Fluoxetine Inhibits Multidrug Resistance Extrusion Pumps and Enhances Responses to Chemotherapy in Syngeneic and in Human Xenograft Mouse Tumor Models

Chemotherapy frequently fails cancer patients due to inherent or acquired multidrug resistance (MDR; refs. 1, 2, 3). In the dominant mechanism, intracellular levels of cytotoxic drugs are reduced below lethal thresholds by active extrusion of the cytotoxic drug(s) from the tumor cell, operated by ATP-dependent pumps such as P-glycoprotein and multidrug resistance-associated protein (1, 2, 3, 4, 5, 6, 7). Inhibition of the extrusion is a main approach to overturn MDR (1, 2, 3, 4, 5, 6, 7, 8, 9).

The search for effective inhibitors (also named chemosensitizers, MDR modulators, and MDR reversal agents) is now into the third generation. First-generation candidates were drugs already approved for other noncancerous indications, such as verapamil, cyclosporine A, and progesterone (8, 9, 10, 11, 12, 13). Unfortunately, although successful in vitro, patients could not benefit from these chemosensitizers, because their clinically relevant doses exceed safety limits resulting in unacceptable adverse effects, frequent poor solubility, and toxicity (8, 9, 10, 11, 12, 13). Chemical derivatization of first-generation molecules and combinatorial chemistry lead to second- and third-generation chemosensitizers, such as VX-710, PSC833, XR9051, XR9576, MS-209, GF120918, R101933, LY335979, and OC144–093 (ONT-093; refs. 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32), some of which are in clinical trials (25, 27, 28, 30, 31, 32). Several of the latter, whereas more potent and less toxic than first-generation compounds, may still be prone to adverse effects, poor solubility, and unfavorable changes in pharmacokinetics of the anticancer drugs. Moreover, given tumor diversity, it is rational to assume that more than one chemosensitizer will be needed in the clinic.

Here we report for the first time that fluoxetine (Prozac), the well-known antidepressant (33), acts as a highly effective chemosensitizer, joining the arsenal of MDR modulators with prospects of reaching the clinic. Seemingly back to first-generation candidates in terms of using drugs approved for noncancerous indications, we suggest it should be viewed as a fourth-generation chemosensitizer. Unlike first-generation molecules, fluoxetine modulates MDR at low doses, free of the severe dose-related drawbacks experienced with chemosensitizers of previous generations and free of solubility limitations.

This report presents results of in vitro and in vivo studies. Activities and mechanism of chemosensitization by fluoxetine were studied in eight cell lines, covering drug-sensitive and drug-resistant cells (inherent and acquired MDR) Verapamil and Cyclosporin A were included as in vitro benchmarks, the rationale for this choice stemming from their acknowledged in vitro activities as MDR reversal agents and their availability. The in vivo studies were conducted in three syngeneic and one human xenograft mouse tumor models, exploring pharmacokinetics, biodistribution, and therapeutic responses.

Paclitaxel, rhodamine-123 (Rh-123), verapamil, cyclosporine A, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT), and Trypan-blue were from Sigma Chemical Co. (St. Louis, MO). Doxorubicin and Vinblastine were a kind gift from TEVA Pharmaceutical Ltd. (Natania, Israel). Mitomycin C was a kind gift from Dexon Ltd. (Or Akiva, Israel). Fluoxetine was a kind gift from Unipharm (Ramat Gan, Israel). Cisplatin was from TEVA Pharmaceutical Ltd. Materials for cell cultures (specified under Materials and Methods) were from Biological Industries (Beit Haemek, Israel). [³H]Vinblastine sulfate (specific activity 17 Ci/mmol) was from Amersham (Buckinghamshire, England). [³H]paclitaxel (specific activity 20 Ci/mmol) was from American Radiolabeled Chemicals Inc., (St. Louis, MO).

