Mechanism of DNA capture by the MukBEF SMC complex and its inhibition by a viral DNA mimic

Abstract

Ring-like structural maintenance of chromosome (SMC) complexes are crucial for genome organization and operate through mechanisms of DNA entrapment and loop extrusion. Here, we explore the DNA loading process of the bacterial SMC complex MukBEF. Using cryoelectron microscopy (cryo-EM), we demonstrate that ATP binding opens one of MukBEF's three potential DNA entry gates, exposing a DNA capture site that positions DNA at the open neck gate. We discover that the gp5.9 protein of bacteriophage T7 blocks this capture site by DNA mimicry, thereby preventing DNA loading and inactivating MukBEF. We propose a comprehensive and unidirectional loading mechanism in which DNA is first captured at the complex's periphery and then ingested through the DNA entry gate, powered by a single cycle of ATP hydrolysis. These findings illuminate a fundamental aspect of how ubiquitous DNA organizers are primed for genome maintenance and demonstrate how this process can be disrupted by viruses.

Keywords: DNA mimics; MukBEF; SMC complexes; Wadjet; bacteriophages; cohesin; condensin; cryo-EM.

Graphical abstract

Figure 1. Reconstitution of DNA loading

(A) Phylogenetic tree of SMC proteins inferred from chained alignments of head and hinge regions. (B) Architecture of MukBEF (left) and simplified geometry of the complexes indicating DNA entrapment (right). (C) Concept of the in vitro loading assay. MukBEF^6C is loaded onto plasmid DNA in the presence of ATP, then gates are closed by BMOE-mediated cysteine crosslinking, and protein/DNA catenanes are probed after SDS denaturation. (D) BMOE crosslinking of P. thracensis MukBEF^6C containing cysteine residues in the three gate interfaces. A Coomassie-stained SDS-PAGE gel shows cross-linked products. (E) Loading time course of MukBEF^6C on negatively supercoiled DNA (pFB527) in the presence of 1 mM ATP and an ATP regeneration system. Reactions were terminated by BMOE crosslinking at the indicated times; samples were denatured by SDS treatment and resolved by agarose gel electrophoresis. (F) Loading reaction as in (E) after 60 min, using different combinations of ATP and MukBEF^6C or the ATP-hydrolysis-deficient E1407Q (EQ) mutant complex. ATP was used at 5 mM without a regeneration system. (G) Loading reactions in the presence of topoisomerases. Reactions were performed with 5 mM ATP as in (F), but an additional 30 mM NaCl was included in the reaction buffer, and DNA was nicked after BMOE treatment to adjust electrophoretic mobility. The experiment used pUC19 as the DNA substrate. (H) Loading on relaxed DNA substrates. DNA was relaxed by Topo I or nicking, purified, and loading was performed with 5 mM ATP as in (F). Samples were nicked after BMOE treatment to make electrophoretic mobility comparable. The experiment used pUC19 as the DNA substrate. See also Figure S1 and Data S1.

Figure 2. Mechanism of gate opening and DNA capture

(A) Structure of the open-gate state. Cryo-EM density of the MukBEF monomer in the nucleotide-bound form (left; PDB: 9GM7) and a focused refinement of the head module with open neck gate (right; PDB: 9GM8). (B) Comparison of apo (left; PDB: 7NYY) and open-gate state (right; PDB: 9GM8). Heads engage upon nucleotide binding, resulting in a swing-out of the MukF MD. (C) Comparison of the engaged MukB heads in the open-gate state (top; PDB: 9GM8) and the DNA-clamped unloading state (bottom; PDB: 7NYW). Binding of MukE and DNA to the top of the heads is mutually exclusive. (D) Structure of the DNA capture state. Focused classification of (A) reveals DNA captured at the open gate. Cryo-EM density (left) and cartoon model (right; PDB: 9GM9) are shown. (E) The DNA capture state in the context of the MukBEF dimer. Cryo-EM density of the dimer (blurred with a σ = 22 Å Gaussian filter to make low-density regions interpretable), cartoon model representation (left; PDB: 9GMA), and close up of the dimeric DNA-capture interface (right) are shown. See also Figure S2.

