Identification of C-terminal hydrophobic residues important for dimerization and all known functions of ParB of Pseudomonas aeruginosa

The ParB protein of Pseudomonas aeruginosa is important for growth, cell division, nucleoid segregation and different types of motility. To further understand its function we have demonstrated a vital role of the hydrophobic residues in the C terminus of ParBP.a.. By in silico modelling of the C-terminal domain (amino acids 242–290) the hydrophobic residues L282, V285 and I289 (but not L286) are engaged in leucine-zipper-like structure formation, whereas the charged residues R290 and Q266 are implicated in forming a salt bridge involved in protein stabilization. Five parB mutant alleles were constructed and their functionality was defined in vivo and in vitro. In agreement with model predictions, the substitution L286A had no effect on mutant protein activities. Two ParBs with single substitutions L282A or V285A and deletions of two or seven C-terminal amino acids were impaired in both dimerization and DNA binding and were not able to silence genes adjacent to parS, suggesting that dimerization through the C terminus is a prerequisite for spreading on DNA. The defect in dimerization also correlated with loss of ability to interact with partner protein ParA. Reverse genetics demonstrated that a parB mutant producing ParB lacking the two C-terminal amino acids as well as mutants producing ParB with single substitution L282A or V285A had defects similar to those of a parB null mutant. Thus so far all the properties of ParB seem to depend on dimerization.


INTRODUCTION
The essential role of par loci in plasmid partitioning has long been appreciated (Williams & Thomas, 1992), while the function of chromosomally encoded par loci in the segregation of bacterial chromosomes is less clear. The chromosomally encoded par loci are highly conserved and belong to type I partitioning systems (Gerdes et al., 2000). Besides the high level of identity in amino acid sequences of the chromosomal homologues of ParA and ParB, the parS sequences are also extremely well conserved at least between so called primary chromosomes. The localization of the parAB genes, as well as the majority of parS sites, in close vicinity to the oriC region could indicate a role of par systems in replication/segregation of chromosomes, as genes known to be crucial for these processes are situated within the oriC domain (20 % of the chromosome around oriC). Moreover, the chromosomal par systems are able to promote active segregation and stabilization of otherwise unstable replicons even in heterologous host cells (Bartosik et al., 2004;Godfrin-Estevenon et al., 2002;Lin & Grossman, 1998;Yamaichi & Niki, 2000). Although these features of chromosomally encoded par loci suggest that they should play a similar biological function in chromosome segregation to that which plasmid par systems play for plasmid DNA, studies on par mutants in different bacteria have revealed a more complex picture. With the exception of Caulobacter crescentus, the chromosomal parA and parB genes are not essential for cell viability (Mohl & Gober, 1997). However, mutations in the parA (soj) and parB (spo0J) genes lead to defects in the sporulation of Bacillus subtilis (Cervin et al., 1998;Quisel et al., 1999;Quisel & Grossman, 2000) and Streptomyces coelicolor (Jakimowicz et al., 2002(Jakimowicz et al., , 2006(Jakimowicz et al., , 2007Kim et al., 2000) and in vegetative chromosome partitioning of B. subtilis (Ireton et al., 1994), Pseudomonas aeruginosa (Lasocki et al., 2007;Bartosik et al., 2009) and Pseudomonas putida (Lewis et al., 2002;Godfrin-Estevenon et al., 2002). Several studies have indicated a role for par genes in origin localization and segregation (Bowman et al., 2008;Ebersbach et al., 2008;Figge et al., 2003;Fogel & Waldor, 2006;Glaser et al., 1997;Jakimowicz et al., 2007;Lee et al., 2003;Ptacin et al., 2010;Saint-Dic et al., 2006;Sharpe & Errington, 1996;Toro et al., 2008;Viollier et al., 2004), in the separation of sister origins and in the regulation of replication (Kadoya et al., 2011;Lee & Grossman, 2006;Murray & Errington, 2008;Ogura et al., 2003;Scholefield et al., 2011;Webb et al., 1997), structural maintenance of chromosomes (SMC), complex recruitment to the nucleoid (Gruber & Errington, 2009;Sullivan et al., 2009) or in cell division initiation and cell cycle coordination (Autret & Errington, 2003;Figge et al., 2003;Jakimowicz et al., 2007;Marston & Errington, 1999;Mohl et al., 2001;Real et al., 2005;Schofield et al., 2010;Thanbichler & Shapiro, 2006).
