JXB Advance Access originally published online on November 22, 2004
Journal of Experimental Botany 2005 56(409):1-14; doi:10.1093/jxb/eri025
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
REVIEW ARTICLE |
The repair of double-strand breaks in plants: mechanisms and consequences for genome evolution
Botany II, University of Karlsruhe, D-76128 Karlsruhe, Germany
* Fax: +49 721 6084874. E-mail: holger.puchta{at}bio.uka.de
Received 30 July 2004; Accepted 9 September 2004
 |
Abstract |
|---|
 Top Abstract IntroductionMeans to make endsDSB repair by homologous...DSB repair by non-homologous...Biotechnological applications:...OutlookReferences |
|---|
The efficient repair of double-strand breaks (DSBs) in genomic DNA is important for the survival of all organisms. In recent years, basic mechanisms of DSB repair in somatic plant cells have been elucidated. DSBs are mainly repaired by non-homologous end-joining (NHEJ). The repair can be associated with deletions, but also insertions due to copying genomic sequences from elsewhere into the break. Species-specific differences of NHEJ have been reported and an inverse correlation of deletion size to genome size has been postulated, indicating that NHEJ might contribute significantly to evolution of genome size. DSB repair by homologous recombination (HR) might also influence genome organization. Whereas homology present in an allelic or an ectopic position is hardly used for repair, the use of homologous sequences in close proximity to the break is frequent. A ‘single-strand annealing’ mechanism that leads to sequence deletions between direct repeats is particularly efficient. This might explain the accumulation of single long terminal repeats of retroelements in cereal genomes. The conservative ‘synthesis-dependent strand annealing’ mechanism, resulting in conversions without crossovers is also prominent and seems to be significant for the evolution of tandemly arranged gene families such as resistance genes. Induction of DSBs could be used as a means for the controlled manipulation of plant genomes in an analogous way for the use of marker gene excision and site-specific integration.
Key words: Arabidopsis, deletion formation, gene conversion, homologous recombination, I-SceI, non-homologous end-joining, single-strand annealing, synthesis-dependent strand-annealing, tobacco
|
Introduction |
|---|
|
TopAbstractIntroductionMeans to make endsDSB repair by homologous...DSB repair by non-homologous...Biotechnological applications:...OutlookReferences |
|---|
Double-strand breaks (DSBs) have to be eliminated before genomes can be replicated. Therefore, the repair of DSBs is critical for the survival of all organisms. Generally, DSBs can be repaired via two different pathways, either via homologous recombination (HR) or via non-homologous end-joining (NHEJ; also known as illegitimate recombination). Whereas in the former pathway the sequences are linked in regions that are identical to each other, in the latter the sequence information does not play a major role in the rejoining of the two double strands. An important question of recent research has been to find out under what conditions each pathway is used (for a general review see Paques and Haber, 1999
). Many studies have been performed using bacteria and, in the last decade, the basic mechanisms of DSB repair has been determined in yeast. In recent years the focus of research has shifted to higher eukaryotes, particularly as rare cutting restriction endonucleases like I-SceI (Perrin et al., 1993) have become available for the induction of unique site-specific genomic DSBs (Jasin, 1996). The first report on the use of the endonuclease I-SceI in a higher eukaryote, tobacco, dates back to 1993 (Puchta et al., 1993). Since then, in the mammalian field, more than 100 studies on DSB repair using restriction endonucleases have appeared in the literature, and on plants many studies, not only based on restriction enzymes but also on transposable elements as means to induce breaks, have been published. Based on these data a broader view of the pathways involved in DSB repair in plants is now possible. After a short introduction of the means to induce breaks, the main part of this review will focus on the role of the two basic pathways of DSB repair, HR and NHEJ. In addition, besides presenting a detailed picture of the ways that breaks are repaired, the possible evolutionary consequences for plant genomes will be discussed. Being able to predict the outcome of DSB repair reactions, the induction of DSB can also be used for the controlled induction of genomic changes in plants, a point that will be described in the final section of this review. As many recent publications have described the current understanding of the enzyme machinery involved in DNA repair in plants and the problems of gene targeting, these topics will not be addressed (Britt and May, 2003; Hanin and Paszkowski, 2003; Hartung and Puchta, 2004; Hays, 2002; Puchta, 2002, 2003a; Ray and Langer, 2002; Reiss, 2003). |
Means to make ends |
|---|
|
TopAbstractIntroductionMeans to make endsDSB repair by homologous...DSB repair by non-homologous...Biotechnological applications:...OutlookReferences |
|---|
The prerequisite for the study of DSB repair is the production of DNA breaks in the plant nucleus at a given time point. In principle, DNA that has already been broken can be transferred into the nucleus or a break can be induced in vivo (e.g. by radiation, application of chemicals, during the replication of transposons, or by the use of restriction endonucleases). Plasmid molecules linearized by restriction enzymes in vitro (‘broken’ DNA) are used as a substrate for the transient transformation of plant cells. Transfer can be mediated by PEG transfection, electroporation or particle bombardment (for transfer techniques see Potrykus and Spangenberg, 1995
). Indeed, by using extrachromosomal DNA, many basic features of HR and NHEJ can be characterized in plant cells. For NHEJ the production of filler DNAs has been demonstrated (Gorbunova and Levy, 1997) and for HR the occurrence of the non-conservative single-strand annealing mechanism has been proven (Puchta and Hohn, 1991; De Groot et al., 1992; Bilang et al., 1992). Although studies on extrachromosomal recombination have been productive in the determination of specific repair pathways it has to be kept in mind that the respective reactions occur early after transformation (Puchta et al., 1992), when the non-replicating plasmid DNA, in contrast to chromosomal DNA, is most probably not associated with chromatin. Moreover, hundreds or even thousands of DNA molecules enter a single cell simultaneously. Thus, the repair enzyme machinery might be outnumbered by the number of breaks that have to be processed. As a consequence, the repair process itself might differ, at least for specific pathways, from a reaction in which only one or a few DSBs are being processed in a single cell (for a detailed discussion see also Puchta and Meyer, 1994). Most of the more recent studies that are discussed in this review concentrate on the induction of genomic DSBs in vivo.
