|Home | About | Journals | Submit | Contact Us | Français|
Transcription proceeding downstream of the λ phage replication origin was previously shown to support initial steps of the λ primosome assembly in vitro and to regulate frequency and directionality of λ DNA replication in vivo. In this report, the data are presented indicating that the RNA polymerase β subunit makes a direct contact with the λO protein, a replication initiator of λ phage. These results suggest that the role of RNA polymerase during the initiation of λ phage DNA replication may be more complex than solely influencing DNA topology. Results demonstrated in this study also show that gyrase supercoiling activity stimulates the formation of a complex between λO and RNA polymerase, suggesting that the introduction of negative supercoils by DNA gyrase, besides lowering the energy required for DNA strand separation, may play an additional role in modeling protein–protein interactions at early steps of DNA replication initiation.
Bacteriophage λ DNA replication starts from binding of λO replication initiator protein to four partially symmetrical iterons present at oriλ (1). Interaction of λO with the origin sequence and formation of higher-order nucleoprotein structure induces DNA helix destabilization within the AT-rich region positioned immediately to the right from λO-binding sites (2,3). Subsequent steps, involving delivery of the host DnaB helicase to preprimosomal complex by the phage λP protein (4,5) and liberation of the helicase from inhibitory interaction with λP (6,7) by the chaperone proteins DnaK, DnaJ and GrpE, are well characterized biochemically (5,8,9). After the rearrangement of the λO–λP–DnaB complex by the heat shock proteins, replication starts in the presence of primase, DNA polymerase III, gyrase and SSB proteins (6,10).
In vivo and in a crude extract, initiation of replication of λ phage and plasmid DNA replication is dependent on transcription carried by bacterial RNA polymerase. In a system consisting of purified enzymes this dependence of λ DNA replication on ‘transcriptional activation’ is alleviated (11); however, it can be restored by the addition of bacterial histone-like protein, HU (12). HU blocks initiation of replication starting from oriλ in vitro in the absence of RNA polymerase activity (12). In the case of wild-type λ sequence, activation of oriλ is provided by transcription initiated at the pR promoter, located ~1000bp upstream of the origin of replication. However, in vivo results demonstrated that replication of oriλ-containing plasmids was still efficient when the promoter was situated much closer to the replication initiation point or even downstream of it (13,14). Activation of oriλ by the transcription proceeding in its vicinity plays a key role in the control of frequency of replication initiation and the switch from θ to rolling circle mode during λ phage DNA replication (15–18).
Similar dependence of the replication initiation on the transcriptional activation was found in the case of the host chromosomal DNA replication (19). DnaA, a bacterial replication initiator protein, binds to specific DnaA boxes within oriC and unwinds the DNA duplex in an AT-rich region. This reaction is stimulated by HU protein and transcription process; however, as in the case of λ DNA replication, high concentrations of HU are inhibitory to DNA unwinding step, but RNA polymerase activity overcomes this negative influence (20). Allele-specific suppression of certain dnaA (ts) mutations by the changes in the rpoB gene, coding for the β subunit of RNA polymerase, suggested direct interaction between DnaA and RNA polymerase (21), which has been recently confirmed also in vitro (22). Observation that mutations in the rpoB gene suppress, in an allele-specific manner, the inability of some λ phage P mutants to plate on certain dnaB strains, implied also the possibility of direct protein–protein interactions between RNA polymerase and the λ replication complex (23).
λO replication initiator binds as a dimer to a partially symmetric sequence and induces a strong DNA bend (24,25). Interaction of λO with four iterons within oriλ results in the formation of a large nucleoprotein complex (O-some) and condensation of a DNA fragment encompassing all iterons and their intervening spacers (2). It was also proved that λO, by binding to its recognition sequences, greatly stimulates transcription-coupled supercoiling of plasmid DNA by the gyrase. This property of λO was attributed to its capability of inducing strong DNA bends (26). The process of transcription-dependent stimulation of supercoiling was proved to rely on the twin-domain model (27), according to which counter-rotation of the transcription machinery and template, as DNA is threaded through RNA polymerase active site, generates waves of supercoiling, positive in front of elongating enzyme, and negative behind it. This concept, originally postulated by Liu and Wang (28), became an attractive mechanism for regulation of local processes involving DNA transactions, by influencing DNA topology in a sequence- and context-dependent manner (29,30). Transcription was also shown to regulate chromatin structure and, thus, selection of replication origin sites and replication initiation timing during eukaryotic chromosomes doubling (31 and references therein). Therefore, studies on coordination of replication and transcription machineries seem to be of general importance.
