Diastereoselectivity studies on the photo-activated cycloaddition of 5-(1,2-dioxyethyl)-2(5 H )-furanones to alkenes

A series of 2(5 H )-furanones, bearing a 1,2-dioxyethyl substituent at the γ -carbonyl position, have been prepared and explored as substrates in photochemical reactions with alkenes. Compared to the simpler oxymethyl analogues, the homologation of the side chain is highly beneficial to the antifacial selectivity of the [2+2] cycloadditions. Most reactions occur in synthetically useful yields, giving access to new polyfunctionalized cyclobutane-fused furanones.


Introduction
There are many natural products, which incorporate in their substructure a 2(5H)-furanone subunit. 1 Many of these compounds display a variety of biological activities and have thus attracted the interest of synthetic organic chemists.The furanone ring is also present in some unnatural drugs including antifungal, antibacterial and anti-inflammatory agents.Moreover, several chiral substituted 2(5H)-furanones, which are readily available in enantiomerically pure form from chiral pool materials, have been used as the starting substrates for the preparation of an assortment of targeted compounds of challenging structures with diverse complexity and potential utility.Among them, we focused our attention on several compounds with a polysubstituted cyclobutane framework [2][3][4] and over the years, we have developed enantioselective synthetic approaches to various pheromones [5][6][7][8][9][10] and cyclobutane nucleoside analogues, [11][12][13] some of them built on a 3-oxabicyclo[3.2.1]heptane scaffold.In these syntheses, the cyclobutane core was generated through a photo-activated [2+2] cycloaddition of a 2(5H)furanone derivative to ethylene or another alkene. 14he [2+2] photocycloaddition is in fact one of the processes more extensively applied to generate cyclobutane rings [15][16][17][18] and a critical aspect is the control of the reaction stereochemistry.0][21][22] Among the studied substrates, the pivaloyl derivative 1a displayed the higher antifacial selectivity, although the presence of a vinylic methyl group, 1b, was detrimental to the diastereoselectivity, increasing the amount of the isomer derived from the competitive synfacial pathway.In further studies on the C 2 -symmetric bislactones 4 (Figure 1), we found higher degrees of antifacial selectivity in their photoreactions with ethylene even when methyl groups were attached to the β-carbonyl position, and we observed that the protective groups of the central diol unit had a noticeable influence on the diastereofacial selectivity, which is almost complete with the TMS protection. 23Unfortunately, the elaborated preparation of these lactones restricts their synthetic application.For this reason, we considered of interest to explore the performance of the more accessible 2(5H)-furanones 5, bearing also a 1,2-dioxyethyl unit as the substituent at the γ-carbonyl position, as substrates in photochemical reactions with alkenes.To the best of our knowledge, there is no report dealing with [2+2] photocycloadditions of these oxymethyl homologues of 1.We anticipated that the facial selectivity of their cycloadditions could be significantly improved in respect to that of lactones 1, provided that a favorable combination of steric and electronic factors diminished the accessibility of the syn face.Moreover, lactones 5 were visualized as interesting chiral synthons with good opportunities for subsequent diastereoselective transformations.In this article we describe the preparation of lactones 5a-h and their [2+2] photocycloadditions to alkenes.

