Chemical Composition and Biological Activity of Extracts from the Aerial Parts of Epilobium parviflorum Schreb.

Abstract

Epilobium parviflorum Schreb. is used in folk and modern medicine for the treatment of prostate diseases. It is also known to alleviate gastrointestinal ailments. The aim of the present study is to define the chemical composition of diverse extracts from the herb, to test their inhibitory properties toward post-proline-specific peptidases and to elucidate the mechanisms of their antitumor activity on colorectal carcinoma cells in vitro. The extractions were performed using mono- or biphasic systems of solvents. Their chemical compositions were defined by LC-HRMS. Inhibitory properties towards prolyloligopeptidase (POP) and fibroblast activation protein (FAP) were studied by kinetic assays on human recombinant enzymes. Antioxidant activity was measured by three methods. Genotoxicity to HT-29 colorectal carcinoma cells was analyzed with the comet assay. FACS analyses and flow cytometry were used to evaluate the extracts effect on the cell cycle and their pro-apoptotic properties on HT-29 cells. The extract derived using 80% ethanol was chosen for the next studies due to its efficient and selective inhibition of POP. It contains mainly oenotein B and myricetin-3-O-rhamnoside. Its antioxidant and moderate genotoxic activities can contribute to the antitumor effect on HT-29 cells. The extract has a small effect on the cell cycle but a pronounced pro-apoptotic action on those cells. In conclusion, the 80% ethanol extract of E. parviflorum concentrates the ellagitannin oenotein B, which is a selective inhibitor of POP. Antitumor activity of the extract towards HT-29 cells may be due to the inhibition of POP, the antioxidant, genotoxic and pro-apoptotic activities.

1. Introduction

Epilobium parviflorum Schreb. (smallflower hairy willowherb) is a perennial flowering plant of Onagraceae family widely distributed all over Europe, West Asia, North Africa and is also cultivated in North America [1]. It is usually found in humid areas around rivers and mountain lakes on semi-shaded places at an altitude up to 2500 m. In Bulgaria, it is spread throughout the country, mainly in humid mountain meadows. E. parviflorum is well-known as a medicinal plant normally used in both traditional and modern medicine. Traditionally, the herb is applied in the form of infusions and decoctions to relieve discomfort in stomach and kidney diseases. In other countries, it has a number of various applications, recently summarized by Granica et al. [2]. It is found to be most suitable for healing prostate, kidney and bladder disorders. Additionally, teas are applied for alleviating the symptoms of stomach ulcers, gastritis and other gastrointestinal ailments [2].

Previous studies have shown that E. parviflorum extracts exhibit significant antioxidant and anti-inflammatory properties [3,4]. The high content of polyphenols, acting as free radical scavengers and modulating oxidative stress pathways, is largely responsible for these effects. The extracts from the E. parviflorum have also been reported to influence prostaglandin biosynthesis and reduce edema in animal models, supporting their traditional use in inflammatory conditions [5]. Most of the studies on the in vitro effects of E. parviflorum extracts are now devoted to the elucidation of their mechanisms of action in prostate benign hyperplasia and prostatic cancers. Furthermore, antitumor activity of the methanol extract from the roots of the herb on the human luminal subtype of breast cancer (MCF-7 cell line) has also been documented [6]. A similar methanol extract has been shown to possess a high cytotoxicity against human malignant melanoma cells of A375 and COLO-679 cell lines [7]. Although the herb has long been used to treat diseases of the digestive system, little research has been devoted to its mechanism of action in such illnesses. Thus, Akbudak et al. compared the cytotoxic effects of water and ethanol extracts from the aerial parts of E. parviflorum on the human colorectal cancer cell line HT-29 [8]. The authors determined that the aqueous extract is more effective (IC₅₀ = 43.6 μg/mL) than the ethanol extract (IC₅₀ = 191.0 μg/mL). Recently, we obtained the 80% aqueous ethanol extract from the aerial parts of the same herb of Bulgarian origin in which IC₅₀ on HT-29 cells was 31.2 μg/mL and the selectivity index relative to MCF-10A cells (non-tumorigenic human breast epithelial cells) was 2.6 [9]. Those differences are most probably due to the dissimilar chemical compositions of the extracts obtained by varied solvents (pure ethanol or water and aqueous ethanol). Compositional analyses of several Epilobium species revealed abundant polyphenols and high antioxidant capacity, emphasizing the importance of solvent and extraction procedure selection [10,11]. Moreover, the secondary metabolites differ largely depending on the geographic region and the climate [12]. In a pilot study, using the LC-HPMS analyses of extracts from E. parviflorum aerial parts, we found components not described so far in this plant species. We supposed that those compounds are characteristic only for the herb of Bulgarian origin.

Our preliminary studies also showed that some extracts of E. parviflorum aerial parts inhibit the human recombinant prolyloligopeptidase activity in vitro (unpublished data). Prolyloligopeptidase (POP, EC 3.4.21.26) belongs to the S9 family of post-proline-specific peptidases (PPSP) together with dipeptidyl peptidase IV (DPPIV, EC 3.4.14.5) and fibroblast activation protein α (FAP, EC 3.4.21.B28). It is a cytosolic enzyme, recently associated with the pathogenesis of cancer. Increased levels of POP have been found in a variety of solid tumors. In colorectal cancer, elevated serum levels of the enzyme have been suggested as a biomarker for distant metastases and low survival rates [13]. It is found that the administration of selective POP inhibitors may result in inhibition of angiogenesis and cell proliferation and restriction of tumor growth [14]. However, the design of such inhibitors is a challenge because of the similarity of the substrates of PPSPs and their uniform catalytic mechanism.

The current research was motivated by the observation that Epilobium parviflorum cultivated in Bulgaria exhibits a distinct phytochemical composition, as well as by the limited information available on its effects in gastrointestinal disorders, including colorectal cancer.

The aim of the present study is to characterize the chemical composition of extracts from the aerial parts of E. parviflorum from Bulgarian sources, to evaluate their inhibitory activity toward post-proline-specific peptidases, and investigate the mechanisms responsible for the detected antitumor effects on colorectal carcinoma cells in vitro. This work may provide a deeper understanding of the mechanisms underlying the antitumor activity of E. parviflorum and contribute to its potential therapeutic application in gastrointestinal diseases.

2. Materials and Methods

2.1. Materials

Dried and crushed stems, twigs, leaves and flowers of E. parviflorum were grown and supplied by Dicrassin Ltd. (Bulgaria) [15] and HBH, Alin^® brand [16], which is licensed to distribute the herb by strictly following the International legislation and also, the Bulgarian Law of Biological Diversity. The plant is grown in Yunatsite village, Pazardzhik municipality (Dicrassin Ltd.) and in Alino village, located in the Samokov municipality, at the foot of Plana Mountain (HBH, Alin^® brand). The distribution method for these companies is in the form of dry crushed material, packaged and provided to herbal pharmacies.

