The molecular basis of the regulation of specific shapes and their role for the bacterial fitness remain largely unknown. We focused in this study on the Gram-negative and spiral-shaped Helicobacter pylori. To colonize its unique niche, H. pylori needs to reach quickly the human gastric mucosa, by swimming to and through the mucus layer. For that reason, the specific shape of H. pylori is predicted to be necessary for optimal motility in vivo, and consequently for its colonization ability. Here, we describe the involvement of a PG-modifying enzyme, HdpA (HP0506), in the mouse colonization ability of this bacterium, by regulating its shape. Indeed, the inactivation of the hp0506 gene led to a stocky and branched phenotype, affecting H. pylori colonization capacity despite a normal motility phenotype in vitro. In contrast, the overexpression of the hp0506 gene induced the transformation of H. pylori from rod to dividing cocci shaped bacteria. Furthermore, we demonstrated by PG analysis and enzymology, that HdpA carried both d,d-carboxypeptidase and d,d-endopeptidase activities. Thus, HdpA is the first enzyme belonging to the M23-peptidase family able to perform the d,d-carboxypeptidation and regulate cell shape.
Since the discovery of microbes by Antonie van Leeuwenhoek, the different shapes of bacteria have puzzled us. As discussed in a recent review (Young, 2006), the existence of a large variety of shapes raises many hypotheses relative to their specific biological value, from the nutrient uptake, to motility and predation for example. In fact, the mechanisms by which bacteria are able to select a particular morphology have only started to be dissected the last couple of years. From the work of Alexander Fleming both on lysozyme and penicillin G, researchers have highlighted the importance of the bacterial cell wall in conferring a specific cell shape. Indeed, purified cell walls keep the shape of the original bacterium. This cell shape and rigidity are given by the major cell wall component, the peptidoglycan (PG), a heteropolymer composed of glycan chains cross-linked by short and flexible unique stem peptides.
In the last 30 years, a collection of evidence from many groups has shown that both enzymes involved in PG metabolism such as the penicillin-binding proteins (PBPs), as well as cytoskeletal proteins (i.e. FtsZ, MreB, the Min system, etc.), are required for cell shape maintenance. However, these studies have been mainly restricted to few bacterial models such as Escherichia coli, Bacillus subtilis or Caulobacter crescentus. In particular, very few studies have addressed the selective value of bacterial shape in their natural niche and how bacteria with altered shape survive in their natural habitat. Some studies have shown that bacteria can undergo significant morphological changes during colonization of their niches. For example, E. coli strains responsible for urinary track infection are found as intracellular microbiofilms of cells with ovoid/coccoid shape (Anderson et al., 2003). In contrast, these same bacteria undergo filamentation during epithelial cell exit into the bladder (Justice et al., 2004). Recently, the E. coli MHD79 mutant lacking six lytic transglycosylases, which grows as small chains of 6–8 bacteria, was shown to be able to colonize the gut of germ-free mice as well as its parental strain (Bouskra et al., 2008). However, very few studies have addressed the impact of morphology on the ability of a bacterium to occupy its niche.
We have chosen to study the role of the bacterial shape and in particular the impact of the PG metabolism in Helicobacter pylori, a member of the ε-proteobacteria. H. pylori is an attractive alternative model because of its compact genome and reduced redundancy of genes involved in PG metabolism (Tomb et al., 1997). For example, genome analysis revealed the presence of only three PBPs (PBP1, 2 and 3) and three clearly identifiable PG hydrolases (Slt, MltD and AmiA). We have shown that Slt and MltD are both non-redundant lytic transglycosylases (Chaput et al., 2007). Furthermore, H. pylori is found under different shapes from spiral/rod to coccoid forms making it a fascinating model to study the interplay between PG metabolism and cell shape. In particular, it is believed that the spiral/rod shape gives a selective advantage for this motile bacterium to colonize its unique niche, the human stomach, favouring movement in the viscous mucus layer. The coccoid form has been proposed to be a persistence form of H. pylori. Accordingly, we have shown that the morphological transformation of H. pylori into coccoid bacteria is regulated by the AmiA protein (Chaput et al., 2006). Interestingly, the morphological transition of H. pylori is associated to a modification of the PG composition leading to an escape of the host immune response.
Previous work has highlighted the unique feature of H. pylori PG characterized by a very low carboxypeptidase activity. The predominant fraction of the muropeptides keeps the pentapeptide intact when compared with other bacterial models such as E. coli (Costa et al., 1999). Indeed, H. pylori genome has genes encoding only high-molecular-weight PBPs (PBP 1, 2 and 3) and remarkably lacks low-molecular-weight PBPs, which usually function as d,d-carboxy- and d,d-endopeptidases. Intriguingly, those low-molecular-weight PBPs were shown to regulate the pole formation in E. coli, and we wondered how H. pylori defined its poles. Hence, we searched for new enzymes involved in the PG metabolism of H. pylori.
