Identification of the periplasmic DNA receptor for natural transformation of Helicobacter pylori

Horizontal gene transfer through natural transformation is a major driver of antibiotic resistance spreading in many pathogenic bacterial species. In the case of Gram-negative bacteria, and in particular of Helicobacter pylori, the mechanisms underlying the handling of the incoming DNA within the periplasm are poorly understood. Here we identify the protein ComH as the periplasmic receptor for the transforming DNA during natural transformation in H. pylori. ComH is a DNA-binding protein required for the import of DNA into the periplasm. Its C-terminal domain displays strong affinity for double-stranded DNA and is sufficient for the accumulation of DNA in the periplasm, but not for DNA internalisation into the cytoplasm. The N-terminal region of the protein allows the interaction of ComH with a periplasmic domain of the inner-membrane channel ComEC, which is known to mediate the translocation of DNA into the cytoplasm. Our results indicate that ComH is involved in the import of DNA into the periplasm and its delivery to the inner membrane translocator ComEC.

H orizontal gene transfer (HGT) is a key driver of bacterial evolution. In pathogenic bacteria HGT favours the adaptation to new hosts and the spread of virulence factors and antibiotic resistance traits. One of the major modes of HGT in many bacterial species is natural transformation (NT), a pathway by which bacteria import DNA from the environment and internalise it into the cytoplasm where it can integrate into the bacterial genome. DNA uptake and integration by naturally transformable bacterial species is a complex process involving a dedicated set of proteins 1,2 . Unlike other HGT mechanisms such as transduction or conjugation, NT relies exclusively on proteins coded by the recipient cell.
In Gram-negative bacteria, where the incoming DNA has to pass through two membranes and the periplasmic space in between, work on Helicobacter pylori 3,4 , Neisseria gonorrhoeae 5 and Vibrio cholerae 6,7 showed that DNA transport through the bacterial envelope is a multi-step process. After binding to the cell surface, double-stranded DNA (dsDNA) is internalised into the periplasm. With the exception of H. pylori, both Gram-positive and Gram-negative transformable bacteria use a (pseudo-) pilus similar to Type IV pili as DNA import machinery 1,8,9 . However, these (pseudo-) pili are not sufficient for the transport of DNA through the outer membrane. Indeed, the protein ComE(A), also conserved amongst most naturally transformable Grampositive and Gram-negative bacteria, is required for DNA uptake 1,4,5,[10][11][12] . In V. cholerae and N. gonorrhoeae, ComE(A) is the periplasmic DNA receptor 6,10 . Once in the periplasm, the transforming DNA (tDNA) is delivered to the inner membrane transporter ComEC 13 . During this step, the dsDNA is processed to yield a single-stranded DNA (ssDNA) suitable for its ComECmediated translocation into the cytoplasm. The periplasmic proteins responsible for processing the tDNA and delivering it to ComEC are yet to be defined in any Gram-negative transformable species. Finally, through a still unknown mechanism, ComEC allows the ssDNA to cross the inner membrane and reach the cytoplasm.
H. pylori infection of the human stomach triggers a chronic gastritis that can evolve into a series of severe pathologies such as gastroduodenal ulcers and cancer 14,15 . Infecting about half of the world population, H. pylori is one of the most successful bacterial pathogens. Its amazing genetic diversity and variability are certainly major contributors to this success by allowing the emergence of new alleles. Furthermore, the new alleles, as well as antibiotic resistance genes, can rapidly propagate through natural transformation, a very efficient mechanism in H. pylori [16][17][18][19][20][21] . Amongst naturally transformable bacteria, H. pylori is unique and divergent in terms of the composition of its competence machinery. Rather than a canonical Type IV (pseudo-) pilus, H. pylori employs a type-IV secretion system (T4SS), ComB, for initial DNA uptake during transformation 22,23 . More surprisingly, no orthologue of the conserved ComE(A) DNA receptor has been identified 24 , raising the question of how tDNA is imported into the periplasm. A recent study showed that H. pylori import large amounts of tDNA into its periplasm 25 , suggesting the presence of an efficient DNA receptor protein in that bacterial compartment.
In this study, we identify ComH, a previously uncharacterised H. pylori protein, as the periplasmic DNA receptor essential for NT. We show that ComH is required for the transfer of external DNA to the periplasm. ComH interacts with DNA through its Cterminal domain of unknown fold and with ComEC putative oligonucleotide binding (OB) fold through its N-terminal domain, providing a carrier for the transforming DNA between the outer and inner membranes.

