Introduction
Sustained proliferative capacity is a hallmark of cancer [
1]. The cell cycle is the process by which mammalian cells regulate proliferation and has 4 functional phases: S phase when DNA replication occurs; M phase (mitosis): when DNA and cellular components are divided to form 2 daughter cells; the G2 phase between S and M when cells prepare for mitosis; and the G1 phase after mitosis and before phase S when cells commit and prepare for another round of DNA and cellular replication. The G1 restriction point (R) was originally described as the point where cell proliferation becomes independent of mitogens and growth factors [
2], and the normal function of the restriction point is essential for maintaining control of cellular proliferation [
3,
4]. The restriction point is controlled by the retinoblastoma pathway (CDK4/6-cyclin D1-Rb-p16/ink4a). The retinoblastoma protein (Rb) is a tumor suppressor that inhibits proliferation through binding to and suppressing the activity of the E2F family of transcription factors [
5]. In early G1, when conditions are favorable for proliferation, D-type cyclin levels increase through transcriptional and posttranscriptional mechanisms [
6]. Increased cyclin D drives the formation of active kinase heterodimers with CDK4 and CDK6 (CDK4/6) catalytic subunits. Active CDK4/6 then phosphorylates Rb, partially relieving suppression of E2F to allow expression of genes required for passage through the restriction point [
7,
8]. This includes expression of cyclin E, which activates another kinase (CDK2) leading to hyperphosphorylation of Rb, fully releasing the suppression of E2F: allowing cells to exit the G1 phase and initiate DNA replication. Additional restriction point control occurs through the action of the endogenous CDK inhibitors p16/ink4a and p21cip1; p16/ink4a blocks the binding of D-type cyclins to CDK4/6; CDK4/6 cyclin D complexes are stabilized by p21cip1, also sequestering it and preventing the inhibition of CDK2/cyclin E [
4]. Phosphorylation of Rb by CDK4/6 also leads to transcription of genes involved in cell cycle-independent activities including signal transduction, DNA repair transcriptional control, and mRNA processing [
9]. CDK4/6 have also recently been shown to phosphorylate other proteins, including FOXM1 and E2F1 [
10,
11], which modulate additional cellular processes that are cell cycle -independent.
The central role of the Rb pathway in controlling cellular proliferation is highlighted by its frequent dysregulation in human cancer. Aberrant expression of D-type cyclins results in activation of CDK4/6, allowing cells to circumvent the restriction point. In mantle cell lymphoma (MCL), cyclin D1 is upregulated by the (11;14) chromosomal translocation seen in this tumor, and cyclin D1 is overexpressed in many breast, head and neck, prostate, and melanoma tumors [
12,
13]. D-type cyclins have also been shown to be downstream effectors of Flt3 in acute myeloid leukemia (AML) [
14]. CDK4 is amplified or overexpressed in several tumors, including soft tissue sarcomas, glioblastoma, and melanoma [
15‐
18]. CDK6 has also been shown to be overexpressed in tumors [
4]. Both p16/ink4a and Rb act as tumor suppressors and are found to be functionally inactivated in numerous tumor types [
19,
20].
While CDK4/6 are considered highly validated targets for therapeutic intervention, progress toward development of inhibitors for these kinases has been limited by issues with potency, selectivity, and poor pharmacological/physiochemical properties. Examples of these initial efforts include flavopiridol, UCN-01, and CYC202 (roscovitine) [
21]. Fry et al. recently reported a CDK4/6 inhibitor, PD 0332991, possessing improved potency and selectivity over previous inhibitors [
22]. Here, we report the identification of LY2835219, a potent inhibitor of CDK4/6 that inhibits phosphorylation of Rb and induces a G1 cell cycle arrest in Rb-proficient tumor cells in vitro. In vivo, LY2835219 inhibits phosphorylation of Rb by CDK4/6 causing a G1 arrest resulting in antitumor activity in human tumor xenograft models. Physiochemical and pharmacological properties of the compound allow for oral administration, and no tumor outgrowth or significant adverse activities were seen with prolonged administration. When LY2835219 is used in combination with gemcitabine, enhanced inhibition of tumor growth is seen and is associated with a reduction of ribonucleotide reductase (RR) expression without a significant inhibition of Rb phosphorylation, suggesting a cell cycle independent mechanism of action for LY2835219 when combined with gemcitabine.
