|Home | About | Journals | Submit | Contact Us | Français|
RecQ-like DNA helicases pair with cognate topoisomerase III enzymes to function in the maintenance of genomic integrity in many organisms. These proteins play roles in stabilizing stalled replication forks, the S phase checkpoint response, and suppressing genetic crossovers and their inactivation results in hyperrecombination, gross chromosomal rearrangements, chromosome segregation defects, and human disease. Biochemical activities associated with these enzymes include the ability to resolve double Holliday junctions, a process thought to lead to the suppression of crossover formation. Using E.coli RecQ and topoisomerase III, we demonstrate a second activity for this pair of enzymes that could account for their role in maintaining genomic stability: resolution of converging replication forks. This resolution reaction is specific for the RecQ-topoisomerase III pair and is mediated by interaction of both of these enzymes with the single-stranded DNA-binding protein, SSB.
The RecQ family of 3′ → 5′ DNA helicases play an important role in the maintenance of genomic integrity. Inactivation of many of the human homologs, such as BLM, WRN, and RECQ4, results in hyperrecombination, genome rearrangements, and propensity for cancer (Bennett and Keck, 2004; Cobb and Bjergbaek, 2006; Hickson, 2003; Mankouri and Hickson, 2004; Sharma et al., 2006; Wu, 2007; Wu and Hickson, 2006). The genome-stabilizing activity of RecQ helicases is often linked to the activity of topoisomerase III, a type IA enzyme that breaks and reseals single strands of DNA. This relationship was first established when mutations in the yeast gene Sgs1 were isolated as suppressors of the slow growth and hyperrecombination phenotype of loss of function mutations in Top3, the gene encoding the yeast Top3 (Gangloff et al., 1994). Sequence analysis revealed that Sgs1 was a member of the RecQ family. Sgs1 was also isolated by two-hybrid analysis as a protein that interacted with both toposiomerases II and III (Watt et al., 1995). In these original reports, the authors recognized that the marriage of a DNA helicase with a type I topoisomerase could produce an unique DNA unlinking activity where the helicase would unwind a segment of duplex DNA and the topoisomerase would act on the single-stranded (ss) DNA, perhaps akin to the reverse DNA gyrase that had recently been reported to be composed of helicase and type I topoisomerase domains in a single polypeptide chain (Confalonieri et al., 1993). A more likely role for the Sgs1-Top3 activities, however, was considered to be at the point where replication forks converge.
Positive supercoiling accumulates ahead of a progressing replication fork, requiring the action of topoisomerases to remove the positive supercoils to allow continued progress. Converging replication forks present a particular topological problem in circular DNAs that are replicated bidirectionally and linear chromosomes where free rotation of the parental DNA is constrained by attachment to cellular structures. Under these circumstances, progress of the forks decreases the parental duplex DNA available for the binding of type II topoisomerases that act to remove the positive supercoils (Figure 1). A point is reached wherein either the topoisomerase can no longer bind, or so much positive supercoiling has built up that it presents a barrier to further replication. In their analysis of the replication of SV40 DNA, Sundin and Varshavsky (Sundin and Varshavsky, 1980; Sundin and Varshavsky, 1981) concluded that the major pathway of resolution of these late θ-type replication intermediates [Figure 1 (ii)] was that the remaining region of parental DNA was unwound by continued synthesis of only the nascent leading strands at the forks, converting the remaining parental duplex turns directly into catenanes between the sister molecules, the lagging-strand gaps were then filled in, the nascent strands sealed, and the catenated sister molecules then unlinked by a type II topoisomerase [Figure 1, pathway A, (ii) → (iii) → (iv) → (v)]. In this pathway, while it was possible that a type I topoisomerase could support replication fork progression, only a type II topoisomerase could segregate the linked sister molecules. Studies on SV40 DNA replication in vitro in crude extracts confirmed the existence of this pathway (Yang et al., 1987).
Our observation that a type I topoisomerase could decatenate replicating pBR322 DNA (Minden and Marians, 1986) led us to propose a second pathway of resolution of the late replication intermediate [LRI, (ii) in Figure 1]. Here the topoisomerase would bind to a single-stranded region on the lagging-strand template and unlink the parental DNA as replication fork progression continued, resulting in sister molecules that were completely resolved topologically coincident with the completion of replication [Figure 1, pathway B, (ii) → (vi) → (vii) → (v)]. Our findings that E. coli topoisomerase III (Topo III) required binding to single-stranded DNA for activity (DiGate and Marians, 1988), could segregate replicating pBR322 and oriC sister plasmid DNA molecules in vitro (Hiasa et al., 1994), and could replace topoisomerase IV as the cellular decatenase when overproduced (Nurse et al., 2003) were consistent with this pathway. A popular review by Wang (1991) shortly after the discovery of yeast Top3 (Wallis et al., 1989) discussed the participation of type I and type II topoisomerases in these pathways of chromosome segregation. In considering how Sgs1 and Top3 might act on such LRIs, Rothstein and Gangloff (1995) modified pathway B suggesting that resolution could be accomplished by Sgs1-catalyzed unwinding of the remaining parental duplex DNA between the converging replication forks in concert with Top3-catalyzed unlinking of the single strands.
