Cyclin D As a Therapeutic Target in Cancer

Home , CDKN2B, CDKN2C, CDKN2D, Cyclin A, Cyclin B1, Cyclin D

REVIEWS

Cyclin D as a therapeutic target in cancer

Elizabeth A. Musgrove*‡, C. Elizabeth Caldon*, Jane Barraclough*, Andrew Stone* and Robert L. Sutherland*‡ Abstract | Cyclin D1, and to a lesser extent the other D‑type cyclins, is frequently deregulated in cancer and is a biomarker of cancer phenotype and disease progression. The ability of these cyclins to activate the cyclin-dependent kinases (CDKs) CDK4 and CDK6 is the most extensively documented mechanism for their oncogenic actions and provides an attractive therapeutic target. Is this an effective means of targeting the cyclin D oncogenes, and how might the patient subgroups that are most likely to benefit be identified?

7 INK4 family The cyclins were named because of their periodic, cell stability . Since their discovery, mammalian cyclins and This family of CDK inhibitor cycle-dependent pattern of expression. The synthesis of CDKs have been the focus of widespread attention as proteins specifically prevent individual cyclins, and consequent cyclin-dependent potential oncogenes, and there is an extensive body of the activation of CDK4 and kinase (CDK) activation at specific cell cycle stages, literature documenting their deregulation in cancer and CDK6, generally by inhibiting cyclin D association. coordinates the sequential completion of DNA replica- oncogenic capacity in experimental models (reviewed tion and cell division1,2. These kinases also underlie the in REF. 7). CIP and KIP family checkpoints that halt cell cycle progression in response This Review focuses on the role of D‑type cyclins as This family of CDK inhibitor to DNA damage and defects in the mitotic spindle. oncogenes and their potential as therapeutic targets in proteins bind cyclin–CDK Consequently, CDK activity is tightly regulated at mul- human cancer. The emphasis is on cyclin D1, because of complexes and are potent inhibitors of cyclin E–CDK2 tiple levels through several mechanisms. These include the weight of evidence for its widespread role in human and cyclin A–CDK2. They the abundance of the regulatory cyclin subunits; their cancer and the greater depth of its functional character act as assembly factors for association with the catalytic CDK subunit; activating ization compared with cyclin D2 and cyclin D3. cyclin D–CDK4 and cyclin D– and inhibiting phosphorylation events; and the abun- We summarize the progress that has been made with CDK6, but can also inhibit the activity of these kinases. dance of members of two families of CDK inhibitory inhibitors of the cyclin D‑associated kinases CDK4 and proteins — the INK4 family, which comprises INK4A CDK6 as potential cancer therapeutics and we evaluate (also known as p16), INK4B (also known as p15 and other means of targeting cancers with altered cyclin D CDKN2B), INK4C (also known as p18 and CDKN2C) expression, and how the patient subgroups that are most and INK4D (also known as p19 and CDKN2D), and likely to respond to therapeutic interventions that target the CIP and KIP family, which comprises p21 (also known cyclin D might be identified. as CDKN1A), p27 (also known as CDKN1B) and p57 (also known as CDKN1C)1,2. Biological functions of D‑type cyclins Cyclin D1, cyclin D2 and cyclin D3 are closely related CDK activation. The earliest known and best-understood G1 cyclins (BOX 1). They activate CDK4 and CDK6, which function of cyclin D is to promote cell proliferation *Cancer Research Program, Garvan Institute of Medical are phylogenetically distinct from the canonical CDKs, as a regulatory partner for CDK4 or CDK6. Extensive Research, 384 Victoria CDK1 and CDK2, and have different substrate speci- research into the underlying mechanisms led to the RB Street, Darlinghurst, Sydney ficity3,4. Extracellular signals, including growth factor pathway model (reviewed in REF. 7), in which the activa- NSW 2010, Australia. receptor activation and integrin-derived adhesion sig- tion of cyclin D–CDK4/CDK6 initiates the release of the ‡St Vincent’s Clinical School, Faculty of Medicine, nalling, influence cyclin D transcription, translation RB‑dependent cell cycle-inhibitory ‘brake’ that governs University of New South and protein degradation, thereby integrating mitogenic, cell cycle transitions during quiescence, senescence and Wales, Sydney NSW 2052, differentiation and attachment signalling with the cell differentiation (FIG. 1). The specific inhibition of cyclin D– Australia. cycle machinery5. The deregulation of cyclin expres- CDK4/CDK6 through the induction of INK4A promotes Correspondence to R.L.S. sion or CDK activation can directly lead to some of RB‑dependent cell cycle arrest in some circumstances, e‑mail: 6 8 [email protected] the hallmarks of cancer by causing proliferation that such as during senescence . The simple model in which doi:10.1038/nrc3090 is independent of normal extracellular cues, or by over- RB phosphorylation by CDKs, including cyclin D– Published online 7 July 2011 riding checkpoints that ensure genomic integrity and CDK4/CDK6, promotes the activation of E2F‑responsive

At a glance regulation of cell growth occurs through effects on both protein synthesis (through tuberous sclerosis complex • Cyclin D–cyclin-dependent kinase 4 (CDK4) or CDK6 activation promotes cell cycle (TSC) regulation of mTOR)19 and ribosome biogenesis progression through the phosphorylation of substrates, including RB and (through phosphorylation of MEP50, a co-activator of transcription factors with roles in proliferation and differentiation. These kinase the arginine methyltransferase PRMT5, which in turn complexes also target substrates with roles in centrosome duplication, mitochondrial 20,21 function, cell growth, cell adhesion and motility, and cytoskeletal modelling. methylates ribosome proteins) . With the exception • D-type cyclins have non-catalytic roles in which interactions with chromatin- of TSC regulation of mTOR, whether these substrates modifying enzymes and diverse transcription factors, including steroid hormone are targets for cyclin D–CDK complexes other than receptors, leads to the transcriptional regulation of suites of genes that are involved cyclin D1–CDK4 has not been examined in detail. in proliferation and differentiation. Independently of CDK activation, the D-type CDK4 and CDK6 have overlapping, but not identical, cyclins also facilitate efficient DNA repair and indirectly activate CDK2 through the substrate specificity in vitro22. The cyclin component of sequestration of CDK inhibitors. the kinase complex also confers some substrate specific- • CCND1 is an established human oncogene that is commonly overexpressed through ity22, and functional differences between cyclin D1 and copy number alterations, or more rarely by mutation, or as a consequence of the cyclin D2 have been identified23. Consequently, these deregulation of mitogenic signalling downstream of oncogenes such as ERBB2. roles may not be characteristic of D‑type cyclin–CDK CCND1 overexpression causes a number of potentially oncogenic responses in complexes other than cyclin D1–CDK4. experimental models and is associated with poor patient outcome. Some CDK4/CDK6 substrates have roles in cellular • Cyclin D1 and its associated CDKs are potential therapeutic targets. Promising results processes that are less directly involved in cell cycle from early CDK inhibitors in experimental systems were not followed by evidence for control, in particular cell motility, cell adhesion and efficacy in clinical trials. Possible reasons for this disappointing outcome include poor 24 pharmacokinetics, suboptimal dosing schedules and clinical testing in unselected cytoskeletal remodelling . Fibroblasts, epithelial cells patient populations. Second-generation, more selective inhibitors of CDK4 and CDK6 and macrophages have increased adhesion and reduced are now undergoing clinical testing. migration in the absence of cyclin D1 (REFS 25–27), and • Possible alternative approaches to targeting cyclin D1 include the use of compounds thymocyte adhesion is decreased in the absence of either that affect CCND1 transcription or cyclin D1 protein turnover, and the use of CDK4 or cyclin D3 (REF. 28). In fibroblasts and epithelial combination therapies that simultaneously target multiple end points of cyclin D1 cells, the expression of a cyclin D1 point mutant (K112E) action. Central to the effective use of these novel approaches is the better selection (BOX 1) that cannot activate CDK4 or CDK6 does not of patient subgroups that are likely to respond. decrease adherence or enhance motility, unlike the wild-type protein26,27, indicating that these effects are dependent on CDK activity. genes that are necessary for DNA synthesis has subse- quently been elaborated to take into account factors such Non-catalytic functions. Not all actions of cyclin D– as combinatorial interactions between RB family proteins CDK4/CDK6 depend on substrate phosphorylation. and various members of the E2F family, which includes One major non-catalytic function of the D‑cyclins is both activator and repressor proteins, as well as some transcriptional regulation. Cyclin D1 is tethered to the atypical E2Fs that repress transcription independently promoters of many genes during normal development, of RB family members (reviewed in REF. 9). probably through interactions with various transcription Evidence that cyclin D1 is not necessary for pro- factors29. It also binds regulators of histone acetylation liferation in cultured cells in the absence of func- and methylation30–32 (FIG. 2), seeming to act as a bridge that tional RB suggested that RB is the principal substrate links DNA-bound transcription factors with chromatin- for cyclin D1–CDK4, and presumably for other modifying enzymes and the transcriptional machinery cyclin D‑associated CDKs10, at least in terms of this spe- in order to regulate cell proliferation and differentia- cific end point of cyclin D1 action. However, cyclin D1 tion (reviewed in REFS 29,33,34). Cyclin D1‑responsive has functions other than facilitating the initiation of genes also include some that promote migration DNA synthesis, and these observations do not exclude and invasion, such as thrombospondin and the Rho the possibility that these functions might depend on effector ROCK2 (REF. 26). Cyclin D–transcription substrates other than RB. Indeed, more recent studies factor interactions commonly occur through motifs in have shown that cyclin D–CDK4 has physiologically regions that are poorly conserved between the D‑cyclins, relevant substrates in addition to RB family members pointing to specificity in the transcriptional effects of (FIG. 1). These include transcription factors, such as the the individual D‑type cyclins, although this has not transforming growth factor-β (TGFβ)-responsive tran- been widely tested and much of the published literature scriptional modulator SMAD3 (REF. 11); members of examines only cyclin D1. However, ectopic expression of the RUNX family12,13; GATA4 (REF. 14) and the MEF2 each of the D‑type cyclins in hepatocytes led to distinct family (REF. 15), which have roles in the proliferation transcriptional profiles35. and differentiation of specific cell lineages; and BRCA1 Multiple members of the steroid hormone recep- (REF. 16), which coordinates DNA damage repair, ubiq- tor superfamily and their co-regulators interact with uitylation and transcriptional regulation to maintain cyclin D1 (reviewed in REFS 33,34) (FIG. 2). Cyclin D1 genomic stability. Other CDK4 targets have roles in enhances oestrogen receptor-α (ERα) activity, through processes that are tightly linked to, and coordinated interactions with ERα and its co-regulators SRC1 (also with, chromosomal DNA replication and segregation, known as NCOA1) and AIB1 (also known as NCOA3). such as centrosome duplication and separation17, mito- Cyclin D2 and cyclin D3 interact poorly with ERα36,37. chondrial function18 and cell growth. Cyclin D–CDK4 In contrast to its activation of ERα signalling, cyclin D1

44,45 DNA damage response inhibits the activity of the androgen receptor (AR), that are involved in the DNA damage response . By A global cellular response that thyroid hormone receptor-β and peroxisome prolifera- binding BRCA2 and the recombinase RAD51, cyclin D1 halts cell cycle progression tor activated receptor-γ (PPARγ), which has a crucial facilitates the recruitment of RAD51 to sites of DNA while damaged DNA is role in fatty acid metabolism, energy homeostasis and damage and so promotes homologous recombination- repaired, or that triggers cell 44,45 death by apoptosis if the adipogenesis. Cyclin D3 interacts with cellular retinoic mediated DNA repair . Both the ability of cyclin D1 to damage is too extensive for acid-binding protein 2 (CRABP2) and retinoic acid enhance the DNA damage response and the formation of repair. receptor-α (RARα); it is the only D‑cyclin to interact RAD51 foci require p21 (REFS 45,46), suggesting that p21 with these proteins38 but, like cyclin D1, cyclin D3 also may also be present in the cyclin D1–RAD51–BRCA2 Cyclin box inhibits AR and PPARγ39,40. Both direct cyclin D3–AR complex. Importantly, decreased cyclin D1 expression, A domain that is characteristic of cyclins and has high binding and cyclin D3–CDK11 phosphorylation of but not treatment with a CDK4/CDK6 inhibitor, impairs 40,41 sequence conservation across AR have been implicated in these effects . DNA repair even in cells that lack RB and so do not the cyclin family. It mediates A second well-described non-catalytic role of cyc- require cyclin D1 for proliferation44. In addition, both cyclin–CDK binding. lin D is the sequestration of p21 and p27 by cyclin D– the cyclin D1 K112E mutant and wild-type cyclin D1 CDK4/CDK6, leading to the indirect activation of CDK2 can restore an efficient DNA damage response in cells (REF. 42). The interaction with p21 and p27 has a key role lacking all three D-type cyclins44. Thus, cyclin D1 facili- in coordinating CDK activity during G1 phase of the cell tation of DNA repair is independent of CDK4/CDK6 cycle, but may also be important for other end points. activation and distinct from cyclin D1 regulation of Cyclin D1 cannot promote migration following p27 proliferation. knockdown26, and p27 has CDK-independent effects on Many of the non-catalytic effects of the D‑type cyclins cell migration through RHOA and stathmin43, suggesting have been elucidated using cell culture models, and that cyclin D1 association with p27 could contribute to simultaneous knockout of all three D‑type cyclins has cyclin D1 effects on migration independently of CDK4. a very similar phenotype to knockout of both Cdk4 and Similarly, cyclin D1 interaction with p21 contributes to its Cdk6 (REFS 47,48), leading to questions over the degree to emerging non-catalytic function in DNA repair. which non-catalytic effects contribute to the normal cel- Cyclin D1 regulates the expression of genes that lular functions of cyclin D1. However, defects in retinal are involved in DNA replication and the DNA damage and mammary gland development in mice lacking Ccnd1 checkpoint, and it also interacts with a number of proteins are largely restored when the K112E cyclin D1 point

