The dynamic, mitosis-like segregation of bacterial chromosomes and plasmids often involves proteins of the ParA (ATPase) and ParB (DNA-binding protein) families. The conversion of multigenomic aerial hyphae of the mycelial organism Streptomyces coelicolor into chains of unigenomic spores requires the synchronous segregation of multiple chromosomes, providing an unusual context for chromosome segregation. Correct spatial organization of the oriC-proximal region prior to septum formation is achieved by the assembly of ParB into segregation complexes (Jakimowicz et al., 2005; J Bacteriol 187: 3572–3580). Here, we focus on the contribution of ParA to sporulation-associated chromosome segregation. Elimination of ParA strongly affects not only chromosome segregation but also septation. In wild type hyphae about to undergo sporulation, immunostained ParA was observed as a stretched double-helical filament, which accompanies the formation of ParB foci. We show that ParA mediates efficient assembly of ParB complexes in vivo and in vitro, and that ATP binding is crucial for ParA dimerization and interaction with ParB but not for ParA localization in vivo. We suggest that S. coelicolor ParA provides scaffolding for proper distribution of ParB complexes and consequently controls synchronized segregation of several dozens of chromosomes, possibly mediating a segregation and septation checkpoint.
Studies of bacterial chromosome segregation have focused mostly on rod-shaped bacteria dividing by binary fission, particularly Escherichia coli, Bacillus subtilis and Caulobacter crescentus (reviewed recently by Errington et al., 2005; Leonard et al., 2005a; Hayes and Barilla, 2006). Bacterial chromosomes are actively segregated, as illustrated by the rapid movement of oriC regions towards cell poles soon after the start of replication (Webb et al., 1997; Teleman et al., 1998; Lewis, 2004; Viollier et al., 2004; Fogel and Waldor, 2005). However, in Streptomyces movement of chromosomes towards cell poles cannot account for all aspects of segregation. These organisms are filamentous, growing by tip extension and hyphal branching to form a dense mycelial network of vegetative hyphae in which occasional septa separate adjacent multigenomic compartments (Flardh, 2003a). Older Streptomyces colonies differentiate to form aerial hyphae, which subsequently develop into chains of spores (Chater, 2001). Rapid growth of aerial hyphae is accompanied by intensive replication of the chromosomes, such that 50 or more non-segregated chromosomes may be present in one long tip compartment (Ruban-Osmialowska et al., 2006). Sporulation then starts with the synchronous placement of regularly spaced Z-rings along the multigenomic compartment, converting it into a chain of unigenomic prespore compartments, which further mature into spores (Schwedock et al., 1997). The early stages of sporulation septation often occur over non-segregated chromosomes, and separation of nucleoids is not observed until septal constriction has started (Flardh, 2003a). In contrast to the movement of chromosomes towards poles in rod-shaped bacteria, dozens of Streptomyces chromosomes are condensed and uniformly aligned along the hyphal tip ensuring that each prespore compartment receives a single copy.
Recent experiments indicate that the chromosomally encoded ParAB proteins are involved in proper organization of the oriC-proximal region rather than its localization, but their roles remain less clear than those of the plasmid proteins. In C. crescentus, the parAB genes are essential for viability (Mohl and Gober, 1997), but in some other bacteria (B. subtilis, Pseudomonas putida) deletion of the parAB homologues results in minor segregation defects (Ireton et al., 1994; Lewis et al., 2002). Interestingly, deletion of Vibrio cholerae parA alone only has a mild effect on cell division (Fogel and Waldor, 2006), and in B. subtilis deletion of soj (=parA) does not affect cellular morphology or chromosome segregation (Ireton et al., 1994). The B. subtilis Soj is involved in the formation of compact Spo0J (ParB) (Marston and Errington, 1999) and ParAI is required for proper localization of ParBI nucleoprotein complexes during segregation of V. cholerae chromosome I (Fogel and Waldor, 2006). On the other hand, C. crescentus ParA dissociates ParB from its binding site in vitro, and overexpression of ParA results in a severe division defect leading to filamentous cells devoid of ParB foci (Easter and Gober, 2002).
Structural and in vitro studies revealed that the ParA homologue of Thermus thermophilus forms a sandwich dimer and can oligomerize to form filaments (Leonard et al., 2005b). Self-interaction of ParA is consistent with previous in vivo localization studies: B. subtilis GFP–Soj fusion protein formed a large fluorescent patch oscillating within nucleoids between cell poles (Marston and Errington, 1999; Quisel et al., 1999 ); and dynamic localization of ParAI extending from the pole towards the septum was also found in V. cholerae (Fogel and Waldor, 2006). C. crescentus ParA localizes to the cell pole but has not been assayed for oscillation in living cells (Mohl and Gober, 1997). Dynamic localization of Soj/ParA in B. subtilis and V. cholerae is disrupted by mutation in the nucleotide binding site and by spo0J/parB mutations (Marston and Errington, 1999; Quisel et al., 1999; Autret et al., 2001; Fogel and Waldor, 2006). Thus, although ParA and ParB do not fully colocalize, genetic evidence suggests functional interaction of the two proteins in all bacteria studied. Direct interaction of ParA and ParB was confirmed in B. subtilis by co-immunoprecipitation (Ogura et al., 2003) and in Pseudomonas aeruginosa by yeast two-hybrid screening (Bartosik et al., 2004).
