Cholinesterase inhibitors isolated from bilberry fruit

Cholinesterases (ChEs) are key enzymes in the pathogenesis of Alzheimer’s disease (AD). A growing body of evidence suggests that plants deliver compounds able to inhibit ChEs (e.g., huperzine A, galanthamine, and physostigmine), thus playing a beneﬁcial therapeutic role in the treatment of AD. Screening for cholinesterase inhibitors (ChEIs) in selected fruits and vegetables showed that extract prepared from bilberry fruit effectively inhibited the activity of acetyl-and butyrylcholinesterase.The puriﬁcation of ChEIs from bilberry fruit followed by HPLC-UV, FT-IR, NMR, and LC–MS demonstrated that the studied compounds were derivatives of chlorogenic and benzoic acids.These results conﬁrm that bilberry fruit may serve as a useful source of ChEIs, leading to the attenuation of memory deﬁcit caused by AD.


Introduction
Cholinesterases (ChEs) are key enzymes participating in the pathogenesis of Alzheimer's disease (AD). This neurodegenerative disorder can be characterized by pathological lesions of the central nervous system (CNS), such as extracellular senile plaques (SPs) formed by amyloid-β (Aβ), as well as intracellular neurofibrillary tangles (NFTs), which are aggregates of hyperphosphorylated tau (τ) protein (Castellani, Rolston, & Smith, 2010). Oxidative stress is also reported to cause devastating changes in brain tissue (Butterfield, Reed, Newman, & Sultana, 2007). Elevated acetylcholinesterase (AChE) and butyrylcholinesterase (BChE) activities (Rao, Sridhar, & Das, 2007) contribute to acetylcholine (ACh) deficits (Schliebs & Arendt, 2011). As a result, memory loss is observed. Additionally, there is evidence that ChEs contribute to the formation of SP and NFT as well as to an increment of the toxicity of Aβ deposits (Ciro, Park, Burkhard, Yan, & Geula, 2012). Currently, cholinesterase inhibitors (ChEIs) are used for the elevation of ACh levels at cerebral cortex synapses.

Plant material
Test samples of fresh bilberry fruit (Vaccinium myrtillus L.) in harvest maturity were purchased at the local market and were immediately frozen (−20°C) until use. Fruits were harvested in forests in Lublin province (51°14′ 29

Determination of ChE inhibitory activities
Inhibition of ChE was evaluated in each fraction obtained at every stage of purification, based on the modified method of Ellman, Courtney, Andres, and Featherstone (1961 or Tris-HCl buffer instead of a studied sample were simultaneously examined as a positive and a negative control, respectively. A blank sample containing DTNB and ATCh (or BTCh) and completed to the final reaction volume with Tris-HCl buffer was used to control the spontaneous hydrolysis of the substrate. The inhibitory activity of the studied sample was calculated using the calibration curves of eserine (0.00-5.34 mmol). The inhibitory activity was expressed in eserine equivalents (µmol Es), as proposed earlier (Szwajgier & Borowiec, 2012a, 2012b. The samples were analyzed in two or eight repeats.

Acid hydrolysis
The acid hydrolysis of a combined fraction (obtained as described in Section 2.5., 50 mL) was performed. The reaction mixture consisted of 50 mL of the fraction, 200 mg of ascorbic acid, and 1.525 mL of 35-38% HCl (final HCl concentration: 1 mol). After incubation (1 h, 85°C, water bath), the mixture was cooled down and frozen (24 h, −20°C). Immediately before analysis, the thawed solution was filtered (syringe filters 0.45 µm, Merck Millipore, Darmstadt, Germany) followed by separations using analytical HPLC.

Ion exchange
Ionic compounds were removed from fraction Bi using an AG 501-X8 mixed bed resin (Bio-Rad) following the producer's instructions. One gram of the resin was added to 20 mL of fraction Bi dissolved in deionized water (0.15 mg dry mass/mL) followed by stirring (24 h, 4°C, magnetic mixer), decantation, filtration (syringe filters 0.45 µm, Merck Millipore), and drying under nitrogen.

