Mechanism of actin filament nucleation by the bacterial effector VopL.

Vibrio parahaemolyticus protein L (VopL) is an actin nucleation factor that induces stress fibers when injected into eukaryotic host cells. VopL contains three N-terminal Wiskott-Aldrich homology 2 (WH2) motifs and a unique VopL C-terminal domain (VCD). We describe crystallographic and biochemical analyses of filament nucleation by VopL. The WH2 element of VopL does not nucleate on its own and requires the VCD for activity. The VCD forms a U-shaped dimer in the crystal, stabilized by a terminal coiled coil. Dimerization of the WH2 motifs contributes strongly to nucleation activity, as do contacts of the VCD to actin. Our data lead to a model in which VopL stabilizes primarily lateral (short-pitch) contacts between actin monomers to create the base of a two-stranded filament. Stabilization of lateral contacts may be a common feature of actin filament nucleation by WH2-based factors.

Actin cytoskeletal dynamics are important in numerous cellular processes, including migration, division and maintenance of morphology[1-3]. These functions all require rapid assembly of new actin filaments de novo from actin monomers. However, actin assembly is intrinsically slow, due to kinetic and thermodynamic barriers to forming the actin dimers and trimers needed to initiate filament growth[4]. Cells have therefore developed specialized factors to stabilize these actin nuclei and thus catalyze filament assembly[4]. Three classes of so-called actin nucleation factors have been identified: Arp2/3 complex, formin proteins and WASP Homology domain 2 (WH2)-based nucleators[5-11]. Arp2/3 complex binds to the side of an existing filament and initiates growth of a new filament from its actin related protein 2 (Arp2) and Arp3 subunits[12,13]. Alignment of the actin homologs Arp2 and Arp3 to resemble the first two monomers at the base of an actin filament is believed to be a key aspect of Arp2/3-mediated nucleation[14]. Formin proteins nucleate linear filaments through a conserved formin homology 2 (FH2) domain, which is also thought to organize three actin monomers into a structure that resembles the base of an actin filament[15,16]. After nucleation, formins remain associated with the growing filament barbed end as new monomers are added[5]. Several mechanisms have been proposed to account for this processive activity[15-18]. The WH2-based actin nucleators have been identified most recently, and include Spire, Corbon blue (Cobl) and Leiomodin (Lmod)[7-9,11]. These proteins all contain a series of tandem WH2 motifs, each capable of binding an actin monomer. These proteins are also thought to act through organization of multiple actin monomers into a stable structure that resembles an actin filament. However, the nature of that nucleus—whether the monomers are aligned longitudinally to resemble a single strand of the paired actin filament, or laterally to resemble both strands—appears to differ between the various factors, due to differences in the WH2 domains, the linkers between them and the presence of additional actin-binding elements[7-10,19].

During infection, many bacterial and viral pathogens hijack the actin cytoskeleton of eukaryotic host cells to increase their pathogenicity or mobility[20-22]. In many cases, these pathogens target pathways that control actin filament nucleation[23-28]. Several bacteria have evolved WH2-based actin nucleation factors, which they inject into host cells through type III secretion systems to promote actin assembly. These factors include VopL, VopF, TARP and Sca2[29-33]. VopL and its homolog VopF are effector proteins of the gram negative bacteria Vibrio parahaemolyticus and Vibrio cholera, respectively[30,32,34]. When introduced into eukaryotic cells, VopL and VopF induce actin stress bers and filopodia, respectively. In vitro, both proteins potently induce nucleation of actin filaments that, at least for VopL, grow at their barbed ends[30]. Both consist of one or two proline-rich (P) motifs and three tandem WH2 motifs, followed by either a VopL/VopF C-terminal domain (VCD) that lacks homology to known proteins. Mutagenesis of the WH2 motifs impairs the ability of VopL to induce stress fibers in cells, implicating these elements in actin nucleation[30]. The mechanisms by which the VopL/VopF WH2 motifs and C-terminal domain assemble an actin nucleus remain unclear, as does the relationship between these bacterial proteins and eukaryotic WH2-based nucleation factors.

