Genes of Helicobacter pylori Regulated by Attachment to AGS Cells

Abstract

Reciprocal interactions between Helicobacter pylori and cells of the gastric epithelium to which it adheres may affect colonization. Changes in gene expression of H. pylori induced by adhesion to AGS gastric cancer cells by coculture were compared to changes in gene expression of H. pylori cultured without AGS cells by using cDNA filter macroarrays. Adhesion was quantitatively verified by confocal microscopy of green fluorescent protein-expressing bacteria. Four experiments showed that 22 and 21 H. pylori genes were consistently up- and down-regulated, respectively. The up-regulated genes included pathogenicity island, motility, outer membrane protein, and translational genes. The σ²⁸ factor antagonist flgM, flgG, the stress response gene, flaA, omp11, and the superoxide dismutase gene (sodB) were down-regulated. The up-regulation of cag3, flgB, tonB, rho, and deaD was confirmed by quantitative PCR, and the up-regulation of lpxD, omp6, secG, fabH, HP1285, HP0222, and HP0836 was confirmed by reverse transcription (RT)-PCR. The down-regulation of flaA, sodB, and HP0874 was confirmed by quantitative PCR, and the down-regulation of omp11 was confirmed by RT-PCR. The alteration of gene expression in H. pylori after adhesion to gastric cells in vitro suggests that changes in motility, outer membrane composition, and stress responses, among other changes, may be involved in gastric colonization.

One-half of the world's population is infected by Helicobacter pylori (25). This gram-negative bacterium is responsible for gastritis and often causes duodenal ulcers, and it is the cause of the majority of gastric ulcers. Infection is also linked to gastric cancer and mucosa-associated lymphoid tissue lymphoma (7, 13, 33, 40). However, less than 20% of infected individuals present with clinical symptoms. The reasons for the variable clinical effects are not known but perhaps depend on the bacterial strain or the host cell response.

Several approaches have been used to detect differentially expressed genes, including subtractive hybridization (49), differential display (34, 50), serial analysis of gene expression (54), and, more recently, cDNA arrays (12, 15, 20). The advantage of the cDNA array technology is that it allows simultaneous examination of all the genes expressed by organisms whose sequences are complete, since the DNA chips contain all the open reading frames (ORFs) of these organisms.

In several studies the workers have used microarray technology to investigate H. pylori-induced changes in host gene expression, such as changes in AGS cell gene expression (1, 8, 21, 37, 38, 47), or the strain-specific gene diversity of H. pylori without host cells (4, 28, 29, 31, 44). Less is known about host cell-induced gene expression in H. pylori. This is likely due to the difficulty of obtaining sufficient RNA from the bacteria compared to the amount of RNA of the host cells. However, it has been shown by array analysis that inactivation of an H. pylori DNA methyltranferase (hpyIM) alters dnaK operon expression following AGS cell adherence (11).

A goal of genomic expression studies of pathogenic bacteria is to identify bacterial genes that are differentially regulated as a result of infection within the host. In this pool of genes are those genes that allow a microbe to adapt to host-specific microenvironments or that encode virulence determinants or have other functions. Ideally, studies of this kind should compare the expression profiles of bacteria under standardized in vivo conditions of infection. Unfortunately, this technically formidable goal has not been achieved by using nonamplification methods because the number of organisms within infected tissues is often small and RNA from host cells is so much more abundant than bacterial RNA. Recently, Graham et al. examined H. pylori gene expression in human tissue by using selective capture of transcribed sequences (19). To identify bacterial genes that are differentially regulated as a result of infection within the host, the gene expression of H. pylori affected by gastric cells should be compared with that of H. pylori in the absence of eukaryotic cells under the same conditions. Since it is not possible to faithfully replicate in vivo conditions, including the varying hydrogen ion, bicarbonate, and urea concentrations, in vitro methods have to provide initial information concerning specific host-regulated genes that can subsequently be identified in vivo by methods such as real-time PCR.

To overcome the difficulty of obtaining sufficient RNA from the bacteria instead of the host cells, green fluorescent protein (GFP)-expressing H. pylori in the log phase was used with AGS cells, and full adhesion to these cells was verified by confocal microscopy. Then ³³P-labeled cDNA derived from the coculture RNA was hybridized to H. pylori gene array membranes only after the quantity and quality of bacterial rRNA in the preparation had been confirmed to be comparable to the quantity and quality of eukaryotic rRNA by capillary electrophoresis.

Adhesion resulted in consistent 2.2- to 5.2-fold up-regulation of 22 H. pylori genes and simultaneous 2.2- to 5.7-fold down-regulation of 21 H. pylori genes. Up- or down-regulation of some of these genes was confirmed by quantitative PCR (qPCR) and reverse transcription (RT)-PCR analyses. Since AGS cells are a gastric cancer-derived cell line, the effects of a mucous layer and receptor phenotypes on H. pylori could be different from the effects of either in vivo gastric cells or primary cell lines. Nevertheless, in vitro data are a prerequisite for analysis of expression of these genes in the gastric environment.

MATERIALS AND METHODS

Bacterial strain and coculture conditions.

H. pylori strain 69a expressing GFP (a gift from Reiner Haas, Munich, Germany) was used to quantify the attachment of H. pylori to AGS cells (ATCC CRL 1739), a human gastric cancer cell line. This bacterial strain was constructed as described previously (26), and the GFP gene remained on the pHel2 plasmid; hence, the presence of GFP did not affect transcription. This strain is cag pathogenicity island (PAI) positive and vacA⁺, and it is able to infect mice and gerbils. Hence, this strain is of pathological interest.

