Published online before print
January 14, 2003, 10.1101/gr.593403
Vol 13, Issue 2, 195-205, February 2003
LETTER
Centromere Satellites From Arabidopsis Populations: Maintenance of Conserved and Variable Domains
Sarah E. Hall1,2,
Gregory Kettler2,3 and
Daphne Preuss2,3,4
1Committee on Genetics, 2Howard Hughes Medical
Institute, 3Department of Molecular Genetics and Cell
Biology, University of Chicago, Chicago, Illinois 60637, USA
 |
ABSTRACT
|
|---|
The rapid evolution of centromere sequences between species has led
to a debate over whether centromere activity is sequence-dependent. The
Arabidopsis thaliana centromere regions contain  20,000
copies of a 178-bp satellite repeat. Here, we analyzed satellites from
41 Arabidopsis ecotypes, providing the first broad population
survey of satellite variation within a species. We found highly
conserved segments and consistent sequence lengths in the
Arabidopsis satellites and in the published collection of
human  -satellites, supporting models for a functional role. Despite
this conservation, polymorphisms are significantly enriched at some
sites, yielding variation that could restrict binding proteins to a
subset of repeat monomers. Some satellite regions vary considerably; at
certain bases, consensus sequences derived from each ecotype diverge
significantly from the Arabidopsis consensus, indicating
substitutions sweep through a genome in less than 5 million years. Such
rapid changes generate more variation within the set of
Arabidopsis satellites than in genes from the chromosome arms
or from the recombinationally suppressed centromere regions. These
studies highlight a balance between the mechanisms that maintain
particular satellite domains and the forces that disperse sequence
changes throughout the satellite repeats in the
genome.
[Supplemental material is available online at
www.genome.org.]
The large heterochromatic constrictions known as
centromeres play many roles in multicellular eukaryotes, holding sister
chromatids together during the early stages of mitosis and, at later
stages, assembling the kinetochores that bind to microtubules and
mediate chromosome separation. These roles are conserved across
eukaryotes, yet the DNA sequences that mediate centromere function have
remained undefined in many cases. Overall, the sequence composition of
the centromere region varies considerably across species, raising the
possibility that epigenetic modifications, and not DNA sequence per se,
govern centromere function (Choo 2000 ; Henikoff et al. 2001).
Genetically, the DNA sequence composition of a centromere is defined as
the portion of homologous chromosomes that segregate to opposite poles
in meiosis I. The boundaries of such genetic intervals are defined by
recombination events, and these intervals are often large, given the
limited recombination in the centromere regions. In
Arabidopsis, the genetically defined centromere regions
contain large satellite arrays comprised of thousands of copies of
180-bp repeats (Martinez-Zapater et al. 1986; Copenhaver et al.
1999). We examined the patterns of satellite sequence evolution across
Arabidopsis populations and found striking conservation of
repeat sequence length as well as significantly conserved and variable
regions within the repeats. We applied the same analysis to the
previously published collections of human -satellite DNA (Choo et
al. 1991), finding similar patterns, albeit a higher degree of sequence
variation. The presence of satellite domains with strikingly different
rates of nucleotide substitution strongly indicates a
sequence-dependent role for Arabidopsis centromere satellites.
Direct evidence of sequence-based centromere function comes from the
budding yeast Saccharomyces cerevisiae, in which centromere
function was first genetically defined by tetrad analysis, and
subsequently reduced to a minimal functional region using
minichromosomes (Clarke and Carbon 1980; Cottarel et al. 1989). In this
species, the minimal DNA sequence necessary and sufficient to confer
all centromere functions is only 125 bp in length. These centromeres
are present on every chromosome and contain three conserved DNA
elements (CDE): the 5'-CDEI (8 bp), a central, A + T-rich CDEII
(78–86 bp), and a 3'-CDEIII (25 bp; Cottarel et al. 1989). Different
protein complexes assemble on each of the CDE regions; mutations in
either the CDE or various centromere-binding proteins reduce the
efficiency of chromosome segregation (Clarke 1998). Although many of
the proteins that assemble on the centromere DNA of yeast are well
understood, the machinery that mediates attachment of these regions to
microtubules for chromosome segregation remains unclear.
The centromere regions of budding yeast lack repetitive DNA, yet most
other eukaryotes examined, including the fission yeast
Schizosaccharomyces pombe, have numerous repeats at the
centromere. Typically, eukaryotic centromere regions contain repeat
units ranging in length from 150 to 210 bases, approximately the
length required to form a single nucleosome (Henikoff et al. 2001). In
humans, the centromere-specific -satellites are AT-rich 171-bp
repeats, tandemly arrayed in a head-to-tail arrangement. They are
sufficiently variable to allow classification into distinct
chromosome-specific subfamilies (Waye and Willard 1987; Choo et al.
1991). Synthetic minichromosomes that contain -satellite arrays
recruit essential centromere-binding proteins and are transmitted
through mitosis, indicating that -satellite arrays are sufficient to
confer centromere function in human cell lines (Willard 1998, 2001;
Yang et al. 2000; Schueler et al. 2001; Grimes et al. 2002).
Neocentromeres that completely lack -satellite DNA also have been
characterized (DuSart et al. 1997; Barry et al. 2000). These
centromeres form at a low frequency following disruption of a natural
centromere, and indicate that, under some circumstances, -satellite
sequences are not necessary for centromere function.
Human -satellite sequences share 90% identity with those from
gorilla, chimpanzee, or orangutan (Durfy and Willard 1990; Baldini et
al. 1991; Haaf and Willard 1998), and chimpanzee and gorilla
satellites, like those of human, are organized into higher-order arrays
(Durfy and Willard 1990; Baldini et al. 1991; Haaf and Willard 1998).
Human satellites can acquire centromere activity when introduced into
African green monkey cells (Haaf et al. 1992), indicating either that
the features required for centromere function are conserved between the
two species or that the satellite DNA sequence itself is unimportant.
