Total Synthesis of the Bridged Indole Alkaloid Apparicine

An indole-templated ring-closing metathesis or a 2-indolylacyl radical cyclization constitute the central steps of two alternative approaches developed to assemble the tricyclic ABC substructure of the indole alkaloid apparicine. From this key intermediate, an intramolecular vinyl halide Heck reaction accomplished the closure of the strained 1-azabicyclo[4.2.2]decane framework of the alkaloid with concomitant incorporation of the exocyclic alkylidene substituents.


Introduction
Apparicine ( Figure 1) is a fairly widespread monoterpenoid indole alkaloid, first isolated from Aspidosperma dasycarpon more than 40 years ago. 1,2 Its structural elucidation, 2 carried out by chemical degradation and early spectroscopic techniques, revealed a particular skeleton with a bridged 1-azabicyclo [4.2.2]decane framework fused to the indole ring and two exocyclic alkylidene (16-methylene and 20E-ethylidene) substituents. 3 The same arrangement was also found in vallesamine 4 and later in a small number of alkaloids, including 16(S)hydroxy-16,22-dihydroapparicine 5 or ervaticine, 6 which differ from apparicine in the substitution at C-16. 7 The apparicine alkaloids are biogenetically defined by the presence of only one carbon (C-6) connecting the indole 3position and the aliphatic nitrogen, which is the result of the C-5 excision from the original two-carbon tryptamine bridge of the alkaloid stemmadenine. 8 The fragmentation-iminium hydrolysis-recyclization route depicted in Scheme 1, which involves the operation of a stemmadenine N-oxide equivalent, 9 has been proposed to rationalize this biogenetic relationship. Such a route appears to be likely since stemmadenine itself 10 and, more  ; Djerassi, C. Tetrahedron 1965, 21, 1141-1166 (2) Joule, J. A.; Monteiro, H.; Durham, L. J.; Gilbert, B.; Djerassi, C. J. Chem. Soc. 1965, 4773-4780. (3) The E configuration of the ethylidene group was established some years later: Akhter, L.; Brown, R. T.; Moorcroft, D. Tetrahedron Lett. 1978, 19, 4137-4140. (  Chem. Commun. 1978, 947-948. recently, pericine (subincanadine E) 11 have been transformed in vitro into the respective C-5 nor-alkaloids (vallesamine and apparicine) by treatment of the N-oxides with trifluoroacetic anhydride (modified Polonovski reaction). 12 Their synthetically challenging structures make apparicine alkaloids attractive targets for synthesis. However, progress in this area has been limited to the approach developed by Joule's group in the late 1970s, which allowed the construction of the ring skeleton of apparicine (i.e., 20-deethylideneervaticine) but proved unsuitable for the total synthesis of the alkaloid. 13 We envisaged apparicine to be accessible via tricyclic ABC substructures containing the central eight-membered ring (e.g., azocinoindoles A, Scheme 2), from which the carbon skeleton would be completed by inserting an ethylideneethano unit between the aliphatic nitrogen and C-5. In particular, it was planned that after N-alkylation with the appropriate haloalkenyl halide, an intramolecular Heck reaction 14 upon a 2-vinylindole moiety would serve to close the piperidine ring and at the same time install the requisite 20E-ethylidene and (when R 2 = Me) 16-methylene appendages. It should be noted that similar Heck couplings of vinyl halides and elaborated cycloalkenes have proved to be useful for the assembly of the bridged core of several indole alkaloids, including pentacyclic Strychnos alkaloids, 15 strychnine, 15c,16 minfiensine, 17 apogeissoschizine, 18 and ervitsine. 19 However, to the best of our knowledge, there are no reported vinyl halide Heck reactions involving (aza)cyclooctene rings to produce strained bridged systems. 20 The power of ring-closing metathesis (RCM) 21 to synthesize medium-sized rings 22 and our own work on RCM of indole-containing dienes 23 made it our method of choice to assemble the indolo fused eight-membered ring of the key intermediates A and also to install the double bond required for the Heck reaction, either directly or after an isomerization step. In the course of our work, an alternative approach to A based on 2-indolylacyl radical cyclization 24 and manipulation of the resulting ketone was also investigated. 25 This Article deals with the development of the above indole annulation chemistry and its application to complete the first total synthesis of (()-apparicine. 26

Resuts and Discussion