Bacteria were grown for 16 h under microaerobic conditions (5% O₂, 10% CO₂, 85% N₂) at 37°C on Trypticase soy agar II plates supplemented with 5% sheep blood. The bacteria were harvested and resuspended in RPMI 1640 medium (BRL/Life Technologies Inc., Gaithersburg, Md.).

AGS cells were grown to confluence in tissue culture flasks. Each AGS monolayer was washed twice with RPMI 1640 medium. H. pylori was added at a multiplicity of infection of more than 1,000:1 and incubated at 37°C in a microaerobic atmosphere for a total of 4 h. An identical amount of H. pylori was added to a flask without AGS cells and incubated in the same way as the H. pylori-AGS cell coculture. This allowed analysis of gene regulation in the organism resulting from attachment to and culture with the AGS cells. A noninfected flask of AGS cells served as a negative control for RNA isolation to ensure that no contaminating signals derived from eukaryotic RNA were present. The RPMI 1640 medium did not include any additives, such as fetal bovine serum or antimicrobial agents.

After 1 h of coincubation, the H. pylori-AGS cell coculture was washed twice with RPMI 1640 medium to remove unattached H. pylori cells and debris, which eliminated unattached organisms from the analysis. The RPMI 1640 wash medium was prewarmed at 37°C to avoid stressing either H. pylori or the AGS cells. One hour was sufficient time for tight attachment of H. pylori to AGS cells, as confirmed by confocal microscopy. After washing, incubation was continued for an additional 3 h. At the end of the incubation period, the H. pylori-AGS cell coculture was washed two times with phosphate-buffered saline (pH 7.4, 37°C) to remove detached H. pylori cells. All of the control flasks were treated in the same way.

Confocal microscopy.

Confocal microscopy experiments were performed with a Zeiss LSM 510 microscope (Carl Zeiss Inc., Jena, Germany). The AGS cells were grown on 25-mm coverslips, and H. pylori strain 69a expressing GFP was added at the same density that was used for the coculture. The coverslips were treated in the same way that the cocultures were treated, with two washes after 1 h of incubation and two washes after an additional 3 h of incubation. The AGS cells were then stained with the fluorescent cationic dye dicarbocyanine [DiSC₃(5)] (red emission) to contrast them with the GFP-expressing microbes. The specific membrane attachment of H. pylori strain 69a expressing GFP was characterized by observing an H. pylori-AGS cell coculture in the z axis of the microscope field. The excitation wavelength for GFP was 488 nm, and the emission wavelength was 530 nm; for DiSC₃(5) the excitation and emission wavelengths were 543 and 610 nm, respectively. This allowed excellent visualization of the quantity of adherent organisms.

RNA extraction and DNase digestion.

For RNA extraction, AGS cells alone and H. pylori and AGS cells in coculture were treated with the TRIzol reagent (Invitrogen/Life Technologies Inc.) and incubated at room temperature for 5 min to permit complete dissociation of nucleoprotein complexes. After the lysates were transferred to 1.7-ml Eppendorf tubes and chloroform (one-fifth the volume of TRIzol reagent) was added, the tubes were mixed by inversion for 15 s and incubated at room temperature for 2 to 3 min. After centrifugation for 15 min at 11,500 × g and 4°C, the colorless upper aqueous phase containing RNA was collected. Further purification was performed with an RNeasy mini kit (Qiagen Inc., Valencia, Calif.) by following the manufacturer's instructions. The culture containing only H. pylori was centrifuged and also washed twice with phosphate-buffered saline, and TRIzol was directly applied to the pellet; the preparation was then treated just like the H. pylori and AGS cells in coculture were treated.

To remove all the genomic DNA, RNA was treated with DNase I (Invitrogen/Life Technologies Inc.) and RNasin (Promega, Madison, Wis.) for 20 min at 37°C. The RNA was precipitated with ethanol following phenol extraction.

Identification of the source of the RNA.

Since the amount of prokaryotic RNA in the coculture was a relatively small fraction of the total RNA, it was important to ensure that similar quantities of RNA from H. pylori were used to generate the cDNAs that were hybridized. The RNA extract was evaluated with an RNA 6000 Nano assay by using an Agilent 2100 bioanalyzer (Agilent Technologies, Mountain View, Calif.). The ratio of the quantity of H. pylori RNA obtained from H. pylori alone to the quantity of RNA obtained from an H. pylori-AGS cell coculture was calculated by using the relative peak ratio for prokaryotic 23S rRNA in the two samples, and the quantities were adjusted so that they were equal in the cDNA generation reaction mixture. The quantity of AGS cell RNA was adjusted so that it was equal to the quantity in the H. pylori-AGS cell coculture by using the 28S rRNA peak to check whether any contaminating signal on the H. pylori filter array could have come from AGS cells.

Preparation of ³³P-labeled cDNA, gene array hybridization, and analysis of the arrays.

After denaturation of RNA for 5 min at 70°C and chilling on ice for 3 to 4 min, a 6-mer random primer (a generous gift from Sigma-Genosys Inc., The Woodlands, Tex.), [³³P]dCTP, dATP, dGTP, dTTP, and avian myeloblastosis virus reverse transcriptase were added to initiate the cDNA synthesis reaction. Labeled cDNA was purified by using Sephadex G-25 spin columns, and the level of ³³P-labeled cDNA was determined by scintillation counting.

