Dbf4

Abstract

Together with cyclin-dependent kinases, the Dbf4-dependent kinase (DDK) is essential to activate the Mcm2-7 helicase and, hence, initiate DNA replication in eukaryotes. Beyond its role as the regulatory subunit of the DDK complex, the Dbf4 protein also regulates the activity of other cell cycle kinases to mediate the checkpoint response and prevent premature mitotic exit under stress. Two features that are unusual in DNA replication proteins characterize Dbf4. The first is its evolutionary divergence; the second is how its conserved motifs are combined to form distinct functional units. This structural plasticity appears to be at odds with the conserved functions of Dbf4. In this review, we summarize recent genetic, biochemical and structural work delineating the multiple interactions mediated by Dbf4 and its various functions during the cell cycle. We also discuss how the limited sequence conservation of Dbf4 may be an advantage to regulate the activities of multiple cell cycle kinases.

Introduction

Dbf4 (Dumbbell former 4 protein) is the regulatory subunit of the Cdc7 kinase which plays critical roles in DNA replication,¹ activation of the replication checkpoint,² meiosis,³ translesion synthesis,⁴ histone homeostasis,⁵ and mitotic exit.⁶ Levels of Cdc7 are constant throughout the cell cycle, however it only becomes functional when bound to its regulatory subunit.^7,8 Dbf4 (ASK1 in humans) is an unstable protein whose level of expression peaks late in G₁ and S phase and is degraded at the end of mitosis. Therefore, the kinase activity of Cdc7 fluctuates in a highly predictable manner during the cell cycle, peaking when DNA synthesis begins and disappearing by the end of mitosis.^7,9-12 The interaction between Dbf4 and Cdc7 is robust and is required for the functions of the latter in various chromosomal transactions. Hence activation of Cdc7 has been regarded as the most important role of Dbf4. However, Dbf4 also interacts with other essential cell-cycle kinases including Rad53^13-16 and Cdc5,^6,17 thereby modulating several steps in the cell cycle. While the multiple interactions of Dbf4 have been demonstrated genetically and in some cases biochemically, the mechanisms by which these Dbf4-mediated interactions regulate the cell cycle remain elusive. In the following sections, we will discuss how recent genetic, biochemical and structural work has advanced our understanding of the complex network of interactions mediated by Dbf4.

Dbf4 has a Modular Architecture

DBF4 was originally identified in budding yeast as a gene whose mutation resulted in the arrest of cells with buds creating a characteristic dumbbell morphology.^18,19 Due to limited sequence conservation, Dbf4 homologs were difficult to identify in other species. This was unexpected because proteins involved in DNA replication are among the most highly conserved across the eukaryotic kingdom due to their essential role. Interestingly, certain organisms appeared to lack Dbf4 completely based on their sequenced genomes, particularly protozoan parasites.²⁰ Recent work with the protozoa Giardia duodenalis has empirically identified a Dbf4 homolog in a pull-down assay with Cdc7,²¹ indicating that other organisms initially presumed to lack Dbf4 may also encode highly divergent homologs.

Work from Masai and Arai revealed that only three short sequences (< 100 amino acids) were conserved among all homologs, and were named N, M and C based on their relative location within the polypeptide chain.²² Each of these motifs is associated with a different function of Dbf4 during the cell cycle by participating in distinct protein-protein interactions (Fig. 1).

Figure 1. Cell cycle kinases modulated by Dbf4. Domain organization of Dbf4 and its interaction partners, Cdc5, Cdc7 and Rad53. Known functional domains found in these proteins are color-coded and the regions of interaction are indicated with gray lines.

