transgene. This time we chose 21ºC as a permissive GAL80 TS shift tempera-  Control (no PHP) animals raised entirely at 21°C took ~120 hours after the egg laying 2 6 5 period to reach the wandering third instar stage, while control (PHP) animals raised entirely at 2 6 6 29°C took ~96 hours to reach the same stage (Fig. 3A). A 1-day recovery condition was utilized 2 6 7 -exposing animals to the UAS-GluRIIA M/R challenge for ~72 hours (29ºC) and allowing them to 2 6 8 recover at 21°C for 1 day prior to recording. We also tested 2-and 3-day recovery conditions,  at 29ºC showed reduced mEPSP frequency, reduced mEPSP size, slightly below-normal EPSP 2 8 0 amplitudes, and increased QC, indicating robust PHP (Fig. 3 ingly, the 2-day recovery EPSP amplitudes revealed restored levels of excitation, due to an in-  We further analyzed the aggregate data from the reversibility experiment. We wished to 2 9 2 test for hallmarks of PHP and reversal related to recovery time. Prior studies of homeostatic 2 9 3 plasticity at the Drosophila NMJ have shown that by plotting hundreds individual recording val-2 9 4 ues, quantal content inversely scales with quantal size across genotypes -and as a result, gle genotype -this also proved to be the case (Fig. 4A).

9 8
Next, we plotted mEPSP and QC values versus the number of animals spent at the re-2 9 9 covery temperature, with a specific recovery time value for each NMJ, time locked to the egg  Consistent with the observation that PHP was present -but not perfect -at 29ºC (Figs. 1-3),  Homeostatic potentiation was robust, yet imperfect, at the 29ºC condition. We hoped to 3 0 9 test a condition that was both compatible with perfect PHP and the reversibility assay. We at-3 1 0 tempted a new temperature swap, changing a few parameters. For one change, we lowered the 3 1 1 restrictive Gal80 TS temperature from 29ºC to 28.5ºC. Other studies using the TARGET system 3 1 2 in Drosophila melanogaster have reported that 28.5ºC is somewhat effective at impairing there was a formal possibility that imperfect compensation at 29ºC reflected a GAL4 driver-3 1 5 specific phenomenon, rather than a temperature-specific phenomenon. Therefore, we replaced 3 1 6 the MHC-Gal4 muscle driver with the BG57-Gal4 muscle driver (Budnik et al., 1996). Finally, we 3 1 7 returned to 25ºC 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 25ºC  showed completely normal NMJ EPSP amplitudes because of a perfect, offsetting homeostatic  Table 1). Of note, the 28.5ºC NMJs had markedly diminished that of animals raised at 25ºC throughout life, indicating a complete reversal of PHP (Fig. 5). Gal4 >> UAS-GluRIIA M/R animals, as well as driver-specific controls. For the driver controls, 3 3 7 quantal size was somewhat diminished at 30ºC (Figs. 6A-D, Table 1). This was consistent with 3 3 8 the idea that quantal size is generally diminished at very high temperatures (Ueda and Wu, 2015) -though it could also be the case that high levels of muscle-driven GAL4 protein at high 3 4 0 temperatures contributes to this phenotype. Evoked EPSP amplitudes were still robust for driver  Table 1). For the dominant-negative NMJs, there was a significant reduc-  Table 1). These data  To extend this line of inquiry, we examined a second homeostatic challenge to NMJ showed relatively normal physiology at 30ºC (Figs., 6E, F, Table 1). NMJs from GluRIIA SP16 de-3 5 0 letion animals raised at 30ºC showed significant PHP, with reduced mEPSP amplitudes and ro- Table 1). We noted that EPSP amplitudes 3 5 2 were somewhat reduced compared to WT (Table 1). Collectively, our data show that normal 3 5 3 baseline neurotransmission is not disrupted at 30ºC. Moreover, PHP is possible and robust (yet imperfect) at 30ºC for GluRIIA SP16 mutants, but it is abolished in the case of the dominant-  We present evidence that presynaptic homeostatic potentiation (PHP) at the Drosophila 3 5 9 melanogaster neuromuscular synapse is a reversible process. In doing so, we confirm prior find- ings showing that there is a tight inverse relationship between quantal amplitude and quantal 3 6 1 content at the NMJ (Fig. 4). We complement those findings by conducting temperature shift ex-3 6 2 periments. We find that PHP is measurable at an early stage of larval development (Fig. 2) and  For the Drosophila NMJ, homeostatic potentiation is a robust and sensitive process. One assumption supported by all available data is that the larval NMJ is capable of modulating its signaling has previously been demonstrated at the Drosophila NMJ by application of  Why is reversibility slow after dominant-negative GluRIIA M/R removal?
There was a robust expression of presynaptic homeostatic potentiation (PHP) for NMJs transgene was halted, this expression of PHP was erased over a slow 48-to 72-hour period 3 8 1 (Fig. 3). 24 hours of halted dominant-negative expression provided no relief (Fig. 3).

