Syntheses of Highly Functionalized Spirocyclohexenes by Formal [4+2] Annulation of Arylidene Azlactones with Allenoates

Abstract A straightforward phosphine‐catalyzed formal [4+2] annulation between α‐branched allenoates and arylidene azlactones has been developed to access highly functionalized spirocyclohexenes. This cyclization favors the γ‐addition of the phosphine‐activated allenoates over a β′‐addition pathway. Detailed computational studies support the proposed mechanism and provide a reasonable explanation for the observed regioselectivity and the noted effect of the catalyst.


Reactiond evelopment
Our initial screening of the reactionc onditions wasc arriedo ut by treating parenta cceptor 1a with diethyla llenoate ester 4a as our modelr eaction. An overview of the most interesting results from ad etailed screening of different catalysts and conditions is shown in Ta ble 1. All of thesereactions were performed for 20 hi nt he presence of 3equiv of allenoate 4a and 1equiv of 1a.O ther 4a/1a ratios were examined, but lower yields were afforded, most likely the result of the allenoates participating in side reactions. [14] Initial experiments that used either catalytic amountso f PBu 3 ,P Ph 3 ,o r1 ,4-diazabicyclo[2.2.2]octane (DABCO) in the presence of Cs 2 CO 3 in CH 2 Cl 2 revealed that only PBu 3 led to the targeted [4+ +2] annulation reaction (Table1,e ntries 1-3). The formation of adduct 5a,w hich originates from the g-addition of the allenoate to the acceptor, wasc learly favored over the generation of b'-addition product 6a (Table 1, entry 1). We also realizedt hat the use of degassed solvents and the addition of molecular sieves had ab eneficial effect on the product yield ( Table 1, entry 1v s. 4). Accordingly,a ll furtherr eactions were performedi nd ry,d egassed solvents and in the presenceo f 4 molecular sieves (MS).
Product 5a was always obtained as am ixture of two diastereomers, and the relative configurationo ft he major diastereomer was unambiguously proven by single-crystal X-ray diffraction studies ( Figure 1). [15] Notably,t he other two possible diastereomers of 5a were not detected, but we later observed the formation of additional diastereomers of some of the other derivatives (see Scheme 2).
Using as toichiometric amount of PBu 3 had no positive effect on the yield (Table 1, entry 5). Among the different examined solvents, THF and toluenel ed to better resultst han those initially obtained by using dichloromethane (DCM, Ta ble 1, entries 6a nd 7). We also found that the reaction afforded as lightly higher yield (but with somewhat lower regioselectivity) in the absence of ab ase ( Table 1, entry 8). The yield was furtheri ncreasedb yc arrying out the reaction at an elevated temperature, and under theseconditions, THF gave aslightly better yield than that obtained in toluene (Table 1, entries 9 and 10).
[f] Cy = cyclohexyl, n.r. = no reaction. utilized in such allenoate cyclizationsa re typicallys terically demandinga nd contain aryl groups. [12] Nevertheless,w ed id perform af ew reactions with chiral phosphines such as compounds P2 and P3 (both of which have been successfully used in [3+ +2] annulations of azlactones 1 with allenes 2 as shown in Scheme 1 [12b, c] ), but with absolutely no success( Ta ble 1, entries 16 and1 7). This same lack of reactivity was also observed with other known chiral phosphines. [16] Next, we investigatedt he scope of the racemic protocol for this reaction by using various substituted acceptors 1 and allenoates 4 in the presence of PBu 3 (Scheme 2) The reactions with PBu 3 were easier to handle and more robust, as PEt 3 was found to be more sensitive towards oxidation. The isolated yields, in most cases, were within the same range as that obtained for the modelr eaction. However,t he resulting diastereoselectivity was clearly influenced by the electronic and steric properties of the startingm aterials, and, in some cases, we observed the formation of at hird diastereomer (Scheme 2, see compounds 5g and 5i). For all of the substrates, the formationo fg-addition product 5' was favored overt hat of b'-addition product 6,a lthought he selectivity was not as pronounced in some cases. This was especially observed by changing the ester group of the allenoate (Scheme 2, see products 5b/6b and 5d/6d)a nd by introducing an ortho-methoxy group to the aryl group of acceptor 1 (Scheme 2, products 5f/6f). In all cases, the isolationo fb'-adduct 6 was difficult, as these compounds coeluted with the allenoate degradation products, the mixture of which could not be separated by columnc hromatography.T herefore, no isolated yields for compound 6 have been reported.
As for the potential transformations of spiro products 5',w e carriedo ut aq uick test to determine if the selectiveh ydrolysis of the azlactone is possible. Under ambient acidic conditions, we found that the azlactone was easily hydrolyzed without a reactionoccurring at the other ester functionalities. [16] Computational studies To gain af undamental understanding of the mechanism and the origin of the regio-and stereoselectiviy of this formal [4+ +2] annulation, we have examined the free energy profileo f the reactionb etween acceptor 1a and allenoate 4p with trimethylphosphine as the catalyst( Scheme 3). The calculations were carried out at the M06-2X-D3/6-311 + G**//B3LYP-D3/6-31G* level of theory,w hich included ac ontinuumd escription of tetrahydrofuran as the solvent. [17,18] All potential mechanismsf or the formation of 5p from 1a and 4p were investigated. [18] The mechanistic sequence of the predicted most favored pathway is shown in Scheme3.T he first step involves formation of ylide Z by the addition of PMe 3 to allenoate 4p.T his step is only slightly endergonic (1.9 kcal mol À1 )b ut occurs through as ignificant free energy barrier (25.9 kcal mol À1 ). The resulting ylide then undergoesa ddition to 1a to form isoenergetic int1-Z throughal ow-lying transition state (15.0 kcal mol À1 ). Next, migration of the double bond occurst oy ield int2 (four diastereomers are possible and the respective energy values/ranges for the different intermediates and transition states are shown in Scheme 3). Many mechanisms can be envisaged fort his double-bond migration, [18] but our calculations indicate that the mostf avored pathway consists of two successive intramolecular proton transfer reactions that involvet he azlactone moiety.I ntermediate int2 can then undergo ring closure to form zwitterion int3,w hich eventually undergoes rapid elimination to yield product 5p (as diastereomers).
