Physiology of DNA Uptake

Many H. pylori strains are able to take up and recombine DNA into the chromosome under in vitro conditions (20, 39, 66). The transformation rates determined range between 10⁻³ and 10⁻⁶ (66), depending on the recipient strain and the resistance marker used, e.g., streptomycin, rifampin, and metronidazole. It is not yet clear how transformation competence is regulated in H. pylori. Actively growing H. pylori is more competent than slowly growing or resting cultures (20), and transformation rates have been shown to be highest during early logarithmic phase (27), which is in agreement with the naturally competent bacteria Campylobacter jejuni and Acinetobacter calcoaceticus (32). This is in contrast to Haemophilus influenzae, a gram-negative bacterium in which natural transformation competence is induced by starvation (13). Wang et al. reported transformation rates obtained from cultures grown on agar plates to be 100 to 1,000 times higher than cultures grown in liquid medium (66). In our experience, transformation frequencies are higher when H. pylori is grown in liquid medium.

Transformation frequencies in Campylobacter species increase with higher DNA concentrations and are saturable by DNA concentrations of more than 1 μg per ml (67). In contrast, Wang et al. reported that transformation rates in H. pylori could not be saturated by DNA concentrations of up to 50 μg per ml (66), but Israel et al. described a saturation of transformation by using less than 1 μg of homologous DNA (27). In the gram-negative bacteria H. influenzae and Neisseria gonorrhoeae, specific 8- to 10-mer nucleotide sequences, called DNA uptake signals, are involved in DNA binding and uptake (11, 18). Thus, natural transformation can be inhibited with homologous but not heterologous DNA. In A. calcoaceticus, however, DNA uptake is inhibited by heterologous chromosomal DNA, suggesting that these gram-negative bacteria bind DNA nonspecifically (41). Gram-positive bacteria, such as Bacillus subtilis or Streptococcus pneumoniae, take up DNA without any specificity (56).

The genome sequence of H. pylori lacks potential uptake signals (49). Thus, H. pylori does not seem to recognize and take up DNA by specific sequences. Moreover, DNA uptake in H. pylori is inhibited competitively by the heterologous chromosomal DNA of C. jejuni or Escherichia coli (23, 66). However, other experiments suggested that H. pylori is able to differentiate between foreign and self DNA (27). Thus, it is not yet clear what type of transformation is employed by H. pylori.

Restriction/Modification Systems

In the complete genome sequences of two unrelated H. pylori strains the presence of at least 11 restriction/modification systems was identified on the basis of sequence similarity to endonucleases, methyltransferases, and specificity subunits (62). Furthermore, seven adenine-specific and four cytosine-specific methyltransferases and one of unknown specificity were found. The presence of diverse restriction/modification systems is reflected in the findings of several groups that genomic DNA of different H. pylori isolates showed high diversity of modification patterns (34, 51).

One reason for the different DNA modification patterns of various H. pylori isolates might be that some of the restriction/modification enzymes are encoded by genes involved in slipped-strand mispairing regulation (58), which is able to switch genes on and off very rapidly by changing the reading frame through the insertion or deletion of homopolymeric tracts or dinucleotides (62). This poses the question whether the diverse modification patterns of chromosomal H. pylori DNA have an impact on the transformation frequencies of the DNA of different recipients. Wang et al. provided the first experimental evidence for restriction and modification as a mechanism to control natural transformation of H. pylori chromosomal or plasmid DNA (66). Transformation of plasmid pUOA26 into plasmid-free cells of H. pylori gave frequencies of 1 × 10⁻⁴ transformants per viable cell when plasmid DNA was isolated from the same strain. However, introduction of the same plasmid into a different H. pylori strain, NCTC 11639, resulted in transformation frequencies of 1 × 10⁻⁷ per viable cell.

Similar data were obtained with the shuttle vector pHel2 isolated from H. pylori strain P1, which transformed the homologous strain 100-fold more efficiently than several other strains. The same shuttle plasmid was about 100-fold less effective in transforming strain P1 when it was isolated from E. coli DH5α, which has a different modification pattern than that of P1 (22) (see^{Addendum in Proof}, below). Thus, the fact that many fresh clinical isolates of H. pylori are not transformable with chromosomal or plasmid DNA might be a problem of efficient restriction of the incoming DNA, rather than the incompetence of the strain to take up and recombine the DNA.