Discussion
The Rb pathway is dysregulated in more than 80 % of human cancers, highlighting its critical role in controlling cellular proliferation [
4]. Pathway alterations found in human tumors include overexpression of D-type cyclins, activating mutations and amplification of CDK4, and Rb and p16/ink4a inactivation by deletion, mutation, or epigenetic silencing. Data for the role of the Rb pathway in primary human tumors are supported by extensive functional analysis in genetically engineered mouse models [
32]. The importance of the pathway in tumorigenesis and recent success developing small molecule kinase inhibitors, make CDK4/6 a highly validated cancer drug target [
33,
34]. Until recently, CDK inhibitors have shown limited activity in clinical trials because of issues with potency, selectivity and poor pharmacological/physiochemical properties. CDK4/6 inhibitors with improved pharmacokinetic and pharmacodynamic properties have recently been described and are now entering advance stages of clinical development. These recent results support the clinical potential for selective targeting of CDK4/6 in the treatment of cancer [
35].
We describe the identification and preclinical characterization of LY2835219, a potent inhibitor of CDK4/6 with antiproliferative activity in a variety of tumor models in vitro and in vivo. In addition to monitoring p-Rb, we phenotypically assessed the cell cycle status in cellular and in vivo assays; as we previously showed p-Rb alone was not considered sufficient to determine cell cycle activity [
36]. In vitro LY2835219 inhibits Rb phosphorylation by CDK4/6, resulting in a sustained G1 arrest at concentrations 50-fold greater than the cellular IC
50 for p-Rb inhibition, indicating no off-target activity. Consistent with the cell cycle mechanism of action of LY2835219, cell cycle arrest occurred only in Rb-proficient cells. Cell cycle activity was further tested in vivo by monitoring markers for specific cell cycle compartments. LY2835219 effectively inhibited phosphorylation of Rb at serine 780, and significant inhibition of all markers was achieved in colo-205 xenografts at doses of 12.5 mg/kg and higher, indicating a block at the G1 restriction point preventing cells from cycling into the S and M compartments (monitored by TopoIIα and pHH3, respectively). Maximum inhibition of p-Rb and the cell cycle occurred 24 h after LY2835219 administration and was sustained after maximal plasma exposure (Fig.
3). Release from the G1 block was seen through the temporal repopulation of S and M phases, indicated by the sequential loss of p-Rb inhibition (indicating active CDK4/6, and cells entering late G1 phase) followed by TopoIIα (S phase), and finally pHH3 (M phase). In colo-205 xenograft dose response target inhibition studies the TED
70 for LY2835219 is 14 mg/kg for in vivo inhibition of both pRb and TopoII (Fig.
3a). This value is in agreement with the observed minimally efficacious dose of 12.5 mg/kg for tumor growth inhibition. Due to the cell cycle-dependent nature of the PD effect and the observed indirect PK/PD relationship in mouse models, plasma concentrations do not directly correlate with in vivo target inhibition in xenografts, as maximal target inhibition is observed at 24 h post-acute oral dose (Fig.
3b). An integrated and semi-mechanistic pharmacokinetics/pharmacodynamics (PK/PD) model was developed to describe the indirect PK/PD relationship between LY2835219 plasma concentration, inhibition of pRb and subsequent cell cycle arrest, and in vivo efficacy in colo-205 bearing mice [
37]. This PK/PD analysis concludes that sustained, continuous in vivo pRb inhibition and cell cycle arrest are required for robust efficacy in colo-205, and supports a continuous dosing strategy achieving minimum steady state trough plasma concentrations of approximately 200 ng/mL (approximately 400 nM). In this context, it is important to note that the desired cell cycle phenotype in preclinical models in vitro (Fig.
