Salidroside was obtained from Aladdin (Shanghai, China). The fatty acid vinyl esters as acyl donors were provided by Tokyo Chemical Industry, TCI (Shanghai) Development Co., Ltd. All microbial strains (Pseudomonas fluorescens, Pseudomonas stutzeri, Pseudomonas aeruginosa, Aspergillus niger, Aspergillus oryzae and Rhizopus oryzae) were obtained from Guangdong Institute of Microbiology (Guangzhou, China). Other chemicals were of analytical grade.

Bacterial and fungal strains were cultivated as described previously (Yang L. M. et al., 2020). The bacterial strains were activated in the medium contained 1% sucrose, 1% beef extract, 1% peptone, 0.5% NaCl, 0.5% K₂HPO₄ and 0.02% MgSO₄ 7H₂O. The fungal strains were activated on potato dextrose agar (PDA) medium. Then, the activated bacterial suspension and fungal spore suspension were inoculated into the same fermentation broth respectively, which contained 0.1% soybean oil, 0.2% tryptone, 0.5% (NH₄)₂SO₄ and 0.02% MgSO₄ 7H₂O. After cultivation, the cells were collected by filtration or freeze-centrifugation and freeze-dried for 24 h.

In a typical experiment, salidroside (20 mM), vinyl caprylate and whole-cell catalyst preparation were added in organic solvents and incubated at 200 rpm. The samples were collected at predetermined time intervals and detected by high-performance liquid chromatography (HPLC). The control experiments without whole-cell catalysts displayed no acylation action. The conversion (C) was measured as the ratio of transformed to initial salidroside. The initial reaction rate (V₀) refers to the substrate consumption per unit time in the initial stage, in which the substrate concentration decreased linearly with the reaction time. The experiments were carried out in triplicates.

The operational stability of Aspergillus oryzae cells during the batch reaction was investigated. The reaction was carried out in anhydrous acetone containing 20 mM salidroside, 200 mM vinyl caprylate, 20 mg/ml whole-cell biocatalysts for 12 h at 40°C and 200 rpm. After each batch synthetic reaction, whole-cell biocatalysts were separated by filtration, washed with reaction medium and utilized in the next fresh reaction.

All samples were analyzed by RP-HPLC on an Agilent Zorbax Eclipse Plus C18 column (4.6 mm × 250 mm, 5 μm) using Shimadzu LC-200C pump with the DAD detector at 275 nm. The flow rate was 1.0 ml/min. The mobile phase contained a mixture of methanol and water at a 1.0 ml/min flow rate. The volumetric ratio of methanol to water and the retention times for its ester derivatives were 60/40 and 3.018 min (salidroside-6'-acetate), 60/40 and 4.278 min (salidroside-6'-butyrate), 80/20 and 3.265 min (salidroside-6'-hexanoate), 80/20 and 4.355 min (salidroside-6'-caprylate), 80/20 and 6.782 min (salidroside-6'-decanoate), 80/20 and 6.621 min (salidroside-6'-undecenoate), 90/10 and 4.760 min (salidroside-6'-laurate), 90/10 and 6.623 min (salidroside-6'-myristate), 90/10 and 9.981 min (salidroside-6'-palmitate), respectively.

All the synthetic products were purified through flash column chromatography using a mixture of ethyl acetate and petroleum ether as the mobile phase. The structural identification of the ester derivatives was determined by ¹³C NMR and ¹H NMR (Bruker DRX-400 NMR Spectrometer) at 100 MHz and 400 MHz, respectively, with DMSO-d₆ being the solvent. All the NMR spectroscopic results are shown in the supplementary information.