While an attractive model as a role for the RecQ-like helicase-topoisomerase III pairs, the action of these enzymes in resolving converging replication forks received little support. Most observed defects in chromosome segregation in yeast and S. pombe strains mutated in top3 and sgs1 (rqh1 in S. pombe) have been attributed to unresolved recombinant structures linking the sisters that arise from homologous recombination at stalled replication forks. The top3 phenotypes can be suppressed by mutation in homologous recombination genes (Oakley et al., 2002; Shor et al., 2002) and Sgs1-Top3 have been shown to suppress genetic crossing over (Ira et al., 2003). These observations focused attention on the likelihood that the Sgs1-Top3 pair acted to resolve double Holliday junctions [DHJ(s)] in a manner that prevents crossing over. This model has received considerable support, including the biochemical demonstration that such structures can be resolved in vitro by the human (Wu and Hickson, 2003) and Drosophila (Plank et al., 2006) BLM-Topo IIIα pairs.
Here we demonstrate the ability of the E. coli RecQ-Topo III pair to resolve stalled, converging replication forks in vitro in an LRI structure produced by the replication of oriC plasmid DNA. The resolution reaction displays functional specificity for the RecQ-Topo III pair and functional cooperation between the two proteins that is mediated by interaction with the single-stranded DNA-binding protein (SSB).
To generate an LRI structure with two converging, stalled replication forks we made use of the activity of Tus to block replication fork progression when bound to its DNA binding sequence Ter (Hill and Marians, 1990; Lee et al., 1989). The Tus-Ter system acts to form a replication termination zone on the E. coli chromosome (Kuempel et al., 1989). We engineered a bidirectionally replicating oriC plasmid template carrying two TerB sites oriented to block the oncoming replication forks [Figure 2A (i)]. Replication of the template in vitro with purified proteins yields form II DNA as the major product in the absence of Tus and, as we have demonstrated previously (Hiasa and Marians, 1994), an LRI in the presence of Tus [Figure 2A (ii)]. Because the LRI is essentially twice the size of the input template DNA, it migrates slowly during neutral agarose gel electrophoresis (Figure 2B). Our previous studies (Hill and Marians, 1990) on the replication fork-blocking activity of Tus demonstrated that at the stalled fork, the 3′-OH of the nascent leading strand was ahead of the 5′-end of the last Okazaki fragment (creating a variably-sized gap in the nascent lagging strand) and that the leading strand progressed until it physically collided with Tus bound to the DNA. In the LRI substrate used in this report, the 3′-OH ends of the two nascent leading strands are separated by 130 bp of unreplicated parental duplex DNA [Figure 2A (iv)]. Resolution of the LRI by RecQ and Topo III will yield gapped, form II sister DNA molecules [Figure 2A (iii)].
The LRI was converted to form II DNA in the presence of RecQ, Topo III, and SSB (Figure 2C). All three proteins were required for the resolution reaction. SSB was included in these reactions because the protein will be bound to any ssDNA at a replication fork in the cell. As we show below, SSB in fact appears to mediate functional cooperation between Topo III and RecQ in the resolution reaction. As expected, the combination of Topo III and SSB could not resolve the LRI, nor could the combination of RecQ and SSB. In the latter case, the parental duplex DNA in the LRI was unwound; however, because there was no topoisomerase present to unlink the parental strands, each duplex turn of the unreplicated parental DNA in the LRI that is unwound gives rise to a ss catenane between the sister molecules. In the data shown in this paper, a ladder of bands is generated (these ladders can be seen more clearly in Figures 4, ,6,6, and and7),7), each rung differing by steps of one in the number of linkages between the sister DNA molecules. [Restriction enzyme analysis of the LRI and the ss catenanes is given in Supplemental Figure 1. A ladder of bands is generated as opposed to one band on the 13th rung of the ladder because of the manner in which replication terminates during preparation of the LRI (see the legend to Fig. 2).] This observation suggested that the resolution reaction need not be concerted, i.e., RecQ could unwind the parental duplex first followed by the action of Topo III to unlink the catenated DNA, as opposed to a requirement for a complex of RecQ and Topo III acting together to unwind and unlink in unison. That the reaction is not concerted and can be separated into unwinding and unlinking steps is shown in Figures 6 and and77.
