Comparative genomic analysis of Helicobacter pylori isolates from gastric cancer and gastritis in China

Abstract

Background

This study aimed to explore and compare the genomic characteristics and pathogenicity of Helicobacter pylori (H. pylori) strains derived from the gastric cancer (GC) and gastritis in the Chinese population.

Methods

We performed whole genome sequencing on 12 H. pylori strains obtained from GC and gastritis patients in China. Additionally, we retrieved sequencing data for 20 H. pylori strains from various regions worldwide from public databases to serve as reference genomes. An evolutionary tree was constructed based on comparative genomics, and we analyzed the differences in virulence factors (VFs) and gene functions.

Results

In the GC strains, we identified 1,544 to 1,640 coding genes, with a total length ranging from 1,549,790 to 1,605,249 bp. In the gastritis strains, we found 1,552 to 1,668 coding genes, with a total length spanning from 1,552,426 to 1,665,981 bp. The average length of coding genes was approximately 1,594 (90.91%) for GC strains and 1,589 (90.81%) for gastritis strains. We observed a high degree of consistency in the VFs predicted for both cohorts; however, there was a significant difference in their cagA status. Clustering analysis showed significant core single nucleotide polymorphisms (SNPs) differences between GC and gastritis strains, but no major differences in homologous proteins or gene islands. Subsequent pan-genomic and Average Nucleotide Identity (ANI) analyses indicated high homology among GC, gastritis, and other reference H. pylori strains. Furthermore, gene function annotation results showed substantial similarity in gene functions between the H. pylori strains from GC and gastritis patients, with specific functions primarily concentrated in metabolic processes, transcription, and DNA repair.

Conclusions

H. pylori strains derived from GC and gastritis patients exhibit differences in virulence factors and SNPs, yet they demonstrate high genomic homology across other levels in the Chinese population.

Background

Helicobacter pylori (H. pylori) is a highly successful gram-negative bacterium that chronically infects the human gastric mucosa, affecting an estimated 4.3 billion people worldwide in 2022 [1]. H. pylori infection is linked to gastritis and gastric cancer (GC), with approximately 1–3% of infected individuals eventually developing GC [2, 3]. The prevalence of H. pylori infection is notably higher in developing countries compared to developed nations [4]. Currently, China has one of the highest infection rates worldwide, exceeding 50% [5]. Meanwhile, according to GLOBOCAN 2022, the age-standardized rate (ASR) of GC prevalence in Eastern Asia is alarmingly high at 16.1% [6]. However, there is minimal knowledge about the origin of H. pylori genomes and pathogenicity differences in patients with gastritis and GC in this region, which limits the application of molecular diagnostics, such as screening for early GC, prevention of precancerous lesions, and vaccine drug development.

Previously, several studies applied the first-generation sequencing method to initially explore and compare the genetic differences of H. pylori from GC and benign gastric diseases [7,8,9]. In 2009, by comparing H. pylori from GC and benign gastric ulcers, it has been first found that highly differentiated alleles and strain specific genes may be useful biomarkers for analyzing the geographic distribution of H. pylori and identifying strains capable of inducing malignant or precancerous gastric lesions [10]. Following this, subsequent research has increasingly focused on specific genes and virulence factors (VFs), such as the vacA and cagA genes, which may serve as biomarkers for GC risk [11, 12]. Despite the utilization of whole-genome sequencing (WGS) in some studies, there is still a significant gap in research regarding the specific genomic differences between H. pylori strains in GC and gastritis patients, particularly regarding the genetic variations of H. pylori strains within the Chinese population. In this study, we isolated and cultured H. pylori strains from patients with GC and gastritis (GC: n = 6, gastritis: n = 6), and employed whole-genome sequencing methods to thoroughly explore and compare the genomic and pathogenic differences between these two cohorts in the Chinese population.