Acetonitrile, formic acid and methanol of the highest possible grade were from Merck EAD, Sofia, Bulgaria. Ethylenediaminetetraacetic acid (EDTA), 3-pentanone, 37% hydrochloric acid and trifluoroacetic acid were products of ACROS Organics—Labimex Ltd., Sofia, Bulgaria. The organic solvents acetonitrile, ethanol, ethyl acetate, hexane and isopropanol of simply pure or analytical grade were purchased from Fisher Scientific, Loughborough, Leicestershire, UK., whereas dithiotreitol was also from Fisher Scientific, but in the Waltham, MA, USA. The POP and FAP common substrate benzyloxycarbonyl-glycyl-prolyl-4-methylcoumarin-7-amide (Z-Gly-Pro-AMC) was supplied by Bachem, Switzerland. The inhibition analyses were performed on the recombinant human enzymes: Prolyl Oligopeptidase (rhPOP) from R&D Systems, Minneapolis, MN, USA and Fibroblast Activation Protein α (rhFAP) from Enzo Life Sciences-FOT Ltd., Sofia, Bulgaria. RNase A was supplied by Roche Diagnostics GmbH, Mannheim, Germany. Most of the other reagents including those for cell culturing were from Sigma-Aldrich, Darmstadt, Germany: NaCl, Na2EDTA, Tris, tritonX-100, NaOH, dimethyl sulfoxide (DMSO), propidium iodide (PI), Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS) and antibiotics (penicillin and streptomycin). The disposable consumables were supplied by Orange Scientific, Braine-l’Alleud, Belgium. Annexin-V-FITC/PI apoptosis detection kit, phosphate-buffered saline (PBS), agarose (low melt), agarose (normal melt), 1,1-diphenyl-2-picrylhydrazyl (DPPH), riboflavin, methionine, nitro-blue tetrazolium (NBT), potassium persulfate, 3-ethylbenzothiazoline-6-sulfonic acid (ABTS) and (±)-6-hydroxy-2,5,7,8-tetramethylchromone-2-carboxylic acid (Trolox) were also purchased from Sigma-Aldrich, Darmstadt, Germany.

Experimental Design and Workflow

The schematic overview of the experimental design, linking each step of the study to the corresponding extracts and methods used, was provided in

Table 1. Experimental design of the study.

Step	Extract/Sample	Method/Assay	Purpose/Objective
1	Aerial parts from two Bulgarian regions	Preparation of extracts: 80% methanol, 80% ethanol, 85% isopropanol, 80% acetonitrile, ethyl acetate/water, 3-pentanone/water	Generate diverse extracts for yield evaluation
2	80% ethanol extracts from two regions	UPLC-DAD	Compare chromatographic profiles from different geographic origins
3	Selected extracts (80% ethanol and ethyl acetate/water)	LC-HRMS (negative ionization)	Identify phytochemical compounds (nonvolatile compounds)
4	All extracts	MS and MS/MS data analyses	Identify all the registered compounds
5	80% ethanol and ethyl acetate/water extracts	Enzyme inhibition assays	Evaluate inhibition of post-proline-specific peptidases (POP, FAP)
5	Oenothein B (from extract)	Molecular docking	Predict binding to POP and FAP enzymes
6	80% ethanol extract (selected)	Antioxidant assays: ABTS, DPPH, NBT	Assess antioxidant potential
7	80% ethanol extract	Cell-based assays in HT-29 cells: cell cycle distribution, apoptosis, genotoxicity	Investigate mechanisms underlying antitumor activity

. The design starts with extraction of aerial parts from two Bulgarian regions, followed by chemical characterization, molecular docking, enzyme inhibition assays, antioxidant evaluation, and in vitro studies in colorectal carcinoma cells.

2.2. Methods

2.2.1. Solid–Liquid Extraction

Solid–liquid extractions were performed as described before with some changes [17]. All extractions were performed in triplicate to ensure reproducibility of the extraction yields. Briefly, samples (5 g) of E. parviflorum aerial parts in ground state were suspended in 50 mL of the corresponding aqueous solvent: 80% methanol, 80% ethanol, 85% isopropanol or 80% acetonitrile, acidified by 0.25 mL 1 M hydrochloric acid and stirred for 3 h at RT (room temperature). The filtrated solid residues were extracted two more times for 4 h and overnight in turn. All the filtrates were combined and concentrated in vacuo. Then, they were treated with acetonitrile (5 mL) and the solvent was removed on a rotary evaporator. The same procedure was repeated by successively using acetonitrile, acetone and hexane. Lastly, hexane (10 mL) was added to the solid residues, the suspension was filtrated, the solids were washed with hexane and dried in vacuo.

2.2.2. Biphasic System Extraction

All biphasic extractions were also performed in triplicate to assess reproducibility. Two biphasic systems were used for the extraction process: ethyl acetate/water and 3-pentanone/water in which the organic solvent and water were in ratio 4:1 (40 mL non-polar solvent and 10 mL water). The aqueous layer was acidified by 0.25 mL 1 M hydrochloric acid to pH 3.0. E. parviflorum material (5 g) was extracted at different temperatures as described before [17]—RT for ethyl acetate/water and 60 °C for 3-pentanone/water. Then, a further extraction with 20 mL organic solvent was performed at the corresponding temperature for 2 h. The organic layers were combined, dried over anhydrous sodium sulfate and the solvents were evaporated in vacuo at 40 °C. Then, the extracts were treated with hexane as above.

2.2.3. UPLC-DAD Analyses

Basic Manual system of brand device Thermo Scientific™ UltiMate™ 3000 (Waltham, MA, USA) was used for the analytical purposes. Two types of detectors Dionex UltiMate 3000, both connected to the chromatograph, were used: diode array detector and fluorescence detector. The analytes were separated by means of a Pyramid column Nucleodur C18 from Dueren, Macherey-Nagel, Germany (dimensions: 125 × 4 mm; 3.5 µm). The analyzed materials (10.0 µL each) were injected automatically. The column temperature was maintained at 45 °C and the flow rate at 0.500 mL/min. A linear gradient of two eluents was applied: solvent A—0.1% trifluoroacetic acid in water and solvent B—0.1% trifluoroacetic acid in acetonitrile. The run time was 38 min, and the flow rate was 0.5 mL/min. The gradient elution program was as follows: 0–5 min, 5% B; 5–25 min, 5–30% B; 25–30 min, 30–90% B; 30–33 min, 90% B; and 33–35 min, 90–5% B. Equilibration time was 3 min. The chromatogram was monitored at 215, 260, 280 and 360 nm. The range 200–400 nm encompassed all the individual peaks in the UV spectra. Data were recorded by Thermo Scientific™ Dionex™ Chromeleon™ CDS, version 7.x software.

