Skip to main content
NCBI home page
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2014 Sep 2;111(39):14112–14117. doi: 10.1073/pnas.1407562111

Mu transpososome and RecBCD nuclease collaborate in the repair of simple Mu insertions

Wonyoung Choi 1,1, Sooin Jang 1, Rasika M Harshey 1,2
PMCID: PMC4191772  PMID: 25197059

Significance

Transposition is not finished until the gaps left in the target in the wake of transposition are repaired, yet this is the least understood aspect of transposition. This study describes to our knowledge the first in vitro assay for repair of Mu insertions, which shows that repair is controlled by both transpososome and host factors. The long flanking DNA (FD) attached to Mu is shortened in two stages, requiring collaboration between the transpososome and RecBCD at both stages. The transpososome initially regulates the timing of RecBCD entry into the FD, and then exploits RecBCD action on this DNA to promote endonucleolytic cleavage within the transpososome. The final product has short flanks, and is likely the substrate for gap repair by a host polymerase.

Keywords: conservative integration, nonreplicative transposition

Abstract

The genome of transposable phage Mu is packaged as a linear segment, flanked by several hundred base pairs of non-Mu DNA. The linear ends are held together and protected from nucleases by the phage N protein. After transposition into the Escherichia coli chromosome, the flanking DNA (FD) is degraded, and the 5-bp gaps left in the target are repaired to generate a simple Mu insertion. Our study provides insights into this repair pathway. The data suggest that the first event in repair is removal of the FD by the RecBCD exonuclease, whose entry past the N-protein block is licensed by the transpososome. In vitro experiments reveal that, when RecBCD is allowed entry into the FD, it degrades this DNA until it arrives at the transpososome, which presents a barrier for further RecBCD movement. RecBCD action is required for stimulating endonucleolytic cleavage within the transpososome-protected DNA, leaving 4-nt flanks outside both Mu ends. This end product of collaboration between the transpososome and RecBCD resembles the intermediate products of Tn7 and retroviral and retrotransposon transposition, and may hint at a common gap-repair mechanism in these diverse transposons.

The repair of transposon insertions is the least understood aspect of transposon biology. Transposable phage Mu provides an excellent system to study this process because of its high transposition frequency. Mu is a temperate phage, which uses transposition not only to integrate into its Escherichia coli host to generate prophages but also to amplify its genome during the lytic cycle, where transposition is coupled to replication (Fig. 1A) (1–4).

Fig. 1.

Fig. 1.

Open in a new tab

Two transposition pathways during the Mu life cycle. The chemical steps of single-stranded DNA cleavage at Mu ends followed by strand transfer (ST) of the cleaved ends to phosphodiester bonds spaced 5 bp apart on the target are the same in both the lytic (A) and infection (B) phases of transposition. During the lytic phase, Mu transposition is intramolecular, and the ST intermediate is resolved by target-primed replication through Mu (A). During the infection phase, transposition is intermolecular, and the Ɵ ST intermediate is resolved by removal of the flanking DNA and repair of the 5-bp gaps left in the target (B). The FD is covalently closed in A but noncovalently closed by phage N protein (oval) bound to the tips of the FD in B. The target DNA flanking Mu is red in all of the figures. Arrowheads indicate the 3′ ends.

This study focuses on the transposition event that occurs immediately after infection, which is followed by repair rather than replication of the Mu insertion. The infecting Mu genome is linear, and is linked to several hundred base pairs of non–Mu-flanking DNA. The tips of this DNA are held together by an injected phage protein, N, which converts it into a noncovalently closed circular form and protects the DNA against exonucleases (5–7). After integration, the flanking DNA (FD) is removed (8), and the 5-bp gaps generated in the target are repaired to yield a simple Mu insertion (Fig. 1B) (9). Unlike the noncovalently closed configuration of the infecting Mu genome, replicating Mu is initially part of a covalently closed circular E. coli chromosome (Fig. 1A). At the end of the lytic phase, Mu replicas are packaged such that host DNA linked to either side of the insertion is included in the phage head; this is the source of the FD in the infecting Mu genome.

The chemical steps of Mu transposition during the replication and repair pathways are the same, namely the MuA transposase nicks Mu DNA at each end and then joins the nicked ends to target DNA cleaved 5 bp apart (Fig. 1) (2, 8). However, the resulting branched strand-transfer intermediate is resolved alternatively by target-primed replication during the lytic phase (Fig. 1A) (10) and by FD removal and limited replicative repair during the infection phase (Fig. 1B) (8, 11). The alternate fates of a similar strand-transfer joint have long been a matter of speculation. Although much has been learned about the proteins that promote the transition from transposition to replication (10, 12, 13), little is known about those that assist in repair, other than that the gap-filling E. coli polymerase PolA is not required and that the double-strand break (DSB) repair machinery is somehow involved (14).

We show in this study that the noncovalently closed configuration of the FD in the infecting Mu donor substrate controls the fate of the strand-transfer joint. Our data show that the RecBCD exonuclease is required in vivo for degradation of the FD. This occurs only after integration of infecting Mu, suggesting that there is a timed mechanism to remove N and allow RecBCD access into the FD. Attempts to recapitulate this reaction in vitro have revealed that the FD is processed in two stages: First, RecBCD degrades the long DNA, and next, a specific endonucleolytic cleavage leaves short 4-nt flanks within the transpososome-protected DNA. Earlier, we had found that a cryptic endonuclease activity of MuA was required for removal of the flanks in vivo, and had interpreted the data to suggest that degradation of this DNA is initiated by endonucleolytic cleavage (15). In light of the results obtained in the present study, we reinterpret our earlier data and propose a new model for repair. We expect that this first biochemical analysis, to our knowledge, of the initial steps of Mu repair will reveal pathways common to the repair of all transposon insertions.

