Host-Mimicking Strategies in DNA Methylation for Improved Bacterial Transformation

In 1928, Griffith [1] reported that soluble substances from virulent pneumococcal cells trans‐ formed non-virulent pneumococcus to virulent forms. This substance has now been demon‐ strated to be DNA [2-4]. This is considered to be the first report on genetic transformation of bacteria by exogenous DNA. Subsequently, natural competence of Bacillus subtilis was re‐ ported in 1958 by Young and Spizizen [5]. They also demonstrated genetic transformation of natural competent B. subtilis cells using exogenous DNA. It was in 1970 that genetic trans‐ formation of Escherichia coli using chemically competent cells was reported [6]. Thus, genetic transformation of common bacterial models was established at an early stage in the develop‐ ment of bacteriology. The alternative view is that bacterial models such as B. subtilis and E. coli have become the mainstay of this field because of high transformation ability. Genetic transformation techniques remain important for studying numerous bacteria and for the ad‐ vancement of bacteriology, biochemistry, applied microbiology, and microbial biotechnolo‐ gy. Moreover, recent developments in the search for new bacteria and genome sequencing have provided numerous effective bacteria that are useful for biological studies and indus‐ trial applications. With these developments, there is a greater demand for establishing ge‐ netic transformation methods for more bacteria.


Introduction
In 1928, Griffith [1] reported that soluble substances from virulent pneumococcal cells transformed non-virulent pneumococcus to virulent forms.This substance has now been demonstrated to be DNA [2][3][4].This is considered to be the first report on genetic transformation of bacteria by exogenous DNA.Subsequently, natural competence of Bacillus subtilis was reported in 1958 by Young and Spizizen [5].They also demonstrated genetic transformation of natural competent B. subtilis cells using exogenous DNA.It was in 1970 that genetic transformation of Escherichia coli using chemically competent cells was reported [6].Thus, genetic transformation of common bacterial models was established at an early stage in the development of bacteriology.The alternative view is that bacterial models such as B. subtilis and E. coli have become the mainstay of this field because of high transformation ability.Genetic transformation techniques remain important for studying numerous bacteria and for the advancement of bacteriology, biochemistry, applied microbiology, and microbial biotechnology.Moreover, recent developments in the search for new bacteria and genome sequencing have provided numerous effective bacteria that are useful for biological studies and industrial applications.With these developments, there is a greater demand for establishing genetic transformation methods for more bacteria.
DNA introduction is an essential process for transforming target bacterium by exogenous DNA.Various methods for introducing DNA into bacteria have been developed to date, including chemotransformation, electroporation, sonopolation, tribos, and conjugational transfer [7].In spite of these developments, it is often difficult to establish transformation methods for target bacterium.A possible reason is the difficulty faced while exploring suitable conditions for introducing DNA, which requires not only theoretical understanding but also a trial and error approach.Circumventing bacterial RM systems is a major challenge.These systems defend bacteria against transformation by exogenous DNA, such as bacteriophages, and effectively hamper genetic transformation by exogenous plasmids.RM systems selectively digest exogenous DNA by differentiating them from host-endogenous DNA on the basis of host-specific DNA methylation [8].Therefore, DNA that imitates the methylation patterns of the host bacterium (host-mimicking DNA) is incorporated into the bacterium without restriction.Thus, RM systems can be overcome by theoretical host-mimicking strategies (Figure 1), rather than exploring conditions for introducing DNA.This chapter explains host-mimicking strategies and provides tips for establishing transformation methods for new bacteria.

RM systems
In 1962, Arber and Dussoix [9] found that bacteriophage λ carried specificity for E. coli strains in which they were produced.For examples, bacteriophage λ from E. coli K-12 was efficiently transfected into K-12 and C strains, but not strains B and K(P1) (efficiencies were <10 −4 fold).Meanwhile, bacteriophage λ produced by E. coli C efficiently transformed strain C, but not strains K-12, B, and K(P1) (efficiencies were <4×10 −4 fold).Thus, bacteriophage λ vectors readily infected E. coli strains that produced them but the infection to other strains was "restricted".This occurrence, termed restriction, is explained by E. coli RM systems that act as host defense against exogenous DNA.RM systems selectively digest exogenous DNA and greatly influence the efficiency of genetic transformation by exogenous DNA.Numerous RM systems have been found in bacteria and archaea, and are classified into four main types (type I-IV) [10].Their general properties are summarized in Table 1.The REBASE database (rebase.neb.com)[11] has accumulated large amounts of information about RM systems, including types, gene sequences, recognition sites, and origin organisms.Among the four types of RM systems, type II consists of RE and MT.In this type, RE cuts exogenous DNA at specific sites but not endogenous DNA that has been methylated by MT (Figure 1).Type I and III systems also cut exogenous DNA by similar mechanisms, but comprise protein complexes of some subunits.Type IV is known as a modification-specific restriction system to cut DNA with heterologous modifications.Intriguingly, RM systems behave as selfish elements like viruses and transposons [12], implying that RM systems have been irreversibly distributed in bacteria.The following sections describe more details of RM systems.

