Published online February 18, 2004

Nucleic Acids Research, 2004, Vol. 32, No. 3 e31

DOI: 10.1093/nar/gnh035

Large-scale determination of the methylation status

of retrotransposons in different tissues using a
methylation tags approach
Konstantin Khodosevich*, Yuri Lebedev and Eugene D. Sverdlov

Shemyakin-Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Science, 16/10 Miklukho-Maklaya

St., 117997 Moscow, Russia

Received December 10, 2003; Revised and Accepted January 24, 2004

ABSTRACT expression, although this suggestion is still not reliably proven

(1). Even less is known about functions of CpG dinucleotides
A technique for simultaneous determination of the located outside CpG islands. One of the important functions
methylation status of numerous loci containing usually assigned to methylation is suppression of the activity
retroelements (REs) is reported. It is based on the of retroelements (REs), including that of long interspersed
observation that methylated and unmethylated elements (LINEs), short interspersed elements (SINEs) and
areas in the genome are usually extended, long terminal repeats (LTRs).
and therefore the methylation of particular methyl- Due to evolutionary selection, REs are mostly located in
sensitive restriction endonuclease recognition sites `inactive' genomic regions such as heterochromatin and are
might re¯ect the methylation status of DNA regions heavily methylated. Moreover, it is quite possible that newly
around them. The method includes dot-blot hybridi- integrated REs methylated by the cellular methylation
zation of repeat ¯anking sequences arrayed on a machinery become centers of spreading methylation over
surrounding genomic regions, thus suppressing their func-
solid support with speci®cally ampli®ed ¯anking
tional activity (4). On the other hand, a multitude of REs were
regions of presumably unmethylated repeats. A `domesticated' by mammalian genomes and might be
multitude of ¯anking regions of REs adjacent to involved in modulation of expression of resident genes (5,6).
unmethylated restriction sites are ampli®ed simul- The involvement of REs in regulation of gene expression has
taneously, providing a complex hybridization probe. been demonstrated in a number of studies (7±10). REs of this
The technique thus allows the determination of the group can be suggested to be active non-methylated regulatory
methylation status of restriction sites, which serve units located in active non-methylated genome areas.
as tags of the methylation status of the surrounding Determination of the methylation status of REs and their
regions. The validity of the technique was con®rmed adjacent regions may be used to discriminate between
by various means, including bisul®te sequencing. potentially active and inactive REs. Furthermore, insofar as
The technique was successfully applied to the each type of RE represents a group of repetitive elements with
conserved sequence features, REs can be used as universal
identi®cation of methylation patterns of the regions
genomic anchors allowing the design of easily detectable tags
surrounding 38 human-speci®c HERV-K(HML-2) of the methylation status of the loci where they are located.
long terminal repeats in cerebellum- and lymph In this work, we developed a new technique for detection of
node-derived genomic DNAs. The described tags of the methylation status of genomic loci containing
technique can be readily adapted to the use of DNA interspersed repeats, and successfully applied it to 38 LTRs of
microarray technology. human endogenous retroviruses (HERVs) belonging to the
most functionally active family, HERV-K(HML-2), and
located in different genomic loci (11±13). The methylation
INTRODUCTION patterns obtained in such a way for cerebellum- and lymph
node-derived genomic DNAs were compared and shown to be
It is widely recognized that DNA methylation has profound
different. The validity of this technique for determining the
effects on gene expression. DNA methylation plays an
methylation status was con®rmed by bisul®te sequencing and
important role in embryonic development, X-chromosome
PCR ampli®cation.
inactivation, imprinting, and suppression of repetitive
sequences in the genome (1). Approximately 80% of CpG
residues in the mammalian genome are methylated in most MATERIALS AND METHODS
adult tissues (2). CpG dinucleotides are distributed non-
randomly along DNA and clustered within so-called CpG Genomic DNA isolation, restriction and adaptor ligation
islands that are usually protected from methylation in normal Genomic DNA from human tissues was isolated using
cells (3) and possibly are involved in tissue speci®city of gene proteinase K digestion and phenol extraction as described

*To whom correspondence should be addressed. Tel: +7 095 330 6329; Fax: +7 095 330 6538; Email: [email protected]

Nucleic Acids Research, Vol. 32 No. 3 ã Oxford University Press 2004; all rights reserved
e31 Nucleic Acids Research, 2004, Vol. 32, No. 3 PAGE 2 OF 8

