From Hamamelitannin Synthesis to the Study of Enzymatic Acylations of D-Hamamelose

The bioactive natural substance, hamamelitannin, was effectively synthesized in two ways. The chemical acylation of 2,3-O-isopropylidene-α,β-D-hamamelofuranose promoted by Bu2SnO using 3,4,5-tri-O-acetylgalloyl chloride, followed by the deprotection provided hamamelitannin in 79%. Pilot enzymatic benzoylation of D-hamamelose using vinyl benzoate (4 equiv.) and Lipozyme TL IM as a biocatalyst in t-butyl methyl ether (t-BuMeO) gave mainly benzoylated furanoses (89%), of which tribenzoates reached (52%). Enzymatic galloylation of 2,3-O-isopropylidene-α,β-D-hamamelofuranose with vinyl gallate under the catalysis of Lipozyme TL IM in t-butyl alcohol (t-BuOH) or t-BuMeO provided only the 5-O-galloylated product. The reaction in t-BuMeO proceeded in a shorter reaction time (61 h) and higher yield (82%). The more hydrophobic vinyl 3,4,5-tri-O-acetylgallate in the same reactions gave large amounts of acetylated products. Vinyl gallate and triacetylgallate in the enzymatic acylation of D-hamamelose with Lipozyme TL IM in t-BuMeO yielded 2′,5-diacylated hamamelofuranoses in a yield below 20%. The use of other vinyl gallates hydrophobized by methylation or benzylation provided 2′,5-diacylated hamamelofuranoses in good yields (65–84%). The reaction with silylated vinyl gallate did not proceed. The best results were obtained with vinyl 2,3,5-tri-O-benzyl gallate, and the only product, 2′,5-diacylated hamamelofuranoside precipitated from the reaction mixture (84% in 96 h). After debenzylation, hamamelitannin was obtained an 82% yield from hamamelose in two steps. This synthesis is preparatively undemanding and opens the way to multigram preparations of bioactive hamamelitannin and its analogues.


Introduction
Tannins are polyphenolic plant secondary metabolites that have enormous structural diversity [1]. Hydrolyzable tannins are often isolated from plants for their remarkable therapeutic effects [2]. Their structure generally consists of a central sugar core, typically a glucose unit, to which galloyl groups, often meta-depsidically bonded (gallotannins) or C-C bonded dehydrodigalloyl units (ellagitannins), are attached [3]. The total synthesis of such compounds tends to be complex [4][5][6].
Extracts and distillates of witch hazel bark, twigs, and leaves containing 1 are widely used as components of skin care products and in dermatological treatment of sunburn, irritated skin, acne, atopic eczema, and to promote wound healing through anti-inflammatory effects [14][15][16][17][18]. Pure gallotannin 1 inhibits the activity of α-TNF (tumor necrosis factor) [19], Extracts and distillates of witch hazel bark, twigs, and leaves containing 1 are used as components of skin care products and in dermatological treatment of s irritated skin, acne, atopic eczema, and to promote wound healing through anti matory effects [14][15][16][17][18]. Pure gallotannin 1 inhibits the activity of α-TNF (tumor factor) [19], autoactivation of plasma hyaluronan-binding protein [20] and exhib scavenging and protective activity against cell damage by active oxygen and p [21][22][23]. It also appears to be a promising chemotherapeutic agent, which might in the treatment of colon cancer without compromising the viability of healthy co [24]. Very recently, 1 [25][26][27], cyclodextrin-hamamelitannin complexes [28], or d synthetic analogues of hamamelitannin analogues [29][30][31] in combination with an were studied as perspective suppressors of staphylococcal infections by inhibiti lence of bacterial biofilms through quorum sensing mechanisms.
The antiviral efficacy against influenza A virus and human papillomavirus of from Hamamelis virginiana bark extract has also been demonstrated [32]. Gallotann also become a relatively successful molecule in various in silico screening model at studying the inhibition of proteins important in the process of carcinogenesis sclerosis, or SARS-CoV-2 disease [33][34][35][36].
