FM Dye Cycling at the Synapse: Comparing High Potassium Depolarization, Electrical and Channelrhodopsin Stimulation

Figure 1 shows the work-flow for the activity-dependent FM dye imaging protocol. The experiment always begins with the same larval glue dissection, regardless of the stimulation method subsequently used. Figure 1a is a schematic of a dissected larva, showing the ventral nerve cord (VNC), radiating nerves and repeated hemisegmental muscle pattern. The VNC is removed and the preparation bathed in a 4 µM solution of FM1-43 (Figure 1b, pink). The preparation is then stimulated in the presence of FM dye using one of the three possible options to load the dye via activity-dependent SV endocytosis (Figure 1cI-III). Next, FM dye cycling is arrested using Ca²⁺-free saline and a specific NMJ terminal that has been loaded with FM dye is imaged using a confocal microscope ("loaded image"; Figure 1d). In this case, the muscle 4 NMJ is selected with the schematic showing placement of the nerve, NMJ terminal branching and FM loaded synaptic boutons. The synaptic preparation is stimulated for a second time using the same method selected above, but in the absence of FM dye to drive SV exocytosis and FM1-43 release (Figure 1e). The same identified NMJ (muscle 4 NMJ in this example) is then re-imaged using the HRP:647 synaptic label and residual FM1-43 vesicle signal ("unloaded image"; Figure 1f). The experimenter determines the strength and duration of loading and unloading phases to optimize the measurements for a specific assay or mutant condition. Fluorescence intensity is quantified after loading and unloading (Figure 1d, 1f), to obtain the measurements of synaptic dye endocytosis, exocytosis and the percentage of loaded FM dye released. Analyses can be done for the entire NMJ or single boutons.

FM dye cycling can be stimulated in three ways: 1) high [K⁺] saline depolarization of the entire preparation, 2) suction electrode stimulation of the motor nerve, or 3) light driven activation of targeted channelrhodopsin (ChR2; Figure 2). For a direct side-by-side comparison of all three methods, we stimulated with each approach for 5 min in the presence of FM1-43 (loading) and then for 2 min in the absence of FM1-43 (unloading), and imaged HRP-labeled NMJs using identical confocal settings (Figure 2). The high [K⁺] saline depolarization method follows the widely-used FM1-43 protocol for the Drosophila larval NMJ²³, except we use glue instead of pins for the larval dissection. The glue avoids the interference with the microscope objectives and enables more precise determination of body wall tension, but does require a moderate learning curve. All three methods produce strong and consistent fluorescent signals throughout the NMJ synaptic bouton arbor and within individual synaptic boutons (insets). The high [K⁺] saline depolarization method causes all NMJs to be FM-loaded throughout the larva as it depolarizes every neuron in the animal (Figure 2A, middle). The nerve suction electrode electrical stimulation method drives only in a single hemisegment of the larvae for which the corresponding innervating motor nerve has been stimulated (Figure 2B, middle). Unstimulated segments in the same animal are distinguishable using the HRP label but contain no observable fluorescent dye loading, and serve as excellent internal controls. The optogenetic ChR2 method produces targeted depolarization dependent on the transgenic Gal4 driver employed (Figure 2C). Here, the driver was a vesicular glutamate transporter (vglut) Gal4, so all glutamatergic neurons are depolarized with blue light stimulation, including all of the glutamatergic motor neurons. As a result, all NMJs are loaded with FM1-43 dye in this example (Figure 2C, middle). Cell-specific drivers can be used to label specific NMJs and leave others unlabeled for an internal control.

