In this study we demonstrate a significant, immediate response from AGS cells to the exposure to a H. pylori strain obtained from a clinical setting. More than 6000 human genes showed statistically significant differential regulation during the first 24 h of co-incubation.
H. pylori infection has been associated with both stimulation and inhibition of apoptosis. Some cell culture experiments demonstrate up-regulation of genes associated with apoptosis [7, 8], whereas some in vivo studies demonstrate proliferation and apoptosis inhibition [9, 10]. VacA toxin has been shown to cause apoptosis in several studies [30–33], whereas the role of CagA is conflicting. CagA has been associated with both stimulation and inhibition of apoptosis [11, 12, 34]. Biliary cells exposed to cagA
H. pylori at a very low inoculum (MOI 1:1) demonstrated increased cell growth, whereas at MOI of 200:1, apoptosis was stimulated . CagA may even directly antagonize the pro-apoptotic effect of VacA, as seen in AGS cells . Apoptosis occurs after a number of cellular events, leading to activation of caspase-3, which is thought to constitute the basic effector of apoptosis. In the present study, both inhibitory and stimulatory genes showed significant differential expression, demonstrating the complexity of the influence of H. pylori on apoptosis: caspase inhibitors HSPA5 and DHCR24 showed similar late down-regulation as heat shock genes HSPA1B, HSPB1, which are also associated with apoptosis stimulation (cluster E, Table 3). On the other hand, TNFAIP3, BIRC2, BIRC3 and SERPINB2, also associated with apoptosis inhibition, demonstrated early and persistent up-regulation grouped together in cluster A. However, positive regulators of apoptosis PTPRH, TNFRSF12A, IL24, GADD45A, TRIB3, DDIT4, PHLDA4, PP1R15A and SQSTM1 were all up-regulated in similar pattern after 6-12 h (cluster C). MCL1, an anti-apoptotic gene expressed in response to CagA injection , demonstrated increasing up-regulation over the course of the study. There were no significant changes in BCL-2 and very little increase in BAX expression in our study, two important genes that determine the sensitivity of cells to other apoptotic stimuli [36–39]. Noteworthy, there was marked up-regulation of TP53BP2, an important tumor suppressor gene (TSG) in human cancer, primarily stimulating p53 promotion of apoptosis genes. On the other hand, TP53BP2 is coding ASPP2 protein, which has also been shown to stimulate apoptosis independently of p53 [40–42]. However, Buti et al. recently demonstrated that CagA injected into gastric epithelial cells targeted ASPP2 protein to inhibit p53-mediated apoptosis . The increased TP53BP2 expression seen in our study, might therefore potentiate this effect by increasing the CagA-ASPP2 interaction to cause increased inhibition of p53-mediated apoptosis. In fact, the current study showed that p53 target genes involved in apoptosis  such as FAS, DR4, TNFRSF10B (also referred to as DR5/KILLER), DCR1, DCR2, P53AIP1, CASP6, APAF1 and BNIP3L did not show any significant increase, and BNIP3L, CASP6 and APAF1, BID and BAX showed only little increase. p53 target genes regulating non-apoptotic cellular processes including MDM2, GADD45A, CDKN1A (also known as P21 WAF1/CIP1), EGFR, CCND1, CCNG2 and TGFA demonstrated moderate to marked up-regulation. This differential gene expression identified among the p53 target genes in this study, may indicate selective inhibition of p53-mediated apoptosis due to increased CagA-ASPP2 interaction, consistent with Buti's findings.
