INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Helicobacter pylori is now recognized as an important organism associated with peptic ulcer disease, gastric adenocarcinoma, and gastric mucosal-associated lymphoid tumor-type lymphoma (9, 16, 17, 23). Although putative virulence factors like cytotoxins (3), adhesins (12), and flagella (6) have been identified, the mechanisms by which H. pylori contribute to these diverse clinical outcomes remain poorly understood. Recently, H. pylori has been shown to synthesize two heat shock protein homologs with differing antigenic characteristics. Heat shock protein A (HspA) is a 13-kDa protein of the GroES class, and heat shock protein B (HspB) is a GroEL homolog of 58 kDa (5, 13, 22); the genes encoding these two proteins form a bicistronic operon (22). While HspB and the first 90 amino acids of HspA (A domain) are highly homologous to other bacterial heat shock proteins (11, 20), HspA contains a unique 27-amino-acid histidine-rich carboxyl terminus (B domain). Experimental studies have shown that this histidine-rich region is involved in urease activity, presumably secondary to nickel binding (22). While HspA is essentially a cytoplasmic protein, H. pylori cells often lyse and expose the internal antigens.

Although all H. pylori strains studied possess hspA, in two previous studies only 40% of persons infected with H. pylori had detectable levels of serum antibody against this protein (19, 22). These two studies involved North American and European patients, but the immunologic responses to H. pylori HspA among Asian populations have not been determined. In one of the studies, an association between HspA seropositivity and proximal gastric carcinoma was found, but this also could have reflected the advanced age of these patients (19).

H. pylori is highly diverse at the genomic level (1, 2, 14). Kansau et al. (10) demonstrated diversity in the deduced hspA-encoded peptide sequences among 32 strains studied. They reported nucleotide polymorphism for the region encoding the HspA B domain, but the diversity of this region for strains from non-European populations has not been explored. The aims of this study were to investigate whether in Asian patients the anti-HspA serologic responses are also heterogeneous and whether variation correlated with clinical outcome. A secondary aim was to examine the extent of variation in the nucleotide and amino acid sequences of HspA among H. pylori strains collected from different geographic locales and to determine whether this variation might help explain differences in host responses.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Patients studied. Between January 1994 and December 1996, 179 Hong Kong patients who were of Chinese descent and who presented with upper digestive tract symptoms were enrolled in this study after written informed consent was obtained. All were examined by esophagogastroduodenoscopy for investigation of symptoms, and demographic data were recorded. The presence of H. pylori was determined by culture and/or histological examination of the gastric mucosal biopsy specimens (15). In total, 154 patients (mean age, 52.3 ± 17.0 years; 94 males and 60 females) were confirmed to be carrying H. pylori. The diagnoses among these patients, were as follows: duodenal ulcer, n = 60; gastric ulcer, n = 29; gastric adenocarcinoma, n = 29; and unremarkable endoscopy, n = 36. For the remaining 25 (14%) patients (mean age, 45.6 ± 13.4 years; 17 males and 8 females), neither the presence of H. pylori nor any endoscopic abnormality was detected.

Serologic methods. Recombinant HspA produced as a fusion protein with the maltose binding protein MalE (MBP-HspA) or MalE alone (MBP) were harvested from DH5- Escherichia coli strains carrying pILL933 or pMAL-2, respectively, as described previously (10). The cells were induced with isopropyl--D-thiogalactopyranoside and lysed by passage through a French pressure cell, and the recombinant proteins were purified to homogeneity by large-scale affinity chromatography. The presence of anti-HspA immunoglobulin G (IgG) in patient sera diluted 1:100 was determined in parallel enzyme-linked immunosorbent assays (ELISAs) as described previously (19). Goat anti-human IgG conjugated with horseradish peroxidase was used as the secondary antibody and was used at a dilution of 1:4,000. For each patient, the optical density (405 nm) that resulted from the serologic reaction with MBP alone was subtracted from that obtained from MBP-HspA to calculate a net optical density. The ratio of the net optical density of the tested serum samples to that of a standard positive control specimen on the same plate was defined as the optical density ratio (ODR). The cutoff ODR for seropositivity was 0.182, which was determined for a group of 40 asymptomatic volunteers known to be H. pylori negative. The cutoff value was equivalent to 3 standard deviations above the mean ODR for these 40 serum samples (data not shown). The presence of serum IgG antibodies to H. pylori CagA was also determined by ELISA, as reported previously (18).

Characterization of H. pylori isolates from Hong Kong patients. H. pylori isolates from 14 Hong Kong patients were used to analyze the hspA nucleotide sequence; 7 patients each were HspA seropositive or HspA seronegative and were randomly selected. After growth on Trypticase soy agar with 5% sheep blood under microaerobic conditions for 48 to 72 h, bacterial cells were collected for DNA extraction as described previously (25). A 487-bp segment containing the 384-bp hspA gene was amplified by PCR with the primers (5'-TGCGCTATAGTTGTGTCGC and 5'-GCTATCTGAAAATTTGATTTCTTTTGC) described by Kansau et al. (10).

