Accurate DNA partition at cell division is vital to all living organisms. In bacteria, this process can involve partition loci, which are found on both chromosomes and plasmids. The initial step in Escherichia coli plasmid R1 partition involves the formation of a partition complex between the DNA‐binding protein ParR and its cognate centromere site parC on the DNA. The partition complex is recognized by a second partition protein, the actin‐like ATPase ParM, which forms filaments required for the active bidirectional movement of DNA replicates. Here, we present the 2.8 Å crystal structure of ParR from E. coli plasmid pB171. ParR forms a tight dimer resembling a large family of dimeric ribbon–helix–helix (RHH)2 site‐specific DNA‐binding proteins. Crystallographic and electron microscopic data further indicate that ParR dimers assemble into a helix structure with DNA‐binding sites facing outward. Genetic and biochemical experiments support a structural arrangement in which the centromere‐like parC DNA is wrapped around a ParR protein scaffold. This structure holds implications for how ParM polymerization drives active DNA transport during plasmid partition.
Faithful inheritance of genetic information requires accurate partition of replicated DNA molecules into daughter cell compartments before cell division. Akin to mitosis in eukaryotes the process of DNA partition in prokaryotes is a highly dynamic process in which the DNA replicates are separated and transported to opposite sides of the cell division plane (Gerdes et al, 2004). To a large extent, our knowledge about DNA partition in bacteria is derived from studies of how low‐copy‐number plasmids are segregated. Such plasmids rely exclusively on partition loci (par) for their specific distribution to progeny in every cell cycle. In addition, partition loci have been identified on a large number of bacterial chromosomes and appear to play an important role in the spatial organization of chromosomal DNA as well (Gerdes et al, 2000; Yamaichi and Niki, 2000; Hayes, 2000; Wu and Errington, 2003; Errington et al, 2005; Fogel and Waldor, 2006; Lee and Grossman, 2006; Yamaichi et al, 2007). Partition systems are generally composed of three essential elements, all specified by the same genetic locus: two proteins, an ATPase and a DNA‐binding adaptor protein, that act together on a cis‐acting centromere‐like region on the DNA. When classified according to the nature of their ATPase component, partition systems fall into two main categories: type I systems that employ Walker‐type ATPases termed ParA, ParF or SopA, and type II systems whose ATPases, termed ParM, are homologs of actin. Type I partition systems further fall into two subgroups type Ia and type Ib based on the size of the Walker ATPases. The adaptor proteins of type Ia, type Ib and type II partition systems (termed ParB, ParG and ParR, respectively) show no homology even though they have similar functions in the partition process, and the centromere‐like sites they recognize are diverse (Gerdes et al, 2000; Hayes, 2000).
The earliest step in a general mechanistic outline of plasmid partition involves the formation of a partition complex between the adaptor protein and its cognate centromere‐like site on the DNA. In parMRC from plasmid R1, which is typical of partitioning systems with actin‐like ATPases (type II), the dimeric ParR adaptor protein, which is present in less than 1000 copies in the cell (Møller‐Jensen et al, 2002), binds cooperatively to two sets of five 11‐bp direct repeats denoted parC (Dam and Gerdes, 1994; Møller‐Jensen et al, 2003). The parC DNA region has been shown to be intrinsically curved (Hoischen et al, 2004). Apart from its function in plasmid partition, ParR binding serves to autoregulate the expression of partition proteins as the parC site overlaps with the par‐promoter (Jensen et al, 1994). A minimum of two direct repeats of DNA is required for binding of ParR and footprinting analysis have shown that the partition complex extends over the entire parC region (Møller‐Jensen et al, 2003). Formation of the ParR/parC DNA partition complex has been shown by electron microscopy to mediate pairing of plasmid replicates (Jensen et al, 1998).