In this work, we characterize a novel PG-modifying enzyme belonging to the M23-family of zinc-metallopeptidases. In particular, we focus this work on one member of this family, HP0506. We show that HP0506 carries both d,d-carboxy- and d,d-endo-peptidase activity. We show that HP0506 is important for the cell shape maintenance and that it is required to define pole formation of H. pylori. Deletion of hp0506 leads to bacteria with multiple poles each with fully functional flagella. Most importantly, despite a normal motility phenotype in vitro, the abnormal shape of the hp0506 mutant has a major impact on the ability of H. pylori to colonize mouse gastric mucosa, showing the selective value of its shape for colonizing and surviving in its natural niche.
Characterization of the hp0506 mutant
Looking for missing activities involved in the PG metabolism of H. pylori, we searched for uncharacterized peptidases in its genome. The hp0506 gene (1209 bp) was identified from the MEROPS peptidase database (gene nomenclature from the first sequenced genome, strain 26695). It encodes for the putative periplasmic peptidase HP0506 (45.8 kDa) carrying a cleavable signal peptide located between amino acids 1 and 28. It belongs to the M23B family of the lysostaphin-like β-lytic metallopeptidases. The hp0506 mutant was constructed in the N6 background by allelic replacement with a non-polar kanamycin resistance cassette (N6hp0506ΩKm). First, we compared its growth with that of the parental strain by following the colony-forming units (cfu) per millilitre every 4 h, during 38 h (Fig. 1A). We observed that the mutant presented a slight but significant growth delay compared with the wild-type, suggesting that even if the deletion was not lethal, it probably had a cost for the bacteria in vitro. Nevertheless, generation times of N6 and N6hp0506ΩKm (calculated during the exponential growth phase, hour 0 to 26) were similar (5 h), suggesting that the mutation had a higher impact at the beginning of the growth. Accordingly, we observed that the mutant had a particular morphological phenotype, compared with the wild-type. We measured the length and the width of 50 bacteria for each strain (Fig. 1B). The results showed that the disruption of hp0506 led to a significant decrease of the length (an average of 3.3 µm for the wild-type and 2.3 µm for the mutant), accompanied by a significant increase of the width of the bacteria (an average of 0.55 µm for the wild-type and 0.78 µm for the mutant), leading to a stocky phenotype. This suggested that HP0506 was involved in the control of the cell shape, particularly concerning the regulation of the cell diameter and the length of the bacteria.
Moreover, we also observed that a fraction of the population (between 15% and 40% of the population depending of the growth phase) had a branched and irregular phenotype, which was defined with the emergence of at least one supplementary pole (Fig. 1C). These supplementary poles seemed to be irregular in length and were located at different sites of the bacteria. Furthermore, analysis of both strains by scanning electronic microscopy showed the presence of flagella on these supplementary poles, suggesting that they were functional.
Complementation of the hp0506 mutant
To ensure the link between the deletion of the gene and the morphological phenotype, we wanted to complement the construction. For this, we first constructed a new mutant with a non-polar gentamicin cassette. By this way, we confirmed the phenotype previously described, using another resistance cassette (N6hp0506ΩGm, Fig. 1D). Then, we replaced the complete deletion of hp0506 by the gentamicin cassette with a wild-type copy of hp0506 carrying downstream a kanamycin resistance cassette. Hence, we obtained a complemented strain with a wild-type hp0506, and a non-polar kanamycin cassette inserted between hp0506 and hp0507 (N6hp0506-Km). Pictures taken by phase-contrast microscopy showed the recovery of a wild-type phenotype (Fig. 1D).
Site-directed mutagenesis of the M23 catalytic domain
To determine if the morphological phenotype was linked to the disruption of the activity of HP0506, or rather to the absence of HP0506, we decided to inactivate its activity by site-directed mutagenesis. Based on predictions from the MEROPS peptidases database, we substituted one (H259) of the three predicted amino acids involved in metal binding (H259, D263 and H341 of the M23 domain) with an alanine (N6hp0506*Km; Fig. 1E). This single mutation led to the same stocky and branched phenotype, suggesting that the morphological abnormalities were linked to the disruption of the activity of the protein. However, we cannot exclude that the mutation had also an impact on the protein stability.
