Cascading MutS and MutL sliding clamps control DNA diffusion to activate mismatch repair

Abstract

Mismatched nucleotides arise from polymerase misincorporation errors, recombination between heteroallelic parents and chemical or physical DNA damage. Highly conserved MutS (MSH) and MutL (MLH/PMS) homologues initiate mismatch repair and, in higher eukaryotes, act as DNA damage sensors that can trigger apoptosis. Defects in human mismatch repair genes cause Lynch syndrome or hereditary non-polyposis colorectal cancer and 10-40% of related sporadic tumours. However, the collaborative mechanics of MSH and MLH/PMS proteins have not been resolved in any organism. We visualized Escherichia coli (Ec) ensemble mismatch repair and confirmed that EcMutS mismatch recognition results in the formation of stable ATP-bound sliding clamps that randomly diffuse along the DNA with intermittent backbone contact. The EcMutS sliding clamps act as a platform to recruit EcMutL onto the mismatched DNA, forming an EcMutS-EcMutL search complex that then closely follows the DNA backbone. ATP binding by EcMutL establishes a second long-lived DNA clamp that oscillates between the principal EcMutS-EcMutL search complex and unrestricted EcMutS and EcMutL sliding clamps. The EcMutH endonuclease that targets mismatch repair excision only binds clamped EcMutL, increasing its DNA association kinetics by more than 1,000-fold. The assembly of an EcMutS-EcMutL-EcMutH search complex illustrates how sequential stable sliding clamps can modulate one-dimensional diffusion mechanics along the DNA to direct mismatch repair.

Extended Data Figure 1. The construction of mismatched DNA used in single-molecule total internal reflection fluorescence (smTIRF) microscopy

a, A schematic illustration for the construction of a 17.3-kb mismatched DNA. L or R (blue) indicates the orientation of the DNA relative to the L and R cos end of λ-phage DNA. P (red) indicates the 5′-phosphate of the DNA. b, A schematic illustration of 17.3-kb mismatched DNA observation by prism-based smTIRF microscopy. c, Representative mismatched DNA visualized by smTIRF microscopy in the absence of flow. The DNA was stained with Sytox Orange and a 40 × 85 µm field of view is shown. d, A schematic illustration of the DNA length determination. e, The length distribution of the mismatched DNA observed by smTIRF microscopy. Gaussian fit of the data are shown along with the mean ± s.d.

Extended Data Figure 2. The fluorophore-labelled E. coli MMR proteins used in these studies and the formation of an EcMutS sliding clamp on DNA

a, Coomassie stained (top) and fluorescent (bottom) SDS–PAGE gels of labelled MMR proteins. For gel source data, see Supplementary Fig. 1. b, A crystal structure of the N-terminal domain of EcMutL bound to AMP-PNP (top) and magnification of the binding domain (bottom; PDB ID: 1B63). AMP-PNP is shown in green and Arg-95 (R95) is shown in magenta. c, An illustration of the kymograph construction of three separate EcMutS sliding clamps on a single mismatched DNA. d, The distribution of diffusion coefficients for the EcMutS sliding clamp. The data were fit to a Gaussian with the mean ± s.d. e, The distribution of dwell times (mean ± s.e.m.) for the EcMutS sliding clamp.

Extended Data Figure 3. EcMutL does not bind DNA in physiological ionic conditions

a, b, Representative kymographs and dwell-times (mean ± s.e.m.) for EcMutL binding to a mismatched DNA at various conditions. c–h, The distributions of dwell times (mean ± s.e.m.) for EcMutL on mismatched DNA at different biochemical conditions as indicated.