Box 1 | D‑type cyclins

The three D-type cyclins — AR and THRβ1 cyclin D1, cyclin D2 and cyclin D3 MYOD — are closely related. Overall, the TAFII 250 DMP1 human cyclin D2 and cyclin D3 p21 and p27 p21 and p27 proteins are 62% and 51%, p300, CBP and PPARγ ERα respectively, identical to human RB CDK SRC1 cyclin D1, and 62% identical to D1 295 aa each other (see the Figure). A K112 T286 second human cyclin D1 isoform, D1b 274 aa cyclin D1b, arises through K112 alternative splicing. It is identical to the full-length, canonical cyclin D1 D2 54% 78% 56% 48% 290 aa protein for the first 240 amino acids K111 T280 but diverges at the carboxy‑ D3 49% 73% 46% 36% 292 aa terminal and therefore lacks some K112 T283 key interaction domains69. Cyclin box PEST The greatest homology between LXCXE motif LLXXXL motif the D‑cyclins occurs in the cyclin box that mediates cyclin-dependent kinase (CDK) binding and is necessary for interaction with the CDK inhibitors Nature Reviews | Cancer p21, p27 and p57. All three D‑type cyclins share an RB‑binding LXCXE motif at the extreme amino‑terminal, a C‑terminal PEST domain that is rich in proline, glutamate, serine and threonine and that is characteristic of proteins that are rapidly turned over, and a threonine residue near the C terminus (T286 in cyclin D1) that, when phosphorylated, triggers ubiquitin-mediated degradation. T286 phosphorylation also promotes cyclin D1 nuclear export. Mutations within these highly conserved regions have been widely used to probe cyclin D1 functions; for example, the T286A mutant, which is stable and constitutively located in the nucleus, and the K112E mutant, which does not activate cyclin-dependent kinase 4 (CDK4) or CDK6. The K112E mutant retains the ability to bind CDK4 and sequester CDK inhibitors in some experimental models49, but not others27. The region between the cyclin box and the C terminus contains domains that are responsible for transcription factor– cyclin D1 interactions and is relatively poorly conserved. One key interaction domain is the C‑terminal leucine-rich motif of cyclin D1 (LLXXXL), which binds an LXXLL motif in the steroid receptor co-activators SRC1 and AIB1 (REF. 176). The corresponding region in cyclin D2 and cyclin D3 does not contain an LLXXXL motif, although it is leucine-rich. aa, amino acids; AR, androgen receptor; ERα, oestrogen receptor-α; PPARγ, peroxisome proliferator activated receptor-γ; THRβ1, thyroid hormone receptor β1.

Senescence

Centrosome Genetic INK4A duplication instability • Proliferation • Cell growth • Migration Other CDK Cyclin D Cyclin E • DNA damage targets • Centrosome CDK4 and CDK2 duplication CDK6 • Diﬀerentiation • Mitochondrial function p21 and p27 RB redistribution RB, p107 Cell cycle and p130 progression p130 E2F E2F Quiescence

E2F1

Diﬀerentiation Senescence Apoptosis Figure 1 | CDK-dependent functions of cyclin D. The activation of cyclin D–cyclin-dependent kinase 4 (CDK4) or CDK6 initiates the phosphorylation of the tumour suppressor protein RB and other RB family membersNature (suchReviews as p107 | Cancer and p130), resulting in the release of E2F transcription factors. In turn, this leads to the transcriptional activation of E2F‑responsive genes that are essential for DNA synthesis, including cyclin E and cyclin A, which further promote RB phosphorylation by activating CDK2. Cyclin D–CDK4/CDK6 complexes also indirectly activate cyclin E–CDK2 by sequestering the CDK inhibitors p21 and p27. Both INK4A and RB also affect cellular processes other than cell cycle progression, such as centrosome replication, so that their loss can lead to centrosome amplification and genomic instability. Similarly, cyclin D–CDK4/CDK6 phosphorylates substrates in addition to RB, thereby regulating a diverse set of end points (shown in the blue boxes).

mutant is knocked in to the Ccnd1 locus49. Similarly, perhaps acting as a ‘safety net’ for rapidly proliferating deletion of the gene encoding p27 (Cdkn1b) can rescue cells, but sustained cyclin D1–CDK4 activity following many of the developmental defects that are observed fol- DNA damage leads to the inappropriate re-replication lowing the deletion of Ccnd1 or Cdk4 (REFS 50,51). Thus, of DNA and chromosomal damage20,55,56. Collectively, the ability of the cyclin D1–CDK4 complex to sequester these observations raise the question of which p27 is required during development, but the kinase molecular functions of cyclin D1 are crucial during activity of this complex is not essential. This provides oncogenesis. strong support for the idea that non-catalytic functions Cells lacking all three D-type cyclins are resistant to of cyclin D1, particularly its ability to sequester CDK transformation by various oncogenes in vitro, and mice inhibitors, are physiologically relevant. lacking cyclin D1 are resistant to the effects of some, although not all, mammary oncogenes47,57. The CDK Oncogenic consequences of cyclin D deregulation. More activation function of cyclin D1 is largely dispensible than 100 proteins that interact with cyclin D1 in human during mammary development, but it is required for cancer cell lines have been identified44. Proteins that are mammary oncogenesis, as mice lacking cyclin D1 or involved in cell cycle control and transcriptional regula- CDK4, or expressing the CDK4/CDK6‑specific inhibition are prominent among these interactors. However, tor INK4A or the cyclin D1 K112E mutant, are resistant proteins that are involved in DNA repair, RNA metabo- to mammary cancers induced by ERBB2 (also known lism, protein folding, cell structure and cell organization as HER2 and Neu)49,57–59. The targeted deletion of are also enriched in the list of cyclin D1‑interacting pro- either Cdk6 or Ccnd3 causes resistance to lymphoma teins44. Consequently, the deregulation of cyclin D1 will genesis60,61, germline deletion of Cdk4 prevents carcinogen- not only promote mitogen-independent proliferation, and MYC-induced skin cancer7,62, and deletion of but may also affect other cellular processes, both directly Ccnd2 prevents colorectal adenomas in adenomatous and indirectly, in ways that have potentially oncogenic polyposis coli (Apc)-deficient mice63. The frequent sim- consequences. These consequences include angiogen- ilarities in phenotype between the deletion of D‑type esis, through the regulation of vascular endothelial cyclins and the deletion of CDK4 or CDK6 indicate growth factor (VEGF) expression52, centrosome dupli- that CDK activation makes a substantial contribu- cation53 and the DNA damage response. Nuclear cyc- tion to the requirement for the D‑type cyclins during lin D1 (but not cyclin D2) is rapidly degraded after DNA oncogenesis. However, it is important to note that these damage or replication stress as part of the S phase models of tumour initiation do not address the func- DNA damage checkpoint54, but the remaining low levels tions of the D‑type cyclins that may be required for contribute to efficient DNA repair44. High levels of cyclin the continued proliferation of established tumours or D1 prime cells for an enhanced DNA damage response45, their metastatic spread. Similarly, germline deletion of

BRCA2, RAD51 p27 Cyclin D1 and p21

DNA damage Migration response STAT3, DMP1 PPARγ BMYB and TAFII250 Nuclear hormone Transcriptional Cell cycle Adipogenesis receptors regulation progression

ERα, SRC1, Chromatin NEUROD, C/EBPβ GRIP1 modiﬁcation and MYOD and AIB1

Cell cycle AR C/EBPβ Tissue-speciﬁc progression HATs, PCAF and HDACs diﬀerentiation p300/CBP

Cell growth Androgen receptor-dependent Diﬀerentiation cell cycle progression Figure 2 | CDK-independent functions of cyclin D1. Although p21 and p27 are constituents of cyclin D– cyclin-dependent kinase 4 (CDK4) or CDK6 complexes, cyclin D1 can bind p21 or p27 independently of CDK4 or CDK6 binding, leading to effects on migration71 and the DNA damage response45, respectively. It also hasNatur effectse Re onviews the DNA | Cancer damage response through interactions with RAD51 and BRCA2 (REF. 44). Cyclin D1 regulates cell proliferation, cell growth and differentiation by binding representatives of several transcription factor families. These include nuclear hormone receptor family members (oestrogen receptor-α (ERα), androgen receptor (AR) and peroxisome proliferator-activated receptor-γ (PPARγ)) and their co-activators (SRC1, GRIP1 and AIB1), BMYB and the MYB-related transcription factor DMP1, as well as the helix–loop–helix transcription factors neurogenic differentiation factor 1 (NEUROD1), MYOD and C/EBPβ (reviewed in REFS 29,33,34). In addition, cyclin D1 binds chromatin-modifying enzymes, including histone acetyltransferases (HATs) such as P/CAF30, p300/CBP31 and histone deacetylases (HDACs)32. More general effects on 177 transcription can also result from cyclin D1 binding to TAFII250 (also known as TAF1) , a subunit of the basal transcriptional machinery. STAT3, signal transducer and activator of transcription 3.

these genes is not necessarily informative regarding the Cyclin D overexpression in cancer likely therapeutic success of inhibiting function rather CCND1 is a well-established human oncogene: a recent than expression. census concluded that there was substantial evidence Cancer-associated mutations in cyclin D1 that lead to for the involvement of CCND1 amplification and over- constitutive nuclear localization and impaired degrada- expression in breast cancer and significant evidence tion increase the transforming activity of cyclin D1 in for its involvement in lung cancer, melanoma and oral cell culture and reduce the latency of cyclin D1‑driven squamous cell carcinomas74. The criteria used included tumour development in animal models64. This has led deregulated expression, particularly when correlated to the view that the oncogenic actions of cyclin D1 are with clinical outcome, and biological consequences predominantly nuclear. However, in some cancers, such of altered expression; for example, the demonstration as lung, prostate and ovarian cancer, overexpressed that cyclin D1 overexpression in the mammary gland is cyclin D1 is exclusively cytoplasmic in a significant tumorigenic, albeit with a long latency and incomplete proportion of cases65–67. Cytoplasmic cyclin D1 inhibits penetrance75. As illustrated by the examples in TABLE 1, apoptosis following low-level DNA damage68, suggesting many common cancers have CCND1 amplification that cytoplasmic localization of cyclin D1 may also be rates of 15–40%, and higher rates of CCND1 mRNA relevant to its role as an oncogene. and protein overexpression. A translocation that jux- The cyclin D1b isoform (BOX 1) is present at low levels taposes CCND1 with the immunoglobulin heavy chain in many normal cells, but can be overexpressed in human locus (IGH), leading to cyclin D1 overexpression, is cancer and has biological properties that are different diagnostic of mantle cell lymphoma (MCL) (TABLE 1), from those of full-length cyclin D1 (reviewed in REF. 69). and the small proportion (<10% of MCLs) that lack this Cyclin D1b is constitutively localized to the nucleus, is translocation often display overexpression of CCND2 deficient in promoting RB phosphorylation and elicits or CCND3 (REF. 76). CCND2 or CCND3 amplification a transcriptional response that only overlaps with that of has been reported, but is rare compared with CCND1 full-length cyclin D1 by 33%. It also has a more potent amplification. CCND2 amplification is present in only transforming ability than full-length cyclin D1, perhaps 2% of gliomas77, and this is the only cancer in which because high levels of cyclin D1b do not trigger a DNA the evidence for cyclin D2 involvement is classed as damage response45,69–71. Cyclin D1b does not bind ERα72, significant74, although CCND1 amplification is one of and although it does bind AR in androgen-dependent the most common copy-number alterations in human prostate cancer cells, it stimulates proliferation, in cancer78. Although cyclin D2 is overexpressed in contrast to the inhibition of androgen-stimulated some cancers, CCND2 is frequently methylated, with proliferation by full-length cyclin D1 (REF. 73). loss of cyclin D2 expression in pancreatic, breast and

