Exocyst dynamics during vesicle tethering and fusion

Abstract

The exocyst is a conserved octameric complex that tethers exocytic vesicles to the plasma membrane prior to fusion. Exocyst assembly and delivery mechanisms remain unclear, especially in mammalian cells. Here we tagged multiple endogenous exocyst subunits with sfGFP or Halo using Cas9 gene-editing, to create single and double knock-in lines of mammary epithelial cells, and interrogated exocyst dynamics by high-speed imaging and correlation spectroscopy. We discovered that mammalian exocyst is comprised of tetrameric subcomplexes that can associate independently with vesicles and plasma membrane and are in dynamic equilibrium with octamer and monomers. Membrane arrival times are similar for subunits and vesicles, but with a small delay (~80msec) between subcomplexes. Departure of SEC3 occurs prior to fusion, whereas other subunits depart just after fusion. About 9 exocyst complexes are associated per vesicle. These data reveal the mammalian exocyst as a remarkably dynamic two-part complex and provide important insights into assembly/disassembly mechanisms.

Fig. 1

Establishment and validation of tagged exocyst subunit cell lines. a Schematic showing a region of subunit gene targeted by sgRNA and tags to be inserted C-terminally, in-frame with the coding region. b Strategy to isolate CRISPR/Cas9-mediated sfGFP-tagged clones of exocyst subunits in NMuMG cells. c Western blots with subunit specific antibodies show successful incorporation of sfGFP in both alleles (EXO70, SEC5, and SEC8), or in one allele (SEC3 and SEC6). A shRNA specific to SEC3 was used to confirm the identity of multiple bands in the blot. d Confocal images of live wild-type NMuMG cells and endogenously tagged SEC3-GFP, SEC5-GFP, SEC6-GFP, SEC8-GFP, and EXO70-GFP cell lines. Scale bar = 10 μm. e Protein–protein interaction heat map for endogenous SEC3-GFP, SEC5-GFP, and EXO70-GFP pulldown using GFP-Trap nanobodies and MS. Yellow indicates baits, blue the protein identified with high confidence, and gray denotes preys that were not identified. The experiments were repeated three times for EXO70-GFP and twice for SEC3-GFP and SEC5-GFP. All experiments were also partially confirmed by IP–WB in at least three independent experiments, with similar results. f Western blot analysis to assess the relative abundance of exocyst subunits fused to sfGFP, using anti-GFP antibodies (SEC3 and SEC6 quantifications corrected for heterozygosity). GAPDH was the loading control. Y axis shows the ratio of GFP/GAPDH. Quantification shows pooled data from three experiments

Fig. 2

Mammalian exocyst complex comprises two subcomplexes that can localize to the PM independently. a Quantitative full-scan LC–MS/MS analysis of endogenous EXO70-GFP pulldowns from intact or Sec8-depleted cells. Inputs for the samples were calibrated and normalized using MRM–MS analysis on the same samples. The schematic shows the experimental design in which EXO70-GFP was captured using GFP-Trap beads and SEC8 was depleted by shRNA. b Heatmap summarizing relative binding of SEC6, SEC8, EXO70, and EXO84 to SEC3-GFP in SEC10-depleted cells compared to control shRNA-treated cells, as assessed by western blot. Also see Figure S2H-I. c TIRFM images of EXO70-GFP and SEC3-GFP in control or SEC10 shRNA-treated cells. Scale bars 20 µm. d Quantification of relative fluorescence intensities in C. e Quantification of fluorescence intensities from TIRFM images of SEC5-GFP cells treated with control or SEC10 shRNA. f TIRFM images of double knock-in NMuMG cells, showing localizations of EXO70-GFP and SEC5-Sc at the PM in control, SEC8 shRNA, and SEC3 shRNA-treated cells. Scale bars = 20 µm. g Quantification of relative fluorescence intensities in D. Center lines show the median; box limits indicate 25th and 75th percentiles; whiskers extend 1.5X the IQR from the 25th and 75th percentiles. Experiments were repeated at least three times with similar results

