ABSTRACT
In order to analyse further the genomic distribution of repetitive sequences in
the
Arabidopsis
genome, we have identified and characterized seven novel repetitive sequences.
Analysis of genomic representation, genomic location and DNA sequence divided
the seven repeated sequences into two classes. The first was represented by
three cosmid subclones (182A, 74A, 191A) carrying sequences that hybridised to
up to 20 genomic fragments and showed sequence homology to the genes,
Arabidopsis
CCR2
,
Arabidopsis
MYB
and to various ATP-binding transport proteins. These multi-gene families mapped to various positions within the genome, as
judged by hybridization to YAC clones constituting the
Arabidopsis
physical map. The second class was represented by four cosmid subclones (106B,
164A, 163A, 278A) that hybridised to between 20 and 300 genomic fragments. One
of these, 106B, is a diverged, partial copy of the LTR of the
Arabidopsi
s retrotransposon
Athila.
The other three sequences showed no homology to known genes or proteins. The
distribution of these sequences on chromosome 4 was analysed and sequences
hybridizing to 106B, 164A and 163A were found exclusively at the centromeric
region of this chromosome. Their detailed arrangement at the centromeric region
of chromosome 4, relative to other repeated sequence families and single copy
sequences, was determined.
Many large plant genomes have a high content of repetitive DNA, both tandem and
dispersed. Tandemly repeated sequences are generally found at the centromeres
and telomeres (
1
,
2
) and are often associated with constitutively condensed heterochromatin (
3
). Retrotransposon families form a major component of dispersed repetitive DNA (
4
) and can occur at very high copy numbers in some genomes. For example, the
BIS
-1 retroelement constitutes ~5% of the barley genome (
5
) and similar examples have been shown in the wheat and lily genomes (
6
).
In contrast with this picture, the genome of
Arabidopsis thaliana
is relatively small (~100 Mb, reviewed in
7
) and has a low repetitive DNA content (~25%). Since
Arabidopsis
is an opportunistic wild plant, it is argued that there has been strong
selective pressure for a short generation time resulting in reduced cell cycle
time and genome size. If this is the case then any repetitive DNA that has been
maintained may be functionally significant.
Of the repetitive DNA in the
Arabidopsis
nuclear genome,
~10% is highly repetitive and a further 10% middle repetitive (
7
). Three tandemly repeated sequences constituting ~2% of the genome have been characterised, a 180 bp
Hin
dIII repeat, a related 500 bp
Hin
dIII repeat and a 160 bp repeat (
8
-
10
).
In situ
hybridisation experiments demonstrated that the 180 bp tandemly repeated
sequences co-localised with the heterochromatin surrounding the centromeres on all five
chromosome pairs (
11
,
12
). Schmidt
et al
. (
13
) have recently shown that at least six other classes of repeated sequences
flank these tandemly repeated sequences around the centromere on chromosome 4.
The telomeric repeats have also been defined, each telomere carrying ~350 copies of a 7 bp tandemly repeated sequence (
14
).
The major component (7-8% of the genome) of the middle repetitive fraction of the
Arabidopsis
nuclear genome is the rDNA, localised within the nucleolus organizing regions
on chromosomes 2 and 4. The rest of the middle repetitive DNA is made up of
dispersed repeat elements. If these averaged 1-2 kb in length and were distributed randomly, then ~600 dispersed repeats would be distributed, one per 125 kb (
15
). This long interspersion pattern distinguishes
Arabidopsis
from large plant genomes such as maize where ~ 80% of the genome consists of repetitive sequences (
16
) which are found interspersed with blocks of middle repetitive and low copy
sequences (e.g. the
Adh1
gene;
17
,
18
). There has been little work addressing the identification of these dispersed
repeated sequences in the
Arabidopsis
genome. The only two characterised retrotransposon families in
Arabidopsis
are the
Ta
family (
19
,
20
) and the
Athila
family (
21
,
22
) which occur in 15 and ~150 copies respectively. Most gene families so far characterized in
Arabidopsis
have relatively few members, however, there are some exceptions, for example,
the [beta]-tubulin gene family which has nine members (
23
).