In Pseudomonas aeruginosa, a facultative pathogen, the par locus has been identified~8 kb from oriC; eight out of ten parS sequences are located in the ori domain. The ParABS system of P. aeruginosa can stabilize an unstable plasmid in Escherichia coli (Bartosik et al., 2004). Neither ParA nor ParB of P. aeruginosa is essential for the viability of the cells. Their lack causes visible phenotypic defects (more severe in the parA mutant) which include an over 200-fold increase in the number of anucleate cells, longer cells, slower growth rate and perturbations in colony formation and motilities (Bartosik et al., 2009;Lasocki et al., 2007). The increased frequency of chromosome loss observed in actively dividing cells of a P. aeruginosa parB mutant (Bartosik et al., 2009) has been also reported for P. putida parB mutants (Godfrin-Estevenon et al., 2002;Lewis et al., 2002), but only in the transient and stationary phase of culture growth. The other phenotypic defects (motility defects, changed colony morphology, increase in cell size) caused by the lack of ParB P.a. seem to be unique among ParB representatives and so far unexplained.
In terms of molecular functions, previous studies (Bartosik et al., 2004) have shown that ParB of P. aeruginosa conforms to the behaviour of other chromosome-and plasmid-encoded homologues. It demonstrates the ability to interact with ParA and to dimerize and to bind centromere-like sequences (parS). ParB P.a. has the ability to spread on DNA and silence genes adjacent to the parS sites and it has been shown that this effect is dependent on a putative helix-turn-helix (H-T-H) motif, the ability of ParB to dimerize and also on an intact N terminus (Bartosik et al., 2004;Kusiak et al., 2011). ParB also forms regularly distributed foci in P. aeruginosa cells which colocalize with the nucleoid and undergo dynamic changes (Bartosik et al., 2009). In this study, we dissected the C terminus of ParB from P. aeruginosa demonstrating the vital role of the hydrophobic residues and C-tip of the protein in dimerization and all known functions of this protein.
Bacteria were grown in Luria broth (Kahn et al., 1979) at 37 or 30 uC or on Luria agar (Luria broth with 1.5 % w/v agar) supplemented with antibiotics as appropriate: benzylpenicillin sodium salt at 150 mg ml 21 in liquid medium and 300 mg ml 21 on agar plates for penicillin resistance in E. coli, kanamycin sulphate at 50 mg ml 21 for kanamycin resistance in E. coli, carbenicillin at 300 mg ml 21 for carbenicillin resistance in P. aeruginosa, rifampicin at 300 mg ml 21 for rifampicin resistance in P. aeruginosa. Some experiments were performed in M9 minimal medium (Sambrook et al., 1989) and on MacConkey Agar Base (Difco) supplemented with 1 % maltose. The Luria agar used for blue/white screening contained 0.1 mM IPTG and 40 mg X-Gal ml 21 .
Plasmid DNA isolation, analysis, cloning and manipulation of DNA. Plasmid DNA was isolated and analysed by standard procedures (Sambrook et al., 1989). The plasmids used in this study are listed in Table 1. Standard PCR (Mullis et al., 1986) was performed as described previously (Lasocki et al., 2007) with the primers listed in Table S1 (available with the online version of this paper). PCR sitedirected mutagenesis (modified method of Stratagene, QuikChange Site-Directed Mutagenesis) was used to make parB alleles producing ParBs with single amino acid substitutions: L282A, V285A and L286A. In this procedure, supercoiled double stranded plasmid DNA with parB P.a. (either pKLB2 or pKLB2.8) and pairs of synthetic oligonucleotide primers #21 and #22, #23 and #24 and #25 and #26 containing the desired mutation (Table S1) were used. The mutagenic oligonucleotide primers were designed to either introduce or remove the restriction site. The plasmid DNA was sequenced to verify the presence of the mutation.
