Evolution of Helicobacter: Acquisition by Gastric Species of Two Histidine-Rich Proteins Essential for Colonization

Metal acquisition and intracellular trafficking are crucial for all cells and metal ions have been recognized as virulence determinants in bacterial pathogens. Virulence of the human gastric pathogen Helicobacter pylori is dependent on nickel, cofactor of two enzymes essential for in vivo colonization, urease and [NiFe] hydrogenase. We found that two small paralogous nickel-binding proteins with high content in Histidine (Hpn and Hpn-2) play a central role in maintaining non-toxic intracellular nickel content and in controlling its intracellular trafficking. Measurements of metal resistance, intracellular nickel contents, urease activities and interactomic analysis were performed. We observed that Hpn acts as a nickel-sequestration protein, while Hpn-2 is not. In vivo, Hpn and Hpn-2 form homo-multimers, interact with each other, Hpn interacts with the UreA urease subunit while Hpn and Hpn-2 interact with the HypAB hydrogenase maturation proteins. In addition, Hpn-2 is directly or indirectly restricting urease activity while Hpn is required for full urease activation. Based on these data, we present a model where Hpn and Hpn-2 participate in a common pathway of controlled nickel transfer to urease. Using bioinformatics and top-down proteomics to identify the predicted proteins, we established that Hpn-2 is only expressed by H. pylori and its closely related species Helicobacter acinonychis. Hpn was detected in every gastric Helicobacter species tested and is absent from the enterohepatic Helicobacter species. Our phylogenomic analysis revealed that Hpn acquisition was concomitant with the specialization of Helicobacter to colonization of the gastric environment and the duplication at the origin of hpn-2 occurred in the common ancestor of H. pylori and H. acinonychis. Finally, Hpn and Hpn-2 were found to be required for colonization of the mouse model by H. pylori. Our data show that during evolution of the Helicobacter genus, acquisition of Hpn and Hpn-2 by gastric Helicobacter species constituted a decisive evolutionary event to allow Helicobacter to colonize the hostile gastric environment, in which no other bacteria persistently thrives. This acquisition was key for the emergence of one of the most successful bacterial pathogens, H. pylori.


Introduction
Helicobacter pylori is a gram-negative bacterium that colonizes the stomach of about half of the human population.Infection by this pathogen causes the development of gastro-duodenal ulcers, MALT lymphoma and gastric carcinoma [1,2] and can lead to gastric cancer, which is responsible for about 800,000 deaths worldwide every year [3].Virulence of H. pylori directly depends on its capacity to persistently colonize the stomach, a hostile and acidic niche.Survival under such conditions relies on the activity of the nickel-containing urease, an enzyme that catalyzes hydrolysis of urea into ammonia and bicarbonate, two buffering compounds that allow the bacterium to maintain its cytoplasmic pH close to neutrality [4].The only other nickeldependent enzyme in H. pylori, the [NiFe] hydrogenase [5], has also been shown to be important for colonization, presumably because it provides an alternative respiratory pathway, allowing H. pylori to use molecular hydrogen as an energy source [6].In this regard, the transition metal ion Ni(II), which is an essential constituent of the active site of urease and [NiFe]hydrogenase, can be considered as an essential determinant for the virulence of H. pylori and for in vivo colonization [6,7].
Urease is very abundant in H. pylori accounting for 10% of total soluble proteins and its maturation requires four accessory proteins for nickel delivery into the active site (UreE-F-G-H) [8].Under neutral pH and/or low nickel availability conditions, only a fraction of the urease pool is found in a nickel-loaded form.Upon exposure to acidic and/or nickelreplete conditions, the remaining pool of urease becomes activated by the nickel maturation machinery allowing a rapid adaptation to pH variations [9].Urease requires up to 24 nickel ions per fully active enzymatic complex [10], while [NiFe] hydrogenase depends on a binuclear [NiFe] center, coordinated by CO and CN -ligands.
H. pylori therefore needs to acquire large amounts of nickel in order to maturate and activate these two essential enzymes and consequently to survive within the stomach (for a review, see [11]).Accordingly, nickel intracellular concentration of H. pylori cells is about 50-fold higher than in Escherichia coli [12].The nickel concentration measured in the human body is very low (0.5 nM) [13] and, as expected from its vital need, H. pylori possesses dedicated nickel uptake mechanisms.We previously identified the FrpB4 protein as the first TonB-dependent transporter mediating energized nickel uptake across the outer membrane [14].Once nickel reaches the periplasmic space, it is transported through the inner membrane by the high affinity transporter NixA [15].Then, nickel accumulates in the cytoplasm where its homeostasis and trafficking is controlled by several dedicated nickel-binding proteins that are only partially characterized [16,17].As for other metal ions, the intracellular nickel concentration needs to be tightly controlled at the level of its uptake, storage and efflux.Indeed, non-physiological accumulation of free Ni(II) ions generates toxic effects such as interference with other metalbinding proteins or metal-dependent enzymes thereby disrupting essential catalytic functions that can lead to cell death [18].
Nickel trafficking displays several originalities in H. pylori.First, the maturation factors required for nickel incorporation into urease (UreE-F-G-H) and [NiFe] hydrogenase (HypA-HypB) are functionally overlapping as shown by genetic [19] and interactomic approaches [20].This suggests the existence of a molecular cross-talk as well as a controlled nickel distribution strategy between these two enzymes [20].Second, H. pylori possesses three atypical histidine-rich (His-rich) proteins, namely the heat shock protein A (HspA) and the two paralogous proteins Hpn and Hpn-2.Because poly-His peptides are known to bind nickel, it was proposed that the His-rich proteins found in H. pylori could serve as nickel stores [21,22].We recently showed that expression of hspA, hpn and hpn-2 genes is upregulated by the nickel-responsive transcriptional regulator NikR in response to nickel [23].
HspA, the sole homologue of the GroES chaperonin in H. pylori, displays a conserved C-terminal histidine and cysteine-rich (His-and Cys-rich) extension [24], absent from non-Helicobacter bacteria that behaves like a nickel sequestration domain [25].This extension participates in nickel resistance and allows HspA to function as a dedicated chaperone for the maturation of [NiFe]-hydrogenase [25].Hpn and Hpn-2 are atypical homologous proteins, because they contain an exceptionally high number of His residues, mostly clustered into stretches (Fig 1).Hpn, (HP1427 in strain 26695 [26]), was the first to be described [22].It is a small (MW: 7.08 kDa) protein representing about 2% of total proteins in H. pylori [22].It is composed of 60 amino acid residues among which 28 His and two Cys pairs (Fig 1).Hpn-2 (also designated Fig 1.Sequences of the Hpn and Hpn-2 proteins of H. pylori strain B128 [27,28].Histidine residues are highlighted in blue and Glutamine residues are in green.The triangle corresponds to the position at which Hpn is truncated in Hpn-ΔC mutant and at which Hpn-2 is truncated in the Hpn-2ΔC mutant.The black stars the positions at which the Hpn and Hpn-2 sequence are interrupted in the study of Seshadri et al. [29].doi:10.1371/journal.ppat.1005312.g001 Hpn-like or Hpnl), is a 66 amino-acid long protein (HP1432 in strain 26695) with a MW of 8.07 kDa [26]).Hpn-2 is similar to Hpn in its N-terminal half and central part ( Fig 1).It is composed of 16 His residues (out of 34 amino acid residues), while its C-terminal region is strongly enriched in glutamine (Gln) residues (21 Gln out of 32 amino acid residues) (Fig 1).
Metal binding properties of Hpn and Hpn-2 have been characterized with purified recombinant proteins produced in E. coli.As expected from their high His content, both proteins can bind in vitro Ni(II) as well as other metal ions zinc (Zn(II)), cobalt (Co(II)), copper (Cu(II)) and bismuth (Bi(III)) and some residues important for binding were identified [30][31][32][33][34].In solution, Hpn exists predominantly as a 20-mer [35] and each monomer binds 5 Ni(II) with a K d of 7.1 μM.Purified Hpn can assemble in vitro into amyloid-like fibers, interact with membrane mimics and is cytotoxic to gastric epithelial cell cultures [36] [37].The actual existence of such fibers in H. pylori needs however to be established.Hpn-2 appears to be a 22-mer in its native state, with one monomer binding 2 Ni(II) with a K d of 3.8 μM [38].The metal-binding capacity of Hpn and Hpn-2 has been addressed in vivo by FRET in E. coli using overexpressed engineered recombinant fluorescent proteins [39,40].These studies revealed that Hpn did not bind nickel or zinc in E. coli while it was able to interact with Bi 3+ .In E. coli Hpn-2 is able to bind nickel and its C-terminal Gln-rich domain has little effect on metal binding [40].Interestingly, when expressed in E. coli, Hpn and Hpn-2 both provided protection against nickel toxicity [38,41].
The in vivo function of Hpn and Hpn-2 has also been addressed in H. pylori.Initial studies reported that Hpn is an important player in protection against nickel and bismuth toxicity [42] and that complete deletion of the hpn gene does not alter urease activity [22].More recently, inactivation of hpn and hpn-2 genes by a cassette inserted approximately in the middle of the ORFs (Fig 1) suggested that Hpn-2 plays a major role in protection against nickel, while Hpn has only a minor role.The authors concluded that both proteins compete with the nickeldependent urease maturation machinery under low nickel conditions [29].Given these discrepancies, the respective roles of Hpn and Hpn-2 in H. pylori remained to be precisely established and were addressed by the present study.
The Helicobacter genus is composed of two subgroups, the enterohepatic species that infect the liver or gastrointestinal tract of mammals and some birds and a small group of gastric Helicobacter species (including H. pylori) [43].Previous studies suggested that Hpn homologues are expressed in the gastric H. mustelae [22] and H. felis species [42] but the corresponding genes were neither identified nor annotated and Hpn homologues were never reported in other Helicobacter species.Phylogenetic distribution and evolutionary history of the atypical Hpn and Hpn-2 proteins consequently remained unknown and was established for the first time during our study.
In the present study, by combining assays to measure metal resistance, nickel accumulation and urease activity together with interactomic analysis, we provide evidence that Hpn and Hpn-2 act in a common pathway to control both nickel accumulation and urease activity in H. pylori.Using phylogenomic approaches, we determined the taxonomic distribution of hpn and hpn-2 in the Helicobacter genus.Most interestingly, we showed that speciation of the gastric Helicobacter species from the other Helicobacter species occurred concomitantly with the emergence of hpn, while the duplication event at the origin of hpn-2 took place in the common ancestors of H. pylori and H. acinonychis species.Using top-down proteomics, which is based on the analysis of intact proteins, without any digestion step, we were able to demonstrate the expression of these proteins in gastric Helicobacter species.In conclusion, our data suggest that acquisition of the nickel-binding protein Hpn was decisive for the gastric adaptation of Helicobacter.Accordingly, we observed that Hpn and Hpn-2 are both essential for colonization of the mouse stomach by H. pylori.