FM dye unloading was achieved via the same method that was used to load dye. In the first method, the larval preparation was simply bathed in high [K⁺] saline in the absence of FM1-43 to depolarize cells, drive SV exocytosis, and unload the synaptic terminals (Figure 2A, right). We typically select a shorter unloading period (2 min), so unloading is partial and terminals remain visible in the FM channel. Suction electrode electrical stimulation unloading is considerably more challenging, because it involves returning to the same hemisegment, identifying the same nerve, and re-stimulating that nerve in the absence of FM1-43 to drive dye unloading (Figure 2B, right). Note that the dye unloading was again partial. The ChR2 unloading entails a second period of blue LED illumination of the larval preparation in the absence of FM1-43 to activate the channels, depolarize the motor neurons and stimulate SV exocytosis dye release (Figure 2C, right). With this timescale of unloading (2 min), each of these depolarization methods unloaded only a percentage of the loaded FM1-43 dye, with the high [K⁺] strongest and ChR2 weakest in driving dye release (Figure 2). We measured the LED light intensity used (140 µW/mm²), which was in the published range (20-1000 µW/mm²)^24,25,26. ChR2 is maximally activated by ~1 mW illumination from 50-200 mW light sources, with ChR2 effectiveness also dependent on the ATR concentration fed to larva^27,28. Thus, ChR2 stimulation strength can be manipulated. Moreover, the relationship may not remain true with other stimulation strengths, timescales or genotypes. If the experimenter is working with a mutant that has diminished SV endocytosis or unusually fast exocytosis, the stimulation parameters may need to be altered to maintain an observable signal after unloading. The anti-HRP label allows one to identify the NMJ even in the complete absence of the FM signal, but this is not ideal for quantification. In principle, the unloading stimulation could also be of a different type from the loading stimulation.

We recently studied the loss of the secreted synaptic deacylase Notum effects on the SV cycle using electrical stimulation⁴. Here, we extend this analysis to compare 1) high [K⁺] depolarization and 2) suction electrode motor nerve stimulation methods (Figure 3). We find that both methods give a similar result, showing reduced FM1-43 dye loading in a Notum null mutant; however, the degree of the phenotype is distinct between the two methods. After the depolarization with high [K⁺] saline for five min, there is a large decrease in loaded fluorescence intensity between genetic background control (w¹¹¹⁸) with high levels and Notum null mutants (Notum^KO) with low levels (Figure 3A, top). In quantified measurements, normalized FM fluorescence intensity within HRP-outlined NMJ terminals is significantly reduced in the absence of Notum (w¹¹¹⁸ 1.0 ± 0.05 vs. Notum^KO 0.57 ± 0.07; n≥13, p<0.0001; Figure 3B). After electrical stimulation at 20 Hz for 1 min, we find an insignificant decrease in loaded FM1-43 intensity between controls and Notum nulls when quantifying whole NMJ fluorescence (w¹¹¹⁸ 1.0 ± 0.05 vs. Notum^KO 0.86 ± 0.06; n = 8, p = 0.10; Figure 3A, B). When measuring dye incorporation on a bouton-per-bouton basis, we find a significant decrease in dye loaded with both of the methods. In quantified measurements after stimulation with high [K⁺] saline, normalized FM fluorescence intensity per bouton is significantly decreased (w¹¹¹⁸ 1.0 ± 0.02 vs. Notum^KO 0.52 ± 0.02; n≥241, p<0.0001; Figure 3C). After electrical stimulation, normalized fluorescence intensity per bouton is also significantly decreased, albeit to a lower degree (w¹¹¹⁸ 1.0 ± 0.02 vs. Notum^KO 0.83 ± 0.02; n≥127, p<0.0001; Figure 3C).

The result differences between the two stimulation paradigms could be due to a number of factors. First, the stimulation strength is presumed to be greater with high [K⁺] compared to electrical motor nerve stimulation. Electrophysiology recordings cannot be done in the presence of the high [K⁺] saline, but the larval neuromusculature is clearly being robustly stimulated, as muscle contractions are strong and continual throughout the 5 min bathing period. However, we do not know the strength or frequency of the stimulation. In contrast, the electrical stimulation is much more controlled with the user choosing the exact voltage strength and frequency of stimulation. Second, the stimulation duration was longer with high [K⁺] compared to electrical nerve stimulation (Figure 3). We chose 1 min of 20 Hz electrical stimulation after repeated trials. After 1 min of dye loading, there was a strong and reliable FM1-43 signal, so we chose that paradigm for our study⁴, although 5 min of stimulation gave an even stronger fluorescent signal (Figure 2). Thus, the length of the stimulation paradigm is likely contributing to the magnitude variability in FM1-43 loading between high [K⁺] and electrical stimulation (Figure 3B, C). Although the high [K⁺] method showed a more robust phenotype for Notum mutants, we still used the electrical method because of the greater control⁴. There are many parameters to control such as strength, frequency and duration of stimulation, and those settings must be decided based on the mutant and the question. For example, the stimulation method choice may depend on the SV pool interrogated⁷ and the activity-dependent mechanism investigated¹⁰.

