Article | Published:

Structure of the eukaryotic MCM complex at 3.8 Å

Nature volume 524, pages 186191 (13 August 2015) | Download Citation

Abstract

DNA replication in eukaryotes is strictly regulated by several mechanisms. A central step in this replication is the assembly of the heterohexameric minichromosome maintenance (MCM2–7) helicase complex at replication origins during G1 phase as an inactive double hexamer. Here, using cryo-electron microscopy, we report a near-atomic structure of the MCM2–7 double hexamer purified from yeast G1 chromatin. Our structure shows that two single hexamers, arranged in a tilted and twisted fashion through interdigitated amino-terminal domain interactions, form a kinked central channel. Four constricted rings consisting of conserved interior β-hairpins from the two single hexamers create a narrow passageway that tightly fits duplex DNA. This narrow passageway, reinforced by the offset of the two single hexamers at the double hexamer interface, is flanked by two pairs of gate-forming subunits, MCM2 and MCM5. These unusual features of the twisted and tilted single hexamers suggest a concerted mechanism for the melting of origin DNA that requires structural deformation of the intervening DNA.

Main

For DNA to be replicated, two strands of the duplex DNA must be separated so that each can serve as a template for the synthesis of daughter strands. In both prokaryotes and eukaryotes, DNA unwinding is carried out by specialized helicases that encircle and translocate along one of the DNA strands1. However, the mechanisms for the initial melting or unwinding of origin DNA are markedly different1,2. In bacteria, the origin recognition protein DnaA initiates origin melting and then recruits hexameric helicase DnaB3 to unwind DNA by translocation on the lagging strand in the 5′–3′ direction2. By contrast, in eukaryotes, the origin recognition complex first binds replication origins without effecting initial origin melting, and then loads two MCM2–7 (ref. 4) single hexamers with the help of the DNA replication proteins Cdc6 and Cdt1 onto double-stranded origin DNA to form a double hexamer5,6. This inactive assembly of proteins is known as the pre-replicative complex. Subsequent activation of MCM2–7 complex takes place in the S phase, and requires several factors and cell-cycle-specific kinases7,8,9,10,11, resulting in the formation of an active replicative helicase, the Cdc45–MCM2–7–GINS (CMG) complex12. CMG translocates along the leading strand in a 3′–5′ direction to unwind duplex DNA (steric exclusion model)13,14,15,16,17. However, how origin DNA is melted before active replication elongation is unknown. This process probably requires the reconfiguration of MCM2–7 helicase, a complex molecular motor that has defied high-resolution structural analysis for decades. At present, much of the mechanistic insights came from low-resolution structures of the MCM2–7 complex in functional forms from different species5,16,18,19,20,21,22, as well as crystal structures of simpler archaeal versions, in non-functional oligomers23,24,25, truncations26,27,28,29 or a chimaeric hexamer30.

In this study, we purified the endogenous MCM2–7 double hexamer from G1 chromatin of budding yeast (Extended Data Fig. 1a–c), and determined its cryo-EM structure at an overall resolution of 3.8 Å (gold-standard Fourier shell correlation 0.143 criteria) (Extended Data Fig. 1k). Except for peripheral regions, the core of the map is better than 3.5 Å (Extended Data Fig. 1j), which enabled atomic model building for 80% of sequences of this 1.2-megadalton (MDa) complex. Our structure reveals rich details for the organization of this large complex, and informs many functional aspects of this replicative helicase, particularly in the initial origin melting.

Overall structure and domain organization

A first glimpse of the structure is the tilted arrangement of two single hexamers, with a 14° wedge in between (Fig. 1a–c), a feature already noticed from low-resolution data of the MCM2–7 double hexamer5,18 and the SV40 large tumour antigen31. The two single hexamers also have a twisted arrangement (Fig. 1a, side panel), resulting in the misalignment of two hexamer axes. The quality of the density map allowed an independent assignment of six subunits, being 2-6-4-7-3-5 (viewed from the carboxy-terminal domain (CTD) ring) (Fig. 1d), consistent with the well-established model21,32,33. Notably, when viewed from the single hexamer axis, the gravity centres of three major structural components—NTD-A (A subdomain of N-terminal domain (NTD)), oligonucleotide/oligosaccharide-binding (OB)-fold (C subdomain of NTD), and CTD—fall onto three eccentric circles (Fig. 1d). While the circles of NTD-As and OBs are nearly concentric, the CTD circle exhibits apparent rotational and translational offsets, indicating a relative shift and twist between the NTD and CTD rings within the single hexamer. Also, the NTD-As and OBs for each subunit are nearly vertically arranged (as indicated by their centres falling along on the same radial lines), with slight rotations in opposite directions for MCM4 and MCM5. Notably, the CTDs of all six subunits have left-handed twists to varying extents (Fig. 1d, f) with respect to their OBs and NTDs. Furthermore, the distances between neighbouring CTDs are different (Fig. 1e), showing a 4-Å difference between tightly (4:7, 7:3 and 5:2) and loosely (6:4, 3:5 and 2:6) packed groups. While six OBs form a plane perpendicular to the hexamer axis, the CTDs and NTD-As display marked axial variations (Fig. 1f, g).

Figure 1: Overall structure and domain organization of the MCM2–7 double hexamer.
Figure 1

ac, Side-views of the cryo-EM density map superimposed with the atomic model. Unsharpened map (a) is displayed from the two-fold axis, and sharpened map (b, c) displayed with indicated rotations relative to a along the cylinder axis. The side panels of a and b illustrate the tilted arrangement of the two single hexamers. d, f, Top (d) and side (f) views of the organization of the indicated structural domains. Small coloured balls denote gravity centres (see Methods) of these domains. e, Distances between the gravity centres of adjacent CTDs. g, Radial projection of f. hl, Side and top views of the segmented maps (sharpened) of ZFs at the hexamer interface.

