Enzymology of the G₂/Mitosis Transition

The transition between the G₂ phase and mitosis is the most profound morphologic and physiological change that occurs during the life of a growing cell, matched only by the dramatic changes that occur during death by apoptosis (see Chapter 46). Entry into mitosis is controlled by a network of stimulatory and inhibitory protein kinases and phosphatases, presided over by Cdk1–cyclin B1. (Chapter 40 introduced the components involved in the G₂/M transition.)

Cdk1, the driving force for entry into mitosis, is present at a constant level throughout the cell cycle. Cdk1 regulation is multifaceted (see Fig. 40-14), including binding of cyclin cofactors, inhibition and activation by phosphorylation, binding of inhibitory molecules, and changes in subcellular localization.

Mammalian cells have at least three B-type cyclins: B1, B2, and B3. Cyclin B1 is essential for triggering the G₂/M transition, and disruption of its gene has lethal consequences. Cyclin B1, newly synthesized during the latter part of the cell cycle, binds Cdk1 and shuttles it in and out of the nucleus. Importin b carries the Cdk1–cyclin B1 complex into the nucleus, and then Crm1 rapidly exports it back to the cytoplasm (see Chapter 16). Cdk1–cyclin B2 associates with the Golgi apparatus during interphase and might function in Golgi disassembly during mitosis (see Fig. 44-4). Cyclin B3 appears to function only during meiosis in mammals.

As cells approach the G₂/M transition, a combination of one stimulatory and two inhibitory kinases control activation of Cdk1–cyclin B1. The activating kinase, called CAK (Cdk-activating kinase; actually Cdk7–cyclin H), is located in the nucleus. CAK phosphorylates Cdk1 on T¹⁶¹, allowing the refolding of the active site cleft that is required for the enzyme to bind substrates (see Chapter 40). Two other kinases, Wee1 and Myt1, counteract the action of CAK (Fig. 43-1). Wee1 is localized to the nucleus, where it inhibits Cdk1 by phosphoryla-ting Y¹⁵ adjacent to the ATP-binding site. The role of Wee1 as a mitotic inhibitor is clearly demonstrated in S. pombe: Overexpression of functional Wee1 delays or prevents the entry of cells into mitosis (Fig. 43-2). Myt1 associated with the Golgi apparatus and endoplasmic reticulum inhibits Cdk1 while it resides in the cytoplasm by phosphorylating T¹⁴ and Y¹⁵. Together, Wee1 and Myt1 ensure that Cdk1 remains inactive as it shuttles into and out of the nucleus (Fig. 43-3). This combination of stimulatory and inhibitory modifica-tions holds Cdk1–cyclin B1 poised for a burst of activation.

Figure 43-1 Regulation of Cdk1 by cyclin binding and protein phosphorylation from the late S phase through mid-mitosis.

(Structure for Wee1 kinase provided by E. N. Baker prior to re-lease to the Protein Data Bank; based on Squire CJ, Dickson JM, Ivanovic I, Baker EN: Structure of human Wee1A kinase: Kinase domain complexed with inhibitor PD0407824. Structure 13:541–550, 2005. PDB file: 1X8B.)

Figure 43-2 The effect of changing cellular levels of Wee1 protein on cell-cycle progression in fission yeast. A, Cells that lack functional Wee1 protein enter mitosis too soon in the cell cycle and are smaller than wild-type cells (B). C–D, Cells that express excess Wee1 protein are too effective at inactivating Cdk1 and are severely delayed in their ability to enter mitosis (hence their larger size).

(From Russell P, Nurse P: Negative regulation of mitosis by Wee1⁺, a gene encoding a protein kinase homolog. Cell 49:559–567, 1987.)

Figure 43-3 Summary of the patterns of Myt1 and Wee1 phosphorylation of Cdk1 in the cytoplasm and nucleus in interphase and mitosis. NES, nuclear export sequence.

(Based on an original figure by Helen Piwnica-Worms. 14-3-3 protein, PDB file: 1YWT.)

