INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Helicobacter pylori induces gastric inflammation in virtually all colonized individuals, and such gastritis increases the risk for peptic ulcer disease and distal gastric adenocarcinoma (8, 14, 15, 21, 23, 27). However, only a minority of persons carrying H. pylori develop clinical sequelae, suggesting that particular bacterial products may contribute to pathogenesis (4). The first strain-specific gene identified in H. pylori was cagA, a component of the cag pathogenicity island (1, 7, 9, 30); persons colonized with H. pylori cagA-positive strains are at increased risk for developing peptic ulceration and distal gastric cancer compared to persons harboring cagA-negative strains (5, 10-12, 22, 24, 29). The gene vacA represents a second H. pylori locus of heterogeneity, and strains that possess a vacA s1 signal sequence allele are also associated with peptic ulcer disease (3, 32). However, the majority of persons colonized with cagA⁺ vacA s1 strains remain asymptomatic, suggesting that additional H. pylori genes may also be important in disease pathogenesis. Recently, iceA (a designation derived from the phrase "induced by contact with epithelium") was identified in H. pylori following transcriptional up-regulation on contact with gastric epithelial cells (26). iceA exists as two distinct genotypes, iceA1 and iceA2, and only iceA1 RNA is induced following adherence in vitro (26). The deduced H. pylori iceA1 product (Fig. 1A) demonstrates strong homology to a restriction endonuclease, NlaIII, in Neisseria lactamica (19); however, mutations and deletions found in the majority of iceA1 sequences preclude translation of a full-length homolog. In vivo, carriage of H. pylori iceA1 strains has been found in some (20, 26, 32) but not all (18, 36) studies to be associated with peptic ulceration and enhanced acute neutrophilic infiltration. However, substantial heterogeneity among gastric inflammatory scores exists within iceA1-colonized populations (20, 26).

View larger version (24K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the genetic organization of iceA1 (A) and iceA2 (B) and the flanking genes, cysE and hpyIM. (A) The hatched regions represent continuous ORFs homologous to nlaIIIR. The top schematic represents cysE, prototype iceA1, and hpyIM from H. pylori strain 60190, and the bottom schematic is a variant from strain CH4 containing an ORF homologous to the complete N. lactamica nlaIIIR ORF. Two potential ATG start sites are shown, as is the major transcriptional start site P1 (large arrows) recently identified in strain 60190 (13). The presence of minor transcripts originating in the intergenic region between the cysE and iceA1 ORFs is indicated by the small arrows with single asterisks. Read-through transcription from cysE into iceA1 is indicated by the small arrows with two asterisks. The size of the intergenic region between the end of the cysE ORF and the first iceA1 ATG codon is 84 nucleotides for strain 60190 and 25 nucleotides for strain CH4. The positions of PCR primers are indicated by arrows below the 60190 schematic. (B) The top schematic represents cysE, prototype iceA2, and hpyIM from H. pylori strain J178, encoding a protein of 59 aa. The diagrams below represent subsequently identified iceA2 variants (16). Each of the five iceA2 peptide motifs, of 14, 13, 16, 6, and 10 aa, respectively, is represented by a box. The existence of two distinct 16-aa domains is indicated by different patterns. The positions of PCR primers are indicated by arrows. The total number of amino acids in each iceA2 ORF is shown for each variant.