To characterise repetitive DNA within the
Arabidopsis
genome further, 300 cosmids were examined for the presence of novel repeated
sequences and the results are described here. The sequences identified were
analysed to determine their representation in the genome, what proportion were genic, whether they were clustered in the
genome and whether they were associated with specific chromosomal areas.
A five genome equivalent cosmid library was constructed (C. Lister and C. Dean,
unpublished) by cloning size fractionated
Sau
3A partially digested Columbia genomic DNA into the
Bam
HI polylinker site of the pLAFR3 vector (
24
). The average insert size of the library was 25 kb (I. Bancroft and C. Dean,
personal communication). It has been used extensively in the generation of
cosmid contigs for use in the EC genome sequencing project and these
experiments have shown that the majority of inserts faithfully represent
genomic DNA. Cosmid DNA was prepared using a method adapted from (
25
) and (
26
).
Plant genomic DNA was prepared from the Columbia ecotype using the protocol from
(
27
).
DNA was blotted and fixed onto Hybond-N nylon membrane following the recommended protocol (Amersham). Columbia
genomic DNA (100 ng) labelled with [[alpha]-
32
P]dCTP by random primer extension, was used to probe Southern blots. Fragments
used as probes were gel purified away from vector sequences twice using a Qiaex
gel extraction kit (Qiagen). All hybridisations were performed at 65oC and washing was carried out at 65oC in 0.1* SSC, 0.1% SDS.
YAC colony and Southern blot hybridisation protocols were performed as described
in (
28
).
Standard protocols were used to subclone the restriction fragments into the
Bluescript SK
+
(Stratagene) vector as outlined in (
29
). Ligations were transformed by electroporation into SURE recombination
deficient, electrocompetent cells (Stratagene).
Double-stranded DNA sequencing was performed on 2 [mu]g of DNA from each subclone using a Pharmacia T7 sequencing kit.
Sequencing and PCR primers were made using a Pharmacia LKB Gene Assembler Plus
oligo-machine.
Sequencing data was compiled and analysed using the UWGCG sequencing packages (
30
). Database searches were performed using FASTA and BLASTX programmes.
Copy number was determined by comparing the hybridisation intensity (on a
Southern blot) of dilutions of each subclone insert (equivalent to 1-100 copies assuming a genome size of 100 Mb) with 1 [mu]g of Columbia genomic DNA digested with the relevant enzyme to the
subclone.
To identify uncharacterised repetitive DNA sequences, 300 randomly selected
cosmids from a library containing DNA from the
Arabidopsis
ecotype Columbia were hybridised with Columbia total genomic DNA. Seventy-two cosmids showed a stronger hybridisation signal than cosmids containing
low copy sequences and therefore were considered to contain repetitive DNA. The
average cosmid size was ~ 25 kb and so the 72 cosmids represent ~1.8% of the genome.
The 72 cosmids were hybridised with probes containing the total chloroplast
genome and coding and intergenic regions of the rDNA and 5S rDNA. The cosmids
were also probed with three characterised repetitive DNA sequences; a high copy
tandemly repeated 180 bp sequence in pAL1 (
8
), a 500 bp repeat (
9
) and the minisatellite arabms1 (
31
). Of the initial 72 cosmids, 65 hybridised to known repetitive sequences (27 to
chloroplast DNA, 22 to the 180/500 bp tandem repeat family, 13 to rDNA and
three to arabms1) and were not studied further. They may, however, carry novel
repetitive sequences linked to those previously characterized.
Table 1
The smallest restriction fragment within each of the seven remaining cosmids
that hybridised strongly to
Arabidopsis
genomic DNA was subcloned (summarized in Table
1
). These were called 164A, 106B, 278A, 163A, 74A, 191A and 182A. Only part of a
large repetitive element or one of multiple repetitive elements from each
cosmid would be analysed using this approach. The copy number of the cross-hybridizing sequences was estimated by determining the number of
restriction fragments in the
Arabidopsis
genome hybridising to each subclone. Reconstruction experiments showed that
164A hybridised to ~150 fragments (five of which were the same size as the subclone) and 106B
to ~300 (~100 of which were the same size as the subclone) (Fig.
