A network of cis and trans interactions is required for ParB spreading

Abstract

Most bacteria utilize the highly conserved parABS partitioning system in plasmid and chromosome segregation. This system depends on a DNA-binding protein ParB, which binds specifically to the centromere DNA sequence parS and to adjacent non-specific DNA over multiple kilobases in a phenomenon called spreading. Previous single-molecule experiments in combination with genetic, biochemical and computational studies have argued that ParB spreading requires cooperative interactions between ParB dimers including DNA bridging and possible nearest-neighbor interactions. A recent structure of a ParB homolog co-crystallized with parS revealed that ParB dimers tetramerize to form a higher order nucleoprotein complex. Using this structure as a guide, we systematically ablated a series of proposed intermolecular interactions in the Bacillus subtilis ParB (BsSpo0J) and characterized their effect on spreading using both in vivo and in vitro assays. In particular, we measured DNA compaction mediated by BsSpo0J using a recently developed single-molecule method to simultaneously visualize protein binding on single DNA molecules and changes in DNA conformation without protein labeling. Our results indicate that residues acting as hubs for multiple interactions frequently led to the most severe spreading defects when mutated, and that a network of both cis and trans interactions between ParB dimers is necessary for spreading.

Figure 1.

Crystal structure of HpSpo0J reveals a network of cis and trans interactions for ParB spreading (see colors online). (A) Crystal structure of the C-terminally truncated HpSpo0J–parS complex (30) (PDB code: 4UMK). (B) Cartoon representation of the crystal structure (figure is not drawn to scale). Each chain (chain A in blue, chain B in orange, chain C in red and chain D in green) is a C-terminally truncated HpSpo0J monomer that is part of a dimer bound to half of a 24-bp parS DNA duplex. Only two DNA molecules were available from the original PDB file. Note that tetramerization of the HpSpo0J monomers is asymmetric due to chain C failing to interact with chain D in the crystal structure. (C) Multiple interactions in trans between chain A (blue) and chain B (orange) coordinated by the two R89 residues on each chain. (D) Multiple interactions in cis between chain B (orange) and chain D (green) coordinated by the single R89 residue on chain D. (E) Multiple interactions in cis between chain A (blue) and chain C (red) coordinated by the single R89 residue on chain C. Yellow dashed lines indicate hydrogen bonds, and magenta dashed lines indicate hydrophobic interactions. Figures were prepared in PyMOL. (F) A 2D network map generated from the crystal structure (see ‘Materials and Methods’ section) indicating cis (blue) and trans (green) interactions within the HpSpo0J–parS tetrameric complex. Interactions between residues within the same HpSpo0J monomer are shown in gray. Highly conserved residues that act as hubs for multiple interactions are circled in magenta. Residue number corresponds to that in HpSpo0J.

Figure 2.

In vivo characterization of BsSpo0J spreading (see colors online). Localization of mGFPmut3-BsSpo0J variants. See Supplementary Figures S3–5 for images of all characterized mutants. Nucleoid (false-colored red) was labeled with HBsu-mCherry. Scale bar = 5 μm.

Figure 3.

Single-molecule DNA compaction by BsSpo0J (see colors online). (A) Schematic of the single-molecule protein-induced fluorescence enhancement (PIFE) assay (35). A 20-kb dsDNA sparsely labeled with Cy3 dyes (green) is tethered to the surface of a functionalized glass coverslip and extended by buffer flow. Step 1: binding of unlabeled proteins to DNA enhances the fluorescence of nearby Cy3 dyes (red) due to PIFE. Step 2: changes in DNA conformation mediated by proteins can be simultaneously detected as a change in length of flow-stretched DNAs. PEG, polyethylene glycol; SA, streptavidin. Figures are not drawn to scale. (B) Demonstration of the single-molecule PIFE assay with wild-type BsSpo0J (100 nM). Trajectories of individual DNAs are shown in gray and the average over all trajectories (n = 38) is shown in red. The fold increase in integrated fluorescence intensity over time was calculated by dividing each trajectory by the value averaged for the first 5–10 s before protein binding. DNA length was normalized to the maximum values in individual trajectories. Time zero was defined as the starting point of protein association. t_lag is lag time between protein binding and the initiation of DNA compaction. Insert: kymograph of a single DNA. Scale bar = 6 s. (C) Fold increase in the integrated fluorescence intensity and DNA length trajectories for wild-type BsSpo0J (black; the same curve is reproduced in each panel) and mutants (red) at a protein concentration of 100 nM. Each trajectory was averaged over 20–30 Cy3-labeled DNAs. See Supplementary Figures S9 and 12 for trajectories of all characterized mutants.

Figure 4.

In vitro characterization of the specific binding of BsSpo0J to parS DNA duplexes. EMSA of wild-type BsSpo0J and mutants binding to either a radioactively labeled 39-bp parS DNA substrate supplemented with cold 39-bp scrambled parS competitor DNA as shown in (A, C and E), or a labeled 24-bp parS substrate without competitor DNA as shown in (B, D and F) (see ‘Materials and Methods’ section). Protein concentrations were 0.2, 0.4, 0.8 μM in (A and B) and 0.2, 0.4, 0.8 and 1.0 μM in (C–F). Asterisk and arrow indicate position of the wells and free DNA respectively in each gel. See Supplementary Figures S13 and 14 for results of all characterized mutants.

Figure 5.

Thermal stability and oligomeric state of BsSpo0J (see colors online). (A) Thermal denaturation curves for wild-type BsSpo0J and mutants at a protein concentration of 100 μg ml⁻¹ measured with differential scanning fluorimetry (see ‘Materials and Methods’ section). Fluorescence intensities were normalized to the maximum in each curve. Only one replicate of each protein is shown. See Supplementary Figures S18 and 19 for results of all characterized mutants. (B) Melting temperatures of wild-type BsSpo0J (highlighted by the red dashed line for reference) and mutants arranged in groups (see Table 2) determined from a Boltzmann fitting on the thermal denaturation curves (see ‘Materials and Methods’ section). Error bars are standard errors of the mean between three replicates. Group II mutants were significantly less stable compared to the wild-type protein (**P < 0.001, based on an unpaired t-test with unequal variances). (C) Chromatograms of wild-type BsSpo0J and mutants measured in size exclusion chromatography. All mutants showed a major peak at 12.7 ml overlapping with the wild-type protein, corresponding to a BsSpo0J dimer. Group II mutants (G77S and P62A) displayed a significant peak at 8.5 ml, corresponding to void volume (Vo). G77S also showed a second major peak at 13.4 ml with a lower molecular weight. All proteins showed a minor peak at 15.5 ml with a low molecular weight, likely corresponding to a monomeric protein. Absorbance at 280 nm was normalized to the maximum in each curve.

Figure 6.

Model of ParB spreading (see colors online). (A) ParB proteins dimerized through their C-terminal domains bind specifically to parS (red) and non-specifically to DNA that can be up to tens of kilobases away from parS. (B) ParB dimers bound to DNA can interact with each other through their N-terminal domains both in cis and in trans. Interactions in cis may occur between ParB dimers that are bound directly adjacent to each other on DNA or far apart but brought into close proximity through DNA looping. (C) ParB dimers nucleated at parS oligomerize into a tetrameric structure through interactions both in cis and in trans. A higher-order partition complex is assembled through DNA bridging that connects distant DNA loci and traps large DNA loops over multiple kilobases.