Inter-hexamer interface

The head-to-head stacking of the two hexamers is largely mediated by their zinc-fingers (ZFs; B subdomain of NTD), as expected from previous studies26,34,35. Notably, consistent with sequence analysis36 (Extended Data Fig. 2), the ZF of MCM3 is a degenerate version without zinc binding (Extended Data Fig. 3a–c). Twelve ZFs arrange into two stacked rings at the interface (Fig. 1h–l), with an apparent centre shift (Fig. 1l and Supplementary Video 1). Inter-ZF interactions are versatile, displaying completely different patterns at opposite sides of the wedged interface (Fig. 1b, subpanel). Although ZFs are more horizontally arranged at the thin 3-5-7 edge (Fig. 1h), they are nearly vertical at the thick 2-4-6 edge (Fig. 1j). ZF interactions are largely from their polar residues, dominated by two pairs of ZF5:ZF3′ (Fig. 1h) and two pairs of ZF6:ZF2′ (Fig. 1j), as measured in buried surfaces (Extended Data Table 1a). Different ZF orientations at the hexamer interface perfectly explain the observed tilt and twist between the two single hexamers, because this unique arrangement would enable comprehensive close contacts for both edges and leads to the stabilization of the double hexamer.

Eukaryotic MCM proteins distinguish themselves from archaeal counterparts by many subunit-specific sequence extensions at their N and C termini (NTE and CTE, respectively) and insertions within functional domains. Both MCM4 and MCM6 have a very long linker between their OBs and CTDs (Extended Data Fig. 4a), which could be the underlying basis of the observed twist between NTD and CTD rings in the single hexamer (Fig. 1d). Notably, many sequence insertions and extensions also markedly contribute to the double hexamer stabilization (Fig. 2). For example, an insertion located on the β-turn of the OB from MCM6 (Extended Data Fig. 4h) interacts with the ZF of MCM2 on the other hexamer (Fig. 2e). The most unique inter-hexamer interactions involve MCM3, MCM5 and MCM7. Compared with archaeal MCMs, they have longer sequences at their N termini (Extended Data Fig. 2), which form extended strands or loops (Extended Data Fig. 4e, g and i). MCM7 also has a long insertion (70 residues) at its NTD-A (N-terminal insertion (NTI)) (Extended Data Fig. 4a), folding into a helix–turn–helix motif (Extended Data Fig. 4i). The N terminus of MCM5 extends into the space between ZFs of MCM3 and MCM7 from the other hexamer, and forms interactions with β-strands of both ZFs (Fig. 2f and Supplementary Video 2). Furthermore, the long NTI of MCM7 extends towards the opposite MCM5 and interacts with its NTD-A (Fig. 2c, g). The N terminus of MCM7 also interacts with the N terminus of MCM3 from the other hexamer (Fig. 2h). On the basis of the calculated buried surfaces for the above interfaces (Fig. 2e–h), the contribution of NTIs and NTEs to the double hexamer stability is even greater than the ZF interactions (Extended Data Table 1a). Importantly, most insertions and extensions involved in inter-hexamer stabilization (Fig. 2) are conserved in higher eukaryotes, suggesting a universal importance of these eukaryotic-specific sequences.

Figure 2: Inter-hexamer interactions contributed by NTEs and NTIs.
Figure 2

ad, Side views of the MCM2–7 double hexamer, with indicated rotations around the cylinder axis. Atomic structure is superimposed with the unsharpened map. The sequence elements involved in inter-hexamer interactions are highlighted in blue and red representing single hexamer 1 (SH1) and SH2, respectively. eh, Zoomed-in views of the boxed regions from bd. Buried areas (Å2) of these interfaces in eh are labelled. BS, buried surface.

MCM2, 4 and 6 have very long NTEs, which are targets of cellular signalling kinases7,8,9,11. These NTEs are highly disordered in our structure, and their involvement in inter-hexamer interaction is unknown.

Intersubunit interaction

The intersubunit interactions are very similar, and can be categorized into three tiers based on their axial locations (Extended Data Fig. 5a–c). The first one is between two contacting CTDs (ATPase domains), largely composed of hydrophobic interactions, as exemplified by a tight stacking between two surface-exposed helices from two respective CTDs (Extended Data Fig. 5a). The second tier, on the neck region of the hexamer, involves four conserved hairpins or loops from two adjacent subunits, including allosteric communication loop (ACL) and helix-2-insert (H2I) of the first subunit, and H2I and presensor 1 β-hairpin (PS1-HP) of the flanking second subunit (Extended Data Fig. 5b). Atomic interactions at the neck interfaces are versatile, involving different residues from these loops. However, a large proportion of them are polar residues, indicating electrostatic or hydrogen-bonding interactions dominate these interfaces. The third tier, contributed by the ZF of one subunit and two loops from the OB of a flanking subunit (Extended Data Fig. 5c), is largely composed of hydrophobic interactions between respective β-strands or loops (for example, see Extended Data Fig. 6d). Perturbation of this interface by an MCM4 mutation (Phe391Ile) causes pre-replicative complex assembly defects in yeast and mammary carcinoma in mouse37. In addition, a mutation on MCM5 NTD-A (Phe83Leu) that results in Dbf4-dependent kinase (DDK)-independent activation38 is close to this interface.

Intersubunit interactions are further enhanced by the NTIs of MCM3, 5, 6 and 7, which contact the NTD-As of their neighbouring MCM7, 3, 2 and 4, respectively (Extended Data Fig. 5d–i). Compared with other pairs, the interface of MCM5–MCM2 is without NTI involvement, a feature that may facilitate the gap opening observed between them during hexamer loading and activation16,19,20,21. At lower contour levels, four CTEs containing the winged-helix DNA binding motif could be identified for MCM4, 5, 6 and 7 (Extended Data Fig. 3d, e). The flexibility of these winged-helix-containing CTEs suggests that they are not involved in intersubunit interaction, contrasting the role of winged-helix motifs in the origin recognition complex structure39.

