Molecular Insights on the Assembling of ParB1 and ParB1-parS1 Partition Complex

Abstract

ParB, which is a condensate-forming DNA clamp and a CTP switch, is a key member of the bacterial chromosome segregation apparatus. Molecular features that govern the assembly of ParB for partition assembly condensate formation that leads to DNA segregation are sparsely understood. We report that ParB1 from the multipartite bacteria Vibrio cholerae is a CTPase. The AI-predicted structural model of the ParB1-parS1 nucleoprotein complex consistent with small-angle X-ray scattering revealed how parS1 loading primes ParB1 for sliding, with the formation of a lumen for housing the sliding DNA. The CTPase domains in this dimeric ParB1-parS1 structure are accessible for the formation of higher-order assemblies. Furthermore, we showed that a truncated N-terminal segment of ParB1 containing the CTPase domain undergoes concentration-dependent oligomerization. A flexible linker joining this oligomerization-prone N-terminal segment and the C-terminal dimerization domain appears to intrinsically restrain the self-association of full-length ParB1. Our results suggest that the oligomerization propensity of the N-terminal segment of ParB1 is a key factor for higher-order partition assembly formation.

Introduction

Even though ATP or GTP-dependent protein switches are abundant in nature, only a handful of CTP switches are known. Recently, it was shown that ParB is a CTP switch that acts as a DNA clamp. − ParB is a component of the bacterial ParABS chromosome segregation or partition cassette that binds palindromic parS site(s) near the chromosomal origin. − Multiple ParBs can spread around the parS site(s) to form a large partition assembly or partition complex probably by sliding on the DNA and bridging. − ,− Additionally, ParB participates in the loading of SMC proteins on DNA. Earlier, several mechanisms of DNA segregation have been proposed. , Lately, it has been reported that ParBs can form phase-separated condensates, which are regulated by CTP. , Biological roles of these condensates in ParB-mediated DNA segregation are just beginning to emerge. −

A canonical ParB is a dimer containing three domains. , The N-terminal domain of ParB binds and hydrolyzes CTP (CTP binding domain or CTPBD). − A central domain containing helix-turn-helix motif (HTHD) specifically binds palindromic parS DNA. The C-terminal dimerization domain (CTDD) of ParB contains a weak DNA-binding site. Recent crystal structures of the nucleotide-bound and parS-bound truncated ParB family of proteins aided in understanding the CTP-dependent DNA clamping mechanism. − ,,, Binding of two nucleotides at the dimer interface of CTPBD forms a locked “N-gate,” thus forming a closed ParB clamp. − , This CTP-driven “N-gate” closure displaces parS DNA from the central HTHD, which is required for ParB sliding. Biochemical analysis shows that CTP hydrolysis is catalyzed by parS, suggesting a coupled conformational change in ParB.

ParABS-mediated DNA segregations are reported in multichromosomal bacteria, such as Vibrio cholerae and Burkholderia cenocepacia. − Both chromosomes of V. cholerae (chr1 and chr2) contain chromosome partition cassettes (ParABS1 and ParABS2) that participate in the segregations of cognate chromosomes. − While chr1 in V. cholerae undergoes asymmetric segregation, chr2 undergoes symmetric segregation. Multiple parS sequences that serve as the sites for partition assembly formation are reported in V. cholerae chromosomes. , The chr1 contains three origin-proximal parS1 sites that are highly similar to the consensus chromosomal parS sequence. The parS2 sites in chr2 were more scattered. How these partition assemblies are formed in multipartite V. cholerae and whether CTP plays any role in these processes are sparsely known.

Interconversion between conformational states associated with different binding partners as stimuli is critical for the functioning of protein switches like ParB. Small-angle X-ray scattering (SAXS) is a powerful tool for probing such binding-dependent conformational changes in proteins. Advanced methodologies, such as size-exclusion column chromatography coupled with SAXS (SEC-SAXS) and evolving factor analysis, can be particularly useful for obtaining the scattering profile of an aggregation-free protein and the exclusive scattering profile of a DNA–protein complex from a mixture, which can aid in the sensitive detection of conformational changes. In addition, SAXS is a fitting tool for probing molecular features of proteins like ParB that can contribute to self-association and condensate formation. To understand the molecular details of partition assembly formation in multipartite bacteria, we probed conformational switching and self-assembling of ParB1 from V. cholerae (vcParB1) using SAXS/SEC-SAXS and other biophysical tools.

We report that vcParB1, like many of its homologues, has CTP-hydrolyzing activity. Using SEC-SAXS, we showed that DNA, or CTP, induces a conformational change in full-length vcParB1. The Alphafold structural model of the DNA-loaded form of dimeric vcParB1 concurrent with SAXS data shows the formation of a lumen region below the parS1-binding site for capturing the sliding DNA and CTPBD domains accessible for additional oligomerization. Further, we show that a truncated vcParB1 (vcParB1trunc) containing the CTPBD and HTHD, and none of the putative intrinsically unstructured regions (IDRs), self-associates at high concentration. By comparing self-aggregation of a triple mutant that conferred flexibility in the short linker region joining the CTDD and HTHD of vcParB1 (vcParB1-linker) and the wild-type vcParB1, we showed that the linker restrains the self-associating tendency of full-length vcParB1. The assembly properties of vcParB1 are discussed in the context of higher-order partition assembly formation.

