Syn1A Binds with High Affinity to the Fusogenic Lipid Phosphatidic Acid
We initially set out to determine whether Syn1A specifically interacts with lipids that might decrease the energetic requirements for membrane fusion or exert other important functions in exocytosis. In these experiments, bacterially expressed soluble Syn1A (aa 1-267) was overlaid in decreasing concentrations onto PIP strips. A demonstrates that Syn1A bound to multiple acidic phospholipids in a dose-dependent manner. Although Syn1A bound with highest apparent affinity to the fusogenic lipid, PA, interactions with several PIPs, including PI(3)P, PI(4)P, PI(5)P, PI(3,4)P2, PI(4,5)P2, and PI(3,4,5)P3, were also observed (B). Notably, all of the lipids to which Syn1A bound were acidic, although Syn1A did not interact with every acidic phospholipid tested (e.g., phosphatidylserine [PS] or LPA). Furthermore, Syn1A exhibited a greater apparent affinity for the monophosphate lipid PA than the polyphosphate inositol lipids, which contain a greater negative valence compared with PA. These data suggest that although electrostatic interactions have an important role in mediating Syn1A's lipid interactions, other structural features may ultimately underlie the specificity.
Figure 1. Syn1A directly binds a subset of acidic phospholipids that includes the fusogenic lipid phosphatidic acid. (A) Protein–lipid overlay demonstrating concentration-dependent binding between soluble Syn1A(1-267) and multiple acidic phospholipids. (more ...)
Although the PIP strip assay was used as an initial screen for Syn1A's lipid-binding properties, it should be noted that this assay can be prone to false positives or false negatives, and thus may not accurately report on the specificity of protein–lipid interactions (Downes et al., 2005
). Thus, a correlative and more physiologically relevant series of liposome flotation binding experiments was also performed, in which Syn1A was mixed with liposomes of defined composition and loaded under a sucrose density gradient. After ultracentrifugation, Syn1A bound to liposomes floats to the top of the gradient with the liposome fraction. C shows Syn1A immunoreactivity in the collected gradient fractions of a representative experiment and demonstrates that Syn1A specifically bound to liposomes containing PA but not to those containing LPA. Furthermore, Syn1A did not bind to control liposomes (containing only PC and PE). Together, the data demonstrate that Syn1A specifically binds the fusogenic lipid, PA.
The Polybasic Juxtamembrane Region in Syn1A Comprises a Lipid Interaction Domain
Sequence analysis of Syn1A revealed a polybasic juxtamembrane region within Syn1A that is highly conserved across multiple species (A, top). To determine if this region is responsible for Syn1A's interactions with acidic phospholipids, a series of progressive neutralizing mutations was generated within this region, in which one (R262A), two (R262A/R263A), or all five (5RK/A) basic residues were neutralized to alanines (A, bottom). Mutant proteins were purified and tested for lipid-binding capacity using nitrocellulose blots on which PA had been spotted in increasing amounts. Binding curves were fit using the Hill equation, which allowed determination of the apparent binding affinity (EC50) of each protein for PA (B). Progressive neutralizing mutations resulted in a progressive reduction in apparent binding affinity to PA, with the 5RK/A mutant demonstrating the greatest reduction in apparent binding affinity (EC50 ~ 3.2 × 105 fmol), followed by the R262A/R263A mutant (EC50 ~ 6.1 × 104 fmol), and lastly, the R262A mutant (EC50 ~ 2.9 × 103 fmol), which demonstrated binding similar to the WT protein (EC50 ~ 2.0 × 103 fmol). Thus, complete neutralization of the polybasic juxtamembrane region in Syn1A resulted in a greater than 2 log shift in EC50.
Figure 2. A polybasic juxtamembrane sequence in Syn1A comprises the lipid-binding domain. (A) Top, sequence alignment of juxtamembrane regions of Syn1A across species. Note the five highly conserved basic residues between positions 260-265 (NCBI accession numbers: (more ...)
