Endomembrane breaching is a crucial strategy employed by intracellular pathogens enclosed within vacuoles to access the nutrient-rich cytosol for intracellular replication. While bacteria use various mechanisms to compromise host membranes, the specific processes and factors involved are often unknown. Shigella flexneri, a major human pathogen, accesses the cytosol relying on the Type Three Secretion System (T3SS) and secreted effectors. Using in-cell correlative light and electron microscopy, we tracked the sequential steps of Shigella host cell entry. Moreover, we captured the T3SS, which projects a needle from the bacterial surface, in the process of puncturing holes in the vacuolar membrane. This initial puncture ensures disruption of the vacuole. Together this introduces the concept of mechanoporation via a bacterial secretion system as a crucial process for bacterial pathogen-induced membrane damage.

Shigella T3SS drives multistep cytosolic access

We aimed to analyze the implication of the T3SS in the early steps of Shigella cytosolic access at high spatiotemporal resolution. For this purpose, we established a double reporter HeLa cell line expressing eGFP-Lysenin and mOrange-Galectin-3 to simultaneously track pathogen-induced vacuolar damage, and subsequent vacuolar rupture using time-lapse confocal imaging (Fig 1A). Upon vacuolar damage, sphingomyelins translocate to the cytosolic membrane leaflet, triggering detection by Lysenin [17]. This is followed by irreversible rupture of the vacuolar membrane, exposing glycans to the cytosol and enabling Galectin-3 to diffuse into the vacuolar lumen [18]. This sequence of events precedes later stages of vacuolar disassembly.

We used our dual reporter to investigate the direct role of the Shigella T3SS in membrane breaching and elucidate its implication in vacuolar damage or rupture. To this end, we employed engineered Escherichia coli strains expressing functional Shigella T3SSs [14,19], including the translocon pore proteins IpaB and IpaC (S1 Fig). To bypass the need for specific Shigella effectors for host cell entry, we utilized two E. coli systems: E. coli mT3_Shigella [19] expressing the structural components of the T3SS plus a reduced set of entry effectors (IpaA, IpgB1, IcsB, IpgD), and E. coli mT3_ShigellaΔeff_Inv [14] expressing only structural components of the T3SS while lacking entry effectors but exploiting Yersinia Invasin (Inv) [20] for receptor-mediated entry.

At 2 and 3 h post-infection respectively, immunofluorescence showed that both E. coli mT3_Shigella and E. coli mT3_ShigellaΔeff_Inv, invaded our reporter cell line (S2 Fig). Time-lapse microscopy revealed the dynamics of Lysenin and Galectin-3 recruitment at vacuoles containing Shigella, E. coli mT3_Shigella and E. coli mT3_ShigellaΔeff_Inv (Fig 1B, 1D, and 1F). Shortly after epithelial cell entry (average 4 ± 3 min), Shigella vacuoles were uniformly Lysenin-positive, indicating membrane damage. The abrupt onset of the Lysenin signal around the entire vacuole is due to the rapid lateral diffusion of sphingomyelins within the cytoplasmic membrane leaflet upon flipping at sites of membrane injury [21,22]. This step was always followed by progressive Galectin-3 recruitment (average 7 ± 4 min after entry, Fig 1C), consistent with the kinetics of Lysenin and Galectin-8 recruitment at the vacuole [17], thereby validating our dual-reporter system. Monitoring E. coli mT3_Shigella cytosolic access, we noted extended delays of Galectin-3 recruitment after the onset of Lysenin signal (Fig 1E). For both, E. coli mT3_Shigella and E. coli mT3_ShigellaΔeff_Inv infections, the vacuolar membrane remained around the bacteria even after rupture (Fig 1D and 1F), suggesting defects in vacuole disassembly, likely due to the absence of specific effectors involved in membrane unpeeling [12].

By examining the delay between vacuolar damage and rupture through Lysenin and Galectin-3 recruitment, we found that for Shigella, Galectin-3 was recruited on average 3 ± 3 min after Lysenin (Fig 1G). In contrast, we observed a 5-fold delay in Galectin-3 recruitment to E. coli mT3_Shigella vacuoles, with rupture being detected on average 16 ± 9 min after Lysenin. These results indicate that cytosolic access by Shigella occurs in sequential steps: vacuolar damage followed by rupture, each with distinct kinetics. Notably, E. coli mT3_ShigellaΔeff_Inv vacuoles ruptured on average only 2 ± 3 min after initial membrane damage, similar to Shigella, indicating that receptor-mediated entry through Inv promotes vacuole damage and rupture. Together, our results using engineered E. coli strains as a model for invasion using a minimal T3SS, hints towards an important role of the T3SS in initiating membrane injury.

