Abstract
Mutations in the gene encoding vesicle-associated membrane protein B (VAPB) cause a familial form of amyotrophic lateral sclerosis (ALS). Expression of an ALS-related variant of vapb (vapbP58S) in Drosophila motor neurons results in morphologic changes at the larval neuromuscular junction (NMJ) characterized by the appearance of fewer, but larger, presynaptic boutons. Although diminished microtubule stability is known to underlie these morphologic changes, a mechanism for the loss of presynaptic microtubules has been lacking. By studying flies of both sexes, we demonstrate the suppression of vapbP58S-induced changes in NMJ morphology by either a loss of endoplasmic reticulum (ER) Ca2+ release channels or the inhibition Ca2+/calmodulin (CaM)-activated kinase II (CaMKII). These data suggest that decreased stability of presynaptic microtubules at vapbP58S NMJs results from hyperactivation of CaMKII because of elevated cytosolic [Ca2+]. We attribute the Ca2+ dyshomeostasis to delayed extrusion of cytosolic Ca2+. Suggesting that this defect in Ca2+ extrusion arose from an insufficient response to the bioenergetic demand of neural activity, depolarization-induced mitochondrial ATP production was diminished in vapbP58S neurons. These findings point to bioenergetic dysfunction as a potential cause for the synaptic defects in vapbP58S-expressing motor neurons.
SIGNIFICANCE STATEMENT Whether the synchrony between the rates of ATP production and demand is lost in degenerating neurons remains poorly understood. We report that expression of a gene equivalent to an amyotrophic lateral sclerosis (ALS)-causing variant of vesicle-associated membrane protein B (VAPB) in fly neurons decouples mitochondrial ATP production from neuronal activity. Consequently, levels of ATP in mutant neurons are unable to keep up with the bioenergetic burden of neuronal activity. Reduced rate of Ca2+ extrusion, which could result from insufficient energy to power Ca2+ ATPases, results in the accumulation of residual Ca2+ in mutant neurons and leads to alterations in synaptic vesicle (SV) release and synapse development. These findings suggest that synaptic defects in a model of ALS arise from the loss of activity-induced ATP production.
Introduction
Amyotrophic lateral sclerosis (ALS) is an untreatable neurodegenerative disease characterized by the progressive loss of motor function leading to paralysis and death by respiratory failure (Taylor et al., 2016). Mutations in many genes have been implicated in the onset of the familial forms of the disease, which result in one or more of the following, disrupted nucleocytoplasmic transport, proteostatic imbalance, alterations in RNA metabolism, genome instability, mitochondrial dysfunction, aberrant Ca2+ homeostasis, neuronal hyperexcitability, and neuroinflammation (Ling et al., 2013; Selfridge et al., 2013; Taylor et al., 2016; Lin et al., 2017; Frere and Slutsky, 2018). In this study, we sought to examine the mechanisms of neuronal dysfunction in flies expressing a missense variant of vapb, which is equivalent to the human variant that causes a familial form of ALS (ALS8) in humans (Nishimura et al., 2004, 2005; Kanekura et al., 2006; Marques et al., 2006; Landers et al., 2008). We chose vesicle-associated membrane protein B (VAPB) because of the importance of this protein to neuronal viability. In addition to the missense variant that functions via a dominant-negative mechanism (Ratnaparkhi et al., 2008), spinal cords from patients with sporadic cases of ALS exhibit decreased VAPB expression (Anagnostou et al., 2010; Mitne-Neto et al., 2011). Defective VAPB function is also observed in Parkinson's disease (Kun-Rodrigues et al., 2015; Paillusson et al., 2017; Boczonadi et al., 2018).
At the molecular level, VAPB is a single-pass endoplasmic reticulum (ER) membrane protein that has been proposed to regulate mTOR signaling, autophagy, lysosomal acidification, proteasomal degradation/ER quality control, and formation of interorganellar contacts that link the ER to endosomes, peroxisomes, Golgi, and mitochondria (Peretti et al., 2008; De Vos et al., 2012; Deivasigamani et al., 2014; Moustaqim-Barrette et al., 2014; Roulin et al., 2014; Dong et al., 2016; Stoica et al., 2016; Hua et al., 2017; Zhao et al., 2018; Chaplot et al., 2019; Mao et al., 2019; Şentürk et al., 2019). Via its role in the formation of ER–mitochondria contacts, VAPB orchestrates the transfer of Ca2+ from ER into the mitochondrial matrix, and thereby, regulates ATP synthesis (De Vos et al., 2012; Stoica et al., 2016; Gomez-Suaga et al., 2017; Smith et al., 2017; Xu et al., 2020; Wong et al., 2021).
Expression of the ALS-related variants of vapb in Drosophila neurons has been shown to result in morphologic alterations in both axons and dendrites (Chai et al., 2008; Ratnaparkhi et al., 2008; Kamemura et al., 2021). Although it is known that defects in the development of the axon termini stem from diminished stability of presynaptic microtubules (Chai et al., 2008; Ratnaparkhi et al., 2008), exactly how mutant VAPB disrupts the microtubule cytoskeleton has remained unknown. In order to uncover the underlying molecular mechanism, we sought to examine genetic interactions between the vapb variant and genes encoding ER Ca2+ channels whose absence elicits phenotypes that ostensibly resemble those evoked by mutant vapb (Wong et al., 2014). These studies revealed that the phenotypes arising from the expression of mutant vapb are a consequence of presynaptic Ca2+ dyshomeostasis, which could stem from the inability of the neurons to synthesize the ATP molecules needed for Ca2+ extrusion. Our findings agree with the insights gleaned from in silico modeling of ALS neurons (Le Masson et al., 2014), and provide a possible explanation for the presynaptic defects associated with the expression of the ALS-causing variant of vapb in Drosophila motor neurons.
Materials and Methods
Drosophila husbandry
Flies were reared at 21°C on standard fly food (1 l of food contained: 95 g of agar, 275 g of Brewer's yeast, 520 g of cornmeal, 110 g of sugar, 45 g of propionic acid, and 36 g of Tegosept) unless otherwise stated. The following fly lines were obtained from Bloomington Drosophila Stock Center: ok371-GAL4 and d42-GAL4 (Parkes et al., 1998; Mahr and Aberle, 2006), RyR16 (Sullivan et al., 2000), UAS-Dcr1 (Dietzl et al., 2007), UAS-CKII-I.Ala (UAS-CKIIala; Joiner and Griffith, 1997), UAS-mCherry-mito-OMM (Vagnoni and Bullock, 2016), and UAS-itprRNAi (TRiP.JF01957; Wong et al., 2021). Other strains used in the study were: UAS-vapbWT and UAS-vapbP58S (Tsuda et al., 2008), UAS-GCaMP5G-tdTomato (Wong et al., 2021), Canton-S, UAS-PercevalHR (Wong et al., 2021), iav1 (also called iavhypoB-1; Wong et al., 2014), and UAS-iav (Wong et al., 2014). Canton-S flies were used as the wild-type controls in the various crosses.