FT-IR spectroscopy
FT-IR spectra were collected using Nicolet 6700 FT-IR Spectrometer (Thermo Scientific). The Smart iTR ATR sampling accessory was used. Fraction Bi was applied on ATR as freezedried powder. The spectra were collected over the range 4000-650 cm −1 . The sample was examined twice. Two hundred scans were averaged with a spectral resolution of 4 cm −1 . The final average spectrum was calculated and normalized for the sample. The baseline correction was obtained using Omnic Soft-ware (Thermo Scientific) and the OriginPro 8.5 program (OriginLab Corp., Northampton, MA, USA).

NMR spectroscopy
NMR analyses were performed using a Bruker Avance 500 Spectrometer (Bruker, Rheinstetten, Germany) in D2O and methanol-d4 solutions. Due to the complex composition and low concentration of the Bi fraction (about 0.4 mg dry mass/mL) and even after a long acquisition time, only the most intensive signals were observed in 1 H-NMR, 13 C-NMR, and DEPT-135 spectra. The recorded spectra were obtained using frequency 500.

LC-MS
LC-MS analysis of the Bi fraction was performed using a Shimadzu LC-MS-IT-TOF instrument with ES ionization (heat block and CDL temperature: 200°C, argon gas flow: 1500 mL/ min), connected to a Shimadzu Prominence chromatograph (Shimadzu, Kyoto, Japan) consisting of two LC-20AD XR pumps and an SPD-20A UV detector (254 and 354 nm). LC-MS grade solvents (containing 0.1%v/v formic acid) were deionized water (A) and methanol (B). The separations were performed for 15 min using isocratic elution with A:B 9:1v/v (0.15 mL/min) and Kinetex C18 100A (100 mm × 2.1 mm, 2.6 µm, Phenomenex, Torrance, CA, USA). Signals were analyzed using Lab-Solutions LC-MS software (Shimadzu). The main components of the isolated fractions were characterized after the isolation of the precursor ions (pseudomolecular ions) in an ion trap and then collided with argon to induce the fragmentation. Also, the elemental composition of the isolated compounds was studied. Formula Predictor software (Shimadzu) or the ChemBioDraw program (PerkinElmer, Waltham, MA, USA) was used for chemical structure drawing and analysis.