Table 1. Oligonucleotides and primers used in hybridization probe Table 2. Primers used in bisul®te sequencing
construction
Primers Structure (5¢±3¢)
Primers Structure (5¢±3¢)
LTR11bisfor1 TATAAGTTGGATTTGTATTAGAGGATTTG
A1A2 AGCAGCGAACTCAGTACAACAAGTCGACGCGTGCCC- LTR11bisfor2 GGTTTAGAGTTTAGGAGTTTTGGTTGTTAG
GGGCTGGT LTR11bisrev1 CTTACTCACACTAAACCTAACAATAAATACTC
A1 AGCAGCGAACTCAGTACAACA LTR11bisrev2 ACATCCCCAACCTACATCTCCCTC
A2 AGTCGACGCGTGCCCGGGCTGGT LTR12bisfor1 GTAGGATATGAAGTATGGAGAATAGGTAAG
a1 CGACCAGCCC LTR12bisfor2 TATTTTTGTAGGGATGTATAGTGTTGGG
T1 CCACCTTACGAGAAACACCCACAG LTR12bisrev1 CACCTATAAACCATCCTTTACTATTATAAAC
T2 TGTTTCAGAGAGCACGGGGTTGGG LTR12bisrev2 AAAATCCTTTTACCTCCATTCATCTTC
T3 CGAGAAACACCCACAGGTGTG LTR27bisfor1 TTTTGTTATTGAGAAGTATATAGTTTAGTGG
T4 TCTGATCTCTCTTGCTTTTCCCCAC LTR27bisfor2 GGAAGAAGAGTTAATTGGTGGTATAAGG
LTR27bisrev1 AAATCTCTAATACCATTCCTCCTACC
LTR27bisrev2 ATAAAAATCCCAATATTAAAATCACATCC

The PCR primers were designed to be fully complementary to the

deaminated DNA strand and did not include CG dinucleotides. First round
(14). A 1 mg aliquot of the genomic DNA from each tissue was ampli®cations (with primers for1 and rev1) were performed in 25 ml of PCR
digested with 70 U of HpaII restriction enzyme (New England buffer for Taq DNA polymerase (Promega) containing 200 mM of each
Biolabs) at 37°C overnight in 100 ml of the reaction mixture. dNTP, 1.5 mM MgCl2, 0.4 mM of each primer and 1 U of Taq DNA
polymerase (Promega), with individual 3 ml beads in a thermal cycler (MJ
The completeness of HpaII digestion was con®rmed by PCR Research) as follows: 35 cycles at 95°C for 20 s, 63°C for 30 s, and 72°C
with primers to either side of four previously identi®ed non- for 1.5 min. Ampli®cations using nested primers (for2 and rev2) were
methylated CCGG sites located within exons, introns or performed using 1 ml of the PCR mixture after the ®rst round under PCR
promoter regions of housekeeping genes (see table 1 of the conditions as follows: 22 cycles at 95°C for 20 s, 60°C for 30 s, and 72°C
for 1.5 min.
Supplementary Material available at NAR Online). The
digested DNAs were ethanol precipitated, dissolved in 10 ml
of sterile water and ligated to an excess of adaptor (2 mM) at
16°C overnight using T4 DNA ligase (Promega). To form the each) were then spotted and immobilized on nylon membranes
adaptor, 300 pmol of each of the oligonucleotides A1A2 and (Hybond N; Amersham) according to the manufacturer's
a1 (Table 1) were annealed in 20 ml of TM (10 mM Tris±HCl, recommendations. The same amount of l phage DNA was
pH 7.8, 10 mM MgCl2) buffer. The ligation was terminated by also spotted on the membranes as a negative hybridization
incubation at 65°C for 15 min. The ligates were then puri®ed control.
from excess oligonucleotides by passing them through a
QIAquick DNA Puri®cation Kit (Qiagen), and their 3¢ ends Dot-blot hybridization
were ®lled in by the Klenow fragment of DNA polymerase I Hybridization probes were prepared by random labeling of the
(Fermentas). Finally, the ligates were ethanol precipitated and second round suppressive PCR products (see Preparation of
redissolved in 40 ml of sterile water. hybridization probes). The DNA fragments dotted on the
membrane and the probes had the same sequences at their 5¢ or
Preparation of hybridization probes 3¢ termini, i.e. 25 and ~40 bp when ampli®ed from the LTR U3
Fractions of LTR-U5 and LTR-U3 ¯anking sequences were or U5 terminal regions, respectively. To prevent cross-
ampli®ed using a PCR suppression approach (15). In the ®rst hybridization between these sequences, the hybridization
round of suppressive PCR, A1 primer was used with T1 or T2 mixture was supplemented with a 200-fold molar excess of
primers (Table 1). Primers T1 and T3, and T2 and T4 (Table 1) unlabeled competitive oligonucleotides corresponding to the
correspond to the most conserved parts of the U5 and U3 common parts of the spotted ¯anks. After random labeling,
regions of HERV-K(HML-2) LTRs, respectively. The PCR probes were diluted, mixed with competitive oligonucleotides
mixture contained 10 ng of the ligate in 25 ml of PCR Buffer (20 pmol each), denatured at 100°C for 5 min, immediately
for AdvantageÔ Taq (BD Biosciences, Clontech) containing chilled on ice and added to the hybridization mixture. Dot-blot
200 mM of each dNTP, 0.4 mM of each primer and 3 U of hybridization was performed at 68°C overnight in a buffer
Advantage Taq DNA polymerase (BD Biosciences, containing 63 SSC, 53 Denhardt's, 0.5% SDS and 100 mg/ml
Clontech). PCR was carried out in a thermal cycler (MJ salmon sperm DNA. The membranes were washed under high
Research) as follows: 22 cycles at 95°C for 20 s, 65°C for 30 s, stringency conditions and exposed to X-ray ®lm (Renex,
72°C for 2 min. The PCR products were diluted 1000-fold and Russia) for 2 days.
re-ampli®ed with A2, T3 and T4 primers (Table 1) in the
second PCR round (17 cycles at 95°C for 20 s, 65°C for 30 s, Ampli®cation of the second round PCR products
72°C for 2 min). The results of dot-blot hybridizations were veri®ed by PCR to
check a correlation of signal intensity with the abundance of
Hybridization membranes preparation the corresponding ¯anking DNA in the amplicon used as a
Primers against ¯anking sequences of 38 human-speci®c hybridization probe. PCR ampli®cations used T3 or T4 LTR
HERV-K(HML-2) LTRs were designed using the `Gene primers (Table 1) with the primers speci®c for particular LTR
Runner (Version 3.00)' program (Hastings Software, Inc.). ¯anking sequences (primer sequences and PCR conditions are
They were synthesized and used for PCR ampli®cation of shown in table 2 of the Supplementary Material), and the
human placenta DNA. Aliquots of the PCR products (50 ng second round PCR products as templates.
PAGE 3 OF 8 Nucleic Acids Research, 2004, Vol. 32, No. 3 e31

