Rapid evolution of the Helicobacter pylori AlpA adhesin in a high gastric cancer risk region from Colombia

Introduction

Helicobacter pylori, a Gram-negative bacterium, has persistently colonized the stomach of half of the human population (Perez-Perez, Rothenbacher & Brenner, 2004; Khalifa, Sharaf & Aziz, 2010). This infection produces an asymptomatic inflammation of the gastric epithelium, but in some patients, it progresses toward a more severe clinical disease, such as ulcers and gastric cancer (Yakirevich & Resnick, 2013).

Gastric cancer is the fifth most common cancer worldwide (Jemal et al., 2010; Bertuccio et al., 2009; Forman & Sierra, 2014), and it is the second leading cause of cancer deaths (Ferlay et al., 2013); an infection with H. pylori is the strongest factor risk for its development (Helicobacter and Cancer Collaborative Group, 2001). In Colombia, the prevalence of this infection is universally high (Matta et al., 2017). However, the gastric cancer risk increases along with the altitudinal gradient (Torres et al., 2013). Thus, it is higher in the Andes region than along the Pacific coast (Kodaman et al., 2014); this phenomenon has been called the Colombian enigma (Correa & Piazuelo, 2010).

The bacterium has coevolved with its human host for a period of 60,000 years, since the first major migratory wave outside Africa (Linz et al., 2007). The bacterium has followed a similar dispersal pattern as its human host. Currently, seven populations of H. pylori have been identified: hpEurope, hpNEAfrica, hpAfrica1, hpAfrica2, hpAsia2, hpSahul and hpEastAsia (Yamaoka, 2009; Falush et al., 2003; Achtman et al., 1999; Moodley et al., 2009; Devi et al., 2007). A recent human migratory event was the colonization of the Americas 525 years ago. During this colonization, new pathogens arrived to the American continent, including new strains of H. pylori, which caused the disappearance of 80% of the native population (Bianchine & Russo, 1992; Parrish et al., 2008). It has been reported that H. pylori has followed unique evolutionary pathways in Latin-America (Gutiérrez-Escobar et al., 2017; Muñoz Ramírez et al., 2017) and that the strains followed rapid adaptive processes in different countries of the region (Thorell et al., 2017), establishing local independent lineages.

The bacterium has virulence factors that correlate with the risk of developing gastric diseases (Parsonnet et al., 1997; Nomura et al., 2002; Mahdavi et al., 2002; Yamaoka et al., 2002a; Yamaoka et al., 2002b). H. pylori attach to the gastric epithelium thorough adhesins that contribute to the initial steps of the infection (Matsuo, Kido & Yamaoka, 2017). AlpA (∼56 kDa) is an adhesin that is encoded by the locus alpAB (Alm et al., 2000), which is essential for adherence to the human gastric epithelium (Odenbreit et al., 1999). This adhesin is expressed by all clinical isolates (Odenbreit et al., 2009) and is recognized in sera from infected patients (Xue et al., 2005), and in addition, it induces the secretion of interleukin-8 (IL-8) (Lu et al., 2007). AlpA represents an emerging virulence factor of H. pylori that has been gaining attention because of its potential as a vaccine target (Sun et al., 2010).

The aim of the present study was to describe the genetic diversity and microevolution of the adhesin AlpA at the population level in the high gastric cancer risk region of Colombia. Population genetics statistics and phylogenetic methods were performed to compare H. pylori strains from Colombian clinical isolates against strains from different geographical backgrounds.

Materials and Methods

Results

A total of 142 sequences were used in this study: 86 were obtained from Colombian patients with different gastric diseases, and 34 sequences were obtained from reference strains from GenBank. The phylogenetic tree of AlpA using only the reference pool showed three clades: one clustering sequences from HspAmerind and HspAsia populations, the second clustering sequences from HpEurope and HspSouthIndia, and the last clustering sequences exclusively from HspWestAfrica. The tree does not reflect a clear separation between HspAmerind and HspAsia nor between HpEurope and HspSouthIndia, which suggests that recombination and gene conversion gave rise to its evolutionary pattern (Fig. 1).

The phylogenetic tree including Colombian isolates showed an intricate pattern of clades. Five major clades were detected: (1) cluster sequences from HspWestAfrica, HspColombia and HpEurope; (2) cluster sequences from HpEurope, HspSouthIndia and HspColombia; (3) cluster sequences from HspColombia; (4) cluster sequences from HspAsia, HspAmerind; and (5) cluster sequences from HpEurope, HspAsia and HspColombia (Fig. 2). The phylogenetic tree of the Colombian isolates showed three major clades, called Col1, Col2 and Col3; a more detailed analysis allowed us to show that the phylogenetic tree had seven subclades, called 1 to 7, which indicates that an intense evolutionary process is taking place between Colombian isolates with respect to AlpA (Fig. 3).

