Diverse Partners of the Partitioning ParB Protein in Pseudomonas aeruginosa

Chromosomally encoded ParB proteins are composed of three domains: the N-terminal domain (NTD) involved in CTP binding and hydrolysis, the central DNA-binding domain, and the C-terminal part mediating dimer formation (7, 40 to 44). Previous experiments with the use of the yeast two-hybrid system suggested involvement of the Pae ParB C-terminal domain for the interactions with ParA (45), which was in contrast to what was acknowledged for other bacterial species. To clarify this discrepancy, here we have systematically analyzed ParA interactions with various truncated variants of ParB using the bacterial two-hybrid (BACTH) system, based on reconstitution of adenylate cyclase activity by interacting proteins fused with CyaT18 and CyaT25 subunits (Fig. 1A). The analysis showed that ParA interacts with the NTD of ParB but not with the remaining two domains (Fig. 1A). Removal of 37 or even 6 amino acids from the N terminus of ParB abrogated the interactions, and, concomitantly, the ParB_1-36 fragment interacted with ParA (Fig. 1A). This suggests that the N-terminal part of ParB is sufficient for interactions with ParA. The interactions of the dimerization domain with ParA reported previously with the use of the yeast two-hybrid system might represent a false-positive observation, possibly due to auto-activation of reporter genes, triggered by the C terminus of ParB (46).

The sequence alignment of bacterial ParBs provided a logo illuminating amino acid conservation (Fig. 1B; full logo in Fig. S1 in the supplemental material). The N terminus of ParB contains a conserved stretch of 12 amino acids with several positively charged and aliphatic amino acids. Alanine substitution of 5 highly conserved amino acids (gray in Fig. 1B) showed that altering L8, R10, G11, and L12 but not G9 blocked the ParB interactions with ParA (Fig. 1C). The sensitivity and indirect relation between reporter activities and the extent of protein–protein interactions in BACTH could mask the residual interactions between ParA and ParB variants; nevertheless, the results clearly indicate the importance of residues at the N terminus of Pae ParB for interactions with ParA.

To check the effect of these amino acid substitutions on chromosome segregation, we have engineered Pae PAO1161 derivatives with missense parB mutations resulting in production of ParB variants L8A, R10A, G11A, or L12A. Analysis of the culture growth in the rich medium showed lower growth rates of strains producing these ParB variants, similar to that observed for strains lacking parB or expressing ParA L84K protein, defective in interactions with ParB (36) (Fig. 1D). Interestingly, the point mutants differed in the extent of the growth retardation, and the most severe growth defect was observed for the strain producing ParB G11A (Fig. 1D). The effect seemed not to be related to the level of ParB, as it was only slightly reduced in mutants expressing ParB L8A, R10A, G11A, or L12A, compared to native ParB in PAO1161 in contrast to cells lacking ParA (Fig. S2) (33). Concomitantly, analysis of the frequency of anucleate cells in the cultures exponentially growing in rich medium showed that, whereas less than 0.1% of cells lacking nucleoid were observed for the wild-type (WT) strain and around 3% of such cells were observed for the ΔparB strain, the cultures of strains producing ParA and ParB variants with disrupted interactions with their cognate partners contained significantly more (5 to 10%) anucleate cells (Fig. 1E), similar to what was observed for the ParA deficient strain (33).