Figure 1. Scheme of the method. After restriction and adaptor ligation (stage 1), an unmethylated pool of RE ¯anks is selectively PCR ampli®ed (stage 2)
and hybridized (stage 3) with an (micro)array of the whole pool of the same RE-¯anking sequences. Methylated and non-methylated HpaII (HhaI) restriction
sites are shown by short vertical lines with black circles or without circles, respectively. REs and oligonucleotide adaptors are shown as green and
black-and-white boxes, respectively. Primer positions and orientations are indicated by arrows.

Bisul®te sequencing using primers T3 or T4 (Table 1) and primers corresponding to

LTR-¯anking sequences (primer sequences are available on
RE ¯anks were sequenced using agarose beads by a modi®ed
request). The PCR mixture contained 10 ng of the digested
method for bisul®te-based cytosine methylation analysis, as
DNA in 25 ml of PCR buffer for Taq DNA polymerase
described (16). Lymph node-derived DNA was digested with
(Promega) containing 200 mM of each dNTP, 0.4 mM of each
EcoRI restriction endonuclease, denaturated and then treated
primer and 1 U of Taq DNA polymerase (Promega). PCR was
with bisul®te to convert all unmethylated cytosines into carried out in a thermal cycler (MJ Research) as follows: 28
uracils. This reaction did not affect methylated cytosine cycles at 95°C for 20 s, 61°C for 30 s, and 72°C for 1 min.
residues. For each conversion, 100 ng of lymph node DNA
was taken. The converted DNA fragments were subjected to
two rounds of PCR ampli®cation using a nested primer
approach (Table 2) and then cloned into a TA cloning vector RESULTS
(Promega). Four clones for each RE-¯anking fragment were Technique rationale
sequenced. The cytosine residues converted with bisul®te are
The technique we have developed is aimed at whole-genome
identi®ed in the course of sequencing as Ts.
identi®cation of DNA methylation patterns of interspersed
repeats (interspersed repeat-containing loci) and is based on
PCR of genomic DNA digested by HpaII
the DNA array technique that can be easily adapted for DNA
Cerebellum- and lymph node-derived DNAs were digested as microchip technology. The technique uses two basic proper-
described above in Genomic DNA isolation, restriction and ties of interspersed repeats: (i) characteristic conservative
adaptor ligation. The digested DNAs were then PCR ampli®ed sequence features that can be used for selective simultaneous
e31 Nucleic Acids Research, 2004, Vol. 32, No. 3 PAGE 4 OF 8

