In Depth Analysis of the Helicobacter pylori cag Pathogenicity Island Transcriptional Responses

The severity of symptoms elicited by the widespread human pathogen Helicobacter pylori is strongly influenced by the genetic diversity of the infecting strain. Among the most important pathogen factors that carry an increased risk for gastric cancer are specific genotypes of the cag pathogenicity island (cag-PAI), encoding a type IV secretion system (T4SS) responsible for the translocation of the CagA effector oncoprotein. To date, little is known about the regulatory events important for the expression of a functional cag-T4SS. Here we demonstrate that the cag-PAI cistrons are subjected to a complex network of direct and indirect transcriptional regulations. We show that promoters of cag operons encoding structural T4SS components display homogeneous transcript levels, while promoters of cag operons encoding accessory factors vary considerably in their basal transcription levels and responses. Most cag promoters are transcriptionally responsive to growth-phase, pH and other stress-factors, although in many cases in a pleiotropic fashion. Interestingly, transcription from the Pcagζ promoter controlling the expression of transglycolase and T4SS stabilizing factors, is triggered by co-culture with a gastric cell line, providing an explanation for the increased formation of the secretion system observed upon bacterial contact with host cells. Finally, we demonstrate that the highly transcribed cagA oncogene is repressed by iron limitation through a direct apo-Fur regulation mechanism. Together the results shed light on regulatory aspects of the cag-PAI, which may be involved in relevant molecular and etiological aspects of H. pylori pathogenesis.


Introduction
Helicobacter pylori is a major human pathogen that colonizes the gastric epithelium of more than half of humankind, worldwide. Etiology of H. pylori infections ranges from mild to acute symptoms, including gastric inflammations and duodenal diseases. While symptomatic infections can be successfully treated with antimicrobial drugs, the untreated and asymptomatic infections persist over decades, promoting the long term chronic inflammation and insurgence of peptic ulcers and gastric cancer [1,2].
The severity of symptoms largely depends on the genetic diversity of the infecting strain [2], and particularly on specific genotypes of virulence-associated genes, such as the cag pathogenicity island (cag-PAI). The latter is a 38 kb multi-operon locus coding for 28 putative ORFs, six of which were identified as homologues of the basic type IV secretion system (T4SS) represented by the Agrobacterium tumefaciens virB operon [3]. The H. pylori cag-T4SS promotes injection of the CagA bacterial oncoprotein into host cells [4], as well as a CagA-independent induction of interleukin-8 secretion via the host AP-1 and NF-kB signaling pathways [5]. Notably, chemokine induction requires direct contact of the bacteria with epithelial host cells [6]. Host integrin receptors are engaged by the CagL protein to promote CagA translocation across the host cell membrane and to activate the Src tyrosine kinase for CagA phosphorylation [7], resulting in cytoskeletal rearrangements of the gastric epithelium [8]. Seventeen genes of the cag-PAI are essential for CagA translocation, including three genes HP0524 (virD homologue), HP0526, and HP0540 that are dispensable for IL-8 induction [9]. Moreover, several genes of the cag-PAI are not necessary for either CagA translocation or IL-8 induction, and their role in Cag-T4SS assembly and virulence remains to be elucidated [10,11].
Proper stoichiometric ratio of gene products and appropriate expression in time and space in the host allows the assembly of functional secretion systems [12]. Polarity effects modulating transcription and translation can contribute to this regulation, but in many bacteria the assembly requires finely tuned regulatory mechanisms. For example, it has been shown that dedicated transcriptional regulators are frequently encompassed within the PAIs of type III and type IV secretion systems. Transcriptional regulators such as Fur, HilA, HilC, HilD, InvF and ExsA, controlling Salmonella enterica pathogenicity island (SPI1) and Pseudomonas aeruginosa T3SS gene expression, respectively, are informative examples to date [13][14][15].
H. pylori is unusual in this respect. In fact, despite multiple operons with oppositely phased ORFs and frequent intergenic regions spanning .70 bp, which advocate the existence of multiple cag promoters, the cag-PAI lacks genes with annotated regulatory function [16]. In the past, only one promoter region of the divergent cagAB genes has been analyzed in detail [17]. In addition, it has also been shown that some of the H. pylori cag genes may be responsive to acid pH [18][19][20], or free iron [21][22][23][24],while others may be induced upon contact with the host cells [25]. However, little is known about the regulatory events behind these processes and the regulators that transduce these signals are not known.
As the expression of a functional cag-T4SS is intimately linked to the etiology of H. pylori, a systematic study of cag gene transcription will provide insights into timing and regulation of H. pylori virulence. Here we functionally characterize the main cag promoters and their transcriptional responses after different stress signals, demonstrating a direct regulatory role of apo-Fur on cagA and a consistent transcriptional induction of the cagfedc cistron upon H. pylori interaction with host cells.

