To characterize the par system of Corynebacterium glutamicum pCGR2 and to manipulate the par components to effectively manipulate plasmid copy number.
To characterize the par system of Corynebacterium glutamicum pCGR2 and to manipulate the par components to effectively manipulate plasmid copy number.
ParB binds sequence specifically to centromere-binding sites around the parAB operon and serves as an autorepressor. A small ORF (orf4, later named parC) downstream of parAB encodes a protein with 23·7% sequence identity with ParB. ParB is also implicated in the repression of parC transcription. Nonetheless, this ParC protein does not bind to centromere-binding sites and is not essential for plasmid stability. Introduction of a frameshift mutation within ParC implicated the protein in regulation of both parAB and parC. Electrophoretic Mobility Shift Assay confirmed a previously unreported ParC–ParB–parS partition complex. ParC also interacts directly with ParB without the mediation of the centromere sites. Deletion of the par components resulted in different plasmid copy numbers.
A previously unreported ParC–ParB–parS partition complex is formed in pCGR2, where interaction of ParC with ParB–parS may affect the level of repression by ParB. Modifying the par components and antisense RNA enables manipulation of plasmid copy number to varying degrees.
Genetically manipulating the par components, in combination with deactivation of antisense RNA, is a novel approach to artificially elevate plasmid copy number. This approach can be applied for development of new genetic engineering tools.
Bacterial low-copy-number plasmids ensure their stable segregation to daughter cells by utilizing partition (par) systems during cell division (Ghosh et al. 2006). Partition cassettes generally consist of three components; (i) two operon genes, one encoding a motor protein (typically an ATPase) and the other a centromere-binding protein, and (ii) a centromere-like site that is generally located either upstream or downstream of the par operon (Gordon and Wright 2000; Hayes and Barilla 2006b; Schumacher 2007). Type I par systems employ Walker-type (P-loop) ATPases and can be further divided into Ia and Ib subgroups based on characteristics such as location of the centromere-like site, mechanism of transcriptional regulation and size of the par genes, whereas type II par systems employ actin-like ATPases as the motor protein (Gerdes et al. 2000). The partition reaction generally proceeds first by assembly of the centromere-binding proteins at the centromere site to form the partition complex, followed by recruitment of the motor protein to create an active segregosome to mediate plasmid separation (Hayes and Barilla 2006a). The molecular mechanism of type II partitioning is well understood with R1 plasmid (Garner et al. 2004; Becker et al. 2006; Campbell and Mullins 2007), but that of type I loci is less well understood. Although centromere-binding proteins have little sequence homology even within a given family, the proteins share common DNA-binding motifs: helix-turn-helix (HTH) motifs in type Ia (Leonard et al. 2004; Schumacher and Funnell 2005), and ribbon-helix-helix (RHH) motifs in types Ib and II (Weihofen et al. 2006; Møller-Jensen et al. 2007; Schumacher et al. 2007).
Corynebacterium glutamicum is a high-GC content, gram-positive, traditional industrial amino acid-producing bacterium (Kinoshita et al. 1957). The pCG1 family represents the largest plasmid family of Corynebacterium species. The family so far includes a total of 8 plasmids including the small cryptic plasmids (approx. 3–5 Kb; pCG1, pGA1 and pSR1) as well as larger plasmids (approx. 20–30 Kb; pCG4, pAG1, pGA2, pTET3 and pCGR2) (Okibe et al. 2010). Fourteen additional pCG1 family plasmids have so far been reported for the entire Corynebacterium genus. Except for a few small pCG1 plasmids, the presence of putative partitioning genes in Corynebacterium pCG1 plasmids appears to be widespread (Okibe et al. 2010). Involvement of the type Ib partitioning system in the larger Coryne. glutamicum pCG1 family plasmids has been elucidated based on nucleotide sequence annotations (Tauch et al. 2003). However, despite the considerable number of studies so far conducted on industrially important Corynebacterium species, surprisingly, few have focused on their plasmid partition systems. In fact, no studies that we know of characterized in detail the partitioning mechanism in Corynebacterium plasmids.
Tsuchida et al. (2010) identified the 8th member of the pCG1 family plasmids, pCGR2 and reported that its parA and parB genes are involved in maintenance of its stability. The stability of the plasmid is lowered upon deletion of the parA gene. The effect becomes more apparent when both parA and parB genes are deleted from the plasmid. In a previous study, we reported that the plasmid copy number of pCGR2 is negatively controlled by its antisense RNA, CrrI, and that similar antisense RNA control systems are likely to exist in related plasmids across Corynebacterium species (Okibe et al. 2010). Abolishing synthesis of CrrI by promoter mutagenesis engenders a 7-fold increase in copy number of a pCGR2-derived shuttle vector. This control is surprisingly strongly accentuated by ParB.
