Controlling DNA degradation from a distance: a new role for the Mu transposition enhancer

Summary

Phage Mu is unique among transposable elements in employing a transposition enhancer. The enhancer DNA segment is the site where the transposase MuA binds and makes bridging interactions with the two Mu ends, interwrapping the ends with the enhancer in a complex topology essential for assembling a catalytically active transpososome. The enhancer is also the site at which regulatory proteins control divergent transcription of genes that determine the phage lysis-lysogeny decision. Here we report a third function for the enhancer - that of regulating degradation of extraneous DNA attached to both ends of infecting Mu. This DNA is protected from nucleases by a phage protein until Mu integrates into the host chromosome, after which it is rapidly degraded. We find that leftward transcription at the enhancer, expected to disrupt its topology within the transpososome, blocks degradation of this DNA. Disruption of the enhancer would lead to the loss or dislocation of two non-catalytic MuA subunits positioned in the transpososome by the enhancer. We provide several lines of support for this inference, and conclude that these subunits are important for activating degradation of the flanking DNA. This work also reveals a role for enhancer topology in phage development.

Introduction

Temperate phage Mu and its relative D108 are the only transposable elements known to require a transposition enhancer (E), which is located at a distance of 1 kb from the left (L) end of their ~37 kb genomes (Chaconas & Harshey, 2002, Morgan et al., 2002) (Fig. 1A). The enhancer is required for transpososome assembly in vitro, and is also the hub of transcriptional activity in vivo, where regulatory proteins bind to control the lysis-lysogeny decision. Indeed, the ~200 bp E segment was first described as an operator (O) DNA region where the lysogenic repressor Rep, and a λ Cro-like protein Ner, battle for control of phage development (Goosen & van de Putte, 1987). The O/E segment comprises three large binding sites - O1, O2 and O3 – for both Rep and the transposase MuA (Fig. 1A, lower panel) (Krause & Higgins, 1986, Leung et al., 1989, Mizuuchi & Mizuuchi, 1989). Two divergent early promoters Pe and Pc whose transcripts overlap, originate within O2 and O3, respectively. Rep is the product of the c gene, synthesized from the leftward Pc transcript; Pc transcription favors lysogeny (Goosen & van de Putte, 1987). Ner, synthesized early from the rightward Pe transcript, binds between Pe and Pc to inhibit Pc and allow Pe transcription (Van Leerdam et al., 1982, Tolias & Dubow, 1986, Goosen & van de Putte, 1987, Strzelecka et al., 1995). The Pe transcript is long, and encodes in addition to ner, the transposition functions A and B, as well as several other genes in the SE (semi-essential) region, not all of whose functions are known (Paolozzi & Symonds, 1987, Morgan et al., 2002); Pe transcription favors lytic development (Fig. 1A). Ner and Rep also autoregulate their own synthesis to maintain lytic or lysogenic states, respectively. The IHF protein binds between O1 and O2 to enhance Pe and depress Pc transcription (Goosen et al., 1984, Krause & Higgins, 1986, van Rijn et al., 1988, Higgins et al., 1989, Alazard et al., 1992, Betermier et al., 1995, Rousseau et al., 1996).

Fig. 1.

Regulation of transcription and transposition by the Mu enhancer and the non-replicative (repair) pathway Mu integration. A. Top: Disposition of relevant genes and protein binding sites on the Mu genome, linked to variable flanking DNA (FD) as packaged in the phage. Middle: The operator (O) region controls transcription from two divergent promoters Pc and Pe. The lysogenic repressor Rep (c gene product), transcribed from Pc, binds to sites O1–O3. Ner, made from the Pe transcript, binds between O2–O3 (stippled boxes in expanded region). IHF binds between O1 and O2. The O sites also bind transposase MuA and serve an essential enhancer (E) function in transpososome assembly. MuA also binds L1 to L3 and R1 to R3 at the L/R ends. HU binds between L1 and L2. Bottom: The −10 regions of Pe and Pc are indicated; the red arrow indicates the start of the Pc deletion in strain CW172 used later this study. B. Left: Topology of the LER transpososome. MuA bridging interactions between the E and L/R sites, assisted by HU and IHF, result in a specific 5-noded topology. Right: MuA subunits are shown as orange ovals. The catalytic subunits at L1 and R1 associated strongly with those at L2 and R2 to form a stable tetrameric core, which holds the nodes indicated by black dots with white circles (left). The E-R crossing indicated with a black dot is held by the MuA subunit at R3. The single E-L crossing is likely contributed by the subunit at L3. The L3, R3 subunits are loosely positioned. C. Non-replicative or repair pathway of Mu integration. The FD attached to infecting Mu (see A) is protected by Mu N protein, which circularizes the DNA non-covalently. After integration (strand transfer), N is removed and the FD is degraded by RecBCD in a process that requires ClpX. In vitro, the final product of RecBCD degradation is +4 nucleotides (Choi et al., 2014). The short-flank product is likely the substrate for the final steps in repair, where the 5 bp target gaps generated after strand transfer are filled and ligated to yield a simple insertion. The transpososome is shown as an orange oval. See text for details.