Results
At first, compounds present in the freeze-dried ultrafiltrate from bilberry juice (molecular weight cutoff <5 kDa) were separated using the preparative HPLC (e.g., Fig. S1). Forty-eight fractions were obtained, among which 1-20 exhibited anti-AChE activity (0.43 ± 0.04-1.16 ± 0.04 µmol Es). After 87 repeats of the separation presented in Fig. S1, fractions 1-20 were combined and concentrated to obtain an inhibitory activity equal to 1.98 ± 0.04 µmol Es. The UV-Vis absorption of this sample revealed the presence of nonprotonated dienes (245 nm) or aromatics and polyalkylaromatics (280 nm; Jiang, Huang, Weitkamp, & Hunger, 2007). The sample obtained after the combination of fractions 1-20 was very complex (Fig. S2a). For this reason, acid hydrolysis of the fraction was carried out reflecting the phenomenon occurring in the human digestive tract where glycosides undergo hydrolysis to aglycones (Stahl et al., 2002). After the hydrolysis, the separation of compounds was substantially improved ( Fig. S2b and S2c). The anti-AChE activity was detected in the fraction representing a peak marked as Bi (0.88 ± 0.02 µmol Es).
After 140 repeats of the separation presented in Fig. S2c, Bi fractions were combined and freeze dried to obtain 1 mg dry mass. It was evaluated that 1.5 kg of bilberry fruit was used in the study, and the approximate yield of fraction Bi was 0.066 mg/100 g of fresh fruit. The final Bi fraction (standardized to obtain 1 mg dry mass/mL) exhibited a high anti-AChE and anti-BChE activity (2.00 ± 0.08 and 1.50 ± 0.02 µmol Es, respectively). After the isolation, the Bi fraction was once more chromatographed in the same way (Fig. 1). The main peak (#4) as well as other minor peaks can be seen in this chromatogram.
The FT-IR spectrum of the Bi fraction ( Fig. S3) revealed the presence of single (O-H, C-H, C-C, C-O) and double (C=O, C=C) bonds. The chemical band assignment suggests that com-pounds in the Bi fraction may contain aromatic phenol rings and functional groups typical for alkanes, alcohols, or carbonyls ( Table 1).
The NMR spectra were registered; however, despite the long acquisition time, only the most intensive signals were recorded. The 13 C-NMR signals obtained using methanol-d4 were at 113.1, 115.9, 117.8, 124.0, and 146.9 ppm. The DEPT-135 spectrum obtained using D2O revealed signals at 115. 6, 117.2, 123.8, 149.4, 161.6, and 169.9 ppm. These results suggest the presence of -CH groups in the Bi fraction. Moreover, multiplets were registered in the 1 H-NMR spectrum with the use of methanol-d4 (Fig. S4), suggesting the presence of sugars as well as vinyl, aliphatic, and aromatic compounds. The 1 H-NMR spectrum obtained in D2O (Fig. S5) revealed the following multiplets, which were partially identified: 3.83 ppm (singlet), 6.33 ppm (doublet, J = 16.1 Hz), 6.87 ppm (doublet, J = 8.2 Hz), 7.43 ppm (doublet, J = 11.7 Hz), and 7.55 ppm (doublet, J = 16.1 Hz). The singlet at 3.83 ppm was assigned to the -COCH3 group. Doublets at 6.33 and 7.55 ppm, as well as at 6.87 and 7.43 ppm, were assigned to two -CH=CH-groups. In both cases, vinyl or heteroaromatic protons were identified. Due to the multiplets in the aromatic region, it was suggested that a number of structurally similar chemical compounds containing sugars or aromatic fragments were in the Bi fraction.  After LC-MS analysis of sample Bi in isocratic mode, the HPLC (Fig. S6a) and the total ion chromatograms (TIC, Fig. S6b) were recorded. It was assumed that the benzene ring caused UV absorption at 254 nm (retention times 7.1-8.1 and 10.0 min; Kumar, 2006). The UV absorption at 354 nm (retention time 10.0 min) was probably caused by compounds containing a chromen fragment (Tsimogiannis, Samiotaki, Panayotou, & Oreopoulou, 2007). Taking into consideration both chromatograms (Fig. S6) and mass spectra of the Bi fraction, an attempt was made to identify the structure of potential ChE inhibitors.