¯anks of HERV-K(HML-2) LTRs obtained by selective PCR

ampli®cation of the genomic DNA (Fig. 1, stage 2) digested
with a methyl-sensitive endonuclease, such as HpaII or HhaI
(Fig. 1, stage 1). Therefore, the selection resulted in only those
genomic fragments bordered by the nearest to the LTR non-
methylated restriction sites.
To make the ampli®cation of the target fragments as
speci®c as possible, we used the PCR suppression effect
through ligation of special PCR suppressive adaptors to the
restriction fragments (Fig. 1, stage 1). This modi®cation
prevents PCR ampli®cation of the fragments lacking REs,
with the adaptors on both termini, but does not preclude the
ampli®cation of RE-containing fragments by PCR initiated
from sites within REs.
Comparative analysis of the distribution of methylation
sites neighboring the LTRs in human cerebellum and
lymph node
The genomic DNAs isolated from cerebellum and lymph node
and digested with methyl-sensitive restriction endonuclease
HpaII were ligated to the suppressive adaptors (see Materials
and Methods). The ®rst round of the PCR-suppressive
selective ampli®cation was followed by nested primer PCR
to improve the selectivity. The amplicons obtained were
labeled with 32P and used as hybridization probes.
Figure 2. Dot-blot hybridization with cerebellum (A and B) and lymph Individual ¯anking regions of 38 human-speci®c HERV-K
node (C) DNA-derived probes. LTRs are numbered from left to right, and (HML-2) LTRs were prepared by PCR ampli®cation using
from top to bottom. In (A) and (C), position A1 corresponds to LTR1, A2
to LTR2, etc., B1 to LTR13, etc., C1 to LTR25, etc., D1 to LTR37, and D2 primers against the outermost parts of the LTR (T3 or T4,
to LTR38. Spots D3 in (A) and (C) and A11 in (B) correspond to 50 ng of Table 1) and a unique region of the adjacent genomic ¯ank,
phage l DNA. Designation of spots in (B) corresponds to that in (A). Rows respectively (primer sequences are available on request). All
A and B in (B) represent the same materials spotted in duplicate. the 38 ¯anking sequences obtained were dotted on Hybond-N
nylon membranes. One of the identical membranes was then
isolation of ¯anking sequences for all REs; and (ii) different hybridized with a human cerebellum DNA-derived probe, and
¯anking sequences that allow reliable discrimination of the other with a probe prepared from human lymph node
individual RE integration sites. Thus, REs can be considered DNA.
natural repetitive universal markers of a multitude of genomic Figure 2 shows the resulting hybridizations of the human-
loci. Here we used these markers for evaluation of the speci®c LTR-¯anking sequences with the human cerebellum
methylation status of genomic regions surrounding HERV-K and lymph node DNA-derived probes. The results of the
LTRs that were integrated in the human genome after the hybridizations were used to identify the methylation status of
divergence of the human and chimpanzee lineages. The CpG dinucleotides within various HpaII cleavage sites in the
principles of the approach used are schematized in Figure 1. LTR-¯anking sequences. Although a similar number of
The sequences representing ¯anks of REs were immobi- positive hybridization signals (approximately 15) was
lized on a solid support as a DNA array. These RE-¯anking observed with the cerebellum (Fig. 2A) and lymph node
sequences can be prepared by any available technique (Fig. 2B) probes, the intensities and patterns of the signals
(chemical synthesis, PCR ampli®cation, etc.). In the present were signi®cantly different (see Table 3). The differences in
work, we used ¯anks of HERV-K(HML-2) LTRs prepared by intensities might be due to different methylation levels of
selective PCR ampli®cation (Fig. 1). various HpaII sites that lead to the different abundance of the
A pool of genomic RE-¯anking regions was prepared by corresponding fragments in the hybridization probes. Some of
selective PCR ampli®cation with one of the primers repre- the HpaII sites in the LTR-¯anking regions have the same
senting a universal RE-speci®c sequence designed from the methylation status in both tissues. For example, the HpaII site
consensus of the RE type under investigation. The other adjacent to LTR28 (C4 position in Fig. 2) seems to be non-
primer was also universal but targeted at the arti®cial adaptors methylated in both cerebellum and lymph node. In contrast,
attached to the termini of the restriction fragments obtained by the HpaII site adjacent to LTR2 (A2 position in Fig. 2) is
digestion of the genomic DNA with a methyl-sensitive apparently heavily methylated in both tissues. Some loci have
restriction endonuclease. The labeled amplicon was used for a different methylation status in these tissues. In particular, the
hybridization with the array (Fig. 1, stage 3). Such a LTR18-adjacent region (B6 position in Fig. 2) is more
hybridization is supposed to reveal only those spots on the methylated, and the ¯ank of LTR27 (C3 position in Fig. 2) is
array that correspond to non-methylated RE-¯anking restric- less methylated in lymph node than in the cerebellum-derived
tion sites. DNA.
In our case, we used HERV-K(HML-2) LTR-speci®c The hybridization signals did not depend on probe or dotted
primers, and the amplicon/hybridization probe thus contained DNA lengths or on the presence of various repetitive elements
PAGE 5 OF 8 Nucleic Acids Research, 2004, Vol. 32, No. 3 e31

