Epigenetics: application of virtual image restrictionlandmark genomic scanning Vi-RLGS Kuniaki Koike1, Tomoki Matsuyama2and Toshikazu Ebisuzaki1 1 Computational Astrophysics Laboratory,
Trang 1Epigenetics: application of virtual image restriction
landmark genomic scanning (Vi-RLGS)
Kuniaki Koike1, Tomoki Matsuyama2and Toshikazu Ebisuzaki1
1 Computational Astrophysics Laboratory, Discovery Research Institute, RIKEN, Saitama, Japan
2 Plant Breeding and Cell Engineering Research Unit, Discovery Research Institute, RIKEN, Saitama, Japan
Restriction landmark genomic scanning (RLGS) uses
restriction enzyme sites as landmarks [1,2] and is used
in the detection of DNA polymorphisms caused by
genetic mutations and hyper- or hypomethylation
changes in cancer cells [3–6], imprinted genes [7,8] and
linkage maps [9,10] This method is an especially
powerful tool in DNA methylation studies using
methylation-sensitive restriction enzymes, as it allows
genome-wide scanning to detect alterations in DNA
methylation after fractionation of DNA fragments by
high-resolution, 2D gel electrophoresis [4–8,11]
The RLGS procedure is shown in Fig 1 First,
restriction landmarks, typically those of rare-cutter
enzymes, are labeled directly with a radioisotope and
subjected to 1D electrophoresis in an agarose gel In
most cases, the DNA is digested with six-cutter
enzymes for clear fractionation in 1D electrophoresis (step I, Fig 1) Next, the fractionated DNA is digested with a four-cutter enzyme and subjected to 2D gel electrophoresis in a polyacrylamide gel (step II, Fig 1) The separated DNA is visualized as a pattern
of spots after exposure to X-ray film (step III, Fig 1) Thousands of spots can be identified with good repro-ducibility Different landmark restriction enzymes allow further extension of the scanning field In addi-tion, the autoradiographic intensity of the spots clearly reflects copy number and allows quantitative analysis,
as labeling occurs only at landmark enzyme sites Therefore, in contrast to PCR-mediated methods, changes in epigenetic alterations caused by DNA methylation can be analyzed using differences in RLGS spots
Keywords
Arabidopsis; DNA methylation; DNA
polymorphism; electrophoresis; epigenetics;
in silico; mutant; 5mC; N6-methyladenine;
Vi-RLGS
Correspondence
T Ebisuzaki, Computational Astrophysics
Laboratory, Discovery Research Institute,
RIKEN, 2-1 Hirosawa, Wako,
Saitama 351-0198, Japan
Fax: +81 48 467 4078
Tel: +81 48 467 9414
E-mail: ebisu@riken.jp
(Received 30 November 2007, revised
28 January 2008, accepted 1 February 2008)
doi:10.1111/j.1742-4658.2008.06329.x
Restriction landmark genomic scanning (RLGS) is a powerful method for the systematic detection of genetic mutations in DNA length and epigenetic alteration due to DNA methylation However, the identification of poly-morphic spots is difficult because the resulting RLGS spots contain very little target DNA and many non-labeled DNA fragments To overcome this, we developed a virtual image restriction landmark genomic scanning (Vi-RLGS) system to compare actual RLGS patterns with computer-simu-lated RLGS patterns (virtual RLGS patterns) Here, we demonstrate in detail the contents of the simulation program (rlgssim), based on the lin-ear relationship between the reciprocal of mobility plotted against DNA fragment length and Vi-RLGS profiling of Arabidopsis thaliana
Abbreviations
5mC, 5¢-methylcytosine; RLGS, restriction landmark genomic scanning; RLGSSIM, restriction landmark genomic scanning simulation software; Vi-RLGS, virtual image restriction landmark genomic scanning.
