The Biology of Cancer - Chapter 8 (PDF)

Chapter 8 pRb and Control of the Cell Cycle Clock This immediately leads one to ask, if the [hypothetical cellular transfor- mation] loci can get into so much mischief, why keep them around? The logical answer is that they have some necessary function during some stage o...

Chapter 8 pRb and Control of the Cell Cycle Clock This immediately leads one to ask, if the [hypothetical cellular transfor- mation] loci can get into so much mischief, why keep them around? The logical answer is that they have some necessary function during some stage of the cell cycle, or some stage of embryogenesis. David E. Comings, geneticist, 1973 T he fate of individual cells throughout the body is dictated by the signals that each receives from its surroundings—a point made repeatedly in earlier chapters. Thus, almost all types of normal cells will not proliferate unless prompted to do so by mitogenic growth factors. Yet other signaling proteins, notably transforming growth factor-β (TGF-β), may overrule the messages conveyed by mitogenic factors and force a halt to proliferation. In addition, extracellular signals may persuade a cell to enter into a post-mitotic, differentiated state from which it will never re-emerge to resume proliferation. These disparate signals are collected by dozens of distinct cell surface receptors and then funneled into the complex signal-processing circuitry that operates largely in the cell cytoplasm. In some way, this mixture of signals must be processed, integrated, and ultimately distilled down to some simple, binary decisions made by the cell as to whether it should proliferate or become quiescent, and whether, as a quiescent cell, it will or will not differentiate. These behaviors suggest the existence of some centrally acting governor that operates inside the cell—a master clearinghouse that receives a wide variety of incoming signals and makes major decisions concerning the fate of the cell. Movies in this chapter 8.1 Animal Cell Division This master governor has been identified. It is the cell cycle clock, which operates 8.2 Regulation of CDK in the cell nucleus. Its name is really a misnomer, since this clock is hardly a device Function for counting the passage of time. Nonetheless, we will use this term here for want of 275 276 Chapter 8: pRb and Control of the Cell Cycle Clock Figure 8.1 The central governor of growth monitors monitors factor of genome TGF-β of cell growth and proliferation The term receptors integrity receptors integrins metabolism “cell cycle clock” denotes a molecular circuitry operating in the cell nucleus that processes and integrates a variety of afferent (incoming) signals originating from outside and inside the cell and decides whether or not the cell should CELL enter into the active cell cycle or retreat CYCLE into a nonproliferating state. In the CLOCK event that active proliferation is decided upon, this circuitry proceeds to program the complex sequence of biochemical changes in a cell that enable it to double its contents and to divide into two programming daughter cells. OR of cell cycle phases enter into enter into active G1 S G2 M G0 cell cycle quiescent state a better one. Rather than counting elapsed time, the cell cycle clock is a network of interacting proteins—a signal-processing circuit—that receives signals from various sources both outside and inside the cell, integrates them, and then decides the cell’s fate. Should the cell cycle clock decide in favor of proliferation, it proceeds to orches- trate the complex transitions that together constitute the cell’s cycle of growth and division. Should it decide in favor of quiescence, it will use its agents to impose this nonproliferative state on the cell (Figure 8.1). The proliferative behavior of cancer cells indicates that the master governor of the cell’s fate is influenced not only by normal proteins but also by oncogene proteins that insert themselves into various TBoC2signaling pathways and disrupt normal control b8.01/8.01 mechanisms. Similarly, the deletion of key tumor suppressor proteins evokes equally profound changes in the control circuitry and thus is equally influential in perturbing decision making by the cell cycle clock. Consequently, sooner or later, the molecular actions of most oncogenes and tumor suppressor genes must be explained in terms of their effects on the cell cycle clock. To this end, we will devote the first half of this chapter to a description of how this molecular machine normally operates and then proceed to study how it is perturbed in human cancer cells. 8.1 Cell growth and division is coordinated by a complex array of regulators When placed into culture under conditions that encourage exponential multiplica- tion, mammalian cells exhibit a complex cycle of growth and division that is usually referred to as the cell cycle. A cell that has recently been formed by the processes of cell division—mitosis and cytokinesis—must decide soon thereafter whether it will once again initiate a new round of active growth and division or retreat into the non- growing state that we previously termed G0. As described earlier (Sections 5.1 and 6.1), this decision is strongly influenced by mitogenic growth factors in the cell’s surround- ings. Their presence in sufficient concentration will encourage a cell recently formed by mitosis to remain in the active growth-and-division cycle; their absence will trigger the default decision to proceed from mitosis into the G0, quiescent state. Withdrawal from the cell cycle may be actively encouraged by the presence of growth- inhibitory factors in the medium. Prominent among these anti-mitogenic factors is transforming growth factor-β (TGF-β). Withdrawal from the cell cycle into the G0, qui- escent state, whether due to the absence of mitogenic growth factors or the presence of anti-mitogens such as TGF-β, is often reversible, in that an encounter by a quiescent  Organization of the Cell Cycle 277 cell with mitogenic growth factors on some later occasion may induce this cell to re- enter into active growth and division. However, some cells leaving the active cell cycle may do so irreversibly, thereby giving up all option of ever re-initiating active growth and division, in which case they are said to have become post-mitotic. Many types of neurons in the brain, for example, are widely assumed to fall into this category. The decision by a cell recently formed by cell division to remain in the active growth- and-division cycle requires that this cell immediately begin to prepare for the next division. Such preparations entail, among other things, the doubling of the cell’s macromolecular constituents to ensure that the two daughter cells resulting from the next round of cell division will each receive an adequate endowment of them. This accumulation of cellular constituents, which drives an increase in cell size, is some- times termed the process of cell growth to distinguish it from the process of cell divi- sion, which yields, via mitosis and cytokinesis, two daughter cells from a mother cell (Figure 8.2). However, in the more common usage and throughout this book, the term “cell growth” implies both the accumulation of cell constituents and the subsequent cell division, that is, the two processes that together yield cell proliferation. The accumulation of a cell’s macromolecules involves, among many other molecules, the duplication of the cell’s genome. In many prokaryotic cells, this duplication—the process of DNA replication—begins immediately after daughter cells are formed by cell division. But in most mammalian cells, the overall program of macromolecular synthesis is organized quite differently. While the