Mechanisms for initiating cellular DNA replication

See allHide authors and affiliations

Science  24 Feb 2017:
Vol. 355, Issue 6327, eaah6317
DOI: 10.1126/science.aah6317

Diverse molecular choreography of replication

Accurate duplication and transmission of genetic information to the next generation requires complex molecular assemblies. Bleichert et al. review replication initiation across the three domains of life, with a focus on origin selection and helicase loading. These processes identify potential origins of replication and prepare them for subsequent bidirectional replication initiation. There are key similarities and multiple differences in replication mechanisms between eukaryotes, prokaryotes, and archaea and many outstanding questions to be answered.

Science, this issue p. eaah6317

Structured Abstract


Cellular life depends on the ability of organisms to accurately duplicate and transmit genetic information between generations. In bacteria, archaea, and eukaryotes, this process uses sophisticated molecular machineries, termed replisomes, which copy chromosomal DNA through semiconservative replication. At the core of these replication factories reside ring-shaped hexameric helicases that assist polymerases in DNA synthesis by processively unzipping the parental DNA double helix. Replicative helicases are loaded onto DNA by dedicated initiator, loader, and accessory proteins during the initiation of DNA replication in a tightly regulated, multistep process. Although initiators and loaders are phylogenetically related between the bacterial and archaeal/eukaryotic lineages, the molecular choreography of DNA replication initiation turns out to be surprisingly diverse across different systems and employs different strategies to prepare replication origins for subsequent replisome assembly.


Early work in bacterial systems established that initiator proteins recognize specific sequence elements (termed replication origins). Nucleotide-dependent oligomerization of the bacterial initiator DnaA at origins facilitates DNA melting and provides an access point for the replicative helicase, DnaB, which is recruited to and loaded onto each of the single-stranded origin regions by interactions with the initiator and other loader proteins to establish bidirectional replication forks. It has recently become clear that the strategies for origin recognition, helicase loading, and duplex DNA melting in archaeal and eukaryal systems, as well as the order of these steps, deviate considerably from the archetypal initiation events followed by bacteria. Although nucleotide-dependent interactions between initiator subunits still control early steps of replication initiation and regulate the association of initiators with origin DNA, they do not appear to contribute to DNA melting. Instead, archaeal/eukaryotic initiators [Orc in archaea or the multisubunit origin recognition complex (ORC) in eukaryotes] and helicase coloading factors cooperate to recruit and load the replicative helicase motor, the minichromosome maintenance (MCM) complex, onto duplex DNA in an inactive, double-hexameric form. Helicase activation, DNA melting, and replisome assembly do not occur until subsequent cell cycle phases, preventing premature replication and exposure of melted, single-stranded DNA regions to DNA damaging factors.


Although the core components of the initiation pathway have been identified, enabling reconstitution of the replication initiation cascade in vitro with recombinant proteins from bacterial and eukaryotic systems, the mechanisms by which these initiation factors function at the molecular level remain ill defined. Moreover, how replication initiation is temporally and spatially linked to the local chromatin environment and coordinated with other cellular (in bacteria and archaea) or nuclear (in eukaryotes) processes is not well understood. Future studies that integrate genetic, biochemical, biophysical, and structural approaches will be pivotal for resolving these questions.

Overview of similarities and differences in the initiation of DNA replication across the three domains of life.

Bacteria and archaea/eukaryotes share a requirement for an adenosine triphosphate (ATP)–dependent initiator factor that helps catalyze the loading of two ring-shaped, hexameric helicases onto a replication origin, thereby nucleating the formation of a bidirectional replication fork. The bacterial initiator forms a multimeric assembly that actively melts the origin before loading single helicase hexamers, whereas in archaea and eukaryotes, the helicase is loaded as an inactive dodecamer that then isomerizes into two active, single hexamers during or following origin melting.


Cellular DNA replication factories depend on ring-shaped hexameric helicases to aid DNA synthesis by processively unzipping the parental DNA helix. Replicative helicases are loaded onto DNA by dedicated initiator, loader, and accessory proteins during the initiation of DNA replication in a tightly regulated, multistep process. We discuss here the molecular choreography of DNA replication initiation across the three domains of life, highlighting similarities and differences in the strategies used to deposit replicative helicases onto DNA and to melt the DNA helix in preparation for replisome assembly. Although initiators and loaders are phylogenetically related, the mechanisms they use for accomplishing similar tasks have diverged considerably and in an unpredictable manner.