Morphology of N6 overexpressing hp0506
To deepen the characterization of HP0506 in cell morphogenesis, we explored the effects of the overproduction of the protein in the same background (N6 strain). We cloned the whole gene into the inducible plasmid pILL2157, allowing a high level of expression (10- to 15-fold induction; Boneca et al., 2008). The constructed plasmid (pMAB1) was introduced into the wild-type N6 strain, generating N6pMAB1. Next, we observed the cell shape after 2 and 4 h of induction, compared with the control grown without IPTG (Fig. 2A). The N6pMAB1 without IPTG did not present any morphological abnormality nor growth defect compared with the wild-type (data not shown). This suggested that the level of expression of hp0506 due to the presence of extra copies of the gene carried by the plasmid was not sufficient to induce morphological modification. In contrast, from 2 h of induction, cells started to transform into round-shaped bacteria, with an intermediate phenotype, suggesting an asymmetrical mechanism of action of HP0506 (half round and half rod-shaped). After 4 h of induction, all bacteria completely adopted a cocci-shape, with irregular sizes. Accordingly, the growth of N6pMAB1 in the presence of IPTG was strongly decreased, although cell viability seemed unaffected (Fig. 2C). We estimated a generation time of 5 h for N6pMAB1 without IPTG and 21 h for N6pMAB1 with IPTG, calculated during the exponential growth phase (hour 0 to 36). Moreover, we also made the observation that cells overexpressing hp0506 were still able to divide, in spite of their abnormal shape. We performed vancomycin-FITC labelling with the aim of visualizing the PG and especially the septa of the bacteria, by specific recognition of the PG muropeptide, the GMpentapeptide. The signal was weaker for the mutant compared with the parental strain, which suggested that the overproduction of HP0506 could affect the pool of GMpentapeptide. Nevertheless, we could observe the presence of septa with or without IPTG, showing cells were capable of dividing, despite the overproduction of the protein (Fig. 2B).
To investigate the biochemical activity of HP0506, we first analysed the muropeptide composition of the mutants. Three independent analyses were performed for each construction described above: N6, N6hp0506ΩKm, N6hp0506*Km and N6pMAB1, with or without inducer. PG was extracted, and then digested with mutanolysin to estimate the relative abundance of each muropeptide by high-pressure liquid chromatography (HPLC) (Fig. 3A and Table 1). The results revealed that the disruption of hp0506, both by total deletion or by site-directed mutagenesis, was linked to an increase of the GMpentapeptide pattern, both in monomeric (peak 5) and dimeric forms (hydrated peak 9 and anhydro peak 13). We also noticed an important reduction of the abundance of GMtetrapeptide pattern, particularly concerning the GMtetrapeptide–tetrapeptideGM (peak 8), the GMtetrapeptide–tripeptideGM (peak 6) and the GanhMtetrapeptide (GanhM4)–tetrapeptideGM (peak 12) forms. In contrast, the overproduction of the protein led to the decrease of the abundance of most muropeptides carrying a GMpentapeptide moiety (peaks 5, 9, 13). Under these conditions, the abundance of muropeptides with GMtetrapeptide moiety (peaks 2, 3, 7) and particularly with GMtripeptide (peaks 1, 6) moiety was increased compared with the wild-type. In addition, the muropeptide composition of the control N6pMAB1 without IPTG presented significant differences compared with the wild-type N6 strain, especially with an increased abundance of GMtripeptide (peak 1) and a decreased abundance of both dimers carrying the pentapeptide moiety (peaks 9 and 13). This suggested that the presence of extra copies of hp0506 led to PG modifications, despite the absence of shape abnormality (Fig. 2A). Furthermore, the results showed modifications of the relative distributions of total monomeric and dimeric muropeptides, with especially an increase of near 30% of the monomers under overexpressing conditions, compared with the parental strain. Altogether, these results strongly suggested that HP0506 carried a d,d-carboxypeptidase activity leading to the generation of GMtetrapeptide from the GMpentapeptide moiety. Moreover, the increase of the monomers (14) after overexpression of HP0506 could also denote the presence of a d,d-endopeptidase activity, able to cleave the cross-link bond of the dimers.
Table 1. Effect of the disruption or the overexpression of the hp0506 gene on the muropeptides composition of the N6 strain.
