Evidence for Conserved Function of γ–Glutamyltranspeptidase in Helicobacter Genus

The confounding consequences of Helicobacter bilis infection in experimental mice populations are well recognized, but the role of this bacterium in human diseases is less known. Limited data are available on virulence determinants of this species. In Helicobacter pylori, γ-glutamyltranspeptidase (γGT) contributes to the colonization of the gastric mucosa and to the pathogenesis of peptic ulcer. The role of γGT in H. bilis infections remains unknown. The annotated genome sequence of H. bilis revealed two putative ggt genes and our aim was to characterize these H. bilis γGT paralogues. We performed a phylogenetic analysis to understand the evolution of Helicobacter γGTs and to predict functional activities of these two genes. In addition, both copies of H. bilis γGTs were expressed as recombinant proteins and their biochemical characteristics were analysed. Functional complementation of Esherichia coli deficient in γGT activity and deletion of γGT in H. bilis were performed. Finally, the inhibitory effect of T-cell and gastric cell proliferation by H. bilis γGT was assessed. Our results indicated that one gene is responsible for γGT activity, while the other showed no γGT activity due to lack of autoprocessing. Although both H. bilis and H. pylori γGTs exhibited a similar affinity to L-Glutamine and γ-Glutamyl-p-nitroanilide, the H. bilis γGT was significantly less active. Nevertheless, H. bilis γGT inhibited T-cell proliferation at a similar level to that observed for H. pylori. Finally, we showed a similar suppressive influence of both H. bilis and H. pylori γGTs on AGS cell proliferation mediated by an apoptosis-independent mechanism. Our data suggest a conserved function of γGT in the Helicobacter genus. Since γGT is present only in a few enterohepatic Helicobacter species, its expression appears not to be essential for colonization of the lower gastrointestinal tract, but it could provide metabolic advantages in colonization capability of different niches.

Introduction c-Glutamyltranspeptidase (cGT) is a threonine N-terminal nucleophile (Ntn) hydrolase that catalyses the transpeptidation and hydrolysis of the c-glutamyl group of glutathione and related compounds [1]. cGT is widely distributed in living organisms and is highly conserved, with mammalian and bacterial homologues often sharing more than 25% of sequence identity [2]. From the ,1000 of whole genome sequenced bacterial species available in MEROPS databases (http://merops.sanger.ac.uk [3]), 540 (,200 genera) possess cGT-like proteins belonging to protease family T03. Moreover, several bacterial species carry multiple copies of genes annotated as cGT, but the majority of these genes lack functional verification.
cGT is found in all gastric Helicobacter species. However, among the 20 validly published enterohepatic Helicobacter species (EHS), only H. aurati, H. bilis, H. canis, H. muridarum and H. trogontum express this enzyme [4]. In H. pylori, cGT represents an important virulence factor, playing a role in colonization [5,6] and pathogenesis [7][8][9][10]. It is constitutively expressed in vivo and in vitro [11] and enables the bacterial cells to use extracellular glutamine and glutathione as a source of glutamate [9,12]. Purified Hp-cGT was shown to be involved in upregulation of growth factors in MKN-28 gastric cells [7], to induce apoptosis in AGS cells [10], to inhibit T-cell proliferation [9] and to contribute to H. pylori-mediated H 2 O 2 generation [8]. By contrast, the biochemical properties of EHS cGT and its role in the colonization of the gut and in the pathogenesis of gastrointestinal and hepatobiliary diseases are completely unknown. EHS is a phenotypically and genotypically heterogeneous phylogroup within the Helicobacter genus [4], including species usually colonizing the intestinal tract and/or the liver of mammals and birds. Although EHS could be considered part of the normal microbiota of rodents, some species cause diseases in these animals [13]. In particular, H bilis, an endemic EHS in most experimental mouse colonies, induces disease in susceptible animals and may substantially confound interpretations of some research studies [14]. Natural H. bilis infection in inbred [15] or outbred mice [16] has been associated with multifocal hepatitis. Moreover, H. bilis has been used experimentally to induce inflammatory bowel disease (IBD) in mdr 2/2 and IL-10 2/2 knock-out mice [16], typhlocolitis in the C3H/HeN mice strain [17] and cholesterol gallstone formation in C57L mice [18]. H. bilis is able to infect and cause diseases in different animal hosts, showing one of the broadest host spectrums in the Helicobacter genus [19]. It was isolated from the aborted fetus of sheep and pig [19] and from chronic hepatobiliary diseases in hamsters [20]. H. bilis has been also isolated from human patients with chronic diarrhoea [21] and pyoderma gangrenosum-like ulcers [22]. In addition, several studies have reported an association of this species with chronic liver diseases [23,24] or biliary tract and gallbladder cancers [25,26] in human, using either PCR or serological tests. Limited data are available on virulence determinants of H. bilis [27][28][29], and no studies to date have described the biochemical and biological proprieties of H. bilis cGT (Hb-cGT).
In contrast to observations in gastric Helicobacter spp., the genome sequence of H. bilis ATCC 43879 revealed the presence of two ggt copies. In this study, we used a phylogenetic and a functional approach to analyse both H. bilis cGT paralogues. Although both genes were phylogenetically related to other Helicobacter cGTs, analysis of the recombinant proteins, western blot using specific antibodies, complementation of E. coli Dggt and mutation in H. bilis clearly showed that only one gene was responsible for H. bilis cGT activity. The cGT of H. bilis exhibited a similar affinity as H. pylori to c-Glutamyl-p-nitroanilide and to L-Glutamine; however, it was significantly less active. Nevertheless, H. bilis cGT inhibited T-cell and gastric cell proliferation at a similar level to that observed for H. pylori cGT. The inhibition observed was mediated by an apoptosis-independent mechanism and suggested a conserved function of cGT in Helicobacter genus.

