Phage Mu enlists the β-sliding clamp for late gene transcription

Significance

For phages, efficient replication and transcription ensure that components necessary for assembly of new phage particles are in place before cell lysis begins. The strategies to do this are diverse, including encoding their own RNA and DNA polymerases, modifying host RNAP to recognize phage promoters, or producing factors that allow RNAP to transcribe through termination signals. We report here that phage Mu appropriates the host β-sliding Clamp, an essential replication factor that interacts with Pol III, to transcribe late phage genes through interaction with the late gene transcription factor MuC. This mechanism would, at the very least, allow Mu to couple late gene expression with ongoing Mu DNA replication.

Abstract

The phage Mu C protein (MuC), along with core RNA polymerase (RNAP) and σ⁷⁰, is required for transcription of phage late genes. We found that overexpression of MuC was lethal in Escherichia coli and observed that host replication was overinitiating under these conditions. Suppressors of MuC lethality mapped to dnaA, diaA, and dnaX. DnaA initiates replication at oriC, assisted by DiaA. Reinitiation is prevented by hydrolysis of ATP-DnaA to ADP-DnaA by Hda and DnaN (β-sliding clamp or Clamp). DnaX, the tau/gamma subunit of Pol III, is part of the Clamp loader complex. Coexpression of hda and dnaN rescued MuC lethality. This result suggested that MuC was interfering with either Hda or DnaN. We noticed that MuC contains two near-consensus Clamp-binding motifs (CBM), one at the N terminus and one at the C-terminus. Changing the consensus residues of either CBM abolished MuC lethality, abrogated MuC-dependent transcription, and reduced plaque-forming units. Inactivation of a ts Clamp through temperature shift specifically inhibited MuC-dependent transcription but not σ⁷⁰-dependent transcription. We show that MuC interacts with the Clamp to activate late gene transcription both in vivo and in vitro. This study demonstrates the involvement of the E. coli Clamp, a processivity factor essential for DNA replication, for transcription by E. coli RNAP. We observed that members of the Mor/MuC family of transcription factors all possess at least one CBM, suggesting that engaging the Clamp for transcription is likely to be more prevalent than hitherto recognized.

Phage Mu amplifies its DNA by a unique mechanism called replicative transposition (1). Two phage-encoded proteins MuA and MuB, and two host proteins IHF and HU, participate in transposition of the 37 Kbp Mu genome. Transposase MuA recognizes and binds to the two Mu ends, and MuB is a AAA+ family protein that facilitates target DNA capture by the Mu ends (2). Synapsis of the ends by MuA is assisted by IHF and HU, which bind and bend DNA to facilitate assembly of the transpososome. Mu ends get joined to target DNA through a nicked end-joining mechanism facilitated by MuA (3). The transpososome is then disassembled, leading ultimately to assembly of DNA polymerase III holoenzyme at the Mu ends and resolution of the transposition product to generate a duplicated copy of Mu (4).

Pol III is the replicative polymerase of the 4.6 Mb Escherichia coli genome (5, 6). It consists of 10 distinct proteins divided into 3 subassemblies: Core Polymerase (α, ε, and θ), Clamp (β), and Clamp Loader Complex or CLC, (τ, γ, δ, δ’, χ, and ψ) (7). DNA synthesis is carried out by the α subunit; ε is the proof-reading subunit (3’-5’ exonuclease activity), while θ serves to stimulate the proof-reading function of ε (7). The β, or the Clamp, is the processivity factor that enables Pol III to synthesize >50 Kbp before disassociating (8). In the absence of the Clamp, Pol III core exhibits a processivity of ~10 bp, too low to support the replication of the E. coli genome (7, 9). Crystal structure of the Clamp reveals the mechanism for processivity: a dimer of the Clamp forms a ring-shaped structure that encircles the DNA, thus enabling the core DNA polymerase to stay associated with the DNA during replication (10). The ring structure of the Clamp is released and loaded onto primer/template junctions by the action of the CLC (11). The minimal pentameric CLC consists of 3 subunits of either τ or γ, one subunit of δ, and one subunit of δ’ (12). In addition to the minimal CLC, two accessory proteins χ and ψ assist in the DNA binding activity of CLC at the primer/template junction (13, 14).

The Clamp is also involved in controlling the rate of replication initiation in E. coli. DnaA, assisted by DiaA, controls replication initiation at the origin of replication (oriC) by binding DnaA boxes at oriC, promoting melting of the duplex and subsequent assembly of DnaB helicase, DnaG primase, and eventually Pol III (15). Replication initiation by DnaA is regulated by its nucleotide-bound state, with only ATP-DnaA being proficient in replication initiation (16). One mechanism employed to prevent reinitiation is to control the nucleotide-bound state of DnaA by RIDA (Regulatory Inhibition of DnaA) (15). RIDA requires Hda (homologue of DnaA) and the Clamp (17–19). Interaction between the Clamp and Hda enables Hda to stimulate the ATP-hydrolysis activity of DnaA, decreasing ATP-DnaA levels, thus preventing replication initiation (20). RIDA and other mechanisms involved in regenerating ATP-DnaA work in concert to ensure timely replication initiation (15).

