Post-Golgi vesicles undergo kiss-and-run exocytosis

To monitor secretion of cargo from these post-Golgi vesicles, cells expressing luminal marker (hGHss-GFP-CAD-hGH) were imaged by TIRFM. During transit through Golgi, this protein is cleaved by the Golgi specific protease furin (Fig. 1B). Thus, the fluorescent vesicle luminal marker is hGH signal sequence fused to GFP-CAD (hGHss-GFP; the hGH itself remains non-fluorescent) (Rivera et al., 2000). The fluorescence of these vesicles increase as they entered the evanescent field, and then drops rapidly to background level as it spreads laterally, indicating cargo exocytosis (Fig 2A). We also observed exocytic events where vesicle fluorescence decreased, but remained above the background (Fig 2B). This observation is consistent with the vesicle transiently opening its fusion pore to the outside, releasing only a part of its cargo. Notwithstanding the fact that most of these vesicles do not move following fusion, the term “kiss-and-run fusion” is used to describe such partial fusion fates (Harata et al., 2006). We will thus refer to these fusion fates as kiss-and-run fusion. To test if these partially fusing vesicles open a transient fusion pore we used FM 4–64 dye, which can access the vesicle membrane only when it contacts the extracellular media and has been used to test kiss-and-run fusion of synaptic and other vesicles (Harata et al., 2006). Indeed, FM 4–64 labeled many GFP containing post-Golgi vesicles (Fig. 2C). Similarly, vesicles got labeled with the fluid phase marker Alexa fluor 546-hydrazine (not shown).

Open in a separate window

Figure 2

Imaging individual post-Golgi vesicle exocytosis

Exocytic vesicles are monitored using (A–C) luminal marker hGHss-GFP and (D, E) membrane marker NGFR-GFP. (A, D) Complete release and (B, E) partial (kiss-and-run) release of the cargoes. The dotted circle marks pixels whose intensity was unchanged as the vesicle docked. The plots show fluorescence within the dotted circle (open symbols) and of the entire field (filled symbols). Intensity is normalized after subtracting background prior to the appearance of the vesicle. Based on the measurement of vesicle fluorescence, vesicle was considered to be tethered/docked at the point after appearing in the evanescent field when its fluorescence was relatively stable. Priming marks the period between docking and initiation of fusion, which is indicated by rapid increase in vesicle fluorescence followed by the cargo release. As total fluorescence of the membrane cargo in D and E remain unchanged after release of the cargo from the vesicles, it indicates that the proteins from the vesicle membrane are delivered to the cell membrane and not lost in the cytosol (as would occur if the vesicle had lysed). (C) A cell expressing hGHss-GFP (green) was treated for 20 minutes with AP21988, followed by 10 minutes in FM4–64 (red). Image is of a part of this cell showing many vesicles labeled with both these markers.

The ability of extracellular markers to enter the lumen of the vesicle indicates that partial release of vesicle luminal marker is due to a transient opening to the outside. However, it does not rule out the possibility that this may involve photodamage to the cell and vesicle membrane causing leakage and mixing of luminal content. To resolve between fusion, leakage and lysis we monitored the fate of vesicle membrane proteins during exocytosis. Only when a vesicle exocytoses will its membrane protein be delivered to the cell surface where it diffuses in two dimensions, thus any decrease in vesicle fluorescence will cause a concomitant increase in cell membrane fluorescence resulting in no change in integrated fluorescence. In case of leakage the membrane marker would remain with the vesicle and not diffuse in the cell membrane, while in case of lysis it will diffuse in the cytoplasm in three dimensions (Jaiswal and Simon, 2007; Jaiswal et al., 2007; Schmoranzer et al., 2000). Following ligand addition, CFP tagged CAD-NGFR trafficked to the post-Golgi vesicles together with YFP tagged CAD-hGH (Fig. S1C, D). Based on the fluorescence of the vesicle membrane marker we observed two types of exocytic behaviors. In some cases the fluorescence of the vesicular membrane protein reduced to the pre-docking background (Fig. 2D red symbol), indicating complete release. In other cases the fluorescence of the vesicular membrane protein decreased, but remained above the pre-docking background (Fig. 2E red symbol), indicating partial release. In each of the observed fates of the vesicle (Fig. 2D, E), the integrated (sum of vesicular and cell membrane) fluorescence of the fusing vesicle’s membrane protein remained high (green symbols) even when there was a decrease in the luminal fluorescence (red symbols). This confirms that these events involve merger of the vesicle and cell membrane and thus indicate exocytosis, not leakage or lysis.

