Mobile genetic elements in Methanobacterium thermoformicicum

<p>The identification of the <em>Archaea</em> as a third primary lineage of life and their adaptation to extreme environmental conditions have generated considerable interest in the molecular biology of these organisms. Most progress in the investigation of archaeal mobile genetic elements, i.e. viruses, plasmids and insertion sequences, has been made in the halophilic branch while only limited knowledge was available about mobile elements of methanogens (reviewed in Chapter 1). The aim of this thesis was therefore to get more insight into the molecular genetics of mobile elements from methanogens. Thermophilic species of the genus <em>Methanobacterium</em> were used as model organisms in this study, since they are among the best characterized methanogens.<p>Phylogenetic studies of the species <em>M.thermoformicicum</em> and strains of <em>M.thermoautotrophicum</em> are described in Chapter 2. The comparison of two variable 16S rRNA regions allowed the conclusions that (i) <em>M.thermoformicicum</em> consists of two groups of strains, the Z-245-group (including strains Z-245, FTF, THF, FF1, FF3 and CSM3) and the CB 12-group (consisting of the strains CB12, SF-4, and HN4), which probably constitute different species, (ii) <em>M.thermoautotrophicum Δ</em> H is closely related to the Z-245-group, and (iii) <em>Mthermoautotrophicum</em> Marburg belongs to neither group and most likely represents a different species (Table 1). This classification of <em>M.thermoformicicum</em> and <em>M.thermoautotrophicum</em> strains obtained by comparative 16S rRNA analysis is in line with recently reported results derived from DNA-DNA hybridization studies. Both approaches are based on genotypic characters which generally are more reliable for the determination of phylogenetic relationships than classical phenotypic ones. Therefore, the proposed reclassification of the examined thermophilic <em>Methanobacterium</em> strains may reflect their phylogeny more accurately than the current classification. Moreover, some other characters of the examined methanogens including genetic fingerprints with <em>fdh</em> A (Chapter 2 and 4), the sensitivity to phage ΦF1 (Chapter 3) and the distribution of FR-I (Chapter 6) provide additional evidence for the proposed new taxonomy of <em>M. thermoformicicum</em> and <em>M. thermoautotrophicum</em> (Table 1). Another phylogenetically relevant aspect reported in Chapter 2 was the finding that the non-formate utilizers <em>M.thermoautotrophicum Δ</em> H and Marburg contain sequences similar to the <em>fdh</em> A and <em>fdh</em> B genes from <em>M. formicicum</em> The presence off dhAB-like sequences in strain ΔH and Marburg suggest that both <em>M.thermoformicicum</em> and <em>M.thermoautotrophicum</em> are descendants of a common formate-utilizing ancestor.<p>It has recently been shown that <em>Mthermoautotrophicum Δ</em> H and Marburg differ clearly in their genome size (1.725 versus 1.623 kb, respectively) and the position of restriction sites in the chromosomal DNA for the endonucleases <em>Not</em> I, <em>Pme</em> I <em></em> and <em>Nhe</em> I. These findings are likely to reflect the limited homology of both <em>M. thermoautotrophicum</em> strains on the genomic level. Similarly, also strain ΔH and <em>M.thermoformicicum</em> THF have been found to differ considerably in genome size (1.725 versus 1. 600 kb, respectively) and <em>Not</em> I, restriction pattern. This would contradict the results obtained by DNA-DNA similarity studies and comparative 16S rRNA sequence analysis which indicate a close relationship of these two strains (Chapter 2). However, the results of the genomic analysis of strain THF have to be interpreted with caution for two reasons. Firstly, the <em>Not</em> I-site (5'-GCGGCCGC) comprises the target site of the GGCC-recognizing restriction-modification system <em>Mth</em> TI <em></em> harbored by strain THF (see Chapter 7). As the consequence, the methyltransferase of the <em>Mth</em> TI <em></em> system probably modifies the <em>Not</em> I-sites contained in DNA from strain THF to yield 5'GCGG <sup><font size="-2">me</font></SUP>CCGC which is resistant against cleavage by <em>Not</em> I . The reported two <em>Not</em> I, restriction fragments of 1.350 and 250 kb generated from THF DNA (7) may therefore not reflect the real number of <em>Not</em> I-sites in the chromosome of strain THF. Secondly, the enormous size of the large <em>Not</em> I, fragment probably does not allow an accurate determination of its size.<p><img src="/wda/abstracts/i1603_1.gif" height="352" width="600"/><p>Chapter 3 describes the characterization of the novel archaeal phages ΦF1 and ΦF3 that are able to infect several thermophilic strains of <em>Methanobacterium</em> (Table 1). Both phages differ with respect to their host range and the topology of their double-stranded DNA genomes. While phage ΦF l has a broad host range and contains a linear, approximately 85-kb genome, ΦF3 was specific for <em>M.