The three Cdc25 protein phosphatases remove the inhibitory phosphates from Cdk1–cyclin B1 and trigger the G₂/M transition. Cdc25s are dual-specificity protein phosphatases (see Fig. 25-5) that remove phosphates from serine (S), threonine (T), and tyrosine (Y) residues, including the inhibitory phosphates from T¹⁴ and Y¹⁵ of Cdk1. Cdc25s are regulated by stimulatory and inhibitory phosphorylation, by alterations in their subcellular localization, and by ubiquitin-mediated proteolysis. Cdc25A, the only one of the three to be indispensable for life, functions at both the G₁/S and G₂/M transitions, whereas Cdc25B and Cdc25C have roles in the G₂/M transition. The preferred targets of the individual Cdc25 isoforms are not known.

Cdc25A and Cdc25C are relatively inactive during interphase for three reasons (Fig. 43-4). First, phosphorylation on a serine residue creates a binding site for a member of the 14-3-3 group of adapter proteins (see Fig. 25-10). These proteins bind sites on target proteins containing serines flanked by several other characteristic amino acids, but only when the critical serine is phosphorylated. This is an example of the general mechanism whereby phosphorylation regu-lates interactions between proteins in response to physiological signals (see Chapter 25). Association of 14-3-3 with Cdc25A interferes with its binding to Cdk1–cyclin B1 and inhibits its import into the nucleus. Because Cdc25C has an intrinsic nuclear export sequence, the 14-3-3-bound form of Cdc25C is primarily cytoplasmic.

Figure 43-4 Regulation of Cdc25A and Cdc25C activity in interphase and mitosis.

(Based on an original figure by Helen Piwnica-Worms.)

Second, phosphorylation at other sites targets Cdc25A for ubiquitination by SCF^bTrCP and destruction by proteasomes. Third, full activation of Cdc25s requires phosphorylation of their amino-terminal region. In meiotic cells, phosphorylation of this site is initiated by a protein kinase called Polo (see next paragraph) and then completed by the Cdc25 substrate, Cdk1–cyclin B1, creating a powerful positive feedback amplification loop that provides a burst of Cdk activity and triggers entry into mitosis. Whether this same mechanism operates in mitotic cells is not known.

Since Cdk1–cyclin B1 is inactive unless activated by a Cdc25 phosphatase, a molecular trigger is required to start the amplification cycle. Members of the Polo family of protein kinases are candidates for this role, by activating Cdc25. These kinases and their substrates share amino acid motifs called “polo boxes,” which bind to other polo boxes after they have been phosphorylated. Polo kinases recognize certain substrates only after they have been “primed” by phosphorylation by another kinase. This allows for additional levels of control and rapid amplification of the phosphorylation response. Polo family kinases are involved in a variety of mitotic events, including formation of a bipolar spindle, cytokinesis, and passage through certain cell-cycle checkpoints. They reside at the centrosome during interphase and phosphorylate a number of centrosomal proteins in addition to Cdc25.

Changes in Subcellular Localization at the G₂/M Transition

Cyclin B1 (with associated Cdk1) and Cdc25C have both nuclear import and nuclear export signals, so they shuttle in and out of the nucleus throughout interphase (Fig. 43-5). Cdc25A resides in the nucleus. During interphase, Cdk1–cyclin B1 and Cdc25C spend most of their time in the cytoplasm. Inhibitory phosphorylation of Cdk1 and Cdc25C keeps Cdk1 activity low.

Figure 43-5 Shuttling of components between the nucleus and cytoplasm contributes to the regulation of Cdk1–cyclin B1 during interphase.

(Based on an original figure by Helen Piwnica-Worms.)

Early in prophase, phosphorylation of cyclin B1 inactivates its nuclear export signal and promotes its nuclear import (see Chapter 16), allowing Cdk1–cyclin B1 to accumulate rapidly in the nucleus within five minutes (Figs. 43-3 and 43-6). Cdc25C also stops shuttling at the G₂/M transition, probably as a result of phosphorylation by Polo kinase. The best evidence seems to indicate that the Cdk1–cyclin B that accumulates in the nucleus is already active, but restriction of Cdk–cyclin B complexes to the nucleus together with Cdc25A and Cdc25C significantly increases their local concentration (the volume of the nucleus is much less than the volume of the cytoplasm) and might contribute to the final burst of Cdk1–cyclin B1 activation.

Figure 43-6 Rapid movement of cyclin B1 from the cytoplasm into the nucleus at the onset of prophase and subsequent association with the spindle during mitosis.

