Synthetic studies related to the akuammiline alkaloids

6-Ethyl-2,3-dihydro-1-phenyl-1 H -pyrrolizine 19 and 1-(2-aminophenyl)-6-ethyl-2,3-dihydro-1 H -pyrrolizine 30 were synthesized starting from the pyrroles 4-acetylpyrrol-2-yl phenyl ketone 13 and 4-acetylpyrrol-2-yl 2-(dimethylaminomethylenamino)phenyl ketone 26 , respectively, using vinyltriphenylphosphonium bromide in an intramolecular Wittig reaction for the formation of the second five-membered ring.


Introduction
The variety of indole alkaloid skeleta, 1 of greater or lesser complexity, has been a frequently employed proving ground for novel synthetic methodologies and strategies.Some indole alkaloid types have been the subject of frequent attention -the alkaloid ellipticine 2 probably holds the record for the most frequently synthesised alkaloid ever -the Aspidosperma skeleton 3 has seen Stork's synthesis, an early illustration of enamine β-alkylation, Overman's approach, illustrating his aza-Cope/Mannich sequence, and Magnus' route utilising an indole-2,3-quinodimethane intermediate, amongst others.One of the few groups of monoterpenoid indole alkaloids which has not yet been the subject of a successful total synthesis, is the group characterised by the presence of a C-7-C-16 bond 4 -akuammiline 1 represents this structural type in its simplest form.Approximately 100 indole bases are now known (more than 20 in the last decade) which fall into the akuammiline structural category, and the chemistry and pharmacology of these have been recently reviewed. 5ur personal interest in this group of alkaloids traces back to that period when the Manchester Chemistry Department was priviliged to number amongst its academic staff Arthur Birch and Rod Rickards, and one of the present authors studied Akuamma alkaloids for PhD. 6It is a pleasure to contribute this paper for the issue of Arkivoc which commemorates Professor

Background
The use of plants and plant extracts containing akuammiline alkaloids is widespread in traditional medicine.For example, the aqueous extracts of Picralima nitida (the "akuamma" tree) are used in West Africa as painkillers; 7 in Ghana, pulverised and encapsulated seeds can be found for sale for medicinal purposes in the market.In the rural areas of Saudi Arabia, decoctions of Rhazya stricta are employed as a cure for helminthiasis 8 while in the Philippines the bark of Rauwolfia sumatrana is believed to alleviate the symptoms of malaria. 9rompted by the traditional medicinal use of the seeds of the Picralima nitida tree in the Gold Coast, Henry and Sharp were the first to examine the alkaloidal content. 10These authors adopted the word "akuamma", used by natives in the region to designate the tree, as the prefix for the name given to the major alkaloidal constituent, akuammine 2 (R=HO).Subsequently, Henry isolated, and characterised Ψ-akuammigine 2 (R=H) and akuammiline. 11In 1961, the determination of the X-ray crystal structure of the quaternary salt echitamine 3 12 then allowed the final establishment of the structures and absolute configurations 13 of Ψ-akuammigine, akuammine and akuammiline based on their chemical correlation with echitamine.Studies of extracts of seeds, fruit and bark of Picralima nitida revealed anti-protozoal, antimicrobial and in vitro anti-malarial activity. 14Extracts from Hunteria zeylanica (rich in corymine 4) showed anti-inflammatory activity and stimulation of the CNS, 15 while extracts from Alstonia scholaris (which contain picraline 5 and Ψ-akuammigine amongst other akuammiline alkaloids) showed hepato-protective activity. 16Echitamine chloride has been reported to have antitumour activity coupled with a low toxicity profile. 17It is probably appropriate to apply the term privileged structures, as employed by Evans et al. 18

Synthesis of bicycles 8
Our first plan (Scheme 2) was to close an intermediate of the form 9, via an intramolecular electrophilic attack at the pyrrole α-position (ignoring questions of regioselectivity), to give 10.2-Aminoethyl phenyl ketone hydrochloride 28 was prepared via addition of phthalimide to phenyl vinyl ketone, 29 then hydrolysis with 1:1 AcOH-conc HCl. 30 Reaction of the amine with 2,5dimethoxytetrahydrofuran gave the pyrrole ketone 9a, borohydride reduction of which gave the alcohol 9b.Alcohol 9b was also prepared from 3-hydroxy-3-phenylpropanamine 31 by reaction 32 with 2,5-dimethoxytetrahydrofuran.
Attempts to effect ring closure by conversion of the benzylic alcohol in 9b into a triflate were disappointing.Despite seemingly excellent precedents in which an analogous 5,6-ring system was constructed, 33 the best that could be claimed in our experiments were signals in 1 H NMR spectra, of complex product mixtures, which could have represented some of the desired product.
We explored the possibility of achieving the desired benzylic electrophilic reactivity by decomposition of a trichloroacetimidate; 34,35 the required derivative 9c was easily prepared following a literature precedent, 36 but various attempts to effect the desired acid-catalysed closure were unsuccessful (Scheme 2).

Scheme 2
Katritzky had described 37 the synthesis of bicycles 12a and 12b, the key step, in which the saturated five-membered ring is fused to the pyrrole nucleus, involving intramolecular displacement of tosylate in 2-(benzotriazol-1-ylmethyl)-1-(2-tosyloxyethyl)pyrrole precursors 11a and 11b, subsequent nucleophilic displacement of the benzotriazolyl unit in 12a with phenylmagnesium bromide completing a synthesis of 13a.Although we were able to reproduce the literature report and prepare 13a, in attempting to adapt this sequence for the synthesis of the desired 13b, we were only able to reach 11c, as shown in Scheme 3, failing completely in attempts to bring about ring closure to 12c from this intermediate.We turned to construction of the five-membered ring via an intramolecular Wittig reaction.Linderman and Meyers had shown 38 that addition of pyrrolyl anions to vinyltriphenylphosphonium bromide, directly generates ylides which react with aldehydes to produce substituted 1-allylpyrroles.The process had been earlier carried out in an intramolecular sense starting from 2-formylpyrrole, to generate a five-five bicycle of the type required in the present work, though an attempt to use 2-benzoylpyrrole produced only tars. 39A comparable five-membered ring formation was reported with 2-formylindole.

Scheme 3
In our hands, a preliminary investigation using 2-benzoylpyrrole in reaction with vinyltriphenylphosphonium bromide was encouraging but sounded a warning: there was clear evidence that an initial product 14 was in equilibrium (ca.1:1) with an isomer 15 (Scheme 4).In an attempt to prevent this complication, we turned to the use of a 3-acetylpyrrole, arguing that conjugation with this ketone carbonyl would maintain the first-formed system of double bonds.The acetyl substituent also corresponds to the two-carbon side-chain eventually required in the alkaloid.Of various routes to 4-acetyl-2-benzoylpyrrole 16 we found the best was to start with 3acetylpyrrole, itself available via Friedel-Crafts acetylation of 1-phenylsulfonylpyrrole, then de-N-protection; 40,41  Reaction of 16 with vinyltriphenylphosphonium bromide produced a single, though rather unstable bicyclic alkene 17, which was hydrogenated giving 18, the acetyl group then reduced, producing 19, using 42 t-BuNH 2 BH 3 with AlCl 3 (Scheme 5).That the orientation of double bonds was, as desired, and as shown in 17, 18 and 19, was verified by the 1 H NMR spectra of these substances.For example, in 17 the methylene signal at δ 4.68 was split (J 2.3 Hz) by the adjacent alkene proton and in the spectrum of 18 there was a triplet (J 7.7) at δ 4.40 for the doubly benzylic proton, in addition to multiplet signals for the other four aliphatic protons.
The further development of this successful route, to provide an appropriate bicycle for the alkaloid synthesis, required that the benzenoid ring carry an ortho nitrogen substitutent.We reacted the 2-lithio-1-phenylsulfonylpyrrole with ortho-nitrobenzaldehyde, then immediately oxidised the resulting mixture, but were only able to obtain a 10% yield of ketone 20.However, following work using 1-para-toluenesulfonylpyrrole, 43 Friedel-Crafts ortho-nitrobenzoylation of 1-phenylsulfonylpyrrole proceeded satisfactorily, giving 20, the orientation of substitution being confirmed by the 1 H-NMR spectrum of the product which showed signals for the three orthorelated pyrrole protons at δ 6.34 (1H, m), 6.50 (1H, dd, J 1.8,3.8Hz) and 7.97 (1H, dd, J 1.8,3.8Hz).After N-deprotection, AlCl 3 -catalysed acetylation gave the desired diketone 21, by substitution meta to the nitrobenzoyl group, as confirmed by the small coupling constants for the remaining two pyrrole protons, at δ 6.90 (bs) and 7.92 (bs).Disappointingly, the Wittig cyclisation with nitro-substituted substrate 21 failed completely -only polar material that could not be characterised was obtained (Scheme 6).

Scheme 6
It was clear that an alternative nitrogen substitutent would have to be used -a protected amine was the obvious choice.There appeared to be a rather direct route through to a benzamide following a strong earlier precedent. 44Anthranilic acid was converted into 2-phenyl-4Hbenzoxazin-4-one 22 by reaction with benzoyl chloride and this reacted with the Grignard derivative of pyrrole giving 23.C-Acetylation proceeded normally and the resulting diketone 24a was subjected to the Wittig cyclisation conditions, but once again we were to be disappointed, and no bicyclic product was obtained (Scheme 7).There are two N-hydrogens in 24a and the cyclisation requires that base remove the pyrrole N-hydrogen.By reaction of 24a with one equivalent of NaH and then iodomethane we confirmed that the pyrrole N-proton is the most acidic proton, by the formation of 24b, in which the typical signal for a pyrrole N-hydrogen (δ 10.6 in 24a) was no longer present, but the amide proton signal remained, and at exactly the same shift, δ 11.4, as in 24a.It is not clear why 24a would not take part in the Wittig cyclisation.Hydrolysis of the benzamide revealed the amine 25, but, as expected, no Wittig cyclisation could be achieved with this, either.

Scheme 7
It seemed that it would be necessary to mask both of the side-chain N-hydrogens.Attempts to convert the amino group in 25 into a 1,3,5-triazine, 45,46 a phthalimide, or to effect a condensation (Schiff's base) with benzaldehyde, all failed.However, reaction with Bredereck's reagent [t-BuOCH(NMe 2 ) 2 ] converted 25 efficiently into 26.An interesting practical point concerning the relative mobilities on silica gel chromatography of starting primary amine 25 and product amidine 26 is that the latter was by far the slower, despite the absence of N-hydrogens, presumably because of its greater basicity.To our delight, the Wittig cyclisation strategy was now successful, producing 27 from 26, the relatively low yield undoubtedly reflecting the practical difficulty in separating the product from the triphenylphosphine oxide byproduct.Now, catalytic reduction gave 28 then conversion of acetyl into ethyl, as before, produced a mixture of 29 and deprotected 30, in unoptimised yields of 12% and 40% respectively (Scheme 8).

2-Phenyl-4H-benzoxazin-4-one (22).
To a stirred mixture of anthranilic acid (8.88 g, 64.8 mmol) in THF (130 ml) at 0-5 °C, Na 2 CO 3 (powder, 13.73 g, 129.6 mmol, 2 equiv) was added followed by benzoyl chloride (18.8 ml, 162 mmol, 2.5 equiv).After 10 min, the cold bath was removed and the mixture was stired at rt overnight.Water (130 ml), was added and the mixture was stirred for 10 min prior to filtration.The solid was washed with water and then 50% aq.MeOH.The additional material precipitated from the filtrate was collected by filtration and washed.The combined crops were dried at 50 °C under high vacuum to give 2-phenyl-4Hbenzoxazin-4-one 22 (14.2 g, 98%) as a white solid, mp 119 °C, lit 49   (20).(23).After a solution of EtMgCl (52.5 ml, 2 M solution in THF, 105 mmol, 2.1 equiv) in anhydrous THF (20 ml) was cooled to 0 °C, a solution of freshly distilled pyrrole (7.98 ml, 115 mmol, 2.3 equiv) in dry toluene (7.98 ml) was added dropwise over 20 min, keeping the mixture in an ice bath.After the mixture was stirred to rt for 20 min, a suspension of 2-phenyl-4H-benzoxazin-4-one 22 (11.15g, 50 mmol) in THF (35 ml) was added.After 45 min stirring, the mixture was refluxed for 3 h.Sat.aq.NH 4 Cl (11.5 ml) was then added to the hot mixture over 5 min.After 20 min of stirring, Na 2 SO 4 (11.5 g) was added.The suspension was stirred for 20 min prior to filtration.The collected solid was washed with THF, and the combined organic filtrate and washes were concentrated to dryness.The residue was suspended in toluene (50 ml), and the suspension cooled in an ice bath for 20 min.The solid was collected by filtration and washed with hexane.Drying at rt overnight under high vacuum gave N- [2-(1H-pyrrol-2-ylcarbonyl)