Dysregulation of p27Kip1 due to proteolysis that involves the ubiquitin ligase (SCF) complex with S-phase kinase-associated protein 2 (Skp2) as the substrate-recognition component (SCFSkp2) frequently results in tumorigenesis. In this report, we developed a high-throughput screening system to identify small-molecule inhibitors of p27Kip1 degradation. This system was established by tagging Skp2 with fluorescent monomeric Azami Green (mAG) and CDK subunit 1 (Cks1) (mAGSkp2–Cks1) to bind to p27Kip1 phosphopeptides. We identified two compounds that inhibited the interaction between mAGSkp2–Cks1 and p27Kip1: linichlorin A and gentian violet. Further studies have shown that the compounds inhibit the ubiquitination of p27Kip1in vitro as well as p27Kip1 degradation in HeLa cells. Notably, both compounds exhibited preferential antiproliferative activity against HeLa and tsFT210 cells compared with NIH3T3 cells and delayed the G1 phase progression in tsFT210 cells. Our approach indicates a potential strategy for restoring p27Kip1 levels in human cancers.
Cyclin-dependent kinases (CDK), which function upon activation by cyclin binding, are known to be an active key molecule complexes that regulate the progression of cell cycles. The activities of CDK are constrained by CDK inhibitors (CKI).[1-3] One CKI, p27Kip1, a 198-amino-acid protein that was discovered in cells arrested by TGF-β,[4, 5] has significant functions in governing cell proliferation, cell motility, senescence and apoptosis.[1-3] Unlike other tumor suppressors, for instance p53, p27Kip1 is rarely mutated in human cancers. It is reported that p27Kip1 is usually dysregulated in cancers, despite high or constant p27Kip1 mRNA levels.[3, 6, 7] During the cell cycle, p27Kip1 negatively regulates the G1-S transition, and its levels peak during G1 phase, causing arrest in G1.[1, 3, 8, 9] These fluctuations are not mirrored in its mRNA levels,[9, 10] suggesting that downregulation of p27Kip1 in human cancers, which is associated with many aggressive phenotypes and a poor prognosis in various cancers (e.g. breast, colon, prostate, lung and gastric cancers), is caused primarily by post-transcriptional events.[3, 9]
It is well known that p27Kip1 is degraded through a sequential degradation system, called the ubiquitin-proteasome system (UPS). This process begins with the transfer and covalent attachment of ubiquitin to target proteins through a cascade of enzymatic reactions, followed by degradation of the marked target proteins by the proteasome. Biochemical studies have shown that p27Kip1 is ubiquitinated in vitro and in vivo primarily by SCFSkp2, a ubiquitin ligase complex that contains S-phase kinase-associated protein 2 (Skp2).[13-16] p27Kip1 is recognized by Skp2 only when it is phosphorylated by CDK2/cyclin E on Thr-187.[17-19] Moreover, recognition of p27Kip1 by SCFSkp2 requires an accessory protein, CDK subunit 1 (Cks1), which binds to phosphorylated p27Kip1 and Skp2.[20, 21] Thus, Skp2 and Cks1 are equally important for the recognition of and binding to p27Kip1.
Several published studies suggest that Skp2 has oncogenic activity.[22, 23] Notably, transformed cell lines and human cancers are frequently associated with overexpression of Skp2 and p27Kip1 downregulation.[6, 24, 25] With activated Ras, Skp2 transforms cells and induces lymphomas in transgenic mice, and transgenic expression of Skp2 in mouse prostate causes hyperplasia, dysplasia and low-grade prostate carcinoma with significant p27Kip1 downregulation. In contrast, Skp2 knockout mice are fertile and viable and have elevated p27Kip1 levels. Moreover, siRNA-mediated silencing of Skp2 in oral and lung carcinomas inhibits and suppresses tumor proliferation.[29, 30] Thus, the involvement of Skp2 in many aggressive cancers suggests that targeting it using small-molecule inhibitors is a promising cancer therapy.
Recent advances in chemical biology studies have allowed us to identify such biologically active small molecule inhibitors of various targets by high-throughput compound screening. Several compounds able to prevent p27Kip1 degradation are reported.[31-33] In Rico-Bautista et al., cell-based high-throughput screening using Skp2 overexpressing cells identified two compounds that restored the levels of nuclear p27Kip1 efficiently. While the direct molecular targets of these compounds were yet to be unveiled, one of the compounds turned out to induce the downregulation of Skp2 in cells. The other compounds of this study and a compound found in another study have an activity to increase the mRNA level of p27Kip1. Using an in vitro-reconstituted electrophoresis-based ubiquitination assay as a screening system, a compound that inhibits ubiquitination dependent proteolysis of p27Kip1 is isolated.
