Plant Molecular Biology 51: 523­531, 2003.
 2003 Kluwer Academic Publishers. Printed in the Netherlands.
523
Differences in the processing of DNA ends in Arabidopsis thaliana and
tobacco: possible implications for genome evolution
Nadiya Orel and Holger Puchta
Institut für Pflanzengenetik und Kulturpflanzenforschung (IPK), Corrensstrasse 3, 06466 Gatersleben, Germany
(*author for correspondence; e-mail puchta@bio.uka.de)
Received 29 November 2001; accepted in revised form 8 August 2002
Key words: C value paradox, deletion, DNA recombination, NHEJ, plant genome size
Abstract
Surprising species-specific differences in non-homologous end-joining (NHEJ) of genomic double-strand breaks
(DSBs) have been reported for the two dicotyledonous plants Arabidopsis thaliana and Nicotiana tabacum. In
Arabidopsis deletions were, on average, larger than in tobacco and not associated with insertions. To establish
the molecular basis of the phenomenon we analysed the fate of free DNA ends in both plant species by biolistic
transformation of leaf tissue with linearized plasmid molecules. Southern blotting indicated that, irrespective of
the nature of the ends (blunt, 5 or 3 overhangs), linearized full-length DNA molecules were, on average, more
stable in tobacco than in Arabidopsis. The relative expression of a -glucuronidase gene encoded by the plasmid
was similar in both plant species when the break was distant from the marker gene. However, if a DSB was
introduced between the promoter and the open reading frame of the marker, transient expression was halved in
Arabidopsis as compared to tobacco. These results indicate that free DNA ends are more stable in tobacco than
in Arabidopsis, either due to lower DNA exonuclease activity or due to a better protection of DNA break ends or
both. Exonucleolytic degradation of DNA ends might be a driving force in the evolution of genome size as the
Arabidopsis genome is more than twenty times smaller than the tobacco genome.
Introduction
The elucidation of the complete sequence of the Ara-
bidopsis genome revealed multiple sequence duplica-
tions, although the genome consists of only 125 Mb
and is thus one of the smallest plant genomes known.
This hints to the possibility of a genome duplication
in the evolution of Arabidopsis (Arabidopsis Genome
Initiative). Thus, ample amounts of DNA must have
been lost after the postulated duplication event, result-
ing in the small size of today's Arabidopsis genome.
Recently, programmed sequence elimination in the
first generations after allopolypolydization of wheat
could be demonstrated (Ozkan et al., 2001; Shakled
et al., 2001), but further DNA elimination over longer
evolutionary time periods has to be postulated to ex-
plain the amount of sequence loss (Petrov, 2001). Indi-
cations for species-specific deletions were first found
in insects. Studies on non-coding retrotransposon-like
sequences of insects indicated considerable DNA loss
from the Drosophila genome (Petrov, 1996). Indeed,
the rate of deletions differed drastically between dif-
ferent insect genomes. Cricket, which has a 11-fold
larger genome than Drosophila has a 40-fold lower
rate of DNA loss than Drosophila. This applies for the
number as well as the size of the deletions. Based on
this finding, an inverse correlation between genome
size and deletions size was postulated (Petrov et al.
2000).
But, what kind of process might be responsible
for these deletions? Deletions may occur by different
mechanisms: by replication slippage (as suggested by
Capy, 2000), by unequal crossover (as suggested by
Smith, 1976) or by double-strand break (DSB) repair.
Recently we compared DSB-induced deletions that re-
sult in a loss of function of a chromosomal marker
gene in two related dicotyledonous plant species Ara-
bidopsis and tobacco (Kirik et al., 2000), which differ
524
by more than 20 times in genome size (Bennett and
Leitch, 1997). Whereas the mechanism of junction
formation itself was not different between the two
species, the size classes of the deletions in Arabidop-
sis and tobacco differed remarkably. On average the
deletions were one third larger in Arabidopsis than in
tobacco (Kirik et al., 2000). This parallels with theo-
retical studies on insects sustaining the hypothesis of
an inverse correlation between the genome size and
the average length of deletions (Petrov et al., 2000).