DNA is expressed as a linear triplet code, with numerous noncoding regulatory regions that render the genomes of cells quite large. For example, in humans, each of our 1013 cells carries roughly 2 m of DNA; an individual’s body will synthesize more than a light year’s worth of DNA from conception to death [1016 cell divisions in ~70 years (1)]. At the same time, the error rate by which base-mispairing mutations are accidentally introduced into newly made strands is vanishingly low, less than one mistake per 108 nucleotide-addition events. How such fidelity is achieved without sacrificing speed (a single replication fork can catalyze 25 to 1000 nucleotide incorporation events per second) to meet the demands of cell proliferation remains a frontier question in biology.

The creation of new DNA strands from parental chromosomes depends on a large, dynamic multiprotein complex termed the replisome, which uses an array of DNA unwinding (helicase) and synthesis (primase/polymerase) functionalities, coupled with various processivity (sliding ring clamps and their loaders), protective [single-stranded DNA (ssDNA) binding], and scaffolding factors. Despite the clear functional conservation of core replisomal activities, the helicases and primase/polymerases used for replicating DNA have evolved twice independently, with bacteria using one class of enzymes and archaea/eukaryotes using another (2). This dichotomy agrees well with the proposal that the adoption of DNA as the genetic material of living organisms occurred after the prior evolution of RNA and protein-based self-replicating systems.

Replisome assembly is tightly linked to cell-cycle status (3). Dedicated factors known as initiators play a critical role in recognizing replication start sites (termed origins) and in depositing replicative helicases onto origins, often with the help of specialized loading factors. In eukaryotes, chromosome-bound initiator complexes and loaded replicative helicases outnumber the actual origins that become activated in S phase. Many of these dormant origins are required to rescue stalled forks (additional roles for the remainder may yet be discovered), and the extent to which this choice is stochastic or determined by chromosomal context is not known (46). Timely and accurate replication onset is extremely important to cell viability, because aberrant helicase loading can directly lead to genetic instabilities and cellular transformation, as well as genetic diseases such as primordial dwarfism (711).

Here, we summarize the state of knowledge with respect to how initiation factors operate at the molecular level and the extent to which particular initiation mechanisms overlap or are distinct between bacteria, archaea, and eukaryotes. Given space constraints, we apologize to those colleagues whose work could not be cited.

Phylogeny and architecture of initiation enzymes

Initiators/helicase loaders

Cellular replication initiators are closely related to one another evolutionarily (12), which initially led to speculation that the mechanisms underlying origin recognition and helicase loading might be largely conserved between different cellular systems. In fact, the common subunit-level conservation of initiator structure often belies very different mechanisms of action.

Replication initiators share a similar domain architecture, in which individual protomers are composed of a AAA+ [adenosine triphosphatases (ATPases) associated with various cellular activities] nucleotide-binding-and-hydrolysis domain, coupled to a C-terminal helix-turn-helix (HTH) domain (Fig. 1A). Both domains directly interact with origin DNA regions (1317). Bacteria usually encode a single initiator protein, known as DnaA, whereas archaea rely on one or more paralogous copies of an initiator known as Orc. Many eukaryotes use a six-subunit assembly termed the origin recognition complex (ORC), five subunits of which are predicated on a AAA+/HTH unit. Duplication and specialization of the ancestral Orc1 gene of archaea likely allowed for greater regulatory potential of initiator function in eukaryotes, where commitment to S phase involves activation of many origins and is linked to other aspects of cell metabolism. Interestingly, only a subset of the ORC components has been identified to date in certain unicellular eukaryotes, such as protists, suggesting that initiators in these systems may be functionally more similar to those in the archaeal system than those in other eukaryotes (18, 19).

Fig. 1 Phylogenetic and structural kinship of replication initiators.

(A) Initiators are multidomain proteins, comprising auxiliary, AAA+ ATPase, and HTH domains. Specific domain functions are indicated. (B) Replication initiators and helicase loaders are closely related evolutionarily, forming a separate AAA+ superfamily clade (12). AAA+ ATPases involved in DNA replication belonging to other AAA+ clades are also shown (for completeness, clades not containing DNA replication proteins are included in the diagram). The helix 2 insert, SFIII helicase, HCLR (HslU, ClpX, Lon, RuvB), and Pre-sensor 2 insert are separate clades within the Pre-sensor 1 hairpin superclade. (C) ATPase sites in AAA+ oligomers (shown for DnaA) [Protein Data Bank (PDB): 2HCB (22)] are formed at the interface between adjoining protomers and contain characteristic sequence elements for ATP binding and hydrolysis. In some instances (e.g., in MCMs), the Sensor 2 arginine can be provided in trans. (D) Bacterial DnaA [PDB: 2HCB (22)] and (E) eukaryotic ORC [PDB: 4XGC (16)] assemble into oligomers stabilized by interactions between the AAA+ and HTH domains of adjacent protomers.