Cell monolayers or suspensions were grown in 100 × 20-mm dishes (culture plates and dishes were from Corning Glass, Corning, New York). B16F10.9 and HT29 cells were cultured in DMEM at 37°C in 5% CO₂ supplemented with 10% fetal calf serum, penicillin (10,000 units/mL), streptomycin (10 mg/mL), and l-glutamine (200 mmol/L). C-26, P388, and D122 cells were similarly cultured, except the medium was RPMI 1640. MCF-7 and MCF-7/ADR cells were maintained as monolayer cultures in MEM containing 10% fetal bovine serum, 1 mmol/L sodium pyruvate, 2 mmol/L l-glutamine, and 1.5 g/L of sodium bicarbonate. P388/ADR and MCF-7/ADR were grown in the presence of 0.25 μg Doxorubicin/mL and 0.5 μg Doxorubicin/mL, respectively. Cells were free of Mycoplasma contamination. determined by a Mycoplasma ELISA test (Boehringer Mannheim GmbH, Mannheim, Germany) performed every 3 months. Cell viability was determined by the following: (1) the MTT method, recording the absorbencies in a plate reader, at two wavelength 550 and 650 nm, (2) the Trypan-blue method, using a hemocytometer, and (3) total cell protein by the Bradford method.

Suspensions of P388/ADR or C-26 (1 × 10⁶ cells/mL) in RPMI 1640 were incubated at 37°C for 30 minutes with 5 μmol/L Rh-123 with and without a chemosensitizer, selected from verapamil (15 μmol/L), cyclosporine A (15 μmol/L), and fluoxetine (5 μmol/L). Intracellular fluorescence was determined by Fluorescence Activated Cell Sorter (FACSort, Becton Dickinson, Franklin Lakes, NJ). Excitation and emission were at 485 nm and 547 nm, respectively. Sample sizes were 10⁴ cells.

C-26 cells were seeded onto 24-multiwell culture plates at densities of 5*10⁴ to 5*10⁵ cells/mL, and the experiments were initiated upon confluency. The wells, divided into three groups, were incubated for 10 hours with serum-supplemented growth medium containing 0.1 μg/ml Doxorubicin and the following additions: group 1: none, group 2: 15 μmol/L verapamil, and group 3: 15 μmol/L fluoxetine. Upon end of incubation, the medium from each well was aspirated, and the cells washed with PBS and incubated with efflux medium, PBS, PBS+ verapamil (15 μmol/L), and PBS+fluoxetine (15 μmol/L), for groups 1, 2, and 3, respectively. At selected time points the medium from each well was collected and replaced with fresh efflux medium. Upon termination the cells in each well were dissolved with 5% deoxycholate. Aliquots from all of the collected media and from each detergent-treated well were assayed for Doxorubicin in a fluorescence plate reader (Fluoroskan Ascent FL, TermoLabsystems, Vantaa, Finland) using appropriate calibration curves. The protocol for Doxorubicin efflux from P388/ADR cells was similar to that of the C-26 cells, except for the following: (1) cell seeding was of 1*10⁶ cells/mL into six-well plates, and (2) the cells were divided into four groups, to accommodate testing three chemosensitizers, verapamil, cyclosporine A, and fluoxetine, each at a concentration of 5 μmol/L. Efflux of paclitaxel from MCF-7/ADR cells was done similar to that of Doxorubicin.

Twenty-four hours before an experiment, cells of a given line were seeded onto 96-multiwell plates at a density of 1 × 10⁴ cells/mL. Upon initiation the regular media was replaced by treatment media, consisting of drug-containing regular growth media with or without a chemosensitizer. Doxorubicin, mitomycin C, vinblastine, and paclitaxel were the test drugs, each tested with fluoxetine and one or more of the benchmark chemosensitizers. Four hours after administration, the media from each well was removed and the cells washed and fed with drug-free chemosensitizer-free fresh serum-supplemented media. The experiments were terminated 20 hours later (24 hours from start), and the quantity of viable cells was determined by the MTT method. Cells treated with paclitaxel (with and without chemosensitizer) were incubated with the treatment media for 48 hours, at the end of which the MTT assay was performed.