Results
ComH is essential for the import of transforming DNA. comH was originally identified as a gene required for NT by screening a mutant library of H. pylori for non-transformable strains 26 . We confirmed the effect of comH inactivation on NT by determining the frequencies of integration of a streptomycin resistant (StrepR) marker using as donor total genomic DNA from a StrepR isogenic strain. When compared to the wild-type strain, the ΔcomH mutant displayed >10,000-fold reduction in the yield of recombinant clones, an effect similar to that resulting from inactivation of comB2 or comEC (Table 1), both essential for natural competence 22,23,27 . The values obtained are however slightly higher than the spontaneous mutation frequencies, 6.3 × 10 −9 and 1.09 × 10 −9 for the wild type and comH strains, respectively, suggesting that some transformation is still taking place. To complement the mutant strain we inserted in rdxA, a nonessential locus, the ComH-FLAG coding sequences fused to the comH upstream sequences and the N-terminal signal peptide from ComH. Complementation by the ectopically expressed ComH protein restored the yield of recombinants to levels comparable to those of the wild-type strain (Table 1), ruling out the possibility of a polar effect in the ΔcomH strain.
The transformasome of H. pylori is composed of proteins involved in either tDNA uptake and transport across the bacterial envelope (i.e. components of T4SS and ComEC) or in its handling within the cytoplasm leading to chromosomal integration (i.e. DprA, RecA) 28,29 . To define in which of those two steps ComH is required, electroporation of different H. pylori mutant strains with a 139-mer chemically synthesised ssDNA coding for streptomycin resistance was performed and the recombinant frequencies were determined. ssDNA is a very poor substrate for natural transformation 30 . We did not observe significant number of StrepR colonies using the standard transformation protocol with the ssDNA but electroporation with the same substrate yielded a recombination frequency of 5.28 × 10 -7 ( Table 2). Delivery of the ssDNA into the cytoplasm by electroporation should allow circumventing the initial binding and transport processes. Indeed, recombination frequencies obtained by electroporation with ssDNA of the ΔcomB2 and ΔcomEC strains were similar to that of the wild-type strain, while almost undetectable for the ΔdprA and ΔrecA strains ( Table 2). Electroporation of the ΔcomH strain with ssDNA allowed recombination frequencies close to that of the wild-type strain ( Table 2). Taken together these results show that ComH, like the ComB complex and ComEC, is involved in the delivery of the tDNA into the cytoplasm but dispensable for its integration into the chromosome.
ComH is required for the import of tDNA into the periplasm.
During NT in H. pylori, fluorescently labelled tDNA can be transiently detected in the periplasm as discreet foci in wild-type strains but not in those mutated in comB 4,31 . To better define the role of ComH, we determined the efficiency of tDNA foci formation (Fig. 1a). While~20% of the wild-type bacteria displayed fluorescent tDNA foci, <1% of the ΔcomB2 cells had detectable foci (Fig. 1b). In a ΔcomEC strain, the proportion of bacteria with tDNA foci was similar to that of the wild-type (Fig. 1b), consistently with the role of ComEC in the DNA transport through the inner membrane 4 . When ComH was disabled, only 1% of the cells presented tDNA foci. Wild-type levels were recovered in the ΔcomH strain by ectopic expression of ComH-FLAG (Fig. 1a, b). We also monitored by PCR the presence of internalised DNA in the different mutant strains (Fig. 1c, d). After incubation of the cells in the presence of a 6.3 kb linear dsDNA with no homology to the H. pylori genome, the extracellular and unprotected DNA was removed by extensive washing and nuclease treatment. After isolating total DNA, a 1.1 kb region of the internalised donor dsDNA was amplified by PCR. In agreement with the formation of tDNA foci, amplification of internalised tDNA was observed in the wild-type while only residual amplification was detected in a  ComH is essential for tDNA import into the periplasm. a Fluorescent DNA foci formation in H. pylori wild-type, ΔcomH and ΔcomH redxA::comH-FLAG strains. Z maximum projections of merged images of ATTO-550 (red channel) and differential interference contrast (DIC) are presented. Scale bar = 2.5 µm. b Percentage of bacteria with fluorescent DNA foci from the experiments described in a. The mean and standard deviation values calculated from at least two independent experiments are shown for wild-type (n = 661), ΔcomB2 (n = 633), ΔcomEC (n = 1068), ΔcomH (n = 989) and ΔcomH rdxA:: comH-FLAG (n = 5245) cells. Bars correspond to the mean+/-SD. c The presence of donor DNA in the periplasm was monitored by PCR. The products of amplification were visualised by agarose gel electrophoresis. d The ratios of donor versus recipient amplification were obtained by scanning of the agarose gels using ImageJ and the results were normalised with respect to the wild-type strain. Bars correspond to the mean of at least 3 independent experiments+/-SEM.
ΔcomB2 strain (Fig. 1c, d). Because the level of amplified tDNA in a ΔcomEC strain was similar to that of the wild-type, it can be assumed that it essentially reflects the DNA present in the periplasm (Fig. 1c, d). Disruption of comH led to a dramatic reduction in the levels of internalised DNA, comparable to those of the ΔcomB2 strain, that could be reverted by the ectopic expression of the gene in the ΔcomH rdxA::comH-FLAG strain (Fig. 1c, d). The residual amplification observed in the competence mutant strains is likely due to non-specific DNA binding to the cell surface. Collectively, the foci analysis and the PCR amplification experiments showed that ComH is essential for the presence of tDNA in the periplasm during natural transformation.