Results
Sequence analysis revealed marked differences between two cGT paralogues of H. bilis ATCC 43879 The ggt paralogues HRAG_01341 and HRAG_01828 of the human-associated H. bilis strain ATCC 43879 genome (NCBI ACDN0000000), were named bgh1 (H. bilis ggt homologue 1) and bgh2 (H. bilis ggt homologue 2), respectively. Nucleotide and amino acid similarity between the two H. bilis cGT paralogues, and between each homologue and H. pylori cGT (Hp-cGT; HP1118) were analysed by pairwise global alignment. The two H. bilis cGT paralogues showed 65.8% and 62.0% of nucleotide and amino acid identity, respectively. Moreover, in comparison with Hp-cGT, Bgh2 showed 65.2% of amino acid identity, while Bgh1 showed 53.4%. To detect amino acid positions potentially involved in functional change, conserved sites of both Bgh1 and Bgh2 genes were evaluated on the basis of cGT structural data available for H. pylori [30,31] and E. coli [32]. A multi-alignment including all Helicobacter cGT sequences, other bacterial cGTs and class IV cephalosporin acylase (CA) of Pseudomonas sp. strain SE82, was constructed ( Figure 1). In Bgh2, all the functional sites described for H. pylori are conserved. In Bgh1, by contrast, amino acid substitutions in the 20 kDa subunit potentially involved in functional change were observed. The substitutions were as follows: Asp421Glu, Asp422Asn, Leu432Arg, Tyr433His, Ser452Thr and Gly473Ser (Hp-cGT numeration).
To predict the localization of both H. bilis cGT paralogues, the sequences were submitted to SignalP 3.0, using both Neural networks and Hidden Markov Models (HMM) methods [33], PSLpred (Hybrid Approach Based; [34]), PSORTb v3.0.2 [35] and CELLO v2.5 [36]. These tools were unable to uniformly predict the sub-cellular localization of both Bgh1 and Bgh2. To verify the presence of other potential start codons, we analysed the upstream region of both genes. We identified a second putative start codon 54 bp upstream of the annotated methionine of Bgh2, but no other potential start codons for Bgh1. Using the new predicted start codon, all of the analysis strongly predicted Bgh2 as a periplasmic protein, and SignalP detected the presence of sec signal peptide sequence with a potential cleavage site between positions 30 and 31 (VFA-AS). The same prediction was obtained for all cGTs of Helicobacter spp. and C. jejuni. No twin-arginine signal peptide cleavage site was detected in any cGTs of Helicobacter spp. and C. jejuni. Phylogenetic analysis of Helicobacter cGTs Phylogenetic relationships among epsilon proteobacteria and other bacterial cGTs and Pseudomonas spp. CA were analysed ( Figure 2). The Minimum Evolution tree based on amino acid sequence alignment showed that both H. bilis cGT paralogues cluster with Helicobacter and C. jejuni cGTs, forming, however, two distinct branches: Bgh2 and C. jejuni cGT belonged to a single clade, while Bgh1 formed a separate branch.
To examine the phylogenetic relationships among Helicobacter cGTs in more detail and to identify other Helicobacter species potentially carrying multiple cGT sequences, consensus degenerate hybrid oligonucleotide primers were designed on the basis of highly conserved amino acid motifs (Table 1). A fragment of about 1300 bp, corresponding to an almost complete ggt sequence, was successfully amplified from H. canis, and two fragments, corresponding to an almost complete sequence of two distinct ggt copies, were amplified from H. trogontum and H. bilis genomospecies FL56. From H. aurati and H. muridarum, amplification of only the Nterminal part of a single ggt copy was possible. An unrooted Minimum Evolution tree was built on the basis of an almost complete amino acid sequence of the cGT pro-enzyme of Helicobacter spp. and C. jejuni (Figure 3a). Phylogenetic analysis showed that the two ggt copies amplified from H. trogontum and H. bilis genomospecies FL56 corresponded to orthologues of bgh1 and bgh2, whereas the sequences obtained from H. canis, H. aurati and H. muridarum (Figure 3a,b) clustered with other Helicobacter ggt and bgh2. In Figure 3, the phylogeny of the pro-enzyme was compared with those of the single sub-units. The trees showed differences in both the topology and the bootstrap supporting nodes, suggesting different evolution of the two parts. The same results were obtained comparing the maximum likelihood trees of the nucleotide sequences (data not shown). Finally, to provide a measure for the selective pressure that the gene pair bgh1 and bgh2 was subjected to, the ratio between non-synonymous/synonymous substitution (Ka:Ks) was evaluated by sliding window analysis (window size 50 bp) using SWAKK [37]. A Ka:Ks ratio smaller than one was observed in all positions analysed. The average Ka:Ks ratio was calculated to be 0.175, indicating that the genes are under purifying selection. A similar Ka:Ks ratio was observed also for the bgh1 and bgh2 pairs of H. trogontum and H. bilis genomospecies FL56 (data not shown).