Characterization of other isolates. For comparison with the Hong Kong isolates, we analyzed hspA nucleotide sequences from 48 other strains. We included the hspA sequences of 39 clinical isolates from France, whose amino acid sequences were reported previously (10), as well as from strain 26695, for which the genomic sequence was recently published (24). Two H. pylori isolates collected from each of three members of a Dutch family (27) were also studied. All six strains had previously been analyzed by PCR-based random amplified polymorphic DNA fingerprinting, which showed that all six strains were highly similar but not identical. The pairs of strains from each patient differed in their cagA status. Strains 2a, 3a, and 5a were cagA positive, while strains 2b, 3b, and 5d were cagA negative. In addition, the hspA sequences of two H. pylori isolates from rhesus monkeys, previously designated 31001 and ATCC 51407, respectively (4), were also determined.

Statistical analysis. The chi-square test, Fisher's exact test, and Student's t test were carried out with Epi-Info software, when appropriate. The actual values for the anti-HspA serologic responses of patients with different clinical entities were also compared by a one-way analysis of variance test. Differences were defined as being statistically significant if the P value was less than 0.05.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Serologic responses to HspA. As expected, none of the 25 patients who were not colonized with H. pylori showed an IgG serologic response to H. pylori HspA, with all ODR values being below 0.1. This finding confirmed the high specificity of the IgG HspA ELISA for this Asian population. Of the 154 serum samples collected from persons known to carry H. pylori, 54 (35.1%) were HspA seropositive (Table 1), a proportion close to those reported previously (19, 22). Although patients with either duodenal ulcer or gastric ulcer had seropositivity rates slightly higher than those for the other groups of patients, these differences were not statistically significant. Among the entire group (Fig. 1), a bimodal distribution of the anti-HspA serologic responses was observed, and the distribution was similar to that reported previously (19, 22). The mean (standard deviation) anti-HspA ODRs for patients with stomach cancer, duodenal ulcer, gastric ulcer, and nonulcer dyspepsia were 0.14 (0.30), 0.18 (0.28), 0.26 (0.28), and 0.20 (0.33), respectively, and were not significant (P = 0.51; one-way analysis of variance).

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Distribution of serologic responses against H. pylori HspA for serum specimens from 154 patients known to be carrying H. pylori (black bars) and 25 patients not carrying the organism (white bars).

Patient characteristics and serologic responses. To assess the effect of other patient factors on the serologic responses to HspA, patient demographic data, social habits, and serologic responses to the CagA antigen of H. pylori were studied (Table 2). There were no significant associations between patient gender, smoker status, alcohol use, or anti-CagA IgG response and the results of anti-HspA serology. On average, patients who were seropositive were 4.5 years older than those who were seronegative. To determine whether there was any effect of age on the HspA serology in this population, we subclassified the studied patients into three age groups. In that analysis (Table 3), the risk of seropositivity for HspA was significantly correlated with patient age (P = 0.04).

Nucleotide sequences of H. pylori hspA. Direct DNA sequencing of the 354-bp hspA segment for the 14 Hong Kong strains revealed that no 2 strains had identical nucleotide sequences. The isolates from the seven hosts who were seropositive did not form a distinct cluster but were intermixed with the isolates from the seven seronegative hosts. In contrast, analysis of hspA sequences of the six strains from members of a multigeneration Dutch family (27) revealed complete identity. Neighbor-joining analyses for the sequences of the entire hspA gene and the B domain alone were performed and resulted in similar trees. The tree derived from the B domain alone was a consensus of 1,000 bootstrap replicates and suggested the geographic segregation of H. pylori strains (Fig. 2). All except two of the Hong Kong strains were part of a major (East Asian) branch of the dendrogram (Fig. 2). Six strains isolated in France were also part of the East Asian branch. However, two of these strains (strains F27 and F30) had actually been isolated from patients of Far Eastern origin residing in France. The other three major branches consisted mainly of isolates from persons of European or North African origin; there was no obvious distinction between the isolates from persons in these two groups. However, it is noteworthy that the East Asian branch was not supported in a majority of bootstrap replicates, and the classification of a distinct East Asian clonal grouping awaits the characterization of larger or additional gene segments. Nevertheless, when the proportion of East Asian strains in the Asian cluster is compared with that of those outside the cluster, the difference is statistically significant (15 of 21 versus 1 of 31 [P < 0.001; two-tailed Fisher exact test]).

View larger version (66K): [in this window] [in a new window]   FIG. 2.   Phylogenetic tree of hspA nucleotide sequences of 62 H. pylori strains of different geographic origins. Branches that represent more than one strain are indicated by asterisks. The six strains from the members of a Dutch family (25) with identical sequences are represented by the minor branch listed as 2a. Hong Kong strains are denoted by the prefix "97." Strains with the prefix "F" were isolates studied at the Institut Pasteur (10). The gene encoding the E. coli GroES homolog was used as an outgroup to root the hspA tree. The true branch length for the E. coli sequences was four times longer than shown in the figure. The shaded region refers to the major Asian cluster of hspA in the dendrogram.

