Chemoenzymatic Synthesis of Chiral Building Blocks Based on the Kinetic Resolution of Glycerol‐Derived Cyclic Carbonates

The biocatalytic kinetic resolution of cyclic carbonates derived from glycerol is reported. A selection of 26 esterases and lipases was tested for the asymmetric hydrolysis of the model substrate (epichlorohydrin carbonate) in aqueous medium. Among them, Pig Liver Esterase and Novozym® 435 showed the best selectivity with E=38 and 49, respectively. Both enzymes were employed for the conversion of 12 glycerol derivatives under optimized conditions. The resolution of halogenated carbonates afforded the unconverted enantiomer in up to >99 : 1 er. Furthermore, Novozym® 435 was successfully recycled 10 times without significant loss of activity. Upscaling and isolation of the chiral carbonate was also demonstrated. Subsequent conversion of this chiral building block allowed the direct one‐pot synthesis of (S)‐Guaifenesin, (S)‐Mephenesin and (S)‐Chlorphenesin in up to 89 % yield and 94 : 6 er.


Introduction
Cyclic carbonates have gained increasing attention in recent years due to their attractive synthesis from CO 2 and their applications in several fields.Both at the industrial and academic level they are widely employed for their properties such as low vapor pressure, low toxicity, high flash point and high dielectric constant. [1,2]In particular, in synthetic chemistry, organic carbonates present interesting reactivity.[12] Moreover, the functionalization of the substituents adjacent to the carbonate makes these compounds valuable building blocks. [10][15] Numerous procedures are reported for the direct conversion of glycerol to glycerol carbonate [16][17][18] and for the functionalization of the hydroxyl group in the glycerol carbonate molecule, which leads to various valuable glycerol-based carbonates 1. [19] Conversion of glycerol to cyclic carbonates is also possible via different routes, for example, in the epicerol process glycerol is converted to epichlorohydrin which can be further reacted with CO 2 to the corresponding carbonate. [20]It is also noteworthy that the cycloaddition of CO 2 to epoxides is the most common route for the synthesis of 5-membered cyclic carbonates.This reaction presents several advantages in the context of green chemistry since it is 100 % atom efficient and uses carbon dioxide as inexpensive, non-toxic, and renewable C1-building block. [21,22]ven though the preparation of cyclic carbonates from epoxides has been extensively studied, their asymmetric synthesis remains challenging.[31] In this context, the kinetic resolution of the cyclic carbonates 1 emerges as an alternative (Scheme 1).In this approach, the substrates are selectively hydrolyzed, giving the enantioenriched cyclic carbonates and their corresponding diols.[34][35] Even though these enzymes are widely known for their activity towards esters, lactones and amides, few reports explore the hydrolysis of the carbonate group.In this regard, we have recently reported the kinetic resolution of styrene carbonate with pig liver esterase (PLE), [36] which motivated a deeper study towards the potential of carboxylesterases for the synthesis of chiral cyclic carbonates.Special attention was drawn to the activity of Novozym® 435 (N435), an acrylic resinbased adsorbed form of lipase B from Candida antarctica, which is widely employed due to its promiscuity, stability, recyclability, low production cost and applicability in industrial processes. [37,38]The resulting chiral carbonates can be employed as chiral building blocks, e. g. for the synthesis of terminal aromatic glycerol ethers (TAGEs) with pharmacological activity, such as (S)-Guaifenesin, (S)-Mephenesin and (S)-Chlorphenesin (Scheme 1). [39]