In this report, we developed a high-throughput screening (HTS) system to identify small molecule inhibitors of protein–protein interaction between Skp2–Cks1 and p27Kip1. We identified two small molecule inhibitors of p27Kip1 degradation. These small molecule inhibitors inhibited the in vitro ubiquitination of p27Kip1 by SCFSkp2, stabilized p27Kip1 levels in HeLa cells and inhibited the growth of human and mouse cancer cells.
Materials and Methods
To construct recombinant baculoviruses that express mAGSkp2 (mAG, N-terminally fused fluorescent monomeric Azami Green) and Cks1, hemagglutinin (HA)-tagged mAGSkp2 and Cks1 DNA fragments were inserted into the EcoRI-NotI and SalI-NotI sites of pFast-Bac. The recombinant plasmids were then transformed into DH10Bac cells followed by transfection of the obtained recombinant bacmids into ExpressSF+ Serum-Free Insect Cells (Protein Sciences) using Cellfectin II (Invitrogen, Carlsbad, CA, USA). Recombinant baculovirus that expressed Skp1 was obtained from Dr Keiichi Nakayama (Kyushu University, Fukuoka-shi, Japan).
For western blot analysis to detect endogenous protein levels, commercial antibodies against Skp2 (Santa Cruz), p130 (Santa Cruz) and p21 (Abcam, Cambridge, UK) were used at concentrations recommended by these manufacturers.
Cell culture conditions
Insect cells were grown in Sf-900 II SFM (serum-free medium complete 1×, GIBCO) and cultured at 27°C with constant stirring at 130 rpm. A total of 150 mL of insect cell culture was coinfected with 3 mL of each recombinant baculovirus (mAGSkp2, Cks1 and Skp1) for 72 h before the cells were harvested. HeLa cells were maintained in DMEM (Invitrogen) that contained 10% fetal calf serum, and NIH3T3 cells were maintained in MEM (SIGMA, St. Louis, MO, USA) that contained 10% calf serum; both media were supplemented with 0.5% penicillin/streptomycin (Invitrogen). Cells were maintained in a humidified incubator at 37°C and 5% CO2. The tsFT210 cell line, which is a temperature-sensitive mutant that has been isolated from the mouse mammary carcinoma cell line FM3A, was maintained in RPMI-1640 medium (Invitrogen) that was supplemented with 5% calf serum in a suspension culture at 32°C and 5% CO2. tsFT210 cells can be arrested at G2 phase and are large in size when cultured at 39°C at 5% CO2.[35-37]
Development of binding assay and screening for inhibitors of Skp2–Cks1 and p27Kip1 interaction
A 24-residue (residue 175–197) p27Kip1 phosphopeptide that harbored the target sequence (C-SDGSPNAGSVEQpTPKKPGLRRRQT), which binds Skp1–Skp2–Cks1 with equal affinity compared with the full-length phosphorylated p27Kip1 peptide and its derivatives without phosphorylation, was chemically synthesized, purified and bound covalently to maleimide-activated 96-well plates (Pierce) at 100 μL/well of a 10 μM solution according to the manufacturer's protocol.
Insect cells expressing mAGSkp2, Cks1 and Skp1 were lysed by sonication in lysis buffer (20 mM Tris, pH 7.5, 125 mM NaCl, 0.5% Nonidet P-40, 50 mM NaF, 1 mM EDTA, 200 μM Na3VO4 and 1 mM dithiothreitol), containing Complete Protease Inhibitors (Roche Applied Science, Penzberg, Germany). Screening for inhibitors of the Skp2–Cks1 and p27Kip1 interaction from RIKEN NPDepo (Natural Product Depository) chemical library was performed at high-throughput manner on a Biomek 2000 liquid handling system (Beckman-Coulter, Brea, CA, USA), essentially as described previously. Briefly, 50 μL of insect cell lysate was mixed with chemical compounds at a final concentration of 60 μg/mL. The mixtures were added to 96-well phosphopeptide-bound plates and incubated overnight at 4°C on a shaker. After five rinses with 0.05% Nonidet P-40 in PBS and two rinses with PBS, the reaction was quantified spectrophotometrically at excitation 488 nm and emission 546 nm.