Qualitatively similar results were obtained using liposome flotation assays, in which the relative affinities of the WT and 5RK/A Syn1A proteins for PA-containing liposomes was tested. C shows that 24.3 ± 3.4% (n = 5) of the total WT protein, versus 6.1 ± 2.0% (n = 3) of the total 5RK/A mutant protein, bound to the liposomes (p < 0.01). Moreover, the 5RK/A mutation eliminated binding to all acidic lipids as demonstrated by the protein–lipid overlays shown in D. Thus, Syn1A's juxtamembrane basic residues are critical in mediating Syn1A's interactions with acidic phospholipids.
Additional Structural Determinants Underlie Syn1A's Lipid-binding Properties
To determine whether structure within the juxtamembrane basic region was important for lipid binding, we constructed a Syn1A RKRK mutant, in which the order of the juxtamembrane basic residues was rearranged, while the overall charge was maintained. The apparent PA binding affinity of the RKRK mutant (EC50 ~ 2.6 × 103 fmol) was indistinguishable from WT, indicating that Syn1A–lipid interactions can tolerate small structural changes within the juxtamembrane region (B).
To establish whether domains outside the polybasic juxtamembrane region affect Syn1A's lipid-binding specificity, we synthesized a peptide corresponding to this region (aa 252-265), as well as the corresponding 5RK/A peptide. Protein–lipid overlays demonstrated that the WT peptide closely recapitulated Syn1A's lipid-binding profile, whereas the 5RK/A mutation abrogated the peptide's ability to bind acidic phospholipids (E). One notable difference between the WT peptide and Syn1A (1-267) was that the WT peptide demonstrated a high apparent affinity for PS, a lipid for which Syn1A (1-267) exhibited almost no binding (compare , D and E). Importantly, lipid binding to Syn1A (1-259), in which the polybasic juxtamembrane region had been truncated, was largely eliminated (data not shown). Thus, although other regions in Syn1A may modify its lipid-binding profile, the basic juxtamembrane residues are clearly required for these interactions.
Full-Length Syn1A Juxtamembrane Mutants Traffic Normally to the Plasma Membrane in Live Cells
To assess the relevance of Syn1A–lipid interactions in an in vivo situation, full-length, juxtamembrane mutant Syn1A proteins were next studied in living cells. Initial experiments examined whether these mutant Syn1A proteins were expressed and targeted properly. Enhanced cyan fluorescent protein (ECFP)-tagged, full-length Syn1A mutants were generated and transiently transfected into PC-12 cells. All Syn1A constructs were cotransfected with Munc18-1, which facilitated high levels of targeting of Syn1A to the plasma membrane regions. A demonstrates that the fluorescence signal associated with both the WT and 5RK/A ECFP-Syn1A proteins trafficked normally to the plasma membrane region, as determined by confocal microscopy.
Figure 3. Syn1A 5RK/A targets to plasma membrane regions in PC-12 cells. (A) Representative confocal fluorescence and corresponding DIC images of PC-12 cells transiently transfected with eCFP-tagged Syn1A(WT) and Syn1A (5RK/A) together with Munc18-1. (B) Representative (more ...)
To quantify the extent of surface labeling between mutants, we next used dual-fluorophore–labeled Syn1A constructs, which were tagged with mRFP at the N-terminus and pHluorin at the C-terminus. pHluorin is a pH-sensitive variant of GFP, whose signal is quenched within acidic intracellular compartments, but which becomes highly fluorescent upon exposure to the neutral, extracellular solution (which in this case, occurs upon insertion of Syn1A's C-terminal transmembrane domain into the plasma membrane). We thus reasoned that the pHluorin signal would report only on the pool of exogenous Syn1A that had been correctly inserted into the plasma membrane, whereas the mRFP signal would report on the entire pool of exogenous Syn1A expressed within a cell. B shows representative epifluorescence images of PC-12 cells transiently cotransfected with the mRFP-Syn1A-pHluorin constructs and Munc18-1. The notable fluorescence of the pHluorin label at the plasma membrane region demonstrates that both the WT and 5RK/A Syn1A constructs were correctly trafficked and inserted into the plasma membrane. Importantly, measurements of the average pHluorin and mRFP fluorescence intensities across a large number of cells were comparable between the WT and 5RK/A conditions, demonstrating that both the surface levels and total expression levels of WT and 5RK/A Syn1A proteins were similar (C). D shows a scatterplot in which the mean pHluorin intensity was plotted against the mean mRFP intensity for each individual cell. Notably, there occurred a large overlap in the distribution of points between the WT and 5RK/A cells, and linear fits of these data for cells expressing moderate levels of the Syn1A (RFP mean intensities between 20 and 170) were not significantly different. Thus, the WT and 5RK/A Syn1A constructs demonstrate not only comparable expression levels, but also similar abilities to traffic and insert within the plasma membrane.
