Cargo crowding at actin-rich regions along axons causes local traffic jams in neurons

Abstract

Steady axonal cargo flow is central to the functioning of healthy neurons. However, a substantial fraction of cargo in axons remains stationary up to several minutes. We examine the transport of precursors of synaptic vesicles (pre-SVs), endosomes and mitochondria in Caenorhabditis elegans touch receptor neurons, showing that stationary cargo are predominantly present at actin-rich regions along the neuronal process. Stationary vesicles at actin-rich regions increase the propensity of moving vesicles to stall at the same location, resulting in traffic jams arising from physical crowding. Such local traffic jams at actin-rich regions are likely to be a general feature of axonal transport since they also occur in Drosophila neurons. Repeated touch stimulation of C. elegans reduces the density of stationary pre-SVs, indicating that these traffic jams can act as both sources and sinks of vesicles. This suggests that vesicles trapped in actin-rich regions are functional reservoirs that may contribute to maintaining robust cargo flow in the neuron.

INTRODUCTION

Microtubule-based molecular motors transport diverse neuronal cargo along the axon, both in the anterograde direction, away from the cell body toward the synapse, as well as in the retrograde direction toward the cell body. Time-lapse fluorescence imaging of in vivo cargo transport shows that mobile and stationary cargo co-exist along the axon (1–4). Although a significant fraction of vesicular cargo and mitochondria in neurons remains stationary over imaging time, studies thus far have largely focused on understanding the characteristics, mechanisms of motion and functions of moving cargo (5–11). Moving vesicular cargo typically pause for up to 15 seconds due to motor-based mechanisms (12–14). Cargo are also known to pause for up to several minutes (1–4, 15) but we know little about the factors that cause cargo to remain immobile over these much longer timescales.

Mechanisms such as a tug of war between motors and motor accumulations at the ends of microtubules have been implicated in stalling of cargo in vitro (12, 16–19) The microtubule binding protein, Syntaphilin, has also been implicated in docking mitochondria to microtubules for up to several minutes (3). Such studies suggest that microtubule-associated proteins and motor-based mechanisms both play roles in interrupting smooth cargo transport. However, it is unclear if similar mechanisms contribute to stalling in vivo, nor is it clear if mechanisms of stalling depend on cargo or neuron type.

Constraints common to all types of moving cargo in axons include the narrow geometry of the axon and its cytoskeletal architecture. The axon is densely packed with microtubules of variable lengths, actin, intermediate filaments, cytoskeleton associated proteins as well as stalled cargo.20–24, 36, 74–77 An in vivo study suggests that synaptic vesicles tend to stop at microtubule ends due to the absence of a track to move ahead (25) Other in vitro studies, performed with motors and microtubules alone, show that crowding agents such as Polyethylene Glycol (PEG) can affect motor movement, as can crowding caused by motor accumulation at the end of the track (19, 26). Additionally, cargo sizes are varied, ranging from 30 nm to 3 μm (20–22) and their motion is likely further constrained in cytoskeletally crowded environments in comparison to motors unbound to cargo. Since a network of filamentous actin is present under the plasma membrane in addition to deep actin along the axon (27, 28) actin-enriched regions could be a potential source of such impediments to cargo motion. Stationary endosomes have also been shown to be present at actin-rich regions in cultured hippocampal neurons (28).

In this study, we show that actin-rich regions along the axon act as hotspots that initiate local traffic jams by stalling moving cargo. This local cargo crowding leads to additional vesicle stalling in these regions. Vesicles trapped at these local traffic jams could form functional reservoirs along the neuronal process. Such reservoirs might underlie the robustness of cargo transport within the physically crowded environment of the axon.

Results

Multiple vesicular cargo in neurons are immobile for long durations

We tracked fluorescently labeled precursors of synaptic vesicles (pre-SVs) and endosomes using time-lapse fluorescence imaging in neurons of Caenorhabditis elegans. We used the following markers: (1) RAB-3 and Synaptobrevin-1 (SNB-1) for pre-SVs in touch receptor neurons (TRNs), (2) RAB-5 for endosomes in TRNs and (3) RAB-3 for pre-SVs in commissures of motor neurons.

Each cargo in a defined neuron has distinct movement parameters (Table S1, Supporting information). We label a tagged vesicle as stationary if it remains immobile for at least 15 times the maximum pause duration of its cognate moving vesicle in the neuron-type imaged (Table S2, materials and methods for details and terminology related to stationary cargo). This cut-off was chosen since we wished to exclude vesicles stalled due to either motor pausing or stalled as a consequence of a tug of war between oppositely directed motors. Both these processes typically occur at timescales of 0.5–15 seconds (12, 13). The timescales of pausing could additionally depend on the nature of the cargo motor complex, the type of cargo, the size of the cargo and the axoplasmic environment of the neuron where these cargo move. We therefore independently calculated cut-offs for different cargo types and neuron types based on our measured pause times for each cargo type imaged (Tables S1 and S2). We consistently observe stationary vesicles in all cargo types mentioned above (Figure 1A–E). These are seen in kymographs as long vertical lines (Figure 1C–E, red arrows). In TRNs of C. elegans, stationary pre-SVs are observed irrespective of the type of pre-SV protein tagged (RAB-3 or SNB-1), their position in the axon or the dosage of the marker (Figure S1A–G). Typically about 5 RAB-3-marked pre-SVs are observed to be stationary across a 10 μm stretch of the TRN process (Figure 1F). This density is independent of both GFP::RAB-3 concentration and the presence of endogenous RAB-3 (Figure S1H). Motor neuron commissures in C. elegans have a lower density of stationary pre-SVs compared to that in TRNs (Figure 1F), while the density of RAB-5-marked stationary endosomal compartments is ~2 ± 1.0 across 10 μm of the TRN process (Figure 1F).

Figure 1.