Interestingly, we were not able to find insertions asso-
ciated with deletions in Arabidopsis, whereas this was
the case in almost half of the repair events analysed in
tobacco (Salomon and Puchta, 1998).
What kind of enzymatic difference between the
two plant species could be responsible for such a phe-
nomenon? A possible explanation is that the efficiency
of the degradation of DNA ends might differ between
the two species. This could be either due to a lower ex-
onuclease activity or due to a better protection of DNA
ends from degradation in tobacco. To test this hypothe-
sis we performed comparative analysis of the stability
of linearized plasmid molecules transferred by parti-
cle bombardement into Arabidopsis and tobacco cells.
Southern blot analysis in combination with transient
expression studies with -glucuronidase as marker
gene indicated that plasmid DNA is exonucleolyti-
cally degraded to a higher extent in Arabidopsis than
in tobacco. The small genome size of Arabidopsis
might therefore result from reduced stability of free
DNA ends in this plant species over evolutionary time
periods.
Material and methods
Cloning procedures and DNA preparation
The plasmid pGUS23 is a pUC7 derivative that
carries the -glucuronidase ORF under the con-
trol of the 35S promoter (Puchta and Hohn, 1991).
A specific polylinker was inserted into the Acc65I
site of pGUS23 between promoter and ORF of the
plasmid by using the annealed oligonucleotides 5 -
GTACAGCATGCGGGCCCTTAATTAACCGCGGA
GGCCTGTTAACATCGATACCGGTTCTAGA-3 and
5 -GTACTTCTAGAACCGGTATCGATGTTAACAG
GCCTCCGCGGTTAATTAAGGGCCCGCATGC-3 .
From the resulting clone pGUS231 plasmid DNA was
prepared with the Qiagen plasmid Maxi Kit (Qiagen,
Hilden, Germany) and linearized by digestion with
one of the following restriction enzymes: XbaI, StuI,
SmaI, AatII and ScaI (all NEB, Beverly, MA). Com-
pletion of digestion was tested by retransformation
of the linearized plasmid DNA into Escherichia coli.
For transformation experiments only plasmid prepa-
rations were used in which the number of colonies
was decreased by more than 99% as compared to the
unrestricted plasmid.
DNA preparation and biolistic plant transformation
Arabidopsis thaliana (ecotype Columbia-O) and Nico-
tiana tabacum (SR1) plants were grown under stan-
dard conditions on GM medium in Sigma boxes. Plant
leaves (one for tobacco and eight or nine for Arabidop-
sis) were arranged to form a 3 cm diameter circle
in the centre of a petri dish containing callus induc-
tion medium (CIM) and incubated for 24 h before
bombardment in a growth chamber under standard
conditions (Seki et al., 1999).
5 g gold particles (1 m diameter, BioRad, Her-
cules, CA) were kept overnight in 1 ml absolute
ethanol. After washing two times in sterile water, the
particles were resuspended in 100 l of 50% glycerol.
A mixture required for four bombardments was pre-
pared as follows: 25 l of gold particle suspension
was mixed with 10 l of plasmid DNA and 35 l of
1 M Ca(NO2)3. After 10 min at room temperature and
brief centrifugation, the supernatant was discarded and
the DNA-coated gold was re-suspended in 100 l of
absolute ethanol. Per macro carrier disc 25 l of the
plasmid/gold suspension were transferred. For each
transformation about 200 ng plasmid DNA was used.
After complete evaporation of the ethanol from the
disc, biolistic transformation was performed. Leaves
were bombarded by means of the Biolistic PDS1000-
He Particle Delivery System (BioRad) with a pressure
of 7.6 MPa and a gap size of 0.25 .
Plant DNA extraction and Southern analysis
Two days after transformation, total DNA was isolated
from Arabidopsis and tobacco leaves as described (Sa-
lomon and Puchta, 1998). For Southern analysis, 5 g
of genomic DNA was digested with HindIII overnight,
fractionated on a 1% agarose gel and transferred onto
a nitrocellulose hybridization membrane Hybond N
(Amersham, UK). Plasmid DNA of pGUS231 was
labelled with a random priming labelling kit (Amer-
sham, UK) and [-P32] dATP (Amersham, UK).
