Investigation of antitumor effects of synthetic epothilone analogs in human myeloma models in vitro and in vivo

Abstract

26-Trifluoro-(E)-9,10-dehydro-12,13-desoxyepothilone B [Fludelone (Flu)] has shown broad antitumor activity in solid tumor models. In the present study, we showed, in vitro, that Flu significantly inhibited multiple myeloma (MM) cell proliferation (with 1-15 nM IC₅₀), whereas normal human bone marrow stromal cells (HS-27A and HS-5 lines) were relatively resistant (10- to 15-fold higher IC₅₀). Cell-cycle analysis demonstrated that Flu caused G₂/M phase arrest and induced cell apoptosis. After Flu treatment, caspase-3, -8, and -9 were activated, cytochrome c and second mitochondrial-derived activator of caspase were released to the cytosol, and c-Jun N-terminal kinase was activated, indicating that mitochondria were involved in the apoptosis. Flu toxicity to human hematopoietic stem cells was evaluated by CD34⁺ cell-apoptosis measurements and hematopoietic-progenitor assays. There was no significant toxicity to noncycling human CD34⁺ cells. We compared the efficacy of Flu with the epothilone analog 12,13-desoxyepothilone B (dEpoB) in xenograft nonobese diabetic/severe combined immunodeficient mouse models with subcutaneous or disseminated MM. Flu caused tumor disappearance in RPMI 8226 subcutaneous xenografts after only five doses of the drug (20 mg/kg of body weight), with no sign of relapse after 100 d of observation. In a disseminated CAG MM model, mice treated with Flu had a significantly decreased tumor burden, as determined by bioluminescence imaging, and prolonged overall survival vs. mice treated with dEpoB or vehicle control, indicating that Flu may be a promising agent for MM therapy.

The discovery, in the early 1970s, of paclitaxel in the bark of the Pacific yew (Taxus brevifolia) (1) heralded a revolution in the field of tumor therapy. Paclitaxel binds to microtubules, stabilizing them and disrupting their polymerization dynamics (2, 3). Despite paclitaxel's efficacy against various solid tumors (4), poor aqueous solubility and the development of pleiotropic drug resistance (5-8) hamper its clinical utility.

Epothilones have emerged as a highly promising family of microtubule-stabilizing agents. The epothilones have higher solubility in water than paclitaxel and are poorer substrates for the P-glycoprotein, proving more effective than paclitaxel against tumors showing multidrug resistance (5, 6). Epothilones also have a simpler molecular architecture, which has allowed for the total synthesis of the natural epothilones and many synthetic analogs.

Our group achieved the total synthesis, and the structure-activity relationship evaluation, of several epothilones (9-11). This work soon led to the synthesis of 12,13-desoxyepothilone B (dEpoB), which proved highly effective in various in vivo xenograft tumor models (10, 11). More recently, we reported the total synthesis of (E)-9,10-dehydro-12, 13-desoxyepothilone B (9) and 26-trifluoro-(E)-9,10-dehydro-12,13-desoxyepothilone B, named Fludelone (Flu) (12). In mice, Flu appears to have a particularly broad therapeutic index in vivo. It seems to be curative over a range of tumor types, without tumor relapse upon suspension of treatment. Moreover, Flu is orally available and is completely curative against human tumor xenografts by parenteral and oral therapy. In addition, treatment with Flu gave complete tumor remission against Taxol-resistant tumors (10, 12). These very promising antitumor data were obtained in s.c. solid tumor models; however, drug efficacy in orthotopic models has not been investigated. In the present study, we focused on the evaluation of the anti-multiple-myeloma (MM) effects of the second generation analog Flu, in comparison with dEpoB, in a disseminated MM model. Our data demonstrated that Flu had a profound antitumor activity in human MM both in vitro and in vivo, indicating that this compound might be a promising agent against MM, particularly in late-stage refractory disease.

Materials and Methods

Cell Lines and Primary Specimens. For information about cell lines and primary specimens, see Supporting Materials and Methods, which is published as supporting information on the PNAS web site.