Table 3. Probe and dotted DNA characteristics for analyzed LTRs

Accession no. Length of probe fragments (bp); Length of dotted fragments (bp); Repeats within ¯ank sequences Hybr signal
positions in corresponding positions in corresponding for dotted fragments
accession no. accession no.
Cerebellum Lymph node

LTR1 AC044819 1170 (130 901±131 973) 280 (130 901±131 181) 1±250 bp HERV7 ± ±
LTR2 AC025420 1910 (46 613±48 529) 120 (46 613±46 736) 1±180 bp L2 ± ±
LTR3 AC010267 1430 (106 778±108 212) 130 (106 778±106 909) 1±90 bp MSTA ± ±
LTR4 AC002508 430 (33 657±34 088) 300 (33 657±33 952) 20±150 bp MLTJ ± ±
LTR5 AL354855 490 (71 833±72 321) 210 (71 833±72 042) 120±250 bp L1 ± +
LTR6 AC012146 990 (24 102±25 089) 70 (24 102±24 171) No repeats ± ±
LTR7 AL671681 1120 (167 436±168 552) 100 (167 436±167 534) 1±120 bp MLT1A1 + +
LTR8 AL139421 530 (69 211±69 744) 820 (69 211±70 031) No repeats ± +
LTR9 AL592220 320 (93 880±94 203) 270 (93 933±94 203) 1±60 bp AluSg ± ±
LTR10 AL359701 520 (61 942±62 464) 650 (61 817±62 464) 1±700 bp HERV9 + ±
LTR11 AC021987 1100 (20 313±21 412) 50 (21 360±21 412) 1±20bp LTR26 + +
LTR12 AC074117 530 (154 755±155 289) 160 (155 132±155 289) No repeats ± ±
LTR13 AL139404 600 (60 431±61 027) 50 (60 977±61 027) 1±50 bp LTR49 ± ±
LTR14 AC084028 50 (141 209±141 258) 150 (141 209±141 356) 1±120 bp MLT1A1 ± ±
LTR15 AL162412 1610 (66 382±67 997) 210 (67 791±67 997) 150±250 bp AluSx ± ±
LTR16 AL139090 580 (1536±2115) 280 (1536±1816) No repeats + ±
LTR17 AC024884 170 (43 103±43 169) 40 (43 103±43 144) No repeats ± ±
LTR18 AC000389 180 (9548±9731) 680 (9052±9731) No repeats + +
LTR19 AC113425 240 (41 008±41 248) 202 (41 046±41 248) No repeats ± +
LTR20 AC021294 1410 (145 000±146 412) 200 (146 216±146 412) 1±200 bp L2 ± ±
LTR21 AC091895 1080 (102 415±103 493) 210 (102 415±102 625) 1±240 bp L1PA13 + ±
LTR22 AL162723 210 (19 165±19 370) 130 (19 240±19 370) 1±50 bp MSTA ± ±
LTR23 AC099661 1590 (149 214±150 807) 160 (150 643±150 807) No repeats ± +
LTR24 AC074019 950 (111 321±112 273) 330 (111 947±112 273) 1±300 bp L1MC ± ±
LTR25 AL359644 550 (93 908±94 457) 40 (94 417±94 457) No repeats ± ±
LTR26 AP000812 1020 (74 214±75 234) 400 (74 838±75 234) 1±40 bp MER61 + +
LTR27 AC002400 380 (109 107±109 490) 330 (109 107±109 437) 1±150 bp L2 + +
LTR28 U47924 410 (97 456±97 869) 660 (97 206±97 869) 250±500 bp AluSg + +
LTR29 AP001591 700 (52 405±53 106) 110 (52 405±52 510) 70±150 bp MER58A + +
LTR30 Z84493 90 (8120±8210) 520 (7689±8210) No repeats + +
LTR31 AC002350 850 (45 768±46 614) 260 (45 768±46 027) No repeats + +
LTR32 AC073898 900 (78 284±79 177) 150 (79 029±79 177) No repeats ± +
LTR33 Z80898 310 (6071±6380) 270 (6071±6337) 1±250 bp L1PA1 + +
LTR34 AC007326 1120 (35 759±36 874) 770 (36 100±36 874) 1±150, 450±550 bp AluJ, ± ±
150±450 bp ERV1
LTR35 AC023074 1060 (119 305±120 365) 330 (120 037±120 365) 120±160 bp L1PA8A ± +
LTR36 AC006432 1810 (10 078±11 894) 410 (10 078±10 490) No repeats + ±
LTR37 AL109763 210 (72 566±72 773) 670 (72 566±73 233) 1±700 bp L2 + +
LTR38 AP001631 120 (108 235±108 355) 110 (108 235±108 345) No repeats + ±