To quantify the number of post-Golgi vesicles undergoing partial or complete fusion, we monitored the release of the luminal cargo or membrane cargo following the addition of CAD disaggregating ligand. Based on luminal contents (hGHss-GFP), we observed 32.7±8.0 fusions/cell in a minute of which 73% (23.8±7.1) were kiss-and-run fusion and the rest 8.9±2.9 were complete fusions (Fig. 3A gray bars). However, using membrane cargo (LNGFR-GFP) we observed 12.2±2.2 fusions/cell in a minute of which 17% (2.1±0.8) were kiss-and-run fusions and the rest 10.1±2 were complete fusions (Fig. 3A black bars). While the difference in number of complete fusions reported by luminal and membrane markers is not statistically significant (p value = 0.4), there is 11-fold difference in the number of vesicles that partially released their content (p value = 10⁻⁶).

Open in a separate window

Figure 3

Post-Golgi vesicles can exocytose without releasing membrane cargo

(A) Vesicles undergoing partial (kiss-and-run) or complete exocytosis were measured for one minute in cells expressing only luminal (n = 378 vesicles from 9 cells) or only membrane (n = 135 vesicles from 6 cells) markers. The error bars represent standard deviation. (B) The vesicle labeled with both, luminal (hGH-YFP) as well as membrane (NGFR-CFP) marker (supplementary figure 2A) that undergoes complete fusion. Open symbol shows vesicle-associated fluorescence and filled symbol shows fluorescence of the entire field of view. (C) A dual labeled vesicle that exocytoses to partially release the luminal cargo (red symbol), but does not deliver any membrane cargo to the cell surface (green symbol). Vesicle is shown in figure S2B. (D) Total fluorescence of luminal and membrane cargo for a vesicle that undergoes multiple partial fusions prior to complete fusion. This vesicle is shown in figure S2C and a schematic for the various fusion stages are marked by numbers as described in figure E. (E) Schematic showing states of the vesicle imaged in D – 1&2: docking and priming. 3: first fusion with the cell membrane resulting in partial release of luminal marker (rapid increase and decrease in YFP fluorescence) and no release of the membrane marker (increase and no subsequent decrease in the vesicle associated CFP fluorescence). 4: A period where none of the two markers were released from the vesicle. 5: Fusion where the luminal and membrane markers are released partially albeit to different extents. 6: A period when both the markers are retained by the vesicle. 7: Complete release of luminal and membrane cargoes resulting in return of their vesicle associated fluorescence to pre-docking background. Membrane cargo is marked in green and luminal cargo in red, the shaded yellow region depicts the evanescent field. (F) Release time for hGHss-GFP was quantified for vesicles that underwent partial or complete fusion in five cells (n = 169 vesicles). Release time denotes the period during which vesicle fluorescence increased and reached a post-fusion plateau (see Fig. 2).

Above results are consistent with a population of vesicles that undergo kiss-and-run fusion resulting in partial release of their luminal cargo but no release of the membrane cargo. To test this prediction we co-expressed the membrane and luminal markers as CFP and YFP tagged proteins. Both these cargoes were sequestered in the ER and AP21988 addition caused both cargoes to package in the same post-Golgi vesicles (Fig. S1C, D; Fig. S2 A–C).