The buried surfaces of the six subunit interfaces are sharply different (Extended Data Table 1a), with the smallest at 2:6, rather than at the gate-forming 2:5 interface. The weak 2:6 interface gives rise to a unique side channel (13 Å), enough to accommodate single-stranded DNA (ssDNA) (Extended Data Fig. 7 and Supplementary Video 1), in contrast to archaeal MCM structure with six side channels23. A ssDNA extrusion model has been proposed for the function of side channels in DNA unwinding23,35,36. However, unwinding studies13,14,15,17 generally conflict with this model. A definitive function for this unique 2:6 side channel remains to be examined.

ATPase active centres

Remarkable conformational differences lie at the six ATPase centres of the CTD ring. Comparisons of them indicate that two active centres, 2:6 and 5:2, are apparent outliers. Their ATP-binding pockets are less compact, with sensor elements (sensors 2 and 3, and arginine finger) in MCM2 and MCM6, respectively, considerably shifted away from nucleotides, and the displacements are as large as 4–5 Å measured by Cα atoms of sensor 3 residues (Fig. 3a, b). Further analysis was done by comparisons of representative centres from the compact and loose groups with an active ATPase centre from papillomavirus E1 crystal structure40 or an inactive one from an archaeal MCM structure30. Indeed, while the conformational differences of the four compact centres relative to that of E1 are small (for example, dimers of 7:3 and 4:7; Extended Data Fig. 8h, i), the loose ones display sharp differences from that of E1 (Fig. 3c, d). Moreover, the nucleotide occupancies at the centres of 3:5 and 6:4 dimers are comparatively low (Extended Data Fig. 9), consistent with their reported nearly null ATPase activities33. On the basis of the active centre arrangement and nucleotide occupancy, it appears that only dimers of 7:3 and 4:7 are active. This observation agrees with the extremely low ATPase activity observed in MCM2–7 as a double hexamer18, and the reported activity of 7:3 dimer comparable to that of the whole MCM2–7 complex33. Previously, individual active centres were proposed to have distinct roles in regulating helicase activities36,41,42. Supported by our data, allosteric regulation of these ATPase centres orchestrated by the orientation changes between adjacent CTDs, might be the basis for factor-dependent control of helicase activities during different replication stages.

Figure 3: Inactive ATPase centres of 2:6 and 5:2 dimers.
Figure 3

a, b, Comparison of the ATP-binding pockets from 2:6 and 5:2 dimers with that from 7:3 dimer. Dimers (cis:trans) are presented with cis and trans ATPase elements on the left and right, respectively. c, d, Same as a and b, but the ATPase centres of 2:6 and 5:2 dimers are compared with that of the hexameric E1 helicase40. Motifs of Walker A (WA), Walker B (WB), arginine finger (AF), sensor 2 and sensor 3 are displayed in stick model. The directions and distances of sensor 3 movements are marked. All alignments were done using the Walker A and B motifs as a reference.

Axial displacement of interior hairpin loops

As in many hexameric AAA+ machineries, the central-pored chamber of MCM2–7 complex is decorated with layers of hairpin loops. For archaeal MCM, four layers of conserved loops essential for DNA binding and/or unwinding have been described35,43, with two of them located innermost (Fig. 4a). The first one, composed of six H2Is, was previously shown to undergo axial movement depending on the nucleotide-binding states of the ATPase domains44. The other, composed of β-turn motifs of the six OBs, was shown to coordinate the binding of ssDNA to the MCM–ssDNA binding motifs on the channel surface of OBs in the crystal structure of an archaeal MCM NTD homohexamer27. In the MCM2–7 complex, these two layers of loops are placed in axially staggered positions (Fig. 4b, c), and particularly, six H2Is roughly display a helical trajectory (Fig. 4c). An alignment with the ssDNA-bound archaeal MCM hexamer27 precisely placed ssDNA between these two layers of loops (Fig. 4b, c). In addition, when a double-stranded DNA (dsDNA) is placed in the channel, the H2Is show very close contact with it, capable of inserting their terminal loops into its major or minor grooves consecutively (Fig. 4d). At a very low threshold, an extra piece of fragmented density, which might be the residual dsDNA, could be identified within the channel at the H2I ring, but its sub-stoichiometric occupancy prevented positive identification and further analysis. Nevertheless, the snug fitting of the helically arranged H2Is and dsDNA suggests that H2I might be involved in the initial melting of origin DNA. These observations, together with repeated reports of the axial displacement of interior loops of AAA+ hexameric machines40,45, suggest that the MCM2–7 complex uses a conserved mechanism involving cycles of ATP binding and hydrolysis to control the axial positions of the interior loops to facilitate DNA translocation and unwinding.

Figure 4: Spatial distribution of conserved hairpins in the single hexamer.
Figure 4

a, Bottom view (from the hexamer interface) of the MCM2–7 hexamer. The six MCM subunits are shown in alternating grey scales, with four hairpins highlighted in indicated colours. All four hairpins are at the subunit interface, and extend towards the adjacent subunit. b, c, Side views of the inner surface of the hexamer, docked with a model ssDNA from an archaeal MCM structure (PDB accession code 4POG)27. d, Contacts between H2Is and dsDNA illustrated through fitting a dsDNA fragment into the central channel of the hexamer.

Central channel and model of initial origin melting

The diameter of the central channel in MCM2–7 single hexamer is not uniform (Fig. 5a, b), about 30 Å at the C-terminal end and 40 Å at the N-terminal end, but with two constriction sites (25 Å) at H2Is and β-turn motifs that are just wide enough to accommodate dsDNA (Fig. 5c, d). However, owing to the twisted stacking between two ZF rings in the double hexamer (Fig. 5f), the channel is partially blocked at the hexamer interface by the ZF rings, splitting the wide channel at the double hexamer interface into a main central channel and two minor channels (Fig. 5f and Supplementary Video 1). The overlapping central channel is just about the size of dsDNA (Fig. 5f, g), while the minor channels are not wide enough for the passage of dsDNA (Fig. 5h) but accessible from the outside. Notably, gate-forming subunits MCM2 and MCM5 participate in the formation of both channels. The overlapping central channel is delineated by ZFs from two MCM2–MCM5 dimers and a vertically arranged MCM6 dimer (Fig. 5g), and the minor channel involves ZFs from MCM2 and 5 of one single hexamer, and MCM3 and 7 of the other single hexamer (Fig. 5h).

