|Home | About | Journals | Submit | Contact Us | Français|
A tricyclic precursor for the synthesis of the prodrugs of pro-1,2,9,9a-tetrahydrocyclopropa[c]benz-[e]indole-4-one tetramethoxyindolecarboxamide (CBI-TMI) was prepared using the ring-closing metathesis approach. The tricyclic intermediate was converted to an advanced precursor of a CBI-TMI prodrug equipped with a linker presumably suitable for activation using the aldolase catalytic antibody 38C2. An attempted 38C2-catalyzed two-step activation of the hydroxy-pro-CBI intermediate involving retro-aldol and the β-elimination reactions was also examined.
Selective delivery of cytotoxins to tumor and cancer-associated cells without affecting the normal cells remains one of the important unmet goals.1 We are developing several alternative approaches, including the chemical programming2 of the catalytic aldolase antibody (Ab) 38C23 using inhibitors of the cancer-associated antigens/receptors, and use of the chemically programmed Ab 38C2 (cpAb or cp38C2)4 and inhibitors to selectively deliver a cytotoxin to tumor site. In one approach, we are using chemically labeled Ab 38C2 (clAb or cl38C2), that possess both the catalytic and the cell-targeting properties for a selective prodrug activation at the tumor site.5 The other counterpart of this approach is a prodrug that should be nontoxic and efficiently activated using Ab 38C2 or the cl38C2 to afford the therapeutically effective cytotoxins for the desired outcome. Unfortunately, all prodrugs available so far lack these criteria, mainly due to a combination of the drugs examined and that the prodrugs could not be activated efficiently, and they cannot be advanced to preclinical studies.6, 7 With new doxorubicin prodrugs, we have made some improvements with respect to their toxicities and the activation rates, but they might also not meet the desired criteria. We argue that the desired prodrugs should (1) have sub- or low nanomolar cytotoxic drug counterpart, (2) be at least 1000–10000 fold less toxic than the drug used, (3) have low or subnanomolar binding to the catalyst, viz. cl38C2, and (4) be activated rapidly. With these hypothesis, we are seeking new prodrugs from a highly potent cytotoxin, such as CBI-TMI (1).8
CBI-TMI (1) is a highly potent synthetic analog9 of the naturally occurring antitumor antibiotic, duocarmycin SA (2)10 (Figure 1). CBI (1,2,9,9a-tetrahydro-cyclopropa[c]-benz[e]indole-4-one) derivatives are both chemically stable, and also synthetically easily accessible. Like the parent molecule, CBI-TMI selectively binds and alkylates duplex DNA at the N-3 position of adenines in the minor groove.11 Figure 1 also shows structures of several seco-CBI-TMI derivatives (3–8), some of which were previously prepared and evaluated.12 These seco-CBI-TMIs are indeed the prodrug form of the corresponding active CBI-TMI drugs, as the latter are produced in situ before reacting with DNA. In fact, most duocarmycin analogs and prodrugs were developed on this principle in that the cyclopropane ring was generated by Winstein cyclization13 of the chloromethyl group or an analogous reactive function and facilitated by the free phenolic hydroxy or amine functional groups in 3–5. Similarly, the isomeric compounds 6–8 could also undergo cyclization reaction affording 1 or its corresponding imine analogs. Because these compounds could cause indiscriminate toxicity to normal cells, we decided to explore a prodrug approach for their delivery.14 In this strategy, a prodrug functionalized with a linker can be selectively activated using tumor-associated proteases (TAPs)15 or a non-endogenous enzyme, including catalytic antibody (Ab).16 Here, we report our preliminary study toward the synthesis of a CBI-TMI prodrug 6a and its activation using Ab 38C2.17
Figure 2 shows structures of the CBI-TMI prodrug, 6a, as well as several designed analogs, 6b, 7c, and 8d, as the viable candidates of the Ab 38C2-catalyzed activation. Like the previously studied prodrugs of doxorubicin, etoposide, camptothecin, and enediynes, these prodrugs also possessed an ‘aldol-Michael’ linker, which on treatment with Ab 38C2 would start the activation by the retro-aldol reaction. The resultant labile intermediates would undergo β-elimination to give 6 from 6a. Other prodrugs would continue subsequent transformation, including the decarboxylation reaction and urea formation, under the physiologic conditions to give 6, 7, or 8, which should serve as the precursor of compound 1 or its imine analogs.