In this report, we demonstrate evidence that the λO protein, like the host replication initiator, DnaA, interacts directly with bacterial RNA polymerase. Moreover, the results of our studies indicate that the β subunit of RNA polymerase is a contact site for λO. In addition, we present data suggesting that gyrase activity enhances this interaction.
Escherichia coli, strain C600 (supE44 hsdR thi-1 thr-1 leuB6 lacY1 tonA21) (32), was used for plasmid purification and strain MM294 (supE44 hsdR endA1 pro thi) (33) was used for the λO and λP protein overexpression.
The following plasmids were employed: pRLM4 (oriλ, cro(ts), kanR) was described previously (34); pLamberA (oriλ, colE1, ampR) was constructed by removing the 916-bp HindIII–NcoI fragment from a plasmid pLamber (35); plasmids: pGP1-2 (T7 RNA polymerase gene under the control of λ repressor CI 857(ts); kan R) (36) and pEW1 (O and P genes under the control of a T7 phage promoter) (37) were used for overproduction of the λO and λP proteins.
Both λO and λP proteins were prepared from E. coli strain MM294 bearing plasmids pGP1-2 and pEW1. Proteins were purified according to a method described previously (38). Other proteins and antibiotics were from the following sources: E. coli DNA gyrase was obtained from New England BioLabs. Bovine serum albumin (BSA) and novobiocin were from Sigma and E. coli RNA polymerase holoenzyme was provided by Eppicentre.
The following antibodies were used in this study: a polyclonal rabbit antibody against the λO protein (39) and monoclonal mouse antibodies, which were specific for α, σ, β and β′ subunits of E. coli RNA polymerase, were obtained from Neoclone; secondary HRP-conjugated anti-rabbit and anti-mouse antibodies were obtained from Sigma.
To analyze a direct interaction between λO protein and E. coli RNA polymerase, surface plasmon resonance (SPR) analyses were performed using a double-stranded 141-bp linear DNA fragment containing the four iterons from oriλ. The DNA fragment was obtained by polymerase chain reaction (PCR) amplification with following program: initial denaturation at 94°C for 3min; 36 cycles of denaturation at 94°C for 20s; annealing at 55°C for 30s and extension at 72°C for 30s. The following primers were used: forward: 5′ - biotin - TCA AGC AGC AAG GCG GCA TGT TTG G-3′ and reversed: 5′ TGT CCC CCT GTT TTG AGG GAT AG-3′. Then, the DNA fragment in TNE buffer (10mM Tris pH 7.6; 300mM sodium chloride and 1mM EDTA) was immobilized on a streptavidin matrix-coated Sensor Chip S.A. (Biacore) by biotin covalent linkage, following the manufacturer’s instructions. Experiments were carried out on a Biacore 2000 instrument by a 5-min injection of λO protein (50nM) in HBS buffer (10mM HEPES- KOH pH 7.4; 150mM sodium chloride; 3mM EDTA and 0.005% surfactant P20), followed with a 5-min injection of E. coli RNA polymerase (25nM) and BSA (BSA; 25nM) in binding buffer (10mM HEPES pH 7.6; 10mM MgCl2; 20mM KCl; 0.5mM DTT and 0.1mM EDTA). Next, the HBS buffer was injected over the chip surface and the dissociation phase was recorded. The analyses were performed at 15µl/min flow speed. Following the completion of each protein co-injection, the surface of the chip was regenerated by applying 0.2% sodium dodecyl sulfate (SDS), which releases all bound protein without affecting the binding capacity of the immobilized DNA. Data were evaluated using Biacore AB’s BIA evaluation software. The results are presented as the sensogram, obtained after subtraction of the background response signal and correction of the buffer effect. As a control of nonspecific interactions, an empty reference cell was used.
