To whom correspondence should be addressed: Dept. of Biomolecular Chemistry, University of Wisconsin-Madison School of Medicine and Public Health, 440 Henry Mall, Madison, WI 53706. Tel.: 608-262-3761; Fax: 608-262-5253;
* This work was supported in part by National Institutes of Health Grants GM110567 and GM088151 (to A. A.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. ♦ This article was selected as a Paper of the Week. This article contains supplemental Videos S1–S10. 1 Supported by Grant BB/J018236/1 from the Biotechnology and Biological Sciences Research Council. 2 Investigator of the Howard Hughes Medical Institute.
The majority of biosynthetic secretory proteins initiate their journey through the endomembrane system from specific subdomains of the endoplasmic reticulum. At these locations, coated transport carriers are generated, with the Sar1 GTPase playing a critical role in membrane bending, recruitment of coat components, and nascent vesicle formation. How these events are appropriately coordinated remains poorly understood. Here, we demonstrate that Sar1 acts as the curvature-sensing component of the COPII coat complex and highlight the ability of Sar1 to bind more avidly to membranes of high curvature. Additionally, using an atomic force microscopy-based approach, we further show that the intrinsic GTPase activity of Sar1 is necessary for remodeling lipid bilayers. Consistent with this idea, Sar1-mediated membrane remodeling is dramatically accelerated in the presence of its guanine nucleotide-activating protein (GAP), Sec23-Sec24, and blocked upon addition of guanosine-5′-[(β,γ)-imido]triphosphate, a poorly hydrolysable analog of GTP. Our results also indicate that Sar1 GTPase activity is stimulated by membranes that exhibit elevated curvature, potentially enabling Sar1 membrane scission activity to be spatially restricted to highly bent membranes that are characteristic of a bud neck. Taken together, our data support a stepwise model in which the amino-terminal amphipathic helix of GTP-bound Sar1 stably penetrates the endoplasmic reticulum membrane, promoting local membrane deformation. As membrane bending increases, Sar1 membrane binding is elevated, ultimately culminating in GTP hydrolysis, which may destabilize the bilayer sufficiently to facilitate membrane fission.
lumen and subsequently packaged into coated vesicles or tubules for export. Components of the cytosolic coat protein II (COPII) complex play key roles in ER membrane remodeling and scission to bud transport carriers (
). In particular, the conserved Sar1 GTPase has been implicated in multiple steps of this process. Current models suggest that in the presence of its guanine nucleotide exchange factor (GEF) Sec12, which is enriched at subdomains of the ER that synthesize COPII-coated carriers, GTP-bound Sar1 exposes an amphipathic helix that penetrates the outer leaflet of the membrane (
). Recent studies have highlighted the flexibility exhibited by the inner and outer COPII layers, enabling the formation of small vesicles (∼50–100 nm in diameter) and elongated tubules (∼300–400 nm in length), which can capture and traffic both small and large cargoes, respectively, away from the ER (
). However, mechanisms governing the location and timing of carrier scission have not been clearly defined. In particular, how GTP hydrolysis on Sar1 contributes to this process continues to be debated.
In cell-free assays conducted more than 2 decades ago, poorly hydrolysable analogs of GTP were shown to be sufficient to promote COPII-mediated vesicle formation and release, suggesting that GTP hydrolysis was dispensable for the budding reaction (
). However, in both cases, Sar1-dependent vesicle release was measured following differential centrifugation and/or prolonged periods of incubation with COPII components and nucleotide analogs, which may have led to the unintentional detachment of nascent transport carriers from ER membranes. Using synthetic giant unilamellar vesicles (GUVs), another report showed that Sar1 alone was capable of deforming membranes and generating small vesicles independently of GTP hydrolysis (
). However, a high concentration of Sar1 (>7 μm) was necessary for this effect, which may not be physiologically relevant in the context of COPII-mediated budding in vivo. Moreover, the electron microscopy (EM)-based methods employed in this study required mechanical perturbations, including fixation and dehydration in negative staining preparations and liquid blotting in cryogenic approaches, which can perturb samples sufficiently to induce artifacts (
), raising the possibility that nucleotide hydrolysis contributes to the membrane fission reaction necessary to detach transport carriers. Although difficult to reconcile, these conflicting reports highlight the need for minimally invasive techniques to study the role of Sar1 in membrane remodeling, while still maintaining sufficiently high spatial and temporal resolution to interrogate the process.
