Chapter 16: Cell-Cycle Regulation and Hematologic Disorders

Mitosis is the final step of a defined program—the cell cycle—that can be separated into four phases: the G₁, S, G₂, and M phases (Fig. 16–1). A number of surveillance systems (checkpoints) control the cell cycle and interrupt its progression when DNA damage occurs or when cells have failed to complete a necessary event.¹ These checkpoints have been given an empirical definition: When the occurrence of event B is dependent on the completion of prior event A, the dependence is a result of a checkpoint if a loss-of-function mutation can be found that relieves the dependence.¹ Three major cell-cycle checkpoints have been discovered: the DNA damage checkpoint, the replication checkpoint, and the spindle-pole body duplication checkpoint.^2,3,4 The functional consequence of failure to “satisfy” the requirements of a cell-cycle checkpoint is usually death by apoptosis. However, small numbers of genetically altered cells may survive. Cells with defective checkpoints have an advantage when selection favors multiple genetic changes. Cancer cells often are missing one or more checkpoints, which facilitates a greater rate of genomic evolution.⁵

A disturbance of cell-cycle regulation is an important pathway in the development of many hematologic malignancies as a result of mutations in tumor-suppressor genes or oncogenes. Until the end of the 20th century, it was believed that the only mechanism by which the “gatekeepers” of the cell cycle could be inactivated was deletion or mutation (gain-of-function or loss-of-function mutations). Progress in the understanding of the regulation of gene expression put emphasis on another mechanism of gene inactivation, called epigenetic regulation (Chap. 10). This term summarizes several molecular modifications, including histone deacetylation, CpG-island hypermethylation, ubiquitination, and phosphorylation, etc.

Table 16–1 lists Cdks, associated partners, and their functions.

Early experiments on the control of mitosis in human cells provided evidence for the existence of factors called M-phase and S-phase promoting factors.⁶ The key element of S-phase promoting factor was thought to be cell division cycle (cdc) 2. Experiments performed in Xenopus eggs showed that cdc2 is an M-phase–specific histone H1 kinase,⁷ but is just one subunit of a regulatory complex. A second component is cyclin B, which is synthesized in interphase and degraded in mid-mitosis. More than 10 members of the mammalian cyclin family have been identified. Most of these cyclins interact with a group of cdc2-related kinases called cyclin-dependent kinases (cdks),^8,9 while others are involved in alternate splicing processes.¹⁰ Phosphorylation of tyrosine 15 is the key event in regulating human cdc2 activity. Threonine 14 also is phosphorylated in G₂ phase. Dephosphorylation at both phosphorylation sites is required for mitotic initiation. Cdc2 interacts with cyclin B in mitosis, whereas the cdc2/cyclin A complex is formed before mitosis and is required for progression through late G₂ phase.¹¹ Thus, cyclins A and B are also called the mitotic cyclins, because they are upregulated in late G₂ or G₂/M phase and undergo proteolysis in M phase. The exit from mitosis is characterized by the abrupt ubiquitination and subsequent degradation of cyclin B. Cells with a defective cyclin B degradation mechanism or without mitotic cyclin B easily become aneuploid. There is evidence that cyclin A acts at the G₂/M transition and binds cdk2 in S phase. Cyclin A is mandatory for the downregulation of anaphase-promoting complex (APC).¹² Overexpression of cyclin A in G₁ phase leads to an accelerated entry into S phase.¹³ Because cdc2 is able to interact with mitotic and G₁ cyclins, it is likely that one protein kinase potentially can fulfill several different functions in the cell cycle at various checkpoints. Notably, there is increasing evidence that cdc2 is directly involved in regulating the DNA damage response (DDR), including DNA damage checkpoint activation and DNA repair (particularly homologous recombination [HR]).¹⁴ There are several cdc2-related protein kinases in humans that interact with the corresponding cyclins. Originally, three cdc2-related proteins were isolated, which were able to replace deficient cdc28 function in budding yeast: cdk1, cdk2, and cdk3.^15,16,17 Other cdks are termed cdk4,¹⁸ cdk5,¹⁹ and cdk6.²⁰ Cdk4/6 has been in the focus of tumor-suppressor gene research for the past several years, because it complexes with cyclin D (a G₁ cyclin). This complex is an important element in the p16^INK4A-retinoblastoma (RB) gene pathway, which is commonly disrupted in cancer. Interestingly, the neural cdk5 is also expressed at high levels in certain cancer types (e.g., myeloma), potentially representing a therapeutic target in these diseases.²¹ Moreover, cdk5 may serve as a prognostic marker and help to identify patients most likely to respond to treatment (e.g., with bortezomib).²² In addition, inhibition of cdk1 and/or cdk5 may contribute to disruption of the unfolded protein response (UPR) secondary to endoplasmic reticulum (ER) stress inducers.²³ Three transcription-regulatory cdks have been characterized. Cdk7 interacts with cyclin H and is responsible for phosphorylating cdks on threonine residues.²⁴ Cdk8 interacts with cyclin C, Med12, and Med13, and forms a complex called “cdk8” subcomplex. Several substrates for cdk8, when complexed with the above-mentioned proteins, have been detected, including RNA polymerase II and histone H3.²⁵ Cdk8 directly antagonizes the repression of β-catenin transcription by the transcription factor E2F1 in colorectal cancer.²⁶ The suppression of β-catenin by E2F1 contributes to apoptosis. Therefore, overexpression of cdk8 (and RB) accounts for a reduced rate of apoptosis and increased cell growth.²⁶ Cdk9 binds cyclin T and displays a tissue-specific expression pattern.²⁷ That cdk9/cyclin T specifically interacts with the tat element of HIV-1 links this cyclin-dependent kinase directly to the replication pathway of HIV, and circumstantially to HIV-1–related malignancies (e.g., Kaposi sarcoma).²⁸ Cdk10²⁹ and cdk11³⁰ define a novel class of cyclin-dependent kinases; both cdks interact with apoptosis-related factors³⁰ or transcription factors such as ets.³¹ Cdk10 has two isoforms with different functions. A role at the G₂/M transition has been suggested for the first isoform of cdk10, whereas the alternative splicing form interacts with the N-terminus of the Ets2 transcription factor. This interaction affects the G₂/M transition in mice.³² Two known cyclin-dependent kinases, cdk12 and cdk13, interact with both forms of cyclin L (cyclin L₁ and L₂). This complex seems to be involved in alterative RNA-splicing,^10,33 and thus is involved in the pre–messenger RNA processing machinery. All cyclins share an approximately 150-amino-acid region, called the cyclin box, which interacts with the cdks.³⁴ The G₁ cyclins (C, D, and E) and the mitotic cyclins (A and B)³⁵ form distinct categories, although cyclin H, cyclins L1 and L2, and the type T cyclins (T₁, T₂a, and T₂b) fall outside these two major groups.

Cyclin A binds and activates cdk2 mainly in S phase. However, microinjection of anti–cyclin A antibodies into cells causes cell cycle arrest just before S phase.¹¹ This observation, together with the finding that overexpression of cyclin A leads to accelerated S-phase entry, suggests that cyclin A is involved in transformation.¹³ Cyclin A is able to compensate for loss of cyclin E function. Cyclin E is important for the duplication of centrosomes. In cyclin E-defective cells, cyclin A can take over the function of cyclin E in S phase, whereas cyclin A is important for centrosome amplification in G₂-arrested cells, irrespective of whether cyclin E is present.³⁶ The importance of cyclin A in cell division is underlined by other reports.³⁷ In addition to its role at the G₁/S boundary, cyclin A acts in late G₂ phase, where it complexes with cdk1. Cyclin E, the other cyclin that interacts with cdk2, may control the progression from G₁ to S phase, and the time point when cdk2 “switches” from cyclin E to cyclin A binding is right after prereplication complex assembly terminates while DNA replication initiates. Cells overexpressing cyclin E progress much faster through G₁ into S phase, but the time required for DNA synthesis remains normal.³⁸ On the other hand, a bifurcation in cdk2 activity determines whether cells immediately commit to the next cell cycle or enter a transient state of quiescence as they exit mitosis.³⁹ Cyclin E levels are also regulated by environmental factors, including transforming growth factor-β (TGF-β) and irradiation. These effects are, in part, mediated by small proteins, the cyclin-dependent kinase inhibitors. Cyclin E accumulates at the G₁/S boundary of the cell cycle, where it stimulates functions associated with entry into and progression through S phase.⁴⁰ In normal cells, cyclin E levels are highly regulated so that peak cyclin E–cdk2 kinase activity occurs only for a short interval near the G₁/S boundary.⁴⁰ Cyclin E–cdk2 complexes become active during S phase and are then rapidly ubiquitinated after phosphorylation.⁴¹ The poor prognostic implications of overexpression of cyclin E has been observed in a variety of human malignancies,⁴² leading to a high cyclin E level throughout the cell cycle. The direct linkage between cyclin E overexpression and tumorigenesis is not completely understood. It has been suggested that the cyclin E–cdk2 complex phosphorylates and inactivates the RB protein⁴³ or leads to genomic instability via generation of aneuploid cells.⁴⁴ Cyclin E overexpression delays progression through early phases of mitosis and causes mitosis to be executed aberrantly, thus dysregulating mitotic progression.⁴⁵