Stationary vesicular cargo are juxtaposed to immobile F-actin-enriched regions in Caenorhabditis elegans touch receptor neurons (TRNs). (A) Schematic representation marking the region imaged in TRNs of C. elegans (C.e). (B) Time-lapse images showing retrograde (left) and anterograde motion (right) of GFP::RAB-3 tagged pre-SVs (white arrow) in a C.eTRN. Schematic representations of the neurons imaged and corresponding kymograph representations of time-lapse movies of (C) pre-SVs tagged with GFP::RAB-3 in a C.e TRN, (D) endosomal cargo tagged with eGFP::RAB-5 in a C.e TRN, (E) pre-SVs tagged with GFP::RAB-3 in a C.e motor neuron commissure (Comm.); Arrows red: stationary vesicular cargo, yellow: retrogradely moving vesicles and white: anterograde moving vesicles. Cell body (cb) is on the right-hand side in all the kymographs. (F) Density of stationary vesicular cargo per 10 μm observed in 75 seconds time-lapse movies of different cargo types and neurons of C.e (n = 10 animals, number of 10 μm bins analysed ≥15, data represented as mean ± SEM). (G) A Representative kymograph showing actin trails (red arrow, red line in overlay), and immobile F-actin-enriched regions (green arrow, green lines in overlay) in C.e TRN. Lower panel shows the overlay of annotated actin dynamics. (H) Representative kymographs showing juxtaposition between GFP::utCH and mCherry::RAB-3. Right panel shows the annotations of respective markers and yellow/red arrows represent an example where the two markers overlap. Cell body (cb) is on the right-hand side in the kymograph. Note only high-intensity stationary pre-SVs and F-actin-marked regions are annotated in the kymographs. Data were acquired using simultaneous dual-color time-lapse imaging. (I) Percentage of stationary pre-SVs marked by RAB-3 juxtaposed to immobile GFP::utCH-rich regions, utCH−: without Utrophin and utCH+: with Utrophin (n ≥ 12 animals for each, n = 375 utCH-rich regions, n = 228 stationary RAB-3 pre-SVs)

In summary, we observe that various types of vesicular cargo can be stationary for several minutes, and that the density of stationary vesicular cargo is specific to neuron and cargo type.

RAB-3-marked vesicles accumulate and remain immobile for long durations at actin-rich regions

In cultured hippocampal neurons, both cortical actin and deep actin are known to be present throughout the neuronal process (27, 28). Stationary endosomes have been observed in these regions (28). Heterogeneities in the actin cytoskeleton could influence steady cargo flow. We therefore investigated whether stationary pre-SVs along the neuronal process were present in actin-enriched regions. To test this, we examined the juxtaposition of stationary RAB-3-marked pre-SVs with actin-rich regions in C. elegans TRNs.

We used a transgenic line that expresses the GFP tagged calponin homology domain of F-actin-binding protein Utrophin (utCH) as a marker of F-actin-rich regions (29–31). We observed mobile pools of F-actin, manifest as trails in our kymographs with typical trail lengths of 1.5 ± 0.05 μm (Figure 1G, S2A–C, D red arrow), in addition to immobile pools observed as high-intensity vertical lines with average lifetime of 45 ± 0.05 seconds (Figure 1G, S2A–C, E green arrow). These observations recapitulate ones made by Ganguly et al in cultured hippocampal neurons (28). The frequency of actin trails is 5 ± 0.42/100 μm/minute (Figure S2F) and their velocity of extension is 0.42 ± 0.02 μm/s (Figure S2G). We also used Coronin-1::mCherry (COR-1), an orthologue of a mammalian actin-binding protein associated predominantly with F-actin, that is also known to cross-link actin and microtubules in yeast (32–34). About half of the utCH-rich regions are enriched with COR-1 as well (Figure S3A,B). This suggests that utCH and COR-1 mark an overlapping subset of actin-rich regions along the TRN process.

We tested the juxtaposition of RAB-3-marked stationary pre-SVs with actin-rich regions using dual-color, time-lapse imaging experiments. Ninety percent of stationary pre-SVs are associated with utCH-rich regions (Figure 1H,I). To test whether the overlap between the 2 markers is significant and does not occur by chance, we calculated the Manders Overlap coefficient (MOC) (section 4) measuring the spatial overlap between mCherry::RAB-3 and utCH::GFP that suggests significant overlap (Table S3). Nearly 45% of stationary pre-SVs are juxtaposed to COR-1-enriched regions (Figure S3C,D). This difference in extent of juxtaposition of stationary pre-SVs vis a vis the 2 actin markers likely arises due to a lower density of COR-1 (2.5 ± 1.1/10 μm) compared with utCH (10.6 ± 4.5/10 μm). The density of stationary pre-SVs in a region depends linearly on the density of immobile utCH or COR-1 in the same region (Figure S3E,F). Although 90% of all stationary pre-SVs are associated with utCH-marked actin-rich regions, only 59% of utCH-marked regions associate with stationary pre-SVs. Seventy-three percent of COR-1 rich regions associate with stationary pre-SVs. Thus, there are several actin-rich regions that do not have stationary pre-SVs associated with them.

It has been proposed that actin is nucleated from endosomes along the neuronal process (28). We therefore examined the locations where actin polymerizes, observed as utCH trails in our kymographs (Figure 1G, red arrow). Eighty nine percent of utCH trails occur from utCH-enriched regions, 53% of which are associated with stationary pre-SVs (Figure S4A). Only 6% of all the actin-polymerizing events occur at stationary pre-SVs not associated with actin (Figure S4A). Thus, it appears that in C. elegans TRNs pre-existing actin-rich regions act as major sites of actin nucleation unlike endosomes observed by Ganguly et al in 2015. However, we cannot exclude the possibility that cargo stalled at such locations may also contribute to actin polymerization.

In summary, nearly 90% of the stationary pre-SVs are present at utCH-marked F-actin-enriched regions. Only about 10% of stationary pre-SVs are not associated with actin-rich regions. These stationary pre-SVs may be present at the ends of microtubules, which have been previously reported to lead to short-term cargo pausing in vivo35 in other C. elegans neurons, or may be associated with the few COR-1-enriched sites lacking utCH.

Multiple types of cargo accumulate at actin-rich regions

To examine whether multiple cargo types are associated with actin-rich regions, we investigated 2 additional cargo, RAB-5 and mitochondria in C. elegans TRNs. We observe that ~91% of stationary mitochondria and ~63% of stationary RAB-5-marked endosomes are associated respectively with utCH-rich and COR-1-rich regions (Figure 2A–C). Moreover, 95% of stationary mitochondria and 56% of stationary RAB-5-marked endosomes, respectively, also juxtapose with RAB-3-marked stationary pre-SVs (Figure 2D,F). The MOC measuring the spatial overlap between RAB-5-marked stationary endosomes and RAB-3-marked stationary pre-SVs suggests that the overlap between the 2 markers is significant despite their difference in abundance and does not occur by chance (Table S3).

Figure 2.

