Bioactive Phenolics of the Genus Artemisia (Asteraceae): HPLC-DAD-ESI-TQ-MS/MS Profile of the Siberian Species and Their Inhibitory Potential Against α-Amylase and α-Glucosidase

Artemisia genus of Asteraceae family is a source of medicinal plants known worldwide and used as ethnopharmacological remedies for the treatment of diabetes in Northern Asia (Siberia). The aim of this study was to determine the phenolic profile of 12 Siberian Artemisia species (A. anethifolia, A. commutata, A. desertorum, A. integrifolia, A. latifolia, A. leucophylla, A. macrocephala, A. messerschmidtiana, A. palustris, A. sericea, A. tanacetifolia, A. umbrosa) and to test the efficacy of plant extracts and pure compounds for antidiabetic potential. Finally, by HPLC-DAD-ESI-TQ-MS/MS technique, 112 individual phenolic compounds were detected in Artemisia extracts in a wide range of concentrations. Some species accumulated rare plant phenolics, such as coumarin-hemiterpene ethers (lacarol derivatives) from A. latifolia and A. tanacetifolia; melilotoside from A. tanacetifolia; dihydrochalcones (davidigenin analogs) from A. palustris; chrysoeriol glucosides from A. anethifolia, A. sericea, and A. umbrosa; eriodictyol glycosides from A. messerschmidtiana; and some uncommon flavones and flavonols. The predominant phenolic group from Artemisia species herb was caffeoylquinic acid (CQAs), and in all species, the major CQAs were 5-O-CQA (20.28–127.99 μg/g) and 3,5-di-O-CQA (7.35–243.61 μg/g). In a series of in vitro bioassays, all studied Artemisia extracts showed inhibitory activity against principal enzymes of carbohydrate metabolism, such as α-amylase (IC50 = 150.24–384.14 μg/mL) and α-glucosidase (IC50 = 214.42–754.12 μg/mL). Although many phenolic compounds can be inhibitors, experimental evidence suggests that the CQAs were key to the biological response of Artemisia extracts. Mono-, di- and tri-substituted CQAs were assayed and showed inhibition of α-amylase and α-glucosidase, with IC50 values of 40.57–172.47 μM and 61.08–1240.35 μM, respectively, and they were more effective than acarbose, a well-known enzyme inhibitor. The results obtained in this study reveal that Siberian Artemisia species and CQAs possess a pronounced inhibitory activity against α-amylase and α-glucosidase and could become a complement to synthetic antidiabetic drugs for controlling blood glucose level.