accumulation of RNA and proteins is initiated immediately after cell division and proceeds continuously until the next cell division, the task of replicating the DNA is deferred for a number of hours (often as DNA Akt/PKB actin (A) (B) (C) Figure 8.2 Growth versus proliferation Alterations of certain (brown) are labeled with an antibody against phosphorylated S6, signaling proteins, such as the one encoded by the TSC1 tumor a ribosomal protein important in regulating protein synthesis and suppressor gene, allow the processes of cell growth and division thus cell growth; S6 phosphorylation and functional activation is to be uncoupled from one another. (A) In this scanning electron deregulated in cells lacking TSC1 function. (C) The Akt/PKB protein micrograph of a Drosophila eye, the ommatidial cells in the is frequently hyperactivated in human cancers. In an imaginal disc upper portion of the eye have been deprived of the fly ortholog of a developing Drosophila larva, Akt/PKB (green) that has been of TSC1; these cells are physically larger than the wild-type cells hyperactivated in a portion of cells (upper right) causes a great shown (below), because they have grown more during the cell increase in nuclear and overall cell size relative to cells with normal cycles that led to their formation. (B) The same behavior can be Akt/PKB (lower left); similar increases in mammalian cells are also seen in the brains of patients suffering from tuberous sclerosis, in observed in response to hyperactive Akt/PKB. (A, from X. Gao which TSC1 function has been lost through a germ-line mutation and D. Pan, Genes Dev. 15:1383–1392, 2001. B, courtesy of and subsequent somatic loss of heterozygosity. Seen here are the J.A. Chan and D.J. Kwiatkowski. C, from J. Verdu et al., Nat. Cell giant cells present in a benign growth (a “tuber”). The giant cells Biol. 1:500–506, 1999.) 278 Chapter 8: pRb and Control of the Cell Cycle Clock (A) (B) prophase, metaphase, anaphase, telophase G0 M G2 interphase prophase G1 S Figure 8.3 The mammalian cell cycle (A) Immunofluorescence is used here to illustrate the four distinct subphases of mitosis (M phase) in newt lung cells (top to bottom). During the prophase of mitosis, the chromosomes (blue), which were invisible microscopically during interphase prometaphase (the period encompassing G1 through S and G2), begin to condense and become visible, while the centrosomes (light green radiating bodies) at the poles of the cell begin to assemble (top two images). During metaphase, the chromosomes align along a plane that bisects the cell and become attached to the microtubule fibers of the mitotic spindle (light green, third image). At the same time, the nuclear membrane has disappeared. During anaphase, the two halves of each chromosome—the chromatids—are pulled apart by the mitotic spindle (i.e., they segregate) to the two opposite poles of the cell (fourth image). During telophase, shortly after the chromatids cluster into the two sets seen here (bottom image), the chromatids de-condense and a new nuclear membrane forms around each set of chromatids (now called chromosomes; not shown). At the same time, during the process of cytokinesis, the cytoplasm of the mother cell divides, yielding two daughter cells. (B) The mammalian growth-and-division cycle is divided into four metaphase phases—G1, S (during which DNA is replicated), G2, and M (mitosis). A fifth state, G0 (G zero), denotes the resting, nonproliferating state of cells that have withdrawn from the active cell cycle. While exit from the active cell cycle into G0 is depicted here as occurring in early G1, it is unclear when during G1 this actually occurs. (A, courtesy of Conly Rieder.) many as 12 to 15) after emergence of new daughter cells from mitosis and cytokinesis. During this period between the birth of a daughter cell and the subsequent onset of DNA synthesis, which is termed the G1 (first gap) phase of the cell cycle (Figure 8.3), cells make critical decisions about growth versus quiescence, and whether, as quies- cent cells, they will differentiate. anaphase In many types of cultured mammalian cells, the DNA synthesis that follows G1 often requires 6 to 8 hours to reach completion. This period of DNA synthesis is termed the S (synthetic) phase, and its length is determined in part by the enormous amount of cellular DNA (~6.4 × 109 base pairs per diploid genome) that must be replicated with fidelity during this time. The actual length of S phase varies greatly among different kinds of cells, being much shorter in certain cell types, such as rapidly dividing embry- onic cells and lymphocytes. Having passed through S phase, a cell might be thought fit to enter directly into mito- sis (M phase). However, most mammalian cells spend 3 to 5 hours in a second gap phase, termed G2, preparing themselves, in some still-poorly understood fashion, for entrance into M phase and cell division. M phase itself usually encompasses an hour telophase  Organization of the cell cycle 279 or so, and includes four distinct subphases—prophase, metaphase, anaphase, and telophase; this culminates in cytokinesis, the division of the cytoplasm that allows the formation of two new cells (see Figure 8.3A). While these times are commonly observed when studying mammalian cells in culture, they do not reflect the behavior of all cell types under all conditions. For example, actively proliferating lymphocytes may double in 5 hours, and some cells in the early embryo may do so even more rap- idly. As is the case with S phase, M phase must proceed with great precision. M phase begins with the two recently duplicated DNA helices within each chromosome; these are carried in the sister chromatids of the chromosome, which are aligned adjacent to one another in the nucleus. The allocation during mitosis of the duplicated chroma- tids to the two future daughter cells must occur flawlessly to ensure that each daugh- ter receives exactly one diploid complement of chromatids—no more and no less. Once present within the nuclei of recently separated daughter cells, these chromatids become the chromosomes of the newly born cells. This means that the endowment of one genome’s worth of genetic material to each daughter cell depends on the precise execution of two processes—the faithful replica- tion of a cell’s genome during S phase, and the proper allocation of the resulting dupli- cated DNAs to daughter cells during M phase. As will be discussed later, defects in either of these processes can have disastrous consequences for the cell and the organ- ism, one of which is the disease of cancer. Like virtually all machinery, the machine that executes the various steps of the cell cycle is subject to malfunction. This fallibility contrasts with the stringent requirement of the cell to have the various phases of the cell cycle proceed flawlessly. For this rea- son, the cell deploys a series of surveillance mechanisms that monitor each step in cell cycle progression and permit the cell to proceed to the next step in the cycle only if a prerequisite step has been completed successfully. In addition, if specific steps in the execution of a process go awry, these monitors rapidly call a halt to further advance through the cell cycle until these problems have been successfully addressed. Yet other monitors ensure that once a particular step of the cell cycle has been completed, it is not repeated until the cell passes through the next cell cycle. These monitoring mechanisms are termed variously checkpoints or checkpoint controls (Figure 8.4). entrance into M anaphase blocked if blocked if DNA chromatids are not replication is properly assembled not completed on mitotic spindle Figure 8.4 Examples of checkpoints in the cell cycle Checkpoints impose quality control to ensure that a cell has M properly completed all the requisite steps G2 of one phase of the cell cycle before it is allowed to advance into the next phase. A cell will not be permitted to enter into S phase until all the steps of G1 have been completed. It will be blocked from entering G2 until all of its chromosomal G1 DNA has been properly replicated. Similarly, a cell is not permitted to enter into anaphase (when the paired S chromatids are pulled apart) until all of its chromosomes are properly assembled on the mitotic spindle during metaphase. In addition, a cell is not allowed to advance into S or M if its DNA has been damaged and not yet repaired. Other DNA damage DNA damage controls (not shown) ensure that once checkpoint: DNA checkpoint: entrance a specific step in the cell cycle has been replication halted if into S is blocked if completed, it is not repeated until the genome is damaged genome is damaged next cell cycle. 