HP1421 is required for natural transformation

A BLASTP homology search for VirB11-type ATPases pointed, in addition to the Cag-associated and well characterised HP0525 (Cagα), to the protein coded by hp1421. Sequence alignment of the two H. pylori genes with A. tumefaciens VirB11 show a high degree of homology and, in particular, a strong conservation of residues essential for the VirB11 and HP0525 ATPase activities (S1 Fig). While HP1421 has been proposed to be a VirB11 homologue for the ComB system [20], a transposon insertion in codon 20 of hp1421 did not result in a defect in plasmid transformation [16].Since that mutant preserved most of the open reading frame (ORF), including the region harbouring the putative active sites, production of an active protein could not be ruled out. Furthermore, transformation with genomic DNA was not tested on that strain. We thus decided to create a null mutant by replacing the whole hp1421 ORF by a non-polar antibiotic resistance cassette and test its capacity to be transformed by genomic DNA of a streptomycin resistant (Strep^R) but otherwise isogenic strain.

Fig 1A shows that inactivation of hp1421 resulted in transformation levels below the detection limit of the assay, which is defined by the frequency of spontaneous Strep^R mutations. In order to rule out polar effects caused by the insertion of the cassette, we reintroduced in non-essential loci constructs expressing the hp1421 ORF with a FLAG tag coding sequence fused 5’ of it. In the first case, the gene was under the control of the strong ureA promoter, while in the other, introduced into the rdx locus, the weaker comH promoter was used. In both cases, the NT capacity was completely restored to wild type levels, confirming the essential role of HP1421 in NT.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. HP1421 is essential for natural transformation.

(A) Natural transformation frequencies for indicated H. pylori strains. Data represent Mean + SEM (n = 7 → 17). Different lowercase letters indicate significant differences (p < 0.05) between treatments (SNK test). The horizontal line 1.99X10^-7 indicates frequencies of revertant due to spontaneous mutations in absence of donor DNA in the wild type strain. (B) Transformation frequencies after electroporation with streptomycin resistance genomic DNA donor. Data represent Mean + SEM (n = 4). Different lowercase letters indicate significant differences (p < 0.05) between treatments (Tukey’s Test).

https://doi.org/10.1371/journal.ppat.1014140.g001

HP1421 participates in the initial steps of NT

The NT process can be divided in four distinct steps [10]: capture of the tDNA at the surface of the cell, its uptake into the periplasm, its transport through the inner membrane and finally its integration into the bacterial genome by homologous recombination. To establish where in this process HP1421 acts, we first tested the capacity of the Δhp1421 strain to be transformed by electroporation. As shown in Fig 1B, introducing the tDNA by this method resulted in transformation frequencies of the Δhp1421 strain equivalent to those of the wild type, indicating that HP1421 is required at some point upstream of the recombination step. Inactivating comB2, a gene known to be required for the entrance of the tDNA into the periplasm [22], led to the same result. These results are consistent with a role for HP1421 linked to the ComB system.

To confirm that HP1421 is required for the tDNA entry to the cell, we used the whole-cell duplex PCR-based DNA uptake assay that allows the detection of heterologous tDNA presence either in the periplasm or in the cytoplasm [14,23,24]. The assay showed that the tDNA cannot be detected inside the Δhp1421 cells but the signal was recovered after complementation of the knockout strain with HP1421 expressed from the ureA locus (Fig 2A and 2B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. HP1421 is involved in capture and/or uptake of transforming DNA.

(A) Internalized donor DNA was monitored by PCR. Amplification products were visualized by 1% agarose gel electrophoresis. (B) Percentage ratios of donor versus recipient amplification, results were normalized with respect to the wild-type strain. Data represent Mean + SEM (n = 3). Different lowercase letters indicate significant differences (p < 0.05) between treatments (Tukey’s Test). (C) Fluorescent cy3-λDNA foci formation in H. pylori wild-type, ΔcomB2 Δhp1421 and Δhp1421 ureA::FLAG::hp1421 strains. Z maximum projections of merged images of Cy 3.5 (red channel) and bright field are presented. (D) Percentage of bacteria with fluorescent DNA foci. Data represent Mean + SEM (n = 4→8). Different lowercase letters indicate significant differences (p < 0.05) between treatments (Tukey’s Test).

https://doi.org/10.1371/journal.ppat.1014140.g002

Canonical T4SS systems provide a direct passage between the cytosol and the extracellular compartment. During NT, the delivery of the tDNA from the surface of the cell to the cytoplasm is a two-step mechanism [12,13]. In H. pylori, the uptake across the outer membrane is mediated by ComB while the passage through the inner membrane depends on the conserved ComEC protein [22,25]. If HP1421 is a VirB11 homologue associated with ComB, we expected HP1421 to be implicated in the first of those steps. To validate this hypothesis, we monitored the accumulation of tDNA in the periplasm using a fluorescently labelled tDNA. Previous work had shown that fluorescent tDNA foci are detected and stabilised in H. pylori comEC mutants [22,25]. In the absence of HP1421, but otherwise wild-type background, periplasmic foci were undetectable, as in a ΔcomB2 mutant (Fig 2C and 2D). Again, complementation by expression of the wild-type HP1421 protein resulted in the recovery of the foci. Both, the uptake and the foci formation experiments (Fig 2) for the hp1421 mutant strains reveal phenotypes identical to those obtained by the inactivation of comB2 known to be essential for the uptake of the tDNA [16]. Taken together, these results implicate HP1421 in the initial step of NT leading to the delivery of the tDNA into the periplasmic space.

