Common fragile sites (CFSs) are chromosomal regions that are prone to form breaks or gaps during mitosis, in particular following replication stress. The mechanisms modulating CFS expression and promoting safe chromatid transmission to daughter cells are not clear. Now CFS expression is shown to reflect the activity of the MUS81–EME1 resolvase complex which cooperates with the dissolving action of the BLM helicase to prevent uncontrolled chromosome breakage and to promote genome integrity.
Cancer cells are characterized by chromosomal abnormalities. For over 40 years we have known that some regions in mammalian chromosomes are prone to breakage. Fragile sites are extended genomic regions that are susceptible to gaps and/or breaks detectable in metaphase chromosomes. They have been described in several mammalian genomes, are present on all human chromosomes, and have been linked to human pathologies, mainly neurological diseases and cancer1.
Two papers in this issue, by Ying et al.2 and Naim et al.3, provide evidence for the existence of an active mechanism responsible for the controlled production of DNA breaks at these chromosomal locations, and indicate that this process promotes, rather than compromises, the maintenance of genome integrity.
Fragile sites are generally stable in cultured cells and express their intrinsic fragility following exposure to replication stress, notably by exposure to low levels of the DNA polymerase inhibitor aphidicholin. The so-called common fragile sites (CFSs) are present in all individuals, whereas rare fragile sites are only found in less than 5% of the population. The identification of a number of CFSs has allowed the establishment of their involvement in sister chromatid exchanges, deletions and translocations observed in cancer cells, and their preferential targeting by viral insertions1. The mechanisms responsible for CFS expression and for the safe transmission of these chromosomal regions to the daughter cells have been the focus of several studies in the past 10 years or so. Much evidence has been gathered that links CFSs to the presence of an unreplicated region of DNA, and genome-wide analyses of CFSs has revealed an enrichment for highly flexible A/T-rich sequences, which may form unusual secondary structures4. The working model is that CFSs are regions that replicate late and represent a problem for DNA polymerases. Replication fork stalling at CFSs may be caused by a specific DNA sequence, a particular secondary structure or the reduced number of activated replication origins in the area5,6,7. More recently, expression of CFSs has been related to collisions between the transcription and replication machineries8. The Werner syndrome protein (WRN) is important for resolving the above-mentioned situations, promoting replication fork restart and suppressing CFS expression9. If the incompletely replicated tracts escape detection by the ATR-dependent checkpoint, they are allowed into mitosis and accumulate breaks and gaps. Indeed, inhibition of the ATR-mediated response results in an increase in CFS expression10. Attempts to segregate replication intermediates lead to sister chromatid entanglement and non-disjunction, resulting in the presence of ultra-fine bridges (UFBs) in anaphase cells. Although most UFBs originate from centromeric regions, CFS-specific UFBs have been described, characterized by the Fanconi anemia proteins FANCD2/FANCI binding to the extremities of the bridge, and by BLM (Bloom syndrome helicase) and PICH (Plk1-interacting checkpoint helicase) binding along the bridge itself11. The BLM helicase was shown to be responsible for dissolving these structures and allowing sister chromatid segregation. Dysfunctional BLM, as seen in cells from Bloom's syndrome patients, results in increased numbers of UFBs, elevated levels of sister chromatid exchanges, and chromosome abnormalities, indicating that BLM is required to untangle incompletely replicated sister chromatids and promote completion of anaphase.
