Preclinical Evaluation of Amino Acid Prodrugs of Novel Antitumor 2-(4-Amino-3-Methylphenyl)Benzothiazoles ¹

Novel 2-(4-aminophenyl)benzothiazoles (Fig. 1) possess potent and intriguing antitumor properties (1, 2). In the NCI

In vitro, induction of CYP1A1-catalyzed biotransformation of compound 1 is critical for execution of antitumor activity (4–6). Covalent binding, detected between 1 and recombinant CYP1A1, requires metabolism and is significantly reduced by glutathione (4). This suggests formation of an electrophilic, reactive intermediate species. Paradoxically, the major C-6 oxidation biotransformation product, liberated into nutrient media, is devoid of antitumor activity (7), possessing mitogenic properties at μm concentrations (8). Moreover, this metabolite antagonizes cellular uptake of 1, covalent binding between CYP1A1 and 1, CYP1A1 activity, and growth inhibition induced by 1. Thus, biotransformation to hydroxy derivatives is likely to effect the unique biphasic dose-response relationship characteristic of growth inhibition in vitro.

Ab initio Frontier Molecular Orbital Calculations have been used to predict the presence or absence of exportable hydroxy metabolites of 2-(4-amino-3-methylphenyl)fluorobenzothiazoles (9), and 2-(4-amino-3-methylphenyl)-5-fluorobenzothiazole (2) has been synthesized, which successfully blocks C-6 oxidation (8). Thus, compound 2 is the favored analogue for clinical consideration, demonstrating enhanced efficacy in vitro and superior potency in vivo against human breast and ovarian tumor xenografts implanted in nude mice. Lipophilicity, however, poses a pharmaceutical challenge, as an aqueous i.v. formulation is desired to minimize the possibility of first pass deactivating metabolism and improve drug bioavailability.

The primary aromatic amine, dapsone, has been derivatized as amino acid amides (10). Similarly, the exocyclic primary amine function of 2-(4-amino-3-methylphenyl)benzothiazoles has been conjugated successfully to alanine and lysine residues as mono- and dihydrochloride salts, respectively (11), fulfilling the criteria for a suitable prodrug: water solubility and chemical stability. In vitro, selective antitumor activity is retained as parent amine is rapidly and quantitatively regenerated in the presence of carcinoma cells.

Objectives of initial murine pharmacokinetic studies with 1a and 1b were to characterize the disposition of the compounds to allow optimization of dosing regimens for in vivo efficacy studies. Doses of 25–30 mg/kg were selected MTDs for repeated daily injection in mice. To approximate the doses of the initial studies, equimolar target doses (70 μm/kg) of prodrugs of fluorinated derivatives were selected. Male CD2F1 mice were given i.v. doses of 26.5 mg/kg 1a, 31 mg/kg 1b, 25.6 mg/kg 2a and 3a, or 32.2 mg/kg 2b and 3b as a 30-s infusion in the tail vein, using a vehicle of sterile water and a dose volume of 1 μl/gram body weight. The equivalent molar concentration of compound 2 (18 mg/kg) was similarly administered in a vehicle of DMSO. Groups of three mice were exsanguinated via the suborbital sinus at intervals from 2 min to 6 h. Plasma proteins were precipitated by the addition of 150 μl of acetonitrile to 50 μl of plasma. Samples were vortexed for 1 min and centrifuged for 5 min (14,000 × g). Supernatant (150 μl) was removed and mixed with 100 μl of 10% acetonitrile in 1% acetic acid. HPLC separations were effected on a J‘Sphere ODS-AQ analytical column (4.6 × 150 mm; YMC, Inc., Wilmington, NC). The column was eluted at a flow rate of 1 ml/min with sequential linear gradients. The mobile phase was formed by increasing the acetonitrile concentration in 1% acetic acid over 20 min by mixing acetonitrile-water-acetic acid solutions: (10:90:1, volume for volume and 80:20:1, volume for volume). Samples (230 μl) were injected into the system and concurrently analyzed for concentrations of prodrug and parent amine. Column eluant was monitored at 324 nm using HP 1050 series UV diode array detector.

