For the expression of SNG-1 fusion constructs and channelrhodopsins in C. elegans, the punc-17 promotor (cholinergic motor neurons) and a short version of the ptdc-1 promotor (RIM interneurons) were used. pcDNA3-SypHluorin 4x (S4x) was a gift from Stephen Heinemann & Yongling Zhu (Addgene plasmid #37005¹; RRID:Addgene_37005). pJB14 (TOPO vector::2xpHluorin) was generated using the TOPO cloning kit (Thermo Fisher Scientific Inc., USA) by amplifying two copies of pHluorin cDNA from the pcDNA3-SypHluorin 4x (S4x) plasmid with primers oBJ51 (5′-ATATCGAACCGTCTTCAGATATGGATCTAGCCACC-3′) and oBJ62 (5′-TATATTCGCCGTCTTCTCCACCGCATGTGATTCGA GCTCC-3′). pJB10 (punc-17::sng-1::unc-54-3′UTR) was generated through Gibson assembly by digesting pRM348 with BmtI and BsiWI (punc-17 and backbone), by amplifying pAG52 (sng-1) with primers oBJ58 (5′-TCAGGAGGACCCTTGGCTAGATGGAGAACGTGCGTGCTTA TG-3′) and oBJ59 (5′-ATGACTCGAGCTAATAACCATATCCTTC CGACTGAG-3′) and by amplifying pAH03 (unc-54-3′UTR) with oBJ60 (5′-ATATGGTTATTAGCTCGAGTC ATGGTCGACAAG-3′), and oBJ61 (5′-AAACGCGCGAG ACGAAAGGGCCCAAACAGTTATGTTTGGTATATTGGG-3′). pJB11 (punc-17:: sng-1::2xpHluorin::unc-54-3′UTR) was generated by digestion of pJB10 and pJB14 with BbsI and subsequent ligation to introduce two copies of pHluorin cDNA into the sequence encoding the first intraluminal loop of SNG-1. pDisplay-pHuji was a gift from Robert Campbell (Addgene plasmid #61556²; RRID:Addgene_61556). pDisplay-pHuji was amplified using primers oBJ104 (5′-GCAGAAGAAAACCATGGGCTG-3′) and oBJ105 (5′-CAGCCCATGGTTTTCTTCTGC-3′) to remove the BbsI restriction site to generate pJB24. pJB25 (punc-17::sng-1::pHuji::unc-54-3′UTR) was generated via Gibson assembly by digesting pJB10 with BbsI and amplifying pJB24 with primers oBJ107 (5′-ATATCGAAAAGTCTTCAGGTGGAGGTGGAAGTATGGTGAG CAAGGGCGAG-3′) and oBJ108 (5′-TATATTCGCCGTC TTCGGTGGAGGTGGAAGTCTTGTACAGCTCGTCCATG-3′) which contain the sequence for a GGGGS linker to add in front of the coding region of pHuji. pJB26 (punc-17::ChR2(H134R)::myc) was generated by amplifying ChR2(H134R)::myc using primers oBJ113 (5′-GAACGCTAGCACCACTAGATCCATCTAGAG-3′) and oBJ114 (5′-GCATGCTAGCCACCAGACAAGTTGGTAA-3′) which was introduced into pRM348 by restriction digest with NheI and subsequent ligation. pMSE01 (punc-17::ChrimsonSA::unc-54-3′UTR) was generated by amplifying pDV07 (punc-17::ChrimsonWT::unc-54-3′UTR) with oMSE16 (5′-CGAGTGGCTGCTGGCTTGCCCCGTGAT-3′) and oMSE017 (5′-ATCACGGGGCAAGCCAGCAGCCACTCG-3′) to introduce the point mutation (S169A). pMSE23 (ptdc-1s::ChrimsonSA::unc-54-3′UTR) was generated by amplifying ptdc-1s from pXY07 (ptdc-1s::GFP) with primers oMSE105 (5′-TCCCGGCCGCCATGGCCGCGATTTCTGTATGAGCCGCCCG-3′) and oMSE106 (5′-AAAGACTTTCGATGAATTACTTGGGCG GTCCTGAAAAATG-3′), amplifying the ChrimsonSA backbone from pMSE01 with oMSE107 (5′-CATTTTTCAGGACCGCCCAAGTAATTCATCGAAAGTCTTT CTATTTTCCGCATCTCTTGTTCAAGGGATTGG-3′) and oMSE108 (5′-CGGGCGGCTCATACAGAAATCGCGGCCATGGC GGCCG-3′) and combined the fragments using Gibson assembly. pMSE24 (ptdc-1s::sng-1::pHluorin::unc-54-3′UTR) was generated by amplifying the sng-1::pHluorin backbone from pBJ11 with oMSE114 (5′-AGGGTCGACCATGACTCGAGCTAATAACCATATCCTTC-3′) and oMSE115 (5′-GTAATTCATCGAAAGTCTTTCTATTTTCCG CATCTCTTGTTCAAGGGATTGG-3′) and with oMSE108 and oMSE113 (5′-GAAGGATATGGTTATTAGCTCGAGTCATGGT CGACCCT-3′) and fusing these fragments with the ptdc-1s using Gibson assembly.