Bacterial transformation. Competent cells of E. coli were prepared by using a standard CaCl 2 method (Sambrook et al., 1989). Transformation of P. aeruginosa strains was done according to a previously published method (Irani & Rowe, 1997).
Introduction of par mutant alleles into P. aeruginosa PAO1161.
Mutant parB alleles were cut out as EcoRI-SalI fragments from pMMB5.2, pMMB6.2, pJMB100 and pJMB101.1 and inserted into pAKE600, suicide vector unable to replicate in P. aeruginosa strains (El-Sayed et al., 2001), to create pJMB400, pJMB401, pJMB404 and pJMB405, respectively. To provide the region of homology downstream of truncated parBs the 389 bp DNA fragment corresponding to genomic DNA adjacent to the 39 end of parB was amplified by PCR with the pair of primers # 11 and # 12 (Table S1) and then introduced as a SalI-BamHI fragment into pJMB400, pJMB401, pJMB404 and pJMB405 to form pJMB402, pJMB403, pJMB406 and pJMB407, respectively.
Transformants of E. coli strain S17-1 with pAKE600 derivatives were used as the donors in conjugation with the recipient strain P. aeruginosa PAO1161 Rif R . The transconjugants with pAKE600 derivatives integrated into the chromosome were treated as described previously (Lasocki et al., 2007). The PCR products (chromosomal DNA isolated from the putative mutants as the templates with primers # 1 and # 12) were screened by SalI digestion and finally the presence of modifications was confirmed by sequencing of PCR fragments.
'Silencing' assay. E. coli DH5a(pABB811parS 2/3 ) and DH5a(pGB2) cells were transformed with the appropriate pGBT30tacp-parB derivatives. Undiluted and 10-and 100-fold dilutions of the initial transformation mixture were plated to select for incoming plasmid (Luria agar supplemented with penicillin) or for both incoming and resident plasmid (Luria agar with penicillin and streptomycin) with and without 0.5 mM IPTG to induce ParB production. After 24 h incubation at 37 uC the colonies were counted and the number of different class transformants in the original transformation mixture was estimated.
Analysis of protein-DNA interactions by electrophoretic mobility shift assay (EMSA). To check the ability of mutated ParB P.a. proteins to bind parS DNA in vitro, a nonradioactive EMSA (Leonard et al., 2004) was performed. A 16 bp dsDNA fragment (annealed oligonucleotides, # 17 and # 18, Table S1) containing the parS 2/3 (Bartosik et al., 2004) was used in EMSA. Samples (5.6 pmol) of parS 2/3 oligonucleotides without or with increased concentration of purified His 6 -tagged ParB and its derivatives were incubated under conditions described previously (Kusiak et al., 2011). Negative control of the binding reaction was provided by use of unrelated dsDNA with palindromic sequence (annealed primers # 19 and # 20, Table S1) and the same amounts of ParB in the incubation mixture. The samples were analysed on 10 % (w/v) non-denaturing polyacrylamide gel in TBE buffer (Sambrook et al., 1989). DNA bands were stained with 0.5 mg ethidium bromide ml 21 and visualized on a UV transilluminator.
Growth experiments and sample preparation for Western blotting. The growth of bacteria was monitored by measuring OD 600 ; the cultures were diluted and plated on Luria agar to establish c.f.u. ml 21 . Bacteria were harvested, resuspended in sonication buffer (50 mM phosphate buffer, pH 8.0, and 300 mM NaCl) and disrupted by sonication. Crude extracts from the same number of cells were analysed by SDS-PAGE followed by Western blotting performed as described previously (Bartosik et al., 2004).
Bacterial adenylate cyclase two-hybrid system (BACTH system). The interactions between ParB mutant derivatives and either wild-type ParB or ParA were analysed by using the bacterial twohybrid system BACTH (Karimova et al., 1998). The C-terminal parB mutant alleles have been cloned as EcoRI-HincII fragments from pET28mod derivatives into pLKB4 (derivative of pUT18C) to create translational fusions with the T18 catalytic domain of Bordetella pertussis adenylate cyclase, CyaT18-ParB. The wild-type parA and wild-type parB alleles have been cloned into pLKB2 (modified pKT25) to produce translational fusions: CyaT25-ParA and CyaAT25-ParB. E. coli BTH101 cya strain was co-transformed with both pLKB4 and pLKB2 derivatives and plated on indicator MacConkey base medium supplemented with 1 % maltose (as the only carbon source), penicillin, kanamycin and 0.5 mM IPTG. The plates were incubated for 48 h at 27 uC.