Reactions were performed in the buffer containing: 10mM HEPES-KOH pH 7.6; 10mM MgCl2; 20mM KCl; 0.5mM DTT, 0.1mM EDTA and 50µg/l poly (dI:dC). Supercoiled λ plasmid DNA (2.8nM) (purified by ultracentrifugation in a cesium chloride/ethidium bromide gradient) was mixed with λO protein (0.5µM) and after a 10-min incubation at 30°C reaction was applied onto a 1-ml Sepharose 4B-CL column, equilibrated with a reaction buffer. DNA–λO complexes were separated from unbound protein fraction. Next, to the reaction mixture RNA polymerase (22nM) and NTPs (0.5mM each) were added. After a 10-min incubation at 30°C, protein complexes were cross-linked with disuccinimidyl glutarate (DSG) reagent (a final concentration of 0.02µM) and incubation was prolonged for an hour at 30°C. Reactions were quenched by addition of Tris–HCl pH 7.5 to 50mM and incubation for next 10min at 30°C. Subsequently, the cross-linked complexes were separated electrophoretically in 7.5% SDS-polyacrylamide gels, transferred onto polyvinylidene fluoride (PVDF) membrane (Millipore) and detected by immunoblot analysis using anti-λO polyclonal antibodies or monoclonal antibodies specific for various subunits of E. coli RNA polymerase. An ECL-Western Blotting Substrate Kit (Pierce) was used to detect antibody–antigen complexes.
Size-exclusion chromatography was performed to investigate the influence of transcription on pre-formed O-some complexes. The Sepharose 4B-CL (Sigma) column (0.5×8cm) was equilibrated with a buffer containing: 10mM Tris–HCl pH 7.6; 10mM MgCl2; 20mM KCl; 0.5mM DTT; 0.1mM EDTA. The 120-µl reaction mixture (in the same buffer) supplemented with 10µg of the supercoiled pRLM4 plasmid and 6µg of λO protein was incubated for 10min at 30°C and loaded on a Sepharose 4B-CL column at room temperature. Fractions containing O-some were collected. The presence of plasmid DNA in the O-some fractions was initially confirmed by ethidium bromide staining and visualization under ultraviolet (UV) light. During subsequent repetitions of the experiment, the presence of λO in the void volume fraction was confirmed by amide black staining; fractions containing λO were pooled and divided into two parts. Subsequently, to one of the O-some-containing samples (~100–120µl) 2.5µg of RNAP polymerase was added. After 10min at 30°C, rNTPs (2mM each) were added to the mixture and incubation was continued for the next 10min at 30°C. The reaction mixture containing only O-some was also incubated at 30°C. Finally, the two mixtures (separately) were loaded on a Sepharose 4B column. Two-drop fractions were collected and analyzed by SDS- polyacrylamide gel electrophoresis (PAGE), followed by silver staining. Each fraction content was quantitated densitometrically and compared to a sample containing known amount of the λO protein.
λO protein binding to its DNA recognition sequences is a primary step in the initiation of DNA replication. Nucleoprotein structure called O-some, formed due to λO oligomerization and its DNA-bending properties, serves as a landing pad for other proteins involved in the initiation of replication. As RNA polymerase activity may protect λ replication complex from disassembly during unfavorable conditions (40), we wanted to test whether RNA polymerase binding to promoter regions present in λ plasmid DNA (mainly to the strongest pR promoter) and transcription activity may influence O-some stability or the affinity of λO protein binding to the oriλ-containing plasmid DNA. To study this possibility closer, we performed reactions, using supercoiled λ plasmid DNA, and purified proteins: λO and RNA polymerase. λO–DNA complexes, formed in the presence and absence of active RNA polymerase, were resolved by gel filtration (Sepharose 4B-CL). To test whether efficient transcription occurred under the conditions we used, total RNA synthesis was tested by measuring amount of [α-32P]UTP incorporated into acid-insoluble material (data not shown). Subsequent analysis of fractions, containing λO bound to DNA sequences, demonstrated that the formation of such a nucleoprotein structure is enhanced by the transcription process. Assembly of O-some was stimulated by 70% in fractions containing RNA polymerase and nucleotides, in comparison to those where only λO and DNA were present (Figure 1A). In addition, the formation of nucleoprotein complexes by the O-protein was facilitated by RNA polymerase even in the absence of transcription process, although this effect was less pronounced than the one observed in the reactions where RNA polymerase was allowed to transcribe (Figure 1A). This result suggested that RNA polymerase could facilitate λO protein binding to the oriλ-containing plasmid. However, this result could also mean that, under the conditions we used, RNA polymerase activity enhanced stability of the formed O-some structure. To verify the latter hypothesis, we performed analogous gel filtration experiment in which nucleoprotein complexes of the λO protein were formed without RNA polymerase and separated by size-exclusion chromatography from unbound protein. Subsequently O-some containing fractions were collected and divided into equal portions. One of the samples was supplemented with RNAP and NTPs and both were subjected to a second round of gel filtration on separate columns. The amount of λO-containing nucleoprotein complexes was similar in both cases (Figure 1B), indicating that RNA polymerase