Quantification of the vesicle and cell membrane-associated fluorescence indicates vesicles that fully released their YFP-tagged luminal cargo (Fig 3B, red line) also fully released their CFP-tagged membrane marker cargo (Fig. 3B open blue circles) and the released membrane cargo is delivered to the cell membrane: The integrated fluorescence of the vesicle and cell membrane remained relatively constant after exocytosis (Fig. 3B green line, filled symbols). Thus vesicle luminal and membrane cargoes behaved similarly when co-expressed or expressed independently – their fluorescence returned to pre-fusion level (Figs. 2A and 2D). Vesicles partially releasing their luminal content delivered either little or no membrane protein (Fig. 3C–E, Fig. S2B–C). The fusion events where luminal cargo is partially released and membrane cargo is not released (state 3 in Fig. 3E) account for the reduced number of partial fusions when only the vesicle membrane protein was used as exocytic marker (Fig. 3A).

Vesicles that fuse partially without releasing membrane cargo can subsequently fuse with partial or complete release of the membrane cargo (Fig. 3D). We confirmed that the fluorescence signals we observed originated from an individual fusing vesicle (Fig. S2 D–E). Above results indicate that the nature of vesicle fusion is regulated dynamically. To examine if restriction of fusion pore (during kiss-and-run fusion) slows release of luminal content, we compared the time between start and end of luminal cargo release by individual exocytic vesicles from five cells. Cargo released faster during complete fusions (1.11 ± 0.30 s) compared to kiss-and-run fusions (1.49 ± 0.36 s) (p value = 0.01; Fig. 3F).

It has been previously reported that dense core vesicles (DCVs) can partially release their luminal contents together with release of some but not other membrane proteins (Tsuboi et al., 2004). Thus, incomplete release of membrane proteins could reflect a limitation of the reporter protein used and not the nature of the fusion. However, our vesicle membrane marker (NGFR-GFP) can be retained or fully released by the same vesicle (Fig. 3D) and our observation that ~ 17% (2/12; Fig. 3A) post-Golgi vesicles labeled by this marker undergo kiss-and-run fusion is in agreement with the results of 14% previously reported using LDLR-GFP (Schmoranzer and Simon, 2003). Together, these results suggest that the incomplete release we observe is not due to a limitation of our reporter. Instead, it is consistent with post-Golgi vesicles partially releasing their contents similar to the kiss-and-run fusion proposed for synaptic, dense core and other regulated exocytic vesicles (Harata et al., 2006).

Ca²⁺ does not trigger nor regulate the nature of post-Golgi vesicle exocytosis

Unlike regulated exocytic vesicles, exocytosis of post-Golgi vesicles is unaffected when extracellular Ca²⁺ is chelated using EGTA (Lew and Simon, 1991). This indicates that Ca²⁺ influx is not required for post-Golgi vesicle exocytosis, but does not rule out that transient localized Ca²⁺ microdomains formed by the efflux from the docked post-Golgi vesicle may trigger its exocytosis. Such a mode of Ca²⁺-triggering has been identified for other membrane fusion reactions (Peters and Mayer, 1998). To test this possibility we used two reagents - ionomycin to rapidly discharge vesicular Ca²⁺ (Williams et al., 1985) and BAPTA-AM to rapidly chelate cytosolic Ca²⁺ and prevent the formation of Ca²⁺ microdomains (Peters and Mayer, 1998). Increasing Ca²⁺ by ionomycin triggered lysosomal exocytosis in HT1080 cells as we have previously observed for other cells (Jaiswal et al., 2002). BAPTA-AM prevents Ca²⁺-microdomain formation which triggers yeast vacuole fusion (Peters and Mayer, 1998). Using the Ca²⁺-sensitive ratiometric dye pair fura-red and fluo-3 we found that both these agents stably altered the cytosolic Ca²⁺ within 10 sec (Fig. 4A). Calcium is required for trafficking of biosynthetic cargo from ER to Golgi (Beckers and Balch, 1989) and this treatment reduced the number of fluorescently-labeled post-Golgi vesicles that newly appeared in the TIR field (video 2). Thus, assaying total cargo released from cells with altered cytosolic calcium cannot distinguish between effects of Ca²⁺ on vesicle exocytosis and effects on transport through the biosynthetic pathway.