Hybridization was performed according to the instruc-
tions at 65 C.
525
Histological staining for GUS expression
Two days after transformation, Arabidopsis and to-
bacco leaves were infiltrated with -glucuronidase
staining solution (0.1 M sodium phosphate buffer pH
7.0 with 0.05% X-Glu (Duchefa, Haarlem, Nether-
lands) dissolved in dimethyl formamide, in the pres-
ence of 0.1% sodium azide) and incubated at 37 C
for two days. Blue sectors on the leaves were counted
using a binocular microscope. Only those experiments
were taken into account for the analysis in which a
certain minimum number of sectors could be detected
(at least 10 per cm2 of tobacco leaf).
Results
Rationale
To test the hypothesis that differences found in DSB
repair between Arabidopsis and tobacco are due to
differences in the processing of DNA ends, we de-
cided to test the stability of plasmid DNA transformed
into leaf cells of both organisms. For this purpose,
we inserted between promoter and ORF of the -
glucuronidase gene of the plasmid pGUS23 (Puchta
and Hohn, 1991) a polylinker that contained recog-
nition sites for multiple restriction enzymes including
XbaI and StuI, resulting in the plasmid pGUS231 (Fig-
ure 1). Prior to plant transformation, the plasmid was
linearized with specific restriction enzymes. Five dif-
ferent restriction enzymes were used: ScaI, resulting
in linearized plasmid DNA with blunt ends, as well as
AatII, resulting in linearized DNA with 5 overhangs,
both enzymes cut in the pUC backbone of pGUS231;
XbaI resulting in a linearized DNA molecule with
3 overhangs and StuI resulting in a linearized DNA
molecule with blunt ends, both enzymes cut between
35S promoter and -glucuronidase ORF; and SmaI re-
sulting in a linearized DNA molecule with blunt ends,
which enzyme cuts at the start of the -glucuronidase
ORF (for details see Figure 1). Particle bombardement
was used to transfer the DNA into tobacco and Ara-
bidopsis leaves (Seki et al., 1999). To avoid variation
due to differences in the DNA preparations, in all cases
the same batch of the DNA-gold preparation was used
for transformation of both species. For this purpose
leaves of 8-week old tobacco seedlings and leaves of
6-week old Arabidopsis seedlings, both in a similar
phase of their development, were cultured for one day
on CIM medium and then bombarded in parallel with
gold particles coated with DNA. Next, the leaves were
Figure 1. The plasmid pGUS231. The promoter (35S) and open
reading frame (GUS) of the -glucuronidase expression cassette are
indicated as well as the recognition sites of the restriction enzymes
applied for linearization of the plasmid in this study.
cultivated for two days on CIM medium before further
analysis on the stability of the transferred DNA either
by Southern blotting or by transient expression studies
was performed.
Southern blot analysis
In a first set of experiments plasmid stability was
analysed by Southern blotting. To elucidate how dif-
ferent types of breaks influence the stability of a
DNA molecule in plant cells, pGUS231 with 3 -
(XbaI), 5 -(AtaII) overhangs or blunt ends (SmaI, StuI)
was prepared. After the incubation period the leaves
were harvested and total DNA was extracted and di-
gested with the restriction enzyme HindIII, which
has no recognition sequence within pGUS231. After
gel electrophoresis, Southern blotting was performed
and the respective membranes were hybridized with
a plasmid-specific probe under highly stringent con-
ditions. The obtained radioactive signals on the blot
were quantified by the use of a phosphoimager. The
quotient of signal intensity of the linearized plasmid
DNA (bracket a in Figure 2) and that of the smear
produced by degraded plasmid DNA, in the same lane
(bracket b in Figure 2) was calculated. In pilot exper-
iments different incubation times of the leaves after
bombardment were tested. After 8, 24 and 48 h, be-
sides degraded DNA, the linearized plasmid band was
clearly visible for both plant species, whereas after
72 h the resulting band became too weak for reliable
526
quantification. For our further analysis we therefore
chose the longest possible incubation period for which
a reliable quantification could be performed (2 days).