Analysis of RNAi-mediated gene knock-down
Flies expressing the relevant RNAi transgenes or UAS-Luc under the control of heat shock–inducible promoter (hs-GAL4) and hs-GAL4/+ controls were heat shocked on three consecutive days by placing them in a 37°C water bath for 1-h each time. The day after the third heat shock, RNA was extracted from whole-fly extracts using RNeasy mini kit (QIAGEN) by following the manufacturer's instructions. Using the high-capacity cDNA reverse transcription kit (Applied Biosystems), 1 μg of total RNA was reverse-transcribed. Real-time qPCR was performed using PowerUP SYBR Green Master Mix (Applied Biosystems) by following the manufacturer's instructions. The primers used were as follows:
RpL32 forward: TACAGGCCCAAGATCGTGAA
RpL32 reverse: TCTCCTTGCGCTTCTTGGA
vapb (Vap33) forward: TGAAGTGCGTTTTCGAGATGC
vapb (Vap33) reverse: CTGAGCTAGTATTGGCACCCG
Neuromuscular junction (NMJ) immunohistochemistry and confocal microscopy
Dissection and immunostaining of NMJ was performed as described previously (Wong et al., 2014, 2015). Briefly, wandering third instar larvae were filleted in ice-cold PBS to remove all visceral organs except the brain and nerves. For microtubule staining, larvae were dissected in HL-3 (70 mm NaCl, 5 mm KCl, 1 mm CaCl2, 20 mm MgCl2, 10 mm NaHCO3, 115 mm sucrose, 5 mm trehalose, and 5 mm HEPES; pH 7.2), before treated with HL-3 containing 50 μm nocodazole or 0.1% DMSO for 30 min at room temperature. The fillets were fixed in 4% paraformaldehyde in PBS for 30 min. The fixed fillets were washed with 0.1% Triton X-100 in PBS before incubation with primary antibodies overnight at 4°C. Primary antibody dilutions were 1:100 mouse anti-discs large (DLG), 1:50 mouse anti-Futsch, and 1:200 mouse anti-α-tubulin. The monoclonal antibodies against DLG (4F3), Futsch (22C10), and α-tubulin (12G10) were obtained from the Developmental Studies Hybridoma Bank (DSHB) developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA 52 242. The samples were then washed and probed with 1:200 FITC-conjugated anti-HRP (Jackson ImmunoResearch) and Alexa Fluor 568-conjugated anti-mouse secondary antibodies (ThermoFisher) at room temperature for 1.5 h, and then mounted on glass slides with Vectashield (Vector Labs). Confocal images were obtained using a Nikon A1 Confocal Laser Microscope System (Nikon). For NMJ bouton counting, a 60× oil objective was used to focus on the NMJs on abdominal segment 3.
To count the number of microtubule loops within the type 1b NMJ boutons in larvae stained with anti-α-tubulin antibody, we applied an HRP mask to remove the α-tubulin signal outside the boutons. Briefly, confocal images were opened in the Fiji image processing package. We first separated the merged Z-stacks into the anti-HRP and anti-tubulin channels. Next, we applied the “Make Binary” function to the anti-HRP channel using default settings, and having checked “Calculate threshold for each image” and “Black background (of binary masks).” We then generated a maximum intensity Z-projection of the binary image, and inverted the color to generate the mask. Using the “Image Calculator” dialogue window, we subtracted the mask from the anti-α-tubulin stack to remove the anti-α-tubulin signal outside the boutons.
To analyze mitochondria within NMJ boutons, we first separated merged Z-stack images into the mCherry-mito-OMM and anti-HRP signals. We generated an HRP mask exactly as described above to remove any mCherry signal outside the NMJs. To determine the number, volume, and surface area of the mitochondria at the NMJ, we used the “3D Object Counter v2.0” dialogue box with “Threshold” set at 50 and minimum “Size filter” set to 1. We then determined the aspect ratio (volume/surface area ratio) for each entity identified by the software. Operating under the assumption that the aspect ratio of each mitochondrion should be distinct, akin to a fingerprint, we filtered out all the counted particles whose aspect ratio was duplicated in the dataset for that NMJ. The remaining particles were counted in the analyses.
Larval crawling assay
Larval crawling assay was performed as reported previously (Kashima et al., 2017). Briefly, a 100 mm Petri dish half-filled with 2% agar was placed on an LED drawing pad in a dark room. A black paper strip was wrapped around the circumference of the Petri dish to form a 20-mm-tall wall. Late third-instar larvae were collected from fly food-containing vials, briefly rinsed with water, and placed in a 35-mm Petri dish. A paint brush was used to transfer one larva to the center of the agar plate. A digital camera connected to a laptop computer was mounted at ∼80 cm above the Petri dish to capture time-lapse images at 2-s intervals (0.5 frame per second). Image recording continued until the larva reached the perimeter of the dish or for a maximum of 5 min. We used each larva in the assay only once. For analysis, we superimposed the time-lapse images of every 10 s (five frames) using ImageJ (NIH). A translucent ruler with millimeter marks on the LED pad was used as calibration scale. Position of each larva was tracked on the superimposed image to measure the total crawling distance in millimeters. Mean velocity was calculated by dividing the distance by the duration of the time. At least six larvae were recorded and analyzed for each genotype.
NMJ electrophysiology
Wandering third instar larvae were dissected in ice-cold HL-3 (70 mm NaCl, 5 mm KCl, 20 mm MgCl2.6H2O, 10 mm NaHCO3, 115 mm sucrose, 5 mm trehalose, and 5 mm HEPES; pH 7.2) and rinsed with HL-3 containing 0.5 mm Ca2+. Recordings were made from body-wall muscle 6 (abdominal segment 3) with sharp electrodes filled with a 2:1 mixture of 2 m potassium acetate and 2 m potassium chloride. Data were collected from samples with resting membrane potential below −60 mV. Excitatory postsynaptic junctional potentials (EJPs) were evoked by directly stimulating the A3 hemisegmental nerve through a glass capillary electrode (internal diameter, ∼10 μm) at 0.2 Hz. Stimulus pulses were generated by pClamp 10 software (Molecular Devices Inc). The applied currents were 6 ± 3 μA with fixed stimulus duration at 0.3 ms. A total of 20–30 evoked EJPs were recorded for each muscle for analysis. Miniature EJP (mEJP) events were collected for 5 min. Both EJPs and mEJPs were amplified with an Axoclamp 900A amplifier (Molecular Devices) under bridge mode, filtered at 10 kHz and digitized at 10 kHz (EJPs) and 40 kHz (mEJPs) with pClamp 10. Experiments were performed at 21°C. EJPs and paired-pulse stimulation were analyzed with pClamp 8.0 software (Molecular Devices). The mEJPs were analyzed using the Mini Analysis Program (Synaptosoft Inc.). The EJPs paired-pulse amplitudes were corrected by nonlinear summation (Feeney et al., 1998). Paired-pulse ratio was calculated as the ratio of second to first peak. The quantal content of evoked release was calculated from individual muscles by the ratio of the average EJP amplitude over the average mEJP amplitude. For high-frequency stimulation, nerves were stimulated for 10 min at 10 Hz. By fitting the decay portions of each high-frequency trace to first order exponential function, we obtained the rate constant of decays. Rundowns were extracted from the rate-constants of decay.
Dissociation of Drosophila neurons
We dissociated primary motor neurons from Drosophila as described previously (Wong et al., 2021). Briefly, the exterior of wandering third instar larvae was sterilized by brief submersion in ethanol, and then washed with sterilized H2O before dissection in filtered Schneider's medium (S0146; Sigma-Aldrich) containing 10% fetal bovine serum (FBS), antibiotic/antimycotic solution (A5955; Sigma-Aldrich) and 50 μg/ml of insulin (I6634; Sigma-Aldrich). Brains dissected from these larvae were washed in separate wells containing filtered Schneider's medium before being transferred to a filtered HL-3 solution (70 mm NaCl, 5 mm KCl, 1 mm CaCl2, 20 mm MgCl2, 10 mm NaHCO3, 115 mm sucrose, 5 mm trehalose, and 5 mm HEPES) supplemented with 0.423 mm L-cysteine (Calbiochem) and 5 U/ml papain (Worthington; note, after L-cysteine addition but before papain addition, the pH of the solution was recalibrated to 7.4). The brains were then enzymatically digested in the papain solution for 20 min before transfer to a 1.5-ml tube containing 1 ml of filtered Schneider's medium. Cells were centrifuged at 100 G for 1 min before decantation of Schneider's medium. The solution was replaced with 1 ml of fresh filtered Schneider's medium. This process was repeated twice before neurons were dissociated by pipetting repeatedly until the solution was homogeneous. The solution with dissociated neurons was then placed on 35-mm glass bottom dishes (D35-10-0-N; Cellvis) that had been coated with concanavalin A (C2010; Sigma-Aldrich). Cells were cultured in Schneider's medium supplemented with 10% FBS, antibiotic/antimycotic solution (A5955; Sigma-Aldrich) and 50 μg/ml of insulin (I6634; Sigma-Aldrich) at room temperature in a humidified container at room temperature. After each day in culture, cells were washed with PBS to remove any yeast contamination or debris remaining from dissociation.