HP1421 is a cytoplasmic ATPase

Sequence homology with ATPases of the VirB11 family (S1 Fig) strongly suggests that HP1421 has an ATPase activity. To confirm this, we expressed in Escherichia coli the HP1421 protein fused through its N-terminal end to a maltose binding protein (MBP) tagged with six histidine residues (6-His-MBP-HP1421, herein MBP-HP1421). The fusion protein was purified (S2A Fig) but attempts to obtain HP1421 without the MBP were unsuccessful, the protein precipitating after cleavage of the MBP. The ATPase activity was measured by determining the release of phosphate as a function of the substrate (ATP) concentration (Fig 3A). MBP-HP1421 displayed a strong ATPase activity, with a calculated K_m of 29.8 μM ATP, a value in agreement with those obtained for other enzymes of the VirB11 family [26,27]. In parallel, an equivalent construct was used to express an Asp box motif mutant of the protein (E176A). E176 is highly conserved amongst the proteins of the VirB11 family (S1 Fig) and shown to be required for ATPase activity [28]. As expected, HP1421 E176A showed no enzymatic activity (Fig 3A). Since the purification protocol was identical to the one used for the wild-type protein, these results rule out the presence of contaminating activities and confirm that HP1421 is a bona fide ATPase.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. HP1421 is a cytoplasmic hexameric ATPase.

(A) ATPase activity of purified HP1421 versions as a function of ATP concentration, Pi released during ATPase enzymatic reactions was followed by spectrometry. (B) Bacterial Two Hybrid test to assess the interaction between the indicated alleles of HP1421. pKT25 and pUT18 pair and T25-Zip T18-Zip pair were included as negative and positive control for the interaction, respectively. (C) Chromatogram for size exclusion chromatography (SEC) of 6xHis-MBP-HP1421 using a Superdex 200 column. Aligned with the peak an AlphaFold3 prediction made using 6 copies of HP1421 and 6 ATP molecules (pTM = 0.77, ipTM = 0.75, and chain-pair ipTM = 0.73 between any two monomers of HP1421 interacting in the model) is shown. One copy of HP1421 is highlighted from the model for visualization purpose. (D) Western blot of subcelular fractions for indicated H. pylori strains.

https://doi.org/10.1371/journal.ppat.1014140.g003

As it is the case with most ATPases, HP1421 is expected to form multimers. To test whether the HP1421 monomers interact between themselves, a bacterial adenylate-cyclase two hybrid (BACTH) assay was carried out. This analysis revealed that indeed HP1421 form at least dimers (Fig 3B). To establish the degree of multimerization, gel filtration assays on the purified MBP-HP1421 were performed with a Superdex 200 column. The main peak containing the fusion protein eluted at a volume corresponding to a molecular weight of approximately 460kDa which is close to that of an MBP-HP1421 hexamer, 473kDa (Figs 3C and S2B). This is in agreement with a structural model of a HP1421 hexamer predicted by AlphaFold3 [29] with a high level of confidence (Fig 3C). This predicted HP1421 hexamer is structurally very similar to the experimental structure of the HP0525 hexamer (PDB identifier: 6BGE, (S3 Fig). Additionally, including 6 ATP molecules in the AlphaFold3 model of the HP1421 showed that ATP was confidently predicted as bound into a pocket on the surface of the HP1421 hexamer and in contact with residue E176 in the model (Fig 4A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. HP1421 ATPase activity is crucial to for NT.

(A) Modelling of ATP binding in HP1421 pocket in hexameric context. AlphaFold3 model of HP1421 with 6 ATP all confidently predicted to bind into a HP1421 pocket, in contact with E176 residue (chain-pair ipTM between HP1421 and the ATP molecule bound to its surface = 0.87 to 0.89). (B) Natural transformation frequencies for indicated H. pylori strains. Data represent Mean + SEM (n = 4 → 5), the horizontal line 1.99X10^-7 indicates frequencies of revertant due to spontaneous mutations in absence of donor DNA in the wild type strain. (C) Fluorescent cy3-λDNA foci formation in H. pylori strains. Z maximum projections of merged images of Cy 3.5 (red channel) and bright field are presented.

https://doi.org/10.1371/journal.ppat.1014140.g004

T4SS-associated VirB11-like ATPases were reported to be localised in both, soluble and membrane fractions [30–32]. To examine HP1421 distribution within the cell, the soluble and membrane fractions from a H. pylori strain expressing the active FLAG tagged protein (Fig 1A) were separated. Western blot analysis showed that while the majority of HP1421 is present in the soluble extract, a fraction of the protein can be detected associated with the membranes (Fig 3D). Interestingly, in Legionella pneumophila Dot/Icm T4SS, DotB, the VirB11 homologue, is found essentially in the cytoplasm with only in a fraction of bacteria displaying DotB foci at the cellular pole. However, the DotB E191K mutant, equivalent to HP1421 E176K and previously shown to bind but not hydrolyse ATP [28], was almost exclusively found at the bacterial pole [33]. When we analysed the localisation of HP1421 E176K, also ATPase dead (S6A Fig), the balance between the two fractions was inverted, the majority of the mutant protein being associated with the membranes (Fig 3D).

The active site mutants provide a tool to determine if HP1421 ATPase activity is needed for its function in NT. We thus tested complementing the Δhp1421 strain phenotypes with either HP1421 E176K or HP1421 E176A expressed from the ureA promoter. Unlike that of the wild-type protein (Fig 1A), the expression of the active site mutants did not allow the recovery of the transformation capacity (Fig 4B) nor the accumulation of fluorescently labelled tDNA in the periplasm (Fig 4C). Together, these results show that HP1421 is an hexameric ATPase whose enzymatic activity is indispensable for NT.

