ABSTRACT
The type I restriction and modification (R-M) enzyme from Salmonella enterica serovar kaduna (StySKI) recognises the DNA sequence 5'-CGAT(N)7GTTA, an unusual target for a type I R-M system in that it comprises two tetranucleotide components. The amino target recognition domain (TRD) of StySKI recognises 5'-CGAT and shows 36% amino acid identity with the carboxy TRD of EcoR124I which recognises the complementary, but degenerate, sequence 5'-RTCG. Current models predict that the amino and carboxy TRDs of the specificity subunit are in inverted orientations within a structure with 2-fold rotational symmetry. The complementary target sequences recognised by the amino TRD of StySKI and the carboxy TRD of EcoR124I are consistent with the predicted inverted positions of the TRDs. Amino TRDs of similar amino acid sequence have been shown to recognise the same nucleotide sequence. The similarity reported here, the first example of one between amino and carboxy TRDs, while consistent with a conserved mechanism of target recognition, offers additional flexibility in the evolution of sequence specificity by increasing the potential diversity of DNA targets for a given number of TRDs. StySKI identifies the first member of the IB family in Salmonella species.
The type I restriction and modification (R-M) systems of enteric bacteria, the first restriction systems to be characterised, comprise a diverse group of complex enzymes that possess both endonuclease and methyltransferase (MTase) activities (for reviews see 1 -7 ). The importance of these enzymes in vivo as a host barrier to phage infection is highlighted by the variety of phage-encoded anti-restriction systems specific to type I enzymes (3 ), but DNA breakage may also promote homology-dependent recombination between the incoming DNA and the bacterial chromosome (6 ,8 ).
The type I R-M systems are encoded by three genes, hsdR, hsdM and hsdS; the HsdS and HsdM subunits together form an active MTase which with the addition of the HsdR subunit forms the complete R-M system (1 -5 ). These enzymes bind to characteristic bipartite DNA target sequences, typically a 3 bp component separated from a 4-5 bp component by a non-specific spacer of 6-8 bp (1 -5 ). The MTase transfers a methyl group from S-adenosyl methionine (AdoMet), the methyl donor and co-factor, to the N6 position of a specific adenine residue on each strand of the DNA target. DNA methylation by the type II Mtase M.HhaI has been shown to involve a base-flipping mechanism (9 ). The similarities seen between the predicted secondary structure of the AdoMet-binding domain of EcoKI (10 ) and those of M.HhaI (9 ) and other type II Mtases (11 ,12 ) suggest that base-flipping is a prerequisite to DNA methylation by EcoKI. Upon binding to unmethylated DNA targets, the R-M enzyme initiates the complex restriction reaction which involves ATP-dependent DNA translocation followed by double strand cleavage at a non-specific site distant from the enzyme-bound target sequence (13 -17 ).
Unlike type II systems, which have separate MTases and endonucleases, the multi-subunit complexes of type I systems are endowed with both activities. One subunit, HsdS, determines the DNA specificity of the system and consequently changes within the hsdS gene that confer a new specificity concomitantly affect both restriction and modification. In contrast, an alteration in the specificity of a type II R-M system requires changes in both the endonuclease and the MTase. The known type I systems of enteric bacteria have been divided into four families (IA, IB, IC and ID) on the basis of a number of characteristics, most notably high sequence conservation within but not between families (18 ). The hsd genes for members of the IA, IB and ID families are closely linked to the serB locus, at 98.5 minutes on the chromosome of Escherichia coli K-12, and they behave as alleles in tests dependent on genetic recombination (19 ,20 ). The IA and ID families have members from both E.coli and Salmonella, the IB family has a member from Citrobacter freundii in addition to those from E.coli (18 ,21 ,22 ). The majority of the type IC systems characterised to date are encoded by plasmids that are known to be transferred readily between E.coli and Salmonella (2 ).