Another consistent observation from these studies indicates that Sar1 is capable of associating with lipid bilayers that exhibit a range of curvatures, in a manner dependent on the amino-terminal amphipathic helix of Sar1. Specifically, Sar1 has been shown to bind to both highly curved liposomes (nanometers in diameter) and flat GUVs (microns in diameter) (
). Depending on the size of the tubule, Sar1 can form structurally unique scaffolds, varying from ordered lattices on membranes possessing shallow curvature to fragmented arrays on bilayers of higher curvature (
). The difference in its membrane binding properties may be related to its dual role in the formation and budding of COPII-coated transport carriers. However, apart from structural studies and largely qualitative measurements of bilayer association, the biophysical characteristics of Sar1 membrane binding have not been clearly defined. Additionally, it remains unknown how membrane curvature impacts GTP hydrolysis on Sar1. Here, we investigate these properties using Sar1 isoforms isolated from three distinct species. Our data demonstrate that membrane curvature strongly influences both the affinity of Sar1 for membranes and its rate of GTP hydrolysis in an evolutionarily conserved manner. Collectively, our data are most consistent with a model in which Sar1 promotes membrane scission at bud necks in a curvature-dependent process that involves GTP hydrolysis.
Caenorhabditis elegans Growth, Maintenance, RNA Interference, and Live Imaging
All C. elegans strains used in this study were derived from the Bristol strain N2, which was described previously (
). Double-stranded RNA (dsRNA) was synthesized from templates prepared by PCR to amplify C. elegans genomic DNA. For RNAi experiments, early L4 stage hermaphrodites were soaked in dsRNA for 24 h at 20 °C within a humidified chamber. Animals were then allowed to recover for 24–48 h before analyzing them for embryo production or mounting onto a 10% agarose pad in a 4-μl suspension of polystyrene beads to immobilize them for imaging (
). Images were acquired on a swept-field confocal microscope (Nikon Ti-E), using a Nikon ×60, 1.4 numerical aperture Planapo oil objective lens and a Roper CoolSnap HQ2 CCD camera. Acquisition parameters were controlled by Nikon Elements software, and image analysis was conducted using Metamorph software.
Recombinant Protein Expression, Purification, and Mass Determination
All C. elegans COPII components were amplified from a C. elegans cDNA library, and Sanger sequencing was used to confirm their identity. Proteins were expressed as His6-SUMO fusions and purified using nickel-nitrilotriacetic acid-agarose resin in Sar1 buffer (25 mm HEPES, pH 7.2, 100 mm NaCl, and 1 mm MgCl2). Sumo protease was used to remove the His6-Sumo tag, and the cleaved proteins were subjected to size-exclusion chromatography, which was coupled to a Wyatt mini-DAWN TREOS three-angle light scattering detector and a Wyatt Optilab T-rEX refractive index detector. Data were collected at a flow rate of 0.5 ml/min and analyzed using ASTRA software to determine molecular mass (
). For C. elegans SAR-1, protein purification was carried out in the presence of either GDP or GTP (500 μm each). Specifically, a 30-fold molar excess of nucleotide was present during Sumo protease cleavage (16 h at 4 °C) to ensure complete incorporation onto SAR-1, as described previously (
). Furthermore, nucleotide-bound SAR-1 was additionally gel-filtered in the presence of the appropriate nucleotide (500 μm) prior to use in all assays.C. elegans SEC-23/SEC-24.2 and Saccharomyces cerevisiae Sec23p-Sec24p complexes were purified similarly, with the exception of the buffer used (25 mm HEPES, pH 7.2, 160 mm KOAc, and 1 mm MgCl2). Human and yeast forms of Sar1 were purified as described previously (
). When S. cerevisiae proteins were utilized, 20% ergosterol was used in place of cholesterol. Lipids were dried, resuspended in Sar1 buffer, and subjected to extrusion through a nitrocellulose filter (Whatman) of a desired pore size. Liposomes were analyzed by dynamic light scattering using a Wyatt DynaPro NanoStar to determine their average diameters.
Stopped-flow Rapid Kinetic Analysis and Co-sedimentation Analysis
Stopped-flow rapid kinetic analysis and co-sedimentation experiments were carried out as described previously (
). Briefly, SAR-1 (8 μm) was incubated with GTP (100 μm) in Sar1 buffer before being injected into the stopped-flow chamber. An equal volume of Sar1 buffer containing a desired liposome dilution was simultaneously injected to initiate SAR-1 binding. Liposomes contained the following phospholipid composition (lipid, mol %): DOPC, 71; DOPE, 15; DOPS, 8; DOPA, 5; dansyl-PE, 1, and were supplemented with 20% cholesterol (major/minor mix lacking phosphoinositides and DAG). For co-sedimentation analysis, SAR-1 (4 μm) was incubated with 1 mm GMP-PNP in Sar1 buffer and supplemented with a desired liposome concentration. Liposomes were composed of 66% DOPC, 21% DOPE, 8% DOPS, 5% DOPA, and supplemented with 20% cholesterol. Samples were mixed in a TLA100 tube for 15 min at room temperature and then centrifuged at 100,000 × g for 45 min. The supernatant was collected, mixed with sample buffer, and resolved on an SDS-polyacrylamide gel for densitometry to determine SAR-1 depletion.
