This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Shapiro, G. I. Search for Related Content PubMed PubMed Citation Articles by Shapiro, G. I.

Cyclin-Dependent Kinase Pathways As Targets for Cancer Treatment

Geoffrey I. Shapiro

From the Department of Medical Oncology, Dana-Farber Cancer Institute; and the Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA

Address reprint requests to Geoffrey I. Shapiro, MD, PhD, Dana Farber Cancer Institute, Dana 810A, 44 Binney St, Boston, MA 02115; e-mail: geoffrey_shapiro{at}dfci.harvard.edu

	ABSTRACT

TOP ABSTRACT INTRODUCTION G1->S PROGRESSION AND CANCER... S-PHASE PROGRESSION AND E2F-1... CDK1 PARTICIPATES IN THE... TRANSCRIPTIONAL CDKS AND THE... SMALL-MOLECULE CDK INHIBITORS Appendix Author's Disclosures of... Author Contributions GLOSSARY REFERENCES

 
Cyclin-dependent kinases (cdks) are critical regulators of cell cycle progression and RNA transcription. A variety of genetic and epigenetic events cause universal overactivity of the cell cycle cdks in human cancer, and their inhibition can lead to both cell cycle arrest and apoptosis. However, built-in redundancy may limit the effects of highly selective cdk inhibition. Cdk4/6 inhibition has been shown to induce potent G₁ arrest in vitro and tumor regression in vivo; cdk2/1 inhibition has the most potent effects during the S and G₂ phases and induces E2F transcription factor–dependent cell death. Modulation of cdk2 and cdk1 activities also affects survival checkpoint responses after exposure to DNA-damaging and microtubule-stabilizing agents. The transcriptional cdks phosphorylate the carboxy-terminal domain of RNA polymerase II, facilitating efficient transcriptional initiation and elongation. Inhibition of these cdks primarily affects the accumulation of transcripts with short half-lives, including those encoding antiapoptosis family members, cell cycle regulators, as well as p53 and nuclear factor-kappa B–responsive gene targets. These effects may account for apoptosis induced by cdk9 inhibitors, especially in malignant hematopoietic cells, and may also potentiate cytotoxicity mediated by disruption of a variety of pathways in many transformed cell types. Current work is focusing on overcoming pharmacokinetic barriers that hindered development of flavopiridol, a pan-cdk inhibitor, as well as assessing novel classes of compounds potently targeting groups of cell cycle cdks (cdk4/6 or cdk2/1) with variable effects on the transcriptional cdks 7 and 9. These efforts will establish whether the strategy of cdk inhibition is able to produce therapeutic benefit in the majority of human tumors.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION G1->S PROGRESSION AND CANCER... S-PHASE PROGRESSION AND E2F-1... CDK1 PARTICIPATES IN THE... TRANSCRIPTIONAL CDKS AND THE... SMALL-MOLECULE CDK INHIBITORS Appendix Author's Disclosures of... Author Contributions GLOSSARY REFERENCES

 
The cyclin-dependent kinases (cdks) are heterodimeric complexes composed of a catalytic kinase subunit and a regulatory cyclin subunit, and comprise a family divided into two groups based on their roles in cell cycle progression and transcriptional regulation.^1,2 Members of the first group comprise core components of the cell cycle machinery, and include cyclin D–dependent kinases 4 and 6, as well as cyclin E–cdk2 complexes, which sequentially phosphorylate the retinoblastoma protein (Rb), to facilitate the G₁S transition.³ Cyclin A–dependent kinases 2 and 1 and cyclin B–cdk1 complexes are required for orderly S-phase progression and the G₂M transition, respectively.⁴ cdks are regulated by positive phosphorylation, directed by cdk-activating kinase (CAK; cyclin H/cdk7/MAT1),⁵ as well as negative phosphorylation events,⁶ and by their association with cyclins and endogenous Cip/Kip or INK4 (inhibitor of cdk4) inhibitors.⁷ In malignant cells, altered expression of cdks and their modulators, including overexpression of cyclins and loss of expression of cdk inhibitors, results in deregulated cdk activity, providing a selective growth advantage.^8,9 In contrast to cdks governing the transitions between cell cycle phases, transcriptional cdks, including cyclin H–cdk7, and cyclin T–cdk9 (pTEFb), promote initiation and elongation of nascent RNA transcripts by phosphorylating the carboxy-terminal domain (CTD) of RNA polymerase II.^10-12 Because of their critical role in cell cycle progression and cellular transcription, as well as the association of their activities with apoptotic pathways, the cdks comprise an attractive set of targets for novel anticancer drug development.

	G1S PROGRESSION AND CANCER CELL CYCLES

TOP ABSTRACT INTRODUCTION G1->S PROGRESSION AND CANCER... S-PHASE PROGRESSION AND E2F-1... CDK1 PARTICIPATES IN THE... TRANSCRIPTIONAL CDKS AND THE... SMALL-MOLECULE CDK INHIBITORS Appendix Author's Disclosures of... Author Contributions GLOSSARY REFERENCES

 
The G₁S Transition
The retinoblastoma susceptibility protein, Rb, plays a central role in the G₁S transition (Fig 1). ¹³ In its hypophosphorylated state, Rb prevents progression from G₁ to S through its interaction with E2F transcription family members. This interaction not only blocks transcriptional activation of E2F, but also actively represses transcription by recruiting histone deacetylases to the promoters of genes required for S-phase entry.¹⁴ During cell cycle progression, Rb is inactivated by sequential phosphorylation mediated by cyclin D–dependent kinases 4 and 6 and cyclin E–cdk2 complexes.^15,16 In response to mitogenic stimulation, cells synthesize D-type cyclins that assemble with cdks 4 and 6, a process that requires contribution of a Cip/Kip family member. Although Cip/Kip family members promote activity of cyclin D–dependent kinases, they are also potent inhibitors of cdk2.¹⁷ Therefore cyclin D–dependent kinases facilitate G₁ progression by first phosphorylating Rb, relieving transcriptional repression by the Rb-E2F complex, and by sequestering Cip/Kip proteins, facilitating activation of cyclin E–cdk2. Cyclin E–cdk2-mediated Rb phosphorylation disrupts the binding of Rb to E2F, allowing E2F activation and the transcription of genes necessary for S-phase entry and progression, including cyclin E itself.^18,19 Although Rb is the primary target of cyclin D–dependent kinases, cyclin E–cdk2 phosphorylates other targets as well, including p27^Kip1,^20,21 which further facilitates S-phase entry, p220 nuclear protein mapped to the Ataxia Telanglectasia locus (^NPAT), which stimulates replication-dependent histone gene transcription,^22,23 and nucleophosmin, which regulates centrosome duplication.²⁴

View larger version (30K): [in this window] [in a new window]   Fig 1. The G1S transition. In response to mitogenic signals, cyclin D/cyclin-dependent kinase (cdk) 4/6/Cip/Kip complexes assemble, sequestering Cip/Kip proteins from cyclin E–cdk2. Cyclin D–and E–dependent kinases phosphorylate the retinoblastoma protein (Rb), resulting in release of E2F, which is necessary for transcription of genes required for S-phase progression, including cyclin E itself, creating a positive feedback loop at the G1-S boundary. Cyclin E–cdk2 also phosphorylates p220NPAT, nucleophosmin, and p27Kip1, the latter ubiquitinated and degraded. As cells age, p16INK4 is induced, which inhibits cdk4/6, causing release and degradation of D cyclins and redistribution of Cip/Kip proteins to cyclin E–cdk2. Antiproliferative signals induce Cip/Kip proteins that also inhibit cyclin E–cdk2. Recently identified complexes are outlined in blue. Cells exit quiescence and proliferate in the absence of cdk4/6, compensated by cyclin D–cdk2. In the absence of cdk2, cyclin D–dependent kinases phosphorylate Rb at cdk2 sites; cyclin E–cdk1 complexes may also functionally substitute. NPAT, nuclear protein mapped to the Ataxia Telanglectasia locus.

