A Short Synthesis of (±)‐3‐Demethoxyerythratidinone by Ligand‐Controlled Selective Heck Cyclization of Equilibrating Enamines

Abstract A short, 5‐step total synthesis of (±)‐3‐demethoxyerythratidinone from a simple pyrrole derivative is described. Features include the formation of gram quantities of a key tricylic aziridine from a challenging photochemical cascade reaction through the use of flow photochemistry. The final step involved a highly unusual Heck cyclization whereby ligand control enabled efficient formation of the natural product in 69 % yield from the minor isomer present in an equilibrating mixture of labile enamines.

Abstract: As hort, 5-step total synthesis of (AE)-3-demethoxyerythratidinone from as imple pyrrole derivative is described. Features include the formation of gram quantities of ak ey tricylic aziridine from ac hallenging photochemical cascade reaction through the use of flowphotochemistry.The final step involved ah ighly unusual Heckc yclization whereby ligand control enabled efficient formation of the natural product in 69 %y ield from the minor isomer present in an equilibrating mixture of labile enamines.
The alkaloid (+ +)-3-demethoxyerythratidinone 1 is one of over 100 natural products produced by the Erythrina genus of flowering plants. [1] The Erythrina alkaloids display ab road range of pharmacological activities including hypotensive, sedative,n euromuscular blocking,C NS depressing and curare-like activities.T he key structural feature of this family is the tetracyclic tetrahydroisoquinoline core.S ince the first total synthesis of 1 by Tsuda in 1984, [2] this tetracyclic alkaloid has been used by others in order to demonstrate the utility of various synthetic methodologies. [3] Av ery elegant synthesis was recently reported by Reisman, where chiral sulfinyl imine chemistry was used to control the stereochemistry of the key quaternary center, [3d] giving (À)-3-demethoxyerythratidinone in just six steps overall ( Figure 1).
Individually both photochemical and Pd-catalyzed crosscoupling methodologies have been shown to be powerful techniques in organic synthesis as well as valuable tools in the synthesis of complex molecular architectures. [4] We have previously reported the photochemical transformation of simple N-butenyl-substituted pyrroles into complex tricyclic aziridines 2. [5] We recently demonstrated that these strained photochemical products undergo arange of thermal and Pdcatalyzed ring-opening/annulation reactions to produce ab road range of fused polyheterocycles,i nj ust two steps from simple pyrroles (Scheme 1). [6] We were therefore keen to exploit the functionality and inherent strain in these aziridines as part of an alkaloid synthesis,i np articular the aziridine carboxylates 3.H erein we report as hort total synthesis of (AE)-1 utilizing ah ighly unusual and selective, ligand controlled intramolecular Heck-reaction onto one of apair of equilibrating enamine intermediates.
Our initial strategy to 1 (Scheme 1) involved aryl cyclization onto the iminium ion 4 which itself would be generated by in situ decarboxylation of the amino acid 5. [7] Although iminium ion cyclizations are one of the most frequently used approaches to such alkaloids, [8] these have usually involved the intermediary of N-acyliminium ions. [9] Ther equisite amino-acid 5 should be obtained by Pd 0 -catalyzed acetate ring-opening of the aziridine 3 followed by N-alkylation with the iodide 6.
Irradiation of pyrrole 7 (254 nm) gave the aziridine (AE)-8 in a39% yield (Scheme 2). As before,wefound that this two- photon process suffers from low productivity due to al ikely low overall quantum yield and as such was limited to ca. 60 mg quantities of (AE)-8 at at ime using a6 Wb atch reactor. [5] Fortunately,w ef ound that using our three-lamp FEP-flow reactor gave routine access to 1.91 go fa ziridine (AE)-8 in as ingle 373 min run (productivity of 7.37 g/24 h). This highlights once again the value of flow photochemistry for scaling-up high dilution/low quantum yield reactions-this would simply not be possible on laboratory batch scale. [10] We then investigated the Pd-catalyzed Tsuji-Trost type reaction of (AE)-8 with acetate as nucleophile,t his proceeded in excellent yield (81 %) to give as ingle diastereomer of the (S N 2')ring-opened amine (AE)-9.Itwas interesting to observe that after 3h the regioisomeric ratio (NMR) of S N 2' to S N 2 ring-opened product was 1:15. However,a fter 16 hn one of the S N 2r ing opened product remained, indicating that this reaction is under thermodynamic control via the initially formed S N 2p roduct. Thed iastereomer obtained is in agreement with the classic "double-inversion" Tsuji-Trost mechanism for soft nucleophiles,w here an et retention of stereochemistry is observed. [11] Amine (AE)-9 was then alkylated with the iodide 10 a (X = H) in good yield using phase transfer conditions.T he tertbutyl group in (AE)-11 a was cleaved in 86 %yield by treatment with TFAt og ive the amino-acid (AE)-12.I na na ttempt to promote the key iminium ion cyclization (cf. 4 to 5)t he acid (AE)-12 was subjected to an umber of decarboxylation conditions, [7,12] including both POCl 3 and oxalyl chloride with catalytic DMF in CH 2 Cl 2 .I na ll instances the reaction was unsuccessful and no product could be detected amongst the complex reaction mixtures obtained. As it was suspected that this may have been related to elimination of the acetate group,atelescoped sequence was investigated, where (AE)-11 a was hydrolyzed and the resultant alcohol oxidized to the enone (AE)-13 with TPAP (74 %over two steps). In an attempt to access the desired amino acid, tert-butyl ester (AE)-13 was treated with TFA. However,u pon cleavage of the tert-butyl ester the resultant acid underwent spontaneous decarboxylation to generate the isolable,b ut labile enamine (AE)-14 in excellent yield. This is clearly aresult of the conjugated enone system enabling (AE)-13 to behave as avinylogous b-ketoester, sensitive to decarboxylation under acid conditions.U nder acidic conditions it was postulated that (AE)-14 should be in equilibrium with the desired iminium species (cf. 4). However,d espite further investigation no cyclization to (AE)-15 was observed when (AE)-14 was treated with ar ange of acid catalysts including PPA, p-toluene sulfonic acid, TFAa nd camphor sulfonic acid. [13] Intrigued by the ease of the decarboxylation of (AE)-13 to (AE)-14,w ee lected to change our strategy and focus on an approach involving aH eck cyclization onto an enamine to generate the quaternary spirocycliccentre of (AE)-1. [14,15] Such as trategy should generate ah igh enough quantity of the isomeric natural product (AE)-17 to convert it to (AE)-1 by areduction/ dehydrogenation sequence.
