Inhibition of all-TRANS-retinoic acid metabolism by R116010 induces antitumour activity

All-trans-retinoic acid is a potent inhibitor of cell proliferation and inducer of differentiation. However, the clinical use of all-trans-retinoic acid in the treatment of cancer is significantly hampered by its toxicity and the prompt emergence of resistance, believed to be caused by increased all-trans-retinoic acid metabolism. Inhibitors of all-trans-retinoic acid metabolism may therefore prove valuable in the treatment of cancer. In this study, we characterize R116010 as a new anticancer drug that is a potent inhibitor of all-trans-retinoic acid metabolism. In vitro, R116010 potently inhibits all-trans-retinoic acid metabolism in intact T47D cells with an IC50-value of 8.7 nM. In addition, R116010 is a selective inhibitor as indicated by its inhibition profile for several other cytochrome P450-mediated reactions. In T47D cell proliferation assays, R116010 by itself has no effect on cell proliferation. However, in combination with all-trans-retinoic acid, R116010 enhances the all-trans-retinoic acid-mediated antiproliferative activity in a concentration-dependent manner. In vivo, the growth of murine oestrogen-independent TA3-Ha mammary tumours is significantly inhibited by R116010 at doses as low as 0.16 mg kg−1. In conclusion, R116010 is a highly potent and selective inhibitor of all-trans-retinoic acid metabolism, which is able to enhance the biological activity of all-trans-retinoic acid, thereby exhibiting antitumour activity. R116010 represents a novel and promising anticancer drug with an unique mechanism of action. British Journal of Cancer (2002) 86, 605–611. DOI: 10.1038/sj/bjc/6600056 www.bjcancer.com © 2002 Cancer Research UK

All-trans-retinoic acid (RA) is a naturally occurring retinoid that is well-known to inhibit cell proliferation and to induce differentiation (Hong and Itri, 1994;Moon et al, 1994). However, the clinical use of RA in the treatment of cancer is significantly hampered by the prompt emergence of resistance, which is believed to be caused by increased RA metabolism (Muindi et al, 1992Lee et al, 1993;Miller, 1998). RA is metabolized by CYP26A1, an inducible cytochrome P450-dependent enzyme, that inactivates RA by 4-hydroxylation of the b-ionone ring (White et al, 1997). This tightly-controlled negative feedback mechanism limits the availability of RA and thereby also its biological activity. Therefore, increasing levels of RA by inhibition of its metabolism might prove an innovative approach to cancer treatment.
Previously, we have identified liarozole-fumarate as an inhibitor of RA metabolism (Wouters et al, 1992). In vitro, liarozole-fumarate has been shown to enhance both the antiproliferative and differentiation-inducing activity of RA (Wouters et al, 1992;Van heusden et al, 1996Van heusden et al, , 1998. In vivo, liarozole-fumarate increases endogenous RA levels both in plasma and tissues (Smets et al, 1995). In several preclinical tumour models, liarozole-fumarate has been shown to exhibit antitumour activity (Van Ginckel et al, 1990;Dijkman et al, 1994;Smets et al, 1995), linked to increased levels of RA within the tumour (Smets et al, 1995). In cancer patients, liarozole-fumarate has been shown to increase the half-life of orally administered RA and 13-cis-retinoic acid (Westarp et al, 1993;Achkar et al, 1994;Miller et al, 1994). However, one of the limitations to the use of liarozole-fumarate was its lack of specificity. It was shown to inhibit also other cytochrome P450-mediated reactions (Vanden Bossche, 1992) and affect plasma hormone levels in volunteers (Bruynseels et al, 1990).
This lack of specificity might partly explain the limited risk benefit ratio observed in cancer patients. A more potent and selective inhibitor is therefore required to improve the clinical efficacy of inhibitors of RA metabolism. In the present study, we identify R116010 ( Figure 1) as a highly potent and selective second-generation inhibitor of RA metabolism that induces antitumour activity. Research & Development, Val-de-Reuil, France). Compounds were dissolved to an initial concentration of 10 mM in DMSO. Further dilutions were made in culture medium. [11, 12-3 H(N)]RA (30 Ci mmol 71 ) was obtained from NEN Life Science Products (Boston, MA, USA). Unlabelled RA was purchased from Serva (Heidelberg, Germany) and dissolved to an initial concentration of 4 mM in ethanol. Further dilutions were made in culture medium. All-trans-retinoic acid was used in a dark-room with yellow illumination. Final solvent concentrations in the in vitro assays were always less than 0.5% (v/v). This solvent concentration had no effect in the respective assays.

