The stereochemistry of addition of allyl sulfone carbanions to aldehydes. Formation of dihydrofurans 1

The reaction of the mono-anion of the bromoallyl sulfone 1 with aldehydes 2 was examined with the aim of obtaining selectively substituted tetrahydrofurans. At –100ºC syn - and anti - open chain adducts 3 and 4 were isolated together with a low yield of 4-methylene-2,3-disubstituted tetrahydrofuran 5 . In the presence of HMPA or at higher temperature the reaction led to formation of 2,5-dihydrofurans 6 . The stereochemical results are consistent with initial addition of 1 to the aldehyde involving Li ion chelation.


Introduction
Recently we have shown 2,3 that 2-(bromomethyl)-3-phenylsulfonyl-1-propene 1 reacts with one molar equivalent of LDA to generate a lithiated α-allylsulfone carbanion that is stable at low temperature and undergoes regioselective and stereoselective additions via the α-carbon to Michael acceptors such as unsaturated esters, ketones, sulfones and nitro compounds, followed by cyclization to methylene-cyclopentenes.For instance, in the case of unsaturated esters, the primary Michael adduct was not even isolable at -78ºC and the incipient carbanion immediately underwent intramolecular reaction with the allylic bromide to afford stereoselectively substituted methylenecyclopentane derivatives (eq.1). 2 The high stereoselectivity in the product can be rationalized on the basis of Li-ion chelation by the sulfone and the ester function during addition.
The addition of lithiated 1 to the C=N of sulfinimines proceeded less stereoselectively and led with double bond rearrangement to 2-arylpyrrolines.3c Br PhSO 2 eq. 1  LDA In view of general interest in formation of stereoselectively substituted tetrahydrofurans, such as the naturally occurring polyether antibiotics, 4 we investigated the reaction of 1 with aldehydes as a potential entry into substituted tetrahydrofurans.We report here our findings, using the mono-lithio derivative of the bromoallyl sulfone 1 and of the hydroxyallylsulfone 8 in reactions with aldehydes 2.

Results and Discussion
The bromoallylsulfone 1 was deprotonated using 1.1 equiv. of LDA in THF at -100ºC.The monolithio derivative thus generated was treated with p-nitrobenzaldehyde 2d at the same temperature and quenched after 30 min with HOAc to give a mixture of two open chain synand antiisomers, 5 3 and 4, as well as the trans-2,3-disubstituted-4-methylenetetrahydrofuran 5 (see eq. 2 and Table 1, entry 2).

(anti) 4
The synand antiisomers 3 and 4 were separated by flash-column chromatography and their structures in the preferred conformation were assigned as 3A and 4A, respectively, on the basis of NMR data.Similarly, the structure assignment of 5 is based on NMR (see discussion below).Addition of LiBr to the reaction mixture did not alter the results: however, addition of a solvating agent (TMEDA or especially HMPA) at -100ºC did change the ratio of products and led to isolation of the dihydrofuran 6d, with the syn isomer 3 remaining unchanged (see Table 1).In two cases, 7 a Michael adduct of the anion derived from product 5 to the aldehyde 2d, was isolated in ca.5% yield.

a-d
Reaction at higher temperature (-40ºC) for a longer period of time (3 h) in the absence of HMPA afforded the syn-isomer 3 together with rearranged dihydrofuran 6d.When the monolithio derivative of 1 was treated with aldehyde 2d followed by HMPA in excess, and allowed to react at -40ºC for 3 h, the only product isolated was the dihydrofuran 6d (see Table 1).reported the formation of the dianion of the chloro analog of 1 using 2 equiv. of BuLi and DMPU and its reaction with alkyl halides and aldehydes (eq.3).With alkyl halides, dialkylation took place.Allowing the reaction with aldehydes to proceed to room temperature, led to dihydrofuran analogs of 6, while the reaction with propanal and quenching at -40ºC gave a mixture of 6 and open-chain diastereomeric adducts.However, no separation of the latter, or their stereochemical elucidation, was reported.

