Substrate Interaction with the EssC Coupling Protein of the Type VIIb Secretion System

The emergence of antibiotic-resistant bacteria poses a rising problem in antibiotic treatment (S. Boyle-Vavra and R. S. Daum, Lab Invest 87:3–9, 2007, https://doi.org/10.1038/labinvest.3700501). We have used the multidrug-resistant S. aureus USA300_FPR3757 as a model organism to study the T7SSb. Effector proteins of this system have been associated with abscess formation and bacterial persistence in mouse models (M. L. Burts, A. C. DeDent, and D. M. Missiakas, Mol Microbiol 69:736–746, 2008, https://doi.org/10.1111/j.1365-2958.2008.06324.x; M. L. Burts, W. A. Williams, K. DeBord, and D. M. Missiakas, Proc Natl Acad Sci U S A 102:1169–1174, 2005, https://doi.org/10.1073/pnas.0405620102). We determined the structure of the essential ATPase domain D3 of the T7SSb at atomic resolution and validated a surface-exposed pocket as a potential drug target to block secretion. Furthermore, our study provides new mechanistic insights into the T7SSb substrate transport.

ABSTRACT

Staphylococcus aureus employs the type VIIb secretion system (T7SSb) to secrete effector proteins that either have antibacterial activities or promote bacterial persistence in mouse infection models. Here, we present the crystal structure of the ATPase domain D3 of the EssC coupling protein from S. aureus USA300_FPR3757, an integral component of the T7SSb complex, resolved at a 1.7-Å resolution. EssC-D3 shares structural homology with FtsK/SpoIII-like ATPase domains of T7SSa and T7SSb and exhibits a conserved pocket on the surface with differential amino acid composition. In T7SSa, substrate EsxB interacts with the D3 domain through this pocket. Here, we identify amino acids in this pocket that are essential for effector protein secretion in the T7SSb. Our results reveal that the adjacent ATPase domain D2 is a substrate binding site on EssC and that substrates bound to D2 require domain D3 for further transport. Point mutations in the Walker B motif of domain D3 have diametric effects on secretion activity, either abolishing or boosting it, pointing to a critical role of domain D3 in the substrate transport. Finally, we identify ATPase domain D3 as a virulence determinant of S. aureus USA300_FPR3757 using an invertebrate in vivo infection model.

IMPORTANCE The emergence of antibiotic-resistant bacteria poses a rising problem in antibiotic treatment (S. Boyle-Vavra and R. S. Daum, Lab Invest 87:3–9, 2007, https://doi.org/10.1038/labinvest.3700501). We have used the multidrug-resistant S. aureus USA300_FPR3757 as a model organism to study the T7SSb. Effector proteins of this system have been associated with abscess formation and bacterial persistence in mouse models (M. L. Burts, A. C. DeDent, and D. M. Missiakas, Mol Microbiol 69:736–746, 2008, https://doi.org/10.1111/j.1365-2958.2008.06324.x; M. L. Burts, W. A. Williams, K. DeBord, and D. M. Missiakas, Proc Natl Acad Sci U S A 102:1169–1174, 2005, https://doi.org/10.1073/pnas.0405620102). We determined the structure of the essential ATPase domain D3 of the T7SSb at atomic resolution and validated a surface-exposed pocket as a potential drug target to block secretion. Furthermore, our study provides new mechanistic insights into the T7SSb substrate transport.

INTRODUCTION

Staphylococcus aureus is both a commensal bacterium that colonizes approximately 30% of the human population, where it is found primarily in the nares, and also an opportunistic human pathogen that causes a wide range of clinical manifestations that vary in anatomical site and severity. S. aureus is the leading cause of bacteremia, endocarditis, osteomyelitis, and skin and soft tissue, pulmonary, and device-related infections (reviewed in reference 1). While some infections remain superficial, serious diseases are initiated when S. aureus breaches the skin or mucosal barrier to gain access to tissues or the bloodstream. Tissue invasion typically manifests itself in the formation of abscesses, which are accompanied by severe inflammation of the surrounding tissue.

Type VII secretion systems (T7SS) have been found in a wide range of Gram-positive bacteria. While T7SSa are present in actinobacteria where they mediate the secretion of virulence factors in pathogens such as Mycobacterium marinum, Mycobacterium bovis, or Mycobacterium tuberculosis (2), T7SSb are widespread in Firmicutes (3) and have been found in a number of clinical S. aureus isolates (4). Active T7SSb have been described in S. aureus strains USA300_FPR3757 (5), RN6390 (6), COL (6), Newman (7), SA113 (6), and STE398 (8) as well as in Streptococcus intermedius (9), Listeria monocytogenes (10), Bacillus anthracis (11), and Bacillus subtilis (12).

The staphylococcal T7SSb mediates the secretion of four small proteins (10 to 15 kDa), designated EsxA, EsxB, EsxC, and EsxD, with unknown function (5, 7, 13) and a 68-kDa nuclease, designated EssD/EsaD (hereafter named EsaD), with antibacterial activity in bacterial competition assays (14, 15). The secreted Esx substrates form heterodimers (EsxA/EsxC and EsxB/EsxD) as well as homodimers (EsxA, EsxB, and EsxC) (6, 16) while EsaD is a monomer (15). EsxA and EsxB belong to the WXG100 protein family, which has a WXG signature and a typical length of around 100 amino acids.

An unresolved phenomenon of the T7SSb is the codependent secretion of proteins that do not interact with each other (5, 7). For example, in S. aureus USA300, the deletion of one of the four Esx substrates abolishes secretion of the other three Esx substrates (5).

Mutations in the T7SSb locus that abolish protein secretion reduce both the number of abscess lesions and their bacterial load, thereby facilitating the clearance of abscesses in mouse models (7, 8, 13, 17).