Motif N (residues 135–179 in S. cerevisiae) does not constitute an independent folding unit but instead is only a small section of a unique BRCT domain.²³ In fungi, this domain (residues 105–220) has an additional N-terminal helix used to build a unique interaction surface for the Rad53 kinase that is absent in the BRCT domains of Dbf4 from higher eukaryotes.¹⁶ While this domain is essential in embryonic mouse cells,²⁴ studies in yeast have shown that motif N is not required for viability, although its deletion causes hypersensitivity to agents that cause fork stalling and defects in late origin activation.^14,15

Motif M is a short proline-rich region that has no resemblance to any known protein fold based on primary sequence.²² This motif includes residues Tyr²⁷³-Phe³²⁵ (numbering refers to S. cerevisiae Dbf4) and is required for the interaction of Dbf4 with the replicative helicase Mcm2-7 in budding yeast.¹⁵ Accordingly, studies in budding and fission yeast have demonstrated that this motif is essential for Dbf4 (Dfp1 in Schizosaccharomyces pombe) function and cell cycle progression.^15,25 A point mutation within this motif decreases the affinity of Dbf4 for Mcm2 and reciprocally increases resistance to replication stress compared with wild-type cells.¹⁵ In certain fungi, a point mutation in motif M prevents the replication checkpoint from arresting the cell cycle in S phase presumably by preventing the suppression of late origin firing,^26,27 in turn suggesting that this mutation could also affect the interaction between Dbf4 and Rad53.

Motif C is a zinc-finger domain located at the C-terminus of Dbf4, likely embedded in a strict structural context because removal, but not point mutations, of the motif confer temperature sensitivity in budding yeast.^28-30 The integrity of the zinc finger is also required for normal replication, as mutants within this motif show delayed entry and progression through S-phase.^28,30 However, there is conflicting evidence regarding the effect of deleting motif C. Using yeast two-hybrid assays, Duncker and coworkers showed that deletion of motif C abolishes the interaction of Dbf4 with Mcm2, Orc2, Rad53 and Cdc7, however point mutations within this motif only affected binding to Mcm2.²⁸ Conversely, Weinreich and coworkers showed C-terminal truncations of Dbf4 that removed motif C and up to ~250 residues before this motif were viable and retain the interaction with Cdc7 in a pull down assay, although these Dbf4 variants were lethal in combination with point mutations in either motif N or M.³⁰ Both studies agree, however, that point mutations within the zinc finger did not affect the interaction between Dbf4 and Cdc7.^28,30 The different behavior of Dbf4 variants lacking motif C in comparison to those encompassing point mutations within the motif could reflect that the structural integrity of the zinc-finger is somewhat stabilized in the context of the Dbf4-Cdc7 complex. This, in turn, would imply that the Dbf4-Cdc7 interaction relies on the structural integrity of motif C rather than its ability to bind zinc.

While only these three motifs are conserved among Dbf4 homologs, the intervening regions of the protein are also important for the cellular roles of Dbf4. For instance, the 20 amino acids upstream of motif N are important for the interaction of Dbf4 with the mitotic kinase Cdc5,¹⁷ and the region connecting motifs M and C, which is not conserved among different species, is important for binding and activating Cdc7.³⁰ Therefore, the crosstalk between conserved motifs of Dbf4 to define complex functional modules, rather than the motifs themselves, is emerging as the critical feature that tunes the roles of Dbf4 during replication initiation and checkpoint activation.³¹

Organization of Dbf4 Functional Modules

Based on the increasing number of genome sequences available, it is clear that there are at least three types of Dbf4 homologs, typified by fungi, animals and insects. It is possible that protozoa constitute an additional class, but more Dbf4 homologs should be identified from this phylum before this can clarified.