Establishing a core-proteome-based phylogeny of the Helicobacter species
To begin our study of Hpn and Hpn-2, we decided to search for their presence and examine the taxonomic distribution of the corresponding genes in the available Helicobacter genomes.As a first step for this phylogenomic analysis, we establish a core-proteome-based phylogeny on 330 proteomes of Helicobacter from various geographical origins associated with different pathologies available at the NCBI (see Table 1 and S1 Table), including 305 H. pylori strains, 7 gastric and 11 enterohepatic non-pylori Helicobacter species (representing 10 and 15 strains of each respectively, see Table 1 and S1 Table).The core-proteome-based phylogeny of these strains was inferred using the 281 proteins present in single copy in at least 320 out of the 330 strains.The resulting phylogeny is relatively well resolved (most SH-like support (SH) values > 0.95) and showed a clear separation between the enterohepatic and the gastric strains (S1 Fig and Fig 2).While strains belonging to the same species clearly group together (i.e. for H. suis, H. bizzozeronii, H. bilis and H. cinaedi), our phylogeny revealed that the two H. cetorum strains MIT 00 7128 and MIT 99 5656 do not, the latter being more closely related to H. acinonychis and H. pylori (SH = 1.0,S1 Fig and Fig 2).This suggests that the two H. cetorum strains may rather correspond to two different species than to different strains of a same species.Regarding enterohepatic Helicobacter, we observed important discrepancies between our core-proteome-based phylogeny and 16S rRNA phylogenies (S1 Fig and Fig 2) that likely result from the use of a much larger number of positions in the core proteome phylogeny (1,500 nucleotide positions and 82,741 amino acid positions, respectively).We observed that H. cinaedi strongly grouped with H. hepaticus (SH = 1.0), while according to 16S rRNA trees it is more closely related to H. bilis [44].Our tree supported also two robust clusters: H. fennelliae, H. macacae, and H. canis on the one hand and H. pullorum, H. canadensis and H. rodentium on the other hand (all SH = 1.0).In contrast, these species were described as dispersed in 16S rRNA trees [44].Regarding non H. pylori gastric species, we observed a basal branching of H. felis (SH = 1.0), whereas this species represents the sister-lineage of H. bizzozeronii in 16S rRNA trees [44].Finally, the large-feline strain H. acinonychis Sheeba [45] does not form a separate lineage, but instead emerges from within H. pylori strains, more precisely as a close relative of strains SouthAfrica7, SouthAfrica50 and SouthAfrica20 (SH-like support = 0.96).This is in agreement with previous observations suggesting that H. acinonychis results from a host jump from early humans to large felines about 200,000 years ago [45].This novel core-proteome based phylogenetic tree provided us with a solid framework to study the distribution and evolutionary history of the hpn and hpn-2 genes in the Helicobacter genus.
hpn and hpn-2 genes are specific to the gastric Helicobacter species Surprisingly, Hpn and Hpn-2 proteins were only annotated in a few sequenced genomes of H. pylori (11 and 56 out of 305 in total, respectively) and in none of the other Helicobacter species (S1 Table ).In several H. pylori genomes, longer ORFs in different frames overlapping the hpn and hpn-2 predicted regions were annotated instead.We suspected that the hpn and hpn-2 genes might have been missed because they encode small proteins with a highly biased amino acid composition.Therefore, we searched for ORFs predicted to encode these proteins in the corresponding genomic sequences.We found ORFs corresponding to Hpn and Hpn-2 in every H. pylori strain and in Helicobacter acinonychis (Fig 2 and S1 Table ).Importantly, Hpn coding genes were detected in every gastric Helicobacter species, but in none of the enterohepatic species (Fig 2 and S1 Table ).

Scenario of the emergence and evolution of Hpn and Hpn-2 in the Helicobacter species
The strong similarity between Hpn and Hpn-2 (see multiple alignments in Suppl S2 Fig) indicates that these proteins are homologous and derived from a common ancestral sequence.Using the core-proteome based phylogenetic tree, we established the taxonomic distribution of hpn and hpn-2.It suggests that the hpn gene emerged in the last common ancestor shared by all gastric Helicobacter species whereas hpn-2 originated later, likely through the duplication of hpn, in the lineage leading to H. We inferred the phylogeny of Hpn and Hpn-2 sequences present in the 75 H. pylori strains and in H. acinonychis strain Sheeba and compared it to the phylogeny of these 76 strains using the 281 core-genome markers mentioned above.The three topologies were compared with an approximately unbiased (AU) test [46].The phylogeny of the 76 strains and that of Hpn-2 was significantly rejected by Hpn sequences, while  the phylogeny of the 76 strains and that of Hpn was significantly rejected by Hpn-2 sequences (all p-values = 0.0).This indicates that the evolutionary histories of Hpn and Hpn-2 are different and do not follow the phylogeny of the strains, suggesting that horizontal gene transfer events had indeed occurred.
The hpn genomic context in the 100 strains was examined.In H. pylori, H. cetorum and H. acinonychis genomes, hpn is located in a conserved region containing a gene cluster (cluster 1, in blue on Fig 3) composed of four genes corresponding to a 23S rRNA methyl-transferase (rMNT), a polysialic acid capsule expression protein (KpsF), the essential RNase J ribonuclease (RNAse J) [47] and the 16S rRNA dimethyladenosine transferase (KsgA).In contrast, in the other gastric species H. mustelae, H. felis, H. bizzozeronii, H. heilmannii and H. suis, we observed that hpn belongs to very different genomic regions (Fig 3).In H. pylori, the position of hpn and hpn-2 relative to cluster 1 varies from one strain to another.In some strains, hpn and hpn-2 surround cluster 1 and are expressed on different strands, whereas in other strains the two genes are adjacent and expressed on the same strand (Fig 3).These differences likely result from chromosomal inversions that might be the consequence of genomic recombination events.Therefore, we hypothesized that this inversion event has led to a duplication of hpn that gave rise to its paralogue hpn-2.In accordance with our conclusion of a more recent origin of Hpn-2, expansion of Gln residues occurred unequally in both the N and C-terminal regions of Hpn-2 whereas Hpn has remained nearly unchanged (S2 Fig) .Finally, the genomic organization of the regions encompassing hpn and hpn-2 is not fully consistent with the phylogeny of the strains.Certain closely related strains do not have the same genomic organization (e.g H. pylori NQ4060 and NQ4200, Figs 2 and 3), while distantly related strains may present a similar genomic organization (e.g.H. pylori F30 and B28).This suggests that comparable genomic reorganization events of this region occurred independently in different H. pylori strains, and/or that additional horizontal gene transfers took place at this locus (in agreement with the AU tests).