In the parABS system of plasmid P1, which employs a Walker type ATPase (type Ia), the partition complex is formed by association of a ParB dimer and the host‐encoded Integration Host Factor with specific recognition sequences in the centromeric DNA region, parS (Bouet et al, 2000). Formation of the partition complex is thought to mediate specific pairing of plasmid replicates at parS sites (Edgar et al, 2001). A recent crystal structure of ParB bound to DNA provided mechanistic insight into how ParB‐mediated plasmid pairing might occur by demonstrating that ParB dimers bound to DNA via their N‐terminal helix–turn–helix (HTH) domain could associate in trans with a different DNA molecule via their dimerization domain and thus serve as a molecular linkage between two plasmid centromers (Schumacher and Funnell, 2005).
The partition complexes are recognized by partition ATPases that convert energy from ATP hydrolysis into active intracellular transport of plasmid molecules. Both ParA and ParM type ATPases have been shown to form filamentous structures, although ATP‐dependent dynamics remain to be demonstrated for ParA proteins (Møller‐Jensen et al, 2002, 2003; Barilla et al, 2005; Leonard et al, 2005; Lim et al, 2005; Becker et al, 2006; Ebersbach et al, 2006; Bouet et al, 2007). In vitro ParM filaments nucleate spontaneously in the presence of ATP and extend bidirectionally. In the absence of ParR and parC DNA, the ParM filament growth phase is succeeded by rapid unidirectional collapse, a phenomenon reminiscent of the dynamic instability displayed by microtubules in eukaryotes (Garner et al, 2004). In a recent study, active DNA segregation was reconstituted from purified ParM, ParR and parC DNA components, indicating that these essential components are sufficient to form a bipolar, plasmid‐segregating spindle (Garner et al, 2007). ParR/parC complexes interacted with both ends of the ParM filaments in the reconstituted plasmid segregation assay thereby stabilizing them against depolymerization, while allowing for insertion of new ParM monomers at the filament tip (Garner et al, 2007). This finding is consistent with previously proposed insertional polymerization models for plasmid movement (Møller‐Jensen et al, 2003; Garner et al, 2004).
An important question concerns the overall architecture of partition complexes and how they interact with the force‐generating ATPases during the DNA partition process. In this work, we present the crystal structure of the adaptor protein ParR from plasmid pB171. This protein binds to the parC1 centromere‐like region of the double partition locus of Escherichia coli pB171 (Ebersbach and Gerdes, 2001) and the binding site is characterized by two identical 10‐bp high‐affinity sequences separated by 31 bp (Ringgaard et al, 2007). ParR forms a tight dimer, which further assembles into a ring or helix structure with putative DNA‐binding regions facing outward. The position of the DNA‐binding domain is confirmed by introduction of site‐specific amino‐acid substitutions on ParR, followed by functional analysis in vivo and in vitro. Electron microscopic examination of ParR/parC DNA complexes further reveal the formation of ParR rings, suggesting that the crystalline packing of ParR protein into a helix is physiologically relevant. This partition complex architecture leads us to propose a model for active plasmid movement in which ParR protein encircle or bind to growing ParM filament tips either as a helix or a ring through interactions involving the ParR C‐terminus, while parC DNA is wrapped on the outside through interactions with the ParR N‐terminus.
ParR forms ring‐like structures on parC DNA
In a previous study, plasmid R1 ParR was shown by electron microscopy (EM) to bind to plasmid parC regions and mediate site‐specific pairing of DNA molecules (Jensen et al, 1998). Here, we further examine the ParR/parC DNA partition complex structure using improved staining techniques for EM imaging. Specifically, we have omitted the use of aldehyde fixatives before negative staining. We now demonstrate that plasmid R1 ParR forms ring‐shaped complexes on pre‐linearized DNA containing parC regions. Images of ParR/parC DNA complexes shown in Figure 1 were obtained by rotary shadowing (panels B–J) and negative staining (panels K–R). Figure 1O–R shows enlarged images of negatively stained ParR rings on DNA. Ring structures always formed at the parC site as judged from the conserved distance to the ScaI restriction site and they never formed on control DNA without the parC region or in the absence of DNA (data not shown). Also, no ring structures were observed on parC DNA in the absence of ParR protein (Figure 1A). The ParR rings measured between 15 and 20 nm in diameter, which is consistent with the 15 nm ParR helix diameter shown in Figure 2A plus the DNA wrapped around it. Figure 1I and J shows examples of DNA pairing by close association of ParR rings.