Despite this functional conservation, there is considerable divergence
in array content between primate species (Waye and Willard 1989; Durfy
and Willard 1990; Warburton and Willard 1990; Haaf and Willard 1997,
1998). For example, chromosome-specific -satellites have been
reported in humans, yet homologous primate chromosomes generally do not
share the same satellite subfamilies (Haaf and Willard 1997, 1998).
Because satellites are present in thousands of copies, their divergence
between species would require genome-wide homogenization, a process
known as molecular drive (Dover 1982). Several mechanisms that could
account for this homogenization have been postulated, including gene
conversion and unequal crossing over (Smith 1976; Dover 1982; Stephan
1986; Charlesworth et al. 1994). In one model, the ancestor of closely
related species contained a "library" of satellite variants within
its genome, and as new species emerged, one satellite was predominately
used as a template, resulting in the conversion of the other genomic
copies (Mestrovic et al. 1998). Although it is attractive to postulate
that selection could drive the choice of one satellite over others,
chance could also account for biased amplification (Dover 1982; Nijman
and Lenstra 2001). The continuous homogenization of satellite sequences
within a genome can lead to a smaller within-species variation than
between-species variation, an observation known as concerted evolution
(Elder and Turner 1995).
To date, the diversification of satellites has been measured between
closely related species, but not between the populations of an
individual species. To provide a detailed understanding of the initial
steps of satellite divergence, we characterized the satellites in
geographically separated Arabidopsis thaliana populations
(accessions or ecotypes). The genome sequencing project for the
Columbia ecotype of Arabidopsis (The Arabidopsis
Genome Initiative 2000) provided >5 Mb of assembled sequence from the
genetically defined centromere intervals, and the unsequenced gaps
within each centromere region are thought to be comprised primarily of
satellite repeats (The Arabidopsis Genome Initiative 2000;
Kumekawa et al. 2000, 2001; Hosouchi et al. 2002). The sequenced
regions that flank these gaps contain numerous repetitive elements
including retroelements, transposons, microsatellites, middle
repetitive DNA, and tandemly organized satellites. The satellite
repeats are not found elsewhere in the genome, but are restricted to
the genetically defined centromere regions (Copenhaver et al. 1999; The
Arabidopsis Genome Initiative 2000). Heslop-Harrison et al.
(1999) analyzed 20 centromere satellite sequences from the Columbia
ecotype and reported an ecotype-specific consensus, noting two regions
of >99% conservation. By examining the sequence of satellites from
other ecotypes, we explored whether the previously defined consensus
could be extended to the species as a whole.
The Brassicaceae family, of which Arabidopsis is a member, has
expanded from a common ancestor to 3350 species within a time frame of
40–50 million years (Al-Shehbaz 1984; Koch et al. 2001). Satellites
homologous to the Arabidopsis 180-bp repeats have been
discovered in other members of the Brassicaceae (Hallden et al. 1987),
indicating that centromere satellites existed in a common ancestor. As
described below, the satellite repeats from Arabidopsis are
evolving rapidly among isolated populations, yet they contain highly
conserved motifs. These studies set the stage for comparing satellite
evolution patterns among the thousands of available species in the
Brassicaceae family; identification of broadly conserved domains would
imply possible selection for specific DNA sequence motifs.
|
RESULTS
|
|---|
Collection of Centromere Satellite Sequences From 41 Ecotypes
We used Polymerase Chain Reaction (PCR) to clone at least 10
satellite repeat sequences from each of 41 Arabidopsis
ecotypes (Table 1); 457 clones were
sequenced, resulting in 1029 whole or partial repeat sequences (GenBank
accession nos. AF494837–AF495294). To reduce potential bias introduced
by a particular PCR primer, two different PCR amplifications were
performed for each ecotype using nonoverlapping primer sets (Fig.
1; see Methods). For most
ecotypes, significant sequence differences were not detected between
these amplifications; however in C24, Est-0, Mv-0, and Nok-0, the two
primer sets amplified different repeat classes that varied at 4, 7, 12,
and 9 sites, respectively. The satellite repeats were aligned using the
location of the HindIII restriction site as the arbitrary
beginning of the repeat. In total, the satellite repeats have an
A + T content of 62.5%, similar to the genomic average of 65.1%
(The Arabidopsis Genome Initiative 2000).
Based on migration through agarose gels, the Arabidopsis
satellites were originally termed 180-bp repeats (Martinez-Zapater et
al. 1986); sequence analysis of the Arabidopsis satellite
repeats instead showed a mean length of 178 bp. Of the repeats we
examined, 72% were 178 bp, 18% were 177 bp, and 8% were 179 bp. In
addition, three outliers were observed at 176 bp, 182 bp, and 192 bp;
the insertion and deletion events that gave rise to these variants
differed in size and were scattered throughout the repeats. Consensus
sequences derived for each ecotype were also 178 bp, demonstrating that
repeat length is conserved across populations (Fig. 1). Similarly,
analysis of -satellite repeat length in primates has shown that
-satellite monomers are fairly constant in length, varying from 168
bp to 172 bp among species (Durfy and Willard 1990; Baldini et al.
1991; Fanning et al. 1993; Alves et al. 1994; Warburton et al. 1996;
Haaf and Willard 1997, 1998).
Consensus Satellite Sequences From Individual Arabidopsis Ecotypes and the A. thaliana Species
We derived a consensus for the satellite repeat sequences from each
Arabidopsis ecotype (Fig. 1), defining a consensus nucleotide
as the base that occurs three times more often than any other at a
given site; this definition was previously used to derive a consensus
for -satellite DNA (Waye and Willard 1987). In those cases in which
these criteria were not met, the site was noted as polymorphic and the
most predominant bases were indicated by the standard IUPAC symbols
(Fig. 1). Next, the set of sequences from the 41 ecotypes was compiled
to derive a consensus for the species; 13 of the 178 nucleotides
comprising the repeat consensus were defined as polymorphic (Fig. 1,
asterisks). These polymorphisms were also observed within individual
ecotypes, indicating that they predate ecotype divergence. A consensus
was previously reported from 20 satellite sequences from the
Columbia ecotype (Heslop-Harrison et al. 1999); the consensus we
defined for Arabidopsis differs at 15 sites, 13 of which
reflect bases that are commonly polymorphic within the species.