may play a role in λO binding to DNA, rather than in stabilization of O-some structure during chromatography. Additionally, an experiment was performed in which pre-formed O-some complex were challenged with an excess of biotinylated DNA competitor (containing iteron sequences) in the presence and absence of RNA polymerase and transcription process. The amount of λO protein bound to competitor DNA was monitored by the separation of biotinylated DNA using magnetic beads, SDS-PAGE and blotting. The results of this experiment confirmed that RNA polymerase and transcription process do not substantially influence O-some stability (for details see Figure S1 in Supplementary Data). Stimulation of the λO protein binding to iterons by the presence of RNA polymerase and also by transcription process was further confirmed by electrophoretic mobility shift assay (EMSA) performed with supercoiled λ plasmid (see Figure S2 Supplementary Data).
During experiments testing the influence of RNA polymerase activity on λO protein binding to oriλ, we obtained evidence suggesting that the λO initiator may interact directly with the host transcription machinery. In addition, similarity of the initial steps of replication at the E. coli and phage λ origin with respect to their dependence on transcriptional activation and recent confirmation of the direct contact between the host initiator protein DnaA and RNA polymerase (RNAP) (22), as well as genetic data suggesting that transcriptional machinery may directly associate with the components of the phage replication complex (23), encouraged us to follow that trail. Therefore, we analyzed the formation of the putative complex between RNAP and the λO protein bound at the iterons, using SPR technique. A DNA fragment encompassing all four λO binding sites of oriλ was linked to the sensor chip. λO and RNAP were injected sequentially over the sensor matrix and the signal, reflecting a change of mass on the immobilized DNA fragment, was measured. Analysis of the sensograms showed that λO bound efficiently to the sensor-attached DNA. Subsequent injection of RNAP resulted in detection of RNA polymerase binding to the pre-formed λO–DNA complex (Figure 2). The response signal, seen after injection of RNAP, was relatively high, indicating that the complex between the host and phage protein is formed with high binding affinity. As RNA polymerase displays some unspecific binding affinity to double-stranded DNA, additional control was made to assess the enzyme binding to sensor-attached DNA, containing oriλ sequence. In the absence of the λO protein, RNA polymerase bound to the DNA fragment; however, this interaction was considerably weaker than that observed with O-containing nucleoprotein complex (Figure 2).
To further confirm the interaction between λO and RNA polymerase in vitro, cross-linking experiments with DSG were performed. DSG is a homobifunctional protein cross-linking reagent with amine-reactive NHS esters at both ends of the 7.7-Å long arm. The λO protein was incubated together with the supercoiled λ plasmid DNA and, subsequently, the fractions containing ‘O-some’ were separated by gel filtration from the unbound protein. The plasmid contained all λO-binding sites and λ phage promoters in their native orientation and relative position. ‘O-some’ fractions were collected and divided into individual reactions, supplemented with RNAP and nucleotides before adding DSG. The cross-linked complexes were separated in SDS-PAGE gels, blotted and probed with antibodies against λO or RNA polymerase (specific against subunits α, σ, β and β′) (Figure 3A and B). Monomeric form of the λO protein (32kDa) was observed when no DSG was present in the reaction (Figure 3A, lane 1). In a control reaction without DNA, large fraction of λO migrated, as expected, at the dimer position in the presence of the cross-linker (Figure 3A, lane 3). DNA-bound λO protein cross-linked, forming several species of increasing molecular mass, representing λO oligomers (lane 6). When a mixture of ‘O-some’ and RNAP was subjected to cross-linking, the formation of a new complex was observed, absent from the sample containing only λO and DNA. This complex migrated at the position similar to RNAP β subunit and was also detected in a sample containing nucleotides (Figure 3A and B; lanes 7 and 8). Under these conditions, no discrete cross-linked complex of RNAP subunits was detected on a blot subjected to anti-RNAP antibodies, in the reactions where only RNAP was present, regardless of the presence or absence of plasmid DNA (Figure 3B, lanes 4 and 5). Instead, a portion of RNAP formed large cross-linked complexes that did not enter the gel. In addition, an increase in the efficiency of formation of the λO dimer and trimer was observed in the presence of DSG and RNA polymerase, in comparison to the sample where only ‘O-some’ was subjected to cross-linking (Figure 3A compare lanes 6 and 7). Moreover, much more of a large oligomeric form of the λO protein was also detected in the presence of RNA polymerase and transcription process (Figure 3A, compare lanes 6, 7 and 8). This cross-linked oligomer had an apparent molecular weight of about 250kDa, matching the molecular weight of four λO protein dimers, present in the O-some structure. These results may suggest that some changes in the conformation of O-some structure occur in the presence of RNA polymerase activity.