Multiple types of cargo are present at the same locations along the C. elegans touch receptor neuron process. Representative kymographs (left) and their respective annotations (right) showing juxtaposition between (A) tagRFP::MLS-marked mitochondria associated protein (Mito) and utCH::GFP, (B) COR-1::mCherry and eGFP::RAB-5. Orange arrows represent examples where the 2 markers overlap. (C) Percentage of mitochondria (Mito) and eGFP::RAB5-marked endosomes juxtaposed to immobile actin-rich regions (Actin−: without Actin and Actin+: with Actin, n = 10 animals, n = 58 mitochondria, n = 234 utCH-rich regions, n = 95 RAB-5-marked vesicles, n = 173 COR-1 enriched regions). Representative kymographs (left) and their respective annotations (right) showing juxtaposition between (D) GFP::RAB-3 and tagRFP::MLS-marked mitochondria (Mito), E) RAB-3::mCherry and eGFP::RAB-5. Orange arrows represent examples where the two markers overlap. (F) Percentage of mitochondria (Mito) and eGFP::RAB-5-marked endosomes present with stationary pre-SVs, (stationary pre-SVs+) and without RAB-3-marked stationary pre-SVs, (stationary pre-SVs−) (n = 10 animals, n = 70 mitochondria, n = 200 RAB-3-marked vesicles; n = 145 RAB-5-marked vesicles, n = 178 RAB-3-marked vesicles). Note: in all the overlay images representing juxtaposition between two markers, only high-intensity F-actin signals and stationary pre-SVs are annotated. In all the kymographs, cell body (cb) is on the right-hand side. Data were acquired using either simultaneous or sequential dual-color time-lapse imaging.

Our data suggest that multiple types of cargo stall at the same regions. We also attempted 3 color imaging to directly observe 2 different types of cargo stalled at a location enriched with actin. However, the use of 3 markers whose expression was driven in the same cell resulted in a nearly 99% reduction in flux of all tagged cargo types along the neuronal process. This made quantitation and interpretation of results difficult and hence we did not attempt any further three-color imaging experiments. We also observe a reduction in flux in dual-color imaging compared to imaging the same transgene as a single marker (Figure S4B). However, the density of stationary vesicular cargo in animals expressing 2 transgenes and other vesicle behavior remains unchanged (Figure S4C,D).

Since multiple cargo, RAB-3 and RAB-5-marked vesicles as well as mitochondria all halt at the same actin-rich locations, we infer that cargo experience common constraints to their motion in a complex environment.

Disrupting the actin cytoskeleton reduces the density of stationary RAB-3-marked vesicles

To test the contribution of the actin cytoskeleton in accumulation of stationary vesicular cargo, we injected LatrunculinA (LatA) into the body cavity of C. elegans. We first characterized the effects of LatA injections on actin-rich regions marked by utCH::GFP. We observe that in wild-type animals, the width of each utCH-marked region extends typically to ~1.2 ± 0.8 μm along the neuronal process, with ~46% spanning a 0.6 to 1.4 μm region (Figure S5A). On LatA injection, utCH-rich regions become narrower and the number of utCH-enriched regions spanning 0.6 to 1.4 μm drops to 28% of all utCH-marked regions (Figure 3A, S5A). We also observe that the density of immobile COR-1 and immobile utCH-marked F-actin-enriched regions reduces by 30% (Figure S5C). This reduction in density of actin-rich regions after LatA treatment is concomitant with a corresponding ~30% reduction in the density of RAB-3-marked stationary pre-SVs (Figures 3B,C, S5B,C) without appreciable changes in the flux of pre-SVs (Figure S5D). The extent of reduction in density of stationary pre-SVs is directly proportional to the concentration of LatA injected (Figure 3C). We further see that the fraction of motile vesicles increases from 70% to about 78% to 88% across various LatA treatments (Figure S5E). This small increase in the fraction of motile vesicles could arise due to the fewer actin-rich regions which leads to mobilization of stationary vesicles.

Figure 3.

Perturbation of actin-rich regions reduces the density of stationary pre-SVs in C. elegans touch receptor neurons (TRNs). Representative kymographs of utCH::GFP in injection controls (inj. Controls) and 30 μM Latrunculin A-(LatA) treated animals. Images acquired from C.e TRN. Red arrows show actin trails. Schematic on the right shows actin trails (red sloped lines) originating from immobile actin-rich regions (red straight lines) in injection controls. Overlay shows a few immobile actin-rich regions in LatA treated animals, no actin trails were observed. Representative kymographs of utCH::GFP and RAB-3::mCherry from (A) control and (B) LatA-treated animals. Note: same criterion as mentioned in Figures 1 and 2 was used to annotate overlay images representing juxtaposition between 2 markers. Red arrows represent an example where the 2 markers overlap. Both the green arrow and the dark green line in (A) represent actin trails. Cell body (cb) is on the right-hand side in kymographs. The juxtaposition between utCH::GFP and mCherry::RAB-3 tested using MOC is significant, *P < .05, for all the LatA-treated animals, similar to injection controls (Table S3). Data were acquired using simultaneous dual-color time-lapse imaging. (C) LatA-concentration-dependent reduction in the density of GFP::RAB-3 marked stationary pre-SVs (n ≥ 7 animals for each concentration, data represented as mean ± SEM, one-way ANOVA with Dunn’s multiple comparison test was used for comparison. All the values were compared with buffer injected controls, **P < .01, ***P < .001).

Nearly half of the actin-rich regions that persist after LatA treatment continue to have stationary RAB-3-marked vesicles associated with them. On the other hand, the percentage of stationary RAB-3-marked pre-SVs that are not associated with actin-rich regions almost doubles (Figure S5F). This may reflect actin-independent sources of stationary vesicles that persist after LatA treatment, for example those reported at microtubule ends (35).

Our data suggest that accumulation of stationary pre-SVs along the neuronal process depends on the presence of actin-rich regions and that the density of stationary pre-SVs reduces upon actin disruption.

Cargo crowding at actin-rich regions leads to increased frequency of stalling of moving vesicles

We showed that vesicular cargo are stationary for long periods at regions enriched in F-actin. In the narrow geometry of a C. elegans neuron (22, 37) such actin-rich locations alone or actin-rich locations associated with stationary pre-SVs can both impede the movement of moving vesicles. To test this hypothesis, we compared the behavior of moving pre-SVs when they encounter actin-rich regions with and without pre-existing stationary pre-SVs.

We analysed dual-color movies with F-actin-enriched regions marked by utCH::GFP and RAB-3::mCherry-marked pre-SVs (Movie S1). At stable F-actin-enriched locations that are associated with stationary pre-SVs, 79% of the anterogradely and retrogradely moving pre-SVs stop. Seventeen percent of moving pre-SVs stop on encountering a location enriched with F-actin alone, while only 5% of moving pre-SVs stop (Figure 4A) at a location that has neither F-actin nor a pre-existing stationary pre-SV (Figure 4A). This suggests that actin-rich regions along the neuronal process likely act as hotspots where pre-SVs tend to stall. Consistent with this, we observe vesicles stopping at locations where actin enrichment appears transiently. In about half of the cases when actin enrichment is lost, stationary pre-SVs mobilize (44% ± 11.9%). Vesicles stalled at actin-rich regions appear to act as more effective roadblocks for moving vesicles, stalling them at higher frequencies than in actin-rich regions alone (Figure 4A). Stationary pre-SVs alone in the absence of actin are also able to cause substantial stalling of other moving pre-SVs although less than in regions that contain actin (compare 80% to 60% in Figure 4A,B).