Altered Expression of Cell Cycle Proteins in Human Cancer
The D-cyclin–cdk4/6–INK4–Rb pathway is universally disrupted in human cancer. Although Rb loss occurs commonly in some tumor types, the majority of human cancers retain wild-type Rb. Instead, most tumors increase activity of cyclin D–dependent kinases by multiple mechanisms. Most commonly, p16^INK4A is inactivated by gene deletion, point mutation, or transcriptional silencing by methylation.^26,27 This permits escape from senescence during the evolution to malignancy, and leaves the cancer cell with increased cyclin D–dependent kinase activity and a growth advantage. p16^INK4A has been proved definitively to be a tumor suppressor, given that mice with targeted disruption of p16^INK4A, and retention of the tumor suppressor p19^ARF, encoded by an overlapping reading frame, are predisposed to tumorigenesis.²⁸ Alternatively, both amplification of cdk4 and cdk4^R24C mutation resulting in the loss of INK4 binding also occur. The cdk4^R24C mutation was first described in malignant melanoma,²⁹ and knock-in mice expressing this mutant develop tumors with almost complete penetrance³⁰ and melanoma with high frequency.³¹ Loss of Rb, inactivation of p16^INK4A, and amplification or mutation of cdk4 are usually mutually exclusive events^32-35; in mouse models, expression of cdk4^R24C and loss of INK4 proteins do not cooperate in tumor development, which is consistent with cdk4 inhibition as the primary function of INK4 family members.³⁶

Overexpression of cyclin D1 is also common. Mammary hyperplasia and carcinoma occur in mouse mammary tumor virus (MMTV) –cyclin D1 transgenic mice,^37,38 and cyclin D1 is a critical mediator of breast cancer induction by Ras or Neu oncogenes, given that MMTV-v-Ha-ras and MMTV-c-neu transgenic mice are protected from breast cancer in a cyclin D1 knockout background.³⁹ Cyclin D1 overexpression can occur as a result of gene rearrangement, including the chromosome 11p15;q13 inversion first described in a parathyroid adenoma⁴⁰ or the t11;14(q13;q32) translocation in mantle-cell lymphoma in which the coding region of cyclin D1 is juxtaposed to the immunoglobulin heavy chain gene, resulting in high levels of cyclin D1 in lymphoid cells, where normally only cyclins D2 and D3 are expressed.⁴¹ Similar translocations targeting D cyclins have been described in multiple myeloma.^42,43 Gene amplification is another common mechanism leading to aberrant overexpression of cyclin D1, although overexpression occurs commonly in the absence of amplification as well.⁴⁴ Alternative splicing of cyclin D1 creates a transcript encoding cyclin D1b, which lacks the carboxy-terminal sequence containing the Thr²⁸⁶ phosphorylated by glycogen synthase kinase-3 ß.^45,46 Because Thr²⁸⁶ phosphorylation is required for nuclear export, cyclin D1b is constitutively nuclear and exhibits enhanced transforming ability in vitro⁴⁷; its presence in the majority of mantle-cell lymphomas suggests the importance of persistent nuclear expression for oncogenicity rather than simple overexpression.⁴⁸ Finally, mutations in cyclin D1 at the Thr²⁸⁶ residue or small adjacent deletions that prevent nuclear export, and also stabilize cyclin D1, have been reported in endometrial carcinomas.⁴⁹

Cyclin D1 overexpression often accompanies loss of p16^INK4A, suggesting that these events may cooperate in promoting transformation. Cyclin D1 may have cdk-independent transcriptional functions of potential tumorigenic relevance.^50,51 Nonetheless, mice are protected by MMTV-neu–driven tumorigenesis not only in the absence of cyclin D1, but also in the absence of cdk4⁵² or the presence of p16^INK4A,⁵³ indicating that in this instance, the cdk-dependent effects of cyclin D1 are critical. Loss of p16^INK4A or overexpression of cyclin D1 increases the amount of cdk4 available for assembly with cyclin D and Cip/Kip proteins; the sequestration of Cip/Kip proteins in cyclin D–dependent kinase complexes promotes activation of cyclin E–cdk2 and hence augments phosphorylation of and inactivation of Rb. Therefore, increased activity of cyclin E–cdk2 is a consequence of cyclin D–cdk4/6–INK4 pathway alterations. High levels of cyclin E,⁵⁴ including hyperactive low molecular weight isoforms⁵⁵ generated by elastase⁵⁶ or calpain,⁵⁷ as well as low levels of p27^Kip1 resulting from increased proteasomal degradation, also contribute to increased cyclin E–cdk2 activity in transformed cells and tend to define tumors that carry a worse prognosis.^58-62

Targeting the Cyclin D–cdk4/6–INK4 Pathway
The frequency of cyclin D–cdk4/6–INK4 pathway alterations suggests that acceleration of G₁ progression provides a proliferative and perhaps survival advantage to cancer cells. Preclinical data suggest that inhibition of cyclin D–dependent kinase activity may have therapeutic benefit. For example, antisense-mediated reduction of cyclin D expression results in inhibition of tumor growth, reversal of oncogenicity, and in some cases, transformed cell death.^63-66 Similarly, ectopic expression of p16^INK4A using inducible promoters or adenoviral gene delivery systems has been shown to induce Rb-dependent G₁ arrest,^67,68 as well as apoptosis in vitro.⁶⁹ In some but not all instances, cooperation with p53 appears necessary for the apoptotic response.⁷⁰ In vivo, adenovirus vector-mediated expression of p16^INK4A in non–small-cell lung cancer cell lines lacking p16^INK4A potently inhibits their growth when they are injected as xenografts in nude mice, and also slows tumor growth when injected into established xenografts.⁷¹ Similar experiments with established mesothelioma xenografts resulted in tumor regression.⁷²