INTRODUCTION
Ethnopharmacology constitutes the scientific basis for the development of active therapeutics based on traditional medicines of various ethnic groups. In recent years, the preservation of local knowledge, the promotion of indigenous medical systems in primary health care and the conservation of biodiversity have become more of a concern to all scientists working at the interface of social and natural sciences. A wide range of innovations in phytochemical analysis allowed an ever-faster analysis and isolation of bioactive natural products (Kumar and Pandey, 2013;Tǎnase et al., 2018) and their identification/structure elucidation (Heinrich, 2010). Novel treatment strategies are needed for all diseases, and recently herbal medicines from such traditions have received attention in the area of prevention or treatment of chronic metabolic disorders, such as diabetes mellitus. Many plant species of various families have been noted for their antidiabetic potential. Among them were known plant extracts that inhibited carbohydrate hydrolysis enzymes, such as α-amylase (Etxeberria et al., 2012;Xiao et al., 2013b) and α-glucosidase (Xiao et al., 2013a), which play a key role in carbohydrate digestion (Yin et al., 2014). The inhibition of these enzymes is one of the possible ways to avert diabetes.
Asteraceae (Compositae) is one of the largest and widespread families of plants, with about 33,000 accepted species. The significance of the Asteraceae family for the curative aims has been described over the centuries. Due to the availability of a large variety of species in this family, it is important in ethnopharmacological medicine throughout the world. Artemisia is a large, diverse genus of plants with more than 480 species belonging to Asteraceae. Recent years have witnessed a widespread increase of interest in research of Artemisia phytocomponents with antimalarial, cytotoxic, antihepatotoxic, antibacterial and antioxidant activity. (Tan et al., 1998;Bora and Sharma, 2011;Abad et al., 2012;Ivanescu et al., 2015;Pandey and Singh, 2017). It was observed that the various Artemisia aqueous and alcoholic extracts possessed an antidiabetic effect caused by hypoglycaemic action (Dabe and Kefale, 2017). Pharmacological data provides convincing evidence that the extracts of A. absinthium (Daradka and Abas, 2014), A. afra (Afolayan and Sunmonu, 2011), A. amygdalina (Ghazanfar et al., 2014), A. dracundulus , A. judaica (Nofal et al., 2009), A. herbaalba (Awad et al., 2012), A. ludoviciana (Anaya-Eugenio et al., 2014), and A. sphaerocephala (Xing et al., 2009) were effective in streptozotocin-and alloxan-induced hyperglycemia experimental animal models due to their significant ability to reduce blood glucose level and protect against metabolic aberrations caused by diabetes. The only reported application of Artemisia drugs in humans was conducted on type II diabetic individuals and used A. absinthium capsules for 30 days (Li et al., 2015). As a result, it was observed that blood glucose level was reduced by 32% compared with the baseline value. This suggests that other Artemisia plants may have anti-diabetic properties.
As shown previously by our team, the plants growing at extreme natural habitats, such as Northern Asia or Siberia, are able to accumulate the phenolic components with antidiabetic potential (Olennikov and Kashchenko, 2014, 2017Olennikov et al., 2015aOlennikov et al., , 2017aKashchenko et al., 2017a,b). It is evident that Siberian species of Artemisia genus need to be included in our continuing search for the best antidiabetic agents of plant origin.
As a part of our ongoing research of antidiabetic plant constituents, the present work aimed to conduct chromatomass-spectrometric profiling of 12 Siberian Artemisia species, traditionally used in Siberia as antidiabetic herbs, with high performance liquid chromatography with diode array and electrospray triple quadrupole mass-spectrometric detection (HPLC-DAD-ESI-TQ-MS/MS), as well as analysis of plant extracts and pure components for their inhibitory activity against digestive enzymes (α-amylase, α-glucosidase). This multidisciplinary approach is an essential basis for the advancement of bioactive compounds of the Artemisia genus for use in tomorrow's medicines.

Total Extract Preparation
For preparation of the total extract accurately-weighed dried and grounded sample of Artemisia herb (100 g) was transferred to a conical glass flask (2 L). After that, 1.5 L of 60% ethanol solution was added with stirring and put in an ultrasonic bath. The extraction conditions were 90 min at 45 • C, ultrasound power of 100 W, the frequency 35 kHz. The extraction was repeated three times. The obtained extracts were filtered through a cellulose filter and combined. The filtrates were evaporated in vacuo at 50 • C until dryness with the use of a rotary evaporator. The total extracts were stored at 4 • C until further chemical composition analysis and bioactivity assays.

SPE Fractionation of Artemisia Extracts
The sample of milled Artemisia extract (10 g) was dissolved in 20 mL of 70% ethanol, and then added to 100 mL of distilled water and the mixture was filtered under reduced pressure. Polyamide column (15 g) was prepared: primed with 400 mL methanol followed by 800 mL tridistilled water (td-water). An aliquot (100 mL) of Artemisia extract was loaded on polyamide column. Sequential elution was done with 300 mL of td-water (dephenolized fraction) and 500 mL of 70% ethanol (flavonoid-enriched fraction). Fractions were concentrated to dryness under reduced pressure and stored at 4 • C until further chemical composition analysis and bioactivity assays.

