Ellipticine derivative NSC 338258 represents a potential new antineoplastic agent for the treatment of multiple myeloma

High-risk multiple myeloma can be correlated with amplification and overexpression of the cell cycle regulator CKS1B. Herein, we used the COMPARE algorithm to correlate high expression of CKS1B mRNA in the NCI-60 cell line panel with the concentration causing 50% growth inhibition (GI₅₀) of >40,000 synthetic compounds. This led to the identification of NSC 338258 (EPED3), a highly stable, hydrophilic derivative of the plant alkaloid ellipticine. In vitro, this synthetic anticancer compound exhibits dramatic cytotoxic activity against myeloma cells grown in suspension or in coculture with stromal cells. EPED3-induced cell cycle arrest and an apoptotic progression that appear to be a consequence of the instantaneous effect of the drug on cytoplasmic organelles, particularly mitochondria. Disruption of mitochondria and cytoplasmic distribution of cytochrome c initiated the intracellular proteolytic cascade through the intrinsic apoptotic pathway. EPED3 is able to induce apoptosis in myeloma cells with de novo or acquired resistance to commonly administered antimyeloma agents. Collectively, our data suggest that EPED3 targets mitochondrial function to rapidly deplete chemical energy and initiate apoptosis in myeloma cells at nanomolar concentrations while leaving stromal cells unharmed. [Mol Cancer Ther 2008;7(3):500–9]

Multiple myeloma is a malignancy of plasma cells that reside primarily in the bone marrow. Although initially responsive to many chemotherapeutic agents, patients with multiple myeloma frequently experience disease relapse that is resistant to multiple drugs and ultimately incurable in most cases (1). With the use of high-resolution, genome-wide molecular genetic analyses done on large, uniformly treated patient cohorts, several genetic entities defining disease subtypes are emerging (2, 3). One of the most consistent findings in these studies is the amplification and overexpression of genes, especially CDC28 protein kinase regulatory subunit 1B (CKS1B), localized to chromosome 1q21. We and others have shown a strong correlation between a poor prognosis and gain/amplification of chromosome arm 1q21 (2, 4, 5). To identify antineoplastic agents with activity in tumors with this high-risk molecular marker, we used the COMPARE algorithm (http://dtp.nci.nih.gov; ref. 6), which identifies correlations between gene expression patterns and drug activity (7–10). By this approach, we identified ellipticines and several other synthetic compounds that showed a strong correlation between their median growth inhibition (GI₅₀) in the NCI-60 human tumor cell lines (11) and CKS1B expression. We subsequently investigated the cytotoxic activity of >20 of these compounds and found that 9-dimethyl amino-ethoxy ellipticine (NSC 338258 or EPED3) is a highly effective in vitro antimyeloma agent. EPED3 is a hydrophilic derivative of ellipticine (Fig. 1), a hydrophobic cell-permeable alkaloid originally derived from Apocyanaceae plants (12).

Ellipticine is a polycyclic molecule that intercalates between DNA base pairs and induces G₂-M-phase cell cycle arrest (13). Previous molecular studies suggested that ellipticine acted primarily through its binding to nucleic acids and inhibition of topoisomerase II (topo II) and topo II–mediated DNA damage (14). Cytochrome P450-dependent oxidation at the 7- or 9-carbon position has been shown to promote DNA intercalation and adduct formation, thereby enhancing the cytotoxic potential of these agents (15). Additionally, selection for ellipticine resistance in a Chinese hamster lung cell line resulted in a phenotype that was cross-resistant to all known topoisomerase-targeting agents as well as several additional P-glycoprotein substrates (16, 17). More recently, Vendôme et al. (18) investigated the potential of ellipticine analogues as tyrosine kinase inhibitors and found that a mutant c-kit kinase activity could be inhibited by ellipticine derivatives with substitutions at the 9-carbon position. The ellipticine derivative EPED3 was not examined in this study, although a structure-activity relationship analysis showed that agents with groups at the C-9 position that could form a hydrogen bond were the most active. Using COMPARE to correlate the cytotoxic activity of EPED3 and other ellipticine derivatives with gene expression, Huang et al. (19) suggested that EPED3 might inhibit multiple drug-resistant protein 1/P-glycoprotein–mediated drug efflux. Thus, the molecular target of EPED3 and its mechanism of action as an antineoplastic agent are largely unknown. Because of our interest in developing agents for treating molecularly defined high-risk multiple myeloma, we conducted in vitro experiments in clinically relevant settings. The disease relies on dynamic interactions between myeloma cells and bone marrow–derived stromal cells, and this interaction can protect malignant cells from drug-induced apoptosis (20). With this in mind, we examined EPED3 cytotoxicity in myeloma cell lines cocultured with stromal cells to simulate the in vivo setting. In these coculture conditions, low concentrations of EPED3 caused rapid cell cycle arrest and massive apoptosis of myeloma cells, whereas stromal cells remained viable.