A classical way to induce DSBs is by the application of X-rays to the organism under study. Indeed irradiation has been used to induce mutations in plants for decades (e.g. in Arabidopsis: Redei and Li, 1969; Shirley et al., 1992). Using transgenic tobacco plants that carry intrachromosomal recombination substrates, it was demonstrated that the application of X-rays strongly enhances homologous recombination frequencies of the marker (Tovar and Lichtenstein, 1992; Lebel et al., 1993), a phenomenon that has been explained by the induction of DSBs into the marker DNA. However, one disadvantage of the use of X-rays is that, beside DNA breaks, other kinds of damage may be induced in the cell. In addition, although the number of breaks can be influenced by the application of different doses, it is not possible to influence where in the genome a break is induced. So in only some of the irradiated cells may a break occur at a specific locus. Thus, for a defined site the frequencies of recombination obtained by X-rays are many fold lower than with a site-specific break introduced uniquely at a precise location. For the induction of a site-specific DSB, transposable elements and rare cutting restriction endonucleases have been used in plants. DNA transposons are excised by the transposase out of the donor locus before integration somewhere else in the genome takes place. During this process the transient DSB is mainly repaired by NHEJ (Rinehart et al., 1997). If homologous sequences are close to a transposon a strong increase in HR has been detected (Athma and Peterson, 1991; Lowe et al., 1992; Xiao and Peterson, 2000). When a restriction site of a rare cutting nuclease like HO or I-SceI was included within the marker gene rather than a transposon, a similar enhancement of homologous recombination could be observed after expression of the enzyme, indicating that indeed the DSB is the inducing principal (Chiurazzi et al., 1996; Siebert and Puchta, 2002; Orel et al., 2003). By regulation of expression of the respective transposase (in the case of a non-autonomous element), or endonuclease, the time point of break formation can be controlled. Expression can be achieved by transient transformation (Puchta, 1999a) or by the control of a constitutive, organ-specific or inducible promoter after stable integration of the respective open reading frame into the plant genome (Schmidt-Puchta et al., 2004).
|
DSB repair by homologous recombination |
|---|
|
TopAbstractIntroductionMeans to make endsDSB repair by homologous...DSB repair by non-homologous...Biotechnological applications:...OutlookReferences |
|---|
The majority of studies on DSB repair in plants concentrate on homologous recombination. In part, this is because there is a general interest in HR for its application to genome manipulation (‘gene targeting’), but also the consequence of DSBs having been demonstrated to be intermediates in meiotic recombination in yeast. In meiosis, DSBs are induced by the Spo11 protein, which is thought to originate from the subunit A of the archaebacterial topoisomerase VI. These breaks are repaired exclusively by HR using the sequence of the allelic chromosome as a matrix (for review see Keeney, 2001
). As the focus of this is on DSB repair in somatic cells, readers are advised to consult recent reviews on meiotic recombination in plants for a deeper insight into this rapidly evolving field (Caryl et al., 2002; Jones et al., 2003). In somatic plant cells HR, in general, is a minor pathway, the efficiency of which is dramatically influenced by the availability of a homologous matrix. This not only influences the frequency of repair via HR versus NHEJ, but also may depend on the site of the homologous sequence in the genome as different kinds of HR pathways might be used. In principle, homologous information for repair can be copied from elsewhere in the genome (‘ectopic’), from the homologue as in meiosis (‘allelic’) or from the same chromosome (‘intrachromosomal’) (Fig. 1). Moreover, after replication the information could also be taken from the sister chromatid.
|
Ectopic recombination
In two independent studies, one based on DSB induction by the transposase of Ac, and the other by the endonuclease I-SceI, the question of how often an ectopic sequence can be used as the matrix for the repair of a DSB in plants was analysed (Shalev and Levy, 1997
; Puchta, 1999b). In both systems the transgenic donor locus and an unlinked acceptor locus at which the DSB was induced carried parts of non-functional marker genes, namely a ß-glucuronidase and a kanamycin resistance gene, respectively. Restoration of the marker was only possible via copying of information from the ectopic donor to the acceptor locus. Indeed, in both systems, restoration of the marker by homologous DSB repair could be detected. It was estimated that around one out of 10 000 breaks is repaired by HR. The detailed molecular analysis of several of the events indicated that the repair was not necessarily linked to homologous recombination of both ends of the break. DSB repair by a combination of HR and NHEJ was also found (Puchta, 1999b).