Full-Length Syn1A Juxtamembrane Mutants Demonstrate Intact Protein–Protein Interactions within Live Cells
Having determined that the full-length juxtamembrane neutralization mutants of Syn1A were capable of trafficking correctly, we next asked whether these mutants were capable of forming appropriate protein–protein interactions in vivo. The above data suggested that both WT and 5RK/A Syn1A constructs interacted similarly with Munc18-1, as coexpression of these constructs with Munc18-1 greatly facilitated trafficking of both constructs to the plasma membrane. To confirm the normal Munc18-1 interaction properties between these Syn1A constructs, we used a sensitized emission FRET approach to compare binding of the CFP-tagged WT Syn1A or 5RK/A mutant to citrine-Munc18-1 in live cells. That this FRET stoichiometry approach is an accurate reporter of the Syn1A-Munc18-1 interaction was established in our earlier publication (Liu et al., 2004
). A demonstrates that both WT and 5RK/A proteins associated with Munc18-1 similarly, across a wide range of molar ratios. This result is quantified in B, where at equimolar ratios (molar ratio between 0.9 and 1.1), the 5RK/A mutant exhibited a FRET efficiency (ED) with Munc18-1 that was similar to that of the WT protein (WT = 24.2 ± 0.01, 5RK/A = 24.8 ± 0.01). In contrast, a Syn1A mutant (I233A) that was previously shown to have reduced binding to Munc18-1 (Kee et al., 1995
), demonstrated a reduced FRET efficiency (I233A = 5.6 ± 0.01) compared with WT. Thus, neutralizing mutations within the juxtamembrane region of Syn1A do not appear to affect Syn1A's interaction with Munc18-1.
Figure 4. Syn1A 5RK/A interacts normally with Munc18-1 and SNAP25. (A) Analysis of direct CFP-Syn1A-citrine-Munc18-1 interactions by sensitized emission FRET imaging. Graph shows a pixel-by-pixel plot of FRET efficiency versus molar ratio (Syn1A:Munc18-1). Both (more ...)
We next asked whether Syn1A 5RK/A possessed the ability to interact normally with SNARE proteins, in particular, the Q-SNARE SNAP25. For these experiments, HEK293-S3 cells were transiently transfected with Syn1A (WT or mutant), Munc18-1, and EGFP-SNAP25. Co-IP experiments were performed to determine whether the 5RK/A mutant was capable of pulling down equal amounts of EGFP-SNAP25 compared with the WT Syn1A. C shows a blot from a representative experiment, which demonstrates that both the WT and 5RK/A Syn1A proteins bound a similar amount of EGFP-SNAP25. In contrast, a mutant of Syn1A (L205A/E206A) which was previously shown to have reduced binding to SNARE proteins (Dulubova et al., 1999
), demonstrated a marked reduction in binding to EGFP-SNAP25. To quantify these results, we measured the integrated densities of the Syn1A and EGFP-SNAP25 bands and determined the ratio of EGFP-SNAP25 to Syn1A in the immunoprecipitated fractions for each condition. For each experiment, the SNAP25:Syn1A ratio for each treatment was normalized to the ratio from the WT treatment, to allow comparison of treatments across experiments. D shows the average results for three independent experiments and again demonstrates that the 5RK/A Syn1A displays similar interactions with SNAP25 compared with the WT Syn1A.