In-cell analysis of T3SS organization

We reasoned that close interactions between T3SSs and the vacuole would be critical to promote host endomembrane injury. We turned to in-cell cryo-electron tomography (cryo-ET), a technique allowing the visualization of bacterial molecular machineries within infected cells [23]. To follow the involvement of the T3SS in the steps leading to vacuolar rupture, we took advantage of cryo-correlative light and electron cryo-microscopy (cryo-CLEM), imaging Shigella-infected HeLa cells expressing our stage-specific double fluorescent reporter to target precise invasion steps. Vitrified cells were first imaged using cryo-fluorescence microscopy (cryo-fLM) to localize fluorescently tagged Shigella. Guided by the cryo-fLM information, cells were thinned into lamellae at infection sites using cryo-focused ion beam (cryo-FIB) milling [24]. Then, zones of lamellae with intracellular bacteria were imaged by cryo-ET. To precisely determine the infection stages at both the single-cell and single-bacterium level, we developed a high-throughput correlation procedure (see Materials and methods section). Following data acquisition, all imaging modalities were correlated to provide detailed insight into the invasion stage of individual bacteria (S3A–S3C Fig). This cryo-CLEM approach enabled both consistent data collection of intracellular Shigella and allowed for precise assessment of vacuolar integrity surrounding each bacterium. Together, integrating in-cell ultrastructural data and functional fluorescence information was crucial for quantitative image analysis of the successive Shigella internalization stages.

To examine the key stage of Shigella cytosolic access, we imaged cells after short infection times (10 min post-infection). At this time point, correlative analysis showed that bacteria were predominantly found entrapped in intact vacuoles, negative for both Lysenin and Galectin-3 (Fig 2A). We also captured the transient stage of vacuole damage, with vacuoles positive for Lysenin but negative for Galectin-3 (Fig 2B). At both stages, the vacuolar membrane tightly surrounded the bacterium, consistent with previously reported volume-imaging of Shigella before rupture [11]. Of note, in most of our tomograms, we could spot spherical electron densities within the tight vacuolar lumen possibly resembling small bacterial outer membrane vesicles [25,26] (Fig 2B, insets 1 and 2 outlined arrowheads). We observed the membranes from ruptured vacuoles, positive for both Lysenin and Galectin-3, still in proximity to the invading bacteria but displayed pronounced morphological differences (Fig 2C). Indeed, it was reported that remnants of Shigella ruptured vacuole can either disassemble as entire segments, small pieces or fragments, or large sections may remain associated with the bacterium [12,13,27]. In accordance, segmentation of vacuolar membranes from Shigella exposed to the cytosol showed significant disruptions including interruptions of varying sizes, delamination, and fragmented membrane remnants (Fig 2C). Notably, damaged and ruptured vacuoles were coated by electron-dense layers that may reflect the accumulation of our fluorescently tagged markers Lysenin and Galectin-3 (S4 Fig).

Throughout the sequential infection stages, we imaged a fair number of densities with the shape and size matching Shigella T3SSs (n = 122) for which we distinctively identified the major substructures: the bacterial membrane spanning basal bodies and protruding needles. These complexes were exclusively found for T3SS-expressing bacterial strains (S1 and S5 Figs). We also found putative basal bodies without needles (n = 25, S6A Fig), suggesting that not all T3SSs are fully assembled prior to bacterial entry. When T3SSs needle complexes and putative basal bodies were identified, they were frequently present in clusters. To gain general insights into the organization of T3SSs at the bacterial surface, we assessed their distribution and found that T3SSs were located to both the poles and the bodies of the bacterial surfaces (41% placed on homogeneously curved membranes and 59% placed on straight membrane segments, respectively; Fig 2D). This observation contrasts with previous reports on the polar secretion of IpaC at host cell invasion [28] and questions a potential redistribution of the Shigella T3SSs from a restricted zone to the entire bacterial surface upon host entry, as observed for the T3SSs of Chlamydia elementary bodies [29]. To further characterize the localization distribution of T3SSs, we annotated each T3SS, and measured the cartesian distances between nearest neighbors of individual T3SSs. The median T3SS inter-distance was 767.7 nm, but 25% of total T3SSs were only 195 ± 92.5 nm apart (Fig 2E), agreeing with our observation that T3SSs were frequently present in clusters.

T3SSs displayed a range of morphological features in relation to the vacuole membrane, either contacting it or not. Notably, some unengaged T3SSs exhibited a bulb-like density at the needle tip (S3D Fig, Damaged vacuole, inset 5), which we hypothesize to be the needle tip complex, described to be wider than the needle filament, in the case of the Salmonella SPI-1 T3SS [30]. Nonetheless, T3SSs were mainly found associated with the vacuole during the early stages of cytosolic access (intact: 82%, damaged: 73%) and remained mostly engaged to vacuolar membrane remnants after vacuolar rupture (rupture: 69%; Fig 2F). These T3SSs displayed significant bending, with the entire structure tilted or with needles deviating from the basal body axis (Fig 2C, insets). Previous work has shown that Shigella reroutes the dynein motor complex to generate forces to efficiently remove vacuolar membrane remnants from the bacteria [12]. We speculate that the observed T3SSs distortions may be induced by the forces generated during this process, which would in turn also pull on T3SSs connected to vacuolar remnants. This interpretation was further supported by deformed bacterial membranes observed around tilted T3SSs (Fig 2C, arrowheads). Notably, despite this marked phenotype suggesting the presence of mechanical constraints at the T3SS-vacuole interface, bacteria with distant vacuolar membrane remnants displayed seemingly intact T3SSs at their surface with needles exposed to the host cytosol (S6B Fig). Possibly these T3SSs can be recognized by the inflammasome [31] and may be important for priming Shigella secondary infections [32].