2.2.4. LC-HRMS Analyses

The mass spectrometer Q Exactive hybrid quadrupole-Orbitrap was supplied by Thetmo Scientific Co., Waltham, MA, USA. It was furnished with TurboFlow^® LC system, heated electrospray HESI II on IonMax^® also from Thetmo Scientific Co., Waltham, MA, USA.

The chromatographic separation of the samples was carried out by HPLC column Nucleoshell RP-18, (50 × 4 mm; 2.7 µm) analytical column (Dueren, Macherey-Nagel, Germany). The used eluents were A: 0.1% formic acid in water and B: 0.1% formic acid in acetonitrile at flow rate of 300 μL/min. During the run time of 36 min, the eluents were applied gradually, as follows: 5% B (2 min), 5–35% B (30 min), 35–90% B (3 min), 90–5% B (3 min). The equilibration was conducted in 2 min. The volume of the analytes was 10.0 μL.

Mass spectrometry settings: full-scan spectra over the m/z range 100–1700 were acquired in negative ion mode at resolution settings of 70,000. The mass spectrometer operating parameters used in a negative ionization mode were spray voltage—4.0 kV; sheath gas flow rate 32, capillary temperature 320 °C; probe heater temperature 300 °C; auxiliary gas flow 10 units; sweep gas 2 units and S-lens RF level 50 (units refer to arbitrary values set by the Q Exactive Tune 2.4 software). All Ion Fragmentation mode of operation of mass analyzer was used for extracts’ compounds identification. Optimized values of the collision energy were HCD 25%. The software program for instrument control was applied for the sensitivity optimization of MS to the target materials. Data were gained and processed with the Xcalibur 2.4^® software package, supplied by Thetmo Scientific Co., Waltham, MA, USA. Mass Frontier 5.1 Software program of the same brand was used to obtain the theoretical m/z values.

2.2.5. POP Inhibition Assay

The kinetic assays were performed exactly as described before [17]. The concentrations of different extracts of E. parviflorum obtained as above varied from 2.5 to 25 μg/mL. The concentration of the extract necessary to inhibit 50% of the activity of POP was denoted as IC₅₀. Data processing was carried out with the program EnzFilter V2.

2.2.6. FAP Inhibition Assay

The kinetic assay of FAP activity was studied exactly as described in Ivanov et al. [17]. Concentrations of the extracts of E. parviflorum were in the interval 2.5 μg/mL to 110 μg/mL. The samples contained FAP in concentration 6.15 μg/mL. Definitions of IC₅₀ and data processing were performed as above (Section 2.2.5).

2.2.7. Molecular Docking

In order to elucidate the molecular bases of oenothein B action on human FAP and POP, we performed molecular docking studies. This procedure can be used to predict the binding interaction of the compound in the active sites of the enzymes. The 3D crystal structures of the FAP (PDB id: 1Z68) and POP (PDB id: 3DDU) were retrieved from the Protein Data Bank [18,19]. All the ions and water molecules were removed. By applying the 1-Click docking online tool, the Mcule platform binding energies (kcal/mol) for oenothein B were calculated. [20]. In this platform, a version of AutoDock Vina 1.2.7. is available, which can be routinely used for docking operations like preparation of ligand and target for the docking procedure [21,22]. In our studies, we choose to regard the enzyme molecules as rigid structures, whereas the ligands (inhibitors) were accepted to be flexible and were optimized by the above-mentioned docking tool. For each enzyme, four possible positions were generated using the stated program and (after calculating the binding energy of each one) we chose the one with the lowest binding energy, since it corresponded to the highest binding affinity. The visualization of different positions of oenothein B in active sites of enzymes was presented by BIOVIA Discovery Studio Visualizer v25.1.0.142884 v25.1.0.142884 [23].

2.2.8. Antioxidant Potential

The antioxidant activity of the extract was evaluated by three different methods (see below) and was presented graphically in percent of inhibition calculated by the following formula:

% antioxidant activity = [(Ac − As) ÷ Ac] × 100,

where

• Ac—absorbance of control solution of the corresponding ROS in methanol (or other solvent).
• As—absorbance of the solution containing both ROS and the extract in a given concentration.

DPPH Radical Scavenging Assay (DPPH^●)

The procedure was performed as described by Brand-Williams et al. [24]. Dry extract was dissolved in DMSO and added to 0.2 mM methanolic DPPH^● solution to obtain twice increasing concentrations of the tested material (from 0.004 to 0.5 mg/mL). The incubation was performed for 30 min at 25 °C. After that, the absorbance of the samples was measured at 517 nm on ELISA reader (TECAN, SunriseTM). A solution of the extract in methanol (in a corresponding concentration) served as a blank and the solution of DPPH^● in methanol served as a control.

Scavenging of Superoxide Anion Radical (O₂^•−) (NBT Test)

The radical anion (O₂^•−) was generated in situ by a photochemical reaction from riboflavin according to the method described before [25]. The reaction mixture contained 1.17 × 10⁻⁶ M riboflavin, 0.2 mM methionine, 2 × 10⁻⁵ M KCN and 5.6 × 10⁻⁵ M tetrazolium salt NBT, all dissolved in 50 mM phosphate buffer, pH 7.8. Different amounts of the extract solution in DMSO were added to obtain a series in the range 0.002 to 0.033 mg/mL. After 7 min of incubation at 25 °C, the absorbance was measured at 560 nm. Blanks and controls were as described above.

ABTS Radical Scavenging Activity (ABTS^+●)

The assay was performed by the method of Re et al. with the slight modifications proposed by Raynova et al. [26,27]. The radicals ABTS^•+ were generated in dark by mixing equimolar solutions of ABTS and potassium persulfate 12–16 h before starting the test. Briefly, methanol was added to the ABTS^•+ solution until the absorbance at 734 nm reached a value of 1.100. To 1.5 mL of this solution the extract dissolved in DMSO was added to obtain solutions in a concentration range from 0.015 to 1.0 mg/mL. The samples were incubated at RT for 15 min, after which the absorbance was measured at 734 nm. Blanks and controls as well as the Trolox equivalent calculations were performed as described in Raynova et al. [21].

2.2.9. Cells Culturing

Human colorectal carcinoma cells of HT-29 cell line (ATCC, Manassas, VA, USA) were cultured in Dulbecco’s Modified Eagle’s medium–high glucose (DMEM 4.5 g/L glucose), supplied with 10% fetal bovine serum and antibiotics in usual concentrations in a humidified atmosphere with 5% CO² at 37.5 °C. The cells were routinely grown as monolayers in 25 cm² tissue culture flasks. Cell separation was performed by trypsinization using Trypsin-0.25% EDTA.