DNA modifications involved in DNA restriction
Several DNA modifications have been elucidated and in almost all cases the modifications that are involved in restriction are nucleobase methylations.The main forms of methylated nucleobases are N 6 -methyladenine ( 6m A; Figure 2), 5-methylcytosine ( 5m C), and N 4 -methylcytosine ( 4m C).These modifications are performed for double-stranded DNA using methyltransferases or other methyltransfer machinery.In addition, some bacteriophage DNA contain 5-hydroxymethylcytosine ( 5hm C) instead of cytosine.This modification is incorporated during phage DNA replication using 5-hydroxymethyldeoxycytidine triphosphate as the substrate [13].The hydroxyl group is further glucosylated to produce β-glucosyl-5-hydroxymethylcytosine ( ghm C) in a phage-specific pattern by glucosyltransferase [14].This modification has various biological functions, including circumvention of restriction barriers of RM systems in phage hosts [13,14].The modified cytosine 5hm C is also found in mammalians [15][16][17].Unlike phages, it is produced by oxidation of 5m C in double-stranded DNA [18].In bacteria, there are no known type I-III RM systems involving 5hm C or ghm C.However, several type IV systems that restrict DNA containing 5hm C and/ or ghm C have been reported [19][20][21].In addition to the methyl-based modifications described above, sulfur modification (phosphorothioation) of DNA backbones has been observed [22,23].Notably, it is suggested that this modification is involved in DNA restriction in Salmonella enterica [24].The gene cluster for this RM system consists of eight genes, of which four are involved in phosphorothioation, while seven genes are essential for restricting unphosphorothioated DNA.Similar gene clusters, along with DNA containing phosphorothiol bonds, are found in many bacteria, implying that phosphorothioation is a widespread DNA modification [24,25].A type IV system that restricts phosphorothioated DNA has been also reported [26].
The cleavage positions are distal from recognition sites and are variable.It is believed that the complex of type I RM system, while binding to recognition sites, translocates (or pulls) the DNA along in an ATP dependent fashion, and cleaves DNA when the translocation is impended by collision and/or by stalling with another translocating complex [27,31,32].
Electron microscopy analysis has been used to detect ATP-dependent formation of loop DNA during DNA cleavage by a type I RM complex [32].The R subunit is responsible for ATP hydrolysis, translocation, and endonuclease activity, but not DNA binding.The binding depends on subunits M and S, which are therefore essential for both endonuclease and methyltransferase activities.
The complex of type I RM system acts similar to methyltransferase when it recognizes DNA that is hemimethylated (methylated on one strand) at recognition sites [27].The methylation is performed by M and S subunits using S-adenosyl-L-methionine as the methyl donor [8].
The M subunit has the binding site for S-adenosyl-L-methionine, while the S subunit is essential for determining recognition sites.The R subunit is unnecessary for methyltransferase activity.In all cases reported so far, methylation by type I RM systems occurs in adenine to produce 6m A.

Type II RM systems
Type II RM systems are extremely diverse and are currently classified into 11 subfamilies [8].Generally, these comprise two enzymes, RE and MT.Cleavages of exogenous DNA at unmethylated recognition sites is carried out by RE, which spares endogenous DNA that has been methylated by the cognate MT.Most REs require Mg 2+ ions as a cofactor for cleavage.Although RE may form monomers, dimers, or tetramers, it functions without forming complexes with the cognate MT.The recognition of cleavage sites is highly precise and recognition sequences are often palindromic.Such sites include 5′-GAATTC-3′ for EcoRI and 5′-GGATCC-3′ for BamHI, which are cleaved symmetrically within the sites.Because of these useful properties, more than 3,500 REs have been characterized, and many are widely utilized in recombinant DNA technology [8].The enzyme MT catalyzes methylation at recognition sites using S-adenosyl-L-methionine as the substrate and generally acts as a monomer.The nucleobases produced are 6m A, 5m C, or 4m C. For example, EcoRI and BamHI methyltransferases produce 5′-GA 6m ATTC-3′ and 5′-GGAT 4m CC-3′, respectively.

