Both ATM and ATR promote the efficient and accurate processing of programmed meiotic double-strand breaks


  • Kevin M. Culligan,

    Corresponding author
    1. Department of Biochemistry and Molecular Biology, University of New Hampshire, Durham, NH 03824, USA, and
    Search for more papers by this author
  • Anne B. Britt

    1. Department of Biochemistry and Molecular Biology, University of New Hampshire, Durham, NH 03824, USA, and
    2. Section of Plant Biology, University of California, Davis, CA 95616, USA
    Search for more papers by this author

*(fax +1 603 862 4013; e-mail; fax +1 530 752 5410; e-mail


The ATM and ATR protein kinases play central roles in the cellular response to double-strand breaks (DSBs) by regulating DNA repair, cell-cycle arrest and apoptosis. During meiosis, SPO11-dependent DSBs are generated, initiating recombination between homologous chromosomes. Previous studies in mice and plants have shown that defects in ATM result in the appearance of abnormally fragmented chromosomes. However, the role of ATR in promoting normal meiosis has not yet been elucidated. Employing null Arabidopsis mutants of ATR and ATM, we demonstrate here that although atr mutants display no obvious defects in any phase of meiotic progression, the combination of defects in atr and atm exacerbates the fragmentation observed in the atm single mutant, prevents complete synapsis of chromosomes, and results in extensive and persistent interactions between non-homologous DNAs. The observed non-homologous interactions require the induction of programmed breaks: the combination of either the atm single or the atr atm double mutant with a spo11 defect eliminates the ectopic interactions observed in the double mutant, as well as significantly reducing the fragmentation seen in atm or in atr atm. Our results suggest that ATM is required for the efficient processing of SPO11-dependent DSBs during meiosis. They also indicate that ATM and ATR act redundantly to inhibit sustained interactions between non-homologous chromatids, and that these ectopic interactions require SPO11 activity.


Meiosis is a specialized form of cell division involving one round of replication followed by two rounds of cell division, thereby reducing chromosomal content to produce haploid gametes. One hallmark process of meiotic cells is the recombination that occurs between homologous chromosomes during the prophase of the first meiotic division. Sites of meiotic recombination are initiated by the formation of DNA double-strand breaks (DSBs) (Sun et al., 1989). In yeasts, plants and most animals this activity is mediated by a topoisomerase-like endonuclease, which is termed SPO11 (Keeney and Neale, 2006; Keeney et al., 1997). Of the many SPO11-dependent DSBs that are formed (Moens et al., 2002), only a select few result in the exchange of chromosome arms that is required, in most organisms (Page and Hawley, 2003), for reductional division. However, all SPO11-generated DSBs must be repaired through homologous recombination to prevent mutagenesis and chromosome fragmentation.

Studies in yeasts and animals suggest that the molecular events leading to SPO11-dependent homologous recombination in meiotic cells are analogous to DSB repair in mitotic cells (Cromie et al., 2006; Haber et al., 2004; Hunter and Kleckner, 2001; Merker et al., 2003). Accordingly, most mutants defective in the repair of DSBs [via homologous recombination (HR) pathways] also display fertility defects. Models for homology-dependent mitotic DSB repair by homologous recombination are continually refined and expanded, as we gain further insight into the molecular players and their biochemical activities (Haber et al., 2004). Although there are multiple pathways for homology-dependent (RAD51-dependent) repair, all exhibit a requirement for the MRE11–RAD50–NBS1 (MRN) complex in 5′ end degradation, although the exonuclease activity of the complex itself is not required for this process. This hinge-like protein complex interacts with free chromosome ends, and activates a phosphoinositide-3 kinase (PI3K)-like protein kinase: termed ATM in animals, or MEC1 in yeast (MEC1 and, to a lesser degree, TEL1 activities in yeast perform the functions of both ATM and ATR in animals; see the discussion below). ATM (or MEC1) in turn activates the 5′ exonuclease activity fostered by the MRN complex, producing the free 3′ ends required for homologous recombination. Thus, PI3K-kinase like protein kinases play a role not only in signaling the presence of mitotic DSBs, but also in activating their repair, in both yeasts and mammals (Abraham, 2001; Jazayeri et al., 2006).