2.2.10. Flow Cytometric Analysis of the Cell Cycle

Cells were treated with 31.2 μg/mL 80% ethanol extract of E. parviflorum for 24 h. Controls represented untreated with the extract cells. In both cases, cells at about 80% confluence were trypsinized with Trypsin-EDTA, followed by centrifugation for 10 min at 1000 rpm and washing with PBS. Cold ethanol was used as a fixative, which was added dropwise at vigorous stirring. Cells were fixed for no less than 12 h in a freezer at −20 °C. After that, they were washed with PBS, treated with 20 µg/mL RNase A solution for 30 min, and stained by addition of 20 µg/mL solution of PI. Cells in different stages of the cell cycle were recorded by flow cytometer Becton Dickinson (San Jose, CA, USA) when 10,000 events were detected from every sample. The percentage of cells in G1, S and G2/M phases was defined with the software FlowJoTM v10.8 (BD Biosciences, San Jose, CA, USA). Data were presented as mean ± SD of three independent experiments.

2.2.11. Flow Cytometric Analysis of Apoptosis

The analysis was performed using the Annexin V-FITC/PI Apoptosis Detection Kit. The procedure strictly followed the recommendations of the kit supplier. Thus, cells of the line HT-29 (1 × 10⁵ cells/well) seeded in a 6-well plate were treated with 31.2 μg/mL (IC₅₀) 80% ethanol extract for 24 h. As above, untreated cells were used as controls. After reaching about 80% confluence, all the cells (treated and controls) were separated with Trypsin-EDTA. The cell suspension was centrifuged at 1000 rpm for 10 min and washed twice with PBS. Then, they were suspended again in binding buffer (0.01 M Hepes/NaOH, pH 7.4) to which 0.14 M NaCl and 2.5 mM CaCl₂ were added. Further on, Annexin V-FITC (5 µL) and PI (5 µL) were applied to 100 µL of the suspension. After 15 min incubation at RT, 10,000 cells were recorded with a flow cytometer (BD LSR II) and analyzed using Diva 6.1.1. software (BD Biosciences, San Jose, CA, USA).

2.2.12. Alkaline Comet Assay

The alkaline comet assay was performed by roughly following a previously described procedure (Walsh and Kato, 2023) with some modification [28]. Cells were seeded (1 × 10⁵ cells/well) in 6-well plates and treated with 80 μg/mL 80% ethanol extract for 4 and 24 h. Cells treated with 150 mM H₂O₂ for 3 min in dark were used as a positive control. After the incubation, cells were trypsinized, re-suspended in PBS, centrifuged at ~150–300× g for 5 min at 4 °C and washed with ice-cold PBS. The cell suspensions were mixed with 1% low melt agarose and placed on the slides pre-coated with 1% (wt/vol) normal melt agarose solution. The slides with encapsulated cells were placed in standard lysis solution (89 mL of lysis stock solution: 2.5 M NaCl, 100 mM EDTA, 10 mM Tris base, pH 10–10.5 with freshly added 10 mL of DMSO and 1 mL of Triton X-100) for at least 1 h at 4 °C in dark. Then, they were transferred directly to the electrophoresis tank containing cold (4 °C) solution (0.3 M NaOH and 1.0 mM Na2EDTA, pH > 13) and incubated for 20–40 min. The electrophoresis was performed at 25 V (~1 V/cm) for 30 min, pH > 13 at 4 °C. The gels were neutralized by washing in cold neutralizing solution (400 mM Tris/HCl, pH 7.5). Then, they were washed in cold (4 °C) dH₂O, drained and stained by silver nitrate according to the procedure describe by Nadin et al. [29]. Stained comet slides were photographed on a microscope Leica DM 5000B (Wetzlar, Germany) and analyzed using the free internet software CaspLab version 1.2.2. One hundred cells were randomly selected from each sample and the percentage of DNA in the tail of comets (% tail DNA) was measured. These cells were scored according to % tail DNA into the following five classes/categories as proposed by Noroozi et al. [30]: 0 class (no damage)—1–5%; 1 class (low damage)—5–25%; 2 class (medium damage)—>25–50%; 3 class (high damage)—>50–75%; 4 class (very high damage)—>75–90%. The DNA damage index was calculated using the following formula:

DNA damage index = 0 × (n) + 1 × (n) + 2 × (n) + 3 × (n) + 4 × (n),

where n is the number of cells in each category.

2.2.13. Statistical Analyses

The results of all the chemical and biological analyses were expressed as mean values ± SD of three replicates for each experiment. Statistical analyses were performed by the one-way analysis of variance (ANOVA) test using GraphPad Prism 8.0 software; p < 0.05 was considered statistically significant.

3. Results and Discussion

3.1. Extractions and UPLC-DAD Analyses

Conventional methods of extraction applied hitherto for plants of Epilobium genus include aqueous methanol, ethanol or acetone or simply pure water [2,31,32]. The amount of the organic solvent in aqueous medium influences total yield and, also, the yields of particular compounds. Furthermore, the use of mild acidic conditions favors the extraction of polyphenolic compounds from plants [33]. In our previous study [17], we applied biphasic systems such as ethyl acetate/water and 3-pentanone/water where the aqueous layers were acidified. In this manner, we obtained good yields of polyphenols from the leaves of Cotinus coggygria Scop. In the present study, we use the same monophasic organic solvent/water systems and biphasic systems. The extraction yield of aerial parts of Epilobium parviflorum varied significantly among solvents (one-way ANOVA, p < 0.05). The highest yield was obtained with 80% acetonitrile (17.0 ± 0.5%), followed by 80% ethanol (14.1 ± 0.35%) and 80% ethanol (Ea ^a, 12.5 ± 0.2%). Methanol (9.5 ± 0.3%) and isopropanol (8.4 ± 0.4%) showed intermediate yields, while ethyl acetate/water (3.2 ± 0.1%) and 3-pentanone/water at 60 °C (2.2 ± 0.05%) produced the lowest yields. Bonferroni post hoc analysis indicated that all extracts differed significantly from each other (p < 0.05), as indicated by different letters in

Table 2. Extraction solvents and yields of several extracts from the dried aerial parts of E. parviflorum samples, 5 g each.

Extract	Extraction Solvent	Temperature (°C)	Yield
			(g)	(%)
M	80% methanol	r.t.	0.475 ± 0.015	9.5 ± 0.3 a
E	80% ethanol	r.t.	0.706 ± 0.026	14.1 ± 0.35 b
Ea a	80% ethanol	r.t.	0.625 ± 0.015	12.5 ± 0.2 c
I	85% isopropanol	r.t.	0.421 ± 0.022	8.4 ± 0.4 d
N	80% acetonitrile	r.t.	0.852 ± 0.031	17.0 ± 0.5 e
AW	Ethyl acetate/water	r.t.	0.158 ± 0.008	3.2 ± 0.1 f
PW	3-pentanone/water	60	0.110 ± 0.005	2.2 ± 0.05 g

^a Extract from aerial part, purchased from Alin^®; r.t.—room temperature. Data are mean ± SD of three independent samples. Statistical differences were determined by one-way ANOVA followed by Bonferroni post hoc test. Column values with different small letters indicate significant differences (p < 0.05).

