Fluorescence Resonance Energy Transfer Reports Properties of Syntaxin1A Interaction with Munc18-1 in Vivo*

Syntaxin1A, a neural-specific N-ethylmaleimide-sensitive factor attachment protein receptor protein essential to neurotransmitter release, in isolation forms a closed conformation with an N-terminal α-helix bundle folded upon the SNARE motif (H3 domain), thereby limiting interaction of the H3 domain with cognate SNAREs. Munc18-1, a neural-specific member of the Sec1/Munc18 protein family, binds to syntaxin1A, stabilizing this closed conformation. We used fluorescence resonance energy transfer (FRET) to characterize the Munc18-1/syntaxin1A interaction in intact cells. Enhanced cyan fluorescent protein-Munc18-1 and a citrine variant of enhanced yellow fluorescent protein-syntaxin1A, or mutants of these proteins, were expressed as donor and acceptor pairs in human embryonic kidney HEK293-S3 and adrenal chromaffin cells. Apparent FRET efficiency was measured using two independent approaches with complementary results that unambiguously verified FRET and provided a spatial map of FRET efficiency. In addition, enhanced cyan fluorescent protein-Munc18-1 and a citrine variant of enhanced yellow fluorescent protein-syntaxin1A colocalized with a Golgi marker and exhibited FRET at early expression times, whereas a strong plasma membrane colocalization, with similar FRET values, was apparent at later times. Trafficking of syntaxin1A to the plasma membrane was dependent on the presence of Munc18-1. Both syntaxin1A(L165A/E166A), a constitutively open conformation mutant, and syntaxin1A(I233A), an H3 domain point mutant, demonstrated apparent FRET efficiency that was reduced ∼70% from control. In contrast, the H3 domain mutant syntaxin1A(I209A) had no effect. By using phosphomimetic mutants of Munc18-1, we also established that Ser-313, a Munc18-1 protein kinase C phosphorylation site, and Thr-574, a cyclin-dependent kinase 5 phosphorylation site, regulate Munc18-1/syntaxin1A interaction in HEK293-S3 and chromaffin cells. We conclude that FRET imaging in living cells may allow correlated regulation of Munc18-1/syntaxin1A interactions to Ca2+-regulated secretory events.

Synaptic transmission is triggered by Ca 2ϩ -dependent fusion of synaptic vesicles with the presynaptic membrane. Central to the vesicular targeting, docking, and fusion process that underlie this regulated exocytotic event are soluble N-ethylmaleimide-sensitive factor and its attachment protein (SNAP) 1 and membrane-bound receptors (SNAREs) (1)(2)(3). Synaptic vesicles contain the v-SNARE synaptobrevin (also termed VAMP) that interacts with high specificity to the integral presynaptic plasma membrane t-SNAREs syntaxin1A and SNAP25. These SNARE proteins assemble into a ternary complex via interaction of their SNARE motifs into parallel four-helix bundles (4,5). SNARE complex assembly is believed to facilitate close membrane apposition and is a step essential to the fusion of synaptic vesicles with the plasma membrane (6). The assembly of trans-SNARE complexes for Ca 2ϩ -regulated exocytosis is subject to strict spatial and temporal regulation (7).
Members of the Sec1/Munc18 (SM) gene family have also emerged as proteins essential for membrane trafficking and fusion (8). This is evidenced by the accumulation of vesicles in loss of function SM mutants in yeast (9,10), Drosophila (11,12) and Caenorhabditis elegans (13). SM proteins bind with high affinity to members of the syntaxin gene family and exhibit strong specificity to a membrane transport pathway, although members of the family are highly homologous and may interact with more than one syntaxin isoform (14). Consistent with a strict specificity, the way SM proteins interact with syntaxins is fundamentally different among different SM/syntaxin pairs. For example, yeast Sec1p binds to assembled SNARE complexes at the cell membrane, yeast Sly1p binds to a conserved N-terminal domain of syntaxin for fusion of transport vesicles from the endoplasmic reticulum to Golgi, whereas interaction of mammalian Munc18-1 with syntaxin at the plasma membrane requires the complete cytosolic domain of syntaxin and is absent in assembled SNARE complexes (15)(16)(17)(18). Multiple functions have been proposed for SM proteins, including those of a trafficking factor for syntaxin (19), a regulator of vesicle docking (20,21), a facilitator of priming processes required for assembly of SNARE complexes (22), and a mediator of the kinetics of expansion of the fusion pore (23,24). However, despite considerable evidence demonstrating their necessity for normal membrane trafficking and fusion, the precise function(s) of SM proteins remains poorly defined.
In the case of synaptic transmission, the SM protein Munc18-1 binds with high affinity to the plasma membrane t-SNARE syntaxin1A. Syntaxin1A consists of an N-terminal domain composed of an antiparallel three-helix bundle (termed the Habc domain) that is connected via a short flexible linker to a SNARE motif (H3 domain) residing just proximal to a singletransit transmembrane domain (25,26). Analysis of the crystal structure of syntaxin1A showed that, in the absence of additional protein interactions, syntaxin1A adopts a folded conformation, whereby the Habc domain folds onto the H3 domain, and the structure is stabilized by a hydrophobic groove of the Habc domain orientated toward the H3 domain (22). The folding of syntaxin1A into this conformation hinders interaction of its SNARE motif with its cognate SNAREs and, thereby, inhibits assembly of SNARE core complexes (27,28). Paradoxical to its necessity for exocytosis, Munc18-1 in vitro interacts with and stabilizes this folded conformation of syntaxin1A and inhibits the association of VAMP/synaptobrevin and SNAP25 with syntaxin1A (28,29). However, in vivo evidence also suggests that the Munc18-1 interaction is essential to facilitate trafficking of syntaxin1A from the Golgi to the plasma membrane as well as to promote a conformational change in syntaxin1A required for interaction of its SNARE motif with cognate SNARE proteins and/or to nucleate zippering of SNARE protein complexes together (2,6,19,30). Conformational opening of syntaxin1A is a critical regulatory step in SNARE complex formation, because opening of the yeast syntaxin ortholog Sso1 increases by 2000-fold assembly of the SNARE core complex (31), and isolated H3 domains of syntaxin1A form SNARE complexes 7-fold faster than the complete syntaxin1A protein (32). Mice lacking the neural-specific Munc18-1 isoform demonstrate a complete loss of synaptic transmission (33). Thus, there is compelling evidence to support regulation of the Munc18-1/syntaxin1A interaction as a critical determinant in the SNARE complex assembly that is necessary for the Ca 2ϩ -dependent exocytotic process of neurotransmitter release and synaptic transmission.
A number of mechanisms regulating the assembly and disassociation of the Munc18-1/syntaxin1A complex have been proposed. Among these, regulation by protein phosphorylation has been best characterized. For example, Munc18-1 has been shown both in vitro and in vivo to be a substrate for protein kinase C (PKC), cyclin-dependent kinase 5 (Cdk5), and deathassociated protein kinase (34 -37). Phosphorylation by PKC occurs on Ser-306 and Ser-313 in vitro and results in a reduction in the affinity of Munc18-1 for syntaxin1A and inhibition of formation of Munc18-1/syntaxin1A complexes (34). Functionally, PKC phosphorylation of Ser-313 has been reported to alter the dynamics of vesicle release events (23,24). By comparison, Cdk5 phosphorylates Munc18-1 at a single site in vitro (Thr-576), and this phosphorylation is capable of mediating disassembly of preassembled Munc18-1/syntaxin1A complexes (38). Pharmacological inhibition of Cdk5 inhibited induced secretion from adrenal chromaffin cells (38) and pancreatic beta cells (39). In addition to protein phosphorylation, a number of direct protein-mediated mechanisms have also been suggested as important regulators of the Munc18-1/syntaxin1A complex. Included among these are proteins of the Munc13 and tomosyn gene families. Munc13 binds to the N terminus of syntaxin1A, and this interaction has been proposed to stabilize an open conformation of syntaxin that is necessary for pairing the syntaxin1A SNARE motif with cognate SNAREs (40,41). Tomosyn contains a single C-terminal V-SNARE motif that can substitute for the VAMP SNARE in syntaxin1A SNARE complexes (42,43). In addition, tomosyn has been shown to compete with Munc18-1 for binding syntaxin1A in vitro. Munc18-1 also interacts with other proteins, such as Mints (44,45) and DOC2 (46), although their role in modulating the Munc18-1/ syntaxin1A interaction is unknown.