Motility assays. The swimming, swarming and twitching assays were performed according to the method of Rashid & Kornberg (2000) with modifications described previously (Lasocki et al., 2007). All sets of plates were standardized by using the same volume of medium.
DAPI staining and immunofluorescence microscopy. The DAPI staining procedure and immunofluorescence microscopy were carried out as previously described (Bartosik et al., 2004;Bignell et al., 1999). Cells were examined using an Eclipse E800 light fluorescence microscope (Nikon) fitted with an ORCA ER CCD camera (Hamamatsu). Images were captured and manipulated on PC Windows XP Professional PL with the Lucia General 5.0 (Laboratory Imaging).
The 3D structure of ParB P.a. has not yet been solved. The sequence of the C terminus of ParB P.a. aligns well with the C-terminal part of the ParB homologue -KorB of RK2 (IncP-1a) (Fig. S1) and moreover the two domains are functionally interchangeable. Replacement of 61 amino acids from the C terminus of ParB by the C-terminal 100 amino acids of KorB (ParB1-229-KorB258-358) restores its ability to dimerize and bind parS with high affinity in vitro as well as to transcriptionally silence genes near a parS site in vivo (Bartosik et al., 2004). Therefore, on the basis of crystallographic analysis of the C-terminal part of KorB of RK2/RP4 (Delbrück et al., 2002) a model of the C-terminal domain of ParB P.a. was built in silico ( Fig. 1). According to this model the hydrophobic residues L282, V285 and I289 (but not L286) are engaged in a leucine-zipper-like structure, whereas the charged R290 and Q266 are implicated in forming a salt bridge involved in stabilization of the ParB dimer. To verify this model, two alleles of parB with Cterminal deletions of either seven (parB1-283) or two Spreading on DNA in vivo -'silencing test' in E. coli The ParB P.a. protein recognizes the parS sequence as a dimer then self-associates and spreads on DNA causing transcriptional silencing of genes adjacent to parS (Bartosik et al., 2004). The plasmid pGB2 (Churchward et al., 1984) used for the 'silencing test' is an Sm R stable replicon based on pSC101 in which a multiple cloning site (MCS) is inserted approximately 200 bp upstream of the promoter for the initiator gene repA. The presence of parS close to the repA promoter in pABB811 does not influence plasmid stability unless wild-type ParB is produced in trans from pKLB2 (tacp-parB) (Bartosik et al., 2004). The transformation frequency of E. coli DH5a(pABB811) with pKLB2 in the absence of IPTG with selection for incoming and resident plasmid was two-to threefold lower than the transformation frequency when only the incoming plasmid is selected. Addition of 0.5 mM IPTG to transformation plates with double selection (conditions of ParB overproduction) decreases the number of transformants more than 100-fold (Table 2) in comparison with the number of transformants grown on double selection plates without IPTG.  Fig. 1. Model of a dimer of C termini of ParB P.a. (amino acids 242-290). The mutagenized residues are shown as sticks in the red subunit (labelled according to their position in the ParB P.a. sequence). The indicated residues L282 and V285 in the grey subunit are possibly involved in a leucine zipper formation. The distance between R290 of one monomer and Q266 of another facilitates the electrostatic interactions (magnification at the left).
The silencing test was repeated to establish the effect of overproducing the modified ParB proteins on the stability of pABB811 in E. coli DH5a(pABB811). The numbers of transformants with selection for either incoming plasmid (Pn) or both incoming and the resident plasmids (Pn Sm) with and without IPTG present are shown in Table 2. Only ParBL286A (pJMB602) caused significant instability of pABB811 and a loss of streptomycin resistance of the recipient strain when ParB was overproduced during growth with IPTG (more than 100-fold decrease in the number of transformants on Pn Sm IPTG plates, effect observed for wild-type ParB delivered from pKLB2). The other plasmids tested had very little impact on stability of pABB811 (two-to threefold decrease in the number of double transformants grown in the presence of inducer in comparison with the number of transformants selected for incoming plasmid). The deletions of seven (pJMB604) or two amino acids I289 and R290 from the C terminus (pJMB603) as well as the single amino acid substitutions V285A or L282A impaired the silencing property of ParB. Western blotting on extracts from analysed transformants has shown a level of ParB overproduction for all mutant derivatives similar to wild-type ParB (Fig. S3).

