BI 6727, A Polo-like Kinase Inhibitor with Improved Pharmacokinetic Profile and Broad Antitumor Activity

In eukaryotes from yeast to men, orthologues of Polo-like kinase 1 (Plk1) control multiple essential steps of mitosis (reviewed in refs. 1, 2). In vertebrate cells, Plk1 has been implicated in the activation of Cdk1-cyclin B at mitotic entry (3), centrosome maturation (4), release of cohesin from chromosome arms (5), formation and maintenance of the bipolar spindle (6–8), mitotic exit (9), and cytokinesis (7, 10–13). In addition, Plk1 is a target of the DNA damage checkpoint (14, 15) and seems to be required for mitotic reentry after DNA damage checkpoint recovery (16). Plk1 mRNA and protein expression, its catalytic activity, and subcellular localization are tightly regulated during cell cycle progression (17). Plk1 is located at centrosomes and kinetochores during early mitosis, and a fraction translocates to the midzone/midbody late in mitosis (18). Based on its temporal and spatial expression patterns as well as genetic and biological evidence, Plk1 seems to function uniquely in the regulation of the cell division cycle during late G₂ and M phases. In previous studies, we have used small interfering RNA (siRNA) and a chemical biology approach with BI 2536 (6, 8, 13, 19) to validate this role of Plk1 in human cells. We observed that the cellular effects of down-regulating Plk1 mRNA and protein with siRNA are identical to selective inhibition of its serine/threonine kinase activity. Cellular effects of Plk1 inhibition include a distinctive “polo arrest” phenotype in prometaphase, accumulation of phospho–histone H3, and formation of aberrant mitotic spindles followed by a surge in apoptosis (6). Using BI 2536, these cellular phenotypes in tissue culture were linked to identical effects in vivo (19). Of note, three additional Polo-like kinase family members (Plk2/Snk, Plk3/Fnk, and Plk4/SAK) have been identified, but these are thought to function during the G₁ and S phases of the cell cycle (reviewed in ref. 17) and their inhibition apparently does not contribute to the BI 2536–induced cellular phenotype.

Consistent with its biological role, Plk1 mRNA and protein are highly expressed in proliferating tissues and Plk1 seems to be overexpressed in a broad spectrum of cancers, with highest expression levels being correlated with poor prognosis in several cancer types (20). Taken together, these observations highlight Plk1 as an attractive molecular target for the antimitotic approach to cancer therapy (21). Clearly, in this category of antiproliferative agents, the approved tubulin-binding taxanes and Vinca alkaloids serve as reference points for clinical efficacy, side effects, and convenience of drug administration. Three lines of continued drug development can be discerned. First, efforts are made to improve the formulation of existing microtubule-stabilizing (taxanes) and destabilizing (Vinca alkaloids) drugs, to develop close analogues suitable for oral administration, and to develop novel chemical series of antimicrotubule agents, like epothilones, to overcome taxane resistance (22). Second, mitotic kinesins have been identified as microtubule-associated motor proteins and ATPases like Eg5 or CENP-E represent novel drug targets that provide specificity for the cell division–related function of microtubules (23, 24). Finally, the cell cycle–regulatory kinases CDK1, Aurora B, and Plk1 are tubulin-independent targets for mitotic inhibitors that are readily accessible to medicinal chemistry approaches (25).

Of these three potential mitotic kinase targets, Aurora B has thus far been targeted extensively by drug candidates. The results of the first clinical trials with these Aurora B inhibitors are pending (26). Plk1 inhibitors have emerged more recently (7, 19, 27–29) with BI 2536 as the first compound to enter clinical studies. ON01910 has been reported earlier to act via Plk1 (30), but this notion was not corroborated in subsequent studies (6, 29) and its mode of action remains to be defined. While the first clinical studies of Plk1 inhibitors are ongoing, we and others (7, 19, 27–29) are continuing to broaden the range of clinical candidates targeting Plk1. These efforts are aided by the lessons learned from our previous program on the chemical optimization of dihydropteridinones and emerging X-ray crystallographic data regarding the basis of high Plk1 kinase selectivity (31, 32).