Whilst the DNA and amino acid sequences of hsd systems within a family are highly conserved, the hsdS genes contain two large regions of sequence (~450 bp) which are usually quite different even in members of the same family of type I systems (22 -25 ), and new combinations of these so-called variable sequences within the hsdS gene generate different DNA specificities (24 -27 ). Each variable sequence encodes a domain that is responsible for recognising half of the DNA target (27 ), and these domains are referred to as target recognition domains (TRDs). The domain boundaries predicted from sequence comparisons are reinforced by analyses of the products of limited proteolysis of both EcoKI (29 ) and EcoR124I (30 ) Mtases. As new type I systems have been discovered and their DNA specificity determined, it has become clear that the amino acid sequences of TRDs may show significant similarity if an identical target is recognised. Two members of the IA family share a common trinucleotide component in their specific recognition sequence; EcoKI recognises 5'-AAC(N)6GTGC and StySPI recognises 5'-AAC(N)6GTRC. The amino TRDs which are responsible for the recognition of the 5'-AAC share 90% identity (31 ). Also, two members of the IB family EcoAI and EcoEI share a trinucleotide target sequence with StyLTIII of the IA family. These three amino TRDs, which all recognise the sequence 5'-GAG, have 40% amino acid identity, identifying possible regions within the TRD that are involved in DNA recognition (28 ).
Only a small number of amino acids within each TRD is expected to be critical for determining the DNA specificity of a target half-site. The remaining amino acids may serve a structural role for which some changes may be tolerated. Analyses of TRDs that recognise similar or identical DNA targets, and consequently have amino acid identities, should reveal regions of importance for specific interactions with DNA. This expectation has stimulated the search for new type I R-M systems. Thirty-seven of the ECOR collection of wild-type E.coli strains (32 ) have been screened using probes specific for members of the IA, IB or ID families, and 17 of them were positive (18 ). There has been no comparable screen of Salmonella species for type I systems, although sequence analysis of the characterised IA and IB family hsd genes implies horizontal transfer of type I R-M genes between E.coli and Salmonella (33 ). Bullas et al. (19 ) have screened many strains of Salmonella for the restriction of bacteriophages, and they identified 12 Salmonella strains with previously unknown restriction specificities. The hsd genes identified were found to be linked to the serB locus and included those from Salmonella enterica serovar kaduna (S.enterica ser kaduna) encoding the StySKI R-M system.
We have cloned the genes encoding the MTase component of StySKI, and the nucleotide sequences of the hsd genes identify StySKI as a member of the IB family. The amino TRD of the HsdSsubunit of StySKI shows 36% identity with a carboxy TRD sequence from an IC family member, EcoR124I, implying that these two target recognition domains might recognise the same DNA target. The target sequences determined by these two TRDs are not identical, but rather one is a degenerate complement of the other. The complementarity is consistent with the suggestion that these two TRDs orient inversely on their DNA targets.
LB4023 (19 ), an E.coli K-12 derivative in which the hsd region of the chromosome has been replaced with that of S.enterica serovar kaduna, was used as the source of the kaduna-specific hsd genes and as a [lambda]-sensitive strain proficient in kaduna-specific restriction. A derivative sensitive phage to M13 was made by the acquisition of F'kan from EH55 (34 ). Mutations were induced in phages by propagation in NM720, an hsd[Delta] mutD5 strain harbouring F'kan. The hsd[Delta] F' strain NM522 (23 ) was used as a standard strain for M13, and ED8654 (35 ) for [lambda] phages. NM782 is a derivative of NM522 in which the hsd region of E.coli K-12 has been replaced with that encoding EcoAI,and NM789 is a derivative of NM782 in which hsdS has been inactivated by a deletion.
The library of DNA fragments was made in the [lambda] vector EMBL3cos (36 ). The SalI fragment encoding the modification component of EcoAI was transferred from the [lambda]hsd phage described by Fuller-Pace et al. (37 ) to a derivative of EMBL3 (NM1249) in which the cI- allele has been replaced by cI857. This enabled lysogens to be made at 32o following integration of [lambda]hsd phages by homology-dependent recombination. A derivative of the [lambda]hsdMS phage was made in which a BamHI fragment containing the 5' half of the hsdS gene was replaced with the 5' half of the hsdS gene conferring the StySKI specificity. The hybrid hsdS gene was transferred to the chromosome of strain NM789 following the homology-dependent integration and excision of the genome of the [lambda]hsdcI857 phage. The specificity of the resulting strain (NM821) is determined by the chimaeric HsdS polypeptide comprising the amino TRD of StySKI and the carboxy TRD of EcoAI.
The plasmid pAB3 used as a probe specific for a sequence close to the hsd genes of S.enterica serovar kaduna includes DNA from immediately downstream of the hsdS gene of serovar typhimurium LT2 (20 ).