In this study, we looked closely at the pCGR2 centromere-binding protein, ParB, with particular focus on its correlation with antisense RNA-mediated plasmid copy number control. In so doing, we identified a previously unknown interaction between ParB and its noncentromere-binding homologue, ParC. By identifying modules that are associated with the function of ParB, and by manipulating them, it was possible to modify plasmid copy number at various levels. The mechanism that may be involved in the drastic increase in plasmid copy number upon simultaneous deactivation of CrrI and ParB is discussed.
Bacterial strains and plasmids used are listed in Table 1. For genetic manipulation, Escherichia coli strains were grown at 37°C in LB medium (Sambrook et al. 1989). Coryne. glutamicum was cultivated at 33°C in rich medium (A medium with 4% glucose; Inui et al. 2007). Where appropriate, E. coli medium was supplemented with 50 μg ampicillin or kanamycin ml−1 and Coryne. glutamicum medium with 50 μg kanamycin ml−1.
|Strain/plasmid||Relevant genotype and property||Source/reference|
|HST02||F' [traD36, proA+B+, lacIq, lacZΔM15]/Δ(lac-proAB), recA, endA, gyrA96, thi, e14− (mcrA−), supE44, relA, ΔdeoR, Δ(mrr-hsdRMS-mcrBC)||Takara|
|BL21(DE3)||F−, ompT, hsdSB(rB− mB−), gal(λcI 857, ind1, Sam7, nin5, lacUV5-T7gene1), dcm(DE3)||Takara|
|Rosetta2 (DE3)||F−, ompT hsdSB(rB− mB−), gal dcm(DE3) pRARE2 (CamR)||Merck|
|Corynebacterium glutamicum R||Wild type||Yukawa et al. (2007), JCM 18229|
|pCRD321||Kmr,; E. coli – Coryne. glutamicum shuttle vector based on pCGR2||Tsuchida et al. (2010)|
|pCRB62||Kmr; E. coli – Coryne. glutamicum shuttle vector based on pCGR2||GenBank HM126493|
|pCRB62mt||Kmr; pCRB62; crrI promoter mutated||Okibe et al. (2010)|
|pCRB62(Bfs)||Kmr; pCRB62; parB ORF frame shifted||Okibe et al. (2010)|
|pCRB62mt(Bfs)||Kmr; pCRB62; crrI promoter mutated, parB ORF frame shifted||Okibe et al. (2010)|
|pCRB62(Cfs)||Kmr; pCRB62; parC ORF frame shifted||This study|
|CRB62mt(Cfs)||Kmr; pCRB62; crrI promoter mutated, parC ORF frame shifted||This study|
|pCRB62(BCfs)||Kmr; pCRB62; parB and parC ORFs frame shifted||This study|
|pCRB62mt(BCfs)||Kmr; pCRB62; crrI promoter mutated, parB and parC ORFs frame shifted||This study|
|pCRB62(DR1up−)||Kmr; pCRB62; a part of DR1 removed||This study|
|pCRB62mt(DR1up−)||Kmr; pCRB62; crrI promoter mutated, a part of DR1 removed||This study|
|pCRB62(DR2−)||Kmr; pCRB62; DR2 removed||This study|
|pCRB62mt(DR2−)||Kmr; pCRB62; crrI promoter mutated, DR2 removed||This study|
|pCRB62(DR12−)||Kmr; pCRB62; a part of DR1 and DR2 removed||This study|
|pCRB62mt(DR12−)||Kmr; pCRB62; crrI promoter mutated, a part of DR1 and DR2 removed||This study|
|pET22b(+)||Apr; E. coli expression vector||Novagen|
|pET22b-parB||Apr; pET22b(+) with the parB gene cloned into the NdeI/XhoI sites (His•Tag added).||This study|
|pET22b-parC||Apr; pET22b(+) with the parC gene cloned into the NdeI/XhoI sites (His•Tag added).||This study|
|pET22b-ParC-Strep||Apr; pET22b(+) with the parC gene fused with Strep•TagII cloned into the NdeI/XhoI site (Strep•TagII added; His•Tag not added).||This study|
Restriction endonucleases were purchased from Takara (Osaka, Japan). DNA ligation was performed using Mighty Mix (Takara). Polymerase chain reaction (PCR) was performed using PrimeSTAR GXL DNA Polymerase (Takara), and the resulting PCR fragments were purified using NucleoSpin Extract II (Macherey-Nagel, Düren, Germany). Escherichia coli was transformed by the CaCl2 procedure (Sambrook et al. 1989). Coryne. glutamicum was transformed as previously described (Vertès et al. 1993). DNA sequences were determined using BigDye Terminator v3·1 Cycle Sequencing Kit with ABI 3130xl/3730xl genetic analyser (Applied Biosystems, Foster City, CA, USA). Sequence data were analysed using Genetyx software (Genetyx).
