A new class of genome rare cutters
Alexei G.
Veselkov
,
Vadim V.
Demidov
,
Peter E.
Nielsen
1,
*
and
Maxim D.
Frank-Kamenetskii
Center for Advanced Biotechnology and Department of Biomedical Engineering,
Boston University,
36 Cummington Street,
Boston
, MA 02215,
USA
and
1
Center for Biomolecular Recognition, Panum Institute, University of Copenhagen,
Blegdamsvej 3, DK-2200
Copenhagen
N,
Denmark
Received April 19, 1996;
Accepted May 8, 1996
ABSTRACT
Although significant efforts have been directed at developing efficient
techniques for rare and super rare genome cutting, only limited success has
been achieved. Here we propose a new approach to solve this problem. We
demonstrate that peptide nucleic acid `clamps' (bis-PNAs) bind strongly and sequence specifically to short homopyrimidine
sites on [lambda]
and yeast genomic DNAs. Such binding efficiently shields
methylation/restriction sites which overlap with the bis-PNA binding sites from enzymatic methylation. After removing the bis-PNA, the genomic DNAs are quantitatively cleaved by restriction
enzymes into a limited number of pieces of lengths from several hundred kbp to
several Mbp. By combining various bis-PNAs with different methylation/restriction enzyme pairs, a huge new class
of genome rare cutters can be created. These cutters cover the range of
recognition specificities where very few, if any, cutters are now available.
INTRODUCTION
While pulsed-field gel electrophoresis (PFGE) solves the problem of separation of large
DNA molecules (
1
,
2
), a recent optical mapping technology allows a rapid construction of ordered
maps for rare cut individual chromosomes (
3
,
4
) and the problem of cloning of large DNA fragments has also been solved (
5
,
6
), cleavage of genomic DNA into a limited number of large pieces still remains a
problem. Most restriction enzymes recognize only 4 or 6 nt (
7
), so the target sites are met too often in a particular genome. Although
significant efforts have been directed towards developing efficient methods for
rare and super rare genome cutting, only partial success has been achieved.
Very few naturally occurring rare cutting restriction and intron-encoded endonucleases and other enzymes with recognition sites of 8 bp or
longer have been identified (
7
-
9
). Therefore, further widening of the arsenal of rare DNA cutters is very
desirable. Among the different approaches, the Achilles' cleavage general
strategy has proven to be most efficient and versatile to increase the
selectivity of restriction enzyme cutting (
10
,
11
). In principle, this strategy permits the conversion of any restriction enzyme
into a rare cutter. To this end, a sequence-specific tool is required, which shields one or a few of the
methylation/restriction sites for the choosen pair of methylation/restriction
enzymes on the genomic DNA. However, up to now this approach was limited to
cleavage of genomic DNA at predetermined sites (
12
-
18
). We have recently proposed a new approach, which offers a general solution to
the problem (
19
).
Our approach assumes a ligand which sequence specifically targets short DNA
sites overlapping with restriction/methylation sites. After methylation of
accessible sites and removal of the ligand, DNA is cut with the restriction
enzyme in a very limited number of sites. To meet the goal of rare cleavage of
an arbitrary genome it is necessary to be able to cut DNA of unknown sequence.
Therefore, the ligand must target a short DNA region. If the target region is
15 bp or longer, at the most one cutting site in the entire genome is expected
statistically, given its size equivalent to the human one. To cut an unknown
genome into pieces in the range from several hundred kbp to several Mbp one
needs the total length corresponding to the ligand target site plus the non-overlapping part of the restriction site to be 8-12 bp long. This requirement for targeting short sequences makes
our approach radically different from the existing approaches, which permit
cutting of DNA only at predetermined sites consisting of 20 bp or more (
12
-
18
). Within the framework of these methods shorter ligands either do not bind to
DNA or are displaced at the methylation stage.
We achieve the goal of targeting short DNA sequences by using peptide-nucleic acid (PNA), which sequence specifically forms exceptionally
stable complexes with duplex DNA. PNA is an oligonucleotide mimic in which the
common DNA nucleobases are attached via a linker to an
N
-(2-aminoethyl)glycine backbone (
20
). As a result of sequence-specific binding to duplex DNA of homopyrimidine PNAs, a stable P-loop structure emerges consisting of a PNA
2
-DNA triplex and the displaced DNA strand (
21
,
22
). Furthermore, it has been shown that a restriction site adjacent to the P-loop is completely protected from cleavage by restriction endonuclease (
23
). The X-ray crystallographic structure of the triplex of the oligonucleotide with
two PNA strands (
24
) has indicated why the DNA-PNA
2
triplex is so exceptionally stable. It appeared that, in addition to classical
Hoogsteen pairing, this triplex is stabilized by hydrogen bonds formed by the peptide NH group of the PNA residues, which forms a Hoogsteen pair with an
oxygen of the phosphate groups of the oligonucleotide.
