Determination of phenolic compounds in water-ethanol extracts of Populus tremula L. leaves using high-performance liquid chromatography

: The analytical task of determining the phenolic compound content of water-ethanol extracts of Populus tremula L. (common aspen) leaves is complicated by the heterogeneity of compound groups having different polarities and appearing in varying concentrations. The purpose of the present work is to study the conditions of solid-phase extraction and high-performance liquid chromatography used to analyse the content of different groups of phenolic compounds in water-ethanol extracts of leaves from the P. tremula plant. In order to facilitate the derivation of phenolic compounds, an exhaustive extraction process was carried out using ethanol. Solid-phase extraction was carried out using a Diapak C16 cartridge, after which the eluates were passed through a membrane filter having a pore diameter of 0.45 μm. The high-performance liquid chromatography method was used to determine the content of phenolic acids and flavonoid glycosides, as well as salicin and individual flavonoid glycoside components: hyperoside, rutin, astragalin and two unidentified flavonoid glycosides in aqueous (analyte 1) and aqueous-alcoholic fractions (analyte 2) in two systems along the gradient elution. The requirement of analysing the primary aqueous eluate together or in parallel with the main aqueous-alcoholic fraction in the preparation of P. tremula leaf extracts for high-performance liquid chromatography using solid-phase extraction cartridges was substantiated. For separating the extract to determine the hydroxycinnamic and hydroxybenzoic acid content, it is preferable to use system 2; for determining the phenologlycoside (salicin) content, system 1 is more effective. Flavonoid glycosides (hyper-oside, rutin, astragalin and two unidentified flavonoids) make the most significant contribution to the differ-ence between the aqueous and aqueous-alcoholic fractions.


INTRODUCTION
The study of phenolic compounds (PCs) in plant material is of current interest in order to identify new economically-significant sources of biologicallyactive substances, primarily for medicinal use. PCs are also widely used as chemotaxonomic markers, as well as, more recently, within the framework of ecological research in the selection of bioindication objects and methods for determining environmental wellbeing. Moreover, for taxonomic and bioindication purposes, it is necessary to obtain the most complete information on the content of individual phenolic substances, which presupposes exhaustive extraction and careful sample preparation. However, the complex matrix composition inherent in crude plant extracts complicates the analytical problem, negatively impacting on analysis results [1]. In terms of significantly simplifying the analysis procedure and improving its metrological characteristics, the most efficient and versatile method for isolation, purification and concentration of phenolic substances from plant samples having a complex composition is solid-phase extraction (SPE) [2,3].
An analysis of publications indexed in the Web of Science citation database (© Clarivate) for 2013-2018 showed that, when performing tasks requiring maximum preservation of the extract for the subsequent study of PS, membrane filters are most typically used during preliminary sample preparation, often in conjunction with centrifugation. Mention of the use of solid-phase extraction (SPE), including the use of cartridges such as Phenomenex Strata-X and C 18 (Torrance, CA, USA), Agilent SampliQ (Agilent Technologies, CA, USA) separately or together with other methods of purification and / or fractionation, was found in about a third of the reviewed publications [4][5][6][7][8][9][10][11][12][13]. In the Russian Federation, domestically-produced Diapak cartridges are successfully used (CJSC BioKhimMak ST, Moscow) for SPE purposes. For polyphenolic compounds, a sorbent with a grafted C18 phase is optimally used for sample preparation within the framework of HPLC analysis [1].
According to the literature data, various groups of phenolic compounds are present in the leaves of Populus tremula L. (common aspen): flavonoids (hyperoside, rutin, quercitrin, isoquercitrin, astragalin), nine phenolic glycosides, including salicin and tremulacin, as well as chlorogenic acid and esters of pcoumaric, ferulic and cinnamic acids [14][15][16][17]. Phenol glycosides constitute a significant proportion of PCs in aspen leaves, significantly exceeding the flavonoid and phenol carboxylic acid content [15]. The separation completeness of compounds and quantitative determination accuracy can be influenced by the chemical nature of individual components in the intact sample, as well as their number and content [5,18]. A matrix with the joint presence of a sufficiently large In our study examining the content of different groups of phenolic compounds in water-ethanol extracts of leaves of P. tremula, we set the goal of studying the conditions of SPE and HPLC analysis.