The B-type cyclins associate with cdk1 to form the classical mitotic cyclin–cdk complexes.⁴⁶ Cyclin B is synthesized in S phase and accumulates, and in the midst of M phase is ubiquitinated and degraded, allowing the cell to exit from mitosis. The G₂/M checkpoint is very often defective in malignant cells, leading to uncontrolled M-phase entry and aneuploidy. The cellular localization of the cdk1–cyclin B complexes is strictly cell-cycle–dependent. Although the complexes accumulate in the cytoplasm during G₂ and S phase, they move to the nucleus in mitosis and bind to the mitotic spindle.^47,48 The cyclin B family has different family members with distinct functions. At mitotic entry, cyclin B₁–cdk1 promotes chromosome condensation, nuclear membrane dissolution, mitotic aster assembly, and Golgi breakdown, whereas cyclin B₂–cdk1 can only induce Golgi disassembly.⁴⁹ At prophase, cyclin B₁ accumulates in the nucleus⁵⁰ and then localizes to condensed chromatin, spindle microtubules, centrosomes, and chromatin during prometaphase.⁵¹ Distinct sequence elements are responsible for the localization of cyclin B₁ to the chromatin, centrosomes, and kinetochores during mitosis.⁵²

The three cyclin D molecules—D₁, D₂, and D₃—function mainly in G₁ phase, where they bind cdk4 and cdk6. These complexes phosphorylate RB, restraining its inhibitory effects on E2F and related transcription factors. Cyclin D₁ is the major D cyclin in most cell types. All three cyclin D molecules act in late G₁ phase, just before entry into S phase. Cyclin D₁ also exhibits a variety of non–cell-cycle regulatory functions. For example, cyclin D₁ regulates microRNA biogenesis by induction of Dicer, a central regulator of microRNA maturation.⁵³ Many tumors have high cyclin D₁ levels without amplification or mutation of the cyclin D₁ structural gene. Instead, cyclin D levels may be regulated by a feedback loop dependent on RB. Alterations of the RB gene in cancer may secondarily cause upregulation of cyclin D transcription. As a result of its central role in cell-cycle control, the cyclin D–cdk4 complex is an important target for anticancer drugs. Mice lacking cyclin D₁ are completely resistant to ErbB-2–driven breast cancer.⁵⁴ ErbB-2–induced mammary tumor development is also prohibited by the inactivation of the cyclin D₁ partner cdk4, underlining the role of this complex in human malignancies.⁵⁵ As aberrations of the p16–cdk4–cyclin D-RB pathway are common in the majority of cancers, the development of selective cdk4 inhibitors (e.g., palbociclib) launched promising efforts to target tumors displaying either cyclin D₁ overexpression (e.g., breast cancer, mantle cell lymphoma, multiple myeloma) or cdk4 amplification (e.g., liposarcoma).⁵⁶ Moreover, cyclin D₁–cdk4 is also involved in regulation of glucose metabolism in postmitotic cells, suggesting a novel cell-cycle–independent function of this complex.⁵⁷ However, cdk6, a functional homologue of cdk4, may also play an important role in tumorigenesis under certain circumstances. For example, acute myelogenous leukemia (AML) cells carrying mixed-lineage leukemia (MLL) rearrangements (e.g., MLL-AF9, MLL-AF4, and MLL-AF6) specifically rely on cdk6, rather than cdk4, to proliferate,⁵⁸ suggesting that cdk6 might represent a target in MLL-driven leukemia.⁵⁹ Interestingly, SUMOylation stabilizes the cdk6 protein, which may contribute to progression of some tumors (e.g., glioblastoma).⁶⁰ Notably, cyclin D–dependent cdk4/6 also phosphorylates a variety of substrates (e.g., RB1 and its relatives RBL1 and RBL2, SMAD2, SMAD3, FOXM1, MEP50, etc.), forming a central node in a complex signaling network that governs the overall transcriptional and biologic response of cells to activation or inhibition of the kinases.⁶¹

Cdk7 plays dual functions in regulation of cell cycle and gene transcription. In the former case, cdk7, together with its partner cyclin H, acts as a cdk-activating kinase, which fully activates various cell-cycle–regulatory cdks (e.g., cdk1, cdk2, cdk4, and cdk6) by phosphorylating their T loop in a context-specific manner.⁶² For example, whereas cdk7 is required to determine cyclin specificity and activation order of cdc1 and cdk2 during S and G₂ phases, it is also required to maintain the activity of cdk4 as cells exiting quiescence and G₁ progression through the restriction point.⁶³ Cdk7, as a component of the general transcription factor TFIIH (cdk7/cyclin H/Mat1 complex), phosphorylates the carboxy-terminal domain (CTD, serines 5 and 7) of RNA polymerase II, which is responsible for transcription initiation and promoter clearance, a critical step in switching initiation to elongation during transcription.⁶⁴ As transcription factors that co-opt the general transcriptional machinery to sustain the oncogenic state, pharmacologic inhibition of cdk7 may represent a novel approach to treat tumor types (e.g., T-cell acute lymphoblastic leukemia [T-ALL]) that are particularly dependent on transcription.⁶⁵

Cdk9, as a catalytic subunit, partners with the regulatory subunit cyclin T, an 87-kDa cyclin C–related protein with three isoforms, to form a complex known as positive transcription elongation factor b (P-TEFb).⁶⁶ The so-called cdk9-related pathway consists of two cdk9 isoforms (cdk9–42 and cdk9–55), cyclin T₁, cyclin T_2a, cyclin T_2b, and cyclin K.²⁷ Cdk9 and its binding partner cyclin T₁ comprise the P-TEFb.⁶⁷ P-TEFb hyperphosphorylates the CTD (primarily serine 2) of RNA polymerase II, essential for transcription elongation. P-TEFb also phosphorylates the negative transcription elongation factors (N-TEFs), including DRB-sensitivity inducing factor (DSIF) and the negative elongation factor (NELF), to release the transcription block (a pause immediately after transcription initiation) of both N-TEFs on hypophosphorylated forms of RNA polymerase II. In addition, cdk9 also plays a role in ribosomal RNA processing through activation of RNA polymerase II.⁶⁸ In normal cells, the activity of P-TEFb is stringently maintained in a functional equilibrium to accommodate transcriptional demands for different biologic activities.⁶⁹ As a rule in oncogenic transformation, upregulated antiapoptotic or prosurvival proteins in transformed cells must be sustained by constitutive RNA polymerase II activity that governs transcription elongation, in which cdk9 is the primary processivity factor.⁷⁰ In other words, transformed cells are addicted to transcription because of the requirement for continuous production of antiapoptotic proteins, particularly those with short half-lives. Of note, abnormal activities in the cdk9-related pathway occur in many human malignancies.⁷¹ For example, high levels of cdk9/cyclin T₁ expression are found in several types of hematologic malignancies, including B- and T-cell precursor-derived lymphomas, anaplastic large cell lymphoma, and follicular lymphomas, whereas strong nuclear staining for both proteins are observed in Hodgkin and Reed-Sternberg cells of classical Hodgkin lymphoma. In this context, selective cdk9 inhibitors preferentially target malignant cells in preclinical hematologic tumor models, including leukemia and multiple myeloma.^72,73 Mcl-1, the Bcl-2 family antiapoptotic protein with an estimated half-life of less than 3 hours,⁷⁴ represents one of the most common downstream targets for cdk9 inhibition.⁷⁵ Moreover, cdk9 inhibition disrupts the process of cytoprotective autophagy, for example, through downregulation of the adaptor protein SQSTM1/p62, resulting in an inefficient form of autophagy from cargo-loading failure, which, in turn, triggers apoptosis via upregulation of the BH3-only protein NBK/Bik.⁷⁶ Moreover, cdk inhibitors also induce upregulation of other BH3-only proteins such as Bim and Noxa.⁷⁷ Although it remains to be determined whether these events stem from inhibition of specific cdk(s), it is now clear that in addition to cell cycle-regulatory cdks, transcription-regulatory cdks, such as cdk9, represent another class of therapeutic targets. In addition, P-TEFb forms a complex with the HIV tat protein that binds the transactivation response element. The modification of RNA polymerase II by cdk9/cyclin T facilitates the efficient multiplication of the viral genome.⁷⁸ Other binding partners of cdk9 include tumor necrosis factor receptor-associated factor 2,⁷⁹ as well as inhibitory MAQ1 (or HEXIM1) and 7SK small nuclear RNA.⁸⁰ Furthermore, cdk9 is expressed throughout the cell cycle⁸¹ and is also involved in viral (HIV, herpes) replication.²⁷

Other members of the transcription-regulatory cdk subfamily include cdk8 and cdk12. Cdk8 is a subunit of the large Mediator complex (~1.2 MDa) composed of 25 to 30 proteins, which acts as a molecular bridge between DNA-binding transcription factors and RNA polymerase II. Cdk8 binds to cyclin C, MED12, and MED13 in the cyclin C–cdk8 module of the Mediator.⁸² The cdk8/cyclin C pair facilitates phosphorylation of both serine 2 and serine 5 at the CTD of RNA polymerase II.⁸³ Cdk8 can perform both positive and negative functions in transcriptional regulation during different transcription stages (e.g., preinitiation and elongation), which provide a mechanism to respond to different promoter contexts (e.g., transcription factors or cdk8 module binding).⁸² Unlike cdk7 and cdk9, which govern global gene expression, cdk8 promotes only gene-specific transcription. Recently, cdk8 expression has been detected in 70 percent colorectal cancers and correlated with β-catenin activation, suggesting that cdk8 may act as a oncogene in certain types of cancer (e.g., colorectal and pancreatic cancer).^84,85 Cdk12 and cdk13 have been identified as CTD kinases, both of which are unusually large proteins that contain a central kinase domain and share the same partner, cyclin K.⁸⁶ Cdk12/cyclin K₁ phosphorylates the CTD (preferably serine 2) of RNA polymerase II.⁸⁷ Interestingly, cdk12/cyclin K only regulates expression of a small subset of genes, predominantly long genes with high exon numbers and DDR genes, including critical regulators of genomic stability, for example, BRCA1, ATR (ataxia-telangiectasia mutated [ATM] and Rad3 related), FANCI, and FANCD2.⁸⁸

