Tid1/Rdh54 promotes dissociation of Dmc1 from nonrecombinogenic sites on meiotic chromatin

DNA strand exchange is the molecular process that defines homologous recombination (Roca and Cox 1990; Eggleston and Kowalczykowski 1991; Kowalczykowski 1991; Folta-Stogniew et al. 2004). Members of the RecA family of recombinases promote DNA strand exchange by first assembling helical filaments on single-strand DNA (ssDNA). These filaments carry out a homology search of duplex DNAs. Recognition of homology is associated with DNA strand exchange, the process through which nucleotides on the incoming ssDNA strand form base pairs with nucleotides of the complementary strand of the target duplex. Homologous recombination events involve formation and repair of DNA double-strand breaks (DSBs). In somatic cells, DSBs form spontaneously, while in meiotic cells, formation of these breaks is programmed (Paques and Haber 1999; Symington 2002). SPO11, a meiosis-specific transesterase, is required for formation of programmed meiotic breaks (Keeney et al. 1997). After breaks form, nucleases process DNA ends to yield 3′ ssDNA tails, which are the substrate for recombinase assembly into presynaptic filaments (Cox and Lehman 1982; Sung 1994; Sugawara et al. 2003; Wolner et al. 2003; Fukuda and Ohya 2006). DNA strand exchange then yields D-loops with a hybrid segment of DNA terminating with the 3′ end of the invading strand. Such ends serve as primers for the DNA repair synthesis needed to complete the recombination event (Paques and Haber 1999; Symington 2002). Eukaryotes encode two major RecA-like recombinases, Rad51 and Dmc1 (Aboussekhra et al. 1992; Bishop et al. 1992; Shinohara et al. 1992). Rad51 is required for recombinational repair of DSBs in somatic cells. Dmc1 is a meiosis-specific recombinase found in a variety of organisms including budding and fission yeast, plants, and mammals (Villeneuve and Hillers 2001). In budding yeast, both Dmc1 and Rad51 play important roles in meiotic DNA strand exchange. The two recombinases have partially redundant and distinct functions; efficient, properly regulated meiotic recombination requires both proteins (Rockmill et al. 1995; Schwacha and Kleckner 1997; Shinohara et al. 1997a). Rad51 and Dmc1 share several properties with bacterial RecA, including the abilities to form helical presynaptic filaments and to promote strand exchange in purified systems (Eggleston and Kowalczykowski 1991; Sehorn and Sung 2004). All RecA-like recombinases are DNA-dependent ATPases, and all require ATP as a cofactor during DNA strand exchange; however, hydrolysis of ATP is not essential for homology recognition and DNA strand exchange (Menetski et al. 1990; Sung and Stratton 1996; Campbell and Davis 1999a; Masson et al. 1999; Hong et al. 2001; Bugreev et al. 2005). The ability of recombinases to function without ATP hydrolysis led to two different explanations for the function of energy in RecA-mediated reactions. In one model, RecA helical filaments are rotary motors that use ATP hydrolysis to drive unidirectional branch migration to extend heteroduplexes even across regions of imperfect homology (Cox 2003; Cox et al. 2005). A second model is based largely on the finding that ATP hydrolysis converts a form of recombinase with high affinity for DNA (RecAATP) to a form of recombinase with low affinity (RecAADP). Thus, ATP hydrolysis promotes the dynamics of RecA–DNA interactions, allowing release of recombination products after strand exchange and/or making it possible to correct discontinuities that form during assembly of presynaptic helical filaments on ssDNA (Menetski et al. 1988; Kowalczykowski 1991). Although the core DNA strand exchange reaction is likely to proceed by a mechanism conserved since the divergence of prokaryotes and eukaryotes, certain biochemical properties are significantly different (Eggleston and Kowalczykowski 1991). Three biochemical differences are particularly relevant to the present study. First, the ATPase activity of the eukaryotic recombinases is ∼80- to 200-fold weaker than that of RecA (discussed further below). Second, the eukaryotic recombinases show relatively little binding preference for ssDNA relative to double-strand DNA (dsDNA) (Ogawa et al. 1993; Sung 1994; Li et al. 1997; Hong et al. 