Reagents. Flu and dEpoB, were synthesized in-house (the BioOrganic Laboratory at Memorial Sloan-Kettering Cancer Center). Paclitaxel was purchased from Sigma. For in vivo mouse-model experiments, Flu and dEpoB were dissolved in 50% DMSO/50% ethanol as a stock solution (30 mg/ml) and diluted with normal saline at the time of injection.

Cell-Viability Assay. Cell-viability measurement was assessed by a colorimetric assay using 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT, Sigma) and phenazine methosulfate (PMS, Sigma) according to the manufacturer's protocol.

Cell-Cycle Analysis. Cells were fixed with 66.6% (vol/vol) ethanol at-20°C overnight, resuspended in 50 μg/ml RNase A (diluted in PBS), and incubated for 30 min at 37°C. The samples were resuspended in 25 μg/ml propidium iodide and 38 mM sodium citrate buffer, pH 7.2. Flow cytometry was performed on a FACSCalibur automated system (Becton Dickinson), and data were analyzed by using the program flowjo (Tree Star, Ashland, OR).

Annexin-V-Binding and DNA-Fragmentation Assays. Based on cell-cycle analysis, the optimal concentration of dEpoB and Flu causing complete cell-cycle arrest was 50 nM; therefore, this concentration was used in the following in vitro study. MM cells were treated with or without 40 μM pancaspase inhibitor Z-Val-Ala-Asp-(OMe)-CH₂F (z-VAD-fmk, Calbiochem) for 1 h before adding the compounds. Cells were stained with annexin V phycoerythrin and 7-aminoactinomycin D (7-AAD, Sigma), following standard protocol. A DNA-fragmentation assay was performed by using a Suicide-Track DNA ladder isolation kit (Oncogene Research Products, San Diego).

Caspase-8 and -9 Fluorometric Assays. Caspase-8 and -9 activities were measured with assay kits, following the manufacturer's instructions (Calbiochem). MM cells treated with tumor-necrosis-factor-related apoptosis-inducing ligand (50 ng/ml) were used as a positive control. The result was expressed as the relative fluorescent units in three independent experiments.

Nonradioactive Stress-Protein-Activated Kinase (SPAK)/c-jun N-Terminal Kinase (JNK) Assay. The SPAK/JNK activity was determined by using a commercially available kit (Cell Signaling Technology, Beverly, MA). In brief, cells were lysed, and equal amounts of protein were incubated overnight with glutathione S-transferase (GST)-c-jun fusion (amino acids 1-89) protein beads. The immune complexes were resuspended in 50 μl of kinase assay buffer, supplemented with 200 μM ATP, and incubated for 30 min at 30°C, allowing SPAK to phosphorylate the c-jun substrate. Reactions were terminated by the addition of 25 μl of 3× SDS sample buffer and subjected to standard immunoblotting analysis.

Preparation of Cytosolic and Mitochondrial Extracts from CAG Cells. A mitochondrial fractionation kit (Active Motif, Carlsbad, CA) was used for the isolation of mitochondrial or cytosolic fractions, following the manufacturer's instructions.

Immunoblotting Analysis. Protein lysates were prepared by using Laemmli buffer, and protein concentration was determined by Lowry protein assay (Bio-Rad). An equal amount of protein (50 μg) was fractionated on a precast 4-20% Tris-glycine gel (Bio-Rad). After transfer, membranes were processed for standard Western blotting assay. The primary antibodies used were as follows: poly(adenosine diphosphate-ribose) polymerase (PARP, BD PharMingen); caspase-3 (BD PharMingen); anti-phospho-SPAK/JNK and anti-SPAK/JNK (Cell Signaling Technology); rabbit polyclonal Ab against actin (Sigma); and monoclonal second mitochondrial-derived activator of caspase (Smac)/Diablo antibody and rabbit polyclonal cytochrome c antibody (Cell Signaling Technology).