Class I: Fungi

Class I Dbf4 proteins are found in fungi. They have motif C at the extreme C-terminus of the protein and include a long linker connecting motifs M and C together. The linker itself does not appear to be necessary, as it can be replaced with a much shorter sequence that brings motifs M and C into close proximity. In S. pombe, these motifs can even be expressed as separate polypeptides and still activate the Cdc7 kinase.³²

Class II: Animals

Class II proteins are of a similar length to class I, but have a short linker between motifs M and C. They also include a C-terminal tail that lacks any additional conserved motifs. This C-terminal tail is able to interact with Cdc7,³³ but this interaction is not necessary for the activation of its kinase activity.³⁴ In fact, the presence of this tail decreases the kinase activity of Cdc7, while the interaction between this region and the lens epithelium-derived growth factor (LEDGF) stimulates Cdc7 activity.³⁴ As LEDGF binds to transcriptionally active DNA in eukaryotes, the Dbf4-Cdc7 complex (often referred as Dbf4-Dependent Kinase or DDK) may use this interaction to target early origins, which are generally found in topologically open DNA.

Class III: Insects

Dbf4 proteins in this class are only found in certain insects. They are twice the size of Dbf4 from other classes, primarily due to the presence of a long C-terminal extension. In contrast to the C-terminal extension found in class II, this region contains three additional conserved motifs.³⁵ However, the role of these additional motifs is unclear. The lifestyle of insects requires DNA replication to purposely over-replicate specific regions of their genome in follicle cells, something that most cells spend considerable amounts of energy to avoid. These regions contain genes that encode for the building blocks of the chorion, which creates the eggshell for embryos. Lowering Dbf4 protein levels in fruit flies results in female sterility due to interference with chorion synthesis, suggesting that the C-terminal extension present in this Dbf4 class may be involved in this process.³⁵

Dbf4 Expression During Cell Cycle and Development

Dbf4 levels peak at the G₁/S boundary primarily due to the inactivation of the anaphase-promoting complex (APC).¹² However, the expression levels of Dbf4 are still limiting compared with other DNA replication initiation proteins.³⁶ Importantly, overexpression of Dbf4 does not lead to an accelerated S phase by overcoming this limitation. Instead, a study conducted in S. pombe revealed that overexpression of DDK did not alter cell cycle kinetics, although it caused an increased efficiency of origin firing.³⁷ This triggered replication fork stalling and activated the replication checkpoint to maintain genomic stability and slow S phase progression. Consequently, cells overexpressing DDK become hypersensitive to the ribonucleotide reductase inhibitor hydroxyurea.³⁷

Along with Sld3, Dbf4 represents an important control for origin firing.^36,38-40 Both of these proteins play key roles in activating the Mcm2-7 helicase. During G₁ phase, the origin recognition complex (ORC), Cdt1 and Cdc6 load Mcm2-7 at origins.⁴¹ However, it must associate with two additional factors before Mcm2-7 becomes a processive helicase: the tetrameric GINS complex and Cdc45.^42,43 Cdc45 binds Mcm2-7 after DDK phosphorylates the Mcm4 and Mcm6 subunits of the complex.^44,45 This relieves an inhibitory action of the Mcm4 N-terminus⁴⁴ and induces conformational changes that ultimately alter Mcm5.⁴⁶ Mcm5 and its neighboring subunit Mcm2 do not associate, which leaves a gate in the hexameric ring. Cdc45 closes this gate by interacting with both subunits, in addition to ATP occupying the composite active site at their interface.⁴³ Cdc45 binds to Sld3 creating the CMS complex, which may induce DNA melting.⁴⁷ Sld2 is also phosphorylated leading to the formation of the pre-loading complex, which consists of Sld2, Dpb11, Pol ε and GINS.⁴⁸ The Dpb11 component is then able to bind phosphorylated Sld3,^48-50 and ultimately GINS makes contact with both the Mcm2-7 and Cdc45 to form the CMG (Cdc45-Mcm-GINS) complex,⁴³ thought to be the functional replicative helicase. Thus, by controlling the assembly of the Mcm2-7 helicase, both Dbf4 and Sld3 are critical targets of the replication checkpoint to inhibit late origins.

