ABSTRACT
The genomic region encoding the type IIS restriction-modification (R-M) system
Hph
I (enzymes recognizing the asymmetric sequence 5
'
-GGTGA-3
'
/5
'
-TCACC-3
'
) from
Haemophilus parahaemolyticus
were cloned into
Escherichia coli
and sequenced. Sequence analysis of the R-M
Hph
I system revealed three adjacent genes aligned in the same orientation: a
cytosine 5 methyltransferase (gene
hphIMC
), an adenine N6 methyltransferase (
hphIMA
) and the
Hph
I restriction endonuclease (gene
hphIR
). Either methyltransferase is capable of protecting plasmid DNA
in vivo
against the action of the cognate restriction endonuclease.
hphIMA
methylation renders plasmid DNA resistant to R.
Hind
III at overlapping sites, suggesting that the adenine methyltransferase modifies
the 3
'
-terminal A residue on the GGTGA strand. Strong homology was found between
the N-terminal part of the m6A methyltransferasease and an unidentified reading
frame interrupted by an incomplete
galE
gene of
Neisseria meningitidis
. The
Hph
I R-M genes are flanked by a copy of a 56 bp direct nucleotide repeat on each
side. Similar sequences have also been identified in the non-coding regions of
H.influenzae
Rd DNA. Possible involvement of the repeat sequences in the mobility of the
Hph
I R-M system is discussed.
The majority of restriction endonucleases operate within short symmetrical
nucleotide sequences (
1
). A subclass of the type II enzymes, so-called type IIS ENases (
2
) recognize asymmetrical nucleotide sequences and cleave DNA at a specified
distance from it. The spatial separation of the DNA recognition and cleavage
sites suggests that the type IIS ENases are likely organized as two-domain enzymes, where one domain is responsible for DNA recognition and
the other one carries the catalytic center that makes a double strand break in
DNA. This suggestion was supported experimentally by analysis of the domain
structure of the
Fok
I restriction endonuclease (
3
). This finding opens up new prospects for efforts to change the specificity of
restriction enzymes that might be more feasible than in the case of type II
restriction endonucleases, which are organized as single domain homodimeric
proteins (
4
). However, until now the sequences of only three type IIS restriction-modification (R-M) systems,
Fok
I,
Mbo
II and
Sts
I (
5
-
7
), have been published. Here we report cloning and sequence analysis of the
genomic region encoding the type IIS R-M system
Hph
I from
Haemophilus parahaemolyticus
(EMBL accession no. X85374). This R-M system recognizes an asymmetrical sequence, 5'-GGTGA-3'/5'-TCACC-3'. The restriction
endonuclease makes a staggered cut at the eighth base pair downstream of its
recognition sequence on the upper strand, producing single base 3' protruding ends (
8
,
9
). Earlier studies of this system reported the identification of only one type
of methylation within the recognition sequence, namely modification of the
inner C on the lower strand, 5'-T
m5
CACC-3' (
10
). Our results provide evidence for the presence of a second methylase that
modifies an A base within the complementary strand, yielding 5'-GGTG
m6
A-3'.
Haemophilus parahaemolyticus
(ATCC no. 49700) was used as a source of DNA for cloning of the
Hph
I R-M system.
Escherichia coli
strain ER2267 of genotype
e14
-
(
mcrA
-
)
endA1 supE44 thi-1
[Delta](
mcrC-mrr
)
114
::
IS10
[Delta](
argF-lac
)
U169 recA1/ F
'
proA
+
B
+
lacIq
[Delta](
lacZ
)
M15 zzf
::mini-Tn
10
(Km
R
) was used as host for all cloning and subcloning procedures.
Citrobacter freundii
RFL9 (Ap
R
Cm
S
) isolated in this laboratory is a wild-type strain producing the
Cfr
9I restriction endonuclease (
11
). It was used as host for the construction and propagation of the positive
selection vector pBR-R. The T7 expression system (
12
) employed for expression of R.