Experimental Section

Protein Expression and Purification

Either pET15b or pET28a vectors containing genes of target protein constructs (Figure A and Table S1) were transformed into BL21­(DE3) competent cells. Cells were grown at 37 °C in Luria–Bertani broth (Difco, Fisher Scientific, cat 244620) with added antibiotics (100 μg/mL ampicillin, Sisco Research Laboratory, cat 61314 or 50 μg/mL kanamycin monosulfate, Gold Biotechnology, cat K-120–5) until OD₆₀₀ reached 0.6. Next, Isopropyl-β-D-thiogalactoside (IPTG, 1 mM; Gold Biotechnology, cat I2481C50) was used for induction, and cells were grown for another 15 h at 25 °C. Cells were harvested and stored at −80 °C.

1.

CTPase activity of vcParB1. (A) Schematic diagrams showing the domain organizations of the vcParB1 constructs used in this study (vcParB1 and vcParB1-N129A in the left panel; vcParB1-linker and vcParB1trunc in the right panel; mutations are depicted by red stars). (B) Multiple sequence alignment of the N-terminal region of vcParB1 (residues 89 to 106, encompassing the conserved GxxRxxA motif) with homologous ParBs is shown. Sequence conservation scores (calculated in Jalview) are shown in the lower panel. (C) A bar graph depicting the result of the CTPase activity assay of vcParB1 (green) and vcParB1-N129A mutant (red), with and without added specific (parS1) and nonspecific DNA (nsDNA16) control (CTP, protein dimer, and DNA duplex concentrations are shown in parentheses).

For protein purification, cells were lysed by sonication (Sonics Vibra-Cell, VCX750, Sonics & Material) in the lysis buffer containing 100 mM Tris-HCl (Sigma-Aldrich, cat T6066), pH 7.5, 1 M NaCl (Sigma-Aldrich, cat S3014), 5% glycerol (Sigma-Aldrich, cat G7757), 5 mM β-mercaptoethanol (BME, Sigma-Aldrich, cat M6250), 0.2 mM ethylenediaminetetraacetic acid (EDTA, Sigma-Aldrich cat EDS), and protease inhibitor cocktail (Sigma-Aldrich, cat P8849). Proteins were purified using immobilized metal affinity column chromatography (IMAC sepharose 6 fast flow, Cytiva, cat 17092107), followed by ion exchange column chromatography (Q sepharose Fast Flow, Cytiva, cat 17051010) and size-exclusion column chromatography (HiLoad 16/600 Superdex 200 pg, Cytiva, cat 28989335). Proteins were loaded onto a Ni²⁺-charged IMAC Sepharose Fast Flow column in lysis buffer. Unbound and weakly bound proteins were eluted by wash buffer (20 mM Tris-HCl, pH 7.5, 1 M NaCl, 5% glycerol), followed by a binding buffer containing 20 mM Tris-HCl, pH 7.5, 300 mM NaCl, 5% glycerol, and 20 mM imidazole (Sigma-Aldrich, cat I2399). Bound proteins were eluted in the binding buffer with increased concentration of imidazole in steps up to a maximum concentration of 500 mM. Next, the collected protein fractions were pulled and dialyzed in the ion exchange buffer (20 mM Tris-HCl, pH 7.5, 20 mM NaCl, 5 mM BME, 5% glycerol, 1 mM EDTA). A gradient of 20 mM to 500 mM NaCl was used for eluting the protein. Buffer for the size-exclusion column chromatography run contained 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 5% glycerol, 5 mM BME, and 1 mM EDTA. Purity levels of the proteins were evaluated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Fractions from the peak region were collected and further concentrated. Trypsin digested (in-gel digestion, Trypsin Singles, Proteomics Grade, cat T7575) proteins were further confirmed by MS/MS ion search from one or more peptides in the MASCOT database server. Protein concentrations were determined using the BCA assay (Pierce BCA Protein Assay Kit, cat 23225).

Oligonucleotide Preparation

Commercially synthesized oligonucleotides were obtained in PAGE-purified form from the Sigma-Aldrich oligonucleotide synthesis facility (Table S2). Equimolar concentrations of oligonucleotides were mixed and annealed by heating at 95 °C (10 min), followed by cooling at room temperature (45 min) and at 4 °C (overnight).

CTPase Activity Assay

To detect CTPase activity of vcParB1 constructs, a malachite green assay (malachite green oxalate salt, Sigma-Aldrich, cat M6880, ammonium molybdate tetrahydrate, Sigma, cat 09878) was performed. Standard curves were prepared using potassium phosphate monobasic (Sigma-Aldrich, cat P5379). Protein and cytidine triphosphate (CTP; Thermo Scientific, cat R0451, 1 mM) were incubated in the reaction buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM MgCl₂) for 60 min at room temperature. OD₆₅₀ was measured in a plate reader (Thermo Fisher Scientific, Multiskan Go) after adding the malachite green working reagent and sodium citrate. OD obtained from the reaction buffer with only CTP was used as a blank for subtraction. Next, these experiments were repeated in the presence of oligonucleotides (parS1 and nsDNA16). All experiments were performed three to four times and are plotted in Excel (Microsoft).