Frequency of bgh1 and bgh2 genes in H. bilis strains
To determine the frequency of bgh1 and bgh2 in H. bilis, 33 H. bilis strains from our collection [38], including H. bilis type strain ATCC 51630 T , H. bilis ATCC 49314 and H. bilis ATCC 49320, were subjected to a specific BclI RFLP-PCR for bgh1 and a specific PCR for bgh2. All strains were positive for the presence of both genes. Furthermore, bgh1 orthologues of H. bilis type strain ATCC 51630 T and three canine strains [38] were sequenced and uncorrected distance matrices were constructed on the basis of both nucleotide and amino acid sequences. The overall nucleotide sequence identity observed varied from 94.7% to 97.3%, while the amino acid identity ranged from 97.5% to 100%.

Transcription of bgh1 and bgh2 by H. bilis CCUG 23435
The transcription of both bgh1 and bgh2 was evaluated in H. bilis CCUG 23435 at different time points. Bacterial growth was

Expression, purification and autoprocessing of recombinant Bgh1 and Bgh2 proteins
To further analyse the biochemical characteristics of both H. bilis cGT paralogues, we expressed recombinant His-tagged Bgh1 and Bgh2 proteins in E. coli. Bgh2 was expressed without the signal peptide, while for Bgh1 the entire sequence was used. Recombinant expression of Bgh2 resulted in a soluble protein that was directly purified by Ni-affinity chromatography. By contrast, Bgh1 was expressed as an insoluble protein of ,70 KDa and was purified after refolding. The purity of the proteins on SDS-PAGE was .90% (Figure 4a). Bgh2 showed a catalytic activity for the substrate analogue L-c-glutamyl-p-nitroanilide (gGpNA) and was synthesized as a pro-form of ,60 KDa, which undergoes autocatalytic processing to generate two subunits of ,40 and ,20 KDa (Figure 4a). By contrast, after solubilisation and refolding, recombinant Bgh1 showed no significant maturation after 24 h of incubation at 37uC and no activity for gGpNA ( Figure 4b). To estimate the secondary structure of Bgh1 and to verify the absence of a random folding, a Circular Dichroism (CD) spectrum of the recombinant protein was calculated. The spectrum minimum calculated for recombinant Bgh1 was 206 nm, indicating the absence of random coil. The secondary structure of Bgh1 was predicted by K2D3 [39] to contain 30.23% of alfa helix and 16.82% of beta sheet. The CD spectrum for Bgh1resulted similar to the spectrum predicted by K2D3 and to those for proper folded proteins, e.g. lysozyme [40], indicating that recombinant Bgh1 has a defined secondary structure ( Figure S1).

Autoprocessing of Bgh1 and Bgh2 in H. bilis
In order to evaluate the maturation of both Bgh1 and Bgh2 in H. bilis, antisera against both recombinant proteins were produced in mice and western blot analysis on H. bilis whole lysate was performed. The antisera showed high specificity for the corresponding H. bilis cGT paralogue in ELISA test at 1:900 dilution selected for the subsequent western blots (data not shown). The western blot performed on H. bilis whole lysate using the antiserum against Bgh1 clearly showed a single band without indication of autoprocessing ( Figure 5a). On the contrary, the western blot carried out using antiserum against Bgh2 revealed the expression of the pro-form and its corresponding two subunits ( Figure 5b). These results confirmed that both proteins are expressed in H. bilis and that only Bgh2 undergoes autocatalytic processing.

Complementation of E. coli Dggt
To evaluate the capability of H. bilis cGT paralogues to functionally complement an E. coli cGT-deficient strain, Bgh1 or Bgh2 (carried on plasmid pMRg3 and pMRg5, respectively) were introduced in E. coli CY128 [41]. E. coli DH5a grown in LB overnight at 25uC was used as positive control in cGT assay. Only Bgh2 was found to successfully complement E. coli CY128, while Bgh1 was unable to restore cGT activity (data not shown). The induction of pMRg3 in E. coli CY128 resulted in a clear inhibition of growth. This finding was consistent with formation of inclusion bodies, as a consequence of the overexpression of Bgh1. However, when the protein was expressed overnight at a lower temperature (15uC) with low amount of inducer (0.002% instead of 0.2%), there was no inhibition of the growth rate of E. coli, suggesting a low accumulation of insoluble proteins. The successful expression of Bgh1 under improved conditions was confirmed by SDS-PAGE (data not shown). Nevertheless, Bgh1 was not able to complement E. coli Dggt, while Bgh2, expressed under the same conditions, restored cGT activity in the mutant strain.