A variety of pCGR2 derivatives were constructed by inverse PCR (Table 1) using primer sets listed in Table S1. A frameshift mutation was introduced into the parC gene (by exchanging 5 nucleotides after ATG with SpeI restriction site) to construct pCRB62(Cfs), pCRB62mt(Cfs), pCRB62(BCfs) and pCRB62mt(BCfs). An upstream part of the DR1 region (Fig. 1, corresponding to Comp1 in Fig. 2a) was deleted to construct pCRB62(DR1up−) and pCRB62mt(DR1up−). The DR2 region (Fig. 1, corresponding to Comp3 in Fig. 3a) was deleted to construct pCRB62(DR2−), pCRB62mt(DR2−), pCRB62(DR12−) and pCRB62mt(DR12−). pCRB62 was used as template DNA for pCRB62(Cfs), pCRB62(DR1up−) and pCRB62(DR2−). pCRB62mt was used as template DNA for pCRB62mt(Cfs), pCRB62mt(DR1up−) and pCRB62mt(DR2−). pCRB62(Bfs), pCRB62mt(Bfs), pCRB62(DR1up−) and pCRB62mt(DR1up−) were used as template DNAs for pCRB62(BCfs), pCRB62mt(BCfs), pCRB62(DR12−) and pCRB62mt(DR12−), respectively.
Plasmid DNA was extracted by alkaline lysis method (Sambrook et al. 1989) or using a NucleoSpin Plasmid QuickPure kit (Macherey-Nagel). The DNA extraction procedures were modified by pretreating corynebacteria with 4 mg lysozyme ml−1 (Sigma-Aldrich, Steinheim, Germany) at 37°C for 1 h. Coryne. glutamicum harbouring each plasmid was grown in 100 ml rich media, and 1 ml each of late-log phase culture (OD610 of 6·0) was withdrawn for extraction of total DNA and RNA for quantitative RT-PCR (Tsuchida et al. 2009).
Plasmid copy number was calculated as a ratio of quantified aph (a single-copy gene on the plasmids) to dnaA (a single-copy gene on Coryne. glutamicum R chromosome) as described elsewhere (Tsuchida et al. 2009), with the modification that serial dilution series of plasmids containing dnaA or aph sequence were made ranging from 1 × 104 to 1 × 108 or 1 × 105 to 1 × 109 copies μl−1, respectively, to produce standard curves.
mRNA was quantified using the ABI 7500 Fast Real-Time PCR System (Applied Biosystems). Each RT-PCR reaction mixture (20 μl) contained 10 μl Power SYBR Green Master Mix (Applied Biosystems), 150 nmol l−1 forward/reverse primer, 5 U MuLV reverse transcriptase (Applied Biosystems), 8 U RNase inhibitor and 100 ng total RNA (1 ng total RNA for the reference 16S rRNA gene). Transcripts of the 16S rRNA, repA, parB and parC genes were quantified using primer sets shown in Table S1. The target gene transcripts were normalized to the reference gene transcripts (16S rRNA) from the same RNA samples. Reactions were performed as follows: 50°C for 30 min, 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 30 s. The results, referred to as CT (cycle threshold) values, were means of triplicate experiments.
The 5′-ends of the parAB operon and parC were localized using a SMART Rapid amplification of cDNA ends (RACE) cDNA amplification kit (Clontech, Mountain View, CA, USA). For determination of the parC 5′-end, total RNA extracted from Coryne. glutamicum harbouring pCRB62 was polyA-tailed prior to the RACE reaction, using the Poly(A)-Tailing Kit (Applied Biosystems), and the 1st strand synthesis performed using the primer from the kit, followed by the PCR using the gene-specific primer, ParC_5RACE (Table S1). The 5′-end of the parAB operon was determined using primers Par_1st and Par_5RACE for the 1st strand synthesis and a consequent PCR, respectively (Table S1). The resulting PCR products were TA-cloned using a pGEM-T Easy Vector System (Promega, Fitchburg, WI, USA) and sequenced.