Enzymatic activity of the recombinant HP0506-His6
To characterize the biochemical activity of HP0506, we purified the recombinant HP0506 with a His6-tag, and without its signal peptide (Fig. 3B). We incubated recombinant HP0506-His6 with different substrates in presence of MgCl2 and penicillin G (1 mM). As a control of this selective inhibitory activity of the penicillin G, we tested the relative affinities of an analogue of the penicillin G linked to Bodipyl FL dye, Bocillin, for HP0506-His6 and for the recombinant PBP5 from E. coli (Fig. 3B). The results confirmed that HP0506-His6 from H. pylori did not bind to Bocillin, in contrast to PBP5 from E. coli, meaning that the penicillin G could be used as a selective inhibitor of potential contaminating low-molecular-weight PBPs from E. coli. Therefore, we used this condition to test the activity of HP0506-His6 on different substrates. First, we showed that HP0506-His6 was able to release the GanhM4 from the GanhMpentapeptide (GanhM5), suggesting it has a d,d-carboxypeptidase activity (Fig. 3C). Second, we observed that the protein could perform the d,d-endopeptidase reaction, by cleaving the cross-link bond of the dimer GanhM5–tetrapeptideGanhM, and releasing the monomers GanhM4 and GanhM5 (Fig. 3D).
Role of HP0506 in the mouse colonization ability
To evaluate the colonization ability of the hp0506 mutant, we decided to use the X47 strain, since the N6 strain does not colonize the gastric mucosa of mice. First, we confirmed that the deletion of hp0506 led to the same morphological phenotype in the genetic background of X47, compared with the N6 strain (Fig. 4A; length and width average of 3.5 and 0.43 µm for X47 and of 2.5 and 0.73 µm for X47hp0506ΩKm respectively). Then, the ability of the mutant to colonize the mouse gastric mucosa was compared with that of the parental strain after 1, 4 and 7 days of infection. Results revealed that from the beginning of the infection, the mutant was strongly affected in its colonization ability (Fig. 4B). Furthermore, this impaired colonization worsened with time, resulting in a decrease of more than four log of the bacterial load per stomach between the mutant and the parental strain, after 7 days of infection. To ensure that this colonization defect was not correlated with a defect of the motility of the mutant, we found that it was as motile as the parental strain (Fig. 4C). Thus, these results showed that the deletion of the hp0506 gene altered the virulence of the bacteria by decreasing its colonization ability, but not its motility in vitro.
In this study, we characterized a peptidase involved in the metabolism of the murein sacculus of H. pylori, HP0506. This peptidase belongs to the M23B family of the lysostaphin-like β-lytic metallopeptidases, mostly composed of endopeptidases. We showed, both with PG analysis and enzymology, that the protein carried not only a d,d-endopeptidase (d,d-EPase) activity but also a d,d-carboxypeptidase (d,d-CPase) activity. For that reason, we decided to rename this protein HdpA, for Helicobacterd,d-peptidase A. Most of the d,d-CPase previously described belong to the S11 or S13 families of D-Ala-D-Ala-peptidase A or C. Hence, HdpA presents unique features compared with others classical d,d-CPases. Indeed, it does not have a transmembrane domain, nor penicillin binding domain, contrary to d,d-CPases which are all PBPs of low molecular weight. It is not surprising to find bifunctional (or multifunctional) enzymes in H. pylori, given the low redundancy of its genome. But the mechanism by which HdpA achieves both activities remains unclear, although, HdpA appeared to function preferentially as a d,d-EPase, i.e. it would cleave preferentially dimers over monomers (see Table 1). Another simple explanation would be that, because of the differential locations of each substrate, HdpA would be able to perform rather the d,d-carboxypeptidation or the d,d-endopeptidation. In all cases, it would be interesting to characterize the crystal structure of the protein alone and in association with its different substrates.
Second, we showed that HdpA was involved in the regulation of the morphology of H. pylori. A recent study (Sycuro et al., 2010) described the implication of the same gene (named csd3, coding for HPG_464), in the shape regulation of the H. pylori G27 strain. The authors showed that csd3 was involved in the regulation of the cell curvature, correlated with a putative bifunctional activity of d,d-CPase/d,d-EPase, predicted from PG analysis. Here, we confirmed the activity of the protein by enzymology, but our mutants presented a distinct morphological phenotype, potentially due to the high degree of genetic and shape diversity (from spiral to rod) within the H. pylori species. Indeed, the disruption of the hp0506 gene (both by total deletion and site-directed mutagenesis) led to the occurrence of a stocky branched phenotype in N6 and X47 backgrounds. A similar phenotype was reported in E. coli, accompanying the simultaneous inactivation of the d,d-CPase DacA (PBP5), with at least two others low-molecular-weight PBPs (Nelson and Young, 2001). But in H. pylori, the single deletion of hp0506 is sufficient to obtain such a phenotype. In addition, N6hp0506ΩK2 presented a slight growth defect, whereas the combination of the deletion of all low-molecular-weight PBPs of E. coli has no effect on growth (Denome et al., 1999). These both observations suggest that HdpA would play an even more central role in the shape regulation than collectively the others d,d-CPases previously described, and that could be linked to the low redundancy of the genome of H. pylori. Nevertheless, PG analysis of the hp0506 mutant revealed the presence of residual amounts of tetrapeptides despite its inactivation. It suggests that H. pylori might harbour at least another d,d-CPase, potentially involved in a different step of the cell cycle, since it is unable to compensate for the phenotype of the hp0506 mutant. Sycuro et al. (2010) described two additional proteins belonging to the M23-family (HP1543 and HP1544) predicted to act preferentially as d,d-EPases. Nevertheless, we cannot exclude they also carry a d,d-CPase activity. Additionally, the MEROPS peptidase database listed HP0750 as a fourth metallopeptidase belonging to this family, which could also be a candidate to perform the d,d-carboxypeptidation.