These observations have motivated the development of cdk4/6 inhibitors that might achieve selectivity for transformed cells. In cells that retain Rb, a cdk4/6 inhibitor should reduce Rb phosphorylation and induce G₁ arrest; in tumors lacking Rb, in which p16^INK4A is present at high levels and already associated with cdk4/6, such an agent would be ineffective. Until recently, compounds with cdk4/6 inhibitory activity also potently inhibited other cdks. However, compounds specific for cdk4/6 have now been described. Of particular interest is PD 0332991,⁷³ of the pyridopyrimidine class, members of which were optimized by testing against cyclin D1–cdk4, cyclin A–cdk2, fibroblast growth factor receptor, and platelet-derived growth factor receptor. Compounds with high selectivity for cdk4 were then further tested for their ability to induce G₁ arrest (without arrest in other cell cycle phases) in an Rb-positive breast carcinoma line. This strategy identified a subset of pyridopyrimidines with exceptional selectivity for cdk4; PD 0332991 was selected for its superior pharmacokinetic properties and has entered phase I clinical trial.^74,75 As expected for a specific cdk4/6 inhibitor, this compound induced exclusive concentration-dependent G₁ arrest in Rb-positive cell lines in vitro, with dephosphorylation of Rb at known cdk4-specific phosphorylation sites including Ser⁷⁸⁰ and Ser⁷⁹⁵. The concentrations that inhibited cellular proliferation by 50% (IC₅₀) against a panel of Rb-positive cell lines ranged from 40 to 400 nmol/L, whereas IC₅₀ values against two Rb-negative cell lines was more than 3 µmol/L.

In breast and colon carcinoma cells, the drug is cytostatic; after G₁ arrest and cessation of proliferation, cells did not die, even after prolonged exposure. Strikingly, however, in mice bearing Colo-205 colon carcinoma xenografts, PD 0332991 produced rapid tumor regression.⁷³ It is possible that the tumor represents a balance of proliferating and apoptotic cells, and inhibition of the proliferative compartment allows the naturally dying cells to predominate. p16^INK4A replacement has also been associated with downregulation of vascular endothelial growth factor,⁷⁶ and suppression of angiogenesis could have also contributed to the in vivo effects. Suppression of Rb phosphorylation at Ser⁷⁸⁰ was also demonstrated in vivo, with a corresponding decrease in expression of E2F-1–dependent genes and Ki67 staining. As expected, PD 0332991 was inactive against Rb-negative tumor xenografts.

New Insights Into Cell Cycle Biology
Recent experiments with knockout mouse embryonic fibroblasts (MEFs), as well as antisense and siRNA-mediated depletion of cdks in tumor cells, have challenged multiple aspects of the model for G₁ progression depicted in Figure 1.⁷⁷ For example, quiescent, serum-starved cdk4/cdk6-null fibroblasts enter S phase with kinetics similar to those of wild-type MEFs (albeit with lower efficiency), and early-passage double-mutant MEFs displayed a doubling time similar to that of wild-type MEFs.⁷⁸ Therefore, cyclin D–cdk4/6 complexes are not essential for exit from quiescence after exposure to mitogenic stimuli or for the proliferation of embryonic fibroblasts. Compensation for the loss of cdk4/6 occurred with accumulation of cyclin D–cdk2 complexes capable of phosphorylating Rb and inducing cell proliferation. Furthermore, introduction of a retrovirus encoding small interfering of RNA (siRNA) targeting cdk2 with approximately 70% knockdown did not affect proliferation of wild-type MEFs but efficiently blocked proliferation of cdk4/cdk6-null MEFs, which provides further evidence indicating that cdk2 can compensate for the loss of cdk4 and cdk6.

These results suggest that a similar bypass may occur in tumor cells after continuous exposure to a specific cdk4/6 inhibitor. Although continuous dosing beyond 14 days was not performed in preclinical in vivo experiments with PD 0332991, rechallenge experiments were performed in responding xenografts. Colo-205 colon xenografts that had undergone complete regression during 14 days of dosing were permitted to grow back; tumors that re-emerged were collected and reimplanted into naïve mice. After tumors grew to 100 to 150 mg, mice were treated with PD 0332991. Importantly, tumors responded with equal sensitivity and again underwent substantial regression, demonstrating that no resistance had emerged during the initial brief treatment period.⁷³

The model in Figure 1 places cdk2 in a central role at the G₁-S boundary. However, several lines of evidence now indicate that cdk2 is not essential for cell proliferation. This evidence was first demonstrated in a variety of cancer cell lines that were able to proliferate after specific and acute depletion of cdk2 by siRNA or antisense oligonucleotides.⁷⁹ In addition, cdk2 knockout mice are viable.^80,81 Although there was a defect in germ cell meiotic cell division, the survival of mice for up to 2 years indicated that cdk2 is not required for mitotic cell division or survival of most cell types. Interestingly, cdk2–/– MEFs did display delayed entry into S phase, consistent with a role of cdk2 in the timing of S-phase entry^80,81 Nonetheless, the effects of cdk2 loss at the G₁-S boundary are less potent than anticipated, with little effect on overall proliferation of either normal or malignant mammalian cells. Furthermore, Cip/Kip proteins are able to induce G₁ arrest in cdk2–/– cells, again suggesting that cdk2 is dispensable and easily bypassed.⁸² In the absence of cdk2, there is compensation from other cdk family members. For example, cdk4 can phosphorylate Rb even at cdk2-preferred sites, including Thr⁸²¹.^80,83 G₁ arrest by Cip/Kip inhibitors in cdk2–/– cells is likely mediated by inhibition of cyclin E–cdk1 complexes, recently demonstrated in both wild-type and knockout cells.⁸⁴

These data suggest that highly selective inhibition of cdk2 may not be useful therapeutically. One exception may be malignant melanoma. In melanocytes, cdk2 undergoes transcriptional regulation by the melanocyte lineage transcription factor microphthalmia-associated transcription factor ( MITF).⁸⁵ Microarray data sets reveal a tight correlation in expression for MITF and cdk2 in primary human melanomas, but not other malignancies, defining melanomas with high versus low levels of cdk2. Low levels of MITF and cdk2 expression in melanoma cell lines accurately predict increased susceptibility to G₁ arrest induced by siRNA targeting cdk2 or decreased proliferation in response to roscovitine, a cdk2 inhibitor.⁸⁵ Therefore, subsets of melanomas may be sensitive to a selective cdk2 inhibitor.

	S-PHASE PROGRESSION AND E2F-1–DEPENDENT APOPTOSIS

TOP ABSTRACT INTRODUCTION G1->S PROGRESSION AND CANCER... S-PHASE PROGRESSION AND E2F-1... CDK1 PARTICIPATES IN THE... TRANSCRIPTIONAL CDKS AND THE... SMALL-MOLECULE CDK INHIBITORS Appendix Author's Disclosures of... Author Contributions GLOSSARY REFERENCES

 
S-Phase Events
After cdk-mediated phosphorylation of Rb during G₁, E2F activity is derepressed and E2F is released. E2F bound to its heterodimeric partner, DP-1, directs transcription of genes required for S phase. However, this transcription is activated only transiently. Orderly S-phase progression requires the downregulation of E2F-1 activity, accomplished in part by cdk-mediated phosphorylation^86-89 (Fig 2A). Inhibition of cdk activity during S phase results in inappropriately persistent E2F, which is known to cause S-phase delay and apoptosis. E2F-1–induced apoptosis can occur by both p53-dependent and p53-independent mechanisms,⁹⁰ the latter involving the activities of E2F-1 transcriptional targets, such as p73, Apaf-1, or caspase 3; repression of Mcl-1⁹¹; or the ability of E2F-1 to interact with death receptor and nuclear factor-kappa B (NF-B) pathways.⁹²