In general, the intensity of the linear band was be-
tween 1% and 5% of the total intensity of the signal
in the respective lane. These values obtained for to-
bacco and Arabidopsis were then set into relation. For
example, if in a particular experiment the linearized
band represented 3% of total signal intensity in to-
bacco and 2% of total signal intensity in Arabidopsis,
a ratio of 1.5 was established. Any value over 1 thus
indicates that the full-length linear DNA was more
stable in tobacco than in Arabidopsis. The higher the
values the more linear plasmid molecules remained
undegraded in tobacco as compared to Arabidopsis. A
value of 2 would indicate that twice as many linear
plasmid molecules remained undegraded in tobacco
than in Arabidopsis. A series of experiments was per-
formed using differently restricted plasmids. For each
restriction enzyme nine independent experiments are
depicted in Table 1. In spite of quite large varia-
tion between single experiments, which are typical
of transient transformation assays (e.g. Puchta and
Hohn, 1991), our results clearly demonstrate that, in-
dependent of the respective DNA end, in most cases
(29 out of 36) the linearized DNA was less degraded
in tobacco than in Arabidopsis (Table 1). Linearized
plasmid DNA was, on average, 1.4­1.9 times more
stable in tobacco than in Arabidopsis. No major differ-
ences between blunt and overhanging DNA ends could
be detected (Table 1). However, due the fact that the
subtraction of the standard deviations from the means,
given in Table 1, result in a value above 1 in only
two of four cases, these results can only be taken as
a first indication of differences in processing of DNA
molecules in the two plant species and needed to be
further corroborated by transient expression analysis.
Transient expression
To sustain the results described above and to further
characterize the mode of nuclear degradation transient
expression experiments were performed. A difference
in DNA degradation could be due either to differences
of exonucleolytic or endonucleolytic processing of the
transfected DNA. To clarify the nature of nucleolytic
activity we performed transient expression assays after
biolistic bombardement of Arabidopsis and tobacco
with linearized pGUS231. The rationale was that in
case of endonucleolytic degradation, no difference
in the relation of the expression rates of a marker
Figure 2. Representative Southern analysis of DNA extracted from
leaves two days after biolistic transformation with linearized DNA
of the plasmid pGUS231. After isolation the DNA was restricted
with HindIII, blotted and hybridized with labelled pGUS231. The
intensity of the signals was determined by the use of a phosophoim-
ager. The linear band (a) was put into relation with the smear
resulting from degraded DNA (b). In lanes 1­3 Arabidopsis DNA
and in lanes 4­6 tobacco DNA was loaded. Lanes 1, 4: DNA of
plant leaves bombarded with pGUS231 restricted with XbaI; lanes
2, 5: DNA of plant leaves bombarded with pGUS231 restricted with
SmaI; lanes 3, 6: DNA of plant leaves bombarded with pGUS231
restricted with AtaII.
Table 1. The relative stability of linearized DNA molecules of the
plasmid pGUS231 in leaf cells of tobacco in relation to Arabidopsis.
In the bottom line the means and standard deviations are given.
