A synthetic approach to (±)-aristomakine

We describe the synthesis of trans -11b-methyl-2,3,4,6,11,11b-hexahydro-1 H -benzo[ a ]carbazol-3-ol ( 2 ) in five steps from the Wieland-Miescher ketone 3 in 17% overall yield. The N-benzyl analogue ( trans- 11-Benzyl-11b-methyl-2,3,4,6,11,11b-hexahydro-1 H -benzo[ a ]carbazol-3-ol) 15 was likewise prepared. Attempts thus far to fashion (±)-aristomakine ( 1 ) from 2 , 15 , or derivatives have not been successful.


Introduction
In view of our interest in the synthesis of alkaloids of the plant genus Aristotelia, namely (-)-hobartine and (+)aristoteline, 1 we wished to pursue a synthesis of (±)-aristomakine (1), an unusual alkaloid from Aristotelia serrata containing an N-isopropyl group. 2 A biomimetic synthesis of this alkaloid was described by Burkard and Borschberg in 1990, 3 the only synthesis of this rare alkaloid reported to date.
We envisioned a route to 1 that involved the elaboration of trans-11b-methyl-2, 3,4,6,11,11b-hexahydro-1H-benzo[a]carbazol-3-ol (2), which could be prepared from the venerable Wieland-Miescher ketone (3) via Fischer indolization of 3 or a suitable derivative.The obvious challenges lay in the regioselective dehydration of the axial hydroxyl group and the regio-and stereoselective amination of the olefin in 2. We now describe our progress to this end.All compounds are obtained as racemates.