Here we set out to understand the mechanism of VopL-mediated actin nucleation. To achieve this, we used a combination of biochemical and crystallographic analyses. We show that although the individual WH2 domains can bind actin with nM affinity, the (WH2)3 element (W3) alone cannot nucleate filaments, and requires the VCD for activity. The VCD is a high affinity dimer in solution and forms a U-shaped dimeric structure in the crystal. Dimerization dramatically enhances filament nucleation by W3. The VCD alone retains some nucleation activity, which can be decreased by mutation of exposed surfaces, suggesting that the domain contacts actin during nucleation. Our data suggest a model for VopL-mediated nucleation in which contacts of both the WH2 motifs and the VCD stabilize inter-strand contacts between the initial two or more actin monomers to create a filament nucleus.

Results

WH2 motifs and VCD are both needed for actin nucleation

Each WH2 motif in VopL contains both the N-terminal hydrophobic residues and C-terminal basic sequence that are important for binding to actin monomers in eukaryotic WH2 motifs[35]. We used isothermal titration calorimetry to measure the binding affinity of each motif (named WH2a, WH2b and WH2c from N- to C-terminus, Fig. 1a) for monomeric actin (Supplementary Fig. 1). The dissociation equilibrium constants (KD) range from 5 to 15 nM, explaining the nM potency of VopL in actin assembly assays[30], and indicating little difference between the ability of the motifs to bind actin. Note that these assays were performed in low salt buffer to prevent actin polymerization (see Methods), and affinities could be somewhat different in the presence of higher KCl and MgCl2 concentrations.

Figure 1

Both the WH2 motifs and VCD contribute to actin nucleation by VopL. (a) Domain structures of VopL constructs used in this work. All constructs lack the N-terminal secretion sequence needed for VopL translocation into host cells, whose deletion does not affect activity (not shown). P: Proline-rich motif. WH2: WASP Homology 2 domain. VCD: VopL C-terminal domain. GGS: (Gly-Gly-Ser) linker. The half-time (t1/2) of actin assembly for each VopL protein is given by the average and standard deviation measured in three experiments. Assays contained 50 nM or 5 nM VopL proteins, as indicated. (b) Actin assembly assays contained the indicated concentrations of VopL constructs W3-C, W3 or W4. (c) Actin assembly assays contained 50 nM W3-C, W2-C, W1-C or VCD. Figures in this manuscript were all generated using Adobe Illustrator CS2.

Linear WH2 arrays in other nucleation factors such as Spire and Cobl have substantial activity. But a VopL fragment containing only the three WH2 motifs without the VCD (construct W3) was completely inactive in pyrene-actin assembly assays (Fig. 1b, half times of actin assembly (t1/2) are tabulated in Fig. 1a). High concentrations of W3 inhibited assembly, likely by sequestration of actin monomers (Supplementary Fig. 2). By contrast, the VCD alone retained some actin assembly activity, albeit with ~100-fold lower potency than W3-C, which contains the three WH2 motifs and the VCD (Figs. 1c, a; 500 nM VCD has activity similar to 5 nM W3-C (not shown)). To further dissect the roles of the individual WH2 motifs, we compared the activity of constructs with zero, one, two or three WH2 motifs (Figs. 1a, c; constructs VCD, W1-C, W2-C, W3-C, respectively). In this series, activity decreased in the following order: W3-C > W2-C ≫ VCD > W1-C. Thus, proteins containing three or two WH2 motifs had appreciably higher activity than those with fewer. A single WH2 motif appears to inhibit activity relative to the bare VCD (see below). Although these activities could in principle reflect differences in rates of actin filament elongation (e.g. by binding of the WH2 motifs to the filament barbed end[36]), Dominguez and colleagues observed that VopL did not change elongation rates in single filament assays[37]. Thus, the differences in activity likely reflect differences in nucleation potency. Together, these data demonstrate that actin binding by the WH2 motifs alone is insufficient to promote nucleation, and both the WH2 motifs and VCD play important roles in VopL function.