Here, we report on BI 6727, a dihydropteridinone derivative with high potency and selectivity. In preclinical studies, a high volume of distribution, indicating good tissue penetration, and a long terminal half-life have emerged as distinct features of BI 6727, which may have a favorable effect on antitumor efficacy in vivo. The results of ongoing clinical studies in cancer patients will allow a comparison with human pharmacokinetic and pharmacodynamic parameters in the future.

Chemistry. The dihydropteridinone BI 6727 was derived from a chemical lead optimization program designed to generate Plk inhibitors (Patent Application WO 2004/076454).

Crystal structure. Protein expression of the T210V mutant kinase domain was done using a baculovirus/insect cell expression system. A homogeneity of >95% as judged by Coomassie-stained SDS-PAGE was achieved after a chromatographic purification procedure including affinity, ion exchange, and gel filtration chromatography. Protein crystallization was done using the sitting drop method with a reservoir containing 3% to 16% PEG 3350-6000 and 0.2 to 0.5 mol/L sodium malonate (31). Protein-ligand complexation was done before crystallization by incubation of the ligand for 2 h at 4°C. Crystals were frozen using the Proteros Free Mounting System. Data were collected at the SLS beam line X06SA (Swiss Light Source, Paul Scherrer Institute) using the PILATUS 6M detector. Images were processed with XDS; structure solution was achieved by Patterson search methods and model building; and refinement was done with COOT and REFMAC. Protein crystallography was done by Proteros biostructures GmbH. The structure has been deposited in the Protein Data Bank under the access code 3FC2.

In vitro kinase assays. Recombinant human Plk1 (residues 1-603) was expressed as NH₂-terminal, GST-tagged fusion protein using a baculoviral expression system (BaculoGold, BD Biosciences) and purified by affinity chromatography using glutathione-agarose (Amersham Biosciences). NH₂-terminal His₆-tagged recombinant human Plk3 (residues 19-301) was purchased from Upstate Cell Signaling Solutions, whereas recombinant human Plk2 was obtained from Invitrogen. Enzyme activity assays for Plk1, Plk2, and Plk3 were done in the presence of serially diluted inhibitor using 20 ng of recombinant kinase and 10 μg casein from bovine milk (Sigma) as substrate. Kinase reactions were done in a final volume of 60 μL for 45 min at 30°C [15 mmol/L MgCl₂, 25 mmol/L MOPS (pH 7.0), 1 mmol/L DTT, 1% DMSO, 7.5 μmol/L ATP, 0.3 μCi γ-³²P-ATP]. Reactions were terminated by the addition of 125 μL of ice-cold 5% TCA. After transferring the precipitates to MultiScreen mixed ester cellulose filter plates (Millipore), plates were washed with 1% TCA and quantified radiometrically. Dose-response curves were used for calculating IC₅₀ values. To establish a kinase selectivity profile, additional kinase assays were done by contract research organizations or reagents were purchased from commercial sources and assays were done according to the supplier's instructions. Appropriate positive and negative controls were included in the assay design.

Cell culture and proliferation assays. Cancer cell lines were obtained from the American Type Culture Collection or the German National Resource Center for Biological Material and were cultured according to the supplier's instructions. Cell proliferation assays were done by incubating cells in the presence of various concentrations of BI 6727 for 72 h and cell growth was assessed by measuring Alamar blue dye (Serotec) conversion in a fluorescence spectrophotometer. Effective concentrations at which cellular growth was inhibited by 50% (EC₅₀) were extrapolated from the dose-response curve fit.

Cell cycle analysis. Logarithmically growing NCI-H460 cells were incubated with 0.1% DMSO or various concentrations of BI 6727. To determine the DNA content, cell suspensions were fixed in 80% ethanol, treated for 5 min with 0.25% Triton X-100 in PBS, and incubated with 0.1% RNase and 10 μg/mL propidium iodide in PBS for 20 min at room temperature. Cell cycle profiles were determined by flow cytometric analysis (FACSCanto, BD Biosciences).