This result was reminiscent of studies in which an I-SceI-induced genomic break was repaired by the use of homologous sequences on an incoming T-DNA from Agrobacterium tumefaciens (Puchta et al., 1996). Integration of the T-DNA into the acceptor locus was mediated either by HR alone or by a combination of HR and NHEJ. As the same acceptor locus and the same donor sequences were used in both studies (T-DNA: Puchta et al., 1996; ectopic DNA: Puchta, 1999b), a comparison is possible and indicates that T-DNA is a better template for homologous DSB repair than an ectopic sequence. This is most probably due to the fact that the incoming T-DNA is more easily accessible for the copying processes than a chromosomal site. The original single-stranded T-DNA seems to be transformed to a double-stranded molecule in the nucleus (Tinland et al., 1994; Tzfira et al., 2003) and thus might be activated by the presence of factors involved in DNA repair or replication. Moreover, in contrast to a chromatin-packed ectopic locus, there might be fewer constraints of a steric nature for an interaction between the acceptor sequences and a ‘free’ T-DNA.
The fact that a break can be repaired by a combination of HR and NHEJ clearly demonstrates that, in somatic cells, the classical double-strand break repair (DSBR) model of recombination (Szostak et al., 1983) is not appropriate, as it predicts that both ends of the break would be repaired by HR in all cases. By contrast, other models such as the one-sided invasion (OSI) model (Belmaaza and Chartrand, 1994) or the synthesis-dependent strand-annealing (SDSA) model of recombination (Formosa and Alberts, 1986; Nassif et al., 1994; Keeler and Gloor, 1997) could explore the combination of HR and NHEJ on the repair of a DSB. However, whereas the OSI model could account for only the combination of HR and NHEJ, the SDSA model has the capacity to explain both kinds of events, repair by HR alone as well as by HR and NHEJ. To discriminate further between the models, experiments were performed with a T-DNA that carried homology to only one end of the break. The frequencies were compared with experiments in which a T-DNA was used harbouring homology to both ends of the break in the target locus (Fig. 2). According to the DSBR model the reaction should only proceed when homology to both ends of a break is present. The recombination frequencies obtained with the repair construct harbouring homology at both ends were only one-third higher than with the one-ended construct (Puchta, 1998). Even this small difference could be explained by one-sided invasion of the unique second homologous end of the respective T-DNA. Hence, homology to only one end of the DSB is sufficient for an efficient homologous recombination reaction and DSB repair reactions in plant cells are initiated by one-sided initiation events. Using Ockham's razor (postulating the least number of entities required for the explanation of a phenomenon) the best suited model for repair of genomic DSB in somatic plant cells is the SDSA model (Fig. 3) as it is based on one-sided initiation and it is able to describe recombination events due to HR as well as due to a combination of HR and NHEJ events. The application of the SDSA model seems to be even broader (see below): during repair of breaks by NHEJ, in some cases sequences from elsewhere in the genome are copied into the break (Rubin and Levy, 1997; Salomon and Puchta, 1998; Gorbunova and Levy, 1999). The respective junctions between broken DNA and inserted sequence are characterized by some base pairs (bp) of common homology, indicating a template switch, which is most easily explained by a SDSA mechanism (Fig. 4).
|
|
|
Using the SDSA model as the central paradigm of somatic DSB repair, two other important predictions can be made: somatic DSB repair events between ectopic sequences are not accompanied by crossovers and, in gene targeting experiments, not only perfect integration events at the target locus by homologous recombination will be found. Indeed, since the early days of gene targeting in plants, recombinants were characterized in which only one end of the targeting vector integrated by homology (Risseeuw et al., 1995
) or after copying sequences from the transgene locus the vector can be found integrated elsewhere in the genome (‘ectopic targeting’, Offringa et al., 1993). Such cases were found repeatedly and represent a major class of targeting events (Reiss et al., 2000; Hanin et al., 2001; Hohn and Puchta, 2003). Recently, it was demonstrated that ectopic recombination could be overcome by the integration of counter selection markers into the targeting vector outside the homologous regions. Using an efficient selection procedure an enrichment of ‘perfect’ targeted recombinants has been achieved for rice (Terada et al., 2002).
The SDSA model describes double-strand break repair without formation of double Holliday junctions during the repair process. Consequently, this reaction will lead either to the non-homologous repair of one end of the break or in the case of perfect homologous recombination to the exclusive occurrence of gene conversion events not accompanied by crossovers (Fig. 3). Thus, due to their biological role, pathways of gene conversion at least partly differ in somatic and meiotic cells in higher eukaryotes. Whereas in meiosis double Holliday junctions occur (Schwacha and Kleckner, 1996) and crossovers are a prerequisite for alterations in genetic coupling arrangements, in somatic cells the prevalence of gene conversion events avoids unpredictable deleterious genomic change by crossovers between ectopic sequences (Richardson et al., 1998). This might be particularly important for preserving the stability of the plant genome, as many species contain ample amounts of repetitive DNA and somatic change can be transferred to the germ line (Walbot, 1996).
