S monolignol and syringaresinol were synthesized according to procedure described in one of our previous articles (Jaufurally et al., 2016). 0.4 g of syringaresinol (0.96 mmol) was placed in a 500 mL flask and dissolved in 180 mL of ethanol (EtOH). Then, 200 mL of deionized water was added (1 g/L of total solvents) prior the introduction of 17.6 mg of laccase (0.1 U/mg regarding total substrates). 2 g of S monolignol (9.51 mmol) was dissolved in 20 mL of EtOH (100 g/L) and added dropwise at 0.9 mL/h using a syringe pump. The mixture was stirred in the dark at ambient temperature for 24 h. EtOH was removed under reduced pressure, the mixture was cooled to 0°C, and was then filtered on a glass filter. The same procedure was repeated 10 times and the products gathered in order to obtain a sufficient batch that was dissolved in acetone, then precipitated into diethyl ether. After filtration, the recovered S-DHPs (approximately 10 g) were dried under vacuum.

Alkali lignins Protobind 1000 (PB1000) was magnetically stirred in ethyl acetate (10 L/kg) for 1 h at room temperature. The mixture was then filtered and flushed with ethyl acetate on a glass filter. The procedure was repeated a second time and the filtrates were combined prior to evaporation under reduced pressure to eliminate ethyl acetate and recover F1 as a brown solid (36% wt.). The dried insoluble residue was magnetically stirred in methyl ethyl ketone (MEK) (10 L/kg) for 1 h at room temperature. The mixture was then filtered and flushed with MEK on a glass filter. The procedure was repeated a second time and the filtrates were combined prior to evaporation under reduced pressure to eliminate MEK and recover F2 as a brown solid (26% wt.) and an insoluble residue F3 (37% wt.).

The substrate to be acetylated (200 mg) was placed in the presence of supported CAL-B (20 mg, 10% mass of the substrate) and mixed with ethyl acetate:acetonitrile (1:1 vol, 20 mL, 10 g/L) in a 25 mL round bottom flask equipped with a Dean–Stark apparatus. The mixture was magnetically stirred under reflux for 24 h and aliquots were periodically collected for kinetics study. After cooling to room temperature, the mixture was filtered to recover the supported lipase and the filtrate was concentrated under reduced pressure to recover the acetylated compound.

S-DHPs (200 mg) was placed in a 25 mL round bottom flask under argon flow in the presence of anhydrous pyridine (2 mL). After cooling to 0°C, acetic anhydride (2 mL, 10 mL/g) was added and the mixture was magnetically stirred at room temperature for 18 h. The reaction was cooled at 0°C before adding methanol dropwise (2 mL). Toluene (2 mL) was then added and the solvents were evaporated under reduced pressure to recover the acetylated DHPs.

The substrate to be hexanoylated was placed in a round bottom flask in the presence of ethyl hexanoate (1:1 mass ratio to substrate), supported CAL-B (10% mass of substrate), and MEK (25 g/L). The mixture was magnetically stirred under reflux for 24 h using a Dean–Stark apparatus. After cooling to room temperature, the mixture was filtered to recover the supported lipase and the filtrate was concentrated under reduced pressure. In the case of the lignins substrate, the mixture was then precipitated into hexane under magnetic stirring and hexanoylated lignins were recovered by filtration on a glass filter, then dried under vacuum overnight. The parameters that have been changed for the purpose of the study are indicated in the discussion.

All NMR spectra were recorded on an Ascend™ 400 spectrometer (Bruker).

Liquid chromatography analyses were performed using a HPLC system (Thermo Fisher Scientific) equipped with an ACCELA 600 pump, an ACCELA auto sampler, a C18 column EC 50/2 Nucleoshell RP 18, 2.7 μm (Macherey-Nagel), and an ACCELA photodiode array (PDA) detector recording over the range 250–450 nm. Samples were dissolved in acetonitrile (ACN) at 1 mg/mL, filtered on a 0.45 µm PET microdisk (Macherey-Nagel), and 1 µL was analyzed using a mobile phase consisting of water/ACN +1‰ HCOOH eluted at 250 μL/min, applying a gradient from 80/20 to 30/70 within 15 min.

