Isolation and characterization of a HpyC1I restriction-modification system in Helicobacter pylori.

Using transposon shuttle mutagenesis, we identified six Helicobacter pylori mutants from the NTUH-C1 strain that exhibited decreased adherence and cell elongation. Inverse polymerase chain reaction and DNA sequencing revealed that the same locus was interrupted in these six mutants. Nucleotide and amino acid sequences showed no homologies with H. pylori 26695 and J99 strains. This novel open reading frame contained 1617 base pairs. The amino acid sequence shared 24% identity with a putative nicking enzyme in Bacillus halodurans and 23 and 20% identity with type IIS restriction endonucleases PleI and MlyI, respectively. The purified protein, HpyC1I, showed endonuclease activity with the recognition and cleavage site 5'-CCATC(4/5)-3'. Two open reading frames were located upstream of the gene encoding HpyC1I. Together, HpyC1I and these two putative methyltransferases (M1.HpyC1I and M2.HpyC1I) function as a restriction-modification (R-M) system. The HpyC1I R-M genes were found in 9 of the 15 H. pylori strains tested. When compared with the full genome, significantly lower G + C content of HpyC1I R-M genes implied that these genes might have been acquired by horizontal gene transfer. Plasmid DNA transformation efficiencies and chromosomal DNA digestion assays demonstrated protection from HpyC1I digestion by the R-M system. In conclusion, we have identified a novel R-M system present in approximately 60% of H. pylori strains. Disruption of this R-M system results in cell elongation and susceptibility to HpyC1I digestion.

by protecting against mucosal shedding into the gastric lumen. The most well defined H. pylori adhesin-receptor interaction to date is that between the Lewis b (Le b ) 1 blood group antigenbinding adhesin, BabA, and the H, Le b , and related ABO antigens (6,7). Further, it has been suggested that the interaction between H. pylori adhesin, SabA, and cellular receptor, sialylated Lewis x (sLe x ), promoted persistent infection (8). However, Le b antigens are only abundant among individuals with the O blood type. Seroepidemiology studies revealed that infection rates are the same in patients with different blood types (9), and some clinical H. pylori strains did not bind either Le b or sLe x antigens (8). Therefore, we used an H. pylori mutant library to identify genes involved in cell adhesion. After initial screening, we isolated mutants with decreased adherence. These mutants appeared elongated on Gram stain when compared with wild type controls. However, further characterization revealed that the interrupted operon was a novel restriction-modification (R-M) system in H. pylori.
R-M systems in bacteria protect against invasion of foreign DNA (10). The restriction endonuclease recognizes a specific sequence, and the cognate methyltransferase modifies the same sequence to differentiate self-DNA from foreign DNA (10). Thousands of restriction enzymes have been purified and characterized. Restriction enzymes are traditionally divided into three types according to subunit composition, cleavage site, sequence specificity, and cofactor requirements (10). Type II R-M systems, consisting of a restriction endonuclease and a paired methylase, are the most well known and have great practical value (10).
Based on sequence similarities, there are more than 20 putative R-M systems in H. pylori 26695 strain. A previous study showed 14 Type II R-M systems with biochemical activity in H. pylori 26695 strain (11). Comparing the complete sequences of 26695 and J99 strains, the R-M systems of these two strains are diverse. The difference of R-M systems results in the barrier of interstrain plasmid DNA transfer (12) and chromosomal DNA transformation (13), but the biological significance of so many R-M systems in H. pylori is still unclear.

EXPERIMENTAL PROCEDURES
Bacteria Strains and Culture Condition-Clinical isolates were obtained at National Taiwan University Hospital (NTUH) as previously described (14). H. pylori strains were grown on Columbia blood agar plates containing 5% sheep blood and chloramphenicol (4 g/ml) or kanamycin (10 g/ml) and incubated for 2-3 days in microaerophilic conditions (5% O 2 , 10% CO 2 , 85% N 2 ) at 37°C. For bacterial growth curve examination, H. pylori were grown in the Brucella broth containing 5% fetal calf serum.
