Biogenesis of the membrane complex employs three Acinetobacter-specific envelope-associated proteins

The presence of specific A. baumannii T6SS components and the absence of TssJ, prompted us to investigate their role in the functioning and assembly of the trans-envelope spanning complex. Bioinformatic analysis revealed five genes encode for membrane-associated proteins (Fig 1A, green stars). Four are A. baumannii-specific proteins and two of them (TslA/AsaA and TagX/AsaE) have already been characterized (see Introduction). The protein encoded by the A1S_1301 locus has an unknow function (Fig 1A, green stars), but bioinformatic analysis suggested a putative implication in phospholipid biogenesis due to its structural homology with β-ketoacyl synthase. We renamed the protein encoded by the A1S_1301 (asaB) gene TsmK (Type six secretion membrane complex assembly pseudo-Ketoacyl synthase). To test the importance of this protein in T6SS functioning, we deleted the corresponding gene in the A. baumannii ATCC 17978 strain and performed phenotypic assessment of the T6SS activity. For comparison, we have also designed mutant strains that are deleted for genes previously reported to be essential to A. baumannii T6SS functioning. As previously described, deletion of tagN had no impact on T6SS functioning demonstrated by the T6SS-dependent secretion of the Hcp protein (Fig 1B) and the killing of a prey E. coli strain (Fig 1C), which were both at levels comparable with the parental A. baumannii ATCC 17978 strain. As observed by Ringel et al. [24], deletion of tslA (asaA) had an intermediate effect and did not abolish the Hcp secretion and the T6SS-dependent killing of E. coli (Figs 1B, 1C and S1A). Interestingly, the deletion of tsmK completely abolished T6SS functioning (Figs 1B, 1C and S1A) to the same extent as the deletion of tssM or tagX (asaE), both encoding essential components of the T6SS [7,19]. The defect in Hcp secretion of the tsmK deletion mutant was complemented to WT level by a plasmid with an inducible copy of the gene, indicating no polar effect of the deletion (S1B Fig). To investigate the involvement of the three proteins TslA, TsmK and TagX in the biogenesis of the A. baumannii T6SS MC, we needed to monitor its assembly using a visual and quantitative assay. The A. baumannii MC has never been studied nor observed directly. In order to monitor the assembly and dynamic of the MC, we generated different fluorescent chimera with conserved T6SS components (S2A and S2B Fig). Only the chromosomally expressed TssM^sfGFP was functional and able to assemble readily detectable foci using live fluorescence microscopy (Fig 2A). The fluorescent foci were detected as local maxima by MicrobeJ software and localized relatively to the medial axis of the cell. The position along the medial axis was normalized to [-1, +1] and the distribution presented as histograms (Fig 2A, right panel). Analysis of foci distribution showed that the A. baumannii MC was randomly distributed around the cell periphery with a preferential accumulation at the bacterial poles (n > 1000 cells, two biological replicates). Using a different fluorescent fusion (mCherry) did not alter the pattern of distribution of TssM (S2C Fig, left panel). As expected, mutations affecting conserved essential T6SS components of the tail (TssB-C or ClpV) did not alter this localization pattern (S2C Fig, right panel). To investigate the involvement of the three proteins TslA, TsmK and TagX in the biogenesis of the A. baumannii T6SS MC, we generated in-frame deletions of the different genes in the recombinant A. baumannii ATCC 17978::TssM^sfGFP strain. Live fluorescent recording and quantification of MC foci showed that the parental strain displayed an average of 1.0 foci per cell (1.01 ± 0.09) (Figs 2B and S2D). However, a severe loss of MC assembly was observed in the different mutants (Figs 2B and S2D). In the tslA and tsmK mutants we counted an average 0.4 (0.37 ± 0.06) and 0.5 foci (0.51 ± 0.04) per cell, respectively. Similarly, in the tagX mutant we observed an average of 0.4 (0.40 ± 0.01) foci per cell. Overall, the decreased number of average foci per cell in the mutants can be interpreted by an impaired stability of the MC. Control experiments showed that TssM production was identical in all the various mutants (S2E Fig), and that their deletion did not trigger any polar effects on the expression of tssM. In conclusion, TslA, TagX, and the uncharacterized TsmK proteins are required for both functioning of the T6SS in A. baumannii and for the assembly of the trans-envelope MC.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. A. baumannii T6SS essential for the MC biogenesis.

(A) Fluorescent cluster of the TssM^sfGFP protein in A. baumannii tssM-sfGFP strain. On the left panel, the images represent the average over a time lapse acquisition of 61 frames (10s / frame). scale = 2 μm. The right panel represents the histograms of cluster distribution of the main axis of the cell with a preferential accumulation at the poles of the cell. (n > 1000 cells, two biological replicates). Comparison of the cellular distribution of TssM-sfGFP foci between the WT and tsmK mutant (right panels). (B) Statistical analysis of the amount of fluorescence foci in the different mutants. On the left, a graphical representation of the percentage of cells with zero (n = 0), one (n = 1), two (n = 2), or more than three (n = 3) foci in the different A. baumannii mutants. On the right, a graphical representation of the average number of foci per cell in the different mutants. The experiment was performed in triplicate on three different groups of cells. (C-E) Dynamic study of the fluorescence foci of the TssM-sfGFP in WT and tsmK mutant. Fluorescence microscopy in TIRF mode captured an average of 61 images every 10 seconds, revealing multiple fluorescent foci within the cells. Cell masks were generated using the Cellpose cyto2 model, and FIJI’s MicrobeJ plugin, set in "rod-shape" mode. The lower panel shows the average of images processed with a bandpass FFT filter and background subtraction that enhances the foci contrast. (D) Kymographs derived from intensity profiles measured along the cell contour, representing time (61 images every 10 seconds) on the vertical axis and cell contour length in microns on the horizontal axis. Traces of foci movement along the cell contour are evident, with fixed and stable foci yielding vertical traces, some corresponding to the whole acquisition duration (the height of the kymograph = 610 seconds). (E). Violin plots for the quantitative analysis of foci traces between the WT and tsmK mutant (3 biological replications). The difference in means was tested by t.test (R software) and yields a p value < 0.0001. (F) Biogenesis of the A. baumannii-T6SS MC and the original degradation of the tail after contraction. Time-lapse fluorescence microscopy recordings showing localization and dynamics of the ^mCherryClpV and TssM^sfGFP fusions proteins. Individual images were taken every 40 sec. The positions of the foci are indicated by an asterisk. The scale bars are 1 μm. The lines (from top to bottom) represent the phase contrast, the mCherry channel, the sfGFP channel and the superposition of the two channels. Below, a schematic representation of the sequential biogenesis of the T6SS membrane complex.

https://doi.org/10.1371/journal.ppat.1011687.g002

Since TsmK is a newly described T6SS component, we further characterized its role in MC biogenesis. Even though the tsmK mutant exhibited decreased number of TssM^sfGFP foci, their overall cellular distribution remained unchanged (Fig 2A). Interestingly, the dynamic pattern of MC assembly is dramatically altered in the tsmK mutant. We recorded traces of foci movement along the cell contour, with fixed and stable foci yielding vertical traces, some corresponding to the whole acquisition duration (610 seconds) (Fig 2C and 2D). A quantitative analysis of foci traces revealed significant differences between the WT and tsmK mutant: WT exhibited larger traces (108 pixels, sd = 113) compared to the mutant (17.2 px, sd = 48.7), along with a higher number of traces per cell (13.7, sd = 8.7) versus (3.22, sd = 3.9) in the tsmK mutant (WT n = 2850, tsmK n = 3189, for 3 biological replications) (Fig 2E). We thus concluded that TsmK is important for the stability and dynamic of the TssM-containing membrane complex (MC).