Despite numerous reports on the medical effects of hamamelitannin, only o synthesis of 1 has been published so far, as early as 1969 [37]. The authors obta target di-O-acyl-glycoside in only 22% yield by conventional acylation of the p benzyl β-D-hamamelofuranoside with tri-O-benzylgalloyl chloride in a pyridine/ drofuran mixture. Acylation proceeded for 71 h at −40° to rt and afforded three p The main product was hydrogenated over 10% palladium on charcoal to give com 1 with a yield of 58%. The starting branched sugar D-hamamelose (2-C-hydroxym ribose) was prepared from D-arabinose in several steps via methyl 3,4-O-isoprop β-D-erythro-pentopyranosidulose [38].
Regioselective acylation of polyhydroxylated molecules like sugars is often lenge [39]. The solution is to use biocatalysts, especially lipases, and perform e catalyzed acylation [40,41]. Lipases tolerate a wide range of substrates and are work in an aqueous environment as well as in organic solvents [42]. Although th primarily evolved to hydrolyze triacylglycerols with long fatty acids, some of th tolerate phenolic substrates, making them similar in reactivity to feruloyl estera The antiviral efficacy against influenza A virus and human papillomavirus of tannins from Hamamelis virginiana bark extract has also been demonstrated [32]. Gallotannin 1 has also become a relatively successful molecule in various in silico screening models aimed at studying the inhibition of proteins important in the process of carcinogenesis, atherosclerosis, or SARS-CoV-2 disease [33][34][35][36].
Despite numerous reports on the medical effects of hamamelitannin, only one total synthesis of 1 has been published so far, as early as 1969 [37]. The authors obtained the target di-O-acyl-glycoside in only 22% yield by conventional acylation of the prepared benzyl β-D-hamamelofuranoside with tri-O-benzylgalloyl chloride in a pyridine/tetrahydrofuran mixture. Acylation proceeded for 71 h at −40 • to rt and afforded three products. The main product was hydrogenated over 10% palladium on charcoal to give compound 1 with a yield of 58%. The starting branched sugar D-hamamelose (2-C-hydroxymethyl-D-ribose) was prepared from D-arabinose in several steps via methyl 3,4-O-isopropylidene-β-Derythro-pentopyranosidulose [38].
Regioselective acylation of polyhydroxylated molecules like sugars is often a challenge [39]. The solution is to use biocatalysts, especially lipases, and perform enzymecatalyzed acylation [40,41]. Lipases tolerate a wide range of substrates and are able to work in an aqueous environment as well as in organic solvents [42]. Although they have primarily evolved to hydrolyze triacylglycerols with long fatty acids, some of them also tolerate phenolic substrates, making them similar in reactivity to feruloyl esterases [43]. In aprotic organic solvents, they can catalyze esterifications or transesterifications. The choice of a suitable solvent in reactions is important for their speed as well as selectivity [44,45]. Several reaction steps can be saved by the appropriate selection of the biocatalyst and reaction conditions in the acylation of carbohydrates. They work under mild conditions, so reactions of this type do not consume much energy. Moreover, they are commonly commercially available and can be used multiple times.
In this study, we report two simpler and more efficient syntheses of hamamelitannin 1 from different starting compounds by conventional as well as lipase-promoted galloylation. The regioselectivity of various methods of galloylation of hamamelofuranose, including the enzymatic procedure, was studied.

Enzymatic Benzoylation of 8
D-Hamamelose (0.36 g, 2 mmol) was suspended in t-BuMeO (40 mL). Molecular sieves 4Å (2g), vinyl benzoate (1.11 mL, 4 equiv.), and Lipozyme TL IM (0.4 g) were added. The tightly closed reaction mixture was shaken on a vibrating shaker at 450 rpm in an incubator at 37 • C. After 50 h, the reaction was filtered through Celite 545, the filter cake was washed several times with EtOAc, and the filtrate was concentrated. The reaction mixture was purified on a silica gel column eluted with toluene/EtOAc (3:1→1:2). Several products were obtained during the elution in the order: 13 (3%), 11c (35%), 12c (17%), 9c (37%), and 10c (1%).  Table 2, the reaction was filtered through Celite 545, the filter cake was washed several times with EtOAc, and the filtrate was concentrated. The reaction mixture was purified on a silica gel column eluted with toluene/EtOAc (3:1→1:2). The data of the products are presented in Table 2.   Table 3. HRMS (ESI): m/z calcd for   Diacylated compound 6a (0.1 mmol) or triacylated compound 7 (0.1 mmol) were dissolved in CH 3 CN (2 mL), and 3M HCl (2 mL) was added. The reaction mixture was stirred at laboratory temperature for 72 h. After the reaction, the organic solvent was removed under reduced pressure and the aqueous residue was extracted with ethyl acetate (3 × 50 mL). The organic layer was washed with brine (3 × 50 mL), dried (Na 2 SO 4 ), and concentrated under reduced pressure. The product 1 from the deprotection of digallate 6a was obtained as a pure compound in high yield (94%), while the reaction mixture obtained from the deprotection of compound 7 contained the product 1 and gallic acid and after chromatography on silica gel (0.1%CH 3 COOH in EtOAc) only 56% of product 1 was obtained.