Identification of Hpn and Hpn-2 proteins in H. pylori and in non-pylori Helicobacter species by top-down proteomics
Our survey of Helicobacter genomes led us to identify ORFs predicted to encode Hpn and Hpn-2 proteins in a misannotated DNA region.These proteins were not identified by the various proteomic studies published on H. pylori [48,49].We speculated that these proteins have been missed because of their small size (<10 kDa) and because of the absence of Lys and Arg residues in their primary sequence, that precludes the use for their identification of classical bottom-up proteomics strategies based on trypsin digestion to generate peptidic fragments suitable for MS analysis.These two features make them however ideal candidates for top-down proteomics approaches [50,51].Top-down proteomics is an emerging technology based on the analysis of intact proteins and thus does not involve enzymatic digestion steps.Therefore, we decided to use this technology to search for the Hpn and Hpn-2 proteins in the proteomes of H. pylori and of non-pylori Helicobacter species.
This study was performed with five selected representative strains belonging to the gastric Helicobacter group, namely H. pylori B128 (as a positive control), H. acinonychis (str.Sheeba), H. felis (ATCC 49179), H. bizzozeronii (CCUG35545) and H. mustelae (12198).In order to identify Hpn and/or Hpn-2 in total protein extracts from liquid cultures of these strains, we performed selective purification of the nickel-binding proteins, taking advantage of their property to strongly bind Ni-NTA agarose before their analysis by LC-MS/MS.Because the genomic regions containing the ORFs potentially encoding Hpn and Hpn-2 were misannotated, we manually implemented the sequences of these proteins in the corresponding databases before searching the proteomics data.
The top-down proteomics analysis procedure enabled us to unambiguously identify both Hpn and Hpn-2 proteins in H. pylori and in H. acinonychis.As a negative control, none of these proteins was detected in the enterohepatic species H. hepaticus.The Hpn protein (but not Hpn-2) was also identified in the gastric Helicobacter strains H. felis, H. bizzozeronii and H. mustelae.Interestingly, several proteoforms of Hpn were found in H. mustelae corresponding to N-terminal variants of the 3 first amino acid residues.The data are presented in Table 2.These results fully validate the in silico detection of hpn and hpn-2 genes in the genomes of gastric Helicobacter group and confirms that both proteins are produced.

Hpn but not Hpn-2 plays a major role in protection against nickel overload
The His-rich composition of Hpn and Hpn-2 and previous data prompted us to examine the importance of these proteins in the sensitivity of H. pylori to nickel overload [29,42].This was evaluated in different genetic backgrounds (B128, SS1, X47-2AL and 26695) using mutants carrying complete deletions of hpn or hpn-2 and of both genes (     These data indicate that the Hpn protein is crucial to protect H. pylori against nickel overload while Hpn-2 is not required for this function.

Hpn but not Hpn-2 contributes to H. pylori intracellular nickel storage
To evaluate the role of Hpn and Hpn-2 in the control of nickel homeostasis in H. pylori, the total intracellular nickel content was measured by Induced-Coupled-Plasma Optical Emission Spectrometry (ICP-OES).Nickel was measured in the parental B128 strain, isogenic mutants and complemented strains exposed to NiCl 2 (200 μM) as in our previous studies [14].When bacteria were incubated without nickel, the intracellular nickel was below the detection limit of ICP-OES.Upon exposure to nickel, the wild-type strain contained 0.3 μg of nickel per mg of proteins (Fig 5).In all experiments, the intracellular nickel content of the Δhpn-2 mutant was identical to that measured in the wild type strain.In contrast, we observed a decrease in the intracellular nickel content in the mutant strains Δhpn (1.6 fold), and ΔhpnΔhpn-2 (1.8 fold) as compared to the wild-type strain.Trans-complementation of Δhpn with a wild type hpn copy restored wild type nickel content under these conditions.This result indicates that Hpn but not Hpn-2 participates in the accumulation of nickel in vivo.

Both Hpn and Hpn-2 control urease activity
We then evaluated the influence of the Hpn and Hpn-2 proteins on the activity of hydrogenase and urease, the two nickel-containing enzymes of H. pylori.Hydrogenase activity, measured as previously described [25], was not altered in the mutant strains grown in BBβ medium supplemented or not with nickel.Urease activity was monitored by measuring NH 3 production of intact bacteria incubated in BBβ with 5 mM urea (Fig 6).Nickel concentration in BBβ medium is about 0.2 μM [52].We compared urease activities of wild type and mutant strains grown without or with 10 μM NiCl 2 .As expected, urease activity of the wild type strain was enhanced (about 3-fold) in the presence of 10 μM NiCl 2 .In non-supplemented medium, urease activity of the Δhpn mutant was identical to the parental strain.Interestingly, nickel-dependent urease activation was lost in this latter mutant, and restored to wild type level upon complementation (Δhpn + phpn) (Fig 6).In addition, we found that under nickel-restricted conditions, urease activity of the Δhpn-2 mutant was significantly enhanced (2-fold) as compared to the wild type strain, a phenotype that was restored by complementation with a wild type hpn-2 gene.Upon nickel addition, urease activity of the Δhpn-2 mutant increased to wild-type levels.We concluded that Hpn is necessary for full urease activation by nickel, while Hpn-2 directly or indirectly restricts basal urease activity under low nickel conditions.Surprisingly, in contrast to the single Δhpn-2 mutant, basal urease activity of the ΔhpnΔhpn-2 double mutant was similar to that of the wild type strain.This indicates that hpn deletion suppresses the phenotype of the Δhpn-2 mutant.Upon nickel addition, urease activity of the double mutant is slightly induced and reached a lower level than the wild type strain.
Western blotting using anti-UreA antibody was performed to test whether the differences in urease activity of the mutants were caused by changes in the amounts of urease (S4 Fig) .Under nickel-restricted conditions, UreA levels were similar in all strains tested.As expected, the expression of UreA was induced in the presence of nickel (2-fold after 8 hours).We concluded that the differences in urease activities displayed by the mutants are not due to variation in urease amounts but rather to modification in nickel loading into the enzyme.
Taken together, these results suggest that Hpn and Hpn-2 genetically interact in a common pathway to control urease activation, possibly at the level of nickel insertion.

