The Helicobacter pylori methylome is acid-responsive due to regulation by the two-component system ArsRS and the type I DNA methyltransferase HsdM1 (HP0463)

ABSTRACT In addition to its role in genome protection, DNA methylation can regulate gene expression. In this study, we characterized the impact of acidity, phase variation, and the ArsRS TCS on the expression of the Type I m6A DNA methyltransferase HsdM1 (HP0463) of Helicobacter pylori 26695 and their subsequent effects on the methylome. Transcription of hsdM1 increases at least fourfold in the absence of the sensory histidine kinase ArsS, the major acid-sensing protein of H. pylori. hsdM1 exists in the phase-variable operon hsdR1-hsdM1. Phase-locking hsdR1 (HP0464), the restriction endonuclease gene, has significant impacts on the transcription of hsdM1. To determine the impacts of methyltransferase transcription patterns on the methylome, we conducted methylome sequencing on samples cultured at pH 7 or pH 5. We found differentially methylated motifs between these growth conditions and that deletions of arsS and/or hsdM1 interfere with the epigenetic acid response. Deletion of arsS leads to altered activity of HsdM1 and multiple other methyltransferases under both pH conditions indicating that the ArsRS TCS, in addition to direct effects on regulon transcription during acid acclimation, may also indirectly impact gene expression via regulation of the methylome. We determined the target motif of HsdM1 (HP0463) to be the complementary bipartite sequence pair 5′-TCAm6AVN6TGY-3′ and 3′-AGTN6GAm6ACA-5′. This complex regulation of DNA methyltransferases, and thus differential methylation patterns, may have implications for the decades-long persistent infection by H. pylori. IMPORTANCE This study expands the possibilities for complex, epigenomic regulation in Helicobacter pylori. We demonstrate that the H. pylori methylome is plastic and acid sensitive via the two-component system ArsRS and the DNA methyltransferase HsdM1. The control of a methyltransferase by ArsRS may allow for a layered response to changing acidity. Likely, an early response whereby ArsR~P affects regulon expression, including the methyltransferase hsdM1. Then, a somewhat later effect as the altered methylome, due to altered HsdM1 expression, subsequently alters the expression of other genes involved in acclimation. The intermediate methylation of certain motifs supports the hypothesis that methyltransferases play a regulatory role. Untangling this additional web of regulation could play a key role in understanding H. pylori colonization and persistence.

world, infecting about half of the world's population (2).Infection generally occurs in childhood and can persist for the host's lifetime if untreated (3).Infection can occur via oral-oral transmission, fecal-oral transmission, or can be food-borne or water-borne (4).While most cases are asymptomatic, 10%-15% of infections lead to peptic ulcer disease or gastric cancer (5).H. pylori is the leading cause of gastric adenocarcinoma, which is the second leading cause of cancer-related deaths worldwide (3,5).
As the pH of the human stomach drastically fluctuates throughout the day and the lifetime of the host, acid acclimation strategies are vital for H. pylori's long-term survival and colonization (6)(7)(8).The main acid-sensing and response system in H. pylori is the two-component system (TCS) ArsRS which exhibits widespread gene regulation (9,10).Multiple studies have investigated the global changes in gene expression in response to acid conditions and agree that ArsRS has a regulon of at least 100 genes in multiple H. pylori strains (7-9, 11, 12).
Restriction-modification (R-M) systems are best known for their role as primitive bacterial immune systems in which DNA methylation is a mechanism to label the genome as self (13).There is a growing body of research demonstrating that R-M systems have functions beyond genome protection, particularly in pathogenesis (1,(14)(15)(16)(17)(18).The traditional function of degrading foreign DNA may not provide sufficient explanation for the high level of specificity in sequence recognition, the diversity in kind and number of R-M systems in the bacterial domain, or the independent evolution of restriction endonucleases and methyltransferases in relation to each other (14).
There are three classes of restriction-modification (R-M) systems in H. pylori: Types I, II, and III (19).They differ in the subunit composition of their enzymes, motif recognition, and recombination ability (13).This study focuses on a Type I DNA methyltransferase in Helicobacter pylori strain 26695, HsdM1 (HP0463).Type I R-M systems are complexes composed of three subunits: a DNA methyltransferase (MTase), a restriction endonu clease (REase), and a specificity subunit (S) (20,21).Type I R-M systems control sequence recognition via two target recognition domains, TRD1 and TRD2, within the S subunit that together allow the protein to recognize a bipartite target sequence (22).The TRDs are flanked by repeat sequences, the length of which determines the distance between the target sequences (22).Type II R-M systems are the most abundant and well-charac terized class, and thus are used as the control in this study.Type II R-M systems are composed of separate REase and DNA MTase subunits which exhibit identical DNA binding specificity (23).Type II systems bind short, generally palindromic, motifs (13).
Within H. pylori's relatively small genome of 1.64-1.67Mb, there are over 20 restric tion-modification systems, although there is a variable, strain-specific fraction of these RM systems among isolates (19,26).This makes H. pylori an organism of interest for R-M system studies as the average number of MTase genes in a prokaryotic genome is five (27).Methylome analysis across 541 H. pylori genomes found that methylation patterns are directly subject to natural selection and linked phenotypes to specific methylation sites (28).In a comparative genomics study between H. pylori strains 26695 and J99, the two earliest sequenced isolates, nine Type II MTases were conserved across both strains without the presence of cognate restriction enzymes (26).Our study was conducted using H. pylori strain 26695 and focuses primarily on its Type I R-M systems along with two selected Type II systems.H. pylori strain 26695 possesses three Type I MTases, all m6A-specific, 16 Type II MTases of all three methylation types, and three Type III MTases, all m6A-specific (19).As these systems are over-represented in the H. pylori genome, the possibilities for complex genome regulation are significant.
Despite the highly variable nature of R-M systems among H. pylori genomes, HsdM1 (HP0463), the primary methyltransferase of interest in this study, is a widespread Type I MTase among the sequenced genomes.The BLAST search of the IMG Database for orthologs of HsdM1 using 51 genomes from geographically diverse H. pylori isolates revealed that 49 genomes possessed this MTase (29).This includes strains frequently cited in H. pylori investigations such as 26695, J99, G27, P12, SS1, and 7.13.The wide spread nature of HsdM1 is perhaps not surprising when considering the two most genetically related members of the Helicobacter genus, H. cetorum and H. acinonychis, each possesses orthologs of this Type I methyltransferase (data not shown).
DNA methylation has been hypothesized to disrupt simultaneous transcription and translation in bacteria by changing the secondary structure of DNA and potentially physically blocking the binding of the NusG/RNA polymerase and NusG/ribosome complexes (30).Methylation by Type I R-M systems has been shown to decrease transcription of operons, and evidence shows that methylation at start and stop codons can increase or decrease transcription, respectively, by altering both the thermodynamic stability and curvature of DNA, affecting DNA-protein interactions (15,31,32).Type I R-M systems are capable of altering their sequence specificity via homologous recombi nation and point mutations, as well as methylating promoter sequences of other R-M systems in a hierarchical manner (27).This further suggests that bacteria use these DNA methyltransferases as a way of finely tuning their gene expression levels, potentially allowing them to react in more nuanced manners to a changing environment (32).
The principal operon under investigation in the current study is hsdR1-hsdM1 (locus tags HP0464-0463) within H. pylori strain 26695, which consists of genes for the REase and DNA MTase subunits, respectively, of a Type I restriction-modification system (33).HsdM1 has been demonstrated to be regulated by at least three other DNA MTases in H. pylori strains 26695 and P1227.In addition, this operon is a phase variable based upon a repetitive nucleotide sequence in hsdR133.Across all H. pylori strains, restriction-modification genes are phase variable at a disproportionately high rate (19).Phase variable R-M systems have been shown to switch activity during the early infection stages (33)(34)(35).This strongly suggests that phase variation has evolved to allow H. pylori to respond to the various constantly changing environments of the stomach, likely through transcriptional regulation, which is critical to the long-term persistence characteristic of H. pylori infection (33)(34)(35).Thus, R-M systems are likely a mechanism through which the bacterium controls gene expression which may play a key role in colonization and persistence.