MATERIALS AND METHODS
We examined mature undamaged leaf blades from aspen trees of between 10 and 15 years old, collected from the 1st to the 5th August, 2015, on the territory of the Kedrovsky coal mine in southwestern Siberia. According to the results of differential spectrophotometry with AlCl 3 , the different total flavonoid glycoside content in terms of rutin following a single cold extraction with 95% ethanol was as follows: In order to extract phenolic compounds, an exact weighed portion (0.120-0.200 g) of the crushed air-dry material was first obtained by cold extraction with 70% ethanol in darkness for 48 h. Next, an exhaustive extraction was carried out three times with 50% ethanol while heating in a water bath: 1) 30 ml of the extractantfor 30 minutes; 2) 20 ml of extractantfor 20 minutes; 3) 10 ml of extractantfor 10 minutes. The combined extract was evaporated to dryness and brought to a volume of 3 ml with 50% ethanol.
For SPE, 1 ml of the extract was diluted with bidistilled water to 5 ml and passed through a Diapak C16 concentrating cartridge (CJSC BioKhim-Mak ST). The substances were washed off the cartridge with a small amount (5 ml) of solvent with an increasing concentration of ethanol (40, 70 and 96%).
During the preparation of P. tremula samples, it was observed that the aqueous residue had a yellowish colour when the sample was washed through a TFE cartridge, indicating a significant content of coloured compounds, possibly of a phenolic nature. For this reason, the water residue was not discarded following sorption on the cartridge, but analysed as the first fraction (analyte 1 -A1) separately from the water-alcohol fraction (analyte 2 -A2). Following SPE, the eluates were passed through a membrane filter having a pore diameter of 0.45 μm.
The components were analysed on an Agilent 1200 liquid chromatograph fitted with a diode array detector and a ChemStation system for collecting and processing chromatographic data. The substances were separated on a Zorbax SB-C18 column having dimensions 4.6×150 mm and a particle diameter of 5 μm using a gradient elution mode. Two systems were used for the chromatographic procedure. System 1 (S1, developed for the separation of phenolic substances, primarily flavonoid glycosides): in the mobile phase, the methanol content in an aqueous solution of orthophosphoric acid (0.1%) varied: from 32 to 33% in 27 minutes; then up to 46%by 38 minutes; then up to 56%by 50 minutes; and up to 100%by 54 minutes. System 2 (S2, developed for the separation of phenolic substances, primarily phenolic acids): in the mobile phase, the methanol content in an aqueous solution of orthophosphoric acid (0.1%) varied: from 19 to 70%in 30 minutes; then to 100%by 32 min. The flow rate of the eluent is 1 ml/min. Column temperature -26 o C. Volume of injected sample -10 μl. Detection was carried out at analytical wavelengths λ = 255, 270, 290, 340, 350, 360, 370 nm. To prepare the mobile phases, we used methyl alcohol (extra pure grade), orthophosphoric acid (extra pure grade) and bidistilled water. Standard solutions were prepared at a concentration of 10 μg/ml in ethyl alcohol. Standard samples of salicin (MP Biomedicals LLC), rutin, hyperoside and astragalin (FLUKA Analytical) were used as taps. Each variant was analysed in 4 replications.
The content of individual components (C x , %) in terms of absolutely dry matter (ADM) was calculated by the formula: where С stthe concentration of a standard solution of a phenolic compound (PC), μg/ml; S 1the area of the PC peak in the analysed sample, a.u.; S 2peak area of the standard PC, a.u.; V 1volume of the eluate after washing out the PC from the concentrating cartridge, ml; V 2total extract volume, ml; Msample weight, mg; Braw material moisture (%). The salicin content was calculated in terms of salicin; phenolic acidsin terms of gallic and chlorogenic acids. The flavonoid glycoside content was calculated for quercetin using the coefficient known from the literature for converting the concentration -2.504 [19,20].
Statistical data processingcalculation of the average value of the feature (M), its error (m M ), variance (ANOVA, Duncan's test) and the method of principal components were carried out using the Statistica 10 application package.