Cell Adherence Assay-SC-M1, a cell line established from primary * This work was supported by the National Science Council, Taiwan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the DDBJ/GenBank TM (15), was used in this study. This cell line was proved to be Le b -negative and sLe x -positive by monoclonal antibodies against Le b (Seikagaku, Tokyo, Japan) and sLe x (Chemicon, Temecula, CA), respectively. The cells were grown in a 24-well culture plate (RPMI 1640 (Invitrogen) medium, supplemented with 10% fetal calf serum and 5% CO 2 , 37°C) and infected with H. pylori (multiplicity of infection ϭ 1:100). After a 30-min co-cultivation at 37°C, nonadherent bacteria were removed by washing three times with phosphatebuffered saline buffer. SC-M1 cells with adherent H. pylori were trypsinized, serially diluted in normal saline, and spread on the Columbia blood agar plates. Recovered adherent bacterial colonies were counted. Wild type H. pylori NTUH-C1 strain served as a positive control, and each mutant strain was compared with the wild type strain.
Observation of Bacterial Morphology-The morphologies of H. pylori wild type and mutant strains were observed under a light microscope after Gram stain and captured by CoolSnap-pro software (Media Cybernetics, Silver Spring, MD). More than 10 fields were examined on each slide, and the lengths of bacteria were measured in 30 bacteria of 5-10 different fields by CoolSnap-pro software (Media Cybernetics).
Inverse PCR and DNA Sequencing-To identify genetic loci interrupted by the transposon, genomic DNA of mutant strains were extracted and subjected to inverse PCR and DNA sequencing as previously described (16).
Complementation of hpyC1IR-The upstream 198 base pair and hpyC1IR coding sequence was cloned into pGEM-T easy (Promega). A cat gene was ligated into the SalI site of pGEM-T easy/hpyC1IR (nucleotides Ϫ198 to 1617). The plasmid was transformed into a hpyC1IR mutant strain to generate the hpyC1IR complementation strain by chromosomal integration. The gene alignment of the hpyC1IR complementation strain was confirmed by PCR using different combinations of primers.
Constructions of hpyC1IM1 and hpyC1IM2 Mutants-The gene encoding M1.HpyC1I was amplified from genomic DNA of wild type NTUH-C1 strain by PCR and then cloned into pGEM-T easy (Promega). MA-26R (5Ј-CGCTTTAAAATCAGCGTCTTGCC-3Ј) and MA-27F (5Ј-C-TTGAAGAACTGAAAGAATAC-3Ј) primers were used for perform inverse PCR with Pfu polymerase and then phosphorylated by a polynucleotide kinase. A blunt-end cat gene was ligated to the inverse PCR product to generate the hpyC1IM1 disrupted plasmid. This plasmid was then transformed into wild type to generate the hpyC1IM1 knockout mutants such that the insertion site of the cat gene is at nucleotide 26. Transposon shuttle mutagenesis was adopted for generating hpyC1IM2 knockout mutants as previously described (16).
Transformation-Plasmid pHel2 (Cam r ) purified from Escherichia coli strain JM109 was transformed into the wild type H. pylori NTUH-C1 strain and hpyC1IM2 and hpyC1IR knockout mutant strains by electroporation. Plasmid pHel3 (Kan r ) purified from E. coli strain ET12567 was also transformed into wild type and hpyC1IM1 knockout mutant by electroporation. (Plasmids pHel2 and pHel3 were gifts from Dr. R. Haas (Max-Planck-Institute fü r Biologie, Tü bingen, Germany) (18).) H. pylori were washed twice with electroporation buffer (9% sucrose and 15% glycerol) prior to transformation. Electroporation was performed with 1 ϫ 10 9 colony-forming unit H. pylori and 1 g of plasmid DNA as previously described (16). Following overnight recovery, the bacteria were selected on a Columbia blood agar plate containing 10 g/ml kanamycin and 4 g/ml chloramphenicol, and the colonyforming units were counted. This experiment was repeated three times. On the other hand, the pHel3 and pHel2 were purified from hpyC1IM1 and hpyC1IM2 mutants, respectively, and then transformed into wild type, hpyC1IM1 mutant, and hpyC1IM2 mutant by natural transformation. Natural transformation was performed with 5 ϫ 10 9 colonyforming unit H. pylori and 1 g of plasmid DNA as previously described (16).