The VopL VCD is a U-shaped dimer stabilized by a coiled-coil

The W3-C and VCD proteins form high affinity dimers in solution, based on multi-angle laser light scattering (MALLS) analyses (Fig 2a). We crystallized the VCD in space group P212121 with two molecules in the asymmetric unit, and determined its structure to 2.36 Å resolution. Data collection and structure refinement statistics are listed in Table 1. Both molecules in the asymmetric unit are well-defined in the electron density map, except for residues 323-338. The VCD is mainly α-helical except for two anti-parallel β strands formed by residues 355-360 and 364-369 (Fig. 3a). Each subunit is composed of three structural regions, which we name the “arm”, “base” and “C-helix”, respectively, based on the shape of the structure shown in Figure 3a. The base consists of residues 244-278 and 395-456, and forms an elongated cluster of five α-helices. The arm is essentially an insert in the base that consists of residues 283-384. These form a bundle of three helices capped distally to the base by the two-stranded β-sheet, and with an additional perpendicular helix proximal to the base. Finally, the C-helix (residues 462-484) emerges perpendicularly to the end of the arm opposite to the base.

Figure 2

Dimerization is necessary for actin nucleation by the VopL WH2 motifs. (a) Size exclusion chromatography—multi-angle laser light scattering (MALLS) analysis of VopL constructs W3-C (red) and VCD (blue). Normalized scattered light in the chromatographic elution (right y-axis) is superimposed on the molecular weight distribution (left y-axis). (b) MALLS analysis of VCDΔh. Data plotted as in (a). (c) Actin polymerization assays performed with 4 μM actin and 50 nM W3-C, W2-C, W1-C, VCD, W3-GST, W2-GST or W1-GST.

Table 1

Data collection, phasing and refinement statistics

	Native VopL			Selenomethione-VopL
Data collection
Space group	P212121				P212121
Cell dimensions
a, b, c (Å)	55.15	89.81	102.98		56.19	88.64	101.36
a, b, c (°)	90	90	90		90	90	90
				Peak	Inflection			Remote
Wavelength	0.9792			0.9794	0.9795			0.9717
Resolution (Å)	50.0-2.30			50.0-3.10	50.0-3.17			50.0-3.21
Rmerge	0.034(0.618)			0.074(0.713)	0.075(0.788)			0.068(0.665)
I/σ(I)	26.4(1.4)			64.2(5.1)	40.5(3.8)			43.7(4.5)
Completeness (%)	99.1(99.0)			99.8(100)	99.9(100)			99.9(100)
Redundancy	2.8(2.5)			23.5(24.2)	11.8(11.9)			11.7(12.1)
Refinement
Resolution (Å)	29-2.30
No. reflections	23040
Rwork/Rfree	0.221(0.276)/0.275(0.334)
No. atoms
Protein	3510
Ligand/ion	9
Water	36
B-factors
Protein	88.92
Ligand/ion	115.4
Water	64.9
R.m.s deviations
Bond lengths (Å)	0.003
Bond angles (°)	0.564

One seleno-methionine and one native crystal were used for data collection.

Values in parentheses are for highest-resolution shell.

Figure 3

Structure of the VCD dimer. (a) Ribbon diagram of the structure; monomers are colored blue and green. The N- and C-termini of each monomer are indicated. Red dashed lines indicate regions not observed in the electron density map for the green monomer (analogous regions not shown in the blue monomer). Boxed regions are enlarged in (b) and (c), which show detailed views of dimer interface. Black dashed lines indicate hydrogen bonds. In all panels sidechains of residues discussed in the text are shown as sticks.

