Bioscience Reports

Review Article

Control of cell cycle progression by phosphorylation of cyclin-dependent kinase (CDK) substrates

Randy Suryadinata , Martin Sadowski , Boris Sarcevic

Abstract

The eukaryotic cell cycle is a fundamental evolutionarily conserved process that regulates cell division from simple unicellular organisms, such as yeast, through to higher multicellular organisms, such as humans. The cell cycle comprises several phases, including the S-phase (DNA synthesis phase) and M-phase (mitotic phase). During S-phase, the genetic material is replicated, and is then segregated into two identical daughter cells following mitotic M-phase and cytokinesis. The S- and M-phases are separated by two gap phases (G1 and G2) that govern the readiness of cells to enter S- or M-phase. Genetic and biochemical studies demonstrate that cell division in eukaryotes is mediated by CDKs (cyclin-dependent kinases). Active CDKs comprise a protein kinase subunit whose catalytic activity is dependent on association with a regulatory cyclin subunit. Cell-cycle-stage-dependent accumulation and proteolytic degradation of different cyclin subunits regulates their association with CDKs to control different stages of cell division. CDKs promote cell cycle progression by phosphorylating critical downstream substrates to alter their activity. Here, we will review some of the well-characterized CDK substrates to provide mechanistic insights into how these kinases control different stages of cell division.

  • cell cycle
  • cyclin-dependent kinase (CDK)
  • phosphorylation
  • protein kinase
  • substrate specificity
  • target substrate

INTRODUCTION

The phosphorylation of proteins on serine, threonine and tyrosine residues by protein kinases is a fundamental post-translational modification that is important for regulating protein function in nearly all eukaryotic pathways, such as cell proliferation and differentiation, cell mobility and motility, metabolism and apoptosis. The human genome encodes for over 500 protein kinases, which represents approx. 2% of all human genes, exemplifying the central role of protein phosphorylation in regulating cellular function [1]. Deregulation of protein kinase activity plays a central role in many human diseases (reviewed in [26]).

Ultimately, protein kinases mediate their effects by phosphorylation of specific downstream substrate(s) to alter their biochemical and biological function(s) and elicit specific cellular responses. Considerable effort has been directed at identifying and characterizing protein kinase substrates to understand how protein kinases mediate their biological effects. Substrates need to satisfy several criteria to validate that they are bona fide substrates of a particular protein kinase. These include establishing that the substrate is phosphorylated by the protein kinase on the same sites in vitro and in vivo. The substrate must be localized in the same cellular compartment as the protein kinase during the period in which it is active. Finally, phosphorylation should be of sufficient stoichiometry to induce biochemical changes in the substrate to mediate biological effects.

CDKs: KEY REGULATORS OF THE CELL CYCLE

The critical role of protein kinases in central biological processes is exemplified by the CDKs (cyclin-dependent kinases), which control cell cycle progression in eukaryotic organisms. This evolutionarily conserved process is fundamental in regulating cell division from unicellular organisms, such as yeast, through to higher organisms, such as mammals. The cell division cycle is characterized with four distinct phases, including two gap phases (G1 and G2) that prepare and assess the readiness of cells to enter the S-phase (DNA synthesis phase) or the M-phase (mitotic phase) [7] (Figure 1). Strict control of S-phase is important to ensure that cells only undergo a single round of chromosomal DNA replication, while M-phase governs cytokinesis, where the duplicated genetic material is segregated into two identical daughter cells [7]. In mammalian cells, commitment to cell division is regulated by mitogenic factors that act in the G1-phase of the cell cycle to activate CDKs and transverse a point, termed the restriction point (R), after which they are committed to a round of cell division independent of growth factors [8].

Figure 1 Cyclin–CDK regulation of the mammalian cell cycle

The cell cycle consists of a DNA synthesis (S) phase and a mitotic (M) phase, separated by two gap (G1 and G2) phases. In mammalian cells different cyclin–CDK complexes regulate progression of cells through the different phases of the cell cycle.

Progression of cells through the cell cycle is tightly regulated by cyclin–CDK dimeric kinase complexes, where the CDK is the catalytic kinase subunit and the cyclin is the activating subunit. Schizosaccharomyces pombe and Saccharomyces cerevisiae CDCs (cell division cycles) are regulated by a single CDK protein, p34cdc2 and Cdc28 respectively, coupled with several different cyclins that are expressed at different cell cycle phases [9,10]. In contrast, higher organisms possess different CDKs that associate with the different cyclins and are important for progression through different cell cycle phases. Therefore, mammalian cell culture studies have demonstrated that CDK4 and CDK6 in association with D-type cyclins are activated and important for progression during the G1 cell cycle phase [11,12]. The cyclin E–CDK2 complex is important during transition from G1- into S-phase [13]. Cyclin A–CDK2 and cyclin A–CDK1 are important during S- and G2-phase progression respectively [1417]. Finally, progression of cells through mitosis is dependent on cyclin B–CDK1 [18] (Figure 1). Given the importance of cyclin–CDKs in mediating cell cycle progression, these kinases are tightly regulated at several levels to ensure that cell cycle progression occurs only under appropriate circumstances when cell division is required. Therefore, in addition to the binding of activating cyclins, CDKs are regulated by the binding of inactivating proteins (cyclin-dependent kinase inhibitory proteins) and activating and inactivating phosphorylation events on critical residues of the CDK subunit (reviewed in [3,1922]).