Protein interaction networks between the nickel trafficking proteins
Previous studies showed that H. pylori Hpn and Hpn-2, purified from E. coli, each form homomultimers in vitro [35,38].Our data suggest that in H. pylori Hpn and Hpn-2 act in a common pathway to regulate urease activity.We consequently explored the in vivo interactions of these two proteins with each other, and with (i) the urease structural subunits (UreA-B), (ii) the urease accessory proteins involved in nickel incorporation (UreE-F-G-H), (iii) the urea channel (UreI) and finally (iv) the HypA-B urease/hydrogenase maturation subunits that were known to interact with each other [53].In addition, we tested the influence of the C-terminal regions of Hpn and of Hpn-2 on the interactions (deletion sites represented in Fig 1).For this large study, we used the bacterial two-hybrid approach that tested pairwise interactions in E. coli [54].The hpn, hpn-2, hpnΔCter and hpn-2ΔCter genes were cloned in frame at their 5' or 3' extremities, with fragments encoding either the N-terminal (T25) or the C-terminal part (T18) of the Bordetella pertussis adenylate cyclase resulting into four different constructs for each protein.The ureA, ureB, ureI, ureE, ureF, ureG, ureH, hypA and hypB genes were cloned in frame at their 3' extremities with the T25 and T18 fragments.The different plasmids were transformed into E. coli in combination of pairs and the protein-protein interactions were evaluated by measuring the expression of the lacZ reporter gene.The results are illustrated in Fig 7 and the ß-galactosidase activity data are in S2 Table .As a control, background adenylate cyclase activity was measured for each fusion plasmid by co-transforming E. coli with the compatible empty vector.Only the two combinations pUT18/pKNT25(Hpn) and pUT18(Hpn)/ pKNT25 presented relatively high background ß-galactosidase activity 400 and 600 units respectively (light green in S2 Table ).Therefore, interactions with these fusions were only considered as positive when ß-galactosidase activity was significantly higher than 600 units, thereby avoiding false positives.
First, all four combinations of plasmids expressing Hpn fusions scored positive in ß-galactosidase assays, showing that Hpn is able to form homo-oligomers in vivo.The C-terminal region of Hpn region is important but not essential for this interaction.
Concerning the interaction of Hpn-2 with itself, we observed that co-transformation of the plasmids expressing pKNT25(Hpn-2) fusion protein with either pUT18(Hpn-2) or pUT18 (Hpn-2ΔC) was lethal (materials and methods).We suspect that this is due to very strong Hpn-2/Hpn-2 interaction, resulting in extremely high adenylate cyclase activities that are toxic to E. coli, possibly through depletion of the ATP pool (as already observed by D. Ladant, personal communication).The pKNT25(Hpn-2ΔC) fusion interacted with both pUT18(Hpn-2) or pUT18(Hpn-2ΔC).In this context, our results strongly suggest that Hpn-2 is able to form homomultimers and that its C-terminal Gln-rich region is important but not essential for this interaction (Fig 7 ).
We then examined the interactions between Hpn and Hpn-2.Four out of the eight combinations of Hpn and Hpn-2 fusions scored highly positive, indicating very strong interaction between Hpn and Hpn-2.
The interactions of Hpn and Hpn-2 with the urease structural subunits and accessory proteins were then tested.No interaction was detected with UreB, UreE, UreH, UreF and UreG, neither with the UreI urea channel protein.Importantly, UreA presented strong interaction with Hpn but not with Hpn-2, an interaction that was apparently reduced with the HpnΔC fusion (Fig 7).Finally, both Hpn and Hpn-2 interact with HypA and HypB, the two hydrogenase accessory proteins.The ß-galactosidase activities of the Hpn/HypA and Hpn/HypB combinations were higher than those with Hpn-2.Deletion of the Hpn-2 C-terminal region resulted in a complete loss of its interaction with HypB.
In conclusion, these data show that Hpn and Hpn-2 form both homo-and hetero-multimeric protein complexes in vivo (Fig 7).In addition, Hpn strongly interacts with the UreA structural subunit and both Hpn and Hpn-2 interact with HypA and HypB.The Hpn/Hpn-2 protein-protein interaction network summarized in Fig 7 is perfectly in line with the data presented above suggesting that Hpn and Hpn-2 participate in a common pathway of nickel transfer.

Hpn and Hpn-2 are essential for colonization of the mouse model
The role of Hpn and Hpn-2 during gastric colonization was evaluated using the mouse model.Three different H. pylori strains able to colonize the mouse stomach were tested: SS1, X47-2AL and B128.Mutants carrying complete deletions of hpn, hpn-2 and of both genes were constructed.In the SS1 background, the Δhpn and the Δhpn-2 mutants were complemented in cis by the corresponding wild type genes.All strains were orogastrically inoculated to seven NMRI mice per strain.One month later, colonization was assessed by quantitative cultures of stomach homogenates ( Fig 8).We observed that depending on the genetic background, the single Δhpn and Δhpn-2 mutants were either strongly affected or completely deficient in their ability to colonize the mouse stomach.The complemented SS1 mutants recovered the capacity to colonize the murine stomach.Importantly, the double ΔhpnΔhpn-2 mutant was totally unable to colonize the mouse stomach in every genetic background.
These results revealed for the first time a crucial role of Hpn and Hpn-2 during in vivo colonization.

Discussion
The relation between metal trafficking and bacterial virulence has attracted large interest these last years.In addition to uptake and efflux, it is now admitted that cytoplasmic concentration of transition metals is often controlled by specific proteins that can detoxify and/or act as metal  (A-C) the tables represent the results of the βgalactosidase activity measurements on E. coli strains containing the pair-wise combinations of the different constructs.T18 and T25 correspond to the two fragments of the adenylate cyclase and the protein extremity at which the fusion was made (N ter or C ter ) is shown.Panel A: interaction between Hpn, HpnΔCter, Hpn-2 and Hpn-2ΔCter.Panel B and C: interaction of Hpn, HpnΔCter, Hpn-2 and Hpn-2ΔCter with UreA, HypA and HypB.Red squares correspond to β-galactosidase activity that are > 3,000 U, orange squares between 2,000 and 3,000 U, yellow squares between 500 and 2,000 U and light yellow squares, between 240 and 500 U. None of the potential false positive interactions was presented (interactions of pUT18/pKNT25(Hpn) and pUT18 reservoirs that can be mobilized when necessary [55].In H. pylori, nickel is required for stomach colonization as it is the cofactor of two enzymes essential in the animal model, the very abundant urease and the [NiFe] hydrogenase [11].Large amounts of nickel must be acquired by H. pylori in order to supply these enzymes with their metal cofactor.This bacterium expresses three His-rich proteins that were proposed to serve as nickel stores, HspA, Hpn and Hpn-2.We previously showed that HspA is a nickel sequestration protein and a specialized Ni (II) chaperone for hydrogenase [25].However, little information was available regarding the in vivo role of the two other His-rich proteins of H. pylori, Hpn and Hpn-2.In addition, nothing was known on the evolution and phylogenetic distribution of these unusual proteins in bacteria.

Role of Hpn in the control of intracellular nickel content of H. pylori
In a previous study, Hpn-2 and Hpn were reported to play respectively a major and a minor role in nickel tolerance [29].In the present work, we found in agreement with Mobley et al [42], that a mutant carrying a complete hpn deletion is much more sensitive to the toxic effects of nickel than the corresponding wild type strain.Lower intracellular nickel amounts (Hpn)/pKNT25 with ß-galactosidase activity below 600 units).Black squares represent combinations that lead to a lethal phenotype.The β-galactosidase activity values for each strain and the controls are indicated in S2 Table .(D) Schematic representation of the protein-protein interaction network of Hpn and Hpn-2.Thick arrows correspond to strong interactions, thin arrows to moderately strong interactions, according to the ß-galactosidase.* Indicates that this interaction was established before [53].accumulated in the hpn deletion mutant, a phenotype that was restored to wild type concentrations in the complemented strain.This suggests that Hpn is sequestering nickel in vivo in agreement with its capacity to bind nickel in vitro [35].In contrast, we observed that Hpn-2 is not required for protection against the deleterious effects of nickel and has no influence on the intracellular nickel content.Strengthening our conclusions, we observed identical phenotypes for the mutants in four different genetic backgrounds (B128, X47-2AL, SS1 and 26695) and recomplemented the Hpn phenotype in two of these strains (Fig 4).One plausible explanation for the discrepancy between our data and Seshadri's work [29] is that their hpn and hpn-2 mutants were constructed by insertion mutagenesis implying that N-terminal fragments of both proteins are most likely still expressed and may partially retain a function because of their repetitive composition (for Hpn a fragment with 14 His out of 27 and for Hpn-2 a fragment with 26 His out of 27, shown in Fig 1).In contrast, our analysis was performed with complete deletion mutants and allowed us to unambiguously attribute a major role of Hpn in nickel sequestration.
Small iron-binding proteins that are able to form oligomeric structures, such as ferritins [56], have long been known to be involved in storage and detoxification of this metal.Given the in vitro behavior of Hpn, the phenotypes associated with its depletion and the fact that it represents about 2% of the total protein content of H. pylori, it is tempting to attribute a similar function to Hpn.In vitro, Hpn can bind copper, zinc and bismuth in addition to nickel [32,35,40].However, our phenotypic analysis of the Δhpn mutant indicates that besides nickel storage/detoxification, Hpn might additionally be involved in protection against bismuth (our data and previous work by [42]).
Finally, further experiments are needed to understand how metal ions bound to Hpn are mobilized and delivered to other proteins in vivo, whether, as we suspect, Hpn-2 is involved in this process and finally if metal binding to Hpn and Hpn-2 controls their multimerization and interaction.