Trang 2The most important step in developing analyses using
RLGS profiles is the cloning of target spots However,
the amount of DNA available for cloning in a single
spot is very small For example, 1.5 lg of mouse
geno-mic DNA used in an RLGS analysis results in
atto-moles (10)18) of target DNA available for ligation when
all of the DNA molecules are recovered from the
poly-acrylamide gel In addition, the isolated gel fragment
contains 2000· more non-labeled than labeled DNA
fragments [12] In most trials, non-target DNA has been
amplified using PCR-adapter methods Restriction
trap-per-based methods are limited to the purification of
NotI landmarks [13] Therefore, the most difficult step
in RLGS analysis is recovering the target DNA
To overcome this problem, we developed a novel
in silico system for identifying spots using computer
simulation software (rlgssim) and designated virtual
image restriction landmark genomic scanning
(Vi-RLGS) based on organisms for which the entire
genomic DNA sequence is known [14,15]
Algorithm First, rlgssim reads sequences to generate a pattern The program can read sequences in GenBank or FASTA format Next, the program generates the elec-trophoresis pattern
Figure 2 illustrates the simulation procedure that follows:
Step 1 The sequence is cut into fragments; for exam-ple, ‘Fragment-A’ is cut with restriction enzyme A Step 2 Restriction enzyme B cuts Fragment-A into
‘Fragment-AB’ and ‘Fragment-BB’
Step 3 The X-dimensional mobility of Fragment-AB is calculated
Step 4 Restriction enzyme C cuts Fragment-AB into further fragments, including ‘Fragment-AC’
Step 5 The Y-dimensional mobility of Fragment-AC
is calculated
Fig 1 RLGS procedure The actual RLGS pattern is generated by NotI–EcoRV–MboI (restriction enzymes A–B–C, respectively) in rice ‘a’,
‘b’ and ‘c’ indicate the respective restriction enzyme end sites.
Trang 3These steps generate 2D mobility (X,Y) values for
each fragment of a given DNA sequence and
combina-tion of restriccombina-tion enzymes Because the electrophoresis
time is constant for each fragment, we can plot the 2D
mobility (X,Y) of each fragment to generate a virtual
2D electrophoresis image
A fragment containing the origin or end point of
each sequence might not be valid because each
sequence (clone) may be divided from a single long
sequence Therefore, we added information to indicate
if a fragment included the origin or end point of each
sequence
Implementation
The main components of the simulation engine are a
sequence-reading module and
electrophoresis-simula-tion module We developed these modules in the
C++ language
Sequence-reading module
We implemented the sequence multi-format reading
module, which is capable of reading GenBank, FASTA
and original sequence formats, in our laboratory
This module is designed for object-oriented
program-ming The sequence reader consists of a main reader
module and specific format (GenBank, FASTA and original format) parsers The main reader module is completely separate from specific format parsers, allow-ing us to easily add new format types to the sequence reader module The main reader module reads a sequence file to a memory buffer, then determines the sequence format using the format parsers The format parsers check the validity of the sequence, and send this information to the main reader module If the sequence format can be determined, the main reader module reads the sequence; otherwise, it reports error information to the user Unknown or discontinuous parts of the sequence (such as ‘N’) are reported to the user to help determine the validity of a result spot
The 2D electrophoresis-simulation engine The main components of the simulation engine are a restriction enzyme component, mobility calculator and 2D electrophoresis component The restriction enzyme component splits a sequence into fragments using given recognition sequences and cut positions as parameters The component finds the recognition sequence in the main sequence, and then splits it into fragments at the given position
The mobility calculator calculates the mobility of the fragment sequence The mobility is determined
Fig 2 Simulation procedure The letters ‘i’, ‘j’ and ‘k’ indicate the DNA fragments resulting from digestion by restriction enzymes A, B and
C, respectively ‘a’, ‘b’ and ‘c’ correspond to the restriction enzyme sites in Fig 1 Figure 3 shows their computational handling.