10 mg of samples was weighed and tetrahydrofuran (THF) containing 5% of toluene (internal standard) was added in order to reach a concentration of 5 mg/mL. After 5 min of shaking, samples were filtered on 0.45 µm GHP microfilters (PALL) prior to analysis by size-exclusion chromatography (SEC). SEC analyses were performed using THF stabilized with BHT as the eluent at 1 mL/min (Ultimate 3000 Pump, Dionex). 10 μL was injected (Ultimate 3000 Autosampler, Dionex) on either a PLgel Mixed C column (5 μm, 7.5 mm × 600 mm, Agilent) in the case of DHPs, or a PLgel Mixed E column (3 μm, 7.5 mm × 600 mm, Agilent) in the case of lignins, and the signal was observed at 280 nm (Ultimate 3000 UV/vis detector, Dionex). The molar mass distributions were determined using a calibration curve based on 10 polystyrene standards ranging from 580 to 364,000 g/mol (Agilent) for the Mixed C column (Supplementary Figure S1) or from 162 to 22,290 g/mol (Agilent) for the Mixed E column (Supplementary Figure S2).

In order to test the feasibility and selectivity of the reaction, the CAL-B activity was first tested on commercial guaiacylglycerol-β-guaiacyl ether dimer, which is commonly used as representative of the β-O-4 linkage between two coniferyl alcohols found in lignin oligomers as the major linkage between two p-hydroxycinnamic subunits. This commercial stereopure dimer exhibits a free phenolic group and both primary and secondary aliphatic hydroxy groups. The acetylation of the dimer (Figure 2) was first investigated in the presence of ethyl acetate and supported CAL-B (10% wt of dimer) in acetonitrile (10 g/L, 50:50 acetonitrile:ethyl acetate) and with continuous removal of the by-produced ethanol by azeotropic distillation. The reaction was easily followed by HPLC and ¹H NMR (Supplementary Figure S4). These first results showed only acetylation of the primary hydroxyl in approximately 90% yield after 8 h of reaction. The reaction could also be monitored by ³¹P NMR after phosphorylation (Supplementary Figure S5), where the relative integration of the signal corresponding to the phosphorylated primary OH (triplet at 147.5 ppm) progressively decreased, while the integration of secondary (doublet at 148.2 ppm) and phenolic OH (singlet at 139.5 ppm) remained equal, even if their chemical shift slightly increased (to 148.3 and 139.6 ppm, respectively), thus confirming the selectivity. To further ensure this selectivity, the reaction was conducted over a longer time period (13 days): no peracetylated product was observed by HPLC, ¹H NMR, or ³¹P NMR. If CAL-B is well-known to be strictly inactive toward phenols, it may induce acylation of secondary hydroxyls (Pion et al., 2013). In this current case, inactivity toward the secondary hydroxyl might be due either to stereoselectivity toward the involved stereopure commercial dimer, or to the proximity of the hindered phenolic moiety (Uppenberg et al., 1995). Acetonitrile, first selected because it is known for being tolerated by CAL-B (Dutta Banik et al., 2016) has however a rather low ability to solubilize lignins. Thus, it was thereafter replaced by methyl ethyl ketone (MEK), showing similar acetylation kinetics on the commercial dimer (Figure 3). MEK was therefore chosen as the reaction solvent for this process and to efficiently solubilize a large fraction of Protobind 1000 (60% wt), which is our targeted lignin substrate.

However, direct monitoring of the reaction on lignins fractions by ¹H NMR or HPLC would be unfeasible due to the complexity of the lignins’ structure and their poor solubility in the appropriate solvents (Wen et al., 2013; Wurzer et al., 2021). Therefore, to challenge further analytical methodologies (SEC, ³¹P NMR), we turned to the acetylation of a substrate of intermediate complexity: dehydrogenative dehydropolymers (DHPs) (Lahive et al., 2020) of sinapyl alcohol (Figure 4). Such S-type DHPs, exhibiting only two types of inter subunit linkages (β-O-4 and syringaresinol types), are indeed good candidates to adapt our analytical procedures to a more complex substrate representative of lignins.

DHPs were treated in the presence of CAL-B (10% wt compared to DHPs) in ethyl acetate:MEK (50:50 vol, 10 g/L) under reflux for 2 days. Afterwards, the mixture was cooled to room temperature, filtered to recover supported CAL-B, and the solvent evaporated under vacuum to obtain the resulting acetylated DHPs.