hsdM1 transcription is acid-responsive
The Type I DNA methyltransferase hsdM1 (HP0463) exists in the same operon as its cognate restriction endonuclease gene hsdR1 (HP0464) while hsdS1b (HP0462), the cognate specificity subunit of this Type I restriction-modification system, has its promoter (36) (Fig. 1).We demonstrated that hsdM1 transcription is significantly upregulated in the absence of the H. pylori 26695 major acid-sensing protein, the sensory histidine kinase ArsS (Fig. 2A).These data demonstrate that hsdM1 is controlled by the two-component system (TCS) ArsRS.
Given that ArsRS is the major acid-sensing and response mechanism of H. pylori, and hsdM1 mRNA is impacted by the absence of ArsS, we hypothesized that hsdM1 expres sion is acid-responsive.We quantified hsdM1 mRNA in both the H. pylori 26695 control mutant and an arsS null mutant cultured at pH 7 and pH 5. Our data indicate that hsdM1 transcription is induced at pH 5 in an ArsS-dependent manner (Fig. 2B).
The increased transcription of hsdM1 under pH 5 conditions peaks at 30 minutes but is still evident at 60 minutes (Fig. 2B).hsdM1 induction at pH 5 is not evident in the ΔarsS mutant indicating that hsdM1 acid-induced expression is ArsRS dependent (Fig. 2B).We speculate that this may be due to the inability of this mutant to mount an acid acclimation response in the absence of a functional ArsRS TCS and thus its physiology is impaired by the continued low pH exposure.
H. pylori 26695 has two other complete Type I R-M systems plus two Type I orphan specificity subunits (Table S1) and 16 Type II MTases, including orphan MTases and those within complete R-M systems.As part of our effort to examine the effects of environmental pH on the potential to methylate the H. pylori genome, and to determine if the acid response seen in hsdM1 was unique, we investigated four other MTases.We conducted qRT-PCR on the other Type I DNA MTases, hsdM2 (HP0850) and hsdM3 (HP1403), as well as two selected Type II DNA methyltransferases, M.HpyAI (HP1208) and M.HpyAII (HP1368).The results of our qRT-PCR analyses indicate that there are no significant changes in hsdM1 expression among any of these four DNA MTases in response to pH 5 conditions and that the ablation of the sensory histidine kinase ArsS has no apparent effect on these methyltransferases (Fig. S1 and S2).This indicates that the expression patterns revealed in the current study of hsdM1 are not universal among H. pylori restriction-modification systems or even among the Type I systems (Fig. S1).The expression patterns seen in M.HpyAI and M.HpyAII (Fig. S2) serve as a control to ensure that neither our in vitro acid shock itself nor the deletion of arsS are causing genomewide changes in MTase expression.Therefore, ArsRS regulation and acid sensitivity are unique to hsdM1 among Type I DNA methyltransferases, and possibly among other families of MTases, in H. pylori 26695.

hsdM1 (HP0463) expression varies due to phase variation in the upstream gene, hsdR1 (HP0464)
A major means of bacterial acclimation to changing environments is the alternative expression states of operons and their resulting proteins.This process of phase variation is commonly the result of hyper-mutable repetitive DNA sequences that undergo slipped strand mispairing to alter the length of DNA repeats and thus turn on or off the expression of the affected gene (16).A 2005 study by Salaun et al. (37) identified six phase variable restriction-modification (R-M) systems in H. pylori 26695 and seven in the conspecific strain J99.Several of these phase variable loci were variably present among the strains examined in their study.In addition, they observed variation in the phase of these R-M systems among the studied strains.Other investigators have demonstrated that the phase variable nature of these R-M systems affects the expression of other genes, some related to virulence (33,38).We hypothesized that the length of the poly-C tract located within hsdR1 (HP0464), encoding the restriction endonuclease subunit of the H. pylori 26695 Type I R-M system hsdR1-hsdM1-hsdS1b (HP0464-HP0463-HP0462) (Fig. 1), might affect the transcription of the downstream methyltransferase, hsdM1 (HP0463).Interestingly, our examination of sequenced strains of H. pylori in the Integrated Microbial Genome database found that only about 25% of sequenced strains are annotated as hsdR1 (HP0464) phase on (data not shown) (29).Therefore, we created phase-stabilized versions of hsdR1 in H. pylori 26695.This was achieved by replacing the 3rd position cytosine of each proline codon in the poly-cytosine tract encoding prolines 205-209 such that they still encoded tandem prolines but slipped strand mispairing would be much less frequent during genome replication.This created a phase-lock ON allele of hsdR1.We created an alternate version of this stabilized hsdR1 that was phase-lock OFF by adding another cytosine after the stabilized proline codons such that HsdR1 could not be expressed as a full-length protein.