Fig. 3

Sec3 departs from the exocyst complex prior to vesicle fusion. a Schematic representation of the data. b NMuMG cells were transduced with mApple-Rab11 and VAMP2-pHluorin lentivirus and imaged by TIRFM. Kymographs show arrival and departure of mApple-Rab11 and vesicle fusion, highlighted by VAMP2-pHluorin flashes. Speed = 5 Hz. Scale bar = 5 s. c Distribution of mApple-Rab11 arrival times prior to vesicle fusions. The inset shows box and whisker plot; center line = −14.6 s (median; 95% CI: −16.6 to −12.5). Error bars = Tukey’s range. d Intensity trace over time for VAMP2-pHluorin and TfR-pHuji. The peak intensities denote flashes resulting from vesicle fusion. Speed = 5 Hz. e Effects of exocyst subunits depletion on vesicle fusion activities in NMuMG cells assessed using VAMP2-pHluorin. n = 32, 32, 18, 22, 21, 22. P-values compared to control. f Kymographs showing an itinerary of SEC3-GFP, SEC5-GFP, SEC6-GFP, and EXO70-GFP arrivals/departures with respect to the vesicle fusion marker TfR-pHuji. Square images show snapshots in the green and red channels at the time of vesicle fusions. Scale bar = 5 s. g, h Quantification of exocyst subunit arrival and departure times at and from the vesicle fusion site shown in E. Data points shown as dots. g Median arrival times in seconds: −11.7 (95% CI: −12.9 to −9.1; EXO70), −11.6 (95% CI: −14.6 to −10.2; SEC3), −10.6 (95% CI: −14.0 to −8.2; SEC5), and −10.3 (95% CI: −12.3 to −8.8; SEC6). n = 146, 132, 108, and 132 objects in the order data shown. h Median departure times in seconds: 1.4 (95% CI: 1.2–1.5; EXO70), −0.4 (95% CI: −0.98 to −0.69; SEC3), 1.3 (95% CI: 1.21.6; SEC5), and 1.3 (95% CI: 1.21.7; SEC6). n = 145, 117, 108, and 132 particles from 33, 19, 29, and 26 cells in the order data are shown. Center lines = medians; hinges extend from 25th to 75th percentiles. Statistical significance measured by Kruskal–Wallis test followed by Dunn’s post hoc analysis. Experiments were repeated three times with similar results

Fig. 4

Coincidence measurements between exocyst subunit pairs by TIRF in live cells. a Fluorescence intensity trajectories (intensity values were normalized to a common scale) for the SEC5-Halo + JF585/SEC8-GFP pair, and SEC5-Halo + JF585/EXO70-GFP pair over 20 s. The fluorophores were excited simultaneously and images were captured at 12.5 Hz (left). (Right) Graphs show the displacement of the normalized intensities of the two fluorophores at each time point. b Quantification of the delay in coincidence between EXO70-GFP and SEC5-Halo + JF585 or SEC8-GFP and SEC5-Halo, measured by TIRFM. Outliers are represented by black circles; data points are plotted as magenta circles. The n denotes number of particles analyzed from a total of three experiments. c Distribution of the order of arrival of EXO70-GFP and SEC5-HaloJF585 shown in panels a and b. d Distributions of the duration of the indicated subunits were together in the TIRF field. e Comparison on the distribution of residence times of Rab11 and the indicated pairs of exocyst subunits. White circles = medians; box limits indicate the 25th−75th percentiles; whiskers extend 1.5X the interquartile ranges from 25th to 75th percentiles; polygons represent density estimates of data and extend to extreme values. Statistical significance measured by Kruskal–Wallis test followed by Dunn’s post hoc analysis