We report here that the Clamp is required for late Mu gene transcription. Mu employs a transcription program typical of most phages, with early genes encoding replication/transposition functions and late genes specifying phage morphogenesis and lysis functions. All Mu genes are transcribed by host RNAP and σ⁷⁰. Switching from early to middle and late genes is mediated by transcription factors Mor and MuC, respectively (21). MuC is a 16.5kD polypeptide that binds 4 specific DNA sequences on the Mu genome to activate transcription (22–24). These sites reveal a bipartite consensus sequence 5’-TTAT-[N_5-6]-ATAA-3’ (25). MuC-dependent promoters possess a recognizable -10 sequence but lack a recognizable -35 sequence (22, 23). MuC interacts with core RNAP and σ⁷⁰ to recognize Mu late promoters (24, 26). However, it does not require the C-terminal domain of the α subunit nor the C-terminal domain of σ⁷⁰ (26).

We serendipitously observed that overexpression of MuC is lethal. Identity of the suppressors of this lethality led us to hypothesize that overexpression of MuC was disrupting RIDA. We found that MuC interacts with the Clamp through two Clamp-Binding Motifs (CBMs) and that this interaction stimulates MuC transcription, thus linking late transcription to replication. Interestingly, phage T4, which uses host RNAP but encodes its own DNA polymerase, Clamp, CLC, and transcription factors, uses the same strategy as Mu for late gene transcription (27). We discuss why these two phages, very different in both their genome size and lifestyle, may have converged on the same strategy.

Results

MuC Lethality Is Suppressed by Mutations in Host Replication Genes.

This study began with an initial curiosity about the peculiar Ribosome Binding Site (RBS) of MuC, shown to be inefficient for expression of the gene (22, 23). Wondering whether this RBS was a regulatory sequence activated exclusively by a phage product, we devised tests that involved cloning MuC in various vectors and stumbled on the observation that induction of MuC was lethal, resulting in a 3-log reduction in CFU (Fig. 1B). This MuC-overexpression phenotype had not been reported previously (23, 28), so we decided to investigate its mechanism by suppressor analysis. To avoid isolation of loss-of-function MuC mutants, we monitored retention of MuC activity through its ability to activate P_mom-lacZ (one of the four C-dependent promoters; Fig. 1A) from a compatible plasmid, monitoring β-galactosidase (lacZ) activity on plates supplemented with X-gal (Materials and Methods). We isolated 14 blue survivor colonies and performed whole genome sequencing to identify the suppressor mutations (Table 1). We were surprised to find that a majority of the suppressors carried mutations in three genes involved in DNA replication—dnaA, diaA, and dnaX. Two distinct mutations were recovered in dnaA: a small deletion of the linker domain II (Δ96-118), thought to be involved in DnaA–phospholipid interaction (29), and a missense mutation in the ATP-binding domain III (R342H). Of the two mutations in diaA, one was a missense mutation (G148A) and the other generated a premature stop codon at position 94. The missense mutation in dnaX (P380A) would affect the τ/γ subunit of the CLC. This mutation occurs concurrently with a diaA mutation, so we were unable to directly attribute the contribution of this allele to resistance to MuC toxicity.

Fig. 1.

MuC lethality is suppressed by mutations is host replication genes. (A) Transcription program of phage Mu. Genes involved in regulating transcription of Mu-specific genes are color-coded. Known promoters of each transcription factor are labeled and indicated by an arrow. The cartoons are self-explanatory. (B) Overexpression of MuC is lethal. MuC was provided on a p15A-origin vector under P_tet control (Left). Serial dilutions of the tester strains, without and with the inducer aTc (Right). One of three biological replicates is shown. (C) MuC lethality is suppressed by deletion of diaA. P_tet-MuC was introduced into a ΔdiaA background and MuC lethality was assayed as in B. aTc, anhydrotetracycline.

Table 1.

Suppressors of MuC lethality and their associated mutations

Strain ID	Amino acid change	Gene name	Codon number	Variant frequency
1	Q -> K	flhB	25	100.00%
	G -> A	diaA	148	100.00%
2	Q -> K	flhB	25	99.40%
	G -> A	diaA	148	100.00%
3	Frame shift	sapB	228	100.00%
4	L -> *	diaA	94	98.40%
5	P -> A	dnaX	380	100.00%
	L -> *	diaA	94	100.00%
6	A -> P	zntA	675	99.50%
	A -> P	ydhK	392	100.00%
	L -> *	diaA	94	100.00%
7	A -> P	yqgE	13	100.00%
	L -> *	diaA	94	99.50%
8	Δ96-118	dnaA	N/A	100.00%
9	R -> H	dnaA	342	100.00%
10	G -> A	fiu	217	99.60%
	Δ96-118	dnaA	96	100.00%
13	Frame shift	sapF	50	100.00%
14	L -> P	trkA	6	98.60%
	Δ419-425	rsmB	425	99.50%

Variant frequency indicates the percentage of reads with the indicated mutation relative to the total number of reads. See Materials and Methods for cut-off thresholds.