Debenzylation of 9g
Diacylated hamamelofuranose 9g (0.206 g, 0.02 mmol) was dissolved in MeOH (10 mL) and 10% Pd/C (0.06 g) was added. The reaction mixture was intensively stirred at room temperature (25 • C) under hydrogen atmosphere for 18 h and then filtered through Celite 545. After washing the celite cake with MeOH, the filtrate was concentrated to give 0.094 g (97%) of product 1.

Synthesis of Hamamelitannin by Acylation of Acceptor 2 Prepared from D-Ribose
D-Ribose was used as the starting material for the preparation of isopropylidenated D-hamamelofuranose 2 (Scheme 1). Treatment of D-ribose with anhydrous acetone in the presence of a catalytic amount of concentrated sulfuric acid yielded the corresponding 2,3-acetonide 2 [46] in 91% yield. The reaction of 2 with aqueous formaldehyde in the presence of potassium carbonate in MeOH introduced the branching hydroxymethyl group through a Ho crossed aldol reaction [47].
x FOR PEER REVIEW 12 of 20 Scheme 1. Conventional and enzymatic acylation of acceptor 2 using acetylated galloyl donors. Reagents and conditions are shown in Table 1.
Recently, we have been intensively dealing with enzymatic acylations of sugars by various phenolic acid donors [43,48,60,61]. The commercial lipase Lipozyme TL IM was shown to be the most effective catalyst in terms of its substrate specificity, reactivity, and stability in these reactions. The disadvantage of its use was the longer reaction time, reaching several days. We have, therefore, galloylated 2 under our optimised conditions for galloylation of methyl β-D-glucopyranoside [61] catalysed by Lipozyme TL IM using 3 equivalents of vinyl gallate 4b as an acyl donor at 37 °C and dry t-butyl alcohol (t-BuOH) as a solvent (Scheme 2). The 2,3-isopropyl-D-hamamelofuranose 2 was surprisingly better accepted by the enzyme than methyl β-D-glucopyranoside. The reaction proceeded regioselectively to the 5-OH position of compound 2 and the maximum yield (66%) of monogallate 5b was obtained after 242 h (Scheme 2). t-BuOH is quite a polar solvent (octanol/water partition coefficient as log Pow: 0.30). According to our experience, the lipasecatalyzed transesterification proceed faster in less polar solvents such as for example tbutyl methyl ether (t-BuMeO) (octanol/water partition coefficient as log Pow: 1.06).
The same reaction of 2 with 3 equiv. of 4b was repeated in t-BuMeO. Under these conditions, after 61 h it gave 5-O-gallate 5b in a yield of 82%. The reaction was significantly faster with a higher yield and the product was again the monoacylated product 5b (Scheme 2). Regarding the increase in reaction rate, we can hypothesize that t-BuMeO opens the hydrophobic lid in the active center of Lipozyme TL IM more efficiently than t-BuOH, and the substrate binding site is more accessible [62]. In the initial stage of the work, an acetyl-protecting group was selected for the galloyl moiety because the deacetylation of phenols and the removal of the isopropylidene group from the sugar can be performed in one step under acidic conditions. In the first step, standards of the desired triacetylgallates of hamamelofuranose 8 were chemically prepared. After the preparation of 3,4,5-tri-O-acetylgalloyl chloride (3) from gallic acid in two steps, we looked for optimal conditions for the acylation of hamamelose 2. Acetylated phenols are sensitive to both acidic and basic conditions, but deprotected gallates can oligomerize under basic conditions. Therefore, in the synthesis, it was necessary to find slightly basic or neutral conditions that allow 2 ,5-di-O-acylation of compound 2 and to which the acetyl groups are inert. Our secondary goal was to regioselectively achieve 2 ,5-di-O-acylated hamamelose 6a in maximum yields, using a minimum of equivalents of acyl reagent 3. A high content of monoacylated or triacylated products was undesirable.