The human myeloma cell lines H929, OPM-2, RPMI-8226, and U266 were purchased from the American Type Culture Collection. The ARK myeloma cell line was developed in our laboratories, and the remaining myeloma cell lines were kindly provided by Dr. Michael Kuehl (Genetics Department, Medicine Branch, National Cancer Institute). The RPMI-8226/Dox1V drug-resistant variant was the kind gift of Dr. William S. Dalton (H. Lee Moffitt Cancer Center) and is maintained under chronic selective pressure (1 × 10⁸ mol/L doxorubicin and 10 μg/mL verapamil; Sigma-Aldrich). The cell line is characterized by reduced expression and activity of topo II and shows cross-resistance to other topo II poisons, such as mitoxantrone and etoposide; however, it is not resistant to chemotherapeutic agents not known to target topo II, including melphalan, vincristine, cytarabine, and dexamethasone (21). HS-5 (American Type Culture Collection) is a long-term human bone marrow stromal cell line transformed with human papillomavirus.

All cell lines were maintained in RPMI 1640 (Invitrogen) supplemented with 10% fetal bovine serum (Atlanta Biological), 100 units/mL penicillin/streptomycin, 2 mmol/L l-glutamine, and 1 mmol/L sodium pyruvate (Invitrogen). Cells were incubated at 37°C with 5% carbon dioxide.

Myeloma cell lines were maintained at >90% viability and seeded in 96-well tissue culture plates at 10³ cells per well in a 50 μL aliquot. All compounds obtained from the Developmental Therapeutics Program (DTP) of the National Cancer Institute were initially dissolved in 100% DMSO (Sigma-Aldrich) at 1 mg/mL, further diluted in fresh medium to 1 μg/mL, and added to each well in a 50 μL aliquot. All experiments were conducted in triplicate. With the use of the luminescent assay kit CellTiter-Glo (Promega) according to the manufacturer's instructions, cell viability was assayed every 24 h after addition of the DTP compounds for 5 days. Luminescence intensity was measured by a computerized luminometer (Promega), with each well being read five times within 30 min. The statistical sampling distribution per cell line for each compound was based on a sample of 15 measurements.

The monolayer of HS-5 cells were trypsinized (Invitrogen), seeded in six-well plates, and allowed to reach 75% confluence. For indirect contact coculture, myeloma cells were suspended in TC inserts with 3.0 μm pore size track-etched polyethylene terephthalate membranes (Becton-Dickinson Labware) at 10⁶/mL of total volume of medium (5 mL/well) as HS-5 cells adhered to the bottom of the plates. Antimyeloma agents were added individually at serial 2-fold titration; control wells received medium only.

We used a 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) reduction assay kit (Promega) to determine cell proliferation and survival. Briefly, myeloma cells were seeded in 96-well plates at 2 × 10⁴ per well. Antimyeloma agents were diluted in a serial 2-fold titration according to the potency of each agent. All tests were set in triplicate or quadruplicate. MTT assays were done 96 h after addition of the agents by adding the dye solution to each well (15 μL/well) and further incubating at 37°C and 5% carbon dioxide for 2 h; the reactions were then stopped with 100 μL of the solubilization solution/stop mix. Plates were read simultaneously at wavelengths of 570 and 650 nm in a plate reader; the A₅₇₀ values were corrected by the A₆₅₀ values.

Cell cycle arrest was measured with the use of the DNA-Prep Reagent System kit (Beckman Coulter). Myeloma cells were harvested from coculture TC inserts and washed in PBS; 5 × 10⁵ cells were stained according to the manufacturer's instructions. Samples were analyzed with the EPICS XL-MCL flow cytometer equipped with EXPO32 ADC software (Beckman Coulter).

Apoptosis was detected by staining with Annexin V-FITC (Beckman Coulter). Myeloma cells were harvested from coculture TC inserts and washed in PBS; 5 × 10⁵ cells were incubated on ice with 100 μL binding buffer and 10 μL Annexin V-FITC for 15 min. Samples were analyzed with the EPICS XL-MCL flow cytometer equipped with EXPO32 ADC software (Beckman Coulter).