Figure 5: Central channel and its implication in origin melting.
Figure 5

a, b, Cut-away views of the density map (unsharpened) with two dsDNA fragments fitted in the central channel. cf, Surface representation (top view) of the atomic models of the six H2Is (c), β-turns (d), ZFs (e) from one single hexamer, and ZFs from both single hexamers (f). Red asterisks mark the positions of two minor channels. g, h, Double hexamer interface centred for the views of the narrowed central channel (g), and minor channel (h). ZFs not part of the constricted central channel are in high transparency. i, Model for initial origin melting (see text).

The structure of an already constricted central channel of the single hexamer that opens to a larger channel at the NTD only to be occluded by the offset of the ZF rings at the double hexamer interface invites speculations for functions. First, the kink in the central channel created by the offset of the two single hexamer rings will probably cause deformation of trapped duplex DNA (Fig. 5i) to create a nucleation centre for DNA melting. Second, the tight grip of the duplex DNA on either ends by the helically positioned H2Is serves to hold the kinked DNA in place such that a slight left-handed rotation between the two single hexamers, as previously proposed18, could further deform the origin DNA at the nucleation point. Third, possible relative rotation between the NTD and CTD rings within single hexamers upon helicase activation, might further lower the activation energy of DNA melting. We envision that initial melting involves allosteric conformational changes, in combination with dsDNA translocation in opposite directions by the coupled single hexamers32. The dsDNA being pumped into the central channel provides the slack necessary for strand separation. This initial melting step requires the activation of the MCM2–7 helicase activity most likely by DDK phosphorylation and binding of Cdc45 and GINS12. Recent studies showed that DDK phosphorylation of the NTEs of MCM2–7 does not cause double hexamer separation18,46, but promotes MCM2–MCM5 gate opening47. Opening of the MCM2–MCM5 gates at this point would merge central and minor channels, creating an expanded N-terminal chamber for strand separation. The ssDNA looping out through this chamber would be accessible to replication factors lurking nearby (Fig. 5i). Further strand separation towards the CTD ring may be facilitated by the interior β-hairpin loops and the MCM–ssDNA binding motifs27 on the inner surface of OBs.

This structure-informed hypothesis on the initial origin melting is in accordance with previous data. First, many factors required for helicase activation, such as Sld2, Sld3, Cdc45 and Mcm10, have well-defined ssDNA binding activity11,16,32,48,49,50. Second, a similar replicative helicase SV40 large tumour antigen initiates origin melting as a dsDNA pump32,35, and conformational rearrangements of the two single hexamers were observed during this process31. Our structure suggests that, in addition to its role in processive fork unwinding, MCM2–7 is also actively involved in origin DNA melting. In transitioning from the initial origin melting state to the fork unwinding state, MCM2–7 essentially translocates first on dsDNA (dsDNA pump) and then along ssDNA (steric exclusion).

In summary, the fine structural details provided in this work will serve as a rich source of information for designing and interpreting biochemical studies aimed at dissecting the mechanistic functions of the MCM2–7 complex. In particular, it will provide a framework for future study of the eukaryotic-specific assembly, activation and regulation of this helicase family.

Methods

No statistical methods were used to predetermine sample size.

Yeast strain

One-step PCR-based approach51 with pTF272 (pFA6a-TEV-6×Gly-3×Flag-HphMX, Addgene) as DNA template was used to generate MCM4-TEV-3×Flag tagging modification in the W303-1a background strain. The resulting strain showed no growth defect compared to its parent W303-1a strain.

Sample purification

Forty litres of log-phase G1 yeast cells (3 × 107–4 × 107 cells per ml) were collected and processed for spheroplasting to isolate crude chromatin as described previously52 with the following modifications for a large-scale preparation. Spheroplasting was performed in 200 ml of spheroplasting buffer containing sufficient amount of lyticase that was purified from an Escherichia coli strain bearing lyticase expressing plasmid pUV5-G1S (gift from S. Gasser). The spheroplasts were lysed with extraction buffer EBX (50 mM HEPES/KOH, pH 7.5, 100 mM K-glutamate, 10 mM magnesium acetate, 0.25% Triton X-100, 3 mM ATP, 1 mM dithiothreitol (DTT), 1 mM EDTA, 2 mM NaF, 1 mM NaVO4, 1 mM phenylmethanesulfonylfluoride (PMSF), 2 μg ml−1 pepstatin A and 1× protease inhibitor cocktail (Roche)). The lysate was layered onto the top of equal volume of EBX buffer containing 30% sucrose and centrifuged at 25,000g (Hitachi R20A2) for 15 min. To solubilize chromatin fractions, the crude chromatin was digested in 40 ml of freshly made benzonase buffer (50 mM HEPES/KOH, pH 7.5, 100 mM K-glutamate, 8 mM MgCl2, 0.02% NP-40, 3 mM ATP, 1 mM EDTA, 2 mM NaF, 1 mM NaVO4, 1 mM PMSF, 2 μg ml−1 pepstatin A and 1× protease inhibitor cocktail (Roche)) with 1 U μl−1 of benzonase (71206-3; Merck Biosciences) for 10 min at 37 °C, and then 1 h on ice. The suspension was then centrifuged for 20 min at 25,000g. The clear phase was recovered, and subjected to anti-Flag immunoprecipitation with 1 ml bed volume of washed anti-Flag M2 agarose (Sigma) at 4 °C for 2 h. Beads were recovered, and washed extensively with benzonase buffer and then tobacco etch virus (TEV) buffer (50 mM HEPES/KOH, pH 7.5, 100 mM K-glutamate, 8mM MgCl2, 0.02% NP-40, 3 mM ATP). MCM2–7 complexes were cleaved from the M2 agarose by incubation for overnight at 4 °C in TEV buffer with 100 U ml−1 of AcTEV protease (Life Technology). His-tagged TEV protease was removed by incubating the eluate with a TALON metal affinity resin (Clontech) for 30 min at 4 °C. The MCM2–7 complexes were then applied on the top of 20–40% glycerol gradient in buffer EBX with protease inhibitors. The glycerol gradient was centrifuged in a TLS-55 rotor (Beckman Optima TLX ultracentrifuge) at 175,000g for 6.5 h. The fractions were collected from the top of the gradient after centrifugation. The fractions containing the MCM2–7 double hexamers were pooled and processed for electron microscopy analysis.