Chemical Composition Assays
The total flavonoid content was estimated as rutin equivalents after spectrophotometric procedure after 5% AlCl 3 addition (Chirikova et al., 2010). The total caffeoylquinic acid content was determined by the colorimetric Arnow method using 3-O-caffeoylquinic acid as the standard (Olennikov et al., 2011).

Enzyme Inhibition Assays
The α-glucosidase inhibition assay was performed using spectrophotometric method . α-Glucosidase from Saccharomyces cerevisiae was dissolved in phosphate buffer (pH 6.8) containing bovine serum albumin (2 mg/mL) up to 0.5 U/mL concentration. Solution (10 µL) of sample in DMSO-phosphate buffer, pH 6.8 (1:9) at varying concentrations (10, 100, 250, 500, and 1,000 µg/mL) was premixed with 490 µL of phosphate buffer (pH 6.8) and 250 µL 5 mM p-nitrophenyl α-D-glucopyranoside. After preincubating at for 5 min, 250 µL α-glucosidase (0.5 U/mL) was added and incubated at for 15 min. The reaction was terminated by the addition of 2,000 µL Na 2 CO 3 (200 mM). Absorbance was measured at 400 nm. A solution of acarbose (20 mg/mL) was used as a positive control (PC), and water was used as a negative control (NC). The ability to inhibit α-glucosidase was calculated using the following equation: where A 400 NC is the absorbance of the negative control (water) at 400 nm, A 400 PC is the absorbance of the positive control (acarbose) at 400 nm and A 400 Sample is the absorbance of the sample solution at 400 nm. The IC 50 value is the effective concentration at which α-glucosidase activity was inhibited by 50%. Values are expressed as mean obtained from five independent experiments. α-Amylase inhibitory activity was assayed according to a previously published spectrophotometric protocol (Olennikov and Kashchenko, 2014). Sample solution in DMSO (10 µL) at varying concentrations (10, 100, 250, 500, and 1,000 µg/mL), 30 µL of phosphate buffer (pH 5.0) and 10 µL of α-amylase from Aspergillus oryzae (3 U/mL) were incubated for 20 min at . Then 10 µL of 2% starch solution, 40 µL of phosphate buffer (pH 5.0) and 100 µL of the reagent were added and incubated for 30 min at . The reagent was a solution of K 2 HPO 4 (0.8 mM), KH 2 PO 4 (0.4 mM), phenol (220 mM), 4aminoantipyrine (1.5 µM), glucose oxidase from Aspergillus oryzae (3 U/mL), and peroxidase from horseradish (0.3 U/mL) in deionized water. Absorbance was measured at 510 nm. A solution of acarbose (20 mg/mL) was used as a positive control (PC), and water was used as a negative control (NC). The ability to inhibit α-amylase was calculated using the following equation: where A 510 NC is the absorbance of the negative control (water) at 510 nm, A 510 PC is the absorbance of the positive control (acarbose) at 510 nm and A 510 Sample is the absorbance of the sample solution at 510 nm. The IC 50 value is the effective concentration at which amylase activity was inhibited by 50%. Values are expressed as mean obtained from five independent experiments.