The role of ATM in the maintenance of genomic stability in plants has been genetically characterized in both mitotic and meiotic cells. For example, atm mutants in plants are sensitive to ionizing radiation, and display reduced fertility. Analysis of the meiotic stages in Arabidopsis atm mutants showed fragmentation of chromosomes in the early prophase I, which probably explains the reduced fertility of the mutants (Garcia et al., 2003), with similar results having been observed in mammals (Barlow et al., 1997; Xu et al., 1996). However, it has not yet been determined (in mammals or plants) whether the meiotic fragmentation results from persisting programmed DSBs, or from some other source of damage.

In mammals, ATM and the related protein kinase ATR have been shown to associate along synapsed meiotic chromosomes (Keegan et al., 1996; Moens et al., 1999). Furthermore, ATM is required for γ-H2AX phosphorylation at sites of DSBs in meiotic cells (Bellani et al., 2005), whereas ATR has been shown to co-localize with γ-H2AX during the early prophase I (Perera et al., 2004). These studies are clearly consistent with a role for ATM in the processing of meiotic DSBs, but leave an open question as to the significance of the presence of ATR. Because null mutant alleles of ATR are lethal in animals, the role of ATR in processing DSBs during meiosis has been difficult to assess directly.

In addition to the roles of ATM and ATR in processing DSBs, both kinases are key players in mitotic cell-cycle arrest and apoptotic responses to DSBs (Abraham, 2001; Shiloh, 2003). It is possible that ATM and/or ATR could play similar roles during the meiotic cell cycle, monitoring the progression of DSB repair, before progressing into the next meiotic cell stage, as has been proposed previously (Barlow et al., 1998; Hochwagen and Amon, 2006). In support of this, mutants of MEC1, the ATM/ATR homolog in Saccharomyces cerevisiae, progress prematurely into the first meiotic division in the presence of unrepaired DSBs, unlike wild-type (WT) cells (Lydall et al., 1996). However, the ATM-dependent damage response pathway is not essential for meiotic arrest or apoptosis responses, as ATM-deficient mice themselves display meiotic arrest and apoptotic degeneration early in prophase I (Barlow et al., 1997; Xu et al., 1996), as well as chromosomal fragmentation. One possibility is that ATR-dependent pathways can also regulate meiotic progression.

Plants display neither apoptosis nor a terminal meiotic arrest in response to unrepaired DSBs; various homologous recombination mutants of Arabidopsis, such as dmc1 (Couteau et al., 1999) and rad51 (Li et al., 2004), the latter of which displays an abundance of chromosomal fragmentation in early prophase I, complete meiosis. However, these studies did not measure the rate of progression of the cell cycle, and it is possible that a transient arrest occurs. Because Arabidopsis atr single and atr atm double mutants are viable (Culligan et al., 2004), Arabidopsis provides a valuable model system for studying the role of ATR and ATM during the upstream meiotic recombination process.

Employing null Arabidopsis ATR and ATM mutants, we show here that both ATR and ATM can contribute to the repair of SPO11-dependent DSBs. Although the atr single mutant does not display meiotic defects, the atr mutation in the atm background compounded the fragmentation observed in the single atm mutant, suggesting a partially redundant role for ATR and ATM during meiosis. Mutations inducing defects in SPO11 reversed both the fragmentation and the extensive ectopic interactions observed in the atm single and atr atm double mutant lines. These results suggest that ATM (and possibly ATR) facilitate the timely processing of SPO11-dependent DSBs during meiosis, and that both proteins play a critical role in preventing persistent and indiscriminant interactions between chromosomes during meiosis.


