Abstract
The double tubular structure of the type VI secretion system (T6SS) is considered as one of the longest straight and rigid intracellular structures in bacterial cells. Contraction of the T6SS outer sheath occurs almost instantly and releases sufficient power to inject the inner needle-like Hcp tube and its associated effectors into target bacterial cells through piercing the stiff cell envelope. The molecular mechanism triggering T6SS contraction remains elusive. Here we report that the double tubular T6SS structure is strikingly flexible and elastic, forming U-, circular-, or S-shapes while maintaining functional for contraction and substrate delivery. We show that physical contact with cytoplasmic membrane induced a range of T6SS structure deformation, but the resultant mechanical pressing force on the T6SS baseplate did not trigger contraction. Our results also reveal a stalling intermediate stage of sheath-tube extension following which the structure contracts or resumes to extend. These observations suggest that the recruitment equilibrium of sheath-tube precursors to the extending structure is key to stability/contraction and lead us to propose a model of T6SS contraction, termed ESCAPE (extension-stall-contraction and precursor equilibrium). Our data highlight the remarkable flexibility of the double tubular T6SS structure and its length control mechanism distinct from the other evolutionarily related contractile cell-puncturing systems.
Introduction
Bacteria have evolved with various mechanisms to protect themselves and outcompete neighboring species in complex communities 1. The antagonistic interactions mediated by the activities of those bacterial weapons could dictate community structures and dynamic changes 2–4. The type VI secretion system (T6SS) is one such lethal weapon used by many Gram-negative bacteria to deliver toxic effectors into neighboring bacterial and eukaryotic cells5–8. Encoded by 13 conserved genes, the T6SS structure consists of a membrane complex TssJLM, a baseplate TssK-TssEFG, and a double tubular structure comprising an outer VipA/B (TssB/C) sheath encircling an inner needle-like Hcp tube 9–15. At the tip of the tube sits a spike complex VgrG/PAAR that not only sharpens the tube but also serves as a main loading site for effectors 16–19. The inner tube and the outer sheath share the same six-start helical structures with each ring composed of hexamers of Hcp and heterodimeric VipA/B, respectively 13,14. The T6SS structure assembles from the baseplate and polymerizes inward with new subunits added to the distal end, often spanning across the cell 10. The extended sheath is in a high energy conformation, and its contraction is proposed to initiate from the baseplate side to the distal end, shortening the sheath by half within two milliseconds 10,20. After contraction, the N-terminus of VipB is exposed and recognized by the AAA+ (ATPase Associated with various cellular Activities) ATPase ClpV that disassembles the contracted sheath and enables recycling of sheath subunits for another assembly 21–24. By contrast, the inner Hcp tube is ejected and requires de novo synthesis, hence under distinct transcriptional regulation from the sheath subunits 25.
Size determination of a cellular organelle is a critical control for its proper function, and yet it often remains poorly defined mechanistically due to the complexity 26–28. Molecular ruler proteins have been discovered for length control of the type III secretion system (T3SS) and its related flagellum hook that form extracellular structures protruding from the outer membrane 29–31, as well as for length control of phage tail 32. By contrast, it remains elusive regarding the mechanisms that dictate the intracellular T6SS sheath-tube length, terminate extension and trigger contraction. The T6SS extended sheath exhibits a quite narrow length distribution in normal rod cells 10,33, suggesting random contraction is rare and the size is somehow controlled. However, a specific ruler protein is likely absent because the extended sheath-tube structure size has been found to decrease when Hcp is limited and increase in spheroplast cells 10,34. Because the T6SS double tubular structure polymerizes from the membrane anchor and ends at the opposite side of the cytoplasmic membrane and because phage sheath contraction is believed to be triggered by baseplate changes resulting from phage fiber attachment to target receptors 20, it is possible that sheath contact with the cytoplasmic membrane at the distal end may generate a physical force to the T6SS baseplate, which subsequently causes baseplate conformational change leading to contraction. Although short yet contractible T6SS structures have been found 34, cellular geometry makes it difficult to rule out possible membrane contact. However, testing this physical force-driven contraction hypothesis directly poses a technical challenge because it is unfeasible to either remove the cytoplasmic membrane in a living cell or to assemble a functional T6SS in vitro.