To verify that the mutations in the three DNA replication genes recovered were related to MuC lethality, we chose diaA, the only one that is nonessential (30). Deletion of diaA permitted growth of cells overexpressing MuC (Fig. 1 C, Top panel) compared to the WT strain (Fig. 1 C, Bottom panel). Providing DiaA in trans restored MuC lethality (SI Appendix, Fig. S1, Top panel). To better understand the mechanism of ΔdiaA resistance to MuC lethality, we introduced into it two point mutations with known phenotypes. DiaA(N83D) affects DiaA oligomerization, while DiaA(F191S) is defective for DnaA binding (31). Like WT diaA, the N83D mutant showed sensitivity to MuC in a ΔdiaA background (SI Appendix, Fig. S1, Middle panel, compare to Top panel). However, DiaA(F191S) showed resistance to MuC lethality like that of a ΔdiaA strain (SI Appendix, Fig. S1, Bottom panel), suggesting that DiaA per se is not the direct target of MuC, but the lethality is likely due to deregulation of DnaA initiation at oriC. We tested that next.

MuC Overexpression Causes Overinitiation of Host Replication.

To determine whether MuC overexpression interfered with RIDA, we labeled DNA with Sytox green and performed flow-cytometry. As observed before (32), WT cells growing in LB contained 4 chromosomes (−aTc); we also observed a small population containing 8 chromosomes (Fig. 2A, green peaks). With MuC overexpression (+aTc), we observed loss of a discrete chromosome copy number, accompanied by an increase in DNA levels in cells (Fig. 2A, blue graph) relative to uninduced control. This is indicative of overinitiation of replication: replication initiates before Pol III can complete replication, increasing the amount of DNA content in cells, with distinct chromosome peaks absent. We also examined the DNA content of ΔdiaA with or without MuC overexpression to determine if the absence of diaA could suppress the observed increase in DNA content found in WT. In the absence of MuC induction, ΔdiaA cells (−aTc) have a variable number of chromosome copies (Fig. 2A, red), ranging from 1 to 4, consistent with previous findings (30). With MuC overexpression (+aTc), ΔdiaA cells still show an increase in DNA content (Fig. 2A, orange graph, red arrow) and loss of discrete chromosome copies, but not as severe as that of WT (i.e. orange ΔdiaA curve is left-shifted compared to blue WT curve, indicating less DNA content). We conclude that MuC overexpression causes overinitiation of replication.

Fig. 2.

MuC overexpression causes replication overinitiation. (A) Flow Cytometry of MuC-overexpressing strains. WT and ΔdiaA strains with or without aTc induction of MuC (differentially color-coded) were stained with Sytox Green and sorted (Materials and Methods). 10,000 cells were analyzed for each sample. The y-axis, “unit area”, is the normalized number of cells. Arrows indicate the peak of the distribution; note the left-shifted peak of ΔdiaA compared to that of WT. The data shown are one of three biological replicates. (B) Scheme to determine relative ATP-DnaA levels. lacZ is fused to P_dnaA; LacZ activity is inversely correlated with ATP-DnaA repression of P_dnaA activity (see text). (C) ATP-DnaA activity in the presence of MuC-overexpression constructs. Relative transcription of lacZ (induced/uninduced for MuC expression) fused to P_dnaA, using the Miller assay. The first measurement was taken at an O.D. of 0.3 for both MuC variants and continued for the indicated times. (D) Suppression of MuC lethality by DnaA(L366K). The mutant was cloned downstream of P_lacZ (IPTG-inducible), while MuC is under P_tet control. (E) Suppression of MuC lethality by Hda-DnaN. hda and dnaN were cloned downstream of P_lacZ. Other descriptions are as in D.

MuC Overexpression Interferes with RIDA.

Replication initiation in E. coli is controlled by the nucleotide-bound state of DnaA (15, 16). DiaA helps load ATP-DnaA at oriC, where it forms spiral-like oligomers that cause the region to melt, and initiates replication by promoting stepwise assembly of Pol III (33); ADP-DnaA is not proficient in these activities. Strains compromised for RIDA have been shown to accumulate ATP-DnaA (18). We took several approaches to test whether overinitiation of replication by MuC overexpression is due to an increase in ATP-DnaA levels. First, we exploited the ability of DnaA to autoregulate itself to indirectly measure ATP-DnaA levels when MuC was overexpressed: ATP-DnaA binds and represses transcription from its promoter, while ADP-DnaA is capable of binding but not repression (34). We fused the promoter region of dnaA to lacZ, with the expectation that high levels of ATP-DnaA would repress transcription from P_dnaA, resulting in low β-galactosidase activity, and vice versa (Fig. 2B). We observed that with MuC overexpression (black line, Fig. 2C; note that the y-axis plots a ratio), β-galactosidase activity decreased on average by 58% across the time points. To verify that the increased ATP-DnaA level is linked to MuC lethality, we tested a MuC mutant that is nontoxic (L15A) (SI Appendix, Fig. S2A) but is transcriptionally active (SI Appendix, Fig. S2B). Overexpression of MuC(L15A) retained 90% of uninduced lacZ activity on average across time points (Fig. 2C, blue line), supporting our hypothesis that MuC overexpression is linked to accumulation of ATP-DnaA.