Results and Discussion
Benzyl and pivaloyl were chosen as the protective groups of the primary alcohol.Benzyl was selected because it was expected that an aromatic residue could be involved into a beneficial πstacking interaction with the carbon-carbon double bond of the lactone, shielding the syn face more effectively and hence preventing the approach of the alkene.On the other hand, the pivaloyl group had previously displayed the better diasteroselectivity in the former type 1 lactone series.The secondary hydroxyl was either unprotected or derivatized to a sterically demanding group.
Lactone 5a was prepared from (+)-dimethyl L-tartrate by a previously described procedure 24 and it was then converted into the new lactones 5b, 5c and 5d, following standard methodologies.The synthesis of the pivaloyl derivative 5e was accomplished through a similar sequence to that described for 5a (Scheme 2).Thus, (+)-dimethyl L-tartrate, 6, was transformed into 2,3-O-isopropylidene L-threitol (7) by a described procedure which involves acetalization followed by reduction. 25After several attempts of monopivaloylation of the symmetric diol 7, the best regioselectivity was obtained by absorbing the diol over silica gel and treating a suspension of this silica gel in hexane with pivaloyl chloride and pyridine. 26Under these conditions, the starting diol was recovered in part and the monopivaloate 8 was isolated in 41% yield (61% over consumed 7), along with a minor quantitiy of the dipivaloyl derivative (15% yield).The alcohol 8 was then subjected to the Swern oxidation to furnish the corresponding aldehyde, which without any purification was reacted with Ph 3 P=CHCO 2 Me in dry methanol, 27 delivering a 7:1 mixture of the expected (Z)-and (E)-α,β-unsaturated esters.On treatment with methanolic hydrogen chloride, this mixture afforded the targeted lactone 5e in 51% yield for the three steps, along with the (E)-α,β-unsaturated ester 9 in 7% yield.Lactones 5f and 5g were prepared from 5e following standard procedures.Finally, the furanone 5h, bearing a methyl group at the β-carbonyl position, was prepared by treatment of 5f with diazomethane, followed by pyrolysis of the corresponding pyrazoline 10 in refluxing 1,4-dioxane, in overall 55% yield (Scheme 3).For the photochemical study, the furanones 5a-h in acetone solutions saturated with ethylene were irradiated through a Pyrex vessel with a 125W high-pressure mercury lamp at -20 °C (Table 1).The progress of the cycloaddition was monitored by GC and the irradiation was prolonged until complete conversion of the starting furanone.The cycloadducts anti 11a-h and syn 12a-h were then purified through silica gel column chromatography and individually characterized.
The photocycloaddition of 5a to ethylene delivered the two expected cyclobutane diastereomers 11a and 12a in good yield with a very good degree of antifacial selectivity (Table 1, entry 1).Introduction of the bulky pivaloyl group at the secondary alcohol as in the reference substrate 1a diminished the rate of the cycloaddition and did not produce a substantial improvement of the facial discrimination (Table 1, entry 2).This could be in agreement with a greater influence of the expected π stacking interaction with the participation of the benzyl group (Figure 2, A) over the steric barrier exerted by the pivaloyl residue.Nevertheless, the free hydroxyl group in 5a can enable intramolecular hydrogen bonding with the carbonyl oxygen of the lactone (Figure 2, B) that can lead to very efficient hindrance of the syn face.The photoreaction of the benzoyl derivative (entry 3) was faster but occurred with lower facial selectivity.This observation argues against the π stacking hypothesis but a competitive interaction between the two aromatic rings in this particular substrate cannot be totally discarded (Figure 2, C).In agreement with the precedents, the TBDMS derivative 5d (entry 4) displayed the highest diastereoselectivity within this series, delivering exclusively the anti cycloadduct 11d, although in slightly lower yield, despite the complete consumption of the starting furanone.The cycloadduct yields within the primary pivaloyl series of furanones were also good (Table 1, entries 5-7), with very good diastereoselectivities for the diester derivatives 5f and 5g and somewhat lower for the substrate bearing the secondary free alcohol 5e.Apparently, the size of the acyl group does not play a decisive role in the stereochemical outcome of the reaction.We conclude that, in the photocycloadditon to ethylene, the efficiency of the process in all the cases is superior to that previously found for 1a, both in terms of yield and antifacial selectivity.To evaluate the influence of a methyl group attached to the β-carbonyl position on these new substrates, lactone 5h was irradiated under the same conditions (Table 1, entry 8).This reaction delivered the corresponding cycloadducts 11h and 12h in good yield although the anti:syn selectivity decreased, as it has been observed for the parallel process from 1b.The structural elucidation of the new cyclobutanes was supported by NMR analysis of pure isolated samples, including mono-and bidimensional experiments.8][9] Thus, for the anti isomers 11 the J 4,5 values were in the range from 0 to 2.3 Hz, while for the syn isomers 12 oscillate between 5.2 and 5.7 Hz (Table 2).These coupling constants were determined on the signals corresponding to H-4, because in most cases the signal of H-5 overlapped with other cyclobutane protons.Unfortunately, for the TBDMS derivative 11d, the signal of H-4 is also masked by that of the benzyl protons and, hence, J 4,5 could not be determined, but we assumed that the only product isolated from the photocycloaddition of 5d should have the anti configuration.The stereochemical assignment of the isomers 11h/12h, lacking the proton at C-5, was deduced from the 13 C chemical shift of the β-methyl group, which is more sterically compressed in the anti isomer (17.5 ppm) compared to the syn (22.0 ppm).
Previously, we developed an alternative entry to fused cyclobutane furanones that avoided the use of ethylene 28 and we decided to explore this option also on the new substrates.As such, lactones 5a and 5e were irradiated in acetonitrile solutions containing an excess of (Z)-1,2dichloroethylene and, without isolating the individual isomers, the product mixture of dichlorocyclobutanes was reduced by treatment with tributyltin hydride and AIBN in THF (Scheme 4).This protocol applied to 5e gave a 7:1 mixture of 11e and 12e in 67% overall yield.Hence, the antifacial selectivity is similar to that obtained with ethylene from the same substrate (Table 1, entry 5) and, although the total yield is slightly lower, this procedure may have practical advantages when working on a larger scale.Unfortunately, the same protocol applied to 5a did not lead to any identifiable products.To broaden the synthetic applicability of the cycloadducts, we investigated the photocycloaddition of 5b and 5f to 1,1-diethoxyethylene (Scheme 5).In contrast to the previous reactions, this cycloaddition is amenable to produce regioisomers, depending on the head to tail (HT) or head to head (HH) orientation of the two reagents.Moreover, each orientation may occur, as before, through an anti-or syn-facial approach, overall producing up to four isomers.In previous studies with 1a and other similar lactones, we observed that, compared to acetonitrile, the less polar solvents favored the HT regioisomer and decreased the antifacial selectivity. 29The photocycloadditions of lactones 5b and 5f were assayed in acetonitrile, diethyl ether and hexane, in the presence of an excess of 1,1-diethoxyethylene (Table 3).In all these solvents, the irradiation of 5b was completely regioselective, furnishing exclusively the HT adducts 13b and 14b in moderate yields (Table 3, entries 1-3).Conversely, in the irradiation of lactone 5f under identical conditions (Table 3, entries 4-6), the HH-anti cycloadduct 15f was detected as a minor product, with the higher proportion in acetonitrile as the solvent, as expected.The antifacial selectivity was quite similar in all the cases, not being influenced by the solvent polarity.a Yield of isolated product as a mixture of stereoisomers after column chromatography purification.b Isomer ratio from GC analysis of the isolated mixture of products.