In an initial study, rats received 25 mg/kg 1a and 1b i.v. HPLC methods were developed to allow simultaneous detection of 1, 1a, and 1b in addition to resolution of degradation products and metabolites. Separation was effected at room temperature on a Primesphere 5C-18 HC110 column (250 × 3.2 mm) using a mobile phase of 40% acetonitrile in 5 mm potassium dihydrogen phosphate (pH 7) delivered at a flow rate of 0.8 ml/min. Compounds were detected at 320 nm.

A single male Sprague Dawley rat received either 25 mg/kg 1a or 31 mg/kg 1b in sterile water as a 30-s infusion in a catheter implanted in the jugular vein (dose volume 1 μl/gram body weight). Blood samples were taken through the catheter at selected intervals from 3 min up to 6 h. Plasma samples were concurrently monitored for concentrations of 1, 1a, and 1b by HPLC analysis.

Compound 2b (2, 4, and 14.3 mg/kg) was administered to one male and one female beagle dog per treatment group as a 1-h civ. This concentration (14.3 mg/kg 2) is the molar equivalent of 7.5 mg/kg 1, used previously in mouse efficacy studies performed at the NCI.

Plots of plasma concentration against time were constructed. Polyexponential, nonlinear regression analysis was performed on the plasma concentration-time profiles using an iterative computer-assisted curve-fitting program (Table Curve; Jandel Scientific, San Rafael, CA). Computer analyses of parameters, including slopes and intercepts of the regression lines, allowed calculation of secondary pharmacokinetic parameters, such as area under the concentration-time profile and efficiency of prodrug conversion to parent amine.

Female B6D2F1 nude mice bearing s.c. MCF-7 (ER+) and MT-1 (ER−) breast xenograft tumors were treated with 1a and 1b administered i.v. in a vehicle of saline/Tween 80. The target schedule of three times once weekly with doses ranging from 5 to 25 mg/kg/injection, guided by previous studies (1), enabled MTD determination.

IGROV-1 ovarian and MCF-7 breast xenografts were transplanted s.c. into flanks of NCR-Nu female nude mice. Animals were treated i.v. with 2b in saline according to the following schedules: 20 mg/kg single dose, 20 mg/kg days 0 and 7, 4 mg/kg days 0 and 1–4. Drinking water of mice bearing MCF-7 xenografts was supplemented with 60-day release 17β-estradiol pellets (0.72 mg/pellet; Innovative Research), maintaining blood levels of 300–400 pg/ml. Antitumor activity in MCF-7 xenografts of prodrug 2b was compared with parent amine 2. To examine the selective nature of growth inhibition in vivo, mice bearing MCF-7 and MDA-MB-435 tumors in opposite flanks were treated with 5 mg/kg compound 2 in DMSO (10%) plus oil, administered i.p. on 4 consecutive days (days 0–3).

A male and one female dog received one of the following treatments: vehicle alone (5% dextrose in water), 1-h civ 2 mg/kg compound 2b, 1-h civ 4 mg/kg 2b, and 24-h civ 14.3 mg/kg 2b. Hematology and clinical chemistry parameters were measured, and histopathology evaluation was conducted on one dog that became moribund.

Tissues were recovered from mice 24 h after treatment with 2b (20 mg/kg, i.p.) or vehicle alone (DMSO + oil) and snap frozen. Tumors, lungs, and livers were homogenized in buffer containing 20 mm Tris-HCl (pH 7.4), 2 mm EDTA (pH 7.4), 2 mm EGTA (pH 7.4), 6 mm β-mercaptoethanol, 10 μg/ml leupeptin, 2 μg/ml aprotinin, and 1% NP40. Homogenates were sonicated (MSE Soniprep 150 3 × 10 s, amplitude 20 μm) and prepared for examination of CYP1A1 protein expression. After protein determination (n = 3; Ref. 12) and addition of sample buffer, samples were boiled at 95°C for 5 min, and solubilized proteins (50 μg) were separated by SDS polyacrylamide gel (10%) electrophoresis. Proteins were electroblotted to polyvinylidene difluoride membranes and probed for CYP1A1 protein with polyclonal antiserum specific for human CYP1A1/1A2 (Gentest Corp.). Secondary antibody was conjugated to alkaline phosphatase, and CYP1A1 was detected after brief (<10 min) incubation with bromochloroindolyl phosphate and nitroblue tetrazolium in alkaline phosphatase buffer. Molecular weight markers and a positive control of recombinant CYP1A1 (Gentest Corp.), included in blots, confirmed detection of M_r 52,000 CYP1A1 protein.