Biolayer Interferometry

ParB1-DNA interactions were detected by biolayer interferometry in our in-house Octet K2 (ForteBio, USA). Biotinylated duplex DNA (bt-parS1, Table S2) containing a single parS1 sequence and nonspecific DNA (bt-nsDNA) were used for coating the streptavidin biosensor (SA biosensor, Sartorius). Protein (vcParB1trunc) concentrations were optimized before the final runs in the reaction buffer containing 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM MgCl₂, 1 mM BME, 0.05% Tween20 (Sigma-Aldrich, cat P9416), and 0.5% bovine serum albumin (Sigma-Aldrich, cat A2153). Binding experiments were repeated three times. Data analysis was performed using software Data Analysis HT 11.1 (Sartorius). A heterogeneous binding model was used for data analysis, which provided two on-rates and two off-rates. It was not possible to obtain a reliable dissociation constant due to much variability in the data, which is probably due to the self-associating nature of truncated vcParB1trunc.

Analytical Ultracentrifugation

To find out about the assembly state of vcParB1trunc, analytical ultracentrifugation (AUC) runs were performed in duplicate at 42,000 rpm at 20 °C using ProteomeLab XL-I protein characterization system (Beckman Coulter). An AN-50 Ti rotor was used and scans were taken at 280 nm. Purified vcParB1trunc was used for the AUC experiments with the following reference buffer: 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 5% glycerol, and 5 mM BME. SEDFIT was used for the data analysis.

Dynamic Light Scattering

To evaluate self-association in vcParB1 and vcParB1-linker constructs, a dynamic light scattering (DLS) experiment was performed using Prometheus Panta (NanoTemper Technologies), with a laser wavelength of 405 nm, available for demonstration. Purified proteins (vcParB1 and vcParB1-linker) were used in a buffer containing 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 5% glycerol, and 5 mM BME at 20 °C for the DLS runs. Data analysis was performed using the Panta analysis software available with the instrument (NanoTemper Technologies). DLS data analysis provided cumulative polydispersity indices of vcParB1 and vcParB1-linker.

Analytical Size-Exclusion Column Chromatography

To find out about concentration-dependent changes in oligomeric state, size-exclusion column chromatography (SEC) runs were performed using vcParB1trunc at two different concentrations (70 and 370 μM). HiLoad 16/600 Superdex 75 pg column (Cytiva) was used for the SEC runs at a 0.3 mL/min flow rate and at 10 °C. The following buffer was used for the SEC runs: 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 5% glycerol, and 5 mM BME. The column was standardized using the gel filtration marker kit MWGF200 (Sigma-Aldrich). Excel (Microsoft) was used for plotting the results.

Small-Angle X-ray Scattering and Size-Exclusion Column Chromatography Coupled with SAXS Experiments

Purified proteins and protein–DNA complexes (2:1 stoichiometry) were snap-frozen under liquid nitrogen and shipped to the synchrotron beamline in a dry shipper. All SAXS and SEC-SAXS data were collected at the beamline BM29, European Synchrotron Radiation Facility (ESRF), Grenoble, France. Coupling SAXS with SEC helped us to obtain SAXS data from proteins without aggregation. Further, SEC-SAXS helped in separating protein/protein–DNA complexes from unbound DNA. Batch-mode SAXS experiments were performed for evaluating self-association in vcParB1 and vcParB1-linker. Experimental parameters are presented in Table S3.

SAXS and SEC-SAXS Data Analysis

Resultant SAXS and SEC-SAXS data were analyzed using BioXTAS RAW 2.2.1 and ATSAS suite of software. , Following buffer subtraction, peak regions of protein and protein–DNA complexes from the SEC-SAXS elution profiles were selected based on singular value decomposition (SVD) analysis using BIOXTAS RAW. Averaged scattering profiles were used for subsequent analysis. Regularized alternating least-squares (REGALS) analysis was performed for the SEC-SAXS data set of the vcParB1-parS1 complex. SVD and REGALS analysis of the SEC-SAXS data helped in obtaining SAXS profiles from distinct scatterers. Pair distribution functions were calculated using GNOM. Second virial coefficients (B22) were estimated from concentration series of batch-mode SAXS data, using the following equation

$$\frac{1}{I(c,0)}=\frac{1}{F(0)}+\frac{2·M·B22·c}{F(0)}$$ 1

where I(c,0) is the forward scattering intensity at concentration c, M is the mass, and F(0) is the form factor at zero scattering angle. Excel (Microsoft) was used for preparing the graphs and analyzing data.

Structure Modeling and Analysis

Structures of dimeric vcParB1 and vcParB1-parS1 assembly were predicted using Alphafold3. AFsample2 was used for generating conformationally varied structures of dimeric vcParB1-parS1. CRYSOL was used for calculating theoretical scattering profiles from atomic coordinates. Top-ranked predicted models, along with the experimental SAXS data, were used for minimal ensemble modeling in the BilboMD server using suggested protocols with CHARMM. Multiple sequence alignments were prepared in Jalview, and flexible regions were predicted using the “hot loop” option in Jalview.

