Regioselective chemoenzymatic syntheses of ferulate conjugates as chromogenic substrates for feruloyl esterases

Generally, carbohydrate-active enzymes are studied using chromogenic substrates that provide quick and easy color-based detection of enzyme-mediated hydrolysis. For feruloyl esterases, commercially available chromogenic ferulate derivatives are both costly and limited in terms of their experimental application. In this study, we describe solutions for these two issues, using a chemoenzymatic approach to synthesize different ferulate compounds. The overall synthetic routes towards commercially available 5-bromo-4-chloro-3-indolyl and 4-nitrophenyl 5-O-feruloyl-α-ʟ-arabinofuranosides were significantly shortened (from 7 or 8 to 4–6 steps), and the transesterification yields were enhanced (from 46 to 73% and from 47 to 86%, respectively). This was achieved using enzymatic (immobilized Lipozyme® TL IM from Thermomyces lanuginosus) transesterification of unprotected vinyl ferulate to the primary hydroxy group of α‐ʟ‐arabinofuranosides. Moreover, a novel feruloylated 4-nitrocatechol-1-yl-substituted butanetriol analog, containing a cleavable hydroxylated linker, was also synthesized in 32% overall yield in 3 steps (convergent synthesis). The latter route combined the regioselective functionalization of 4-nitrocatechol and enzymatic transferuloylation. The use of this strategy to characterize type A feruloyl esterase from Aspergillus niger reveals the advantages of this substrate for the characterizations of feruloyl esterases.


Introduction
The development of white biotechnology is underpinned by advances in enzyme discovery and engineering, areas that are being driven by metagenomics and in vitro directed enzyme evolution.These techniques procure massive discovery or creation of new enzyme-encoding sequences, filling up databases with a wealth of information.
However, while resolving an early step in the discovery pipeline, these techniques progressively create a new bottleneck regarding enzyme characterization.Therefore, there is a pressing need to extend the enzymologist's toolbox, providing informationally rich high-throughput screens that can not only attribute an activity to putative enzymes, but also procure some qualitative details on enzyme properties.In this respect, the availability of easy to use chromogenic substrates that can provide both qualitative and quantitative assays and be compatible with automatized protocols is a crucial issue.Feruloyl esterases (Faes; EC 3.1.1.73and family CE1 of the CAZy classification [1]) are of interest, both because of their role in the deconstruction of complex plant-based materials and also as synthetic tools for the preparation of bioactive compounds with potential antioxidant properties [2][3][4][5].Operating via a two-step serine protease mechanism involving a conserved Ser-His-Asp/Glu catalytic triad [6,7], Faes catalyse the hydrolysis of ester bonds linking hydroxycinnamoyl groups to the glycosyl moieties of plant-based polysaccharides, such arabinoxylans and arabinans.In this respect, Faes are important components of plant cell wall-degrading enzymatic arsenals, since the hydrolysis of trans-ferulate-polysaccharide linkages contributes to breakdown of intermolecular bonds that structure the lignocellulosic matrix.Moreover, Faes are useful tools to obtain commercially relevant ferulic acid, which represents up to 3% (w/w) of plant cell wall dry weight [8].
So far, the detection and characterization of Faes has mainly relied either on the use of HPLC or UV-visible spectrophotometry, using natural or synthetic compounds [9,10].
In this work, we revisit the preparation of simple feruloylated substrates, such as 5bromo-4-chloro-3-indolyl and 4-nitrophenyl O-5-feruloylated α-L-arabinofuranosides 1a and 1b.Although these substrates are commercially available, their synthesis involves 7-8 steps [15][16][17].This engenders rather high costs (i.e., as of July 29 th , 2020, €2500 and €778 per 100 mg for 1a and 1b respectively) that are approximately 19-and 14-fold higher than the non-feruloylated precursors.Therefore, our aim was to simplify synthesis in order to reduce cost.Furthermore, we describe the short synthesis of new feruloylated chromogenic substrate 12, a molecule that obviates the need for a glycosyl moiety while containing a cleavable hydroxylated linker that mimics natural geometry and physico-chemical properties of osidic linkages.