Immunofluorescence analysis. NCI-H460 cells were plated in four-well chamber slides. The next day, cells were treated for 24 h with various concentrations of BI 6727 or 0.1% DMSO as vehicle control. After fixation and permeabilization of the cells with 4% formaldehyde/Triton X-100 (1:200) for 5 min at room temperature, immunofluorescence staining was done according to standard procedures. Briefly, antitubulin (Sigma T6199) and anti–phospho-histone H3 (Upstate 07-145) antibodies were diluted 1:500 in PBS/2% bovine serum albumin. 4′,6-Diamidino-2-phenylindole dihydrochloride (Roche) was used at a final concentration of 0.05 μg/mL. Cells were mounted in Mowiol and analyzed using a Zeiss Axioplan 2 fluorescence microscope.

Efficacy studies in mice. Female BomTac:NMRI-Foxn1^nu mice (Taconic) were grafted s.c. with 2 × 10⁶ HCT 116 human colon carcinoma cells (ATCC CCL-247), 1 × 10⁶ NCI-H460 non–small cell lung cancer cells (ATCC HTB-177), or CXB1 human colon carcinoma tumor pieces derived from patient material by serial transplantation in nude mice. When tumors had reached a volume of ∼50 to 100 mm³, animals were randomized into treatment and control groups of 10 mice each. BI 6727 was formulated in hydrochloric acid (0.1 N), diluted with 0.9% NaCl, and injected i.v. into the tail vein at the indicated dose and schedule. For oral treatment, BI 6727 was resuspended in 0.5% Natrosol 250 hydroxyethyl-cellulose and given intragastrally via gavage needle. An administration volume of 10 mL per kilogram of body weight was used for both administration routes. Tumor volumes were determined thrice a week using a caliper. The results were converted to tumor volume (mm³) by the formula length × width² × π/6. The weight of the mice was determined as an indicator of tolerability on the same days. Median tumor volumes on the last day of the experiment were used to calculate treated versus control values (= tumor volume_{treated mice} × 100/tumor volume_{control mice}).

Immunohistochemistry. For ex vivo analysis of histone H3 phosphorylation and apoptosis in tumors, female BomTac:NMRI-Foxn1^nu mice were grafted s.c. with 2 × 10⁶ HCT 116 human colon carcinoma cells as described above. The average tumor volume was ∼200 mm³ at the start of the experiment. Tumors were excised from nude mice 6, 24, or 48 h after i.v. administration of a single dose of 40 mg/kg BI 6727 or vehicle control (n = 3 per treatment group and time point, total 18 mice) and were processed for immunohistochemistry. In addition, samples from the small and large intestines were obtained at the same time points.

Histone H3 phosphorylation was determined by the avidin-biotin immunoperoxidase method on frozen sections using an antibody to Ser¹⁰-phosphorylated histone H3 (Upstate) followed by hematoxylin counterstaining. Apoptosis was quantified with a terminal deoxyribonucleotide transferase–mediated nick-end labeling (TUNEL) assay, using the ApopTag Fluorescein in situ apoptosis detection kit (Chemicon International Inc.) and analyzed by fluorescence microscopy.

For image analysis, representative pictures were taken from each tumor with a Zeiss Axiophot microscope at 10× magnification. The mitotic index and the number of apoptotic cells were quantified on digital images using the Definiens Developer image analysis software (Definiens AG). Briefly, the area occupied by the tumor cells arrested in mitosis (pH3-positive cells) was calculated as a percentage of the total “tumor cell area” (#tumor cells + tumor stroma, excluding normal tissues). In addition, the area occupied by apoptotic tumor cells (TUNEL-positive cells) was also calculated as a percentage of the total tumor cell area. The statistical significance was determined by the Student's t test.

Pharmacokinetic studies. Female BomTac:NMRI-Foxn1^nu mice (Taconic) and male Wistar rats of the strain Crl:WI (Han; Charles River) were used for the pharmacokinetic studies. EDTA plasma was prepared by centrifugation of the collected blood at 10,000 rpm at 4°C. Before analysis, the plasma samples were added to an internal standard and 0.1 N NaOH. The analytes were extracted by liquid-liquid extraction using ethyl acetate. The resulting solution was dried in a warm nitrogen stream and dissolved in 25% methanol/0.1% formic acid.