Methods

H. Pylori isolates

Twelve strains of H. pylori were isolated from tissue samples of patients undergoing gastric surgery (open surgery and endoscopic resection) at Fudan University Shanghai Cancer Center (FUSCC) (Supplemental Table S1). Based on the diagnosis, the isolates were categorized into two subgroups: GC (n = 6) and gastritis (n = 6). With the consent of all patients, at least 2 pieces of tissue were collected from the margin of either benign lesions or cancerous tissue for bacterial culture and pathological examination. Two experienced clinical pathologists performed pathological assessments of the gastric tissue specimens. This study was approved by the FUSCC review board in accordance with Chinese bioethical regulations, and written informed consent was obtained from all participants (Ethics Approval Number: FUSCC-D-2022-144). In addition, genome sequences of 20 H. pylori reference strains were downloaded from GenBank, with detailed information provided in Supplemental Table S2.

Genomic DNA extraction and sequencing

The DNA of H. pylori isolates were extracted using a bacterial genomic DNA extraction kit (Beyotime Biotechnology Co., Ltd., China) according to the manufacturer’s instructions [13]. The DNA content and purity were assessed using a Qubit fluorometer and a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Carlsbad, USA). WGS of the H. pylori strains was performed at Shanghai PersonalBio Technology Company (SPTC, Shanghai) using the Illumina NovaSeq 6000 platform.

Data preparation and genome assembly

Raw sequencing data were filtered using fastp (version 0.20.0, https://github.com/OpenGene/fastp) to remove sequencing junctions and primers from the reads. Additionally, reads with a constant quality value ≤ 5 bases and those containing more than 5% N bases were discarded, along with any repeated contaminants, resulting in the acquisition of clean reads [14]. Subsequently, the clean data were assembled using the default parameters of Unicycler (v0.4.8), leading to the generation of the genome sequence [15].

Genomic structural and functional annotation

Structural annotation of the genome sequence was performed using prokka(v 1.12) software. Following the generation of the gene set for the sequenced strains, the genes were compared and annotated across multiple databases to determine their functions and related descriptions. The amino acid sequences of the genes from each strain were matched with entries in each database to obtain corresponding functional annotation information. For general functional analysis, three key databases were selected: the Kyoto Encyclopedia of Genes and Genomes (KEGG), Clusters of Orthologous Groups (COG), and Gene Ontology (GO). Additionally, the Pathogen Virulence Factor Database (VFDB) was utilized to assess pathogenicity.

Comparative genomic analysis

OrthoFinder (v2.3.5) software was used to compare the predicted protein sequences of 32 strains of H. pylori and identify homologous gene clusters.

Then, the basic situation of core and non-core genes was analyzed, and the differences within species were studied from the point of view of specific gene sequences [16]. The single-copy homologous genes identified by OrthoFinder were aligned using MAFFT (v 7.427). To identify core SNPs, Snippy (v4.6.0) and Gubbins (v2.4.1) were utilized, and a phylogenetic tree was constructed using FastTree (v2.1.10) and iTOL. Average Nucleotide Identity (ANI) is an index that compares the relationship between two genomes at the nucleotide level, calculating their similarity or evolutionary distance for species classification and kinship comparison. ANI analysis was conducted on the 32 sorted strains in pairwise comparisons using FastANI (v1.31), and the results were organized into a matrix for visualization. Pan-genome is a general term for all genes of a species, reflecting all genetic information at the gene level of the species [17].

Statistical analyses

Statistical analyses were performed using SPSS software version 19.0. Differences between the groups were compared using Chi-square or Pearson Chi-square tests, Fisher’s exact method, or Kruskal-Wallis analysis according to the data type and the number of comparisons. Statistical significance were considered clinically significant at a p value < 0.05. All P-values were two-sided. When the results of Pearson Chi-square tests showed a statistical difference, the intra-group pairwise comparison was adjusted using the Bonferroni formula.