(Courtesy of Christina Karlsson and Jonathon Pines, Wellcome/CRC Institute, Cambridge, England.)

Cdk1–Cyclin A and the Initiation of Prophase

As is noted in Chapter 42, Cdk2–cyclin A plays a critical role during the S phase. Several lines of evidence have revealed that a Cdk–cyclin A complex also helps to trigger the G₂/M transition. First, inactivation of cyclin A, either by mutation in Drosophila or by injection of anti–cyclin A antibodies into cultured cells, arrests the cell cycle in G₂. Second, Cdk–cyclin A activity peaks at G₂/M, before the peak of Cdk1–cyclin B1 activity. Finally, if activated Cdk2–cyclin A complexes are injected into cells just after completion of the S phase, cells enter mitosis prematurely.

Cdk–cyclin A kinase is most likely to regulate several events at the transition from the G₂ phase to prophase, including changes in microtubule behavior and chromosome condensation. Late in G₂, the half-life of microtubules drops dramatically from about 10 minutes to about 30 seconds (see Table 44-1). This, coupled with centrosomes’ enhanced ability to initiate microtubule polymerization, completely transforms the organization of the microtubule cytoskeleton. Centrosomes take on the appearance of spindle poles and migrate apart over the surface of the nucleus. At the same time, chromatin begins to condense in the nucleus. The mechanism of chromosome condensation is not known, but a protein complex called condensin is required for this condensation to begin during prophase (see Fig. 13-19). Condensin does not associate tightly with chromosomes during interphase; association with chromosomes requires phosphorylation of two of its subunits by a Cdk. This occurs at G₂/M. As chromosomes condense, several mitosis-specific kinetochore proteins also move into the nucleus and associate with kinetochores (see Chapter 13).

These events occur while most Cdk1–cyclin B1 is in the cytoplasm. It is most likely, therefore, that Cdk1–cyclin A triggers at least the nuclear events of prophase (Fig. 43-7). In fact, microinjection of a specific inhibitor of Cdk1–cyclin A causes prophase cells to return rapidly to interphase; chromosomes decondense, rounded prophase cells flatten, and the interphase microtubule network returns. Commitment to mitosis appears to be irreversible only after Cdk1–cyclin B1 enters the nucleus.

Figure 43-7 Locations and patterns of activation of Cdk1 complexed to cyclin A versus cyclin B across the cell cycle.

Summary of the Main Events of the G₂/M Transition

Synthesis of cyclin B1 in the latter portion of the S and G₂ phases leads to assembly of Cdk1–cyclin B heterodimers that shuttle into and out of the nucleus, spending most of their time in the cytoplasm associated with microtubules. In the late S and G₂ phases, activation of Cdk1–cyclin A initiates mitotic prophase, beginning with changes in microtubule dynamics and chromosome condensation. Several events trigger entry into the active phase of mitosis. Cdc25A becomes stabilized and no longer binds 14-3-3 proteins. This results in the accumulation of Cdc25A and allows more efficient interactions between Cdc25A and Cdk1–cyclin B. The phosphate-binding site for 14-3-3 proteins on Cdc25C becomes dephosphorylated, allowing it to accumulate in the nucleus. In addition, phosphorylation of cyclin B1 blocks its export from the nucleus and promotes its import, thus causing Cdk1–cyclin B1 to accumulate rapidly in the nucleus. The inhibitory kinase Wee1 is also dephosphorylated, and its activity drops. Cdc25A and Cdc25C activate Cdk1–cyclin B1 by removing inhibitory phosphates on T¹⁴ and Y¹⁵. This starts in the cytoplasm and then may be stimulated as the proteins concentrate in the nucleus. There, the action of Cdk1–cyclin B1 on the nuclear lamina triggers nuclear envelope breakdown and drives the cell into mitosis. Fig. 43-8 summarizes the network regulating Cdk1–cyclin B activation.

Figure 43-8 Summary of the Cdk1 feedback regulation mechanism at the G₂/M transition.

Why did such an elaborate system evolve to regulate the G₂/M transition? The answer appears to lie in the exquisite sensitivity provided by the interlocking network of stimulatory and inhibitory activities. On the one hand, this network ensures a rapid, almost explosive, final transition into mitosis. On the other, it provides a number of ways to delay the G₂/M transi-tion if the cell detects damage to chromosomes. Mitosis with chromosomal damage can lead to cell death or cancer.