Restriction was estimated from the reduced efficiency of plating (e.o.p.) of unmodified phages on a restricting strain relative to that on a nonrestricting strain; modified phages plated with equal efficiency on both strains. Modification was assessed by the protection against the relevant restriction system e.g. StySKI (LB4023), EcoAI (NM782) or the hybrid derivative (NM821).
Deletion derivatives of [lambda] phages were selected on BBL agar supplemented with 0.2 mM EDTA (37 ). Deletion and substitution derivatives of [lambda] used to locate targets for either StySKI or the hybrid system included deletions in the b2 region (e.g. [lambda]b527, [lambda]b538 and [lambda]b221) and deletions or substitutions in the immunity region ([lambda]imm434, [lambda]imm21 and [lambda]cI[Delta]KH54). The coordinates for these deletions and substitutions are given in Appendix I of Lambda II (38 ).
Lysates of [lambda] and M13 were made on an hsd[Delta] strain (the mutD strain NM720 for [lambda] and NM789 for M13), and then transferred to the appropriate restricting strain, either LB4023 F'kan encoding StySKI, or NM821 specifying the hybrid R-M system. After five cycles of enrichment in which phages were grown on an hsd[Delta] strain and then challenged by transfer to a restricting strain, phages were isolated that were resistant to restriction.
These were as described in reference 33 . The -40 primer was supplied with the Amersham Sequence® kit, other oligonucleotides, purchased from Oswell DNA Service (University of Southampton), were as follows. 5'-CACTGACCGGAACGCCAA (T3941) and 5'-GCCAGAATATCCCTGCCA (T3942) were used to amplify and sequence the region of [lambda] to the right of gene J which contains a target for StySKI. 5'-GGCGAGCCTGGCTAACCG (T5482) and 5'-GTAAGCGCATTGGCCCGC (T5483) were used to amplify and sequence the region of [lambda] in gene O including a second target for StySKI. 5'-AATTGTCGANCTGCTCCGTTAGGC (T6731) and its complementary olignucleotide 5'-AATTGCCTAACGGAGCAGNTCGAC (T6732) were used to generate a putative target sequence for StySKI with a single degeneracy at position 4 of the target sequence. 5'-AATTGACCKATACTGAGAGTTACGC (T10012) and its complementary oligonucleotide 5'-AATTGCGTAACTCTCAGTATMGGTC (T10013) generated a putative target sequence for StySKI, but with a single degeneracy at position 2. Both pairs of complementary oligonucleotides have 5' extensions that base pair with vector cut with EcoRI.
Nucleotide sequence of M13 single-stranded DNA was obtained using the di-deoxy chain termination sequencing method (39 ) with 5-[[alpha]-35S]thiotriphosphate (40 ) using reagents and methods as recommended in the Amersham Sequenase® sequencing kit. The nucleotide sequence of PCR products were obtained using the Amersham PCR product sequencing kit; the PCR products were treated with alkaline phophatase and exonuclease III to remove dNTPs and primers respectively. The enzymes were inactivated by incubation at 80oC before the sequencing reactions were carried out. The sequence comparisons and alignments used the GAP and PILEUP programs of the Wisconsin Package, Version 8 of the Genetics Computer Group, WI, USA.
A derivative of E.coli K-12 (LB4023) in which the resident hsd genes encoding EcoKI have been replaced by those from S.enterica ser kaduna (19 ) was used as the source of bacterial DNA, and as a [lambda]-sensitive strain restricting with the specificity of StySKI. A library of large DNA fragments, generated by digesting the DNA of LB4023 with Sau3A, was made in a [lambda] replacement vector. Probes specific for either type IA or IB systems had failed to identify hsd genes in digests of DNA isolated from LB4023 (41 ), therefore a DNA fragment from immediately downstream of the hsdS gene of S.enterica ser typhimurium was used as a probe to identify [lambda] clones with inserts from the homologous region of the chromosome of LB4023 (20 ).
Three independent clones were isolated and the genome of one of the three ([lambda]4023) was found to be resistant to restriction by LB4023, consistent with the modification of its target sequences for StySKI. [lambda]4023 was predicted to include the genes encoding the modification component of StySKI, and was the basis of all subsequent analyses.