Expression and Purification of HpyC1I Protein-The gene encoding HpyC1I was amplified from DNA of wild type NTUH-C1 strain by PCR and cloned into a TA vector, pGEM-T easy (Promega). The pGEM-T easy/hpyC1IR was digested by NotI (New England Biolabs, Beverly, MA) and ligated in frame into pET28c plasmid (Novagen, Darmstadt, Germany). The resulting pET28c/hpyC1IR plasmid was transformed into an E. coli strain BL21(DE3). The HpyC1I protein was expressed and purified per the manufacturer's instruction under 1 mM isopropyl-1-thio-␤-D-galactopyranoside induction at 25°C (Qiagen, Hilden, Germany).
Recognition and Cleavage Site of HpyC1I-To determine the recognition and cleavage site of HpyC1I, cloning and sequencing of the HpyC1I digestion products from bacteriophage DNA (New England Biolabs) were performed (19). The HpyC1I-digested fragments were blunted by T4 DNA polymerase and cloned into the EcoRV (New England Biolabs) site of pBR322 plasmid. Because the EcoRV site of pBR322 is in the tetracycline resistance gene, the Amp r , Tc s transformants were selected. Plasmid DNA was isolated from Amp r , Tc s colonies, and the restriction fragment-vector junctions were sequenced.
Diversity of HpyC1I R-M System among H. pylori Strains-Genomic DNAs of 15 randomly selected H. pylori strains, including four ATCC strains and 11 clinical isolates were extracted. To examine the modification of CCATC sequences, the DNAs were digested with purified HpyC1I protein. To determine the presence of the HpyC1I R-M system, hpyC1IM1-F (5Ј-TTATGGGGCAAGACGCTGAT-3Ј) and hpyC1IR-R (5Ј-TTAAGCTTTCAAATGGTCATTGATCTG-3Ј) primers corresponding to the 5Ј-end of hpyC1IM1 and 3Ј-end of hpyC1IR, respectively, were used for PCR. To characterize the location of the HpyC1I R-M system, HP1498-F (5Ј-TCACGGCTGTTTATGTGG-3Ј) and HP1501-R (5Ј-CCT-TACGCCGTCAGTATTC-3Ј) primers derived from the conserved region were used for PCR.
RNA Expression Profiles of Wild Type and hpyC1IR Mutant-RNAs from wild type NTUH-C1 strain and hpyC1IR mutant strain were extracted, and microarray hybridization was performed as previously described (20). The density of each microarray signal was standardized with the internal control, 23 S rRNA. The up-regulated and downregulated genes were so defined when the variations of standardized expression levels were greater than twice the S.D.

RESULTS
Screening the Mutant Library by Adherence Assay-A total of 1500 H. pylori mutant strains were obtained from a clinical isolate NTUH-C1 by transposon shuttle mutagenesis (16). Diversity of these mutants has been confirmed by restriction pattern and random sequencing (16). To identify genes involved in adhesion, we screened each mutant strain by a 24well culture plate in duplicate. Six mutant strains revealed a 5-10-fold decrease of the recovered adherent bacteria counts compared with wild type strain after a 30-min incubation with the SC-M1 cells (Fig. 1).
Observation with Light and Electron Microscopy-To ensure the basic bacterial morphology integrity, we observed these mutant strains with decreased adherent ability under light microscopy. Light microscopic observation revealed an elongated morphology (Fig. 2). The length of wild type strain was 4.3 Ϯ 0.82 m, and that of the mutants was 8.7 Ϯ 1.50 m (mean Ϯ S.D.). The elongated phenotype was also observed by electron microscopy (data not shown). Therefore, the decreased adherent ability might be caused by aberrant morphology or other indirect effects.
Identifying the Interrupted Gene by Inverse PCR and Sequencing-By inverse PCR and sequencing, the mini-TnKm insertion site for each of the mutants was determined and compared with the NCBI BLAST data bases (available on the FIG. 1. Adherence assays of H. pylori NTUH-C1 wild type and six mutants with decreased adherence. The adherence ability of wild type strain was defined as 100%, and those of the mutants were calculated proportionally. These data were the mean of three independent experiments.