Safety Statement

No unexpected safety hazards were encountered.

Results

ParB1 from V. cholerae Is a CTPase

Recently, it has been shown that many ParB family enzymes have CTPase activity that aids them to act as DNA clamps. − The CTPase active site located at the CTPBD domain of ParB contains a signature GxxRxxA sequence motif. − Like many other ParBs, the N-terminal region of vcParB1 contains this signature GxxRxxA sequence motif (Figure B). To detect CTPase activity in vcParB1, we performed a malachite green assay (Figure C). Addition of a 16-meric DNA duplex containing the palindromized parS1 sequence appears to mildly enhance the CTPase activity of vcParB1 by a few folds (Figure C). A nonspecific duplex DNA of equal length (nsDNA16) as a control showed no such effect on the CTPase activity of vcParB1 (Figure C). Interaction between parS1 and a truncated version of vcParB1 (vcParB1trunc) containing the parS1-binding HTHD was experimentally verified (Figure S1). A mutation in the residue N112S at the CTPase active site reportedly made bsParB binding competent but hydrolysis deficient. A mutated version of vcParB1 containing an alanine in the corresponding position (vcParB1-N129A) showed reduced CTPase activity when compared with the activity of the wild type (Figure C). Thus, vcParB1 has CTP-hydrolyzing activity, which is mildly enhanced in the presence of cognate parS1 DNA.

DNA Binding Induces Conformational Changes in ParB1

To understand the effect of DNA on the conformation of full-length vcParB1, we performed a series of SEC-SAXS runs for vcParB1, parS1, nsDNA16, and vcParB1-DNA assemblies (Figures A and S2). Based on SVD analysis surrounding the peak region (Figure S3A,B), an averaged SAXS profile of vcParB1 (vcParB1-peak) was obtained. The eluted vcParB1 is a dimer, with an estimated molecular mass of 59.5 kDa (theoretical monomeric mass: 34.1 kDa, mass was estimated from the averaged SAXS data using a Bayesian method). The vcParB1-peak profile was analyzed to obtain sizes (Table and Figure S4A).

2.

SEC-SAXS profiles of vcParB1. (A) SEC-SAXS profiles, along with R _g distribution, for vcParB1 (solid blue line, R _g as a blue circle), vcParB1-parS1 (solid red line, R _g as a red square), and vcParB1-nsDNA16 (solid green line, R _g as a green triangle) are shown (x-axis: SEC-SAXS frame number; primary vertical axis in the left: integrated intensity in arbitrary unit; secondary vertical axis in the right: R _g in nm, obtained using the Guinier equation, calculated as a moving average of 5 frames). The elution profile for parS1 (dark blue solid line) is shown. The elution profile for nsDNA16, which elutes at the same place as parS1, is not shown. Note that the shown R _g values are obtained from weak scattering profiles averaged over 5 frames and, therefore, are noisy. (B) Normalized pair distribution functions (P(r)/I(0) versus r, P(r) is the pair distribution function, I(0) is the forward scattering intensity, and r is the pairwise distance between atoms in nm) of vcParB1-peak (blue), vcParB1-parS1-peak (red), and vcParB1-CTP (black) are shown. In the inset, a magnified version of the P(r) function plot is shown for the region where r = D _max. (C) SEC-SAXS elution profiles of CTP-treated vcParB1 (solid black line, with R _g as a gray diamond) and vcParB1 (solid blue line, with R _g shown as a blue circle) are shown (plotted in the same way as in 2A).

1. SEC-SAXS Data Analysis .

data set	frames used for averaging	Guinier R g (nm)	q·R g range	real space R g (nm)	D max (nm)	injected protein concentration(μM)
vcParB1-peak	522–565	4.05 (±0.03)	0.5–1.27	4.34 (±0.03)	17.0	293
vcParb1-parS1-peak	527–564	3.89 (±0.05)	0.56–1.3	4.16 (±0.04)	15.0	126 (62.8)
vcParB1-parS1-deconv (REGALS)		3.83 (±0.04)	0.47 −1.28	4.05 (±0.04)	15.0
vcParB1-CTP-peak	527–564	3.93 (±0.04)	0.525–1.29	4.24 (±0.05)	16.0	175.8 (718.0)

^aBioXTAS RAW and ATSAS were used for computing R _g and maximum particle diameter (D _max). ,

^bFrame numbers for averaging to obtain averaged SAXS profiles were decided based on SVD analysis (Figure S3).

^c q · R _g range for Guinier R _g calculations (R _g is the radius of gyration, and q is the momentum transfer in nm^–1).

^dReal space R _g and D _max values were obtained using GNOM from the ATSAS suite. D _max values were rounded off.

^eDNA or CTP concentrations are in parentheses. 100 μL of samples were injected for the SEC-SAXS runs.