ParR belongs to the MetJ/Arc superfamily of DNA binding proteins
Attempts to determine the structure of plasmid R1 ParR by X‐ray crystallography proved unsuccessful as the protein formed needle‐shaped crystals of insufficient size and quality for structure determination. Instead, we have determined the structure of the homologous ParR from E. coli plasmid pB171. This protein has 25% sequence identity and 44% similarity over all residues to plasmid R1 ParR (Figure 3). The structure contains two monomers in the asymmetric unit, each consisting of a short N‐terminal β‐strand, followed by four or five α‐helices (Figure 2A). The two monomers of the asymmetric unit are related by a pseudo‐two‐fold axis and form a tight antiparallel homodimer in which one monomer is distinguishable by a short α‐helix H5 near the C‐terminus. The polypeptide chains were traceable in the electron density from Lys6 to Leu95, leaving the C‐terminal 35 amino acids disordered. The dimer N‐termini form a ribbon–helix–helix (RHH)2 structure in which a short two‐stranded antiparallel β‐ribbon (S1) is flanked by two sets of α‐helices (H1 and H2). The compact RHH2 motif, which is held together by a core of hydrophobic interactions, is followed by short α‐helices (H3–H5). The side and top view of a ParR dimer structure is shown in Figure 2A. The family of bacterial RHH2 proteins is typified by the MetJ and Arc transcriptional repressors, both of which have been crystallized in complex with their respective operator DNA (Somers and Phillips, 1992; Raumann et al, 1994b). These proteins contain a two‐stranded antiparallel β‐ribbon structure that interacts with base‐determinants in the DNA major groove. This interaction is stabilized by interactions between the DNA backbone phosphates and α‐helices H1 and H2 (Raumann et al, 1994a). Other members of this protein family include CopG, a repressor involved in plasmid copy‐number‐control (Gomis‐Ruth et al, 1998), and the plasmid‐encoded partition adaptor proteins ParG (Golovanov et al, 2003) and Omega (Weihofen et al, 2006), the structures of which are shown in Figure 2C.
ParR dimers multimerize to form a helical DNA‐binding scaffold
A unifying feature of RHH2 proteins is the cooperative nature of their DNA binding. In all cases, the minimum binding entity is comprised by a dimer of dimers and the complexes are stabilized by cooperative interdimer contacts. In our ParR crystal lattice, the proteins assemble into a continuous helix structure consisting of 12 ParR dimers per full 360° turn. ParR dimers are arranged with their N‐termini facing outward and their C‐termini pointing towards the helix center. As shown in Figure 2B, a helical turn along the screw axis results in a translation of 13 nm. The view along the screw axis demonstrates that the ParR helix has a diameter of 15 nm and is made up from six symmetrical pairs of dimers such that every ParR dimer is related to its nearest neighbor by a 30° rotation and a two‐fold symmetry axis. The dimers are held together mostly by interactions between amino‐acid side chains in H3, H4 and H5. The N‐terminal RHH2 domains align on the helix exterior and form regularly spaced basic patches as can be seen in the surface potential representation in Figure 2B. The fact that the RHH2 DNA‐binding sites are positioned 3.5 nm apart, which corresponds almost exactly to one helical turn of B‐helix DNA and is in register with the repeat structure of the parC1 centromere‐like site, strongly advocates a physiological relevance of the crystalline packing of ParR. Thus, the crystal structure suggests a partition complex architecture in which the centromere‐like parC1 region wraps around a helical scaffold formed by ParR protein.
