Structural basis for the initiation of COPII vesicle biogenesis

Summary

The first stage of the eukaryotic secretory pathway is the packaging of cargo proteins into COPII vesicles exiting the endoplasmic reticulum (ER). The cytoplasmic COPII vesicle coat machinery is recruited to the ER membrane by the activated, GTP-bound, form of the conserved Sar1 GTPase. Activation of Sar1 on the surface of the ER by Sec12, a membrane-anchored GEF (guanine nucleotide exchange factor), is therefore the initiating step of the secretory pathway. Here we report the structure of the complex between Sar1 and the cytoplasmic GEF domain of Sec12, both from Saccharomyces cerevisiae. This structure, representing a key nucleotide-free activation intermediate, reveals how the potassium ion-binding K-loop disrupts the nucleotide binding site of Sar1. We propose an unexpected orientation of the GEF domain relative to the membrane surface and postulate a mechanism for how Sec12 facilitates membrane insertion of the amphipathic helix exposed by Sar1 upon GTP-binding.

eTOC blurb

The first step of the eukaryotic secretory pathway is formation of “COPII” vesicles at the endoplasmic reticulum. The COPII coat machinery is recruited by the Sar1 GTPase, which is activated via nucleotide exchange by Sec12. Joiner et al. determined the structure of the Sar1-Sec12 complex, representing a key activation intermediate.

Introduction

The eukaryotic secretory pathway is essential for protein secretion and trafficking of virtually all membrane proteins that localize to the plasma membrane, lysosomes, endosomes, and the Golgi complex. The first step of the secretory pathway, after protein synthesis and quality control measures within the ER, is the concentration and packaging of cargo into vesicles for transport to the Golgi complex. These vesicles are generated by the conserved coat protein complex II (COPII) machinery that interacts directly with the cytoplasmic portions of cargos and sculpts the ER membrane into transport carriers (Russell and Stagg, 2010; Jensen and Schekman, 2011; Zanetti et al., 2012; Barlowe and Miller, 2013; Barlowe, 2019).

The generation of COPII-coated vesicles at the ER is initiated by activation of Sar1, a small Ras-like GTPase of the Arf family (Oka et al., 1991; Barlowe et al., 1993; Huang et al., 2001; Bi et al., 2002). Activated, GTP-bound Sar1 triggers formation of the COPII coat by directly interacting with the inner layer of the coat complex, the Sec23/Sec24 heterodimer. Sar1 binds to the ER membrane through its N-terminal amphipathic helix, which is exposed only upon GTP-binding and plays a primary role in membrane remodeling during vesicle formation and fission (Antonny et al., 1997; Aridor et al., 2001; Huang et al., 2001; Bielli et al., 2005; Lee et al., 2005; Hariri et al., 2014; Schwieger et al., 2017).

Sec12 is the GEF responsible for activating Sar1 via nucleotide exchange (Nakano et al., 1988; d’Enfert et al., 1991; Barlowe and Schekman, 1993; Weissman et al., 2001; Futai et al., 2004). Among established GEFs for Arf family members, Sec12 is the only known integral membrane protein. Sec12 is anchored to the ER through its C-terminal transmembrane domain and the nucleotide exchange activity resides in the N-terminal cytosolic β-propeller domain (Barlowe and Schekman, 1993; Futai et al., 2004; McMahon et al., 2012).

Although the structures of Sec12 and Sar1 are known, and the kinetics of Sec12 GEF activity towards Sar1 have been characterized, the actual mechanism used by Sec12 to activate Sar1 via nucleotide exchange has not been determined (Huang et al., 2001; Bi et al., 2002; Futai et al., 2004; Rao et al., 2006; McMahon et al., 2012). In contrast to many small GTPases of the Ras superfamily, Sar1 and most other Arf GTPase family members are distinguished by the direct coupling of activation to insertion into the cytoplasmic leaflet of a membrane bilayer (Antonny et al., 1997). Despite extensive structural characterization of Arf family GEFs (Cherfils et al., 1998; Goldberg, 1998; Renault et al., 2003; DiNitto et al., 2007; Aizel et al., 2013; Malaby et al., 2013; Padovani et al., 2014; Galindo et al., 2016; Richardson et al., 2016; Karandur et al., 2017; Das et al., 2019), it has remained unresolved how these GEFs couple nucleotide exchange with membrane insertion of their substrates.

Here, we report the crystal structure of the yeast Sar1-Sec12 complex representing a key intermediate of the nucleotide exchange reaction. This structure provides detailed insights into the mechanism of nucleotide exchange, and in particular explains the vital role of the potassium ion-binding K-loop of Sec12 in rearranging the Sar1 nucleotide binding site. We identify an extensive, positively charged surface on the complex which also contains the Sec12 transmembrane domain. The coexistence of these features on the same surface suggests a specific orientation of the complex on the membrane. We therefore propose that Sec12 positions Sar1 in an orientation that allows its N-terminal helix to insert directly into the ER membrane upon GTP binding.

Results

The structure of the Sar1-Sec12 nucleotide exchange intermediate

GEFs activate their GTPase substrates by competing with nucleotide for binding to the GTPase. Therefore, GEFs typically make extensive contacts with their substrate in order to displace bound GDP or GTP. As this binding is reversible and cellular GTP concentrations are much higher than GDP concentrations, subsequently the GEF is displaced by GTP resulting in activated GTPase (Goody and Hofmann-Goody, 2002). Nucleotide-free GTPase bound to its GEF is a critical intermediate in this pathway. To determine the structural and mechanistic basis for Sec12 GEF activity, we sought to capture the nucleotide-free intermediate of the Sar1-Sec12 guanine nucleotide exchange reaction. We incubated the purified cytosolic domain of S. cerevisiae Sec12 (residues 1-354) with a purified S. cerevisiae Sar1 construct lacking its N-terminal membrane-inserting amphipathic helix (residues 24-190) and a small amount of alkaline phosphatase. The phosphatase was included to hydrolyze guanine nucleotide which would otherwise compete with Sec12 for binding to Sar1. We purified the stable nucleotide-free Sar1-Sec12 complex (Figure S1) and grew crystals for X-ray diffraction studies. We collected an X-ray diffraction dataset using a single crystal and solved the structure by molecular replacement using the previously reported structures of yeast Sec12 (McMahon et al., 2012) and hamster Sar1-GDP (Huang et al., 2001). Due to significant anisotropy in the data, we used STARANISO (Tickle et al., 2018) for anisotropic correction. After model building and refinement against the resulting structure factors extending to 2.3Å (Figure S2), the final atomic model (Figure 1A) exhibits appropriate geometry and refinement statistics (Table 1).

Figure 1: The structure of a Sar1-Sec12 nucleotide exchange intermediate.

A. Sar1 (residues 24-190) is orange and Sec12 (residues 1-354) is blue. The purple sphere depicts the bound K⁺ ion in the K-loop of Sec12. Left, surface representation; right, ribbon depictions. The N- and C- termini are labeled and dashed lines indicate regions lacking electron density which were unmodeled.

B. Superposition of Sar1-bound Sec12 (blue, this study), and Sec12 alone (pink, (McMahon et al., 2012), PDB: 4H5I). The pink sphere depicts a K⁺ ion that is observed only in the unbound crystal structure.

C. Comparison of Sar1-GDP (purple, (Huang et al., 2001), PDB: 1F6B), nucleotide-free Sar1 (orange, this study), and Sar1-GTP (green, (Bi et al., 2002), PDB: 1M2O). Regions of Sar1 are highlighted by coloring: the phosphate loop (residues 30-37) is magenta, the switch I region (residues 46-57) is yellow, the switch II region (residues 76-92) is blue, the Ω loop (residues 154-169) is tan, and the nucleotide is black.

