Mixed Infection with cagA Positive and cagA Negative Strains of Helicobacter pylori Lowers Disease Burden in The Gambia

Background The prevalence of Helicobacter pylori including strains with putatively virulent genotypes is high, whereas the H. pylori-associated disease burden is low, in Africa compared to developed countries. In this study, we investigated the prevalence of virulence-related H. pylori genotypes and their association with gastroduodenal diseases in The Gambia. Methods and Findings DNA extracted from biopsies and H. pylori cultures from 169 subjects with abdominal pain, dyspepsia or other gastroduodenal diseases were tested by PCR for H. pylori. The H. pylori positive samples were further tested for the cagA oncogene and vacA toxin gene. One hundred and twenty one subjects (71.6%) were H. pylori positive. The cagA gene and more toxigenic s1 and m1 alleles of the vacA gene were found in 61.2%, 76.9% and 45.5% respectively of Gambian patients harbouring H. pylori. There was a high prevalence of cagA positive strains in patients with overt gastric diseases than those with non-ulcerative dyspepsia (NUD) (p = 0.05); however, mixed infection by cagA positive and cagA negative strains was more common in patients with NUD compared to patients with gastric disease (24.5% versus 0%; p = 0.002). Conclusion This study shows that the prevalence of H. pylori is high in dyspeptic patients in The Gambia and that many strains are of the putatively more virulent cagA+, vacAs1 and vacAm1 genotypes. This study has also shown significantly lower disease burden in Gambians infected with a mixture of cag-positive and cag-negative strains, relative to those containing only cag-positive or only cag-negative strains, which suggests that harbouring both cag-positive and cag-negative strains is protective.


Introduction
Helicobacter pylori is a genetically diverse microaerophilic gram negative bacterial species that chronically infects the human gastric mucosa, often starting in infancy and lasting for life [1]. About 50% of the world's adult population is colonized, with prevalences of over 80% in many developing countries including The Gambia [2][3][4]. Earlier reports indicated high prevalence of H. pylori colonization, but a low frequency of H. pylori-associated disease in Africa [2,5,6], a phenomenon that was called the ''African enigma'' [2]. DNA sequencing of housekeeping and virulence genes has shown that different sets of genotypes predominate in different human populations [7]. Of particular interests have been H. pylori's cagA oncogene and toxigenic s1 and m1 alleles of its vacA gene, which have been implicated in gastroduodenal diseases caused by this pathogen both in epidemiologic [8,9], experimental animal and cell culture infection [10] studies. This said, several studies from different world regions have not detected such an association [8,11,12], an outcome suggesting the possibility of other virulence-modulating factors.
Individuals can be colonized by either a single or multiple strains of H. pylori, and even colonization by what is initially a single strain can, over time, lead to the emergence of multiple H. pylori subpopulations, due variously to mutation or to genetic recombination either between duplicate sequences in the single strain's genome or with DNAs from other transiently colonizing strains [8]. The prevalence of such mixed infections has been reported to vary (5-68%) [13][14][15][16][17] depending on geographical region, whether in a developed or developing country (low and high overall infection risk, respectively), and probably also methods of analysis. The H. pylori virulence-associated vacuolating cytotoxin (vacA) and cag pathogenicity island (cag PAI) genes, and also the cag empty site in strains lacking the cag PAI, are typically found in only one copy per genome [17][18][19][20]. Accordingly, detection of both the cagA gene and the cag empty site, or of both s1 and s2 (signal sequence; at 59 end of gene) or both m1 and m2 (middle region) alleles of vacA in a biopsy or in pool of H. pylori from a person indicates mixed infection.
We wondered if having mixed infection might influence the risk of gastric disease; for example, if strains of different genotypes might occupy a broader range of niches in the stomach as has been seen during experimental infection [21] and thereby impact on clinical outcome. In this study, we investigated the genotypes of H. pylori in The Gambia and the relation of apparently single versus mixed infections to gastroduodenal diseases.

Ethics statement
Ethical approval of this study was obtained from the joint Medical Research Council (MRC) Unit, The Gambia/Gambia Government Ethics Committee and Division of Microbiology Infectious Diseases (DMID) International Review Board.

Patients
Clinical data from the MRC Unit in The Gambia revealed that of 428 patients with gastric complaints investigated by gastric endoscopy between 2003-2008, 8 (1.9%) had gastric carcinoma, 20 (4.7%) and 15 (3.5%) had gastric and duodenal ulcers respectively, and that the others (89.9%) did not have such overt disease (diagnosed as non-ulcer dyspepsia; NUD (data not shown). All patients referred for endoscopy to the MRC Unit during the years 2003 to 2008 were initially considered eligible for inclusion: All 169 subjects who agreed to join this study provided written informed consent; in addition, for children less than 18 years, antral biopsies were obtained only after informed written parental consent. Patients with severe oesophago-gastroduodenal disease, including those with gastro-oesophageal varices, a small number with advanced gastric cancer and those on H. pylori eradication therapy, were excluded from the study. One hundred and twenty one patients from whose biopsies we successfully amplified virulence genes were analysed here. The mean age of these subjects was 35, ranging from 9 to 80 years. All the subjects were Gambians and most of them (75) came from the Greater Banjul Area, 38 from the West Coast region, 5 from Lower River region and 3 from North Bank region of The Gambia.