Streptomyces coelicolor chromosomes contain a parAB operon that is developmentally regulated, with one of its two promoters strongly upregulated shortly before sporulation septation (Kim et al., 2000; Jakimowicz et al., 2006). Up to now, studies on the role of this operon in chromosome segregation have focused mostly on ParB. Deletion of parB, or of the whole parAB operon, or elimination of the ParB DNA-binding motif, resulted in relatively frequent anucleate spores (about 15%), but did not visibly affect colony growth or sporulation. ParB binds to about 20 parS sites near oriC, and also affects the sequences flanking recognition sites, suggesting non-specific binding and/or oligomerization of the protein (Jakimowicz et al., 2002). This results in formation of large complexes that could be visualized in vivo as bright ParB–EGFP foci, which behave differently during vegetative growth and in sporulating aerial hyphae. Most vegetative hyphal tips and young aerial hyphae have a bright focus close to the tip, other foci being smaller and irregularly spaced through the length of the hyphae. However, during sporulation, arrays of regularly spaced ParB complexes assemble along the aerial hyphal tip compartment. The complexes form before DNA segregation and septation, and disassemble soon after these processes are completed (Jakimowicz et al., 2005). We have postulated that ParB complexes assist DNA condensation and segregation, most likely by mediating proper organization and/or positioning of the oriC-proximal part of the chromosome.
Here, we focus on the contribution of ParA to sporulation-associated chromosome segregation. We show that ParA forms extensive filaments in vivo even when its ATPase activity is eliminated, but that ATP binding is needed for the filaments to function in the assembly of ParB complexes and hence in chromosome segregation.
Deletion of parA results in irregular septation and chromosome segregation
The role of ParB in efficient segregation of multiple chromosomes into prespore compartments, and the fact that parAB genes form a developmentally regulated operon, suggested that ParA might also be important in this process. To test this, J3306, a strain carrying a non-polar, unmarked, in-frame parA deletion, was constructed by PCR targeting of the wild type strain M145. In S. coelicolor, the parB start codon and the parA stop codon overlap, indicating a possible translational coupling, so the last 36 bp of parA adjacent to parB as well as the parA start codon were left intact, to avoid disturbance of parB translation. Western blot analysis indicated that ParB was expressed in the parA deletion strain, though at a slightly lower level than in the wild type (Fig. S1).
J3306 showed no obvious defect in growth or morphology when cultured in liquid or on solid medium, but severe disturbances in chromosome segregation and septum placement in aerial hyphae were seen (Fig. 1) when the mutant was analysed by fluorescence microscopy after staining with DAPI (to visualize DNA) and wheat germ agglutinin (WGA) conjugated with Texas red (to visualize cell walls). Prespore compartments were of uneven size compared with the wild type, with a significant fraction (16%) of minicompartments smaller than 0.7 μm and increased occurrence of very long compartments (more than 2 or even 3 μm). The average length in μm of prespore compartments was 1.3 ± 0.3 in J3306 and 1.4 ± 0.1 in the wild type M145; thus the variance of distances between septa in the parA deletion strain was three times higher than in the wild type. Most minicompartments, and some normal-sized compartments, lacked DNA, resulting overall in 26.1% of J3306 prespores being anucleate, in comparison with 1.8% in the wild type (Fig. 1). Interestingly, the segregation defect of J3306 was more pronounced than in a parB deletion strain J2537 (17.4%; Fig. 1, Kim et al., 2000), and the latter strain did not produce so many prespore minicompartments (10.4%). This makes it unlikely that the slight deficiency of ParB in the parA mutant might be solely responsible for these phenotypic effects (see also below).
In order to verify that the phenotypic effects were attributable to parA deletion, a wild type copy of parA with its own promoter was cloned into the integrating vector pSET152 and introduced in J3306, creating J3307. Segregation efficiency in J3307 (6.6% DNA-free compartments, 6.7% minicompartments) was intermediate between that of the wild type and that of J3306 (Fig. 1A). Strain J3308, in which a 5.3 kb fragment of chromosomal DNA encompassing the whole parAB operon was used for complementation, displayed a phenotype more closely resembling the wild type (4.4% DNA-free compartments, 5.0% minicompartments). Thus, although ParA deficiency is responsible for much of the segregation defect, complementation was incomplete. We hypothesize that this may reflect disturbances in the balance of ParA and ParB levels, and/or the ectopic location of the complementing gene and associated ParB binding sites in relation to the oriC–ParB complex.
In conclusion, elimination of ParA had even greater effects on sporulation-associated chromosome partitioning than that of ParB, in contrast with the greater effects of ParB elimination in other bacterial systems. Moreover, unexpectedly, it had significant effects on the positioning of septa.