Type III RM systems
Type III RM systems operate with multi-subunit machinery comprising two R subunits and two M subunits (R 2 M 2 ) [33].Subunit M contains recognition domain for binding to specific sites and also a methyltransferase domain.It can thereby bind at recognition sites independently, and methylate DNA using S-adenosyl-L-methionine as a substrate.The nucleobase produced is 6m A in all cases reported so far.Unlike type I and II RM systems, full modification is actually hemimethylation (methylation on one strand).The recognition sequences are asymmetric, such as 5′-CAGCAG-3′ for EcoP15I (methylated adenine is underlined).
Subunit R has an ATP dependent DNA helicase and endonuclease domains that are responsible for DNA cleavage.This subunit is unable to bind to DNA and therefore requires subunit M to cleave DNA.Two unmethylated sites that are inversely oriented (head-to-head orientation) serve as the target for DNA cleavage.Cleavage occurs at 25-27 bp downstream of one of the recognition sites, which is chosen randomly from the two sites.Even DNA with 3.5 kb between the two sites is cleaved.The cleavage requires ATP similar to type I RM systems.However, the amount of ATP consumed is only ~1% of that required for cleavage by type I RM systems.This fact makes it difficult to transpose the translocation model that is proposed in type I RM systems to type III RM systems.Thus, some alternative models have now been proposed [33].

Sulfolobus acidocaldarius
In vivo (HG) Electroporation [38] Thermoanaerobacter sp.X514 In vivo (IG) Sonoporation [66] Table 2. Microbial transformation using host-mimicking DNA.DS: in vivo methylation in a strain that is related to the target bacterium and is deficient in restriction and proficient in methylation; EX: in vitro methylation using a crude extract of the target bacterium; GM: in vivo methylation using methyltransferase genes in the target bacterial genome; HG: in vivo methylation using heterologous genes; IG: in vivo methylation using E. coli intrinsic genes; MF: methyl-free DNA; MT: in vitro methylation using commercially available methyltransferases; and SD: transformation using DNA with specifically abolished recognition sites.

Host-mimicking strategies
Circumvention of RM systems is critical for establishing transformation methods for target bacteria (Table 2).This is true not only for bacteria but also archaea [38].This section describes some strategies for circumventing RM systems and focuses on host-mimicking.Because all types of RM systems digest exogenous DNA after distinguishing it from endogenous DNA on the basis of host-specific methylation patterns, DNA modification that mimic these patterns evade digestion.A general flowchart for producing host-mimicking DNA is shown in Figure 3.The details are described in the following sections.

A brief survey of functional RM systems
As mentioned earlier, type I-III RM systems involve DNA methylation in their functions.Therefore, the presence of methylated DNA in bacterial chromosomes indicates a type I-III RM system in the bacterium.Hence analysis of methylated DNA is an effective survey method to identify functional RM systems in target bacterium.Methylated DNA in chromosomes can be analyzed using high-performance liquid chromatography (HPLC) [15,18,[39][40][41][42][43][44][45].This method determines the presence of deoxynucleosides and methylated deoxynucleosides.Deoxynucleosides are prepared by hydrolyzing chromosomal DNA with nuclease P1 and alkaline phosphatase [39][40][41], are separated using reverse mode C 18 -based silica columns, and are detected by ultraviolet absorption at 260 and/or 280 nm.Authentic 6m dA, 5m dC, and 4m dC are commercially available or can be prepared from methylated 2′-deoxynu-cleoside-5′-triphosphate by dephosphorylation using E. coli alkaline phosphatase [40].Note that this analysis requires large amounts of DNA (>10 µg).Contamination with RNA, proteins, and chemicals that absorb ultraviolet radiation may also complicate accurate analysis.
In some reports [18,42,43], deoxynucleosides are more accurately identified by combining HPLC and mass spectrometry analysis.In addition, immunochemical methods are available for analyzing methylated DNA [46][47][48][49].Several anti-5m C antibodies have been developed and are commercially available.Although commercially available antibodies against other methylated DNA are limited, Kong et al. [49] have reported successful production of rabbit polyclonal antibodies against 6m A and 4m C. Analysis of DNA using HPLC allows determination of deoxynucleoside composition and methylation frequency in chromosomes.The coefficients in Table 3 may be used for estimation.For example, HPLC analysis of S. griseus chromosomes revealed 5m dC but not 6m dA or 4m dC [40].The composition ratio of 5m dC to dC was 0.7 mol%, suggesting that S. griseus possesses approximately one 5m C per 0.5 kb of chromosomal DNA with 67% GC content.The chromosome of G. kaustophilus contains 6m dA and the composition ratio to dA is 2.0 mol% [41].This suggests that G. kaustophilus possesses approximately one 6m A per 0.1 kb of chromosomal DNA with 52% GC content.Although DNA methylation unrelated to RM systems have been observed, such as E. coli Dam and Dcm methylation [67,68], high frequency methylations imply that the bacteria may harbor a considerable type I-III RM system.Meanwhile, the chromosomes of S. avermitilis, S. coelicolor, and S. lividans contain no methylated DNA (my unpublished data).This observation suggests that these bacteria harbor no type I-III RM systems.However, the possibility remains that a functional type IV system exists in these species.Potent methyl-specific restrictions have been observed in S. avermitilis [61], S. coelicolor [35], S. bambergiensis [62], and S. natalensis [65].