The MC is preferentially, but not exclusively localized to the cell pole. We wondered if other T6SS components are enriched at this specific location. Fluorescent labelling of sheath subunits has been often used in other bacteria to monitor T6SS dynamic. However, it is not possible to genetically engineer TssB^sfGFP fusion at the chromosomal T6SS locus in A. baumannii due to the translational coupling of both tssB and tssC genes. However, the dynamic of the recycling ATPase ClpV was previously reported as a proxy to monitor the sheath dynamic, i.e. assembly, contraction and disassembly [5,25]. We constructed the dually labelled strain A. baumannii::TssM^sfGFP::^mCherryClpV and observed single cell behavior using live fluorescence microscopy. TssM^sfGFP forms static foci that are unstable over time, appearing and disappearing during the observation. The ^mCherryClpV signal is highly evolutive and dynamic, forming diffuse signal first (panel 0s), followed by long rod-shaped signal (panel 40s), then short rod-shaped signal (panel 80s) and ultimately evolves as diffuse signal (panel 120s) (Figs 2F and S2F). We concluded that ClpV follows the same behavior as the T6SS sheath. The ^mCherryClpV signal (long rod-shaped) can be observed colocalized with that of TssM^sfGFP. We thus conclude that the polar MC foci are used for TTC dynamic assembly, consequently they are functional spots for the A. baumannii T6SS. Interestingly, after contraction of the sheath, when ^mCherryClpV behaves as short rod-shaped or diffuse signals, the TssM^sfGFP signal completely disappears. These results suggest that upon contraction of the sheath, the A. baumannii MC disassembles. We concluded that the A. baumannii T6SS MC undergoes a cycle of assembly and disassembly coordinated with the sheath dynamic.

An intricate interaction network assembles the A. baumannii membrane super-complex

We sought to determine how the canonical TssL-TssM T6SS MC components connect with the A. baumannii T6SS-specific proteins. We have thus determined the protein-protein interaction network (PPIN) by co-purification experiments using full-length membrane proteins in detergent micelles from heterologous production in E. coli (see Material & methods). Bioinformatic analysis identified two transmembrane helices in TsmK that delineates a possible large cytoplasmic domain (S3A Fig), accounting for 87% of the full-length protein. The tssL, tssM, tsmK, tagX and tslA genes were co-expressed in E. coli BL21(DE3). Constructs were designed to add StrepII, 8His, VSVG and Flag tags at the amino N-terminus of TssL (^STssL, 30.7 kDa), N-terminus of TssM (^HTssM, 138.7 kDa), carboxy C-terminus of TsmK (TsmK^F, 54.8 kDa), C-terminus of TagX (TagX^V, 24.5 kDa) and C-terminus of TslA (TslA^F, 25.4 kDa), respectively (S1 Table). Total membrane fractions were isolated and solubilized using detergents. Affinity chromatography was performed on either Strep- or His- Trap-HP column. By interacting binarily with TssM, TssL and TagX, the newly identified TsmK appeared to occupy a central position in MC PPIN (Fig 3A and 3B). We also demonstrated that TagX interacts with TssM. In addition, by pulling down on TssL^S we isolated a protein complex composed of the three proteins TssL, TssM, and TsmK (Fig 3B). We further demonstrated that TagX and TslA were recruited to the complex, producing a large four and five-proteins membrane super-complex, respectively. When comparing the purification of TssL-TssM and TssL-TssM-TsmK complexes, we observed that the level of TssL and TssM were increased in the presence of TsmK (Fig 3B). All the tested proteins (except the Strep- or His-tagged baits) do not interact nonspecifically with the affinity resins (Fig 3C). We concluded that TsmK stabilizes the complex. We thus demonstrated an intricate PPIN connecting the 4 transmembrane proteins and TslA, revealing the original composition of the A. baumannii transenvelope T6SS super MC. By interacting simultaneously with TssL, TssM and TagX, the protein TsmK appeared to be a cornerstone in the A. baumannii MC. TslA and TagX can further interact with this core-complex, leading to the assembly of an envelope-spanning complex in A. baumannii T6SS.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Determination of the protein-protein interaction network (PPIN) of A. baumannii T6SS membrane complex.

(A) Western blot probing binary interaction between TagX, TssM, TssL and TsmK respectively. (B) Co-purification of the 3-, 4- and 5-protein complexes. (A-D) Detergent-solubilized extract of E. coli BL21(DE3) cells producing the indicated protein were submitted to an affinity purification step on Strep- or His- Trap. The detergent-solubilized total membrane (L), and eluate (E) were subjected to denaturing 12.5%-polyacrylamide gel electrophoresis (PAGE) and immunodetected with the appropriate antibody. The composition of the protein mix is indicated on the top of each panel, the protein that binds to the column is underlined. Immunodetected proteins are indicated on the right. Molecular weight markers (in kDa) are indicated on the left. Tags: H, 6×His; S, Strep-tag; F, FLAG; HA, Hemagglutinin; V, VSVG. (C) Negative controls. (D) Western blot probing the binary interaction between TsmK and TssK (left panel), including negative control (TssK alone, right panel). (E) Native pull-down of TssM from A. baumannii cells. Western blot probing the identified interaction partners (left panel). Mass spectrometry analysis of TssM copurified interactants (right panel).(F) Schematic interaction network determined by co-purification and topology of the A. baumannii T6SS-MC full-length proteins. Interactions between proteins are represented with arrows: TsmK is in the center of the interaction network (red arrows). Indirect interactions (native pull-down are shown by dashed double-head arrows.

https://doi.org/10.1371/journal.ppat.1011687.g003

We then asked if the large and bulky cytoplasmic domain of TsmK could contact the baseplate (BP). To answer to this question we co-produced in E. coli both TsmK and TssK. Pulling down experiment using TsmK^STREP as a bait allowed the co-purification of TssK (Fig 3D). No TssK was retained on the STREP column in the absence of TsmK. Consequently, TsmK is able to directly interact with TssK, a central component of the BP.

To validate the PPIN observed in E. coli, we sought to confirm these results using native pull-down directly in A. baumannii. For this purpose, A. baumannii ATCC 17978 was genetically modified to introduce an in-frame DNA sequence encoding a STREPII tag to TssM-Cter (see Material & methods). The A. baumannii::TssM^STREP retains a fully functional T6SS, as evaluated by Hcp secretion. In addition, to confirm the TssM-TsmK interaction observed in E. coli, we transform the A. baumannii::TssM^STREP strain with the plasmid pVRL2-TsmK^HA. The detection of TsmK^HA is rendered possible by anti-HA western blot. Remarkably, TsmK co-purifies with TssM directly in A. baumannii (Fig 3E, left panel). We then analyzed the purified sample using mass spectrometry–based proteomics (see Material & methods). We further demonstrated that T6SS proteins belonging to the three functional sub-complexes are also co-purified by TssM in A. baumannii: TssL and TsmK (membrane complex, MC); TssK and TssF (baseplate, BP); TssB, TssC and ClpV (tail-tube complex, TTC) (Fig 3E, right panel). Together, these results enforce the presence of an entangled protein protein interaction network that build the T6SS nanomachine, which was already beginning to be unraveled with our reconstituted approach in E. coli (Fig 3F).