Initial Studies. We set out to study the indole-templated RCM en route to apparicine, directly targeting 6-methylazocino [4,3-b]indoles (A, R 2 = Me, Scheme 2) with the trisubstituted 5,6-double bond functionality required for the Heck coupling. To this end, 2-isopropenylindoles 3, which are equipped with Boc or Ts groups at the aliphatic nitrogen SCHEME 1. Biosynthesis of Apparicine Alkaloids SCHEME 2. Synthetic Strategy JOCArticle and a robust MOM group at the indole nitrogen, were selected as the starting dienes (Scheme 3). These compounds were efficiently prepared from the known 2-chloroindole-3-carbaldehyde 1 27 by a Stille coupling with (isopropenyl)tributylstannane, followed by reductive amination of aldehyde 2 with 3-butenylamine and subsequent acylation or sulfonylation of the resulting secondary amine with di-tert-butyl dicarbonate or tosyl chloride, respectively. Unfortunately, exposure of dienes 3 to the second-generation Grubbs catalyst B in CH 2 Cl 2 or toluene did not deliver the expected eight-membered ring. Instead, carbamate 3a mainly underwent an intermolecular metathesis reaction leading to dimer 4a, even when working under high dilution conditions (0.007 M). Sulfonamide 3b, in turn, led to the respective dimer 4b along with variable amounts of the ring-contracted product 5, coming from the competitive isomerization of the terminal double bond followed by RCM with liberation of propene. Azepinoindole 5 was the only isolated product (75%) when cyclization of 3b was performed in refluxing toluene. In both cases, the use of other metathesis catalysts, either based on ruthenium (first-generation Grubbs or second-generation Hoveyda-Grubbs catalysts) or molybdenum (Schrock's catalyst) did not lead to any improvement.
Because this unsuccessful result was probably due to the presence of a geminal disubstituted terminal alkene moiety in dienes 3, we turned our attention to more easily available 6-demethyl tricyclic substructures (A, R 2 =H, Scheme 2). These model azocinoindoles would also serve as precursors for closing the piperidine ring of apparicine by a reductive Heck cyclization or a tandem Heck cyclization-capture, which could also allow the introduction of the remaining carbon atom at C-16. The implementation of this new synthetic plan is depicted in Scheme 4.
Thus, the required RCM substrates 7a and 7b were uneventfully prepared from 2-vinylindole-3-carbaldehyde 6 27 by reductive amination with 3-butenylamine followed by N-acylation or sulfonylation, as in the above isopropenyl series. As anticipated, RCM of dienes 7a or 7b, involving two terminal monosubstituted alkene units, took place with the use of ruthenium complex B under standard conditions (0.01 M, toluene, 60°C or CH 2 Cl 2 , reflux) to give azocinoindoles 8a or 8b in acceptable yields. At this point, access to the more advanced synthetic intermediate 9 required the manipulation of the aliphatic nitrogen of 8a or 8b to install the iodoalkenyl chain for eventual cyclization. As our first attempts to induce N-desulfonylation of 8b under reductive conditions (Mg, NH 4 Cl, MeOH or Na/ naphthalenide, THF) proved problematic, affording the unchanged starting product or complex reaction mixtures, we focused on the more labile carbamate function of 8a. Removal of the Boc group under standard acidic conditions (TFA, Me 3 SiI, ZnBr 2 ) was also troublesome, leading to partial decomposition, but the deprotection took place cleanly upon exposure of 8a to a mild acidic protocol (1.2 M HCl in MeOH at rt). The resulting secondary amine was directly subjected to alkylation with (Z)-2iodo-2-butenyl tosylate in hot acetonitrile in the presence of K 2 CO 3 to give 9 in 60% isolated yield over the two steps.
We next studied the key formation of the piperidine ring by Pd-catalyzed cyclization of the vinyl iodide upon the 2vinylindole moiety. Our expectation was that the initially formed alkylpalladium intermediate C (Scheme 5), in which no β-hydrogen is available for elimination, would be stable enough to be reduced or trapped with a suitable quencher. However, when 9 was subjected to a number of standard conditions for reductive Heck reactions, the desired tetracyclic system D (Q = H) was never detected. The only observed process under the phosphine-free conditions 28 [Pd(OAc) 2 , SCHEME 3. Attempted Direct RCM Synthesis of 6-Methyl-1,2,3,4-tetrahydroazocino [4,3-b]indoles SCHEME 4. Studies in the 6-Demethyl Series K 2 CO 3 , TBACl, HCO 2 Na, DMF, 80°C] previously used by Overman 17a on a related substrate was N-dealkylation. After this unsuccessful result, a variety of palladium precatalysts [Pd(OAc) 2 , Pd(PPh 3 ) 4 , Pd 2 (dba) 3 ], ligands (PPh 3 , dppe), and cosolvents (toluene, CH 3 CN, THF) were examined in the presence of Et 3 N or diisopropylethylamine as the potential reductants. 16b,29 Whereas short reaction times left the starting product unchanged, prolonged heating gave low yields (5-10%) of the unexpected tetracycle 10, coming from an apparent 7-endo cyclization with inversion of the ethylidene configuration. 30 The yield of 10 was raised to 30% on exposure of 9 to Pd(PPh 3 ) 4 in 1:1 THF-Et 3 N in a sealed tube at 90°C for 24 h. On the other hand, under cationic conditions [Pd(OAc) 2 , PPh 3 , Ag 2 CO 3 , 1:1 toluene-Et 3 N, 90°C] the cyclization proceeded readily to give tetracycle 10 in 46% isolated yield. Significantly, this result was not substantially altered when the reaction was carried out in the presence of HCO 2 Na as the reductant or KCN, K 4 [Fe(CN) 6 ], TMSCN, or tributylvinylstannane as trapping agents.