Cell culture
Human T47D breast cancer cells -purchased from the American Type Culture Collection (Rockville, MD, USA) -were cultured in RPMI 1640 medium supplemented with 10% (v/v) FBS, 2 mM Lglutamine, 50 mg ml 71 gentamicin, and 10 mg ml 71 insulin (all reagents from Life Technologies, Gent, Belgium). Cells were grown in a humidified incubator (5% CO 2 , 95% air) at 378C and were Mycoplasma-free as tested by the Mycoplasma T.C. kit (Gen-Probe Incorporated, CA, USA).
Microcolumn assay for RA metabolism RA metabolism was quantitatively determined using the microcolumn assay as described previously (Krekels et al, 1997). Briefly, T47D cells were pretreated for 16 h with 1 mM RA to induce RA metabolism. Cells were then washed twice with culture medium, harvested and resuspended at 4610 6 cells ml 71 . An aliquot of this cell suspension (450 ml) was incubated for 90 min in the presence of 0.1 mM [11, 12-3 H(N)]RA, after which 2 ml acetonitrile was added. After centrifugation for 10 min at 780 g, the resulting deproteinized supernatant was acidified with 2.5 ml of 40 mM acetic acid and applied to a 3 ml C18 Bond Elut LRC column (Varian, Harbor City, CA, USA; pretreated with 4 ml distilled water) under a vacuum of 127 mmHg using VAC ELUT SPS-24 and the effluent was collected. The column was eluted with 1 ml 40% acetonitrile and the effluent was collected in the same vial. Radioactivity in the collected effluent, containing the polar metabolites, was determined in a Packard Tricarb 4530 liquid scintillation analyzer. Optiphase 'Hi Safe II' (Wallac, Milton Keynes, UK) was used as a scintillator. Data are presented as mean+s.d. of three or four independent experiments. The IC 50 -values were calculated by non-linear regression analysis using SigmaPlot 4.01 software.

HPLC analysis
The T47D cell suspension was prepared as described above and 450 ml was incubated with 0.1 mM [11, 12-3 H(N)]RA for 90 min. After centrifugation for 10 min at 780 g, the supernatant was analyzed for the presence of RA metabolites. Reverse-phase HPLC analysis was carried out as described previously (Van heusden et al, 1998).

In vitro cytochrome P450 isozyme specificity
Metabolism of RA by CYP26A1 was measured in microsomes of T47D human breast cancer cells that were pretreated for 16 h with 1 mM RA to induce RA metabolism. Microsomes were prepared exactly as described by Han and Choi (1996), and RA metabolism was measured as described by Van Wauwe et al (1994). The conversion of androstenedione to estrone by aromatase (CYP19) in human placental microsomes, the conversion of 17-hydroxy-20-dihydroprogesterone to testosterone by 17,20-lyase (CYP17) in rat testicular S10 fractions, and the 2a-, 7a-, 6b-, and 16b-hydroxylation of testosterone (CYP2C11, CYP2A1, CYP3A, CYP2B1/2) in rat liver microsomes were carried out as described .
Proliferation experiments (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide assay) T47D cells were seeded in Falcon 1 96-well cell culture plates (Life Technologies, Merelbeke, Belgium) at 2500 cells per well in a total volume of 150 ml. Cells were allowed to adhere to plastic for 24 h. Drugs and/or medium were added to a final volume of 200 ml (day 0). Cells were grown under these conditions for 7 days, with renewal of medium and drugs on days 2 and 5. On day 7, medium was renewed and 25 ml MTT-solution was added, and cells were further incubated for 2 h in a cell incubator. The medium was aspirated and 25 ml Sorensen glycine buffer (0.1 M glycine, 0.1 M NaCl; pH 10.5) was added together with 100 ml DMSO to solubilize the blue MTT-formazan product. After shaking for 10 min on a microplate shaker, the absorbance at 540 nm was determined using a E max 96-well spectrophotometer (DPC, Grimbergen, Belgium). Data are presented as mean+s.d. of three or four independent experiments. Within an experiment, the result of each experimental condition is the mean of six replicate wells. The IC 50 -values were calculated by non-linear regression analysis using SigmaPlot 4.01 software.