The comB Locus

Natural transformation competence in most gram-negative bacteria is associated with production of type IV pili. In gram-positive bacteria, such as B. subtilis or S. pneumoniae, genetic competence also depends on a number of type IV pilin-like proteins and a corresponding prepilin peptidase (14). Functional assembly of the type IV pilus is, however, not necessary for genetic transformation competence, since gram-positive bacteria do not assemble pili at all. In N. gonorrhoeae, one of the best-studied gram-negative systems of genetic transformation competence, several additional genes like pilC, pilT, comL, or comA are necessary for development of genetic competence (14).

H. pylori is a gram-negative bacterium that displays natural competence for genetic transformation, but neither a type IV pilus nor any type IV pilin-like genes have been identified in the two published complete genome sequences. Thus, it appears that the Helicobacter system for natural transformation is organized differently from that of other bacteria. An intensive genetic screen for H. pylori genes involved in transformation competence by transposon shuttle mutagenesis using the blaM reporter transposon TnMax9 resulted in identification of the comB operon (40). Sequencing of the mutated genes located the transposon insertion to the ATG start codon of comB1, designated as HP38 in the H. pylori 26695 genome (24, 62). Further sequence analyses revealed that the comB locus consists of four genes, comB1, comB2, comB3, and a preceding short open reading frame designated as orf2, encoding a 37-amino acid peptide (24). The comB genes are organized in an operon shown in Fig. 1.

Figure 1

Arrangement of gene loci involved in H. pylori transformation competence. Gene arrangement of the four loci that to date have been shown to be necessary for H. pylori transformation competence. Gene orientations, numbers, and designations are shown according (more...)

The comB locus was detected in 10 out of 10 H. pylori strains by Southern hybridizations, independently of whether the strains were transformable or not, indicating that the presence of the comB operon is necessary but not sufficient for transformation (24). Another important element in successful transformations is the modification status of the DNA described in the previous section.

McGowan et al. identified a set of four genes (HP43–HP46) located downstream of the comB genes, involved in the biosynthesis of polysaccharides, especially O-antigens in H. pylori (36). Interestingly, expression of wbcJ (HP45) was demonstrated to be pH-dependent, showing a higher expression at pH 4 to 5 than at pH 7. Since comB seems to be cotranscribed with the lipopolysaccharide (LPS) gene cluster, the authors speculated that the pH-regulated LPS operon might also be involved in natural competence. Hofreuter et al. demonstrated, however, that the knockout of HP43 did not have a drastic effect on the transformation competence in vitro, so that the contribution of the LPS gene cluster to genetic competence would be only marginal(24).

The ComB2 and ComB3 proteins are exported via a sec-dependent signal sequence, but ComB1 does not carry such a signal. ComB1, ComB2, and ComB3 show significant homologies to VirB8, VirB9, and VirB10 of Agrobacterium tumefaciens (23), structural proteins that are part of the type IV secretion apparatus of A. tumefaciens involved in transfer of T-DNA into plant cells. ComB1 has a putative transmembrane domain close to its N terminus and ComB3 near its C terminus. Membrane fractionations and immunoblotting experiments suggested that ComB1 and ComB3 are localized in the cytoplasmic membrane, possibly forming a complex. ComB2 seems to be peripherally membrane-associated (23). According to the model of DNA transformation in gram-negative bacteria, the pilin-like molecules are involved in bridging the inner and outer membrane to form a kind of channel, through which the DNA might be transported. Since type IV pilin-like proteins are missing in H. pylori, it might be possible that the VirB8, VirB9, and VirB10 homologs of H. pylori substitute for this function of DNA transport through the periplasmic space.

Other Loci Involved in Transformation Competence

The genome sequences of H. pylori strains 26695 and J99 contain three genes each with homologies to transformation competence genes: dprA (HP333), comL (HP1378), and comEC (HP1361). Another gene, which is necessary for transformation competence in H. pylori, is the recA gene (50).