Hph
I was purchased from Novagen and includes
E.coli
BL21(DE3) (
E.coli
B strain
F
-
ompT
[
lon
]
hsdS
B
r
B
-
m
B
-
with a [lambda] prophage DE3 carrying the T7 RNA polymerase gene) and plasmid pET-21b. Plasmids pBR322 (
13
), pBR329 (
14
), pUC19 (
15
) and pACYC184 (
16
) were used as vectors in cloning and subcloning experiments. pCfr9IRM2.3 (
17
) was used as the source of the gene of the
Cfr
9I restriction endonuclease. Phage [lambda]
vir
stocks were prepared according to Sambrook
et al
. (
18
).
Haemophilus parahaemolyticus
cells were grown as described by Kleid
et al.
(
8
).
Escherichia coli
and
C.freundii
cells were grown in LB medium containing ampicillin (Ap, 60 [mu]g/ml), kanamycin (Km, 50 [mu]g/ml) and chloramphenicol (Cm, 30 [mu]g/ml) as required. Cells were transformed using the CaCl
2
-heat shock method (
18
). Transformants were selected by plating onto LB agar supplemented with
appropriate antibiotics.
Restriction enzymes, T4 DNA ligase, DNA polymerase I large fragment (Klenow),
Bal31, bacterial alkaline phosphatase (BAP), the ExoIII/S1 Deletion Kit, the
DNA Labeling Kit (version 2.0),
Hin
dIII or
Eco
RI-digested and BAP-dephosphorylated pBR322 and oligonucleotides were products of MBI
Fermentas. The DNA Sequencing Kit was purchased from Pharmacia. All enzymes and
kits were used according to manufacturers' recommendations. [[alpha]-
33
P]dATP was obtained from Izotop (St Petersburg, Russia).
Haemophilus parahaemolyticus
genomic DNA was extracted and purified as described by Marmur (
19
). Plasmid DNAs were prepared by the alkaline lysis procedure (
20
) and were further purified by binding to glass powder (
21
). Restriction and deletion mapping, agarose gel electrophoresis, isolation of
individual DNA restriction fragments from agarose gels, subcloning of DNA
fragments and Bal31 deletions were carried out by standard procedures (
18
). ExoIII/S1 deletions were performed following the manufacturer's
recommendations.
The 1.2 kb
Bst
XI-
Bam
HI fragment of pCfr9IRM2.3 (
17
), containing the
cfr9IR
gene, was blunt-ended with Klenow fragment and purified by electrophoresis in an agarose
gel. It was ligated into the pBR329 vector, which was digested with
Eco
32I and
Eco
88I, blunt-ended with Klenow fragment and dephosphorylated with BAP. The ligation mix
was used to transform competent
C.freundii
RFL9 cells. Plasmids were isolated from Cm
R
transformants and screened with restriction endonucleases for inserts of
appropriate size and orientation.
Haemophilus parahaemolyticus
genomic DNA was fragmented by sonication. The sonication products were
fractionated on a low melting point agarose gel and 1-8 kb fragments were isolated. Aliquots of 15 [mu]g of the recovered DNA fragments were blunt-ended with Klenow fragment and ligated with T4 DNA ligase at 8oC for 24 h in a 100 [mu]l reaction volume to 5 [mu]g
Eco
47III-cleaved, dephosphorylated pBR-R vector DNA. The ligation mixture was used to transform competent
E.coli
ER2267 cells. Plasmid DNA (10 [mu]g) isolated from the pooled 10 000 Ap
R
Km
R
transformants was digested with excess
Hph
I (30 U for 3 h) and then transformed back into ER2267. Plasmid DNA of the
resulting 19 individual transformants was then screened for resistance to R.
Hph
I digestion.