We also noticed that the overproduction of HdpA led to the transformation from rod-shaped to viable cocci-shaped bacteria. Such a morphological transformation was described for the first time in 1982, with the overproduction of the PBP5 in E. coli (Markiewicz et al., 1982). Since, a few others means of obtaining spherical E. coli were reported. In all these cases, E. coli died after the morphological transition, if not associated with the overproduction of FtsZ, with the increase of the concentration of ppGpp, or with the reduction of the growth rate (Bendezu and de Boer, 2008; Young, 2008). But in this study, we showed that spherical H. pylori N6 overexpressing HP0506 were able to continue dividing and maintained viability. In addition, the labelling of the septum with the vancomycin-FITC proved that cell division and, presumably, the FtsZ ring were normal.
Finally, we could observe that HdpA was involved in the regulation of cell diameter and length. The exact contribution of the protein in this shape regulation remains mysterious. These results are in favour of functional links between cytoskeletal proteins and the PG biosynthetic machinery. However, a recent study has shown that H. pylori MreB is not required for rod-shape maintenance in this bacterium (Waidner et al., 2009), further highlighting the need to study other bacterial models to fully understand and appreciate cell shape diversity in the bacterial world.
As previously discussed, the complete molecular mechanisms required for the genesis of specific bacterial shape still raises many questions. But the other crucial question is why do bacteria need to retain one particular shape among many options? In this study, we raised once again the question by watching the consequences of different shape transitions of H. pylori, on the virulence of the bacteria. Indeed, the colonization assays performed with the deleted mutant revealed a strong defect of colonization since day 1, and that increased with time. Similarly, Sycuro et al. (2010) described an altered colonization ability of the csd1 and csd2 mutants, when tested in a competition assay. Here, we showed that disruption of hp0506 (csd3) is sufficient to strongly affect the mouse colonization of strain X47. Several reasons can explain the impaired colonization of the hp0506 mutant. Motility is an important feature of H. pylori required for efficient colonization of the gastric mucosa (Ottemann and Lowenthal, 2002). In vitro, the normal motility of hp0506 mutant might be a consequence of the fraction of the population with a normal phenotype (around 70% of the bacteria). However, an in vivo bottle-neck effect strongly reduces the amount of bacteria able to reach the gastric mucosa at the beginning of the infection, thus, affecting the colonization of the gastric mucosa. Additionally, we can imagine that the stocky phenotype of most of the population could interfere with the motility of the bacterium in the gastric mucosa, by slowing down its path through the mucus layer, where the viscosity is higher than in brucella broth soft agar plates. Finally, some studies on other rod-shaped bacteria, such as E. coli or P. aeruginosa, have shown that during the first step of the bacterial adhesion, the presence of adhesins specifically localized at the cell poles (Chiang et al., 2005) allowing the initial interaction of the bacteria with the host cells, in spite of the electrostatic repulsions (Agladze et al., 2005). Interestingly, the potential H. pylori adhesin HpaA has been shown to localize specifically at the bacterial poles (Lundstrom et al., 2001). Thus, we could imagine that because of the branched phenotype, the amount of adhesins per bacterial pole would be reduced, leading to an impaired attachment of the branched bacteria.
Altogether, the morphological abnormalities of the mutant were likely to be directly involved in its colonization defect. These could impact on several important features of H. pylori required for efficient colonization of the gastric mucosa: motility in a viscous environment, poor adhesion to target cells, or reduced growth. Individually or collectively, these altered phenotypes can contribute significantly to explain the reduced fitness of the hp0506 mutant in vivo.
Bacterial strains and growth media
Escherichia coli strains BL21 (TOP10, Invitrogen) and BLi5 (Novagen) were used, respectively, for plasmid construction and recombinant protein production. They were cultivated at different temperatures in Luria–Bertani medium supplemented with appropriated antibiotics (20 µg ml−1 kanamycin or 20 µg ml−1 chloramphenicol).