The ParR fold is conserved
A multiple sequence alignment of chromosomal and plasmid‐encoded ParR homologs from E. coli and other closely related gamma proteobacteria is shown in Figure 3. The aligned sequences display a large degree of sequence conservation in the N‐terminal RHH2 domain, as well as in helix H3 and H4, which are involved in interdimer contacts. This alignment includes ParR from R1, which is the prototypical type II partition system and is used for our EM studies. RHH2 secondary structure elements are succeeded by conserved proline residues that induce sharp bends in the polypeptide chain responsible for the wedge shape of the molecule. The aligned sequences differ in the H5 region, where certain proteins contain insertions. Interestingly, the C‐terminus, which is disordered in our ParR crystal structure, shows high sequence conservation, indicating that this part of the protein may be important for the partition function.
The RHH2 motif is required for DNA binding and plasmid partition
To verify that the DNA‐binding interface of ParR of pB171 is in fact the N‐terminal RHH2 domain, we carried out site‐specific mutagenesis of residues Lys6, Arg7 and Lys8, which are presumed to be involved in contact with the DNA. As negative control, we performed amino‐acid substitutions closer to the C‐terminus at Lys85 and Arg92. Lysine residues were mutated into glutamic acid and arginines were substituted by Serine. The position of the mutated amino‐acid residues is shown in Figure 2A in white space fill representation. Mutant parR genes were tested for functionality in their natural genetic context on test plasmids derived from low‐copy‐number plasmid pGE103, which contains the wild‐type (wt) partition locus from pB171 (Ebersbach and Gerdes, 2001). For the plasmid stabilization assay, cells were grown in nonselective medium and the fraction of plasmid‐containing cells was determined by plating on nonselective indicator plates containing X‐gal. As shown in Figure 4A, test plasmids containing the ParR171 K6E, R7S and K8E mutations severely impaired the function of ParR171. By contrast, ParR171 K85E and R92S mutations seemed to have no effect on par activity as the plasmids were stably maintained throughout the assay. To correlate the observed deficiency of N‐terminal mutants in ParR171 with lack of DNA binding, we purified mutant versions of ParR and tested these for binding to parC1 DNA by gelshift analysis and fluorescence anisotropy measurements (Figure 4B and C). As reported previously, wt ParR from pB171 associates specifically with parC1 DNA (Figure 4B, lanes 1–4) (Ringgaard et al, 2007). No retardation was observed with a nonspecific control DNA fragment. The introduction of K6E, R7S and K8E mutations abolished the ability of ParR to bind to parC1 DNA, whereas ParR with K85E and R92S mutations could still bind to the DNA (Figure 4, lanes 5–9). These findings are supported by fluorescence anisotropy measurements showing impaired DNA binding of the K6E, R7S and K8E mutant proteins and retained binding capacity of the K85E and R92S mutant proteins. The fact that K85E‐mutated ParR did only display partial retardation in the gelshift could be explained by a disruptive effect of this amino‐acid substitution on dimer–dimer interaction such that spreading beyond high‐affinity sites is limited (white arrow head). This assumption is corroborated by the fluorescence anisotropy data showing reduced binding cooperativity of this mutant protein. In conclusion, these results demonstrate that the N‐terminal RHH2 domain is required for DNA binding and that disruption of the interactions to parC1 DNA leads to impaired plasmid partition.
RHH2 structures: variations on the same theme
Members of the MetJ/Arc protein superfamily include the plasmid‐encoded repressors Omega (Weihofen et al, 2006), CopG (Gomis‐Ruth et al, 1998), ParG (Golovanov et al, 2003) and ParR. All share the same DNA‐binding RHH2 motif and their mode of DNA recognition is similar. The binding specificity is determined by interactions between side chains of the antiparallel β‐ribbon and bases in the DNA major groove. This interaction is further stabilized by interactions formed between the DNA phosphate backbone and side chains of helix H1 and main chain amines of helix H2, respectively. In most cases, the RHH2 domain is located at or near the N‐terminus, but it can also be located C‐terminally, as in the case of ParG shown in Figure 2C (Golovanov et al, 2003). Our crystal structure of ParR from pB171 showed the presence of an N‐terminal RHH2 domain. Although attempts to co‐crystallize ParR with parC1 DNA proved unsuccessful, we confirmed the involvement of this domain in DNA recognition and plasmid partition by genetic and biochemical data shown in Figure 4.
