CagT is essential for OMC structural stability and non-essential for PR stability

First, we sought to determine the structural organization of the OMCC in the absence of the CagT. CagT is a lipoprotein that has limited sequence similarity to VirB7 homologs, but it shares structural relatedness to X. citri VirB7 within the N-terminal region (Fischer, 2011; Backert et al, 2017; Sgro et al, 2018; Chung et al, 2019; McClain et al, 2020). Two copies of CagT are present in each ASU of the OMC (Sheedlo et al, 2020) (Fig 1A and B). Both copies are localized in the outer layer of this subassembly, and the cysteine residue predicted to be lipidated is positioned to interact with the bacterial outer membrane (Chung et al, 2019; Sheedlo et al, 2020). Studies of a mutant strain engineered to produce CagT with an altered lipobox indicated that CagT lipidation is essential for CagT stability and Cag T4SS activity (McClain et al, 2020). A ∆cagT deletion mutant is defective in both the translocation of CagA into gastric epithelial cells and the induction of IL-8 in AGS gastric epithelial cells (Fischer et al, 2001; Johnson et al, 2014; Frick-Cheng et al, 2016), indicating that CagT is required for Cag T4SS function. Immunoblot analysis of Cag T4SS complexes isolated from the ∆cagT mutant showed that the preparations did not contain CagT or Cag3, but CagY, CagX, and CagM were still detected (Frick-Cheng et al, 2016). Cryo-ET analysis of OMCCs in the ∆cagT bacteria lacked peripheral density, and fewer T4SS complexes were detected in the ∆cagT mutant than in WT bacteria (Hu et al, 2019). Negative stain EM analysis of Cag T4SS complexes purified from the ∆cagT mutant showed no structured OMCCs but did show some thin rings with a ∼19-nm diameter (Frick-Cheng et al, 2016). However, the lack of a high-resolution OMCC structure available at the time of the previous work limited any detailed conclusions that could be made about the role of CagT in the overall organization of OMCC, other than concluding that it was important for complex stability. Now, with the availability of a 3.4 Å resolution structure and the molecular model of the WT Cag T4SS OMCC (Fig 1) (Chung et al, 2019; Sheedlo et al, 2020), we decided to more closely examine how CagT, a conserved T4SS component, contributes to the overall structural organization of the OMCC.

Using mass spectrometry, negative stain EM, and single-particle cryo-EM analyses, we analyzed the composition and structure of Cag T4SS complexes purified from the ∆cagT mutant. Liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis of T4SS complexes purified from the ∆cagT mutant showed that the samples contained CagX, CagY, and CagM, but lacked Cag3 and CagT (Table 1). These results concur with previous immunoblot results (Frick-Cheng et al, 2016). As before, negative stain EM showed that complexes isolated from the ∆cagT mutant differed markedly from WT OMCCs (Fig 2A), with only ∼19-nm-wide thin rings visible in the images of T4SS complexes purified from the ∆cagT mutant (Fig 2B). However, now with a better understanding of the overall Cag T4SS OMCC structure, we recognized that these rings have a size and appearance similar to the PR of the OMCC.

To examine this more closely, we collected a cryo-EM dataset of complexes purified from the ∆cagT mutant and used 2D and 3D single-particle analysis to characterize the complexes (Fig S1A and B, Table S1). The “en face” views of 2D classes of mutant particles contain none of the secondary structural features of the OMC that usually dominate this view of WT OMCCs (Fig 2C and D), indicating that this part of the complex is no longer stably organized in the absence of CagT. However, the averages did show the presence of a structured ring, ∼19 nm in diameter with 17-fold symmetry (Fig 2E). This is the same width and symmetry as the PR in WT OMCCs (Fig 1A). Thus, 2D analysis of the cryo-EM images indicates that T4SS complexes lacking CagT no longer have a structured OMC but appear to retain an organized PR.