Trang 4solely from the length of the fragment reported in
sequence databases The reciprocal of mobility (m)
plotted against fragment length (l) is linear [16,17] We
use the following formulae:
mx¼ 76:368
lx½kBp þ 1:032þ 3:745 ð1Þ
my¼ 15850:043
ly½kBp þ 476:068 3:521 ð2Þ where lx is fragment length for the X direction, mx is
the mobility corresponding to the X direction of the
fragment, lyis fragment length for the Y direction and
my is the mobility corresponding to the Y direction of
the fragment Each coefficient is calculated from actual
surveys of actual RLGS patterns using 100-bp or 1-kb
ladder markers
The 2D electrophoresis engine generates data for
each spot from the sequence and information on
restriction enzymes A, B and C Each spot contains
the following data: (a) sequence of the fragment; (b)
X-direction mobility (mx); (c) Y-direction mobility
(my); (d) marking flag (if set, the spot is visible); and
(e) an edge flag
To visualize these spots, we plot (mx, my) on a 2D
plane for marked spots In Fig 3, we show how to
generate these spots from a given sequence and the
restriction enzymes
The 2D electrophoresis flow The flowchart in Fig 3 summarizes the operation using the designations: RezA, RezB and RezC, restric-tion enzymes A, B and C, respectively; seq, sequence data; wfrag, xfrag and yfrag, sequence array; wfrag[i], ith sequence in the wfrag array; wfrag, fragments split
by restriction enzyme A; xfrag, fragments split by restriction enzyme B; yfrag, fragments split by restric-tion enzyme C; and the subscripts i, j and k denote positions within the sequence
First, the original sequence is split by restriction enzyme A to yield wfrag Next, wfrag fragments are split by restriction enzyme B to yield xfrag At this stage, the X-direction mobility is calculated for each fragment in xfrag Next, the xfrag fragments are split
by restriction enzyme C to yfrag, and the Y-direction mobility is calculated for yfrag fragments A marked flag is set if the fragment has an A-edge, and an edge flag is set if the fragment contains an origin or end point of the original sequence
User interface
To set the parameters and view the result on the screen, we built a graphical user interface (Fig 4) We can specify the restriction enzymes and the list
of sequences graphically and view the generated 2D
Fig 3 Flowchart of the 2D electrophoresis simulation in virtual RLGS.
Trang 5electrophoresis pattern We implemented the graphical
user interface using the Microsoft Foundation Classes
(MFC) library, which provides the framework for
standard Windows OS application programs The user
interface layer also has a function that allows us to
load and store spot data in a storage file This layer
can manage the relative positions of several sequences
and show their electrophoresis images in the same
win-dow Thus, we can easily specify the clone to which a
selected spot belongs This software can be accessed at
RIKEN DRI (contact T Matsuyama or T Ebisuzaki,
http://www.riken.jp/engn/r-world/research/lab/unit/
breeding/index.html)
RLGS profiling using the Vi-RLGS
system
The entire nuclear genomic DNA sequence of the
model plant Arabidopsis thaliana (L.) Heynh
(Colum-bia) is known [15] Because the information available
on TAIR (http://www.arabidopsis.org/) and MIPS (http://mips.gsf.de/proj/plant/jsf/index.jsp) has a high degree of accuracy and NotI is used as a universal restriction landmark enzyme in animal and plant RLGS analysis, we evaluated the system by first per-forming a NotI–Arabidopsis simulation profile using our Vi-RLGS system, and EcoRV and MboI as enzymes B and C, respectively (NotI–EcoRV–MboI) The actual pattern, which is in the range 0.6–7.5 kb
in the first dimension and 50–800 bp in the second dimension, is shown in Fig 5A The virtual pattern corresponding to the range enclosed by the broken line
is shown in Fig 5B The spots indicated by arrows in Fig 5A were cut from the gel and cloned using the PCR-adapter ligation method [18,19] Sequencing con-firmed that their patterns concurred with those expected from Vi-RLGS theory The differences in number and pattern of spots in Fig 5A,B may have
Fig 4 Screenshot of the 2D electrophoresis simulator (RLGSSIM) The right-hand panel is the image resulting from the simulation, and the left-hand panel shows the processed sequence information If the clone (sequence) is checked, the clone has spots in the right panel Clones (sequences) are categorized by group, and the relative position of each clone (sequence) is shown in the left-hand panel The arrow indicates a spot that was selected, and sequence information for the spot is displayed on the screen.
