Native DNA repeats and methylation in Ascobolus
Christophe
Goyon*
,
Jean-Luc
Rossignol
and
Godeleine
Faugeron
Institut de Génétique et Microbiologie, CNRS URA 1354, Université Paris-Sud, Bâtiment 400, F-91405
Orsay
Cedex,
France
Received May 28, 1996;
Revised and Accepted July 11, 1996
EMBL accession nos X99080-X99093 (incl.)
ABSTRACT
We identified two classes of native dispersed DNA repeats in the
Ascobolus
genome. The first class consisted of several kilobase long, methylated repeats. These repeats, named
Mars
(
m
ethylated
A
scobolus
r
epeated
s
equences), fell in one family of
LINE
-like elements and in three families of LTR-containing retrotransposable elements. The methylation features of
Mars
elements were those expected if they were natural targets for the MIP
(methylation induced premeiotically) previously discovered in
Ascobolus
. The second class consisted of short repeats,
~100 bp long, corresponding to 5S rRNA and tRNA genes. As expected from their
size, which was too small to allow MIP to occur, they were unmethylated, as were 26 kb of unique sequences tested. These observations are consistent
with the hypothesis that MIP is targeted at natural DNA repeats and constitutes
a defensive process against the detrimental consequences of the spreading of
mobile elements throughout the genome. The 9 kb tandem repeats harbouring the
28S, 18S and 5.8S rRNA genes displayed methylation features suggesting that
rDNA methylation proceeds through a process other than MIP.
INTRODUCTION
Repeated genes artificially introduced in the filamentous ascomycete
Ascobolus immersus
by integrative transformation are frequently silenced as a consequence of the
process of methylation induced premeiotically (MIP;
1
,
2
). MIP occurs only during the sexual phase of the life cycle, at a premeiotic
stage between fertilization and karyogamy (
1
). It acts efficiently on duplicated kilobase sized DNA segments, whether linked
or unlinked and regardless of their endogenous or exogenous origin (
2
,
3
). MIP leads in both repeats to the methylation of all C residues, whether or
not they belong to symmetrical sequences (
4
). Shorter duplications are poor targets for MIP and/or the maintenance of the
resulting methylation, the minimal size required for a repeat to undergo
detectable MIP being close to 300 bp for a tandem duplication and close to 600
bp for an unlinked duplication (
5
).
The occurrence of MIP suggested that this process was directed against native
DNA repeats. This hypothesis was tested by looking for the presence of DNA
repeats within the
Ascobolus
genome and assessing their methylation status. Among the different native DNA repeats identified, all the kilobase sized repeats, which were
mostly retroelements or relics of retroelements, displayed the methylation expected if they had undergone MIP. In contrast, our
results suggest that the tandemly arranged rDNA repeats are methylated by a
different process.
MATERIALS AND METHODS
Isolation and manipulation of DNA
Ascobolus
DNA was purified by CsCl/EtdBr centrifugation as described previously (
6
). Large scale preparation of plasmid and phage [lambda] DNA was performed using Qiagen columns (Qiagen, Germany). Small scale
preparation of plasmid DNA used for sequencing was performed using the BioRad
Prep-A-Gene plasmid purification kit. Otherwise, standard techniques were
used (
7
). Plasmid vector pBluescript II KS (Stratagene, USA) was used for subcloning.
To study methylation with restriction enzymes two pairs of isoschizomers were
used (
8
): (i)
Sau
3AI and
Nde
II, which recognize the sequence GATC;
Sau
3AI does not cut this site if the C on one of the strands is methylated, while
Nde
II is insensitive to C methylation. (ii)
Eco
RII and
Bst
NI recognize the sequence CCWGG;
Eco
RII does not cut this site if the internal C is methylated, while
Bst
NI is insensitive to C methylation. Another restriction enzyme was used,
Bst
UI, which does not cut its recognition site CGCG if the external C is methylated
(
8
).
Cloning of the rDNA unit
Ascobolus
DNA was digested with a mixture of five restriction enzymes,
Asp
718,
Bam
HI,
Pst
I,
Pvu
II and
Bgl
II. The first four enzymes do not cut the rDNA unit, while
Bgl
II cuts once in the 9 kb unit. After agarose gel electrophoresis, the 9 kb DNA
fraction, hence highly enriched in rDNA units, was extracted from the gel and
cloned in the
met2
-bearing plasmid pCG5 (
9
) cut with
Bam
HI and
Bgl
II and dephosphorylated. Plasmid pCG32 was thus obtained and shown to contain an
rDNA unit by restriction mapping and by using its 9 kb insert as a probe in
Southern hybridization experiments.
