Genomic differentiation within East Asian Helicobacter pylori

The East Asian region, including China, Japan and Korea, accounts for half of gastric cancer deaths. However, different areas have contrasting gastric cancer incidences and the population structure of Helicobacter pylori in this ethnically diverse region is yet unknown. We aimed to investigate genomic differences in H. pylori between these areas to identify sequence polymorphisms associated with increased cancer risk. We analysed 381  H . pylori genomes collected from different areas of the three countries using phylogenetic and population genetic tools to characterize population differentiation. The functional consequences of SNPs with a highest fixation index (Fst) between subpopulations were examined by mapping amino acid changes on 3D protein structure, solved or modelled. Overall, 329/381 genomes belonged to the previously identified hspEAsia population indicating that import of bacteria from other regions of the world has been uncommon. Seven subregional clusters were found within hspEAsia, related to subpopulations with various ethnicities, geographies and gastric cancer risks. Subpopulation-specific amino acid changes were found in multidrug exporters (hefC), transporters (frpB-4), outer membrane proteins (hopI) and several genes involved in host interaction, such as a catalase site, involved in H2O2 entrance, and a flagellin site mimicking host glycosylation. Several of the top hits, including frpB-4, hefC, alpB/hopB and hofC, have been found to be differentiated within the Americas in previous studies, indicating that a handful of genes may be key to local geographic adaptation. H. pylori within East Asia are not homogeneous but have become differentiated geographically at multiple loci that might have facilitated adaptation to local conditions and hosts. This has important implications for further evaluation of these changes in relation to the varying gastric cancer incidence between geographical areas in this region.

HPF57_0337, MscS-1, includes a MscS (mechanosensitive channel of small conductance) family member 1 . When a hypoosmotic shock raises the pressure in the cell, MscS family senses the increase of tension in the membrane by the force from the lipids and releases solutes and water and plays various roles in signal transduction. Fujian-specific H139R is predicted to be on coiled-coil helix, often seen in a periplasmic scaffold, N-distal to the MscS domain. This MscS family member might have a novel type of host interaction.

Methionine importer
Amino acids and oligopeptides are important nutrients for H. pylori and other members of the microbiome as well as for the host human body. MetQ (HP1564, MWE_1780), methionine-binding subunit of a D-methionine ABC transporter (Fig. S2D), has Yunnan-specific TG105A in the Met binding pocket. Oxidative damage of Met on proteins by host-driven oxidative stress is a serious problem for H. pylori 2 . The dependence of cancer cells on exogenous methionine is known as the Hoffman effect 3 .
FecA-1 imports iron(III) dicitrate (Fig. S3B). It determines the survival of Helicobacter pylori in the stomach 5 . Comparison of high cancer areas vs. low cancer areas revealed four clustered amino-acid changes on its upper side and one in the plug with Fst values of 0.44-0.48.

CopA (MWE_1691), a well-conserved copper(I) exporter of the P-type ATPases characterized by a P~protein intermediate, has
Okinawa-specific N392D in the actuator domain involved in structural change between the open and closed configuration. A mutation in the corresponding site in its human homolog ATP7A causes protein disappearance and leads to Menkes syndrome affecting copper metabolism (Fig. S2E) 7 .

Outer membrane proteins
HpaA (Fig. S3F) is an adhesin well studied as a virulence factor. It has Japan-specific T167A and China-specific Q45H and A51V as well as gastric-cancer associated L109 and A112 (detected in GWAS) 8 in the glycolipid-binding ends 9 . T167A is also at the interface between two identical subunits (Fig. S3F (ii)).
OmpA family are outer membrane proteins non-covalently anchored to peptidoglycan and can form 8-stranded beta-barrel in the N-terminus. They have pathogenic roles including adhesion, invasion, and intracellular survival as well as evasion of host defenses and stimulation of pro-inflammatory cytokine production 10 . Omp18 (HP1125) lipoprotein likely of this family is involved in persistent colonization by evading interferon γ signaling 11 . HP0358 with Japan-specific 151 is another protein likely of OmpA family.

RkiP, an oncoprotein candidate mimicking human RKIP
This protein (HP0218) (Fig. S2B) belongs to PEBP family functioning in lipid binding and regulation of signaling pathways. Its human member RKIP (Fig. S2B (iii)) is a tumor suppressor inhibiting MAPK pathway of signal transduction leading to stomach cancer.. In HP0218's solved structure, likely representing a cytoplasmic precursor form, the two monomers have strong interaction only around C65 (and N-terminus, see the paragraph end) from each monomer in the parallel configuration just as the human homolog RKIP. All the other structurally characterized bacterial members have anti-parallel configuration instead. Bonding of the SH pair into S-S would lead to a shape and an electric charge distribution around the ligand binding area closely resembling the human RKIP's. We hypothesized that H. pylori member has evolved a mimic of human RKIP and named them RkiP (from rkiP gene). H. pylori infection increases in RKIP phosphorylation 12 and RKIP enhances cell death in infection 13 . Expression of RkiP is altered by adhesion to human cells 14 . We hypothesize that RkiP interferes with RKIP action as a mimic, which results in increase in RKIP phosphorylation and switches MAPK pathway to oncogenesis. RkiP carries a weak signal of signal peptide (SignalP-5.0) in a region corresponding to the signal peptide of YbcL of E. coli. Japan-specific S42A, modeled by in silico mutagenesis (PyMOL), alters interaction of residue 42 with M7 on this putative signal peptide (Fig. S2B (i)  right). We hypothesize that this change may affect maturation and the above function of RkiP.