To check whether the ParB–ParA interaction is required for ParB complex formation on parSs, in the case of Pae, the parS1-4 cluster adjacent to oriC, we have used fluorescence microscopy. In the PAO1161 strain grown in the rich medium, the majority of the cells contained up to 4 regularly spaced YFP-ParB foci (Fig. 2A, B). In the cells with two foci, they occupied positions at 20 and 80% of relative cell length (Fig. 2C, D), as reported previously (34, 46). ParB foci were still observed in strains producing interactions interactions-defective ParA L84K or ParB L8A, although they demonstrated variable intensity and irregular positioning (Fig. 2A, C, D). The number of foci detected per cell was lower, possibly due to the proximity of individual ParB-parS1-4 complexes triggered by a defect in their separation (Fig. 2B). Similar disturbances were observed when complex localization was analyzed in cells producing ParB R10A, G11A, or L12A (Fig. S3). This confirms that interactions between ParB and ParA are not essential for ParB complex formation on parSs but are required for proper movements of ParB-ori domain complexes. Strikingly, there was a much stronger phenotypic effect of disrupted ParA–ParB interaction in comparison to the lack of ParB alone. Indeed, the phenotype of the ΔparAB strain was similar to ΔparB, suggesting that ParB not interacting with its cognate ParA is detrimental to the cells (Fig. 2E). When the crucial parS sequences were mutated (ΔparS1-4 background), the presence of ParB L8A, R10A, G11A, or L12A did not further increase the anucleate cell content (Fig. 2E), confirming that impaired separation of ParB-parS1-4 complexes was responsible for the negative effect of ParB variants (compare to Fig. 1D).

In other organisms, ParB-parS complexes were shown to recruit SMC, and remarkably, an asymmetry of SMC recruitment was observed recently between the two Pae daughter chromosomes, suggesting presence of a mechanism controlling the process of its loading on newly replicated origins (47). The detrimental effect of ParA absence or disruption of ParB-ParA interactions was largely absent in the Δsmc mutant, implicating that SMC could enhance adhesion of unseparated ParB-parS complexes (Fig. 2E). Identical observation was made when cells were grown in the defined minimal medium, slowing down the growth rate (Fig. S4). The absence of SMC did not affect ParB level (Fig. S2) and significantly restored the separation of ori domains in cells producing ParB G11A variant as visualized by tagging the ori region with an orthogonal P1 ParB-parS system (Fig. 2F). This suggests that in the strains lacking SMC and ParA, ParB complexes could be segregated by other mechanism(s). Finally, to analyze the impact of ParB-ParA interactions and SMC on ParB binding to the chromosome, including the extent of ParB spreading around parSs, we performed chromatin immunoprecipitation and sequencing (ChIP-seq) in WT, parB_G11A, parA_L84K, and Δsmc strains. We calculated the ratio between the number of ChIP-seq reads in the region ±15 kb around the parS1-4 cluster versus the number of reads mapping to sites encompassing half-parSs, specifically bound by Pae ParB (39). The use of ParB binding to the half-parSs, at which ParB is thought not to spread from, constitutes an internal control for each ChIP sample. Large nucleoprotein complexes formed by ParB around oriC were observed in all analyzed mutants, in agreement with the fluorescence microscopy analyses. Concomitantly, there was no significant effect of disrupted ParB–ParA interactions (ParB G11A or ParA L84K) or SMC presence on the ratio, suggesting no major impact of these factors on the ability of ParB to remain bound to the DNA within the complex around parS1-4 (Fig. 2G). Overall, this suggests that ParB interactions with ParA mediated by a conserved motif at the ParB N terminus are required only for rapid partition of newly replicated ori regions, which prevents their cohesion by SMC (Fig. 2H).

The complex phenotype of cells lacking the partitioning proteins in P. aeruginosa suggested their regulatory character (33, 34, 37, 38) and prompted the analysis of the ParB protein interaction network beyond ParA. The candidate ParB-interacting proteins were identified using three screening methods: (i) coimmunoprecipitation (CoIP) with the use of antibodies targeting ParB followed by identification of associated proteins by mass spectrometry; (ii) using ParB as a bait in the bacterial two hybrid screening of the Pae genomic library; and (iii) analysis of ParB interactions with proteins showing sequence similarity to ParA (Fig. 3A).