Construction of a H. bilis mutant deficient in cGT activity
To confirm that Bgh2 is responsible of the cGT activity in H. bilis, a mutant of bgh2 was constructed in CCUG 23435 strain. The bgh2 gene was disrupted by insertion of a chloramphenicol cassette between positions 1182 and 1246, in order to delete the threoninethreonine catalytic dyad. The successful insertion of the choramphenicol resistance cassette and the deletion of bgh2 were confirmed by PCR. RT-PCR and western blot analysis showed the expression of bgh1 in both wild-type and mutant strains (data not shown), indicating that the replacement of bgh2 had no effect on the expression of the bgh1 paralogue gene. Since complementation approaches are currently unavailable for H. bilis, another independent Dbgh2 mutant was used as a control for secondary mutations. Neither of the H. bilis mutants were able to hydrolyse the substrate analogue gGpNA. Moreover, in the supernatant of H. bilis wild-type cGT activity was determined with a gGpNA turnover of 0.48 mM/min at a concentration of 50 mg/mL of total protein. No activity was detected in the supernatant of the mutant strain MR9 ( Figure S2). These results indicate that only bgh2 encodes a functional cGT in H. bilis CCUG 23435 (Hb-cGT). The mutants grew normally in vitro, as described for H. pylori cGTdeficient strains [9], indicating that cGT is not essential for survival and growth of both H. pylori and H. bilis in vitro.  The kinetic parameters of recombinant Bgh2 (Hb-cGT) were measured. The initial experiments were performed at pH 8.0. The K M was determined by titration of the gGpNA concentrations between 1 and 1000 mM. The processed Hb-cGT had a K M of 7.761.2 mM and a k cat of 1.1260.03N10 2 /sec. Compared with K M and k cat values calculated for Hp-cGT (K M = 9.861.5 mM and k cat = 17.760.5N10 2 /sec), the apparent K M value for Hb-cGT was 20% lower, while the k cat value was .10-fold reduced ( Table 2). The pH dependence of the Hb-cGT activity was analysed and compared to Hp-cGT ( Figure 6). The pH profile of Hp-cGT activity was similar to reported data [30] (data not shown), whereas the activity optimum of Hb-cGT was shifted to a more acidic pH between 6.0 and 7.0. By contrast, the K M was increased more than 2-fold at this pH range. Initial experiments for comparing the substrate specificity of Hp-cGT and Hb-cGT for glutathione, D-glutamine, L-glutamine and L-glutamic acid were performed by a competition assay. A better binding affinity to the active cGT and therefore a competitive inhibition of the gGpNAreaction would be expected for its native substrates. As previously described [9,12], Hp-cGT substrate inhibition was observed for both by glutathione and L-glutamine, whereas Hb-cGT was only inhibited by L-glutamine ( Figure S3).

Inhibition of human T-cell and AGS cell proliferation
To evaluate biological properties of recombinant Hb-cGT and compare these with effects already observed for Hp-cGT, the ability of Hb-cGT to inhibit human T-cell and AGS cell proliferation was assessed. Recombinant Hb-cGT efficiently inhibited the proliferation of Jurkat cells in a dose-dependent manner (Figure 7a). Concentrations of Hb-cGT as low as 0.1 mg/ mL induced inhibition of T-cell proliferation. Comparison of the inhibitory potential of supernatants of wild-type and Dggt strains confirmed the results obtained with the recombinant proteins. The supernatants of H. bilis wild type reduced T-cell proliferation up to 60%, while the supernatant of Dggt strain had no significant inhibitory activity. When Jurkat cells were infected with acivicin pre-treated supernatant of H. bilis wild type, no significant reduction in proliferation of T-cells was observed, indicating that only Hb-cGT is responsible for the inhibition of T-cell proliferation in our system ( Figure 7b).
Moreover, both recombinant Hb-cGT and Hp-cGT showed inhibitory effect on AGS cell proliferation in a dose-dependent manner as well, although the minimum concentration of protein required was 25 time higher (2.5 mg/mL) compared to that able to inhibit the proliferation of human T-cells ( Figure 8). No statistically significant differences were observed between the two proteins.

Apoptosis in AGS cell line
The ability of recombinant Hp-cGT and Hb-cGT to induce apoptosis in AGS cells was evaluated using AnnexinV-Propidium Iodide double staining and FACS analysis. After 24 h of incubation, concentrations of up to 5 mg/mL of both recombinant proteins did not induce apoptosis in AGS cells ( Figure S4). The average of percentage of apoptosis observed after treatment with Hp-cGT and Hb-cGT was 6.961.2% and 6.860.9%, respectively. Those values were not statistically different from the percentage of apoptosis measured in the untreated cell (7.361.9).