The tissue samples (100-150 mg) were placed in a 2-mL tube containing Q-Biogene Lysing matrix D (Q-Biogene) and H₂O (6 μL/mg tissue). The mixture was vortexed twice in a Thermo Savant FastPrep FP120 for 30 s at 6.5 m/s. Calibration and quality control samples were added to the analytes after vortexing. Homogenates were diluted 1:1 with 50% methanol/0.1% formic acid containing internal standard for mass spectrometry. Then, acetonitrile was added. After centrifugation for 5 min at 10,000 rpm, the supernatant was used for quantification by liquid chromatography/tandem mass spectrometry.

Quantification of the analytes was done by high-performance liquid chromatography/tandem mass spectrometry (Agilent 1100, Applied Biosystems API3000, ESI+) with a solvent gradient of 95% A to 10% A in 1.5 min [A: 5 mmol/L ammonium acetate (pH 4.0), B: acetonitrile with 0.1% formic acid; Luna C18 3μ 2 × 30].

For the conversion of the tissue concentrations in nmol/mg to nmol/L, a tissue density of 1 was assumed. The area under the curve in the given time interval was calculated from the plasma concentration-time profiles by the trapezoidal rule. Pharmacokinetic parameters were calculated by noncompartmental analysis and terminal half-life was estimated from the last two time points of the concentration-time profile after i.v. administration.

Crystal structure of BI 6727 bound to Plk1. We have screened a diverse library of small molecules for their ability to inhibit the catalytic activity of Plk1 and identified a series of dihydropteridinones that were then further tailored for potency, selectivity, ability to inhibit cell proliferation in vitro, and a suitable pharmacokinetic profile. BI 6727 is a derivative of this chemical series (Fig. 1A, left); the structure of a cocrystal with the T210V mutant kinase domain of Plk1 (Fig. 1A, right) shows a binding mode similar to that previously reported for BI 2536 (32). BI 6727 binds to the hinge region between the NH₂-terminal and COOH-terminal lobes of the kinase domain via two hydrogen bonds from the dihydropteridinone core to the backbone NH and carbonyl group of Cys¹³³. This places BI 6727 in the ATP binding pocket of Plk1. The solvent-exposed basic benzamide-derived moiety of the compound was selected based on an extensive optimization program to further improve the pharmacokinetic properties of dihydropteridinones without changing the biochemical and cellular profile.³

Characterization of kinase specificity and cellular activity. BI 6727 potently inhibited Plk1 as well as the two closely related kinases Plk2 and Plk3 (IC₅₀ values 0.87, 5, and 56 nmol/L, respectively). In contrast, assays using a panel of >50 other kinases (listed in ref. 19) failed to identify any inhibitory activity at concentrations up to 10 μmol/L.

BI 6727 inhibited proliferation of multiple cell lines derived from various cancer tissues, including carcinomas of the colon (HCT 116, EC₅₀ = 23 nmol/L) and lung (NCI-H460, EC₅₀ = 21 nmol/L), melanoma (BRO, EC₅₀ = 11 nmol/L), and hematopoietic cancers (GRANTA-519, EC₅₀ = 15 nmol/L; HL-60, EC₅₀ = 32 nmol/L; THP-1, EC₅₀ = 36 nmol/L and Raji, EC₅₀ = 37 nmol/L) with EC₅₀ values of 11 to 37 nmol/L. Representative dose-response curves for NCI-H460 and HCT 116 cells are shown in Fig. 1B. Immunofluorescence microscopy (Fig. 1C) and fluorescence-activated cell sorting analysis for DNA content (Fig. 2A and B) were used to analyze the cellular effects of BI 6727 in more detail. Immunofluorescence analysis of NCI-H460 cells treated for 24 hours with 100 nmol/L BI 6727 showed an accumulation of mitotic cells with monopolar spindles and positive staining for histone H3 phosphoserine 10 (Fig. 1C), confirming that cells are arrested early in the M phase. Consistent with these data, treatment of NCI-H460 cells with increasing concentrations of BI 6727 for 24 hours resulted in accumulation of cells with 4N DNA content (Fig. 2A), indicative of a cell cycle block in G₂-M phase, a phenotype expected for Plk1 inhibition (8). Starting at a concentration of 100 nmol/L, treated cells showed a prominent G₂-M peak in the fluorescence-activated cell sorting profile. Using this inhibitor concentration, a time course experiment was done (Fig. 2B). G₂-M arrest at 24 hours was followed by induction of apoptosis at 48 hours as shown by the increase in the sub-G₁ peak of the fluorescence-activated cell sorting profile. Induction of apoptosis was further confirmed by Western blot analysis of poly(ADP)ribose polymerase cleavage in NCI-H460 cells treated with 100 nmol/L BI 6727. Cleaved poly(ADP)ribose polymerase fragment was already detected after 24 hours of treatment; after 48 hours, all detectable poly(ADP)ribose polymerase protein had been converted to the fragment (Fig. 2C).