Bacterial strains and growth conditions
All H. pylori strains used are listed in Table 1. Bacteria were recovered from 280uC glycerol stocks and propagated on BBLBrucella (BD) agar plates containing 5% fetal calf serum (Oxoid), 0.2% cyclodextrin, and Dent's or Skirrow's antibiotic supplement. Cultures were grown for 24-48 hours at 37uC in a water-jacketed thermal incubator (9% CO 2 , 91% air atmosphere, and 95% humidity) or in jars using CampyGen (Oxoid) gas-packs. Liquid cultures were grown in BBL Brucella Broth supplemented with 5% fetal calf serum and Dent's or Skirrow's antibiotic supplement at 37uC with gentle agitation (125 rpm), in glass or tissue-culture flasks with vented cap. When required, Brucella agar plates or liquid broth were supplemented with chloramphenicol (30 mg ml 21 ) and kanamycin (25 mg ml 21 ). H. pylori transformants were obtained by double homologous recombination of the naturally competent G27 strain using 5 mg of transforming DNA, as previously described [26]; positive clones were selected on Brucella agar plates supplemented with chloramphenicol, according to the resistance phenotype conferred by the cat cassette (Cm R ). E. coli DH5a cultures for cloning purposes were grown in Luria-Bertani broth. Ampicillin (100 mg ml 21 ), chloramphenicol (30 mg ml 21 ) and kanamycin (25 mg ml 21 ) were added when required.

DNA manipulations
DNA amplification, restriction digests and ligations were all carried out with standard molecular techniques, with enzymes purchased from New England Biolabs. Large preparations of plasmid DNA were carried out with a NucleoBond Xtra Midi plasmid purification kit (Macherey-Nagel). DNA fragments for cloning purposes were extracted and purified using Qiaquick Kits (Qiagen, Inc.)

Construction of an arsSmutant
The H. pylori G27-derivative arsS knock-out mutant was obtained replacing the arsS gene (ORF HP0164HP0165) from position 3 to position 1290 of the coding sequence with a Campylobacter coli cat chloramphenicol resistance cassette (cat) by double homologous recombination using the pBS::DarsS suicidal vector. Primers 163f_Xba and 163r_Bgl (Table 2) were used to amplify and clone a 460 bp XbaI-BglII fragment encompassing the region upstream of arsS, corresponding to 359 bp of the 59 region of the HP0163 open reading frame (ORF), 9 bp of the intergenic region and the 57 bp of the 39 region of HP0164. Primers 166f_Bgl and 166r_Hin were used for amplification and cloning of the arsS downstream region: a 616 bp BglII-HindIII fragment carrying 585 bp of the 39 region of the HP0166 ORF and 25 bp of the intergenic region upstream the HP0165 ORF. The cat cassette derived as BglII-BamHI fragment from pBS::cat was inserted between these two fragments and the final construct pBS::DarsS was used to transform H. pylori. The chloramphenicolselected mutant strains were confirmed by PCR.
Generation of Pcag-lux and Pcag-59UTR-lux reporter strains H. pylori G27-derivative strains carrying the transcriptional fusions of the Pcag promoter regions (with or without the 59 untranslated regions) with the luxCDABE reporter operon were obtained as described previously [27]. Briefly, the Pcag promoter regions were PCR amplified from H. pylori G27 genomic DNA, digested and cloned into the unique BamHI site of the vector pVCC. Constructs carrying the insert in the desired orientation were identified by digestion and checked by sequencing, then the plasmids were used to transform G27lux acceptor strain carrying the promoterless Photorhabdus luminescens luxCDABE operon in the vacA locus. The chloramphenicol-selected mutant strains were expanded and the correct insertion was confirmed by PCR. pVCC-derivative constructs used for H. pylori transformation and the corresponding mutant strains are reported in Table 1.