ATR and ATM perform synergistic functions in plant meiosis

To define the individual roles of ATR and ATM in Arabidopsis meiosis, we stained pollen mother cells (PMCs) of WT, atr and atm single mutants, and the atr atm double mutant with 4′,6-diamidino-2-phenylindole (DAPI) to compare the meiotic progression in each line. Figure S1 shows examples of WT meiosis stages I and II. During leptonema of meiosis I, chromosomes appear diffuse but begin to condense. Here, SPO11-dependent DSBs are formed to initiate the crossing over (recombination) of homologs. Sister chromosomes continue to condense in zygonema, appearing as thin strands, and begin to pair with their homolog. This pairing continues until completion in pachynema, forming bivalents with thicker strands. During diplonema/diakinesis, the bivalents condense and are connected by chiasmata. The bivalents further condense and align along the central plane of the cell in metaphase I, and the resulting recombinant homologs are pulled to opposite cell-spindle poles during anaphase I. During meiosis II, the sister chromatids condense and migrate to the metaphase plate, then separate during anaphase II to produce four recombinant sets of chromatids, and the resulting gametophytes (pollen tetrad) are formed.

A previous study (Garcia et al., 2003) employing atm T-DNA insertion mutants showed that ATM is required for normal meiotic progression, resulting in a semisterile phenotype. To determine the role of ATR in meiotic progression, we compared the single atr and atm mutants with WT during meiosis I. As shown in Figure 1, the atr mutant appears very similar to WT, forming paired (normal) bivalents during pachytene and metaphase I, followed by the normal separation of univalents during anaphase I. Meiosis II was also indistinguishable from that of WT (data not shown). This indicates that ATR is not essential for meiosis in plants. Although atm mutants display normal pairing of bivalents in pachytene, similar to atr and WT, chromosome fragmentation is apparent shortly thereafter in metaphase I (e.g. Figure 1m, arrow indicates a chromosome fragment excluded from the meiotic division spindle). Other types of fragments are apparent in anaphase I, when the chromosomes separate: some fragments seem to be attached to the spindle, perhaps torn between two chromosomes. Additional fragments could also be seen at later stages in meiosis II (data not shown; Garcia et al., 2003).

Figure 1.

 Comparison of meiosis-I stages of pollen mother cells (PMCs) in wild type (WT), atr and atm single mutants, and in the double mutant atr atm.
(a–e) Leptotene/early zygotene. Chromosome loops (bouquet) are apparent in all lines, but free chromosome ends and fragments are sometime present in the atr atm double mutant [arrows in (d) and (e)].
(f–j) Pachytene. Chromosome pairing is complete in WT, and in the atr and atm mutants. Fully paired pachytene chromosomes were absent in the atr atm double mutant, but occasionally showed partial pairing [circle in (j)]. Fragments were also seen at ‘pachytene’ stages in the atr atm double mutant [arrows in (i)].
(k–o) Metaphase I. Chromosomes (bivalents) aligned normally at the metaphase plate in WT, and in atr and atm mutants, but atm often displayed fragments excluded from the metaphase plate [arrow in (m)]. Normal alignment of homolog pairs (bivalents) was not readily observed in the atr atm double mutant, showing entangled chromosomes (n) and fragmentation (o).
(p–t) Anaphase I. Homologous chromosomes (univalents) are pulled to opposite poles in WT, and in atr and atm single mutants, but extensive fragmentation was observed in atm during homolog separation. Most chromosomal material remained at the metaphase plate in the atr atm double mutant.

In contrast to semisterile atm mutants, atr atm mutants are completely sterile (Culligan et al., 2004). This indicates that, athough ATR is not required for normal chromosome segregation in WT cells, it does becomes important in the atm mutant background. To better understand the nature of ATR’s role in atm-defective meiosis, we compared the meiotic progression of WT and the atr atm double mutant. In the zygonema of atr atm, we occasionally observed what appeared to be free chromosome ends and fragments (Figure 1d,e). This is in contrast to WT or single mutants, in which the condensing chromosomes form complete loops anchored by their telomeres, which cluster because of their association with the nucleolus (Armstrong et al., 2001). In addition, fully synapsed (pachytene) chromosomes were never observed in the double mutant: only a few of the >100 early prophase (zygonema in appearance) nuclei observed displayed any synapsis at all, and the synapsis was never complete (i.e. Figure 1j). In contrast, WT, atr or atm clearly displayed both partially (zygonema) and fully synapsed (pachynema) stages (n > 100 for each line). Additionally, we did not observe distinct metaphase-I bivalents in the double mutant. Instead, the bulk of the chromatin appeared as a clump at the metaphase plate (Figure 1n) and fragments were sometimes observed (Figure 1o). In the anaphase-I-stage meiocytes of the double mutant, the majority of the chromosomal DNA remained at the metaphase plate, whereas the centromeres were apparently pulled in opposite directions to the spindle poles (Figure 1s,t), leaving only thin strands of DNA connecting the chromosome arms and the centromeric regions. The phenotype of the double mutant indicates that either ATR or ATM is sufficient to promote chromosome synapsis. In contrast, whereas synapsis appears to be normal in the atm mutant, ATR does not completely compensate for the loss of ATM: abnormal chromosome fragmentation is still observed in the single atm mutant. As a result, chromosome segregation is often unequal, leading to irregular polyads (not shown; Garcia et al., 2003), and thus to semisterility in the atm mutant. Although the ATR mutant has no meiotic phenotype, it is also possible that ATR normally plays a role in meiosis in WT, and that ATM can completely compensate for the loss of this activity. However, this possibility cannot be tested by simple genetic analysis.