The T7SSb is evolutionarily distantly related to the mycobacterial T7SSa through one or more secreted proteins of the WXG100 superfamily and a membrane-anchored coupling protein, which energizes the substrate transport and interacts with the secreted substrates (3). However, both systems require several additional system-specific components for substrate export (see reference 2 for a review of T7SSa).

In S. aureus, the conserved core machinery consists of four membrane proteins (EssA, EssB, EssC, and EsaA) and the cytosolic component EsaB (Fig. 1), which are required for protein export (6, 7, 18, 19). Pulldown experiments indicated that the four membrane components (EssA, EssB, EssC, and EsaA) assemble into a complex in the cell envelope, suggesting that they form a nanomachine that mediates the secretion of effector proteins (20, 21).

FIG 1.

Organization of the T7SSb in S. aureus USA300. (A) Schematic representation of the S. aureus USA300_FPR3757 T7SSb gene cluster. Genes encoding membrane proteins are presented in green, the membrane-anchored ATPase EssC is in blue, small secreted Esx substrates are in black and gray, the cytosolic component is in orange, and the nuclease complex is in red. (B) Schematic representation of T7SSb proteins and their predicted topologies; colors correspond to those in panel A.

The 171-kDa EssC coupling protein is the only ATPase of the membrane protein complex and a predicted multidomain protein. Two X-ray structures from S. aureus MU50 and Geobacillus thermodenitrificans show that the N terminus of EssC is composed of two forkhead-associated (FHA) domains, to which no function has been assigned yet (22, 23). While the FHA domains are unique to coupling proteins of T7SSb, the C-terminal segment shares sequence and structural homology with T7SSa.

Several partial three-dimensional structures have shed light on the overall domain architecture of the conserved C-terminal segment. These structures comprise the stalk domain and the domain of unknown function (DUF) ATPase domain from Mycobacterium smegmatis (T7SSa) (24, 25) and ATPase domains D1 to D3 from Thermomonospora curvata (T7SSa) (26), as well as ATPase domains D2 and D3 from G. thermodenitrificans (T7SSb) (22). Combined, these structures show that two transmembrane helices (TMHs) extend into a stalk domain that is linked to a linear array of four ATPase domains, termed DUF, D1, D2, and D3.

Mutational analysis indicated that the DUF ATPase activity is essential for the ESX-3 T7SSa secretion system of M. smegmatis (24). Sequence comparisons suggest that the DUF is also present in T7SSb of S. aureus (see Fig. S1 in the supplemental material). The ATPase activity of D1 is essential for substrate secretion in T7SSa and T7SSb, whereas the analyses of ATPase domains D2 and D3 revealed differences. For example, ATPase activities of D2 and D3 are required for substrate secretion in ESX-1 and ESX-5 T7SSa of M. tuberculosis (26–28), but they are dispensable for secretion in the yuk T7SSb of B. subtilis (27).

As FtsK/SpoIII ATPase domains are known to form hexameric ring structures, it was suggested that T7SS coupling proteins form a tube-like structure made of four layers of ATPase domains in the cytoplasm (22, 24). In this model, domain D3 forms the outermost layer, and the DUF is located closest to the membrane.

Analyses of S. aureus genomes revealed four alleles of EssC with sequence variations in the 3′-terminal region (4). Experiments in which four EssC alleles were swapped showed that secretion of EsxC but not EsxA depends on the variable C-terminal segment of EssC encompassing part of domain D2 and the entire domain D3, suggesting that EsxC is targeted to this C-terminal region of EssC for secretion while EsxA binds elsewhere (29).

Other studies showed that domain D3 plays a critical role in substrate targeting. In ESX-1 T7SSa of M. tuberculosis and the yuk T7SSb of B. subtilis, the C terminus of substrate EsxB is required for secretion (30, 31). Cocrystal structures of D3 domains from ESX-1 T7SSa from M. tuberculosis and T7SSa from T. curvata with peptides representing the C-terminal signal peptide of the substrates showed that the C-terminal signal peptides interact with a hydrophobic pocket on the surface of domain D3 (26, 32).

In T7SSb, the secretion of the small Esx substrates is also dependent on EsaE/EssE (hereafter named EsaE) in a yet unknown manner (18). EsaE is a cytoplasmic chaperone, which stabilizes nuclease EsaD and is also required for secretion of EsaD. In the cytoplasm, EsaD is bound to an antitoxin EsaG, which inhibits the nuclease activity of EsaD and is not cosecreted with the nuclease (15). EsaE also interacts with EssC, indicating that it targets the EsaG-EsaD complex to the coupling protein EssC (15, 18).

Here, we set out to investigate the interaction of the small Esx substrates, EsxB and EsxC, with coupling protein EssC.

RESULTS

ATPase domain D3 of EssC from S. aureus USA300_FPR3757 was expressed in Escherichia coli BL21 and purified to homogeneity using nickel affinity and size exclusion chromatography (SEC). EssC-D3, which has a molecular weight of 27 kDa, elutes as oligomer and monomer in SEC (see Fig. S2A in the supplemental material). Octahedral crystals of the monomeric form of EssC-D3 were obtained in 1.4 to 1.8 M ammonium citrate tribasic, pH 7.0. The phase problem was solved by single anomalous dispersion of sulfur atoms (Table 1). After refinement, a model of EssC-D3 was built at 1.7-Å resolution, comprising most of the amino acid sequence with exception of a small flexible loop (amino acids 1449 to 1453).

TABLE 1.