Figure 4.

Stationary cargo at actin-rich regions act as local roadblocks to moving vesicles in Caenorhabditis eleganstouch receptor neurons (TRNs). (A) Percentage of vesicles stopping at or continuing to move through regions (refer to schematic on the right) that are neither actin-rich nor associated with pre-existing stationary vesicles, regions that are only actin-rich, and actin-rich regions with a stationary pre-SV. (Increase in stopping at actin-rich regions associated with stationary cargo compared to regions enriched only with actin is statistically significant, One-way ANOVA followed by multiple comparison, n = 10 animals, number of vesicles analysed ≥100, *P < .05. See section 4 for the details of analysis.) (B) Percentage of vesicles stopping at locations occupied by cognate stationary vesicles in both anterograde and retrograde directions (schematic on right shows cognate vs non-cognate stopping). The bar graphs represent percentage of moving RAB-3-marked pre-SVs stopping at stationary RAB-3-marked pre-SVs lacking actin (utCH−) and percentage of moving RAB-5-marked endosomes stalling at RAB-5-marked stationary vesicles lacking immobile RAB-3-marked pre-SVs. n = 10 animals each, number of vesicles analysed ≥40. (C) Percentage of vesicles stopping at locations occupied by non-cognate stationary cargo in both anterograde and retrograde directions. The bar graphs represent percentage of moving RAB-3-marked pre-SVs stopping at stationary mitochondria (Mito) lacking stalled RAB-3-marked pre-SVs and percentage of moving RAB-5-marked endosomes stalling at RAB-3-marked stationary pre-SVs lacking immobile RAB-5-marked endosomes (n = 12 animals each, number of vesicles analysed ≥40).

We further investigated whether the stopping of moving vesicles in regions where stationary vesicular cargo were present depended on the type of stationary cargo such moving vesicles encountered. We used dual-color time-lapse imaging to label 2 different cargo types. In each case, we assessed stopping at regions lacking their cognate stationary cargo. In C. elegansTRNs, we imaged RAB-3-marked pre-SVs and mitochondria, RAB-3-marked pre-SVs and RAB-5-marked endosomes. We observe that ~6% of mitochondria have no associated stationary pre-SVs at the beginning of the movie (Figure 2F). We tracked every moving RAB-3-marked vesicle encountering a stationary mitochondrion with no associated stationary pre-SVs. We observe that 39% and 27% of the anterogradely and retrogradely moving vesicles respectively, stop after encountering an immobile mitochondria (Figure 4C). We similarly observe that 32% of anterogradely moving and 20% of retrogradely moving RAB-5-marked endosomes stall at sites occupied by stationary RAB-3-marked pre-SVs (Figure 4C). Such stalling of moving vesicles at non-cognate sites suggests that stationary cargo of any type along the neuronal process can act to stall movement of any other cargo, suggesting that physical crowding can impede cargo movement. However, a greater proportion of RAB-5- marked vesicles, 60% stall at locations with stationary RAB-5 vesicles that lack RAB-3 marked stationary pre-SVs (Figure 4B, C). These data suggest that besides physical crowding, there might be additional cargo-specific mechanisms that enhance cargo accretion.

In summary, actin-rich regions, actin rich-regions with stationary vesicles and notably stationary cargo alone all appear to stop vesicles moving through a given region of the neuronal process causing local traffic jams. Our data suggest that cargo crowding, especially at sites enriched with actin, plays a key role in forming local traffic jams along the neuronal process. The role of cargo crowding in stalling may account for the continued presence of stationary vesicles in a location despite loss of actin enrichment.

The presence of stationary pre-SVs locally impedes vesicle motion

Our data suggest that the presence of stationary vesicles can themselves impede cargo movement (Figure 4B). We thus examined: (1) whether the majority of pre-SVs stall where stationary pre-SVs pre-exist, (2) whether number of pre-SVs stopping in a location decreases after stationary pre-SV in that location mobilize and (3) if a moving pre-SV travels a shorter distance in a region with high density of stationary pre-SVs.

Using GFP::RAB-3 and SNB-1::GFP movies in TRNs and GFP::RAB-3 in motor neuron commissures, we observe that only 5% of pre-SVs stall away from pre-existing stationary pre-SVs in both neuron types (Figure 5A). To address the effects on stopping of moving pre-SVs at locations where stationary pre-SVs mobilize, we count the number of moving pre-SVs that stop or continue moving through the same location before and after stationary pre-SV mobilization in our 5-minute time lapse movies of GFP::RAB-3 and SNB-1::GFP (Figure 5B). As expected, there is a substantial decrease in the number of vesicles stopping and an increase in the number of vesicles going through after mobilization of stationary pre-SVs from a location (Figure 5B,C). This increase is ~3-fold with no appreciable change in the number of vesicles that encounter the region before or after stationary pre-SVs mobilize (Table S4). Finally, to assess whether the moving pre-SVs travel shorter distances in regions with a high density of stationary pre-SVs, we chose non-overlapping 20 μm regions on kymographs with differing densities of stationary pre-SVs. We calculate the distance a given pre-SV moves in these regions before stopping, termed as runlength. We find that the average of total run lengths of moving pre-SVs in a region is inversely proportional to the density of stationary pre-SVs (Figure 5D) and that a typical pre-SV travels around 4.8 ± 0.6 μm before stalling at stationary pre-SVs.

Figure 5.

Stationary pre-SVs locally modulate motion of moving vesicles. (A) Percentage of pre-SVs stopping away from stationary pre-SVs in the neuronal process of C.e TRNs and C.e motor neurons commissures (Comm.) (n = 10 animals, number of vesicles analysed ≥200). (B) Kymograph illustrating the quantitation of flux at a location from where a stationary RAB-3-marked pre-SV has mobilized in C.eTRN; (a) when stationary pre-SV was present, (b) stationary pre-SV mobilized and (c) after stationary pre-SV mobilization. Overlay image of events (right) occurring is shown for clear visualization. Red line: stationary pre-SV, green line: mobilization of stationary pre-SV, yellow and magenta line: trajectories of moving vesicles before and after mobilization of stationary pre-SV respectively. (C) Percentage of moving vesicles crossing a site in the presence of a stationary pre-SV, stationary pre-SV+ and after it has mobilized, stationary pre-SV−, in C.e TRN (paired t-test, n = 10 animals, number of vesicles analysed ≥50, number of sites = 20, data represented as Mean ± SEM). (D) Regression plot between the density of stationary pre-SVs and average run length of moving pre-SVs marked by GFP::RAB-3 in the same region of C.e TRN. Dotted lines in the 2 graphs represent 95% confidence band for the best fit line (n = 10 animals analysed, number of regions = 14, n = 80 vesicles, *P < .05).