Ascobolus
strains
The wild-type strain used for DNA isolation was RN42 (
4
). Strain f8-1 resulted from the homologous integration of one copy of plasmid pCG32
into the rDNA locus of a (
met2.
n::
amdS
) strain carrying a deletion of
met2
(
5
). Homologous integration was obtained by targeted transformation (
10
) with plasmid pCG32 linearized by cutting at the single
Sal
I site present in the rDNA insert. The number of integrated pCG32 copies was
measured using a radioimager, as described below for assessing the number of
copies of the repeated sequences.
DNA sequencing
All plasmid inserts were sequenced either partially or totally on one strand
only. In most cases, sequencing was performed on unidirectionally partially
deleted subclones obtained by digestion of the DNA with exonuclease III, using
the double-stranded Nested Deletion kit (Pharmacia). Sequencing primers were the M13
Universal and M13 Reverse primers. In some cases, internal specific primers
were used to complete the sequence. Fragments were sequenced on an Applied
Biosystems Model 373A DNA Sequencing System using the ABI PRISM Ready Reaction
DyeDeoxyterminator Cycle sequencing kit.
Nucleotide sequence accession numbers
The nucleotide sequences have been submitted to the EMBL Nucleotide Sequence
Database under the following accession numbers: X99080 for the
Mars
1 copy present on plasmid pCG20; X99081 for the
Mars
1 copy present on plasmid pCG27; X99082 and X99083 for
Mars
2; X99084 for
Mars
3; X99085 for
Mars
4; X99086 for
Mars
5; X99087 for 5Sa; X99088 for 5Sb1 and 5Sb3; X99089 for 5Sb2; X99090 for 5Sc;
X99091 for the part of plasmid pCG68 that harbours ta and tb; X99092 for the
other part of the insert of plasmid pCG68; X99093 for tc.
PCR amplification, cloning and sequencing of bisulphite-treated DNA
The bisulphite genomic sequencing method was based on that described by Frommer
et al.
(
11
), modified by Goyon
et al
. (
4
). Bisulphite-treated
Ascobolus
DNA was from the same batch as that used in a previous study (
4
), in which ~2% of C residues were resistant to the treatment and remained unconverted.
The primers used for PCR amplification of the transcribed strands of the P1 and
P2 regions were: (P1) cg11, CTACAGCCTGTCAACACT, and cg12, ATGTTGGAGAGTGTGAGT;
(P2) cg13, CACCACACATTATCCCTT, and cg14, CAGAAAGTGCGTGGAGAC. The primers used
for PCR amplification of the non-transcribed strand of the rDNA C6-7/D7 region were: cg15, AGATCTTGGTGGTAGTAG, and cg16,
AAAACTATTCCTTCCACC. Amplification reactions were carried out as follows: 45 s
at 94oC, 20 s at 53oC, 45 s at 72oC for 35 cycles, with an additional 10 min at 72oC at the end of the last cycle. The amplified fragments were
gel purified and cloned without any further treatment into
Eco
RV-digested plasmid pBluescript II KS. Individual cloned molecules were
sequenced using the conditions described above.
Estimation of the copy number of the repeated sequences
Plasmids pCG20, pCG35, pCG45, pCG57, pCG61, pCG68 and pCG80-2 containing repeated DNA sequences were dotted on a nylon membrane
together with a plasmid containing a unique sequence and hybridized with
32
P-labelled total
Ascobolus
DNA. The amount of
32
P-labelled DNA that hybridized to each dot was determined using a
radioimager [SOFI, Scintillating Optical Fiber Imager (
12
); Quartz et Silice, France]. The membrane was then stripped of the first probe,
hybridized with
32
P-labelled pBluescript DNA and analysed again using the radioimager. This
allowed us to normalize the values obtained with the first probe for the
quantity of each plasmid that was effectively dotted on the membrane. The
plasmid containing a unique sequence was used as a reference to define the
standard level of counts expected per length unit for a sequence present as a
single copy in the
Ascobolus
genome. Knowing (or, in some cases, estimating by analogy with known elements)
the length of the repeated elements contained in the plasmid inserts, we
estimated the number of copies of these elements per genome by dividing the
number of counts per length unit of each element by the standard level defined
for a unique sequence.