Interestingly, several outer membrane proteins were additionally copurified with TssM, of which OmpA (Fig 3E). Even though they could represent contaminants, however OmpA was already proposed to ensure the correct membrane localization of the T6SS in A. baumannii [23]. Our result suggests that the function of OmpA could be mediated by its interaction with TssM.

Analysis of the Acinetobacter-specific T6SS protein TsmK

Since TsmK plays a key role in the assembly of the A. baumannii T6SS MC, we sought to investigate its conservation and structural features. We conducted a sequence analysis to examine how the TsmK conservation level compares with the conservation level of TssM, TssB, and TssK in Acinetobacter and in A. baumannii only (S4A Fig). We observed that TsmK had a conservation level lower than that obtained for TssK, TssB, and TssM, thus suggesting a higher sequence heterogeneity (or alternatively sequence divergence) than that associated with other T6SS sequences. To study the presence of TsmK in Acinetobacter and other proteobacterial species, we performed a search of homologous sequences using BLASTp. We obtained 116 hits with a unique taxonomy ID (TaxID) and genome. 70% (81/116) of these hits belong to the genus Acinetobacter. Most of these sequences had a blast e-value below 1e-20 and covered at least 60% of the query sequence (S4B Fig). In S4B Fig, the red vertical and horizontal dotted lines represent the e-value (1e-20) and coverage thresholds (0.6), respectively, and are used to determine if a homolog is a TsmK sequence. On the other hand, the non-Acinetobacter hits (30%) had an e-value much higher than the threshold (1e-20) (most of them have an e-value greater than 1e-05 and a below 60% of coverage). This clear separation allowed us to classify the hits in true TsmK sequences found in Acinetobacter and similar sequences found elsewhere. We studied the co-occurrence of TsmK, TssM and TssJ using a two-step approach (S4C and S4D Fig). First, we analyzed Acinetobacter genomes containing the TsmK sequences. 96% of them contained a TssM gene, while none of them had a TssJ gene. TssJ and TssM were found in 77.1% and 97.1% of the non-Acinetobacter genomes, respectively (see Material and methods). Second, we searched for TsmK, TssM, and TssJ genes in the 1350 Acinetobacter genomes available, using TssB as an indicator of a complete T6SS operon. None of the complete 893 operons had TssJ, 90.5% contained TssM, and 90.0% contained TsmK. Importantly, 82.2% had TssM and TsmK simultaneously (S4D Fig). These two observations let us conclude that co-occurrence of TsmK and TssM, and the simultaneous absence of TssJ in Acinetobacter are statistically robust properties of A. baumannii T6SS. To investigate the function of TsmK, we computed structural models based on its sequence using three ab initio predictors AlphaFold2 [58], TrRosetta [60] and RaptorX [26]. All three models essentially agree on the global fold and arrangement of the core of TsmK (Figs 4A, 4B and S4E). The three models disagree in some parts, referred to as the inconsistent regions (Figs 4B, S4E and S4F). Essentially the inconsistent regions are: 1) the two N-terminal transmembrane helices 2) a long loop between β1 and β4, 3) the minor β-sheet, with the exclusion of β9-β10 hairpin with associated helices (Fig 4B). The inconsistent regions have a structural precision that ranges between 7 and 15 Å, and a low AlphaFold2 confidence score (Fig 4C). On the contrary, the consistent regions are the major β-sheet together with β9-β10 hairpin, with all the associated helical decoration (Fig 4C). The consistent regions have a structural precision that ranges between 0 and 7 Å, and a high AlphaFold2 confidence score (Fig 4C). Using the DALI server [27], we queried the structure of the consistent region of TsmK to find the closest structural homolog. The protein with the highest structural similarity was the Ketoacyl Synthase (KS) domain from Acyltransferase type I polyketide synthase (PKS) (PDB 4TKT, [28]) from Streptomyces platensis (Figs 4D and S4G). TsmK and the KS domain shared a low sequence identity (10%) (S4H Fig). The structural alignment of the KS domain with the TsmK model revealed how the major and minor β-sheets fit the corresponding parts of the KS domain. The resulting structure-based sequence alignment showed how the TsmK model lacks two of the three conserved catalytic residues involved in the fatty acid modification (S4H Fig). A structural inspection of the conserved catalytic residue His 366 found in the TsmK model did not reveal any cysteine in the surroundings that could substitute the catalytic cysteine of PKS. In fact, the unique cysteine found on TsmK was located in β7, far from the hypothetical conserved catalytic site. Taken all together, these results suggest that TsmK had retained structural features of the fatty-acid modifying enzyme, but not its function. One of the functions of TssJ in the T6SS is to stabilize the MC through its interaction with the central pilar of the trans-envelope complex, the protein TssM [19,29]. We thus demonstrated here that TsmK could have been co-opted by the A. baumannii T6SS to fulfill such a stabilizing function.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Structural prediction study of TsmK an Acinetobacter-specific protein related to Ketoacyl synthases.

Analysis of the TsmK predicted structure. (A) AlphaFold2 confident score (predicted LDDT per position) mapped on the model. (B) Topology of the secondary elements of the predicted model for TsmK. The diagram depicts the secondary structure organization of the A. baumannii TsmK model generated with AlphaFold2. The major and minor β-sheets, as well as the long loop, are highlighted with colored squares. Inconsistent regions are represented with pale colors. Five antiparallel β-strands, β1, β7, β6, β4, and β5 assemble into a β-sheet (referred to as the major β-sheet). The α-helices are present at the loops formed between each pair of consecutive β-strand and decorate both sides of the β-sheet. AlphaFold2 predicted a second β-sheet (referred to as minor β-sheet) not present in TrRosetta and RaptorX predictions. This β-sheet is formed with the two β-strands of the converging hairpin (β9 and β10) and the β11, β12, and β8 (C) Structural superimposition of the three structural models of A. baumannii TsmK. The major and minor β-sheets as well as the long loop are highlighted with colored squares. Structural precision ranges between 0 and 15 Å (blue to red) (D) Topology of the secondary elements of the Ketoacyl synthase domain from Acyltransferase type I polyketide synthase (PKS) (PDB 4TKT). The homologous regions between TsmK and the KS domain of the polyketide synthase are highlighted in dashed colors.