Vacuole tightness constrains the T3SS needle complexes

Surprisingly, T3SSs with tilted morphologies were also detected at intact vacuoles indicating that the T3SS-vacuole interface is already under substantial tension before membrane rupture. To assess the potential impact of vacuolar constriction on individual T3SSs, we designed an analytical workflow for unbiased and quantitative analysis of intermembrane distances in three dimensions (3D). For this, membranes were semi-automatically segmented and the distances between the vacuole and bacteria outer membranes were determined in a spatially resolved manner using nearest neighbor distance calculation. During early invasion stages, the relative distance between the vacuole and the bacterial outer membrane showed tight spacing (Fig 3A) in line with the skewed distribution of calculated nearest neighbor distances (Fig 3B, gray and yellow). As expected, the space between the bacterium and the ruptured vacuole was enlarged (Fig 3A) with a wider spread distribution of vacuole-to-bacteria outer membrane nearest neighbor distances (Fig 3B, magenta). Vacuolar constriction did not change significantly from intact (48.0 ± 12.7 nm) to damaged vacuoles (48.0 ± 8.7 nm), indicating that membrane injury events occur in localized zones. The increased vacuole membrane to bacteria distances calculated for ruptured vacuoles (78.0 ± 9.2 nm) reflected general alteration of membrane integrity (Fig 3C).

We then measured the needle lengths of T3SSs (Fig 3D) before cell entry (extracellular: 47.8 ± 14.4 nm) and for intracellular bacteria until vacuole rupture (vacuoles intact: 47.8 ± 10.6 nm, damaged: 49.0 ± 13.0 nm, ruptured: 48.3 ± 11.2 nm). The mean T3SS needle length was constant throughout the infection stages and matched the available luminal space before rupture, yet their measured lengths were longer and more heterogeneous than previously reported [33]. Needles that did not establish contact with the vacuolar membrane were significantly shorter, averaging 42.8 ± 13.4 nm length. Notably, T3SS needles in contact with the host membrane averaged 50.0 ± 9.6 nm, with some extending over 50 nm, exceeding the mean gap between the bacteria and the vacuole (Fig 3E). While T3SSs that did not contact the host vacuole were rather straight (Fig 3F; average base to needle angle of 170.4° ± 7.8°), needles in contact with the vacuole showed pronounced bending with regards to the T3SS base (average 165.5° ± 11.8°). Together, this suggests that T3SSs mediate contacts between bacteria and the vacuole and that the vacuolar tightness exerts substantial tension on T3SS needles.

T3SS-induced mechanoporation of the vacuole

At the early infection stages (Galectin-3 negative) the vacuolar membrane did not exhibit zones that were prominently perturbed. Therefore, we reasoned that early membrane injury events should be assessed at the sites where individual T3SS are in contact with the vacuole. We closely examined the T3SS-vacuole interface prior to vacuolar rupture and correlated the needle length of individual T3SS to the surrounding local vacuole-to-bacteria distances and membrane curvedness [34] as quantitative measure of local membrane deformation in local neighborhood of the T3SS tips. T3SSs not contacting the vacuole had either short needles or were located to zones with relaxed vacuolar membranes (S7A and S7B Fig), indicating that contact requires matching between both the needle length and the inter-membrane space. This interpretation is further supported by the T3SSs with needles lengths fitting the luminal space that showed limited to no constraints on the vacuole or bacterial membranes (S7C Fig). Together, these observations imply an interplay between the needle length and the local vacuolar tightness exerting opposing forces at the T3SS-vacuole interface.

We next focused on T3SSs with longer needles that did not fit into the tight vacuoles. T3SSs with bent insertions across bacterial membranes (Figs 4A and S7D), only slightly deformed the vacuolar membrane. This indicates that the local vacuole constriction exerts a strong mechanical force on T3SSs with long needles, pushing the T3SS sideways resulting in a tilted complex. At the bacterial level, the T3SS tilt was echoed by deformations of the membranes surrounding the basal body (Figs 4A and S7D white arrowheads). In these cases, the local vacuole tightness is the main factor constricting the T3SSs. In other cases, T3SSs with long needles were inserted straight across the bacterial envelope and deformed the vacuolar membrane locally forming bulges, shown by strong increase of the curvedness at the T3SS-vacuole contact site (Fig 4B). Such strong membrane curvedness was not detected away from the T3SS (S7E and S7F Fig), highlighting the reciprocity between needle length and vacuole constriction leading to local membrane deformation. Further examination of the vacuolar membrane morphologies at these zones revealed membrane lesions very close to the longer needle contact sites (Fig 4C and 4D). These T3SSs were also bent, emphasizing the tensions generated by opposing forces necessary to induce membrane injuries. By segmenting the bacterial and host membranes and docking an available atomic model for the T3SS into our tomogram density we could visualize that the T3SS needle physically injured the vacuole membrane. We observed large holes localized near the long T3SS needles (Fig 4C and S10 Video) and punctures (Fig 4D). We could also spot multiple T3SSs contacting and inducing curvature of the vacuolar membrane in the same area (Fig 4D). Even though these membrane injuries were observed, these vacuoles were still negative for the rupture marker Galectin-3 (S8 Fig). Thus, we propose that the Shigella T3SS needle complex initiates vacuole membrane injury by physically deforming it and puncturing a hole, akin to mechanoporation [35].

Cell culture and stable cell line generation

All HeLa epithelial cell lines used in this study were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) supplemented with 10% (v/v) heat-inactivated fetal bovine serum (Sigma-Aldrich) at 37 °C, 5% CO₂ and were continuously monitored for mycoplasma.