Electrophoresis of
H.parahaemolyticus
DNA fragments obtained after digestion with various restriction enzymes (single
and double digests) on agarose gels and Southern transfer were performed as
described (
18
). A DNA Labeling Kit was used for the preparation of two radioactive DNA probes
(1.05 kb,
Eco
32I-
Eco
RI and 1 kb,
Eco
RI-
Bgl
II fragments) from the previously cloned 2 kb fragment which encodes the
functionally active
Hph
I methyltransferase (MTase) (Fig.
1
B). Fragments overlapping with the 2 kb insert were identified and used to
construct a partial genomic map. The cloned
hphIM
gene mapped to a 7.2 kb
Hin
dIII fragment of the
H.parahaemolyticus
genomic DNA, which seemed large enough to contain the complete R-M system. This fragment was cloned into pBR322 as follows. Genomic DNA was
digested with
Hin
dIII and the reaction products resolved on an agarose gel, where 6-8 kb DNA fragments were isolated. Aliquots of 3 [mu]g of the fragments were ligated at 16oC for 16 h in a 50 [mu]l reaction volume with 3 [mu]g
Hin
dIII-cleaved and BAP-dephosphorylated pBR322. Plasmid DNA isolated from the 5000 pooled
Ap
R
Km
R
transformants was subjected to digestion with R.
Hph
I, followed by re-transformation back into ER2267. Eighty five transformants were obtained,
which were tested for resistance to phage infection
in vivo
(see below). Twenty four of these clones were picked randomly for individual
screening of their plasmid DNA with the
Hph
I restriction endonuclease
in vitro
.
To determine restriction activity of individual clones
in vivo
, plating efficiency of [lambda]
vir
on these clones was measured and compared with that of
E.coli
ER2267 cells carrying the pBR322 vector alone (
22
). Screening of restriction-proficient cells was carried out by replicating the transformants onto top-layer agar containing 10
6
phage particles per plate. The endonuclease activity
in vitro
was tested by incubating serial dilutions of cell-free extracts prepared as described (
23
) with 1 [mu]g [lambda] DNA at 37oC for 1 h in a 40 [mu]l reaction mixture containing 10 mM Tris-HCl, pH 7.5 at 37oC, 10 mM MgCl
2
and 0.1 mg/ml bovine serum albumin. Reaction products were analyzed by
electrophoresis in 1% agarose gels. To determine the
Hph
I-specific modification
in vivo
, plasmid DNA was challenged with an excess of the
Hph
I restriction endonuclease followed by agarose gel electrophoresis.
To test for methylation by the
hphIMA
gene product, we first designed a plasmid, pBR-H, which contains the
Hph
I recognition sequence (bold letters) overlapping by 1 bp with the recognition
sequence of
Hin
dIII (boxed). The double-stranded linker
was introduced into
Sal
I/
Pae
I-digested pBR322. The resulting construct was verified by DNA sequencing.
To determine the effect of
Hph
I methylation on cleavage of the overlapping
Hin
dIII site by R.
Hin
dIII, the plasmid pBR-H was transformed into
E.coli
ER2267 harboring plasmid pAC-HphIA3 (Fig.
2
). The plasmid DNA isolated from Ap
R
Cm
R
Km
R
transformants was tested with
Hph
I and
Hin
dIII endonucleases.
To provide the genes for
Hph
I methylases
in
trans
, these genes (
hphIMA
and
hphIMC
) were subcloned into a compatible vector, pACYC184. The 2.75 kb
Psp
1406I-
Vsp
I fragment of pHphIMM1, containing both methylase genes, was isolated (Fig.
