ABSTRACT
The best-studied Helicobacter pylori virulence factor associated with development of peptic ulcer disease or gastric cancer (GC) rather than asymptomatic nonatrophic gastritis (NAG) is the cag pathogenicity island (cagPAI), which encodes a type IV secretion system (T4SS) that injects the CagA oncoprotein into host epithelial cells. Here we used real-time reverse transcription-PCR (RT-PCR) to measure the in vivo expression of genes on the cagPAI and of other virulence genes in patients with NAG, duodenal ulcer (DU), or GC. In vivo expression of H. pylori virulence genes was greater overall in gastric biopsy specimens of patients with GC than in those of patients with NAG or DU. However, since in vitro expression of cagA was not greater in H. pylori strains from patients with GC than in those from patients with NAG or DU, increased expression in GC in vivo is likely a result of environmental conditions in the gastric mucosa, though it may in turn cause more severe pathology. Increased expression of virulence genes in GC may represent a stress response to elevated pH or other environmental conditions in the stomach of patients with GC, which may be less hospitable to H. pylori colonization than the acidic environment in patients with NAG or DU.
INTRODUCTION
Helicobacter pylori infects nearly half of the world's population and causes a chronic nonatrophic gastritis (NAG) in essentially all those who are infected. After decades of inflammation, the infection may lead to peptic ulcer disease in approximately 10 to 15% of cases and to the development of gastric cancer (GC) in 1 to 3% of cases (28, 40). Whether the outcome of H. pylori infection is simply nonatrophic gastritis, which is usually asymptomatic, rather than peptic ulcer or gastric cancer is determined by host, environmental, and bacterial factors, the best studied of which is the cag pathogenicity island (cagPAI). The cagPAI is an approximately 40-kb locus composed of about 27 to 31 genes, many of which are responsible for the synthesis of a type IV secretion system (T4SS) that injects the CagA oncoprotein into host epithelial cells (5, 6, 22, 38, 39, 47, 54). Translocated CagA alters the cell-cell junctions (2), motility, and cytoskeleton arrangement (47, 48) and induces a proinflammatory and antiapoptotic gene expression profile via activation of the NF-κB transcription factor (10, 51). In addition to the cagPAI, other genes are relevant for colonization, persistence, and damage to the gastric mucosa, as they encode adhesins (BabA and SabA) (24, 29, 35), proteins that protect H. pylori from the oxidative burst (KatA, NapA, and arginase) (15, 25, 43), and the pore-forming cytotoxin VacA, which induces in vitro epithelial cell vacuolation (17, 21), inhibition of T cell activation and proliferation (23, 55), and apoptosis (16).
The extent of gastric mucosal damage, and hence disease outcome, may depend not only on the gene content of the particular H. pylori strain but also on the level of in vivo expression of the genes capable of inducing chronic inflammation and gastric mucosal damage. In an effort to better understand H. pylori adaptation to the gastric niche, transcription of individual genes or even the complete genome has been evaluated for different in vitro conditions, such as pH (11, 19, 36, 50, 56, 62), iron concentration (63), or growth phase (32). However, data obtained under these experimental conditions do not reflect the conditions that are present in vivo in the human gastric mucosa, where H. pylori encounters other complex physicochemical factors, such as motility, viscosity of gastric mucin, and the host inflammatory response, to name just a few. Therefore, we and others have studied in vivo expression of H. pylori virulence genes, both in animal models and in humans (8, 9, 12, 27, 45, 46). A few studies have also analyzed the relationship between in vivo expression of individual H. pylori virulence genes and disease. Examples include the association of increased transcription of iceA1 with more severe gastric inflammation (40, 41), higher expression of cagA with intestinal metaplasia and gastric adenocarcinoma (49), and upregulation of urease genes with gastric cancer (64). Nevertheless, despite these attempts, we know very little about in vivo gene expression in the gastric mucosa of infected patients, and even less about how this compares in the different H. pylori-associated diseases. We hypothesized that there are differences in the in vivo expression of H. pylori virulence-associated genes in patients with different clinical manifestations, which might result from the response to physical or chemical differences in the gastric environment or perhaps even be related causally to the development of disease. We therefore sought to measure the in vivo expression of the cagPAI and other H. pylori virulence genes in the gastric mucosa of patients with GC compared to those with NAG and duodenal ulcer (DU).