Due to complexity of the cross-linking pattern obtained with the multisubunit RNA polymerase and λO and the difficulty to distinguish individual complexes after reaction with DSG, we performed that experiment using varying concentrations of the cross-linking reagent. When the efficiency of cross-linking reaction was sufficiently high, a large portion of reacting proteins was lost from the final picture as a result of the formation of high molecular mass complexes, unable to enter the gel. However, under such conditions, we were also able to identify two products of the cross-linking reaction: λO dimer and the second one, resulting from the interaction between λO and RNA polymerase (>250kDa), (Figure 4A–C). Subsequent analysis of cross-linked species by immunoblotting, performed with purified monoclonal antibodies specific for each of the subunits of RNA polymerase (Figure S3 Supplementary Data), revealed that the high molecular mass cross-linked product cross-reacted with antibodies directed against both λO and RNAP β subunit and most probably represents λO tetramer cross-linked to the β subunit. Moreover, we observed the formation of such a complex even in the absence of DNA, which suggest that both proteins are able to interact in solution, even in the absence of specific DNA sequences recognized by these proteins, although with a lower efficiency than in the case of DNA-bound molecules (Figure 4A–C, lane 4). RNAP β subunit alone could form no complex with the λO protein, detected after cross-linking (Figure 4A–C, lane 7). This is most probably due to altered conformation of the free β subunit in comparison to that observed when β is embedded in the RNAP core structure. These results corroborate conclusions from the SPR experiment that the λO initiator protein is able to interact directly with the host RNA polymerase in vitro and indicate that the β subunit of RNA polymerase is a target for interaction with λO.
Results of the experiments presented to this point suggest that λ phage replication initiator protein, λO, interacts with the host RNA polymerase. This conclusion may have important implications for the role of transcription in activating replication initiated by the λ phage proteins. During transcription elongation, RNA polymerase introduces topological changes into DNA template, positive supercoils are generated ahead of the transcription machinery while negative are left in its wake (28). Transient waves of supercoiling produced by RNAP can be stabilized by the action of DNA topoisomerases. It was demonstrated that DNA gyrase rapidly converts transcription-induced positive supercoils into the negative ones (41,42). Moreover, λO protein was shown to greatly enhance transcription-coupled negative supercoiling of plasmid DNA by the gyrase in in vitro reaction (26). Therefore, we aimed to test whether gyrase activity may influence the formation of the λO–RNA polymerase complex in a reaction consisting of purified proteins and a supercoiled λ plasmid DNA. To assess this possibility, λO–plasmid DNA nucleoprotein complexes were isolated and supplemented with RNAP and/or E. coli gyrase and nucleotides. The formation of the complex between RNAP β subunit and the λO protein was analyzed by cross-linking with DSG and subsequent immunoblotting (Figure 5A). In the presence of the active RNA polymerase, a band corresponding to λO–RNAP cross-linked product was observed (Figure 5A, lane 6). Addition of DNA gyrase to the reaction, stimulated interaction between RNA polymerase and λO in a manner dependent on adenosine triphosphates (ATP) concentration (Figure 5A, lanes 7 and 8). Novobiocin, in turn, reduced this stimulatory effect (Figure 5A lane 9). This aminocoumarin antibiotic, which is a competitive inhibitor of the gyrase ATP-ase activity, also inhibits supercoiling activity of gyrase, which is exerted at the expense of ATP. We also tested λO–β subunit complex formation in the presence of transcription and gyrase activity at various ratios of RNAP to the λO protein and at higher concentration of cross-linking reagent, yielding higher molecular mass cross-linked product (Figure 5B). In this case, we also observed the enhancement of the λO–β subunit conjugation by DSG, in the reactions where gyrase was present, in comparison to the ones where transcription was allowed to proceed in the presence of the λO protein only. Stimulation of the λO–β complex formation was assessed densitometrically and summarized in the Figure 5C. These results indicate that ATP-dependent gyrase activity may facilitate the contact between the λO replication initiator and the host RNA polymerase.