Results

Flanking-DNA Removal in Vivo Depends on the Nuclease Activity of RecBCD.

Repair of Mu insertions involves two events: removal of the FD and the filling in of the 5-bp gaps (Fig. 1B). Of these, the former is readily monitored, because this DNA is long. We have therefore used this reaction as an in vivo assay for repair.

An earlier genetic screen designed to identify host factors required for posttransposition recovery (and hence repair) of Mu insertions identified five candidate genes: recA, recB, recC, priA, and dnaT (14). These genes function together in the repair of DSBs, but could additionally participate in removal of the FD. At least one of these genes, priA, was shown earlier to not be involved in the DNA removal function (15). The products of the recB and recC genes are part of the RecBCD exonuclease. To test whether these genes might be involved, we assayed FD removal in these mutants as well as in a number of other E. coli mutants that had been identified as conditionally defective for the recovery of Mu lysogens (14). The assay involves infecting lacZ− hosts with phage propagated on a lacZ+ host, using lacZ sequences as markers for FD (8). At various times after infection, total cellular DNA was isolated, integrated Mu DNA was separated from free Mu DNA by electrophoresis, and Mu and lacZ sequences were monitored in the integrated DNA fraction by PCR.

In a wild-type host, Mu integration is typically detected within 15 min of infection, concomitant with detection of lacZ FD sequences (Fig. 2, WT). The Mu signal increases at the 30- and 50-min time points because of Mu replication during the lytic growth. However, the signal for lacZ (FD) decreases sharply by 30 min and is not detectable at 50 min. This is the last time point assayed because cells begin to lyse around 50 min. Of the 16 mutants identified earlier as being impaired in the recovery of Mu lysogens (14), the majority showed normal kinetics of disappearance of the FD lacZ marker (Fig. S1), with the exception being the recB and recC mutants (Fig. 2). The latter support Mu integration and replication (14), but neither was proficient in FD removal (Fig. 2; see also Fig. S2D). The lacZ signal in these strains persisted up to the onset of lysis at 50 min. In the RecBCD exonuclease complex, the D subunit is encoded by recD (16); RecB and RecD have helicase activities on the two DNA strands, and RecB has the common nuclease activity site (17). Although recD was not identified in our genetic screen, we monitored this mutant for FD removal because without RecD, RecBC is not active as a nuclease (18, 19). The recD mutant showed the same pattern as the recB and recC mutants, in that the FD was not degraded until 50 min. All three mutants support normal Mu lytic growth (Fig. S2 A–C). As an additional test, we introduced a point mutation within recB, which is known to inactivate the nuclease activity of RecB while retaining its helicase activity (20). The recB(D1067A) mutant behaved like the recB null mutant in being defective for FD processing (Fig. 2).

Fig. 2.

Fig. 2.

Open in a new tab

Flanking-DNA removal depends on the RecBCD nuclease. Wild-type (BW25113) or its rec mutant derivatives (SI Methods) were infected with Mu, genomic DNA was isolated at various times, and Mu or FD was detected as described in SI Methods. C, control, where genomic DNA from uninfected cells was mixed with Mu DNA prior to electrophoresis to gauge the extent of contamination of the genomic DNA band with free Mu in subsequent PCRs; M, purified Mu DNA; N, no template control. See also Fig. S2D.

We conclude that the exonuclease activity of RecBCD is required for removal of the FD after transposition of infecting Mu.

Degradation by RecBCD Depends on the Flanking-DNA Configuration.

The Mu transposition reaction is well-established in vitro (2, 3). Addition of MuA/MuB proteins and E. coli HU protein to a mini-Mu donor plasmid leads to Mu end cleavage by the MuA transposase, followed by MuB-assisted strand transfer of the cleaved ends into a target plasmid to generate the Ɵ-shaped strand-transfer intermediate (Fig. 3A, lane 1 and Fig. 3B, Center). When treated with SDS, the Ɵ complex disassembles to generate a series of topoisomers, reflecting supercoils still trapped within the Mu segment of the Ɵ structure (lane 2). Addition of RecBCD to the Ɵ complex did not change its configuration (lanes 3 and 4). Although ClpX is required in vivo for efficient removal of the FD (Fig. S1) (15), addition of ClpX did not promote degradation of the Ɵ complex by RecBCD.

Fig. 3.

Fig. 3.

Open in a new tab

RecBCD degrades the FD in a Ɵ complex in vitro when this DNA is linearized. (A) Ɵ strand-transfer complexes generated using plasmid substrates (lanes 1 and 2; B, Center) were treated with RecBCD before or after XmnI digestion (lanes 3–8) or with cell extracts after XmnI treatment (lanes 9–11). Lane 12: the nicked form of pWY96 used as a 4.8-kb marker for the expected size of the Mu simple insert (4.7 kb) in lanes 8–10. Reactions were run on agarose gels with or without SDS treatment (SI Methods). BCD−, extracts from the recBCD mutant host; D, mini-Mu donor; I, extracts from Mu-infected host BU1384; ln, linear; oc, open circular; S, simple insert; S′, SDS-treated simple insert; U, cell extracts from uninfected host BU1384. See SI Methods for details. Plasmids are described in Table S1. (B) Schematic of Ɵ DNA configurations in vivo and in vitro, and of the expected simple insert product after FD removal. Arrowheads indicate the 3′ ends of Mu. The FD is covalently closed in the lytic phase (Fig. 1), as it is in the Ɵ formed in vitro.