Myeloma cells were in TC inserts cocultured with HS-5 cells and treated with individual antimyeloma agents. An aliquot containing 10⁶ cells was transferred to a 24-well plate and mixed with MitoTracker Red CMXRos (Invitrogen) at a 50 nmol/L final concentration. After 15 min, cells were collected, washed with PBS, and adhered to slides (5 × 10⁴ per spot) by Shandon-Cytospin centrifugation (Thermo Electron). Slides were fixed in 3.7% formaldehyde/PBS for 15 min and rinsed in PBS; the cells were permeabilized in ice-cold acetone for 5 min and rinsed twice in a solution of PBS with 0.1% NP-40 (PBD). Mouse antihuman cytochrome c monoclonal antibody (R&D Systems) was diluted 1:500 with 10 mmol/L Na⁺PO₄ (pH 7.8)/0.15 mol/L NaCl buffer and added to each spot. After 15 min, slides were washed twice in PBD. FITC-conjugated goat anti-mouse IgG H+L chain (BD Biosciences) was added, and slides were incubated 15 min. Finally, to stain cell nuclei, slides were incubated with 0.1 μg/mL 4,6-diamidino-2-phenylindole (Sigma-Aldrich) in PBS and mounted with Antifade solution (Invitrogen). Images were captured with the use of the Genetic Workstation, with a red (for mitochondrial staining), green (for cytochrome c staining), and blue (for nuclear staining) single band-pass filter set (Abbot Vysis).

Myeloma cell lines were indirectly cocultured with HS-5 cells as described above. EPED3 or bortezomib (Millennium Pharmaceuticals) was added to cocultures to final concentrations of 2 and 0.5 μmol/L, respectively. After 6 h of treatment, myeloma cells were harvested from the inserts, washed twice in PBS, and resuspended at 10⁴ cells/μL in protein extraction buffer [PBS with 10 μg/mL aprotinin, 10 μg/mL leupeptin, and 1 mmol/L phenylmethylsulfonyl fluoride (Sigma-Aldrich)]. Cell lysates were prepared by three cycles of freezing in liquid nitrogen followed by thawing at 37°C. Protein concentration was determined by spectrophotometry (NanoDrop Technologies).

For SDS-PAGE, 100 μg protein extract was mixed with NuPAGE LDS sample buffer (Invitrogen) and loaded into each lane of the gel (10-15% polyacrylamide separating gel, 4% stacking gel). After electrophoresis, protein was transferred to Hybond ECL nitrocellulose membrane (Amersham Pharmacia Biotech). Primary goat anti-human caspase-3, caspase-8, and caspase-9 polyclonal antibodies (R&D Systems), rabbit anti-human CKS1B, and mouse anti-human β-tubulin (Invitrogen) were applied and then detected with the WesternBreeze Western blot immunodetection kits (Invitrogen) according to the manufacturer's instructions.

U266 myeloma cells (10⁶/mL) were treated for 6 h with 2 μmol/L EPED3. Cells were harvested, washed twice in PBS, and then fixed in 3% glutaraldehyde solution; postfixation with 1% osmium tetroxide followed. After serially graded ethyl alcohol dehydration, cells were embedded in resin (Embed 812, Electron Microscopy Sciences) and baked overnight at 60°C for polymerization; sectioning then followed. The morphology of U266 cells was examined and captured with a transmission electron microscope (Philips Electronics; ref. 22).

To identify potential anticancer compounds with specific potency in cells with high-risk molecular features, we used the National Cancer Institute's COMPARE program to query growth inhibition data in the DTP open database of ∼44,000 synthetic compounds. The GI₅₀ value (the concentration that causes 50% growth inhibition) of a compound was used as the selection criterion for this analysis.

