An Investigation of a Palladium Catalysed Biaryl Synthesis of Pyrrolophenanthridine Derivatives . Extension of the Heck Reaction

The pyrrolo[3,2,1-de]phenanthridine skeleton has been found in a large number of natural products from the plants of almost all of the genera of the family Amaryllidaceae. Examples of these compounds include assoanine1 (1), oxoassoanine2 (2), hippadine3 (3) and ungeremine4 (4) amongst others5. These compounds (Figure 1) are of interest as synthetic targets not only for the development of synthetic methodology but also for their biological activities6. An important feature of the pyrrolo[3,2,1-de]phenanthridine skeleton is the biaryl linkage and this has often been a central subject of the synthetic methodology employed for the construction of these, and a diverse range of phenanthridine or benzophenanthridine derivatives7. The Heck reaction8 is a versatile methodology for the formation of new C-C bonds, particularly for the coupling of an aromatic halide with an olefin9. The corresponding biaryl synthesis where hydrogen halide is eliminated from the two aromatic components is a recent extension of the Heck reaction10, whereas the corresponding biaryl synthesis using an oxidative addition reaction with stoichiometric quantities of palladium salts has been more extensively studied11.


Introduction
The pyrrolo[3,2,1-de]phenanthridine skeleton has been found in a large number of natural products from the plants of almost all of the genera of the family Amaryllidaceae.Examples of these compounds include assoanine 1 (1), oxoassoanine 2 (2), hippadine 3 (3) and ungeremine 4 (4) amongst others 5 .These compounds (Figure 1) are of interest as synthetic targets not only for the development of synthetic methodology but also for their biological activities 6 .An important feature of the pyrrolo[3,2,1-de]phenanthridine skeleton is the biaryl linkage and this has often been a central subject of the synthetic methodology employed for the construction of these, and a diverse range of phenanthridine or benzophenanthridine derivatives 7 .
The Heck reaction 8 is a versatile methodology for the formation of new C-C bonds, particularly for the coupling of an aromatic halide with an olefin 9 .The corresponding biaryl synthesis where hydrogen halide is eliminated from the two aromatic components is a recent extension of the Heck reaction 10 , whereas the corresponding biaryl synthesis using an oxidative addition reaction with stoichiometric quantities of palladium salts has been more extensively studied 11 .

Results and Discussion
In a general approach for the construction of pyrrolo-phenanthridine alkaloids an appropriately substituted N-benzylisatin derivative (5), that contained the entire carbon framework, was envisaged as a suitable precursor for the obtention of these natural products and their analogs (Scheme 1).
Subsequent reduction of the cyclised isatin derivatives to indoles would yield, for example, dehydroanhydrolycorine 12 (6).Successful cyclization would also furnish substrates to further on going investigations of the use of BH 3 .THF for the reduction of isatin derivatives to indoles 13 .Strategically, oxidation of the derivatives of compound 6 could present problems due to oxidation of the indolic double bond.However, Gribble reduction 14 of the indolic double bond and subsequent oxidation of the indoline would seem to be a viable route for the synthesis of, for example, anhydrolycorin-7-one (7) and hippadine (3), Scheme 1.

a b c
The appropriate N-benzylisatin derivatives 5 for investigation of the synthesis of the pyrrolphenanthridines were readily prepared by benzylation of substituted isatin derivatives 15 8 using basic conditions (Scheme 2).
Initial experiments to determine conditions for the formation of the biaryl linkage focused upon the attempted cyclisations of 7-bromo-and 7-iodo-5-methyl-Nbenzylisatin (5a and 5b) using 10 mol% Pd(OAc) 2 in the presence of 20 mol% of triphenylphosphine in acetonitrile 16 .Substantial substrate consumption was observed but no defined product could be isolated upon workup.It was therefore questioned as to whether the cis dicarbonyl system could possibly interfere in the coupling reaction.An inspection of the isatin literature revealed a study by Martinez and Naarmann that reported the Suzuki couplings of a number of heteroarylboronates with 5-bromo-and 5iodo-isatin using tetrakis-triphenylphosphine palladium as the catalyst, albeit in very modest yields 17 .A plausible explanation for the modest yields obtained by Martinez and Naarmann and the poor mass balance in our reactions, where the substrate was recovered in 52-62% yield, was encountered in the work of Lathourakis and Litinas 18 , where these authors have shown that PPh 3 reacts with isatin to yield a phosphoranylidene that can undergo Wittig reactions.Therefore, to eliminate this source of interference the ketone was transformed into the corresponding ketal (9a-i) using conventional methodology (Scheme 2).
Submission of 9 (a and b) or N-benzyl-5-methyl-7-iodo-2-oxindole, obtained by reduction of 5b with hydrazine hydrate 19 , to the same reaction conditions 16 as used in the first experiments also yielded no defined products but the substrates were recovered in 87-91% yield.The reaction conditions 16 were subsequently modified by replacing acetonitrile with DMF to allow an increased reaction temperature.When 9b was submitted to the new conditions, the formation of defined products could be observed upon TLC analysis. 1H NMR analysis of the crude product after workup of the reaction revealed the conversion of the reaction to be low, 29% (96% mass balance), and the major product was identified as the reduced N-benzyl-5-methylisatin ethylene ketal (25%) from comparison of spectral data with an authentic sample 20 .The structure of the minor component (4%) was suspected to be the desired cyclised product 10a; this suspicion being latter confirmed, vide infra.Subsequent investigations of the reaction conditions were based upon experimental work first reported by Jeffery 21 .Similarly to the initial experiments using the substrates 5a and 5b under Heck conditions, substrate consumption was observed but no defined products could be detected by TLC or upon workup of the reaction.Attention was then focused upon the ketal substrate 9a, which in the presence of 10% Pd(OAc) 2 (0.1mol L -1 DMF), 1 mole equivalent of Bu 4 NBr, and 5.5 mole equivalents of KOAc at 100 o C (bath temp.)resulted in the formation of a single product at 56% conversion of the substrate, Table 1, entry 1 (95% mass balance).This product had the same NMR features as the minor product observed in the previous experiment using this substrate, vide supra.In a similar experiment where PdCl 2 was used as the catalyst (table entry 2) a comparable result was obtained with a 98% mass balance.In contrast to the bromide 9a, the iodide 9b readily furnished the same product as the previous experiments.Isolation and characterisation of this product identified it as the oxopyrrolophenanthridine 10a.Investigation of the catalyst loading revealed a reaction time dependance upon the reaction (table entries 4-7).The ortho-iodobenzyl derivative 9c cyclised under similar conditions to yield the parent ring system 10b (table entry 8).Further evidence for halogen positional versatility can be seen in table entries 9 and 10 where the oxopyrrolophenanthridine derivative 10c was obtained in 77-78% yield from either substrate 9d or 9e.The iodides were found to be superior to the bromides giving better yields and shorter reaction times (compare table entries 1 and 2 with 5, 10 with 11, and 12 with 13).The presence of a nitro group either on the halogenated aromatic Scheme 1. Retrosynthetic analysis for the synthesis of pyrrolophenanthridine derivatives.ring or the aromatic ring to be attacked by the intermediate organopalladium species significantly accelerated the reaction giving increased yields in reduced reaction times (compare table entries 1, 2 and 3 with 9 and 10).Little difference in reaction time or yield was observed in the cyclisations of the unsubstituted 9c and the methylenedioxy substituted 9h (table entries 8 and 13).However, the dimethoxy substituted 9i required a signifcantly longer reaction time to attain an almost equally good yield (table entry 14).
In all successful reactions initiation of the reaction was evidenced by a darkening of the reaction mixture which normally had an onset temperature greater than 80 o C (bath temp.).Both aromatic iodides and bromides have been cyclised successfully in the absence of phosphine ligands, though the experiments revealed that the iodides are better substrates 22 .No special precautions are required for performing these cyclisations which have been routinely performed exposed to the laboratory atmosphere.
The reaction offers a unique opportunity to investigate structural requirements and substituent effects as it has been found to be tolerant to both activation and deactivation of the aromatic nuclei and therefore, mechanistically relevant information can be delineated.These studies as well as the application of this methodology to the synthesis of a number of natural products are currently being investigated.

Experimental
General 1 H and 13 C NMR spectra were recorded using 300 MHz and 200 MHz ( 1 H) Bruker spectrometers and are referenced to TMS.Coupling constants are quoted in Hz.Mass spectra were recorded at NPPN (Nucleo de Pesquisa de Produtos Naturais, Universidade Federal do Rio de Janeiro) using a Hewlett-Packard GC-MS or by direct insertion using a VG Autospec.Infra-red spectra were recorded using a Perkin Elmer 1600 FT-IR (error ± 4cm -1 ) as KBr discs.Melting points were recorded on a hotstage microscope and/or using a MelTemp II capillary apparatus and are reported as uncorrected values.DMF was used as supplied without further treatment.

Table 1 .
Palladium catalysed cyclization of N-benzylisatin derivatives 10e a Z = spiro-2-(1,3-dioxolane)ring; b Reaction conditions used: 10 mole% Pd(OAc) 2 , 1.1 equiv.Bu 4 NBr, 5.5 equiv.KOAc, DMF (substrate concentration 0.1mol L -1 ), 100 o C. Modifications to this procedure are indicated within the square brackets, [modification].All reactions were accompanied by silica gel TLC until substrate was no longer evident or until the reaction appeared to no longer progress; c Isolated yield of purified product.Values in parenthesis are conversions to cyclised product based upon isolation of the reaction product(s) and substrate as calculated from analysis of 1 H NMR spectra.