The parB and parC genes were PCR-amplified from pCRB62 using the primers listed in Table S1 and then cloned into the expression vector pET22b(+) (Merck, Darmstadt, Germany) to yield pET22b-parB and pET22b-parC, respectively (Tables 1 and S1).
Escherichia coli BL21(DE3) transformed with pET22b-parB or pET22b-parC was grown at 37°C in LB medium to an OD600 of 0·5, and protein expression was induced by addition of 0·5 mmol l−1 IPTG. Cultures were further incubated at 37°C for 4 h before harvesting the cells. His-tagged proteins were purified by affinity chromatography on nickel nitrilotriacetic acid agarose (Qiagen, Hilden, Germany). The concentration of purified proteins was determined using a Bio-Rad protein assay (Bio-Rad Laboratories, Berkeley, CA, USA).
Cy3-labelled DNA probes used in Electrophoretic Mobility Shift Assay (EMSA) were generated by PCR using the primer sets shown in Table S1. Short unlabelled competitor DNA probes, Comp1 through 4, were produced by annealing two complementary oligos in Table S1. Unlabelled competitor DNA probe Comp0 was produced by PCR using the primers shown in Table S1.
For EMSA with either ParB or ParC protein, each binding reaction (20 μl) contained the Cy3-labelled DNA probe (10 nmol l−1), His-tagged ParB or ParC protein (amounts as specified in relevant figure legends), 1 μg of poly(dIdC), 0·2 μl of 1% BSA in binding buffer (10 mmol l−1 Tris–HCl pH 8·0, 150 mmol l−1 KCl, 0·5 mmol l−1 EDTA, 0·1% Triton-X100, 12·5% Glycerol, 0·2 mmol l−1 DTT). Unlabelled competitor DNA probes were included in the reaction at concentrations specified in relevant figure legends. The reaction mixture were incubated for 20 min at room temperature and run on a 5 or 8% polyacrylamide gel in 0·5 × TBE at 150 V. For detection of ParB–ParC interaction using EMSA, binding buffer consisted of 10 mmol l−1 Tris–HCl pH 7·5, 50 mmol l−1 KCl, 5 mmol l−1 MgCl2, 0·1% Triton-X100, 12·5% Glycerol, 0·2 mmol l−1 DTT. The reactions were incubated for 1 h at room temperature. Detection of the Cy3-labelled DNA and DNA–protein complexes were performed using the Typhoon TRIO variable-mode imager (GE Healthcare Bioscience, Pittsburgh, PA, USA).
Chemical cross-linking reaction mixtures (20 μl) contained 5 μg of ParB or ParC protein and dimethyl pimelimidate (DMP) at concentrations specified in the relevant figure legend, in a reaction buffer consisting of 50 mmol l−1 phosphate buffer (pH 8·0), 50 mmol l−1 KCl and 5 mmol l−1 MgCl2. The reaction mixtures were incubated at 33°C for 1 h and terminated by adding 1 μl of 0·5 mol l−1 Tris–HCl (pH 6·8) and SDS sample buffer, followed by heating at 95°C for 5 min. Samples were loaded onto a 15% SDS-PAGE gel, and the cross-linked proteins were visualized either using the SeePico CBB Stain Kit (Benebiosis, Seoul, Korea) or by Western blotting with the SuperSignal West HisProbe Kit (Pierce, Rockford, IL, USA). Images were scanned using the LAS-3000 system (Fujifilm, Tokyo, Japan).
The parC gene was PCR-amplified using primers listed in Table S1 and cloned into the expression vector pET22b(+) (Novagen) to yield pET22b-parC-Strep (Table 1). This construct was designed so that the parC gene is fused only with the primer-harboured Strep•TagII sequence and not with the pET22b(+)-harboured His•tag sequence. Escherichia coli Rosetta2 (DE3) transformed with pET22b-parC-Strep was grown at 37°C in 100 ml LB medium to an OD600 of 0·9, and protein expression was induced by addition of 0·7 mmol l−1 IPTG. The culture was further incubated at 37°C for 4 h before harvesting of cells. The cell pellet was resuspended with 3 ml B-PER (Thermo scientific, Waltham, MA, USA) containing 2 μl Benzonase (Qiagen). After incubating for 15 min with gentle rocking at room temperature, the suspension was centrifuged at 14 000 g for 10 min, and the supernatant was collected.