1
). 278A hybridised to ~20 fragments, 163A to ~90 fragments and 191A and 182A hybridised to ~20 restriction fragments each (data not shown).
The repeated sequences were mapped onto the
Arabidopsis
genome by taking advantage of the physical map of chromosome 4 generated in YAC clones (
13
). All the available subclones were hybridized to colony filters carrying YAC
clones from the CIC library (
32
) and positive hybridization was confirmed by Southern blot analysis. The
subclones, 164A and 106B, hybridized to a proportion of the YAC clones to which
the respective whole cosmids had hybridized. The cosmids, CC164 and CC106, had
been found to hybridize to YAC clones that exclusively mapped to the
centromeric region of chromosome 4 and cross-hybridising sequences were not detected elsewhere on the chromosome (
13
). The hybridization profile of each subclone was analysed with respect to the
relative overlap of the YAC clones between markers mi233 and mi87 and around
HY4/nga8. These overlaps created intervals varying between 10 and 300 kb to
which the repetitive sequences could be mapped (Fig.
2
). 164A hybridised to multiple intervals located across the whole centromeric
region including the interval mapping in between the 180 bp repeat loci that
contained the single-copy marker mi87. 163A was found associated with 164A but was detected
only on one side of the 180 bp repeat loci, 106B related sequences were more
clustered around the 180 bp repeat loci, being closely associated with
sequences cross-hybridizing to 164A in this area. There was one 106B locus located someway
within the short arm of chromosome 4, close to the marker mi233.
The DNA sequences were analysed for distinct features such as tandem arrays,
microsatellite sequences and direct and inverted repeat motifs. Four clones,
163A, 106B, 278A and 164A, contained direct and inverted repeat motifs of
between 9 and 24 bp in one or two copies.
Homology searches to all available DNA and peptide sequence databases were
performed. 182A (EMBL accession no. X93610) was found to have significant DNA
homology to the
CCR2
gene (93% in the 276 bp overlap). This gene encodes a glycine-rich protein, contains an RNA-binding motif containing and is a member of a small gene family with
approximately six members (
34
). 182A hybridised to two CIC YAC clones positioned on the physical map of
chromosome 2.
The sequence analysis of 74A (EMBL accession no. X93607) revealed two
distinctive features, a (GA)
38
microsatellite repeat in the middle of the clone and a region showing homology
to the MYB class of transcriptional regulators. The
MYB
genes in
Arabidopsis
and
Zea mays
contain two ~60 amino acid imperfect repeats at the N-terminus, thought to be necessary for DNA binding. The predicted
peptide sequence of 74A shows strong homology to these conserved repeats in the
Arabidopsis
MYB
gene
GL1
(55% in the first repeat and 75% in the overlap region of the second repeat)
,
a gene required for
trichome differentiation in
Arabidopsis
(
35
).
(GA)
n
microsatellites, like the dinucleotide repeat found in 74A, are one of the least
abundant microsatellites in
Arabidopsis
(
36
)
.
They are highly polymorphic and estimated to occur once every 244 kb. To
determine whether the GA microsatellite or the
MYB
homologous sequences contributed to the highly repetitive pattern of this
clone, PCR primers were designed to separately amplify the microsatellite and
the
MYB
DNA binding region. The resulting PCR products were used to probe Columbia
genomic Southern blots. The
MYB
PCR probe hybridised to seven
Eco
RI fragments whereas the microsatellite hybridised to ~100 fragments (data not shown) demonstrating that the (GA)
38
microsatellite accounted for the hybridisation pattern of 74A. This was further
confirmed by analyzing the YAC clones corresponding to 74A using Southern blot
analysis. The
MYB
homologous sequences hybridised to only two of the 16 putative loci for 74A,
which had been anchored on chromosomes 4 and 5.