Table 1 | Cyclin D1 deregulation in cancer Mechanism of deregulation Tumour type Frequency Refs Amplification and overexpression CCND1 amplification Head and neck squamous cell carcinoma 26–39% 112,178 Cyclin D1 overexpression Head and neck squamous cell carcinoma 20–68% 112,178 CCND1 amplification Non-small-cell lung cancer 5–30% 65,74 Cyclin D1 overexpression Non-small-cell lung cancer 18–76% 65,74,88 CCND1 amplification Endometrial cancer 26% 179,180 Cyclin D1 overexpression Endometrial cancer 40–56% 179,180 CCND1 amplification Melanoma 0–25% 181 Cyclin D1 overexpression Melanoma 30–65% 181 CCND1 amplification Pancreatic cancer 25% 182 Cyclin D1 overexpression Pancreatic cancer 42–82% 182 CCND1 amplification Breast cancer 15–20% 74,102 Cyclin D1 overexpression Breast cancer 50–70% 74,102 CCND1 amplification Colorectal cancer 2.5% 183 Cyclin D1 overexpression Colorectal cancer 55% 184 Chromosomal rearrangement and overexpression CCND1: IGH translocation t(11;14)(q13;q32) Mantle cell lymphoma >90% 76 Cyclin D1 overexpression Mantle cell lymphoma >90% 76 CCND1: IGH translocation t(11;14)(q13;q32) Multiple myeloma 16% 185 Cyclin D1 overexpression Multiple myeloma 30–50% 185 Splice variants and transcript aberrations 3′ UTR rearrangements, microdeletions or point mutations Mantle cell lymphoma 4–10% 82,104 Cyclin D1b overexpression Breast cancer 22%* 89,91 Cyclin D1b overexpression Prostate cancer 27%* 90 Mutations affecting nuclear export and proteolysis Cyclin D1 T286R; ∆266–295 Oesophageal cancer 4% 83 Cyclin D1 P287S; P287T; ∆289–292 Endometrial cancer 4% 84 FBXO4 S8R, S12L, P13S, L23Q, G30N and P76T Oesophageal cancer 14% 85 FBXO4, F-box 4; IGH, immunoglobulin heavy chain locus; UTR, untranslated region.*Cyclin D1b overexpression without overexpression of full-length cyclin D1.

prostate cancer79–81, pointing to a potential role as a Many polymorphisms have been identified within the tumour suppressor rather than as an oncogene. CCND1 locus, but the G/A870 polymorphism is the only Mutations have also been implicated in aberrant one that has been investigated in any detail69. The A870 cyclin D1 expression (TABLE 1), although they have rarely allele is present in a large proportion of the population been investigated, and thus the importance of their (AA, 25.0%; AG, 50.0% in Caucasians) and is associated role is not yet clear. Mutations in the 3′ untranslated with a significant, but small, increase in cancer risk87. It region that result in the stabilization of the CCND1 favours the production of cyclin D1b. In some cancers, mRNA have been reported in MCL82, and muta- such as lung cancer, cyclin D1b is coordinately overex- tions and deletions clustering around T286 have been pressed with full-length cyclin D1, although the absolute reported in oesophageal and endometrial cancers83,84. expression level of cyclin D1b is much lower88. In other Phosphorylation at this site governs the turnover and cancers, cyclin D1b is independently overexpressed89–91. nuclear export of the cyclin D1 protein, and mutations The G/A870 polymorphism is not the only determinant in F-box 4 (FBXO4), an SCF E3 ubiquitin ligase that of increased cyclin D1b expression, and although factors targets T286‑phosphorylated cyclin D1 protein for that affect cyclin D1 splicing have been identified92, the degradation, have also been detected in endometrial mechanisms for the cancer-specific overexpression of cancers85: up to 20% of endometrial cancers may dis- cyclin D1b are not known (reviewed in REF. 69). play nuclear overexpression of stable cyclin D1 through Overexpression of cyclin D1 is much more common mutations in CCND1 or FBXO4. An initial study has than can be accounted for by copy number or muta- also indicated that cyclin D1 stabilization might be tional events that affect CCND1 (TABLE 1). Another important in breast cancer86. route to cyclin D1 overexpression is as a consequence

RTKs ERBB2 RHO

RAS Integrin signalling PI3K

MEK–ERK RAC AKT GSK3β FBXO4

WNT NF-κB miRNAs mTOR

Splicing Cyclin D

Transcription Translation Degradation

• CCND1 ampliﬁcation Cyclin D1b • Mutation near cyclin D1 T286 • CCND1–IgH translocation miRNA deletion overexpression • FBXO4 mutation • Oncogenic activation e.g. ERBB2

Figure 3 | Oncogenic activation of cyclin D1. Cyclin D1 abundance is regulated at multiple levels,Natur eache Re ofviews which | Cancer can be affected during oncogenesis (shown in the blue box). In addition to activating mutations that target CCND1, several other oncogenic events can affect cyclin D1 abundance. These include the activation of signalling through receptor tyrosine kinases (RTKs), and the MEK–ERK, WNT and nuclear factor-κB (NF-κB) pathways that lead to increased CCND1 transcription. Post-transcriptional effects on cyclin D1 include protein translation through the PI3K–mTOR pathway and effects on cyclin D1 protein turnover through phosphorylation of cyclin D1 at T286 by kinases, including glycogen synthase kinase 3β (GSK3β), that promote F‑box 4 (FBXO4)-mediated proteolysis. miRNA, microRNA.

of oncogenic activation of mitogenic signalling path- individual E2Fs7–9, there is a substantial body of experi- ways. Many signalling intermediates, including the mental evidence in support of its predictions. Data from RAS–MEK–ERK and PI3K pathways, regulate cyclin D1 clinical studies are less conclusive. This probably partly expression5 (FIG. 3). Overexpression of cyclin D1 through reflects the need to use surrogate markers for key end these pathways could have oncogenic consequences in points, such as proliferation, CDK4/CDK6 activity, RB addition to deregulation of mitogenic signalling, as status and E2F activity, in clinical material. However, it RAS-induced centrosome amplification is dependent on is also possible that some of the non-catalytic functions CDK4 (REFS 17,93). ERBB2‑driven or RAS-driven mam- of cyclin D1 contribute to its role in human cancer101,102. mary cancers in mice express high levels of cyclin D1 In MCL CCND1 mRNA expression correlates with (REFS 94,95), and most ERBB2‑positive human breast a proliferation signature that is comprised of genes that cancers display moderate or strong cyclin D1 expression are expressed at higher levels in dividing cells than in (for example, REFS 58,96,97). A newly emerging alterna- quiescent cells, with the highest levels of proliferation or tive mechanism for cyclin D1 overexpression is the loss cyclin D1 expression associated with the poorest overall of microRNAs (miRNAs) that target CCND1, such as survival103. The expression of this proliferation signature miR‑15a and miR‑16. Their expression is inversely cor- correlated with other markers of proliferation (such as, related with cyclin D1 expression in the small series of Ki67 expression and mitotic index)103. Although MCL prostate and lung cancers that have been examined to is distinguished by almost universal overexpression of date98,99 and an mir‑15A and mir‑16‑1 cluster is located cyclin D1, concomitant deletions of the CDKN2A locus at a common site of deletion (13q14.3) in several malig- (encoding INK4A and ARF), CDK4 amplification or nancies, including prostate cancer. There is, however, microdeletions in RB1 leading to loss of RB protein a second mir‑15A and mir‑16‑1 cluster at 3q26.1, and expression, also occur and are associated with more there are also several other miRNAs that target cyclin D1 proliferative, aggressive disease104. Increasing cyclin D1 (REF. 100), so it is not yet clear how important miRNA expression in the context of CDKN2A deletion is associ- deletion might be as a cause of cyclin D1 overexpression. ated with particularly poor patient outcome in both MCL and head and neck cancer103,105. One possibility is that Correlation with proliferation. The cyclin D–RB–E2F multiple lesions in the cyclin D1–RB–E2F pathway may pathway model depicted in FIG. 1 predicts that cyclin D1 cooperate to increase pathway activation and hence the overexpression leads to increased CDK4/CDK6 activity, likelihood of increased proliferation. Another is that RB phosphorylation, activation of E2F‑responsive genes the contribution of each component to processes other and so increased proliferation. Although this model than cell cycle control (FIG. 1) may offer an additional does not take into account complicating issues such selective advantage during oncogenesis. as RB interactions with a large number of cellular pro- The correlation between high cyclin D1 expression teins, combinatorial interactions between the different and markers of increased proliferation in MCL is ech- E2Fs and RB family members, and distinct roles for oed in many large studies in carcinomas. However, in a

large panel of different cancers there was no correlation with poor patient outcome116, suggesting that it may be between cyclin D1 expression and a signature of genes useful to assess the effects of cyclin D1 deregulation using responsive to E2F1 and E2F2 (REF. 106). There was, how- other parameters in addition to cyclin D1 expression. ever, a correlation between cyclin D3 and the signature In breast cancer, cyclin D1 overexpression is strongly of E2F1- and E2F2‑responsive genes106. Furthermore, associated with the ER‑positive, better prognosis sub- a study of 1,740 ER‑positive breast cancers found that type102. This probably accounts for the observation that in cyclin D1 expression was only moderately correlated unstratified breast cancer samples cyclin D1 overexpres- with a signature of RB inactivation that contained sion either is not significantly correlated with outcome or proliferation-related genes107, and another large study is associated with favourable outcome116. As breast cancer of 779 breast cancers from multiple cohorts found that subtypes display marked differences in phenotype and although expression of cyclin B1 and cyclin E1 was consist- clinicopathological features it is difficult to conclusively ently correlated with Ki67 expression, cyclin D1 expres- determine the effects of cyclin D1 overexpression unless sion was not108. Overall, these observations suggest that ER‑positive and ER-negative cancers are considered sepa- cyclin D1 action in breast cancer, and perhaps other can- rately. In ER‑positive patients, cyclin D1 expression was cers, might not simply be a consequence of increased significantly associated with a shorter time to metastasis proliferation, which is consistent with experimental and reduced patient survival in one large study117 (E.A.M., evidence showing that genes regulating various cellular R.L.S. and C. M. McNeil, E. K. A. Millar and S. A. O’Toole processes in addition to proliferation are responsive to unpublished observations). However, cyclin D1 expres- cyclin D1 (REFS 18,109,110) and that cyclin D1 expression did not correlate with survival in another study, sion is not correlated with CDK4 activity in breast although this study did find that CCND1 amplification cancer cell lines111. C/EBPβ, a transcription factor that was a significant independent predictor of survival118. regulates cellular differentiation, could be involved in Not all studies of cyclin D1 overexpression have con- these effects, as cyclin D1 overexpression antagonized trolled for treatment effects, although in experimental C/EBPβ repression of genes that comprise a cyclin D1 models, cyclin D1 overexpression causes resistance to signature in breast cancer106. However, the relation- some cytotoxic drugs, as well as to targeted therapies such ship between genes with promoters that are bound by as anti-oestrogens, the selective epidermal growth factor cyclin D1 and the expression of genes that correlate receptor (EGFR) tyrosine kinase inhibitor gefitinib, and with cyclin D1 overexpression in cancer has only been inhibitors of BRAF and MEK signalling119–124. It has also examined for selected cyclin D1‑responsive genes109,110. been implicated in acquired radioresistance125. Multiple One recent study was able to distinguish breast cancer studies of large cohorts with well-controlled treatment subgroups that displayed overexpression of cyclin D1 regimens have found a significant association between alone, rather than in combination with cyclin B1 over- high cyclin D1 expression and poor outcome in women expression, which was closely associated with prolifera- treated with tamoxifen96,126,127, with a further study noting tion108. The subgroup with both cyclin D1 and cyclin B1 a borderline association108. However, with the exception overexpression had a significantly poorer outcome. The of endocrine resistance in breast cancer, few studies have difference in phenotype implies a difference in intrinsic addressed the relationship between cyclin D1 expression biology and suggests that it may be informative to dis- and response to therapy in human cancer. tinguish cyclin D1 overexpression that either is or is not associated with increased proliferation. Targeting cyclin D as therapy for cancer Oncoproteins are attractive therapeutic targets as they Relationship with patient outcome. Cyclin D1 overexpres- are causally related to cancer development, and cancer sion is associated with shorter patient survival in many cells often become dependent on them for continued cancers and is often associated with increased metastasis proliferation and survival (oncogene addiction)128. The (REFS 104,112, for example), which is consistent with the D-type cyclins are generally regarded as difficult to target ability of cyclin D1 to enhance migration and invasion. In directly, as they lack intrinsic enzymatic activity and are lung and breast cancer this relationship has been addressed intracellular, although several possible approaches have in multiple large cohorts but remains unclear, at least been suggested (FIG. 4). Given the increasingly routine partly owing to confounding technical and reagent issues, clinical use of specific kinase inhibitors129, a more imme- and the presence of different isoforms and subcellular diately feasible approach has been to target cyclin D by localization of cyclin D1 (REFS 65,88,89,113,114). RB1 inhibiting associated kinases. mutation or deletion results in reduced cyclin D1 expres- sion115, so the highest and lowest extremes of cyclin D1 Therapeutic inhibition of CDK4/CDK6. Early evidence expression are both markers of RB pathway deregulation. of the high incidence of aberrant cyclin expression in

