J. Med. Chem. 2005, 48, 3903-3918
3903
A Novel Class of in Vivo Active Anticancer Agents: Achiral seco-Amino- and seco-Hydroxycyclopropylbenz[e]indolone (seco-CBI) Analogues of the Duocarmycins and CC-1065 Atsushi Sato,†,‡ LuAnne McNulty,† Kari Cox,† Susan Kim,† Adrienne Scott,† Kristen Daniell,† Kaitlin Summerville,† Carly Price,† Stephen Hudson,† Konstantinos Kiakos,§ John A. Hartley,§ Tetsuji Asao,‡ and Moses Lee*,† Department of Chemistry, Furman University, 3300 Poinsett Highway, Greenville, South Carolina 29613, Cancer Research UK Drug-DNA Interactions Research Group, Department of Oncology, University College London, 91 Riding House Street, London, W1W 7BS, U.K., and Taiho Pharmaceutical Co., Ltd., 1-27, Misugidai Hanno-City, Saitama, 357-8527, Japan Received February 24, 2005
One achiral seco-hydroxycyclopropylbenz[e]indolone (seco-CBI) (12) and seven achiral secoamino-CBI (11a-g) analogues of CC-1065 and the duocarmycins were designed, synthesized and evaluated for their DNA-binding and anticancer properties. These compounds contain a core 2-chloroethylnaphthalene structure and they do not have a stereocenter. From thermal cleavage gel analyses, compounds 11a-g and 12 demonstrated similar covalent sequence specificity to adozelesin 3 and the racemic seco-CBI-TMI 4 for binding to the 5′-AAAAA(865)3′ site. Continuous exposure of human (K562) and murine (B16, L1210 and P815) cancer cell lines to the compounds demonstrated their significant cytotoxicity, with IC50 values in the sub-micromolar range. Generally, a good leaving group on the ethyl moiety and a free amino or hydroxyl group on the naphthyl moiety are essential for activity. According to NCI’s cytotoxicity screen, compounds 11a and 12 were active against human cancer cell lines derived from lung, colon, melanoma, renal system, and breast. At the respective doses of 15 and 20 mg/kg (administered via an ip route), compounds 11a and 12 inhibited the growth of murine B16-F0 melanoma in C57BL/6 mice, with minimal toxicity, and 11a gave a significant anticancer effect. The in vivo anticancer activity of compound 11a was confirmed in a human tumor xenograft study (advanced stage SC-OVCAR-3 ovarian cancer growing in scid mice). Finally, compound 11a was not toxic to murine bone marrow cell growth in culture at a dose that was toxic for the previously reported compound 4. Introduction (+)-CC-1065 (1)1,2 and duocarmycin SA (DUMSA, 2)2,3 (Figure 1) are cyclopropylpyrrolo[e]indolone (CPI)containing alkaloids isolated from the fermentation broth of Streptomyces zelensis4 and Streptomyces sp.5 These potent cytotoxic compounds derive their biological activity by binding the minor groove of AT-rich sequences and covalently reacting between the CPI moiety and adenine-N3. This mechanism of action was confirmed by the isolation of adenine-CC-10656 and adenineduocarmycin SA products7 from thermally cleaved DNA-drug adducts. (+)-CC-1065 and (+)-DUMSA preferentially bind 5′-PuNTTA-3′ and 5′-AAAAA-3′ sequences and they react with the N3-position of the underlined adenine residue.6-8 Due to the severe toxicity of (+)-CC-1065 to mice4b and the myelotoxicity of DUMSA,5 neither compound was developed further. Following detailed structureactivity relationship studies on these compounds, adozelesin (3),1b,4c,9 carzelesin,10 bizelesin,11 and KW218912 have been evaluated clinically. Because of severe myelotoxicity, three of the compounds were dropped from * Corresponding author. Phone: (864) 294-3368. Fax: (864) 2943559. E-mail:
[email protected]. † Furman University. ‡ Taiho Pharmaceutical Co., Ltd. § University College London.