Allelic recombination
During meiosis, DSBs are induced and repaired by the use of sequences from the homologue, resulting in either crossover or gene conversion. It was therefore important to test whether such a pathway is also operating in somatic cells. As gene targeting is not feasible for flowering plants it is, in contrast to mammalian (Moynahan and Jasin, 1997) and yeast cells (Hiraoka et al., 2000), not possible to manipulate directly allelic positions in plant genomes. To overcome this limitation a strategy was developed in which the transgene locus carrying the negative selectable marker cytosine deaminase (codA) with an I-SceI originally produced for the analysis of NHEJ (Salomon and Puchta, 1998; see below) was rendered non-functional by the induction of a DSB repair reaction. Plants homozygous for the modified sequence were then crossed with plants homozygous for the original transgene. Thus, plant lines could be obtained with two different alleles at the transgene locus on homologous chromosomes. One allele contained an I-SceI site within the codA gene and the other either a deletion or an insertion within the region of the codA expression cassette that rendered both the I-SceI site and the codA gene non-functional. This approach was similar to the one of Dooner and Martinez-Ferrez (1997) in which naturally occurring alleles of the bz gene of maize were combined. One of the alleles carried an active transposable element. However, the germinal excision of the element did not stimulate meiotic recombination, a fact that might be due to the programmed induction of DSBs at a specific time point during meiosis. After DSB induction by I-SceI expression selection was performed in heteroallelic tobacco plants for cells resistant to 5-fluorocytosine (5-FC) due to a non-functional codA gene. The occurrence of DSB-induced allelic recombination could be demonstrated (Gisler et al., 2002). However, the vast majority of events resulted in a loss of function of the marker gene due to NHEJ. The frequency of allelic recombination was estimated to be about 10–4, similar to that of ectopic recombination obtained after DSB repair in tobacco by two different studies (see above; Shalev and Levy, 1997; Puchta, 1999b). Allelic recombination is induced by DSB at a specific site by at least three orders of magnitude. Thus, compared with meiosis, allelic HR is not a significant DSB repair pathway in somatic cells. As in these cells DSB repair occurs mainly by NHEJ, a process which might lead to major genomic alterations (see below), the inefficiency of allelic HR seems to be a prerequisite for the fixation of major changes in the genome. As somatic changes can be transferred to the germ line, this might be an important factor responsible for the enhanced plasticity postulated for plant genomes.
Intrachromosomal recombination
By contrast with the use of allelic and ectopic positions, sequences close to the break on the same chromosome or on the sister chromatid are more efficient matrixes for repair. Sister chromatids can be used as a template in the G2 stage of the cell cycle, and this pathway works efficiently in mammals (Johnson and Jasin, 2000), however, due to experimental limitations it has not been possible to address this question in plant cells in connection with DSB induction until now (Molinier et al., 2004). By contrast, a number of studies have been published on intrachromosomal DSB repair in plants (Athma and Peterson, 1991; Chiruazzi et al., 1996; Orel et al., 2003; Siebert and Puchta, 2002; Xiao and Peterson, 2000; Xiao et al., 2000) and a more detailed picture of the frequencies of repair as well as on the pathways has been determined. Depending on the structure of the respective chromosomal locus at least two different kinds of pathways may be used to repair the break. Whereas one pathway results in the formation of a deletion between direct repeats, the product of the other is a gene conversion event (Fishman-Lobell et al., 1992). The deletion pathway is described by the single-strand annealing (SSA) model, whereas gene conversion can be explained by the synthesis-dependent strand-annealing (SDSA) model. The SSA model (Lin et al., 1984, 1990) describes a non-conservative reaction, resulting in a loss of information that is positioned between the annealing repeats (Fig. 3). SSA-like models have also been postulated to explain the loss of information during NHEJ (Nicolas et al., 1995). As discussed previously the SDSA model is conservative, it describes transfer of information from a homologous donor sequence to the break site without loss of sequences at the donor locus (Fig. 3). The efficiency of the pathways can be addressed by using a different configuration of marker genes. According to the SDSA model the orientation of the donor sequence to the break has no influence on gene conversion and the break should lie within the homologous region. By contrast, constructs measuring the efficiency of the SSA mechanism should have a break site within two direct repeats. Using the respective markers in a recent study (Orel et al., 2003) the frequencies of both pathways could be directly compared, indicating that the SSA pathway seems to be about five times more efficient than the gene conversion pathway.
However, of equal importance is to obtain an estimate of the number of breaks that are repaired in this situation by HR compared with NHEJ. In an attempt to determine this an assay was set up in tobacco based on interrupted overlapping halves of a ß-glucuronidase gene. Between the overlaps a negative selectable marker gene flanked by two I-SceI sites was inserted into the transgene (Fig. 5). As a consequence of I-SceI expression, after the excision of a negative selectable marker gene, a single genomic DSB had to be repaired. As the marker gene is lost, all resulting repair events (independent of their nature) could be isolated. It was demonstrated that the DSB was repaired in about one in three cases by HR. Thus, SSA is a highly efficient mechanism for DSB repair. SSA-like models can also be used to describe a class of NHEJ events (see below). Thus, SSA and SSA-like mechanisms might be the most prominent mode for the rejoining of broken DNA molecules in higher eukaryotes. If there are homologous sequences available in close proximity to the break they are regularly used for the annealing reaction, if they are not available, short patches of homology, which are present abundantly, might be used.
|
The demonstration that, in plants, DSBs can be repaired efficiently by the use of nearby homologies seems also to be relevant for plant genome evolution: In the retroelement BARE-1 only a 7–42-fold excess of single long terminal repeats (LTRs) over its own internal regions has been found (Vicient et al., 1999
; Kalendar et al., 2000; Shirasu et al., 2000). It seems surprising that such a high incidence of single LTRs was found, as this situation could only have arisen by a homology-dependent deletion of the internal region of the retrotransposon. BARE-1 is active in barley (Kalendar et al., 2000; Vicient et al., 2001) and during their spread retroelements integrate regularly into one another (San Miguel et al., 1996). If one assumes that transient breaks within the internal part of BARE-1 are induced either due to the activity of transposons (Xiao et al., 2000) or due to aberrant integration events of retroelements, a DSB-induced HR between the two flanking LTRs may yield solo LTRs (Fig. 6). Thus, it is tempting to speculate that accumulation of multiple solo LTRs in plant genomes is linked to the presence of active mobile genetic elements that are able to induce breaks during their life cycle.