Data collection and refinement statistics

Data set	Value(s)a for:
	D3 native (PDB)	D3 (S-SAD)
Wavelength (Å)	1.0332	1.7712
Resolution range (Å)	19.75–1.7	124.53–1.9
Space group	P43212	P43212
Cell dimensions: a, b, c (Å); α = β = γ (°)	91.03, 91.03, 124.51; 90	90.750, 90.750, 124.530; 90
Molecules/asymmetric unit	1	1
Rmeas (%)	4.807 (165.2)	6.9 (207.0)
I/σI	25.71 (1.20)	21.85 (0.76)
Completeness (%)	99.66 (98.72)	99.2 (90.1)
Multiplicity	8.7 (8.8)	12.3 (7.3)
CC1/2	1 (0.542)	1 (0.3)
No. of unique reflections	57,937 (5,635)	81,565 (5,506)
Rwork (%)	15.28 (30.37)
Rfree (%)	17.24 (33.92)
Average B factors (Å2), protein	38.57
Average B factors (Å2), solvent	48.76
No. of atoms (nonhydrogen)	2,104
Solvent	97
Ligand (1× glycerol)	6
RMSD bond lengths (Å)	0.007
RMSD bond angles (°)	0.77
Ramachandran favored (%)	96.88
Ramachandran allowed (%)	3.12
Ramachandran outliers (%)	0
Clashscore	6.22

^aNumbers in parentheses are for the highest resolution shell.

The EssC-D3 crystal structure exhibits a typical Rossmann fold of nucleotide hydrolysis domains and comprises a central β-sheet containing 9 β-strands surrounded by 10 α-helices (Fig. 2A and B).

FIG 2.

Structural comparison of EssC-D3 from S. aureus USA300 (T7SSb) with EccC-D3 from T. curvata and M. tuberculosis (T7SSa). (A) The crystal structure of EssC-D3 from S. aureus USA300_FPR3757 shows similarity to FtsK/SpoIII ATPases. (B) View to the ATP binding site of EssC-D3. The Walker B motif of S. aureus USA300 (yellow) and corresponding amino acids in the D3 domain structure from T. curvata (blue). Amino acid K1122 is part of the Walker A motif of domain D3 from T. curvata. A Walker A sequence is not found in EssC-D3 of S. aureus USA300 but a potentially functional equivalent R1302. (C, D) Superimposition of domain D3 from S. aureus USA300_FPR3757 (yellow) with D3 domains from T. curvata (blue) and from M. tuberculosis Esx-1 (magenta), which show also bound signal peptides (green sticks) and ATP molecules (green spheres). The bound ATP molecules clash with α-helix 3, indicating that α-helix 3 needs to undergo conformational changes for accommodation of ATP. (E, F) Amino acid composition inside the substrate binding pockets of domain D3 from S. aureus USA300_FPR3757 (yellow) from T. curvata (blue) and from M. tuberculosis (magenta). The D3 pockets differ in their amino acid composition, but one leucine (L1356 in S. aureus USA300, L1208 in T. curvata, L442 in M. tuberculosis, and L1341 in G. thermodenitrificans) is conserved and essential for substrate binding in the T7SSa of T. curvata. This work shows that the conserved leucine is also essential for secretion in S. aureus USA300_FPR3757.

A similarity search using Dali (33) found two D3 ATPase domain structures of T7SSa coupling proteins from T. curvata (EccC-D3; PDB code 5FV0-A; Z score of 22.4; root mean square deviation [RMSD] of 2.7 Å for 218 C-α positions) (26) and from M. tuberculosis (EccCb-D3; PDB code 6J19-A; Z score of 18.8; RMSD of 3.2 Å for 216 C-α positions) (22), which both show bound signal peptides, and one D3 domain structure of a T7SSb coupling protein from G. thermodenitrificans (EssC-D3; PDB code 4LYA-A; Z score of 22.8; RMSD of 2.6 Å for 219 C-α positions). The two signal peptides are bound to a pocket on the surface of the respective D3 domains, which is also present in D3 domains of the T7SSb coupling proteins from S. aureus USA300_FPR3757 and G. thermodenitrificans (Fig. 2C to F). The D3 pocket has a differential amino acid composition in all four D3 domains. In the T7SSa of T. curvata and of M. tuberculosis, the D3 pocket was identified as the binding site for the C-terminal T7SSa signal sequence (26, 32). Therefore, we hypothesized that the amino acids inside the pocket also mediate substrate specificity in T7SSb, encoded by the cognate T7SSb signal sequence.

ATPase domain D2 mediates substrate binding.

To test whether ATPase domain D3 is crucial for substrate binding, we coexpressed either EssC-ΔD3, which lacks D3, or full-length EssC with substrates EsxB and EsxC in E. coli, respectively.

Subsequent analysis of the purified complexes by blue native polyacrylamide gel electrophoresis (BN-PAGE) and Western blotting showed the comigration of the two substrates with an ∼1-MDa protein species corresponding to hexamers of EssC (171 kDa) and EssC-ΔD3 (144 kDa) (Fig. 3).

FIG 3.

Substrates EsxB and EsxC interact with EssC as well as EssC-ΔD3. Complexes EsxB-EssC and EsxC-EssC were purified by affinity and size exclusion chromatography, respectively (Superose 6 increase; left). SEC fractions were subsequently resolved by BN-PAGE and analyzed by Western blotting using specific antibodies against EsxB, EsxC, and the Strep-Tactin tag II of EssC and EssC-ΔD3, respectively (middle and right, respectively). (A) EssC plus EsxB; (B) EssC-ΔD3 plus EsxB; (C) EssC plus EsxC; (D) EssC-ΔD3 plus EsxC.

To further evaluate the contribution of domain D3 to substrate binding, we labeled purified EssC and EssC-ΔD3 with fluorophore NT-647 (1-MDa oligomer, respectively) and measured binding to the EsxB homodimer (16) using fluorescence quenching (Fig. 4A and B; see also Fig. S2B and C and Fig. S3 in the supplemental material). We obtained similar affinities of EsxB for either full-length EssC (K_d (dissociation constant) = 8.5 ± 1.3 μM) or EssC-ΔD3 (K_d = 1.6 μM ± 0.1 μM). This result raised the possibility either that D3 is dispensable for substrate binding or that the removal of D3 exposed a second binding site on EssC with equivalent affinity for EsxB.