To summarize, the stalling of moving pre-SVs occurs largely at locations with pre-existing stationary pre-SVs, suggesting that the local obstruction of transport might occur preferentially at such locations. Local increases in flux after stationary pre-SVs mobilize, and the dependence of distance traveled on the density of stationary pre-SVs, are consistent with locally impeded transport where the locations with stationary pre-SVs act as physically crowded roadblocks for any other moving vesicle at that location.

Vesicular cargo transport is impeded at actin-rich regions in Drosophila neurons

To test if actin-rich regions associated with stationary vesicular cargo result in local traffic jams of microtubule-based cargo transport in another model system, we examined axons of Drosophila chordotonal neurons. We used dual-color imaging to investigate: (1) the juxtaposition between stationary vesicular cargo and actin-rich regions, marked by RAB-4-mRFP (38) and Lifeact-GFP, respectively, (39) and (2) the behavior of moving vesicles through actin-rich regions with associated stationary RAB-4-marked vesicles.

We observe that about 70% of the stationary RAB-4-marked vesicles are found close to Lifeact-rich regions and that 70% of Lifeact-enriched locations have stationary RAB-4-marked vesicles juxtaposed to them (Figures 6A, B, S6A, B). As observed in C. elegans TRNs, LatA treatment reduces the percentage of Lifeact-rich regions by 30% with a corresponding reduction in the density of RAB-4-marked stationary vesicles (Figure 6C, Figure S5C). These data suggest that actin-rich regions contribute to the presence of stationary vesicular cargo in Drosophila chordotonal neurons as well.

Figure 6.

Regulation of moving cargo at F-actin-enriched locations in chordotonal neurons of Drosophila melanogaster. (A) Left: kymographs made from movies of Lifeact-GFP and RAB-4-mRFP. Right: overlay of 2 markers used for juxtaposition analysis. Cell body is on the right-hand side in kymographs. Note: same criterion as mentioned in Figures 1 and 2 was used to annotate overlay images representing juxtaposition between two markers. Orange arrows represent a location where both markers are present. (B) Percentage of RAB-4 stationary vesicles juxtaposed to Lifeact-rich regions, Lifeact−: without Lifeact and Lifeact+: with Lifeact, (n = 10 animals, n = 160; Lifeact-rich regions, n = 196 vesicles). The juxtaposition between RAB-4-mCherry and Lifeact-GFP tested using MOC is significant, *P < .05, for all animals (Table S3). (C) Density of stationary vesicular cargo marked by RAB-4-mRFP per 10 μm in control animals and animals treated with 50 μM LatA (data represented as mean ± SEM, n = 10 animals and 10 bins analysed, Student’s t-test, *P < .05). (F) Percentage of RAB-4-marked vesicles stopping at or continuing motion (refer to schematic on the right) without interruption after encountering heterogeneous locations along the neuronal process (Increase in stopping at actin-rich regions associated with stationary vesicles compared to regions enriched only with actin is statistically significant, One-way ANOVA followed by multiple comparison, n = 10 animals, n = 200 vesicles, *P < .05. See section 4 for analysis details). All data were acquired using sequential dual-color time-lapse imaging.

To investigate whether actin-rich regions with stationary RAB-4-marked vesicles act to stall moving RAB-4-marked vesicles we carried out the same comparative analysis described earlier. We observe that on encountering a location enriched with both RAB-4-marked stationary vesicles and Lifeact, ~70% of moving RAB-4-marked vesicles stop (Figure 6D). Around 15% of moving RAB-4-marked vesicles stop on encountering regions enriched with Lifeact but without RAB-4-marked stationary vesicles (Figure 6D). The majority of moving vesicles either marked by RAB-4 or SYT-1 stall predominantly at stationary vesicular cargo in Drosophila neurons as well (Figure S6C). Since C. elegans TRNs exhibit similar obstruction of transport at regions that are both actin-rich and have pre-existing stationary vesicles (Figure 4A), impeded vesicular transport at such physically crowded locations can be argued to be a general phenomenon associated with axonal transport of vesicles.

Vesicles stop at and emerge from stationary vesicle clusters forming vesicle reservoirs

Our kymographs show that more than 95% of pre-SVs moving over a 70 to 80 μm length of the TRN will eventually encounter pools of stationary vesicular cargo and stop (Figure 5A). Unless compensated by the remobilization of stationary vesicular cargo, this is likely to lead to a steady reduction in vesicle flux along the neuronal process over time.

To determine if moving vesicles can emerge from stationary vesicles, we first examined the numbers of vesicles present at such vesicular clusters. We find that there are ~2 to 4 vesicles in a stationary vesicle cluster of RAB-3-marked pre-SVs in C. elegans TRNs as well as RAB-4-marked vesicles in Drosophila neurons, based on the distribution of intensities of individually paused vesicles (Figures 7A, S6D, section 4). Consistent with this, we observe that when stationary vesicles disperse completely in C. elegans TRNs, they release 2 to 4 vesicles (Figure 7B). Similar small groups of synaptic vesicles have been observed in Electron microscopy (EM) sections in asynaptic sites along the axons of Hermaphrodite Specific Neuron (HSN) neurons of C. elegans (40) and may be similar to the clusters of stationary pre-SVs seen in this study. Since several vesicles are present at stationary vesicles, we examined whether these locations contribute to flux along the neuron, using photoconversion of Dendra-2::RAB-3 (Figure 7C,D). We find that stationary pre-SV clusters release a vesicle approximately every 20 seconds (Figure 7D), a number comparable to that obtained from our conventional GFP::RAB-3 kymographs (Figure 7E). Vesicles not only stall at locations where other vesicles stall but also emerge from these locations and contribute to cargo flux along the neuron.

Figure 7.