Cell proliferation assay
A total of 4 × 103 HeLa cells/200 μL/well, 1 × 104 NIH3T3 cells/200 μL/well or 1.6 × 104 tsFT210 cells/200 μL/well were seeded into the wells of 96-well plates and were incubated for 24 h in a humidified 37°C incubator with 5% CO2 for HeLa and NIH3T3 cells and in a 32°C incubator for tsFT210 cells. Then, the cells were treated with increasing concentrations of compounds for 48 h, cell growth was measured using Cell Count Reagent SF (Nacalai Tesque, Kyoto, Japan), and IC50 values were calculated, based on the absorbance at 450 nm (Wallac 1420 ARVO; PerkinElmer, Waltham, MA, USA).
Protein stability assay
HeLa cells were treated with compounds at their IC50 concentrations for 18 h, and 20 μg/mL cycloheximide (CHX) was added for the indicated times prior to the preparation of cell lysates. Total protein amounts were measured by Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA, USA), with BSA as standard. Approximately 40 μg of cell lysate was subjected to SDS-PAGE, followed by immunoblot analysis with anti-p27Kip1 (BD Transduction Laboratory, Franklin Lakes, NJ, USA).
RNA extraction, cDNA synthesis and semi-quantitative RT-PCR
Total RNA of compound-treated HeLa cells was extracted using TRizol (Invitrogen) according to the manufacturer's instructions. First-strand cDNA was synthesized using the ImProm-II Reverse Transcription System (Promega, Fitchburg, WI, USA) and was used for semi-quantitative PCR employing intron-spanning primers. Primer sequences for p27Kip1 are FW: 5′-ACCTGCAACCGACGATTCTT-3′ and RV: 5′-CCCTTCCCCAAAATTGCTTC-3′. Primer sequences for β-actin are FW: 5′-CTGGACTTCGAGCAAGA-3′ and RV: 5′-TCCTGCTTGCTGATCCA-3′.
Purification of p27Kip1
The coding region of p27Kip1 was cloned into the pET-28a(+) vector with 8 His-Tag at the N-terminal, using the BamHI and NcoI sites to produce recombinant p27Kip1. Recombinant p27Kip1 was transformed and expressed in BL21 (DE3) cells and purified with HisTrap affinity columns (GE Healthcare, Little Chalfont, UK) according to the manufacturer's protocols.
In vitro ubiquitination assay
The p27Kip1in vitro ubiquitination assay was performed as previously described[21, 42] using SCF complex expressed and immunoprecipitated from HEK293T cells.
Flow cytometric analysis
To obtain G2-synchronized cells, tsFT210 cells were seeded into the wells of 24-well plates at 10 × 104 cells/500 μL/well and preincubated at 39°C for 18 h. Then, the cells were treated with compounds at the indicated concentrations prior to incubation at 32°C for the indicated times. Cell suspensions were harvested, washed with PBS, and resuspended in 500 μL propidium iodide buffer, containing 50 μg/mL propidium iodide (Sigma-Aldrich, St. Louis, MO, USA), 0.1% sodium citrate, 0.2% Nonidet P-40 and 2 μg/mL RNase A (Nacalai Tesque), for at least 30 min in the dark. The DNA content of the cells was analyzed on a Cytomics FC500 (Beckman Coulter).
Development of screening system to identify inhibitors of Skp2–Cks1 and p27Kip1 interaction
We developed a high-throughput screening system using Skp1, Skp2 and Cks1 expressed in the baculovirus protein expression system and phosphopeptides of p27Kip1-derived sequences. In this system, we fused mAG, a fluorescent protein from the stony coral Galaxeidae to the N-terminus of Skp2, allowing us to quantify the binding of mAGSkp2–Cks1 to p27Kip1 by spectrofluorometry (Fig. 1a).
When insect cell lysates expressing Skp1, mAG-Skp2 and Cks1 were incorporated into the 96-well plates in which p27Kip1-derived phosphopeptides were covalently bound, the binding of mAG-Skp2 to p27Kip1 peptides was detected by spectrofluorometry. The binding was observed only to p27Kip1 phosphopeptides (Fig. 1b). The importance of Skp1 and Cks1 in p27Kip1 phosphorylation-dependent binding was also evaluated and we noted that Skp1 is not necessary for p27Kip1 phosphorylation-dependent binding. In contrast, the fluorescent signal was abolished in cell lysates that did not express Cks1. Thus, Cks1 but not Skp1 is essential for p27Kip1 phosphopeptide recognition (Fig. 1b).