The fundamental roles of cyclin–CDK complexes for cell cycle progression and development in mammals has been extensively studied in mouse models, through knockout or overexpression of genes expressing particular cyclin or CDK subunits. These studies have unveiled significant complexity in terms of developmental importance and tissue-specific action and redundancy of CDK action. For example, mice lacking cyclin D1 display abnormalities associated with retinal development and mammary gland development during pregnancy [23,24]. In addition, these mice have reduced body size when compared with the wild-type litter-mates [23,24]. In contrast, cyclin-D2-deficient mice exhibit impaired proliferation of peripheral B-lymphocytes [25,26]. Furthermore, cyclin-D2-deficient female mice are sterile, whereas the males retain their fertility, but exhibit smaller testes [27]. Conversely, cyclin-D3-deficient mice display impairment in the thymus-derived T-lymphocyte development [28]. These studies demonstrate tissue-specific action of D-type cyclins. However, mice lacking all three D-type cyclins die by E16.5 (embryonic day 16.5) [29], indicating that the presence of at least one D-type is essential for viability. Similarly, mice lacking either cyclin E1 or E2 do not show a significant impact on their development; however, double knockout of the E-type cyclins is embryonically lethal by E11.5 due to defects in endoreplication in placental trophoblast cells [30]. Mice lacking either cyclin B1 or cyclin B2 expression demonstrated that, although cyclin-B2-deficient mice display normal developmental patterns, no pups carrying homozygous cyclin B1 deletion survived [31]. Interestingly, cyclin B1 expression compensates the loss-of-function of cyclin B2-null mice, but not vice versa [31]. Mice lacking either CDK4 or CDK6 remain viable, despite displaying defects in the proliferation in specific cell types. CDK4 is important for lactotroph-derived prolactin production and secretion [32]. Secretion of prolactin is important for the female reproductive system as it maintains the luteal phase (late phase of the menstrual cycle), which, in turn, is important for pregnancy. CDK4-deficient female mice are sterile [32]. In addition, CDK4-null mice develop diabetes due to impaired pancreatic β-cell proliferation [33]. In contrast, CDK6 plays a role in the expansion of differentiated haemopoietic population [34]. Deletion of both CDK4 and CDK6 results in lethality during late embryogenesis, although fibroblasts derived from early embryos proliferate in vitro, suggesting that other CDKs can compensate for their loss during G1 cell cycle progression [34]. CDK2-null mice are viable for up to 2 years, indicating that CDK2 is one of several cell cycle genes that appear to be dispensable [35,36]. However, both male and female CDK2-deficient mice are sterile, indicating that CDK2 plays a crucial role in germ cell development [35,36]. Unlike CDK2, CDK4 and CDK6, CDK1 is important for early stage embryonic development [37]. Interestingly, knockout studies demonstrate that CDK1 can interact with cyclins D and E to form active kinase complexes to compensate for the loss of CDK4 and CDK2 [37]. This study [37] challenges the absolute neccessity of CDKs 2, 4 and 6 in regulating cell cycle progression. However, although CDK1 is able to partially compensate the role of CDKs 2, 4 and 6 in general, their presence is necessary to phosphorylate specific target substrates important for proliferation of specialized cell types [37]. Altogether, the whole-animal studies with knockout mice demonstrate tissue-specific roles for different cyclins and CDKs, and significant functional redundancy of the different family members.

APPROACHES FOR IDENTIFYING PROTEIN KINASE SUBSTRATES

The identification of protein kinase substrates has represented a significant challenge, with technical limitations to the various approaches adopted by different laboratories. Indeed, as yet there is no single method that can simply identify the cellular substrate complement of a protein kinase. Irrespective of the approach used, further studies are required to validate that a protein isolated by screening methods is a bona fide substrate of a particular protein kinase. This is further complicated by the possibility of multiple phosphorylation sites, as well as the same sites being targeted by multiple protein kinases.

(i) Radioisotope labelling in vivo

Earlier approaches to identify in vivo protein kinase substrates involved incubating cells with [32P]Pi radioisotope to radioactively label intracellular ATP pools. Following activation of the kinase, phosphoproteins can be separated on one- and two-dimensional polyacrylamide gels and analysed to identify potentially phosphorylated substrates [3840]. A major limitation of this approach is the poor signal-to-noise ratio as many cellular proteins are phosphorylated, leading to high background phosphorylation and difficulty in identifying specifically phosphorylated substrates, particularly low-abundance protein substrates.