Oncogene addiction Consequently, in populations in which both RB1 inacti- myriad cancer types and the sensitivity of cancer cells Heightened dependency of vation and cyclin D1 overexpression are common, it is to acute inhibition of cyclins or CDKs, including cyc- cancer cells on specific difficult to draw definitive conclusions without separately lin D1, led to programmes that aimed to develop small- oncogenes, so that, despite the comparing high and low cyclin D1 expression to interme- molecule CDK inhibitors as therapeutics130,131. Initial presence of multiple genomic diate, presumably normal, levels of expression. Like cyc- enthusiasm was tempered by increasing experimental alterations, inactivation of a single oncogene can be lin D1 overexpression, RB loss does not have a clear-cut evidence indicating that individual CDKs, including 116 sufficient to impair proliferation association with patient outcome . However, a gene sig- CDK2 and CDK4, were largely dispensible during and survival. nature of RB loss seems to be more consistently associated development, and that CDK2 activity was not necessary

• RTK inhibitors cyclin D1 (REF. 132). However, established ERBB2‑driven (trastuzumab) Cytotoxics mammary cancers are sensitive to knockdown of either • Anti-oestrogens (REF. 58) (tamoxifen) cyclin D1 or CDK4 . The first small-molecule CDK inhibitors inhibited CDK1, CDK2 and CDK4 at submicromolar concentra- tions, and often also inhibited transcriptional CDKs, which are CDK family members such as CDK7 and mTOR CDK9 that have functions in transcriptional control inhibitors rather than (or in addition to) cell cycle control7. These Protein degradation pan-CDK inhibitors may also have inhibited antipro- • Radiotherapy (bexarotene and Cyclin D • Immunotherapy liferative CDK family members such as CDK10 and GSK3β inhibitors) CDK11 (REF. 7). This has rarely been tested, but may be important, as both cyclin D1 and cyclin D3 bind CDK11, Competitive and cyclin D3–CDK11 has been implicated in AR regu- binding inhibitors lation29,41. A second generation of CDK inhibitors with improved potency (IC of ~10 nM or less) and selectivity Cyclin D 50 • ATP for CDK4/CDK6 in preference to other CDKs was subse- Cyclin D competitors quently developed, and several of these inhibitors are cur- • Substrate rently in clinical trials (TABLE 2) after showing promising CDK4 mimetics antitumour activity in preclinical models. For example, Blood vessel the selective CDK4/CDK6 inhibitor PD0332991 causes cultured cells to arrest in G1 phase and inhibits the prolif- Invasion eration of xenografts of RB‑positive breast, ovarian, lung, Cyclin E colon and prostate cancer cell lines, glioblastoma cell VEGFR pan-CDK lines, and leukaemia, myeloma and MCL cell lines133–139. inhibitors CDK2 inhibitors That inhibition of CDKs is feasible in a clinical setting Figure 4 | Therapeutic targeting of cyclin D1. Possible therapeutic approaches to has been shown using biomarkers such as decreases in targeting cyclin D1‑dependent cancers range from downregulatingNatur cycline Reviews D1 to | Cancer RB phosphorylation and the expression of E2F‑responsive inhibiting end points of cyclin D1 action. The most immediately feasible approaches are genes140–143. However, the therapeutic efficacy of the first- to target proliferation through the inhibition of cyclin-dependent kinase 4 (CDK4) and generation pan-CDK inhibitors was modest, at least in CDK6, either alone or in combination with other CDKs, or to target cyclin D1 actions part because of poor pharmacokinetics, dose-limiting more generally but less directly by the use of compounds with actions that include toxicity and suboptimal dosing schedules, and several cyclin D1 downregulation or protein degradation, such as cytotoxics, radiotherapy and early trials were discontinued130,131. Improved potency and targeted therapies such as trastuzumab, tamoxifen or mTOR inhibitors. Greater selectivity is expected to reduce the toxicity from off-target specificity for individual functions could be achieved by using competitive inhibitors effects. Early results from Phase I trials of CDK4‑specific of specific protein–protein interaction domains in cyclin D1 or by substrate mimetics. 144–146 Emerging possibilities include broad-based inhibition, either by single molecules with inhibitors indicate that the side effects are tolerable , combined actions against multiple cyclin D1 targets, or by a combination of molecules in contrast to a Phase I trial of a second-generation pan- 143 each targeting a specific function. GSK3β, glycogen synthase kinase 3β; RTK, receptor CDK inhibitor . In vitro resistance to long-term treat- tyrosine kinase; VEGFR, vascular endothelial growth factor receptor. ment with a CDK4/CDK6‑specific inhibitor is associated with increased CDK2 activity in both breast cancer and acute myeloid leukaemia (AML) cell lines147,148, and liver in some cancer cells130,131. The relevance of these data cells overexpressing cyclin D1 are particularly sensitive to the therapeutic use of CDK inhibitors can be ques- to loss or inhibition of CDK2 (REF. 149). Thus, it is unclear tioned, as the knockdown or deletion of CDKs will have whether greater selectivity is preferable to more gener- consequences that are distinct from those that arise alized CDK inhibition130,131, nor is it known whether the from acute inhibition of kinase activity. Conversely, efficacy of second-generation CDK inhibitors in pre- the dependence of ERBB2 and other oncogenes on the clinical models will translate to human studies. The first ability of cyclin D to activate CDK4 and CDK6, as sum- Phase I trials of CDK4‑specific inhibitors do, however, marized elsewhere in this Review, provides support for provide preliminary evidence of antitumour activity144–146. the idea that CDK4/CDK6 inhibitors may be effective The CDK inhibitors currently undergoing clini- therapeutics. Again, whether these results imply that cal testing are ATP competitors that block the active established tumours that display CCND1 amplification site of the kinase. This is only one means of impairing or cyclin D1 overexpression, particularly in the context the activity of these kinases, but the design of inhibitors of ERBB2 amplification, are also dependent on CDK4/ that target kinase substrates and regulatory binding sites CDK6 activity can be questioned. Although functional is notoriously challenging. However, peptides that tar- mammary glands develop in the absence of CDK4/ get cyclin A–CDK2 by mimicking substrate binding or CDK6 activity, they lack the specific mammary pro- p21 binding, by locking the kinase in an inactive con- genitor population that is targeted by ERBB2, pointing formation, or by blocking cyclin–CDK association have to a developmental defect rather than a cell-intrinsic been identified150–155. It may also be possible to identify mechanism for the dependence of ERBB2‑driven mam- low-molecular-mass inhibitors of cyclin–CDK substrate mary oncogenesis on the CDK activation function of interactions using high-throughput screens that are

Table 2 | Selected second-generation CDK4/CDK6 inhibitors

Compound Primary targets (IC50) Clinical trials* Preclinical and clinical data (Company)

BAY1000394 CDK1–cyclin B (7 nM) Phase I: advanced malignancies (NCT01188252) Inhibited proliferation with a mean IC50 of 16 nM (Bayer) CDK2–cyclin E (9 nM) (8–37 nM). Activity independent of functional CDK9–cyclin T1 (<10 nM) p53 or RB. Reduced phosphorylation of RB CDK4–cyclin D1 (11 nM) in vitro and in xenografts, indicating intracellular inhibition of CDK2 and CDK4. Reduced RB phosphorylation in paclitaxel- and cisplatin- refractory xenografts186 P1446‑05 CDK1–cyclin B (n/a) • Phase I: advanced refractory malignancies No published preclinical data (Piramal CDK4–cyclin D1 (n/a) (NCT00840190) Healthcare) CDK9–cyclin T (n/a) • Phase I: advanced refractory malignancies (NCT00772876) PD0332991 CDK4–cyclin D3 (9 nM) • Phase I: advanced cancer (NCT00141297) • Effective in RB-positive carcinoma cell lines (Pfizer) CDK4–cyclin D1 (11 nM) • Phase I: previously treated MCL (NCT00420056) and xenograft models134, and MCL in vitro135 CDK6–cyclin D2 (15 nM) • Phase II: advanced or metastatic liposarcoma • Principal and dose-limiting clinical toxicity is CDK2–cyclin E2 (>10 μM) (NCT01209598) myelosuppression145,146 CDK2–cyclin A (>10 μM) • Phase II: recurrent RB-positive glioblastoma • Preferentially inhibits the proliferation of CDK1–cyclin B (>10 μM) (NCT01227434) luminal ER‑positive human breast cancer CDK5–p25 (>10 μM) 134 • Phase II: refractory solid tumours (NCT01037790) cell lines in vitro. Synergizes with tamoxifen • Phase I: PD0332991 plus bortezomib in relapsed and trastuzumab, and increases tamoxifen MCL (NCT01111188) sensitivity in resistant cells136 • Phase I/II: PD0332991 in combination with • In combination with bortezomib, increases bortezomib and dexamethasone in refractory tumour suppression and improves survival in multiple myeloma (NCT00555906) myeloma cell lines169 • Phase I/II: letrozole plus PD0332991 and • In combination with dexamethasone, letrozole first-line treatment of ER-positive, enhances the killing of myeloma cells170 ERBB2-negative advanced breast cancer in postmenopausal women (NCT00721409) R547 CDK4–cyclin D (n/a) Phase I: advanced solid tumours (NCT00400296) • Growth inhibitory activity in vitro and in (Hoffman-Roche) CDK2–cyclin A (0.1 nM) xenograft models187 CDK5–p35 (0.1 nM) • In Phase I, adverse effects were mild and CDK1–cyclin B (0.2 nM) manageable and included nausea, fatigue, CDK2–cyclin E (0.4 nM) emesis, headache and hypotension144 CDK6–cyclin D3 (4 nM) CDK7–cyclin H (171 nM) GSK3α (46 nM) GSK3β (260 nM)187 Combination therapy RGB‑286638 CDK1–cyclin B (<5 nM) Phase I: relapsed or refractory haematological Cytotoxic in conventional drug-sensitive and (GPC Biotech/ CDK2–cyclin A (<5 nM) malignancies (NCT01168882) resistant multiple myeloma cell lines, as well as Agennix) CDK9–cyclin T (<5 nM) primary cultures of multiple myeloma188 CDK4–cyclin D (44 nM) CDK6–cyclin D (55 nM) GSK3β, SRC, MEK and JNK (<100 nM) ZK304709 CDK2–cyclin E (4 nM) Phase I: advanced solid tumours190 Combined inhibition of cell cycle and (Schering CDK9–cyclin T1 (5 nM) angiogenesis resulted in superior efficacy Pharma AG) CDK1–cyclin B (50 nM) compared wiht standard chemotherapeutic CDK4–cyclin D1 (61 nM) compounds in human tumour xenografts, as CDK7–cyclin H (85 nM) well as orthotopic human pancreatic carcinoma VEGFR1 (10 nM) models189 VEGFR2 (34 nM) VEGFR3 (1 nM) PDGFRβ (27 nM)189 CDK, cyclin-dependent kinase; ER, oestrogen receptor; GSK3β, glycogen synthase kinase 3β; JNK, JUN N‑terminal kinase; MCL, mantle cell lymphoma; n/a, not applicable; PDGFRβ, platelet-derived growth factor receptor-β; VEGFR, vascular endothelial growth factor receptor. *Information from the ClinicalTrials.gov website (see Further information)

similar to those used for the discovery and preclinical interactions. Conversely, mimetics of the domain of cyclin testing of inhibitors of MDM2–p53 binding (nutlins)156. D1 involved in AR repression (BOX 1) have been proposed The resolution of the cyclin D1–CDK4 and cyclin D3– as a therapeutic approach in androgen-dependent pros- CDK4 crystal structures157,158 will facilitate the devel- tate cancer159. However, the potent transforming ability of opment of non-ATP competitive inhibitors of CDK4/ cyclin D1b, which does not bind ERα and has altered AR CDK6, and may also offer the possibility of targeting non- interactions, suggests that caution should be exercised in catalytic functions of cyclin D1, blocking protein–protein selectively targeting these interactions.