Electron microscopy

The MCM2–7 double hexamer was concentrated by ultrafiltration to remove glycerol. Negative staining of the MCM2–7 double hexamer was performed with 2% uranyl acetate. Grids were examined using an FEI T12 microscope operated at 120 kV, and images were recorded using a 4k × 4k charge-coupled device (CCD) camera (UltraScan 4000, Gatan).

For cryo-grid preparation, 4 μl aliquots of samples were applied to a glow-discharged holy carbon grid (Quantifoil R2/2) coated with a thin layer of freshly prepared carbon, and cryo-freezing was performed with an FEI Vitrobot Mark IV (4 °C and 100% humidity). Grids were examined using an FEI Titan Krios operated at 300 kV, and images were recorded using a K2 Summit direct electron detector (Gatan) in counting mode, at a nominal magnification of 22,500×, which renders a final pixel size of 1.32 Å at object scale after post-magnification calibration, and with the defocus ranging from −1.5 to −2.5 μm. Images were collected under low-dose condition in a semi-automatic manner using UCSF-Image4 (written by X. Li and Y. Cheng). For each micrograph stack, a total of 32 frames were collected, with a dose rate of 8.2 counts (10.9 electrons) per physical pixel per second for an exposure time of 8 s.

Image processing

Initial 3D model from negatively stained particles was calculated using RELION53 using a density cylinder as reference. For cryo-EM data, beam-induced motion correction at micrograph level was performed as previously described (written by X. Li)54. Micrographs screening, automatic particle picking and normalization were done with SPIDER55. Program of CTFFIND3 (ref. 56) was used to estimate the contrast transfer function parameters. The 2D, 3D classification and refinement were performed with RELION. A total of 347,801 particles (with a binning factor of two) from 2,230 micrographs were subjected to a cascade of 2D and 3D classification. Analysis of classification structures indicated that there is a C2-axis perpendicular to the cylinder axis of the MCM2–7 double hexamer, reflecting a symmetric arrangement of one single hexamer relatively to the other single hexamer by a simple 180° rotation. A final structurally homogeneous data set composed of 85,365 particles, as classification structures of them have reached to considerably higher resolution, in full window size (300 × 300) were used for high-resolution refinement with C2-symmetry imposed. From the orientation distribution (Extended Data Fig. 1h, i), there is a wide equator belt with a complete distribution of particles, along with two regions with relatively more particles. Nevertheless, this type of uneven distribution did not affect our final reconstruction, as particles from the equator belt have provided sufficient information for a complete sampling of the central slices in the Fourier space. Symmetry-free refinement was also performed, resulting in generally similar but slightly worse density maps. To improve the resolution further, different combinations of movie frames were used for motion correction and frame averaging. The first two frames had large motions, therefore, frames 3–16 were used to sum micrographs. To reduce interpolation errors, particles were rewindowed by offsetting translation parameters determined in the 3D refinement of last round, which improved the resolution to 4.6 Å (gold-standard FSC 0.143 criteria). The final round of refinement was performed with a soft-edged mask applied, resulting in a 4.3-Å map. After correction for the modulation transfer function of K2 detector, and map sharpening using post-processing options of RELION with a B-factor of −100 Å2, the overall resolution of the final density map within the region defined by the soft mask is 3.8 Å for the overall map (Extended Data Fig. 1k), after correction of the effect of soft mask on the FSC curve57. Local resolution map was estimated using blocres in Bsoft58. From the local resolution map, peripheral regions are associated with worst resolution, while the core region is better than 3.5 Å. The statistics of the data collection and structural refinement is provided in Extended Data Table 1b.