Several conventional acylation methods have been used, operating from mildly basic to neutral conditions. The results are summarized in Table 1. In initial experiments, we investigated mild basic reaction conditions working with acetylated galloyl chloride 3, which has been reliably verified in many acylation reactions [53][54][55]. A mixture of two bases-Et 3 N and DMAP (1 equiv. and 0.25 equiv. relative to 1 equiv. of acyl) was used in dichloromethane (Table 1, Entry 1, 2). Using 2.2 equivs of acyl reagent 3 and appropriate equivalents of bases led to the mixture of per-O-acetylated trigallate 7, digallate 6a, and relatively high content of monogallates (24%), while conditions for theoretical attachment of three acyl groups (3.3 equivs of 3) led to 7 as a major product (81%). Digallate 6a and monogallates were formed in minimal quantities.
In one of our previous investigations, we studied ZnO as a convenient catalyst in the 4-O-acetylferuloylation of glycosides [58]. Therefore, we have examined tri-Oacetylgalloylation of 2 in CH 2 Cl 2 ( Table 1, Method C, Entry 6), but we isolated a mixture of acetylated monogallates as the main product fraction. Better results were obtained when we used CH 3 CN as the reaction medium (Table 1, Entry 7, 8). With elevating the reaction temperature to 40 • C, ZnO equivalents and reaction time lead to trigallate 7 a digallate 6a in the summary yield 83% (Table 1, Entry 8). It is interesting that we did not observe compound 7 as the β-anomer and 7α was the exclusive tri-O-acylated product under all conditions; this suggests the neighboring group effect by the C-2 acyloxymethyl group in the rigid 2,3-isopropylidenated furanose ring, as has been already reported [59].
Recently, we have been intensively dealing with enzymatic acylations of sugars by various phenolic acid donors [43,48,60,61]. The commercial lipase Lipozyme TL IM was shown to be the most effective catalyst in terms of its substrate specificity, reactivity, and stability in these reactions. The disadvantage of its use was the longer reaction time, reaching several days. We have, therefore, galloylated 2 under our optimised conditions for galloylation of methyl β-D-glucopyranoside [61] catalysed by Lipozyme TL IM using 3 equivalents of vinyl gallate 4b as an acyl donor at 37 • C and dry t-butyl alcohol (t-BuOH) as a solvent (Scheme 2). The 2,3-isopropyl-D-hamamelofuranose 2 was surprisingly better accepted by the enzyme than methyl β-D-glucopyranoside. The reaction proceeded regioselectively to the 5-OH position of compound 2 and the maximum yield (66%) of monogallate 5b was obtained after 242 h (Scheme 2). t-BuOH is quite a polar solvent (octanol/water partition coefficient as log Pow: 0.30). According to our experience, the lipase-catalyzed transesterification proceed faster in less polar solvents such as for example t-butyl methyl ether (t-BuMeO) (octanol/water partition coefficient as log Pow: 1.06). Scheme 1. Conventional and enzymatic acylation of acceptor 2 using acetylated galloy agents and conditions are shown in Table 1.
Recently, we have been intensively dealing with enzymatic acylations o various phenolic acid donors [43,48,60,61]. The commercial lipase Lipozyme shown to be the most effective catalyst in terms of its substrate specificity, rea stability in these reactions. The disadvantage of its use was the longer rea reaching several days. We have, therefore, galloylated 2 under our optimised for galloylation of methyl β-D-glucopyranoside [61] catalysed by Lipozyme T 3 equivalents of vinyl gallate 4b as an acyl donor at 37 °C and dry t-butyl alcoh as a solvent (Scheme 2). The 2,3-isopropyl-D-hamamelofuranose 2 was surpris accepted by the enzyme than methyl β-D-glucopyranoside. The reaction proc oselectively to the 5-OH position of compound 2 and the maximum yield (6 nogallate 5b was obtained after 242 h (Scheme 2). t-BuOH is quite a polar tanol/water partition coefficient as log Pow: 0.30). According to our experience catalyzed transesterification proceed faster in less polar solvents such as for butyl methyl ether (t-BuMeO) (octanol/water partition coefficient as log Pow: The same reaction of 2 with 3 equiv. of 4b was repeated in t-BuMeO. U conditions, after 61 h it gave 5-O-gallate 5b in a yield of 82%. The reaction was s faster with a higher yield and the product was again the monoacylated (Scheme 2). Regarding the increase in reaction rate, we can hypothesize tha opens the hydrophobic lid in the active center of Lipozyme TL IM more efficie BuOH, and the substrate binding site is more accessible [62]. The same reaction of 2 with 3 equiv. of 4b was repeated in t-BuMeO. Under these conditions, after 61 h it gave 5-O-gallate 5b in a yield of 82%. The reaction was significantly faster with a higher yield and the product was again the monoacylated product 5b (Scheme 2). Regarding the increase in reaction rate, we can hypothesize that t-BuMeO opens the hydrophobic lid in the active center of Lipozyme TL IM more efficiently than t-BuOH, and the substrate binding site is more accessible [62].