HPLC-DAD-ESI-TQ-MS/MS Profiling and Compounds Identification Condition
Reversed-phase high-performance liquid chromatography with diode array detection and electrospray ionization mass spectrometry (RP-HPLC-DAD-ESI-TQ-MS/MS) procedure was used for the phenolic compounds profiling. Experiments were performed on an LCMS 8050 liquid chromatograph coupled with diode-array-detector and triple-quadrupole electrospray ionization detector (Shimadzu, Columbia, MD, USA), using a ProntoSIL-120-5-C18 AQ column (1 × 50 mm, Ø 1 µm; Metrohm AG; Herisau, Switzerland), column temperature was . Eluent A was water and eluent B was acetonitrile. The injection volume was 1 µL, and elution flow was 100 µL/min. Gradient program: 0.0-1.0 min 5-21% B, 1.0-2.0 min 21-38% B, 2.0-2.7 min 38-55% B, 2.7-3.5 min 55-61% B, 3.5-5.0 min 61-94% B. The DAD acquisitions were performed in the range of 200-600 nm and chromatograms were integrated at 280 nm. For ESI-MS, the parameters were set as follows: temperature levels of ESI interface, desolvation line and heat block were 300, 250, and 400 • C, respectively; the flow levels of nebulizing gas (N 2 ), heating gas (air) and collision-induced dissociation gas (Ar) were 3, 10, and 0.3 mL/min, respectively. The capillary voltage was kept at +3 kV (coumarins) in positive mode and at −4.5 kV (phenylpropanoids, dihydrochalcones and flavonoids) in negative mode. ESI-MS spectra were recorded by scanning in the range of m/z 100-1,900. The identification of compounds was done by analysis of their retention time, ultraviolet and mass-spectrometric data comparing the same parameters with the reference samples and / or literature data. The product ion spectra of the selected precursor ions (MS/MS) were used to improve the reliability of identification of Artemisia phenolic compounds.

HPLC-DAD Quantification Condition
Quantification of the phenolic compounds was performed in HPLC-DAD experiments using chromatographic conditions mentioned above. To prepare the stock solutions of reference compounds, 15 mg of 4-O-caffeoylquinic acid, kaempferol-3-O-(6 ′′ -rhamnosyl-)glucoside (nicotiflorin), isorhamnetin-3-O-glucoside, and quercetin were accurately weighed and individually dissolved in DMSO/methanol mixture (1:4) in volumetric flasks (1 mL). The external standard calibration curve was generated using 10 data points, covering the concentration ranges 1-1,000 µg/mL. The calibration curves were created by plotting the peak area vs. the concentration levels. Scopoletin-7-O-neohesperidoside (t R 0.92 min) was used as the internal standards and was dissolved separately in DMSO/methanol mixture (1:4) at concentration 1,000 µg/mL. All the analyses were carried out in triplicate and the data were expressed as mean value ± standard deviation (SD). For preparation of extract solution, an accurately weighed extract of Artemisia plant (10 mg) was placed in an Eppendorf tube, 1 mL of 60% ethanol was added, and the mixture was weighted. Then the sample was extracted in an ultrasonic bath for 10 min at 40 • C. After cooling, the tube weight was reduced to initial sign, and the resultant extract was filtered through a 0.22-µm PTFE syringe filter before injection into the HPLC system for analysis.

Method Validation
For validation of the analytical method, the guidelines established by the International Conference on the Harmonization of Technical Requirements for the Registration of Pharmaceuticals for Human Use (ICH) were employed (2005). The linearity of the method was studied by injecting five known concentrations of the standard compounds in the defined range. Results from each analysis were averaged and subjected to regression analysis. Limits of detection (LOD) and quantification (LOQ) were determined using the following equations: where S YX is a standard deviation of the response (Y intercept) and a is a slope of calibration curve. The precision of the analytical method was evaluated by intra-day, inter-day, and repeatability test. Intra-day assay was determined by assaying the mixture solution containing 15 standards (50 µg/mL) during the same day (five injections), and inter-day assay was analyzed using the same concentration for intra-day precision on four different days (interval of 1 day) in the same laboratory. The repeatability test of the sample was performed on 7-fold experiments of the mixture solution containing 15 standards (100 µg/mL). The stability test was performed with one sample solution, which was stored at room temperature and analyzed at 0, 2, 4, 8, 12, 24, and 48 h. For analysis of recovery data, the appropriate amounts of the powdered sample of 15 standards were weighted and spiked with a known amount of each reference compound and then analyzed. Each sample was analyzed in five times.