A comparison of the SEC-SAXS profiles of vcParB1, vcParB1-parS1, vcParB1-nsDNA16, parS1, and nsDNA16 showed that the eluted DNA peaks were separated from the eluted protein and protein–DNA complex peaks (Figure A). The overall radius of gyration (R _g, which is a measure of size) distribution surrounding the peak region showed that the average size of the vcParB1-parS1 complex, and that of the vcParB1-nsDNA16 complex to a lesser extent, is reduced when compared with the average size of vcParB1 (Figure A). This shift in R _g suggests a change in size upon vcParB1-parS1 assembly, which could be an indication of conformational change.

To obtain a SAXS profile of the vcParB1-parS1 complex from the SEC-SAXS data collected from a mixture of protein, DNA, and DNA–protein complex, we performed REGALS decomposition (Figure S5A–C). The major component (vcParB1-parS1-deconv), which was most compact in the Kratky plot, was assigned to be the protein–DNA complex (Figure S5A). SVD analysis of the vcParB1-parS1 peak region identified a segment containing one component, which was averaged (vcParB1-parS1-peak; Figure S3C,D), and used for subsequent analysis along with the vcParB1-parS1-deconv profile. Pair distribution function (PDF, which is dependent on the shape of the scattering particle), maximum particle diameter (D _max), and real space R _g derived from PDF were obtained from analyzing these SAXS profiles (Table ). A comparison of PDFs obtained from the vcParB1-parS1-peak/vcParB1-parS1-deconv and vcParB1 profiles showed that the vcParB1-parS1 is ∼2 nm or 20 Å smaller in size than vcParB1, supporting conformational change in vcParB1 upon parS1 binding (Table and Figure B). A similar parS-induced shrinking in mycobacterial ParB was reported earlier.

To find the effect of nucleotide on vcParB1 conformation, we performed SEC-SAXS experiments with CTP-treated vcParB1 in the presence of mobile buffer supplemented with an excess of CTP (1.2 mM). We note that the cytosolic concentration of CTP in bacteria is ∼0.3 mM. Addition of a nonhydrolyzable CTP analog induced aggregation in vcParB1, and therefore CTP was used for this SEC-SAXS experiment. SVD analysis aided in obtaining the averaged intensity profile of nucleotide-treated vcParB1 (vcParB1-CTP-peak) from a set of contiguous frames (Figure S3E,F). A small reduction in the size of vcParB1 upon CTP binding suggested that, like parS1, CTP induces conformational change in vcParB1 (Table , Figure B,C).

Structural Modeling of vcParB1-parS1 Complex

To predict the structure of the dimeric vcParB1-parS1 nucleoprotein complex, we used Alphafold3 (40, Figures A,B and S6A). Poorly defined segments with low predicted local distance difference test (pLDDT) scores in these structures, one in the N-terminal region and another short linker joining CTDD and HTHD, appear to correlate well with predicted IDRs (Figures A–C and S6). To evaluate the predicted models, we compared theoretical scattering profiles computed from these predicted models with experimental SAXS data. One of the profiles of vcParB1-parS1 agreed with experimental data, with a χ² value of ∼1.7 (model rank 1, Figures A,B and S6A,B).

3.

Structural modeling of vcParB1-parS1 nucleoprotein assembly. (A, B) Two views of the structure of the dimeric vcParb1-parS1 complex predicted by Alphafold3 (model rank 1, which best-matched with experimental SAXS data) are shown as a cartoon. Proteins are colored by pLDDT score (high pLDDT in blue to low pLDDT in red) and parS1 is shown in gray. Note that all predicted models of vcParB1-parS1 (rank 0 to rank 4) are shown in Figure S6A. The fit between the calculated SAXS profiles of the models shown in Figure S6A and experimental SAXS data is shown in Figure S6B. (C) Schematic diagram of the domain organizations of vcParB1 is shown in the upper panel. Predicted disorder probability of vcParB1 versus residue number, calculated in PrDOS (prdos.hgc.jp), is shown in the lower panel. The threshold is shown as a horizontal red line. (D) Structure of vcParB1-parS1 modified to fit experimental SAXS data in BilboMD is shown as a cartoon (protein chain in cyan and DNA in gray, semitransparent surface of the protein moiety in light gray). In the inset, the region surrounding the lumen of vcParB1-parS1 is shown with electrostatic potential mapped on the semitransparent surface (calculated in CCP4Molecular graphics, colored from −0.5 V in red to +0.5 V in blue). The lysine residues that define the secondary DNA-binding site in CTDD of vcParB1 are shown as yellow sticks. (E) The fit between theoretical scattering profile (solid black line) calculated using the BilboMD model shown in 3D and experimental vcParB1-parS1-peak data (red open circle; χ² ∼ 0.98) is plotted (logI versus q in Å^–1; I is the intensity in arbitrary units, and q is the momentum transfer). (F–H) Cartoon of vcParB1-parS1 dimer (vcParB1 in cyan, parS1 in gray) is shown superimposed with (F) CTDD (brown, pdb code 5noc; RMSD 1.5 Å for 49 aligned Cα atoms), HTHD (crimson, pdb code 6t1f, RMSD 1.6 Å for 109 aligned Cα atoms), and CTPBD (red, pdb code 6t1f, RMSD 1.5 Å for 74 aligned Cα atoms) domains of homologous ParBs, (G) dimeric DNA-bound cvParBtrunc (red, pdb code 6t1f, RMSD 1.6 Å for 109 aligned Cα atoms), and (H) a monomer of truncated, tetrameric hpParB containing HTHD and CTPBD (red, pdb code 4umk, RMSD 2.9 Å for 131 aligned Cα atoms).