There are two potential dimers in the asymmetric unit of the VCD crystals (Figs. 3a and S3). One is formed by coiled-coil interactions between the C-helices of the two subunits, and by contacts between the adjacent ends of the base (Fig. 3a). The second is formed by contacts between the arm from one monomer and the base from another monomer (Supplementary Fig. 3). The former dimer has an appreciably larger (1065 Å2 versus 780 Å2 buried) and more hydrophobic intermolecular interface, and is predicted by PDBePISA (Protein Interfaces, Surfaces and Assemblies)[38] to be the more probable of the two possibilities (complexation significance score of 0.2 versus 0.0). Moreover, a VopL protein lacking the C-helix (VCDΔh, residues 247-460) is monomeric in solution by MALLS analysis (Fig. 2b). Thus, both the structure and mutagenesis data support the functional relevance of the coiled-coil mediated dimer of Figure 3a, and we base our further analyses on it.

The VopL dimer forms a shallow U shape, with a wide platform formed by end to end association of the base, flanked by the perpendicular arm of each subunit (Fig. 3a). The C-terminal coiled coil extends perpendicularly from the center of the base. The N-terminus of the structure is at the base-arm elbow; the WH2 element would emerge from the VCD at this location. As discussed below, a 24 residue linker connects the first ordered residue in the structure with the C-terminus of WH2c. At the intermolecular interface a hydrophobic pocket formed by residues Thr422, Ala426, Leu465 of one molecule partially surrounds the sidechain of Tyr462 in the neighboring molecule (Fig. 3b). The sidechains of Lys421 and Gln418 hydrogen bond to those of Glu417 and Thr422 from the other chain. The C-terminal coiled coil is mediated by packing of the Leu465, Val469, Leu472, Leu476, Leu479 and Leu483 sidechains of the two monomers (Fig. 3c). The Arg468 sidechain also hydrogen bonds to the Glu473 sidechain across this interface.

The base is very similar in structure for each molecule in the asymmetric unit, with all-atom root mean square deviation (rmsd) of 0.36 Å. The same is true of the arm, with rmsd of 0.85 Å. However, the organization of the two domains differs between the two subunits due to rigid body movement about the interdomain hinge (residues 279-283 and 384-395, Supplementary Fig. 4). Thus, in solution VopL likely samples a range of conformations involving fluctuations in the relative orientation of the base and the arm. This notion is consistent with small angle x-ray scattering data by Dominguez and colleagues that suggest an average co-linear organization of the base and arm in solution [37].

Dimerization and actin contacts of the VCD are important for activity

Compared to the inactive W3 protein, the highly active W3-C protein is both dimeric (Fig. 2a) and contains additional potential contact sites to actin. To better dissect these two contributions to activity we performed three different experiments: artificial dimerization of WH2 proteins, mutagenic monomerization of W3-C and increasing the number of WH2 motifs in a linear sequence. In the first experiments, we replaced the C terminal domain of WH2-containing constructs with glutathione S-transferase (GST), which forms high affinity dimers[24]. At a concentration of 50 nM, the engineered W3-GST protein exhibits substantial nucleation activity, which is comparable to that of W2-C (Fig. 2c). Both W2-GST and W1-GST are inactive at this concentration. Similar behaviors are observed in N-terminal GST fusions (data not shown). Thus, dimerization alone can greatly increase the activity of W3, which is completely inactive as a monomer even at high concentrations. Nevertheless, W3-GST and W2-GST remain appreciably less active than their native counterparts dimerized by the VCD. In the second experiments we found that while the activity of the monomeric W3-CΔh is substantially below that of the dimeric W3-C counterpart, the former protein is still more active than W3 alone (Fig. 4a). This residual activity does not appear to result from residual dimerization, as a W3-CΔh protein that also contains K421E and T422A mutations (W3-CΔhKT) designed to weaken dimerization contacts between the base is similarly active (Supplementary Fig. 5). Finally, we generated a W4 construct by adding an additional WH2 motif to the N terminus of W3. As shown in Figure 1b, W4 proved to be completely inactive. Together, the data suggest that a specific spatial organization of the WH2 motifs, rather than merely their total number, is likely important for function. Moreover, while dimerization of the WH2 element is necessary for high activity, the VCD appears to contribute in ways beyond mere dimerization, perhaps through an ability to bind and orient actin molecules to favor nucleation (see below). Thus, the data imply a specific organization of actin in the nucleus, generated by the WH2 elements juxtaposed by the VCD.