ParB P.a. dimerization in vitro
For the in vitro analysis, wild-type ParB as well as modified ParB proteins (ParB1-283, ParB1-288, ParBL282A, ParBV285A and ParBL286A) with a His 6 tag attached to the N terminus were expressed upon induction with IPTG from pET28mod derivatives in E. coli strain BL21(DE3). All ParB variants were present in the soluble cellular fraction and it was possible to purify them in a native form on a Ni 2+ -agarose column. Wild-type ParB P.a. protein has previously been found to dimerize and form higher order complexes in vitro (Bartosik et al., 2004). The dimerization and oligomerization domains are separate in ParB P.a. (Kusiak et al., 2011). The purified His 6 -tagged ParB derivatives were treated with increasing concentrations of the cross-linking agent glutaraldehyde (GA). Wild-type ParB protein dimerized so strongly that even at the lowest glutaraldehyde concentration (0.001 %) dimeric species were visible. The monomeric and dimeric forms were predominant after SDS-PAGE separation and Coomassie blue staining (not shown). To visualize higher-order complexes, a Western blot analysis was performed with anti-ParB antibodies (Fig. 2a). ParBL286A dimerized and formed higher order complexes as efficiently as the wild-type protein. ParBL282A and ParBV285A dimerized but with much lower effectiveness than wild-type ParB. The C-truncated ParBs -ParB1-283 and ParB1-288 -were drastically impaired in dimerization. These results confirmed that the C-terminal fragment of ParB is the major determinant of its ability to form dimers, as deletion of seven (ParB1-283) or even two (ParB1-288) amino acids impaired significantly the ability of monomers to interact.
ParB P.a. DNA binding in vitro (EMSA) A previous study (Bartosik et al., 2004) indicated that self association of ParB is important for efficient DNA binding (deletion of the C terminus in ParB1-229 significantly decreased the DNA binding ability but did not completely stop it from binding parS). All purified ParB P.a. derivatives were tested for binding to parS P.a. using a standard EMSA (Fig. 2b). Non-radioactive EMSA was performed on parS (annealed oligonucleotides # 17 and # 18) and an unrelated palindrome motif (annealed oligonucleotides #19 and # 20) as a control. Wild-type ParB did not bind the control oligonucleotides at tested concentrations. ParBL286A showed affinity towards parS approximately twofold higher than wild-type ParB whereas ParBV285A and ParBL282A bound parS but with twofold lower affinity. ParB1-288 and ParB1-283 hardly shifted the double-stranded parS oligonucleotides at tested concentrations. Therefore the ability to bind parS seems to correlate with the degree of dimerization proficiency as illustrated by comparing Fig. 2(a) and (b). ParB P.a. self association and interaction with ParA P.a. in vivo To check the interactions of mutated ParB proteins with wild-type ParB and ParA in vivo, the bacterial adenylate cyclase two-hybrid system in E. coli (Karimova et al., 1998(Karimova et al., , 2000 was applied. Mutated ParB derivatives were translationally fused to CyaT18 fragment (pUT18C derivatives), whereas wild-type ParB and ParA were fused to CyaT25 fragment (pKT25 derivatives). E. coli BTH101, an adenylatecyclase-deficient strain (cya), was co-transformed with a mixture of appropriate pairs of BACTH system plasmids and plated on MacConkey base medium supplemented with 1 % (w/v) maltose, 0.5 mM IPTG and selective antibiotics.