Vesicles stop at and emerge from stationary vesicles. Approximation of number of vesicles present at a stationary vesicle cluster (A) by using average intensities of individual paused vesicles in both Caenorhabditis elegans TRNs and Drosohila melanogaster chordotonal neurons and (B) by counting number of vesicles that arise when a stationary vesicle disperses into multiple vesicles in C. e TRNs (n ≥ 10 animals, number of stationary vesicles that mobilize ≥ 15, number of stationary vesicles ≥ 20). (C) Representative kymograph showing Dendra-2::RAB-3 photoconversion in C.e TRN. (D) Number of Dendra-2::RAB-3-marked vesicles emerging from a location containing stationary pre-SVs per 20 seconds after photoconversion, (n = 20 animals, number of stationary pre-SVs analysed ≥30). (E) Average number of vesicles emerging from a location containing stationary vesicles in kymographs acquired from different neurons expressing vesicle markers (n ≥ 10 animals, number of stationary vesicules analysed ≥10). (F) Representative kymographs of control and stimulated animals with wild type, mec-4(u253), and mec-10(tm1552) genotypes acquired from C.eTRN. G) Density of stationary pre-SVs per 20 μm in control (Cntrl) and stimulated animals using different stimulation paradigms and across genotypes in C.e TRN (Student’s t-test, *P < .05, **P < .01, n = 10 animals, n ≥ 15 bins for all).

Since the motion of pre-SVs depends on the UNC-104/KIF1A motor (41), we examined whether a change in levels of motor on the cargo surface resulted in altered pre-SV behavior on encountering stationary vesicles. We examined wild-type animals, an unc-104 mutant encoding a Kinesin-3 motor defective in cargo binding with known reduction in levels of motor on the pre-SV surface (42), and animals overexpressing the UNC-104 motor. A reduction in UNC-104/KIF1A does not change the locations where pre-SVs stall with ~82% of all stationary pre-SVs present at stable actin-rich regions, similar to wild type (Figure S6E). Further, the mutants do not show any change in the density of immobile utCH-tagged F-actin when compared to wild type (Figure S6F). In unc-104 animals, the fraction of anterogradely moving RAB-3-marked vesicles that stop at locations with pre-existing stationary pre-SVs significantly increases (Figure S7A). In addition, the number of vesicles that emerge to move in the anterograde direction from a pre-existing pool of stationary RAB-3 vesicles is 9% lower than in wild type (Figure S7A). Retrogradely moving vesicles, as expected, do not show any changes in the fraction of pre-SVs stopping at stationary pre-SVs in unc-104 animals (Figure S7B). Vesicles in animals overexpressing UNC-104 behave similar to wild type suggesting that sufficient motors may already be present on the cargo surface in wild type (Figure S7A,B).

The ability of moving vesicles to incorporate with, as well as emanate from, the locations of stationary pre-SV clusters, suggests that these locations serve a natural function as reservoirs of vesicles along the entire neuronal process. Levels of motors on the cargo surface appear to play a role in both stalling as well as in mobilization of vesicles from stationary vesicle clusters. Motor regulators have also been reported to similarly affect behavior of moving vesicles at vesicle clusters (43, 44). Thus, the behavior of pre-SVs at a physically crowded location can be modulated by the numbers of motors on the cargo.

Density of stationary pre-SV clusters reduces after stimulation

To investigate whether stationary pre-SV clusters can act as reservoirs of vesicles which can be modulated in response to external signals, we repeatedly stimulated C. elegans TRNs using either an eyelash or a Polydimethylsiloxane (PDMS) device with flexible pillars, equivalent to artificial dirt (45, 46). We observe that the density of stationary pre-SV clusters in TRNs reduces after repeated stimulation in both experimental paradigms and is independent of the marker used to tag the vesicles, viz. GFP::RAB-3 or mCherry::RAB-3 (Figure 7F,G). The density of immobile utCH-rich regions is similar in both stimulated and unstimulated animals (Figure S7C). To determine whether this reduction in density depends on the ability of the animal to sense gentle touch, we used mec-4 and mec-10 mutant animals. MEC-4 and MEC-10 encode subunits of the mechanically gated channels expressed in the TRNs that are essential for touch sensitivity (47–50). Upon repeated touch stimulation using an eyelash touch, we observe no reduction in the density of stationary pre-SVs in both mec-4 and mec-10 mutants (Figure 7F,G).

Our data thus suggest that locations with stationary pre-SV clusters along the neuronal process can function as reservoirs of vesicles dependent on activity of the neuron and that this change occurs independent of actin density but may depend on molecular motors.

DISCUSSION

Crowded environments pose special problems for steady transport. The axon is narrow and filled with cytoskeletal elements, associated proteins and other cargo, leading to a congested transport path that cargo must navigate (20–23, 25, 27, 28, 36, 37, 51–53). Cargo accumulation along such crowded paths appears to be a general feature of axonal transport arising from such common impediments to all types of transport. Our experiments show that multiple types of vesicular cargo in different neurons and across different organisms indeed stall for times up to several minutes (Figures 1A–F, 6A). We observe that nearly 90% of stationary pre-SVs and majority of other types of cargo we examined that are stationary are present at actin-rich regions (Figures 1G–I, 2A–C, 6A–B). Such stalling can arise from the physical constraints caused by the actin mesh itself. In the case of C. elegans, for instance, it is known that a filamentous architecture, likely composed of actin, is present in proximity to the plasma membrane and connects microtubules to it (22). The average length of these filaments is around 14 ± 3 nm (22). Though the pore size of the mesh is not known, multiple such filaments can potentially physically constrain cargo motion. However, Myosin associated with cargo can also impede their motion due to tug of war or association with actin-rich regions (54–58). We observe that different types of cargo stall at actin-rich locations (Figures 1G–I, 2A–B), moving vesicles can also stall at stationary cargo themselves in the absence of actin (Figure 4B) and the presence of stationary pre-SVs at actin-rich regions increases the propensity of moving vesicles to stall at the same location (Figures 4A, 6D). These observations suggest that physical crowding itself provides a mechanism for cargo stalling that can operate independently of, or in addition to, specific cargo associated myosin-actin interactions (Figure 8).

Figure 8.

Model of local traffic jams caused by crowding and their influence on transport. Actin and stationary cargo initiate local traffic jams that impede moving cargo. These local traffic jams form vesicle reservoirs that are mobilized according to cellular needs.

Physical crowding might arise since cargo themselves vary in size from 30 nm to 3 μm (22, 56, 59) and thus can locally clog the axonal transport path, leading to local traffic jams. This is expected in TRNs which have an average diameter of ~4 μm that is filled with about 45, 15 protofilament microtubules with an outer diameter of ~30 nm, thus occupying a large volume of the available space (37). Cargo stalling is also likely to depend on a combination of cargo size, the composition of the motors on the cargo surface, the stiffness of the cargo and any actin-binding proteins on the cargo surface.

Additionally, such physical crowding can itself influence motion properties through motors. A previous study shows that stalled cargo slow the motion of other moving cargo in their vicinity (36). Further, the accumulation of motors at the ends of microtubules in vitro itself influences the motion of motors walking along the tracks (19). Thus, cargo stalling might arise both through direct physical crowding and also via indirect and more complex effects mediated through motors. The net consequence of such effects is a local traffic jam and accumulation of cargo.