ComH is in the periplasm where it co-localises with tDNA foci. We then proceeded to determine the subcellular localisation of ComH by fractionation of exponentially growing H. pylori ΔcomH cells expressing a FLAG-tagged version of the protein.
Western blot analysis showed that the majority of ComH-FLAG was present in the periplasm (Fig. 2a) consistently with the presence of a signal peptide (amino acids 1-19) (http://www.cbs.dtu. dk/services/SignalP/) 32 potentially targeting the protein to this compartment. With the fractionation conditions used in this experiment no association of ComH with the membrane fraction was detected. The presence of ComH in the periplasm was confirmed using a strain expressing a HA-tagged version of the protein from the comH locus ( Supplementary Fig. 1).
In view of its localisation and its role in the import of tDNA into the periplasm, we asked whether ComH co-localised with the tDNA during natural transformation. To address this question, we generated a H. pylori strain expressing a ComH-mCherry fusion protein yielding transformation frequencies comparable to those of the wild-type strain (Supplementary Table 1). The expression and stability of ComH-mCherry in H. pylori was verified by Western blot (Supplementary Fig. 2). Confocal microscopy analysis revealed that the majority of the cells showed a rather homogenous and weak fluorescent signal on which interspaced puncta or clusters could be observed (Fig. 2b). This pattern was observed independently of the genetic background. The number of clusters increased with time, reaching a maximum 90 min after putting the cells in contact with DNA. When a comEC disrupted strain (comH-mCherry ΔcomEC) was used for the co-localisation experiment in order to stabilise the tDNA foci, we found that 423 out of the 553 tDNA foci analysed (76%) co-localised with ComH-mCherry clusters (Fig. 2b). This was confirmed by a high Manders' coefficient (0.88 ± 0.02) reflecting the fraction of green signal (DNA foci) with the clusters of red signal (ComH-Cherry) determined using a threshold defined by the JACOP plugin from ImageJ 33 . The detection of ComH clusters without DNA foci could reflect the presence of non-labelled DNA arising from bacterial lysis. This is consistent with the fact that genetic markers can be transferred from one strain to another by simply co-culturing them.
ComH binds double-stranded DNA. Despite the fact that no obvious DNA binding motif could be identified by sequence analysis of the protein, the co-localisation of ComH with tDNA and the inability of ΔcomH strain to promote tDNA entry led us to hypothesise that ComH could be the periplasmic receptor for the incoming DNA. While in other naturally transformable Gram-negative bacteria this is the role of ComE(A) 5,6 , no orthologue of such proteins has been identified in H. pylori. We therefore tested the in vitro DNA binding ability of purified ComH. We first assessed the DNA binding capacity of the purified C-terminally His-tagged fusion protein (ComH-His 6 ) (Supplementary Fig. 3a) by electrophoretic mobility shift assays (EMSA). ComH-His 6 formed discrete nucleoprotein complexes with an 18 bp dsDNA in a concentration dependent manner while no binding to ssDNA (18-mer) was detectable (Fig. 3a). The same preference for dsDNA over ssDNA was observed when Surface Plasmon Resonance (SPR) was used to characterise the DNA interaction properties of the ComH-His 6 . The binding sensorgrams show that ComH-His 6 protein bound dsDNA with high affinity while only residual binding was detected when ssDNA was used as substrate (Fig. 3b). Neither the Langmuir isotherm nor the equilibrium approach using the Proteon Manager software gave satisfactory fittings for the calculation of affinity constants. In order to obtain quantitative data on the binding affinity of ComH for dsDNA, we used the microscale thermophoresis (MST) using a 18 bp dsDNA labelled with a 5'FAM 34 . The observed Kd was 0.05 ± 0.03 µM (Fig. 3c). We also analysed the interaction between ComH and 18 bp dsDNA or 18nt ssDNA using sedimentation velocity analytical ultracentrifugation (SV-AUC). With the dsDNA we observed the formation of a major complex with a sedimentation coefficient (6.5S). and with a Kd of 0.1 ± 0. To determine the DNA binding characteristics of ComH we used the purified MBP-ComH protein ( Supplementary Fig. 3b) in EMSA. MBP-ComH showed affinity for a variety of DNA substrates, as long as they harboured a double-stranded region, but not for ssDNA ( Supplementary Fig. 5a). The nucleoprotein complexes formed by MBP-ComH with a 39 bp dsDNA were stable up to 600 mM of either NaCl or KCl ( Supplementary  Fig. 5b). Purified MBP alone did not display binding to tested dsDNA using this assay ( Supplementary Fig. 5c). Taken together, the results obtained using two different tagged constructs of ComH protein and four different experimental approaches, show that ComH is a novel DNA binding protein, with strong affinity for duplex DNA compared to ssDNA.
ComH binds to dsDNA through its C-terminal domain. The full length ComH protein is composed of 479 amino acids (54.78 kDa). Preliminary bioinformatics analysis of ComH predicted the presence of two structured regions at the N-and Ctermini of the protein connected by a linker sequence. Neither region showed potential structural homology to other proteins and therefore they represent novel and uncharacterised domains of unknown function. To search for the ComH DNA-binding domain, we separately purified the N-terminal (ComH-NTD) (amino acids 20-169) and C-terminal domains (ComH-CTD) (amino acids 170-479) of the protein fused to either His 6 or MBP tags ( Supplementary Fig. 3a, b).