https://doi.org/10.1371/journal.ppat.1011687.g004

A. baumannii TssM undergoes a specific dynamic in the cell envelope

The second function of TssJ is to anchor the MC into the outer membrane, allowing TssM sub-domains to cross the lipid bilayer [19]. The absence of a TssJ-like component in A. baumannii prompted us to investigate the specific sub-cellular location and behavior of TssM. We first determined the membrane localization of TssM using sucrose gradient fractionation of A. baumannii total membranes (Mat. & Methods). TssM appeared strongly associated with both the inner membrane and with the Omp28-containing fraction (Fig 5A). Consequently, even in the absence of a TssJ homolog TssM is still anchored in the outer membrane fraction. To further characterize the location of TssM we have monitored its stability over time using fluorescence microscopy observation of the GFP fusion protein. The time lapse stack was reduced in one image by the average projection. The region of the image in S5A Fig illustrates that almost all cells harbored one or more stable clusters over time that are clearly visible in the average image (S5A Fig). However, observation of the recording also shows very dynamic clusters that move or appear and disappear rapidly and some cells where clusters are not well visible (S1 Movie, Fig 5B sub-panels a-b). To characterize these behaviors, we plotted the kymographs corresponding to the intensity profile of the cell contour. We could record four types of movements: fixed clusters (F), intermittent clusters (I), lateral displacement (LD), cell without clusters (WC) (Fig 5B, sub-panels c), illustrated by representative average images (Fig 5B, sub-panels a-b). The different positions and movements of TssM foci are reminiscent to the TagB localization and dynamic in the P. putida T6SS [18]. Indeed, TagB forms transient foci which accumulate and abruptly disappear over time and then re-appear in the same or neighboring cells. However, TagB is a cytoplasmic protein that is recruited at the baseplate where it stabilizes sheath polymerization. Once the sheath contracts, TagB foci are dispersed. The TssM dynamic can be better understand in the context of its colocalization with the ClpV ATPase (Fig 2G). In conclusion, we propose that TssM foci assemble and then disappear after sheath contraction, which could then explain this intermittent appearance.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Characterization of TssM location.

(A) TssM location determined by sucrose gradient. A. baumannii ATCC 17978 strains were grown, and the outer (OM) and inner (IM) membranes were separated by a continuous sucrose gradient (35%-60%). After centrifugation, 750ul samples were taken from the top to the bottom of the gradient. The samples were subjected to denaturing 12.5%- polyacrylamide gel electrophoresis (PAGE) and immunodetected with the appropriate antibody (including synthetics polyclonal TssM antibody). Immunodetected proteins are indicated on the right. Molecular weight markers (in kDa) are indicated in the left. Tags: GspC^sfGFP (T2SS IM protein), Omp28^mCherry (OM porin). (B) Dynamics of fluorescent clusters of the TssM protein. Cells were recorded in TIRF illumination during 10 min each 10 seconds (a). The column in panel (a) shows the corresponding average image of the 4 cells. Panel (b) shows 6 time points of these 4 cells. The arrows point to the cluster fluctuations (line 2) or their lateral displacement (line 3). The lower panels (c) are kymographs that illustrate 4 types of movements: fixed clusters (F), intermittent clusters (I), lateral displacement (L), cell without clusters (WC). The kymographs were constructed from the intensity profile of the cell contour. Scale: 2 μm (C) In A. baumannii ATCC 17978, the fusion of sfGFP to TssM C-terminal (TssM-sfGFP) and mCherry to a majority porin (Omp28-mCherry) was used to observe the location of the TssM C-terminal. The porin was used as a proxy to label the outer membrane. (D) Structured illumination microscopy (SIM) images of the TssM^sfGFP regarding the Omp28^mCherry label. The scale bar is 1μm. (E) Box plot representing the distances measured between the TssM foci and the mCherry label in a WT, a clpV mutant and a tssB-tssC mutant. The difference in position is equivalent to the distance between the two observed fluorophores and was measured in n = 60 bacteria (WT_ ΔclpV: t 3.1331, df = 105.39, p-value = 0.00224; WT_ ΔtssBΔtssC: t = 0.024481, df = 100.92, p-value = 0.9805; ΔclpV_ΔtssBΔtssC: t = -2.6278, df = 95.461, p-value = 0.01002).

https://doi.org/10.1371/journal.ppat.1011687.g005

The heterogeneity in sub-cellular localization and dynamic of the A. baumannii MC invited us to look at its behavior relative to the cell envelope in vivo. For this purpose, we performed super-resolution 3D structured illumination microscopy (SIM) on intact A. baumannii cells. The Omp28^mCherry porin was used as a proxy to label the outer membrane and TssM location was monitored by following the sfGFP signal, located at its C-terminus (Fig 5C). TssM foci appeared as bright green spots co-localized near the diffuse and linear outer membrane signal of Omp28 (Fig 5D). Assuming that the TssM fluorescent signal is produced by a single particle we can estimate its localization with sub-pixel accuracy by fitting a Gaussian curve to the TssM fluorescent signal along a transversal intensity profile (S5B and S5C Fig). The estimated position of TssM corresponds to the peak of the fitted curve. With the same approach we can estimate the likely position of the OM signal along the same intensity profile (S5C Fig). With these two estimated positions we can relatively localize the two signals for TssM and OM. Wildtype cells (WT) exhibit a relative position slightly shifted inward, compared with a random localization centered at 0 shift. In the absence of ClpV (ΔclpV) TssM showed a localization centered around the Omp28 signal (Fig 5D and 5E) as expected for a random localization inward/outward. Conversely, the localization of TssM without TssBC (ΔtssC, ΔtssB) also presented a slightly inward-shifted signal. In the absence of dynamic recordings and within the limits of SIM microscopy resolution, speculatively this result could mean that ClpV activity is required to maintain the inward positional shift of TssM in the OM.

A C-terminal glycine-serine linker specific to Acinetobacter TssM is crucial for T6SS functioning

We then sought to identify molecular features that could render TssM independent of TssJ for its anchoring to the outer membrane. In EAEC, TssM interacts with TssJ through the C-terminal domain (CTD), notably the β-strand domain (S6A Fig) [29]. We thus focused our attention on the conservation of the CTD. Interestingly, the interaction region between TssJ and TssM is well conserved between A. baumannii and other proteobacterial species (S6A Fig). The multiple sequence analysis of Acinetobacter TssM sequences highlighted the presence of a conserved GxxGxxxGxxG motif in the TssM CTD, which is not observed in other proteobacterial species (Figs 6A, 6B, S6A and S6B). From the amino-acid composition of the motif, rich in glycine and serine residues, and its secondary structure prediction, we hypothesized that this peptide is a linker or a random-coil region. We have named this motif GS-linker in reference to a “Glycine/Serine-rich linker”. We have modelled the structure of A. baumannii using Alphafold2 (AF2) (S6C Fig). Interestingly, this GS-linker has a loop-like conformation which is located at the interface between the last alpha α- and β-domains, at the same position occupied by a short hydrophobic 3(10)-helix for EAEC (S6C Fig sub-panel a-b). To understand the contribution of the GS-linker to TssM-TssM interaction, we monitored the homotypic interaction of part of the TssM periplasmic domain (TssM_930-1228) (Fig 6C, upper panel). A pull-down assay between Strep- and His-tagged versions of TssM revealed that TssM_930-1228 domain is able to homo-multimerize (Fig 6C, lower panel). On the contrary, deletion of GS-linker in TssM_930-1228 blocks this interaction, suggesting that GS-linker is essential for the multimerization of TssM periplasmic domain. To assess the implication of the GS-linker in the functioning of the A. baumannii T6SS, we designed by targeting mutagenesis (i) a TssM mutated protein where the first three Gly were replaced by three Ala residues and (ii) a TssM mutant where the GS-linker was completely deleted (see Material & methods). We investigated the functionality of these TssM variants to restore the Hcp secretion phenotype in the tssM deleted strain by trans-complementation. Expression of the WT tssM allele was able to complement Hcp secretion of a tssM mutant (Fig 6D, upper panel). However, complementation by a “GS^G/A” or “ΔGS” mutated alleles failed to restore the T6SS phenotype in A. baumannii. We checked that the mutations did not affect the level of TssM production, which was equivalent to WT TssM (S6D Fig.). The impact of the mutations of the GS-linker on the functionality of the A. baumannii T6SS was further validated by competition assay (Fig 6D, bottom panel). Contrary the WT tssM allele, the mutated alleles (ΔGS, G/A) both failed to restore killing of the E. coli prey.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. The A. baumannii-specific TssM C-terminal GS-linker is essential for T6SS assembly and functioning.