To date, characterization of specific mechanisms of formation and disassembly of a Munc18-1/syntaxin1A complex have re-lied on in vitro biochemical assays of protein-protein interaction and mechanistic interpretations of alterations in functional responses using genetic approaches in cell or animal models (2,6,8). However, to characterize the temporal and spatial aspects of regulation associated with activation of the secretory pathway requires an experimental approach that directly reports the state of the bimolecular Munc18-1/ syntaxin1A protein interaction within living cells. In the present study, we used fluorescence resonance energy transfer (FRET) as a non-invasive imaging method to characterize the interaction of Munc18-1 with syntaxin1A in both the HEK293-S3 cell line and in primary cultures of bovine adrenal chromaffin cells. As FRET probes we used enhanced cyan fluorescent protein (CFP) and the citrine variant of enhanced yellow fluorescent protein (cYFP) fused to the N termini of Munc18-1 and syntaxin1A, respectively. Two independent FRET methods have been used to provide additional confidence in measurement and quantification of apparent FRET efficiency of the bimolecular interactions. To establish the selectivity of the FRET signal and to establish the sensitivity of the approach to report changes in the affinity of the bimolecular interaction within cells, we have tested the effects of various expression paradigms and mutants of Munc18-1 and syntaxin1A. In addition, we evaluated the effect of PKC and of Cdk5 phosphorylation of Munc18-1 on formation of Munc18-1/ syntaxin1A complexes and on the trafficking of syntaxin1A from the Golgi complex to the plasma membrane.

Chemicals and Expression
Constructs-Mouse monoclonal antibody against syntaxin1A was from Sigma; mouse monoclonal antibody against Munc18-1 and the Golgi marker protein GS27 were from BD Transduction Lab (San Jose, CA); Alexa Fluor-647 and Alexa Fluor-488 goat anti-mouse antibodies were from Molecular Probes (Eugene, OR). All other chemicals were obtained from Sigma. HEK293 cells stably expressing the rat ␣ 1B and human ␤ 1B and ␣ 2 ␦ voltage-gated calcium channel subunits (termed HEK293-S3 cells) were a gift from D. Rock (Warner-Lambert Parke Davis, Ann Arbor, MI). The LoxP sequence from pLP-EGFP-C1 (Clontech) was subcloned into the multiple cloning site regions of pcDNA3.1, pECFP-C1, and pEcYFP-C1 (Q39M mutant of pEYFP-C1) to generate recipient vectors for subcloning using the Crerecombinase-mediated Creator System (Clontech). Rat Munc18-1 and rat syntaxin1A were merged to the C terminus of the enhanced cyan mutant of green fluorescent protein (ECFP) and EcYFP. Mutants of Munc18-1 and syntaxin 1A were constructed using the polymerase chain reaction-based Quickchange Site-Directed Mutagenesis kit from Stratagene (Cedar Creek, TX). The sequence fidelity of all expression constructs was confirmed by DNA sequencing (University of Michigan DNA Sequencing Core).
Cell Culture and Transfections-HEK293-S3 cells were plated and cultured in RPMI 1640 with L-glutamine supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin (Life Technologies, Inc.), 0.4 mg/ml of hygromycin, and 0.6 mg/ml of geneticin at 37°C in 95% O 2 -5% CO 2 for 2 days on coverslips (thickness 1) affixed to the bottom of 35-mm culture dishes before transfection. Transfections were performed using Lipofectamine 2000 (Invitrogen), according to the manufacturer's instructions. One h before transfection, cells were placed in Dulbecco's modified Eagle's medium lacking antibiotics and supplemented with 1% L-glutamine, 1% non-essential amino acids, and 10% fetal bovine serum. After 4 -6 h with DNA-Lipofectamine 2000, the culture medium was changed to RPMI 1640. Cells were used for optical investigation 24 -36 h after transfection.
Chromaffin cells were isolated and cultured from bovine adrenal glands using divalent metal ion-free rinse, collagenase digestion, and gradient centrifugation as described previously (47). Cells were cultured and transfected in 35-mm culture dishes and replated onto collagen-coated glass coverslips 4 h before imaging. Chromaffin cells were transfected using biolistic particle bombardment (gene gun, Bio-Rad) according to the manufacturer's instructions.
Immunocytochemistry-HEK293-S3 cells cultured on coverslips were fixed with 4% paraformaldehyde in physiological saline solution (PSS) for 20 min, rinsed with PSS, and quenched with 50 mM NH 4 Cl-PSS for 12 min. PSS contained 145 mM NaCl, 5.6 mM KCl, 5.6 mM glucose, 0.5 mM MgCl 2 , 2.2 mM CaCl 2 , and 15 mM HEPES (pH 7.4). After rinsing with PSS, cells were permeabilized with 0.2% Triton X-100 in PSS containing 2% normal goat serum for 20 min and then incubated sequentially with Munc18-1 or syntaxin1A mouse monoclonal antibody (1:400 to 1:800 dilution) for 2 h and with Alexa Fluor-488 goat antimouse secondary antibody (1:200 dilution) for 1 h. After rinsing, coverslips were mounted on glass slides with 75% glycerol-PSS. The Golgi marker GS27 was localized in HEK293-S3 cells using a mouse monoclonal antibody (1:200 dilution) and Cy5-conjugated anti-mouse secondary antibody (1:200 dilution). Digital fluorescence images from HEK293-S3 and chromaffin cells were viewed with either an Olympus fluorescence microscope or a Zeiss LSM 510 META confocal microscope (conventional channel mode; 488-nm argon laser line), and images were processed using Photoshop 6.0 software (Adobe Systems Inc.). Controls with no primary antibody were used to confirm the specificity of the immunochemical reactivity.
Confocal Microscopy of CFP and cYFP Constructs and FRET Measurements-Cells were cultured on glass coverslips for 1-2 days after transfection. Cells were then fixed as described above and mounted on slides with 75% glycerol-PSS. Image acquisition and determination of FRET efficiency by acceptor photobleaching were obtained using the lambda mode of a Zeiss LSM 510 Meta confocal microscope and analytical software. Briefly, control emission spectra from cells expressing CFP or cYFP fusion proteins alone were obtained with the Meta system, using the 458-and 514-nm laser lines of the argon laser, respectively. The Meta instrumentation uses an optical grating to record emission spectra (10-nm increments) by projecting the emission on 32 channels of the Meta detector. These reference spectra were used to apply pixel-bypixel linear unmixing calculations (Meta software) to completely decompose or digitally separate the contribution of cYFP-S1A and CFP-M18 in the mixed emission spectrum recorded from cotransfected cells, thereby producing unmixed CFP and cYFP images. The effectiveness of the linear unmixing algorithm was validated using cells that highly overexpressed CFP alone and cYFP alone. After linear unmixing, no cellular fluorescence was found in the YFP channel for cells expressing CFP or in the CFP channel for cells expressing cYFP, even when post-image acquisition contrast settings were boosted to high levels.
