Double strand break-induced recombination in
Chlamydomonas reinhardtii
chloroplasts
Double strand break-induced recombination in Chlamydomonas reinhardtii chloroplasts
F.
Dürrenberger
+
,
A. J.
Thompson
1,[sect]
,
D. L.
Herrin
1
and
J.-D.
Rochaix*
Departments of Molecular Biology and Plant Biology, University of Geneva, 30
quai Ernest-Ansermet, 1211
Geneva
4,
Switzerland
and
1
Department of Botany, University of Texas at Austin,
Austin
, TX 78713,
USA
Received June 3, 1996;
Revised and Accepted July 22, 1996
ABSTRACT
The mechanisms of chloroplast recombination are largely unknown. Using the
chloroplast-encoded homing endonuclease I-
Cre
I from
Chlamydomonas reinhardtii
, an experimental system is described that allows the study of double strand
break (DSB)-induced recombination in chloroplasts. The I-
Cre
I endonuclease is encoded by the chloroplast ribosomal group I intron of
C.reinhardtii
and cleaves specifically intronless copies of the large ribosomal RNA (23S)
gene. To study DSB-induced recombination in chloroplast DNA, the genes encoding the I-
Cre
I endonuclease were deleted and a target site for I-
Cre
I, embedded in a cDNA of the 23S gene, was integrated at an ectopic location.
Endonuclease function was transiently provided by mating the strains containing
the recombination substrate to a wild-type strain. The outcome of DSB repair was analyzed in haploid progeny of
these crosses. Interestingly, resolution of DSB repair strictly depended upon
the relative orientation of the ectopic ribosomal cDNA and the adjacent copy of
the 23S gene. Gene conversion was observed when the 23S cDNA and the
neighbouring copy of the 23S gene were in opposite orientation, leading to
mobilization of the intron to the 23S cDNA. In contrast, arrangement of the 23S
cDNA in direct repeat orientation relative to the proximal 23S gene resulted in
a deletion between the 23S cDNA and the 23S gene. These results demonstrate
that
C.reinhardtii
chloroplasts have an efficient system for DSB repair and that homologous
recombination is strongly stimulated by DSBs in chloroplast DNA.
INTRODUCTION
The importance of DNA repair and recombination for the maintenance of genome
integrity is well documented (
1
-
3
). Repair of DNA double strand breaks (DSBs) has been studied in detail in the
yeast
Saccharomyces cerevisiae
. In
S.cerevisiae
DSBs are predominantly repaired through recombination between the broken DNA and
an intact homologous sequence (reviewed in
4
). Two distinct mechanisms, DSB repair (
5
) and single strand annealing (
6
-
9
), have been proposed for the repair of DSBs. In addition, it has been
demonstrated that certain hot spots for meiotic recombination are associated
with DSBs, suggesting that meiotic recombination is initiated by DSBs (
10
,
11
).
The chloroplast genome of the green unicellular alga
Chlamydomonas reinhardtii
is composed of ~80 copies of a 195 kb circular DNA molecule (
12
,
13
). An inverted repeat of 19 kb which includes the ribosomal RNA genes divides
the chloroplast genome of
C.reinhardtii
into two large single copy regions (Fig.
1
). A characteristic feature of mutant strains with deletions or point mutations
within the inverted repeat is that they are transmitted to both repeated
segments, presumably through an efficient gene conversion mechanism (
14
). Each 23S ribosomal RNA gene of
C.reinhardtii
contains an intron with an internal open reading frame encoding I-
Cre
I, a homing endonuclease (
15
,
16
). This enzyme has previously been shown to recognize a sequence of at least 19
bp and to cleave at the intron insertion site in 23S DNA lacking the ribosomal
intron (
17
,
26
). In addition, integration of a ribosomal cDNA containing the I-
Cre
I target site into the
C.reinhardtii
chloroplast genome results in efficient transfer of the ribosomal intron to
that ectopic homing site (
15
).