(A) Schematic topology of A. baumannii TssM protein. This protein has a huge periplasmic domain composed of three alpha domains (circular shape) and one beta domain (scare shape) on the C-terminal. On the C-terminal, the protein has a GxxGxxxGxxG helix (the “GS-linker”, in yellow) inserted between the last α- and the β- domain of the protein. (B) The 807 aligned TssM sequences were extracted from the set of Acinetobacter genomes with a complete T6SS operon. The height of each letter represents the information content of the corresponding residue at a given position in bits. A. baumannii strain ATCC 17978 TssM (on top of the sequence logo) sequence was used as a reference position. (C) TssM C-terminal oligomerization depends on the “GS-linker” motif. On the top, a schematic representation of the TssM C- terminal constructions used (the periplasmic domain between 930 and 1228 amino acids) to determine the interactions. Soluble extracts of E. coli BL21(DE3) cells producing the indicated protein were submitted to a simple affinity purification step on Strep-Trap. The lysate (total soluble extracts, L), and eluate (E) were subjected to denaturing 12.5%-polyacrylamide gel electrophoresis (PAGE) and immunodetected with the appropriate antibody. Immunodetected proteins are indicated on the right. Molecular weight markers (in kDa) are indicated in the left. Tags: H, 6×His; S, Strep-tag. (D) Upper panel, western blot assays probing for Hcp secretion in whole-cell pellets (P) and supernatants (S), by A. baumannii ATCC 17978 wildtype with the plasmid control, the ΔtssM mutant and the complemented mutant strain. The last two show the phenotype of ΔtssM mutant owning a plasmid pVRL2- with tssM mutated (Gly→Ala substitution in position 1172, 1175 and 1179 or a deletion of the alpha helix between the 1172 and 1179 residues). The TssM expression was induced by adding 0.1% arabinose. Lower panel, bacterial competition experiments. Survival of E. coli rifampicin-resistant after incubation with ATCC 17978 wildtype with the plasmid control, the ΔtssM mutant and the complemented mutant strain. (E) The “Small Domain Interference” (SDI) validate the importance of TssM-CTD. Dot blot quantification of Hcp secretion in wildtype with the plasmid control (pVRL2-) or a plasmid overexpressing the TssM C-terminal peptide, and in the ΔtssM mutant with a plasmid control. The C-terminal peptide was induced by adding 1% arabinose. (F) Outer membrane permeability test. Scheme representing the different protein constructs produced in E. coli BL21 (left panel). Exponential-phase cultures of E. coli overproducing different protein domains were adjusted for OD600 and serial diluted onto LB media supplemented with vancomycin 10μg/mL and 0.05mM IPTG (right panel).

https://doi.org/10.1371/journal.ppat.1011687.g006

Interestingly, the AF2 models of TssM-CTD with ΔGS display a displacement of the β-strand domain with respect to native position observed in the EAEC and WT structures (S6C Fig sub-panel c-d). To complement our knock-out approach, we performed “Small Domain Interference” (SDI) to validate the importance of TssM-CTD. We have recently used SDI or ectopic expression of peptide to surgically dissect the assembly of the T6SS wedge complex [30,31]. If the TssM-CTD is important for the multimerization of the MC and for OM anchoring, then ectopic production of this peptide outside the context of the full length TssM should interfere through competition with the functioning of the T6SS. Indeed, periplasmic overproduction of the TssM C-terminal peptide alone, encompassing the GS-linker reduced significantly (25% decrease) the Hcp secretion in the WT A. baumannii (Fig 6E). Altogether, these results indicated that the TssM GS-linker and its glycine composition is crucial for the functioning of the A. baumannii T6SS.

We then pursued to understand the involvement of TssM_ẞ domain in the crossing of the outer membrane and the specific role played by the GS-linker. We reasoned that anchoring of the TssM_ẞ domain in the outer membrane would increase the permeability of the outer membrane toward toxic compounds, a phenomenon previously observed with membrane-spanning proteins [32]. For that, we compared the permeability of the outer membrane of E. coli producing different periplasmic constructs in the presence of vancomycin, a bulky antibiotic that is normally incapable of crossing the outer membrane of Gram-negative bacteria. Two constructs were designed from the EAEC MC (TssJ, TssM_ẞ domain) and two constructs from the A. baumannii MC (TssM_ẞ domain, TssM_ẞ + ΔGS-linker) (Fig 6F, left panel). Of note, TssM_ẞ constructs are produced into the E. coli periplasm owing to a translational fusion with the N-terminal OmpA signal sequence. Overproduction of both TssJ and the TssM_ẞ domain of EAEC are both needed to trigger vancomycin sensitivity to E. coli, but not the individual proteins (Fig 6F, right panel). Interestingly, the A. baumannii TssM_ẞ domain alone rendered E. coli vulnerable to vancomycin. Conversely, deletion of the GS-linker abolished the outer membrane permeability. Together, these results suggest that the A. baumannii MC is anchor to and crosses the OM using the TssM_ẞ domain and that the GS-linker plays a pivotal role in this process.

We have identified two players that compensate the absence of TssJ functions in A. baumannii T6SS: the protein TsmK for the stabilization of the MC and the C-terminal GS-linker of TssM to help the oligomerization of TssM. To confirm this evolutionary scenario, we used a bio-informatic approach to study the co-occurrence of the three molecular events: the TssM GS-linker, the presence of TsmK, and the absence of TssJ in Acinetobacter. We analyzed the co-occurrence of these three conditions in the set of 807 Acinetobacter genomes containing both a TssM and TssB gene. The predominant configuration was the co-occurrence of the TssM GS-linker with the TsmK in the absence of TssJ, which was found in 88.4% and 82.8% of Acinetobacter baumannii and non-baumannii genomes, respectively. As a control, we also analyzed a set of 1096 non-Acinetobacter genomes containing both a tssM and a tssB gene (Fig 6F). Here, the predominant configuration was the lack of the TssM GS-linker motif with the absence of TsmK and the presence of TssJ, which was found in 61.5% of the genomes. The second most common configuration was the lack of the TssM GS-linker motif with the absence of TsmK and TssJ, which is found in 27.5% of the genomes. We have thus established a connection between the presence of both TsmK and TssM, the TssM GS-linker and the absence of TssJ as a key evolutionary event that shaped the A. baumannii T6SS.

Bacterial strains and growth conditions

The strains, plasmids and nucleotides used in this study are listed in S1 Table. E. coli and A. baumannii ATCC 17978 strains were grown in LB medium at 37°C with shaking. The antibiotics kanamycin (40 μg/ml), streptomycin (50 or 100 μg/ml), chloramphenicol (30 μg/ml), rifampin (100 μg/ml), gentamicin (10 or 50 μg/ml), carbenicillin (400 μg/ml) and tetracycline (5 μg/ml) were added where necessary.