Taken together, these data demonstrate that, despite profound differences in lipid-binding capabilities, the 5RK/A mutant Syn1A behaves nearly identically to the WT Syn1A in live cells, with respect to expression levels, membrane trafficking, and forming appropriate protein–protein interactions.
Use of BoNT-C Knockdown of Syn1A to Specifically Isolate the Functional Phenotypes of Exogenous Syn1A Constructs
To test whether disruption of Syn1A–lipid interactions would have a functional effect on regulated exocytosis, we transfected Syn1A juxtamembrane neutralization mutant constructs into live secretory cells (PC-12 cells or bovine adrenal chromaffin cells). To reduce potentially confounding effects from endogenous Syn1A in these cells, we transfected the cells with the light chain of BoNT-C, which cleaves Syn1A and precludes it from mediating membrane fusion (Schiavo et al., 1995
). Syntaxin 4 was previously reported to be resistant to cleavage by BoNT-C and to contain an Ile in place of Lys at residue 253 of the BoNT-C cleavage site (Schiavo et al., 1995
). We therefore tested whether a similar mutation (K253I) in Syn1A, would generate a BoNT-C–resistant Syn1A. For this analysis, PC-12 cells were cotransfected with an N-terminally tagged CFP-Syn1A and Munc18-1, with or without BoNT-C. Cells were then imaged using conventional fluorescence microscopy. In the absence of BoNT-C, both the WT Syn1A and Syn1AK253I
constructs targeted to plasma membrane regions (A). In the presence of BoNT-C, however, the WT Syn1A signal was redistributed as a diffuse cytosolic signal within the cells, indicating cleavage by BoNT-C, whereas the Syn1AK253I
signal distributed to the plasma membrane, indicating resistance to cleavage by BoNT-C (A). These experiments were quantified by scoring at least 100 random cells from each condition (while blinded to the conditions) as demonstrating either a cytosolic or membrane fluorescence distribution. Importantly, in the presence of BoNT-C, the percent of cells demonstrating a cytosolic fluorescence distribution was 63% for WT Syn1A, compared with only 5% for Syn1AK253I
Figure 5. BoNT-C knockdown allows isolation of functional effects of exogenous Syn1A constructs in live secretory cells. (A) Top, representative epifluorescence images demonstrating that Syn1A (K253I) is resistant to cleavage by BoNT-C. Images compare the subcellular (more ...)
We next tested whether a BoNT-C resistant Syn1AK253I
construct could rescue secretion in BoNT-C–treated cells. hGH secretion assays (Wick et al., 1993
) were performed on PC-12 cells transfected with various combinations of the BoNT-C light chain, Munc18-1, and either WT Syn1A or Syn1AK253I
. B shows that transfection of the BoNT-C light chain into PC-12 cells effectively reduced secretion to 30% of control secretion. Importantly, Syn1AK253I
rescued secretion in BoNT-C–transfected cells, to roughly 78% of control secretion, but only when cotransfected with Munc18-1. This requirement for Munc18-1 was likely the result of enhanced membrane targeting of Syn1A, as Munc18-1 itself was insufficient to rescue the BoNT-C knockdown of secretion (35% of control secretion). Importantly, WT Syn1A was also unable to rescue secretion in BoNT-C–transfected cells, even when cotransfected with Munc18-1 (29% of control secretion). This clearly indicates that rescue of the BoNT-C knockdown is specific to the expression and proper targeting of a functional BoNT-C–resistant Syn1A to the plasma membrane region in these cells. Of note, although higher expression levels of BoNT-C were sufficient to achieve a more complete knockdown of secretion (reduction to <10% of control secretion), we were often unable to rescue this phenotype by coexpression of Syn1AK253I
and Munc18 (data not shown). We attribute this to the fact that at higher concentrations, BoNT-C may also cleave SNAP25 in addition to Syn1A (Williamson et al., 1996
Neutralizing Mutations within Syn1A's Polybasic Juxtamembrane Region Result in a Progressive Inhibition of Syn1A's Secretory Function