Once the genome has been replicated and successfully segregated between the mother cell and bud, the anaphase-promoting complex is activated and Dbf4, as well as the mitotic cyclins, is degraded.^11,12 Low levels of these regulatory subunits persist until late G₁ phase when the cell commits to dividing again. Interestingly, S. cerevisiae cells still transition into S phase normally even when degradation of Dbf4 is prevented by mutating its destruction box.¹¹ Similarly, ectopic expression of Dbf4 and/or Cdc7 to maintain protein levels in human cells does not alter the cell cycle.³³ This is in agreement with a recent study concluding that Cdc6, which is active during G₁ phase, prevents DDK from phosphorylating the Mcm complex.⁵¹ This questions the need to degrade Dbf4 during G₁ phase. In S. pombe, loss of DDK activity is necessary because Cdc7 phosphorylates the transcription factor Ams2, thereby targeting it for ubiquitin-mediated proteolysis.⁵ Since Ams2 activates core histone genes, loss of DDK activity during G₁ phase allows for a burst of histone synthesis in preparation for S phase.

Most eukaryotes express a single Dbf4, however human, mouse and Xenopus cells have a second Dbf4-like protein, referred to as Drf1 (Dbf4 Related Factor 1) or ASKL1 (Activator of S-phase Kinase Like 1).^52-54 Drf1 is also able to bind and activate Cdc7, however this represents an alternate DDK complex as Drf1 and Dbf4 cannot bind to the same Cdc7 molecule simultaneously.⁵² Cells change the distribution of the two DDK complexes by expressing more of one regulatory subunit than the other. For example, Drf1 is highly expressed in embryonic cells whereas adult cells contain mostly Dbf4.⁵⁵ The consequence of this switch between embryonic and adult cells is unknown.

Cancer cells express both Drf1 and Dbf4, however siRNA knockdown of the Drf1 regulatory unit causes a reduction in replication rate while maintaining viability.⁵⁴ This is in contrast to siRNA against Dbf4 or Cdc7 in these same cells, which inevitably results in cell death indicating the different roles of the two DDK complexes.⁵⁴ The fission yeast S. pombe also contains a second Dbf4 protein (Spo4), however Spo4 regulates a second Cdc7-like kinase (Spo6) and this complex is specifically involved in meiosis.⁵⁶

Interactions Mediated by Dbf4

Functionally, DDK is most similar to the cyclin-dependent kinases (CDKs), and together these kinase complexes define the main components of the cell cycle clock. As is a common theme with kinases that regulate decisions in the cell cycle, both DDK and CDKs play multiple roles in different pathways depending on the state of the cell. For a CDK, this is achieved by associating with different cyclin subunits. For example, the single catalytic subunit from budding yeast (Cdc28) ushers this organism through cell division by associating with a different cyclin for each distinct phase. While the D-type cyclins control the decision to enter the cell cycle, the A-type and E-type cyclins control DNA replication in S phase and the B-type cyclins oversee mitosis. This is due to the different substrate binding specificities of the different cyclins.

In contrast, Cdc7 in budding yeast has only one regulatory subunit, Dbf4, which must negotiate a number of protein interactions that change with the phases of the cell cycle, some of which are critical for life. The fact that a single regulatory subunit is able to display such functional diversity and stay in tune with the cell cycle suggests an exquisite level of regulation for the interactions that Dbf4 mediates. We will examine the interactions with Cdc7, Rad53 and Cdc5 to illustrate the molecular versatility of the Dbf4 scaffold.

Interaction with Cdc7

Cdc7 is the kinase subunit of the DDK complex and Dbf4 regulates its activity in an analogous manner to the cyclin-dependent kinases. The activities of DDK and CDKs are necessary throughout S phase to initiate DNA replication from origins. Although functionally similar to CDKs, structurally Dbf4 and Cdc7 bear little in common to cyclins and their catalytic subunits. Based on sequence alignments, Cdc7 is most closely related to the α subunit of Casein Kinase 2 (CK2). Cdc7 homologs are characterized by the presence of three insertions in the canonical kinase fold (Fig. 2). Despite these insertions, it has been proposed that Cdc7 and CK2 are similar because structurally related compounds inhibit both kinases.^29,57