Synthesis of Hamamelitannin by Enzymatic Acylation of D-Hamamelose
Promising results with enzymatic galloylation of compound 2 prompted our increased efforts to prepare hamamelitannin 1 via direct enzymatic acylation of D-hamamelose (8). D-hamamelose, although a rare branched sugar, is commercially available. One of the ways to prepare 8 is molybdic acid-catalyzed isomerization of D-fructose by Bílik reaction [63,64].

Synthesis of Hamamelitannin by Enzymatic Acylation of D-Hamamelose
Promising results with enzymatic galloylation of compound 2 prompted our increased efforts to prepare hamamelitannin 1 via direct enzymatic acylation of D-hamamelose (8). D-hamamelose, although a rare branched sugar, is commercially available. One of the ways to prepare 8 is molybdic acid-catalyzed isomerization of D-fructose by Bílik reaction [63,64].
When proceeding the reaction with 3 equivalents of acetylated vinyl gallate 4a and gallate 4b under similar conditions (Lipozyme TL IM, t-BuMeO, 37 • C), the desired products were obtained in both cases, however, in very low yields. Acetylated 4a gave 13% of diacyl 9a after 41 h and the rest were various UV-inactive acetylated products. The reaction with vinyl gallate 4b was allowed to react for a longer time (292 h), and again obtained only 15% of the acylation product 1 (Scheme 4). The monoacylated product's content was not visible on TLC. In both cases, we did not observe the presence of aromatic triacylated products. Vinyl gallate 4b was not sufficiently reactive in t-BuOH and no galloylation product with hamamelose 8 was observed even after 10 days.
The structure of the starting compounds in lipase-catalyzed esterifications or transesterifications, especially in the case of phenolic compounds, influences the course of the reaction [48,66]. Therefore, we decided to test other more hydrophobic and stable vinyl esters of gallic acid derivatives in the investigated enzyme reaction. Vinyl esters of syringic acid (4d) and 3,4,5-trimethoxybenzoic acid (4e) were prepared according to our previous work [48]. Two new vinyl esters were also prepared-3,4,5-tri-O-(t-butyldimethylsilyl)gallate (4f) and 3,4,5-tri-O-benzylgallate (4g). The silyl derivative 4f was prepared by silylation of vinyl gallate and the benzylated derivative 4g was prepared by transesterification of 3,4,5-tribenzylgallic acid with vinyl acetate. Silyl and benzyl protective groups are widely used in the syntheses, as they can be effectively removed under mild, relatively neutral conditions [67].
Biomolecules 2023, 13, x FOR PEER REVIEW 14 of 2 products. Vinyl gallate 4b was not sufficiently reactive in t-BuOH and no galloylatio product with hamamelose 8 was observed even after 10 days. The structure of the starting compounds in lipase-catalyzed esterifications or trans esterifications, especially in the case of phenolic compounds, influences the course of th reaction [48,66]. Therefore, we decided to test other more hydrophobic and stable viny esters of gallic acid derivatives in the investigated enzyme reaction. Vinyl esters of syrin gic acid (4d) and 3,4,5-trimethoxybenzoic acid (4e) were prepared according to our prev ous work [48]. Two new vinyl esters were also prepared-3,4,5-tri-O-(t-butyldimethyls lyl)gallate (4f) and 3,4,5-tri-O-benzylgallate (4g). The silyl derivative 4f was prepared b silylation of vinyl gallate and the benzylated derivative 4g was prepared by transesterif cation of 3,4,5-tribenzylgallic acid with vinyl acetate. Silyl and benzyl protective group are widely used in the syntheses, as they can be effectively removed under mild, relativel neutral conditions [67].