For the purpose of analytical comparisons, completely acetylated DHPs were prepared through a chemical procedure in pyridine using acetic anhydride. From ¹H-¹³C HMBC NMR (in C₅D₅N) of the chemically acetylated DHPs, three acetate groups were observed: one phenolic (2.29 ppm/168.5 ppm) and two aliphatic, corresponding to primary and secondary hydroxyls (1.94 ppm/170.6 ppm and 2.13 ppm/170.0 ppm, respectively), while the enzymatically acetylate showed a single primary aliphatic acetate signal (1.94 ppm/170.6 ppm, Supplementary Figure S6). This confirmed the transposability of the selective enzymatic process to more complex substrates, where acetylation is restricted to primary aliphatic hydroxyls. In addition, another characteristic signal was observed at 2.11 ppm/173.3 ppm, corresponding to residual acetic acid (confirmed by comparison with pure acetic acid, which also appears at 134.6 ppm in ³¹P NMR, Supplementary Figure S7).

The molar mass distribution of the samples was also determined through size exclusion chromatography (SEC), assuming that the acetylation could induce an increase in apparent molar mass. Chemically acetylated DHPs showed a slight increase in molar mass distribution when compared to the starting material (Figure 5). Surprisingly, this increase in apparent molar mass was even more significant in the case of enzymatic acetylation, while it should have a lower molar mass gain since its acetylation was partial. Such an observation suggests that side reactions, such as intermolecular cross-coupling reactions, may occur in these reaction conditions in the case of DHPs.

Therefore, to better understand this unexpected result, two control reactions were run: one in the absence of any acyl donor and another in the absence of the catalyst CAL-B. In the two cases where CAL-B is present (control without acyl donor and enzymatic reaction), a significant increase in molar mass is observed, considering that lipase treatment is able to induce cross-coupling reactions between DHPs residues. In the absence of CAL-B, a similar molar mass increase is observed, but is far less important and probably also due to chemical modifications of the DHP structure upon thermal treatment (Figure 6). The hydroxyl group content of the different samples was then assessed through ³¹P NMR after phosphorylation, a method commonly used for the quantification of the different types of lignin hydroxyls (Table 1) (Meng et al., 2019).

In the case of enzymatic acetylation, the aliphatic hydroxyl content significantly decreased (−40%, Entry 2), suggesting efficient acetylation, as observed by the appearance of an ester spot in ¹H-¹³C HMBC NMR. Nevertheless, even if lower, this phenomenon was also observed in controls (−20% and −17%, Entries 3 and 4), where no acetate was observed through ¹H-¹³C HMBC NMR, thus suggesting that this decrease in aliphatic hydroxyls can also be due to spontaneous side reactions. Moreover, this hypothesis is reinforced by an increase in the free phenol content, probably due to the cleavage of some β-O-4 bonds inside the oligomers. From such global observations it can be assumed once more that DHPs (and probably also lignin fractions) may undergo dramatic structural changes in these reaction conditions (temperature, solvent), including for instance β-O-4 bond cleavage and recondensation reactions. The balance between these two pathways, going from control reactions to the real reaction may be to some extent due to the generation of acetic acid in the medium containing the ethyl acetate and CAL-B, but also to the specific activity of CAL-B on lignin oligomers in the absence of any acyl donor.

Although confirming the occurrence of side reactions impacting the hydroxyl content in all cases, ³¹P NMR appeared to be nevertheless unsuitable for precise quantification of the acylation yield for DHPs (and probably for lignin fractions), contrary to our expectations. Indeed, the acetylation yield, when estimated on the basis of ³¹P NMR data, seemed largely overestimated (approximately 50% mol of the putative primary hydroxy groups). We thus concluded that another analytical method was needed to quantify the acylation yield. We envisaged the use of an indirect method based on either saponification or transesterification methods, allowing the reformation of acids or esters from the acylated lignins, which can be further quantified by GC-MS. However, such a process is incompatible with acetylated compounds since it generates either acetic acid or methyl acetate, which are difficult to quantify by GC-MS. Another longer acyl donor was therefore chosen, ethyl hexanoate, as its excess can still be easily removed by either evaporation or hexane washing, and since its corresponding acid or methyl ester are easily quantifiable by GC-MS.

The hexanoylation process was first tested on the commercial dimer in order to assess the efficiency of this approach with a longer chain and a more apolar acyl donor. Guaiacylglycerol-β-guaiacyl ether dimer was thus reacted with ethyl hexanoate in a ratio of 5:1 mol to ensure an excess of acyl donor at the same conditions (10% wt CAL-B, MEK, 10 g/L, reflux). The kinetics of hexanoylation was monitored by ¹H NMR (Supplementary Figure S8) and HPLC and was found to be very similar as the acetylation (Figure 7), with a hexanoylation yield of 76% after 9 h, thereby proving the versatility of the method in terms of acyl donor. Similar control experiment ran without CAL-B induced no structural changes of guaiacylglycerol-β-guaiacyl ether dimer (results not shown), thus confirming that hexanoylation was indeed only catalyzed by CAL-B.