Interestingly, there were notable sequence differences between the
consensus derived for the species as a whole and the consensus of
individual ecotypes, indicating rapid divergence within the past 5
million years (Koch et al. 2001). We used a  2 test to
identify those substitutions that are significantly different from the
species consensus (Fig. 1, shading). In some cases, deviations were
uniquely present in one ecotype consensus (Fig. 1, 11 cases, yellow);
such substitutions were observed in Est-0, Gre-0, and Can-0. The
substitutions seen in these ecotypes provide evidence for
homogenization of new mutations across the genome in a short time
frame. Termed molecular drive, such homogenization processes serve to
increase the relative abundance of a particular variant; they can
result from selective forces or as a consequence of random chance
(Dover 1982). All of the substitutions observed in the consensus
sequences for Est-0, Gre-0, and Can-0 were observed as minor sequence
variants in some of the other ecotypes, indicating they were likely
present in the ancestral population. Although the mechanisms behind
this sequence divergence and subsequent amplification are unknown, it
is of interest that in at least one case, Est-0, four unique
substitutions are in close proximity, implying they originated from a
single event (Fig. 1).
In addition to these ecotype-specific substitutions, we also identified
152 statistically significant deviations from the species consensus
that correspond to nucleotide substitutions commonly found in multiple
ecotypes (Fig. 1, pink). In many cases, these changes reflect sites
that are more variable than in the species as a whole. Conversely, five
changes correspond to the fixation of a single nucleotide at a site
that is highly polymorphic in the population. Although broader sampling
is required to interpret the significance of these events, they provide
additional evidence that the Arabidopsis satellites are
dynamic; new mutations are likely emerging continuously, replacing, by
an undetermined mechanism, the predominant variants in the population
(Nijman and Lenstra 2001).
Measuring Sequence Variation Across the Arabidopsis Centromere Satellite Repeats
Despite mechanisms that homogenize repeat arrays across a genome,
satellite repeats nonetheless accumulate variation at an appreciable
rate. For example, human -satellite repeats have a high degree of
sequence heterogeneity, and variable sites are distributed among the
satellite monomer classes in a nonrandom manner (Waye and Willard 1987;
Choo et al. 1991). We used our entire data set of Arabidopsis
satellite repeats to measure nucleotide variability, calculating the
occurrence of the most frequent base as a percentage of all the
nucleotides sequenced at each site (Fig.
2). Averaging these data across all of the
sites in the repeat showed that most nucleotides are highly conserved,
within 1 SD of a mean of 90.3 ± 9.8% (Fig. 2A). However, 21 sites
showed more variation; 13 of these corresponded to polymorphisms
identified previously (Fig. 2A, filled circles). We replotted these
data (Fig. 2B), taking into account frequent polymorphisms (see
Methods); this adjusted plot highlighted additional sites that
exhibited unusually high variability.
We identified conserved and variable segments within the satellite
repeats by examining nucleotide occurrence frequencies over a sliding
window of 15 bases. The 15-bp conserved domains C1, C2, and C3, and the
25-bp variable domain, V1, comprised of two overlapping windows,
exhibited variation significantly different than the mean (Fig.
3A). Much of the variation within the
satellite repeats is clustered near the V1 region, which contains 5 of
the 8 highly variable sites (>2 SD from mean) and 3 of the 13
polymorphic sites. Strikingly, the same regions we identified as highly
conserved or highly variable in the species as a whole showed similar
patterns in the consensus sequences of individual ecotypes (Fig. 1).
Thus, because these patterns occur repeatedly across
Arabidopsis populations, they do not reflect chance variation
within our sample of sequences. These nonrandom patterns of evolution
within the Arabidopsis satellites strongly indicate biological
constraints on satellite sequences. Whereas highly conserved domains
may reflect important protein-binding sites, regions that exhibit
extreme variation may point to areas where strict sequence consensus is
not important. Alternatively, some sites may be under selection to
remain polymorphic, creating a diversity of repeat monomers within
arrays. In humans, such polymorphisms are organized into higher-order
repeat units that might be important in the formation and structure of
a centromere (Willard and Waye 1987; see Discussion).
View larger version (46K):
[in this window]
[in a new window]
|
Figure 3. Identification of significantly conserved and variable domains. The
percent occurrence of the most frequent base (Fig. 2A,C) was subjected
to a z-score analysis, measured over a sliding window of 15
bp. This process sets the average at zero (solid line); dashed lines
indicate ±1.2 SD. Significantly conserved windows (light gray) and
significantly variable regions (dark gray) were merged when the sliding
windows overlapped, and the entire window was represented as conserved
(C1, C2, C3) and variable (V1, V2) regions (Figs. 1 and 2).
|
|
Lastly, we examined 950 repeats from the Columbia ecotype that were
sequenced by the Arabidopsis Genome Project. These repeats are
located on the edges of the satellite arrays; recent examinations of
human -satellites show that repeats on array edges are more variable
than the repeats in the array core (Schueler et al. 2001). The Columbia
sequences from the array edges differed from the species consensus at
only 20 sites (Fig. 4), 18 of which were
frequently polymorphic in the random sample of Arabidopsis
populations (Fig. 1). Surprisingly, we found that the overall
conservation of nucleotides within this large set of Columbia repeats
was 89.4% ± 3.9%, similar to the Arabidopsis species
average (90.3% ± 9.8%) and the Columbia consensus average derived
from random sampling (91.3% ± 14.4%). We assessed monitored
nucleotide conservation across these sequences, applying the same
criteria used to generate Figure 3. The Columbia satellites from the
array edges have an expanded C3 region and V1 region, and do not
display any conservation above average in the C1 region, whereas the C2
region remains unchanged. Thus, in contrast to expectations based on
human repeats, the edges of Arabidopsis satellite arrays are
not more variable than sequences collected randomly from the genome.
These observations may reflect a fundamental difference in the
mechanisms that maintain human and Arabidopsis arrays.