Compound A. Retention times 6.6 and 7.9 min. After the insightful analysis of pseudomolecular ions (m/z): 267 [M + H] + and 289.1261 [M + Na] + , the only possible formula C11H22O7 (accuracy 1.04 ppm) was proposed. The fragmentation spectrum of this compound was not recorded. Nevertheless, it was assumed that the observed mass and formula could match the simple glycoside presented in Fig. 2a.  Fig. S9). The proposed structure can be an isomer of methylacetamide-deoxy-O-fucopyranosyl-glucopyranoside (Fig. 2d).  (Fig. 2e). The observed fragmentation pattern matched both compounds closely (Fig. S10). Glucovanillin is also known as vanillin (or 4-hydroxy-3-methoxybenzaldehyde) glucoside. None of these compounds have been found in bilberry fruit yet. Nevertheless, it is known that bilberry fruit is a source of vanillic acid (Díaz-García, Obón, Castellar, Collado, & Alacid, 2013).
Compound F. Retention time 7.8 min. This compound was characterized by the following ions (low intensity): 397.1095 and 373.1149 Da. The nature of the concomitant inorganic positive ion was not determined. After the fragmentation, the positive ion 367 Da was obtained. The proposed formula is C 16H22O10 (accuracy: 2.52 and 2.41 ppm in the range of positive and negative ions, respectively), and the proposed compounds are geniposidic acid or 3,4,5-trihydroxy-6-(hydroxymethyl)oxan-2-yl 3,4,5-trimethoxybenzoate (Fig. 2f). Geniposidic acid is the main constituent in Eucommia ulmoides and can efficiently protect PC-12 cells against the cytotoxicity caused by Aβ25-35 (Zhou et al., 2009). However, the ChE inhibitory activity of geniposidic acid has not yet been tested. Also, the presence of this compound in bilberry or any other plant belonging to the Vaccinium genus has not yet been reported.
Compound Based on a precise assessment of the masses of ions (353.0845 and 329.0877 Da), the molecular formula C14H18O9 was proposed (accuracy: 1.13 and 0.3 ppm for positive and negative ions, respectively). Since the fragmentation of the sodium adduct ion was uncertain, the fragmentation path of anion 329.0877 Da leading to ions 167.0387 Da (100%) and 152.0173 Da (15%) was analyzed. In the case of anion 167.0387 Da, a methoxy derivative of hydroxybenzoic acid C8H7O4 (Fig. S11) can be proposed. Therefore, the tested compound may be an ester or ether of a methoxy-hydroxybenzoic acid bound to hexose (C8H7O3-O-C6H11O5). In contrast, it could also be the wellknown 4-(β-D-glucopyranosyloxy)-3-methoxybenzoic acid (Fig. 2g), previously isolated from Carthamus oxyacantha (Johansen, Wubshet, Nyberg, & Jaroszewski, 2011). It is important to note that ion 353 Da was also eluted at 7.5 min. Therefore, it was suggested that more than one isomer of this compound was present in the Bi fraction. Therefore, it was suggested that compound C16H18O9 could be an isomer of chlorogenic acid (e.g., 4-O-caffeoylquinic acid or 5-O-caffeoylquinic acid; Fig. 2h). This result was supported by the NMR spectra (Figs. S5 and S12), which showed signals derived from -CH groups of caffeic acid. The result of the fragmentation of the real HPLC standard of chlorogenic acid was similar to that observed after the fragmentation of the studied ion 355 Da. Indeed, the following fragmentation ions were obtained: 163, 145, and 135 Da (Fig. S13). However, during the presented comparison of the unknown compound with the standard of chlorogenic acid, different retention times were recorded. For this reason, we assume that the isolated compound was not chlorogenic acid.  Fig. 2j), which has already been detected in bilberry fruit (Rieger, Müller, Guttenberger, & Bucar, 2008).

