Enantioselective Catalytic [4+1]‐Cyclization of ortho‐Hydroxy‐para‐Quinone Methides with Allenoates

Abstract The first highly asymmetric catalytic synthesis of densely functionalized dihydrobenzofurans is reported, which starts from ortho‐hydroxy‐containing para‐quinone methides. The reaction relies on an unprecedented formal [4+1]‐annulation of these quinone methides with allenoates in the presence of a commercially available chiral phosphine catalyst. The chiral dihydrobenzofurans were obtained as single diastereomers in yields up to 90 % and with enantiomeric ratios up to 95:5.


Results and Discussion
Initial optimization of the racemicreaction We startedo ur investigations by carrying out the racemic reaction between p-QM 3a and diethyla llenoate 4a in the presence of PPh 3 (Table 1g ives an overview of the most significant screening results). Our first reactionsw ere carried out in analogy to the conditions developed for the annulation of o-QMs 2 [8] (pleasen otet hat we previously used at wofold excesso f the quinonem ethide 2 to compensate for its competing decomposition under the reaction conditions). Gratifyingly the targeted dihydrobenzofuran 5a could be obtained as as ingle diastereomer in this initial attempt already (entry 1). The relative configuration of the product 5a was confirmed by NOESY experiments( as shown in Scheme 4) andw ealso observed the same correlations for other products 5 later (Scheme2). As the reactionw as found to be rather slow,w ith significant amounts of unreacted 3a being recovered (indicating its increased stability compared to o-QMs 2), we next increased the amounto f base (entry 2), which however had ad etrimental effect (complete decomposition of startingm aterials).
As decomposition of the acceptor 3a was not very fast in the first attempts with 2equivalents of base, we next used an excess of allenoate 4a,w hich led to am easurablei ncrease in yield (entry 3). The screening of differents olvents revealed that toluene allows for as lightly highery ield (entry 4), but also accompanied with am ore pronounced formation of variousn ot identified side-or decomposition products. Other solvents did not give satisfactory results (see entry 5f or one example) and so further optimizations with CH 2 Cl 2 were carried out. Very interestingly,l owering the amount of base (entry 6) significantly improved the yield and suppressed side product formation. By testingo ther bases, K 2 CO 3 turned out to be the mostp romising (entry 7). It should be noted that other simple trialkylphosphinesw ere tested as well, [15] but in analogy to our previous observation [8] these did not allow for this [4+ +1]-annulation.
With these first high yielding conditions set, we next lowered the amount of PPh 3 .G ratifyingly,a nd in sharp contrast to the reactionw ith o-QMs 2, [8] the use of 20 mol %P Ph 3 allowed for the same yield as when using as toichiometric amount (comparee ntries 7a nd 8). Further lowering of the catalyst amount unfortunately slowed down the reactionm easurably (entry 9). Considering the beneficial effecto fu sing less base when using Cs 2 CO 3 (entry 6), we finally also loweredt he amount of K 2 CO 3 (entries 10, 11), and much to our surprise the reaction proceeded well even without any base (entry 11;t he reaction was reproduced several times on differents cales and also on 1mmolscale).

Development of an asymmetric catalyticprotocol
Havinge stablished high yielding and robustc atalytic procedures for the racemic synthesis of 5a we next focusedo nt he use of chiral phosphine catalysts. As already mentioned before, we weren ot able to identify as uited asymmetric catalystf or our previous [4+ +1]-annulation of o-QMs 2.H owever,g iven the fact that p-QM 3a performed very well in the racemic reaction and also allowed for ac atalytic approach, we were confident that the well-described bulky chiral phosphines A [16] or B [17] may allow for at ruly catalytic enantioselective protocol (Table 2). We first used the binaphthyl-based phosphines A1-3, but unfortunately neither of them allowed for anyp roduct formation (entries 1-3). Gratifyingly however,b ys witching to the commercially available chiral spiro phosphine B ((R)-SITCP) [17] we observed av ery clean and reasonably enantioselective product formation when using 20 mol %o ft his catalyst under base-free conditions in CH 2 Cl 2 (entry 4). Lowering the reaction temperature to 0 8Cu nfortunately did not allow for product formation anymore (entry 5). When carrying out the reactioni n the presence of two equivalentso fK 2 CO 3 ,t he outcomew as only slightly affected in this solvent (entry 6). Interestingly,w hen changing to toluene (other solvents like THF were found to be not suited), we were able to improve the enantioselectivity significantly (entries 7-9). At room temperaturer eactions in the presence of K 2 CO 3 as well as under base-free conditions performed very similarly,w ith as lightly higher e.r.i nt he absence of base (comparee ntries 7a nd 8). However,w hen we furtheri nvestigated the application scope, we realized that the base-mediated conditions werem ore robust when using differently substituted starting materials 3, while not all of those allowed for good conversionsu nder base-free conditions. Other bases were found to be less satisfactory (with for example, Cs 2 CO 3 giving lower yields and K 3 PO 4 giving no product at all). We thust ested if any further improvement in the presence of 2equivalents of K 2 CO 3 may be possible (entries 8-11). However,l oweringo ft he reaction temperature was possible to somee xtent only (entries 9, 10), but reducing the catalystloadingto10mol %was unfortunately not possible anymore (entry 11). Accordingly, the best-suited and most robust catalytic enantioselective approach to access 5a as as ingle diastereomer was to carry out the reaction in toluene at 10 8Ci nt he presence of 2equivalents of K 2 CO 3 by using 20 mol %o fthe commerciallya vailable phosphine catalyst B (entry 9, the reaction was reproducedb yd ifferent persons on 0.05-0.1mmol 3a scale giving identicalresults).