MATERIALS AND METHODS
Patient selection.Adult patients were recruited from those undergoing endoscopy because of gastroduodenal disease or possible gastric cancer in hospitals of the Instituto Mexicano del Seguro Social (IMSS), Mexico City, Mexico. We screened 274 consecutive patients for possible inclusion in the study and selected cases that fulfilled the following criteria: absence of treatment with antimicrobials or proton pump inhibitors during the previous 14 days, positive H. pylori culture with a cagPAI+ strain, and biopsy specimens that yielded sufficient RNA for quantitation of in vivo gene expression. We selected consecutive cases from patients with NAG (n = 10; mean age, 50.4 years; 2 females and 8 males), DU (n = 10; mean age, 59.5 years; 7 females and 3 males), and GC (n = 11; mean age, 60.2 years; 8 females and 3 males). Each participant provided informed consent, and the study was approved by the ethical committee of the National Council for Research at IMSS.
Gastric biopsy specimens.Patients underwent endoscopy with collection of four gastric biopsy specimens from the antrum or corpus, one of which was used for H. pylori culture, one for histologic examination, and the other two for extraction of total RNA. In GC cases, biopsy specimens were obtained from the tumor as well, but they were used only for histopathology, not for analysis of gene expression. Gastric biopsy specimens for histology were fixed with formalin, embedded in paraffin, and stained with hematoxylin-eosin. Biopsy specimens used for RNA extraction were placed in TRIzol (Invitrogen, Carlsbad, CA), frozen immediately in liquid nitrogen, and transported to the laboratory, where they were stored at −70°C until use.
Diagnostic criteria.Diagnosis was based on endoscopy findings and on histopathology by a single experienced pathologist. NAG and GC were documented for each of the biopsy specimens by using accepted histologic criteria (18, 34, 37); diagnosis of DU was based on endoscopy findings.
H. pylori culture and DNA extraction.Gastric biopsy specimens were homogenized, inoculated onto blood agar plates, and incubated at 37°C for 3 to 10 days under a 9% CO2 atmosphere as previously described (26). H. pylori colonies were identified by microscopic morphology and by positive tests for catalase, urease, and oxidase. From the primary growth, single colonies were propagated in blood agar for an additional 48 h, harvested in phosphate-buffered saline (PBS), and then used for extraction of genomic DNA by use of a Wizard genomic DNA purification kit (Promega, Madison, WI).
RNA extraction.To study in vivo gene expression, frozen biopsy specimens were thawed and used for extraction of total RNA by the TRIzol protocol (4). Isolated RNA samples were treated with DNase I (Roche Diagnostics, Mannheim, Germany), purified with an RNeasy minikit (Qiagen, Austin, TX), and suspended in PCR-grade water. To evaluate in vitro gene expression, single-colony H. pylori isolates were grown for 48 h on blood agar plates (pH 7.0) at 37°C in 9% CO2 as previously described (26), suspended in 1 ml TRIzol, and used for RNA extraction following the same protocol as that for biopsy specimens. All RNA samples were quantified using a Nanodrop ND-1000 spectrophotometer and then maintained at −70°C until use.
Primer design.We designed primers for genes on the cagPAI and for other selected virulence genes, including napA, sabA, katA, vacA, babA, and rocF, using Oligo 6.0 software and the published sequences of H. pylori 26695 (61) and J99 (1) (Table 1). All primers had a calculated melting temperature (Tm) of 68 to 70°C, amplified products of 100 to 300 bp, and ended with a double-dA 3′ terminus. The amplification efficiency for each set of primers was first tested with DNA from each H. pylori patient isolate. If primer efficiency was <85% for a given gene in any isolate, a new primer was designed and tested.
Primer sequences for real-time RT-PCRa
qRT-PCR.To determine the in vivo expression of virulence genes, we performed quantitative reverse transcriptase PCR (qRT-PCR) assays using a Roche LightCycler 2.0 thermal cycler. We previously determined that DNA equivalent to 105 bacterial cells (0.186 ng) yielded a cycle threshold (CT) of 18.5 (9). This value was arbitrarily defined as 97% efficiency, so a primer pair with 97% efficiency should give a standard yield of 1.9418.5 copies after 18.5 PCR cycles. Cycle efficiency as a function of the observed CT was defined as follows: cycle efficiency = (1/2)(1.9418.5)1/CT observed.