Although RHH2 proteins generally exist as dimers in solution, they bind to DNA in a cooperative fashion and form tetramers or higher order oligomers through quaternary dimer–dimer (protein–protein) interactions. These quaternary relationships are quite diverse and reflect the different compositions of their binding sites on the DNA. Thus, protein–DNA complexes formed by this family display a large degree of structural diversity as illustrated by the structures of Omega (Weihofen et al, 2006) and Arc (Raumann et al, 1994b) in complex with their respective DNA operators in Figure 2C. Omega dimers bound to adjacent heptad DNA sequences interact through hydrophobic side chains in helix H1, forming a left‐handed matrix around straight DNA in which each dimer is related to its neighbor by a 7‐bp translation and a 252° rotation (Weihofen et al, 2006). By contrast, two Arc dimers bound to a 21‐bp operator site interact using the loop between helices H1 and H2 to form a tetramer that binds on one face of the double helix, thereby inducing a 50° bend in the DNA (Raumann et al, 1994b). Thus, binding site length and spacing between adjacent binding sites are important parameters in determining how RHH2 domains interact on the DNA. Although the RHH2 domain structure is essentially conserved in proteins of this family, they use it in different ways and for different purposes.
ParR is a bifunctional protein, which acts both as a transcriptional repressor and a partition protein. It blocks transcription from the par promoter by binding cooperatively to the iterated recognition sequences in parC of plasmid R1 (Dam and Gerdes, 1994; Ringgaard et al, 2007) and by arranging the repressor/operator complex further into a specialized hyperstructure that is capable of interacting with polymerizing ParM, ParR functions in plasmid partition as well. In ParR, hydrophobic residues in helices H3, H4 and H5 form extensive dimer–dimer interactions, which shape the scaffold. According to the crystal structure, pB171 ParR will bind on one face of the parC1 DNA and thereby induce a 30° bend in the DNA per dimer. This is in accordance with the 31 bp (or three helical turns of double helix) spacing of the two high‐affinity binding sites in the parC1 centromere‐like region (Ringgaard et al, 2007).
ParR packed as a continuous helix in the crystals, but appeared as a ring structure in the EM projections. The screw axis in the crystal may be caused by packing restraints on the one hand, but on the other hand, a ParR helix structure could also be compressed into a closed ring upon application to EM grids. Thus, the EM pictures are projections possibly hiding the helical nature of the ParR arcs. Yet another possibility could be that ParR from plasmids pB171 and R1 form different superstructures. We find this unlikely, however, given the close homology between the two proteins. Despite considerable effort, we were unable to produce negatively stained complexes between ParR171 and DNA for electron microscopic visualization in the absence of aldehyde fixatives, thus making it impossible to discern whether this protein formed ring‐shaped complexes (data not shown). Based on the present data, it is not possible to distinguish between ring and helix structures and both would function in the proposed mechanism of plasmid partition. The diameter of a closed ParR ring structure would perhaps be too small to encircle the ParM filament. A comparison of the ParM filament cross‐section and the ParR scaffold diameter is shown in Figure 5B. Another point arguing in favor of a helical ParR scaffold is the fact that wrapping of DNA around a disc‐shaped scaffold would result in a steric clash between DNA entering and exiting the complex.
Implications for the plasmid partition mechanism
Previous studies of the partition system from plasmid R1 have provided detailed information about the mechanism of active plasmid distribution in the cell. Pairing of replicated plasmids by formation of the ParR/parC complex leads to the formation of dynamic ParM filaments that push the plasmid copies apart at the expense of ATP hydrolysis (Møller‐Jensen et al, 2002, 2003). The ParM filament structure is similar to F‐actin and employs essentially the same mechanism of ATP‐driven polymerization as actin (van den Ent et al, 2002). In the absence of ParR and parC DNA, ParM filaments were found to be extremely dynamic and transient, switching between stages of bidirectional growth and shortening (Garner et al, 2004). This dynamic instability of ParM filaments led to the proposal that bidirectional plasmid transport could involve capping of both filament ends by ParR/parC complexes, which stabilize the growing ParM filament (Garner et al, 2004). Bipolar stabilization of ParM filaments by ParR/parC complexes has been shown directly in a reconstituted DNA motility assay, which demonstrated the rapid decay of ParM polymers with uncapped tips (Garner et al, 2007). These findings are consistent with immunofluorescence‐imaging of E. coli cells demonstrating that plasmids are indeed localized at the filament ends during partition (Møller‐Jensen et al, 2003).