The cdk10 gene encodes two different cdk-like putative kinases; it is postulated that they exert their function at the G₂/M transition.²⁹ These two isoforms predominate in human tissues, except in brain and muscle, and the relative isoform levels do not vary during the cell cycle.²⁹ Cdk10 interacts with the N-terminus of the Ets2 transcription factor, which contains the highly conserved pointed transactivation domain. The pointed domain is implicated in protein–protein interactions and Ets2 requires an intact pointed domain to bind Cdk10, which inhibits Ets2 transactivation in mammalian cells.³¹ This could be an important factor for the development of follicular lymphoma, because cdk10 is overexpressed in this cancer.⁸⁹ In addition, cdk10 silencing increases Ets2-driven transcription of c-RAF, resulting in mitogen-activated protein kinase (MAPK) pathway activation and loss of tumor cell reliance upon estrogen signaling.⁹⁰ Cdk10 promoters are frequently hypermethylated in malignant tumors, resulting in low expression levels of cdk10 and impaired cell-cycle regulation.⁹⁰

Cdk11 is associated with cyclin L.⁹¹ It is part of the large family of p34(cdc2)-related kinases whose functions appear to be linked with cell-cycle progression, tumorigenesis, and apoptotic signaling. Cdk11 interacts with the p47 subunit of eukaryotic initiation factor 3 during apoptosis and is therefore directly involved in cell death mechanisms.⁹² Casein kinase 2 phosphorylates the cdk11 aminoterminal domain, suggesting that cdk11 participates in signaling pathways that include casein kinase 2 and that its function may help to coordinate the regulation of RNA transcription and processing events.⁹¹ So far two isoforms of cdk11 have been identified, a larger p110 and a smaller p46 isoform. During Fas- or tumor necrosis factor-α–induced apoptosis, the caspase-processed p46 isoform is generated from the larger p110 isoform and it promotes apoptosis when it is ectopically expressed in human cells. Cdk11 also stabilizes the microtubule assembly of cells⁹³; cdk11 is therefore mandatory for the maintenance of sister chromatid cohesion⁹⁴ and its disruption can contribute to the development of cancer.⁹⁵

SUBSTRATES AND INHIBITORS OF CYCLIN-DEPENDENT KINASES

Many cyclin–cdk substrates have been identified by immunoprecipitation or two-hybrid assays, but only a few of them are thought to exert a direct function in cell-cycle control. The regulation of the cell cycle has been studied extensively during the last decade and a consensus paradigm of cell-cycle regulation has been suggested.^50,96 According to this paradigm, the important switch of the cell cycle is the RB family of proteins (Fig. 16–2). In its hypophosphorylated state, RB binds to and inhibits a class of transcription factors, of which the best characterized is the E2F transcription factor. Hyperphosphorylation causes RB to detach from its binding site, permitting transcriptional activation of genes necessary for DNA synthesis and cell division. This phosphorylation of RB is regulated in a cell-cycle–dependent manner.⁹⁷ A widely accepted model suggests that RB is phosphorylated by different regulators such as cyclin E/cdk2 at the so-called “R” point, a time point during G₁ when cell cycle progression becomes independent of exogenous stimuli.⁹⁸ Interference with RB function impairs G₁ checkpoint regulation and fosters unrestrained cell growth, a nearly universal characteristic of malignancy. RB controls the activity of several other cell-cycle regulatory elements such as Skp2.⁹⁹ The Skp2 regulation follows an autocrine loop where Skp2 triggers degradation of the cdk inhibitor p27 ^kip1, followed by cyclin E/cdk2 activation, consecutive cdk2-induced RB-phosphorylation, and further E2F-dependent Skp2 expression.¹⁰⁰ Causes of reduced RB activity include changes in the structural gene, the sequestration and inactivation of the protein by viral oncogene products, and hyperphosphorylation of RB as a result of increased cdk4 and cyclin D activity or deletion of the gene for the p16^INK4A inhibitor of cdk4. Deletions, mutations, and translocations of RB are common in various malignancies, while homozygous deletions of the p16^INK4A gene are even more frequent. Many different transforming viruses (papillomavirus, simian virus 40) produce proteins that interact with RB. Both cyclin D₁–cdk4 and cyclin D₁(D₂, D₃)–cdk6 complexes are able to phosphorylate RB.^101,102 The time point of RB phosphorylation correlates strongly with the appearance of the cyclin D₁–cdk4 complex.¹⁰³ The link between RB and cyclin D is supported by the observation that loss of RB function leads to a decrease in the cellular cyclin D level.¹⁰⁴ However, cyclin D is not the only cyclin that is involved in the RB regulatory pathway.^99,102 Ectopic expression of both cyclin A and cyclin E restores RB hyperphosphorylation and causes cell-cycle arrest in cancer cell lines. Perhaps the cdk2–cyclin A complex contributes to additional phosphorylation of RB, whereas the cdk2–cyclin E complex prolongs the phosphorylation time.¹⁰⁵

The key regulatory element for the G₁-to-S transition is the RB–E2F complex. After RB is phosphorylated by cdk4 and/or cdk6 complexes during G₁ phase and cdk2 at G₁/S interphase, E2F proteins are released and promote the transcription of genes essential for the transition to S phase.^99,106 As mentioned above, the p16^INK4A/cyclin D₁/cdk4/RB/E2F cascade is probably one of the most important cascades in cell-cycle control, and is frequently affected in human cancer. For example, this pathway is defective in nearly 100 percent of AML cell lines and most of the primary AML samples, although the exact mechanism of inactivation is not always clear. Two RB-related pocket proteins, p107 and p130, also form complexes with the transcription factor E2F,¹⁰⁷ bind to the region of the adenovirus E1A protein required for transformation, and are able to induce G₁ arrest when they are overexpressed in human malignant cell lines.^108,109 Unlike RB, the p107 and p130 proteins contain a so-called spacer region that interacts with cdk2/cyclin A and cdk2/cyclin E,¹¹⁰ although it seems to be unlikely that these two complexes regulate the activity of p107 and p130.¹⁰⁵ Instead, p107 may bind and inactivate the cyclin A and cyclin E complexes. Thus, p107 may regulate the cell cycle by several different mechanisms. Because both p107 and p130 are regulated through phosphorylation, efficient cell-cycle entry is accompanied by phosphorylation of all the RB-related proteins.¹¹¹

In addition to its cell-cycle regulatory properties, RB also influences hematopoietic differentiation.¹¹² RB interacts with the transcription factor PU.1, which blocks erythroid differentiation in the proerythroblast stage when ectopically overexpressed in marrow cells,^113,114 and represses GATA-1 activity.¹¹⁵ An important event in this differentiation process is the interaction between hematopoietic stem cells and the microenvironment of the marrow. In addition, hypophosphorylated RB promotes monocytic over neutrophilic differentiation in bipotent progenitor cells, an event that is switched to neutrophilic differentiation if RB expression is inhibited. This finding points to an important property of RB independent of cell-cycle control.¹¹⁶

Besides regulation by phosphorylation, specific protein inhibitors of cdk enzymatic activity have been identified.¹¹⁷ The cyclin-dependent kinase inhibitors (CDKIs) cause cells to arrest in G₁ phase, followed by differentiation and/or senescence. The first CDKI identified was p21^cip1.¹¹⁸ It binds to several cyclin/cdk complexes, including cyclin A/cdk2, cyclin D/cdk4, and cyclin E/cdk2 (see Fig. 16–2).^97,119 Several different cell-cycle regulatory pathways are affected by p21^cip1. In addition, a basal p21^cip1–cdk2 axis determines quiescent and cycling cell states and thus controls population heterogeneity in both normal cells and tumors, which can make anticancer treatment selectivity challenging.¹²⁰ The molecule has a p53 binding site in its promoter, and an increase in p53 levels results in transcriptional activation of p21^cip1, slowing cell-cycle progression. In addition to this p53-dependent pathway, p21^cip1 is also regulated in a p53-independent manner. For example, histone deacetylase inhibitors (HDACIs) are able to induce p21^cip1 expression in p53^null leukemia cells through an alternative nuclear factor (NF)-κB–dependent mechanism, which may limit the antileukemia activity of these agents.^121,122 Several binding partners of p21^cip1, including Pim-1, have been identified. Pim-1 associates with and phosphorylates p21^cip1 in vivo, which influences the subcellular localization of p21^cip1.¹²³ p21^cip1 is phosphorylated by Pim-1 at two distinct sites, Thr145 and Ser146; phosphorylation on Thr¹⁴⁵ results in a nuclear localization of p21^cip1 and a disruption of the cell cycle, while phosphorylation on Ser¹⁴⁶ leads to a cytoplasmic localization of p21^cip1,124 suggesting that overexpression of Pim-1 in certain tumors plays a key role in tumorigenesis.¹²⁵ Like RB, expression and function of p21^cip1 is also affected by several different mechanisms, including mutation and histone deacetylation (see “The Role of Histone Deacetylases in Cell-Cycle Regulation” below). Other members of the p21^cip1 family of CDKIs include p27^kip1 and p57^kip2.^101,126 As a cdk inhibitor, p27^kip1 has tumor-suppressor activity. Besides cdks, p27^kip1 regulates additional cellular processes, including cell motility, some of which seem to mediate the oncogenic activities of p27^kip1. For example, the constellation of high p27^kip1 and low Myc expression is characteristic of chronic lymphocytic leukemia (CLL) cells.¹²⁷ These activities of p27^kip1 are regulated through multiple phosphorylation sites. The multiple functions of p27^kip1 are dependent on a number of different conditions, and dictate whether the protein displays anti- or protumorigenic properties.¹²⁸ High-level expression of p27^kip1 leads to a cell-cycle block in G₁ phase after treatment of cells with TGF-β. One major difference between p21^cip1 and p27^kip1 is that the former binds predominantly to cdk2 whereas the latter binds cdk4.