(ii) Radioisotope labelling in vitro

Due to the limitations of these in vivo-based approaches, most laboratories have utilized in vitro-based phosphorylation approaches, where kinases are incubated with proteins or cellular extracts under appropriate phosphorylation conditions to identify putative substrates. These approaches also suffer from background phosphorylation issues, since cell lysates contain numerous endogenous kinases and substrates, making it difficult to identify specific substrates of the exogenous kinase. Different strategies have been adopted to circumvent these limitations. For example, several groups have employed chromatographic techniques to separate cellular extracts into different fractions to reduce the background phosphorylation and enable detection of substrates phosphorylated by added kinase [4143] (reviewed in [44]). An alternative in vitro approach involves engineering a mutant protein kinase in the ATP-binding domain that specifically utilizes an analogue of ATP that cannot be utilized by other cellular kinases [4547]. Incubation of cellular extract with the engineered kinase and the ATP analogue thus results in specific phosphorylation of substrate proteins by the engineered kinase with no background phosphorylation from other kinases. Phosphorylated proteins can then be separated on polyacrylamide gels, isolated and identified by MS. These techniques allow for phosphorylation of substrates in their native conformation, but a potential limitation is that low-abundance proteins may be harder to detect or identify by MS. This approach has successfully been used to identify substrates of Raf1, JNK (c-Jun N-terminal kinase) and CDKs [4749].

(iii) Phosphorylation of phage-display or cDNA expression libraries

In addition to phosphorylation of proteins from cell extracts, other approaches have involved phosphorylation screening of recombinant proteins expressed from λ-phage cDNA expression libraries [50,51]. This approach involves expression of recombinant proteins from phage cDNA expression libraries that are immobilized on nitrocellulose filters and incubated with recombinant protein kinase in the presence of [γ-32P]ATP. Phosphorylated proteins are visualized by autoradiography and identified by sequencing of the cDNA of the corresponding phage clone. In principle, this method allows for the equivalent expression of different recombinant proteins and thus is not biased towards the detection of highly abundant proteins. However, a caveat of this technique is that bacterial expression may result in partially expressed protein and incorrect folding, leading to false positives. This approach has been used to identify substrates of MAPK (mitogen-activated protein kinase) [52] and CDKs [53]. In addition to phosphorylation of bacterially expressed proteins, other groups have phosphorylated proteins following in vitro transcription and translation from a cDNA library in a reticulocyte lysate system [54]. Protein expression in this system avoids the issues of incorrect protein folding that may occur with bacterial expression. Expressed proteins were mixed with mitotic cell extracts, and phosphorylated substrates were identified using a mitosis-specific anti-MPM-2 (mitotic protein monoclonal 2) phospho-specific antibody or by reduced mobility on SDS/PAGE.

(iv) Phosphorylation of peptides

Apart from phosphorylation of proteins, a number of laboratories have utilized peptide-based phosphorylation screening approaches to identify the optimal phosphorylation motif of a particular protein kinase. Protein kinases phosphorylate residues in substrates within a defined consensus sequence context [55,56]. Initially, these were established using synthetic peptides corresponding to local phosphorylation site sequences of native substrates [55], but were subsequently superseded by combinatorial peptide libraries to determine the consensus phosphorylation sequences for PKA (protein kinase A), CDKs, SLK1 (STE20-like kinase 1), protein kinases CK1 and CK2, ERK1 (extracellular-signal-regulated kinase 1), calmodulin-dependent kinase, phosphorylase kinase and NIMA (never in mitosis gene A of Aspergillus nidulans) protein kinase has been defined [5759]. By determining the consensus phosphorylation motif for a protein kinase, bioinformatic approaches have been used to identify substrates of IKKβ and IKKε (IκB kinases β and ε) [60,61]. The potential limitations associated with this method is that, in addition to substrate consensus phosphorylation motifs, kinases often require other structural aspects to recognize substrates, such as binding motifs [6264]. In addition, if a consensus motif for a particular kinase is not highly specific or stringent, bioinformatic searches are difficult as the motif will be present in many proteins, necessitating the use of additional selection criteria.

(v) Yeast two-hybrid screen

Another approach that has been used to identify protein kinase substrates is the yeast two-hybrid interaction screen, which relies on interaction between the kinase and substrate [6568]. However, this technique is prone to isolation of interacting proteins that may not be substrates of a protein kinase, but binding proteins or regulatory subunits.