RNA preparation
To measure metal-dependant transcriptional responses, cultures of the wild-type and mutant strains were grown to mid-log or latelog phase (optical density at 600 nm [OD 600 ] 0.5-0.6 and 1.7, respectively) and treated for 30 minutes with either 1 mM (NH 4 ) 2 Fe(SO 4 ) 2 , 1 mM NiSO 4 , or 100 mM 2,2-dipyridyl chelator (Sigma-Aldrich) prior to RNA extraction. For acid exposure experiment, mid-log cultures (OD 600 0.5-0.6) of the wild-type and mutant strains were divided in two subcultures and treated with either 1 M HCl to adjust the pH from 7.0 to 5.2 (acid shock) or equal volume of sterile water (control sample). Subcultures were grown for 30-90 minutes before RNA extraction. The volume of 1 M HCl required to achieve a pH of 5.2 was determined on aliquots of the growing cultures. To follow the expression of cag genes over time, an overnight culture of wild type strain was diluted to a starting OD 600 of 0.08 and cultured to an OD 600 of 1.75 for approximately 15 hours. Aliquots of 10 ml from this master culture were harvested at different time points for RNA extraction. Heat-shock experiments were performed as previously described [28]. Total RNA was extracted using a hot-phenol procedure [23]. RNA integrity and purity were ensured by electrophoresis on 1% agarose gels.

Primer extensions
Primer extension analyses were performed with 12 mg of total RNA and 0.1 pmol of 59-end-labeled primers as described previously [29]. The oligonucleotides used for primer extension reactions are listed in Table 2. Quantification of the signals from extension products obtained was performed using a Storm phosphor-imager (Amersham-GE) and Image Quant Software (Molecular Dynamics).

Overexpression and purification of recombinant His 6 -Fur
Recombinant His 6 -Fur was overexpressed and purified under native conditions [26]. Thrombin protease (10 U/mg) was used to remove the N-terminal histidine tag according to the instructions

DNase I footprinting
Plasmid pGEM-PcagAB used for the generation of the footprinting DNA was obtained by cloning the 403 bp Lux547R-Lux546F amplicon in pGEM-T Easy (Promega). 1 pmol of pGEM-PcagAB was linearized with NcoI, dephosphorylated with calf intestinal phosphatase and labeled at the 59 ends with [c-32 P]ATP (6,000 Ci/mmol; PerkinElmer) and T4 polynucleotide kinase (all enzymes by New England Biolabs). The labeled DNA probe was further digested with SalI and the products were separated by native polyacrylamide gel electrophoresis and purified as described previously [29]. The binding reactions between approximately 20 fmol of labeled probe and increasing concentrations of Fur were carried out at room temperature for 15 min in a final volume of  and 5 mM MgCl 2 was added to the reaction mixture and digestion was allowed to occur for 85 s. The reaction was then stopped, and the samples were extracted and purified [26]. Samples were resuspended in 5 ml of formamide loading buffer, denatured at 100uC for 3 min, separated on 8 M urea-6% acrylamide sequencing gels and autoradiographed.
AGS cell culture and infection assay AGS cells, a human adenocarcinoma epithelial cell line (ATCC CRL 1739), were grown in RPMI-1650 medium with 10% fetal bovine serum (FBS) in tissue-culture flasks. For the infection assay, cells were seeded in 24-well plates (Orange Scientific) and cultured for 1-2 days to reach 60-80% confluence. Before the infection, the wasted medium was replaced with fresh RPMI-1650 with 5% FBS conditioned in the bacterial incubator (9% CO 2 , 91% air atmosphere, and 95% humidity). Cells were infected with G27 Pcag-lux strains at a multiplicity of infection (MOI) of 5, while other 24-well plates filled with medium but without AGS cells were infected with the same amount of bacterial culture and used as control sample. The plates were placed inside the bacterial incubator and luminescence was measured at regular time intervals with Victor3V (1420) multilabel reader (Perkin Elmer), with bottom trail pre heated at 37uC. Luminescence was measured with an integration time of 2 seconds (normal aperture) in the absence of optical filters. The luminescence values of wells filled with plain growth medium were used as blank control and subtracted from the values of the experimental samples. Each infection assay was performed in quadruplicate and the assay was repeated in four independent biological replicates. average values and standard deviations were calculated.