Model building

Six monomers of the crystal structure of a chimaeric archaeal MCM (PDB code 4R7Y)30 (Sulfolobus solfataricus NTD fused with Pyrococcus furiosus CTD) hexamer were manually docked to the density map of the MCM double hexamer using Chimera59. The docking also confirmed the handedness of the density map. The rigid-body docking was performed by dividing the crystal structure of the monomer into four pieces (NTD-A, OB-fold, ZF and CTD) (Extended Data Fig. 4b, c). Sequence alignments of the yeast MCM proteins with crystal template were initially performed using BLAST60 and manually adjusted according to the secondary structure prediction of these sequences (PSIPRED)61. The predicted secondary structural information of the eukaryotic subunit-specific sequences was used to assign the six MCM proteins into the cryo-EM density map. Initial atomic coordinates of the OB-fold subdomains and CTDs of MCM2–7 proteins were then generated using CHAINSAW62 in the CCP4 suite63. Models were manually adjusted and built in Coot64. Only minor changes were required for modelling the OB-fold subdomains and CTDs of the yeast MCM proteins owing to their high sequence identity to the template (Extended Data Fig. 2). The NTD-As of the yeast MCM proteins contain many sequence insertions, and the modelling of these sequences was similar to that described above, but involved multiple rounds of realignment of sequences and largely facilitated by the predicted secondary structure. In many cases, the modelling of NTD-A required complete retracing of the main-chain based solely on densities and secondary structural predictions. For regions independent of known template (eukaryotic-specific sequences, Extended Data Fig. 4), poly-alanine models were built first using Coot. Clearly resolved bulky residues (Phe, Tyr, Trp and Arg) were then used as markers to assign the primary sequences. As a result, we derived an atomic model of the MCM2–7 double hexamer, for 80% of its sequences, from a near-atomic cryo-EM density map, integrated with structural information from other sources. Further model refinement was done by alternating rounds of model rebuilding in Coot and real-space refinement (phenix.real_space_refine)65 in Phenix66, with secondary structure and stereochemical constraints applied. Similar to a previous cryo-EM work with comparable 3.8-Å resolution67, during the real-space refinement, knowledge-based restraints, including Ramachandran potentials and rotamer correction, were applied to ensure a proper balance between density-fitting and stereochemical and rotamer distributions. Owing to the resolution limitation, local densities at the ATP-binding sites could not unambiguously distinguish between ATP and ADP. For modelling purpose, ADP was docked to the active centres and similarly refined in Phenix. The atomic model was cross-validated according to previously described procedures68,69. Specifically, the coordinates of the final model were randomly displaced by 0.2 Å using the PDB tools of Phenix. The displaced model was refined against the Half1 map (produced from a half set of all particles during refinement by RELION). The refined model from Half1 map was compared with the maps of Half1, Half2 in Fourier space to produce two FSC curves, FSCwork (model versus Half1 map) and FSCfree (model versus Half2 map), respectively (Extended Data Fig. 1l). Another FSC curve between the refined model from Half1 and the final density map (model versus merge) from all particles was also produced. As indicated by these curves, the agreement between FSCwork and FSCfree (no large separation) indicated that the model was not overfitted. MolProbity70 (http://molprobity.biochem.duke.edu/) was used to evaluate the final model, and final statistics of the model was provided in Extended Data Table 1b. Notably, application of knowledge-based restraints during the real-space refinement has improved the stereochemical and rotamer statistics of the model. Comparisons of the representative density with the atomic model for selected areas are shown in Extended Data Fig. 6 and Supplementary Video 2.

Gravity centres of individual domains (Fig. 1) were determined with segmented maps of the conserved core regions of these domains (minus variable loops and linkers) using SPIDER. To determine the cylinder axis of the hexamer, a plane perpendicular to the axis was determined by least-square fitting of six centres of the OBs in Chimera. Pymol71 and Chimera were used for structural analysis and figure preparation. Interface areas of the intersubunit and inter-hexamer interactions were calculated by PISA72, and provided in Extended Data Table 1a.

Accessions

Primary accessions

Electron Microscopy Data Bank

Protein Data Bank

Data deposits

The cryo-EM density map has been deposited in the Electron Microscopy Data Bank (EMDB) under accession number EMD-6338; and the atomic model has been deposited in the Protein Data Bank (PDB) under accession number 3JA8.

References

  1. 1.

    , & Principles and concepts of DNA replication in bacteria, archaea, and eukarya. Cold Spring Harb. Perspect. Biol. 5, a010108 (2013)

  2. 2.

    , & Mechanisms for initiating cellular DNA replication. Annu. Rev. Biochem. 82, 25–54 (2013)

  3. 3.

    & A structural framework for replication origin opening by AAA+ initiation factors. Curr. Opin. Struct. Biol. 23, 144–153 (2013)

  4. 4.

    MCM proteins in DNA replication. Annu. Rev. Biochem. 68, 649–686 (1999)

  5. 5.

    et al. Concerted loading of Mcm2–7 double hexamers around DNA during DNA replication origin licensing. Cell 139, 719–730 (2009)

  6. 6.

    et al. A double-hexameric MCM2–7 complex is loaded onto origin DNA during licensing of eukaryotic DNA replication. Proc. Natl Acad. Sci. USA 106, 20240–20245 (2009)

  7. 7.

    , & Regulating DNA replication in eukarya. Cold Spring Harb. Perspect. Biol. 5, a012930 (2013)

  8. 8.

    et al. Eukaryotic origin-dependent DNA replication in vitro reveals sequential action of DDK and S-CDK kinases. Cell 146, 80–91 (2011)

  9. 9.

    , , , & Regulated eukaryotic DNA replication origin firing with purified proteins. Nature 519, 431–435 (2015)

  10. 10.

    & Helicase activation and establishment of replication forks at chromosomal origins of replication. Cold Spring Harb. Perspect. Biol. 5, a01037 (2013)

  11. 11.

    , & Switch on the engine: how the eukaryotic replicative helicase MCM2–7 becomes activated. Chromosoma 124, 13–26 (2015)

  12. 12.

    , , & Activation of the MCM2–7 helicase by association with Cdc45 and GINS proteins. Mol. Cell 37, 247–258 (2010)

  13. 13.

    et al. Selective bypass of a lagging strand roadblock by the eukaryotic replicative DNA helicase. Cell 146, 931–941 (2011)

  14. 14.

    , , & MCM forked substrate specificity involves dynamic interaction with the 5′-tail. J. Biol. Chem. 282, 34229–34234 (2007)

  15. 15.

    , , & Organization of the archaeal MCM complex on DNA and implications for the helicase mechanism. Nature Struct. Mol. Biol. 12, 756–762 (2005)

  16. 16.

    et al. DNA binding polarity, dimerization, and ATPase ring remodeling in the CMG helicase of the eukaryotic replisome. eLife 3, e03273 (2014)

  17. 17.