Resolution of DHJs by the hBLM-hTopo IIIα pair also does not require a physical interaction between the two proteins, however, it does require the HRDC (helicase and RNaseD C-terminal) domain of RecQ (Wu et al., 2005). The HDRC domain, found in many members of the RecQ helicase family (ScSgs1, EcRecQ, and hBLM included), can bind DNA but the mode of DNA binding seems to differ among different members of the family. In the case of hBLM, a variant protein deleted for the HDRC was able to unwind forked DNA substrates but was defective in either binding or resolving DHJs, suggesting that the domain was responsible for specific recognition of the substrate. We tested whether this domain was required for resolution of the LRI.
RecQΔHRDC (consisting of the N-terminal 523 amino acids of RecQ) supported LRI resolution nearly as efficiently as intact RecQ (Figure 3), with less than a two-fold difference in specific activity between the two proteins. Because RecQ unwinding of the LRI is initiated at a forked structure, this observation is consistent with the demonstration by Wu et al., (2005) that removal of the HRDC domain from EcRecQ inactivated the ability of the enzyme to unwind a DHJ but not a forked structure. However, whether RecQ is recognizing either a structural feature of the LRI or something else is unclear. Binding and unwinding by RecQ of forked substrates with a complete nascent leading strand (i.e., with the 3′-OH of the leading strand at the fork branch point), such as in the LRI, is poor. In the absence of SSB, preferred RecQ substrates have just the opposite of the fork structure in the LRI: a complete nascent lagging strand (i.e., with the 5′-end of the lagging strand at the fork branch point) and no nascent leading strand (Hishida et al., 2004). These observations suggest that RecQ binding to a specific structural element during resolution of the LRI in the presence of SSB is not likely to be a major component of recognition.
In yeast (Gangloff et al., 1994) and S. pombe (Goodwin et al., 1999), Sgs1 (Rqh1 in S. pombe) and Top3 interact genetically, and the human (Johnson et al., 2000; Wu et al., 2000), Drosophilla (Plank et al., 2006), yeast (Bennett et al., 2000), and S. pombe (Laursen et al., 2003) RecQ-like helicase and topoisomerase pairs interact physically. The hTopo IIIα-interaction domain of hBLM is essential for suppression of sister chromatid exchanges in vivo (Hu et al., 2001), even though it is not required for biochemical resolution of DHJs (Wu et al., 2005). On the other hand, DmBLM does not interact physically with DmTopo IIIβ and the latter enzyme cannot substitute for DmTopo IIIα in a DHJ resolution reaction (Plank et al., 2006). These observations suggest that interaction between RecQ-like helicase-topoisomerase III pairs is required for some manifestation of their activity, be it biochemical or specific localization in the cell. However, neither a physical nor genetic interaction has been demonstrated between EcRecQ and EcTopo III.
To assess whether there was functional specificity to the RecQ-Topo III pair, we investigated the ability of other E. coli topoisomerases and DNA helicases to support the LRI resolution reaction. E. coli possess four topoisomerases, two type IA, Topo I and Topo III, and two type II, DNA gyrase and Topo IV (Champoux, 2001). Topo III was clearly very efficient in supporting the resolution reaction, with half-maximal activity obtaining at about 1 nM concentration (Figures 4A and B). As expected, Topo IV could not support the resolution reaction at all (Figure 4C). Topo IV cleaves and reseals both strands of the duplex at the same time, it will not act on ssDNA and thus should not be able to unlink the ss catenanes generated by RecQ unwinding of the LRI. Topo I, on the other hand, is a type I enzyme and, in theory, could unlink the ss catenanes. Topo I activity in the resolution reaction was very weak. At high concentrations of Topo I, reduction of the number of linkages between the two sister DNA molecules could be observed (a change in the distribution of the linkages is evident), but the enzyme appeared unable, even at a concentration of 100 nM, of completely decatenating the substrate (Figure 4C). The relative inactivity of Topo I compared to Topo III in the resolution reaction may relate to differences in preferred DNA binding sites or to the relative ability of the two enzymes to interact with DNA in the presence of SSB. Topo I does prefer to bind to regions where single-stranded and double-stranded DNA abut (Kirkegaard and Wang, 1985), whereas Topo III does not show this preference and will bind to any region of ssDNA.