We first tested the DNA binding characteristics of ComH-CTD-His 6 and ComH-NTD-His 6 using EMSA. As the full-length protein, ComH-CTD-His 6 bound dsDNA while it had essentially no-affinity for ssDNA (Fig. 4a, b). On the other hand, ComH-NTD-His 6 had basically no DNA binding capacity to either dsDNA or ssDNA in the same protein concentration range (Fig. 4b, c). Similar results were obtained when we tested the DNA binding capacities of the MBP fused domains. Both, fulllength MBP-ComH and MBP-ComH-CTD formed discrete nucleo-protein complexes when incubated in the presence of dsDNA while the MBP-ComH-NTD displayed negligible DNA binding affinity ( Supplementary Fig. 6a, b). In order to obtain quantitative data on the relative binding affinity of full-length ComH versus the N-and C-terminal domains for dsDNA, the interactions of the His 6 -tagged domains were analysed by MST. The Kd obtained for ComH-CTD was 3.4 ± 1.7 µM (Fig. 4d). While this value reflects a lower affinity than the full-length protein, the ComH-NTD had a much weaker affinity (Kd > 40 µM) ( Supplementary Fig. 6c) in agreement with the EMSA analyses. Finally, we addressed the question of whether ComH needed free DNA ends for loading using His 6 tagged constructs. Both, ComH-His 6 and ComH-CTD-His 6 displayed concentration dependent binding to linear and covalently closed supercoiled plasmid DNA irrespective of their topology ( Supplementary  Fig. 7a, b).
In conclusion, the DNA binding properties of the truncated version of ComH suggest that the CTD harbours a novel DNA binding motif responsible for the affinity of ComH for dsDNA.
ComH-CTD is sufficient to allow tDNA entry into the periplasm. Since the CTD of ComH is sufficient for DNA binding, we tested whether expression of this domain in a ΔcomH strain was sufficient to revert the mutant phenotype. For this purpose, as we did for the full length protein, we inserted in rdxA the ComH-CTD-FLAG coding sequences fused to the comH upstream sequences and the N-terminal signal peptide from ComH (Fig. 5a). We then confirmed that ComH-CTD-FLAG was expressed and exported to the bacterial periplasm like the fulllength form (Fig. 5b). The ectopic expression of ComH-CTD-FLAG resulted in a partial restoration of tDNA foci formation (Fig. 5c). The capacity of the CTD to allow the import of the tDNA to the periplasm was confirmed by the duplex PCR assay. Indeed, ectopic expression of ComH-CTD in the ΔcomH strain allowed significant amplification of the tDNA retained by the bacteria after nuclease treatment, albeit to lower levels than those observed for the mutant strain complemented with the full-length ComH (Fig. 5d). We then assessed the capacity of the truncated form of ComH to allow the whole natural transformation process up to the chromosomal integration of the tDNA. Unlike that of the fulllength ComH-FLAG, the ectopic expression of ComH-CTD-FLAG in the ΔcomH strain did not restore the capacity to obtain recombinants (Fig. 5e).
Taken together, these results suggest that while the CTD of ComH is sufficient for the normal detection of the tDNA in the periplasm of competent cells during NT, it does not support its translocation into the cytoplasm. These results were further confirmed following the internalisation of fluorescently labelled DNA into the bacterial cytoplasm, as previously described 31 ( Supplementary Movies 1-3). While in a wild-type strain the volume of internalised DNA increased over time, in a ΔcomEC mutant the tDNA remained blocked in the periplasm, reflecting the impairment in this mutant of the transport through the inner membrane. When the ΔcomH rdxA::comH-CTD-FLAG strain was tested in this assay, it displayed a phenotype very similar to that of the ΔcomEC mutant ( Supplementary Fig. 8).
ComH NTD interacts with the periplasmic OB domain of ComEC. The similitude of phenotypes between the ΔcomEC and ΔcomH rdxA::comH-CTD-FLAG strains (periplasmic foci formation but no translocation of the tDNA into the cytoplasm) raised the possibility that ComH might act in close association with ComEC to allow the transport of tDNA through the inner membrane. Topological prediction of H. pylori ComEC structure suggests that this inner membrane protein has 9 transmembrane segments and an oligonucleotide binding (OB) domain 35 present on the periplasmic side of the membrane (Fig. 6a). We therefore tested a potential interaction between ComH and the ComEC-OB domain. Extracts from E. coli bacteria expressing His 6 -tagged recombinant fusions of the ComEC-OB domain were incubated with E. coli extracts containing either MBP or MBP-ComH. The products of amylose affinity pull-down from the mixtures were then eluted from the resin by competition with maltose and probed by immunoblot for the presence of the ComEC-OB domain using anti-His tag antibodies. As shown in Fig. 6b, ComEC-OB-His 6 was recovered from the pull-down when coincubated with full-length MBP-ComH, indicating a direct interaction between ComH and the OB domain from ComEC. To rule out the possibility that the interaction was mediated by DNA present in the extracts, the pull-down experiments were repeated with the addition of benzonase. The interaction between ComH and the ComEC OB domain persisted even after degradation of the DNA (Supplementary Fig. 9a, b). Interestingly, the His 6tagged OB domain was also recovered when MBP-ComH-NTD was used as bait but not in the case of MBP-ComH-CTD or MBP alone. These results showed that through its NTD, ComH is able to interact with ComEC, providing a link between the handling of tDNA in the periplasm and its passage through the inner membrane.