.

Using the UPLC-DAD method, identical chromatographic profiles of the extracts obtained with monophasic systems were observed. On the other hand, qualitative differences between the extracts in monophasic and biphasic systems exist. UV chromatograms of 80% ethanol and ethyl acetate/water extracts from the herb can be seen in Supplementary Figure S1.

Further on, we compared the composition of 80% ethanol extracts obtained in the same way from the herb purchased from two Bulgarian companies with different geographic regions of activity: Alin^® Samokov municipality and Dicrassin Ltd. Pazardzhik municipality. The UPLC-DAD analyses showed a noticeable similarity between the chromatographic profiles (Supplementary Figure S2). This result gives a reason to accept that the herbs grown in different regions of the country have similar chemical compositions.

3.2. Chemical Composition of the Extracts

Nonvolatile compounds in the extracts from aerial part of E. parviflorum as determined by LC-HRMS analyses in negative ionization mode are presented in Figure 1. All the registered compounds were routinely identified by MS and MS/MS data analyses and, in comparison, to the records in previous studies. All the data about the extracts’ compositions are summarized in

Table 3. Identification of phytochemical compounds in E. parviflorum extracts by LC-HRMS in negative mode.

Peak	RT (min)	Experimental m/z Error (ppm)	Molecular Formula	MS/MS Fragments m/z, (R.I., %)	Proposed Compound	Sample
1	0.64	181.0709 (−4.37)	C6H14O6	-	Hexan-1,2,3,4,5,6-hexol	AW, E, I, M, N, PW
2	0.80	331.0674 (1.12)	C13H16O10	211.0244 (9), 169.0129 (85), 151.0026 (5), 125.0232 (100), 124.0154 (32), 123.0075 (16)	O-Galloylglucose	E, Ea, I, M, N
3	1.18	169.0132 (−6.01)	C7H6O5	125.0232 (100), 124.0154 (33), 123.0075 (30)	Gallic acid	AW, E, Ea, I, M, N, PW
4	1.61	633.0740 (1.11)	C27H22O18	300.9992 (100.00), 275.0201 (56), 257.0094 (7), 249.0405 (24), 247.0251 (5), 231.0295 (14), 229.0137 (5), 203.0344 (6), 169.0132 (18), 125.0230 (8), 123.0074 (5)	Galloyl-HHDP-glucose	E, I
5	2.36	153.0182 (−7.07)	C7H6O4	-	Dihydroxybenzoic acid (isomer I)	PW
6	2.95	137.0232 (−8.96)	C7H6O3	-	Hydroxybenzoic acid (isomer I)	AW
7	3.57	285.0618 (0.73)	C12H14O8	153.0180 (100)	Dihydroxybenzoyl pentose	E, Ea, I, M, N
8	3.64	183.0291 (−4.63)	C8H8O5	181.0136 (20), 169.0135 (15), 153.0183 (33), 139.0026 (32), 125.0240 (18), 124.0155 (100), 123.0075 (91)	3,4-Dihydroxy-5-methoxybenzoic acid (3-O-Methylgallic acid)	AW
9	3.77	137.0232 (−8.96)	C7H6O3	-	Hydroxybenzoic acid (isomer II)	PW
10	4.41	153.0182 (−7.07)	C7H6O4	-	Dihydroxybenzoic acid (isomer II)	PW
11	5.27	179.0341 (−4.81)	C9H8O4	135.0440 (100), 134.0361 (64), 133.0283 (43), 117.0336 (27)	Caffeic acid	AW
12	5.43	295.0460 (0.27)	C13H12O8	163.0390 (22), 161.0234 (9), 133.0283 (3), 119.0489 (100), 117.0333 (12)	p-Coumaroyltartaric acid (coutaric acid)	E, I, M
13	5.70	1567.1446 (−0.02) 783.0701 [M-2H]2−	C68H48O44	765.0568 (7), 613.0427 (5), 597.0533 (6), 472.9778 (5), 450.9950 (16), 445.0413 (6), 427.0304 (8), 425.0150 (5), 399.0354 (5), 300.9989 (100), 299.9906 (17), 298.9835 (20), 275.0199 (67), 273.0045 (33), 247.0249 (45), 245.0091 (11), 231.0295 (30), 229.0139 (15), 169.0132 (32), 125.0230 (13), 123.0074 (16)	Oenothein B	AW, E, Ea, I, M, N
14	5.79	177.0184 (−5.16)	C9H6O4	149.0233 (21), 148.0153 (80), 133.0283 (100), 132.0204 (19), 121.0285 (5)	6,7-Dihydroxycoumarin (Aesculetin, Esculetin)	PW
15	6.26	179.0341 (−4.81)	C9H8O4	177.0184 (11), 151.0386 (6), 150.0350 (7), 149.0230 (12), 135.0440 (62), 134.0360 (49), 133.0283 (100.00), 132.0206 (18), 125.0234 (45), 121.0283 (13)	6,7-dihydroxy-3,4-dihydrocoumarin	PW
16	6.48	1567.1445 (−0.02) 783.0701 [M-2H]2−	C68H48O44	765.0568 (6), 613.0427 (5), 597.0533 (5), 472.9778 (5), 450.9950 (14), 445.0413 (5), 427.0308 (7), 425.0150 (5), 399.0354 (5), 300.9989 (100), 299.9906 (19), 298.9835 (20), 275.0199 (66), 273.0045 (33), 247.0249 (44), 245.0091 (10), 231.0295 (32), 229.0139 (15), 169.0132 (36), 125.0230 (11), 123.0074 (10)	Oenothein B (isomer I)	E, Ea, I, M, N
17		325.0567 (0.69)	C14H14O9	193.0498 (45), 191.0343 (43), 189.0185 (32), 163.0392 (20), 149.0590 (5), 135.0425 (7), 134.0360 (100 ), 133.0282 (14)	Feruloyltartaric acid (fertaric acid)	E, Ea, I, M
18	7.64	1567.1445 (−0.02) 783.0701 [M-2H]2−	C68H48O44	765.0567 (5), 613.0428 (5), 597.0527 (5), 472.9776 (5), 450.9950 (18), 445.0414 (5), 427.0311 (8), 425.0150 (5), 399.0353 (5), 300.9989 (100), 299.9905 (18), 298.9836 (20), 275.0199 (67), 273.0045 (33), 247.0249 (42), 245.0091 (11), 231.0295 (28), 229.0139 (15), 169.0132 (40), 125.0230 (5), 123.0074 (5)	Oenothein B (isomer II)	E, Ea, I, M
19	8.04	163.0390 (−6.00)	C9H8O3	145.0282 (7), 135.0440 (11), 119.0489 (100)	p-Coumaric acid	AW
20	8.19	392.0387 [M-H]•− (0.56)	C17H14O11	223.0243 (5), 211.0237 (9), 193.0134 (8), 169.0132 (100), 151.0025 (8), 125.0231 (81), 123.0074 (36)	3-Galloyloxy-2-oxopropyl gallate (1,3-Digalloyoxyacetone)	E, Ea, I
21	8.74	225.1129 (−1.25)	C12H18O4	159.0442 (37), 151.0384 (49), 149.0233 (100), 147.0804 (54), 133.0648 (29)	2-[hydroxypentenyl]-3-oxo-cyclopentyl]acetic acid	PW
22	9.14	163.0390 (−6.00)	C9H8O3	119.0489 (100)	Hydroxycinnamic acid	PW
23	10.01	631.0947 (−1.04)	C28H24O17	479.0838 (5), 317.0280 (28), 316.0226 (100), 287.0199 (7), 275.0198 (9), 271.02526 (14), 270.0173 (5), 247.0250 (5), 178.99751 (5), 169.0132 (12), 151.0024 (5), 125.0230 (5)	Myricetin-3-O-(O-galloyl)-glucopyranoside	AW, E, Ea, I, M, N, PW