To determine the reliability of our screening system, we performed a peptide competitive binding assay. Insect cell lysates expressing mAG-Skp2 and Cks1, mixed with p27Kip1 phosphopeptide solution at the indicated concentrations, were added to 96-well plates to which p27Kip1 phosphopeptides were bound (Fig. 1c). In the presence of 10 μM phosphopeptides, the interaction between mAG-Skp2–Cks1 and well-bound p27Kip1 phosphopeptides was inhibited by approximately 20%. This binding decreased further with higher concentrations of the p27Kip1 phosphopeptide solution, indicating the reliability and efficacy of our screening system. Thus, we used this system for small-molecule inhibitor screening.
Identification of linichlorin A and gentian violet as specific inhibitors of Skp2–Cks1 and p27Kip1 interaction
We screened approximately 20 000 compounds in the RIKEN NPDepo chemical library at 60 μg/mL. We identified 258 compounds that reduced the fluorescence of mAGSkp2 to <80% of control levels as primary hits. In a secondary screening, all 258 hit compounds were re-examined at the same concentration to confirm reproducibility of their inhibitory activities and, at the same time, compounds that inhibit other phosphorylation dependent protein interactions, such as polo box domain dependent interaction, were excluded. After these selections, 30 compounds were considered positive in our assay. Next, we measured the effects of these compounds on cell growth. Among the 30 compounds, 15 compounds showed strong growth inhibition on HeLa cells while the other 15 compounds exhibited weaker growth inhibition. Among the 15 stronger compounds, linichlorin A and gentian violet were found to show the strongest effect on cell growth (Fig. 1d). We also observed that the inhibitory activity of linichlorin A and gentian violet obtained from the pulled down assay (Fig. 2a) corresponds to the inhibitory activity obtained in HTS. Thus, these two compounds were selected for further study.
Linichlorin A and gentian violet inhibit in vitro ubiquitination of p27Kip1
As shown in Figure 2(b), we have constructed the in vitro ubiquitination reaction of p27Kip1. The ubiquitination was observed only when the reagents and enzymes (ubiquitin, p27Kip1 and CDK2/Cyclin E) were present in the reaction. After adding linichlorin A or gentian violet to the ubiquitination reaction, in vitro p27Kip1 ubiquitination declined by 70–80% (Fig. 2c), as quantified on an ImageQuant TL (GE Healthcare). Thus, linichlorin A and gentian violet inhibited p27Kip1 ubiquitination in vitro.
p27Kip1 stabilization by linichlorin A and gentian violet
To examine the effect of these compounds on the stability of p27Kip1, exponentially growing HeLa cells were treated with the compounds for 18 h. Then, 20 μg/mL CHX was administered to the cells to block the protein translation, and p27Kip1 levels were measured by immunoblotting. As shown in Figure 3(a) and (b), p27Kip1 was degraded rapidly after the addition of CHX in control cells (DMSO). In contrast, the degradation of p27Kip1 was almost completely inhibited in the presence of either compound, demonstrating that linichlorin A and gentian violet stabilize p27Kip1 in HeLa cells after the protein translation. In fact, as expected, no significant change in the p27Kip1 mRNA expression level was observed in the compound treated cells (Fig. 3c). In addition, the expression level of Skp2 protein was not affected under these conditions, although it was decreased in the presence of a higher concentration of gentian violet (Fig. S1).
Linichlorin A and gentian violet inhibit cell growth and delay cell cycle progression
Next, we examined the effect of these compounds on the growth of tumor and non-tumor cells. HeLa (human tumor cells), tsFT210 (mouse tumor cells) and NIH3T3 (mouse immortalized cells) were exposed to increasing concentrations of linichlorin A and gentian violet for 48 h and their viability was measured by WST-8 assay (Fig. 4a). The IC50 values for linichlorin A in HeLa, tsFT210 and NIH3T3 cells were 3.2, 1.6 and 12.7 μM, respectively, and 0.4, 0.6 and 5.3 μM for gentian violet, respectively. This shows that the compounds inhibited growth to a greater extent in HeLa and tsFT210 cells compared with NIH3T3 cells. When we compared the expression level of p27Kip1 between two mouse cells (tsFT210 cells and NIH3T3 cells) by immunoblotting, the endogenous expression level of p27Kip1 in growing cells was higher in NIH3T3 cells (Fig. 4b). However, in the range of the compound concentration examined, p27Kip1 expression level was increased by the compounds dose-dependently only in tsFT210 cells, as it is in HeLa cells (Fig. 4b). Although the mechanism of this difference in the drug sensitivity remains to be elucidated, this difference may explain the stronger growth inhibitory effect of these compounds in cancer cell lines.