Control of urease activity by Hpn and Hpn-2
Phenotypic and enzymatic assays on H. pylori strains revealed that Hpn is a key player in nickel homeostasis, as well as in the control of urease activity.The basal urease activity of the Δhpn mutant is similar to that of the wild-type strain, in contrast no increase in urease activity of this mutant was observed under conditions of high exogenous nickel.This suggests that Hpn is involved in activation of the cellular urease pool, when nickel increases.Given that Hpn is a nickel-binding protein involved in protection against metal overload, it is tempting to assign a direct role of Hpn in nickel delivery to urease or urease maturation machinery.Our two-hybrid experiments suggest that Hpn may achieve this role through direct interaction with the structural protein UreA, but not with the UreE-F-G-H accessory proteins.In bacteria, nickel incorporation into urease requires UreE that acts as a nickel chaperone [9].Compared to other bacteria, the H. pylori UreE has a lower nickel-binding capacity (1 Ni(II) ion per monomer) and a poor content in His residues [57] [58].These molecular features might limit the overall nickel incorporation capacity of the urease maturation machinery, when intracellular nickel concentration increases.Based on our enzymatic and two-hybrid experiments, we propose the following model illustrated in Fig 9 .Hpn would directly interact with urease (UreA) and increase the nickel incorporation capacity of the maturation machinery, thereby compensating for this limitation.We observed that Hpn also interacts with HypA and HypB (Fig 7), proteins that have been both reported to participate in nickel incorporation into both urease and hydrogenase [19], which could represent an alternative and/or additional pathway influencing the nickel incorporation capacity.
The role of Hpn-2 is more intriguing.Our data revealed that Hpn-2 restricts basal urease activity, since the Δhpn-2 mutant presented enhanced urease activity under low nickel conditions, whereas at higher nickel concentrations the mutant behaved like a wild type strain (Fig 6).This phenotype depends on the presence of Hpn.Indeed, the Δhpn mutation was found to be dominant over the Δhpn-2 mutation in the ΔhpnΔhpn-2 mutant suggesting that the two proteins are involved in a common pathway of nickel transfer to urease (Fig 6).In the Δhpn-2 mutant, urease activity restriction under low nickel conditions could be due to a yet unknown, direct or indirect, regulatory or physiological mechanisms.However, since our twohybrid experiments provided evidence for a direct interaction of Hpn-2 with Hpn (but not with UreA), and also with HypA and HypB, we propose the following model (Fig 9).This model takes into account these interactions and the fact that Hpn and Hpn-2 were shown to have similar affinities for nickel, but to differ in their metal binding capacity (5 and 2 nickel ions per monomer respectively) and the fact that Hpn is much more abundant than Hpn-2 in H. pylori.When nickel availability is low, the abundant Hpn proteins would not be saturated with nickel.Since Hpn and Hpn-2 strongly interact, we propose that competition for nickel through direct interaction would occur, resulting in limited transfer of nickel from Hpn to urease, thereby restricting its activation.At higher nickel concentrations, the nickel-bound Hpn pool would strongly increase, while Hpn-2 might readily become saturated with nickel because of its lower abundance.Under these conditions, competition with Hpn for nickel would become negligible and would fail to impact urease activation.However, given the importance and complexity of nickel trafficking in H. pylori, we cannot exclude the participation of additional proteins responding to nickel, this will be the subject of future investigations.

In vivo essentiality of Hpn and Hpn-2
Our results reveal for the first time that Hpn and Hpn-2 proteins are individually required for full colonization of mice by H. pylori.This was demonstrated with three different H. pylori mouse-adapted strains (B128, X47-2AL and SS1).Complementation of the mutations by wild type hpn or hpn-2 genes restored full colonization.In addition, the Δhpn-Δhpn-2 double mutants were totally unable to establish infection.Our results revealed that Hpn and Hpn-2 are involved in nickel storage and regulation of urease activity.The essentiality of these proteins during gastric colonization is possibly associated to the higher solubility of nickel under acid conditions and the strict dependence of H. pylori on high and tightly controlled urease activity for acid resistance.We cannot exclude that these proteins have additional essential functions in vivo.Further work is required to explore this possibility.
Our results contrast with those of Benoit et al [59] who reported wild type colonization of the mouse model by a Δhpn-Δhpn-2 double mutant and observed an anecdotic 3-fold lowering of colonization loads in mice that were fed with a Ni-deficient chow.Unfortunately, this paper does not mention the mouse lineage used for colonization and the H. pylori mutants were not further characterized.Given the lack of information in the other study, the reason for the discrepancy between our results consequently remains enigmatic, one possibility might be the difference in mouse lineage and/or in diet.
Our data are consolidated by the fact that we obtained consistent results with well-characterized mutants in three different genetic backgrounds as well as restoration of the observed phenotypes in complemented mutants.

Evolutionary history of Hpn and Hpn-2, two proteins required for in vivo colonization
The experimental data presented here support the existence of a nickel trafficking pathway in H. pylori.Our phylogenomic analysis showed that hpn and hpn-2 genes, encoding two key players of this pathway, are present in every available H. pylori complete genome as well as in H. acinonychis.These two genes are thus part of the H. pylori and H. acinonychis core-genome underlining their important role for these organisms.hpn-2 gene is restricted to these two species.In contrast, hpn genes were predicted in every available genome of gastric Helicobacter species, whereas it was absent from the genomes of enterohepatic Helicobacter species ( Model of the role of Hpn and Hpn-2 in nickel trafficking H. pylori.In H. pylori, nickel is transported across the outer membrane by FrpB4, a TonBdependent transporter.Once in the periplasm, uptake through the inner membrane occurs through the NixA permease (green).In the cytoplasm, nickel can follow several pathways, via the UreE/F/G/H or HypA/B accessory proteins dedicated to urease and/or hydrogenase maturation, the HspA co-chaperonine and Hpn or Hpn-2.(A) Under low nickel concentrations, and because Hpn is abundant, the Hpn cellular pool is not saturated with nickel, while Hpn-2, which is less abundant, can compete with Hpn (similar affinities for nickel).Nickel transfer from Hpn to the urease maturation machinery is low.This results in the "basal" urease activity level monitored in wild type H. pylori cells.(B) Under high nickel conditions, the Hpn protein pool stores more nickel and becomes saturated with this metal.Since Hpn-2 is less abundant, it rapidly becomes outcompeted.Under such conditions, interactions between Hpn and urease and/ or hydrogenase maturation proteins enables Hpn-stored nickel transfer towards urease, leading to an increase in nickel incorporation, and enhanced urease activity.doi:10.1371/journal.ppat.1005312.g009(S2 Fig) .We propose that these motifs are functionally important, as cysteine pairs are often involved in metal binding [30].Our survey of the genomic regions encompassing hpn and hpn-2 highlighted several chromosomal inversions of a gene cluster composed of 4 genes (cluster 1 in Fig 3).Depending on the orientation of this gene cluster, hpn is either neighbor to hpn-2, with both genes located on the same strand, or hpn and hpn-2 surround this cluster and are located on different strands.Some examples of DNA duplication associated with inversions have previously been reported in H. pylori [60], which could account for a putative mechanism explaining our observations.We consequently hypothesize that an inversion event has triggered the duplication of hpn that gave rise to its paralogue, hpn-2.In accordance with a more recent emergence of Hpn-2, following a duplication event, expansion of Gln residues occurred unequally in both the N and C-terminal regions of Hpn-2 in H. pylori and H. acinonychis, whereas Hpn has remained nearly unchanged.Because of the unusual amino acid composition of these proteins, the corresponding genes escaped currently used automated-annotation procedures and were not annotated in most Helicobacter genomes.Using a cutting edge technology, namely top-down proteomics, we confirmed the expression of these genes in several gastric Helicobacter species.Together with the in vivo essentiality of Hpn and Hpn-2, our observations indicate that the acquisition of Hpn in the common ancestor of gastric Helicobacter has been a decisive step for their adaptation to the human stomach, a niche that no other bacterium colonizes and in which metals are more soluble than in the intestine.Finally, we hypothesize that Hpn-2 provides an additional advantage for gastric colonization.