Chemoenzymatic synthesis of 5-O-feruloylated α-L-arabinofuranosides
The synthesis of chromogenic 5-O-feruloylated α-L-arabinofuranosides 1a and 1b is usually achieved using a multi-step pathway that involves trapping the furanose conformation, anomeric activation, glycosidation, regioselective deprotection of the primary hydroxyl group, feruloylation and final deprotection to yield the target molecule [12,[15][16][17].Additionally, the temporary protection of functional groups is sometimes used during synthesis in order to facilitate certain steps.Using an alternative approach, we employed one-step regioselective transesterification of the unprotected vinyl ferulate 2 (synthesized in 56% in-house yield and up to 77% previously reported yield [24,25] in one-step and at gram scale) using immobilized Lipolase 100T from T. lanuginosus [26,27] and readily available and reasonably cheap 5-bromo-4-chloro-3indolyl or 4-nitrophenylα-L-arabinofuranosides.This afforded the corresponding feruloylated derivatives, 1a and 1b (Figures 1A-B).Yields (73 and 86% for the indolyl and 4-nitrophenyl derivatives respectively) characterizing regioselective enzymatic feruloylation of the primary hydroxyl group compare favourably with previously reported overall yields (46 and 47% respectively, in three steps) [15][16][17]28], which relate to enzymatic selective O-5-deacetylation of the primary hydroxyl group, its esterification and final deprotection of the 2,3-O-acetyl groups of the glycoside and Oacetyl group of the ferulate moiety.That fact that lipase-catalysed transesterification obviates the need for require protection/deprotection is a considerable advantage, because the final deprotection in the chemical pathway is complicated by the presence of another ester linkage within the molecules [12][13][14][15].In principle, the method described herein is generic and thus applicable to other chromogenic α-Larabinofuranosidic compounds, such as 4-nitrocatechol (4NTC), 2-chloro-4-nitrophenyl and umbelliferyl derivatives.To demonstrate the suitability of 1a for qualitative in situ screening of microbial colonies growing on solid agar medium, the compound was incorporated (300 µg/mL) into top agar along with an α-L-arabinofuranosidase from Thermobacillus xylanilyticus (TxAbf).This provided the means to reveal coloured microbial colonies expressing the type A Fae(+) from Aspergillus niger (AnFaeA, Figure 1C).Coloration is the result of successive reactions: (i) release of the free 5-bromo-4-chloro-indoxyl-3-ol by an enzyme cascade, wherein TxAbf-catalysed cleavage of the glycosidic bond was made possible by the prior release by AnFaeA of the ferulate moiety [21]; and (ii) its spontaneous oxidation and subsequent dimerization forming the blue 5-bromo-4-chloroindigo-like insoluble dye [15].In an alternative demonstration, 1a was also used for colony level detection of FAE using yeast cells (Yarrowia lipolytica [29]) that actually coexpressed AnFaeA and TxAbf(+) (data not shown).

Investigating enzymatic transferuloylation reaction of non-glycosidic motifs
While lipase-catalysed transesterification reactions onto glycosidic structures have been extensively described, data related to the regioselectivity of transesterification on hydroxylated alkyl and/or aryl moieties are more sparse [30,31].The extent to which Lipolase 100T catalyses feruloyl transfer reactions involving substituted benzylic alcohols was thus investigated to establish its usefulness for the preparation of various polyhydroxylated compounds of interest (e.g., antioxidants, chromogenic molecules for screening, etc.) [4,27].Accordingly, we observed that transesterification only occurred when using 'primary' benzylic alcohols; no phenol acylation was detected in the case of hydroxynitrobenzylic alcohol (Table 1, no side-product of 4 and 9 with transfer on aromatic 'secondary' alcohol; i.e., phenol), 2-chloro-4-nitrophenol or 4NTC (data not shown).Additionally, the exact position of the benzyl alcohol affected transfer, with ortho-substitutions (R 1 ) displaying a hindered electron withdrawing group.This led to low target product yields (5, 6, 8 and 9), lower reactivity and/or greater hydrolysis of the vinyl ferulate into ferulic acid.

Synthesis of L-arabinofuranoside free 4-nitrocatechol-1-yl-linkerferulate chromogenic substrate (12) and its evaluation as chromogenic substrate for Fae assays
To synthesize the chromogenic feruloylated-butanetriol 4-nitrocatechol-1-yl  As expected, investigation of the stability of the chromogenic substrate 12 using UVvisible spectrophotometry revealed that, unlike 4-nitrocatechol-1-yl ferulate (4NTC-Fe) that undergoes spontaneous hydrolysis even at neutral pH and 40 °C [13], the presence of the alkyl-like linker procures higher stability over a wide pH range (up to pH 9.0), irrespective of temperature.This is because in compound 12 the ferulate moiety is not directly linked to the good leaving group 4NTC (pK a = 6.61 [34]).Instead it is bonded to the linker whose pK a can be assimilated either with that of glycerol (pK a = 13.61) or L-arabinose with (pK a =11.31) [35], meaning that it is a poor leaving group.Moreover, our observations regarding linker stability are consistent with the known stability of ester linkages under basic conditions.
The usefulness of 4NTC-linker-Fe 12 for the characterization of Faes was investigated (Figure 2), measuring 4NTC release by AnFaeA [36] at 40 °C.The enzyme-catalysed reaction leads to cleavage of the ester bond linking the ferulate to the linker-4NTC moiety and thus accumulation of linker-NTC.Therefore, working in discontinuous mode, 4NTC is quantified by submitting samples removed from the reaction to the oxidative action of sodium periodate at 0°C and reading absorbance at 530 nm (in alkaline conditions) [23,32,37,38].Importantly, it is vital to include a stoichiometric amount of ethylene glycol to avoid further oxidation of free NTC by sodium periodate (Figure 2B).The specific activity (SA) of AnFaeA on 4NTC-linker-Fe 12 was determined to be 3 IU/mg of protein, a value comparable to that measured on destarched wheat bran (3 IU/mg) [39] containing 5-O-feruloylated α-L-arabinofuranosyl moieties, but lower than that (40 IU/mg, unpublished data) measured using more labile 4NTC-Fe.Therefore, although 4NTC-Fe is a practical synthetic probe for both high-throughput screening and preliminary characterisation of Fae activity [13], 4NTC-linker-Fe 12 is almost certainly a better analogue of ferulate linkages found in plant-based structures.