Figure 4

The VCD contributes to actin assembly activity by dimerization and contacts to actin. Assays were performed with 4 μM actin and 50 nM (a, b) or 5 nM (c) of the indicated VopL proteins. (a) Comparison of W3-C, W3-CΔh and W3. (b) Comparison of W3-C with the VCD arm mutants, W3-CKRR and W3-CDVP. (c) Comparison of W3-C with the VCD base mutants, W3-CKYR, W3-CYR and W3-CEDE.

To test the hypothesis that the VCD might contact actin, we mutated residues at the distal end of the arm and at the base of the VCD “U” structure. Residues Lys323, Arg347 and Arg354 form a conserved basic patch at the distal end of the arm (Fig. 3a). Mutations of these three residues to glutamic acid in W3-C (construct W3-CKRR) or the VCD alone (construct VCDKRR) substantially decreased activity (Figs. 4b, Supplementary Fig. 6, respectively). We also mutated to glycine the three residues Asp326, Val327 and Pro333 in the loop Tyr322-Gly338 near the end of the arm (constructs W3-CDVP and VCDDVP, Fig. 3a), which is not observed in the electron density and is likely flexible in solution. These mutations also produced a small decrease in activity (Figs. 4b and S6). Thus, the distal end of the arm appears to contact actin during nucleation.

The surface of the base that faces into the VCD “U” consists of three distinct patches: a basic patch containing residues Lys421, Tyr425, and Arg428; an acidic patch containing Glu408, Asp413, and Glu417; and a mixed patch containing Leu249 and Glu251 (Supplementary Fig. 7). Mutation of the first patch (KYR to AAD (W3-CKYR) or to KAD (W3-CYR)), or second patch (EDE to KKA (W3-CEDE)) produced small but reproducible decreases in activity (Fig. 4c). Mutation of the third patch had no effect on activity (data not shown). Thus, the base may also contact actin during nucleation, although these interactions do not appear as important for activity as those to the arm.

Multi-angle laser light scattering analyses indicate that the wild type and mutant proteins are largely (and similarly) dimeric at concentrations as low as 500 nM (Supplementary Fig. 8a). In addition, the activities of several VopL proteins are proportional to concentration between 5 nM and 50 nM, suggesting the proteins are mostly dimeric even at low concentrations (Supplementary Fig. 8b). These data support the argument that the mutations perturb contacts to actin rather than causing loss of dimerization. Together, our biochemical analyses suggest the VCD likely plays a direct role in organizing actin monomers in the VopL nucleus, in addition to dimerizing and orienting the WH2 elements.