View larger version (26K): [in this window] [in a new window]   Fig 2. S-phase progression and E2F-1–dependent apoptosis. (A) After retinoblastoma protein (Rb) phosphorylation, E2F, along with a heterodimeric DP family member partner, directs transcription of S-phase genes. Transcription is activated transiently. E2F-1 activity is in part limited by phosphorylation, mediated by cyclin A–cyclin-dependent kinase (cdk) 2, cyclin A–cdk1, and cyclin H–cdk7. Appropriately timed neutralization of E2F activity is required for proper S-phase progression. (B) Inhibition of cdk activity during S phase results in inappropriate persistence of E2F activity. In normal cells, this is tolerable, and ultimately cells still progress through the cycle. However, transformed cells have high baseline levels of E2F activity, so that even a small reduction in cdk activity results in accumulation and persistence of high enough E2F activity to surpass the threshold required to induce apoptosis. Targeting of cyclin A or cdk2/1 during S phase has induced apoptosis selectively in transformed cells.

The targeting of E2F-1 phosphorylation via cdk inhibition may lead to death selectively in transformed cells. The disrupted cyclin D–cdk4/6–INK4–Rb pathway in tumor cells produces high levels of E2F activity. A small reduction in cdk activity during S phase may lead to persistence of E2F activity that has little effect on normal cells, but may leave transformed cells with inappropriately persistent E2F activity at a high enough level to surpass the threshold required to induce apoptosis⁹⁵ (Fig 2B).

Inhibition of cdk2 and cdk1
It is of interest that several approaches aimed at targeting cdk2 induce S and G₂ arrest, often followed by apoptosis. In fact, the majority of these approaches (ie, in which cdk4/6 is not concomitantly inhibited) have produced only weak G₁ arrest, so that the absence of potent G₁ arrest in cdk2–/– MEFs^80,81 or after siRNA ablation in established tumor cell lines⁷⁹ should not be surprising. The inducible expression of a dominant negative cdk2 mutant (dn-cdk2) in U2OS osteosarcoma cells is particularly instructive. In exponentially growing cells, low-level expression of dn-cdk2 resulted in G₂ arrest; induction of higher levels caused arrest during both the S and G₂ phases. Effects on G₁ progression were only observed when cells were synchronized and released from a nocodazole-induced mitotic block.⁹⁶ Similarly, the overall weak effect on the G₁-S boundary can be unmasked when cells are synchronized by passage directly before transient introduction of the dominant negative mutant.⁹⁷ S and G₂ cell cycle effects have also been described after ectopic expression of p27^{Kip1 98} or introduction of peptides capable of inhibiting cdk2 activity.⁹⁵

Although cell death was not observed in U2OS cells inducibly expressing dn-cdk2, experiments only characterized the effects for 24 hours and may not have been long enough to detect apoptosis. However, other approaches targeting cdk2 have resulted in profound apoptosis, including the introduction of cdk2 inhibitory peptides,⁹⁵ or after proteasomal degradation of cyclin A–cdk2.^99,100 Ectopic expression of p27^Kip1 has also induced apoptosis.^98,101 In addition, the inhibitory peptides, capable of blocking the interaction of cyclin A–cdk2 with E2F-1, caused abrupt cell death in tumor cells but not in nontransformed cells. Furthermore, rat fibroblasts engineered to inducibly express ectopic E2F-1 underwent apoptosis after peptide treatment only when the transgene was induced, indicating that elevated E2F-1 was sufficient to sensitize nontransformed cells to apoptosis after cyclin A-cdk2 inhibition.⁹⁵

The profound effects on S and G₂ cell cycle progression and apoptosis afforded by cdk2 inhibitory peptides, targeted cyclin A degradation, or ectopic expression of p27^Kip1 still need to be reconciled with the absence of similar cell cycle effects or cell death after introduction of antisense and siRNA targeting cdk2. One possibility is that the former approaches are targeting both cdk1 and cdk2. Ectopic p27^Kip1 expression would be expected to inhibit both cdks. The reported cdk2 inhibitory peptides were capable of cdk1 inhibition at high concentration, and the proteasomal degradation of cyclin A should affect cyclin A–cdk1 activity as well.^95,99 The scenario that both cdk2 and cdk1 inhibition are required also makes sense because both of these cdks play critical roles in the phosphorylation and modulation of E2F-1 activity. This hypothesis ultimately can be tested by additional work with cell lines engineered to express reduced levels of both cdk2 and cdk1 together.⁸⁴ In the absence of concomitant cdk4/6 inhibition, the primary effects of cdk2/cdk1 ablation may be during S and G₂ and associated with apoptosis.

The principles emerging from these approaches have been borne out with small-molecule cdk inhibitors. In one set of experiments, the pan-cdk inhibitor flavopiridol has been used. Because this compound inhibits multiple cdks, including cdks 4/6, 2, 1, and 7, it induces G₁ and G₂ arrest in many exponentially growing tumor cell types.¹⁰² However, if cells are first recruited to S phase, either by synchronization or by chemotherapy-induced S phase delay, they are sensitized to flavopiridol, so that cytotoxicity occurs.¹⁰³ Cell death is E2F-1 dependent, given that it is inhibited in E2F-1–/– cells,^104,105 and is also selective for transformed cells.¹⁰³ These results have provided rationale for clinical trials using the sequential combination of gemcitabine and flavopiridol, in which gemcitabine is administered at a fixed-dose rate to maximize both the incorporation of difluorodeoxycytidine triphosphate into DNA and the possibility of retardation of S-phase progression. After 16 to 24 hours, flavopiridol is administered.¹⁰⁶ Interestingly, such chemotherapy/flavopiridol combinations may be less sequence dependent in Rb-negative cells, in which G₁ arrest induced by flavopiridol is inherently less potent. In experiments in which adriamycin and flavopiridol were applied simultaneously to osteosarcoma cells, the drug combination sensitized Rb-negative SAOS-2 cells, but not a derivative in which Rb expression had been restored.¹⁰⁷

Several compounds that demonstrate nanomolar or low micromolar potency for inhibition of both cdk2 and cdk1, with significantly lower activity against cdk4/6, have been reported, including seliciclib (Cyclacel Ltd, Dundee, United Kingdom; CYC202, Cyclacel; R-roscovitine, Cyclacel),^108,109 BMS-387032 (SNS-032),^110,111 SU9516,¹¹² AZ703,^113,114 and amino imidazopyridine 1d.¹¹⁵ As expected, these compounds have been reported to induce S and G₂ arrest followed by apoptosis, with variable effects at the G₁-S boundary that are weak and more prominent after synchronization. Several of these compounds cause induction of E2F-1,¹¹⁶ and apoptosis is compromised in E2F-1 –/– cells. In the case of AZ703, introduction of a dominant negative E2F-1 mutant, capable of binding DNA and blocking transactivation, inhibited the apoptotic response, suggesting a dependence on E2F-1 transcriptional targets.¹¹⁷

Alteration of cdk activity during the S and G₂ phases may induce cell death by mechanisms other than the E2F-1–centric mechanism detailed here, including an interface with pathways mediating DNA damage and repair, which contribute to the cytotoxic synergy of chemotherapy agents and cdk modulators. These pathways are briefly presented in the Appendix (online only).

