Molecular architecture of the DNA-binding sites of the P-loop ATPases MipZ and ParA from Caulobacter crescentus

Abstract

The spatiotemporal regulation of chromosome segregation and cell division in Caulobacter crescentus is mediated by two different P-loop ATPases, ParA and MipZ. Both of these proteins form dynamic concentration gradients that control the positioning of regulatory targets within the cell. Their proper localization depends on their nucleotide-dependent cycling between a monomeric and a dimeric state and on the ability of the dimeric species to associate with the nucleoid. In this study, we use a combination of genetic screening, biochemical analysis and hydrogen/deuterium exchange mass spectrometry to comprehensively map the residues mediating the interactions of MipZ and ParA with DNA. We show that MipZ has non-specific DNA-binding activity that relies on an array of positively charged and hydrophobic residues lining both sides of the dimer interface. Extending our analysis to ParA, we find that the MipZ and ParA DNA-binding sites differ markedly in composition, although their relative positions on the dimer surface and their mode of DNA binding are conserved. In line with previous experimental work, bioinformatic analysis suggests that the same principles may apply to other members of the P-loop ATPase family. P-loop ATPases thus share common mechanistic features, although their functions have diverged considerably during the course of evolution.

Figure 1.

Identification of residues critical for DNA binding by MipZ. (A) Surface representation of the MipZ dimer structure (PDB ID: 2XJ9) with the 51 amino acids selected for alanine-scanning mutagenesis marked in purple. The two subunits are depicted in green and blue, respectively. (B) Cartoon representation of the MipZ dimer structure with nine candidate amino acids possibly involved in DNA binding shown in purple. A surface representation of the structure is shown in the background. For clarity, the ATP molecules and Mg²⁺ ions sandwiched between the two subunits are not shown.

Figure 2.

In vitro characterization of MipZ variants with DNA-binding defects. (A) ATPase activities of purified wild-type MipZ and the indicated mutant derivatives. MipZ (6 μM) was incubated with ATP (1 mM), and the rate of hydrolysis was determined. Shown are the average turnover numbers (k_cat) (±SD) from at least three independent experiments. (B) Electrophoretic mobility shift assay analyzing the interaction of different MipZ variants with a linearized plasmid (pMCS-2). MipZ (WT) or the indicated variants (10 μM) were incubated with DNA (10 nM) and ATPγS (0.46 mM). Protein-DNA complexes were then separated from free DNA by standard agarose gel electrophoresis. (C) Biolayer interferometric analysis of the DNA-binding activity of different MipZ variants. A biotinylated dsDNA oligonucleotide (rand-biotin/rand-rev; 37.5 μM) was immobilized on the sensor surface and probed with MipZ (WT) or the indicated variants (4 μM) in the presence of ATPγS (1 mM). All analyses were performed at least three times, and representative results are shown.

Figure 3.

Global distribution of chromosomal MipZ binding sites. (A) ChIP-seq analysis of the distribution of different MipZ variants across the C. crescentus chromosome. Cells producing a wild-type (BH64), monomeric (K13A; BH100) or dimeric (D42A; BH99) MipZ-eYFP fusion in place of the native protein were fixed with formaldehyde and subjected to ChIP-seq analysis with an anti-GFP antibody. A representative 50 kb window of the chromosome is shown for visualization. Data were normalized using wild-type strain NA1000 as a reference. (B) Biolayer interferometric analysis of the interaction of MipZ with DNA of different GC content. Biotinylated dsDNA oligonucleotides with GC contents of 17% (ATrich-biotin/ATrich-rev), 56% (GC56-biotin/GC56-rev) or 79% (GCrich-biotin/GCrich-rev) were immobilized on biosensors and probed with wild-type MipZ (4 μM) in the presence of ATPγS (1 mM).

Figure 4.

Hydrogen/deuterium exchange (HDX) analysis of the MipZ-DNA interaction. (A) MipZ-D42A was incubated in deuterated buffer containing 1 mM ATPγS in the absence or presence of a 14 bp dsDNA oligonucleotide (ran14-up/ran14-lo). The heat plot shows the maximal differences in deuterium uptake between the DNA·MipZ-D42A complex and MipZ-D42A alone at different incubation times for a series of representatives peptides (see Dataset S2 for the full list of peptides). The color code is given on the right. All experiments were performed in the presence of ATPγS to further increase the stability of the dimer. The four different regions that are protected upon DNA binding to MipZ or ParA (see Figure 5) are indicated at the bottom. (B) Mapping of the maximum differences in HDX observed upon DNA binding onto the crystal structure of the MipZ dimer (PDB ID: 2XJ9). The color code is given at the bottom. The DNA-binding residues identified in the mutant screen are shown in red. A surface representation of the structure is shown in the background. For clarity, the ATP and Mg²⁺ ligands are not shown. (C) Electrostatic surface potential of the MipZ dimer. The color code is given at the bottom.

Figure 5.

Hydrogen/deuterium exchange (HDX) analysis of the C. crescentus ParA–DNA interaction. (A) ParA was incubated in deuterated buffer containing 1 mM ATPγS in the absence or presence of a 14 bp dsDNA oligonucleotide (ran14-up/ran14-lo). The heat plot shows the maximal differences in deuterium uptake between the DNA·ParA complex and MipZ alone at different incubation times for a series of representatives peptides (see Dataset S2 for the full list of peptides). The color code is given on the right. All experiments were performed in the presence of ATPγS to lock the protein in the dimeric state. The four different regions that are protected upon DNA binding to MipZ (see Figure 4) or ParA are indicated at the bottom. (B) Mapping of the maximum differences in HDX observed upon DNA binding onto a structural model of ParA, generated with HpParA (PDB ID: 6IUB) (39) as a template. A surface representation of the structure is shown in the background. For clarity, the ATP and Mg²⁺ ligands are not shown. (C) Electrostatic surface potential of the ParA dimer. The color code is given at the bottom.

Figure 6.

Variability of the DNA-binding regions of P-loop ATPases. (A) Comparison of the DNA-binding regions of MipZ and ParA from C. crescentus. Shown are surface representations of the dimer structures of MipZ (PDB ID: 2XJ9) and ParA (modeled as described in Figure 5B). Residues shown to be important for DNA binding are highlighted in purple. Short dsDNA oligonucleotides were modeled into the experimentally verified DNA-binding interfaces. The two subunits of each dimer are depicted in green and blue, respectively. (B) Alignment of the DNA-binding regions of the indicated P-loop ATPases (see Supplementary Figure S10 for details), corresponding to region R4 from Figures 4 and 5. Conserved residues are highlighted in blue. Residues proven to be involved in DNA-binding are colored orange. Residues predicted to have a role in DNA binding (see also Supplementary Figure S12) are shown in cyan. The corresponding secondary structural elements of MipZ from C. crescentus (13) are indicated at the bottom. See Supplementary Figure S10 for an alignment of the full-length sequences.