Type VI secretion system sheath inter‐subunit interactions modulate its contraction

Abstract Secretion systems are essential for bacteria to survive and manipulate their environment. The bacterial type VI secretion system (T6SS) generates the force needed for protein translocation by the contraction of a long polymer called sheath. The sheath is a six‐start helical assembly of interconnected VipA/VipB subunits. The mechanism of T6SS sheath contraction is unknown. Here, we show that elongating the N‐terminal VipA linker or eliminating charge of a specific VipB residue abolishes sheath contraction and delivery of effectors into target cells. Mass spectrometry analysis identified the inner tube protein Hcp, spike protein VgrG, and other components of the T6SS baseplate significantly enriched in samples of the stable non‐contractile sheaths. The ability to lock the T6SS in the pre‐firing state opens new possibilities for understanding its mode of action.


Introduction
Bacteria have evolved various protein nanomachines to translocate macromolecules across biological membranes. A subset of these nanomachines is composed of a rigid tube surrounded by a contractile sheath, which is attached to a baseplate. The sheath is initially assembled in a high-energy, extended state and then quickly transits to a low-energy, contracted state. The sheath contraction pushes the inner tube, through a membrane complex into the target cell, where it delivers its cargo. The bacterial type VI secretion system (T6SS) uses this mechanism to deliver proteins across membranes [1][2][3][4][5].
Biogenesis of the T6SS starts by formation of an integral membrane complex of TssJ, TssL, and TssM on which the baseplate assembles [6,7]. TssE, TssF, TssG, and TssK constitute the baseplate with a trimer of the spike protein VgrG in the center [6,8]. Baseplate assembly is required to initiate polymerization of the hexameric tube protein Hcp. The tube serves as a template for rapid polymerization of the extended sheath composed of heterodimeric protomers of VipA (TssB) and VipB (TssC) [9][10][11][12][13]. In addition, two different classes of TssA proteins are important for the initiation of T6SS sheath assembly and its elongation [14,15], which progresses throughout the whole cell and thus allows the use of live-cell fluorescence microscopy to monitor sheath dynamics [1,7,[16][17][18][19]. Contraction of the T6SS sheath pushes the spike and the inner tube out of the cell into a neighboring cell, which leads to the delivery of effectors that are associated with the spike or the inner tube [20][21][22]. Among substrates of the T6SS are anti-eukaryotic as well as anti-bacterial effectors that are used for host interaction and interbacterial competition [23][24][25].
The sheath can be described as a six-start helix and thus with six strands or as a stack of rings composed of six VipA/VipB heterodimeric subunits, which are interconnected by N-and C-terminal linkers in the inner domain of the sheath [26,27]. Inner domains of sheaths of contractile phage, R-type pyocin, and T6SS likely maintain the connectivity of the sheath during contraction and are evolutionarily related to each other; however, the surface-exposed domains are distinct [12,[26][27][28][29][30][31][32]. Specifically, T6SS sheath contains a surface-exposed Domain 3 that plays a crucial role in sheath recycling [26,27,31,32]. The mechanism of T6SS sheath contraction is largely unknown because the sheath contracts during isolation from cells, and thus, characterization of the extended sheath was not possible until recently [1,32].
Here, we identified two structural features of the T6SS sheath that play a critical role in its contraction. Using live-cell fluorescence microscopy, we show that a single negatively charged residue located on the surface of the middle domain of the T6SS sheath is required for sheath contraction but not sheath assembly. We further show that the VipA N-terminal linker is critical for sheath contraction. Insertion of two and more amino acid residues into this linker completely abrogated contraction and allowed us to isolate non-contractile sheaths from cells for mass spectrometry and electron microscopy analysis. This analysis revealed that non-contracted sheaths are stably associated with the inner Hcp tube and components of the T6SS baseplate. Overall, comparison of T6SS sheaths of Francisella novicida and Vibrio cholerae and the sheath of R-type pyocins shows that conserved structural features are involved in sheath contraction and provides insights into T6SS assembly and its mode of action.