The CoIP analysis performed with the use of anti-FLAG antibodies and extracts from PAO1161 cells producing FLAG-ParB, showed 250 candidates among which 40 proteins were found exclusively in one or more CoIP samples and not in controls (Table S1). These encompassed SMC (PA1527) and its partner ScpB (PA3197). The remaining 210 candidate proteins were also detected in the corresponding control samples, although with significantly lower scores. These were kept in the analysis as this group encompassed ParA and ParB itself (Table S1). Among them, multiple transcriptional regulators were found, proteins involved in the genome integrity and topology maintenance (e.g., UvrD, GyrA, GyrB, ParC, and ParE), nucleoid associated proteins (NAPs) (e.g., HupB, PA1533, and NdpA [PA3849]), as well as proteases ClpP, ClpP2, Lon, and various enzymes (Table S1).

Verification of the interactions between ParB and selected candidate partners, identified in the CoIP by BACTH, confirmed several enzymes as ParB partners (Fig. 3B). These included AruC (PA0895), a part of the catabolic arginine succinyltransferase pathway (48), as well as Pae CTP-synthase PA3637 (PyrG), and others (Table 1, Fig. 3B). While it is tempting to speculate that ParB complexes are involved in control of subcellular localization of metabolic activities, further studies with individual proteins are required to address this relation. An intriguing candidate partner of ParB, identified by CoIP and confirmed by BACTH (Fig. 3C), was NdpA/Yejk (PA3849) protein, representing a poorly characterized class of nucleoid associated proteins (49, 50). Interactions with ParB were also observed for the paralogous NpdA2 protein (Fig. S5A, Fig. 3C). NdpA2 is encoded within PAPI-1 family integrative conjugative elements present in genomes of strains like PA14 or PAO1161 but not in PAO1. Analysis of strains lacking ndpA, ndpA2, both genes, or the strain overproducing NdpA did not show defects in growth or elevated amount of anucleate cells, suggesting that under conditions tested, NdpA proteins may have a function unrelated to chromosome segregation (Fig. S5B, C).

A genome-wide BACTH-based screening was also performed with the use of the vector encoding CyaT25-ParB as a bait and the Pae genomic library cloned into the compatible plasmid, allowing fusion of random fragments to CyaT18 (51), which resulted in 37 candidate proteins (Table S2). These encompassed various N-terminal truncations of ParB protein, validating the approach, as the genomic library construction protocol allowed mostly C-terminal regions of prey proteins to be fused with CyaT18, and the C-terminal domain of ParB mediates protein dimerization (52). A large fraction (24%) of these were the known or putative transcriptional regulators. For most of the partners identified by BACTH screening, the interactions could not be confirmed when full-length candidate proteins were used in the BACTH assays (Table S2), with the exception of MexZ (PA2020), TetR family transcriptional regulator being the repressor of multidrug efflux pump gene mexXY in Pae (53) (Fig. 3B).

Three proteins were identified in both CoIP and BACTH screenings: molybdopterin biosynthetic protein B2 (MoaB2, PA3029), PqsR-mediated PQS regulator PmpR (PA0964), as well as PA4465, but only for the PA4465 or the last one the interactions between ParB and full-length protein were observed with BACTH (Fig. 3D). The sequence of the Pae PAO1 PA4465 protein shows 50% identity (99% coverage) to E. coli K-12 RNase adapter protein RapZ (UniProt, P0A894) and 36% identity (99% coverage) with B. subtilis YvcJ (UniProt, O06973), two proteins that were previously shown to display ATPase and GTPase activities (54 to 56). This suggested that a protein with an NTPase activity, other than ParA, may interact with ParB in Pae.