3D reconstructions of complexes purified from the ∆cagT mutant confirmed the 2D analysis. Although there is amorphous density in the 3D reconstruction that can be attributed to the OMC, the density is extremely noisy with no apparent secondary structural features or symmetry (Fig S1). Although the mass spectrometry analysis detected CagM, CagY, and CagX in the samples, including CagY and CagX peptides from regions localized in the OMC, neither the 2D averages nor the 3D reconstruction showed any defined structural features that can be attributed to these proteins in the OMC. However, the ∆cagT complexes still have a structurally defined PR. A 3D structure of the PR with 17-fold symmetry applied reached 8.0 Å resolution (Fig S2A–C) and had secondary structural features that made it possible to place molecular models for CagY (residues 1,469–1,603) and CagX (residues 32–130, 261–323) built from the WT PR EM density map into the ∆cagT PR density map (Fig 2F–H). Although there are some minor differences between the high-resolution WT PR model and the ∆cagT PR map (Fig S3A), likely because of the heterogeneity in the ∆cagT particles, the overall structural integrity of the PR is preserved.

Overall, these analyses show that CagT is required for the stable association of Cag3 with the OMCC and is essential for the structural organization of CagM, CagX, and CagY within the OMC. However, even in the absence of an organized OMC, the overall structural organization of the PR is maintained.

Loss of the CagY AP region does not alter the structural organization of the OMCC

The importance of CagT in the OMC structure might be explained solely by the central position of the 28 copies of CagT found in each OMC (Fig 1) (Sheedlo et al, 2020). However, CagT is also a lipoprotein, with the position of post-translational lipid moieties likely providing a mechanism to anchor the OMC to the outer membrane (McClain et al, 2020; Sheedlo et al, 2020). Thus, CagT might contribute to the OMC structure through its interactions with the outer membrane. Another Cag T4SS protein that contacts the outer membrane is CagY (Chung et al, 2019; Sheedlo et al, 2020). The CagY AP motif is highly conserved among VirB10 family members, is predicted to form a channel in the outer membrane, and is required for H. pylori Cag T4SS activity (Tran et al, 2023). Specifically, an H. pylori mutant lacking the CagY AP (residues 1,793–1,863) is defective in IL-8 induction and CagA translocation (Tran et al, 2023). VirB10 AP regions are predicted to form a channel through the outer membrane, and cryo-ET analysis of the H. pylori T4SS showed that the OMCC region containing the CagY AP is localized adjacent to the outer membrane (Chang et al, 2018; Hu et al, 2019). If OMC interactions with the outer membrane contribute to OMC assembly or overall stability, the CagY AP motif would be predicted to be important for OMC organization. To test this hypothesis, we investigated whether the presence or absence of the CagY AP region influences the overall structure of the OMCC.

To examine the structural contribution of the conserved AP region of CagY to OMCC organization, we used both negative stain EM and single-particle cryo-EM analyses to investigate the structural organization of Cag T4SS OMCCs purified from a CagY∆AP mutant strain. Negative stain analysis showed that OMCCs from the CagY∆AP mutant globally resembled WT OMCCs (Figs 2A and 3A). To determine whether there were more nuanced changes in the architecture of the Cag T4SS not evident by low-resolution negative stain imaging, we then examined these OMCCs using cryo-EM analysis (Fig S4A and B, Table S1). 2D analysis of particles in vitrified ice showed that the “en face” views of the CagY∆AP OMCCs contained clear secondary structural elements and were similar in diameter to WT OMCCs (Figs 2C and 3B); however, the innermost central density, corresponding to where 14 copies of the CagY AP region form an α-helical channel, is missing in the mutant complexes (Fig 3B). 2D classes representing side views of the mutant complexes showed pairs of CagY∆AP OMCCs interacting where the channel is usually located (Fig 3B). This interaction between OMCCs has not been observed in analyses of WT or ∆cag3 OMCCs (Chung et al, 2019; Sheedlo et al, 2020). 3D single-particle cryo-EM analysis of OMCCs purified from the CagY∆AP mutant showed that although the complexes clearly lack the AP extension, the maps of the OMC and the PR (at the resolutions of 3.8 and 6.6 Å, respectively) (Fig S5A–F) are very similar to the corresponding maps of the WT OMC and PR (Figs 3C–H and S3B and C). CagY in the CagY∆AP OMC, other than missing the helix–loop–helix regions (corresponding to the deleted amino acids 1,793–1,863), adopts the same conformation as CagY in the WT OMC (Fig 3C–H). The CagY∆AP PR also has the same overall organization as the WT PR, although there are some minor differences that cannot be directly modeled at the current resolution of the map (Fig S3C). Thus, although the CagY AP region is required for the biological function of the H. pylori Cag T4SS (Tran et al, 2023) and likely interacts with the OM (Chang et al, 2018; Hu et al, 2019), it is not required for the overall structural stability and organization of the OMCC.