	CDK1 PARTICIPATES IN THE MITOTIC CHECKPOINT

TOP ABSTRACT INTRODUCTION G1->S PROGRESSION AND CANCER... S-PHASE PROGRESSION AND E2F-1... CDK1 PARTICIPATES IN THE... TRANSCRIPTIONAL CDKS AND THE... SMALL-MOLECULE CDK INHIBITORS Appendix Author's Disclosures of... Author Contributions GLOSSARY REFERENCES

 
Cdk1 Inhibition Compromises Survival After Engagement of the Mitotic Checkpoint
In addition to its known role at the G₂-M boundary, cyclin B–cdk1 recently has been implicated in cell survival during mitotic checkpoint (also known as spindle assembly checkpoint) activation. In response to microtubule stabilization by paclitaxel, spindle assembly checkpoint activation and mitotic arrest are associated with an increase in expression of survivin, an inhibitor of apoptosis protein and a mitotic regulator. Survivin is expressed in a cell cycle–dependent manner and is localized to various components of the mitotic apparatus where it contributes to the regulation of spindle microtubule function and cell viability. The increased expression of survivin during spindle checkpoint activation is related to its stabilization, afforded by phosphorylation on Thr³⁴ by cyclin B–cdk1.¹¹⁸ Expression of a dominant negative kinase–dead cdk1 mutant suppressed phosphorylation of survivin at Thr³⁴ and led to massive apoptosis in paclitaxel-treated cells. Similar data were obtained using cdk1 conditional knockout cells; overexpression of survivin reversed these effects.

It has also been proposed that after engagement of the mitotic checkpoint by taxanes, vinca alkaloids, or inhibitors of KSP (hsEg5 or kinesin-5, a member of the kinesin family of mitotic spindle motor proteins), cell death requires an outcome known as adaptation, in which cells exit mitotic arrest in the presence of drug, fail cytokinesis, and enter G₁.^119,120 Escape to G₁ despite continued mitotic checkpoint signaling provokes apoptosis. Because reduction in cyclin B–cdk1 activity is required for exit from mitosis, its inhibition after mitotic checkpoint engagement facilitates mitotic slippage and exit, and speeds cell death.

These concepts have been tested using small-molecule cdk inhibitors, including purvalanol A, flavopiridol, NU6140, and roscovitine. For example, pharmacologic ablation of cdk1 activity in mitotically arrested cells resulted in diminished phosphorylation on Thr³⁴ and survivin depletion followed by apoptosis.^118,121,122 In addition, flavopiridol has been shown to accelerate mitotic exit of paclitaxel-treated cells, promoting a pronounced decline in MPM-2 antibody reactivity, indicating depletion of mitotic phosphoprotein markers.¹²³ Similarly, application of purvalanol or roscovitine to mitotically arrested transformed cells promoted mitotic slippage, with a rapid reduction in expression of phospho-histone H3 and cyclin B, indicating passage to interphase of cells maintaining a 4N DNA content, followed by induction of substantial apoptosis with no recovery.^124,125 In contrast, release from a mitotic block in the absence of roscovitine induced cell death in only a proportion of cells, with normal recycling of the remainder of cells over time. The same events were not observed in nonmalignant cells, in which roscovitine treatment delayed the MG₁ transition after release from mitotic arrest without compromising cell survival. This may reflect the inherent full competence of the mitotic checkpoint in nontransformed cells, more likely to remain engaged and less susceptible to slippage.¹²⁰ This difference suggests that cell death associated with cdk1 inhibition after activation of the mitotic checkpoint may be selective for transformed cells.

The proposed mechanisms account for previous observations that demonstrated sequence-dependent cytotoxic synergy in vitro between paclitaxel and flavopiridol.^123,126 When flavopiridol is applied first, arrest at the G₂-M boundary prevents mitotic entry and subsequent paclitaxel-mediated cell death, but when applied after cells are arrested in mitosis, there is dramatic enhancement of apoptosis. These results have been verified in vivo; when treatment of mice bearing breast or gastric cancer xenografts with paclitaxel or docetaxel is followed by purvalanol A or flavopiridol,^118,127 there is significant enhancement of chemotherapy-induced anticancer activity, with both tumor growth delay and tumor regression observed. Paclitaxel and docetaxel have been studied in phase I and II combinations with flavopiridol,^128,129 although the optimal schedules of these agents, and the most appropriate interval between them, may be critical to the clinical outcome.^127,130

	TRANSCRIPTIONAL CDKS AND THE POTENTIAL OF CDK7/9 INHIBITION

TOP ABSTRACT INTRODUCTION G1->S PROGRESSION AND CANCER... S-PHASE PROGRESSION AND E2F-1... CDK1 PARTICIPATES IN THE... TRANSCRIPTIONAL CDKS AND THE... SMALL-MOLECULE CDK INHIBITORS Appendix Author's Disclosures of... Author Contributions GLOSSARY REFERENCES

 
Phosphorylation of the CTD of RNA Polymerase II
The CTD of RNA polymerase II is regulated by phosphorylation mediated by cdks. The human RNA polymerase II CTD contains 52 tandem repeats of the consensus heptapeptide sequence N-Tyr¹-Ser²-Pro³-Thr⁴-Ser⁵-Pro⁶-Ser⁷-C. Cyclin T-cdk9 (also called P-TEFb) preferentially phosphorylates the Ser2 sites of this sequence to promote transcriptional elongation. It is likely that cyclin T-cdk9 can phosphorylate the Ser5 position as well.¹¹ Cyclin H–cdk7/MAT1, in the complex of transcription factor TFIIH, preferentially phosphorylates Ser5, which facilitates promoter clearance and transcriptional initiation.¹² Cyclin H–cdk7 therefore plays a role both as a cell cycle and a transcriptional cdk; it acts as both CAK and a CTD kinase.

Flavopiridol and Seliciclib
Flavopiridol is the most potent known inhibitor of cdk9.^131-133 Whereas the IC₅₀ values for other cdks range from 100 to 400 nmol/L with Ki values between 40 and 70 nmol/L, the binding of flavopiridol to the adenosine triphosphate (ATP) binding site of cdk9 is significantly tighter, with 1:1 stoichiometry and a Ki value of 3 nmol/L. In fact, competition with ATP could not be demonstrated. Therefore, the inhibition by flavopiridol of cdk 9, as well as cdk7, has profound effects on cellular transcription.^134,135

The transcripts that are most sensitive to flavopiridol-mediated cyclin T–cdk9 P-TEFb inhibition and cyclin H–cdk7/MAT1 inhibition are those with short half-lives.^134,136 The rapid degradation of these mRNAs allows their levels to decrease when transcriptional initiation and elongation are inhibited. Several gene classes are relatively enriched for unstable mRNAs, including immediate early transcription factor genes and cytokine genes. Other gene classes encoding labile mRNAs include transcripts for apoptosis regulators, such as those encoding the antiapoptotic proteins Bcl-2, Mcl-1, and XIAP; cell cycle regulators, which encode cyclin D1, c-myc, and mitotic regulatory kinases; as well as NF-B–responsive gene transcripts and those contributing to the p53 pathway (Fig 3).¹³⁴

View larger version (24K): [in this window] [in a new window]   Fig 3. The transcriptional cdks 7 and 9 phosphorylate the carboxy-terminal domain (CTD) of RNA polymerase II (Pol II), facilitating the initiation of transcription and efficient elongation. Inhibition of these cyclin-dependent kinases (cdks) preferentially affects mRNAs with short half-lives, including those encoding antiapoptotic proteins, cell cycle regulators, and p53 and nuclear factor-kappa B (NF-B) pathway components. VEGF, vascular endothelial growth factor.