RESULTS AND DISCUSSION
Hyperoside, rutin, astragalin, as well as gallic-, chlorogenic-, and cinnamic acids, were previously found in the leaves of P. tremula from ecotopes having varying degrees of technogenic load [21].
In this work, the composition and content of phenolic acids (PA) and flavonoids glycosides (FG) in both aqueous (A1) and aqueous-alcoholic fractions (A2) were determined in two systems differing in the elution gradient. Data on the content of differrent groups of PC are presented in Table 1.  Note. N is the number of phenolic components; R is the degree of analyte recovery.
Earlier, possible losses of gallic and caffeic acids in the process of sample preparation of plant extracts using SPE were reported [1,22]. In all investigated P. tremula samples, the aqueous eluate (A1) contains a significant amount of PAin most cases, exceeding that of the aqueous-alcoholic analytes (see Table 1). In addition, the presence of significant amounts of phenolicincluding flavonoidcomponents was observed in the aqueous fractions of all samples (Fig. 1). Thus, in the first two samples, the FG content in the aqueous analyte is comparable toor only 1.3-1.5 times less thanthe FG content in the target aqueous-alcoholic analyte. In O3, however, the FG content in the aqueous analyte exceeds that observed in the aqueous-alcoholic analyte by 1.4-1.5 times. The lowest number of compounds for all studied variants of determination was noted for O3, in which the highest content of both PC groups is observed.
It is known that substances having different polarities are concentrated and desorbed from the sorbent of the cartridge to varying degrees of extraction. The recovery factor was analysed in the sys-tem having the highest total PC content and number of components, i.e. S2 (see Table 1). In an earlier work, it was indicated that the degree of extraction of rutin from an aqueous decoction of St. John's wort was 49% [1]. In our samples, the degree of extraction from the combined fractions (aqueous and aqueous-alcoholic) was for different samples: for FG -64-105%; for PA -65-91%. The different recovery rates are possibly related to intermolecular effects in the complex natural matrix of the aspen leaf extract, as well as to the different number and content of the determined components in the samples.
To elucidate the relationship between the PC composition and content of the aspen leaf extract with the sample preparation and HPLC analysis conditions, we determined the content of some individual substances (salicin, hyperoside, rutin and astragalin), as well as two unidentified substances (FG1 and FG2), classified as glycosides of flavonoids, according to spectral data (λ max = 255, 355 nm) ( Table 2). The highest salicin content (best recovery and separation) was noted in aqueous eluates, especially during chromatography in S1. Two glycosides FG1 and FG2, which were close in retention times, were not separated in the aqueous-alcoholic eluate of either system. For O3, a different ratio of the main flavonoid componentsrutin (prevailing) and hyperosidewas established than in other studied samples, in which the content of hyperoside exceeded the amount of rutin by 2 or more times (see Table 1).
To identify the most significant influencing factors from the existing number of explanatory variables (the content of analysed individual substances and their sums), the principal component analysis (PCA) method was applied.
We took 9 signs: content salicin, hyperoside, rutin, astragalin, flavonoid glycosides FG1 and FG2, the sum of five glycosides, the sum of FG and the sum of FC. The data visualisation is graphically presented in the scatterplot (Fig. 2).
For the PC of aspen leaves, sufficiently clear scattering regions can be observed on a threedimensional diagram constructed using the first three factors (F1-F2-F3). The compact area of dispersion of indicators of the PC content in the О2 aspen leaves is separated from others by the first factor. The most significant correlation coefficients noted for the first factor are the sum of FG, the sum of five glycosides and the sum of PC, as well as the salicin and rutin content. Different samples have  figure). In addition, according to F2-F3, there are two scattering regions, represented by (1) aqueous and (2) aqueousalcoholic fractions, although some points are located nearby. The second factor correlates most strongly with the astragalin and FG2 content, while the third correlates most strongly with the hyperoside, FG1 and rutin content. According to F1-F3, the region with aqueous-alcoholic eluates is also localised, except for O1 in the aqueous-alcoholic fraction S1.
Thus, the differences between the aqueous and aqueous-alcoholic fractions, as well as the differences of the samples from each other, is mainly due to the content of individual flavonoid components: hyperoside, astragalin, rutin, FG1 and FG2. Differences in the PC content of aspen leaves largely depend on the analysed fraction (aqueous or aqueous-alcoholic) (see Fig. 2). The salicin and PC content correlates with the sample preparation conditions and elution gradient in HPLC analysis. Salicin is best detected in the S1 aqueous eluate (see Table 2). S2 is preferable for use at a sufficiently high content of PA in the test material, in particular, for the analysis of the aqueous fraction of P. tremula leaf extracts (see Table 1).

CONCLUSIONS
1. In order to study the PC composition and content by means of HPLC when preparing Populus tremula extracts using SPE cartridges, it is necessary to analyse the aqueous eluate together or in parallel with the main aqueous-alcoholic fraction.
2. In order to separate the extract for determining the hydroxycinnamic and hydroxybenzoic acid content, it is preferable to use system 2; for determining the phenologlycoside (salicin) content, system 1 is more effective.
3. Preferred conditions for sample preparation and HPLC analysis for optimal separation of the component composition of flavonoid glycosides have not been identified. The content of individual flavonoids (hyperoside, astragalin, rutin, and two unidentified flavonoids) makes the most significant contribution to the differences between aqueous and aqueous-alcoholic fractions.