in these DNAs (see Table 3). This ®nding suggests that the were in good accord with those presented in Figure 2.
signal intensity correlates with the abundance of the corres- Figure 2A and B demonstrates the hybridization reproduci-
ponding ¯ank DNA in the amplicon used as the hybridization bility as well as the total reproducibility of relative intensities
probe. To verify this correlation, seven LTR ¯anks were PCR of the signals in independent experiments.
re-ampli®ed using speci®c primers with the amplicon as a
template. The amount of each individual ¯ank DNA was PCR ampli®cation through HpaII sites con®rms the
estimated by the minimal number of PCR rounds suf®cient to methylation status determined by the DNA array
visualize the band on electrophoregrams of the PCR products. The methylation status of some HpaII sites was additionally
For all of the investigated LTRs, the number of PCR cycles veri®ed by PCR through them using primers hybridizing on
was found to be in a good reciprocal correlation with the either side of the sites and DNA digested with HpaII. In this
intensities of the corresponding hybridization signals on the case, the DNA fragments with an internal methylated HpaII
array. The results demonstrated that the intensity of the site are supposed to be ampli®ed, while the ampli®cation of
hybridization signals was mostly proportional to the content of the fragments with an unmethylated site will fail because such
the corresponding product in the hybridization probe sites are not protected from digestion with HpaII.
(amplicon). As an example, Figure 3 presents PCR ampli®cations for
To con®rm the technique reproducibility, we repeated the three arbitrarily chosen LTR ¯anks. It can be seen that the
hybridization using independently prepared probes and ®lters. HpaII sites adjacent to LTR30 and LTR28 are unmethylated in
In all cases, the relative intensities of the hybridization signals both lymph node and cerebellum DNAs, while the ¯ank of
e31 Nucleic Acids Research, 2004, Vol. 32, No. 3 PAGE 6 OF 8

Figure 3. PCR through HpaII sites of interest. (A) Cerebellum- and lymph node-derived DNAs were digested by the methyl-sensitive restriction endonuclease
HpaII and ampli®ed with primers speci®c to either side of the HpaII site (e.g. P1 and P2, or P3 and P2, where P2 is the primer against the LTR sequence).
Methylated or unmethylated CpG/CCGG sites are designated by ®lled and empty circles, respectively. (B) Gel electrophoresis of the PCR products after
ampli®cation for the LTR28 (lanes 1±3), LTR30 (lanes 4±6) and LTR14 (lanes 7±9) ¯anks through adjacent HpaII sites. Lanes 2, 5 and 8, and 3, 6 and 9
represent PCR products generated by ampli®cation of HpaII-digested DNAs from cerebellum and lymph node, respectively. Lanes 1, 4 and 7 correspond to
the ampli®ed DNA fragments from native placenta. Lane 10, length marker.

Figure 4. Results of bisul®te sequencing. For each of three LTR-¯anking sequences, four clones were sequenced. The methylation results obtained for each
of the clones are schematically presented at the bottom of the ®gure as lines of circles. Filled (black) and empty circles designate methylated and
unmethylated CpGs, respectively.

LTR14 is methylated in both tissues. These and our previous conclusions to be drawn. (i) For all the sequences analyzed,
data on LTR methylation analysis (17) are in good accord with the results on the methylation status of HpaII sites obtained
the results of the DNA array hybridization. with all the techniques used (¯ank DNA arrays, PCR
ampli®cation of HpaII-digested genomic DNAs, and bisul®te
Con®rmation of the correlation of the methylation status sequencing) agreed well with each other. (ii) The methylation
of HpaII sites and neighboring CpGs by bisul®te status of HpaII sites in all three studied cases coincided with
sequencing the methylation status of the nearest CpG dinucleotides, thus
Three arbitrarily chosen LTR ¯anks were taken for bisul®te demonstrating that methylation follows the cooperative prin-
sequencing (Table 2). CpG dinucleotides within the ¯ank of ciple due to which methylated and unmethylated CpGs are
LTR12 were found to be predominantly methylated, whereas clustered. This feature of methylation was previously reported
the LTR11 ¯ank was unmethylated in lymph node-derived as `methylation spread' (4). Due to this property, the
DNA. The results shown in Figure 4 allow one the following methylation status of a particular CpG within such a cluster
PAGE 7 OF 8 Nucleic Acids Research, 2004, Vol. 32, No. 3 e31