Results and Discussion
The Wieland-Miescher ketone (3) was prepared in 48% yield from the condensation of 2-methyl-1,3cyclohexanedione with methyl vinyl ketone, 4 but attempts to effect direct indolization of 3 invariably led to intractable tars.Complications in this reaction may arise upon attack of the phenylhydrazine on the more reactive unconjugated carbonyl of 3, followed by ring fragmentation.However, as shown in Scheme 1, protection of the conjugated ketone and subsequent Fischer indolization smoothly led to the desired indole ring system (6 and 7).Thus, thioketals 4 and 5 were easily prepared by treating 3 with 1,2-ethane dithiol or 1,3-propane dithiol and boron trifluoride etherate in dry ether via the method of Smith. 5Under the relatively mild conditions developed by Dave 6 (glacial acetic acid at 100 °C) 4 cyclized smoothly to give indole 6 in 71% yield after purification.Ketal 5 also reacted with phenylhydrazine to give indole 7 but the yield was only 41%.Removal of the protecting group to give 8 was attempted under a variety of conditions.Only the treatment of 6 with methyl iodide 7 or silver nitrate 8 afforded 8 in 55% yield (Scheme 1).The results are summarized in Table 1.Origin material refers to material that did not move on TLC.Starting material and origin material ---9 HCl/DMSO 15 Tar ---While the deprotection of 6 with silver nitrate was comparable in yield to the reaction with methyl iodide, the former reaction required 1-2 hours while the latter required 24-30 hours for completion.Another advantage of silver ion deprotection is that the ethane dithiol bis-silver salt precipitates during the course of the reaction and is simply removed by filtration.Methyl iodide, on the other hand, generates the dimethyl thioether of ethane dithiol, which must be removed by column chromatography.Similarly, thioketal 7 could be deprotected with silver nitrate in 48% yield, but since 7 was obtained in only 41% yield, we did not study this deprotection step in depth.
Standard methodology [16][17][18][19] was then employed for the two-step conversion of enone 8 to the target alcohol 2 (Scheme 2).The unstable enol acetate 9 was prepared in almost quantitative yield by refluxing 8 in isopropenyl acetate in the presence of a catalytic amount of p-toluenesulfonic acid.Subsequent hydrolysis/reduction of 9/10 with sodium borohydride in aqueous ethanol afforded 2 in 82% yield from 8.Although systems similar to 9 are known to be stable 16,17 the instability of 9 may be attributed to a double bond migration into conjugation with the indole nucleus.The UV spectrum of 2 (and that of 11) is that of an indole and not a 3-vinylindole.
It was important to determine the stereochemistry of the hydroxyl group of 2, although in principle either isomer could be utilized in subsequent reactions.Precedents from terpene chemistry 16,17,[20][21][22] indicated that for most cyclohexanone reductions the predominant product is the equatorial alcohol because of minimization of torsional strain in the transition state. 23,24For example, Spencer observed the axial methine proton of 10--methyl-∆ 5 (10) -8,9-octal-2-ol to resonate at 3.50, with a small signal due to the equatorial methine proton resonating at 4.05. 21The high field (300 MHz) NMR spectrum of 2 shows the corresponding methine proton at 4.25 indicative of an equatorial proton and hence an axial hydroxyl group.To confirm this unexpected stereochemistry, we synthesized acetate 11.Treatment of 2 with acetic anhydride and 4dimethylaminopyridine (DMAP) in dichloromethane cleanly gave acetate 11.Alternatively, 11 was prepared by allowing 2 to react with acetic anhydride in dry pyridine, but this method produced several side products.The proton NMR of 11 shows the (equatorial) methine proton shifted downfield as expected to  5.25.The two vinyl protons at  5.42 (major) and 5.71 integrate for a factor of 6:1 in favor of the axial acetate 11.On the basis of these NMR data we conclude that the axial alcohol 2 is the major reduction product from ketone To preclude possible side reactions in the remaining reactions leading to aristomakine (due to the acidic indole NH), we protected the aforementioned compounds as N-benzyl analogues.Unfortunately, direct benzylation of indole thioketal 6 with benzyl bromide (KOH/DMSO, butyllithium, or tetra-n-butyl-ammonium hydrogen sulfate/KOH) failed in each case.However, the desired N-benzylindole thioketal 12 was synthesized in 53% yield using 1-benzyl-1-phenylhydrazine 25 in a Fischer indolization with ketone 4 (Scheme 3).Ketone deprotection, enol acetate formation, and reduction as previously described afforded alcohol 15.As before, the acetate of 15 was synthesized (acetic anhydride/DMAP/CH 2 Cl 2 ) to confirm the stereochemistry.The NMR spectra of 15 and 16 show the methine protons at 4.20 and 5.30, respectively, indicating an equatorial orientation for the protons, consistent with the results for 2 and 11.For 15 there is a small resonance at  3.45 which may be the axial proton of the epimer of 15.However, no such resonance appears in the spectrum for acetate 16.Moreover, no second vinyl proton resonances are observed for either 15 or 16 leading us to believe that the reduction goes highly, if not exclusively, equatorially to afford the axial hydroxyl group (15).Scheme 3. Synthesis of racemic tetracyclic alcohol 15 and acetate 16.
Our synthesis of aristomakine (1) was predicated on the regio-and stereoselective amination of the double bond of 2 or 15, followed by dehydration of the axial hydroxyl group.Our initial approach was to add an isopropylamino group to the beta face of the double bond of 2 via an intermolecular nitrene insertion, 26 followed by dehydration.However, our attempts to oxidize isopropyl amine with N-chlorosuccinimide (NCS) or lead tetraacetate in the presence of 2 led only to starting material.Control experiments with the more reactive 2norbornene failed to give the expected aziridine, suggesting that intramolecular formation of acetone imine by a 1,2-H shift may supersede an intermolecular nitrene addition.This pathway is known for the pyrolysis of nalkyl azides.For instance, n-butyl azide affords an 89% yield of its imine. 27We then employed Brown's hydroboration-amination method with 2. 28,29 Thus, Brown showed that 1-methylcyclohexene gives trans-2methylcyclohexylamine upon hydroboration and treatment of the resulting borane with hydroxylamine-Osulfonic acid. 29Unfortunately, these conditions with 2 yielded a complex mixture.
In conclusion, we have developed an efficient syntheses of key intermediates (2/15) in our efforts toward the total synthesis of aristomakine and other Aristotelia alkaloids.Intermediates 2 and/or 15 could potentially be used to prepare several members of the Aristotelia family.