Several lipase-producing strains including bacteria and fungi were screened as the biocatalysts for the synthesis of salidroside caprylate (Table 1). Results revealed that all the tested strains showed different catalytic activities. Aspergillus oryzae and Pseudomonas aeruginosa demonstrate significant outstanding catalytic activity, affording 95.72% and 93.64% conversion after 24 h, respectively, and both gave substantially superior result to those obtained with the other strains that were tested. The reaction catalyzed by Rhizopus oryzae, Pseudomonas fluorescens and Pseudomonas stutzeri proceeded with 20.83%, 11.53% and 9.47 conversion rate after 24 h, respectively. These results indicated that catalytic properties of different microorganisms in the same induction medium were different, which was consistent with previous studies (Xin et al., 2017; Hao et al., 2021). Interestingly, in the regioselective synthesis of helicid aliphatic esters, Pseudomonas aeruginosa exhibited excellent catalytic performance, while Aspergillus oryzae displayed no catalytic activity (Yang W. et al., 2019). In the synthesis of esculin esters, Pseudomonas stutzeri demonstrated the highest catalytic activity (Hao et al., 2021). These results indicated that the whole cell catalysts from different sources have certain substrate specificity. In addition, whole cell catalysts from Pseudomonas aeruginosa displayed comparable catalytic activity to an immobilized lipase Novozyme 435 in synthesis of salidroside esters (Yang et al., 2021). From the results in Table 1, it was clear that salidroside was the appropriate substrate for the whole cell catalysts from Aspergillus oryzae and Pseudomonas aeruginosa.

High regioselectivity is one of the remarkable characteristics of biocatalysts. It is worth mentioning that all microbial whole-cell catalysts tested displayed absolute regioselectivities toward the 6′-hydroxyl of glucose in salidroside, which was comparable to the excellent selectivity of an immobilized lipase Novozyme 435 with salidroside 6′- caprylate being the sole ester product (Yang et al., 2021). Similarly, we recently found that Aspergillus oryzae whole cells exhibited excellent regioselectivities toward the primary hydroxyl group at allose moiety of helicid, a structural analogue of salidroside (Yang R. L. et al., 2020). The reason may be that the primary hydroxyl group has the less steric hindrance effect than the other hydroxyls. Based on the greater reaction efficiency and regioselectivity, Aspergillus oryzae strain was shortlisted for the subsequent investigation.

In order to further improve the reaction efficiency, the synthesis of salidroside caprylate was used as a model reaction to investigate the influence of several key variables (reaction medium, molar ratio of substrate, whole-cell dosage) on the reaction. Like enzyme-mediated biotransformation, the nature of reaction medium has a significant effect on the whole-cell biocatalysis, which can impact the biocatalyst activity and stability. The presence of water may cause the hydrolysis of both the ester products and the acyl donors (vinyl esters), so several traditional organic solvents with different polarities from -0.23 to 1.85 were selected as reaction media (Table 2). Salidroside as a phenolic glycoside has high solubility in strongly polar solvents such as dimethyl sulfoxide (DMSO) and dimethylformamide (DMF) with 77.0 mM and 101.2 mM, respectively, while relatively low solubility in less polar organic solvents (25.3–42.1 mM). As speculated, Aspergillus oryzae has no catalytic activity in DMSO and DMF with strong polarity, which may inactivate lipases by destroying the membrane of whole-cell catalysts (Yang R. L. et al., 2019; Yang L. M. et al., 2020; Hao et al., 2021). Among the solvents tested (Figure 2), good conversions were obtained in 2-methyltetrahydrofuran (86.97%), tert-butanol (78.92%) and tetrahydrofuran (72.35%). And the highest conversion efficiency (95.72%) and initial reaction rate (9.30 mM/h) were obtained in acetone. Moreover, there was no significant correlation between the catalytic activity of the cells and the log P of the tested solvents, the commonly used solvent parameter in non-aqueous enzymology. As shown in Figure 1A, the reaction rate accelerated obviously (9.3 mM/h to 15.36 mM/h) with increasing the whole-cell dosage from 10 to 30 mg/ml, and then no obvious improvements occurred with further increasement of catalyst amount. The molar ratio of vinyl caprylate to salidroside had a great influence on the initial reaction rate and the maximal conversion (Figure 1B), which improved significantly with the increase of vinyl caprylate concentration up to 10 equivalents of salidroside concentration. A high conversion (>99%) and good initial reaction rate (15.36 mM/h) could be acquired with the molar ratio of vinyl caprylate to salidroside as 10. Figure 1C depicted that Aspergillus oryzae whole-cells had good biocatalysis performance in the temperature range of 30°C–55°C. Figure 1D showed the reaction process of salidroside acylation with vinyl caprylate under the above-obtained conditions. Salidroside conversion increased sharply at the initial stage of the reaction with 95% at 6 h, and then a slower rise, reaching 99% at 12 h.