2
) and blunt-ended by treatment with Klenow polymerase. The fragment was ligated to
Eco
32I-digested and dephosphorylated pACYC184. After transformation of
E.coli
ER2267 several of the Cm
R
Km
R
transformants were tested for the presence of
Hph
I-specific methylation and for the desired orientation of the subcloned
fragment. The resulting 7.0 kb plasmid, designated pAC-HphIMM, was introduced into the recipient cells and was expressed in
trans
during subsequent endonuclease expression experiments. Subcloning of the
hphIR
gene into the expression vector pET-21b was carried out as follows. A 1.3 kb
Taq
I fragment of pHphIMM1 was gel purified, treated with Klenow fragment and
ligated to
Xba
I-digested, blunt-ended and dephosphorylated pET-21b. The ligate was used to transform
E.coli
ER2267 cells that contained the methylation-proficient plasmid pAC-HphIMM. Individual Ap
R
Cm
R
Km
R
clones were screened for the correct size and orientation of the
hphIR
gene. One isolate (pET-R1) was used in the expression experiments. Expression of the
hphIR
gene was induced by adjusting the culture [
E.coli
BL21(DE3) pre-transformed with plasmids pET-R1 and pAC-HphIMM; see Fig.
2
] to 1 mM isopropyl-[beta]-D-thiogalactoside (IPTG) at an OD
600
of ~0.5. After 3 h induction the cells were harvested by centrifugation and
lysed (
23
). Crude cell extracts were used for determination of R.
Hph
I activity.
DNA sequencing was done in both directions from a series of nested deletions
generated using the ExoIII/S1 or Bal31 methods (
24
). Sequencing reactions were carried out using a DNA Sequencing Kit, [[alpha]-
33
P]dATP, M13/pUC (direct, reverse) or pBR322 (
Sal
I site, ccw) standard sequencing primers and double-stranded, supercoiled plasmid DNA as templates. The reaction products were
resolved by electrophoresis on wedge-shaped polyacrylamide gels.
The comparison of deduced amino acid sequences with the EMBL and GenBank
databases translated in six reading frames was done using the BLAST (
25
) program to identify similar, potentially related sequences.
The lethality of a restriction endonuclease that is expressed without its
accompanying methylase has been used as the basis for the construction of
positive selection systems in several laboratories (
26
,
27
). The pBR-R vector described in this work (Fig.
1
A) permits cloning of any blunt-ended fragments (
Eco
47III site) as well as fragments generated by the restriction endonucleases
Bam
HI,
Bcl
I,
Bgl
II,
Bst
YI,
Mbo
I,
Sau
3A (
Bgl
II site),
Nla
III,
Nsp
I or
Pae
I (
Pae
I site). Our system possesses certain advantages over the other known systems:
(i) propagation of this plasmid in
C.freundii
RFL9, which lacks any natural plasmids, does not require subsequent separation
of a companion methylase-producing plasmid, as in the case of pKG2 (
27
); (ii) pBR-R is considerably smaller than the positive selection vectors pLV57 and
pLV59 and it does not require a temperature shift for inactivation of a
companion methylase, as do pLV57 and pLV59 (
26
); (iii) despite of its similar size to the vectors pKGW and pKGS (
27
), two antibiotic resistance genes are located on pBR-R (Ap and Cm). Expression of the
cfr9IR
gene from the constitutive P
tet
promoter allows transformation of common
E.coli
strains and, in contrast with pKGW and pKGS, requires no addition of IPTG (
27
). The survival frequency of cells transformed with this plasmid was found to be
10
-4
as compared with that of cells transformed with a control vector, pBR329. The
efficiency of pBR-R as a cloning vector was demonstrated by the construction of a
H.parahaemolyticus
library using DNA fragments generated by sonication.
Selection of the gene coding for M.
Hph
I was based on the resistance of self-modifying recombinant plasmids to digestion by R.
Hph
I (
28
). Nineteen transformants obtained after the selection procedure described in
Materials and Methods were further screened for the presence of the
Hph
I-specific modification. Eight modification-positive clones contained the same plasmid with a 2 kb insert
(designated pHphIM2.0; Fig.
1
B). None of these clones showed any
Hph
I endonuclease activity as assayed both
in vivo
and
in vitro
. The 2.5 kb
Eco
32I fragment from pHphIM2.0 was subcloned into pUC19 for sequence determination
(not shown). Sequence analysis revealed the presence of a complete ORF
potentially encoding a m6A methylase. This conclusion was based on the analysis
of conserved amino acid motifs typical for DNA MTases (
29
). The ORF corresponding to the presumed m6A methylase is flanked by two
truncated ORFs. The upstream ORF (Fig.