In contrast to iceA1, iceA2 has no homology to known proteins, and its structure reveals patterns of repeated peptide cassettes (Fig. 1B). In its most common form, iceA2 can encode a protein of 59 amino acids (aa) with two conserved outer domains of 14 and 10 aa, respectively, that flank three internal peptide domains of 13, 16, and 6 aa, respectively (16). Sequence analysis of several H. pylori iceA2 strains shows that the internal 35-aa cassette (composed of the 13-, 16-, and 6-aa domains) may be absent or repeated up to three times, resulting in deduced proteins of 24, 59, 94, or 129 aa (Fig. 1B) (16). Although substantial differences exist between the iceA1 and iceA2 sequences, both genes are transcribed in vivo (26), leading to the hypothesis that levels of iceA transcription within the host environment may contribute to disease development. Therefore, the aims of this study were to quantitate specific iceA1 and iceA2 transcription levels in gastric mucosa and to determine whether such levels were related to bacterial numbers in the same biopsy specimen, iceA sequence heterogeneity, gastric inflammation, peptic ulcer disease, and the presence of other H. pylori genotypes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Clinical specimens. Patients at the Nashville Department of Veterans Affairs Medical Center Gastroenterology Clinic were prospectively enrolled in this study after giving written informed consent. The study was approved by the Vanderbilt University and Nashville Department of Veterans Affairs Institutional Review Boards. Exclusion criteria included the following: use of nonsteroidal medications; history of gastric surgery; malignancy; infection with human immunodeficiency virus; positivity for hepatitis B surface antigen; active gastrointestinal bleeding; use of steroids or other immunosuppressive drugs, antibiotics, or bismuth compounds before endoscopy; and detection of H. pylori by histology but not culture. Patients with peptic ulcer disease were included in the study, and ulceration was defined as a mucosal break at least 1.0 cm in diameter, with depth and covered by exudate, noted at the study endoscopy. Five gastric biopsy specimens were taken from the greater curvature of the antrum, between 2 and 5 cm from the pylorus, and distributed as follows: culture (n = 2), histological examination (n = 2), and reverse transcriptase (RT) PCR (n = 1). Biopsy specimens for RNA isolation were immediately frozen at 70°C. To isolate H. pylori, gastric biopsy specimens were immediately placed in normal saline at 4°C and homogenized in 250 µl of normal saline with a tissue grinder (Micro Kontes, Vineland, N.J.). Fifty microliters was plated onto Trypticase soy agar with 5% sheep blood (BBL) and incubated for 96 h under microaerobic conditions, as described previously (25). The iceA, cagA, and vacA genotypes were determined by PCR of H. pylori genomic DNA, as previously reported (26).

Quantitation of H. pylori gene expression in vivo. Individual gastric biopsy specimens were placed in sterile vials (Nunc, Roskilde, Denmark) and stored at 70°C until they were used for RNA preparation for RT-PCR. Total biopsy specimen RNA was isolated (25) and treated with DNase (1 U/µg), and 1 µg of the RNA was reverse transcribed to cDNA using random hexamers as primers (Pharmacia LKB Biotech, Piscataway, N.J.) as previously described (25). H. pylori iceA1 (bHP1 and uHP1 [Table 1]), iceA2 (bHP2 and uHP2 [Table 1]), and 16S rRNA (bHP16S and uHP16S [Table 1]) primers were designed based on known H. pylori genomic sequences (26). iceA1 genes demonstrate considerable strain variation, with insertions, deletions, and potential differences in transcription start sites (16). These sequence variants could affect the ability to detect iceA1 transcripts in gastric mucosa if they compromise hybridization of the oligonucleotides used either for amplification or detection of the amplified products. To assure that transcripts from all potential sequence variants would be detected, the iceA1 amplification primers (bHP1 and uHP1 [Table 1]) and the fluorescein-conjugated detection oligonucleotide (fHP1 [Table 1]) were designed to hybridize to conserved regions downstream of the most 3' transcription start site in the iceA1 gene (P₁ [Fig. 1A]) (16). iceA2 genes also show extensive strain variation in sequence. In contrast to iceA1, iceA2 sequences contain variable numbers (0, 1, 2, or 3) of a 35-aa cassette that possesses one of two distinct 16-aa domains (Fig. 1B). Since all iceA2 sequences determined to date contain a conserved 14-aa domain (Fig. 1B), the iceA2 fluorescein-conjugated detection oligonucleotide (fHP2 [Fig. 1B and Table 1]) was designed to hybridize within this motif region (Fig. 1B and Table 1).