|
If all experimental data are taken into account, a detailed picture can be obtained of how DSBs are repaired in somatic plant cells. The most efficient way to repair a break is by NHEJ (see below). If homologous sequences are available close to the break, the repair can take place in up to one-third of the cases by a single-strand annealing pathway. In about one out of 15 breaks, repair could also proceed via the SDSA pathway. This is two to three orders of magnitude more efficient than a gene conversion event that uses homology from an allelic or an ectopic site. DSB-induced gene conversion between members of tandemly arranged gene families might well play an important role in plant genome evolution. Various resistance genes are organized in tandem arrays in plant genomes and gene conversion events between the various members have been demonstrated by sequence analysis (Parniske et al., 1997
; Parniske and Jones, 1999). Such events might not necessarily be due to meiotic recombination but may, at least partially, also occur during vegetative growth. Indeed, pathogen attack is inducing intrachromosomal recombination between tandemly repeated sequences (Lucht et al., 2002). Moreover, certain sites in plant genomes might be less stable for repeated sequences than others. This is documented by findings that certain transgene sequences are deleted from the genome at high rates (Zubko et al., 2000), a phenomenon that has been postulated to be due to transient DSB-induction (Puchta, 2000). It seems possible that various gene conversion events, detected in tandemly arranged resistance gene clusters, might be correlated with higher rates of transient breaks in these regions of the plant genome (Ramakrishna et al., 2002). |
DSB repair by non-homologous end joining |
|---|
|
TopAbstractIntroductionMeans to make endsDSB repair by homologous...DSB repair by non-homologous...Biotechnological applications:...OutlookReferences |
|---|
In higher eukaryotes, NHEJ seems to be the main mode of DSB repair (Sargent et al., 1997
). NHEJ might lead, in a certain fraction of cases, to genomic changes such as deletions, insertions or various kinds of genomic rearrangements (Pipiras et al., 1998). As genomic alterations in meristematic cells can be transferred to the offspring, it is especially important to understand this pathway in plants. Different classes of NHEJ repair events were characterized in eukaryotes: on one side the repair of the break was accompanied by incorporation of filler DNAs, or the break ends were rejoined with or without deletions. The resulting junctions of the latter pathway contained either no or several nucleotides of sequence identity between the break ends. Junctions without homologies were explained by simple ligation, whereas small patches of homologous nucleotides were considered to be a prerequisite for the operation of a single-strand annealing-like mechanism (Nicolas et al., 1995; Mason et al., 1996; Lehman et al., 1994). In plants, all different kinds of reaction products have been detected with transfected plasmid molecules as well as with an induced DSB in the genome (Gorbunova and Levy, 1997; Salomon and Puchta, 1998). The repair of a genomic DSB was studied using tobacco harbouring a single I-SceI site between promoter and open reading frame of the cytosine deaminase gene (Fig. 4). After the selection of 5-FC resistant clones besides deletions (see below) a broad spectrum of unique and abundant genomic sequences was found inserted into chromosomal breaks of tobacco (Salomon and Puchta, 1998), a phenomenon drastically different from yeast, where only cDNA of retrotransposons or mitochondrial DNA was found to be inserted into breaks (Moore and Haber, 1996; Teng et al., 1996; Ricchetti et al., 1999). The presence of stretches of several nucleotides homologous between break site and copied sequences strongly argues for the occurrence of a synthesis-dependent strand annealing mediated copying of ectopic chromosomal DNA into the break. In most cases apparently only one 3' end was actively invading a genomic donor sequence (Salomon and Puchta, 1998), however, there were also indications for a combination of sequences of different origin due to independent invasion events of both 3' ends of the break (Gorbunova and Levy, 1997). Older models explain the amplification of genomic sequences by unequal homologous recombination between repetitive sequences in tandem or by replicative transposition of retroelements. However, these models do not explain how transposition-inactive sequences can be transferred into new positions within the plant genome. Using DSB repair, nearly any kind of genomic sequence that is available for a copying process can be inserted into new genomic positions. This phenomenon might be of general importance for higher plants and their genome evolution. DSB repair is also a prominent source of deletions. The explanation for the large differences in the nuclear DNA content of eukaryotes, particularly plants, known as the ‘C value paradox’, has been a matter of debate for some time (Cavalier-Smith, 1985; Dove and Flavell, 1988). Even closely related species with a similar phenotype may differ significantly in their diploid genome size. One mechanism responsible for these size differences could be related to a species-specific increase/reduction of repetitive sequences. In principle, genomes may become larger via duplications and insertions or smaller via deletions. Species-specific spread of retrotransposons was postulated as a main route for the enlargement of plant genomes (SanMiguel et al., 1996, 1998; Bennetzen and Kellog, 1997). Alternatively, deletions might reduce genome size and counterbalance enlargements (Petrov et al., 1996, 2000; Petrov, 2001). A comparison of deletion formation in somatic cells of tobacco and Arabidopsis, two dicotyledonous plant species with an over 20-fold difference in genome size (Bennett and Leitch, 1997), was performed using the same construct as described above (Kirik et al., 2000). Indeed, surprisingly large differences in DSB repair were found: whereas the mechanism of junction formation itself was not different between the two species. In tobacco, almost every second deletion event associated with the loss of function of the marker was accompanied by the insertion of filler sequences. By contrast, no insertions were isolated from Arabidopsis. The overall length of the deletions was about one-third shorter in tobacco than in Arabidopsis. Thus, as suggested theoretically before for insects (Petrov et al., 2000), an inverse correlation between genome size and the medium length of deletions could be detected. Although just two species were compared, and a final conclusion on the matter is, of course, not yet justified, it is nevertheless tempting to speculate that species-specific differences in DSB repair pathways could significantly contribute to the evolution of eukaryotic genome size. As Arabidopsis has undergone a number of segmental duplications or possibly a complete genome duplication during evolution (Arabidopsis Genome Initiative, 2000) deletion formation must have played a prominent role resulting in the small size of the present day genome of this plant species. Indeed, evidence that Arabidopsis has lost DNA over evolutionary time periods is supported by a recently completed elegant computer study based on the analysis of deletions within LTRs of retroelements (Devos et al., 2002).