In the context of transposition, the O site is referred to as E for ‘enhancer’ because it enhances transposition over 100-fold in vivo and is absolutely essential for assembly of the transpososome in vitro (Leung et al., 1989, Mizuuchi & Mizuuchi, 1989, Chaconas & Harshey, 2002). MuA has distinct domains for binding both the O as well as the end sites at L (L1–L3) and R (R1–R3) (Fig. 1A), and makes bridging interactions between specific pairs of E and L/R sites (Allison & Chaconas, 1992, Jiang et al., 1999). On supercoiled DNA, E is critical for MuA-mediated synapsis of the L and R Mu ends, where an initial ER synapse captures L to form an LER synapse that has 5 supercoils trapped within it (Fig. 1B, left) (Pathania et al., 2002, Yin et al., 2007, Harshey & Jayaram, 2006). IHF plays an important role at the enhancer as well, introducing a bend between O1 and O2 to optimize LER interactions; HU does the same between L1 and L2 (Surette & Chaconas, 1989, Surette & Chaconas, 1992, Allison & Chaconas, 1992, Lavoie & Chaconas, 1993). Once the transpososome is assembled, the enhancer is not required for the cutting and joining reactions of transposition in vitro (Mizuuchi et al., 1992, Surette & Chaconas, 1992, Kobryn et al., 2002), yet remains associated with the transpososome (Pathania et al., 2003). Several studies have deduced that in the 6-subunit transpososome, the catalytic MuA subunits occupy L1 and R1, and that these subunits form a stable tetrameric complex with the neighboring subunits at L2 and R2; it follows that the loosely held subunits occupy L3 and R3 (Fig. 1B, right) (Chaconas & Harshey, 2002, Harshey & Jayaram, 2006, Montano et al., 2012). Topological studies suggest that the L3/R3 subunits depend on E for their continued association with the transpososome (Harshey & Jayaram, 2006, Yin et al., 2007).

We show in this study that O/E has a third role in controlling removal of the extraneous DNA attached to the L and R ends, during post-integration repair of the infecting Mu genome. This DNA is derived from packaging host chromosomal DNA linked to either side of Mu during the previous round of lytic growth (Fig. S1) (Symonds et al., 1987). 60 – 150 bp of host sequences flank the L end of Mu (George & Bukhari, 1981), and 0.5 – 3 kb flank the R end (Fig. 1A) (Howe, 1987). Upon infection of a new host, the linear genome circularizes non-covalently with assistance from an injected phage protein N, which binds the flanking DNA (FD) termini and protects them from degradation (Fig. S1) (Harshey & Bukhari, 1983, Puspurs et al., 1983, Gloor & Chaconas, 1988). After the N-protected Mu integrates into the host chromosome, the FD is degraded by RecBCD exonuclease (Figs. 1C and S1) (Chaconas et al., 1983, Au et al., 2006, Choi et al., 2014). Since RecBCD only acts on linear DNA ends, post-integration events must remove the N protein. Some, but not all players in this process have been identified in vivo. We know that the transposase/transpososome is involved because a patch of basic residues within the C-terminal domain of MuA that controls a cryptic nuclease activity (Wu & Chaconas, 1995), is required for FD removal, and because removal is delayed in an E. coli clpX mutant (Choi & Harshey, 2010). ClpX interacts with the C-terminus of MuA in a strand-transfer transpososome (i.e. the MuA complex in which the Mu ends are joined to the target), and remodels it in preparation for replication (Kruklitis et al., 1996, Levchenko et al., 1997, Nakai et al., 2001, Abdelhakim et al., 2010). The shortened FD is eventually repaired, resulting in a simple Mu insertion (Fig. 1C).

This study shows that the enhancer configuration in the transpososome is important for repair of the Mu integrant. We demonstrate that the enhancer is required in vivo for integration of infecting Mu into the E. coli chromosome, and that it plays a post-strand transfer role in FD degradation. This new role for the enhancer was first revealed by the finding that leftward transcription from Pc, but not the product of the Pc transcript (Rep), blocks FD degradation. We surmised that Pc transcription disrupts a critical E-R contact within the strand transfer transpososome, thereby disrupting the native configuration of the enhancer. We have tested this deduction in several different ways and conclude that two non-catalytic MuA subunits held in the complex by the enhancer are important for the post-integration events that control RecBCD entry into the FD.

Results

The enhancer is required for integration of infecting Mu DNA

The importance of E for transposition has been demonstrated using mini-Mu plasmids, but has not been critically tested for the whole phage genome because the E and O functions are inseparable. To determine whether E is required for integration of infecting Mu, we exploited the transposase of the Mu-related phage D108 (D108 A). Mu and D108 phages differ mainly in their operator/enhancer (O/E) region. They also differ in their Rep proteins (product of c gene; Fig. 1A) and in the Iα domain of their transposases which bind O/E (Fig. 2A) (Toussaint et al., 1983, Mizuuchi et al., 1986, Jiang et al., 1999). The importance of O/E for lytic growth has been demonstrated by isolation of hybrids between Mu and D108, which formed plaques only if the Iα region of the transposase matched its cognate E (Toussaint et al., 1983). The hybrid phages had retained the O-binding Rep and Ner proteins as well, so both O and E functions were contributing to growth of the hybrids. To determine if the E function alone was required for integration of infecting Mu, integration-defective phage carrying an amber mutation in the transposase gene A (Aam1093) (O'Day et al., 1978), were used to infect a sup⁻host, complemented for integration by either MuA or D108 A supplied from a plasmid. Thus, the infecting phage have their cognate Rep(O) function, but are being queried with regard to A(E) function by providing A proteins that differ only in the E-binding Iα domains. The functionality of D108 A expressed from a plasmid was ascertained by complementation of a D108 A⁻ lysogen for prophage induction and lytic growth only when D108 A, and not MuA, was provided (Fig. S2). If the E function is dispensable for Mu integration, as it is during enhancer-independent transposition reaction conditions in vitro (Chaconas & Harshey, 2002), we would expect either MuA or D108 A to support integration of the A amber Mu phage, because the rest of the protein (Iβγ, catalytic domain II and regulatory domain III; Fig. 2B) is similar in the two transposases and promotes catalysis using either Mu or D108 ends (Toussaint et al., 1983); if the E function is not dispensable, integration should be MuA-specific. The results are shown in Fig. 2B. At various times after infection, total cellular DNA was isolated and subjected to pulse field gel electrophoresis (PFGE) to separate chromosomally integrated Mu from free Mu DNA; the isolated chromosomal DNA was tested for Mu integration by PCR, using Mu-specific primers (Au et al., 2006). When the complementing transposase was MuA, the Mu Aam phage integrated into the host chromosome with kinetics similar to that observed for wild type phage, where the Mu integration signal is normally detected within 15 min after infection, and increases thereafter due to Mu replication (see Fig. 3A top panel and Fig. 6; (Choi & Harshey, 2010)). When the complementing transposase was D108 A, however, Mu integration was not detected. We conclude that the E function, i.e. transpososome assembly function, is essential for integration of infecting Mu.