After FM1-43 loading into the NMJ terminal, one can employ fluorescence dye photoconversion to produce the electron-dense signal for electron microscopy (Figure 4). In this method, the dye-loaded preparation is exposed to intense fluorescent light in the presence of diaminobenzidine (DAB), with the reactive oxygen species from the FM dye oxidizing the DAB to create a dark precipitate (Figure 4A)¹¹. Please see the JoVE article on photoconversion of styryl dyes for complete details¹². The benefit of this method is that SVs are clearly revealed at an ultrastructural level, although the SV cycle is of course arrested with static electron microscope imaging. In the absence of stimulation, synaptic boutons at the Drosophila NMJ are densely loaded with vesicles (~40 nm in diameter; Figure 4B), with the rare occurrence of enlarged organelles (>100 nm in diameter) presumed to be cycling endosomes. High [K⁺] saline depolarizing stimulation strongly changes this profile, with a partial depletion of the SV population and the rapid accumulation of numerous enlarged organelles (>100 nm in diameter) thought to derive from bulk endocytosis of the plasma membrane (Figure 4C, left). Although similar compartments exist in unstimulated controls, the high density of these organelles after high [K⁺] saline depolarization raises the concern that this may be a non-physiological response. Suction electrode electrical stimulation of the motor nerve is relatively more efficient in depleting the SV population and does not produce more of the enlarged endosomal vesicles (Figure 4D, left). This suggests that the bulk endocytosis driven by high [K⁺] depolarization helps maintain the SV population during more intense demand.

FM1-43 photoconversion can be accomplished using either high [K⁺] saline depolarization or suction electrode electrical stimulation of the motor nerve, to compare synaptic ultrastructure relative to a resting bouton (Figure 4). With high [K⁺] depolarizing stimulation, both individual SVs and enlarged vesicles (>100 nm in diameter) can be labeled with the electron-dense DAB precipitate following light-driven photoconversion (Figure 4C, right). For unknown reasons, the enlarged vesicle membrane is typically less densely labeled than the smaller SV membrane. Moreover, SVs often appear filled with the DAB precipitate, rather than just the membrane, but this does make them much easier to distinguish from the unlabeled vesicles. With suction electrode stimulation of the motor nerve, the cycling SV population can also be marked relative to unlabeled SVs (Figure 4D, right). With electrical stimulation, enlarged vesicles (>100 nm in diameter) are not detectably formed in boutons and, consistently, the membranes of the presumed endosomes are not labeled by FM1-43 photoconversion. As above, cycling SVs are typically filled with the DAB precipitate in a somewhat all-or-none fashion, and therefore more easily distinguishable from SVs that have not been formed during the stimulation (Figure 4D, right). We have not yet attempted photoconversion following ChR2 stimulation. With different time courses of stimulation, FM1-43 dye photoconversion allows one to determine where SVs are formed, how SVs are trafficked, and the timing of movement between different spatial pools within the synaptic bouton. The comparison of high [K⁺] and nerve stimulation allows the dissection of bulk and single SV endocytosis mechanisms.