(vi) Structural basis for substrate predicition: Predikin

In addition to direct experimental approaches, an in silico predictive approach termed ‘Predikin’ has been developed to identify putative protein kinase substrates [69,70]. This approach uses a set of rules based on either the known crystal structure of a protein kinase or defining the substrate determining residues in the catalytic domain of a protein kinase based on its primary sequence. This information is then used to predict which heptapeptide sequences are the best substrate for a particular protein kinase, which can then be used to search sequence databases. A limitation of this approach is that some kinases utilize substrate-specificity determinants in addition to the sequences proximal to the phosophorylated residue, such as anchoring or adaptor proteins, scaffolds or subunits. Residues within the consensus site may be phosphorylated which may affect specificity. In addition this approach does not take into account if a specific protein kinase and putative substrate are co-localized to ensure catalysis.

SUBSTRATES OF CDKS

Since CDKs promote cell division, defining their substrates is critical to understanding the mechanistic basis of their action. Although numerous CDK substrates have been described in the literature to date, a detailed understanding of how CDK-mediated phosphorylation controls their function and cell cycle progression at a mechanistic level has been well characterized for only some of these. It is not within the scope of this review to discuss all CDK substrates reported to date. A relatively comprehensive catalogue of CDK substrates described to date can be found at a public database (www.phosphosite.org). Rather, we will describe some of the key substrates whose CDK-mediated phosphorylation has been well characterized to provide mechanistic insights into the key role of these kinases in controlling cell cycle progression.

Analysis of the CDK phosphorylation sites in proteins and peptides has revealed that these enzymes are proline-directed serine/threonine kinases. Hence, CDKs require a proline immediately C-terminal to the phosphorylated serine or threonine residue (S/T-P) [5759]. Although this recognition motif is the minimal requirement for CDKs to phosphorylate substrates, in many substrates the optimal phosphorylation motif generally consists of a positively charged lysine or arginine three positions downstream of the phosphorylated site (S/T-P-X-R/K) [5759]. Although the consensus phosphorylation motif for different CDKs is conserved, CDKs can be specific to particular substrates. A major mechanism for regulating substrate specificity of CDKs is via the bound cyclin subunit. For example, cyclins E and A possess a hydrophobic patch, Met-Arg-Ala-Ile-Leu (M-R-A-I-L), that recognizes and interacts with an Arg-X-Leu (R-X-L) cy motif (cyclin-binding motif) within substrates [6264]. Mutation of either the cy motif in the substrate or the hydrophobic patch in the cyclin impairs cyclin E–CDK2 or cyclin A–CDK2-mediated phosphorylation of particular substrates [63,64,71]. Although cyclin E–CDK2 and cyclin A–CDK2 have overlapping substrates, each can also target unique substrates, indicating that other distinct features of cyclins E and A can control substrate specificity [42]. In addition to regulating phosphorylation of different substrates, different CDKs can phosphorylate the same substrate on different sites to regulate different aspects of substrate function. For example, cyclin D1–CDK4 and cyclin E–CDK2 phosphorylate different sites on pRb (retinoblastoma protein) to regulate different aspects of its transcriptional inhibitory activity [72]. The mechanism underpinning this selectivity is unclear, but may be related to the cyclin subunits binding to different regions of pRb. Another level of CDK-mediated substrate specificity involves differential subcellular localization. Both cyclins E and A localize and are active with their respective CDKs in the nucleus [7375]. Conversely, cyclin B1 exhibits dynamic subcellular trafficking between the cytoplasm and nucleus during the cell cycle. Synthesis of cyclins B1 and B2 begins during interphase and their localization is restricted to the cytoplasm. As the cells progress into prophase, cyclin B1 is abruptly translocated into the nucleus, whereas cyclin B2 remains in the cytoplasm [76,77]. The ability of cyclins B1 and B2 to maintain their cytoplasmic localization during interphase is dependent on an N-terminal region of 42 amino acid residues, namely the CRS (cytoplasmic-retention sequence) [77]. The dynamic cell-cycle-dependent cytoplasmic and nuclear localization of cyclin B1 is related to the diverse range of substrate targets of cyclin B–CDK1 complexes, whose phosphorylation is required for the major cytoskeletal and nuclear alterations that occur during the mitotic transition. For example, phosphorylation of the nuclear lamins is thought to contribute to the breakdown of the nuclear envelope during mitosis [78,79]. Other cytoskeletal proteins include the centrosomal protein of 55 kDa, Cep55 [80], and caldesmon [8183].

(i) CDK substrates during G1–S-phase

pRb

A major substrate of cyclin D–CDK4/6 in G1 phase is pRb, a tumour suppressor protein of 928 amino acids [84]. pRb inhibits G1–S-phase cell cycle progression, and this inhibitory effect is reversed by CDK-mediated phosphorylation (reviewed in [85]). pRb inhibits cell cycle progression largely through binding and inhibition of the E2F transcription factors whose activities are required for cell cycle progression [86,87]). Transcriptionally active E2F induces the expression of numerous genes important for cell cycle progression, such as CDC6 [88], DHFR (dihydrofolate reductase) [89], TK (thymidine kinase) [90] and DNA polymerase α [91]. The synthesis of cyclin E, which occurs during late G1, is also regulated by the E2F family of transcription factors [9295]. Initial phosphorylation of pRb by cyclin D–CDK4/6 complexes leads to partial activation of E2F, which allows the transcription of cyclin E gene by E2F family members. Cyclin E then interacts with CDK2, forming an active complex that further phosphorylates pRb, resulting in full activation of E2F. This provides a positive feedback loop mechanism, whereby phosphorylation of pRb by cyclin E–CDK2 induces further transcription of the cyclin E gene (reviewed in [96]).