clinical studies,9-11 and presently, only bizelesin remains in clinical trial.11 Consequently, there is a strong interest in the design and development of novel analogues of CC-1065 and the duocarmycins that maintain antitumor activity but have reduced toxicity. In numerous attempts to design novel duocarmycins and CC-1065 analogues with reduced toxicity to bone marrow cells, a wide range of compounds related to the CPI subunit have been synthesized and tested.1 Most drug design studies were done on the basis that the cytotoxic potency of the CPI-containing compounds was directly related to their solvolytic stability. DUMSA, one of the most cytotoxic analogues, is also the most solvolytically stable member of this class of compounds.2,3,13 Examples of some analogues that have been investigated are cyclopropylbenz[e]indolone (CBI),1.14 cyclopropylpyrazolo[e]indolone (CPzI),15 cyclopropylfurano[e]indolone (CFI),16 and cyclopropylindolone (CI).17 seco-Hydroxy-CBI-TMI (4)14 and its amino congener (5) are examples of CBI analogues of the duocarmycins that have been reported.18 Furthermore, our laboratories have recently reported a seco-iso-CFI-TMI (isocyclopropylfurano[e]indolone) analogue to have potent anticancer activity but with relatively low toxicity to murine bone marrow cells, when compared to compound 4.19 These results suggest the potential of designing compounds with more
10.1021/jm050179u CCC: $30.25 © 2005 American Chemical Society Published on Web 05/05/2005
3904
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11
Sato et al.
Figure 1. Structures of (+)-CC-1065 (1), (+)-duocarmycin SA (2), adozelesin (3), seco-hydroxy-CBI-TMI (4), and seco-aminoCBI-TMI (5). Also depicted are structures of the achiral seco-amino and hydroxy analogues of CC-1065 and the duocarmycins 6-8, along with their chiral counterparts 9 and 10.
favorable toxicity profiles than the currently investigated analogues of the duocarmycins and CC-1065. We have undertaken a program to investigate whether the chiral center present in the natural products is needed for DNA sequence recognition and biological activity. Our group has reported studies on the design, synthesis, and biological properties of three types of achiral analogues represented by seco-hydroxy-CI-TMI (6),20 seco-amino-CI-TMI (7),21 and seco-duocarmycin (8).22a These compounds, which lack a stereocenter, were found to have similar DNA-binding preference, as well as having comparable cytotoxic potency to their racemic counterparts, agents 9 (IC50 0.15 µM with K562 cells)20,21 and 10 (IC50 1.38 nM with P388 cells).22b Moreover, compound 6 was able to inhibit the growth of a human advanced stage SC UACC-257 melanoma grown in severe combined immunodeficiency (scid) mice.20 These results have suggested that achiral analogues of CC1065 and the duocarmycins could form a novel class of anticancer agents. Because CBI-containing compounds, e.g. 414 and 5,18 are significantly more cytotoxic than their corresponding CI-containing analogues, e.g. 9, it was important to study the DNA-binding and biological properties of a novel class of achiral seco-CBI analogues. In this report, we describe the synthesis of seven secoamino-CBI analogues (11a-g) and a seco-hydroxy-CBI analogue (12) (Figure 2), which contain a substituted 2-chloroethylnaphthalene core. The DNA sequence specific alkylation and anticancer properties of the target compounds are also reported.
Figure 2. Target achiral seco-amino-CBI analogues (11a-g) and an achiral seco-hydroxy-CBI-TMI analogue (12).
Results and Discussion Synthesis. The synthesis of compounds 11a-e is shown in Scheme 1, and it follows a similar strategy that was used for the preparation of compounds 620 and 7.21 Reaction of 1-chloro-2,4-dinitronaphthalene (13)23 with tert-butyl ethyl malonate provided diester 14 in 54% yield. Subsequent acid-promoted removal of the
Anticancer Analogues of the Duocarmycins
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11 3905
Scheme 1
tert-butyl group, followed by decarboxylation of compound 14, gave ester 15 in 70% yield, which was selectively reduced using sodium sulfide. The amino product 16 was isolated in 11% yield after silica gel chromatography, and it was allowed to react with benzyl chloroformate to give the N-benzyl-protected compound 17 in 29% yield. Hydrolysis of the ester, followed by selective reduction of the carboxylic acid group in 18 with borane, generated alcohol 19 in 49%. Reaction of alcohol 19 with acetic anhydride provided acetate 20 in 55% yield. Reduction of the nitro group by catalytic hydrogenation over Adam’s catalyst at 55 psi, followed by coupling of the amine with 5,6,7-trimethoxyindole2-carboxylic acid in the presence of PyBOP and Hunig’s base, gave compound 21a in 81% yield. Methanolysis of the acetate group in compound 21a gave alcohol 22a in 82% yield. Mesylation of compound 22a, followed by chlorination, gave product 23a in 62% yield. Final catalytic hydrogenation of compound 23a over 10% palladium-on-carbon provided the target achiral compound 11a in 74% yield. Synthesis of compound 11b was accomplished in 74% by catalytic hydrogenation of intermediate 21a. Likewise, hydrogenation of intermediate 22a gave compound 11c in 71% yield. The preparation of compounds 11d,e was achieved by reduction of compound 20, followed by coupling of the amine with the appropriate carboxylic acids in the presence of EDCI. The intermediates 21d,e were obtained in quantitative yields. Subsequent re-