Caenorhabditis elegans strains were kept under standard conditions on nematode growth medium (NGM) plates seeded with the Escherichia coli strain OP50, obtained from the Caenorhabditis Genetics Center (CGC, University of Minnesota, USA), at 20°C (Brenner, 1974). The N2 Bristol strain was provided by the CGC and used as wild type. Transgenic animals were generated as described previously (Fire, 1986). An overview of transgenic and mutant strains used or generated in this work can be found in Table 1. For experiments, well-fed L4 larvae were picked ∼18 h before the assays. For RIM experiments, only animals showing marker fluorescence were used. Animals were supplemented with ATR (Sigma-Aldrich, USA) by adding 100 μl OP50 containing 200 μM ATR to 10 ml NGM agar dishes. Experiments were performed on at least two separate days with animals picked from different plates.

Body length assays were performed as described previously (Liewald et al., 2008; Seidenthal et al., 2022). Briefly, light from a 50 W HBO lamp was filtered with a 450–490 nm bandpass excitation filter for ChR2(H134R) and with a 568.5–591.5 nm filter for ChrimsonSA stimulation. ChR2(H134R) was stimulated at 340 μW/mm². ChrimsonSA was stimulated and tested for crosstalk at a light intensity of 40 μW/mm². Brightfield light was filtered with a 665–715 nm filter to avoid unwanted activation of channelrhodopsins. Videos of single animals were acquired and then analyzed using the WormRuler software (Seidenthal et al., 2022). Body length of each worm was normalized to the 5 s period before stimulation and values >120 or <80% of the initial body length were discarded as these are biomechanically impossible and result from artifacts in the background correction.

Videos of crawling animals were acquired as described previously (Vettkötter et al., 2022) and crawling speed measured using the multi-worm-tracker (MWT) setup (Swierczek et al., 2011). Animals were washed three times with M9 buffer to remove OP50 bacteria. They were then transferred to unseeded NGM plates and kept in darkness for 15 min. A light stimulus was applied using a custom-build LED ring (Alustar 3W 30°, ledxon, 623 nm) which was controlled by an Arduino Uno (Arduino, Italy) device running a custom-written Arduino script. Videos were acquired using a high-resolution camera (Falcon 4M30, DALSA) and crawling speed of single animals as well as reversal count (in bins of 10 s) were extracted using “Choreography” software (Swierczek et al., 2011) and summarized using a custom Python script.

For fluorescence imaging, animals were placed upon 7% agarose pads in M9 buffer. Animals were immobilized using a 20 mM Levamisole-hydrochloride (Sigma-Aldrich, USA) solution in M9 and visualized on an Axio Observer Z1 microscope (Zeiss, Germany) equipped with a 100× oil objective. Fluorescent proteins and channelrhodopsins were excited using a 460 and a 590 nm LED system (Lumen 100, Prior Scientific, UK) coupled via a beamsplitter. pHuji and ChR2(H134R) were excited using a double band pass filter (460–500 and 570–600 nm) combined with a 605 nm beam splitter (AHF Analysentechnik, Germany). 460 nm LED light to stimulate ChR2(H134R) was set to 340 μW/mm² intensity. pHuji fluorescence was excited at 280 μW/mm² and filtered using a 615–680 nm emission filter and visualized using an EMCCD camera (Evolve 512 Delta, Teledyne Photometrics, USA). pHluorin and ChrimsonSA were excited using a 450–490/555–590 nm double band pass filter combined with a GFP/mCherry beamsplitter (AHF Analysentechnik, Germany). 590 nm LED light intensity to stimulate ChrimsonSA and excite pHluorin was set to 40 μW/mm². pHluorin fluorescence was filtered using a 502.5–537.5 nm band pass emission filter and visualized using a sCMOS camera (Kinetix 22, Teledyne Photometrics, USA). For increased ChrimsonSA stimulation at 280 μW/mm², we used the same excitation filter as for pHuji experiments but combined with a FITC/TxRed double band pass beamsplitter and a 515–555 nm emission filter (AHF Analysentechnik, Germany). pHluorin fluorescence was again excited with 40 μW/mm². The dorsal nerve cord (DNC) was visualized using the basal pHuji or pHluorin fluorescence. For cholinergic neurons, a region in the posterior third of the animal was imaged, where an abundance of synaptic puncta can be found. For RIM experiments, fluorescent neuronal extensions in the head region were visualized. Videos were captured using the μManager v.1.4.22 software (Edelstein et al., 2014). pHuji – pOpsicle experiments were performed with 50 ms exposure time (20 fps), pHluorin – pOpsicle experiments with 200 ms exposure (5 fps). Stimulation of channelrhodopsins was triggered using a custom written Autohotkey script to activate and deactivate LEDs. Representative images displaying entire worms were acquired using a 40× oil objective and stitched together using the ImageJ Stitching Plugin (Preibisch et al., 2009). The representative image of RIM neurons (Figure 5A) was made using the Z Project function to generate a projection of slices acquired throughout the head region.