Extended Data Figure 4. Representative kymographs of EcMutS–EcMutL complex and EcMutL particles

a, Representative kymographs showing the loading of EcMutL (green) on DNA by EcMutS sliding clamp (red). b, Representative kymographs of a diffusing EcMutS–EcMutL complex (merged channels) and a non-diffusing EcMutS (red) in the same field of view. The static kymograph of a non-diffusing EcMutS indicates that the change in protein position caused by microscope stage drifting is negligible. c, Representative kymographs of oscillating EcMutS–EcMutL complexes. Two channels (red, EcMutS; green, EcMutL) were merged. d, The distribution of the association times (mean ± s.e.m.) for EcMutS–EcMutL complexes during the oscillating phase. e, Representative kymographs of fast-diffusing EcMutL dissociation from EcMutS–EcMutL complexes. Two channels (red, EcMutS; green, EcMutL) were merged.

Extended Data Figure 5. EcMutH lifetime on DNA and diffusion coefficient of EcMutS–EcMutL–EcMutH and/or EcMutL–EcMutH complex

a, A schematic illustration of EcMutH endonuclease assay (left) and a comparison of labelled or unlabelled EcMutH endonuclease activities (right). For gel source data, see Supplementary Fig. 1. b, c, Representative kymographs and dwell times (mean ± s.e.m.) showing EcMutH on a single mismatched DNA under various ionic and magnesium conditions. d–g, The distributions of dwell times (mean ± s.e.m.) for EcMutH on a single mismatched DNA at different biochemical conditions as indicated. h, Box plots showing D for oscillating (EcMutS)–EcMutLCy3–EcMutHAF647 ((S)–L–H) complex; the established EcMutSAF555–EcMutL–EcMutHAF647 complex (S–L–H); and free EcMutLCy3–EcMutHAF647 complex (L–H) at 100 mM NaCl. i, Box plots showing D for established EcMutS–EcMutL–EcMutH complex at different NaCl concentrations. Two-sample t-test showed no significant difference between diffusion coefficients (P > 0.1). j, Box plots showing D for free EcMutL–EcMutH complex at different NaCl concentrations. Two-sample t-test showed significant differences between diffusion coefficients (P < 0.05). k, Top left, representative kymographs showing FRET between C-terminal AF555-labelled EcMutS and N-terminal AF647-labelled EcMutL (N-ter). Bottom, fluorescent intensities of EcMutS–AF555 (donor, green), EcMutL–AF647 (acceptor, red) and FRET (blue) between them when only the green laser was used for illumination. Right, a schematic illustration of kymographs. Experimental FRET measure (E_{EcMutS–EcMutL} = 0.48 ± 0.05; mean ± s.d.) and theoretical FRET (E_{EcMutS–EcMutL} = 0.56) based on crosslink structure appeared comparable. n = number of molecules throughout.

Extended Data Figure 6. The interactions and kinetic properties of the molecular switch/sliding clamp mechanism for E. coli MMR

a, Illustration of the kinetics and diffusion properties of EcMutS. b, Illustration of the kinetics and diffusion of EcMutS with EcMutL. c, Illustration of the kinetics and diffusion of EcMutS, EcMutL and EcMutH. d, Oscillation dynamics of the EcMutS–EcMutL–EcMutH complex (see main text). Coil, 1D-diffusion search along the backbone; dashed straight arrow, rotation-independent 1D-diffusion; black curved arrow, binding; red curved arrow, oscillating complex; dashed curved arrow, binding-dissociation; binding times and ATP (•) are indicated.

Figure 1. The formation of an EcMutS–EcMutL complex alters the diffusion properties of EcMutS

a, Representative kymographs and illustration showing an EcMutL loaded by an EcMutS sliding clamp. Blue line indicates the mismatch position. b, The distribution of dwell times for the EcMutS–EcMutL complex (τ_{on•EcMutS–EcMutL}; mean ± s.e.m.). c, Box plots of D for the EcMutS–EcMutL complex at different NaCl concentrations (n = number of molecules; Methods). d, The distribution of the starting positions for EcMutS (top) or EcMutS–EcMutL complexes on DNA (bottom). There are two possible orientations of the mismatched DNA with mismatch position (blue star; middle panel). Diamonds in top and bottom panels represent individual starting events. Gaussian fit to the top panel distributions is shown as black lines with the mean ± s.d.