E. coli possess at least one dozen DNA helicases. In considering which ones to assay we decided to test those that have been demonstrated to possess activity on stalled fork structures similar to the one that is present in the LRI. Thus we compared the activity of UvrD, Rep, RuvAB, RecG, and PriA to that of RecQ (Figure 5). Interestingly, of these half-dozen DNA helicases that are active at replication forks, only RecQ could support the resolution reaction. Some apparent processing of the LRI could be observed at high concentrations of UvrD. However, coincidently, an ~ 2.0 kb DNA species is also observed. This size corresponds to the distance that the counter-clockwise moving fork travels on the pBROTB I 535-80 template DNA from oriC to the Ter site, indicating that rather than binding to the leading-strand template and unwinding the unreplicated parental DNA directly, as RecQ does, UvrD was binding the lagging-strand template and unwinding all of the nascent DNA before coming around to the unreplicated parental DNA. (The 3.0 kb fragment from the clockwise moving fork was not observed, suggesting some specificity to UvrD binding.)
The data presented in Figures 4 and and55 support the concept of functional cooperation between RecQ and Topo III: of the two type I E. coli topoisomerases that could possibly support the resolution reaction, only Topo III was capable of doing so. Similarly, of a number of E. coli DNA helicases known to be able to modify the structure of stalled replication forks, only RecQ supported the resolution reaction. To provide additional support for this argument, we sought evidence of direct enzymatic synergy between the two proteins.
As shown in Figure 2, in the presence of SSB, RecQ will unwind the LRI, irrespective of whether Topo III is present. To assess the effect of Topo III on RecQ unwinding, we compared the extent of unwinding by RecQ alone and in the presence of either Topo I or Topo III (Figure 6). The total amount of LRI unwound (the sum of ss catenated DNA and form II present as products of the reaction) by RecQ was unaffected by the presence of Topo I. In contrast, at low concentrations of RecQ, the presence of Topo III stimulated unwinding by greater than 4-fold, strengthening our argument that the RecQ-Topo III pair manifest a functional specificity in the LRI resolution reaction.
The functional specificity of the RecQ-Topo III pair demonstrated above seems inconsistent with the lack of a physical or genetic interaction between RecQ and Topo III and the likelihood that structure-specific binding of the substrate by RecQ was not a significant factor in substrate recognition. However, both RecQ and Topo III were shown to interact with SSB in a proteome-wide interaction screen (Butland et al., 2005) and SSB was captured by RecQ in a TAP-tag affinity purification analysis (Shereda et al., 2007). We therefore investigated the role of SSB in mediating the LRI resolution reaction.
Several proteins that are involved in the maintenance of genomic integrity are known to interact with the extreme C-terminus of SSB. Therefore, the ability of two variant SSBs, one lacking the last eight C-terminal amino acid residues [SSBΔC8 (Shereda et al., 2007)] and the other being SSB113 [SSBP176S, originally isolated as lexC113 (Glassberg et al., 1979; Meyer et al., 1982)] to support the resolution reaction was investigated. The P176S substitution is known to disrupt the interaction of SSB with a number of replication proteins (Kelman et al., 1998; Yuzhakov et al., 1999). Under standard conditions, SSBΔC8 did not support the resolution reaction, whereas SSB113 gave 50% of the activity of the wild type (Figure 7A). This observation indicated that the C-terminal domain of SSB played a central role in the resolution reaction and that mutation of the penultimate C-terminal residue could account for about half of the SSB-dependent stimulation of the resolution reaction.
SSB is known to stimulate both the DNA unwinding activity of RecQ (Harmon and Kowalczykowski, 2001; Shereda et al., 2007; Umezu and Nakayama, 1993) and the superhelical DNA relaxation activity of Topo III (Harmon et al., 2003). In order to distinguish the effects of the SSB variants on either RecQ or Topo III, we divided the reaction into two steps. Unwinding of the LRI by RecQ could be observed in the presence of wild-type SSB (Figure 7B). As with the complete resolution reaction, SSB113 supported about one-half the RecQ-catalyzed unwinding as the wild-type, whereas SSBΔC8 was significantly less active, particularly at lower concentrations of RecQ. To observe the effect of the variant SSBs on Topo III unlinking of the LRI, we first treated the LRI with RecQ and SSB to unwind about 40% of the linkages. This unwound LRI was purified free of protein and used as a substrate for Topo III unlinking of the unwound parental duplex DNA (Figure 7C). When considering this figure, it is important to note that progressive unlinking of the ss catenanes causes a change in the distribution of the ladder of bands, shifting it toward one where the mobility of the bands has decreased (which is quite obvious when comparing the substrate itself to the lane containing Topo III and SSBΔC8). Furthermore, sister DNA molecules linked only once co-migrate with the LRI, so an increase in the fraction of the substrate present as LRI compared to the starting material indicates unlinking by Topo III. As expected, the presence of wild-type SSB stimulated unlinking somewhat. On the other hand, the activity of SSB113 was similar to that of the wild type, whereas SSBΔC8 inhibited compared to when the wild-type protein was present.