RESULTS
Isolation of
Ascobolus
repeated DNA sequences
Forty [lambda] clones were picked at random from a genomic library made from
Ascobolus
DNA (
13
). It could be assumed that these clones represented a faithful sampling of the
Ascobolus
genome since the library did not show any large bias in the genome
representativeness. Indeed, 20 unique sequences were found to be all present ~30 times (V.Colot, personal communication). DNA of the 40 phages was
digested with
Eco
RI (
Eco
RI cut on each side of the
Bam
HI cloning site, releasing the two [lambda] arms plus the
Eco
RI fragments from the insert) and analysed by Southern hybridization, using
total
Ascobolus
DNA as probe. Fragments containing sequences that were repeated in the genome
(and therefore proportionally over-represented in the probe) were expected to show a stronger hybridization
signal than fragments containing unique sequences. Out of the 40 phages
studied, 27 contained only fragments displaying weak hybridization signals, of
similar intensities from phage to phage, corresponding to the hybridization of
unique sequences. The remaining 13 phages showed one or more fragments
displaying a stronger signal, indicating that they contained repeated DNA.
These 13 phages were probed with
Ascobolus
mitochondrial DNA and rDNA (coding for the 18S, 5.8S and 26S rRNAs) and shown
not to contain any of these sequences. Cross-hybridizations were then performed, which resulted in the preliminary
distribution of the 13 phages into eight classes, one class being constituted
of three and another one of four cross-hybridizing phages, the other six being each constituted of a single
phage.
Eco
RI fragments containing repeated DNA from 11 of the 13 phages were then cloned
in a plasmid vector. The two remaining phages belonged to the three phage class
and the four phage class respectively. When used as probes in Southern
hybridization experiments, these two phages displayed, for the repeated part,
the same hybridization patterns as their cross-hybridizing counterparts. When there was a choice among several fragments,
the most strongly hybridizing one was selected for further analyses. In some
cases, large fragments were further subcloned, after characterization by
Southern hybridization of a shorter strongly hybridizing fragment bearing most
of the repeated sequence. Then, DNA sequencing was performed as extensively as
necessary to gather enough information to establish the nature of the repeated
DNAs and to study their methylation status. Six repeated sequences corresponded
to retrotransposable elements, four to 5S rRNA genes and two to tRNA genes. The
remaining one was unidentifiable.
The
Mars
elements
The 1.85 and 2.47 kb
Eco
RI fragments carried by plasmids pCG27 and pCG20 respectively came from two ([lambda]16 and [lambda]36) of the three cross-hybridizing phages. These fragments were sequenced
totally.
Except for its 5' first 177 bp, the pCG27 insert displayed 97% nucleotide (nt) identity
with the 3' region of the pCG20 insert. Among the 45 nt differences, 32 (71%) were
transitions and 13 (29%) were transversions. The pCG27 insert is likely to be a
truncated copy of an element belonging to the same family as the one partially
contained in the pCG20 insert. This family was named
Mars
1 (methylated
Ascobolus
repeated sequence 1; the methylation status will be described below). The
Mars
1 region present in pCG20 is repeated ~60 times in the
Ascobolus
strain used (not shown). Conceptual translation reveals an open reading frame
(ORF) of 746 amino acids (aa) in the pCG20 insert. The 3' part of this ORF is found with 98% aa identity in the pCG27 insert. The
pCG20 insert ORF contains the eight blocks of conserved residues from the
reverse transcriptase (RT) domain (Fig.
1
A and B) that are found in the
LINE
-like elements (
14
), which are non-LTR retrotransposons. The first of these blocks is found exclusively in
this class of retrotransposons. This, added to the fact that a 5' truncated copy was found in the pCG27 insert, indicates that
Mars
1 is a
LINE
-like element. Pairwise comparison (Fig.
1
B) of the RT domain of
Mars
1 with other
LINE
-like elements contained in databases revealed that
Mars
1 RT is most related to the
Neurospora crassa Tad
RT (
15
), sharing with it 35% aa identity and 53% aa similarity.
The 5S rRNA genes
Phages [lambda]6, [lambda]7, [lambda]29 and [lambda]37, which cross-hybridized weakly, were expected to
harbour the same family of repeated elements. All but [lambda]7 were subcloned, resulting in a 2 kb [lambda]6
Hin
cII subfragment (pCG38), a 2.6 kb [lambda]29
Eco
RI subfragment (pCG61) and a 0.8 kb [lambda]37
Bst
UI subfragment (pCG80-2) carrying the repeats (Fig.