Example images were processed, and fluorescence was quantified using ImageJ v1.53 (Schindelin et al., 2012). A region of interest (ROI) was placed on the DNC or RIM neuron using the Segmented Line tool. Pixel width of the line was adjusted according to the width of the fluorescent signal. A background ROI was set in close proximity to the imaging ROI, inside the worm (but avoiding gut autofluorescence) and fluorescence was quantified using the Multi Measure function. XY-drift was corrected using the Template Matching ImageJ plugin, if necessary. Animals that moved excessively or drifted in the focal plane were discarded. Fluorescence was normalized to the average fluorescence before stimulation (F₀) to compare different animals:

A custom written python script was used for background subtraction, normalization and (if needed) filtering of animals according to whether they show a strong response during stimulation (available on GitHub³). For this, the maximum background corrected fluorescence during stimulation was calculated (as a moving average of 1 s). If this was higher than the average background corrected fluorescence before the stimulation +3 × standard deviations (of background corrected fluorescence before stimulation; minimum of two arbitrary units of fluorescence increase), the animal was counted as a strong responder (adapted from Choi et al., 2021); see Supplementary Figure 6A for example traces that fit or do not fit these parameters; animals not fitting the cut-off showed no discernable light-evoked effect on DNC fluorescence. Also, animals that showed an increase of the fluorescence after the end of the stimulus, or animals showing spontaneous events, were excluded. These measures were necessary for the calculation of fluorescence rise and decay kinetics because data from “non-responders” could not be properly fitted. Fluorescence was not corrected for bleaching since bleaching rates were difficult to estimate robustly in the presence of alternating light protocols. Attempts to correct for bleaching led to a progressive deviation toward the end of the acquisition. However, the measured background fluorescence bleached with a similar rate as the fluorescence in the DNC. Thus, subtraction of background fluorescence was sufficient to correct for bleaching in most cases. Fluorescence signal increases in the pOpsicle assay were calculated using the mean of the normalized fluorescence at the first second after stimulation (±SEM) for pHuji experiments or the mean of seconds 15–20 (±SEM) for pHluorin experiments. Regression analysis, to calculate the rate of fluorescence rise and decay, was performed using Graphpad Prism 9.4.1. The equations “Plateau followed by one-phase exponential association” (1) and “Plateau followed by one-phase exponential decay” (2) were fitted to the fluorescence increase during stimulation and decay after stimulation and the time constants τ calculated for each animal. The curve fits were constrained between minimum and maximum fluorescence values to receive reasonable τ values:

where. t, time (in seconds); t₀, timepoint of experimental intervention; set to the first timepoint after the start or stop of stimulation; f₀, value of f(t) at t₀; set to 0 (for Equation 1) or maximum fluorescence values (for Equation 2, as a moving average of five frame intervals); Max, maximum fluorescence during or immediately after stimulation as a moving average of five frame intervals (for Equation 1); Min, minimum fluorescence after stimulation as a moving average of 5 s intervals (for Equation 2); τ, time constant (in seconds; higher τ values indicate a slower rise or decay).

As above, each fit was inspected. Individual fits that showed no increase during stimulation were discarded from analysis as they could not be fitted properly (Supplementary Figure 6A). Similarly, data sets that displayed an increase rather than a decay after stimulation were also discarded.