pRb contains two functional domains, consisting of a so-called pocket region and a C-terminal region [97,98]. The pocket of pRb is approx. 45 kDa, comprising two domains, A and B, that are separated by a spacer region. pRb binding inhibits E2F transcriptional activity by two mechanisms. One mechanism is through direct inhibition of the E2F transactivation domain, where the pRb A–B interface provides docking for E2F [97]. The C-terminal pRb domain also binds E2F to repress activity [99]. In addition to inhibition of transactivation activity, pRb inhibits E2Fs through active repression by recruiting members of the HDAC (histone deacetylase) family to E2F promoters [100,101]. HDAC-mediated deacetylation of the core histones leads to a tighter association between the core histones and DNA on chromatin, thus impairing access of transcriptional co-activators and leading to transcriptional inhibition [102]. pRb contains 16 consensus CDK phosphorylation sites, and several studies have demonstrated that cumulative phosphorylation on multiple sites is required to achieve full inactivation of pRb-mediated E2F inhibitory function to allow G1–S-phase cell cycle progression [72,103,104]. Both cyclin D–CDK4/6 and cyclin E–CDK2 inactivate pRb through sequential phosphorylation of different sites, leading to progressive loss of pRb-mediated E2F inhibitory function. Therefore, initial phosphorylation by cyclin D–CDK4/6 on the pRb C-terminal serine residues at positions 788 and 795 destabilizes its interaction with E2F [98] and induces dissociation of HDACs [72]. Subsequent phosphorylation during late G1 by cyclin E-CDK2 on threonine residues 821 and 826 and residues in the pocket region results in an intramolecular interaction between the pRb C-terminal and pocket regions, leading to dissociation from E2F to allow transcription [72,98] (Figure 2).

Figure 2 Cumulative phosphorylation of the pRb tumour suppressor by cyclin D–CDK4/6 and cyclin E–CDK2 alleviates its inhibitory activity towards the E2F transcription factor

The inhibitory activity of pRb on E2F involves two mechanisms. pRb binds directly to the transactivation domain of E2F and, in addition, recruits HDAC. Phosphorylation (P) by cyclin D–CDK4/6 in early-mid G1 phase destabilizes the C-terminal (C-term) portion of pRb, resulting in the dissociation of HDAC. Further phosphorylation by cyclin E–CDK2 results in conformational changes to pRb, liberating E2F which activates transcription of genes necessary for progression of cells into S-phase.

NPAT (nuclear protein mapped to the AT locus)

In addition to regulating the transcription of genes important for G1–S-phase cell cycle progression through phosphorylation of pRb, CDK-mediated phosphorylation of the NPAT transcriptional regulator is important for increased expression of histones, required for assembling with newly synthesized DNA into chromatin prior to mitosis. NPAT was initially identified as a cyclin E–CDK2 substrate using an in vitro phosphorylation screen of λ-phage cDNA expression libraries [51]. Overexpression of active cyclin E–CDK2 increases NPAT phosphorylation in U2OS osteosarcoma cells, and co-expression of NPAT with cyclin E–CDK2 co-operates in promoting S-phase cell cycle entry. This effect is not observed with cyclin A–CDK2 co-expression, indicating that NPAT is specifically targeted by cyclin E–CDK2 to promote S-phase entry [51,105]. NPAT is associated with histone gene clusters and activates histone gene transcription [105,106]. Mutation of the NPAT cyclin E–CDK2 phosphorylation sites at Ser-775, -779 and -1100 and Thr-1270 and -1350 suppresses its transcription from the histone H2B promoter, underlying the critical role of cyclin E–CDK2 phosphorylation in regulating NPAT-dependent histone gene transcription [105,106].