CagM, an H. pylori-specific T4SS component, is required for a structured OMC but not PR stability

We next characterized the role of the H. pylori-specific T4SS protein CagM in the structural organization of the OMCC. CagM does not have obvious homologs in any non-H. pylori species (Sheedlo et al, 2020). As with CagT, there are two copies of CagM found in each ASU of the OMC (Fig 1A); however, unlike CagT, both copies of CagM are found in the inner layer, rather than the outer layer, of this subassembly (Fig 1) (Chung et al, 2019; Sheedlo et al, 2020). CagM is essential for Cag T4SS function, because a ∆cagM mutant is defective in the translocation of CagA and induction of IL-8 in AGS gastric cells (Fischer et al, 2001; Johnson et al, 2014; Frick-Cheng et al, 2016). Based on immunoblot analysis, OMCCs purified from the ∆cagM mutant only contain CagX and CagY, with CagM, CagT, and Cag3 not detected (Frick-Cheng et al, 2016). To carefully examine the role of this species-specific T4SS protein in the structural organization of the Cag T4SS OMCC, we purified T4SS complexes from the ∆cagM mutant and examined their composition and structure using mass spectrometry, negative stain EM, and single-particle cryo-EM analyses.

LC-MS/MS analysis of complexes purified from the ∆cagM mutant showed that the preparations contained CagX and CagY, but lacked CagT, CagM, and Cag3 (Table 1), matching the previous immunoblot analysis (Frick-Cheng et al, 2016). Examination of the mutant particles by negative stain revealed thin rings similar to complexes purified from the ∆cagT mutant (Figs 2B and 4A). We speculate that the ability to visualize these structures in the current study but not in a previous study (Frick-Cheng et al, 2016) is attributable to the use of larger volume bacterial cultures and improved purification strategies.

We collected a cryo-EM dataset and analyzed the resulting particles using 2D and 3D single-particle cryo-EM approaches (Fig S6A and B, Table S1). In the 2D averages, the ∆cagM complexes have no structured OMC (Fig 4B), in contrast to the well-defined OMC visible in WT OMCCs (Fig 2C); however, there is density for the PR with clear secondary structural features and 17-fold symmetry (Fig 4B and C). 3D reconstruction of these particles confirmed the observations based on analysis of the 2D averages and showed there was no structured density for the OMC (Fig S6A). However, the ∆cagM complexes still have a structurally defined PR. The 3D reconstruction of the PR from the ∆cagM mutant reached 8.5 Å with 17-fold symmetry imposed (Fig S7A–C), and secondary structural features expected at this resolution were visible in the map (Fig 4D–F). We can place molecular models of the regions of CagY (residues 1,469–1,603) and CagX (residues 32–130, 261–323) found in the PR directly into this density (Fig 4F), showing that these parts of CagX and CagY remain structured in the ∆cagM background. Although there are some subtle differences between the 3D structure of the ∆cagM PR and WT PR (Fig S3D), this analysis shows that the ∆cagM PR, even without a structured OMC, retains its overall structural organization. Therefore, these results indicate that CagM is required for CagT and Cag3 association with the OMCC, is required for the structural organization of the OMC, and is not required for maintaining the structural organization of the PR.

Bacterial strains, plasmids, and cell culture

H. pylori strains were cultured on trypticase soy agar plates containing 5% sheep blood in ambient air supplemented with 5% CO₂. Liquid cultures of H. pylori were grown in Brucella broth supplemented with 10% vol/vol heat-inactivated FBS. CagY∆AP (deletion of CagY amino acids 1,793–1,863), unmarked ∆cagT, and unmarked ∆cagM mutant strains, each containing sequences encoding HA-CagF, have been described previously (Frick-Cheng et al, 2016; Tran et al, 2023).