Sar1 GTPase Activity Assays
The GTPase/GAP/GEF-Glo system (Promega) was used to measure the concentration of GTP remaining after incubation with Sar1 isoforms (
). Sar1 (3.1 μm, unless otherwise noted) was added to the reaction before the addition of GTP (5 μm) in a 25-μl reaction (Corning Costar 3912 solid white 96-well plate) in the supplied GEF buffer containing MgCl2. The addition of GTP initiated the reaction, which was conducted at 25 °C. After 2 h, an equal volume of the GTPase-Glo reagent was added to stop the reaction, incubated for 30 min, and followed by the addition of an equal volume of the detection reagent for 10 min. Luminescence was measured on a 96-well plate reader. All reactions were performed in triplicate and repeated independently on three separate occasions. Reactions with S. cerevisiae proteins were performed similarly, except that the initial reactions were carried out at 30 °C. Malachite Green GTPase assays were conducted by first incubating SAR-1GDP with liposomes (1.27 mm) of a specific diameter for 10 min at room temperature. The GTPase reaction was initiated with the addition of GTP (300 μm) to a final reaction volume of 20 μl. In some cases, recombinant Sec23-Sec24 (10 μm, unless otherwise noted) was also present in the reaction prior to GTP addition. The reaction was allowed to continue for 1 h at room temperature. Malachite Green reagent (w/v %: Malachite Green, 0.027; poly(vinyl alcohol), 0.773; ammonium molybdate, 1.01) was added to the reaction to a final volume of 440 μl and allowed to incubate at room temperature for 2 h for the reaction with free inorganic phosphate to proceed to completion (indicated by a colorimetric change). After incubation with the Malachite Green reagent, the reaction was quenched with a sodium citrate solution (w/v %: 34; 50 μl) and incubated at room temperature for 5 min before absorbance measurements were taken (636 nm).
Fluorescence Quenching Assays
SAR1BW86F,W192F (SAR1B Trp reporter) was engineered using site-directed mutagenesis and purified similarly to the wild type protein. Acrylamide quenching measurements were carried out as described previously (
). Briefly, a 2.0 m stock of acrylamide was dissolved in Sar1 buffer (with the addition of 1 mm DTT). The aqueous acrylamide quencher was added to the SAR1B-Trp reporter (1 μm) and supplemented with either 1 mm GDP or 1 mm GTP in the presence or absence of 2 mm lipids (mol %: DOPC, 75; DOPE, 12; DOPS, 8; DOPA, 5) as indicated. Emission spectra were subtracted for blank, lipids, nucleotide, and acrylamide and integrated from 300 to 400 nm to calculate fluorescence intensity. The fluorescence intensity was plotted versus quencher concentration (0–140 mm), and the degree of quenching was analyzed using the Stern-Volmer equation (Equation 1),
where Fo and F are the fluorescence intensities in the absence and presence of acrylamide, respectively; KSV is the Stern-Volmer constant for collisional quenching, and [Q] is the concentration of aqueous quencher.
To measure the depth of penetration of the SAR1B Trp reporter into the membrane bilayer, parallax analysis was performed (
). Briefly, the fluorescence of the SAR1B-Trp reporter was individually measured in the presence of two separate membrane-embedded quenchers of distinct depths. The distance of the single Trp reporter from the bilayer center (ZCF) was determined using parallax analysis as given by Equation 2,
where LC1 represents the distance from the bilayer center to the shallow quencher; C is the mole fraction of the quencher divided by the lipid area; F1 and F2 are the relative fluorescence intensities of the shallow (6,7-Br2-PC) and deep quenchers (11,12-Br2-PC), respectively, and L is the difference in depth between the two quenchers. Brominated lipid concentrations were varied by substituting DOPC.
Atomic Force Microscopy
AFM imaging was performed using a Bruker Dimension FastScan instrument. All imaging was conducted under fluid using FastScan D cantilevers (Bruker). Their resonant frequencies under fluid were 110–140 kHz, and the actual scanning frequencies were ∼5% below the maximal resonance peak. Lipid mixtures containing either a “basic mix” of lipids (70% DOPC, 15% DOPE, and 15% DOPS) or the major/minor mix (52% DOPC, 21% DOPE, 8% DOPS, 5% DOPA, 9% PI, 2.2% PI4P, 0.8% PI(4,5)P2, 2% DAG, and supplemented with 20% cholesterol) were dried under nitrogen and hydrated in Biotechnology Performance Certified water overnight. Suspensions were probe-sonicated at an amplitude of 10 μA until the mixture became transparent. Liposomes were incubated in the presence or absence of proteins for 30 min and then placed on freshly cleaved mica. In each case, 40 μl of the liposome mixture and an equal volume of buffer (50 mm HEPES, pH 7.6, 100 mm NaCl, and 1 mm MgCl2) were applied to the mica surface. The mica was washed three times with the same buffer and placed in the fluid cell of the atomic force microscope. The assembled lipid bilayer was immersed in 150 μl of buffer, and all imaging was performed at room temperature. For the buffer-exchange experiments, buffer containing 1 mm GTP, GMP-PNP, or GDP was introduced into the fluid cell of the microscope. AFM images were acquired at a rate of four frames/min and plane-fitted to remove tilt. Each scan line was fitted to a first-order equation.