The analysis of the nucleotide diversity (2.5-fold) and the average number of nucleotide differences (3.7-fold) showed them to be higher in the reference pool than in the Colombian isolates. Similarly, the theta estimator showed that the reference pool was significantly more diverse than the Colombian counterparts. The number of haplotypes was 3-fold higher in the Colombian isolates than that observed in the reference pool, but the haplotype diversity was similarly higher in both populations; the total number of haplotypes was 134, with an extreme value for the haplotype diversity. Recombination events were 1.3-fold higher in the Colombian isolates than in the reference pool (Table 1).

To test the deviations of the neutral model of molecular evolution, the Tajima and Z-test were employed. The Tajima’s D test was negative and the Z-test showed significant results for neutrality, indicating that in overall average the synonymous substitution rate was similar the non-synonymous substitution rate (p = 0.005) (Table 2). However, the sliding windows analysis showed that several regions of the alpA has an ω value above 1, which suggests that episodic positive diversifying selection is also shaping the microevolutionary patterns of this gene in Colombia (Fig. 4).

The test of genetic differentiation showed that the Colombian population was well-differentiated from the reference pool, but as expected, the Nm obtained from FsT was 3.63, which indicates a moderate gene flow between the Colombian isolates (Slatkin, 1985; Slatkin, 1987) (Table 3). To assess the isolation between pairs of populations, the Colombian isolates were organized into seven groups based on the histopathological diagnosis, and then, the alpA DNA sequences were compared pairwise with the reference pool clustered according to their phylogeographic classification. Population isolation was found between the Colombian populations and all subpopulations (Table 4). Two gene conversion algorithms were applied to the population to test whether this process was involved in the evolutionary pattern of alpA. Betran’s algorithm found 54 conversion tracks between the sequences: 20.5% for the reference pool and 44.4% for the Colombian isolates, and the genconv algorithm found six tracks of gene conversion between the sequences that belong exclusively to the Colombian isolates (Fig. 5).

Functional divergence analysis among the AlpA proteins sequences based on the three major clades found in the phylogenetic tree of the Colombian isolates was performed using Gu’s type-I method. The pairwise comparisons between the Col1/Col3 clades showed five sites with a θD > 0.8 (Fig. 3B), which indicates that the protein presents sites with different evolutionary rates. The analysis of positive and negative selection covered 540 codon sites of the AlpA protein. A similar number of sites were detected by the FEL and IFEL algorithms. FEL identified that 3.3% of the sites were under positive selection and 21.4% evolve under purifying selection. IFEL showed that 2.7% of the sites evolve under positive selection, with 12.2% under purifying selection, and finally, MEME showed that 5% of the adhesin was under episodic diversifying selection (Fig. 6). The relax test showed that when the internal branches were compared against the external branches, a significant pattern of natural selection intensification was detected (K = 29.55, p = 5.48e⁻¹⁰, LR = 38.50).

Discussion

In this study, we describe the genetic diversity and microevolution of AlpA in Colombian isolates of H. pylori obtained from a high gastric cancer region composed of the cities of Bogotá, Tunja and the surrounding towns. The phylogenetic tree using the reference AlpA sequences revealed that the strains clustered according to their geographic origin; two main clades were observed: the Eastern and the Western clades. Then, within each main clade, there was no separation between the H. pylori subpopulations. This geographical segregation has been observed for other virulence factors of H. pylori (Cao et al., 2005; Maeda et al., 1998; Oleastro et al., 2009a; Van Doorn et al., 1999) and it is an indicative of the pre-colonization period. However, when the sequences of Colombian isolates are added to the analysis, the main East and Western clades fade, which indicates that the sequences share information by recombination and diversifying selection induced by the post-colonization period, when the host and bacterial populations mix together in Latin-America, and for the induction of different divergence rates between the paralogous/homologous members of different strains as a consequence of gene conversion (Santoyo & Romero, 2005).

The phylogenetic tree of Colombian isolates showed tree dominant clades entitled Col1, Col2 and Col3. The clades Col1 and Col2 have few branching events with a low number of members, but the Col3 clade showed five subclades. There was no association between the disease state and belonging to a specific clade (p = 0.245 Chi-square 2.81), which indicates that the H. pylori strains interchange DNA randomly between the Colombian strains. It is a well-established fact that the bacterium is naturally competent (Dorer et al., 2013).

The nucleotide diversity and the average number of nucleotide differences was lower in the Colombian alpA alleles that in the reference pool. However, the Colombian isolates showed a higher number of haplotypes and recombination events. The alpA gene from Colombian isolates has shown strong patterns of genetic differentiation, but it maintains the genetic flow that has produced new allelic variants of alpA in Colombia. The arrival of the HpEurope to Latin-America induced the replacement of the native HspAmerind population from urban zones (Yamaoka et al., 2002a; Yamaoka et al., 2002b) and perhaps produced a selective bottleneck. After the bottleneck, a subsequent population expansion emerged with the new recombinant H. pylori subtypes in Colombia. It has been proposed that recombination among H. pylori strains can induce the evolution of different subclones and genotypes (Blaser & Berg, 2001; Blaser & Atherton, 2004).