See also Figures S1 and S2.

Table 1:

Diffraction data and refined model statistics

Wavelength	0.979180
Unit cell	50.333 93.62 115.423 90 90 90
Space group	P 21 21 21
Scaling approach	Isotropic/Spherical	Anisotropic/Ellipsoidal*
RCSB PDB code	n/a	6×90
Resolution range, scaling and merging (outer shell)	57.71 – 2.26 (2.34 – 2.26)	57.71 – 2.26 (2.33 – 2.26)
Total reflections	174650 (16809)	132029 (6507)
Unique reflections	26271 (2524)	20215 (1012)
Multiplicity	6.6 (6.5)	6.5 (6.4)
Spherical completeness, %	99.06 (97.07)	76.6 (43.6)
Ellipsoidal completeness, %	N/A	95.3 (98.6)
Mean I/sigma-I	20.0 (1.2)	20.1 (2.4)
Wilson B-factor	66.3	49.2
R-merge	0.0553 (1.278)	0.039 (0.741)
R-meas	0.0600 (1.388)	0.043 (0.806)
R-pim	0.0231 (0.533)	0.017 (0.313)
CC1/2	0.999 (0.844)	1.000 (0.854)
CC*	1 (0.957)	N.D.
Resolution range, refinement (outer shell)	57.71 – 2.26 (2.34 – 2.26)	57.71 – 2.26 (2.34 – 2.26)
Reflections used in refinement	26101 (2514)	20180 (1152)
Reflections used for R-free	1222 (122)	921 (55)
R-work	0.2557 (0.4923)	0.1985 (0.2886)
R-free	0.3051 (0.5320)	0.2571 (0.3598)
Number of non-hydrogen atoms	3756	3826
Macromolecules	3753	3753
Ligands	1	1
Solvent	2	72
Protein residues	485	485
RMS(bonds)	0.004	0.005
RMS(angles)	0.97	0.78
Ramachandran favored (%)	94.13	93.92
Ramachandran allowed (%)	5.87	6.08
Ramachandran outliers (%)	0	0
Rotamer outliers (%)	0.48	0.72
Clashscore	4.11	3.45
Average B-factor	95.59	66.16
Macromolecules	95.6	66.47
Ligands	93.6	61.80
Solvent	72.5	50.42
TLS groups	0	8

^* Note the resolution range of the outer shell used by STARANISO is slightly different than that used by XDS for the isotropically scaled data merging statistics.

The structure of Sec12 in the Sar1-Sec12 complex is almost identical to the previously published structure of Sec12 alone (McMahon et al., 2012) (Figure 1B). Sec12 consists of seven β-sheets, or “blades”, arranged in a circular fashion to form a β-propeller. On one side of the first propeller blade, there is an extended loop, termed the K-loop, which binds to a potassium ion. This structural feature was previously demonstrated to be essential for GEF activity (McMahon et al., 2012). Consistent with these findings, we observe the K-loop makes direct contact with Sar1, and its conformation is essentially unchanged by Sar1-binding. This interface is discussed further below. In both the Sar1-bound and unbound structures, electron density is absent for the last ten residues of Sec12 (residues 345-354), suggesting these residues serve as a flexible linker connecting the catalytic cytosolic domain to the transmembrane domain (McMahon et al., 2012).

Sar1, like other small GTPases, adopts different structural conformations in its inactive (GDP-bound) and active (GTP-bound) forms (Huang et al., 2001; Bi et al., 2002; Rao et al., 2006; Bi et al., 2007). When bound to Sec12 as a nucleotide-free intermediate in the exchange reaction, we observed that Sar1 adopts a conformation significantly different from either its GDP- or GTP-bound forms (Figure 1C). The most significant observable differences between the structures correspond to the “switch” and “interswitch” regions, whereas the conformation of the phosphate loop remains largely unaltered by Sec12 binding. Some regions of Sar1 lack electron density in complex with Sec12, indicative of their flexibility: residues 48-65, which encompass the entirety of the switch I region and part of β3; and residues 152-162, which correspond to the Sar1-specific “Ω loop” (Huang et al., 2001), are disordered when bound to Sec12. The switch II region remains ordered but has shifted significantly due to displacement by the K-loop of Sec12. The switch II region moves by ~8Å between GDP-bound and the Sec12-bound state and by ~10Å from the Sec12-bound state to the GTP-bound state (Figure 1C). For comparison, the switch II region shifts ~8Å between the GDP- and GTP-bound states (Figure 1C). The consequence of these structural changes is the loss of the nucleotide binding site, particularly the portion responsible for interacting with the Mg²⁺ ion required to stably bind to the negatively charged phosphate groups.

The Sar1-Sec12 interface and the mechanism of nucleotide release

In the crystal structure, Sar1 is bound on one face of the Sec12 β-propeller, making contact with a surface that extends from the center of the β-propeller to the K-loop at the far edge of the first propeller blade (Figure 2A,B). This interface is extensive (1,040A², Table S1) and displays a high degree of sequence conservation (Figure 2C).

Figure 2: Sar1-Sec12 interaction surfaces.

A. Transparent surface and ribbon representation of the Sar1-Sec12 complex in which the interfacial surface is colored green. The purple sphere depicts the bound K⁺ ion in the K-loop of Sec12. The perspective on the left is the same as that shown in Figure 1A.

B. The complex (colored as in A) has been pulled apart in order to visualize the interface area (green) on each protein separately.

C. Same as in B, but colored based on sequence conservation with maroon indicating highest conservation and teal representing low conservation.

D. Ribbon representation of Sar1-GTP (green) bound to Sec23 (yellow) (Bi et al., 2002) next to nucleotide-free Sar1 (orange) bound to Sec12 (blue) (this study). The perspective of the Sar1-Sec12 complex is a 180 degree rotation from that shown in A.

E. As in D, but with Sar1 of each model superimposed.

There is significant overlap between the surface that Sar1 uses to bind to Sec12 and the surface that Sar1 uses to bind to its primary effector, Sec23 (Bi et al., 2002) (Figure 2D,E). As a consequence, we expect that Sar1-GTP bound to Sec23 in the COPII “pre-budding” complex will be protected from engaging with Sec12 again and undergoing unnecessary nucleotide exchange.

Previous studies investigated the functional importance of several conserved Sec12 residues (Futai et al., 2004; McMahon et al., 2012) (Table S2) that we now establish to lie at the Sar1 binding interface. The Sar1-Sec12 interface involves over 25 residues of Sar1, most of which belong to the switch II region, and over 30 residues of Sec12 from 6 out of the 7 blades of the β-propeller (Figure 3A). There are multiple types of interactions between residues of the two proteins, for example: hydrogen bonds are formed between the sidechains of Sec12 Tyr132 and Sar1 Lys85 and Asp86, and between the mainchain carbonyl oxygen of Sec12 Asn35 and the Sar1 Asp32 sidechain (Figure 3A,B); an aromatic Pi-stacking interaction is observed between Sec12 Phe310 and Sar1 Phe72 (Figure 3A,B); and electrostatic interactions are observed between Sec12 Asp71 and Sar1 residues Gln79 and Arg82 (Figure 3C).

Figure 3: Sec12 interacts extensively with the Sar1 switch II region.