RT -PCR
T47D cells were seeded in Nunc TM 6-well plates at a concentration of 2610 6 cells per well in 5 ml medium. Cells were allowed to adhere to plastic for 24 h. After indicated periods of time, total RNA was isolated using UltraSpec-II (Biotecx Laboratories, TX, USA) according to manufacturer's instructions. Three micrograms of total RNA was heated at 658C for 5 min and rapidly cooled on ice. Then, 1500 ng oligo d(T) 12 -18 primer (Life Technologies, MD, USA) was added and cDNA was synthesized using 200 U ml 71 M-MLV RT (Life Technologies, MD, USA) in a RT reaction buffer (Life Technologies, MD, USA) supplemented with 20 U ml 71 RNAsin (Life Technologies, MD, USA), and 500 mM each dNTP (Perkin Elmer, CA, USA). The 60 ml reaction mixture was incubated at 378C for 1 h, followed by a 4 min incubation at 908C to inactivate the RT enzyme. Two microlitres of the RT reaction mix was used for the PCR reaction. Amplification was performed in a total volume of 50 ml containing 2 ml cDNA, 5 ml 106PCR reaction buffer (Boehringer Mannheim, Germany), 1 ml sense primer (20 mM), 1 ml antisense primer (20 mM), 4 ml dNTPs (each at a concentration of 2.5 mM; Perkin Elmer, CA, USA) and 0.25 ml AmpliTaq Gold (5 U ml 71 ; Perkin Elmer, CA, USA). AmpliTaq Gold was activated by a 10 min incubation at 948C. The samples were then subjected to 30 or 24 cycles of amplification for CYP26A1 or b 2 -microglobulin, respectively. Each cycle consisted of 1 min of denaturation at 928C, 1 min of annealing at 558C (CYP26A1) or 508C (b 2 -microglobulin), and 2 min of extension at 728C. At the end of the incubation an extra extension Experimental Therapeutics step was included (10 min at 728C). Ten microlitres of the PCR products were separated on a 1.3% (w/v) agarose gel, and visualized by UV light illumination after ethidium bromide staining. Specific primers were obtained from Eurogentec (Seraing, Belgium). Their sequences were as follows: CYP26A1 sense primer 5'-GCTGAAGAGTAAGGGTTTAC-3', and antisense primer 5'-CT-TGGGAATCTGGTATCCAT-3' (yielding a PCR-product of 184 bp); b 2 -microglobulin sense primer 5'-ACCCCCACTGAAA-AAGATGA-3', and antisense primer 5'-ATCTTCAAACCTCCAT-GATG-3' (yielding a PCR product of 114 bp).

TA3-Ha mammary carcinoma model
Oestrogen-independent TA3-Ha murine mammary carcinoma cells were grown by weekly intraperitoneal passage in vivo. Cells (1.5610 6 ) were injected intraperitoneally in syngeneic A/J mice. After 1 week, the peritoneum was rinsed with sterile saline and the cells further diluted in MEM-Rega3 cell culture medium (Life Technologies, Merelbeke, Belgium). The next in vivo passages (1610 4 cells) were done in allogeneic C3D2F1 mice, but every 5th passage syngeneic A/J mice were used as the host.   10 -10 10 -9 10 -8 10 -7 10 -6 10 -5 Figure 3 Inhibition of RA metabolism by R116010 and liarozole-fumarate. T47D human breast cancer cells were pretreated with 1 mM RA to induce RA metabolism. Concentration-response curves showing the inhibition by R116010 (*) and liarozole-fumarate (*) were determined using the microcolumn assay as described in Materials and Methods. Results are presented as mean+s.d. for R116010 (n=4) and liarozole-fumarate (n=3).
subcutaneous tumours were excised 24 h after the last treatment and weighed to validate the antitumoural effects of the treatment. Data are represented as box plots showing median group value, 25/ 75th percentile, 10/90th percentile and outliers. Groups were statistically compared to the vehicle-treated groups using the Mann -Whitney U-test. Significance was defined at the level of P50.05.
All animal experiments have been carried out with ethical committee approval. The ethical guidelines that were followed meet the standards required by the UKCCCR Guidelines.