Eight micrograms of purified His-tagged ParB protein (bait) prepared in 200 μl Buffer A (50 mmol l−1 Tris–HCl, pH 8·0; 300 mmol l−1 NaCl; 0·01% Tween 20) was incubated with 10 μl Dynabeads His-Tag Isolation & Pulldown (Invitrogen, Darmstadt, Germany). After 10-min incubation at room temperature with gentle rocking, the suspension was washed thoroughly 4 times with 400 μl Buffer A. Three hundred microlitres of the cell lysate expressing the Strep-tagged ParC protein (prey) was then added to the bead/protein complex and incubated for 30 min with gentle rocking at room temperature. The suspension was washed thoroughly 4 times with 600 μl Buffer B (20 mmol l−1 Tris–HCl, pH 7·5; 70 mmol l−1 NaCl; 0·01% Tween 20). Finally, 20 μl of SDS sample buffer was added to the suspension and heated at 95°C for 10 min to elute the proteins. Two microlitres of this eluent was loaded into each lane of a 15% SDS-PAGE gel. The proteins were visualized either with Bio-Safe CBB G-250 stain (Bio-Rad) or by Western blotting using the SuperSignal West HisProbe Kit (Pierce) and Strep•TagII Antibody HRP Conjugate (Novagen), according to the manufacturer's instruction. Images were scanned using the LAS-3000 system (Fujifilm).
The parB gene of pCGR2 encodes a 75 amino acid residue protein with an estimated molecular weight of 8·2 kDa. The DNA-binding activity of this small ParB protein was assessed by EMSA using a purified His-tagged ParB protein. First, Cy3-labelled DNA probes, Probe 1 through 5, were designed to estimate the ParB binding region (Fig. 1). Probe 1 contained large, obvious 180-bp-long direct repeats (DR0), but no specific ParB binding was observed in this region (data not shown).
Probe 2 covers the region upstream of parAB (Fig. 2a). When incubated in increasing concentrations of ParB, progressively longer smeared ladders of shifted bands were visible. At 1000 nmol l−1 ParB Probe2 was totally super-shifted (Fig. 2b(1)). 5′-RACE analysis revealed the major transcriptional start site of the parAB operon to be localized at the initiation of translation codon (Fig. 2a), indicating that the parAB operon expressed leaderless mRNA transcripts. Putative –35 and –10 regions of the parAB operon are indicated in Fig. 2a. Small, incomplete 7-bp-long direct/inverted repeats were found to stretch over and beyond the parAB promoter region (Fig. 2a). When Probe 2 was shortened on the downstream end to form Probe 2-1, a decrease in the ParB binding ability was evident (Fig. 2b(2)). Further shortening to form Probe2-2 resulted in ParB not being able to bind the probe (Fig. 2b(3)), indicating that a minimum number of repeats are necessary for effective binding of ParB. To confirm this, a competition experiment was performed using unlabelled probes, Comp1, Comp2 and Comp0 (Fig. 2b(4),(5) and (6), respectively). Although Comp1 was a poor competitor of Probe2–ParB complex (Fig. 2b(4)), the fact that Comp0 (Fig. 2b(6)) was a stronger competitor than Comp2 (Fig. 2b(5)) implied a role of Comp1 in ParB binding.
Probe 3, which covered the entire parAB operon, showed no specific binding of ParB (data not shown). Immediately downstream of parAB lies a series of short 7-bp-long incomplete direct repeats (DR2). Probe 4, designed to cover this region (Fig. 1, 3a), complexed with ParB (Fig. 3b-(1)). Unlabelled probe Comp3, designed to cover seven direct repeats (Fig. 3a), competed better with the Probe4–ParB complex than Comp1 which, at higher concentrations, allowed the ParB–Probe4 complex to become slowly relaxed (Fig. 3b).
Probe5 covered the region starting immediately downstream of DR2 to the start codon of the parC gene (Figs 1 and 3a). It complexed with ParB, showing progressively longer ladders of shifted bands (Fig. 3c(1)). Probe 5 was progressively shortened on the upstream end to produce Probe5-1 through 5-4 (Fig. 3a,c(2) through 3c(5)) and on the downstream end to produce Probe5-5 and 5-6 (Fig. 3a,c(6),(7)). Probe5-1 and 5-2 showed almost identical shift patterns with Probe5, but ParB binding became significantly weak as the probe length approached that of Probe5-3 and 5-4 (Fig. 3c(4) and (5), respectively). Probe5-5 showed almost the same shift pattern as Probe5 (Fig. 3c(6)), and almost no ParB binding was detected with Probe5-6 (Fig. 3c(7)). From these results, the region covered by unlabelled probe Comp4 (Fig. 3a) was thought to serve as the major ParB binding site, where the presence of a series of 7-bp-long incomplete direct repeats was evident. The Comp4 region was indeed confirmed to harbour the ParB binding site, as it competed with the ParB–Probe5 complex (Fig. 3c(9)). When Comp1 from the region upstream of parAB was used in the competition experiment in comparison, weak bands corresponding to the free Probe5 fragment were observed at higher Comp1 concentrations (Fig. 3c(8), lanes 5–7), indicating that Comp1 could compete with Probe5 to some extent.