To investigate the interactions between RecQ and SSB; and Topo III and SSB, we used a co-precipitation assay. This assay, developed originally to score the RecQ-SSB interaction (Genschel et al., 2000) exploits the fact that SSB is very insoluble in (NH4)2SO4 solution, precipitating at 27% saturation; whereas both RecQ and Topo III are soluble at that saturation level (Figure 7D). As demonstrated previously using this assay (Shereda et al., 2007), RecQ co-precipitates with SSB (Figure 7D), but not with SSBΔC8 (Figure 7E). Interestingly, the Topo III-SSB interaction resulted in solubilization of SSB (Figure 7D); however, as expected, Topo III also did not interact with SSBΔC8 (Figure 7E). No interaction between RecQ and SSB113 could be detected, whereas SSB113 still interacted to a significant extent with Topo III (Figure 7F). Taken together, the data in Figure 7 indicate that Pro176 is important for interaction with RecQ, but not with Topo III, and that there are other residues in the C-terminal eight amino acids of SSB that are important for interaction with both RecQ and Topo III. Whether these are the same or different amino acid residues remains to be determined. We conclude that the LRI resolution reaction is facilitated by interaction of both RecQ and Topo III with the C-terminal tail of SSB. These interactions help generate the functional specificity observed for the RecQ-Topo III pair in the LRI resolution reaction.
RecQ-like DNA helicases clearly play an important role in preserving genomic integrity. Of the five human family members, inactivation of three cause the rare genetic disorders Bloom syndrome (BLM), Werner syndrome (WRN), and Rothmund-Thomson syndrome (RecQ4), which carry serious health consequences including premature aging and increased cancer incidence. In eukaryotes, various members of this family of DNA helicases have been shown to play a role in stabilizing replisomes, processing of stalled replication forks, the S phase checkpoint response, Okazaki fragment processing, suppression of crossovers, chromosome segregation, double-strand break repair, and telomere maintenance. Several aspects of the activity of the RecQ helicases in these processes are linked to their interactions with topoisomerase III.
What are the biochemical activities that underlie the participation of RecQ helicases and Topo III in these pathways of maintaining genome stability? Like other DNA helicases, the RecQ family members can operate on a number of different substrates, yielding different outcomes. Several activities that appear relevant to the role of this family of helicases in preserving genomic integrity are: the ability to branch migrate a Holliday junction; to unwind the invading strand in D loops; to unwind G quartets; and to regress forked structures.
Biochemical cooperation between RecQ-like DNA helicases and Topo III has been observed in two different reactions. Harmon et al. (1999) showed that in the presence of EcSSB, the combination of EcRecQ and EcTopo III could fully catenate supercoiled plasmid DNAs. This intermolecular reaction is stimulated by volume excluded conditions that effectively increase the concentration of the DNA (Harmon et al., 2003). The notion is that as RecQ unwinds the DNA it creates a substrate that can be acted on by Topo III. Sequential strand passage events lead to the formation of full, double-stranded catenanes. Under conditions where the DNA is present at dilute concentration, the strand passage events tend to be intramolecular, leading to relaxation of positive supercoils and the generation of negative supercoils. Specificity was demonstrated for EcTopo III (neither EcTopo I nor EcTopo IV substituted) and partially so for EcRecQ (EcUvrD did not substitute).
Wu and Hickson (2003) devised a substrate composed of two circular oligonucleotides joined together by a DHJ and showed that a combination of hBLM and hTopo IIIα could resolve the substrate into two circular molecules, a direct demonstration of the activity presumed responsible for suppressing crossover formation during recombination. Here no other RecQ-like helicase could substitute for hBLM (Wu et al., 2005), however, both EcTopo I and EcTopo III could substitute for hTopo IIIα (Wu et al., 2006), suggesting a lack of specificity. This substrate, where the two circular oligonucleotides were topologically linked only by the two strand exchanges in the non-mobile Holliday junctions, was representative of the very end stage of the resolution of DHJs. Plank et al. (2006) constructed a substrate representative of an earlier stage in the process, where two small double-stranded DNA circles were joined by mobile DHJs separated by 165 bp of homologous DNA. Because the substrate was topologically constrained, the Holliday junctions could not migrate spontaneously. This substrate could be resolved by DmBLM and DmTopo IIIα in the presence of RPA. Resolution occurred by BLM-catalyzed convergent branch migration of the two Holliday junctions toward each other as Topo III acted to remove the linkages between the circular DNAs.
