Seliciclib (CYC202, R-Roscovitine) Induces Cell Death in Multiple Myeloma Cells by Inhibition of RNA Polymerase II–Dependent Transcription and Down-regulation of Mcl-1

Seliciclib (CYC202, R-roscovitine) is a cyclin-dependent kinase (CDK) inhibitor that competes for the ATP binding site on the kinase. It has greatest activity against CDK2/cyclin E, CDK7/cyclin H, and CDK9/cyclin T. Seliciclib induces apoptosis from all phases of the cell cycle in tumor cell lines, reduces tumor growth in xenografts in nude mice and is currently in phase II clinical trials. This study investigated the mechanism of cell death in multiple myeloma cells treated with seliciclib. In myeloma cells treated in vitro, seliciclib induced rapid dephosphorylation of the carboxyl-terminal domain of the large subunit of RNA polymerase II. Phosphorylation at these sites is crucial for RNA polymerase II–dependent transcription. Inhibition of transcription would be predicted to exert its greatest effect on gene products where both mRNA and protein have short half-lives, resulting in rapid decline of the protein levels. One such gene product is the antiapoptotic factor Mcl-1, crucial for the survival of a range of cell types including multiple myeloma. As hypothesized, following the inhibition of RNA polymerase II phosphorylation, seliciclib caused rapid Mcl-1 down-regulation, which preceded the induction of apoptosis. The importance of Mcl-1 was confirmed by short interfering RNA, demonstrating that reducing Mcl-1 levels alone was sufficient to induce apoptosis. These results suggest that seliciclib causes myeloma cell death by disrupting the balance between cell survival and apoptosis through the inhibition of transcription and down-regulation of Mcl-1. This study provides the scientific rationale for the clinical development of seliciclib for the treatment of multiple myeloma.

There are more than 500 protein kinases encoded by the human genome and this class of enzymes has been an area of intensive research for the development of novel anticancer agents (1). One family of kinases that has attracted particular attention is the cyclin-dependent kinases (CDK), which regulate two processes essential for cancer cell survival; cell cycle progression and gene transcription (2). CDKs control entry into each stage of the cell cycle by phosphorylating key substrates such as pRb and condensin (3, 4). CDKs usually form heterodimers with cyclins to create an active complex and because most cyclin levels oscillate throughout the cell cycle, this contributes to the temporal activation of specific CDKs; for example, CDK2/cyclin E for entry into S phase and CDK2/cyclin A for exit out of S phase. CDKs also regulate transcription by phosphorylating the carboxyl-terminal domain of the large subunit of RNA polymerase II (5). In humans the carboxyl-terminal domain contains 52 repeats of a heptapeptide (YSPTSPS) that can be phosphorylated on serines, threonines, and tyrosines (5). A number of CDKs have been shown to phosphorylate these sites including: CDK7/cyclin H/Mat1, part of the TFIIH complex that activates transcriptional initiation and CDK9/cyclin T, also called P-TEFb, that can activate transcriptional elongation (6). In addition, CDK8/cyclin C, CDK1/cyclin B, and CDK2/cyclin E have been shown to phosphorylate the carboxyl-terminal domain in vitro (6, 7).

Seliciclib (CYC202, R-roscovitine) is a 2,6,9-substituted purine analogue that competes with ATP for binding to the active site on CDKs. Seliciclib has potent in vitro activity against CDK2/cyclin E (IC₅₀ = 0.1 μmol/L), CDK7/cyclin H (IC₅₀ = 0.36 μmol/L), and CDK9/cyclin T (IC₅₀ = 0.81 μmol/L; refs. 8, 9). A broad range of tumor cell types are inhibited by seliciclib in vitro, with an average 72-hour IC₅₀ value of 15 μmol/L. Treatment of cell lines has wide-ranging effects including: accumulation of cells in G₁ and G₂ phases, inhibition of rRNA processing, inhibition of RNA polymerase II–dependent transcription, disruption of nucleoli, and induction of apoptosis from all stages of the cell cycle (8, 10–15). Human tumor xenografts in nude mice treated with seliciclib show significantly reduced growth compared with controls (8). Seliciclib is currently in phase II clinical trials in combination with gemcitabine/cisplatin for non–small cell lung cancer and as a single agent for B cell malignancies including multiple myeloma.