The cellular levels of a number of cell-cycle regulators, including p21^cip1 and p27^kip1, are regulated by ubiquitination and subsequent proteolysis. Polyubiquitinated proteins are degraded by the 26S proteasome complex. There are two major ubiquitination systems in the cell; they are designated SCF and APC.^108,110 SCF is named for three of its core components, Skp, Cullin, and an F-box–containing protein. Important examples of SCF substrates are Cln1, Sic1, Wee1, Cdc6/Cdc18, E2F, cyclin D₁, cyclin E, p21^cip1, p27^kip1, and p57^kip2.¹²⁹

A second group of CDKIs belongs to the inhibitor of the kinase 4 (INK4) family and includes p15^INK4B, p16^INK4A, p18^INK4C, and p19^INK4D.^{104,107,130,131} They all bind and inhibit the cyclin D₁–cdk4 and/or cyclin D₁–cdk6 complex, which regulate cell-cycle progression via RB.^104,130 TGF-β is also a potent inducer of p15^INK4B,¹⁰⁴ one of the mechanisms by which the cytokine regulates the proliferation of hematopoietic cells (Chap. 16). p16^INK4A is probably the most important CDKI, because the gene is inactivated by several mechanisms (deletion, mutation, hypermethylation) in many different human cancers.¹³² Surprisingly, p16^INK4A and p14^ARF are overexpressed in some cases of human hematologic malignancies.¹³³ This overexpression is probably a result of defects downstream of p16^INK4A, particularly caused by mutations in the RB gene.¹³⁴ In hematologic malignancies, the highest frequencies of p14^ARF, p15^INK4B, or p16^INK4A inactivations are found in T-ALL,¹³⁵ secondary high-grade lymphomas, and mantle cell lymphoma (MCL).^136,137 The potency of p14^ARF and p16^INK4A in terms of tumorigenicity becomes obvious because the reexpression of both genes by either retroviral transfection or demethylation of the promoter regions results in a complete reversion of the malignant phenotype.^138,139 p14^ARF has multiple tumor-suppressor functions, some of which are mediated by signaling to p53. On the other hand, it has been shown that p14^ARF is able to drive tumor progression in a p53-independent fashion, especially in myc-driven lymphomas.¹⁴⁰

Perturbations of the cell cycle, for example, overexpression of cyclins or underexpression of endogenous CDKIs, are nearly universal in human malignancies, and cdks, as critical regulators of cell-cycle progression and RNA transcription, represent attractive targets for anticancer drug development, as their inhibition can lead to both cell-cycle arrest and apoptosis.^141,142 Additionally, pharmacologic inhibition of transcriptional cdks, for example, cdk9, affects proteins with short half-lives, for example, the antiapoptotic proteins Mcl-1 and XIAP (X-linked inhibitor of apoptosis), cell-cycle regulators, p53, and NF-κB–responsive gene products.¹⁴² Hematologic malignancies may be particularly susceptible to cdk inhibition and apoptosis induction. A number of pharmacologic CDKIs have been developed and subjected to various phases of preclinical and clinical testing in hematologic malignancies (for a review, see Ref. 143). The selective cdk4/6 inhibitor palbociclib has been approved by the FDA as a potential treatment for hormone-responsive metastatic breast cancer. In MCL, a disease characterized by cyclin D₁ overexpression, this agent has demonstrated encouraging clinical activity.¹⁴⁴ Whether selective or broad-spectrum inhibition is the superior therapeutic strategy continues to be debated,¹⁴³ and may be tumor type-specific. The “pan-CDKI” flavopiridol (alvocidib), the first CDKI to enter the clinic, has shown promising activity in patients with genetically high-risk chronic lymphocytic leukemia (CLL), particularly when administered by a pharmacologically derived “hybrid” schedule,^145,146 but recent developmental efforts for this agent have focused on AML, where it was recently granted “orphan drug” status.¹⁴⁷ Flavopiridol induces apoptosis in primary leukemic blasts and recruits surviving leukemic cells into a proliferative state, thereby priming such cells for killing by S-phase-active cytotoxic agents such as cytarabine.^148,149 These observations led to the design of “timed sequential therapy” regimens, for example, FLAM (flavopiridol, cytarabine, mitoxantrone), which exhibited promising clinical activity in AML patients with nonfavorable cytogenetics.¹⁵⁰ Interestingly, both “bolus” and “hybrid” schedules of administration of flavopiridol produce comparably encouraging results in this setting.¹⁵¹ Other rational combinations involving pan-CDKIs such as flavopiridol or roscovitine involve those with proteasome inhibitors (PIs),^152,153 HDACIs¹⁵⁴ or BH3-mimetics.⁷⁵ The first strategy has been explored in phase I trials in patients with recurrent or refractory indolent B-cell neoplasms,^155,156 and a phase I trial of bortezomib and dexamethasone plus the pan-CDKI dinaciclib in patients with relapsed myeloma is ongoing as of this writing (NCT01711528). Synergism in preclinical studies in leukemia between pan-CDKIs and HDACIs is based predominantly on reciprocal effects on the cytoprotective NF-κB pathway and the endogenous cdk inhibitor p21^WAF1/CIP1, besides Mcl-1/XIAP downregulation.¹⁵⁴ Some of these phenomena have been recapitulated in patients with AML.¹⁵⁷

The cell cycle progresses in an orderly fashion and is monitored by safety mechanisms known as cell-cycle checkpoints, which, upon activation, function to halt cell division.¹⁵⁸ When DNA damage occurs, distinct, albeit overlapping and cooperating, checkpoint pathways are activated, which block S-phase entry (the G₁/S-phase checkpoint), delay S-phase progression (the intra–S- or S-phase checkpoint), or prevent mitotic entry (the G₂/M-phase checkpoint). These events direct phase-specific DNA repair mechanisms through repair-specific gene transcription. If repair fails, checkpoints trigger apoptosis.¹⁵⁹ Checkpoints are thus important quality control measures that ensure the proper sequence of cell-cycle events and allow cells to respond to DNA damage.¹⁶⁰

The G₁/S checkpoint is the first defense against genomic stress in cycling cells. In response to DNA damage, the G₁/S checkpoint prevents cells from entering the S-phase by inhibiting the initiation of DNA replication. At this checkpoint, the checkpoint kinase Chk2 is activated by the proximal transducer ATM to phosphorylate (and thereby inhibit) the cdc25A phosphatase, thus preventing activation of cyclin E(A)/cdk2 and temporarily halting the cell cycle. G₁ arrest is sustained by ATM/Chk2-mediated phosphorylation of murine double minute protein 2 (MDM2) and p53, resulting in p53 stabilization and accumulation. p53 transcriptionally activates the endogenous cdk inhibitor p21, which, in turn, inhibits cyclin E(A)/cdk2 and preserves the association of Rb with E2F.¹⁶⁰

The intra-S or S-phase checkpoint is activated in response to structural DNA damage as well as stalled replication forks. Upon activation of this checkpoint, ATR/Chk1 and ATM/Chk2 phosphorylate cdc25A, resulting in enhanced proteolysis of the phosphatase and inhibition of its function through 14-3-3 σ binding. Cyclin E(A)/cdk2 is thereby inhibited, and progression through the S-phase halted.¹⁵⁸

The G₂/M checkpoint prevents mitotic entry of cells that have either incurred DNA damage during G₂, or that have escaped the G₁/S and intra-S checkpoints despite earlier genomic insults. The key downstream target of the G₂/M checkpoint is the promitotic cyclin B/cdk1 (cdc2) complex. During interphase, this complex is inactivated through phosphorylation by Myt1 and Wee1.¹⁵⁸ Chk1 may phosphorylate Wee1, mediating binding of Wee1 to 14-3-3 proteins, which, in turn, may stimulate the kinase activity of Wee1 against cdk1 (cdc2). Thus, both Chk1 and 14-3-3 proteins may act together as positive regulators of Wee1.¹⁶¹ Activation of cyclin B/cdk1 (cdc2) requires dephosphorylation by the cdc25 phosphatases (A, B, and C). Notably, phosphorylation/inactivation of cdk1 (cdc2) involves two inhibitory sites, for example, Tyr15 and Thr14, and dephosphorylation of both sites is necessary for full cdk1 (cdc2) activation. Initiation of the G₂/M checkpoint is mediated by ATR/Chk1, which phosphorylates (and thereby inhibits) cdc25A, B, and C, whereas maintenance of this checkpoint requires p53 and its downstream effectors p21, 14-3-3 σ, and growth arrest and DNA damage (GADD) 45.¹⁵⁸