Figure 1: Flowchart of FM1-43 dye loading protocol at the Drosophila NMJ. The larval glue dissection produces a flattened neuromusculature preparation, with the ventral nerve cord (VNC) projecting segmental nerves from the ventral midline (Vm) to the hemisegmentally repeated body wall muscle arrays (step a). The VNC is cut free, and the entire larval dissection is then incubated in the pink FM1-43 solution (4 µM) in preparation for stimulation (step b). FM1-43 is then loaded with a selected stimulation paradigm (step 3); with the options of high [K⁺] depolarization of the entire larva (cI), suction electrode stimulation of a single motor nerve (cII), or light-driven activation of highly targeted channelrhodopsin (cIII). FM1-43 incorporation is arrested using Ca²⁺-free saline and the dye-loaded NMJ imaged (step d). A second stimulation is then done without FM1-43 in the bath to drive dye synaptic vesicle exocytosis (step e). The same NMJ is then re-imaged to assay the unloaded synaptic terminal (step f). Fluorescent intensity is measured from both loaded and unloaded NMJs to quantify SV endocytosis and SV exocytosis levels. The bottom panel shows the construction parameters and dimensions for the transparent acrylic chamber used for these studies. Please click here to view a larger version of this figure.

Figure 2: FM dye loading and unloading comparison of all stimulation methods. Comparison of FM1-43 dye loading and unloading in the wandering third instar NMJ with 1) high [K⁺] depolarization of the entire larval preparation (top), 2) suction electrode electrical stimulation of the motor nerve (middle) and 3) light-driven activation of the targeted channelrhodopsin (ChR2) only in motor neurons (bottom). (A) The larval NMJ labeled with the anti-HRP:647 presynaptic membrane marker (blue, left), loaded with FM1-43 via high [K⁺] depolarization for 5 min (middle) and then unloaded via high [K⁺] depolarization for 2 min. (B) Comparison with suction electrode electrical nerve stimulation with the same stimuli periods for both FM1-43 dye loading and unloading. (C) Targeted vglut-Gal4>UAS-ChR2-H134R expression in motor neurons activated with blue (470 nm) light for the same stimuli periods of FM1-43 dye loading and unloading. Asterisks refer to insets displaying higher magnification boutons. The scale bar is 10 µm, with inset synaptic boutons enlarged 3.5X from main panels. Please click here to view a larger version of this figure.

Figure 3: Notum null mutants show reduced FM dye loading in synaptic boutons. FM1-43 loaded into synaptic terminals of the wandering third instar NMJ comparing the genetic control (w¹¹¹⁸) to Notum null mutants (Notum^KO). (A) NMJ boutons labeled with high [K⁺] depolarization of entire larval preparation (top) or suction electrode stimulation of one motor nerve (bottom) in w¹¹¹⁸ (left) and Notum^KO (right). (B) Quantified loaded FM1-43 fluorescence per NMJ as a scatter plot comparing the high [K⁺] depolarization and suction electrode stimulation in w¹¹¹⁸ controls vs. Notum^KO mutants. (C) Quantified loaded fluorescence on a bouton-per-bouton basis as a box-and-whisker plot comparing high [K⁺] depolarization and suction electrode stimulation in w¹¹¹⁸ controls vs. Notum^KO. Student's two-tailed t-tests were performed for each comparison with p-values displayed on the graphs. Bars show mean ± SEM made with Prism (Version 7.0 for Windows). The electrical stimulation data has been adapted with permission from Kopke et al., Development 144(19):3499-510, 2017. Please click here to view a larger version of this figure.

Figure 4: Synaptic ultrastructure with FM dye photoconversion to mark vesicles. Fluorescent FM1-43 can be photoconverted to an oxidized diaminobenzidine (DAB) electron-dense precipitate for visualization via transmission electron microscopy (TEM). (A) A schematic diagram of the photoconversion method to label the Drosophila NMJ following activity-dependent FM1-43 dye loading. (B) Representative TEM image of a wildtype synaptic bouton at rest (unstimulated). Note dense population of uniform-sized SVs. (C) Synaptic boutons that have been stimulated with high [K⁺] saline depolarization for 5 min, without photoconversion (left) or with FM1-43 photoconversion (right). Note the presence of bulk endosomal structures, labeled and unlabeled. (D) Synaptic boutons that have been electrically stimulated with a nerve suction electrode at 20 Hz for 5 min with no photoconversion (left) or with FM1-43 photoconversion (right). Note that the dye-loaded vesicles appear much darker than adjacent unloaded vesicles. Please click here to view a larger version of this figure.