For the preparation of C. elegans primary cell culture, established protocols were adapted and modified (Christensen et al., 2002; Strange et al., 2007). Gravid adult worms were grown on enriched peptone plates with nystatin (NEP agar) seeded with Na22 E. coli (CGC; Zhang et al., 2011). Worms were washed off the plate using double-distilled water (ddH₂O) and transferred to 15 ml centrifuge tubes. A total of 2 ml of household bleach as well as 1 ml of 5 M NaOH solution were added to 7 ml of worm suspension. The solution was vortexed for at least 5 min at maximum speed to get rid of adult worm bodies. All the following steps were performed under a sterile workbench. The solution, now containing only eggs, was centrifuged at 500 g for 1 min. Excess liquid was removed, and the pellet was resuspended in ddH₂O. Washing was repeated three times. The egg pellet was resuspended in 500 μl freshly thawed chitinase (1 U/ml, Sigma-Aldrich, USA) and transferred to a 1.5 ml tube. The tube was placed into a shaker for 90 min at room temperature to digest the chitin shell of the eggs. The chitinase reaction was stopped with 800 μl L-15 full medium (Gibco, USA) containing 10% fetal calf serum (FCS) as well as Pen/Strep (50 U/ml penicillin + 50 μg/ml streptomycin; Sigma-Aldrich, USA). After centrifugation at 900 g, excess liquid was discarded, and the pellet was resuspended in 500 μl L-15 full medium. Using a 2 ml syringe with an 18-gauge needle, the solution was aspirated and released back into the tube 15–20 times to dissociate the cells. After dissociation, 1 ml L-15 full medium was added to the tube and taken up into the syringe. With the cell solution inside the syringe, the needle was replaced by a 5 μm filter (Millipore, Germany). The solution was released through the filter into a fresh tube. The original tube was refilled with 1 ml L-15 full medium, the filter was replaced by the needle and the procedure was repeated to release the solution into another tube. This was repeated four to six times, depending on the initial number of eggs (more eggs = less repetitions). Filtered cell solutions were centrifuged at 900 g and most of the supernatant was discarded. Cell pellets were resuspended in the remaining medium and pooled. A total of 500–1,000 μl L-15 full medium were added to the suspension and the solution was seeded on 1–2 peanut lectin (Sigma-Aldrich, USA) coated glass bottom petri dishes (ibidi, Germany). Petri dishes were filled with 1 ml L-15 full medium and cells were allowed to adhere for 24 h in a 20°C sterile incubator (Memmert, Germany) before exchanging the medium. Treatment with ATR was performed after two days. Medium was replaced with a 10 μM ATR solution in L15 full medium and the cells incubated for at least 12 h at 20°C. Prior to measurement, cells were washed two times with L-15 full medium without ATR. Neurons were imaged three to four days after seeding. The filter setup was identical to the one used to visualize pHluorin in intact animals, yet neurons were visualized using a 40× objective and 200 ms exposure time. To reduce bleaching during NH₄Cl treatment frame rate was reduced to one frame every 5 s. The camera shutter and 460 nm LED were controlled using a custom-written Arduino script (kindly provided by Ichiro Aoki). Buffers used to either de-quench or quench pHluorin fluorescence were adapted from Dittman and Kaplan (2006) and added manually by pipetting 1 ml of the respective solution onto the petri dish and then removing the same amount of liquid. Before starting a new acquisition, cells were washed three times using the control saline buffer (pH 7.4). Fluorescence was quantified as in living animals. A ROI was set on top of neurite extensions using the ImageJ “Segmented line” tool with a background ROI set in proximity. The releasable fraction of SVs (RF) was calculated for individual cells as the ratio of the mean normalized fluorescence during optogenetic stimulation to the normalized fluorescence during treatment with NH₄Cl containing buffer (mean of 85–90 s; Egashira et al., 2016):

Statistical analysis and plotting of graphs were done using Graphpad Prism 9.4.1. Unpaired t-tests were performed if two normally distributed datasets were compared and one-way ANOVA (with Dunnett’s or Tukey’s correction) or two-way ANOVA (with Šídák’s correction) for three or more datasets. Fluorescence rise and decay constants were compared using the Mann–Whitney test (for two datasets) and the Kruskal–Wallis tests (with Dunn’s correction) or by applying multiple Mann–Whitney tests (with Holm–Šídák’s correction) for three or more datasets.