For measurement of FRET efficiency at the plasma membrane by the acceptor photobleach method, a mixed spectral image (2-m optical section) of cotransfected cells was collected, and an area of the plasma membrane was then selected for photobleaching of cYFP. A protocol was then used which recorded pre-and post-bleach images using 458-nm excitation, with an intermediate bleach of the selected area with the 514 laser (acceptor bleach). FRET was resolved as an increase in the CFP (donor) signal after photobleaching of cYFP (acceptor). CFP and cYFP digital images were subsequently generated after linear unmixing as described above. Reduction of the cYFP signal after photobleaching for CFP-Munc18-1 and cYFP-syntaxin1A cotransfected HEK293-S3 cells was on average 87 Ϯ 1.1% (n ϭ 37). For determination of FRET efficiency for the plasma membrane region in the selected bleach area, the average pixel intensity (I) of the CFP signal was determined from the unmixed pre-and post-bleach images using Zeiss software. Relative FRET efficiency was calculated as (1 Ϫ [CFP I pre-bleach / CFP I post-bleach ]) ϫ 100%. For control purposes, an area of the plasma membrane that was not photobleached was also analyzed for FRET. Application of the photobleach protocol on CFP-only expressing cells, followed by application of the linear unmixing algorithm for CFP ϩ cYFP signals, demonstrated an apparent 9.5 Ϯ 1.9% (n ϭ 13) increase in CFP signal that was not subtracted from the FRET efficiency values reported. The cause of the change in CFP upon photobleach remains undetermined, although one possibility is a small consistent photoactivation by the 514-nm bleach. This effect on CFP, together with somewhat less than complete bleach of all acceptors and variation in CFP/ cYFP ratios in cotransfected cells, may impact FRET efficiency positively and/or negatively. Therefore, the FRET values reported by acceptor photobleach are relative estimates that serve to confirm FRET data from a filter-based approach while providing the advantage of high spatial resolution. Digital images were exported to Photoshop 6.0 for photographic processing.
For experiments with HEK293-S3 cells in which the Golgi marker GS27 was localized together with CFP-Munc18-1 and cYFP-syntaxin1A, the channel mode of the Zeiss confocal was first used to record Cy5 immunofluorescence of GS27 (HeNe2 633-nm laser line, 644 -725-nm emission filter), followed by capture of CFP and cYFP fluorescence in the lambda mode, linear unmixing, and calculation of FRET efficiency as described above.
Measurement of FRET Stoichiometry by Sensitized Emission on a Conventional Fluorescence Microscope-FRET imaging was performed on transfected cells 24 -36 h after transfection. The microscope consisted of an inverted fluorescence microscope (Olympus, IX71) interfaced to a TILL-Photonics Polychrome IV xenon lamp-based monochrometer (Till-Photonics, Grafelfing, Germany), a polychroic mirror that allowed detection of multiple fluorophores (436 -500 nm, Chroma Technology Corp., Brattleboro, VT), a Planapo ϫ60 water immersion objective (1.2 NA), a multispec microimager (Optical Insights, Sante Fe, NM) containing a dichroic splitter (505dcxr) and emission filters (D465/30 and HQ535/30) to allow simultaneous two-channel monitoring of emission fluorescence, and a cooled digital CCD camera (TILL-I-MAGO QE). Control of image acquisition and excitation wavelengths was carried out under software control (TILL-Vision). Pixel-by-pixel alignment of images was performed by hardware adjustments of the multispec microimager and offline positioning of images using TILL-Vision software. All analysis of the acquired and adjusted images was performed using Metamorph image-processing software (version 6.1, Universal Imaging, Inc., Malvern, PA).
Individual images were acquired with an exposure time of 250 ms, and a single or 10-frame averaged image was generated for each filter set combination. No differences were found between data for single and frame-averaged acquisition, but as a result of lowered image noise, the FRET values reported and the images shown were based on the 10frame averaged images. Images corresponding to three excitation/emission wavelength settings were collected. The CFP fusion protein was used as a FRET donor with the cYFP fusion protein as the FRET acceptor. These include acceptor excitation/acceptor emission (AA), donor excitation/donor emission (DD), and donor excitation/acceptor emission (DA). In addition, for each excitation/emission filter combination, an averaged background and shade correction image was collected before an experiment for offline correction of cell images for camera bias and/or uneven illumination or fluorescence signal across the field imaged. The background image was a 20-frame average of images collected with the illumination light source blocked. The shade correction image was a 20-frame averaged image of purified solutions of bacterially expressed CFP and cYFP (ϳ50 nM) in phosphate-buffered saline (Life Technologies, Inc.) buffer sandwiched between a coverslip and glass slide precoated with bovine serum albumin (1 mg/ml in phosphate-buffered saline). Background and shade images were used to generate corrected images (C I ) of the cells for each excitation/emission setting according to the equation, where S max is the shade image maximum value, S is the shade image, and B is the background.
Quantification of FRET from the corrected cell images (i.e. I DD , I DA , I AA ) was determined according to a recently described procedure termed FRET stoichiometry (48). This method determines three parameters for each pixel of a cell image. The first parameter is the apparent FRET efficiency (E A ), which is the product of the FRET efficiency of the specific bimolecular interaction, termed characteristic FRET efficiency (E C ), and the fraction of acceptor in complex with the donor. E C per se is dependent upon orientation and distance between the fluorescent probes. Because E A is proportional to the fraction of acceptor in complex, it can be used to measure changes in the fraction of the acceptor in complex. E A was determined according to the relation, where I DA , I DD , and I AA represent shade and background corrected fluorescent intensity corresponding to wavelengths of donor excitationacceptor emission (DA), donor excitation-donor emission (DD), and acceptor excitation-acceptor emission (AA). The proportionality constants ␣, ␤, and ␥ were determined as described below. The second parameter determined is E D , which is the apparent donor efficiency and is the product of E C and the fraction of the donor in complex. E D was determined according to the relation.
The third parameter is the acceptor-donor molar ratio (RATIO). RATIO indicates the mole fraction of acceptor to donor in each pixel of the collected cell image. RATIO was calculated according to the formula A RATIO of 1 indicates equal mole fractions of acceptor to donor. Equations used to determine these three parameters are derived in Hoppe et al. (48) and require determination of four constants identified as ␣, ␤, ␥, and on the optical system used for measurements. The proportionality constant ␣ relates the intensity of acceptor emission with donor excitation to that with acceptor excitation and is determined from transfected cells expressing only cYFP. ␤ relates the intensity of donor emission to that of acceptor emission with donor excitation and is determined from transfected cells expressing CFP alone. The constant ␥ was determined from cells expressing a CFP-cYFP fusion protein construct in which the two fluorophores are directly linked by a 16amino acid linker (48). This constant defines the ratio of the extinction coefficient of the acceptor to the donor at the donor excitation. The proportionality constant relates the intensity of acceptor emission resulting from donor energy transfer to the concomitant decrease in donor fluorescence. In our optical system, ␣ and ␤ were 0.015 and 0.942, respectively, and ␥ and were 0.067 and 0.014, respectively. An E C value of 0.37 determined for the linked CFP-cYFP construct was taken from the literature (48).