The double expressing mOrange-Galectin-3 and eGFP-Lysenin stable cell line was generated using the Sleeping Beauty (SB) transposon system [55]. Briefly, N-terminally eGFP tagged Lysenin was PCR amplified from M6P-GFP-LyseninW20A plasmid [17] (kindly provided by Felix Randow) using the primers Forward_eGFP (5′aggcctctgaggccaccatggtgagcaagggcgag 3′) and Reverse_Lysenin (5′ aggcctgacaggcctcagcccacgacttccagga 3′). The obtained amplicon was cloned into the SB transposon (pSBbi-BLA, addgene plasmid #60526) using the Sfil sites. The resulting pSBbi eGFP-LyseninW20A plasmid was confirmed by sequencing and co-transfected with the vector encoding the SB100X transposase, pCMV(CAT)T7-SB100X (Addgene plasmid #34879) in the monoclonal HeLa pSBbi mOrange Galectin-3 [13] (puromycin resistant). Forty-eight hours post-transfection, cells were switched to selection media containing 10 µg/mL blasticidin (Gibco) and 1.7 µg/mL puromycin (Gibco). After 10 days of culturing in selection media, cells were sorted to select double fluorescence positive clones (eGFP-Lysenin and mOrange-Galectin-3) using a BD FACSAria III Cell Sorter (BD Biosciences). A polyclonal subpopulation with high expression of eGFP-Lysenin was further selected for expansion.

Bacterial strains and infection for fluorescence microscopy

Escherichia coli DH10β (Thermo Scientific), derivative minimal T3SS (mT3) strains: E. coli mT3 (pLLX13 ipaJ thru spa40, TET^R, KAN^R + pNG162-VirB, SPEC^R) [19] and E. coli mT3_ShigellaΔeff_Inv (pLLX13 ipaJ thru spa40ΔipaA, ΔipgB1, ΔicsB, ΔipgD, TET^R, KAN^R + pNG162-VirB, SPEC^R + pRI203, CM^R, AMP^R) [14] were grown in lysogeny broth (LB) at 37 °C with shaking. For the mT3 strains, LB was supplemented according to strain specification with 20 µg/mL tetracycline, 50 µg/mL kanamycin, 100 µg/mL ampicillin, and 100 µg/mL spectinomycin (all from Sigma-Aldrich). For E. coli infection experiments, bacteria were diluted in LB at 1:50 from an overnight culture and grown at 37 °C. In the case of the mT3 strains, T3SS expression was induced with 1mM IPTG (Thermo Scientific). At an optical density of 600 nm (OD₆₀₀) ~0.45 bacteria were harvested by spinning and washed twice in EM buffer (120 mM NaCl, 7 mM KCl, 1.8 mM CaCl₂, 0.8 mM MgCl₂, 5 mM glucose, 25 mM HEPES, pH 7.3). Bacteria were diluted to a multiplicity of infection (MOI) of 100 in EM buffer and spun down on the cells for 10 min at 180g, at room temperature (RT).

Shigella strains were derivatives of the WT Shigella flexneri M90T, when indicated they carried the pGG2-eGFP or pGG2-TagBFP [56] plasmids, for constitutive expression of eGFP or TagBFP, respectively. Shigella flexneri BS176, a M90T derivative lacking the virulence plasmid, was used as control for secretion assays [57]. Shigella strains were plated on trypticase soy (TCS) agar supplemented with 0.01% Congo Red (Sigma-Aldrich) and liquid cultures were grown in TCS broth at 220 rpm at 37 °C, when applicable the medium was supplemented with 100 µg/mL ampicillin. On the day of the infection, bacteria were sub-cultured from an overnight culture in fresh TCS at 1:100 dilution until OD₆₀₀ reached ~0.45. Bacteria were harvested, washed once in EM buffer, and incubated for 15 min at 37 °C with shaking in EM buffer supplemented with 1 µg/mL poly-l-lysine hydrobromide (Sigma-Aldrich). The bacterial solution was washed twice, and the bacteria were diluted to the appropriate MOI.

Congo Red induction assay

Congo Red effector protein secretion assay was carried out as described previously [58] with some modifications. Briefly, overnight bacterial cultures were diluted 1:50 and grown in fresh media as described previously. Bacterial suspensions were centrifuged, pellets resuspended in PBS to adjust 1 mL to an OD₆₀₀ = 2, and incubated with or without 1% Congo Red (Sigma) at 37 °C with shaking (500 rpm) for 30 min. Samples were then centrifuged at 14,000g for 15 min at 4 °C. The supernatants were collected, and proteins precipitated for 30 min on ice using 1:1 (v/v) trichloroacetic acid at a final concentration of 20%. Proteins were recovered by centrifugation at 14,000g for 15 min at 4 °C and the pellets washed twice with ice-cold acetone. Protein samples were denatured in Laemmli buffer, boiled for 10 min at 95 °C and run in a NuPAGE 10% Bis–Tris Gel (Invitrogen) with a protein ladder (26619, Thermo Scientific). The gel was fixed with 50% methanol, 7% acetic acid for 15 min at RT, washed in milliQ water and stained with GelCode Blue Stain reagent (Themo Scientific) for 1 h. After destaining, gel was imaged using an iBright 1500 imaging system (Invitrogen).