Deletion derivatives of [lambda]4023 were selected by isolating plaques on medium containing the chelating agent EDTA. The modification phenotypes of the deletion derivatives were determined. Some derivatives were sensitive to restriction by LB4023, as would be expected if hsdM or hsdS were inactivated by a deletion. DNA preparations from the parental phage and deletion derivatives were analysed by restriction digests. The deletion mutants localise hsdM and S to the left-hand part of the restriction map (Fig. 1 ). Each of the two adjacent SalI fragments (2.3 and 2.6 kb) located in this region was transferred to M13mp18 and mp19 and the nucleotide sequence of the bacterial DNA was determined.
The sequence obtained (EMBL database accession no. Y11005) includes three ORFs. The amino acid sequences predicted from these ORFs are consistent with the identification of the hsdM and hsdS genes preceded by 660 bp of the 3' end of hsdR. The hsdM gene and the 3' end of hsdR share 70-95% identity with the hsdM and hsdR genes encoding EcoAI and EcoEI, consistent with S.enterica ser kaduna having a type IB R-M system. The hsdS gene includes the conserved regions characteristic of the type IB family, whilst two regions predicted to encode the two TRDs are not conserved, as expected if they are responsible for the recognition of different DNA targets. The hsd genes of members of the same family have always been sufficiently similar to be identified with family-specific DNA probes, this is so for the cloned StySKI hsd genes (data not shown), despite poor hybridisation of DNA in chromosomal digests (41 ).
The predicted amino acid sequences of the subunits of StySKI are compared with those of other type I R-M enzymes in Table 1 . The high levels of sequence identity found in comparisons between each of two members of the IB family contrasts with the low levels of identity with members of each of the three other families of type I R-M enzymes. The figures are consistent with those of previous comparisons within and between families (33 ,42 ).
Table 1
An unexpected similarity was detected between the predicted HsdS sequences of StySKI and EcoR124I (an IC family member). The amino TRD sequence of StySKI shows 36% identity with the carboxy TRD sequence of EcoR124I. This is the first report of extensive similarity between an amino TRD sequence and a carboxy TRD sequence. Similarity of TRD sequences is not usual unless they recognise the same sequence. The level of identity of these TRDs is comparable to the 40% detected for two amino TRD sequences responsible for the recognition of 5'-GAG; EcoAI of the IB family and StyLTIII of the IA family.
In order to determine whether the identity between the EcoR124I carboxy TRD and the StySKI amino TRD is functionally significant it was essential to determine the target sequence recognised by StySKI. Deletions and substitutions were used to identify and localise the targets for StySKI in the genome of phage [lambda]. Unmodified phages were tested for their sensitivity to restriction by LB4023. Wild-type [lambda] has an e.o.p. on LB4023 of ~10-3. Some deletions in the b2 region of [lambda], including b538 and b527, had no effect on the e.o.p.; others in which the deletion extended closer to gene J, including b189 and b221, resulted in an increased e.o.p. (10-1 to 10-2). Similarly, substituting imm434 for imm[lambda] had no effect, but an imm21 substitution increased the e.o.p. (10-1 to 10-2). These results are consistent with one StySKI target in the b2 region close to the J gene and a second in the immunity region. In subsequent genetic crosses it was found that not all imm21 phages had an increased e.o.p., a finding explained by the final location of the second target close to, rather than within, the immunity region.
More precise positions of the predicted target sequences were identified by mutations within the [lambda]b527 genome which conferred resistance to attack by StySKI. Mutations were induced by growth of the phage in a mutD strain (NM720), to increase the chance that restriction targets were lost as the result of base substitutions rather than deletions. [lambda]b527 was chosen as the starting phage since it lacks a normal attachment site and is therefore less likely than [lambda]+ to yield deletions as the result of Int-mediated recombination. Mutant phages were isolated following multiple cycles of enrichment for phages with increased e.o.p. on LB4023 (see Materials and Methods). Two classes of mutants were isolated: those that had an increased e.o.p. on LB4023 when compared with the original phage (e.o.p. of 10-1 to 10-2 rather than 10-3), and those that were completely resistant to restriction by StySKI, as indicated by an e.o.p. of 1 on LB4023. Mutants that had lost their sensitivity to LB4023 restriction were presumed to have lost their targets for StySKI. The simple interpretation of the results is that [lambda]b527 has two targets for StySKI; phages with an e.o.p. of 10-1 to 10-2 retain one target and those with an e.o.p. of 1 have lost both. A 1.8 kb SmaI-EcoRI fragment from the left-hand end of the b2 region of a [lambda]b527 derivative resistant to restriction by StySKI was cloned in M13mp19 and the nucleotide sequence of the insert determined. A single base change of adenine to guanine was detected at position 20936 in the nucleotide sequence of [lambda] (37 ,39 ; GenEMBL sequence data base accession no. J202495).