Second, we rescued MuC toxicity by exploiting the interaction between DnaA and phospholipid, known to regulate ATP-DnaA levels. DnaA interacts with acidic phospholipids in vitro, facilitating the exchange of ATP and ADP (35, 36). This interaction is thought to be important for initiating replication because a mutation in pgsA, which blocks the first committed step of phospholipid biosynthesis, is suppressed by DnaA mutations in an amphipathic helix near domain III, but only in the presence of WT DnaA, possibly by making mixed DnaA oligomers (37). One of these mutants, DnaA(L366K) has been shown to have an altered interaction with phospholipids without any significant change in its affinity for ATP or for oriC binding in vitro (29, 38). We hypothesized that DnaA(L366K), which could bypass the requirement for phospholipids in nucleotide exchange, could rescue the accumulation of ATP-DnaA caused by MuC by releasing ATP from DnaA. We therefore coexpressed either DnaA or DnaA(L366K) (IPTG-inducible) with MuC (aTc inducible) and assessed their ability to rescue MuC lethality. DnaA(L366K) induction rescued lethality (Fig. 2D, +aTc and +IPTG panel), but DnaA did not (SI Appendix, Fig. S2C, +aTc and +IPTG panel). Without IPTG induction, DnaA(L366K) did not alleviate the effect of MuC (Fig. 2D, +aTc and −IPTG panel). These observations lend support to overinitiation of replication by MuC being due to ATP-DnaA accumulation.

Finally, since phospholipid-dependent nucleotide exchange is not specific to ATP-DnaA, i.e. both ATP and ADP forms are exchanged (39), we examined whether specific hydrolysis of ATP-DnaA to ADP-DnaA by the Clamp-Hda complex would rescue the MuC phenotype. Hda is homologous to the ATP-binding domain (domain III) of DnaA and promotes hydrolysis of ATP-DnaA in conjunction with the Clamp (20). We introduced hda and dnaN (Clamp) individually or together and assessed their ability to rescue. Expression of either hda or dnaN alone did not rescue MuC lethality (SI Appendix, Fig. S2D, +aTc +IPTG panels). However, coexpression of hda and dnaN provided full rescue (Fig. 2E, +aTc and +IPTG panel).

Taken together, we can conclude that MuC overexpression causes replication overinitiation from accumulation of ATP-DnaA, not by interacting directly with DnaA (because overexpression of DnaA alone does not rescue MuC lethality; SI Appendix, Fig. S2C), but by interfering with RIDA through interacting either with the Clamp or with Hda.

MuC–Clamp Interaction Is Required for Lethality.

Many DNA polymerases in E. coli, as well some transposases, interact with the Clamp (40–42), giving rise to a consensus binding motif of QL[$\begin{array}{c}S\\ D\end{array}$]LF in prokaryotic systems (Fig. 3 A, Top). We observed that MuC contains two near-consensus putative CBMs: one at the N terminus and one at the C-terminus (Fig. 3 A, Bottom), deviating from the consensus sequence only at the second position (QHDLF at the N terminus, and QPDLF at the C-terminus), suggesting that MuC interacts with the Clamp. To test this interaction in vivo, we used the bacterial two-hybrid BACTH system. Fusion of MuC-T25 and T18-DnaN produced positive β-galactosidase activity, indicative of interaction (Fig. 3B, 4th bar from Top). We verified that this interaction is dependent on CBMs by constructing point mutations in them. At the N-terminal CBM, changing the nonconsensus residue H to a consensus residue L (i.e. H3L; 3 refers to its position in MuC), showed a 1.6-fold increased interaction with the Clamp (Fig. 3B). We also substituted a conserved L residue of both CBMs with an A residue. This substitution at either CBM abolished interaction between MuC and the Clamp (Fig. 3B).

Fig. 3.

MuC interaction with the Clamp is required for lethality. (A) MuC possesses two putative CBMs. Top. CBMs from different systems (adapted from 40-42) and the consensus CBM. Bottom. Putative CBMs in MuC. The coding sequence of MuC is shown, with putative CBMs underlined. Red residues are identical to that of the consensus CBM. (B) MuC interacts with the Clamp in vivo. Miller units from BACTH assay of indicated MuC variants and the Clamp (DnaN) are shown. Data points for each replicate are shown as black circles. Representative images of the assay are shown on the Right. The leucine zipper (ZIP) fused to either the T18 or T25 fragment was used as positive control for interaction. ***P < 0.01, student t’s test (two-tailed). (C) MuC lethality requires Clamp interaction. MuC lethality assay for variants shown in B. Other descriptions as in B.