The structure of the ParR/parC partition complex presented here immediately suggests how the ParR protein scaffold can serve as a link between plasmid DNA and the dynamic ParM filament during active transport in the cell. We propose that ParR interacts with ParM through its C‐terminus located in the helix or ring interior and thus encircle a growing ParM filament, while contacting the plasmid DNA with the helix/ring exterior. As the ring is quite small, it might also be possible that the ring or arc of ParR binds via the C‐termini to the side of the filaments. Figure 5A shows a model of ParM‐mediated plasmid partition that takes the new ParR structure into account. In Figure 5B, the ParM filament and ParR helix cross‐sections are shown for comparison. With a diameter of about 6 nm, the ParM filament would fit inside the inner ring formed by ParR alpha‐helix 3. This arrangement would require the flexible C‐terminal part to give way. Alternatively, the helix may open up to accommodate the ParM filament as shown in Figure 5A or it might just bind to the side of the filament. Due to the ParR dimer symmetry, the partition complex is identical in both orientations on the filament and hence will be able to interact with both ends of the polar ParM filament, where ParM monomers exist in the ATP‐bound form. If the binding affinity of ParR for ParM‐ATP exceeds that for ParM‐ADP, this would explain how continuous sliding of partition complexes toward growing filament ends with ATP‐bound ParM could take place. An important part of this model is that the energy of ParM ATP hydrolysis would be used for displacing the C‐termini from the ADP‐containing filament, so that they can re‐bind the ATP caps, effectively driving ParR binding along the filament like a motor. Having many ParM‐binding sites on the ParR superstructure makes this movement more processive, so ParR (and hence the plasmids) stick to the filament ends at all times.
Materials and methods
Stains and plasmids
E. coli strain DH5α (Invitrogen) was used for cloning and plasmid stability assays. E. coli strain BL21AI (Invitrogen) was used for arabinose‐induced protein overexpression from T7 promoters. Genes encoding R1 ParR (GI:134956) and pB171 ParR (GI:6009443) were cloned in vector pHis17 (without any additional residues) (van den Ent and Löwe, 2000) resulting in plasmids pJMJ100 and pJMJ101, respectively. Expression plasmids pJMJ106, pJMJ107, pJMJ108, pJMJ185, pJMJ192 were derived from pJMJ101 by PCR mutagenesis. R1 test plasmids pJMJ206, pJMJ207, pJMJ208, pJMJ285, pJMJ292 were derived from pGE103, which contains a functional par1 locus from E. coli pB171 (Ebersbach and Gerdes, 2001) by PCR mutagenesis.
Protein expression and purification
R1 ParR and pB171 ParR proteins were overexpressed in E. coli BL21‐AI cells (Invitrogen). Cells were grown at 37°C in 2 × TY medium supplemented with 100 μg/ml ampicillin and 0.2% glucose. At mid‐exponential growth, protein expression was induced by addition of 0.4% arabinose. After 6 h of induction, the cells were harvested, resuspended in buffer A (50 mM Tris–HCl, pH 7; 100 mM KCl; 1 mM EDTA; 1 mM DTT) containing 5 μg/ml DNaseI and 1 mg/ml lysozyme, and lysed by sonication. The lysate was cleared by centrifugation for 45 min at 100 000 g. The cleared lysate was then loaded on a 5 ml HiTrap (GE Healthcare) heparin column and protein was eluted with a linear gradient of buffer B (buffer A+1 M NaCl). Fractions containing partially pure protein were pooled, diluted 100‐fold in Buffer A and loaded onto a 5 ml HiTrap SP HP cation exchange column. Again, protein was eluted with a linear gradient of buffer B. Purified protein was gel filtrated in buffer C (20 mM Tris–HCl pH 9; 50 mM KCl; 1 mM EDTA; 1 mM NaN3), concentrated to 10 mg/ml and snap‐frozen in liquid nitrogen. Purified R1 ParR and pB171 ParR tend to precipitate reversibly at pH below 8.5. The correct identity of the purified protein was verified by electrospray ionization mass spectrometry. The typical yields exceeded 10 mg protein per liter culture.