</em><em>thermoformicicum</em> FF3 and contained a circular, approximately 36-kb genome. No similarity was found between the genomes of ΦF1 and ΦF3 nor between both phages and genomic DNA from different <em>Methanobacterium</em> strains or from phage ΨM1 of <em>M. thermoautotrophicum</em> Marburg.<p>The isolation and initial molecular analysis of three plasmids, pFV 1 (13.5 kb), pFZ 1 (11.0 kb) and pFZ2 (11.0 kb), harbored by <em>M.thermoformicicum</em> strains THF, Z-245 and FTF, respectively, is reported in Chapter 4. Using cloned pFZ1 DNA as hybridization probe, the plasmids were found to constitute a family of highly related elements. Whereas pFZ1 and pFZ2 are probably identical, only partial similarity was observed between pFZ1 and pFV1. Furthermore, genetic fingerprinting experiments using <em>fdh</em> A from <em>M.formicicum</em> as hybridization probe allowed a subdivision of the species <em>M.thermoformicicum</em> into the CB12-group and the Z-245-group, comprising among others the plasmid-harboring strains Z-245, M and THF (Table 1).<p>The relatedness of the plasmid DNA was confirmed by sequence analysis of pFV I and pFZ I (Chapter 5). Comparison of the primary sequence of pFV1 (13513 bp; see also Appendix A) and pFZ I (110 14 bp; see also Appendix B) revealed a modular organization of the plasmid genomes: a backbone structure, conserved on the sequence level and in overall gene order, which is interspersed with plasmid-specific elements. The organization of the <em>M.thermoformicicum</em> plasmids resembles that of prokaryotic chromosomes both of which are subjected to DNA rearrangements, in particular insertions and/or deletions, thereby retaining a basic genome organization. The high sequence similarity and the comparable genetic maps suggest that pFV1 and pFZ1 (and pFZ2) have originated from a common ancestor. If so, the plasmids probably share homologous functions necessary for plasmid replication and maintenance which should be located within regions with high interplasmid similarity, i.e. the plasmid backbone. Those essential functions may include two large palindromic regions and a putative gene encoding a NTP-binding protein which are contained within the conserved backbone structure. In contrast, the plasmid-specific sequence blocks probably represent accessory elements which do not specify essential plasmid functions.<p>Both plasmids pFV1 and pFZ1 harbor sequences with similarity to chromosomal DNA of different thermophilic methanogens (Chapter 6). One of those, named FR-I, represents an accessory element of pFV1 with chromosomal counterparts in several <em>M.thermoformicicum</em> strains and <em>M.thermoautotrophicum Δ</em> H (Table 1). Comparison of the plasmid-derived and chromosomal FR-I elements revealed that FR-I has a size of 1.5 kb and exists in variants which differ in the organization of two subfragments. Remarkably, the corresponding subfragments of either FR-I element were identical on the nucleotide sequence level. Although FR-I lacks terminal inverted repeats, the presence of terminal direct repeats and its occurrence in multiple copies suggest that FR-I represents a new type of archaeal insertion sequences. A second element, termed FR-II, is part of both plasmids pFV I and pFZ 1, and the chromosome of <em>M.thermoformicicum</em> THF, CSM3 and HN4. Sequence analysis of the two plasmid- and one chromosome-derived FR-II elements showed that they are highly similar and may code for a protein with yet unknown function. In contrast to FR-I, FR-II is present in a single chromosomal copy and does not contain terminal repeats. Its presence in plasmid- and chromosomal DNA suggests, however, that FR-II is mobile or has been mobilized.<p>Each plasmid contains an accessory element encoding components of a type 11 R/M system: the GGCC-recognizing system <em>Mth</em> TI <em></em> carried by pFV1 from strain THF (Chapter 7) and the CTAG- recognizing systems <em>Mth</em> ZI <em></em> and <em>Mth</em> FI <em>located</em> on pFZ1 and pFZ2 from strain Z-245 and FTF, respectively (Chapter 8). These findings demonstrated for the first time plasmidencoded enzymatic activities for methanogens. The R/M systems <em>Mth</em> TI <em></em> and <em>Mth</em> ZI have been characterized in detail by their cloning and expression in <em>Escherichia coli.</em> Strikingly, both R/M modules are located within the same plasmid backbone region suggesting that this part of the plasmids has undergone substantial genomic rearrangements: either deletion or insertion of a single R/M module from an ancestral plasmid that contained both or none of the R/M cassettes. The presence of R/M cassettes would support the concept of a modular evolution, i.e., the mobilization of functional units within the same or between different replicons. Characterization of methanogenic R/M systems with the same specificity as the plasmid-located <em>Mth</em> TI and <em>Mth</em> ZI systems may provide insight in the relationships of archaeal R/M systems and possible mechanisms of dissemination of R/M modules.