The experiments followed Dantzig et al. (21) with minor changes. Briefly, 1 × 10⁶ cells/mL of the MCF-7/ADR and of the C-26 lines were incubated for 24 hours with fluoxetine (5 μmol/L) or verapamil (15 μmol/L), before being washed zero or three times with serum-supplemented growth media. Incubation continued for selected periods (from 0 to 24 hours), after which Doxorubicin was added. Twenty four hours later, the cells were detached, washed twice, and replated at the concentration of 2.5 × 10⁵ cells/mL. Three days later the experiment was terminated, and cell viability was determined using the MTT method. This type of experiment was repeated over many Doxorubicin doses to allow for determination of IC₅₀. Control experiments were conducted under similar conditions but without chemosensitizer.

Animals were obtained from the animal breeding center, Tel Aviv University (Tel Aviv, Israel) and were maintained and treated according to NIH guidelines. All of the animal protocols were approved by the Tel-Aviv Institutional Animal Care and Use Committee.

In all of the syngeneic mouse tumor models, fluoxetine was administered p.o. by inclusion in the drinking water, at the dose of 0.04 mg/kg body/day, from tumor inoculation and on. Drug administration was by injection of 100 μL of Doxorubicin-buffered (PBS) solution to the lateral tail vein using 26-gauge needles. Doxorubicin doses and administration regimen were model-specific, as detailed below.

The protocol was adapted from previous methods (6, 23). P388/WT and (separately) P388/ADR murine leukemia cells were propagated in the peritoneum of 6–8 weeks old BDF₁ female mice by weekly transfer of 0.5 mL of peritoneal fluid containing 5 × 10⁵ cells. In each experiment the animals were divided into four treatment groups (n = 5): saline, fluoxetine, Doxorubicin, and Doxorubicin+fluoxetine. Doxorubicin dose was 3 mg/kg body. Injections were on days 1, 5, and 9 from tumor inoculation. Survival of all of the animals was monitored continuously.

The protocols were adapted from methods described previously (34). B16F10.9 cells (5 × 10⁵/50 μL PBS) were administered i.v. into 10- to 12-week-old male C57BL/6 mice. Treatment groups, animals/group, and Doxorubicin administration regimen were as listed in the section above, except the Doxorubicin dose was 10 mg/kg body. An additional group of untreated tumor-free mice served as control. Two independent experiments were run, one to evaluate survival and the other to evaluate lung metastatic burden. In the latter, terminated 21 days after tumor injection, the lungs of all of the animals were removed, weighted, and fixed in Bouin’s solution. Increase in lung weight was calculated using the formula (34): Lung weight increase (%) = 100 × (tumor lung weight − normal lung weight)/normal lung weight. Surface metastases were counted, using a dissecting microscope, by a pathologist blinded to the experimental groups involved. For survival evaluation animals were monitored daily, and the experiment was terminated on day 75.

Animal species, tumor cell line, and tumor inoculation were essentially as described in the section above. Two independent experiments were run, one for pharmacokinetics and one for biodistribution. In each run, the animals were divided into two groups (n = 5), one treated with Doxorubicin and the other with Doxorubicin+fluoxetine. A single dose of Doxorubicin (10 mg/kg body) was injected at day 10 from tumor inoculation. Blood samples for pharmacokinetics were drawn up to 6 hours after drug injection. For drug biodistribution, animals were sacrificed 6 hours after drug injection, lungs and selected organs removed, viewed by the pathologist, weighed, and processed to extract and assay the drug (by fluorescence, as detailed under drug efflux).

The protocol was adapted from methods described previously (34). C-26 cells (5 × 10⁵/30 μL Hank’s Buffer) were injected into the right-hind foot pad of 6- to 8-week-old female BALB/c mice. The number of treatment groups and animals/group were similar to those listed in Intraperitoneal Ascites Model section above. Doxorubicin dose was 10 mg/kg body, and treatments were given on days 5, 12, 19, and 26 after tumor inoculation. Tumor size was measured, using an electronic caliper, every other day for the next 29 days, and the tumor volume was calculated according to the formula: tumor volume = 1/2(width)² × length. Animal survival was monitored continuously.