1
B) contained conserved motifs VI-X characteristic of m5C methylases (
30
), while the downstream ORF did not show any similarities with known protein
sequences. As restriction endonucleases usually share little conservation, we
assumed that the latter ORF might encode the N-terminal part of the
Hph
I ENase. It therefore seemed likely that the entire
Hph
I R-M system is located in DNA regions extending in both directions around the
cloned fragment.
To select for a DNA fragment large enough to include the entire m5C methylase
and
Hph
I restriction endonuclease genes a restriction map of the
H.parahaemolyticus
genomic R-M locus was determined (not shown). Two DNA fragments prepared from
pHphIM2.0 (1.05 kb
Eco
32I-
Eco
RI and 1 kb
Eco
RI-
Bgl
II) were used as probes in Southern hybridization. The 7.2 kb genomic
Hin
dIII fragment was cut with R.
Eco
RI into two pieces. The 2.2 kb subfragment hybridized with the 1.05 kb probe,
while the 5.0 kb subfragment hybridized with the 1 kb probe (not shown). This
indicated that the
Hin
dIII fragment includes the cloned 2 kb region of pHphIM2.0 and flanking
sequences. Cloning of the 7.2 kb
Hin
dIII fragment and selection of R.
Hph
I-resistant clones was carried out as described in Materials and Methods.
Eighty five transformants were analyzed, but none of them demonstrated
resistance to [lambda]
vir
infection. Nine out of 24 clones tested were resistant to cleavage
in vitro
with R.
Hph
I. Restriction mapping revealed the presence of two types of recombinant
plasmids that differed in the orientation of the same insert (designated
pHphIMM1 and pHphIMM2; Fig.
2
). These clones were tested for endonuclease activity both
in vivo
, using the plating efficiency test (Materials and Methods) and
in vitro
. In neither case was endonuclease activity detected, indicating that the R.
Hph
I gene was either incomplete or inactive in
E.coli
cells. To analyze the regions surrounding the MTase genes, sequencing of the
4.8 kb
Hin
dIII-
Mlu
I fragment from pHphIMM1 and pHphIMM2 (Fig.
2
) was carried out.
A 4790 bp region of the cloned 7.2 kb
Hin
dIII DNA fragment encompassing the genes for
Hph
I MTases as well as adjacent regions was sequenced on both strands. Five ORFs
(three complete, one truncated and one partially sequenced), all oriented in
the same direction, were identified (Fig.
2
). The first ORF (ORFX), 258 bp long (nt 3-260, termination codon included) is truncated and may represent the 3'-terminal part of an unknown gene. The next three ORFs encode
the
Hph
I R-M enzymes: ORF2 (
hphIMC
), m5C MTase; ORF3 (
hphIMA
), m6A MTase; ORF4 (
hphIR
),
Hph
I restriction endonuclease (see next paragraph). The position of a translation
initiation codon in
hphIMC
could not be unambiguously inferred from the nucleotide sequence alone due to
the presence of four potential initiation sites upstream of the first conserved
motif of the m5C MTase (
30
) (ATG at position 562, ATG at position 634, TTG at position 640 and ATG at
position 688); none of the start codons is preceded by a putative Shine-Dalgarno sequence either. This ORF (
hphIMC
) ends with translation termination codon TAA at position 1680. ORF3 (
hphIMA
) is 1011 bp long and extends from nt 1673 to 2683 (336 amino acids). This ORF
overlaps by 8 nt with the preceding ORF and by 1 nt with the downstream one.