View larger version (13K):
[in this window]
[in a new window]
|
Figure 4. Comparison of Columbia ecotype satellite consensus sequences. The Col-0
consensus was derived from PCR-amplified sequences obtained in this
study; the Col-edges consensus was derived using the 950 satellite
sequences available from The Arabidopsis Genome Project
(GenBank). The two consensus sequences were aligned with the species
satellite consensus (Fig. 1); dots represent identity to the species
consensus, and changes from the consensus are indicated.
|
|
Comparisons With Sequence Variation Across Human -Satellite Repeats
To compare the composition of the Arabidopsis satellites to
human -satellite DNA, we reexamined the set of 293 human sequences
compiled previously (Choo et al. 1991). In -satellite DNA, 15
polymorphic sites have been identified, similar to the number in the
Arabidopsis satellites. As with Arabidopsis, the
percent occurrence of bases at these polymorphic sites is within 1 SD
of the mean when the second-most-frequent base is considered (Fig. 2D).
Interestingly, the average percent occurrence of the most abundant
bases in the -satellites was 84.0% ± 10.7% (Fig. 2C),
indicating these repeats are significantly more variable
(P < 0.0001) than the collection of Arabidopsis
repeats, as determined by a univariate ANOVA test. This difference may
reflect the dissimilarities in population structure and mating patterns
between humans and Arabidopsis, the nearly fivefold difference
in chromosome (and centromere) number (23 vs. 5, respectively), or a
disparity in the functional roles of the repeats, accompanied by
different selective pressures.
Using the same criteria as with the Arabidopsis satellites, we
identified three regions of conservation (C1, C2, C3) and two regions
of variability (V1, V2) in the -satellite repeats (Fig. 3B). Whereas
windows of significant conservation or variability in the
Arabidopsis satellites tended to cluster, these windows were
more scattered in -satellites. Interestingly, the binding site for
the 17-bp centromere protein B (CENP-B box), defined by DNA
footprinting (Muro et al. 1992), resides in one of the variable
regions, V2, which contains five polymorphic sites (Alexandrov et al.
1993; Rovanova et al. 1996). The average occurrence of the most
frequent base across the entire CENP-B box is 78%, and when the
nucleotides essential for CENP-B binding are considered (Tanaka et al.
2001), this percentage drops to only 68%, making this region notably
more variable than the rest of the -satellite repeat. This
observation supports the model that many -satellite repeats cannot
tightly bind CENP-B, resulting in protein phasing and higher-order
chromatin structure (Yoda et al. 1998). In fact, when we surveyed a set
of 880 -satellites from GenBank, only 23% had all of the
nucleotides essential for CENP-B binding.
Variation of Single-Copy Sequences From the Recombinationally Suppressed Arabidopsis Centromeres
To better appreciate the diversity of the Arabidopsis
centromere regions, we examined the sequence variation of three
single-copy loci that are tightly linked to three different genetically
defined centromeres (CEN2, CEN3, and CEN4),
and DNA sequences from eight single-copy genes located in the
chromosome arms (Supplemental Fig. 5, GenBank accession nos.
AF494760–AF494836, AF495295–AF495335, AF495337–AF495375; available
online at http://www.genome.org). For each intron and exon, we
determined the sequence variability at each site by measuring the
average occurrence of the most common nucleotide (Table
2). The exons from chromosome arms and from
recombinationally supressed centromere regions displayed a similar rate
of variation, with an average occurrence of the most frequent
nucleotide ranging from 99.5% to 99.9% and 99.7% to 99.9%,
respectively. Although two of the three centromeric introns had
variation (95.6% and 95.8%) that was significantly different from
that of introns from the chromosome arms (ANOVA,
P < 0.0001), much of this variation is attributed to a
single large deletion event, and therefore may not be representative of
intron variation in the centromeric region.
The variation of orthologous single-copy sequences cannot be directly
compared with the variation among repetitive paralogous satellites;
nonetheless, it is of interest that none of the intron or exon
sequences showed as much variation as the collection of
Arabidopsis satellite repeats (Table 2; ANOVA,
P < 0.0001). Similarly, the nucleotide diversity of
satellite repeats was substantially higher than that of genes (Table
2). Finally, we compared transition and transversion frequencies for
the Arabidopsis and -satellite repeats to the set of
single-copy sequences (Table 3). Using the
species consensus for each sequence, we tabulated the number of
transitions and transversions for each individual sequence relative to
the consensus. As expected, exons from both the chromosome arms and
centromere regions showed more conservative changes than introns,
having 68.3% and 75.0% transitions versus 56.6% and 60.0%
transitions, respectively. In contrast, the Arabidopsis
satellite repeats and -satellite repeats had fewer transitions than
either exons or introns (40.8% and 38.0% transitions, respectively)
approaching the theoretical 33% value for a sequence that is mutating
at random.
|
DISCUSSION
|
|---|
Satellite DNA and Centromere Functions
The development of human minichromosomes supports the idea of a
functional role for satellite DNA in centromeres (Willard 2001). In
contrast, the lack of sequence conservation across species and the
prevalence of neocentromeres that lack satellite repeats have raised
questions as to whether any specific satellite sequences are required
for function (Choo 2000; Henikoff et al. 2001). In this study, we
demonstrated that the centromeric satellite repeats of
Arabidopsis have domains that are highly conserved, whereas
other portions of these repeats vary considerably. Thus, the
preservation of both conserved and variable domains across 41 different
populations, along with a strict conservation of sequence length,
strongly indicates that the evolution of the satellite repeats is
constrained.
Conserved Domains in Satellite Repeats
We used a statistical test to define three regions (C1, C2, C3) of
high average conservation (87.9% ± 1.5%, 88.4% ± 1.4%,
88.4% ± 1.8% conserved, respectively) and two variable regions (V1
and V2, 78.0% ± 3.3%, 76.1% ± 3.1% conserved, respectively)
in human -satellite DNA. None of these domains had been defined
previously. Similarly, we defined three conserved regions (C1,
95.2% ± 0.9%; C2, 94.6% ± 0.7%; C3, 95.0% ± 0.9%) and
one variable region (83.8% ± 2.4%) in the Arabidopsis
satellite repeat consensus. These domains are distinct from the two
conserved regions (Box A and Box B, 99% conservation) previously
derived for 20 Columbia satellite sequences (Fig. 2A; Heslop-Harrison
et al. 1999). Although Box A and Box B were not highly conserved across
Arabidopsis or in the sample of Columbia satellites we
obtained, some of these differences can be attributed to polymorphic
sites, and others are more likely the result of the bias inherent in a
smaller data set. The centromere satellite consensus sequence presented
here was derived from 1029 repeats from 41 Arabidopsis
ecotypes, and consequently more broadly reflects the species as a
whole.
