Structural and dynamic basis of DNA capture and translocation by mitochondrial Twinkle helicase

Abstract

Twinkle is a mitochondrial replicative helicase which can self-load onto and unwind mitochondrial DNA. Nearly 60 mutations on Twinkle have been linked to human mitochondrial diseases. Using cryo-electron microscopy (cryo-EM) and high-speed atomic force microscopy (HS-AFM), we obtained the atomic-resolution structure of a vertebrate Twinkle homolog with DNA and captured in real-time how Twinkle is self-loaded onto DNA. Our data highlight the important role of the non-catalytic N-terminal domain of Twinkle. The N-terminal domain directly contacts the C-terminal helicase domain, and the contact interface is a hotspot for disease-related mutations. Mutations at the interface destabilize Twinkle hexamer and reduce helicase activity. With HS-AFM, we observed that a highly dynamic Twinkle domain, which is likely to be the N-terminal domain, can protrude ∼5 nm to transiently capture nearby DNA and initialize Twinkle loading onto DNA. Moreover, structural analysis and subunit doping experiments suggest that Twinkle hydrolyzes ATP stochastically, which is distinct from related helicases from bacteriophages.

Figure 1.

Overall structure of LcTwinkle. (A) Domain structures of Hs and LcTwinkle. The disease-related Twinkle mutations are marked as red bars. MTS, mitochondrial targeting sequence; Zn, zinc-binding domain; Pri, primase; Hel, helicase; C-tail, C-terminal tail. (B) E325Q LcTwinkle–DNA complex formation revealed by size exclusion chromatography. (C) Helicase activity of WT and E325Q LcTwinkle. Twinkle hexamers at 25–400 nM were used for the assays. (D, E) Two views of the LcTwinkle–DNA complex. The subunits A to F are color-coded from blue to red, following their order on DNA. DNA is colored orange. The NTDs from subunits E and F are disordered and their positions are marked by semi-transparent triangles. The disordered part of the N–C loop in subunit F is indicated by a dashed line. (F) Structure of LcTwinkle–DNA₂. The mobile subunit F and the structural gaps between F and its neighboring subunits are highlighted. (G, H) Structures of the apo LcTwinkle hexamer (G) and heptamer (H). The structural gaps in the oligomers are marked.

Figure 2.

ATP- and DNA-binding interfaces of LcTwinkle. (A) ATP- and DNA-binding sites in the LcTwinkle–DNA complex. The CTDs are shown as transparent surfaces and structural elements involved in ATP and DNA binding are shown in the cartoon. (B) Zoom-in view of the ATP- and DNA-binding sites between subunits C and D. Key residues are shown as sticks. WA, WB, D1 and D2 stand for Walker A, Walker B, LoopD1 and LoopD2 motifs, respectively. The position of the predicted water molecule for the nucleophilic attack is highlighted by a dotted circle. (C) Helicase activity of the WT and the active site mutants of LcTwinkle. (D) Fluorescence anisotropy DNA binding of the WT and the active site mutants of LcTwinkle. (E) DNA-binding loops from different subunits of the LcTwinkle–DNA complex take the same configuration. (F) Structural alignments of ATP-binding sites across the hexamer suggest that they are superimposable. The five ATP-binding sites were extracted from the LcTwinkle–DNA complex, and subunits at the 5′ side of the DNA in each dimer are aligned. The 5′-side subunit is shown as gray surfaces, and the active site residues on the 3′ side subunits are drawn as cartoons and sticks. (G, H) Subunit doping experiments of ATPase (G) and helicase (H) activities with WT and E325Q LcTwinkle. Mixed LcTwinkle at 0.2 μM was used for the helicase assay and 0.05 μM for the ATPase assay.

Figure 3.

Disease-related mutations on LcTwinkle. (A) Disease-related mutations on the N–C interface and the N–C linker. Two adjacent subunits C (green) and D (yellow) are selected to illustrate the positions of the disease-related mutations. The mutations on subunit C are highlighted in blue and the mutations on subunit D are highlighted in orange. The corresponding human disease-related mutations are indicated in parentheses. The locations of each panel relative to the overall structure are indicated on the small diagrams in the top right corner in (A). (B) Disease-related mutations on the CTD. (C) Disease-related mutations on the NTD.

Figure 4.

Real-time HS-AFM imaging of LcTwinkle shows DNA capture by N-protrusion and at the central channel. (A) LcTwinkle switches between open (blue triangle) and closed conformations in the absence of DNA (panels I and II). Nearby DNA induces LcTwinkle N-protrusion (panels III and IV). The N-protrusion is highlighted by green (N-protrusion on the x- and y-plane) and yellow (N-protrusion characterized by an increased height; brighter z-scale) arrows. See also Supplementary Video S2. xy scale bar = 20 nm. (B, C) LcTwinkle conformers showing N-protrusions that capture DNA in the vicinity (distance to DNA <20 nm). See also Supplementary Videos S3 and S5. The open gap in LcTwinkle is indicated as blue triangles in panels I and II in (C). xy scale bar = 20 nm. (D) DNA capture and release revealed by HS-AFM imaging. The green and purple arrows highlight DNA binding at the N-protrusion and the CTD central channel, respectively. See also see Supplementary Video S6. xy scale bar = 50 nm. (E) Box plots showing the distance distribution between DNA and the protruding subunit on LcTwinkle (10.7 ± 3.7 nm, n = 27) and the protruding length of the LcTwinkle conformers (4.9 ± 1.8 nm, n = 27). The reported mean and standard deviation are from HS-AFM images collected from three independent experiments. (F) Two views of the electrostatic surfaces of an LcTwinkle subunit. The negatively charged surfaces are colored blue, and the positively charged surfaces are colored red.

Figure 5.

Model of LcTwinkle loading and translocation. Twinkle in the absence of DNA is prone to ring opening (State I). When DNA is nearby, the positively charged NTD captures the negatively charged DNA over long distances and the CTD hexamer opens further for DNA loading (State II). After loading, Twinkle forms a lock-washer-shaped hexamer with the NTDs attaching to the side of the CTDs and the DNA binding to the CTD central channel (State III). ATP hydrolysis is stochastic in Twinkle. When ATP is hydrolyzed at the subunit interfaces near the DNA 5′ end (State IV and State VI), one or two subunits dissociate from their neighbor and travel to the other end of the DNA (State V and State VII). However, when ATP is hydrolyzed in the middle or close to the DNA 3′ end (State VIII), the translocation is unfavored, leading to a futile cycle of ATP hydrolysis.