The sequence similarity of the AAA+ domain in replication initiators is sufficiently preserved that the enzymes can be grouped into their own subfamily, or clade, within the AAA+ superlineage (12) (Fig. 1B). By contrast, the HTH regions of these proteins have apparently arisen through two different fusion events: one in bacteria that used a NarL/FixJ-type HTH fold and another in archaea/eukaryotes that incorporated a winged-helix (WH) domain (13, 20). The core AAA+/HTH element of initiators is often appended with distinct auxiliary folds, which offer additional functionalities that can affect origin recognition, helicase recruitment, and initiator interactions (Fig. 1A).

The AAA+ domain plays a central role in controlling higher-order initiator assembly and function (Fig. 1C). AAA+ proteins in general form homo- and heterooligomers in which the adenosine triphosphate (ATP)–binding site of one subunit is complemented by catalytically functional amino acids from a partner protomer (notably a basic side chain such as arginine, termed an “arginine finger”) (21). This arrangement allows initiators such as ORC and DnaA to assemble into large cracked rings or helical arrays, respectively (16, 22) (Fig. 1, D and E). The bipartite binding and hydrolysis of ATP further allows for allosteric coupling between subunits to switch initiators between different functional states (Fig. 1C).

Although some initiators appear capable of independently carrying out both origin recognition and helicase recruitment/deposition [e.g., Helicobacter pylori and Pseudomonas spp. DnaA (2325) and Sulfolobus Orc1 (26)], many are aided by specialized helicase-loading factors. Some, such as eukaryotic Cdt1 (27, 28), are noncatalytic. Others, such as DnaC/DnaI of bacteria and Cdc6 of eukaryotes, are AAA+ ATPases and close evolutionary cousins of replication initiators (12, 29) (Fig. 1B). Like DnaA and ORC, ATP-dependent helicase loading factors can assemble into larger, homomeric and heteromeric arrays (e.g., DnaC can form a spiral assembly, while Cdc6 appears capable of docking into an open ORC ring) (16, 30, 31). Here again, the joint AAA+ ATPase site is critical for controlling the activity of these enzymes.

Replicative helicases

Helicases comprise the front end of the replisome. Tasked with unwinding the entire genome, replicative helicases have evolved to couple high rates of speed (when working alongside cognate polymerases) with highly processive movement. Cellular replicative helicases universally form donut-shaped hexamers and are thought to encircle one of the two DNA strands during translocation to physically pry apart a downstream duplex (32, 33) (Fig. 2, A and B). The use of six independent yet coupled ATPase sites provides a highly coordinated means for generating directional movement and the motive force necessary to help separate parental DNA (32).

Fig. 2 Bacterial and archaeal/eukaryotic replicative helicases are hexameric motors but belong to different evolutionary ASCE ATPase lineages

(12). (A) Strand displacement by replicative helicase movement (arrows) results in duplex DNA unwinding. (B) Active replicative helicases form hexamers that encircle ssDNA. E. coli DnaB bound to ssDNA [PDB: 4ESV (156)] and eukaryotic Mcm2-7 (complexed with GINS and Cdc45) [PDB: 3JC5 (157)] are shown. (C) Topology diagrams of the motor domains of DnaB and MCM highlight the related but distinct folds (RecA versus AAA+, respectively) of these two ATPases, with ATP binding/hydrolysis and DNA binding motifs oriented differently with respect to each other. (D) The disparate locations of conserved sequence motifs and DNA binding loops cause the RecA-like and AAA+ folds to be positioned almost perpendicular to each other in the context of the DnaB and Mcm2-7 hexamers (indicated by green arrows) (35). PDBs: DnaB, 4ESV (156); Mcm2-7, 3JA8 (40). Only the five central β strands, the Walker A motif with its preceding helix, and the DNA binding loops are shown and colored as in (C). Motor domains are depicted in gray.

Unlike initiator/loaders, cellular replicative helicases actually derive from two distinct ATPase branches. One, exemplified by the DnaB-family helicases of bacteria, uses a RecA-type ATPase fold (34). The other lineage, the MCMs (minichromosome maintenance proteins), is a AAA+ ATPase, albeit from a clade distinct from that comprised by replication initiators (12) (Fig. 1B). Although both DnaB- and MCM-class helicases share a distant ancestral ATP binding module [the additional strand catalytic glutamate (ASCE) or P-loop ATPase fold] (12), their ATPase subunits are oriented in a physically orthogonal way with respect to one another and to substrate nucleic acid in a hexameric ring (35) (Fig. 2, C and D). Consequently, key functional loops and amino acids within bacterial and archaeal/eukaryotic replicative helicases reside on separate secondary structural elements that have evolved convergently, and the two motors travel with opposite polarity on DNA (Fig. 2A).

