CRISPR-like sequences in Helicobacter pylori and application in genotyping

Clustered regularly interspaced short palindromic repeats (CRISPR) are detected in around 40% of bacteria and many archaea [1, 2]. CRISPR together with the CRISPR-associated genes (cas) are a prokaryotic defense system against invasive bacteriophages or genetic elements. Cas proteins function to degrade foreign nucleic acids of bacteriophages or plasmids. The number of CRISPR loci detected vary between and within bacterial species and strains, 1–4 in Escherichia coli [3], 3 in Yersinia pestis [4], 2 in Salmonella Typhimurium [5], and 1–2 in Staphylococcus aureus [6]. CRISPR loci contain multiple direct repeat (DRs) sequences from 21 to 48 bp long separated by variable spacer sequences 21–72 bp in length [7]. DR sequences are commonly conserved whereas spacer sequences are diverse, and derived from bacteriophages or plasmids. The variable number of DRs and spacers have been used as a typing tool in epidemiologic and evolutionary analysis of bacterial strains [8].

Helicobacter pylori is a Gram negative bacterium that causes peptic ulcer, gastric cancer and mucosa associated lymphoid tissue lymphoma. The risk of disease is associated with H. pylori harboring the cytotoxin associated gene A and vacuolating cytotoxin A, encoded by cagA and vacA genes respectively. The cagA gene encodes the bacterial oncoprotein that causes abnormal cellular signals leading to deregulation of cell growth, cell turnover, cell to cell contact, and elongation of epithelial cells. The vacA gene encodes the pore-forming toxin that induces epithelial cell apoptosis, and inhibits leukocyte activation by massive vacuolization, and disruption of the endosome [9]. Allelic variation of vacA occurs in a signal region (s1/s2) and a middle region (m1/m2) resulting in different levels of vacuolating cytotoxicity. The vacuolating activity is high in the s1m1 genotype whereas the intermediate and absent activities are associated with s1m2 and s2m2 genotypes, respectively. H. pylori carrying the cagA and vacA s1m1 allele has been frequently isolated from patients with severe gastric diseases including peptic ulcers, atrophic gastritis, and gastric cancer [10]. The gene encoded for the virulence-associated protein D (vapD) is present in all virulent strains of ovine footrot bacteria, Dichelobacter nodosus, although the VapD protein function in this bacterium is unknown [11]. This gene has been reported in some H. pylori isolates with 64.9% nucleotide identity to the vapD gene of D. nodosus [12]. The vap region of D. nodosus has been demonstrated to harbor genetic element of bacteriophage suggesting the possibility that gene in this region may be transferred among bacteria [13].

In studies of H. pylori, biotyping, serotyping and hemagglutinin typing, have been reported to possess low discriminatory power index (DI) compared to genotyping such as RAPD [14], pulse field gel electrophoresis (PFGE) [15], restriction fragment length polymorphism-PCR (RFLP-PCR) [16], and repetitive extragenic palindromic PCR fingerprinting (REP-PCR) [17]. PFGE is not widely used for H. pylori because inter-patient variation is rare in the fingerprints obtained [18]. The DI of PFGE is between 0.24 and 0.88 whereas RAPD analysis reveals excellent DI (between 0.99 and 1). Thus, RAPD is recommended for H. pylori typing [19].

Analysis of the CRISPR-Cas systems in H. pylori has not been clearly demonstrated. The polymorphism detected in CRISPR loci has been applied as a genetic marker for typing many bacteria, such as Campylobacter fetus [20] and S. Typhimurium [5]. In the present study, the CRISPR sequences and virulence genes of this bacterium were investigated and CRISPR-virulence typing was compared to the RAPD.

Type II CRISPR-Cas systems are present in Helicobacter cinaedi (NC017761), Helicobacter mustelae (NC013949) and Wolinella succinogenes (NC005090) [28]. In Helicobacter cetorum (NC017737) a type III CRISPR-Cas system is detected. However, no cas gene could be identified in H. pylori, thus CRISPR-like sequences is assigned for the CRISPR loci detected in this work.