To determine if MuC–Clamp interaction is responsible for MuC lethality, we examined the ability of the CBM mutants to inhibit growth. Substitution of L→A in either of the CBMs abolished MuC lethality (Fig. 3 C, Right). By contrast, the H3L variant, which showed a stronger interaction with Clamp (Fig. 3B), was more toxic than WT (Fig. 3 C, Left). The increased lethality of the H3L variant was verified more quantitatively in liquid growth. Indeed, upon the induction of MuC(H3L), growth ceased immediately compared to the slowing of growth in WT MuC, or no effect of the L→A substitutions (SI Appendix, Fig. S3A). To ascertain that the observed reduction in MuC–Clamp interactions of the CBM variants is not due to reduced stability, we fused (3X-)FLAG tag to the C-terminus of MuC. Although reduced in toxicity compared to MuC (SI Appendix, Fig. S3B), MuC-FLAG was still toxic and hence could be used to assess MuC stability. Introduction of L5A and L133A substitutions into MuC-FLAG showed that neither substitution affected MuC levels (SI Appendix, Fig. S3 B, Bottom), ruling out the possibility that the reduction of MuC–Clamp interaction in the two-hybrid system is due to reduced stability of the CBM variants.

Mor, the transcription factor that activates the middle operon of Mu (which includes MuC; Fig. 1A), and MuC are highly similar (43, 44). So, we looked for a CBM in Mor and found a putative CBM with only the sequence DLF of the consensus QL[$\begin{array}{c}S\\ D\end{array}$]LF at the N terminus (SI Appendix, Fig. S3C). We confirmed that Mor overexpression is also toxic, albeit to a lesser extent than that of MuC (SI Appendix, Fig. S3 D, Top). We constructed a CBM variant of Mor (L5A) and found that this variant is nontoxic (SI Appendix, Fig. S3 D, Bottom). We also determined the ability of this Mor variant to promote transcription in vivo by fusing the Middle of promoter of Mu, P_middle, (45) to lacZ. Consistent with its loss of toxicity, Mor L5A was unable to promote P_middle transcription (SI Appendix, Fig. S3E). Since Mor/C defines a family of transcription factors (46), we examined 100 other members of this family to determine if they possess CBMs. We found a CBM at the N terminus of the Mor/C superfamily with the consensus Qx[D/E][L/M]F, and at the C-terminus with the consensus QPDLF (SI Appendix, Fig. S3F). We infer from this finding that Clamp binding is a general property of the Mor/MuC superfamily of transcription factors.

We interpret these results to mean that MuC lethality is a result of its interaction with the Clamp, preventing the Clamp from participating in RIDA. Yet, MuC is a transcription factor, so we hypothesize that the MuC–Clamp interaction is directly related to the role of MuC in Mu late gene transcription.

MuC Interaction With the Clamp Is Required for Transcription In Vivo.

To determine if the Clamp is involved in MuC transcription, we first measured transcription from P_mom using MuC CBM variants. Consistent with their effects on lethality (SI Appendix, Fig. S3A), the L5A and L133A variants were completely defective for P_mom transcription (Fig. 4A), while the H3L variant showed higher transcription, reaching a maximum of 130% of the WT (Fig. 4A, blue triangle). These results suggest that the affinity of MuC for the Clamp is directly correlated with its transcriptional activity.

Fig. 4.

MuC interacts with the Clamp for transcription in vivo. (A) Transcriptional activity of MuC variants. Activity of indicated variants compared to the WT MuC (dotted line). (B) Effect of the loss of MuC–Clamp interaction on phage titers. Spot-plaque assay of Mu lysates carrying indicated MuC variants expressed from the phage genome. (C) The Clamp is specific for MuC transcription. In a dnaNts background (illustrated by cartoon), relative transcriptional activities from MuC-dependent (P_mom) and σ⁷⁰-dependent (P_tet) promoters (empty circle and square, respectively) were monitored at 42 °C vs 30 °C. Filled symbols indicate complementation with WT dnaN in the same strains. MuC was expressed from P_tet in (A and C) without aTc induction.

Since MuC is required for late gene transcription (22), we sought to determine the effect of CBM substitutions on phage propagation. We introduced MuC CBM variants into Mu prophages and assayed their ability to either lyse the host cell or to plaque. We observed that the L5A variant (filled triangle) exhibited a 20-minute delay in lysis compared to the WT (SI Appendix, Fig. S4A), while the variant with both CBMs mutated (L5A L133A; filled square) showed a 40-minute delay in lysis. In contrast, the ΔMuC prophage did not lyse the host (SI Appendix, Fig. S4A, open triangle). Consistent with the delays in lysis, lysate from the L5A variant showed a 1-log decrease in PFU (Fig. 4B), while that from the double mutant showed a 2-log decrease compared to the WT (Fig. 4B). These results indicate the MuC–Clamp interaction contributes to Mu growth.