Discussion
During natural transformation H. pylori displays a remarkable efficiency to import tDNA into the periplasm of the bacteria and its subsequent recombination in the genome 25,36 . Pioneering work in H. pylori showed that this pathogen uses a two-step DNA uptake mechanism in which first the dsDNA is transported from the surface of the cell to the periplasm and subsequently it is transferred to the cytoplasm through the inner membrane 4 . Similar two-step DNA uptake mechanisms were described for the Gram-negative bacteria N. gonorrhoeae and V. cholerae 4,5,10 . While the first step is mediated by species-specific DNA uptake machineries, the ComB type IV secretion system in the case of H. pylori or type IV (pseudo-) pili for the other naturally transformable bacteria, the second step of DNA translocation across the cytoplasmic membrane is mediated by the conserved ComEC (ComA in Neisseria) channel 3 . In N. gonorrhoeae and V. cholerae it is proposed that the periplasmic DNA binding protein ComEA (ComE in Neisseria) is required for DNA traversal of the outer membrane by retaining it within the periplasm 5-7 .
In the absence of a comEA orthologue in H. pylori, we searched for proteins able to bind the transforming DNA in the periplasm.
Here, we showed that ComH is the H. pylori periplasmic DNA receptor. The comH gene (hp1527) is critical for natural transformation, the transformation frequencies of the comH null mutants being close to the detection limit 26,36 (Table 1). Amino acid sequence analysis of ComH indicated that it is unique to H. pylori with apparently no homology to known protein motifs or domains, making it difficult to speculate on its function during competence 26    ARTICLE putative signal peptide that would target the protein to the periplasm. Despite complete sequence dissimilarity to ComEA proteins, we found that ComH shows the essential characteristics of periplasmic DNA receptor proteins. Indeed, ComH is present in the periplasm, binds dsDNA with high affinity and is required for the formation of tDNA foci, all properties shared with ComEA. While most of the ComEA proteins harbour conserved structural helix-hairpin-helix (HhH) motifs crucial for the DNA interaction 6 , ComH has no obvious DNA binding domain. We show that its C-terminal domain is responsible for the dsDNA binding and thus contains a novel DNA binding fold. Further structural studies should allow its definition. As it is the case for ComEA with the Type IV pili involved in the DNA uptake in other species, whether ComH interacts with the ComB components remains to be determined. How the DNA is handed to the ComEC inner-membrane channel is still unknown. It has been proposed that the presence in the periplasmic space of ComEC loops or domains would allow the interaction of the membrane-embedded channel with other DNA-uptake machinery proteins 13,37 . Interestingly, the capacity of ComH CTD to bind dsDNA is sufficient for the import of the incoming DNA into the periplasm but not for its translocation into the cytoplasm, indicating a separation of functions. Truncation of the NTD leads to the persistence of DNA foci in the periplasm, a phenotype observed in comEC mutants 4,31 . These results suggest that this domain is implicated in the delivery of the DNA to the inner membrane pore. Our observation of a physical interaction between ComH NTD and the OB domain from H. pylori ComEC, predicted to be exposed to the periplasm, is consistent with a role for ComH in the handing of the DNA to the inner membrane ComEC channel. It would be interesting to investigate if in the case of other naturally transformable bacteria ComEA is also implicated in the delivery of the tDNA to ComEC (ComA in Neisseria).
Based on the results presented here, we propose that ComH represents a new class of DNA binding proteins and is not only the missing DNA receptor for natural transformation in H. pylori but also the protein responsible for delivering the DNA to ComEC. Its role in the translocation of tDNA through the bacterial envelope can be divided in two steps as represented in Fig. 7. By its capacity for binding dsDNA through its CTD, ComH acts as a receptor for the incoming DNA into the periplasm (Fig. 7a). Subsequently, ComH could deliver the DNA to the inner-membrane channel through the interaction of the NTD with ComEC OB domain (Fig. 7b). Although we have here unveiled the role of a new player in the mechanism of transforming DNA internalisation in H. pylori, several questions remain open. In S. pneumoniae, the nuclease EndA is required to generate the single stranded DNA capable of crossing the innermembrane but a protein with similar function remains to be identified in other species 2 . It has been proposed that ComEC could harbour such an activity 35,38 . Furthermore, the need of other yet unknown proteins to process the dsDNA or to coordinate the different steps of the DNA passage through the periplasm cannot be ruled out.