Given that multiple sites along the neuronal process are prone to traffic jams, processes that resolve these traffic jams to maintain cargo flow must exist. We observe that stationary pre-SVs do mobilize resulting in a 3-fold increase in cargo flow through the same location (Figure 5B,C). Cargo emerge from already existing stationary pre-SVs at regular intervals (Figure 7C–E). Thus, stationary vesicular cargo clusters can function as dynamic reservoirs where vesicles can both stall and emerge. The mobilization of such vesicles could depend on the state of motors as well as subtle changes in the surrounding cytoskeletal architecture. Our data suggest that levels of motor on the cargo surface affect both stalling and emergence of cargo from such locations (Figure S7A,B).

We also see that repeated touch stimulation leads to a reduction in the density of stationary pre-SV clusters without affecting the density of actin-rich regions (Figure 7F,G, S7C). These data suggest that stationary pre-SV clusters can act as functional reservoirs of vesicles and that the process of mobilization of stationary vesicles from such locations could be mediated by motors without affecting actin, as suggested by reduced mobilization observed in a motor mutant (Figure S7A,B). Touch stimulation is shown to increase calcium levels in the TRN (50). In mec-4 and mec-10 mutants, there are reduced calcium transients after stimulation (47, 50). Calcium influx is also known to influence both motor behavior and cargo motion in a neuron (60, 61). The observed reduction in density of stationary pre-SV clusters after stimulation could therefore be mediated through calcium. Vesicle capture and release may be selectively modulated from stationary vesicle pools in response to Ca²⁺concentration changes arising from external stimuli or internal stores (Figure 8). The ability of stationary pools of vesicles to function as dynamic vesicle reservoirs may thus provide an additional functional layer of control of cargo flux in response to cellular requirements, with large variations in flux potentially buffered by emergence or halting of vesicles at stationary vesicle clusters.

MATERIALS AND METHODS

Strains

Following strains were used in the study:

jsIs37	pmec-7::SNB-1::GFP (ref 62)	NM664
jsIs821	pmec-7::GFP::RAB-3 (ref 42)	NM2689
jsIs682	prab-3::GFP::RAB-3 (ref 63)	NM2415
wyIs291	punc-86::GFP::utCH; PODR-1::GFP (ref 30, 31)
twnEx195	pmec-7::COR-1::mCherry, pttx-3::mCherry (ref 34)	CLP945
tbIs263	pmec-7::COR-1::mCherry, pttx-3::mCherry	TT753
jsIs1073	pmec-7::tagRFP-MITO unc-119 (ref 64)	NM2057
jsIs1239	[unc-119(+) pmec-7::DENDRA-2::RAB-3 RIM-3, unc-119]	NM2357
oxSi266	unc-119(ed-9); oxSi266[ prund-1::eGFP::RAB-5, unc-119] (ref 65)	EG6193
tbIs227	[ pmec-4::mcherry::RAB-3; punc122::GFP]	TT1884
tbIs147	Pan neuronal UNC-104::GFP integrated in unc-104(e1265) background (ref 42)	TT343
rab-3(js49)	complete loss-of-function allele that has a nonsense mutation in the tryptophan 76 codon. (ref 66)	NM791
unc-104 (e1265tb120)	Suppressor of unc-104(e1265), UNC-104(D1497N M1540I) (ref 42)	TT385
mec-4(u253)	Deletion in exon 3, shown to be null (ref 67)	TU253
mec-10(tm1552)	Deletion in exon 8 and intron, shown to be null (gene knock-out consortium ref 68)	ZB2551

jsIs1239 was created by inserting the _pmec-7::Dendra2-RAB-3 Rim3’sequences from plasmid NM2357 into chromosome II at the mos ttTi5605 insertion site using MosSCI in an unc-119(ed3) mutant background (69). The structure of insertion was confirmed by PCR and restriction digestion of the resulting PCR product. NM2357 was constructed by PCR amplification of dendra2 from pDendra2-C (Evrogen) using oligonucleotides TTTAgctagcgtcgacggtacCATGAACACCCCGGGAATT and TTCATGTACACGCCGCTGTC, digesting the resulting product with NheI and BsrGI, and inserting into similarly digested NM2211 (pCFJ150-pmec- 7::GFP::RAB-3 Rim3′) replacing GFP sequences with RAB-3. NM2211 was created by inserting the pmec-7-GFP::RAB-3 sequences from a BssHII/XbaI fragment from NM1028 (70) and inserting the fragment into BssHII/AvrII-digested CFJ150.

The strain mec-10(tm1552) was confirmed using PCR with the following primer set: FP 5′TGGGAGGGAGCTTCATCTTA3′RP 5′GTAGGGTCTGCAACTAGCTC3′.

Worm maintenance

Worms were maintained on Nematode Growth Medium (NGM) agar plates seeded with OP50 Escherichia coli strain, at a temperature of 20°C (71). L4s or 1-day adults from contamination-free and non-crowded plates were used for imaging.

Drosophila stocks

Drosophila stocks expressing Cha19bGal-4 UAS-GFP were raised on a standard corn meal medium at 25°C. Cha19bGal-4 drives the expression of UAS-RAB-4-mRFP and UAS-Lifeact-GFP specifically in cholinergic neurons of Drosophila (38, 39). For dual-color experiments and LatA treatment, filet prep was done on third instar larvae as described by Parton et al in 2010 (72) and Brent et al in 2009 (73).

Dynamic imaging and analysis

Time-lapse imaging

Live single worms mounted on glass slides with agar pads were anesthetized using 5 mM Tetramisole (Sigma-Aldrich) prepared in M9 buffer. For Latrunculin A injection, different concentrations of Latrunculin A (Sigma-Aldrich) solutions were prepared in 1× Phosphate Buffered Saline (PBS). Latrunculin A or 1× PBS (as control) was injected into the pseudocoelom of L4 worms using a microinjection apparatus. The worms were anaesthetized using 5 mM Tetramisole immediately after injection and imaged. Olympus IX83 microscope with a spinning disc (Perkin Elmer Ultraview) was used for imaging. Time-lapse fluorescence images of specific regions of the neuronal processes of TRNs were acquired at either 5 fps (frames per second) or 3 fps using a 60×, 1.42NA objective, and an Andor (iXon DU897-UVB)/Hamamatsu (SZK) monochrome camera. The typical length of neuron imaged is ~80 μm in C. elegans TRNs and ~50 μm in Drosophila chordotonal neurons.

Analysis

Kymographs were made using the ImageJ MultipleKymograph plugin Plugins were downloaded from the NIH website with the following links; http://www.rsbweb.nih.gov/ij/ and http://www.emblheidelberg.de/eamnet/html/body_kymograph.html. In a kymograph, cargo moving in the retrograde direction (toward the cell body) and anterograde direction (away from the cell body) were identified by sloped lines. Vertical lines represented stationary cargo. A cargo is counted as moving if it has been displaced by at least 3 pixels in successive time frames (72).