The results of in vivo BACTH analysis confirmed the conclusions from the in vitro dimerization studies presented above. The two short deletion mutants ParB1-283 and ParB1-288 were unable to associate with wild-type ParB whereas interactions between ParBL286A and wild-type ParB were similar to self-association of wild-type ParB as demonstrated by BTH101 (pLKB702)(pLKB233) transformants (Fig. 3). Interactions of ParBL282A and ParBV285A with wild-type ParB were weaker than control interactions between pLKB433 and pLKB233 but still very clear. The analysis of interactions of mutant ParB derivatives and ParA showed a correlation between the efficiency of ParB dimerization and the ability to interact with ParA in the BACTH system. ParBL286A demonstrated interactions with ParBV285A ParBL282A ParB (  ParA similar to those for wild-type ParB. Visibly weaker interactions between ParA and ParBL282A or ParBV285A and no interactions between ParA and ParB1-283 or ParB1-288 were detected. This strongly suggests that the dimer form of ParB is required for interactions with ParA.

Introduction of mutant parB alleles into the P. aeruginosa chromosome
The C-terminal modifications of ParB (ParB1-283 and ParB1-288 as well as ParBL282A and ParBV285A) cause defects in dimerization and in turn defects in DNA binding, transcriptional silencing and interaction with ParA. To determine the phenotypic effect of these mutations all four alleles were introduced into the P. aeruginosa chromosome using suicide vector pAKE600 and allele exchange via homologous recombination (El-Sayed et al., 2001). The new P. aeruginosa PAO1161 Rif R parB mutants, parB1-288 and parB1-283, parB282 and parB285 were tested for various forms of motility: swimming, swarming and twitching. A drastic defect in swarming, slight defect in swimming and no effect on twitching have been observed previously for the P. aeruginosa PAO1161 Rif R parB null strain (Bartosik et al., 2009). Neither of the new tested mutants was disturbed in twitching (data not shown) but all four were strongly impaired in swarming and slightly affected in swimming (Fig. 4). The parB282 and parB285 mutants demonstrated lesser defects in swimming when compared with parB null, parB1-288 and parB1-283 mutants.
To look at the effects of parB mutations on P. aeruginosa PAO1161 Rif R growth, time-course experiments in both rich (Luria broth) and minimal (M9) medium were conducted. Each growth experiment was performed using cells freshly taken from a deep-frozen stock to reduce the possibility of accumulation of secondary mutations. The new PAO1161 Rif R parB mutants (short deletions and single amino acid substitutions) demonstrated changes in the growth rate similar to the parB null strain. They showed~10 % longer mean generation time (mgt) in comparison with wild-type PAO1161Rif R when grown in Luria broth or minimal medium (M9) at 37 u C, and~20 % longer mgt when grown in Luria broth at 30 u C (Table 3).
The parB mutant strains were also examined for the frequency of anucleate cell formation. Bacterial cells were collected from cultures at late exponential growth phase  Fig. 4. Motility assays for P. aeruginosa PAO1161 Rif R , parB mutants and merodiploid strains. Volume-standardized plates for swarming and swimming were inoculated with a sterile toothpick using material from a single colony and incubated for 24 h at 30 6C. The zones of growth/spreading are indicated in mm below the images; the boundaries of the swimming zones are marked by arrows.
(OD 600 0.8). The cells were fixed and DAPI-stained to visualize chromosomes. The number of cells without chromosomes and the mean cell length were estimated using fluorescence microscopy combined with appropriate software. The frequency of anucleate cell formation for the parB null mutant was more than 100-fold higher than for the wild-type strain under the same growth conditions on the sample of at least 1000 cells (Table 3). The short C-terminal deletion mutants PAO1161 Rif R parB1-288, parB1-283 as well as the substitution mutants parB282 and parB285 produced anucleate cells at similar frequencies to those observed for the PAO1161 Rif R parB null mutant. Measurements of cell length showed that all parB mutants produce cells up to 10 % longer on average than those of the wild-type similarly to the PAO1161 Rif R parB null mutant (Table 3).