Measurement of Efficiency of Syntaxin1A Trafficking to Plasma Membrane-The relative amount of expressed cYFP-syntaxin1A associated with plasma membrane regions was quantified offline across treatments from background-and shade-corrected cYFP images. For each imaged cell, the total cell area and average fluorescence intensity of the cYFP signal were determined. Area and average intensity of the "cytoplasm region" was then determined for each cell by constricting the region measured by five pixels from the outer perimeter of the plasma membrane. Averaged cYFP signal at the plasma membrane region was then calculated according to the formula, where A cell is the total area of cell, I cell is the averaged cYFP intensity of the cell, A cyto is the total area of the cytoplasm, and I cyto is the averaged cYFP intensity of the cytoplasm. Measurements of images were performed using Photoshop 6.0 software.
Electrophysiological Measurements of Steady-state Inactivation of Voltage-gated Calcium Channels-Recordings were made using an Axopatch 200A amplifier (Axon Instruments, Foster City, CA) through an ITC-16 AD/DA interface (Instrutech Corp., Great Neck, NY) under the control of Pulse Control integrated into IGOR PRO software version 4.02a (Wavemetrics Inc., Lake Oswego, OR). Patch pipettes (2-4 M⍀) were pulled on a Sutter Instruments P-87 microelectrode puller using borosilicate glass (1.5-mm outside diameter, type 617000, AM Systems, Carlsborg, WA) and coated with elastomer (Sylgard, Dow Corning, Midland, MI) to minimize capacitance. All recordings were made using the amphotericin B perforated patch-clamp method under voltageclamp at room temperature. The patch pipette recording solution contained 115 mM CsOH, 115 mM CH 4 O 3 S, 1 mM MgCl 2 , 13 mM CsCl, 30 mM HEPES, 1 mM CaCl 2 , and 1 mM EGTA (pH 7.3), with 0.26 mM amphotericin B added immediately before recording. The extracellular solution consisted of 120 mM tetraethylammonium chloride, 10 mM BaCl 2 , 1 mM MgCl 2 , 10 mM glucose, and 10 mM HEPES (pH 7.3). Ba 2ϩ was used as charge carrier to minimize Ca 2ϩ -induced current inactivation or activation of Ca 2ϩ -dependent signaling pathways. Whole-cell conductance and series resistance were compensated electronically, and evoked currents were filtered at 5 kHz. All currents were leak subtracted (P/4 protocol) before analysis.
To study the properties of voltage-dependent steady-state inactivation, a three-pulse voltage protocol was used. Two 20-ms test pulses to 0 mV from a Ϫ100-mV holding potential were separated by a conditioning inactivating voltage step. The voltage dependence of inactivation was measured in response to 3-s conditioning steps that ranged from Ϫ110 to 0 mV. Application of each three-pulse protocol was separated by a 25-s period at the holding potential to allow recovery from voltagedependent inactivation. Voltage-dependent steady-state inactivation data were fitted with a Boltzmann function of the form where I/I 0 is the current normalized to maximum current, Aϱ and A o are the maximum and base of the curve, V h is the midpoint of voltage dependence of inactivation, and k is the slope factor. Curve fits to data were made using an iterative, non-linear, least-squares fitting algorithm provided in the IGOR PRO software.

Expression and Functional Analysis of Munc18-1 and
Syntaxin1A FRET Probes-HEK293-S3 cells were initially used to establish FRET measurements between CFP-Munc18-1 and cYFP-syntaxin1A. These cells allowed high transient transfection efficiency and protein overexpression, and they lack endogenous syntaxin1A and Munc18-1 proteins, which likely facilitated interactions between the expressed neuralspecific recombinant proteins. Also, HEK293-S3 cells lack neuronal proteins that modify the molecular mechanism of Ca 2ϩregulated exocytosis in neurons and may affect Munc18-1 interactions with syntaxin1A.
Initially, we tested whether the addition of CFP and cYFP tags to Munc18-1 and syntaxin1A, respectively, altered the subcellular targeting or function of these proteins. As shown in the confocal images of Fig 1A, CFP-Munc18-1 fluorescence localized predominantly to the cytosolic compartment without specific labeling of the nucleus or plasma membrane. Examination of the expression of a non-tagged Munc18-1 by immunocytochemistry using anti-Munc18 antibody revealed a subcellular distribution similar to that of CFP-Munc18-1. Expression of cYFP-syntaxin1A in the absence of Munc18-1 consistently resulted in labeling of the perinuclear membrane, numerous small punctate intensities of cytosolic fluorescence, and often a few sites of intense fluorescence likely representing protein compartmentalization within organelles such as the proteins in transiently transfected HEK293-S3 cells as visualized by confocal fluorescence microscopy. Expression of non-tagged Munc18-1 and syntaxin1A was detected using anti-Munc18 and anti-syntaxin1A monoclonal antibodies, respectively, and Alexa Fluor-488-labeled secondary antibody. Cells were fixed 24 -36 h after transfection and, for non-tagged proteins, processed for immunocytochemistry before imaging. For cotransfection conditions, the fluoroprotein signal or antigen immunoreactivity detected for each cell image is underlined. Immunofluorescence of Alexa-408 corresponding to anti-Munc18 is green; anti-syntaxin1A is red. B, effects of cYFP-syntaxin1A and CFP-Munc18-1 expression on voltage-dependent, steady-state inactivation properties of N-type calcium channels. Averaged normalized current (I/I o ) for control and each (co)expression condition is plotted against the conditioning prepulse potential. Coexpression conditions: •, control; OE, syntaxin1A; E, Munc18-1/syntaxin1A. Fits of the I/I o curves are as described in "Materials and Methods." C, averaged potential of halfinactivation of steady-state inactivation (Vh) for each condition. Values were determined from the fits of the curves. Numbers above columns indicate the number of observations. Significant differences are relative to the syntaxin1A condition. endoplasmic reticulum and Golgi. A similar pattern in subcellular distribution upon expression of non-tagged syntaxin1A was observed by immunocytochemical analysis. We next imaged syntaxin1A and Munc18-1 distribution when coexpressed either as fluorescent protein-tagged fusion proteins or as nontagged proteins. Confocal images of the cells (Fig. 1A) showed a dramatic difference in Munc18-1 and syntaxin1A subcellular distribution compared with that with expression of each protein alone. Upon coexpression, both the tagged and non-tagged syntaxin1A and Munc18-1 were efficiently trafficked to the plasma membrane region, where they demonstrated considerable overlap in labeling. This distribution of syntaxin1A is consistent with previous reports that established Munc18-1 as important for proper membrane targeting of syntaxin1A (19,49).
To further test whether normal functional interactions were retained in the fluorescent protein-tagged Munc18-1 and syntaxin1A proteins, the regulatory effect of these proteins on aspecificcellularfunctionwasdetermined.Regulationofvoltagedependent steady-state inactivation of N-type (␣ 1B , Ca V2.2 ) and P/Q-type (␣ 1A , Ca V2.1 ) voltage-gated calcium channels by syntaxin1A has been well documented (50,51). Syntaxin1A interaction with a domain in the 2-3-transmembrane linker of N-type Ca 2ϩ channels (termed the synprint site) promotes entry into this inactivated state, shifts the half-inactivation potential negatively by 10 -20 mV, and enhances strong tonic inhibition by G-protein ␤␥ subunits (52). These effects of syntaxin1A are reversed upon coexpression of Munc18-1 (53,54). Fig 1B compares the effects of expression of cYFP-syntaxin1A on steady-state inactivation of N-type calcium channels stably expressed in a HEK293-S3 cell line with cells that were not transfected or transfected with an empty parental plasmid. Expression of cYFP-syntaxin1A resulted in an averaged negative shift in half-inactivation potential of ϳ10 mV from control, without a change in averaged current density or current-voltage relations (data not shown). Coexpression of CFP-Munc18-1 with cYFP-syntaxin1A resulted in an elimination of the enhancing effect of cYFP-syntaxin1A on voltage-dependent steady-state inactivation. The addition of the fluoroprotein tags to the syntaxin1A and Munc18-1 proteins did not, therefore, alter their subcellular distribution, functional interactions, or capacity to regulate voltage-gated calcium channel activity.
