In this paper we shed light on the mechanism by which Filamin A (FLNA) is required for efficient DNA repair. Our data indicates that lack of FLNA impacts on HR and NHEJ. FLNA is an actin-binding protein and its inactivation leads to an array of disorders such as otopalatodigital spectrum disorder, Melnick-Needles syndrome and periventricular heterotopia.26
Although of unclear significance, at least two families carrying germline mutations in BRCA1
have been shown to manifest ventricular heterotopia.33,34
FLNA interacts with a variety of proteins, including BRCA2,25
and deficiency in FLNA leads to sensitivity to DNA damage and a defect in the recovery from G2
Thus, we investigated further its role in the DNA damage response.
FLNA binds BRCA1 using its extreme C-terminus which contains its dimerization domain. BRCA1 interaction with FLNA is mediated by a 30 amino acid region in the N-terminus of BRCA1 which contains a conserved domain called Motif 2.21
Introduction of the Y179C mutation in Motif 2 significantly decreases the interaction. Analyses by the Align GV-GD method or by a yeast-based recombination assay suggest that Y179C may act as a deleterious mutation.35,36
On the other hand, this variant has been found co-occurring in trans with a known deleterious mutation, which indicates that it is unlikely to have severe effects.37
Thus, the Y179C may constitute a hypomorphic mutation with moderate effects on breast cancer predisposition. Of note, Motif 2 is close to the region that has been implicated in binding of BRCA1 to Ku86.38
In order to dissect the molecular role of FLNA in the DDR we took advantage of a well-characterized genetically-defined system. A melanoma cell line lacking FLNA was isolated and subsequently reconstituted with FLNA yielding a pair of cell lines in which the only difference is the presence or absence of FLNA.29
When we irradiated FLNA-
cells, we noticed that FLNA-
took much longer to resolve DSBs (). To elucidate the mechanism underlying the repair defect we systematically investigated the proficiency of damage signaling in FLNA-
Initially we investigated the recruitment and activation kinetics of the upstream kinases, as well as their downstream substrates after DNA damage. We found that FLNA deficiency led to the hyperactivation of ATM as judged by phosphorylation of ATM S1981 and CHK2 T68, surrogate markers of ATM activation.39-41
Similarly, lack of FLNA also led to a hyperactivation of ATR, as measured by CHK1 S317 phosphorylation, a marker for ATR activation.42
Moreover, we also found sustained levels of phosphorylation of NBS1 S343 to be higher in FLNA-
cells. Although the role of NBS1 phosphorylation in the DNA damage signaling is poorly understood, it is generally thought to reflect ATM and ATR activation.43,44
We also determined that major mediator proteins BRCA1, MDC1 and 53BP1 formed IR-induced foci irrespective of FLNA status. However, BRCA1 foci formation was significantly impaired in FLNA-deficient cells. In addition, Rad51 foci formation displayed a delayed kinetics in cells lacking FLNA. These data indicate that FLNA-deficient cells have impaired homologous recombination. Indeed, during the preparation of this manuscript Yue et al. showed that FLNA-deficient cells have a reduced ability to repair I-SceI-induced DSBs.45
During the course of our experiments we noticed a consistent increase in the number of FLNA-
cells displaying IR-induced RPA foci. These foci progressively increased in size at later time points after IR. RPA is a ssDNA binding protein and participates in DNA metabolism processes where there is generation of ssDNA such as replication, repair and recombination.46,47
Phosphorylation leads to inability of RPA to associate with the replication centers and leads to the association with DNA damage-induced foci instead.48
Interestingly, lack of NHEJ proteins DNA-PKcs and Ku86, which together with Ku70 form the active DNA-PK complex, leads to accumulation of ssDNA in S phase.49
Thus, we further investigated how the lack of FLNA impacted on DNA-PK complex formation.
Remarkably, Ku86 failed to interact with DNA-PKcs in the absence of FLNA. The reduced stability of the interaction is not due to Ku86 failure to load onto chromatin, as FLNA-
cells displayed sustained higher levels of chromatin-bound Ku86 than FLNA+
cells after damage. Ku86 is one of the first molecules to bind DNA ends after DSBs50
and recruits DNA-PKcs via its C-terminus.3
Taken together these results establish that lack of FLNA results in an unstable association of Ku86 and DNA-PKcs impairing the function of the complex. This impaired DNA-PK activity leads to a continuous build up of ssDNA and Ku86 on chromatin.