Phase I and II clinical trials Cyclin D1 as a therapeutic target. The multiplicity of combinations are currently in Phase I and Phase II clini- The first stages of clinical testing cyclin D effects on cancer cell biology, and evidence for cal trials (TABLE 2). Similarly, a CDK4/CDK6 inhibitor in humans. Phase I trials include their CDK-independent actions on end points such as enhanced the activity of an FLT3 inhibitor in AML cell tests of safety, tolerability, and cell migration and the DNA damage response, provide an lines that expressed a mutant form of the FLT3 recep- pharmacokinetics; Phase II 148 trials begin to assess efficacy. impetus for targeting cyclin D rather than, or in addition tor tyrosine kinase , and acted synergistically with the to, CDK4/CDK6 activity. However, although cyclin D1 BCR-ABL kinase inhibitor imatinib in leukaemia cell has been identified as a target for cell-based immuno- lines171. Although it has been suggested that CDK4/ therapy in MCL160, most currently feasible approaches to CDK6 inhibitors could be used to induce reversible qui- inhibiting D‑type cyclins are less direct (FIG. 4), although escence, thereby protecting normal cells from radiation- potentially effective nonetheless. In a well-studied exam- induced toxicity172, CDK4/CDK6 inhibition enhanced ple, the RXR activator bexarotene enhances the effects the effects of radiotherapy in gliobastoma xenografts137. of the EGFR inhibitor erlotinib in lung cancer, and There is evidence that resistance to therapies directed this is thought to be due to the cooperative repression at ER signalling, ERBB2, EGFR and BRAF is accompa- of cyclin D1 expression161,162. Cyclin D1 expression is a nied by increased cyclin D1 expression121–123, suggesting biomarker of therapeutic response to this combination, that the addition of a cyclin D‑targeted therapy might which is currently undergoing further clinical testing reduce therapeutic resistance. Consistent with this after promising results from Phase I and Phase II clinical idea, the CDK4/CDK6 inhibitor PD0332991 was syn- trials162–164. Several other approaches show potential but ergistic with both the anti-oestrogen tamoxifen and the are less well developed. The translation of CCND1 mRNA ERBB2‑targeted therapy trastuzumab in ER‑positive is mTOR-dependent5, suggesting that mTOR inhibitors breast cancer cell lines and was also effective in might inhibit cell cycle progression partly through effects anti-oestrogen-resistant cell lines136. A Phase I/II trial of on cyclin D1 abundance. In Phase II clinical trials mTOR the combination of letrozole, which like tamoxifen targets inhibitors were particularly effective in MCL, which is ER signalling, with PD0332991 is in progress (TABLE 2). characterized by almost universal overexpression of cyclin Finally, consistent with the CDK-independent role of D1, although this may not be accompanied by a reduction cyclin D1 in homologous recombination, depletion in cyclin D1 expression (reviewed in REF. 165). Diverse of cyclin D1 enhanced sensitivity to radiation treatment, compounds, some with potential therapeutic application, and to inhibition of poly(ADP ribose) polymerase 1 lead to enhanced cyclin D1 degradation166. Promoting (PARP1), on which cells that are deficient in homologous cyclin D1 degradation by knockdown of USP2, a deu- recombination become dependent44. Although these biquitylating enzyme that specifically targets cyclin D1, studies are preliminary, they do suggest some treatment inhibited proliferation in cancer cells overexpressing combinations that are worthy of further investigation. cyclin D1 but not in normal fibroblasts, suggesting that targeting cyclin D1 degradation through this mechanism Patient selection. Whole-genome transcript profiling and could be an effective, cancer-specific therapy167. sequencing efforts have documented numerous distinct molecular phenotypes, often accounting for <10% of a par- Cyclin D–CDK inhibition in combination therapy. To ticular cancer173. Consequently, it is increasingly apparent date, CDK inhibitors have had limited success when used that testing novel targeted therapeutic strategies in unse- as single agents, but they may find more clinical use when lected patients may underestimate efficacy, and that many combined with other drugs. In addition, it may be more candidate therapeutics, the development of which was effective to simultaneously target multiple functions of halted because of an apparent lack of efficacy, could poten- cyclin D1, either by using combinations of therapies each tially be reassessed, revived and used effectively if respon- targeting a specific function or by targeting multiple path- sive subgroups could be identified173. This may well be the ways using a single molecule. Examples of multiple case for CDK inhibitors, for which essentially all previ- pathway targeting, such as combining CDK inhibition ous and ongoing trials are in unselected patients. Recent with VEGFR inhibition to target angiogenesis as well as preclinical studies have begun to address this issue, by proliferation, are now in Phase I clinical trials (TABLE 2). examining the relationship between cyclin D1 expression, Several preclinical studies have indicated that pan- RB pathway inactivation and response to CDK4/CDK6 CDK inhibitors function in synergy with cytotoxic drugs inhibition, and by undertaking more global analyses (cisplatin, 5‑fluorouracil, doxorubicin and pacilitaxel) of genes that are differentially expressed in sensitive especially when the cytotoxic agent is administered and resistant cell lines136,138,139. CDK4/CDK6 inhibitors first168. This suggests that CDK inhibitors may be more are generally ineffective in cells lacking RB133,134,136,137,139. effective when cells are synchronized or arrested in spe- However, even when there is no apparent dysfunction cific cell phases, and that one limitation of previous clini- of RB itself, response to CDK4/CDK6 inhibition can be cal studies might have been poor pharmacokinetics, so variable. In a panel of breast cancer cell lines there was that the effective dose was not maintained long enough a strong correlation between sensitivity to CDK4/CDK6 to allow a majority of the target cell population to enter inhibition and the luminal, ER‑positive phenotype, and the sensitive phase of the cell cycle. with high expression of cyclin D1 and RB but low expres- CDK4/CDK6 inhibitors may also lead to enhanced sion of INK4A136. Sensitivity to CDK4/CDK6 inhibition is efficacy when used in combination with other therapies also correlated with high expression of RB and low expres- — for example, the proteasome inhibitor bortezomib or sion of INK4A in ovarian cancer139, as well as deletion of dexamethasone in multiple myeloma169,170 — and these both INK4A and INK4C in glioblastoma138. However, in

ovarian cancer cell lines there was no relationship between CDK4/CDK6 activity alone may only be partially effec- CDK4/CDK6 inhibitor sensitivity and expression of any tive in cyclin D1‑dependent cancers. However, much of the D-type cyclins, CDK4 or CDK6 (REF. 139), and sen- work still needs to be done to establish whether cyc- sitivity was not correlated with CDK6 or CDK4 expres- lin D1 itself can be effectively targeted, and if so, whether sion in glioblastoma138. These observations raise questions this will be more useful therapeutically than inhibiting about whether the measurement of cyclin D levels is likely CDK4/CDK6 activity. to be a generally useful biomarker of response to CDK4/ The most effective use of potential therapies directed CDK6 inhibitors, although theses observations do begin towards cyclin D1 or CDK4/CDK6 will rely on improved to identify a biomarker profile that can be used to better patient selection on the basis of genomic and/or pro- direct these compounds to patients who are most likely teomic signatures of cyclin D1 and/or CDK4/CDK6 to respond. dependence, as well as on the parallel development of Cancers displaying activation of specific oncogenic biomarkers of therapeutic response. Patient subgroups pathways may also be particularly sensitive to CDK4/ that are particularly likely to benefit will also include CDK6 inhibition. The evidence for a specific dependence those in which resistance to cytotoxics and targeted of ERBB2‑driven carcinogenesis on CDK4/CDK6 activity, therapies is commonly accompanied by increased and the sensitivity of ten of 16 ERBB2‑amplified breast cyclin D1 and/or CDK4/CDK6 activity. Important ques- cancer cell lines to CDK4/CDK6 inhibition, suggests that tions include whether cancers are more or less addicted breast cancers overexpressing ERBB2 may be effectively to cyclin D1 overexpression as a result of CCND1 trans- targeted using CDK4/CDK6 inhibitors, possibly in com- location, amplification or mutation, compared with cyc- bination with specific ERBB2‑targeted therapies, such as lin D1 overexpression secondary to another oncogenic trastuzumab136. KRAS-driven non-small-cell lung cancer event. As amplification of 11q13 can involve amplicons is particularly dependent on CDK4, but not on CDK2 in addition to the one harbouring CCND1, the effect of (REF. 174), and is also sensitive to the cyclin D1‑degrading co-amplification of other oncogenes and overexpression combination of bexarotene and erlotinib164, although of neighbouring genes is also worthy of continued inves- KRAS-mutant cancers usually respond poorly to erlo- tigation. Furthermore, given that the use of most targeted tinib175. Thus, KRAS-mutant lung cancer is another therapies is limited by the development of resistance, it specific disease subtype that is resistant to other targeted will be important to better understand mechanisms of therapies, but which may be responsive to the inhibition of resistance to inhibition of cyclin D or CDK4/CDK6: for the cyclin D1–CDK4/CDK6 pathway in the clinic. example, the identification of CDKs that compensate for CDK4/CDK6 could help to further tailor the com- Conclusions bination of CDKs to be targeted. Similarly, a better Several decades of work have established that the dereg- understanding of the degree to which the deregulation ulation of the cyclin D–RB–E2F pathway is central of cyclin D or CDK4/CDK6 contributes to resistance to to the development of most human cancers, with ampli- other targeted therapies could allow the rational design of fication, mutation and overexpression of cyclin D a therapeutic combinations that might minimize the devel- major contributor. Cyclin D1 has both catalytic and opment of resistance or that could be useful in resistant non-catalytic roles that are important in both normal disease. The first steps have been taken towards these and neoplastic cells, with the implication that targeting goals, but much remains to be achieved.

1. Sherr, C. J. & Roberts, J. M. Living with or without 10. Lukas, J. et al. DNA tumor virus oncoproteins and 18. Wang, C. et al. Cyclin D1 repression of nuclear cyclins and cyclin-dependent kinases. Genes Dev. 18, retinoblastoma gene mutations share the ability to respiratory factor 1 integrates nuclear DNA synthesis 2699–2711 (2004). relieve the cell’s requirement for cyclin D1 function in and mitochondrial function. Proc. Natl Acad. Sci. USA 2. Besson, A., Dowdy, S. F. & Roberts, J. M. CDK G1. J. Cell Biol. 125, 625–638 (1994). 103, 11567–11572 (2006). inhibitors: cell cycle regulators and beyond. Dev. Cell 11. Matsuura, I. et al. Cyclin-dependent kinases regulate 19. Zacharek, S. J., Xiong, Y. & Shumway, S. D. 14, 159–169 (2008). the antiproliferative function of Smads. Nature 430, Negative regulation of TSC1‑TSC2 by mammalian 3. Matsushime, H. et al. Identification and properties of 226–231 (2004). D‑type cyclins. Cancer Res. 65, 11354–11360 an atypical catalytic subunit (p34PSK–J3/cdk4) for 12. Shen, R. et al. Cyclin D1‑cdk4 induce runx2 (2005). mammalian D type G1 cyclins. Cell 71, 323–334 ubiquitination and degradation. J. Biol. Chem. 281, 20. Aggarwal, P. et al. Nuclear cyclin D1/CDK4 kinase (1992). 16347–16353 (2006). regulates CUL4 expression and triggers neoplastic