The atr atm double mutant, unlike rad51, exhibits ectopic interchromosomal interactions during anaphase I

Like the atr atm double mutant described here, homologous recombination mutants, such as rad51, display abnormal chromosomal fragmentation during meiotic progression, and fail to fully synapse (Li et al., 2004, 2005; Puizina et al., 2004). It is possible that the chromosomal fragmentation in atr atm is simply caused by the inability to initiate or activate homologous recombination complexes. In this case, we would expect to see a similar progression of meiotic stages in both the atr atm mutant and the rad51 mutant. To test this, we directly compared meiotic stages of rad51 (atrad51-1; Li et al., 2004) with atr atm mutant plants grown under identical conditions. During meiosis I, both rad51 and atr atm displayed the normal bouquet (condensation chromosome) formation during leptonema, but no pachynema stages were observed, as above. During metaphase I, both rad51 and atr atm displayed what appeared to be randomly distributed, broken and disorganized chromatin at the metaphase plate (Figures 1n,o and 2a,b). However, during anaphase I, the rad51 chromosomes appeared to untangle into multiple smaller fragments that remained near the metaphase plate (probably because of a lack of attachment to a centromere), whereas the atr atm chromosomes never separated from their entangled mass at the metaphase plate (Figures 1s,t and 2c,e versus Figure 2d,f). This trend continued during anaphase II (not shown), perhaps because the chromosomes failed to fully separate during meiosis I. This suggests that the meiotic progression of rad51 and atr atm is fundamentally different in that atr atm, but not rad51, chromosomes exhibit a persistent, irresolvable ectopic interaction.

Figure 2.

 Comparison of meiotic stages between the atr atm double mutant and the rad51 single mutant.
(a–b) Metaphase I.
(c–f) Anaphase I.

Defects in spo11-1 partially rescue atr- and atm-dependent chromosomal fragmentation, and eliminate the abnormal chromosomal interactions observed in the double mutant

Both ATM and ATR can induce a programmed response to DSBs, although they do so via different mechanisms. DSBs are induced in a programmed manner early in meiosis I, and SPO11 catalyzes the formation of the breaks. To determine whether the chromosome fragmentation observed in atm and atr atm mutants is primarily caused by a defect in response to programmed SPO11-dependent DSBs, we generated double- and triple-mutant combinations of atr and atm with spo11-1 (atr spo11, atm spo11 and atr atm spo11). The Arabidopsis genome encodes three SPO11 homologs. Two of these homologs, SPO11-1 and SPO11-2, have essential functions in meiosis (Stacey et al., 2006), and the phenotype of both mutants (sterility, with random segregation of chromosomes) is identical, and is also identical to their double mutant. It is not yet clear why in Arabidopsis two such similar genes are both functionally essential, rather than being redundant; perhaps the enzyme is a heteromeric complex. Nonetheless, based on its sterile phenotype, it is likely that spo11-1 mutants are defective in the formation of DSBs required for meiotic recombination and ordered chromosome segregation.

