From resection to resolution: biochemical investigation of early and late steps of homologous recombination

Abstract

Cells utilize two major pathways to repair DNA double strand breaks (DSB) including homologous recombination (HR) and non-homologous end joining (NHEJ). DNA end resection of the 5'-terminated DNA at DSBs is the key event, which mechanistically determines the repair pathway choice by promoting HR while inhibiting NHEJ. Numerous studies have established the role of the MRE11-RAD50-NBS1 (MRN) complex and CtIP in the initiation of resection. However, the nucleolytic polarity of the only nuclease in the complex, MRE11, is the opposite (3' to 5') to the direction of resection (5' to 3'). To solve this enigma, a so-called bidirectional resection model has been proposed. MRE11 was anticipated to make incision(s) close to the DSB by its endonucleolytic activity, which creates an entry site for its 3' to 5' exonuclease that proceeds back towards the DSB. Although this model solves the polarity paradox, the mechanistic understanding of such nicking activity by MRE11 was undefined. Here I show that CtIP stimulates the endonuclease activity of MRN on double-stranded DNA (dsDNA). Phosphorylation of CtIP (pCtIP) at T847 is absolutely required for this stimulatory effect. The blocking of DNA ends, especially at the 5' end, is required to observe this MRN-pCtIP activity. In agreement with the proposed model, the position of the first incision by MRN-pCtIP complex maps to [symbol about] 20 nucleotides away from the 5' DNA end. The nicking activity is intrinsic to MRE11, as nuclease-dead (H129L D130V) variant of MRE11 does not show any clipping activity. Additionally, RAD50 mutants deficient in ATP binding (RAD50K42A) or ATP hydrolysis (RAD50K42R), fail to show any clipping activity, indicating the essential role of RAD50 and its ATPase activity. Interestingly, the removal of NBS1 from MRN complex results in the loss of the clipping activity. Therefore, unlike in yeast, my work establishes NBS1 as the indispensable component of the MRN complex together with RAD50 to stimulate MRE11-catalyzed clipping activity in humans. CtIP tetramerizes by making a dimer of dimers through its amino-terminus and such oligomerization is important for its in vivo functions in HR. The analysis of pCtIP L27E mutant, in which tetramerization is abolished while dimerization is preserved, revealed a reduced capacity to promote MRN, suggesting that proper oligomeric structure of CtIP is likely essential for its optimal activity. Furthermore, the deletion of 160 amino acids from the amino-terminus of CtIP (1-160∆ pCtIP), which disrupts the CtIP dimerization and consequently its tetramerization as well, reduces the pCtIP capacity to stimulate MRN drastically. In contrast to yeast, pCtIP also stimulates the MRN endonuclease on ssDNA circular plasmid. In terms of cofactors, Mg2+, Mn2+ and ATP were found to be important for the optimum activity of of MRN-pCtIP. My second project was focused on meiotic homologous recombination. Meiotic HR favours the formation of double Holliday junctions (dHJ) as intermediates of programmed DSB repair. These are, by a yet unknown mechanism, processed exclusively to preferentially to crossovers (CO to facilitate the production of genetic diversity. Meiotic cells have a dedicated and biased mechanism in place to ensure the formation of obligate COs. MLH1-MLH3 (MutLγ) has been strongly implicated in meiosis as a putative endonuclease responsible for cleaving dHJs in a biased manner to produces COs. Additionally, MSH4-MSH5 (MutSγ) has also been shown to function in the MutLγ mediated pathway. Despite the availability of extensive genetic data, the mechanistic understanding of the biased cleavage of dHJs by MutLγ and its partners remains elusive. Here, in collaboration with Nicolas Weyland, I set out to study and characterize the biochemical behaviour of human MLH1-MLH3 in conjunction with human MSH4-MSH5. Previously, we showed that hMutLγ prefers binding to HJs and similar structures. Further analysis indicated that it presumably binds to the core of a HJ. Here I could show that hMutLγ nicks super-coiled dsDNA non-specifically in the presence of Mn2+. No such activity with nuclease-dead hMutLγ (MLH1-MLH3D1223N) was observed, proving that the observed activity is intrinsic to hMutLγ. The presence of ATP further simulates the nicking activity of hMutLγ. Using purified hMutSγ, I could confirm that it prefers binding to HJs over dsDNA and slides upon HJ arms upon ATP binding. We could also show that hMutLγ directly interacts with hMutSγ in vitro. Furthermore, DNA binding analysis of hMutLγ and hMutSγ revealed that hMutLγ binds cooperatively to HJ with hMutSγ. No such observation with dsDNA emphasizes the specific nature of the observed effect with HJ. The data also indicates that hMutSγ further stabilizes the hMutLγ-DNA complex.