RecBCD requires a free double-stranded DNA end for entry (21). Such an end is not available on the Ɵ complex in vitro, but is potentially available in the Ɵ complex in vivo if the N protein were to be removed (Fig. 3B). To mimic such an in vivo substrate, the FD on the in vitro Ɵ complex was linearized with the restriction enzyme XmnI (Fig. 3B, Center); the XmnI cut site was positioned to leave 250 bp and 2 kb of DNA outside the Mu L end and Mu R end, respectively, approximating the normal Mu FD lengths. XmnI digestion of the Ɵ complex (Fig. 3A, lane 5) resulted in a slower-migrating DNA band when deproteinized with SDS (lane 6). When RecBCD was added to the XmnI-cut Ɵ, a faster-migrating band (S) was detected (lane 7). SDS treatment converted this band into a slower-migrating species (S′) (lane 8), which had the size expected for a simple insert of the Mu segment into the target plasmid (Fig. 3B, Right) (∼4.7 kb; lane 12).

To determine whether RecBCD is the only nuclease involved in removal of the FD in vivo, the XmnI-cut Ɵ complex was incubated with cell extracts prepared from uninfected (U) or Mu-infected (I) cells (Fig. 3A, lanes 9 and 10). Both extracts converted the cut Ɵ complex to the S′ product, showing that Mu proteins are not involved. Extracts prepared from a recBCD mutant (BCD−) failed to generate the S′ product (lane 11), showing that RecBCD is the major nuclease in cell extracts.

We conclude that RecBCD degrades the DNA flanking Mu in a Ɵ complex generated in vitro only when an entry site for the nuclease is provided. RecBCD is the primary exonuclease activity in E. coli that efficiently degrades this DNA. Taken together with data in Fig. 2, we infer that after integration of infecting Mu DNA, RecBCD must gain access into the FD.

The Transpososome Prevents RecBCD from Degrading Mu.

To test the role of the transpososome in generating the RecBCD-dependent S/S′ product, the XmnI-cleaved Ɵ complex (Fig. 4, lane 1) was treated with RecBCD, where it generated the expected S′ product (lane 3). When the transpososome was removed from the Ɵ complex by phenol treatment (similar to the SDS treatment in lane 2), RecBCD degraded the DNA in its entirety (lane 4).

Fig. 4.

Fig. 4.

Open in a new tab

Transpososome blocks RecBCD access to Mu sequences. XmnI-cut Ɵ complex (lane 1) was treated with RecBCD either before (lane 3) or after (lane 4) removal of the transpososome with phenol. ClpX-treated XmnI-cut Ɵ complex (lane 5), degraded completely when treated with RecBCD (lane 6). Lane 2: control showing the position of the deproteinized complex in lane 1. S′, SDS-treated simple insert.

ClpX is known to weaken the tightly bound transpososome for transition into replication (10). When added to the XmnI-cleaved Ɵ complex at concentrations observed to destabilize the transpososome (Fig. 4, lane 5) (15), RecBCD degraded the DNA completely (lane 6).

We conclude that RecBCD, which is known to evict most DNA-bound proteins (22), is stopped by the MuA transpososome. Either removing the transpososome entirely or destabilizing it with ClpX allows RecBCD entry through the host–Mu junction and into the rest of the Mu DNA. We infer that the transpososome is not similarly destabilized by ClpX in vivo, because such an action would be lethal to the survival of the Mu insertion.

RecBCD Does Not Degrade the Flanking DNA Protected Within the Transpososome.

The transpososome is reported to protect 10–13 bp of the FD (23, 24). To determine how much of this DNA is degraded by RecBCD, the S′ product was isolated from agarose gels and subjected to primer-extension analysis with fluorescently labeled oligonucleotides that annealed within the Mu L or R end (Fig. 5A). If RecBCD stops at the Mu–host junction, the labeled primer will be extended to this junction (labeled “0”). However, the product of primer extension on the S′ template was larger than the 0 marker (Fig. 5B, lanes M) and corresponded to +19 bp outside both the L- and R-end Mu–host junctions, as determined by size analysis (SI Methods) (Fig. 5B, lanes +). Without RecBCD, the primer-extension products were long and did not enter the gel (lanes −).

Fig. 5.

Fig. 5.

Open in a new tab

RecBCD stops 19 bp outside the Mu–host junction. (A) Schematic for detecting FD length after RecBCD treatment on the S′ simple insert (Figs. 3 and 4) by extension of 6-FAM (carboxyfluorescein amidite)–labeled primers (short arrows) hybridizing to the L and R ends. 0, Mu–host junction. Arrowheads on Mu indicate 3′ ends. (B) Primer-extension products after RecBCD treatment (lanes +) or without RecBCD (lanes −), resolved on 8% denaturing polyacrylamide gels. Lanes M contain a marker that matches a product extended to precisely the Mu–host junction. The size of the product in lanes + was determined as described in SI Methods.