Full-Length Text
Journal of Bacteriology DNA sequencing of PCR amplicons from wild-type H. pylori 26695 and both phaselocked ON and OFF hsdR1 mutants demonstrated that slipped strand mispairing and thus the generation of variant hsdR1 alleles was reduced greatly (data not shown).We subsequently used qRT-PCR to quantify hsdM1 mRNA in the hsdR1 phase-lock ON and phase-lock OFF H. pylori 26695 mutants and found an average fourfold decrease in hsdM1 mRNA when the upstream gene in the hsdR1-hsdM1 operon was made phase-lock OFF (Fig. 3).Thus, within H. pylori populations, variants exist that express, and fail to express, this R-M system.We speculate that populations may therefore possess differing degrees of genomic methylation.

Identification of consistently methylated motifs in H. pylori strain 26695
To characterize the impact of the HsdM1 methyltransferase on the epigenome, as well as the effects of acid and the ArsRS TCS, we cultured the H. pylori 26695 control mutant, isogenic ΔarsS mutant, and isogenic ΔhsdM1 mutant via serial passage in pH 7 or pH 5 liquid media, subculturing every 24 hours for 3 days.
Motifs that exhibited ≥97.0%methylation in all sequenced samples are shown in Table 1.For ease of data interpretation, these motifs are not included in the H. pylori mutant-specific methylome tables (Tables 2 to 4).As methylation of these motifs is consistently near 100% in the H. pylori 26695 control mutant, isogenic ΔarsS mutant, and isogenic ΔhsdM1 mutant grown in pH 7 and pH 5 liquid media, we propose that these methyltransferases are not regulated by acid, HsdM1, or the two-component system ArsRS.

Determination of the HsdM1 (HP0463) methylation motif
Type I restriction-modification systems recognize bipartite target sequences (22).We determined the target motif of HsdM1 to be the complementary bipartite sequence pair 5′-TCA m6A VN 6 TGY-3′ and 3′-AGTN 6 GA m6A CA-5′.This determination was made because these are the only sequences that are methylated in the H. pylori 26695 control and not in the isogenic ΔhsdM1 mutants cultured under either pH condition (Tables 2 and 3).This motif assignment has the same architecture as the target motif of HsdM2/M.HpyAXIII (HP0850), a previously characterized Type I m6A DNA methyltransferase in H. pylori 26695 can recognize and methylate its target motif on both DNA strands (39).In the control H. pylori 26695 mutant, 5′-(HH)TCA m6A VN 6 TGY-3′ is only methylated at pH 7 while 5′-(H)TCA m6A VN 6 TGY-3′ and 3′-AGTN 6 GA m6A CA-5′ are only methylated at pH 5 (Table 3).Also, the HsdM1 forward target motif, 5′-TCA m6A VN 6 TGY-3′, is not methylated in the ΔarsS mutant when grown in either pH 7 or pH 5 conditions (Table 3).These data demon strate that HsdM1 methylation is acid dependent and that functionality is altered in the absence of arsS.This generates some intriguing possibilities for the ability of HsdM1 to recognize and selectively modify unmethylated, hemi-methylated, and fully methylated DNA, and the possible role of hemi-methylation in H. pylori epigenetic regulation.The motif 5′-KASCGAA m6A CB-3 is methylated only at pH 5 in the ΔhsdM1 mutant.In addition, ΔhsdM1 is the only H. pylori mutant within which this motif is methylated.As 5′-CGAA m6A C-3′ are the nucleotides consistently called, we hypothesize that this is the core motif.This motif does not align with any currently annotated methyltransferases in    H. pylori 26695, thus likely belongs to one of the many uncharacterized methyltransfera ses in H. pylori 26695.
There are multiple methylated motifs across our samples that PacBio Sequel Single Molecule, Real-Time (SMRT) Sequencing identified as including nucleotides that are not part of the established target motifs.To preserve the integrity of the methylome sequencing data without confusing, we have put these nucleotides in parentheses to indicate that they are not part of the accepted core motif.Interestingly, these extra nucleotides change between conditions for certain motifs, indicating that the genetic environment of the target motif may have an impact on methyltransferase activity.For example, in H. pylori 26695 control mutant cells grown at pH 7, the forward strand of the HsdM1 target motif is methylated only when it possesses two leading H's (Table 3).However, when the same mutant is grown at pH 5, that target motif is methylated when it has only one leading H (Table 3).M.HpyAV (HP0054) is a Type II m6A/m5C MTase with the m6A target sequence 3′-GGA m6A AG-5′ (19,39).In our samples, M.HpyAV (HP0054) methylates 3′-GGA m6A AG(D)−5' at both pH 7 and pH 5 in ΔhsdM1, but methylates 3′-GGA m6A AG(CY)−5' at pH 5 in the control (Tables 2 and 3).
Together, the differential methylation patterns of HsdM1 (HP0463), M.HpyAV (HP0054), and M.HpyAVIII (HP1121) lead us to hypothesize that the varying nucleotides surrounding the core target motif may play a role in guiding selective methylation.
Therefore, we come to the novel conclusion that the target motif of the Type I DNA methyltransferase HP0463 (HsdM1) is the complementary bipartite sequence pair 5′-TCA m6A VN 6 TGY-3′ and 3′-AGTN 6 GA m6A CA-5′, HsdM1 activity is acid-responsive in an ArsS-dependent manner, and selective methylation may be guided by the genetic environment of the target motif.