The fourth ORF, 1140 bp long (nt 2683-3822, the
hphIR
gene), encodes the
Hph
I ENase of 379 amino acid residues. The last ORF (
menB
) was not completely sequenced; it extends from nt 3983 to 4789 and represents
the 5'-terminal part of the
menB
gene (see below). A putative Shine-Dalgarno sequence was found only upstream of the last gene,
menB
(AAGGA at positions 3969-3973).
Further analysis of the sequence revealed two copies of a 56 bp stretch that
flanked the
Hph
I R-M system on both sides (Fig.
3
). Restriction-modification systems, although not essential for cell viability, are
widespread in bacteria (
1
). Dissimilar systems are often present in related bacterial strains and,
vice versa
, homologous systems can be found in taxonomically distant microorganisms (
17
,
31
,
32
). These observations suggest that many R-M genes may have enjoyed a certain degree of mobility, at least in the
recent evolutionary past. Our finding of a 56 bp long direct repeat flanking
the
Hph
I R-M genes may be important in this context. Its presence suggests models for
R-M mobility involving recombination of a circular DNA containing the 56 bp
sequence and the
Hph
I R-M gene cassette with a corresponding (although somewhat divergent) 56 bp sequence in the chromosome. One class of models
invokes site-specific recombination, such as with a temperate phage (i.e. [lambda]) (
33
) and with the 56 bp sequence constituting (part of) the attachment site; a
related model would be one based on `integrons', a recently discovered class of
mobile DNA segments that underlie site-specific insertion of certain antibiotic resistance genes into new
plasmids or transposons (
34
). There is no obvious similarity between the
Hph
I 56 bp repeat element and the 59 bp repeats of currently known integrons
however and thus any possible `integron' involving the
Hph
I R-M system would differ in specificity from the drug resistance integrons
described to date. It is of course formally possible that the 56 bp repeats are
not components of a site-specific recombination system. Rather, since the repeats exceed the
minimum length needed for RecA protein-mediated DNA pairing (
35
), this suggests generalized recombination for both insertion of the
Hph
I R-M genes into particular chromosomal sites or their excision and loss from
the cell. Of note, two sequences highly related to the
Hph
I repeats (Fig.
3
) are present in intergenic regions of
H.influenzae
strain Rd, whose genome has been entirely sequenced (
36
).
The deletion and subcloning experiments were carried out to investigate the
expression of
Hph
I R-M genes (Fig.
2
). We found that both MTases are expressed in
E.coli
and protect plasmid DNA against R.
Hph
I when cloned together or subcloned separately. Moreover, a similar degree of
DNA protection was detected when the plasmids carrying the MTase genes in
different orientations were isolated, treated with R.
Hph
I and analyzed. These findings indicate that the endogenous promoters operate
efficiently in
E.coli
. No activity of the restriction endonuclease was detected in crude extracts
from cells carrying the initial plasmids, pHphIMM1 and pHphIMM2. We subcloned
the presumed
hphIR
gene (ORF4) into the expression vector pET-21b using cells harboring the methylase genes in
trans
as recipient. In this system, the restriction endonuclease activity was readily
detectable upon induction with IPTG, indicating that translation initiation
functions normally and that there are no strong transcription termination
signals immediately upstream of the
hphIR
gene. Apparently, expression of this gene from its own promoter in
E.coli
is so weak that it could not be detected by the methods used.
No amino acid sequences similar to that of R.
Hph
I were found in the EMBL and GenBank databases. The sequence motif PD(X)
17
EGK was found in the central part of R.
Hph
I (amino acids 140-161), which matches well the consensus P(E/D)X
9-18
(E/D)XK motif essential for catalytic activity and Mg
2+
binding in the
Eco
RI,
Eco
RV and
Pvu
II restriction endonucleases (
4
).