Initially, 24 DTP compounds (Supplementary Table S1)³

were identified by a strong correlation of cytotoxicity with expression levels of the gene CKS1B, a high-risk feature in multiple myeloma, in the NCI-60 cancer cell panel. These compounds were screened for effectiveness against four myeloma cell lines: ARK, KMS11, RPMI-8226, and U266. The compounds were added to cells at a final concentration 0.5 μg/mL. Cell viability was determined with CellTiter-Glo assays every 24 h over the following 5 days and compared with untreated controls (medium only or 0.5% DMSO). Two ellipticine derivatives, EPED3 and NSC 69187, most significantly inhibited cell viability in the four myeloma cell lines (Fig. 2A). With this finding in hand, we extended our study to additional ellipticine derivatives from the DTP. A substantial number of ellipticine derivatives had significant Pearson correlation coefficients, indicating the general tendency of the compound toward cell-killing activity in association with high CKS1B gene expression levels in the NCI-60. Among the ellipticines with high Pearson correlation coefficients, we focused on those with a high SD, which measures the degree of difference in the response of the cell line to the compound of interest. Of 18 ellipticines listed in Supplementary Table S2,³ we tested 12 compounds (insoluble ellipticines were excluded) for growth inhibition of the same four cell lines. The data indicated that, among the ellipticine derivatives, EPED3 most significantly inhibited myeloma cell growth (Fig. 2B). This suggests that the enhanced antineoplastic activity of EPED3 is related to the specific modifications made to produce EPED3 (Fig. 1).

The cytotoxicity of EPED3 was further tested with a panel of 12 myeloma cell lines in a serial 2-fold titration (50-3,200 nmol/L). Two commonly used antimyeloma agents, melphalan (Alkeran; Celgene) and bortezomib (Velcade, Millennium Pharmaceuticals), were simultaneously set for direct comparison at concentrations ranging from 2 to 128 μmol/L and 0.5 to 32 nmol/L, respectively. Cytotoxicity was analyzed by MTT assay. Absorbance was read after 96-h incubation, and the 50% inhibitory concentration (IC₅₀) of EPED3, melphalan, or bortezomib for each myeloma cell line was calculated with the use of OriginPro 7.5 software (OriginLab). EPED3 inhibited myeloma cell growth at nanomolar concentrations in all cell lines tested (Table 1A). Additionally, to determine whether human serum albumin would affect drug activity in future clinical trials, MTT assays were done in the presence of 2% human serum albumin. No inhibition of cytotoxicity was seen in these assays.

Previous studies showed that bone marrow–derived stromal cells contributed to the survival of multiple myeloma cells via both soluble factors and direct cell contact (20, 23, 24). To determine the effect of the sensitivity of myeloma cells to EPED3 in the presence of protective stromal cells, we treated 12 myeloma cell lines in coculture with HS-5 cells. EPED3 was diluted in a 2-fold series (range, 125-4,000 nmol/L). Myeloma cell proliferation was determined with MTT assays 96 h after the addition of EPED3. From each TC insert, myeloma cells were homogenously suspended, and aliquots were transferred to a 96-well plate for MTT assay. The adherent HS-5 cells were simultaneously assayed by MTT dye reduction after removal of the TC inserts from each plate (n = 12). The IC₅₀ values of EPED3 in each myeloma cell line, as well as HS-5 cells, were calculated with OriginPro 7.5 software (Table 1A; Supplementary Fig. S1).³ Myeloma cell lines showed a mean 3.2-fold resistance (range, 1.46-6.92 μmol/L increase of IC₅₀; Table 1B) to EPED3 in coculture with HS-5 stromal cells compared with myeloma cells alone. HS-5 cells showed no significant cell death at the highest dose of EPED3 examined (Supplementary Fig. S1).³ The cytotoxicity of EPED3 to HS-5 cells was also tested without the presence of myeloma cells. HS-5 cells were initially seeded in 96-well TC plates at 2-fold incremental densities and exposed to serially diluted EPED3 for 96 h. The MTT assays indicated that HS-5 cells at higher density were more insensitive to EPED3 (Supplementary Fig. S2).³

The cytotoxicity of EPED3 was compared in vitro with those of ellipticine and four other, commonly used antimyeloma agents in four myeloma cell lines: JJN3, L363, OCI-MY5, and U266. Ellipticine, doxorubicin, etoposide, dexamethasone, bortezomib, and EPED3 were each applied in a 2-fold dilution series at concentrations ranging from 0.1 to 6.4 μmol/L. Cell proliferation was assessed with MTT assays 24 and 48 h after initiation of treatments. In contrast to all other agents, except bortezomib, area-under-the-curve calculations indicated that EPED3 efficacy was dose dependent but not time dependent (Supplementary Table S3).³