Helicobacter pylori strains N6 (Ferrero et al., 1992) and X47 (Ermak et al., 1998) were cultivated on blood agar plates or BHI with 10% of decomplemented fetal calf serum (FCS), both supplemented or not, with an antibiotic-antifungal mix (0.31 µg ml−1 polymyxin B, 2.5 µg ml−1 amphotericin B, 12.5 µg ml−1 vancomycin and 6.25 µg ml−1 trimethoprim) and appropriated antibiotics (20 µg ml−1 kanamycin, 5 µg ml−1 apramycin or 5 µg ml−1 chloramphenicol), at 37°C in a microaerophilic atmosphere.
Construction of the deleted mutants
Deleted mutants N6hp0506ΩKm and X47hp0506ΩKm were constructed by allelic replacement of the hp0506 whole gene with a non-polar kanamycin resistance cassette (Skouloubris et al., 1998). The 3′ end 500 bp of hp0505 and the 5′ end 500 bp of hp0507 were, respectively, amplified by high fidelity PCR (Phusion polymerase, Finnzymes) with the following pairs of primers (5′-gatgataatgagttgagcgaa-3′/5′-CATTTATTCCTCCTAGTTAGTtaatgccttcaatcaaatctg-3′, and 5′-TAGTACCTGGAGGGAATAATgtcttattttaagaatgct-3′/5′-gcacttcagcgtaatagagcg-3′). The kanamycin resistance cassette was purified from the pUC18K-2 plasmid by enzymatic restriction (KpnI, BamHI). The fragments were then assembled by amplification with the primers (5′-gatgataatgagttgagcgaa-3′/5′-gcacttcagcgtaatagagcg-3′), cloned into the TOPO Zero Blunt (Invitrogen), and checked by sequencing. The double recombination on the chromosome of H. pylori strains (N6, X47) was performed by natural transformation with the plasmid purified from E. coli (5 ng of DNA, 16 h) (Skouloubris et al., 1998). The mutants were selected on blood agar plates supplemented with 20 µg ml−1 kanamycin.
Complementation of the hp0506 mutant
Construction of the N6hp0506ΩGm mutant. The N6hp0506ΩGm mutant was constructed by allelic replacement of the whole gene with a non-polar gentamicin resistance cassette. The 3′ end 500 bp of hp0505 and the 5′ end 500 bp of hp0507 were, respectively, amplified by high fidelity PCR (Phusion polymerase, Finnzymes) with the following pairs of primers (5′-gatgataatgagttgagcgaa-3′/5′-AGTTAGTCACCCGGGTACCGtaatgccttcaatcaaatctgtatt-3′, and 5′-TAGTACCTGGAGGGAATAATGatgtcttattttaagaatgctttc-3′/5′-gcacttcagcgtaatagagcg-3′). The gentamicin resistance cassette was amplified from the pUC1813Gm plasmid (Bury-Mone et al., 2003), using the primers (5′-gctcggtacccgggtgact-3′/5′-ctctagaggatccccgtgtc-3′). The fragments were then assembled by amplification with the primers (5′-gatgataatgagttgagcgaa-3′/5′-gcacttcagcgtaatagagcg-3′), cloned into the TOPO Zero Blunt (Invitrogen), and inserted into the chromosome of H. pylori N6 strain by natural transformation with the plasmid purified from E. coli (5 ng of DNA, 16 h). The mutant was selected on blood agar plates supplemented with 5 µg ml−1 apramycin.
Complementation. The N6hp0506ΩGm mutant was complemented by replacement of the gentamicin cassette with a wild-type copy of hp0506 carrying downstream a kanamycin resistance cassette (N6hp0506-Km). The hp0505-hp0506 fragment was amplified with the pair of primers (5′-gatgataatgagttgagcgaa-3′/5′-cattttagccatttaaaaaccctctaaaagata-3′), and then assembled with the construction obtained for the site-directed mutagenesis, using the primers (5′-gatgataatgagttgagcgaa-3′/5′-ccccccaccctttgaaactttc-3′). The construction was cloned into the TOPO Zero Blunt (Invitrogen) and inserted into the chromosome of H. pylori N6 strain by natural transformation with the plasmid purified from E. coli (5 ng of DNA, 16 h). The mutant was selected on blood agar plates supplemented with 20 µg ml−1 kanamycin.