Next, we used BilboMD, which generates a conformational ensemble of highly populated structures that concur with SAXS data, to obtain a minimal ensemble of the vcParB1-parS1 complex starting from the Alphafold3 model rank 1. The best (most populated) model from the BilboMD, which fits the experimental SAXS data with a χ² value of 0.98 and R _g of 3.9 nm, contains a single conformer (Figure D–E). This model is consistent with an “N-gate open” form of vcParB1-parS1 in solution, in which the CTPase domains are far from each other. The parS1-bound model of vcParB1 revealed a lumen region surrounded by HTHD, linker region, and CTDD (Figure D).

Individual domains of the vcParB1 model are comparable to the individual domains of the known ParB structures (Figure F). The structure of the C-terminal dimeric CTDD of vcParB1 was strikingly similar to the NMR structures of CTDD of ParB from Bacillus subtilis, with which it bears low sequence identity (Figure F). The CTPBD in the predicted models showed variability in its orientation (Figure S6C). Surprisingly, the CTPBD domains in the best model of vcParB1-parS1 adopted an open form when compared with the structure of truncated, parS-bound ParB from Caulobacter vibrioides (cvParB, Figure G). The CTPBD-HTHD segment in this open form of vcParB1-parS1 is roughly similar to the equivalent segment in the structure of tetrameric, truncated ParB from Helicobacter pylori that participates in DNA bridging (hpParB, Figure H).

None of the computed scattering profiles obtained from the predicted models of DNA-free vcParB1 agreed with experimental SAXS data obtained from the free form of vcParB1 or CTP-treated vcParB1 (Figure S7A–D). Therefore, we used AFsample2 to generate conformationally variable models of vcParB1 (Figure S8A). Since these AFsample2 models, either individually or combined as an ensemble, did not concur with experimental SAXS data, we modeled the minimal ensemble of vcParB1 using BilboMD. The best minimal ensemble obtained from BilboMD (χ² ∼ 1.25) contains two single conformers with widely different conformation (Figure S8B). In one of these two conformers of vcParB1, the two CTPBD domains are in physical proximity (Figure S6B). In the other conformer, the CTPBD domains are far from each other (Figure S6B). Probably due to conformational plasticity in solution, it was not possible to obtain a converged model of DNA-free vcParB1.

Truncated Form of vcParB1 Devoid of the Putative Intrinsically Unstructured Regions Is Highly Heterogeneous

Surprisingly, the SEC-SAXS elution profile of a truncated construct of vcParB1 (vcParB1trunc) without the IDR and CTDD showed noticeable heterogeneity in size (Figure A). Distribution of R _g around the eluted peak with a distinct slope, and SVD analysis of the peak region, indicated the presence of multiple components that are being separated while eluting (Figure A,B). We note that SAXS is a weight-average method, and the presence of a small amount of higher-order oligomer can change the SAXS profile significantly. As it was not possible to obtain an averaged SEC-SAXS profile of vcParB1trunc due to a large variation in R _g across the peak, a number of SAXS profiles averaged over a few frames from different sections of the eluted peak were used for further analysis (Figure S9). Kratky plots obtained from these SAXS profiles (sequentially numbered from I to VI from left to right) suggested the presence of folded proteins (Figure C). Comparison of PDFs obtained from these SAXS profiles, with a large variation in D _max, further supported heterogeneity in vcParB1trunc (Figure d).

4.

Heterogeneity in a truncated segment of vcParB1. (A) SEC-SAXS profile of vcParB1trunc with integrated intensity (primary vertical axis, blue line, arbitrary unit) and moving average R _g over 5 frames (secondary vertical axis, red diamond, in nm) versus frame numbers. (B) Result of SVD of the vcParB1trunc SEC-SAXS data set encompassing the peak region. A semilog plot of the singular value versus the index is shown. (C) Dimensionless Kratky plots, (qR _g)² I(q)/I(0) versus qR _g (I(q) is the intensity in arbitrary units, and q is the momentum transfer) and (D) normalized PDF (normalized by I(0)) of averaged SAXS profiles of segments obtained from the above data set, with segment ranges I to VI (early to late eluting segments). PDFs were calculated and shown as described in Figure B. The following frames were used for averaging: I (699–704), II (707–717), III (728–740), IV (754–775), V (794–805), and VI (816–826). (E) Plot of sedimentation coefficient distribution c(s) versus sedimentation coefficient (s) is shown for vcParB1trunc (blue). (F) SEC profiles (normalized to a maximum value of 1) of vcParB1trunc from the same batch at two concentrations are shown in blue (70 μM) and orange (370 μM). Vertical black lines show the peak positions of cytochrome c (12.5 kDa) and alcohol dehydrogenase (150 kDa).