Multiple myeloma is a disease of malignant B cells that have extended survival and a low proliferation rate which accumulate in the bone marrow causing osteolytic bone lesions (16). Treatment with current chemotherapeutic agents initially reduces tumor burden but the disease remains incurable, with the average patient survival time being 3 years. Therefore, there is an urgent need for novel therapies ideally based on an understanding of the biology of the disease. Typically, the disease begins with monoclonal gamopathies of undetermined significance that progresses to myeloma in a small percentage of cases (16, 17). Genetic changes that give rise to multiple myeloma can involve genes that regulate both the cell cycle and apoptosis. These include overexpression of antiapoptotic Bcl-2 and down-regulation of proapoptotic Bik (18). A recent study analyzing identical twins, one with multiple myeloma and one without, identified overexpression of Mcl-1, c-FLIP, and Dad-1 in the malignant cells of the multiple myeloma patient, emphasizing the importance of these antiapoptotic proteins in the development of this disease (19).

Mcl-1, an antiapoptotic member of the Bcl-2 family (20), was first identified as a protein up-regulated in a human myeloblastic leukemia cell line induced to differentiate along the monocyte lineage (21). Mcl-1 is essential for the survival of multiple myeloma cells (22–24) as well as B cell lymphoma cells (25). Mcl-1 is thought to act by antagonizing proapoptotic proteins such as Bim (26).

In this study, the mechanism of cell-killing by seliciclib was explored; testing the hypothesis that the ability of seliciclib to induce myeloma cell death occurred by inhibiting transcription and down-regulating critical proteins required for cell survival. This study shows that seliciclib inhibits the kinases which phosphorylate the carboxyl-terminal domain of RNA polymerase II resulting in inhibition of transcription, down-regulation of the levels of Mcl-1 mRNA and protein, and rapid induction of apoptosis. These data provide scientific rationale for the clinical development of seliciclib as an agent for treating multiple myeloma.

Cell lines and reagents. NCI-H929, LP-1, RPMI 8226, OPM2, and U266 cells were purchased from the DSMZ (Germany). Cells were cultured at 37°C with 5% CO₂ in RPMI 1640 containing 10% FCS and 2 ng/mL of IL-6 (R&D systems, Abingdon, United Kingdom) an important survival factor in vivo (27). Media for H929 cells was supplemented with 50 μmol/L 2-mercaptoethanol (BDH, Poole, United Kingdom). Cells were kept at a density of between 0.2 and 1 × 10⁶ cells/mL. All reagents were purchased from Sigma (Poole, United Kingdom) unless stated otherwise.

Cytotoxicity assays. Cells were seeded in 96-well plates appropriately for their doubling time (5,000 cells per well except for H929 which were seeded at 8,000 cells per well) and incubated overnight. Stock solutions of compounds were prepared in DMSO at 100 mmol/L. A dilution series for seliciclib or 5,6-dichlorobenzimidazole riboside (DRB) was prepared in medium, added to cells and incubated for 72 hours. A 20% stock of alamar blue (Roche, Lewes, United Kingdom) was prepared in medium, added to each well and incubated for 2 hours. Absorbance was read at 488 to 595 nm and data were analyzed (Excel Fit v4.0) to determine the IC₅₀ (concentration of compound that inhibited cell growth by 50%) for each compound.

To test the effect of exposure time to seliciclib on viability, cells were seeded in flasks at 0.3 × 10⁵ cells/mL, treated with DMSO or seliciclib, and aliquots removed at specific time points. These cells were centrifuged at 800 × g for 5 minutes, washed and replated into 96-well plates in the absence of drug. Cell viability was determined, using alamar blue, 72 hours after the start of the experiment.

Preparation and analysis of cell lysates by immunoblotting. Cells were seeded at 0.4 × 10⁶ cells/mL in T25 flasks and treated with either 30 μmol/L seliciclib, 60 μmol/L DRB, 300 nmol/L MG132, or DMSO. Cells were removed at a variety of time points, washed twice with PBS, and the cell pellets snap-frozen in liquid nitrogen prior to storage at −70°C. Cells were resuspended in high salt buffer [50 mmol/L Tris-HCl (pH 7.6), 1% NP40, 0.4 mol/L NaCl, 5 mmol/L EDTA, 5 mmol/L DTT, 1 mmol/L phenylmethylsulfonyl fluoride, 2 mmol/L NaF, 1 mmol/L Na₃VO₄, 20 mmol/L β-glycerol phosphate, 5 mmol/L sodium pyrophosphate, and a protease inhibitor cocktail] and allowed to lyse for 15 minutes prior to sonication for 5 seconds at 5 amp (Sanyo soniprep 150). The protein concentration of each lysate was determined using a Bradford reagent assay (Bio-Rad, Hemel Hempstead, United Kingdom).