The studied enzymatic acylations were performed according to previously imple mented conditions. (1 mmol of 8, 3 equiv. of vinyl ester, 0.2 g of Lipozyme TL IM, 20 m of solvent, 37 °C). To compare the effect of solvent on the reaction time, the compositio of products. and product yields with individual acyl donors; these were carried out in t BuOH as well as in t-BuMeO. The reaction with the benzylated gallate 4g was also carrie out in CH3CN. Acylations were monitored by TLC chromatography. The reactions wer stopped when the concentration of the products no longer increased. The lipase and mo lecular sieves were filtered off and the filtrate was purified by chromatography after con centration.
The results of the reaction (Scheme 5) summarized in Table 2 showed that Lipozym TL IM catalyzes acylations with all acyl donors except silylated gallate 4f. In the case o 4f, we did not observe any product in both solvents even after hundreds of hours (Tabl 2, entries 5, 6). Acyl donor 4f is probably too large to interact with the active site of th enzyme. Acylation with syringate 4d proceeded as within the longest reaction times (mor than 200 h), while hydrophobic 4e and 4g reacted faster (tens of hours). This is consisten with our previous experience [48], and it appears that the transesterification activity o Lipozyme TL IM, similarly to Lipolase 100T (both are lipases from Thermomyces lanugino sus), corresponds to the hydrolytic activity of type A feruloylesterase [68]. In general, re actions in t-BuOH proceeded slower and a higher quantity of monoacylated product were isolated (Table 2, entries 1, 3, 7). On the contrary, we have observed only negligibl amounts of monoacyl-hamameloses in t-BuMeO and mostly 2',5-di-O-acyls of α,β-D The studied enzymatic acylations were performed according to previously implemented conditions. (1 mmol of 8, 3 equiv. of vinyl ester, 0.2 g of Lipozyme TL IM, 20 mL of solvent, 37 • C). To compare the effect of solvent on the reaction time, the composition of products. and product yields with individual acyl donors; these were carried out in t-BuOH as well as in t-BuMeO. The reaction with the benzylated gallate 4g was also carried out in CH 3 CN. Acylations were monitored by TLC chromatography. The reactions were stopped when the concentration of the products no longer increased. The lipase and molecular sieves were filtered off and the filtrate was purified by chromatography after concentration.
The results of the reaction (Scheme 5) summarized in Table 2 showed that Lipozyme TL IM catalyzes acylations with all acyl donors except silylated gallate 4f. In the case of 4f, we did not observe any product in both solvents even after hundreds of hours (Table 2, entries 5, 6). Acyl donor 4f is probably too large to interact with the active site of the enzyme. Acylation with syringate 4d proceeded as within the longest reaction times (more than 200 h), while hydrophobic 4e and 4g reacted faster (tens of hours). This is consistent with our previous experience [48], and it appears that the transesterification activity of Lipozyme TL IM, similarly to Lipolase 100T (both are lipases from Thermomyces lanuginosus), corresponds to the hydrolytic activity of type A feruloylesterase [68]. In general, reactions in t-BuOH proceeded slower and a higher quantity of monoacylated products were isolated (Table 2, entries 1, 3, 7). On the contrary, we have observed only negligible amounts of monoacylhamameloses in t-BuMeO and mostly 2 ,5-di-O-acyls of α,β-D-hamamelofuranose (9d-e, 9g) (Scheme 5) were isolated (Table 2, entries 2 , 4, 9). The reactivity of trimethoxybenzoate 4e was similar to that observed for benzoate 4c. Products with acylated secondary hydroxyls were also observed. We isolated a significant proportion of triacyls in t-BuMeO for 4e ( Table 2, entry 4), and the highest yields of diacyls for t-BuOH were achieved using 4e ( Table 2, entry 3). This suggests that the reaction was directed towards the products that were more soluble in the used solvent. solvent, which serves as an adjuvant, were known from enzymatic syntheses of su acids esters [69]. The low solubility of 9g in the reaction medium probably protect unwanted acylations to the secondary hydroxyls. Reactions in more polar sol BuOH, CH3CN) proceeded more slowly with low product conversion. This could that the lipase does not have a favorable conformation for the bulky acyl 4g or vated by the products. We also examined the structural composition of our mixtures of monoacylat ucts isolated from reactions in t-BuOH. Theoretical structures in the mixture of h lose monoacyls are shown in Scheme 6. They could not be separated individu their mixtures were analyzed by NMR, and the H-1 and C-1 signals for in anomers and conformations were assigned with the support of literature data. T able literature and experimental values are listed in Table 3. These data demonst the initial acylation is not selective (Entries 9-14). In the mixture, 5-O-monoacyla predominating 2'-O-acylated hamameloses are visible. Generally, the primary 2' sition is sterically less favorable than the primary 5-OH position. It is possible acylation takes place first in the 2'-OH position if hamamelose is present in the medium in pyranose form. This acylated pyranose is then transformed into fura mutarotation. Scheme 6. Theoretical products of non-regioselective monoacylation of D-hamamelose.