An ORF of heretofore unknown function, orf4, is present downstream of the parAB operon in pCGR2. orf4 encodes a 93 amino acid residue protein with an estimated molecular weight of 10·6 kDa. The sequence, christened ParC, shares 23·7% identity with ParB. 5′-RACE analysis of parC revealed a major transcriptional start site 101 bp upstream of the translation start codon (Fig. 3a). Consequently, the –35 and –10 regions of parC were elucidated (Fig. 3a). Because of similarities in size and amino acid sequence between ParB and ParC, it was important to establish whether or not ParC can also bind DNA. His-tagged ParC protein was therefore purified and its DNA-binding activity was determined by EMSA. Interestingly, ParC could bind neither Probe1 through 5 nor the region downstream of parC (data not shown). This suggests that unlike ParB, ParC may not be integral to plasmid partitioning.
ParB and ParC showed appreciable homology with helix-turn-helix proteins such as CopG from pMV158 (Gomis-Ruth et al. 1998) and type Ib ParB proteins from pB171 and pCT745 (Galli et al. 2001; Ringgaard et al. 2007a). Their amino acid sequences (Fig. 4a) feature C-terminal RHH (ribbon-helix-helix) motifs. However, ParC differs from the others in its lack of the first predicted helix.
Cross-linking experiments using DMP were performed to compare oligomeric states of ParB and ParC (Fig. 4b,c). With increasing concentrations of DMP, increasing amounts of higher-order ParB multimers (dimers to pentamers) were formed (Fig. 4b). In contrast, with ParC, only dimers (as double bands) became increasingly evident as the DMP concentration increased (Fig. 4c). DMP is a homo-bifunctional imidoester that specifically reacts with primary amine groups (i.e. ε-amino groups of lysine residues). ParB and ParC both include 8 lysine residues but at different locations. The presence of such residues in ParC may have produced differently cross-linked dimers, resulting in formation of the double bands.
In a previous study, we reported on the antisense RNA (CrrI) of pCGR2, inactivation of which resulted in an increase in the plasmid copy number (Okibe et al. 2010; see copy numbers of pCRB62 and pCRB62mt in Fig. 5a). This antisense RNA–mediated control was accentuated by ParB; the simultaneous deactivation of CrrI and ParB resulted in an even more drastic increase in copy number (Okibe et al. 2010; see pCRB62 (Bfs) and pCRB62 (Bfs)mt copy numbers in Fig. 5b). In this study, this observation was investigated in more detail. Under the control of CrrI, the absence of ParB (pCRB62(Bfs)) resulted in increased parB (47-fold) and parC (30-fold) mRNA levels, although the copy number remained almost unchanged from the original pCRB62 (Fig. 5b). When CrrI is disabled, deactivating ParB (pCRB62mt(Bfs)) resulted in markedly increased parB (120-fold) and parC (71-fold) mRNA levels, while increasing the copy number only about 12-fold (Fig. 5b). Given that ParB binds specifically to the promoter regions of the parAB operon and the parC gene, the results indicate that ParB negatively regulates its own promoter as well as that of parC.
It has so far in this study been established that ParB binds the region upstream (DR1) and downstream (DR2 and DR3) of the parAB operon. We christen the regions parS1 and parS2, respectively. The parC gene was also found to express a ParB homologue whose transcription is likely to be under negative regulation of ParB (see above section). To determine whether or not deactivating ParB-associated modules affects plasmid copy number, a variety of pCGR2 derivatives were constructed (Tables 1 and S1, Fig. 5). In addition to plasmids from our previous study harbouring parB sequences sporting single frameshift mutations (Fig. 5b), plasmids with parC sequences sporting single frameshift mutations (Fig. 5c) as well as parB and parC sequences sporting double frameshift mutations (Fig. 5d) were constructed. Moreover, to investigate the effects of deleting ParB binding sites, care was taken not to affect the promoter sequence of the parAB operon and the parC gene. This was achieved by deletion of part of the DR1 region (Comp1 sequence from Figs 2a and 5e), the entire DR2 region (Comp3 sequence from Figs 3a and 5f), or both regions (Fig. 5g).