To compare the ability of full-length Syn1A juxtamembrane neutralization mutants to rescue secretion, we used the BoNT-C knockdown assay in cotransfected PC-12 cells. A demonstrates that neutralizing mutations within Syn1A's polybasic juxtamembrane domain resulted in a progressive decrease in secretory function. When normalized to the extent of rescue seen with Syn1AK253I, the Syn1AK253I(5RK/A) mutant was only capable of rescuing secretion to 67 ± 3% of the Syn1AK253I control level (n = 20). The Syn1AK253I(R262A/R263A) mutant rescued secretion to 77 ± 4% of the Syn1AK253I control level (n = 16), whereas the Syn1AK253I(R262A) mutant exhibited a phenotype similar to Syn1AK253I, rescuing secretion to 97 ± 4% (n = 16) of the Syn1AK253I control level. Considering that the residual baseline secretion after BoNT-C knockdown accounts for roughly 45% of the rescued control secretion (B and A, left, dotted line), the actual deficit in secretion resulting from neutralization of Syn1A's juxtamembrane region was quite substantial, with the Syn1AK253I(5RK/A) mutant rescuing secretion to only 43% of the Syn1AK253I control (A, right). Importantly, this decline in secretory function of the juxtamembrane Syn1A mutants correlates well with the decrease in apparent affinity of these mutants for binding phosphatidic acid (B) and presumably, with their affinity for other acidic phospholipids as well.
Figure 6. Neutralization of Syn1A's polybasic juxtamembrane region results in a decrease in evoked secretion. (A) Progressive neutralizing mutations of Syn1A's polybasic juxtamembrane domain result in a graded reduction in evoked hGH secretion. Transfected PC-12 (more ...)
Neutralizing Mutations within Syn1A's Polybasic Juxtamembrane Region Result in a Reduction in Fusion Event Frequency
We next sought to determine the mechanism by which neutralization of Syn1A's juxtamembrane region resulted in secretory inhibition. Importantly, both Syn1AK253I and Syn1AK253I(5RK/A) cells demonstrated comparable calcium fluxes upon depolarization, as measured using the calcium indicator Fura2 (B). PC-12 cells were transfected with BoNT-C, Syn1AK253I or Syn1AK253I(5RK/A), Munc18-1, and RFP (to identify transfected cells), preloaded with the Fura2 AM ester, and depolarized using a brief, 5-s local perfusion with 100 mM K+. Changes in intracellular [Ca2+], reported by a change in F340/F380, were comparable between Syn1AK253I (control) and Syn1AK253I(5RK/A) cells in all parameters analyzed, including the baseline calcium levels (control: 0.91 ± 0.01, n = 51; 5RK/A: 0.91 ± 0.01, n = 59) and the peak change in calcium (control: 0.20 ± 0.02, n = 51; 5RK/A: 0.19 ± 0.01; n = 59), as well as the kinetics of the calcium fluxes. This suggests that the decrease in evoked secretion seen in the 5RK/A cells occurs downstream of calcium influx.
To probe the temporal resolution and analyze the dynamics of individual fusion events, we next used carbon fiber amperometry. For these single-cell experiments, bovine adrenal chromaffin cells were used rather than PC-12 cells, as we found the exocytotic responses produced by these cells to be far more robust than those of PC-12 cells. Chromaffin cells were biolistically transfected with BoNT-C, either Syn1AK253I
(control) or Syn1AK253I
(5RK/A), and GFP. Coexpression of Munc18-1 was not necessary in these experiments, as we have previously shown that Syn1A can traffic to the plasma membrane in chromaffin cells without the need for Munc18-1 overexpression (Gladycheva et al., 2007
). Transfected cells were stimulated to secrete by local perfusion of 100 mM K+
for 60 s, during which time amperometric spikes were recorded. Representative amperometric traces are displayed in C. In agreement with the hGH data above, the average number of spikes (fusion events) per cell was substantially reduced in the 5RK/A cells compared with control (control: 33.6 ± 4.0 spikes/cell, n = 86; 5RK/A: 24.3 ± 2.6, n = 82; D). However, the frequency distribution of the spikes, when normalized to the total number of spikes for each condition, was identical between control and 5RK/A cells (E). This suggests that, rather than affecting a specific subpopulation of vesicles, the decrease in fusion events observed in the 5RK/A condition results from a generalized decrease in fusogenicity.