The H. pylori methylome is acid-responsive
There are several studies characterizing both the transcription levels and protein dynamics of methyltransferases in acid conditions (9,(45)(46)(47).In addition, the H. pylori methylome in neutral conditions has been determined (39).We sought to determine the H. pylori methylome in both neutral and acid growth conditions and demonstrate that the H. pylori methylome is acid-responsive.
In the H. pylori 26695 control mutant, multiple sequences are methylated when grown at pH 5 and not when grown at pH 7 (Table 3).The forward strand of the Type I m6A DNA MTase HsdM1 (HP0463) bipartite target motif, 5′-TCA m6A VN 6 TGY-3, is methylated under both growth conditions but with different leading motifs.When cultured at pH 7, this motif is methylated with two leading Hs, denoting either an A, C, or T. When cultured at pH 5, this motif is methylated with only one leading H. Interestingly, the reverse strand of this bipartite sequence pair, 3′-AGTN 6 GA m6A CA-5, is methylated only when grown at pH 5.In addition, the target motif of M.HpyAV (HP0054), 3′-GGA m6A AG-5′, is methylated when grown at both pH 7 and pH 5 with a leading D nucleotide, denoting either an A, G, or T.However, when that leading nucleotide is a C, the motif is only methylated when grown at pH 5.As aforementioned, the M.HpyAVIII (HP1121) target motif also exhibits two states of surrounding nucleotides, both of which are only methylated when grown at pH 5.These data further support our hypothesis that the genetic context of the target motif helps guide selective methylation under certain conditions.Importantly, there are over 2,200 additional instances of methylation in the genome of the H. pylori 26695 control mutant when grown at pH 5 compared to the genome of the same mutant grown at pH 7.
These data demonstrate for the first time that the H. pylori methylome is plastic and acid-responsive and that the acid-responsiveness of the H. pylori methylome is partially dependent on HsdM1 (HP0463) activity.

Methylation of the H. pylori genome is partially dependent upon the ArsRS TCS
When the ArsS acid-sensing histidine kinase of the ArsRS two-component system is deleted (ΔarsS mutant), methylation of the forward strand of the Type I m6A DNA MTase HsdM1 (HP0463) bipartite target motif, 5′-TCA m6A VN 6 TGY-3′, is absent, while methylation of the complementary strand of this bipartite target motif, 3′-AGTN 6 GA m6A CA-5′, remains (Table 4).This supports our earlier conclusion that hsdM1 expression is ArsS-depend ent.This also indicates that HsdM1 may engage in hemi-methylation.The ability of methyltransferases to differentiate between unmethylated and hemimethylated DNA has previously been shown and could have exciting implications for the transcriptional and/or translational impacts (48).Interestingly, our qRT-PCR data demonstrated that hsdM1 expression is upregulated at least fourfold in ΔarsS, so it is surprising that HsdM1 activity appears to decline rather than increase.This suggests that post-transcriptional regulation may be occurring.
In the control mutant, SMRT sequencing suggested that M.HpyAVIII (HP1121) motifs are only methylated at pH 5 (Table 3).In ΔarsS, M.HpyAVIII (HP1121) motifs are methyla ted at both pH 5 and pH 7 with variations in the genetic environment (Table 4).Both 5′-GC m5C GC-3′ motif contexts that are methylated in the H. pylori control mutant are also methylated in the isogenic ΔarsS mutant but two additional genetic environments are methylated in the genome of the ΔarsS mutant.Of particular interest, 5′-SC m5C GC(GVNH) −3′ is unique to the ΔarsS mutant and is present 1,996 times in the genome which is over 1,000 more occurrences than any other 5′-GC m5C GC-3′ motif environment in any mutant (Tables 2 to 4).It is also the only 5′-GC m5C GC-3′ motif that is methylated in pH 7 conditions in the ΔarsS mutant (Table 4).This supports our hypothesis that the genomic environment of target motifs plays a role in selective methylation.However, as aforemen tioned, SMRT sequencing has higher rates of error when determining the sites of m5C modifications (40).Therefore, additional experiments to improve the detection of m5C modifications, such as Tet1 treatment-paired SMRT sequencing (40), will be necessary to determine if M.HpyAVIII (HP1121) is regulated by ArsRS and/or acid.
M.HpyAXIV (HP1517) is active in the H. pylori 26695 control and ΔhsdM1 isogenic mutant grown at both pH 7 and pH 5 conditions (Tables 2 and 3), suggesting that the expression of this N6 adenosine methyltransferase is independent of pH or the hierarchical influence of HsdM1.However, the M.HpyAXIV target motif is methylated in the ΔarsS mutant only in H. pylori cells grown at pH 5 (Table 4).In addition, there is a loss of methylation of 3′-GGA m6A AG-5′, the target motif of M.HpyAV (HP0054), with CY leading nucleotides when grown at pH 5 in the ΔarsS mutant compared to the control mutant (Tables 3 and 4).Interestingly, methylation of the same core motif with a leading D nucleotide is consistent between the ΔarsS and control mutants when grown at both pH 7 and pH 5 (Tables 3 and 4).This loss of methylation in the ΔarsS mutant suggests the expression of M.HpyAXIV (HP1517) and M.HpyAV (HP0054) in H. pylori 26695 may be regulated by the ArsRS TCS.