Discussion
Two Helicobacter cGT genes, identified in the shotgun genome sequence of the human strain H. bilis ATCC 43879 on the basis of sequence homology, were subjected to functional analysis to   confirm their annotations. Although the two genes bgh1 and bgh2 (H. bilis ggt homologue 1 and 2) showed significant homology and both are classified as T03.01 peptidases in MEROPS, the evaluation of conserved amino acid residues indicated a marked difference between these two paralogues, especially in the substrate-binding pocket. In fact, all critical functional residues are present in Bgh2, as well as in all the other Helicobacter cGTs, but not in Bgh1. The amino acid substitutions observed in Bgh1 could potentially affect the enzyme maturation, catalytic activity and substrate specificity, suggesting different biochemical properties. Moreover, the absence of a signal peptide in Bgh1, in contrast to observations for Bgh2 and for the other Helicobacter cGTs, suggests a different sub-cellular localization of this protein.
We first evaluated the phylogeny of Helicobacter cGTs in relation with T03 proteases of other bacterial species. The Minimum Evolution tree clearly showed that both H. bilis T03.01 peptidase paralogues were phylogenetically associated with other e-proteobacteria cGTs. Nevertheless, Bgh1 formed an independent branch, indicating a different evolution. Moreover, among Helicobacter species, only H. trogontum showed two distinct cGT paralogues, as observed in H. bilis. In addition, the data suggested that both genes are not restricted to H. bilis ATCC 43879, but are common features among H. bilis genomospecies and universally shared by all members of the taxon.
A more detailed phylogenetic analysis further revealed that the two subunits of cGTs have followed different evolutionary paths. The tree based on the C-terminal part resembled the topology of the tree of 16S rRNA and other housekeeping genes and includes H. bilis in the same branch as other enterohepatic Helicobacter spp. [38]. On the contrary the phylogenetic analysis of the N-terminal part separates the cGTs of H. bilis and related species (H. aurati, H. muridarum and H. trogontum) from others enterohepatic Helicobacter species. On the basis of our phylogenetic analysis we can hypothesise that differing phylogeny of the subunits reflects a possible divergence of substrate specificity between the cGTs of H. bilis and related species and the other Helicobacter or Campylobacter cGTs. More studies are needed to verify this hypothesis.
In the trees based on the sequences of both subunits, the position of the clade including the orthologues of bgh1 indicated that the paralogues potentially evolved by gene duplication from a common ancestral sequence probably before the speciation event of H. bilis and related species. The hypothesis of gene duplication is well supported also by the low Ka:Ks ratio between the two paralogues (0,175), indicating that the two genes are under strong purifying selection [42].
Both sequence and phylogenetic analyses reveal clear differences between the two H. bilis T03.01 peptidases paralogues. To compare their biochemical properties, both genes were overexpressed in E. coli and the recombinant proteins were purified by affinity chromatography. The overexpression of Bgh1 resulted in formation of inclusion bodies due to the accumulation of an insoluble protein of ,70 KDa. After solubilisation and refolding, recombinant Bgh1 showed no significant maturation into two subunits and, subsequently, no cGT activity. In contrast, recombinant Bgh2 was expressed in soluble form as a pro-enzyme of ,60 KDa, which undergoes autocatalytic processing to generate two subunits of ,40 and ,20 KDa and showed cGT activity. No clear molecular evidence explains the absence of an autocatalytic process for Bgh1 since all of the amino acid residues critical for Hp-cGT [2,43] and Ec-cGT [44,45] autoprocessing  are also conserved in Bgh1. We, therefore, cannot exclude that the absence of maturation could be due to misfolding in vitro. However, the CD spectrum of purified Bgh1 showed that the protein has a defined set of secondary structures indicating a nonrandom folding of the recombinant protein. Moreover, Bgh1 was unable to restore cGT activity after complementation of E. coli Dggt mutant. Furthermore, the H. bilis strain in which bgh2 was disrupted but still able to transcribe bgh1 resulted in deficient cGT activity. This data clearly indicate bgh1 does not encode a functional H. bilis cGT. As reported previously, the single amino acid substitution D433N converts Ec-cGT to a class IV cephalosporin acylase [46]. Since in Bgh1 the same substitution is present, we decided to check the potential cephalosporin acylase activity of H. bilis CCUG 23435, which expresses both cGT paralogues, as previously described for E. coli [46]. H. bilis did not show any cephalosporin acylase activity (data not shown), indicating that glutaryl-7-aminocephalosporanic acid is unlikely to be a substrate for Bgh1. To elucidate whether the lack of enzyme activity of Bgh1 is caused by the absence of autocatalytic maturation, western blot analysis on H. bilis whole lysate using specific antisera for both cGT paralogues were performed. The results showed that H. bilis expresses Bgh1 which does not undergo autocatalytic processing, confirming the results obtained with the recombinant protein. However, further studies are needed to determinate potential substrates for Bgh1 and to investigate its role in the metabolism of H. bilis.
Analysis of the recombinant protein, western blot on H. bilis whole lysate, complementation of E. coli Dggt and deletion of the gene in H. bilis CCUG 23435 clearly showed that only bgh2 encodes the functional H. bilis cGT. Hb-cGT exhibited similar affinity to the substrate analogue gGpNA at pH 8.0 compared with that of Hp-cGT; however, it was significantly less proficient, having more than 10-fold less k cat . Moreover, in contrast to observations for Hp-cGT [30], the affinity of Hb-cGT to gGpNA varied in relation to the pH, being stronger at acidic and alkaline conditions and decreasing significantly at pH 6.0 and 7.0. In addition, Hb-cGT showed the highest activity at pH 6.0, which is lower compared to previous reports for H. pylori and other bacterial cGTs [30]. The results of the competition assay suggested that Hb-cGT and Hp-cGT have similar substrate specificity, utilizing not only the substrate analogue gGpNA but also L-glutamine. Nevertheless, Hb-cGT differs from Hp-cGT in its inability to bind glutathione as substrate. Therefore, we can speculate that H. bilis cGT lost its ability to interfere the host redox system. However, the enzyme enables the bacterial cells to use extracellular glutamine as the source of glutamate. This hypothesis is supported by the presence in the H. bilis genome of a putative glutamate transporter gene (HRAG_00091; gltS) showing ,60% of amino acid identity with H. pylori GltS homologue [47]. Glutamate transporters gltS as well as ggt genes are missing in the EHS H. hepaticus genomes [47]. Both H. hepaticus and H. bilis are urease-positive Helicobacter spp., commensal bacterial species and/ or opportunistic pathogens of mice, sharing the same niches in the lower gastrointestinal tract and causing similar diseases [48]. However, H. bilis, but not H. hepaticus, was also isolated from the stomach of different hosts [19]. Therefore, the expression of cGT activity by Helicobacter spp. does not appear to be essential for colonization of the lower gastrointestinal tract of mice, but it could provide metabolic advantages in colonization capability of other tissues or in the adaptation of different hosts.
Although the importance of cGT for physiology and pathobiology of mammalian tissues has been known for many years, its function in prokaryotic metabolism and its role in interaction with the host have been studied for only a few bacterial species [6,9,10,12,47,[49][50][51]. Recently it has been showed that cGT has an important role in the interaction between H. pylori and the host by inhibiting T-cell proliferation [9], inducing apoptosis in gastric cells [10] and contributing to H. pylori-mediated H 2 O 2 generation [8]. Here, we showed that H. bilis cGT inhibited T-cell proliferation similar to what was observed for H. pylori [9,10]; the inhibitory effect of T-cell proliferation was completely abolished in the cGT-deficient mutants of H. bilis. Our data demonstrate that this activity is not limited to H. pylori but is conserved among the genus, confirming the potential immune suppressive role of secreted Helicobacter cGTs. The conserved function of this enzyme among Helicobacter spp. with different niches and host specificity opens a new scenario on the role of cGT in the pathogenesis of Helicobacter associated diseases.
We detected AGS cell proliferation inhibition after treatment with both recombinant cGTs, as observed for T-cells, but only when a high concentration was used (2.5 mg/mL). However, in contrast to what was previously described [10], we showed that the suppressive influence of both Hb-cGT and Hp-cGT on AGS cells was mediated by an apoptosis-independent mechanism. The discrepancy between our results and those of Shibayama and colleagues [10] could attribute to different methodologies applied, in particular to the use of serum starvation by these authors. An increase of apoptosis was verified in AGS cells treated with both Hp-cGT and Hb-cGT under the same conditions used by these authors (data not shown), apparently confirming the apoptosis inducing potential of these enzymes. However, if this phenomenon is only detectable upon stress conditions (i.e. serum starvation), we cannot exclude that the observed apoptosis is a consequence of a nonspecific activity of cGT. Further studies are needed to establish the real role of cGT in the apoptosis of epithelial cells.