Live-cell imaging of fly primary neurons
Live imaging of dissociated neurons was performed as described previously (Wong et al., 2021). Briefly, culture media on plates of dissociated neurons was first replaced with HL-3 (70 mm NaCl, 5 mm KCl, 1 mm CaCl2, 20 mm MgCl2, 10 mm NaHCO3, 115 mm sucrose, 5 mm trehalose, and 5 mm HEPES; pH 7.2, room temperature). For measurements of cytosolic Ca2+, GCaMP5G and tdTomato were sequentially excited at 488 and 561 nm, respectively, by an A1 laser confocal microscope with a 40× objective (Nikon). Emission signals at 525 and 595 nm were recorded. Backgrounds were measured from a cell-free region of interest (ROI). Baselines were established for 1 min before addition of muscarine (1 mm). Cytosolic Ca2+ transients were evoked by store-depletion with the SERCA inhibitor, thapsigargin (TG), as described. Amplitudes of the GCaMP5G/tdTomato ratio represented cytosolic free [Ca2+]. More than 50 cells from a minimum of three independently conducted experiments for each condition were used for the calculations.
PercevalHR signals were recorded by measuring the ratio of fluorescence emissions at 525 nm sequentially excited at 487.5 and 407.8 nm. An A1R laser confocal microscope with 40× objective (Nikon) was used for measurement. Background emission signals were measured from a cell-free ROI. Baselines were established for 2 min before the bath was replaced with high [K+] (52 mm) HL-3. Oligomycin A (oligoA; 10 μm) were added as needed and signals were recorded. Amplitudes of the emission ratio represented the cytosolic [ATP]/[ADP] ratio (Tantama et al., 2013; Wong et al., 2021). Data were quantified as change in the PercevalHR ratio over the baseline that was set to the value at the time point immediately before the addition of the drug. We determined the cumulative change in the ratio using a custom R code and the area under the curve (AUC) per unit time. Cells from a minimum of three to four independently conducted experiments for each condition were used for the calculations.
Analysis of fly lifespan
Newly eclosed adult flies of both sexes were collected and transferred to vials containing standard fly food (≤15 flies per vial). Flies were kept at room temperature (∼21°C) and transferred to new vials twice per week. Dead flies at the bottom of the old vials were counted after each transfer until all the animals in a cohort died.
Experimental design and statistical analyses
Excel (Microsoft), and Prism 8 (GraphPad) were used for statistical analyses and curve-fitting. We used either parametric or nonparametric tests of statistical significance on the basis of whether the data were normally distributed. We assessed the normality of data using the Shapiro-Wilk test. Descriptions of the data distribution are provided in the tables. Multiple comparisons were made by ANOVA or Kruskal–Wallis tests. Custom R code used for quantifying [Ca2+] and PercevalHR data. Statistical significance was defined as a p < 0.05. P-values were shown on the figures as asterisks: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Lifespan (Kaplan–Meier) curves were generated using Prism 8. We used the log-rank (Mantel–Cox) test to determine p-values. Other test-specific values are provided in the tables.
Results
Expression of an ALS-related variant of vapb results in defective presynaptic bouton development
Either the deletion of Drosophila vapb (also called Vap33) or the expression of a transgene equivalent to an ALS-causing variant (vapbP58S) disrupts presynaptic microtubules at the larval NMJ resulting in the appearance of fewer, but morphologically larger, boutons (Pennetta et al., 2002; Chai et al., 2008; Ratnaparkhi et al., 2008). We confirmed that in comparison to the GAL4 and UAS controls, ectopic expression of vapbP58S, but not wild-type vapb (vapbWT), in motor neurons (using VGlutok371-GAL4, herein referred to as ok371-GAL4) led to a significant reduction in bouton number (Fig. 1A–E; Brand and Perrimon, 1993; Mahr and Aberle, 2006; Tsuda et al., 2008). NMJ boutons in neurons expressing vapbP58S also exhibited significant increases in bouton area (Fig. 1A,B,F). Although animals expressing vapbWT in motor neurons exhibited a slight, yet significant, increase in bouton area (Fig. 1D,F) relative to the UAS controls, bouton numbers in these animals were not significantly altered (Fig. 1E). Animals expressing vapbP58S, however, exhibited significantly fewer and larger boutons in relation to the GAL4/UAS controls and those expressing vapbWT (Fig. 1E,F). These data agree with prior findings regarding the effects of VAPBP58S on Drosophila larval NMJ synapse morphology (Chai et al., 2008; Ratnaparkhi et al., 2008).
Expression of vapbP58S in Drosophila motor neurons led to significant changes in presynaptic bouton development at the larval NMJ. A–D, Representative confocal images of larval NMJs dissected from animals of the indicated genotypes stained with antibodies against HRP (green) and DLG (magenta). Scale bar shown in A on the top left applies to all panels. Please note that in all figures, UAS-transgene/+ refers to the presence of the noted UAS-transgene construct without a GAL4 driver, and driver>transgene refers to the UAS-transgene, whose expression was driven using the driver-GAL4. E, F, Bar graphs showing quantification of the larval NMJ bouton numbers (E) and relative bouton area (F) in animals of the indicated genotypes. Values represent mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; n.s., not significant, t tests with Bonferroni correction. Dots represent values from distinct NMJs. Refer to Table 1 for statistical information.
Statistical information for the data shown in Figure 1
VAPBP58S-induced defects in presynaptic bouton development are ameliorated by lowering the expression of genes encoding ER Ca2+ release channels
We previously showed that decreased abundance of ER Ca2+ release channels, Inactive (Iav), ryanodine receptor (RyR), and inositol trisphosphate receptor (IP3R), result in NMJs that exhibit fewer boutons that are larger in size (Fig. 2A; Wong et al., 2014). The ostensible similarities between the morphology of NMJ boutons in iav1 mutants (also called iavhypoB-1, strong hypomorphs of iav; Homyk and Sheppard, 1977; Wong et al., 2014) and animals expressing vapbP58S implies a common underlying mechanism. Therefore, we speculated that the coincidence of iav1 and vapbP58S would enhance the respective phenotypes leading to an even greater reduction in bouton number (Fig. 2A). To our surprise, expression of vapbP58S in the iav1 background restored, rather than worsened, both bouton number and morphology to control levels (Fig. 2B,C,J,K). Whereas bouton numbers and area elicited by vapbP58S were not significantly different from those in iav1, these parameters were restored to wild-type levels on the coincidence of vapbP58S and iav1 (Fig. 2J,K). In contrast, overexpression of vapbWT did nothing to alter the paucity of boutons in iav1 (Fig. 2D,E,L). Therefore, decreased abundance of Iav rescued the synaptic growth phenotype stemming from the expression of vapbP58S.
VAPBP58S-induced defects in presynaptic bouton development are ameliorated on concomitant loss of ER Ca2+ release channels. A, Model showing that the development of control NMJs on muscles 6 and 7 of the larval body wall muscles leads to the appearance of numerous small boutons. Animals either expressing vapbP58S in motor neurons or harboring mutations that diminish ER Ca2+ release are characterized by the appearance of fewer, but morphologically larger, boutons. Ostensible similarities in bouton development phenotypes raises the question of genetic interactions between these conditions. Image was created with BioRender. B–I, Representative confocal images of larval NMJs dissected from animals of the indicated genotypes stained with antibodies against HRP (green) and DLG (magenta). Scale bar shown in B applies to all panels. J–N, Bar graphs showing quantification of the larval NMJ bouton numbers (J, L–N) and relative bouton area (K) in animals of the indicated genotypes. Values represent mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001; n.s., not significant, paired t tests with Bonferroni correction. Dots represent values from distinct NMJs. Quantification of bouton numbers in Figure 1 and this figure were from the same experiment. Refer to Table 2 for statistical information.