Figure 2. The formation of an oscillating EcMutS–EcMutL complex and fast-diffusing EcMutL

a, Representative kymographs and illustrations showing the different types of EcMutS–EcMutL complex dissociations (coloured letters). The frequency (%) of each dissociation type is shown (right, coloured numbers). b, Representative kymographs and illustration showing an oscillating EcMutS–EcMutL complex (white arrowheads indicates dissociation events). c, Dissociation time distribution (τ_{off•EcMutS↔EcMutL}; mean ± s.e.m.) for the oscillating EcMutS–EcMutL complex. d, Illustration of oscillating EcMutS–EcMutL complex with lifetimes and calculated diffusion distances. Oscillations are indicated for illustration only and should be stochastic. e, Representative kymographs and illustration showing the dissociation and fast diffusion of EcMutL from an EcMutS–EcMutL complex. f, Dwell time distribution (mean ± s.e.m.) of fast-diffusing EcMutL on the mismatched DNA. g, Box plots of D for fast-diffusing EcMutL at different NaCl concentrations. h, The frequency of EcMutS–EcMutL complex and fast-diffusing EcMutL under various conditions (mean ± s.d.); n = number of DNA molecules.

Figure 3. ATP binding by EcMutL results in formation of a ring-like sliding clamp

a, Representative kymographs showing the formation of two separate EcMutS–EcMutL(R95F) complexes on a mismatched DNA. b, Dwell time distribution for the EcMutS–EcMutL(R95F) complex (τ_{on•EcMutS–EcMutL(R95F)}; mean ± s.e.m.). c, Box plots of D for the EcMutS sliding clamp (S); EcMutS–EcMutL complex (S–L); EcMutS–EcMutL(R95F) complex (S–L(R95F)); and EcMutL particle (L) at 100 mM NaCl. d, The frequency of EcMutS–EcMutL complex (blue) and fast-diffusing EcMutL (grey) with wild-type EcMutL or EcMutL(R95F) (mean ± s.d.). e, Pie charts showing the distributions of dissociation types for EcMutS–EcMutL or EcMutS–EcMutL(R95F) complexes. Frequency and dissociation types are indicated within the pie chart (see Fig. 2a). f, Illustration of ATP and AMP-PNP pre-incubation studies. g, The frequency of EcMutS–EcMutL complex and fast-diffusing EcMutL under various EcMutL or EcMutL(R95F) pre-incubation conditions (mean ± s.d.). h, Illustration showing AMP-PNP induced EcMutL ring-like clamp closure (top) that does not affect EcMutL(R95F) (bottom). i, Representative kymographs showing the absence of EcMutS–EcMutL complexes and the presence of EcMutS–EcMutL(R95F) complexes following pre-incubation of EcMutL or EcMutL(R95F) with AMP-PNP.

Figure 4. EcMutH binds EcMutL sliding clamps

a, The frequency of EcMutH–AF647 on the mismatched DNA in the presence of other MMR components (mean ± s.d.). b, Representative kymographs of EcMutH–AF647 co-localization and diffusion with EcMutL–Cy3 (and unlabelled EcMutS) on a single mismatched DNA. c, Representative kymographs of EcMutL(R95F)–Cy3 (and unlabelled EcMutS) diffusion on a single mismatched DNA. Note the absence of EcMutH–AF647. d, Representative kymographs of EcMutH–AF647 co-localization and diffusion with EcMutS–AF555 (and unlabelled EcMutL) on a single mismatched DNA. Arrowheads indicate association of EcMutS with EcMutL–EcMutH. e, Representative kymographs of EcMutH–AF647 co-localization and diffusion with EcMutL–Cy3 on a single mismatched DNA following the dissociation of EcMutS. Arrowheads indicate association of EcMutH with EcMutL sliding clamp. f, Dwell time distribution of the EcMutL–EcMutH complex (τ_{on•EcMutL–EcMutH}; mean ± s.e.m.).