The protocol was adapted from methods described previously (35). Athymic nude mice (6 weeks old) were housed in barrier facilities on a 12-hour light/dark cycle. Food and water were supplied ad libitum. On day zero 2.5 × 10⁶ cells in 0.1 mL of PBS were implanted s.c by injection just above the right femoral joint. When tumor volumes reached 75 mm³ (day 0 of treatment), the mice were randomly separated into four groups, similar to those of the syngeneic mice. Saline and Doxorubicin were given every 2 days by i.p. injection. Doxorubicin dose was 4 mg/kg body, and fluoxetine administration (similar in all respects to that given to the syngeneic mice) was started on day zero of treatment. Tumor volume was calculated as (length × width) [(length + width)/2]. The experiment was terminated upon reaching tumor volumes of 2,500 mm³.

The eight cell lines used in this study fall into three categories: (1) drug-sensitive: HT29 (human colon carcinoma), P388/WT (murine leukemia), and MCF-7 (human breast carcinoma); (2) inherently resistant (P-glycoprotein): C-26 (Colon carcinoma ref. 26), B16F10.9 (highly aggressive mouse melanoma, ref. 36), and D122 (highly aggressive mouse Lewis lung carcinoma, refs. 37, 38); and (3) acquired-resistant (P-glycoprotein): P388/ADR (derived from P388/WT) and MCF-7/ADR (derived from MCF-7). The effects of fluoxetine alone were tested over the dose range of 2 to 50 μmol/L and found to be nontoxic up to 40 μmol/L. Consequently, in all of the subsequent in vitro experiments fluoxetine doses were well below toxicity limits. Each cell line was tested with three drugs: vinblastine, mitomycin C, and doxorubicin.

Cytotoxicity trends were quite similar among the three drugs tested, with obvious quantitative differences from one drug to the other (Table 1). As expected, and confirming the MDR nature of the resistant lines, IC₅₀ values of drug-sensitive cells treated with drug alone were significantly lower than those of drug-resistant cells, irrespective of whether the resistance was acquired or inherent (Table 1 rows 1–3 versus rows 4–7). The three drug-sensitive lines were unaffected by the combination treatment of fluoxetine + chemotherapeutic drug (Table 1, rows 1–3). Drug-resistant cells were, on the other hand, quite affected by the combination treatment, the addition of fluoxetine resulting in 15- to 69-fold increases in drug sensitivity (Table 1, RF columns). We found similar trends when the cytotoxicity experiments were repeated with other fluoxetine doses, within the range of 2 to 30 μmol/L, and/or using longer exposures to the treatment media. Comparing, for the acquired-MDR cell lines, the parent versus the resensitized cells, shows mixed trends: P388/ADR cells were resensitized to higher levels than the parent cells, 7.8-, 8.7-, and 3.1-fold, for mitomycin C, doxorubicin, and vinblastine, respectively. Higher resensitization of MCF-7/ADR cells was seen with mitomycin C (6-fold) but not with the other two drugs. These mixed trends indicate that in the transition from a sensitive to a resistant culture, more than the resistance characteristics may have changed. Two major approaches provide MDR acquisition: the selective growth used in this study and MDR1 gene transfection. The former may be more active in changing culture characteristics but is better in modeling acquired-MDR in real life. To verify that fluoxetine resensitized P-glycoprotein-generated MDR, we repeated studies of the type reported in Table 1, testing cisplatin. IC₅₀ values of cisplatin in MCF-7 cells were 35 and 40 ng/mL, with and without fluoxetine, respectively, and those for MCF-7/ADR cells were 60 and 55 ng/mL, with and without fluoxetine, respectively.