2) and in vivo (Fig.
3) is retained when LY2835219 is dosed up to single digit micromolar concentrations (Fig.
3b). In a phase 1 study in advanced cancer patients, pharmacodynamic activity as well as clinical efficacy were observed. Plasma concentrations in this study reached a mean maximum steady state value of 562 nM [
38,
39], consistent with the target plasma concentrations from our PK/PD model. In total, these data suggest that, in vivo, LY2835219 inhibits CDK4/6 at plasma concentrations clinically achievable, resulting in a G1 arrest that can be sustained with continuous oral dosing.
CDK4 and CDK6 are functionally redundant and ubiquitously expressed, requiring inhibition of both targets to inhibit proliferation of tumors arising from different cellular origins. LY2835219 is a potent inhibitor of both CDK4 and CDK6 kinases with potent cellular inhibition of pRb (Fig.
2a) of approximately 100 nM and has good selectivity over other closely related kinases (approximately 1000-fold selective for CDK1 and ≥ 100-fold selective for CDK2 and CDK5). Biochemical kinase profiling is best interpreted in the context of the functional cellular potency and selectivity for a given drug candidate (see Table
1 for cellular profiling). Factors to consider are the expected potency shift due to the higher intracellular ATP concentration and Km, conformational differences in the native state of protein kinases with respect to conditions and constructs used for biochemical assays, and compensation by other kinases not seen in biochemical assays. For LY2835219 among other kinases in the CMGC kinome family, some activity below 100 nM was observed against CDK9 and PIM1 kinases in biochemical assays. However, the inhibition of CDK9 seen in biochemical assays did not translate into cellular CDK9 activity (inhibition of pCTD and a G2/M arrest in U2OS cells, Figure S
1). Similarly, while activity against PIM1 was seen in biochemical assays, monitoring of the cell cycle in vitro and in vivo shows LY2835219 induces a G1 cell-cycle arrest, even in MV4-11 and Jeko-1 cells and xenografts, which express high levels of PIM1 [
40‐
43]. In MV4-11 cells and unlike PD0332991, LY2835219 showed significant inhibition of pBAD (Ser112), and p4E-BP1 (Thr37 and 46), known substrates of PIM kinases, at concentrations equal or higher than 625 nM after 24 h of incubation (Figure S
3). Both compounds showed similar potent inhibition of pRb (Ser780) in MV4-11. Although the presence of functional cellular PIM1 activity for LY2835219 in these hematological tumor types that are presumably sensitive to PIM inhibition seems possible, the very similar MV4-11 and Jeko-1 xenograft efficacy observed for this molecule with respect to another CDK4/6 inhibitor without PIM1 activity (Fig.
4c, Table
2) suggests a modest effect of this activity compared to CDK4/6 for this particular tumor type, possibly due to compensation by PIM2 or other kinases. More work is be needed to further elucidate the potential role of this PIM1 activity in other tumor types. Overall, these results demonstrate that, in MV4-11, inhibition of PIM1 does not abrogate G1 cell-cycle arrest that arises from the inhibition of CDK4/6.
In vivo target inhibition studies showed LY2835219 produced a cell cycle arrest 24 h after dosing, and was reversible (Fig.
3). The arrest was maintained even after peak plasma concentrations were achieved and indicates target inhibition and antitumor activity can be maintained with daily dosing schedules. This was confirmed in efficacy studies. Significant antitumor activity was seen when LY2835219 was administered orally in several different human tumor xenografts. In colo-205, doses of 12.5 mg/kg and higher inhibited tumor growth and were associated with inhibition of p-Rb, TopoIIα, and pHH3. This shows in vivo antitumor activity results from inhibition of CDK4/6 resulting in a G1 arrest. The antitumor activity of LY2835219 was confirmed in several additional human xenografts representing different human histologies, and was dose-dependent. LY2835219 antitumor activity was further confirmed by comparison of in vivo efficacy to another CDK4/6 inhibitor, PD0332991 (which has similar activity against CDK4/6). In all models tested, both compounds gave similar antitumor activity at similar doses (Fig.