HpyC1I R-M System in H. pylori
World Wide Web at www.ncbi.nlm.nih.gov/BLAST) as well as the H. pylori genome data base (available on the World Wide Web at www.tigr.org). The same locus in these six mutant strains was interrupted by the mini-TnKm. The transposon insertion site of these six mutants was at nucleotide 773 of this locus. Because initial characterizations were the same in these six mutants, mutant 6 ( Fig. 1) was selected for subsequent studies. This open reading frame (ORF) contained 1617 base pairs but did not have any homologies with the published sequences of H. pylori 26695 and J99 strains (21,22). The nucleotide and protein sequences were compared with NCBI BLAST data bases. The nucleotide sequences only shared short partial sequences with J166 strain-specific C8 sequences (23). The amino acid sequences have 24% identity with a putative nicking enzyme in Bacillus halodurans, and 23 and 20% identity with two Type IIS restriction endonucleases PleI and MlyI, respectively. There were two ORFs located upstream of this locus. Both of the upstream ORFs contained a methyltransferase domain. Based on protein function predictions and gene alignments, we proposed that these three ORFs formed an operon and functioned as a R-M system (DDBJ/EMBL/Gen-Bank TM accession number AB118944). This 3.3-kb fragment was absent in both 26695 and J99 strains. This DNA fragment was located between HP1498 and HP1501 of H. pylori 26695 strain and between jhp1391 and jhp1394 of J99 strain (Fig. 3). The G ϩ C content of the 3299-bp R-M genes in NTUH-C1 strain was 30.7%. The G ϩ C contents of HP1499-HP1500 and jhp1392-jhp1393 (the replaced regions in 26695 and J99 strains) were 32.1 and 31.1%, respectively. In contrast, the G ϩ C content of conserved flanking regions including HP1498 and HP1501 was 37.2%, which matches the average G ϩ C content in 26695 and J99 strains (37.3%). The significantly lower G ϩ C content of this R-M system implied that this R-M system might have been acquired by horizontal gene transfer.
Reknockout and Complementation of hpyC1IR-Because the transposon insertion site of these six mutants was same, we reknocked out hpyC1IR in a fresh wild type NTUH-C1 strain with a cat cassette inserted at a different site. The morphology of hpyC1IR reknockout mutant was also elongated to a same degree as mutant 6 ( Fig. 2). Other phenotypic characteristics and assays of the reknockout mutant were the same as those of the original mutants. The hpyC1IR complementation strain restored the same morphology and adherence as wild type (Figs. 2 and 4).
Transformation Frequency in Mutants-R-M system differences in H. pylori are a barrier to interstrain plasmid transfer (12). We purified pHel2 and pHel3 from E. coli and transformed wild type and R-M knockout mutants. Plasmid pHel2 (Cam r ) from E. coli strain JM109 was introduced into wild type NTUH-C1 strain, and hpyC1IM2 and hpyC1IR knockout mutant strains by electroporation. Plasmid pHel3 (Kan r ) from E. coli strain ET12567 was introduced into wild type NTUH-C1 strain and hpyC1IM1 knockout mutant strain. By transforming plasmids from unrelated hosts, the transformation efficiencies of hpyC1IM1 (1.3 ϫ 10 Ϫ5 ), hpyC1IM2 (3.2 ϫ 10 Ϫ6 ), and hpyC1IR (2.7 ϫ 10 Ϫ6 ) mutants were significantly higher than those of wild type (8.4 ϫ 10 Ϫ8 , 1.0 ϫ 10 Ϫ9 ). The transformation efficiencies of R-M knockout mutants are 100 -1000-fold higher than H. pylori wild-type strain (Table I). On the other hand, the pHel3 and pHel2 were purified from hpyC1IM1 and hpyC1IM2 mutants, respectively, and transformed wild type, and hpyC1IM1 and hpyC1IM2 mutants. The transformations into wild type (2.0 ϫ 10 Ϫ9 , 3.3 ϫ 10 Ϫ7 ) were more difficult than back to the hpyC1IM1 (1.2 ϫ 10 Ϫ5 ) or the hpyC1IM2 (1.4 ϫ 10 Ϫ5 ) mutants. The transformation efficiency of wild type was 100 -10,000-fold lower than hpyC1IM1 and hpyC1IM2 mutants (Table I, bottom). Because the HpyC1I R-M system was an operon, there was no endonuclease activity in these mutants. Therefore, plasmids isolated from either unrelated hosts or methylase mutants were easily transformed into the R-M mutants.
