J. Med. Chem. 1997, 40, 2335-2346
2335
2-Cyclopropylindoloquinones and Their Analogues as Bioreductively Activated Antitumor Agents: Structure-Activity in Vitro and Efficacy in Vivo Matthew A. Naylor,*,† Mohammed Jaffar,‡ John Nolan,§ Miriam A. Stephens,§ Susan Butler,§ Kantilal B. Patel,† Steven A. Everett,† Gerald E. Adams,† and Ian J. Stratford‡ Gray Laboratory Cancer Research Trust, P.O. Box 100, Mount Vernon Hospital, Northwood, Middlesex, HA6 2JR, United Kingdom, Department of Pharmacy, University of Manchester, Oxford Road, Manchester M13 9PL, United Kingdom, Departments of Medicinal Chemistry and Experimental Cancer Therapy, Division of Experimental Oncology, Medical Research Council, Harwell, Oxon OX11 ORD, United Kingdom Received December 11, 1996X
A series of 2-cycloalkyl- and 2-alkyl-3-(hydroxymethyl)-1-methylindoloquinones and corresponding carbamates have been synthesized and substituted in the 5-position with a variety of substituted and unsubstituted aziridines. Cytotoxicity against hypoxic cells in vitro was dependent upon the presence of a 5-aziridinyl or a substituted aziridinyl substituent for 3-hydroxymethyl analogues. The activity of 5-methoxy derivatives was dependent upon the presence of a 3-(carbamoyloxy)methyl substituent. Increasing the steric bulk at the 2-position reduced the compounds’ effectiveness against hypoxic cells. A 2-cyclopropyl substituent was up to 2 orders of magnitude more effective than a 2-isopropyl substituent, suggesting possible radical ring-opening reactions contributing to toxicity. Nonfused 2-cyclopropylmitosenes were more effective than related fused cyclopropamitosenes reported previously. The reduction potentials of the quinone/semiquinone one-electron couples were in the range -286 to -380 mV. The semiquinone radicals reacted with oxygen with rate constants 2-8 × 108 dm3 mol-1 s-1. The involvement of the two-electron reduced hydroquinone in the mediation of cytotoxicity is implicated. The most effective compounds in vitro were the 2-cyclopropyl and 5-(2methylaziridinyl) derivatives, and of these, 5-(aziridin-1-yl)-2-cyclopropyl-3-(hydroxymethyl)1-methylindole-4,7-dione (21) and 3-(hydroxymethyl)-5-(2-methylaziridin-1-yl)-1,2-dimethylindole4,7-dione (54) were evaluated in vivo. Both compounds showed antitumor activity both as single agents and in combination with radiation, with some substantial improvements over EO9 (3) at maximum tolerated doses and as single agents against the RIF-1 tumor model and comparable efficacy in the KHT tumor model. Introduction Fused cyclopropamitosenes and closely related indoloquinones have recently been evaluated as novel bioreductive anticancer agents targeted toward both solid tumors with defined hypoxic fractions and tumor tissues that may be rich in the required activating enzymes.1-3 The reduction of quinones bearing appropriate leaving groups to generate alkylating species, originally termed bioreductive alkylation,4 is now well established, although the novel cyclopropamitosenes were originally designed as analogues of the prototype quinone bioreductive alkylating agent mitomycin c (MMC, 1, Figure 1).5 These compounds were expected to exhibit much reduced electophilicty at C-1 due to the inertness of the 1,2-cyclopropane compared to the aziridine in MMC. Certain indoloquinones (e.g. 2, Figure 1) in this series were found to be highly potent cytotoxins compared with MMC and in some cases with substantially higher hypoxic cytotoxicty ratios (HCR).1-3 The cyclopropamitosenes were found to be more rapidly reduced than MMC by DT-diaphorase (NAD(P)H: (quinone acceptor) oxidoreductase (EC 1.6.99.2)),3 an important activator of mitosenes and an enzyme hyperexpressed in some tumors.6-8 However, this could not explain fully the greater potency (aerobic and hypoxic) †
Gray Laboratory. University of Manchester. Medical Research Council. X Abstract published in Advance ACS Abstracts, June 15, 1997. ‡ §
S0022-2623(96)00842-4 CCC: $14.00
Figure 1. Structures of known lead compounds.