Lysate (30 μg) was mixed with gel loading buffer containing reducing agent and separated in either 3% to 8% or 10% polyacrylamide gels using denaturing electrophoretic conditions (Invitrogen, Glasgow, United Kingdom). Proteins were transferred to nitrocellulose membranes (Hybond ECL, Amersham, Chalfont St. Giles, United Kingdom) using wet electrophoretic transfer. Membranes were stained with Ponceau S to confirm equal loading before blocking in 5% nonfat milk in PBS with 0.1% Tween 20 (PBSTM) for 1 hour. Membranes were incubated overnight at 4°C with primary antibody, diluted in PBSTM. Membranes were washed in PBS and 0.1% Tween 20 (PBST) and incubated for 1 hour in PBSTM containing horseradish peroxidase-conjugated secondary antibody. Membranes were washed and incubated with enhanced chemiluminescence solution (Amersham) and exposed to X-ray film (Amersham).

Antibodies used in this study were: RNA Polymerase II (MMR-126R, Covance, Cambridge, United Kingdom), RNA Polymerase II carboxyl-terminal domain phosphoserine 2 (MMR-129R, Covance), RNA Polymerase II carboxyl-terminal domain phosphoserine 5 (MMR-134R, Covance), PARP (H250 Santa Cruz Biotechnology, Santa Cruz, CA), Hdm2 (SMP14, Santa Cruz), p53 (DO1 Calbiochem, Nottingham, United Kingdom), Mcl-1 (S-19, Santa Cruz), Bcl-2 (clone 124 Upstate, Milton Keynes, United Kingdom), XIAP (610763, Becton Dickinson, Cowley, United Kingdom), survivin (NB500-201, Novus Biologicals, Littleton, CO), pRb (554136, Becton Dickinson), and pRb phospho 249/252 (44-584, Biosource, Nivelles, Belgium).

Terminal deoxynucleotidyltransferase dUTP nick end labeling assays. Cells were seeded, treated with DMSO or 30 μmol/L seliciclib and harvested at specific time points. Cells were washed in PBS, fixed for 5 minutes on ice in 1% w/v paraformaldehyde, washed in cold PBS and resuspended in 70% ethanol (−20°C), prior to storage at −20°C. Labeling was carried out using the APO-DIRECT kit (Becton Dickinson). Briefly, 1 million cells were incubated for 1 hour at 37°C in the presence of terminal deoxynucleotide transferase enzyme and FITC-dUTP to label DNA breaks; washed and then incubated in propidium iodide (5 μg/mL) and RNase (200 μg/mL) for 30 minutes at room temperature to label DNA. Cells were analyzed using a Becton Dickinson LSR flow cytometer. The argon laser was used as an excitation source (488 nm). Terminal deoxynucleotidyltransferase dUTP nick end labeling (TUNEL)-positive cells were designated on the basis of green fluorescence (530 ± 28 nm) acquired on a logarithmic scale. DNA content was determined using red fluorescence (575 ± 26 nm) on a linear scale and pulse width analysis was used to exclude cell doublets and aggregates from the analysis. Cells exhibiting < 2n DNA content were designated as sub-G₁ cells. Only cells with 2n-4n DNA content were included in the TUNEL analysis.

Real-time PCR. Cells were treated and harvested identically to those prepared for immunoblotting. Cell pellets were lysed and RNA extracted using RNeasy kits (Qiagen, Crawley, United Kingdom). The RNA was quantified and normalized to 1 mg/mL prior to storage at −70°C. Quantitative real-time PCR was done on a Roche LightCycler using equal amounts of total RNA. The sequence of the primers were as follows:

• Mcl-1 (5′-GGTGGCATCAGGAATGT-3′)

• (5′-ATCAATGGGGAGCACT-3′)

• 28S rRNA (5′-AAGCAGGAGGTGTCAGAAA-3′)

• (5′-GTAAAACTAACCTGTCTCACG-3′)

The RNA Master SYBR Green I kit (Roche) was used to determine the relative amount of each mRNA. Samples were prepared in duplicate and PCR reactions were also run in duplicate. The change in expression in the presence of compound compared with the DMSO control was calculated using the formula: fold change in expression = 2^ΔCt, where 2 is the maximum efficiency of the PCR reaction and ΔCt is the change in crossing point values (sample Ct − DMSO control Ct).