5
A). Total or partial sequencing of these three subfragments followed by
comparison with sequences contained in databases revealed that pCG38 and pCG80-2 each contain one copy (named 5Sa and 5Sc respectively) of the 119 bp 5S
rRNA genes and pCG61 contains at least three copies (5Sb1, 5Sb2 and 5Sb3) of
them. We estimated from dot-blot hybridization that the
Ascobolus
strain used contains at least 60 of these genes (not shown). Figure
5
B shows the alignment of sequences of the five cloned copies together with the
most similar 5S rRNA gene found in databases, the 5S rDNA gene isotype beta
from
N.crassa
(5S[beta]). The five
Ascobolus
copies are all divergent, representing five different isotypes. If the 5Sa copy
is taken as a reference, the other copies present from 91.5 to 96% identities.
Identities between the 5S[beta] gene from
N.crassa
and the
Ascobolus
copies range from 76 to 80%. In
Ascobolus
5S rRNA genes appear not to be arranged in tandem, since we showed that large
genomic segments could contain only one 5S rRNA gene. Even the three linked
copies contained in pCG61 are not tandemly arranged, being separated by unique
sequences and potentially transcribed from different strands. This situation is
similar to that found in
N.crassa
, in which 5S rRNA genes are dispersed (
18
).
The tRNA genes
Two other non-cross-hybridizing repeated DNA sequences present on [lambda]30 and [lambda]35 were characterized, being respectively harboured
(Fig.
6
A) by a 3.3 kb
Eco
RI subfragment (pCG68) and a 2 kb
Bam
HI-
Bst
UI subfragment (pCG77-2). The pCG77-2 insert was totally sequenced; 2.9 kb of the pCG68 insert was
sequenced. Comparison with sequences contained in databases revealed that pCG68
contained two genes (ta and tb) that might both code for tRNA
Asn
and that pCG77-2 contained a relic of a tRNA gene (tc) (Fig.
6
A).
Both genes of pCG68 are interrupted by an intron, of 37 bp in ta and of 6 bp in
tb. The two exons, 74 bp in total, differ by 1 bp located in the T[Psi]C loop and present 78% identity with the
Saccharomyces cerevisiae
tRNA
Asn
gene (Fig.
6
B). The predicted cloverleaf tRNA structure of ta is shown in Figure
6
C. The tb gene is likely to be non-functional, since Haselbeck and Green (
19
) demonstrated in
Xenopus
oocytes that a pre-tRNA with a 6 nt intron was incompletely spliced. We estimated from dot-blot hybridization that
Ascobolus
contains about nine copies of tRNA genes homologous to ta and tb (not shown).
The tc sequence from pCG77-2 resembles a tRNA
Gln
gene (not shown). However, it contains one mismatch in the T[Psi]C stem and lacks 1 nt in the acceptor stem. In addition, the tc sequence is
interrupted by a 17 bp intron that is located at the junction between the D
stem and the anticodon stem instead of being located at the invariable
position, i.e. 1 nt 3' of the anticodon sequence. It is thus very unlikely that tc encodes a
functional tRNA. A 287 bp PCR product including tc was used for probing genomic
DNA digests (not shown). About 10 hybridization bands showed up, indicating
that tc or part of it is indeed repeated in the
Ascobolus
genome.
We analysed methylation at the single
Nde
II/
Sau
3AI restriction site (Fig.
6
A and B) present in tb only.
Sau
3AI and
Nde
II genomic DNA digests were analysed by Southern hybridization (Fig.
3
). Besides fragments due to hybridization of other tRNA gene copies homologous
to ta and tb, two hybridizing
Nde
II fragments (315 and 195 bp) characteristic of the genomic region containing ta
and tb were expected (Fig.
6
A). These two fragments were observed in both
Nde
II and
Sau
3AI digests (Fig.
3
), indicating that the GATC site in tb was not methylated. Additionally, both
digests displayed identical hybridization patterns, indicating that none of the
analysed GATC sites were methylated. This result suggests that this family of
tRNA genes is unmethylated, since it is likely that some of the hybridizing
copies contained one GATC site, similarly to tb.
The 18S, 5.8S and 26S rRNA genes (rDNA)
The 9 kb rDNA repeats encoding the 18S, 5.8S and 26S rRNAs are reiterated ~110 times at a single locus in
Ascobolus
(V.Colot, personal communication). We cloned one unit and used it as a probe in
Southern hybridization performed with genomic DNA digested with
Nde
II and
Sau
3AI (Fig.