FIG 4.

Comparison of affinity measurements between EsxB and EssC variants reveals domain D2 as a substrate binding site. Purified EssC, EssC-ΔD3, fragment D2D3, and D3 were labeled with fluorophore NT-647, and purified EsxB was titrated to the respective variant until no further increase in fluorescence was observed. All experiments were carried out as technical triplicates. Error bars represent the standard deviation from the mean fluorescence. (A) EssC plus EsxB; (B) EssC-ΔD3 plus EsxB; (C) D2D3 plus EsxB; (D) D3 plu EsxB.

Using fluorescence quenching, we measured a similar affinity of EsxB for a fragment consisting of domains D2D3 (K_d = 1.3 ± 0.4 μM), which elutes as a hexamer in SEC (∼360-kDa oligomer peak), but we did not detect binding of EsxB to monomeric D3 (Fig. 4C and D; see also Fig. S2A and D). These results indicate that EsxB interacts with domain D2.

Substrate transport from ATPase domain D2 requires ATPase domain D3.

To investigate whether D3 is required for secretion, we established an assay to monitor the T7SSb-dependent secretion of substrate EsxC into the culture medium (5). We first constructed an in-frame deletion of the essC gene of S. aureus USA300_FPR3757 and then transformed the deletion strain with plasmid pLac expressing either the EssC wild type or variants under the control of the T7 promoter (see Fig. S4 in the supplemental material) and then tested for the presence of EsxC into the culture medium using polyclonal antiserum against EsxC (Fig. 5). Complementation of the S. aureus USA300ΔessC with plasmid pLac carrying essC restored secretion of EsxC while only basal levels of EsxC were detected in the culture medium of the essC deletion strain. The deletion of domain D3 abolished EsxC secretion, indicating that substrate transport from domain D2 requires domain D3. This result is in line with a previous report that showed that the conserved C-terminal D3 domain is essential for secretion in T7SSb from S. aureus RN6390 (22).

FIG 5.

Structure and function of EssC-D3. Culture supernatants from wild-type S. aureus USA300 and complemented S. aureus USA300ΔessC were analyzed for substrate secretion using specific antibodies against EsxC and for cell lysis using specific antibody against RNA polymerase subunit (150 kDa). Cell pellets were analyzed for EsxC expression and RNA polymerase β subunit, which was used as loading control.

The Walker B motif of domain D3 has a significant effect on secretion activity.

Both purified EssC-D3 and its homolog from G. thermodenitrificans appear incapable of binding ATP due to an extension of α-helix 3 that occludes a potential ATP binding site. Superimpositions show that α-helix 3 in both D3 domains clashes with the bound ATP in the D3 structures of T. curvata and M. tuberculosis (Fig. 2C and D). While the helix extension of EssC-D3 from G. thermodenitrificans carries a Walker A motif, the extension of EssC-D3 does not contain a Walker A motif but a possible functionally equivalent arginine in position 1302, suggesting that a conformational change to a loop could restore the ATP binding site. In agreement with our observation, fluorescence quenching experiments showed that the isolated monomeric D3 domain binds ATP with low micromolar affinity (K_d = 397 ± 175 μM), while no interaction with the full-length EssC hexamer was detected (Fig. 6).

FIG 6.

Monomeric EssC-D3 binds ATP with low millimolar affinity. ATP was titrated to either hexameric EssC or monomeric EssC-D3, both labeled with fluorophore NT-647 until no increase in fluorescence was observed. (A) EssC plus ATP. (B) D3 plus ATP.

The structural comparisons with EssC-D3 from T. curvata, M. tuberculosis, and G. thermodenitrificans revealed a degenerated Walker B motif (hhhhND; h indicates hydrophobic) in EssC-D3 composed of amino acids N1379 and D1380 (Fig. 2B). To test whether the catalytic activity of domain D3 is a requirement for secretion, we mutated both amino acids of the Walker B motif to alanine and monitored EsxC secretion into the culture medium. The mutation D1380A abrogated EsxC secretion while the mutation N1379A unexpectedly increased secretion significantly compared to that of the deletion strain complemented with wild-type essC. Our analysis indicates that domain D3 may have a significant impact on T7SSb secretion activity. Furthermore, we conclude that α-helix 3 of EssC-D3 needs to undergo conformational changes that permit ATP binding and hydrolysis in vivo.

The surface-exposed pocket on ATPase domain D3 is crucial for secretion.

Since we did not detect substrate binding to the isolated D3 domain, we investigated whether the conserved pocket on D3 is involved in secretion. Therefore, we mutated six solvent exposed amino acids in the cleft of D3 and measured their impact on the secretion activity. In the center of the D3 pocket, amino acids L1356 and F1370 compose a small hydrophobic patch, which is surrounded by polar amino acids C1325, N1342, and Q1363 as well as Y1376, which is exposed only with its hydroxyl group. Structural comparisons further revealed that the amino acids N1342, L1356, and S1374 of EssC-D3 superimpose with three hydrophobic amino acids in the pockets of D3 from M. tuberculosis (A427, L442, and V471) and T. curvata (I1163, L1208, and I1179) (Fig. 2C to F). In T. curvata, these residues were shown to be essential for binding substrate EsxB (Fig. 2E). Further structural comparison showed that leucine (L1356 in S. aureus, L1208 in T. curvata, L1341 in G. thermodenitrificans [not shown], L442 in M. tuberculosis) is the only invariant amino acid in the binding pockets of the D3 domain structures from S. aureus USA300, G. thermodenitrificans, T. curvata, and M. tuberculosis.

The mutations S1374A and L1356A caused a strong reduction of EsxC secretion. An L1356A/S1374A double mutant did not cause any further drop in secretion. Mutations C1325A decreased secretion of EsxC moderately, while Y1376A caused a slight increase in secretion. The mutations N1342A, Q1363A, F1370A, and Y1376F did not affect EsxC secretion significantly. Taken together, these results demonstrate that the conserved cleft on domain D3 plays an essential role in secretion.