A. Closeup view of the Sec12 and Sar1 interface with the molecules pulled apart to more clearly see all the residues involved in making contact. The switch II region (blue) of Sar1 contributes many residues to the interaction surface with Sec12 (gray). Residues of the Sec12 K-loop and the surrounding propeller blades comprise the interaction surface. The purple sphere depicts the bound K⁺ ion in the K-loop of Sec12, and the Sar1 P-loop is pink.

B. The interface viewed from a slightly different perspective.

C. A broader perspective of the Sar1-Sec12 interface with zoomed inset highlighting the interaction between Sec12 Asp71 and Sar1 Gln79 and Arg82.

D. The sec12-D71A mutant is unable to complement the loss of wild-type SEC12.

E. Representative, normalized traces from in vitro nucleotide exchange assay measuring the kinetics of Sec12ΔC activation of full-length Sar1 in the presence of synthetic liposomes.

F. Quantification of E. Wild-type exchange rate is similar to the previously reported rate (McMahon et al., 2012), but the D71A mutant is unable to activate FL-Sar1. Error bars indicate the 95% confidence intervals, n = 3.

See also Figure S3 and Tables S1 and S2.

While prior investigations tested the importance of some of these Sec12 residues, we identified many whose functional significance had not been interrogated. We generated several new mutant Sec12 alleles and screened them for their ability to complement a sec12Δ yeast strain (Figure 3D and Table S2). We observed that the sec12-D71A mutant was inviable (Figure 3D), suggesting that the electrostatic interactions between Sec12 Asp71 and Sar1 Gln79 and Arg82 (Figure 3C) are required for recognition of Sar1 by Sec12. To test this hypothesis, we measured the nucleotide exchange activity of the D71A mutant using a quantitative biochemical reconstitution system (Futai et al., 2004; Richardson and Fromme, 2015). We monitored the kinetics of Sar1 activation by the purified Sec12 cytoplasmic domain in the presence of synthetic liposomes. The liposomes are essential in this reaction because the activation of full-length Sar1 only occurs on the surface of a membrane. We observed that the Sec12ΔC[D71A] construct was well behaved (Figure S3) but lost virtually all exchange activity (Figure 3E,F, Table S2), providing an explanation for the loss of function of this mutant in vivo.

In order to gain further insight into why certain interface residues appear to be more critical than others, we constructed a structural animation to visualize the conformational transitions that Sar1 undergoes during the nucleotide exchange reaction (Supplemental Video S1). This analysis used “morphing” to highlight structural transitions between crystal structures of Sar1-GDP (Huang et al., 2001) (Figure 4A), Sar1-Sec12 (this study, Figure 4B), and Sar1-GTP (Bi et al., 2002) (Figure 4C). Importantly, morphing does not reveal the precise conformational pathway between these steps, but helps to visualize the conformational differences between each experimentally determined structural snapshot and therefore can reveal key features of the Sec12 GEF mechanism.

Figure 4: The structural basis for nucleotide release.

A. Sar1-GDP (Huang et al., 2001) colored as in Figure 1, centered on the nucleotide binding pocket. Asp75 (Asp73 in yeast) is highlighted for its role in coordinating the magnesium ion for nucleotide binding. The numbers in parentheses indicate the equivalent residue numbers in yeast Sar1.

B. Sar1-Sec12 complex colored as in Figure 3, perspective as in A. Sar1 Asp73 (Asp75 in hamster) is displaced by Sec12 Ile38, disrupting a hydrogen bond network necessary for magnesium ion stabilization.

C. Sar1-GTP (Bi et al., 2002) colored as in Figure 1, perspective as in A and B.

D. Sar1-Sec12 complex colored as in B, next to Sar1-GDP (Huang et al., 2001) colored as in A. The Sar1 molecules are aligned, highlighting the conformational changes that occur in the switch II region as a result of Sec12 binding.

See also Supplementary Video S1.

Our analysis indicates that the nucleotide release mechanism involves destabilization of nucleotide binding by steric displacement of the Sar1 switch I, switch II, and interswitch regions. The Sec12 K-loop plays an indispensable role, occupying a location previously held by the switch II α-helix and part of the nucleotide binding site. The displaced switch II a-helix is stabilized in a new position by binding to the center of the Sec12 propeller fold (Figure 4D). Key residues mediating this interaction are Sec12 Asp71 and Tyr132 (Figure 3A,B). Sar1 switch II residue Tyr87, which in the Sar1-GDP structure is nestled on the surface of the GTPase, protrudes from Sar1 and points towards the center of the Sec12 propeller (Figure 3B). Accordingly, substitution mutations in Sec12 Tyr132 and several K-loop residues severely impacted Sec12 GEF activity (Table S2) (McMahon et al., 2012), and the importance of Sec12 Asp71 was demonstrated above.

Our structural results highlight Sec12 Ile38, which is part of the K-loop, as a critical and conserved residue for the nucleotide exchange mechanism. The Ile38 sidechain is not involved in binding to the Sec12 potassium ion (McMahon et al., 2012); instead it appears to be critical for displacing the Sar1 interswitch and switch II regions, thus repositioning Sar1 Asp73 (Figure 4B). Sar1 Asp73 is important for binding the hydrated Mg²⁺ ion and therefore stabilizing bound nucleotide, in both the Sar1-GDP and Sar1-GTP structures (Huang et al., 2001; Bi et al., 2002) (Figure 4A,C). Sec12 I38A and I38D substitution mutations disrupted nucleotide exchange (McMahon et al., 2012), likely because the mutant sidechains do not possess the bulk required to displace Sar1 Asp73 and interfere with Mg²⁺ ion binding.

Sec12 and Sar1 interaction surfaces are distinct from structurally similar proteins

We compared the structure of the Sar1-Sec12 complex to that of the Arf1 GTPase bound to the GEF domain of the Arf-GEF Gea2 (Goldberg, 1998) (Figure 5A). Yeast Arf1 and Sar1 share ~35% amino acid sequence identity and the nucleotide-bound conformations of these proteins are structurally very similar. In contrast, the Gea2 GEF domain adopts a “Sec7” fold composed primarily of α-helices, sharing no structural resemblance with the β-propeller fold of the Sec12 cytoplasmic domain.

Figure 5: Sec12 and Sar1 interactions are distinct from structurally similar proteins.

A. A ribbon representation of the structure of the Arf1-Gea2 complex (Goldberg, 1998) (Gea2 in yellow, Arf1 in gold) is shown next to the structure of Sar1-Sec12 (Sar1 in orange, Sec12 in blue).

B. As in A, but with GTPases superimposed.

C. A ribbon representation of the structure of the Ran-Rcc1 complex (Renault et al., 2001) (Ran in red, Rcc1 in pink) shown alongside our structure of Sar1-Sec12 (Sar1 in orange, Sec12 in blue).

D. As in C, but with propeller domains superimposed.

E. As in C, but the structures have been rotated in distinct 90° turns so that the GTPases lie behind the GEFs.

F. As in C and D, but with the GTPases superimposed.

See also Figure S4.

In spite of the similarity between Arf1 and Sar1, there are differences in their GEF-bound structures and the interfaces they use to interact with their GEFs (Figure 5B). Arf1 uses residues from its switch I, switch II, and phosphate loop regions to interact with Gea2 (Goldberg, 1998), while Sar1 uses several residues from the phosphate loop, the interswitch region, switch II, and from helix α3.