Tr eatment of (AE)-9 with 10 b gave the aryl iodide (AE)-11 b. Hydrolysis of the acetate and oxidation with TPAP gave the enone (AE)-16 in 74 %overall yield. Cleavage/decarboxylation of (AE)-16 gave the enamines (AE)-18 and 19, which although labile to isolation and separation, were formed in essentially quantitative yield as a4 :1 mixture,r espectively.I nitial attempts at the Heck cyclization were carried out on this mixture using Pd 0 /DMF conditions described by Waldmann [16] and led to a6 5% yield of (AE)-17 as as ingle diastereomer ( Table 1, entry 1). However,d uring reaction optimization ap eculiar observation was noted when conditions described by Orito [17] were employed (Table 1,  This raised the attractive prospect of modifying the reaction conditions such that (AE)-1 could be produced as the sole product from the minor component (19)b yi ns itu equilibration of this enamine mixture,t hus considerably shortening the overall synthesis of this natural product. We postulated that varying the Pd ligand might affect reaction selectivity and so abrief survey was undertaken ( Table 1).
It is clear that electron-rich phosphines (entries 4a nd 5) likely favor the formation of (AE)-1 by cyclization of the organopalladium-enamine isomer 21.C onversely,c omparatively electron poor ligands (entries 2a nd 3) likely favor cyclization to (AE)-17 via the isomer (AE)-20.Itispossible that L n HPdI from b-hydride elimination may serve as aconvenient catalyst for the isomerization of (AE)-18 to 19 and different ligands will affect the reactivity [18,19] of such ac atalyst e.g. reductive elimination vs.e namine isomerization. Waldmann [16] previously observed isomerized products from Heck cyclization onto dihydro-4-pyridones (enaminones) and attributed these to isomerization of initially formed apalladio-ketones via s-p-s-allyl rearrangement. In our case the same reaction conditions ( Table 1, entry 1) lead only to (AE)-17 and so it is likely that ap athway involving isomerization of (AE)-18 and 19 is plausible.
As 19 is achiral then this opened up the possibility of effecting an asymmetric synthesis of (+ +)-1 directly from the mixture of enamines.After screening arange of chiral ligands it became clear that most resulted in amixture of (AE)-17 and 1 with little ee observed for the latter (see the Supporting Information). Use of (À)-DIOP gave 55 %yield of (AE)-17 and 40 %o f1 with an ee of 25 %i nf avor of (À)-1 ( Table 1, entry 7). (R)-i Pr-PHOX ( Table 1, entry 8) gave a3 2% isolated yield of 1 with an ee of 38 %i nf avor of the natural enantiomer (+ +)-1.A ttempts to increase this by use of additives (e.g.A gs alts) resulted in inferior results or inhibition of reaction.
In conclusion, we have developed ashort 5-step sequence to (AE)-3-demethoxyerythratidinone in 15 %o verall yield from the simple pyrrole carboxylate 7.N otable features include the use of ap owerful two-photon cycloadditionrearrangement reaction to provide ar eactive aziridine 8,t he key intermediate of the synthesis.T his was produced in gram quantities using flow-photochemistry,which would have been very difficult to achieve in batch due to the high dilution and irradiation times required. This study also uncovered ahighly unusual and selective,ligand-controlled intramolecular Heck reaction. By use of electron rich phosphines (AE)-3demethoxyerythratidinone (1)w as formed as the major product from Heck cyclization onto the minor component of apair of enamine isomers.Use of electron poor phosphine ligands gave an isomer of the natural product by cyclization onto the major enamine isomer in the mixture.The generality of such as witching process and the mechanistic understanding merits further investigation. Use of chiral ligands for the asymmetric synthesis of (+ +)-1 yielded mixed results;u pt o 38 % ee was observed in favor of (+)-1 but at the expense of product selectivity. [a] 1equiv of TBAI.