Figure 2. Sequence alignment of Cdc7. Alignment of Cdc7 sequences from Saccharomyces cerevisiae (Sc), Caenorhabditis elegans (Cs), Schyzosaccharomyces pombe (Sp), Homo sapiens (Hs), Mus musculus (Mn) and Xenopus laevis (Xl). The predicted secondary structure of S. cerevisiae Cdc7 is included above the alignment with the conserved kinase regions shown in blue and the Cdc7-specific insertions indicated in red. Conserved hydrophobic, polar, positively charged and negatively charged are highlighted in yellow, green, pink and blue, respectively. The inset shows a ribbon diagram of the conserved catalytic core of human Cdc7 that was determined by deleting the characteristic Cdc7 inserts (PDB ID: 4F99). The nucleotide is shown in green and the deleted regions indicated as red dotted lines with the missing residues indicated.

The crystal structure of Cdc7’s catalytic core bound to Dbf4 has recently confirmed this idea.²⁹ The catalytic core of Cdc7 and its regulatory subunit, Dbf4, interact through a bipartite interface involving motifs M and C of Dbf4, but not the intervening region between the two motifs,²⁹ though this may not be the case for the interaction of full-length Dbf4 and Cdc7.^30,31 Motif C interacts with the N-terminal lobe of Cdc7. Zinc-binding causes motif C to adopt a very compact structure that encloses the N-terminus of helix αC, however the interaction is further stabilized by extensive interactions between the helix following the zinc finger and helix αC of Cdc7 (Fig. 3) thereby explaining why Dbf4 variants lacking motif C, but not those encompassing point mutations that impair zinc binding, abrogate the interaction with Cdc7.²⁸ Due to its role at stabilizing helix αC, motif C is strictly required for the kinase activity of Cdc7. Notably, the Cdc7-Dbf4 complex crystallized in a nucleotide-bound conformation that is strikingly similar to that of the activated CDK2-Cyclin A complex, confirming that Dbf4-dependent and cyclin-dependent kinases are activated similarly despite their unrelated regulatory subunits.²⁹

Figure 3. Structure of the catalytic core of the human DDK complex. 130° away views of the crystal structure by Hughes and coworkers, with the catalytic core of Cdc7 shown as a semi-transparent surface and Dbf4 depicted as a purple ribbon diagram. The Cdc7 insertions are colored in pink and helix αC in blue. The nucleotide is shown as brown sticks with the coordinating Mg²⁺ metal ion shown as an orange sphere. The side chains defining the zinc finger of motif C are shown as color-coded sticks and the Zn²⁺ metal ion is shown as a teal sphere.

This structure provides the general strokes to explain the Dbf4-dependent activity of Cdc7 and a framework to design Cdc7 inhibitors.²⁹ However, crystallization came at the price of truncating features that are unique to Cdc7, probably the reason why the specific activity of this complex is only 10% that of DDK.²⁹ Kinase insert I connects the second and third strands of the β-sheet found in the N-terminal lobe of Cdc7 and it can be as short as three amino acids in the human homolog, although it is much longer in yeast (18 amino acids in S. cerevisiae). Dbf4’s motif C runs through the back surface of the N-terminal lobe of Cdc7 and interacts with this loop, thereby confirming previously published data showing the species dependency of the Dbf4-Cdc7 interaction.³¹

All kinases have their active site in a cleft between the N- and C-terminal lobes of their kinase domain, and the activity is often controlled by the orientation of a region called the activation segment.⁵⁸ This segment is normally disordered in an inactive kinase, but it is stabilized in the active form.⁵⁸ CK2 is constitutively active—and hence the exception to the rule—because its activation segment is properly oriented even in the absence of substrate.^58,59 In Cdc7, the kinase insert II defines the activation loop, but it has no role in binding to Dbf4 in humans.⁶⁰ Similarly to the ERK2 kinase, insert II is apparently involved in regulating nuclear import.^61,62 However, as the activation loop is the main substrate-binding site, insert II may modulate the activity of Cdc7 by other mechanisms. Motif C, which is spatially close to insert II, is immediately preceded by a strictly conserved KEKKKK sequence that is important for Cdc7 activity.²⁹ It is conceivable that these two motifs interact either directly or indirectly to enhance substrate binding or catalytic activity, however neither insert II nor the KEKKKK motif are present in the Dbf4—Cdc7 structure and, thus, their roles remain unclear.