Fig. 5

Assessment of fractional binding using single-molecule approach. a Diagram showing experimental workflow. b Cleared cell lysates from the indicated double knock-in cells were spread between quartz slides and glass coverslips for TIRF imaging. The GFP and Halo + JF585 fluorophores were excited simultaneously with 488-nm and 561-nm lasers at 12.5 Hz for 40 s. Scale bar = 1 µm. c Number of photobleaching steps detected for SEC5-Halo and SEC8-GFP in experiments shown in panel b. d Quantification of the fraction of subunit A bound to (–> ) subunit B from experiments shown in panel b. Numbers of particles analyzed for each pair are shown in Supplementary Fig. 4c. e Fraction of SEC5-Halo and SEC8-GFP bound to each other in SEC3 or SEC6 shRNA-treated cells compared to control. Center line denotes mean. f Western blot analysis of SEC3 and SEC6 shRNA knockdown efficiencies in the experiment shown in e. Error bars denote  ± SD. Statistical significance was computed by one-way (d) or two-way (e) ANOVA, followed by Tukey’s multiple comparison tests. Experiments were repeated three times and data were pooled

Fig. 6

Dual-color FCCS of exocyst subunits in the membrane and cytosol. a Schematic of dual-color cross-correlation spectroscopy (FCCS). Top diagram shows cross-correlation between co-diffusing molecules. Bottom shows molecules that do not co-localize. b SEC5-Halo and EXO70-GFP FCS measurements in the cytosol and the PM. EXO70-GFP and Akt-Halo fluorescence cross-correlation was used as a negative control. HaloTag was labeled using JF646 Halo ligand (200 nM for 1.5 h). c SEC5-Halo and SEC8-GFP, or SEC8-GFP and Akt-Halo FCS measurements in the cytosol or the PM. d SEC5-Halo and SEC3-GFP, or SEC3-GFP and Akt-Halo FCS measurements in the cytosol or the PM. e Statistics of the fraction of GFP-tagged exocyst subunit bound in the cytosol. SEC8-GFP+SEC5-Halo: 92 ± 3.86%, SEC3-GFP + SEC5-Halo: 67 ± 3.84%, and EXO70-GFP+SEC5-Halo: 64 ± 3.46% (mean ± SEM). f Fractional coincidence between SEC5-Halo and SEC8-GFP or SEC5-Halo and EXO70-GFP was analyzed using TIRFM. Calculations were corrected for HaloTag labeling efficiency and expression levels of SEC8-GFP and EXO70-GFP with respect to SEC5-GFP. Center line denotes median. SEC5-Halo bound to SEC8-GFP was 85 ± 3.3% (mean ± SEM). SEC8-GFP bound to SEC5-Halo was 85 ± 5.1%. SEC5-Halo bound to EXO70-GFP was 60 ± 3.2% and EXO70-GFP bound to SEC5-Halo was 69.8 ± 2.9%. Object detection parameters were set between 0.30 and 0.35 µm and contrast was adjusted to faithfully detect objects of interest. White circles show the medians; box limits indicate the 25th and 75th percentiles as determined by R software; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles; polygons represent density estimates of data and extend to extreme values. p values were calculated using Kruskal–Wallis test followed by Dunn’s multiple comparison test. The hydrodynamic radius (R_H) for the slow diffusing fraction was median 466 nm (95% CI: 355.2–606.8), and that of the fast diffusing component, 14.82 nm (95% CI: 13.22–17.27). FCS measurements were fitted (lines) to the model described in Methods (Eqs 2–4). Curves shown are representative, taken from a single cell. Statistics are average from > 25 measurements for each condition from two experiments and > 5 cells. p-values were calculated by one-way ANOVA followed by Tukey’s post hoc test