DprA was first described in H. influenzae (29), but it is also involved in transformation competence in B. subtilis (43) and S. pneumoniae (8). In H. influenzae, this protein is necessary for transformation with chromosomal DNA, but not with plasmid DNA (29). In contrast, mutation of dprA in H. pylori reduces transformation efficiencies for both chromosomal and plasmid DNA 100-fold (2, 54). The H. pylori ComL homolog does not seem to be involved in transformation competence (54). ComEC is an integral membrane protein required for DNA transport originally described in B. subtilis (43). ComEC-homologs are also found in other gram-negative bacteria. No experimental data are available for the H. pylori ComEC homolog, and it remains to be elucidated whether it is necessary for transformation of H. pylori or not.

Owing to the homologies of the ComB proteins to type IV secretion system components, it may be speculated that other type IV secretion homologs on the cag pathogenicity island also have a function in transformation competence. Such an influence of virB genes on DNA uptake (instead of DNA transfer) has been demonstrated for A. tumefaciens (5, 10). However, a deletion of the cag pathogenicity island genes does not have an effect on the transformation efficiency of H. pylori (25).

Recently, a novel H. pylori-specific competence gene has been identified and named comH (53). The ComH protein is an exported protein that seems to be necessary for DNA binding or DNA uptake since electroporation of comH mutants is still possible. ComH is necessary for transformation with both plasmpids and chromosomal DNA. Like the comB genes, comH homologs are present in competent as well as in less competent H. pylori isolates but absent in other Helicobacter species such as H. mustelae and H. felis.

The RecA Protein

The importance of recombination for natural transformation competence has already been mentioned, but (homologous) recombination is also possible during all other states of transient or permanent diploidy, i.e., during conjugation or transduction or after DNA replication. In E. coli, recombination follows one of several interrelated pathways dependent on the RecA protein (55), which is thus a central component of all homologous recombination processes. A recA homolog of H. pylori has been cloned and characterized (50, 60), and it shows good homology to other bacterial recA genes. H. pylori recA mutants have been shown to have a defect in natural transformation; in the repair of DNA, as suggested by their increased sensitivity to UV light (50); and in acid tolerance (60).

In E. coli, RecA is not only involved in homologous recombination, but it also regulates the SOS response by acting as a coprotease cleaving the LexA repressor and the UmuD protein when activated by exposure to single-stranded DNA. LexA binding motifs upstream of the recA gene or other SOS genes are not present in H. pylori, and there are no homologs to lexA or umuC/umuD in its genome. This suggests that there is no SOS response and no trans-lesion synthesis pathway present in H. pylori, although this has not yet been established experimentally.

An interesting feature of the H. pylori RecA protein is a putative posttranslational modification that does not occur when the H. pylori recA gene is expressed in E. coli. This modification is dependent on one or more genes putatively cotranscribed with recA (17). The function of the modification is not yet known.

H. pylori recA mutants are unable to be transformed with chromosomal DNA, but they retain a low transformation competence for plasmid DNA (50), suggesting that the uptake of plasmids can be achieved independent of RecA, albeit less efficiently. In other naturally competent bacteria, such as N. gonorrhoeae, plasmids are linearized during uptake (4) and have to be recircularized in a RecA-dependent fashion.

Other Components of Homologous Recombination

There are several genes involved in homologous recombination in E. coli. According to the double strand-break repair model, homologous recombination in E. coli proceeds in four steps with different genes involved in each step (30). These steps are initiation (presynapsis), DNA strand transfer (synapsis), processing of heteroduplex DNA (postsynapsis), and resolution of junctions. Initiation involves a processing of DNA with double strand breaks to produce single-stranded DNA and in E. coli is mainly dependent on the RecBCD protein complex in conjunction with χ sites, but also may be performed in the absence of RecBCD by RecJ in cooperation with the recombination-specific DNA helicase RecQ (33). Homologs to RecBCD are not present in H. pylori, but there the genes HP1089 and HP1553 encode proteins with a RecB-like nuclease domain (3). Such a domain is also found in several other helicases such as the RecE protein, and one of the reading frames (HP1553) is indeed homologous to the PcrA helicase II of Staphylococcus aureus. In the absence of RecBCD, exposure of single-stranded DNA tails in H. pylori may be achieved by a RecJ homolog (HP348), possibly together with the RecN homolog (HP1393) and with the UvrD helicase (HP1478) (37), or the ATP-dependent helicase Rep/PcrA (HP 911), because there is no homolog to the RecQ helicase present (Table 1). Together with RecF, RecO, and RecR, these genes belong to the RecF pathway (55), which in E. coli may be involved in replication fork reactivation. In H. pylori, there is no evidence for genes encoding RecF or RecO, suggesting a predominance of the RecF recombination pathway over RecBCD. A similar situation can be inferred from the genome sequence of C. jejuni (42).