Finally, in the targeted approach, we used BACTH to probe ParB interactions with proteins showing sequence homology to ParA (Fig. 4A). These encompassed PA3244, encoding a homolog of E. coli MinD protein, a component of the Pae MinCDE system (57). The construction and analysis of Pae min mutants confirmed the predicted function of these genes since their deletions resulted in the extreme cell elongation and presence of minicells (Fig. 4B, C). The systematic analysis of interactions between the Min and Par proteins system revealed self-interactions of MinC, MinD, and MinE and the interactions between MinD-MinC and MinD-MinE, but no interactions between MinC and MinE, similarly as for E. coli homologs (58). Despite the ParA and MinD similarities, no interactions with any of the Par proteins were detected (Fig. 4D). Interestingly, MinE interacted with ParB, whereas MinC interacted with ParA (Fig. 4D), suggesting an intricate cross-talk between Par and Min systems in Pae (Fig. 4E).

Probing ParB interactions with the four other proteins showing sequence homology to ParA (Fig. 4A) revealed three other partners: PA3481, PA5028, and FleN (PA1454) (Fig. 4F). PA3481 is a Mrp/ApbC/NBP35 subfamily putative NTPase, with 48% amino acid sequence identity to E. coli Mrp, possibly involved in metabolism of Fe–S clusters. PA5028 is an uncharacterized orphan P-loop NTPase, and FleN (PA1454) is a regulator of the expression of flagellar genes in Pae (59, 60). Importantly, interactions between ParB and these three proteins were also confirmed using CoIP, when corresponding protein pairs were produced in a heterologous host E. coli (Fig. 4G). The fourth ParA homolog, PA1462, was identified as a candidate ParB partner in CoIP analysis (Table S1); however, in the BACTH analysis, the interactions between fusion proteins were not confirmed (Fig. 4F). Overall, these data indicate that at least four proteins with presumed NTPase activity, PA4465, PA3481, PA5028, and FleN, may interact with partitioning protein ParB from Pae.

The newly identified ParB partners, PA4465, PA3481, PA5028, and FleN, do not compete with ParA for ParB binding, as the analysis of their interaction with ParB 1 to 36 and 37 to 290 fragments showed that for all of them, the region of ParB involved in the interactions is different than for ParA (Fig. 5A). To check the significance of these proteins in Pae and their cooperation with ParB, we analyzed (i) the impact of lack or excess of these proteins on culture growth and chromosome segregation and (ii) the impact of a partner absence on the other protein cellular distribution. Despite using various approaches, a PAO1161 ΔPA3481 mutant could not be constructed. Previously, the genes encoding PA3481 orthologs (e.g., PA14_19065) were indeed listed as essential in transposon mutagenesis analysis (61).

The growth analysis of ΔPA5028, ΔfleN, or ΔPA4465 mutants showed no differences relative to the parental PAO1161 (WT) strain (Fig. S6A). The absence of these genes did not result in the appearance of anucleate cells in the mutant cultures (Fig. 5B) and did not alter anucleate cell content in populations of strains carrying additional deletion of parAB genes (ΔparAB, Fig. 5B). The condensin complex MksBEF may support chromosome segregation in the absence of a functional ParAB-parS system in the Pae (47, 62), hence it was hypothesized that minor defects of the ParAB-parS system (caused, e.g., by lack of partner proteins) could be masked by the action of MksBEF. However, deletion of PA4465, PA5028, fleN, or all three of them in PAO1161 ΔmksBEF also did not result in the appearance of anucleate cells in cultures (Fig. 5B).

The analysis of the growth of strains overproducing the proteins showed a negative effect of PA5028 and FleN excess on culture growth, whereas no impact was noted in the case of PA3481 and PA4465 (Fig. S6B). Interestingly, the excess of PA5028 frequently resulted in formation of cell chains, cells with guillotined nucleoids (Fig. 5C), and the increase in the number of anucleate cells (up to 0.5%). The absence of any of these proteins, however, did not affect the cellular distribution of ParB complexes as well as the numbers of detectable YFP-ParB foci (Fig. 5D, E). This suggests that analyzed proteins do not play a major role in Pae chromosome segregation under the conditions tested. Nevertheless, the observation that PA5028 abundance affects both cell morphology and DNA distribution suggests a role of this protein in the Pae cell cycle.