Results and Discussion
The kinetic resolution of 4-(chloromethyl)-1,3-dioxolan-2-one (1 a) was chosen as model reaction for the screening of a series of 26 commercially available hydrolases (Scheme 2).This substrate is an interesting building block since it can be easily prepared from epichlorohydrin derived from glycerol [20] and the substitution of the chlorine atom can be used for further functionalization.The enzyme screening was conducted under similar conditions to the previously reported kinetic resolution of styrene carbonate (4 mg/mL of enzyme, 23 °C, 24 h, NaPi buffer pH 7.0). [36]The absolute configuration of the unreacted enantiomer ((S)-1 a) was determined by chiral GC, comparing the retention time of the unreacted enantiomer with that of an enantiopure standard.For the evaluation of the results, the conversion of 1 a, the enantiomeric purity of the unreacted (S)-1 a and the resulting E-values were determined (Figure 1).Notably, when PLE was employed (S)-1 a was obtained with an er > 99 : 1 (E = 38), while the control reaction without enzyme gave only 5 % conversion, indicating that the non-catalyzed hydrolysis occurs slowly.
Even though PLE is one of the most used esterases in synthesis due to its wide applicability for the asymmetric hydrolysis of esters, it is prepared by extraction of the pig liver obtaining a batch-depending mixture, which presents several disadvantages. [40]43] Thus, six recombinant isoenzymes of PLE were tested in parallel (rPLE1À rPLE6).While rPLE1À rPLE4 gave low conversions (< 20 %), PLE5 and PLE6 converted 1 a in 43 % and 27 %, respectively.However, all rPLE showed low selectivity for the preparation of (S)-1 a (E < 3).Other recombinant esterases were studied, obtaining also poor selectivity for the asymmetric hydrolysis (E < 4).Among them, only the esterases from Bacillus subtilis A and Paenibacillus barcinonensis gave enantiomeric ratios significantly different from 50 : 50.Finally, lipases were evaluated for the kinetic resolution of 1 a.All of them gave low conversion of the substrate (< 20 %) and low selectivity (E < 2) except for the immobilized Lipase B from Candida antarctica, Novozym® 435, which despite the low conversion under the reaction conditions (14 %), showed good selectivity (E = 49).
Due to these promising results, all further experiments were carried out with PLE and N435 with the objective of preparing and isolating pure (S)-1 a.The focus was set on achieving at least an er > 95 : 5, while minimizing the conversion to less than 75 % (Table 1).Under the conditions tested for the screening, PLE gave (S)-1 a with excellent purity (er > 99 : 1), but required high conversion (82 %) thus limiting the yield of the desired carbonate (Entry 2).A lower enzyme concentration (2 mg/mL) gave a reduced conversion (57 %), as expected, but the er obtained was also considerably decreased (85 : 15, Entry 1).When the concentration of the substrate was increased from 10 to 40 mM (Entry 3) 1 a was converted in 65 % obtaining (S)-1 a with very good enantiomeric purity (er = 93 : 7).This also highlights the potential applicability of this biocatalyst for the synthesis of chiral building blocks since the enzyme tolerates high concentrations of substrate.Finally, the reaction was carried out for 16 h (Entry 4) obtaining (S)-1 a in high er of 98 : 2 with a conversion of 73 %.Subsequently, N435 was also evaluated for the synthesis of (S)-1 a. Increasing the enzyme concentration from 4 to 10 and 20 mg/mL resulted in an increment of the (S)-1 a purity from 58 % to 71 % and 83 % (Entries 5-7) with a conversion below 40 % in all cases.However, a further increase to 30 mg/mL of N435 did not show significant changes (Entry 8).The effect of the temperature was also studied for N435.However, higher temperature led to higher conversion of both enantiomers at the expense of a loss of selectivity.When the reaction was performed at 40 °C with 10 mg/mL of N435 the substrate was converted in 67 %, and an er of 88 : 12 was obtained (Entry 9).An enzyme concentration of 20 mg/mL at 40 °C resulted in a conversion of 73 % and an excellent er of 97 : 3 of (S)-1 a (Entry 10).Overall, both PLE and N435 allowed the preparation of pure (S)-1 a with er � 97 : 3 at 73 % conversion of the initial substrate (Entries 4 and 7).The optimized conditions for PLE (Entry 4) and N435 (Entry 10) were chosen for further experiments.
The results obtained with PLE and N435 for the model reaction motivated us to explore the activity of these enzymes towards different substrates.Therefore, halogenated derivatives (1 a-1 d) were tested (Figure 2).These carbonates can be easily prepared by transformation of glycerol carbonate by nucleophilic substitution. [19]Under the conditions previously chosen for PLE and N435, the fluorine derivative (1 b) gave a poorer result (er < 73 : 7) than its chlorinated analogous 1 a (er � 97 : 3), which can be explained considering that fluorinated compounds are known to be enzyme inhibitors. [44]Notably, the bromine (1 c) and iodine (1 d) derivatives gave excellent er values (� 99 : 1) with both PLE and N435.Overall, despite the high conversions (72-84 %) for 1 a, 1 c and 1 d, the corresponding chiral carbonates were obtained in excellent enantiomeric purity.
Furthermore, glycerol carbonate 1 e and its derivatives 1 f-1 l were also evaluated (Figure 3).Kinetic resolution of glycerol carbonate (1 e) afforded the racemic mixture despite a conversion of 18 % with PLE and 38 % with N435.Low selectivity was found for the methoxy-substituted carbonate, 1 f, despite the significant conversion (32 % for PLE and 45 % for N435).For the tert-butoxy-and allyloxy-substituted carbonates, 1 g and Table 1.Optimization of the reaction conditions for the kinetic resolution of 1 a with PLE and N435.

Entry
Enzyme (mg/mL) Conversion 1 a (%) [a] er (S)-   1 h, the conversion with PLE was low (4 and 16 %, respectively).N435 led to higher conversion of 1 g and 1 h giving an er of 63 : 37 and 66 : 34, respectively.Furthermore, we became interested in phenoxy-substituted derivatives (1 i-1 l) since the hydrolysis of the carbonate moiety in these compounds gives the corresponding diol.Such structural motives can be found in various pharmaceuticals. [45]For the phenoxy-derivative 1 i the conversion by PLE was < 5 % but with N435 it gave a moderate er (77 : 23) at a conversion of 42 %.The guaiacol derivative 1 j was significantly converted by both enzymes, but with poor selectivity, affording the carbonate in er = 54 : 46 with PLE and er = 65 : 35 with N435.A similar result was obtained for 1 l, which despite the high conversion (79 % and 87 %) gave poor er values. 1 k was obtained as a racemic mixture after using PLE (11 % conversion), while N435 converted 1 k in 29 % with 57 : 43 er.Considering the insolubility of 1 i-1 l it was hypothesized that a solvent might have a favorable effect on the reaction.As DMSO is a suitable co-solvent for hydrolases, [32,46] the kinetic resolution of 1 i was carried out using 10 % of DMSO.Unfortunately, no significant improvement was obtained with PLE, and a detriment in the selectivity was found for N435 (see Supporting Information).
It is worth to highlight that for the substrates studied (1 a-1 l) both enzymes showed identical enantiopreference, yielding the same enantiomer.For substrates 1 a, 1 i and 1 j the absolute configuration of the obtained enantiomer was assigned by comparison with the retention times (chiral GC) of the corresponding chiral standard (see Supporting Information).Considering this and the fact that enantiocomplementary enzymes that favor the conversion of opposite enantiomers are very rare, [47,48] it can be assumed that the same configuration of the chiral center is obtained for all substrates ((S)-1 a-1 d and (R)-1 e-1 l).Overall, N435 gave a higher conversion rate and higher selectivity for most substrates.The different results obtained for these substrates underline the high impact of the substituent on enzyme activity and selectivity towards the hydrolysis of cyclic carbonates.
We were also interested in evaluating the synthetic potential of this procedure.The potential recyclability of N435 and the good results obtained for 1 a with this enzyme motivated us to perform the kinetic resolution of 1 a on a larger scale and to study the recyclability of N435 at the same time.Thus, the kinetic resolution of 1 a was scaled up 40 times using 20 mg/mL of N435, at 40 °C (Figure 4).After 24 h, N435 was simply recovered by filtration and (S)-1 a and (R)-2 a were isolated from the aqueous mixture by extraction.The procedure was repeated 10 times isolating (S)-1 a in yields between 15 %  and 24 % (with 50 % being the maximum yield).In the first 5 cycles the carbonate was obtained in excellent er (� 95 : 5) and after 10 cycles the selectivity of the enzyme only dropped slightly giving er values � 92 : 8.As a result of the successful multiple usages of N435, 529 mg of pure (S)-1 a were isolated.On the other hand, (R)-2 a was isolated in yields up to 47 % and er values up to 62 : 38 (see Supporting Information).The diol could in principle also be reconverted to the carbonate and resubmitted to the kinetic resolution.The synthesis of chiral vicinal diols is of great relevance in synthetic chemistry. [49,50]ne of the most common routes for their preparation is the Sharpless asymmetric dihydroxylation of alkenes with OsO 4 . [51]nother well studied reaction is the hydrolytic kinetic resolution of epoxides with Co-complexes. [52]In this work, cyclic carbonates were envisioned as protected vicinal diols by taking advantage of the carbonate moiety, which uses CO 2 as protecting group giving the diol after hydrolysis with a base. [3]dditionally, the substitution of a suitable leaving group allows the use of these glycerol-derived cyclic carbonates as building blocks.Thus, the substitution of the chlorine in (S)-1 a by a phenoxide together with the hydrolysis of the carbonate should facilitate the direct synthesis of (S)-TAGEs (terminal aromatic glycerol ethers) in one-step (Scheme 3).Moreover, TAGEs serve as precursors for the preparation of other APIs, such as βblockers, for which is known that their pharmacological activity changes with different enantiomers.For example, (S)-propanol and (S)-atenolol are more potent than the corresponding (R)enantiomers. [53,54]To evaluate the feasibility of this approach for the preparation of (S)-TAGEs, the obtained product (S)-1 a (er = 94 : 6) was used as starting material with different phenols as nucleophiles under basic conditions.The reaction was carried out with NaOH in ethanol by modification of a protocol reported by Egri et al. for 3-chloropropane-1,2-diol. [55]The substitution with phenol (3 a) as a model compound afforded (S)-2 i in 86 % yield and 90 : 10 er.Guaifenesin (2 j) is an expectorant commonly used for the treatment of mucus and phlegm. [56,57]With this procedure (S)-Guaifenesin ((S)-2 j) was prepared from (S)-1 a and guaiacol (3 b) in 89 % yield with 91 : 9 er (Scheme 3).Mephenesin (2 k) is a muscle relaxant, which was obtained from o-cresol (3 c) and (S)-1 a giving (S)-2 k in 81 % yield with total retention of the configuration (94 : 6 er).Chlorphenesin (2 l) is used as antiseptic and preservative since it presents antifungal and antibacterial properties [58] and its carbamate acts as muscle relaxant. [59]Similarly to (S)-2 k, (S)-Chlorphenesin ((S)-2 l) was obtained from (S)-1 a and pchlorophenol (3 d) with 94 : 6 er, in 88 % yield.
Furthermore, we were able to isolate crystals of (S)-2 i and (S)-2 j suitable for X-ray crystallographic analysis.The molecular structures confirmed the absolute configuration of the chiral center (Figure 5, and Supporting Information). [60]he feasible reaction pathways leading to the formation of (S)-2 i or (R)-2 i are shown in Scheme 4. Attack at the Clsubstituted carbon by aromatic nucleophiles leads to the desired reaction (path-1) and the formation of carbonate (R)-1 i. [13,61] Subsequently, basic hydrolysis yields the desired vicinal diol (S)-2 i. Due to steric interactions the substitution at the chiral center is not feasible.However, the attack of the phenoxide anion on the less hindered methylene carbon is plausible (Scheme 4, path-2).This leads to linear intermediate Int-a, which can undergo an intramolecular nucleophilic substitution, giving (S)-1 i.The subsequent basic hydrolysis yields (R)-2 i.Thus, even though the substitution does not occur at the chiral center, the observed partial racemization is a consequence of the two different pathways.However, the lability of the chloride favors path-1 as path-2 was only observed in the synthesis of (S)-2 i and (S)-2 j, which explains the slightly lower er values of 90 : 10 and 91 : 9, respectively, compared to the er of the starting material (S)-1 a (94 : 6).In contrast (S)-2 k and (S)-2 l were obtained with complete retention of the configuration.

Conclusions
PLE and N435 proved to be efficient and highly enantioselective for the kinetic resolution of cyclic carbonates derived from glycerol.Among the substrates studied, the best results were found for halogenated carbonates.In particular, cyclic carbonates with chloro-, bromo-and iodomethylene-substituents gave moderate to high conversions affording the remaining (S)-Scheme 3. Synthesis of (S)-TAGEs by conversion of (S)-1 a with phenols 3 under basic conditions.Reaction conditions: 1.0 mmol of (S)-1 a added to 3.0 mmol of phenol and 3.0 mmol of NaOH in EtOH at 100 °C for 3 h.Isolated yields are informed.Er values were determined by chiral GC-FID.enantiomers in excellent enantiomeric ratios of up to � 99 : 1.Moreover, the reaction was successfully scaled up, which allowed the isolation of (S)-1 a. Notably, N435 could be easily recycled and 9 times reused without significant loss on the activity and selectivity.Finally, the potential of the obtained cyclic carbonates as building blocks was demonstrated by synthesizing pharmaceutically relevant (S)-TAGEs from (S)-1 a and simple phenols, under basic conditions.

Experimental Section
General procedure 1 (GP1) for screening of reaction conditions for the kinetic resolution of 1 a 0-100 mg of enzyme were added to a solution of the carbonate 1 a (0.05 mmol, 10 mM) in 5.0 mL of NaPi buffer (0.1 M, pH 7.0).The mixture was shaken at 300 rpm at 23-40 °C.After 8-24 h, the mixture was filtered and extracted with EtOAc (3×25 mL).The organic layers were combined and dried over Na 2 SO 4 .All volatiles were removed in vacuum.The resulting crude mixture was transferred to a 5.0 mL volumetric flask with EtOH and analyzed by chiral GC-FID using hexadecane as internal standard.
General procedure 2 (GP2) for the kinetic resolution of 1 with PLE 20 mg of PLE were added to a solution or suspension of the carbonate 1 (0.05 mmol, 10 mM) in 5.0 mL of NaPi buffer (0.1 M, pH 7.0).The mixture was shaken at 300 rpm at 23 °C.After 16 h (1 a-1 d) or 24 h (1 d-1 l), the mixture was filtered and extracted with EtOAc (3×25 mL).The organic layers were combined and dried over Na 2 SO 4 .All volatiles were removed in vacuum.The resulting crude mixture was transferred to 5.0 mL volumetric flask with EtOH and analyzed by chiral GC-FID using hexadecane as internal standard.

General procedure 3 (GP3) for the kinetic resolution of 1 with N435
100 mg of PLE were added to a solution or suspension of the carbonate 1 (0.05 mmol, 10 mM) in 5.0 mL of NaPi buffer (0.1 M, pH 7.0).The mixture was shaken at 300 rpm at 40 °C.After 24 h, the mixture was filtered and extracted with EtOAc (3×25 mL).The organic layers were combined and dried over Na 2 SO 4 .All volatiles were removed in vacuum.The resulting crude mixture was transferred to 5.0 mL volumetric flask with EtOH and analyzed by chiral GC-FID using hexadecane as internal standard.

General procedure 5 (GP5) for the synthesis of (S)-TAGEs by substitution
In a pressure tube, the phenol 3 (3.0mmol) was dissolved in EtOH (0.5 mL) and a solution of NaOH (120 mg, 3.0 mmol) in H 2 O (0.4 mL) added.The reaction mixture was stirred at 100 °C for 30 minutes.Subsequently, a solution of (S)-1 a (136 mg, 1.0 mmol) in EtOH (1.0 mL) was added dropwise and the mixture was stirred at 100 °C for 3 h.Subsequently, H 2 O (2 mL) were added, and the reaction mixture was further stirred at 100 °C for 30 min.The mixture was cooled to room temperature, neutralized with 1.0 mL of HCl (1 M) and extracted with EtOAc (3×5 mL).The organic layers were combined and dried over Na 2 SO 4 .All volatiles were removed in vacuum.The resulting crude products were purified by flash Scheme 4. Proposed pathway for the synthesis of (S)-TAGEs by substitution of (S)-1 a with phenol under basic conditions.column chromatography (SiO 2 , cHex/EtOAc = 1 : 1).The er values of the diols were measured as er values of their corresponding carbonates by derivatization of 0.10 mmol of diol 2 in dimethylcarbonate (0.5 mL), with K 2 CO 3 (0.15 mmol) and molecular sieves (3 Å, 50 mg) at 110 °C for 24 h, according to a reported procedure. [62]The resulting product was transferred to 5.0 mL volumetric flask with EtOH and analyzed by chiral GC-FID using hexadecane as internal standard.

Figure 1 .
Figure 1.Results of the enzymatic kinetic resolution of 1 a with selected esterases and lipases.Reaction conditions: 10 mmol/L of 1 a in 5.0 mL of NaPi buffer (0.1 M, pH 7.0), 4 mg/mL of enzyme, 23 °C, 24 h.Conversions (&), er values (*) were determined by chiral GC-FID using hexadecane as the internal standard.E-values (*) were calculated from the conversion of 1 a and ee of (S)-1 a. [a] 20 h.
Reaction conditions: 10 mmol/L of 1 a in 5.0 mL of NaPi buffer (0.1 M, pH 7.0), 23 °C, 24 h.Conversions and er values were determined by chiral GC-FID using hexadecane as the internal standard.E-values (*) were calculated from the conversion of 1 a and ee of (S)-1 a. [b] 40 mM of 1 a. [c] 16 h.[d] 40 °C.

Figure 5 .
Figure 5. Molecular structure of (S)-2 i in the crystal.Displacement ellipsoids correspond to 50 % probability at 110 K.