eq. 3
Cl Our stereochemical results are best explained via the "cyclic model" in which the lithio allylsulfone carbanion reacts by an α-carbon attack on the aldehyde leading to a chelated intermediate.As can be seen, the syn-chelated structure 3B (likely as a chair conformer of a 6membered ring with two equatorial groups) should be favored over syn-3C or over anti-4A 7 because of gauche interactions; this explains why the syn product predominates.When chelation of the Li cation with sulfone-and alcohol-oxygens is disrupted, as in the presence of HMPA, rotation can occur to 3C and 4B, respectively, 7 conformers needed for subsequent cyclization.In this case, it is the anti-isomer in its anionic form (4B or 4C) which is expected to ring-close to a tetrahydrofuran 5, while steric hindrance inherent in vicinally cis- disubstituted five-membered rings should raise the transition state energy for the cyclization of syn-3C.Indeed, only cyclization of the antiisomer to the transtetrahydrofuran 5 was observed at -100ºC (Table 1, entry 4).At higher temperature and in the presence of HMPA and base, ring closure of both diastereomers took place followed by isomerisation to the dihydrofuran 6, as well as side reactions.The above series of events is consistent with the products which we had observed in the reaction of 1 with ω-nitrostyrene, in which the anti-adduct ring-closed more readily (to the trans-substituted cyclopentane) than did the synisomer.3b It was hoped that chelation to the hydroxyallylsulfone 8 might lead to higher stereoselectivity during the addition to the carbonyl group by complexation, as shown in 9.In fact, the hydroxyallylsulfone 8 had been shown to add regio-and stereoselectively to nitro-olefins. 8ence, we also examined the condensation of the Li derivative of 8 with hexanal 2b.In this case, however, two diastereomers 10 (syn-) and 11 (anti-) were formed in essentially equal amounts and addition of TMEDA did not change the results.Apparently, in this system chelation of the Li cation from the aldehyde to the sulfone or the hydroxy group are either not important, or more likely, equal.

Structure assignment
The configurational assignment of 3A as the syn-isomer is based on a value of 9 Hz for the coupling constant between H a and H b indicating an antirelationship.In the second isomer 4A the corresponding coupling constant is 2.5 Hz, consistent with a gauche orientation of H a and H b .
In both isomers the hydroxylic proton exhibits a doublet with coupling to H b of 2 Hz.This phenomenon is known in cases where the exchange rate of the hydroxylic proton is very low and may be due to hydrogen bonding.The determination of configuration of the trans-substituted tetrahydrofuran 5 is also based on a 4 Hz coupling between H a and H b .Although configurational assignment in 5-membered rings based on NMR should be made with caution in cyclopentanes that are rather flat, trans-vicinal hydrogens usually show a small coupling constant of 2-4 Hz (dihedral angle near 120º) while cis-vicinal hydrogens exhibit a larger coupling constant of 8-11 Hz (dihedral angle closer to 0º).In the case of 5 the assumption that the ring is rather flat is well founded since (a) there is an sp 2 hybridized center in the ring, (b) the 5-membered ring contains an oxygen atom, (c) each of the vinylic protons shows three similar allylic couplings (with H a , H c , H d ).This is consistent with having C 3 , C 4 , C 5 in nearly the same plane.More convincing is the NOE (2%) observed in 5 between H a and the orthohydrogens of the cis-p-nitrophenyl ring, as well as between H a and H d and between H b and H c .
In conclusion, the lithio allylsulfone carbanion derived from 1 added stereoselectively to aldehydes to produce preferentially the syn-adduct at low temperature (-100ºC).This is best explained by a chelated intermediate 3B, while at higher temperature and in the presence of HMPA, ring-closure and finally isomerisation to the dihydrofuran 6 took place.There was no stereochemical preference in the addition of hydroxysulfone 8 to aldehyde 2b.

Experimental Section
General anhydrous experimental techniques and analytical measurements were as previously described. 8

General procedure for the reaction of 1 with p-nitrobenzaldehyde (2d); Formation of 3-7
To a stirred solution of LDA (prepared from 0.14 mL (1 mmol) of diisopropylamine and 0.64 mL of n-BuLi (0.92 mmol, 1.475 N in hexane) in 4 mL of THF) was added dropwise at -100ºC a solution of 1 (200 mg, 0.72 mmol) in 1 mL of THF.After stirring for 10 min at the above temperature, p-nitrobenzaldehyde (121 mg, 0.08 mmol) in 1 mL of THF was added dropwise.After 15 min, at -100ºC, the reaction mixture was quenched with aqueous (20%) AcOH, poured into water, and extracted with CH 2 Cl 2 .The extracts were washed successively with saturated NaHCO 3 solution and brine, dried (MgSO 4 ), and evaporated under reduced pressure.Chromatographic purification of the residue (ether/petroleum ether, 1:2) gave compounds 3-7, as viscous oils.Yields are reported in Table 1.

Table 2 . Dihydrofurans 6 from reaction of 1 with
While this work was in progress, Najera et al.