But what are the molecular causes for the differences in DSB repair between Arabidopsis and tobacco? 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. Depending on the efficiency of DNA degradation more or less information will be lost. To address this question, particle bombardment was used to test the stability of linearized plasmid DNA in tobacco and Arabidopsis cells (Orel and Puchta, 2003). Indeed indications were found that plasmid DNA is degraded to a lesser extent in tobacco than Arabidopsis. If a break was induced between promoter and open reading frame of a marker gene the expression level has reduced by a factor of two in Arabidopsis compared with tobacco. This observation was in contrast to the similar expression levels obtained for both species when the break was introduced a long distance from the marker gene. Whereas in the first case exonucleolytic digestion of few nucleotides would result in the destruction of the expression cassette, in the latter case the removal of 2 kb would be a prerequisite for the destruction of the ORF. Because endonucleolytic degradation would destroy all the information on a plasmid at the same rate, independent of a pre-existing break site, no differences should be found in the presence of such a nuclease activity. However, as the restoration of the marker ORF required a ligation 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 a Southern blot analysis revealed differences in the degradation of linearized plasmid DNA between Arabidopsis and tobacco, it seems that most of the effect is due to a more efficient exonucleolytic degradation of DNA in Arabidopsis. However one question remains: What is the reason for 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).
|
Biotechnological applications: marker gene excision and site-specific integration |
|---|
|
TopAbstractIntroductionMeans to make endsDSB repair by homologous...DSB repair by non-homologous...Biotechnological applications:...OutlookReferences |
|---|
After the elucidation of the main mechanisms of DSB repair in plants it is of course an obvious step to use the induction of breaks at specific sites for the controlled manipulation of plant genomes. Two major applications have been developed: The excision of sequences from the genome and the integration of transgenes into specific sites of the genome.
Marker gene excision
Site-specific recombinases have been used efficiently for the excision of selectable marker genes (for a review see Hohn et al., 2001; Hare and Chua, 2002; Puchta 2003a, b). However, using this approach, recognition sites of the recombinases are left behind in the genome. Thus, for the performance of multiple genomic changes, a combination of different site-specific recombination systems has to be applied. The development of alternative approaches for site-specific alterations of genomes is of great interest in biotechnology. A very promising approach would be the combined elimination of the transgene sequences and the respective recognition sequence. Therefore, the use of highly specific restriction endonucleases to remove transgenic sequences from the plant genome is a useful alternative to the established site-specific techniques. The removal of the recognition site of the enzyme makes the process irreversible. It has been demonstrated that a marker gene can be efficiently eliminated from the plant genome either via HR or via NHEJ (including religation) by inducing DSBs by a rare cutting restriction enzyme (Fig. 5; Siebert and Puchta, 2002). Thus, any sequence flanked by restriction sites could, in principle, be eliminated from the plant genome. In praxis, the open reading frame of the endonuclease, under an inducible promoter, could be included in the ‘I-SceI cassette’ of the transgene to eliminate all superfluous sequence from the genome in one step (Puchta, 2003b). The use of several other rare cutting endonucleases might also be useful for this purpose.
DSB-induced site-specific integration
Using co-transformation it could be demonstrated that different T-DNAs integrate into a single genomic locus (De Block and Debrouwer, 1991; De Neve et al., 1997). This indicates that a surplus of extrachromosomal DNA ends is recognized by the cell as breaks to be ligated, yielding a tandem arrangement of transgenes, which can then be linked to free genomic ends. It seems that the number of free genomic ends might be rate-limiting for the integration step of transgenes and indeed X-rays, which induce DSBs in the genome, can enhance transgene integration in plant cells (Köhler et al., 1989). T-DNA integration into a DSB was demonstrated directly in a study on NHEJ using the codA gene harbouring an I-SceI site (see above; Salomon and Puchta, 1998). In three out of four cases almost the entire sequence of the 2.4 kb long T-DNA was integrated. More detailed analysis on the frequencies of such events was performed recently using different site-specific recombinases. In approximately 1% of the transformants the T-DNA was found inserted into the break (Chilton and Que, 2003; Tzfira et al., 2003). Where I-SceI sites flanked the transformation marker on the T-DNA, integration of the DNA into the genomic I-SceI site by simple ligation could be demonstrated. This indicates that the T-DNA became double-stranded and was then cut, prior to integration, by the enzyme. These studies clearly demonstrate that DNA can integrate during DSB repair into the plant genome. Nevertheless, taking the frequencies, as well as the integration patterns, into account, it seems that DSB repair is not the only and not even the main mechanism of T-DNA integration. Various studies on T-DNA integration indicate that there are, in principle, two different classes of integration patterns (Fig. 7; Brunaud et al., 2002). In most cases, the right border is conserved in the recombinants and contains no or very little microhomology to the target locus whereas longer homologies exist between the left border (combined with truncations) and the preinsertion site. In these events the two ends behave differently, and it has been postulated that integration occurs via single-strand annealing followed by the right border being linked to genomic sequences directly or indirectly by the VirD2 protein that is covalently bound to the 5' end of the T-strand (Tinland, 1996). In a second class of events, not only the left but also the right border is partly truncated in the recombinants, and common nucleotides can be found in most cases at the junction site also between the right border and the preinsertion. This pattern of integration might be indicative for DSB repair. T-DNA molecules are transferred as single strands to the nucleus, where some of them become double-stranded before integration (Tinland et al., 1994; Tzfira et al., 2003). By interaction with break ends, a single- or double-stranded T-DNA could be copied into the genome via synthesis-dependent strand annealing. Alternatively, a double-stranded T-DNA could also be integrated at the break by two consecutive single-strand annealing reactions.