4. Meyerson, M. & Harlow, E. Identification of a G1 13. Zhang, L., Fried, F. B., Guo, H. & Friedman, A. D. growth via activation of the PRMT5 methyltransferase. kinase activity for cdk6, a novel cyclin D partner. Mol. Cyclin-dependent kinase phosphorylation of RUNX1/ Cancer Cell 18, 329–340 (2010). Cell. Biol. 14, 2077–2086 (1994). AML1 on 3 sites increases transactivation potency 21. Ren, J. et al. Methylation of ribosomal protein S10 by 5. Musgrove, E. A. Cyclins: roles in mitogenic signaling and stimulates cell proliferation. Blood 111, protein-arginine methyltransferase 5 regulates and oncogenic transformation. Growth Factors 24, 1193–1200 (2008). ribosome biogenesis. J. Biol. Chem. 285, 13–19 (2006). 14. Nakajima, K. et al. Coordinated regulation of 12695–12705 (2010). 6. Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: differentiation and proliferation of embryonic 22. Sarcevic, B., Lilischkis, R. & Sutherland, R. L. the next generation. Cell 144, 646–674 (2011). cardiomyocytes by a jumonji (Jarid2)-cyclin D1 Differential phosphorylation of T‑47D human breast 7. Malumbres, M. & Barbacid, M. Cell cycle, CDKs and pathway. Development 138, 1771–1782 (2011). cancer cell substrates by D1‑, D3‑, E‑, and A‑cyclin‑CDK cancer: a changing paradigm. Nature Rev. Cancer 9, 15. Lazaro, J. B., Bailey, P. J. & Lassar, A. B. Cyclin D‑cdk4 complexes. J. Biol. Chem. 272, 33327–33337 (1997). 153–166 (2009). activity modulates the subnuclear localization and 23. Baker, G. L., Landis, M. W. & Hinds, P. W. Multiple References 6 and 7 are comprehensive reviews that interaction of MEF2 with SRC-family coactivators functions of D‑type cyclins can antagonize pRb- place aberrations in cell cycle control within the during skeletal muscle differentiation. Genes Dev. 16, mediated suppression of proliferation. Cell Cycle 4, wider context of oncogenesis. 1792–1805 (2002). 330–338 (2005). 8. Gil, J. & Peters, G. Regulation of the 16. Kehn, K. et al. Functional consequences of cyclin D1/ 24. Zhong, Z. et al. Cyclin D1/cyclin-dependent kinase 4 INK4b‑ARF‑INK4a tumour suppressor locus: all for BRCA1 interaction in breast cancer cells. Oncogene interacts with filamin A and affects the migration and one or one for all. Nature Rev. Mol. Cell. Biol. 7, 26, 5060–5069 (2007). invasion potential of breast cancer cells. Cancer Res. 667–677 (2006). 17. Adon, A. M. et al. Cdk2 and Cdk4 regulate the 70, 2105–2114 (2010). 9. Chen, H. Z., Tsai, S. Y. & Leone, G. Emerging roles of centrosome cycle and are critical mediators of 25. Neumeister, P. et al. Cyclin D1 governs adhesion and E2Fs in cancer: an exit from cell cycle control. Nature centrosome amplification in p53‑null cells. Mol. Cell. motility of macrophages. Mol. Biol. Cell 14, Rev. Cancer 9, 785–797 (2009). Biol. 30, 694–710 (2010). 2005–2015 (2003).

This publication showed that cyclin D1 was not 49. Landis, M. W., Pawlyk, B. S., Li, T., Sicinski, P. & Hinds, 74. Santarius, T., Shipley, J., Brewer, D., Stratton, M. R. & solely a regulator of cell proliferation, but also had P. W. Cyclin D1‑dependent kinase activity in murine Cooper, C. S. A census of amplified and overexpressed effects on cellular migration. development and mammary tumorigenesis. Cancer human cancer genes. Nature Rev. Cancer 10, 59–64 26. Li, Z. et al. Cyclin D1 regulates cellular migration Cell 9, 13–22 (2006). (2010). through the inhibition of thrombospondin 1 and ROCK 50. Geng, Y. et al. Deletion of the p27Kip1 gene restores 75. Wang, T. C. et al. Mammary hyperplasia and signaling. Mol. Cell. Biol. 26, 4240–4256 (2006). normal development in cyclin D1‑deficient mice. Proc. carcinoma in MMTV-cyclin D1 transgenic mice. Nature 27. Li, Z. et al. Cyclin D1 induction of cellular migration Natl Acad. Sci. USA 98, 194–199 (2001). 369, 669–671 (1994). requires p27(KIP1). Cancer Res. 66, 9986–9994 51. Tsutsui, T. et al. Targeted disruption of CDK4 delays This paper shows that cyclin D1 overexpression in (2006). cell cycle entry with enhanced p27Kip1 activity. Mol. the mammary gland is sufficient for tumour 28. Chow, Y. H. et al. Role of Cdk4 in lymphocyte function Cell. Biol. 19, 7011–7019 (1999). formation, the first experimental evidence for its and allergen response. Cell Cycle 9, 4922–4930 52. Yasui, M. et al. Antisense to cyclin D1 inhibits vascular oncogenic capacity in vivo. (2010). endothelial growth factor-stimulated growth of vascular 76. Bertoni, F., Rinaldi, A., Zucca, E. & Cavalli, F. Update 29. Bienvenu, F. et al. Transcriptional role of cyclin D1 in endothelial cells: implication of tumor vascularization. on the molecular biology of mantle cell lymphoma. development revealed by a genetic-proteomic screen. Clin. Cancer Res. 12, 4720–4729 (2006). Hematol. Oncol. 24, 22–27 (2006). Nature 463, 374–378 (2010). 53. Nelsen, C. J. et al. Short term cyclin D1 77. The Cancer Genome Atlas Research Network. This publication conclusively shows cyclin D1 overexpression induces centrosome amplification, Comprehensive genomic characterization defines binding to DNA during normal mouse development, mitotic spindle abnormalities, and aneuploidy. J. Biol. human glioblastoma genes and core pathways. Nature underlining the importance of its effects as a Chem. 280, 768–776 (2005). 455, 1061–1068 (2008). transcriptional regulator. 54. Pontano, L. L. et al. Genotoxic stress-induced cyclin 78. Beroukhim, R. et al. The landscape of somatic copy- 30. McMahon, C., Suthiphongchai, T., DiRenzo, J. & Ewen, D1 phosphorylation and proteolysis are required for number alteration across human cancers. Nature 463, M. E. P/CAF associates with cyclin D1 and potentiates genomic stability. Mol. Cell. Biol. 28, 7245–7258 899–905 (2010). its activation of the estrogen receptor. Proc. Natl (2008). 79. Matsubayashi, H. et al. Methylation of cyclin D2 is Acad. Sci. USA 96, 5382–5387 (1999). 55. Shields, B. J., Hauser, C., Bukczynska, P. E., Court, observed frequently in pancreatic cancer but is also an 31. Reutens, A. T. et al. Cyclin D1 binds the androgen N. W. & Tiganis, T. DNA replication stalling attenuates age-related phenomenon in gastrointestinal tissues. receptor and regulates hormone-dependent signaling tyrosine kinase signaling to suppress S phase Clin. Cancer Res. 9, 1446–1452 (2003). in a p300/CBP-associated factor (P/CAF)-dependent progression. Cancer Cell 14, 166–179 (2008). 80. Evron, E. et al. Loss of cyclin D2 expression in the manner. Mol. Endocrinol. 15, 797–811 (2001). 56. Aggarwal, P. et al. Nuclear accumulation of cyclin D1 majority of breast cancers is associated with promoter 32. Fu, M. et al. Cyclin D1 inhibits peroxisome proliferator- during S phase inhibits Cul4‑dependent Cdt1 hypermethylation. Cancer Res. 61, 2782–2787 activated receptor γ-mediated adipogenesis through proteolysis and triggers p53‑dependent DNA (2001). histone deacetylase recruitment. J. Biol. Chem. 280, rereplication. Genes Dev. 21, 2908–2922 (2007). 81. Padar, A. et al. Inactivation of cyclin D2 gene in 16934–16941 (2005). 57. Yu, Q., Geng, Y. & Sicinski, P. Specific protection prostate cancers by aberrant promoter methylation. 33. Coqueret, O. Linking cyclins to transcriptional control. against breast cancers by cyclin D1 ablation. Nature Clin. Cancer Res. 9, 4730–4734 (2003). Gene 299, 35–55 (2002). 411, 1017–1021 (2001). 82. Wiestner, A. et al. Point mutations and genomic 34. Fu, M., Wang, C., Li, Z., Sakamaki, T. & Pestell, R. G. This paper shows that cyclin D1 is essential for deletions in CCND1 create stable truncated cyclin D1 Minireview: cyclin D1: normal and abnormal functions. some, but not all, oncogenic pathways, and was mRNAs that are associated with increased Endocrinology 145, 5439–5447 (2004). followed by a series of publications that together proliferation rate and shorter survival. Blood 109, 35. Mullany, L. K. et al. Distinct proliferative and showed that ERRB2‑driven mammary oncogenesis 4599–4606 (2007). transcriptional effects of the D‑type cyclins in vivo. Cell required the ability of cyclin D1 to activate CDK4. 83. Benzeno, S. et al. Identification of mutations that Cycle 7, 2215–2224 (2008). 58. Yu, Q. et al. Requirement for CDK4 kinase function in disrupt phosphorylation-dependent nuclear export of 36. Neuman, E. et al. Cyclin D1 stimulation of estrogen breast cancer. Cancer Cell 9, 23–32 (2006). cyclin D1. Oncogene 25, 6291–6303 (2006). receptor transcriptional activity independent of cdk4. 59. Yang, C. et al. The role of the cyclin D1‑dependent 84. Moreno-Bueno, G. et al. Cyclin D1 gene (CCND1) Mol. Cell. Biol. 17, 5338–5347 (1997). kinases in ErbB2‑mediated breast cancer. Am. mutations in endometrial cancer. Oncogene 22, 37. Zwijsen, R. M. L. et al. CDK-independent activation of J. Pathol. 164, 1031–1038 (2004). 6115–6118 (2003). estrogen receptor by cyclin D1. Cell 88, 405–415 60. Hu, M. G. et al. A requirement for cyclin-dependent 85. Barbash, O. et al. Mutations in Fbx4 inhibit (1997). kinase 6 in thymocyte development and dimerization of the SCF(Fbx4) ligase and contribute to 38. Despouy, G. et al. Cyclin D3 is a cofactor of retinoic tumorigenesis. Cancer Res. 69, 810–818 (2009). cyclin D1 overexpression in human cancer. Cancer Cell acid receptors, modulating their activity in the 61. Sicinska, E. et al. Requirement for cyclin D3 in 14, 68–78 (2008). presence of cellular retinoic acid-binding protein II. lymphocyte development and T cell leukemias. Cancer 86. Russell, A. et al. Cyclin D1 and D3 associate with the J. Biol. Chem. 278, 6355–6362 (2003). Cell 4, 451–461 (2003). SCF complex and are coordinately elevated in breast 39. Sarruf, D. A. et al. Cyclin D3 promotes adipogenesis 62. Berthet, C. & Kaldis, P. Cell-specific responses to loss cancer. Oncogene 18, 1983–1991 (1999). through activation of peroxisome proliferator- of cyclin-dependent kinases. Oncogene 26, 87. Pabalan, N. et al. Cyclin D1 Pro241Pro activated receptor γ. Mol. Cell. Biol. 25, 9985–9995 4469–4477 (2007). (CCND1‑G870A) polymorphism is associated with (2005). 63. Cole, A. M. et al. Cyclin D2‑cyclin‑dependent kinase increased cancer risk in human populations: a meta- 40. Knudsen, K. E., Cavenee, W. K. & Arden, K. C. D‑type 4/6 is required for efficient proliferation and analysis. Cancer Epidemiol. Biomarkers Prev. 17, cyclins complex with the androgen receptor and inhibit tumorigenesis following Apc loss. Cancer Res. 70, 2773–2781 (2008). its transcriptional transactivation ability. Cancer Res. 8149–8158 (2010). 88. Li, R. et al. Expression of cyclin D1 splice variants is 59, 2297–2301 (1999). 64. Kim, J. K. & Diehl, J. A. Nuclear cyclin D1: an differentially associated with outcome in non-small cell References 36, 37 and 40 were the first to oncogenic driver in human cancer. J. Cell. Physiol. lung cancer patients. Hum. Pathol. 39, 1792–1801 document kinase-independent functions of cyclin 220, 292–296 (2009). (2008). D1 in both promoting and inhibiting steroid 65. Gautschi, O., Ratschiller, D., Gugger, M., Betticher, 89. Millar, E. K. et al. Cyclin D1b protein expression in hormone receptor transcriptional activity. D. C. & Heighway, J. Cyclin D1 in non-small cell lung breast cancer is independent of cyclin D1a and 41. Zong, H. et al. Cyclin D3/CDK11p58 complex is cancer: a key driver of malignant transformation. Lung associated with poor disease outcome. Oncogene 28, involved in the repression of androgen receptor. Mol. Cancer 55, 1–14 (2007). 1812–1820 (2009). Cell. Biol. 27, 7125–7142 (2007). 66. Comstock, C. E., Revelo, M. P., Buncher, C. R. & 90. Comstock, C. E. et al. Cyclin D1 splice variants: 42. Sherr, C. J. & Roberts, J. M. CDK inhibitors: positive Knudsen, K. E. Impact of differential cyclin D1 polymorphism, risk, and isoform-specific regulation in