Open in a separate window

Figure 4

Effect of calcium on trafficking and exocytosis of post-Golgi cargo

(A) Six HT1080 cells labeled with Fura-red and Fluo-3 and were imaged using epifluorescence excitation starting 30s prior to the addition of calcium altering agents and their averaged fluorecen is plotted. (B) Cells expressing luminal cargo hGHss-GFP were imaged by TIR-FM and the time for which the exocytic vesicles remained tethered at the cell surface prior to fusion was measured for 107 vesicles. (C) Rate of exocytosis of 450 hGHss-GFP vesicles from 10 cells was quantified, those present in the TIR field at the start of imaging were termed as tethered/docked, rest were classified as recruited. (D) Exocytic fate of the tethered and newly recruited vesicles. (E) Tethered post-Golgi vesicles that underwent exocytosis were quantified in nine cells each that were treated for 20–40 minutes with AP21988 only (untreated) or with AP21988 followed by 100μM BAPTA-AM (for 5 min) or 10μM ionomycin.

To resolve if calcium affected steps after transport, we took advantage of the observation that most post-Golgi vesicles pause prior to fusion, as if briefly tethered to the cell membrane (step 2 in Fig. 3E; Fig. 4B n = 107) (Schmoranzer et al., 2000). Exocytosis of these cargo carrying post-Golgi vesicles is independent of effects at the step of cargo packaging. Thus, we monitored the exocytosis of membrane tethered post-Golgi vesicles to test for effects of calcium release from their lumen on exocytosis. In cells where calcium has not been altered, over one-third of all vesicles that exocytose during a two minute imaging period were tethered at the cell surface at the start of imaging (Fig. 4C). Vesicles that were tethered or freshly recruited were equally likely to partially (or completely) release their contents (Fig. 4D). We next quantified the effect of transient treatment of ionomycin or BAPTA-AM on exocytosis of tethered vesicles (Fig. 4E). Neither chelating the cytosolic Ca²⁺ (with BAPTA-AM) nor increasing it (with ionomycin) affected the rate of exocytosis of post-Golgi vesicles (Fig. 4E p value = >0.4 for both treatments). To test if the acidic lumen of secretory vesicles was hindering the ionomycin-induced release of calcium, we treated cells with ionomycin in the presence of 25 mM NH₄Cl or 100 μM of the Na⁺/H⁺ exchanger monensin. Neither agent alone or together with ionomycin had any affect on the rates of post-Golgi vesicle fusion (data not shown). These results indicated that post-Golgi vesicle exocytosis is insensitive to calcium influx and insensitive to any local Ca²⁺ efflux from the secreting vesicle. To test if Ca²⁺ affects the nature of post-Golgi vesicle fusion we monitored post-Golgi vesicle exocytosis in BAPTA and ionomycin treated cells. We detected no changes in the rate of partial or complete release of cargo by either treatment (Fig. 4E; p values between 0.2 – 0.8). This ruled out a role of Ca²⁺ or Ca²⁺-dependent exocytic machinery in regulating the choice between kiss-and-run and complete fusion of post-Golgi vesicles.

Thomas Kirschhausen provided us dynasore and DsRed-CLC and Claire Atkinson generated Dynamin2(aa)-mCherry. We thank lab members and others who offered helpful suggestions during the course of this work and Patrick Bhola for comments on the manuscript. Knut Wittkowski helped us with some of the statistical analysis and the work was funded by grants from NSF (BES-0620813) and NIH (P20 GM072015 to SMS and R01AR055686 to JKJ).