280 Chapter 8: pRb and Control of the Cell Cycle Clock (A) (B) Figure 8.5 Consequences of loss of checkpoint controls Loss of critical cell cycle checkpoint control mechanisms is often manifested as an altered karyotype. (A) The normal human karyotype (left) is contrasted with that of a cell that has been deprived of the Rad17 checkpoint protein (right), which is responsible for preventing inadvertent re-replication of already-replicated chromosomal DNA, resulting in endoreduplication and increases in the ploidy of the cell. (B) The Bub1 protein normally prevents separation of chromosomes in the event that one or more chromosomal pairs are not properly aligned on the metaphase plate. In its absence, cells gain or lose chromosomes, as seen in this spectral karyotyping (SKY) analysis, which indicates that this human cell has only (C) one Chromosome 1 (yellow, arrow, right) and one Chromosome 6 (red, arrow, left). One checkpoint ensures that a cell cannot advance from G1 into S if the genome is (Several other red-colored chromosomes in need of repair. Another, operating in S phase, will slow or pause DNA replication are evident but can be identified as in response to DNA damage. (In mammalian cells, this may cause a doubling of the belonging to other chromosome pairs time required to complete DNA synthesis.) A third will not permit the cell to proceed by their size and shading.) (C) The ATR through G2 to M until the DNA replication of S phase has been completed. DNA dam- (ataxia telangiectasia and Rad3 related) protein kinase is responsible for, among age will trigger another checkpoint control that blocks entrance into M phase. During other functions, halting further DNA M phase, highly efficient checkpoint controls block anaphase; these blocks are only replication until stalled replication forks removed once all the chromosomes have been properly attached to the mitotic spin- are repaired. In its absence, fragile dle. Yet other checkpoint controls, not cited here, have been reported. For example, sites—sites in the chromosome prone a decatenation checkpoint in late G2 prevents entrance into M until the pair of DNA to breakage—become visible upon helices replicated in the previous S phase have been untangled from one another. karyotypic analysis. Here, fragile sites on Defects in some of these checkpoint controls can be observed because of their effects human Chromosomes 3 and 16 (white on cells’ chromosomes (Figure 8.5). arrows) are apparent in cells lacking ATR function. (A, from X. Wang et al., Genes The operations of these checkpoints also influence the formation of cancers. As tumor Dev. 17:965–970, 2003. B, from A. development (often called tumor progression, discussed in Chapter 11) proceeds, Musio et al., Cancer Res. 63:2855–2863, incipient cancer cells benefit from experimenting with various combinations and per- 2003. C, from A.M. Casper et al., Am. J. mutations of mutant alleles, retaining those that will afford them the greatest prolif- Hum. Genet. 75:654–660, 2004.) erative advantage. An increased mutability of their genomes accelerates the rate at which theseBoC2 cells b8.05/8.05 can acquire advantageous combinations of alleles and thus hastens the overall pace of tumor progression. Such mutability and resulting genomic insta- bility is incompatible with normal cell cycle progression, since checkpoint controls usually block advance of the cell through its cycle if its DNA has been damaged or its chromosomes are in disarray. So, in addition to acquiring altered growth-controlling genes (that is, activated oncogenes and inactivated tumor suppressor genes), many types of cancer cells have inactivated one or more of their checkpoint controls. With  Important decisions are made during G1 281 Sidebar 8.1 Embryonic stem (ES) cells show highly autono- termed LIF (leukemia inhibitory factor), which is needed to mous behavior Our perceptions about the behavior of normal prevent their differentiation, mouse ES cells proliferating in mammalian cells have been conditioned by decades of work vitro appear able to drive their own proliferation through inter- with a wide variety of somatic cells present in embryonic and nally generated signals. (Indeed, a constitutively activated Ras- adult tissues. However, cells in the very early embryo clearly like protein, termed E-Ras, has been reported to be expressed operate under a very different set of rules. The pRb pathway specifically in these cells.) and the cell cycle clock machinery to be described in this ES cells seem to preserve much of the cell-autonomous chapter appear to be operative in one form or another in virtu- behavior that we associate with the single-cell ancestors of ally all types of adult cells. In contrast, a variety of experiments metazoa, that is, the behavior of cells that have not yet become indicate that pRb-imposed growth control is not functional in dependent on their neighbors for signals controlling growth early embryonic cells, including their cultured derivatives, em- and survival. The most stunning indication of their extreme bryonic stem (ES) cells (Supplementary Sidebar 8.1). The same autonomy is their ability to form benign tumors (teratomas) can be said about the p53 pathway (see Chapter 9). It seems when introduced into many anatomical sites in an adult organ- that the mitogenic signals required to keep more differenti- ism. Since these cells are genetically wild type, they represent ated cells proliferating are not required by cultured ES cells for the only example of a wild-type cell that is tumorigenic. their proliferation. For example, aside from the growth factor these controls relaxed, incipient cancer cells can more rapidly accumulate the mutant genes and altered karyotypes that propel their neoplastic growth. The breakdown of the controls responsible for maintaining the cellular genome in an intact state is one of the main subjects of Chapter 12. 