and negative regulators of G1 phase progression. expression and localisation in prostate cancer. Br. prostate cancer. Clin. Cancer Res. 15, 5338–5349 Genes. Dev. 13, 1501–1512 (1999). J. Cancer 96, 970–979 (2007). (2009). 43. Besson, A., Assoian, R. K. & Roberts, J. M. 67. Dhar, K. K. et al. Expression and subcellular 91. Abramson, V. G. et al. Cyclin D1b in human breast Regulation of the cytoskeleton: an oncogenic function localization of cyclin D1 protein in epithelial ovarian carcinoma and coexpression with cyclin D1a is for CDK inhibitors? Nature Rev. Cancer 4, 948–955 tumour cells. Br. J. Cancer 81, 1174–1181 (1999). associated with poor outcome. Anticancer Res. 30, (2004). 68. Ahmed, K. M., Fan, M., Nantajit, D., Cao, N. & Li, J. J. 1279–1285 (2010). 44. Jirawatnotai, S. et al. A function for cyclin D1 in DNA Cyclin D1 in low-dose radiation-induced adaptive 92. Sanchez, G., Delattre, O., Auboeuf, D. & Dutertre, M. repair uncovered by protein interactome analyses in resistance. Oncogene 27, 6738–6748 (2008). Coupled alteration of transcription and splicing by a human cancers. Nature 474, 230–234 (2011). 69. Knudsen, K. E., Diehl, J. A., Haiman, C. A. & Knudsen, single oncogene: boosting the effect on cyclin D1 45. Li, Z. et al. Alternative cyclin D1 splice forms E. S. Cyclin D1: polymorphism, aberrant splicing and activity. Cell Cycle 7, 2299–2305 (2008). differentially regulate the DNA damage response. cancer risk. Oncogene 25, 1620–1628 (2006). 93. Zeng, X. et al. The Ras oncogene signals centrosome Cancer Res. 70, 8802–8811 (2010). 70. Kim, C. J. et al. Cyclin D1b variant promotes cell amplification in mammary epithelial cells through cyclin References 44 and 45 show that cyclin D1 invasiveness independent of binding to CDK4 in D1/Cdk4 and Nek2. Oncogene 29, 5103–5112 promotes efficient DNA repair, independently of human bladder cancer cells. Mol. Carcinog. 48, (2010). CDK4/CDK6 activity, through binding to RAD51 953–964 (2009). 94. Lee, R. J. et al. Cyclin D1 is required for and BRCA2. 71. Li, Z. et al. Alternate cyclin D1 mRNA splicing transformation by activated Neu and is induced 46. Raderschall, E. et al. Formation of higher-order modulates p27KIP1 binding and cell migration. through an E2F‑dependent signaling pathway. Mol. nuclear Rad51 structures is functionally linked to p21 J. Biol. Chem. 283, 7007–7015 (2008). Cell. Biol. 20, 672–83 (2000). expression and protection from DNA damage-induced 72. Zhu, J., Sen, S., Wei, C. & Frazier, M. L. Cyclin D1b 95. Desai, K. V. et al. Initiating oncogenic event apoptosis. J. Cell Sci. 115, 153–164 (2002). represses breast cancer cell growth by antagonizing the determines gene-expression patterns of human breast 47. Kozar, K. et al. Mouse development and cell action of cyclin D1a on estrogen receptor α-mediated cancer models. Proc. Natl Acad. Sci. USA 99, proliferation in the absence of D‑cyclins. Cell 118, transcription. Int. J. Oncol. 36, 39–48 (2010). 6967–6972 (2002). 477–491 (2004). 73. Burd, C. J. et al. Cyclin D1b variant influences prostate 96. Ahnstrom, M., Nordenskjold, B., Rutqvist, L. E., 48. Malumbres, M. et al. Mammalian cells cycle without cancer growth through aberrant androgen receptor Skoog, L. & Stal, O. Role of cyclin D1 in ErbB2‑positive the D‑type cyclin-dependent kinases Cdk4 and Cdk6. regulation. Proc. Natl Acad. Sci. USA 103, breast cancer and tamoxifen resistance. Breast Cancer Cell 118, 493–504 (2004). 2190–2195 (2006). Res. Treat. 91, 145–151 (2005).

97. Reis-Filho, J. S. et al. Cyclin D1 protein overexpression 120. Kornmann, M. et al. Inhibition of cyclin D1 expression 140. Tan, A. R. et al. Phase I trial of the cyclin-dependent and CCND1 amplification in breast carcinomas: an in human pancreatic cancer cells is associated with kinase inhibitor flavopiridol in combination with immunohistochemical and chromogenic in situ increased chemosensitivity and decreased expression docetaxel in patients with metastatic breast cancer. hybridisation analysis. Mod. Pathol. 19, 999–1009 of multiple chemoresistance genes. Cancer Res. 59, Clin. Cancer Res. 10, 5038–5047 (2004). (2006). 3505–3511 (1999). 141. Haddad, R. I. et al. A phase II clinical and 98. Bandi, N. et al. miR‑15a and miR‑16 are implicated in 121. Musgrove, E. A. & Sutherland, R. L. Biological pharmacodynamic study of E7070 in patients with cell cycle regulation in a Rb-dependent manner and determinants of endocrine resistance in breast cancer. metastatic, recurrent, or refractory squamous cell are frequently deleted or down-regulated in non-small Nature Rev. Cancer 9, 631–643 (2009). carcinoma of the head and neck: modulation of cell lung cancer. Cancer Res. 69, 5553–5559 (2009). 122. Kalish, L. H. et al. Deregulated cyclin D1 expression is retinoblastoma protein phosphorylation by a novel 99. Bonci, D. et al. The miR‑15a‑miR‑16‑1 cluster controls associated with decreased efficacy of the selective chloroindolyl sulfonamide cell cycle inhibitor. Clin. prostate cancer by targeting multiple oncogenic epidermal growth factor receptor tyrosine kinase Cancer Res. 10, 4680–4687 (2004). activities. Nature Med. 14, 1271–1277 (2008). inhibitor gefitinib in head and neck squamous cell 142. Berkofsky-Fessler, W. et al. Preclinical biomarkers for 100. Jiang, Q., Feng, M. G. & Mo, Y. Y. Systematic carcinoma cell lines. Clin. Cancer Res. 10, a cyclin-dependent kinase inhibitor translate to validation of predicted microRNAs for cyclin D1. BMC 7764–7774 (2004). candidate pharmacodynamic biomarkers in phase I Cancer 9, 194 (2009). 123. Smalley, K. S. et al. Increased cyclin D1 expression can patients. Mol. Cancer Ther. 8, 2517–2525 (2009). 101. Ewen, M. E. & Lamb, J. The activities of cyclin D1 that mediate BRAF inhibitor resistance in BRAF 143. Locatelli, G. et al. Transcriptional analysis of an E2F drive tumorigenesis. Trends Mol. Med. 10, 158–162 V600E‑mutated melanomas. Mol. Cancer Ther. 7, gene signature as a biomarker of activity of the cyclin- (2004). 2876–2883 (2008). dependent kinase inhibitor PHA‑793887 in tumor and 102. Arnold, A. & Papanikolaou, A. Cyclin D1 in breast 124. Noel, E. E. et al. The association of CCND1 skin biopsies from a phase I clinical study. Mol. Cancer cancer pathogenesis. J. Clin. Oncol. 23, 4215–4224 overexpression and cisplatin resistance in testicular Ther. 9, 1265–1273 (2010). (2005). germ cell tumors and other cancers. Am. J. Pathol. 144. Diab, S. et al. A phase I study of R547, a novel, 103. Rosenwald, A. et al. The proliferation gene expression 176, 2607–2615 (2010). selective inhibitor of cell cycle and transcriptional signature is a quantitative integrator of oncogenic 125. Shimura, T. et al. Acquired radioresistance of human cyclin dependent kinases (CDKs). ASCO Meeting Abstr. events that predicts survival in mantle cell lymphoma. tumor cells by DNA-PK/AKT/GSK3β‑mediated cyclin 25, 3528 (2007). Cancer Cell 3, 185–197 (2003). D1 overexpression. Oncogene 29, 4826–4837 145. Schwartz, G. K. et al. Phase I study of PD 0332991, This analysis found a correlation between cyclin D1 (2010). a cyclin-dependent kinase inhibitor, administered in expression and a proliferation gene expression 126. Rudas, M. et al. Cyclin D1 expression in breast cancer 3‑week cycles (Schedule 2/1). Br. J. Cancer 104, signature. patients receiving adjuvant tamoxifen-based therapy. 1862–1868 (2011). 104. Jares, P., Colomer, D. & Campo, E. Genetic and Clin. Cancer Res. 14, 1767–1774 (2008). 146. O’Dwyer, P. J. et al. A phase I dose escalation trial of a molecular pathogenesis of mantle cell lymphoma: 127. Stendahl, M. et al. Cyclin D1 overexpression is a daily oral CDK 4/6 inhibitor PD‑0332991. ASCO perspectives for new targeted therapeutics. Nature negative predictive factor for tamoxifen response in Meeting Abstr. 25, 3550 (2007). Rev. Cancer 7, 750–762 (2007). postmenopausal breast cancer patients. Br. J. Cancer 147. Dean, J. L., Thangavel, C., McClendon, A. K., Reed, 105. Bova, R. J. et al. Cyclin D1 and p16INK14A expression 90, 1942–1948 (2004). C. A. & Knudsen, E. S. Therapeutic CDK4/6 inhibition predict reduced survival in carcinoma of the anterior 128. Weinstein, I. B. & Joe, A. K. Mechanisms of disease: in breast cancer: key mechanisms of response and tongue. Clin. Cancer Res. 5, 2810–2819 (1999). oncogene addiction-a rationale for molecular targeting failure. Oncogene 29, 4018–4032 (2010). 106. Lamb, J. et al. A mechanism of cyclin D1 action in cancer therapy. Nature Clin. Pract. Oncol. 3, 148. Wang, L. et al. Pharmacologic inhibition of CDK4/6: encoded in the patterns of gene expression in human 448–457 (2006). mechanistic evidence for selective activity or acquired cancer. Cell 114, 323–334 (2003). 129. Krause, D. S. & Van Etten, R. A. Tyrosine kinases as resistance in acute myeloid leukemia. Blood 110, In contrast to reference 103, by demonstrating targets for cancer therapy. N. Engl. J. Med. 353, 2075–2083 (2007). that cyclin D1 expression was not correlated with 172–187 (2005). 149. Hanse, E. A. et al. Cdk2 plays a critical role in an E2F1 and E2F2‑activated gene signature, this 130. Lapenna, S. & Giordano, A. Cell cycle kinases as hepatocyte cell cycle progression and survival in the paper called into question the idea that the therapeutic targets for cancer. Nature Rev. Drug setting of cyclin D1 expression in vivo. Cell Cycle 8, proliferative effects of cyclin D1 were responsible Discov. 8, 547–566 (2009). 2802–2809 (2009). for its oncogenic action. 131. Shapiro, G. I. Cyclin-dependent kinase pathways as 150. Bagella, L. et al. A small molecule based on the pRb2/ 107. Ertel, A. et al. RB‑pathway disruption in breast cancer: targets for cancer treatment. J. Clin. Oncol. 24, p130 spacer domain leads to inhibition of cdk2 differential association with disease subtypes, disease- 1770–1783 (2006). activity, cell cycle arrest and tumor growth reduction specific prognosis and therapeutic response. Cell Cycle 132. Jeselsohn, R. et al. Cyclin D1 kinase activity is in vivo. Oncogene 26, 1829–1839 (2007). 9, 4153–4163 (2010). required for the self-renewal of mammary stem and 151. Arris, C. E. et al. Identification of novel purine and 108. Agarwal, R. et al. Integrative analysis of cyclin protein progenitor cells that are targets of MMTV‑ErbB2 pyrimidine cyclin-dependent kinase inhibitors with levels identifies cyclin b1 as a classifier and predictor tumorigenesis. Cancer Cell 17, 65–76 (2010). distinct molecular interactions and tumor cell growth of outcomes in breast cancer. Clin. Cancer Res. 15, 133. Fry, D. W. et al. Cell cycle and biochemical effects of PD inhibition profiles. J. Med. Chem. 43, 2797–2804 3654–3662 (2009). 0183812. A potent inhibitor of the cyclin D‑dependent (2000). 109. Sakamaki, T. et al. Cyclin D1 determines mitochondrial kinases Cdk4 and Cdk6. J. Biol. Chem. 276, 152. Gondeau, C. et al. Design of a novel class of peptide function in vivo. Mol. Cell. Biol. 26, 5449–5469 (2006). 16617–16623 (2001). inhibitors of cyclin-dependent kinase/cyclin 110. Yang, C. et al. Identification of cyclin D1- and estrogen- This publication shows that a specific inhibitor of activation. J. Biol. Chem. 280, 13793–13800 regulated genes contributing to breast carcinogenesis CDK4 and CDK6 has anti-proliferative effects, and (2005). and progression. Cancer Res. 66, 11649–11658 was followed by a series of publications 153. Canela, N. et al. Identification of an hexapeptide that (2006). demonstrating antitumour effects in various binds to a surface pocket in cyclin A and inhibits the 111. Sweeney, K. J., Swarbrick, A., Sutherland, R. L. & malignancies. catalytic activity of the complex cyclin-dependent Musgrove, E. A. Lack of relationship between CDK 134. Fry, D. W. et al. Specific inhibition of cyclin-dependent kinase 2‑cyclin A. J. Biol. Chem. 281, 35942–35953