Because ellipticine has been reported to inhibit topo II and in vitro selection for ellipticine resistance is reported to result in a phenotype with cross-resistance to topoisomerase-targeting agents, we examined the activity of EPED3 in a drug-resistant variant of the RPMI-8226 myeloma cell line. The RPMI-8226/Dox1V cell line was selected for resistance to doxorubicin in the presence of the ATPase inhibitor verapamil. This selection scheme prevents the emergence of P-glycoprotein-mediated multidrug resistance. Rather, resistance is due to a reduction in the expression and activity of topo IIα. RPMI-8226/Dox1V cells and the parental line (RPMI-8226) were each incubated with serial dilutions of EPED3 or doxorubicin, as described previously (21), and cell growth was assessed with MTT assays. Linear regression analysis of 96-h dose-response curves was used to compare the cytotoxic activity of EPED3 in RPMI-8226/Dox1V and RPMI-8226 cells. EPED3 cytotoxicity in the doxorubicin-resistant myeloma cells was not significantly different from its activity in the parental cells (mean IC₅₀, 131.3 ± 36.0 nmol/L, n = 4, versus 150.3 ± 40.6 nmol/L, n = 4, respectively; P = 0.7; Fig. 2C). In contrast, RPMI-8226/Dox1V cells showed 7.3-fold higher resistance to the cytotoxic effects of doxorubicin, a known topo II inhibitor, than did parental RPMI-8226 cells (IC₅₀, 293.5 ± 34.01 nmol/L, n = 4, versus 40 ± 17.68 nmol/L, n = 4, respectively; P < 0.0001; Fig. 2D).

To investigate the effects of EPED3 on the cell cycle and apoptosis of myeloma cells, U266 cells were exposed to 2 μmol/L EPED3 or 0.5 μmol/L bortezomib in TC insert coculture with HS-5 cells. Cells were harvested and stained for DNA content or with Annexin V for flow cytometry analyses. After treating U266 myeloma cells for 12 h, we found that 67.5% of the EPED3-treated population, compared with 49.5% of the untreated control population and 59.8% of the bortezomib-treated population, was in G₀-G₁ phase. Furthermore, 7.2% of the EPED3-treated population, in contrast to 25.4% of the control population and 18.9% of the bortezomib-treated population, was in G₂-M phase (Fig. 3A-C). The addition of Z-VAD-FMK, a pan-caspase inhibitor, to the cocultures did not significantly change the cell cycle distribution of cell populations when U266 cells were exposed to EPED3 or bortezomib (Fig. 3D-F), suggesting that the reduced number of cells in G₂-M phase was not due to cell death.

Based on simultaneous flow cytometry analysis of Annexin V-FITC staining, the degree of programmed cell death resulting from EPED3 treatment at 45.5% was marginal compared with that of 51.4% with bortezomib (Fig. 3G-I). However, the cytotoxicity of both EPED3 and bortezomib was blocked by addition of Z-VAD-FMK, which resulted in decreased cell death in both EPED3-treated (27.8%) and bortezomib-treated (27.7%) cells (Fig. 3J-L). These data suggest that the cytotoxicity of both agents operates in part through caspase-dependent apoptotic mechanisms.

We used EPED3 autofluorescence to visually examine its intracellular distribution in myeloma cells. The EPED3 molecule emits fluorescence when exposed to UV light at a wavelength of ≤500 nm. Under fluorescence microscopy, cells that have taken up EPED3 are therefore discernible by the characteristic golden light emitted by the compound. When cells were treated with EPED3, we observed nearly instantaneous permeation, with cytosolic distribution of the agent within 15 min of exposure (data not shown). Six hours after treatment, large vacuoles formed within the cells, indicating ongoing apoptosis; disintegration of the plasma membrane, another hallmark of apoptosis, was visualized as the intensity of EPED3 fluorescence dimmed in the cytosol (Fig. 4A). The uptake of EPED3 was also found in HS-5 cells in coculture with myeloma cells (Supplementary Fig. S3).³

To investigate the effects of EPED3 on mitochondrial integrity and cytochrome c, we used immunohistochemistry. Disruption of mitochondria in EPED3-treated U266 myeloma cells was shown by the loss of staining with MitoTracker Red CMXRos and subsequent release of cytochrome c from the mitochondrial membrane into the cytosol. In contrast, mitochondrial disruption and cytochrome c release were not seen in cells similarly treated with bortezomib (Fig. 4B).