Plasmid construction

PCRs were performed using Q5 high-fidelity DNA polymerase (New England Biolabs). Restriction enzymes from New England Biolabs was used according to the manufacturer’s instructions. Custom oligonucleotides were synthetized by IDT and are listed in S1 Table. A. baumannii 17978 chromosomal DNA were used as templates for PCRs. E. coli DH5α was used for cloning procedures. Briefly, the sequence encoding full-length gene was PCR-amplified using primers (FV and RV). The PCR introduced a 5’ and a 3’ restriction sites and usually a C-terminal TAG-extension. Restrictions were performed at 37°C with 1X NEBuffer r3.1 and the ligation at 16°C with T4 DNA Ligase Buffer (10X). Product was then transformed into competent E. coli DH5α, and recombinant strains were selected on the appropriate antibiotic. All the plasmids were engineered by restriction cloning except for the pVRL2-[TssM(GS^G/A)^His], pVRL2-[TssM(ΔGS)^His], pACYC-[TssM_930-1228(ΔGS)^Strep] and pRSF-[^HISTssM_930-1228(ΔGS)^FLAG]. The pVRL2-[TssM(GS^G/A)^His], pVRL2-[TssM(ΔGS)^His], pACYC-[TssM_930-1228 (ΔGS)^Strep] and pRSF-[^HISTssM_930-1228(ΔGS)^FLAG] vector plasmids were engineered by a two-fragments sequence- and ligation-independent cloning (SLIC) strategy [49,50]; and the pRSF- [TssM ẞ^HISAb(ΔGS), pRSF-[TssJ^STREPEAEC] and pRSF-[TssM ẞ^HISEAEC+TssJ^STREPEAEC] were constructed by megapriming. Briefly, each DNA fragments were amplified by PCR using two pairs of oligonucleotides (FW1/RV1 and FW2/RV2). PCR products were digested (DpnI), cleaned (Macherey-Nagel PCR cleaning kit), and mixed with the T4 DNA polymerase and its buffer (NEBuffer r2.1) at room temperature. The reaction was stopped after 2 min 45 s, and the mixture was put on ice. The annealing product was transformed into competent E. coli DH5α, and recombinant strains were selected on the appropriate antibiotic. The pCter-peptide is a synthetic plasmid designed by Genecust that encompasses the DNA sequences encoding for an in frame A. baumannii DsbA signal sequence fused to TssM C-terminal from amino acid 1166.

Construction of A. baumannii mutants

The primers used in this study are listed in S1 Table in the supplemental material. Mutants were constructed as described previously [51]. Briefly, an antibiotic resistance cassette was amplified using 2 sets of primers. The first one (A-primer) was composed of ~55-bp oligonucleotide homology to the flanking regions of the targeted gene with an additional 3’ 18 to 25 nucleotides of homology to the FRT site-flanked kanamycin resistance cassette from plasmid pKD4. The second (B-primer) was composed of ~55-bp oligonucleotide homology to the flanking regions of the targeted gene in 5’ of the previously amplified region with an additional 3’ 18 to 25 nucleotides of homology to the A primer. The PCR product was electroporated into a competent T6+ A. baumannii ATCC 17978 carrying pAT04, which expresses the RecAb recombinase. Mutants were selected on 12 ug/mL kanamycin, and integration of the resistance marker was confirmed by PCR. To remove the kanamycin resistance cassette, electrocompetent mutants were transformed with pAT03, which expresses the FLP recombinase, to remove the FRT-flanked resistance cassette. Additional mutations were made by repeating this process, and PCR, sequencing and complementation were used to confirm the validity of all the mutants.

SDS-PAGE, protein transfer, immunostaining, and antibodies

SDS–PAGE was performed on Bio-Rad Mini-PROTEAN systems using standard protocols. For immunostaining, proteins were transferred onto 0.2-μm nitrocellulose membranes (Amersham Protran) Immunoblots were probed with primary antibodies and goat secondary antibodies coupled to alkaline phosphatase and developed in alkaline buffer in presence of 5-bromo-4-chloro-3- indolylphosphate and nitro-blue tetrazolium. The anti-HA (HA-7 clone, Sigma Aldrich), anti-Flag (Sigma Aldrich, A2220-1ml or F3165-.2MG), anti-StrepII (Bio-Rad), anti VSV-G (Sigma Aldrich, SAB4200695-100UL), anti-6His (Sigma Aldrich, SAB1305538-400UL), anti-StrepII (Biorad MCA2489), anti-GFP (Thermofischer, MA5-15256), RNA-pol II (Ozyme 664906) monoclonal antibodies, rabbit polyclonal mcherry antibodies (OriGene TA150125) mouse secondary antibodies (Interchim 115-055-003) and rat secondary antibodies (Interchim115-055-045) were purchased as indicated. Synthetics polyclonalrabbit antibodies were designed by GeneScript for Hcp (epitope: SVGGHTAERVEHSDC) and TssM detection (epitope: CAPAQAAAPAKTENP).

Hcp secretion

Overnight cultures were back diluted in fresh LB medium to an OD600 of 0.05 and grown at 37°C with shaking (induction of the pVRL2 plasmid expression was performed by adding 0.1% of arabinose 1h after starting or by adding 1% arabinose and induction of the pVRL1 plasmid expression was performed by adding 2mM of ITPG 1h after starting, since the beginning of the culture when necessary) until they reached an OD600 of 1.5 approximately. The cells were then pelleted by centrifugation and resuspended in Laemmli buffer with β-mercaptoethanol to a final OD600 of 0.02 while the supernatant fraction was centrifuged twice again (as above) to pellet residual cells. Supernatant proteins were subsequently precipitated with trichloroacetic acid 10% and resuspended in Laemmli buffer with β-mercaptoethanol to a final OD600 of 0.1. Detection of Hcp in normalized samples of whole cells and culture supernatants by Western blotting was performed as described previously, with an anti-Hcp, anti-RNA polymerase antibodies, which were used as a lysis control or an anti-TssM, used to control the expression level of the protein in mutants complemented with different plasmids.

Bacterial killing

Cultures of A. baumannii and E. coli MG1655R were grown overnight, and the E. coli cells were washed three times with fresh LB to remove rifampin. The cells were then resuspended to an OD600 of 1.0, 100 μl of E. coli was mixed with 10 μl of A. baumannii, and a 10 μl sample was spotted onto a dry LB agar plate. After a 4 h incubation at 37°C, spots were cut from the agar and resuspended by mixing with 500 μl of LB broth. This mixture was serially diluted 10-fold, and dilutions were plated onto rifampin-containing LB agar to determine the number of E. coli CFU remaining. Controls consisted of E. coli MG1655R mixed with rifampin sensitive LB medium. Experiments were performed twice in technical triplicate. Competitions for these experiments were done at a (1:10) A. baumannii to E. coli ratio, and E. coli bacteria were enumerated on agar plates containing rifampicin.

Purification of soluble protein for interaction studies

Plasmids expressing the genes combination of interest were co-transformed into E. coli BL21(DE3). Cells were grown at 37°C in lysogeny broth (LB) to an OD600 ~ 0.8 and the expression of the TssM_930-1228 genes was induced with 1.0 mM IPTG for 3 hours at 37°C. Cell pellets were resuspended in 50 mM Tris-HCl pH8.0, 150 mM NaCl, 1 mM EDTA and supplemented with 100 μg/ml of DNase I, 100 μg/mL of lysozyme and 10 mM of MgCl2 (lysis buffer). The soluble fraction was then collected by centrifugation at 41 000 g for 45 min. The supernatant was loaded onto a 1-ml Strep-Trap-HP (GE Healthcare) column and then washed with 50 mM Tris-HCl pH 8.0, 150 mM NaCl (soluble affinity buffer) at 4°C. The co-purified proteins were eluted in the soluble affinity buffer supplemented with 2.5 mM desthiobiotin (Sigma Aldrich). The lysate, flow through, wash and elution fractions were collected, resuspended in Laemmli loading buffer supplemented with 300 mM 2-Mercaptoethanol, heated for 10 min at 96°C prior to analyses by SDS-PAGE and immunoblotting.