Seliciclib (R-roscovitine) also inhibits cyclin T-cdk9 and cyclin H-cdk7 in addition to cyclin E-cdk2,^108,109 and affects RNA polymerase II CTD phosphorylation, which is associated with a decrease in Mcl-1 as well as other antiapoptotic proteins in CLL cells.^143,144 The proapoptotic activity of both flavopiridol and seliciclib in multiple myeloma cell lines occurs by a similar mechanism.^145-147 Of note, Mcl-1 is also a target of drugs capable of prematurely terminating transcription, including 8-chloro-adenosine and fludarabine.^148,149

Currently, these results are being translated clinically in trials of flavopiridol and seliciclib in CLL, lymphoma, and multiple myeloma. For flavopiridol, novel drug schedules appear to be overcoming pharmacokinetic barriers.¹⁵⁰ Initial trials, both in hematopoietic malignancies and solid tumors, used 24- to 72-hour continuous infusions to reflect preclinical observations that prolonged exposure enhanced apoptotic effects in vitro and repeated low-concentration drug treatment demonstrated antitumor activity in vivo.^102,151,152 Although these schedules produced nanomolar concentrations, consistent with concentrations capable of producing preclinical effects, prolonged flavopiridol infusions have been largely inactive in multiple settings.^129,153,154 Subsequent data suggested the superiority of bolus administration designed to achieve high micromolar concentrations, which led to cures of lymphoma xenografts.¹⁵⁵ This prompted the development of 1-hour infusions, which have achieved micromolar maximal plasma drug concentrations, although with a short half-life.¹⁵⁶ Interestingly, although 24- or 72-hour continuous infusions were inactive in both CLL and mantle-cell lymphoma, 1-hour bolus infusions achieved low response rates, suggestive of activity.^157-160 Higher than expected concentrations of flavopiridol may be required in vivo because of 92% to 95% plasma protein binding.^161,162

More recently, a phase I study in relapsed CLL patients used a 30-minute bolus dose followed by a 4-hour infusion, pharmacokinetically modeled to achieve and sustain micromolar concentrations for several hours.¹⁶³ In a preliminary presentation of this study, a 41% response rate was achieved in 22 assessable patients; eight of nine responders had fludarabine-refractory disease, bulky lymphadenopathy, and either 11q or 17p deletion. Dose-limiting toxicity was life-threatening and fatal tumor lysis syndrome. Active investigation into the development of predictors for this hyper-acute syndrome is ongoing. Considering the improvement in response rate between continuous infusion and bolus flavopiridol in mantle-cell lymphoma, trials of bolus/infusion flavopiridol in this disease are anticipated. Combinations of flavopiridol with fludarabine and rituximab have demonstrated substantial but manageable toxicity, with high response rates.¹⁶⁴

Although the reduction in antiapoptotic proteins by inhibitors of cdks 7 and 9 may be adequate to induce significant cell death in some instances, their depletion may sensitize cancer cells to other apoptotic stimuli emanating from damage of DNA, modulation of microtubular stability, effects of tumor necrosis factor–related apoptosis inducing ligand or from signal transduction inhibition. For example, flavopiridol has been shown to potentiate imatinib-mediated apoptosis in BCR-ABL–positive leukemia cells¹⁶⁵; flavopiridol-mediated reduction in levels of Mcl-1 may have contributed to the observed synergism, and a phase I combination trial of imatinib and flavopiridol is ongoing. Similarly, flavopiridol-mediated downregulation of XIAP underlies its synergism with tumor necrosis factor–related apoptosis inducing ligand in leukemia cells,¹⁶⁶ and the attenuation of expression of mRNAs and proteins encoding multiple antiapoptosis regulators accounts for synergism with epothilones in breast cancer cells.¹⁶⁷

In addition to antiapoptotic genes, cell cycle regulators are also transcriptional targets of inhibitors of cdks 7 and 9. For example, cyclin D1 is readily depleted in flavopiridol-¹⁶⁸ and seliciclib-treated cells.¹⁶⁹ It also will be important to demonstrate the repression of transcription of cyclin D1b; nonetheless, depletion of D-cyclins adds rationale for the use of these agents in mantle-cell lymphoma and multiple myeloma, as well as other tumors that may be particularly dependent on cyclin D1, such as Her-2–positive breast cancer³⁹ or esophageal cancer.¹⁷⁰ Similarly, transcriptional repression of c-myc by inhibitors of cdk7 and cdk9 may be particularly pertinent in Burkitt lymphoma and small-cell lung cancer.

The inhibition of CTD phosphorylation may also have profound effects on p53-dependent and -independent expression of p21^Waf1/Cip1. For example, flavopiridol has been shown to reduce expression of mdm2, causing p53 stabilization.^171,172 In wild-type p53-expressing cells prone to apoptosis, including acute lymphocytic leukemia or germ cell tumor cells,¹⁷³ induction of p53 may lead to cell death. In addition, transcriptional repression can also prevent the induction of p21^Waf1/Cip1, so that p53 induction can occur without a concomitant increase in p21^Waf1/Cip1. Although this combination of events would be expected to produce apoptosis even in cells prone to p53-mediated cell cycle arrest,¹⁷⁴ flavopiridol-induced cell death is p53 independent,^172,175 indicating that other mechanisms or transcriptional targets must be contributing when the drug is used alone.

Interestingly, however, flavopiridol can also prevent induction of p21^Waf1/Cip1 in response to DNA damage. p21^Waf1/Cip1 mediates cell cycle arrest and also prevents apoptosis by binding to caspase 3. Therefore, in p53 wild-type colon carcinoma cell lines that arrest after irinotecan-mediated stimulation of p53 and p21^Waf1/Cip1, flavopiridol-mediated inhibition of the transcriptional induction of p21^Waf1/Cip1 can result in cell death.¹⁷⁶ In the context of a phase I trial of irinotecan and flavopiridol, patients with responsive or stable disease tended to express wild-type p53 and have no change or decrease in p21^Waf1/Cip1 levels post-treatment.¹⁷⁷ Therefore, in this setting, clinical benefit was correlated with p53 status and the ability of flavopiridol to inhibit a critical p53 effector than can prevent apoptosis.

p53-independent induction of p21^Waf1/Cip1 is characteristic of histone deacetylase inhibitors¹⁷⁸ and contributes to cytostatic and differentiation responses to these agents. In both leukemia and solid tumor models, abrogation of p21^Waf1/Cip1 by flavopiridol has resulted in enhanced cytotoxicity when it was combined with sodium butyrate,^179,180 suberoylanilide hydroxamic acid,¹⁸¹ or depsipeptide¹⁸²; the latter two combinations are also under clinical evaluation.