Trang 6been due to gaps in the Arabidopsis genomic DNA
sequence, for example, from highly repeated sequences
such as centromeres, telomeres, ribosomal RNA gene
clusters and their flanking-region DNA sequences that
have not yet been reported, and DNA modification by
DNA methylation
In plant DNA, 5¢-methylcytosine (5mC) occurs at
cytosine residues in symmetrical sequences, CpG and
CpNpG (where N is any nucleotide), due to the
actions of MET1, DRM1⁄ 2 and CMT3 [20–22] In
Arabidopsis, the 5mC content (5mC⁄ 5mC+C) is
5.2% [14,23] Because only NotI is sensitive to 5mC,
and EcoRV and MboI are insensitive, this difference
results in the recognition of DNA methylation at NotI
sites The spots indicated by black arrowheads in
Fig 5B were present in virtual RLGS patterns but
absent in actual RLGS patterns The influence of 5mC
at NotI landmarks was confirmed using the bisulfite
sequencing method and Vi-RLGS analysis of
20%-reduced 5mC hypomethylated Arabidopsis plants gen-erated using 5-aza-2¢-deoxycytidine Therefore, the masked spots resulted mainly from DNA methylation However, genome-wide detection of methylated regions
in Arabidopsis genomic DNA can be realized only by gathering information from spots present in virtual RLGS patterns but absent in actual RLGS patterns [14]
However, the spot indicated by the white arrowhead
in Fig 5B has slightly different mobility in the 2D Vi-RLGS profile This phenomenon was observed mainly in 2D high molecular mass regions ( 500 bp)
We speculate that they resulted from sequence gaps or sliding due to methylation in the flanking regions In addition, in 2D polyacrylamide gels, the electrophore-sis mobility of short DNA fragments was affected by their base composition and sequence For example, curved DNA caused by short adenine tracts may move aberrantly However, when we reduced the
polyacryl-II
1F10F5 5K18P6 3T4P13 5F2G14 5MPI10 5K18P6 5MDC12 4F13C5 5F18A17 1F19G10 4T11J8 3F4F8 5F2G14 2F24C20 5F2K13 5MTH12 5MIK19 1F1019 3MMB12 1T23F18 3T8H10 3K7M2 4T13J8 2T9F8 3F3C22
4T16H5 4T805 4FCA1 4F20M13 5MIK19 2F14M4 1T19D16 2T1014 3F15G16 4F17A13 3F16L2 3F1C9 1T14N5 1F12K11 1F3M18 2T9J23 3T7M13 4F4D11 3F9F8 4FI10 5MWF20 1F10K1
D
3K10D20
1F9L1
1F3M18
5MKD15
1F1019
4FCA1
1T3F24
1F2K11
3K7L4
1F16F4
5F502
5K17N15
4F4I10
3MYM9
5MWF20
5MWF20
5K19B1
5MKD15
3F26K24
1F19G10
2T29F13 1F25P22 4F26F21 5F8L15 5K17H15 5K19B13 1T23K23 1T3F24 5F7K24 3MFD22 2F3L12 3F8A24 4T805 4F17A13 4T30C3 3F16L2 2T29F13 3MLN21 4T15F16 3MYM9 3T18B22 3T4P13
I
C
2D (bp)
A
500
200
1.0 2.0 5.0 1D (kb)
I II
B
T20O10
T26N6 F17A13 FCA2 F7A10 T300
F16L2 T29F13 MBK21 MLN21
T15F16 MYM9 F3L12
T22N4 F17A17T18B22
F2O106
K17N1524 F4I10
MKD15
T18B22 T20J72 F26K24 F15F4
MYM9 J24
MWF20 MWF20 K19B1
00
7
MTH12 U22
Fig 5 Vi-RLGS profiling of Arabidopsis (A) Actual RLGS pattern, (B) virtual RLGS pattern, corresponding to the sequence indicated by the broken line The spots indicated by arrowheads are absent or have slightly different mobility in (C) (C,D) Vi-RLGS profiles generated by matching the actual with the virtual RLGS patterns.
