Structure-based generation of a new class of potent Cdk4 inhibitors: new de novo design strategy and library design.

As a first step in structure-based design of highly selective and potent Cdk4 inhibitors, we performed structure-based generation of a novel series of Cdk4 inhibitors. A Cdk4 homology model was constructed according to X-ray analysis of an activated form of Cdk2. Using this model, we applied a new de novo design strategy which combined the de novo design program LEGEND with our in-house structure selection supporting system SEEDS to generate new scaffold candidates. In this way, four classes of scaffold candidates including diarylurea were identified. By constructing diarylurea informer libraries based on the structural requirements of Cdk inhibitors in the ATP binding pocket of the Cdk4 model, we were able to identify a potent Cdk4 inhibitor N-(9-oxo-9H-fluoren-4-yl)-N'-pyridin-2-ylurea 15 (IC(50) = 0.10 microM), together with preliminary SAR. We performed a docking study between 15 and the Cdk4 model and selected a reasonable binding mode which is consistent with the SAR. Further modification based on the proposed binding mode provided a more potent compound, N-[(9bR)-5-oxo-2,3,5,9b-tetrahydro-1H-pyrrolo[2,1-a]isoindol-9-yl]-N'-pyridin-2-ylurea 26a (IC(50) = 0.042 microM), X-ray analysis of which was accomplished by the soaking method. The predicted binding mode of 15 in Cdk4 was validated by X-ray analysis of the Cdk2-26a complex.

A novel approach for the development of selective Cdk4 inhibitors: library design based on locations of Cdk4 specific amino acid residues.

Identification of a selective inhibitor for a particular protein kinase without inhibition of other kinases is critical for use as a biological tool or drug. However, this is very difficult because there are hundreds of homologous kinases and their kinase domains including the ATP binding pocket have a common folding pattern. To address this issue, we applied the following structure-based approach for designing selective Cdk4 inhibitors: (1) identification of specifically altered amino acid residues around the ATP binding pocket in Cdk4 by comparison of 390 representative kinases, (2) prediction of appropriate positions to introduce substituents in lead compounds based on the locations of the altered amino acid residues and the binding modes of lead compounds, and (3) library design to interact with the altered amino acid residues supported by de novo design programs. Accordingly, Asp99, Thr102, and Gln98 of Cdk4, which are located in the p16 binding region, were selected as first target residues for specific interactions with Cdk4. Subsequently, the 5-position of the pyrazole ring in the pyrazol-3-ylurea class of lead compound (2a) was predicted to be a suitable position to introduce substituents. We then designed a chemical library of pyrazol-3-ylurea substituted with alkylaminomethyl groups based on the output structures of de novo design programs. Thus we identified a highly selective and potent Cdk4 inhibitor, 15b, substituted with a 5-chloroindan-2-ylaminomethyl group. Compound 15b showed higher selectivity on Cdk4 over those on not only Cdk1/2 (780-fold/190-fold) but also many other kinases (>430-fold) that have been tested thus far. The structural basis for Cdk4 selective inhibition by 15b was analyzed by combining molecular modeling and the X-ray analysis of the Cdk4 mimic Cdk2-inhibitor complex. The results suggest that the hydrogen bond with the carboxyl group of Asp99 and hydrophobic van der Waals contact with the side chains of Thr102 and Gln98 are important. Compound 15b was found to cause cell cycle arrest of the Rb(+) cancer cell line in the G(1) phase, indicating that it is a good biological tool.

Crystallographic approach to identification of cyclin-dependent kinase 4 (CDK4)-specific inhibitors by using CDK4 mimic CDK2 protein.

Genetic alteration of one or more components of the p16(INK4A)-CDK4,6/cyclin D-retinoblastoma pathway is found in more than half of all human cancers. Therefore, CDK4 is an attractive target for the development of a novel anticancer agent. However, it is difficult to make CDK4-specific inhibitors that do not possess activity for other kinases, especially CDK2, because the CDK family has high structural homology. The three-dimensional structure of CDK2, particularly that bound with the inhibitor, has provided useful information for the synthesis of CDK2-specific inhibitors. The same approach used to make CDK4-specific inhibitors was hindered by the failure to obtain a crystal structure of CDK4. To overcome this problem, we synthesized a CDK4 mimic CDK2 protein in which the ATP binding pocket of CDK2 was replaced with that of CDK4. This CDK4 mimic CDK2 was crystallized both in the free and inhibitor-bound form. The structural information thus obtained was found to be useful for synthesis of a CDK4-specific inhibitor that does not have substantial CDK2 activity. Namely, the data suggest that CDK4 has additional space that will accommodate a large substituent such as the CDK4 selective inhibitor. Inhibitors designed to bind into this large cavity should be selective for CDK4 without having substantial CDK2 activity. This design principle was confirmed in the x-ray crystal structure of the CDK4 mimic CDK2 with a new CDK4 selective inhibitor bound.