In order to check the intracellular concentration of the mutated ParBs, equal numbers of cells of the PAO1161 Rif R strain and parB mutants from the same growth phases were collected and analysed by Western blotting with anti-ParB antibodies. The amount of ParBs truncated at the C terminus was approximately five-to sixfold lower than the amount of wild-type ParB in actively dividing cells of the PAO1161 Rif R strain. A similar decrease was observed for two ParBs with amino acid substitutions at the C terminus (ParBL282A and ParBV285A) probably due to the lower stability of monomeric ParB (Fig. 5a). To exclude the possibility that it was the decreased cellular concentration of ParB rather than the specific mutation that was responsible for the observed defects in growth and nucleoid segregation, the medium-copy-number broad-host-range plasmid pBBR1-MCS1 carrying the mutated parB alleles under control of tacp (the plasmid series from pJMB501 to pJMB504) was introduced into the appropriate chromosomal mutants. Western blotting of extracts from defined numbers of cells of such transformants grown in the absence of IPTG showed that the level of mutant ParBs was similar or even higher when related to wild-type ParB in PAO1161 grown under the same conditions (Fig. 5a). The merodiploid strains were also tested for motility (Fig. 4), growth rate and anucleate cell production (Table 3). No suppression of the defects was observed by increasing production of mutant ParB derivatives, confirming that specific changes in ParB and not a decreased level were responsible for the mutant phenotypes.
To visualize the localization and ability of the ParB mutant derivatives to form intracellular foci, immunofluorescence microscopy was applied (Fig. 5b). Fixed cells from the exponential growth phase were incubated first with purified anti-ParB antibodies and then with FITC-conjugated antirabbit IgG. Cells of the PAO1161 Rif R parB null mutant were also examined as a control for the specificity of the antibodies used. The majority of the actively dividing cells of the wild-type strain of P. aeruginosa contained from two to four regularly spaced ParB foci as expected from the number of ori domains. In cells of mutants parB1-283, parB1-288, parB282 and parB285 no such strictly organized foci were observed, but instead, multiple irregularly distributed signals appeared in the region of the nucleoid. The fluorescence signals in transformants of the PAO1161 Rif R parB1-283 and PAO1161 Rif R parB1-288, in which truncated ParBs were also supplied from the plasmids, showed multiple and dispersed fluorescent foci, similar to those seen in the mutants (data not shown). The diminished ability of ParB derivatives to dimerize leads to the defect in foci compaction (probably oriC domain organization) and nucleoid segregation.

DISCUSSION
The work described in this paper adds important details to our understanding of ParB from P. aeruginosa, which is a key representative of the large family of ParB proteins encoded by both plasmids and chromosomes. Our earlier in vivo studies revealed that overproduced ParB of P. aeruginosa is able to silence the expression of genes adjacent to the parS site (Bartosik et al., 2004) and that spreading activity relies on dimer formation by the C terminus, DNA binding and N-terminal polymerization domain (Bartosik et al., 2004;Kusiak et al., 2011). Accumulating evidence has shown that this activity is a common feature of ParB family members of type IA (Bingle et al., 2005;Rodionov et al., 1999;Schumacher et al., 2007Schumacher et al., , 2010, also including the chromosomal homologues (Bartosik et al., 2004;Breier & Grossman, 2007). This silencing is thought to be a consequence of spreading on DNA due to ParB-ParB interactions through the N-terminal polymerization domain (Kusiak et al., 2011). However, the physiological role of both plasmidic (Rodionov & Yarmolinsky, 2004) and chromosomal ParBs spreading on DNA is unclear. The recent studies on Spo0J of B. subtilis showed that Spo0J spreads around each parS site on chromosomal DNA over dozens of kilobases (Breier & Grossman, 2007) but under the conditions tested, this process did not significantly affect expression of the majority of genes near parS, with the exception of some sporulation genes. The crystallographic studies on ParB homologues of type IA (Delbrück et al., 2002; Khare et al., 2004;Leonard et al., 2004;Schumacher et al., 2007) combined with further experimental verification should help to elucidate the exact role of ParB spreading on DNA and whether this role is universal for all ParB homologues. The discoveries of interactions of chromosomal ParB homologues with DnaA (control of initiation of replication), different proteins involved in chromosome organization and ori domain localization (SMC, PopZ, TipN) and cytokinesis (FtsZ, MipZ) suggest an important biological role of Par proteins in a wide