Motion property calculations

Pause time is defined as the length of time for which a cargo stays stationary between 2 consecutive runs. Cargo pausing in proximity of another stalled cargo is not included in this analysis. The Measure tool in ImageJ was used to measure pause time and the pixel value was multiplied by a conversion factor of either 0.2 seconds (for 5fps) or 0.3 seconds (for 3 fps) depending on the frame rate of the acquired movie.

Total run length is the sum of segment run lengths of a given vesicle. Segment run length is the distance moved by the cargo between pauses.

Cut-offs for stationary vesicular cargo

To calculate cut-offs for stationary vesicular cargo, pause times were measured from kymographs. Maximum pause time for each vesicular cargo was quantified in a kymograph. Median of maximum pause values acquired from different kymographs was used for calculating cut-off for a specific vesicle type. A vesicular cargo was defined as stationary if it remained immobile for at least 15 times the maximum observed pause time of that cargo (Table S2). To denote different stationary cargo types, we use the term stationary before the specific cargo type, for example, stationary pre-SVs or stationary endosomes. Stationary vesicles or stationary vesicular cargo indicates all stalled vesicles, that is, both stationary pre-SVs and endosomes. Stationary cargo refers to all stalled cargo types including stalled vesicles and stalled mitochondria.

Density calculation

To calculate the density of stationary vesicular cargo in a particular kymograph, several 20 μm stretches in sharp focus along the neuronal process were selected. The numbers of stationary vesicles were counted in each stretch. The density was represented as number of stationary vesicles per 10 μm of the axonal length.

Approximation of number of vesicles at a stationary vesicular cargo

We measure the intensities of vesicles when they pause between 2 consecutive run lengths. For measuring intensities of both paused and stationary vesicles, we draw a line of thickness 2 pixels spanning the stationary vesicles. Measure tool in ImageJ is used for quantifying the average intensities. Multiple such vesicles are sampled in a kymograph. We divide the intensity of a stationary vesicles with the average intensity of paused vesicles to approximate the number of vesicles at a stationary vesicle cluster in the same kymograph.

Photobleaching and photoconversion

Photobleaching experiments were performed with a spinning disc confocal system with an attached photokinesis unit (Perkin Elmer Ultraview). Time-lapse images were acquired by using a 100×, 1.63 Numerical Aperture (NA) oil immersion, objective lens. Anesthetized animals were imaged along the neuronal process of PLM neurons. The Spot function in software was used to randomly photoconvert stationary pre-SVs while a 5-minute movie was being acquired. Photoconversion of Dendra-2 was done using a 405-nm LASER at 60% transmission for 100 ms, 10 iterations of 10 ms each. Kymographs were then used to analyse the events occurring at individual stationary pre-SVs.

Dual-color imaging

Both simultaneous and sequential imaging of red and green markers was done for a total of 2 minutes in a Zeiss LSM 5 live scanning confocal system or Perkin Elmer spinning disc microscope using a 60×, 1.42 NA, oil immersion objective lens. Multispec beads were used to check the alignment of both cameras for dual-color imaging. To check bleed through, laser power in 1 channel (eg, green) was slowly increased and an increase in intensity in the other channel (eg, red) was examined. This confirmation was done for all the fluorophore combinations used. Sequential or simultaneous imaging conditions were then accordingly chosen to collect unbiased data for a given pair of fluorophores depending on the extent of bleed through. Time-lapse images were acquired at a maximum frame rate of 1.5 or 3 fps. Alternate frames were separated using Image-J> > Stacks> > Tools> > Substack command. Kymographs were plotted for separated frames and overlaid for analysis.

Statistical analysis

Graph pad Prism 5.0 was used to perform the Student’s t-test or 1-way ANOVA to compare between 2 groups or multiple groups, respectively. Kruskal-Wallis and Dunn’s comparison post-tests were applied for Student’s t-test and 1-way ANOVA, respectively, since not all the groups compared passed the normality tests. Two independent normality tests were used: D’Agostino and Pearson omnibus normality test and Kolmogorov-Smirnov test (with Dallal-Wilkinson-Lillie for P-value). In case of comparison of flux with and without a stationary cargo at the same site, paired t-test was used. To test the statistical significance of data like stop and go or juxtaposition, represented as grouped plots in Figures 4 and S6, percentage values from individual animals were compiled and tested for significance using the same statistical tests listed above. However for ease of representation, the errors bars are not included in the graphs and the data are represented as a sum of values from all individual animals. We have mentioned the P-values in these cases in the legend. Pearson’s correlation analysis was used to determine correlation between the density of stationary pre-SVs and run length of moving cargo in a region. Correlation values were plotted on the graph and were fit using linear regression. P-values of <.001 is designated by ***, <.01 is designated by ** and <.05 is designated by *. All data are represented as mean ± SEM in the graphs unless stated otherwise.

Statistical analysis of marker co-localization

We analyse about 7 snapshots obtained from multiple kymographs from kymograph trajectories separated by at least 15 seconds in time to ensure that they are approximately independent and across multiple worms (n = 7–10). These yield spatial signals for the 2 fluorescent markers whose co-localization we wish to assess. We compute the Pearson’s Correlation Coefficient, the Mander’s Overlap Coefficient and the Mander’s Correlation Coefficient, all measures of the correlations between the 2 markers. We use the MOC and the Mander’s correlation coefficients (MCC) to estimate marker co-localization. The MOC and the MCC are the most appropriate for our experimental situation, where we expect a strong overlap between signals from 2 markers only at isolated points along the axon, identified with actin accumulations, but very little overlap or correlation elsewhere, since at least one of these markers is associated with mobile cargo. We note that the MCC’s directly measure co-localization, since they compute the fraction of total probe fluorescence that co-localizes with the fluorescence of a second probe, irrespective of any linear relationship between their measured intensities. We also develop an approach based on the histogram of values of the scaled correlation function as a function of pixel location. For marker locations which are uncorrelated, the histogram should exhibit values equally distributed about zero, with vanishing skewness. If, on the other hand, these markers are correlated positively, we expect that the histogram will skew toward positive values. Thus an increased co-localization of the 2 markers is associated with a positive value of the mean and skewness of the histogram of correlation function values, which we examine using 1-tailed t-tests of significance, computing the t-statistic and the Zg1 statistic from the data.

Abbreviations:

SC

Stationary cargo

C.e

Caenorhabditis elegans

D. m

Drosophila melanogaster

pre-SVs

precursors of synaptic vesicles