Statistical and Multivariative Analysis
Statistical analyses were performed using a one-way analysis of variance (ANOVA), and the significance of the mean difference was determined by Duncan's multiple range test. Differences at p < 0.05 were considered statistically significant. The results are presented as mean values ± SD (standard deviations) of the three replicates. Advanced Grapher 2.2 (Alentum Software Inc., Ramat-Gan, Israel) was used to perform linear regression analysis and to generate graphs. Principal component analysis (PCA) based on a data matrix (15 markers × 12 samples) was performed using Graphs 2.0 utility for Microsoft Excel (Komi NTc URO RAN, Syktyvkar, Russia) to generate an overview for groups clustering.

Phenolic Group Content and Inhibitory Activity Against α-Amylase and α-Glucosidase of 12 Artemisia Species
At the preliminary stage of the investigation, the extracts of Artemisia species were obtained and the yields of the extracts were determined ( Table 1). The average value of extraction yield of Artemisia species was 29.26 %. Further, the characterization of the phenolic group content was performed. There were significant variations in flavonoid and caffeoylquinic acids (CQAs) contents of Artemisia species. The maximum flavonoid content was observed in A. palustris extract (202.67 mg/g), in turn, minimal value was evident for A. desertorum extract (2.46 mg/g). The content of CQAs in A. commutata extract (514.65 mg/g) considerably exceeded the values of CQAs for other Artemisia species. Minimal CQAs value was detected in A. sericea extract (26.46 mg/g).
After determination of the phenolic contents, the investigation of α-amylase and α-glucosidase inhibitory activity of the extracts examined was conducted. Enzymatic analysis showed that A. commutata extract demonstrated an ability to inhibit α-amylase activity with the highest value (IC 50 = 150.24 µg/mL) and the lowest value for A. sericea extract (IC 50 = 384.14 µg/L). A similar trend was seen in the α-glucosidase inhibition experiments. A. commutata extract was the most active (IC 50 = 150.24 µg/mL) and A. sericea extract displayed considerably worse inhibitor activity of α-glucosidase (IC 50 = 754.12 µg/mL). The remaining Artemisia species showed inhibitory activity when present in concentrations higher than 207.12 µg/mL for α-amylase, and 325.63 µg/mL in the case of α-glucosidase. Acarbose inhibited α-amylase and α-glucosidase with IC 50 values of 311.24 and 1209.59 µg/mL, respectively.
To understand the links among all the studied chemical parameters and biological potential, linear regression analysis was used (Figure 2). The highest correlation was observed between total CQAs content and enzyme inhibiting activity (r 2 = 0.5315-0.6611) opposite total flavonoid content demonstrating weak relationships (r 2 = 0.0056-0.2330).
Importantly, the elimination of the phenolic compounds from Artemisia extracts after SPE procedure caused a drastic reduction of α-amylase and α-glucosidase inhibiting activity of the resultant dephenolized extracts (Supplementary Table S1).