moval of the acetate group, chlorination, and deprotection of the benzyl group afforded the target compounds 11d,e in an overall yield of 5 and 24%, respectively. The pyrido compounds 11f,g were prepared using a different strategy and it is shown in Scheme 2. Amino ester 16 was reduced by DiBAL-H to give an alcohol intermediate that was protected by reaction with acetic anhydride and di-tert-butyl dicarbonate [(BOC)2O] to afford compound 24 in an overall 10% yield. Reduction of the nitro group on compound 24, followed by coupling to the appropriate pyrido carboxylic acids,12e,24 gave amides 25f,g in 56 and 61% yields, respectively. Subsequent transformation of the acetate group to a chloride and removal of the BOC protecting group gave the desired products 11f,g as hydrochloride salts in an overall yield of 79 and 84%, respectively. Our strategy for synthesizing compound 12 is illustrated in Scheme 3. Diazotization of amino ester 16 and subsequent protection of the naphthol by reaction with benzyl bromide produced compound 27 in 26% overall yield. Reaction of ester 27 with DIBAL-H, followed by treatment of the resultant product 28 with acetic anhydride gave acetate 3 in an overall 62% yield. Selective reduction of the nitro group followed by coupling of the amine with 5,6,7-trimethoxyindole-2carboxylic acid in the presence PyBOP and Hunig’s base gave amide 30 in 36% yield. Standard conversion of the acetate into a chloride 31, followed by removal of the benzyl protecting group by catalytic hydrogenation, gave
3906
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11
Sato et al.
Scheme 2
Scheme 3
compound 12 in an overall yield of 63%. All intermediates and products were characterized by FT-IR, 500 MHz 1H NMR, MS, and accurate mass measurements. Compounds 11a-e and 12 were also characterized by elemental analysis. Thermal Cleavage Analysis. The sequence specificity of alkylation by compounds 11a,d-g and 12 was assessed by a thermally induced DNA strand cleavage experiment, which is commonly used to probe sequence specific covalent bonding with purine-N3 in the minor groove.2,3,14,19 The 191 base pair DNA fragment used in these studies was obtained by PCR amplification from the pUC18 plasmid that was linearized with Hind III. A 5′-32P labeled primer was used as the forward primer so that each final probe copy was singly end-labeled. Representative results from the thermally induced DNA strand break experiment on compounds 11a, 12 along with 3 and 4 are depicted in Figure 3. The results indicate that the achiral seco-CBI compounds 11a and 12 have similar DNA sequence selectivity for the 3′-A(865)AAAA cluster. Like compound 4, the achiral compounds displayed a dose-response DNA alkylation effect. However, the seco-hydroxy analogue 12 was around 10-fold less reactive than 11a. It is also worthy to note that the achiral compounds were more sequence discriminating than compounds 3 and 4. The latter compounds produced additional minor bands on the gel: A(764), A(772), and A(843). Taq Polymerase Stop Assay. The covalent sequence specificity of compounds 11a, 12, and adozelesin (3)1b-d was confirmed by the results obtained from a PCR-based assay (data given in the Supporting Infor-
mation).25 The studies were performed using a singly radiolabeled primer 5′-32P-CTCACTCAAAGGCGGTAATAC-3′ and Hind III linearized pUC18 plasmid DNA.
Figure 3. Thermal cleavage gel analysis on compounds 11a and 12, compared to seco-CBI-TMI (4) and adozelesin (3). A pUC18 plasmid was linearized with Hind III and amplified using a 5′-32P-CTGTCGGGTTT-3′ primer. The drug-DNA adducts were heated at 90 °C for 30 min prior to loading onto the gel.
Anticancer Analogues of the Duocarmycins
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11 3907
Table 1. Cytotoxicity of Compounds 11a-g and 12a compd
K562
L1210
B16
P815
11a 11b 11c 11d 11e 11f 11g 12 8 4 23a 31 6 7
0.045
0.18 53 41 0.087 0.39 0.55 0.18 0.055 4.2 0.0084
0.15 54 60 0.043 0.26 0.75 0.062 0.027 6.8 0.078 5.4 >100
0.068
0.15 0.25 0.0028 1.4320 1.3021
58 1.520
0.025 5.4 0.0086 5.620
a IC 50 values are given in µM. The cells were continuously treated with compounds for 3 days, except K562 cells, which were treated for 4 days.