VopL activity is sensitive to the length of linker between WH2c and C

It is somewhat paradoxical that W1-C has lower activity than both W2-C and VCD (Fig. 1c). That is, adding one WH2 motif to the VCD is inhibitory but adding two WH2 motifs is stimulatory. We hypothesized that this effect might arise because the natural 24-residue linker between the C-terminus of WH2c (Thr223, of the basic sequence LRKT) and the beginning of the VCD (Arg247) is too short, causing sub-optimal orientation of the initial bound actin monomer(s). To test this idea, we inserted 9- and 40-residue linkers into W1-C, to give a total of 33 and 64 residues, respectively, between the WH2c motif and VCD (W1-9-C and W1-40-C, Fig. 1a). The linker in W1-40-C is identical in length to that between WH2b and the C-terminus of W2-C. As shown in Figure 5, increasing linker length in W1-C progressively increases activity. In related experiments, when the WH2c motif in W2-C or W3-C is replaced by an equal length Gly-Gly-Ser linker (to give constructs W1-L-C and W2-L-C), activity decreases only modestly. Thus, a smaller number of WH2 motifs (e.g. W3-C versus W2-L-C) can be largely compensated for by increasing the spacing of those motifs from the VCD. These results are consistent with the idea that the natural linker between WH2c and C is too short (in one or both molecules in the dimer), likely preventing optimal organization of actin monomers for nucleation activity.

Figure 5

Increasing the linker between WH2c and the VCD increases actin assembly activity. Assays contained of each 50 nM VopL protein, W3-C, W2-C, W1-C, W2-L-C or W1-L-C.

Discussion

Our data demonstrate that multiple WH2 motifs and the VCD of VopL are both required for high potency actin filament nucleation. Dimerization, either by the VCD or artificially by GST, is necessary for the WH2 motifs to nucleate. The VCD itself also appears to contact actin, since it alone can nucleate weakly, and mutagenesis of conserved residues in the base and arm decreases activity. These key data, along with the geometry of the VCD structure, suggest a preliminary model for the actin nucleus assembled by VopL. A central concept in this model is that the WH2 motifs recruit actin monomers, while the VCD provides organization that leads to a productive nucleus.

The requirement for dimerization of the WH2 element suggests that in the nucleus, at least one actin monomer is contributed or recruited by each subunit. An individual multi-WH2 protein should be able to assemble a single strand of the nascent paired actin filament (i.e. stabilize long-pitch contacts between actin monomers). But such activity is evidently insufficient for nucleation by VopL. This observation suggests that in VopL dimers, the paired WH2 elements act by stabilizing lateral (short-pitch) actin-actin contacts. Thus, the nucleus should minimally contain a short-pitch actin dimer.

How might such a structure interact with the VCD? One possibility is suggested by geometric considerations. In a short pitch actin dimer, the two subunits are axially displaced on average by 27.5 Å[39]. In the VCD, the distal end of the arm is ~30 Å from the platform created by the base. Thus, the axial displacement of an actin dimer could be accommodated by only small changes in the structural organization of the VCD. In this configuration the penultimate actin monomer would contact the distal end of the arm and the terminal monomer would contact the base (Fig. 6). While our image in Figure 6 has the plane of the terminal monomer perpendicular to the plane of the VCD, other orientations with more acute angles are also possible. Such an arrangement would be consistent with several observations. First, the VCD alone has nucleation activity, suggesting that it may contact both actins in a short pitch dimer. Our mutagenesis data (Fig. 4b) suggest that residues at both the distal end of the arm and the VCD base formed by base contribute to nucleation activity, probably through contacts to actin. Second, in order for WH2c to bind this penultimate actin, the WH2c-VCD linker would need to extend from the N-terminus of the VCD, around the arm to its tip, a distance of ~50 Å, and then approximately half the length of an actin monomer to the N-terminus of the WH2 motif[35], ~28 Å. While these distances are obviously very approximate, they would be consistent with the idea that the natural 24 residue linker (~80 Å fully extended) may be poorly able to position a WH2c-bound actin on the VCD, explaining the low activity of W1-C, and the progressive increase in activity as the linker is lengthened (Fig. 5). In native VopL, the penultimate monomer may be recruited by WH2b. Third, SAXS analyses from Dominguez and colleagues on the W1-C:actin complex suggest that an ordered actin contacts the base [37]. This terminal monomer could be recruited by the WH2c of the opposing VCD. Finally, the short-pitch dimer bound to the VCD would be further stabilized by additional actin monomers recruited by additional WH2 motifs, consistent with the activity series W3-C > W2-C > W1-L-C (Fig. 5). While this model remains speculative, it makes testable predictions that will be explored in future work.