Methods
H. pylori growth conditions. H. pylori cultures were grown under microaerophilic conditions (5% O 2 , 10% CO 2 , using the MAC-MIC system from AES Chemunex) at 37°C. Blood agar base medium (BAB) supplemented with 10% defibrillated horse blood (AES) was used for plate cultures. Liquid culture were grown in brain heart infusion media (BHI) supplemented with 10% defibrillated and decomplemented fetal bovine serum (Invitrogen, Carlsbad, CA, USA) with constant shaking (180 rpm). Antibiotic mix containing polymyxin B (0.155 mg/ml), vancomycin (6.25 mg/ml), trimethoprim (3.125 mg/ml), and amphotericin B (1.25 mg/ ml) was added to both plate and liquid cultures.  Table 3) and cloned in pjET1.2 (2974 bp) vector by blunt end ligation. Further, the cassette encoding either the antibiotic resistance or indicated protein tag were introduced in the gene construct by either classical restriction-ligation method or using sequence-and ligation-independent cloning (SLIC). All the plasmids harbouring different gene constructs (listed in Supplementary Table 4) were verified by DNA sequencing and introduced into H. pylori by natural transformation. Transformants were selected by using appropriate required antibiotics: kanamycin (20 μg/ml), apramycin (12.5 μg/ml), and chloramphenicol (8 μg/ml). The correct constructions were verified by PCR using locus and gene specific primers. The glycerol stocks of H. pylori were prepared in BHI media supplemented with 12.5% glycerol and stored in -80°C. The details of different constructions generated in this study are given below. Construction of comH::Km/Cm. hpcomH (hp1527) locus was amplified using primers Op81 and Op82 and ligated to blunt pjET1.2 vector to generate p1196 (pJet1.2-comH).The PCR fragments generated by amplification of p1196 (using primers Op83 and Op84), non-polar kanamycin (using primers KpnI-Km-for and BamH-Km-rev), and chloramphenicol (using primers KpnI-Cm-for and BamHI-Cm-rev) resistance cassette were digested using KpnI and BamHI and ligated separately to generate p1198 (pJet1.2-comH::Km) and p1295 (pJet1.2-comH::Cm). p1198 and p1295 were used for allelic replacement of native comH locus.
Constructions of translational fusions. +/-300 bp at the end of hp1527 locus was amplified (using primers Op86 and Op104) and cloned in pJet 1.2 vector to generate plasmid p1225. DNA sequence coding for linker-mCherry-Cm (amplified from plasmid p1146 using primers Op107 and Op108) was inserted before the STOP codon of comH gene (p1225 amplified using Op105 and Op106) using SLIC to generate p1236 (pJet1.2-comH-linker-Cm). Further, BirA-HA coding sequence was replaced with mCherry of p1236 to generate p1320 (pJet1.2-comH-BirA-HA-Cm). These plasmids were then used for allelic replacement of native comH locus of the wild-type strain.
Determination of recombination frequencies. Natural transformation assays were performed as described in 39 . Briefly, exponentially growing H. pylori cells (optical density of 4.0 at 600 nm) were incubated with isogenic streptomycin resistance total chromosomal DNA (200 ng) on Blood Agar Base (BAB) plates for 24 h at 37°C. Next day, serial dilutions of H. pylori cells were plated on BAB plates with or without streptomycin (10 μg/ml) and incubated for 4-5 days at 37°C. The recombination frequencies were calculated as the number of streptomycin resistance colonies per recipient colony-forming unit. P values were calculated using the Mann-Whitney U test on GraphPad Prism software.

Determination of spontaneous mutation frequencies.
Exponentially growing H. pylori cells (optical density of 4.0 at 600 nm) were spotted on Blood Agar Base (BAB) plates for 24 h at 37°C (no exogenous DNA was added). Next day, the spots were collected in peptone water. 200 µl of non-diluted sample was directly plated on BAB plates with streptomycin (10 μg/ml). In total 20 µl of 10 -5 dilution was plated on BAB plates. The plates were incubated for 6 days at 37°C. The mutation frequencies were calculated as the number of streptomycin resistance colonies per colony-forming unit.
Determination of recombination frequencies after electroporation. Chemically synthesised 139-mer ssDNA (Op245) (Supplementary Table 5) carrying A128G mutation in the hp1197 gene was used as substrate for electroporation experiment. Electro-competent cells were prepared by re-suspending exponentially growing H. pylori cells (optical density of 10 OD/ml at 600 nm) in 1 ml of ice-cold 15% Glycerol, 9% Sucrose. After centrifugation (2204 g) for 3 min), the bacterial pellet was re-suspended in 250 µl of ice-cold 15% Glycerol, 9% Sucrose. 50 µl of these electro-competent cells were mixed with 1 µg of salt free ssDNA in a clean electroporation cuvette. The electroporation was performed at 2.5 kV cm −1 , and 25 µF. Immediately, 100 µl BHI was added and 50 µl cells were spotted on BAB plates. Next day, serial dilutions of H. pylori cells were plated on BAB plates with or without streptomycin (10 μg/ml) and incubated for 4-5 days at 37°C. The recombination frequencies were calculated as the number of streptomycin resistance colonies per recipient colony-forming unit. P values were calculated using the Mann-Whitney U test on GraphPad Prism software.