Transcription was shown to play a pivotal role in the regulation of bacteriophage λ replication. In vitro, transcription was necessary in the step preceding preprimosome rearrangement by the heat shock proteins and this process was also shown to be able to overcome inhibitory action of the histone-like HU protein (12,43). Both in vivo and in vitro, activity of RNA polymerase stimulated bidirectional replication starting at the λ origin (16,44). In addition, it was demonstrated that efficient transcription process was required for the λ replication complex to survive heat shock-provoked disassembly by GroELS proteins after transient DNA relaxation by gyrase (40). These results suggested that transcription stabilizes replication complex under unfavorable conditions.
Results presented in this work strongly indicate that the λO protein interacts directly with bacterial RNA polymerase (Figures 2 and and3).3). On RNAP part, β subunit seems to be responsible for the contact with λ phage replication initiator (Figure 4), similarly as it was suggested for the host initiator DnaA protein (21,22). This interaction seems to facilitate λO protein binding to iteron-containing DNA as we observed enhanced formation of λO nucleoprotein complex even in the absence of nucleotides (data not shown) and transcription process further supported association of λO with oriλ (Figure 1). It is unlikely that this stimulation plays a primary role in the transcriptional activation process; however, it may be helpful under conditions of limiting λO protein. Although we used under mild conditions, O-some structure is very stable and therefore we did not observe any enhancement of this nucleoprotein complex stability by transcription (Figure 1B); it cannot be excluded that direct interaction of RNA polymerase and λO may support replication complex survival under more harsh circumstances. λO–RNAP association may also help in organizing replication hyperstructure consisting of phage and host proteins necessary for phage DNA replication and progeny production, in keeping with the model of a ‘factory’ where many proteins are engaged in various DNA transactions (45).
It was demonstrated that λO protein, most probably due to its DNA-bending properties, stimulates transcription-coupled DNA supercoiling by gyrase in vitro (26,27). Here, we show that gyrase activity—in turn—stimulates formation of a complex between RNA polymerase β subunit and the λO protein. We have not determined whether this stimulatory influence of gyrase relies on, connected with supercoiling, increase in probability of juxtaposition of RNA polymerase and λO position on DNA. Supercoiling could also actively affect conformation of the λO complex in a way favoring the interaction. Topological state of the loop formed by Lac repressor bound to adjacent operator sites was also shown to influence repressor conformation (46). Similarly, DNA supercoiling was implicated in assisting in octamerization of the λ CI protein and, hence, facilitating autoregulatory repression of pRM promoter by aiding CI in binding to distant OL3 and OR3 operators and DNA looping (47,48). In addition, it was demonstrated that several promoters are sensitive to the supercoiling level (49), so the stimulatory effect of gyrase activity on the formation of RNA polymerase–λO complex may result from altered transcriptional activity of promoters located on plasmid DNA. Whatever mechanism operates in gyrase-mediated stimulation of λO and RNA polymerase interaction, this observation emphasizes the importance of DNA supercoiling, not only in lowering the energy necessary for DNA strand separation but also in modeling of protein–DNA and protein–protein interactions.
The Ministry of Science and Higher Education of Poland (grant No. N N301 4140 33). Funding for open access charge: The University of Gdańsk.
Conflict of interest statement. None declared.
Supplementary Data are available at NAR online.
We are particularly grateful to Prof. Igor Konieczny (Faculty of Biotechnology, University of Gdańsk) for his advice regarding SPR experiments.