The ³³P-labeled cDNA preparations were hybridized at 65°C overnight to Panorama H. pylori gene array membranes (Sigma-Genosys Inc.), which contained 1,681 PCR-amplified ORFs representing all putative H. pylori strain 26695 protein-encoding genes (1,590 genes) and 91 protein-encoding genes of H. pylori strain J99 that are unique to this strain. Strain 69a harbors a few genes that are unique to this strain, and these genes were not expressed on this array membrane. After washing, the membrane was exposed on a phosphorimager screen (Molecular Dynamics, Sunnyvale, Calif.). To determine the intensity of each spot on the array, image files were analyzed by using the ImageQuant package (Molecular Dynamics). Relative signals from different arrays were normalized by comparing the signal from a gene to the total signal from all spots on the array. The cutoff for significant up- or down-regulation of genes was defined as a twofold difference. The consistently up- and down-regulated genes found in four separate experiments were identified, and the statistical difference between the H. pylori-AGS cell coculture and H. pylori alone was analyzed by the Mann-Whitney U test. A P value of <0.05 was considered statistically significant. The nature of some of the additional genes with unknown functions was determined by homology by using the Blast search program.

Real-time qPCR.

Unique primers were designed for 100- to 300-bp regions of selected up- and down-regulated genes seen in the array. Primer design was aided by the Primer 3 software available at http://www-genome.wi.mit.edu/genomesoftware/other/primer3.html (43). A total of 7 of 22 up-regulated genes (cag3, flgB, tonB, rho, deaD, pflA, and thrS) and 7 of 21 down-regulated genes (flaA, sodB, HP0874, HP1286, thioredoxin gene, ferrodoxin gene, and HP0920) were selected on the basis of two criteria: (i) the function of the gene was known, and (ii) the level of RNA expression of the gene in H. pylori alone was not so low that it increased the validity of the qPCR. The RNA expression level for each gene was calculated by comparing the signal from the gene to the total signal from all spots on the array. The primers used in real-time reactions are listed in Table 1. Standard PCR performed with genomic DNA from H. pylori strain 69a expressing GFP as the template was used to check that all primer pairs resulted in amplification of a single product.

TABLE 1.

qPCR primers

Gene	Forward primer sequence (5′ - 3′)	Reverse primer sequence (5′ - 3′)
ureA (HP0073)a	AACCGGATGATGTGATGGAT	GGTCTGTCGCCAACATTTTT
cag3 (HP0522)	TTTGGGGATTGGTTTGGTAA	TTAGCGTTGCGTTTGTTGAG
flgB (HP1559)	GGCCAAAGGATTTGGATTTT	CTTGCCCATTTCACTCGTTT
tonB (HP1341)	TAAAATGGCGCAGATTAGGG	GATGCACGGTTTCTTCTGGT
rho (HP0550)	GTTACGCAAGGCGGATACAT	CGCCATTCCTTAAAGCAAAA
deaD (HP0247)	CTACACCAGGAAGGCTGCTC	TTTTAATCGGCTCTGGCATC
flaA (HP0601)	GAACTCTAACCGAACAGGCGTTA	CCCAAATGAATCCCATTTAAGGT
sodB (HP0389)	GGATTGCCTAAGCCCCAAAG	CCAATTCCAACCAGAGCCAA
HP0874	AGAGTTTGCCCTGGTTTGCA	CTAGGAGCGTTTTTCACCACG

^aThe designations in parentheses are the locus names in H. pylori strain 26695.

The template used for qPCR analysis was the same cDNA that was used for hybridization to the arrays for H. pylori alone and for the H. pylori-AGS cell cocultures. Genomic DNA from H. pylori strain 69a expressing GFP that was serially diluted 10-fold was used to obtain the standard curve. Each reaction mixture, which contained the SYBR Green label (Qiagen Inc.) and the primers, was added to a 96-well plate along with 1 μl of cDNA or genomic DNA template; the final reaction volume was 50 μl/well. The negative control contained the reaction mixture but no DNA. The reactions were performed with an Icycler real-time PCR machine (Bio-Rad, Hercules, Calif.) for 40 cycles (92°C for 30 s, 55°C for 30 s, 72°C for 40 s). Data were collected during the extension step and were expressed in arbitrary fluorescence units per cycle. A melting curve was used at the end to confirm that there was only one peak and only one product.

RT-PCR analysis.

Primers were designed for 15 of the 22 up-regulated genes (lpxD, omp6, secG, fabH, HP1285, HP0222, HP0836, HP0582, infC, HP0906, HP0118, omp19, HP0757, HP1430, and HP0973) and for 14 down-regulated genes (omp11, flgM, flgG, HP0359, HP1548, HP1326, HP0241, HP0113, HP0218, HP1510, HP0427, HP0682, HP1408, and JHP0332) as an alternative to qPCR. The cDNA templates for RT-PCR were synthesized by RT reactions from H. pylori alone and the H. pylori-AGS cell coculture by using avian myeloblastosis virus transcriptase (Sigma-Genosys Inc.). The cDNAs were used as the templates, and the PCRs were performed with the primer pairs shown in Table 2.

TABLE 2.