Expression and Purification of Novel Restriction Endonuclease-Based on earlier crystal structures, all Type II and Type IIS restriction endonucleases have a structurally similar catalytic core that spatially brings together three essential charged residues, typically two acidic residues (Asp or Glu) and one lysine residue (Lys), forming a P(D/E)X n (D/E)XK motif (24). The amino acid sequence analysis of this novel restriction endonuclease, HpyC1I, showed such a PD-DTK motif (Pro 435 , Asp 436 , Asp 441 , Thr 442 , Lys 443 ) in the carboxyl terminus. HpyC1I shared amino acid similarities with a putative nicking enzyme in B. halodurans and Type IIS restriction endonucleases, PleI and MlyI. Because the putative nicking enzyme of

B. halodurans
was not yet characterized in detail (25), PleI and MlyI were chosen to digest the chromosomal DNA of wild type NTUH-C1 strain. DNA from wild type NTUH-C1 strain was digested by both PleI and MlyI. These results indicated that the PleI and MlyI R-M system were both absent in H. pylori NTUH-C1 strain. To assay its activity, HpyC1I was expressed in E. coli. His tag fusion protein was generated by using pET28c plasmid and was purified by Ni 2ϩ -NTA agarose. The endonuclease activity of purified protein was detected by cleavage of DNA. The optimal reaction conditions were under 1ϫ NEB buffer 1 (10 mM Bis-Tris propane-HCl, 10 mM MgCl 2 , 1 mM dithiothreitol, pH 7.0) supplemented with 100 g/ml bovine serum albumin and incubated at 37°C. About 60 ng of purified protein (0.1 l) could digest 1 g of DNA in 1 h at 37°C.
The Recognition and Cleavage Site of HpyC1I-To determine the recognition and cleavage site of HpyC1I, phage DNA was digested with HpyC1I and then blunted by T4 DNA polymerase (19). The HpyC1I restriction fragments were cloned into a pBR322 plasmid and then sequenced. Comparisons of the 10 junction sequences indicated that HpyC1I was a Type IIS re-   HpyC1I R-M System in H. pylori striction endonuclease, because a putative nonpalindromic recognition sequence was identified in the cloned inserts at a constant distance from the junction (Table II). The enzyme recognized a 5-bp asymmetric sequence CCATC and cleaved DNA downstream, after nucleotides 4 and 5 in the top and the bottom strand, respectively. The double strand cleavage of HpyC1I produces a one-base 5Ј-protruding end. REBASE searches (available on the World Wide Web at rebase.neb.com) revealed that the recognition and cleavage site of HpyC1I was identical with restriction endonuclease BccI (25). Therefore, HpyC1I was an isoschizomer of BccI. The reaction condition, R-M gene alignment, and digestion pattern of , pBR322, and phiX174 DNA by HpyC1I were identical with BccI (Fig. 5). DNA of wild type H. pylori NTUH-C1 strain was resistant to BccI digestion (data not shown). Therefore, the CCATC sequences of NTUH-C1 strain were modified, and these modifications were resistant to HpyC1I and BccI digestion. The Diversity of HpyC1I R-M Systems among H. pylori Strains-Chromosomal DNAs from 15 H. pylori strains, including four ATCC strains and 11 clinical isolates, were examined for the prevalence of HpyC1I R-M systems. The chromosomal DNAs of the nine strains with HpyC1I R-M system were resistant to HpyC1I digestion (Fig. 6A), whereas those from the other six strains without the HpyC1I R-M system were digested by HpyC1I. These results of DNA digestion were consistent with PCR results of HpyC1I R-M genes. HpyC1IM1-F and hpyC1IR-R primers (corresponding to the 5Ј-end of the gene encoding M1.HpyC1I and 3Ј-end of the gene encoding HpyC1I, respectively) were used to amplify HpyC1I R-M genes (Fig. 6B). PCR products with the predicted size of 3.3 kb were only observed in the nine strains that were resistant to HpyC1I digestion and not in the other six strains. To examine the chromosomal locations of the HpyC1I R-M genes among these strains, HP1498-F and HP1501-R primers corresponding to the conserved HP1498 and HP1501 genes were used for PCR (Fig.  6C). DNAs from the nine strains (including wild type NTUH-C1 strain) yielded PCR products with a predicted size of ϳ4.5 kb. These results indicated that the size and location of the integrated region were the same in the nine strains harboring the HpyC1I R-M genes. DNAs from the other six strains without the HpyC1I R-M genes yielded a predicted size of about 2.1 kb as did 26695 and J99 strains.