J-104,871, a novel farnesyltransferase inhibitor, blocks Ras farnesylation in vivo in a farnesyl pyrophosphate-competitive manner.

Farnesylation of the activated ras oncogene product by protein farnesyltransferase (FTase) is a critical step for its oncogenic function. Because squalene synthase and FTase recruit farnesyl pyrophosphate as a common substrate, we modified squalene synthase (SS) inhibitors to develop FTase inhibitors. Among the compounds tested, a novel FTase inhibitor termed J-104,871 inhibited rat brain FTase with an IC50 of 3.9 nM in the presence of 0.6 microM farnesyl pyrophosphate (FPP), whereas it scarcely inhibited rat brain protein geranylgeranyltransferase-I or SS. The in vitro inhibition of rat brain FTase by J-104,871 depends on the FPP concentration but not on the concentration of Ras peptide. Thus, in vitro studies strongly suggest that J-series compounds have an FPP-competitive nature. J-104,871 also inhibited Ras processing in activated H-ras-transformed NIH3T3 cells with an IC50 value of 3.1 microM. We tested the effects of lovastatin and zaragozic acid A, which modify cellular FPP levels, on Ras processing of J-104,871. Lovastatin, a hepatic hydroxymenthyl coenzyme A reductase inhibitor that reduced the cellular FPP pool, increased the activity of J-104,871, whereas 3 microM zaragozic acid A, an SS inhibitor that raised the FPP level, completely abrogated the activity of J-104,871 even at 100 microM. These results suggest that J-104,871 inhibits FTase in an FPP-competitive manner in whole cells as well as in the in vitro system. Furthermore, J-104,871 suppressed tumor growth in nude mice transplanted with activated H-ras-transformed NIH3T3 cells.

Effects of exogenous p16INK4a on growth of cells with various status of cell-cycle regulators.

p16INK4a, a protein that inhibits cyclin-dependent kinase 4 (Cdk4) and Cdk6, is deficient in many human cancers and in established lines of tumor cells. It has been reported that transfection with cDNA for p16INK4a inhibits the growth of cell lines that express retinoblastoma protein (pRB). However, it is unclear whether the introduction of cDNA for p16INK4a affects the growth of cells that express p16INK4a protein. Moreover, the effects of other cell-cycle regulators on the inhibition of cell growth by p16INK4a remain unknown. In this study, cDNA for p16INK4a was used to transfect human cell lines that had various status of expression of RB pathway-related proteins, such as members of the RB family proteins and Cdk-inhibitory proteins. We found that status of p107, p130, p15INK4b, p18INK4c, p21Cip1, p27Kip1, cyclin D1, and Cdk4 were not correlated with the growth-inhibitory activity of exogenous p16INK4a. By contrast, transfection with cDNA for p16INK4a had a significant effect on the growth of cells depended on the status not only of pRB but also of p16INK4a. Although exogenous p16INK4a inhibited the growth of cells that expressed pRB but did not express p16INK4a (pRB+/p16- cells), it had little affect on either pRB+/p16+ cells or pRB-/p16+ cells. Moreover, transfection with cDNA for p16INK4a also inhibited the activity of the E2 promoter of the dehydrofolate reductase gene in the same manner that depended on the absence of p16INK4a, as well as on the presence of pRB. These results suggest that deregulation of the RB pathway by p16INK4a deficiency plays a very important role in the proliferation of cells that lack p16INK4a protein.

Expression of p16INK4a suppresses the unbounded and anchorage-independent growth of a glioblastoma cell line that lacks p16INK4a.

p16INK4a is a inhibitory protein of Cyclin-dependent kinase 4(Cdk4).p16 negatively regulates the cell cycle progression from G1 to S phase. Functional p16 is absent from many human cancers, as well as from many established lines of tumor cells. However, it is not clear whether expression of p16 in p16-deficient tumor cells can suppress their anchorage-independent growth. Therefore, we introduced a cDNA for p16INK4a into the human glioblastoma cell line T98G, which lacks a gene for p16INK4a. We isolated several clones that stably expressed various amounts of p16 protein. The doubling time of the various clones was generally prolonged. Clones with high-level expression of p16 protein had characteristics of restricted growth, such as contact inhibition, while the parental T98G cells had no such characteristics. Furthermore, the efficiency of colony formation in soft agar was dramatically decreased in the case of cells that expressed exogenous p16. Our observations suggest that the expression of p16 protein restricts the unbounded growth and the anchorage-independent growth of tumor cells.