We estimated relative gene expression normalized to the concentration of 16S rRNA, which we previously showed may be used as a surrogate of the bacterial load in vivo (9). Briefly, qRT-PCR was performed in a 20-μl reaction mixture containing 2 U of thermostable recombinant Tth DNA polymerase (Applied Biosystems), 100 ng of RNA from a gastric biopsy specimen, 0.25 μM (each) corresponding primers (Table 1), 0.3 mmol/liter (each) dATP, dCTP, and dGTP, 0.05 mmol/liter dTTP, 0.5 mmol/liter dUTP, 0.4× SYBR green (Roche), 2.4 mmol/liter Mn(OAC)2, 0.4 U of uracil DNA glycosylase (Roche), 5% dimethyl sulfoxide (DMSO; Sigma), and 1× Tricine buffer. Each sample was run in duplicate under the following PCR conditions: 3 min at 50°C, 30 min at 60°C, and then 40 cycles of 2-step amplification with 20 s at 95°C and 1 min at 59°C. The CT values for 16S rRNA and for each virulence gene were corrected for primer efficiency and used to estimate relative expression as follows: relative expression = 2(CT 16S rRNA − CT target gene).
As quality controls, data were filtered to include only amplifications for which duplicate assays differed by ≤2 cycles, showed a single Tm value, and yielded a single band of the correct size on agarose gel electrophoresis. Since transcript levels in some cases were very low (high CT), we also required that the CT for the sample be at least 2 cycles lower than that for the no-template control, which ensured that low expression levels could be distinguished from zero.
Statistical analysis.Differences in relative expression across genes and across patient groups were analyzed with the following linear model, which allows testing of more than two treatments simultaneously, increasing the power of the tests (42): yijk = μ + αi +δj + αδik + εijk, where yijk is the expression level, αi is the group effect (NAG, DU, and GC) for an i value of 1, 2, or 3, δj is the gene effect for a j value of 1, 2,…32, αδij is the interaction between the group and gene, and εijk is the experimental error for patient isolate differences with a k value of 1, 2,…ri (ri = 11, 10, and 10, respectively).
The model considered variance heterogeneity among genes. Because there was no interaction effect between genes and groups due to the low power of the test, results are shown as means and standard deviations (SD) for all genes and for every group (NAG, DU, and GC). In addition, log ratio expression among groups was calculated and plotted. The Pearson correlation was used to explore coexpression among genes.
RESULTS AND DISCUSSION
H. pylori virulence gene expression in the gastric mucosa of patients with NAG, DU, and GC ranges over 2 orders of magnitude.Quantitative RT-PCR was used to determine gene expression in the gastric mucosa. Expression was calculated based on the cycle threshold (corrected for primer efficiency) for the gene of interest relative to that for the 16S rRNA, which we have previously shown is strongly correlated (R = 0.80) with bacterial CFU per gram of tissue and therefore serves to control for differences in bacterial number in each sample (8, 9). Gene transcript levels were typically 1 to 3 log lower than the 16S rRNA level, which was expected since stable RNA is most abundant in the cell. Based on a previous estimate of ∼250 copies of 16S rRNA per H. pylori cell (9), our results correspond to transcript abundances that range from about 0.25 to 25 mRNA copies/cell, which is within reasonable agreement with estimates for Escherichia coli and Saccharomyces cerevisiae, allowing for the much smaller cell volume of the H. pylori cell (31, 65).
cag26 (cagA) is the most highly expressed gene on the H. pylori cagPAI.Relative in vivo expression was determined for each gene on the cagPAI except for cag2 (HP0521), a locus that is highly polymorphic (7) and for which efficient primers could not be obtained. Since gene expression was generally lower than that for 16S rRNA, relative expression was typically represented as a value between 0 and 1. Expression was highest for cag26, similar to the results found in a nonhuman primate model (8). The cag26 gene encodes CagA, which is translocated by the T4SS into host cells, where it exerts a number of biological activities, including alteration of the cytoskeleton and cell junctions, activation of transcription factors leading to synthesis of inflammatory mediators, alteration of the cell cycle and polarity, and increased proliferation (2, 10, 47, 51, 57, 58). CagA was recently reported as the first bacterial oncoprotein, based on these results and on development of tumors in transgenic mice expressing cag26 (39).