Using Western blotting, we also investigated the activation of caspases involved in the mitochondrial apoptotic pathway. JJN3, L363, OPM-2, and U266 cells were indirectly cocultured with HS-5 cells and treated with EPED3 or bortezomib for 6 h. Proteins were then extracted from myeloma cells and analyzed for activation of caspase-8, caspase-9, and caspase-3. None of the myeloma cell lines examined showed obvious activation of caspase-8, suggesting that the extrinsic apoptotic pathway is not involved in EPED3 cytotoxicity. In contrast, the strong activation of caspase-9 followed by the cleavage of caspase-3 indicated that the intrinsic apoptotic pathway was involved in triggering the proteolytic cascade in all myeloma cell lines (Fig. 4C).

Electron microscopy was used to examine subcellular morphology following EPED3 exposure. The electron microscopic observation revealed that U266 cells underwent severe degenerative changes 6 h after treatment with 2 μmol/L EPED3. Ultrastructural alterations, including pyknosis (condensation and reduction in nuclear size associated with hyperchromatosis), destruction of mitochondrial inner cristae, and formation of multiple cytosolic vacuoles, reflected the apoptotic state of the treated cells (Fig. 5).

For ∼15 years, the DTP of the National Cancer Institute has used the NCI-60 panel to screen potential anticancer agents for their ability to inhibit growth in established cancer cell lines (7). Early in this process, it was recognized that compounds with similar mechanisms had similar patterns of sensitivity. This led to development of the COMPARE algorithm (6), which compiles a list of compounds with patterns of growth inhibition similar to that of a “seed” compound supplied by an investigator. This approach has been useful in the investigation of potential mechanisms of action for novel compounds. More recently, molecular characterization of the NCI-60 has provided an opportunity to apply COMPARE analysis to gene expression signatures (9, 10). In the present study, the COMPARE algorithm was used to correlate gene expression profiles with the potency of anticancer compounds (10). Correlation of the expression levels of the CKS1B gene, whose overexpression is associated with rapid failure of high-dose therapies (25), with the activity of antineoplastic agents in the in the NCI-60 panel led to the identification of EPED3 as a potential new, rationally identified antimyeloma agent.

EPED3 is an analogue of ellipticine with a dimethyl amino-ethoxy substitution at the C-9 position. This radical not only tremendously improves the hydrophilicity of the compound, allowing rapid cellular uptake, but also prevents intracellular metabolism by the hydroxylation at C-9 of ellipticine (14). Although initially categorized as a topo II inhibitor that localizes to the nucleus, ellipticine has more recently been shown to localize in the cytoplasm and accumulate in mitochondria (26). Similarly, EPED3, which displays intrinsic fluorescence identical to ellipticine, was found to rapidly accumulate in the mitochondria in this study. Fluorescence microscopy studies show that cellular uptake of EPED3 is rapid and unrestrained. Under timed observations, EPED3 passively crosses the plasma membrane and becomes evenly distributed throughout the cytosol within minutes, suggesting that EPED3 is a lipophilic molecule that crosses the plasma membrane independent of receptor or ion channel mechanisms. Subsequent to its entry into cells, EPED3 immediately initiates an enormous loss of mitochondrial membrane potential, release of cytochrome c, and formation of large vacuoles (Figs. 4 and 5). The CellTiter-Glo assay, which principally quantifies ATP and signals metabolic activity, showed that EPED3-treated cells produce the lowest ATP level than all agents tested (data not shown), suggesting an inhibition of cellular chemical energy production following treatment. It is likely that, like ellipticine, EPED3 localizes to the mitochondrial inner membrane, thereby disrupting the membrane potential by uncoupling mitochondrial oxidative phosphorylation and crippling energy production in the cell. The exact mechanism of mitochondrial membrane permeability transition is not yet fully defined but may involve the generation of reactive oxygen species (data not shown).

The bone marrow microenvironment augments myelomagenesis and can protect myeloma cells from apoptosis (20, 27, 28). Although most myeloma cell lines were initially cultivated from either peripheral blood (e.g., L363, LP-1, OPM-2, RPMI-8226, and U266) or pleural effusion (e.g., ARK, ARP1, CAG, EJM, and H929), all encode the common genetic mutations that exist in primary myeloma (29). In vitro, soluble factors generated by intracellular interactions have been documented to contribute to protecting myeloma cells from chemotherapeutic agents (20). Thus, we used an indirect myeloma cell-stromal cell coculture system in our systematic screen to discover new synthetic compounds capable of overcoming the soluble factors present in the bone marrow microenvironment that confound therapeutic efficacy. HS-5, a long-term stromal cell line that can support the proliferation of hematopoietic progenitor cells in coculture conditions, was initially tested to better understand the activity of EPED3. HS-5 cells were set at incremental density and treated by EPED3 in 2-fold titration. Our results indicate that HS-5 cells respond to EPED3 only at high dosage and display confluence-dependent resistance as is reported for many cell lines (30).