activity and G1 cyclin expression in breast cancer cells. kinase 4/6 by PD 0332991 and associated antitumor (2006). Oncogene 16, 2865–2878 (1998). activity in human tumor xenografts. Mol. Cancer Ther. 154. Adams, P. D. et al. Identification of a cyclin‑cdk2 112. Thomas, G. R., Nadiminti, H. & Regalado, J. Molecular 3, 1427–1438 (2004). recognition motif present in substrates and p21‑like predictors of clinical outcome in patients with head 135. Marzec, M. et al. Mantle cell lymphoma cells express cyclin-dependent kinase inhibitors. Mol. Cell. Biol. 16, and neck squamous cell carcinoma. Int. J. Exp. Pathol. predominantly cyclin D1a isoform and are highly 6623–6633 (1996). 86, 347–363 (2005). sensitive to selective inhibition of CDK4 kinase activity. 155. Ball, K. L., Lain, S., Fahraeus, R., Smythe, C. & 113. Roy, P. G. & Thompson, A. M. Cyclin D1 and breast Blood 108, 1744–1750 (2006). Lane, D. P. Cell-cycle arrest and inhibition of Cdk4 cancer. Breast 15, 718–727 (2006). 136. Finn, R. S. et al. PD 0332991, a selective cyclin D activity by small peptides based on the carboxy- 114. Taneja, P. et al. Classical and novel prognostic markers kinase 4/6 inhibitor, preferentially inhibits proliferation terminal domain of p21WAF1. Curr. Biol. 7, 71–80 for breast cancer and their clinical significance. Clin. of luminal estrogen receptor-positive human breast (1997). Med. Insights Oncol. 4, 15–34 (2010). cancer cell lines in vitro. Breast Cancer Res. 11, R77 156. Vassilev, L. T. et al. In vivo activation of the p53 115. Lukas, J., Bartkova, J., Rodhe, M., Strauss, M. & (2009). pathway by small-molecule antagonists of MDM2.

Bartek, J. Cyclin D1 is dispensible for G1 control in In this study a panel of breast cancer cell lines was Science 303, 844–848 (2004). retinoblastoma gene-deficient cells independently of used to identify a gene expression signature 157. Day, P. J. et al. Crystal structure of human CDK4 in cdk4 activity. Mol. Cell. Biol. 15, 2600–2611 (1995). correlated with response to CDK4/CDK6 inhibition. complex with a D‑type cyclin. Proc. Natl Acad. Sci. 116. Knudsen, E. S. & Knudsen, K. E. Tailoring to RB: It was the first to use a large-scale, unbiased USA 106, 4166–4170 (2009). tumour suppressor status and therapeutic response. approach that aimed to develop criteria for patient 158. Takaki, T. et al. The structure of CDK4/cyclin D3 has Nature Rev. Cancer 8, 714–724 (2008). selection in clinical studies of CDK4/CDK6 implications for models of CDK activation. Proc. Natl 117. Kenny, F. S. et al. Overexpression of cyclin D1 inhibition. Acad. Sci. USA 106, 4171–4176 (2009). messenger RNA predicts for poor prognosis in 137. Michaud, K. et al. Pharmacologic inhibition of cyclin- 159. Schiewer, M. J. et al. Cyclin D1 repressor domain estrogen receptor-positive breast cancer. Clin. Cancer dependent kinases 4 and 6 arrests the growth of mediates proliferation and survival in prostate cancer. Res. 5, 2069–2076 (1999). glioblastoma multiforme intracranial xenografts. Oncogene 28, 1016–1027 (2009). 118. Elsheikh, S. et al. CCND1 amplification and cyclin D1 Cancer Res. 70, 3228–3238 (2010). 160. Wang, M. et al. Cyclin D1 as a universally expressed expression in breast cancer and their relation with 138. Wiedemeyer, W. R. et al. Pattern of retinoblastoma mantle cell lymphoma-associated tumor antigen for proteomic subgroups and patient outcome. Breast pathway inactivation dictates response to CDK4/6 immunotherapy. Leukemia 23, 1320–1328 (2009). Cancer Res. Treat. 109, 325–335 (2008). inhibition in GBM. Proc. Natl Acad. Sci. USA 107, 161. Dragnev, K. H. et al. Bexarotene and erlotinib for 119. Biliran, H. Jr. et al. Overexpression of cyclin D1 11501–11506 (2010). aerodigestive tract cancer. J. Clin. Oncol. 23, promotes tumor cell growth and confers resistance to 139. Konecny, G. E. et al. Expression of p16 and 8757–8764 (2005). cisplatin-mediated apoptosis in an elastase-myc Retinoblastoma determines response to CDK 4/6 162. Kim, E. S., Lee, J. J. & Wistuba, II. Cotargeting cyclin transgene-expressing pancreatic tumor cell line. Clin. inhibition in ovarian cancer. Clin. Cancer Res. 17, D1 starts a new chapter in lung cancer prevention and Cancer Res. 11, 6075–6086 (2005). 1591–1602 (2011). therapy. Cancer Prev. Res. 4, 779–782 (2011).

163. Kim, E. S. et al. The BATTLE trial: personalising This study identifies an interaction between CDKs 185. Bergsagel, P. L. & Kuehl, W. M. Molecular pathogenesis therapy for lung cancer. Cancer Discovery 1, 44–53 and RAS signalling that could be used as a basis and a consequent classification of multiple myeloma. (2011). for the rational design of combination therapies. J. Clin. Oncol. 23, 6333–6338 (2005). 164. Dragnev, K. H. et al. Bexarotene plus erlotinib The synthetic-lethal approach adopted in this 186. Siemeister, G. et al. Pharmacologic profile of the oral suppress lung carcinogenesis independent of study merits wider application, given the tissue novel pan-CDK inhibitor BAY 1000394 in KRAS mutations in two clinical trials and specificity of dependence on individual D‑type chemosensitive and chemorefractory tumor models. transgenic models. Cancer Prev. Res. 4, 818–828 cyclins, CDK4 and CDK6 that is apparent in Cancer Res. Abstr. 70, 3883 (2010). (2011). knockout mice. 187. DePinto, W. et al. In vitro and in vivo activity of R547: 165. Sabatini, D. M. mTOR and cancer: insights into a 175. Roberts, P. J., Stinchcombe, T. E., Der, C. J. & Socinski, a potent and selective cyclin-dependent kinase complex relationship. Nature Rev. Cancer 6, M. A. Personalized medicine in non‑small‑cell lung inhibitor currently in phase I clinical trials. Mol. Cancer 729–734 (2006). cancer: is KRAS a useful marker in selecting patients Ther. 5, 2644–2658 (2006). 166. Alao, J. P. The regulation of cyclin D1 degradation: for epidermal growth factor receptor-targeted 188. Cirstea, D. et al. RGB 286638, a Novel Multi-Targeted roles in cancer development and the potential for therapy? J. Clin. Oncol. 28, 4769–4777 (2010). Small Molecule Inhibitor, Induces Multiple Myeloma therapeutic invention. Mol. Cancer 6, 24 (2007). 176. Zwijsen, R. M., Buckle, R. S., Hijmans, E. M., (MM) Cell Death through Abrogation of 167. Shan, J., Zhao, W. & Gu, W. Suppression of cancer cell Loomans, C. J. & Bernards, R. Ligand-independent CDKDependent and Independent Survival growth by promoting cyclin D1 degradation. Mol. Cell recruitment of steroid receptor coactivators to Mechanisms. ASH Annual Meeting Abstr. 112, 2759 36, 469–476 (2009). estrogen receptor by cyclin D1. Genes Dev. 12, (2008). This study identifies USP2 as a specific 3488–3498 (1998). 189. Siemeister, G. et al. Molecular and pharmacodynamic deubiquitylase for cyclin D1, and suggests that 177. Siegert, J. L., Rushton, J. J., Sellers, W. R., Kaelin, characteristics of the novel multi-target tumor growth targeting it may be an effective therapy in cyclin W. G. Jr. & Robbins, P. D. Cyclin D1 suppresses inhibitor ZK 304709. Biomed. Pharmacother. 60, D1‑dependent cancers. retinoblastoma protein-mediated inhibition of 269–272 (2006). 168. Kelland, L. R. in Inhibitors of Cyclin-Dependent TAFII250 kinase activity. Oncogene 19, 5703–5711 190. Scott, E. N. et al. A phase I dose escalation study of Kinases as Anti-Tumor Agents (eds Smith, P. J. & Yue, (2000). the pharmacokinetics and tolerability of ZK 304709, E. W.) 371–388 (CRC Press, 2007). 178. Hardisson, D. Molecular pathogenesis of head and an oral multi-targeted growth inhibitor (MTGI), in 169. Menu, E. et al. A novel therapeutic combination using neck squamous cell carcinoma. Eur. Arch. patients with advanced solid tumours. Cancer PD 0332991 and bortezomib: study in the 5T33MM Otorhinolaryngol. 260, 502–508 (2003). Chemother. Pharmacol. 64, 425–429 (2009). myeloma model. Cancer Res. 68, 5519–5523 179. Moreno-Bueno, G. et al. Molecular alterations (2008). associated with cyclin D1 overexpression in Acknowledgements 170. Baughn, L. B. et al. A novel orally active small endometrial cancer. Int. J. Cancer 110, 194–200 The authors are grateful to C. M. McNeil and C. M. Sergio for molecule potently induces G1 arrest in primary (2004). assistance with literature searches, A. V. Biankin for thought- myeloma cells and prevents tumor growth by specific 180. Wu, W. et al. Correlation of cyclin D1 and cyclin D3 provoking discussions and I. Rooman for helpful comments. inhibition of cyclin-dependent kinase 4/6. Cancer Res. overexpression with the loss of PTEN expression in The authors’ research is supported by the National Health 66, 7661–7667 (2006). endometrial carcinoma. Int. J. Gynecol. Cancer 16, and Medical Research Council of Australia, Cancer Institute 171. Kuo, T. C., Chavarria-Smith, J. E., Huang, D. & 1668–1672 (2006). New South Wales, National Breast Cancer Foundation, Cure Schlissel, M. S. Forced expression of CDK6 confers 181. Li, W. et al. The role of cell cycle regulatory proteins in Cancer Australia Foundation, the Australian Cancer Research resistance of pro‑B ALL to Gleevec treatment. Mol. the pathogenesis of melanoma. Pathology 38, Foundation, the Petre Foundation, Young Garvan and the RT Cell. Biol. 31, 2566–2576 (2011). 287–301 (2006). Hall Trust. 172. Johnson, S. M. et al. Mitigation of hematologic 182. Garcea, G., Neal, C. P., Pattenden, C. J., Steward, W. P. radiation toxicity in mice through pharmacological & Berry, D. P. Molecular prognostic markers in Competing interests statement quiescence induced by CDK4/6 inhibition. J. Clin. pancreatic cancer: a systematic review. Eur. J. Cancer The authors declare no competing financial interests. Invest. 120, 2528–2536 (2010). 41, 2213–2236 (2005). 173. Biankin, A. V. & Hudson, T. J. Somatic variation and 183. Toncheva, D. et al. Tissue microarray analysis of cyclin cancer: therapies lost in the mix. Hum. Genet. 5 Jun D1 gene amplification and gain in colorectal FURTHER INFORMATION 2011 (doi:10.1007/s00439‑011‑1010‑0). carcinomas. Tumour Biol. 25, 157–160 (2004). Robert L. Sutherland’s homepage: http://www.garvan.org. 174. Puyol, M. et al. A synthetic lethal interaction between 184. McKay, J. A. et al. Cyclin D1 protein expression and au/about-us/our-people/professor-rob-sutherland K‑Ras oncogenes and Cdk4 unveils a therapeutic gene polymorphism in colorectal cancer. Aberdeen ClinicalTrials.gov: http://clinicaltrials.gov/ strategy for non-small cell lung carcinoma. Cancer Cell Colorectal Initiative. Int. J. Cancer 88, 77–81 ALL LINKS ARE ACTIVE IN THE ONLINE PDF 18, 63–73 (2010). (2000).