The G₂ Checkpoint

Separation of sister chromatids during mitosis is a potential danger point for a cell. If DNA is damaged after it is replicated, the cell can use information present in sister chromatids (which have one good copy and one bad copy) to guide the repair process. How-ever, once sisters separate, such a corrective mechanism is impossible. In addition, if a cell enters mitosis before completing replication of its chromosomes, the attempt to separate sister chromatids causes extensive chromosomal damage. To minimize these hazards, a checkpoint operates in the G₂ phase to block mitotic entry if DNA is damaged or DNA replication is in-complete.

Exposure of cells to agents that damage DNA, including certain chemicals or ionizing radiation, halts the cell cycle temporarily in the G₂ phase. This G₂ delay gives cells an opportunity to repair damaged DNA before entering mitosis. Studies of radiation induced G₂ delay in budding yeast identified a major cell-cycle checkpoint in G₂ sensitive to the status of the cellular DNA. Cells that are defective in this checkpoint are much more sensitive to radiation injury than are wild-type cells because they continue to divide, despite the presence of broken or otherwise damaged chromosomes (Fig. 43-9). This continued division in the face of DNA damage leads to cell death, presumably due to accumulated chromosomal defects, as well as chromosome loss.

Figure 43-9 mutation of genes required for the g₂ checkpoint. Cells that are defective in the G₂ checkpoint (Rad9 mutants of budding yeast) cannot delay their entry into mitosis in the presence of damaged DNA and therefore divide themselves to death. Rad-9 is a component of the PCNA-like 9-1-1 damage sensor complex.

(Courtesy of Ted Weinert, University of Arizona, Tucson.)

The G₂ checkpoint also monitors the completion of DNA replication. Remarkably, this aspect of the checkpoint even works in vitro in cell-free extracts. As is described in Box 40-3, highly concentrated extracts made from Xenopus eggs can be induced to undergo a cyclic alteration in cell-cycle phases, even in the absence of added nuclei. Passage through the different phases can be followed by monitoring the activity of Cdk1–cyclin B kinase, which is high in mitosis and low in interphase. In the absence of nuclei, an inhibitor of DNA polymerases (such as the fungal toxin aphidicolin) has no effect on such extracts, which continue to cycle unabated between the S and M phases. On the other hand, if the extract contains more than 400 nuclei/mL undergoing synchronous cycles of DNA replication and mitosis, aphidicolin brings the cycling to a halt before M phase. This experiment appears to reconstitute the G₂ checkpoint in vitro (Fig. 43-10). Furthermore, it reveals that partly replicated nuclei release an inhibitory signal that stops the cycle.

Figure 43-10 reconstitution of a cell-cycle checkpoint in vitro. A, Cell-cycle transitions in a cell-free extract. B,Panel 1: DNA replication was measured by adding ³²P-labeled dCTP to the extract and subsequently isolating the DNA, running it in a gel, and detecting incorporation of the radioactive label by autoradiography.Panel 2: Cdk1–cyclin B activity was assayed on the basis of the ability of active enzyme to phosphorylate added histone H1 (which was similarly subjected to gel electrophoresis and autoradiography). These panels show that DNA replication and Cdk activity alternate in these in vitro cell cycles. Cdk activity declines precipitously at the end of mitosis. Panel 3: If the concentration of nuclei is increased beyond a threshold level and an inhibitor of DNA replication is added, the G₂ checkpoint detects the partly replicated DNA and stops the extract from entering mitosis. Cdk activity now remains at about 10% of the level needed to trigger mitosis.

(B, Redrawn from Dasso M, Newport JW: Completion of DNA replication is monitored by a feedback system that controls the initiation of mitosis in vitro: Studies in Xenopus. Cell 61:811–823, 1990.)

The G₂ Checkpoint and Cancer

Defects in the G₂ checkpoint are associated with cancer, since failure to correct damage to tumor suppressor genes can compromise the G₁ checkpoint in subsequent cell cycles. For example, suppose that radiation damages a nucleotide base in the gene for the retinoblastoma susceptibility protein (Rb; see Fig. 41-10). If this happens in the S or G₂ phase, informa-tion in the undamaged copy of the gene can be used to repair the damage. Delaying the cell cycle increases the chance of successful repair. On the other hand, if the checkpoint fails and the cell enters mitosis, sister chromatids separate from one another, and the damage is never repaired. As a consequence, one daughter cell from that division has a defective pRb gene, impairing its ability to regulate its growth at the restriction point in the next G₁ phase. Loss of tumor suppressor genes such as pRb and p53 is a major cause of cancer (see Chapter 41).