Purification of full-length membrane protein for interaction studies

Plasmids expressing the genes combination of interest were co-transformed into E. coli BL21(DE3) and cells were treated as described below. Cells were grown at 37°C in lysogeny broth (LB) to an OD600 ~ 0.8 and the expression of the genes was induced with 1.0 mM IPTG for 3 hours at 37°C. Cell pellets were resuspended in 50 mM Tris-HCl pH8.0, 50 mM NaCl, 1 mM EDTA and supplemented with 100 μg/mL of DNase I, 100 μg/mL of lysozyme, EDTA-free protease inhibitor (Roche) and 10 mM of MgCl2 (lysis buffer). Cells were broken using an Emulsiflex-C5 (Avestin) and clarified by ultracentrifugation at 10 000 g for 15 min. The membrane fraction was then collected by centrifugation at 41 000 g for 45 min. Membranes were mechanically homogenized and solubilized in 50 mM Tris-HCl pH 8.0, 50 mM NaCl, 0.5% w/v n-dodecyl-β-D-maltopyranoside (DDM, Anatrace), 0.75% w/v Decyl Maltose Neopentyl Glycol (DM-NPG, Anatrace), at 4°C overnight. The suspension was clarified by centrifugation at 35 000 g for 35 min. For His-tag affinity, the supernatant was loaded onto a 5 mL His-Trap-HP (GE Healthcare) column equilibrated in affinity buffer (50 mM Tris-HCl pH 8.0, 50 mM NaCl, 0.05% w/v DM-NPG) supplemented with 20 mM imidazole. The column was then washed using the affinity buffer supplemented with 50 mM imidazole and protein complexes were eluted in the same buffer supplemented with 200–250 mM imidazole. For strep-tag affinity, the supernatant was loaded onto a 5 mL Strep-Trap-HP column (GE Healthcare), washed with affinity buffer, and eluted in affinity buffer supplemented with 2.5 mM desthiobiotin (Sigma Aldrich). The lysate, flow through, wash and elution fractions were collected, resuspended in Laemmli loading buffer supplemented with 300 mM β-mercaptoethanol, heated for 10 min at 96°C prior to analyses by SDS-PAGE and immunoblotting.

Native purification of A. baumannii TssM

A. baumannii::tssM^STREP strain was transformed with the pVRL2-tsmK^HA plasmid. Growth was performed in LB medium at 37°C and induction was performed with 2mM IPTG 1h after starting, during 5H. See above for a full description of the procedure (membrane extraction and purification on STREP column).

Inner and outer membrane separation using sucrose gradient

The cells were pelleted, 250 OD₆₀₀ were resuspended in the lysis buffer, and then broken using an Emulsiflex-C5 (Avestin) and clarified by ultracentrifugation at 10 000 g for 15 min. The crude membrane fraction was isolated by ultracentrifugation at 100 000 g for 45 min and then resuspend in 0.5 mL. Outer and inner membranes were separated by differential centrifugation using a continuous sucrose gradient by using a peristatic pump (the gradient was produced by mixing two solutions, one with 35% and the other with 60% sucrose). We ultracentrifuge the tubes in a swing-out rotor ON at 274 000 g at 4°C. After centrifugation, 750ul samples were taken from the top to the bottom of the gradient.

Homologous search

The search of TsmK homologs was performed by BLASTp [53] based on the Uniref90 database [52] using as a query the TsmK sequence of A. baumannii ATCC 17978 strain. To infer the presence of a particular gene on a specific genome, we look for homologs of this gene on the given genome using BLASTp. When inferring TsmK presence, we used the already mentioned A. baumannii ATCC 17978 TsmK sequence as a query. In the case of TssM, Tssj, and TssB, we queried the TssM, TssJ, and TssB sequences from a selected list of pathogenic T6SSs (S2 Table). Homologous sequences whose blast e-value was higher than 1-e20 or whose query coverage was below 60% were discarded and hence not used to determine the presence of a particular gene on a given genome. Uniref90 and the different genomes were converted to blast databases using the makeblastdb application [53].

TaxID and genome assignment

To assign a TaxID to a given sequence, we used UniProt (UniProt Consortium 2021) annotations and the NCBI taxonomy database (UniProt Consortium 2021). Sequences that could not be assigned a TaxID were discarded. A given TaxID might have multiple subtree TaxIDs. When this happens, a unique genome could not be assigned to it. In this case, we discarded the TaxID and any sequence associated with it. EDirect NCBI tools [54] were used to download the unique genome associated with a TaxID with no subtree with multiple TaxIDs. Acinetobacter genomes were obtained from a list of 6041 Acinetobacter

TaxIDs collected from the NCBI taxonomy database. Apart from TaxIDs with multiple TaxIDs, the ones whose associated genome has less than 500 genes were discarded. As a result, we ended up with a set of 893 Acinetobacter genomes carrying a TssB gene (used as an indicator of a complete T6SS operon). To guarantee the diversity of the set of non-Acinetobacter genomes containing a tssM gene, we used the TssM sequences from the list of pathogenic T6SSs (S2 Table) as queries during a BLASTp search on Uniref90. The resulting homologs whose length was less than 50% or more than 150% of the query length were discarded. Next, we merged the remaining sequences and clustered them by 0.9 sequence identity using cd-hit [55]. Then, we assigned a unique TaxID and genome to each TssM homolog. Finally, we ended up with a set of 1096 non-Acinetobacter genomes carrying both a TssM and a TssB gene.

Protein conservation

We used a previously developed closeness metric [31] to compute the conservation level of proteins of interest. As input, we used the TssM, TsmK, TssB, and TssK sequences obtained from the 893 Acinetobacter genomes aligned with Mafft [56]. In the case of only A. baumannii, we filter the sequences using the Uniprot annotations and the NCBI taxonomy database. A. baumannii TsmK models. Two transmembrane helices in the N-terminal region of A. baumannii TsmK (from I5 to W24 and from F36 to I57) were predicted using the TMHMM server [57]. The sequence of the transmembrane helices was not included in the structural modeling of A. baumannii TsmK (from Y58 to S470) performed by AlphaFold2 [58,59], TrRosetta [60] and RaptorX [26]. To identify consistent regions between the three TsmK models, we conducted a structural alignment using TMalign [61]. The structural homology search for the consistent model of TsmK was performed by the Dali server [27].

Bioinformatic data analysis, visualization, and storage

To generate the multiple sequence alignment (MSA) figures (Figs 6B, S4H, S6A and S6B) and compute the residue conservation, we used JalView [62]. The molecular visualizations of TsmK models (Figs 4A, 4C, S4E and S4F) were rendered with UCSF Chimera [63]. MSA residue conservation was represented on a sequence logo plot using the WebLogo server (Fig 6B) [64]. Secondary structure topological diagrams (Fig 4B and 4D) were assembled with TopDraw [65]. All Python and shell scripts developed explicitly for this work are available with the initial genomic data (TsmK, TssM, TssJ, and TssB sequences) and the resulting homologous sequences at https://github.com/IFilella/AbAsaBTssM.