Nevertheless, this study was not designed to assess whether the overall sum of inhibitory and stimulatory signals facilitated apoptosis or proliferation of epithelial cells. The current results illustrate the complexity of apoptosis regulation in epithelial cells in response to H. pylori exposure, and the cluster analysis suggests that there is some biological coordination of gene expression regulating apoptosis. This may explain some of the complex carcinogenic mechanism of H. pylori in gastric adenocarcinoma. There is strong association between H. pylori infecton, in particular the cagA
+ genotype , and gastric adenocarcinoma [45, 46], and also other cancers have been suggested to harbour a role for H. pylori [47, 48]. Furthermore, the present study shows that several cancer-related KEGG pathways are impacted in AGS cells during 24 h of cagA
H. pylori infection, in particular pathways in cancer, bladder cancer, prostate cancer, small cell lung cancer and the MAPK pathway. Several individual oncogenes and cancer related genes were also increased during, and at the end of the study, including ANGPT2, CEBPB, ECGF1, MMP7, MMP10, JUN, FOSB, EGFR, CTNNB1, ANXA1, CD55, CLDN1, KLK6, KRT7, LCN2, MYC, PIM1, PIM2, PIM3 and ATF3.
IL-8 appears paramount in the acute inflammatory response to H. pylori infection, as this gene is involved in all significant response pathways in the initial cellular response to infection. Several authors have demonstrated increase in IL-8 in response to H. pylori in both in vivo  and in vitro [50, 51] studies. IL-8 is a key chemokine in accumulating neutrophils. Gastric mucosal IL-8 levels have shown a positive correlation with the degree of stomach corpus inflammation , and IL-8 is also highly increased in gastric cancer [53, 54]. Our findings are supported by other authors who have demonstrated that IL-8 mRNA in vitro peaks between 2 and 4 h before decreasing over the next hours under similar conditions [55, 56]. Protein studies have shown steady state IL-8 levels after 3 h [50, 57, 58], which is also in harmony with our ELISA results, where marked IL-8 levels were detectable at 3 h and continuing to increase at 6 h before reaching a steady level. H. pylori-induced IL-8 secretion may be explained by both stimulation of the MAPK signaling system [59, 60], and NF-κB activation through several pathways [61, 62]. In the present study, MAPK signaling was ranked relatively high from 3 h onwards, based on IF calculations, and the cluster analysis showed that increasingly more genes in the MAPK pathway were affected after 6 h of H. pylori exposure. Regulators of NF-κB; TNFAIP3, RELB and BIRC3, which could also have explained the IL-8 expression, show increasing expression after 3 h (Additional file 1: Table S1), identical to the findings of Guillemin et al. . Therefore, it is interesting that the great increase in IL-8 mRNA, which peaked as early as 3 h as shown in both the microarray and rt-PCR data, occurs before the onset of both MAPK and NF-κB signaling. The rt-PCR data, but not the microarray analysis, also demonstrated a second increase in IL-8 mRNA at 24 h, although with noteworthy variance between experiments. While it is possible that this second surge may be explained by MAPK and/or NF-κB activation, it is unlikely that MAPK or NF-κB signaling explain the initial, powerful IL-8 mRNA peak seen at 3 h. The present study is the first to demonstrate that among more than 38 000 human genes, IL-8 was the single most up-regulated gene by gastric epithelial cells in response to H. pylori exposure in vitro, and it appears feasible that mechanisms other than MAPK or NF-κB activation may be responsible for this up-regulation.
Although histopathological studies indicate that MOI around 10:1 appear in H. pylori-colonized gastric mucosa, laboratory conditions can never replicate the complex physiology of the human stomach. Much higher MOI have normally been used to study in vitro gastric epithelial cell response to H. pylori colonization, and MOI of 300:1 was our incoulum of choice, as we wanted a sufficient inoculum to induce a biological response from AGS cells, both at the mRNA and protein levels, as indicated by other experiments [35, 63–71]. However, it is worth noting that in a recent report by Ritter et al., a marked IL-8 response from AGS cells exposed to cagA
H. pylori was seen at MOI ranging from 10:1 to 100:1 . The IL-8 response was higher at MOI 100:1 compared to 10:1 in all the bacterial strains tested. The response to MOI 300:1 was not assessed. Neither cagA nor vacA status seemed to affect the IL-8 response at the higher inoculum. Ritter's study also showed that different cellular pathways were activated in response to high or low MOI. In some other studies, where non-gastric cells were exposed to cagA
H. pylori, low MOI was associated with apoptosis inhibition and cell growth, whereas high MOI stimulated apoptosis and inhibited survival [35, 72, 73]. Hence, the choice of MOI may be crucial for the study outcome. Nevertheless, based on our immunofluorescence studies, where we found sufficient bacterial adhesion to AGS cells, typical morphological changes, and most importantly, a marked IL-8 mRNA and protein response to MOI 300:1, we concluded that under our experimental conditions, 300:1 was adequate to elicit a biological response without overloading the system.