Deprotections for Obtaining Hamamelitannin 1
In the final step, the tri-O-acetylgalloylated compounds 6a, 7 and benzylgal 9g were deprotected. Acidic conditions-3M HCl in CH3CN-were found to be s The acylation of 8 with the bulky benzylated acyl donor 4g had a different reaction course. The reaction in t-BuMeO proceeded with the highest yield (84%) of 2 ,5-diacyl 9g ( Table 2, entry 9). The desired main product 9g was the only one precipitated from the reaction mixture. After the end of the reaction, it was filtered together with the immobilized enzyme and molecular sieves. It was then washed with hot ethyl acetate from the filter cake. A similar reaction system, in which the starting monosaccharide (D-hamamelose), biocatalyst and the product (9g) were insoluble or almost insoluble in the reaction solvent, which serves as an adjuvant, were known from enzymatic syntheses of sugar fatty acids esters [69]. The low solubility of 9g in the reaction medium probably protects it from unwanted acylations to the secondary hydroxyls. Reactions in more polar solvents (t-BuOH, CH 3 CN) proceeded more slowly with low product conversion. This could indicate that the lipase does not have a favorable conformation for the bulky acyl 4g or is inactivated by the products.
We also examined the structural composition of our mixtures of monoacylated products isolated from reactions in t-BuOH. Theoretical structures in the mixture of hamamelose monoacyls are shown in Scheme 6. They could not be separated individually, but their mixtures were analyzed by NMR, and the H-1 and C-1 signals for individual anomers and conformations were assigned with the support of literature data. The available literature and experimental values are listed in Table 3. These data demonstrate that the initial acylation is not selective (Entries 9-14). In the mixture, 5-O-monoacylated and predominating 2 -O-acylated hamameloses are visible. Generally, the primary 2 -OH position is sterically less favorable than the primary 5-OH position. It is possible that the acylation takes place first in the 2 -OH position if hamamelose is present in the reaction medium in pyranose form. This acylated pyranose is then transformed into furanose via mutarotation.
rected towards the products that were more soluble in the used solvent.
The acylation of 8 with the bulky benzylated acyl donor 4g had a different rea course. The reaction in t-BuMeO proceeded with the highest yield (84%) of 2',5-diac (Table 2, entry 9). The desired main product 9g was the only one precipitated from reaction mixture. After the end of the reaction, it was filtered together with the imm lized enzyme and molecular sieves. It was then washed with hot ethyl acetate from filter cake. A similar reaction system, in which the starting monosaccharide (D-ham lose), biocatalyst and the product (9g) were insoluble or almost insoluble in the rea solvent, which serves as an adjuvant, were known from enzymatic syntheses of sugar acids esters [69]. The low solubility of 9g in the reaction medium probably protects it unwanted acylations to the secondary hydroxyls. Reactions in more polar solven BuOH, CH3CN) proceeded more slowly with low product conversion. This could ind that the lipase does not have a favorable conformation for the bulky acyl 4g or is in vated by the products. We also examined the structural composition of our mixtures of monoacylated p ucts isolated from reactions in t-BuOH. Theoretical structures in the mixture of ham lose monoacyls are shown in Scheme 6. They could not be separated individually their mixtures were analyzed by NMR, and the H-1 and C-1 signals for indiv anomers and conformations were assigned with the support of literature data. The a able literature and experimental values are listed in Table 3. These data demonstrate the initial acylation is not selective (Entries 9-14). In the mixture, 5-O-monoacylated predominating 2'-O-acylated hamameloses are visible. Generally, the primary 2'-OH sition is sterically less favorable than the primary 5-OH position. It is possible tha acylation takes place first in the 2'-OH position if hamamelose is present in the rea medium in pyranose form. This acylated pyranose is then transformed into furanos mutarotation. Scheme 6. Theoretical products of non-regioselective monoacylation of D-hamamelose.