T4SS OMCC purification

OMCCs were purified from mutant H. pylori strains using a purification method targeting HA-CagF, as described previously (Frick-Cheng et al, 2016; Chung et al, 2019). In brief, strains were grown in liquid culture for about 20 h, the bacteria were pelleted at 3,300g for 15 min at 4°C, and the pellet was resuspended in RIPA buffer (50 mM Hepes, 100 mM NaCl, 1% NP-40, and 0.025% deoxycholate supplemented with 1 mM phenylmethylsulfonyl fluoride and protease inhibitors [Roche]) and sonicated on ice (at 25% amp, 10 s on and 10 s off, five times). The suspension was then incubated for 1 h at 4°C. The insoluble material was pelleted, the bacterial lysate was incubated with anti-HA antibodies non-covalently linked to protein G Dynabeads (Invitrogen) for 30 min, and then the complexes were eluted with HA peptide for 1 h at RT.

LC-MS/MS analysis of Cag T4SS OMCCs

To analyze the protein content of the immunopurified samples, samples were digested and prepared for analysis using S-traps (Protifi), following the manufacturer’s recommended protocol. The resulting tryptic peptides were analyzed by data-dependent LC-MS/MS (McClain et al, 2023). Briefly, peptides were autosampled onto a 200 mm by 0.1 mm (Jupiter 3 micron, 300A) self-packed analytical column coupled directly to an LTQ linear ion trap mass spectrometer (Thermo Fisher Scientific) using a nanoelectrospray source, and resolved using an aqueous-to-organic gradient. Both the intact masses (MS) and fragmentation patterns (MS/MS) of the peptides were collected in a data-dependent manner using dynamic exclusion to maximize the depth of coverage. Resulting peptide MS/MS spectral data were searched against an H. pylori database to which sequences of common contaminants and reversed versions of each protein had been added, using SEQUEST. Peptide spectral matches were collated, filtered, and compared using Scaffold (Proteome Software). Protein identifications required a minimum of two unique peptides per protein and were filtered to a 1% false discovery rate for peptides and a 5% false discovery rate for proteins.

Negative stain EM sample preparation and data collection

Negative stain EM was carried out using established methods (Ohi et al, 2004). 400-mesh copper grids covered with carbon-coated collodion film (Electron Microscopy Sciences) were glow-discharged for 30 s at 5 mA in a PELCO easiGlow glow discharge unit (Ted Pella). 3.5 μl of the Cag T4SS sample (as purified) was adsorbed to the grids and incubated for 1 min at RT. The grids were then washed twice with water, negatively stained with 0.75% (wt/vol) uranyl formate solution, and blotted until dry. Negative stain images were taken using a Tecnai Spirit T12 transmission electron microscope (Thermo Fisher Scientific) operated at 120 kV and at a nominal magnification of 26,000x (2.3 Å/pixel). Images were acquired with Leginon (Suloway et al, 2005) on a 4K × 4K Rio complementary metal-oxide semiconductor camera (Gatan) at −1.5-μm defocus value.

Cryo-EM sample preparation

Cryo-EM samples were prepared as described previously (Chung et al, 2019; Sheedlo et al, 2020). In brief, 3.5 μl of the Cag T4SS OMCC sample (as purified) was applied to a glow-discharged Quantifoil R 2/2 UT 200-mesh copper grid (Quantifoil). The sample was applied to a grid three to four times, incubated for 60 s each before blotting, and vitrified by plunge freezing in a slurry of liquid ethane using Vitrobot (Thermo Fisher Scientific) at 20°C and 100% humidity.

Cryo-EM data collection

Images were collected on a Titan Krios electron microscope (Thermo Fisher Scientific) equipped with K3 Summit Direct Electron Detector (Gatan) operated at 300 kV with a nominal pixel size of 1.08 Å per pixel. The Bioquantum energy filter (Gatan) was inserted with a slit width of 20 eV. Micrographs were acquired using SerialEM software (Mastronarde, 2005). The electron dose totaled 60 e/Ų, and the defocus range was −1 to −2.5 μm.