Since many of the genes on the cagPAI are implicated in formation of the T4SS, we hypothesized that their expression may be correlated with that of cagA. As expected, expression of cagA was positively correlated with expression of other cagPAI genes (Table 2). These included cag7, cag10, and cag12, which constitute the core of the T4SS genes; cag5, virB11, and cag23 (homologs of virD4, virB11, and virB3, respectively), which encode ATPases that provide the energy necessary for the secretory machine (14, 44, 53); and cag18 and cag25, which may encode components of the pilus structure (3, 33). Expression of cag25 was particularly high and similar to that observed for cagA, which is perhaps not surprising since cag25 is an orthologue of virB2, encoding the pilus subunit of the T4SS in Agrobacterium tumefaciens. High-level expression would be expected if cag25 also encodes the repeating pilus subunit of the T4SS in H. pylori, though experimental evidence for this is still lacking (59). Expression of cagA also correlated highly with that of some genes on the PAI that are not implicated in formation of the T4SS and whose function is unknown (cag1, cag6, and cag17). In contrast, among the 6 other virulence genes outside the cagPAI, only katA showed a significant correlation (R = 0.98; P < 0.0001) with expression of cagA.
Analysis of correlation between expression of cag26 (cagA) and expression of other cagPAI genes
In vivo expression of H. pylori virulence genes is higher in patients with GC than in those with NAG or DU.The relative expression of genes on the cagPAI (Fig. 1A) and of six other putative virulence genes (Fig. 1B) was measured in patients with NAG, DU, and GC. For 17 of the 24 cagPAI genes (71%) and 5 of the 6 other virulence genes (83%), expression was higher in patients with GC than in those with DU or NAG. For 15 of 24 cagPAI genes (63%), >2-fold increases were observed for patients with GC compared to those with NAG or DU (Table 3). Among these genes, 9 are thought to be involved in the function and assembly of the H. pylori T4SS: cag5, cag6, virB11, cag8, cag12, cag22, cag23, cag25, and cag26.
Relative in vivo expression of H. pylori genes on the cagPAI (A) and of other putative virulence genes (B). Bars represent mean (±SD) relative expression in vivo for each gene in patients with NAG (n = 10), DU (n = 10), and GC (n = 11). Expression of cag2 was not measured because it is highly polymorphic and efficient primers could not be determined. NAG, nonatrophic gastritis; DU, duodenal ulcer; GC, gastric cancer.
Mean relative in vivo expression of cagPAI and other H. pylori virulence genes in patients with GC compared to those with NAG and DU
To better visualize the differences among patient groups, we constructed a scattergram, which illustrates that expression of most cagPAI and other virulence genes was grouped in the area where the ratio was higher for GC patients than for NAG and DU patients (Fig. 2). Mean in vivo expression of all cagPAI and other virulence genes in patients with GC (mean = 0.051) was significantly (P < 0.01) higher than that in patients with NAG (mean = 0.019) and DU (mean = 0.019), which had identical values. These results were illustrated in a correlation analysis (Table 4) which demonstrated that overall gene expression in patients with NAG and DU was much more highly correlated (R = 0.95) than that of either group and patients with GC (R = 0.69 and 0.57, respectively).
Scattergram of the ratios of mean relative gene expression in patients with NAG/DU compared to those with NAG/GC. Cardinal numbers represent the cagPAI gene designations. Non-cagPAI genes are designated as follows: v11 = virB11, R = rocF, S = sabA, V = vacA, B = babA, K = katA, and N = napA. Most genes showed higher expression in patients with GC, so they have ratios of <1.0 on the x axis and are located above the diagonal line. No data are shown for cag19 or cag24 because expression was not detected in patients with GC. NAG, nonatrophic gastritis; DU, duodenal ulcer; GC, gastric cancer.
Correlation coefficients for relative in vivo expression of H. pylori virulence genes among patient groups
Expression of H. pylori virulence genes in vitro compared with in vivo.Higher expression of virulence genes in GC patients might be an intrinsic property of these strains, or it might be a result of different environmental conditions in the stomach of patients with GC. To distinguish between these two possibilities, we evaluated cagA transcripts from 48 H. pylori isolates grown in vitro (9 from GC, 9 from NAG, and 6 from DU) and compared the results to in vivo expression. In general, expression of cagA was higher in vitro than in vivo. Although in vitro expression of cagA correlated significantly with in vivo expression for patients with NAG, this was not true for patients with GC or DU (Fig. 3). Mean (±SD) relative expression of cagA in vitro from patients with GC, NAG, and DU was 0.74 (±1.01), 0.74 (±1.23), and 0.51 (±0.67), respectively, and there were no statistically significant differences among patient groups. These results suggest that increased expression in GC patients is more likely a bacterial response to environmental conditions, rather than causally related to development of gastric cancer. For example, cagA expression may be higher in GC because the atrophic gastritis that precedes gastric cancer results in increased gastric pH (52), which has been shown by some investigators to induce cagA expression in vitro (36). It is unclear whether this increased cagA expression in GC patients could in turn contribute to pathology or is simply a marker of elevated gastric pH or some other change in gastric physiology in patients with GC.