Results

Clinical characteristics of patient information and strains isolation

Between 1 March 2021 and 31 October 2022, thirty-nine GC and gastritis patients were hospitalized in our department for surgery and twenty patients tested with H. pylori positive. Next, H. pylori isolate from each patient (from ten single colony isolates stored for each patient) was selected and sub-cultured. Finally, twelve H. pylori isolates were chosen through a three-step randomization procedure for WGS analysis (Fig. 1). Of these patients, six underwent gastrectomy for GC and six underwent Endoscopic submucosal dissection (ESD) for gastritis, respectively. The average age of the patients was 46.2 years, with a range of 37 to 69 years. 75% (8/12) of the patients were male (Supplemental Table S1).

Fig. 1

H. pylori isolation, culture in this study and selection of strains for the present study. WGS, Whole Genome Sequencing

The whole genome of H. pylori strains were sequenced and analyzed using Illumina NovaSeq sequencing platforms in this study. Table 1 presents the general characteristics for each of the 12 genome sequences. The number of contigs ranged from 24 to 43, with a high (310×-1040×) genome coverage. The mean genome size was 1.57 Mb, and the average G + C content was 38.8%. Each genome contains between 1544 and 1688 annotated average coding genes sequence (CDS), with approximately 91% of the genome allocated to coding regions.

Table 1 Genome statistics of the whole genome sequences of the 12 H. pylori isolates

Circle diagram of the chromosome of H. Pylori isolates strain

The genome sequence of H. pylori in GC strains consisted of a circular chromosome with an average total length of 1,579,639 bp and an average G + C content of 38.8% (Table 1; Fig. 2A). In the six GC stains of H. pylori, between 1544 and 1640 coding genes were predicted, with a total length ranging from 1,549,790 to 1,605,249 bp, accounting for 90.63–91.16% of the total genome. In the gastritis stains of H. pylori (Fig. 2B), the 1552–1668 coding genes, with a total length of 1,552,426-1,665,981 bp, were predicted in the genome, which accounted for 90.14-91.19% of the total genome. In addition, the average length of coding genes in GC and gastritis strain, was approximately 1594 (90.91%) and 1589 genes (90.81%), respectively. Based on the VFDB (Virulence Factors of Pathogenic Bacteria) database, these virulence genes were divided into 7 VF (virulence factor) classes including acid resistance, adherence, immune evasion, immune modulator, motility and so forth. In detail, each VF class contained several kinds of VFs and the related genes as shown in Table 2. We compared the VFs of H. pylori derived from GC and gastritis, and found that the VFs predicted by the two cohort strains had a high consistency, but there were also obvious differences. For example, cagA, a well-known virulence factor related gene, was found in GC strains compared to those derived from gastritis.

Fig. 2

Circle diagram of the chromosome ofH. pyloriisolates strain in GC and gastritis patients. (A) Circle diagram of the chromosome of H. pylori isolates strain, the plot displays the genomes of GC cohort as circular chromosomes of 1,539,122 bp. (B) Circle diagram of the chromosome of H. pylori isolates strain, the plot displays the genomes of gastritis cohort as circular chromosomes of 1,591,513 bp. Circle (from outside to inside): circle 1 (virulence factor prediction); circle 2–7 (size of the genome in GC and gastritis cohort); circle 8 (size of the reference genomes); circle 9 (GC skew); circle 10 (GC content). GC, Gastric Cancer. GI, Gastritis. VF, Virulence Factor

Table 2 Comparison of virulence factor prediction of H. Pylori from GC and gastritis

Phylogenetic tree of the 12 H. pylori isolates and the reference genomes

The genomes of our H. pylori isolates were analyzed for phylogenetic relationships using core genome single nucleotide polymorphisms (SNPs) in conjunction with 20 public reference genomes (Supplemental Table S1). The core genome analysis revealed four distinct populations based on geographic origin. Specifically, 30.0% (6/20) of the isolates were from Asia, 30.0% (6/20) from America, 30.0% (6/20) from Europe, and 10.0% (2/20) from Australia.