It is predicted that the G1–S transition will be retarded when p27Kip1 is stabilized. To confirm this, we examined the compounds' effects on the cell cycle progression using tsFT210 cells that have temperature sensitive Cdc2 mutation. After tsFT210 cells were synchronized at G2 phase by incubation at 39°C (0 h), the cells were treated with compounds at their respective IC50 values (linichlorin A 1.6 μM and gentian violet 0.6 μM) prior to incubation at 32°C for 8 and 12 h. As a result, the compounds delayed the initiation of S phase (Fig. 4c) and the levels of p27Kip1 were significantly increased in the compound-treated cells (Fig. 4d).
Low levels of p27Kip1 and overexpression of Skp2 in many human carcinomas and lymphomas are caused primarily by the increased proteolysis of p27Kip1 through ubiquitination by SCFSkp2 E3 ligase.[43-46] Thus, discovery of small molecule inhibitors through the development of screening systems that specifically target SCFSkp2 E3 ligase by blocking the interaction between the F-box protein (Skp2) and its substrate (p27Kip1) presents an ideal cancer therapeutic.
We developed an HTS system for inhibitors of the interaction of Skp2–Cks1 and p27Kip1. Through HTS and further study, we identified linichlorin A and gentian violet as the most potent compounds that inhibited the ubiquitin-proteasome dependent degradation of p27Kip1 (Fig. 1d). Linichlorin A is a sesquiterpene lactone that was first isolated from Centaurea linifolia Vahl and does not have any reported biological activity, whereas gentian violet (also known as crystal violet) is a triphenylmethane-classed dye that has antifungal, antibacterial and antiparasitic activities. While both compounds have inhibitory effects on the binding between mAGSkp2–Cks1 and p27Kip1 phosphopeptides, these compounds did not show any effect on the phosphorylation-dependent binding of other F-box proteins, β-TrCP1 and β-TrCP2. In addition, these compounds also did not show any effect on other phosphorylation dependent protein–protein interactions of PBD, Pin1, 14-3-3 and Mdc1 (Watanabe N and Osada H, unpublished results). Therefore, we consider that the inhibitory effect of these compounds is specific to Skp2–p27Kip1 interaction.
From the structure of Skp2 and Cks1 complexed with p27Kip1 phosphopeptides, it is predicted that these compounds bind to the pocket created by Skp2 and Cks1 rather than binding to the phosphorylated peptides. In fact, no inhibition on binding was observed in our screening system when the compounds were pre-mixed with phosphorylated peptides in 96-well plates and washed before the addition of Skp2–Cks1 containing lysates, indicating that the compounds do not bind to the phosphopeptides but to the Skp2–Cks1 complex, as predicted. In this regard, the increase in the expression level of two other known Skp2–Cks1 target proteins, p21Cip1() and p130, by these compounds supports this idea (Fig. S2).
According to earlier findings,[13, 14, 17, 18, 20] the ubiquitination of p27Kip1 is triggered by its phosphorylation by CDK2/cyclin E, followed by recognition of phosphorylated p27Kip1 by SCFSkp2 E3 ligase and an adapter protein, Cks1, which polyubiquitinates p27Kip1 and targets it for proteasomal degradation. Here, we found that linichlorin A and gentian violet had actually inhibited p27Kip1 ubiquitination in vitro (Fig. 2c), as predicted by their Skp2–p27Kip1 binding inhibitory activities, followed by stabilization of p27Kip1 level in the compound-treated HeLa cells (Fig. 3a,b).
We have also demonstrated that linichlorin A and gentian violet have substantial, selective antiproliferative activity against cancer and transformed cells in the micromolar range (Fig. 4a). In many types of tumor cells, it is reported that Skp2 is overexpressed and the level of p27Kip1 is downregulated. This downregulation of p27Kip1 is considered to be responsible for the accelerated growth of tumor cells. Thus, the inhibition on Skp2-dependent degradation of p27Kip1 may exhibit larger effects on the growth of tumor cells.
Although further analyses using cells with reduced or overexpressed levels of p27Kip1 and/or Skp2 are necessary, our results strongly suggest that these compounds inhibit the growth of cancer cells at G1 phase of the cell cycle by stabilizing p27Kip1 through the inhibition of the interaction between Skp2–Cks1 and p27Kip1. Taken together, our study demonstrates a potential strategy for restoring p27Kip1 levels in cancers using small molecule inhibitors.
This work was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. We thank the members of the RIKEN Chemical Biology Core Facility (especially Emiko Sanada) for technical assistance, support and discussions. L-C. Ooi is grateful for financial support from the RIKEN International Program Associate and from the Universiti Sains Malaysia Fellowship Scheme.