Fluoxetine chemosensitization was similar in trend to that of the two benchmarks, verapamil and cyclosporine A, but at equal doses, fluoxetine was more active, as shown for doxorubicin (Fig. 1,A) and for paclitaxel (Fig. 1,B). The effect of fluoxetine persisted much longer compared with verapamil (Table 2). Combined treatment of verapamil + doxorubicin generated 2- to 3-fold reduction in IC₅₀ values compared with doxorubicin alone (Table 2), but responses reverted to control values upon verapamil washout before drug administration (Table 2). Results of similar experiments with fluoxetine were quite different (Table 2): a 3-fold lower fluoxetine dose (5 versus 15 μmol/L) generated a 40- to 70-fold decrease in doxorubicin IC₅₀s, and it took a 24-hour gap between fluoxetine washout and cell exposure to doxorubicin for return to control IC₅₀s.

Insight into the mechanism(s) by which fluoxetine modulates MDR was pursued through the effects of fluoxetine on drug efflux from and on drug accumulation within MDR-cells.

The efflux of doxorubicin or paclitaxel from MDR-cells was quite fast (Fig. 2) for both acquired and inherent (P388/ADR and MCF-7/ADR) and inherent (C-26 line) MDR. One to 2 hours sufficed for complete depletion (Fig. 2). Fluoxetine and the benchmark chemosensitizers verapamil and cyclosporine A slowed down the efflux and, at equimolar doses, fluoxetine was the most effective of the three (Fig. 2). Similar trends were found for mitomycin C efflux from C-26 cells (data not shown).

Accumulation of Rh-123, a well-established P-glycoprotein substrate, was rather poor in both inherent and acquired MDR cells (Fig. 3). Adding 15 μmol/L verapamil or cyclosporine A to the incubation medium generated mild increases in intracellular Rh-123 accumulation, whereas adding 5 μmol/L fluoxetine generated a substantial, 2 to 3 orders of magnitude, increase in intracellular accumulation (Fig. 3).

Fluoxetine did not alter the plasma pharmacokinetics of i.v. administered doxorubicin (Fig. 4,A) nor did it affect drug accumulation in spleen, kidneys, and liver (Fig. 4,B), organs found by pathology to be tumor-free. On the other hand, doxorubicin accumulation in the tumor-bearing lungs was 12-fold (P < 0.01) higher in the animals treated with doxorubicin + fluoxetine compared with those treated with the drug alone (Fig. 4 B).

The lung metastatic burden was evaluated by two measures: the increase in lung weight compared with a control group of healthy animals (i.e., no tumor inoculation and no treatment), and the number of lung metastasis. Good correlation was obtained between both measures for each of the four treatment groups. Upon termination of the experiment, day 21, the lungs of the healthy mice weighed 0.21(±0.03) g. Animals receiving saline or fluoxetine had the highest increase in lung weight and the largest number of lung metastasis (Fig. 5,A). Doxorubicin was better than saline or fluoxetine, but the animals still had a high metastatic burden (Fig. 5,A). Only the combination of doxorubicin + fluoxetine generated a significant reduction in lung metastatic burden (Fig. 5,A). The effects of each treatment group on the metastatic disease were mirrored by the survival data (Fig. 5 B): all animals treated with saline, fluoxetine, or doxorubicin did not survive beyond day 25, whereas animals treated with doxorubicin and fluoxetine were long survivors, the increase in life span >3-fold.

Over a period of 20 days animals inoculated with P388/ADR cells showed the following changes in body weight: mild increase (11%) when treated with saline or fluoxetine, 20% loss when treated with doxorubicin, and a milder loss of 13% when treated with doxorubicin + fluoxetine. Similar trends and body loss values were observed for the animals inoculated with the parent line P388/WT. Survival of animals implanted with either type of cells (sensitive or MDR) was unaffected by treatment with saline or with fluoxetine, and their rather poor survival reflects the aggressive nature of these tumors (Fig. 6). Treatment with doxorubicin increased survival of animals implanted with the drug-sensitive cells, but adding fluoxetine to the doxorubicin made no change compared with those receiving doxorubicin alone (Fig. 6). Responses of the animals bearing the MDR ascites tumor were quite different: poor survival upon treatment with saline, fluoxetine, or doxorubicin and a 2.5-fold increase in life span upon treatment with doxorubicin + fluoxetine (Fig. 6).