You et al. performed a similar microarray study published in 2010 , where AGS cells were exposed to H. pylori for 6 h. A relatively stable number of 300-400 genes were reported to be differentially expressed at each of the sample points, whereas our data showed a progressive increase in the number of genes from 0.5 to 24 h. In addition, key biological processes like chemotaxis, TLR signaling and epithelial cell signaling were reported as down-regulated. This is in contrast to our results, and also the findings of most other similar microarray studies [19–23, 25, 26, 28, 29, 68, 75], where these particular processes are regularly increased. However, many of these studies do indeed show somewhat conflicting results, possibly explained by differences in incubation conditions, bacterial strains and obsolete or proprietary cDNA arrays and technology.
We have previously suggested a potential role for OMPLA in inflammation [14, 16]. OMPLA+ variants were found to yield increased hemolysis, adherence and release of urease and VacA compared to the OMPLA- variant. One of the aims of the present study was therefore to investigate the role of OMPLA on the gastric epithelial cell inflammatory response. We compared the gene expression profile of H. pylori OMPLA+ exposed cells against OMPLA- exposed cells at the 6 different time points. No significant difference was detected at any of the time points.
No other studies have directly investigated the role of OMPLA on the gastric epithelial cell inflammatory response, as the pldA/OMPLA status is unknown in most strains. Among the few full genome sequenced H. pylori strains, G27 carries a C7 repeat in the pldA gene  and B38 carries a C9 repeat, both giving rise to a truncated and inactive OMPLA . Several experiments have demonstrated the ability of G27 to induce a significant IL-8 response [29, 78], supporting our current observation that OMPLA-
H. pylori is indeed capable of inducing significant inflammation. One surprising result has been reported in a study of pH-regulated gene expression in the G27-strain , where Merrell et al. reported that cagA was consistently suppressed by low pH in H. pylori G27. Previous studies of other H. pylori strains, however, had suggested that cagA expression was induced at low pH.
Although the pldA phase variation did not appear to affect the inflammatory response in this study, phase variation of the pldA gene probably serves a purpose in other aspects of H. pylori. OMPLA activity is associated with increased survival at low pH [13, 80]. The mechanism behind this property is not yet known. One possibility might be that OMPLA has adapted an as yet unknown function needed for this specific environment, in addition to phospholipase activity. Dorrell et al. have showed that a pldA knockout mutant was unable to colonize mice . Salaün et al. have assessed changes in a spectrum of H. pylori phase-variable genes in a mouse model of gastric colonization . pldA was among the most rapidly changing genes, with changes occurring within the first 3 days of colonization. The change in pldA showed a phenotypic selection from an initial inoculum which consisted of a mixture of ON and OFF phenotypes, to an exclusively ON population.
Wernegreen et al. have postulated that evolutionary selection will interrupt a slippery tract, such as the C-tract in the pldA gene, thus removing the possibility of phase variation . When selection does not happen, the sequence feature must be to some benefit for the bacterium. It seems clear that the normal gastric environment is optimal for the OMPLA+ phenotype, but for what niche the OMPLA- phenotype is adaptive is currently unknown. One could speculate that the properties of the OMPLA- variant could be useful when transferring from one human stomach to another.