ParA assembles into helical structures along the aerial hyphae at the time of Z-ring formation
Our data suggest a role for ParA in the correct segregation of DNA during later stages of aerial development. Because of the length of multigenomic, sporulating aerial hyphae, it seemed possible that this role might require ParA to form a structure along the compartments. To localize ParA in hyphae, antibodies raised against purified recombinant ParA (see Experimental procedures) were used in immunofluorescence microscopy. ParA was detected at the tips of about 50% of vegetative hyphae, and at the tips of short, possibly nascent aerial hyphae (Fig. 2A1). In longer aerial hyphae the localization of ParA varied. It was either accumulated at the tips of hyphae extending for a variable distance from the tip (Fig. 2A2, see also Fig. 9A), or was spread along the aerial hyphae (the latter distribution was almost always observed in long branches, Fig. 2A3, see also Fig. 9B). Interestingly, the pattern of the fluorescence signal indicated a spiral structure. Deconvolution microscopy confirmed the presence of a pair of stretched helical filaments with a pitch about 1.4 μm (Fig. 2, and Fig. 4A, i and o), which corresponds to the length of prespore compartments. We could not detect such a signal in the parA deletion strain (Fig. 2).
In order to correlate ParA localization with particular stages of aerial hyphal development, ParA was immunolocalized in strain J3310, which expresses ParB–EGFP instead of the wild type ParB (Jakimowicz et al., 2005), and in strain K202, in which, in addition to FtsZ encoded by the normal chromosomal copy, FtsZ–EGFP fusion protein is expressed from an integrated plasmid (Grantcharova et al., 2005). During aerial hyphal development, regularly distributed ParB complexes have been shown to be formed before DNA condensation and septation, and they disassemble shortly after cytokinesis is complete (Jakimowicz et al., 2005), while FtsZ forms spiral-shaped filaments along hyphae that are gradually replaced by FtsZ rings (Grantcharova et al., 2005). Assembly of ParB–oriC complexes seems to be independent of synchronous Z-ring formation (Jakimowicz et al., 2006). Here, we have analysed hyphae after 44 h growth, at the time point at which the highest induction of the developmentally specific p2 promoter of parAB was observed (Kim et al., 2000). However, it is important to bear in mind that there is some developmental asynchrony between aerial hyphae during colony maturation. ParA immunolocalization was classified into three patterns: A1 – only tip-associated; A2 – abundant at the tips but somewhat extended along the hyphae; or A3 – evenly spread along the hyphae. Hyphae containing ParB–EGFP foci were classified as: B1 – hyphae with irregular, weak ParB foci and a tip-associated ParB complex; B2 – unseptated hyphae containing arrays of regular ParB complexes; or B3 – hyphae with ParB complexes and evident septa. Similarly, hyphae exhibiting FtsZ signals fell into three subclasses: Z1 – spirals along hyphae; Z2 – some spirals and some irregular Z-rings; and Z3 – regular Z-rings. A0 or Z0 indicates no particular fluorescence signal for ParA or FtsZ respectively. The results for hyphae exhibiting any fluorescence signal (105 for K202 and 106 for J3310) are summarized in Figs 3 and 4. It was found that ParA assembles at tips of aerial hyphae (group A1) at a very early stage of their development, before the assembly of ParB complexes and formation of FtsZ spirals (Fig. 3A, a, and Fig. 4A, a and d). Spreading of ParA filaments from the tips preceded the formation of FtsZ spirals (A2/Z0 – 26%, A3/Z0 – 10%, A3/Z1 – 20% of hyphae), which was followed by Z-ring assembly (A3/Z2 – 11% of hyphae) (Fig. 3A, d, g and j). Finally, when the Z-rings were present, ParA signals disappeared (A0/Z3 – 16%) (Fig. 3A, m). Remarkably, although ParA and FtsZ helices were both present in some hyphae, they did not colocalize (Fig. 3A, g and j). ParA filaments had stretched along hyphae before regular ParB foci were formed (A2/B1 – 27%, A3/B1 – 18% of hyphae) (Fig. 4A, d), were present during ParB assembly (A3/B2 – 16%) (Fig. 4A, g, j and m), and shortly afterwards disassembled. ParA was never detected in hyphae containing septa (A0/B3 – 10%) (Fig. 4A, m). ParA filaments were not dependent on ParB – they had the same appearance in the parB deletion strain as in the wild type (Fig. 2).
Thus, in the long apical compartments of aerial hyphae, the ParA-dependent partitioning of many chromosomes does indeed involve the formation of a pair of developmentally specific extended helical filaments, at a very early stage of aerial hyphal sporulation. We have provided evidence that chromosomal ParA can form extended helical filaments in vivo, reinforcing previous studies which revealed the dynamic movement of (mainly plasmid-encoded) ParA homologues in rod-shaped cells along a helical path (Ebersbach and Gerdes, 2004; Adachi et al., 2006).
ParA assists the formation of ParB foci in vivo
ParA and ParB did not colocalize, but ParB foci assembled in the presence of ParA, raising the possibility that ParB localization or complex formation might depend on ParA. Therefore, we analysed ParB localization in a parA deletion strain. The parA deletion was introduced into strain J3310, which expressed ParB–EGFP protein instead of the wild type ParB, to give strain J3318. The formation of ParB–EGFP complexes was then monitored microscopically. Foci were present in both vegetative and aerial hyphae, but they usually appeared smaller, more scattered and less intensely fluorescent than in the parental strain J3310 (Fig. 5A). We measured the relative intensity (maximal pixel value with background signal subtracted) of foci in images of aerial hyphae of parA and its parent strain J3310 taken with the same camera settings. The average signal intensity of ParB foci in ΔparA was about 58% of the wild type (average of 350 units for 430 foci in J3318 and 600 units for 420 foci in J3310) (Fig. 5B). Additionally, there were fewer tip-associated foci in both vegetative and growing aerial hyphae in J3318. In parent strain J3310, about 90% of vegetative hyphal tips had ParB complexes associated with them, at a constant distance of 1.4 μm from the tip (Jakimowicz et al., 2005), while in J3318 more than 50% of the short branches did not contain any ParB foci, and in those hyphae that did possess ParB complexes, the average distance from tip to focus was 2.5 μm. These results indicated that ParA filaments are required for proper localization of ParB foci, suggesting they may provide a scaffolding to arrange complexes along hyphae, prior to their capture by prespore compartments. The fact that the ParB foci are more scattered and less intensely fluorescent in the absence of ParA may also suggest that ParA assists ParB–DNA interaction and the formation of high-order and/or more compact nucleoprotein complexes.