The increase in tumor volume with time was fast and exponential for animals treated with saline, fluoxetine, or doxorubicin, and tumor responses were quite similar among these three treatments (Fig. 7,A). Tumor growth rate and tumor volumes were both significantly smaller (P < 0.001) in the animals treated with doxorubicin and fluoxetine than in the other three groups (Fig. 7,A). These responses were mirrored by the effects on animal survival (Fig. 7,B). All of the animals treated with saline, doxorubicin, or fluoxetine, died between days 35 and 45, whereas animals treated with doxorubicin + fluoxetine were long survivors, with a 2- to 3-fold increase in life span (Fig. 7 B).

On the day of treatment initiation, tumor volumes in all of the treatment groups were 70 to 75 mm³. In animals treated with saline, fluoxetine, or doxorubicin, tumors continued to grow rapidly with no significant distinction between the three groups (Fig. 8,A). Response of animals treated with doxorubicin and fluoxetine were strikingly different (Fig. 8,A): tumor growth was arrested upon the first treatment and thereafter, with time and completion of full treatment course, there was complete tumor regression. All of the animals in this group were tumor-free on day 28. These results are mirrored by the survival data (Fig. 8 B): all animals treated by saline, fluoxetine, or drug were dead by day 38, with only slight differences between the three groups, whereas all of the animals treated with doxorubicin + fluoxetine were alive on day 48 when the experiment was terminated.

The choices we made for the in vitro studies were based on MDR variables observed in the clinic: resistance toward more than one drug and a series of cell lines that include both drug-sensitive and drug-resistant systems, the latter exhibiting either inherent or acquired resistance. A chemosensitizer is expected to overturn the poor response of drug-resistant cells to anticancer drugs and at the same time to have little or no effect on drug-sensitive cells (23, 26, 29, 39). Fluoxetine, tested in vitro over the matrix of selected variables, met both expectations. In drug-sensitive cells cytotoxicity generated by a given drug was unaffected by the presence of fluoxetine in the treatment media (Table 1; Fig. 1,B). In drug-resistant cells, acquired and inherent, combination treatment of fluoxetine with an anticancer drug (be it mitomycin C, doxorubicin, vinblastine, or paclitaxel) enhanced cytotoxicity by 2 to 3 orders of magnitude (Table 1; Fig. 1,B), similar to trends observed with other chemosensitizers (23, 26, 29, 37). Fluoxetine was also similar in trend but more potent than the in vitro benchmarks verapamil and cyclosporine A (Table 2; Fig. 1). The ability of fluoxetine to enhance cytotoxicity for several drugs that differ from one another in mechanism of action and in cellular localization of the site of drug action (5) carries the implication that fluoxetine exerts its beneficial effect in pathways or processes that are independent of the particular mechanism(s) by which each drug causes cell demise.

Bearing in mind the dose-related problems with previous-generation chemosensitizers, we deliberately set out to test, in vivo, a low fluoxetine dose: 0.04 mg/kg body/day compared (for example) to the ranges of 1.6, 5 to 30, 2 to 8, and 200 to 300 mg/kg body/dose for Tiamulin, OC144–093, XR9756, and MS-209, respectively (23, 26, 29, 39). One of the advantages of fluoxetine was the ability to administer it through the drinking water, which was less stressful and friendlier than direct (usually forced) oral administrations or injections (i.v. and i.p.; refs. 23, 26, 29, 39).

A MDR reversal agent should be able to truly increase, in vivo, accumulation of an anticancer drug at the tumor and there alone, without altering the pharmacokinetics of the drug. We found fluoxetine to pass both tests. Similar to trends found with OC144–093 (23) and XR9576 (26), and unlike some other chemosensitizers, the pharmacokinetics of doxorubicin were the same, with and without fluoxetine (Fig. 4,A). Fluoxetine generated a substantial 12-fold increase of doxorubicin accumulation in the tumor and at the same time did not affect the accumulation of the drug in other healthy organs (Fig. 4 B).