Subcellular fractionation. To obtain the subcellular fractions, 20 ml of bacterial cultures at OD 600 1.5 were pelleted, washed in PBS and resuspended in 2 ml buffer C (20% Sucrose, 1 mM EDTA, 30 mM Tris-HCl, pH8) and further incubated for 10 min at room temperature and centrifuged at 5700 × g. The pellet was resuspended in 1 ml of ice-cold water and incubated 10 min at 4°C. After centrifugation at 5,766 g, the supernatant corresponding to the periplasmic fraction was recovered. The pellet was resuspended in 1 ml of buffer A (10 mM Tris-HCl, pH 7.5, 1 mM DTT, 1× protease inhibitor cocktail) and sonicated. After centrifugation at 147,420 × g the cytoplasmic (supernatant) and membrane (pellet resuspended in 0.2 ml of 30 mM Tris pH 8, 100 mM NaCl) fractions were recovered. The presence of proteins in different bacterial compartments was monitored by immunoblotting.
Uncropped images are provided as Source Data. As controls for the fractionation experiments, we determined the localisation of MotB and NikR, membrane and cytoplasmic proteins, respectively.
Duplex PCR assay. The presence of transforming DNA in the periplasm of H. pylori was detected using protocol adapted from ref. 6  Fluorescently labelled DNA substrates for microscopy. DNA substrates were prepared, as described in ref. 31  Fluorescence microscopy experiments. Both live and fixed H. pylori cells were prepared as described 31 . To monitor the fractions of bacteria that formed DNA foci, exponentially growing H. pylori cells were incubated with 200 ng of ATTO-550-dUTP labelled dsDNA (408 bp) for 7 min at 37°C, the unbound DNA was washed and the bacteria were resuspended in BHI and covered with low melting agarose (1.4%) supplemented with 10% fetal bovine serum and imaged immediately. To monitor the time dependent internalisation of ATTO-550-dUTP labelled dsDNA (408 bp) in GFP expressing bacteria, the samples were prepared as described above and the bacteria that formed DNA foci were observed under live conditions [gas mixture (10% CO 2 , 3%O 2 ), humidity (90%)] at 37°C for 3 h. To monitor in vivo co-localisation of ComH-mCherry with DNA, 1 µg of ATTO-488-dUTP labelled dsDNA (2 kb) was incubated with exponentially growing H. pylori cells for 90 min at 37°C under microaerophilic conditions. The bacteria harvested by centrifugation (323 g for 3 min) were subjected to DNaseI (50 U/ml) treatment at 37°C for 5 min followed by 1× PBS wash before fixation with formaldehyde (4%). All the images were captured with a 60x objective using inverted Nikon A1R confocal laser scanning microscope system equipped with an environmental chamber. The images were processed and analysed using NIS-element software (Nikon Corp., Tokyo, Japan) and ImageJ software. The bacteria that formed DNA and ComH-mCherry foci were counted manually. For large-scale purification, the plasmids carrying different recombinant DNA constructs were transformed into E. coli BL21 expression strain and grown in LB media containing ampicillin (100 μg/ml) at 37°C. Induction of the proteins were achieved by IPTG (0.5 mM) at an optical density of 0.6 at 600 nm for 4-5 h at 30°C. The cells harvested by centrifugation were resuspended in lysis buffer (50 mM Tris pH 8.0, 500 mM NaCl, 1 mM DTT, 0.1 mM EDTA, 0.2% NP40, and 1× protease inhibitor cocktail). 5 mM imidazole was added in case of the His 6tagged proteins. After sonication, the soluble fractions were separated from the cell debris by centrifugation at 18,637 × g for 45 min. Standard Ni 2 +-NTA, and Amylose affinity chromatography was used for the first step of purification of proteins. The fractions containing the proteins were pooled, and desalted using HiTrap Desalting Column (GE Healthcare). Full-length ComH, and ComH-NTD were further purified using Resource-S, while the ComH-CTD was further purified using Resource-Q ion-exchange chromatography. Purity of the proteins was analysed by SDS-PAGE and the pure fractions were pooled and dialysed twice for 4-6 h against storage buffer (50 mM Tris, 100 mM NaCl, 1 mM DTT, 0.1 mM EDTA, and 10% glycerol). The proteins were quantitated using Bradford assay and small aliquots were frozen immediately in liquid N 2 , and stored at −80°C.