Checkpoint dysfunction is common in human cancers and is considered a pathologic hallmark of neoplastic transformation.¹⁶² Conversely, agents used for cancer treatment, such as cytotoxic chemotherapy and ionizing radiation, activate cell-cycle checkpoints.¹⁵⁸ Cancer cells are particularly dependent upon the S- and G₂/M-phase checkpoints for repair of DNA damage because of preexisting defects in G₁/S checkpoint mechanisms, such as p53 and RB mutations. Because the S-phase checkpoint facilitates slowing, rather than arrest, of the cell cycle, a cancer cell harboring DNA damage may progress through the S-phase checkpoint, only to halt at the G₂/M checkpoint. The latter is, therefore, a key guardian of the cancer cell genome, and its abrogation can lead to enhanced tumor cell death while sparing normal cells, which maintain an intact G₁/S-phase checkpoint. G₂/M checkpoint abrogation prevents cancer cells from repairing DNA damage, forcing them into a premature and lethal mitosis (“mitotic catastrophe”).¹⁶⁰

Based upon these concepts, synergism between DNA-damaging agents (cytotoxic chemotherapy) and G₂/M checkpoint abrogators have been examined both preclinically and in clinical studies. In human AML cell lines, the Chk1 inhibitor MK-8776 (SCH900776) markedly increased cytarabine-induced apoptosis, with negligible impact on normal myeloid progenitors,¹⁶³ leading to a phase I trial of the combination in patients with relapsed and refractory acute leukemias.¹⁶⁴ Chk1 inhibitors increase HDACI lethality in human leukemia cells.¹⁶⁵ This may reflect the ability of HDACIs to induce DNA damage in AML cells,¹⁶⁶ inhibit both homologous recombination and nonhomologous end-joining mechanisms of DNA repair,¹⁶⁷ and downregulate/inactivate Chk1.¹⁶⁸ However, clinical development of MK-8776 has been halted. Based upon the ability of Hsp90 inhibitors to downregulate Chk1, a phase I study of the combination of cytarabine and the Hsp90 inhibitor tanespimycin was conducted in adults with recurrent or refractory acute leukemia; however, the combination exhibited limited clinical activity.¹⁶⁹

Wee1 is a validated therapeutic target in AML.¹⁷⁰ The combination of cytarabine and the Wee1 inhibitor AZD1775 synergistically induces apoptosis in myeloid cell lines.¹⁷¹ Similarly, genome-wide short hairpin RNA screens strongly implicate cell-cycle checkpoint proteins, particularly Wee1, as critical mediators of AML cell survival after cytarabine exposure.¹⁷² In AML cell lines, synergistic inhibition of proliferation by the combination of AZD1775 and Ara-C occurred regardless of p53 functionality.¹⁷³ Another combinatorial strategy in AML involves coadministration of AZD1775 with HDACIs.¹⁷⁴ Besides inducing DNA damage and inhibiting DNA repair, HDACIs downregulate several proteins critical to checkpoint function, such as ATR, Chk1, and Wee1,^{168,175,176,177} and Wee1 inhibition promotes premature mitotic entry of cells bearing unrepaired DNA damage.¹⁷⁸ Efforts to extend these pre-clinical findings to the clinic are underway.

Table 16–2 lists the common genomic aberrations seen in hematologic malignancies. Tables 16–3 and 16–4 list common somatic mutations encountered in the major myeloid and lymphoid malignancies, respectively.

The complicated cell-cycle network has its parallel in the several different oncogenes and tumor-suppressor genes that influence carcinogenesis and tumor progression. The products of oncogenes, the oncoproteins, lead to or facilitate the transformation of a normal into a malignant cell. Oncogenes can be carried into the cell by viruses or they can arise from mutations in normal cellular genes. In addition, they can also arise from leukemia- or lymphoma-associated translocations where two usually separated genes are fused together and form a novel fusion protein. The familiar concept of this kind of protooncogene activation can be blurred by the fusion proteins because they possess unique capabilities not shared by either of the individual fusion partners. Oncoproteins can interact directly with cell-cycle regulatory proteins or control their activity through phosphorylation and dephosphorylation. Not all mutations in oncogenes lead to an altered function of the resulting product. The nomenclature in the oncogene and tumor-suppressor gene field is not always clear. As a general guideline, if a mutation causes a functional loss of the gene product (loss of function), and the recessive loss of function leads directly to uncontrolled cell division, the underlying gene can be named a tumor-suppressor gene. On the other hand, if the mutation leads to an altered gene product (gain of function) that interacts abnormally with other proteins to influence the cell cycle, this gene is an oncogene, acting in a dominant fashion. Translocations are typical of oncogenes, whereas homozygous deletions and hypermethylation of CpG-nucleotide repeats in the promoter regions (“epigenetic silencing”) are characteristic features of tumor-suppressor genes.

Numerous oncogenes/oncogene candidates are described in the literature, and most of them are involved in the pathogenesis and development of many different tumors types, especially the hematologic malignancies. Among the chromosomal translocations, some of the most well-studied are found in AML and other myeloid neoplasms. These include t(8;21)(q22;q22), leading to the AML1-ETO or RUNX1-RUNX1T1 rearrangement, del4(q12;q12), t(5;12)(q31-q32;p13), leading to the TEL-platelet derived growth factor receptor (PDGFR) β rearrangement, t(15;17)(q22;12), leading to the promyelocytic leukemia (PML)-retinoic acid receptor (RAR) α rearrangement, inv16(p13;q22) or t(16;16)(p13;q22), leading to the CBFβ-MYH11 rearrangement, t(9;22)(q34;q11), leading to the BCR-ABL rearrangement, t(3;3)(q21;q26), t(8;16)(p11;p13), t(6;9)(p23;q34), t(7;11)(p15;p15), t(9;11)(p22;q23), t(6;11)(q27;q23), t(11;19)(q23;p13.1), and t(11;19)(q23;p13.3), all of which translocate the MLL gene located at 11q23, t(16;21)(p11;q22), and t(1;22) (p13;q13).^179,180 Leukemias carrying MLL rearrangements are driven by dysregulated epigenetic mechanisms in which fusion proteins containing N-terminal sequences of MLL can cause human leukemia without the requirement for a “second hit.”¹⁸¹ MLL-rearranged leukemias provide a paradigm for how epigenetic dysregulation can lead to cancer through inappropriate chromatin structure with subsequent activation of target genes with oncogenic activity. An improved molecular understanding of how MLL fusions upregulate binding targets has led to the identification of a number of potential mechanism-based therapeutic vulnerabilities for this poor-prognosis malignancy. Potential novel therapeutic approaches include inhibition of P-TEFb (cdk9/cyclin T), the histone modifying enzymes DOT1L (methyltransferase) or TIP60 (acetyltransferase), and disruption of the interaction of MLL fusions with other epigenetic systems such as CpG island methylation and polycomb genes, for example, PRC2.¹⁸¹ In contrast, in secondary myeloid leukemias, recurrent numerical and unbalanced cytogenetic abnormalities predominate such as del(5q), del(7q), 7/del(7q), and del(20q), and are often associated with a poor prognosis.^180,182 Table 16–2 lists some of the fusion partners. Apart from the chromosomal translocations in AML as described above, there also aberrant fusion proteins in acute lymphoid leukemia (ALL), such as t(9;22), which is also found in chronic myelogenous leukemia (CML), t(4;11) in prolymphoblastic leukemia, and t(12;21) in childhood ALL. Some lymphomas are characterized by chromosomal translocations that juxtapose an oncogene to the immunoglobulin heavy-chain gene, which then drives overexpression of the oncogene, for example, t(8;14) in Burkitt lymphoma, t(11;14) in MCL, or t(14;18) in follicular lymphoma; for a review see Ref. 125.¹⁸³ The same is true of myeloma, where abnormalities such as t(4;14), t(6;14), t(11;14), t(14;16), and t(14;20) are frequently seen. An excellent overview of chromosomal rearrangements in cancer and the affected genes is available.¹⁸⁰