|
Homologous integration into a genomic DSB was also achieved by the use of I-SceI. In up to 1% of the transformation events the target vector integrated into the homologous target locus (Puchta et al., 1996
). Besides events in which both ends of the target vector were integrated by HR, combinations of HR and NHEJ were also found and, indeed, targeting with a vector with homology only to one end of the break resulted in similar frequencies (Puchta, 1998; see also Fig. 2). However, this strategy is not applicable for targeting genes at will, as such a break can only be induced at the specific recognition sites of the rare cutting restriction enzyme. The technique can only be used for the integration of transgenes at a specific predetermined site in the genome and is, as such, an alternative to site-specific integration mediated by sequence specific recombinases (for a review see Ow, 2002). However, an alternative would be to construct restriction endonucleases that recognize specific sequences in the gene of interest. Indeed, chimeric endonucleases (Kim et al., 1996; Chevalier et al., 2002; Epinat et al., 2003) and zinc finger nucleases (for a review see Carroll, 2004) are being developed to fulfil this task. The application of zinc finger nucleases for gene targeting in vertebrates has already been demonstrated (Bibikova et al., 2003; Porteus and Baltimore, 2003) and it may be only a matter of time before the first reports of the use of the technique will appear for plants.
Another gene targeting approach using DSB induction for the production of a recombination intermediate has been developed recently for Drosophila melanogaster by the group of Kent Golic (Rong and Golic, 2000; Rong et al., 2002). In this method the construct for targeting is integrated in the host genome flanked by two recognitions sites of a site-specific recombinase and includes a site for a rare cutting restriction endonuclease. By induced expression of the site-specific recombinase a DNA circle is excised from the genome. This circle is then linearized after the restriction enzyme (in this case I-SceI) has been expressed resulting in an ‘activated’ DNA molecule with both ends homologous to the target sequence (for a simplified scheme see Fig. 8). In the female germline of Drosophila, gene targeting occurred in about one out of 500 cases. As gene targeting is a technique urgently needed in plant biotechnology there is considerable interest in determining whether the strategy is also applicable to plants (for discussion see also Kumar and Fladung, 2001). On the one hand the set-up of the system seems to be very complex, as besides construction of a donor sequence with sites for a recombinase and a restriction enzyme, expression cassettes of both enzymes have to be included into the transgene construct or supplied in trans. On the other hand, if the reaction occurs in an efficient way in planta, every single seedling should represent an excision event. Thus, by the use of respective marker genes, large numbers of plants can be produced and easily screened.
|
One has to see if and when the two new approaches, both based on the induction of DSBs, will help to solve the long-lasting problem of gene targeting in plants.
|
Outlook |
|---|
|
TopAbstractIntroductionMeans to make endsDSB repair by homologous...DSB repair by non-homologous...Biotechnological applications:...OutlookReferences |
|---|
Over the last ten years a substantial body of information has accumulated relating to DSB repair in plants. The characterization of different pathways involved in DSB repair and the estimation of the respective frequencies makes it now possible for us to draw a quite detailed picture of the changes that occur during DSB repair in the plant genome. As sequenced genomes of an increasing number of plant species become available and, as a consequence, the evolutionary changes can be analysed at the whole genome level, current knowledge of DSB repair pathways will help to explain why genomes evolved in size and organization, resulting in the complexity that is found today. However, although major insights have been made, this picture of DSB repair in plant cells is far from complete, and neither questions on DSB-induced sister chromatid exchange nor on DSB-induced translocation have been addressed until now. A major challenge will be to refine techniques of genome manipulation further and this holds true particularly for the set-up of a reliable targeting technique. The induction of meiotic recombination by site-specific DSBs may be an important step along this path over the coming years.
|
Acknowledgements |
|---|
I would like to thank the different members of my group, especially Siegfried Salomon, Angela Kirik, Ralph Siebert, Brigit Gisler, Nadja Orel, Waltraud Schmidt-Puchta, Petra Oswald, and Christa Fricke, who over the years have carried out the work on DSB repair in plants, and the Deutsche Forschungsgemeinschaft and the European Union for funding my research in this area.
|
References |
|---|
|
TopAbstractIntroductionMeans to make endsDSB repair by homologous...DSB repair by non-homologous...Biotechnological applications:...OutlookReferences |
|---|
Arabidopsis Genome Initiative. 2000. Analysis of the genome sequence of the flowering plant Arabidopsis thaliana. Nature 408, 796–815.[CrossRef][Medline]
Athma P, Peterson TA. 1991. Ac induces homologous recombination at the maize P locus. Genetics 128, 163–173.[Abstract]
Belmaaza A, Chartrand P. 1994. One-sided invasion events in homologous recombination at double-strand breaks. Mutation Research 314, 199–208.[ISI][Medline]
Bennett MD, Leitch IJ. 1997. Nuclear DNA amounts in angiosperms—583 new estimates. Annals of Botany 80, 169–196.