We first analyzed seed set in spo11, atr spo11-1, atm spo11-1 and atr atm spo11-1 mutants, as shown in Figure 3. As previously observed (Grelon et al., 2001), spo11-1 mutants are nearly sterile, producing approximately two seeds per silique, or about 5% of the seeds produced in the WT (Figure 3). More than half of these seeds appear abnormal (dark and/or shrunken). In contrast, the atr atm double mutant (WT for SPO11) produces no seeds at all, either normal or abnormal, compared with the atr and atm single mutants, which produce approximately 50 and approximately six seeds per silique, respectively (Culligan et al., 2004; Garcia et al., 2003). When spo11-1 is combined with atr atm, however, fertility (seed set) is partially restored, to about half that of spo11-1 (Figure 3). The progeny of the triple mutant were, however, largely defective, with only 20% germination, a rate similar to that observed for the spo11-1 single mutant (Table 1). These data show that the spo11-1 mutation partially rescues the complete sterility of the atr atm double mutant. This suggests that the defect in the double mutant lies in its inability to properly process and/or respond to SPO11-generated DSBs.

Figure 3.

 Quantification of the number of seeds per silique in the spo11 mutant combination lines and in the atr atm double mutant.
In the wild type (WT), siliques generally produce approximately 50 ‘normal’ seeds. Siliques from each line were individually dissected before opening, and seeds were counted as normal (full size, not wrinkled or shrunken) or abnormal (shrunken, wrinkled or >50% smaller than in WT).

Table 1.   Seed germination of spo11-1-1 mutant combination lines in comparison with wild type (WT)
LineTotal seeds plantedTotal germinatedPercentage germination
spo11, atr45810122%
spo11, atm4068621%
spo11, atr, atm77515820%

We next observed meiotic progression in the DAPI-stained PMCs of spo11-1, atr spo11-1, atm spo1-1 and atr atm spo11-1, as shown in Figure 4. As previously observed (Grelon et al., 2001), spo11-1 mutants fail to synapse, lacking fully paired chromosomes during pachynema, with either none or few bivalents at metaphase. Because of the lack of chiasmata, univalents segregate randomly during anaphase I. In all mutant combinations, we observed this general pattern: we did not observe any fully synapsed pachytene-stage meiocytes (data not shown), and very few, if any, bivalents. Most spo11-1 meiocytes showed 10 univalents, and these displayed random segregation during anaphase I (Figure 4), and unequal chromosome segregation, as expected (Grelon et al., 2001; data not shown). Most strikingly, the spo11-1 mutation completely suppressed the abnormal, indiscriminate ‘stickiness’ observed in the atr atm double mutant. Loss of this adhesion is likely to account for the enhanced fertility (versus atr atm) observed in the triple mutant. This suggests that the induction of programmed DSBs is required for the subsequent events leading to abnormal chromosome cohesion.

Figure 4.

 The spo11-1-1 mutation partially rescues chromosomal fragmentation in atm single and atr atm double mutants. A comparison of anaphase I in spo11 mutant combination lines is shown, with three representative examples for each line.
(a–c) spo11 mutants display the typical random segregation of 10 univalents.
(d–f) The spo11 atr double mutant is similar to the spo11 single mutant.
(g–i) The spo11 atm double mutant shows much less fragmentation compared with the atm single mutant. Arrows in (h) and (i) show that fragments are still present.
(j–l) The spo11 atr atm triple mutant also displays much less fragmentation compared with the atr atm double mutant. In (k) there appear to be 11 chromatids, suggesting a single fragmented chromosome, and in (l) there are multiple smaller fragments (arrow).

However, the spo11-1 mutation did not completely eliminate the presence of chromosome fragments in the atr and atm mutants. As shown in Table 2, the low level of fragmentation observed in atr mutants remained constant (although the rarity of these events make a change of frequency hard to detect), whereas the fragments observed in the atm single and atr atm double mutants dropped to about a third of the SPO11 level. It is possible that these fragments result from residual programmed break activity, perhaps encoded by SPO11-1, as the allele we used (spo11-1-1) shows some residual activity (Stacey et al., 2006). It is also possible that these residual breaks are encoded by SPO11-2, although this seems unlikely given the identical degree of semisterility observed in the spo11-1 spo11-2 double mutants. Alternatively, these breaks may have been induced spontaneously in the mutants at some point during the meiotic cell cycle.