Additional cdk7/9-related targets of interest in flavopiridol-treated cells include suppression of NF-B–mediated transcription, perhaps underlying the preclinical synergism observed when flavopiridol is combined with either tumor necrosis factor alpha^172,183 or bortezomib.¹⁸⁴ Finally, flavopiridol has been shown to prevent the hypoxia-mediated induction of vascular endothelial growth factor, which may translate to antiangiogenic effects, and may account for responses and stable disease that have been observed in renal cell carcinoma.^185-187

	SMALL-MOLECULE CDK INHIBITORS

TOP ABSTRACT INTRODUCTION G1->S PROGRESSION AND CANCER... S-PHASE PROGRESSION AND E2F-1... CDK1 PARTICIPATES IN THE... TRANSCRIPTIONAL CDKS AND THE... SMALL-MOLECULE CDK INHIBITORS Appendix Author's Disclosures of... Author Contributions GLOSSARY REFERENCES

 
A growing number of cdk inhibitors representing multiple chemical classes currently are in clinical trial. The assessment of these drugs, along with pharmacokinetically superior schedules of flavopiridol, will help determine whether preclinical predictions regarding cell cycle inhibition will translate therapeutically. These drugs may be classified based on their effects against the cell cycle cdks as either pan-cdk inhibitors or more selective cdk inhibitors, with varying potency against the transcriptional cdks (Fig 4).

View larger version (59K): [in this window] [in a new window]   Fig 4. Small-molecule cyclin-dependent kinase (cdk) inhibitors may be classified based on effects on the cell cycle cdks. Pan-cdk inhibitors, including flavopiridol and AG-024322,203,204 inhibit cdks 4/6, 2, and 1. Other compounds are highly selective, including the pyridopyrimidine PD 0332991,73 and triaminopyrimidine, CINK4,205 and inhibitors of cdk4/6. Several other compounds inhibit cdk2 and cdk1 more selectively, including seliciclib109; BMS-387032, an aminothiazole111; PNU-252808, another thiazole derivative200,201; imidazo[1,2-a]pyridines, including AZ703114,117 and aminoimidazo[1,2-a]pyridine-1d115; and the purine-based NU6102202 and NU6140.122 Seliciclib, BMS-387032, and NU6140 are most selective for cdk2, whereas AZ703 and NU6102 inhibit cdk2 and cdk1 with similar potencies. Many compounds inhibit cdk9, including flavopiridol, seliciclib, BMS-387032, and AZ703. CINK4, PNU-252808, AZ703, NU6102, and NU6140 have been studied preclinically; the others listed have entered clinical trials.206

In the evaluation of all of these drugs, it will be important to include pharmacodynamic measures of whether the cdk targets are hit. This could include assessment of Rb and p27^Kip1 phosphorylation¹⁹³ as markers of cdk4 and/or cdk2 inhibition, as well as CTD phosphorylation, depletion of cyclin D1 or Mcl-1, or induction of p53 as markers of cdk7/9 inhibition. Such analyses have been done in recent trials including flavopiridol¹⁹⁴ and E7070, another cell cycle modulatory agent.¹⁹⁵ Results of these studies suggest that cdk inhibition can be achieved in tumor cells. However, correlation of effects on cell cycle–related phosphorylation events and effects on standard antiproliferative markers requires further clarification.^194-198 In addition, fluorothymidine positron emission tomography scanning may represent a noninvasive technology for monitoring cell cycle arrest in response to these agents.¹⁹⁹

Results on the group of compounds currently under study will determine whether more promiscuous cdk inhibition is preferable to selective cdk inhibition. For example, both seliciclib and BMS-387032 are relatively selective for cdk2. However, these and the majority of compounds that inhibit cdk2 also inhibit cdk1; compounds more equipotent for cdk2 and cdk1 may be superior. Whether a selective cdk2/cdk1 inhibitor will be therapeutically superior to a cdk2/cdk1/cdk4/6 inhibitor remains to be determined. In addition, given that inhibition of cdk2 and cdk1 may potentiate apoptosis induced by DNA-damaging agents that affect S-phase progression or by microtubule stabilizing agents, chemotherapy combination trials will continue to be important. Although this approach requires commitment to randomized studies, cdk inhibitors may ultimately find their place in combination regimens. General principles of cdk inhibition are summarized in Table 1.

One of the lessons from the flavopiridol experience is that for new compounds, it will be important to determine their relative potencies against both the cell cycle and the transcriptional cdks. Effects on cdk7/9 may be particularly relevant to inducing apoptosis in malignant hematopoietic cells and may sensitize solid tumor cells to a variety of apoptotic stimuli, including those induced by inhibition of cell cycle–related cdks. For example, it has recently been shown that cell death induced by the imidazo[1,2-a]pyridine cdk1/2 inhibitor AZ703 is potentiated in cells depleted of cdk9,¹¹⁷ indicating that cdks 1, 2, and 9 represent a promising cdk subset, inhibition of which may induce potent cytotoxic effects. Combined inhibition of cell cycle and transcriptional cdk activities may also be achieved with an inhibitor of cyclin H–cdk7,^208,209 a component of cdk-activating kinase that also phosphorylates E2F-1, and is a CTD kinase as well. This combination of activities may generate cytotoxic responses by causing perturbations of cell cycle progression, stabilization of E2F-1, and compromised expression of transcripts required by dividing cells, including those encoding anti-apoptotic proteins.²⁰⁷ Alternatively, detailed knowledge of the structure of cdk7 may pave way to inhibitors that can selectively target CAK or CTD kinase activity, which may help define therapeutic opportunities that are either cell cycle or transcription directed, each with value in specific disease contexts.²¹¹ Finally, the continued discovery of critical transcripts, repression of which mediates responses to transcriptional cdk inhibitors, either alone or in combination with other agents, will also identify novel targets for anticancer drug development.¹³⁵

	Appendix

TOP ABSTRACT INTRODUCTION G1->S PROGRESSION AND CANCER... S-PHASE PROGRESSION AND E2F-1... CDK1 PARTICIPATES IN THE... TRANSCRIPTIONAL CDKS AND THE... SMALL-MOLECULE CDK INHIBITORS Appendix Author's Disclosures of... Author Contributions GLOSSARY REFERENCES

 
The Appendix is included in the full-text version of this article, available online at www.jco.org. It is not included in the PDF version (via Adobe® Reader®).