The central ATPase elements of both DnaB-class enzymes and MCMs have been augmented with various unrelated folds that serve essential functions during replication initiation and DNA unwinding. For example, the N termini of both helicases contain evolutionarily distinct scaffolding domains that can serve as binding sites for ssDNA and that provide a collar of additional contacts that hold the hexameric rings together (3641). MCMs also retain a highly mobile WH fold C terminal to their central ATPase core of variable and enigmatic function (42, 43).

Initiator-dependent origin recognition

Origins are cis-acting DNA sequences that mark sites of initiator-dependent replisome assembly. In prokaryotes and budding yeast such as Saccharomyces cerevisiae, origins are identified by conserved DNA sequence motifs (4447). By contrast, ORC in multicellular organisms shows no DNA sequence specificity for binding (48, 49). In S.pombe and metazoans, origin specificity is fluid and appears to rely on local context such as chromatin status and/or accessibility, rather than on sequence. For example, origin recognition in fission yeast depends on an A/T hook domain appended to Orc4 (50), whereas in metazoans, chromatin accessibility, epigenetic features, and ancillary ORC interactions with other proteins are influencing factors (5156).


Bacterial origins are generally specified by a single locus that contains a number of defined initiator binding sequences. Some of these sequences (e.g., DnaA boxes) conform to a fairly uniform consensus motif that can be occupied by DnaA throughout the cell cycle, whereas others are degenerate and bound only during initiation (44, 5759). The DnaA box appears to be recognized exclusively by the HTH domain of DnaA (13, 60); closely spaced arrays of DnaA boxes in turn promote the ATP-dependent self-assembly of multiple DnaA subunits into a higher-order helical oligomer through lateral AAA+/AAA+ interactions (22, 61, 62). The precise structure of an assembled DnaA/origin nucleoprotein complex has yet to be established, but it likely involves the wrapping of DnaA box–containing DNA segments about the exterior of the DnaA spiral (Fig. 3A). There is evidence in some bacterial species that these extended DnaA/origin superstructures can accommodate short local bends or large loops of DNA as well (6365).

Fig. 3 Initiators bind origin DNA and mark sites for starting DNA replication.

(A) In bacteria, the DnaA initiator is recruited to replication origins by DnaA boxes and other sites using its HTH domain. DnaA binding to duplex DNA facilitates DnaA filament formation and promotes duplex melting; ssDNA is sequestered by the AAA+ domains of a DnaA filament, creating a single-stranded substrate for loading two DnaB hexamers onto complementary DNA strands in opposite directions. Structures of DnaA bound to duplex [PDB: 1J1V (13)] and ssDNA [PDB: 3R8F (14)] are shown as insets. (B) The archaeal initiator Orc binds to specific duplex DNA sequence elements at replication origins, termed mini-ORBs, using its WH domain, and to adjacent DNA using its AAA+ fold. A crystal structure of archaeal Orc1 bound to DNA is shown at right [PDB: 2QBY (17)]. (C) In eukaryotes, the initiator ORC engages duplex DNA in a sequence-independent (S. cerevisiae excepted) but ATP-dependent manner, likely using DNA binding elements that line the ORC’s central channel. A structural model for DNA-ORC (based on the crystal structure of apo-ORC) [PDB: 4XGC (16)] is shown at right, docked into the electron microscopy (EM) structure of an ORC-DNA-Cdc6 complex [Electron Microscopy Data Bank (EMDB): 5381 (31)]. Initiator recruitment to chromosomes is facilitated by secondary DNA- or nucleosome-binding elements.


Like bacteria, the replication origins of archaea are generally demarcated by the presence of select sequence elements. These repeats, known as origin recognition boxes [ORBs (47)], serve as the principal means for localizing Orc-family initiators to multiple origins in archaeal genomes [although not all of them seem to rely on Orc for initiation (66)]. Unlike DnaA, archaeal Orc proteins use a bipartite strategy for ORB recognition, in which the initiator WH domain associates with a conserved origin submotif (known as a mini-ORB), and the outer edge of the central AAA+ ATPase fold binds to a second, less-well-conserved origin segment (15, 17) (Fig. 3B). Multiple ORBs can often be found in close proximity to one another in archaeal origins, sometimes in overlapping pairs, other times in tandem, linear arrays. Although Orc proteins have been shown to bind tightly to ORB sequences (47, 6770), evidence for the formation of higher-order, ATP-controlled Orc oligomers (such as those formed by DnaA or ORC) has been limited, although many members of this family appear to have all the requisite conserved catalytic amino acids necessary to form a canonical, bipartite AAA+-type ATPase site.