The precise mechanism by which the fusion proteins lead to tumorigenesis is not always well understood. Nevertheless, in patients with AML, abnormal expression of the transcription factor RUNX1 (AML1) is able to promote cell-cycle progression by shortening G₁ phase and by repressing p21^cip1 promoter activity. RUNX1 is absolutely required for the establishment of adult-type hematopoiesis¹⁸⁴; it regulates genes specific to the lymphoid, myeloid, and megakaryocyte lineages,¹⁸⁵ and mice lacking RUNX1 do not develop definitive hematopoiesis, indicating a role in adult hematopoietic stem cell formation.¹⁸⁶ In contrast, the fusion product AML1/ETO, derived from the t(8;21), slows cell-cycle progression, suggesting that one gene in different “fusion situations” can cause different effects on the cell cycle.¹⁸⁷ Activation of the RUNX1-repression domain or fusing the gene to ETO results in downregulation of cdk4 and Myc, directly linking this fusion protein to cell-cycle checkpoints.¹⁸⁷ Additional evidence for the direct involvement of RUNX1 in cell-cycle control comes from the observation that the transcription factor binds to the p19^INK4D promoter and downregulates p19^INK4D expression in megakaryocytes.¹⁸⁸ Inhibiting the oligomerization domain of ETO interferes with RUNX1/ETO oncogenic activity and these cells lose their progenitor cell characteristics, arrest cell-cycle progression, and undergo cell death.¹⁸⁹ Another interesting chromosomal translocation fusion product that affects cell-cycle control is found in patients with acute PML (APL) or its variant form (vAPL). The PML-RARα, which results from t(15;17)(q22;12), upregulates cyclin A₁ expression, whereas PML itself seems to be a negative regulator of cell growth because its overexpression leads to growth suppression and G₁ arrest in a variety of different cell types.¹⁹⁰ PML is crucial for the growth-inhibiting activity of retinoic acid and its absence abrogates the retinoic acid-dependent transactivation of p21^cip1.¹⁹¹ Another mechanism by which PML elicits irreversible growth arrest is believed to involve activation of the tumor-suppressor pathway p16^INK4A/RB.¹⁹² Recent data point toward a linkage between PML and the nucleoporins, especially Nup98 and Nup214. In some AML specimens, these nucleoporins are expressed as oncogenic fusion proteins and become directed—complexed with PML—to common cytoplasmic compartments during the M-to-G₁ transition of the cell cycle. In APL cells, the loss of function of normal PML causes an increase in cytoplasmic-bound versus nuclear-membrane-bound nucleoporins.¹⁹³ Consequently, PML by itself is a tumor-suppressor gene that positively regulates cell-cycle progression. Further evidence for a tumor-suppressor gene function of PML comes from transgenic mice models where PML^–/– mouse embryonic fibroblasts are enriched in S phase and the G₀/G₁ phase is minimized.¹⁹⁴ In APL, this regulatory role is disrupted by the fusion to RARα. One mechanism by which this fusion protein (and also the PML Kruppel-like zinc finger [PLZF]-RARα fusion derived from the rare t[11;17]) affects cell-cycle control is its strong interaction with SMRT or N-CoR, two corepressor elements that are important for the recruitment of HDACs, as described below in “The Role of Histone Deacetylases in Cell-Cycle Regulation.” In accordance with this is the finding that retrovirally transduced PML-RARα induces a maturation arrest in the corresponding cells, implying that these cells are unable to express certain transcription factors as a consequence of the conformational changes caused by the recruitment of HDACs.¹⁹⁵ A variant of this chromosomal translocation results in a fusion protein between RARα and the PLZF protein, which is observed in a subset of patients with APL.^195,196

The translocation t(9:22), which fuses the BCR gene to the c-ABL gene, is a characteristic feature of CML (Chap. 88). The chromosome 9 breakpoints, where the c-abl gene is located, involve a large region of about 200 kb, but fusion genes invariably include the abl exon 2. The corresponding breakpoints on chromosome 22 are located in a much smaller region, including the BCR gene.¹⁹⁷ The bcr-abl fusion protein localizes to the cytoskeleton and displays enhanced tyrosine kinase activity.¹⁹⁸ It is also found in some cases of ALL and in occasional cases of AML.^199,200 Bcr-abl not only regulates cell proliferation, apoptosis, differentiation, and adhesion, but also induces resistance to cytostatic drugs by modulation of DNA repair mechanisms, cell-cycle checkpoints, and the Bcl-2 family of apoptosis regulators. Upon DNA damage, bcr-abl enhances repair of DNA lesions and prolongs activation of cell-cycle checkpoints (e.g., G₂/M), providing more time for repair of otherwise lethal lesions, so that these cells have a significant survival advantage.¹⁹⁹ The bcr-abl fusion product is so far the only oncogenic product that is sufficient to induce malignant growth in vivo without the presence of other abnormal molecular changes. Several reports have shown that bcr-abl–positive cells display pronounced G₂/M delay in response to various chemotherapeutics and irradiation. The exact mechanism of G₂/M delay in bcr-abl–positive cells has not been characterized in detail, but it seems that the cdc2-cyclin B₁ regulation is affected. In addition, bcr-abl, through both kinase-dependent and kinase-independent mechanisms, converts p27^kip1 from a nuclear tumor suppressor to a cytoplasmic oncogene, which may contribute to bcr-abl tyrosine kinase inhibitor (TKI)-resistance.²⁰¹ The bcr-abl signal transduction process involves adapter molecules such as GRB2 and GAB2, as well as signaling pathways (e.g., phosphatidylinositol 3′-kinase [PI3K], Janus-associated kinase [JAK]–signal transducer and activator of transcription [STAT]).¹⁹⁸ In addition, although there is no direct evidence that the abnormal bcr-abl product affects the M checkpoint itself, some data suggest that bcr-abl–positive CML cells contain elevated MAD2 and BUB1 levels, proteins that inhibit the APC and therefore cause mitotic spindle arrest.²⁰² Amplification of the fusion sequence is frequently used to detect minimal residual disease in patients under therapy with interferon-α, TKIs,²⁰³ and after stem cell transplantation.²⁰⁴ The etv6 gene is the only known non-bcr fusion partner of abl, sometimes observed as etv6-abl in ALL or myeloproliferative syndromes (t[9;12][q34;p13]).²⁰⁵ The affected cells show only a minor response to imatinib.

Mutant-activated receptor protein-tyrosine kinases (rPTKs) comprise a family of very-well-characterized oncogenes. The constitutive activation of rPTK usually is achieved by mutations that lead to the dimerization and activation of their cytoplasmic catalytic domains.²⁰⁶ Other possible causes of rPTK dimerization are chromosomal translocations that create chimeric proteins. In the t(2;5) translocation, found in several anaplastic large cell lymphomas, N-terminal nucleophosmin sequences on the long arm of chromosome 5 are fused to the cytoplasmic domain of the ALK protein on chromosome 2.^207,208 The characteristic translocation of chronic myelomonocytic leukemia, t(5;12), fuses sequences from the transcription factor TEL to the cytoplasmic domain of the PDGFRβ (TEL-PDGFβR), resulting in the formation of a TEL-PDGFβR fusion protein and constitutive activation of the receptor tyrosine kinase (RTK),²⁰⁹ while targeting Id1 (inhibitor of DNA-binding 1) inhibits growth of leukemia cells expressing oncogenic FLT3-ITD and BCR-ABL tyrosine kinases.²¹⁰ Patients with the t(5;12) translocation respond to imatinib, as the drug also inhibits the PDGFR. The chromosomal area surrounding the TEL gene is a fragile site, because the TEL gene is involved in several other translocations in human acute leukemias (e.g., t[12;9]). One of the TGF-β receptors also is involved in oncogenesis, and mutations are frequently found in colon cancer. TGF-β receptor signaling acts through the SMAD family of transcription factors.

Two important oncogene families encode the Ras and Rho family proteins. Ras itself is a G protein, and activating mutations in H-Ras, K-Ras, and N-Ras have been found in nearly all kinds of human cancers. Several different Ras mutations are able to transform normal cells in tissue culture.^211,212 Mutations in many different Ras-related pathways have been identified in cancer (e.g., Raf1, p110 PI3K, Rin1, Mekk1), and numerous downstream signaling effects of those mutations have also been defined. The Ras and the Rho families of oncoproteins are linked by a small G protein called Rac, which is required for transformation by Ras.^213,214 The normal formation of actin filaments is required for G₁/S-phase entry. Recent data have shown the Rho-guanosine triphosphatases play a key role in the Wnt-signaling pathway, where they are involved in cellular polarization processes.²¹⁵ Thus, alterations in the Rho pathway may lead to premature entry into M phase by interference with cytoskeletal organization. The Ras/Raf/MEK/ERK cascade couples signals from the surface to the intracellular space and triggers cell proliferation signals that influence the cell cycle. Abnormal activation of this cascade occurs in several leukemias because of activating mutations in the Ras protooncogene.²¹⁶ Ectopic overexpression of Raf proteins is associated with cell proliferation, whereas overexpression of activated Raf is associated with cell-cycle arrest in G₁ phase.^217,218 Different Raf genes have different functions in cells, although A-Raf and B-Raf share three conserved domains termed CR1, CR2, and CR3.²¹⁹ A-Raf is able to upregulate the expression of cyclin D₁, cdk2, cyclin E, and cdk4, whereas B-Raf and Raf-1 induce p21^cip1, leading to a G₁ arrest.^216,219 The mode of action of these Raf molecules is not fully understood but one explanation why they act differently may be because they activate different downstream pathways, namely the MAPK (MEK [MAP/ERK (extracellular signal-regulated kinase)]). The three different MAPK cascades are the ERK–, c-Jun N-terminal kinase (JNK)/stress-activated protein kinase (SAPK)–, and p38 pathways. The MAP kinase pathways consist of three types of kinases in a series, MAPKKK, MAPKK, and MAPK (ERK), each sequentially activating the next kinase. The MAPK cascades all transmit responses from several different surface receptors to the nucleus.²²⁰ One explanation for the oncogenic effects of the MAPK pathway is that ERK activates c-Myc via phosphorylation on serine 62.²²¹ In addition, repression of c-Myc is required for terminal differentiation of many cell types, including hematopoietic cells. Thus, deregulated expression of c-Myc in both M1 AML cells and in normal myeloid cells derived from murine marrow blocks terminal differentiation and its associated growth arrest, and also induces apoptosis, which is dependent on the extrinsic Fas/CD95 pathway. New data suggest a linkage between c-Myc downregulation, the p16^INK4A/cyclinD1/RB, and SMAC/Diablo apoptotic pathways.²²² Several different transcription factors have been implicated in the downregulation of c-Myc expression during differentiation, including CCAAT/enhancer binding protein (C/EBP) α, CTCF, BLIMP-1, and RFX1. Alterations in the expression and/or function of these transcription factors, or of the c-Myc and Max interacting proteins, such as MM-1 and Mxi1, can influence the neoplastic process.^223,224