24	10.90	479.0832 (0.22)	C21H20O13	317.0274 (11), 316.0226 (46), 287.0201 (65), 271.0251 (100), 270.0173 (15), 259.0248 (18), 242.0218 (15), 214.0267 (17), 178.9974 (6), 151.0022 (10), 137.0232 (5)	Myricetin-3-O-glucoside	AW, E, Ea, I, M, N, PW
25	12.04	433.0413 (0.16)	C19H14O12	300.9991 (100), 299.9917 (74), 283.9965 (8), 282.9889 (7), 273.0041 (7), 257.0088 (5), 245.0092 (6), 229.0136 (10), 228.0064 (7), 216.0061 (9), 200.0108 (6)	Ellagic acid pentoside	E, Ea, I, M, N
26		313.1297 (1.45)	C15H22O7	229.0141 (100), 173.0235 (26), 157.0283 (11), 149.0596 (24)	Unidentified	E, I, M
27	12.26	497.3344 (0.10)	C25H46N4O6	451.3308 (10), 433.3195 (14), 333.2305 (38), 324.2669 (18), 265.1482 (39), 242.1874 (27), 225.1607 (68), 224.1767 (100), 207.1500 (65)	Unidentified	AW, E, Ea, I, M, N, PW
28	12.66	300.9991 (0.30)	C14H6O8	283.9966 (21), 257.0098 (6), 245.0089 (13). 244.0009 (12), 229.0136 (14), 228.0059 (19), 217.0130 (17), 216.0060 (36), 201.0175 (23), 200.0107 (74), 199.0030 (35), 189.0187 (26), 185.0239 (17), 173.0232 (32), 172.0157 (63), 171.0078 (28), 163.0391 (30), 161.0234 (53), 160.0155 (41), 145.0283 (100.00), 144.0204 (30), 133.0283 (50), 132.0204 (42), 129.0333 (15)	Ellagic acid	AW, E, Ea, I, M, AW, E,
29		463.0886 (0.91)	C21H20O12	317.0278 (14), 316.0228 (55), 287.0200 (65), 271.0251 (100), 270.0172 (14), 259,0250 (20), 243.0293 (10), 242.0219 (15), 214.0267 (18), 178.9970 (7), 151.0024 (12), 137.0232 (5)	Myricetin-3-O-rhamnoside (Myricitrin)	Ea, I, M, N, PW, N, PW
30	12.82	297.1346 (0.84)	C15H22O6	189.0185 (51), 177.0186 (17), 175.0395 (15), 161.0235 (100), 149.0229 (12), 133.0285 (94), 129.0332 (31)	Unidentified	E, I, M, N
31	13.73	297.1346 (0.84)	C15H22O6	177.0183 (30), 175.0392 (51), 149.0233 (36), 133.0282 (100)	Unidentified	E, I, M, N
32	13.92	297.1345 (0.63)	C15H22O6	177.0183 (14), 175.0392 (15), 161.0235 (100), 133.0281 (46)	Unidentified	E, I, M, N
33	14.46	187.0968 (−4.07)	C9H16O4	141.0909 (100), 125.0959 (38), 123.0801 (17)	Nonanedioic acid (Azelaic acid)	AW, E, Ea, I, PW
34		313.1297 (1.45)	C15H22O7	255.0300 (100), 227.0347 (86), 149.0596 (15)	Unidentified	E, I, M
35	15.20	297.1345 (0.63)	C15H22O6	177.0183 (34), 175.0393 (35), 149.0231 (19), 121.0282 (100)	Unidentified	E, I, M, N
36	15.31	447.0935 (0.50)	C21H20O11	301.0336 (13), 300.0280 (29), 271.0254 (100), 255.0300 (48), 243.0299 (22), 227.0346 (10), 178.9980 (5), 163.0027 (6), 151.0025 (15)	Quercetin-3-O-rhamnoside (Quercitrin)	AW, E, Ea, I, N, PW
37	16.17	317.0302 (−0.25)	C15H10O8	287.0189 (8), 271.0249 (26), 287.0208 (8), 259.0236 (5), 255.0296 (18), 243.0296 (8), 227.0345 (8), 178.9978 (16), 169.0134 (18), 151.0025 (72), 137.0232 (100), 123.0075 (6), 121.0284 (6)	Myricetin	AW, PW
38	16.53	359.0775 (0.67)	C18H16O8	197.0447 (5), 179.0343 (8), 161.02345 (34), 135.0440 (69), 133.0284 (100), 132.0205 (31), 123.0439 (8)	Rosmarinic acid	AW, PW
39	17.63	431.0985 (0.60)	C21H20O10	285.0401 (17), 284.0333 (17), 255.0300 (100), 227.0348 (84), 229.0510 (11)	Kaempferol-3-O-rhamnoside	AW, PW
40	18.16	711.3970 (−0.68)	C37H60O13	503.3386 (100), 485.3278 (19), 473.3263 (5), 453.3015 (10), 441.3412 (5), 421.3123 (5), 409.3109 (7)	19-α-Hydroxyasiatic acid derivative (I)	AW, Ea, PW
41	19.23	287.0565 (1.21)	C15H12O6	151.0390 (7), 151.0026 (16), 135.0440 (100)	Eriodictyol	AW, I, PW
42	19.40	711.3970 (−0.68)	C37H60O13	503.3389 (100), 485.3277 (24), 459.3482 (5), 441.3412 (5)	19-α-Hydroxyasiatic acid derivative (II)	AW, Ea, I, PW
43	20.33	637.1782 (1.27)	C29H34O16	283.0613 (100), 268.0380 (49)	Acacetin-O-heptosylglucoside	E, I, M, N
44	20.48	301.0355 (0.58)	C15H10O7	271.0259 (98), 255.0307 (19), 243.0299 (5), 227.0357 (5), 178.9984 (6), 151.0025 (100), 121.0282 (90)	Quercetin	AW
45	20.58	285.0407 (0.87)	C15H10O6	227. 0356 (5), 175.0391 (8), 151.0026 (15), 145.0282 (9), 133.0283 (100), 132.0204 (45), 121.0282 (14)	Luteolin	AW
46	24.25	327.2179 (0.71)	C18H32O5	239.1287 (6), 229.1448 (34), 221.1177 (12), 211.1337 (51), 209.1187 (10), 193.1225 (5), 185.1169, (8), 183.1387 (9), 181.1234 (5), 171.1018 (100), 165.1277 (47), 137.0962 (24), 135.0804 (12), 127.0752 (11)	9,12,13-trihydroxyoctadeca-10,15-dienoic acid	AW, E, Ea, I, PW
47	24.96	695.4022 (−0.37)	C38H56N4O8	487.3440 (100), 469.3331 (24)	Unidentified	AW, Ea, PW
48	25.56	695.4022 (−0.37)	C38H56N4O8	487.3441 (100), 473.3286 (8)	Unidentified	AW, Ea, PW
49	26.37	329.2337 (1.07)	C18H34O5	229.1449 (62), 211.1342 (100), 209.1181 (10), 183.1383 (17), 171.1022 (70), 165.1279 (51), 139.1116 (33), 137.0958 (10), 127.0752 (11)	9,12,13-trihydroxyoctadec-10-enoic acid (isomer I)	AW, E, Ea, I, PW
50	26.69	329.2336 (0.80)	C18H34O5	229.1458 (7), 211.1341 (27), 201.1125 (22), 171.1019 (100), 165.1276 (34), 155.1064 (28), 139.1114 (56)	9,10,13- trihydroxyoctadec-11-enoic acid (isomer II)	AW, E, Ea, I, PW AW, E, Ea, I, PW
51		343.0461 (0.49)	C17H12O8	328.0231 (5), 313.0000 (14), 297.9756 (34), 285.0038 (6), 269.9809 (100.00), 241.9856 (24), 213.9904 (33), 197.9950 (30), 185.9949 (45), 173.9950 (5), 169.9998 (7), 157.9997 (22), 145.9997 (5), 142.0048 (8), 130.0047 (11)	2,3,8-Tri-O-methylellagic acid
52	27.30	416.1618 (0.60)	C24H23N3O4	346.1571 (5), 319.1455 (6), 263.0829 (8), 249.0671 (80), 237.0668 (10), 222.0557 (5), 221.0716 (5), 220.0771 (5), 219.0557 (14), 196.0397 (20), 194.0603 (100)	Unidentified	AW
53	27.71	287.2230 (0.76)	C16H32O4	201.1120 (31), 171.1019 (65), 155.1065 (16), 153.0908 (19), 137.0961 (22), 135.0804 (19), 125.0958 (100)	9,10-Dihydroxyhexadecanoic acid	AW, PW
54	28.54	213.0552 (−2.38)	C13H10O3	169.0654 (48), 135.0075 (100)	Unidentified	AW, PW