Comparison of HspA amino acid sequences. The translated amino acid sequences of HspA from all 14 Hong Kong strains examined were highly similar to that of the major variant reported previously (10). Although there was a high degree of conservation in domain A compared to the published sequences (10), variations were chiefly detected at residues 68, 75, and 90. There was greater variation in domain B, again almost exclusively confined to three residues: residues 98, 99, and 110 (Fig. 3). None of the HspA molecules had the signature (95ANSxxxxxHxHA107) of the minor variant that was previously described to be present in 10% of the European isolates. Only 2 of the 14 Hong Kong strains had the same primary HspA sequences; the other 12 strains had different substitutions at one or more amino acids. There was no association between the variation within domain B of the H. pylori isolate and the presence of serologic responses by the 14 Hong Kong patients.

View larger version (38K): [in this window] [in a new window]   FIG. 3.   Translated HspA amino acid sequences of 14 H. pylori strains from Hong Kong patients in comparison with that of reference strain 85P. The vertical line separates domains A and B between residues 90 and 91. The A domain represents the conserved portion of HspA that is homologous to GroES.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

That a minority of persons colonized with H. pylori in Hong Kong have serum IgG responses to HspA confirms previous studies with non-Asian populations, and the fact that the positivity rate is approximately the same (35% versus 38 to 39%) indicates that the immunologic response to H. pylori HspA is probably not related to ethnicity (19, 22). The bimodal distribution of the serologic results also paralleled that reported previously (19). The role of serum IgG antibodies in relation to mediation of the inflammation and tissue injury associated with H. pylori is unknown. It has been suggested that the dichotomous response to certain H. pylori antigens is a host factor-related phenomenon and may be correlated to the clinical diagnosis for the patients. Therefore, we next sought to determine whether there is any relationship between patient characteristics and the serologic results. Among all the patient factors studied, only age correlated with HspA seropositivity, confirming an age-related effect described previously (19). Although the mechanism behind this relationship remains unknown, one possibility is that the immunologic response to HspA reflects prolonged exposure to the organism. Unlike the previous study (19), we did not demonstrate any association between gastric cancer and HspA serology, even though the mean age of the cancer patients was significantly higher than that of the patients with duodenal ulcers or an unremarkable endoscopy. The age-related responses may reflect a breach of tolerance, since human cells also possess GroES-like antigens, but they are unlikely to be related to atrophic gastritis since gastric cancer patients were not more frequently seropositive. Alternatively, H. pylori HspA expression may be heightened in strains present in the stomachs of older persons.

Few previous studies on the genetic diversity of H. pylori have studied geographic segregation (7, 8). While the patients in those studies were residents within the same area where endoscopy was performed, it is now clear that most H. pylori infections are acquired during early childhood (21). That each of the 14 strains from the Hong Kong patients showed a different hspA nucleotide sequence, whereas the 6 strains from the Dutch family were identical, despite other genetic differences among the Dutch strains, confirms that the Dutch strains are closely related (27). In addition, it suggests that the rate of mutation in hspA is not unusually high. In the present study, the results of the phylogenetic analysis of hspA nucleotide sequences may be indicative of geographic clustering. It may suggest that H. pylori diversification has continued since the time that Asian and European human populations separated from each other. However, the majority of the bootstrap replicates did not support H. pylori strain clustering by geographic origins on the basis of the hspA sequences. This also was observed by van der Ende et al. (28), confirming the sequences of glmM of Dutch and Chinese H. pylori isolates, and are consistent with the multilocus enzyme electrophoresis analysis of 74 H. pylori isolates without geographic clustering by Go et al. (7). Further analysis with larger numbers of strains might better clarify the population structure. The segregation between Dutch and Chinese H. pylori isolates in a dendrogram on the basis of cagA sequences may be explained by the location of cagA on a genetic element with a different evolutionary background (28).

That the A domain is highly conserved at the amino acid level confirms previous findings (10). The near invariance of the B-domain histidines is also consistent with an important functional role, perhaps related to nickel binding (10). However, the remaining amino acids in the B domain, which show more than 10-fold variation compared with those for the A-domain and the B-domain histidines, are thus not as strongly conserved. The presence of substantial variation at different sites within a single gene suggests that selection varies at particular codons, reflecting different functional constraints. The high degree of similarity among hspA sequences from human and monkey strains suggests a very close relationship between H. pylori in humans and other primates, a finding consistent with an ancient origin for H. pylori.

In conclusion, we confirm and extend previous studies (10, 22) that the diversity of hspA does not account for the bimodal distribution of serologic responses to HspA among persons colonized with H. pylori. The basis for this phenomenon, although related to patient age, remains unexplained.