Fluorescence microscopy and immunolabeling

Imaging was carried out on a Nikon Ti-E inverted microscope equipped with a Perfect Focus System (PFS), a spinning disk confocal system (CSU-W1, Yokogawa), and an ORCA flash 4.0 camera (Hamamatsu). Time-lapse imaging was performed using a CFI S Plan Fluor ELWD 40×/0.60 air immersion objective (MRH08430, Nikon) and microscopy of fixed samples with a CFI Plan Apo VC 60×/1.2 water immersion objective (MRD07602, Nikon) or CFI Plan Apo 60×/1.4 oil immersion objective (MRD01605, Nikon). For fluorescent microscopy experiments, 40,000 cells were seeded into 8-well glass bottom microslides (ibidi). For time-lapse imaging, the microscope chamber was heated at 37 °C, and bacterial invasion was monitored immediately after bacteria addition to the cells. Images were recorded every minute for 2.5 h (Shigella) or 3.5 h E. coli mT3_Shigella. For microscopy of fixed samples, cells were infected with E. coli strains at 37 °C, 5% CO₂ for 2 h, while Shigella infections were carried out for 30 min at 37 °C. The inside-out staining protocol was adapted from another study [59]. Infected cells were washed three times in DPBS (Gibco) and fixed in 3.7% (wt/vol) Paraformaldehyde (PFA, Electron Microscopy Science), diluted in DPBS for 20 min at RT. All antibodies were diluted in an immunolabelling solution (2% BSA in DPBS) and incubated for 30 min at RT. First, the extracellular bacteria were labeled with a rabbit polyclonal antibody to E. coli (1:2,000; ab137967, Abcam) or a rabbit polyclonal antibody to Shigella (1:2,000; ab65282 Abcam). Cells were then washed three times with DPBS and permeabilized with 0.1% Triton X-100 in DPBS for 10 min at RT. Permeabilization solution was washed three times with DPBS. The extracellular bacteria were labeled with goat anti-rabbit AlexaFluor 647 (1:2,000; A32733, Invitrogen), and the total bacteria were stained with DAPI (1:2,000, Fisher Scientific). Finally, samples were washed three times with DPBS and stored hydrated at 4 °C, protected from light. z-stacks of individual positions were recorded with a 0.3 µm step size.

Time-lapse image analysis

Images were analyzed using FIJI [60]. Images from the time-lapse series were corrected for the intensity decay due to photobleaching using the FIJI bleach correction plugin [61]. Vacuole damage and rupture times were manually quantified.

Shigella epithelial cell infection for cryo-electron microscopy

Cryo-EM gold grids (Quantifoil R2/2 Au 200 mesh, Quantifoil) were placed in the slots of a custom-made PDMS stencil (Alvéole) on a 35 mm dish (µ-Dish 35 mm, ibidi). The montage was plasma-cleaned for 45 s and sterilized under UV irradiation in a laminar flow hood for 15 min. Cell culture medium was added to the dish and equilibrated for 15 min at 37 °C. 85,000–100,000 HeLa cells were seeded and grown for 24 h at 37 °C, 5% CO₂. Before the infection, cells were washed three times with DMEM and incubated with bacterial suspension at a high MOI (100–400). Infection was synchronized at RT for 15 min and cells were infected at 37 °C for 10 or 20 min. Cells were washed three times with DPBS (Gibco) and fixed for 15 min in 2% PFA (Electron Microscopy Science), 0.05% Glutaraldehyde (Sigma-Aldrich) in 0.1 M HEPES (Gibco) followed by 15 min in 4% PFA, 0.1% Glutaraldehyde in 0.1 M HEPES. Fixed cells were washed three times with DPBS and immediately vitrified. Grids were blotted from the backside and plunged into liquid ethane at liquid nitrogen temperature using a Leica EM GP automatic plunger with the following settings: humidity 98%, blot time: 8 s, 20 °C. The vitrified grids were stored in sealed boxes in liquid nitrogen until further use.

Fluorescence-guided cryo-FIB milling

Grids clipped into cryo-FIB milling Autogrids (Thermo Fisher Scientific) were imaged in light microscopy using a Leica THUNDER Imager EM Cryo-CLEM (Leica Microsystems). A global map of the grid was acquired to aid later correlations and individual z-stacks with 0.35 µm spacing were recorded at the regions of interest. Cryo-FIB lamellae were prepared using an Aquilos 2 dual-beam cryo-FIB scanning electron microscope (cryo-FIB-SEM; Thermo Fisher Scientific) instrument equipped with a cryo-transfer system, a cryo-stage, and a 45° pre-tilted shuttle (Thermo Fisher Scientific). In order to locate region of interest for the lamellae milling, an SEM map of the grid was acquired and cryo-fluorescence images were loaded at the correct relative position and orientation within the SEM map using Maps 3.2 (Thermo Fisher Scientific). Before milling an organometallic platinum layer was deposited on the cells for 1 min using the gas injection system. Lamellae were automatically milled overnight to 1 µm with a reducing current (1–0.1 nA) using AutoTEM (v.2.2, Thermo Fisher Scientific) with a 10° milling angle. The next day, lamellae were manually polished to ~200 nm and stored in liquid nitrogen.