ParA stimulates binding of ParB to DNA in vitro
As Western blots, confirmed by transcriptional analysis, indicated that the ParB level in the parA deletion strain was slightly lowered in comparison with the wild type (Fig. S1), it was possible that reduced levels of ParB–EGFP per se might be the cause of its failure to form proper complexes, rather than some involvement of ParA in the formation of the complexes. This question could be addressed by using purified recombinant proteins for in vitro binding studies of the effects of ParA on ParB–DNA binding and/or oligomerization. ParB was purified as described earlier (Jakimowicz et al., 2002), and ParA was cleaved from an overexpressed glutathione S-transferase (GST) fusion protein. The purified ParA protein was used for in vitro studies of ParA–ParB interaction. Approaches involving affinity chromatography and gel supershift did not show formation of any stable ParA–ParB complexes. However, the application of DNase I footprinting and surface plasmon resonance (SPR) indicated that ParA influenced the affinity of ParB for parS-containing DNA. In these experiments ParA alone did not show DNA binding. In the footprinting experiment, incubation of increasing amounts of the ParB protein with a radiolabelled 510 bp DNA fragment containing two parS sequences separated by 280 bp resulted in slight protection of the parS sequence and also, more obviously, in enhanced DNase I sensitivity in certain positions flanking the binding site. Such a pattern has been observed previously, and was taken to suggest that the large nucleoprotein complex affects sequences adjacent to the binding site (Jakimowicz et al., 2002). Addition of ParA–ATP enhanced the ParB-dependent parS protection and DNase I sensitivity of the DNA fragment (Fig. 6A). The effect was more obvious at low ParB concentrations (60 nM and 200 nM), at which ParB alone did not bind efficiently to the DNA. We obtained similar results when binding of ParB to a 180 bp DNA fragment containing two parS sites was analysed by SPR. Increasing concentrations of ParA-ATP pre-mixed with ParB visibly increased ParB–DNA binding (Fig. 6B), promoting faster ParB–parS association (50 times increased ka at ratio ParB : ParA 1:1), and also enhancing non-specific binding to a control DNA fragment that did not contain parS sites (data not shown). Thus, our in vitro data supported the in vivo observation that, although ParA is not indispensable for ParB–DNA binding, in its presence, formation of the nucleoprotein complex is more efficient.
The ATP binding site of ParA is essential for its function during sporulation, but not for localization
Like other ParA family proteins, ParA from S. coelicolor contains a Walker A motif in its N-terminal domain. Indeed, purified recombinant ParA exhibited ATPase activity, which was elevated in the presence of ParB (about 150% of ParA intrinsic activity) (Fig. 7). It was shown previously for other ParA proteins that a lysine residue present in the Walker motif A (GXXXXGKT/S in the N-terminal part of ParA) is essential for nucleotide binding and protein dimerization (Leonard et al., 2005b). We therefore expected that the ATP binding of ParA would be necessary for the assembly of filaments, although it was alternatively possible that dimerization might be quite distinct from the formation of extended structures. In order to find out the biological consequences of abolishing ATP binding by ParA in S. coelicolor, we constructed a strain J3344 in which a K44E mutation was introduced into the Walker A motif of ParA. The mutation was inserted into parA at its normal chromosomal locus. The J3344 phenotype was similar to that of the parA deletion mutant, but if anything was slightly more severe (see Fig. 1), the ParAK44E mutation caused irregular septation, leading to a high frequency of minicompartments (16%), occurrence of elongated compartments (variance of spacing between the septa was 0.3 μm in the mutant compared with 0.1 μm in the wild type) and highly disturbed chromosome positioning, with almost 30% of empty compartments. Surprisingly, immunolocalization of ParAK44E in J3344 showed that it was arranged in aerial hyphae in apparently the same helical filaments as in the wild type strain (Fig. 2). Thus, ATP binding appears to be needed for ParA to play its part in DNA partitioning, but only at a stage after ParA filaments have formed.