Application scope
Having established ah igh yieldinga nd robustc atalytic procedure for the synthesiso fd ihydrobenzofuran 5a,w en ext tested the use of differently substituted quinonem ethides 3 and allenoates 4 (Scheme 2).
First, we could show that replacemento fo ne of the allenoate ethyl ester groups for ab enzyl ester was tolerated very well (see product 5b). Then it turned out that ad imethylbased p-QM 3 can be used as well to obtain the enantioenriched product 5c (albeit with as lightly lower selectivity than for the parent tBu-based 5a). Interestingly,s ubstituents in the 5a nd 6-position of the benzofuran backbonew ere very well tolerated (see compounds 5d-f, 5i-k, 5m). In contrast, substituentsi np ositions 4a nd 8t urned out to be more limiting and product 5g was only accessible with ar ather low enantiomeric ratio of 79:21. Surprisingly,c ompound 5h was not formed at all under the asymmetric conditions (even with longerr eaction times). We were however able to obtain racemic 5h in high yield when using PPh 3 as an achiral catalyst. Very interestingly,w hile we found initially that benzyle ster containing allenoates were tolerated similarly well as ethyl ester-based ones (see targets 5a and 5b), we found that tertbutyl esters resulted in somewhat lower enantiomeric ratios compared to ethyl and benzyl esters (compare 5k and 5l as well as 5m, 5n,a nd 5o). All asymmetric reactions were initially carried out on 0.05 mmol scale of the limiting agent 3 and we also reproduced selected reactions on up to 0.2 mmol scale withouta ffecting the outcome, thus indicating that the asymmetric procedure is of similar robustness as the racemic one (Table 1, entry 11). All substrate combinations gave the (+ +)-enantiomer as the major product,b ut unfortunately,i th as not been possible to obtain suitable crystals of any of the products 5 to determine the absolute configuration by single-crystal X-ray analysis.
It has been described by others that the tert-butyl groups of the phenol derivatives obtained by addition of nucleophilest o QMs can be cleaved off under (Lewis) acidic conditions. [7a, 11c] We thuscarried out afew (unoptimized) test reactions to see if as imilar debutylation is also possible on the highly functionalized diester-containing dihydrobenzofuranes 5 (Scheme 3). Carrying out the reaction at elevated temperature only led to decomposition. In contrast, at room temperature, the slow formation of the debutylated diester 6a was observed by MS. Interestinglyh owever,t he major product was found to be the debutylated monoester 7a that was formed in around 30 %a fter one day and around5 0-60 %a fter 3days (accompanied with some decomposition products)a nd which could also be isolated after columnc hromatography (NMR clearly confirmed that the ester group on the stereogenic center was hydrolyzed). It should be noted that no further attempts to optimize this reaction were undertaken, but this result clearly shows that the highly functionalized compounds 5a can be used for further transformations and that the two ester groups have different reactivities.