DISCUSSION
It is becoming increasingly clear that restriction-modification (R-M) systems, particularly DNA methyltransferases, play an important role in gene regulation beyond their role in genome protection.Helicobacter pylori is an important organism in which to study the different functions of methyltransferases as it possesses over 20 R-M systems, a significant over-representation for its small genome of 1.64-1.67Mb (19,26).Our current study demonstrates regulatory mechanisms and patterns of the Type I m6A DNA methyltransferase HsdM1 (HP0463) as well as the role of phase variation, acid, and the acid acclimation facilitator, the two-component system (TCS) ArsRS, in H. pylori epigenetics.
The methylated DNA motif of interest determined in the current study has compara tively low abundances and varying degrees of methylation compared to most annotated motifs.The most ubiquitous motifs in the genome, such as 5′-CA m6A TG-3′ which appears 14,780 times, and 5′-GA m6A TC-3′ which appears 10,830 times, are both methylated in all our samples to at least 98.8% methylation.This may indicate that these motifs play a more traditional genome protection role.The intermediate levels of methylation of the HsdM1 (HP0463) target motifs, between 35.7% and 44.0%methylation, challenge the idea that methylation is an on/off switch and supports the growing hypothesis that certain methyltransferases have a primarily regulatory role (32).
Existing studies of the H. pylori methylome focus only on the core target motifs of the methyltransferases when characterizing methyltransferase activity.However, we speculate that the story may be much more complex.In this study, we included all nucleotides that the PacBio Sequel I SMRT Cell 8M and PacBio SMRTlink Analysis software annotated as part of each methylated motif, even if they are not part of the annotated core motifs of previous studies.We found that the genetic environment of certain methylated motifs changed between pH 7 and pH 5 conditions and among our three H. pylori 26695 mutants.Therefore, we make the novel proposal that the genetic environment of the methyltransferase DNA target motifs may play a role in guiding selective methylation, and thus could mediate differential selective methylation in varying conditions.
It is worth noting that the PacBio SMRTlink Analysis software annotated 5′-GC m5C GC-3′ as m4C methylation but we attribute methylation of this motif to the Type II m5C DNA methyltransferase M.HpyAVIII (HP1121) as this specificity has been previously demonstra ted (19,39,44).As m5C detection requires significantly higher coverage, the PacBio SMRTLink analysis software will only label modifications as either m6A or m4C (41).Thus, SMRTLink has been proven to mistakenly annotate m5C methylation as another type, particularly in H. pylori (40,41,49).H. pylori 26695 has been shown multiple times to only have one m4C MTase, M2.HpyAII (HP1368), with the target motif 3′-CTTC m4C T-5′ (20,39,44), which was correctly identified in our data.Therefore, we feel it is appropriate to attribute methylation of the 5′-GCGC-3′ motif to the m5C methyltransferase M.HpyAVIII (HP1121) but acknowledge that additional experimentation is required to determine the possible regulatory patterns of this motif with greater confidence.
The expression patterns of hsdM1 do not appear to follow the traditional model of two-component system (TCS) regulation.In neutral pH conditions, the standard view of ArsRS TCS functioning predicts a high quantity of unphosphorylated response regulator, ArsR, and a low quantity of phosphorylated ArsR (ArsR~P) (10).The sensory histidine kinase protein ArsS adopts different conformations based on the level of acidity via varying degrees of protonation of the signal input domain, thus producing varying concentrations of ArsR~P (10).In the traditional model, response regulators are dimerized and act as transcription factors when phosphorylated (9).This model predicts that ArsR represses hsdM1 expression under neutral conditions in vitro, as evidenced by the de-repression of hsdM1 expression in our ΔarsS mutant (Fig. 2A).Therefore, we would expect hsdM1 to be repressed in acidic conditions as there is a substantially higher concentration of ArsR~P predicted due to activated ArsS.However, this simple model is not sufficient to explain our demonstration that acidic pH induces hsdM1 transcription in an ArsS-dependent manner, rather than repressing it.This leads us to speculate that non-phosphorylated ArsR may also be acting as a transcriptional regulator of hsdM1.Non-phosphorylated ArsR has been found to act as a transcriptional regulator, and ΔarsR mutants are not viable, which indicates that ArsR in the non-phosphorylated form has an essential role within H. pylori (9,50).It is tempting to speculate that ArsR and ArsR~P may bind the same DNA sites with different affinities, or how ArsR and ArsR~P bind the DNA results in different DNA conformations and thus different transcriptional outcomes.
Preliminary analysis of the genomic locations of methylations reveals the presence of differentially modified bases in certain promoter regions.The antisense promoter region of M.HpyAVIII (HP1121), annotated by Sharma et al. 2010 (36), contains a modified adenine nucleotide in H. pylori mutant ΔhsdM1 grown at pH 7 and pH 5 conditions and ΔarsS at pH 5 conditions.In addition, HP0731, a gene involved in toxin production (51), contains a methylated 5'-(H)TCA m6A VN 6 TGY-3′ motif, the target of the methyltrans ferase HsdM1 (HP0463), within its antisense promoter (36) in the H. pylori 26695 control mutant grown at both pH 5 and pH 7 conditions.Lastly, we noticed a modified guanine nucleotide within the primary promoter of the HP0052-HP0056 operon in the H. pylori 26695 control and ΔarsS mutants at both pH conditions.This guanine methylation is likely an oxidative modification, rather than due to the action of a methyltransferase (52).While 8-oxoG modifications can, in some cases, cause lesions preventing binding of DNA and RNA polymerase, we speculate a potential role HsdM1 may play in regulating transcription of an 8-oxoG repair system, as the guanine base modification was present in high enough frequency to be reported in both the control and ΔarsS samples but not in the ΔhsdM1 samples (53).These results suggest that any change in M.HpyAV (HP0054) transcription by hsdM1 is indirect, and hsdM1 may instead be acting upon a hypothetical 8-oxoG repair system.HP0054 encodes the Type II m6A/m5C MTase M.HpyAV, with the target motifs 5′-CC m5C TTC-3′ and 3′-GGA m6A AG-5′ (19,39).M.HpyAV (HP0054) is the only methyltransferase in H. pylori 26695 that can be both an m6A and m5C MTase (19,39).The presence of methylation in promoter regions supports the hypothesis that methylation plays a regulatory role.In addition, the possible regulation of the hypotheti cal 8-oxoG repair system and an m6A/m5C methyltransferase adds another layer to the web of complex, hierarchical methyltransferase regulation.
Interestingly, we noted some disagreement between our data and those of some existing studies.For example, experiments conducted in H. pylori 26695 by Banajaree & Rao, 2011 (46) and Narayanan et al. 2020 (47) both found that M.HpyAXI (HP0593) has an acid optimum at pH 5.5 in solution and can only form a functional tetramer in acidic conditions.RNA sequencing data from Wen et al. 2003 (45) also found M.HpyAXI (HP0593) to be transcriptionally upregulated 2.5-fold in acid conditions.Therefore, we would expect its target motif, 5′-GCA m6A G-3′, to have significantly higher methylation at pH 5 conditions.However, our methylome sequencing revealed an equal percentage of methylation of the M.HpyAXI (HP0593) target motif, 5′-GCA m6A G-3′, when cultured at pH 7 and pH 5 in all mutants (Table 1), thus indicating M.HpyAXI (HP0593) enzymatic activity in vivo is unaffected by acidic pH.This may be because, as the gastric pH drastically fluctuates, H. pylori has mechanisms for intracellular buffering, such as the urease gene cluster (45,54).Therefore, despite an extracellular pH of 5, the intracellular pH likely does not drop to pH 5. Thus, Banajaree and Rao's 2011 (46) and Narayanan et al. 's 2020 (47) studies of protein dynamics in vitro may not translate in vivo, at least not in non-extreme extracellular pHs.Wen et al. 2003 (44) also demonstrate transcriptional upregulation of M.HpyAIV (HP1352), M.HpyAVII (HP0478), and M.HpyAI (HP1208) in acidic conditions in H. pylori 26695.None of the target motifs of these methyltransferases exhibit a differential percentage of methylated motifs when cultured at pH 7 vs pH 5 in any of our samples.Such differences between predicted DNA methylation based on transcriptional data and actual methylation patterns, both within this study and in comparison to previous studies, support the growing hypothesis of complex, layered epigenetic regulation.There are likely regulatory mechanisms that manifest at the translational and/or protein levels.For example, hierarchical regulation between H. pylori DNA methyltransferases via methylation at translational start and stop codons has been previously demonstrated (27).This may also help explain the significant variance between studies characterizing the ArsRS regulon (2,9,11,12,45).Of particular interest, as we identified differential methylation in antisense promoters, antisense RNA has been shown to regulate R-M systems in E. coli (55).As our data indicate that the forward and reverse DNA target motifs of HsdM1 can be either fully methylated or hemimethylated in different condi tions, it is noteworthy the Dam methylase in E. coli exhibits a similar ability to fully methylate and hemimethylate target motifs (56,57).
We show, for the first time, that the H. pylori 26695 methylome is altered in prolonged acidic conditions, and that methylome plasticity is partially conferred by the two-compo nent system ArsRS and the DNA methyltransferase HsdM1 (HP0463).We propose that a two-component system may also indirectly impact gene expression by demonstrating that its differential transcriptional activation of methyltransferase genes has an impact on the epigenome, and thus likely on the transcription of other genes.We character ize the target motif of the Type I m6A DNA methyltransferase HsdM1 (HP0463) and demonstrate that both its transcription and protein activity, as shown by differential methylation patterns of its target motif, are regulated by the two-component system ArsRS in an acid-sensitive manner.In addition, we propose that the genetic environment of methyltransferase target motifs plays a role in guiding selective methylation.These methyltransferase dynamics result in an acid-sensitive, plastic epigenome in H. pylori 26695.This study expands the possibilities for H. pylori's ability to sense and respond to the dynamic environment of the human gastric epithelium which may play a key role in the decades-long persistent infection by this pathogen.