Statistical information for the data shown in Figure 2
Next, we asked whether the rescue of vapbP58S-induced NMJ phenotypes was specific to the loss of Iav, or whether similar suppression would occur in response to lowered abundance of other ER Ca2+ channels. Simultaneous knock-down of the gene encoding the fly IP3R (itpr) using the d42-GAL4 motor neuron driver and a validated RNAi line (UAS-itprRNAi) ameliorated the effect of vapbP58S expression on bouton number (Fig. 2G,M; Venkatesh and Hasan, 1997; Parkes et al., 1998; Wong et al., 2014, 2021). The d42-GAL4/+ and UAS-vapbP58S/+ control animals did not exhibit a decrease in bouton numbers (Fig. 2M). Absence of comparable suppression in animals coexpressing vapbP58S and a neutral UAS transgene (UAS-Dcr1; Fig. 2F,M) demonstrates that the suppression brought about by coexpression of itprRNAi was not because of GAL4 dilution stemming from the presence of a second UAS transgene. Concomitant absence of one copy of RyR (RyR16/+), which mimics the iav1 NMJ growth phenotype, also prevented vapbP58S-induced alterations in bouton numbers (Fig. 2H,I,N; Sullivan et al., 2000; Wong et al., 2014). Taken together, these data demonstrate that VAPBP58S-induced defects in presynaptic bouton development are ameliorated by cell autonomous reduction in the expression of genes encoding ER Ca2+ channels.
CaMKII overactivation underlies alterations in presynaptic development in motor neurons expressing vapbP58S
The aforementioned findings are consistent with the notion of VAPBP58S inducing an increase in cytosolic [Ca2+] such that the attenuation of ER Ca2+ release restores homeostasis, and thus, mitigates the bouton development phenotypes. Our data also imply that either an increase or decrease in presynaptic [Ca2+] results in morphologically identical bouton phenotypes. Indeed, we previously showed that either the absence or overexpression of iav results in comparable reduction in the number of NMJ boutons (Wong et al., 2014). How might this bell-shaped relationship between presynaptic [Ca2+] and NMJ bouton development come about?
We posit that the key to understanding these outcomes is the stability of presynaptic microtubules. We and others have shown that destabilization of presynaptic microtubules, which manifests as loss of characteristic intrabouton microtubule loops, results in an increase bouton size, and a concomitant decrease in bouton number (Roos et al., 2000; Pennetta et al., 2002; Viquez et al., 2006; Wong et al., 2014). Indeed, the number of presynaptic microtubule loops at NMJs of animals expressing vapbP58S in motor neurons was significantly lower than that in animals expressing vapbWT (Fig. 3A,B). Treatment of NMJs with the microtubule depolymerizing agent, nocodazole, led to a reduction in the number of microtubule loops at the NMJs of animals expressing vapbWT, whereas we observed no further decrease in the number of microtubule loops in nocodazole-treated NMJs in animals expressing vapbP58S (Fig. 3A,B). The number of microtubule loops in nocodazole-treated vapbWT NMJs was statistically indistinguishable from those in nocodazole-treated or -untreated vapbP58S NMJs (Fig. 3B). These data are consistent with microtubule depolymerization underlying the fewer microtubule loops within the vapbP58S NMJs.
CaMKII underlies the alterations in presynaptic bouton development in motor neurons expressing vapbP58S. A, Representative confocal images of larval NMJs dissected from animals of the genotypes indicated on the left, stained with antibodies against HRP (green) and tubulin (magenta) as indicated on the top. Images derived by application of an HRP mask to the anti-tubulin images are shown. Whether or not the larvae were treated with nocodazole is indicated with + or – on the right. Scale bar shown in top left image applies to all the images in the panel. B, Bar graph showing quantification of the number of microtubule loops within each NMJ. Values represent mean ± SEM; *p < 0.05; t tests; and n.s., not significant, Kruskal–Wallis. Dots represent values from distinct NMJs. C, Representative confocal images of NMJs expressing vapbP58S either with or without CKIIala as indicated. Samples were stained with antibodies against HRP (green) and Futsch (magenta). White arrowheads point to individual boutons and highlight recovery of Futsch loops in neurons expressing both vapbP58S and CKIIala. Higher magnification images (high-mag) represent the regions shown in the box overlaid on the merged images. Scale bar shown on the top left applies to all panels. D–F, Bar graphs showing quantification of the number of Futsch loops per NMJ (D) and larval NMJ bouton numbers (E, F) in animals of the indicated genotypes. Values represent mean ± SEM; *p < 0.05, **p < 0.01, ****p < 0.0001; n.s., not significant, paired t tests with Bonferroni correction. Dots represent values from distinct NMJs. Quantification of bouton numbers in Figures 1 and 2 and this figure were from the same experiment. G, Model showing that either an increase or decrease in presynaptic [Ca2+] could result in Futsch hyperphosphorylation and depolymerization of presynaptic microtubules. Elevated [Ca2+] would lead to persistent activation of CaMKII, which could induce Futsch phosphorylation and attendant disruption of presynaptic microtubules leading to the appearance of fewer, but larger, boutons. Decreased expression of ER Ca2+ channels results in lower presynaptic resting [Ca2+] and diminished calcineurin activity (Wong et al., 2014), which also results in Futsch phosphorylation and attendant disruption of presynaptic microtubules leading to the appearance of fewer, but larger, boutons. Image was created with BioRender. Refer to Table 3 for statistical information.
Statistical information for the data shown in Figure 3
We showed previously that reduction in presynaptic resting [Ca2+] and the attendant attenuation of the Ca2+/calmodulin (CaM) responsive phosphatase, calcineurin (CanA1), destabilize presynaptic microtubules (Wong et al., 2014). In the event of a Ca2+-responsive kinase having the same target as CanA1, increased activity of that kinase could also result in hyperphosphorylation of the target. If so, one might expect that either an increase or decrease in cytosolic [Ca2+] would result in higher fractions of that target being phosphorylated. Since activation of the Ca2+/CaM-dependent protein kinase II (CaMKII or CKII) by high [Ca2+] destabilizes microtubules by phosphorylating microtubule-associated proteins (MAPs; Baratier et al., 2006; McVicker et al., 2015; Oka et al., 2017), we asked whether elevated ER Ca2+ release compels the observed changes in NMJ development because of unrestrained CaMKII activity. Mirroring the decrease in the number of presynaptic microtubule loops, NMJs in animals expressing vapbP58S also exhibited fewer loops of the fly MAP1b homolog, Futsch (Hummel et al., 2000; Roos et al., 2000; Fig. 3C,D). Coexpression of the CaMKII inhibitory peptide, CaMKIIala (CKIIala; Joiner and Griffith, 1997), with vapbP58S led to recovery of the numbers of Futsch loops (Fig. 3C,D) and boutons (Fig. 3E). In iav1 synapses, which exhibit lower resting [Ca2+] (Wong et al., 2014), expression of CKIIala did not restore the bouton number (Fig. 3E). Suggesting a specific rescue of the vapbP58S phenotype rather than a general increase in bouton number, overexpression of CKIIala without vapbP58S was not sufficient to elevate the number of presynaptic boutons (Fig. 3E). Indicating the sufficiency of CaMKII downstream of elevated [Ca2+], in iav-overexpressing neurons, which exhibit higher presynaptic [Ca2+] (Wong et al., 2014), bouton number was restored to wild-type level on coexpression of CKIIala (Fig. 3F). Taken together, these data suggest a biphasic dependency of presynaptic microtubule stability and bouton number on cytosolic [Ca2+], and that vapbP58S-induced alterations in presynaptic development stem from aberrant CaMKII activation (Fig. 3G).
Next, we asked whether vapbP58S-induced phenotypes other than NMJ development also stem from elevated presynaptic [Ca2+] and CaMKII activation. Since vapbP58S is equivalent to the ALS-causing variant of human VAPB (Nishimura et al., 2004, 2005; Kanekura et al., 2006; Marques et al., 2006; Landers et al., 2008), we examined locomotor activity in larvae expressing this transgene in motor neurons. To this end, we placed wandering third instar larvae in an open field and recorded their movement toward the periphery of the field (Fig. 4A). In comparison to larvae expressing vapbWT, those expressing vapbP58S exhibited a significant reduction in velocity (Fig. 4A,B). Upon coexpression of either itprRNAi or CKIIala, the velocities of vapbP58S-expressing larvae were no longer significantly different from the animals expressing vapbWT with itprRNAi or CKIIala, respectively (Fig. 4B).
vapbP58S-induced alterations in the velocity of larval peristalsis and adult longevity stem from overlapping and distinct underlying mechanisms. A, Images showing the trajectory of larval crawling in an open field. Each colored line represents the path taken by distinct individual larvae. For each larva, duration for which video capture and mean velocity of crawling are noted. Genotypes of the larvae are indicated. B, Bar graph showing quantification of crawling velocity of larvae of the indicated genotypes. Values represent mean ± SEM; *p < 0.05; n.s., not significant, t tests. Dots represent values from distinct larvae. C, Lifespan of flies of the indicated genotypes; ****p < 0.0001; n.s., not significant, log-rank tests. Refer to Table 4 for statistical information.