Pharmacokinetic profile of BI 6727. To determine the pharmacokinetic profile of BI 6727 in rodents (Fig. 2D), a single dose of 10 mg/kg was given as an i.v. 5-minute infusion to male Wistar rats (n = 4). Plasma concentrations of BI 6727 were measured at various time points after start of the infusion using a high-performance liquid chromatography/tandem mass spectrometry assay and various pharmacokinetic parameters were calculated (Table 1). The total plasma clearance was low at 8.9 mL/min/kg, corresponding to <20% of the hepatic blood flow. In contrast, the volume of distribution at the steady state (V_ss) was very high at 22 L/kg, indicating excellent tissue penetration. A similar experiment was done in mice (n = 3), giving 35 mg/kg BI 6727 i.v. (Fig. 2D; Table 1). Plasma clearance was low at 10 mL/min/kg corresponding to ∼10% of the hepatic blood flow. Similar to rats, the V_ss was high at 7.6 L/kg. As part of our ADME analysis, we noted good oral bioavailability (F) of BI 6727 across species (mouse F = 41%, rat F = 55%, and dog F = 53%).

Efficacy of BI 6727 in human tumor xenograft models using i.v. and oral treatment schedules. We next investigated the ability of BI 6727 to inhibit growth of s.c. human carcinoma xenografts in nude mice. Consecutive cycles of treatment with BI 6727 given oral or i.v. for up to 6 weeks were efficacious in multiple xenograft models (Table 2). Treatment was well tolerated as judged by clinical signs and body weight changes (Fig. 3A and D and data not shown). Figure 3A and B shows results of experiments using the HCT 116 colon carcinoma model. We tested efficacy of BI 6727 after oral administration comparing three different administration schedules (once a week, twice a week, or daily; Fig. 3A). At a total weekly dose of 50 mg/kg, all treatment schedules showed comparable efficacy and were well tolerated. In the same model, i.v. administration of a daily dose of 15, 20, or 25 mg/kg on 2 consecutive days per week led to significant tumor growth delay and even tumor regression (Fig. 3B). Similar efficacy was observed in an independent experiment where 20 or 30 mg/kg BI 6727, respectively, were given i.v. once a week (Table 2). To test the efficacy of BI 6727 in a more rigorous setting, treatment of HCT 116 tumors was initiated when an average tumor size of ∼400 mm³ had been reached (Fig. 3C). Five cycles of 20 mg/kg BI 6727 given on 2 consecutive days per week resulted in tumor regression, whereas all vehicle-treated control animals showed progressive disease.

BI 6727 was also active in a non–small cell lung carcinoma xenograft model derived from NCI-H460 cells (Table 2). Oral doses of 70 mg/kg given once weekly or 10 mg/kg daily significantly delayed tumor growth (Table 2). Efficacy of BI 6727 was further studied in a model of taxane-resistant colon carcinoma (Fig. 3D). This model is based on serial transplantation of tumor fragments rather than inoculation of cultured cells, and the tissue architecture preserves typical features of human colon carcinoma. The efficacy of BI 6727 was compared with docetaxel in this model. Because docetaxel is given i.v. in clinical practice, we chose the i.v. administration route for both compounds. Whereas docetaxel given once a week at the maximally tolerated dose (note progressive weight loss in Fig. 3D, right) did not show efficacy, 15 mg/kg BI 6727 given i.v. on 2 consecutive days significantly suppressed tumor growth and was well tolerated (Fig. 3D, left and right). In summary, BI 6727 shows efficacy in multiple xenograft models using i.v. or oral administration schedules (Table 2).