Syn1A's Polybasic Juxtamembrane Region Regulates Fusion Pore Dynamics
We hypothesized that the generalized decrease in fusogenicity seen in 5RK/A cells might also be manifest in individual fusion events, particularly with regards to fusion pore dynamics. Many amperometric spikes contain a PSF, which is believed to represent the initial flux of catecholamine through the fusion pore, before dilation of the fusion pore and full fusion of the vesicle with the plasma membrane (Chow et al., 1992
; Zhou et al., 1996
). Analysis of PSF can thus elucidate details surrounding the late stages of vesicle fusion, especially those regarding the formation and expansion of the fusion pore. Changes in PSF amplitudes are believed to represent changes in fusion pore diameter, whereas changes in PSF duration are thought to represent changes in stability of the fusion pore and kinetics of fusion pore expansion.
Representative PSF are shown in A. Interestingly, the Syn1AK253I
(5RK/A) cells demonstrated decreased PSF amplitudes, in addition to increased PSF durations, compared with control Syn1AK253I
(control) PSF (, B and C). In other words, the fusion pores in the 5RK/A cells were not only smaller in diameter, but also took longer to expand to full fusion. The mean PSF amplitude for control cells was 7.66 ± 0.27 pA, which was slightly reduced to 6.18 ± 0.24 pA for 5RK/A cells (n = 584 control PSF, n = 576 5RK/A PSF; B, left). The mean PSF duration for control cells was 7.28 ± 0.74 ms, which was lengthened to 12.33 ± 1.05 ms in 5RK/A cells (n = 584 control PSF, n = 576 5RK/A PSF; C, left). Qualitatively, these results also held true under a more stringent analysis scheme, in which the median PSF parameters for each cell were first determined, and the medians then were averaged across cells. Median PSF amplitudes were 6.16 ± 0.43 and 4.42 ± 0.22 pA for control and 5RK/A cells, respectively (n = 66 control cells, n = 78 5RK/A cells; B, right). Median PSF durations were 5.44 ± 0.54 ms for control, compared with 8.23 ± 0.95 ms for 5RK/A cells (n = 66 control cells, n = 78 5RK/A cells; C, right). In all analysis schemes, the differences between control and 5RK/A PSF parameters were statistically significant (p < 0.05). Thus, neutralization of Syn1A's polybasic juxtamembrane region results in a decrease in fusion pore diameter, as well as a delay in fusion pore expansion. As formation and expansion of the fusion pore have been modeled to be among the most energetically expensive steps in the membrane fusion process (Chernomordik and Kozlov, 2003
; Cohen and Melikyan, 2004
), these data suggest that energetic inefficiencies in the fusion process may, in part, account for the secretory defects observed with the 5RK/A mutant.
Figure 7. Neutralization of Syn1A's polybasic juxtamembrane region results in a decrease in fusion pore diameter and a lengthening of fusion pore duration. (A) Representative PSF for control (Syn1AK253I) or Syn1AK253I (5RK/A) expression conditions (+ BoNT-C) during (more ...)
Manipulation of Intracellular Phosphatidic Acid Levels Differentially Regulates Secretion from Control and 5RK/A Cells
Thus far, we have demonstrated that Syn1A binds multiple acidic phospholipids, including the fusogenic lipid, phosphatidic acid, and that these interactions are mediated through Syn1A's polybasic juxtamembrane domain. Moreover, we have shown that neutralization of Syn1A's polybasic juxtamembrane domain (5RK/A mutant) results in a significant decrease in stimulated secretion, in addition to effects on fusion pore amplitude and duration that are suggestive of an energetic defect in the fusion process. To determine whether the functional effects seen with the 5RK/A mutant are a direct result of the inability of this mutant to bind lipids, we performed hGH secretion assays to test how various alterations in membrane lipid composition affected the abilities of the Syn1AK253I(control) or Syn1AK253I(5RK/A) mutant to rescue BoNT-C–mediated knockdown of secretion in PC-12 cells. As in prior experiments, all results have been normalized to the level of rescued secretion observed in the Syn1AK253I control cells in the absence of any lipid manipulations.