Domain D3 is a virulence determinant of the type VIIb secretion system.

Finally, we used an invertebrate in vivo infection model to investigate whether effector protein secretion correlates with virulence. To do this, we used infection of Galleria mellonella larvae. This infection model produces only an innate immune response consisting of neutrophils, macrophages, and host-defense peptides (34, 35). Thus, the Galleria in vivo infection model represents a simplified model to study the bacterial protection against the innate immune response in the absence of the adaptive branch. Our Galleria assays used cohorts of 25 larvae challenged with bacterial suspension (5 × 10⁶ CFU in 20 μl); the progression of infection was monitored for 48 h at 37°C, and dead larvae were counted.

First, we monitored the survival of G. mellonella after infection with the S. aureus USA300ΔessC (Fig. 7). Higher survival rates of G. mellonella were observed after infection with S. aureus USA300ΔessC, which is defective in EsxC secretion compared to that in the wild-type strain. In agreement, complementation of the deletion strain with pLac:EssC-ΔD3 lacking domain D3, which is also defective in secretion, showed similarly higher survival rates of G. mellonella. In contrast, complementation with plasmid pLac:EssC encoding full-length EssC caused a drop in the survival rate to wild-type levels. Overall, defects in secretion attenuated but did not abolish virulence of the complemented S. aureus USA300ΔessC strains. Our results indicate a correlation between virulence and secretion of effector proteins of the T7SSb, and we identify EssC-D3 as a virulence determinant of the T7SSb.

FIG 7.

Domain D3 is a virulence determinant of S. aureus USA300_FPR3757. Galleria mellonella survival was monitored over 48 h upon infection with S. aureus USA300_FPR3757. Survival rates of G. mellonella larvae improved significantly when essC was deleted from the genome (S. aureus USA300ΔessC) or when S. aureus USA300ΔessC was complemented with pLac carrying essC-Δd3. No difference in survival rates was observed upon infections with either wild-type S. aureus USA300 or S. aureus USA300ΔessC complemented with pLac carrying wild-type essC.

DISCUSSION

Here, we analyzed substrate interaction with coupling protein EssC of the T7SSb from S. aureus USA300_FPR3757. Structural data of the coupling protein from T. curvata show that the ATPase domains D1, D2, and D3 exhibit similar surface-exposed pockets, but the pockets of ATPase domains D1 and D2 are occupied with the linkers of their preceding ATPase domains.

Located at the distal end, ATPase domain D3 has no preceding ATPase domain, and thus, its pocket is accessible and can interact with the C terminus of substrate EsxB (26). Cocrystal structures of D3 domains with peptides representing the C-terminal cognate signal sequences show a rather small number of interactions between pocket and peptide (26, 32). Consistent with this, the relatively low affinities of ∼10 μM and ∼30 μM between these D3 domains and their cognate signal peptides might reflect the transient nature of this interaction (26, 32).

Although we could not detect binding between EsxB and the monomeric D3 domain of S. aureus USA300_ FPR3757, we identified amino acids in the D3 pocket that are crucial for secretion, indicating substrate interaction in vivo analogous to the T7SSa.

Our binding studies indicate that domain D2 is a binding site for substrates EsxB and EsxC. Based on the coupling protein structures from T. curvata and G. thermodenitrificans (22, 26), it seems plausible that removal of D3 has exposed the putative binding pocket on ATPase domain D2, which may otherwise become accessible only through conformational changes during transport.

Our results are in line with a study that suggested that a segment comprising domains D2 or D3 targets EsxB, EsxC, and EsxD, whereas EsxA binds elsewhere (29). Secretion of the small Esx substrates also requires chaperone EsaE by a yet unknown mechanism (18). EsaE targets the toxin-antitoxin complex EsaD-EsaG to the multimeric form of EssC (15). This interaction could further stabilize the hexameric ATPase ring structure of EssC in the cytoplasm and, thus, indirectly promote secretion of the small Esx substrates. Consistent with this model, a pulldown experiment using EsaE as bait identified EsaG, EsaD, EssC and in addition substrate EsxC (18). As our comigration studies indicated that EsxC interacts directly with EssC, it seems possible that EssC can be loaded with multiple substrate implying that EsxC and EsaE (bound to EsaD-EsaG) target different binding sites on EssC.

We further observed that substrate transport from domain D2 requires the presence of D3. Domain D3 forms oligomers, and thus, it might be required for stabilizing a hexameric ring structure of EssC in the cytoplasm.

Our data also highlight a critical role of the D3 ATPase domain in the secretion activity of T7SSb. Mutations in the Walker B motif have diametric effects on secretion. We speculate that these mutations might mimic conformational changes that are triggered upon substrate engagement. Indeed, inspection of the ATP binding site of D3 along with ATP binding assays confirmed the inaccessibility of the ATP binding in the absence of a bound substrate and indicate the necessity to undergo conformational changes in order to bind and hydrolyze ATP in vivo.

Overall, the discovery of a second substrate binding site points to a secretion model in which the small effector proteins EsxB, EsxC, and EsxD are transferred subsequently along the structurally similar ATPase domains D3, D2, D1, and DUF toward the membrane pore. It remains elusive where EsxA and the much bigger EsaE-EsaD-EsaG complex are targeted and how their transport is connected to EsxB, EsxC, and EsxD.

MATERIALS AND METHODS

Cloning.

Cloning was carried out using the Phusion polymerase (Invitrogen) and In-Fusion cloning (Clontech). Vectors were linearized by PCR or restriction enzymes. All primers are listed in Table S1 in the supplemental material.