Statistical information for the data shown in Figure 4
Expression of vapbP58S in motor neurons shortens Drosophila lifespan via overactivation of PLCβ–IP3R signaling, and either the deletion of the gene encoding a fly PLCβ (norpA; Bloomquist et al., 1988) or the concomitant knock-down of itpr strongly suppresses the effects of VAPBP58S on animal longevity (Wong et al., 2021). In contrast to the ameliorative effects on the NMJ bouton phenotype, neither iav1 nor the RyR16/+ mutations influenced the lifespan of animals expressing vapbP58S (Wong et al., 2021). Therefore, the effects of vapbP58S on NMJ bouton development and adult lifespan occur via distinct mechanisms. In agreement, coexpression of CKIIala, which mitigated the NMJ growth and larval velocity phenotypes in animals expressing vapbP58S, did not influence the effect of mutant VAPB on adult lifespan (Fig. 4C). Collectively, these data show that although larval phenotypes in vapbP58S-expressing animals involves CaMKII, the same is not the case for adult longevity.
Expression of vapbP58S results in diminished extrusion of cytosolic Ca2+
Lower presynaptic resting [Ca2+] in the absence of Iav results in reduced probability of synaptic vesicle (SV) release (Wong et al., 2014). NMJs in iav1 animals, therefore, exhibit reduced amplitudes of evoked EJPs and paired-pulse facilitation in response to a second stimulus applied after a delay of 50 ms (Wong et al., 2014). Conversely, overexpression of iav led to an increase in presynaptic resting [Ca2+], higher EJP amplitudes, and paired-pulse depression indicating elevated SV release probability (Wong et al., 2014). These data point to a dose-dependent effect of ER Ca2+ release on presynaptic resting [Ca2+] and SV release probability. Given that the NMJ bouton phenotypes in animals overexpressing either vapbP58S or iav were suppressed by CKIIala, we asked whether VAPBP58S also elevates SV release probability because of an increase in resting [Ca2+]. If so, expression of vapbP58S would result in increased EJP amplitude and paired-pulse depression. However, EJP amplitudes in vapbP58S neurons were not significantly different from those in vapbWT neurons (Fig. 5A,B), which is in agreement with the report that ectopic expression of the ALS-causing human transgene (hVAPP56S) had no effect on the amplitude of evoked potentials at the Drosophila larval NMJ (Chai et al., 2008). Furthermore, relative to neurons expressing vapbWT, those expressing vapbP58S showed paired-pulse facilitation, rather than paired-pulse depression (Fig. 5A,C). Amplitudes of mini-EJPs (mEJPs) and quantal content were also virtually identical in vapbWT-expressing and vapbP58S-expressing NMJs (Fig. 5D,E). Taken together, these data argue against elevations in presynaptic resting [Ca2+] and SV release probability in vapbP58S-expressing motor neurons.
Expression of vapbP58S results in delayed extrusion of cytosolic Ca2+. A, Representative EJP (left) and paired-pulse EJP (right) traces recorded from larval NMJs isolated from animals expressing vapbWT (blue trace) or vapbP58S (magenta trace) in motor neurons using ok731-GAL4. B, Bar graph showing quantification of the EJP amplitudes from the data shown in A. Values represent mean ± SEM; n.s., not significant, paired t tests. Dots represent values from distinct NMJs. C, Bar graph showing quantification of the paired-pulse ratio (fractional change in the amplitude of the second EJP to that of the first EJP when the two stimulatory pulses were applied 50 ms apart) of the data shown in A. Values represent mean ± SEM; *p < 0.05, t test. Dots represent values from distinct NMJs. D, Bar graph showing quantification of the mini EJP amplitudes. Values represent mean ± SEM; n.s., not significant, paired t tests. Dots represent values from distinct NMJs. E, Bar graph showing quantification of quantal content (ratio of amplitudes of EJP and mini EJP). Values represent mean ± SEM; n.s., not significant, paired t tests. Dots represent values from distinct NMJs. F, Model showing that the rates of [Ca2+] elevation and extrusion are necessary for maintaining the fidelity of synaptic transmission. Paired-pulse facilitation without a change in the amplitude of the first pulse, as seen in A, could be explained by VAPBP58S decreasing the rates of Ca2+ extrusion. Image was created with BioRender. G, Left, Traces showing the decay of GCaMP5G/tdTomato ratio after TG-induced cytosolic [Ca2+] elevations in motor neurons dissociated from animals of the indicated genotypes. Values were fit to a first order exponential function. Values represent mean ± SEM of traces from >50 neurons of each genotype. Right, Box plots showing quantification of half-lives of decay of the signals in individual neurons of the indicated genotypes; *p < 0.05, Mann–Whitney test. Refer to Table 5 for statistical information.
Statistical information for the data shown in Figure 5
What then could explain the hallmarks of elevated [Ca2+] in vapbP58S-expressing motor neurons? Elevations in cytosolic [Ca2+] could stem either from greater Ca2+ influx or from diminished Ca2+ extrusion from the cytosol. Since the former is ruled out by the aforementioned data, we asked whether the effects of VAPBP58S stem from diminished Ca2+ extrusion (Fig. 5F). We reasoned that diminished Ca2+ extrusion would promote the accumulation of residual Ca2+, boost SV release in response to rapidly delivered stimuli, and thereby, explain the paired-pulse facilitation in vapbP58S-expressing neurons (Fig. 5A,C). To examine the rates of Ca2+ extrusion, we performed live-cell imaging of larval motor neurons expressing GCaMP5G-tdTomato (Daniels et al., 2014; Wong et al., 2014, 2021). To ensure that our assessment of Ca2+ extrusion focused on transfer to the extracellular medium rather than uptake into the ER, we applied the SERCA inhibitor, TG (Lytton et al., 1991; Wong et al., 2021). Incidentally, TG-treatment evokes cytosolic Ca2+ transients because of leak of ER stores into the cytosol (Wong et al., 2021), and the kinetics of the return of these transients to baseline reflect the rate of Ca2+ extrusion. We found that the decay of TG-evoked GCaMP5G-tdTomato elevations followed a first order exponential function (Fig. 5G). Half-life of this decay was significantly higher in neurons expressing vapbP58S compared with those expressing vapbWT (Fig. 5G). These data argue in favor of diminished Ca2+ extrusion in neurons expressing vapbP58S.
Rates of ATP production are unable to keep up with demand in vapbP58S-expressing motor neurons
Electrochemical homeostasis in healthy neurons is maintained by pumps powered by mitochondrially-derived ATP (Fig. 6A). While SERCA and PMCA restore depolarization-induced changes in cytosolic [Ca2+] to resting levels, the Na+/K+ ATPase restores the membrane potential (Fergestad et al., 2006; Chouhan et al., 2012; Ivannikov and Macleod, 2013; Le Masson et al., 2014). The purported coupling of ATP production to neuronal activity, therefore, ensures the availability of ATP for maintenance of Ca2+ homeostasis and for membrane repolarization (Fig. 6A, left; Le Masson et al., 2014; Rangaraju et al., 2014; Wong et al., 2021). In ALS neurons, defects in mitochondrial function are predicted to result in a paucity of ATP, which in turn, attenuates the activity of Ca2+ and Na+/K+ ATPases (Fig. 6A, right; Le Masson et al., 2014). In silico modeling has revealed that the outcome of mitochondrial dysfunction and diminished pump activity is a “deadly loop” comprised of a rapidly burgeoning demand for ATP that remains unmet, and Ca2+ dyshomeostasis (Le Masson et al., 2014). Since our findings of diminished Ca2+ extrusion in neurons expressing vapbP58S (Fig. 5G) aligns with the notion of Ca2+ dyshomeostasis, we asked whether the expression of vapbP58S compromises mitochondrial ATP production, which in principle, could preclude a suitable response to depolarization and [Ca2+] elevation.