191A (EMBL accession no. X93606) showed significant nucleotide sequence homology
to the
Arabidopsis
EST at 8149 DNA, 85% in the 69 bp overlap. It also showed predicted peptide
homology to various ATP-binding transport proteins, including the
Drosophila
white protein (sp:P10090). Comparison with other ATP-binding transport proteins identified a putative ATP binding site. 191A
hybridized to ~18 fragments in Columbia genomic DNA but to only one CIC YAC clone, the
position of which on the physical map is unknown. This would suggest that all
the associated sequences to 191A are clustered within one region.
The subclone 106B (EMBL accession no. X93611) was found to have significant DNA
sequence homology to the long terminal repeats (LTR) of the recently
characterised
Athila
retrotransposon (EMBL accession no. X81801) (
21
,
22
). The clone showed 70% DNA sequence homology to the left LTR and 69% homology
to the right LTR. The level of homology suggests that the sequence in clone
106B represents a diverged copy of an LTR from this group of retroelements.
The remaining three subclones, 163A (EMBL accession no. X93608), 164A (EMBL
accession no. X92080) and 278A (EMBL accession no. X92081) had no significant
homology to any database entries or to each other.
We describe here the identification of novel middle repetitive DNA sequences in
the
A.thaliana
ecotype Columbia genome. This fraction of the
Arabidopsis
genome is relatively uncharacterised. Analysis of genomic representation,
chromosomal location and DNA and predicted peptide sequence divided the
repeated sequences into two classes summarised in Table
1
. The first class represented gene sequences, with up to 20 associated genomic
fragments. From the proportion of the repeats that fell into this class, it is
likely that a high proportion of the dispersed repeated sequences in the
Arabidopsis
genome are made up of large gene families.
The second class of sequences had no homology to known genes. They were present
in between ~20 and 300 copies and were clustered in the genome. No highly repetitive
DNA sequences other than those previously characterised by (
8
) and (
9
) were detected in the initial cosmid screening experiments.
One of the subclones, 106B, was found to be a diverged, partial sequence of the
LTR of the
Athila
retroelement. It is possible that the cosmid CC106 contains an intact copy of a
related element. The
Athila
retrotransposon is found in up to 150 copies in the
Arabidopsis
ecotype C24 (
21
,
22
). It has been estimated that up to 200 retroelements may be present in the
Arabidopsis genome (
37
), thus
Athila
is likely to constitute a significant number of these. Since the estimated copy
number of 106B is ~300, this may indicate that all the LTRs are associated with intact
elements (one at each end). Diverged copies of the
Athila
element have been found and are likely to represent inactive remnants of old
integration events (
22
). One
Athila
element was found integrated within the 180 bp repeat tandem array and from a
distribution analysis of
Athila
within YAC and [lambda] libraries Pelissier
et al
. (
22
) concluded that the elements were concentrated at the centromeric regions. Our
analysis of the distribution of repeated sequences, including 106B, across the
centromeric region of the physical map of chromosome 4 shows that sequences
related to 106B and hence the LTRs of
Athila
are clustered in this region particularly around the 180 bp repeat arrays. We
have also detected sequences related to 106B up to 1.5 Mb away from the 180 bp
repeat arrays.
Only a relatively small fraction of the genome, ~7.5% (300 cosmids of average size 25 kb), was analysed in this study and it
is clear from other studies that a large number of other middle repetitive
elements, not detected in this study, also co-localize with the paracentromeric heterochromatin in chromosome 4. These
include sequences present on two RFLP markers, m456 and mi167, and two YAC end-probes, CIC5C6LE and CIC6D7RE (
13
). Preliminary analysis of cosmid CC106 revealed that it hybridized to five
additional intervals, as compared with 106B, across the centromeric region of
chromosome 4. This suggests the presence of middle repetitive sequences, other
than 106B, in CC106. Thus a lot more analysis is required before we will be
close to fully characterizing the repetitive DNA of the
Arabidopsis
genome.
