Reaction with other carbon nucleophiles

The formation of C-C bonds is one of the most important objectives in organic chemistry.

Organometallic reagents such as the Grignard reagent are nucleophilic due to the polarisation of the metal carbon bond. Grignard reagents are often reacted with aldehydes and ketones to form secondary and tertiary alcohols. An examination of 119a shows that a Grignard reagent could react at three possible sites, at the cyclopropane ring, the ketone group and the ester functionality.

Simple alkyl Grignard reagents, butyl and ethyl, react only at the cyclopropane ring. No products from reaction with the ketone or the ester functionality were recovered. Treatment of 119a with ethyl magnesium bromide led to a single diastereoisomer which was assigned the anti configuration, due to ³J(1’-H/2’-H) > 10Hz. This finding can be explained by assuming that the cyclopropane ring undergoes partial opening followed by attack from the hard nucleophilic Grignard reagent onto a carbenium ionic transition state 148 from underneath.

O O

O

119a

O O

O H

O O

OH RMgBr R

149 149a R = Et, 61% yield

149b R = Bu, 42% yield

Ph

MgBr δ+

δ-R

-148

Ring opening of 3,5-dispirodihydrofuran-4,12-diones with nucleophiles 4 8

Experiments were undertaken with more complicated Grignard reagents, however each attempt was met with failure.^[150] Attempts with benzyl magnesium bromide, cyclopropane magnesium bromide and isopropyl magnesium bromide gave no reaction at all. In each case the starting material could be recovered and no products arising from attack on the ketone or ester functionality could be detected. The use of allyl magnesium bromide however led to 122a. Grignard reagents are basic compounds and we suspect that in the case of allyl magnesium bromide the Grignard reagent has deprotonated 119a leading to a retro oxa-ene reaction back to 122a.^[151]

O O

O H₂C

H

retro oxa-ene reaction

O O

OH

119a 122a

MgBr

Organolithium reagents are highly reactive hard nucleophiles and it was suspected that organolithium compounds would react at the hard electrophilic ketone group. The reaction of methyl and butyl lithium with a sample of 119a-βββββ led to 150 in high yields. Only one diastereoisomer was observed suggesting that the phenyl ring prevents addition of the alkyl lithium reagent from one face of the molecule. No products arising from attack on the cyclopropane ring or the ester functionality were detected. The reaction of 119a-βββββ with phenyl lithium led to an incomplete reaction. NMR spectroscopy of the mixture showed the presence of the desired product mixed with the starting material. These two compounds proved impossible to separate by normal chromatography. Despite the expected change in polarity these compounds exhibit almost identical R_f values in a range of solvent mixtures. An attempt was also made to react 119c with organolithium reagents; the result was a complicated mixture of products, probably arising from a lithium chlorine exchange which subsequently led to undefined reactions.

O O

O

RLi, THF, 16h, rt

O O

OH R

119a 150

150a R = Me, 83% yield 150b R = n-Bu, 65% yield

150 is a stable molecule and can be handled with ease under ambient conditions. Theoretically it should be possible to react 150 with an acidic compound; protonation of the hydroxy group, a rearrangement and elimination of water should lead to 4-alkyl-α,β-butenolides of type 151. This reaction was attempted with HBF₄ in THF which led to complete decomposition, probably as a result of using such a strong acid. The use of HBF₄ with more suitable solvents such as DCM,

or toluene, and the use of milder acids or dehydrating agents have not yet been attempted.

Another possiblity would be the conversion of the alcohol functionality to a tosylate group (an excellent leaving group); such systems should be amenable to ring opening reactions to generate butenolides.

O O

OH

150a

H⁺

O O

H₂C

OH₂ R H

O O

rt, 16h HBF₄, THF

-H₂O

151

Organocopper compounds are well known to undergo nucleophilic displacements with halides and sulfonates. They also react with epoxides, add to alkynes and undergo conjugate additions to α,β-unsaturated systems.^[152-155] Organocuprates were therefore suitable candidates for reaction with 119a. The simple dimethyl lithium cuprate 152 was formed from copper bromide and methyl lithium. Solutions of 152 were formed at -10^oC upon which a solution of 119a was introduced. Unfortunately no reaction was observed even when temperatures were raised to 0^oC and later raising the solution to room temperature gave no reaction. Addition to the cyclopropane probably requires higher reaction temperatures however unfortunately simple organocuprates are not normally stable above room temperature. More stable mixed cuprates obtained from alkyllithium compounds and copper cyanide were not tested as these mixed cuprates are often not as reactive. It should be noted that compounds of type 153 have been synthesised by other members of our working group starting from 119a by using Knochel and Normant type cuprates.^[147,197]

O O

O

Me₂CuLi 152, DCM -10^oC - rt

O O

OH

119a 153

The Baylis-Hillman reaction continues to be an interesting carbon-carbon bond forming reaction, with the reaction between α,β-unsaturated esters and aldehydes, utilising tertiary amines or phosphines as catalysts, leading to densely functionalised molecules.^[156-158] Examination of the Baylis-Hillman mechanism leads to the possibility of an interesting analogous reaction with 119a.

The first step of the Baylis-Hillman reaction normally involves the conjugate addition of the nucleophilic reagent DABCO 154 to an α,β-unsaturated ester e.g. ethyl acrylate 155. This results in an enolate 156 which is normally reacted with an aldehyde. We proposed to replace the aldehyde with 119a which we expected to react by a similar mechanism resulting in 157

Ring opening of 3,5-dispirodihydrofuran-4,12-diones with nucleophiles 5 0

which should subsequently eliminate DABCO 154 leading to interesting compounds of type 158.

One of the main disadvantages of the Baylis-Hillman reaction is that it normally requires several days of reaction time. Many examples exist in the literature for acceleration of this reaction. We therefore performed this reaction under ultrasound conditions in DCM. The temperature was kept between 35 - 40^oC for 72h with no reaction having taken place.The reaction was once again performed this time utilising microwave conditions which in recent years has been shown to vastly accelerate the reaction. Heating of DCM solutions in a sealed vessel placed in a 300 W focussed single-mode microwave reactor (CEM) at 100 °C for 1 h led to recovery after column chromatography of 55% of 119a with ~15% of compound 122a also recovered. No products corresponding to 158 were recovered. 122a was probably produced due to the basic nature of the DABCO catalyst leading to a reverse oxa-ene rearrangement already discussed on Page 48.