To enable an all-optical method for analysis of SV exo- and endocytosis, we used proteins with presumably low optical crosstalk. First, we tested the recently described red fluorescent pHuji protein which shows a 22-fold increase of fluorescence when transferred from intravesicular to extracellular pH (Shen et al., 2014), combined with the blue light-gated ChR2(H134R; hereafter called ChR2) for stimulation of neurotransmission (Nagel et al., 2005). Overlap of the pHuji and ChR2 excitation spectra is minimal (Azimi Hashemi et al., 2014; Shen et al., 2014; Lambert, 2019; Supplementary Figure 1A). C. elegans proteins can be targeted to, e.g., cholinergic or GABAergic motor neurons using specific promoters, and for subcellular localization of pHuji we chose the ubiquitous (in neurons) integral SV membrane protein synaptogyrin-1 (SNG-1; Zhao and Nonet, 2001; Abraham et al., 2006). SNG-1 was previously used to target proteins to SVs (Liu et al., 2019; Vettkötter et al., 2022). We inserted pHuji into the second intraluminal loop of SNG-1 (Figure 1) and co-expressed it with ChR2 in cholinergic motor neurons using the promotor of unc-17, encoding the vesicular acetylcholine transporter (Alfonso et al., 1993; Miller et al., 1996; Liewald et al., 2008). SNG-1::pHuji could be observed throughout the cholinergic nervous system, i.e., nerve ring, ventral and dorsal nerve cords (VNC/DNC; Supplementary Figure 1B) and dorsal neuronal extensions (Supplementary Figure 1C). Fluorescent puncta in the DNC indicate accumulation of SNG-1::pHuji at NMJs (Sieburth et al., 2005). Next, we analyzed whether neurotransmission can be activated in these animals using ChR2, which typically leads to muscle contraction and reduction of body length through acetylcholine release (Liewald et al., 2008). As expected, blue light caused a decrease of body length of animals when grown in the presence of ATR (Supplementary Figure 1D), but not in no-ATR controls. Next, we quantified fluorescence of SNG-1::pHuji in the posterior DNC where NMJs are abundant (Sieburth et al., 2005) before, during and after ChR2 photostimulation (Supplementary Figures 1C, E). Since pHuji fluorescence was relatively dim, we could not quantify fluorescence during the stimulation period, as blue light increased autofluorescence and/or further excited pHuji, thus erroneously increasing the measured signal. Alternating light protocols also failed, likely due to the photoswitching nature of pHuji (Liu et al., 2021). However, we could compare the fluorescence before and after stimulation. Animals treated with ATR showed a significantly increased fluorescence after stimulation (19 ± 3% ΔF/F₀), compared to animals without ATR (2 ± 2%, ^***p < 0.001; Supplementary Figure 1E). This increase gradually declined toward the fluorescence level in animals grown without ATR, most likely indicating recycling of externalized SNG-1::pHuji (and thus of SVs). We termed the pHuji-ChR2 combination “red pOpsicle.” To determine whether the red-pOpsicle assay reports on altered SV exo- and endocytosis, we crossed the line carrying the ChR2 and SNG-1::pHuji transgene into mutants known to affect the SV cycle (see section “1. Introduction”; Supplementary Figure 2). Compared to wild type, snb-1(md247) mutants showed a significantly lower fluorescence increase after stimulation with blue light [wild type: 15 ± 2%, snb-1(md247): 2 ± 2%; Supplementary Figures 2A, B], indicating lower amounts of SVs containing SNG-1::pHuji being exocytosed. Deletion mutants of SV recycling factors, unc-26(s1710) and unc-57(e406), displayed a higher increase in pHuji fluorescence than wild type animals [Supplementary Figure 2C: wild type 17 ± 3%, unc-26(s1710) 31 ± 5%, *p < 0.05; Supplementary Figure 2D: wild type 16 ± 3%, unc-57(e406) 30 ± 4%, *p < 0.05]. This was surprising since both mutants were previously shown to be depleted of ready-to-release SVs and should therefore not be able to externalize as many SVs as wild type animals (Harris et al., 2000; Schuske et al., 2003). Nevertheless, they showed prolonged increased fluorescence during acquisition compared to wild type. snt-1(md290) knockout mutants displayed a similar signal increase and decay trend as wild type [Supplementary Figure 2E: wild type 14 ± 3%, snt-1(md290) 18 ± 3%, ns p > 0.05]. We calculated the kinetics of fluorescence decay after stimulation using a one-phase exponential decay model (Supplementary Figures 2F–H). However, no significant difference in the rate of the fluorescence decay was observed. We thus wondered if the pHuji/ChR2 combination works properly. One concern was that photophobic reactions to blue light, mediated by the photoreceptor LITE-1, may affect the measured fluorescence increase and recycling rate (Edwards et al., 2008). However, we did not observe any significant difference between lite-1(ce314) knockout mutants and wild type animals in the pOpsicle assay, independent of ATR treatment (Supplementary Figures 2I, J). We suggest that the previously reported photo-switching behavior of pHuji in blue light (Liu et al., 2021) likely explains the slight increase in fluorescence in animals without ATR after stimulation and could also influence the decay kinetics in animals with ATR (Supplementary Figures 1E, 2I). Since pHuji appeared unable to properly track SV recycling, we investigated another protein combination for the pOpsicle assay.