For protein expression, the essC gene was codon optimized for E. coli. The fragments essC-d3 and essC-d2d3 were cloned into the pET-16b vector (Novagen) using primers X1 to X4 to generate EssC-D3/EssC-D2D3 with a tobacco etch virus (TEV) cleavage site and a His₁₀ tag at the N terminus. The gene esxB of S. aureus USA300 was cloned into pET-16b using the primers X1/X2 to linearize the vector and X17/X18 to integrate esxB including the TEV cleavage.

The essC gene (codon optimized for E. coli) was cloned into the pASG-IBA103 (X5 to X8) and then transferred with the Twin-Strep II into pASK-IBA3C (X9/X10, RE HindIII, and XbaI). A TEV cleavage site was inserted into essC-Twin-Strep II-pASK-IBA3C (X11/X12, RE XbaI, and AfeI) (IBA Lifesciences). To produce untagged EsxB and EsxC, stop codons were introduced at the 5′ end of the genes of plasmids esxB-pET-20b and esxC-pET-20b, respectively, using site-directed mutagenesis (X19/X20 or X23/X24).

The markerless gene deletion of essC was generated using the primers BN156 to BN159 and integrated into the pMAD vector (36). The gene essC from S. aureus USA300_FPR3757 and the truncated variant essC-Δd3 were cloned into the pLac vector using primers X29 to X31 (37). The point mutation C1235A of essC-USA300 was generated by splitting essC into two fragments, which were amplified using the primers X32/X33 and X34/X35 containing the point mutation. The amplified fragments were integrated into the pLac vector, which was linearized by PCR using the primers X27/X28. All point mutations in EssC were generated as described for C1235A using the primers listed in Table S1 (X32 to X75).

Expression and purification of EssC-D3.

E. coli BL21 Star (Thermo Fisher) was transformed with the plasmid pET-16b carrying essC-d3 (nucleotides [nt] 3742 to 4440). Bacteria were grown in LB medium containing 100 μg/ml ampicillin at 37°C. Overexpression of EssC-D3 was induced by addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to a final concentration of 1 mM at an optical density at 600 nm (OD₆₀₀) of 0.6 for 3 h at 37°C.

Purification of EssC-D3.

Bacteria were harvested by centrifugation at 4,000 × g for 15 min, resuspended in 50 mM Tris HCl (pH 8.0), 300 mM NaCl, 10 mM MgCl₂, and 3 mM dithiothreitol (DTT), and broken by three passages through an EmulsiFlex-C3 homogenizer.

After ultracentrifugation (100,000 × g, 1 h, 4°C) to remove debris, the supernatant was supplemented with 20 mM imidazole and loaded onto a 5-ml Ni-nitrilotriacetic acid (Ni-NTA) column (GE Healthcare) equilibrated with 50 mM Tris HCl (pH 8.0), 300 mM NaCl, 10 mM MgCl₂, and 3 mM DTT. EssC-D3 was eluted in the same buffer containing a step gradient of 250 mM imidazole.

Afterwards the His₁₀ tag was cleaved using TEV protease (1:10). The sample was dialyzed in 50 mM Tris HCl (pH 8.0), 300 mM NaCl, 10 mM MgCl₂, and 3 mM DTT and applied to a 5-ml Ni-NTA column. The flowthrough was collected and concentrated to 2 ml using a 10-kDa centrifugal concentrator (Millipore).

The sample was gel filtrated on a Superdex 75 column (GE Healthcare) using 20 mM Tris HCl (pH 8.0), 150 mM NaCl, 10 mM MgCl₂, and 3 mM DTT.

Purification of EssC, EssC-ΔD3, EssC-D2D3, EsxB, and EsxC.

E. coli BL21 Star (Thermo Fisher) was transformed with plasmid pASK-IBA3C (IBA Lifesciences) carrying either essC or essC-Δd3 (nt 1 to 3742) fused to a Twin-Strep-Tactin II tag at the 5′ end under the control of the tet promoter and expressed/purified as described in “Comigration experiments” below.

E. coli BL21 Star (Thermo Fisher) was transformed with the plasmid pET-16b carrying essC-d2d3 (nt 2962 to 4440). Bacteria were grown in LB medium containing 100 μg/ml ampicillin at 37°C. Overexpression of EssC-D2D3 was induced by addition of IPTG to a final concentration of 1 mM at an OD₆₀₀ of 0.6 for 2 h at 37°C, and bacteria were purified as described in the section Purification of EssC-D3. The sample was gel filtrated on a Superose 6 increase column (GE Healthcare) using 20 mM Tris HCl (pH 8.0), 150 mM NaCl, 10 mM MgCl₂, and 3 mM DTT.

E. coli BL21 Star was transformed with esxB-pET-16b encoding a TEV cleavage site and His₁₀ tag at the N terminus under the control of the T7 promoter. Bacteria were grown in an LB medium containing 100 μg/ml ampicillin at 37°C. Protein overexpression was induced by addition of IPTG to a final concentration of 1 mM at an OD₆₀₀ of 0.6 for 4 h at 26°C. Bacteria were harvested; resuspended in 50 mM Tris HCl (pH 8.0), 300 mM NaCl, and 3 mM DTT; and disrupted using an Emulsiflex-C3 homogenizer. Bacterial membranes were separated by ultracentrifugation (100,000 × g, 1 h, 4°C), and the supernatant was supplemented with 20 mM imidazole and loaded onto a 5 ml Ni-NTA column (GE Healthcare) equilibrated with 50 mM Tris HCl (pH 8.0), 300 mM NaCl, and 3 mM DTT. EsxB was eluted in the buffer used for equilibration supplemented with 250 mM imidazole. Afterward, the His₁₀ tag was cleaved using TEV protease (1:10). The sample was dialyzed in 50 mM Tris HCl (pH 8.0), 300 mM NaCl, and 3 mM DTT and applied to a 5 ml Ni-NTA column. The flowthrough was collected and concentrated to 2 ml using a 10-kDa centrifugal concentrator (Millipore). The sample was gel filtrated on a Superdex 75 column (GE Healthcare) using 20 mM Tris HCl (pH 8.0), 150 mM NaCl, and 3 mM DTT. E. coli BL21 Star was transformed with esxC-pET-20b encoding a His₆ tag at the C terminus under the control of the T7 promoter. Bacteria were grown in LB medium containing 100 μg/ml ampicillin at 37°C. Protein overexpression was induced by addition of IPTG to a final concentration of 1 mM at an OD₆₀₀ of 0.6 for 14 h at 26°C. Protein purification was performed as described for EsxB.