To further examine if the Clamp is specific to MuC transcriptional activity, we exploited a temperature sensitive (ts) allele of DnaN, dnaN159, (Fig. 4 C, Left). We observed that upon shifting to nonpermissive temperature, MuC transcription decreased with time in vivo (Fig. 4 C, Right, open circle), reaching 80% reduction in transcriptional activity after 2 h (compared to the permissive condition; note that the y-axis is a ratio). This decrease does not occur with P_tet, a σ⁷⁰-dependent promoter (Fig. 4 C, Right, empty square). In fact, P_tet activity increased by 2-fold at 80 min and then leveled off (MuC also requires σ⁷⁰ (24)). The decrease in MuC transcription due to Clamp inactivation could be rescued by providing WT dnaN in trans (Fig. 4 B, Right, filled circle), which shows an ~1.3-fold stimulation of P_mom transcription at the nonpermissive temperature. A g eneral stimulation of transcription from P_tet was not observed when WT dnaN was provided in trans (Fig. 4 C, Right, filled square). Since the Clamp is known to require the CLC for Pol III activity (5), we also determined whether the Clamp loader is required for MuC transcription in vivo using a ts allele of the τ/γ subunit (DnaX), dnaX2016, of the CLC. We observed a similar decrease in P_mom transcription upon temperature shift, but not so in P_tet (SI Appendix, Fig. S4B). Thus, both the Clamp and Clamp loader are essential for Mu late gene transcription in vivo.

MuC Interaction With the Clamp is Required for Transcription in Vitro.

To better demonstrate that the Clamp assists MuC in activating Mu late gene transcription, we reconstituted the reaction in vitro from purified RNAP core, σ⁷⁰, MuC and its variants, DnaN, and CLC (γ-Complex) (SI Appendix, Fig. S4C). Transcription was assayed using a convenient aptamer-based (Baby spinach) reporter system, which fluoresces when bound to DFHBI, and where the output is a result of multiple rounds of transcription (47, 48) (SI Appendix, Materials and Methods). At first, we did not observe any difference in MuC activity when the Clamp was added to the reaction. Previous work (49) has shown that, in addition to protein components, Clamp-dependent transcription requires a nicked substrate. To determine if there is a similar requirement for a nicked substrate for activation of MuC transcription by the Clamp, we introduced a nick 88 bp, and 98 bp upstream of a linear substrate containing P_mom, on the nontemplate and template strands, respectively. We found that a substrate nicked on the nontemplate strand exhibited greater transcriptional activity than the no-nick substrate (2-fold increase in RFU at the final timepoint) (Fig. 5 B, Top). By contrast, a substrate that is nicked on the template-strand exhibited similar transcriptional activity to that of the no-nick substrate (Fig. 5 B, Bottom). To verify that the CBM on MuC is responsible for the observed activation of transcription, we purified and tested two MuC variants carrying defective CBMs: L5A, and L5A L133A. These variants directed transcription of P_mom at similar levels to that of the WT MuC on a no-nick substrate (Fig. 5 C, Bottom). However, on the substrate nicked on the nontemplate strand, the Clamp did not activate transcription with either MuC variant compared to WT MuC (Fig. 5 C, Top). To determine the contribution of the Clamp to MuC transcription, we removed the Clamp from the reaction on the nicked nontemplate strand substrate and observed that transcriptional activity reduced by 2-fold in the absence of the Clamp compared to its presence (Fig. 5D), similar to the extent of decrease when a non-nicked template was used.

Fig. 5.

MuC interacts with the Clamp for transcription in vitro. (A) Scheme of in vitro transcription. Transcription from a P_mom substrate with purified core RNAP, σ⁷⁰, MuC, Clamp, and CLC, is monitored through the fluorescence of Baby Spinach aptamer placed downstream as shown. (B) Transcription time course on the P_mom substrate, either nicked on the nontemplate (Top) or template strand (Bottom). (C) Transcription on the P_mom substrate with MuC variants carrying indicated substitutions in the CBM. Top: all nicked on nontemplate strand; WT (blue), L5A (green), and L5A L133A (black). Bottom: As above but with a non-nicked template. (D) The Clamp is required for P_mom activation. As in B (Top), + and – Clamp addition. The solid line indicates the mean of three replicates, and the circle the SD.

We conclude from the in vitro data that the activation of MuC transcription requires the Clamp and a substrate that is nicked on the nontemplate strand.

Discussion

We report here the finding that phage Mu appropriates the host β Clamp, a DNA replication processivity factor, for transcription of its late (and likely also middle) genes by the host RNAP, thus linking its replication to transcription. Starting with the fortuitous observation that the late gene–specific transcription factor MuC is lethal when overproduced (Fig. 1B), we first showed that host replication was overinitiating under these conditions (Fig. 2A). Suppressors of the lethality indeed mapped to genes involved in replication initiation—dnaA, diaA, dnaX (Table 1). Several experiments revealed that overinitiation was due to accumulation of ATP-DnaA, and thus interference with RIDA (Fig. 1 B and D and SI Appendix, Fig. S1). Lethality was rescued by complementation with both Hda and DnaN, proteins that maintain RIDA by stimulating the ATP-hydrolysis activity of DnaA (Fig. 2E). This prompted us to closely examine the MuC sequence, where we found not one, but two CBMs (Fig. 3A). Indeed, MuC interacted with DnaN in the BACTH assay (Fig. 3B), and mutating any one of the CBMs abrogated MuC lethality (Fig. 3C), showing definitively that MuC not only interacts with the Clamp, but that this interaction was the cause of lethality when MuC was overproduced. We then showed that the Clamp was required for Mu late gene transcription in vivo (Fig. 4 A and C) and for Mu amplification (Fig. 4B). Finally, we set up an in vitro transcription assay, demonstrating that the Clamp stimulated transcription from a Mu late promoter and that this stimulation required a nick on the nontemplate strand (Fig. 5). These findings provide an explanation for the MuC-lethality phenotype: excess MuC soaks up available Clamps, interfering with RIDA, which leads to overinitiation (Fig. 6).