As long as CrrI functioned normally, no marked change in both the plasmid copy number (2·3–2·7) and the rep mRNA level was observed. On the contrary, where CrrI was deactivated, plasmid copy number became highly dependent on the plasmid sequence modified. The highest copy number was achieved when ParB was modified; 2·1 × 102 of pCRB62mt(Bfs) (Fig. 5b) and 2·1 × 102 of pCRB62mt(BCfs) (Fig. 5d). Modification of ParC resulted in a comparatively smaller increase in the copy number (5·2 × 10 of pCRB62mt(Cfs); Fig. 5c). Modifying both ParB and ParC did not show any cumulative effects (Fig. 5d).
When each of the direct repeat regions was deleted, the plasmid copy number increased from 1·8 × 10 [pCRB62mt; Fig. 5a) to 2·5 × 10 (pCRB62mt(DR1up−]; Fig. 5e) and to 2·5 × 10 (pCRB62mt(DR2−); Fig. 5f). Deleting both direct repeat regions resulted in higher copy numbers than those realized from deleting each individual region [3·8 × 10 copies for pCRB62mt(DR21−); Fig. 5g]. Because deletion of the region containing the direct repeats did not affect transcription of either parB or parC (and consequently the ParB and ParC protein levels), the observed copy number increase was likely due to ParB protein not being able to bind to the corresponding regions. This deletion effect was, however, smaller compared with when parB was frameshifted [2·1 × 102 copies for pCRB62mt(Bfs)], implying that simultaneous and complete deletion of DR1, DR2 as well as DR3 regions (and possibly other unknown ParB binding sites) likely result in a further increase in the plasmid copy number.
The observation that deactivation of not only ParB but also ParC resulted in a major increase in the plasmid copy number raised the interesting question of how a centromere-nonbinding protein, ParC, is involved in such a phenomenon. In the search of a solution to the question, formation of a partition complex containing ParC and ParB was investigated. As was shown in previous experiments, ParC itself was unable to bind either parS1 or parS1 (Fig. 6a(1),b(1)). However, addition of ParC to ParB–parS1 and ParB–parS2 complexes both resulted in a further shift in mobility, although to a different extent: ParC–ParB–parS1 complexes were shown to form more readily than ParC–ParB–parS2 complexes (Fig. 6a(2),b(2)). In order to obtain direct evidence of the ParC–ParB interaction, a pull-down assay was performed (Fig. 6c). Strep-tagged ParC was pulled-down specifically from the cell lysate by binding the Strep-tagged ParC to His-tagged ParB protein. This clearly confirmed that ParC directly interacts with ParB without requiring mediation of DNA. From these results, it was concluded likely that the copy number increase by deactivation of ParC was indirectly caused by interaction of ParB and ParC.
pCGR2 contains the parAB operon that encodes a 199 amino acid residue motor protein with Walker-type ATPase motif (ParA) and a 75 amino acid residue centromere-binding protein (ParB). The parAB operon expresses leaderless mRNA as was also reported for transcription of the odx gene in Coryne. glutamicum (Klaffl and Eikmanns 2010), as well as of different genes in other bacterial species (Brock et al. 2008). Based on a previous implication of involvement of the par locus in pCGR2 plasmid stability (Tsuchida et al. 2010), the results here confirm that pCGR2 employs the type Ib par system for stable segregation. Nevertheless, the majority of the type Ib centromere-binding proteins characterized so far are of gram-negative bacterial origin (Yin et al. 2006). Tsuchida et al. (2010) originally identified orf4 downstream of the parAB operon and reported that it had no effect on plasmid stability. Interestingly, orf4 encodes a 93 amino acid residue protein (10·6 kDa), named ParC in this study, that shares appreciable identity with ParB. Nonetheless, ParC does not share the fundamental function of ParB to bind to centromere sites, and it is not essential for the partitioning process.
Although there is significant lack of sequence homology among centromere-binding proteins of different types (type Ia, Ib and II), they share common DNA-binding domains: Type Ia family contains HTH (helix-turn-helix) folds, while type Ib and type II contain RHH (ribbon-helix-helix) motifs (Schumacher 2008). Ringgaard et al. (2007b) reported that the C-terminal RHH motif of pB171 ParB is responsible for DNA binding as well as dimerization. In this study, the presence of RHH motif is predicted in ParB, but not in ParC. Furthermore, ParC self-associates to form dimers but not higher-order multimers like ParB does. These differences may be caused by the lack of the RHH motif in ParC, but further study is necessary to clarify this.