The G₂ checkpoint minimizes this problem by detecting DNA damage and either delaying entry into mitosis until the damage is fixed or triggering cell suicide by apoptosis. The checkpoint works by modulating the activities of the components that control the G₂/M transition.

How the G₂ Checkpoint Works

The minimal machinery of a DNA damage checkpoint (see Fig. 40-4) involves sensors that detect DNA damage, transducers (usually protein kinases) that produce a biochemical signal as a result of the detected damage, and effectors (both protein kinases and transcriptional activators) that either directly or indirectly block cell-cycle progression. The sensors are not yet as well characterized as are the transducers and effectors. This section discusses this machinery as a prelude to consideration in Box 43-1 of the mechanisms that repair damaged DNA.

Figure 43-13 EXAMPLES OF DNA DAMAGE AND THE REPAIR PATHWAYS THAT RESPOND TO DIFFERENT TYPES OF LESION.

Figure 43-14 pathways for the repair of base damage, bulky adducts, such as thymidine dimers formed by ultraviolet light, or mismatched bases. For detailed descriptions, see the text. The inset in the panel on base excision repair shows human 3-methyladenine DNA glycosylase complexed to DNA (PDB file: 1BNK). This enzyme scans the DNA for bases that are not strongly H-bonded, uses its “finger” to swing them up into the pocket for scanning, and—if bad—catalyzes excision of that base.

(Inset illustration by Graham Johnson (www.fivth.com) for the Howard Hughes Medical Institute, copyright 2004, all rights reserved. Reference: Lau AY, Scharer OD, Samson L, et al: Crystal structure of a human alkylbase-DNA repair enzyme complexed to DNA: Mechanisms for nucleotide flipping and base excision. Cell 95:249–258, 1998.)

Figure 43-15 pathways for the repair of dna double-strand breaks. A double-strand break is recognized by the PI3-kinase family members ATM or ATR and a cell-cycle delay ensues (see Fig. 40-4). The break is then bound by repair factors in the homologous recombination or the nonhomologous end-joining pathway of DNA repair. It is not known what factors control the “choice” between these pathways. A, Homologous recombination pathway of DNA repair. The MRN complex chews back the DNA at a break, leaving a single-stranded overhang that is stabilized by RPA (not shown). It is believed that the MRN complex also plays a role in keeping the broken ends in proximity to one another. Next, RAD51 forms a nucleoprotein filament on the single-stranded DNA, displacing the RPA. The RAD51 nucleoprotein filament then initiates homology searching and repairs the DNA break by inserting the extended single-stranded DNA into homologous sequences (usually on the sister chromatid [blue]) and allowing homologous recombination and DNA repair/resynthesis to occur. Capture of the second single-stranded DNA end allows the formation of a joint molecule with a double Holliday junction. Resolution of this Holliday junction structure results in accurate, templated repair of the double-strand break. B, A Holliday junction formed by four complementary oligonucleotides complexed to the enzyme Cre (not shown). The Holliday junction is a dynamic structure (arrows) that can migrate along the DNA. (PDB file: 3CRX.) C, Nonhomologous end-joining pathway of DNA repair. Nonhomologous end-joining is initiated by break recognition by the Ku70/Ku80 heterodimer, which recruits DNA-PK and tethers the broken ends. The breaks are then processed in a reaction involving the MRN complex and other repair factors. DNA-PK’s precise role is not yet entirely clear. Next, DNA ligase IV/XRCC4 is recruited to the processed double-strand break, which is ligated back together.

(B, Reference: Gopaul DN, Guo F, Van Duyne GD: Structure of the Holliday junction intermediate in Cre-loxP site-specific recombination. EMBO J 17:4175–4187, 1998.)