Transfer of short interfering RNA by reversible cell permeabilization. Streptolysin O was used to reversibly permeabilize H929 cells using a modification of a published protocol (28). In brief, streptolysin O was suspended at 1,000 units/mL in Mg²⁺- and Ca²⁺-free PBS containing 0.01% bovine serum albumin and activated by adding DTT to 5 mmol/L prior to incubating for 2 hours at 37°C. Cells were washed twice in RPMI 1640 and resuspended at 25 × 10⁶ cells/mL. Activated streptolysin O was added to the wells of a 48-well plate at 14 units/10⁶ cells. The relevant short interfering RNAs (siRNAs) were added to each well (final concentration 10 μmol/L) followed by 5 × 10⁶ cells and incubated at 37°C 5% CO₂ with occasional mixing. After 10 minutes, growth medium containing 10% FCS (1 mL) was added to each well and the cells were returned to 37°C. After 30 minutes, growth medium containing 10% FCS (9 mL) was added and the cells were incubated for 24 hours before harvesting. The Mcl-1 siRNA was a smart pool consisting of four independent siRNAs designed using cDNA sequence: GI33519457 (Dharmacon, Lafayette, CO).

Annexin V staining. Cells were centrifuged at 500 × g for 5 minutes, washed twice in PBS and once in annexin buffer [10 mmol/L Hepes (pH 7.4), 2.5 mmol/L CaCl₂, and 140 mmol/L NaCl]. Cells were resuspended at 1 × 10⁶/mL and 100 μL was transferred to a 5 mL tube prior to incubation for 10 minutes in the dark at room temperature with 5 μL of annexin stain (Becton Dickinson) and 10 μL of propidium iodide (50 mg/mL). Annexin buffer (1 mL) was added and the cells were analyzed by flow cytometry. Annexin V–positive cells were designated on the basis of green fluorescence and propidium iodide–positive cells were designated on the basis of red fluorescence.

The effect of seliciclib and DRB on multiple myeloma cell viability. The IC₅₀ for seliciclib was determined in five myeloma cell lines. Cells were seeded in triplicate in 96-well plates, treated with serial dilutions of seliciclib and tested for viability after 72 hours. The average IC₅₀ after seliciclib treatment was 15.5 μmol/L (Table 1) with the most sensitive cell lines being H929 (8.84 μmol/L) and LP-1 (12.68 μmol/L), and the most resistant being RPMI 8226 (19.5 μmol/L). The IC₅₀ was also determined following 24 hours exposure to seliciclib, these values were equivalent to the 72-hour IC₅₀, indicating that a relatively short exposure to seliciclib was sufficient to kill cells (data not shown). In addition to the seliciclib IC₅₀ for these myeloma cell lines, the IC₅₀ of DRB was also determined (Table 1). DRB is a well-characterized inhibitor of transcription, that principally inhibits CDK9 (29), and in this respect, DRB has a similar mode-of-action to another established CDK inhibitor, flavopiridol (30–32). In general, the IC₅₀ values for DRB were higher than those for seliciclib but a similar profile of sensitivities between the cell lines was detected (Table 1).

Seliciclib induces rapid apoptosis in multiple myeloma cells. H929, LP-1, and RPMI 8226 cells were treated with either DMSO (control) or 30 μmol/L seliciclib (twice the average IC₅₀) and harvested at specific time points. Cells were fixed and stained using TUNEL to identify the proportion of cells containing fragmented DNA, a late stage marker of apoptosis (33). Induction of TUNEL was increased following 3 hours of seliciclib treatment and continued to accumulate to ≥60% by 24 hours (Fig. 1). As reported previously (8), apoptosis occurred from all phases of the cell cycle (data not shown).

Exposure time required to reduce myeloma cell viability. To determine the minimum exposure time required to induce maximum cytotoxicity, cells were treated with DMSO or seliciclib; at specific time points, cells were washed and returned to fresh medium in the absence of seliciclib. After a total of 72 hours of growth, cell viability was determined. An 8-hour treatment with either 20 or 30 μmol/L seliciclib reduced the 72-hour viability by at least 50% in all three cells lines (Fig. 2). Treatment of RPMI 8226 for 8 hours induced a maximal effect at both concentrations (Fig. 2C) but longer exposure was required to induce a maximal cytotoxicity in H929 and LP-1 cells, 16 and 24 hours, respectively (Fig. 2A and B). These data, in addition to the previous results, indicate that exposure to seliciclib for ≥8 hours markedly reduces cell viability by inducing apoptosis.