Crystallization of EssC-D3.

EssC-D3 was concentrated to 10 mg/ml for crystallization using centrifugal concentrators (Millipore). Octahedral crystals of EssC-D3 were grown in hanging drops at 18°C using the vapor diffusion method. A volume of 1.5 μl of EssC-D3 (10 mg/ml) was mixed with an equal volume of 1.4 to 1.8 M ammonium citrate and equilibrated against 600 μl reservoir solution. Crystals were transferred into mother liquor supplemented with 25% (vol/vol) glycerol as cryoprotectant and then flash frozen in liquid nitrogen.

Structure determination of EssC-D3.

Native data were collected at 100 K at beamline 14.1 at the Helmholtz Zentrum (Berlin, Germany) and were indexed, integrated, and scaled to a 1.7-Å resolution using the XDS software package (38). Data collection and model refinement statistics are reported in Table 1.

Single-wavelength anomalous dispersion data of sulfur atoms (S-SAD) were collected at beamline id23eh1 of the European Synchrotron Radiation Facility in Grenoble (France), indexed, integrated, and scaled to a 1.9-Å resolution using the XDS software package (38), and the phase problem was solved using SHELX (39). The initial model for EssC-D3 was built by Autobuild (40) and refined by alternating rounds of model building with Coot (41) and refinement cycles with Phenix (42, 43). The refinement protocol included initial rigid body refinement, cartesian, and individual B factor refinement.

Comigration experiments.

E. coli BL21 Star (Thermo Fisher) was cotransformed with the plasmid pASK-IBA3C (IBA Lifesciences) carrying either essC or essC-Δd3 (nt 1 to 3742) fused to a Twin-Strep-Tactin II tag at the 5′ end under the control of the tet promoter and with plasmid pET-20b encoding untagged substrates EsxB or EsxC, respectively. Bacteria were grown in LB medium containing 100 μg/ml ampicillin and 25 μg/ml chloramphenicol at 37°C. Protein overexpression was induced by addition of IPTG to a final concentration of 1 mM and 2 μg/ml anhydrotetracycline at an OD₆₀₀ of 0.6 for 20 h at 18°C. Bacteria were harvested; resuspended in 50 mM Tris HCl (pH 8.0), 300 mM NaCl, 10 mM MgCl₂, and 3 mM DTT; and disrupted using an Emulsiflex-C3 homogenizer. Bacterial membranes were isolated by ultracentrifugation (100,000 × g, 1 h, 4°C) and then solubilized in 50 ml of 50 mM Tris HCl (pH 8.0), 300 mM NaCl, 10 mM MgCl₂, 3 mM DTT, and 0.5% n-dodecyl-β-d-maltoside (DDM) (1 h, stirring at 4°C). After removal of insoluble material by ultracentrifugation (100,000 × g, 1 h, 4°C), the supernatant was loaded on a Strep-Tactin column (1 ml; IBA Lifesciences) equilibrated with 50 mM Tris HCl (pH 8.0), 300 mM NaCl, 10 mM MgCl₂, 3 mM DTT, and 0.05% DDM. The column was washed with 30 ml of equilibration buffer ensuring the UV absorption went back to the baseline following elution of EssC-Strep in the same buffer supplemented with 2.5 mM d-desthiobiotin.

The peak fraction was concentrated using a 100-kDa centrifugal concentrator (Millipore), and 500 μg of the sample was applied to a Superose 6 increase column (10/300 GL, GE Healthcare). Centrifugal concentrator and SEC column were equilibrated with 20 mM Tris HCl (pH 8.0), 150 mM NaCl, 10 mM MgCl₂, 3 mM DTT, and 0.05% DDM. Ten microliters of fractions was used for BN-PAGE analysis.

Generation of the S. aureus USA300ΔessC deletion strain.

The markerless gene deletion for essC was generated using the temperature-sensitive pMAD vector (36). This plasmid replicates in S. aureus only when growing at less than 30°C. The plasmid was transferred into S. aureus RN4420 using electroporation. After recovery for 1 h at 30°C, the bacteria were spread on tryptic soy broth (TSB) plates containing 2 μg/ml erythromycin and 100 mM 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) for selection and incubated at 30°C for 48 h. The blue colonies were grown in TSB medium containing 2 μg/ml erythromycin at 30°C for 6 h and then used for phage production. The bacterial suspension was mixed with 5 mM CaCl₂ and incubated at 57°C for 90 s. A volume of 300 μl bacteria was incubated with 100 μl of ϕ11 phage lysate in different dilutions (10⁻³, 10⁻⁴, and 10⁻⁵) for 15 min at room temperature (RT), mixed with 3 ml TSB-soft agar supplemented with 20 mM MgSO₄, and spread on LB plates. The plates were incubated for 16 to 18 h at 30°C, and phage hollows were harvested by scraping of the soft agar, which was taken up in 3 ml TSB medium. Agar pieces were spun down (4,000 × g for 10 min) and the supernatant filtered twice (0.2 μm). A culture of S. aureus USA300_FPR3757 was grown in TSB medium for 6 h at 37°C, mixed with 5 mM NaCl, and incubated at 57°C for 90 s, and a 300-μl bacterial suspension was incubated with 100 μl phage lysate containing the plasmid for 15 min at RT. The bacterial suspension was spread on TSB plates containing 100 mM X-Gal, 150 μg/ml erythromycin, and 20 mM MgSO₄ and incubated for 2 to 3 days at 30°C. The blue colonies were used for the first recombination.