Measurement of the Interaction of CFP-Munc18-1 and cYFP-Syntaxin1A in Intact HEK293-S3 Cells Using an Acceptor Photobleach FRET Technique-In FRET, the non-radiative transfer of donor energy to the acceptor quenches the emission of light from the donor (55). Thus, we initially examined whether the emission of light by the donor (CFP-Munc18-1) became more intense after photobleach of the acceptor (cYFP-syntaxin1A), when the proteins were coexpressed in HEK293-S3 cells. Fig 2A provides a  from the Zeiss Meta confocal microscope of the mixed emission spectra of the CFP-Munc18-1 donor and cYFP-syntaxin1A acceptor fluorophores (excitation, 458-nm line of argon laser) taken from cell images before and after localized photobleach for photobleached and non-photobleached regions of the cell. A selective increase in the peak emission intensity corresponding to the region of CFP emission was observed in the photobleached, and not in the non-photobleached, regions of the cell. These results are consistent with a FRET signal originating from the transfer of resonant energy from CFP to cYFP and the dequenching of this energy transfer upon photobleaching of the acceptor fluorophore. Fig 2B shows a representative set of spectrally unmixed images (CFP and cYFP) taken from HEK293-S3 cells before and after photobleach of the acceptor (514-nm line of argon laser). The CFP images show an increase in the donor emission (also pseudocolored intensity images) after photobleach that occurred only in the region of the cell exposed to the photobleach. In addition to interaction of CFP-Munc18-1 and cYFP-syntaxin1A at plasma membrane regions, cell images also frequently showed discrete regions of CFP and cYFP fluorescence within the cytosol. We, therefore, tested the hypothesis that these regions of CFP and cYFP fluorescence resulted from targeting of cYFP-syntaxin1A to the Golgi complex and its interaction with CFP-Munc18-1 because syntaxin1A is sorted for trafficking to the plasma membrane. Initially, we evaluated whether the CFP and cYFP fluorescence colocalized with immunofluorescence against a Golgi antigen (GS27 protein) and whether FRET occurred between CFP-Munc18-1 and cYFP-syntaxin1A within the Golgi region. To facilitate accumulation of fluorescent signal within the Golgi, imaging was performed on cells that were fixed 6 h after cotransfection. Confocal microscopy of these cells showed a predominant colocalization of immunofluorescence against GS27 protein with CFP and cYFP fluorescence (Fig. 3). At this early time point after transfection, few cells showed significant localization of CFP-Munc18-1 or cYFP-syntaxin1A at the plasma membrane. In addition, immunofluorescence against Golgi marker GS27 in control, nontransfected HEK293-S3 cells demonstrated a similar subcellular localization and immunofluorescent signal to that observed in the cells cotransfected with CFP-Munc18-1 and cYFP-syntaxin1A, albeit with slightly increased perinuclear labeling of the transfected cells. These data indicate that overexpression of Munc18-1 with syntaxin1A did not lead to significant disruption or reorganization of the Golgi complex. Importantly, the colocalization of CFP-Munc18-1 and cYFP-syntaxin1A at the Golgi region represented a direct bimolecular interaction as determined by measurement of FRET using the acceptor photobleach approach. Indeed, averaged relative FRET efficiency for this interaction (27.3 Ϯ 0.9%, Ϯ S.E., n ϭ 8) was not significantly different from that measured for plasma membrane regions after 24 -48 h of expression. These data demonstrate that Munc18-1 interacts with syntaxin1A at the region associated with the Golgi complex and suggests that the interaction between these proteins is maintained during trafficking and after delivery of syntaxin1A to the plasma membrane.
Measurement of CFP-Munc18-1 and cYFP-Syntaxin1A FRET by Sensitized Emission-As an alternative and independent measure of FRET between CFP-Munc18-1 and cYFP-syntaxin1A, we used a filter-based approach with a conventional fluorescence microscope. This approach measured FRET as an increase in acceptor fluorescence resulting from donor excitation (i.e. sensitized emission FRET). Three FRET parameters were quantitatively determined: E A , E D , and RATIO (48). E A is the apparent efficiency of acceptor interaction with donor, whereas E D is the apparent efficiency of donor interaction with acceptor. RATIO is the molar ratio of cYFP to CFP.  1%, n ϭ 27). An additional experiment evaluated FRET between Munc18-1 and syntaxin1A that were linked to nondimerizing mutant forms of CFP and cYFP fluoroproteins (A206K mutants, (56)), respectively, as green fluorescent proteins, and their variants may self-aggregate and thus FRET. Syntaxin1A distribution and FRET values were nearly identical to those observed with the non-monomeric CFP and cYFP fusion proteins. By comparison, a negative control involving coexpression of CFP ϩ cYFP-syntaxin1A showed no specific FRET signal (E A , 2.7 Ϯ 0.5%; E D , 1.2 Ϯ 0.1%, n ϭ 14), whereas a positive control with the linked CFP-cYFP construct demonstrated spatially uniform FRET throughout the cell cytoplasm. The averaged FRET efficiency values E A and E D for all treatments are shown in Fig.  4B. Differences in the FRET values between treatments occurred without substantial variations in averaged molar RA-TIO values (Fig 4C).
CFP-Munc18-1/cYFP-Syntaxin1A FRET Is Sensitive to Affinity of Protein Interaction-To determine whether FRET between CFP-Munc18-1 and cYFP-syntaxin1A was responsive to changes in the level of interaction of these proteins, we measured the effects of mutants of Munc18-1 and syntaxin1A. Initially, the effect on FRET was determined using CFP-Munc18-1 and a cYFP-tagged mutant of syntaxin1A (L165A, E166A) that assumes a constitutively open conformation (syntaxin1A "open" mutant) and that has been reported not to bind to Munc18-1 in vitro (28). The results demonstrated a strongly reduced apparent FRET efficiency (E A and E D ) as compared with the wild-type protein pair (Fig. 5). E A was reduced from 31.6 Ϯ 0.9% to 13.1 Ϯ 1.0%, whereas E D decreased from 27.0 Ϯ 1.1% to 8.9 Ϯ 0.6%. Images of cYFP-syntaxin1A for the syntaxin1A "open" mutant also showed localized densities of cYFP fluorescence with no apparent increased signal at the plasma membrane. It is important to note, however, that FRET between the syntaxin1A "open" mutant and Munc18-1 was significantly above the background value (2.7 Ϯ 0.5%, from CFP ϩ cYFP-syntaxin1A treatment), indicating that the syntaxin1A "open" mutant retained a capacity to interact with Munc18-1 in vivo.