To address the impact of ParB on the cellular distribution of ParB partners, fluorescence microscopy analyses were performed. The ParA-CFP fluorescence signal displayed an asymmetric distribution, often showing comet-like profiles along the long axis of the cells in the parental (WT) strain (Fig. 6A). Quantitative analysis of the fluorescence distribution asymmetry, analyzed by calculation of the ratio between mean fluorescence signal in each half of the cell, confirmed asymmetric distribution of ParA-CFP as the signal was more unevenly distributed than in cells with a free fluorescent protein (sfGFP) or not expressing such a protein (control for auto-fluorescence, Fig. 6B). Concomitantly, lack of ParB led to a less smooth ParA-CFP signal, possibly due to ParA association with the nucleoid (Fig. 6A). Reduced asymmetry of ParA-CFP signals in the cells was observed in ΔparB as well as in the strain with mutated parS1-4 sequences (Fig. 6B). The strains with disrupted ParA–ParB interactions displayed various degrees of ParA-CFP asymmetry (Fig. 6B). These data confirm that ParB affects the cellular distribution of its cognate ATPase partner, ParA. A similar analysis with the other identified NTPase partners showed that FleN was equally distributed in the cells independent of the terminus to which fluorescent protein was attached (Fig. S6). The other three proteins showed discrete distribution in the cells. YFP-PA3481 could be observed in the majority of the cells as one or two foci at cell poles (Fig. 6C, D), and the number of foci increased with the cell length (Fig. 6E). Identical distribution and foci numbers were observed when the analysis was performed in ParB-deficient cells (Fig. 6F). Similar analysis of YFP-PA5028 showed that for this protein the signal is mostly diffused in the cells; however, also 1 to 4 discrete foci could be observed (Fig. 6G, H). These were mainly localized at cell poles, but foci at midcell were also observed (Fig. 6I). The foci number and positioning were independent of ParB (Fig. 6H, I). Finally, sfGFP-PA4465 foci were also observed in ~7% of cells, and they were located at cell poles (Fig. 6J, K, L), and such distribution was not dependent on ParB. Overall, this indicates that at least three putative NTPases interacting with Pae ParB may be located at the cell poles.

Strains, plasmids, and oligonucleotides used and constructed in this study are described in Tables S4 to S8. P. aeruginosa strains were grown at 37°C or 28°C, in L broth (LB; 1% bactotrypton, 0.5% yeast extract, 0.5% NaCl), supplemented with antibiotics when necessary: 75 μg mL⁻¹ chloramphenicol, carbenicillin at 300 μg mL⁻¹, rifampicin at 300 μg mL⁻¹, or M9 medium (94) supplemented with 100 μg mL⁻¹ leucine and 0.25% citrate or 0.5% glucose. IPTG (isopropyl-β-D-thiogalactopyranoside) at indicated concentrations was used for tacp induction. Allele exchange in the P. aeruginosa chromosome was conducted with the use of suicide vector pAKE600 (95) as described in the supplemental material.

The bacterial two-hybrid system, based on the reconstitution of Bordetella pertussis adenylate cyclase (CyaA), was used (51). E. coli BTH101 cya- were transformed with pairs of complementary vectors containing cyaT18 and cyaT25 fragments fused with indicated genes (fragments). Transformants were selected on McConkey agar base (BD, 281810) supplemented with 1% maltose, appropriate antibiotics, and 0.15 mM IPTG. After 3 days at 28°C, random colonies were restreaked to a fresh plate, incubated for 2 days, photographed, and used to inoculate cultures (LB with antibiotics, 0.15 mM IPTG) grown overnight at 28°C before performing β-galactosidase activity assays (96). A library containing random genomic DNA fragments of PAO1161 fused to the C-terminal part of CyaT18 was prepared by fragmentation of PAO1161 genomic DNA by nebulization, blunting of DNA with a mixture of T4 DNA polymerase and Klenow fragment, and ligation with SmaI digested pUT18C (51). The library consists of plasmids from around 250,000 colonies, with 40% of colonies containing an insert and an average insert size of 800 bp, which overall yields a 10× coverage of the genome. Identification of ParB partners using BACTH was performed by transformation of the library into E. coli BTH101 cells expressing CyaT25-ParB. Cells were plated on MacConkey medium and incubated at 28°C for 4 days, and red colonies were restreaked. Following plasmid isolation and retransformation, inserts of colonies showing red appearance on MacConkey plates were sequenced (Table S2).