Taken together, the data in Figs. 3–5 allow us to conclude that RecBCD digests most of the FD in the Ɵ complex, except for the terminal 19 bp proximal to Mu. Resistance of this terminal DNA to RecBCD is consistent with arrest of the nuclease at the transpososome boundary, because the transpososome protects 10–13 bp of this sequence.

A Host Factor(s) Stimulates Flanking-DNA Cleavage Within the Bound Transpososome.

We surmised that the +19 RecBCD product was likely not the end point of the reaction, because a cryptic endonuclease activity (MuANuc) encoded by the BAN (DNA binding and endonuclease) region within the C-terminal domain of MuA (25) was shown earlier to be important for removal of the FD in vivo (15). We therefore decided to look for other host factors in the E. coli extracts that might stimulate MuANuc and process the +19 product further.

When the RecBCD-treated Ɵ complex (Fig. 6A, lane 2; +19 primer-extension product) was incubated with cell extracts from an uninfected host and analyzed by primer extension, a shorter +4 product was observed (lane 3). This product was also generated by incubating the complex directly (i.e., without prior RecBCD treatment) with cell extracts (which have RecBCD) prepared from uninfected (lane 4) or Mu-infected cells (lane 5). Extracts prepared from a ClpX mutant still generated the +4 product (lane 6), showing that this reaction is independent of ClpX. Extracts from a RecBCD mutant did not generate the +4 product (lane 7), suggesting that the FD has to be recessed to +19 by RecBCD before formation of the +4 product.

Fig. 6.

Fig. 6.

Open in a new tab

Host factor stimulates cleavage within the transpososome-protected +19 RecBCD product. (A) Primer-extension analysis of the Ɵ complex. Lane 1: 0 length marker as in Fig 5. The XmnI-cut Ɵ complex was incubated with (lanes 2 and 3) or without (lanes 4–7) RecBCD, followed by incubation with cell extracts from uninfected, infected, or mutant (clpX, recBCD) strains. Primer extension was performed for the L end. The +4 product was detected in all reactions with RecBCD, whether added as pure enzyme (lane 3) or in wild-type cell extracts (lanes 4–6), but not in a recBCD mutant (lane 7). (B) Linearized Ɵ complex assembled with either wild-type MuA (lane 3) or a MuA BAN domain mutant, RQRQQ (lane 4), and subjected to primer extension after treatment with wild-type cell extracts. Lanes 1 and 2 are as in A. (C, Left) The XmnI-cut Ɵ complex incubated without (lane 1) or with (lane 2) wild-type cell extracts, resolved by agarose gel electrophoresis. Lanes 3 and 4 are similar to lanes 1 and 2, except that the reactions were treated with 1% SDS. The XmnI reaction did not go to completion, as seen in lane 2, and as reflected in the residual Ɵ topoisomers after SDS treatment in lanes 3 and 4. The gel was stained with ethidium bromide (Left) or immunoblotted with anti-MuA antibodies (Right). Other symbols are as in Fig. 3.

To test the role of MuANuc in the +4 reaction, a MuA mutant (RQRQQ) defective for both BAN region activities (nonspecific DNA binding and MuANuc), which was reported to be defective for FD removal in vivo (15), was tested. When the linearized Ɵ complex assembled from this mutant was incubated with wild-type cell extracts, it yielded a +4 product similar to that generated by a wild-type MuA complex (Fig. 6B, lanes 3 and 4). Because the RQRQQ mutant was defective for both BAN activities, that is, DNA binding as well as MuANuc (15), we considered the possibility that the primary defect in this mutant was loss of DNA binding, and that the nuclease activity resided elsewhere in the BAN peptide (residues 575–600) (25). We therefore targeted serine, threonine, aspartic acid, and glutamic acid residues in this region for mutagenesis, because of their potential to either serve as nucleophiles or coordinate metal ions that would facilitate hydrolysis of the phosphodiester bond. Both BAN activities were assayed in purified C-terminal domains of MuA bearing these mutations, because these activities are cryptic and not observed in a full-length protein (15, 25). Of the following single and double mutants generated—S583A, T585A, S583A T585A, and D596A E599A—none were defective for either BAN activity. They were therefore not tested further in full-length proteins.

We expect the cell extract-dependent +4 product to be generated within the transpososome bound to the +19 product (Fig. 3, lane 7; S complex), not only because the +4 position is within the transpososome-protected FD but also because DNA not bound stably to the transpososome will be completely degraded by RecBCD in these extracts (Fig. 4). To ascertain that the cell extract-treated S product still has bound MuA, the reaction was electrophoresed on agarose gels (Fig. 6C, Left), transferred to a PVDF membrane, and probed with MuA antibody (Fig. 6C, Right). The Western blot showed that MuA was associated with the Ɵ complex as expected (lanes 1 and 2), and also with the cell extract-generated S product (lane 2) before SDS treatment (compare lanes 1 and 2 with lanes 3 and 4).

We conclude that the +4 FD cleavage takes place within the transpososome. The cleavage activity is observed in a transpososome assembled using the RQRQQ MuA mutant, which does not display the BAN activity observed in an isolated C-terminal domain. We reconcile these apparently conflicting data in Discussion. We call the +4 cleavage activity of the cell extracts “CSF,” for cleavage stimulating factor.