RNA Expression Profiles of Wild Type and hpyC1IR Mutant Strain-The RNA expression profiles of wild type and hpyC1IR mutant were analyzed by microarray hybridization. Compared with the expression level of wild type, 10 genes were up-regulated and seven genes were down-regulated in hpyC1IR mutant. Five of the 17 genes had no data base match and no predicted function (Table III). DISCUSSION The initial aim of our study was to identify adhesins in H. pylori. However, the mutants with decreased adherence re-vealed elongated cell structure; the interrupted locus was predicted to be a restriction endonuclease. Therefore, this locus was unlikely to be an adhesin, and the decrease of cell adherence might be caused by changes in cell structure or other indirect changes caused by the knockout of R-M system.
The transposon insertion of these six mutants was at the same site in hpyC1IR. Therefore, these six mutants might have descended from one insertion mutation event. However, reknocked out hpyC1IR in a fresh wild type strain caused the same phenotypic changes, and complementation restored the morphology of wild type; both findings indicated that disruption of hpyC1IR rather than other chromosomal mutations caused these phenotypic alterations.
H. pylori is a genetically diverse species (26). Strain-specific genes may be of great interest biologically, and some may be associated with drug resistance, bacterial surface structure, or restriction-modification. Comparison of 26695 and J99 genome sequences revealed that more than 20 putative R-M systems  could be identified in each strain (11). Unique genes of the J166 strain that were identified by PCR-based subtraction hybridization were predominantly (7 of 18) R-M genes (23). The HpyC1I R-M genes were present in NTUH-C1 strain but absent in both 26695 and J99 strains. We also proved that the HpyC1I R-M system were present in ϳ60% H. pylori strains.  Compared with the G ϩ C content of the conserved flanking regions (ϳ37.2%) and whole genome (ϳ37.3%) of 26695 and J99 strains, the integrated R-M system has significantly lower G ϩ C content (30.7%). This R-M system might be acquired by horizontal transfer during evolution. R-M systems in bacteria were responsible for cleavage of unmodified foreign DNAs to protect their own chromosomal DNA. H. pylori strains have diverse R-M systems resulting in barriers of interstrain DNA transformation (12). Transformation efficiency results also demonstrated CCATC-unmodified plasmids from E. coli, and hpyC1IM1 and hpyC1IM2 mutants were easily digested by the wild type NTUH-C1 strain. However, small amounts of unmodified plasmid DNA could still be transformed. This indicated that either some plasmids might be protected by the methylase in wild type, or a DNA recombination event might have taken place prior to digestion by the restriction endonuclease.
The Type II R-M system consists of DNA methyltransferase and restriction endonuclease. An endonuclease always contains an N-terminal domain for DNA binding and a C-terminal domain for DNA cleavage. The conserved catalytic residues were identified in HpyC1I. Amino acid sequence comparison revealed 23 and 20% similarity with Type IIS restriction endonucleases PleI and MlyI, respectively. However, the HpyC1I recognition site was different from PleI and MlyI because DNA from wild type was digestible by PleI and MlyI. Detailed characterization of its activity and recognition site showed HpyC1I to be a BccI isoschizomer; however, NCBI BLAST and REBASE data base searches did not find nucleotide or amino acid similarities with BccI.
Knockout mutants of hpyC1IR showed cell elongation and decreased adherence to the gastric cancer cell line (SC-M1). The mechanism for the morphological change remained unclear. However, R-M systems could affect many cellular genes as shown by microarray study. There were 10 genes with increased expression and seven genes with decreased RNA expression levels in the hpyC1IR mutant. These results may shed light on the regulatory role of HpyC1I R-M system. Interestingly, a methylase gene of the Type I R-M system, hsdM, was up-regulated in the hpyC1IR mutant. On the other hand, a restriction endonuclease of the Type IIS R-M system, mboIIR, had a decreased expression level in hpyC1IR mutant.
In conclusion, we have identified a novel HpyC1I R-M system in H. pylori and have documented its R-M function and recognition site. This R-M system is present in ϳ60% of the H. pylori isolates. Inactivation of HpyC1I resulted in cell elongation.