Conclusion
The use of immobilized Lipolase 100T provides the means to perform regioselective transesterification of vinyl ferulate 2 to the primary hydroxyl group of benzylic alcohols and polyhydroxylated compounds.Three compounds suitable for the detection and/or characterization of Fae activity were synthesized in a straightforward protocol that holds the potential to greatly reduce the cost of substrates 1a and 1b.Moreover, the enzymedriven convergent synthesis of 12 affords a novel substrate that is highly suitable for the characterization of feruloyl esterases.

General procedure for enzymatic transesterification
The enzymatic transesterification steps were performed according to a published protocol [26].Briefly, Lipolase 100T (1 g) was added to a solution of alcohol (0.30 mmol, 1 equiv.)and vinyl ferulate 2 (100 mg, 0.45 mmol, 1.5 equiv.) in acetonitrile ( 6mL).The reaction mixture was stirred overnight at 37 °C, then filtered, filter cake washed with acetone and the filtrate was evaporated to dryness.The residue was recovered in ethyl acetate, washed with saturated aqueous sodium hydrogencarbonate (three times).Combined organic phases were dried over anhydrous sodium sulfate, filtered and concentrated under reduced pressure.Flash chromatography (gradient of ethyl acetate in petroleum ether from 0 to 50%) afforded pure ferulates 1a, 1b, 3-9 and 12.

Araf-Fe (1b)
Y. lipolytica AnFaeA(+) strain was used to inoculate solid YNB medium (1.7 g/L YNB without casamino acid, 5 g/L ammonium chloride, 20 mL/L oleic acid, 10 g/L Dglucose, 2 g/L casamino acid, and 15 g/L bacto agar in 100 mM citrate-phosphate buffer pH 5. Petri dishes were incubated for 48 h at 30 °C and then overlayed with a preparation of molten 1% (w/v) top agar containing chromogenic substrate 1a (300 µg/mL and 0.5% DMSO) and TxAbf (2 IU/mL).Once the top agar was solid, incubation at 37 °C for 1 h allows color to develop.Y. lipolytica strains that contain no AnFaeA gene were also checked to remain colorless after addition of chromogenic substrate and auxiliary enzyme.

Liquid medium-based colorimetric assays using 4NTC-linker-Fe (12)
In a typical experiment, discontinuous enzyme assays were performed in triplicate in buffered conditions (100 mM sodium phosphate pH 6.0) in the presence of 1.8 mM 12

Figure 2 :
Figure 2: (A) Principle of the 4NTC release after action of Fae on 12 in the presence of

and 3 .
6% DMSO, final concentrations.For the assay, this solution was preincubated at 40 °C before AnFaeA addition.Aliquots (25 µL) were stopped by cooling (at 0 °C) every 6 min over a 24 min-period and mixed with 45 µL of cooled 10 mM NaIO 4 solution (pH 2.0).After keeping 5 min at 0 °C throughout, 45 µL of ethylene glycol were added, followed by 135 µL of 2 M Na 2 CO 3 after 5 min.The optical densities (OD) at 530 nm were recorded on a microplate reader Infinite M200 PRO (TECAN).One unit (IU) of Fae specific activity (SA, expressed in µmol/min/mg or IU/mg) corresponds to the amount of released 4NTC (in µmol) per minute per milligram of protein.Negative controls containing all of the reactants except the enzyme were always included in order to monitor and correct for spontaneous hydrolysis of the substrate.Control reactions containing 12, 11 and 4NTC without enzyme and 4NTC without both enzyme and ethylene glycol were also prepared.
a Yields of isolated ferulates after purification step; b Ratio (in %), determined by 1 H NMR, of the different species of feruloyl derivatives within the crude reaction mixture: remaining vinyl ferulate, ferulic acid (hydrolysis product) and ferulates 3-9; c The expected ferulates were confirmed by mass spectrometry analysis (HRMS) but the low purity of samples after purification prevented fine structural characterization by NMR.