Results
Physical force of membrane contact results in curved structures but does not trigger contraction
Here we designed an alternative strategy by inducing irregular intercellular membrane vesicles in cell wall-deficient cells, as previously reported in Listeria L-form cells 35,36. We postulate that, if such internal membrane barriers could terminate T6SS sheath growth and trigger contraction, it would strongly support membrane contact as a sufficient trigger (Figure 1A). We found that ampicillin-treated Vibrio cholerae spheroplast cells can accumulate internal membrane vesicles (Figure 1B) and decided to use it as a model.
Surprisingly, under vesicle-inducing conditions, we observed highly curved T6SS structures instead of vesicle-triggered early contraction as we initially expected (Figure 1C&D, Supplemental Figures S1A&B, Movie S1). Upon contact, the membrane did not trigger contraction but rather guided a polymerizing straight sheath to bend along and continue to extend. After the sheath’s distal end contacted the membrane, we observed an arching deformation suggesting there is continual addition of VipA/B and Hcp precursors to the growing structure (Figure 1C, D &E). The curved sheaths can still increase their length on average by twofold in comparison with the initial straight length or the linear distance from baseplate to the distal end (Figure 1E). The deformation also suggests that sheath growth generates a bending force that directly presses on the T6SS baseplate-membrane anchor and is counteracted by the physical strength of the cytoplasmic membrane (Figure 1F). Notably, such force is strong enough to bend the sheath-needle double tubular structure but insufficient to trigger contraction (Figure 1F). There are also curved extending sheaths that eventually stopped extending without meeting any physical barriers and in some cases, the polymerizing sheaths fully encircled the whole circumference of the cell (Figure 1G).
To rule out the possibility that the curved extended sheaths are a result of some unknown properties of ampicillin other than targeting cell-wall synthesis, we examined the T6SS formation in the deletion mutant of mrcA, encoding the cell-wall synthesis enzyme PBP1a whose absence is known to disrupt the cell wall and causes a spherical shape in V. cholerae 37,38. We found that curved T6SS sheaths were similarly formed in both, ampicillin-treated cells and the mrcA mutant (Figure 1G, Supplemental Figure S1C). We measured the curvature of 120 curved sheaths and found that the average curvature of the flexible sheaths formed in ampicillin-treated cells is 0.64 μm−1 (± 0.13; n=80), which is comparable in the mrcA mutant (0.69 μm−1 ± 0.17; n=40) (Figure 1G). When we measured the curvature at each point along individual sheaths, maximum curvature points reached up to 2.3 μm−1 (Mean= 1.22 μm−1 ± 0.34; n=120) (Supplemental figure S1D). These results demonstrate the remarkable flexibility of the T6SS double tubular extending sheath/Hcp structure, previously described as a rigid structure, and indicate that contact with the cytoplasmic membrane is not a direct trigger for contraction.
Determination of curvy limit of extended sheath
Considering the six-start helical strands of the sheath and its inner Hcp tube comprising hundreds of rings of VipA/B and Hcp hexamers, we then asked what might have changed at each sheath ring that collectively account for the observed curvature. For simplicity, we postulate that the bending extending force leads to the stretching of sheath interaction with increased height/rise on one side of each sheath ring when curving (Figure 2A). We define the ratio of height change as a stretching factor (S) that is an intrinsic parameter of the sheath-tube structure dependent on the strength of the protomer interactions.
Because the width of the sheath (w = 20 nm) and rise of each ring (h = 3.8 nm) are known in V. cholerae 13, and the radius for the curved sheaths (r) can be measured (Figure 1G), we could estimate the stretching factor (S) equal to the ratio of sheath width to the radius (w/r) or the product of curvature and sheath width (c × w) (Figure 2A). When considering average curvature (c ≤ 1 μm−1) and assuming that the stretching force is equally distributed along the sheath, we estimated that S is less than 2%, and the maximum change of each ring is 0.08 nm and about 0.04% change relative to the sheath width. If local maximum curvature is considered (c = 2.27 μm−1) (Supplemental Figure S1F), 4% of stretching could be reached at certain points of the sheath, which corresponds to a rise change of 0.16 nm on each affected ring. However, because curvature is measured based on the 2-D projection of the 3-D sheath structure, such maximum curvature is likely an overestimation. By comparison, the rise of sheath decreases from 3.80 nm to 3.78 nm after contraction in V. cholerae, and the T6SS extended sheath rise is 3.7 nm in Myxococcus xanthus 13,39. Importantly, w and S are intrinsic parameters of the sheath/Hcp structure that ultimately dictate the flexibility potential of the T6SS and define a tolerable range of curvature (Figure 2B). Indeed, we noticed that curvy sheath formation is dependent on the geometry of the cell-sheath contact region. If the contact angle is close to orthogonal, the theoretical curvature at the contacting point will far exceed the maximum tolerable curvature value and thus no curved structures can be formed (Supplemental Figure S1E).