Cdk Inhibition, DNA Damage and Repair
It has recently been shown that cdk inhibition during S phase, mediated by pharmacologic inhibitors dn-cdk2 and siRNA targeting cdk2, elicits an intra–S phase checkpoint that shares components of the pathway activated by double-strand DNA breaks (Zhu Y, Alvarez C, Doll R, et al. Mol Cell Biol 24:6268-6277, 2004). Cdk inhibition has been shown to trigger the accumulation of activated forms of the phosphatidylinositol 3-kinase family member ataxia-telangiectasia mutated (ATM) and the checkpoint kinase chk2, as well as nuclear foci containing phosphosphorylated substrates of ATM, including histone H2AX, often a marker for double-strand breaks (Maude SL, Enders GH. Cancer Res 65:780-786, 2005). These findings raise the possibility that cdk2 inhibition can induce or predispose to DNA damage, although the link is not considered solidified. Nonethelesss, roscovitine has been reported to induce nucleolar fragmentation (David-Pfeuty T. Oncogene 18:7409-7422, 1999).

One mechanism by which cdk inhibition could mediate DNA damage appears related to reduced expression of chk1 (Maude and Enders). Chk1, along with chk2, is typically activated by DNA damage in order to constrain cdk activity (Fig A1). However, after cdk inhibition during S phase, the response to cell cycle slowing involves downregulation of chk1, perhaps part of a negative feedback loop promoting cell cycle recovery. Reduction in cdk2 activity may slow or stall DNA replication forks. This block in replication is detected by ATR (ATM and Rad3-related), which primarily activates chk1. Stalled replication forms are dependent on the ATR-chk1 pathway for stabilization (Tercero JA, Diffley JF. Nature 412:553-557, 2001; Lopes M, Cotta-Ramusino C, Pellicioli A, et al. Nature 412:557-561, 2001); when chk1 activity is compromised, double-strand breaks may occur. This mechanism may in part explain the enhancement of topoisomerase II inhibition by cdk inhibitors; cytotoxicity induced by etoposide or doxorubicin is dramatically enhanced by roscovitine in a variety of transformed cell lines, with an increased frequency of double-strand breaks noted after combination treatment (Maude and Enders; Crescenzi E, Palumbo G, Brady HJM. Clin Cancer Res11:8158-8171, 2005).

View larger version (24K): [in this window] [in a new window]   Fig A1. Modulation of cyclin-dependent kinase 1 (cdk1) activity at the G2 checkpoint. After DNA damage, activation of the ataxia-telangiectasia mutated (ATM)/ATM and Rad3-related (ATR) kinases occurs, resulting in phosphorylation and activation of the checkpoint kinases chk1 and chk2, causing phosphorylation of cdc25C phosphatase on Ser-216, resulting in its binding to 14-3-3 proteins and ultimate cytoplasmic sequestration. This prevents cdc25C from removing inhibitory phosphates from cdk1; the latter remains inactive, preventing the entry of damaged cells into mitosis, contributing to G2 checkpoint control. UCN-01 inhibits chk1, facilitating persistent activity of cdc25C with dephosphorylation and activation of cdk1, and effective abrogation of the G2 checkpoint, with entry into mitosis. The ATM, chk2 and chk1 kinases also phosphorylate p53 after DNA damage, contributing to its stabilization. Stabilization of p53 after DNA damage occurs even in the presence of UCN-01, and can mediate G2 arrest via transcriptional induction of p21Waf1/Cip1 and 14-3-3, which contribute to cdk1 inhibition and sequestration, respectively. Therefore, in the presence of p53, UCN-01 does not prevent the arrest of cells in G2 after DNA damage. In the absence of p53, the checkpoint is completely dependent on the cdc25C pathway, and is severely compromised by UCN-01-mediated chk1 inhibition. Inappropriate entry into mitosis after DNA damage is frequently lethal. Thus, UCN-01 predisposes to mitotic catastrophe after DNA damage selectively in cells deficient in p53.

The ability of roscovitine to sensitize to doxorubicin did not occur in all cell lines (Crescenzi, Palumbo, and Brady). In some cell lines, an intra–S phase response to DNA damage involves Rb dephosphorylation and activation, so that Rb plays a role during both G1 and S phase transit (Knudsen KE, Booth D, Naderi S, et al. Mol Cell Biol 20:7751-7763, 2000). Rb dephosphorylation (activation) during S phase after DNA damage induces a prolonged, irreversible arrest and a senescence response to protect from genotoxic stress. In these cell lines, the coapplication of roscovitine with doxorubicin decreased the frequency of double-strand breaks and promoted the arrest response. Many, but not all, tumor cell lines demonstrate defective Rb activation during S phase after DNA damaging treatments (Broceno C, Wilkie S, Mittnacht S. Proc Natl Acad Sci U S A 99:14200-14205, 2002); in these cells, more prone to G2 progression than to cytostasis, the combination of doxorubicin and roscovitine or dn-cdk2 increases the degree of DNA damage and apoptosis. In this regard, the combination of DNA damage and cdk inhibition may induce synergistic cytotoxicity selectively in transformed cells, where the Rb response during S phase is more likely to be compromised.

These results suggest that a direct inhibitor of chk1, such as UCN-01 (Graves PR, Yu L, Schwarz JK, et al. J Biol Chem 275:5600-5605, 2000; Busby EC, Leistritz DF, Abraham RT, et al. Cancer Res 60:2108-2112, 2000) may directly mediate or amplify DNA damage responses during S phase (Syljuasen RG, Sorensen CS, Hansen LT, et al. Mol Cell Biol 25:3553-3562, 2005). In fact, this has been observed with concurrent treatment of a topoisomerase I inhibitor with UCN-01; the addition of UCN-01 resulted in increased phosphorylation of H2AX, suggesting augmentation of DNA double-strand breakage (Tse AN, Schwartz GK. Cancer Res 64:6635-6644, 2004). Although this phenomenon occurred in both wild-type and p53-negative colon carcinoma cells, the complexity of checkpoint defects in tumor cells may very well still permit potentiation that is not observed in nontransformed cells (Levesque AA, Kohn EA, Bresnick E, et al. Oncogene 24:3786-3796, 2005). UCN-01 may also influence DNA repair processes (Jiang H, Yang LY. Cancer Res 59:4529-4534, 1999).

G2 Checkpoint Modulation and cdk1 Activation In mammalian cells, cyclin B-cdk1 controls entrance into mitosis. In response to DNA damage, a cascade of events is activated to maintain the inhibitory phosphorylations of cdk1 at Thr¹⁴ and Tyr¹⁵ to promote G2 arrest, permitting repair before mitotic entry. This cascade is initiated by activation of ATM/ATR after DNA damage, resulting in the phosphorylation and activation of the checkpoint kinases chk1 and chk2, which phosphorylate the cdc25C phosphatase resulting in its cytoplasmic sequestration so that it cannot dephosphorylate and activate cdk1 (Weinert T. Science 277:1450-1451, 1997). These events are particularly critical in p53-negative cells, as the induction of p53 after DNA damage results in transcriptional activation of p21^Waf1/Cip and 14-3-3, which serve to inactivate cyclin B-cdk1 as well as cause its cytoplasmic sequestration (Bunz F, Dutriaux A, Lengauer C, et al. Science 282:1497-1501, 1998; Chan TA, Hermeking H, Lengauer C, et al. Nature 401:616-620, 1999; Fig A1).

Therefore, checkpoint kinase inhibitors are