The sensors that detect DNA damage have not yet been entirely characterized (Fig. 43-11). The two apical kinases ATM and ATR (see Fig. 40-4) are mobilized quickly to the sites of DNA damage. However, ATM is activated before it gets to the DNA, and the identity of the earliest sensor of DNA damage remains unclear. The 9-1-1 protein complex (described later) and the enzymes that are involved in processing damaged DNA are also possible sensors of the DNA damage.

Figure 43-11 Hypothetical models for how ATM, ATR, and the 9-1-1 complex might act as DNA damage sensors. A, ATM as sensor. B, ATR as sensor. C, The 9-1-1 complex as sensor.

Two kinases, ATM and ATR (see Fig. 40-4), are key transducers of the DNA damage response. ATM responds to DNA damage any time during the cell cycle, but ATR appears to respond only if DNA is damaged prior to the completion of replication. Thus, ATM is essential for the G₂ DNA damage checkpoint to arrest the cell cycle when DNA damage occurs after replication is complete.

The activation of ATM involves autophosphorylation, which causes inactive dimers to dissociate into active monomers. Subsequently the Nbs1 subunit of the MRN complex (Fig. 43-11A; see also Fig. 12-13) recruits ATM monomers to the DNA break and initiates the full-blown checkpoint response. The MRN complex has a key role in repair of double-strand breaks in DNA. The Nbs1 gene is mutated in humans with Nijmegen breakage syndrome (Table 43-1).

Table 43-1 KEY DNA REPAIR GENE DEFECTS ASSOCIATED WITH HUMAN DISEASE

NER, nucleotide excision repair; MMR, mismatch repair; DSB, double-strand break; HR, homologous recombination; NHEJ, nonhomologous end joining.

* This list is an outline. For an updated list of the approximately 130 human DNA repair genes known to date, the reader is referred to http://www.cgal.icnet.uk/DNA_Repair_Genes.html#refs.

ATR exists in cells in a constitutive complex with a second subunit, ATR-interacting protein (ATRIP). ATRIP binds to the single-strand DNA binding protein RPA, an essential component of the DNA replication fork (Fig. 43-11B). DNA damage often results directly or indirectly in the presence of single-stranded DNA, which can be produced by the damaging event itself or as a by-product of the DNA repair process. This single-stranded DNA binds RPA and then, through the action of ATRIP, recruits and activates ATR.

Interestingly, the interactions of ATR with ATRIP and ATM with NBS1 both involve short peptide sequences at the very C-terminus of ATRIP and NBS1. These are the only features that NBS1 and ATRIP have in common. Somehow, this binding may alter the structures of the ATM and ATR in a way that turns on the checkpoint response. Since the kinases are already activated prior to recruitment to the sites of DNA damage, activation of the checkpoint response may require additional functions, such as acquiring selectivity for certain key substrates.

ATR depends on two other protein complexes to mount a checkpoint response. One is the trimeric 9-1-1 complex, which gets its name from its subunits Rad9, Hus1, and Rad1 (Fig. 43-11C). The 9-1-1 complex resembles proliferating cell nuclear antigen (PCNA), the doughnut-shaped processivity factor that is indispensable during DNA replication (see Fig. 42-11). PCNA is loaded onto DNA by the pentameric replication factor C (RFC) ATPase during DNA replication and anchors DNA polymerases and other factors to DNA. A similar enzyme composed of one special subunit, Rad17, plus the four small subunits of RFC is believed to load the 9-1-1 complex onto DNA at or near sites of damage. This speculation is supported by the observation that mutants in four of the RFC subunits are defective in G₂ checkpoint control in yeasts and Drosophila.

The role of the 9-1-1 complex is not certain. It could be involved in recognition of damage as it slides along the DNA; it could act as a sliding clamp tethering specialized polymerases that repair the damaged DNA; or it could act as a mobile landing pad for other factors involved in processing damaged DNA.

When DNA damage activates ATM and ATR, they phosphorylate several important substrates, including the tumor suppressor protein p53, and two protein checkpoint kinases called Chk1 and Chk2. When activated by phosphorylation, Chk1 phosphorylates the Cdc25A protein phosphatase (Fig. 43-12). This has at least two consequences. First, it produces binding sites for a 14-3-3 protein, which blocks Cdc25A from activating Cdk1–cyclin B. Second, phosphorylation by Chk1 targets Cdc25A for ubiquitin-mediated proteoly-sis. This ensures that levels of Cdc25A remain low. In addition, activation of the G₂ checkpoint maintains Cdc25C in a 14-3-3-bound form. This prevents Cdc25C from accumulating in the nucleus and from activat-ing Cdk1–cyclin B as it shuttles in and out of the nucleus.