In many eukaryotes, including fungi, plants, and metazoans, the hexameric ORC assembly is primarily responsible for specifying prospective origins and sites of replicative helicase loading (71, 72). The five subunits of ORC that contain a AAA+ domain (Orc1 to 5) form a notched ring, whereas the sixth (Orc6) binds to the rest of the particle by a short C-terminal α helix that interacts with an auxiliary domain in Orc3 (16, 73). Only one of ORC’s subunits (Orc1) exhibits ATPase activity (7476). Two others—Orc4 and Orc5—bind but do not appear to hydrolyze ATP, whereas Orc2 and Orc3 seem to have lost the ability to associate with nucleotide entirely (16, 7477).

Structural evidence indicates that the ORC ring encircles DNA, likely using the AAA+ and WH folds of the Orc1 to 5 subunits as binding determinants (16, 78) (Fig. 3C). This interaction may account for the observed coupling between ORC’s ATPase activity and productive origin engagement (49, 71, 75). ATP binding by ORC appears sufficient for helicase loading in vitro, whereas ATP hydrolysis by ORC is inversely correlated with DNA binding and is required for multiple rounds of helicase loading, suggesting that ATP hydrolysis helps to reset ORC for subsequent loading events (74, 75, 7981).

There is increasing evidence, particularly outside of budding yeast, that ORC has multiple means of associating with DNA before the formation of a helicase-loading competent complex with DNA. This capacity derives in part from auxiliary DNA- and chromatin-binding domains such as a bromo-adjacent homology (BAH) domain (found in many Orc1 subunits), a transcription factor IIB (TFIIB)–like domain (Orc6), and an A/T hook repeat (S. pombe Orc4) (50, 56, 82, 83) (Fig. 3C). Thus, in those eukaryl lineages that rely on the existence of relatively hard-wired origin sequences (such as S. cerevisiae), origin specification appears predominantly carried out by the AAA+/WH domains of ORC that encircle DNA. By contrast, local chromatin cues that attract other ORC regions—many of which remain to be elaborated—probably serve this role in most other eukaryotes.

From origin melting to helicase loading and vice versa

Once a DNA region has been selected and marked by the appropriate initiator as an origin, the next step along the initiation pathway involves a series of events that culminate in the placement of two copies of a replicative helicase around complementary single DNA strands. Insight into how hexameric helicases become loaded onto DNA has been influenced by landmark studies of polymerase sliding clamps and their loaders [reviewed in (84, 85)]. Sliding clamps form rings that encircle DNA and latch onto polymerases to improve the processivity of strand synthesis. The clamp loaders themselves are AAA+ ATPases that descend from a clade closely related to that of replication initiators and helicase loaders (12) (Fig. 1B). Clamp loaders operate by forming notched rings that first bind to and stabilize an open-ring state of a sliding clamp and subsequently associate with target DNA (which enters through the notch); the loaded clamp is then released in a closed-ring form (Fig. 4A). ATP binding and hydrolysis control the clamp binding and release cycle, respectively.

Fig. 4 The E. coli helicase loader DnaC follows a clamp loader–like strategy for replicative helicase loading.

(A) Clamp loaders are pentameric AAA+ assemblies that help open ring-shaped sliding clamps. An open clamp-loader–clamp complex binds a primer-template junction within its central pore; ATP hydrolysis by the loader leads to clamp closure and loader dissociation, trapping the clamp on DNA. (B) E. coli DnaC, a monomer in solution, uses its AAA+ domains to self-oligomerize upon binding DnaB (forming a DnaB6-DnaC6 complex), promoting helicase ring opening. ssDNA enters through the crack in the ring to bind the central channel of the DnaBC complex. DnaB ring closure and DnaC release leave the DnaB helicase loaded on ssDNA. Structures of intermediate states for clamp and DnaB loaders are shown in (A) and (B) [PDB: 3U60 (158); EMDB: 2321 (87)].

Because replicative helicases are also rings and the ATP-dependent initiator/helicase-loader proteins are evolutionary cousins of clamp loaders, it has often been assumed that the mechanisms for loading the two classes of proteins would largely parallel each other (78, 86, 87). However, there is substantial variation as to which replicative helicases favor forming naturally open- or closed-ring states and which favor forming stable hexamers versus dissociated monomers (88). As a consequence, a considerable number of divergent, parallel strategies have evolved for catalyzing what is an essential, ancient, and physically challenging process. Even the timing by which helicase loading occurs with respect to origin melting varies in different organisms.


In bacteria, origin melting precedes helicase recruitment and loading (Fig. 3A). The formation of the ATP-dependent nucleoprotein complex between DnaA and duplex origin DNA serves as a precursor for the subsequent melting of adenosine-thymidine (AT)–rich regions situated adjacent to the primary assembly locus (8991). Melting likely occurs by direct interactions between the AT-rich duplex DNA and the AAA+ ATPase domains of DnaA; one of the two melted strands is sequestered and stabilized by a DNA-stretching mechanism akin to that used by the RecA homologous recombination protein (14). Precisely how DnaA’s ATPase domain engages the duplex AT-rich region just before melting has yet to be established but may involve the recognition of a repeating trinucleotide motif by DnaA’s AAA+ domains that has been shown to be important for subsequent stable ssDNA-DnaA filament formation (92).