TssM C-terminal “GS-linker motif” identification

We used four different amino acid segments to infer the presence of the GxxGxxxG motif in the TssM C-terminal domain: GxxxGxxxG, GxxGxxxG, GxxxGxxG, and GxxGxxG, where x stands for any of the 20 bacterial amino acids. To validate the presence of the GxxGxxxG motif, the last 100 residues of TssM should possess at least one of these segments. Importantly, these segments do not exclude longer segments with an additional intermediate region (xx or xxx) followed by a fourth Gly.

Fluorescence microscopy

Sample preparation: Cells were spotted on thin pad supplemented with 3% agarose in PBS 1x, made inside a gene frame (Thermo Scientific- Gene Frame) taped to a microscope slide. An SR cover glasses (Marienfeld Superior—Precision cover glasses No. 1.5H) was then used to seal the preparation. Imaging was performed on a DeltaVision OMX SR Imaging system (GEHealthcare) with an Olympus PlanApo N 60× 1.42 NA oil immersion objective and liquid-cooled sCMOs cameras (IMM Microscopy Platform).

Data analysis—Clusters density quantification and clusters per cell quantification: Images were acquired using excitation at 488 nm during 500 ms at 100%T for sf-GFP and excitation at 568nm during 500 ms at 100%T for mCherry. For each acquisition a BrightField image was generated. The masks of cells were obtained directly from a brightfield image by thresholding Images were filtered with FFT band pass (1 to 5 pixels) with FIJI [66]. The fluorescent clusters were detected by local maxima algorithm in MicrobeJ plugin ([67]; http://www.microbej.com/) that yields the localization density maps and the cell histograms. The statistics of the number of clusters per cell were calculated with R (R Core Team (2023). _R: A Language and Environment for Statistical Computing_. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/).

Data analysis—TIRF: Total Internal Reflection Fluorescence (TIRF) Imaging was realized using the RING-TIRF mode from DeltaVision OMX, with illumination angle around 80° and an Olympus PlanApo N 60× 1.49 NA oil immersion objective. Time-lapse recordings were analyzed with the following workflow: 1—projection of the mean of fluorescence channel. 2—detection of the cell masks with cellpose [68] with the pre-trained model “cyto2” 3—filtering of the image by a bandpass FFT filter of FIJI/ImageJ, settings = minimum = 1 pixel, max = 5 pixels 4—cell tracking and contour intensity analysis and creation of kymographs with the MicrobeJ plugin on the FFT filtered fluorescence channel. (Fig 5B) The images were then analyzed using ImageJ (http://imagej.nih.gov/ij/) and the MicrobeJ plugin (http://www.microbej.com/).

Data analysis–Kymographs: Dynamics of fluorescent clusters of the TssM protein. Time-lapse was generated by acquiring 61 images every 10 seconds using excitation at 488 nm during 100 ms at 15%T. The masks of cells were obtained from the projection of the mean fluorescence channel, with Cellpose [69] with the pre-trained model “cyto2”. In parallel fluorescence channel images were filtered with FFT band pass (1 to 5 pixels) with FIJI. Cell tracking, contour intensity analysis and creation of kymographs were then done with the MicrobeJ plugin on theses filtered images.

Data analysis– 3DSIM: Images were acquired using an excitation wavelength of 488 nm during 200 ms at 10% Transmission for the sfGFP reporter and an excitation of 568nm during 200 ms at 5%Transmission for the mCherry reporter. For each acquisition, images of 512 × 512 pixels (pixel size of 80 nm) were acquired with a sample thickness of 1 μm (9 sections, Δz = 125 nm) centered at the middle of the bacteria. At each z position, SI images (five phases at three angles) were acquired for 3D-SIM imaging. Image analysis was performed using Fiji (version 1.53c) to extract intensity profiles which were then processed with python (version 3.9.5). The best z position of each stack was first selected so that sfGFP-spot and the membrane appear clearly in the image. Then intensity profiles were wisely plotted through sfGFP-spot with an orientation perpendicular to the membrane and directed towards the interior of the bacteria. The size of each profile has been set at 20 pixels (800 nm). Finally, fifty profiles were recovered for each mutant. The distance between the TssM and Omp28 was then measured with python (version 3.9.5) by fitting a gaussian curve to each profile and calculating the difference between the green maximum peak and the red maximum peak. For the heat-map, the images were acquired using excitation at 488 nm during 500 ms at 100% Transmission. The cell segmentation of the bright-field images was done using machine learning [70]. The images were then analyzed using ImageJ (http://imagej.nih.gov/ij/) and the MicrobeJ plugin (http://www.microbej.com/). Box plot Fig 5E representing the distances measured between the TssM^sfGFP foci and the mCherry label in a WT, a clpV mutant and a tssB-tssC mutant. The difference in position is equivalent to the distance between the two observed fluorophores and was measured in n = 60 bacteria (WT_ ΔclpV: t 3.1331, df = 105.39, p-value = 0.00224; WT_ ΔtssBΔtssC: t = 0.024481, df = 100.92, p-value = 0.9805; ΔclpV_ΔtssBΔtssC: t = -2.6278, df = 95.461, p-value = 0.01002).

Mass Spectrometry–Based Proteomics

The list of identified proteins provided in Fig 3E (right panel) was established from mass spectrometry–based proteomics on excised SDS-PAGE gel bands (duplicates) containing the purified proteins. In-gel trypsin digestion and MSMS analysis were performed on a Q-Exactive plus mass spectrometer (ThermoScientific) as previously described [71], but slight modifications were brought: tryptic peptides were separated by a Vanquish_Neo liquid nanochromatography (ThermoScientific) on a C18 column (Easy-Spray PepMap Neo, 1500bar 75 μm x 500 mm, 2 μm, 100 Ǻ, ThermoScientific) at 300nl/min by a linear gradient from 4% to 25% of mobile phase B (0.1% formic acid (FA) (vol/vol) / 80% acetonitrile (vol/vol)) in mobile phase A (0.1% FA (vol/vol)) for 90 min, then to 50% of B in 20 min followed by a chase at 99% B. Spectra were processed by the Proteome Discoverer software 3.0 SP1 (version 3.0.1.27 from ThermoFisher) using the Sequest HT algorithm and the Chimerys identification node with the database extracted from NCBI, Acinetobacter baumannii strain ATCC 17978 (Taxon ID 400667; 5376 entries, last modified 02/13/2023). In this study, proteins were identified at “high level” protein FDR confidence.

Outer membrane integrity assays

The permeability of the outer membrane was analysed by using the sensitivity to the antibiotic vancomycin. Briefly, E. coli BL21was transformed by a series of pRSF-Duet plasmid encoding…. Next, a colony of E. coli BL21 bearing the different pRSF-Duet plasmids was inoculated into 5 mL of LB media and incubated overnight at 37°C with shaking at 200 rpm. The overnight culture was diluted with sterile M9 medium to an OD600 = 0.1 and incubated in a 96-well NunclonΔ surface plate without shaking at 37°C until the culture reached an OD600 = 0.5. The cells were harvested by centrifugation (15,000 x g for 2 min), washed twice with assay buffer (5 mM HEPES, 5 mM glucose, pH 7.2) and resuspended in assay buffer to a final OD600 = 1.