Response to ATP burden of depolarization in motor neurons depends on the presence of functional vapb. A, Model adapted from Le Masson et al. (2014) showing the role for neuronal ATP in mediating ionic homeostasis. In healthy neurons, ATP derived from mitochondrial oxidative phosphorylation (OXPHOS) powers Ca2+ ATPases, such as SERCA and PMCA, to maintain Ca2+ homeostasis, and the Na+/K+ ATPase, which is needed for setting the resting membrane potential and for repolarization of the membrane potential after bouts of depolarization. In ALS neurons, a decrease in mitochondrial ATP production would attenuate the activities of Ca2+- and Na+/K+-ATPases. This would be predicted to set in motion a self-reinforcing “deadly loop” comprised of continuously increasing ATP consumption coupled with diminished ATP availability, Ca2+ dyshomeostasis, and the eventual loss of membrane potential. Image was created with BioRender. B, D, Representative traces showing normalized PercevalHR ratios in Drosophila motor neurons coexpressing the indicated transgenes. Values represent mean ± SEM of traces from >20 neurons of each genotype. Arrows indicate treatments. C, Bar graph showing relative vapb expression in animals of the indicated genotypes. Values were normalized to the mean in hs-GAL4/+, and represent mean ± SEM; ****p < 0.0001; n.s., not significant, t tests with Bonferroni correction. E, Bar graph showing the cumulative change in the PercevalHR ratio per unit time in neurons dissociated from animals of the indicated genotypes. Values < 0 denote a net decrease in the PercevalHR ratio (i.e., [ATP]/[ADP] ratio) after application of high [K+]. Values represent median ± 95% CIs; *p < 0.05, ****p < 0.0001; n.s., not significant, Mann–Whitney tests with Bonferroni correction. F, Bar graph showing the recovery of PercevalHR ratio during a 30-s window depicted by the green lines in the traces shown in B and D. Values represent median ± 95% CIs; ****p < 0.0001; n.s., not significant, Mann–Whitney tests with Bonferroni correction. G, Box plots showing quantification of half-lives of decay of the GCaMP5G/tdTomato ratio after TG-induced cytosolic [Ca2+] elevations in motor neurons dissociated from animals of the indicated genotypes; ***p < 0.001, Mann–Whitney test. Refer to Table 6 for statistical information.
Statistical information for the data shown in Figure 6
Given that mitochondrial Ca2+ uptake, needed for energizing those organelles and stimulating TCA dehydrogenases, is disrupted in vapbP58S-expressing neurons, it stood to reason that “on-demand” ATP production would be attenuated in these neurons (Duchen, 1992; McCormack and Denton, 1993; Dumollard et al., 2004; Cárdenas et al., 2010; Ding et al., 2018; Wong et al., 2021). To test this idea, we monitored the cytosolic [ATP]/[ADP] ratio in larval motor neurons using the genetically-encoded sensor, PercevalHR (Tantama et al., 2013; Wong et al., 2021). In neurons coexpressing vapbWT, depolarization by the application of high [K+] (52 mm) led to a transient decrease in the PercevalHR ratio (i.e., [ATP]/[ADP] ratio), whereas the ratio remained unchanged in neurons not challenged with high [K+] (Fig. 6B). In vapbP58S-expressing neurons, however, high [K+] led to a more sustained decline in the PercevalHR ratio (Fig. 6B, see below for quantification of cumulative decline).
Since prior reports have indicated that vapbP58S exerts its effects in fly neurons via a dominant-negative mode of action (Ratnaparkhi et al., 2008), we asked whether a reduction in vapb expression would result in a similar defect in sustaining the [ATP]/[ADP] ratio on depolarization. We obtained a transgenic line for vapb knock-down, vapbRNAi, which significantly reduced vapb mRNA levels (Fig. 6C). Upon coexpression with PercevalHR in motor neurons, vapbRNAi led to a sustained decrease in the [ATP]/[ADP] ratio in response to depolarization with high [K+] (Fig. 6D). Comparison of the cumulative change in the PercevalHR ratio per unit time revealed that neurons expressing vapbP58S or vapbRNAi exhibited a significantly greater cumulative decrease in PercevalHR ratio in response to high [K+] than did neurons expressing vapbWT (Fig. 6E). In contrast, the cumulative changes in vapbP58S or vapbRNAi were not statistically significant (Fig. 6E). These data indicate that neurons expressing vapbP58S or vapbRNAi were less capable of responding to the ATP burden of depolarization than were neurons with expressing vapbWT. Recovery of the PercevalHR ratio during a 30-s window after application of high [K+] (Fig. 6B,D, green lines), was significantly greater in neurons overexpressing vapbWT than they were in neurons expressing vapbP58S or vapbRNAi (Fig. 6F). These data show that the rate at which the [ATP]/[ADP] ratio recovers after depolarization is diminished in neurons expressing vapbP58S or vapbRNAi. Finally, we asked whether knock-down of vapb also delays the extrusion of cytosolic Ca2+ as we had observed in neurons expressing vapbP58S. Relative to neurons not expressing the RNAi transgene, those expressing vapbRNAi exhibited delayed recovery from TG-induced [Ca2+] elevation (Fig. 6G). Together, these data indicate that either the knock-down of vapb or the expression of vapbP58S in Drosophila motor neurons compromised the neurons' response to the bioenergetic burden of depolarization, and delayed the rates at which they extrude Ca2+. These conclusions agree with the predictions made by in silico modeling of ALS neurons (Fig. 6A; Le Masson et al., 2014).
Next, we examined the role of OXPHOS in the observed bioenergetic alterations. We found that application of the ATP synthase inhibitor, oligoA, led to the expected decrease of the [ATP]/[ADP] ratio in neurons expressing vapbWT, vapbP58S, or vapbRNAi, regardless of whether the neurons were exposed to normal or high [K+] (Fig. 7A,B). The cumulative decreases in ratio after inhibition of OXPHOS were comparable in the vapbWT and vapbP58S neurons (Fig. 7A,B). These data imply that on cessation of ATP production, the [ATP]/[ADP] ratio comes to rest at similar values in neurons expressing either vapbWT or vapbP58S. These data argue that neuronal [ADP] is not significantly altered in neurons of the two genotypes. In agreement, the cumulative decrease in [ATP]/[ADP] ratio was reduced in depolarized neurons because the burden of depolarization had lowered the [ATP]/[ADP] ratio, even before the application of oligoA. Taken together, our data indicate that the greater cumulative decline in the [ATP]/[ADP] ratio in depolarized vapbP58S neurons, as evident from Figure 6E, reflects the diminished rates of ATP synthesis rather than changes in substrate (i.e., ADP) availability. In neurons expressing vapbRNAi1, however, oligoA-induced reduction in the [ATP]/[ADP] ratio was significantly higher than that in the other two genotypes, both at normal and high [K+] (Fig. 7A,B). These data argue in favor of higher [ADP] in those neurons, which is likely a compensatory response to vapb knock-down.
oligoA-induced changes in the [ATP]/[ADP] ratio, and effects of high-frequency stimulation in neurons expressing vapb variants. A, Traces showing PercevalHR ratio after application of oligoA in neurons dissociated from animals of the indicated genotypes. Values represent mean ± SEM of traces from >20 neurons of each genotype. Arrows indicate treatments. B, Bar graph showing the cumulative change in the PercevalHR ratio per unit time in neurons dissociated from animals of the indicated genotypes. Values represent median ± 95% CIs; ***p < 0.001, ****p < 0.0001; n.s., not significant, Kruskal–Wallis and Mann–Whitney tests. C, Traces showing EJP amplitudes recorded from NMJs of animals of the indicated genotypes stimulated at 10 Hz for 10 min. Traces connecting filled circles represent raw values that were determined experimentally, and the traces connecting filled-squares represent calculated rundowns fit to first order exponential decay. Green arrows indicate the onset of decay. All values represent mean ± SEM of EJP values recorded from five NMJ preparations of each genotype; **p < 0.01; n.s., not significant, paired t tests. D, Representative confocal images of larval NMJs dissected from animals of the genotypes indicated on the top and stained with antibodies against HRP (green). Mitochondria were labeled by mCherry-mito-OMM (magenta). Scale bar shown in top left image applies to all the images in the panel. E–H, Bar graphs showing quantification of the parameters of mitochondrial morphology in animals of the indicated genotypes. Values were normalized to the control mean (E) or to the control median (F–H). Values represent mean ± SEM (E) or median ± 95% CIs (F–H). In E, n.s., not significant, t test, and circles represent values from distinct NMJs. In F–H, *p < 0.05, **p < 0.01; Mann–Whitney tests. Refer to Table 7 for statistical information.