Mapping of cag promoters
Recent studies have provided insights on the transcriptional organization of the H. pylori cag pathogenicity island, with the mapping of transcriptional start sites (TSS) and the identification of putative promoter regions. In strain 26695, out of 40 putative 59-end of RNA transcripts identified [20], 14 map within the 300 bp upstream of annotated ORFs, and are predicted to contain the promoter regions of Cag protein coding sequences. These results were recently confirmed in different strains by promotertrap and reverse transcription analyses of ORFs and intergenic regions [30]. The positions of the 59 end of RNA transcripts and transcriptional units identified are schematically represented reported in Fig. 1A. To study their regulation we set out to map the 59 end of these transcripts by primer extension analyses on total RNA extracted from H. pylori strain G27 grown to mid-log phase using oligonucleotides mapping downstream of the 14 aforementioned predicted promoters (Fig. 1A, Table 2). The remaining 26 internal and antisense TSSs deserve more dedicated studies and have been deliberately excluded from the current study.
To evaluate the transcript levels at the cag promoters and compare their relative mRNA abundance, we used H. pylori G27 isogenic strains harboring the transcriptional fusions of the cag promoters with a lux reporter system [27]. For each of the 11 cag promoters, the region upstream of the transcriptional start site was placed upstream of the promoterless lux reporter operon, generating 11 Pcag-lux strains (Table 1). Total RNA was extracted from the Pcag-lux strains grown to mid log phase and used to quantify transcript levels by primer extensions with a common lux specific primer, with data normalized to the mRNA level at the Pcagf promoter (Fig. 1C). For the sake of clarity we will synthesize the results by subdividing the cag genes/operons in three different functional classes: i) structural cag genes or operons coding for components forming the structural core of the secretion system, essential for CagA translocation and IL-8 induction (dark grey block arrows in Fig. 1A); ii) effector cag genes coding for secreted components responsible for cytotoxic effects on host cells, such as CagA (white block arrow); and iii) accessory cag genes or operons, which may have modulatory or ancillary functions (light grey block arrows).
Interestingly, the promoters of operons encoding structural Cag components (Pcagf, PcagV, PcagU, PcagM, PcagF, PcagC), essential for a functionally assembled cag-T4SS, are all characterized by a reasonably conserved -10 box in addition to an extended TG element or -35 box (Fig. 1B). These promoters display similar basal transcription levels, suggesting that the messenger RNAs of these structural operons are generated with similar stoichiometric ratios. On the other hand, the PcagA promoter responsible for the expression of the CagA toxin, has a perfect match to the -10 TATAAT box in addition to an extended TGn, and a putative UP element [17], which together appear to confer significantly higher transcription levels with respect to the operons encoding structural components (Fig. 1C). This observation is congruent with the different (effector) role for the cagA gene product, in accordance with previous observations [30]. The promoters of monocistronic operons encoding putative accessory components (PcagS, PcagQ, PcagP, PcagB), not essential for T4SS function, exhibit conserved -10 boxes, but lack extra elements. Interestingly, they denote extremely disparate basal transcription levels, with PcagQ and PcagP matching or even exceeding the high basal levels of cagA transcripts (intriguingly PcagP lacks conserved -35 or TG elements, tentatively indicating at a dedicated activator involved in the high transcription levels). This suggests that the genes under the control of these promoters are actively expressed, and likely play an important functions for the cag secretion system. These transcript levels verified in the G27 strain, show some differences compared to the cag transcript levels reported in other H. pylori strains [20,[30][31][32]. In particular, PcagC and Pcagf were previously reported with a 10-to 1000-fold higher expression level with respect to the other structural promoters in the C57 and 26695 strains [33,34], suggesting strain-specific transcriptional variations.

Growth-phase regulation of the cag promoters
To study the transcriptional regulation of the selected cag promoters during growth, we carried out time course experiments. Aliquots of bacterial cultures were collected at different time points and used to extract total RNA for quantitative primer extension experiments at the 11 cag promoters (Fig 2).
Transcription from the Pcagf, PcagV, PcagF and PcagA promoters showed no significant variation in the amount of mRNA during the early exponential growth stages of the bacteria, while their amount increased in late logarithmic growth phase with an up to five-fold increase of transcript levels from Pcagf, PcagV, PcagF and PcagA ( Fig. 2A). By contrast, transcription from PcagS, PcagP, PcagM and PcagB promoters exhibited a progressive reduction of mRNA levels during the time course experiment, reaching up to ten-fold down-regulation at OD = 1.7 (Fig 2B). No significant variations of transcripts deriving from PcagU, PcagQ and PcagC were recorded during the same time-course experiment (Fig. 2C).
We conclude that during bacterial growth, transcription from Pcagf, PcagV, PcagF and PcagA promoters increases at late logphase, in agreement with previous observations [32,35,36], while transcription from PcagS, PcagP, PcagM and PcagB promoters is decreased. These observations suggest that the transition from exponential to stationary phase prompts a modulation in the expression of the CagA toxin and of specific structural components of the T4SS, which together may impact on the assembly or function of the secretion system. In other pathogenic bacteria, such as Brucella abortus and Legionella pneumophila, the regulation of essential components of the T4SS is growth-phase dependent [37] or quorum-sensing responsive [38]. Thereby, the virulence mechanisms are fine-tuned according to the bacterial load and the nutrient availability in the host niche. The growth-phase dependent regulation of particular cag promoters indicate that H. pylori may adopt similar strategies to control virulence, as cagspecific responses to stress signals encountered in the host niche after infection are frequently mimicked by the stationary phase conditions of planktonic laboratory cultures [31].