Trang 7amide gel concentration from 5 to 4%, the aberrant
motion was dispelled, and the actual RLGS spots were
similar to Vi-RLGS patterns In addition, we
con-firmed that the DNA fragments of Arabidopsis
chloro-plast DNA (accession number: AP000426, position:
58 165–58 777), which showed aberrant mobility in
Vi-RLGS profiling, were used in PAGE with the usual
slab-style gels at a high temperature (45C) and the
differential mobility was also dispelled (data not
shown) These results demonstrate that Vi-RLGS
pro-filing of the actual pattern and the virtual pattern
pos-sess inherent problems caused by secondary structure;
these differences need to be noted for spot
identifica-tion However, as the secondary structure of DNA,
such as DNA curvature caused by the architecture of
nucleoprotein complexes, is thought to be involved in
biological processes [24], Vi-RLGS analysis may also
be suitable for detecting regions that indicate a
rela-tionship with gene expression
Despite these anomalies, a comparison of virtual and
actual RLGS patterns showed that Vi-RLGS identified
85.5% of the spots in the actual RLGS pattern, and
was able to generate Vi-RLGS profiles that allowed
effective and rapid quantitative analysis of 5mC status
at NotI sites and their flanking regions (Fig 5C,D) In
theory, this method allows the study of epigenetic
changes caused by 5mC in all organisms, including
humans, animals, plants and micro-organisms such as
bacteria, fungi and algae that possess bulk DNA
sequences, even without detailed DNA sequence
infor-mation In fact, many actual RLGS spots of mouse
genomic DNA have been verified, and the Vi-RLGS
system is effective for epigenetic studies of
tissue-specific differentially methylated regions and embryonic
stem cells [25,26] Various in silico systems have been
developed and have demonstrated high validity in
RLGS analysis, similar to the Vi-RLGS system [27,28]
In addition, bacterial genomic DNA possesses
N6-methyladenine at GATC and GANTC sites caused by
deoxyadenosine methyltransferase and
cell-cycle-regu-lated methyltransferase [29] Recently, Vi-RLGS
analy-ses of N6-methyladenine alterations in the genomic
DNA of the symbiont Mesorhizobium loti and the plant
pathogen Xanthomonas oryzae have been reported
[30,31] Our system using N6-methyladenine-sensitive
restriction enzymes, such as MboI and HinfI, is capable
of expanding the genomic analysis of microorganisms
as their whole genomic DNA sequences are reported
Conclusions
Cloning target RLGS spots is very time- and
labor-intensive Vi-RLGS saves considerable time and effort,
and utilizes a simple procedure for identifying target spots in RLGS analysis and screening of suitable restriction enzymes The typical RLGS analysis is lim-ited to between 0.5 and 6.0 kb in the first dimension and no more than 100 bp in the second dimension because of constraints in cloning procedure efficiency and the reproducibility of spot signal intensity Our Vi-RLGS analysis overcomes these limitations and expands the scanning field and ability to detect changes in genomic DNA
Moreover, in typical analyses, the cloning step requires a large amount of genomic DNA Thus, in plant studies, obtaining sufficient DNA for RLGS analysis of particular tissues and organs at various developmental stages is difficult Vi-RLGS profiles generated in advance overcome these problems and allow instant analysis of genome status at each devel-opmental stage, as target spots can be identified from
a single profile with a very small amount of applied DNA