Histology. Sections of biopsy specimens were examined without knowledge of the experimental PCR results by one experienced histopathologist (K.T.T.). Histologic parameters were scored from 0 to 3 as outlined in the Sydney System and included assessment of the degree of activity (neutrophil infiltration), lymphocyte and plasma cell infiltration, epithelial degeneration, mucus depletion, epithelial erosion, glandular atrophy, intestinal metaplasia, and colonization density, as previously described (26). Modified Giemsa stain was used for identification of H. pylori (17).

Sequence analysis. Sequence analysis was performed on genomic DNA from 38 of the 41 H. pylori strains isolated (7 iceA1 and 31 iceA2) using an ALF-Express automated sequencer and DNAMAN PC Gene Software (Lynn on Biosoft).

Statistics. Mean 16S rRNA transcript levels, and inflammation and injury scores were compared between groups using the Mann-Whitney U rank sum test. Differences in H. pylori strain-specific genotypes (cagA and vacA) and the prevalence of peptic ulcer disease were compared among groups using the Mantel-Haenszel test. A P value of <0.05 was defined as significant.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Quantitation of iceA1 and iceA2 transcripts in vitro. Substantial heterogeneity exists among inflammatory scores within iceA1-colonized populations (26), raising the possibility that levels of iceA transcription within the host environment may contribute to such differences or to disease development. Therefore, we developed a quantitative RT-PCR assay to measure RNA transcripts for iceA1, iceA2, and H. pylori 16S rRNA in gastric biopsy specimens. We first determined the sensitivity of the quantitative PCR assay for detection of H. pylori iceA sequences by amplifying serial dilutions of plasmid DNA containing the iceA1 or iceA2 inserts from H. pylori strains 60190 (cag⁺ vacA s1a/m1 [where s1a is the signal sequence allele and m1 is the midregion allele] iceA1) and J178 (cag⁺ vacA s1a/m1 iceA2) (26), respectively, with a lower limit of detection by PCR for both iceA1 and iceA2 of 10²² mol of plasmid DNA (Fig. 2A). Preliminary studies of eight gastric biopsy specimens demonstrated that all reverse-transcribed samples with detectable iceA cDNA gave amplification within this linear range. Samples that had not been reverse transcribed showed no amplification above negative controls, indicating the absence of detection of contaminating DNA. There is extensive sequence variation in iceA1 and iceA2 genes. Therefore, we next established that the primers used to amplify and detect the prototype sequences (Fig. 2A) would detect all sequence variants in the 41 isolated strains and that these primers were specific. Accordingly, quantitative PCR was performed using 10 ng of genomic DNA from each of the 41 isolates (Fig. 2B). The mean relative light units for each iceA1 or iceA2 DNA sample tested were >1,000-fold higher than baseline in the respective iceA1 or iceA2 PCRs (Fig. 2B). These results demonstrated that the reagents could detect transcripts of all iceA1 and iceA2 sequence variants if they were expressed in these biopsy specimens. In addition, the primers used demonstrated absolute specificity for the appropriate genotypes, as no amplification above background was seen with the reciprocal gene.

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Sensitivity (A) and specificity (B) of quantitative iceA1 and iceA2 PCR on DNA. (A) Dilutions of plasmid containing either 60190 iceA1 or J178 iceA2 DNA were used as templates for PCR with iceA1- or iceA2-specific oligonucleotide primers (Table 1). The results shown (mean ± SD) are for dilutions corresponding to plasmid DNA from 1019 to 1023 mol from a representative standard curve. A line representing mean relative light unit values for negative controls with no template DNA which were run concomitantly is shown for reference. The limit of detection for iceA1 and iceA2 by this assay was 1022 mol of plasmid DNA. (B) Genomic DNA (10 ng) from each of the 41 H. pylori strains was used as a template for quantitative PCR with biotinylated iceA1- and iceA2-specific primers (Table 1). Streptavidin-bound PCR products were hybridized with an allele-specific internal fluorescein-conjugated probe (Table 1), and chemiluminescence was quantitated using a luminometer. Mean relative light units + SD for iceA PCRs are shown for iceA1 and iceA2 strains and for negative controls (solid columns) run concomitantly and consisting of reactions with no template DNA.