To identify ParB partners with coimmunoprecipitation, Pae strain with flag-parB was used along with WT control strain. Exponentially growing cells (OD₆₀₀ ~0.8) were suspended in lysis buffer (10 mM Tris-HCl, pH = 8.0, 20% sucrose, 40 mM EDTA, 0.5 mg mL⁻¹ lysozyme) and incubated 30 min on ice. Subsequently, 1 volume of 2× IP buffer (1.5 M Tris, pH = 7.0, 0.3 M NaCl, 0.2% Triton X-100) was added along with PMSF (phenylmethylsulfonyl fluoride; final concentration 1 mM) and protease inhibitor cocktail (P8465, Sigma, diluted 100×), and cells were disrupted by sonication. Clarified lysates were incubated with Dynabeads (14311D, Thermo) coupled with anti-FLAG antibodies (MA1-91878, Thermo) and washed with 1× IP buffer, and proteins were eluted from the beads by sequential vortexing for 5 min at 22°C after addition of 0.1 M Glycine (pH 3.5). pH of the samples was neutralized, and proteins in fractions with highest content of FLAG-ParB (as judged by Western blotting), and corresponding control samples, were identified by mass spectrometry analysis (Mass Spectrometry Laboratory IBB PAS). Data were analyzed using MascotScan software (Matrix Science, Inc., USA). Analyses of ParB interactions with partners upon their expression in a heterologous host with the use of coimmunoprecipitation were conducted as described previously for the analysis of ParB–ParA interactions (36). E. coli BL21 cells transformed with pABB1.2 and pET28a derivatives encoding ParB and partners, respectively, were used. Samples were subjected to Western Blot analysis with use of anti-His₆ antibodies.

Cells expressing fluorescent protein fusions were cultured and fixed by addition of 1 volume of 2.8% formaldehyde, 0.04% glutaraldehyde. For DNA staining, cells were harvested and fixed in formaldehyde, followed by fixation in EtOH, overnight incubation in solution containing RNase, and staining with propidium iodide/SYTO9 mixture. Fluorescence microscopy images were captured using a Zeiss Imager M2 and analyzed using ImageJ and MicrobeJ (97). Detailed microscopy protocols and description of image quantification are included in the supplemental material.

ChIP-seq analysis was performed essentially as described before (39). Polyclonal antibodies against ParB and cells growing exponentially in LB medium (OD₆₀₀ 0.4 to 0.5) were used. Reads quality controlled using fastp (98) were mapped to the P. aeruginosa PAO1161 genome (accession number CP032126 [80]) using Bowtie (v.2.3.4.3 [99]). The presented data represent the ratio between the number of reads mapping to the parS1-4 ± 15 kbp region and reads mapping to regions encompassing half-parS sequences, defined as range of peaks called for combined WT replicates using MACS2 v.2.1.2 (100) with fold enrichment of >3 (Table S3 in the supplemental material). Remaining parts of the genome were used to calculate the theoretical coverage background, which was subtracted from both values, before calculating the ratio. The read coverage of half-parS containing regions provides an internal control for each sample, correcting for unequal ChIP efficiency. Calculation of the number of reads mapping to indicated genome parts in ChIP samples was done using the plotEnrichement function from deeptools (v.3.3.0 [101]).

Raw sequencing data were deposited in the NCBI’s Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/) under accession number GSE213881.