The presence of highly conserved domains within the satellites
indicates that some repeat regions may be under selective pressure to
maintain a particular DNA sequence, whereas other regions of the repeat
evolve without constraint. One explanation for the differential rates
of substitution in the Arabidopsis satellites could be the
interaction of DNA-binding proteins with satellite DNA. In humans,
centromere-binding proteins A, B, C, E, G, and H have been identified.
Of those proteins, CENP-A, CENP-B, and CENP-C have been shown to have
DNA-binding activity (Choo 2000). CENP-A is a histone H3-like protein
that is found at active centromeres and is associated with
-satellite arrays in humans (Smith 2002). In addition, CENP-A
homologs in Drosophila and Arabidopsis appear to be
evolving adaptively, which could correlate with the sequence divergence
of satellite arrays in the centromere (Malik and Henikoff 2001; Talbert
et al. 2002). The association of CENP-A homologs with corresponding
centromeric DNA could influence the maintenance of conserved sequence
domains in the repeats.
Both CENP-B and CENP-C have been shown to associate with a subset of
-satellite repeats. However, the localizations of the two proteins
on -satellite arrays are distinct and nonoverlapping. CENP-C is
found only at active centromeres, and the exact binding site of CENP-C
within the -satellite is still unknown (Politi et al. 2002). CENP-B
is found associated with -satellite arrays at both active and
inactive centromeres; it binds to -satellite monomers at a specific
17-bp sequence named the CENP-B box (Muro et al. 1992). Interestingly,
the CENP-B box in the -satellite repeats overlaps with the highly
variable V2 region, and contains five polymorphic sites in its
consensus. Combining the insights from the recently solved
CENP-B/CENP-B box cocrystal (Tanaka et al. 2001) with the survey of
published -satellite sequences, we found four of the nine bases
essential for CENP-B binding are also polymorphic; CENP-B would be
unable to interact with a highly common base at each of these four
sites. Ikeno et al. (1994) analyzed a higher order -satellite array
comprised of 11 repeats, and found that CENP-B-binding sites are
located in alternating repeat monomers. Taken together, these results
raise the possibility that polymorphisms serve to phase CENP-B binding
within the satellite arrays, potentially aiding in the assembly of the
-satellite DNA into a higher-order structure recognized by other
centromere-binding proteins (Yoda et al. 1998; Choo 2000). Although
centromere-binding proteins from plants are less well characterized, it
is possible that a similar phasing mechanism could be operating, given
the patterns of nonrandom variation that we observed within the
Arabidopsis satellite repeats.
Conservation of Satellite Sequence Length
A requirement for uniform nucleosome phasing and the subsequent
propagation of centromeric heterochromatin has often been ascribed as
the source of the uniform satellite length observed within a species
and between closely related species (Henikoff et al. 2001). In
primates, satellite monomers vary from 168 to 172 bp (Durfy and Willard
1990; Baldini et al. 1991; Fanning et al. 1993; Alves et al. 1994;
Warburton et al. 1996; Haaf and Willard 1997, 1998). Average centromere
satellite lengths have also been determined for a wide range of other
species, including maize (156 bp; Ananiev et al. 1998), rice (159 bp;
Dong et al. 1998), and insects in the genus Palorus (143 bp;
Mestrovic et al. 1998). We found that Arabidopsis centromere
satellites were remarkably conserved within all 41 ecotypes
(178 ± 0.1 bp). The highly invariant length of Arabidopsis
satellite repeats indicates a rigid length requirement. Because
nucleosome arrays can accommodate insertions of several base pairs
without a dramatic alteration in phasing patterns (Simpson 1991), other
explanations, such as length requirements that modulate higher-order
structures across entire arrays, may be more appropriate. CENP-B, known
to bind as a dimer, may require rigid monomer length so that CENP-B
boxes are in appropriate locations within a centromere structure for
protein binding (Yoda et al. 1998). Alternatively, the length
requirement could be a result of the satellite array interaction with
specialized centromere histones, such as CENP-A (Talbert et al. 2002).
If nucleosome phasing is involved, then the diversity of satellite
lengths among Arabidopsis, maize, and rice would require
invoking a species-specific nucleosome length restriction.
The Diversity of Satellites Across Populations
Despite the presence of conserved domains, many portions of the
satellite repeats exhibit notable variation. Considered as a whole, the
centromere satellite repeats are more variable across the
Arabidopsis population than any other single-copy sequence
examined, including noncoding DNA. Interestingly, the
Arabidopsis satellite repeats were significantly less variable
than -satellite repeats. Reproductive strategies may explain some of
this difference; because Arabidopsis is self-pollinating, it
is expected to have less heterozygosity and less genetic diversity than
individuals in an outcrossing population (Charlesworth and Wright
2001).
The vast number of satellite copies in the genome provides tremendous
redundancy and an enhanced opportunity for divergence; they can undergo
various mechanisms of evolution, homogenizing new changes through gene
conversion, unequal exchange, and transposition (Dover 1982). Moreover,
recombination and repair enzymes may have a limited access to
heterochromatic satellites, increasing the rate of nucleotide
substitution relative to the rest of the genome. If satellite sequences
indeed provide critical functions, this redundancy and high rate of
change could allow organisms to sample substitutions, even in
functional domains, without deleterious effects.
Evolution of Satellite DNA
Although the function of satellite DNA remains questionable (Csink
and Henikoff 1998), satellite evolution has attracted much attention.