Bennetzen JL, Kellog EA. 1997. Do plants have a one-way ticket to the genomic obesity? The Plant Cell 9, 1509–1514.[CrossRef][ISI][Medline]
Bibikova M, Beumer K, Trautman JK, Carroll D. 2003. Enhancing gene targeting with designed zinc finger nucleases. Science 300, 764.
Bilang R, Peterhans A, Bogucki A, Paszkowski J. 1992. Single-stranded DNA as recombination substrate in plants assessed by stable and transient expression. Molecular and Cellular Biology 12, 329–336.
Britt AB, May GD. 2003. Re-engineering plant gene targeting. Trends in Plant Science 8, 90–95.[CrossRef][ISI][Medline]
Brunaud V, Balzergue S, Dubreucq B, et al. 2002. T-DNA integration into the Arabidopsis genome depends on sequences of pre-insertion sites. EMBO Reports 3, 1152–1157.[CrossRef][ISI][Medline]
Carroll D. 2004. Using nucleases to stimulate homologous recombination. Methods in Molecular Biology 262, 195–207.[Medline]
Caryl AP, Jones GH, Franklin FC. 2002. Dissecting plant meiosis using Arabidopsis thaliana mutants. Journal of Experimental Botany 54, 25–38.
Cavalier-Smith T. 1985. The evolution of genome size. Chichester: John Wiley and Sons.
Chevalier BS, Kortemme T, Chadsey MS, Baker D, Monnat RJ, Stoddard BL. 2002. Design, activity, and structure of a highly specific artificial endonuclease. Molecular Cell 10, 895–905.[CrossRef][ISI][Medline]
Chilton MD, Que Q. 2003. Targeted integration of T-DNA into the tobacco genome at double-stranded breaks: new insights on the mechanism of T-DNA integration. Plant Physiology 133, 956–965.
Chiurazzi M, Ray A, Viret J-F, Perera R, Wang X-H, Lloyd A, Signer ER. 1996. Enhancement of somatic intrachromosomal homologous recombination in Arabidopsis by HO-endonuclease. The Plant Cell 8, 2057–2066.[Abstract]
De Block M, Debrouwer D. 1991. Two T-DNA's co-transformed into Brassica napus by a double Agrobacterium tumefaciens infection are mainly integrated at the same locus. Theoretical and Applied Genetics 82, 477–484.
De Groot MJA, Offringa R, Does MP, Hooykaas PJJ, van den Elzen PJM. 1992. Mechanisms of intermolecular homologous recombination in plants as studied with single- and double-stranded DNA molecules. Nucleic Acids Research 20, 2785–2794.
De Neve M, De Buck S, Jacobs A, Van Montagu M, Depicker A. 1997. T-DNA integration patterns in co-transformed plant cells suggests that T-DNA repeats originate from co-integration of separate T-DNAs. The Plant Journal 11, 15–29.[CrossRef][ISI][Medline]
Devos KM, Brown JK, Bennetzen JL. 2002. Genome size reduction through illegitimate recombination counteracts genome expansion in Arabidopsis. Genome Research 12, 1075–1079.
Dooner KH, Martinez-Ferrez IM. 1997. Germinal excision of the maize transposon activator do not stimulate meiotic recombination or homology-dependent repair at the bz locus. Genetics 147, 1923–1932.[Abstract]
Dove GA, Flavell RB. 1988. Genome evolution. London: Academic Press.
Epinat JC, Arnould S, Chames P, Rochaix P, Desfontaines D, Puzin C, Patin A, Zanghellini A, Paques F, Lacroix E. 2003. A novel engineered meganuclease induces homologous recombination in yeast and mammalian cells. Nucleic Acids Research 31, 2952–2962.
Fishman-Lobell J, Rudin N, Haber JE. 1992. Two alternative pathways of double-strand break repair that are kinetically separable and independently modulated. Molecular and Cellular Biology 12, 1292–1303.
Formosa T, Alberts BM. 1986. DNA synthesis dependent on genetic recombination: characterization of a reaction catalyzed by purified bacteriophage T4 proteins. Cell 47, 793–806.[CrossRef][ISI][Medline]
Gisler B, Salomon S, Puchta H. 2002. The role of double-strand break-induced allelic homologous recombination in somatic plant cells. The Plant Journal 32, 277–284.[CrossRef][ISI][Medline]
Gorbunova V, Levy AA. 1997. Non-homologous DNA end-joining in plant cells is associated with deletions and filler DNA insertions. Nucleic Acids Research 25, 4650–4657.
Gorbunova V, Levy AA. 1999. How plants make ends meet: DNA double-strand break repair. Trends in Plant Science 4, 263–269.[CrossRef][ISI][Medline]
Hanin M, Volrath S, Bogucki A, Briker M, Ward E, Paszkowski J. 2001. Gene targeting in Arabidopsis. The Plant Journal 28, 671–677.[CrossRef][ISI][Medline]
Hanin M, Paszkowski J. 2003. Plant genome modification by homologous recombination. Current Opinion in Plant Biology 6, 157–162.[CrossRef][ISI][Medline]
Hare PD, Chua NH. 2002. Excision of selectable marker genes from transgenic plants. Nature Biotechnology 20, 575–586.[ISI]