ParA dimerization and interaction with ParB depend on the Walker A motif
The observed ParA filaments might involve either interactions between ParA monomers, or interactions of ParA with another cellular component. We therefore sought evidence of direct interactions of ParA, using a bacterial two-hybrid system (Karimova et al., 1998; 2000). The S. coelicolor parA gene and its K44E mutant form were expressed as fusion proteins with the T25 and T18 subdomains of Bordetella pertussis adenylate cyclase in E. coli strain BTH101. Interaction of fusion proteins was indicated by a red appearance on McConkey medium, resulting from the restoration of the cAMP signal cascade and the consequent fermentation of maltose (Fig. 8). This visual test, which was confirmed by quantitative β-galactosidase enzyme assay in liquid cultures (data not shown), provided strong evidence of ParA–ParA interaction. However, in contrast with the ATPase independence of ParA filament formation in vivo, the Walker A motif K44E mutation impaired the two-hybrid interaction. This indicated that the ParA dimerization revealed by the two-hybrid analysis was not the same as the interaction needed for filamentation, and that the latter interaction might require another cellular component. Two-hybrid evidence was also sought for possible interactions with FtsZ, MreB or DivIVA, other proteins that exhibit a similar localization pattern in S. coelicolor aerial hyphae (Schwedock et al., 1997; Flardh, 2003b; Mazza et al., 2006), but there was no signal in the experiments with any of them (data not shown). These negative results might either reflect a technical limitation of the methods employed, or indicate that spiral filament involves the recruitment of ParA by some other cytoskeletal protein. Additionally, the two-hybrid assay system revealed a direct ParA–ParB interaction, which was also dependent on the integrity of the Walker A motif, thus suggesting that ParB may interact with ParA dimers.
In rod-shaped bacteria dividing by binary fission, nascent chromosomes are segregated into the two halves of the elongating cell. In contrast, Streptomyces chromosome segregation during sporulation requires precise positioning of several dozen chromosomes along the hyphal compartment. Our results demonstrated that ParA is required for efficient chromosome segregation and regular septation during sporulation (Fig. 1), and that it has a weak ATPase activity that is needed for these functions (Fig. 7). Here we set these basic aspects of Streptomyces ParA function in the context of the sequence of molecular and cellular events that lead to the formation of prespore compartments, and relate our results to what is known about the behaviour of ParA orthologues in other bacteria.
ATPase-independent formation of extended ParA filaments in aerial hyphae
In young aerial hyphae ParA accumulates at the tips, but as hyphae elongate ParA extends back from the tip in the form of pair of helical filaments, reaching tens of microns in length in full-length hyphae (Fig. 2). This is the first observation that any ParA homologue can form such extended structures. Previously, short and curved but highly dynamic structures have been observed in studies of chromosomally encoded ParAI of V. cholerae, while ParA of plasmid pB171 and SopA of plasmid F were found to oscillate along a helical track between nucleoid poles in E. coli (Ebersbach and Gerdes, 2004; Adachi et al., 2006; Fogel and Waldor, 2006). In the absence of ParB, ParA of pB171 formed stationary helices and in fixed wild type cells, the helical signal was seen as a drop-shaped structure in one half of the cell, with the brightest signal at one pole as if ParA was accumulating or disintegrating: a pattern closely resembling that of ParA in young Streptomyces aerial hyphae (Ebersbach and Gerdes, 2004).
Surprisingly, the formation of ParA helical filaments in Streptomyces hyphae was unaffected by mutation of the ATP-binding Walker A conserved lysine (Hayashi et al., 2001), which eliminated ParA–ATP binding and ATPase activity. Mutation of the crucial lysine in the Walker A motif has severe effects in vivo on all ParA homologues tested so far. In B. subtilis and V. cholerae the corresponding mutation (K16Q or K16E respectively) caused loss of cell pole-associated dynamic localization of the protein (Quisel et al., 1999; Fogel and Waldor, 2006). Similarly, corresponding mutations in plasmid ParA homologues (TP228 ParF and pB171 ParA and others studied) entirely abrogated plasmid segregation, probably due to inefficient protein polymerization (Ebersbach and Gerdes, 2001; Li et al., 2004; Barilla et al., 2005).
This surprising difference of Streptomyces ParA from other bacteria leads us to suggest that its filament formation may not be intrinsic property, but instead involves association of ParA with some other cytoskeletal scaffolding. However, limited two-hybrid studies failed to reveal evidence of interactions with some candidate proteins (FtsZ, MreB, DivIVA).
ParA mediates formation of ParB complexes in an ATPase-requiring manner
Our previous studies showed that ParB of S. coelicolor binds to numerous parS sites scattered over a 400 kb chromosomal segment containing oriC, and assembles into a large nucleoprotein complex that is necessary for proper chromosome organization and/or localization in pre-sporulating aerial hyphae. As the parA mutants studied here showed disturbed chromosome segregation into spores, it seemed possible that ParA might be involved in the formation or activity of ParB complexes. Indeed, our in vivo observations clearly indicated an influence of ParA on ParB complexes. ParA filament formation along the hyphae preceded the assembly of arrays of high-order ParB complexes, and the ParB complexes, both in aerial hyphae and at the tips of vegetative hyphae, were formed inefficiently in a parA deletion mutant. This is reminiscent of observations in B. subtilis, in which ParA is required for the coalescence of scattered ParB complexes around oriC into larger foci (Glaser et al., 1997; Lewis and Errington, 1997; Lin et al., 1997; Marston and Errington, 1999), while in V. cholerae ParBI foci are formed but mislocalized in the absence of ParAI (Fogel and Waldor, 2006). In C. crescentus, overproduction of ParA abolished ParB–DNA binding activity (Mohl and Gober, 1997).