Many studies have compared the satellites from closely related species
(Waye and Willard 1989; Grebenstein et al. 1996; Alix et al. 1998;
Mestrovic et al. 1998; Rajagopal et al. 1999; Landais et al. 2000;
Nijman and Lenstra 2001). In these analyses, homogenization of
satellite repeats within the genome has typically occurred, resulting
in less variation within a species than between closely related
species. This type of change, termed concerted evolution (Elder and
Turner 1995), likely relies on mechanisms of molecular drive: unequal
exchange, gene conversion, or transposition (Smith 1976; Dover 1982;
Stephan 1986; Charlesworth et al. 1994).
The results presented here indicate a balance between the stochastic
and selective pressures that drive satellite diversity. Our finding of
significantly conserved and variable regions across ecotypes indicates
a strong bias in the turnover of satellite sequences (Mestrovic et al.
1998). Molecular drive may account for the homogenization of 11
substitutions observed in individual Arabidopsis ecotypes that
differ from the species consensus (Fig. 1, yellow shading). Because
these 11 substitutions also occur at a low frequency in other ecotypes,
they were likely present in the ancestral parent, and homogenization of
the variant occurred since the ecotype populations diverged (Nijman and
Lenstra 2001). In addition, the precise conservation of satellite
length is particularly striking. Taken together, these observations
indicate a model in which higher-order structures have a strict
requirement for sequence length and conservation of particular repeat
regions. Satellite evolution may progress in a manner that retains all
of these features, maintaining essential protein-binding sites,
structural domains, and sites for epigenetic modification.
|
METHODS
|
|---|
Source of DNA Sequences Analyzed
Arabidopsis centromere satellite repeat sequences were
from ecotypes obtained primarily from the Arabidopsis
Biological Resource Center (ABRC), Ohio State University; Kz-8 was
obtained from Joy Bergelson, University of Chicago. DNA was extracted
from a rosette leaf of an individual plant as described (McKinney et
al. 1995). Two sets of primers (F2: 5'-AGCTTCTTCTTGCTTCTCA; R2:
5'-CCAATCACAAAACCT CAGC; and F4: 5'-GAGTCTTTGGCTTTGTATCTTC; R4:
5'-GTATACCTGAAACCGATGTGG; Fig. 1) were used to amplify satellite
repeats; PCR was performed as recommended (PanVera Corporation).
Amplification products were separated by gel electrophoresis, and the
ladder of repeats was visualized after ethidium bromide staining. Bands
measuring 180 bp, 360 bp, and 540 bp were purified, cloned (TOPO TA
kit, Invitrogen), and sequenced using the M13 forward and reverse
primers. A minimum of 10 clones was sequenced for each of the ecotypes.
The resulting 457 clones sequenced gave 1029 whole or partial repeat
sequences. For analysis of repeat length, 176 internal repeats derived
from 360-bp or 540-bp amplified bands were considered. For this study,
we did not include the 950 satellite repeat sequences from the Columbia
ecotype that were deposited in GenBank by the Arabidopsis
Genome Sequencing Project; many of these sequences reside at the
borders of satellite arrays and consequently contain biases not
representative of the genome. Furthermore, to consider all ecotypes
equally, we also excluded a set of 624 satellite repeat sequences that
were obtained on random clones from the Landsberg ecotype
(http://www.tigr.org/tdb/e2k1/ath1/atgenome/Ler.shtml).
The analysis of human -satellite DNA described here relied on data
compiled previously by Choo et al. (1991). This earlier study derived a
consensus of available human satellite sequences, and tabulated the
variation at each site.
The sequence variation among Arabidopsis ecotypes was analyzed
for 12 genes (Supplemental Fig. 5, available online at
http://www.genome.org; Table 2). These genes are all expressed, with
known EST or cDNA counterparts (The Arabidopsis Genome
Initiative 2000). We performed PCR and DNA sequencing in four cases
(ARP6, MCM5-like, SL15-like, and
ACTIN8; GenBank accession nos. AF494760–AF494836,
AF495295–AF495335, AF495337–AF495375); for the remainder, we analyzed
sequence available in GenBank (for references, see Supplemental Fig. 5,
available online at http://www.genome.org).
Sequence Analysis
Prior to analysis, primer and vector sequences were trimmed,
concatenated satellite repeat arrays were separated, and sequences were
aligned using Seqman (DNAStar). Polymorphic sites within the consensus
are indicated (Fig. 1) by IUPAC symbols: (B) C or G or T; (D) A or G or
T; (H) A or C or T; (K) G or T; (M) A or C; (R) A or G; (S) C or G; (V)
A or C or G; (W) A or T; (Y) C or T. A 2 test determined
the significance of nucleotide differences observed in individual
ecotypes. For the expected nucleotide occurrence at a given site, we
used the overall nucleotide frequencies, as determined for the 41
ecotypes combined, at each position. The data were divided into two
classes (consensus and nonconsensus) for a single degree of freedom;
differences from the consensus were considered significant when
P  0.0001. Nucleotides within a given ecotype that showed
a significant deviation in frequency from the overall species consensus
are indicated in Figure 1 (shading). Some substitutions in ecotype
consensus sequences are not shaded, as the nucleotide frequency does
not significantly differ from the overall consensus.
The percent occurrence of the most frequent base at each site was
calculated for Arabidopsis satellite repeats, -satellite
repeats, and gene sequences; for the satellites, this is plotted in
Figure 2. At polymorphic sites (i.e., sites where the most common base
is not three times more frequent than any other, filled circles in Fig.
2) either the percent occurrence of the most frequent base was
calculated (Fig. 2A,C), or the percent occurrence of the polymorphic
nucleotides, considered as a group, was considered (Fig. 2B,D). The
average percent occurrence and standard deviations are also depicted in
Figure 2; for these calculations, polymorphic sites were not treated
differently from other nucleotides. A univariate ANOVA test
( = 0.05) with a Bonferroni adjustment (Sokal and Rohlf 1997) was
used to determine if the average values for the percent occurrence
differed when Arabidopsis satellites, -satellites, and
genes were considered.