The 3′ end 500 bp of hp0506 and the 5′ end 500 bp of hp0507 were, respectively, amplified by high fidelity PCR (Phusion polymerase, Finnzymes) with the pairs of primers (5′-cgtatgggagattccatcctgtc-3′/5′-cattttagccatttaaaaaccctctaaaagata-3′, and 5′-ggatgaattgttttagatgtcttattttaagaatgctt-3′/5′-ccccccaccctttgaaactttc-3′). The kanamycin resistance cassette was purified from the plasmid by enzymatic restriction (KpnI, BamHI). The fragments were then assembled by amplification with the primers (5′-cgtatgggagattccatcctgtc-3′/5′-ccccccaccctttgaaactttc-3′), cloned into the TOPO Zero Blunt (Invitrogen), and checked by sequencing. The substitution of H259 (CAT) with an alanine (GCA) was introduced by reverse PCR with the following pair of primers (5′-gcataatccacgccgtaTGCaggccgtctg-3′/5′-gcctGCAtacggcgtggattatgcggctaa-3′), and checked by sequencing. The double recombination on the chromosome of H. pylori N6 strain was performed by natural transformation with the plasmid purified from E. coli (5 ng of DNA, 16 h). The mutant N6hp0506*Km was selected on blood agar plates supplemented with 20 µg ml−1 kanamycin.
Construction of the overexpressing mutant
The whole hp0506 gene was amplified by high fidelity PCR (Phusion polymerase, Finnzymes) with the primers (5′-ggaattcCATATGgtattttttcataagaaaatt-3′/5′-cgcGGATCCttaaaaaccctctaaaagata-3′), before cloning into the pILL2157 plasmid (Boneca et al., 2008) by enzymatic restriction (NdeI, BamHI). The construction, named pMAB1, was checked by sequencing, and then introduced into the N6 strain by natural transformation with the plasmid purified from E. coli (5 ng of DNA, 16 h). The mutant was selected on blood agar plates supplemented with 5 µg ml−1 chloramphenicol.
N6pMAB1 was grown in BHI with 10% FCS supplemented or not with 1 mM IPTG. Cultures were stopped in exponential growth. Cells were washed twice with PBS and then resuspended in 100 µl of PBS. Bacteria were incubated with BodipyFL-conjugated vancomycin (Pierce) to a final concentration of 3 µg ml−1, supplemented by cold vancomycin (Sigma) at the same concentration for 20 min at room temperature. Pictures were obtained without fixation on an inverted epifluorescence microscope Axio Observer (Zeiss), using the Axiovision software.
Bacteria from exponential cultures were washed with PBS and loaded on cover slides by centrifugation (1000 r.p.m., 5 min). Bacteria were then fixed with 4% of paraformaldehyde for 30 min and washed with PBS three times for 5 min. Cover-slides were fixed on slides with Dako fluorescence mounting medium (Prolong Gold, Invitrogen). Images were acquired with an inverted microscope Axio Observer (Zeiss), in phase contrast, using the Axiovision software.
Scanning electronic microscopy
Bacteria from exponential cultures were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.2). They were then washed three times for 5 min (each time) in 0.2 M cacodylate buffer (pH 7.2), post-fixed for 1 h in 1% (w/v) osmium tetroxide in 0.2 M cacodylate buffer (pH 7.2), and then rinsed with distilled water. Samples were dehydrated through a graded series of 25%, 50%, 75% and 95% ethanol solution for 5 min (each time). Samples were then dehydrated for 10 min (each time) in 100% ethanol followed by critical point drying with CO2. Dried specimens were sputtered with 10 nm gold palladium, with a GATAN Ion Beam Coater and were examined and photographed with a JEOL JSM 6700F field emission scanning electron microscope operating at 5 kV. Images were acquired with the upper SE detector (SEI).
Peptidoglycan extraction and analysis
The PG from N6, N6hp0506ΩKm, N6hp0506*Km and induced or non-induced N6pMAB1 were extracted from 100 ml of culture stopped in exponential growth, in an ice-ethanol bath. Then the crude sacculus was immediatetly extracted by boiling in SDS (4% final concentration). After cooling, the SDS was removed by several ultracentrifugation (39 000 r.p.m., 30 min, 18°C), by washing with distilled water. The total elimination of the SDS was checked as described (Hayashi, 1975). The PG suspensions were digested with 184 units of mutanolysin (Sigma-Aldrich) in 12.5 mM sodium phosphate buffer (pH5.8), overnight at 37°C. Samples were then reduced for 30 min in 0.25 M of borate buffer pH 9 to which a spatula end of sodium borohydride was added. The reaction was stopped by adjusting the final pH to 2 with orthophosphoric acid. Muropeptides were analysed by HPLC, using a Hypersil ODS18 reverse-phase column with a 0–15% methanol gradient in sodium phosphate buffer (pH 4.3–4.9; Glauner, 1988). Chromatograms were obtained by monitoring of muropeptides elution at 206 nm, and relative area of each peak was estimated using the LabSolution software (Shimadzu).