Although chloroplasts fuse in zygotes of
C.reinhardtii
, one of the parental chloroplast genomes is selectively degraded. This process
is genetically controlled by the nuclear mating type loci and results in mostly
uniparental inheritance of chloroplast genes. However, the fact that 1-10% of meiotic zygotes transmit chloroplast markers from both parents
permits the genetic analysis of chloroplast DNA recombination (
13
). Due to the low frequency of biparental transmission of chloroplast genes it
seems unlikely that the main function of the chloroplast recombination system
is promoting genetic diversity. Furthermore, studies of
C.reinhardtii
chloroplast transformants expressing dominant negative versions of the
Escherichia coli
RecA protein indicate that a RecA-mediated recombination system exists in chloroplasts. These experiments
lead to the suggestion that in chloroplasts, which have to cope with
significant photooxidative stress, the primary role of DNA recombination is in
DNA repair (
18
).
The occurrence of a rare-cutting homing endonuclease in the chloroplast of
C.reinhardtii
presents an excellent opportunity to study DSB-induced recombination in chloroplasts (
15
). Several studies, mainly in yeast and bacteriophages, have revealed that
homing endonucleases are paramount for the genetic mobility of introns which
they inhabit. These intron-encoded endonucleases cleave both strands of the recipient (intronless)
DNA duplex at a specific site close to the intron insertion site. Subsequent
repair of the DSB in the recipient allele using the intron-containing allele as a template is thought to proceed by a mechanism
similar to the DSB repair mechanism originally proposed for recombination of
nuclear DNA in yeast (
5
). In genetic terms, the net result of intron-homing is the unidirectional gene conversion of an allele lacking a given
intron to the corresponding intron-containing form (reviewed in
19
,
20
).
To gain more insight into the molecular mechanisms underlying chloroplast recombination, we have established a system in
C.reinhardtii
which allows us to induce transient DSBs with the I-
Cre
I endonuclease at a specific target site within the chloroplast genome and to
assess the consequences of these breaks on homologous recombination between
partially repeated ribosomal sequences. We discuss our findings with respect to
current models for DSB-induced recombination.
MATERIALS AND METHODS
Construction of the
[Delta]
I-
Cre
I c23S recombination tester strains
In a first step, FuD50, a
C.reinhardtii
strain carrying a deletion of the 3' end of the chloroplast
atp
B gene and some downstream sequences (
21
), was transformed with pES7.2 with a microprojectile gun as described previously (
16
). pES7.2 contains a deletion in the 23S intron that removes 360 bp of the 489
bp I-
Cre
I coding sequence. In addition, the flanking 16S and 23S ribosomal sequences
present in pES7.2 contain point mutations that confer resistance to
spectinomycin and erythromycin, respectively (
16
). Thus, chloroplast transformants were selected on TAP plates (
22
) containing spectinomycin and erythromycin. Homoplasmicity for the I-
Cre
I deletion was tested by Southern blotting. The deletion was present in both
segments of the inverted repeat, most likely as the result of an efficient copy
correction mechanism (
14
). One of the resulting [Delta]
atp
B [Delta]I-
Cre
I transformants was subjected to a second round of chloroplast transformation
with plasmids pINTc23S.2 and pINTc23S.3 in order to provide the I-
Cre
I target site contained within the 600 bp 23S cDNA, as described previously (
15
). Site-specific integration of the 23S cDNA at the
atp
B locus is accompanied by replacement of the mutant
atp
B gene with the functional
atp
B gene delivered by the transforming pINT:c23S plasmids, thereby restoring
photoautotrophic growth. Photosynthetic proficient [Delta]I-
Cre
I
c23S
transformants were recovered on HSM plates (
22
). Precise insertion of the 23S cDNA either in the same or in the opposite
orientation relative to the proximal 23S gene was verified by Southern
blotting.