The first and second recombinations in S. aureus USA300_FPR3757 were performed as described in reference 37. Genes essC, essC-ΔD3, and essC containing the respective point mutations were inserted into S. aureus USA300ΔessC using the temperature-sensitive pLAC vector. Transformation of S. aureus RN4420 with the respective plasmids, phage production, and transduction were carried out as described above.

Protein secretion assay.

Staphylococci were grown in TSB medium at 30°C overnight. The precultures were washed in 1× PBS, resuspended in 1 ml TSB medium, and used to inoculate 30 ml TSB medium. The OD₆₀₀ of the cultures was 0.05 at the start, and the cultures were grown until an OD₆₀₀ of 1 at 28°C under shaking (160 rpm). The cultures were centrifuged at 2,770 × g for 10 min at RT, 30 ml supernatant was passed through a 0.2-μm filter, and 3 μg of LprG-His₆ was added as precipitation control. Proteins in the supernatant were precipitated by addition of 5% trichloroacetic acid (overnight, 4°C), centrifuged (6,000 × g, 4°C, 15 min), and washed with 100% ice-cold acetone. The precipitate was air-dried; resuspended in 125 μl 1× Laemmli buffer, 10 mM Tris HCl, pH 8.0; and incubated at RT overnight.

The bacterial pellet was resuspended in 1 ml of 20 mM Tris HCl (pH 7.5), 10 mM EDTA, 50 μg/ml lysostaphin, and 1 mM phenylmethylsulfonyl fluoride (PMSF). The samples were incubated for 10 min at 37°C and mechanically lysed by addition of 250-mg glass beads per sample using a FastPrep shaker. The lysis step was performed twice for 40 s with a speed of 6.5 m/s. In between the steps, the samples were rested on ice. The lysates were centrifuged for 10 min by 4,000 × g at 4°C, and the supernatants were transferred into fresh tubes and centrifuged for 35 min by 100,000 × g at 4°C. The pellets were resuspended in 150 mM NaCl and 50 mM Tris HCl, pH 8.0, and incubated for 10 min at 50°C. The supernatants of the cytosolic fractions as well as the pellets were mixed with 4× LDS sample buffer. All samples (supernatant, cytosolic fraction, and membrane fraction) were boiled for 10 min at 100°C before separation by SDS-PAGE and subsequent Western blotting analysis. Polyclonal antibodies against EssC and EsxC were kindly provided by Tracy Palmer (Centre for Bacterial Cell Biology, Newcastle University, England).

Blue native polyacrylamide gel electrophoresis.

Proteins were separated on Novex 3 to 12% blue native gradient gels according to the manufacturer’s protocol (Invitrogen).

Western blotting.

Western blotting was carried out following the manufacturer’s protocol (Invitrogen). Polyvinylidene difluoride (PVDF) membranes were incubated either with directly conjugated horseradish peroxidase (HRP) antibodies for 1 h or with primary antibodies overnight at 4°C (see Table S2 in the supplemental material). The membranes were incubated for 2 h with secondary antibodies at RT (Table S2) and exposed using the enhanced chemiluminescence substrate kit 12 (Pierce) and the ImageQuant LAS 4000 biomolecular imager.

Protein binding assay.

Purified EssC, EssC-ΔD3, and EssC-D2D3, corresponding to the hexamers and the EssC-D3 monomer, were N-hydroxysuccinimide (NHS) labeled with fluorophore NT following the manufacturer’s protocol (Monolith protein labeling kit RED-NHS 2nd generation MO-L011; NanoTemper), and concentrations of the NHS-labeled proteins were determined. The binding assay was carried out according to the manufacturer’s protocol using the Monolith NT.115. Because of quenching effects, the fluorescence intensity was used for data analysis instead of thermophoresis. EsxB was titrated at the following concentrations: 0.65 mM to 19.8 nM to 0.5 nM EssC, 500 μM to 16.8 nM to 20 nM EssC-ΔD3, 1 mM to 33.6 nM to 10 nm D2D3, 0.1 mM to 3.1 pM to 1 nM D3. ATP was titrated in a concentration range of 100 mM to 3.1 nM to a concentration of 6 nM either EssC-D3 or EssC.

G. mellonella survival assay.

Wax moth larvae (G. mellonella) were used as an in vivo model to assess the pathogenicity of different S. aureus USA300_FPR3757 mutants. Infection assays were performed as previously described (34). Briefly, bacterial overnight cultures were diluted 1:100 in 10 ml brain heart infusion (BHI) medium and grown for 3 h at 37°C. Cultures were adjusted to an OD₆₀₀ of 0.6 (∼1 × 10⁷ CFU/ml) in 1 ml BHI, and the cells were harvested by centrifugation at 4,000 × g for 10 min at 4°C. The cell pellets were washed three times with 10 mM MgSO₄ and finally suspended in 1 ml 10 mM MgSO₄.

Larvae were inoculated with 20 μl of bacterial suspension into the last proleg using an insulin pen (BD Micro-Fine, U100, 0.3 ml; Becton, Dickinson). Larvae injected with 10 mM MgSO₄ were used as a control to exclude negative effects of the physical trauma on larvae survival. Three independent experiments were performed using cohorts of 25 larvae per strain and experiment. After injection, larvae were kept in petri dishes at 37°C. Larval survival was assessed after 16 h, 20 h, 24 h, and 48 h of inoculation, with larvae being scored as dead if they did not respond to touch or had turned black.

Data availability.

The model of the EssC-D3 domain of S. aureus has been deposited in the Protein Data Bank under PDB code 6TV1.