Table 1

H. pylori genes that may be involved in homologous recombination and recombinational repair.

The second step of recombination, homologous pairing and DNA strand exchange, is mediated by RecA, which has to be assembled in a filamentous form on the single-stranded DNA in the presence of the competing single-strand binding protein SSB (HP1245). This assembly is supported by the RecF, RecO, and RecR proteins (64), only one of which (RecR, HP925) seems to be encoded in the H. pylori genome. Homologs to recO and recF are also absent from many other genomes, whereas a recR homolog is almost always present (16).

Heteroduplex extension, or Holliday junction branch migration, is mediated by RuvAB. The recombination reaction is completed by removal of thpe junctions by the RuvC endonuclease (68). Branch migration and junction resolution can also be performed by RecG (69), but both systems may have different functions in (recombinational) DNA repair. The four genes ruvA, ruvB, ruvC, and recG (HP883, 1059, 877, and 1523, respectively) are present in the H. pylori genome, indicating that Holliday junction branch migration and resolution may be achieved in a way similar to E. coli.

Macrodiversity of the Genomes

H. pylori has been characterized as an organism with an enormously high degree of genomic diversity between strains in the sequences of single genes (microdiversity). This microdiversity has been attributed to frequent recombination between alleles (59). Macrodiversity, i.e., the different arrangement of genes, seems to be less pronounced in the genomes, although various translocations and inversions can be found comparing the two published genome sequences (1). The arrangement of marker genes along the chromosome has been found to be quite variable between strains (6, 28), with strains 26695 and J99 being rather conserved in comparison to other strains characterized.

The occurrence of sequence repeats in the chromosome of H. pylori (62) and of plasmids containing sequence homologies to the chromosome (1) shows that there are regions of stable diploidy, possibly making recombination a mechanism regulating macrodiversity as well. Thus, the diversity in gene arrangement between the published genome sequences can partly be explained by recombination between sequence repeats such as IS605 or IS606 insertions, among other possible mechanisms. An interesting possibility is a horizontal gene transfer from one H. pylori strain to another by conjugation of plasmids. Although conjugation has not been characterized in H. pylori except for the presence of a DNase-insensitive DNA transfer mechanism (31), the presence of homologous regions between H. pylori plasmids and the chromosome (1) suggests that plasmids may be transferred from strain to strain. Moreover, most of the plasmids sequenced so far contain repeat sequences that may be considered as recombination hot-spots (12, 38). H. pylori may be able to rapidly transfer and shuffle gene modules between strains by recombination between plasmids and the chromosome.

Recombinational Repair

In most organisms, the various types of DNA damage are repaired by several mechanisms using different sets of enzymes. DNA repair is not only necessary in response to exogenous damaging agents, but it is also a housekeeping function during or after DNA replication. Recombinational repair and the mismatch repair system are postreplication repair mechanisms. Other repair mechanisms are independent of DNA replication and include nucleotide excision repair, which is thought to recognize nonspecifically errors in DNA structure, and several damage-specific repair systems.

During replication, recombination is involved in repair of double-strand breaks and single-strand gaps. Besides RecA, E. coli cells need other recombination proteins such as RuvABC and RecG for double-strand break repair. As discussed above, orthologous genes coding for these proteins have been identified in H. pylori. Unless there are joining blocks, single-strand gaps may be repaired very easily using DNA polymerase I and DNA ligase, encoded by HP1470 and HP615, respectively. For oriC-independent reinitiation of the replisome, the PriA protein is needed, a homolog of which is also encoded in the H. pylori chromosome by HP387 (48). Other components of the restart primosome may be the DnaG and DnaB proteins, also encoded in the H. pylori genome by HP12 and HP1362, respectively. Homologs to PriB and PriC are lacking, but this is also the case with many other bacteria, whereas PriA seems to be highly conserved among bacteria (9). However, error-prone trans-lesion synthesis, which depends on the DinB (DNA polymerase IV) or UmuCD (DNA polymerase V) proteins, does not seem to be present in H. pylori.