Both Sec12 and the Sec7-domain catalyze nucleotide exchange by displacing a bound Mg²⁺ ion to weaken the bonding network of the nucleotide, but they utilize different structural approaches. Gea2 and other Sec7-domain proteins use a catalytic glutamate finger to displace the magnesium and to sterically and electrostatically repel the bound nucleotide (Beraud-Dufour et al., 1998; Goldberg, 1998). However, Sec12 does not employ an acidic residue and instead relies on a bulky hydrophobic residue (Ile38) to disrupt nucleotide and Mg²⁺ binding through steric displacement (Figure 4B).

There is one other structurally characterized GEF that adopts a β-propeller fold, Rcc1, the GEF for the Ran GTPase (Renault et al., 1998, 2001; Makde et al., 2010). Rcc1 uses a “β-wedge” protrusion from one blade of its β-propeller to interact with Ran, similar to the structure and function of the K-loop of Sec12, although the Rcc1 wedge and Sec12 K-loop lie on opposite faces of the β-propeller fold (Figure 5C,D, Figure S4). In spite of the use of comparable structural protrusions, the interactions between the GEFs and their substrate GTPases are fairly different. Ran binds to Rcc1 closer to the center of the propeller blades than does Sar1 to Sec12 (Figure 5E). Accordingly, Ran interacts with all seven blades of Rcc1, whereas Sar1 interacts with only six of the seven blades of Sec12. Both Sar1 and Ran interact with their GEFs using residues from their phosphate loop, interswitch, switch II and helix α3 regions. However, superposition of Ran and Sar1 in their GEF complex structures (Figure 5F) reveals that Rcc1 and Sec12 interact with significantly different surfaces of their GTPase substrates to achieve similar outcomes.

Sec12 positions Sar1 for membrane insertion during activation

In vivo, Sec12 is anchored to the ER membrane by its C-terminal transmembrane domain. However, the orientation of the β-propeller domain with respect to the ER membrane is unknown. As activation of Sar1 coincides with the release of its amphipathic α-helix, the ability of the Sec12 cytosolic domain to orient it towards the ER membrane is important to avoid nonproductive aggregation of this helix. The cytoplasmic GEF domain appears able to directly interact with membranes in vitro, as it can perform membrane-dependent activation of Sar1 without its transmembrane domain (Futai et al., 2004). Only 10 residues lie between the structured portion of the cytosolic domain and the predicted transmembrane domain, suggesting that the GEF domain is constrained within a short distance of the ER membrane surface. To identify potential membrane binding surfaces, we analyzed the electrostatic surface potential of the Sar1-Sec12 complex using APBS in Chimera (Baker et al., 2001; Pettersen et al., 2004; Dolinsky et al., 2007). On one surface of the complex, we observed a prominent region of positive charge, while the remainder of the surface was predominantly negatively charged (Figure 6A). Although the residues on the positive surface are not strictly conserved, examination of homologous sequences indicates that the positively charged character of this surface of Sec12 appears to be conserved across evolution (Figure S5). Notably, the C-terminus of the Sec12 β-propeller domain, which is directly connected to the transmembrane domain by a short linker, is also located on the same face of the complex (Figure 6B), This positive surface is therefore a strong candidate for an ER membrane interaction surface. We note this surface represents an edge, rather than one of the two faces, of the β-propeller fold. Interestingly, Rcc1 also uses an edge of its propeller fold to bind to the negatively charged phosphate backbone of nucleosomal DNA (Makde et al., 2010).

Figure 6: The Sec12 β-propeller domain contains a likely membrane-binding surface.

A. Surface rendering of Sar1-Sec12 complex colored by electrostatic potential from −5kT/e (red) to +5kT/e (blue). A substantial positively-charged patch is observed on one face of the complex. The location of the nucleotide binding pocket is indicated by a nucleotide (dark gray) that has been added for visualization purposes. Note that this complex is nucleotide-free.

B. Similar to A, but with transparent surface and ribbon representations of Sec12 (blue) and Sar1 (orange). The N-terminal residues of the Sar1 construct (G24 and K25) are represented as orange spheres, while the C-terminal residue of the Sec12 construct (N344) is depicted as blue spheres.

C. In vitro membrane pelleting assay of Sec12ΔC comparing major-minor liposomes to PC liposomes. While Sec12ΔC exhibits a small amount of pelleting in the absence of synthetic liposomes, it is strongly enriched in the membrane-bound fraction in the presence of major-minor liposomes and slightly enriched in the presence of PC liposomes. (S=supernatant, P=pellet)

D. Quantification of C. Average % membrane-binding between three assays is plotted, the error bars represent the 95% confidence intervals. The value for each replicate is depicted. The negative values arise from subtraction of the average amount of protein that pellets in the absence of liposomes.

E. In vitro membrane pelleting assay of wild-type Sec12ΔC compared with Sec12ΔC[11E] on major-minor liposomes. Both constructs exhibit a small amount of pelleting in the absence of synthetic liposomes, however the 11E mutant is unable to bind to the major-minor liposomes. (S=supernatant, P=pellet)

F. Quantification of E. Average % membrane binding between three assays is plotted, the error bars represent the 95% confidence intervals. The value for each replicate is depicted. The negative values arise from subtraction of the average amount of protein that pellets in the absence of liposomes.

G. Representative, normalized traces from in vitro nucleotide exchange assay measuring the kinetics of Sec12ΔC activation of full-length Sar1 in the presence of synthetic liposomes.

H. Quantification of G. Wild-type exchange rate is similar to the previously reported rate (McMahon et al., 2012), but the 11E mutant is unable to activate FL-Sar1. Error bars indicate the 95% confidence intervals, n = 3.

See also Figure S5.

With this in mind, we tested whether the purified cytosolic domain of Sec12 (lacking its TMD which would normally mediate membrane interactions) could stably interact with membranes using an in vitro membrane pelleting assay. In agreement with our hypothesis that the clustered positive charges serve as the membrane-binding surface, we observed that Sec12ΔC bound robustly to anionic “major-minor” liposome membranes (Futai 2004), but not to neutral phosphatidylcholine (PC) liposomes (Figure 6C, D). To directly probe whether the basic patch (Figure 6A) was responsible for mediating the binding to membranes, we mutated 11 of the 12 positive residues on that surface (Figure 6A, Figure S5) from lysine or arginine to glutamate, altering its overall charge character from positive to negative. Thermal stability analysis indicated this mutant construct was well behaved (Figure S3). While both wild-type Sec12ΔC and the charge-reversal mutant Sec12ΔC[11E] pelleted slightly in the absence of liposomes, only wild-type pelleting was greatly enriched in the presence of major-minor liposomes (Figure 6E, F). Having established that the cytosolic domain of Sec12 binds directly to membranes and having verified the surface that mediates this interaction, we next asked whether direct membrane binding by the cytosolic domain was required for the function of Sec12, catalyzing nucleotide exchange on Sar1 at the membrane surface. Utilizing the same GEF reconstitution system described above (Figure 3E, F), we determined that the charge-reversal mutant Sec12ΔC[11E] was completely deficient at catalyzing nucleotide exchange (Figure 6G, H). When these charge reversal mutations were introduced into the full-length Sec12 protein in vivo, cell viability was maintained (Table S2), indicating the core GEF activity of the mutant remained intact and that limiting the diffusion of the cytosolic domain, via TMD anchoring to the ER, is sufficient for proper orientation. Taken together, these results suggest that the orientation of the Sec12 cytosolic domain at the ER membrane surface is maintained by a cluster of positive charges on one surface and that preservation of this orientation is important for the activation of Sar1.