If the YAC clones hybridizing to the 180 bp tandem repeat loci do represent at
least the core of the centromere (
13
), then it is interesting to compare the arrangement of the repetitive sequences
relative with other characterized centromeric regions. A recent study
characterising the centromere of
Drosophila
minichromosome
Dp1187
found that a central core of ~220 kb containing complex DNA, of single copy and middle repetitive
sequences, was essential for centromere function (
38
). A region of ~200 kb, either side of the essential core, containing highly repetitive DNA
sequences was also required for completely normal inheritance. This region was
believed to be involved in sister chromatid adhesion and indirectly assist in
kinetochore formation. The centromeric region of the fission yeast
Schizosaccharomyces pombe
has also been characterised and displays some of the features of the
Drosophila
centromere. A central core of 6.8 kb of single copy DNA is directly flanked by
moderately repetitive sequences one of which, the K-type repeat contains a region 2.1 kb long which is critical for centromere
formation. The central region is flanked by ~100 kb of satellite DNA (
39
). The examples of the
Drosophila
and
S.pombe
centromeres allow us to draw a comparison with the centromeric region of chromosome 4. 164A cross-hybridizing sequences are found in the ~100 kb interval between the 180 bp satellite repeat loci. The RFLP
marker mi87, that maps to this interval, hybridises to a single genomic
fragment so at least some of this DNA is not repetitive. This region might
equate to the essential centromere core. Families of other repetitive sequences
are also present up to 1.5 Mb away from the 180 bp tandem repeat loci on each
chromosome arm and these may represent the non-essential functional flanking DNA. It will be interesting to see if the
asymmetry of the 163A hybridizing sequences is functionally significant.
The repeated sequences comprising the heterochromatin around centromeres are
generally species-specific (
40
). However, some centromeric repeats have been shown to be chromosome-specific, for example the pBcKB4 and pBoKB1 repeats in
Brassica
chromosomes (
41
). The repetitive element in 278A would appear to be an example of this class of
repeat. It is associated with 164A, 106B and 180 bp repeat loci sequences but
is not present on the YAC contigs currently available for the centromeric
region of chromosome 4. The availability of the physical maps for the other
chromosomes, in combination with
in situ
hybridisation analysis, will allow the distribution of repeats such as these to
be analysed and to further examine the organization of
Arabidopsis
centromeres.
The authors would like to thank Joanne West, Karina Love and Zoë Lenehan for their contributions to the physical map, and Dr Clare Lister
for the cosmid library and her advice. We should also like to thank Dr Keith
Mitchelson for the arabms1 probe and Dr Richard Macknight for proofreading.
This work was funded by a BBSRC grant (PG208/PG0608) to C.D. and a BBSRC
studentship to H.L.T.
*To whom correspondence should be addressed
Present addresses:
+
Molecular Genetics Laboratory, Instituto Giannina Gaslini, Largo G. Gaslini 5,
16148 Genova-Quarto, Genova, Italy and
[sect]
Max-Delbrück-Laboratorium in der MPG, Carl-von-Linne-Weg 10, 50829 Köln, Germany
[para]
X92080, X92081, X93606-X93608, X93610 and X93611
Subclone
Restriction fragment
Size (kb)
Homology
Location
Copy number
cloned
164A
Hin
dIII
0.9
centromeric on chr 4, also mapping to other centromeric regions
~150
106B
Eco
RI
0.4
centromeric on chr 4, also mapping to other centromeric regions
~300
278A
Eco
RI/
Bgl
II
1.3
centromeric but not on the present chr 4 contigs
~20
163A
Hin
dIII
0.45
centromeric on chr 4, also mapping to other centromeric regions
~90
74A
Eco
RI/
Bgl
II
0.7
MYB
dispersed on chr 4 and 5 and unmapped on other chromosomes
~7 (for the MYB homology)
191A
Eco
RI
0.7
WHITE
unknown
~18
182A
Bgl
II
1.8
CCR2
chr 2
~20
REFERENCES
Return