Crosses
Crosses were performed as described (
13
). The wild-type
C.reinhardtii
strain was 137c (
13
). Strains 2A1 and 4C2, harbouring the 23S cDNA in inverted and direct
orientation, respectively, were used as the recombination tester strains. The
structure of the 23S cDNA was analyzed only in tetrads showing 2:2 segregation
for the nuclear mating type loci. Segregant B9A was employed as the
mt
-
parent for the control crosses in which both mating partners carried the
deletion in the I-
Cre
I gene. B9A (
mt
-
[Delta]I-
Cre
I) was isolated from the cross between strains 2A1 and wild-type, and had the 23S cDNA with flanking sequences precisely replaced with
the corresponding allelic wild-type
atp
B locus, whereas the deletion in the endonuclease gene was maintained (based on
Southern analysis and resistance to spectinomycin and erythromycin).
Southern analysis
Total DNA was prepared from 10 ml cultures of light grown
C.reinhardtii
cells using a previously published protocol (
22
) with modifications: cell pellets were resuspended in 350 [mu]l TEN (50 mM EDTA, 20 mM Tris-HCl, pH 8.0, 100 mM NaCl). After addition of 25 [mu]l proteinase K (5 mg/[mu]l) and 50 [mu]l 10% SDS, the lysates were incubated for 2 h at 55oC. After two phenol-chloroform extractions, 70 [mu]l 5 M NaCl were added to the aequeous
phase and gently mixed. Sixty-five microlitres of 10% CTAB (Hexadecyltrimethylammoniumbromide, Sigma)
solution prepared in 0.7 M NaCl was added and the mixtures were incubated for
10 min at 65oC. After a final chloroform extraction, the nucleic acids were precipitated
by addition of 1 ml ethanol, recovered by a 1 min centrifugation in a
microcentrifuge, washed with 70% ethanol, dried, and resuspended in 50 [mu]l TE (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.2 mg/[mu]l RNase A). Total DNA (1-2 [mu]l) was digested with restriction enzymes, fractionated
on agarose gels and transferred to Hybond N
+
(Amersham) membrane. Double-stranded DNA probes for Southern hybridizations were labeled by random
priming with [[alpha]-
32
P]dATP (
23
). The c23S probe was the 0.6 kb
Bam
HI fragment of pc23S (
15
). The ribosomal intron probe was the 0.2 kb
Hin
fI fragment from pKS:HB100 (
15
) which maps outside of the deletion in the I-
Cre
I gene. The IR probe was a 1.2 kb
Bam
HI-
Sph
I fragment isolated from the p
atp
B-INT chloroplast transformation vector and corresponds to a
Bam
HI-
Kpn
I fragment of the same size in wild-type chloroplast DNA (
25
). The
atp
B probe was a 0.5 kb
Pst
I-
Kpn
I fragment isolated from p
atp
B3'NT (J.-D. Rochaix, unpublished results) corresponding to the coding region
of the
atp
B gene deleted in FuD50. The HB2.5 probe was isolated from pCWHB2.5, a plasmid
carrying the HB150 subclone of the R24 chloroplast restriction fragment (
25
).
PCR analysis
Selective amplification of the 23S cDNA was carried out as described previously
(
15
). One primer hybridized to the
psb
D sequence present at the 3' end of the
atp
B gene (primer 3'psbD). The second primer was specific either for the 5' end of the 23S cDNA (primer 5'23S.1) in amplifications involving transformants carrying the 23S cDNA in inverted orientation, or for the
3' end (primer 3'23S.2) when transformants were analyzed with the 23S cDNA in
direct repeat orientation.