Bacterial strains, cell line, culture condition and DNA extraction
The bacterial strains and plasmids used in this study are listed in Table 3 and 4. All Helicobacter strains were routinely grown on HP medium (LabM Limited, Lancashire, UK) containing 5% bovine blood at 37uC in a microaerobic atmosphere (10% CO 2 ; 5% O 2 ) (Thermo Forma, Series II Water Jacketed Incubator; Thermo Fisher Scientific, Waltham, MA USA). E. coli strains were cultivated on Luria-Bertani (LB) agar or broth supplemented with 100 mg/L of ampicillin, 15 mg/L of kanamycin or 20 mg/L of chloramphenicol when needed. Jurkat T-cells (DSMZ, ACC 282, Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Braunschweig, Germany) were cultured in RPMI 1640 (Invitrogen Corporation, Carlsbad, CA, USA) with 10% of foetal bovine serum (FBS; Gibco, Invitrogen). AGS cells (CLS, 300408; Cell Lines Service, Eppelheim, Germany) were grown in Dulbecco's Modified Eagle's growth medium (DMEM high glucose, 11971, Invitrogen) supplemented with 10% FBS and 200 U/mL of penicillin and 0.2 mg/mL of streptomycin (Pen Strep, Gibco, Invitrogen). Cell lines were incubated at 37uC with 5% CO 2 . Bacterial DNA extraction was performed as described earlier [19].

PCR, cloning and sequencing
Consensus DEgenerate Hybrid Oligonucleotide Primers [iCO-DEHOP [52]] were designed on the basis of highly conserved amino acid sequences of available Helicobacter and C. jejuni cGTs. Standard PCR was done in a 50 mL reaction mixture containing 50 ng of DNA template, 25 pmol of each primer and 1.25 U of DyNAzyme TM (Finnzymes Oy, Espoo, Finland). The 59 consensus  clamp region contains the sequencing primers used for direct sequencing of the PCR products. If secondary peaks or multiple sequences were present, the PCR products were inserted into pGEMH-T and sequenced on both strands using M13 primers.
The area corresponding to nucleotides 702-1569 (amino acids 234-523) of bgh1 was used to design specific primers for RFLP-PCR protocol. The PCR was done in a 50 mL reaction mixture, as described above, and then the products were digested by BclI (NEB, New England Biolabs, Ipswich, MA, USA). BclI specifically cuts the PCR fragment in two pieces of 606 and 260 bp. EMBL accession numbers are listed in Table 3. All oligonucleotides used in this study are listed in Table 1.

Phylogenetic analysis
Phylogenetic analyses of cGT amino acid sequences were conducted in MEGA4 [53] using the Minimum Evolution (ME) method. The amino acid sequences were aligned using MAFFT [54]. The evolutionary distances were computed using the Dayhoff matrix-based method, and the ME tree was searched using the Close-Neighbor-Interchange (CNI) algorithm (level 1). All positions containing alignment gaps and missing data were eliminated in pairwise sequence comparisons [53]. A multi-alignment of ggt sequences of Helicobacter spp and C. jejuni was built in MEGA4 on the basis of the amino acid sequence alignment. Maximum likelihood (ML) trees were constructed using PHYML [55] with a nucleotide evolution model selected by FindModel (http://www. hiv.lanl.gov/). Trees were visualized by iTOL [56]. Nonsynonymous/synonymous substitution (Ka:Ks) was evaluated by sliding window analysis (window size of 50 bp) using SWAKK [37].