Table 2.   Quantification of fragmentation during anaphase I
LineTotal observed, anaphase ITotal observed, fragmented (at least one fragment at anaphase I)Percentage fragmented
Wild type10400%
atr, atm2929100%
spo11, atr6123%
spo11, atm821518%
spo11, atr, atm721926%


The fragmentation observed in atm lines largely results from SPO11 activity

Garcia et al. previously described the semisterile phenotype and meiotic prophase defects of the Arabidopsis atm-1 mutant. Although homologs do synapse, and reductional segregation occurs in this mutant, there is extensive fragmentation that is probably responsible for the partial sterility. Essentially the same phenotype is also observed in atm knock-out mice (Xu and Baltimore, 1996): synapsis is delayed, but still occurs, together with extensive fragmentation. Given the well-established role of ATM in DSB signaling, it has been suggested that this fragmentation may be caused by a defect in the signaling and/or processing of programmed meiotic DSBs. However, it is also possible that these breaks reflect the accumulated spontaneous somatic DNA damage that has persisted ‘unchecked’ in the absence of ATM. Here, we demonstrate that atm lines defective in SPO11, the endonuclease that induces meiotic breaks, display a nearly complete correction of the fragmentation phenotype, consistent with the idea that these breaks are largely to the result of programmed SPO11 activity. This suggests that ATM, in WT cells, plays an important role in the processing of programmed breaks, the signaling of their presence (presumably in order to slow progression through meiosis) or in both processes. The defect in RAD51 assembly in atm−/− mice during meiosis (Barlow et al., 1998) is consistent with the notion that ATM plays a role in the processing of programmed meiotic DSBs. Arabidopsis mutants defective in XRCC3, a RAD51 paralog that is probably involved in the processing, but not in the initiation, of meiotic recombination events, display a meiotic phenotype identical to that of atm, and identical in its suppression by spo11 (Bleuyard and White, 2004).

A possible role for ATR in meiosis?

In partially synapsed chromosomes of mouse spermatocytes, ATR foci are associated with non-synapsed regions, whereas ATM clusters along synapsed regions (Keegan et al., 1996). However, the function of ATR in meiosis in higher eukaryotes is unclear, as atr mutants are lethal in animals (in yeast, the functions of ATR and ATM are to some degree fused into the single MEC1 protein, and this protein will be discussed later). As we reported earlier, the Arabidopsis ATR knock-out is not only viable, but is apparently fully fertile. Our analysis of meiotic stages from atr plants does suggest that there is a slight, possibly significant, increase in chromosomal fragmentation in atr, and that this is not suppressed by spo11. This fragmentation may result from the accumulation of undetected/unrepaired DNA damage in the atr line. Although it is not clear whether this fragmentation is meiosis specific, it does suggest that ATR plays some role in protecting genetic integrity in meiotic, as well as mitotic, cells. Of course, it is always possible that ATR plays a regular and important role in the process of meiosis, but that its failure to perform this role can be compensated for by another protein (such as ATM) that adequately copes with the consequences of this failure. This scenario is suggested by the extreme phenotype of the double mutant, described below.

Plants lacking both ATM and ATR display complete infertility and a strikingly defective meiotic phenotype

The Arabidopsis genome is distributed on only five chromosomes. Thus, in the complete failure of meiosis, where chromosomes distribute randomly, the odds of getting a complete chromosomal complement in the megaspore mother cell (with or without ‘extra’ chromosomes) is about 1 in 32. Given the fact that only pollen with a complete (or more than complete) genetic complement in the pollen-tube nucleus can germinate and grow to fertilize the megagametophyte, and that there is a great excess of pollen produced, we should expect to find two or three seeds per silique in the complete absence of organized meiotic segregation. This is approximately what we and others (Figure 3; Grelon et al., 2001) find in the spo11-1-1 mutant. These seeds suffer from poor germination (Table 1); however, this suggests that, although at least one copy of every chromosome must be present, there is significant aneuploidy.

In order to reduce fertility to zero, Arabidopsis must not only fail to simply segregate chromosomes properly, but must also fail to produce functional chromosomes. This appears to be the case for the atr atm double mutant.