Curved sheath structure is elastic and restores to straight forms after contraction
Though extended sheaths are curved, contraction reduces the curvature greatly, leading to either straight contracted sheaths (70%, n=34) or much less curved contracted sheaths (30%, n=15) prior to their disassembly (Supplemental Figures S2A,B&C). Interestingly, curved sheaths with a higher curvature value and longer sheaths tended to break at the highest curvy points into straight short pieces due to contraction, prior to disassembly (Figure 2C, D&E, Supplemental Figure S1D). The contracted and break-free sheath pieces were similarly disassembled, suggesting the disassembly process does not require attachment to baseplate (Supplemental Figure S2A). Considering contraction separates at least one end of the sheath from the membrane, the straightening of contracted sheaths suggests that the more compact contracted structures might be less flexible to adopt the curved form and membrane contact is required to confine the sheath into the curvy form.
Next, we asked if the curved extended structures are elastic rather than fixed into a permanent shape. We postulated that removal of membrane constraint would straighten the sheaths. To test this and avoid the complexity of contraction, we used the vipA-N3 mutant in which the sheath-Hcp tube is locked into the extended form 40. The non-contractile extended sheath could assemble curved structures similar to wild type while no contraction was observed (Figure 2F, Movie S2). When cells were treated with detergent TritonX and polymyxin B to induce lysis, the curved structures straightened after escaping from the cells (Figure 2G, Supplemental Figure S2D, Movie S3), indicating that the extended sheath structures are indeed elastic.
Curved T6SS structures maintain functionality
To test if the curved T6SS sheaths represent functional T6SS structures in delivering effectors, we used a sensitive T6SS complement assay using the T6SS null hcp1/2 and vgrG2 mutants as reporter 34 The reporter strains cannot assemble T6SS sheath/Hcp structure due to the lack of key structural components unless the missing components are provided through delivery by a donor cell 34. Indeed, under the curvy-sheath inducing conditions, we readily detected sheath formation in both reporter strains while none detected when the donor was the T6SS mutant (Figure 3A, Supplemental Figures S3A&B). More sheaths were observed in the hcp mutant while longer sheaths were observed in the vgrG2 mutant (Figure 3B), reflecting the fact that one VgrG trimer and a stack of Hcp hexameric rings are needed to assemble a functional extended sheath-Hcp structure.
The H1-T6SS in Pseudomonas aeruginosa PAO1 can respond to external T6SS attacks and thus can be used as a reporter for T6SS delivery as well 41. Indeed, under the curvy-sheath inducing condition, significantly more T6SS activities in PAO1 were observed when PAO1 cells were neighboring wild type V. cholerae than when surrounding the hcp mutant (Figure 3C, Supplemental Figure S3C).
Next we tested whether the curved sheath could contract normally and be disassembled by the ATPase ClpV. The otherwise buried VipB N-terminal domain is exposed only after sheath contraction and specifically recognized by ClpV that disassembles the contracted sheath 21–24. The curved extended sheaths were not recognized by ClpV but contraction quickly led to ClpV coating of the sheath, indicating the VipB N-terminus is inaccessible in the curved extended sheath the same as in the straight extended sheath (Figure 3D, Supplemental Figure S3D, Movie S4). These results collectively indicate that the curved extended sheath-tube structures remain functional.