Expression of Bgh1 and Bgh2 and purification of recombinant proteins
Primers pairs were designed to amplify complete sequences of bgh1 and bgh2 genes using PhusionH High-Fidelity DNA Polymerase (Finnzymes). Bgh1 and Bgh2 were expressed as 66His-tagged proteins by inserting the genes respectively in NdeI and HindIII or NheI and HindIII restriction sites of pet28b+ vector (NovagenH, Merck KGaA, Darmstadt, Germany). The 66His-tag was fused in the N-terminal part of the proteins. The resulting expression constructs pMRg1 and pMRg2 were sequenced and confirmed to be identical to bgh1 and bgh2, respectively. Both expression constructs were used to transform E. coli Rosetta 2 (NovagenH). Expression of recombinant protein was induced with 0.5 mM IPTG for 4 h at 27uC. After centrifugation, pellets were solubilized in ice-cold binding buffer (50 mM Tris-HCl, 500 mM NaCl, 10 mM imidazole, pH 7.4) containing protease inhibitors and then lysed by subsequent sonication (5 times for 30 s) on ice. The supernatant was loaded on the HisTrap TM column (GE Healthcare Biosciences, Pittsburgh, PA, USA) at 1 mL/min at 4uC, and bound protein was eluted with an imidazole-gradient (0-100 mM imidazole) using the elution buffer (50 mM Tris-HCl, 500 mM NaCl, 1000 mM imidazole, pH 7.4). Eluates were collected and then tested for purity by SDS-PAGE. Since purification of a soluble form of Bgh1 was not possible, the sonicated cells of E. coli (4 times for 1 min) were homogenized, solubilized in 6 M Gua-HCl, 1 mM DTT, 10% Glucose pH 8.5 and stirred for 1 h at 4uC. Following centrifugation, the supernatant containing the unfolded protein was loaded on a HisTrapHP TM column (GE Healthcare) at 4uC. The bound protein was refolded with a gradient of 100 mM HEPES, 5% Glucose, 250 mM NaCl, 10 mM MgCl 2 , 1 mM DTT pH 8.5 (0-100% of HEPES buffer). After the refolding, the column was washed extensively with 5 column volumes of HEPES buffer, and bound protein was eluted with an imidazole-gradient (0-1000 mM) using HEPES buffer with 1000 mM imidazole. Eluate was collected and then tested for purity by SDS-PAGE. For further purification, fractions with the recombinant proteins from Ni-Sepharose affinity chromatography were pooled, dialysed overnight against 100 nM HEPES, 140 mM NaCl, 2.5% Sucrose pH 7.0, at 4uC and processed to the second and final purification step. The dialysed samples were loaded on a Superdex TM 75 column (GE Healthcare) and equilibrated with 100 nM HEPES, 140 mM NaCl, 2.5% Sucrose pH 7.0. All fractions collected were analysed by SDS-PAGE for the presence of recombinant proteins. Fractions containing the protein were pooled, aliquoted and stored at 280uC until further use.

Circular Dichroism (CD) spectrum of the recombinant Bgh1
To estimate the secondary structure of Bgh1 a CD spectrum of the recombinant protein was calculated. The measurements were performed on a JASCO J-810 spectropolarimeter equipped (Jasco Labor-und-Datentechnik GmbH, Gross-Umstadt, Germany) using a 0.2-mm-pathlength quartz-glass cuvette 1-mm. CD spectra were recorded with 50 ml of the sample at final concentration of 15 mM in 20 mM KH 2 PO 4 pH 7.5, 50 mM K 2 SO 4 . All CD measurements were corrected by subtracting the buffer spectrum. The following parameters were used: data pitch, 0.1 nm; continuous scanning mode at 100 nm/min; response, 4 sec; band width, 1 nm; accumulation, 16. Smoothing was made by the method Means-Movement. Data were collected using Spectra manager v1.54.03 (Jasco). For the western blot, H. bilis overnight culture was resuspended in 1 mL BHI (BD). Following lysis (sonication 4630 sec), the protein-concentration in cleared lysates was determined by BCA-Assay and adjusted to 1 mg/mL. Per lane 5 mg of total protein were loaded. After blocking with 5% BSA in Tris-buffered saline (TBS), the Whatman ProtranH Nitrocellulose membranes (GE Healthcare) were incubated for 1 h at RT with the sera of the immunized mice (1:800 diluted in TBS). Following washing, HRPconjugated secondary anti-mouse IgG antibody (Promega) (diluted 1:3000 in TBS) was added and detected by the acridan-based ECL-system (Thermo Fisher Scientific).

Immunization of mice and Western-Blot on whole H. bilis lysate
Complementation of E. coli Dggt bgh1 and bgh2 were amplified as described above and were inserted between KpnI and HindIII restriction sites of pBAD24 expression vector [57], resulting in pMRg3 and pMRg5, respectively. Both expression constructs were used to transform E. coli CY128 Dggt. The recombinant strains were grown in LB medium with ampicillin at 37uC to OD 600 <0.4, and the expression of the corresponding gene was induced with 0.2% Larabinose for 4 h at 25uC. The basal expression of pMRg3 and pMRg5 vectors was inhibited by using 0.2% of glucose instead of L-arabinose. Expressions were verified by SDS-PAGE electrophoresis and Coomassie Blue stain. A total of 10 9 cells were harvested by centrifugation and stored at 270uC before the cGT assay.

Construction of H. bilis isogenic mutant
Chromosomal inactivation of bgh2 was performed in H. bilis CCUG 23435. Deletion was introduced by allelic exchange using vector pUC119 in which ,1100 bp of the 59 -end and ,500 bp of the 39 -end of the target gene and the chloramphenicol resistance gene from pUOA14 [58] were cloned. The resultant plasmid, pMRg9, was constructed and amplified in E. coli DH5a and used as a suicide plasmid in H. bilis CCUG 23435. H. bilis mutant was obtained by electroporation as described for H. hepaticus [59]. The mutant strain, H. bilis MR9 (bgh2::cam), was selected on an HP medium containing 5% bovine blood supplemented with chloramphenicol (20 mg/mL). The site of recombination was verified by PCR.