Mechanistic considerations
Mechanistically this is ar ather complexr eaction and it should be admitted thats of ar,w eo nly have some hints that may allow us to postulate the mechanistic scenariod epicted in Scheme 4. This proposal is also based on our recent observations made for the racemic [4+ +1]-annulation of o-QMs 2 where we found that intramolecular rapid protont ransfers are crucial to explain the outcome of this [4+ +1]-cyclization. [8] Addition of the phosphine to the allenoate is supposed to give the required zwitterion I after proton transfer on the primary addition product. Following the reactionb etween PPh 3 and 4a by 31 PNMR shows the appearance of two new signals around2 7ppm (the parentP Ph 3 peak is at À5ppm) substantiating the formation of alkylated phosphine species (these addition productsd ecomposed very quickly in the absence of any electrophile). Upon addition of the quinonem ethide 3a immediatelyastrong red color evolves, which can be rationalized by the 1,6-addition of I to 3a to give the phenolate II.S imilar color changes can also be observed when adding different nucleophiles to other p-QMs,s ubstantiating the assumed initial 1,6-addition. With respectt ot he nature of the electrophile 3a one could however also postulate that ap rototropic shift from the phenol to the para-QM moiety gives an ortho-QM in situ, which then reacts with I to give III directly. [18] However,w e found compound 3a being rather stable under basic conditions and we never observed any other species or got any experimental hint that supports this pathway, but it should not be ruled out completely with the current state of knowledge. The phenolate II then needs to undergo twop roton transfer reactions towards the betaine IV,w hich can then finally react to the product 5a via an Sn2'-type cyclization.W eh aver ecently shown for the cyclizations of o-QMs 2 that these proton transfers are rather likely processes and we reason that the presence of ab ase is beneficialf or these reactions, which would be an explanation why the herein presented [4+ +1]-cyclization is more robust under basic conditions. This beneficial effect of base became especially pronounced in those cases where no electron-donating ring substituent para to the OH group is present (these reactions usually proceeded ab it slower as well). This observation supports as cenario where the final ring closure may be the rate-determining step, which also rationalizes why slightly larger amounts of catalyst were necessary to obtain satisfying catalyst turnover.
With respectt ot he observed stereoselectivity it is likely that the catalyst controlst he absolutec onfigurationi nt he 1,6-addition step. An alternative may be al ess selective1 ,6-addition followed by base-mediated isomerization of the benzylic position on one of the chiral catalyst-bound intermediates II or III. However,a st he observed enantioselectivity wasm ore or less the same under basic and base-free conditions (compare with Ta ble 2), this option seems less likely.T he diastereoselectivity is then controlled in the final proton transfer-cyclization sequence. Given the fact that S N 2' reactions usually proceed with a cis-orientation of nucleophile and leaving group [19] the proton transfer towards IV is supposed to be highly selective, and may be steered by electrostatic attraction between the phenolate anion andt he phosphoniumc ation in the nonpolar reactionsolvent. However,its hould clearly be pointed out that this is just am echanistic hypothesis and although we were able to observe the presence of some alkylated phosphonium speciesb y 31 PNMR during the reaction, noneo ft hese intermediates could be isolatedo rm ore carefullyanalyzed.

Conclusions
The first highly asymmetric catalytic formal [4+ +1]-annulation of o-hydroxy-p-quinonem ethides 3 with allenoates 4 has been developed. The outcome of this reaction is in sharp contrastt o other recently reported reactions between quinone methides 3 and allenoates. [12] Key to success was the use of the commercially availablec hiral phosphine B as ac atalystu nder carefully optimized reactionc onditions. This methodology allowed for the so far unprecedented synthesis of the chiral dihydrobenzofurans 5 as single diastereomers in yields up to 90 %a nd with enantiomeric ratios up to 95:5.

Experimental Section
General details can be found in the online Supporting Information. This document also contains detailed synthesis procedures and analytical data of novel compounds and reaction products as well as copies of NMR spectra and HPLC traces.

General asymmetric [4+ +1]-cyclization procedure
Am ixture of the para-quinone methide 3 (0.05-0.2 mmol), K 2 CO 3 (2 equiv), and chiral phosphine B (20 mol %) was cooled to 10 8C and as olution of the allenoate 4 (2 equiv) in dry toluene (20 mL per mmol 4)w as added. The resulting mixture was stirred at 10 8C under an Ar atmosphere for approximately 20 h. The mixture was diluted by adding CH 2 Cl 2 (5 mL), filtrated over ap ad of Na 2 SO 4 and the residue was rinsed with CH 2 Cl 2 (5 5mL). The combined organic layers were evaporated to dryness (under reduce pressure) and the products were purified by silica gel column chromatography (gradient of heptanes and EtOAc) giving the corresponding dihydrobenzofurans 5 in the reported yields and enantiopurities (Syntheses of racemic samples were carried out in analogy using PPh 3 instead).