Experiments on oncoproteins have focused on apoptosis, the lethal response of a cell to either DNA damage or to signaling through cell surface “death” receptors. Key regulators of apoptosis induced by DNA damage are the multiple members of the Bcl-2 family of proteins, which include Bcl-2, Bcl-x_L, Mcl-1, Bax, Bak, Bim, and Bad, among others. Bcl-2 is involved in the t(14;18) chromosomal translocation, which is found classically in follicular lymphoma.²²⁵ The disruption of these loci increases expression of Bcl-2, and results in the uncontrolled accumulation of malignant B cells, because of an impaired balance between growth and apoptosis.^226,227 It also has been shown that Bcl-x_L, Bax, and Bad are involved in the regulation of AML cells. For example, the ratio between Bax and Bcl-2 is a prognostic factor in this myeloid neoplasm.²²⁸ Additionally, Mcl-1 may be even more critical to the development and maintenance of AML than Bcl-2 or Bcl-_XL.²²⁹ The nuclear HDAC complex, which regulates the structural conformation of DNA and therefore the activation of several genes, is targeted by ETO, the fusion partner of the RUNX1 gene in some patients with AML. The t(8;21) translocation that occurs in such patients allows the formation of a stable complex between the HDAC complex and ETO, contributing to leukemogenesis.^230,231 PLZF, PLZF-RARα, and BCL-6 are other oncogenes that target the HDAC complex.^232,233

Almost every cancer harbors one or more abnormalities of tumor-suppressor genes. These include mutations, translocations, deletions, and epigenetic modifications. In addition, at least two epigenetic mechanisms—the hypermethylation of CpG islands in the promoter region and the aberrant acetylation of histones (especially histone H4)—can silence tumor-suppressor genes in a variety of human cancer cell lines and primary tumors.

The products of three important tumor-suppressor genes (RB, P53, and p16^INK4A) are interconnected biochemically. The RB gene maps to chromosome 13q14 and has several downstream effectors, among which the transcription factor E2F is the best characterized.²³⁴ The RB gene family consists of three closely related proteins, RB, p107, and p130. All three proteins are able to interact with several E2F family members.

Transcriptional activation and repression are mediated via complexes consisting of RB family members, E2F family members, and so-called DP proteins.²³⁵ Besides its role in cell-cycle control, RB can modulate RNA polymerase activity, thus linking cell-cycle progression to transcriptional regulation. Many cellular proteins have been identified that bind to RB. These proteins can be divided into different groups, including transcription factors, growth factors, protein kinases, protein phosphatases, and nuclear matrix proteins. Mutations of RB are frequent in cancer e.g., leukemias; soft-tissue sarcomas; and breast, esophagus, prostate, and renal carcinomas.²³⁶ Several viral or oncoproteins can bind to and inactivate RB.^112,237

The p53 gene has been called a “guardian” of the genome because it transmits signals arising from various forms of DNA damage, leading to cell-cycle arrest or apoptosis. p53 protein is also the target of leukemogenic mutations. Damaging factors, such as hypoxic stress, chemicals, or irradiation either alter the p53 protein itself or can stabilize its cellular inhibitor, MDM2 (in humans the homologue is termed HDM2).²³⁸ The MDM2 protein inhibits p53 transcription and stimulates p53 degradation.^239,240 For the former, MDM2 is able to bind to the transactivation domain of p53 through a p53-interacting domain on the MDM2 amino terminus, which inhibits p53 from binding to its transcriptional coactivators, thereby preventing activation of p53 transcriptional targets. For the latter, MDM2 binds to p53 through a RING-domain and ubiquitinates p53 through the E3 ubiquitin ligase activity of MDM2, causing p53 nuclear export and ultimate degradation.²⁴¹ Moreover, MDM2 is able to affect chromosomal stability independently of p53.²⁴² The MDM2 binding region contains several phosphorylation sites.²⁴³ These residues (e.g., serine 395) are phosphorylated by DNA damage-activated kinases (e.g., ATM and Chk2), which is required for p53 activation.²⁴⁴ The p14^ARF tumor-suppressor gene, which is encoded within the p16^INK4A locus by alternate splicing, controls MDM2 activity.²⁴⁵ In de novo AML, although p53 mutations are not common,²⁴⁶ overexpression of MDM2 is frequently encountered,²⁴⁷ and there is current interest in exploring MDM2 antagonists, for example, RO5503781, in combination with cytarabine in patients with relapsed or refractory AML. Additionally, the MDM2 antagonist nutlin-3A causes p53-dependent apoptosis in AML cells by disrupting the p53-MDM2 interaction²⁴⁸ and synergizes with BH3-mimetics,²⁴⁹ MEK/ERK inhibitors,^250,251 aurora kinase inhibitors,²⁴⁹ cdk1 inhibitors,²⁵² XIAP antagonists,²⁵³ FLT3 inhibitors,^254,255 and inhibitors of nuclear export, such as CRM1.²⁵⁶ The p14^ARF gene shares exons 2 and 3 with p16^INK4A but has a distinct exon 1. The discovery that two important tumor-suppressor genes are encoded by the same chromosomal locus and share several exons was unexpected and is unique in human biology. The p16^INK4A gene function depends on p53, because overexpression of p16^INK4A causes cell-cycle arrest in p53 wild-type cells but not in p53-deficient cells.²⁵⁷ The transcription of p16^INK4A is regulated by E2F, which is under the control of RB.²⁵⁸ This indicates the existence of yet another feedback loop, which links the RB pathway to p53.²⁵⁹ The Ras protein is another identified p16^INK4A factor involved in MDM2-p53-p21-RB regulation.^260,261 Different signaling routes that connect DNA damage with p53 include a cascade of Ser/Thr kinases, for example, ATM, ATR, Chk1 and Chk2, which phosphorylate p53.²⁶² Abnormalities of p53 are found in slightly more than 50 percent of all human tumors and, surprisingly, even in some normal cells. It is unclear if these “normal” cells represent a pool of premalignant cells in an otherwise healthy individual or, more likely, p53 changes are just one step in multistage tumorigenesis. Two different p53 homologues, p63 and p73, have been described, which show DNA binding, transactivation, and oligomerization domains similar to p53.²⁶³ This similarity in the DNA binding domain allows p63 and p73 to regulate p53 target genes, induce cell-cycle arrest and apoptosis, and therefore act as tumor suppressors.²⁶⁴ The p73 gene has been localized to chromosome 1p36, a common region of cytogenetic changes in cancer. p73 protein can also bind p53, inhibiting its transcriptional regulatory activity.²⁶⁵ Although p53 mutations are found frequently in almost all cancers, p63 and p73 mutations are much more rare.^264,266 However, the p73 gene is inactivated by hypermethylation of CpG islands in its promoter region in both leukemias and lymphomas.²⁶⁷

Homozygous deletions of the p16^INK4A/p14^ARF gene locus on human chromosome 9p21 have been detected in gliomas,^107,268 primary cancers of lung,^107,269 bladder,²⁷⁰ head and neck,²⁷¹ as well as in acute T-cell leukemias^272,273 and mesotheliomas.²⁷⁴ Because inherited mutations of p16^INK4A exon 2 may interfere with its expression and/or function, without causing an amino acid change in p14^ARF, it is clear that p16^INK4A inactivation alone is an important step in the evolution of malignant disease. However, in established tumor cell lines, nearly all chromosome 9p21 deletions disable the entire p16^INK4A/p14^ARF locus. Both proteins act as suppressors of the G₁-S transition, even though they function in two different pathways: p16^INK4A acts as an inhibitor of cyclin D₁/cdk4/6 complexes whereas p14^ARF stabilizes p53 by inhibition of MDM2. Several models provide insight into the different modes of action of p16^INK4A and p14^ARF on cell-cycle regulation. Interestingly, if the entire p19^ARF/p16^INK4A locus is disrupted in mice (the mouse homologue of p14^ARF is p19^ARF), the mice develop lymphomas, lymphoid leukemias, and sarcomas, suggesting that these tumor-suppressor genes do not act in a lineage-specific manner on cell-cycle regulation but in a more general way. Retroviral expression of p16^INK4A restores the normal phenotype in some cell types underlining the strong tumor-suppressor potency of p16^INK4A. The p15^INK4B gene, also located on chromosome 9p21, approximately 20 kb centromeric of p16^INK4A, is deleted somewhat less frequently. Analyses of primary tumors, however, show that not all 9p21 deletions encompass these three tumor suppressor genes. One mechanism for disruption of the p15^INK4B/p14^ARF/p16^INKA region in T-cell leukemias may be the action of an illegitimate variable diversity joining (V[D]J) recombinase.²⁷⁵ Several other binding partners of p16^INK4A have been identified.²⁷⁶ The RB gene interacts with one of these factors, BRG1, to remodel chromatin structures. BRG1 also acts upstream of RB with p16^INK4A and functions as a tumor suppressor.²⁷⁶