CSF Activity Is Present in Yeast and Mammalian Cell Extracts.

Attempts to purify the CSF from E. coli extracts were unsuccessful. The activity was lost by heating at 75 °C for 10 min but present after RNase treatment, suggesting that a protein factor(s) is likely involved (Fig. S3). Mass spectrometry analysis of the +4 activity that came through several fractionation columns (SI Methods) yielded a large number of proteins. Nonetheless, we tested some candidate proteins common to three purification attempts, such as DnaK, GroEL, Elongation factor G, and RNA polymerase, alone and together, without success. ClpX alone is not a candidate for this factor (Fig. 6A, lane 6). We wondered whether these data reflected stimulation of an endonucleolytic activity within the transpososome by a macromolecular crowding effect, where high protein concentrations reduce the volume of solvent available, increasing the effective concentration of the reactive proteins. Such an effect can also be achieved by adding polymers such as polyethylene glycol or polycations such as spermine. However, neither of these reagents promoted formation of the +4 product. To test whether other cell extracts could substitute for E. coli extracts, we prepared these from the yeast Saccharomyces cerevisiae and from human embryonic kidney 293 cells. Interestingly, both extracts shortened the +19 RecBCD product (Fig. 7). The length of this product was slightly larger than +4 with yeast cell extracts (Fig. 7, lanes 2 and 3) and slightly smaller than +4 with human cell extracts (Fig. 7, lanes 4 and 5). The significance of these results is discussed below.

Fig. 7.

Fig. 7.

Open in a new tab

CSF activity in yeast and human cell extracts. The +19 RecBCD product was treated with the cell extracts prepared from E. coli (BU1384; lane 1), yeast (S. cerevisiae; lanes 2 and 3), or human embryonic kidney 293 cells (lanes 4 and 5) for 30 min at 37 °C. Primer extension was performed as described in Fig 5.

Discussion

This study provides new insights into the repair of Mu DNA insertions, and offers a solution to the long-standing question of how a common strand-transfer intermediate is resolved differentially by either repair or replication: The choice appears to be related to the linear DNA ends flanking the infecting genome, which are normally protected by the N protein. Our data suggest that N is removed after integration, and that the timing of its removal is controlled by the transpososome. RecBCD degrades the bulk of the FD, and this action is required for initiating endonucleolytic events that leave short 4-nt flanks within the transpososome-protected DNA. Earlier in vivo data showed a requirement for the ClpX chaperone as well as the BAN activity of MuA for removal of the FD (15); these requirements were not seen in vitro. We discuss these results below, and propose a model that reconciles the in vivo and in vitro data (Fig. 8).

Fig. 8.

Fig. 8.

Open in a new tab

Model for repair of Mu insertions. The model is based on synthesis of in vivo and in vitro data. After strand transfer, MuA transpososome (BAN region) and ClpX interact mutually or with host factors to control N removal. RecBCD degrades the FD until it reaches the transpososome but cannot get past it. The +19 bp of FD that remain untouched at both ends are shortened to +4 by endonucleolytic cleavage (MuANuc?; see text for alternate candidates) that depends on host factors (CSF). A polymerase/ligase must complete gap repair (not shown). The transpososome (orange oval) is drawn separately on the two ends for clarity; in reality, the two ends are in proximity.

RecBCD Access and the Mu Transpososome.

When integration of infecting Mu is blocked, the unintegrated N-linked Mu genome is indefinitely stable (5, 6, 26). Thus, the FD is not degraded before integration, a sensible outcome for the infecting genome, which depends on integration for survival. Our finding that RecBCD is required for removal of this DNA after integration (Fig. 2) suggests that postintegration signals control RecBCD entry past the N-protein block. What could be the nature of these signals? Previous work showed that integration of Mu in a clpX mutant delayed removal of the FD for at least an hour (Fig. S1) (15). ClpX, a chaperone that recognizes specific protein sequences (27), interacts with MuA in a strand-transfer transpososome and remodels it for transition to replication (10, 28). It is therefore plausible that ClpX bound at the transpososome, or a ClpX-remodeled transpososome, or proteins recruited to the remodeled transpososome, can remove the N protein by either destabilizing or degrading it.

FD was also stable when the infecting DNA encoded MuA variants carrying several substitutions in a patch of basic residues (RRRQK) within the C-terminal BAN region (15). Earlier, we had interpreted these results to suggest that removal of the FD is initiated by the endonucleolytic activity of MuANuc (Fig. 1B). In light of our findings that RecBCD is required for degradation of this DNA despite the presence of wild-type MuANuc activity (Fig. 2), we reinterpret our earlier findings to suggest that the BAN domain of MuA also plays a critical role in the strand-transfer configuration of the transpososome important for N removal. This role could be direct or manifest after ClpX remodeling. For example, mutations in the basic residue patch could be impaired for conformational changes in MuA that regulate N removal.

In summary, the in vivo data implicate the transpososome in authorizing RecBCD access into the FD.

RecBCD Arrest by the Mu Transpososome: A Role for MuANuc Activity?