Production and purification of the recombinant His6-tagged Hp506
Plasmid construction. The hp0506 gene was fused to a C-terminal 6-His tag by cloning it without its signal peptide into the expression plasmid TOPO pET151-D (Invitrogen), using the primers (5′-caccgccgatggaatggctaaaaag-3′/5′-ttaaaaaccctctaaaagata-3′). The plasmid was transformed in the BLi5 E. coli strain, selected with 100 µg ml−1 carbenicillin, and verified by sequencing.
Production and purification. The strain was cultivated in 200 ml of Luria–Bertani medium supplemented with 100 µg ml−1 carbenicillin. When the OD600 reached approximately 0.6, the expression of the recombinant protein was induced by adding 1 mM IPTG and incubating 3 h at 30°C. Cells were then centrifugated and resuspended in 5 ml of lysis buffer (20 mM HEPES, 500 mM NaCl, 10% glycerol, pH 8) supplemented with a protease inhibitor mix (Complete without EDTA, Roche). The bacteria were sonicated (4 × 15 s, 50%), and centrifuged 20 min at 7500 r.p.m., 4°C. The supernatant was then incubated with 2 ml of Ni-NTA agarose (Qiagen) for 1 h at 4°C. To remove contaminating proteins, beads were washed with wash buffer (20 mM HEPES, 1 M/500 mM NaCl, 10% glycerol, 20 mM imidazol, pH 8). The recombinant protein was eluted with 3 ml of elution buffer (20 mM HEPES, 150 mM NaCl, 10% glycerol, 200 mM imidazol, pH 8).
Anhydromuropeptides purification. To purify anhydromuropeptides, PG from strain N6 was digested with the recombinant Slt70 from E. coli in 350 mM sodium acetate pH 6.5 at 37°C for 18 h. Anhydromuropeptides were separated and purified by HPLC using a Hypersil ODS18 reverse-phase column with a 0–50% acetonitrile gradient in 0.1% trifluoroacetic acid.
Enzymatic tests. The enzymatic tests were performed by incubating 5 µg of HP0506-His6 with the substrates GanhM5 or GanhM5–GanhM4, purified as described above, in 20 mM MES buffer pH 6.0, supplemented with 1 mM MgCl2, 1 mM penicillin G (Sigma-Aldrich) to inactivate traces of penicillin-binding proteins with d,d-carboxypeptidase activity from the E. coli producing strain, and EDTA for the negative control. After 2 h of incubation at 37°C, the samples were analysed by HPLC. Muropeptides and anhydromuropeptides were analysed by MALDI-TOF mass spectometry as previously described (Antignac et al., 2003).
Bocillin affinity. The affinities of HP0506-His6 and the recombinant PBP5-His6 from E. coli for penicillin were estimated using an analogue of the penicillin G linked to Bodipyl FL dye (Bocillin, Invitrogen). One microgram of each protein was incubated with 100 µM of Bocillin in PBS, during 15 min at 37°C and protected from the light. Proteins were then detected on a SDS-polyacrylamide (10%) electrophoretic gel, using a PharosFX Plus Molecular Imager (Bio-Rad, 488–530 nm).
Mouse colonization assays
Five-week-old female mice (C57BL/6J, Charles River) were infected with wild-type X47 or X47hp0506ΩKm by oral route with feeding needles (2.108 bacteria per mouse, 16 mice per condition, from two distinct experiments). After 1, 4 and 7 days of infection, mice were euthanized with CO2 and the stomachs were ground and homogenized in peptone broth. The samples were then diluted and spread on blood agar plates supplemented with 200 µg ml−1 bacitracin, and 10 µg ml−1 nalidixic acid, to inhibit the growth of resident bacteria from the mouse forestomach. The colonies forming units were enumerated after 8 days of incubation.
The motility of the bacteria was measured on brucella broth plates supplemented with 0.35% of bacteriological agar, 10% of FCS, and with the antibiotic-antifungal mix. The medium was inoculated with 2 µl of exponential liquid cultures at mid-height of the agar and incubated at 37°C for 5 days.
Mathilde BONIS was supported by a PhD fellowship (Ministère de l'Enseignement Supérieur et de la Recherche, France). This study was supported by the ERC starting grant (PGNfromSHAPEtoVIR n°202283). We would like to thank Dominique Mengin-Lecreulx for kindly providing recombinant Slt70 and PBP5 from E. coli.