The phylogenomics of our H. pylori strains were analyzed following the flow diagram outlined in Supplementary Figure S1. Pangenome analysis using OrthoFinder identified a total of 2,492 genes across 32 H. pylori strains, of which 1,054 genes (42.4%) were conserved and present in all strains (100%) (Supplementary Table S3). We used snippy to identify SNPs of all species using CP003904.1_Helicobacter_pylori_26695 (1,667,892 bp) as a reference sequence. A total of 206,091 core SNPs were identified. Gubbins were then used to reassemble 188,039 SNPs, and Fasttree was subsequently used to construct core SNP evolutionary tree. Additionally, ModelFinder was applied to select the best-fit substitution model based on the grouping relationships and cross-validation errors, resulting in the phylogenetic tree depicted in Supplementary Figure S2. The phylogenetic analysis of 20 strains, as illustrated in Fig. 3 and Supplementary Figure S3, revealed four distinct H. pylori populations: HP-Asia (Japan-1, Japan-2, India-1, India-2, Bangladesh-1, and Bangladesh-2), HP-America (United states-1, United states-2, Mexico-1, Mexico-2, Brazil-1, and Brazil-2), HP-Europe (France-1, France-2, Belgium-1, Belgium-2, Spain-1, and Spain-2), and HP-Oceania (Australia-1, and Australia-2). According to the clustering with reference strains, our strains (GC and gastritis strains) were assigned to HP-Asia population, especially with high homology with Japanese strains (Fig. 3A). Furthermore, we also mapped the evolutionary tree of strains based on homologous proteins, corroborating the high similarity of our strains with those from Asia, especially Japanese strains (Fig. 3B). Notably, the clustering results indicated significant differences in core SNPs between the GC and gastritis strains, while no substantial differences were observed in the homologous proteins between the two cohorts.

Fig. 3

Phylogenetic tree of the 12H. pylori isolates and the reference genomes. (A) Core SNP based phylogenetic tree of the 12 H. pylori isolates and the reference genomes. (B) Homologous protein based phylogenetic tree of the 12 H. pylori isolates and the reference genomes. GC, Gastric Cancer. GI, Gastritis. SNP, Single Nucleotide Polymorphism

Synteny in genome organization of the 12 H. pylori isolates and the reference strains

To illustrate the utility of IslandCompare, we conducted an analysis focused on the prediction, comparison, and exploratory visualization of gene islands associated with GC, gastritis, and the reference of H. pylori strains. The gene-island prediction visualization features an interactive linear display covering each genome (Fig. 4). Gene islands are uniformly colored according to their sequence clusters, with the highlighted colors highlighting similar gene islands across all genomes. Through the results of gene island prediction, the location of each gene island, the relationship between each other and the possible evolution sequence of all strains in the cluster were visualized. The prediction results of H. pylori strains for GC and gastritis based on gene islands showed that the locations of gene islands for 91.7% (11/12) strains were concentrated in 1-1.4 Mb and were mainly divided into two distinct segments. In the upper segment, four strains (GC-05, GC-06, gastritis-02, and gastritis-05) cluster closely with gastritis-03 and India-1. Conversely, in the lower segment, five strains (GC-01, GC-02, gastritis-01, gastritis-04, and gastritis-06) are grouped with other strains of Asian origin, including Japan-2, India-2, and Bangladesh-1, as well as with GC-03 and GC-04. This clustering highlights the evolutionary relationships and shared genetic characteristics among the strains.

Fig. 4

Synteny in genome organization of the 12 H. pylori isolates and the reference strains. A phylogeny (left) indicates the relationship between the isolates in the analysis, with comparative visualization and zoom- in functionality available. GC, Gastric Cancer. GI, Gastritis

Genome homology and difference of H. Pylori in GC and gastritis patients

We used FastANI software to perform pairwise ANI analysis on 32 sorted H. pylori strains, and the results were visualized accordingly. In Fig. 5A, the ANI analysis indicates that the GC and gastritis strains examined in our study exhibit a high degree of homology with strains of Asian origin, including Japan-2, India-1, India-2, Bangladesh-1, and Bangladesh-2. Our analysis identified that the H. pylori strains associated with GC and gastritis share a total of 1,103 genes. Additionally, we found 160 genes that are specific to GC strains and 144 genes that are specific to gastritis strains (Fig. 5B).