Cryo-EM data analysis

cryoSPARC v.4.2.1 was used for image processing of all cryo-EM datasets (Punjani et al, 2017). Fig S1 shows data processing steps for cryo-EM analysis of T4SS complexes purified from the ∆cagT mutant. For analysis of the ∆cagT OMCC, 310,039 particles were selected by template picking in cryoSPARC. After iterative 2D classification, classes with clear secondary structural features were retained, corresponding to 8,415 particles. These particles were used in a reference-free initial 3D reconstruction (ab initio model in cryoSPARC) designating two 3D classes, no applied symmetry (C1), and a resolution range of 35–12 Å (the default cryoSPARC setting). The 3D classes resembled previous PR structures and did not have density resembling the OMC. In an effort to visualize any structured OMC density, one of the ab initio 3D volumes was used as the initial model in a non-uniform refinement that used 8,415 particles and imposed C14 or C17 symmetry. Both reconstructions contained artifacts and lacked any secondary structural features associated with either the OMC or the PR. Because there were 2D averages with 17-fold symmetry, we did another ab initio reconstruction applying C17 symmetry, designating two 3D classes, and choosing a resolution range of 20–4 Å. The best class contained 5,706 particles. These particles were used in a subsequent local refinement, using the ab initio model as the initial volume, and a solvent mask that was created from the ab initio model that was padded and dilated 10 pixels each. The rotation search extent was 10°. After local refinement, the final resolution was 8 Å (Fig S2).

Fig S4 shows data processing steps for cryo-EM analysis of OMCCs purified from the CagY∆AP mutant. For analysis of the CagY∆AP OMCC, 633,415 particles were selected by template picking in cryoSPARC. After iterative 2D classification, classes with clear secondary structural features were retained, containing a combined 22,643 particles. These particles were subjected to an ab initio 3D reconstruction designating one 3D class not applying symmetry, and using a resolution range of 35–12 Å. This model was used as a reference for a non-uniform 3D refinement with C1 symmetry, resulting in a density map with a global resolution of 6.8 Å. For the focused refinement of the CagY∆AP OMC, the C1 ab initio 3D structure was used as the initial model for a non-uniform refinement imposing C14 symmetry. This resulting 3D volume was used to create soft masks for both particle subtraction and local refinement. After particle subtraction and refinement, a final local refinement was performed, which yielded a map with a final resolution of 3.8 Å. For the focused refinement of the CagY∆AP PR, the C1 ab initio model was used as the initial model for a non-uniform refinement imposing C17 symmetry. This resulting volume was used to create soft masks for both particle subtraction and local refinement. After particle subtraction and refinement, a final local refinement was performed, which yielded a final resolution of 6.6 Å (Fig S5).

Fig S6 shows data processing steps for cryo-EM analysis of T4SS complexes purified from the ∆cagM mutant. For analysis of the ∆cagM OMCC, 78,268 particles were selected by template picking in cryoSPARC. After iterative 2D classification, 2D classes with clear secondary structural features were retained, containing a combined 4,251 particles. These particles were subjected to a reference-free initial 3D reconstruction (ab initio model in cryoSPARC) designating two 3D classes, no applied symmetry, and a resolution range of 35–12 Å. The best 3D class contained 3,009 particles and had features with 17-fold symmetry. To attempt to visualize any OMC density in this 3D map, the volume was used in a non-uniform refinement imposing C14 symmetry. However, the C14 reconstruction contained artifacts and lacked any secondary structural features. To determine a higher resolution structure of the ∆cagM PR, the best volume was used as an initial 3D model for non-uniform refinement with C17 symmetry. This map was used to create a soft mask for local refinement. Local refinement with C17 imposed symmetry resulted in a map with 4.1 Å resolution calculated for the “tight mask” at FSC = 0.143; however, this resolution does not match the secondary features present in the map or the resolutions shown in the local resolution map (Fig S7). Taking these observations into consideration, the more accurate resolution estimation appears to be 8.5 Å as calculated by cryoSPARC when using the criterion of a “loose mask” at FSC = 0.5. This resolution matches the features seen in this map and agrees with the local resolution calculations.

Map-to-model cross-correlation values for the ASU were calculated using Phenix (Liebschner et al, 2019). Map–map correlations between mutant and WT maps were done in chimera using the “measure correlation” command. ChimeraX was used to make figures of maps and models (Pettersen et al, 2021).