Phenolic Profile of 12 Artemisia Species by HPLC-DAD-ESI-TQ-MS/MS
Phenolome profiling of the Artemisia species was implemented after high performance liquid chromatography (HPLC) separation of total methanolic extracts ( Figure 3) and flavonoidenriched SPE fractions (Figure 4) using diode array detection (DAD) and triple quadrupole electrospray ionization detection (TQ-ESI-MS), with both positive and negative mode.
The identification data summarized in Table 2 allowed us to detect a total of 112 phenolic compounds in 12 Artemisia extracts, including phenylpropanoid quinates and glycosides, simple phenolic acid, coumarins, dihydrochalcones and flavonoids (flavones, flavonols, flavanones) both in glycosidic and aglycone form.
Frontiers in Pharmacology | www.frontiersin.org The only simple phenolic acid O-glycoside from A. sericea had a UV-maxima at 255, 293 nm and MS-ions (m/z 315→153), and was identified as protocatechuic acid-O-glycoside (Chen et al., 2012).         A total of 7 coumarins, of simple structure and hemiterpenecoupled substances, were identified using MS detection with positive ionization mode. Component 14, discovered in five Artemisia extracts, had a pseudomolecular ion with m/z 325 [M+H] + , and an adduct ion with m/z 347 [M+Na] + as well as a deglucosidated fragment with m/z 163. Similarity to the standard UV-pattern and t R value, allowed us to identify 14 as umbelliferone-7-O-glucoside or skimmin. The values of m/z for the pseudomolecular and adduct ions and deglucosidated fragment of 18, were 30 Da more than 14, indicating a close similarity to 14 structures, with additional methoxyfunction. After comparing with the reference compound, it was summarized as scopoletin-7-O-glucoside or scopolin, and was detected in A. anethifolia, A. integrifolia, A. latifolia and A. tanacetifolia. Chromatographic behavior and spectral data of 13, from A. messerschmidtiana, was identical to esculetin.
Concentration linear ranges from 1 to 500 µg/mL were appropriate for quantification and LOQ and LOD parameters and were 0.12-0.87 and 0.36-2.64 µg/mL, respectively. The values of intra-and inter-day precision, repeatability and stability were determined for all compounds analyzed, and their means of RSD did not exceed 3% (
The obtained data on the content of 15 phenolic compounds in the extracts of 12 Artemisia species were subjected to principal component analysis (PCA) (Figure 6B). The final scores plot of PCA, as a two-component model, cumulatively consider 93.6% of total variables (PC1, 84.4%; PC2, 9.2%). In the space of the main axes of the ordination, four groups were formed, while the largest number of the samples was located at the center of the diagram. The main part of this group consisted of three pairs elements as A. anethifolia-A. integrifolia, A. tanacetifolia-A. latifolia and A. messerschmidtiana-A. leucophylla. Also, near the centrally located group, the elements of another pair were located A. umbrosa-A. desertorum. In contrast to the main group of elements, four species (A. commutata, A. palustris, A. sericea, A. macrocephala) were not associated into groups. Inhibitory Activity of Pure Caffeoylquinic Acids Against α-Amylase and α-Glucosidase Because a high correlation was obtained between the caffeoylquinic acid (CQA) content in Artemisia extracts and extract's enzyme inhibitory activity, some pure compounds were assayed for their inhibitory potential against α-amylase and α-glucosidase. Ten compounds with various degrees of substitution of the quinic acid skeleton were divided into three groups. There were four mono-substituted CQAs ( The inhibitory effect of the tested CQAs is summarized in Table 6. The trisubstituted caffeoylated analog of quinic acid, 3,4,5tri-O-CQA, as well as 4,5-di-O-CQA, exhibited stronger activity as α-amylase inhibitors with IC 50 values of 40.57 and 42.32 µM, respectively, than acarbose (IC 50 = 482.54 µM). For the α-glucosidase inhibition, the general trend of activity was similar. Two compounds, 3,4,5-tri-O-CQA and 4,5-di-O-CQA, were the most active (IC 50 = 61.08 and 62.14 µM, respectively) greatly exceeding the activity of the reference inhibitor acarbose (IC 50 = 1875.33 µM). The additional di-CQAs showed average activities with IC 50 range 184.34-983.53 µM. The potency of mono-substituted CQAs against α-glucosidase