    , , & Steric exclusion and wrapping of the excluded DNA strand occurs along discrete external binding paths during MCM helicase unwinding. Nucleic Acids Res. 39, 6585–6595 (2011)

  18. 18.

    et al. Structural and mechanistic insights into Mcm2–7 double-hexamer assembly and function. Genes Dev. 28, 2291–2303 (2014)

  19. 19.

    et al. A unique DNA entry gate serves for regulated loading of the eukaryotic replicative helicase MCM2–7 onto DNA. Genes Dev. 28, 1653–1666 (2014)

  20. 20.

    et al. Cryo-EM structure of a helicase loading intermediate containing ORC-Cdc6-Cdt1–MCM2–7 bound to DNA. Nature Struct. Mol. Biol. 20, 944–951 (2013)

  21. 21.

    et al. The structural basis for MCM2–7 helicase activation by GINS and Cdc45. Nature Struct. Mol. Biol. 18, 471–477 (2011)

  22. 22.

    et al. DNA induces conformational changes in a recombinant human minichromosome maintenance complex. J. Biol. Chem. 290, 7973–7979 (2015)

  23. 23.

    et al. Crystal structure of a near-full-length archaeal MCM: functional insights for an AAA+ hexameric helicase. Proc. Natl Acad. Sci. USA 105, 20191–20196 (2008)

  24. 24.

    et al. Insights into the architecture of the replicative helicase from the structure of an archaeal MCM homolog. Structure 17, 211–222 (2009)

  25. 25.

    et al. Mini-chromosome maintenance complexes form a filament to remodel DNA structure and topology. Nucleic Acids Res. 41, 3446–3456 (2013)

  26. 26.

    et al. The structure and function of MCM from archaeal M. thermoautotrophicum. Nature Struct. Biol. 10, 160–167 (2003)

  27. 27.

    , , , & A conserved MCM single-stranded DNA binding element is essential for replication initiation. eLife 3, e01993 (2014)

  28. 28.

    , , , & The 1.8-Å crystal structure of the N-terminal domain of an archaeal MCM as a right-handed filament. J. Mol. Biol. 426, 1512–1523 (2014)

  29. 29.

    , , , & Structural analysis of the Sulfolobus solfataricus MCM protein N-terminal domain. Nucleic Acids Res. 36, 3235–3243 (2008)

  30. 30.

    , , & Analysis of the crystal structure of an active MCM hexamer. eLife 3, e03433 (2014)

  31. 31.

    et al. Conformational rearrangements of SV40 large T antigen during early replication events. J. Mol. Biol. 397, 1276–1286 (2010)

  32. 32.

    & The eukaryotic Mcm2–7 replicative helicase. Subcell. Biochem. 62, 113–134 (2012)

  33. 33.

    , & Subunit organization of Mcm2–7 and the unequal role of active sites in ATP hydrolysis and viability. Mol. Cell. Biol. 28, 5865–5873 (2008)

  34. 34.

    et al. The ORC/Cdc6/MCM2–7 complex facilitates MCM2–7 dimerization during prereplicative complex formation. Nucleic Acids Res. 42, 2257–2269 (2014)

  35. 35.

    & MCM structure and mechanics: what we have learned from archaeal MCM. Subcell. Biochem. 62, 89–111 (2012)

  36. 36.

    & The Mcm complex: unwinding the mechanism of a replicative helicase. Microbiol. Mol. Biol. Rev. 73, 652–683 (2009)

  37. 37.

    et al. A viable allele of Mcm4 causes chromosome instability and mammary adenocarcinomas in mice. Nature Genet. 39, 93–98 (2007)

  38. 38.

    , , , & mcm5/cdc46-bob1 bypasses the requirement for the S phase activator Cdc7p. Proc. Natl Acad. Sci. USA 94, 3151–3155 (1997)

  39. 39.

    , & Crystal structure of the eukaryotic origin recognition complex. Nature 519, 321–326 (2015)

  40. 40.

    & Mechanism of DNA translocation in a replicative hexameric helicase. Nature 442, 270–275 (2006)

  41. 41.

    , & Multiple functions for Mcm2–7 ATPase motifs during replication initiation. Mol. Cell 55, 655–665 (2014)

  42. 42.

    , , , & Origin licensing requires ATP binding and hydrolysis by the MCM replicative helicase. Mol. Cell 55, 666–677 (2014)

  43. 43.

    & The minichromosome maintenance replicative helicase. Cold Spring Harb. Perspect. Biol. 5, a012807 (2013)

  44. 44.

    & Minichromosome maintenance helicase activity is controlled by N- and C-terminal motifs and requires the ATPase domain helix-2 insert. Proc. Natl Acad. Sci. USA 103, 7613–7618 (2006)

  45. 45.

    , , , & Mechanisms of conformational change for a replicative hexameric helicase of SV40 large tumor antigen. Cell 119, 47–60 (2004)

  46. 46.

    et al. Prereplicative complexes assembled in vitro support origin-dependent and independent DNA replication. EMBO J. 33, 605–620 (2014)

  47. 47.

    & The Dbf4-Cdc7 kinase promotes Mcm2–7 ring opening to allow for single-stranded DNA extrusion and helicase assembly. J. Biol. Chem. 290, 1210–1221 (2015)

  48. 48.

    & Cdc45 protein-single-stranded DNA interaction is important for stalling the helicase during replication stress. J. Biol. Chem. 288, 7550–7563 (2013)

  49. 49.

    et al. Primer utilization by DNA polymerase α-primase is influenced by its interaction with Mcm10p. J. Biol. Chem. 279, 16144–16153 (2004)

  50. 50.

    , , , & Novel DNA binding properties of the Mcm10 protein from Saccharomyces cerevisiae. J. Biol. Chem. 284, 25412–25420 (2009)

  51. 51.

    et al. A versatile toolbox for PCR-based tagging of yeast genes: new fluorescent proteins, more markers and promoter substitution cassettes. Yeast 21, 947–962 (2004)

  52. 52.

    , , & Cdc14p resets the competency of replication licensing by dephosphorylating multiple initiation proteins during mitotic exit in budding yeast. J. Cell Sci. 123, 3933–3943 (2010)

  53. 53.

    A Bayesian view on cryo-EM structure determination. J. Mol. Biol. 415, 406–418 (2012)

  54. 54.

    et al. Electron counting and beam-induced motion correction enable near-atomic-resolution single-particle cryo-EM. Nature Methods 10, 584–590 (2013)

  55. 55.

    et al. SPIDER image processing for single-particle reconstruction of biological macromolecules from electron micrographs. Nature Protocols 3, 1941–1974 (2008)

  56. 56.