Once DnaA has properly opened a replication origin, two copies of a DnaB-family helicase are recruited and loaded onto the ssDNA. Thus far, three varying approaches have emerged by which this process occurs (88). In certain Gram-negative bacteria (such as E. coli), six copies of the DnaC helicase loader first bind to and crack open a DnaB hexamer in an ATP-facilitated manner that is highly analogous to how clamp loaders open a sliding clamp (87) (Fig. 4B). An auxiliary N-terminal domain on DnaA helps to localize DnaB-DnaC complexes to the melted origin (by binding to the DnaB N-terminal collar), where one copy of the complex loads onto each of the two single DNA strands in opposite orientations (9396); interactions between DnaA and DnaC may assist this process (30). The role of ATP hydrolysis and the precise order of events following DnaB-DnaC loading remain poorly understood, but it has been observed that ATP-DnaC inhibits DnaB helicase activity, that ATP hydrolysis by DnaC occurs after DnaB loading, and that RNA primer synthesis by the DnaG primase protein helps to eject DnaC (97, 98). Large conformational changes in the DnaB N-terminal region appear not only to help coordinate DnaC release but also to activate the helicase for productive DNA unwinding (37).

In comparison with the ring-opening mechanism used by E. coli, in some Gram-positive bacteria (such as Bacillus subtilis), the replicative helicase appears to be assembled into a hexamer around ssDNA by both a DnaC-family helicase loading factor (known as DnaI) and noncatalytic cochaperones (99). During the loading process, the helicase/helicase-loader complex probably still progresses through a 6:6 intermediate that encircles ssDNA (100, 101), and primase can bind to this dodecamer before helicase loader dissociation. However, precisely when certain factors bind to and release another, or how ATP turnover controls the helicase deposition and release cycle, has not been defined.

Although the use of a dedicated DnaC/DnaI-family helicase loader represents the textbook view of replication initiation in bacteria, there are a number of bacterial species (perhaps a majority) that appear to lack such proteins (23, 24, 102). In such organisms, a recently identified factor, DciA, has been suggested to take the place of DnaC/I protein in the loading process (103). Alternatively, in H. pylori (which also lacks DnaC/I), the replicative helicase has been found capable of assembling into not just a hexameric ring but a “head-to-head” (N terminus to N terminus) double hexamer (104, 105). Expression of this protein in E. coli can overcome a need for DnaC during replication (24), suggesting that it may independently load onto DNA when the correct origin substrate is presented.


The exact order of DNA melting and helicase loading in archaea has yet to be firmly established. Although early studies detected the ATP-dependent deformation of origin DNA by archaeal Orc (67, 106), it is unclear whether this activity reflects bone fide DNA melting (66). Recent work has reported that Orc prebound to ORB sequences can promote the loading of pre-opened Mcm helicases onto duplex DNA; ATP binding to Orc seems to be required for loading, but how nucleotide controls initiator activity is not well understood (26). Once loaded, structural and biochemical studies have found that archaeal Mcms can form double hexamers (as are formed in eukaryotes) (107110) and that archaea have homologs of the eukaryotic Mcm-associated factors GINS (Go, Ichi, Ni, San) and Cdc45 (111113). Nonetheless, the current data indicate that archaeal Orc does not operate by a clamp-loader type mechanism, but rather serves more as a recruitment factor for chaperoning and aligning the helicase around replication origins (26). This divergence is consistent with the notion that DNA replication evolved independently in bacterial and archaeal lineages.


In eukaryotes, helicase loading occurs before origin melting, an order opposite to that of bacteria (Fig. 5). The replicative helicase is first loaded in an inactive state in G1, whereas DNA opening is coupled to helicase activation in S phase (114, 115). ORC and Cdc6, together with Cdt1, collaborate to deposit two copies of the Mcm2 to 7 (Mcm2-7) complex around duplex DNA as a double hexamer that is stabilized by interactions between the Mcm2-7 N-terminal domains (116118). Mcm2-7 loading occurs when cyclin-dependent kinase (CDK) activity is low, because CDK inhibits several loading factors by phosphorylation (119122). During S phase, new rounds of Mcm2-7 loading by ORC are blocked, and a second kinase (the Dbf4-dependent kinase, or DDK) phosphorylates the latent helicase (123), preparing the double hexamer for activation and eventual separation (124, 125). Several other noncatalytic proteins (Sld2, Sld3, Sld7, and Dbp11 in budding yeast), some of which require CDK phosphorylation, then help chaperone two accessory/scaffolding factors—the GINS tetramer and Cdc45—onto Mcm2-7, forming an 11-subunit complex, the CMG (Cdc45, Mcm2-7, GINS), that encircles ssDNA and is now active for DNA unwinding (126128). Overall, the coupling of helicase loading and activation with distinct phases of the cell cycle prevents both the premature formation of ssDNA and rereplication, helping to ensure that a given origin can support only one round of initiation.