RNA extraction and RT-PCR
To analyse the expression of both bgh1 and bgh2 during growth, H. bilis wild-type CCUG 23435 was cultivated in Brain Heart Infusion (BHI; BD, NJ USA) containing 20% FBS and Vitox supplement (Oxoid Ltd., Cambridge, UK) at 37uC microaerobically by continuous shaking at 150 rpm. OD 600 was measured at 8, 16 and 24 h and the same amount of cells was treated with RNAprotect Bacteria Reagent (Qiagen GmbH, Hamburg Germany). RNA was extracted using RNeasy Mini Kit (Qiagen) and treated with TURBO DNase TM (Applied Biosystems/Ambion, Austin, TX, USA). Moreover, H. bilis CCUG 23435 and the mutant MR9 were cultivated in YT62 medium (16 g/L Tryptone, 10 g/L Yeast Extract, 5 g/L NaCl) containing 10% FBS at 37uC in a microaerobic atmosphere. After 24 h RNA was extracted as described above. cDNAs were synthesized from 1 mg of total RNA using SuperScriptH III Reverse Transcriptase (Invitrogen) using random hexamers (Finnzymes). RT-PCR reactions were carried out in a total volume of 50 mL as described above, using 1 mL of cDNA and 1 mL of RNA as negative control and 20 pmol of specific primers for bgh1 and bgh2. Thermocycling conditions were 94uC for 2 min followed 40 cycles of 94uC for 45 s, 58uC for 45 s and 72uC for 1 min. The PCR products were sequenced using the same primers.

T-cell and AGS proliferation assay
For measurement of the effect of H. bilis cGT on T-cell proliferation, Jurkat cells were seeded at 5610 3 per well in a complete culture medium in 96-well microtiter plates and a variable amount of protein (0.001 to 10 mg/mL) or sterile-filtrated culture supernatant (1 mg/mL total protein) was added in a total volume of 100 ml. As a control for the inhibition of T-cell proliferation, the recombinant cGT from H. pylori 26695 (HP-cGT) was used. The HP-cGT was produced as previously described [9]. Additionally, supernatant were preincubated with 50 mM Acivicin for 30 min at 37uC to inactivate the cGT. Cells were incubated at 37uC for 48 h and analysed for proliferation as described previously [9]. Results are expressed as the mean6SEM from three experiments.
For the measurement of the effect of recombinant H. bilis and H. pylori cGT on proliferation of AGS cells, 5610 3 cells per well were seeded in culture medium in a 96-well plate and treated with 0.5-5 mg/mL of protein in a total volume of 100 ml. Cells were incubated at 37uC for 48 h and analysed for proliferation as described above.

Apoptosis in AGS cells
Apoptosis in gastric epithelial cells was determined by Fluorescence-activated cell sorting (FACS). AGS cells were seeded in a 6-well plate till they reached a confluence of 60-70%, and treated for 24 hours with 0.5-5 mg/mL of purified recombinant cGT proteins from H. pylori and H. bilis in DMEM with 10% of FBS. Untreated cells were used as a negative control and cells treated with 10 mM of Staurosporine (LC laboratories, Woburn, MA, USA) as a positive control for apoptosis. Cells were harvested with 0.25% Trypsin-EDTA (Invitrogen) and 2610 5 cells were washed once with PBS. The pellets were resuspended in 10 mg/ mL of AnnexinV-FITC antibody (BD, Becton, Dickinson and Co., NJ USA) in FACS buffer and incubated for 20 minutes in dark at room temperature. Cell suspensions were then diluted with FACS binding buffer and 10 ml of 50 mg/mL of Propidium Iodide (Sigma-Aldrich, Inc, St Louis, MO, USA) was added before acquiring samples. Cell debris was excluded by scatter gating (forward vs. side scatter). Approximately 50,000 events were collected per sample using CyAn ADP FACS cytometer (Beckman Coulter, Inc., CA, USA) and data was analysed with FloJo software (Tree Star, Inc, Oregon, USA). The percentage of apoptotic cells is expressed as the double positive cell population (AnnexinV+PI) and pre-apoptotic cells stained only for AnnexinV.

cGT enzymatic and competition assays
For determination of cGT activity, ,10 9 bacteria were suspended in 400 mL of reaction buffer containing 1 mM of L-cglutamyl-p-nitroanilide (gGpNA) as described previously [6]. Additionally, culture supernatants, containing 50 mg/mL protein were analysed by diluting 50 mL of supernatant in 150 mL of reaction buffer. The kinetic parameters for the enzymatic reaction of the recombinant proteins were measured by following the cleavage of gGpNA. To determine the exchange rate of the substrate to 4-p-nitroaniline, the reaction was monitored at 405 nm and calculated by using the reported extinction coefficient of 8800 M/cm. Enzymatic activity was measured in 0.1 M Tris HCl with a pH range from 5.0 to 9.0, 20 mM glycyl-glycine and gGpNA at a concentration range from 1 mM to 2.5 mM. The experiments were carried out at 37uC for 30 min using a Mithras LB940 Well-Reader. The initial rates were determined at each gGpNA concentration and entered in the Michaelis-Menten equation. To investigate the affinity of the postulated cGTsubstrates (i.e. L-Glutamine, D-Glutamine, L-Glutamic acid and Glutathione), we made a competition experiment in which the hydrolysis and transpeptidation of gGpNA was analyzed by measuring the absorption at 405 nm in the presence of cGTsubstrates. For this assay we used 250 ng of the Hb-cGT or Hp-cGT per reaction with 100 mM gGpNA, 10 mM GlyGly and the substrates in a concentration-range between 0-5 mM in 100 mM Tris/HCl pH 8.0. The measurement was performed for 15 min at 37uC in a total volume of 200 ml.