We have examined another reaction thought to be mediated by the RecQ-Topo III pair, resolution of converging replication forks. Almost completely replicated sister DNA molecules joined by an ~ 130 bp region of unreplicated parental DNA were very efficiently resolved into monomeric gapped form II DNAs in the presence of EcSSB by the combination of EcRecQ and EcTopo III. This reaction differs from the DHJ resolution reaction in that it is the DNA unwinding activity of RecQ, not its branch migration activity, that is important for separation of the two strands of the unreplicated duplex DNA. Functional cooperation between EcRecQ and EcTopo III could be demonstrated in that of a half dozen E. coli DNA helicases known to be active on structures modeling stalled replication forks, only EcRecQ was active in the LRI resolution reaction. Similarly, of the two E. coli type I topoisomerases, only EcTopo III was active in the resolution reaction. Furthermore, we demonstrated that EcTopo III stimulated EcRecQ unwinding of the unreplicated parental DNA in the LRI, whereas EcTopo I did not.
The LRI resolution reaction reported here is more akin to the catenation reaction (Harmon et al., 1999) than the DHJ resolution reaction (Plank et al., 2006; Wu and Hickson, 2003). However, we note that the catenation reaction of Harmon et al. (1999) is considerably less efficient than the LRI resolution reaction, requiring markedly higher concentrations of RecQ and Topo III and a molecular crowding agent for robust product formation. We conclude that the combination of RecQ and Topo III is uniquely suited to resolving converging replication forks.
It seems reasonable that a complex of the RecQ helicase family member and cognate Topo III would be the most efficient agent of manifesting these biochemical activities that can resolve converging replication forks and DHJs. However, unlike many eukaryotic RecQ helicase-Topo III pairs, EcRecQ and EcTopo III do not interact physically, suggesting that complex formation was not required. Consistent with this argument, the domain of hBLM required for interaction with hTopo III was not required for resolution of the oligonucleotide DHJ substrate (Wu et al., 2005). On the other hand, neither DmTopo I, DmTopo II, nor DmTopo IIIβ substituted for DmTopo IIIα in the resolution reaction established by Plank et al. (2006). Given that DmTopo IIIβ does not interact with DmBLM whereas DmTopo IIIα does, the authors argued for complex formation between the two proteins.
The key to complex formation during these resolution reactions could be the presence of an SSB. In the case of DHJ resolution by the Drosophila proteins, both DmBLM and DmTopo IIIα interact with RPA. hRmi1, a protein shown to exist in complex with hBLM and hTopo IIIα (Raynard et al., 2006; Wu et al., 2006; Yin et al., 2005), contains an OB-fold and could be a specialized SSB (Chen and Brill, 2007). Inclusion of hRmi1 in the DHJ oligonucleotide resolution reaction specifically stimulates the activity of hTopo IIIα, but not that of other type IA enzymes (Wu et al., 2006). EcSSB stimulates EcRecQ helicase activity (Harmon and Kowalczykowski, 2001; Shereda et al., 2007; Umezu and Nakayama, 1993) and EcTopo III superhelical DNA relaxation (Harmon et al., 2003). Consistent with these observations, we found that the C-terminal tail of EcSSB mediated the LRI resolution reaction via interaction with both RecQ and Topo III. In general, deletion of the C-terminal domain, which yields an SSB that binds ssDNA roughly as well as the wild type (Shereda et al., 2007), inhibited the resolution reaction and ablated physical interaction between SSB and either RecQ or Topo III. The penultimate C-terminal amino acid, Pro176, clearly played a role in stimulating EcRecQ unwinding (and SSB113 and RecQ did not interact), whereas it was not necessary for stimulation of EcTopo III unlinking (SSB113 and Topo III retained significant interaction). Additional, as yet undefined, amino acid residues of the C-terminal tail of EcSSB were also involved in mediating the activity of both EcTopo III and EcSSB. At least in the case of the LRI resolution reaction demonstrated here, access to the substrate for both EcTopo III and EcRecQ may therefore be via interaction with EcSSB.
Failure to resolve converging replication forks in vivo would likely lead to defects in chromosome segregation, and there is considerable data demonstrating such defects for cells where either the RecQ or Topo III activities have been inactivated. In S. pombe, inactivation of Top3 leads to inviability, with many cells displaying a cut phenotype, where the septum has bisected nuclear masses that have failed to separate completely (Goodwin et al., 1999; Oh et al., 2002). Similar defects are observed if the RecQ helicase is inactivated (Win et al., 2005). In E. coli, topB is required in strains where topA (encoding Topo I) is inactivated and that carry compensatory mutations in the genes encoding the subunits of DNA gyrase (Zhu et al., 2001). Cells mutated in both type I topoisomerase genes filament and display aberrant nucleoid structure. And overproduction of EcTopo III can rescue the lethality of inactivating the primary cellular decatenase, Topo IV (Nurse et al., 2003).