Using immunohistochemical methods, we were able to show that tumor response to treatment with BI 6727 is accompanied by an increase in the mitotic index as well as an increase in apoptosis (Fig. 4). HCT 116 tumor-bearing nude mice were injected with a single dose of 40 mg/kg BI 6727 or vehicle control and tumor tissue as well as pieces of the small and large intestines were removed 6, 24, and 48 hours after administration of BI 6727. The mitotic index was determined on frozen sections by staining with an antibody to histone H3 phosphoserine 10 (pH3), and apoptotic cells were visualized using a TUNEL assay. Twenty-four hours after administration of BI 6727, a significant (13-fold) increase in mitotic cells was observed in these tumors compared with tumors from control mice (Fig. 4A). Apoptotic cells in the BI 6727–treated tumors were detected as early as 24 hours after administration of BI 6727 and their number increased further at 48 hours (4.5-fold increase compared with control tumors; Fig. 4B and data not shown). In contrast, the mitotic index and the number of apoptotic cells did not increase in the small intestine and large intestine samples examined after treatment with a single dose of BI 6727 compared with samples from control animals (data not shown), suggesting that the proliferation rate of the intestinal cells is not high enough to be detectably affected by a single dose of BI 6727. Nevertheless, BI 6727 is expected to inhibit mitosis of all dividing cancer as well as normal cells, as reported for BI 2536 (19).

Drug exposure in tissues. To further evaluate tissue exposure and residence time of BI 6727 in vivo, which is likely more decisive for efficacy than plasma exposure, a single dose of 35 mg/kg BI 6727 was given i.v. to nude mice bearing s.c. xenografts of the human HCT 116 colon carcinoma. Tumors, multiple organs, as well as plasma samples were collected at various time points after compound administration, and the concentration of BI 6727 in tissue extracts was determined (Fig. 4C). Although exposure differences between organs—possibly reflecting different perfusion rates—were observed, overall a high tissue exposure was determined in these extracts, consistent with the high volume of distribution of BI 6727. Interestingly, drug concentrations in tumor tissue were much higher (a maximum of 32 μmol/L was reached 8 hours postadministration) than in plasma (7 μmol/L at 1 hour) and longer lasting. One hundred sixty-eight hours after administration, a concentration of 4 μmol/L was measured in the tumor, whereas plasma levels were only 8 nmol/L. Although it is clear that concentrations measured in the tissue extracts do not reflect free compound levels, the data suggest that tumor cells are efficiently exposed to BI 6727 in vivo.

Targeting mitosis is an established approach for cancer therapy, and taxanes as well as Vinca alkaloids are currently used to treat a broad variety of cancer types. However, these drugs have significant limitations as they are ineffective for many cancer types and patients suffer from toxicities related to the broad requirement for microtubule function in critical cellular processes unrelated to mitosis. New mitosis-specific targets for cancer therapy have emerged in recent years, including the mitotic kinesins and the Plk and Aurora families of serine/threonine kinases, and a first wave of inhibitors are currently under evaluation in early clinical trials (33, 34). These novel antimitotic agents have the potential to provide similar or improved efficacy compared with microtubule-targeting agents combined with improved tolerability and are being explored to overcome primary and secondary resistance to these drugs.