8.2 Cells make decisions about growth and quiescence during a specific period in the G1 phase As mentioned earlier, virtually all normal cell types in the body require external sig- nals, such as those conveyed by mitogenic growth factors, before they will undertake to grow and divide. The only exceptions to this rule appear to be very early embry- onic cells, which seem able to proliferate without receiving growth-stimulating sig- nals from elsewhere (Sidebar 8.1). The rationale for this behavior of the normal cells throughout our tissues is a simple one: since these cells participate in the formation of precisely structured tissues, their proliferation must, by necessity, be coordinated with neighboring cells in those tissues. Put differently, the body cannot give each of its almost 1014 component cells the license to decide on its own whether to grow and divide. To do so would invite chaos. Evidence accumulated over the past quarter century indicates that cells consult their extracellular environment and its growth-regulating signals during a discrete window of time in the active cell cycle, namely, from the onset of G1 phase until an hour or two before the G1-to-S transition (Figure 8.6). The operations of the G1 decision-making machinery are indicated by the responses of cultured cells to extracellular signals. If we were to remove serum and thus growth factors from cells before they had completed M G2 Figure 8.6 Responsiveness to extracellular signals during the cell cycle Cells respond to extracellular period during which cells are responsive mitogens and inhibitory factors (such as to mitogenic GFs TGF-β) only in a discrete window of time G1 and to TGF-β that begins at the onset of G1 and ends just before the end of G1. The end of this time window is often designated as the S restriction (R) point, which denotes the point in time when the cell must make the commitment to advance through the remainder of the cell cycle through M phase, to remain in G1, or to retreat from R point the active cell cycle into G0. 282 Chapter 8: pRb and Control of the Cell Cycle Clock 80 to 90% of G1, they would fail to proceed further into the cell cycle and would, with great likelihood, revert to the G0 state. However, once these cells had moved through this G1 decision-making period and advanced into the final hours of G1 (the remain- ing 10 to 20% of G1), the removal of serum would no longer affect their progress and they would proceed through the remainder of G1 and onward through the S, G2, and M phases. Similarly, anti-mitogenic factors, such as TGF-β, are able to impose their growth-inhibitory effects only during this period in the early and mid-G1 phase. Once a cell has entered into late G1, it seems to be oblivious to the presence of this negative factor in its surroundings. This schedule of total dependence on extracellular signals followed by entrance in late G1 into a state of relative independence indicates that a weighty decision must be made toward the end of G1. Precisely at this point, a cell must make up its mind whether it will remain in G1, retreat from the active cycle into G0, or advance into late G1 and thereafter into the remaining phases of the cell cycle. This critical decision is made at a transition that has been called the restriction point or R point (see Figure 8.6). In most mammalian cells studied to date, the R point occurs several hours before the G1/S phase transition. If a cell should decide at the R point to continue advancing through its growth-and- division cycle, it commits itself to proceed beyond G1 into S phase and then to com- plete a rigidly programmed series of events (the entire S, G2, and M phases) that ena- ble it to divide into two daughter cells. This decision will be respected even if growth factors are no longer present in the extracellular space during these remaining phases of the cell cycle. We know that this series of later steps (S, G2, and M phases) proceeds on a fixed schedule, because a cell that enters S will, in the absence of a major disaster, invariably complete S and, having done so, enter into G2 and then advance into M phase. For those interested specifically in understanding the deregulated proliferation of cancer cells, this fixed program holds relatively little interest, since the late G1 → S → G2 → M progression proceeds similarly in normal cells and cancer cells. Students of cancer therefore focus largely on the G0/G1 transition and on the one period in the life of an actively growing cell—the time window encompassing most of G1—when a cell is given the license to make decisions about its fate. The commitment to advance through the R point and continue all the way to M phase is, as hinted, not an absolute one. Metabolic, genetic, or physical disasters may inter- vene during S, G2, or M and force the cell to call a halt, often temporary, to its further advance through the cell cycle until these conditions have been addressed. Still, in the great majority of cases, cells in living tissues seem to succeed in avoiding these various disasters. This leaves the R-point decision as the critical determinant of whether cells will grow or not. An increasingly large body of evidence indicates that deregulation of the R-point decision-making machinery accompanies the formation of most if not all types of cancer cells. However, yet other decision points in late G1 may also contrib- ute to the deregulated proliferation of certain cancer cells. Prominent among these in mammals is the monitoring by cells of their attachment to the extracellular matrix (ECM) after the R point and before the G1/S transition. If a cell does not enjoy the proper attachment to the ECM, much of it mediated by integrins (see Section 5.9), it will halt further advance through the cell cycle until proper tethering has been achieved. Alternatively, if it has lost all attachment to the ECM, it may activate anoikis, a form of the apoptotic cell suicide program. When grown in monolayer culture in Petri dishes, normal cells achieve the required attachment by adhering to extracellular matrix that they have deposited on the glass or plastic surfaces of these dishes. This requirement for attachment reflects the phe- notype of anchorage dependence (see Section 3.5). Tumorigenic cells, which almost always have lost this dependence and have therefore become anchorage-independ- ent, have inactivated or lost this late G1 checkpoint. In some fashion, still poorly understood, oncoproteins like Ras and Src are able to mislead a cell into thinking that it has achieved extensive anchorage (see, for example, Section 6.9) when, in fact, none may exist at all.  Cyclin–CDK complexes regulate cell cycle progression 283 Yet other, still-fragmentary evidence suggests additional checkpoints that operate before and after the R point. For example, one important checkpoint control operating between the R point and the onset of S phase gauges whether a cell has access to ade- quate levels of nutrients and halts cell-cycle progression until such nutrients become available; another appears to determine whether reactive oxygen species (which in other contexts are thought to be toxic for the cell) are present in adequately high lev- els before progression is permitted. Such findings suggest that successful advance through late G1 depends on passing through a succession of checkpoints in addition to the R point described here, not all of which have been well characterized. 