Bacterial mutants
H. pylori strain 26695 ΔrdxA, a metronidazole-resistant mutant, served as the control in all experiments and is referred to hereafter as the control mutant.The ΔrdxA mutant was used as part of a counter-selection procedure to create the ΔhsdM1 mutant described below as well as the ΔarsS mutant.The counter-selection procedure is detailed in Loh et al. 2011 (58) and the ΔarsS construction is detailed in Loh et al. 2021 (12).This controls for any potential impact of the ΔrdxA locus.This locus is unrelated to any studied functions as it encodes an oxygen-insensitive NADPH nitro-reductase and its deletion confers metronidazole resistance.
H. pylori 26695 ΔrdxA-ΔhsdM1, referred to hereafter as ΔhsdM1, is the deletion of hsdM1 (locus tag; HP0463).A 2081 bp fragment of H. pylori strain 26695, containing the coding sequence of hsdM1 (HP0463), was amplified by PCR using; 5′-CACGAAC AAACGCCTTCATAAGCACC-3′ and 5′-TAAGCGTGAAATGCTGCGGCTGCC-3′.The amplicon cloned into pGEM T-EASY (Promega).An 804 bp portion of the hsdM1 coding sequence was deleted by inverse PCR with the inclusion of a unique BamHI site in each of the following primers; 5′-GGGGGATCCGGCGAGTTTGGGGAGTCTTTTATCGG-3′ and 5′-G GGGGATCCTGCTTAGTAATAAGGGTAAGGGGGC-3′.Introduced BamHI sites are indicated in bold italics and underlined.After amplification and BamHI digestion, the CAT/rdxA fragment encoding from pMM674 (58) was cloned into the introduced BamHI site.Natural transformation of this plasmid, containing a deletion within hsdM1 along with an intact copy of H. pylori rdxA (HP0954) and the gene for chloramphenicol resistance, into H. pylori 26695 ΔrdxA was used for allelic replacement of the native hsdM1 gene.
Subsequent counter-selection by the method of Loh et al. 2011 (58) was used to introduce a markerless 804 bp deletion allele of hsdM1.The sequence was confirmed by PCR amplification and sequencing.
H. pylori 26695 hsdR1 phase-lock variants were created.HsdR1 (HP0464) is a predicted Type I restriction endonuclease.The region of the gene encoding amino acids 205-210 (TPPPPQ) possesses a 15 nt poly-cytosine tract that we hypothesized facilitates phase variation of hsdR1 via Slipped Strand Mispairing (SSM) resulting in an increased mutation rate creating sub-populations of H. pylori 26695 with hsdR1 phase-lock OFF and phase-lock ON.To limit or prevent this region from undergoing SSM, hsdM1 sequences were synthesized (GeneWiz) in which the 3rd position of each of the four proline codons (206-209) was changed to a T, A, or G (CCT 206 CCA 207 CCG 208 CCT 209 ).This preserved the proline coding capacity while greatly reducing the mutation rate.Another version of this sequence synthesized possessed an additional C placed after the modified proline codon 209 to create a phase-lock OFF version of hsdR1.Both versions of hsdR1, phaselocked ON and phase-locked OFF, were cloned into pUC-GW-Amp.To facilitate screening of allelic replacement of these mutant forms of hsdR1 into H. pylori 26695, a mutant version of each of the hsdR1 had a unique BamHI site inserted by site-directed muta genesis of a C at position 1033 of hsdR1 to a G using oligonucleotide-directed muta genesis (5′-CATCAGAGACTTTTTTAGCGGATCCAACCTAAACAAAAAGAC-3′) (Quick-Change Lightning, Agilent).The mutated site is shown in bold and italics and the generated BamHI site is underlined.A CAT-rdxA gene cassette was cloned into this BamHI site from pMM674 ( 58

RNA extraction
H. pylori samples for RNA extraction were centrifuged at 6,000 × g for 7 minutes and then resuspended in RNAzol RT (Molecular Research Center).Samples then underwent RNA extraction according to the manufacturer's suggested protocol.The RNA was quantified for concentration and purity using a Nanodrop (ThermoFisher), and the purified RNA samples were stored at −80°C.