C. elegans Sar1 Regulates the Secretion of ER-derived Transport Carriers to Maintain Germ Line Tissue Architecture
The Sar1 GTPase plays a conserved role in protein and lipid secretion in several organisms, including yeast, Drosophila, and mammals (
). In humans, several distinct mutations affecting one of the Sar1 isoforms (SAR1B) have been implicated in chylomicron retention disease (Anderson disease), an early onset, inherited lipid malabsorption disorder characterized by hypocholesterolemia (
). However, the contribution of Sar1 to other developmental processes, including tissue morphogenesis, remains largely uncharacterized. To investigate a potential role for Sar1 GTPases in this process and to verify that C. elegans SAR-1 regulates protein secretion from the ER, we conducted a series of depletion experiments using the C. elegans germ line as a model system. An important feature of the C. elegans reproductive system is its amenability to RNA interference (RNAi)-mediated depletion (
). Further examination of germ line tissue depleted of SAR-1 revealed a dramatic defect in the organization of membranes that normally partition individual nuclei into compartments within the syncytial stem cell niche (Fig. 1B). A nearly identical phenotype was observed upon depletion of the COPII subunit SEC-23, suggesting that the defect was a consequence of inhibiting COPII-mediated secretion (Fig. 1B). Consistent with this idea, trafficking of integral membrane secretory cargoes, including the secretory vesicle-associated SNARE synaptobrevin (SNB-1), from the ER was potently blocked following depletion of SAR-1 or SEC-23 (Fig. 1C). These data highlight a conserved function for C. elegans SAR-1 and COPII in protein secretion and further demonstrate a key role for COPII-mediated vesicle transport during tissue morphogenesis.
Stable Membrane Penetration of Sar1 Is Dependent upon an Association with GTP
Because our findings indicated that C. elegans SAR-1 is functionally conserved, we decided to purify a recombinant form of the GTPase and analyze its membrane binding capabilities in vitro. For these experiments, nucleotide was loaded onto SAR-1 by purifying the protein in the presence of a large molar excess of GTP or GDP (500 μm), as described previously (
). Because previous work indicates that Sar1 GTPases exhibit minimal intrinsic activity in the absence of a guanine nucleotide-activating protein (GAP), we decided to use GTP in these experiments as opposed to a poorly hydrolysable analog (
). Size-exclusion chromatography and multiangle light scattering conducted in the presence of nucleotide (500 μm GTP or GDP) indicated that recombinant SAR-1 loaded with either GTP or GDP formed mainly monodisperse monomers in solution, with small populations of dimers and larger oligomers (Fig. 2, A–C, and Table 1). Similar results were obtained using recombinant forms of yeast Sar1p and human SAR1B (Fig. 2, A–C, and Table 1). These data indicate that Sar1 isoforms exist largely as monomers in solution, but they are also capable of self-association.
TABLE 1Molecular masses of monomeric Sar1 isoforms determined experimentally using size-exclusion chromatography-multiangle light scattering
Predicted molecular mass (kDa) based on amino acid content
Experimentally determined mass (kDa) when bound to GDP
Experimentally determined mass (kDa) when bound to GTP
We next investigated the manner by which Sar1 binds to lipid bilayers. Although the amphipathic amino terminus of Sar1 is known to play an essential role in membrane association and tubulation, relatively little is known about its dynamic properties. To address this issue, we used a technique that measures tryptophan fluorescence quenching by aqueous acrylamide, both in the presence and absence of membranes (
). For these experiments, we used a form of human SAR1B, which harbors only a single naturally occurring tryptophan reporter at position 7 within the amino-terminal amphipathic helix (SAR1BW86F,W192F), enabling us to examine its accessibility. The protein was purified in the presence of GDP and subsequently incubated with a 100-fold molar excess of GTP for 30 min at room temperature, which was shown previously to be sufficient for nucleotide exchange (
). This form of SAR1B (1 μm) exhibited efficient quenching of fluorescence upon the addition of acrylamide, suggesting that the amphipathic helix was exposed in solution (Fig. 2, D and E, and Table 2). We also measured changes in tryptophan fluorescence when GDP-bound SAR1BW86F,W192F was incubated with increasing concentrations of acrylamide. In contrast to the idea that the amphipathic helix is buried within the core of Sar1 when bound to GDP, we found that the addition of acrylamide resulted in equally efficient quenching with nearly identical kinetics to the GTP-bound protein (Fig. 2, D and E, and Table 2). These data argue that the amphipathic helix of Sar1 is solvent-exposed irrespective of its nucleotide-bound state.
TABLE 2Stern-Volmer constants determined for the SAR1B amphipathic helix under various conditions
When examined in the presence of membranes, we found that the GTP-bound form of SAR1B was highly resistant to tryptophan fluorescence quenching, suggesting that the amphipathic helix becomes buried within the hydrophobic core of the lipid bilayer under these conditions and is protected from acrylamide in solution (Fig. 2, D and E, and Table 2). Surprisingly, GDP-bound SAR1B also exhibited reduced fluorescence quenching when membranes were present, although the effect was not as pronounced as compared with GTP-bound SAR1B (Fig. 2, D and E, and Table 2). These data are consistent with the idea that SAR1B can associate with membranes, irrespective of the bound nucleotide. However, GTP-binding appears to promote more stable membrane penetration of the SAR1B amphipathic helix, which likely supports membrane bending during the formation of COPII-coated transport carriers.