Stalling is an unstable intermediate stage between extension and contraction
Imaging analysis shows that in both rod (n=150) and spheroplast (n=80) cells, there is a stalling stage (~ 50 seconds, median value) during which the extended sheath maintains unchanged (Figure 4A, Supplemental Figure S4A). After the stalling, the extended sheath could continue to extend or undergo contraction. T6SS sheath assembly is also strictly dependent on, and proportional to, the availability of Hcp under curvy sheath inducing conditions, indicating that the observed curvy structures are not due to abnormal assembly of empty sheath (Figure 4B, Supplemental Figure S4D). Interestingly, despite the length difference, the number of sheaths per cell remains comparable under different Hcp inducing levels (Figure 4C), suggesting that the T6SS assemblies initiating from different baseplate complexes could compete for limited amounts of Hcp. In addition, VipA-sfGFP precursor levels fluctuated in inverse correlation with the total length of polymerized sheath, indicating an intracellular recycling balance between free and polymerized sheath components (Figure 4D, Supplemental Figures S4B&C).
ESCAPE (extension-stall-contraction and precursor equilibrium) model for sheath-tube extension and contraction
Based on these observations, we propose that the T6SS sheath-tube length and contraction are dependent on an equilibrium of stabilizing precursor addition and destabilizing tendency for contraction (Figure 4E). Analogous to an enzymatic reaction, precursors of VipA/B sheath and Hcp tube tend to polymerize to form the extended sheath-tube structure. However, polymerization enters a stalling stage when no new precursor is added, either due to depletion of precursor or limitation of physical space (most likely in rod cells). When precursors can be recruited again, either by recycling of VipA/B sheath subunits and de novo synthesis of Hcp, or the distal end is freed due to membrane flexibility, the stalling structure restarts extension. Otherwise contraction occurs and ejects the inner Hcp tube and effectors outward.
Discussion
Contractile needle-like cell-puncturing devices are often involved in delivering toxins into target cells. The T6SS is mechanistically and evolutionarily related to R-type pyocins 42,43, Photorhabdus virulence cassettes (PVC) 44, Serratia anti-feed prophage 45, and contractile phage tail 46,47. Their built-in strength for puncturing and distinct concentric double tubes implicate the structural rigidity. The sheath is assembled into a high energy extended state and contraction completes almost instantly (less than 2 milliseconds for T6SS) 10, which enables a powerful penetration to breach the target cell. Here we report that the T6SS structure is not rigid but rather strikingly flexible and elastic. The double tubular sheath-tube structure could accommodate physical force to deform while maintaining connectivity and functionality. It is generally believed that the baseplate of evolutionarily related contractile phages switches to contraction by sensing physical changes resulting from the attachment of tail fibers to receptors 20,48,49. By contrast, we demonstrate that the pressing force, resulting from sheath-membrane contact and acting on the T6SS membrane-baseplate complex, could cause straight to curvy deformation but could not trigger contraction (Figures 1C&D).
Based on our results and previous findings that T6SS sheath length correlates with Hcp abundance and physical space 10,34, we propose a precursor-equilibrium-mediated stability model (Figure 4E). The effect of precursor abundance on sheath stability could be direct and/or indirect. In the former scenario, precursor recruitment could directly transmit a stabilizing signal from the distal end to the baseplate of an extending sheath. Such stabilization may involve a conserved T6SS protein TssA that directly interacts with the sheath-tube precursors Hcp and VipB 50,51. It has been proposed that a dodecameric TssA complex functions as a cap at the sheath distal end to recruit sheath-tube precursors and stabilize the structure 10,51, while another model suggests that TssA remains attached to the baseplate 50. Notably, there are two distantly related TssA homologs in V. cholerae. Deletion of tssA1 but not tssA2 results in abolished T6SS secretion 52. However, the exact functions of these two TssAs remain elusive and warrant further studies. Alternatively, precursor might stabilize the extended sheath indirectly by inactivating some unknown contraction-triggering proteins or preventing their interaction with the extending sheath.