To determine whether the secretory defect in the 5RK/A cells might be related to the energetics of fusion, we first tested the effects of externally applied LPC (1 μM), an inverted-cone–shaped lipid that facilitates fusion pore formation and expansion by inducing positive curvature to the outer leaflet of the membrane bilayer. In these experiments, using an LPC concentration that was low enough so as not to affect stimulated secretion from the Syn1AK253I control cells, we observed a statistically significant and substantial (>50%) increase in secretion from the Syn1AK253I(5RK/A) cells (38.6 vs. 59.8% rescued secretion in the absence or presence of LPC, respectively; ). This result suggests that the exocytotic defect in the 5RK/A mutant occurs at a late step in fusion and is likely energetic in nature.
Figure 8. Functional phenotypes of Syn1AK253I control and 5RK/A mutant are differentially regulated by manipulation of phosphatidic acid levels in live PC-12 cells. Graph demonstrates averaged results from hGH secretion assays utilizing the BoNT-C knockdown and (more ...)
The partial rescue of the 5RK/A secretory phenotype by externally applied LPC (which induces positive curvature on the outer bilayer leaflet) nicely complemented our previous result demonstrating that Syn1A interacts with phosphatidic acid, a cone-shaped, fusogenic lipid (which induces negative curvature on the inner bilayer leaflet). We thus investigated whether specific Syn1A-phosphatidic acid interactions are important for regulated exocytosis. For these experiments, the BoNT-C knockdown and rescue assay was repeated, while overexpressing either phospholipase D1 (PLD1) or an siRNA construct previously demonstrated to target PLD1 (Zeniou-Meyer et al., 2007
). PLD1 is a stimulus-activated enzyme that cleaves phosphatidylcholine to generate free choline and phosphatidic acid. Overexpression of PLD1 has been shown to enhance regulated exocytosis, whereas knockdown of PLD1 activity results in a decrease in secretion (Humeau et al., 2001
; Vitale et al., 2001
; Hughes et al., 2004
; Huang et al., 2005
; Zeniou-Meyer et al., 2007
). Although the extent to which PLD1 overexpression and knockdown affected control secretion in our BoNT-C knockdown experiments () was slightly less than has been previously reported, this likely resulted from the BoNT-C knockdown assay's requirement for simultaneous overexpression of multiple constructs. As such, changes in PLD1 levels achieved in this system were likely more limited than when PLD1 or siRNA-PLD1 are the only constructs being overexpressed (as in prior studies).
Importantly, we found that overexpression of PLD1 resulted in a near complete phenotypic rescue of the 5RK/A mutant [117 vs. 103% rescued secretion, for PLD1-treated Syn1AK253I control cells or PLD1-treated Syn1AK253I(5RK/A) cells, respectively; ]. Also of substantial interest was the finding that siRNA-mediated knockdown of PLD1 drastically reduced secretion from Syn1AK253I control cells, while having little effect on secretion from Syn1AK253I(5RK/A) cells [74% reduction in secretion for Syn1AK253I control cells, compared with a 24% reduction for Syn1AK253I(5RK/A) cells, in the presence of siRNA-PLD1; ]. These data provide strong evidence that the loss of interaction between Syn1A and phosphatidic acid largely accounts for the secretory defects seen with the Syn1A 5RK/A mutation and rule out the possibility that this secretory defect resulted from disruption of untested Syn1A–protein interactions or from alterations in Syn1A structure. More importantly, these results demonstrate that Syn1A–lipid interactions are critically important in regulating the overall levels of evoked secretion in live neuroendocrine cells.