of the compounds compared to the current lead clinical agent of this general type EO9 (3, Figure 1),9 which is reduced 2 orders of magnitude more rapidly than 2 by DT-diaphorase.3,10 The nature of the 5-substituent (in particular aziridines and substituted aziridines) and of the potential leaving group on the 3-methylene substituent was therefore identified as an important feature both in terms of oxic and hypoxic potency and ability to act as substrates for reductase enzymes. The reduction potential does not vary greatly among 5-aziridinyl and 5-methoxy derivatives.2,11 However, the relative rates of reduction of this type of compound by one-electron reductases, which will be important under hypoxic conditions, is unknown. This may be related to the oneelectron reduction potential of the quinone/semiquinone couple (E(Q/ Q•-)), but there are little published data on indoloquinones in aqueous solution. Such data would also be an indicator of the redox-cycling propensities of such compounds and thus be related to aerobic toxicity © 1997 American Chemical Society
2336
Journal of Medicinal Chemistry, 1997, Vol. 40, No. 15
mediated through reactive oxygen species as well as a measure of ease of reduction by one-electron reducing enzymes such as NADPH:cytochrome P-450 reductase and xanthine oxidase. Intramolecular ionic ring opening of the fused cyclopropane was considered unlikely,3 but radical ring opening of the cyclopropane following one-electron reduction to give a reactive H atom abstractor has been suggested as a possible explanation for hypoxic potency, although there was no direct evidence for this mechanism.3 We have therefore designed and synthesized the analogues of 2 in which the cyclopropane rings are not fused with the indoloquinone ring system and for which isopropyl analogues could be synthesized that are structurally very closely related but unable to undergo radical ring-opening reactions. These derivatives were also designed as closer analogues of EO9 (3), initially retaining a hydroxymethyl substituent at the 3-position together with an alkyl substituent at the 2-position, to give a series of compounds which have remained unevaluated to date. We have sought to obtain further structure-activity data for this series of bioreductively activated drugs as hypoxic cell cytotoxins with the aim of optimizing the hypoxic-cytotoxicity ratios and achieving activity in vivo, hitherto unseen with the fused cyclopropamitosenes such as 2. The relationship of these biological effects to the redox properties of the drugs was also studied.
Naylor et al.
Scheme 1. Synthesis of 2-Cycloalkyl- and 2-Alkylindoloquinonesa
Synthetic Chemistry 2-Alkyl- and 2-cycloalkyl-substituted indoles were synthesized in 14 steps from the common precursor 3-chlorophenol as shown in Scheme 1. The crucial step was the 1,5-electrocyclization of the imine (e.g. 8) to the 2,3-dihydroindole derivatives, which was successfully carried out using isobutyraldehyde or cycloalkane carboxaldeydes, but not with acrolein in a proposed alternative route, using zinc acetate in methanol as has been employed in the synthesis of 3 via a 2-acrylate derivative.9 Nitration at the desired 4-position could be achieved only subsequent to the N-methylation step in order to avoid increasing yields of the 6-nitro isomer. The subsequent six steps, including nitration, oxidations (DDQ and Fremy’s salt), and reductions (Sn/HCl, Na2S2O4, and DIBAL-H or LiAlH4), left the 2-cyclopropyl moiety intact, and the desired indoloquinones were obtained. Substitution of the 5-methoxy substituent was successful in high-yielding reactions with aziridine and 2-methylaziridine. Comparable 1,2-dimethyl analogues 52-56 were synthesized from commercially available 2-methyl-5-methoxyindole in seven steps (Scheme 2). Substitution of the 5-position with aziridine and 2-methylaziridine was again successful in high yield, as was the substitution with cis-2,3-dimethylaziridine. However, 2,2-dimethylaziridine reacted with difficulty, and the resulting 5-(2,2-dimethylaziridinyl) analogue was unstable in aqueous solution and on silica gel, ring opening via an SN1 mechanism to give 57 (Scheme 2). The 2-unsubstituted analogues 62-66 were obtained in six steps from 5-methoxyindole-3-carboxaldehyde (Scheme 3). Results and Discussion
a Reagents: (i) H SO /NaNO /H O/C H N; (ii) K Fe(CN) /KOH/ 2 4 2 2 5 5 3 6 H2O; (iii) NaH/THF/(MeO)2SO2; (iv) NCCH2CO2Et; (v) HCl/EtOH/ H2O; (vi) H2/PtO2/PhMe/EtOH; (vii) RCHO/MeOH; (viii) Zn(OAc)2/ MeOH; (ix) Ac2O; (x) KOH/EtOH/H2O/0 °C; (xi) (MeO)2SO2/DMF/ K2CO3; (xii) DDQ/PhMe/reflux; (xiii) 4%KOH/MeOH; (xiv) NaH/ DMF/MeI/60 °C; (xv) fuming HNO3/AcOH/4 °C; (xvi) Sn/HCl/ EtOH/H2O; (xvii) Fremy’s salt/Me2CO/NaH2PO4/Na2HPO4/pH 6.0; (xviii) 1H-aziridine; (xix) Na2S2O4/H2O/EtOH/CHCl3 then LiAlH4/ THF/30 °C (DIBAL-H/PhMe/-30 °C then 0 °C for 20) then FeCl3/
In previous studies the presence of a hydroxymethyl substituent at the 3-position of indoloquinone anticancer
(xxii) NH3/CH2Cl2/-78 °C; (xxiii) R1-CHCH2NH.