Statistical information for the data shown in Figure 7
Loss of synaptic transmission during high-frequency stimulation in vapbP58S-expressing motor neurons
Many vapbP58S phenotypes we describe in this study, including elevated [Ca2+] and limited [ATP], are also observed in drp1-deficient animals (Verstreken et al., 2005). Since the paucity of ATP in drp1 mutant NMJs precludes the maintenance of synaptic transmission during high-frequency stimulation (Verstreken et al., 2005), we asked whether the shortage of ATP in vapbP58S motor neurons would result in similar rundown of SV release. Although stimulation of the vapbP58S NMJs at 10 Hz led to EJP amplitudes that initially plateaued at values that were ∼35% higher than baseline, those values transitioned to periods of sustained decay after ∼5 min of high-frequency stimulation (Fig. 7C, left, traces with filled circles, green arrow indicates onset of decay). EJP amplitudes in 100% of the traces recorded from vapbP58S animals stimulated at 10 Hz exhibited sustained periods of decay after initial, transient elevations. We reasoned that superimposed on the liminal increase in EJP amplitudes were the rundowns of SV release, which become apparent after ∼5 min of high-frequency stimulation. We extracted the rundown components from the traces using the rate-constants of the decay (Fig. 7C, left, traces with filled squares). Over the 10 min duration of the high-frequency stimulation, calculated rundowns in vapbP58S NMJs dropped significantly lower than baseline (Fig. 7C, left). The decay phase of high-frequency traces recorded from vapbWT NMJs was relatively less-pronounced (Fig. 7C, right), with only 40% of the traces exhibiting sustained periods of declining EJP values. Calculated rundown in vapbWT NMJs was not significantly lower than baseline (Fig. 7C, right). Therefore, high-frequency stimulation leads to a greater loss of synaptic transmission in vapbP58S-expressing motor neurons. In agreement, calculated rundowns in vapbP58S were predicted to plateau at 3.82 mV [95% confidence interval (CI), 3.69–3.94, p < 0.0001], whereas rundowns were predicted to plateau at 11.96 mV (95% CI, 11.81–12.11, p < 0.0001) in vapbWT.
Bioenergetic dysfunction and attendant changes in SV cycling at the drp1 mutant NMJs stem from the absence of local mitochondria because of defects in mitochondrial trafficking to the axon termini (Verstreken et al., 2005). In contrast, coexpression of the mitochondrial marker, mCherry-mito-OMM (Vagnoni and Bullock, 2016), with the vapb variants belied putative defects in the availability of mitochondria at the NMJ (Fig. 7D,E). Mitochondria in the vapbP58S-expressing neurons did, however, exhibit subtle morphologic changes, slight yet significant increases in the relative volume and surface-area of each mitochondrion at the termini (Fig. 7F,G). Mitochondrial aspect ratio (i.e., the volume/surface area ratio of each mitochondrion) was ∼2% lower in vapbP58S NMJs (Fig. 7H), which argues against meaningful changes in fission–fusion dynamics. Collectively, our data argue against changes in mitochondrial trafficking or dynamics as the underlying cause for rundown of synaptic transmission in vapbP58S NMJs.
Discussion
Our findings explain the characteristic changes to NMJ morphology observed in animals that harbor vapb deletions or express an ALS-causing variant of the gene (vapbP58S; Pennetta et al., 2002; Chai et al., 2008; Ratnaparkhi et al., 2008). While it was known that the reduction in NMJ bouton number and increase in bouton size stem from diminished stability of presynaptic microtubules (Pennetta et al., 2002; Chai et al., 2008; Ratnaparkhi et al., 2008), the molecular underpinnings of these cytoskeletal changes had remained unclear. Our finding that nocodazole decreased the number of presynaptic microtubule loops in neurons expressing vapbWT, but not in neurons expressing vapbP58S, argues in favor of microtubule-depolymerization being constitutively elevated in the latter. The additional finding that the morphologic defects in vapbP58S NMJs were attenuated by inhibition of CaMKII implicates aberrant CaMKII activation in the observed NMJ phenotypes. These conclusions agree with prior reports of CaMKII regulating microtubule stability by phosphorylating microtubule-associated proteins (Baratier et al., 2006; McVicker et al., 2015; Oka et al., 2017).
We speculate that the MAP1b homolog, Futsch (Hummel et al., 2000; Roos et al., 2000), is the relevant CaMKII target in this context. Previous studies have shown that hyperphosphorylation of Futsch, which provokes microtubule destabilization, results in the loss of Futsch-loops within NMJ boutons (Gögel et al., 2006; Viquez et al., 2006; Wong et al., 2014). The potential involvement of Futsch also agrees with prior reports of reduced Futsch abundance at the NMJ in a fly model of TDP-43-induced proteinopathy, primarily because of diminished translation of futsch mRNA (Coyne et al., 2014). Therefore, microtubule destabilization with the underlying involvement of Futsch could be a common occurrence in Drosophila models of ALS (Pennetta et al., 2002; Ratnaparkhi et al., 2008; Coyne et al., 2014). Interestingly, expression of a gain-of-function variant of vapb results in the opposite phenotype, increased number of synaptic boutons and Futsch loops (Sanhueza et al., 2014). Given the likelihood that VAPBP58S acts via a dominant-negative mode of action (Ratnaparkhi et al., 2008), there appears to be a direct correlation between the functionality and dosage of VAPB, Futsch-dependent microtubule stability, and bouton number at the larval NMJ (Pennetta et al., 2002; Sanhueza et al., 2014).
Ectopic expression of an overactive TRPV4 channel variant that elevates intracellular [Ca2+] and induces neuropathology via CaMKII hyperactivation (Woolums et al., 2020) sets a precedent for the notion that aberrant CaMKII activation in motor neurons implies higher cytosolic [Ca2+]. Blocking the sources of presynaptic Ca2+ elevation, then, would be expected to restore Ca2+ homeostasis, and thereby, ameliorate the consequences of CaMKII hyperactivation. Indeed, reduced expression of ER Ca2+ release channels in Drosophila neurons counteracts the phenotypes stemming from activating mutations in a voltage-gated Ca2+ channel (Brusich et al., 2018). We found that lowering the abundance of any one of the three presynaptic ER Ca2+ release channels, Iav, RyR, and IP3R, restored bouton morphology and number in neurons expressing vapbP58S. These data also point to the existence of a biphasic, bell-shaped relationship between presynaptic [Ca2+] and NMJ morphology, whereby either a supraphysiological increase or decrease in [Ca2+] elicits morphologically indistinguishable changes at the NMJ.
Do the effects of ER Ca2+ channel knock-down or CKIIala expression on NMJ development in larvae expressing vapbP58S correlate with functional alterations in locomotion?