The kinase insert III is the only insertion located in the C-terminal lobe of Cdc7 and it is important for binding to Dbf4’s motif M.^29,60 The β-hairpin at the N-terminus of motif M forms an anti-parallel β-sheet with the β-strand in the kinase insert III and interacts extensively with helices αEF and αG (Fig. 3). The rest of motif M adopts an extended conformation that tracks the surface of the C-terminal lobe of the kinase. Variations in length and sequence of insert III seem to correlate with variations on the sequence and spacing between the conserved motifs of Dbf4 in different organisms. Indeed, it has been recently shown that the interactions that hold the DDK complex are species specific while substrate recognition is not,³¹ reinforcing the idea that crosstalk between domains rather than sequence conservation is what allows Dbf4 to regulate the activity of its binding partners.

Cdc7 also has a C-terminal extension that is necessary for the interaction with motif C of Dbf4,⁶⁰ but this region is also disordered in the structure of the catalytic core of Cdc7 bound to Dbf4.²⁹

Interaction with the effector kinase Rad53

The S-phase checkpoint pathway coordinates DNA replication, DNA repair and cell cycle progression, by regulating several processes including stabilization and restarting of stalled forks and firing of late replication origins.⁶³ Accumulation of ssDNA at stalled forks triggers the recruitment of the Mec1 kinase,^64,65 which creates phosphoepitopes at the site of damage and leads to the recruitment and phosphorylation of the effector kinase Rad53.^66-68 Rad53 is further hyperphosphorylated by upstream kinases leading to its full activition.^69-71 Aside from directly activating Rad53, these additional kinases also create phosphoepitopes in both Rad53 and its substrates that promote interactions between them. DDK is required to maintain the hyperphosphorylated state of Rad53, thereby sustaining the checkpoint response.²

Rad53 is composed of a kinase domain flanked by two forkhead associated (FHA) domains that are each preceded by a serine cluster domain (SCD) (Fig. 1). FHA domains are small phosphoepitope binding domains that mediate protein-protein interactions with a strict preference for phosphothreonine (pThr) containing epitopes. A conserved arginine residue (Arg70 in the FHA1 domain of S. cerevisiae Rad53) that recognizes the pThr moiety, and a variable pocket that recognizes a specific type of residue three amino acids away from the pThr are responsible for the strict substrate specificity of FHA domains.⁷² The substrate specificity of FHA domains has been characterized using short phosphopeptides, however full-length binding partners may have additional interactions beyond the phosphoepitope binding site^73-75 or even interact in a phosphorylation-independent manner.⁷⁶

The FHA1 domain of Rad53 interacts with the N-terminal region of Dbf4 (Fig. 1). The minimal fragment of Dbf4 necessary for the interaction with Rad53 surrounds motif N and folds as a modified BRCT domain that includes an additional helix necessary for the stability and functionality of the domain.¹⁶ Therefore, this domain is often referred to as HBRCT domain to distinguish it from the non-functional BRCT domain and describe the relative location of the additional helix.¹⁶ Since a variant of Rad53 encompassing an Arg70Ala mutation abrogated the interaction with Dbf4 in a yeast two-hybrid assay,¹³ it was presumed that the Dbf4-Rad53 interaction was dependent on Dbf4 phosphorylation. However, all the threonine residues within the HBRCT domain of Dbf4 can be mutated without affecting the interaction, indicating that the FHA1 and HBRCT domains of Rad53 and Dbf4, respectively, interact in a phosphorylation independent manner.¹⁶ The structural characterization of the Rad53-Dbf4 complex has now confirmed that the interaction between the FHA1 and HBRCT domains does not involve the phosphopeptide binding site of FHA1 (L.A.M. and A.G., unpublished data), adding to the growing number of non-canonical interactions mediated by FHA domains (reviewed in ref. 77).