Cryo-ET sample preparation for bacterial cells

Bacteria were grown as described above and 1 mL of subculture was centrifuged at 6,010g for 2 min in a 2 mL Eppendorf tube. The bacterial pellet was washed twice in DPBS, then fixed, washed three times and resuspended in DPBS to an OD600 of 10. Immediately after resuspension, bacteria were plunge-frozen. To this end, 1 µ L of DPBS was applied on the backside of a glow-discharged copper grid (Quanifoil R2/2 Cu 200 mesh, Quanifoil), while 4 μL of the bacterial suspension was deposited on the film side and the grid was back-blotted for 4 seconds. Cryo-FIB milling was carried out as described above, without fluorescence-guided targeting.

Tilt series data collection and reconstruction

Cryo-ET datasets were collected using SerialEM software [62] (version 4.1.0 to 4.1.4) on a 300 kV cold-Field Emission Gun Titan Krios transmission electron microscope (Thermo Fisher Scientific) equipped with a Falcon 4i direct electron detection camera (Thermo Fisher Scientific) and Selectris X an energy filter (Thermo Fisher Scientific). For tilt series acquisition the microscope was set in nanoprobe mode, energy filter at 20 eV (zero loss) and an objective aperture of 100. Tilt series were acquired with a dose dose-symmetric acquisition scheme [63] ranging from +70° to −50° starting from a 10° pretilt with 3° increments at the nominal magnification of 26,000× and 42,000× corresponding to a pixel size of 3.104 and 4.81 Å respectively with defocus ranging from 3 to 5 µm. The total dose applied was 140 e⁻/Ų with an electron dose starting at 3.5 e⁻/Ų at 10° pretilt and exposure time-varying to be 1.15 times higher at high tilts. The cold-FEG was flashed before each tilt series acquisition and frames were saved in eer format. Tilt series were processed with our in-house reconstruction script. First, the frames were motion-corrected and gain-corrected using MotionCor2 [64]. 2D contrast transfer function (CTF) was calculated with ctfplotter and the correction was applied with phase-flipping using ctfphaseflip [65]. Tilt series were dose-weighted and aligned in AreTomo [66] and tomograms were reconstructed using IMOD by weighted back-projection [67] to a pixel size ~10 Å corresponding to a binning of 2 or 3 depending on the dataset and applying and “exact filter”. Finally, reconstructed tomograms were filtered using isotropic reconstruction software IsoNet [68] with the following parameters (make_mask --density_percentage 50 --std_percentage 5; refine --iterations 30 --noise_start_iter 10,15,20,25 --noise_level 0.05,0.1,0.15,0.2). Tomograms presented in the supplementary movies were denoised with Topaz-Denoise [69].

Tomogram analysis

Statistical analysis

Two-tailed unpaired t-tests and ANOVA with Tukey’s multiple comparisons tests were performed using GraphPad Prism version 10.2.2 for Windows and 10.4.1 for Mac, GraphPad Software, Boston, Massachusetts USA (www.graphpad.com). With a p-value considered significant if p < 0.05 with * < 0.05 and **** < 0.0001.

Figure preparation

All figures were prepared using Adobe Illustrator (v. 27.8.1) and some elements (bacteria) of Figs 1A, 4, and S3 were created with BioRender.com.

S1 Fig. Ipa protein secretion profiles from the bacterial strains used in this study.

SDS-PAGE showing secretion of Ipa proteins after Congo Red induction (+) or not (−). IpaA, B, C, and D migration bands are marked by asterisks. Shigella BS176 that does not carry the invasion plasmid encoding the T3SS does not secrete Ipa proteins. E. coli mT3_Shigella secretes IpaA, B, C and D. E. coli mT3_ShigellaΔeff_Inv that only encodes for T3SS, including needle tip protein (IpaD) and translocon pore proteins (IpaB and C) does not secrete the effector IpaA.

https://doi.org/10.1371/journal.pbio.3003135.s001

(TIF)

S2 Fig. Synthetic Escherichia coli strains expressing Shigella T3SS invade epithelial cells with minimal effector set or Yersinia invasin expression.

Representative microscopy images of HeLa eGFP-Lysenin mOrange-Galectin-3 cells infected with either DH10β E. coli or E. coli mT3_Shigella, for 2 h, E. coli mT3_ShigellaΔeff_Inv for 3 h or WT Shigella for 30 min. Extracellular bacteria were stained with an antibody directed against LPS and total bacteria (extra and intracellular) were labeled with DAPI. The First panel shows full field of view, inset is marked by a dashed box. E. coli DH10β did not invade HeLa cells while Shigella, E. coli mT3_Shigella and E. coli mT3_ShigellaΔeff_Inv did. eGFP-Lysenin: yellow, mOrange-Galectin-3: magenta, DAPI: cyan, and α-LPS: green. Intracellular bacteria: cyan, Extracellular bacteria: cyan and green. Scale bars are 50 µm for the large field of view and 8 µm for all the insets.

https://doi.org/10.1371/journal.pbio.3003135.s002

(TIF)

S3 Fig. Correlative cryo-ET recapitulates the successive steps of Shigella vacuole damage and rupture.