For purification of mutant pB171 ParR proteins, the pB171 parR reading frame was cloned into pOP‐TM (gift from Olga Perisic) to create an in‐frame MBP‐ParR gene fusion containing a TEV cleavage site between the two protein moieties. Site‐specific amino‐acid substitutions were introduced by PCR. TEV cleavage leaves a glycine‐serine‐histidine residual at the pB171 ParR amino terminus. MBP‐ParR was overexpressed using BL21‐AI as described above. The cells were lysed in buffer D (50 mM Tris–HCl pH 7.5; 500 mM NaCl; 1 mM EDTA; 1 mM DTT) containing 5 μg/ml DNaseI and 1 mg/ml lysozyme and lysed by sonication. Cleared lysates were loaded onto amylose resin (New England Biolabs), washed in buffer D and eluted by addition of 15 mM maltose in buffer D. Purified fusion protein was dialyzed into buffer E (50 mM Tris–HCl, pH 8.8; 25 mM KCl; 1 mM EDTA; 1 mM DTT) and incubated with TEV protease for 4 h at room temperature. The cleavage products were separated on a HiTrap Q HP ion exchange column using a linear gradient from 25 mM to 1 M NaCl in buffer E. All mutant proteins behaved similarly during purification and displayed similar column elution profiles, indicating that overall protein folding was not affected by the amino‐acid substitutions. Fractions containing pure mutant pB171 ParR protein were pooled dialyzed into buffer E, concentrated and frozen.
Crystallization, data collection and structure determination
Initial crystallization conditions were found using our in‐house 100 nl high‐throughput crystallization screen of 1500 standard conditions (Stock et al, 2005). Crystals were subsequently grown by sitting drop vapor diffusion, adding 1 μl of 1.4 M sodium acetate, 0.1 M sodium cacodylate, pH 6.5–1 μl of protein solution at 10 mg/ml. The crystals were cryoprotected with 25% ethylene glycol. Heavy‐atom derivatives for isomorphous replacement were prepared by soaking crystals overnight in pre‐equilibrated drops containing K2RuCl6, K2PtCl6, K2Au(CN)2 and K2OsCl6 (Table I). Diffraction data for native (NATI) and derivatized crystals were collected on beamline ID14eh4 (ESRF, Grenoble, France). Crystallographic data are presented in Table I. The structure of pB171 ParR was solved by isomorphous replacement. Initial sites were found using SHELXDE (Uson and Sheldrick, 1999) and refined with SHARP (de La Fortelle and Bricogne, 1997) and marginal sites were added. The final figure of merit using four derivates was 0.46 up to 3.1 Å resolution. After density modification, the density was of very good quality and could be built in one session. Refinement was performed using CNS 1.1 (Brunger et al, 1998). Refinement statistics are presented in Table II. The coordinates have been deposited in the Protein Data Bank with accession code 2jd3.
Negatively stained samples were prepared on glow‐discharged carbon‐coated grids according to a modified deep stain protocol (Stoops et al, 1992). Briefly, a 383 bp PCR product encoding the parC region was constructed using the primers: SR14: GCGAAA GGGGGATGTGCTGC; SR15: CCCAGGCTTTACACTTTATGC with pMD330 (Dam and Gerdes, 1994) as template and the PCR product was purified using a Qiaquick PCR‐purification kit. Standard reactions for negative stain were performed in a total volume of 15 μl 20 mM Tris pH 7.5, 50 mM KCl, 2 mM MgCl2. parC DNA was added to a final concentration of 10 ng/μl and R1 ParR was added to a concentration of 1 μM. The samples were incubated 15 min at room temperature and diluted fivefold in buffer. Samples (3 μl) were loaded on glow‐discharged carbon‐coated grids. After 30 s, grids were rinsed with 2% uranyl acetate, blotted dry with filter paper, and further dried with a hairdryer.