Consider the above data comprehensively, the optimum conditions of reaction medium, Aspergillus oryzae amount, molar ratio of vinyl caprylate to salidroside and reaction temperature were acetone, 30 mg/ml, 10°C and 40°C, respectively, with initial rate of 15.36 mM/h and the highest substrate conversion of 99%. Additionally, the regioselectivity of the reaction maintained 99% under the above conditions.

The reusability of biocatalyst is of great importance to the cost efficiency in industrial production. As shown in Figure 2, the Aspergillus oryzae whole cells remained 51% of its original activity after six consecutive batches. And the Aspergillus oryzae lost only 25% of its activity after three reaction cycles, which might be that the whole-cell provided a natural protective environment for intracellular enzyme with acylation activity.

The synthesis of various salidroside aliphatic esters mediated by Aspergillus oryzae cells was explored in the above-mentioned optimal reaction conditions, with fatty acid vinyl esters of different chain lengths as acyl donors (Table 3). It is worth mentioning that 6′-monoesters of salidroside were exclusively achieved in all cases by NMR and HPLC analysis, which was similar to that in helicid esters synthesis catalyzed by Aspergillus oryzae cells, also affording an excellent selectivity for 6′-OH at allose moiety (Yang R. L. et al., 2020). However, in the synthesis of acetyl ester (a short alkyl acyl group) of salidroside by the commercial lipase Novozyme 435, 6′-O-acyl salidroside and 3′,6′-O-diacyl salidroside were the main two products, also with a small amount (<5%) of 2′,6′-O-diacryloyl salidroside (Yu et al., 2008). Thus Aspergillus oryzae cells exhibited superior regioselectivity to Novozyme 435 in the synthesis for short-chain aliphatic ester of salidroside, which highlighted the excellent regioselectivity of Aspergillus oryzae cells.

As shown in Table 3, Aspergillus oryzae cells exhibited excellent catalytic activities with 96%–99% conversions in the salidroside acylation with different chain-length acyl donors. Furthermore, the initial reaction rate increased from 3.37 mM/h to 17.31 mM/h with the elongation of chain length of acyl donors from C2 to C11, which indicated that the interaction between medium-chain acyl groups and hydrophobic acyl binding sites of intracellular acylase was stronger than that of short-chain groups. Nevertheless, the initial reaction rate decreased with further extending chain length from C12 to C16, owing to the higher steric hindrance of the longer-chain acyl donors. Previous studies have found that the biocatalysts from different sources showed different catalytic properties for the different acyl donors due to the specific structure of the lipase active site and the acyl (Wang and Zong, 2009; Yang et al., 2013; Li X. F. et al., 2018; Yang W. et al., 2019). For example, the Pseudomonas aeruginosa cells were more specific toward the medium-chain acyl donors, as well as Candida antarctica lipase B (Novozyme 435) and Thermomyces lanuginosus lipase (Lipozyme TLL) (Wang and Zong, 2009; Yang et al., 2013; Yang R. L. et al., 2019), while the whole cells of Candida parapsilosis was most efficient for the short chain acyl donor, vinyl propionate. (Li G. Y. et al., 2018).