Oligonucleotide primers were used to amplify, by the polymerase chain reaction (PCR), a segment of the [lambda] genome that includes the base-pair identified by mutation. The nucleotide sequences of the amplified DNA from each of five restriction-resistant derivatives of [lambda]b527 identified two additional base changes (Target 1 in Fig. 2 a). A putative target sequence of 5'-CGAT(N)7GTTA was deduced as the only sequence to occur twice in [lambda] that is compatible with the three mutations identified in the first target sequence; the other occurrence of 5'-CGAT(N)7GTTA is close to the immunity region in gene O (bp 39823-39837). The BglII-SmaI fragment that includes this second putative target sequence (Target 2 in Fig. 2 a) was transferred from a restriction-resistant derivative of [lambda]b527 to M13mp18. The nucleotide sequence indicated a single change within the predicted target sequence. Oligonucleotide primers were chosen to amplify the region of [lambda] identified by this mutation, and the sequences of products derived from five restriction-resistant derivatives of [lambda]b527 were determined. Three different mutations were detected (Fig. 2 a). The predicted target sequence, CGAT(N)7GTTA, was changed in all the restriction-resistant derivatives of [lambda]b527. Degeneracies at all but the second and fourth positions within the bipartite, octameric sequence require more than two targets in the [lambda] genome. The target sequence compatible with the data is 5'-CKAB(N)7GTTA, where K is either G or T and B is T, C or G.
Each natural type I R-M system studied prior to StySKI recognizes a target sequence that includes a trinucleotide component specified by the amino TRD of the HsdS subunit, and by convention the target sequence is written beginning with the component specified by the amino TRD. The target site of StySKI differs in that it comprises two tetranucleotide components and the data presented do not orient the TRDs with respect to the two components of the target sequence. Formally, the target sequence for StySKI could be either 5'-CGAT(N)7GTTA or 5'-TAAC(N)7ATCG. The component of the target sequence specified by the amino TRD of StySKI must be established. This was done by the analysis of a hybrid system.
A chimaeric HsdS subunit was made in which the amino TRD of the EcoAI system was replaced with that of the StySKI system. This was achieved in a [lambda]hsd phage encoding the modification component of EcoAI, by replacing the BamHI fragment encoding the amino TRD of EcoAI with that encoding the amino TRD of StySKI. The resulting phage no longer encoded the modification system of EcoAI, but conferred a new specificity. Lysogens of this phage in a strain (NM789) that retains hsdR and hsdM of EcoAI,but has a deletion within hsdS, were restriction-proficient. A lysogen was cured of its prophage and a strain (NM821) that had replaced the hsdS[Delta] with a hybrid hsdS gene was used to deduce the target recognised by the hybrid R-M system.
If the amino TRD of the new system recognizes the sequence 5'-CGAT and the spacing of N=7 is retained, the predicted specificity of the chimaeric subunit is 5'-CGAT(N)7GTCA, where GTCA is the component specified by the carboxy TRD of EcoAI.
This target sequence occurs once within the sequence of M13mp18 or -19 and twice within that of phage [lambda], where one of the two targets is within the immunity region in DNA deleted by the mutation cI[Delta]KH54. Consistent with the predictions, both wild-type [lambda] and M13mp18 were restricted by the hybrid system. Furthermore, the e.o.p. of wild-type [lambda] was less than that for either [lambda]cI[Delta]KH54 or [lambda]imm434, as expected if one target is within the immunity region. This distribution of targets in [lambda] and M13mp18 is inconsistent with a spacing of N=6 rather than N=7, and with the alternative combinations of target sequences. Nevertheless, a derivative of M13mp18 was selected that was resistant to restriction by the hybrid system, and the sequence of this mutant M13 was checked in the region of the predicted target sequence. The three restriction-resistant derivatives of M13mp18 tested had a mutation in the 3' component of the target sequence; 5'-GTCA was changed to 5'-GTTA. This mutation confirms the predicted target sequence for the hybrid system and consequently defines the orientation of the components of the target sequence for StySKI. Concomitantly, conversion of the target sequence to 5'-CGAT(N)7GTTA creates a new target sequence for StySKI. As expected, the mutant phage has acquired sensitivity to restriction by StySKI.