RESULTS
Experimental system for the study of DSB repair and recombination in chloroplast
DNA
To study DSB-induced homologous recombination of chloroplast DNA, we decided to
establish an experimental system that permits the induction of a site-specific DSB within one repeat of a pair of homologous sequences in
C.reinhardtii
chloroplast DNA. The ~20 bp recognition sequence of the
C.reinhardtii
chloroplast homing endonuclease I-
Cre
I spans the intron insertion site of the ribosomal intron (
17
,
26
) and accordingly is disrupted in wild-type chloroplast DNA of
C.reinhardtii
in which both 23S genes carry the ribosomal intron. Hence, we expected the 195
kb chloroplast DNA circles of
C.reinhardtii
to be cut by this rare cutting endonuclease only if a I-
Cre
I target site is introduced via chloroplast transformation. Therefore, strains
were constructed carrying a 0.6 kb cDNA of the 23S gene either as an inverted
or a direct repeat relative to the proximal 23S gene located 8 kb away from the
23S cDNA (Figs
3
and
7
, respectively). The 23S cDNA (
c23S
) contains the target site for I-
Cre
I and has 0.4 and 0.2 kb of perfect homology to upstream and downstream flanking
sequences, respectively, of the ribosomal intron in the 23S genes located in
both large inverted repeats of the chloroplast DNA. However, previous
experiments disclosed that targeted integration of the 23S cDNA into wild-type chloroplast DNA of
C.reinhardtii
led in all cases to mobilization of the ribosomal intron to the ectopic cDNA and
therefore to disruption of the I-
Cre
I recognition sequence (
15
). Thus, a strain was constructed with a deletion in the intron open reading
frame encoding I-
Cre
I ([Delta]I-
Cre
I). Subsequently the 23S cDNA was introduced into the [Delta]I-
Cre
I strain (Materials and Methods).
I-
Cre
I-mediated intron-homing to the 23S cDNA in inverted repeat orientation
In
C.reinhardtii
the chloroplast genome is inherited in a uniparental fashion from the mating-type plus (
mt
+
) parent (
13
). The [Delta]I-
Cre
I strains containing the 23S cDNA as a recombination substrate were therefore
constructed in a
mt
+
nuclear background to transmit the modified chloroplast DNA to most of the
progeny. The outcome of DSB repair was analyzed in the haploid progeny from the
cross between the [Delta]I-
Cre
I strain carrying the 23S cDNA in inverted orientation (2A1) and a wild-type strain (
mt
-
). Eight tetrads were analyzed by Southern blotting and PCR (data not shown) for
the fate of the 23S cDNA. Figure
3
shows restriction maps of part of the large inverted repeat proximal to the 23S
cDNA for 2A1, wild-type and their progeny.
I-
Cre
I-mediated formation of a deletion between the 23S gene and the 23S cDNA in
direct repeat orientation
DISCUSSION
Group I introns are a prominent feature of
Chlamydomonas
chloroplast genomes where they are frequently encountered in LSU rRNA genes (
27
). Besides I-
Cre
I, two additional homing endonucleases have been identified in chloroplasts of
the genus
Chlamydomonas
. Both, I-
Ceu
I and I-
Chu
I are encoded by ribosomal introns of
C.eugametos
and
C.humicola
, respectively (
28
,
29
). Unidirectional gene conversion in interspecific crosses has been reported for
the
C.eugametos
intron encoding I-
Ceu
I (
30
). Dependence of intron-homing on endonuclease function has been demonstrated for mobile group I
introns from T4 phage and yeast mitochondria (
31
-
34
). However, the role of intron-encoded endonucleases in mobility of chloroplast introns has not been
directly assessed so far. Here we present two lines of experimental evidence
that intron mobility in chloroplasts requires homing endonuclease function.
First, mobility of the
C.reinhardtii
ribosomal intron to an ectopic homing site upon transformation of chloroplasts
with the 23S cDNA is abolished in strains carrying deletions in their I-
Cre
I genes. Secondly, the ribosomal intron can be mobilized during crosses to its
artificial homing site only if endonuclease function is provided by the mating partner. Considering only progeny that had maintained the 23S cDNA, the efficiency of intron homing
was 80% (Table
1
). Surprisingly, intron mobility did not occur if the 23S cDNA had the same
orientation as the proximal ribosomal gene, suggesting that recombination
mechanisms for the repair of DSBs at the homing site differ for the two
orientations of the 23S cDNA (see below).