To determine the extent to which the SAR1B amino-terminal amphipathic helix inserts into lipid bilayers, we measured tryptophan fluorescence quenching mediated by brominated phospholipids that are modified at specific positions along their acyl chains (
). Parallax analysis demonstrated that the single tryptophan reporter in the helix penetrates ∼5 Å into the hydrophobic core of the outer leaflet, just below the hydrophilic headgroups, in the presence of GTP (Fig. 2, F and G; see “Experimental Procedures” for calculations). These data provide a mechanistic basis for the ability of Sar1 isoforms to bind and bend membranes.
Sar1 Is the Curvature-sensing Component of the COPII Complex
The formation of all transport carriers requires stepwise changes in local membrane curvature that ultimately result in membrane scission. We hypothesized that a curvature-sensing mechanism may exist to facilitate budding of COPII-coated carriers from the ER. Based on its ability to bind and bend membranes, Sar1 is a logical candidate for acting as a curvature-sensitive module within the COPII complex. To test this idea, we used two independent approaches to explore the ability of Sar1 isoforms to discriminate between membranes of varying curvatures. First, we compared the binding affinities of SAR-1 for liposomes ranging in size from 105 to 225 nm in diameter, using a co-sedimentation assay. Chemically defined liposomes composed of 66% DOPC, 21% DOPE, 8% DOPS, 5% DOPA, and supplemented with 20% cholesterol, were utilized for these experiments as described previously (
). Because this assay was carried out under equilibrium conditions, we loaded SAR-1 (4 μm) with the poorly hydrolysable GTP analog GMP-PNP to ensure that it remained in an active state throughout the experiment. To monitor binding, we mixed SAR-1 with different concentrations of liposomes and tracked depletion of the protein from the supernatants after centrifugation. The resulting titration curves indicated a 2-fold difference in the apparent dissociation constants between the smallest and largest liposomes, suggesting that SAR-1 prefers membranes of higher curvature (Fig. 3, A and B).
Because Sar1 is known to remodel membranes, which may influence the binding affinities determined following the prolonged incubation periods needed for sedimentation assays, we decided to use a second approach that measures the association between proteins and membranes instantaneously. Specifically, a rapid mixing fluorescence resonance energy transfer (FRET)-based assay was used to monitor the association of Sar1 with liposomes ranging in average size from 105 to 225 nm in diameter. Importantly, we took advantage of two native tryptophan residues within the C. elegans SAR-1 amino terminus, which could act as FRET donors for headgroup-dansylated phosphatidylethanolamine (PE) that was incorporated into liposomes. Using stopped-flow spectroscopy, GTP-loaded SAR-1 (8 μm) was rapidly mixed with liposomes of varying diameters, enabling us to calculate relative binding affinities. Our data demonstrated again that SAR-1 binds preferentially to more highly curved membranes, exhibiting an ∼4.6-fold higher affinity for 105-nm liposomes as compared with 225-nm liposomes (Fig. 3, C–E). The difference in apparent binding affinities determined using the two methods likely reflects the ability of SAR-1 to tubulate the larger liposomes over time, thereby increasing its affinity for the 225-nm liposomes during the 1-h co-sedimentation process. Taken together, these data strongly suggest that Sar1 binds to membranes in a curvature-sensitive manner.
Sar1 Remodels Supported Lipid Bilayers in a Manner Dependent on GTP Hydrolysis
Current approaches to study the role of Sar1 in membrane remodeling include the use of COPII budding assays in semi-intact cells followed by differential centrifugation or imaging of synthetic liposomes and GUVs by light and electron microscopy (
). An explanation for these discrepancies remains unclear but may depend on sample handling. Nevertheless, these data suggest that cell-free assays for COPII budding exhibit substantial variability and are therefore difficult to interpret. EM-based studies using recombinant Sar1 and chemically defined liposomes or GUVs have yielded similarly conflicting results regarding the importance of GTP hydrolysis in membrane scission (
). Again, EM methods employed may be highly sensitive to sample manipulation. Although live imaging assays cause the least perturbation, the resolution afforded by this approach is typically limited and has only demonstrated that Sar1 tubulates membranes in the presence of GTP (
To study the action of Sar1 on lipid bilayers, we took an alternative approach that uses AFM imaging under fluid. In this method, Sar1 assembly was monitored at nanometer resolution on supported lipid bilayers (SLBs) assembled on mica, with very limited mechanical perturbation. We used SLBs composed of either a basic mix of lipids (70% phosphatidylcholine, 15% PE, and 15% phosphatidylserine) or the major/minor mix (52% DOPC, 21% DOPE, 8% DOPS, 5% DOPA, 9% PI, 2.2% PI4P, 0.8% PI(4,5)P2, 2% DAG), which was shown previously to be important for maximal Sar1 membrane binding and facilitating Sec23-Sec24-mediated GAP activity on Sar1 (
). Gaps in the bilayer that formed spontaneously, irrespective of protein addition or lipid composition, produced edges with high curvature (Fig. 4A), enabling us to determine the distribution of Sar1 in the presence of both flat and highly curved membrane surfaces. We previously used a similar AFM-based assay to study the formation and activity of the ESCRT-III complex, another protein machinery implicated in membrane bending and scission, and demonstrated that it assembled in a curvature-dependent manner and promoted remodeling of the bent bilayer edges (
). Importantly, by imaging in a fast-scan mode, we were able to visualize the effects of various nucleotides, nucleotide analogs, and other proteins on Sar1 membrane association and remodeling.