HCl/H2O/0 °C; (xx) R1-CHCH2NH; (xxi) PhCO2Cl/C5H5N/0 °C;
Bioreductively Activated Antitumor Agents
Journal of Medicinal Chemistry, 1997, Vol. 40, No. 15 2337
Scheme 2. Synthesis of 2-Methyl Analoguesa
Scheme 4. Potential Mechanisms of Toxicity for 2-Cyclopropyl- and 2-Alkyl-Substituted Indoloquinones
a Reagents: (i) NaH/DMF/MeI/60 °C; (ii) HCON(Me)Ph/POCl 3 then NaOAc/H2O; (iii) fuming HNO3/AcOH/4 °C; (iv) Sn/HCl/H2O/ EtOH; (v) Fremy’s salt/Me2CO/NaH2PO4/Na2HPO4/pH 6.0; (vi)
NaBH4/MeOH/Ar then air; (vii) R(R1)CCH(R2)NH; (viii) H2O/45 °C.
Scheme 3a
a Reagents: (i) NaH/DMF/MeI; (ii) concentrated HNO /AcOH/0 3 °C; (iii) Sn/HCl/H2O/EtOH; (iv) Fremy’s salt/Me2CO/NaH2PO4/
Na2HPO4/pH 6.0; (v) NaBH4/MeOH/Ar; (vi) R-CHCH2NH; (vii) PhCO2Cl/C5H5N/0 °C; (viii) 1H-aziridine.
drugs has been thought to give inactive drugs, with the majority of compounds evaluated consequently possessing carbamate or acetate derived leaving groups.12-14 The success of the mitosenediol EO9 (3) and earlier in vivo activity of related compounds, however,7,9,14-16 suggests the importance of this class of compound, the structure of which has yet to be optimized. The in vitro biological data presented in this study (Table 1) add credence to this suggestion, indeed substitution with the carbamate moiety has in many cases actually reduced
potency (see compounds 21 and 25, 38 and 42) while in other cases potency shows the expected increase (e.g. compounds 20 and 24, 22 and 26, 61 and 66). Exceptional compounds are always 5-aziridinyl derivatives, suggesting that the dominance of aziridine-mediated toxicity may outweigh effects due to the nature of the potential leaving group at the 3-position in many cases. 5-Methoxy derivatives generally showed improved potency and HCR when carbamate replaced hydroxy on the 3-methylene substituent, indicating that the nature of this potential leaving group1-3,12-14 can dominate the bioreductive properties of non-aziridinyl analogues. That the 3-hydroxymethyl substituent is required for hypoxia-selectivity is indicated when comparing the 5-aziridinyl-3-methyl derivative 44 (HCR ) 1.77) with the corresponding 3-hydroxymethyl analogue 37 (HCR ) 32.5). There is a clear trend of increasing aerobic and hypoxic potency on reducing the steric bulk of substituent R2 (see series of compounds 21, 37, 46, and 53 for example). Thus, the least potent 5-aziridinyl analogue tested (both hypoxic and oxic) was the 2-cyclohexyl derivative 46 while the most potent compound tested (both hypoxic and oxic) was 66, which has both aziridine and carbamate functionalities and no 2-alkyl substituent. The low HCR for this latter compound, however, demonstrates the clear need for these structure-activity studies to obtain both optimal potency and optimal hypoxia-selectivity. Certain aziridinyl 2-cyclopropyl analogues do not fit into this pattern and also show much increased potency, particularly under hypoxic conditions, compared to corresponding isopropyl derivatives (e.g. compare 21 with 37 and 25 with 41). Reactions of the semiquinone radical can dominate under severe hypoxia following one-electron reduction, and these data provide some indirect, though compelling, evidence for a contribution from semiquinone induced radical ring opening of the cyclopropane under such hypoxic conditions, possibly leading to further cytotoxic reactions. The full elucidation of the mechanism of such toxicity, hypothesized in Scheme 4, clearly requires further study.