XbaI 3 StuI blunt SmaI blunt AtaII 5
overhang overhang
2.48 2.25 0.96 2.64
1.88 0.97 1.32 0.63
1.70 2.80 2.17 0.97
1.28 0.94 1.00 1.95
1.52 1.17 1.64 1.74
1.92 2.63 2.15 1.48
0.63 3.15 1.54 1.86
1.46 1.55 1.18 1.23
1.54 1.34 1.75 0.46
1.60  0.50 (9) 1.91  0.92 (9) 1.52  0.45 (9) 1.42  0.66 (9)
527
gene should be detectable between Arabidopsis and
tobacco, irrespective where the break was induced
in the plasmid molecule. However in case of exonu-
cleolytic degradation, different positions of the break
resulting in a linear plasmid molecule are expected
to reveal differences between the two species. The
closer the break site is to a DNA sequence required
for expression the higher the efficiency of marker de-
struction by exonucleolitic degradation resulting in a
reduced expression activity is expected to be. There-
fore we compared the XbaI and StuI restriction sites
of pGUS231 lying between promoter and ORF of the
-glucuronidase gene with the ScaI site located within
the pUC backbone (Figure 1). Whereas in case of XbaI
and StuI sites exonucleolytic degradation of less than
100 bp at either end would result in a loss of func-
tion of the -glucuronidase gene, in case of the ScaI
site only degradation of more than 1 kb at one end
or more than 2 kb at the other would be required to
result in a loss of function of the enzyme. Since StuI
as well as ScaI digestion results in blunt ends, not the
nature of the break but its position differs between the
respective DNA molecules. Conversely, XbaI-digested
(3 overhanging ends) and StuI-digested (blunt ends)
DNA molecules differ in respect to the DNA ends
but hardly in the position of the break site. How-
ever, for the latter DNAs a religation step (Gorbunova
and Levy, 1997) to fuse the promoter to the ORF
is required for transcription of a functional mRNA.
Bombardement with circular plasmid DNA was in-
cluded as transcription control. After histochemical
staining and detection of blue sectors the average num-
ber of spots per cm2 leaf was determined. The data
are presented in Table 2. In general independent of
plant species the highest numbers of spots were de-
tected after shooting with circular pGUS231, followed
by the plasmid cut outside of the marker gene. Re-
ligation of pGUS231 harbouring sticky ends occurred
quite efficiently, depending on experiment in the range
of 10­30% of the molecules the ORF was restored,
whereas religation of the blunt-ended plasmid gave in
most experiments lower rates. All values were above
1%, the putative maximal `background' value, that
could arise due to traces of undigested plasmid DNA
in the preparation, as determined by retransformation
of the respective linearized plasmid preparations into
E. coli. To directly compare the results obtained from
tobacco and Arabidopsis the values obtained with the
linearized plasmid molecules were set into relation to
the circular ones (Table 2, second last row). By this
procedure variations between tobacco and Arabidopsis
due to transcriptional differences were excluded from
the calculation. These values were then set into rela-
tion to one another (Table 2, last row). For example,
a value of 2 indicates the presence of twice as many
functional -glucuronidase genes in tobacco than in
Arabidopsis after bombardement with the respective
DNA. In repeated experiments (for statistical evalua-
tion, see Table 3) this relation was found to be around 1
in case of the ScaI-digested pGUS231 and around 2 for
both XbaI- and StuI-digested pGUS231. This indicates
that in case of a break between promoter and ORF,
on average, twice more functional -glucuronidase
expression cassettes were restored in tobacco than in
Arabidopsis irrespective of the nature of the DNA
ends. The calculated standard deviations indicate that
the detected differences are highly significant.
Discussion
Double-strand breaks (DSBs) are critical lesions in
genomes. Efficient repair of DSBs is necessary for the
survival of all organisms. In principle, DSBs can be re-
paired via non-homologous end joining (NHEJ) or via
homologous recombination. Although homologousre-
combination is used for DSB repair in plants (Chi-
urazzi et al., 1996; Shalev and Levy, 1997; Puchta,
1999; Xiao and Peterson, 2000; Siebert and Puchta,
2002) NHEJ seems to be the main mode of repair
(for reviews, see Puchta and Hohn, 1996; Vergunst
and Hooykaas, 1999; Gorbunova and Levy, 1999;
Mengiste and Paszkowski, 1999). Error-prone DSB
repair may result either in deletions, insertions or
various other kinds of genomic rearrangements (Pipi-
ras et al., 1998; Salomon and Puchta, 1998; Jasin,
2000). In plants genomic alterations in somatic cells
of meristems can be transferred to the offspring (Wal-
bot, 1996) and therefore relevant for evolution. Thus,
somatic DSB repair might have in evolutionary terms
an impact on genome size and genome organization.