Fig. 2.

O/E binding specificities of Mu and D108 transposase A proteins, and requirement for E during Mu integration. A. Domain organization deduced for the transposase MuA, extrapolated to D108 A. Three major domains mapped by partial proteolysis (Nakayama et al., 1987), have been assigned DNA binding, Catalysis and Allosteric control functions (Chaconas & Harshey, 2002). Domain Iα binds three sites (O1-O3) in E, and Iβγ binds three sites within L and R (see Fig. 1A). DDE are the catalytic residues for transposition (Montano & Rice, 2011). Domain III has overlapping recognition sites for two regulatory proteins, ClpX and MuB (Levchenko et al., 1997). MuA and D108 A proteins differ mainly in their E binding Iα domains. See text. B. Integration of MuAam phage in a sup⁻host (BU40) at various times after infection, complemented by plasmids in the host strain expressing either wild type MuA or D108 A from pWY12 and pWY15, respectively, with the level of transposase expression that is attained in the absence of inducer IPTG (see Fig. S2). Genomic DNA was isolated as described in Methods, and Mu integration was detected by primers specific to Mu. N, no template control; M, input Mu; C, control genomic DNA isolated after PFGE by mixing with free Mu DNA prior to electrophoresis in order to gauge the extent of contamination of the genomic DNA with free Mu DNA (Au et al., 2006).

Fig. 3.

Role of regulatory proteins Rep and Ner in FD removal. A. Kinetics of detection of integrated Mu and FD (lacZ) sequences at two temperatures that inactivate the ts Rep. FDs were amplified across the Mu R end – lacZ junction as described in Methods. Other descriptions as in Fig. 2B. B. Strain BU1384 carrying three different plasmids was infected with ner⁻ phage (from CW54) at 37°C, with or without the indicated plasmid. pWY62 expresses Ner, A, and B from the natural promoter Pe. pIL137 expresses A and B from Ptet, but uninduced expression levels were sufficient to support Mu integration. Ner expression in pWY59 was induced with 0.1% arabinose for this plasmid. Other descriptions are as Fig. 2B.

Fig. 6.

FD removal is blocked in phages carrying deletion of R3 and L3 sites. Infection, DNA isolation and PCR assays were as described in Figs. 3–4, except that the FD sequence monitored was malF, because the original prophage was located in this gene in strain MP1999. The ΔR3 strain lyses slowly (Fig. S5), so FD removal could be monitored for a longer time in this strain.

Ner, but not Rep, is required for flanking DNA removal

Upon Mu infection, the FD signal is detected concomitantly with detection of integrated Mu, but diminishes significantly by 30 min when Mu is actively replicating, and is typically undetectable thereafter (Au et al., 2006, Choi & Harshey, 2010). Since FD removal is seen only after integration (strand transfer), it is possible that the configuration of the strand transfer transpososome must play a role in controlling the timing of degradation. Given that the enhancer remains associated with the strand transfer transpososome in vitro (Pathania et al., 2003), we wished to test if this association was important for FD removal, in which case proteins that bind O/E might influence this process (Fig. 1A,B). We therefore monitored the effects of inactivating Rep and Ner proteins in the FD removal assay as described below. An earlier study had shown that IHF activity was not important, since Mu infection in an IHF mutant strain showed normal FD removal kinetics (Au et al., 2006).

The methodology for detecting FD after integration of infecting Mu has been described (Au et al., 2006, Choi & Harshey, 2010). Briefly, lacZ sequences are used as FD markers. These sequences are enriched in progeny phage derived from induction of a Mu prophage integrated in lacZ. When a lac⁻ host is infected with these phage, lacZ FD linked to integrated Mu is readily detected by PCR. The FD at both Mu ends is degraded (Au et al., 2006), so either end can be monitored for its presence.

To test the role of Rep in FD removal, phage carrying the temperature-sensitive Rep allele, cts62, were used for infection. This Rep variant has reduced DNA-binding to O at all temperatures compared to the wild-type Rep, but the binding defect is most exacerbated at higher temperatures (Vogel et al., 1991). If Rep played a role in FD removal, we would expect to see differences in FD removal kinetics at higher temperatures. As observed earlier, both Mu and lacZ FD sequences were detected within 15 min of infection (Fig. 3A, compare left and right panels). While the Mu signal continued to increase during the time course monitored, the FD signal was maximal at 15 min, decreased by 30 min, and was undetectable by 50 min. Although the infected cultures lysed faster at 42°C, than at 37°C (Fig. S3A), the Mu and FD profiles were similar at the two temperatures. Thus, the inability of Rep to associate stably with DNA does not affect FD removal.