8.3 Cyclins and cyclin-dependent kinases constitute the core components of the cell cycle clock The existence of the R point leaves us with two major questions that we will spend much of this chapter answering. First, what is the nature of the molecular machinery that decides whether or not a cell in G1 will continue to advance through the cell cycle or will exit into a nongrowing state? Second, how does this machinery, which we call the cell cycle clock, implement these decisions once they have been made? We begin with the second question and will return later to the first. As described earlier (Chapters 5 and 6), when signals are broadcast from a single mas- ter control protein to many downstream responders, the signal-emitting functions are often delegated to protein kinases. These enzymes are ideally suited to this task. By phosphorylating multiple distinct targets (that is, substrates), a kinase can cre- ate covalent modifications that serve to switch on or off various activities inherent in these substrate proteins. Indeed, the cell cycle clock uses a group of protein kinases to execute the various steps of cell cycle progression. For example, phosphorylation of centrosome-associated proteins at the G1/S boundary allows their duplication in preparation for M phase. Phosphorylation of other proteins prior to S phase enables DNA replication sites along the chromosomes to be activated. Phosphorylation of histone proteins in anticipation of S and M phases places the chromatin in configu- rations that permit these two phases to proceed normally. And the phosphorylation of proteins forming the nuclear membrane (sometimes called the nuclear envelope), such as lamin and nucleoporins, causes their dissociation and the dissolution of this membrane early in M phase. The kinases deployed by the cell cycle machinery are called collectively cyclin- dependent kinases (CDKs) to indicate that these enzymes never act on their own; instead, they depend on associated regulatory subunits, the cyclin proteins, for proper functioning. Bimolecular complexes of CDKs and their cyclin partners are respon- sible for sending out the signals from the cell cycle clock to dozens if not hundreds of responder molecules that carry out the actual work of moving the cell through its growth-and-division cycle. The CDKs are serine/threonine kinases, in contrast to the tyrosine kinases that are associated with growth factor receptors and with nonreceptor kinase molecules like Src. The CDKs show about 40% amino acid sequence identity with one another and therefore are considered to form a distinct subfamily within the large (approximately 430) throng of Ser/Thr kinases encoded by the human genome (see Supplementary Sidebar 16.8). The cyclins associated with the CDKs activate the catalytic activity of their CDK partners (Figure 8.7). (In the well-studied example of the binding of cyc- lin A to CDK2, the association of the two proteins increases the enzymatic activity of CDK2 a staggering 400,000-fold!) At the same time, the cyclins serve as guide dogs for the CDKs by helping the cyclin–CDK complexes recognize appropriate protein sub- strates in the cell. The cyclins, for their part, also constitute a distinct family of cel- lular proteins that share in common an approximately 100–amino acid residue–long domain that is involved in the binding and functional activation of CDKs. It is actually cyclin–CDK complexes that constitute the engine of the cell cycle clock machinery. During much of the G1 phase of the cell cycle, two similarly acting CDKs— CDK4 and CDK6—are guided by and depend upon their association with a trio of 284 Chapter 8: pRb and Control of the Cell Cycle Clock (A) (B) PSTAIRE helix PSTAIRE helix ATP cyclin E activation cyclin A CDK2 loop activation CDK2 loop Figure 8.7 Cyclin–CDK complexes related cyclins (D1, D2, and D3) that collectively are called the D-type cyclins (Figure X-ray crystallographic analyses have 8.8). After the R point in late G1 the E-type cyclins (E1 and E2) associate with CDK2 to revealed the structures of cyclin–cyclin- enable the phosphorylation of appropriate substrates required for entry into S phase. dependent kinase (CDK) complexes, As cells enter into S phase, the A-type cyclins (A1 and A2) replace E cyclins as the part- such as these formed by CDK2 with ners of CDK2 and thereby enable S phase to progress (see Figure 8.7). Later in S phase, two of its alternative partners, cyclins the A-type cyclins switch partners, leaving CDK2 and associating instead with another A and E. In each case, the cyclin CDK called either CDC2 or CDK1. (We will use CDC2 here.) As the cell moves further activates the CDK molecule through into G2 phase, the A-type cyclins are replaced as CDC2 partners by the B-type cyclins stereochemical shifts of the CDK catalytic (B1 and B2). Finally, at the onset of M phase, the complexes of CDC2 with the B-type site and directs the resulting activated complex to appropriate substrates for cyclins trigger many of the events of the prophase, metaphase, anaphase, and telo- phosphorylation. (A) During the late phase that together constitute the complex program of mitosis. [Another cyclin–CDK G1 phase of the cell cycle, cyclin E pair—cyclin C–CDK3—is implicated in the movement of cells from the G0 into the TBoC2 b8.07/8.07 directs CDK2 to substrate proteins that G1 phase (see Figure 8.3A) of the cell cycle. However, because most strains of inbred must be phosphorylated in preparation for entrance into S phase. Another B CDC2 segment (orange) is involved with cyclin E’s association with centrosomes. The PSTAIRE α-helix is present in all CDKs and is essential for binding of cyclins. The activation loop, sometimes termed M a T-loop, must be phosphorylated on a G2 threonine residue by a CDK-activating kinase (CAK) in order for the catalytic D CDK4/6 function of a CDK to become activated (see Figure 5.17). (B) During the course of S phase, cyclin A replaces cyclin E and A CDC2 G1 directs CDK2 to substrates that must be phosphorylated in order for S phase to R point proceed. (A, from R. Honda et al., EMBO S J. 24:452–463, 2005. B, from P.D. Jeffrey et al., Nature 376:313–320, 1995.) E CDK2 A CDK2 Figure 8.8 Pairing of cyclins with cyclin-dependent kinases Each type of cyclin pairs with a specific cyclin-dependent kinase (CDK) or set of CDKs. The D-type cyclins (D1, D2, and D3) bind CDK4 or CDK6, the E-type (E1 and E2) bind CDK2, the A-type cyclins (A1 and A2) bind CDK2 or CDC2, and the B-type cyclins (B1 and B2) bind CDC2. The brackets indicate the periods during the cell cycle when these various cyclin–CDK complexes are active.  Cyclin–CDK complexes regulate cell cycle progression 285 B B B Figure 8.9 Cell cycle–dependent fluctuations in cyclin B levels The cyclic fluctuations in the levels of cyclin B in early frog and sea urchin embryos gave cyclins their name. These fluctuations were noticeable because the cell cycles in these early embryos are synchronous, i.e., all cells enter into M phase simultaneously. In these early S M S M S M embryos, the G1 and G2 phases of the cell cycle (orange, pink chevrons) are laboratory mice lack CDK3, it seems to be unnecessary for normal cell cycle control virtually absent and the cells, in effect, and remains poorly studied.] alternate between M and S phases. (Although cyclin B levels are already As is the case with all well-regulated systems, the activities of the various cyclin–CDK substantial prior to the onset of M phase, complexes must be modulated in order to impose control on specific steps in the cell cyclin B molecules are unable to form cycle. The most important way of achieving this regulation depends upon changing catalytically active B–CDC2 complexes the levels and availability of cyclins during various phases of the cell cycle. In contrast, until the G2-to-M transition.) the levels of most CDKs vary only minimally. The first insights into cyclin and CDK control came from studies of the governors of TBoC2 b8.09/8.09 mitosis in early frog and sea urchin embryos. As these experiments showed, levels of B-type cyclins increase strongly in anticipation of mitosis, allowing B cyclins and CDC2 to form complexes that initiate entrance into M phase. At the end of M phase, cyclin B levels plummet because of the scheduled degradation of this protein. Early in the next cell cycle, cyclin B is virtually undetectable in cells, and accumulates gradu- ally later in this cycle in anticipation of the next M phase. Because the growth-and- division cycles of all of the cells in these early embryos are synchronous (that is, take place coordinately), all cells in the embryo go through S and M at the same time. This results in repeated rounds of cycling of the levels of these cyclin proteins—the behav- ior that inspired their name (Figure 8.9). This