Detection of iceA transcripts in gastric biopsy specimens. Of the 41 patients studied, all were positive for H. pylori by both serology and culture. Of these, 10 were colonized with iceA1 strains and 31 harbored iceA2 strains (Table 2). Fifteen (37%) of the 41 patients had duodenal ulcer disease; no gastric ulcers were identified (Table 2). Quantitative RT-PCR to measure iceA RNA expression in gastric tissue revealed that 4 (40%) of 10 iceA1- and 19 (61%) of 31 iceA2-colonized persons had detectable iceA RNA. The patients were stratified into the following four groups: iceA1 RNA positive, iceA2 RNA positive, iceA1 RNA negative, and iceA2 RNA negative. To confirm that detection of iceA1 and iceA2 expression in vivo reflected the genotype of the colonizing strain, the results of PCR using bacterial genomic DNA as a template were compared to RT-PCR results on biopsy specimen cDNA. In all 23 iceA RNA-positive tissue samples, the biopsy specimen iceA status corresponded to the iceA genotype of the strain isolated (P < 0.0001), demonstrating the high specificity of these assays.

Quantitation of constitutively expressed H. pylori 16S rRNA transcripts in gastric tissue. H. pylori iceA expression in vivo may be dependent upon local host conditions, such as pH, osmolarity, or nutrient concentrations, or, as found in vitro, bacterial growth phase (13). Alternatively, the amount of iceA mRNA may simply reflect bacterial colonization density. Therefore, to determine whether bacterial density correlated with levels of mucosal iceA RNA in the same tissue specimen, we also quantitated levels of the constitutively expressed 16S rRNA. All 41 gastric biopsy samples contained detectable H. pylori 16S rRNA, although considerable variation was observed (Fig. 3). H. pylori 16S rRNA transcript levels were similar among persons colonized by iceA1 strains and persons colonized with iceA2 strains (P = 0.34). When patients were stratified on the basis of iceA expression in vivo, mucosal levels of 16S rRNA were not significantly different of iceA1 RNA-positive and iceA1 RNA-negative patients (P = 0.1 [Fig. 3]). H. pylori 16S rRNA transcript levels among iceA2 RNA-positive persons were higher (P = 0.008) than those of iceA2 RNA-negative persons (Fig. 3). However, among the iceA2 RNA-positive group, levels of iceA2 RNA did not correlate with the 16S rRNA levels from the same patients (P = 0.6). In addition, there were no significant associations found between the number of organisms quantified histologically in adjacent antral biopsies and iceA1 or iceA2 expression levels (P = 0.25 and 0.41, respectively) (data not shown). Since the differences in 16S rRNA expression are quite small, these results suggest that the presence of iceA transcripts does not simply reflect bacterial load.

View larger version (12K): [in this window] [in a new window]   FIG. 3.   Relationship of H. pylori 16S rRNA levels and detectable expression of iceA1 or iceA2 RNA in gastric tissue from 41 H. pylori+ persons. Mean values () ± SD are shown adjacent to the data points (). Among persons carrying iceA2 strains, 16S cDNA levels were significantly higher in iceA2 RNA-positive than in RNA-negative persons (5.4 × 1019 versus 3.7 × 1019 mol.; P = 0.008); however, 16S cDNA levels were not significantly different among persons carrying iceA1 strains stratified on the basis of iceA1 expression in vivo (P = 0.1). P values were determined by the Mann-Whitney U test.