Establishment of Subcutaneous and Disseminated MM Xenografts and Therapy. Nonobese diabetic (NOD)/severe combined immunodeficient (SCID) mice (The Jackson Laboratory) were housed and maintained in facilities under an institute-approved animal protocol. For the s.c. xenograft MM RPMI 8226 mouse model, 10- to 12-wk-old female mice were sublethally irradiated with 3 cGy from a cesium γ-radiation source and inoculated with 10 × 10⁶ tumor cells in the right flank. When tumor volumes approached 100 mm³, the mice were divided into experimental cohorts of five to eight mice each. Injections (i.p.) of Flu or dEpoB (20 mg/kg of body weight, five doses (one dose every 2 d) then three doses (one dose every 3 d), were administered. Control mice were treated with the same amount of vehicle under the same protocol. Tumor volume was calculated by using the formula mm³ = 4/3πr³, where r = (length + width)/4, measuring the two largest perpendicular axes of the tumor.

For establishing a disseminated xenograft NOD/SCID mouse model, 10 × 10⁶ CAG cells, stably expressing the HSV-TK-eGFP-luciferase fusion protein (13), were injected intravenously via the tail vein. The tumor distribution was followed by serial whole-body noninvasive imaging of visible light emitted by luciferase-expressing MM cells, upon injection of mice with luciferin. At 7-10 d after tumor injection, a group of NOD/SCID mice with established disseminated MM was divided into four cohorts, with a statistically equivalent tumor burden in each of the cohorts. The cohorts were treated with either Flu or dEpoB alone (20 mg/kg of body weight, five doses, with one dose every 2 d followed by 6 d of rest and then five doses, with one dose every 3 d).

Bioluminescence Imaging (BLI). Mice were anesthetized with isoflurane (Baxter Healthcare, Deerfield, IL), and d-luciferin (Xenogen, Alameda, CA) in PBS was administered, at a dose of 75 mg/kg of body weight, by retroorbital injection. The BLI with a charge-coupled device camera (IVIS, Xenogen) was initiated 2 min after the injection of luciferin. Dorsal and ventral images were acquired from each animal at each time point to better determine the origin of photon emission. The data were expressed as photon emission (photons per second per cm² per steradian).

Statistical Analysis. All analyses were done by using the program stata 7.0 (StataCorp, College Station, TX), with P < 0.05 considered to be significant. For comparing tumor-associated parameters, the nonpaired Student t test was used. Log-rank tests were used to calculate the statistical significance of the difference in Kaplan-Meier survival curves.

Results

Epothilone Inhibits the in Vitro Growth of MM Cell Lines and Patient CD138⁺ Cells. We first evaluated the direct effect of Flu and dEpoB on proliferation and survival of MM cell lines. Both drugs decreased the survival of RPMI 8226, CAG, H929, MOLP-5, and Dex-sensitive MM.1S lines in a dose-dependent manner. As shown in Table 1, the IC₅₀ in MM cell lines after 72-h incubation was between 6 and 14.4 nM for Flu, and between 37 and 68.6 nM for dEpoB. MOLP-5 is a stroma-dependent MM line that was sensitive to Flu (IC₅₀, 14.4 nM) but relatively resistant to dEpoB (IC₅₀, 68.6 nM). In addition, three of four MM lines tested in vitro were sensitive to paclitaxel, with IC₅₀ values between 9 and 11.5 nM, whereas one line, H929, was relatively resistant (IC₅₀, 75 ± 10 nM). We also evaluated the drug sensitivity of Dex-resistant MM.1R cells and Dox-resistant RPMI 8226/Dox40 cell lines. Flu still showed an IC₅₀ comparable to that obtained on the Dex- or Dox-sensitive parent lines; however, paclitaxel was not as effective as Flu in killing these MM cell lines, showing much higher IC₅₀ values. In contrast to MM lines, two lymphoma lines showed 5- to 10-fold higher IC₅₀ after Flu or dEpoB treatment. Two human marrow stroma lines, HS-27A and HS-5, were relatively resistant to both dEpoB and Flu (IC₅₀ ≈ 100 nM).