theme of dramatic cell cycle phase–dependent changes in the levels of cyclin B is repeated by other cyclins as well. Cyclin E levels are low throughout most of G1, rise abruptly after a cell has progressed through the R point, and collapse as the cell enters S phase (Figure 8.10), while cyclin A increases in concert with the cell’s entrance into S phase. (While there are at least two subtypes of cyclin A, cyclin B, and cyclin E, we will refer to these simply as cyclins A, B, and E, respectively, since the two subtypes of each of these cyclins appear to operate identically.) The collapse of various cyclin species as cells advance from one cell cycle phase to the next is due to their rapid degradation, this being triggered by the actions of highly coordinated ubiquitin ligases, which attach polyubiquitin chains to these cyclins (see Supplementary Sidebar 7.5). This polyubiquitylation leads to proteolytic break- Figure 8.10 Fluctuation of cyclin down in the proteasomes. The cyclins’ gradual accumulation followed by their rapid levels during the cell cycle The levels destruction has an important functional consequence for the cell cycle, because it dic- of most of the mammalian cyclins tates that the cell cycle clock can move in only one direction, much like a ratchet. This fluctuate dramatically as cells progress ensures, for example, that cells that have exited one M phase cannot inadvertently slip through the cell cycle. For most of backward into another one, but instead must advance through G1, S, and G2 until they these cyclins, these fluctuations are have once again accumulated the B cyclins required for entrance into M phase. tightly coordinated with the schedule of advances through the various cell cycle phases. However, in the case of cyclin B nuclear D1 cyclin E cyclin A the D-type cyclins, extracellular signals, notably those conveyed by growth factors, strongly influence their levels. (While cyclin D1—and possibly other D-type cyclins—is present in other cell cycle phases besides G1, following the G1/S transition it is exported from the nucleus into the cytoplasm, where M G1 S G2 M G1 it can no longer influence cell cycle progression. The precise time point at which cyclin D–CDK4/6 complexes lose R point activity is not well defined.) 286 Chapter 8: pRb and Control of the Cell Cycle Clock (A) 15 min 30 min Figure 8.11 Control of cyclin D1 levels 11 h 14 h 17 h 23 h (A) Cyclin D1 was discovered as a protein 0h 1h 2h 4h 6h 8h whose levels are strongly induced by exposure of macrophages to the mitogen 4.5 kb CSF-1 (colony-stimulating factor-1). Here, 3.8 kb macrophages that had been starved of CSF-1 were exposed to fresh CSF-1 and the amounts of cyclin D1 mRNA at subsequent times thereafter were determined by RNA (Northern) blotting. (B) growth factor ECM (B) The control of cyclin D1 levels by extracellular mitogens can be explained, tyrosine kinase integrins in part, by a signal transduction cascade receptors (RTKs) that leads from growth factor receptors (RTKs) to the AP-1 transcription factor Grb2 (TF), one of a number of factors that modulate the transcription of the cyclin D1 gene. (AP-1 is a heterodimeric TF Sos formed from Fos and Jun subunits, each encoded by a proto-oncogene.) In Ras addition, a number of other signaling cascades converge on the promoter of Raf Ral-GDS Ral-GDS this gene, not all of which are illustrated in this figure. (A, from H. Matsushime et MEK Rac al., Cell 65:701–713, 1991.) GSK-3β PAK Erk MEK p90Rsk MKK4 Elk-1/SRF JNK Fos Jun AP-1 growth HER2/Neu Sp1 D1 Tcf/Lef β-catenin Frizzled Wnts factor STAT Gli NF-κB Jak Smoothened IKK cytokine various Patched receptor receptors cytokine various Hedgehog ligands The sole exception to these well-programmed fluctuations in the levels of cyclins is presented by the D-type cyclins. The levels of these three, similarly structured cyclins are not found to vary dramatically as a cell advances through the various phases of its growth-and-division cycle. Instead, the levels of D-type cyclins are controlled largely by extracellular signals, specifically those conveyed by a variety of mitogenic growth TBoC2 factors. In the case of cyclin D1—the b8.11/8.11 best-studied of the three D-type cyclins—growth factor activation of tyrosine kinase receptors and the resulting stimulation of several downstream signaling cascades results in rapid accumulation of cyclin D1 (Figure 8.11). Conversely, removal of growth factors from a cell’s medium results in an equally rapid collapse of its cyclin D1 levels with a half-life of about 30 minutes.  Extracellular signals regulate D-type cyclins 287 Table 8.1 Induction of D-type cyclin expression by extracellular signals Source of signal Signaling intermediaries Type of cyclin RANK receptor NF-κB pathway D1 Prolactin receptor Jak/STAT D1 Estrogen receptor AP-1 TF (?) D1 Focal adhesion kinase D1 HER2/Neu receptor E2F and Sp1 TFs D1 Wnts–Frizzled receptor β-catenin and Tcf/Lef TFs D1 Sidebar 8.2 Cyclins have other jobs Bcr/Abl D2 besides cell cycle control Two dec- FSH receptor cyclic AMP D2 ades of cell cycle research have cre- ated the impression that cell cycle Various mitogens Myc D2 control is the sole function of cyclins. In fact, cyclin D1 has been shown to Interleukin-4, 7 receptor D2 associate with both the estrogen re- Interleukin-5 receptor STAT3/5 D3 ceptor (ER) and the transcription factor C/EBPβ. By binding the ER, Mitogens E2A TF D3 cyclin D1 may mimic the normal lig- and of this receptor—estrogen—in Abbreviations: RANK, receptor activator of NF-κB; FSH, follicle-stimulating hormone. stimulating the receptor’s transcrip- tional activities. The great majority (>70%) of breast cancers express the ER, and its expression explains the mitogenic effects that estrogen has The distinctive behavior of the D-type cyclins has been rationalized as follows: They on the cells in these tumors. Since serve to convey signals from the extracellular environment to the cell cycle clock oper- cyclin D1 is overexpressed in most ating in the cell nucleus. Because the levels of D-type cyclins fluctuate together with of these tumors and can activate this the levels of extracellular mitogens, D-type cyclins continuously inform the cell cycle receptor, the D1–ER complexes may clock of current conditions in the environment around the cell. also play an important role in driv- ing the proliferation of cells in these After D-type cyclins are synthesized in the cytoplasm and migrate to the nucleus, they tumors. The association of cyclin D1 assemble in complexes with their two alternative CDK partners, CDK4 and CDK6. with C/EBPβ results in activation Since these two CDKs function similarly to one another, we will refer to them hereaf- of this transcription factor, which ter as CDK4/6. The cyclin D–CDK4/6 complexes seem to have similar if not identical is thought to play a key role in pro- enzymatic activities and substrate specificities, independent of whether they contain gramming the differentiation of a va- cyclin D1, D2, or D3. riety of cell types. Knockout of cyclin D1 in the mouse germ line results These similarities provoke the question of why mammalian cells express three appar- in viable mice that have severely ently redundant cyclins. What appears to be a redundancy actually offers the cell underdeveloped mammary glands. refined sensory input and enhanced flexibility of response. The promoter of each of This effect can be reversed by provid- the three encoding genes is under the control of a different set of signaling pathways ing the mammary glands with a mu- and thus under the control of a different set of cell surface receptors (Table 8.1). For tant cyclin D1 that can bind C/EBPβ example, the promoter of the cyclin D1 gene (properly termed, in humans, CCND1) but cannot activate CDK4/6. Hence, carries sites for binding by the AP-1, Tcf/Lef, and NF-κB transcription factors (see Fig- in this case, the non-CDK-associ- ure 8.11B), which in turn are activated by a variety of growth factor receptors. In con- ated functions of this cyclin are far trast, the cyclin D2 promoter is responsive to activation by the Myc transcription factor more important than its effects on and to extracellular signals that stimulate increases in the concentration of intracellu- CDK4/6 activation. More globally, lar cyclic adenosine monophosphate (cAMP). The cyclin D3 gene has been found to be the use of chromatin immunopre- responsive to STAT3 and STAT5 transcription factors, and these in turn often respond cipitation (ChIP; see Supplementary to interleukin receptors active in various types of hematopoietic cells; the E2A tran- Sidebar 8.3) has revealed at least 700 scription factor, which is active in programming the differentiation of lymphocytes, sites in the genome, most affiliated also controls cyclin D3 expression. with gene promoters, to which cyc- lin D1 is bound, indicating that the This arrangement enables a diverse set of extracellular signals to influence the two associations of cyclin D1 with activities of CDK4/6 by controlling the levels of its various D-type cyclin partners. In transcription factors, as cited above, addition, more recent research has indicated that certain cyclins have functions in are likely only the tips of a far larger the cell that are apparently unrelated to their role in promoting cell cycle progression iceberg. (Sidebar 8.2). 288 Chapter 8: pRb and Control of the Cell Cycle Clock PROGRAM INFLUENCED BY EXTRACELLULAR SIGNALS CELL-AUTONOMOUS PROGRAM R point D D–CDK4/6 E E–CDK2 A A–CDK2 A–CDC2 B B–CDC2 mitogens CDK4/6 CDK2 CDC2 G1 S G2 M Figure 8.12 Control of cyclin levels during the cell cycle predictable schedule that appears to be independent of extracellular While extracellular signals strongly influence the levels of D-type physiologic signals. In part, this coordination is achieved because cyclins during most of the G1 phase of the cell cycle, the levels of the cyclin–CDK complexes in one phase of the cell cycle are the remaining cyclins are controlled by intracellular signaling and responsible for activating those in the subsequent phase (indicated precisely coordinated with cell cycle advance. Thus, after cells pass here) and for shutting down those that were active in the previous through the R point and cyclin E–CDK2 complexes are activated, phase (not shown). the activation of the remaining cyclin–CDK complexes occurs on a Once they are formed, the cyclin D–CDK4/6 complexes are capable of ushering a cell all the way from the beginning of the G1 phase up to and perhaps through the R-point gate. After the cell has moved through the R point, the remaining cyclins—E, A, and B—behave in a pre-programmed fashion, executing the fixed program that begins at the R point and extends all the way to the end of M phase (Figure 8.12). Indeed, once the cell has passed through its R point, its cell cycle machinery takes on a life of its own that is quite autonomous and no longer responsive to extracellular signals. In ways that remain poorly understood, cyclin–CDK complexes in later phases sup- press the activities of the cyclin–CDK complexes that have preceded them in earlier phases of the cell cycle. TBoC2 For example, when cyclin A is activated by the actions of cyclin b8.12/8.12 E–CDK2 complexes during the G1/S transition, the activities of cyclin A–CDK2 result in, among other things, the inactivation of the transcription factor that served previ- ously during the R-point transition to induce cyclin E expression. In late S and early G2 phases, the cyclin A–CDC2 complex begins to prepare for the activation of cyclin B–CDC2 complexes that are required later for entrance into mitosis. Once the latter complexes are activated, they seem to cause a shutdown of cyclin A synthesis, and so forth. While the program depicted in Figure 8.12 has been validated in a number of cultured cell types, it may not apply to all cells, as indicated by studies of genetically modified mice (see Supplementary Sidebar 8.2). 8.4 Cyclin–CDK complexes are also regulated by CDK inhibitors The scheme of cell cycle progression laid out above implies that physiologic signals are able to influence the activity of the cell cycle clock only through their modulation of cyclin levels. In truth, there are several other layers of control that modulate the activ- ity of the cyclin–CDK complexes and thereby regulate advance through the cell cycle. The most important of these additional controls is imposed by a class of proteins that are termed generically CDK inhibitors or simply CKIs. To date, seven of these pro- teins have been found that are able to antagonize the activities of the cyclin–CDK complexes. A group of four of these, the INK4 proteins (named originally as inhibi- tors of CDK4), target specifically the CDK4 and CDK6 complexes; they have no effect  CDK inhibitors also regulate CDKs 289 on CDC2 and CDK2. These inhibitors are p16INK4A, p15INK4B, p18INK4C, and p19INK4D. The three remaining CDK inhibitors, p21Cip1 (sometimes termed p21Waf1), p27Kip1, and p57Kip2, are more widely acting, being able to inhibit all of the other cyclin–CDK complexes that form at later stages of the cell cycle (Figure 8.13). (Each one of this trio of CKIs has been found to affect other specific processes besides cell cycle pro- gression, including transcriptional regulation, apoptosis, cell fate determination, cell migration, and cytoskeletal organization; however, since these other functions appear to play relatively minor roles in cancer development, we will not discuss them fur- ther.) p57Kip2 plays a key role in certain tissues during embryogenesis but a minor role in cancer development; for this reason, our discussion will focus on its two cousins, p21Cip1 and p27Kip1. (A) p57Kip2 p16INK4A p27Kip1 p15INK4B p18INK4C p19INK4D p21Cip1 D–CDK4/6 E–CDK2 A–CDK2 A–CDC2 B–CDC2 (B) (C) p27Kip1 p16INK4A CDK6 cyc A CDK2 Figure 8.13 Actions of CDK inhibitors (A) The CDK inhibitors block the actions of CDKs at various points in the cell cycle. The four INK4 proteins (p16INK4A, p15INK4B, p18INK4C, and p19INK4D) are specialized to inhibit the D–CDK4 and D–CDK6 complexes that are active in early and mid-G1. Conversely, the three Cip/Kip CKIs (p21Cip1, p27Kip1, and p57Kip2) can inhibit the remaining cyclin–CDK complexes that are active throughout the cell cycle. Under certain conditions, signals originating within the cell (e.g., signals indicating damage to the cell’s DNA) can halt advance through the cell cycle by blocking the actions of cyclin/CDK complexes that are active in the S and G2TBoC2 of the cell cycle. Paradoxically, p21Cip1 and p27Kip1 are phasesb8.13/8.13 known to promote formation of D–CDK4/6 complexes during the G1 phase (see Figure 8.17). (B) This depiction illustrates how one domain of p27Kip1 (light green) blocks cyclin A–CDK2 function by obstructing the ATP-binding site in the catalytic cleft of the CDK (see also Figure 8.7). (C) Inhibitors of the INK4 class, such as p16INK4A shown here (brown), bind to CDK6 (reddish) and to CDK4 (not shown). These CDK inhibitors distort the cyclin-binding site of CDK6, reducing its affinity for D-type cyclins. At the same time, they distort the ATP-binding site and thereby compromise catalytic activity. Identical interactions likely characterize the responses of CDK4 to p16INK4A. (B, from A.A. Russo et al., Nature 382:325–331, 1996. C, courtesy of N.P. Pavletich and from A.A. Russo et al., Nature 395:237–243, 1998.) 