Figure 6

Model for a minimal actin nucleus assembled by VopL. VCD is indicated by blue blocks with ribbons representing the coiled coil. Initial actin monomers are indicated by yellow boxes. Actin-bound WH2c motifs are shown as red cylinders. Red lines indicate linkers between WH2c and VCD. Additional actin monomers assembled by WH2a and WH2b would bind to the upper surfaces of the WH2c-bound actins, making additional short- and long-pitch contacts, further stabilizing the assembly. Actins are implicitly shown with barbed ends directed away from the VCD, but our current data do not speak to the orientation of actin monomers in the VopL nucleus.

The stabilization of short-pitch contacts in the nascent actin filament appears to be a common feature of WH2-based nucleation factors, both prokaryotic and eukaryotic. In Cobl and Lmod, the stoichiometry of actin binding and linker length dependencies of activity have led to models in which WH2 and other motifs stabilize a short-pitch actin trimer[8,9]. In Spire, a tandem array of WH2 motifs was shown by electron microscopy to organize a linear actin structure, which was proposed to nucleate through serving as a single, long-pitch strand of an actin filament[7]. However, Spire alone is a relatively weak nucleation factor, whose activity is appreciably enhanced through dimerization mediated by the formin protein, Cappuccino[40]. Similarly, the actin nucleation factor TARP from the pathogen Chlamydia trachomatis requires oligomerization of its WH2 motif by an adjacent poly-proline region for activity[33]. In both Spire and TARP, higher potency likely arises, as in VopL, from the ability of dimers to organize lateral contacts between actin monomers in the nascent filament. These experimental findings are all consistent with computational analyses of actin nucleation, which suggest that a stabilized short-pitch dimer will recruit a third monomer with much greater affinity than will a long-pitch dimer[4]. Thus, short-pitch dimers should act as more effective nuclei.

Together, our findings provide an initial mechanistic model for the potent actin nucleation activity of VopL. Yet many questions remain to be answered. What is the conformation of the VopL-bound actin nucleus, and what contacts to both WH2 motifs and the VCD stabilize it? Does VopL remain bound to filaments after nucleation? If so, does it bind filament ends or sides; does it remain static like the Arp2/3 complex or process like formins? If not, what triggers its release from the nascent filament? Why do VopL and VopF, which presumably nucleate through very similar mechanisms, produce different actin structures in cells? Our work here provides an initial framework to address these questions and others in the future.

Methods

Molecular Biology and Protein Purification

We cloned VopL fragments (W3-C: 115–484, W2-C: 155–484, W1-C: 194–484, VCD: 247–484 and WH2 peptides: WH2a: 129–157, WH2b: 158–187, WH2c: 199–226, W3: 115–246) into pGEX vector by PCR and verified them by DNA sequencing. Proteins were expressed in Escherichia coli BL21(DE3) T1R cells grown in LB media and induced with 1mM IPTG at optical density (OD600) of 0.8. Proteins were purified with successive glutathione-sepharose affinity chromatography, tobacco etch virus (TEV) protease cleavage, followed by strong anion exchange and gel filtration chromatographies. GST fusions of VopL WH2 fragments (W3-GST, W2-GST and W1-GST respectively) were expressed from pET11a and purified using glutathione-sepharose affinity, strong anion exchange and gel filtration chromatographies.

Actin Polymerization Assay

Actin was prepared from rabbit muscle. All actin polymerization assays contained 4μM actin (5% pyrene labeled) and 5 nM–500 nM VopL proteins in KMEI-G Buffer (10 mM Imidazole (pH 7.0), 50 mM KCl, 1 mM MgCl2, and 1 mM EGTA), and were performed at 20 ° C as decribed[41]. Assays were performed using a Jobin Yvon Horiba Fluorimeter (www.jyhoriba.com). The half-time (t1/2) of actin assembly was measured as the time for 50% polymerization.