As a complementary experiment, we determined the effect on apparent FRET efficiency of a site-directed mutant of Munc18-1 (R39C) that exhibits strongly reduced affinity for interaction with syntaxin1A in vitro (57). FRET images from cells coexpressing the CFP-tagged version of this mutant with cYFP-syntaxin1A (Fig. 5) demonstrated that expression of the Munc18-1 (R39C) mutant resulted in a significant reduction in FRET (E A , 15.3 Ϯ 1.3%; E D , 25.2 Ϯ 1.0; n ϭ 27) as compared with wild-type proteins. However, the reduction in FRET is less than that observed with coexpression of cYFP-syntaxin "open" mutant ϩ CFP-Munc18-1. Comparison of cYFP and the RATIO images between treatments indicates that the amount of syntaxin1A at the plasma membrane region with Munc18-1 (R39C) expression is less than that with coexpression of wildtype proteins but greater than with the syntaxin1A "open" condition. Notably, coexpression of both syntaxin1A "open" and Munc18-1 (R39C) mutants resulted in a lower apparent FRET efficiency (E A of 7.0 Ϯ 0.5%; E D of 7.8 Ϯ 0.2%; n ϭ 11) than when only one of the FRET pair was mutated. These results further substantiate that the syntaxin1A "open" mutant can bind to Munc18-1 in vivo.
To evaluate the effect of mutations directly within the SNARE motif of syntaxin1A on apparent FRET efficiency in vivo, we examined two site-directed mutants of cYFP-syntaxin1A, i.e. cYFP-syntaxin1A (I233A) and cYFP-syntaxin1A (I209A). The I233A mutation of syntaxin1A rests within the N-terminal domain of the SNARE motif that normally occupies the central cavity of Munc18-1 during interaction of Munc18-1 with syntaxin1A. Previously reported in vitro binding assays demonstrated that this mutation reduces syntaxin1A/Munc18-1 binding without reduction in syntaxin1A binding to its cognate t-SNARE SNAP25 (58). As shown in Fig. 5, measurement of FRET between CFP-Munc18-1 and cYFP-syntaxin1A (I233A) showed that apparent FRET efficiency (E A , 9.9 Ϯ 0.5%; E D , 15.8 Ϯ 1.1%) was reduced from that of the control CFP-Munc18-1 and cYFP-syntaxin1A proteins. The I209A mutation of syntaxin1A is also within the SNARE motif of syntaxin1A, but as a result of its placement, this mutant has been reported to retain full capacity for Munc18-1 binding while reducing binding to SNAP25 (59). Coexpression of cYFP-Munc18-1 with a CFP-syntaxin1A (I209A) mutant resulted in apparent FRET efficiency values (E A , 31.9 Ϯ 1.5%; E D , 34.2 Ϯ 1.8%) that were not significantly different from control. Taken together, the data demonstrated that FRET is sensitive to expected differences in interaction between Munc18-1 and syntaxin1A.
Relationship of Apparent FRET Efficiency to RATIO Allows Unambiguous Comparison of FRET between Treatments-In the present study, determination of the molar ratio (cYFP/CFP)  Fig. 6B. Comparison of the relationships between the wild-type FRET protein pair and CFP-Munc18-1 ϩ cYFP-syntaxin1A "open" showed a marked decrease in apparent FRET efficiency throughout the range of RATIO values. In contrast, the data for the wild-type FRET protein pair showed considerable overlap with that for CFP-Munc18-1 (R39C) ϩ cYFP-syntaxin1A. Importantly, however, in each case apparent FRET efficiency was sensitive to the relative expression levels of the acceptor and donor proteins, and the apparent FRET efficiency/RATIO relationship resembled that of a single-site binding curve. Therefore, to calculate a value for E A and E D that can be unambiguously compared between treatments, determination of averaged FRET efficiency was restricted to include only those FRET efficiency values where RATIO fell within the range 0.9 to 1.1. This ratio was chosen based on an established 1:1 stoichiometry for Munc18-1 and syntaxin1A interaction. Fig. 6C shows the E A and E D values calculated over this restricted ratio range for each of the treatments. The results demonstrated the expected correspondence between the E A and E D values when the molar ratio approaches 1. In addition, the results confirmed a strong reduc-tion in FRET signal with the syntaxin1A "open" mutant (67%) and an ϳ30% reduction with the Munc18-1 (R39C) mutant from control.
Effects of Mutations of Munc18-1 Phosphorylation Sites on CFP-Munc18-1/cYFP-Syntaxin1A FRET-To determine the effects of Munc18-1 phosphorylation by PKC and Cdk5 on Munc18-1/syntaxin1A interaction in intact cells, we measured FRET using phosphomimetic mutants of the PKC and Cdk5 sites of Munc18-1. Fig. 7A compares averaged E A values in the 0.9 -1.1 RATIO interval between the control CFP-Munc18-1 ϩ cYFP-syntaxin1A FRET pair and a series of phosphomimetic mutants of CFP-Munc18-1 coexpressed with cYFP-syntaxin1A. To better mimic electrostatics of a PO 4 2Ϫ group at Ser-313, the effects of double dGlu mutations at Ser-312 and Ser-313 were also tested. Results showed that the PKC phosphomimetic mutants demonstrated the strongest reductions in FRET when expressed as double Ser-312-Ser-306 or triple Ser-306-Ser-312-Ser-313 dGlu mutants. The Ser-306-Ser-313 double dGlu mutation showed no statistically significant change in apparent FRET efficiency values from control, demonstrating that within the cell environment this phosphomimetic mutant is capable of interacting with syntaxin1A. This suggests that effects of the Ser-306-Ser-313 phosphomimetic mutant on secretory responsiveness result from actions independent of Munc18-1/syntaxin1A binding. Coexpression of Munc18-1 mutants carrying a single dGlu mutation at either the Ser-306 or Ser-313 site also had no significant effect on the measured FRET. The Cdk5 phosphomimetic mutant demonstrated a reduction in FRET. We next examined whether phosphorylation altered plasma membrane targeting of syntaxin1A by analysis of subcellular distribution of cYFP. As shown in Fig. 7B, coexpression of the control CFP-Munc18-1 ϩ cYFP-syntaxin1A resulted in a pronounced targeting of cYFP to the plasma membrane regions. By comparison, expression of Munc18-1 with either the PKC phosphorylation site mutations (Ser-306-Ser-313 or the Ser-306-Ser-312-Ser-313) or the Cdk5 phosphorylation site mutation (Thr-574) resulted in restriction of coexpressed syntaxin1A to perinuclear regions and areas of intense fluorescence within the cytoplasm that likely reflect accumulation in Golgi and endoplasmic reticulum compartments. Averaged results quantifying the effect on trafficking of cYFP-syntaxin1A to the plasma membrane for each of the mutants of Munc18-1 are shown in Fig 7C. None of the single dGlu mutations of the PKC sites differed from control in targeting cYFP-syntaxin1A. To verify that the Munc18-1 and its mutant constructs were expressed at similar levels relative to syntaxin1A, the averaged molar cYFP/CFP ratios were compared across treatments (Fig 7D). The results showed similar expression ratios, indicating that differences in FRET efficiency or syntaxin1A targeting across treatments did not result from changes in relative expression of the proteins.