Rotary‐shadowed samples were prepared on glow‐discharged carbon‐coated grids essentially as described by Williams (1977). pMD330 (Dam and Gerdes, 1994) was linearized by ScaI (NEB) cleavage for 5 h at 37°C and purified with a Qiaquick PCR‐purification kit. Standard reactions for rotary shadowing were performed in a total volume of 15 μl 20 mM Tris pH 7.5, 50 mM KCl, 2 mM MgCl2. Linearized pMD330 was added to a final concentration of 1 ng/μl and R1 ParR was added to a final concentration of 500 nM. Reactions were incubated 15 min at room temperature and a 3 μl sample was loaded on a glow‐discharged carbon‐coated grid. After 30 s, grids were rinsed with 0.05% uranyl acetate, blotted with filter paper and dried in a stream of air. Rotary shadowing was performed in an Edwards E306A coating system by using a platinum source with an oblique angle of 6° and a sample‐to‐source distance of 8 cm.
Electron microscopy was performed at 80 kV using a Philips EM208 transmission electron microscope. Images were photographed at a magnification of × 32–40 k and negatives were scanned at 6 μm/pixel using a MRC‐KZA scanner.
Determination of plasmid stability
Plasmid loss curves were determined as described by Gerdes et al (1985). E. coli DH5α cells were grown in nonselective medium and the fraction of plasmid‐bearing cells was determined by plating on nonselective indicator plates containing X‐gal. As the β‐galactosidase gene is located on the plasmid in these strains, the plasmid content can be determined by scoring blue and white colonies.
Electrophoretic mobility shift assay
Gelshift analysis was carried out essentially as described by Ringgaard et al (2007). Briefly, 354 bp parC DNA fragments obtained by PCR using primers 5′‐GTATACGTTCATCTATAGCCC, 5′‐GATCTCCGTTTAACAGGCAG and pGE3 (Ebersbach and Gerdes, 2001) as template. Control fragments of 318 bp were generated from pUC19 DNA using primers 5′‐CGACAGGTTTCCCGACTGG, and 5′‐CAGCTGGCGAAAGGGGGATG. DNA fragments were end‐labeled with 32P and purified from a 1% agarose gel. The binding reaction contained 2 nM 32P‐labeled DNA in 10 mM Tris–HCl pH 7.5, 50 mM KCl, 50 mM NaCl, 1 mM MgCl2, 0.5 mM DTT, 0.5 mM EDTA, 0.1 mg/ml BSA and 0.1 mg/ml sonicated salmon sperm competitor DNA. Two micromolar wt or mutant ParR171 was added and the reactions were incubated for 30 min at 25°C. Following addition of glycerol to a concentration of 3%, the samples were loaded on a 5% poly‐acrylamide gel.
Fluorescence anisotropy was recorded at 298 K using a Perkin‐Elmer LS55 luminescence spectrometer. An 88‐bp fluorescein‐labeled PCR product containing the parC1 region from plasmid pB171 was generated using primers JMJ32 (5′‐GATAGTGCTCAAATTGAGTATT ACC‐3′) and JMJ33 (5′‐GTATACGTTCATCTATAGCCCC‐3′) and used at a concentration of 2 nM for binding reactions. Wt and mutant ParR171 protein was serially titrated into the cuvette at concentrations ranging from 0–5 μM in binding buffer (20 mM Tris–HCl pH 8.8; 25 mM NaCl; 1 mM DTT; 1 mM EDTA) and allowed to equilibrate for 2 min before measurement.
We gratefully acknowledge support from the staff at beamline ID14eh4 at the ESRF (Grenoble, France). JMJ was supported by fellowships from EMBO and the Danish Research Council.
- Copyright © 2007 European Molecular Biology Organization