Upon GTP-binding to Sar1, the resulting conformational change triggers folding of the N-terminal 23 amino acids of Sar1 into an amphipathic helix that inserts into the ER membrane (Antonny et al., 1997; Huang et al., 2001). Intriguingly, Sar1 Gly24, which is the N-terminus of the Sar1 construct used for crystallization, lies on the same face of the Sar1-Sec12 complex as the positive surface and the C-terminus of the Sec12 β-propeller domain (Figure 6B and Supplemental Video S1). Therefore, our structure leads us to a mechanistic proposal for how Sec12 can directly couple Sar1 activation to Sar1 membrane insertion (Figure 7). Due to the orientation of the Sec12 β-propeller domain on the ER membrane, when Sec12 binds to Sar1 and displaces nucleotide, it positions Sar1 so that the Sar1 N-terminus faces the membrane. Upon GTP-binding, Sar1 reveals its N-terminal amphipathic helix which inserts into the ER membrane as Sar1 dissociates from Sec12.

Figure 7: Structural model for Sar1 activation by Sec12 on the ER membrane.

A. Colored in purple, Sar1-GDP (Huang et al., 2001) is cytosolic but samples the membrane surface (Antonny et al., 1997). The ER membrane is depicted as stylized lipids, approximately to scale. All proteins are shown as ribbons with a transparent surface.

B. The Sar1-Sec12 complex (this study, Sar1 in orange, Sec12 in blue) interacts with the ER via its positively charged surface and the transmembrane domain of Sec12. A transmembrane domain and linker (gray) are modeled to represent how Sec12 is attached to the ER membrane. This attachment and the positively-charged patch (Figure 6) provide physical constraints for the cytosolic domain’s interaction with the membrane and position Sar1 with its N-terminus towards the membrane. The N-terminus of Sar1 and the C-terminus of Sec12 are represented as spheres.

C. Membrane-bound Sar1-GTP (Bi et al., 2002) is colored in green. The N-terminal amphipathic helix is modeled in gray to indicate how Sar1 interacts with the ER membrane in its active conformation.

D. As in C, but with Sec23 (yellow) (Bi et al., 2002) bound to Sar1-GTP, portraying recruitment of the first effector in COPII vesicle coat formation.

Discussion

Activation of Sar1 by Sec12 on the ER membrane surface is the essential initiating event of the eukaryotic secretory pathway. Here we have reported the crystal structure of a Sar1-Sec12 complex. This complex represents a pivotal, nucleotide-free intermediate in the transition from the inactive (GDP-bound) to active (GTP-bound) state during GEF-mediated activation. We provide a detailed mechanistic explanation for nucleotide release and propose a mechanism for how Sec12 couples Sar1 activation with membrane insertion of its N-terminal amphipathic helix.

Although many structures of GEF-GTPase complexes have been determined, Sar1 and Sec12 serve indispensable roles to initiate the secretory pathway and possess distinguishing structural features that are vital to their function. The N-terminal amphipathic α-helix is the defining feature of Sar1 and most other Arf family members. In addition to serving as a membrane anchor, this helix has been demonstrated to serve a critical role in membrane remodeling events including induction of membrane curvature and driving membrane fission (Bielli et al., 2005; Lee et al., 2005; Drin and Antonny, 2010; Boucrot et al., 2012; Schwieger et al., 2017). Work from several laboratories has provided crucial structural insights into how Sec7-domain Arf-GEFs exchange nucleotide, and models for how Sec7-domain Arf-GEFs with PH-domains are organized on the membrane surface (Cherfils et al., 1998; Goldberg, 1998; Renault et al., 2003; DiNitto et al., 2007; Aizel et al., 2013; Malaby et al., 2013; Padovani et al., 2014; Galindo et al., 2016; Richardson et al., 2016; Karandur et al., 2017; Das et al., 2019). However, it remains unclear whether the Sec7 domain itself interacts directly with the membrane, and therefore a detailed mechanistic description of how Arf-GEFs deal with insertion of the substrate amphipathic α-helix has not been reported.

Among the known Arf-GEFs, Sec12 is unique as an integral membrane protein. Why did evolution favor a transmembrane domain GEF for Sar1 but peripheral membrane protein GEFs for other Arf family GTPases? One possibility is the location of the GEFs and GTPases. Transmembrane domain proteins localizing to the Golgi complex, endosomes, and plasma membrane are initially synthesized at the ER before trafficking to their destination organelle. Perhaps the transient localization of a plasma membrane or endosomal Arf-GEF at the ER would result in deleterious activation of Arf substrates ectopically at the ER.

We took advantage of the constraint provided by the Sec12 transmembrane domain, together with a striking positively charged region, to identify the likely membrane-interacting surface of the Sec12 β-propeller GEF domain. Our biochemical experiments confirmed that this surface does bind directly to membranes. When bound to Sec12, Sar1 is oriented such that its N-terminus is pointed towards the membrane. Upon GTP binding to Sar1, Sec12 will dissociate while the Sar1 N-terminal amphipathic α-helix takes shape in just the right place to insert into the membrane. We think the orientation of the complex on the membrane is important because optimal positioning would greatly increase the frequency of productive encounters between the folding amphipathic α-helix and the membrane surface. This would bias outcomes towards successful insertion events rather than aggregation-prone exposure of the hydrophobic surface of this helix to the cytoplasm.

Although the proposed membrane binding surface was critical for GEF activity of the isolated cytoplasmic domain in vitro, we found it was not required for viability in the context of the full-length protein in vivo. Our interpretation of this result is that the covalent anchoring of the Sec12 cytoplasmic domain to the ER membrane via its TMD constrains the possible orientations of the complex, and the positive surface of Sec12 provides an additional driving force to direct the N-terminus of Sar1 towards the membrane for insertion. Alternatively, the cationic surface of Sec12 may serve to concentrate Sec12 at regions of the ER locally enriched with anionic lipids, or localized enrichment of anionic lipids may stimulate Sec12 activity.

We observed an additional interesting feature regarding the surface electrostatics of the Sar1-Sec12 complex. The Sar1 nucleotide binding pocket is a small positively charged “bullseye” within the large negatively charged surface of the Sar1-Sec12 complex that faces away from the membrane (Figure 6A). Because nucleotides bear a strong negative charge due to their phosphate groups, we suspect the membrane-distal surface acts as an electrostatic “funnel” (Carpenter and Lightstone, 2016) to guide GTP to its binding pocket in Sar1.

It has been known for some time that cellular K⁺ levels diminish in response to metabolic stress (Findlay, 1994) and depletion of cellular K⁺ to ~25% of normal levels inhibits ER-Golgi transport, likely by blocking exit from the ER (Judah et al., 1989). Sec12 is a tantalizing candidate for the focal point of this regulation of trafficking. The potassium ion-binding K-loop is essential for Sec12 function in cells (McMahon et al., 2012), and our structure demonstrates how the K-loop acts to directly disrupt the Sar1 nucleotide-binding pocket by displacing the switch and interswitch regions.

STAR Methods

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Chris Fromme (icf14@cornell.edu).

Materials Availability

Requests for plasmids and strains generated in this study should be directed to the lead contact.

Data and Code Availability

The X-ray diffraction intensities, structure factors, and coordinates of the refined model have been deposited in the RCSB PDB (Accession code: 6×90). The deposited structure factor file contains both the anisotropically corrected intensities and structure factors (datablock 1) and original uncorrected intensities (datablock 2).

