Our study reveals several novel observations. First, we uncover a DSB-independent function of the MRX complex for DNA integrity at protein–DNA barriers in the rDNA, where absence of the MRX complex leads to increased levels of ssDNA. Second, we find that absence of Mre11 does not trigger a checkpoint response from within the rDNA, whereas its absence leads to checkpoint activation when ectopically placed RFBs are present. Third, relieving rDNA silencing is sufficient to provoke checkpoint activation in the rDNA to the same extent as seen for the ectopic site, strongly suggesting that checkpoint activation is governed by chromatin context at least in the rDNA.
In our Fob-block system, all events caused by higher levels of stalled replication forks in the rDNA are unproblematic for wild-type cells but become detrimental when cells lack MRX. This is in agreement with genetic data reporting that absence of Rrm3, a helicase required for helping replication forks traverse protein–DNA barriers, leads to synthetic lethality with mre11Δ
). Although overexpression of Fob1 likely affects the level of rDNA, silencing deletion of SIR2
does not suppress the growth defect observed in the absence of the MRX complex. This supports the idea that the MRX complex is required at the rDNA owing to replication fork stalling and not owing to an affected rDNA silencing on overexpression of Fob1.
We provide evidence for a DSB independent role of MRX at protein–DNA barriers based on three facts. First, RAD52
is not required for proper cell growth in our Fob-block system, which argues against DSB formation. Second, the growth defect observed for GAL-FOB1 mre11Δ
cells is not suppressed by a YKU70
deletion. It has previously been reported that suppression of mre11Δ
phenotypes by a YKU70
deletion is restricted to events at DSB ends (37
). Third, the endo- and exonuclease activities of Mre11p are not required.
At present, we do not know the exact mechanism by which MRX protects forks at RFBs; however, several modes of action can be envisioned, which are not mutually exclusive. MRX may preserve the conformation of newly synthesized DNA behind forks at RFBs, thus preventing replisome collapse. This function would be analogous to what has been suggested for MRX at HU-stalled forks and is supported by structural data showing that the MRX complex can bridge two DNA duplexes held in the same distance as newly synthesized sister chromatids (39
). Alternatively, the MRX complex may prevent fork reversal at RFBs (A). At so-called terminal forks, absence of the Mre11 has been suggested to lead to fork reversal, where the generated structure is prone to cleavage (46
). If fork reversal occurs at RFBs in the absence of Mre11, it does not seem to be prone to cleavage, as we fail to detect higher levels of DSBs at the RFBs in the rDNA. However, our PFGE analysis supports the idea that absence of the MRX complex could lead to more branched structures in the rDNA, as a large portion of Chr. XII is retained in the wells. The retention of DNA in the wells in the DSB assays more likely stems from an inhibition of restriction enzyme digestion owing to the presence of ssDNA, as branched structures will run into this type of gel. Furthermore, we detect more RPA in the rDNA at late time points (after bulk DNA synthesis) in the absence of Mre11. This indicates that processing occurs in these cells generating more ssDNA compared with wild-type cells. This can be explained if more reversed forks are generated in the absence of MRX, which are processed to give ssDNA. Indeed, regions of ssDNA are frequently detected on the regressed arm of a reversed fork (47
). It has been suggested that reversed forks are formed in yeast cells at RFBs in the rDNA (48
), and there are accumulating evidence that stalled replication forks are very prone to fork reversal (47
). If fork reversal is a frequent incidence at RFBs even in wild-type cells, it is attractive to think of the MRX complex as a ‘protector’ of the structure (A). On fork reversal, a double-stranded end is exposed, which can be recognized by the MRX complex. Binding of the MRX complex to this end could potentially protect the structure from further processing. Our ChIP data would also be consistent with this hypothesis.
Figure 9. (A) Suggested functions of the MRX complex on replication fork stalling at a protein–DNA barrier. The MRX complex may hinder fork reversal (left) or protect a reversed fork from further processing (right). (B) Model for the cellular consequences (more ...)
The observed growth defect is identical for GAL-FOB1 mre11Δ and GAL-FOB1 eRFB mre11Δ strains. Thus, the ectopically RFBs do not contribute additionally to a growth defect. It is tempting to believe that the growth defect stems from problems arising directly at the RFBs, but we cannot rule out that it originates owing to other problems in the rDNA. Replication in the rDNA is not only challenged by the unidirectional mode of replication but also owing to its repetitive nature, unusual secondary structures that may affect DNA replication (e.g. creating more stalling) are also more likely to be generated in this region. Together, this would create a stronger need for the MRX complex in this region either to suppress fork reversal or protect a reversed fork. Thus, we cannot exclude that the rightward-moving fork in the rDNA encounters problems in the absence of the MRX complex, although this is not the case for the rightward-moving fork in the ARS606–ARS607 region ().
The cellular implications in the rDNA on FOB1 induction and in the absence of Mre11 are to our surprise checkpoint blind (B), which is in striking contrast to the checkpoint activation arising when forks stall at ectopically placed barriers in the absence of Mre11 (B). We believe that it is a structure arising at the ectopic barrier, which is checkpoint activating, as we fail to detect unreplicated DNA in this region, supporting that a fully competent fork emanates from ARS606. Opposed to the rDNA and to our surprise, we fail to detect more RPA at this location; however, this is probably due to limitations in our ChIP experiments.
Why is it that abnormal structures at ectopically placed RFBs are checkpoint activating but checkpoint blind in the rDNA? It has been shown that a DSB induced in the rDNA by the endonuclease I-Sce
I is checkpoint activating; thus, the unique heterochromatic structure found in the rDNA is not enough to suppress a checkpoint response under normal circumstances (52
). Furthermore, we can rule out a general function of Mre11 for checkpoint activation in the rDNA, as we can detect checkpoint activation when a DSB is induced by the endonuclease I-Sce
I in the absence of Mrell (Supplementary Figure S4
). However, we considered the idea that overexpression of Fob1 in our Fob-block system could lead to a significant higher level of heterochromatic structure in the rDNA, which may impact checkpoint activation. Indeed, when SIR2
is deleted, we are able to detect Rad53 activation to the same level as seen in strains with ectopically placed RFBs. Thus, disturbance of a balanced rDNA silencing may adversely affect the checkpoint response. This effect may be direct in that the heterochromatic structure hinders access of checkpoint sensors and thereby suppress the checkpoint response. However, it is also easy to imagine that unbalanced rDNA silencing encumbers proper processing of the DNA at the RFBs into a strong checkpoint activating structure. Another attractive possibility is that SIR2
in general suppresses checkpoint in the vicinity of RFBs.
Our finding that a SIR2
deletion restores checkpoint activation points to a significant influence of chromatin structure on checkpoint activation in the rDNA. In line with this, it has been reported that heterochromatin could pose a barrier to the DNA damage response pathway (53–57
), and more recent, it was furthermore shown that heterochromatin induced by oncogenic stress restrains DNA damage response (58
It is reasonable to picture that there is a delicate balance in the rDNA to know whether a checkpoint response is activated. In the Rdna, both replication dependent and independent double strand breaks occur in each cell cycle, which are not checkpoint activating (41
), whereas an I-Sce
I generated DSB causes checkpoint activation [Supplementary Figure S3
)]. Thus, in the rDNA, there may be an inherent way to distinguish between natural or aberrant damage. Alternatively, as RFBs are a natural integrated part of the rDNA, and a hotspot for DSBs, it is attractive to suggest that Sir2 in general suppresses checkpoint activation in the vicinity of the RFBs. Future studies will hopefully uncover the underlying mechanism controlling this phenomenon and unravel whether this is evolutionarily conserved.