Centrosome duplication

Chromosomal segregation during mitosis is dependent on the formation of bipolar mitotic spindle [107,108]. Microtubules which are the major component of the mitotic spindle attach to kinetochores of sister chromatids at the metaphase plate and to centrosomes at the spindle poles. During mitosis, sister chromatids are separated and retracted by microtubules from the metaphase plate towards the centrosomes at the spindle poles to segregate the genetic material prior to cytokinesis to generate two identical daughter cells. Centrosomes thus play a pivotal role in separating sister chomatids during mitosis. Following cytokinesis, each daughter cell inherits one centrosome, and therefore it is necessary to duplicate the centrosome prior to the next round of mitosis to establish a bipolar mitotic spindle. The centrosome consists of a pair of core centrioles, surrounded by an amorphous mass of proteins termed PCMs (pericentriolar materials; providing support to the core centrioles [109]). Cyclin E–CDK2 promotes centrosome duplication [110,111] by phosphorylation of the centrosomal proteins NPM (nucleophosmin)/B23 [108,112] and CP110 (centrosomael protein of 110 kDa) [107]. Cyclin E–CDK2-mediated phosphorylation of NPM/B23 results in its dissociation from the centrosome to trigger centriole duplication. Microinjection of an anti-NPM/B23 antibody inhibits cyclin E–CDK2-mediated NPM/B23 phosphorylation and centriole duplication. The importance of cyclin E–CDK2-mediated NPM phosphorylation in centrosome duplication is exemplified by studies demonstrating that expression of a NPM/B23 T199A phosphosite mutant remains associated with centrosomes and inhibits their duplication [112]. Duplicated centrosomes remain free of NPM/B23 until mitosis. When the nuclear membrane breaks down during mitosis, NPM/B23 relocalizes to centrosomes, and upon cytokinesis each daughter cell receives one centrosome bound by NPM/B23. Therefore NPM/B23 is thought to constitute a ‘licencing system’ for centrosome duplication. By phosphorylation of NPM/B23 and other substrates important for G1–S-phase cell cycle progression, cyclin E–CDK2 co-ordinates centrosome and DNA duplication in preparation for mitosis. Another cyclin E–CDK2 centrosomal substrate is CP110. CP110 is expressed during G1 phase, peaks at S-phase and its levels diminish as cells enter mitosis [107]. Depletion of CP110 suppresses centrosome reduplication, whereas expression of a CP110 CDK phosphosite alanine mutant to mimic a constitutively non-phosphorylated form leads to impaired centrosome separation and polyploidy [107].

(ii) CDK substrates during S-phase

During S-phase, CDKs play important roles in both promoting DNA replication and ensuring that only one round of replication occurs per cell cycle. As cells enter S-phase, chromosomal DNA replication is initiated at numerous origins concurrently and a conserved mechanism termed the licencing system ensures that DNA replication occurs only once per cell cycle. The DNA replication origins consist of initiator proteins that interact with replicator elements within the DNA [113]. The initiator protein complex, which is collectively referred to as the ORC (origin recognition complex), consists of six polypeptides that serve as a foundation for protein–protein interactions important for DNA replication, including the licencing protein complex, consisting of the MCM (mini-chromosome maintenance) proteins 2–7 [114] that are evolutionarily conserved. Their C-terminus display sequence homology to hexameric DNA helicases and function to unwind the DNA for replication. All six MCM proteins are required for DNA replication, with MCM 4, 6 and 7 involved in ATP binding and catalysis, whereas MCM 2, 3 and 5 possess regulatory functions [115]. Studies in yeast show that during DNA replication initiation, CDC6 and CDT1 (chromatin licencing and DNA replication factor 1) proteins mediate the interaction between the MCM proteins and the ORC, thereby creating the pre-RC (pre-initiation replicative complex) [116,117]. In the absence of CDC6 cells enter a pseudo-mitotic state in the absence of replicated DNA [118]. DNA polymerase α, δ and ε are also recruited to the origins by CDC45, which initiates DNA synthesis (reviewed in [119]). The ORC is regulated by CDKs to prevent re-licencing of replicated DNA prior to mitosis to prevent polyploidy. The phosphorylation status of the ORC varies during the cell cycle with the ORC being hypophosphorylated during early G1, increased phosphorylation during the G1–S transition and being hyperphosphorylated when cells are in S-phase through to mitosis [120123]. Several proteins are phosphorylated by CDKs to regulate the ORC. Hence, cyclin A–CDK1 phosphorylates the ORC subunit Orc1 during mitosis, preventing its interaction with chromatin [123]. Another major target of CDKs in preventing DNA re-replication in budding yeast is CDC6 (the S. pombe homologue is cdc18). CDK-mediated phosphorylation of CDC6 results in its ubiquitin-mediated degradation at the G1–S-phase transition [124]. In higher eukaryotes, CDK-mediated phosphorylation controls nuclear transport of excess CDC6 during S-phase [125] (Figure 3). Another licencing protein which is regulated by CDK complexes is CDT1. In higher eukaryotes, CDK2 and/or CDK4 complexes and cyclin A–CDK1-mediated phosphorylation of CDT1 results in its interaction with the F-box protein Skp2 (S-phase kinase-associated protein 2) of the SCF (Skp1/cullin/F-box) ubiquitin ligase complex, which mediates ubiquitin-mediated proteolysis of CDT1 [126,127] (Figure 3). Furthermore, in vitro phosphorylation of CDT1 reduces its capacity to bind to DNA, whereas inactivation of CDK1 in murine FT210 cells resulted in de-phosphorylation and re-binding of CDT1 to chromatin [127]. In contrast, CDK-mediated phosphorylation of S. cerevisiae CDT1 resulted in its nuclear export [128]. The MCM proteins are also targeted by the S. cerevisiae CDK complex during S, G2 and early mitosis, which lead to their export from the nucleus to the cytoplasm, ensuring that re-replication of the chromosomal DNA does not occur in a single round of cell division [129].