Quantitative real-time PCR
All samples that underwent RNA extraction were used as templates for the synthesis of cDNA for quantitative real-time PCR (qRT-PCR).cDNA synthesis (iScript, Bio-Rad) was performed as specified by the manufacturer (Bio-Rad).TaqMan custom gene expression assays for the Type I MTases hsdM1 (HP0463), hsdM2 (HP0850), and hsdM3 (HP1403), and the Type II MTases M.HpyAI (HP1208) and M.HpyAII (HP1368) were synthesized by ThermoFisher.The sequences of all probes and reporters used are found in Table S2.All reporters have FAM as the fluorophore at the 5′ end and NQR as the quencher at the 3′ end.qRT-PCR was performed on the StepOne Plus System or QuantStudio3 System (Thermo Fisher).Target and normalizing genes in each sample were analyzed in technical triplicates.qRT-PCR data were analyzed by calculating Relative Quantity (RQ) values against GyrB using the 2 −ΔΔCT method (59).Due to the extreme sensitivity of qPCR, technical replicates with an RQ standard deviation over 0.25 were considered technical errors and were not included (59).The DNA Gyrase B subunit gene (gyrB) was used as the normalizing gene as previously described (60,61).

Acid shock experiments
H. pylori cells were harvested from plates and resuspended in 2.5 mL of Sulfite-Free Brucella Broth containing cholesterol (Gibco-BRL) (SFBB-chol) at pH 7.These mutants were cultured in an incubator at 37°C in a 5% CO 2 /95% ambient air environment with shaking at 150 rpm.The growth of the two overnight cultures was quantified via OD 600 .
New broth cultures were begun by subculturing each overnight culture to achieve a final OD 600 of 0.4.These cultures were incubated for 7 hours shaking in a 5% CO 2 incubator at 37°C under the conditions described before.
After 7 hours, the OD 600 of the two cultures was taken and 1 OD 600 unit (~10 9 cells) from the control and ΔarsS mutant cultures were collected as the time zero samples.The control and ΔarsS mutant cultures were each split into two equal aliquots and harvested by centrifugation at 5,000 × g, 20°C, for 10 minutes.The supernatant media was decanted, and one of the two pellets of each mutant H. pylori was resuspended in the same volume of SFBB-chol at pH 7, while the other pellet was resuspended in the same volume of media, but at pH 5, and then both were transferred to a six-well plate.The plate was incubated for 60 minutes under the same conditions described for broth cultures above with samples taken at 30 minutes and 60 minutes.To collect the H. pylori cells, 1 OD 600 unit of the cultures was harvested by centrifugation at 6,000 × g.The supernatant was decanted, cells were resuspended in 1 mL RNAzol (Sigma-Aldrich), and samples were stored at −80°C.All samples were subsequently subjected to RNA extraction and cDNA synthesis in preparation for qRT-PCR to determine the concentra tions of mRNA of interest.

Acid growth experiment
We conducted a 3-day, serial passage acid growth experiment using the H. pylori 26695 control mutant and isogenic ΔarsS and ΔhsdM1 mutants.All mutants were initially grown on Trypticase Soy Agar/5% sheep's blood plates.Overnight broth cultures and subcul tures were performed as in the acid shock experiments.After 7 hours of subculture, each mutant was split into two equal aliquots and H. pylori cells were isolated by centrifuga tion at 5,000 × g, 20°C, for 10 minutes.The supernatant media was decanted, and one of the two pellets of each H. pylori mutant was resuspended in SFBB-cholesterol-vanco mycin at pH 7, while the other pellet was resuspended in the same media at pH 5 and transferred to a six-well plate.These cultures were grown overnight under standard broth culture conditions.The next morning, each culture was subcultured by transferring 0.2 OD 600 units of each culture into 6 mL of new media to make a new overnight culture.This subculture procedure was repeated the next morning, and samples were collected the morning after that so that each mutant was growing in pH 5 or pH 7 broth for 72 hours total.Culture was harvested by centrifugation at 4,500 rpm, 20°C, for 10 minutes.The supernatant was decanted, and the pellets were stored at −20°C.The samples were then prepped for methylome sequencing, described below.