Relationship of iceA expression in vivo to histologic parameters of inflammation and injury. Substantial variability among inflammatory scores exists within iceA1- and iceA2-colonized patients (26). This heterogeneity prompted us to investigate whether levels of iceA mRNA expression in vivo may be associated with the degree of histopathological change. Acute inflammatory scores were significantly higher in iceA1 RNA-positive patients (2.3 ± 0.3 [mean ± standard deviation {SD}]) than in iceA1 RNA-negative (0.6 ± 0.3), iceA2 RNA-positive (0.7 ± 0.1), or iceA2 RNA-negative (1.2 ± 0.3) subjects (P = 0.01, 0.002, and 0.03, respectively) (Fig. 4). However, the presence of detectable iceA transcripts was not related to other parameters of histopathology. Since iceA1 has been associated with the presence of cagA and vacA s1 alleles, as well as peptic ulcer disease (26, 32), we also examined the relationship between iceA expression and other genetic and clinical markers of H. pylori virulence. Mucosal expression of iceA1 and iceA2 was not related to cagA or vacA genotypes or duodenal ulcer disease (data not shown), suggesting that expression of iceA1 is associated with neutrophilic infiltration independent of the presence of other H. pylori virulence-related genes.

View larger version (14K): [in this window] [in a new window]   FIG. 4.   Relationship between acute-inflammatory scores and expression of iceA1 or iceA2 RNA detected in gastric mucosa from 41 H. pylori+ persons. Acute inflammation was scored from 0 to 3 by a single pathologist blinded to mucosal levels of iceA RNA. Mean values () ± SD are shown adjacent to the data points (). Each data point reflects averaged scores from two antral biopsy specimens (one data point per patient). Biopsies from four (40%) of 10 iceA1- and 19 (61%) of 31 iceA2-colonized persons had detectable mucosal iceA RNA. Acute inflammatory scores were significantly higher in iceA1 RNA-positive patients compared to all other groups. Statistical significance was determined by the Mann-Whitney U test.

iceA2 gene structures correlate with iceA2 expression in vivo. We next sought to determine whether sequence heterogeneity within iceA1 and iceA2 might contribute to variation in mucosal iceA expression. Among seven iceA1 sequences, only one contained a continuous open reading frame (ORF) homologous to the NlaIII full-length ORF and which might be predicted to encode a functional restriction endonuclease (16). Gastric tissue from the patient colonized with this strain did not show iceA1 RNA expression. Among the four iceA1 RNA-positive samples, there were no variations in the corresponding iceA1 DNA sequences that were consistently associated with differences in mucosal iceA1 RNA (data not shown). Among the 31 iceA2 strains, 21 possessed a single 35-aa internal cassette while 10 harbored two cassettes. Among the 19 iceA2 RNA-positive samples, expression levels were higher among persons colonized with strains possessing one versus two internal 35-aa cassettes (Fig. 5A). To evaluate the relationship between cassette structure and iceA2 and 16S rRNA expression in each biopsy specimen, we also derived an iceA2/16S rRNA ratio and compared these ratios with iceA2 cassette status among the 19 iceA2 RNA-positive strains. Consistent with the results for iceA2 RNA alone, iceA2/16S rRNA transcription ratios were significantly higher among strains containing one versus two cassettes (P = 0.018) (Fig. 5B). These results suggest that particular iceA2 gene structures, such as the presence of a single 35-aa internal cassette, are associated with enhanced iceA2 transcription and that this difference is not dependent upon variations in bacterial density.

View larger version (14K): [in this window] [in a new window]   FIG. 5.   Relationship of iceA2 gene structure and levels of iceA2 expression in vivo. (A) Among patients harboring iceA2 strains, 19 had detectable levels of iceA2 transcripts. iceA2 from corresponding H. pylori genomic DNA was sequenced using an ALF-Express automated sequencer. Mean values () ± SD are shown adjacent to the data points (). (B) The iceA2/16S rRNA ratio was determined by dividing iceA2 transcript levels by the corresponding 16S rRNA levels. Statistical significance was determined by the Mann-Whitney U test.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