H. pylori has one of the highest mutation and recombination rates between bacterial species (Suerbaum & Josenhans, 2007). It has been identified that H. pylori has followed unique evolutionary pathways in Latin-America (Muñoz Ramírez et al., 2017), and the strains have followed rapid adaptive processes in different countries of the region (Gutiérrez-Escobar et al., 2017; Thorell et al., 2017). Another factor that contributes to the rapid evolution of the Colombian H. pylori alpA subtypes was the mestizo host. Perhaps the immune and inflammatory responses of this new host and its genetic heterogeneity represented by the variability and distribution of receptors to which the bacterium can adhere could be a selective factor for the divergence among Colombian H. pylori strains (Dubois et al., 1999; Oleastro et al., 2009b). In fact, the host imposes a selective pressure that induces variation within the bacterium (Thompson et al., 2004).

Gene conversion was a major process that gave rise to the allelic variation of alpA in Colombia, and the paralogous interchange of the DNA fragments close to the 3′ region of the gene was considerable higher between the Colombian isolates than in the reference pool. In total, 60 recombination events were detected by the two applied algorithms, and despite the total number of sequences used in the analysis, only Colombian alpA alleles showed statistical signals of this type of recombination. Gene conversion has been identified previously in other H. pylori adhesins, for example, homB and homA (Dorer et al., 2013; Dubois et al., 1999; Oleastro et al., 2009a; Oleastro et al., 2009b; Santoyo & Romero, 2005; Yamaoka et al., 2002a; Yamaoka et al., 2002b), sabB and omp27 (Talarico et al., 2012) and babA (Hennig, Allen & Cover, 2006), and a Latin-American type of babA has also been recently reported (Thorell et al., 2016).

The interplay between positive selection and recombination has been detected in bacterial genomes (Lefebure & Stanhope, 2007; Joseph et al., 2011; Orsi, Sun & Wiedmann, 2008). The analysis of functional divergence analysis of the protein AlpA based on the comparison of the tree clades Col1, Col2 and Col3 showed that the protein residues 23V, 65V, 190Q, 219A and 446G have a significant evolutionary site variation rate with a θD value of 0.8, which confers functional differences between the members of the clades Col1/Col3. In Colombia, positive, episodic diversifying, purifying selection and recombination of alpA has given rise to the presence of rare alleles and new haplotypes in the emerging population, which have been maintained at the protein level by natural selection.

The action of natural selection might purge diversified genes from the population by strong purifying selection (Lynch & Conery, 2000) or lead to evolutionary novelties by positive selection (Wertheim, Murrell & Smith, 2015). When the internal branches of the phylogenetic tree of Colombian alpA alleles were compared against the external ones, we determined that there was a significant natural selection intensification operating in the alpA alleles (K = 29.55, p = 5.48e⁻¹⁰, LR = 38.50), which means that positive selection is leading the microevolution at the high gastric cancer risk region of Colombia.

The Mestizo populations in the mountain zones in Colombia have an admixture of European and Amerind ancestries in a similar fashion that the bacterium mirroring the colonization process (Kodaman et al., 2014). When the Colombian H. pylori strains and human host ancestries are compared, a difference in the intensification of the disease aggressiveness is observed. Deleterious duplicated alleles of alpA were purged out of the population, but those strains with fixed alpA alleles under positive selection give advantages to the new types of strains in this region.

H. pylori is a bacterium that can display very fast local adaptive processes via mutation and recombination (Cao et al., 2015; Furuta et al., 2015). The phylogenetics, population genetics and protein evolutionary analysis suggest that alpA in Colombia has functionally divergent variants that are the result of gene conversion in a staggeringly short period of time. AlpA proteins from the Colombian population show evidence of functional divergence, positive selection and episodic positive selection at specific sites. It is possible that the polymorphism of this adhesin in Colombia reflects the phylogeography and historical generation of Mestizos in Latin-America because H. pylori is a reliable biological marker of human migratory events (Templeton, 2007).

Conclusion

The molecular evolution of virulence factors of H. pylori is currently gaining attention in the scientific community due to the new genomics and evolutionary findings around the world. Currently, Latin-American countries have emerged as evolutionary laboratories for H. pylori. To our knowledge, this study is the first study that presents statistically-supported evidence that alpA alleles from a high gastric cancer risk area from Colombia owe their variation patterns to gene conversion and purifying selection. In addition, a fast process of gene diversification followed by positive and relaxed selection has shaped three protein variants for the AlpA adhesin in Colombia.

Supplemental Information