The p15^INK4B/p16^INK4A/p14^ARF locus on chromosome 9p21 is a hotspot in the development of human cancer and approximately 50 percent of all human malignancies show abnormalities in at least one of these tumor-suppressor genes. Another gene lies about 100 kb telomeric of p16^INK4A, and this gene, methylthioadenosine phosphorylase (MTAP), encodes an important enzyme in purine metabolism. Some early gliomas show MTAP deletions without deletions of other genes on 9p21, suggesting that MTAP by itself has tumor-suppressor properties. The reexpression of MTAP in breast cancer cells severely inhibits their ability to form colonies in soft agar or collagen, supporting this hypothesis.²⁷⁷ In addition, MTAP-expressing cells are suppressed for tumor formation when implanted into severe combined immune deficiency (SCID) mice. Recent findings suggest that the enzyme ornithine decarboxylase (ODC) is overexpressed in MTAP-deleted tumors, providing evidence for a new pathway in tumorigenesis. Overexpression of ODC has been observed in many tumors and is linked to the Ras pathway.²⁷⁸ Reexpression of MTAP in ODC overexpressing cells decreases ODC levels and inhibits tumor cell growth.²⁷⁹ In addition, high levels of 5′-deoxy-5′-(methylthio) adenosine (MTA) induce matrix metalloproteinase and growth factor gene expression in melanoma cells, leading to enhanced invasion and vasculogenic mimicry. In addition, MTA induced the secretion of β-fibroblast growth factor and the upregulation of activator protein-1, demonstrating a tumor-supporting role of MTA, which is increased in MTAP-deficient cells.²⁸⁰

The mechanisms by which the above-mentioned genes are inactivated are rather different. Especially in permanent cell lines, p15^INK4B/ p14^ARF/p16^INK4A and MTAP are homozygously deleted. One allele of MTAP is also deleted in AML lines, but not in primary AML samples. Mutations in p15^INK4B/p14^ARF/p16^INK4A genes are rare, and if present, occur in exon 2. Hypermethylation of CpG islands in the promoter areas of p15INK4B/p14^ARF/p16^INK4A are frequently found in hematologic malignancies.^281,282,283 The availability of demethylating agents such as 5-azacytidine and 5-aza-2′-deoxycytidine (decitabine) makes this phenomenon an interesting target for chemotherapy.^284,285 Decitabine has been used to treat patients suffering from different hematologic malignancies and was reported to have activity in advanced myelodysplastic syndrome (MDS), accompanied by demethylation of the p16^INK4A promoter.²⁸⁵ Azacytidine and decitabine are approved for the treatment of MDS^286,287,288 and also widely used for the treatment of AML, particularly older patients considered unfit for cytotoxic chemotherapy, particularly when blast counts are low.^289,290,291 However, p16^INK4A and p15^INK4B are not the only targets of these demethylating agents in hematologic malignancies.²⁹² Transcriptional regulation by methylation is mediated by a multiprotein complex consisting of a MeCP2, a methylcytosine-binding protein with a transcriptional repressor domain that binds the corepressor mSin3A, which is itself one element of a multiprotein complex that includes HDAC1 and HDAC2.^293,294 Therefore, reexpression of silenced genes can be achieved by demethylating DNA or by destabilizing HDACs, and it could be demonstrated that both mechanisms are tightly linked. HDACIs and demethylating agents act synergistically to induce genes silenced in cancer by hypermethylation.¹³⁹ Another new mechanism of gene regulation and inactivation in vivo is degradation by microRNAs. This has also been shown for several members of the p16^INK4A/cdk4/cyclin D₁/RB pathway.²⁹⁵

HDACs catalyze the deacetylation of lysine residues in the histone N-terminal tails and are found in large multiprotein complexes with transcriptional corepressors. Human HDACs are grouped into three classes based on their similarity to known yeast factors: class I HDACs are similar to the yeast transcriptional repressor yRPD3; class II HDACs are similar to the yeast transcriptional repressor yHDA1; and class III HDACs are similar to the yeast transcriptional repressor ySIR2 (Table 16–5; Fig. 16–3).^296,297 Eleven different HDACs have been identified so far. The physiologic counterparts of the HDACs are histone acetyl transferases (HATs). In the nucleosome, positively charged hypoacetylated histones bind tightly to the phosphate backbone of the DNA and maintain the chromatin in an inactive, silent state. Both HAT and HDAC are recruited to target genes in complexes with sequence-specific transcription factors and their cofactors. Examples of these cofactors include NCoR or SMRT (Fig. 16–4). Several different transcription factors are assembled with these complexes, including Bcl-6, MAD1, PML, and ETO.²⁹⁶ HDACs are involved in different cellular mechanisms, including proliferation and differentiation. Irregular activation of HDACs leads to the loss of cell-cycle control.²⁹⁸ Gene silencing by HDAC complexes is an important mechanism in the development of AML, most notably APL. The PML-RARα fusion protein is an oncoprotein that represses retinoic acid-dependent transcription by recruitment of HDAC to RAR-regulated genes (Fig. 16–4B), halting myeloid maturation because of cell-cycle arrest. In the PML-RARα fusion protein, the RARα is not responsive to physiologic concentrations of retinoic acid and supraphysiologic doses of all-trans-retinoic acid are necessary to overcome the tight HDAC-recruitment and the consequent cell-cycle block.²⁹⁶ The rare translocation t(11;17) fuses the RARα gene to the PLZF gene, which directly interacts with the NCoR–mSin3a–HDAC complex to suppress gene transcription. This block can only be overcome by the addition of a HDACI. Another well-known example of transcriptional silencing by the recruitment of an HDAC repressor is the AML1-ETO fusion protein which results from the t(8;21) translocation. As already described, the addition of an HDACI can relieve ETO-mediated transcriptional repression.²⁹⁹ Although 11 HDACs have been described, only limited information is available about their redundant biologic and physiologic functions. As shown in Figure  16–4B, inhibitors of HDAC activity lead to the reexpression of silenced genes and to the induction of differentiation. Most of these inhibitors, such as depsipeptide (romidepsin), belinostat or vorinostat,³⁰⁰ do not exhibit isoenzyme selectivity and may therefore be of limited therapeutic value, at least as single agents. These drugs are currently approved for patients with previously treated peripheral and cutaneous T-cell lymphomas, although they continue to be studied for other indications, for example, vorinostat for AML in combination with chemotherapy (NCT01802333). The pan-HDACI panobinostat, in combination with bortezomib and dexamethasone, has been approved in the treatment of patients with relapsed or refractory myeloma,³⁰¹ while the class I–selective HDACI entinostat is currently being studied in phase III clinical trials in advanced hormone-responsive breast cancer in conjunction with aromatase inhibitors (NCT02115282). Finally, pracinostat (pan-HDACI) and mocetinostat (isotype-selective) have been granted “orphan drug” status for AML,¹⁴⁷ and for MDS and diffuse large B-cell lymphoma with specific mutations in HATs (e.g., CREBBP and EP300), respectively. However, the HDACI valproic acid, an established antiepileptic agent, is the first drug within this group that selectively inhibits one HDAC, namely HDAC2.³⁰² Valproic acid induces proteasomal degradation of HDAC2. Basal and valproic acid-induced HDAC2 turnover strongly depend on the E2 ubiquitin conjugase Ubc8 and the E3 ubiquitin ligase RLIM. Thus, polyubiquitination and proteasomal degradation provide an isoenzyme-selective mechanism for downregulation of HDAC2.³⁰² This also underlines the importance of another cell-cycle element, the proteasome.

The proteasome is a 2.4 MDa, multicentric protease complex with an important role in cellular protein regulation. Its structure consists of a cylindrical core, the so-called 20S particle, composed of four stacked rings with a total of seven proteins in each ring. The second part of the proteasome, two copies of a 19S particle, is bound to the 20S core. Only proteins that have been ubiquitinated can be degraded in the proteasome. The ubiquitination of different substrate proteins involves the sequential action of three enzymes: E1 (an ATP-dependent ubiquitin-activating enzyme), E2 (a ubiquitin-conjugating enzyme), and E3 (ubiquitin-protein ligase). The ubiquitin-proteasome pathway plays a critical role in the degradation of intracellular proteins involved in cell-cycle control, transcription activation, apoptosis, and tumor growth through an ATP-dependent mechanism.³⁰³ Proteins such as HDAC2 are tagged with several ubiquitin molecules and then degraded in the machinery.³⁰² Several tumors depend on rapid cell cycling, which requires expression and degradation of numerous regulatory proteins. Some of the proteins that undergo proteasome-mediated degradation include cyclins (cyclins A, B, D, E), endogenous cdk inhibitors (p27^kip1, p21^cip1), p53, RB, cdc25 phosphatase, and others.³⁰⁴ The rapid turnover of these proteins triggers the rapid growth rate of certain human malignancies, thus the proteasome is an excellent new target for the development of new drugs, as attested to by the success of the proteasome inhibitors bortezomib^305,306 and carfilzomib³⁰⁷ in patients with MCL and myeloma. These agents inhibit the proteolytic activity of the proteasome and so cells accumulate in the G₂-M phase of the cell cycle with a decrease of cells in G₁.^304,308 For example, p27^kip1, p21^cip1are upregulated in myeloma cells after the treatment with bortezomib, leading to cell-cycle arrest and apoptosis.³⁰⁹

The proteasome is also required for activation of the nuclear transcription factor NF-κB, which in response to environmental stress or cytotoxic agents, plays a role in maintaining cell viability through the transcription of inhibitors of apoptosis, and the DDR including DNA-damage checkpoints³¹⁰ and DNA repair.³¹¹ Based on these observations, targeting the proteasome has become a successful approach to cancer treatment³¹² and with a better understanding of the human cell-cycle machinery, it will be possible in the future to identify new targets for antineoplastic therapies.