Deprotections for Obtaining Hamamelitannin 1
In the final step, the tri-O-acetylgalloylated compounds 6a, 7 and benzylgalloylated 9g were deprotected. Acidic conditions-3M HCl in CH 3 CN-were found to be sufficient for simultaneous deisopropylidenation and deacetylation, while the galloyl groups were retained. The product 1 from the deprotection of diacyl 6a was obtained as a pure compound in high yield (94%, conditions (a) in Scheme 7), while the reaction mixture obtained from the deprotection of compound 7 contained 1 and gallic acid. Gallic acid originated from deacylation of anomeric gallate moiety sensitive to acidic conditions. The disadvantage of this method was the long reaction time (3 days at laboratory temperature). Debenzylation of 9g by reductive cleavage with molecular hydrogen over 10% Pd/C proceeded smoothly. After 18 h, the reaction mixture was filtered through Celite 545, and after the concentration of the filtrate, hamamelitannin 1 was obtained with a 97% yield and satisfactory purity (conditions (b), Scheme 3).
for simultaneous deisopropylidenation and deacetylation, while the galloyl groups were retained. The product 1 from the deprotection of diacyl 6a was obtained as a pure compound in high yield (94%, conditions (a) in Scheme 7), while the reaction mixture obtained from the deprotection of compound 7 contained 1 and gallic acid. Gallic acid originated from deacylation of anomeric gallate moiety sensitive to acidic conditions. The disadvantage of this method was the long reaction time (3 days at laboratory temperature). Debenzylation of 9g by reductive cleavage with molecular hydrogen over 10% Pd/C proceeded smoothly. After 18 h, the reaction mixture was filtered through Celite 545, and after the concentration of the filtrate, hamamelitannin 1 was obtained with a 97% yield and satisfactory purity (conditions (b), Scheme 3).

Conclusions
2,3-Isopropylhamamelofuranose 2 and D-hamamelose 8 were studied as acceptors in chemoenzymatic galloylations with the aim of developing an efficient preparation of hamamelitannin. The chemical preparation of hamamelitannin from furanose 2 proceeded smoothly. Base-catalyzed acylation of 2 with acetylated galloyl chloride 3 provided 81% of 1,2 ,5-trigallate 7. The Bu 2 SnO-promoted reaction yielded 84% of 2 ,5-digallate 6a regioselectively. Enzymatic reactions using vinyl gallate 4b or its acetylated analogue 4a catalyzed by Lipozyme TL IM provided mainly 5-O-galloyl derivatives. Reaction condition using acyl donor 4b in t-BuMeO afforded 82% 5-O-gallate 5b after 61 h. The pilot enzymatic benzoylation of hamamelose 8 using vinyl benzoate and Lipozyme TL IM as a biocatalyst gave mainly benzoylated furanoses (89%), of which mainly tribenzoates (52%). Similar reactions with vinyl gallate 4b and its acetylated analogue 4a gave 2 ,5-diacylated hamameloses but in yields below 20%. Acetylated vinyl gallate 4a also appeared as an acetyl donor. The Lipozyme TL IM, in its presence in the reaction mixture, also performed acetylation of acceptors 2 and 8. Hamamelose 8 in t-BuMeO with vinyl gallates, where phenolic groups were hydrophobized with methyl or benzyl moiety, readily afforded 2 ,5-diacylated hamamelofuranoses (65-84%), with the exception of the reaction with the silylated gallate 4f. The best results were obtained with tribenzylated gallate 4g, where the desired 2 ,5-diacyl 9g precipitated from the reaction mixture. Similar reactions in more polar solvent t-BuOH gave mainly monoacyls and proceeded more slowly. They did not proceed on secondary hydroxyls and were not regioselective on primary hydroxyls. Finally, after deacetylation and deisopropylidenation of compound 6a under acidic conditions, product 1 was obtained in 94% yield (79% after two steps). Similarly, after the reductive debenzylation of compound 9g, hamamelitannin 1 was obtained with a yield of 97% (82% after two steps from hamamelose). The accomplished syntheses (especially the enzymatic method) open the way to multigram preparations of bioactive hamamelitannin and its analogs.