Recombination functions required for mobility of the phage
td
intron have been described in some detail. The recombinases and nucleases
required for efficient mobility of this prokaryotic group I intron are
consistent with the functions predicted by the DSB repair (DSBR) pathway for
intron inheritance (
35
). Assuming that mobility of eukaryotic group I introns proceeds by a similar
DSBR mechanism as described for their prokaryotic counterparts, our results for
intron mobility to the 23S cDNA in inverted repeat orientation are most readily
explained by the DSBR model for intron-homing. The single-stranded 3' ends of the 23S cDNA generated by cleavage of I-
Cre
I (
17
,
26
), and presumably further exposed by subsequent recession of the 5' termini, invade homologues stretches of the 23S gene flanking the
ribosomal intron. DNA repair synthesis using the 23S gene as a template and
subsequent resolution leads to copying of the ribosomal intron to the cDNA.
Since in all examined cases intron mobility was not associated with inversions
between the 23S cDNA and the 23S gene, DSBR was not accompanied by crossing
over of flanking DNA (data not shown). The reason could be that such an
inversion might have deleterious effects. First, the segment of chloroplast DNA
bordered by the 23S cDNA and the proximal 23S gene contains the strong promoter
of the ribosomal RNA unit which would lead to increased transcription outside
of the large inverted repeat and might thereby interfere with expression of
nearby essential genes. Second, the inversion would create a chloroplast genome
with two ribosomal direct repeats which may undergo recombination with each
other and compromise the stability of the chloroplast genome.
Interestingly, only homing of the intron carrying the deleted I-
Cre
I gene was observed among the analyzed progeny. This is in contrast with typical
(allelic) intron-homing, where both endonuclease and intron are contributed
in
trans
. The difference might be due to the fact that in
C.reinhardtii
the chloroplast DNA from the
mt
-
mating partner is degraded in the zygote (
36
), thereby increasing the number of [Delta]I-
Cre
I introns which can be used as donors for conversion of the 23S cDNA to the
intron-containing form. Alternatively, this result might indicate a preference of
the system for intramolecular gene conversion, if the choice between inter- and intramolecular reaction is offered.
Chloroplast genes are mainly uniparentally transmitted to meiotic progeny of
C.reinhardtii
such that most of the progeny has the same chloroplast genotype as the
mt
+
parent. However, between 1 and 10% of meiotic zygotes transmit chloroplast
markers from the
mt
-
parent (
13
). Surprisingly, up to 25% of the meiotic progeny derived from crosses involving
the
mt
+
chloroplast recombination tester strain carrying the 23S cDNA in inverted
orientation (strain 2A1) had lost the cDNA and acquired the
mt
-
(i.e. wild-type) configuration at the
atp
B locus (Table
1
). These recombination events were mostly independent of I-
Cre
I endonuclease function, since similar frequencies were obtained whether or not
the
mt
-
parent provided endonuclease function (25 and 21%, respectively, Table
1
). They appear to be induced during the meiotic cycle because they were never
observed in cells grown vegetatively. However, endonuclease-independent loss of the 23S cDNA seems to be orientation-dependent, since this recombination event was detected in only 14%
of the progeny derived from the cross between the recombination tester strain
carrying the 23S cDNA in direct repeat orientation (4C2) and an
mt-
strain deficient in I-
Cre
I . In addition, several
aad
A reporter gene fusion constructs integrated at the same site of the
C.reinhardtii
chloroplast genome (
atp
B locus) were transmitted with >95% frequencies to meiotic progeny, and showed
therefore little recombination with
mt
-
chloroplast DNA (
37
,
38
). Thus, the relatively high frequency of I-
Cre
I-independent loss of the 23S cDNA appears to be caused by the nature and
orientation of the inserted ribosomal cDNA rather than the site of insertion.