HP1421 interaction with ComB4 is critical for NT

Structural and biochemical approaches revealed an interaction between VirB11 orthologues and their cognate VirB4 in various T4SS models [34–36]. In silico modelling of the H. pylori proteins predicted with good confidence that such an interaction could also occur for the HP1421-ComB4 pair. We performed AlphaFold3 modeling of the interaction between a HP1421 hexamer and a ComB4 C-terminal domain hexamer, yielding a very confident structural model (ipTM = 0.81, pTM = 0.82, Figs 5 and S4A). Since AlphaFold3 was recently shown to sometimes produce false positive predictions with high confidence scores [37], we performed complementary modelling for this interface. We confirmed that the AlphaFold3 double hexamer model is highly structurally similar to models of the HP1421-ComB4 interface using single copies of the full-length proteins, obtained with high confidence by AlphaFold2 [38,39] and AlphaFold3 (ipTM = 0.67 with AlphaFold2, 0.66 with AlphaFold3 and pTM = 0.78 with AlphaFold2, 0.75 with AlphaFold3) and to an AlphaFold2 model using only the N-terminal domain of HP1421 and the C-terminal domain of ComB4 (ipTM = 0.88, pTM = 0.89, S4B Fig). The HP1421 ATPase activity is mediated by the C-terminal domain and the model hints that this activity should not be affected by the binding of HP1421 to ComB4. In addition, the HP1421-ComB4 modeled interface is perfectly compatible with the formation of a full-length ComB4 hexamer (S5 Fig). Taken together, these observations reinforce our confidence in the modelled HP1421-ComB4 interaction.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. HP1421 and ComB4 double hexamer complex and interface.

HP1421 full length (blue) and ComB4 C-terminal domain (salmon) hexamers confidently predicted as interacting with AlphaFold3 (pTM = 0.82, ipTM = 0.81). One copy of each protein was highlighted for visualisation purposes. B) Zoom on the interface highlight important residues (R8 and R60 on HP1421, E548 and D559 on ComB4).

https://doi.org/10.1371/journal.ppat.1014140.g005

In order to empirically probe the interaction, a split luciferase assay was carried out with H. pylori strain expressing codon-optimized long (lgbit) and small (smbit) nanoluciferase fragments (S7 Fig) fused to the N-terminal end of HP1421 and the C-terminal end of ComB4, respectively. Proximity between lgbit-HP1421 and ComB4-smbit would reconstitute the luciferase and produce a luminescent signal. This was indeed the case, strongly suggesting that HP1421 and ComB4 interact in the cellular context (Fig 6A). The model presented in Fig 5 displayed a strong electrostatic complementary of the interaction interfaces (S8 Fig). We hypothesized that switching charges in some key interface residues could abolish the interaction and thus validate the model and provide a tool to assess the biological role of the interaction. Two amino-acid changes were chosen for each of the proteins based on the structural model of the HP1421-ComB4 interface (Fig 5). For HP1421, arginine 8 and 60 were substituted by aspartic acid and glutamic acid, respectively. For ComB4, glutamic acid 548 and aspartic acid 559 were replaced by arginine. The substitutions were introduced in the split luciferase constructs and the assays were carried out. As shown on Fig 6A, substituting either the arginine on the N-terminal domain of HP1421 with acidic amino-acids or the acidic amino-acids in the C-terminal domain of ComB4 with arginine abolished the luminescent signal, indicating a disruption of the interaction.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. HP1421-ComB4 interaction is necessary to allow NT.

(A) Split-luciferase assay of interaction between ComB4::SmBit and LgBit::HP1421 variants, data represent Mean + SEM (n = 5 → 8). Different lowercase letters indicate significant differences (p < 0.05) between treatments (Tukey’s test). (B) Natural transformation frequencies for indicated H. pylori strains. Data represent Mean + SEM (n = 3 → 6). Different lowercase letters indicate significant differences (p < 0.05) between treatments (SNK test). The horizontal line 1.99X10^-7 indicates frequencies of revertant due to spontaneous mutations in absence of donor DNA in the wild type strain. (C) Western blot of total fractions for indicated H. pylori strains. (D) Western-blot of subcelular fractions for indicated H. pylori strains.

https://doi.org/10.1371/journal.ppat.1014140.g006

We then tested those interaction mutants for their capacity to sustain NT. For that purpose, we constructed H. pylori strains that expressed either both wild-type genes or one or the other interaction mutants. A strain expressing the HP1421 R8D-R60E variant was completely impaired for transformation (Fig 6B and 6C). The same result was obtained with the ComB4 E548R-D559R variant (Fig 6B and 6C). To rule out an effect of the amino-acid substitutions on the biochemical functions of HP1421, we expressed in E. coli and purified MBP-HP1421 R8D-R60E to assess its enzymatic activity and its capacity to assemble as hexamers, two properties required for its function in other T4SS [40]. As shown in S6A Fig, HP1421 R8D-R60E was able to hydrolyse ATP with the same specific activity than that of the wild-type protein. When the purified mutant was subjected to a gel filtration assay to determine its oligomerization status, the elution profile obtained indicated the presence of an approximately 460 KDa complex corresponding to a hexamer, as it was found for the wild-type protein MBP-HP1421 (S6B Fig). Taken together, these results reinforce our confidence in the proposed structural model and show that the HP1421 interaction with ComB4 is essential for the functionality of the ComB system.