Acquired drug resistance is a major obstacle to sustained successes in myeloma treatment. Understanding the cytotoxic mechanism of a putative therapeutic agent will allow the design of therapeutic strategies to overcome acquired drug resistance in patients. In vitro selection has historically been used to establish model systems for investigating the mechanisms of acquired drug resistance commonly seen in patients (30, 31). Our data indicate that EPED3 can overcome acquired drug resistance in myeloma cells selected for resistance to doxorubicin in the presence of the P-glycoprotein inhibitor verapamil. RPMI-8226/Dox1V cell line is cross-resistant to a variety of topo II–dependent therapeutic agents presumably due to reduction of topo II expression as well as activity (21). The cells exhibited a high tolerance to doxorubicin (Fig. 2D), which functions through topo II inhibition, but EPED3 treatment resulted in prompt, dose-dependent killing of RPMI-8226/Dox1V cells. Collectively, these data accentuate the much greater potency of EPED3 in killing myeloma cells, even those insensitive to topo II inhibitors, through topo II–dependent and time-independent mechanisms.

Dose-response analysis indicated that the efficacy of EPED3 was not schedule dependent, whereas doxorubicin is well known to be most effective in the DNA synthesis (S) phase. The IC₅₀ and the area under the curve of myeloma cells responding to doxorubicin declined as drug exposure was prolonged, whereas the IC₅₀ and the area under the curve of EPED3 were steady during the time course (Supplementary Table S3).³

Two other ellipticine derivatives that are clinically administered in the treatment of breast cancer, Celiptium (NSC 264137) and Detalliptinium (NSC 311152), also appear to act independently of topo II. DNA cleavage activity of both these ellipticine derivatives was compared in vivo with that of 4-(9-acridinylamino)-methanesulfonic-m-anisidide, a putative topo II inhibitor. Although IC₅₀ dosages of all three agents were within close range, Celiptium and Detalliptinium were 50 times less potent for DNA strand breakage than 4-(9-acridinylamino)-methanesulfonic-m-anisidide (14). This is consistent with our finding that the cytotoxicity of the ellipticine derivative EPED3 is topo II independent.

Flow cytometry analyses in our studies showed that myeloma cells accumulate in the G₀-G₁ phase with EPED3 treatment. The mechanism of cell cycle arrest is not yet known but may be related to intracellular energy depletion. Although a high CKS1B gene expression profile was the initial “seed” in our search for more effective antimyeloma agents, the gene expression profiling results showed no significant changes in CKS1B gene expression after myeloma cells were exposed to 2 μmol/L EPED3 for 6 h.⁴

For confirmation, we conducted Western blotting assays and found that CKS1B protein levels were unchanged in myeloma cell lines treated with EPED3 (Fig. 4C). These data suggest that the gene transcription or mRNA stability of CKS1B is not a primary target of EPED3 but that myeloma cells with CKS1B gene expression signature are more likely to be sensitive to EPED3. Future studies employing COMPARE algorithm analyses will be aimed at using multiple gene expression signatures of high-risk myeloma instead of the single gene signature. Overall, multiple in vitro–derived lines of evidence point to reduced cell viability, cell cycle arrest, and apoptosis in myeloma cells treated with EPED3.

We thank the DTP of the National Cancer Institute for providing the synthetic compounds in this study; Simona Colla, Sushil Gupta, Yongsheng Huang, Bob Kordsmeier, Chris Randolph, Peter Stewart, Yan Xiao, David Williams, Hong Wei Xu, and Fenghuang Zhan (Donna D. and Donald M. Lambert Laboratory of Myeloma Genetics) as well as the physicians and nurses at the Myeloma Institute for Research and Therapy for contributions to establishing and maintaining the multiple myeloma gene expression profile database since 1999; the Office of Grants and Scientific Publications at the University of Arkansas for Medical Sciences for excellent editorial assistance; and Profs. Dehui Chen and Shaojun Liu and Yi Yang (Institute of Basic Medical Sciences, Academy of Military Medical Sciences, China) for their expertise in the cell ultrastructure studies.