R116010 is a potent inhibitor of RA metabolism
Human T47D breast cancer cells, cultured under control conditions, are unable to metabolize RA into more polar metabolites ( Figure  2A). After pretreatment for 16 h with 1 mM RA, T47D cells show extensive RA metabolism ( Figure 2B), converting RA into highly polar metabolites (retention time: 3 -10 min) and several metabolites with intermediate polarity (retention time: 15 -25 min), including 4-OH-RA, and 16-OH-RA. RA metabolism is completely inhibited by R116010 at a concentration of 1 mM ( Figure 2C).
A concentration-response curve (Figure 3), as measured with the quantitative microcolumn method, shows that R116010 inhibits the formation of polar [ 3 H]RA metabolites with a calculated IC 50 -value of 8.7+1.4 nM. In comparison, R116010 is over 100-fold more potent than liarozole-fumarate, a previously identified inhibitor of RA metabolism, that inhibits RA metabolism with an IC 50 -value of 1.4 mM (Figure 3).

Auto-induction of RA metabolism correlates with induction of CYP26A1
CYP26A1 is a cytochrome P450-dependent enzyme that specifically metabolizes RA (White et al, 1997). Human T47D breast cancer cells, cultured under control conditions, barely express detectable CYP26A1 mRNA levels ( Figure 4). CYP26A1 mRNA expression levels can be induced in a concentration-and time-dependent manner by treatment with RA ( Figure 4). The expression levels of CYP26A1 closely correlate with the observed enzyme activity ( Figure 2B and data not shown).

R116010 is a selective inhibitor of RA metabolism
The selectivity of R116010 towards CYP26A1 was tested by determining its inhibition profile against other cytochrome P450-mediated reactions. As shown in Table 1, R116010 inhibits CYP26A1-mediated RA metabolism in microsomes from T47D cells with an IC 50 -value of 4 nM. In contrast, even at high micromolar concentrations (10 mM), R116010 barely inhibits other cytochrome P450-mediated reactions, except for limited activity against 17,20-lyase (IC 50 =0.25 mM).

R116010 enhances the antiproliferative activity of RA
RA inhibits T47D cell proliferation in a concentration-dependent manner ( Figure 5A,B) with a calculated IC 50 -value of Experimental Therapeutics  2.0+0.5 nM (Table 2). R116010 by itself has no effect on T47D cell proliferation ( Figure 5A). However, in combination with RA, R116010 enhances the antiproliferative activity of RA in a concentration-dependent manner ( Figure 5A). At a concentration of 0.01 mM R116010 enhances the antiproliferative activity of RA by 1.25-fold (IC 50 -value=1.6+0.6 nM), at 0.1 mM R116010 by 2.6-fold (IC 50 -value=0.77+0.18 nM) and at 1 mM R116010 by three-fold (IC 50 -value=0.62+0.19 nM).
In contrast, liarozole-fumarate, tested up to a concentration of 10 mM, is unable to enhance the antiproliferative activity of RA ( Figure 5B). At this concentration (10 mM), liarozole-fumarate by itself decreases T47D cell proliferation by approximately 25% (Figure 5B). This decrease in T47D cell proliferation at 10 mM explains the apparent enhancement of RA activity, which is actually the effect of liarozole-fumarate alone ( Figure 5B).

R116010 inhibits the growth of murine TA3-Ha mammary tumours in vivo
Mice subcutaneously inoculated with oestrogen-independent TA3-Ha cells were treated twice-daily with RA ( Figure 6A) or R116010 ( Figure 6B) from day 1 until day 21. As shown in Figure  6A, RA inhibits TA3-Ha tumour growth in a dose-dependent manner. The lowest active dose is 2.5 mg kg 71 . At a dose of 5 mg kg 71 , mice lost 5% of body weight (data not shown) and toxicity was observed, which consisted mainly of hair loss, and impaired movement due to bone fractures.
R116010 potently inhibits TA3-Ha tumour growth at doses as low as 0.16 mg kg 71 ( Figure 6B). R116010 is approximately 60fold more potent than liarozole-fumarate which in this model inhibits tumour growth only at 10 mg kg 71 (data not shown). Toxicity, related to hypervitaminosis A, was observed at a dose of 5 mg kg 71 for R116010 and at 20 mg kg 71 for liarozole fumarate. At these doses hair loss and bone fractures were observed, and body weight decreased by *15% (data not shown).