Experimental model and subject details

All recombinant plasmids and yeast strains were generated using standard molecular biology techniques and are listed in the Key Resources Table. Plasmids were constructed using the DH5α strain of E. coli (New England Biolabs). Recombinant proteins were purified from the Rosetta2 strain of E.coli (Novagen). Cell viability assays were performed in the BY4742 strain of S. cerevisiae. Yeast strains were grown in YPD or synthetic dropout media as described below. E. coli strains were grown on Luria broth plates and in Luria broth liquid media (for cloning) or in terrific broth liquid media (for protein expression) as described below.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and Virus Strains
Rosetta 2, E. coli	Novagen	Cat# 71400
DH5α, E. coli	New England Biolabs	Cat# C2987I
Chemicals, Peptides, and Recombinant Proteins
GTP	Thermo Fisher	Cat# R0461
GDP	Sigma-Aldrich	Cat# G7127
5-fluoroorotic acid (5-FOA)	United States Biological	Cat# F5050
Phenylmethylsulfonyl fluoride (PMSF)	MP Biomedicals	Cat# 02195381
GelCode Blue Safe Protein Stain	Thermo Fisher	Cat# 24594
1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC)	Avanti Polar Lipids, Inc.	Cat# 850375
1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE)	Avanti Polar Lipids, Inc.	Cat# 850725
1,2-dioleoyl-sn-glycero-3-phospho-L-serine (DOPS) sodium salt	Avanti Polar Lipids, Inc.	Cat# 840035
1,2-dioleoyl-sn-glycero-3-phosphate (DOPA) sodium salt	Avanti Polar Lipids, Inc.	Cat# 840875
L-α-phosphatidylinositol (PI, Liver) sodium salt	Avanti Polar Lipids, Inc.	Cat# 840042
L-α-phosphatidylinositol-4-phosphate (PI(4)P, Brain) ammonium salt	Avanti Polar Lipids, Inc.	Cat# 840045
1,2-dipalmitoyl-sn-glycero-3-cytidine diphosphate (CDP-DAG) ammonium salt	Avanti Polar Lipids, Inc.	Cat# 870510
Cholesterol (ovine)	Avanti Polar Lipids, Inc.	Cat# 700000
PI(4,5P2) diC16	Cell Signals	Cat# 918
1,1’-Dioctadecyl-3,3,3ˈ,3ˈ-Tetramethylindotricarbocyanine Iodide (DiR’; DilC18(7))	Thermo Fisher	Cat# D12731
Bacto Yeast extract	VWR	Cat# 90004-090
Bacto Tryptone	Thermo Fisher	Cat# 211705
Glycerol	VWR	Cat# BDH1172-1LP
Potassium phosphate monobasic	VWR	Cat# EM-PX1565-5
Potassium phosphate dibasic	VWR	Cat# BDH9266-2.5KG
Isopropyl β- d-1-thiogalactopyranoside (IPTG)	IBI	Cat# IBI-IB02125
Tris(hydroxymethyl)aminomethane base	VWR	Cat# 76348-650
Sodium chloride	VWR	Cat# 97061-268
Imidazole	VWR	Cat# IC102033.5
Dithiothreitol	VWR	Cat# 97061-338
Calf intestinal alkaline phosphatase	Sigma	Cat #P4978
Potassium chloride	Fisher Scientific	Cat# BP366-500
Potassium acetate	VWR	Cat# BDH9254-500G
HEPES	VWR	Cat# IB01133
Magnesium chloride	VWR	Cat# 97061-356
SYPRO Orange	Life Technologies	Cat# S6651
Deposited Data
Sec12(1-354):Sar1(24-190)	This study	PDB:6X90
Sec12(1-354)	McMahon et al., 2012	PDB:4H5I
Sar1-GDP	Huang et al., 2001	PDB:1F6B
Sar1-GTP	Bi et al., 2002	PDB:1M2O
Experimental Models: Organisms/Strains
BY4742 MATα ura3-52 his3-Δ200 leu2-3,112 lys2-801 trp1-Δ901 suc2-Δ9	Brachmann et al., 1998	n/a
BY4743 Diploid, ura3-Δ0/ura3-Δ0, his3-Δ1/his3-Δ1, leu2-Δ0/leu2-Δ0, met15-Δ0/MET15, lys2-Δ0/LYS2 SEC12/sec12Δ::KanMX	Giaever et al., 2002	n/a
BY4742 sec12Δ::KanMX, +pRS416-SEC12	This study	CFY4048
Recombinant DNA
Yeast centromeric vector with LEU2 marker	Sikorski and Hieter, 1989	pRS415
Yeast centromeric vector with URA3 marker	Sikorski and Hieter, 1989	pRS416
Sar1-6xHis in pET21b	McMahon et al., 2012	CFB3379
Sec12(1-354)-6xHis in pET21b	McMahon et al., 2012	CFB3380
Sar1(24-190)-6xHis in pET21b	This study	CFB3440
pRS416 with SEC12 (Sec12 promoter-Sec12-Sec12 terminator)	This study	CFB4021
pRS415 with Sec12-3xHA (Sec12 promoter-Sec12-3xHA-Sec12 terminator)	This study	CFB4042
pRS415 with Sec12[H308A, S309A, F310A]-3xHA	This study	CFB4073
pRS415 with Sec12[T313A]-3xHA	This study	CFB4080
pRS415 with Sec12[A329E, A330D, N331A]-3xHA	This study	CFB4081
pRS415 with Sec12[F310A]-3xHA	This study	CFB4082
pRS415 with Sec12[P13A, Y15A]-3xHA	This study	CFB4085
pRS415 with Sec12[D130A, D131A, Y132A, K134A]-3xHA	This study	CFB4086
pRS415 with Sec12[S150A]-3xHA	This study	CFB4087
pRS415 with Sec12[D70A, D71A, S72A]-3xHA	This study	CFB4112
pRS415 with Sec12[D70A, D71A]-3xHA	This study	CFB4113
pRS415 with Sec12[S72A]-3xHA	This study	CFB4114
pRS415 with Sec12[D71A]-3xHA	This study	CFB4117
Sec12(1-354)[D71A]-6xHis in pET21b	This study	CFB4181
pRS415 with Sec12[K2E, K237E, K238E, K240E, K250E, R258E, K260E, R265E, K294E, K296E, K302E, K305E]-3xHA	This study	CFB4239
Sec12(1-354)[K237E, K238E, K240E, K250E, R258E, K260E, R265E, K294E, K296E, K302E, K305E]-6xHis in pET21b	This study	CFB4248
Software and Algorithms
RAPD pipeline	Northeastern Collaborative Access Team (NE-CAT) facility at the Advanced Photon Source	http://necat.chem.cornell.edu/
XDS	Kabsch, 2010	http://xds.mpimf-heidelberg.mpg.de/
Phaser	McCoy et al., 2007	https://www.phenix-online.org/documentation/index.html
Phenix	Liebschner et al., 2019	https://www.phenix-online.org/documentation/index.html
Coot	Emsley et al., 2010	http://www2.mrc-lmb.cam.ac.uk/personal/pemsley/coot
STARANISO estimation server	Tickle et al., 2018	http://staraniso.globalphasing.org/cgi-bin/staraniso.cgi
SBGrid	Morin et al., 2013	https://sbgrid.org
UCSF Chimera	Pettersen et al., 2004	https://www.cgl.ucsf.edu/chimera/
UCSF ChimeraX	Goddard et al., 2018	https://www.cgl.ucsf.edu/chimerax/
Pymol	Schrodinger, LLC	https://pymol.org/2/
Clustal Omega	Madeira et al., 2019	https://www.ebi.ac.uk/Tools/msa/clustalo/
Prism 8/9	GraphPad	https://www.graphpad.com/scientific-software/prism/
Fiji	Schindelin et al., 2012	https://imagej.net/Fiji
Consurf web server	Armon et al., 2001; Landau et al., 2005; Ashkenazy et al., 2010; Celniker et al., 2013; Ashkenazy et al., 2016	https://consurf.tau.ac.il
Other
MonoQ 5/50 GL	GE Healthcare	Cat# 17516601
Ni-NTA Agarose Resin	Qiagen	Cat# 30210
Superdex 200 Increase 10/300	GE Healthcare	Cat# 28990944
Amicon Ultra-4 Centrifugal Filter Units	Millipore Sigma	Cat# UFC801024
TLA 100.3 rotor	Beckman Coulter	Cat # 349490
Fluorometer	Photon Technology International	Model# QM4
0.1μm Whatman Nuclepore Track-Etched Membranes	Sigma-Aldrich	Cat# WHA800309
0.4μm Whatman Nuclepore Track-Etched Membranes	Sigma-Aldrich	Cat# WHA800282
Mini-extruder	Avanti Polar Lipids, Inc.	Cat# 610000
Filter supports	Whatman	Cat# 230300