Electrophoretic mobility shift assay. HPLC purified, chemically synthesised oligonucleotides (5′-Cy5 or 3′-TAMRA labelled) (Eurogentec) were used for the EMSA experiments (Supplementary Table 5). Different DNA substrates were prepared by annealing appropriate combinations of oligonucleotides (Supplementary Table 5). Indicated concentrations of purified proteins were incubated with DNA substrates (30 nM) in binding buffer (10 mM Tris-HCl pH 7.5, 50 mM KCl, 0.1 mM MgCl 2 , 1 mM DTT, 0.1 µg/µl BSA). The NaCl concentration was maintained at constant amount (~100 mM) by adding protein storage buffer to the reaction mixture. After incubation of 30 min on ice, the reaction mix was separated by native-PAGE (6%). The gels were visualised by using Typhoon and quantified using ImageJ. Percentage binding was estimated by measuring the depletion in substrate DNA considering DNA without protein as 100%.
Surface plasmon resonance. SPR binding kinetics were analysed using the BIAcore3000 optical biosensor (GE Healthcare) at 25°C. A 50-mer 5′-biotinylated DNA (Op249) was immobilised on a streptavidin-coated surface biosensor chip. The double-stranded was generated by hybridising the complementary DNA strand (Op250). Variable concentrations of ComH-His6 (0-5 µM) in binding buffer (PBS, and 0.005% Tween20) were injected into the biosensor chip for 240 s followed by dissociation time of 900 s.
Microscale thermophoresis. DNA binding kinetics using microscale thermophoresis were carried out on a NanoTemper´s Monolith NT.115 instruments. Briefly, an 18-mer 5'-fluorescently labelled oligonucleotide with a FAM probe was hybridised with the complementary DNA. Increasing concentrations of ComH-His 6 , ComH-CTD-His 6 and ComH-NTD-His 6 were used (typically 0.3 nM to 5 µM) in buffer 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 10 mM MgCl 2 , 0.05% v/v Tween20. The signals to noise ratio are above 12 as recommended (signal is defined as the response amplitude (difference of Fnorm between DNA alone and DNAprotein complex) and the noise, defined as the standard deviation of errors between replicates). They range between 26 to 42 for three interactions. The difference of Fnorm is positive for ComH and ComH-NTD and negative for ComH-CTD as already observed for other projects.
Sedimentation velocity analytical ultracentrifugation (SV-AUC). The interactions between ComH-His6 and dsDNA or ssDNA were measured using an Optima proteome-lab XLI (Beckman Coulter, Palo Alto, USA) with an An-50 Ti rotor at 20°C equipped with a fluorescent detector (Aviv) by doing an isoterm analysis, a titration of dsDNA by ComH-His6. 400 μL of samples were centrifuged at 128,297 × g. Sedimentation profiles were collected every 5 min and analysed as described [40][41][42] . A 18-mer 5'-fluorescently labelled oligonucleotide with a FAM probe was used as ssDNA. The 18 bp dsDNA was formed by hybridising the 18nt ssDNA with the complementary DNA. Increasing concentrations of ComH-His6 (between 3 nM to 2 µM) were added to 60 nM of fluorescently labelled oligonucleotids in buffer Tris-HCl 50 mM pH 7.4, NaCl 150 mM, MgCl2 10 mM, Tween 20 0.05% v/v. We observed a transient population with a sedimentation coefficient of 3.8S and a high sedimentation complex with sedimentation coefficient of 6.5S. The K D for 3.5S and 6.5S complexes are 0.1 and 0.3 µM and may correspond to ComH-dsDNA and [ComH] 2 -dsDNA, respectively.A 18-mer 5'-fluorescently labelled oligonucleotide with a FAM probe was used as ssDNA. The 18 bp dsDNA was formed by hybridising the 18nt ssDNA with the complementary DNA. Increasing concentrations of ComH-His6 (between 3 nM to 2 µM) were added to 60 nM of fluorescently labelled oligonucleotids in buffer 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 10 mM MgCl 2 , 0.05% v/v Tween 20. We observed a transient population with a sedimentation coefficient of 3.8S and a high sedimentation complex with sedimentation coefficient of 6.5S. The K D for 3.5S and 6.5S complexes are 0.1 and 0.3 µM and may correspond to ComH-dsDNA and [ComH] 2 -dsDNA, respectively.
MBP pull-down assay. The coding sequence (amino acids 30-156) of hpcomEC amplified using primers HpComEC-30_F and HpComEC-156_R2 was ligated to plasmid pnEAvH (digested using NdeI + BamHI) using SLIC to generate plasmid p1397. The interaction between MBP-ComH and ComEC-OB-His 6 was tested by amylose affinity pull-down assay. The plasmids carrying these recombinant constructs were transformed in E. coli BL21 cells. The expression of the proteins was induced by IPTG (0.5 mM) and cultures were grown for 3 h at 37°C. Equal amounts of these E. coli cultures were mixed together and sonicated. After centrifugation, the soluble fraction of the cell lysate was incubated with amylose resin for overnight at 4°C to allow the interaction between the proteins. The resin was washed, and the bound proteins were eluted using maltose (10 mM). Equal amounts of the elution fractions were separated by SDS-PAGE. The presence of ComEC-OB-His 6 in the eluted fractions was detected by Western blots using anti-His antibodies. Uncropped gel images are provided as Source Data.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.