The Vi-RLGS system, consisting of electrophoresis, restriction enzyme digestion and in silico analysis, has good reproducibility and high resolution ability at both unique gene regions and methylated repetitive sequences [21,32] Thus, the Vi-RLGS system offers a different spectrum of mutation detection than that of microarrays and will be a valuable tool for detecting both genetic mutations in DNA lengths and epigenetic alterations with DNA modification in post-genomic sequencing research
Acknowledgements
We thank Dr D J Smiraglia for his critical reading of this manuscript This study was supported by a grant from the Research Collaborations with Industry Pro-gram, RIKEN TM is supported by a grant for Basic Science Research Projects from the Sumitomo Founda-tion
References
1 Hatada I, Hayashizaki Y, Hirotsune S, Komatsubara H
& Mukai T (1991) A genomic scanning method for higher organisms using restriction sites as landmarks Proc Natl Acad Sci USA 88, 9523–9527
2 Hayashizaki Y & Watanabe S (1997) Restriction Land-mark Genomic Scanning (RLGS) Springer, Tokyo
3 Hirotsune S, Hatada I, Komatsubara H, Nagai H, Kuma K, Kobayakawa K, Kawara T, Nakagawara A, Fujii K, Mukai T et al (1992) New approach for detec-tion of amplificadetec-tion in cancer DNA using restricdetec-tion landmark genomic scanning Cancer Res 52, 3642–3647
Trang 84 Costello JF, Fru¨hwald MC, Smiraglia DJ, Rush LJ,
Robertson GP, Gao X, Wright FA, Feramisco JD,
Pelt-oma¨ki P, Lang JC et al (2000) Aberrant CpG-island
methylation has non-random and tumour-type-specific
patterns Nat Genet 24, 132–138
5 Zardo G, Tiirikainen MI, Hong C, Misra A, Feuerstein
BG, Volik S, Collins CC, Lamborn KR, Bollen A,
Pin-kel D et al (2002) Integrated genomic and epigenomic
analyses pinpoint biallelic gene inactivation in tumors
Nat Genet 32, 453–458
6 Smiraglia DJ & Plass C (2002) The study of aberrant
methylation in cancer via restriction landmark genomic
scanning Oncogene 21, 5414–5426
7 Hayashizaki Y, Shibata H, Hirotsune S, Sugino H,
Okazaki Y, Sasaki N, Hirose K, Imoto H, Okuizumi
H, Muramatsu M et al (1994) Identification and
char-acterization of an imprinted U2af binding protein
related sequences on mouse chromosome 11 detected by
efficient genomic screening using restriction landmark
genomic scanning (RLGS-M) Nat Genet 6, 33–40
8 Plass C, Shibata H, Kalcheva I, Mullins L,
Kotelevts-eva N, Mullins J, Okazaki Y, Held WA, Hayashizaki Y
& Chapman VM (1996) Identification of Grf1 on mouse
chromosome 9 as an imprinted gene by RLGS-M Nat
Genet 14, 106–109
9 Hayashizaki Y, Hirotsune S, Okazaki Y, Shibata H,
Akasako A, Muramatsu M, Kawai J, Hirasawa T,
Shiroishi T, Watanabe S et al (1994) A genetic linkage
map of the mouse using restriction landmark genomic
scanning (RLGS) Genetics 138, 1207–1238
10 Okazaki Y, Okuizumi H, Ohsumi T, Nomura O,
Tak-ada S, Kamiya M, Sasaki N, Matsuda Y, Nishimura
M, Takagi O et al (1996) A genetic linkage map of the
Syrian hamster and localization of cardiomyopathy
locus on chromosome 9qa2.1-b1 using RLGS
spot-mapping Nat Genet 13, 87–90
11 Kawai J, Hirotsune S, Hirose K, Fushiki S, Watanabe
S & Hayashizaki Y (1993) Methylation profiles of
geno-mic DNA of mouse developmental brain detected by
restriction landmark genomic scanning (RLGS) method
Nucleic Acids Res 21, 5604–5608
12 Ohsumi T, Okazaki Y, Hirotsune S, Shibata H,
Mura-matsu M, Suzuki H, Taga C, Watanabe S &
Hayashi-zaki Y (1995) A spot cloning method for restriction
landmark genomic scanning Electrophoresis 16, 203–
209
13 Hayashizaki Y, Hirotune S, Hatada I, Takamatsu S,
Miyamoto C, Furuichi Y & Mukai T (1992) A new
method for constructing NotI linking and boundary