Table 1. Cell growth inhibition IC₅₀, nM by epothilones and paclitaxel against a panel of human tumors and normal human cell populations as determined by XTT assay.

		IC50, nM
Histology	Cell line	dEpoB	Flu	Paclitaxel
Myeloma	RPMI 8226/S	37 ± 2	7.6 ± 1.2	11.5 ± 2.8
	RPMI 8226/Dox40	38 ± 3	31 ± 2	6,000 ± 100
	CAG	61.3 ± 4.2	12 ± 1.8	11 ± 2
	NCI-H929	43 ± 5	9 ± 2	75 ± 10
	MOLP-5	68.6 ± 6	14.4 ± 3	NA
	MM.1S	52.3 ± 7	6 ± 1	9 ± 2
	MM.1R	74 ± 4.5	14.2 ± 1.6	68 ± 8
Lymphoma	RL	90 ± 11	80 ± 11	NA
	SKI-DLBCL1	72 ± 9.8	60 ± 4.2	NA
BM stroma	HS-27A	100 ± 10	102 ± 8	89 ± 8
	HS-5	100 ± 8	96 ± 7	NA

The results are expressed as mean ± SD from three independent experiments. NA, no data available; BM, bone marrow.

All MM lines exposed to Flu (10 × IC₅₀) for 24 h exhibited typical morphological changes, characterized by chromatin condensation, development of ring-like structures, a reduction of cell volume, membrane blebbing, and appearance of apoptotic bodies (Fig. 1A). The Dox-sensitive and -resistant RPMI 8226 lines showed similar morphology after Flu treatment.

Fig. 1.

Flu showed high potency in growth inhibition of MM cells in vitro. (A) Morphological changes of MM cells after treatment with Flu (10 × IC₅₀) for 24 h. (B) RPMI 8226 cells were pulse-exposed to either Flu or dEpoB at 10 × IC₅₀ for 1, 2, 4, 8, or 24 h, and the drug was washed out and the cells incubated for up to 48 h. (C) Effect of epothilones and paclitaxel on in vitro proliferation of purified bone marrow CD138⁺ cells (NPCs, normal plasma cells). The data are given as mean ± SD (n = 3). Note that significant inhibition is seen only with Flu on MM cells at 72 h (P < 0.01). (D) Morphology of primary MM cells treated with Flu at 48 and 72 h.

It has been reported that epothilone B was only minimally effective against MM and colon cancer cells after 4-h drug exposure in vitro, possibly because of the kinetics of its intracellular accumulation and retention (14, 15). To investigate the time course of efficacy of Flu and dEpoB on MM cells, RPMI 8226 cells were pulse-exposed to 125 nM drugs for 1, 2, 4, 8, and 24 h, followed by drug wash-out and continued incubation in drug-free medium for up to 48 h. As shown in Fig. 1B, both Flu (Right) and dEpoB (Left) were active, even with short-term drug exposure. However, the effect was most dramatic in cells treated by Flu, because the cell proliferation was completely inhibited at 24 h with only 1-h pulse exposure, whereas the same degree of inhibition with dEpoB required 8 h of pulse exposure. By 48 h, all cells had died after 4- to 8-h exposure to Flu, but 24-h exposure to dEpoB was needed for comparable toxicity (Fig. 1B).

We evaluated the drug sensitivity of MM cells from three patients and normal plasma cells from three healthy donors. The MM CD138⁺ cells were incubated in a 24-well-plate with a preestablished confluent monolayer of marrow stromal cells in the presence of IL-6, sIL-6R, and IGF-I for up to 72 h to facilitate viability and proliferation. Flu, dEpoB, paclitaxel (50 nM), or control vehicle was added to the culture. The viability of MM cells was significantly decreased after 48-h incubation with Flu and even more so after 72-h incubation, relative to controls (Fig. 1C; P < 0.01); however, incubation with the same concentration of dEpoB or paclitaxel had no significant effect on MM cell viability (P > 0.05). Flu had no significant inhibitory effects on normal plasma cells, indicating that these cells are resistant to Flu treatment, presumably because of their G₀ state. Whereas untreated MM cells appeared as typically round or elliptical, treated cells showed apoptotic features, such as chromatin condensation, ring-like nuclear structure, cell shrinkage, and apoptotic bodies (Fig. 1D).