Protein Crystallization and X-ray Diffraction Data Collection

VCD crystals (50 × 150 × 200 μm) were grown at 20°C using the hanging-drop vapor-diffusion method. Drops contained 1 μl of 10–15 mg ml−1 protein solution (20 mM Tris, pH 7.5, 100 mM NaCl, 1 mM DTT) plus 1 μl of reservoir solution (100 mM Tris pH 7.5, 20% PEG 1500 (w/v)) and were equilibrated against 200μl of reservoir solution. Crystals of selenomethionine-labeled protein were obtained under similar conditions. Crystals were cryoprotected by addition of 30% (v/v) ethylene glycol to reservoir solution. Native crystals were soaked in a cryosolution that also contained 5% PEG20,000 (w/v) for 30 minutes prior to flash cooling. Native and anomalous dispersion data were collected at the Advanced Photon Source 19ID beamline at 100 K. Native VCD crystallized in space group P212121 with unit cell parameters of a = 56 Å, b =88 Å, c = 101 Å, containing two molecules in each asymmetric unit with ~50% solvent. Crystals diffracted X-rays to a minimum Bragg spacing of 2.30 Å. Data indexing and scaling were carried out using the HKL3000 suite of programs[42]. Data collection statistics are in Table 1.

Phase Determination and Structure Refinement

We used multi-wavelength anomalous diffraction data to a resolution of 3.1–3.21 Å from a selenomethionine VopL crystal to obtain experimental phases. The peak, reflection and remote wavelengths were 0.9794 Å, 0.9795 Å and 0.9717 Å, respectively. Ten of ten expected selenium sites were located using the program SHELXD[43]. Phases were refined with the program MLPHARE[44] and further improved to a figure-of-merit of 0.90 by density modification using the program DM[45]. An initial structure model was built using the program Buccaneer[46] and further improved by manual model building using the program Coot[47]. We collected a native data set at the same temperature and beamline, at a wavelength of 0.9792 Å. An electron density map of native VopL was obtained through molecular replacement using the selenomethionine VopL as a search model using the program Phaser[48]. The native structure model was again initially built using Buccaneer and improved with manual model building. Structure refinement with native data was carried out using the program PHENIX[49], consisting of refinements of individual sites, individual B-factors and TLS parameters. The current model contains two VopL chains in one asymmetric unit. Statistics of refinement and structural quality are listed in Table 1. In the refined structure, 97.5%, 2.5% and 0.0% of residues are in the favored, allowed and disallowed regions of the Ramachandran plot, respectively, according to the Molprobity structural validation tools[50].

Multi-Angle Laser Light Scattering (MALLS)

We applied 500 μl of 40 μM VopL proteins at a flow rate of 0.5 ml min−1 to either a Superdex 200 10/300 or a Superdex 75 10/300 column (Amersham Biosciences) equilibrated in buffer containing 20 mM Tris pH 8.0, 100 mM NaCl and 2 mM DTT. The chromatography column was combined in line with a three angle light scattering detector (miniDAWN TREOS) coupled to a refractive index detector (Optilab rEX, Wyatt Technology). Data analysis and molecular weight calculations were carried out using the program ASTRA V (Wyatt technology).

Isothermal Titration Calorimetry (ITC)

All ITC experiments were performed using a VP-ITC MicroCalorimeter at 20°C. Prior to each experiment, WH2 peptides and actin were dialyzed extensively in the same beaker against 2 liters of G-Buffer for at least 24 hours. G-Buffer contains 2 mM Tris, pH 8.0, 0.1 mM CaCl2, 0.5 mM DTT, 1 mM NaN3 and 0.2 mM ATP. In each experiment, 100 μM of WH2 peptide was titrated into 10 μM actin in G-Buffer. Data were analyzed and fit to a single site model using the program Origins7, Microcal for ITC.