H. pylori iceA1 strains have been associated with duodenal ulcer disease and enhanced neutrophilic infiltration of the gastric mucosa, yet considerable heterogeneity exists in the levels of inflammation among persons harboring iceA1 strains (20, 26, 32). Although iceA1 mRNA expression is induced following adherence to gastric epithelial cells in vitro, this model may differ from events occurring within the gastric mucosa, since it does not include potential contributions by other cells or soluble mediators to bacterial gene expression. H. pylori iceA2 strains have not been linked with disease outcome, but iceA2 mRNA has been detected within gastric tissue (26). To explore the hypothesis that the amount of iceA expression within the host environment, and not iceA genotype per se, may contribute to increased pathology, we quantitated iceA expression within gastric tissue and related this to the clinical and histologic outcome, the presence of additional H. pylori virulence-associated genes, and iceA gene structure. The results showed that, in a limited number of biopsy samples, expression of iceA1 but not iceA2 RNA was significantly related to the severity of acute mucosal inflammation. Although iceA expression and inflammation were quantified in different tissue samples, these biopsy specimens were harvested from the same antral site during endoscopy to more completely insure that comparisons between these variables were valid. In our population, iceA1 genotypes are present in approximately 20% of colonized persons, in contrast to recent reports from Europe (32) and East Asia (18, 20), where iceA1 is the predominant genotype. Since we detected transcripts in 40% of iceA1-infected biopsy specimens, only 10% of our total population would be expected to be classified as iceA1 RNA positive. Of interest, a recent report from Japan (20) has similarly demonstrated that although H. pylori iceA1 strains are associated with more severe gastritis, heterogeneity exists among inflammatory scores. We therefore look forward to the possibility of having our current results tested in a geographically distinct host population.

We found that in approximately one-half of the patients, iceA transcripts could be detected in vivo. One explanation for failure to detect iceA transcription may be that sequence variation inherent to iceA1 or iceA2 reduced oligonucleotide hybridization. However, amplification of iceA1 and iceA2 was successful using DNA from every strain as templates for the PCRs (Fig. 2B). Another possibility is that variation in H. pylori colonization densities affected the ability to detect iceA mRNA. The fact that there were no substantial differences in the amount of constitutively expressed 16S rRNA between biopsies with and without iceA transcripts, though, likely reflects similar H. pylori colonization levels. Finally, it is possible that some patients were colonized with multiple H. pylori strains, which could have contributed to the present findings. However, as we have previously reported, this appears to be an uncommon event in our patient population (33). Collectively, these data suggest that strain differences in iceA expression do not simply reflect bacterial load and that expression of these genes is regulated rather than constitutive. Factors that may contribute to regulation of iceA expression include differences in the gastric environment between colonized hosts and intrinsic strain differences, such as the structures of the genes themselves or activation of transcription factors necessary for their expression.

H. pylori iceA1 has significant homology to nlaIIIR, which encodes the restriction endonuclease NlaIII in N. lactamica (19), and is adjacent to hpyIM, the H. pylori homolog of the cognate N. lactamica DNA methyltransferase M.NlaIII. These data suggest that iceA1 may be part of a DNA restriction-modification system (26, 34), and data for one strain (CH4) possessing such a full-length ORF demonstrate that it has functional restriction endonuclease activity (16). There are several putative mechanisms by which restriction-modification systems may enhance virulence. For example, DNA ends generated by restriction may stimulate recombination events, and for organisms that colonize potentially hostile niches, such gene exchange may be more important than de novo mutations for maintenance of long-term infection. DNA methylation could also regulate expression of H. pylori genes that are required for inducing gastric mucosal inflammation. Recent studies suggest that iceA1 and iceA2 strains differ in the level of expression of hpyIM, the DNA methyltransferase immediately downstream of iceA (Q. Xu, J. P. Donahue, and M. J. Blaser, submitted for publication). Finally, iceA1 may be directly involved in induction of gastritis, although recent data from our laboratory have demonstrated that inactivation of iceA1 does not alter interleukin-8 production by gastric epithelial cells in vitro (R. M. Peek, Jr., unpublished data). Although this does not preclude a direct role for iceA1 in the establishment of gastritis, definitive evidence will require experimental challenge with wild-type and isogenic mutant strains in animal models of H. pylori-induced gastric inflammation, and these experiments are under way in our laboratory.