2338
Journal of Medicinal Chemistry, 1997, Vol. 40, No. 15
Naylor et al.
Table 1. Structures of Compounds and in Vitro Biological Data Comparing Differential Aerobic/Hypoxic Cytotoxicities
compd 1 2 3 4 5 18 20 24 19 21 25 22 26 35 36 37 38 40 41 42 43 44 45 46 52 53 54 55 57 62 63 64 65 66 a
type MMC A EO9 A A B B B B B B B B B B B B B B B B B B B B B B B B B B B B B
R
R1
R2
C50 (air), µM
C50 (N2), µM
0.8b Azc
CH2OCONH2
MeO Az MeO MeO MeO Az Az Az 2-Me-Az 2-Me-Az MeO MeO Az Az MeO Az 2-Me-Az MeO Az MeO Az MeO Az 2-Me-Az 2,3-Me2-Az Me2C(OH)CH2NHMeO Az 2-Me-Az MeO Az
CH2OH CH2OH CO2Me CH2OH CH2OCONH2 CO2Me CH2OH CH2OCONH2 CH2OH CH2OCONH2 CO2Me CH2OH CH2OH CH2OH CH2OCONH2 CH2OCONH2 CH2OCONH2 Me Me CH2OH CH2OH CH2OH CH2OH CH2OH CH2OH CH2OH CH2OH CH2OH CH2OH CH2OCONH2 CH2OCONH2
c-Pr c-Pr c-Pr c-Pr c-Pr c-Pr c-Pr c-Pr CH(Me)2 CH(Me)2 CH(Me)2 CH(Me)2 CH(Me)2 CH(Me)2 CH(Me)2 CH(Me)2 CH(Me)2 c-Hex c-Hex Me Me Me Me Me H H H H H
0.003d 0.19 ( 0.027 108.4 ( 5.9 0.965 ( 0.11 103.5 ( 5.2 540 ( 52 29 ( 5.6 1.72 ( 0.18 0.603 ( 0.099 3.33 ( 0.75 130 ( 9.8 12.5 ( 2.9 150 ( 15 800 ( 80 25.7 ( 3.0 127.8 ( 13.5 20.9 ( 1.67 47.5 ( 14.6 20.6 ( 2.5 1.236 ( 0.189 0.333 ( 0.014 122.4 ( 17.5 34.5 ( 6.0 1077 ( 44 0.149 ( 0.011 94 ( 7.6 202 ( 11 1260 ( 126 220 ( 20 0.153 ( 0.013 4.42 ( 1.17 3.1 ( 0.24 0.00019 ( 0.000032
HCR ) hypoxic cytotoxicity ratio (C50 (air)/C50 (N2)). b Reference 32. c Az ) aziridin-1-yl.
There is evidence in the present results for clear advantages in terms of both hypoxia-selectivity and hypoxic potency to compounds with a 2-cyclopropyl substituent rather than a 1,2-fused cyclopropane system (compare compounds 5 and 21). This effect seems to be lessened when more potent leaving groups are present such as in carbamate 24, which is comparable in its potencies to its fused analogue2 (compare 2 with 25). An increase in HCR upon alkyl substitution of the aziridinyl moiety has been demonstrated in this study with some (compare 41 and 42, 53 and 54, 63 and 64) but not all types of compound (compare 21 and 22, 25 and 26). Significantly, the compounds which do not fall into this pattern are again the 2-cyclopropane derivatives, where toxicity is not dominated by the reactivity of the aziridine moiety to the same degree as other 2-alkyl derivatives, possibly due to a contribution from the cyclopropane ring (e.g. Scheme 4). Generally, a single methyl substituent on the 5-aziridine ring is optimal. With multiple substitution, the potency and HCR is much reduced due to progressive deactivation of the aziridine counteracting increasing pKa. Redox chemical studies (Table 2) demonstrated that analogues containing an aziridine group exhibited a similar reactivity toward oxygen as EO9 (3)17 (k2 (see Experimental Redox Chemistry Section, eq 2) was generally found to be in the order of 2 × 108 M-1 s-1).