290 Chapter 8: pRb and Control of the Cell Cycle Clock (A) (B) exposure to TGF-β plasma membrane 0 2 4 6 8 14 hrs p15INK4B mRNA TGF-β DNA damage P relatively p21Cip1 weak p15INK4B D CDK4/6 E–CDK2 A–CDK2 A–CDC2 B–CDC2 Figure 8.14 Control of cell cycle The actions of these two classes of CDK inhibitors are nicely illustrated by p15INK4B progression by TGF-β (A) TGF-β (top and the pair p21Cip1 and p27Kip1. When TGF-β is applied to epithelial cells, it elicits a left) controls the cell cycle machinery in number of downstream responses that antagonize cell proliferation. Among these are part through its ability to modulate the substantial increases in the levels of p15INK4B, which proceeds to block the formation levels of CDK inhibitors. It acts to strongly of cyclin D–CDK4/6 complexes (Figure 8.14) and to inhibit those that have already induce increased expression of p15INK4B formed. Without active D–CDK4/6 complexes, the cell is unable to advance through and, weakly, of p21Cip1. The former can early and mid-G1 and reach the R point. Once a cell has passed through the R point, the block the actions of cyclin D–CDK4/6 actions of the D–CDK4/6 complexes seem to become unnecessary. This may explain complexes, while the latter can block why TGF-β is growth-inhibitory during early and mid-G1 and loses most (perhaps all) the actions of the remaining cyclin–CDK complexes that are active throughout the of its growth-inhibitory powers once a cell has passed through the R point. remainder of the cell cycle. Independent p21Cip1, a more widely acting CDK inhibitor, is also induced by TGF-β, albeit weakly. of this, damage to cellular DNA causes Far more important are increases in the levels of p21Cip1 that occur in response to vari- strong, rapid increases in p21Cip1, which ous physiologic stresses (see Figure 8.14A); once present at significant levels, p21Cip1 in turn can shut down the cyclin–CDK complexes that are active in the phases can act throughout much of the cell cycle to stop a cell in its tracks. Prominent among of the cell cycle after the cell has passed these TBoC2 stressesb8.14/8.14 is damage to the cell’s genome. As long as the genomic DNA remains through the R point in the late G1 phase. in an unrepaired state, the p21Cip1 that has been induced will shut down the activity (B) When TGF-β is applied to human of already-formed cyclin–CDK complexes—such as E–CDK2, A–CDK2, A–CDC2, and keratinocytes, it evokes a dramatic B–CDC2—that happened to be active when this damage was first incurred; once the 30-fold induction of p15INK4B mRNA damage is repaired, the p21Cip1-imposed block may then be relieved. Such a strategy synthesis as demonstrated here by RNA makes special sense in G1: if a cell’s genome becomes damaged during this period (Northern) blotting analysis. These cells through the actions of various mutagenic agents, p21Cip1 will block advance through were exposed to TGF-β for the time the R point (by inhibiting E–CDK2 complexes) until the damage has been repaired, periods (in hours) indicated (above). ensuring that the cell does not progress into S phase and inadvertently copy still-dam- (B, from G.J. Hannon and D. Beach, aged DNA sequences. In addition, p21Cip1 can inhibit the functions of a key compo- Nature 371:257–261, 1994.) nent of the cell’s DNA replication apparatus, termed PCNA (proliferating-cell nuclear antigen); this ensures that already-initiated DNA synthesis is halted until DNA repair has been completed. We will return to the mechanisms controlling p21Cip1 expression in the next chapter (Section 9.9). While DNA damage and, to a much lesser extent, TGF-β can elicit increases in the lev- els of p21Cip1 (thereby blocking cell cycle advance), mitogens act in an opposing fash- ion to mute the actions of this CDK inhibitor and in this way favor cell cycle advance. One mechanism by which they do so depends on the phosphatidylinositol 3-kinase (PI3K) pathway, which is activated directly or indirectly by the mitogens that stimu- late many receptor tyrosine kinases (Figure 8.15A). Akt/PKB, the important kinase activated downstream of mitogen-activated PI3K (see Section 6.6), phosphorylates p21Cip1 molecules located in the nucleus, thereby causing them to be exported into the cytoplasm, where they can no longer engage and inhibit cyclin–CDK complexes (see Figure 8.15B). Similarly, Akt/PKB phosphorylates p27Kip1 (which functions much like p21Cip1) and prevents its export from its cytoplasmic site of synthesis into the nucleus,  Mitogens antagonize CDK inhibitors 291 (A) mitogens Figure 8.15 Control of cell cycle plasma advance by extracellular signals membrane Countervailing extracellular signals influence the cell cycle machinery, in part through their ability to control the levels and intracellular localization of CDK inhibitors. (A) During the G1 phase, TGF-β induces p15INK4B and TGF-β PI3K (weakly) p21Cip1 expression, negatively affecting the D–CDK4/6 and E–CDK2 Akt/PKB translocation complexes, respectively. Conversely, to cytoplasm mitogens, acting through Akt/PKB, cause P the phosphorylation and cytoplasmic relatively localization of both p21Cip1 and p27Kip1, weak p27Kip1 which prevents these CDK inhibitors p15INK4B p21Kip1 from entering the nucleus and blocking the activities of various cyclin–CDK complexes operating there. (The growth factor receptors can also reduce the levels of the two CKIs depicted here, D CDK4/6 E–CDK2 A–CDK2 A–CDC2 B–CDC2 p21Cip1 and p27Kip1; these CKIs can inhibit, in turn, cyclin–CDK complexes (B) active in late G1, S, and G2. It is unclear, p21Kip1 DNA overlay however, whether these cyclin–CDK complexes are actually affected by mitogens in the S and G2 phases of constitutively active Akt/PKB the cell cycle.) (B) Ectopic expression of a constitutively active Akt/PKB (upper row) causes p21Cip1 (orange, left) to be localized largely in the cytoplasm. Compare its localization with that dominant- negative of the cell nuclei (blue, middle); the Akt/PKB overlap of these two images is also seen (right). Conversely, expression of a dominant-negative Akt/PKB (lower row), (C) T157A which interferes with ongoing Akt/PKB wild-type p27Kip1 mutant p27Kip1 function, allows p21Cip1 to localize to wild-type T157A + constit. active + constit. p27Kip1 mutant p27Kip1 Akt/PKB active Akt/PKB the cell nucleus, where it can interact with cyclin–CDK complexes and inhibit cell cycle progression. (C) In normal p27Cip1 cells (first panel), ectopically expressed wild-type p27Kip1 (green) is found to be exclusively nuclear. This localization is not changed if a p27Kip1 mutant is expressed that lacks the threonine residue normally phosphorylated by Akt/PKB (second panel). (The T157A mutant carries an where it normally does its critical work (see Figure 8.15C). Taken together, these sig- alanine in place of this threonine.) If a naling responses illustrate how extracellular growth-inhibitory signals (conveyed by mutant, constitutively active Akt/PKB is TGF-β) impede the advance of the cell cycle clock, while growth-promoting signals expressed, however, much of wild-type promote its forward progress. p27Kip1 is now seen in the cytoplasm (third panel). But if the p27Kip1 mutant These effects on intracellular localization appear to have clinical consequences. For that cannot be phosphorylated by Akt/ example, in low-grade (that is, less advanced) human mammary carcinomas, levels PKB is expressed, it resists the actions of activated Akt/PKB are low and p27Kip1 is able to carry out its anti-proliferative func- of constitutively activated Akt/PKB and tions in the cell nucleus. In high-grade tumors, however, activated Akt/PKB is abun- remains in the nucleus (fourth panel). dant and much of p27Kip1 is now found in the cell cytoplasm (Figure 8.16A). This intra-