Why are the T6SS structures curved and why are they not seen in normal rod cells? All sheath-tube structures are straight while extending until their distal ends touch the membrane. Continual addition of precursors at the distal end would generate a conflict between the tendency to remain straight and the physical confinement of the membrane. Such conflict results in a mechanical bending force exerted on the sheath from the membrane contacting points, mainly the membrane anchor and the distal end (as seen in Figure 1c). The resulting stress could force a range of angle changes at the membrane contacting points in spheroplast cells that lack a rigid cell wall. Such angle change is key to accommodate the continual addition of precursors but is much less likely to occur in rod cells considering that the T6SS essential TssJLM membrane complex spans across the two membranes and the rigid cell wall. We also found straight sheaths at early time points but curvy sheaths at later time points after ampicillin treatment, further supporting that cell wall inhibits curvy sheath formation. Notably, a previous report using similar ampicillin-treatment in V. cholerae found long but straight sheath when imaged at earlier time points (40 min) 10. It is also possible that some unknown factors controlling the sheath shape are depleted or enriched at later time points in cell wall deficient cells. Such factors may be identified using genetic screening of random transposon mutants combined with high throughput imaging tools.
The six-start helical structure of both the inner and outer tubes 13 likely provide the structural basis for the stretching, curvy formation and elasticity. Using time-lapse fluorescence microscopy, we could monitor the dynamic assembly of curved extended sheath and provide a theoretical estimate on the stretching potential of the T6SS structure. The observed flexible T6SS structures are phenotypically reminiscent of the long and flexible tails of non-contractile phages53, suggesting a common ancestry.
Considering the relatedness of all known contractile double-tubular structures, the remarkable flexibility of T6SS sheath-tube might represent a general nature of elasticity among similar contractile injection systems.
Materials and Methods
Bacterial strains and growth conditions
All strains and plasmids used in this study are listed in Tables S1 and S2, respectively. Cells were grown aerobically at 37 °C in LB (1% [wt/vol] tryptone, 0.5% [wt/vol] yeast extract, 0.5% [wt/vol] NaCl). Antibiotics and inducers were used at the following concentrations: chloramphenicol 25 μg/mL (for E. coli) and 2.5 μg/mL (for V. cholerae), 20 μg/mL gentamycin, 100 μg/mL streptomycin and 25 μg/mL irgasan. Gene expression vectors and precise knockout mutants were constructed as previously described 54. All constructs were verified by sequencing. All primers used in this study are provided in Table S3.
Induction of Membrane Vesicle (MV) formation by ampicillin treatment
Overnight cultures were diluted 1/100 in 2 mL of fresh LB and grown aerobically at 37 °C to an OD600 of 0.6-0.8. Ampicillin was added to the cultures at 500 μg/mL and further incubated for 45 min. Culture was centrifuged and pellet was resuspended in 0.5X PBS buffer. Cells were spotted onto 1% agarose-pads supplemented with diluted LB (1:5) and PBS buffer (1:2), and ampicillin (10 μg/mL). Spotted microscopy slides were incubated at 37 °C and imaged at different time points.
V. cholerae AmrcA spheroplasts
Overnight culture of the ΔmrcA strain was centrifuged and the pellet was resuspended in fresh LB. After incubation at 37°C for 1 h, cells were concentrated to OD600 =10 and spotted onto 1% agarose-0.5X PBS pads supplemented with diluted LB (1:5). The spotted microscopy slides were incubated at 37 C prior to imaging.
Cell lysis induction
V. cholerae VipA-N3-sfGFP cells were grown to OD600=0.6-0.8 and treated with 500 μg/mL ampicillin for 45 min. Cells were then collected and spotted onto an agarose pad under membrane vesicle inducing conditions, as explained before. After incubation at 37 □C of the spotted microscope slides, the coverslip was carefully removed and 1 μl of 80 μg/mL colistin-0.1% Triton X plus 1 μl of 10 μg/mL DAPI were deposited on top of the agarose pad. A new coverslip was added and cell lysis was quickly monitored by time-lapse fluorescence microscopy at a frame rate of 10 seconds per frame until all observed cells were lysed (approximately 30 minutes).
Functionality assays under curvy structure conditions
For the T6SS interbacterial complementation assay, Δhcp1/2 VipA-sfGFP and ΔvgrG2 VipA-sfGFP strains were used as recipient cells. Wild type VipA-mCherry (T6SS +) or ΔvipB (T6SS -) were used as donor strains. Both, donor and recipient cells were treated similarly with 500 μg/mL of ampicillin, harvested and then mixed to a 1:3 ratio (recipient to donor) for imaging. Sheath formation in the recipient strains was monitored using time-lapse fluorescence microscopy at a frame rate of 10 seconds per frame for 10 minutes.