0.4b
d
0.003d 0.0038 ( 0.00057 60.6 ( 7.4 0.073 ( 0.011 59.8 ( 11.4 820 ( 83 0.44 ( 0.023 0.93 ( 0.067 0.0058 ( 0.0013 0.074 ( 0.015 5.6 ( 0.6 0.459 ( 0.096 150 ( 15 200 ( 20 0.79 ( 0.14 13.9 ( 1.5 0.117 ( 0.018 5.46 ( 0.43 0.87 ( 0.215 0.657 ( 0.063 0.188 ( 0.011 20.5 ( 5.3 0.93 ( 0.078 284.8 ( 37.6 0.0116 ( 0.0008 0.5 ( 0.07 14.2 ( 1.7 500 ( 120 240 ( 23 0.0086 ( 0.0009 0.0179 ( 0.02 0.037 ( 0.007 0.00013 ( 0.000025
HCRa 2.0b 1.0d 50.3 1.8 13.2 1.7 0.65 65.9 1.8 103.5 45.0 23.4 27.23 1.0 4.0 32.5 9.2 178.6 8.7 23.7 1.88 1.77 6.0 37.0 3.78 12.8 188 14.2 2.5 0.92 15.4 24.7 83.8 1.46
Reference 2.
However EO79 (the 5-methoxy analogue of EO9 (3)) and other 5-methoxy derivatives selected from this study 20, 24, 40, and 62 were less electron-affinic than corresponding 5-aziridinyl compounds and reacted typically 2-4 times faster with oxygen. Thus 5-aziridinyl semiquinone radicals will be longer-lived than 5-methoxy analogues in the presence of oxygen, which, in addition to reductive aziridine activation, may also contribute to their greater effectiveness compared to 5-methoxyindoloquinones following one-electron reduction. This will also be relevant following two-electron reduction since it will slow down the rate of reoxidation of the hydroquinone in sequential one-electron steps, and influence the lifetime of semiquinones formed through a comproportionation reaction between an indoloquinone and its hydroquinone. The one-electron reduction potentials at pH 8.5 for the aziridinylindoloquinones 53 (-294 mV) and 21 (-286 mV) were similar to that of EO9 (3), where E (Q/ Q•-) ) -265 mV (lit.17 -253 mV). The redox potentials of 5-methoxyindoloquinones were generally 25-50 mV lower than the corresponding 5-aziridinyl analogues. This is consistent with recent half-wave potential reduction data on related indoloquinones2 (e.g. compound 2 Eredox ) -1.360 V and its 5-MeO analogue Eredox ) -1.395 V). These differences in reduction potential were reflected in the corresponding rates of reaction of semiquinone radicals with oxygen (see Table 2). How-
Bioreductively Activated Antitumor Agents
Journal of Medicinal Chemistry, 1997, Vol. 40, No. 15 2339
Table 2. One-Electron Reduction Potentials at pH 8.5 (E Q/Q•-) for Representative Compounds and Rate Constants for Reaction of Semiquinone Radicals with Oxygen
compd
R
R1
R2
K3a
E (Q/Q•-), mV
108k2(Q•- + O2), dm3 mol-1 s-1
EO9(3) 21 53 54 EO7 62 20 40 24
Azb Az Az 2-Me-Az MeO MeO MeO MeO MeO
CH2OH CH2OH CH2OH CH2OH CH2OH CH2OH CH2OH CH2OCONH2 CH2OCONH2
CHdCHCH2OH c-Pr Me Me CHdCHCH2OH H c-Pr CH(CH3)2 c-Pr
39.6 ( 5.7 41.2 ( 1.4 30.1 ( 1.5 19.7 ( 1.7 15.5 ( 1.5 6.8 ( 0.4 6.2 ( 0.2 20.3 ( 0.7 19.8 ( 1.7
-265 ( 5c -286 ( 4c -294 ( 4c -305 ( 4c -309 ( 5c -332 ( 4d -334 ( 4c -380 ( 4d -377 ( 9d
1.7 ( 0.1 2.3 ( 0.1 2.8 ( 0.1 2.4 ( 0.1 4.5 ( 0.1 4.4 ( 0.1 5.1 ( 0.2 6.3 ( 0.1 8.2 ( 0.2
a Means of measurements using five different concentrations of indoloquinones. b Az ) aziridin-1-yl. cPotentials vs E(BV2+/BV•-) ) -374 mV at pH 8.5. d Potentials vs E(MV2+/MV•-) ) -450 mV at pH 8.5 (will be the same at pH 7.4).