Since two PNA molecules bind to the DNA single strand forming the P-loop, PNA `clamps' (bis-PNAs) consisting of two PNA molecules connected with a flexible
linker proved to be much more efficient in targeting short regions on duplex
DNA than monomeric PNAs (
25
,
26
). Further stabilization of the P-loop at neutral pH can be achieved by replacing cytosines with pseudo-isocytosines in that half of the bis-PNA, designed for Hoogsteen recognition (
25
). We have demonstrated that such bis-PNAs, carrying extra positive charges to promote additional electrostatic
stabilization of the PNA-DNA complexes bound strongly and sequence specifically to quite short
target sites on genomic DNA and efficiently protected a limited number of
methylation sites overlapping with them (
19
). After removing the bis-PNA, the restriction enzyme made very few cuts in the yeast genome (
19
). Here we present the detailed data, which strongly suggest that short,
positively charged bis-PNAs combined with various methylation/restriction enzyme pairs provide a
huge new class of genome rare cutters. This opens new opportunities for genome
mapping, cloning and sequencing.
MATERIALS AND METHODS
PNA oligomerization was performed as previously described (
27
) and the PNA monomers and egl linker were obtained from PerSeptive Biosystems.
The PNA oligomers were purified by reversed phase HPLC and characterized by
MALDI-TOF mass spectroscopy. The following bis-PNAs (J is pseudo-isocytosine and egl is 8-amino-3,6-dioxaoctanoic acid) were used:
PNA
1
, H-T
2
JTJ
2
T
2
-egl
3
-T
2
C
2
TCT
2
-LysNH
2
;
PNA
2
, H-Lys
3
-T
2
J
2
T
3
- egl
3
-T
3
C
2
T
2
-LysNH
2
;
PNA
3
, H-Lys
2
-TJ
4
T
2
J-egl
3
-CT
2
C
4
T- LysNH
2
.
All enzymes were from New England Biolabs except methylase
Cvi
BIII and endonuclease
Sfu
I, which were purchased from Boehringer Mannheim. [lambda] phage DNA was obtained from New England Biolabs.
Saccharomyces cerevisiae
yeast strains PSY316 and BP1 (
28
) were kindly provided by Dr N.Silverman (MIT, Cambridge, MA). Yeast DNA was
isolated as described (
29
).
For PARC analysis, [lambda] or yeast DNA embedded in a 0.65% low melting agarose plug was incubated
with shaking at 37oC in buffer solution (20 mM MES, 10 mM NaCl, 2 mM EDTA, pH 6.3) containing
the desired PNA for 5-10 h. Then the samples were equilibrated in buffer containing either 25
mM sodium citrate, 10 mM NaCl and 5 mM [beta]-mercaptoethanol, pH 7.2, or 25 mM sodium citrate and 5 mM [beta]-mercaptoethanol, pH 7.0, and were incubated overnight
with 1 U methylase
Cvi
BIII, with 15 U methylase
Hha
I or with 8-12 U methylase
Hpa
II at 4oC in the same buffers containing 100 [mu]g/ml BSA. After that, S-adenosylmethionine was added up to 300 [mu]M and methylation was performed for 1.5-4 h at 37oC. Methylation was stopped and the PNA-DNA complex was dissociated by incubation
for 50 min at 58oC in 1% SDS, 500 mM NaCl, 10 mM EDTA, 50 mM Tris-HCl, pH 8.9. Then the samples were equilibrated in an appropriate
buffer and digestion with a desired restriction endonuclease was performed
under optimal conditions in the presence of 100 [mu]g/ml BSA. The digestion was stopped by incubating for 10-20 min in 50 mM EDTA, pH 8. Finally, the samples were equilibrated
with TE buffer, pH 7.5, and loaded onto a 1% agarose gel. The PFGE was run on a
CHEF Mapper system (BioRad).
RESULTS AND DISCUSSION
Cleavage of
[lambda]
DNA using various bis-PNAs
The PNA-assisted rare cleavage (PARC) approach we describe follows the general
scheme of the Achilles' cleavage strategy and uses PNA to protect a limited
number of genomic sites from enzymatic methylation (Fig.