Figure 3 CDK regulation of CDC6 and CDT1 prevents re-replication during DNA synthesis

During S-phase, DNA replication initiation occurs at ORCs at multiple sites, which serve as a foundation for protein—protein interactions. The interaction between the ORC and the licencing complex is mediated by CDC6 and CDT1. CDK-mediated phosphorylation (P) of the ORC, as well as the licencing complex, including CDC6 and CDT1, prevents re-licencing and DNA re-replication prior to mitosis. CDK-mediated phosphorylation of CDT1 leads to its ubiquitination (Ub) and proteosomal degradation, whereas phosphorylation of CDC6 results in its transport out of the nucleus.

Apart from governing DNA replication licencing, CDKs are also crucial for regulating DNA synthesis. In S. cerevisiae, CDK-mediated phosphorylation of Sld2 is necessary for initiating DNA replication. Phosphorylated Sld2 interacts with DNA replication protein Dpb11 to promote DNA replication [130]. A phosphosite mutant of Sld2 where all six CDK consensus phosphorylation sites are mutated to alanine interferes with Sld2–Dpb11 complex formation and prevents DNA replication [130]. The ability of Sld2 to interact with Dpb11 is due to the presence of the BRCT [BRCA1 (breast cancer early-onset 1) C-terminus] domains located at the C-terminus of Dpb11 [131]. In addition, Dpb11 also possesses N-terminal BRCT domains which interact with another CDK substrate, Sld3 [132]. The primary amino acid sequence of Sld3 contains 12 potential CDK phosphorylation sites, two of which (Thr-600 and Ser-622) are essential for cell survival [132]. Furthermore, both Thr-600 and Ser-622 have been demonstrated to be phosphorylated during the G1–S-phase transition and during S-phase at the same time as ORC6 [132]. The ability of Sld2 and Sld3 and the effect of their phosphorylation by CDKs in regulating DNA replication remain unclear. However, it is thought that they may target downstream replication proteins such as CDC45 and the GINS (from the Japanese go-ichi-ni-san meaning 5-1-2-3, after the four related subunits of the complex Sld5, Psf1, Psf2 and Psf3) complex. CDC45 interacts with Sld3 and together with GINS forms a complex with the MCM proteins at the replication fork (reviewed in [133]). In addition, CDC45 is also involved in the recruitment of DNA polymerases (reviewed in [119]).

(iii) CDK substrates during M-phase

APC/C (anaphase-promoting complex/cyclosome)

Progression through mitosis is primarily governed by cyclin B–CDK1 and inactivation of this kinase during late mitosis is necessary for cell cycle exit. A major target of cyclin B–CDK1 in regulating mitotic progression is the APC/C ubiquitin ligase. Cyclin B–CDK1 and APC/C interplay in feedback regulation to control mitotic progression and exit. The APC/C, which is an evolutionarily conserved ubiquitin ligase of the RING family consisting of at least 12 core subunits, is critical for mitosis [134].

Activation of the APC/C requires the binding of one of two activator proteins, CDC20 or CDH1 (CDC20 homologue 1). Both proteins are active during different phases of mitosis [135]. CDC20 binding activates APC/C in early mitosis, whereas CDH1 activates APC/C in late mitosis through to G1 phase [135]. During early mitosis, cyclin B–CDK1-mediated phosphorylation on the APC/C core subunits CDC27, CDC16, CDC23 and APC7 plays a major part in activation of the APC/C and binding of CDC20 to APC/C [136]. In addition to phosphorylation of the APC/C core subunits facilitating CDC20 binding, concurrent cyclin B–CDK1-mediated phosphorylation of CDH1 prevents its interaction with the APC/C [137], thus maintaining APC/CCDC20 as the major ubiquitin ligase complex during mitosis. Once activated, APC/CCDC20 initiates the ubiquitination and proteasomal degradation of the securin proteins and mitotic B-type cyclins during the metaphase–anaphase transition. Securin is an anaphase inhibitor protein that blocks sister chromatid separation [138140]. Following ubiquitin-mediated degradation of securin, the separase enzyme is activated, which in turn cleaves the SCC1 (sister chromatid cohesion protein 1) subunit of the cohesion complex, thus allowing the separation of the two sister chromatids [141144]. In addition, APC/CCDC20 also ubiquitinates cyclin B, leading to its degradation and inactivation of cyclin B–CDK1 during late anaphase [145147]. This in turn relieves the phosphorylation of CDH1, thus allowing CDH1 interaction with APC/C. As cells exit mitosis, active APC/CCDH1 targets substrates such as the CDC20 protein [148,149]. Upon re-entering the G1 phase, the APC/CCDH1 remains active to prevent the early activation of mitotic APC/CCDC20 complex, which can then be re-activated following increased cyclin B synthesis prior to the next round of mitosis. The interplay of cyclin B–CDK1 and the APC/C in controlling mitotis represents an important illustration of how CDK-mediated phosphorylation and ubiquitin-mediated proteolysis control cell cycle progression (Figure 4).