Imaging of FRET between CFP-Munc18-1 and cYFP-Syntaxin1A in Adrenal Chromaffin Cells-To examine the properties and regulation of Munc18-1 and syntaxin1A in a cell environment where these proteins are endogenous and where components normally regulating their interaction in the Ca 2ϩdependent exocytotic pathway are functionally active, we used neuroendocrine adrenal chromaffin cells. A comparison of the subcellular distribution of CFP-Munc18-1 and cYFP-syntaxin1A when expressed alone or in combination in chromaffin cells is shown by confocal microscopy in Fig 8A. Unlike expression in HEK293-S3 cells, the expression of CFP-Munc18-1 in chromaffin cells resulted in specific enrichment of fluorescence intensity at the plasma membrane regions, although the fluorescent signal was also present throughout the cytosol. The presence of endogenous syntaxin1A in the plasma membrane likely promoted the targeting and association of Munc18-1 to this domain. In contrast, overexpression of cYFP-syntaxin1A in chromaffin cells resulted in intense fluorescent labeling of the perinuclear region, an effect similar to that observed for cYFP-syntaxin1A expression in HEK293-S3 cells. Because Munc18-1 is critical for trafficking of syntaxin1A, these results suggest that Munc18-1 may be a limiting factor for syntaxin1A trafficking in chromaffin cells. Coexpression of CFP-Munc18-1 and cYFP-syntaxin1A resulted in strong fluorescent labeling at or near the plasma membrane without accumulation of cYFP-syntaxin1A at the perinuclear region. Overlap in the subcellular distribution between CFP-Munc18-1 and cYFP-syntaxin1A occurred as expected. To confirm that the overlap in the distribution reflected a true bimolecular interaction, we first evaluated FRET between the CFP-Munc18-1 and cYFP-syntaxin1A using the acceptor photobleach method. Fig 8B shows a representative set of spectrally unmixed images (CFP and cYFP) taken from transfected and fixed adrenal chromaffin cells before and after photobleach of the acceptor. With photobleaching, donor emission increased only in the region of the cell exposed to the photobleach. Fig. 8C compares the mixed emission spectra of CFP-Munc18-1 and cYFP-syntaxin1A taken before and after photobleach for bleached and non-bleached regions of the cell. Consistent with the incidence of FRET, the peak of CFP emission intensity was selectively increased in the photobleached region of the cell. The averaged relative FRET efficiency was 36.8 Ϯ 2.2% (n ϭ 14). As a final test for FRET, we imaged transfected, living chromaffin cells and measured FRET by sensitized emission. 0.9 to 1.1). Taken together, the results showed that FRET between CFP-Munc18-1 and cYFP-syntaxin1A can be quantified in neuroendocrine chromaffin cells and, potentially, can be used to evaluate dynamic changes in the interactions of these proteins.

DISCUSSION
In addition to the cyclical formation and disassembly of SNARE protein complexes, the pathways regulating membrane fusion and exocytosis require the involvement of SM proteins. Despite the critical importance of SM proteins and the existence of several proposed models that indicate a critical role in the regulation of protein-protein interactions, the exact molecular mechanism by which SM proteins act to regulate the exocytotic pathway remains enigmatic. Although biochemical properties of the syntaxin1A/Munc18-1 interaction have been characterized and the crystal structure of the heterodimer complex has been determined (22), little is known of the temporal and spatial aspects of the interaction or of its regulation in intact cells. The investigations of the present study sought unambiguous evidence that FRET could be used to visualize and quantify temporal and spatial aspects of the bimolecular Munc18-1/syntaxin1A interaction in intact cells. FRET between donor and acceptor labeled molecules results in several measurable parameters, including sensitized acceptor fluorescence, quenching of donor fluorescence, and a decrease in donor lifetime. In the present study, we have measured FRET between N-terminal fusions of CFP and cYFP to Munc18-1 and syntaxin1A, respectively, in intact HEK293-S3 and adrenal chromaffin cells using both a sensitized acceptor fluorescence stoichiometry method and by dequenching of donor fluorescence when the acceptor fluorophore was photobleached. These independent measures of FRET resulted in highly complementary results and unambiguously validated the signal measured in both cell types as FRET. The measured FRET efficiency was comparable with that obtained on expression of a directly linked CFP-cYFP fusion protein that assumes a FRET configuration.
The fact that FRET can occur is not necessarily reflective of specific protein-protein interactions. For example, the propensity of the green fluorescent protein family of proteins to directly dimerize or oligomerize when overexpressed may result in FRET (56,60). In addition, FRET can occur as a result of random proximity effects that arise from high-density compartmentalization of donors and acceptors into specific subcellular domains (61,62). In the present study, we found no significant differences in the apparent FRET efficiency values between experiments using the normal enhanced versions of the CFP and cYFP fluorophores and nondimerizing mutants of fluoroproteins (A206K mutation, (56)). Also, when expressed alone CFP-Munc18-1 appeared as a soluble protein localized within the cytosol, thereby making it highly unlikely that the FRET signal with membrane delimited cYFP-syntaxin1A resulted from nonspecific proximity interactions. Consistent with in vitro binding data, mutants of Munc18-1 and syntaxin1A that exhibit reduced binding interactions also demonstrated significant reductions in apparent FRET efficiency values, even when determined over equivalent cYFP/CFP ratios. Notably, however, a syntaxin1A mutant (L165A, E166A) that is con- strained into a constitutively open conformation and that fails to exhibit measurable binding interactions in vitro (28) retained the capacity for interaction when coexpressed with CFP-Munc18-1 within intact cells. These latter results demonstrate that in vitro binding data between Munc18-1 and mutants of syntaxin1A may quantitatively differ from what occurs within the cellular environment and suggest caution using in vitro binding data to interpret secretory effects resulting from expression of Munc18-1 or syntaxin1A mutants in eukaryotic cells and extrapolation of the interpretation to mechanistic models of the secretory pathway.
Determination of apparent FRET efficiency for each pixel of an image allows a subcellular mapping of energy transfer efficiency for a cell. However, because FRET between donor and acceptor fluorophore-tagged proteins within cells results from a bimolecular binding interaction, the measured FRET efficiency can vary from cell to cell and between experiments based on differences in the stoichiometry of expression of the donor and acceptor labeled proteins and on the degree of saturation of the binding reaction for the interacting proteins. These variables have often resulted in the determination of relative rather than absolute FRET efficiency values, thereby hindering direct and quantitative comparisons of FRET efficiency between cells and treatment groups. The present investigations implemented a recently developed FRET stoichiometry approach that improves the quantification of FRET in living cells to determine the molar concentration ratios (acceptor/donor) and fractions of interacting molecules throughout a cell image (48,63). Indeed, a hyperbolic relationship was found between apparent FRET efficiency and the ratio of acceptor to donor-labeled protein (Fig. 6). That is, the apparent efficiency E D was a saturable function of molar ratio (cYFP/CFP), approaching zero as the ratio approaches zero and saturating with increasing molar ratio values as the donor becomes saturated with acceptor. E A was a similar function of the inverse of the molar ratio. Using this information, it was possible to directly compare apparent FRET efficiency (E A , E D ) between cells and across treatments at given molar ratios. Yet, E A and E D are defined as the product of characteristic efficiency of the interaction (E C ) and the fraction of acceptor (f A ) or donor (f D ) in complex. Therefore, changes in E A and E D between treatments could result from both changes in E C and/or f A and f D , even when considered at identical molar ratios. However, an estimation of E C is given by the level at which the E A versus 1/(molar ratio) or E D versus molar ratio relationships saturate. By this estimate, the reduction in apparent efficiency for the syntaxin1A/Munc18 (R39C) treatment from control likely primarily resulted from a change in the fraction of molecules in complex, because the E C values for both conditions were similar. By comparison, strongly reduced FRET efficiencies for syntaxin1A (L165A/E166A) and syntaxin1A (I233A) conditions from control were accompanied by corresponding decreases in E C , indicating that the change in the fraction of molecules in complex with these mutants may not be as great as that for the Munc18 R39C mutant condition. Recently reported effects of these syntaxin1A mutants to increase quantal size and slow kinetics of release from transfected adrenal chromaffin cells (64) are consistent with the observed reduction in apparent FRET efficiency. It should also be noted that at a molar RATIO of 1, the averaged E D and E A values for each treatment were not statistically different, reflecting the 1:1 binding stoichiometry between Munc18-1 and syntaxin1A and arguing against significant homo-oligomerization via the H3 domain SNARE motifs of the expressed cYFP-syntaxin1A, when coexpressed with CFP-Munc18-1.