Fig. 6.

Model for MuC lethality. MuC interacts with the clamp to activate transcription from late gene promoters during Mu growth. In the absence of late promoters, MuC overexpression causes a depletion of Clamps that are involved in RIDA. Perturbation of RIDA leads to overinitiation of replication and to lethality.

While the model proposed in Fig. 6 explains our major findings, we are puzzled at why the MuC defect was restricted to replication initiation and not elongation, given the major role of the Clamp in elongation (40, 50). Perhaps MuC is unable to access the pool of Clamps that are engaged in replication because the protein binding pocket of the Clamp is constantly occupied by either Pol III or other Clamp-interacting partners. If it is simply the case that MuC is depleting the pool of available Clamp for RIDA, we wonder why overexpression of dnaN alone was not sufficient to rescue MuC lethality (SI Appendix, Fig. S2D), and that Hda was also required (Fig. 2E); perhaps Hda activity or expression is somehow compromised by MuC interaction with the Clamp.

Phage T4 Employs a Mechanism Analogous to Mu.

The utilization of the Clamp for transcription by MuC is reminiscent of, and perhaps analogous to, the coupling of DNA replication to activation of late gene transcription of phage T4, which encodes its own replication apparatus, borrowing only RNAP from the host (27). DNA structures transiently generated by T4 replication in the form of nicks or primer/template junctions promote the loading of the Clamp by a Clamp loader, both encoded by T4. Activation of transcription requires two RNAP binding proteins (gp33/gp55), which could be thought of as a split sigma factor (27, 51, 52), the T4 Clamp (gp45) (53), and the T4 Clamp loader complex (gp44/gp62) (27).

There are several analogies between the T4 and Mu systems: Like MuC, the gp33/gp55 promoters have a recognizable -10, but a -35 consensus is not evident (22, 23). T4 and E. coli Clamps are similar in that they both form a doughnut-like structure which encircles the DNA (54, 55). The nicked DNA or primer/template junctions for Clamp loading in T4 (27) are available during T4 replication by its own DNA Polymerase. Mu on the other hand generates only one nick at each end from its nick-join mechanism of transposition and uses the E. coli machinery for replication (1, 3). The requirement for the Clamp and Clamp loader in transcription activation is absolute for both systems in vivo (Figs. 4D and SI Appendix, S4B) (27). In vitro, the T4 Clamp loader stimulated transcription substantially, but could be omitted provided that a crowding agent (such as PEG or Ficoll) was included in the reaction (56). These results indicate the essentiality of the Clamp for linking replication to transcription of late gene products in both phages.

Despite the similarities between the T4 system and the Mu late gene transcription apparatus, the protein players are nothing alike. The gp33/gp55 complex effectively replaces σ⁷⁰, while MuC requires σ⁷⁰ to recognize late promoters (27, 57). Furthermore, Gp33 alone represses basal transcription from T4 late promoters (27) while Mu has no known equivalent protein. A structural approach would provide insights into how such dissimilar systems at the protein-component level produce similar activation mechanism(s).

Significance of Using the Replication Clamp for Transcription.

Phages employ sophisticated mechanisms to ensure that their genetic information is transcribed and translated efficiently and in a timely manner, allowing for the successful production of progeny phages before the host cell is destroyed. They do this by encoding proteins that modify the host RNAP to alter promoter specificity to prioritize transcription of phage genes (58), by subverting the function of host sigma factors to redirect RNAP to phage-specific promoters, and by encoding proteins that help RNAP read-through terminators, ensuring timely expression of phage genes (59). Using the Clamp to enable transcription appears to be an unusual strategy reported first for T4 and now for Mu—two phages that could not be more different. T4 is a lytic phage with a large genome (169 Kbp) while Mu is a temperate phage with a moderate genome size (37 Kbp). T4 initiates DNA replication from specialized structures that form in its genome at different times: R-loops at initial times where the annealed RNA is a primer for leading-strand synthesis in one direction, and D-loops at later times when replication initiation becomes dependent on recombination proteins leading to formation of branched recombination intermediates (60), providing ample opportunities for Clamp loading. On the other hand, the nick-join mechanism of Mu transposition leaves only one nick in the target DNA that is used to prime replication across Mu (1). T4 has ~40 late promoters, while Mu has only 4 (23, 27). It is surmised that by sliding along the large genome, the T4 Clamp allows the transcription machinery to efficiently find late promoters, which might be located far from the Clamp-loading sites, a problem that Mu does not share with T4. Oddly, the only attribute these two phages share is that they are both Myoviruses, which use contractile tails to inject their genetic material into the host (21, 61). The Mor/MuC family of transcription factors identified by BLAST-P as containing CBMs (SI Appendix, Fig. S3F), provide no clue about the morphology of phages that harbor them. Could it be that the Clamp has an additional role in T4 that is shared with Mu? An in vitro T4 DNA packaging assay found that packaging is coupled to the replication-dependent late transcription machinery (62), i.e. requires gp45, gp44/gp62, and gp33/gp55.