To assess the effect of Ner on FD removal, ner⁻ phage were prepared as described in Methods; Ner is required for lytic growth. Infection with these phage resulted in normal integration, even in the absence of Ner in the host (Fig. 3B; top ‘no vector’ panel, Mu), but the cells did not enter lytic growth (Fig. S3B), allowing observation of the FD for longer times. Interestingly, the FD signal persisted until the last time point examined (90 min; Fig. 3B, top panel, FD). To test if complementation with Ner would restore FD removal in ner⁻ infections, Ner was provided in the recipient host. Different plasmids were tested as the source of Ner because induction of ner⁻ prophage had been observed to be sensitive to Ner levels (see Methods). When ner, A and B were expressed from the natural Pe promoter (pWY62), FD was processed normally (Fig. 3B, second panel), and infected cells went through lytic growth (Fig. S3B). To separate the contributing effects of Ner from that of A and B proteins, FD removal in the ner⁻ Mu integrants was tested under three other conditions. When only A and B were provided i.e. no Ner (pIL137), the FD remained attached (Fig. 3B, third panel). When Ner alone was provided from the arabinose-inducible pBAD promoter (pWY59), FD removal depended on the induction level of Ner. In the absence of inducer, where there is leaky expression from this promoter (Guzman et al., 1995), a wild-type pattern of FD processing was observed (Fig. 3B, fourth panel); this level of Ner was not sufficient to promote lytic growth (Fig. S3B). Thus FD removal is not dependent on lytic growth. When Ner synthesis was induced with 0.1% arabinose, FD removal was inhibited (Fig. 3B, bottom panel). High levels of Ner are known to shut off Pe; as expected, cells did not progress through the lytic cycle (Fig. S3B) (Goosen & van de Putte, 1987, Van Leerdam et al., 1982). The absolute levels of Ner under all these conditions proved problematic to monitor. We therefore used Pc transcript levels as an indicator of Ner levels (Fig. S3C). The data were consistent with the known inhibitory effect of Ner on Pc transcription.

In summary, of the two transcription regulators Rep and Ner that function at O, Ner appears to play a role in the post-integration removal of the DNA flanking Mu ends. FD removal is sensitive to Ner levels, and is observed only when Ner levels are low; FD is not removed in the absence of Ner or when Ner levels are high.

Ner influences flanking DNA removal by modulating Pc transcription

Absence of Ner is expected to favor Pc transcription, generating Rep, which will bind O and block not only Pe transcription, but also E function (Fig. 1A) (Mizuuchi & Mizuuchi, 1989). However, the infection experiments are carried out at high temperature where Rep is non-functional, so a Rep effect can be ruled out. Transcription from Pc per se, i.e. in the absence of functional Rep, was reported to inhibit lytic growth in a ner mutant; this inhibition was relieved by deletion of Pc (Goosen & van de Putte, 1986). These data led us to test whether the effect of Ner on FD removal was related to Pc transcription. To do so, we constructed a prophage that inactivated both Pc and ner as described under Methods. The ΔPc Δner phage integrated as efficiently as the Δner phage under these conditions (Fig. 4, compare the Mu signal in the first and second panels), but went through lytic growth (Fig. S4; (Goosen & van de Putte, 1986)). There was a striking difference, however, in their kinetics of FD removal. Whereas the FD signal from integrated ner⁻phage persisted until the last time point tested (90 min), this signal from the integrated ΔPc Δner phage showed wild-type kinetics of disappearance. Thus, the deletion of Pc alleviated the FD removal defect caused by the deletion of ner. Therefore, absence of Ner inhibits FD removal by promoting Pc transcription.

Fig. 4.

FD removal in wild type host infected with ner⁻ phage or ΔPc ner⁻ phage. Infection conditions as in Fig. 3. The ΔPc Δner phage do not express B protein, which is essential for efficient integration (Chaconas & Harshey, 2002). MuB was therefore supplied in the recipient host from plasmid pIL137, which encodes both A and B; no inducer was added. To provide Ner, A and B, pWY96 and pIL137 were used. The expression of Ner was induced from pWY96 with 0.1 % arabinose. Other descriptions as in Fig. 2B.

A plausible explanation for why Pc transcription, but not the product of this transcription, would affect distant events at the FD is that this transcription interferes with E function by perturbing its topology within the transpososome. This would also explain why high levels of Ner, which are expected to block Pc transcription, would have the same effect (Fig. 3B, last panel). In this case, Ner binding to O would perturb E topology. The latter proposition can be tested in vitro, and is described below. The reason why similar levels of Ner did not inhibit FD removal in the ΔPc Δner phage infection (Fig. 4, last panel) might be attributed to deletion of DNA immediately adjacent to the Ner binding site between O2 and O3, which might have altered the stability of Ner on DNA (Fig. 1A, bottom panel).

Ner disrupts E-R crossings

In vitro experiments have deduced a 5-noded topology of the DNA segments bound by the transpososome (Harshey & Jayaram, 2006). Given the high degree of correspondence between in vitro and in vivo requirements for all the cis and trans elements/factors essential for transposition (Chaconas & Harshey, 2002), the topology of the transpososome, which is dictated by these elements/factors, is also expected to be similar in vivo. However, an in vivo variable is transcription across E, which would be expected to modulate the DNA crossings. Of the 5 DNA crossings between L, E and R, one is contributed by R3-E (Fig. 1B left, black dot) (Yin et al., 2007, Yin et al., 2005). Trancription from Pc, or Ner binding, could impact this crossing. The latter proposition was tested by adding Ner to the transpososome assembly reaction, and assessing the configuration of the R3-E crossing by difference topology using Cre recombinase. Two mini-Mu plasmids pSP(R)Dir and pSP(R)In, which differ only in the orientation of loxP sites flanking the R end, either direct (Dir) or inverted (In), were employed (Fig. 5). Transpososomes in which Mu ends had been cleaved (Type I reaction) were assembled, so the Cre recombination products would be naturally nicked; removal of supercoils is essential for analysis of the products by gel electrophoresis (Fig. 5A,B, lane 2). Under wild type conditions, the transpososomes will yield predominantly 4Cat and 5Knot products after Cre recombination on the Dir and In substrates, respectively, because R crosses E and L four times (Fig. 5 A,B, lanes 3; see schematic in this figure and also Fig. 1B) (Pathania et al., 2002); the extra crossing in the In substrate comes from the need to align loxP sites in an antiparallel configuration for recombination (Guo et al., 1997, Kilbride et al., 1999, Grainge et al., 2002). If Ner were to disrupt only the distal E-R crossing, Cre recombination would yield 4-Cat and 3-Knot products; if both E-R crossings were disrupted, Cre would give 2-Cat and 3-Knot products. The results show that addition of Ner gave 2-Cat and 3-Knot products (Fig. 5A,B, lanes 4 and 5), suggesting that Ner disrupted both E-R crossings, leaving the two L-R crossings intact. Ner altered the E-R topology only if it was included in the reaction from the start, and not if added after transpososome assembly. We conclude that Ner directly influences the integrity of the E-R crossings.