In this study, we have also used spatial information contributed by imaging of FRET efficiency to map subcellular regions of CFP-Munc18-1/cYFP-syntaxin1A interactions. The confocal images of FRET efficiency demonstrated prominent interac- Pseudocolored intensity images at the bottom show a region of cell membrane (above diagonal white line) before and after photobleaching. C, representative example of mixed emission spectra of CFP-Munc18-1 donor and cYFP-syntaxin1A acceptor fluorophores taken before (black line) and after (red line) photobleaching. Spectra are shown for a region that was photobleached (upper plot) and for another region in the same cell that was not photobleached (lower plot). D, representative image set comparing cYFP-syntaxin1A distribution (cYFP-S1A), apparent FRET efficiency (EA, ED), and molar ratio (Ratio) for cells cotransfected with CFP-Munc18-1 ϩ cYFP-syntaxin1A. tions at the plasma membrane regions for both cotransfected HEK293-S3 and chromaffin cells. This contrasted strongly with a FRET efficiency signal approaching background for most of the cytoplasm, with the exception of occasional densities of intense FRET that are likely related to the Golgi and/or membrane trafficking vesicles in the cytoplasm. The resolution of subcellular and regional differences was, as expected, lower in apparent FRET efficiency images using sensitized emission methods on a conventional fluorescence microscope as a result of effects of out-of-field focus fluorescence combined with FRET calculations based in part on a ratio of fluorescence intensities. However, even FRET efficiency images from sensitized emission showed localized regions within the cytosol, and particularly at the plasma membrane, of heightened FRET efficiency. We have also demonstrated, with shorter periods of expression after transfection, that FRET between CFP-Munc18-1 and cYFP-syntaxin1A colocalized with immunoreactivity against GS27, a Golgi marker protein. In addition, FRET efficiency was similar at the Golgi as that seen in the plasma membrane region. The importance of Munc18-1 in trafficking syntaxin1A to the plasma membrane has been previously well documented (19,49). Our observations suggest that interactions between Munc18-1 and syntaxin1A that form at the Golgi complex are maintained at nearly equivalent levels during transit and delivery to the plasma membrane.
By itself, syntaxin1A switches between a closed, SNARE pairing, inactive configuration and an open, SNARE pairing competent configuration with Ͻ1 ms of relaxation time (32). Therefore, regulatory proteins such as Munc18 have been proposed as necessary to stabilize syntaxin1A in a conformation, consistent with the activation state of the secretory mechanism (27)(28)(29)65). Knock-out of Munc18-1 gene expression in mice (33) and loss-of-function mutations for SM proteins in C. elegans (13,21) and Drosophila (11,12) result in an elimination or strong reduction in evoked exocytotic responses that underlie synaptic neurotransmission. Although an interaction between syntaxin1A and Munc18-1 is believed essential to regulated exocytosis, there are studies indicating that each protein may subserve different roles and exert functions at different steps in the secretory pathway. For example, Munc18-1 function is critical to docking of secretory vesicles in C. elegans neurons (21) and in bovine adrenal chromaffin cells (20). By comparison, botulinum toxin-mediated ablation of syntaxin1A did not disrupt vesicle docking (66, 67) (except (68)), yet regulated secretion was strongly reduced (69,70), consistent with a role of syntaxin1A as a SNARE. Thus, Munc18 and syntaxin1A may have multiple roles in the secretory process, with certain functions requiring direct interactions between these proteins, such as syntaxin1A trafficking to the plasma membrane, whereas others, such as vesicle docking, may require independent action.
On the basis of binding studies, interactions between Munc18-1 and syntaxin1A are regulated by PKC and Cdk5 phosphorylation of Munc18. Phosphorylation by PKC inhibits Munc18-1 binding to syntaxin1A in vitro (34) and in rat brain nerve terminals in situ (36). Phosphorylation by Cdk5 has likewise been reported to inhibit the Munc18-1/syntaxin1A interaction in vitro, and pharmacological block of Cdk5 activity reduced secretory responsiveness in adrenal chromaffin cells (35,38). Moreover, although a high constitutive phosphatase activity in rat brain nerve terminals was reported to keep the phosphorylation of Munc18-1 low in a basal state, depolarizing stimuli were found to result in rapid PKC and Cdk5 phosphorylation of Munc18-1 (36). The proposed result of both PKC and Cdk5 phosphorylation of Munc18-1 has been to enhance equilibrium syntaxin1A pairing with cognate SNAREs to facilitate secretory responses. Our studies used dGlu phosphomimetic mutants of Munc18-1 at Ser-306 and Ser-313 (24), the two major PKC phosphorylation sites on Munc18-1 (34), with FRET to further define the in vivo properties of this regulation. The results showed that single dGlu substitutions at Ser-306, Ser-312, or Ser-313 resulted in no significant change in apparent FRET efficiency or in the trafficking of syntaxin1A to the plasma membrane. On the other hand, the double dGlu mutation at Ser-312 and Ser-313, which may better mimic the negative charges of a PO 4 2Ϫ residue occurring by PKC phosphorylation at Ser-313, resulted in significant reductions in both apparent FRET efficiency and in the trafficking of syntaxin1A to the plasma membrane region. The data also suggest that PKC-mediated regulation of the Munc18-1/syntaxin1A interaction and rapid effects of PKC on secretion may be best mimicked by this double dGlu mutation, because the dGlu mutation of Ser-306 alone resulted in no significant effect on FRET efficiency relative to control, and when Ser-306 was mutated together with Ser-312 and Ser-313, the extent of inhibition was not further increased. Moreover, the double Ser-306-Ser-313 mutation affected only targeting of syntaxin1A to the plasma membrane and, therefore, would not result in rapid regulation of the secretory mechanism. By comparison, the single dGlu substitution at a Cdk5 site Thr-574 reduced FRET efficiency and syntaxin1A targeting, supporting acute regulation of secretory responsiveness. Spatially averaged cYFP/CFP RATIO values were similar across all dGlu treatments, indicating that effects did not result from relative protein expression differences. Our findings for PKC are consistent with the lack of a measurable effect of single and of significant effects of double dGlu mutations on single-vesicle release kinetics (24). Our results also are in support of Munc18-1 phosphorylation as one component of a long-established enhancement of exocytosis by PKC in adrenal chromaffin cells. In addition to the short-term positive effects of Munc18 phosphorylation on exocytosis, our data indicate that sustained phosphorylation would ultimately become detrimental to the exocytotic pathway as a result of restricted trafficking of syntaxin1A to sites of exocytosis at the plasma membrane. The constitutively high phosphatase activity reported in rat brain nerve terminals is, therefore, likely to be critical not only for limiting rapid effects of Munc18-1 on exocytosis but also in allowing continued trafficking of syntaxin1A to the plasma membrane. This is critical as the Munc18-1/syntaxin1A interaction is also important in secretory granule docking (20,21) and may also facilitate priming processes required for assembly of SNARE complexes (8,71).
In vivo FRET analysis of Munc18-1/syntaxin1A interactions should promote further understanding of the precise timing at which these complexes assemble and function within the regulated exocytotic pathway in neurons and neuroendocrine cells.