Figure 43-12 how the g₂ checkpoint blocks the g₂ → m transition following activation of atm and/or atr by dna damage. Dotted lines show activities that are switched off by the checkpoint. The dashed line between ATM/ATR and the kinase that inhibits Cdc25C indicates that this pathway is not yet known.

(Based on an original figure by Helen Piwnica-Worms.)

The transcription factor p53 is phosphorylated and activated by the ATM/ATR kinases following DNA damage. Activated p53 then drives the expression of a number of G₂ checkpoint genes. Although p53 is not required to arrest the cell cycle in G₂ phase in response to DNA damage, it is required to prolong this cell-cycle arrest. p53 stimulates the expression of the Cdk inhibitor p21, which inhibits Cdk1–cyclin A 100-fold better than it inhibits Cdk1–cyclin B1. p21 expression thus provides an effective way of blocking the initiation of prophase by Cdk1–cyclin A. p21 also participates in the G₁ DNA damage checkpoint.

p53 also drives the expression of 14-3-3s, an adapter protein that may interfere with shuttling of Cdk1–cyclin B1 between the nucleus and cytoplasm. Binding of 14-3-3s maintains the Wee1 inhibitory kinase in amore active state, ensuring that the Cdk1–cyclin B1 complex remains inactive. Disruption of the gene for 14-3-3σ is fatal for cells if they sustain DNA damage. Instead of activating their G₂ checkpoint, they enter an aberrant state with characteristics of both mitosis and apoptosis, and then die.

Another substrate of ATM and ATR is the specialized histone isoform H2AX. H2AX phosphorylated by ATM and/or ATR is known as γ-H2AX. γ-H2AX spreads rapidly across sites of DNA damage, called “foci.” These g-H2AX foci recruit key proteins for signaling, DNA repair machinery, and the cohesin complex, which holds replicated sister chromatids together (see Fig. 13-19). In the case of double-strand DNA breaks, a particularly dangerous form of DNA damage, the unbroken sister chromatid provides a template to ensure that the break is repaired correctly.

ATM and ATR signaling often involves proteins known as adapters (also called mediators). The best-known ATR adapter is a protein called claspin, a normal component of replication forks, which is conserved between yeasts and humans. ATR phosphorylates claspin, which recruits Chk1 to stalled replication forks for phosphorylation by ATR. This amplifies the checkpoint response.

Phosphorylation of the ATM adapter, 53BP1, recruits Chk2 to sites of damage where it can be activated by ATM. BRCA1 is another adapter that functions in DNA repair. BRCA1 is of interest because its gene is mutated in 90% of cases of inherited familial breast and ovarian cancer. Both 53BP1 and BRCA1 have protein motifs called BRCT domains that recognize and bind to specific phosphorylated sequences on target proteins. BRCT motifs may recruit proteins to g-H2AX foci by binding to the phosphorylated histone.

Ideally, checkpoint activation has one of two outcomes. If DNA damage is so extensive that it cannot be repaired, the cell commits suicide by a pathway called apoptosis (see Chapter 46). Less serious damage can be repaired by one of the systems described in Box 43-1.

Transition to Mitosis

The complex web of stimulatory and inhibitory activities that regulates Cdk activity in the G₂ phase enables exquisite control of the G₂/M transition. On the one hand, it poises Cdk1–cyclin B in a state in which it is ready for the explosive burst of activation that triggers the G₂/M transition. At the same time, the complex pathways afford many points where the process may be regulated. These are the basis of the G₂ checkpoint control that prevents cells from segregating their chromosomes if genomic DNA cannot meet stringent quality control standards. Eventually, however, if all goes well, Cdk1–cyclin A and Cdk1–cyclin B1 are activated, and the cell embarks on mitosis, probably the most dramatic event of its life.

ACKNOWLEDGMENTS

Thanks go to Jonathon Pines, Helen Piwnica-Worms, and Carl Smythe for their suggestions on revisions to this chapter. We are especially grateful to Ciaran Morrison, who wrote the new box on DNA repair.