Our in vitro footprinting and SPR results, coupled with evidence from two-hybrid analysis, indicated that S. coelicolor ParA interacts directly with ParB to promote ParB–DNA binding. This interaction, and the proper segregation of chromosomes into prespore compartments, both failed when the ATPase (K44E) mutant form was substituted for the wild type ParA. The in vitro effect of ParA on the interaction of ParB with DNA was mirrored by the stimulatory effect of ParB on the ATPase activity of ParA (Fig. 7). This would be consistent with the role for the ATPase function in the assembly of ParB complexes suggested by the in vivo results. The intact Walker A motif was also required for the ParA–ParA dimerization detected by two-hybrid analysis. It is attractive to postulate that the transition from a state in which ParA is in the form of filaments and ParB complexes are not yet present, to a state in which ParB–oriC complexes have formed and ParA filaments have disappeared, is ParB-stimulated and ATPase-dependent. Previously, it was suggested that ATP-dependent formation of the Soj (ParA) dimer in B. subtilis is a prerequisite for interaction with its partner Spo0J (ParB) protein (Leonard et al., 2005a,b). It was also proposed that ParA filament formation is accompanied by ATP hydrolysis while ParA-ADP filament dissociates with the release of ADP in the presence of ParB, which may act as a nucleotide exchange factor (Figge et al., 2003; Ebersbach and Gerdes, 2004; Leonard et al., 2005a; Ebersbach et al., 2006).
ParA ensures regular septation
In addition to their chromosome segregation deficiency, parA deletion or ATPase-defective mutants also showed frequent abnormal spacing between Z-rings (data not shown) and between sporulation septa (Fig. 1). This effect could either be direct, with ParA providing some sort of scaffolding for Z-rings, or perhaps be an indirect result of disturbances in ParB complex formation or segregation of the oriC regions. The significantly more marked effect on septation of parA deletion than of parB deletion does not help to distinguish between the direct or indirect models. The extension of ParA filaments in aerial hyphae coincided with the formation of helical precursors of Z-rings, so some linkage between the spiral ParA and FtsZ structures, perhaps via a third element(s), is possible, even though no direct ParA–FtsZ interaction was found by two-hybrid analysis, the spirals of ParA and FtsZ in aerial hyphae did not superimpose, and Z-rings formed synchronously in a mutant devoid of ParA. In C. crescentus, overexpression of parA led to inhibition of FtsZ ring assembly, and resulted in a cell division defect (Mohl et al., 2001; Easter and Gober, 2002); it was proposed that ParA is a sensor for detecting proper chromosome segregation as a checkpoint for cell division (Figge et al., 2003). Possibly, Streptomyces ParA also plays a role in co-ordinating chromosome segregation with cell division. Apart from ParA, additional proteins (Ssg family, CrgA) have been described that have effects on sporulation septation, but their modes of action and whether they affect the cell division machinery directly or indirectly are not clear (van Wezel et al., 2000; Del Sol et al., 2003; Keijser et al., 2003). Currently, we are screening for ParA interaction with the other proteins engaged in development of aerial hyphae.
On the basis of our results we propose the following model (Fig. 9). During growth of aerial hyphae ParA accumulated at the tips spreads along the hyphae as a pair of helical filaments. Filament formation does not require ATP binding by ParA. With the assistance of ParA filaments, regularly spaced ParB complexes form on the oriC-proximal part of the chromosome, placing this region between the sites of future septa. The ATPase activity of ParA, which is stimulated by ParB, appears to be necessary for the proper assembly of segregation complexes. We suggest that ParA filaments themselves act in some way as a ‘ruler’ for the spacing of the complexes and, directly or indirectly, for the positioning of FtsZ rings, as the spacing of Z-rings is disturbed in parA mutants. Ebersbach et al. (2006) and Adachi et al. (2006) proposed similar functions for ParA of plasmid pB171 and SopA of plasmid F, in positioning plasmids regularly over the bacterial nucleoid. In Streptomyces, the ruler function would require the ParA ATPase activity, as septal positioning is severely disturbed in the ParAK44E ATPase mutant. The apparent ruler function of Streptomyces ParA, positioning (possibly dynamically) several dozens of segregation complexes along the extended filament and ensuring that each spore receives a single copy of chromosome, provides a new perspective on the suggestion of Leonard et al. (2005a) that ParA homologues provide a motor force for pushing or pulling nucleoprotein complexes towards cell poles. After segregation complexes are correctly arrayed, the filaments disassemble. Disassembly does not depend on the formation of segregation complexes or ParA ATPase activity, and ongoing experiments should clarify whether it happens in response to some other event in sporulation septation, such as the ingrowth of FtsZ rings. It is attractive to think that there should be checkpoint cross-talk between the cytoskeletal elements contributing to successful cytokinesis during sporulation.
DNA manipulation, bacterial strains and growth conditions
DNA manipulations were carried out by standard protocols (Sambrook et al., 1989). Enzymes were supplied by Roche, Fermentas or New England BioLabs, isotopes were from Amersham-Pharmacia-Biotech and oligonucleotides were from Invitrogen. The S. coelicolor and E. coli strains are listed in Table 1. Culture conditions, antibiotic concentrations, and transformation and conjugation methods followed general procedures for E. coli (Sambrook et al., 1989) and Streptomyces (Kieser et al., 2000). S. coelicolor was cultivated in TSB : YEME (1:1) complex liquid medium or on MS (soy flour) agar plates unless otherwise stated.