Yeast two-hybrid analysis revealed that the interaction between Rad53 and the HBRCT domain of Dbf4 is stronger than that with full-length Dbf4, implying that some interactions mediated by Dbf4 may be mutually exclusive.¹⁶ This is an interesting idea because the mitotic kinase Cdc5 (the S. cerevisiae homolog of human Plk1) inhibits hyperphosphorylation of Rad53 leading to checkpoint adaptation and interacts with a region of Dbf4 that is in close proximity to the HBRCT domain.^17,78 If cells encounter persistent or irreparable damage, the checkpoint undergoes adaptation to overcome cell cycle arrest, wherein protein turnover and the activation of phosphatases reduces Rad53 activity.⁷⁹ Cdc5 and Cdc28 (the S. cerevisiae homolog of human Cdk1) phosphorylate Rad53 in a damage-independent manner in cells from G₂/M until the G₁/S transition of the next cycle.⁸⁰ This allows Rad53 to remain active for a longer period of time before adaptation and thus makes Rad53 easier to activate due to the presence of ssDNA.⁸⁰ During S phase, however, this phosphorylation is lost, although the activity of the phosphatases responsible for checkpoint adaptation are not required for this to occur.⁸⁰ Consequently, Rad53 is not activated unless extensive regions of ssDNA are exposed at stalled forks. This indicates an intricate functional interplay between Dbf4, Rad53 and Cdc5 at different stages of the cell cycle.

Interaction with the Cdc5 kinase

Cdc5 regulates mitotic progression at a number of steps including sister chromatid cohesion, spindle dynamics and mitotic exit.⁸¹ Cdc5 levels oscillate during the cell cycle by increasing in G₂ phase and disappearing at the end of mitosis due to the activity of the anaphase-promoting complex.⁸² Cdc5 is composed of a N-terminal kinase domain that belongs to a subfamily of Ser/Thr kinases and two polo-boxes (PB) that mediate protein-protein interactions (Fig. 1). The two polo boxes, and a small motif dubbed the polo cap, fold together to form the polo-box domain (PBD) that functions as a phosphopeptide-binding domain. Cdk1 phosphorylates the activation loop of Cdc5, as well as a number of Cdc5 targets that can then be recognized by the polo-box domain.⁸³

The structural and functional characterization of the human Cdc5 homolog (Plk1) in complex with a variety of binding partners has delineated the mechanisms of phosphopeptide targeting and regulation by the polo-box domain.^84-86 Like FHA domains, the PBD forms a β-sandwich, where each polo box contributes one β-sheet to the β-sandwich and the cleft between the two polo boxes defines the phosphoepitope-binding pocket (Fig. 4).^84,85 The polo-box domains can also bind ligands through alternative surfaces^87,88 and even mediate phosphorylation-independent interactions, as is the case for its interaction with DDK.¹⁷ Dbf4 interacts with the polo-box domain of Cdc5 through a non-canonical interface¹⁷ and this interaction only requires a short sequence in Dbf4 immediately upstream of the HBRCT domain (residues 83–88 in S. cerevisiae Dbf4). Cells carrying a deletion of this region in Dbf4 fail to arrest in mitosis and exit prematurely from the cell cycle.¹⁷ This interaction does not inhibit the kinase activity of Cdc5, but it presumably alters its polo-substrate targeting, thereby inhibiting the polo-mediated activation of the mitotic exit network.⁶ This region of Dbf4 is highly prone to protease degradation implying that it is probably flexible and devoid of secondary structure.¹⁶ In keeping with this idea, the interaction between Dbf4 and the PBD of Cdc5 could be recapitulated using a short peptide.¹⁷ These short Dbf4 peptides do not outcompete Cdc5′s ability to bind to phosphopeptides and, reciprocally, mutation of the phosphopeptide-binding site does not disrupt the interaction between Cdc5 and Dbf4.¹⁷ These results collectively demonstrate that the interaction between Cdc5 and Dbf4 is not mediated by the phosphoepitope binding site of Cdc5’s PBD, and suggest that Dbf4—like in the case of Rad53—has devised a unique mode of interaction with Cdc5.