DISCUSSION
The prompt emergence of resistance to RA treatment is a major hurdle in the clinical development of RA (Muindi et al, 1992;Miller, 1998). Drugs which are able to inhibit RA metabolism may therefore prove valuable in the treatment of cancer. In this study, we have identified R116010 as a highly potent and selective inhibitor of RA metabolism.
Human breast T47D carcinoma cells are known to have an inducible RA metabolism (Han and Choi, 1996;Van der Leede et al, 1997). The capacity to metabolize RA closely correlates with the expression level of CYP26A1, a cytochrome P450-dependent enzyme that specifically metabolizes RA (White et al, 1997). In untreated T47D cells, which are unable to metabolize RA, CYP26A1 expression is barely detectable. Treatment with RA rapidly induces the capacity to metabolize RA (Han and Choi, 1996), which is correlated with a strong induction of CYP26A1, in agreement with a previous report (Sonneveld et al, 1998). In intact T47D cells, R116010 potently inhibits RA metabolism with Experimental Therapeutics an IC 50 -value of 8.7 nM. As such, R116010 is more than 100-fold more potent than liarozole-fumarate, a previously identified inhibitor of RA metabolism (Wouters et al, 1992). As well as being a more potent inhibitor of RA metabolism, R116010 is also more selective as compared to liarozole-fumarate. Although our in vitro data suggest that R116010 might affect androgen levels by inhibiting 17,20-lyase, no change on blood hormone levels could be detected in rats treated with a high dose (10 mg kg 71 ) of R116010 (data not shown). Liarozole-fumarate is much less specific. It is a potent aromatase inhibitor (Bruynseels et al, 1990), and in addition it also inhibits other cytochrome P450mediated reactions, thereby affecting hormone levels both in animals and in humans (Bruynseels et al, 1990). Taken together, these data suggest that R116010 is less likely to produce adverse side effects.
In T47D cell proliferation experiments, R116010 enhances the antiproliferative activity of RA in a concentration-dependent manner. Concentrations effective in enhancing the biological activity of RA, are identical to the concentrations that are needed to inhibit RA metabolism in intact cells. Therefore, these data support the hypothesis that R116010 enhances the biological activity of RA through inhibition of RA metabolism. Liarozole-fumarate is unable to enhance the antiproliferative activity of RA in T47D cells. This is in contrast to our previous data, which clearly showed that liarozole-fumarate is able to enhance both the antiproliferative and differentiation-inducing activity of RA in MCF-7 human breast cancer cells (Wouters et al, 1992;Van heusden et al, 1996). This apparent contradiction might be explained by the fact that T47D cells are known to metabolize RA at a much higher rate than MCF-7 cells (Van der Leede et al, 1997; unpublished results) and that liarozole-fumarate is therefore not potent enough to effectively inhibit RA metabolism in T47D cells.
In vivo, R116010 has been shown to inhibit the growth of orthotopically implanted androgen-independent rat prostatic Dunning R3327/PIF-1 carcinoma cells (Van Ginckel et al, 1999), and of subcutaneous mouse Lewis lung (3LL) tumours (Smets et al, manuscript in preparation). In this study, the in vivo activity of R116010 has been evaluated in the murine oestrogen-independent TA3-Ha mammary carcinoma model. An oestrogen-independent model was chosen to exclude any hormonal mechanism of action. In addition, TA3-Ha cells are known to actively metabolize RA (Garrabrant and End, 1995) and TA3-Ha tumours are sensitive to RA treatment . Indeed, treatment with RA inhibits TA3-Ha tumour growth in a dose-dependent manner. However, keeping in mind that the lowest active dose is 2.5 mg kg 71 and that the maximum tolerable dose is 5 mg kg 71 , RA does not show a real therapeutic window. Similarly, liarozole-fumarate has no therapeutic window since a dose of 10 mg kg 71 has to be used to induce antitumour activity, and the maximum tolerated dose is 20 mg kg 71 (data not shown). In sharp contrast, R116010 inhibits TA3-Ha tumour growth at doses as low as 0.16 mg kg 71 . With a maximum tolerated dose of 5 mg kg 71 , R116010 exhibits a 30-fold therapeutic window. We speculate that this larger therapeutic window is due to the intratumoural action of R116010, thereby increasing intracellular levels of RA, while RA mainly remains in the circulation, and much higher doses are needed to produce a biological response. Given that R116010 has a broader therapeutic window, one can argue that R116010 may be a more effective anticancer drug. The observation of retinoid-like side effects at higher doses reinforces the hypothesis that R116010 exerts its antitumour activity via RA.
In conclusion, R116010 has been identified as a potent and selective inhibitor of RA metabolism. Inhibition of RA metabolism leads to an enhancement of the antiproliferative activity of RA both in vitro and in vivo. Inhibitors of RA metabolism may therefore represent a novel class of anticancer agents.