Figure 4 Interplay of cyclin B–CDK1 and the APC/C in controlling mitosis

Early mitosis: Cyclin B–CDK1 becomes active following increased synthesis of cyclin B in late G2 phase. During early mitosis, cyclin B–CDK1-mediated phosphorylation (P) of core APC/C subunits stimulates the binding of CDC20. Concurrently, cyclin B–CDK1 phosphorylates CDH1, preventing its interaction with the APC/C. CDC20 or CDH1 binding to the APC/C during mitosis differentially regulates the substrate preference of this ubiquitin ligase. The APC/CCDC20 complex ubiquitinates (Ub) several substrates, including the anaphase inhibitor protein securin, which is an inhibitor of the separase protease. Ubiquitination and proteasomal degradation of securin leads to activation of the separase enzyme which proteolyses SCC1, allowing the separation of sister chromatids. Late mitosis: through feedback regulation, the APC/CCDC20 complex also ubiquitinates cyclin B as cells progress through mitosis, reducing the activity of cyclin B–CDK1. Reduced cyclin B–CDK1 activity allows CDH1 interaction with the APC/C during late mitosis. The APCCDH1 targets numerous substrates, including CDC20 and cyclin B, to abolish CDK1 activity and mediate mitotic exit.

Emi1 (early mitotic inhibitor 1)

In addition to CDC20 and CDH1, an inhibitor to the APC/C, Emi1, is a CDK substrate. Emi1 contains an F-box domain, a ZBR (zinc-binding domain), five potential CDK phosphorylation sites and two potential NLSs (nuclear localization sequences) [150]. Emi1 interacts with CDC20 and inhibits the APC/C to induce mitotic arrest and stabilization of cyclin B [150]. Emi1 levels oscillate through the cell cycle where its levels are increased during S-phase and depleted in M-phase [150]. Emi1 destruction at early mitosis is mediated by the SCF ubiquitin ligase complex [151,152]. Synergistic phosphorylation by Plk1 (polo-like kinase 1) and cyclin B–CDK1 is required for SCF-mediated ubiquitination of Emi1 in vitro [153].

CONCLUDING REMARKS

Genetic and biochemical studies over the last three decades have provided significant insights into the molecular mechanisms of cell cycle progression and revealed that CDKs are pivotal regulators of this fundamental process. These kinases phosphorylate numerous substrates to orchestrate the dynamic changes in cell behaviour during cell division, such as DNA replication, mitosis and cytokinesis. How CDKs control these processes at a mechanistic level has been well described for a few substrates reported to date. The characterization of the full repertoire of CDK substrates and the understanding of how their phosphorylation controls cell division awaits further study. Given the central role of de-regulated CDK activity in many human cancers, these studies will provide a better understanding of mechanisms of oncogenesis.

FUNDING

Work in the laboratory on CDK substrates is supported by The Cancer Council Victoria [grant number 502619]; The Australian Research Council [grant number DP0771366]; The National Breast Cancer Foundation [grant number KCF7]; and The Breast Cancer Research Program from The Department of Defence Congressionally Directed Medical Research Programs [grant number BC020800].

Acknowledgments

We thank Bruce E. Kemp for critical reading of the manuscript prior to submission.

Abbreviations: APC/C, anaphase-promoting complex/cyclosome; BRCT, BRCA1 (breast cancer early-onset 1) C-terminus; CDC, cell division cycle; CDH1, CDC20 homologue 1; CDK, cyclin-dependent kinase; CDT1, chromatin licencing and DNA replication factor 1; CP110, centrosomal protein of 110 kDa; cy motif, cyclin-binding motif; E16.5 etc., embryonic day 16.5, etc.; Emi1, early mitotic inhibitor 1; G1, Gap1 phase; G2, Gap2 phase; GINS, from the Japanese go-ichi-ni-san meaning 5-1-2-3, after the four related subunits of the complex Sld5, Psf1, Psf2 and Psf3; HDAC, histone deacetylase; MCM, minichromosome maintenance; M-phase, mitosis; NPAT, nuclear protein mapped to the AT locus; NPM, nucleophosmin; ORC, origin of recognition complex; pRb, retinoblastoma protein; SCC1, sister chromatid cohesion protein 1; SCF, Skp1/cullin/F-box; Skp, S-phase kinase-associated protein; S-phase, DNA synthesis phase

References

View Abstract