Materials and Methods

For a more detailed description, please refer to SI Appendix, Expanded Materials and Methods.

Media, Strains, Phages, and Plasmids.

Unless conditions are specified, all strains are grown in LB at 30 °C with shaking and supplemented with the appropriate antibiotic(s). Induction of genes under control of P_lacZ was done with IPTG (Fisher), and induction of P_tet was achieved with anhydrotetracycline (aTc). 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) and o-nitrophenyl-β-D-Galactoside (ONPG) was purchased from Sigma. Cells were routinely transformed using either electroporation or heat shock. Mu lysate was stored in Mu Buffer. Strains and Plasmids employed in this study are listed in SI Appendix, Table S1. Primers, purchased from Integrated DNA Technology (IDT), used in this study are listed in SI Appendix, Table S2.

General Molecular Techniques.

Routine PCR reactions were performed with Taq DNA Polymerase (NEB), according to the manufacturer’s instructions. PCR fragments for plasmid or strain construction were generated with Phusion DNA Polymerase (NEB) according to the manufacturer’s instructions. Gibson Master Mix (isothermal) was made according to (63). gRNA plasmids for editing MuC were generated with Golden Gate.

Plasmid Construction.

Plasmids were constructed using Gibson assembly. Successful clones were verified by sequencing the junction PCR product by either Eton Biosciences or Azenta (GeneWiz). For generating point mutants, primers carrying the desired mutation were used to amplify the plasmid of interest followed by self-ligation. Positive clones carrying the desired mutation were identified by PCR of the target sequence followed by sequencing.

Strain/Prophage Construction.

For construction of deletion strains, pKD4 was used as the template, and the procedure was based on (64). Construction of Mu prophage with desired point mutants in the N terminus was based on (65). For C-terminus mutants, the desired mutations were introduced into the homology ends of the primer, and the PCR product was used to insert the substitution into the prophage as described for deletion strains.

Flow Cytometry.

The Sytox Green staining procedure to determine chromosome equivalents was based on published methodology (30, 32), cells were subjected to fixation and subsequent staining and analyzed on LSRFortessa SORP Flow Cytometer (BD Biosciences) using filter settings for FITC. Chromosome-equivalents were determined by using cells from an overnight culture where the copy number is 1.

MuC Lethality Assay.

Strains to be examined were grown overnight in LB at 30 °C in the appropriate antibiotic(s). The cells were washed and plated with the appropriate antibiotics and inducer(s) or were diluted in liquid culture supplemented with antibiotics and inducer.

Isolation, Sequencing, and Analysis of MuC Suppressors.

Mutants were selected on LB agar with appropriate antibiotics and inducer. Mutants were confirmed by restreaking and confirming that the plasmid retains lethality in a WT strain. Sequencing was performed by Novogene on the NovaSeq platform, paired-end reads 150. Reads were mapped to the genome of BW25113 (GCA_000750555.1) using default parameters on Geneious. SNPs, indels, and large rearrangement were detected with Geneious using default parameters. SNPs and indels present in the parent strain were discarded from analysis. SNPs and indels were scored as a hit if the variant frequency was greater than 75%. Mutants and the associated mutations are shown in Table 1.

Mu Phage Preparation.

When required, lysogen of the indicated prophage was grown overnight, diluted 1:100 in fresh LB, shifted to 42 °C until lysis was complete (typically 1 h). Chloroform was added and the aqueous phase was collected (top layer) and used to assay for plaquing.

Miller Assay.

See SI Appendix, Expanded Materials and Methods.

Alignment of the Mor/MuC Family of Transcription Factors.

BlastP was performed on MuC sequence (66). Aligned sequences (100 total) were downloaded and aligned with ClustalW on Geneious. Sequence logo threshold for the consensus was set to 95%.

Temperature Shift Experiments.

MuC-specific transcription was assayed with pKH33 (P_mom-lacZ). σ⁷⁰ transcription was assayed with pKH34 (P_tet-lacZ). BW25113 carrying desired alleles in dnaNts or dnaXts was transformed with the appropriate plasmid. To measure transcriptional activity, cells were washed, diluted, and grown to an O.D. _600nm of 0.2 at permissive temperature, before being either shifted to nonpermissive temperature or kept at permissive temperature. Miller assay was performed for each condition.

BACTH (Bacterial Adenylate Cyclase-Based Two-Hybrid) Assay.

The host strain for the assay, BTH101, was transformed with indicated plasmids and selected on plates supplemented with appropriate selection. Single colonies were picked and grown overnight in LB with antibiotics. The plates were incubated at 30 °C for 16 to 24 h, since color-development time appeared to be variable. For quantification of β-galactosidase activity, cells on the plate were scraped off and diluted to an O.D. between 0.4-0.6 in PBS. Miller assay was performed on this dilution as described above.

Protein Purification and In Vitro Transcription.

See SI Appendix, Expanded Materials and Methods.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.