It has been reported that H. pylori and several Helicobacter spp. possess vacA-like genes. In H. pylori, these genes are designated as imaA (immunomodulatory autotransporter A), faaA (flagella-associated autotransporter A) and vlpC (VacA-like protein C-Vlp C) [29]. It is of interest that all the CRISPR-like sequences of H. pylori identified in this work are located in the vlpC gene (Fig. 4). Analysis of faaA and imaA of 5 H. pylori isolates (26695-1MET, XZ274, F57, India7, and SNT49) indicated the absence of any CRISPR-like sequences. Although the function of Vlp C has not been clearly illustrated, mutation of the vlpC gene is associated with the resistance of H. pylori to metronidazole [30]. Analysis of H. pylori 26695-1MET Vlp C using the PHYRE2 program (the protein homology/analogy recognition engine v 2.0) indicated that this protein was homologous to the viral A-type inclusion protein (E = 2 × 10⁻⁶⁹) of the Trichomonas vaginalis virus (TVV). Interestingly, the presence of TVV in T. vaginalis causes resistance to metronidazole [31]. The NCBI database indicates that only H. pylori and H. cetorum harbor this gene. Analysis of vlpC of H. pylori F57 and H. cetorum MIT 99-5656 revealed the homology of nucleic sequences to be between 72 and 80% and this gene contained 2 regions, a high variable region and a low variable region (Fig. 4). The CRISPR-like sequences were identified only in H. pylori and were located in the high variable region (between 777 and 1030 nucleotide sequences) with AT-rich leader sequences and contained 2 regions (between 819–916 and 968–1030) homologous to H. cetorum sequences. AT-rich leader sequences assumed to be a transcriptional promoter are assigned as one of the CRISPR feature structures [22, 32]. Thus, it is possible that the CRISPR-like sequences were inserted into this part of the vlpC gene, which is not the conserved domain or functional region of the gene. Analysis of 4210 CRISPR arrays revealed 89% were cas-positive in which 33% CRISPR arrays were adjacent to a cas locus and 56% were located outside of cas loci, the rest 11% were orphans (no cas detected) [33]. It has been demonstrated that orphan CRISPR may involve in gene regulation. In Listeria monocytogenes, orphan CRISPR affects virulence through feoAB iron transport system [34]. In H. pylori the presence of CRISPR-like sequences in vlpC gene may influence its expression. In vivo investigation of H. pylori three vacA-like genes revealed that they were up regulated during colonization of mouse stomach; however, in vitro evaluation of VacA-like proteins in broth culture supernatant using Western blot analysis, ImaA and FaaA were detected but Vlp C was not detected which might be due to it was expressed in the low level [35]. The CRISPR-like sequences of H. pylori are detected in the coding region of a vlpC gene, it is uncommon that CRISPR locus of H. pylori located in the coding region, however, it has been reported that CRISPR loci I, III, IV and V of Thermotaga maritima were detected in the genes encoded for the hypothetical protein [36]. It is possible that H. pylori acquires those sequences after vlpC gene is already presented in its genome because no CRISPR-like sequences is detected in the vlpC gene of H. cetorum.

In this work, we detected 23–31 bp DRs in H. pylori and the number of spacers was between 1 and 3, with 25–37 bp. Twenty-eight spacer patterns were classified from H. pylori isolates. Interestingly, H. pylori designated as PBH03 and SKH03 possessed the same virulence gene pattern (Fig. 3) and s1 and s2 spacers of PBH03 were organized in the spacer pattern 28 and 16 respectively which were the same patterns of s1 and s3 of SKH03 respectively (Fig. 1). Thus, it is possible that those bacterial isolates might be invaded by the same kind of phage. In this work, some spacers were 83–100% similar to genetic elements encoding phage DNA polymerase (PBH03-s2, PBH06-s3, SKH03-s3 and SKH04-s3), phage tail protein (PBH13-s2, SKH02-s2 and SKH05-s2) or phage DNA packing protein (SKH04-s1) whereas some spacers were similar to hypothetical proteins derived from bacterial plasmid (PBH06-s1, SKH03-s1, SKH06-s1) (Additional file 2: Table S1). These genetic elements could be bacterial signatures for typing.