Flu Treatment Blocked MM Cells at G₂/M Phase, Resulting in Apoptosis. To confirm that synthetic epothilone analogs and paclitaxel shared a similar mechanism of induction of cell-cycle arrest (16), we examined the cell-cycle profile of Flu-treated MM cell lines. As shown in Fig. 2A, the addition of 125 nM Flu induced a shift of OPM-2 cells from G₁ to G₂/M as early as 6 h. The cell cycle was completely blocked at G₂/M by 24 h, followed by a sharp increase in sub-G₁ cells (data not shown). To determine the minimal concentration of Flu sufficient to induce cell-cycle arrest, serial concentrations (7.8-125 nM) were incubated with cells for 24 h. Complete cell-cycle arrest of RPMI 8226 cells in the G₂/M phase was seen at a concentration of 31.3 nM Flu (Fig. 2B).

Fig. 2.

Epothilones induce MM cells to undergo cell-cycle arrest at the G₂/M phase, followed by apoptosis. (A) Cell-cycle profile of OPM-2 MM cells. (B) Influence of Flu dose on RPMI 8226 cell-cycle arrest. (C) Annexin-V-binding assay on RPMI 8226 MM cells. (D) DNA fragmentation assay on RPMI 8226 and CAG cells.

To determine the consequence of Flu-induced cell-cycle arrest at the G₂/M phase, we evaluated annexin V staining of treated cells. Cells treated with Flu for only 12 h showed a >10-fold increase in annexin binding, compared with untreated controls (data not shown). After 24 h, Flu-treated cells were 100% annexin-V-positive vs. control cells (<10% positive). Pretreatment with a pancaspase inhibitor, z-VAD-fmk, reduced the number of annexin-binding cells by ≈60%, indicating that the apoptotic process requires caspase activation (Fig. 2C). We also performed DNA-fragmentation assays on RPMI 8226 and CAG cells treated by either Flu or dEpoB. Typical DNA laddering was detected within 24 h (Fig. 2D).

To evaluate the dynamic changes of microtubules in 50 nM Flu-treated RPMI 8226 cells, we showed that Flu and dEpoB shared with paclitaxel the capacity to enhance microtubule-bundle formation in tumor cells, without appreciably changing the total mass of microtubules in the cell shortly after exposure (6-12 h). At later stages (≈24 h), microtubules were disrupted, and cell apoptosis occurred (refs. 16 and 17; and see Supporting Materials and Methods and Fig. 5, which is published as supporting information on the PNAS web site).

Epothilones Induce Apoptosis by Activation of Caspase-3, -8, and -9. To determine whether caspase is involved in epothilone-mediated apoptosis, we first examined the activation of caspase-3 and PARP by standard Western blotting assay. Immunoblots of whole-cell lysates showed typical 17-kDa products of caspase-3 or 85-kDa products of PARP cleavage, accompanied by a decrease in the detection of the uncleaved form of caspase-3 or PARP (Fig. 3 A and B). The activation of caspase-3 was further confirmed by immunohistochemistry, in which only the cleaved caspase-3 was stained (see Supporting Materials and Methods and Fig. 6, which is published as supporting information on the PNAS web site).

Fig. 3.

Flu-induced MM-cell apoptosis is associated with the activation of caspase pathway. (A) Caspase-3 was activated after epothilone treatment in CAG cells. The cells treated with tumor-necrosis-factor-related apoptosis-inducing ligand (TRAIL) were used as a positive control. (B) Cleavage of PARP in CAG cells. (C and D) Caspase-8 and -9 activities were increased and inhibited by specific caspase inhibitors. FMK, fluoromethyl ketone. (E) Detection of cytosolic second mitochondrial-derived activator of caspase (SMAC) (Top) and cytochrome c (Middle) in CAG cells. (F) Activation of phosphorylated JNK in CAG cells (Top and Middle, mean ± SEM) and the total JNK (Bottom). All blots are representative of three independent experiments.