AUC experiments confirmed that purified vcParB1trunc (∼70 μM) is a monomer (estimated mass: 23.4 kDa, theoretical mass: 23.9 kDa; Figure E). A high concentration (∼386 μM) of vcParB1trunc was used for the SEC-SAXS experiment, which is unsuitable for AUC. On the other hand, SAXS experiments require highly concentrated proteins. Therefore, an SEC run with vcParB1trunc at two concentrations (∼70 and ∼370 μM) was performed and elution profiles were compared, which was consistent with the oligomerization of vcParB1trunc at higher concentrations (Figure F). According to previous reports, CTPBD of ParBs can oligomerize in many ways, such as a dimer and a tetramer. , At high concentration, vcParB1trunc probably exists as a mixture of oligomers that are formed by intermolecular interactions.

Elevated Flexibility in the Linker Region Promotes Self-Association of vcParB1

Next, we asked what molecular features could be responsible for balancing between the intramolecular clamp formation within the dimeric vcParB1 and intermolecular self-association that is required for higher-order partition assembly formation. A short flexible linker region has been identified in many ParB homologues that join CTDD and HTHD (Figures A–C and A,B). A number of predicted ParB structures downloaded from the Alphafold2 database (alphafold.ebi.ac.uk) showed very low pLDDT scores in this linker region, further supporting predicted flexibility in this region (Figure B). We hypothesized that this linker is restricting the N-terminal region of wild-type vcParB1 dimer from self-interaction with neighboring dimers by limiting the rotational freedom around the linker. To test this hypothesis, we designed a glycine-rich triple mutant (vcParB1-linker; P235G, D237G, K239A, Figure A) for elevating flexibility in this linker region. We expect that a highly flexible linker will allow the N-terminal segment of vcParB1-linker to liberally rotate about the linker to sample wider conformational space, thus enhancing chances of intermolecular interactions for vcParB1-linker when compared with the wild type.

5.

Mutation in the linker region makes vcParB1 more association-prone than the wild type. (A) Multiple sequence alignment of the linker region of vcParB1 (Uniprot accession number: Q9KNG7, residues 209–279) and other ParB homologues is shown with predicted flexibility (hot loop, colored in red, calculated in Jalview). Sequence conservation scores are shown below the sequence alignment. The positions of the three mutations in the vcParB1-linker construct are marked with blue stars. (B) Superposed Alphafold predicted structures of vcParB1 and ParB homologues (Uniprot accession numbers: Q9PB63, Q9RYD8, Q87BY1, Q92JI0, Q1CVJ4, Q9JW77), colored by pLDDT score (90 in blue to 50 in red), are shown. Predicted structures were superimposed using the Cα atoms of HTHD. The CTPBD domains are not shown. (C) Dimensionless Kratky plots (as described in Figure C) and (D) normalized PDFs (as described in Figure B) of SAXS data obtained from vcParB1 and vcParB1-linker. Real space R _g and I(0) values from the PDFs were used for calculating the Kratky plots. PDF obtained from the vcParB1-peak SEC-SAXS data is plotted as a dashed black line in 5D.

Purified vcParB1-linker showed weak CTPase activity (Figure S10A). Cumulative polydispersity indices obtained from DLS experiments supported a higher level of intermolecular interactions in vcParB1-linker compared with wild-type vcParB1 (Figure S10B).

To further compare the self-associating properties of vcParB1 and vcParB1-linker, we obtained batch-mode SAXS data. Kratky plots obtained from these SAXS data suggested a high level of self-association (as opposed to self-avoiding behavior) in the vcParB1 constructs (Figure C). D _max values obtained from the vcParB1-linker data sets were consistently higher than those from the vcParB1 data sets, suggesting a higher degree of self-association in the linker mutant (Figure D). To quantitate self-interaction in vcParB1 and vcParB1-linker, we calculated the second virial coefficient (B22) from SAXS data. , B22 for vcParB1 and vcParB1-linker were estimated to be −3.99 × 10^–4 mol·mL·g^–2 and −4.66 × 10^–4 mol·mL·g^–2, respectively. Negative B22 values imply attractive interaction, and it appears from the B22 values that the vcParB1-linker is more association-prone than vcParB1 within our experimental concentration regime. Thus, imparting more flexibility in the linker region made vcParB1 more association-prone.

Discussion

Since the recent serendipitous discovery that many ParBs act as CTP-dependent conformational switches and DNA clamps, the number of CTP switches reported in the literature is steadily growing. Further, it has been shown that some ParBs can self-associate to form condensates, which is reportedly aided by CTP. − However, molecular-level details of how ParBs associate to form partition assembly and how CTPs affect this process are little known for multipartite bacteria.

We report that ParB1 from V. cholerae is a CTPase that undergoes a change in conformational state upon DNA or CTP binding (Figures and ). AI-predicted structures of vcParB1 in DNA-bound and DNA-free states were evaluated and altered, using experimental SAXS data (Figures and S6–S8). A converged structural model of the DNA-free form of vcParB1 consistent with SAXS data was not obtained, which is probably due to conformational plasticity in solution. A model of the DNA-bound form of vcParB1 that agrees with SAXS data revealed a lumen poised for capturing the dislodged parS1, suggesting that vcParB1 can function as a sliding clamp (Figure D–E).