We tested the more commonly used green fluorescent pHluorin as a pH-sensitive fluorophore which moreover features brighter fluorescence (Miesenbock et al., 1998; Li et al., 2022). Since pHluorin has overlapping excitation spectra with ChR2, we needed to exchange ChR2 with a red light activated channelrhodopsin. The recently described ChrimsonSA (for super red-shifted and accelerated), which is a mutated variant (S169A) of Chrimson, seemed to be a suitable candidate (Oda et al., 2018). Previously, we could show that ChrimsonSA can be used to depolarize C. elegans motor neurons upon stimulation with red light (Seidenthal et al., 2022; Figure 2A). Two copies of pHluorin were inserted into the second intraluminal loop of SNG-1 as multiple pHluorin insertions have been shown to increase the signal-to-noise ratio (Zhu et al., 2009). We co-expressed this fusion construct with ChrimsonSA in cholinergic motor neurons and observed basal pHluorin fluorescence throughout the cholinergic nervous system (Figure 2B). SNG-1::pHluorin also localizes to neuronal extensions of DA and DB neurons innervating dorsal muscle cells (Figure 2C). Fluorescent puncta in the DNC indicate an accumulation of pHluorin at NMJs. We then tested for crosstalk of pHluorin imaging light with ChrimsonSA activation using contraction assays (Figures 2D, E). The intensity of blue light used to visualize pHluorin (470 nm, 40 μW/mm²) did not lead to a significant decrease in body length of animals treated with ATR. In contrast to this, 40 μW/mm² of 580 nm light was sufficient to trigger contraction of body wall muscles. This indicates that ChrimsonSA is functioning properly to depolarize cholinergic motor neurons yet is not pre-activated by blue light. Next, we performed pOpsicle assays with pHluorin animals (“green pOpsicle”). We observed spontaneous pHluorin increases in some animals, independent of treatment with ATR or light stimulation (Supplementary Figure 3A), verifying that pHluorin imaging can report on exocytosis. During photostimulation, DNC fluorescence gradually increased by 21 ± 2% upon continuous 590 nm stimulation in ATR-treated animals (Figures 2F–I, see Supplementary Figure 3B for a representation of all experiments, and Supplementary Movie 1). The signal approached a plateau after ∼10 s, indicating that most SNG-1::pHluorin was externalized at this point. Animals without ATR showed no increase upon stimulation (0 ± 1%, ^***p < 0.001; Figures 2H, I). The increase in fluorescence was especially high in synaptic puncta along the DNC, indicating locations of highly active SV release sites (Figures 2F, G). Following the end of the stimulation, like in red pOpsicle experiments, the signal gradually decreased toward levels before stimulation. We also tested a higher light intensity to stimulate ChrimsonSA (280 μW/mm²) and observed a slightly higher increase in fluorescence yet overall similar trend (Supplementary Figure 3C; 30 ± 3%, *p < 0.05 compared to 40 μW/mm²). Similarly, stronger light stimulation led to a linearly increasing contraction in behavioral experiments (Supplementary Figure 3D). For subsequent analyses we used 40 μW/mm² since it seemed sufficient to release most SVs and stronger stimulation may trigger unphysiological effects.