Relative expression of cagA in vivo in patients with NAG, DU, and GC versus that from the same strains determined in vitro. cagA expression was generally greater in vitro than in vivo. NAG, nonatrophic gastritis; DU, duodenal ulcer; GC, gastric cancer.
Unexpectedly, we were unable to detect in vivo expression of cag19 and cag24 in any of the 11 patients with GC, as the corresponding products were not observed in agarose gels and analysis of the Tm indicated a lack of a specific product even when primers efficiently amplified the correct products from DNAs from H. pylori isolates cultured from the corresponding patients. To determine if the absence of in vivo expression of cag19 and cag24 was an intrinsic property of the strains or a response to in vivo conditions of GC, we determined their in vitro expression in five H. pylori isolates from patients with GC. In vitro transcripts were readily detected for both cag19 and cag24 for all five strains (Fig. 4), typically at a level 1 to 2 log below the 16S rRNA level, similar to the levels of most other cagPAI genes in vivo. These results suggest that, like the increased expression of most cagPAI genes in GC, the absence of expression of cag19 and cag24 in GC is a result of environmental conditions in the stomach of patients with GC. Since the functions of the proteins encoded by cag19 (CagI) and cag24 (CagD) are unknown and the data are conflicting regarding their role in the induction of interleukin-8 (IL-8) and translocation of CagA (13, 30), the biological significance of this observation is unclear.
Relative expression in vitro of cag19 and cag24 in five H. pylori strains recovered from patients with GC. Although no expression was detected for cag19 or cag24 in vivo (Fig. 1A), strains from these patients expressed these genes in vitro.
Perspectives.In vivo expression of H. pylori cagPAI and other virulence genes was greater overall in patients with GC than in those with NAG or DU. However, since in vitro expression of the cagA-encoded oncoprotein was not greater in H. pylori strains from patients with GC, increased expression in GC in vivo is likely a result of environmental conditions in the gastric mucosa, though it may in turn contribute to more severe pathology. Recent analysis of the primary transcriptome of H. pylori (50) demonstrated that under acid stress, several cagPAI genes are upregulated in vitro. However, it is difficult to extrapolate these findings to the in vivo conditions, since H. pylori can increase or decrease acid production depending on whether infection is predominant in the antrum or the corpus (20). Furthermore, it is likely that multiple factors affect gene expression in vivo, so it should not be surprising that in vivo expression cannot readily be predicted from the results of in vitro manipulations. The elevated pH and perhaps other environmental conditions in the stomach of patients with GC may be less hospitable to H. pylori, and the increased expression of virulence genes may represent a sort of stress response. This interpretation is consistent with the observation of a major switch in gene expression during the late-log-to-stationary-phase transition, with induction of virulence gene expression, including that of cagA and other genes on the cagPAI (8, 60). Failure to express cag19 and cag24 in vivo should be studied in patients with precancerous lesions (atrophic gastritis, metaplasia, or dysplasia) to determine if this might serve as a biomarker of risk for development of gastric cancer.
ACKNOWLEDGMENTS
This work was supported in part by a Fogarty International Research Collaboration Award (FIRCA) to J.V.S. (5R03TW007460) from the National Institutes of Health (NIAID). Javier Torres is a recipient of an exclusivity scholarship from Fundacion IMSS, Mexico. This material is based upon work supported by a grant from the University of California Institute (UC MEXUS) and the Consejo Nacional de Ciencia y Tecnología de México (CONACyT) for Mexico and the United States and by a grant for Mexico from the Instituto Mexicano del Seguro Social.
FOOTNOTES
- Received 24 August 2011.
- Returned for modification 25 September 2011.
- Accepted 11 November 2011.
- Accepted manuscript posted online 28 November 2011.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.