Fig. 5.

Effect of Ner on the E-R DNA crossings in an assembled MuA synapse. A and B (left panels) show products of Cre recombination on supercoiled pSP(R) Dir and In substrates, respectively, after assembling a Type I (cleaved) transpososome with MuA and HU proteins as described in Methods, in the presence or absence of Ner (lanes 4, 5, 6: 56, 28, 14 ug/ml Ner). Reactions were deproteinized before electrophoresis, as described in Methods. OC, open circular; SC, supercoiled. The numbers of the right indicate catenane or knot nodes in Dir and In substrates, respectively; these are derived from the known position of these bands characterized previously (Pathania et al., 2002, Yin et al., 2005). The band below the SC position on the Dir substrate is the larger of two deletion circles generated from the free plasmid (i.e. not converted to Type I) after Cre recombination. The right panels show a self-explanatory schematic tracking catenane and knot product formation after Cre recombination on an assembled Mu synapse.

Deletion of R3 or L3, which disrupt crossings with E, also block removal of flanking DNA

The results in Figs. 3–5 suggest that E-R interactions play a critical role in FD removal. To test this directly, we deleted the R3 site on the Mu genome. This site is the most distal of the three MuA binding sites at the R ends (Fig. 1A). As a control, we separately deleted the most distal site L3 at the L end; R3 and L3 sites cannot be simultaneously deleted (Allison & Chaconas, 1992). The lysis profiles of prophages carrying these single site deletions are shown in Fig. S5. Lysis was slightly delayed in the ΔL3 Mu, but substantially delayed in the ΔR3 strain; however, phage titers from these strains were comparable to wild-type. Phage isolated from both deletion strains were used in infection experiments to monitor FD removal. Compared to wild-type, both ΔL3 and ΔR3 phages integrated normally (Fig. 6, Mu panels), but neither degraded their FD (Fig. 6, FD panels). The ΔR3 phage was followed for a longer time because lysis is delayed in this mutant (Fig. S5).

The similar effect of deleting either R3 or L3 on FD removal could stem from the destabilization of E within the transpososome. Although these deletions would disrupt one of three crossings E makes with the L and R ends (Fig. 1B), all three may be needed to anchor E within the transpososome in the face of transcriptional activity at O. This scenario is also applicable to the Ner results (Figs. 3–5). Loss of even one E crossing by the R3 or L3 deletion might alter the disposition of the other non-catalytic MuA subunit. Our data suggest that these subunits play an important role in events that lead to FD degradation.

Discussion

Enhancers are known to regulate transcription, site-specific recombination, and transposition from a distance (Bulger & Groudine, 2011, Craig, 1985, Harshey & Jayaram, 2006). A common theme in these cases is that proteins bound at the enhancer, or direct contacts between the enhancer and the target protein, activate the catalytic potential of the target protein, generally an RNA polymerase, a recombinase, or a transposase. This study reports a new role for the phage Mu transposition enhancer – activating DNA degradation from a distance – thus directly implicating the transpososome in this process. We infer that enhancer-assisted positioning of two non-catalytic MuA subunits within the transpososome is important for this activation. The data also suggest that the native enhancer topology is important for commitment to the lytic phase of development.

Controlling flanking DNA degradation by enhancer topology

During the infection phase of Mu transposition, Mu DNA is linear, and the FD is non-covalently closed by MuN (Fig. S1). After integration, N is likely removed, FD is degraded by RecBCD, and the insertion is repaired (Fig. 1C) (Choi et al., 2014). While the delay in FD removal in a ClpX mutant could be interpreted as indicative of a role for a ClpX-remodeled strand transfer transpososome in FD degradation, it could also be consistent with an independent role for ClpX in this function (Choi & Harshey, 2010) The results of the present study provide direct evidence that the transpososome plays a role by showing that the enhancer topology influences FD degradation.

The enhancer, which is located at a distance from Mu ends (Fig. 1A), is brought into proximity with these ends by pairwise interactions between the enhancer and ends promoted by MuA, culminating in the assembly of a 6-subunit transpososome (Fig. 1B) (Pathania et al., 2003). The stable catalytic core of this complex is a tetrameric unit (Montano et al., 2012, Lavoie et al., 1991, Yuan et al., 2005). Of the two loosely held subunits (Lavoie et al., 1991, Kuo et al., 1991, Baker & Mizuuchi, 1992), one is expected to be positioned at R3 by the enhancer because under high salt conditions, or when the R3 site is deleted, the distal E-R crossing (E-R3) is lost (Yin et al., 2005, Yin et al., 2007). We surmise that the subunit at L3 is also positioned by the enhancer, although this could not be directly demonstrated (Yin et al., 2007); absence of the E-L crossing does not affect E-R crossings in vitro (Yin et al., 2005). Deletion of either the L3 or the R3 site does not inhibit assembly of a functional transpososome in vitro (Allison & Chaconas, 1992), nor does it interfere with integration of Mu phages carrying these deletions in vivo, showing that the tetrameric catalytic unit is assembled and is functional under these conditions (Fig. 6). However, post-integration, FD removal is blocked in these phages (Fig. 6). A logical mechanism by which deletion of a single MuA binding site at one end affects FD removal at both ends must involve the enhancer which juxtaposes the two ends. The selective effect of these deletions on FD degradation, implicates the non-catalytic subunits positioned by the enhancer at L3 and R3 in this process. Loss of either one of these subunits must alter the positioning of the other by altering enhancer topology.