We next explored what initiator caspases are activated, to determine whether the death-receptor pathway (caspase-8 activation) or the mitochondrial-injury pathway (caspase-9 activation) was activated by Flu. As illustrated in Fig. 3, the activity of caspase-8 showed a 3- to 4-fold increase (Fig. 3C), whereas caspase-9 increased 8- to 10-fold, relative to untreated controls (Fig. 3D).

Epothilones Promote the Release of Mitochondrial Proteins Cytochrome c and Second Mitochondrial-Derived Activator of Caspase (Smac) and Activation of Phosphorylated JNK. Activation of caspase-9 is closely associated with mitochondrial injury in the apoptotic pathway. We therefore examined the role of mitochondria in triggering apoptosis in MM cells after treatment with Flu. Treatment of CAG cells with Flu or dEpoB induced the release of both cytochrome c and Smac detected at 12 h and, particularly, at 24 h of incubation (Fig. 3E).

We next looked at an upstream activator of this pathway. Several reports have linked JNK to apoptosis (18, 19), in which stress stimuli activate JNK (20, 21). We demonstrated significant increases (7- to 10-fold) in GST-Jun phosphorylation, indicating activation of JNK (Fig. 3F Top and Middle). Moreover, cotreatment of CAG cells with SP600125 (BioSource International, Camarillo, CA), a specific inhibitor of JNK (22), blocked Flu-induced JNK activity (Fig. 3F Top). Activation of JNK in response to Flu was not associated with changes in total JNK protein levels (Fig. 3F Bottom).

Flu Is a Potent Inhibitor of s.c. Xenografts in NOD/SCID MM Model. To assess the activity of Flu and dEpoB on MM tumor growth in vivo, we first evaluated its effect in an s.c. RPMI 8226 xenograft MM model in NOD/SCID mice. Based on a previous study on solid tumor xenografts at our institute (12, 23), we established a maximum tolerated dose (MTD) for dEpoB of 15-25 mg/kg of body weight, based on i.p. administration every other day and, for Flu, an MTD of 15-30 mg/kg of body weight, based on 6-h i.v. infusion. For the present study, we chose to administer both agents by an i.p. route at 20 mg/kg of body weight, every second day. Animals treated with Flu alone showed a dramatic inhibition of RPMI 8226 tumor growth (Fig. 4A). After three doses of Flu, there was significant tumor shrinkage, and after five doses, the tumor was undetectable. We have followed the mice for >100 d, and there has been no recurrence of tumor. In contrast to the remarkable effect of Flu, dEpoB showed no significant effect on tumor growth, compared with vehicle controls (Fig. 4A). Quantification of tumor burden by BLI at day 50 (just before killing moribund mice in the control and dEpoB groups) showed an inhibition comparable to that obtained by tumor-diameter measurement (Fig. 4 B and C). There was a significant treatment-related body-weight loss (≈13%) noted in comparison with the vehicle-treated control animals (see Fig. 7, which is published as supporting information on the PNAS web site). However, after treatment was suspended, the body weight recovered to control values within 1 wk.

Fig. 4.

Therapy with Flu and dEpoB on NOD/SCID mouse models with subcutaneous or disseminated MM xenografts. (A) Suppression of growth of RPMI 8226 MM xenografts in an s.c. mouse model. Data plotted are the representative tumor volume of two rounds of experiments (n = 10 for each group). (B) BLI of mice with s.c. RPMI 8226 xenografts at day 50. (C) Quantification of dorsal and ventral tumor burden by BLI. (D) In vivo therapy in disseminated CAG MM xenografts. Shown is the representative BLI at day 40, and all images are displayed in the same scale. (E) Quantification of dorsal and ventral tumor burden in mice with disseminated CAG xenografts by BLI at day 40. Data plotted are the percentage of tumor burden (mean ± SD) of mice, relative to control (n = 10 in each group). (F) Kaplan-Meier survival curve of disseminated MM mice treated with Flu or dEpoB (log-rank test, ^*, P < 0.005 for Flu cohort vs. control or dEpoB cohort; n = 10 per treatment group).