A summary of the pharmacokinetic results obtained after prodrug administration to mice, rats, and dogs is given in Table 1.

Plasma concentrations of 1 achieved after injection of 1a exceeded those reached after injection of 1b. The area under the plasma concentration-time profile was greater by >40% and efficiency of conversion, or bioavailable fraction, of prodrug to parent was 24% with 1a and 18% with 1b.

Concentrations of 2b and 2 in plasma, after i.v. administration of 32.2 mg/kg 2b are shown in Fig. 2A Plasma concentrations of 2b declined in a biexponential manner with elimination half-lives (t_1/2β) for the initial and terminal phases of 2 and 43 min, respectively. The total rate of plasma clearance was 191 ml/min/kg. Plasma levels of 2 rapidly increased, maintaining concentrations between 2 and 4 μm between 13 min and 120 min after administration of 2b (Table 1, duration of maximum amine plasma concentrations). A peak concentration of 4.11 μm 2 was achieved after 1 h and remained above 1 μm for 4 h after treatment, thereafter decreasing with t_1/2β of 110 min. The efficiency of conversion, or bioavailable fraction, of prodrug to parent compound 2 was 42% with 2a and 45% with 2b. By comparison, the pharmacokinetic properties of 2 have been determined after equimolar i.v. administration. Plasma concentrations declined in a biexponential manner with initial and terminal t_1/2β of 19 and 99 min, respectively. Total plasma clearance rate was 41 ml/min/kg, indicating a more rapid rate of elimination of 2 than for compound 1 (t_1/2β 187 min, plasma clearance rate 22 ml/min/kg; results not shown).

Both plasma concentration and area under the plasma concentration-time profile of 3, after injection of 3b, was substantially higher than after injection of 3a. However, the area under the plasma concentration-time profile of 2, after administration of equimolar doses of 2a and 2b, exceeded that for 3.

Although 1a and 1b were stable in rat plasma in vitro, both these compounds were readily removed from plasma and converted to parent amine 1 when either prodrug was given to rats i.v. at 25 mg/kg; compound 1 was detectable in plasma as early as 3 min postadministration. Regenerated 1 was further N-acetylated, this being the main metabolic route of 1 in rats (13), forming the inactive N-acetyl derivative 9 min after dosing. N-acetylation was extensive, such that negligible liberated 1 could be detected after 360 min (result not shown).

In a separate study, summarized in Table 1, the efficiency of conversion, or bioavailable fraction of prodrug to parent, was 87% with 1a and 106% with 1b, far greater than observed in mice.

Prodrug 2b (14.3 mg/kg) was rapidly converted to parent amine 2 with t_1/2β of 17 and 30 min in one male and one female dog (Fig. 2B), respectively. Plasma concentrations of 2 were between 2 and 5 μm between 60 and 300 min after the start of the infusion. At doses of 2 and 4 mg/kg 2b, peak plasma concentrations of 2 were 0.5 and 0.9 μm, respectively.

Gastrointestinal toxicity (emesis, diarrhea, and low food consumption) and lethargy were exhibited by two dogs for 5–7 days after treatment with 14.3 mg/kg 2b, and by day 7, the dogs had lost between 9 and 14% body weight. Increased levels of serum bile acids, γ glutamyltransferase, aspartate aminotransferase, alanine aminotransferase, and alkaline phosphatase were measured. On day 5, the male dog that received a 24-h infusion exhibited labored respiration and was sacrificed because of moribund condition. Histopathology examination revealed depletion/atrophy in the bone marrow, lymphoid tissues and gut-associated lymphoid tissue, intestinal lesions, and extensive acute inflammation in the alveolar spaces with edema and hemorrhage; pulmonary toxicity was dose limiting. At doses of 2 and 4 mg/kg 2b (1-h civ), emesis and soft stool occurred on the day of dosing, but during days 1–5, no effects on clinical pathology parameters and no lesions in lung or liver tissues were observed. Moreover, plasma concentrations of 2 reached a peak of 0.9 μm, an efficacious concentration in vitro against sensitive breast cancer cell lines (MCF-7, ZR 75, T47D; 8).