Discussion
The presented results confirm that bilberry fruit is an interesting source of ChEIs, including derivatives of chlorogenic and benzoic acids, purified within the framework of this study. The presence of phenolic acids in bilberry fruit is well documented (e.g., Ochmian, Oszmiań ski, & Skupień , 2009). The exact determination of the structural formulas of some studied compounds was not possible, but functional groups were well characterized. The presence of at least one -OH group substituted in the phenol ring in meta-or para-position was considered. The presence of a -NH2 group (at position C1 in quinic acid) and/or a hexose (meta-substitution in the phenol ring) was confirmed in the chlorogenic acid derivatives. A hexose as well as a -OCH3 group (in the meta-or para-position of the phenol ring) was probably present in the derivatives of benzoic acid. Due to the different retention times of chlorogenic acid and its isomer found in the Bi fraction, it seems that the position of functional groups in a molecule is essential for ChE inhibitory activity. The inhibition of ChE by phenolic acids (including chlorogenic acid as well as 3-or 4-hydroxybenzoic acids) has already been reported (Szwajgier, 2013;Szwajgier & Borowiec, 2012a). It was demonstrated that the presence of a -OH group and/ or a -OCH3 group in the phenol ring increased the anti-ChE activity. Moreover, the methyl or ethyl esters of phenolic acids were more effective ChEIs than the corresponding free phenolic acids. Additionally, it was shown that chlorogenic acid exhibited higher anti-ChE activity (especially anti-BChE) than caffeic acid (Szwajgier, 2013). Kwon et al. (2010) demonstrated that the administration of chlorogenic acid to mice (3, 6, or 9 mg/kg, per os) inhibited, ex vivo, the AChE activity in the hippocampus and the frontal cortex as well as scopolamineinduced memory impairment (examined using a few behavioral tests). On the other hand, Orhan, Kartal, Tosun, and Sener (2007) reported on the inhibition of BChE by chlorogenic acid, but free caffeic and quinic acids were inefficient in this context. Akhtar et al. (2011) showed that anti-AChE activity could be due to the presence of a -OH group in ortho-position in the phenol ring, but the methylation was not efficient. Katalinić et al. (2010) indicated that BChE inhibitory activity increased proportionally with an increasing number of -OH groups in the phenol ring of flavonoids, wherein the position of -OH groups was important in this context. Also, the methylation of the phenol ring in para-position contributed to increased anti-AChE activity as well as exerting a general neuroprotective action (Uriarte-Pueyo & Calvo, 2010).
The preparative and analytical HPLC used in this study led to sufficient purification of the Bi fraction for structure elucidation. Previously, the LC-MS method coupled with biochemical detection was applied for the rapid online analysis of the anti-AChE activity of complex biological matrices (de Jong, Derks, Bruneel, Niessen, & Irth, 2006). Rhee et al. (2004) separated ChEI from Nerine bowdenii using two-step HPLC (preparative scale with µ-Bondapack C18 and analytical scale with Lichrospher 60 RP).
In the present study, acid hydrolysis was carried out in order to remove nonphenolic parts. However, the structure elucidation indicated that glycosidic bonds were still in the Bi fraction. It is worth mentioning that the ChE inhibitory activity of phenolic glycosides has previously been reported. Hillhouse, Ming, French, and Towers (2004) showed that the flavonoid glycosides isolated from Rhodiola rosea (gossypetin-7-O-Lrhamnopyranoside and rhodioflavonoside) were effective AChE inhibitors. Rollinger, Hornick, Langer, Stuppner, and Prast (2004) reported that scopoletin (coumarin), as well as scopolin (glucoside), could be used as potential AChE inhibitors that increase the extracellular ACh concentration in rat brains. On the other hand, the fitting of a flavonoid molecule to the active site of BChE was limited by a sugar moiety. As a result, the inhibitory activity of flavonoid glycosides was lower in comparison with that of aglycones (Katalinić et al., 2010). In another work (Uriarte-Pueyo & Calvo, 2010), it was demonstrated that the mono-acetylation of the sugar moiety in flavones from Galeopsis ladanum L. (Lamiaceae) contributed to the increase of the anti-AChE activity of these compounds.
It should be stressed that phenolic glycosides are effectively hydrolyzed in the human organism, but phenolic glucosides can also be absorbed from the small intestine without hydrolysis (Hollman, 2004). Additionally, the hydrolysis of phenolic acid esters (e.g., chlorogenic acid) is observed in the digestive tract as a result of the activity of bacterial esterases (Couteau, McCartney, Gibson, Williamson, & Faulds, 2001).
Although phenolic acids (e.g., 5-O-caffeoylquinic acid or 4-hydroxybenzoic acid) have been shown to have effects on CNS in mice after ingestion (Ohnishi et al., 2006), their permeation through the blood-brain barrier has never been proven (Diniz et al., 2007). Lardeau and Poquet (2013) suggested that 5-O-caffeoylquinic acid cannot be considered promising candidate for an entry in the brain and for a direct effect on CNS.
However, an indirect protective effect of these compounds on the brain cannot be excluded (Lee et al., 2012). Therefore, further studies will be required to investigate this issue or to develop other methods of administration of active compounds.

Conclusions
Our results confirm that bilberry fruit is an interesting source of AChE and BChE inhibitors. Selected compounds possessing the inhibitory activity were purified followed by structure elucidation, namely the derivatives of chlorogenic acid (e.g., 4-O-caffeoylquinic acid, 5-O-caffeoylquinic acid, or 3,5-O-dicaffeoylquinic acid) and benzoic acid (e.g., 3-(β-D-gluco-pyranosyloxy)-4-hydroxybenzoic acid or 4-(β-Dglucopyranosyloxy)-3-methoxybenzoic acid). Our study is the first to characterize ChE inhibitors from bilberry fruit in detail. However, further studies are required to confirm the established molecular formulas and structures of compounds. In our opinion, both preparative and analytical-scale HPLC should be improved as critical stages of the purification procedure.