The region of the plasmid starting from position 749 was linearly amplified. All three compounds showed similar covalent sequence selectivity at the A(865) cluster and are consistent with results from the thermal cleavage experiment. On the basis of the sequence specificity results, it became evident that the chiral center present in the CC-1065 and the duocarmycins did not play a significant role in controlling the DNA interaction properties. Coupled to their relative ease of synthesis, the achiral seco-CBI compounds provide an attractive platform for the design and development of a novel class of anticancer agents. Cytotoxicity Studies. An MTT-based growth inhibition assay was used to determine the cytotoxicity of the target achiral compounds.26 The concentrations required for inhibiting the growth of the human K562 chronic myeloid leukemia, murine L1210 leukemia, P815 mastocytoma, and B16-F0 melanoma cells by 50% (IC50 values) for compounds 11a-g and 12 were determined following continuous exposure.19-22 IC50 values are given in Table 1. For comparison, the reported IC50 value of CC-1065 against L1210 cells (3 days exposure) was 30 pM.27 The results reveal several interesting trends. First, compounds 11a,d-g and 12 inhibited the growth of the tumor cells at sub-micromolar concentrations, with compounds 11a and 12 being the most active. These compounds were more active than the corresponding achiral seco-CI analogues (e.g. 6 and 7). They were, however, less active than the racemic seco-CBITMI compound 4. Nonetheless, the achiral seco-CBI compounds have significant activity and are worthy of further biological testing. Second, protection of the hydroxyl or amino group, found in compounds 31 and 23a, led to a dramatic decrease in cytotoxicity. These compounds gave IC50 values of >100 and 5.4 µM, respectively, against B16 cells. In contrast, their corresponding hydroxy and amino counterparts, 12 and 11a, gave IC50 values for B16 cells of 0.027 and 0.15 µM, respectively. Third, for the seco-amino-CBI series of compounds, agents 11b and 11c gave poor cytotoxicity compared to their chloridecontaining counterparts, suggesting that a good leaving group on the ethyl substituent in the naphthalene core was necessary for inhibiting the growth of cancer cells. These results indicate that the seco-achiral hydroxy- or aminonaphthethyl halide compounds must eliminate
HCl to form spiro[2,5]cyclopropanebenzocyclohexadienone intermediates that react with DNA to elicit the cytotoxic activity. This observation is consistent with the biological activity of seco-analogues of CC-1065, the duocarmycins,2,3 as well as the achiral seco-CI and DUMSA compounds (6-8).20-22 Compounds 11a and 12 were subjected to further cytotoxicity testing at the National Cancer Institute, against a panel of 60 human cancer cell lines.28 The viability of the cells after 48 h of continuous exposure to the compounds was determined using a sulforhodamine B assay. Both compounds gave comparable cytotoxicity results, and Figure 4 shows the LC50 values (concentration for killing 50% of the cells) for compound 11a against the 60 different tumor cell lines. The LC50 results for compound 12 are given in the Supporting Information. Bars extending to the right indicate cells more sensitive than the average to the particular compound, whereas bars to the left indicate less sensitive cells. Several observations were made on the results. First, the compounds exhibited activity in the sub-micromolar range, and in certain colon, central nervous system (CNS), and skin cancers, the LC50 values were in the nanomolar range for compound 12. Second, compounds 11a and 12 were not indiscriminately toxic to cells. For example, they were not active against leukemic cells. Third, these compounds showed activity against cells from many human solid tumors, including melanoma, lung, colon, CNS, ovary, and breast. Specific reasons for the unique patterns of cytotoxicity for these compounds are unknown, and experiments are underway to address this issue. The cytotoxicity profiles for agent 11a and 12 are similar to that reported for achiral seco-CI-TMI (6).21 Cell Cycle Studies. Duocarmycin analogues have been demonstrated to induce leukemic cells to undergo apoptosis. The cells demonstrated morphological abnormalities and genomic DNA degradation associated with apoptosis.21,29 Flow cytometry was used to detect apoptotic cells as a sub-G0 peak,30 using propidium iodide staining. Following a 24-hour incubation of P815 cells at the IC50 concentration (0.068 µM) as well as 10-times the IC50 concentration of compound 11a, the results given in Figure 5 show a substantial increase in the percentage of cells in the sub-G0 stage (11% at IC50, and 52% at 10 times IC50 concentration), compared to 1% percent for untreated control cells. These results indicate that compound 11a was capable of damaging DNA and inducing the cells to undergo apoptosis. For comparison, cisplatin at 1000 µM produced 59% percent of sub-G0 population of cells, which is consistent with a literature report.30 Results from our studies suggest that, like their seco-CC-1065, seco-duocarmycins, the achiral analogues, 620 and 11a are likely to exert their cytotoxic activity through the induction of apoptosis. In Vivo Anticancer Studies on B16 Melanoma. This study was conducted according to a similar procedure reported by Li et al.31 For each group of animals in the experiment, eight 5-7-week-old female C57BL/6 mice were inoculated in the left flank with cultured B16F0 murine melanoma cells at day 0. On days 1, 5, and 9, the mice were treated with the appropriate doses, based on their maximum tolerated doses, of compound 11a (15 mg/kg) or compound 12 (15 and 20 mg/kg) or
3908
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11
Sato et al.