METHOD DETAILS

Plasmid construction

Plasmids were constructed using standard molecular biology approaches using the E. coli strain DH5α.

Protein purification

For purification of full-length Sar1, we used a previously reported plasmid (McMahon et al., 2012) using the pET21b (Novagen) backbone that expresses full-length Sar1 from Saccharomyces cerevisiae with a C-terminal hexahistidine tag. For crystallization, we mutated this plasmid to create an N-terminal truncation of the first 23 amino acids (corresponding to the amphipathic helix) of Sar1 (ΔN23-Sar1). Purification of both protein constructs was similar: plasmid was transformed into the Rosetta2 strain of E.coli (Novagen). Two liters of culture were grown in terrific broth for 8-12h at 37°C (until the OD₆₀₀ ~2). The temperature was then reduced to 16°C and 1 hour later IPTG was added to a final concentration of 300μM to induce protein expression. Following overnight expression (14-18h), cells were collected via centrifugation and resuspended in lysis buffer (40mM Tris pH 8, 300mM NaCl, 10% glycerol, 10mM imidazole, 1× PMSF, and 1mM DTT). Cells were lysed by sonication and ΔN23-Sar1 −6×His or full-length Sar1-6×His was purified by nickel affinity chromatography (Ni-NTA resin, Qiagen) and eluted with elution buffer (40mM Tris pH 8, 300mM NaCl, 10% glycerol, 250mM imidazole, and 1mM DTT). The sample was further purified using anion exchange chromatography with a linear gradient (buffer A: 20mM Tris pH 8, 1mM DTT; buffer B: 20mM Tris pH 8, 1M NaCl, 1mM DTT) on a MonoQ column (GE Healthcare). The peak fractions were collected, concentrated, and frozen in liquid nitrogen prior to storage at −80°C.

A similar approach was used to purify the cytoplasmic domain of Sec12. We used a previously reported plasmid (McMahon et al., 2012) with the pET21b (Novagen) backbone that expresses a C-terminal truncation of Sec12 (Sec12ΔC residues 1-354, lacking the transmembrane domain and a short ER lumenal domain) from Saccharomyces cerevisiae with a C-terminal hexahistidine tag. Mutations were introduced via a two-step PCR protocol. For all Sec12ΔC constructs, the purification protocol used was identical to that of Sar1, except 200mM KCl was added to the nickel purification buffers and KCl was added to the MonoQ fractions to a final concentration of ~130mM.

For nucleotide exchange assays, proteins were further purified via gel filtration chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare) prior to usage. Buffer for full-length Sar1: 10mM Tris pH 8.0, 150mM NaCl, 1mM DTT; buffer for Sec12ΔC constructs: 10mM Tris pH 8.0, 100mM NaCl, 100mM KCl, 1mM DTT (Note: two overlapping peaks were observed for Sec12 and both were combined). Proteins were aliquoted, flash frozen, and stored at −80°C. Prior to use, aliquots were thawed, and pre-spun for 3 minutes at ~20,000g.

See Table S4 for plasmid information.

Sar1-Sec12 Complex formation

Purified Sec12ΔC and ΔN23-Sar1 (previously frozen fractions from the anionic exchange step) were mixed in a 1:3 molar ratio and incubated overnight at room temperature with calf intestinal alkaline phosphatase (Sigma). After 12-18h, the sample mixture was subjected to size exclusion chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare) in 10mM Tris pH 7, 250mM KCl, 1mM DTT. The trace displayed two dominant peaks corresponding to ΔN23-Sar1:Sec12ΔC complex and free ΔN23-Sar1 (Figure S1). The fractions containing the complex were pooled and concentrated to ~34mg/mL in a 10kDa molecular weight cutoff concentrator from Millipore.

Crystallization

An Art-Robbins Phoenix robot was used to set up 96-well crystallization screens containing several concentrations of purified ΔN23-Sar1:Sec12ΔC at room temperature and 4°C using the sitting drop vapor diffusion method. Several initial crystallization hits were optimized but found to contain only Sec12. Diffraction data used for Sar1-Sec12 structure determination was collected from a single crystal. This crystal grew from a 5mg/mL protein solution mixed with 200mM ammonium sulfate, 100mM MES pH 6.5, 30% PEG 5000 MME after 3 weeks at 4°C. The crystal was washed in place with a solution of 50% glycerol as cryoprotectant then flash frozen in liquid nitrogen.

Diffraction data collection, structure solution, model building, and refinement

X-ray diffraction data were collected on NE-CAT Beamline 21-IDE using an Eiger 16-M detector. X-ray data were processed by the RAPD pipeline (http://necat.chem.cornell.edu/) which utilizes XDS for scaling and merging (Kabsch, 2010). The structure was solved by molecular replacement using Phaser (McCoy et al., 2007) in Phenix (Liebschner et al., 2019), using previously published structures of yeast Sec12 (PDB:4H5I) (McMahon et al., 2012) and hamster Sar1-GDP (PDB:1F6B) (Huang et al., 2001) as search models.

Iterative rounds of model rebuilding in Coot (Emsley et al., 2010) and refinement in Phenix (reciprocal-space, real-space, and ADP parameter refinements) were performed to yield a model with an R_free value of 0.31 and average atomic B-factor of 96 that we were unable to improve further. We noted that although the CC^1/2 metric indicated the resolution of the data extended to at least 2.26Å, examination of I/σ values indicated the data was weak for many reflections corresponding to frequencies beyond 2.8Å. Subsequent analysis of the data using STARANISO (Tickle et al., 2018) confirmed the presence of significant anisotropy. We therefore used anisotropically corrected structure factors generated by STARANISO for further model refinement, keeping the same set of test reflections used for R_free calculations during the earlier stages of refinement. The anisotropic correction significantly improved the appearance of the electron density map, enabling us to model ~70 additional ordered water molecules. Further refinement (reciprocal-space, TLS, and ADP parameter refinements) resulted in a model with with an R_free value of 0.26 and average atomic B-factor of 66. Data and refinement statistics for both isotropic (spherical) and anisotropic (ellipsoidal) scaling approaches are reported in Table 1. Examples of electron density are shown in Figure S2.

Liposome preparation

Synthetic liposomes were prepared using various lipid components (Table S3). Lipids were mixed, dried by rotary evaporation, rehydrated overnight in HK buffer (20mM HEPES pH 7.4, and 160mM KOAc) at 37°C, then extruded through 100nm or 400nm filters (Whatman) to generate homogenous liposomes. Extruded liposomes were stored at 4°C.