<p>Supporting evidence for a modular evolution is provided by the similarity between the two GGCC-recognizing R/M systems <em>Mth</em> TI <em></em> from <em>M.thermoformicicum</em> THF and <em>Ngo</em> PII from <em>Neisseria gonorrhoeae</em> (Chapter 7). In contrast to other R/M systems with GGCC specificity, <em>Mkh</em> TI <em></em> and <em>Ngo</em> PII comprise endonuclease- and methyltransferase genes which both were similar on the nucleotide and the deduced amino acid sequence level and exhibit an identical organization (Figure 1). These findings suggest that the <em>Mth</em> TI <em></em> and the <em>Ngo</em> PII <em></em> systems are homologous and have been disseminated via horizontal gene transfer. If so, gene transmission would have occurred between members of the evolutionary different domains <em>Archaea</em> and <em>Bacteria.</em><p><img src="/wda/abstracts/i1603_2.gif" height="346" width="600"/><p>An interesting property of the <em>Mkh</em> TI system is the fact that it is present in a thermophilic organism. In contrast, all other known GGCC-recognizing, or, more general, m <sup><font size="-2">5</font></SUP>C-producing RIM systems, have been isolated from mesophilic organisms. The absence of such RIM systems in thermophiles has been attributed to an accelerated deamination rate of m <sup><font size="-2">5</font></SUP>C (m <sup><font size="-2">5</font></SUP>C <sup>_</SUP>>T) at high temperatures generating G-T mismatches in double-stranded DNA which, if not corrected, would result in point mutations (1) (Figure 2). The thermophilic <em>M.thermoformicicum</em> THF apparently contains a mechanism which either avoids deamination of m <sup><font size="-2">5</font></SUP>C or is capable of G-T mismatch correction. The latter possibility seems to be realized in strain THF since the pFV1-located ORF10 has the capacity to code for a protein with significant similarity to <em>E.coli</em> DNA mismatch repair enzymes (Chapter 5; Figure 2). Similar to the organization of functionally related genes into operons, ORF10, <em>mthTIM</em> and <em>mthTIR</em> are located on the same pFV1 module (Figure 2) and may form a functional unit that specifies components of a restriction-modification-repair (RMR) system. Additionally, a putative fourth gene, ORF9, is located on the RMR module (Figure 2). Although no function could be assigned to the deduced ORF9 product, a participation in the activity of the RMR system would be plausible. Besides a possible common regulation, the observed gene clustering may also be necessary to ensure survival of a potential thermophilic recipient which acquires the RMR module via lateral gene transfer. However, if the recipient is a mesophilic organism, the DNA repair functions would probably be dispensable and may be lost after transfer. Assuming that a transfer of the entire RMR module has occurred from the thermophilic <em>M. thermoformicicum</em> THF to the mesophilic <em>N.</em><em>gonorrhoeae,</em> this may explain why the similarity included the genes of the <em>Mth</em> TI <em></em> and <em>Ngo</em> PII systems but did not extend beyond adjacent sequence regions.<p>The major function of RIM systems is proposed to be the protection of the resident DNA from contamination by for instance phage DNA (Figure 2). The same function may be supposed for the discussed R/M systems from <em>M.thermoformicicum</em> However, one would expect that RIM systems which generate thermostable modification products such as m <sup><font size="-2">4</font></SUP>C and m <sup><font size="-2">6</font></SUP>A would be more advantageous for a thermophilic host than those that produce m <sup><font size="-2">5</font></SUP>C since deamination of the latter methylation product increases the chance on point mutations which have to be compensated by mismatch repair. Nevertheless, the m <sup><font size="-2">5</font></SUP>C-generating <em>Mkh</em> TI <em></em> R/M system obviously proved to be successful under thermophilic conditions since it is maintained by <em>M.thermoformicicum</em> THF. A possible explanation for the presence of the <em>Mth</em> TI <em></em> system in strain THF may be that the system provides another advantage to its host, in addition to restriction of incoming phage DNA by the <em>Mkh</em> TI <em></em> endonuclease. As discussed in Chapter 5, deamination of m <sup><font size="-2">5</font></SUP>C may serve as additional Protection mechanism against phages that contained m <sup><font size="-2">5</font></SUP>C-methylated genomes since, in contrast to the host, these phages are not be able to correct mismatches.