VipB residue D333 is important for contraction in vivo
Interactions of charged residues were previously suggested to be important for contraction of T4 phage sheath and R-type pyocin sheath [28,30]. Analysis of interfaces that are expected to be present only in the contracted form of the T6SS sheath of V. cholerae suggested that VipB residues K223 and D333 located on two different VipB-VipB interfaces, one between protomers of a single sheath strand and the second between protomers on two adjacent strands, significantly contribute to the stability of the contracted structure [27]. This suggests that the energy released by the formation of these interactions could contribute to the sheath contraction.
To test this hypothesis, we mutated K223 and D333 to alanine and expressed the mutated vipB in a vipA-msfGFP, DvipB background to allow monitoring of sheath dynamics in live cells. The K223A mutation impaired assembly of the sheath, and no elongated sheaths were observed ( Fig 1C and Movie EV1). This was also reflected in the inability of the VipB-K223A mutant to kill target cells ( Fig 1A). Interestingly, the VipB-D333A mutant assembled into nondynamic sheath structures that were stable in vivo over more than one hour of imaging (Fig 1B and Movie EV1 and EV2). The number of sheath structures per cell was comparable to the number of structures assembled in wild-type cells suggesting that D333 residue is not critical for sheath assembly. Similarly to the extended wild-type sheaths, the VipB-D333A sheaths were stable even in the presence of ClpV. Importantly, the VipB-D333A mutation also completely blocked target cell killing (Fig 1A).

VipA linker is critical for sheath contraction in vivo
Recent atomic models of T6SS sheaths in a contracted state and structures of the R-type pyocin in an extended and contracted state identified intermolecular linkers important for sheath function [26,27,30]. Interestingly, the N-terminal linker of the R-type pyocin sheath appears more stretched in the contracted sheath than in the extended sheath ( Fig EV1). We hypothesized that stretching of the N-terminal sheath linker of a basal ring upon contraction results in pulling on the VipA N-terminal linker of the next sheath ring, which in turn triggers its contraction. Such a mechanism would lead to propagation of contraction through the whole sheath, as suggested earlier for the T4 phage sheath [33]. We decided to test this hypothesis by generating a series of mutant V. cholerae T6SS sheaths with longer VipA linkers. We inserted 1-7 amino acids of the native "AEVELPL" sequence of the linker after residue 25 of VipA wildtype protein (labeled here as VipA-N1 to VipA-N7). To monitor the assembly and contraction, we fused VipA and its variants to msfGFP and expressed it from pBAD24 plasmid in the absence of chromosomal vipA [1,27].   Images are composite images of phase contrast and fluorescence channels. Scale bars are 1 lm. C Ten-minute timelapse images of V. cholerae vipA-msfGFP, vipB À complemented with vipB wild type, vipB-K223A, or vipB-D333A on pBAD24. A dynamic, contracting (VipB wild type, white arrow) and a stable, static (VipB-D333A, black arrow) sheath are highlighted with arrows. VipB-K223A does not assemble T6SS sheaths. Images are composite images of phase contrast and fluorescence channels. Scale bars are 1 lm. D Crop of an electron micrograph of purified VipB-D333A sheath and below a plot of the summed intensities. The inner diameter is marked with black arrowheads and the outer diameter with white or gray arrowheads. Scale bar is 50 nm. All mutant T6SS sheaths assembled with a frequency similar to that of the wild-type sheath ( Fig 2B); however, the frequency of sheath contraction was strongly dependent on the linker length ( Fig 2B, Movie EV3). Whereas insertion of one amino acid (VipA-N1) had almost no effect on sheath dynamics, an elongation by two or more amino acids (VipA-N2-7) reduced the fraction of sheaths that contract within 5 min from 50% (of 159 VipA wild-type structures counted, 85 contracted) to 0% (of 204 VipA-N3 structures counted, none contracted). Many of these mutant sheaths were stable over one hour of imaging ( Fig 2C and Movie EV4). Sheaths that occasionally broke after extensive bending caused by the movement and growth of cells were however quickly disassembled (Movie EV4). VipA with "AGAGA" sequence inserted, labeled as VipA-N5(GA), also assembled stable full-length sheaths ( Fig 2B, Movie EV3), suggesting that VipA linker length, but not its sequence, is specifically critical for sheath stability. Furthermore, the killing of Escherichia coli MG1655 by V. cholerae T6SS (Fig 2A) was strongly dependent on the length of the VipA linker. Whereas an extension of this linker by one amino acid had no effect on the killing efficiency, an elongation by two amino acids completely abolished the killing of target cells.