Environmental regulation at the cag promoters
To study the transcriptional regulation of cag promoters in response to environmental changes, we exposed exponentially growing cultures of H. pylori G27 strain to various stress conditions that challenge the bacterial metabolism or fitness. Total RNA was extracted from treated and untreated samples and transcript levels at the cag promoters were assayed by quantitative primer extensions with cag-specific oligonucleotides ( Table 2). In bacterial cultures exposed to heat shock (30 min at 42uC) we observed a 6to 40-fold reduction of mRNA levels at most cag promoters (Fig. 3A). Exceptions to this finding were at the Pcagf and PcagA promoters that showed unchanged transcript levels (Fig. 3A). Subsequently, we assayed the mRNA levels at all cag promoters in H. pylori strains deleted of the heat shock transcriptional regulatory genes hspR and hrcA [39]. In comparison to the wild type strain, the knock-out DhspR and DhrcA mutants grown in normal conditions or exposed to heat shock treatment showed similar mRNA levels at cag promoters (data not shown). Thus, the observed variation in the mRNA levels after heat shock is not under the direct control of HspR or HrcA, likely reflecting a pleiotropic effect on transcription or mRNA stability.
Bacterial cultures treated with iron (30 min with 1 mM (NH 4 ) 2 Fe(SO 4 ) 2 ), iron chelator (30 min with 100 mM 2,2-dipyridyl), or nickel (30 min with 1 mM NiCl 2 ), showed no significant variations in the transcript levels from most of the cag promoters (data not shown). Exception to this finding was observed at the PcagA promoter that showed a slight increase in the RNA levels after exposure to iron ions and 1.5-fold reduced levels in irondepleted cultures (Fig. 3B -upper panel). This iron-dependent response is in agreement with previous studies [22].

Fur-dependent iron-inducibility of cagA
To further investigate on iron-dependent regulation, we assayed the mRNA levels at the cag promoters in the knock-out mutant of the iron-dependent regulator Fur, exposed to the same conditions as the wild type strain. In RNA extracted from the mutant culture strain we observed unchanged transcript levels at the PcagA promoter (Fig. 3B -middle panel), as well as at the other cag promoters (data not shown). The loss of the iron-dependent response of the PcagA promoter in the Dfur mutant strain suggested that Fur can mediate the iron-dependent regulation at this promoter.
Since intracellular Fur increases during bacterial growth [23], we assayed the iron-dependent response of the PcagA promoter in wild type and Dfur cultures grown to late log-phase (OD 600 = 1.7), with results reported in Fig. 3B (bottom panel). As expected, in the wild type background the PcagA promoter was iron-regulated, with markedly higher differences in the mRNA levels between ironreplete and iron-depleted conditions, while in the Dfur strain, the transcript levels were unchanged. These results suggested that Fur represses the PcagA promoter in response to iron starvation, likely through a direct mechanism. Albeit related indications were previously reported [21,22,40], together with extensive Fur binding within the cag-PAI [23], the mechanism behind the ironinducibility of cagA had not been understood in detail so far.
Thus, to demonstrate direct Fur-promoter interaction and map its operators sites, we set up DNaseI footprinting assays using the PcagA-PcagB intergenic region as probe, both in iron-replete (holo-Fur) and iron-depleted (apo-Fur) conditions. The protection pattern of holo-Fur on the PcagA-PcagB probe (Fig. 3C) shows four areas of DNaseI protection (marked I-IV in Fig. 3C): two high affinity binding sites (holo-operators; II and III) appear at the minimal protein concentration used (21 nM, lane 2), while two low affinity binding sites (I and IV) appear a higher Fur concentrations (84 nM, lane 4). Footprinting of the apo-Fur on the same probe showed a swap of the protection affinities, with two high affinity binding regions (apo-operators I and IV, Fig. 3D) and two lower affinity binding regions (II and III). Both the apooperators and the holo-operators respectively encompass sequences similar to the TCATT-n10-TT and TAATAATnATTATTA consensus motifs recently proposed for discriminative apoand holo-Fur binding [40][41][42]. Given the position of the apo-Fur binding sites with respect to the cagA transcriptional start site and the Furdependent iron-response in stationary phase cultures, we propose that the PcagA promoter is directly regulated by an apo-Fur repression mechanism involving the occlusion of the -10 box to the RNA polymerase. In fact, the position of these boxes suggest a repressor role exerted by apo-Fur, at least on PcagA, in agreement with the apo-Fur-dependent repression of its transcript (Fig. 3B). This mechanism is similar to the FeON regulation mechanism described for the pfr promoter [26,42], and likely important for the repression of the CagA toxin when the intracellular concentration of Fe 2+ ions are limiting. Moreover, the position of the high affinity binding site of holo-Fur upstream the -35 region of the PcagA promoter may indicate that the holo-form of the protein could positively regulate the PcagA promoter, with a class II activation mechanism. Hence, the Fur-mediated regulation of PcagA is dependent by a complex binding of holoand apo-Fur proteins on the corresponding operators. Recent studies suggest that iron Heat-shock response of the Pcag promoters. Primer extension analyses were performed on total RNA extracted from bacterial cultures of H. pylori wild type strain grown to exponentially phase and maintained at 37uC or exposed to 42uC for 30 min. B. Iron-dependent regulation of the PcagA promoter. Liquid cultures of wild type and Dfur strains were grown to OD 600 = 0.5 or OD 600 = 1. limitation may increase cag-dependent virulence [43]. Thus it is possible that additional regulatory mechanism concur to regulate the functionality of the cag-T4SS. For example, previous footprinting analysis with the a-subunit of the RNA polymerase showed a protection pattern on the region spanning from -17 to -70 nucleotides of PcagA [17], suggesting the presence of an UP element recognized by the CTD domain of RpoA. We can speculate that the observed iron-dependent regulation of PcagA could be exerted not only by the binding of apoand holo-Fur to its operator elements, but also by Fur competing for the binding of thea-subunit to the UP element. These evidences add to the documented importance of Fur as central regulatory hub in the H. pylori pathogenesis.