AW—ethyl acetate/water; E—80% ethanol; Ea—80% ethanol (aerial part, purchased from Alin^®); I—85% isopropanol; PW—3-pentanone/water; M—80% methanol; N—80% acetonitrile.

and discussed below. In the extracts, thirteen types of compounds and their derivatives were detected (Figure 2).

3.2.1. Hydroxybenzoic Acids and Derivatives

Compounds 6 and 9 gave [M-H]⁻ ions at m/z 137.0232, corresponding to a molecular formula C₇H₆O₃, which represented two isomers of hydroxybenzoic acid. Compounds 5 and 10 gave [M-H]⁻ ions at m/z 153.0182, corresponding to a molecular formula C₇H₆O₄, which are identified as two isomers of dihydroxybenzoic acid. Compound 7 with [M-H]⁻ ion m/z at 285.0618 (C₁₂H₁₃O₈) gave a fragment ion at m/z 153, indicating the loss of pentose moiety (132 Da). This compound was identified as dihydroxybenzoyl pentose. Compound 3 with [M-H]⁻ ion at m/z 169.0132 was found to correspond to a molecular formula C₇H₆O₅. Observed in the MS/MS, spectrum ions of a higher relative abundance were at m/z 125 ([M-H-CO₂]⁻), 124 ([M-2H-CO₂]^•−) and 123 ([M-3H-CO₂]⁻). This fragmentation is typical for 3,4,5-trihydroxybenzoic acid (gallic acid). Compound 8 with [M-H]⁻ ion at m/z 183.0291 corresponded to the gross formula C₈H₈O₅. Its fragment ions were at m/z 169 [M-H-CH₂]⁻, also corresponding to gallic acid as well as⁻ other fragments at m/z 125, 124 and 123. The neutral loss of 14.0156 Da (carbene) and 30.0107 Da (CH₂O) indicated that this compound was 3-O-methylgallic acid. Compound 20 formed in MS/MS spectrum radical anion [M-2H]^•− at m/z 392.0387 (C₁₇H₁₂O₁₁) and base peak at m/z 169. This indicated a gallic acid derivative. The product ion at m/z 223 (C₁₀H₇O₆) which was observed in MS/MS spectrum was obtained from the molecular radical anion after elimination of the gallic acid radical. Specific fragment ion at m/z 211 was observed as well. Based on the above, we identified this compound as 1,3-Digalloyoxyacetone.

3.2.2. Ellagic Acid and Derivatives

Ion [M-H]⁻ observed at m/z 300.9991 (C₁₄H₅O₈) exhibited fragments at m/z 284 ([M-HO]^•−), 257 ([M-CO₂]⁻), 245 ([M-2CO]⁻), 229 ([M-CO₂-CO]⁻), 217 ([M-3CO]⁻), 201 ([M-CO₂-2CO]⁻), 173 ([M-CO₂-3CO]⁻) and base peak 145 ([M-CO₂-4CO]⁻). Using the previously published MS/MS spectrum [34] as well as those presented in the MassBank of North America spectrum [35], we identified compound 28 as ellagic acid. Compound 25 was identified as ellagic acid pentoside (m/z 433), with a characteristic loss of 132 Da (pentose moiety) and a base peak in MS/MS spectrum at m/z 300.9991. Compound 51 showed an [M-H]- ion at m/z 343.0461, indicating a molecular formula C₁₇H₁₂O₈. The MS/MS spectrum of 51 had fragment ions at m/z 328.0231 [M-H-CH₃]^•−, 313.0000 [M-H-2CH₃]⁻, 297.9756 [M-H-3CH₃]^•−, 269.9809 [M-H-3CH₃-CO]^•− and 241.9856 [M-H-3CH₃-2CO]^•−, which were produced from 2,3,8-tri-O-methylellagic acid [36].