RT-PCR primers

Gene	Forward primer sequence (5′ - 3′)	Reverse primer sequence (5′ - 3′)	Expected size (bp)
ureA (HP0073)	ATGAAACTCACCCCAAAAGAG	CTCCTTAATTGTTTTTACATA	714
lpxD (HP0196)	ATGAAATTAAGCGAATTGTTA	TTAAGATTTAAAAAACCCTTT	1,011
omp6 (HP0229)	ATGAAAAAAACGATTTTACTT	TCAATAGGCCAAAACATAATT	1,452
secG (HP1255)	GTGTTTATGACAAGCGCTCTG	TTATTTTTCATCCTTTTTTTG	606
fabH (HP0202)	ATGGAATTTTACGCCTCTCTT	CTAACTTCCTCCAAAATACAC	996
HP1285	ATGAGTGTTTTAAATGCCAAA	TTATTTTTTATTCTGCCATGC	693
HP0222	ATGGAAAAGACAGAAAACACA	TTACTCTATTTTTCTTAAAGC	222
HP0836	ATGCCCATGCGTTTGCACACT	CTAGAGCCAGGGCTTGAGAGG	360
omp11 (HP0472)	ATGATTAAAAGAATTGCTTGT	TTAGAAAGTAAAGACATAATC	561

The PCR program was as follows: after incubation for 10 min at 95°C, 30 cycles of 92°C for 1 min, 52°C for 1 min, and 72°C for 2 min. PCR products were analyzed by electrophoresis on 1% agarose-ethidium bromide gels. No product was obtained in the RT-PCR with the RNA of AGS cells alone, confirming that there was no cross-reactivity that interfered with interpretation of the RT-PCR data.

RESULTS

Attachment of H. pylori to the AGS cells.

The attachment of H. pylori to AGS cells was characterized by observing the H. pylori-AGS cell coculture in the z axis by confocal microscopy. The organisms were attached to the upper and apical surfaces of the AGS cells. Only AGS cells stained with DiSC₃(5) were seen in the first section on the bottom of a culture dish with very few H. pylori cells on the surface of the culture coverslip (Fig. 1A). Halfway through the cell layer, H. pylori cells were clearly visible as rings on the periphery of AGS cells (Fig. 1B), and attached H. pylori cells were also clearly seen on the apical surfaces of the cells (Fig. 1C). Few if any free organisms were seen. Furthermore, all the GFP-containing organisms remained associated with the AGS cells after the two wash steps, and thus only the gene profile of these organisms was compared to that of the control organisms.

FIG. 1.

Attachment of GFP-expressing H. pylori to AGS cells. Binding of H. pylori to AGS cells was assessed by confocal microscopy (original magnification, ×400). (A) Optical section through the H. pylori-AGS cell coculture close to the coverslip, showing predominantly AGS cells (red fluorescence). (B) Optical section through the middle of the coculture, showing H. pylori cells bound to the periphery of AGS cells (green fluorescence for bacteria and red fluorescence for AGS cells). (C) Optical section through the top of the H. pylori-AGS cell coculture, showing H. pylori adhering to the upper surface of AGS cells. A z-axis section showing an optical section of each image is provided below each panel, and these images show that there was mainly apical and some lateral adhesion of H. pylori to the AGS cells and there was no basal or glass surface adhesion.

RNA analysis.

The characteristic prokaryotic 16S and 23S rRNAs and eukaryotic 18S and 28S rRNAs were visualized after capillary electrophoresis by using an Agilent 2100 bioanalyzer (Fig. 2). In Fig. 2A, the left lane shows the prokaryotic 16S and 23S rRNA bands. The center lane shows a mixture of these bands and the eukaryotic 18S and 28S rRNA bands of the AGS cells in the RNA extracted from the coculture. The right lane shows the eukaryotic 18S and 28S rRNA bands of the AGS cells alone. Figure 2B shows the results of Agilent quantitation of the bands. The amounts of the 16S and 23S rRNAs contributing to the rRNAs in the coculture were almost equivalent to the amounts of the eukaryotic rRNAs and to the amounts of the rRNAs derived from H. pylori alone, which allowed a meaningful comparative analysis of the hybridization of the rRNAs derived from the coculture and the hybridization of the rRNAs derived from H. pylori in the absence of AGS cells.

FIG. 2.

RNA patterns for AGS cells and H. pylori. Prokaryotic and eukaryotic rRNA patterns were analyzed by the RNA 6000 Nano assay. (A) In the RNA from the H. pylori-AGS coculture (Hp/AGS co-culture) (middle lane), characteristic prokaryotic 16S and 23S rRNA (left lane) and eukaryotic 18S and 28S rRNA (right lane) were well visualized. (B) Quantitative analysis of bands with the Agilent apparatus allowed calculation of the relative ratios of the prokaryotic and eukaryotic rRNAs. These ratios were similar, which allowed appropriate array analysis.

To show that the cDNA that hybridized to the array was derived from H. pylori RNA and not from AGS cells, an equal amount of RNA from uninfected AGS cells was used to probe the filters. There was no signal from the membrane hybridized with ³³P-labeled cDNA from AGS cells alone (data not shown), indicating that the hybridization observed for the cDNA of the coculture on the Panorama H. pylori gene array membrane was derived only from the H. pylori RNAs. Indeed, of the 1,681 genes on the array, 1,542 (91.7%) showed reproducible positive hybridization signals under both conditions.

Gene expression of H. pylori adhering to AGS cells.