Although iceA1 may encode a functional restriction endonuclease, the majority of iceA1 sequences examined to date contain nucleotide insertions and/or deletions that would preclude translation of a full-length protein homologous to NlaIII (16). The size of the intergenic region between the end of the cysE ORF and the first iceA1 ATG codon (Fig. 1A) also varies, depending on the strain (16); for example, in 60190, the distance is 84 nucleotides, while in CH4, the distance is 25 nucleotides. In addition to sequence variation, different iceA transcriptional initiation sites exist among H. pylori strains. Northern blot analysis of iceA1 transcripts in strains 60190 and CH4 indicated that the P₁ promoter shown in Fig. 1A is the major site of transcriptional initiation. However, in strain 60190, primer extension data have also identified the 5' terminus of an iceA transcript in the intergenic region between the cysE and iceA1 ORFs, and RT-PCR results demonstrate read-through transcription from cysE into iceA1 (13). For strain CH4, which contains a continuous ORF homologous to the N. lactamica nlaIIIR ORF, we have demonstrated IceA1 restriction endonuclease activity in cell extracts (16). This indicates that transcripts initiated within the cysE-iceA1 intergenic region or upstream within cysE (possibly at the cysE promoter) must also be present in strain CH4. The P₁ promoter resembles an Escherichia coli ⁷⁰ promoter, with nearly consensus 10 and 35 sequences; however, sequences upstream of the putative transcriptional initiation site in the cysE-iceA1 intergenic region do not resemble consensus sequences shown to be recognized by any of the sigma factor homologs known to be present in H. pylori (⁸⁰, ²⁸, and ⁵⁴).

Although our iceA1 strain population was limited in number (n = 10), we did not detect any molecular iceA1 variant that was consistently associated with expression in vivo, and no expression was observed for a strain with a full-length NlaIII-like ORF. These results suggest that additional factors, such as adherence, rather than gene structure may more directly regulate iceA1 expression. H. pylori iceA1 was originally identified by its up-regulation following contact with gastric epithelial cells in vitro (26), and thus one physiological event that could potentially regulate iceA1 expression in vivo is binding to gastric epithelium. Since interaction with host epithelial cells is a stimulus for virulence-associated gene expression for other mucosal pathogens, such as Yersinia pseudotuberculosis and uropathogenic E. coli (28, 37), contact with gastric epithelial cells also may be a paradigm for induction of H. pylori genes linked with pathological outcomes, such as severity of acute inflammation.

In contrast to iceA1, our results indicate that strain differences in gene structure may be more important for iceA2 expression, since higher levels were seen for genes possessing one rather than two 35-aa cassettes. Differences in gene expression related to gene structure have been described for H. pylori vacA and cagA, which demonstrate either a mosaic cassette structure (3, 31) or variations in 3' repeat regions (35), respectively. The genetic structure of the H. pylori iceA2 locus differs considerably from that of iceA1 in that iceA2 contains variable numbers of highly conserved peptide repeat regions (Fig. 1B). The five distinct iceA2 variants that contain zero, one (two forms), two, or three copies of the internal 35-aa cassette have each been isolated from various regions of the world, indicating that the different iceA2 variants are highly conserved (16). Examination of the overall genetic architecture of the iceA2 locus has revealed strong (>90%) homology in the nucleotide sequences encoding the 14- and 10-aa external motifs (16). These findings lead to the hypothesis that these external motifs could facilitate recombination between iceA2 genes containing varying copies of the 35-aa cassette, which could result in differing levels of iceA2 transcription.

In summary, our experiments demonstrate that RT-PCR is an effective approach for quantitation of both constitutively and conditionally expressed H. pylori genes within gastric mucosa. Using this technique, we have demonstrated that iceA1 expression is significantly related to the host mucosal response while iceA2 expression may be more highly influenced by gene structure. An additional advantage of this technique is that generation of random-hexamer-primed cDNA provides the ability to quantitate expression of multiple prokaryotic and eukaryotic genes within the same tissue sample and to relate levels of transcription to inherent pathology.