When the Mu Ɵ intermediate formed in vitro was linearized within the FD, RecBCD degraded the DNA, revealing several interesting properties of the strand-transfer transpososome (Fig. 3). First, RecBCD activity was independent of ClpX (Fig. 6). As suggested above, if the in vivo role for ClpX is to regulate the removal of N, then N is already absent in vitro. Second, RecBCD progress through the FD was arrested by the transpososome, as indicated by the residual +19 bp of flanking sequences (Fig. 5). RecBCD is a powerful motor, which strips off a variety of DNA-bound protein complexes including transcribing RNA polymerases and nucleosomes (22). The arrest of RecBCD by the Mu strand-transfer transpososome is consistent with its extraordinary stability (29). Destabilization of the transpososome with ClpX was sufficient for RecBCD to push through the bound transpososome and degrade the entire DNA (Fig. 4). Such an action in vivo is likely prevented by host factors. Third, a heat-labile activity in cell extracts (Fig. S3) promoted +4 cleavage within the FD bound by the transpososome, only when this DNA was first shortened to +19 bp by RecBCD (Fig. 6 A and C). The +4 position is near the BAN region in the crystal structure of a strand-transfer transpososome (30). However, it remains to be determined whether MuANuc is responsible for the +4 cleavage, whether a host activity is involved, or whether the catalytic DDE residues in MuA perform this function. Similarly, whether the +4 product is the final substrate for gap repair, or whether RecBCD has partial access into a ClpX-remodeled transpososome in vivo such that the final gap-repair substrate has no 5′ overhangs, are questions for the future. In any event, the shortened flanks likely destroy the binding site for restart replication proteins, favoring nonreplicative transposition (10).

Why is the +4 product dependent on RecBCD, and why is there variability in its length depending on the source of the cell extracts—E. coli, yeast, or human (Figs. 6 and 7)? The FD region immediately flanking the Mu junction is single-stranded in an assembled transpososome (31). With RecBCD pushing on this DNA with great force, conformational changes transmitted along the DNA, coupled with the flexibility of single-stranded DNA, may allow this DNA to make alternate contacts with the endonuclease (32–34).

A Model for Repair of Mu DNA Insertions.

Combining results in this study with those reported earlier, we present a scheme for how a Mu insertion is repaired in vivo and consider the broader implications of our findings.

Shortening the long FD: Similarity of the short Mu flanks to Tn7 and retrotransposon flanks.

After integration, the transpososome configuration, modulated perhaps by ClpX and/or host factors, signals that the N protein be removed (Fig. 8). RecBCD gains entry and shortens the FD in two stages. RecBCD first degrades the bulk of the DNA but is intercepted by the transpososome, leaving 19 bp of this DNA untouched. A cellular CSF activity acts on the +19 RecBCD product to stimulate cleavage within the transpososome-bound DNA, leaving +4 overhangs. The similarity of the +4 Mu overhangs to the +3 or +2 overhangs of the 5′ FD in strand-transfer intermediates of Tn7 (35), or retroviral (36) and retrotransposon DNA (37), is striking. Although generated by different mechanisms in each element, the overhangs will encounter a common fate of being removed during repair of the gaps flanking the insertion. Could the similarities be indicative of a shared gap-repair mechanism?

Gap repair.

It is widely assumed that gap-filling polymerases repair the gaps at transposon ends (38, 39). However, recovery of Mu insertions is not dependent on the gap-filling polymerase PolA in E. coli (14). Our current thinking is that Pol III might fill the gaps. In this scenario, a replication fork arriving at a Mu gap would stall, filling the gap on the leading strand but generating a DSB on the lagging strand; this break must be repaired to recover viable cells, explaining the observed requirement for the DSB repair functions (recA, recB, recC, priA, and dnaT) in the recovery of simple Mu insertions (14). We note that the cellular checkpoint protein ATR, which is activated by stalled replication forks, has been implicated in retroviral insertion repair (40). Mu and retroviruses share a similar transposition mechanism (2, 41). Although Mu is unusual in having long flanks that must be degraded, the final substrate for repair resembles that of other transposons, as discussed above. Knowledge of Mu gap repair should therefore be informative for all transposons.

Our work demonstrates that RecBCD plays two roles in the repair of Mu insertions. It not only degrades the FD but is also part of the DSB repair machinery (RecABC, PriA, DnaT) required for recovery of Mu insertions.

Why Mu Uses Two Pathways to Resolve Integration Events.

The ability of RecBCD mutants, as well as some MuANuc mutants, to enter lytic growth (Fig. S2) (15) reveals that the absence of FD removal in these mutants does not prevent Mu replication, suggesting that their N-linked replicas are stable. This means that homologous recombination between two N-linked copies of Mu would be a plausible mechanism for generating a simple insertion (Fig. S4), but this does not happen (9). Why not? Why instead has Mu evolved an elaborate mechanism for removal and repair of the FD? We suggest that even one round of replication would commit Mu to the lytic cycle. Limiting Mu replication to short gaps at the Mu ends gives the phage a chance to enter a prophage state.

Methods

Details of materials and methods can be found in SI Methods. These describe strain construction, preparation of cell extracts, and protein purification. In vivo and in vitro assays for transposition and FD removal and primer extension analysis are also described. Plasmids are listed in Table S1.

Supplementary Material

Supplementary File
pnas.201407562SI.pdf (412.5KB, pdf)

Acknowledgments

We thank Alexandra Grand for construction of the BAN region mutants of MuA, Rudra Saha for help with illustrations, and Makkuni Jayaram and Lynn Zechiedrich and her colleagues for discussions. This work was supported by National Institutes of Health Grant GM33247 and in part by Robert Welch Foundation Grant F-1351.