Streptomyces coelicolor mutants were constructed by PCR targeting (Gust et al., 2003). We used a two-step procedure involving, first, substitution of the parA gene within cosmid H24 with apra-oriT, an apramycin resistance cassette flanked by FRT sites (FLP recombinase recognition targets) and, second, FLP-mediated excision of the cassette, leaving only an in-frame ‘scar’ sequence. The cassette was amplified from pIJ773 with primers pA-del-fw, and pA-del-rv (Table 2) and used for transformation of arabinose-induced BW25113/pIJ790 carrying cosmid H24 (Redenbach et al., 1996), resulting in the deletion of parA through to a position 36 nucleotides upstream of the adjacent parB start codon. The resulting cosmid H24ΔparA-apra was subjected to FLP-mediated cassette excision to generate an in-frame deletion. Subsequently the kan gene in the SuperCos part of cosmid H24ΔparA was exchanged for a vph-oriT cassette, yielding H24ΔparA kan::vio-oriT. This construct was used for conjugation into S. coelicolor J2538 [parAB::apra] (Kim et al., 2000). VioR exconjugants were screened for the loss of both VioR and ApraR, indicating a double-cross-over allelic exchange of the parAB locus of J2538 giving strain J3306.
To introduce a parA deletion into cosmid H24 containing the parB–egfp gene, we used a construct in which the parAB promoter region was replaced by the apra cassette flanked by unique SwaI restriction sites (Jakimowicz et al., 2006). H24 parAB promoter::apra, parB–egfp, kan::vio-oriT was linearized with SwaI and used for co-electroporation of arabinose-induced BW25113/pIJ790 with a PCR product encompassing the promoter region and parA deletion, obtained with H24ΔparA as a template. ApraS transformants were verified by restriction digestion, and were used for conjugation into S. coelicolor strain J2538 as described earlier.
Cosmid H24 was also used in a two-step procedure to construct a point mutation in the parA Walker A motif. First the apra cassette, amplified with oligonuclotides pATPase_Bst_fw and pATPase_Bst_rv flanked by unique BstZ17I restriction sites, was inserted between nucleotides 129 and 135 of parA in cosmid H24. Subsequently, H24BstZApr was linearized with BstZ17I and used for co-electroporation of arabinose-induced BW25113/pIJ790 with the 92 nt oligonucleotide pK_Emut encompassing the region encoding K44. ApraS transformants were screened for the mutations by restriction digestion and clones verified by sequencing. Subsequently the kan gene in the SuperCos part of cosmid H24parAK44E was exchanged for a vio-oriT cassette and used to replace the ApraR-marked parA locus of S. coelicolor strain J2538 (as described above for strain 3306) to give strain J3344.
Strains expressing an additional copy of parA were made by introducing derivatives of the integrative plasmids pSET152 or pIJ82 (pSET152 derivative harbouring a hygromycin resistance gene) (Kieser et al., 2000) containing either wild type or modified parA under control of its own promoter. A 1300 bp fragment encompassing parA and its promoter was amplified on a chromosomal DNA template with the primers pparApSETfw and pparAEcorv, and cloned between BamHI and EcoRI sites in pIJ82 to give pDJ01. Chromosomal DNA of all strains constructed was checked by PCR, and sequencing in the case of unmarked mutations; and cell-free extracts were checked by phosphoimager scanning of SDS-PAGE gels and by Western blotting using antibodies against ParB protein as described previously (Jakimowicz et al., 2002).
Strains for microscopic observations were inoculated in the acute-angled junction of coverslips inserted at 45° in MM agar containing 1% mannitol (Kieser et al., 2000), and cultured for 42–44 h. Staining procedures were as described previously (Schwedock et al., 1997). Samples were fixed for 10 min with paraformaldehyde/glutaraldehyde mixture, digested 2 min with 1 mg ml−1 lysozyme, washed with PBS and blocked with 2% BSA. For immunostaining, samples were incubated with antibody against ParA (1:5000 dilution) overnight, washed six times with PBS and then incubated for 1 h with secondary antibody (anti-rabbit) conjugated with Alexa Fluor546. For DNA staining, samples were incubated with 0.1–1 μg ml−1 DAPI (Molecular Probes) and for cell wall visualization with 10 μg ml−1 WGA-Texas red or Alexa Fluor350 conjugate (Molecular Probes). After five washes with PBS, coverslips were mounted in Vecta-Shield (Vector Laboratories) antifading reagent. Florescence microscopy was carried out using a Nicon Eclipse microscope with a Hammamatsu camera or Zeiss AxioImager M1 with camera AxioCam MRm Rev. 2. For two-dimensional deconvolution, a series of 20 images in different focal planes spaced by 0.1 μm was captured, creating a Z-stack. All the images were analysed by AxioVision Rel. 4.5 software with deconvolution mode.