Fig. 5

Genome homology and difference ofH. pylori in GC and gastritis patients. (A) Genome-wide ANI analysis of 12 H. pylori isolates and the reference strains. (B) Homologous and differential genes of H. pylori in patients with GC and gastritis. (C) GO analysis of H. pylori specific gene in GC patients. (D) KEGG analysis of H. pylori specific gene in GC patients. ANI, Average Nucleotide Homology. GO, Gene Ontology. KEGG, Kyoto Encyclopedia of Genes and Genomes. GC, Gastric Cancer. GI, Gastritis

The pan-genomic analysis of GC, gastritis, and reference strains revealed a substantial number of shared genetic components, with a total of 1,215 core genes identified across all strains. In contrast, the GC and gastritis strains exhibited a limited number of specific genes, ranging from 2 to 8 (Supplementary Figure S4A). GO functional annotation found that the specific genes associated with GC and gastritis were primarily concentrated in categories such as cellular anatomical entity, binding, cellular process, and metabolic process (Fig. 4C and Supplementary Figure S4B). Additionally, COG analysis found that the specific gene functions of GC and gastritis strains were mainly concentrated in metabolism, translation, biogenesis, replication, recombination, and repair (Supplementary Figure S4C-D). Furthermore, presented in Fig. 4D, indicated that the top four functions of specific genes of GC strains were nucleotide metabolism, amino acid metabolism, metabolism of cofactors and vitamins. For gastritis strains, the leading functions of their specific genes included the metabolism of cofactors and vitamins, translation, amino acid metabolism, and replication and repair (Supplementary Figure S4E).

Discussion

H. pylori exhibits a complex and long-term coexistence with humans, playing a significant role in the development of gastritis and GC [18, 19]. Researches have focused on whether the pathogenicity of H. pylori varies by population, source, or genetic specificity, and whether specific highly pathogenic strains exist [20, 21]. Therefore, identifying key factors contributing to strain pathogenicity and understanding region-specific highly pathogenic strains is essential. Here, we sequenced 12 H. pylori isolates of derived from GC or gastritis patients and performed whole genome–based comparative analysis to investigate the genetic structure and differences between GC and gastritis strains. Based on this, further in-depth research will continue, and it is hopeful that humanity will fully master the molecular biological characteristics of GC-causing strains, enabling targeted prevention and early diagnosis of GC, thereby revolutionizing strategies for the prevention and treatment of GC.

East Asia has long been a high-incidence region for gastritis and GC, with a strong link between H. pylori and their development [22, 23]. In this study, 12 H. pylori strains from gastritis and GC patients in China were selected as research objects, and it is representative to explore the pathogenicity and genetic differences of the two cohorts. Meanwhile, we included representative strains from other regions worldwide sourced from public databases as references [18]. Consequently, the findings of our study will provide insights into the homology and difference between the strains of gastritis and GC, and with other reference strains to a certain extent.

In the present study, we used next-generation sequencing technology to sequence the whole genome of H. pylori strains that were isolated from gastritis and GC patients in Shanghai. The genome sequence of GC strain H. pylori consists of a ring chromosome with an average total length of 1,579,639 bp and contains approximately 1,594 CDS. In contrast, the gastritis strains exhibited a smaller average genome size of 1,569,508 bp and an average of 1,589 CDS, indicating lower genomic complexity compared to the GC group. Previous studies have reported that the genome size of H. pylori typically ranges from 1.5 to 1.9 Mb, with around 1,600 coding genes [24, 25]. Additionally, VFs of H. pylori from GC and gastritis, although with high consistency. However, the presence of the cagA gene sequence in GC strains, as compared to gastritis strains, may indicate a heightened level of virulence and aggressiveness in GC strains. While cagA-positive H. pylori strains are associated with various gastrointestinal conditions, including acute gastritis, peptic ulcer disease, and gastric cancer, our findings suggest that the presence of cagA in GC strains may reflect a specific propensity for increased virulence in the context of GC [26]. Globally, cagA-positive strains account for approximately 60% of H. pylori infections in individuals [27]. Phosphorylated CagA interacts with SHP2, Csk, Crk junction protein, and other proteins to activate the ERK/MAPK/JNK/PI3K/JAK-STAT signaling pathway, leading to abnormal expression of epithelial genes and inducing morphological changes in the “hummingbird phenotype” [28, 29].