The essential role of Plk1 during mitosis has been characterized in much detail (1). Although recent publications suggest additional physiologic roles for Plk1, i.e., in processes such as invasion, telomere stabilization, or the regulation of DNA topoisomerase IIα (35–37), the attractiveness of Plk1 as a target in all cancer cells irrespective of mutations in oncogenes or tumor suppressor genes, and therefore its potential relevance for cancer treatment across indications, remains high. In previous studies with BI 2536, we have opened up the evaluation of small-molecule kinase inhibitors to provide detailed insights into Plk1 function in cells and animal models and new options for cancer therapy (6, 13, 19). Like BI 2536, BI 6727 is an ATP-competitive kinase inhibitor from the dihydropteridinone class of compounds. Recently, additional chemical series providing inhibition of Plk1 have been described (7, 27–29). BI 6727 potently inhibits Plk1 kinase activity (IC₅₀ = 0.87 nmol/L) and the phenotype of cancer cells treated with BI 6727 over a wide concentration range is consistent with that induced by Plk1-specific siRNAi (8). At higher concentrations, BI 6727 also inhibits the related Polo-like kinase family members Plk2 and Plk3. Compared with Plk1, much less is known about the physiologic roles of these enzymes, and eventual consequences of inhibition of these enzymes in terms of efficacy and tolerability of drug candidates cannot currently be predicted. Whereas Plk1 transcripts are only expressed in proliferating cells, Plk2 and Plk3 transcripts show broad tissue distribution and are also expressed in postmitotic neurons (38). Plk2 may primarily function as a regulator of G₁- and early S-phase progression, of centriole duplication during the S phase, and may also play a role in the DNA damage checkpoint (17, 39–41). However, Plk2-deficient mouse embryonic fibroblasts show only subtle effects on cell cycle progression and Plk2−/− mice are viable, with embryos showing only slightly retarded overall growth and defects in skeletal development (42). Plk3 is a nucleolar protein; levels peak in the G₁ phase of the cell cycle and may regulate entry into the S phase (43, 44). Nevertheless, Plk3 is apparently not essential for cell cycle progression, as Plk3−/− mice do not show growth defects (45). Interestingly, Plk1 and Plk3 have recently been proposed as tumor suppressor genes based on the fact that Plk1+/− and Plk3−/− mice seem to develop tumors with somewhat increased frequency (45, 46).

In vitro tests with BI 6727 show promising activity against cells resistant to taxanes or Vinca alkaloids. Briefly, BI 6727 is able to inhibit the proliferation of NCI-H460 cells preselected for resistance to high concentrations of vincristine. Similar results were obtained with a melanoma cell line ectopically expressing the MDR1 gene (BROmdr, data not shown), suggesting that BI 6727 is either less sensitive to these resistance mechanisms than taxanes or Vinca alkaloids or that the high tissue permeability of BI 6727 may compensate for transporter-mediated drug efflux. Consistent with these in vitro observations, BI 6727 is highly efficacious not only in standard nude mouse xenograft models of human cancers but also in the taxane-resistant CXB1 xenograft model of colorectal cancer. We consider CXB1 to be particularly attractive for evaluation of drug resistance in vivo because it was neither genetically modified nor selected in vitro and shows histopathologic features much more akin to authentic human cancers than typical cell line–derived models. Moreover, it represents a cancer model where MDR1/P-gp drug efflux pump expression in a polarized epithelial tissue (data not shown) may contribute to drug resistance in a pharmacologically relevant manner.

Our results in xenograft models suggest that BI 6727 is suitable for use in a wide range of treatment schedules. However, in the clinical context, the relevant tissue levels and notably levels of free versus protein-bound drug as well as exchange rates from bound to free drug are difficult to assess and cannot simply be extrapolated from plasma pharmacokinetics. In a pragmatic approach, we have prioritized BI 6727 for clinical development based on pharmacokinetic characteristics indicative of high and sustained exposure in tumor tissue. It is worth noting that early clinical studies of BI 6727 given by i.v. infusion to cancer patients seem to confirm the predicted high volume of distribution (>3,600 liters) and long half-life (about 110 hours; ref. 47). In addition, although it is too early to assess the effect of these distinctive pharmacokinetic properties on efficacy, the phase I results provide first evidence of antitumor activity as well as a favorable safety profile, with reversible hematotoxicity as the main side effect (47).

No potential conflicts of interest were disclosed.

We thank the entire Plk1 Research and Development team for their contributions. All authors are employees of the Boehringer Ingelheim group of companies. We thank Proteros biostructures GmbH, Germany, for generating the cocrystal structure of Plk1 with BI 6727.