Figure 4. NCI cytotoxicity screen data (LC50 values) for compounds 11a against a panel of 60 different human cancer cell lines.
cyclophosphamide (30 mg/kg)29b or vehicle only. The compounds were formulated in PET/glucose and administered using an interperitoneal route. The volume of injection was adjusted according to a ratio of 0.02 mL for every 10 g of body weight. The size (volume) of the tumor was monitored regularly, and the results are depicted in Figure 6, parts A (for 11a) and B (for 12). Measurements of the tumor volumes were ceased when four animals (50%) in the test group had died. The in-vivo anticancer effects for compound 11a were apparent from the results given in Figure 6A. For the control group, four of the mice had died by day 21; however, the fourth mice in the 11a treated group died on day 34, indicating an extension in the life span in the treated group of mice. Moreover, on day 21, the size of the tumor in the treated animals was barely palpable, compared to the large tumors (average 4500 mm3) already visible for the control animals. Furthermore, it is worthy to mention that compound 11a was signifi-
cantly more active than cyclophosphamide, a clinically useful agent, under the conditions employed. As part of this study, the weights of the animals were also monitored regularly. It was found that at the doses administered, compound 11a and cyclophosphamide did not cause any significant weight loss following drug injection. The in vivo results for compound 12 (Figure 6B), at a dose of 15 mg/kg, gave weak anticancer activity, compared to the untreated control mice. However, at a higher dose of 20 mg/kg, a significant reduction in the tumor growth was observed. However, there was no significant extension in the life span of the treated animals. At the 15 mg/kg dose, there was no weight loss observed for the animals following drug injections. However, at 20 mg/kg, some weight loss was observed immediately after the injection (data not shown), but the animals regained their weight within 2-3 days.
Anticancer Analogues of the Duocarmycins
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 11 3909
Figure 5. Flow cytometry analysis of P815 murine mastocytoma cells, stained with propidium iodide: (A) control untreated cells, (B) the cells were treated with compound 11a for 24 h at its IC50 concentration, and (C) the cells were treated with compound 11a for 24 h at 10 times its IC50 value.
Figure 6. (A) In vivo anticancer studies of compound 11a against B16-F0 murine melanoma grown in the flank of C57BL/6 female mice. The mice were treated intraperitoneally. Diamonds represent the control untreated mice, squares represent cyclophosphamide-treated mice at 30 mg/kg. Triangles represent mice treated with compound 11a at a dose of 15 mg/kg. Measurements of tumor size were stopped when half of the animals in each cohort of eight mice died. (B) In vivo anticancer studies of compound 12 against B16-F0 murine melanoma grown in the flank of C57BL/6 female mice. The mice were treated intraperitoneally. Diamonds represent the control untreated mice, and squares represent mice treated with 12 at a dose of 15 mg/kg. Triangles represent mice treated with compound 12 at a dose of 20 mg/kg. Measurements of tumor size were stopped when half of the animals in each cohort of eight mice died.
In Vivo Tumor Xenograft Studies of 11a.32 This in vivo tumor model assay conducted by the National Cancer Institute involved subcutaneous implantation of the advanced stage human SC-OVCAR-3 ovarian cancer fragments (30 mg) into the axillary region of pathogenfree scid mice on experimental day 0. Compound 11a was dissolved in Tween 80 and saline and was administered via an ip route on a q4d × 3 schedule. Drug treatments were initiated on day 6 for at a dose of 11.20 mg/kg. The volume injected was adjusted according to a ratio of 0.01 mL for every 10 g of body weight. Tumor size and body weights were obtained approximately two times per week. From the results, it was apparent that compound 11a was able to inhibit the growth of the human ovarian cancer xenograft. The %T/C value, expressed as the ratio of the change in median tumor weight for treated and control animals, was recorded as 24% on day 33. The net log cell kill was 0.30, indicating that compound 11a was capable of killing the tumor cells. The mean time for doubling of the tumor weight was 25.3 days, compared to 13.0 days for the control untreated animals. Furthermore, the growth delay %T - C/C was 95%, where T and C are the
median times to doubling of the tumor for the treated and control groups, respectively. According to the NCI’s evaluation criteria, compounds that produce %T/C values of >0 and