The formation of unusual Heck cyclization products like 10 has been previously observed 31 and rationalized 32 by considering that the initial 6-exo cyclization is followed by an intramolecular carbopalladation on the exocyclic alkene. The resulting cyclopropane intermediate would undergo rearrangement, with concomitant inversion of the alkene geometry, and final β-hydride elimination. In our case, the cyclopropanation-rearrangement route depicted in Scheme 5 would be fast enough to prevent the quenchers from intercepting the initially formed alkylpalladium intermediate C.
It now became apparent that the presence of a 6-methyl group in the Heck cyclization substrate was crucial to assemble the bridged framework of apparicine, as it would guarantee the β-elimination of the alkylpalladium intermediate arising from cyclization, thus hampering the above undesired route. Hence, we renewed our efforts to synthesize 6-methylazocino [4,3-b]indoles (A, R 2 =Me, Scheme 2), once again tackling the problem of RCM and ready to explore new routes.
RCM-Isomerization Route to Azocinoindole 15. Given that we were unable to directly form the trisubstituted double bond included in the azocine ring by RCM, we decided to change the cyclization site from the 5,6-position to the less crowded 4,5-position by using a 3-(allylaminomethyl)-2allylindole such as 13 (Scheme 6) as the diene. Consequently, the synthesis of the Heck precursor would now require an additional isomerization step of the resulting double bond.
It was planned to install the R-methyl-substituted allyltype chain at the indole 2-position taking advantage of an allylic nucleophilic substitution reaction using a suitable organometallic derivative of indole. Thus, 1-(phenylsulfonyl)indole was allowed to react with n-BuLi and CuCN, and the intermediate organocopper derivative was treated with (E)-4-chloro-2-pentene. The resulting indole 11 was then converted into the RCM precursor 13 by Friedel-Crafts formylation, reductive amination of aldehyde 12 with allylamine, and the subsequent protection of the aliphatic nitrogen with a Boc group. The overall yield of the four steps was 58%. Satisfactorily, ring closure of diene 13 took place with the second-generation Grubbs catalyst B under standard conditions (0.07 M, CH 2 Cl 2 , reflux) to give the desired 6methylazocinoindole 14 in 80% yield.
Attention was then focused on the isomerization step. Considering recent reports on alkene isomerizations mediated SCHEME 5 SCHEME 6. RCM-Isomerization Route to 6-Methylazocinoindole 15  Soc. 1992, 114, 10091-10092. by suitably modified ruthenium-metathesis catalysts, 33-36 we sought to examine if such a protocol could be synthetically useful for our purpose (Scheme 7). Unfortunately, when azocinoindole 14 was treated with catalyst B in refluxing toluene, 35 a slow isomerization of the double bond took place to its N-conjugated counterpart (3,4-position), providing the enecarbamate 16 in 50% yield (not optimized). The directing effect of the carbamate nitrogen was also decisive, although to a lesser extent, in the ruthenium-catalyzed isomerization of the 6-demethyl analogue 17a, 23c which led to the enamide 18a as the major product along with minor amounts of vinylindole 19a. Significantly, the influence of the heteroatom was suppressed in the N-tosyl analogue 17b, 23c which underwent isomerization to afford vinylindole 19b as the only product. Finally, no isomerization was observed upon exposure of azocinoindoles 14 or 17 to catalyst B in hot methanol. 36 Satisfactorily, we fortuitously discovered that the double bond of azocinoindole 14 moved into conjugation with the aromatic ring under the basic conditions used to remove the phenylsulfonyl group. Thus, long exposure of 14 to t-BuOK in refluxing THF brought about the anticipated indole deprotection along with alkene isomerization, affording 15 in 90% yield (Scheme 6). By using shorter reaction times and using NMR spectroscopy, we found that the migration of the double bond took place after the initial indole N-deprotection step, which suggests that the base-induced isomerization is only compatible with the presence of a free indole NH group.