<p><img src="/wda/abstracts/i1603_3.gif" height="429" width="600"/><p>Analysis of the abundance of the tetranucleotide sequences GGCC and CTAG, the targets of the <em>Mth</em> TI and <em>Mth</em> ZI <em></em> R/M systems, in DNA of thermophilic <em>Methanobacterium</em> strains revealed some interesting results. The sequence data used for analysis comprised 48.4 kb of chromosomal and plasmid DNA, the majority of which were obtained from the EMBL and GenBank data bases (Table 2). The tetranucleotide GGCC was found 228 times within 48.4 kb, which statistically corresponds with 1 tetranucleotide per 212 bp. This value is almost equal to the expected random- frequency for a tetranucleotide which is 1 per 256 bp. In contrast, only 13 CTAG tetranucleotides were detected, corresponding with 1 per approximately 3700 bp. A similar low abundance of CTAG (1 per 2700 bp) has been reported for the <em>E.coli</em> genome. This finding clearly shows that the recognition sequence of the <em>Mth</em> ZI <em></em> system, in contrast to that of the <em>Mth</em> TI <em></em> system, is rare in DNA from thermophilic <em>Methanobacterium</em> strains. However, a considerable different result was obtained from analysis of the plasmid DNA harbored by <em>M. thermoformicicum.</em> Compared to the values observed for the genomic DNA from thermophilic members of <em>Methanobacterium,</em> plasmid pFZ1 exhibits a lower frequency of GGCC (1 site per 610 bp; 22 GGCC-sites per pFZ1 genome) but shows a significantly higher abundance-of CTAG (1 per 670 bp; 20 CTAG-sites per pFZ1 genome) (Table 2).<p><img src="/wda/abstracts/i1603_4.gif" height="524" width="600"/><p>While pFZ1 contains an comparable number of GGCC-sites (28 per plasmid genome with a frequency of 1 per 390 bp), the CTAG-frequency is substantially lower in pFZ1 from strain Z-245 (1 per approximately 1800 bp; 6 per pFZ1 genome) (Table 2). These results are interpreted as follows:<p>(i) The low abundance of CTAG in the DNA of thermophilic <em>Methanobacterium,</em> strains and <em>E. coli</em> suggests that a similar mechanism is responsible for this bias. It has recently been reported that the underrepresentation of several tetranucleotides, including CTAG, in the genome of <em>E.coli</em> may be the result of selection against target sequences of the <em>vsr</em> gene product, an enzyme involved in G-T to G-C mismatch correction. Usually, those target sequences of the Vsr endonuclease are generated by deamination of m <sup><font size="-2">5</font></SUP>C in the sequence C <sup><font size="-2">me</font></SUP>C(A/T)GG, the product of the <em>dcm</em> MTase. However, the <em>vsr</em> gene product displays a relative relaxed specificity since it also recognizes G-T mismatches in the sequence C <u>T</u> (A/T)G or <u>T</u> (A/T)GG where the underlined T is mismatched with G, and it does not require methylation of the unmutated strand. Consequently, T-G mismatches produced during recombination or DNA replication would, without determination which of both nucleotides is the incorrect one, sequence-specifically be corrected to C-G pairs. The tetranucleotide C <u>T</u> AG is one of the possible target sequences of the Vsr endonuclease and would be 'repaired' to CCAG, resulting in an elimination of CTAG-sites in time. A similar mechanism may be responsible for the low CTAG abundance in DNA of thermophilic members of <em>Methanobacterium.</em><p>(ii) Compared to the genomic DNA of thermophilic <em>Methanobacteriwn</em> strains, plasmid pFZ1 of strain THF showed a five-fold increased number of CTAG tetranucleotides. This significant difference may indicate that selection against CTAG is absent in strain THF. Alternatively, pFZ1 (and the related pFZ1 and pFZ2) does not originate from DNA of a thermophilic <em>Methanobacterium</em> strain. The relatively low GC content of pFZ1 (41.8%) compared to that of strain THF (49.6%) would support the latter possibility.<p>(iii) Although pFZ1 of strain Z-245 is highly related to pFZ1 the abundance of the tetranucleotide CTAG is clearly lower in pFZ1 Comparison of the nucleotide sequence of plasmid pFZ1 with the one from pFZ1 showed that in particular CTAG-sites located within putative coding regions were absent in pFZ1 The lower frequency of CTAG-sites in pFZ1 correlates with the presence of the CTAG-recognizing <em>Mth</em> ZI <em></em> system that is encoded by pFZ1 A possible explanation would be that the MTase <em>M.Mth</em> ZI <em>provides</em> only an incomplete protection against cleavage of the corresponding ENase which would result in selection against CTAG-sites. Another possibility may be that strain Z-245 contains a Vsr-like activity.<p>(iv) The presence of the GGCC-recognizing <em>Mth</em> TI system encoded by pFZ1 seems to correlate with the underrepresentation of GGCC-sites in the pFZ1 genome. Comparison of the nucleotide sequences of pFZ1 and pFZ1 revealed that 15 GGCC-sites present in pFZ1 were changed in pFV1. Remarkably, ten of those changed GGCC-sites in pFZ1 displayed the sequence GGTC. The same base substitution would be generated by deamination of a single m <sup><font size="-2">5</font></SUP&