Molecular changes in seliciclib-treated multiple myeloma cells. Because both seliciclib and DRB can inhibit CDKs responsible for RNA polymerase II phosphorylation, experiments were done to determine whether the mechanism by which seliciclib induced apoptosis was through an inhibition of transcription, in an analogous manner to DRB. H929 cells were treated with either DMSO, seliciclib (30 μmol/L), or DRB (60 μmol/L) and collected at specific time points; these compound concentrations represent levels of equal potency, that is, twice the average IC₅₀ value. To examine the molecular changes in seliciclib-treated multiple myeloma cells, the level and phosphorylation state of proteins important for RNA polymerase II–dependent transcription and cell survival were studied. The use of low percentage acrylamide gels enabled the detection of two main forms of the RNA polymerase II large subunit. In DMSO-treated cells, the total RNA polymerase II antibody detected both the slow migrating hyperphosphorylated IIO form and the fast migrating hypophosphorylated IIA form, along with other intermediates as previously reported (34). In contrast, phosphoserine 2 and phosphoserine 5 were only detected in the hyperphosphorylated IIO form (Fig. 3, lanes 1-5). Treatment for 1.5 hours with either seliciclib or DRB resulted in the rapid reduction in the levels of phosphoserine 2 and phosphoserine 5 in the absence of changes in the total RNA polymerase II levels (Fig. 3, lanes 6 and 10). Careful analysis of the changes in mobility detected with these antibodies showed subtle differences between the effects of seliciclib and DRB on the mobility of RNA polymerase II. The faster migrating hypophosphorylated IIA form was detected with the anti-phosphoserine 5 antibody after treatment with DRB but not seliciclib. Furthermore, the total RNA polymerase II antibody detected less protein in the hyperphosphorylated IIO position after treatment with DRB compared with seliciclib. This may reflect the differences in kinase specificities, suggesting that DRB exerts greater effects than seliciclib, leading to a more significant increase in the hypophosphorylated IIA form.

The phosphorylation status of pRb was analyzed at the S249/T252 sites, which are reported to be sites for CDK phosphorylation (35, 36). Dephosphorylation at S249/T252 occurred after 1.5 hours of seliciclib treatment in the absence of changes in total pRb levels (Fig. 3, lane 9). Dephosphorylation of pRb occurred with similar kinetics to that of RNA polymerase II after seliciclib treatment. This suggests that S249/T252 and RNA polymerase II may be phosphorylated by the same kinase(s). DRB treatment did not result in rapid changes in S249/T252 phosphorylation, although by 8 hours, dephosphorylation of S249/T252 was detected (Fig. 3). These data again reflect the differences between kinases inhibited by seliciclib and DRB.

The ability of seliciclib to inhibit phosphorylation of the carboxyl-terminal domain of RNA polymerase II in a similar manner to DRB suggests that seliciclib could also be inhibiting transcription. To explore this further, the effects of seliciclib and DRB on the levels of key cell survival proteins were investigated. The levels of Mcl-1 protein rapidly decreased after treatment for 3 hours with either compound (Fig. 3, lanes 7 and 11), whereas little change was detected in the levels of XIAP, survivin, or Bcl-2, consistent with Mcl-1 having a very short-lived mRNA and protein. The decrease in Mcl-1 preceded an increase in the level of cleaved PARP, a well-characterized marker of apoptosis (37) that was detectable following 5 hours of treatment with either seliciclib or DRB (Fig. 3, lanes 8 and 12). Hdm2, like Mcl-1, has a very short-lived mRNA and protein (38), and similarly, levels of Hdm2 rapidly declined after either seliciclib or DRB treatment (Fig. 3, lanes 7 and 11). Because Hdm2 is an E3 ubiquitin ligase for p53 (39, 40), the reduction in the level of Hdm2 was followed by the expected accumulation in p53 levels (Fig. 3, lanes 7 and 11).