All the results presented so far indicate that HP1421 is indeed the VirB11 homologue linked to the ComB T4SS. We therefore propose herein naming it ComB11.

The association of ComB11 with the membrane is not dependent on ComB4

The polar localization of DotB E191K was dependent on DotO, the VirB4 orthologue [33]. In view of the above results revealing the interaction between ComB11 and ComB4, we asked whether the enrichment of ComB11 E176K at the membranes was affected by the absence of ComB4. Western blots of the membrane and soluble fractions obtained from a ΔcomB4 strain expressing the mutant ComB11 showed the same distribution of the ATPase than in a ComB4 proficient strain (Fig 6D) indicating that the interaction between the two ComB components does not drive the association, possibly transient, of ComB11 with the inner membrane.

comB11 is a core gene in H. pylori whose Com system is recent

To analyse the distribution of the ComB system and how it relates to ComB11, we analysed the 364 completely assembled genomes of Helicobacteraceae from RefSeq [41]. To be noted, 326 of these genomes are from H. pylori. We first tested that the protein ComB11 in the focal strain (26695) is sufficiently similar to the VirB11 protein profile of the T4SS model of TXSScan [42] to be regarded as a bona fide homolog. This is indeed the case, since it was significantly matched by the Virb11 protein profile of TXSScan (i-value = 8x10^-39). We then took every protein of the ComB system and searched for homologs in all the genomes of Helicobacteraceae by sequence similarity using diamond blastp [43]. The results show that most of the genes encoding ComB are very frequent in H. pylori. The least common genes are the comB7 homologs present in only 2/3rds of the genomes. It was previously observed that virB7 genes are often missed in genome scans because the gene is small and evolves fast [18]. The most common gene, and the only gene present in every single strain of H. pylori, is comB11 (Fig 7B). While comB11 is the only ubiquitous gene in the species, most of the genes regarded as central for the function of T4SS and that are very conserved in sequence (ComB4, 6, 8–10) are present in the vast majority (95%) of the strains [18,44]. Hence, the ubiquity of comB11, while intriguing, is not that exceptional. It can be due to the presence of another function for ComB11. It may also be related with the fact that comB11 is in an operon with other genes whose inactivation, or silencing due to polar effects, is likely to be deleterious to the bacterial fitness. This may lead to lower rates of pseudogenization of the gene. Interestingly, the comB11 gene in the 26695 reference genome is an operon with genes encoding proteins of unrelated function, notably a flagellar export protein ATP synthase (FliI), an isoleucyl-tRNA synthetase (IleS) or a putative UDP-N-acetylenolpyruvoylglucosamine reductase [45]. We identified bona fide homologs of most of these genes in two other Helicobacteraceae: Helicobacter cetorum and Helicobacter acinonychis, the two closest species in the family for which we had complete genomes [46]. Systems associated with natural transformation are known to be occasionally defective in strains of a species, e.g. comM of A. baumanni was found to be defective or missing in more than 20% of the strains [47]. Yet, when present, the components of orthologous systems are expected to have similar genetic organisation. In the focal strain, the ComB system was split in three loci, one locus with comB2-B3-B4, another locus ca. 20 genes apart with comB6-B7-B8-B9-B10, and finally comB11 which is distant from the others. We observed that the organization of each locus is highly conserved. When genes were present, the two first loci were strictly conserved in both H. pylori and the two close related species. Yet, relative distance between the three loci is variable (Fig 7C) and they are never contiguous (minimal distance between the comB4 and the comB6 loci is 18 genes). The gene comB11 was always distant from the others, although in many genomes it is between 30 and 40 genes apart from the comB6 locus. Hence, the ComB system is present in most H. pylori strains and the two closest species in a conserved genetic organization, suggesting that the acquisition of the system predated the divergence of these three species.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Frequency of the ComB system across H. pylori.

(A) Typical organisation of the loci. (B) frequency of each gene in the strains of H. pylori. Only comB11 is present in all the strains. (C) Distances between key genes of each of the tree loci (see A for reference). Distances are computed as the minimal circular distance between the genes, counting the number of intervening genes.

https://doi.org/10.1371/journal.ppat.1014140.g007

We have observed homologs of comB genes in species more distant from H. pylori. Helicobacter heilmannii (strain type) encodes homologs of comB2-3-4-6-8-9-11 and Helicobacter felis ATCC 49179 has homologs of comB2-3-4-6-8-9-10-11. In both genomes the genes are scattered in different loci, albeit the loci are not similar to those of H. pylori (e.g., in H. heilmannii comB6 and comB8 are isolated and separated from the locus with comB9-10). These scattered loci are almost certainly not part of a conjugative system of an integrative conjugative element, where they are usually co-localised in one single locus [48]. Further work will be needed to test if they are driving natural transformation in these species. Yet, we have observed that all these genomes lack a homolog of the ComH protein with the N-terminal domain, which was shown to be absolutely required for transformation in H. pylori [49] and was only identified in this species, Helicobacter cetorum and Helicobacter acinonychis.