The impact of fluoxetine on tumor responses and on survival was tested in vivo, using five of the cell lines tested in vitro: MCF-7/ADR, B16F10.9, C-26, P388/WT, and P388/ADR. We found common and encouraging trends, as well as in vitro–in vivo correlations, among the results from all of the animal systems, despite the diversities in animal species, tumor models, and measures of tumor response. In lung metastatic disease, the doxorubicin + fluoxetine treatment reduced the metastatic burden 4- to 7-fold better than doxorubicin alone, with a corresponding increase in life span (Fig. 5). In peritoneal ascites, the doxorubicin + fluoxetine treatment did not affect the responses of mice bearing the drug-sensitive tumor P388/WT but had a significant and positive impact on mice bearing the drug-resistant tumor P388/ADR (Fig. 6). These results (for the P388/ADR line) are on a par or higher than those achieved (albeit using doses 10- to 100-fold higher than fluoxetine) with MS-209, Tiamulin, and OC144–093 (23, 29, 30). In solid tumors, the doxorubicin + fluoxetine combination generated a 6- to 7-fold reduction in tumor progression (Fig. 7,A), with a corresponding increase in life span (Fig. 7,B). The effects on tumor size (Fig. 7,A) seem better than reported (same tumor model and drug but different protocol) for XR9576 (26). In human xenografts (MCF-7/ADR) complete tumor regression and significantly longer survival were achieved solely by the doxorubicin + fluoxetine treatment (Fig. 8). Other studies showed MDR reversal of human xenografts in nude mice by OC144–093 and XR9576, with significant arrest or slow-down of tumor progression (23, 26). The differences in animal models, in protocols, and in drugs preclude direct comparisons of those encouraging findings with our results, yet the similarity in trends is clear.

Two independent types of experiments, measuring the impact of chemosensitizer on drug efflux from and on drug accumulation within MDR-cells, can shed light on the mechanism(s) by which a reversal agent modulates MDR (7, 23, 26, 29, 37). Fluoxetine had no effect on drug efflux (paclitaxel and doxorubicin) from drug-sensitive cells such as MCF-7 or P388/WT (data not shown). It did, however, inhibit the efflux of these drugs in the corresponding acquired MDR cells, as well as from the inherent MDR cells (Fig. 2). By both qualitative and quantitative tests fluoxetine induced significant increases in drug accumulation within MDR-cells (Fig. 3). In both efflux and accumulation, fluoxetine was qualitatively similar to and quantitatively better than the established benchmarks verapamil and cyclosporine A (Figs. 2 and 3), clearly indicating that a major mechanism by which fluoxetine modulates MDR is inhibition of extrusion pumps.

The complexities of cancer diseases and of extrusion pump-mediated MDR will probably require an arsenal of more than one clinically approved chemosensitizer. We find it encouraging that several third-generation chemosensitizers, among them OC144–093 (25), XR9576 (27, 30), MS-209 (28), and VX-710 (31, 32), are in clinical trials. Looking at structural features, fluoxetine, verapamil, OC144–093, and other under-development chemosensitizers, share a similarity in having a “head” consisting of two aromatic rings with a spacer between them. Yet fluoxetine, verapamil, and OC144–093 differ from one another in the size and nature of the spacer and in the substituents on the head rings. It will require in-depth theoretical structural analysis of all of these molecules to understand the relationships among structural similarities and dissimilarities to chemosensitization. Clearly, additional mechanistic and in vivo studies are needed to understand and substantiate the activities of fluoxetine as a chemosensitizer. Yet the low-dose range at which fluoxetine was found to modulate MDR in mice and its ability to reverse MDR in human-originating cells (in vitro and in vivo) hold a promise that fluoxetine may reverse MDR in the clinic at doses well below the safe and approved range of 20 to 80 mg/day prescribed for psychiatric therapy (33).

The authors wish to thank Prof. Nathan Nelson for helpful discussions and Drs. Allon Nechmad and Nissim Cohen for their help in the animal experiments.