Fig. 5 Eukaryotic replicative helicase loading proceeds through a mechanism distinct from the canonical clamp-loader reaction.

ORC binds to DNA and recruits its coloader Cdc6, forming a hexameric AAA+ assembly that encircles duplex DNA. One Mcm2-7 hexamer, which can exist in an open-ring state in solution (139141), is recruited to the DNA duplex by ORC-Cdc6, aided by another coloader, Cdt1. After recruitment, the first Mcm2-7 complex is held in place by ORC, while Cdc6 and Cdt1 dissociate (145). With the help of another Cdc6 and Cdt1, ORC facilitates the loading of a second Mcm2-7 complex onto duplex DNA to yield a head-to-head Mcm2-7 double hexamer (116, 117). At the beginning of S phase, helicase activation promotes duplex DNA melting, CMG formation, and bidirectional replisome assembly. Structures of key intermediate states are shown where available [EMDB: 5429 (141) and 5625 (78); PDB: 3JA8 (40)].

In metazoans, the players in the Mcm2-7 loading and activation process are largely preserved with that of budding yeast, except that several of the noncatalytic chaperones have diverged considerably. For example, Sld3 can be found as a domain in a much larger protein known as Treslin (129131); the metazoan ortholog of Dbp11 is thought to correspond to TopBP1 (132), a BRCA1 C-terminal (BRCT) domain DNA damage response protein; and Sld2 has been proposed to reside in the N terminus of the DNA repair helicase, RecQ4 (133). Metazoans also have evolved a new factor, Geminin, which is not found in budding yeast but which binds to Cdt1 to regulate its activity in the Mcm2-7 loading process (134136).

Why are so many different proteins required simply to establish and open a eukaryotic replication origin? Part of the answer likely lies in the need for eukaryotes to tightly regulate origin firing to respond to changing chromatin landscapes (e.g., for progressing through different developmental fates) and to avoid DNA damage and genetic instabilities arising from aberrant rereplication (8, 137). However, some of the complexity may be due to the path eukaryotes took in loading inactive helicase complexes onto duplex DNA and then needing to awaken these helicases through a complex, programmed isomerization event that leads to origin melting and ssDNA encirclement during S phase. Given the evolutionary kinship between replication initiators and the known ability of DnaA to open replication origins directly, one might have assumed that ORC would serve as a “meltase” that forms the initial replication bubble. However, it has become clear that the primary function of the initiator is to help load the Mcm2-7 complex onto duplex DNA (116, 117) and that it is the activation of Mcm2-7, combined with its own ATPase activity and its metamorphosis into the CMG, that drives duplex separation (127, 138).

How the Mcm2-7 complex is loaded by ORC onto DNA is a major question in the field. One model for this event invokes a clamp-loader type of approach, with ORC and Cdc6 collaborating to crack open an otherwise closed Mcm2-7 ring to deposit the helicase onto DNA (78, 86). However, a distinguishing feature of the clamp-loader reaction is that the loader first binds and opens a target clamp before it places it around DNA (84) (Fig. 4A). This order of events (which is echoed by the action of E. coli DnaC on DnaB) (Fig. 4B) is sensible, because there is no free DNA end to thread through a closed clamp (or a closed helicase ring) during loading. For ORC-Cdc6, however, there is compelling evidence that these two factors form a stable, closed-ring complex that encircles duplex DNA before it can associate with Mcm2-7 and Cdt1 (16, 31) (Fig. 5).

A second difference between clamp-loading systems and Mcm2-7 loading in eukaryotes is that Mcm2-7 complexes [like their archaeal relatives (26)] readily form pre-opened rings (139141). In budding yeast, there is evidence that ATP can help stabilize a closed form of the Mcm2-7 ring (139); however, nucleotide alone has proven insufficient to induce Mcm2-7 hexamer closure in other systems studied to date (140, 141). Although this dichotomy may at first seem contradictory, it likely reflects a simple difference in the tendency of free Mcm2-7 hexamers from different organisms to preferentially equilibrate between open versus closed ring forms, rather than to adopt one or the other state in an all-or-none manner. As for Cdt1, this protein tightly associates with Mcm2-7 in budding yeast (although not in metazoans) (116, 142, 143) and appears to block an innate ability of Mcm2-7 to bind DNA on its own (116, 117). How Cdt1 accomplishes this task is not known; it could either sterically prevent access through the Mcm2/5 gate or it may restrain an Mcm2-7 ring in an open state to prevent the helicase from stably closing around DNA.