The StySKI type I R-M system is the first representative of the IB family to be identified in Salmonella. Both E.coli and Salmonella are now shown to have members in each of the three families of allelic hsd genes. Hybrid hsdS genes have been isolated for both the IA (25 ,27 ) and IC (24 ) families of enzymes and have been shown to confer the predicted new specificies (24 ,26 ). Hybrid hsdS genes generated from members of the IB family have not been described previously; the one reported in this paper functions with the anticipated specificity.
The amino TRD sequences of the EcoAI and StyLTIII share only 40% amino acid identity whilst they both recognise the same DNA target (5'-GAG), implying that many amino acids within the TRD are not directly involved in determining DNA specificity. On the assumption that both domains recognise their DNA target by the same mechanism, the residues critical for specific DNA recognition will be amongst the conserved residues. An analysis of point mutations within the 5' variable region of the EcoKI hsdS gene has shown that many changes have no effect on EcoKI activity; currently those that abolish the restriction and modification activity are confined to a short segment of the TRD (M.O'Neill, personal communication). It is possible that the TRDs impart nucleotide specificity in a similar way to the type II endonucleases and MTases (9 ,44 -47 ). For these enzymes, structural studies demonstrate that short loops of the polypeptide extend into the DNA helix to make base-specific contacts.
The amino TRD sequence of StySKI is 36% identical to the carboxy TRD of EcoR124I (Fig. 3 ), implying that these two TRDs are related despite their different positions within the HsdS subunit and their presence in enzymes from different families. The level of identity is similar to that found for the amino TRDs of EcoAI and StyLTIII but is considerably greater than the 20% identity between the carboxy TRD sequences of EcoKI and StySPI, which also recognise a similar but degenerate target sequence. Although 20% identity seems a low figure for two TRDs with a similar target sequence, the carboxy TRDs of EcoKI and StySPI were the most similar of the 10 comparisons that could be made at that time (27 ).
The chromosomally encoded hsd genes can be transferred between strains by conjugation and by transduction. The type I R-M systems also have plasmid-borne members which are readily transferred between bacterial hosts. Type I R-M systems with new specificities can arise spontaneously in vivo (19 ,24 ,54 -56 ). Experiments have shown that the sequences encoding TRDs can be switched between hsd genes of the same family to create new combinations of TRDs, and consequently enzymes with new specificities (24 -27 and the data presented here for the IB family). Homology-dependent recombination can generate new combinations of TRDs within a family of enzymes, but it seems less likely that a sequence encoding a TRD can be transferred between different families of hsd genes in this way, since there is little DNA sequence identity. Similar TRD sequences in different families of enzymes may reflect their existence prior to divergence of the hsd genes into separate families. Alternatively short conserved sequences that flank the variable regions in some hsdS genes may identify the remnants of sites associated with the homology-dependent recombination a considerable time ago, or illegitimate recombination events could be involved in the transfer of segments of DNA coding for TRDs. If the variable regions encoding TRDs can be transferred not only between hsdS genes in different families, but also from the 5' to 3' end of a gene and vice versa, clearly the scope for varying specificity is increased.
Repeated nucleotide sequences, particularly within the hsdS genes of the IB and IC families, support the idea that present hsdS genes have been derived by a gene duplication event (22 -24 ,52 ), in which case the carboxy TRD sequence of EcoR124I may result from the duplication of an ancestral hsdS gene common to the type IB and IC families of R-M systems. Regardless of the evolutionary relationship, our data from natural enzymes indicate that the variable regions have the potential to encode functional TRDs irrespective of their orientation within the subunit. Such an adaptable DNA binding domain extends the repertoire of new specificities that can be generated from the available TRDs.
We thank our colleagues for their helpful discussions and advice, particularly Lynn Powell and David Dryden for constructive comments on the manuscript and Annette Titheradge for the genomic library. We are indebted to Karen Witherspoon for preparing the manuscript and the Medical Research Council for support of the research and a studentship (P.H.T.).
Enzymes
Family
HsdR
HsdM
HsdS
EcoAI
IB
90
97
55
EcoEI
IB
72
91
50
EcoKI
IA
23
30
21
EcoR124I
IC
21
24
30
StySBLI
ID
14
28
17
REFERENCES
Return