The reason for the large differences in the nu-
clear DNA content of eukaryotes, particularly plants,
known as `C value paradox', has been a matter of
debate since long (Cavalier-Smith, 1985; Dove and
Flavell, 1988). Even closely related species with sim-
ilar phenotype may significantly differ as to their
diploid genome size. One mechanism responsible for
these differences could be due to a species-specific
increase/reduction of repetitive sequences. In princi-
ple, genomes may become larger via duplications and
insertions or smaller via deletions. Species-specific
528
Table 2. Expression of -glucuronidase after bombardment of the plasmid pGUS231 in leaf cells of tobacco and Arabidopsis.
529
spread of retrotransposons was postulated as a main
route enlarging plant genomes (SanMiguel et al.,
1996; SanMiguel et al., 1998; Bennetzen and Kel-
log, 1997). Alternatively, deletions might reduce
genome size and counterbalanceenlargements (Petrov,
2001). Recently, an elegant theoretical study (Petrov
et al., 2000) has demonstrated that deletions of signifi-
cantly different extension within retroelements yielded
species-specific genome size alterations in related in-
sect species over evolutionary time periods. We were
able to demonstrate experimentally that the mean size
of deletions obtained from DSB repair is on average a
third larger in Arabidopsis than in tobacco (Kirik et al.,
2000). Thus, for plants and insects an inverse correla-
tion between genome size and deletion size was found.
However, one has to be careful to draw final conclu-
sion from the comparison of only two species each,
but it is tempting to speculate that the phenomenon
might be a general principle of genome evolution.
During DSB repair the size of a deletion depends
on the processing of DNA ends. If broken ends are not
religated directly the processing of such ends might
result in the loss of DNA at the break site (Salomon
at al., 1998; Kirik et al., 2000). Depending on the
efficiency of DNA degradation more or less informa-
tion will be lost. We were able to detect differences as
to the stability of linearized plasmid DNA in tobacco
and Arabidopsis cells. Although the results obtained
by Southern blotting indicate that on average plas-
mid molecules transformed into Arabidopsis cells are
less stable than in tobacco cells, we of course cannot
exclude that this difference might not be caused by
degradation in the nucleus, as plasmid molecules are
also transferred into other compartments of the cell.
However, as two other assays, that are clearly nuclear-
based (chromosomal DSB repair [Kirik et al., 2000]
and the transient expression assay described in this re-
port) demonstrate a similar difference using unrelated
methodologies, we are tempted to speculate that this is
caused by the same kind of phenomenon.
The results presented here indicate that the differ-
ence in the processing of DNA ends between the two
plant species is at least mainly due to an enhanced
exonucleolytic degradation of DNA in Arabidopsis.
We were able to show that if a break was induced
between the promoter and the open reading frame of
a marker gene the expression level was reduced by a
factor of two in Arabidopsis relative to tobacco. This is
in contrast to the similar expression levels obtained for
both species when the break was introduced distantly
from the marker gene. Whereas in the first case ex-
Table 3. Statistical evaluation of the data presented in Table 2.
In the bottom line the means and standard deviations are given.
StuI ScaI XbaI
blunt in ORF blunt out of ORF 3 overhang in ORF
2.43 1.03 1.40
2.40 1.14 2.00
2.35 0.50 1.60
2.15 0.71 1.38
2.18 1.23 2.10
1.97 0.50 1.50
2.250.18 (6) 0.850.32 (6) 1.660.31 (6)
onucleolytic digest of few nucleotides would result in
destruction of the expression cassette, in the latter case
the removal of more than thousand base pairs would be
a prerequisite for the destruction of the ORF. Because
endonucleolytic degradation would destroy all infor-
mation on a plasmid at the same rate independent of a
pre-existing break site, no differences should be found
in presence of such a nuclease activity. However, since
the restoration of the -glucuronidase ORF in case of
the StuI- and XbaI-restricted plasmid requires a liga-
tion step, the data of the expression assay alone do
not suffice to discriminate between a more efficient
exonucleolytic destruction of the marker gene or a less
efficient ligation reaction in Arabidopsis cells. As we
were able to detect by Southern analysis differences
in the degradation of plasmid DNA linearized with the
exactly the same enzymes between Arabidopsis and
tobacco, we think that we are allowed to speculate
that at least most of the effect is due a more efficient
exonucleolytic degradation of DNA in Arabidopsis.