Figure 3. Discovery of a viral MukBEF inhibitor

(A) Expression of gp5.9 is toxic. E. coli cells were transformed with a kanamycin-selectable empty vector control or an equivalent construct containing gp5.9 under an arabinose-inducible promoter. Transformation reactions were plated on LB plus kanamycin with or without arabinose. Plates were incubated at 37°C. (B) As in (A) but using a ΔrecB background. (C) TMT-MS analysis of gp5.9^FLAG pull-downs using pooled signal from WT and ΔrecB extracts. A volcano plot of significance versus pull-down over mock extract is shown, highlighting gp5.9, MukE and MukF levels. (D) Morphology of cells expressing gp5.9. Cells carrying the indicated constructs were grown for 3 h in LB media with or without arabinose, fixed with formal-dehyde, stained with DAPI, and imaged by combined phase contrast (grayscale) and fluorescence (red) microscopy. (E) Analysis of the DAPI intensity distribution of cells from the experiment shown in (D). Expression of gp5.9 causes a relative increase in cells with altered DNA content. (F) Pull-down of recombinant MukBEF or MukE with gp5.9^FLAG. Anti-FLAG beads were charged with extract containing or lacking gp5.9^FLAG, then incubated with recombinant MukBEF proteins, eluted with FLAG peptide, and analyzed by SDS-PAGE and Coomassie staining. (G) Quantification of pull-downs as in (F), normalizing the indicated band intensities for the corresponding gp5.9^FLAG signal. Band intensities for MukB, MukE, and MukF are shown, comparing the signal between MukBEF complex and single subunit pull-downs. Mean ± SD from n = 3 replicates. (H) SEC analysis of mixtures of gp5.9 and E. coli MukEF (top) and P. thracensis MukEF (bottom), respectively. Elution fractions were analyzed by SDS-PAGE and Coomassie staining. gp5.9 forms a stable complex with E. coli MukEF, but not with P. thracensis MukEF. See also Figure S3 and Data S2.

Figure 4. gp5.9 binds the DNA capture site and inhibits loading

(A) Structure of the gp5.9/MukEF interface. A cartoon of the complex analyzed (left) and cryo-EM density from a focused refinement (right) is shown. (B) DNA capture and gp5.9 binding are mutually exclusive. The cartoon representation of (A) is shown (PDB: 9GMD) with DNA from the superimposed capture state structure (PDB: 9GM9). (C) DNA entrapment assay in the presence of gp5.9 as in Figure 1H using nicked plasmid (pUC19). The molar ratio of gp5.9 to MukBEF^6C monomer sites is indicated. E. coli MukBEF^6C is sensitive to gp5.9, whereas P. thracensis MukBEF^6C is not. (D) As in (C), but gp5.9 was added 60 min after reaction start. Samples were then treated with BMOE at the indicated timepoints after addition of gp5.9. See also Figure S4.

Figure 5. Mechanism of DNA entry into MukBEF

(A) Comparison of the DNA capture state (left) with the E. coli Wadjet I holding state (right; PDB: 8Q72), and a model of the equivalent MukBEF holding state (middle). The latter was composed from DNA-bound MukEF (PDB: 9GM9), the apo MukB/MukF interface (PDB: 7NYY), and a remodeled MukF linker. Co-ordinates were superimposed on the DNA. The state transition from capture to holding state requires a rotation of MukB and the MukF linker around the DNA. (B) Comparison of MukF between capture and holding state. The linker wraps around DNA upon the proposed state transition. (C) Stand off and rotate model for transition from the capture to the holding state and gate closure. MukB releases from MukE upon ATP hydrolysis and rotates around the DNA to close the neck gate. (D) Implications of the stand off and rotate model for loading on relaxed (left) and supercoiled (right) DNA. Rotation around a relaxed double-strand is easier than in the context of a compact plectoneme and is consistent with the inhibition of loading on supercoiled DNA. (E) Model of the MukBEF activity cycle. The state of the neck gate and entrapment of DNA are indicated, and PDB IDs that support the states are shown. Parentheses around IDs indicate partial or homologous structures. Three-dimensional models for the tentative states are available in Data S3. See also Figure S5, Data S3, and Video S1.