EMBO reports
VipA-N5-msfGFP A Escherichia coli survival (AE SD, N = 3) after 3-h competition with indicated Vibrio cholerae strains in a 1:10 ratio on plate; the first and third columns are significantly different from the rest (P < 0.01). B Fluorescence timelapse images of V. cholerae vipA À complemented with indicated msfGFP-tagged-vipA variants on pBAD24. Arrows highlight sheaths that contract (wild type, N1, white arrows) or are static (N2 or higher, black arrows). Images are composite images of phase contrast and fluorescence channels. C Long timelapse of V. cholerae vipA À complemented with vipA-N5-msfGFP on pBAD24. Images are composite images of phase contrast and fluorescence channels. The length of the sheaths in vivo did not differ between wild-type and the non-contractile sheaths VipB-D333A or VipA-N3. The mean length of sheaths was between 0.53 lm and 0.59 lm for all measured mutants (Table EV1). This suggests that the length of sheaths is limited by the size of the cell and the orientation of the T6SS inside cells as shown previously [19] rather than by the mutations that were introduced.

Stable sheaths assemble from baseplate and around Hcp
To analyze the stable mutant sheaths in more detail, we isolated them by using an approach similar to the one used for the isolation of the wild-type contracted sheath [1,27]. Mutant sheaths were expressed in a non-flagellated V. cholerae strain, cells were lysed, and sheaths were purified from soluble proteins and cell debris using ultra-centrifugation. The isolated sheaths were analyzed by negative staining electron microscopy.
Analysis of VipB-D333A sheath sample revealed partially fragmented hollow structures with an outer diameter of 260 nm, thus resembling contracted sheaths ( Fig 1D). This suggests that during isolation, the VipB-D333A sheaths contract and the D333A mutation destabilizes the contracted structure, which leads to partial fragmentation. Similarly, VipA-N2 sheaths closely resembled the wild-type contracted sheaths as they appeared hollow, had the inner diameter of 100 Å , and the outer diameter of 260 Å (Figs 3A and B, and EV2). Interestingly, VipA-N3, VipA-N5, and VipA-N5(GA) mutant sheath diameters were %200 Å and thus narrower than wild type (Figs 3A and B, and EV2). Importantly, uranyl acetate stain was clearly unable to penetrate the sheaths, suggesting that VipA-N3, VipA-N5, and VipA-N5(GA) mutant sheaths were filled with additional proteins (Fig 3A and B).
To identify the proteins that were associated with the mutant sheaths, we subjected the VipA wild-type, N1, 2, 3, 5, and 5 (GA) samples to mass spectrometry analysis (Table EV2 and Dataset EV1). Besides VipA and VipB proteins, VipA-N3, VipA-N5, and VipA-N5(GA) sheath preparations contained large amounts of Hcp, as further confirmed by Western blot (Fig 3C,  lower panel). The presence of Hcp in these mutant sheath samples explains the solid appearance on negative-stain EM ( Fig 3A). Altogether, 10 of 16 T6SS-related proteins were identified in all independently purified triplicates. Interestingly, six of these proteins were specifically enriched in samples with a linker elongated by three or more amino acids (P < 0.02). Among the enriched proteins are baseplate proteins TssE, TssF, TssG, and TssK but also the tip component VgrG-3 (Fig EV3). VgrG-1 and VgrG-2 as well as most other proteins of the membrane complex have not been identified in all replicates and thus were excluded from the analysis.
Interestingly, VipA-N2 sheaths are stable in the cell cytosol in the presence of ClpV, however, can only be isolated in a conformation closely resembling contracted sheath. To test whether VipA-N2 sheaths contract during isolation, we imaged the mutant sheaths on an agarose pad containing low concentrations of CelLytic B and EDTA, the lytic agents that we use for bacterial cell lysis. We found that VipA-N2 sheaths shorten their length by %50% shortly before, during, or immediately after lysis, which indicates that VipA-N2 sheaths indeed assemble in an extended state but may contract during cell lysis (Fig EV4 and Movie EV5).