Strains culture

Helicobacter pylori 26695 strains listed on S1 Table were cultured on blood agar base (Millipore) or brucella broth base (Millipore) supplemented with 1.5% bacteriological agar (Sigma-Aldrich), both enriched with 10% of horse defibrinated blood (Thermo Scientific Oxoid). For liquid cultures, brain heart infusion (BHI) (Millipore) media was supplemented with 10% heat-inactivated fetal calf serum (FCS) (Eurobio scientific). All media were supplemented with an antimicrobial cocktail containing polymyxin B (0.155 mg/ml), vancomycin (6.25 mg/ml), trimethoprim (3.125 mg/ml), and amphotericin B (1.25 mg/ml) (all from Sigma/Sigma-Aldrich). Bacteria were grown at 37 °C under microaerophilic conditions obtained with the Whitley Jar Gassing System (Don Whitley scientific, United Kingdom) and a standardized gas mixture (3% O₂, 10% CO₂) (Air Products, France) using agitation when required (150 rpm) (INFORS HT Ecotron-Switzerland). When indicated, media were supplemented with kanamycin (20 μg/ml) (Sigma), apramycin (12.5 μg/ml) (Sigma), chloramphenicol (8 μg/ml) (Fisher) and/or streptomycin (10 μg/ml) (Sigma).

Escherichia coli strains used in this study are listed in S2 Table. Cultures used were carried out in liquid or solid Lysogeny broth (LB-Sigma) supplemented with the adequate antibiotic (ampicillin 100 µg/mL or kanamycin 50 µg/ml).

Plasmid constructions

Constructions for gene disruption or complementation in H. pylori strains were carried out by using Sequence and Ligation Independent Cloning (SLIC) reaction [66]. Simultaneous and independent PCR reactions were performed in order to obtain an insert and a vector with overlapping sequences bigger than 18 bp. Briefly, 150 ng of the purified vector were mixed with the insert following a 1:5 ratio, 2 µL of NEBuffer r2.1 (New England Biolabs), 0.25 µL of T4 DNA polymerase (New England Biolabs) and H₂O up to a final volume of 20 µL. Sample was incubated at room temperature 6 min and reaction stopped by using 2 µL of dCTP 10 mM, followed by a 15 min incubation at 37 °C. Right after 7 µL were used to transform 50 µL of chemo-competent DH5-α E. coli cells. Plasmids and oligonucleotides used in this study are listed in S3 and S4 Tables, respectively.

H. pylori gene inactivation

Genomic sequences including 200 bp upstream and downstream of the gene of interest (chr: 1490990 bp to 1491904 bp for hp1421; chr: 14248 bp to 16611 bp for comB4) were amplified by PCR and cloned in pJET 1.2 backbone using SLIC reaction. A kanamycin non-polar cassette was used to replace the open reading frame for hp1421 gene and apramycin for comB4. Reference sequences were obtained from https://mage.genoscope.cns.fr/ Helicobacter pylori 26695 - chromosome NC_000915.1 [67].

Complementation

pJET 1.2 plasmids containing loci ureA (Chr: 77653 bp to 78221 bp) or rdx (Chr: 1013570 bp to 1014169) fused to the comH promoter (Chr: 1607419 bp to 1607618 bp) in position 1013912 bp were used for hp1421 ectopic complementation. For ureA locus, the gene of interest was inserted on position 77956 bp. Alternatively, hp1421 ORF was inserted downstream of the comH promoter. For both constructs a chloramphenicol non-polar cassette was inserted right after the gene. A FLAG tag was added at N-terminal side to follow up expression. Regarding constructs to test interacting comB4 and hp1421 mutants, comB4 gene (Chr: 14248 bp to 16611 bp) was inserted on a pJET 1.2 plasmid containing loci ureA (Chr: 77653 bp to 78221 bp) at position 77956 bp. Adjacent to the gene, the previous described construct “rdx-pcomH promoter-FLAG-hp1421-chloramphenicol cassette” was fused. An Alfa-tag was added at C-terminal side of comB4 to follow up expression. Different hp1421 and comB4 gene variants were obtained by site directed mutagenesis [68] of these plasmids. Introduction of the various constructs in H. pylori was carried out by natural transformation or electroporation.

For natural transformation, plate-cultured H. pylori cells were taken to 4 OD₆₀₀ in 40 µL of peptone water, containing 600 ng of plasmidic or genomic DNA harboring the construct of interest. The mixture was spotted in two 20 µL aliquots on blood agar brucella and incubated at 37 °C overnight in microaerophilic conditions. In some cases, complementation of KO strains deficient for the first steps of transformation was performed by electroporation (detailed in “H. pylori electroporation”). Grown spots were spread separately on blood agar brucella plates supplemented with the corresponding antibiotic and incubated until colonies were observed (4–6 days). Isolated colonies were transferred to new plates as patches keeping always the selection to escalate the inoculum to finally spread over half plate. Finally, a small fraction of cells was collected to extract gDNA by resuspension with 100 µL 50 mM NaOH followed by 20 minutes heating at 95 °C and a 5 minutes cooling with the addition of 20 µL 1 M Tris-HCl pH 7. PCR verifications were performed by using Kit Taq’Ozyme OneMix (Ozyme). For the complementing constructs, the genes were amplified using Q5 DNA high fidelity polymerase (New England Biolabs) and sequenced (Eurofins Genomics, Cologne, Germany).

Transformation efficiency tests

Plate-cultured H. pylori strains were taken to OD₆₀₀ = 4 in 45 µL final volume of peptone water, containing 600 ng of genomic DNA from a streptomycin resistant 26695 strain. 15 µL aliquots were spotted by triplicate over blood agar brucella and incubated overnight in microaerophilic conditions at 37 °C. Each spot was collected in 500 µl of peptone water and decimal dilutions were performed (up to 10^-6). 10 µL aliquots were spotted by triplicate on blood agar brucella plates in absence (dilutions 10^-5 and 10^-6) or presence of streptomycin (non-diluted to 10^-3). Plates were incubated until colonies appeared and CFU/mL were calculated (n° of colonies * dilution factor * aliquot factor). Transformation efficiency values were finally calculated by dividing the transformant CFU/mL over the total number of bacteria.