H. pylori cagA ⁺, vacA s1m1strains have been associated in the development of peptic ulcer and gastric carcinoma more than H. pylori cagA ⁻ , vacA s1m2 strains [10, 37]. In this work, all H. pylori isolates were cagA ⁺ strains and the occurrence rate of vacA s1m1 and vacA s1m2 strains was 61.1% (22/36) and 38.9% (14/36), respectively. All vacA m2 H. pylori isolates were organized in the CVT-A cluster of CRISPR-virulence typing (Fig. 3). The previous studies of H. pylori in Thailand demonstrated that 95% of isolates were cagA ⁺ whereas the vacA s1m1 and s1m2 genotypes were found in 65 and 35% of total isolates, respectively [38]. In this study, the prevalence rate of pathogenic H. pylori (cagA ⁺ vacA s1m1) in samples collected in the southern region was 71.4% which was significantly higher than the prevalence rate in the samples from the central region (58.6%).

H. pylori VapD protein possesses endoribonuclease activity and may be involved in cell maintenance or the protection of foreign genetic elements [39, 40]. It is of interest that only 5.5% (2/36) of H. pylori obtained in this work were positive for the vapD gene whereas it was present in 38.0% of H. pylori isolates from Mexico and 61.3% from the USA [12, 25]. Using hybridization and confirmation by sequencing, it has been demonstrated that vapD-negative H. pylori isolates possessed high-level genetic diversity in the vapD region [12]. Differences in the detection of this gene in Asian (Thailand) and Western (Mexico and USA) H. pylori isolates should be explored.

Various patterns of spacers in the CRISPR-Cas system have been reported for subtyping C. jejuni, Mycobacterium tuberculosis, Erwinia amylovora, V. parahaemolyticus and S. enterica [8, 20, 41]. Analysis of orphan CRISPR spacers derived from 14 clinical Enterococcus faecalis isolates indicates two groups of clonal strains which correlates to the Multilocus sequence typing (MLST) [42]. Better bacterial discrimination through a combination of CRISPR pattern and virulence gene typing has been demonstrated [43, 44]. In the present study, the DI of CRISPR-virulence typing was higher than that of CRISPR typing alone (Additional file 3: Table S2). Analysis of 20 profiles derived from CRISPR-virulence typing and RAPD revealed that DIs of both techniques were 1. However, RAPD seems to be better than CRISPR-virulence typing because it could differentiate H. pylori PBH12 from the other strains (Fig. 2) whereas only two clusters were obtained from CRISPR-virulence typing (Fig. 3). This might be due to the low number of spacers detected in the tested isolates in this work. However, the reproducibility of RAPD is poor and the method lacks validated interpretation criteria for inter-laboratory comparison [18, 45]. CRISPR-virulence typing is more reliable for long-term global epidemiology and evolutionary studies and the CRISPR locus databank is excellent for inter-laboratory interpretation and inter-laboratory reproducibility. The information of the CRISPR-like sequences in H. pylori acquired in this work expands the basic knowledge for typing of bacteria.

This is the first report of CRISPR-like sequences in H. pylori which was detected in the vlpC gene. Twenty-eight different spacers with the numbers between 1 and 3 were identified and can be used as bacterial signature for genotyping. In this work, CRISPR-virulence typing of H. pylori was performed and compared with the RAPD. Their DIs were not different. However, RAPD is poor reproducible and lacks validated interpretation criteria for inter-laboratory comparison. Therefore, CRISPR-virulence typing is a novel epidemiological tracking tool for H. pylori management.