ever, 5-methoxy compounds bearing a good 3-methylene leaving group also have exquisite hypoxia-selectivity (e.g. 40 and 65); therefore there appears to be little relationship between reduction potential and HCR as determined by the MTT assay in this study, particularly in view of the low reduction potentials determined for the hypoxia-selective carbamates 24 and 40 compared with EO9 (3). However, it should be noted that under conditions of greater oxygen concentration, reduction potential is likely to have a greater influence on reduced drug reactivity. The trends between k2 and E (Q/Q•-) for the indoloquinones in this study are comparable with those published for quinones of similar redox potential.18 Thus although the indoloquinones should be easily reduced by enzymes such as NADPH-cytochrome P-450 and DT-diaphorase, the high values measured for k2 would indicate that even under modest tumor hypoxia (ca. 10 µmol dm-3 O2) the half-life of the semiquinone radicals {∼0.7/(k3[O2]} will be short (∼0.4 ms). Lead compounds 21 and 54 with HCR values in excess of 100 (2-3-fold higher than was obtained for the lead clinical candidate EO9 (3)) were evaluated in vivo in the RIF-1 and KHT rodent tumor models. Both compounds exhibited substantial cell killing both as single agents, with substantially greater effectiveness than EO9 (3) in both tumor models at drug MTDs (an MTD dose of 21 alone was as effective as a radiation dose of 15 Gy alone in the KHT tumor) and in particular after a single dose (up to 15 Gy) of radiation (up to 3.5 log cell kills against the RIF-1 tumor, Table 3) at drug doses well below MTDs (MTD 21 ) 100 mg kg-1 and MTD 54 ) 90 mg kg-1). A lesser effect in the RIF-1 tumor with EO9 (3) could only be achieved at its MTD. These compounds are therefore now being evaluated against human tumor xenografts. A full study of the in vivo biological effects of these compounds will be reported elsewhere. Since despite the oxygen reactivity of the semiquinone radicals the lead indoloquinone compounds 21 and 54 showed antitumor effects in vivo against hypoxic cell populations, and these effects compare favorably with EO9 (3) (Table 3), little relationship between bioreductive cytotoxicity and the oxygen reactivity of oneelectron-reduced radicals is evident. This provides some evidence that the antitumor bioreductive effect is likely to be mediated principally through the hydroquinone,
Table 3. Comparison of Indoloquinones 21 and 54 with EO9 (3) When Used in Combination with X-rays (20 and 50 mg kg-1) and Alone (MTD) in the RIF-1 and KHT Tumors
tumor control RIF-1 RIF-1 RIF-1 RIF-1 RIF-1 RIF-1 RIF-1 RIF-1 RIF-1 RIF-1 RIF-1 KHT KHT KHT KHT KHT KHT KHT KHT KHT KHT KHT a
X-ray dose, Gy
drug
drug dose, mg kg-1
MTD, mg kg-1
15 15 15 15 15 15 15 15 10 10 10 10 10 10 10 10
21 21 21 54 54 54 EO9 (3) EO9 (3) EO9 (3) EO9 (3)
100 20 50 90 20 50 15 5 10 15
100 100 100 90 90 90 15 15 15 15
21 21 21 54 54 54 EO9 (3) EO9 (3) EO9 (3) EO9 (3)
100 20 50 90 20 50 15 5 10 15
100 100 100 90 90 90 15 15 15 15
mean relative surviving fraction 1.0 2.0 × 10-3 5.0 × 10-3 2.4 × 10-5