Using atr and atm Arabidopsis mutants, we show here that ATR plays a partially overlapping role with ATM in the suppression of chromosomal fragmentation during meiosis. Although the atr single mutant does not have an obvious defect in meiosis, chromosomal fragmentation is significantly enhanced in the atr atm double mutant, in comparison with the atm single mutant. Interestingly, a defect in synapsis was observed in the double mutant. This defect was not complete: there appeared to be some pairing, but synapsis was never complete in any nucleus observed, and we do not know whether some or all pairing events were ectopic. This suggests that ATM and ATR might play a mechanistic role in early events in recombination, leading to a defect in homolog recognition. We emphasize that in both single mutants synapsis and the formation of bivalents was normal. Thus, this double mutant reveals a novel and essential role in synapsis that is performed redundantly by ATM and ATR.

As a consequence, perhaps, of ectopic annealing of non-homologous chromosomes, segregation at meiosis I was a complete failure. Although centromeres appeared to be pulled to the poles (Figures 1s,t and 2), the bulk of the chromatin remained at the metaphase plate in an entangled mass. The introduction of the spo11-1-1 mutation into the double mutant line corrected this aspect of the phenotype. In the triple mutant, chromosomes failed to interact and, although some fragmentation was observed, the univalent chromosomes were able to segregate at random. This indicates that the ectopic chromosomal interactions observed required the presence of SPO11-catalyzed breaks, and may have resulted from tentative and imprecise interactions between DNA sequences that, in the presence of ATM and/or ATR, would normally be corrected. However, other plausible hypotheses also fit these observations.

A role for the PI3K-like protein kinases in the homology search is supported by a comparison of the phenotype of the double mutant with that of other Arabidopsis mutants defective in end processing or single-strand introgression, including rad51, rad51c and mre11 (Li et al., 2004, 2005; Puizina et al., 2004). Each of these display a failure of synapsis (like the atr atm double mutant, but not the single mutants), chromosomal fragmentation and the (less severe) formation of anaphase bridges, as well as a correction of the fragmentation and bridge formation by spo11-1-1. In this study, we directly compare the phenotypes of the recombination-deficient mutant rad51 with atr atm, and show that there are major differences between the extents of ectopic chromosomal interactions, which are quite severe in the PI3K-like protein kinase double mutant. Similarly, S. cerevisiae mutants defective in MEC1, a PI3K-like protein kinase, combines many of the functions of ATM and ATR, display extensive meiotic ectopic recombination and ‘clumping’ of meiotic chromosomes into a single protein-dense mass, as do mutants defective in the checkpoint genes RAD17 and RAD24 (Grushcow et al., 1999). Thus, in yeast and plants, PI3K-like protein kinases act to prevent indiscriminant interactions between meiotic chromosomes. Although the effects of atr deficiency on mammalian meiosis have not been tested (because of the immediate lethality of the defect), ATR has been observed to coat non-synapsed regions of chromosomes (Perera et al., 2004). It also coats the sex chromosomes, in which programmed breaks occur but do not lead to homologous interactions. These observations are consistent with the notion that ATR, in mammals as well as plants, prevents sustained ectopic interactions between meiotic chromosomes. The suppression of these ectopic interactions by spo11-1-1 indicates that they require the induction of DSBs for their formation. It would be interesting to determine whether these interactions are RAD51 mediated, and perhaps represent strand-introgression events that require PI3K-like kinase activity for their reversal.

Activation of ATM is normally associated with activation of the MRE11/RAD50/NBS1 (M/R/N) complex, in which NBS1 is required for the signal transduction activity of the complex (D’Amours and Jackson, 2002; Falck et al., 2005). The plant homolog of NBS1 has recently been identified (Akutsu et al., 2007), and its biological significance, including a possible role in meiosis, has been investigated (Waterworth et al., 2007). Waterworth et al. established that the Arabidopsis NBS1 homolog is essential for cross-link repair (demonstrating its functionality), but is not required for meiosis. Interestingly, the atm nbs1 double mutant, like atm atr, reduced the fertility of the atm mutant to zero. The double mutant displayed a defect in synapsis, extensive fragmentation and clumping of the chromosomes: the same phenotype we observed here in atm atr. These data suggest that, in meiosis in plants, NBS1 is required for ATR function. Importantly, Waterworth et al.’s data indicated that NBS1 acts in the absence of ATM. This is quite a novel result, as NBS1 has heretofore been exclusively associated with ATM-dependent signaling. In certain contexts, ATM has been shown to activate ATR, and this activation requires NBS1 (Jazayeri et al., 2006), but in the case of plant meiosis, NBS1 and ATR may be interacting in the absence of mediation by ATM.