Some recent observations also lend support to a role for the RecQ-Topo III pairs in resolving converging replication forks in normally growing cells. Seki et al. (2006) engineered chicken DT40 cells where Top3α was homozygously disrupted and substituted for by mouse Topo IIIα under the control of a Tet-off promoter provided on a plasmid. Depletion of the ectopic Topo IIIα led to accumulation of cells in G2 phase, enlargement of nuclei, chromosome breaks and gaps, and cell death. All of these phenotypes except cell lethality could be suppressed by homozygous disruption of BLM. The authors argued that in the absence of Topo III, BLM unwinds the parental DNA between converging replication forks, converting the topological linkages between the duplex turns to ss catenanes that persist because they cannot be unlinked by any other topoisomerase in the cell. These linkages lead directly to chromosome breakage and mitotic catastrophe.
Chan et al. (2007) have observed that ultrafine DNA bridges arise at high frequency between segregating sister chromosomes in normal human cells undergoing mitosis. These bridges were identified initially because they were decorated with BLM. Subsequent analysis showed that Topo IIIα and RMI1, the cellular partners of BLM, were present on the bridges as well. These observations suggested that incomplete replication was a frequent occurrence, generating a requirement for resolution of converging replication forks.
On the face of it, one would assume that a failure to resolve converging replication forks would not be suppressed by inactivation of proteins involved in recombination, as most phenotypes resulting from inactivation of RecQ-like helicases are (Liberi et al., 2005; Watt et al., 1996), as well as those associated with inactivation of Top3 in yeast (Oakley et al., 2002; Shor et al., 2002). The role of homologous recombination in generating structures that are resolved by the RecQ-Topo III pairs has been a strong argument for reasoning that the major role of these enzymes is in reactions such as DHJ resolution. However, it seems likely that aberrant recombination structures would also be generated at converging replication forks that have stalled. The governing feature, then, would seem to be an issue of relative rates of competing reactions. Given sufficient time, the remaining region of non-replicated parental DNA could be resolved via pathway A (Figure 1). Under these circumstances, the fully catenated sister chromosomes could then be unlinked by type II topoisomerases, yet still be joined together because of recombination that occurred at the stalled forks. Under these circumstances, a defect that arose initially because of the failure to resolve converging replication forks would, in the end, be suppressed by inactivation of recombination proteins. Clearly, additional investigation will be required to assess the participation in the cell of RecQ-Topo III pairs in these various reaction pathways.
Proteins for replication of the oriC template DNA [DnaA, DnaB, DnaC, DnaG, HU, SSB, DNA gyrase, and the DNA polymerase III holoenzyme (as Pol III* and the β subunit)] were as described previously (Marians, 1995; Peng and Marians, 1993a). Tus was purified as described (Hill and Marians, 1990). Topo I and Topo III were as described (Hiasa et al., 1994), as was Topo IV (Peng and Marians, 1993b). SSBΔC8, SSB113, and RecQΔHRDC were the kind gifts of J. Keck (University of Wisconsin). PriA was purified as described (Marians, 1995). UvrD was the kind gift of T. Lohman (Washington University). Rep was purified as described (Heller and Marians, 2005). RuvA was purified from soluble extracts of BL21(DE3)(pAM159) by sequential chromatography on Q Sepharose, ssDNA cellulose, heparin-agarose, and hydroxylapatite. RuvB was purified from soluble extracts of BL21(DE3)(pGTI19) by sequential chromatography on phosphocellulose, Q-Sepharose, hydroxylapatite, butyl-sepharose, and Mono-Q. Plasmids pAM159 and pGTI19 were the kind gifts of R. Lloyd (University of Nottingham). RecG was purified from soluble extracts of BL21(DE3)pET21a-recG [a Novagen plasmid engineered to over-express the RecG open reading frame (with the native GTG initiator codon changed to an ATG)] by sequential chromatography on DE52, ssDNA cellulose, heparin-agarose, and hydroxylapatite. RecQ was purified from soluble extracts of BL21(DE3)pET21a-recQ (a Novagen plasmid engineered to over-express the RecQ open reading frame) by sequential chromatography on Q-Sepharose, heparin-agarose, hydroxylapatite, and Superdex 200.
The plasmid templates pBROTB I 535-80 and pBROTB II 535-80 were derived from plasmids pBROTB535-I and pBROTB535-II (Hiasa and Marians, 1994) that carry oriC in either of two orientations as well as two Ter sites oriented to block the oncoming replication forks. In the case of oriC orientation I, the Ter sites are 2 kb counter-clockwise, and 3 kb clockwise from the origin. In the case of oriC orientation II, the Ter sites are both 2.5 kb from the origin. In pBROTB I 535-80 and pBROTB II 535-80, there is only 80 bp between the two Ter sequences. The sequence between the Ter sites is as follows: the Ter sites are underlined; the nascent leading-strand stop sites (Hill and Marians, 1990) are in bold; and in italics are the recognition sequences for the SpeI, SacII, NsiI, NheI, and BssHII restriction enzymes, respectively, 5′ to 3′ 5′-AATAAGTATGTTGTAATTAAAGTGATCAGGCTTCTGCCGTACTAGTTTGGCCGCGGAT TTATGCATAGATGATTTCGATTGCTAGCAGTAACAAAGTTTGGCGCGCATCACTTTAG TTACAACATACTTATT-3′.