Conclusions
A series of 2(5H)-furanones 5 bearing a 1,2-dioxyethyl unit as the substituent at the γ-carbonyl position were prepared and explored as substrates in photochemical reactions with alkenes compared to the simpler oxymethyl analogues 1.The additional oxymethyl fragment was highly beneficial to the antifacial selectivity of the [2+2] cycloadditions to ethylene, reaching in most cases diastereomeric excesses around 80%. Furthermore, (Z)-dichloroethylene, as a solid surrogate of ethylene, was also used to prepare the same cyclobutane products in a more practical way.The photoreaction of lactone 5b to 1,1-diethoxyethylene showed a complete regioselectivity towards the head to tail orientation, independently of the solvent, while that of 5f was less regioselective and solvent dependent.For this alkene, the antifacial selectivity was lower than that observed for ethylene.Most reactions occurred in synthetically useful yields, giving access to new polyfunctionalized cyclobutane-fused furanones that may be further elaborated to natural or unnatural cyclobutanes of interest.

Experimental Section
General.Unless otherwise noted, analytical grade solvents and commercially available reagents were used without further purification.Solvents were purified and dried by standard procedures.The solutions were concentrated using a rotary evaporator at 15-20 Torr.Analytical thin layer chromatography (TLC) was performed on 0.25 mm silica gel 60-F plates and visualized by ultraviolet irradiation and KMnO 4 stains.Gas chromatography (GC) analysis was performed using a cross-linked capillary column with 5% dimethylsilicone.Flash column chromatography (FCC) was carried out on silica gel (230-400 mesh).Melting points were determined at the hot stage and are uncorrected.Optical rotations were measured on a Propol Automatisches Dr Kermchem polarimeter.General procedure for the photocycloadditions of 2(5H)-furanones to alkenes.Irradiations were performed in a conventional photochemical reactor (two-necked vessel fitted with a Pyrex or quartz immersion-type cooling jacket) using a medium-pressure, 125 W mercury lamp.Methanol at -15 ºC was used for the refrigeration of the immersion well jacket.The vessel was externally cooled at -20 ºC with a dry ice/CCl 4 bath.The progress of the reaction was monitored by GC analysis of aliquot samples.For the reactions with ethylene, this gas was bubbled through the solution for 15 min before turning the lamp on and a slow flow of ethylene was maintained throughout the irradiation.

Table 2 .
Significant NMR data of compounds 11and 12