spectrum of processes, some of them possibly species-specific (Bowman et al., 2008;Donovan et al., 2010;Ebersbach et al., 2008;Gruber & Errington, 2009;Kadoya et al., 2011;Murray & Errington, 2008;Ptacin et al., 2010;Schofield et al., 2010;Scholefield et al., 2011;Sullivan et al., 2009;Thanbichler & Shapiro, 2006;Toro et al., 2008). In P. aeruginosa ParB seems to be involved not only in the chromosome segregation but also in the control of growth rate, cell motilities and colony morphology (Bartosik et al., 2009). Its role in some but not all of these processes depends on interactions with its cognate ParA counterpart (M. Kusiak (Fig. 1). We constructed five parB mutant alleles to define the functionality of proteins modified in the C-terminal region. Both in vitro and in vivo tests on the ability of ParB derivatives to dimerize indicated that the last two amino acids at the C terminus within the conserved region 4 are essential for the ability of ParB P.a. to self-associate. Removal of I289 and R290 rendered ParB inactive in dimer formation. In agreement with the structural prediction in silico two hydrophobic residues L282 and V285 have been confirmed to play a vital role in ParB dimerization. The alanine substitution derivatives ParBL282A and ParBV285A showed detectable changes in self-association in vitro and in association with wild-type ParB in vivo. On the other hand, L286, which should be directed outwards from the putative dimer ( Fig.  1), has been confirmed experimentally not to be involved in self-associations. The alanine substitution derivative ParBL286A behaved like wild-type ParB in all tests with the exception of EMSA when it seemed to bind parS with even higher affinity than wild-type ParB.
The four ParB mutant derivatives impaired to various extents in dimerization were also impaired to similar extents in parS binding, strongly implying that ParB binds to parS as a dimer. None of these mutants was also able to silence genes adjacent to parS, suggesting that dimerization through the C terminus is a prerequisite for spreading on DNA. The necessity of ParB to form dimers before interacting with its ParA partner was confirmed by analysis of mutants in the BACTH system. The observed in vivo heterologous interactions between ParB mutant derivatives and ParA correlated in strength with the ability of ParB mutants to interact with wild-type ParB as a dimer.
When these four parB alleles were introduced into the P. aeruginosa chromosome by allele exchange they caused defects in growth rate and motilities (swarming and swimming) and more than 100-fold increase in the frequency of anucleate cell formation. Immunofluorescence microscopy showed that in contrast with wild-type ParB, which is organized into one-four regularly distributed foci, the modified ParBs formed multiple smaller foci dispersed within the boundaries of the nucleoid.
It has been observed that all modified ParBs are present in lower quantities per cell and are more prone to degradation than wild-type ParB, probably due to their inability to be protected by ParA (Lasocki et al., 2007;Bartosik et al., 2009). The elevation of mutant ParB production to the level observed for wild-type ParB did not suppress the parB mutant phenotypes in the constructed merodiploid strains, suggesting that the decreased level of protein is not the main factor responsible for the visible deficiencies of the mutants.
Despite the fact that ParBs with single amino acid substitution (ParB282 and ParB285) seem to be significantly less impaired in dimerization, DNA binding or interactions with ParA than the truncated derivatives ParB1-283 and ParB1-288, the phenotypes of four new parB mutants were almost identical (with slight difference between the deletion and point mutants in swimming defects) and they resembled the phenotype of the parB null mutant (Bartosik et al., 2009). The data presented suggest that even small changes in the dimerization ability of ParB may translate into lower affinity of parS binding and in turn result in inability to spread on DNA (silencing test). The spreading on DNA has been shown to determine the biological function of ParB in P. aeruginosa (Kusiak et al., 2011).
In conclusion, an in silico model of the ParB P.a. C-terminal dimerization domain has identified the hydrophobic residues L282 and V285 and charged residue R290 as vital for dimerization. Substitution of hydrophobic residues by alanine or removal of the two last amino acids I289 and R290 impairs ParB P.a. in dimerization, parS binding and ParA interaction and renders it inactive in spreading on DNA (transcriptional silencing). Since such truncation of ParB as well as alanine substitution of two hydrophobic residues led to the same deficiencies in growth, genome segregation and motilities as a complete lack of ParB in P. aeruginosa, it is clear that dimerization is a vital prerequisite for the function of ParB in the cells.