Footnotes

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1407562111/-/DCSupplemental.

References

  • 1.Symonds N, Toussaint A, Van de Putte P, Howe MM. Phage Mu. Cold Spring Harbor, NY: Cold Spring Harbor Lab Press; 1987. [Google Scholar]
  • 2.Mizuuchi K. Transpositional recombination: Mechanistic insights from studies of Mu and other elements. Annu Rev Biochem. 1992;61:1011–1051. doi: 10.1146/annurev.bi.61.070192.005051. [DOI] [PubMed] [Google Scholar]
  • 3.Chaconas G, Harshey RM. Transposition of phage Mu DNA. In: Craig NL, Craigie R, Gellert M, Lambowitz AM, editors. Mobile DNA II. Washington, DC: ASM; 2002. pp. 384–402. [Google Scholar]
  • 4.Harshey RM. The Mu story: How a maverick phage moved the field forward. Mob DNA. 2012;3(1):21. doi: 10.1186/1759-8753-3-21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Harshey RM, Bukhari AI. Infecting bacteriophage Mu DNA forms a circular DNA-protein complex. J Mol Biol. 1983;167(2):427–441. doi: 10.1016/s0022-2836(83)80343-x. [DOI] [PubMed] [Google Scholar]
  • 6.Puspurs AH, Trun NJ, Reeve JN. Bacteriophage Mu DNA circularizes following infection of Escherichia coli. EMBO J. 1983;2(3):345–352. doi: 10.1002/j.1460-2075.1983.tb01429.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Gloor G, Chaconas G. Sequence of bacteriophage Mu N and P genes. Nucleic Acids Res. 1988;16(11):5211–5212. doi: 10.1093/nar/16.11.5211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Au TK, Agrawal P, Harshey RM. Chromosomal integration mechanism of infecting Mu virion DNA. J Bacteriol. 2006;188(5):1829–1834. doi: 10.1128/JB.188.5.1829-1834.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Chaconas G, Kennedy DL, Evans D. Predominant integration end products of infecting bacteriophage Mu DNA are simple insertions with no preference for integration of either Mu DNA strand. Virology. 1983;128(1):48–59. doi: 10.1016/0042-6822(83)90317-3. [DOI] [PubMed] [Google Scholar]
  • 10.Nakai H, Doseeva V, Jones JM. Handoff from recombinase to replisome: Insights from transposition. Proc Natl Acad Sci USA. 2001;98(15):8247–8254. doi: 10.1073/pnas.111007898. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Harshey RM. Transposition without duplication of infecting bacteriophage Mu DNA. Nature. 1984;311(5986):580–581. doi: 10.1038/311580a0. [DOI] [PubMed] [Google Scholar]
  • 12.North SH, Nakai H. Host factors that promote transpososome disassembly and the PriA-PriC pathway for restart primosome assembly. Mol Microbiol. 2005;56(6):1601–1616. doi: 10.1111/j.1365-2958.2005.04639.x. [DOI] [PubMed] [Google Scholar]
  • 13.North SH, Kirtland SE, Nakai H. Translation factor IF2 at the interface of transposition and replication by the PriA-PriC pathway. Mol Microbiol. 2007;66(6):1566–1578. doi: 10.1111/j.1365-2958.2007.06022.x. [DOI] [PubMed] [Google Scholar]
  • 14.Jang S, Sandler SJ, Harshey RM. Mu insertions are repaired by the double-strand break repair pathway of Escherichia coli. PLoS Genet. 2012;8(4):e1002642. doi: 10.1371/journal.pgen.1002642. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Choi W, Harshey RM. DNA repair by the cryptic endonuclease activity of Mu transposase. Proc Natl Acad Sci USA. 2010;107(22):10014–10019. doi: 10.1073/pnas.0912615107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Amundsen SK, Taylor AF, Chaudhury AM, Smith GR. recD: The gene for an essential third subunit of exonuclease V. Proc Natl Acad Sci USA. 1986;83(15):5558–5562. doi: 10.1073/pnas.83.15.5558. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Roman LJ, Eggleston AK, Kowalczykowski SC. Processivity of the DNA helicase activity of Escherichia coli recBCD enzyme. J Biol Chem. 1992;267(6):4207–4214. [PubMed] [Google Scholar]
  • 18.Lovett ST, Luisi-DeLuca C, Kolodner RD. The genetic dependence of recombination in recD mutants of Escherichia coli. Genetics. 1988;120(1):37–45. doi: 10.1093/genetics/120.1.37. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Palas KM, Kushner SR. Biochemical and physical characterization of exonuclease V from Escherichia coli. Comparison of the catalytic activities of the RecBC and RecBCD enzymes. J Biol Chem. 1990;265(6):3447–3454. [PubMed] [Google Scholar]
  • 20.Yu M, Souaya J, Julin DA. Identification of the nuclease active site in the multifunctional RecBCD enzyme by creation of a chimeric enzyme. J Mol Biol. 1998;283(4):797–808. doi: 10.1006/jmbi.1998.2127. [DOI] [PubMed] [Google Scholar]