When the phosphorimages of the H. pylori-AGS cell coculture were compared with those of H. pylori alone (n = 4), no statistically significant differences in the levels of expression of the H. pylori ureA, ureB, and Hsp60 genes were detected. The mean changes in ureA, ureB, and Hsp60 expression were −0.6-, −0.7-, and 0.6-fold, respectively (Fig. 3). These genes can be considered housekeeping genes for this organism under these conditions since their levels of expression were not affected. This relative constancy also validates the comparison between the organisms under these two conditions.

FIG. 3.

Quantitation of up- and down-regulation: relative up-regulation (A) and down-regulation (B) of H. pylori genes induced by AGS cells as determined with the ³³P-labeled phosphorimages. The up-regulation of 22 genes ranged from 2.2- to 5.2-fold, and the down-regulation of 21 genes ranged from 2.2- to 5.7-fold. The changes in expression of these 22 up-regulated genes and 21 down-regulated genes were statistically significant (P < 0.05) compared to the changes in the expression of ureA, ureB, and Hsp60. The data are the averages of four experiments, and the error bars indicate the standard errors of the means.

Furthermore, the expression of only 22 of the 1,542 genes analyzed was increased more than twofold, and the expression of only 21 genes was decreased more than twofold in each of the four experiments. Hence, the vast majority of the genes analyzed did not change over the time of culture or coculture. The average changes in the 22 up-regulated genes are shown in Fig. 3A, and the average changes in the 21 down-regulated genes are shown in Fig. 3B. All the changes in these 22 up-regulated genes and 21 down-regulated genes were statistically significant (P < 0.05). The levels of two genes, HP0906 and rho, increased almost fivefold, and HP1326 and omp11 were down-regulated by almost the same factor. The other genes showed less regulation.

When the 22 up-regulated genes were classified by function, as shown in Table 3, several different types of genes were found. A PAI gene, cag3 (HP0522), and genes encoding two chemotaxis and motility gene products, paralyzed flagellar protein (pflA [HP1274]) and flagellar basal body rod protein (flgB [HP1559]), were up-regulated. A cell envelope gene encoding UDP-3-0-(3-hydroxymyristoyl)glucosamine N-acyltransferase(lpxD [HP0196]), two outer membrane protein genes (omp6 [HP0229] and omp19 [HP0896]), a protein secretion gene (secG [HP1255]), and two siderophore-mediated iron transport genes (tonB [HP1341] and HP0582) were also up-regulated, as were genes encoding the transcription termination factor Rho (rho [HP0550]), the ATP-dependent RNA helicase DEAD-box family (deaD [HP0247]), translation initiation factor IF-3 (infC [HP0124]), and threonyl-tRNA synthetase (thrS [HP0123]).

TABLE 3.

Classification of up-regulated genes

Functional class	Gene	Description
Pathogenesis	HP0522	cag PAI cag3
Chemotaxis and motility	HP1274	Paralyzed flagellar protein (pflA)
	HP1559	Flagellar basal-body rod protein (flgB)
Cell envelope	HP0196	UDP-3-0-(3-hydroxymyristoyl)glucosamine N-acyltransferase (lpxD)
	HP0229	Outer membrane protein (omp6)
	HP0896	Outer membrane protein (omp19)
Protein secretion	HP1255	Protein translocation protein, low temperature (secG)
Transport and binding protein	HP1341	Siderophore-mediated iron transport protein (tonB)
	HP0582	Predicted siderophore-mediated iron transport protein
Transcription	HP0550	Transcription termination factor Rho (rho)
Translation	HP0247	ATP-dependent RNA helicase DEAD-box family (deaD)
	HP0124	Translation initiation factor IF-3 (infC)
	HP0123	Threonyl-tRNA synthetase (thrS)
Fatty acid metabolism	HP0202	Beta-ketoacyl-acyl carrier protein synthetase III (fabH)
Central intermediary metabolism	HP1285	Predicted acid phosphatase lipoprotein
Purines, pyrimidines	HP0757	Beta-alanine synthetase homolog
Conserved	HP1430	Predicted ATP/GTP binding protein
Unknown	HP0906	HP predicted coding region HP0906
	HP0222	HP predicted coding region HP0222
	HP0836	HP predicted coding region HP0836
	HP0118	HP predicted coding region HP0118
	HP0973	HP predicted coding region HP0973

Some metabolic genes, such as the genes encoding beta-ketoacyl-acyl carrier protein synthetase III (fabH [HP0202]; involved in fatty acid metabolism), a predicted acid phosphatase lipoprotein (HP1285; involved in central intermediary metabolism), and a beta-alanine synthetase homolog (HP0757), were also up-regulated. A conserved gene, a gene encoding a predicted ATP/GTP binding protein (HP1430), and five genes with unknown functions (HP0906, HP0222, HP0836, HP0118, and HP0973) were also up-regulated. A Blast search did not find any homologues of the five latter genes, suggesting that these genes are specific for H. pylori or other microaerophiles.

When the 21 down-regulated genes were classified, as shown in Table 4, the mRNA levels of three chemotaxis and motility genes, namely, the gene encoding a temporal regulator of flagellar apparatus biogenesis (flgM homolog [HP1122]), a flagellar basal-body rod gene (flgG [HP1585]), and a flagellin A gene (flaA [HP0601]), decreased. HP1122 is identical to the flgM gene, a σ²⁸ factor antagonist that regulates flagellar gene expression (32), in H. pylori strains N6, BO242, SS1, NCTC11637, CC7C, CC48A, BO255, RE8029, and CC29C. A cell envelope gene encoding an outer membrane protein (omp11 [HP0472]) was also down-regulated, as were energy metabolism genes, such as the genes encoding thioredoxin (HP1458) and ferrodoxin (HP0277), the detoxification gene, the superoxide dismutase gene (sodB [HP0389]), the gene encoding a predicted dihydroneopterin aldolase (HP1510), and three conserved genes, including two genes encoding predicted integral membrane proteins (HP0920 and HP1548) and a gene encoding a predicted secreted protein (HP1286).