Flu Treatment Significantly Prolonged Overall Survival in a Disseminated Xenograft MM Model. We used luciferase-based, noninvasive BLI in a disseminated MM model in NOD/SCID mice by using CAG cells stably transfected with a triple-modality fusion reporter gene expressing herpes simplex virus 1 thymidine kinase, eGFP, and firefly luciferase (13). Anatomical distribution and pathophysiological manifestations in this model were consistent with the clinical course of MM in human patients, i.e., hallmarked by major involvement of the axial skeleton, osteolytic bone lesions captured by both pathology and x-ray examinations (e.g., spine, skull, and pelvis), and frequent development of hind-limb paralysis secondary to spinal lesions, without significant tumor spread to lungs, liver, spleen, or kidney. Most importantly, the tumor burden in individual mice can be quantified by real-time photon emission that correlates with tumor-cell number (Fig. 4D).

We used this disseminated CAG MM xenograft model to test Flu and dEpoB therapy and Flu followed by bortezomib. At the end of treatment (day 40), in mice treated with Flu alone, the tumor burden decreased ≈50-fold, compared with the control; however, there was no difference in tumor burden between dEpoB- and vehicle-treated mice (Fig. 4E). We also compared the tumor burden before and after bortezomib treatment. After five doses of bortezomib (0.25 mg/kg of body weight), the median tumor burden (dorsal and ventral) further decreased 10.1- and 12.3- fold, respectively, relative to tumor burden before bortezomib treatment (data not shown). Overall survival was significantly prolonged in the Flu-treated group (Fig. 4F, P < 0.005). There was a significant body-weight loss (≈17%) in the treatment groups compared with vehicle controls (P < 0.01), and, as in the s.c. study, body weight was recovered ≈1 wk after the cessation of treatment. There was no significant difference in hemoglobin levels and white blood cell count between treated and control mice at the end of drug administration. However, the platelet count was significantly lower in vehicle control mice than in mice treated by Flu, probably because of the extensive tumor infiltration in their bone marrow that resulted in suppression of hematopoiesis (see Fig. 8, which is published as supporting information on the PNAS web site).

Discussion

MM accounts for 1% of all cancers and 10% of hematological malignancies. Treatment of MM with conventional chemotherapy is not curative, with a median survival of ≈3 yr (24). Although high-dose chemotherapy with hematopoietic stem-cell support increases the rate of complete remission and event-free survival, almost every patient relapses, mandating viable salvage therapy options.

Drugs that target microtubules are among the most commonly prescribed anticancer therapies in recent years. The great success of paclitaxel in the market is attributable, in part, to the efficacy of these drugs in solid tumors (17). Paclitaxel has also been tried in the treatment of MM but failed in clinical trials (25, 26).

The application of paclitaxel in MM is limited, not only because of the drug's high toxicity but also because of the development of multidrug resistance, because paclitaxel serves as a substrate for the MDR1/P-glycoprotein drug-efflux pump. Although patients with MM at presentation have a low percentage of tumor cells that express MDR1 protein, this percentage increases up to 50% in patients after chemotherapy (27, 28). These findings suggest that efforts should focus on overcoming some of the problems associated with paclitaxel-based therapy, including issues with formulation, administration, and susceptibility to resistance conferred by the drug-efflux protein P-glycoprotein. Epothilones have emerged from these efforts as a class of microtubule-targeting drugs that should be evaluated in MM.

