|Home | About | Journals | Submit | Contact Us | Français|
Acid catalyzed Friedlander reactions of a number of 2,3-dihydro-1H-cyclopenta[b]quinoxaline-1-ones with 2-aminobenzaldehyde yield, unexpectedly, 8H-indolo[3,2-a]phenazine and quinolino[2,3-c]cyclopentadienone[2,3-b]quinoxalines, the structures of derivatives of which were confirmed by X-ray crystallography. Easy routes to novel quinoxaline-based indoles, quinolones, and quinoxaline-1,4-dioxides are reported and proposed mechanisms for the unexpected products are discussed.
Indoloquinoxalines are well known heterocycles generally prepared through the reaction of o-phenylenediamine with isatin.1 In contrast, quinoxaline-based quinolines and quinoxaline-1,4-dioxides are either rare or unknown in the chemical literature.2 Recent reviews of the biological activities of indoles, quinolines and quinoxalines point out their medicinal chemistry importance.3 Indeed, quinoxalines and their 1,4-dioxides have shown activity as anticancer, antifungal, antibacterial and anti-inflammatory agents.4 The biological effects of quinoxalines as inhibitors of platelet-derived growth factor receptor tyrosine kinase [PDGF-RTK] is of particular interest and several were reported to be superior in this regard to quinolines and indoles.5
The objective of the work reported here was to develop synthetic routes to novel quinoxaline-based indoles, quinolines and quinoxaline-1,4-dioxides. We envisaged that introduction of a ketomethylene (–COCH2–) moiety at C2 of a quinoxaline would access a number of hetero-cycles ligated to the quinoxaline ring system. A series of steps involving a Beirut reaction6a (1 → 2-4) followed by Polonovski rearrangement6 (2-4 → 5-7), base hydrolysis (5/7 → 8/9), and HOAc-mediated oxidation/reduction (8/9 → 13/14) gives the expected 1H-cyclopolymethylene[b]quinoxaline-1-one starting materials for this study. In contrast, 6 leads to 12, which undergoes two consecutive base-mediated tautomerizations followed by a N1,N4-dehydration and air oxidation of the resulting 5,10-dihydrophenazine intermediate to yield 1-hydroxyphenazine (15a; Scheme 1).
Somewhat surprisingly, saponification of hydroxy-quinoxaline 5 or 7 leads unexpectedly to quinoxaline N-oxide 10 or 11, respectively, in modest yields (10, 27%; 11, 24%). While the exact details of how these unique dehydroxylation reactions proceed are not clear, we speculate that methanol plays a pivotal role because this reaction does not take place in t-butyl alcohol/KOH. We propose a methoxide-based mechanism for this reaction (Scheme 2), but a radical mechanism cannot be excluded. This speculation is analogous, in its initial steps, to the recent mechanism proposed by Bjorsvik et al. for the deoxygenation of N-heteroarene N-oxides.7
Having secured quinoxalino ketone 13, we attempted the Friedlander reaction6a of it with 2-aminobenzaldehyde under acid catalysis; 13 decomposes under basic conditions. We were surprised to find that the expected quinolinoquinoxaline 16 (yellow) was formed (Scheme 3), but in only trace amounts, and that the major products were indolophenazine 17 (20%, yellow) and quinolino-quinoxalinone 18 (50%, orange). The 1H-NMR of 17 indicated that a rearrangement had taken place as the spectrum showed two sharp doublets at 8.00 and 8.16 ppm, which evidenced only 1,2-splitting and, therefore, independence from other neighboring protons. Initial mechanistic consideration suggested that the structure of 17 might be indolophenazine 19; however, X-ray crystallography (Figure 1) left no doubt that this indolophenazine has structure 17. In addition to indolophenazine 17, quinolinoquinoxaline 18 was also formed as confirmed by X-ray crystallography (Figure 2).
Mechanistic steps to explain the formation of these unexpected products (17 and 18) is presented in Scheme 4. We suggest that both arise from common intermediate , which is formed from 13 through two consecutive 1,3-prototropic shifts – the second of which is induced by conjugation with the carbonyl group. The relative higher yield of quinolinoquinoxaline 18 (50% vs. 20% for 17) can be rationalized on the basis that intermediate  undergoes oxidative (air) aromatization of the dihydroquinoxaline moiety to produce the highly reactive cylcopetadieneone-like intermediate . This intermediate does a 1,4-addition of 2-aminobenzaldehyde followed by aldol condensation, dehydration, and air-oxidation to give quinolinoquinoxaline 18. A competing 1,2-addition of 2-aminobenzaldehyde onto the carbonyl group of intermediate  followed by aldol condensation and a rearrangement eventually leads to indolophenazine 17.
This unique rearrangement in a Friedlander-like reaction is unprecedented, constituting a simple and direct method for the synthesis of the novel parent ring system as well as substituted indolophenazines 17a/b from keto quinoxaline 13a/b. Moreover, the second products of this reaction, quinolinoquinoxalines 18a/b, are also previously unknown heterocycles.
As outlined in Scheme 5, we next turned to the synthesis of novel quinoxaline-based indolo-, quinolino- and quinoxalino-1,4-dioxides from 14. It should be added that although quinoxaline 14 can be synthesized from 9 (77% yield; Scheme 1), it can also be prepared directly from 7 under base catalysis (KOH/MeOH) in 40% yield. Reaction of quinoxaline 14 with phenylhydrazine (Fischer indole reaction6a), 2-aminobenzaldehyde (Friedlander reaction6a), and benzofuran oxide (Beirut reaction6a) yielded 12,13,14-trihydroindolo[2,3-c]cyclo-hepta[2,3-b]quinoxaline (22, 74%), 12,14,15-trihydroquinolino[2,3-c]-cyclohepta[2,3-b]quinoxline (23, 68%), and 13,14,15-trihydroquinoxalino[2,3-c]cyclohepta[2,3-b] quinoxaline 1,6-dioxide (24, 65%), respectively, in good yields.
In conclusion, we have synthesized a variety of novel heterocycles, which include: indolophenazines, quinolinoquinoxalinones, indoloquinoxalines, quinolinoquinoxalines, and quinoxalinoquinoxaline 1,4-dioxides. The work presented here also underscores how C-ring size in Polonovski rearrangement products dramatically affects the outcome of their subsequent base-mediated hydrolysis reactions (5/7 → 8/9 vs. 6 →  → 15a; Scheme 1). A similar C-ring dichotomy is noted in the Friedlander reactions of 13 (→ 17/18; Scheme 3) vs. 14 (→ 23; Scheme 5).
We thank the National Science Foundation (Grant 0840444) for the Dual source X-ray diffractometer and the National Science Foundation (CHE-0910870) and the National Institutes of Health (GM089153) for generous financial support of this work.
Supporting Information Available: Full experimental details and characterization data (1H-NMR, 13C-NMR, IR, and LC/MS) for all new compounds. This material is available free of charge via the Internet at http://pubs.acs.org.