We anticipated that prodrug 6a could be readily prepared by modifying the previously described synthesis of 1 via a tricyclic intermediate, 14a.18 However, we designed an alternative route using the Grubbs ring-closing metathesis (RCM) reaction19 for the synthesis of 14a (Scheme 1). Thus, the required diene precursor 13a for the RCM reaction was prepared starting with the readily available phenol 10.20 The latter was synthesized from a commercial compound, 1,3-dinaphthol, 9, in three steps, including reaction of compound 9 with 4-methoxybenzyl amine (PMB-amine) to give 3(4-methoxybenzyl)amino-1-naphthol, deprotection of the latter compound under acidic conditions using H2SO4 and TFA, and Boc protection of the resulting free amine, 2-amino-1-naphthol, using Boc anhydride, as described by Gieseg, et al. Compound 10 was next protected using benzylbromide and K2CO3 in the presence of tetrabutyl-ammonium iodide, and the resulting product was iodinated using NIS21 to yield the iodonaphthalene derivative 11. The latter product underwent Pd-catalyzed Stille coupling reaction22 with tributylvinylstannane to afford the vinylnaphthalene derivative 12. N-alkylation of compound 12 with allylbromide using NaH in DMF provided the requisite diene 13a for the RCM step that was accomplished using the Grubbs catalyst I affording the tricyclic compound 13a in good yield. Encouraged by this result, we examined whether the above RCM approach could be utilized to prepare the ring expanded analogs of 14a (seven, eight and nine-membered cyclic alkenes) and their corresponding prodrugs. Accordingly, dienes 13b–d were prepared by treatment of 13 with 4-iodo-1-butene, 5-iodo-1-pentene, and 6-iodo-1-hexene, respectively, and then submitted to the RCM reactions using Grubbs catalysts I or II. Somewhat surprisingly, only the dimerization products 15b–d were obtained in good yields under the attempted conditions presumably due to the increased ring strain in seven- to nine-membered tricyclic olefin products.
With the intermediate 14a in hand, synthesis of the desired precursor 25 of the prodrug 6a progressed as shown in Scheme 2, via a coupling of amine 20 with the acid chloride 23. Here, amine 20 was prepared from the intermediate 14a and the acid chloride 23 was obtained from the commercially available aldehyde 21. Thus, intermediate 14a was hydroborated using BH3.SMe2 and the hydroborated product was oxidized with H2O2/NaOH giving alcohol 16. The free hydroxyl group in 16 was protected as TBDPS ether, and the benzyl group was removed by the Pd-catalyzed hydrogenolysis affording intermediate 17. Compound 17 was alkylated with tosylate 18 and cesium carbonate in the presence of 18-Crown-6 to give compound 19. The later product underwent OsO4-catalyzed dihydroxylation and subsequent oxidative cleavage of the resulting diol with Pb(OAc)4 to afford an aldol product. The Boc group in the latter compound was removed using TFA/CH2Cl2 to give a free amine compound 20. Separately, chlorotrimethoxyindolecarboxylate, 23, was prepared by a usual acyl chloride forming reaction of the readily available acid 22 (prepared from aldehyde 21)23 with oxalyl chloride. Next amine 20 was coupled with acid chloride 23 affording amide 24, and the TBDPS protecting group in the latter product was removed using HF.Py giving the desired alcohol precursor, 25, of the prodrug 6a. We expected that compound 25 should be converted to 6a using CCl4 and PPh3, however to our surprise, no such reaction took place.
While we were yet to develop a suitable method for the conversion of 25 to 6a, we examined the 38C2-catalyzed activation of the former as a model to give phenol 27 via a Michael-type adduct, 26. Thus, compound 25 (100 µM) was incubated with a catalytic amount of antibody 38C2 (5 µM) at 37 °C overnight, and then monitored by LCMS analysis. The LCMS showed the formation of the retroaldol intermediate 26 along with the remaining 25 in approximately 3:2 ratio. However, the subsequent β elimination of 26 to give 27 was very slow, which was surprising because the β-ketoalkylether of a phenolic drug was reported to undergo β-elimination reaction to produce free drug under the physiologic conditions. Moreover, antibody 38C2 was also known to catalyze β-elimination reaction of β-alkoxy-ketones and aldehydes to give free alcohol or phenol and the α,β-unsaturated aldehydes or ketones. Whereas there is no clear explanation why intermediate 26 does not undergo β-elimination reaction to produce 27 at pH 7.4 (PBS buffer), the inertness of the former intermediate to antibody 38C2 can be due to the steric hindrance of the substrate. Presumably, the ketone linker in intermediate 26 is very close to the bulky drug molecule, and out of the reach of the reactive lysine residues in the antibody 38C2 binding sites. Therefore the antibody could not catalyze β-elimination reaction in intermediate 26. This result may not be surprising, because a compound bearing a longer linker is activated faster than one coupled to an analogous shorter linker, as found in our recent studies on doxorubicin prodrugs. These observations suggested that new CBI prodrugs equipped with longer or more active linker should be conceived for an effective release of the free drug when antibody 38C2 will be used as a catalyst. In particular, the linker should be conjugated to parent molecule 6 through the carbamate function rather than the ethereal function. Therefore, now we are focusing on new prodrugs of 6–8, such as 6b, 7c and 8d, which are more likely to be activated by Ab 38C2.
In conclusion, we have developed an efficient RCM approach to the synthesis of the CBI skeleton, which can be applied to synthesize a wide variety of CBI analogs and CBI-derived prodrugs. In our approach, the prodrugs would possess appropriate linkers that are susceptible to aldolase antibodies or a tumor-associated protease. The preliminary activation study with the precursor of our CBI-TMI prodrug showed that the ethereal linker was not suitable for an effective release of the drug using Ab 38C2. Further study along this line as well as development of new prodrugs that can be activated using Ab 38C2 or TAPs are in progress and will be described in a due course.
Spectroscopic data and experimental procedures are available.
We thankfully acknowledge the Skaggs Institute for Chemical Biology and National Cancer Institute (R01CA120289) for the financial support.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.