Conclusion
Our data suggest that during evolution of the Helicobacter genus, acquisition of Hpn by gastric Helicobacter species was a decisive evolutionary event to allow Helicobacter to colonize the hostile gastric environment and was important for the emergence of one of the most successful bacterial pathogens, H. pylori.
Since no nickel containing proteins have been described in humans, nickel trafficking, and in particular Hpn and Hpn-2 proteins, constitute privileged targets for the development of novel anti-Helicobacter drugs [17].In support of that, a pilot study has shown that, in H. pylori-infected patients, a nickel-free diet during one month enhanced the efficiency of the standard Helicobacter triple eradication therapy [61].

Ethics statement
Experiments in mice were carried out in strict accordance with the recommendations in the Specific Guide for the Care and the Use of Laboratory Animals of the Institut Pasteur, according to the European Directive (2010/63/UE) and the corresponding French law on animal experimentation (Arrêtés de 1988).The protocol has been approved by the Committee of Central Animal Facility Board of the Institut Pasteur.To follow the new European directives, the project was approved by the CETEA, Comité d'éthique en Expérimentation Animale of the Institut Pasteur (#2013-0051) and submitted for final approval to the Ministère de l'Enseignement Supérieur et de la recherche (#00317).

Molecular techniques
Molecular biology experiments were performed according to standard procedures [71] and the supplier (Fermentas) recommendations.NucleoBond Xtra Midi Kit (Macherey-Nagel) and QIAamp DNA Mini Kit (Qiagen) were used for plasmid preparations and H. pylori genomic DNA extractions, respectively.PCR were performed either with Taq Core DNA polymerase (MP Biomedicals), or with Phusion Hot Start DNA polymerase (Finnzymes) when the product required high fidelity polymerase.The PCR8/GW/TOPO TA cloning kit (Invitrogen) was used to construct in E. coli, the H. pylori suicide plasmids that served for mutagenesis in H. pylori.

Construction of H. pylori mutants
Chromosomal deletions of the entire genes encoding Hpn and Hpn-2 were performed in strains B128, X47-2AL and SS1.Despite strong conservation of these genes with their homologues from H. pylori strain 26695 (gene number hp1432 for hpn and hp1427 for hpn-2 [26]), the corresponding regions in the genomes of strain B128, of its close derivative B8, strain X47-2AL and SS1 were surprisingly wrongly annotated [27,28,65] (for SS1, the sequence of the complete genome was kindly communicated to us by Dr. R. Ferrero).Deletions were introduced by allelic exchange using H. pylori suicide plasmids (see S3 Table) derived from PCR8/GW/TOPO in which about 600 bp of the 5'-end and the 3'-end regions immediately flanking the open reading frame of the target gene and an antibiotic resistance cassette (non-polar kanamycin cassette or non-polar chloramphenicol cassette [72]) were cloned using PCR fragments generated with the primers indicated in S4 Table .The mutants of H. pylori were obtained by natural transformation as described previously [73], with 1 μg of a preparation of the suicide plasmid DNA and selection on the corresponding antibiotic.The Δhpn Δhpn-2 double mutant was constructed in two steps: inactivation of hpn with the non-polar chloramphenicol cassette followed by the deletion of hpn-2 and replacement with the non-polar kanamycin cassette.Deletion of the genes of interest and correct insertion of cassettes were verified by PCR and sequencing of the gene region.
To complement the Δhpn::kan and Δhpn-2::kan mutants in SS1 and B128 strains, we used a strategy described by Langford et al [74] to introduce the wild type hpn and hpn-2 genes into the chromosome.This strategy was preferred to the classical plasmid-complementation to ensure that the wild type copies were lost in the absence of selection pressure during mouse stomach colonization.This system uses the pIR203C04 suicide plasmid vector that contains fragments of an H. pylori strain 26695 intergenic region (hp0203-hp0204) flanking a chloramphenicol acetyltransferase cassette (cat) conferring chloramphenicol resistance and a multiplecloning site.Insertion into this region has been shown by Langford et al [74] to affect neither in vitro growth nor mouse colonization.To ensure that the genes were actually transcribed, we first introduced the PureI promoter into the pIR203C04 plasmid by cloning a PCR fragment obtained by amplification from the chromosome of B128 using the couple of primer PureIUP/ PureIDO and subsequent digestion with PstI and BamHI (for primers see S4 Table ).Primer PureIDO introduced an EcoRV site at the 3' end of the PureI promoter.The hpn and hpn-2 genes were then amplified from the B128 chromosome (PCR using the couple of primer hpnUP/HpnDO or Hpn-2UP/Hpn-2DO) and cloned into the EcoRV/PstI sites of the pIRC (PureI) plasmid, allowing the transcription of the genes from the PureI promoter.The p (PureI::hpn) and p(PureI::hpn-2) plasmids were used to transform strain 26695 selecting for chloramphenicol resistance.We verified the presence of the wild type copy of the hpn or hpn-2 genes at the correct location by sequencing.In the figures, strains labeled as "+ pureI" are controls in which only the PureI promoter is inserted.The chromosomal DNAs from correct chloramphenicol resistant clones (i.e.where the wild type genes and the cat cassette integrated by homologous recombination between hp0203 and hp0204) were then used to introduce the hpn and hpn-2 genes into various strains by natural transformation and selection for chloramphenicol resistance.

Mouse model of colonization
NMRI-specific pathogen-free mice (Charles River Laboratories) were orogastrically inoculated with 10 9 CFU of H. pylori strains prepared in 100 μL of peptone broth.Several strains were tested (i) the wild type B128 parental strain and its isogenic Δhpn, Δhpn-2 or Δhpn Δhpn-2 mutants, (ii) the wild type X47-2AL strain and its isogenic Δhpn, Δhpn-2 mutants and (iii) the wild type SS1 parental strain, its isogenic Δhpn, Δhpn-2 or Δhpn Δhpn-2 mutants and the two complemented strains Δhpn/p(PureI::hpn) and Δhpn-2/p(PureI::hpn-2).Each strain was used to inoculate a group of seven mice and the experiments were reproduced twice for SS1 and B128 strains and their mutants.As described in [72], one month after inoculation, mice were sacrificed and stomachs were crushed in peptone broth.Viable H. pylori colonizing the stomach were enumerated by culture of serial dilutions of homogenized tissue on blood agar plates containing bacitracin (200 μg.mL -1 ) and nalidixic acid (10 μg.mL -1 ).

Measurement of metal sensitivity by disc diffusion and growth assays
H. pylori strains were tested for their sensitivity to nickel by the disc agar diffusion method.Therefore, 2.10 6 CFU of each strain were prepared from liquid culture and were spread on plates containing Brucella broth, 1.5% bacto agar and 10% FCS.Sterile antibiotic assay discs (Whatman) were soaked in 4 μmol NiCl 2 (Sigma) and placed in the center of the plate.The data correspond to at least three independent experiments with two replicates per experiment.Growth inhibition diameters (zones of inhibition, ZOI) were measured after 5 days of culture.For tests of metal sensitivity during growth in liquid, H. pylori cells were inoculated at OD 600 0.05 in 10 mL Brucella-Broth containing 10% FCS and increasing nickel concentrations (0, 5, 10, 30, 100 or 200 μM).Bacterial growth was monitored 17 hours later by measuring OD 600 .The MIC 50 corresponds to the inhibitory nickel concentration that reduced growth by 50%.

Nickel content measurements by Inductively Coupled Plasma Emission Spectrometry (ICP-OES)
Overnight liquid cultures of H. pylori strain were grown until OD 600 0.9 at 37°C in 6 ml Brucella broth (pH adjusted to 7), then 200 μM NiCl 2 were added and the cultures were left to grow until OD 600 6.Then, the 6 ml of culture were centrifuged at 4000 g at 4°C for 25 min through 400 μL of a 1:2 mixture of the silicone oils AR20/AR200 (Wacker) in order to separate the cells from the medium.Cells were lysed with 400 μL 0.2 M NaOH/1% SDS for 60 min at 95°C.Samples were calibrated by protein concentration measurements with the DC Protein Assay kit (BioRad).Then, the samples were mineralized overnight at 80°C with 300 μL of ultrapure 70% nitric acid (JT Baker) and diluted to 1/20 in ultrapure water.Nickel contents were measured by ICP-OES using a Vista MPX spectrometer (Varian).The content of Ni(II) was determined using a curve established with certified ICP grade standards.The measurement of each strain in each condition was performed in triplicates in at least three independent experiments.