  • 21.Kuzminov A. Recombinational repair of DNA damage in Escherichia coli and bacteriophage lambda. Microbiol Mol Biol Rev. 1999;63(4):751–813. doi: 10.1128/mmbr.63.4.751-813.1999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Finkelstein IJ, Visnapuu ML, Greene EC. Single-molecule imaging reveals mechanisms of protein disruption by a DNA translocase. Nature. 2010;468(7326):983–987. doi: 10.1038/nature09561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Lavoie BD, Chan BS, Allison RG, Chaconas G. Structural aspects of a higher order nucleoprotein complex: Induction of an altered DNA structure at the Mu-host junction of the Mu type 1 transpososome. EMBO J. 1991;10(10):3051–3059. doi: 10.1002/j.1460-2075.1991.tb07856.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Mizuuchi M, Baker TA, Mizuuchi K. DNase protection analysis of the stable synaptic complexes involved in Mu transposition. Proc Natl Acad Sci USA. 1991;88(20):9031–9035. doi: 10.1073/pnas.88.20.9031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Wu Z, Chaconas G. A novel DNA binding and nuclease activity in domain III of Mu transposase: Evidence for a catalytic region involved in donor cleavage. EMBO J. 1995;14(15):3835–3843. doi: 10.1002/j.1460-2075.1995.tb00053.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Gloor G, Chaconas G. The bacteriophage Mu N gene encodes the 64-kDa virion protein which is injected with, and circularizes, infecting Mu DNA. J Biol Chem. 1986;261(35):16682–16688. [PubMed] [Google Scholar]
  • 27.Baker TA, Sauer RT. ClpXP, an ATP-powered unfolding and protein-degradation machine. Biochim Biophys Acta. 2012;1823(1):15–28. doi: 10.1016/j.bbamcr.2011.06.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Abdelhakim AH, Sauer RT, Baker TA. The AAA+ ClpX machine unfolds a keystone subunit to remodel the Mu transpososome. Proc Natl Acad Sci USA. 2010;107(6):2437–2442. doi: 10.1073/pnas.0910905106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Surette MG, Buch SJ, Chaconas G. Transpososomes: Stable protein-DNA complexes involved in the in vitro transposition of bacteriophage Mu DNA. Cell. 1987;49(2):253–262. doi: 10.1016/0092-8674(87)90566-6. [DOI] [PubMed] [Google Scholar]
  • 30.Montaño SP, Pigli YZ, Rice PA. The μ transpososome structure sheds light on DDE recombinase evolution. Nature. 2012;491(7424):413–417. doi: 10.1038/nature11602. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Wang Z, Namgoong SY, Zhang X, Harshey RM. Kinetic and structural probing of the precleavage synaptic complex (type 0) formed during phage Mu transposition. Action of metal ions and reagents specific to single-stranded DNA. J Biol Chem. 1996;271(16):9619–9626. doi: 10.1074/jbc.271.16.9619. [DOI] [PubMed] [Google Scholar]
  • 32.Kouzine F, Sanford S, Elisha-Feil Z, Levens D. The functional response of upstream DNA to dynamic supercoiling in vivo. Nat Struct Mol Biol. 2008;15(2):146–154. doi: 10.1038/nsmb.1372. [DOI] [PubMed] [Google Scholar]
  • 33.Lavelle C. DNA torsional stress propagates through chromatin fiber and participates in transcriptional regulation. Nat Struct Mol Biol. 2008;15(2):123–125. doi: 10.1038/nsmb0208-123. [DOI] [PubMed] [Google Scholar]
  • 34.Fogg JM, et al. Bullied no more: When and how DNA shoves proteins around. Q Rev Biophys. 2012;45(3):257–299. doi: 10.1017/S0033583512000054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Bainton R, Gamas P, Craig NL. Tn7 transposition in vitro proceeds through an excised transposon intermediate generated by staggered breaks in DNA. Cell. 1991;65(5):805–816. doi: 10.1016/0092-8674(91)90388-f. [DOI] [PubMed] [Google Scholar]
  • 36.Craigie R. Retroviral DNA integration. In: Craig NL, Craigie R, Gellert M, Lambowitz AM, editors. Mobile DNA II. Washington, DC: ASM; 2002. pp. 613–630. [Google Scholar]
  • 37.Sandmeyer SB, Aye M, Menees T. Ty3, a position-specific, gypsy-like element in Saccharomyces cerevisiae. In: Craig NL, Craigie R, Gellert M, Lambowitz AM, editors. Mobile DNA II. Washington, DC: ASM; 2002. pp. 663–683. [Google Scholar]
  • 38.Syvanen M, Hopkins JD, Clements M. A new class of mutants in DNA polymerase I that affects gene transposition. J Mol Biol. 1982;158(2):203–212. doi: 10.1016/0022-2836(82)90429-6. [DOI] [PubMed] [Google Scholar]
  • 39.Craig NL, Craigie R, Gellert M, Lambowitz AM, editors. Mobile DNA II. Washington, DC: ASM; 2002. [Google Scholar]
  • 40.Yang YX, Guen V, Richard J, Cohen EA, Berthoux L. Cell context-dependent involvement of ATR in early stages of retroviral replication. Virology. 2010;396(2):272–279. doi: 10.1016/j.virol.2009.10.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Haren L, Ton-Hoang B, Chandler M. Integrating DNA: Transposases and retroviral integrases. Annu Rev Microbiol. 1999;53:245–281. doi: 10.1146/annurev.micro.53.1.245. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary File
pnas.201407562SI.pdf (412.5KB, pdf)

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES