Novel Conversions of a Multifunctional, Bio-sourced Lactone Carboxylic Acid

The plant-derived compounds furfuryl alcohol and itaconic anhydride are known to undergo a Diels-Alder reaction at room temperature and in bulk to efficiently give an alkene-containing lactone carboxylic acid. Reported here is the conversion of this substance to a variety of derivatives via hydrogenation, epoxidation, or halolactonization reactions. Most notable is the formation of a set of three related acrylate or methacrylate esters (see graphical abstract) produced by direct acylative ring opening of ether bonds using Sc(OTf)3 and (meth)acrylic anhydride. These esters are viewed as promising candidates for use as biorenewable monomers in reversible addition-fragmentation chain transfer (RAFT) polymerization reactions.


Introduction
Professor Samir Zard's impressive pioneering advances in and contributions to numerous aspects of radical chemistry (among other subjects) are renowned.Perhaps at the top of the Zard team's accomplishments are his and his coworkers' insightful contributions 1,2 to the underpinnings 3,4 of reversible addition-fragmentation chain transfer (RAFT) polymerization.6][7] Among the first examples of very effective RAFT monomers [1][2][3][4] were acrylate and methacrylate esters, a class of monomers still widely used.
][10][11] With an eye toward developing an efficient method for achieving novel bio-sourced monomers, we previously reported a method for the efficient production of the lactone acid 1 (Figure 1). 12,13Researchers at York University, carrying out concurrent independent studies, made a similar discovery. 14The process involves merely mixing furfuryl alcohol (2; sourced, e.g., from hemicelluloses) with itaconic anhydride (3; sourced, e.g., from starches) in the absence of solvent at ambient temperature.We determined that the metastable intermediates produced by a number of competing, yet reversible, Diels-Alder (DA) cycloadditions (giving the four isomeric 1:1 DA adducts 4) were driven thermodynamically to the predominant product 1 in 94% yield by its selective intramolecular anhydride ringopening and crystallization within the bulk mixture.[16] Figure 1.The known Diels-Alder reaction of a neat mixture of itaconic anhydride (3) and furfuryl alcohol (2) at room temperature [12][13][14] proceeds by reversible formation of the four isomeric anhydrides 4a-d, one of which selectively lactonizes to produce the crystalline lactone acid 1. 12,13 We first explored Bronsted acid-catalyzed reactions of 1 (Figure 2a).Treatment with 50 mol% of triflic acid in refluxing chloroform for 30 minutes gave rise to a mixture of the dilactone 5 and isochromanone (6).This reaction was always accompanied by formation of a considerable amount of a dark-colored resinous substance suggestive of highly conjugated oligomeric material.In contrast, purified 5 could be further (and quite cleanly) converted into 6 under the same conditions with very little darkening of the reaction mixture.Generation of 5 can be rationalized in straightforward fashion by the acid-catalyzed ring-opening of the strained allylic ether in 1 (cf.7a) followed by dehydration of the resulting allylic alcohol 7b.The further conversion of dilactone 5 to 6 is more unusual; the elements of CO2 are lost in the process.This can be rationalized by the acid-catalyzed opening of the protonated lactone 7c to the pentadienyl cation 7d, which can then fragment to the acylium ion 7e in an event driven by the aromatization of the benzene ring.Loss of carbon dioxide and the proton furnishes isochromanone (6).
We attempted, unsuccessfully, to achieve ring-opening transesterification polymerization (ROTEP) of 6.This is consistent with the thermodynamic reluctance of 6 to undergo methanolysis in 2 vol% methanolic chloroform, a proxy we recently reported for evaluating the suitability of any lactone to function as a ROTEP monomer. 17Only 6.5% of the methyl ester 8 was formed at equilibrium; for comparison, -butyrolactone (another reluctant ROTEP monomer) and -valerolactone (a competent ROTEP monomer) produced their corresponding methyl esters to the extent of 18.7% and 86.8%, respectively, in the same assay. 17 The alkene in 1 could be smoothly hydrogenated to the saturated derivative 9 (Figure 2b); no evidence of hydrogenolysis of an allylic C-O bond was seen.Triflic acid treatment of 9 also led to formation of a dilactone followed by dehydration to produce the isomeric alkenes 10a and 10b in an equilibrium ratio of ca.1:1.As anticipated, the rate of the ring-opening reaction of the saturated bicyclic ether 9 was considerably slower (in situ 1 H NMR) than that of the allylic analog 1.
We next explored reactions of the alkene 1 with electrophilic reagents (Figure 3).Predictably, epoxidation using peracetic acid led to efficient formation of the exo-epoxide 11.We initially used in situ generated performic acid at 60 °C19 to effect this epoxidation, but on one occasion we observed a strong exotherm.We do not recommend use of this unstable reagent 20 if alternative expoxidation reagents will suffice.
We also carried out the halolactonization reactions of the alkene acid 1 using NBS or NIS in acetone.In the first case, the major product was the bromovalerolactone derivative 12a, accompanied by a small amount of the (largely coeluting) seven-membered lactone 12b.These were present in a 6:1 ratio in the crude product mixture ( 1 H NMR). Use of NIS gave 13a the iodo analog of 12a; the seven-membered isomer was not definitively identified.With an eye toward converting this core skeleton into a(n) (meth)acrylate ester, we wondered whether some of the carboxylic acids 1, 9, or 11 could be transformed by an acid-catalyzed process into an isomeric dilactone alcohol by intramolecular ring-opening of a suitably reactive, cyclic, C-O ether bond by the pendant carboxylic acid (cf. 1 to 7b or the analogous ring-opening of 9 enroute to 10a/b).The hydroxyl group in the alcohol product would then be available for (meth)acrylation.
For 1 it was necessary, of course, to identify conditions milder than those described in Figure 1a in which the now-desired intermediate 7b was further transformed.We found that scandium triflate served as a very good catalyst for this purpose.Treating 1 with 10 mol% Sc(OTf)3 in acetonitrile, which provided a homogenous reaction mixture from the outset, at 70 °C for 5 hours provided the product 7b in 49% yield.The saturated analog 9 could be rearranged under the same conditions to give the dilactone alcohol 14 in 80% yield (Figure 4a).The more labile epoxide substrate 11 was quickly opened to the six-membered lactone alcohol 15 at room temperature (Figure 4c).The use of BF3•OEt2 resulted in an essentially identical yield of 18c, but other than that we did perform any further screening of catalysts.All of these lactonizations can be rationalized by acid activation of the ether bond with concomitant participation by the carboxylic acid.This could be achieved by direct activation of the ether oxygen by the Lewis acidic Sc(III) species or by the proton from a Lewis acidactivated Bronsted acid [RO(H)•Sc(III)], as depicted in the intermediate structures in Figure 4d.Although we presumed it would be a straightforward matter to esterify these three alcohols with methacrylic or acrylic anhydride or chloride, these experiments were rendered unnecessary because of the reaction shown in Figure 5a.Namely, a mixture of the alkene acid 1 and methacrylic anhydride in acetonitrile was treated with, again, 10 mol% Sc(OTf)3.Pleasingly, the methacrylate ester 16b was smoothly formed in this telescoped, one-pot, net acylative ring-opening reaction.Not surprisingly, acrylic anhydride proceeded analogously to give the acrylate ester 16a.Furthermore, both of the substrates 9 and 11 underwent analogous processes, leading to 17a or 17b and 18a or 18b.Not surprisingly, acetate esters (from Ac2O) were directly produced from all three of the ethers 1, 9, and 11.Finally, the benzoate (from BzCl) and pivalate esters (from PivCl) esters 17d and 17e were smoothly formed from the alkane acid 9 (the only ether we examined).Finally, we addressed the question of whether the alcohol products in Figure 4 were intermediates in the formation of the esters in the reactions shown in Figure 5.This meant evaluating the relative rates of the two different classes of ring-opening: namely, the isomerization to the alcohol (kalcohol) vs. the acylative opening to give the esters (kester).We did this for several of the reactions using in situ NMR monitoring.A solution of the ether-containing substrate in CD3CN was split into two portions into separate NMR sample tubes.The anhydride acylating agent was added to one of the two.Finally, an equal volume of a stock solution of Sc(OTf)3 in CD3CN was added to each tube.Using this protocol for reaction mixture preparation ensured that both of the reaction mixtures were subjected to essentially the same conditions (e.g., trace water in the acetonitrile).Reaction progress was then periodically monitored over time by in situ NMR spectroscopy.

© AUTHOR(S)
One example of (a portion of) the data collected from this kind of study is shown in Figure 6.This is for the reaction of the epoxide acid 11 without and with acetic anhydride (2 equiv) to give either the alcohol 15 or the acetate ester 18c.After 30 minutes (panel a) and 7 hours (panel b) at ambient temperature, in the absence of the anhydride the reaction had proceeded to the alcohol 15 to the extent of 16% and 50% conversions, respectively.In contrast, in the presence of the Ac2O, after only 30 minutes the reaction had proceeded to form the acetate ester 18c to 85% conversion (not shown) and after 7 h >99% conversion (panel c, the observable minor resonances are attributable to a mixed anhydride of 11).These types of experiments strongly suggest that the electrophilic Sc(III)-activated anhydride is more effective at promoting the ring opening compared to Sc(III) alone.Similar observations were made with the substrates 1 and 9 as well.

Conclusions
A variety of transformations of the readily available lactone acid 1 are reported.These lead to interesting scaffolds with different arrays of functional groups.Most notably, the acrylate or methacrylate esters 16a and 16b are available in two steps from furfuryl alcohol and itaconic anhydride.The (meth)acrylates 17a/b and 18a/b are efficiently available in three steps from the same commodity chemicals.We are now beginning studies to explore RAFT polymerizations of these (meth)acrylate monomers.

Experimental Section
General Experimental Protocols: Nuclear magnetic resonance (NMR) spectra ( 1 H and 13 C) were recorded on a Bruker HD-500 spectrometer. 1 H chemical shifts in CDCl3 samples are referenced to TMS (δ 0.00), in acetone-d6 samples to residual solvent protons at 2.05, DMSO-d6 to residual solvent protons at 2.50, and in CD3CN to residual solvent protons at 1.94.Data are reported using the following format: chemical shift (ppm) [multiplicity, coupling constant(s) (in Hz),

AUTHOR(S)
integral (to the nearest whole integer), and assignment of the proton].Coupling constant values have been deduced using protocols previously described. 21,22Non-first order multiplets in a 1 H NMR spectrum are designated by 'nfom'.Non-first order doublets in a 1 H NMR spectrum (e.g., present in a 1,4disubstitutedbenzene ring) are designated by 'nfod' and the apparent doublet coupling constant (actually, e.g., J2,3 + J2,3') is indicated as Japp. 13C{ 1 H} NMR chemical shifts are referenced to the carbon atom in CDCl3 to 77.16 ppm, in acetone-d6 to 29.8 ppm, and in DMSO-d6 to 39.5 ppm.Where assigned, carbon chemical shifts were deduced from analysis of HSQC and/or HMBC data.Infrared (IR) spectral data were collected on a Bruker spectrometer (model Alpha II).Samples were deposited as films on a diamond window (solids by evaporation from DCM; liquids by direct application) in the mode of attenuated total reflectance (ATR).Peaks are reported in cm -1 .High-resolution mass spectrometry (HRMS) measurements were made using ESI ionization with a Thermo instrument (model Orbitrap Velos, which has a mass accuracy of ≤3).An external calibrant (Pierce TM LTQ) was used.Samples were injected directly into the ion source.Medium pressure liquid chromatography (MPLC) was used to purify products.Silica gel (normal-phase, 20-40 μm, 60 Å pore size, Teledyne RediSep Rf Gold ® ) columns were hand-packed into Michel-Miller ® glass columns.The equipment used consisted of a Waters HPLC pump (model 510) and a Waters (R401) differential refractive index detector.Preparative flash chromatography was done on silica gel (230-400 mesh) columns.Thin layer chromatography (TLC) was performed on silica-gel-coated, plastic-backed plates (Machery-Nagel) that were visualized by staining with KMnO4 solution.Heating of reactions performed above ambient temperature was done in silicone oil baths that had been preequilibrated to the desired temperature prior to immersion of the reaction vessel.

(±)-(3aR,5S,7aR)-5-Hydroxytetrahydro-3a,7a-(methanooxymethano)benzofuran-2,10(3H)-dione (14).
A mixture of Sc(OTf)3 (5 mg, 0.01mmol) and the lactone acid 9 (21 mg, 0.1 mmol) were taken into a screwcapped culture tube.Acetonitrile-d3 (0.6 mL) was added.The resulting colorless homogenous solution was heated at 80 °C for 5 h.The solvent was evaporated under vacuum from the resulting colorless solution and the crude material was eluted through a silica gel plug with the aid of EtOAc.The eluent was concentrated and gave 14 (17 mg, 80%) as a white crystalline solid.The proton NMR spectrum of this reaction mixture showed a very clean conversion to the set of resonances for the product 14. 1

Figure 2 .
Figure 2. (a) Triflic acid-catalyzed conversion of 1 to a mixture of the dilactone-diene 5 and isochromanone 6 via the proposed sequence proceeding through intermediates 7a-e.(b) Triflic acid-catalyzed conversion of the dihydro derivative of 1, the saturated lactone acid 9, leads to a mixture of the two monoenes 10a and 10b.

Figure 3 .
Figure 3. (a) Epoxidation of 1 with peracetic acid at room temperature.(b) Halolactonizations of 1 with NBS or NIS in acetone at room temperature.

Figure 4 .
Figure 4. Scandium triflate catalyzed ring-opening to dilactone alcohols via cleavage of strained cyclic ether bonds in (a) the alkene 1, (b) the alkane 9, and (c) the epoxide 11.(d) Depiction of two possibilities for the key step in a likely mechanism for the Sc(III)-catalyzed conversion of, for example, 11 to 15.