(A–C) Detailed correlation steps for the identification intact (A), damaged (B), or ruptured (C) vacuoles (shown in Fig 2) in HeLa eGFP-Lysenin mOrange-Galectin-3 cells infected with Tag-BFP Shigella processed through a correlative cryo-ET workflow. (I) Vitrified cells on cryo-EM grids were imaged by cryo-fluorescence microscopy to localize infection sites and target them for lamella milling. Tag-BFP Shigella: cyan, eGFP-Lysenin: yellow, and mOrange-Galectin-3: magenta. Scale bars: 10 µm. (II) Cryo-TEM overview of the region targeted in I after cells were thinned into lamellae using cryo-FIB-milling. Scale bars 10 µm. (III) Cryo-lamellae maps overlayed with the corresponding cryo-fluorescence images. Scale bar 5 µm. The rotation angles from images I and II to III are indicated on the top left. (IV) Inset of III showing the bacteria that was targeted for imaging. Scale bar 1 µm. The rotation angles from images III to IV are indicated on the top left. Square boxes V correspond to regions presented in the tomographic slices of Fig 2A, 2B, and 2C respectively. (D) Additional T3SS insets marked in Fig 2 damaged and ruptured panels. Outlined arrowheads point to undetermined densities in the lumen of the Shigella vacuoles. Scale bars are 50 nm. Vac.: Vacuole, OM: Outer membrane, IM: Inner membrane.

https://doi.org/10.1371/journal.pbio.3003135.s003

(TIF)

S4 Fig. Differential coating of Shigella vacuole upon loss of membrane integrity in cells constitutively expressing reporters of membrane damage and rupture.

Slice through tomograms (scale bars 200 nm) and corresponding cryo-fLM images (scale bars 2 µm) of Shigella (white arrowhead) at different infection stages with vacuole membranes showing different coating patterns that may reflect on recruitment of overexpressed Lysenin (yellow arrowhead) and Galectin-3 (magenta arrowhead). Intact vacuoles: membrane coating is never observed. Damaged vacuoles: cytosolic side of the vacuole membrane uniformly coated with electron-dense layers upon Lysenin recruitment. Ruptured vacuoles: both on the cytosolic and luminal side of Lysenin and Galectin-3 double-positive vacuoles are coated while only the luminal side of the vacuole membrane is coated in cells expressing just the Galectin-3 marker.

https://doi.org/10.1371/journal.pbio.3003135.s004

(TIF)

S5 Fig. T3SS complexes are reliably found in T3SS-expressing bacterial strains.

Cryo-tomograms were acquired from cryo-lamellae of various bacterial strains, with T3SS-like complexes identified exclusively in the tomograms of T3SS-encoding strains. Tomographic slices are shown (scale bars 200 nm) and when applicable T3SS regions are marked by a dashed box and corresponding insets shown (scale bars 50 nm). Z slices of tomograms and insets are indicated below the images. Vac.: Vacuole, OM: Outer membrane, IM: Inner membrane. (A) Shigella flexneri BS1276 (pWR100 deficient strain). No T3SSs were seen in tomograms. (B) WT Shigella flexneri M90T carrying the virulence plasmid pWR100, encoding for the T3SS structural component and most of the Shigella secreted effectors. Note that inset n°1 shows a gray area as the T3SS is on the edge of the tomogram. (C) E. coli mT3_Shigella, engineered E. coli strain expressing Shigella T3SS structural components including the translocon pore (IpaB, IpaC) and secreted effectors (IpaA, IcsB, IpgD).

https://doi.org/10.1371/journal.pbio.3003135.s005

(TIF)

S6 Fig. T3SS assemblies along Shigella infection stages.

(A) Diversity of T3SS architectures might reflect different assembly states. Contingency graph of the percentage of T3SSs with a membrane-spanning basal body and protruding needle (total n = 122) or with only a basal body without needles (total n = 25) plotted according to the infection stage. Representative examples of putative T3SSs assembly states are shown. Scale bars are 50 nm. The data underlying S6 Fig can be found in S4 Data. (B) T3SS needles are exposed to the cytosol after vacuole rupture and disassembly. Cryo-fLM, (scale bars 2 µm) and corresponding tomogram slice (scale bar 200 nm) of Shigella with ruptured vacuole displaying T3SSs with needles exposed to the cytosol after vacuole rupture (insets, scale bars 50 nm). OM: Outer membrane, IM: Inner membrane.

https://doi.org/10.1371/journal.pbio.3003135.s006

(TIF)

S7 Fig. T3SS-vacuole contact depends on both T3SS length and available luminal space.

(A–D) Tomogram slices of the T3SS-vacuole zones before Shigella membrane rupture. White arrowhead points to bacterial membranes deformations. Scale bars 50 nm. Corresponding 3D rendering of vacuole and bacteria membranes with Shigella T3SS 3D map (EMD-15700) fitted. Galectin-3 negative vacuole (gray), bacteria OM (dark blue) and IM (light blue). Next panels show membrane surface reconstructions of the vacuole membrane coloured by 3D distance to the bacteria outer membrane (left) or vacuole membrane curvedness (right). Bacteria membranes are shown in light gray. Vac.: Vacuole, OM: Outer membrane, IM: Inner membrane. The data underlying S7 Fig can be found at https://doi.org/10.5281/zenodo.15065516. (A and B) T3SSs do not establish contact with the vacuole if they are short (A) or if the vacuole is locally relaxed (B). (C) T3SS with needle length correlating with the available vacuole space contact the vacuole without deforming it. (D) T3SS with long needle contacting the vacuole membrane adopts tilted insertion within bacterial membranes. (E) Quantification of the vacuole membrane curvedness (nm⁻¹) as a function of the distance from the T3SS needle tip (exemplified in F), depending on whether or not T3SS establishes contact with the vacuole (T3SS tips with a distance smaller than 5 nm were considered to be in contact). Bars show the 5%–95% percentile range. (F) Annotated graphical representation showing curvedness values and their distances to the T3SS needle tip in bins of width 10 nm, as used for the quantification in (E).

https://doi.org/10.1371/journal.pbio.3003135.s007

(TIF)

S1 Video. Time-lapse microscopy monitoring of Lysenin and Galectin-3 recruitment at the Shigella vacuole, related to Fig 1B.