In the present study, we have evaluated Flu and dEpoB against a panel of human MM and non-Hodgkin lymphoma (NHL) lines. Flu inhibits MM and lymphoma cell proliferation significantly. However, MM cell lines are more sensitive to Flu, with extremely low IC₅₀, whereas two NHL lines were inhibited at doses of Flu that were 5- to 10-fold higher than were effective in MM (Table 1). Compared with dEpoB, Flu has ≈5-fold greater potency on MM cell lines. Importantly, we showed that Flu is effective against paclitaxel-resistant MM cells (8226/Dox40 line) that overexpress the MDR1/Pgp drug-efflux pump (15), as well as the Dex-resistant MM.1R line. In addition, we evaluated the duration of drug exposure in vitro necessary to cause apoptosis in the MM cells, showing that a much shorter duration of exposure to Flu than to dEpoB was needed to produce extensive tumor-cell kill. The retention of efficacy we observed after short drug-exposure times may be related to the ability of Flu to achieve and sustain high intracellular concentrations that, in turn, could be due to the higher affinity of Flu for microtubules, compared with dEpoB or paclitaxel (29). Furthermore, we showed that Flu significantly decreased the viability of primary MM cells in vitro. In contrast to MM cells, normal human plasma cells and bone marrow stromal cells were relatively resistant to Flu.

There have been controversial reports regarding the apoptosis pathway induced by paclitaxel or epothilones (30-32). Our findings clearly demonstrate that activation of both caspase-8 and -9 is required, and caspase-3 activation is the predominant mechanism whereby Flu triggers MM cell apoptosis.

Given the activity of epothilones against MM cells in vitro, it was important to evaluate these compounds in preclinical MM xenograft models. Lin et al. (15) have tested patupilone (epothilone B) in an s.c. RPMI 8226 MM xenograft mouse model and showed a modest prolongation of survival. Because of its high toxicity, the clinical application of epothilone B is limited (33, 34). Our in vivo data in both subcutaneous and disseminated MM xenograft models showed Flu to be much more effective than dEpoB. In an s.c. model, treatment with Flu alone causes tumor disappearance after only five doses of 20 mg/kg of body weight over 10 d. With an additional three doses of consolidation, there was no tumor recurrence within 100 d of observation, indicating that Flu was curative in this model. In a disseminated MM model with high tumor mass, Flu also demonstrated remarkable potency, with overall survival significantly extended in mice treated by Flu alone (≈68 d) vs. mice treated with vehicle alone (<40 d) (P < 0.005).

Myelotoxicity and drug damage to hematopoietic stem cells is a frequent dose-limiting toxicity with cancer chemotherapeutic agents. We observed that human CD34⁺ hematopoietic stem/progenitor cells were not sensitive to short-term exposure to either Flu or dEpoB, unless activated into cycle by cytokine stimulation (see Fig. 9, which is published as supporting information on the PNAS web site). We also monitored the peripheral blood cell count in mice that had received multiple doses of either compound and showed no significant leukopenia. Indeed, platelet counts were even higher in mice treated with Flu, presumably because of the reduction of total tumor mass infiltrating the bone marrow. This finding, together with rapid recovery of body weight after the end of treatment, suggests that the side effects of Flu at therapeutic doses are tolerable.

Because Flu antitumor toxicity targets proliferating cells, newly diagnosed MM would not be expected to show rapid tumor debulking with this agent. However, advanced, more aggressive, refractory, or relapsing disease with an increased S-phase fraction should be responsive. In addition, MM stem cells, with greater proliferation potential than the bulk of the tumor mass (35, 36), might be responsive to Flu. Failure to eradicate these surviving tumor stem cells has proven to be the obstacle to achieving a curative therapy in this disease.

The data presented above indicate the superior performance of the fully synthetic epothilone analog Flu, in which metabolic stability and bioavailability features were incorporated into the molecule through chemical synthesis. Flu was superior to the earlier generation epothilone analog dEpoB in our MM xenograft models. Flu caused complete tumor regression in s.c. tumors, with no subsequent recurrence, and significantly prolonged the overall survival of mice with disseminated tumor, with tolerable side effects. We anticipate that Flu may prove to be a promising agent in MM therapy.