Fig. 7

Measurements of diffusivity of exocyst subunits. a TIRFM images of untagged wild-type or indicated exocyst subunits fused to sfGFP (top). Heat map of above images (bottom). Scale bar = 20 μm. b Quantification of exocyst subunits localization at the TIRF field. Membrane localization index = density of spots * intensity of cells. Whiskers extend from min to max. Values for SEC3 and SEC6 were corrected for heterozygosity. n = 31, 13, 13, 13, 32 fields containing 2–4 cells each. c Particle tracking over time for the subunits indicated. For two-color tracking, each channel was tracked using Imaris software tracking algorithm and overlaid. Graphs show mean squared displacements over time. Scale bar = 0.5 μm (SEC3-GFP) and 0.8 μm (rest). d Representation of data in b as geometric means of MSDs < r² >. e Diffusion coefficients of the indicated subunits were measured from < r² > . Mean (µm² s^–1) ± SEM = 0.31 ± 0.05 (SEC3-GFP), 0.04 ± 0.002 (SEC5-Halo), 0.13 ± 0.01 (SEC6-GFP), 0.04 ± 0.004 (SEC8-GFP), and 0.05 ± 0.004 (EXO70-GFP). The effect size in terms of Cohen’s d value between SEC3-GFP and SEC6-GFP is 0.348 and between SEC6-GFP and SEC8-GFP is 0.176. White circles show the medians; box limits indicate the 25th and 75th percentiles; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles; polygons represent density estimates of data and extend to extreme values. Experiments were repeated at least three times with similar results. P-values were computed using one-way ANOVA test followed by Scheffe’s multiple comparison tests

Fig. 8

Exocyst molecule counting at vesicle fusion sites. a Stepwise photobleaching of CD86 fused to one or two sfGFP molecules used to determine intensities of known numbers of GFP molecules. b Standard curve for counting protein molecules, from intensities measured on CD86 fused to one, two, or three sfGFP molecules. Error bars = ±s.d.; pink shaded region denotes 95% confidence band. c Fluorescence landscape of SEC8-GFP shows a typical resolvable distribution of molecules within an ROI from which intensity measurements were taken. d Intensity (y) was measured from the peak and converted to the number of molecules using the regression equation determined in b. Numbers for SEC3 and SEC6 were corrected for heterozygosity. Mean ± SD = 7.6 ± 3.6 (SEC3-GFP), 9.8 ± 3.5 (SEC5-GFP), 9.8 ± 3.4 (SEC6-GFP), 9.35 ± 3.2 (SEC8-GFP), and 9.19 ± 2.8 (EXO70-GFP). Cohen’s d value for the difference between SEC3-GFP and SEC5-GFP is –0.208. Centers indicate median, box limits indicate 25th and 75th percentiles, and whiskers extend 1.5x IQR from 25th to 75th percentiles. n = 96, 50, 51, 50, 58. Coefficient of variation (CV) = 47.9, 35.2, 39.2%, 34.3, and 30.7% in the order indicated in the graph. CV between each subunit was 7.9%. Experiments were repeated at least three times with similar results. P-values were computed using one-way ANOVA test followed by Scheffe’s post hoc test

Fig. 9

Model for exocyst subunit assembly/disassembly and vesicle tethering. a Schematic of subunit interactions within the cell. Subunits are shown as circles, SC1 and SC2 mark the two subcomplexes, and arrow weights represent relative on or off rates. SNP23 associates with EXO70 but is not co-precipitated with SEC5. Detailed molecular interactions with membranes are not illustrated and the diagram is not to scale. Percentages are the estimated proportions of each state, derived from corrected TIRFM and FCCS data. Octameric complex abundance = (SEC5 + EXO70) colocalization or cross-correlation; tetrameric subcomplex abundance = (SEC5 + SEC8)–(SEC5 + EXO70); and free subunit abundance = total–octamer–tetramer. Free SEC3 is calculated from total–(SEC3 + SEC5). We do not know what fraction of octamer is actually a heptamer lacking bound SEC3. FCCS data suggest that about 50% of exocyst is very slowly diffusing within the cell, likely associated with vesicles, estimated from calculating the hydrodynamic radius of the slow diffusing species. We propose that the individual subcomplexes can interact weakly with membranes but bind membrane more efficiently when assembled into an octamer. SEC3 interacts with SC1 less robustly than do the other subunits. b Individual subcomplexes can associate with vesicles at the PM but cannot trigger fusion. Percentages show estimated relative abundances at the membrane, based on TIRFM data. c Interaction of the subcomplexes to form octamer stabilizes tethering, and release of SEC3 precedes fusion. d Fusion is accompanied by rapid release of the exocyst