To further characterize parameters of the “green pOpsicle” assay, e.g., releasable fraction of total pHluorin, we replicated it in neurons of primary cell cultures from dissociated C. elegans embryos (Supplementary Figures 4A, B). pHluorin fluorescence could be observed in neurite extensions and in cell bodies, surrounding nuclei (Supplementary Figure 4A). This fluorescence increased in some neurons when illuminated (and when treated with ATR) even though the signal size was lower than in intact animals, where this was the case for each individual recording (Figure 2I and Supplementary Figure 4B). A total of 18 of 52 neurons showed a strong response (for a definition of “strong response,” see section “Materials and methods”). Next, we calculated the fraction of releasable SVs by ChrimsonSA stimulation. To this end, we first stimulated cultured neurons with light and then applied a pH 7.4 NH₄Cl buffer (Supplementary Figure 4C), which can be used to increase the pH within SVs and calculate the SV-associated portion of pHluorins (Sankaranarayanan et al., 2000; Dittman and Kaplan, 2006; Rose et al., 2013; Egashira et al., 2016). To minimize bleaching, we reduced the exposure to blue light by decreasing the frame rate of pHluorin acquisition. When exposed to NH₄Cl (pH 7.4), pHluorin fluorescence in neurites rapidly increased (Supplementary Figures 4C–E). The same culture was then treated with a low pH buffer (pH 5.6) to quench the surface fraction of pHluorin (Supplementary Figures 4C, F). The fluorescence immediately decreased to levels below the basal fluorescence, demonstrating the pH-dependency of pHluorin fluorescence. The fraction of releasable SVs was determined by calculation of the ratio of maximum fluorescence during light stimulation and during NH₄Cl treatment (Supplementary Figure 4G): 10 s continuous light stimulation released 20.20% (±3.72%) of SNG-1::pHluorin molecules attached to SVs, which is comparable to previous measurements in hippocampal slices that used synaptophysin::pHluorin as a reporter (Rose et al., 2013). To test a different, less vigorous stimulation regime, which may be more reminiscent to endogenous activity, we applied 2 Hz pulsed light (Supplementary Figure 4C). This led to a reduced release of SVs compared to continuous light (Supplementary Figure 4G, 10.61 ± 1.87%). Even though the fraction of releasable SVs by optogenetic stimulation is within a typical range compared to other reports, we conclude that in intact animals the signal size is higher and the assay more reliable in living animals indicating that a higher portion of SVs is exocytosed (Figure 2I and Supplementary Figure 4B).

Next, we used green pOpsicle to analyze mutants affecting SV fusion (snb-1(md247), snt-1(md290)) and/or recycling (unc-26(s1710), unc-57(e406), snt-1(md290)) (Figure 3). In snb-1(md247) mutants, the signal increase during stimulation was almost completely abolished (Figures 3A, B and Supplementary Figure 5A). Only 2 of 19 animals showed a relevant response (compared to 23/23 in wild type), demonstrating that the increase of fluorescence during stimulation reports on SV exocytosis. Synaptojanin (unc-26(s1710)) mutant animals also exhibited a reduced signal compared to wild type (Figures 3C, D and Supplementary Figure 5B). Thus, green pOpsicle faithfully revealed the expected depletion of SVs in unc-26(s1710) mutants (Harris et al., 2000), unlike pHuji that displayed a higher fluorescence after stimulation (Supplementary Figure 2C). To allow calculating the kinetics of fluorescence rise and decay, animals which showed no significant increase in fluorescence or no decay after stimulation were removed from our analyses (Supplementary Figure 6A). We observed a significantly slower rise of fluorescence in unc-26(s1710) mutant compared to wild type animals, as could be seen by the increased time constants of the signal curves obtained from single animals, when fitted to a one-phase exponential association kinetic (τ_{Rise, wild type} = 2.3 s, τ_{Rise, unc–26} = 3.6 s; Figure 3E). This could indicate an exocytosis defect, or it could be due to the reduced number of SVs that are available for fusion in the unc-26(s1710) mutant (Dong et al., 2015). The kinetics of the fluorescence decay after stimulation showed a strong reduction, as expected for the SV recycling mutant (Figure 3F). Consequently, the calculated time constants of one-phase exponential decay were significantly increased in unc-26(s1710) mutants (τ_{Decay, wild type} = 15.2 s, τ_{Decay, unc–26} = 24.2 s), closely matching previous results measured in mammalian cortical neurons lacking synaptojanin after strong stimulation (Cao et al., 2017). Wild type animals on the other hand showed decay time constants which are in the range of previous measurements in C. elegans sensory neurons or mammalian hippocampal neurons (Kwon and Chapman, 2011; Ventimiglia and Bargmann, 2017; Watanabe et al., 2018; Li et al., 2022). Endophilin (unc-57(e406)) mutants exhibited similar trends as unc-26(s1710) with a significantly reduced signal and significantly slower association and decay kinetics compared to wild type (τ_{Rise, wild type} = 2.7 s, τ_{Rise, unc–57} = 3.3 s, τ_{Decay, wild type} = 15.2 s, τ_{Decay, unc–57} = 24.7 s; Figures 3G–J and Supplementary Figure 5C). The time constants obtained for unc-57(e406) mutants closely matched those of unc-26(s1710) mutants, in line with the function of synaptojanin and endophilin in similar cellular processes (Harris et al., 2000; Dong et al., 2015; Watanabe et al., 2018). Synaptotagmin [snt-1(md290)] mutants also displayed a smaller increase in fluorescence (Figures 3K, L and Supplementary Figure 5D). The time constants of fluorescence rise were significantly increased in mutant animals (τ_{Rise,wild type} = 3.1 s, τ_Rise,snt–1 = 5.6 s; Figure 3M), demonstrating the role of SNT-1 in SV fusion (Liewald et al., 2008; Yu et al., 2013). We further observed a significantly delayed decrease of fluorescence after stimulation, indicating that SNT-1 is involved in SV recycling at C. elegans NMJs (Poskanzer et al., 2003; Mochida, 2022; τ_{Decay, wild type} = 11.1 s, τ_{Decay, snt–1} = 31.1 s; Figure 3N). Since the increase in fluorescence was generally lower in mutant animals, we wondered whether higher time constants of decay were caused by a lower activation of synaptic transmission rather than decreased SV recycling rates. We thus compared decay time constants of individual animals with the respective signal size. However, there was no significant correlation within any of the analyzed genotypes (Supplementary Figures 6B–D). Therefore, recycling rates likely do not depend on the amount of SV fusion in this assay.