The consensus motif for phosphorylation by cyclin D1-Cdk4 is different from that for phosphorylation by cyclin A/E-Cdk2.

Cyclin D-Cdk4/6 and cyclin A/E-Cdk2 are suggested to be involved in phosphorylation of the retinoblastoma protein (pRB) during the G1/S transition of the cell cycle. However, it is unclear why several Cdks are needed and how they are different from one another. We found that the consensus amino acid sequence for phosphorylation by cyclin D1-Cdk4 is different from S/T-P-X-K/R, which is the consensus sequence for phosphorylation by cyclin A/E-Cdk2 using various synthetic peptides as substrates. Cyclin D1-Cdk4 efficiently phosphorylated the G1 peptide, RPPTLS780PIPHIPR that contained a part of the sequence of pRB, while cyclins E-Cdk2 and A-Cdk2 did not. To determine the phosphorylation state of pRB in vitro and in vivo, we raised the specific antibody against phospho-Ser780 in pRB. We confirmed that cyclin D1-Cdk4, but not cyclin E-Cdk2, phosphorylated Ser780 in recombinant pRB. The Ser780 in pRB was phosphorylated in the G1 phase in a cell cycle-dependent manner. Furthermore, we found that pRB phosphorylated at Ser780 cannot bind to E2F-1 in vivo. Our data show that cyclin D1-Cdk4 and cyclin A/E Cdk2 phosphorylate different sites of pRB in vivo.

Cyclin-dependent kinase-2 (Cdk2) forms an inactive complex with cyclin D1 since Cdk2 associated with cyclin D1 is not phosphorylated by Cdk7-cyclin-H.

Cyclin-dependent kinases (Cdks) form complexes with cyclins, and as a consequence they generally express kinase activities. One of these Cdks, Cdk2, is known to bind with cyclins A and E, and plays an important role in the progression of the cell cycle via phosphorylation of target proteins such as the product of the retinoblastoma tumor-suppressor gene (pRB). It has been suggested that Cdk2 bound with cyclin D1 and Cdk2-cyclin-D1 complex show neither H1 histone nor pRB kinase activity. However, it is not clear whether Cdk2-cyclin-D1 has unknown targets and why Cdk2 is not activated by binding with cyclin D1. We investigated these questions using Cdk, cyclin and Cdk-cyclin complexes produced in a baculovirus expression system. Cdk2 formed a complex with cyclin D1 in this system. After extensive purification, Cdk2 was still bound to cyclin D1. The Cdk2-cyclin-D1 complex did not phosphorylate any tested substrates, such as H1 histone, pRB, SV40 large T antigen, p53, E2F-1 or a preparation of nuclear proteins from HeLa cells; in contrast, Cdk2-cyclin-E and Cdk2-cyclin-A phosphorylated these proteins. Moreover, the Cdk2-cyclin-D1 complex was not activated by incubation with Cdk4 or cyclin E. Thus, Cdk2 and cyclin D1 formed a stable complex that was not activated. In order to determine why Cdk2-cyclin-D1 lacks kinase activity, we investigated the phosphorylation of Cdk2. Under-shifted Cdk2, the active form of Cdk2, was not detected in the Cdk2-cyclin-D1 complex in the baculovirus system. In human WI-38 cells, cyclin D1 began to form a complex with Cdk2 as well as with Cdk4 from the mid-G1 phase of the cell cycle. The Cdk2 bound to cyclin D1 in human cells was also the inactive form that was slowly migrated. Moreover, we found that Cdk2 bound to cyclin D1 was not phosphorylated by Cdk7-cyclin-H, while Cdk2 bound to cyclin E, as well as free Cdk2, was was phosphorylated by Cdk7-cyclin-H. Additionally, Cdk2 phosphorylated by Cdk7-cyclin-H did not bind to cyclin D1. These results strongly suggest that Cdk2 forms a stable complex with cyclin D1 but is not activated because the Cdk2 molecule in the complex is not phosphorylated by Cdk7-cyclin-H and the phosphorylated Cdk2, an active form, does not bind to cyclin D1.

Differences in substrate specificity between Cdk2-cyclin A and Cdk2-cyclin E in vitro.

Cyclin-dependent kinase 2 (Cdk2), when bound to either cyclin A or cyclin E, recognizes the Ser/Thr-Pro-X-basic amino acid (motif A) as a phosphorylation site. In this study, we designed several peptides based on motif A and examined the substrate specificity of Cdk2-cyclin A and Cdk2-cyclin E using these peptides. Peptides containing a proline residue in the sequence Pro-X-Thr-Pro-X-basic amino acid (motif B) had higher affinity for both Cdk2 complexes than peptides containing motif A. Furthermore, differences in substrate affinity between the two Cdk2 complexes were caused by a proline residue adjacent to or three positions before the threonine residue. Similarly, the presence of different basic amino acids in motif B also had different effects on affinity for each complex. We demonstrate the possibility that the substrate specificity of Cdk2 bound to cyclin might be regulated by the species of cyclin.