In summary, this study reveals a new role for the non-catalytic subunits at L3 and R3 in post-integration degradation of the FD (Fig. 7A). A possible mechanism might involve remodeling of these subunits by ClpX to promote recognition and destabilization of MuN, either directly by looping of the intervening DNA, or indirectly via other factors. RecBCD can now enter and degrade the FD, followed by repair of the short gaps flanking Mu to yield a simple insertion (Fig. 1C). It is apparent that repair of the integrant in the non-replicative pathway of Mu transposition is a complex process actively promoted by the transpososome, in contrast to Mu replication in the lytic cycle which is also a complex process, but promoted by disassembly of the transpososome (Nakai et al., 2001).

Fig. 7.

Model for the role of the enhancer in FD removal. A. The MuA subunits at L3 and R3 sites (dotted ovals) depend on the enhancer to remain associated with the transpososome. ClpX may remodel these subunits for interaction with N (directly or indirectly), promoting removal of N. RecBCD will now enter and degrade the FD. B. The E-R3 DNA crossing (dotted circle) plays a critical role in maintaining the integrity of the transpososome by resisting disruption from transcription originating at Pc. Ner plays an important role at this crossing in allowing transcription from Pe, but inhibiting that from Pc.

Enhancer topology and phage development

The enhancer is known to influence the lysis-lysogeny decision not only because the regulatory proteins Rep and Ner which control this decision bind here, but also because MuA binds in the same region to promote the transposition-replication cycle (Fig. 1A) (Chaconas & Harshey, 2002, Symonds et al., 1987). Our study reports a clear role for Ner in controlling the post-integration removal of FD DNA (Figs 3, 4). At first, the results were puzzling, because both absence of Ner, or high levels of Ner, blocked FD removal; apparently, only the ‘right’ levels of Ner will do. The Ner absence effect was traced to transcription from Pc, not to synthesis of Rep (Fig. 4). We reasoned that Pc transcription likely disrupts the E-R3 node within the transpososome (Fig. 7B, dotted circle), destabilizing E from the complex as argued above, and that high levels of Ner may have a similar effect by binding near this node. Indeed, high concentrations of Ner disrupt E-R crossings (Fig. 5).

The negative effect on Mu replication from Pc transcription, unrelated to synthesis of Rep, has been reported earlier (Goosen & van de Putte, 1986). That this effect is due to disruption of the transpososome topology by disruption of the E-R3 node is supported by the behavior of the ΔR3 phage, which integrate normally (Fig. 6, left panel), but delay replication (Fig. S5). The MuΔR3 case makes a compelling argument for the importance of E topology, because it offers a mechanism by which a single site at the R end, far removed from the control center at E, can participate in the lysis-lysogeny decision. By stabilizing the E-R3 node, the MuA subunit at R3 plays a critical role in discouraging Pc and promoting Pe transcription (Fig. 7B). The native 3-crossings enhancer topology is likely designed to resist the disruptive effects of transcriptional activity at O, and promote commitment to lytic cycle.

Thus, we have described two roles for the enhancer topology beyond assembling a catalytically competent synapse – one in promoting lytic growth, and the other in promoting FD degradation. These findings explain why E remains within the transpososome even after strand transfer is complete (Pathania et al., 2003).

A six-subunit transposososome is unique to Mu

A majority of transposases whose structures have been reported, function as dimers (Montano & Rice, 2011). Mu, Tn7, and Hermes transpososomes are the most complex in requiring multiple transposase subunits (Chaconas & Harshey, 2002, Holder & Craig, 2010, Hickman et al., 2014). In the case of Mu and retroviral integrases, although only two subunits carry out the chemistry of transposition, the stable form of the transpososome is a tetramer (Lavoie et al., 1991, Cherepanov et al., 2011, Montano et al., 2012). Structural studies reveal that in both dimeric and tetrameric complexes, an intertwined network of protein-DNA and protein-protein contacts stabilizes the complex (Montano & Rice, 2011). If two transposase subunits are sufficient for catalysis, and four subunits add stability, why does the Mu transpososome have six subunits, especially when two of these are only loosely bound and not necessary for catalysis or stability? Our study answers this question. While these subunits could be involved in structural transitions within the complex that convert it into a stable transpososome (Mizuuchi & Mizuuchi, 2001, Kobryn et al., 2002), they clearly also play additional regulatory roles. The first of these regulatory roles is aiding FD removal (Fig. 7A). The second role is favoring lytic growth by blocking Pc transcription at E-R3 (Fig. 7B).

General Implications

There are several general implications of this work. First, while Mu/D108 are the only phages which employ an enhancer for transposition, enhancers are widely employed in transcription control. Our study therefore adds to the biological repertoire of enhancer functions. Second, genetic circuits that control on-off switches are currently modeled primarily on the DNA binding affinities of regulatory proteins. Our study adds DNA topology to the mix. Third, we show that the transpososome controls the lysis-lysogeny decision not only by controlling the transcriptional switch, but also by altering the integrity of the flanking DNA substrate. These data highlight the interplay between multi-subunit complexes, for example the transpososome and RNA polymerase, or the transpososome and the RecBCD motor (Choi et al., 2014), in regulating biological outcomes. Lastly, our study addresses why transpososomes are multimers, when a dimer is sufficient for catalysis. This question is relevant to transposable elements beyond Mu.