Urease activity measurements
Urease activity of whole H. pylori cells was tested by measuring the ammonia production using the Ammonia-Assay Kit (Sigma) as described [75].H. pylori bacteria grown on blood agar plates for 24 hours were inoculated at OD 600 0.1 in BBß liquid medium and grown overnight.This preculture was used to inoculate the bacteria at OD 600 0.2 in BBß liquid medium without added nickel or supplemented with 10 μM of NiCl 2 .From this culture, log-phase bacteria (OD 600 0.5) were harvested and washed once with phosphate-buffered saline (Sigma).5.10 7 CFU of bacteria were resuspended in 1 mL of buffer (citric acid, 0.1M; Na 2 HPO 4 , 0.2 M pH 7) containing 5 mM urea.The data correspond to at least three independent experiments with two technical replicates each time.Aliquots were taken after 10 min of incubation at 37°C and centrifuged to pellet the bacteria.The NH 3 concentration in the supernatant was measured immediately with the ammonia-assay kit according to the manufacturer's (SIGMA) recommendations.This assay is based on the following reaction: in the presence of NH 3 , α-ketoglutaric acid and NADPH, the enzyme glutamate dehydrogenase produces glutamate and NADP + .The oxidation of NADPH to NADP + results in a decrease in the absorbance at 340 nm that is proportional to the concentration of NH 3 .One unit (U) of urease activity was defined as the amount of enzyme that hydrolyzes 1 μmol urea per min per mg of total proteins.

Purification of nickel-binding proteins
To identify the concentration range of imidazole required for the elution of Hpn and/or Hpn-2 after binding to the Ni-NTA agarose column, we first tested the procedure on E. coli BL-21 cells harboring plasmids overexpressing either Hpn or Hpn-2.To this aim, we constructed plasmids pRSFDuet1-hpn and pRSFDuet1-hpn-2 (S3 and S4 Tables).pRSFDuet1-hpn is a derivative of vector pRSFDuet1 (Novagen) in which hpn was cloned into MCS1 between the NcoI and HindIII restriction sites.pRSFDuet1-hpn-2 is a derivative of the same vector in which hpn-2 was cloned into MCS2 between NdeI and KpnI restriction sites.Both plasmids were transformed into E. coli BL-21 strain (Novagen) and transformants were selected on LB-Kanamycin.These strains were grown overnight in LB medium at 37°C and then diluted to OD 600 0.1 in 50 mL fresh LB medium containing the appropriate antibiotic.Cultures were grown to OD 600 0.6 and expression of Hpn or Hpn-2 proteins was induced by addition of 0.1 mM IPTG during 3 hours at 37°C.Cells were then harvested by centrifugation (10,000xg, 15 min, 4°C), washed in PBS, and resuspended in buffer 1 (Tris-HCl 25 mM, pH 7.5, containing NaCl 300 mM, Imidazole 5 mM, 2-mercaptoethanol 5 mM and proteases inhibitors).Both Hpn and proteins remained bound on Ni-NTA agarose with concentrations of 100 mM imidazole and were eluted with 500 mM imidazole.Assuming that Hpn and Hpn-2 from the other gastric Helicobacter species would behave similarly, the same protocol was applied to extract the nickel-binding proteins from the five cultured gastric Helicobacter strains before their analysis by LC-MS/MS.Gastric Helicobacter strains were grown in liquid to an OD 600 of 1.5-2.0 and diluted to an OD 600 of 0.05 (H.pylori, H. acinonychis, H. mustelae) or 0.2 (H.bizzozeronii and H. felis) in fresh Brucella broth.Cells were grown to an OD 600 of 1.5 and recovered by centrifugation, washed in PBS, and resuspended in buffer 1.Then, cells were disrupted by sonication, and debris were removed by centrifugation (20,000xg, 30 min at 4°C).The supernatant was recovered was incubated 1h at 4°C with gentle shaking after addition of 0.5-1 mL of Ni-NTA agarose beads (Protino Ni-NTA agarose, Marcherey-Nagel).Extracts were then loaded onto 10 mL poly-prep chromatography columns (Bio Rad) and left at 4°C until all the beads pelleted.The beads were washed with 5 volumes of buffer 1, followed by a 5 volumes wash with buffer 2 (Tris-HCl 25 mM, pH 7.5, containing NaCl 300 mM, Imidazole 25 mM).Elution of Ni-binding proteins started with a 5-volume wash with buffer 3 (Tris-HCl 25 mM, pH 7.5, containing NaCl 300 mM, Imidazole 100 mM) in a first step, and then with a 3-volume wash with buffer 4 (Tris-HCl 25 mM, pH 7.5, containing imidazole 500 mM without NaCl or 2-mercaptoethanol).Aliquots of 0.5 mL were collected during the two last steps, and samples were kept at -20°C until use for LC-MS/MS analyzes.Presence of proteins was assessed by SDS-PAGE 14%.

LC-MS/MS analysis
A nanoflow ultra-high pressure liquid chromatography Dionex Ultimate 3000 RSLCnano (Thermo Fisher) coupled to an Orbitrap Fusion Tribrid Mass Spectrometer (Thermo Fisher San Jose, USA) was used for all LC-MS/MS analyses.Fractions enriched in Ni-binding proteins were acidified with formic acid (~1% by volume) and loaded on a trap column (Dionex Pep-Map300 C 4 5 μm, 5 mm x 300 μm).Intact proteins were further separated on an Easy-Spray analytical column (Thermo Scientific PepSwift monolithic PS-DVB, 25 cm x 200 μm,).Solvent A consisted of 0.1% aqueous formic acid, solvent B consisted of 80% acetonitrile with 0.1% of formic acid.Sample loading was performed with a flow rate of 15 μL/min during 3 min, then a flow rate of 1 μL/min was used for protein separation on the analytical column.A 25 min (for targeted acquisitions) or 60 min (for data dependent analyses) gradient was used with the following conditions: from 1% to 50% solvent B in either 25 min or 60 min, then to 90% B in 1 min.Column compartment and analytical column were kept at 60°C.The "intact protein" mode was enabled on the Orbitrap Tribrid Fusion mass spectrometer.Tandem mass spectrometry experiments were performed with a resolution of 60,000 for MS 1 and 120,000 MS 2 , averaging 3-6 microscans for MS 1 and several microscans for MS 2 (20 microscans for H. mustelae and H. bizzozeronii and 5-8 for the other strains).Fragmentation was performed either with HCD (collision energy 25 eV) or with ETD (8 ms reaction time, 2.0E5 reagent AGC target).The precursors were selected for fragmentation either in data-dependent mode (4 most intense ions, excluding charge states 1-5, 180 s dynamic exclusion) or in a targeted mode with an inclusion list of the theoretical mass-to-charge ratios of Hpn and Hpn-2 (with charge states between 7 and 13).
Data processing was accomplished using Thermo Scientific ProSightPC 3 software.Raw files were firstly deconvoluted with the Xtract algorithm.A three-stage database search was then performed for each strain against the corresponding Uniprot databases re-implemented with the sequences of Hpn and Hpn-2.The "absolute mass" mode was initially used with a precursor ion tolerance window of 4.08 Da (to for potential S-S bonds) and 5 ppm for fragments.Then the "biomarker" mode was used with the same parameters to search for truncated proteoforms.Finally, an "absolute mass" search enlarging the precursor window to 10 kDa was used to search for all post-translationally modified proteoforms.Results are summarized in Table 2.

Bacterial Two-Hybrid tests
The Bacterial Two-Hybrid (BACTH) test is based on the reconstitution of adenylate cyclase activity in a cya -E. coli strain as a result of the interaction between two proteins: a bait and a prey fused to two separate catalytic domains (T18 and T25) of the Bordetella pertussis adenylate cyclase.DNA inserts encoding the proteins of interest were obtained by PCR using the primers listed in S4 Table and chromosomal DNA from H. pylori B128 strain as a template.To detect interactions between the proteins of interest, several plasmids were constructed (S3 Table ) expressing either a N-terminal or a C-terminal fusion of these proteins with the T25 catalytic domain (derived from vectors pKNT25 and pKT25, respectively) or either a N-terminal or a C-terminal fusion with the T18 catalytic domain (derived from vectors pUT18 and pUT18C, respectively).All insert were digested by EcoRI and PstI then cloned into plasmids pUT18 and pKNT25 (PCR with primers UT18UP/UT18DO), pUT18C (PCR with primers UT18cUP/ KT25DO) and pKT25 (PCR with primers KT25UP/KT25DO, S4 Table ).The two plasmids expressing fusions to be tested were co-transformed in E. coli strain BTH101.The screening procedure involved growth of the transformants spread on Luria-Bertani agar plates containing kanamycin and ampicillin plus X-Gal with or without IPTG at a final concentration of 0.4% and 0.1 mM, respectively and incubated at 30°C for 2 days.Five mL of LB medium supplemented with antibiotics and IPTG 5x10 -4 M were inoculated with the transformants clones and incubated overnight at 30°C.Quantification of the interactions in strains carrying each plasmid combination was obtained by measurement of the β-galactosidase activity performed in 10 replicates as described in [64] (S2 Table ).
Negative controls corresponded to strains with plasmids expressing one of the fusion proteins tested again the empty vector used to express the other partner (S2 Table ).Positive controls consisted in a strain containing the plasmid used as a vector to construct the baits and the preys but carrying a GCN4 leucine zipper domain fused with the T18 or T25 domains (Zip-Zip construct in S2 Table ).