is characteristic of the whole cluster. Accordingly, a pro®les and therefore for studying epigenetic effects of REs
methylated/unmethylated recognition site of a methylation- that is still a black box of genome functioning analysis.
sensitive restriction endonuclease within an extended region
can serve as a tag of the methylation status of this region.
SUPPLEMENTARY MATERIAL
Supplementary Material is available at NAR Online.
DISCUSSION
Although REs are often considered inert components of the ACKNOWLEDGEMENTS
genome, there is evidence of their participation in genome The authors thank Boris O. Glotov for critical reading of the
functioning (6,13,18±21). However, genome-wide analysis of manuscript and valuable comments. The work was supported
the functional status of LTRs is still at the very beginning and by INTAS 01-0759 and the Russian Foundation for Basic
restricted to some individual LTRs chosen more or less on a Research 01-04-48900 and 2006.200054 grants, and by the
random basis. The novel technique described here allows one Physico-Chemical Biological Program of the Russian
to perform systematic genome-wide analysis of the methyla- Academy of Sciences.
tion status of CpG sites neighboring LTRs. These sites might
serve as tags of methylation of extended regions harboring the
LTRs. In turn, the methylation status of a genomic region may REFERENCES
re¯ect its functional status. A rationale for such a suggestion is 1. Costello,J.F. and Plass,C. (2001) Methylation matters. J. Med. Genet., 38,
a widely discussed methylation spread effect (4), i.e. an ability 285±303.
2. Jaenisch,R. and Bird,A. (2003) Epigenetic regulation of gene expression:
of methylation to spread over adjacent genome regions. The how the genome integrates intrinsic and environmental signals. Nature
results obtained here with bisul®te sequencing are in agree- Genet., 33 Suppl., 245±254.
ment with this effect. Indeed, Figure 4 clearly demonstrates 3. Bird,A.P. (1986) CpG-rich islands and the function of DNA methylation.
clustering of methylated and unmethylated CpGs. Thus we Nature, 321, 209±213.
can consider HpaII sites with their methylation status as tags 4. Turker,M.S. (2002) Gene silencing in mammalian cells and the spread of
DNA methylation. Oncogene, 21, 5388±5393.
of the methylation status of the wider surrounding region. 5. vandeLagemaat,L.N., Landry,J.R., Mager,D.L. and Medstrand,P. (2003)
The technique described here allowed us to show that some Transposable elements in mammals promote regulatory variation and
human-speci®c LTRs are differentially methylated in human diversi®cation of genes with specialized functions. Trends Genet., 19,
tissues, their methylation status being probably linked with the 530±536.
expression of neighboring genes. The technique can be used 6. Whitelaw,E. and Martin,D.I.K. (2001) Retrotransposons as epigenetic
mediators of phenotypic variation in mammals. Nature Genet., 27,
for analysis of methylation patterns of any low and medium 361±365.
copy number REs of genomes. In particular, it might be useful 7. Mi,S., Lee,X., Li,X., Veldman,G.M., Finnerty,H., Racie,L., LaVallie,E.,
for studying certain subgroups of such important human Tang,X.Y., Edouard,P., Howes,S. et al. (2000) Syncytin is a captive
genome constituents as ERVs, SINEs and LINEs. Although retroviral envelope protein involved in human placental morphogenesis.
Nature, 403, 785±789.
these RE families are very abundant in the human genome,
8. Domansky,A.N., Kopantzev,E.P., Snezhkov,E.V., Lebedev,Y.B.,
being represented by ~4.5 3 105, 1.6 3 106 and 8.7 3 105 Leib-Mosch,C. and Sverdlov,E.D. (2000) Solitary HERV-K LTRs
individual elements (22), respectively, each family consists of possess bi-directional promoter activity and contain a negative regulatory
a lot of less abundant subfamilies ampli®ed during various element in the U5 region. FEBS Lett., 472, 191±195.
evolutionary periods. Using our technique, each subfamily can 9. Landry,J.R., Rouhi,A., Medstrand,P. and Mager,D.L. (2002) The Opitz
syndrome gene Mid1 is transcribed from a human endogenous retroviral
be analyzed separately. The youngest HERV, L1 and Alu promoter. Mol. Biol. Evol., 19, 1934±1942.
subfamilies comprise from tens to hundreds of members (23± 10. Schon,U., Seifarth,W., Baust,C., Hohenadl,C., Er¯e,V. and Leib-
25). Some of these recently inserted REs are still not ®xed in Mosch,C. (2001) Cell type-speci®c expression and promoter activity of
the human species, and it would be interesting to estimate their human endogenous retroviral long terminal repeats. Virology, 279,
impact on the functioning of the host genome. 280±291.
11. Mager,D. and Medstrand,P. (2002) In Gardiner,K. (ed.), Encyclopedia of
The technique can easily be applied to any family of repeats the Human Genome. Nature Publishing Group, http://www.ehgonline.net.
providing that their copy number is not too high, because the 12. Medstrand,P., van de Lagemaat,L.N. and Mager,D.L. (2002)
main restriction of the method is the complexity of Retroelement distributions in the human genome: variations associated
hybridization probes prepared by selective PCR ampli®cation. with age and proximity to genes. Genome Res., 12, 1483±1495.
High complexity of the probes results in weak hybridization 13. Sverdlov,E.D. (2000) Retroviruses and primate evolution. Bioessays, 22,
161±171.
signals. 14. Sambrook,J. and Russell,D.W. (2001) Molecular Cloning: A Laboratory
There are also some limitations to the technique regarding Manual, 3rd edn. Cold Spring Harbor Laboratory Press, Cold Spring
tissue samples due to their heterogeneity that leads to Harbor, NY.
heterogeneous methylation of DNA isolated from different 15. Buzdin,A., Khodosevich,K., Mamedov,I., Vinogradova,T., Lebedev,Y.,
Hunsmann,G. and Sverdlov,E. (2002) A technique for genome-wide
cells of one and the same tissue. However, positive identi®cation of differences in the interspersed repeats integrations
hybridization signals always mean the presence of unmethy- between closely related genomes and its application to detection of
lated sites in a given sample. Sorting of cells using, for human-speci®c integrations of HERV-K LTRs. Genomics, 79, 413±422.
example, a cell sorter might improve the results. 16. Olek,A., Oswald,J. and Walter,J. (1996) A modi®ed and improved
We hope that the technique described here will be helpful method for bisulphite based cytosine methylation analysis. Nucleic Acids
Res., 24, 5064±5066.
for a better understanding of the role of REs in genome 17. Khodosevich,K., Lebedev,Y. and Sverdlov,E. (2004) Tissue-speci®c
functioning and molecular evolution. It is an ef®cient and methylation of human-speci®c long terminal repeats of endogenous
inexpensive tool for genome-wide analysis of methylation retroviruses. Rus. J. Bioorg. Chem., 30, in press.
e31 Nucleic Acids Research, 2004, Vol. 32, No. 3 PAGE 8 OF 8