DHPs were then reacted with an excess of ethyl hexanoate (5:1 wt) in MEK (10 g/L) under reflux. After the biocatalyzed hexanoylation reaction and the removal of the supported CAL-B by filtration, the reaction medium was concentrated under reduced pressure and first cleaned from residual unreacted acyl donor through precipitation in hexane, and submitted to extended drying under vacuum. ¹H-¹³C HMBC NMR proved the formation of an ester bond (Supplementary Figure S9) as well as the absence of residual acyl donor (no ethyl spots visible), which was confirmed by GC-MS analysis of the hexane supernatant. Saponification of the hexanoylated compounds was first tested but led to very tedious workup procedures and unsatisfying errors in the quantification of the resulting hexanoic acid. We thus turned to a transmethylation process, well-known in the field of triglyceride chemistry. Transmethylation aims to cleave the ester bonds and thus release the hexanoate moieties as methyl esters (Supplementary Figure S10). Thereafter, transmethylation followed by GC quantification established that 184 µmol of methyl hexanoate was released per Gram of reacted DHPs. This value, compared to the aliphatic hydroxyl content of the starting material determined by ³¹P NMR (3.32 mmol/g, Table 1 Entry 5), indicates a hexanoylation yield of 5.5% mol of the aliphatic hydroxy groups; nevertheless, such a calculation underestimates the yield as it does not take into account the occurrence of side reactions decreasing the amount of available targeted aliphatic hydroxy groups and does not discriminate primary from secondary hydroxyl content, which may be equivalent according to the putative DHPs structure. Structural modification of DHPs was not investigated further as it was not the aim of this work; however, it appears more reliable to quantify acylation through transmethylation rather than structural analysis (³¹P NMR or SEC), as we expected initially. For these reasons, the hexanoylation yield will be estimated by transmethylation and expressed in µmol/g. In initial attempts, we compared the quantification through peak area measurement inserted in the equation [molar concentration] = f ([peak area]) determined by injection of CS at different concentrations. Nevertheless, the use of an internal standard of close structure (methyl heptanoate, MQS) led to higher repeatability and was preferred in the remainder of our study, as described in the material and methods section.

This method is indirect and it can be applied to any acylated substrate of complex structure, as far as the residual unreacted acyl donor is quantified or, when possible, carefully eliminated. Therefore, this method was retained for hexanoylation yield estimation in the remainder of the study, aiming to transfer this biocatalyzed selective hexanoylation process to technical lignins and derived fractions.

To deal with less heterogeneous samples than the complex technical lignins Protobind 1000, sequential solvent fractionation was applied in order to obtain more defined lignin fractions (Lu et al., 2021; Jin et al., 2022). The first step involved ethyl acetate in order to eliminate lower molar mass compounds F1 and to focus on polymeric chains, as they encounter more difficulties in accessing enzyme active sites. The second step involved MEK in order to extract a substrate soluble F2 in the reaction media (MEK), as well as in analytical solvents, by eliminating the insoluble residue F3. These three fractions showed lower polydispersities than Protobind 1000 and exhibited increasing molar masses as follows: M_w (F1) < M_w (Protobind 1000) < M_w (F2) < M_w (F3), while the phenol content decreased in the opposite manner (Figure 8; Table 2).

The hexanoylation of soda lignins fraction F2 (lignins fraction soluble in MEK but insoluble in ethyl acetate) was thus conducted in MEK with ethyl hexanoate. The first attempt was conducted based on previous experimental parameters: 5 mass equivalent of ethyl hexanoate and 10% wt CAL-B regarding lignins were used in MEK (10 g/L) under reflux for 24 h, which led to a hexanoylation yield of 85 μmol/g. The same experiment conducted in absence of CAL-B showed no hexanoylation; this control informed us that CAL-B activity was responsible for the reaction and confirmed the accuracy of hexanoylation yield quantification through transmethylation followed by GC-MS, inducing no overestimation. In order better understand the reactivity of this new system, different key parameters have been studied to optimize enzymatic hexanoylation: 1) ethyl hexanoate:lignins weight ratio, 2) reaction time, 3) lignins concentration, 4) enzyme load, and 5) the solvent used as the reaction medium. In all following sections, different assays were compared based on their resulting hexanoylation yields determined through transmethylation followed by GC-MS. Some of the following studies were conducted in parallel; therefore, the fixed parameters are not necessarily optimized, aiming to detect tendencies rather than defining an optimized process.