TABLE 4.

Classification of down regulated genes

Functional class	Gene	Description
Chemotaxis and motility	HP1122	Temporal regulator of flagellar apparatus biogenesis (flgM)
	HP1585	Flagellar basal-body rod protein (flgG)
	HP0601	Flagellin A (flaA)
Cell envelope	HP0472	Outer membrane protein (omp11)
Translation	HP0359	HP predicted coding region HP0359 (identical to part of Campylobacter lari tRNAAla and tRNAIle genes)
Energy metabolism	HP1458	Thioredoxin
	HP0277	Ferrodoxin
Detoxification	HP0389	Superoxide dismutase (sodB)
Biosynthesis of cofactors; prosthetic groups, and carriers	HP1510	Predicted dihydroneopterin aldolase
Conserved	HP0920	Predicted integral membrane protein
	HP1548	Predicted integral membrane protein
	HP1286	Predicted secreted protein
Unknown	HP1326	HP predicted coding region HP1326
	HP0241	HP predicted coding region HP0241
	HP0427	HP predicted coding region HP0427
	HP0682	HP predicted coding region HP0682
	HP1408	HP predicted coding region HP1408
	HP0874	HP predicted coding region HP0874
	JHP0332	Putative coding region
	HP0113	HP predicted coding region HP0113
	HP0218	HP predicted coding region HP0218

When a Blast search was performed with HP1326, HP0241, HP0427, HP0682, HP1408, HP0874, JHP0332, HP0113, HP0218, and HP0359, only HP0359 (66 bp) was identical to part (66 bp) of the Campylobacter lari tRNA^Ala and tRNA^Ile genes (632 bp). The nine remaining genes had no homologues in the databases examined.

qPCR.

All qPCR primers for seven selected genes (cag3, flgB, tonB, rho, deaD, pflA, and thrS) confirmed that there was only one product when standard PCR was performed with strain 69a expressing GFP. However, two of the genes, pflA and thrS, showed cross-reactivity with AGS cell genomic DNA in PCR that could have interfered with interpretation of the real-time data, and qPCR could be performed for only five genes (cag3, flgB, tonB, rho, and deaD) with cDNA from H. pylori alone or from the H. pylori-AGS cell coculture. The amounts of H. pylori cDNA from H. pylori alone and from the H. pylori-AGS cell coculture were equalized by using the reference housekeeping gene ureA. The level of ureA (Fig. 4A) did not change in qPCR, whereas cag3 (Fig. 4B), flgB (Fig. 4C), tonB (Fig. 4D), rho (Fig. 4E), and deaD (Fig. 4F) were up-regulated in the coculture, in agreement with the array data. rho and deaD showed particularly prominent up-regulation in qPCR.

FIG. 4.

Real-time PCR (qPCR) of up-regulated genes. Up-regulation of cag3, flgB, tonB, rho, and deaD was confirmed by real-time PCR. When the cDNA encoding ureA, cag3, flgB, tonB, rho, and deaD derived from either H. pylori alone (Hp) or the H. pylori-AGS cell coculture (Hp/AGS co-culture) was amplified by real-time PCR, the expression of ureA (A) did not change as determined by the real-time analysis, whereas cag3 (B), flgB (C), tonB (D), rho (E), and deaD (F) were up-regulated in the coculture with AGS cells.

All of the qPCR primers for the seven down-regulated genes (flaA, sodB, HP0874, HP1286, thioredoxin gene, ferrodoxin gene, and HP0920) confirmed that there was only one PCR product when standard PCR was performed with strain 69a expressing GFP. However, three genes, the thioredoxin and ferrodoxin genes and HP0920, showed cross-reactivity with AGS cell genomic DNA in PCR that could have interfered with interpretation of the real-time data, and the real-time PCR results could be interpreted for only the remaining four genes (flaA, sodB, HP0874, and HP1286). flaA (Fig. 5B), sodB (Fig. 5C), and HP0874 (Fig. 5D) were down-regulated, as determined by this technique, which is in agreement with the array data. However, HP1286 did not show down-regulation. The level of expression of this gene in H. pylori alone was low, and this might explain the absence of measurable down-regulation as determined by qPCR.

FIG. 5.

Real-time PCR (qPCR) of down-regulated genes. qPCR revealed down-regulation of flaA (B), sodB (C), and HP0874 (D) compared to the constant level of ureA (A). Hp, H. pylori control; Hp/AGS co-culture, H. pylori-AGS cell coculture.

RT-PCR analysis.