Shigella infection of HeLa eGFP-Lysenin mOrange-Galectin-3 cells. Vacuolar damage characterized by Lysenin recruitment at the vacuole (merge: yellow, grayscale: second panel) is observed shortly before vacuolar rupture (merge: magenta, grayscale: third panel) indicated by the recruitment of Galectin-3 at the vacuole and is followed by vacuole disassembly. Images were taken every minute and a maximum intensity z-projection is presented. Scale bar 5 µm.

https://doi.org/10.1371/journal.pbio.3003135.s009

(MP4)

S2 Video. Time-lapse microscopy monitoring of Lysenin and Galectin-3 recruitment at the Escherichia coli mT3_Shigella vacuole, related to Fig 1D.

E. coli mT3_Shigella infection of HeLa eGFP-Lysenin mOrange-Galectin-3 cells. Vacuolar damage characterized by Lysenin recruitment at the vacuole (merge: yellow, grayscale: second panel) is observed before vacuolar rupture (merge: magenta, grayscale: third panel) indicated by the recruitment of Galectin-3 at the vacuole that does not disassemble. Images were taken every minute and a maximum intensity z-projection is presented. Scale bar 5 µm.

https://doi.org/10.1371/journal.pbio.3003135.s010

(MP4)

S3 Video. Time-lapse microscopy monitoring of Lysenin and Galectin-3 recruitment at the Escherichia coli E. coli mT3_ShigellaΔeff_Inv vacuole, related to Fig 1F.

E. coli mT3_ShigellaΔeff_Inv infection of HeLa eGFP-Lysenin mOrange-Galectin-3 cells. Vacuolar damage characterized by Lysenin recruitment at the vacuole (merge: yellow, grayscale: second panel) is observed very shortly before vacuolar rupture (merge: magenta, grayscale: third panel) indicated by the recruitment of Galectin-3 at the vacuole that does not disassemble. Images were taken every minute and a maximum intensity z-projection is presented. Scale bar 5 µm.

https://doi.org/10.1371/journal.pbio.3003135.s011

(MP4)

S4 Video. Tomogram and corresponding segmentations of the distinct stages of Shigella cytosolic access, related to Fig 2A.

Sequential slices through the tomogram shown in Fig 2A and segmentation showing Shigella entrapped into an intact vacuole. Shigella OM: Outer membrane (dark blue), IM: Inner membrane (light blue). Intact vacuole (gray. Scale bar 200 nm.

https://doi.org/10.1371/journal.pbio.3003135.s012

(MOV)

S5 Video. Tomogram and corresponding segmentations of Shigella in a damaged vacuole, related to Fig 2B.

Sequential slices through the tomogram shown in Fig 2B and segmentation showing Shigella entrapped either into a damaged vacuole. Shigella OM: Outer membrane (dark blue), IM: Inner membrane (light blue). Damaged vacuole (yellow). Scale bar 200 nm.

https://doi.org/10.1371/journal.pbio.3003135.s013

(MOV)

S6 Video. Tomogram and corresponding segmentations of Shigella cytosolic access, related to Fig 2C.

Sequential slices through the tomogram shown in Fig 2C and segmentation showing Shigella entrapped either into an intact or damaged vacuole or surrounded by ruptured vacuole remnants. Shigella OM: Outer membrane (dark blue), IM: Inner membrane (light blue). Ruptured vacuole (magenta). Scale bar 200 nm.

https://doi.org/10.1371/journal.pbio.3003135.s014

(MOV)

S7 Video. Tomogram and corresponding segmentations of Shigella prior cytosolic access, related to S8A Fig.

Sequential slices through the tomogram of S8A Fig with corresponding segmentations showing Shigella in a Galectin-3 negative vacuole. Shigella OM: Outer membrane (dark blue), IM: Inner membrane (light blue), Galectin-3 negative vacuole (gray). Scale bar 200 nm.

https://doi.org/10.1371/journal.pbio.3003135.s015

(MOV)

S8 Video. Tomogram and corresponding segmentations of Shigella prior cytosolic access, related to S8B Fig.

Sequential slices through the tomogram of S8B Fig with corresponding segmentations showing Shigella in a Galectin-3 negative vacuoles. Shigella OM: Outer membrane (dark blue), IM: Inner membrane (light blue), Galectin-3 negative vacuole (gray). Scale bar 200 nm.

https://doi.org/10.1371/journal.pbio.3003135.s016

(MOV)

S9 Video. Tomogram and corresponding segmentation of Shigella prior cytosolic access, related to S8C Fig.

Sequential slices through the tomogram of S8C Fig with corresponding segmentations showing Shigella in a damaged vacuole. Shigella OM: Outer membrane (dark blue), IM: Inner membrane (light blue), damaged vacuole (yellow). Scale bars are 200 nm.

https://doi.org/10.1371/journal.pbio.3003135.s017

(MOV)