In vivo antitumor activity MTDs of 1a and 1b, established in female B6D2F1 mice after i.v. injection, was 25 mg/kg and resulted in catalepsia and convulsions. MTD for three times once weekly repeat treatments of 1a was 15 mg/kg/i.v. injection and for 1b was between 10 and 15 mg/kg/i.v. injection. Antitumor activity of 1a and 1b was examined in MCF-7 (ER+) and MT-1 (ER−) breast xenografts implanted s.c. into the flanks of nude mice. Significant antitumor effects were elicited by both 1a and 1b in MT-1 carcinoma xenografts adopting a three times once weekly schedule (percentage of treated versus control growth < 50%). In MCF-7 xenografts, tumor growth retardation was not as prominent; however, at a dose of 12.5 mg/kg, 1a caused greater growth retardation than 1b. Negligible hematological effects were observed; however, body weight reduction and toxicity accompanied significant tumor growth inhibition (84%) after treatment with 25 mg/kg 1a.

The selective nature of growth inhibition of 2-(4-aminophenyl)benzothiazoles in vitro has been documented previously (1–3). GI₅₀ values < 10 nm were obtained in sensitive cells (e.g., MCF-7), whereas GI₅₀ values exceeding 100 μm were encountered in inherently resistant cell lines (e.g., MDA-MB-435 mammary carcinoma cells). Mechanisms underlying such stark selectivity have been studied (7, 4). The selective nature of growth inhibition in vivo has now been demonstrated unequivocally in a study where MCF-7 and MDA-MB-435 xenografts were transplanted in opposite flanks of the same mouse. The growth of MCF-7 tumors was only significantly retarded after i.p. treatment with compound 2 (Fig. 3A). Antitumor activity of amine 2 was not compromised after amino acid derivatization: significant in vivo tumor growth inhibition was encountered in nude mice bearing MCF-7 breast and IGROV-1 ovarian xenografts after treatment with 2b.

To optimize the dose, three schedules were compared in the IGROV-1 model and were found to be equiactive (Fig. 3B). In contrast, 20 mg/kg 2b administered on days 0 and 7 elicited superior antitumor activity in MCF-7 xenografts when compared with a single dose of 20 mg/kg, 4 mg/kg administered daily for 5 consecutive days (Fig. 3C), or 10 mg/kg administered three times once weekly (Fig. 3D).

We have observed previously the selective induction of CYP1A1 protein expression in cell lines sensitive to growth inhibitory properties of 1 and reported NADPH-dependent covalent binding between [¹⁴C]1 and recombinant CYP1A1 (4). During this study, we examined CYP1A1 protein expression in human tumor xenografts. Fig. 4A demonstrates induction of CYP1A1 protein expression in homogenates of IGROV-1 and MCF-7 tumors, 24 h after treatment of mice with prodrug 2b (20 mg/kg, i.p.). No CYP1A1 protein was detected in homogenates of IGROV-1 and MCF-7 tumors recovered from untreated mice. In MDA-MB-435 xenografts, CYP1A1 protein expression was neither constitutive nor induced by treatment of mice with 20 mg/kg 2b.

Lung and liver tissues were recovered from animals 24 h after treatment (20 mg/kg 2b). Fig. 4B compares CYP1A1 protein expression in these tissues with expression in lungs and livers of untreated mice. CYP1A1 protein was clearly induced in the lungs of mice treated with 2b; constitutive expression was not detected. In livers, CYP1A1 protein was expressed constitutively and after treatment of mice with 2b.

Data have been presented describing the biological properties of alanyl and lysyl amide derivatives of 2-(4-aminophenyl)benzothiazoles in vivo. Prodrugs undergo rapid and quantitative reversion to parent species⁴ (e.g., 2b, Fig. 2). Plasma concentrations of 2, 2a, and 2b in mice have been compared with the concentrations required to elicit GI₅₀, TGI, and 50% lethality in MCF-7, T47D, and ZR-75 breast cell lines after 45-min exposure to 2, 2a, or 2b.