Immuno blotting
Overnight H. pylori B128 parental strain, Δhpn, Δhpn-2 and Δhpn Δhpn-2 mutants grown in BBß were diluted to OD 600nm 0.2 in fresh BBß either without added nickel or with 10 μM NiCl 2 .Aliquots were taken after 4, 8 and 24 hours culture.Bacteria were pelleted and frozen.Immuno blotting was performed on crude extracts after cell lysis in 1% SDS pH 5 according to standard protocols.Protein amounts in the crude extracts were calibrated using the Bradford DC Protein Assay (Biorad) with bovine serum albumin (BSA) as a standard.20 μg of total extracts were separated by 12.5% SDS-PAGE using the Criterion TGX Stain Free Any kD acrylamide gels (BioRad).Gels were blotted on a nitrocellulose membrane (GE healthcare).The H. pylori UreA protein was specifically detected with an anti-ureA rabbit polyclonal antibody used at a dilution of 1:5000 [76].Goat anti-rabbit IgG-HRP (Santa Cruz) were used as secondary antibodies and the detection was achieved with the ECL reagent (Pierce).Band intensity was quantified with a Pharos imager using Image Quant Software (Molecular Dynamics).
pylori and H. acinonychis (Fig 2 and S1 Fig).To further investigate the evolution of hpn and hpn-2 and determine whether horizontal gene transfer had affected their evolutionary history, we selected a subsample of 100 Helicobacter strains (i.e. 14 enterohepatic strains and 85 gastric strains, including 75 H. pylori) representative of the diversity of the 330 strains (Fig 2 and S1 Fig).

Fig 2 .
Fig 2. Maximum Likelihood phylogeny of a subsample of 100 out of the 330 Helicobacter strains analyzed in this study.The enterohepatic and gastric strains are shown in gray and black, respectively.The tree is based on the 281 protein families present in a single copy in at least 320 out of the 330 strains (82,743 amino acid positions).Numbers at nodes represent bootstrap supports (for clarity only values > 50% are shown).The scale bar indicates the average number of substitutions inferred by site.Orange and brown rectangles show the presence of hpn and hpn-2 in the corresponding strains, while orange and brown arrows indicate the origin of these genes.Strains harboring similar genomic contexts of hpn and hpn-2 are indicated with similar color.Strains analyzed in this study are marked by an orange asterisk.Precise loci organizations are shown in Fig 3.

Fig 3 .
Fig 3. Genomic organization of the region encompassing hpn and hpn-2.Conserved homologous genes are shown with similar colors, hypothetical proteins are shown in gray, while non-conserved genes are in white.Hpn (in red) is located in-between two conserved cluster of genes (clusters 1 and 2), while hpn-2 is located either near hpn on the same strand or on the opposite side of cluster 1 on the opposite strand.Helicobacter strains harboring similar genomic organization are listed on the left with similar colors.doi:10.1371/journal.ppat.1005312.g003

2 - 4 -
Experimental monoisotopic mass.3 -At 5 ppm matching tolerance.Expectation value, calculated by ProSightPC software.The expectation value (e value) is the number of sequences in a database that are expected to have P-scores equal to or better than what was observed simply by chance.A P-score is the probability of obtaining at least as good a match between the observed fragment list and a sequence as by chance.Low E-values represent better matches (less likely to be false positives) than high E-values.5, 6, 7, 8 -forms of the protein with N-terminal Met retained (5) or N-terminal NF (6), NFS (7) or NFSTN (8) residues cleaved.doi:10.1371/journal.ppat.1005312.t002

Fig 4 .
Fig 4. Effect of nickel on the growth of H. pylori B128, SS1 and X47-2AL wild type strain and isogenic mutants.The strains with pureI are controls in which only the pUreI is inserted at the locus at which the hpn or hpn-2 genes are introduced in the complemented strains Δhpn + phpn and Δhpn-2 + phpn-2.Panel A, nickel disk diffusion assay and sensitivity to 200 μM NiCl 2 in liquid medium.Panel B, nickel sensitivity in liquid medium (range of NiCl 2 concentrations) of mutants and MIC 50 for the B128 Δhpn, Δhpn +phpn and Δhpn-Δhpn-2 strains.The data correspond to the mean value of three independent experiments with at least triplicate tests for each strain.Error bars represent the standard deviation.doi:10.1371/journal.ppat.1005312.g004

Fig 5 .
Fig 5. Intracellular nickel content of H. pylori B128 wild type strain, isogenic mutants and complemented strains.The identity of each strain is indicated below the bars.Nickel amounts were measured by Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES) and expressed in μg of nickel.mg of proteins -1 .The data correspond to the mean value of two independent experiments with at least triplicate tests for each strain.Error bars represent the standard deviation.*** indicates that the mean value is significantly different from that of the wild type strain (P < 0.01).doi:10.1371/journal.ppat.1005312.g005

Fig 6 .Fig 7 .
Fig 6.Urease activity of H. pylori B128 wild type strain, isogenic mutants and complemented strains.Strains were grown in Brucella β-cyclodextrin medium without added nickel (grey bars) or with 10 µM nickel (black bars).The identity of each strain is indicated below the bars.Urease activity is expressed as μmol of urea hydrolyzed.min -1 .mg -1 of protein.The data correspond to the mean value of three independent experiments with at least triplicate tests for each strain.Error bars represent the standard deviation.*** indicates that the mean value is significantly different from that of the wild type strain (P < 0.01).doi:10.1371/journal.ppat.1005312.g006

doi: 10 .Fig 8 .
Fig 8. Hpn and Hpn-2 are required for full mouse colonization by H. pylori SS1, B128 and X47-2AL strains.Each point corresponds to the colonization load for one mouse, with the strain indicated below.Each strain was tested in a group of seven mice.Horizontal bars represent the geometric means of the colonization load for the wild type and each mutant.The detection limit is shown by a dashed horizontal line.*** and ** indicate that the geometric mean is significantly different from that of the wild type strain (P < 0.01 and P < 0.05, respectively).doi:10.1371/journal.ppat.1005312.g008 Fig 9.Model of the role of Hpn and Hpn-2 in nickel trafficking H. pylori.In H. pylori, nickel is transported across the outer membrane by FrpB4, a TonBdependent transporter.Once in the periplasm, uptake through the inner membrane occurs through the NixA permease (green).In the cytoplasm, nickel can follow several pathways, via the UreE/F/G/H or HypA/B accessory proteins dedicated to urease and/or hydrogenase maturation, the HspA co-chaperonine and Hpn or Hpn-2.(A) Under low nickel concentrations, and because Hpn is abundant, the Hpn cellular pool is not saturated with nickel, while Hpn-2, which is less abundant, can compete with Hpn (similar affinities for nickel).Nickel transfer from Hpn to the urease maturation machinery is low.This results in the "basal" urease activity level monitored in wild type H. pylori cells.(B) Under high nickel conditions, the Hpn protein pool stores more nickel and becomes saturated with this metal.Since Hpn-2 is less abundant, it rapidly becomes outcompeted.Under such conditions, interactions between Hpn and urease and/ or hydrogenase maturation proteins enables Hpn-stored nickel transfer towards urease, leading to an increase in nickel incorporation, and enhanced urease activity.

Table 1 .
List of the gastric and enterohepatic Helicobacter species analyzed in this study.

Table 2 .
Identification of Hpn and Hpn-2 proteins in samples prepared from different Helicobacter species by top-down proteomics.