Methylation site analysis
When significant DNA methylation is observed in the target bacterium, preliminary determination of DNA methylation sites is generally required to produce host-mimicking DNA.
In S. griseus, bisulfite-based analysis of a plasmid library isolated from this bacterium (Figure 4A) was used to determine the two consensus sites 5′-GAG 5m CTC-3′ and 5′-GC 5m CGGC-3′ [40].Note that this method is not employed for methylation analysis in bacteria that cannot be transformed with exogenous plasmids, and in bacteria that have methylated nucleobases other than 5m C. For determining 6m A consensus sites in G. kaustophilus, chromosome digestion using methyl-sensitive restriction enzymes (Figure 4B) was used to reveal 5′-GG 6m ATC-3′ and 5′-G 6m ATCC-3′ site [41].Methods using methyl-sensitive restriction enzymes help identify several methylation species, including 6m A, 5m C, and 4m C.However, methylation sites that can be identified by these methods are limited due to the lack of commercially available restriction enzymes.Recently, direct detection using real-time DNA sequencing has been reported [74,75].This method potentially enables the exhaustive determination of all methylation sites in chromosomal DNA.Although this method requires special equipments, which has limited availability, it may become one of the most promising methods for methylation analysis in the future.Also, the author has now developed a versatile immunological method for determining consensus sequences with methylated nucleobases.