The ParA protein of S. coelicolor was fused to the C-terminus of GST. The parA gene was amplified using the primers, p2parABamHIfw, pparAEcoRIrv (Table 2). The amplified fragment digested with BamHI, EcoRI was cloned between the BamHI, EcoRI sites of pGEX-6P-2 (Amersham-Pharmacia-Biotech) to give the expression construct DJ09. The GST–ParA fusion protein was overexpressed in E. coli BL21, and purified from soluble proteins in a one-step procedure involving affinity chromatography on glutathione-Sepharose 4B beads (Amersham-Pharmacia-Biotech) as described earlier (Majka et al., 1999). For removal of the GST part, the bound fusion protein was treated with the PreScissionTM protease (Amersham-Pharmacia-Biotech) at 4°C for 12–16 h, and the ParA protein was released from the beads. The purified protein was 95% pure on the basis of SDS-polyacrylamide gel electrophoresis, but according to Western blot analysis (data not shown) it was about 5 kDa larger than ParA in a S. coelicolor cell extract. This suggested that the major natural translation start point was downstream of the one selected for use in the expression construct, and that authentic ParA protein consists of 307 aa. The recombinant ParA protein was used to obtain rabbit polyclonal antibody. The anti-ParA antibody was affinity-purified on a ParA column as described earlier (Jakimowicz et al., 2002). The purified antibody recognized only ParA in Western blots of whole-cell extracts.
DNase I footprinting
For DNase I footprinting experiments a 510-bp-long PCR product obtained with primers pparApSETfw and pparArv and encompassing parAB promoter region with two parS sites was used. The 5′-end-radiolabelled DNA fragments (~10 fmoles) were incubated with different amounts of the ParB protein in the binding buffer (20 mM HEPES/KOH, pH 7.6, 5 mM magnesium acetate, 100 mM sodium chloride, 1 mM EDTA, 4 mM dithiothreitol, 0.2% Triton X-100, 5 mg ml−1 BSA and 1 mM ATP) at 30°C for 30 min. After DNase I digestion (Majka et al., 1999), the cleavage products were separated in an 8% polyacrylamide-urea sequencing gel. Gels were dried and analysed by autoradiography.
For the standard SPR analysis, a 180 bp PCR product containing 2 parS sites was obtained with biotinylated oligonucleotide pbiot4440rv and non-biotinylated oligonucleotides pparApSETfw, and then immobilized on to the Sensor Chip SA surface as described previously (Majka et al., 1999). A total of 200–250 response units (RU) of DNA were immobilized. DNA loosely attached to the surface of the chip was removed with a 2 min pulse of 0.05% SDS. As a control, the same fragment of DNA but containing a mutated parS sequence was used. SPR analysis was performed on a BIAcore 1000, by injecting 50 μl of protein solutions in binding buffer (20 mM HEPES/KOH, 5 mM magnesium acetate, 100 mM sodium chloride, 1 mM EDTA, 4 mM dithiothreitol, 0.2% Triton X-100, 80 mg ml−1 BSA and 1 mM ATP) for 5 min at flow rate 10 μl min−1 at room temperature. Protein (ParB, ParA) injections were followed by SPR buffer (10 mM HEPES/KOH, 100 mM NaCl, 10 mM magnesium acetare, 0.005% BIAcore surfactant P20) for 8 min. After each protein binding analysis, the surface of the chip was regenerated by injection of 40 μl of 0.05% SDS, which releases all bound protein without affecting the binding capacity of the immobilized DNA. The results were plotted as sensorgrams after subtraction of the background response signal obtained in a control experiment. The biaevaluation ver.2.1 program (Pharmacia Biosensor AB) was used for data analysis.
Bacterial two-hybrid system
To construct the recombinant plasmids used in the bacterial two-hybrid system (Karimova et al., 1998; 2000), parA (or its mutant forms) or parB were PCR-amplified using appropriate primers (pETH_SA_F, pETH_SA_R for parA and pETH_SB_F, pETH_SB_R for parB) (Table 2), with chromosomal DNA as a template. The amplified DNA fragments were digested with XbaI and KpnI and subcloned between the corresponding sites of the pKT25 and pUT18C vectors. The resulting recombinant plasmids expressed hybrid proteins in which the polypeptides of interest were fused to the C-termini of the T25 and T18 fragments of adenylate cyclase respectively. For bacterial two-hybrid assays, recombinant pKT25 and pUT18C carrying the parA and parB genes or the empty vector plasmids were used in various combinations to co-transform BTH101 cells. The transformants were plated onto MacConkey/maltose medium and incubated at 30°C for 36–48 h. Interactions between different hybrid proteins were confirmed by measurement of β-galactosidase activity in liquid cultures.
ATPase activity test
ATPase activity was tested according to Majka et al. (1997). Briefly, analysed proteins (5 pmols) were incubated with 7 nM [γ-32P]-ATP at 30°C. Samples were collected at the time points indicated and the nucleotides were resolved by TLC. The radioactive signals were quantified with ImageQuant.
We are grateful to Gilles van Wezel for valuable comments on the manuscript, Robert Drynda for assistance in the laboratory and Klas Flardh for providing strain K202. This work was supported by a Marie Curie Fellowship of the European Community programme Human Potential under contract No. HPMF-CT-2002-01676, and Marie Curie Reintegration Grant MERG-6-CT-2005-014851 and partially by the Ministry of Science and Higher Education (Grant 2P04A 054 29). K.F.C. was supported by an Emeritus Fellowship from the John Innes Foundation. We are grateful to the Foundation for Polish Science (the Novum Programme) for financial support for purchase of a fluorescence microscope.