Supported lipid bilayers composed of either the basic mix or the major/minor mix formed in the absence of proteins exhibited mild movements on the mica surface, indicating that they remained mobile throughout the period of AFM imaging (Fig. 4B and supplemental Video 1). Cross-sections showed that the height of the bilayer was largely uniform across flat regions but exhibited a steep decline at its edges where the mica was exposed (Fig. 4C). When supplemented with SAR-1 (400 nm) that had been purified in the presence of GDP, we observed the appearance of particles, which were distributed across the bilayer and at the periphery of the gaps that formed during bilayer assembly (Fig. 4D). Decoration of the bilayer edges was particularly evident when excessive force was applied onto the bilayer with the AFM tip (Fig. 4E). To verify that these particles corresponded to SAR-1, we varied the concentration of the protein used (100 nm to 1.6 μm) and measured the number of particles observed by AFM. Our findings showed that increased levels of SAR-1 led to the appearance of more particles on the bilayer, which accumulated predominantly at the highly bent edges of the membranes (Fig. 4, F–H). Together, these data highlight our ability to monitor SAR-1 accumulation on supported lipid bilayers using AFM.
To facilitate nucleotide exchange onto SAR-1 in the absence of its GEF, a buffer containing 1 mm GTP (equivalent to ∼2-fold the concentration found in most mammalian cell types) was flowed onto a bilayer composed of the basic mix of lipids. Within moments, the bilayer began to undergo a remodeling process that resulted in its systematic removal from the mica surface, which we quantified by measuring the percentage of the mica surface covered with membrane over time (Fig. 4, I and J, and supplemental Video 2). Based on time-lapse AFM imaging, membrane removal appeared to occur from the highly curved edges of the bilayer (Fig. 4I), consistent with our data indicating that SAR-1 can sense membrane curvature. In contrast to the effect of SAR-1 at bent membranes, its presence on the flat regions of the bilayer failed to affect membrane topology under these conditions. The addition of GTP to protein-free bilayers had no substantial effect on their architecture as compared with buffer alone, and the addition of GDP to bilayers harboring SAR-1 similarly had no effect (supplemental Videos 3 and 4). Notably, bilayer removal mediated by SAR-1 was distinct from that of a detergent, which caused the membrane to dissociate from the mica surface in a manner that did not depend on membrane curvature (supplemental Video 5). These data support the idea that GTP-bound SAR-1 is capable of remodeling curved lipid bilayers.
To address the question of whether GTP hydrolysis is necessary for membrane remodeling and removal mediated by Sar1, SLBs composed of the basic mix and harboring SAR-1GDP (400 nm) were exposed to an excess of GMP-PNP (1 mm). In contrast to the effect of GTP, we observed only a transient change in bilayer appearance, which was arrested within moments after addition of the nucleotide analog (Fig. 5, A and B, and supplemental Video 6). We also examined the ability of a mutant form of Sar1 (H75G), which harbors a point mutation within its catalytic switch II region and renders the protein incapable of GTP hydrolysis at any concentration (
), to act on supported lipid bilayers. Although we found SAR-1H75G was highly enriched along membrane edges, the addition of GTP (1 mm) did not stimulate membrane removal, even upon repeated exposure (Fig. 5, C and D, and supplemental Video 7). Together, these data argue that GTP hydrolysis is necessary for the continuous function of SAR-1 to remodel and sever lipid bilayers.
To further confirm the importance of GTP hydrolysis for SAR-1-mediated membrane remodeling, we also examined the effect of adding C. elegans Sec23-Sec24 (500 nm), the SAR-1 GAP, together with GTP (1 mm) to bilayers. For these studies, we used membranes composed of the major/minor mix, which was shown previously to be necessary for efficient Sec23-Sec24-mediated GAP activity on Sar1 (
). Similar to our findings using bilayers composed of the basic mix, the addition of GTP alone to the major/minor mix membranes containing SAR-1 initiated its steady removal (Fig. 6, A and B, and supplemental Video 8), although the kinetics of the process was slower and multiple additions of GTP were required. The addition of Sec23-Sec24 and GTP to bilayers lacking SAR-1 had no effect on the membrane (Fig. 6, C and D, and supplemental Video 9), consistent with the inability of Sec23-Sec24 to bind lipid bilayers without SAR-1 (
). However, we observed substantial, nonspecific accumulation of Sec23-Sec24 heterodimers on the mica surface (Fig. 6, C and D, and supplemental Video 9). We next added Sec23-Sec24 and GTP to major/minor mix bilayers containing SAR-1 and observed an extremely rapid removal of the membrane (Fig. 6, E and F, and supplemental Video 10). These data are again consistent with the idea that GTP hydrolysis on SAR-1, which is stimulated by Sec23-Sec24, is critical for promoting membrane remodeling by SAR-1.