The double intramolecular proton transfer process can potentially lead to four diastereomeric int2 intermediates. Our calculations, however,p redict the stereoselective formation of int1-OH-ZZ and then of int2 with two methyl ester groups in a Z relationship. These two diastereomers can eventually lead to the four possible diastereomers of 5p.T hus, given the predicted irreversibility of the double proton transfer,t he overall diastereoselectivity of the formal [4+ +2] annulation must depend on the double protont ransfer and cyclization processes. The small energy differences between the diastereomeric Scheme2.Investigation of the scope of the PBu 3 -catalyzed [4+ +2] annulation of allenoate 4' and acceptor 1.Reactions were carried out on 0.1-0.25 mmol scale. The 5'/6' ratio was determined by 1 HNMR analysiso fthe crude reaction mixture.Compound 5' was isolatedi neach casea sadiastereomeric mixture afterc olumnc hromatography.Analytically pure samples of single diastereomers of some derivatives were obtained by preparativeH PLC. [16] Asian J. Org. Chem. 2018, 7,1620-1625 www.AsianJOC.org 2018 The Authors. Published by Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim transition states of these processes (see Supporting Information for full data [18] )m ake it difficult to predict the overall diastereoselectivity of the formal [4+ +2] annulation reaction. This complexity in the origin of the stereoselectivity and the computed small energy differencesb etween the variousd iastereomeric pathways are in good agreement with the observed variations in the diastereomeric ratios, according to small structural changes in the substrates (see Scheme2).
Our resultsi ndicatet hat the rate-determining step of the formal [4+ +2] annulation is the formation of the ylide (see Scheme 3). Experimentally,w eo bserved that the nature of the catalysth as ad rastic influence on the reactivity.N on-hindered trialkylphosphines catalyzed the reaction efficiently,w hereas hindered phosphines such as Ph 2 PMe, PPh 3 , P2,a nd P3 led to low or no conversion (see Ta ble 1). In addition, tertiary amines such as DABCO (Table 1e ntry 3) and Et 3 Nw ere shown as inefficient catalysts for this [4+ +2] annulation reaction. To identify the factors responsible for theseo bservations, we explored the first two steps of the process, that is, the formation of the ylide formation and its addition to azlactone 1a in the presence of PPh 3 or NMe 3 as the catalyst. The obtained results reveal that the inefficiency of these catalysts to promote the formal [4+ +2] annulation reaction is mainly from an increase in the endergonicc haracter of the ylide formation( Figure 2). The destabilizing steric interactions between the ylide and PPh 3 and the intrinsic lower stabilization of ammonium ylides (for NMe 3 )r elative to phosphonium ones can be used to explain this trend. [19] In the latter case, another factor also affects the reactivity,a st his ylide is computed to be less nucleophilic than phosphorous ylides (free energy barriert oa ddition is 13.1, 13.4, and 17.3 kcal mol À1 for X = PMe 3 ,P Ph 3 ,a nd NMe 3 ,r espectively),t hereby providing ar ational explanation for the observedt rend in reactivity in the presenceo fd ifferent nucleophilic catalysts (comparewith the results in Ta ble 1).
The formationo fr egioisomer 6p was also computationally investigated. On the basis of our results, the mechanism depicted in Scheme 4m ostly likelya ccountsf or its formation. The mechanism involves the formation of ylide-regio from ylide Z by ap roton transfer.T his ylide can then add to azlactone 1a to yield int1-regio,w hichc an undergo ac yclization to lead to 6p after ap rotontransfer followed by elimination of Scheme3.Computedg lobalm echanistic sequence and free energy (kcal mol À1 relative to reactants) profile. (A rangeo fvaluesf or the free energies of the four diastereomeric pathways are reported.) Figure 2. Influence of the natureoft he catalyst (free energy valuesink cal mol À1 ).
Scheme4.Proposed mechanism for the formation of 6p. Asian J. Org. Chem. 2018, 7,1 620 -1625 www.AsianJOC.org 2018 The Authors. Published by Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim the phosphine. The rate-determining step of the processi s predicted to be the addition step, which has af ree energy barrier of 28.7 kcal mol À1 .O ur calculations, thus, indicate that because of the lower stabilityo ft he ylide-regio relative to ylide Z,t he formationo f6p is less favored than that of 5p (by 2.8 kcal mol À1 ), which is in good agreement with the experimental outcome.

Conclusions
It was shown that a-branched allenoates 4 can undergo formal [4+ +2] annulations with arylidenea zlactones 1 under phosphine catalysis to access highly functionalized spirocyclohexenes 5 in as traightforwardm anner.T his cyclization predominantly proceeds throught he g-addition of the phosphine-activated allenoates, and b'-addition is clearly disfavored. The reaction requires the use of small sterically less-hindered tertiary phosphines and does not proceed in the presence of tertiary amines or sterically hindered phosphines, which unfortunately made an enantioselective protocol not possible. Detailed computational studies support the proposed mechanism and provide ar easonable explanation for the strong influence of the catalysta sw ell as for the preference of the g-addition over a b'-addition reaction.