In this study, the results of the SNP evolutionary tree revealed that the 12 H. pylori strains from patients with GC and gastritis in China exhibited variability in their lineages. The overlap between GC and gastritis H. pylori lineages sequenced in this study aligns with the co-evolution of H. pylori lineages observed in Japan, particularly among strains from gastritis patients. In addition, the phylogenetic tree based on homologous proteins, along with the visualization of gene island predictions, did not differentiate between H. pylori strains from GC and gastritis. The observation of two distinct lineages has been documented in previous studies. Previous genome-wide association studies of strains of hp-East Asia by Japanese scholars have shown that differences in SNPs between strains of GC and duodenal ulcer can be detected, and potential pathogenic mechanisms such as charge changes in ligand-binding pockets, changes in subunit interactions, and pattern switching DNA methylation have been proposed [30]. Yamaoka et al. identified virulence genes in the genomes of strains CHC155 (GC) and VN1291 (duodenal ulcer) as key risk factors for H. pylori pathogenicity [31]. Another study comparing H. pylori infections related to GC and duodenal ulcers suggested that vacA genotype status could help identify patients at high risk for developing GC [11]. The comparison results of strains VF described above in this study also found that the cagA status of H. pylori strains in GC and gastritis was different, which may be an important genomic feature of the two cohort strains. Understanding the implications of cagA status is crucial for early diagnosis and treatment of GC. The presence of specific cagA variants has been associated with increased virulence and a higher risk of progression from gastritis to GC [28]. By identifying patients with high-risk cagA variants, clinicians can implement targeted surveillance and preventive strategies, allowing for earlier intervention and potentially improving GC patient outcomes.

Subsequently, the results of pan-genomic and ANI analyses suggested that GC, gastritis and other reference H. pylori strains exhibited high levels of homology. This finding aligns with previous research, which has shown that H. pylori lineages are influenced by geographical regions and can be categorized into seven distinct lineages: HP-Africa1, HP-Africa2, HP-Sahul, HP-Europe, HP-Asia2, HP-Amerind and HP-East Asia [18]. However, it is anticipated that these coexisting lineages will converge over time due to the genomic plasticity of the strains [19]. In addition to genome homology, the gene function annotation results of COG, GO and KEGG further suggested that the H. pylori strains of GC and gastritis also had high similarity in gene function, and their specific gene functions primarily focused on processes related to metabolism, transcription, and repair. These findings are consistent with previous gene annotation studies conducted on strains from other regions [13, 32].

There are certain limitations to our study. First of all, due to the limited sample size in the study design and the regional nature of the samples, the obtained research results need to be further confirmed by large sample size and multi-center studies. Secondly, the reference genomes in the study only selected a few representative strains from each continent, which may also have a certain selection bias. Additionally, the selection of strains from GC and chronic atrophic gastritis patients presents another potential limitation. Given that chronic atrophic gastritis can be a precursor to gastric cancer, this choice may impact the accuracy of our findings. Indeed, further study including strains from patients with peptic ulcers, which represent another significant outcome of H. pylori infection, might provide a more comprehensive understanding of the pathogenicity of H. pylori across different clinical conditions. Lastly, the differences between VF and SNP in strains of GC and gastritis have not been further explored, nor have relevant basic experiments been designed for verification, and more detailed analyses are needed in future studies to address these gaps.

Conclusion

In conclusion, GC and gastritis patient-derived H. pylori have some differences in VF and SNP, they also demonstrate a high degree of homology at other genomic levels within the Chinese population.