In addition to qPCR, RT-PCR was performed, particularly when low levels of expression were found on the array. The amounts of H. pylori cDNA from H. pylori alone and from the H. pylori-AGS cell coculture were equalized with the reference housekeeping gene ureA (Fig. 6A). Of the 15 up-regulated genes which were not analyzed by qPCR, 7 showed up-regulation in the H. pylori-AGS cell coculture. lpxD (Fig. 6B), omp6 (Fig. 6C), secG (Fig. 6D), fabH (Fig. 6E), HP1285 (Fig. 6F), HP0222 (Fig. 6G), and HP0836 (Fig. 6H) showed more enhanced bands at the expected size with the RNA from the H. pylori-AGS cell coculture than with the RNA from H. pylori alone. Again, the data are in agreement with the conclusions drawn from the array hybridization analyses. The other four genes, HP0582, infC, HP0906, and HP0118, showed the bands at the expected size when the RNA from the H. pylori-AGS cell coculture was used but not when the RNA from H. pylori alone was used (data not shown). Four genes, omp19, HP0757, HP1430, and HP0973, did not show PCR products when either the RNA from H. pylori alone or the RNA from the H. pylori-AGS cell coculture was used. This was probably due to the very low levels of RNA expression for each of these genes.

FIG. 6.

RT-PCR of up- and down-regulated genes. Panel A shows that the level of ureA did not change under coculture conditions. Panels B through H show that there were increases in the levels of expression of lpxD, omp6, secG, fabH, HP1285, HP0222, and HP0836, respectively. In contrast, panel I shows that there was a decrease in the level of expression of omp11. The RT-PCRs amplified a single product of the expected size for all the genes. Hp, H. pylori; Hp/AGS, H. pylori-AGS cell coculture.

Similarly, 1 of the 14 down-regulated genes which were not analyzed by qPCR, omp11 (HP0472), was down-regulated in the H. pylori-AGS cell coculture (Fig. 6I). Another eight genes, flgM, flgG, HP0359, HP1548, HP1326, HP0241, HP0113, and HP0218, did not show definite down-regulation in the H. pylori-AGS cell coculture. The remaining five genes, HP1510, HP0427, HP0682, HP1408, and JHP0332, did not show PCR products from either RNA of H. pylori alone or RNA of the H. pylori-AGS cell coculture, demonstrating the difficulty of confirmation of down-regulation by qPCR or RT-PCR when RNA expression levels are very low.

Hence, whether qPCR or RT-PCR was used to confirm the microarray data, there was general agreement. The exceptions appeared to result from levels of gene expression too low to detect or quantify by these methods and indicate that repetitive microarray analysis might be a more sensitive method for detecting differences between levels of gene expression in H. pylori under different conditions, as observed previously for acid stress conditions (56).

DISCUSSION

H. pylori is a unique gastric denizen that has adapted in many ways to the unique niche of the mammalian stomach. For example, expression of UreI, an acid-activated transporter, has allowed utilization of a neutral pH optimum intrabacterial urease as a means of combating severe gastric acidity and may be the most important adaptation by this neutralophile for gastric habitation (55).

In functional genomics expression profiling of mRNA is used to provide a condition-specific and time-specific genome-scale snapshot of the transcriptome (16, 24). Since cDNA microarrays for the all the ORFs of two strains of H. pylori (26695 and J99) are available, global gene analysis is possible with an array in which all the known genes are present. Whether fabricated as a slide microarray (9, 14, 45, 56), a high-density oligonucleotide array (35, 36), or nylon membrane macroarrays, this analytical tool provides a way to analyze genes of H. pylori whose expression is different under different conditions. The example analyzed here is the effect of adhesion to AGS cells, whereas previously acid-adaptive genes were used (56). Presumably, both sets of genes must be regulated to optimize colonization efficiency.

In most previous studies (1, 2, 6, 8, 10, 27, 38, 42, 47) the workers characterized the host cell RNA response to attached bacteria since preparation of sufficient RNA from host cells is less challenging. In each of these studies the workers used transformed cell lines, relatively large multiplicities of infection, and a short time course in order to induce and study early events. As H. pylori persists for life in a host stomach, which can result in several gastrointestinal diseases, it is difficult to extrapolate the results of H. pylori gene expression after 4 h of coculture in vitro to the gene expression on an in vivo time scale. However, 4 h is shorter than 24 or 48 h (15, 59), and the results might reflect the early events of gene expression in H. pylori. Furthermore, our experiments were designed to detect genes altered by direct contact with the organisms; hence, a membrane filter separating the bacteria and AGS cells was not appropriate.

The problem that there has been no efficient method available for extraction of bacterial RNA without a decrease in the level of RNA from host cells or degeneration of the host cell RNA was solved here in several ways, including the use of GFP-expressing organisms and confocal microscopy to select only adherent microbes, the use of quantitative comparative RNA analysis to ensure that there was a significant amount of microbial RNA, and the use of either qPCR or RT-PCR to confirm that the microarray data remained consistent for at least four individual experiments. It was also found that short-term culture of H. pylori was necessary to ensure log-phase growth conditions. There was also no contamination of signal from the AGS cells on the bacterial microarray.

Entirely consistent data for 43 genes were obtained in four different experiments, indicating that the results presented here reflect reproducible changes in gene expression induced by adhesion of the GFP-expressing organisms to a gastric cell line. Furthermore, qPCR (Fig. 4 and 5) and RT-PCR (Fig. 6) confirmed many of the microarray results, validating the array approach.