Over 16 phosphorylation sites have been identified in DNA-PKcs although their role is still poorly understood. Nevertheless, DNA-PKcs phosphorylation status is thought to influence its activity.51
DNA-PKcs interacts with Ku86 and free ends of DNA in an unphosphorylated form,52
and autophosphorylation is required for NHEJ progression.53
Thus, we investigated the status of the two major phosphorylation clusters in DNA-PKcs, namely the 2056 and 2609 clusters. Clusters 2056 and 2609 were consistently phosphorylated upon treatment with IR irrespective of FLNA status. The fact that DNA-PKcs is phosphorylated upon damage in the absence of FLNA suggests that DNA-PKcs is interacting with the Ku86/DNA complex albeit transiently. Alternatively, it is possible that phosphorylation of DNA-PKcs is not mediated by autophosphorylation at the synaptic complex but rather via hyperactive ATM and ATR in FLNA-deficient cells.
We showed that FLNA and BRCA1 interact and that FLNA deficiency leads to a marked decrease in BRCA1 foci formation after damage. To investigate further the role of BRCA1 we tested whether expression of the BRCA1 FLNA-interacting fragment in FLNA-proficient cells could also act in a dominant negative fashion leading to a phenotype similar to FLNA-deficient cells. Strikingly, expression of the BRCA1-Ff lead to a defect in DNA repair as judged by CDC2 pY15 and γ-H2AX markers. This effect is specific because expression of BRCA1-Ff containing a mutation that disrupts FLNA/BRCA1 interaction does not lead to the same phenotype. Taken together, these data establish that BRCA1 participates in the FLNA-dependent regulation of the DNA damage response.
Our data shows that absence of FLNA leads to defective DSB repair. The defect is a combined result of compromised HR and NHEJ processes. At this stage we cannot distinguish whether FLNA-deficiency leads to a defective step that is common to both pathways or, alternatively, it impacts different steps in these pathways. In fact, the interplay between these two arms of the DNA repair process is not fully understood,54
in particular after IR, which generates an array of different DNA modifications. The observed phenotype is consistent with a model in which Ku86 recognizes and binds free ends of DNA, but in the absence of FLNA, fails to make a stable complex with DNA-PKcs. We propose that unstable Ku86/DNA-PKcs interaction results in impaired end processing, accumulation of ssDNA, and hyperactivation of DNA damage signaling.
In addition, in FLNA-deficient cells BRCA1 displays impaired foci formation suggesting that FLNA also plays a role in stabilizing BRCA1 at the DSBs. BRCA1 colocalizes with Rad50/Mre11/NBS1 complex at IR-induced foci55,56
and inhibits Mre11 exonuclease activity.57
Thus, the diminished amounts of BRCA1 at IR-foci may lead to an unregulated Mre11 exonuclease activity with formation of the observed extended tracts of RPA-coated ssDNA in FLNA-deficient cells (). BRCA1 has also been implicated in the regulation of Rad51,7,58
although the mechanism by which it happens is obscure.59
The kinetics of Rad51 foci formation in FLNA-deficient cells suggests that there is no problem in the initial recruitment to foci (see , bottom, 3 h time point). The extended plateau observed in Rad51 foci (from 3 to 12 h after IR) may indicate an accumulation of DSBs that do not fulfill the end processing requirements for efficient Rad51 loading. Although further research will be needed to test this proposed model, it provides a tractable system to dissect the interplay between different processes involved in DNA repair.
It is possible that FLNA provides a framework for the assembly of factors in the synaptic complex. While unrepaired DNA in yeast (which lacks recognizable DNA-PKcs and FLNA orthologs) migrates to so-called DNA repair centers,60
the picture is different in mammalian cells where broken chromosome ends are essentially immobile.61,62
It will be interesting to determine whether lack of FLNA affects the mobility of broken ends.