    & Accurate determination of local defocus and specimen tilt in electron microscopy. J. Struct. Biol. 142, 334–347 (2003)

  57. 57.

    et al. High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopy. Ultramicroscopy 135, 24–35 (2013)

  58. 58.

    & Bsoft: image processing and molecular modeling for electron microscopy. J. Struct. Biol. 157, 3–18 (2007)

  59. 59.

    et al. UCSF Chimera–a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612 (2004)

  60. 60.

    Using the Basic Local Alignment Search Tool (BLAST). CSH Protoc. 2007, pdb.top17 (2007)

  61. 61.

    , , , & Scalable web services for the PSIPRED Protein Analysis Workbench. Nucleic Acids Res. 41, W349–W357 (2013)

  62. 62.

    CHAINSAW: a program for mutating pdb files used as templates in molecular replacement. J. Appl. Crystallogr. 41, 641–643 (2008)

  63. 63.

    et al. Overview of the CCP4 suite and current developments. Acta Crystallogr. D 67, 235–242 (2011)

  64. 64.

    , , & Features and development of Coot. Acta Crystallogr. D 66, 486–501 (2010)

  65. 65.

    et al. Towards automated crystallographic structure refinement with phenix.refine. Acta Crystallogr. D 68, 352–367 (2012)

  66. 66.

    et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010)

  67. 67.

    et al. An atomic model of brome mosaic virus using direct electron detection and real-space optimization. Nature Commun. 5, 4808 (2014)

  68. 68.

    et al. Mechanistic insights into the recycling machine of the SNARE complex. Nature 518, 61–67 (2015)

  69. 69.

    et al. Structure of the yeast mitochondrial large ribosomal subunit. Science 343, 1485–1489 (2014)

  70. 70.

    et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D 66, 12–21 (2010)

  71. 71.

    The PyMOL Molecular Graphics System v.1.3r1. (2010)

  72. 72.

    & Inference of macromolecular assemblies from crystalline state. J. Mol. Biol. 372, 774–797 (2007)

  73. 73.

    BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series 41, 95–98 (1999)

  74. 74.

    et al. Characterization and structure determination of the Cdt1 binding domain of human minichromosome maintenance (Mcm) 6. J. Biol. Chem. 285, 12469–12473 (2010)

Download references

Acknowledgements

We thank X. Li for providing programs in data collection, motion correction and framed-based analysis, and J. Wang for advices on modelling and model refinement. We also thank the National Center for Protein Sciences (Beijing, China) for technical support with cryo-EM data collection and for computation resource. This work was supported by the Ministry of Science and Technology of China (2013CB910404 to N.G.), the National Natural Science Foundation of China (31422016 to N.G.), the Research Grants Council of Hong Kong (GRF664013 and HKUST12/CRF/13G to Yu.Z.) and the Hong Kong University of Science & Technology (B.-K.T.).

Author information

Author notes

    • Ningning Li
    •  & Yuanliang Zhai

    These authors contributed equally to this work.

Affiliations

  1. Ministry of Education Key Laboratory of Protein Sciences, Center for Structural Biology, School of Life Sciences, Tsinghua University, Beijing 100084, China

    • Ningning Li
    • , Yixiao Zhang
    • , Wanqiu Li
    • , Maojun Yang
    • , Jianlin Lei
    •  & Ning Gao
  2. Division of Life Science, Hong Kong Universityof Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China

    • Yuanliang Zhai
    •  & Bik-Kwoon Tye
  3. Institute for Advanced Study, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China

    • Yuanliang Zhai
  4. Department of Molecular Biology and Genetics, College of Agriculture and Life Sciences, Cornell University, Ithaca, New York 14853, USA

    • Bik-Kwoon Tye

Authors

  1. Search for Ningning Li in:

  2. Search for Yuanliang Zhai in:

  3. Search for Yixiao Zhang in:

  4. Search for Wanqiu Li in:

  5. Search for Maojun Yang in:

  6. Search for Jianlin Lei in:

  7. Search for Bik-Kwoon Tye in:

  8. Search for Ning Gao in:

Contributions

Yu.Z. purified sample; N.L. collected cryo-EM data (with J.L., Yi.Z. and W.L), performed image processing, and analyzed structures. N.L., N.G. and M.Y. performed atomic modelling. N.L., Yu.Z., B.-K.T. and N.G. designed experiments, interpreted the structure and wrote the manuscript.

Competing interests

The authors declare no competing financial interests.

Corresponding authors

Correspondence to Yuanliang Zhai or Bik-Kwoon Tye or Ning Gao.

Extended data

Supplementary information

Videos

  1. 1.

    Overall structure of the MCM2-7 DH and arrangement of the two ZF rings.

    The cryo-EM density map (unsharpened) of the MCM2-7 DH is first shown in surface representation, followed by superimposition of atomic models for each of the MCM proteins one by one. The unique side-channel between MCM2 and MCM6 is highlighted. Subsequently, only the two rings of ZFs, with their atomic models converted to surface representation, are shown in zoom-in views. The two stacked ZF rings are rotated in different directions to highlight the diameter and wall components of a major channel and two minor channels at the DH interface. At last, the surface representation of two ZF pairs of MCM2:MCM5 is hidden, highlighting the proposed fusion of three channels into a larger one upon the gap opening between MCM2 and MCM5. A thumbnail map of the MCM2-7 DH, with the 2-fold axis displayed as a red rod, is shown on the top right corner to illustrate the orientations of individual movie frames relative to the DH.

  2. 2.

    Sharpened density map of the MCM2-7 DH

    The cryo-EM density map of the MCM2-7 DH (sharpened) is displayed in surface representation, zoomed into selected regions with atomic models superimposed.

About this article

Publication history

Received

Accepted

Published

DOI

https://doi.org/10.1038/nature14685

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.