H. pylori electroporation

Plate-cultured H. pylori cells were taken to OD₆₀₀ = 10 in 1 mL final volume of peptone water, pelleted and resuspended in 1 mL of sterile ice-cold electrocompetent solution (9% sucrose - 15% glycerol). After centrifugation pellets were resuspended in 250 µL electrocompetent solution and 1/5 of the reaction was transferred to an electroporation cuvette and mixed with 600–1000 ng plasmid or genomic DNA. Electroporation was performed on ice, at 2.5 kV cm^− 1; 25 µF; 200 Ω (Gene Pulser II, Bio-Rad), right after 100 µL of BHI were added and 50 µL spotted onto brucella blood agar plates.

Subcellular fractionation

Plate-cultured H. pylori cells were resuspended in peptone water. Cells were washed once with sterile PBS (Gibco) 1X and normalized to OD₆₀₀ = 10 in 1mL of buffer A (Tris-HCl 10 mM pH 7.5 (Sigma); 1X protease inhibitor cocktail (cOmplete, Mini, EDTA-free, Roche). Samples were sonicated on ice (amplitude 20%, 30s, 2s pulses, 2s intervals) using a 3 mm microtip (Branson Digital Sonifier 450) and centrifuged 14000 g 15 min at 4°C (Eppendorf 5417R). The supernatant was collected and 100 µL were stored as total extract. The remaining volume was ultra-centrifuged 119000 g 45min at 4°C (Beckman Coulter Optima XPN-100), and supernatant was stored as soluble extract. The pellet containing total membranes was washed using in 1 mL buffer A and centrifuged using the same conditions as before. The pellet was resuspended in Buffer C (Tris-HCl 10 mM pH 7.5; 1X protease inhibitor cocktail; 1% N-Lauroylsarcosine) to constitute the membrane fraction. Protein identification in different bacterial compartments was done by western blot. Additionally, as control of the experiment MotB (membrane marker) and NikR (cytoplasmic marker) localization, was followed by using primary antibodies designed in rabbit and secondary Goat-anti-rabbit IgG IR700 (Advansta).

Western blots

To verify protein expression in complemented strains, exponential phase liquid cultures (18–24 h) were normalized at OD₆₀₀ = 10 in 1 mL of 20 mM Tris-HCl pH 7.5 at 4°C (Sigma), 1X protease inhibitor (cOmplete, Mini, EDTA-free, Roche), right after samples were sonicated on ice (amplitude 20%, 30s, 2s pulses, 2s intervals) using a 3 mm microtip (Branson Digital Sonifier 450) and centrifuged 14000 g 10 min at 4°C (Eppendorf 5417R). 15 µL of extract were mixed with 5 µL of fresh Laemmli 4X, and heated 5 min at 95 °C for HP1421 constructs and 50 °C 15 min for ComB4 ones. Total volume was loaded into 10–12% hand-casted acrylamide gels (Mini-PROTEAN Tetra 2-Gel, BioRad) and run using Mini-PROTEAN Tetra Vertical Electrophoresis Cell at 0.75 A-150V-30W (PowerPac HC High-Current Power Supply, BioRad) until running front was out. Resolved gels were transferred to 0.2µM nitrocellulose membranes (Trans-Blot Turbo Mini 0.2 µm Nitrocellulose Transfer Packs, BioRad) using Mixed MW program on a Trans-Blot Turbo Transfer System (BioRad). Membranes were blocked using 1X PBS, 0.3% Tween 20 (Sigma), 1% BSA (Sigma-Aldrich). For FLAG-tagged proteins, a 1/5000 dilution of ANTI-FLAG M2 primary antibody produced in mouse were used (Merck), and a 1/10000 dilution of secondary Goat-anti-mouse IgG IR800 (Advansta). Regarding ALFA-tag tagged proteins, a 1/1000 dilution of FluoTag-X2 anti-ALFA ATTO 488 antibodies designed in alpaca (NanoTag Biotechnologies) were used. Membranes were washed with 1X PBS, 0.3% Tween 20 and revealed with iBright FL1500 (Thermo Fisher Scientific).

DNA Uptake

Exponential phase liquid cultures (18–24 h) were normalized at OD₆₀₀ = 1 in 1 mL of BHI-FCS. After a washing step using BHI sterile media, 250 ng of heterologous DNA linear plasmid were added as donor to the standardized suspension in prewarmed BHI-FCS. Samples were incubated 15 min at 37 °C in agitation (180 rpm) under microaerophilic atmosphere. After incubation, suspensions were vortexed twice (2500 rpm/min) during 15 s and centrifuged (8000 rpm- 3 min). Two washing cycles were performed using sterile PBS 1X (Gibco), vortexing at high speed between each step. In order to eliminate free DNA, a nuclease treatment was performed at 37 °C during 20 min by pellet resuspension in 500 µL of a mixture containing 10 mM MgCl₂ (Merck), 50 UI/mL DNase I (Sigma) and 62.5 UI/mL Benzonase (Millipore) in 1X PBS. This step was followed by high speed vortexing, pelleting and 2 washing cycles as detailed before. Cells were resuspended in 200 µL of PBS 1X and 200 µL of phenol:chloroform:isoamyl alcohol (25:24:1) (Sigma) and vortexed 30 s at high speed. Samples were centrifuged 12000 rpm for 1min and 100 µL of the upper phase were collected in a new tube, after an equal volume of chloroform was added and DNA was extracted by vortexing-centrifugation. Two simultaneous PCR reactions using Kit Taq’Ozyme OneMix (Ozyme) were performed using oligos op284 and op285 to detect donor DNA (plasmid) and Op31 and Op32 (S4 Table) for recipient DNA (gDNA). Amplified fragments were resolved in a 1% agarose gel stained with SYBR Safe (Invitrogen), and revealed with transillumination (EBOX VX5/20M, Vilber Lourmat). After image acquisition, by using Fiji software band intensity of each reaction was measured and background subtracted to obtain an absolute valor, with those values a ratio recipient/donor was performed and internalization percentages were calculated taking the wild-type strain as 100%.