Mutant sheaths assemble only from a functional baseplate and the non-contractile phenotype is dominant
To exclude the possibility that the non-contractile sheaths are aberrant polymers that assemble independently of other T6SS components, we imaged their assembly in a strain that lacks the baseplate component TssE. In agreement with the previous observations that TssE is required for efficient sheath assembly [1,34], the frequency of wild-type T6SS sheath assembly decreased by 140-fold in the absence of tssE. Similarly, the assembly of D333A or VipA-N5 sheaths was clearly dependent on the presence of a functional baseplate since the number of structures assembled in the cells lacking tssE was reduced by at least 100-fold (Fig 4A and B). This indicates that stable mutant sheaths assemble from a baseplate similarly to the wild-type sheaths.
To test whether the mutant sheath subunits can block T6SS activity also in the presence of the wild-type subunits, we induced expression of the VipA-N5 mutant from pBAD24 plasmid in a strain expressing wild-type sheath from the chromosome and measured the efficiency of E. coli killing. Low-level induction of VipA-N5 by 0.01% arabinose decreased the T6SS-dependent killing of E. coli by 100-fold. The T6SS activity was almost completely blocked by increasing the concentration of arabinose to 0.1% (Fig 4C)    VipA was expressed from the plasmid (Fig 4C). Similarly, low-level expression of VipB-D333A decreased T6SS activity by 100-fold and high level of expression blocked the T6SS activity completely (Fig 4C). This dominant negative phenotype suggests that the mutant subunits are structurally compatible with the wild-type subunits, co-assemble into the same structures, and thus block T6SS function.

Concluding remarks
The T6SS is a highly dynamic system, and this complicates detailed biochemical and biophysical characterization of its mode of action.
Here, we show that the system can be locked in the pre-contraction state by mutagenesis of the linker connecting sheath subunits or changing interactions contributing to the stability of the contracted state. Importantly, some non-contractile sheaths are stable during isolation from cells, assemble around Hcp tube, associate with many T6SS baseplate components, and co-assemble with wild-type extended sheath. This suggests that the structure of non-contractile sheaths is very similar to the wild-type sheath. Moreover, similarly to the wild-type extended sheath, the non-contractile sheaths form in the presence of ClpV and are therefore different from the previously described polysheath-like structures, which form independently of other T6SS components but only in the absence of ClpV [11]. Early electron micrographs of partially contracted T4 phage particles suggested that sheath contraction progresses in a wave of contracting sheath rings from the baseplate toward the phage head [33]. However, it is currently unclear how contraction of one sheath ring triggers contraction of the next ring. As we show here, insertion of two residues into the VipA N-terminal linker prevents T6SS sheath contraction in vivo, suggesting that the exact length of the linker connecting the subunits is essential for sheath contraction initiation or propagation of the contraction along the sheath (Figs 5  and EV1). In addition, T6SS, T4 phage, and R-type pyocin sheath structures indicated charged residues potentially important for the stability of contracted structures [27,28,30]. Many of these A Fluorescence microscopy images of Vibrio cholerae vipA À (upper panels) or vipA À , tssE À (lower panels) complemented with vipA-msfGFP (left panels) or vipA-N5-msfGFP (right panels) on pBAD24. B Fluorescence microscopy images of V. cholerae vipA À , vipB À (upper panels) or vipA À , vipB À , tssE À (lower panels) complemented with vipA-msfGFP on pBAD24 and vipB (left panels) or vipB-D333A (right panels) on pBAD33. C Escherichia coli survival (AE SD, N = 2) after 3-h competition with indicated V. cholerae strains in a 1:10 ratio on plates at two different arabinose concentrations. interactions specifically form during contraction and were thus proposed to drive sheath contraction [28,30]. Together with our data, this suggests that the newly formed interactions of VipB residue D333 stabilize the quaternary structure of the contracted state and prevent reversal of the contraction, especially when contraction is not yet completed throughout the sheath. Interestingly, many baseplate components are enriched in the samples of the isolated stable non-contractile sheaths as compared to the samples of contracted sheaths (Table EV2 and Dataset EV1 and Fig EV3) [1,27]. This suggests that after sheath contraction, the baseplate is destabilized and dissociates from the sheath. Live-cell imaging of sheath dynamics and localization indeed showed that few seconds after contraction, but before disassembly, sheaths often dissociate from the initial cell envelope attachment site [16]. This is consistent with the observation that T6SS assemble repeatedly inside cells and the components of the baseplate are likely reused for new rounds of assembly. In related contractile nanomachines, which are only used once, the contracted sheaths remain stably associated with the baseplates and likely provide mechanical stability to the contracted particles. In the case of contractile phages, the sheath connects baseplate and the phage head as the DNA is translocated and in the case of R-type pyocin, the sheath might be needed to stabilize the tube, which allows ion leakage and the killing of target bacterial cell [12,30,35].
The approach used here to stabilize the pre-contraction state of T6SS will likely be invaluable for further attempts to dissect T6SS mode of action at the molecular level in various bacteria and may be also used to study related contractile nanomachines with a major relevance for viral infection, bacterial competition, and pathogenicity.