3
). The almost complete or the partial disappearance of some of the
Sau
3AI fragments showed clearly that the rDNA was methylated. However, methylation
at GATC sites was quite heterogeneous, some sites being methylated either in
almost all rDNA repeats or in only a fraction of them and some sites being
totally unmethylated.
A comparison with unique DNA sequences
Eight DNA sequences totalling ~26 kb were shown to be present as single copies in the
Ascobolus
genome. Five of them encompassed the
met2
gene (
10
), the
b2
gene (
13
) and genes encoding actin, [beta]-tubulin and a putative zinc finger protein (unpublished)
respectively. The three others remained uncharacterized (unpublished). Southern
hybridization failed to show any methylation for any of these eight sequences
(not shown).
We compared the G+C content of five unique sequences (totalling 19 kb) and of
the sequenced parts of the five
Mars
repeats (totalling 14 kb).
The G+C content ranged from 0.49 to 0.57 for the unmethylated unique sequences
and from 0.49 to 0.55 for the methylated
Mars
elements. This suggests that C -> T transitions by deamination of methylated C residues is efficiently
prevented in
Ascobolus
, unless the pristine
Mars
elements displayed a G+C content higher than that of the
Ascobolus
genome. Nor was any difference found in the expected to observed ratios of CpG
dinucleotides, in contrast to the observation made by Kricker
et al
. (
20
), who showed that most repeated DNA sequences in vertebrates contain
substantial deficits of CpG dinucleotides compared with most unique or
unmethylated sequences.
DISCUSSION
Besides the rDNA tandem repeats, we identified two classes of native DNA
repeats: the several kilobase long, densely methylated repeats and the short, ~100 bp long, unmethylated repeats.
The first class consists of the
Mars
elements, present at 20-60 copies. Partial DNA sequencing and cross-hybridization experiments indicated that six out of the seven
Mars
elements analysed were likely to be retroelements or relics of them. Three
belonged to the
LINE
-like
Mars
1 group;
Mars
2,
Mars
3 and
Mars
4 elements were related to LTR-retrotransposons. The seventh element,
Mars
5, remained unidentified. Methylation of
Mars
elements and methylation resulting from MIP exhibit the same features: they
densely affect all C residues in the overall population of DNA molecules, with
a preference for C residues belonging to CpG dinucleotides (
4
). These observations fit with the idea that the
Mars
elements are natural targets for MIP. Similar non-canonical methylation has also been found in plants (
21
) and in
N.crassa
(
22
), in which it also affects native DNA repeats (
23
-
25
).
ACKNOWLEDGEMENTS
We thank A. Grégoire, E. Lemichez, F. Malagnac, A. Pokorska, F.-X. Sicot and J. Delaruelle for help with some experiments and
Almuth Collard for the photograph work. We also thank J. L. Barra and V. Rocco
for critical reading of the manuscript and suggestions for improvement. This
work was supported by the Association pour la Recherche sur le Cancer (contract
6200) and the Groupement de Recherches et d'Etudes sur les Génomes (contract 34/93).
REFERENCES
1 Faugeron,G., Rhounim,L. and Rossignol,J.-L. (1990) Genetics, 124, 585-591.MEDLINE Abstract
28 Finnegan,E.J., Bretell,R.I.S. and Dennis,E.S. (1993) In Jost,J.P. and Saluz,H.P. (eds), DNA Methylation: Molecular Biology and Biological Significance. Birkhaüser Verlag, Basel, Switzerland, pp. 218-261.
29 Bird,A.P., Taggart,M.H. and Gehring,C.A. (1981) J. Mol. Biol., 152, 1-17.MEDLINE Abstract
30 Russell,P.J., Rodland,K.D., Cutler,J.E., Rachlin,E.M. and McCloskey,J.A. (1985) In Timberlake,W.E. (ed.), Molecular Genetics of Filamentous Fungi. Alan R.Liss, New York, NY, pp. 321-332.
32 Doerfler,W. (1991) Hoppe-Seyler Z. Biol. Chem., 372, 557-564.
33 Matzke,M., Matzke,A.J.M. and Mittelsten Scheid,O. (1994) In Paszkowski,J. (ed.), Homologous Recombination and Gene Silencing in Plants. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp. 271-307.
34 Flavell,R.B. (1994) Proc. Natl. Acad. Sci. USA, 91, 3490-3496.MEDLINE Abstract