For the P. aeruginosa PAO1 reporter assay, V. cholerae wild-type (T6SS +) and Δhcp1/2 (T6SS -) VipA-mCherry strains were grown to exponential phase and treated with ampicillin. Cells were spotted onto 1% agarose-0.5X PBS pads supplemented with LB and ampicillin and incubated to induce curvy sheaths formation. The glass coverslip was carefully removed and, exponentially growing PAO1 TssB-sfGFP rod cells were spotted on top of the V. cholerae containing-agarose pads. A new glass coverslip was added and cells were immediately imaged every 10 sec using time-lapse fluorescent microscopy.
To image ClpV-mediated disassembly, V. cholerae VipA-mCherry ClpV-sfGFP cells were treated as explained before for curved sheaths formation and visualized after incubation at 37 C on the agarose pad. Cells were imaged by time-lapse fluorescence microscopy for 8 minutes at a frame rate of 5.5 sec per frame.
Hcp complementation assay
Exponentially growing cells (0D600=0.6-0.8) of V. cholerae Δhcp1/2 VipA-sfGFP strain either, empty or carrying a pBAD18-hcp1 plasmid were treated with 500 μg/mL of ampicillin for 45 min and spotted onto agarose pads. To induce Hcp expression, different concentrations (0.001%, 0.01% or 0.1% [wt/vol]) of L-arabinose were added to the agarose pads. Cells were imaged after 2 h incubation at 37□C.
Image acquisition
All images were acquired by a Nikon Ti-E inverted microscope with a Perfect Focus System (PFS) and a CFI Plan Apochromat Lambda 100X oil objective lens. Different fluorescence signals were filtered and excited using Intensilight C-HGFIE (Nikon), ET-GFP (Chroma 49002), and ET-mCherry (Chroma 49008) filter sets. Images were recorded by an ANDOR Clara camera (DR 328G-C01-SIL), pixel size 60 nm and, NIS-Elements AR 4.40 software.
Image analysis
Fiji was used for all image analysis 55. To correct for photobleaching, fluorescence intensity of the acquired time lapse videos was normalized to the same mean intensity for each image in a time series as described previously 41. The plugin “Temporal-Color Code” for Fiji was used to visualize the polymerization trajectory of the curved sheath in Figure 1D. Sheath length of straight and curved sheaths was measured using straight and freehand line tools, respectively. To calculate sheath contact angle with the cell membrane, the angle tool for Fiji was used. The curvature analysis was performed using the plugin “Kappa” for Fiji. The curvature is defined as the reciprocal of the radius (r−1). For sheaths with negative curvature points, the absolute values of curvature were used to build graphs and calculate mean values. Five points were used to define each curved sheath and the scale was set to 0.064 μm/pixel. We measured 120 fully extended curved sheaths (80 from ampicillin treated and 40 from ΔmrcA spheroplasts). To quantify cytosolic VipA-sfGFP fluorescence intensity, the average mean gray value of five circles (0.13×0.13 μm) distributed in the cytosol of the cell minus average fluorescence background outside the cell was calculated at each time point. “KymoResliceWide” plugin for Fiji was used to generate kymographs. Freehand line tool was used to define the sheath area to generate the kymographs.
Quantification and statistical analysis
The unpaired two-tailed t-test as well as one-way ANOVA with multiple comparisons and Tuckey post hoc test were performed using GraphPad Prism version 7.04. Number of cells analyzed as well as significance of each comparison are indicated in figure and figure legends. All experiments were performed with at least three biological replicates. Unless indicated differently, values are expressed as mean±SD.
Contributions
T.G.D. conceived the project. M.S.S, X.L. and T.G.D. designed the experiments. M.S.S., X.L., M.W., and S.H. performed the experiments. M.S.S. and T.G.D. prepared the manuscript.
Competing interests
The authors declare no competing interests.
Acknowledgments
This work was supported by grants from the Canadian Institutes of Health Research and the Natural Sciences and Engineering Research Council (NSERC) to T.G.D. T.G.D. is supported by a Canada Research Chair award. We thank members of the Dong lab for providing reagents, general support and critical reading of the manuscript.