Acidic shock response of cag promoters
To investigate the transcriptional responses to acidic pH, liquid cultures of H. pylori grown to mid-log phase were divided in two subcultures and treated for 30 min or 90 min either with HCl to adjust the pH of the medium to a value of 5.2 (acid shock) or with the same volume of sterile water (untreated sample). The RNAs extracted from three independent cultures were assayed by primer extension experiments and bands were quantified with results reported in Fig. 4.
Upon 30 min acidic treatment, transcript levels from the Pcagf, PcagU, PcagF and PcagA promoters increased a 3.8-, 2.3-, 3.2-and 1.9-fold, respectively (Fig. 4A). In contrast, transcript levels from the PcagS and PcagB promoters decreased 3.1-and 6.2-fold, respectively (Fig. 4B), while no significant variation in the mRNA levels was observed at the PcagV, PcagQ, PcagP, PcagM and PcagC promoters (Fig. 4C). After a 90 min acidic shock treatment, most of the promoters showed a pattern of RNA accumulation similar to the 30 min treatment (data not shown). Exceptions were observed at the PcagV, PcagQ and PcagC promoters for which the mRNA level increased to 1.7-, 3.2-, 6.3-fold, respectively (Fig. 4D). Thus, almost all operons coding for proteins essential for the formation of a functional T4SS are inducible by low pH, including PcagA, PcagV (core), PcagU (core), PcagC (pilum), PcagF (pilum stabilization), Pcagf (transglycosylase and core stabilization). These results are in agreement with previous studies, showing pleiotropic responses of the cag promoters to acidic stress in different H. pylori strains [18,19,44,45]. One exception is represented by cagM, expressing a gene product involved in the stabilization of the T4SS core, whose transcript levels appear unchanged upon acidic treatment. On the other hand, the operons coding for unessential, ancillary cag components (PcagS, PcagQ, PcagP, PcagB) exhibited distinct responses, with PcagS and PcagB being repressed by acidic treatment. Interestingly, these promoters respond to acid treatment, and are co-regulated in the stationary phase of growth, tentatively pointing at a common function, which deserves to be further investigated in the future.
As the acidic-response in H. pylori is primarily controlled by the ArsRS two-component system, together with the metal responsive transcriptional regulators NikR and Fur [2,24,44,46], we cultured wild type, Dfur, DnikR and DarsS strains to mid-log phase, exposed to acidic shock for 30 min and evaluated the mRNA levels at the acid-responsive Pcagf, PcagU, PcagF, PcagA, PcagS and PcagB promoters (Fig. 4 A and B) by quantitative primer extension assays with results reported in Fig. 5. Intriguingly, PcagF and PcagS promoters showed a loss of the pH-inducible response in the DnikR mutant, displaying unchanged transcript levels after acidic treatment with respect to the untreated sample, while in the Dfur and DarsS mutants an acid response similar to the wild type strain was observed. Likely, the acidic response at these promoters is directly or indirectly mediated by NikR. Similarly, transcript levels at the PcagB promoter were unchanged after acidic treatment in the Dfur mutant, while the wild type strain and the other mutants showed a pH-induced reduction in the mRNA levels. These results suggest that Fur is involved in the acid-dependent repression of PcagB. On the other hand, Pcagf appears to loose the pH-inducible response both in fur and nikR knockout strains, suggesting a role for both regulators on its acidic regulation. Finally, variations of transcript levels in the mutant strains similar to that in the wild type strain were observed at the PcagA and PcagU promoters, hence acid response of these promoters is mediated by still unknown factors.
To further investigate on the observed loss of the acidic response of some Pcag promoters in the mutant strains, we performed footprinting assays of recombinant Fur protein on a labeled probe encompassing the Pcagf promoter, while footprinting assays of recombinant NikR were performed on the DNA fragments corresponding to the Pcagf, PcagS, PcagF promoters. No patterns of DNaseI protection were observed on these probes (data not shown), suggesting that NikR and Fur mediate the acid responses at these promoters indirectly.
Pcag response to bacterium-host contact Host cell contacts are potent elicitors of secretion system gene expression in pathogenic bacteria [15]. To assess the possible in vivo effects exerted by bacterium-host contacts on the transcription of the cag promoters, we used co-cultures of AGS cells and H. pylori G27-derived strains carrying the Pcag-lux transcriptional fusions. Bacterial cultures were grown to mid-log phase and used to infect AGS cells cultured in 24-well plates (AGS + sample), while same amounts of bacterial cultures were added to plates containing only the medium (AGSsample). During a time-course experiment, we measured the luminescence of the samples at regular time intervals, and for each time point we calculated the ratio of the signal from the bacteria grown in presence or absence of AGS cells (AGS + /AGSratio). The Pcagf-lux strain exhibited a significant increase of luminescence when co-cultured in the presence of AGS cells, with an AGS + /AGSratio that increased over time (Fig. 6). In contrast, the other Pcag-lux strains showed no significant differences between samples cultured with or without the AGS cells, with an AGS + /AGSratio unchanged during the experiment, as exemplified by the PcagQ-and PcagB-lux strains (Fig. 6). We can conclude that under the experimental conditions tested, the interaction of H. pylori with its host cells exerts a positive transcriptional effect only on expression levels at the Pcagf promoter. Previous studies indicated that contact of H. pylori with host cells provokes the increase of visible T4SS pili extruding from the bacterium at the host-pathogen interaction surface [47], and that the protein composition of the cag-T4SS pili differs if bacteria are grown planktonically or in co-culture with AGS host cells [48]. The finding that the interaction with host cells rapidly induces the transcription of the Pcagf promoter is, therefore, particularly striking. In fact, the cagfedc operon encompasses cagd, a gene that codes for a factor bridging the periplasm across the inner and outer membrane, essential for the stabilization of the T4SS core, as well as cagc, encoding the transglycosylase involved in the local hydrolyzation of the murein layer important for the formation and extrusion of the assembling T4SS. Interestingly, Kim and colleagues reported similar variations of cagd expression in H. pylori 69a strain, co-cultured with AGS cells for 1 hour [25]. These evidences suggest a conserved regulation of the operon, likely due to the modulation of the Pcagf promoter activity. Together, the results indicate that Pcagf induction may modulate the number of pili, their distribution on the bacterial cell surface and their composition after host cell contact. Previous observations of the AGS-induced regulation of other cag promoters (e.g. cagA, cagP and cagS) [20,49], were not confirmed in this study, possibly due to strain-specific responses to host-cell contact, or due to the different reporter system used to monitor the responses.