To determine whether these were general effects of seliciclib on myeloma cells, the studies were expanded to evaluate the cellular effects of seliciclib on LP-1 and RPMI 8226 cells. Cells were treated with 30 μmol/L seliciclib and harvested at various time points. As seen in H929 cells, treatment with seliciclib induced rapid dephosphorylation of serine 2 on the carboxyl-terminal domain of RNA polymerase II in both cell lines (Fig. 4A, and B, lane 7). Dephosphorylation of serine 2 was followed by a reduction in the levels of Mcl-1 that began to decrease after 1.5 hours of treatment (Fig. 4A, and B, lane 7). The loss of Mcl-1 again preceded the onset of apoptosis that was detected by PARP cleavage at 3 to 5 hours (Fig. 4A  and B, lane 8). Taken together, these results suggest that in myeloma cells, seliciclib like DRB, rapidly inhibits phosphorylation of the carboxyl-terminal domain of RNA polymerase II, resulting in down-regulation of Mcl-1 prior to the induction of apoptosis.

Seliciclib inhibits transcription of Mcl-1. To confirm that the loss of Mcl-1 protein was due to a block in transcription, not an activation of a proteolytic pathway or changes in translation rates, the levels of Mcl-1 mRNA following seliciclib treatment were measured by real-time PCR. Total RNA was prepared from cells treated with DMSO or 30 μmol/L seliciclib. Previously, it has been shown that roscovitine does not affect the production of rRNA (13), therefore 28S rRNA was used as a control for total RNA levels. Seliciclib caused a rapid reduction by 2- to 5-fold of Mcl-1 mRNA in H929 cells and LP-1 cells with the levels remaining decreased for several hours (Fig. 5). These results confirm that the loss of Mcl-1 protein correlated with an inhibition of Mcl-1 transcription due to the inhibition of the carboxyl-terminal domain phosphorylation by seliciclib.

Loss of Mcl-1 is specific to seliciclib treatment and not a general prerequisite for the induction of apoptosis. Following seliciclib treatment, the decrease in Mcl-1 protein prior to PARP cleavage suggested that this might be part of the mechanism by which seliciclib induces apoptosis. To show that the loss of Mcl-1 was not just a function of the apoptotic process and thereby not directly related to seliciclib's mode of action, the impact of a second apoptotic-inducing agent on Mcl-1 levels was studied. MG132 is a proteosome inhibitor, which as a class of compounds are known to effectively induce apoptosis in myeloma cells (41, 42). MG132, as a proteosome inhibitor, therefore has a very distinct mode-of-action from seliciclib, thus if both compounds caused a decrease in Mcl-1 levels prior to the induction of apoptosis, this would suggest that the loss of Mcl-1 is a fundamental part of the apoptotic process rather than specifically related to seliciclib's mode-of-action. LP-1 cells were treated with DMSO or 300 nmol/L MG132 and collected at various time points. As an indicator of apoptotic induction, PARP cleavage was clearly detected following 16 hours of MG132 treatment without any apparent change in the levels of Mcl-1 (Fig. 6A). Thus, following MG132 treatment, the induction of PARP cleavage occurred before a decrease in Mcl-1 levels that was in distinct contrast to seliciclib treatment where the Mcl-1 decrease preceded PARP cleavage (Figs. 3 and 4). Correspondingly the loss of Mcl-1 prior to the onset of apoptosis is characteristic for seliciclib; and because Mcl-1 is reported to be essential for myeloma cell survival (22, 23), this could cause the onset of apoptosis seen in seliciclib-treated myeloma cells.

Mcl-1 is required for survival of multiple myeloma cells. To confirm that Mcl-1 is vital for survival of multiple myeloma cells, Mcl-1 expression was inhibited using siRNAs. H929 cells were used in these experiments because the importance of Mcl-1 in the survival of this cell line has not been determined previously unlike LP-1 and RPMI 8226 cells where Mcl-1 antisense has been used to show that it is essential (22, 23). Transfection was accomplished using streptolysin O and transfection efficiency was judged to be 85% using FITC dye penetration (data not shown). Mcl-1 siRNA, but not control gl3 siRNA (43) caused a dramatic reduction in the levels of Mcl-1 protein (Fig. 6B) but no change in the levels of the related protein Bcl-2. Furthermore, following 24 hours of treatment, the Mcl-1 siRNA induced apoptosis detectable by PARP cleavage (Fig. 6) and quantified by flow cytometry using TUNEL (∼70% of cells) and annexin V staining (∼80% of cells). These results show the importance of Mcl-1 for the survival of multiple myeloma cells. Therefore, the loss of Mcl-1 caused by seliciclib's inhibition of transcription would be sufficient to induce apoptosis in these cells.