The resolution of some Holliday junctions during recombinational repair results in the formation of chromosomal dimers, which have to be resolved by the XerCD site-specific recombinases found in many organisms (46). There are two and three homologs belonging to the xerCD family in the genome sequences of strains 26695 and J99, respectively. Interestingly, one homolog in strain 26695 and two homologs in strain J99 are located in the so-called plasticity regions (1, 62). For their function in chromosome segregation, the XerCD proteins are dependent on the cell division-related protein FtsK, a homolog of which is also encoded in the H. pylori genomes.

Mismatch Repair, Nucleotide Excision Repair, and Other Types of Repair

The more specialized repair systems act independently of DNA replication. One such simple system is direct damage repair, which in most cases consists of removing wrong alkyl groups from specific DNA sites. The only H. pylori gene belonging to this group seems to be the ogt gene (HP676) whose product is an O⁶-guanine methyltransferase (45).

Another mechanism is base-excision repair, where a damaged base is removed by the action of a glycosylase. The resulting abasic site is subsequently processed by AP endonucleases, creating a gap that is repaired by DNA polymerase and DNA ligase. The H. pylori genome contains an ortholog of the ung gene (HP1347) encoding a uracil DNA glycosylase, but a corresponding AP endonuclease homolog is lacking. Further genes belonging to this category are two homologs of the nth gene (HP585 and HP602) encoding endonuclease III, a glycosylase with associated AP lyase activity acting primarily on pyrimidine hydrates (44), a homolog to the mutY adenine glycosylase (HP142) that removes adenine residues mispaired to 8-oxoguanine (63), and a homolog to the radA/sms gene (HP223) encoding an ATPase involved in DNA repair (57).

Nucleotide excision repair is a more general system that operates on various types of damage, especially those that cause distortions in DNA (47). The responsible exonuclease UvrABC is present in H. pylori, as inferred from the genome sequence. The H. pylori uvrB gene has been shown to be necessary for acid resistance and for resistance to methylmethane sulfonate and UV radiation (61). The HP1541 reading frame encodes a homolog to the transcription-repair coupling factor Mfd, a protein that is involved in recruiting the UvrABC system to template strand lesion sites.

Another mechanism acting during or after replication is the mismatch repair system for undamaged bases unable to form base pairs. A prerequisite for this type of repair is the differential methylation status of parental and newly synthesized daughter strands. The mismatch is first recognized by binding of the MutS protein, which in turn assembles MutL and the MutH endonuclease. The nick generated by MutH is subsequently processed by exonucleolytic digestion by either exonuclease VII (XseA), RecJ, or exonuclease I (19), with the help of the UvrD protein (DNA helicase II). The MutS and MutL proteins also reduce recombination when the DNAs are sufficiently nonhomologous, a process that may account in part for the maintenance of species barriers (35). The H. pylori genome contains orthologs of the mutS, uvrD, recJ, and xseA genes, but not of mutL and mutH. Since MutS acts in all systems examined so far in combination with MutL, and since the H. pylori MutS homolog belongs to the MutS2 subfamily of proteins that are not involved in mismatch repair, but rather in segregation of chromosomes in eukaryotic cells (15), it is doubtful that a mismatch repair system analogous to that in E. coli exists in H. pylori. Another observation supporting this conclusion is the predominance of transition mutations over transversion mutations in H. pylori (65), whereas the mutHLS mismatch repair system is known to repair specifically transition mutations.

Deletion of genes encoding DNA repair proteins leads to higher mutation frequencies, called mutator phenotypes (26). Thus H. pylori genes such as uvrD, nth, mutY, mutT, mutS, ung, recA, recG, ssb, and others might be called mutator genes by analogy to E. coli. A low level of mutator cells may be beneficial for a population in terms of rapid adaptation properties to external stress that would not be possible in normal wild-type bacteria. Mutation frequencies in H. pylori isolates display a considerable variation, and H. pylori strains with mutation rates even higher than in typical E. coli mutator strains can be isolated (52). Whereas wild-type H. pylori strains develop mutator phenotypes upon subcultivation, a mutation of the mutS gene has recently been shown to prevent such an evolution to a mutator phenotype, indicating that mutS acts as a sort of anti-mutator gene in H. pylori (7). This result shows that ascribing mutator genes only by homology may be misleading.