Figure 4. Structure of human Plk1 bound to a phosphopeptide. Ribbon diagram of Plk1 with the polo cap and the two polo boxes shown in dark and light green respectively. The phosphoepitope is shown as a purple ribbon with the side chain of the phosphorylated threonine shown as color-coded sticks.

An adaptable Decision Maker in the Cell Cycle

A number of kinases including CDKs (Cdc28 regulated by a cyclin subunit), Cdc7, Cdc5 and Rad53 work together to drive cell cycle progression, decide when to start or prevent DNA replication and halt the cell cycle during replication stress or DNA damage. While their primary functions are confined to specific points of the cell cycle, they are inter-regulated through direct and indirect means. At the center of it all is Dbf4 interweaving DNA replication stress and fork stalling with cell cycle pausing and prevention of origin firing while forks are being repaired and re-started (Fig. 5). Its architecture builds upon a loose rubric: two short regions of the protein joined by a variable linker that bind and activate Cdc7 at the onset of S phase. The length and sequence of this linker correlates with that of the Cdc7 kinase inserts, reflecting a concerted evolution of the two subunits of the DDK complex. Most intriguing are the interactions between Dbf4 and either Rad53 or Cdc5, because they occur through non-canonical surfaces that preserve normal phosphoepitope targeting of these two kinases. This strategy is reminiscent of that employed by some lesion bypass polymerases to enhance DNA polymerase switching,⁸⁹ whose activity, like that of Dbf4, is also timed with the cell cycle.

Figure 5. Specific roles of Dbf4 during the cell cycle. Dbf4 exerts its functions at three points of the cell cycle: origin firing (by activating the Cdc7 kinase), replication checkpoint (through its interaction with Rad53) and mitotic exit (through Cdc5).

In contrast to most proteins involved in DNA replication, Dbf4 is highly divergent and, yet, its functions are conserved among eukaryotes. Dbf4 may have exploited this evolutionary plasticity to accommodate the needs of different species. Indeed, the lack of intrinsic enzymatic activity may have enabled a higher evolution rate than the kinases it relates. While this makes studying Dbf4—and extrapolating results from one eukaryotic homolog to another—difficult, it also makes the enterprise extremely valuable. It has been noted that mutating certain motifs cause phenotypes consistent with changes in interactions mediated by distant regions of the protein. For example, a point mutation in motif M makes cells hyper-resistant to replication stress,¹⁵ even though it is motif N that mediates the interaction with Rad53. In addition, removal of motif C destroys interactions mediated by both motif M and the BRCT domain.²⁸ Research during the last decade has reinforced the idea that the conserved motifs in Dbf4 do not function as independent units. Therefore, the crosstalk between motifs is emerging as the critical feature determining Dbf4 function. Consequently, and although much have been learned through isolating particular domains and dissecting their modes of interaction, future studies must assess the function of these regions in the context of the full-length proteins to reveal how different organisms have exploited Dbf4’s plasticity within the evolutionary “rigid” cell cycle engine. This poses technical challenges, but it embraces the Aristotelian philosophy that the whole is greater than the sum of its parts and it will provide the key to understanding how Dbf4-mediated interactions are spatially and temporally regulated during the cell cycle.

Glossary

Abbreviations:

APC

anaphase promoting complex

CDK

Cyclin-dependent kinase

DDK

Dbf4-dependent kinase