Gastroscopy results
Endoscopic examination showed that of the 121 study subjects, 11 had gastric ulcer, 7 had duodenal ulcer, 1 had both gastric and duodenal ulcers, 7 had gastric erosions, 1 had gastric carcinoma and all other subjects (94) who presented with either abdominal pain or dyspepsia had no evidence or history of gastric or duodenal ulcers.
The biopsies collected from patients were stored in Brain Heart Infusion (BHI) broth containing 20% glycerol and transported on ice to the laboratory for processing or stored at 270uC until used. Our previous data demonstrated that in The Gambia detection of mixed H. pylori strains in individual biopsies was best undertaken by PCR amplification directly from biopsy material rather than by bacterial culture [22].

Genomic DNA extraction directly from biopsies
Total genomic DNA was extracted directly from the biopsy material by using a combination of bead-beater and the QIAamp DNA isolation kit (Qiagen, UK) as previously described [22].

PCR amplification
PCR was performed to detect the cagA gene and cag empty site, and the signal sequence (s1 and s2) and middle region (m1 and m2) alleles of the vacA gene, as previously described [23] using the primers listed in table 1 and the following cycling conditions: 30 cycles of 94uC for 1 min, 55uC or 60uC for 1 min and 72uC for 1 min. The amplified genes were detected by electrophoresis in a 1.5% gel with ethidium bromide (500 ng/ml) and bands were visualized using Gel Doc 2000 (Bio-Rad laboratories, Milan, Italy). The presence of a particular gene or allele was inferred when a product of the expected size (table 1) was obtained using appropriate primers.

Statistical analysis
We assessed the prevalences of infection with single vs. multiple strains. For the cagA gene, for example, we noted the occurrence of cagA positive, cag empty site and mixed (cagA positive and cag PAI negative) infections. Prevalences were compared between disease groups and p-values were determined using Fisher's exact test.

Prevalence of H. pylori genotypes
One hundred and twenty one patients of the 169 study participants were inferred to be infected with H. pylori when DNAs extracted from their biopsies were tested by PCR for the presence of H. pylori cagA gene and cag empty site. Seventy four biopsies (61.2%) were positive for the cagA gene only, 21 (17.4%) were positive for the cag empty site only and 23 (19%) were positive for both. In parallel we also tested for the vacA gene presence and allele types. In all, 93 of 121 (76.9%) were positive only for the vacAs1 allele, 23 (19.0%) were positive only for the vacAs2 allele and 1 (0.8%) was positive for both. Only m1 or only m2 alleles of vacA were detected in 55 (45.5%) and 36 (29.8%) of biopsies tested respectively; both m1 and m2 (mixed infections) were found in 22 (18.2%) biopsies and up to 6.6% of biopsy DNAs failed to amplify for individual alleles (table 2).

Association between H. pylori genotypes
Of the 93 H. pylori strains that were positive only for vacAs1, 72 (77.4%) were cagA positive compared with only 1 (4.3%) cagA positive among the 23 strains that were positive only for vacAs2; most (16) of them contained the cag empty site allele only (table 3). Similarly, nearly all s1m1 positive biopsies (92.5%) contained cagA genes, whereas none of those containing only vacA s2m2 allele were cagA positive (table 3).
All 27 subjects with overt gastric diseases were of uniform cagA status (that is, uniquely cagA gene positive or cag empty site positive), whereas only 72.3% (68/94) of NUD were of uniform status; the other 23 contained mixed (cagA positive, cag empty site positive) infections. Three other biopsy samples did not give cagA gene or cag empty site amplification (table 5). This association between uniform cagA status and overt disease was statistically significant (p = 0.002).
In terms of age distribution, no association was found between age and overt gastric disease (24.5%,30 years, 12.