Conserved and variable regions within the Arabidopsis
satellite repeats and -satellites were defined by a sliding-window
analysis of the percent occurrence data; z-scores were used to
define windows of significantly higher or lower variation than the
average. Windows of 5 bp, 10 bp, 15 bp, and 20 bp were initially
analyzed, and results from a 15-bp window analysis are presented. The
average percent occurrence for each window was tabulated, and an
overall average and standard deviation for these window data points
were used to produce a z-score
(z = [x – µ]/ , where x is each
window data point, µ is the average of all windows, and is the
standard deviation). Windows that had a z-score of ±1.2 SD
from the mean (20% of all windows) were considered significant
(Fig. 3). For clusters of windows with significant deviations from the
mean, the window with the largest departure from the mean was used as
the center for the conserved or variable regions; the
Arabidopsis satellite repeat variable region V1 consists of
two independent overlapping windows (depicted in Figs. 1 and 2). This
analysis made it possible to use the same criteria to define conserved
and variable regions in the Arabidopsis and -satellite
repeats.
Nucleotide diversity was calculated for Arabidopsis satellite
repeats and centromere and arm genes using ARLEQUIN software (Schneider
et al. 2000). Insertions and deletions were not considered in the
calculations; the Tajima and Nei method was used by the software.
|
WEB SITE REFERENCES
|
|---|
http://www.arabidopsis.org; The Arabidopsis Information
Resource.
http://www.tigr.org/tdb/e2k1/ath1/atgenome/Ler.shtml; The Institute for
Genomic Research, Landsberg erecta random sequence Database (Ler).
|
Acknowledgements
|
|---|
We thank S. Duffy and K. Thornton for helpful discussions; and
members of the Preuss laboratory, M. Sharp, A. Hall, K. von
Besser, and K. Keith, for critical reading of the manuscript.
This work was supported in part by an NIH Training Grant in Genetics
and Regulation (S.E.H.), and by grants from the National Science
Foundation, the David and Lucile Packard Fellows Program, and the
Howard Hughes Medical Institute.
The publication costs of this
article were defrayed in part by payment of page charges. This article
must therefore be hereby marked "advertisement" in accordance with
18 USC section 1734 solely to indicate this fact.
|
Footnotes
|
|---|
4 Corresponding author. 
E-MAIL dpreuss{at}midway.uchicago.edu; FAX (773) 702-6648.
Article and publication are at
http://www.genome.org/cgi/doi/10.1101/gr.593403. Article published online before print in January 2003.
|
REFERENCES
|
|---|
Alexandrov, I.A., Medvedev, L.I., Mashkova, T.D., Kisselev, L.L., Romanova, L.Y., and Yurov, Y.B. 1993. Definition of a new satellite suprachromosomal family characterized by monomeric organization. Nucleic Acids Res. 21: 2209-2215.[Abstract/Free Full Text]
Alix, K., Baurens, F.-C., Paulet, F., Glaszmann, J.-C., and D'Hont, A. 1998. Isolation and characterization of a satellite DNA family in the Saccharum complex. Genome 41: 854-864.[Medline]
Al-Shehbaz, I.A. 1984. The tribes of Cruciferae (Brassicaceae) in the southeastern United States. J. Arnold Arbor 65: 343-373.
Alves, G., Seuanez, H.N., and Fanning, T. 1994. Satellite DNA in neotropical primates (Platyrrhini). Chromosoma 103: 262-267.[Medline]
Ananiev, E.V., Phillips, R.L., and Rines, H.W. 1998. Chromosome-specific molecular organization of maize (Zea mays L.) centromeric regions. Proc. Natl. Acad. Sci. 95: 13073-13078.[Abstract/Free Full Text]
The Arabidopsis Genome Initiative 2000. Analysis of the genome sequencing of the flowering plant Arabidopsis thaliana. Nature 408: 796-815.[CrossRef][Medline]
Baldini, A., Miller, D.A., Miller, O.J., Ryder, O.A., and Mitchell, A.R. 1991. A chimpanzee-derived chromosome-specific -satellite DNA sequence conserved between chimpanzee and human. Chromosoma 100: 156-161.[CrossRef][Medline]
Barry, A.E., Bateman, M., Howman, E.V., Cancilla, M.R., Tainton, K.M., Irvine, D.V., Saffery, R., and Choo, K.H. 2000. The 10q25 neocentromere and its inactive progenitor have identical primary nucleotide sequence: Further evidence for epigenetic modification. Genome Res. 10: 832-838.[Abstract/Free Full Text]
Charlesworth, B., Sniegowski, P., and Stephan, W. 1994. The evolutionary dynamics of repetitive DNA in eukaryotes. Nature 371: 215-220.[CrossRef][Medline]
Charlesworth, D. and Wright, S.I. 2001. Breeding systems and genome evolution. Curr. Opin. Genet. Dev. 11: 685-690.[CrossRef][Medline]
Choo, K.H. 2000. Centromerization. Trends Cell Biol. 10: 182-188.[CrossRef][Medline]
Choo, K.H., Vissel, B., Nagy, A., Earle, E., and Kalitsis, P. 1991. A survey of the genomic distribution of satellite DNA on all the human chromosomes, and derivation of a new consensus sequence. Nucleic Acids Res. 19: 1179-1182.[Free Full Text]
Clarke, L. 1998. Centromeres: Proteins, protein complexes, and repeated domains at centromeres of simple eukaryotes. Curr. Opin. Genet. Dev. 8: 212-218.[CrossRef][Medline]
Clarke, L. and Carbon, J. 1980. Isolation of a yeast centromere and construction of functional small circular chromosomes. Nature 287: 504-509.[CrossRef][Medline]