However, one question remains: what is the cause of
the enhanced DNA degradation in Arabidopsis? This
could be either due to less active DNA exonucleases
or due to the better protection of DNA ends from
degradation, for example by DNA end binding factors
(Liang et al., 1996) in tobacco.
Purposely, we did not test the stability of circular
plasmid DNA in both plant species, as this approach
would not have allowed us to discriminate between
exo- and endonucleolytic degradation. Linearization
of the plasmid DNA by an endogenous endonuclease
would be a prerequisite of exonucleolitic degrada-
tion, however if the plasmid is linearized, exonuclease
could immediately work on it. Thus, it would be
hard to discriminate to which extent what kind of en-
zyme activity would contribute to DNA degradation
530
in Arabidopsis and tobacco. Using linearized mole-
cules we were working with a more defined system
as endonucleases could work on all DNA molecules
immediately after transformation and independent of
a second reaction.
Theoretical calculations on the evolution of
genome sizes have to take into account the rate of
duplication and the rate of loss of genomic sequences
(Petrov, 2001). DSB repair is one mechanism that may
result in sequence deletions. To evaluate the possible
role of DSB repair in this process one has to take into
account two factors: the number of breaks induced
and the loss of sequences information per repair event.
The loss of sequence information is strongly influ-
enced by the stability of the broken DNA ends. The
more efficient the ends are recessed by exonucleolitic
degradation the more the genome size is supposed to
shrink over an evolutionary time period. In line with
this argumentation we found indications that in a plant
species with a small genome free DNA ends are less
stable than in a plant species with a more than twenty
times larger genome.
Acknowledgements
We would like to thank Ingo Schubert and two anony-
mous referees for useful criticism on the manuscript
and Angela Kirik and Frank Hartung for help and dis-
cussions. The study was partly funded by a grant of
the Deutsche Forschungsgemeinschaft to H.P.
References
Arabidopsis Genome Initiative. 2000. Analysis of the genome se-
quence of the flowering plant Arabidopsis thaliana. Nature 408:
796­815.
Bennett, M.D. and Leitch, I.J. 1997. Nuclear DNA amounts in
angiosperms: 583 new estimates. Annal. Bot. 80: 169­196.
Bennetzen, J.L. and Kellog, E.A. 1997. Do plants have a one-way
ticket to the genomic obesity? Plant Cell 9: 1509­1514.
Capy, P. 2000. Perspectives: evolution. Is bigger better in cricket?
Science 287: 985­986.
Cavalier-Smith, T. 1985. The Evolution of Genome Size. John
Wiley, Chichester, UK.
Chiurazzi, M., Ray, A., Viret., J.-F., Perera, R., Wang, X.-H.,
Lloyd, A. and Signer, E.R.. 1996. Enhancement of somatic in-
trachromosomal homologous recombination in Arabidopsis by
HO-endonuclease. Plant Cell 8: 2057­2066.
Dove, G.A. and Flavell, R.B. 1988. Genome Evolution. Academic
Press, London.
Gorbunova, V. and Levy, A.A. 1999. How plants make ends meet:
DNA double-strand break repair. Trends Plant Sci. 4: 263­269.
Gorbunova, V. and Levy, A.A. 1997. Non-homologous DNA end
joining in plant cells is associated with deletions and filler DNA
insertions. Nucl. Acids Res. 25: 4650­4657.
Jasin, M. 2000. Chromosome breaks and genomic instability. Can-
cer Invest. 18: 78­86.
Kirik, A., Salomon, S. and Puchta, H. 2000. Species-specific
double-strand break repair and genome evolution in plants.
EMBO J. 19: 5562­5566.
Liang, F., Romanienko, P.J., Weaver, D.T., Jeggo, P.A. and
Jasin, M. 1996. Chromosomal double-strand break repair in
Ku80-deficient cells. Proc. Natl. Acad. Sci. USA 93: 8929­8933.