HcpX, an effector candidate
HcpX (Fig. S2A) belongs to SLR protein family, a subfamily of tetratricopeptide repeat (TPR)-containing proteins. They contain a motif of two antiparallel alpha-helices such that its tandem arrays generate a right-handed helical structure with an amphipathic channel that accommodate the complementary region of a target protein. HcpC of this family (HP1098) is secreted and interacts with several human proteins 15 . HcpX, carrying only one S-S bond connecting the helix pair, is well modeled on LpnE, a Legionella effector targeting human inositol polyphosphate 5-phosphatase, OCRL-1 16 . Fujian-specific K229T changes surface shape and surface electric charge presumably affecting its interaction with target proteins.

Jag regulating Cag secretion system
VirB11 (= Cag-alpha = HP0525) ATPase of Cag secretion system forms a hexamer-ring gate to transport CagA and other molecules to human cells 17 . Jag is its unlinked (mapping outside of CagPAI) negative regulator forming its partial lid 18 . Fujian-specific S207N is in its S8-like nucleic-acid binding domain.

DsbI
Dsb (disulfide-bond formation) proteins receive electrons generated by S-S bond formation in the cytoplasmic membrane. DsbI 19 (HP0595, MWE_0921) forms a subfamily of DsbB in epsilon-proteobacteria. H. pylori DsbI is required for colonization in mice 29 . It has Okinawaspecific I374T by the funnel of its predicted 5-balade beta-propeller (Fig. S3D) in its C-half.

Modifiers of membrane lipids
Kdo2-lipid A is an essential component of lipopolysaccharide but stimulates the innate immune response. LpxE, lipid A 1-phosphatase (HP0021, MWE_0025) (Fig. S2C) removes its 1-phosphate to camouflage it and also to confer resistance to host-derived cationic antimicrobial peptides (CAMPs) and to antibiotics polymyxin 20 . Okinawa-specific T95A is by the active site and at the top of a helix that catches the lipid.
The lipids containing cyclopropane-fatty-acid protect bacteria from acidity and antibiotics 21 . CfaS, cyclopropanefatty-acyl-phospholipid synthase (HP0416), transfers a methylene group from S-adenosylmethionine across the cis double-bonds of unsaturated fatty acyl chains for their synthesis. It has Fujian-specific E206A at the end of its methyltransferase domain (Fig. S3E).

Metabolic enzymes Vitamin B5 and B7 syntheses in Okinawa
Enzymes for synthesis of three micronutrients are altered in H. pylori of Okinawa subgroup (Fig. S4A). PanD for pantothenate (vitamin B5) synthesis ( Fig. S4A (i)) is self-processed to produce two chains, which stay together by interaction at Okinawa-specific L20I and E109G. BioD for biotin (vitamin B7) synthesis carries Okinawa-specific N97D. A GWAS of host species specificity of Campylobacter jejuni revealed that vitamin B5 synthesis genes were conserved in cattle isolates than in chicken isolates likely because this vitamin is present in the grass diet of the cattle 22 . Likewise, alteration of the enzymes for synthesis of the three micronutrients (for Q-base, see below) in Okinawa might be related to the unique human diet in Okinawa islands 23 .

Q-base synthesis in Okinawa
Tgt (HPF57_0334) (Fig. S4A (iii)), queuine tRNA-ribosyltransferase, replaces G34 in the anticodon of several tRNAs by a precursor of quine (Q-base), which modifies codon recognition and affects efficiency and accuracy of translation 24 . Hosts and members of their associated microbiota compete for the salvage of Q-base precursor micronutrients 25 . Tgt contributes to virulence of Shigella flexneri and E. coli 26 . Okinawa-specific P250L is by the residues recognizing U33. This residue is tightly attaching to of the three active site residues.
The differentiation in synthesis of Q-base may affect carcinogenesis. Decreased Q-modification of tRNA has been demonstrated for a large number of neoplastic tissues. Q-base modification of tRNA alters translational fidelity. The proteome of cancer cells is not fully encoded by their transcriptome, although the contribution of (mis)translation to such diversity remains to be elucidated 27 . The possibility that Q-base-mediated interaction between H. pylori and human cells promotes carcinogenesis and explains low cancer incidence in Okinawa is testable at several levels.

Fur transiption factor
Fe and Ni import and homeostasis as well as redox homeostasis and acid response are regulated by Fur (Ferric uptake regulator), a transcription factor required for colonization (Fig. S4D). Its Japan-specific P114S is close to Zn at S3 important in DNA binding 28 .