Membrane Curvature Regulates Sar1 GTP Hydrolysis
Given the ability of SAR-1 to bind preferentially to membranes of high curvature and remodel them in a manner that depends on GTP hydrolysis, we questioned whether curvature may also influence the rate of GTP hydrolysis on Sar1. To examine this possibility, we employed a new, facile, and highly reproducible bioluminescence assay for monitoring the enzymatic activity of Sar1 in the presence of liposomes that ranged in average size from 105 to 225 nm in diameter. To first validate the use of this assay, we measured Sar1-mediated GTP hydrolysis over a range of protein concentrations. These data highlight the ability of SAR-1 to hydrolyze GTP in solution, but only at relatively high concentrations (above 4 μm) (Fig. 7A). The mutant form of SAR-1 (H75G) was largely incapable of GTP hydrolysis at any concentration, as predicted by previous work (Fig. 7A). Additionally, the supplementation of the reaction with recombinant Sec23-Sec24 dramatically increased the rate of GTP hydrolysis by wild type SAR-1 but not SAR-1H75G (Fig. 7A). These findings collectively indicate that the bioluminescent assay provides a faithful readout of Sar1 GTPase activity.
To test whether membrane curvature affects Sar1 GTPase activity, we incubated nucleotide-free SAR-1 (3.1 μm) with Sec23-Sec24 heterodimers (255 nm) and liposomes (1 mm) of various curvatures and initiated each reaction by the addition of 5 μm GTP. Based on our sedimentation assays, equivalent amounts of SAR-1 are bound to each liposome population under these conditions, irrespective of their diameter (Fig. 3, A and B). Our studies indicated that GTPase activity was inversely related to liposome size, with the 105-nm liposomes promoting SAR-1 activity significantly better than liposomes greater than 200 nm in diameter (Fig. 7B). Additionally, by comparing these data to GTPase activity measured in the absence of membranes, we found that the largest liposomes used (225 nm in diameter) failed to significantly stimulate GTP hydrolysis on SAR-1, in contrast to the smaller liposomes (Fig. 7B). We also confirmed that SAR-1 exhibits curvature sensitivity in the presence of Sec23-Sec24 using a standard colorimetric assay that measures inorganic phosphate release as a function of GTPase activity (Fig. 7C). To further verify that SAR-1, and not its GAP, exhibits curvature sensitivity, we conducted similar experiments using SAR-1 alone. To detect inorganic phosphate release, a higher concentration of SAR-1 was necessary (14 μm). Under these conditions, we again found that GTP hydrolysis on SAR-1 was increased in the presence of small liposomes (105 nm) as compared with large liposomes (225 nm) (Fig. 7D).
To determine whether this curvature-sensitive property of SAR-1 is conserved, we also measured the ability of yeast Sar1p to hydrolyze GTP in the presence of differently sized liposomes. Strikingly, the impact of curvature was even more pronounced, with a significant difference in Sar1p GTPase activity in the presence of 105-nm liposomes as compared with 225-nm liposomes (Fig. 7E). These data highlight a new regulatory mechanism that controls Sar1 function and demonstrates the role of membrane curvature in influencing GTP hydrolysis on Sar1. Together, our findings suggest that the various curvatures adopted during individual steps of COPII-mediated transport carrier budding directly influence Sar1 distribution and activity.
The amino-terminal region of Sar1 has been shown to play an essential role in membrane deformation and generating COPII-coated transport carriers, both in vivo and in vitro (
). However, the mechanisms by which this region binds and manipulates membranes remain unclear. To date, the intact domain has not been crystallographically resolved, but it is predicted to form an amphipathic α-helix and is capable of destabilizing lipid bilayers on its own (
). In contrast, our data suggest that the amphipathic helix transitions constantly through solvent-exposed states, irrespective of GTP binding. However, GTP loading promotes stable membrane penetration, which is necessary to generate curvature. Thus, our results are consistent with the idea that Sar1 continually samples membranes, but only in the presence of its GEF Sec12 can it stably insert to facilitate the formation of COPII transport carriers.