Mengiste, T. and Paszkowski, J. 1999. Prospects for the precise
engineering of plant genomes by homologous recombination. J.
Biol. Chem. 380: 749­758.
Ozkan, H., Levy, A.A. and Feldman, M. 2001. Allopolyploidy-
induced rapid genome evolution in the wheat (Aegilops-Triticum)
group. Plant Cell 13: 1735­1747.
Petrov, D.A. 2001. Evolution of genome size: new approaches to an
old problem. Trends Genet. 17: 23­28.
Petrov, D.A. 1997. Slow but steady: Reduction of genome size
through biased mutation. Plant Cell 9: 1900­1901.
Petrov, D.A., Lozovskaya, E.R and Hartl, D.L. 1996. High intrinsic
rate of DNA loss in Drosophila. Nature 384: 346­349.
Petrov, D.A., Sangster, T.A., Johnston, J.S., Hartl, D.L. and Shaw,
K.L. 2000. Evidence for DNA loss as a determinant of genome
size. Science 287: 1060­1062.
Pipiras, E., Coquelle, A., Bieth, A. and Debatisse, M. 1998. Intersti-
tial deletions and intrachromosomal amplification initiated from
a double-strand break targeted to a mammalian chromosome.
EMBO J. 17: 325­333.
Puchta, H. 1999. DSB-induced recombination between ectopic
homologous sequences in somatic plant cells. Genetics 152:
1173­1181.
Puchta, H. and Hohn, B. 1991. A transient assay in plant cells
reveals a positive correlation between extrachromosomal recom-
bination rates and length of homologous overlap. Nucl. Acids
Res. 19: 2693­2700.
Puchta, H. and Hohn, B. 1996. From centiMorgans to basepairs:
homologous recombination in plants. Trends Plant Sci. 1: 340­
348.
Salomon, S. and Puchta, H. 1998. Capture of genomic and T-DNA
sequences during double-strand break repair in somatic plant
cells. EMBO J. 17: 6086­6095.
SanMiguel, P., Tikhonov, A., Jin, Y.K., Motchoulskaia, N., Za-
kharov, D., Melake-Berhan, A., Springer, P.S., Edwards, K.J.,
Lee, M., Avramova, Z. and Bennetzen, J.L. 1996. Nested retro-
transposons in the intergenic regions of the maize genome.
Science 274: 765­768.
SanMiguel, P., Gaut, B.S., Tikhonov, A., Nakajima, Y. and Bennet-
zen, J.L. 1998. The paleontology of intergene retrotransposons
of maize. Nature Genet. 20: 43­45.
Seki, M., Iida, A. and Morikawa, H. 1999. Transient expression
of the -glucuronidase gene in tissues of Arabidopsis thaliana
by bombardment-mediated transformation. Mol. Biotechnol. 11:
251­255.
Shaked, H., Kashkush, K., Ozkan, H., Feldman, M. and Levy, A.A.
2001. Sequence elimination and cytosine methylation are rapid
and reproducible responses of the genome to wide hybridization
and allopolyploidy in wheat. Plant Cell 13: 1749­1759.
Shalev, G. and Levy, A.A. 1997. The maize transposable ele-
ment Ac induces recombination between the donor site and an
homologous ectopic sequence. Genetics 146: 1143­1151.
Siebert, R. and Puchta, H. 2002. Efficient repair of genomic double-
strand breaks via homologous recombination between directly
531
repeated sequences in the plant genome. Plant Cell 14: 1121­
1131.
Smith, GP. 1976. Evolution of repeated DNA sequences by unequal
crossover. Science 191: 528­535.
Vergunst, A.C. and Hooykaas, P.J.J. 1999. Recombination in the
plant genome and its application in biotechnology. Crit. Rev.
Plant. Sci. 18: 1­31.
Walbot, V. 1996. Sources and consequences of phenotypic and
genotypic plasticity in flowering plants. Trends Plant Sci. 1:
27­33.
Xiao, Y.L, Peterson, T. 2000. Intrachromosomal homologous re-
combination in Arabidopsis induced by a maize transposon. Mol.
Gen. Genet. 263: 22­29.