DNA foci detection

H. pylori cells collected from blood agar plates were used to inoculate BHI broth at OD₆₀₀ = 0.05-0.1. When cultures reached exponential phase (18-24h), OD was normalized to 4 in 20 µL and 200 ng of λ-DNA (New England Biolabs), labelled with Cy3 using Label IT Nucleic Acid Labeling Kit (Mirus Bio) according to manufacturer instructions, were added to the reaction. After 7 min at 37°C in agitation (150 rpm) and microaerophilic conditions, cells were pelleted and washed with 100 µL of BHI media. In order to remove free DNA, cell suspensions (50µL) were treated with 10 U DNase I (Sigma) 5 minutes at 37°C followed by high speed vortexing. Bacteria were washed once more and resuspended in 20µL BHI, 3.5 µL of this suspension were spread on µ-Slide 8 Well (Ibidi) and 400 µL of soft agar (BHI-FCS 1% agarose) was added to immobilize cells. 2D imaging was performed using an inverted-confocal Nikon A1 microscope at 37°C (The Cube 2, Life Imaging Services), 10% CO₂, 3% O₂ and 85% humidity (The Brick, Life Imaging Services); 100-x oil objective was used and z-stack 0.125 μm acquisition 1248 × 1248-pixel resolution was carried out. Image processing was done using ImageJ software. The used criteria included the manual counting of entire rod shape cells and red foci over the cells in the composite image of bright field and red channel (594 nm). After, foci percentage were calculated (total n° of cells/n° of cells with DNA foci).

Protein purification

E. coli BL21 (DE3) strains, transformed with pRSF duet containing N-terminal 6His-MBP tags fused to hp1421 gene in wild type, E176A, E176K and R8D-R60E versions, were cultured in 500 mL LB at an initial OD₆₀₀ of 0.1 at 37°C in agitation (180 rpm). At OD₆₀₀ = 0.6-0.7, IPTG (Thermo Scientific) was added to 1 mM and cultures were incubated overnight at 18°C in agitation. Cultures were centrifuged and pellets resuspended in buffer 500 mM NaCl (Sigma), 50 mM Tris-HCl (pH 8) (Sigma), 10% glycerol (Sigma-Aldrich), 1 mM DTT (Sigma), 1 mg/mL lysozyme (Sigma), 1x cOmplete EDTA-free protease inhibitor (Roche). The amount of buffer was calculated using the following formula: [(DO₆₀₀ in 1 ml* total volume of culture)/40]. After samples were sonicated on ice (amplitude 20%, 3min, 7s pulses, 7s intervals) using a 13 mm tip probe (Branson Digital Sonifier 450). The protein of interest was then purified from the soluble fraction, obtained by centrifugation at 20000 rpm 60 min at 4°C. 3 mL of equilibrated Ni-NTA Superflow resin (Qiagen) were incubated 3h at 4°C in agitation with the soluble fraction, right after the whole suspension was transferred to a 2.5 × 10 cm Econo-Column (Bio-Rad), and flowthrough was obtained through pumping at 25 rpm (Mini peristaltic pump, Gilson). After, washing steps with 1.5 column volumes was performed with buffer A (20 mM Tris pH8; 500 mM NaCl; 1 mM DTT; 10% glycerol; 20 mM imidazole), followed by 1.5 column volumes of buffer B (20 mM Tris pH8; 100 mM NaCl; 1 mM DTT; 10% glycerol; 20 mM imidazole). Proteins were eluted in steps in 2 mL fractions with: 20 mM Tris pH8; 100 mM NaCl; 1 mM DTT; 10% glycerol; 500 mM imidazole. Purified fractions were aliquoted, and stored at −80 °C.

ATPase activity

All tests were performed using the EnzChek Phosphate Assay Kit (Molecular probes, Invitrogen), following the protocol provided by the company. To determine HP1421 Michaelis constant (Km), a concentration rank of 0–4 mM ATP was used with a fixed amount of protein. Briefly the reaction was incubated 30 min at 22 °C, and absorbance was read at 360 nm on a microplate reader (CLARIOstar – BGM Biotech). Released phosphate was calculated by interpolation and corrected by subtraction of ATPase contamination measurement in absence of ATP. Km values were calculated using nonlinear least squares regression (GraphPad Prism 10.0.1). Regarding specific activity determinations on HP1421 R8DR60E mutants, phosphate release was followed on 5 minutes intervals during 45 min at 37 °C in a rank of 1–10 µM of protein at a fixed 4 mM concentration of ATP. Obtained values were corrected as indicated before, and plotted in function of time, common linear intervals were selected and slopes calculated. Finally, the slope of slope versus protein concentration graph was calculated for each determined genotype.