Methylome sequencing
Samples were prepared via genomic DNA extraction using the Bio-Rad genomic DNA extraction kit via the manufacturer's suggested protocol including the optional RNA elimination step.The extracted DNA was quantified for concentration and purity using a Nanodrop (Thermo Fisher).As an additional quality check, we performed agarose gel electrophoresis to ensure all extracted DNA was above 5 kB and had not been fragmented.
The samples were sequenced by Azenta Life Sciences via SMRT Sequencing on the PacBio Sequel I 8M System.The Sequel I System determines both the genomic sequenc ing and N 6 -methyladenosine and N 4 -methylcytosine methylation status genome-wide using Single Molecule, Real-Time (SMRT) sequencing.Azenta Life Sciences prepared PacBio SMRTbell libraries as per manufacturer instructions.Each high molecular weight DNA sample was evaluated for quality control via a Qubit 4.0 Fluorometer (Thermo Fisher Scientific) and Femto Pulse System (Agilent Technologies).Libraries were constructed using SMRTbell Express Template Prep Kit 3.0 (PacBio) which consists of fragmenting and repairing damaged DNA and ligating barcoded adaptors.Post-ligation, products underwent two rounds of 0.6X AMPure PB cleanup, and purity and integrity of each pool were evaluated via a Qubit 4.0 Fluorometer (Thermo Fisher Scientific) and Femto Pulse System (Agilent Technologies).The library and polymerase were bound using the Sequel Binding Kit 3.0 (PacBio) and loaded on one PacBio Sequel I SMRT Cell 8M using the Sequel Sequencing Kit 3.0 (PacBio), with a default pre-extension time.As the DNA polymerase incorporates the fluorescently labeled nucleotides, the color of the fluorescent pulses and the time between fluorescent pulses, the interpulse duration (IPD), are recorded.Each nucleotide has a distinct color and each DNA modification has a unique IPD.Therefore, both the DNA sequence and modified bases were determined.
Sequenced subread were mapped to the provided reference genome of H. pylori strain 26695 (62).Data analysis was performed in PacBio SMRTLink v6.0 and included summary statistics of the mean concordance, number of subreads, number of subread bases, alignment length mean, number of polymerase reads, and polymerase read length mean for each sequenced sample.Each sequenced sample was also given its own report which included the following: motif, modified position, modification type, % of motifs detected, # of motifs detected, # of motifs in genome, mean QV, mean coverage, partner motif, mean IPD ratio, group tag, and objective score.

Statistical analysis
For graphs comparing two samples, significant differences between the samples were determined via a one-tailed unpaired t-test.For graphs comparing more than two samples, significant differences between all samples were determined via two-way ANOVA with post-hoc Tukey test.Data were expressed as the means ± standard deviation with individual biological replicates overlaid as dots.

FIG 2 Full
FIG 2 hsdM1 transcription is acid-responsive in an ArsS-dependent manner.(A) hsdM1 mRNA levels are expressed as relative quantities in relation to the control mutant.Both the control H. pylori 26695 mutant, possessing an intact arsRS locus, and an isogenic ΔarsS mutant were grown to the mid-logarithmic stage of growth and then harvested.Statistical analysis; unpaired one-tailed t-test.(B) The expression of hsdM1 in the H. pylori 26695 control and ΔarsS mutants cultured at pH 7 or pH 5 at 30 and 60 minutes.After the growth of the control H. pylori mutant and isogenic ΔarsS mutant to the mid-logarithmic phase, equal aliquots were harvested and resuspended in pH 7 or pH 5 broth.hsdM1 mRNA levels are expressed as relative quantities in relation to the time zero sample.Statistical analysis; two-way ANOVA with post hoc Tukey test.This experiment was performed three times, yielding three distinct biological replicates, each represented by a dot.Error bars; standard deviation.*; P ≤ 0.05, **; P ≤ 0.01 ***; P ≤ 0.001.

FIG 3
FIG 3 hsdM1 expression is significantly impacted by phase variation of hsdR1.hsdM1 mRNA levels in the hsdR1 phase-lock OFF mutant are expressed as relative quantities in relation to hsdM1 mRNA in the hsdR1 phase-lock ON mutant.This experiment was performed three times, yielding three distinct biological replicates, each represented by a dot.Error bars; standard deviation.Statistical analysis via unpaired one-tailed t-test.***; P ≤ 0.001.

a
Motif assignments are made based on Krebes et al. 2014 (39).Sequencing was performed on the PacBio Sequel I SMRT Cell 8M.The methylated nucleotide is underlined.IUPAC Ambiguity Codes for Nucleotide Degeneracy; B = C, G, or T. D = A, G, or T. H = A, C, or T. K = G or T. M = A or C. N = any nucleotide.R = A or G. S = G or C. V = A, C, or G. W = A or T. Y = C or T.

a
Motif assignments are made based on Krebes et al. 2014 (39).Nucleotides in parentheses are not part of the core motif.Sequencing was performed on the PacBio Sequel I SMRT Cell 8M.The methylated nucleotide is underlined.ND; Not Detected.IUPAC Ambiguity Codes for Nucleotide Degeneracy; B = C, G, or T. D = A, G, or T. H = A, C, or T. K = G or T. M = A or C. N = any nucleotide.R = A or G. S = G or C. V = A, C, or G. W = A or T. Y = C or T.
Degeneracy; B = C, G, or T. D = A, G, or T. H = A, C, or T. K = G or T. M = A or C. N = any nucleotide.R = A or G. S = G or C. V = A, C, or G. W = A or T. Y = C or T.

a
Motif assignments are made based on Krebes et al. 2014 (39).Nucleotides in parentheses are not part of the core motif.Sequencing was performed on the PacBio Sequel I SMRT Cell 8M.The methylated nucleotide is underlined.Bolded motifs are assigned as the HsdM1 (HP0463) target motif.ND; Not Detected.IUPAC Ambiguity Codes for Nucleotide Degeneracy; B =C, G, or T. D= A, G, or T. H= A, C, or T. K= G or T. M= A or C. N= any nucleotide.R=A or G. S=G or C. V=A, C, or G. W= A or T. Y= C or T.
), a gift of Drs.Mark McClain and Timothy Cover, Vanderbilt University.The chloramphenicol and metronidazole counter-selection of Loh et al. 2011 (58) was used to place each of the hsdR1 phase-locked ON and phase-locked OFF gene cassettes into H. pylori 26695 and was confirmed by PCR and sequencing.

TABLE 1
Motifs exhibiting near 100% methylation in the methylomes of all sequenced samples a

TABLE 2
The differentially methylated motifs of the H. pylori 26695 ΔhsdM1 mutant cultured at pH 7 or pH 5 a

TABLE 3
(39)differentially methylated motifs of H. pylori 26695 control mutant cultured at pH 7 or pH 5 a Motif assignments are made based on Krebes et al. 2014(39).Nucleotides in parentheses are not part of the core motif.Sequencing was performed on the PacBio Sequel I SMRT Cell 8M.The methylated nucleotide is underlined.Bolded motifs are assigned as the HsdM1 (HP0463) target motif.ND; not detected.IUPAC Ambiguity Codes for Nucleotide a

TABLE 4
The differentially methylated motifs of the H. pylori 26695 ΔarsS mutant cultured at pH 7 or pH 5 a