Replication reaction mixtures (500 μl) containing 50 mM HEPES-KOH (pH 8.0), 10 mM Mg(OAc)2, 10 mM DTT, 200 μg/ml bovine serum albumin, 2 mM ATP, 200 μM GTP, CTP, and UTP, 40 μM [α-32P]dATP (2000 cpm/pmol), TTP, dCTP, and dGTP, 350 nM DnaA, 30 nM DnaB, 200 nM DnaC, 240 nM DnaG, 27 nM HU, 10 nM DNA gyrase, 400 nM SSB, 10 nM Pol III*, 30 nM β subunit of the DNA polymerase III holoenzyme, 20–60 nM Tus, 8 units/μl bacteriophage T4 DNA ligase (New England Biolabs), 5 milliunits/μl DNA polymerase I (New England Biolabs), and 2.8 nM plasmid template DNA were incubated at 37 °C for 15 min. NaCl and EDTA were then added to 300 mM and 20 mM, respectively, and the incubation continued for 3 min. SDS and proteinase K were then added to 0.8% and 200 μg/ml, respectively, and the incubation continued for 30 min. The reaction mixture was then loaded onto two 11 ml 10%–30% sucrose gradients formed in 10 mM Tris-HCl (pH 7.5 at 4 °C), 1 mM EDTA, 1 M NaCl, and sedimented at 29,000 rpm for 19.5 h at 4 °C in a Sorvall TH641 rotor. Fractions were collected from the bottom of the tube and those containing the LRI were pooled, dialyzed against 10 mM Tris-HCl (pH 7.5 at 4 °C), 0.1 mM EDTA, concentrated by a factor of ten by lyophilization, and then ethanol precipitated. LRI was resuspended at a concentration of 2.5 nM in 10 mM Tris-HCl (pH 7.5 at 4 °C), 1 mM EDTA.
Standard reaction mixtures (8 μl) contained 50 mM HEPES-KOH (pH 8.0), 1 mM Mg(OAc)2, 10 mM DTT, 100 μg/ml bovine serum albumin, and 1 mM ATP. Reactions were assembled on ice with the indicated proteins (see figure legends) and then warmed at room temperature for 5 min. The reactions were initiated by the addition of LRI to a final concentration of 0.25–0.5 nM. Reaction mixtures were incubated at 37 °C for 7 min. NaCl and EDTA were then added to 300 mM and 20 mM, respectively, and the incubation continued for 3 min. SDS and proteinase K were then added to 0.8% and 200 μg/ml, respectively, and the incubation continued for 30 min. Reactions were analyzed by electrophoresis through either 1% or 1.4% vertical agarose gels at 1.5 V/cm for 15 h using 50 mM Tris-HCl (pH 8.3 at 23 °C), 40 mM NaOAc, 1 mM EDTA as the running buffer. Gels were dried, exposed to phosphorimager screens for quantification using a Fuji FLA7000, and autoradiographed.
Reaction mixtures (20 μl) containing 50 mM Tris-HCl (pH 7.5 at 4 °C), 150 mM NaCl, and either 12.5% glycerol (for SSB) or 14.5% glycerol (for SSBΔC8 and SSB113), and 5 μM of each protein were incubated on ice for 1 h. Reaction mixtures were then brought to 27% saturation of (NH4)2SO4 by the addition of 7 μl of a 100% saturated solution and the incubation continued for an additional 15 min. The reaction mixtures were then centrifuged at 13,000 × g for 1 min. The supernatants were removed and the pellets washed twice with 50 μl of 27% (NH4)2SO4 in binding buffer. The pellet was then dissolved in 54 μl of SDS-PAGE loading dye and 27 μl of 2X SDS-PAGE dye was added to the supernatants. Three microliters of each reaction were then analyzed by SDS-PAGE using a 10%–15% step gel. Gels were stained with Coomassie Brilliant Blue and photographed.
This study was supported by NIH grant GM34558. We thank Bob Lloyd, Jim Keck, and Tim Lohman for reagents, Ryan Heller and Ram Madabhushi for discussion, and Rod Rothstein and John Petrini for critical reading of the manuscript. This paper is dedicated to the memory of Nick Cozzarelli, seer of all things topological, discoverer of Topo III, mentor, and good friend.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.