Experimental procedures

Strains

Strains used in this work were derivatives of E. coli K-12, and are listed in Table 1. All gene disruptions were made by λ Red-mediated homologous recombination (Sawitzke et al., 2007). The position of deletions is listed in Table S1. Primer sequences are listed in Table S2. Some gene deletions were carried out by first replacing the target gene by a kan cassette, which was amplified from pKD4 with 50-nt homology extensions from flanking regions of the target gene (Datsenko & Wanner, 2000), followed by removal of the kan cassette using pCP20, except for deletions of Pc-ner where the kan cassette was left in place. Other deletions were created as follows: first, the DNA to be deleted was replaced by a dual selection cassette - kan-ccdB - into the sequence to be replaced (amplified from pKD45, on which these genes are under a rhamnose-inducible promoter; (Kolmsee & Hengge, 2011)). Selection for the cassettes was on kanamycin. Next, the cassettes were replaced by homologous recombination with appropriate DNA to create the desired mutation, selecting on minimal media plates supplemented with 0.5% rhamnose to eliminate kan-ccdB cassette. To generate the Pc-ner deletion, the region containing Pc and ner gene (260 bp) was replaced by a kan cassette in CW26, which in turn was generated by the deletion of B through the SE region in HM8305 (Morgan et al., 2002). Since deletions were made in the temperature-inducible cts prophage, all incubation steps were performed at 30°C. All constructs were confirmed by DNA sequencing.

Table 1.

Strains and plasmids used in this study.

Strain	Genotype	Source (Ref)
HM8305	F’ pro lac::Mu cts62 / Δpro lac his met rpsL Mur	(Bukhari, 1975)
MP1999	AB1157, recB, recC, sbcB, malF::Mu cts62	(M. Pato)
MH3491	Mu cts62Aam1093 sup+	(O'Day et al., 1979)
BU1384	Δpro lac sup+	(Chaconas, 1984)
BU40	Δpro lac trp-8 sup− Smr	(Chaconas, 1984)
BU2044	F’ pro lac::D108 cts/Δpro lac met recA Smr	Bukhari lab
CW43	BU2044, D108 ΔA	This study
CW54	HM8305, Mu Δner	This study
CW26	HM8305, Mu ΔB through SE	This study
CW172	CW26, Mu ΔPc-ner::kan	This study
CW204	HM8305, Mu R3::kan	This study
RS147	MP1999, ΔL3	This study
RS148	MP1999, ΔR3	This study

Plasmid	Protein (promoter)	Resistance	Replication origin	Induction	Source
pWY12	MuA (under Ptrc)	Amp	pBR322	IPTG	This study
pWY15	D108 A (under Ptrc)	Amp	pBR322	IPTG	This study
pWY38	Mu Ner, A, B (no promoter)	Cam	p15A		This study
pWY59	Mu Ner (under Para)	Cam	p15A	Arabinose	This study
pWY96	Mu Ner (under Para)	Amp	p15A	Arabinose	This study
pWY61	Mu Repts, Ner, A, B (under Pe and Pc)	Kan	p15A	Tm shift	This study
pWY62	Mu Ner, A, B (under Pe)	Kan	p15A		This study
pIL153	Mu A, B (no promoter)	Cam	p15A		(Choi & Harshey, 2010)
pIL137	Mu A, B (under Ptet)	Cam	p15A	Tetracyline	made by I. Lee
pKD45	Source for kan-ccdB cassette	Kan	oriR6K gamma	Rhamnose	(Kolmsee & Hengge, 2011)
pKD46	Lambda-red recombinase	Amp	oriR101	Arabinose	(Datsenko & Wanner, 2000)
pKD4	Source of kan cassette	Kan	oriR6K gamma		(Datsenko & Wanner, 2000)
pCP20	FLP recombinase	Amp, Cam	oriR101	Tm shift	(Datsenko & Wanner, 2000)
pWY72	Ner	Kan	pBR322	IPTG	This study
pSP(R)Dir	Mini-Mu with loxP	Amp	pBR322		(Pathania et al., 2002)
pSP(R)In	Mini-Mu with loxP	Amp	pBR322		(Pathania et al., 2002)

* Δ, deletion of genes/sites

:: indicates either insertion of Mu or D108 into lac, or deletion-substitution when an antibiotic resistance cassette is inserted in the indicated gene. The exact location of the deletion is given in Table S1.

Tm, temperature

Preparation of Δner and ΔPc Δner Mu phage

A ner⁻ prophage strain CW54 was first constructed by deleting ner. This mutant cannot go through lytic growth unless complemented for Ner (Goosen & van de Putte, 1986). Phage production from the mutant strain was attempted by providing Ner from a plasmid (pWY59; Table 1). However, this did not succeed, likely because lytic growth is exquisitely sensitive to Ner levels (Van Leerdam et al., 1982). After trial-and-error, we were able to induce the ner⁻ prophage by providing two different plasmids together - pWY62, which expresses ner, A, B from the natural Pe promoter, and pIL153, which expresses A and B at a basal level from a read-through plasmid promoter (Table 1). The resulting ner⁻ phage were used in infection experiments.

The ΔPc Δner Mu prophage strain CW172 was constructed as follows. Prior to inactivating Pc, we deleted all the genes (with the exception of A) expressed from Pe in the Mu prophage (i.e. B through the SE region; see Fig. 1A). Some of these genes enhance transposition (e.g. B) and others enhance cell killing (kil, in the SE region). They were deleted in order to prevent the induction of lytic growth and cell killing when we eliminated the Rep function by deleting Pc. Next, we introduced a deletion that spanned both the −10 and −35 consensus of Pc, along with ner (see Fig. 1A, lower panel). Induction of lytic growth from this ΔPc Δner phage was achieved by using two plasmids – pWY61 (cts62 ner A B) and pIL153 (A B). In pWY61, Rep made from cts62 prevents AB expression until the culture is shifted to the non-permissive temperature.