The interactions of E2F with pRB and with p107 are regulated via the phosphorylation of pRB and p107 by a cyclin-dependent kinase.

It has been postulated that the product (pRB) of the retinoblastoma gene dissociates from the E2F-pRB complex upon phosphorylation by cyclin-dependent kinase(s) (cdk). However, there is no direct evident for the regulation of formation of the E2F-pRB complex via phosphorylation by purified cdk. Therefore, we investigated the regulation of formation of this complex by phosphorylation using pRB and purified cyclin A-cdk2, cyclin E-cdk2 or cyclin D1-cdk4. Purified pRB was incubated with nuclear extracts prepared from pRB-defective cells and then subjected to gel mobility shift assays. We confirmed that unphosphorylated pRB associated with various types of E2F but pRB has been phosphorylated by cyclin A-cdk2 did not. We found that E2F-pRB complexes were disrupted as a consequence of phosphorylation by cyclin A-cdk2, and the levels of the free forms of E2Fs increased. We also found that not only the E2F-pRB complexes but also the E2F-p107 complexes were disrupted upon phosphorylation by cyclin A-cdk2. Furthermore, E2F-pRB complexes were disrupted through phosphorylation by cyclin D1-cdk4 and cyclin E-cdk2, as well as by cyclin A-cdk2. These results clearly demonstrate that the phosphorylation of pRB and p107 by cdks regulates the formation of complexes between E2F and pRB or p107.

Phosphorylation of E2F-1 by cyclin A-cdk2.

Transcription factor E2F-1 has a putative consensus sequence for phosphorylation by cyclin dependent kinase (Ser-Pro-X-Lys/Arg). Therefore, we studied the phosphorylation of E2F-1 in vivo and in vitro and its biological functions. E2F-1 was prepared by immunoprecipitation with anti-E2F-1 antibody from IMR32 lysates and was effectively phosphorylated by human cyclin A-cdk2 which was expressed in insect cells using baculovirus system. GST-E2F-1 was phosphorylated by cyclin A-cdk2 more efficiently than by cyclin E-cdk2. Cyclin D1-cdk4 phosphorylated pRB but scarcely phosphorylated GST-E2F-1 or H1 histone. The 60 kd protein precipitated with anti-E2F-1 antibody was phosphorylated in vivo. Phospho-peptide mapping indicated that its cleavage profile was identical with that of E2F-1 phosphorylated by cyclin A-cdk2 in vitro. This 60 kd protein, which is likely to be E2F-1, was not phosphorylated during the G0 and early G1 phase. Phosphorylation of E2F-1 began from the S phase while phosphorylation of pRB started nearly at G1/S. The in vivo phosphorylation of E2F-1 was inhibited by butyrolactone I, a cyclin-dependent kinase inhibitor (Kitagawa et al., 1993, Oncogene, 8, 2425-2432). The binding of E2F-1 to E2 promoter was found to be reduced by phosphorylation of E2F-1 by cyclin A-cdk2, suggesting that phosphorylation of E2F-1 may induce shut off of gene expression at the transcriptional level. These results suggest that E2F-1 is phosphorylated by cyclin A-cdk2 in the S phase in vivo as well as in vitro and that its phosphorylation by cyclin A-cdk2 may modulate its activity.

Butyrolactone I, a selective inhibitor of cdk2 and cdc2 kinase.

We screened cdc2 kinase inhibitors from cultured mediums of micro organisms using purified mouse cyclin B-cdc2 kinase and a specific substrate peptide for cdc2 kinase. A selective inhibitor of cdc2 kinase was isolated from the cultured medium of Aspergillus species F-25799, and identified as butyrolactone I. Butyrolactone I inhibited cdc2 and cdk2 kinases but it had little effect on mitogen-activated protein kinase, protein kinase C, cyclic-AMP dependent kinase, casein kinase II, casein kinase I or epidermal growth factor-receptor tyrosine kinase. Its inhibitory effect was found to be due to competition with ATP. Butyrolactone I selectively inhibited the H1 histone phosphorylation in nuclear extracts. It also inhibited the phosphorylation of the product of retinoblastoma susceptibility gene in nuclear extracts and intact cells. Thus butyrolactone I should be very useful for elucidating the function of cdc2 and cdk2 kinases in cell cycle regulation.