Unlike the CTPBD domains in the crystal structure of truncated cvParB-parS, the two CTPBD domains are not in physical proximity in our vcParB1-parS1 model (Figure G). On the other hand, the orientation of the CTPBD and HTHD domains in our vcParB1-parS1 model is similar to the orientation of CTPBD and HTHD in the DNA-bridged form of a truncated hpParB CTPase (Figure H). The two CTPBD domains in our predicted model of vcParB1-parS1 are available for either intramolecular (further conformational change for clamping) or intermolecular (higher-order self-association) interactions (Figure A,B).

6.

Oligomerization-promiscuity can lead to various possible modes of self-association in vcParB1. (A) In the left panel, cartoons of predicted structural models of vcParB1-parS1 (protein in cyan, DNA in gray) are shown as a closed dimer (superposed CTPBD of cvParB is shown in red with bound nucleotide as a green sphere, pdb code 7bm8) and a dimer of an open dimer (cartoon, with superposed dimeric CTPBD of cvParB is shown in red and bound nucleotides as a green sphere, pdb code 7bm8). In the right panel, a possible higher-order assembly of vcParB1 bound to parS1 (superposed on the 3 CTPBD of tetrameric, DNA-bridging form of hpParB, pdb code 4umk, shown as a red worm) is shown as a cyan worm, with parS1 as a gray cartoon. (B) Schematic diagram showing how various modes of self-association of the CTPBD domains of DNA-associated vcParB1 (shown in A) can possibly form the partition assembly.

Quite unexpectedly, a trimmed vcParB1trunc containing CTPBD and HTHD showed concentration-dependent self-association (Figure ). Mutational analysis suggests that the linker region joining the HTHD and CTDD limits the higher-order self-assembling propensity of full-length vcParB1 (Figure ). This linker probably poses a physical barrier to the rampant self-association of the vcParB1 dimer with other dimers by making the interaction interfaces of the N-terminal segments less accessible for intermolecular interactions.

Like other ParBs, vcParB1 forms condensates (unpublished results). ,, Therefore, self-associating propensity of vcParB1 is hardly surprising. Since local concentration of ParBs can be as high as 500 μM within the partition assembly, the observed oligomerizing tendency of vcParB1trunc at high concentrations might be relevant for partition assembly formation. Oligomerization of domains, among other factors, is a known influencer of condensate formation in multidomain proteins. We suggest that the observed oligomerization propensity of CTPBD-containing vcParB1trunc, which is intrinsically curbed by a linker in the full-length form of vcParB1, is a likely enabler of partition assembly formation. It appears that CTP can drive ParB condensate formation. ,,, How the assembling of ParBs is fine-tuned by CTP for partition assembly formation is the next key question to be answered.

Conclusions

ParB1 from multipartite V. cholerae is the latest addition to the family of ParB CTP switches. A change in the conformational state of vcParB1 upon binding parS1 or CTP suggests that it can function as a conformational switch. We combined Alphafold and SAXS to build a low-resolution solution conformational state of vcParB1-parS1 that showed how DNA loading prepares vcParB1 for sliding by forming a lumen. In our model, vcParB1 bound to parS1 assumes an open form poised for self-association in various ways that can potentially lead to clamping and higher-order assembly. Unlike full-length vcParB1, a truncated N-terminal segment of ParB1 showed noticeable heterogeneity. A flexible linker appears to restrain the self-associating propensity in the full-length vcParB1. The observed self-associating tendency of the vcParB1 segment is a likely enabler of partition assembly formation in the physiological concentration regime.

Glossary

Abbreviations

AI

artificial intelligence

ATP

adenosine triphosphate

AUC

analytical ultracentrifugation

B22

second virial coefficient

bt-nsDNA

biotinylated parS1 and nonspecific duplex DNA

CTDD

C-terminal dimerization domain of ParB1

CTP

cytidine triphosphate

CTPBD

CTP binding domain of ParB1

DLS

dynamic light scattering

Dmax

maximum particle diameter

GTP

guanosine triphosphate

HTHD

helix-turn-helix domain of ParB1

IDR

intrinsically disordered region

nsDNA16

16-base-pair nonspecific duplex DNA

PDF

pair distribution function

pLDDT

predicted local distance difference test

RMSD

root mean square deviation

SAXS

small-angle X-ray scattering

SEC

size-exclusion column chromatography

SEC-SAXS

size-exclusion column chromatography coupled with SAXS

SVD

singular value decomposition

SAXS data were deposited to SASBDB (www.sasbdb.org), with following accession numbers: SASDVD3, SASDVE3, SASDVF3, SASDVG3, SASDVH3, SASDVJ3, SASDVK3, SASDVL3, SASDVM3, SASDVN3, SASDVP3, SASDVQ3, SASDVR3, SASDVS3, SASDVT3, SASDVU3, SASDVV3

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsbiomedchemau.5c00284.

• Tables S1–S3: the preparation of constructs, oligonucleotide sequences, and SAXS/SEC-SAXS experimental parameters. Figures S1–S10: experimental results of biolayer interferometry, SAXS and SEC-SAXS data analysis, and dynamic light scattering (PDF)

The authors declare no competing financial interest.