Post-transcriptional regulation
The analyses of the sequences downstream the transcriptional start sites of the Pcag promoters showed that cag transcripts harbor 59 untranslated regions (59UTRs) of different lengths (Fig. 1B). To assess possible post transcriptional effects mediated by the Pcag 59UTRs, in analogy to similar findings reported in A. tumefaciens [50], we used the aforementioned Pcag-lux reporter fusions and ad hoc Pcag-59UTR-lux constructs encompassing also the 5'UTRs downstream of the promoters (Fig. 7A). The luminescence emitted by mid-log growing cultures of these reporter constructs was compared to the corresponding Pcag-lux constructs without the 59UTR region. Except for Pcagf, PcagM, PcagF and PcagC, the luminescence counts of the 5'UTR-less constructs correlated well with the transcript levels assayed in primer extension analysis ( Fig.1C; 7/11 promoters matching). We observed a nearly 1:1 signal ratio between the constructs with or without the 59UTR for Cultures were grown to exponential phase and exposed to acid-shock (pH = 5.2) for 30 min. Transcript levels at the Pcagf, PcagU, PcagF, PcagA, PcagS and PcagB promoters were assayed by quantitative primer extensions. Asterisks mark the significant differences of n-fold variations deriving from the average band intensity of three independent primer extension experiments. Error bars indicate the standard deviation. doi:10.1371/journal.pone.0098416.g005 PcagA, PcagB, PcagC, PcagM, PcagS and PcagP (Fig. 7B), suggesting that most 59UTRs downstream of cag promoters do not affect the stability or the translational efficiency of the nascent messenger RNAs. Intriguingly, the luminescence of PcagV-59UTR-lux constructs decreased significantly with respect to the 59UTR-less construct (Fig. 7B), suggesting that this sequences could contain elements that reduce the translational efficiency or decrease the mRNA abundance.
On the other hand the Pcagf-, PcagU-, PcagQ-, PcagF-59UTR-lux constructs showed a strong increase of luminescence with respect to the corresponding 59UTR-less constructs (Fig. 7B), suggesting that the 59UTR downstream of these promoters could contain elements that enhance mRNA stability or translation.  . Comparison of the Pcag-lux and Pcag-59UTR-lux reporter signals. A. Schematic representation of the Pcag-lux and Pcag-59UTR-lux fusion constructs, obtained transforming the G27lux acceptor strain with the PVCC vector. The promoter sequences with or without the 59untranslated regions (59UTRs) carried by the pVCC vector are inserted upstream the luxCDABE operon by double homologous recombination and selected by cat chloramphenicol resistance. B. Luminescence signals from three independent experiments were normalized according to the optical density of the cultures and the means values were reported in the graph, with Pcag-lux signals on the X-axis and Pcag-59UTR-lux signals on the Y-axis. Error bars indicate the standard deviation. A dashed line was added to the graph, corresponding to the 1:1 ratio of the two signals. Grey dots: cag promoters with 1:1 signal ratio; black dots: cag promoters with altered Pcag-lux/Pcag-59UTR-lux signal ratio. doi:10.1371/journal.pone.0098416.g007