Continuous photostimulation induces maximal depolarization and transmitter release, likely causing bulk endocytosis as the extreme form of ultrafast endocytosis (Kittelmann et al., 2013). To assess whether less vigorous, possibly more physiological activation also affects slower recycling, we applied 2 Hz pulsed stimulation (100 ms light pulses; Figure 4). The mutant strains again showed significantly reduced signal amplitudes compared to wild type (Figures 4A, B and Supplementary Figure 7) but no significantly increased time constants of fluorescence rise (τ_{Rise, wild type} = 4.2 s, τ_{Rise, unc–26} = 6.0 s, τ_{Rise, unc–57} = 4.7 s, τ_{Rise, snt–1} = 4.7 s; Figure 4C). When comparing pulsed and continuous stimulation we observed a tendency toward decreased fluorescence amplitudes and significantly increased rise time constants, indicating that pulsed stimulation leads to a reduced activation of neurotransmission (Supplementary Figures 8A, B). For the recycling phase, pulsed stimulation again resulted in significantly larger decay time constants in unc-57(e406) and unc-26(s1710) mutants compared to wild type (τ_{Decay,wild type} = 20.9 s, τ_{Decay, unc–26} = 41.1 s, τ_{Decay, unc–57} = 45.1 s; Figure 4D), however, snt-1(md290) mutant animals showed no significant difference (τ_{Decay, snt–1} = 28.6 s). Possibly, SNT-1 is dispensable for recycling at lower levels of stimulation. Decay time constants for pulsed stimulation were significantly increased in unc-57(e406) and unc-26(s1710) mutants indicating that a compensatory mechanism, which is independent of endophilin and synaptojanin, may be activated after strong stimulation (Supplementary Figure 8C).

While the green pOpsicle assay worked well in cholinergic motor neurons, it remained to be shown that this system works in other neuronal cell types. The RIM interneuron pair integrates signals from sensory neurons to regulate forward and reversal locomotion, using gap junctions as well as glutamate and tyramine signaling (Piggott et al., 2011; Li et al., 2020; Sordillo and Bargmann, 2021). We expressed SNG-1::pHluorin and ChrimsonSA in RIM using the tdc-1 promotor. Green fluorescence could be observed in neuronal extensions surrounding the pharynx, suggesting correct localization of SNG-1::pHluorin (Figure 5A). To explore ChrimsonSA functionality in this neuron pair, we measured animal crawling speed (Figure 5B). Optogenetic depolarization of RIM neurons previously induced reversals and reduced crawling speed (Guo et al., 2009; Li et al., 2020; Sordillo and Bargmann, 2021). Consistent with this, illumination with red light slowed down crawling speed of RIM pOpsicle animals treated with ATR and the number of reversals was increased (Figures 5B, C). This indicated that ChrimsonSA depolarized RIM neurons when activated. Consequently, SNG-1::pHluorin fluorescence in synaptic puncta was significantly increased by ChrimsonSA activation (Figures 5D, E, Supplementary Figure 9, and Supplementary Movie 2), while control animals without ATR showed no change (+ATR 6 ± 2%, −ATR −1 ± 1%, ^***p < 0.001; Figure 5F). The signal increase was significantly slower than in cholinergic neurons (τ_{Rise, RIM} = 5.7 s, τ_{Rise, cholinergic} = 2.4 s; Figure 5G), as was the decay of fluorescence following the end of stimulation (τ_{Decay, RIM} = 42.3 s, τ_{Decay, cholinergic} = 16.0 s; Figure 5H). These results indicate that different classes of neurons may have diverging kinetics of SV exo- and endocytosis in C. elegans. Thus, the pOpsicle assay can be adapted to different neuronal cell types.