Discussion
H. pylori infection is common in dyspeptic adults in The Gambia [4,22], as is typical of developing countries. The range of H. pylori genotypes implicated in overt gastroduodenal disease as opposed to benign colonization or possibly even beneficial carriage [9,24] had not been extensively investigated in Sub-Saharan Africa.
Here we studied the distribution of H. pylori's main virulence genes, cagA and toxigenic alleles of vacA, in The Gambia, and their possible associations with disease outcome.
The prevalence of gastroduodenal disease (10%) that we detected endoscopically is similar to that reported elsewhere in Sub-Saharan Africa [25] and may be lower than that in Europe and North America. However, these estimated prevalences should be interpreted with caution as they may not be representative of the general population, but instead indicate the prevalence of disease among people with gastric complaints of sufficient severity to prompt diagnostic endoscopy. We found that just over half of Gambian isolates (61.2%) carried the cagA gene (table 2) and that mixed infections (both cagA positive and cag PAI negative) were common. We also found toxigenic vacAs1 and m1 alleles to be abundant, but not universal (76.9% and 45.5%). In contrast, most H. pylori strains from Egypt carried vacAs2 and m2 alleles, (57.1% and 85.7%, respectively) [9], whereas some 90% or more of strains in Japan, coastal China, and India carried vacAs1 alleles [23,[26][27][28]. vacAm1 type alleles were also nearly universal in Japanese main island strains, whereas vacAm2 strains were predominant in coastal China [28,29]. In this study, most of the vacAs1 strains (77.4%) were cagA positive similar to what was observed in a study in South Africa [30]. Up to 6.6% of samples in this study failed to amplify for individual genes consistent with other findings [22,30], perhaps due to PCR inhibitors or potent nucleases in some gastric biopsies as also suggested previously [22,30,31].
Thus, our data reinforce conclusions that different H. pylori genotypes, especially types of genes or alleles implicated in virulent vs. benign infections, predominate in different human populations. Strains of the vacA s1m2 type were most common in coastal China, and also seemed to predominate in Southern Nigeria [32]; and both s1m2 and s2m2 allele types were abundant in South Africa [30]. This contrasts with the predominance of vacA s1m1 strains found in our Gambian study participants. This apparent difference between Nigerian and Gambian strain genotypes could have several explanations, including the distance separating these two West African countries (.3000 km), or climatic differences (high rainfall and humidity in Southern Nigeria vs aridity for much of the year in The Gambia). Such explanations would entail genetic divergence by random genetic drift and selection for adaptation to local conditions respectively.    Table 6. Association between vacA genotypes and disease type. Previous reports of a lower than expected prevalence of H. pylori associated disease in Africa [2,5,6] might in principle reflect the influence of bacterial and/or human genotypes, environment including other infections (e.g. parasitic infections that affect host response to H. pylori), normal gut flora, diet (including antioxidants, salt, spices) or likelihood of seriously ill persons being diagnosed and their cases entered in registries. In accord with this last explanation, have been suggestions that gastroduodenal disease is actually common in Africa, that there is no African enigma [11,12]. The cagA gene and vacAs1 and m1 alleles are often linked to severe disease, and the vacAs2 and m2 alleles with more benign infections (gastritis only) in other populations. This is partially reflected in our results: with respect to cagA we did not find a ''Gambian-H. pylori virulence gene'' enigma; disease associations with vacA were less clear cut. The possible effects of bacterial or human genetic and physiologic differences, food, history of other infections and other environmental and lifestyle factors, on outcomes of chronic H. pylori infections in sub-Saharan Africa merit further more detailed analyses.
Most important, was our finding that co-existence of cagA positive and cagA negative strains was significantly more common amongst patients with NUD than among those with overt disease, which suggests that mixed colonization is protective. In principle, protection against development of overt gastric disease might stem from simple competition -whereby carriage of an avirulent (cagAnegative) strain diminishes the vigour of growth of a coexisting virulent strain, thereby reducing its impact on host tissues. It is also possible that factors in cag-negative strains that diminish the impact of virulence proteins such as CagA might predominate during cagA-positive and cagA-negative mixed infections [33]. Or, more generally, an increased complexity of immune responses during chronic infection by multiple divergent H. pylori strains might effectively diminish the inflammatory action of an individual virulent strain, and thereby resultant pathology in host tissues, as noted with other infections [34][35][36][37]. In accord with this idea, the risk of developing overt disease seemed higher in subjects apparently colonized only with cagA negative H. pylori, than in those with mixed cagA positive and negative strains. Conversely, however, the presence of mixed infections might also stem from increased intrinsic host susceptibility to H. pylori infection and equally the development of a more severe clinical outcome [38,39].
This study has revealed frequent gastro-duodenal disease among Gambians with gastric complaints. Many strains carried cagA + and s1, m1 alleles of vacA, which are disease associated in many European and North American populations. Although cagA status was associated with disease in The Gambia, alleles of vacA were not. Comparison of our data with those from southern Nigeria pointed to a potentially significant difference in linkage of signal sequence (s1 vs. s2) and middle region (m1 vs. m2) alleles, which control the potency and tissue specificity of toxin action respectively (s1m1 most common in The Gambia, vs. s1m2 most common in Nigeria). The possibility that such differences reflect selection for optimal genotypes or random genetic drift in these well-separated West African nations merit further study. We suggest that our most interesting finding is the significantly lower disease burden in Gambians infected with a mixture of cag-positive and cag-negative strains, relative to those containing only cagpositive or only cag-negative strains. The possibility that repeated exposure would be beneficial in Sub-Saharan Africa and in developing countries more generally needs to be considered when developing more effective treatment strategies for treating H. pylori infection and risks of gastroduodenal disease [40].