Bacterial strains and DNA manipulations
Vibrio cholerae 2740-80 parental, vipA-msfGFP, DvipA, DvipB, DtssE strains and the pBAD24-vipA-sfGFP plasmid were described previously [1,27]. vipA mutants on pBAD24 plasmid were generated using standard techniques. Mutant vipA genes encode "A, AE, AEV, AEVE, AEVEL, AGAGA, AEVELP, or AEVELPL" residues inserted right after residue 25 of wild-type vipA. The insertions represent either duplication of the native sequence or a sequence encoding "AGAGA". V. cholerae 2740-80 DvipA-vipB strain was created by replacing vipA and vipB with a gene encoding "MSKEGSVGRLDQA" peptide (first seven residues of vipA and last six residues of vipB fused in frame) and V. cholerae 2740-80 DvipA-vipB-tssE strain was created by replacing vipA, vipB, and tssE with a gene encoding "MSKEGSVRKYRVF" peptide (first seven residues of vipA and last six residues of tssE fused in frame) by allelic exchange as was done previously. vipB (wild type) was cloned into pBAD24 and pBAD33 plasmids using standard techniques. K223A and D333A mutations were introduced into vipB using mutagenic primers. All PCR-generated products were verified by sequencing. Plasmids were transformed into V. cholerae by electroporation. Gentamicin-resistant E. coli MG1655 strain with pUC19 plasmid was used in bacterial killing assays. Antibiotic concentrations used were streptomycin (50 lg/ml), ampicillin (200 lg/ml), chloramphenicol (20 lg/ml), and gentamicin (15 lg/ml). Lysogeny broth (LB) was used for all growth conditions. Liquid cultures were grown aerobically at 37°C.

Bacterial killing assay
Vibrio cholerae 2740-80 strains as indicated and E. coli MG1655 with empty pUC19 plasmid were incubated overnight at 37°C in LB supplemented with appropriate antibiotics. Cultures were diluted 100-fold, and grown to OD 0.8-1.2 in presence of appropriate antibiotics and 0.01% arabinose for strains with pBAD plasmids (or 0.1% arabinose for sheath co-assembly experiments). Cells were washed and mixed at final OD of %10 in 10:1 ratio (V. cholerae to E. coli) as specified, and 5 ll of the mixture was spotted on a pre-dried LB agar plate containing 0.01% arabinose and ampicillin or no antibiotic. After 3 h, bacterial spots were cut out and the cells were re-suspended in 0.5 ml LB. The cellular suspension was serially diluted (1:10) in LB, and 5 ll of the suspensions were spotted on selective plates (gentamicin for E. coli and streptomycin for V. cholerae). Colonies were counted

Fluorescence microscopy
Procedures similar to those described previously [16] were used to detect fluorescence signal in V. cholerae. Overnight cultures of V. cholerae carrying pBAD24 plasmid with the respective inserts were diluted 100-fold into fresh LB supplemented with ampicillin, streptomycin, and 0.01% or 0.03% arabinose and cultivated for 2.5-3.0 h to optical density (OD) at 600 nm of about 0.8-1.2. Cells from 1 ml of the culture were re-suspended in %50 ll LB (to OD %20), spotted on a thin pad of 1% agarose in LB, and covered with a glass coverslip. For experiments in which cell lysis was induced to test the stability of VipA-N2 sheaths, an agarose pad (1% agarose in PBS and LB 1:1) containing 0.0625× CelLytic B and 0.3 mM EDTA was used. Cells were immediately imaged at room temperature. A previously described microscope setup was used [27]. VisiView software (Visitron Systems, Germany) was used to record images. Fiji [36] was used for all image analysis and manipulations as described previously [37]. Bleach correction was used if necessary [38]. Contrast on compared sets of images was adjusted equally. All imaging experiments were performed with three biological replicates.