In addition to its role during the initial stages of COPII carrier formation, Sar1 has also been implicated in membrane sculpting and scission steps (
). Recent work highlighted the propensity of Sar1 to dimerize and form ordered lattices on membrane tubules exhibiting diameters larger than 60 nm, equivalent to the size of the smallest COPII-coated transport carriers observed in vitro or in vivo (
). These data are consistent with our results demonstrating that Sar1 can form dimers in solution. In cases where membrane tubules fall below 60 nm in diameter, ordered arrays of Sar1 are no longer observed, implying that curvature plays an important role in Sar1 membrane association (
). We explored this idea further by examining the relative binding affinities of Sar1 for a range of membrane curvatures, which COPII transport carriers likely adopt during their formation. Our findings indicate that Sar1 associates more tightly with membranes of elevated curvature. Importantly, discrepancies in the relative affinities obtained using equilibrium binding and stopped-flow rapid kinetics further suggest that Sar1 actively bends the membrane, diminishing lipid density in the outer leaflet as compared with the inner leaflet and facilitating membrane penetration of additional molecules of Sar1. A localized increase in Sar1 concentration is likely instrumental in achieving sufficient bending, potentially through molecular crowding, to ultimately drive membrane scission (
In many ways, Sar1 appears to act similarly to epsin, a regulator of clathrin-coated vesicle formation, which harbors an ENTH domain that penetrates the inner leaflet of the plasma membrane to a depth similar to what we have measured for the Sar1 amphipathic helix (
). Analogously, by acting as the curvature sensor during COPII transport carrier formation, Sar1 may possess a similar function. Consistent with this idea, at sufficiently high concentrations, Sar1 is capable of transforming membrane tubules into vesicles independently of GTP hydrolysis (
). However, fission reactions that occur under conditions where the GTPase cycle is halted typically require high levels of Sar1 to be present, raising concerns. To address this issue, we took advantage of an in vitro assay to study the impact of various nucleotides and nucleotide analogs on Sar1-mediated membrane remodeling using a limited concentration of the protein (400 nm). Irrespective of the artificial nature of these experiments, our combined use of supported lipid bilayers and AFM imaging enabled us to visualize the effect of Sar1 under various conditions in real time. Our findings demonstrated that GTP hydrolysis is necessary for membrane remodeling mediated by Sar1, suggesting a requirement to expend energy during this process. These data were particularly surprising in light of our results showing that the intrinsic rate of Sar1 GTP hydrolysis in the absence of its GAP is extremely slow. Nevertheless, this minimal amount of GTP consumption appears to be required for Sar1 to perform work on supported lipid bilayers, as neither GTP-locked Sar1 (H75G) nor wild type Sar1 in the presence of poorly hydrolysable GTP analogs were able to mimic the effect of adding GTP to wild type Sar1 assembled on the edges of membranes.
One possible explanation for these findings stems from our work showing that GTP hydrolysis on Sar1 is enhanced in the presence of membranes with elevated curvature. Although the differences in GTPase activity that we measured may appear modest, we were only able to examine effects over a narrow range of curvatures (∼0.004–0.009 nm−1, where curvature is defined as 1/radius). The edges of supported lipid bilayers are predicted to exhibit the highest degree of curvature feasible (∼0.4 nm−1), which appears to promote substantial GTP hydrolysis on Sar1, sufficient to remodel the bilayer even in the absence of its GAP. An analogous situation may arise during the fission of COPII-coated transport carriers from the ER. Over the course of this final step, the ER membrane must achieve a high degree of curvature, significantly higher than that of vesicles we can generate by extrusion, which may be sufficient to promote GTP hydrolysis on Sar1 and thereby disrupt the bilayer to enable fission (
). The role of Sar1 GAP activity must also be considered, especially in light of its dramatic enhancement of the kinetics of SAR-1-mediated membrane remodeling. It is feasible that Sec23 behaves similarly to ArfGAP1, the GAP that stimulates GTP hydrolysis on the Arf1 GTPase in a curvature-dependent manner and directs COPI-mediated trafficking (
). Thus, our data are more consistent with the idea that Sar1, and not its GAP, acts as the curvature sensor in COPII-mediated membrane transport.
Taken together, our findings support a model in which Sec12-mediated GTP loading onto Sar1 allows for stable membrane penetration of its amphipathic tail to initiate membrane curvature on subdomains of the ER and recruitment of additional coat components. A number of regulatory factors, including Sec16, likely contribute to the control of Sec23-dependent GAP activity to limit release of Sar1 from the membrane (
). Ultimately, we speculate that the progressive ability of Sar1 to bend membranes, likely in coordination with COPII coat assembly, leads to an increased propensity for GTP hydrolysis, which is able to support membrane fission and the release of coated transport carriers.
M. G. H., I. M., R. M. H., E. R. C., J. M. E., and A. A. conceived and designed the experiments. M. G. H., I. M., L. W., E. R. C., J. M. E., and A. A. performed the experiments and analyzed the data. M. G. H. and A. A. wrote the paper.
We thank Elizabeth Miller for providing purified yeast COPII components, Subhanjan Mondal and Said Goueli at Promega Corp., for providing us access to the GTPase-Glo system ahead of release, and members of the Audhya laboratory for critically reading this manuscript.