The H. pylori cag PAI, a 40-kb locus that has been linked to H. pylori virulence, contains 31 genes. It induces epithelial cells to secrete interleukin 8 by activating NF-κB (48), remodeling the host cell surface with pedestal formation (46), phosphorylating secreted CagA in the host cell (51), activating transcription factor AP-1, and expressing the protooncogenes c-fos and c-jun by activation of the extracellular signal-regulated kinase-mitogen-activated protein kinase cascade, resulting in ELK-1 phosphorylation and increased c-fos transcription (39). However, little is known about which specific gene in the cag PAI induces these changes. Adhesion of H. pylori to AGS cells up-regulated only cag3 of the 31 genes of the cag PAI. The protein encoded by cag3 has recently been identified as an antigen in sera from infected patients (22), suggesting that cag3 might encode a protein that contributes to the cag PAI-induced host cell changes.

Bacterial flagella consist of three distinct parts connected in series, a basal body in the membrane, a short curved rod (the hook), and a long helical filament. The annotated genomic sequence of HP26695 contains at least 40 genes, scattered throughout H. pylori genome, which are likely to be involved in flagellar function and/or assembly (52). Some of these flagellar genes of H. pylori were regulated by host cell adhesion, and they also were regulated by acid exposure (56). Two flagellar genes, the genes encoding paralyzed flagellar protein (pflA) and flagellar basal-body rod protein (flgB), were up-regulated (Table 3), and three genes, flgM and the genes encoding flagellar basal-body rod protein (flgG) and flagellin A (flaA), were down-regulated in the organisms adhering to AGS cells (Table 4).

Analysis of the disrupted ORFs in nonmotile mutants of H. pylori showed that there was disruption of ORF HP1274 (encoding paralyzed flagellar protein; pflA) (3). As determined by electron microscopy, the nonmotile mutant had an aberration in the region where the flagella protrude from the bacteria (3). Apparently, the paralyzed flagellar protein is important for H. pylori motility. The induction of this paralyzed flagellar protein by AGS cells may be related to adherence of H. pylori to host cells and the subsequent loss of motility. The finding that not only pflA but also four other flagellar genes were up- or down-regulated by adhesion to AGS cells suggests that H. pylori reacts to host cell attachment by modifying the flagellar apparatus, which may result in adaptation of H. pylori to the host cell microenvironment by reducing the motility of the organism.

Some cell-envelope-protein-encoding genes also responded to cell adhesion differently. lpxD, omp6, and omp19 were up-regulated, and omp11 was down-regulated. This down-regulation of outer membrane protein genes has been found in other bacteria. Expression of omp1, encoding the major outer membrane protein of Chlamydia trachomatis, was also strongly attenuated in persistently infecting synovial Chlamydia (18) and in infected monocytes in vitro (17). These changes in outer membrane proteins may be related to the effects of the proximity of the bacterial outer membrane to the host cell membrane during adhesion, which modifies the composition of the cell envelope as an adaptation to cell adhesion.

The changes in gene expression also extended to inner membrane protein genes, such as the genes encoding the protein translocation protein, the low-temperature secretory complex protein (secG), the siderophore-mediated iron transport protein (tonB), and a predicted siderophore-mediated iron transport protein (HP0582), which were up-regulated (Table 3). In contrast, two conserved integral membrane proteins (HP0920 and HP1548) were down-regulated (Table 4).

Translocation of newly synthesized Escherichia coli proteins from the cytosol across the cytoplasmic membrane is facilitated by the concerted action of SecA ATPase and the SecYEG integral membrane complex (5). It has been reported that SecG of E. coli is defective at 20°C but normal at 37°C (41). However, protein translocation, which was related to the inward H⁺ electrochemical gradient, the proton motive force, was strongly dependent on SecG even at 37°C (23). Hence, induction of secG in H. pylori by AGS cell adhesion (3.0-fold) (Fig. 3A) may reflect an increase in membrane translocation of proteins in response to attachment of H. pylori to host cells.

Four translational genes, the genes encoding transcription termination factor Rho (rho), the ATP-dependent RNA helicase DEAD-box family (deaD), translation initiation factor IF-3 (infC), and threonyl-tRNA synthetase (thrS), were up-regulated (Table 3). Hence, translation of H. pylori genes might be enhanced by adhesion, as it is in other prokaryotes that interact with eukaryotic cells (53).

Transcription termination in response to metabolic changes in bacteria is an important way of regulating gene expression. For instance, a rho mutant strain of Caulobacter crescentus was viable in normal conditions, but its viability was greatly impaired when the cells were exposed to stress, especially oxidative stress (30), and the synthesis of several peptides was altered in this mutant strain, suggesting that the Rho factor is essential for an efficient response to stress in Caulobacter. The finding that rho was prominently up-regulated (4.7-fold) in the H. pylori-AGS coculture system (Fig. 3A) suggests that adhesion to AGS cells is interpreted as a stress condition by H. pylori.

In conclusion, hybridization of bacterial cDNAs prepared from high-quality bacterial RNA derived from bacterium-AGS cell cocultures to filter arrays showed that adhesion to AGS cells up-regulated 22 H. pylori genes and down-regulated 21 genes, and some of the data were confirmed by real-time PCR or RT-PCR. The multiple functions of the regulated genes indicate that adhesion of the organism to the gastric cell surface initiates pleiotropic adaptive responses. Functions can be inferred for many of the genes that were found to change due to adhesion. The change in gene expression when the bacterium is attached to a gastric cell line may reflect the responses found in vivo in addition to the responses to the composition of the gastric juice at the site of infection. A search for regulation of these genes during gastric infection could confirm the significance of these in vitro data, but without the in vitro data presented here, a search for genes that change during infection would be obscured by the changes in expression due to the presence of gastric juice. It may be that a correlation can be established between regulation of the genes in vivo and pathogenesis.