This study reports that the CDK inhibitor, seliciclib (CYC202, R-roscovitine), inhibits myeloma cell growth in a time- and dose-dependent manner (Figs. 1 and 2). Evidence is provided that seliciclib acts by inhibiting kinases that phosphorylate the carboxyl-terminal domain of the large subunit of RNA polymerase II resulting in the inhibition of transcription. This leads to the rapid down-regulation of Mcl-1 mRNA and protein which is required for myeloma cell survival (Figs. 3-6).

The targeting of cyclin-dependent kinases by seliciclib in myeloma cells. Accumulating lines of evidence suggest that CDK7 and CDK9 are implicated in the activation of transcriptional initiation and elongation, respectively, via direct phosphorylation of the carboxyl-terminal domain of RNA polymerase II (6). CDK7 can phosphorylate the serine 5 position whereas CDK9 phosphorylates serines at both the 2 and 5 positions. Phosphorylation of the carboxyl-terminal domain of RNA polymerase II at these sites is central not only to transcription but also mRNA capping, splicing, cleavage, and polyadenylation. This is because the large flexible carboxyl-terminal domain acts as a landing pad for docking of a range of factors required for these processes and it is believed that phosphorylation regulates factor binding (44). The in vitro kinase specificity for seliciclib shows that it principally inhibits CDK2/cyclin E, CDK7/cyclin H, and CDK9/cyclin T (8, 9). In seliciclib-treated myeloma cells, the phosphorylation of the serine 2 and 5 positions of the carboxyl-terminal domain of RNA polymerase II was rapidly inhibited, suggesting that seliciclib directly inhibits the kinases responsible for these phosphorylation events. The potent in vitro inhibition of CDK7 and CDK9 by seliciclib, together with the critical involvement of these kinases in carboxyl-terminal domain phosphorylation, suggests that CDK7 and CDK9 are both cellular targets of seliciclib. Inhibition of carboxyl-terminal domain phosphorylation provides a link between kinase activity and the general transcriptional inhibition by roscovitine previously reported (11, 31). In these reports, measurement of the levels of newly synthesized mRNA showed a 70% decrease after 2 hours of treatment with 25 μmol/L roscovitine (11); whereas gene expression analysis showed that roscovitine inhibited the transcription of a subset of genes but the effect did not seem to be as widespread as other transcriptional inhibitors, including DRB and flavopiridol, at doses that gave equivalent cellular effects (31). The latter result is of particular interest because both DRB and flavopiridol are primarily CDK9 inhibitors, unlike seliciclib, and both have broader effects on transcription (29–31, 45). In vitro characterization of the kinases inhibited by DRB shows a 25-fold selectivity for CDK9 over both CDK7 and CDK2 (29, 46).¹

Seliciclib on the other hand, whereas approximately equipotent as DRB against CDK9 (IC₅₀ = 0.81 μmol/L) is significantly more active against CDK7 (IC₅₀ = 0.36 μmol/L) and even more so against CDK2 (IC₅₀ = 0.1 μmol/L; refs. 8, 9). The differences in kinase specificity between seliciclib and DRB would explain the subtle, but distinct differences seen in the levels of hyperphosphorylated IIO and hypophosphorylated IIA forms of RNA polymerase II following treatment with these compounds (Fig. 3).

The rapid dephosphorylation of pRb at S249/T252 after seliciclib but not DRB treatment identifies this phosphorylation site as a marker of seliciclib activity (Fig. 3). The S249/T252 sites became dephosphorylated at the same rate as the carboxyl-terminal domain of RNA polymerase II suggesting the same kinase(s) maybe responsible. The cellular kinase(s) responsible for these phosphorylations have not been determined yet, although in vitro kinase data show that these sites can be phosphorylated by CDK4/cyclin D (36). It would seem unlikely that S249/T252 are phosphorylated by CDK4/cyclin D in cells, due to the lack of potency of seliciclib towards CDK4/cyclin D (8), especially considering that the original in vitro work did not examine the potential of CDK7 or CDK9 to phosphorylate these sites (36). Seliciclib has also been shown to down-regulate other phosphorylation sites on pRb including S780, S807/S811, and T821 (15).²

Significantly longer exposure was required, however, to reduce the levels of phosphorylation at these sites. Considering the rapid decrease in phosphorylation state of the S249/S252 sites in this study (1.5 hours), these data suggest that the decrease in the phosphorylation of the other pRb sites may be caused by subsequent effects on downstream kinases. For example, CDK7 as part of the CDK-activating kinase complex, phosphorylates and activates a range of kinases (47), therefore, inhibition of CDK7 by seliciclib could indirectly inhibit the activities of other CDKs that phosphorylate pRb. Alternatively, the inhibition of transcription by seliciclib could lead to the down-regulation of the levels of cyclins or CDKs that regulate these sites. Finally, although it is likely that S780, S807/S811, and T821 are targeted by different kinases to the one that phosphorylates S249/T252, the explanation that the kinase is the same but the rate of dephosphorylation of S780, S807/S811, and T821 may be slower, cannot be excluded completely.