In summary, ATR and ATM play central roles in the response to DNA damage in mitotic cells, phosphorylating (in many cases) overlapping targets to initiate cell-cycle arrest, DNA repair, transcriptional regulation and, in animals, apoptosis. Although it is unclear whether ATR and ATM play analogous roles in the processing of DSBs during meiosis, our data indicate that both ATR and ATM can, under some circumstances, participate in the processing of SPO11-dependent breaks, and that the proteins act redundantly to prevent SPO11-dependent persistent ectopic interactions. Although ATR may simply ‘fill in’ for ATM in its absence, it is also possible that ATR plays either a truly redundant role with ATM, or a more specialized role in the processing of a particular and relatively rare subclass of breaks, as is suggested by the distinct contributions of the two kinases to mitotic γ-H2AX focus formation (Friesner et al., 2005).

Experimental procedures

Arabidopsis growth conditions

Seeds were surface sterilized with 10% bleach and were then sown on 1x MS (GIBCO, phytagel agar (Sigma-Aldrich, plates. After 7–10 days of growth on plates, seedlings were transferred to soil and were further grown under cool-white lamps filtered through Mylar film at an intensity of 100–150 μmol m−2 sec−1, with a 24-h light cycle at 21°C.

Genotyping of Arabidopsis mutant lines

For this study, we employed the atr-3 allele (Culligan et al., 2004), the atm-1 allele (Garcia et al., 2003) and the spo11-1-1 allele (Grelon et al., 2001), which are all in the Ws background. To generate the WT, single and multiple mutant combination lines used in this study, we performed a cross of SPO11-1-1/spo11-1-1 with atr-3/atr-3 ATM-1/atm-1. The resulting F1 seeds were sown on soil and were genotyped (see below) to identify an SPO11-1-1/spo11-1-1 ATR-3/atr-3 ATM-1/atm-1 individual. We further genotyped the segregating F2 population for the homozygous WT and mutant lines used in this study: procedures for the atr-3 and atm-1 have been described previously (Culligan et al., 2006). To identify the spo11-1-1 allele in our various mutant-combination lines, we employed PCR amplification to identify WT alleles using primers SPO11-1 (5′-GGATCGGGCCTAAAAGCCAACG-3′) and SPO11-2 (5′-CTTTGAATGCTGATGGATGCATG-3′). To amplify the T-DNA mutant allele, we used primers SPO11-1 (as above) and the T-DNA-specific left-border primer SPO11-LB (5′-CGTGTGCCAGGTGCCCACGGAATAG-3′).

Visualization of meiotic chromosomes

Meiotic chromosome spreads were prepared essentially as described by Ross et al. (1996). The resulting slides were stained with DAPI at a concentration of 1 mm in Vectashield (Vector Laboratories, Chromosomes were visualized using a Nikon Eclipse E600 epifluorescence microscope ( equipped with a mercury lamp. Images were viewed using ImagePro Plus software (Media Cybernetics, Inc.,, and were captured using a Hamamatsudigital camera (C4742-95; Hamamatsu Photonics,, equipped with a UniBlitz shutter driver (model D122).


This manuscript is dedicated in loving memory to Carol Ann Bradshaw Culligan. We thank Dr Neil Jackson for assistance with the preparation of meiotic chromosome spreads. This work is supported by a DOE, Office of Basic Energy Sciences Grant #DE-FG02-05ER15668 to ABB and a United States Department of Agriculture (USDA), HATCH grant (NH00488) to KMC. This is scientific contribution number 2358 from the New Hampshire Agricultural Experiment Station.