18. Brosius,J. (1999) RNAs from all categories generate retrosequences that Initial sequencing and analysis of the human genome. Nature, 409,
may be exapted as novel genes or regulatory elements. Gene, 238, 860±921.
115±134. 23. Salem,A.H., Kilroy,G.E., Watkins,W.S., Jorde,L.B. and Batzer,M.A.
19. Khodosevich,K., Lebedev,Y. and Sverdlov,E. (2003) Endogenous (2003) Recently integrated Alu elements and human genomic diversity.
retroviruses and human evolution. Comp. Funct. Genom., 3, 494±498. Mol. Biol. Evol., 20, 1349±1361.
20. Lower,R., Lower,J. and Kurth,R. (1996) The viruses in all of us: 24. Boissinot,S., Chevret,P. and Furano,A.V. (2000) L1 (LINE-1)
characteristics and biological signi®cance of human endogenous retrotransposon evolution and ampli®cation in recent human history.
retrovirus sequences. Proc. Natl Acad. Sci. USA, 93, 5177±5184. Mol. Biol. Evol., 17, 915±928.
21. Sverdlov,E.D. (1998) Perpetually mobile footprints of ancient infections 25. Gifford,R. and Tristem,M. (2003) The evolution, distribution and
in human genome. FEBS Lett., 428, 1±6. diversity of endogenous retroviruses. Virus Genes, 26, 291±315.
22. Lander,E.S., Linton,L.M., Birren,B., Nusbaum,C., Zody,M.C.,
Baldwin,J., Devon,K., Dewar,K., Doyle,M., FitzHugh,W. et al. (2001)