Conclusions
In conclusion, the general picture emerging for the cag-T4SS regulation can be synthesized as follows: i) despite the lack of dedicated transcriptional regulators encoded within the PAI, the cag cistrons appear to be subjected to a complex network of direct and indirect regulations; ii) operons coding for structural components of the T4SS display homogeneous transcript levels; they are transcriptionally responsive to growth-phase, and indirectly responsive to pH and other stress-factors. In some cases (PcagV, PcagQ, Pcagf and to lesser extent PcagU, PcagF) they are subjected to post-transcriptional control; iii) only Pcagf transcription appears to be triggered immediately after host-cell contact; iv) the cagA effector gene is highly transcribed, and matches the responses to acidic pH and bacterial growth phase together with other co-regulated operons encoding structural cag components. On the contrary, it is not responsive to heat stress as most of the other cag operons, while it is clearly induced by iron in a direct apo-Fur-dependent regulation mechanism; iv) monocistronic operons encoding accessory factors vary considerably in their basal transcription levels and responses, indicating non-constitutive expression of their components which may be involved in physiologically relevant aspects of cag-T4SS maturation and assembly. For example, previous reports indicate that cagP products may be involved in H. pylori adherence to host-cells [51]. The finding that cagP exhibits the highest transcript levels of the cag-PAI, and that it is down-regulated in stationary phase, when the transcription levels of structural and effector cag operons increase, provides an exemplification on how the dissection of cag transcriptional responses may guide our efforts to understand the cag-T4SS function. This knowledge will have important outcomes for the appropriate management of H. pylori infections, as the cag-T4SS is among the most important pathogenetic factors that carry an increased risk for gastric cancer [52].