Alternative Synthesis of 15. Although the RCM-isomerization route depicted in Scheme 6 allowed an efficient synthesis of the key apparicine intermediate 15 [41% overall yield from 2-(phenylsulfonyl)indole by way of four isolated intermediates], we explored the possibility of installing the trisubstituted double bond required for the Heck reaction from a ketone carbonyl group. To this end, the first substrate examined was the N-MOM tricyclic ketone 20 (Scheme 8), since it had already been prepared by RCM followed by removal of the resulting double bond by hydrogenation. 23c Reaction of 20 with MeLi smoothly provided tertiary alcohol 21, which was subjected to several dehydration protocols without success. Thus, the acid-catalyzed dehydration using 3 M H 2 SO 4 in acetone or TsOH in benzene was complicated by the competitive indole deprotection, affording low yields of the endocyclic alkene (15). On the other hand, the use of Martin sulfurane resulted in a cleaner dehydration to the exocyclic alkene 22, in which the N-MOM group remained unaffected.
In search of a more efficient approach, we decided to extend the above organometallic addition-dehydration sequence to an analogous indole unprotected ketone (i.e., 26, Scheme 9). After unsuccessful attempts to remove the N-MOM group of 20, the substrate was efficiently prepared by a more direct route free of indole protecting groups, based on an 8-endo cyclization of a 2-indolylacyl radical upon an amino tethered alkene. 24,37 The synthesis began with the preparation of selenoester 25 as the radical precursor, equipped with a bromovinyl chain to increase both the efficiency and the endo regioselectivity of the ring closure. 37 Thus, reductive amination of aldehyde 23 with 2-bromo-2propenylamine followed by standard protection of the resulting secondary amine with a Boc group led to ester 24, which was converted into 25 by phenylselenation through the corresponding carboxylic acid. 38 Treatment of selenoester 25 with n-Bu 3 SnH as the radical mediator and Et 3 B as the initiator achieved the desired ring closure affording ketone 26 in moderate yield (54%). Finally, to our satisfaction, reaction of 26 with methyllithium followed by dehydration of the resulting tertiary alcohol under mild acid conditions (TsOH, CH 3 CN, rt) smoothly provided the target alkene 15. Using this alternative route, the synthesis of 15 was accomplished from aldehyde 23 in 26% overall yield by way of only three isolated intermediates. Completion of the Synthesis of Apparicine. With azocinoindole 15 in hand, we next sought to manipulate the aliphatic nitrogen to install the haloalkenyl chain for the subsequent Heck reaction. As occurred with the C-6 demethyl analogue 8a (Scheme 4), removal of the N-Boc group of 15 required a mild acid protocol to avoid decomposition. The resulting secondary amine proved to be highly unstable and was directly subjected to alkylation with (Z)-2-iodo-2-butenyl tosylate to give 27 in 30% isolated yield over the two steps (Scheme 10). Attempts to place a phenylsulfonyl group at the indole nitrogen of 15 in order to improve the yield were unsucccessful.
The stage was now set for the completion of the synthesis by intramolecular coupling of the vinyl iodide and the trisubstituted alkene. A variety of experimental conditions were screened, including different solvents, palladium precatalysts, and additives, resulting only in the recovery of the starting material or decomposition products. However, the critical closure of the strained 1-azabicyclo [4.2.2]decane framework with concomitant incorporation of the exocyclic alkylidene substituents took place under cationic conditions, although loss of material was still extensive. Thus, when vinyl iodide 27 was subjected to a specific protocol, using Pd(OAc) 2 /PPh 3 (0.2:0.6 equiv) and Ag 2 CO 3 (2 equiv) in 1:1 toluene-Et 3 N at 80°C for a short reaction time (1.5 h), apparicine was obtained in a consistent, reproducible 15% isolated yield. The 1 H and 13 C NMR spectroscopic data of synthetic apparicine essentially matched those described in the literature for the natural product. 2,7,39 Additionally, the chromatographic (TLC) behavior of synthetic apparicine was identical to an authentic sample.

Conclusion
In summary, the first total synthesis of (()-apparicine has been accomplished by a concise route employing a vinyl halide Heck cyclization to close the bridged piperidine ring in the last synthetic step. The key azocinoindole intermediate 15 has been successfully assembled by developing two alternative procedures, namely, an indole-templated RCM followed by base-induced isomerization and an acyl radical cyclization followed by ketone-alkene functional group interconversion.