Copenhaver, G.P., Nickel, K., Kuromori, T., Benito, M., Kaul, S., Lin, X., Bevan, M., Murphy, G., Harris, B., Parnell, L.D., et al. 1999. Genetic definition and sequence analysis of Arabidopsis centromeres. Science 286: 2468-2474.[Abstract/Free Full Text]
Cottarel, G., Shero, J.H., Hieter, P., and Hegemann, J.H. 1989. A 125-base-pair CEN6 DNA fragment is sufficient for complete meiotic and mitotic centromere functions in Saccharomyces cerevisiae. Mol. Cell. Biol. 9: 3342-3349.[Abstract/Free Full Text]
Csink, A. K. and Henikoff, S. 1998. Something from nothing: The evolution and utility of satellite repeats. Trends Genet. 14: 200-204.[CrossRef][Medline]
Dong, F., Miller, J.T., Jackson, S.A., Wang, G.-L., Ronald, P.C., and Jiang, J. 1998. Rice (Oryza sativa) centromeric regions consist of complex DNA. Proc. Natl. Acad. Sci. 95: 8135-8140.[Abstract/Free Full Text]
Dover, G. 1982. Molecular drive: A cohesive mode of species evolution. Nature 299: 111-117.[CrossRef][Medline]
Durfy, S.J. and Willard, H.F. 1990. Concerted evolution of primate satellite DNA: Evidence for an ancestral sequence shared by gorilla and human X chromosome satellite. J. Mol. Biol. 216: 555-566.[CrossRef][Medline]
DuSart, D., Cancilla, M.R., Earle, E., Mao, J.-I, Saffery, R., Tainton, K.M., Kalitsis, P., Martyn, J., Barry, A.E., and Choo, K.H.A. 1997. A functional neo-centromere formed through activation of a latent human centromere and consisting of non--satellite DNA. Nat. Genet. 16: 144-153.[CrossRef][Medline]
Elder, J.F. and Turner, B.J. 1995. Concerted evolution of repetitive DNA sequences in eukaryotes. Quart. Rev. Biol. 70: 297-320.[CrossRef][Medline]
Fanning, T.G., Seuanez, H.N., and Forman, L. 1993. Satellite DNA sequences in the New World primate Cebus apella (Platyrrhini, Primates). Chromosoma 102: 306-311.[CrossRef][Medline]
Grebenstein, B., Grebenstein, O., Sauer, W., and Hemleben, V. 1996. Distribution and complex organization of satellite DNA sequences in Aveneae species. Genome 39: 1045-1050.[Medline]
Grimes, B.R., Rhoades, A.A., and Willard, H.F. 2002. -Satellite DNA and vector composition influence rates of human artificial chromosome formation. Mol. Therapy 5: 798-805.
Haaf, T. and Willard, H.F. 1997. Chromosome-specific -satellite DNA from the centromere of chimpanzee Chromosome 4. Chromosoma 106: 226-232.[CrossRef][Medline]
___, 1998. Orangutan -satellite monomers are closely related to the human consensus sequence. Mam. Genome 9: 440-447.
Haaf, T., Warburton, P.E., and Willard, H.F. 1992. Integration of human -satellite DNA into simian chromosomes: Centromere protein binding and disruption of normal chromosome segregation. Cell 70: 681-696.[CrossRef][Medline]
Hallden, C., Bryngelsson, T., Sall, T., and Gustafsson, M. 1987. Distribution and evolution of a tandemly repeated DNA sequence in the family Brassicaceae. J. Mol. Evol. 25: 318-323.
Henikoff, S., Ahmad, K., and Malik, H.S. 2001. The centromere paradox: Stable inheritance with rapidly evolving DNA. Science 293: 1098-1102.[Abstract/Free Full Text]
Heslop-Harrison, J.S., Murata, M., Ogura, Y., Schwarzacher, T., and Motoyoshi, F. 1999. Polymorphisms and genomic organization of repetitive DNA from centromeric regions of Arabidopsis chromosomes. Plant Cell 11: 31-42.[Abstract/Free Full Text]
Hosouchi, T., Kumekawa, N., Tsuruoka, H., and Kotani, H. 2002. Physical map-based sizes of the centromeric regions of Arabidopsis thaliana Chromosomes 1, 2, and 3. DNA Res. 9: 117-121.[Abstract]
Ikeno, M., Masumoto, H., and Okazaki, T. 1994. Distribution of CENP-B boxes reflected in CREST centromere antigenic sites on long-range -satellite DNA arrays of human Chromosome 21. Hum. Mol. Genet. 3: 1245-1257.[Abstract/Free Full Text]
Koch, M., Haubold, B., and Mitchell-Olds, T. 2001. Molecular systematics of the Brassicaceae: Evidence from coding plastidic MATK and nuclear CHS sequences. Am. J. Botany 88: 534-544.[Abstract/Free Full Text]
Kumekawa, N., Hosouchi, T., Tsuruoka, H., and Kotami, H. 2000. The size and sequence organization of the centromeric region of Arabidopsis thaliana Chromosome 5. DNA Res. 7: 315-321.[Abstract]
___, 2001. The size and sequence organization of the centromeric region of Arabidopsis thaliana Chromosome 4. DNA Res. 8: 285-290.[Abstract]
Landais, I., Chavigny, P., Castagnone, C., Pizzol, J., Abad, P., and Vanlerberghe-Masutti, F. 2000. Characterization of a highly conserved satellite DNA from the parasitoid wasp Trichogramma brassicae. Gene 255: 65-73.[CrossRef][Medline]
Malik, H.S. and Henikoff, S. 2001. Adaptive evolution of Cid, a centromere-specific histone in Drosophila. Genetics 157: 1293-1298.[Abstract/Free Full Text]
Martinez-Zapater, J.M., Estelle, M.A., and Somerville, C.R. 1986. A highly repeated DNA sequence in Arabidopsis thaliana. Mol. Gen. Genet. 204: 417-423.[CrossRef]
McKinney, E.C., Ali, N., Traut, A., Feldmann, K.A., Belostotsky, D.A., McDowell, J.M., and Meagher, R.B. 1995. Sequence-based identification of T-DNA insertion mutations in Arabidopsis: Actin mutants act2-1 and act4-1. Plant J. 8: 613-622.[CrossRef][Medline]
Mestrovic, N., Plohl, M., Mravinac, B., and Ugarkovic, D. 1998. Evolution of satellite DNAs from the genus Palorus—Experimental evidence for the "library" hypot |
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Top
ABSTRACT
RESULTS