Both the papers by Ying et al.2 and by Naim et al.3 report a mechanism in which the MUS81–EME1 nuclease cleaves unreplicated intermediates at CFSs, physically separating the sister chromatids and resolving the non-disjunction problems. These findings change the current view, according to which CFSs break in an attempt to segregate incompletely replicated chromosomes, causing genome instability. The new data show that formation of breaks at CFSs is instead an active mechanism based on MUS81–EME1 that processes replication intermediates during early mitosis, and suggest that this controlled action of MUS81–EME1 actually supports maintenance of genome integrity. The two papers show that depletion of the DNA-repair protein ERCC1 or the MUS81–EME1 complex results in a reduction in CFS expression. This may be due to a positive effect in replicating the CFS region leading to a reduction in chromosome breakage. Molecular combing experiments reveal that normal replication fork dynamics is not modified in cells lacking MUS81–EME1, but stalled replication fork restart is defective. Moreover, depletion of MUS81 leads to an increase in UFBs at CFSs, indicative of a higher level of incompletely replicated regions, and suggesting that this complex actively removes replication intermediates accumulating at CFSs. Naim et al.3 show that MUS81 cleaves at CFSs in late G2 mitosis, and that the cleavage products can be detected in mitosis and in the ensuing G1. Consistently, at the G2-M checkpoint, CFSs are still undergoing DNA synthesis and recruit MUS81, which is frequently found sandwiched between two FANCD2 sister foci. It is important to note that during S phase, MUS81–EME1 activity is controlled by the ATR–CHK1 kinase signalling pathway12, and is activated at G2-M, possibly by cyclin-dependent kinase (CDK) and polo-like kinase 1 (PLK1)13. Intermediates that have not been resolved by MUS81 form UFBs during anaphase, when BLM, topoisomerase IIIa (TOPIIIa) and the BLM-associated proteins RMI1/2 can dissolve the structure and allow proper chromatid segregation. If the intermediate persists, it will break between anaphase and telophase, giving rise to a checkpoint signal detectable as histone γH2AX foci. Intriguingly, these breaks are transmitted to the daughter cells, which, in the ensuing G1 phase, exhibit the accumulation of 53BP1 (a DNA damage response protein) nuclear bodies containing CFS sequences14,15. The abundance of these structures increases following depletion of MUS81 or replication stress. Interestingly, 53BP1 nuclear bodies are assumed to protect the damaged DNA, the repair of which will be attempted in the following S phase, consistent with the function of 53BP1 in preventing processing of broken chromosomes16 (Fig. 1). Depletion of MUS81 or BLM induce higher levels of 53BP1 nuclear bodies — but, whereas in cells depleted of MUS81 the abundance of UFBs at CFS increases, depletion of BLM does not produce a corresponding increase in CFS expression. This suggests that MUS81 and BLM do not compete for the same substrate, but rather MUS81 initially attempts to cleave the replication intermediate and, if it fails, BLM will then have a chance, possibly acting as a back-up mechanism. Interestingly, downregulation of the condensin subunit SMC2 causes decreased formation of 53BP1 nuclear bodies, but increased levels of UFBs, suggesting that chromosome condensation promotes the formation of chromosomal lesions at UFBs if BLM does not intervene efficiently enough.
A few points remain to be understood and may indicate future developments in this field. First, structures found at replication intermediates can be processed by different enzymes (for example, MUS81–EME1, GEN1 and SLX4–SLX1). The present studies describe the role of MUS81–EME1, but a possible involvement of GEN1 and SLX4, possibly at specific CFSs, should be investigated. Second, the work of Naim et al.3 reports that ERCC1 plays a role analogous to that of MUS81. Given that ERCC1 works in a complex with XPF, the fact that depletion of only ERCC1, but not XPF, causes a decrease in CFS expression is unexpected. The authors suggest that ERCC1 could work independently of XPF in this context — but the possibility that, in the absence of ERCC1, XPF may actually act to limit the activity of MUS81–EME1 cannot be ruled out, particularly since SLX4 activates MUS81 but binds also XPF (ref. 17). Third, the fate of the controlled lesions that may be generated by MUS81–EME1 or BLM is not yet clear. Since the products of MUS81are transmitted through mitosis, it will be relevant to understand how cells prevent rearrangement or loss of genetic information, and what mechanisms are employed to fix the lesions. Finally, a full investigation of the actual nature and physiological role of the 53BP1 nuclear bodies will reveal how cells tolerate unrepaired lesions and may possibly uncover a new mechanism for adaptation to DNA damage.
About this article
Are common fragile sites merely structural domains or highly organized “functional” units susceptible to oncogenic stress?
Cellular and Molecular Life Sciences (2014)