libraries using a restriction trapper Genomics 14, 733–
739
14 Matsuyama T, Kimura MT, Koike K, Abe T, Nakano
T, Asami T, Ebisuzaki T, Held WA, Yoshida S &
Nagase H (2003) Global methylation screening in the
Arabidopsis thalianaand Mus musculus genome:
applications of virtual image restriction landmark geno-mic scanning (Vi-RLGS) Nucleic Acids Res 31, 4490– 4496
15 Arabidopsis Genome Initiative (2000) Analysis of the genome sequence of the flowering plant Arabidopsis tha-liana Nature 408, 796–815
16 Southern EM (1979) Measurement of DNA length by gel electrophoresis Anal Biochem 100, 319–323
17 Okuizumi H, Matsuyama T & Hayashizaki Y (2004) Restriction landmark genome scanning method In Encyclopedia of Molecular Cell Biology and Molecular Medicine, Vol 5, 2nd edn (Meyers RA, ed), pp 413–
439 Wiley–VCH Verlag, Weinheim
18 Matsuyama T, Abe T, Bae C-H, Takahashi Y, Kiuchi
R, Nakano T, Asami T & Yoshida S (2000) Adaptation
of restriction landmark genomic scanning (RLGS)
to plant genome analysis Plant Mol Biol Rep 18, 331– 338
19 Suzuki H, Kawai J, Taga C, Ozawa N & Watanabe S (1994) A PCR-mediated method for cloning spot DNA
on restriction landmark genomic scanning (RLGS) gel DNA Res 1, 245–250
20 Chan SW-L, Henderson IR & Jacobson SE (2005) Gar-dening the genome: DNA methylation in Arabidopsis thaliana Nat Rev 6, 351–360
21 Zhang X, Yazaki J, Sundaresan A, Cokus S, Chan SW, Chen H, Henderson IR, Shinn P, Pellegrini M, Jacob-sen SE et al (2006) Genome-wide high-resolution map-ping and functional analysis of DNA methylation in Arabidopsis Cell 126, 1189–1201
22 Bird A (2007) DNA methylation patterns and epigenetic memory Genes Devel 16, 6–21
23 Messeguer R, Ganal MW, Steffens JC & Tanksley SD (1991) Characterization of the level, target sites and inheritance of cytosine methylation in tomato nuclear DNA Plant Mol Biol 16, 753–770
24 Marini JC, Levene SD, Crothers DM & Englund PT (1982) Bent helical structure in kinetoplast DNA Proc Natl Acad Sci USA 79, 7664–7668
25 Hattori N, Abe T, Hattori N, Suzuki M, Matsuyama
T, Yoshida S, Li E & Shiota K (2004) Preference of DNA methyltransferases for CpG islands in mouse embryonic stem cells Genome Res 14, 1733–1740
26 Song F, Smith JF, Kimura MT, Morrow AD, Matsuy-ama T, Nagase H & Held WA (2005) Association of tis-sue-specific differentially methylated regions (TDMs) with differential gene expression Proc Natl Acad Sci USA 102, 3336–3341
27 Rouillard JM, Erson AE, Kuick R, Asakawa J, Wim-mer K, Muleris M, Petty EM & Hanash S (2001) Vir-tual genome scan: a tool for restriction landmark-based scanning of the human genome Genome Res 11, 1453– 1459
28 Smiraglia D, Kazhiyur-Mannar R, Oakes C, Wu Y, Liang P, Ansari T, Su J, Rush L, Smith L, Yu L et al
Trang 9(2007) Restriction landmark genomic scanning (RLGS)
spot identification by second generation virtual RLGS
in multiple genomes with multiple enzyme
combina-tions BMC Genomics 8, 446
29 Ratel D, Ravanat JL, Berger F & Wion D (2006)
N6-methyladenine: the other methylated base of DNA
BioEssays 28, 309–315
30 Ichida H, Matsuyama T, Abe T & Koba T (2007)
DNA adenine methylation changes dramatically during
establishment of symbiosis FEBS J 274, 951–962
31 Ichida H, Maeda K, Ichise H, Matsuyama T, Abe T, Yoneyama K & Koba T (2007) In silico restriction landmark genome scanning analysis of Xanthomonas oryzaepathovar oryzae MAFF 311018 Biochem Bio-phys Res Commun 363, 852–856
32 Zilberman D, Gehring M, Tran RK, Ballinger T & Henikoff S (2007) Genome-wide analysis of Arabidopsis thalianaDNA methylation uncovers an interdependence between methylation and transcription Nat Genet 39, 61–69