Similar analysis of
hphIMC
revealed the presence of all 10 motifs common to m5C MTases (
30
). This MTase shares the greatest degree of similarity with
Dde
I (
37
) MTase (28% identical and additionally 11% similar amino acids) and
Aqu
I (
38
) MTase (33% identical and 11% similar amino acids). The translation product of
hphIMA
contains conserved motifs typical for
N
6-methyladenine MTases. This MTase belongs to the D
12
class of m6A MTases (
29
), a group where the conserved motifs appear in the order `F-G-G' and then `DPPY'. Similarly to some other members of the D
12
group, the
Hph
I m6A MTase contains a slightly scrambled version of the first motif: F-GG instead of F-G-G (
29
). This MTase shares the greatest, although marginal, degree of similarity with
the N-terminal part of
Fok
I and
Sts
I MTases (21 and 20% identity was observed, which rises to 38 and 36%
respectively when conservative substitutions are taken into account). In
addition, strong homology was found between the N-terminal part of this
Hph
I MTase and an unidentified ORF located upstream of the truncated copy of the
galE
gene in
Neisseria meningitidis
(
39
,
40
) (67% identity; Fig.
4
). Curiously, the conservation is not only observed within the uncharacterized
ORF, but extends downstream in the same reading frame after the termination
codon. The N-terminal part of the
Hph
I-like MTase of
N
.
meningitidis
and the C-terminal part of the truncated GalE protein overlap by the KMPYT
pentapeptide, dowstream of which no further homology is observed. This suggests
that the
Hph
I-like MTase in
N.meningitidis
was disrupted during duplication of the
galE
locus.
Many sequence-specific MTases, a constituent part of the type II R-M systems, are able to modify bases of a certain single type (either
A or C) positioned symmetrically in the complementary strands of a palindromic
recognition sequence (
10
). More complex and poorly investigated are the type IIS enzymes, which interact
with asymmetrical sequences. In such a case two different sequences in
complementary DNA strands are recognized. There have been only a few reports
where specificity of type IIS modification methylases were analyzed in detail (
5
,
45
,
46
). These include: a single monomeric enzyme modifying adenine residues in both
strands of the target DNA (
Fok
I) (
5
); two separate cytosine MTases (
Hga
I), each responsible for the methylation of different DNA strands (
45
); DNA methylases yielding m6A or m5C on complementary strands in the reaction
catalyzed by a single tandemly arranged enzyme or two separate m6A and m5C
MTases (
46
; J. Bitinaitè, personal communication). The
Hph
I R-M system is likely to belong to the latter type. Sequence analysis of
pHphIMM1 revealed two separate genes encoding proteins resembling,
respectively, m5C and m6A DNA modification methylases.
hphIMC
is most likely responsible for the reported earlier (
10
) methylation of the 5'-proximal C base in the bottom strand of the recognition sequence 5'-GGTGA-3'/5'-T
m5
CACC-3'. The unique 3' A residue in the upper strand seemed a plausible target
base for the
Hph
I m6A MTase encoded by
hphIMA
. To verify this assumption, we constructed a plasmid, pBR-H
(see Materials and Methods), incorporating the overlapping
Hph
I-
Hin
dIII site GGTGAAGCTT. It was reported that the methylated sequence 5'-
m6
AAGCTT-3' is resistant to R.
Hin
dIII, while this site is readily cleaved in the unmodified state (
10
). We found that R.
Hin
dIII was unable to cleave the diagnostic site when pBR-H was isolated from ER2267[pAC-HphIM3] cells, while the cleavage of a second
Hin
dIII site originating from pBR322 occurred as expected (not shown). These
results indicate that
N
6-methyladenine-specific M.
Hph
I modifies the 3'-terminal A base in the upper (5'-GGTGA-3') strand. Additional experiments are
necessary to define whether this methylation is specific for the upper strand
or whether modification of the bottom strand (5'-TCACC-3') also occurs.
We thank Elisabeth A. Raleigh for
E.coli
strain ER2267. We are also grateful to Douglas E. Berg and Saulius Klimasauskas
for valuable discussions and linguistic help and to Jean-Francois Tomb for providing data on
H.influenzae
Rd U32824 and U32827 sequences.
REFERENCES
Return