2001). Third, although all recombinases use factors that promote their assembly on tracts of ssDNA (Kowalczykowski 2000; Gasior et al. 2001; Symington 2002; Hayase et al. 2004), optimal activity of the eukaryotic recombinases also requires Rad54 and Tid1/Rdh54, proteins that do not appear to have counterparts in prokaryotes (Petukhova et al. 1998, 2000). Rad54 and Tid1 are members of the Swi2/Snf2 family of helicase-like proteins (Eisen et al. 1995; Klein 1997; Shinohara et al. 1997b). Members of this family are potent DNA-dependent ATPases that translocate along DNA in an ATP hydrolysis-dependent manner. Swi2/Snf2 and other members of the family have been shown to be capable of remodeling chromatin, thereby regulating gene expression at the transcriptional level (Eisen et al. 1995). Rad54 and Tid1 act in vivo to enhance the efficiency of Rad51- and Dmc1-dependent DSB repair in mitosis and meiosis (Klein 1997; Shinohara et al. 1997b). As mentioned above, Rad54 and Tid1 also stimulate DNA-strand exchange activity in purified systems (Petukhova et al. 1998, 1999, 2000). The mechanism underlying this stimulation is still the subject of study. Several activities have been described for Rad54 and Tid1 that may contribute to its ability to stimulate DNA strand exchange in vivo. Both Rad54 and Tid1 locally unwind naked duplex DNA in an ATP hydrolysis-dependent manner (Petukhova et al. 1999, 2000), an activity that may enhance the efficiency of the homology search. Like other members of the Swi2/Snf2 family, Rad54 can displace nucleosomes, an activity that could also contribute to homology searching in vivo (Alexeev et al. 2003; Jaskelioff et al. 2003). Finally, physical assays of recombination intermediates in living cells suggest that Rad54 plays a post-synaptic role during Rad51-mediated recombination (Sugawara et al. 2003). The ability of Rad54 to promote dissociation of Rad51 from duplex DNA in vitro could account for such a late role (Solinger et al. 2002; Wesoly et al. 2006). Dissociation of Rad51 from duplex requires the ATPase activity of Rad54, indicating that this activity involves Rad54 translocation on dsDNA. Recent single-molecule studies show directly that Rad54 is a DNA translocase (Amitani et al. 2006). The biochemical properties of Tid1 have not been as extensively characterized as those of Rad54; however, the partial redundancy of the proteins in vivo and their similar activities in DNA strand exchange and in DNA unwinding reactions indicate that the two could share additional biochemical properties (Klein 1997; Shinohara et al. 1997b, 2000). Tid1 and Rad54 have been shown to play a role in promoting the relative distribution of the two recombinases, Rad51 and Dmc1, during meiotic recombination. In wild-type cells, Rad51 and Dmc1 form DSB-dependent subnuclear foci (Bishop 1994). About 80% of Dmc1 foci localize at, or immediately adjacent to, Rad51 foci. The Swi2/Snf2 like recombination proteins control recombinase colocalization; Dmc1 and Rad51 colocalization is dramatically reduced in tid1Δ single mutants and eliminated in tid1Δ rad54Δ double mutants (Shinohara et al. 2000). These observations led to the proposal that Tid1 and Rad54 share an activity that promotes coordinated assembly of the recombinases at sites of DSBs. In the present study we demonstrate that Tid1 functions to prevent the accumulation of Dmc1 on chromatin in the absence of DSBs. This function depends on the ATPase activity of Tid1. Chromatin immunoprecipitation (ChIP) experiments indicate that Rad54 and Tid1 share an activity required for Dmc1 to load at sites of DSBs, and that in the absence of this shared activity, Dmc1 is sequestered on duplex chromatin. These results suggest that Tid1 and Rad54 indirectly coordinate assembly of the recombinases at DSBs by promoting dissociation of recombinase from non-DSB sites. On the basis of these observations, we propose that a major function of the energy expended during Dmc1-mediated recombination is to maintain a pool of unbound protein available for targeted assembly on ssDNA at sites of DSBs. The need for this energy expenditure is proposed to reflect a mechanism of DNA strand exchange, in which free energy is released as the ssDNA filament acquires a complementary DNA strand from the target duplex DNA.