Production of host-mimicking DNA
If the target bacterium contains no or negligible methylated DNA, methyl-free DNA should be used as host-mimicking DNA because the bacterium may harbor type IV restriction systems, as exemplified by transformation of S. avermitilis [61], S. coelicolor [35], and S. natalensis [65].Methyl-free DNA can be readily produced using E. coli strains deficient in DNA methyltransferase genes (dam − dcm − hsd − ), such as E. coli IR27, ET12567, IBEC58, and HST04 (Table 4).Methyl-restrained DNA from E. coli GM2929 and SCS110 (dam − dcm − hsd + ) were also used for efficient transformation of S. bambergiensis [62] and B. anthracis [50], respectively.Because Hsd mediated methylation is of low frequency in E. coli, deficiency of this methyltion may not be essential for circumventing type IV systems.
In transformations of B. cereus [51], B. weihenstephanensis [51], and H. pylori [58], crude extracts prepared from the respective bacterium were used for DNA methylation.This method has high cost-performance and high versatility but may not be efficient because of low methyltransferase activity in crude extracts and DNA degradation by nucleases.Moreover, type I and III methylation pattern cannot be achieved by this method.Another approach is in vivo methylation by expressing methyltransferase genes in E. coli cells [38,40,41,52,55,56]. Type II and III methylation can be reconstituted by expressing MT and M subunit genes, respectively, and type I methylation can be reconstituted by simultaneous expression of M and S subunit genes.Numerous gene sequences of methyltransferases are accumulated in the REBASE database along with their methylation sites [11].Either plasmids or chromosomal integration can be used as expression vectors for methyltransferase genes.When the genome sequence of the target bacterium has been determined, methyltransferase genes in the genome may be used for in vivo methylation [40,41,52].Methylation site analysis is not essential for this approach; however, functional expression of methyltransferase genes requires confirmation by HPLC analysis of recombinant E. coli chromosomes.A methyltransferase gene of S. griseus, responsible for 5′-GAG 5m CTC-3′ methylation, was found to be nonfunctional in E. coli.Hence, an alternative methyltransferase gene from S. achromogenes (M.SacI) was used for DNA methylation [40].The E. coli host used to produce host-mimicking DNA must be a methylation-deficient strain (Table 4) because the target bacterium may have type IV systems in addition to type I-III RM systems, as exemplified by G. kaustophilus transformation [41].In addition, it is desirable that the E. coli host is deficient in type IV system genes (mcrA, mcrBC, and mrr) because these may restrict heterologous methylation in E. coli cells.In this regard, E. coli strains IR27 and HST04 are appropriate for producing host-mimicking DNA through in vivo methylation.Although in vivo methylation may be more complicated than in vitro methylation, this approach often has excellent cost-performance, versatility, and efficiency.
A derivative of the target bacterium with methylation activity and reduced restriction activity can also be used for the production of host-mimicking DNA.In transformations of Salmonella typhimurium and Staphylococcus aureus, plasmids isolated from E. coli strains were initially introduced and propagated in the restriction-deficient strains LB5000 [59] and RN4220 [60], respectively, and were then used for transformation of other strains.In S. griseus transformation, mutant HH1 that reduces restriction activity compared to the wild-type has been used [63].This approach enables production of perfect host-mimicking DNA, although it is not easy to find a strain that is both deficient in restriction and proficient in methylation.

Alternative methods for circumventing RM systems
In addition to host-mimicking strategies, there are some simple approaches for circumventing RM systems.One is to abolish sites recognized by RM systems.In Clostridium difficile transformation, five CdiI sites in plasmids were abolished and used to demonstrate improved efficiency [56].Similarly, a plasmid with three abolished SacII sites was used for efficient transformation of Bifidobacterium longum [53].When methyltransferase enzymes or genes are unavailable, this approach can be an effective alternative.One other approach is to reduce the restriction activity in the target bacterium temporarily by heat treatment.In B. amyloliquefaciens transformation, heat treatment at 46°C for 6 min increased transformation efficiency [78].More forcible heat treatment (higher temperature and longer time) inactivates RM systems more efficiently but concurrently reduces viability of cells.Although this approach is very simple and is available for most bacteria, the heat conditions required for inactivating RM systems are not predictable and thereby may have to be determined by repeated trials.

Conclusion
In this chapter, RM systems are reviewed and some approaches to produce host-mimicking DNA are described.Analysis of chromosomal DNA using HPLC is a simple method to elucidate functional RM systems in target bacterium, and is therefore highly recommended for establishing bacterial transformation methods.When negligible DNA methylation is observed, methyl-free DNA is suitable for transformation of the bacterium.On the other hand, when significant DNA methylation is observed, a host-mimicking strategy involving methylation needs to be utilized.One weak point of this strategy is that there are no methods to exhaustively, readily, and rapidly determine methylation sites in target bacterium.When this analytical method becomes more widespread, this strategy will become a crucial technique for establishing efficient bacterial transformation methods.

Figure 1 .
Figure 1.Host-mimicking strategies for circumventing restriction-modification (RM) systems in bacteria.RM systems serve to defend bacteria against invasion by exogenous DNA, and thereby hamper genetic transformation by exogenous plasmids.In typical RM systems, restriction endonuclease (RE) digests exogenous DNA but not endogenous DNA that has been methylated by cognate DNA methyltransferase (MT).Host-mimicking DNA that imitates the methylation patterns of the bacterial host is recognized as endogenous DNA by the host because RM systems depend on hostspecific DNA methylation to distinguish between exogenous and endogenous DNA.

Figure 2 .
Figure 2. Chemical structures of modified nucleobases (modification moieties are indicated in red).

Table 1 .
General properties of four types of RM systems.Type I comprises R, M, and S subunits.Methylation is performed by subunits M and S. Type III comprises R and M subunits.The M subunit alone catalyzes methylation.Type II comprises two independent proteins, RE and MT.Type IV comprises only RE and restricts DNA with heterologous modifications.Methylation produces 6m A, 5m C, or 4m C.