8 2
If PHP is a readily reversible homeostatic process, why is there a days-long time lag in dominant-negative GluRIIA M/R subunits that had previously been stably incorporated at the NMJ 3 9 7 likely remained at the synapse. flux (Genç et al., 2017). For almost all of the cases in which a mutation or an experimental con-4 1 2 dition blocks the short-term induction of homeostatic signaling, the same perturbation has also 4 1 3 proven to block its long-term maintenance. Interestingly, the converse is not true. Additional 4 1 4 studies have amassed evidence that the long-term consolidation of homeostatic signaling at the 4 1 5 NMJ can be genetically uncoupled from its induction, and select molecules seem to be dedicat-  As more molecular detail about HSP is elucidated, it will be interesting to test if the rapid ic changes in PHP dynamics at the mouse NMJ were mediated by a calcium-dependent in-  Finally, it is instructive to examine mammalian synaptic preparations to study how ho- 18 tors in order to counteract the perturbation (O'Brien et al., 1998;Turrigiano et al., 1998). Bi-4 3 7 directional scaling suggested that reversible mechanisms likely dictate homeostatic scaling pro-  is an efficient strategy used to stabilize activities in metazoan nervous systems. One advantage  Our data suggest that high temperatures represent a potential limitation on the system. It is not clear what the molecular or anatomical basis of this limit is. We do know that it is not an issue of 4 5 1 NMJ excitation at high temperatures. This is because evoked neurotransmission for WT (or 4 5 2 driver control) NMJs remains remarkably robust over a range of temperatures, including 30ºC 4 5 3 (Table 1). Nor does it seem to be an elimination of PHP in general because PHP was still pre-4 5 4 sent in the case of GluRIIA SP16 animals raised at 30ºC (Fig. 6, Table 1). Rather, the limitation 4 5 5 seems to be on homeostatic signaling that supports PHP at high temperatures in the face of the  shown that NMJ growth plasticity can be additionally affected by mutations that affect neuronal 4 6 4 excitability (Budnik et al., 1990;Lee and Wu, 2010;Zhong et al., 1992). Given the backdrop of 4 6 5 these data, it is not unreasonable to hypothesize that the tolerable limits of synaptic activity 4 6 6 challenge could be different at different temperatures.

6 7
For our experiments, 29-30ºC represents a potential "crash" point for homeostatic poten-4 6 8 tiation at the Drosophila NMJ. We must note, however, that our data suggest that the coping 4 6 9 capacity of the NMJ is dependent on genotype. WT NMJs cope at all temperatures. By contrast, NMJs, there is robust, but imperfect PHP at 30ºC (Fig. 6) -not unlike the compensation seen to the well-documented temperature-induced alterations in NMJ growth -or alternatively, a lim-  We thank members of the Tootle, Lin, Wallrath, and Geyer labs for helpful discussions.   Regulation of synapse structure and function by the Drosophila tumor suppressor gene dlg. Neuron 17, 627-640. Drosophila mutants with altered excitability. J Neurosci 10, 3754-3768.   transmitter at normal, myasthenia gravis and myasthenic syndrome affected human end-plates.           Neuropharmacology 78, 63-74.  ER-resident calcium sensor that stabilizes synaptic transmission and homeostatic plasticity.   Kauwe, G., Tsurudome, K., Penney, J., Mori, M., Gray, L., Calderon, M.R., Elazouzzi, F.,  for clustering glutamate receptors at the neuromuscular junction. Genes Dev 26, 974-987. removal enables neto to stabilize glutamate receptors at the Drosophila neuromuscular junction.