4, Table
2). Finally, antitumor activity of LY2835219 was sustained for up to 56 days with continuous or intermittent dosing schedules in mice bearing colo-205 xenografts without significant adverse events or evidence of acquired resistance (Fig.
4d).
LY2835219 can be combined with the cytotoxic drug gemcitabine, approved to treat several types of tumors, including lung cancer [
44]. In calu 6 lung xenografts both LY2835219 and gemcitabine alone were active in inhibiting tumor growth. Administering the compounds together or in sequence had greater antitumor activity than the individual treatments and was equivalent to the maximum tolerated dose of gemcitabine (150 mg/kg) [
31]. Contrary to other xenograft models, neither inhibition of p-Rb nor a cell cycle arrest was associated with the antitumor activity of LY2835219 in calu-6. However, a significant inhibition of RRM1 was seen when LY2835219 was combined with gemcitabine. In vitro and in vivo chemosensitivity to gemcitabine strongly correlates to RRM1 expression [
45,
46], consistent with the modulation of RRM1 seen for LY2835219 and gemcitabine in our experiments. CDK4/6 can phosphorylate substrates other than Rb, such as FOXM1 and E2F1 [
10,
11], which may explain the inhibition of RRM1 and the absence of inhibition of p-Rb and a cell cycle arrest. Alternatively, LY2835219 in calu-6 cells may result in a low level of p-Rb inhibition not detected by western blot or leading to a cell cycle arrest, but is sufficient to inhibit the expression of RRM1. Consistent with the inhibition of RRM1 in our experiments, expression of RRM1 was previously reported to be highly sensitive to regulation by p-Rb [
28]. Previously reported data that PD0332991 has efficacy in vivo without inhibition of p-Rb [
47] also suggest low levels of inhibition of p-Rb may lead to antitumor activity in vivo. Significantly, H460 contains activating KRAS mutation in codon 61, like calu-6, and shows an attenuated response to LY2835219 (25 mg/kg dose Table
2), consistent with a role for KRAS modifying response to CDK4/6 inhibitors [
48,
49].
In conclusion, we describe the identification and characterization of LY2835219, a potent inhibitor of CDK4/6. Broad antitumor activity was observed in human xenograft tumors of diverse histologic origin that represent human cancers with alterations in the Rb pathway, including MCL, colorectal, lung, glioblastoma and AML. The antitumor activity results predominantly from a G1 cell cycle arrest that correlates with inhibition of p-Rb. In calu-6 lung xenograft tumors, LY2835219 can be combined with gemcitabine resulting in additive antitumor activity and inhibition of RRM1 in the absence of a cell cycle arrest. These results indicate that the antitumor activity of CDK4/6 inhibitors can be context-sensitive, as previously reported in genetically engineered mouse models for glioma and lung cancer [
47,
48]. In addition to its broad antitumor activity, LY2835219 has several additional desirable properties. Efficacy is correlated with tumor Rb status and in vivo target inhibition, allowing for the development of biomarkers for patient stratification and to monitor drug response. A unique feature of LY2835219 is its ability to effectively cross the blood–brain barrier; it has been shown to be active in orthotopic brain tumor xenografts alone or in combination with temozolomide (manuscript in preparation). The Rb pathway is dysregulated in approximately 78 % of glioblastoma and in tumors that commonly metastasize to the brain such as breast, lung, and melanoma [
49,
50]. These properties support further clinical study of LY2835219 in the treatment of both brain metastases and tumors arising in the central nervous system [
51‐
53]. LY2835219, is currently in clinical trials (ClinicalTrials.gov identifiers: NCT01394016, NCT01739309, NCT02079636, NCT02057133, NCT02117648, NCT02014129).