Plasmids

D108 A from pET-D108 A (Yang et al., 1995b) and MuA from the Mu lysogen in HM8305, were amplified and cloned in NcoI and BamHI sites of pTrc99 to generate pWY15 and pWY12, respectively. pWY38 was made by cloning of ner, A, and B genes from the HM8305 together with the ner ribosomal binding site, into HindIII and Bsu36I sites of pACYC184. The ner gene alone was amplified with KpnI and SphI recognition sites for cloning in pBAD33 or pBAD24 (pWY59 and pWY96). pWY61 and pWY62 contain ~4 kb of DNA encoding Rep through B and ~ 3.7 kb of DNA encoding Ner through B. pIL137 has A and B genes under the TetR promoter in pACYC184. pIL153 was made by eliminating the TetR promoter in pIL137, and is equivalent pIL164 (Lee & Harshey, 2001). pSP(R)Dir and pSP(R)In plasmids have been described (Pathania et al., 2002). These plasmids were originally derived from mini-Mu pMK21 (Kim et al., 1995) and carry two loxP sites flanking the R end.

Proteins

MuA and HU proteins were purified as described (Yang et al., 1995a). Cre protein was a gift from Makkuni Jayaram (University of Texas at Austin). Ner was cloned into NcoI and BamHI sites of pET28 to generate pWY72, which was transformed into BL21(DE3)pLys E. coli host. Ner expression was induced with 1 mM IPTG at 37C, and purified as described (Kukolj et al., 1989).

Detection of Mu and FD sequences integrated in the E. coli genome

General procedures for prophage Mu induction, phage purification, and infection have been described (Bukhari & Ljungquist, 1977, Au et al., 2006). Prophage strains were induced by thermal shift at 42°C for 40 min, followed by shift-down to 37°C until lysis. The 0 time point indicates the time of shift-down. Phage production from MuΔner prophage strain CW54 required plasmids pWY62 (Ner, A and B) and pIL153 (A and B). MuΔPcΔner phage were obtained by induction of CW172 in the presence of plasmids pWY61 (Rep, Ner, A, and B) and pIL153 (A and B). Unless otherwise indicated, in all infection experiments strains were grown at 37°C till OD₆₀₀ reached 0.4, phage were added at moi = 5, and cultures incubated at 37°C with shaking, until they lysed. The 0 time point indicates the time of infection. MuA am phage were titered on BU1384 (sup⁺). Titers of Δner and ΔPc Δner phage were obtained on BU1384 complemented by pWY38 (ner, A, and B).

Total cellular DNA was extracted from cells at various times after infection, and subjected to pulse field gel electrophoresis as described (Au et al., 2006). Genomic DNA was excised from the gel and used as template for PCR as described (Choi & Harshey, 2010). Mu integration was detected with primers Owy031 and Owy162 (they amplify a region at the R end between nts 35676–36378; Table S2), and FD DNA was detected with primers Owy033 and lacZ (R) (Table S2). In case of ΔL3 and ΔR3 strains, FD DNA was detected with primers Owy033(F) and RS341(R), which amplify across the R end and malF sequences, because the prophages reside in malF (Table 1 and Table S2). The lacZ-R end primer set amplifies a 501 bp DNA fragment (72 bp inside Mu and 429 bp in lacZ), while the malF-R end primer set amplifies a 572 bp DNA fragment (313 bp inside Mu and 259 bp in malF). The PCR conditions used for ΔL3 and ΔR3 strains were 94°C for 2 min; 30–40 cycles at 94°C for 30 s, 58°C for 30 s, and 72°C for 30 s; and a final extension at 72 °C for 2 min.

Quantification of Pc transcript levels in Δner Mu phage

Plasmids pWY62, pIL137, and pWY59 were introduced individually into the ner⁻ prophage strain CW54 (Table 1). The resulting strains were grown at 30°C until OD₆₀₀ was reached 0.5–0.6. The ts repressor was inactivated by thermal shift to 42°C for 30 min, followed by shift-down to 37°C for 15 min. Arabinose (0.1%) when used was added concomitant with the temperature shift-down. Three ml of culture were harvested for RNA isolation using ToTALLY RNA Kit from Ambion according to their specification. The purity of total RNA was checked by agarose gel electrophoresis and the RNA concentration was determined by measuring the OD₂₆₀ to OD₂₈₀ ratio as calibrated by the spectrophotometer (BioRad). RNA samples were stored at −80°C until use. M-MuLV Reverse Transcriptase (NEB) was used to make cDNA (following NEB protocol) from 1 µg total RNA using a gene-specific primer GSP1 (Table S2) for detecting Pc transcripts. The cDNA obtained was used directly for real-time qPCR analysis. Aliquots with 12.5 µl SYBR master mix (Applied Biosystems Inc; includes dNTPs, enzyme and buffer), 1 µl of each primer (10 µM), 1 µl of cDNA template and 9.5 µl of double distilled H2O were held for 10 min at 95°C, followed by 40 cycles of 15 sec at 95°C and 1 min at 60°C (7900HT; Applied Biosystems). Product integrity was checked using the dissociation curve.

Assembly of Mu Transposition Complexes and Cre Recombination Reactions

Type I Mu transpososomes were assembled on mini-Mu pSP Dir and In substrates with 30 µg/ml plasmid DNA, 10 µg/ml HU, and 7 µg/ml MuA in 20 µl of 20 mM HEPES-KOH (pH 7.6), 140 mM NaCl, and 10 mM MgCl₂ for 30 min at 30C. Ner was added at indicated concentrations prior to addition of MuA and incubated for 15 min at 30C. Cre recombination was initiated by the addition of 0.2 µg of Cre followed by further incubation at 30C for 30 min. Reactions were stopped by SDS addition to a final concentration of 0.5% and further de-proteinized with 5 µg Proteinase K at 30C for 30 min. DNA samples were then column-purified (Zymo Research, USA) and electrophoresed in 1% agarose gel at room temperature.