ParB binds sequence specifically to the region surrounding the parAB operon, where short, 7-8-bp repeat elements are found. The upstream centromere site parS1 overlaps the parB promoter. A marked increase in the parB transcription level was observed subsequent to the introduction of a frameshift mutation into the parB gene, indicating that ParB serves as an autorepressor of the parAB operon. The centromere sites of type Ib centromere-binding proteins are distinct in sequence and repeat length, although they are similarly located upstream of the par operon where autoregulation of the operon and formation of the partition complex take place (Fothergill et al. 2005; Schumacher 2008). Binding of ParB to the upstream parS1 site indeed functions in autoregulation of the parAB operon. Interestingly, ParB also binds to two other unrelated sets of direct repeats in the downstream parS2 site. A set of direct repeats cover the parC promoter region and ParB appears responsible for the parC transcription repression. In a few unique cases (e.g. E. coli pB171 and Salmonella enterica R27 plasmids), a single plasmid may contain double par loci consisting of a type I and a type II par system (Lawley and Taylor 2003; Ringgaard et al. 2007a). In E. coli pB171 centromere-binding proteins from the respective par loci bind cooperatively to form the partition and promoter repression complexes (regulatory cross-talk; Ringgaard et al. 2007a). However, there are no known centromere-binding proteins regulating other independent parB-like genes. At the DNA level, no sequence similarity is found between the parB and parC genes. Therefore, the presence of the parC gene is unlikely to be a result of a gene duplication event during evolution.
Simultaneous deactivation of CrrI and ParC brought about an increase in plasmid copy number. However, unlike what was observed with ParB, parB and parC mRNA levels remained almost unchanged from those of pCRB62mt, implying that ParC does not play the same repressor role that ParB does. To explain the inability of ParC to bind DNA, suspected interaction between ParB and ParC was investigated. The partition complex of pCGR2 was thence shown to involve ParC interacting with the ParB–DNA complex. ParC is also able to interact directly with ParB without the mediation of the centromere site. With this in mind, it is plausible that ParC interaction with ParB molecules bound cooperatively to the centromere sites forms the ParC–ParB–parS partition complex, affecting the level of repression by ParB (Fig. 7a,b). This effect may be more apparent with parS1 than parS2, because ParC–ParB–parS1 complexes were formed more readily than ParC–ParB–parS2 complexes. To our knowledge, this is the first report describing the existence and function of ParC protein in the plasmid portioning system. In pTET3, which is another one of the larger Corynebacterium pCG1 family plasmids, a small orf7 is located downstream of parAB operon (Tauch et al. 2002). Amino acid sequence of Orf7 is 92% identical to that of ParC, suggesting that the same mechanism may be involved in plasmid partitioning of pTET3.
By modifying the ParB, ParC and the centromere sites of the pCGR2 par system, it was possible to manipulate plasmid copy number. Following the higher-order assembly of centromere-binding proteins on the centromere site, centromere pairing between partition complexes takes place. Paired plasmids are then recognized by the motor protein and separated (Hayes and Barilla 2006b; Schumacher 2008). Centromere pairing is an important intermediate step for plasmid partitioning in the common type I loci (Ringgaard et al. 2007b): The fact that ligation of parC–DNA molecules is facilitated by ParB implies that the N-terminus of pB171 ParB is required for centromere pairing. Thus, ParB plays a role to ‘glue’ plasmids together by assembling on the centromere sites that serve as the ‘glue margins’. Formation of the partition complexes may therefore physically interrupt plasmid replication. When functions of antisense RNA and ParB are defeated altogether, uncontrolled numbers of plasmids may be allowed to co-exist within the cell. Deletion of the ParB binding site(s) would decrease the ‘glue margin’ for ParB. This may allow the plasmid to more freely replicate especially in the absence of antisense RNA. Deactivating ParB would completely eliminate the ‘glue’, allowing free plasmid replication in the absence of antisense RNA, whereas deactivation of ParC may enable only partial dissociation of the complex, resulting in only a mild increase in the plasmid copy number in the absence of antisense RNA.
This study identified that ParC is involved in formation of the partition complexes in pCGR2. By modifying the components such as ParB, ParC and the centromere sites, it was possible to manipulate plasmid copy number to varying degrees. Manipulation of antisense RNA in combination with the par locus of plasmids can serve as a novel approach to artificially elevate plasmid copy number.
We would like to thank Dr Crispinus A. Omumasaba for the critical reading of the manuscript. This work was supported by a grant from the New Energy and Industrial Technology Development Organization (NEDO), Japan.