To answer this question, we first asked whether expression of vapbP58S in motor neurons leads to alterations in larval locomotion. We found that crawling velocity in an open field was significantly reduced in larvae expressing vapbP58S relative to those expressing vapbWT. Concomitant knock-down of the gene encoding IP3R or expression of CKIIala prevented this decrease in the velocity of peristalsis in vapbP58S-expressing larvae. Expression of vapbP58S in motor neurons also leads to premature lethality in adult animals (Wong et al., 2021). In contrast to its restorative effects on NMJ development and larval locomotion, CKIIala had no effect on the abbreviated lifespan observed in adult flies expressing vapbP58S in motor neurons. Therefore, the developmental and longevity phenotypes stem from partially overlapping, yet distinct, molecular pathways. While the NMJ and larval motor phenotypes were suppressed by concomitant reductions in the abundance of Iav and RyR, or by the expression of the CaMKII inhibitory peptide, none of these manipulations influenced the effects of VAPBP58S on adult lifespan. Only with the knock-down of the gene encoding IP3Rs do we observe suppression of NMJ development phenotypes, larval locomotion, and adult lethality, though the former two likely involve CMKII attenuation while the latter, as we showed previously (Wong et al., 2021), stems from mitigation of endolysosomal [Ca2+] overload.
Given that the NMJ phenotypes in neurons expressing vapbP58S were suppressed by the loss of ER Ca2+ channels, which are needed for maintaining presynaptic resting [Ca2+] (Wong et al., 2014), we asked whether VAPBP58S was triggering an increase in resting [Ca2+]. Higher presynaptic resting [Ca2+] augments the probability of SV release resulting in increased amplitudes of evoked postsynaptic potentials that is accompanied by paired-pulse depression (Wong et al., 2014). In agreement with prior reports (Chai et al., 2008), we found that expression of vapbP58S induced neither a change in the amplitude of evoked potentials nor the appearance of paired-pulse depression. Rather, vapbP58S NMJs exhibited paired-pulse facilitation, which in combination with the absence of changes in the amplitude of evoked potentials, was consistent with poststimulus accumulation of residual Ca2+ because of delayed extrusion of presynaptic Ca2+. Indeed, direct examination of cytosolic [Ca2+] using GCaMP5G-tdTomato revealed significantly delayed Ca2+ extrusion in vapbP58S-expressing motor neurons. These data indicate that the aberrant activation of CaMKII in vapbP58S-expressing neurons arose from the accumulation of residual Ca2+ because of delayed extrusion.
What could explain the defects in Ca2+ extrusion in neurons expressing vapbP58S?
The plasma membrane-resident Ca2+ ATPase, PMCA, has been shown to play a major role in the extrusion of presynaptic Ca2+ in Drosophila motor neurons (Ivannikov and Macleod, 2013; Rossano et al., 2013). Given the bioenergetic burden of powering PMCA, we reasoned that a shortage of ATP could underlie the Ca2+ dyshomeostasis observed in vapbP58S-expressing neurons. In support of this model, mutated VAPB has been shown to disrupt interorganellar transfer of Ca2+ from the ER to mitochondria in both mammalian cells and Drosophila neurons (De Vos et al., 2012; Stoica et al., 2016; Gomez-Suaga et al., 2017; Smith et al., 2017; Xu et al., 2020; Wong et al., 2021). Given the role of matrix [Ca2+] in ATP production (Duchen, 1992; McCormack and Denton, 1993; Dumollard et al., 2004; Cárdenas et al., 2010; Ding et al., 2018), we asked whether expression of vapbP58S compromised mitochondrial ATP production. By examining the [ATP]/[ADP] ratio in live neurons, we found that the bioenergetic response to depolarization was relatively stunted in vapbP58S-expressing neurons. The cumulative decrease in [ATP]/[ADP] ratio in depolarized vapbP58S-expressing neurons was greater than that in neurons expressing vapbWT. Eventual recovery of the [ATP]/[ADP] ratio in depolarized neurons was faster if they overexpressed vapbWT, which points to the sufficiency of vapb in sculpting the kinetics of neurons' response to the bioenergetic burden of depolarization. Consistent with VAPBP58S having a dominant negative mode of action (Ratnaparkhi et al., 2008), neurons expressing an RNAi line against vapb also exhibit larger cumulative decrease in the [ATP]/[ADP] ratio, and delayed extrusion of cytosolic Ca2+.
Upon cessation of OXPHOS by application of the ATP synthase inhibitor, oligoA, we found that cumulative decreases in the [ATP]/[ADP] ratio were comparable in neurons expression either vapb variant. Strictly speaking, the levels at which the cytosolic [ATP]/[ADP] ratio settles following application of oligoA, are functions of two parameters: (1) levels of a pool of ATP that is either not available for hydrolysis or is generated from a source other than OXPHOS; and (2) free [ADP]. Absence of evidence supporting the presence of nonhydrolyzable depots of ATP in the cytosol of neuronal cell bodies, however, argues that the cumulative decrease in the [ATP]/[ADP] reflects the abundance of ADP available for ATP synthesis. By extension, the sluggish response to bioenergetic demands of depolarization in vapbP58S neurons does not stem from diminished substrate availability. Rather, the rates of ATP synthesis from ADP are likely lower in those neurons, although elevated rates of ATP hydrolysis could also be contributing to the phenotype. Indeed, in silico modeling has predicted that mitochondrial dysfunction in ALS neurons is sufficient to induce a toxic self-reinforcing loop of increased ATP consumption and a progressive decrease in the availability of ATP (Le Masson et al., 2014). In neurons expressing the RNAi line against vapb, oligoA-induced cumulative decrease in [ATP]/[ADP] ratio was significantly greater than in the other genotypes. These data argue in favor of basally higher net [ADP] on vapb knock-down, which could be a compensatory response to sustained defects in ATP production. The fact that these neurons were still unable to meet the bioenergetic demands of depolarization speaks to profound defects in activity-dependent ATP synthesis in neurons with lower vapb expression. The apparent absence of compensation in neurons expressing vapbP58S could imply a role for vapb dosage in initiating the compensatory response. Nevertheless, the paucity of [ATP] in vapbP58S-expressing neurons during periods of depolarization (i.e., neuronal activity), is consistent with diminished Ca2+ extrusion, which in-turn could provoke the hyperactivation of CaMKII and attendant changes in NMJ morphology.
Maintenance of synaptic transmission at Drosophila larval NMJs during periods of high-frequency stimulation depends on a steady supply of ATP. A local shortage of ATP, for instance, because of an absence of presynaptic mitochondria in drp1 mutants, results in rundown of SV release during high-frequency stimulation (Verstreken et al., 2005). These rundowns occur because of the inability of the synapses to meet the energy demands of SV cycling, and diminished recruitment of the reserve pool of SVs that are otherwise mobilized to replace the rapid depletion of the readily-releasable pool of vesicles (Verstreken et al., 2005). In either the vapbWT-expressing or vapbP58S-expressing NMJs, high frequency stimulation led to initial elevation of synaptic transmission. Superimposed over these liminal elevations were rundowns that became obvious after ∼5 min of stimulation, and were significantly greater in vapbP58S-expressing neurons. These data agree with enhanced rundown of SV release in motor neurons expressing vapbP58S. The magnitude of this phenotype, however, was smaller than that observed in the drp1 mutants (Verstreken et al., 2005). This discrepancy is explained by the absence of presynaptic mitochondria in drp1 mutant neurons (Verstreken et al., 2005), while we observed no significant changes in the number of presynaptic mitochondria in vapbP58S. The relative severity of rundown at drp1 mutant NMJs is likely because of constitutively lower [ATP] in those mutants (Verstreken et al., 2005), whereas shortage of ATP in vapbP58S-expressing neurons is contingent on depolarization.
Footnotes
This work was supported by the National Institutes of Health Grants R03AG063251 (to C.-O.W.), RF1AG069076 (to K.V.), RF1AG067414 (to K.V.), and R21AG072176 (to K.V.). We thank the Bloomington Drosophila Stock Center for fly stocks. We also thank Yufang Chao for technical help. Confocal and live cell microscopy were performed at the University of Texas Health Sciences Center Center for Advanced Microscopy, Department of Integrative Biology and Pharmacology at McGovern Medical School, and the Advanced Imaging Core, Department of Biological Sciences, Rutgers University, Newark.
The authors declare no competing financial interests.
- Correspondence should be addressed to Kartik Venkatachalam at kartik.venkatachalam{at}uth.tmc.edu or Ching-On Wong at chingon.wong{at}rutgers.edu