VipA/VipB sheath preparation
Overnight cultures of the indicated strains were diluted 1:1,000 in 0.5 l of fresh LB supplemented with appropriate antibiotics and then shaken at 37°C and 250 rpm to an OD of %1.2. Cells were centrifuged for 20 min at 5,000 g and 4°C, re-suspended in 20 ml PBS, and centrifuged again for 30 min at 3,214 g and 4°C. Pellets were frozen until further processing. The cell pellets were thawed, resuspended in 20 ml of TN buffer (20 mM Tris, 150 mM NaCl, pH 8.3), and lysed by addition of (0.75×) CelLytic ™ B, lysozyme (200 lg/ml), EDTA (5 mM), and incubation at 37°C. DNase (50 lg/ml) and MgCl 2 (10 mM) were added to cleave DNA. After 15-min incubation at 37°C, cell debris was removed by centrifugation for 20 min at 10,000 g. Cleared supernatants were subjected to ultraspeed centrifugation for 1 h at 104,000 g and 4°C, and the resulting pellet was washed with 1 ml TN buffer and subsequently re-suspended in 1 ml TN buffer; insoluble material was removed by centrifugation for 1 min at 10,000 g. Cleared supernatants were subjected to a second round of ultraspeed centrifugation for 1 h at 104,000 g and 4°C, and the resulting pellet was washed with 1 ml TN buffer and subsequently re-suspended in 1 ml TN buffer; insoluble material was removed by centrifugation for 1 min at 10,000 g. The supernatant was subjected to a third round of ultraspeed centrifugation for 1 h at 104,000 g and 4°C, and the resulting pellet was re-suspended in 70 ll of TN buffer for further analysis. Purity of the sample was assessed by Coomassie-stained SDS-PAGE.

Mass spectrometry
Sheath samples (30- Results from the database search were imported into Progenesis. The database search results were filtered, limiting the peptide and protein level FDR to 1%. The Progenesis analysis results were further processed using the SafeQuant R package [40], to obtain protein relative abundances. This analysis included summation of MS1 peak areas per protein followed by global normalization of protein abundances, per LC-MS run, based on the abundances of proteins VipA (VC_A0107) and VipB (VC_A0108). Finally, protein abundance ratios in Table EV2 were calculated with respect to the wild-type condition, upon averaging across biological replicates. Testing for protein differential abundance was done using empirical Bayes method [41]. The resulting P-values, reflecting the probability of detecting a given mean abundance difference across sample conditions by chance alone, were corrected for multiple testing using the Benjamini-Hochberg method [42].

Statistical testing
Unpaired Student's t-test was used to evaluate statistical significance of biological replicates for the bacterial killing assays and mass spectrometry.
Negative staining electron microscopy 300-mesh copper grids were glow-discharged for 20 s, and samples (5 ll, protein concentration approx. 0.1 lg/ml) were adsorbed for 1 min and blotted using Whatman #1 filter paper. The grids were washed five times with ddH 2 O, and once using 2% uranyl acetate, followed by a 20-s staining with 2% uranyl acetate. Grids were imaged on a CM-100 microscope (Philips N.V., Amsterdam, Netherlands) equipped with a Veleta 2 k × 2 k camera (Olympus K.K., Tokyo, Japan) at 80 keV and a magnification of 64,000× (7.4 Å pixel size, 1,376 × 1,032 pixel) or a KeenView camera (Olympus K.K., Tokyo, Japan) at a magnification of 20,000× (9.1 Å pixel size) and the iTEM user interface. Fiji [36] was used for all image analysis.

Molecular analysis
Structures of PA0622 in the contracted state (PDB ID: 3J9R) [30] and of VipA and VipB in the contracted state (PDB ID: 3J9G) [27] were aligned based on their 3D structures using UCSF Chimera [43].
Expanded View for this article is available online.