1
). To check whether short bis-PNAs can be used within our approach, we first tested cleavage of [lambda] DNA and used bis-PNA H-T
2
JTJ
2
T
2
-egl
3
-T
2
C
2
TCT
2
-LysNH
2
(PNA
1
). This PNA contains pseudo-isocytosines in one half of the molecule for stabilization of the P-loop at neutral pH. [lambda] DNA contains a unique site for binding PNA
1
,
AAGAGGA
A
TCGAT
at position 35 050, consisting of a homopurine tract, which has a 1 bp overlap
with the
Cla
I methylation/restriction site (here and below, the PNA binding sites are in
bold and the restriction sites are underlined). We applied the PARC strategy
using the PNA
1
/M
.
Cla
I/
Cla
I combination to [lambda] DNA predigested by
Hin
dIII restriction enzyme (the target site was located on the 9.4 kbp
Hin
dIII fragment of [lambda] DNA).
Cutting of yeast chromosomes using different enzyme pairs
To check whether the PARC approach will work on long genomic DNAs we have
applied PNA
3
to the whole yeast genome, as we have already reported (
19
). Here we present additional data on yeast genome cutting by this technique. We
first tried the M
.
Hpa
II/
Sma
I pair of methylation/restriction enzymes. Figure
3
A and B show the results. When PNA
3
and the M
.
Hpa
II/
Sma
I pair are applied to chromosomes from yeast strain PSY316 (Fig.
3
A, lanes 1 and 2), several additional strong bands appear between chromosomes IV
and XV/VII, below chromosome XI, above and below chromosome IX and above chromosome VI. Chromosomes IV, XVI and II are quantitatively cut. Figure
3
B, lane 2 shows results similar to those in Figure
3
A but obtained for another yeast strain (BP1). It is observed that the cleavage
patterns are very similar for the two strains, which have clear differences in
the lengths of some of their chromosomes (compare, for example, chromosomes
XIII and XVI in lanes 1 and 3 in Fig.
3
B). Another pair of enzymes, M
.
Hha
I/
Hae
II, also yielded additional, although quite different, bands when combined with
PNA
3
. Specifically, we could clearly see several new very distinct bands, below chromosome XIII, above chromosome IX and below chromosome I. Furthermore, chromosomes XVI, III and II are quantitatively cut while the rest
of the chromosomes were uncut (
19
).
A new class of genome cutters
The presented data demonstrate that by combining short bis-PNAs with an appropriate pair of methylation and frequently cutting
restriction enzymes a new class of genome rare cutters can be created. Although
only pyrimidine sequences of PNAs can be used, even pyrimidine 7- and 8mers provide a large pool of hundreds of different sequences.
Our data show that virtually complete digestion occurs when the bis-PNA binding site overlaps the methylation site by at least 1 nt.
Incomplete digestion may occur when the PNA binding site does not overlap the
restriction site but they are located very close to each other. Our preliminary
data (not shown) indicate that if the PNA binding site does not overlap the
restriction site and these two sites are adjacent to each other or separated by
1 nt, incomplete protection against methylation is observed. The protection
effect drops dramatically with increasing distance between the two sites and is
unnoticeable when the sites are separated by 2 nt. It should be noted, however,
that in cases of no overlap not only should protection be much weaker, but the
total recognition sequences are also longer and, therefore, these sites are
found much more rarely than the overlapping situation.
Combining the pool of specific homopyrimidine bis-PNAs with a large pool of known methylation/restriction enzyme pairs one
obtains a huge variety of efficient rare cutters. The most frequent of them are
expected to produce, theoretically, fragments in the Mbp range (because the
shortest recognition sites of 9-10 bp in total correspond to 7-8 bp recognized by PNA overlapping with 2-3 bp of a 4-6 bp restriction site). In practice, they may cut
more often if the number of restriction sites is much larger than theoretically
expected, as is apparently the case for the
Hpa
II restriction site CCGG in the yeast genome (see Fig.
3
C). The proposed strategy provides a solution to the problem of how to
quantitatively cleave an unknown genome into a limited number of pieces with
lengths of from several hundred kbp to several Mbp, thus operating in the range
of recognition specificities where very few, if any, cutters are now available.
ACKNOWLEDGEMENTS
We thank Dr N.Silverman for yeast strains and Drs C.R.Cantor and C.L.Smith for
sharing with us some of their equipment during the early stage of the project.
This work was supported in part by the NIH, PerSeptive Biosystems and the
Danish National Research Foundation.