DISCUSSION
Some studies have shown that plant species of the Artemisia genus, used in the ethnomedical systems of Asia, have antidiabetic properties (Dabe and Kefale, 2017). The people living in the territory of present-day Northern Asia or Siberia, widely used Artemisia extracts for the treatment of diabetes-like conditions. To our knowledge, this is the first report that validates the carbohydrate digestive enzyme inhibiting activity of Artemisia species, popular in Siberia, using ethnopharmacological information. Additionally, we identified the compounds responsible for the biological activity. Twelve Artemisia species selected for the investigation were characterized as concentrators and superconcentrators of flavonoids with quantitative content ranging from 2.46 (A. desertorum) to 202.67 mg/g (A. palustris). A similar conclusion was reached by quantification of caffeoylquinic acids (CQAs) found ranging from 26.46 (A. sericea) to 514.65 mg/g (A. commutata). All studied Artemisia extracts were effective inhibitors of α-amylase (IC 50 = 150.24-384.14 µg/mL; acarbose IC 50 = 311.24 µg/mL) and α-glucosidase (IC 50 = 214.42-754.12 µg/mL; acarbose IC 50 = 1209.59 µg/mL), two digestive enzymes involved in the process of increasing postprandial blood glucose level (Yin et al., 2014). Correlation analysis revealed a direct relationship between the content of CQAs in Artemisia extracts and their enzyme inhibitory activity, in contrast to flavonoids that showed weak correlation links. Removal of the phenolic compounds from Artemisia extracts resulted in a complete loss of inhibitory properties (IC 50 > 2,000 µg/mL) and indicated the leading role of these natural compounds in the manifestation of the hypoglycemic effect. Thus, we can confirm the appropriateness of the application of Siberian Artemisia species as α-amylase and α-glucosidase inhibitors, while identifying the phenolic compounds, and especially CQAs, as the main active compounds.
To establish the metabolomic profile of the phenolic compounds, a detailed chromatographic investigation of all Artemisia extracts was performed. At the preliminary stage of the investigation of HPLC profile of the total extracts of Artemisia, it was found that caffeoylquinic acids (CQA), the predominant compounds, interfered with the chromatographic separation process, disguising minor flavonoid glycosides and other compounds. To address this deficiency, methanol extracts were preliminarily subjected to solid phase extraction (SPE) on polyamide, thereby removing of the major proportion of CQA to obtain the flavonoid-enriched SPE fraction. The effectiveness of that prechromatographic treatment of plant extracts to analyze minor compounds was earlier shown (Svehliková et al., 2004;Olennikov et al., 2015a,b;Kashchenko et al., 2017b). The application of two types of extracts to analyze Artemisia species allowed us to achieve greater specificity and sensitivity of metabolomic profiling.
As far as we are concerned, the present study is the first such large-scale scientific project on the investigation of Siberian Artemisia species, resulting in the presence of over a 100 compounds revealed with the capacities of the HPLC-DAD-ESI-TQ-MS/MS method in 12 Artemisia species. Previously, the information of any phenolic compounds was known only for A. integrifolia (flavonoids; Wang J.-L. et al., 2016), A. palustris (flavonoids; Chemesova et al., 1978) and A. tanacetifolia (coumarins; Szabó et al., 1985). The data on the phenolic compounds of A. anethifolia, A. latifolia, A. macrocephala, and A. umbrosa were collected for the first time, and our research is the first to have studied the following five species: A. commutata, A. desertorum, A. leucophylla, A. messerschmidtiana and A. sericea.
Phenylpropanoid quinates were the most common group of the phenolic compounds found in all types of Artemisia, including mono-and dicaffeoylquinic acids with different types of substitutions, as well as 5-O-coumaroylquinic acid, (22) found only in A. latifolia and A. tanacetifolia. Previously it has been repeatedly shown that monotypic and mixed quinates of cinnamic, caffeic and ferulic acids are obligatory components of Artemisia species, including A. absinthium (Fiamegos et al., 2011), A. annua (Han et al., 2008), A. gmelinii (Könczöl et al., 2012), A. iwayomogi (Lee et al., 2017), A. princeps (Cui et al., 2009), A. vulgaris (Fraisse et al., 2011), and others. In contrast to phenylpropanoid quinates, there were only four species of Artemisia in which the representatives of a rare group of phenylpropanoid glycosides were detected, including glycosides of caffeic (5; A. messerschmidtiana), ferulic (8) and dehydroferulic acids (3; A. anethifolia), which, together with glycoside of protocatechuic acid (2; A. sericea), had not been described previously for the genus Artemisia. This applied equally to the isomers of melilotoside (17, 30) from A. tanacetifolia, previously mentioned only for A. splendens (Afshar et al., 2017).
Coumarins are considered ordinary common group of compounds for the genus Artemisia (Al-Hazimi and Basha, 1991). These compounds were discovered in seven studied species in the present research, including 7-O-glycosides of umbelliferone (4) and scopoletin (18). Only in two species, A. latifolia and A. tanacetifolia, coumarin-hemiterpene esters, such as lacarol (74), desoxylacarol (83) and methyllacarol (91) were identified. Both species are related to the Laciniatae subsection of the Abrotanum section, that contains A. laciniata and A. armeniaca, which 74, 83, and 91 were identified earlier (Hofer et al., 1986;Mojarrab et al., 2011). Such a limited rate of detectability of these compounds, only within the subsection Laciniatae, indicates their considerable chemosistematic significance.
Davididenin (85) and its methyl esters (98, 105, 112) were detected only in A. palustris, which is related to the small Auratae subsection of the of the Abrotanum section, together with the unexplored Asian species A. aurata. Despite the lack of scientific information, in view of their uniqueness, these compounds are promising systematic markers for the species of this subsection.
The flavonoid group of flavones is widely distributed in Siberian Artemisia species and has been identified in all studied species. Specific accumulation of C,O-glycosides of apigenin has been established for species of the section Artemisia (A. integrifolia, A. leucophylla, A. umbrosa), as was also shown earlier for the taxonomically similar species A. vulgaris (Lee et al., 1998). A similar pattern of accumulation was found for two representatives of the subgenus Dracunculus -A. commutata (subsection Commutatae, section Campestris) and A. desertorum (subsection Japonica, section Dracunculus), that can concentrate the flavonoids of this group (Kaneta et al., 1978). A. sericea was the only species capable of concentrating flavone glycosides of various types, such as the glycosides of apigenin, luteolin, 6-hydroxyluteoline and others detected. This species relates was detected for individual CQAs detected in Tussilafo farfara (Gao et al., 2008), Lonicera fulvotomentosa (Tang et al., 2008), Coffea arabica (Narita and Inouye, 2011), and Ilex kudingcha (Xu et al., 2015), but for the first time, it was revealed for compounds characteristic for Siberian Artemisia species.