Pelleting assays

In 40uL reactions, loaded ~4ug of protein along with 500μM of 400nm liposomes in HK buffer. Incubated for 10 minutes at RT. Samples were ultracentrifuged using a TLA100.3 rotor (Beckman Coulter) at roughly 130,000g for 15 minutes at 4°C. The supernatant and pellet fractions were recovered and evaluated separately via SDS-PAGE. Each assay was performed at least three times. The gels shown are representative of all three experiments. Gel densitometry was performed using FIJI (Schindelin et al., 2012). The percent of membrane-bound protein is calculated by dividing the intensity of the pelleted bands by the total intensity of protein in both the supernatant and pelleted fractions. Then the average amount of protein that pellets in the absence of liposomes is subtracted from each sample. The quantification is based on the averages of all three experiments. All direct comparisons were performed using the same batch of liposomes to account for batch variability.

Nucleotide exchange assays

GEF assays were performed essentially as described previously (Futai et al., 2004; Richardson and Fromme, 2015). Intrinsic tryptophan fluorescence of Sar1 was measured in real time using a fluorometer (Photon Technology International). Assays were carried out under the following conditions: 2μM Sar1, 10nM Sec12ΔC construct, 200μM GTP, 333μM major-minor liposomes at 100nm (Futai et al., 2004) in HKM buffer (20mM HEPES pH 7.4, 160mM KoAc, 5mM MgCl₂) for 10 minutes at 30°C.

All assays were performed in triplicate. The raw traces were cropped to contain just the exchange section. These traces were fit by a single exponential using Prism 8 (GraphPad). The rate constant (k) of the intrinsic Sar1 exchange was subtracted from all other rate constants. The resulting rate constant was divided by the [GEF] to calculate the exchange rate (k_exchange). The average of 3 exchange rates was plotted along with the 95% confidence interval. For trace visualization, one representative trace is shown after being normalized to between 0 and 1.

Yeast complementation assays

The BY4741/2/3 strain background (Brachmann et al., 1998, Giaever et al., 2002) was used to construct the SEC12 shuffling strain (CFY4048), in which cell viability was maintained in a sec12Δ::KANMX background by the presence of a SEC12:URA3 centromeric plasmid. A SEC12-3xHA:LEU2 centromeric plasmid was subjected to site-directed mutagenesis and then transformed into CFY4048. These plasmids were built using established centromeric yeast plasmids (Sikorski and Hieter, 1989). Cells were grown overnight at 30°C in -LEU media to saturation, then plated in a dilution series onto -LEU and 5-FOA plates and grown for 2 days at 30°C and 37°C before being imaged. Nonfunctional sec12 alleles are unable to support cell growth. See Table S5 for yeast strain information.

Thermal stability analysis

Wild-type, D71A, and 11E mutant Sec12ΔC constructs were diluted to 10 μM final concentration in HKM buffer containing SYPRO Orange (1:2000 dilution of a 5000X concentrate). The fluorescence signal was measured while the temperature was slowly raised using a Roche LightCycler 480. Each trace represents the average of three replicates and the shaded regions indicate the 95% confidence intervals.

Surface analysis

Conservation analysis was carried out using the Consurf web server at default settings (Armon et al., 2001; Landau et al., 2005; Ashkenazy et al., 2010; Celniker et al., 2013; Ashkenazy et al., 2016). The electrostatic surface potential was calculated using the APBS plugin for Chimera (Pettersen et al., 2004) and the map was colored on a red (−5kT/e) to blue (+5kT/e) scale (Baker et al., 2001; Dolinsky et al., 2007), after charges were added to the structure using the Add Charges feature in Chimera (using the AMBER ff14SB setting for standard residues and AM1-BCC for others).

Software

All software used for the above analyses is maintained in our laboratory by SBGrid (Morin et al., 2013). Molecular graphics created with UCSF Chimera and Chimera× (Pettersen et al., 2004; Goddard et al., 2018). The animated video, depicting transitions between Sar1 conformations in three different crystal structures (GDP-bound, nucleotide-free, and GTP-bound), was generated using the “Morph” feature in Pymol (Schrödinger, LLC). Note that the conformational pathway generated by morphing does not necessarily represent the actual pathway of the GEF reaction, but is helpful for highlighting the structural differences between the states observed in the crystal structures.

Sequence alignment

The multiple sequence alignment shown in Figure S5 was performed using Clustal Omega with default settings (Madeira et al., 2019).

Quantification and statistical analysis

Software used for the analyses described in the STAR methods is maintained in our laboratory by SBGrid (Morin et al., 2013). Molecular graphics created with UCSF Chimera and Chimera× (Pettersen et al., 2004; Goddard et al., 2018). FIJI was utilized for gel analyses (Schindelin et al., 2012). Charts and graphs were created using Prism 9 (GraphPad).

Supplementary Material

2

Legend for Supplemental Video S1 (Related to Figure 4)

Inactive Sar1 is depicted in orange with bound GDP (yellow) and a magnesium ion (green). The view zooms in to focus on the bound magnesium ion, which is coordinated through a hydrogen bond network that includes Sar1 residue Asp73 (hamster residue Asp75). While the distance between Asp73 and the magnesium ion is measured at 3.8 Å, Asp73 directly hydrogen bonds with a water molecule which in turn hydrogen bonds with the magnesium ion (water molecules are not shown). After zooming out, Sec12 approaches and binds to Sar1-GDP. The K loop directly displaces Asp73, disrupting the hydrogen bond network that holds the magnesium ion in place. As the magnesium ion leaves, Sar1 undergoes significant structural rearrangements in its switch I and interswitch regions, while the switch II region moves as a rigid body towards the center of Sec12. The structural rearrangements and the destabilization from the absence of the magnesium ion cause GDP to dissociate. After zooming out again, the entire Sar1-Sec12 structure is visible. The view centers on residue Ile38 of Sec12 and then displays the 3.2 Å measured distance between it and residue Asp73 of Sar1. This distance is appropriate for these residues which are in close contact but do not clash. This is in contrast to the hypothetical distance measured between the GDP/GTP-bound Sar1 conformations which would be approximately 0.5-1.5 Å (not shown). The view zooms out to reveal the entire Sar1-Sec12 structure and rotates 90°. Spheres appear to denote the C-terminal residue of Sec12 (residue 344, which is adjacent to its TMD) and the N-terminal residue of Sar1 (residue 24, which is close to the amphipathic helix). The complex rotates back 90° to show the orientation that it is likely to adopt on the ER membrane, with both termini residing on the same face directed downwards. Then, the view centers on the nucleotide binding pocket of Sar1 as GTP moves into position. Upon GTP binding, structural rearrangements in Sar1 occur and Sec12 dissociates. Ultimately, nucleotide binding is stabilized by a magnesium ion which is coordinated through a hydrogen bond network involving Asp73 of Sar1. In the GTP-bound conformation, the switch I region of Sar1 is structured and visible as a loop at the forefront of the frame. Sar1-GDP (PDB:1F6B) and Sar1-GTP (PDB:1M2O) were the starting and ending models, respectively (Huang et al., 2001; Bi et al., 2002)

Highlights.

• The structure of nucleotide-free Sar1 GTPase bound to its GEF, Sec12, was solved

• The structure illuminates the mechanism of nucleotide exchange

• An unexpected membrane-binding surface on Sec12 cytosolic domain is proposed

• Sec12 may orient Sar1 for membrane insertion of its amphipathic helix