Collectively, the available data indicate to us and others (144) that replicative helicase loading in eukaryotes does not proceed through a conventional clamp-loader type of process. Instead, ORC-Cdc6 may act as a combined recruitment platform and alignment jig, akin to what has been seen in archaea (26). In this scenario, the ATPase cycle of ORC-Cdc6 would not actively open a closed Mcm2-7 ring but rather would help facilitate the timely and/or efficient ejection of Cdt1 from Mcm2-7, either unblocking or derestraining the Mcm2-7 hexamer to allow it to stably encircle duplex DNA.

The mechanism for loading of the second Mcm2-7 hexamer is highly enigmatic. The ORC-Cdc6 assembly is inherently asymmetric, yet a single ORC appears to catalyze the formation of a two-fold symmetric Mcm2-7 double hexamer (145) (Fig. 5). Not only does ORC remain associated with the first loaded Mcm2-7 ring but also a new copy of both Cdc6 and Cdt1 is needed to load the second Mcm2-7 hexamer (145). These findings suggest that the ORC-Cdc6 complex holds the first Mcm2-7 ring (in particular the N-terminal collar of the helicase) in an unknown conformation that, after the binding of a new Cdc6 protomer, recruits a second Cdt1-bound Mcm2-7 hexamer through collar-collar interactions (125, 146). One discordant aspect of this model is that it is difficult to reconcile with the finding that the C-terminal region of Mcm3, which directly interacts with Cdc6 (78), appears necessary for the loading of both hexamers (147). There is also evidence that ATP binding and hydrolysis by Mcm2-7 play key, but as yet poorly understood, roles in the loading reaction (81, 138). Nonetheless, the persistence of the initial ORC–Mcm2-7 complex would seem useful for preventing an otherwise unstable single Mcm2-7 hexamer from losing its grip and dissociating from the DNA. The existence of an ORC–Mcm2-7 intermediate also argues against “clamp-closer” types of helicase loading models [as can be catalyzed by the phage T4 clamp loader (148, 149)], instead implicating the stable binding of the second Mcm2-7 hexamer as a motivating factor for ring sealing.

Once formed, the Mcm2-7 double hexamer is a highly stable particle that persists until the onset of S phase (116118). Preparation of the double hexamer for dissociation is promoted by the action of DDK, which phosphorylates the extreme N termini of Mcm2, Mcm4, and Mcm6 (150153); how this event begins to disrupt the stable, highly interdigitated structure formed by the two Mcm2-7 N-terminal collars is uncertain. Even less is known about the next series of events that lead to CMG formation. ATP turnover by Mcm2-7 hexamers has been proposed to play a role in untwisting of duplex DNA (39, 40, 138), but direct evidence for this action is still lacking. Budding yeast Sld3 and Sld2 chaperones (held in proximity to one another by Dpb11) have been found to engage GINS and Cdc45, respectively, as well as to interact with ssDNA (154, 155); these and other findings suggest that there is a highly coordinated handoff between the dissolution of the double hexamer, the timing of origin melting, the opening of the two Mcm2-7 rings to allow escape of one single strand of DNA from each hexamer, and the transfer of GINS/Cdc45 to Mcm2-7.

Concluding remarks

We have highlighted the similarities and differences that nature has used to solve two complex topological problems that occur during the initiation of DNA replication: the loading of ring-shaped replicative helicases onto replication origins and the melting of duplex DNA in preparation for assembling two independent replisomes. An unanticipated outcome of the data available to date is that eukaryotic and archaeal Orc/Cdc6 proteins appear to work in a manner distinct from the structurally related clamp-loader complexes, which bring processivity factors to DNA polymerases. Likewise, DnaA and ORC, which serve similar roles as replication initiators and share both close homology and certain physical characteristics, also operate on origin DNA in a fundamentally different manner, melting the origin in one instance but not the other. These mechanistic differences are striking and raise the question of how and why bidirectional replication is universally conserved. Studies have shown that the rate-limiting steps for replication occur during initiation, underscoring the complexity of simultaneously constructing two forks. Only by continuing to study DNA replication in multiple model systems will we begin to fully appreciate the power of evolution and the creativity of nature in their persistent tinkering with what is one of the fundamental processes of life.

References and Notes

Acknowledgments: All authors contributed to the writing and editing of the manuscript. This work was supported by NIH grants GM071747 to J.M.B. and CA030490 to M.R.B. and J.M.B.
View Abstract

Navigate This Article