The remarkably potent and selective antitumor activity characteristic of compound 2 in vitro (8) is retained in vivo (Fig. 3A) and after prodrug modification in vitro⁴ and in vivo (Fig. 3, B–D); 2b suppresses significantly the growth of MCF-7 breast and IGROV-1 ovarian xenografts in vivo.

An overview of our comprehension of the mechanisms of action of this class of agent is illustrated in Fig. 5. Planar, hydrophobic 2-(4-aminophenyl)benzothiazoles fulfill structural requirements for binding to the AhR and are potent AhR agonists (14). Enhanced CYP1A1 transcription in sensitive cells (5) is accompanied by translocation of AhR to the nucleus, increase in xenobiotic response element (XRE)-driven luciferase activity, and induced formation of protein-DNA complexes on the XRE sequence of the CYP1A1 promoter. Highly specific induction of CYP1A1 activity and protein expression in sensitive cell lines only follows exposure of cells to 2-(4-aminophenyl)benzothiazoles (4), including IGROV-1 and MCF-7 cells. In vivo, only tumor xenografts whose growth was inhibited by 2b (Fig. 3, B–D) demonstrated inducible CYP1A1 protein (Fig. 4). This observation raises the possibility that monitoring CYP1A1 expression in human tumors is achievable and may provide a biomarker for the identification of sensitive tumor phenotypes.

CYP1A1-mediated metabolism of aminophenylbenzothiazoles generates putative reactive electrophilic species and benzothiazole-derived covalent binding to CYP1A1 (4), which leads to the formation of significant DNA adducts,

Thus, in those organs possessing inducible CYP1A1 (e.g., liver and lung), benzothiazole-induced hepato or pulmonary toxicity must be anticipated. 2b is able to induce CYP1A1 protein expression in the lungs of treated mice (20 mg/kg, Fig. 4B). Indeed, dose-limiting pulmonary lesions were detected in one dog receiving 14.3 mg/kg 2b over 24 h; in addition, increased levels of serum bile acids, γ glutamyltransferase, aspartate aminotransferase, alanine aminotransferase, and alkaline phosphatase were consistent with liver damage. However, doses of 2 and 4 mg/kg were well tolerated and yielded efficacious plasma concentrations of amine 2, indicating that a therapeutic window may be achievable. Preliminary studies have revealed the presence of benzothiazole-derived DNA adducts within the livers but not the lungs of mice receiving a therapeutically effective concentration of 2b (20 mg/kg).⁷ However, in human hepatocyte culture, viability remained > 65% after exposure of hepatocytes to 1 or 10 μm compound 2 (96 h). Future experiments will investigate the potential for DNA repair in murine liver.

To conclude, alanyl- and lysyl-amide hydrochloride salts may be conjugated to the exocyclic primary amine function of 2-(4-aminophenyl)benzothiazoles (11), rendering these prodrugs chemically stable, water soluble, and suitable for parenteral delivery. Of the prodrug candidates examined, 2b presents superior pharmacokinetic properties, with manageable toxicity in dogs at concentrations liberating plasma levels of 2, which are potentially efficacious. In vivo antitumor tests revealed selectivity and efficacy; in addition, CYP1A1 protein expression is induced in sensitive xenograft models after treatment of tumor-bearing mice with 2b. In view of the encouraging preclinical properties expounded by the lysylamide dihydrochloride salt, 2b, combined with the superior efficacy and retained selectivity of compound 2, prodrug 2b has been selected to undergo Phase I clinical evaluation in 2002, under the auspices of the Cancer Research Campaign, United Kingdom.

We thank the Cancer Research Campaign (CRC), United Kingdom, for support to the CRC Experimental Cancer Chemotherapy Research Group, Nottingham, United Kingdom, and the Cancer Research Unit, Bradford, United Kingdom. We gratefully acknowledge extensive collaborations, support, and discussions with the staff of the Developmental Therapeutics Program of the NCI and members of the Screening and Pharmacology Group within the European Organization for Research and Treatment of Cancer.