The mechanism of seliciclib-induced apoptosis in myeloma cells. The effects of seliciclib on RNA polymerase II suggest that one potential mechanism for seliciclib-induced cell death may be through the inhibition of transcription. Cell survival factors are often highly regulated at both the mRNA and protein level, therefore, potentially representing key candidates to be affected by a transcriptional block. Investigations into the effects of seliciclib on a number of short half-life antiapoptotic proteins found that rapid changes in Mcl-1 mRNA and protein were detected (Figs. 3-5). Mcl-1 protein has a reported half-life of 30 minutes (48) potentially as a consequence of having proteolytic degradation motifs that are found in highly regulated proteins (49). In myeloma cells, the loss of Mcl-1 was followed rapidly by the onset of apoptosis as detected by PARP cleavage and TUNEL analysis. Mcl-1 degradation was not a general prerequisite for the initiation of the apoptotic process, because treatment of myeloma cells with the proteosome inhibitor, MG132, induced apoptosis in the absence of changes in Mcl-1 protein levels (Fig. 6A). It should be noted that at later time points following MG132 treatment, Mcl-1 levels were decreased, but these changes occurred several hours after the induction of apoptosis. Mcl-1 degradation at this late stage was probably because Mcl-1 is a target for proteolytic cleavage by caspases (25). Previously, antisense RNA against Mcl-1 has been used to show that Mcl-1 is essential for the survival of some myeloma cell lines including RPMI 8226 and LP-1 (22, 23). In this study, siRNAs were used to show the essential role of Mcl-1 for the survival of H929 myeloma cells and that specific down-regulation of Mcl-1 induced apoptosis in the majority of cells. Therefore, the cytotoxic effect of seliciclib in myeloma cells can be explained by the rapid loss of Mcl-1 following the inhibition of RNA polymerase II phosphorylation.

Seliciclib treatment lead to an induction of p53 in H929 cells (Fig. 3), whereas no such effect was seen in LP-1 and RPMI 8226 cells; probably due to the presence of mutant p53 in these cell lines (50). The p53 protein is central to the apoptotic pathway induced by cellular stress and is tightly regulated by Hdm2 (51), which serves as an E3 ubiquitin ligase, targeting p53 for proteolytic degradation. Both Hdm2 mRNA and protein are known to have short half-lives, and like Mcl-1, Hdm2 protein levels were reduced rapidly by seliciclib treatment. As would be expected, p53 started to accumulate shortly after the loss of Hdm2 (Fig. 3). Roscovitine treatment of other cancer cell lines also causes a decrease in Hdm2 levels and induction of p53 (11, 12). Furthermore, a decrease in Hdm2 and induction of p53 also occurs after flavopiridol (52) or DRB treatment (this study). This work and others (8) show that seliciclib kills cells with similar potency regardless of whether or not they contain wild-type p53. Therefore, although the activation of the p53 pathway is a potent mechanism of cell-killing, it does not seem to be the primary contributor to the overall effects of seliciclib.

The characteristics of multiple myeloma cells—slow growth and evasion of apoptosis—make myeloma a particularly difficult disease to treat. The seliciclib-induced down-regulation of the antiapoptotic protein Mcl-1 seems to overcome the evasion of programmed cell death in multiple myeloma cells by switching the balance towards apoptosis. The use of seliciclib to treat myeloma has intriguing possibilities not only as a single agent but also in combination with other compounds. This study provides the scientific rationale for a clinical study to investigate the potential effects of seliciclib for the treatment of multiple myeloma. In addition, this work also identifies antibodies against a number of phosphoproteins as potential biomarkers of seliciclib activity that directly relate to its mechanism of action. Such biomarkers may be useful to show the mode of action of seliciclib in vivo using patient biopsy samples.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We are grateful to Wayne Jackson for in vitro kinase assay results and to Ian Fleming and other Cyclacel colleagues for insightful discussions and critical reading of the manuscript.