It is intriguing that intron homing during genetic crosses was not observed when
the 23S cDNA was in direct repeat orientation. Instead, repair of DSBs at the
23S cDNA in direct orientation led in all cases to deletion of the DNA flanked
by the cDNA and the neighbouring 23S gene. On the basis of the proposed DSBR
pathway for intron-homing the formation of a deletion between the 23S gene and the 23S cDNA
arranged in the same orientation can be explained as resolution of the
canonical recombination intermediate in the crossing-over configuration. However, if formation of these deletions follows the
classical DSBR mechanism, it has to be postulated that resolution of the
Holliday junctions is nonrandom, since no intron-homing to the 23S cDNA in direct repeat orientation was detected, which
corresponds mechanistically to gene conversion without associated crossing-over (Fig.
8
).
ACKNOWLEDGEMENTS
We thank J. van Dillewijn for help with crosses and tetrad analysis, N. Roggli
for preparing the figures, W. Zerges and K. Redding for critical reading of the
manuscript. This work was supported by grant no. 31-34014.92 from the Swiss National Fund to J.-D.R., grants 92-37301-7682 from the US Department of Agriculture and F-1164 from the Robert A. Welch Foundation to D.H.L.
REFERENCES
1 Weiffenbach, B. and Haber, J.E. (1981) Mol. Cell. Biol.1, 522-534.MEDLINE Abstract
2 Fairhead, C. and Dujon, B. (1993) Mol. Gen. Genet.240, 170-180.MEDLINE Abstract
3 Bennett, C.B., Lewis, A.L., Baldwin, K.K. and Resnick, M.A. (1993) Proc. Natl. Acad. Sci. USA90, 5613-5617.MEDLINE Abstract
20 Lambowitz, A. M. and Belfort, M. (1993) Annu. Rev. Biochem.62, 587-622.
21 Woessner, J.P., Masson, A., Harris, E.H., Bennoun, P., Gillham, N.W. and Boynton, J.E. (1984) Plant Mol. Biol. 3, 177-190.
22 Rochaix, J.D., Mayfield, S., Goldschmidt-Clermont, M. and Erickson, J. (1988) In Shaw, C.H. (ed.), Plant Molecular Biology-A Practical Approach. IRL Press, Oxford, pp. 253-275.
23 Feinberg, A.P. and Vogelstein, B. (1983) Anal. Biochem.132, 6-13.MEDLINE Abstract
35 Clyman, J. and Belfort, M. (1992) Genes Dev.6, 1269-1279.MEDLINE Abstract
36 Gillham, N.W., Boynton, J.E. and Harris, E.H (1991) In Bogorad, L. and Vasil, I.K. (eds), Cell Culture and Somatic Cell Genetics of Plants, vol. 7a. Academic Press, San Diego, pp. 55-92.
37 Zerges, W. and Rochaix, J.D. (1994) Mol. Cell. Biol. 14, 5268-5277.MEDLINE Abstract
38 Nickelsen, J., van Dillewijn, J., Rahire, M. and Rochaix, J.-D. (1994) EMBO J.13, 3182-3191.MEDLINE Abstract
39 Lin, F.L., Sperle, K. and Sternberg, N. (1984) Mol. Cell. Biol. 4, 1020-1034.MEDLINE Abstract
40 Cerutti, H., Osman, M., Grandoni, P. and Jagendorf, A.T. (1992) Proc. Natl. Acad. Sci. USA89, 8068-8072.MEDLINE Abstract
41 Pang, Q., Hays, J.B. and Rajagopal, I. (1993) Nucleic Acids Res.21, 1647-1653.MEDLINE Abstract
42 Eddy, S.R. and Gold, L. (1992) Proc. Natl. Acad. Sci. USA 89, 1544-1547.MEDLINE Abstract
*
To whom correspondence should be addressed
Present addresses:
+
Biotechnology Laboratory, University of British Columbia, Vancouver, BC V6T 1Z3,
Canada and
[sect]
Department of Molecular Biology, Horticulture Research International, Worthing
Road, Littlehampton, West Sussex BN17 6LP, UK