Bacterial two hybrid assay

T25-hp1421 and T18-hp1421 constructs designed in plasmids pKT25 and pUT18C (Euromedex) variants were co-transformed E. coli BTH101 chemocompetent cells. Two to three transformant colonies were resuspended in 100 µl of LB, and 10 µL were spotted on solid LB supplemented with 100 µg/mL ampicillin, 50 µg/mL kanamycin, 0.5 mM IPTG (Sigma) and 40 µg/mL X-gal (Roche). Plates were incubated at 30 °C 24h and images registered.

Split luciferase constructs and tests

comB4 gene (Chr: 14248 bp to 16611 bp) was inserted on a pJET 1.2 plasmid containing loci ureA (Chr: 77653 bp to 78221 bp) at position 77956 bp. Adjacent to this gene the construct “comH promoter (Chr: 1607419 bp to 1607618 bp), hp1421 gene (chr: 1490990 bp to 1491904), chloramphenicol cassette” was added. Nano-luciferase optimized fragment sequences for H. pylori expression Sm-Bit and Lg-Bit [69] were fused to N-terminal hp1421 and C-terminal ComB4 respectively. Random 15 bp linkers were added between the segments and gene sequences to allow flexibility. Different hp1421 and comB4 gene variants were obtained by site directed mutagenesis of these plasmids. All luciferase tests were performed using the Nano-Glo Luciferase Assay System (Promega). Briefly, plate-cultured H. pylori cells were taken to OD₆₀₀ = 0.1 in BHI. 50 µL of each suspension were mixed with 50 µL of reconstituted reaction buffer, by duplicate in Nunc F96 MicroWell White Polystyrene Plates (Thermo Scientific). Plate was gently mixed by tapping and 3 minutes after, readings were performed each 30s up to 5 minutes final using a microplate reader (CLARIOstar – BGM Biotech). Averaged blank corrected values were summed and corrected by OD to normalize. To each replicate, wild type non-specific activity was subtracted to each condition.

AlphaFold modelling

Modeling runs for all considered interactions were performed using the AlphaFold3 web server [29]. HP1421 hexamer with ATP, HP1421-ComB4, ComB4 hexamer were modeled with full-length proteins. The HP1421/ComB4 double hexamer was modeled with shorter delimitations due to AlphaFold3 size limit: only the C-terminal domain of ComB4 was used (from the UniProt sequence reference O24862, residues 399–787 (end)). To reinforce our confidence in the predicted interface, the HP1421-ComB4 interface was also modeled with AlphaFold2 [38] using the ColabFold [39]implementation.

Structural data visualization

Molecular graphics and analyses performed with the UCSF ChimeraX software [70], developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with support from National Institutes of Health R01-GM129325 and the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and Infectious Diseases. The command line version of matchmaker module was used to align protein chains and obtain RMSD values with 5.0 Å pruning iteration cut-off distance (to exclude outlier atoms). The “pair ss” parameter were applied when aligning complexes (hexamers) to ensure a global alignment instead of a single pair based one. The electrostatic potential at HP1421-ComB4 interface was estimated using ChimeraX “electrostatic” (coulombic) module. The pLDDT colouring was obtained directly from AlphaFold pLDDT values in the json output file.

Size exclusion chromatography

The oligomerization state of the protein is determined by passage through a Superdex 200 column. Calibration of the gel filtration column is performed using a cocktail of eight proteins: aprotinin (6.5 kDa), ribonuclease A (13.7 kDa), carbonic anhydrase (29 kDa), ovalbumin (44 kDa), conalbumin (75 kDa), aldolase (158 kDa), ferritin (440 kDa), and thyroglobulin (669 kDa). The column is equilibrated with 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM DTT, and 10% glycerol. The purified protein (100 µg) is loaded onto the column, and the elution volume of the protein is determined. By plotting this value on the column calibration curve, the apparent molecular weight of the protein is calculated.

Phylogenetic analyses

We extracted from RefSeq complete genomes [41] (S5 Table) all the 364 genomes of Helicobacteraceae (last accessed in May 2023). We used the annotations of the genomes regarding the position of every protein coding gene and their functional annotation. We used the proteins of the focal strain (26695) to identify homologs in the other genomes using diamond blastp v2.0.4.142 [43] with the ultra-sensitive option. Within the closely related species of H. pylori, Helicobacter cetorum and Helicobacter acinonychis we regarded proteins as orthologous if they were more than 80% identical. This threshold was verified upon the analysis of the loci loci, by comparing the hits with this threshold with those using a much lower threshold (e-value <0.001, down to less than 30% identity). This confirmed that only homologs at more than 80% identity are found in the loci. Once the list of orthologs was available we computed the minimal distance of each pair in the genome as the number of genes that were identified between them (accounting for the circularity of the genomes).

The test that comB11 is a bona fide virB11 homolog was done by checking that it can be identified by the virB11 HMM protein profile from TXSScan [42], using hmmsearch 3.1b2 with default options [71]. The protein profile is given in S6 Table.

Statistical analysis

Independent experiments were performed using a completely randomized design with at least three biological replicates. RStudio software and Graphad Prism 10 were used to perform statistical analysis. Each dataset was analysed using one way ANOVA and Tukey’s test at P < 0.05. Model assumptions were verified using Shapiro–Wilk and Levene’s tests, including Normal Q-Q- plot and residuals versus fits plot analysis. When residuals normality and homoscedasticity was not proven, boxcox function was used to choose the best data transformation. If assumptions weren’t still verified, non-parametrical Kruskal and Wallis test followed by post-hoc SNK test at P < 0.05 (global P value adjusted to the number of comparisons) was performed.