CONCLUSION
The present paper demonstrated the results of the first comprehensive investigation of the Siberian Artemisia species as a source of bioactive phenolics with special attention paid to their inhibitory potential against of α-glucosidase and αamylase. Bioassay guided screening gave indisputable evidence of the pronounced activity of the extracts from the 12 ethnopharmacologically selected Artemisia species caused by the phenolic compounds. A total 112 phenolics of various groups were unambiguously, or tentatively, identified by HPLC-DAD-ESI-TQ-MS/MS. It was suggested that the flavonoid or coumarin patterns of Artemisia species are as suitable chemotaxonomic tools as other groups of plant phenolics and should be also applied. Quantification data for 15 principal phenolics helped to conclude the caffeoylquinic acids (CQAs) as major constituents, and their concentration levels were in good correlation with enzyme-inhibitory activity of plant extracts. Additional study of pure compounds showed the highest inhibitory potential among di-O-substituted CQAs, indicating that the caffeoyl-substitution at C-5 and C-4 of the quinic acid moiety appears to be a key factor in the α-glucosidase and α-amylase inhibition. This activity can also be a candidate for bioactivity targeting and further in vivo investigations are required to estimate glucose-lowering effect of Artemisia extracts.

AUTHOR CONTRIBUTIONS
The first author DO and CV mainly responsible for the operation of the whole experiment and the writing of the paper. NC, NK, VN, and S-WK mainly responsible for assisting the first author in the experiment. All authors responsible for providing technical guidance and experimental materials.