3.2.3. Hydroxycinnamic Acids and Derivatives

Compounds 11, 19 and A1 with [M-H]⁻ ions at m/z 179 and 163, respectively, generated the most abundant ions [M-CO₂]⁻ at m/z 135 and 119. These compounds were identified as caffeic acid, p-coumaric acid and another isomer of hydroxycinnamic acid. MS/MS fragmentation of [M-H]⁻ ion (peak 12) at m/z 295 produced ions at m/z 163 ([p-coumaric acid-H]⁻), m/z 119 ([p-coumaric acid-H-CO₂]⁻) and an ion with low abundance at m/z 149.0081 ([tartaric acid-H]⁻). Similarly, peak 17 produced MS/MS daughter ions at m/z 193 ([ferulic acid-H]⁻), m/z 163 ([ferulic acid-H-CH₂O]⁻), m/z 149.0590 ([ferulic acid-H-CO₂]⁻) and a peak with low intensity at m/z 149.0081 ([tartaric acid-H]⁻). Therefore, compounds 12 and 17 were identified as coutaric acid and fertaric acid, respectively. Compound 38 gave [M-H]⁻ ion at m/z 359.0775 (molecular formula C₁₈H₁₆O₈) in its MS, and produced fragment ions in their MS/MS spectrum at m/z 197, 179, 161, 135 and 133. This agrees with the fragmentation of rosmarinic acid [37].

3.2.4. Hydrolyzable Gallotannins

Compound 2 gave an [M-H]⁻ ion at m/z 331.0668 corresponding to the molecular formula C₁₃H₁₆O₁₀. The relatively abundant ions in the MS/MS spectrum were at m/z 169 and 125, which is typical for gallic acid and its fragment ion. The neutral loss of 162 Da corresponded to a hexose moiety led to the formation of an ion at m/z 169. The mass spectrum was compared to the data in [38], which led to identification of O-galloyl glucose.

3.2.5. Hydrolyzable Ellagitannins

The negative HRESI-MS of compound 4 showed [M-H]⁻ ion at m/z 633.0740, corresponding to the molecular formula C₂₇H₂₂O₁₈. The fragment ion at m/z 300.9995 as a base peak was associated with [ellagic acid]⁻. The fragment ion at m/z 169.0132 corresponded to [gallic acid]⁻. However, fragment ions typical of ellagic acid were missing in MS/MS spectrum, indicating that there is no ellagic acid moiety in the structure of compound 4. The ion at m/z 300.9995 can be formed as a result of lactonization of hexahydroxydiphenic acid (HHDPA). The ions in MS/MS spectrum at m/z 275 ([HHDPA-H-CO₂-H₂O)]⁻), 257 ([HHDPA-H-CO₂-2H₂O)]⁻), 249 ([HHDPA-H-2CO₂)]⁻) and 231 ([HHDPA-H-2CO₂-H₂O)]⁻) are characteristic of the fragmentation of HHDPA. Thus, the formation of the base peak was a consequence of a neutral loss of 332 Da. There is a gallic acid (170 Da) in the structure of this molecule, meaning that the mass of the remainder is 162 Da, corresponding to hexose moiety. As a result of this analysis, we have identified this substance as galloyl-hexahydroxydiphenoyl (HHDP)-glucose.

In the TIC (total ionic chromatogram) of the extracts from E. parviflorum, three compounds (13, 16 and 18) were found, with low abundance [M-H]⁻ ion at m/z 1567.1445 and an ion at m/z 783.0701 ([M-2H]²⁻) with the highest relative abundance. This finding is in accordance with the molecular formula C₆₈H₄₈O₄₄. In the extracts of plans of Epilobium genus (Onagraceae family), oenothein B is the dominating compound [39,40]. RP-UPLC analysis of extract from the aerial parts of E. parviflorum 80% ethanol extract showed that the main substance with retention time 5.70 min had absorption maximum at 262 nm. The obtained daughter ions in MS/MS spectrum of compound 13 are in accordance with fragmentation of oenothein B [41]. The compounds 16 and 18 had practically identical MS/MS spectra and were probably isomers of oenothein B.

3.2.6. Coumarins

The peak 14 with [M-H]⁻ at m/z 177 was identified as 6,7-dihydroxycoumarin (aesculetin), which showed a base peak at m/z 133 corresponding to [M-H-CO₂]⁻ and fragment ions at m/z 149 ([M-H-CO]⁻) and 148 ([M-2H-CO]^•−). The peak 15 exhibited a deprotonated molecular ion at m/z 179 in the MS spectrum. The main detected fragments were at m/z 135 ([M-H-CO₂]⁻), 134 ([M-2H-CO₂]^•−) and 133 ([M-H-H₂-CO₂]⁻), as well as an ion at m/z 177 of low abundance. This compound was 6,7-dihydroxy-3,4-dihydrocoumarin.

3.2.7. Flavonoids and Derivatives

Compounds 37 and 44 were identified only in the ESI/MS spectrum of the ethyl acetate/water extract by the ions at m/z 317 and 301, respectively. In their MS/MS spectra, fragment ions at m/z 179 (^1,2A⁻) and 151 (^1,3A⁻) were found, which is typical for retro Diels–Alder (RDA) fragmentation of flavon-3-ols. In the MS/MS spectrum of the compound 37, fragment ions at m/z 287 ([M-H-CH₂O]⁻), 259 ([M-H-CH₂O-CO]⁻) and a base peak at m/z 137 (^1,2B⁻) were observed, which was typical for myricetin [17]. On the other hand, in MS/MS spectrum of compound 44, ion at m/z 255 ([M-H-H₂O-CO]⁻) was found, typical for quercetin. As shown in Table 2, there were three compounds (29, 36 and 39) with [M-H]⁻ ions at m/z 463, 447 and 431 with a neutral loss of a mass of 146 Da, corresponding to a rhamnose moiety, which were converted into daughter ions with m/z 317, 301 and 285, respectively. Their MS/MS spectra were practically the same as those in our previous study [17], which gave us a reason to identify them as myricitrin, quercitrin and kaempferol-3-O-rhamnoside. A myricetin derivative 24 with [M-H]⁻ at m/z 479.0832 and molecular formula C₂₁H₂₀O₁₃ was also found in the extracts. Its MS/MS spectrum showed an ion at m/z 317 ([M-H-162]⁻) accompanied by a loss of 162 Da, thus showing a hexose moiety. This compound was identified as myricetin-3-O-glucoside [17]. The analysis of peak 23 (m/z 631.0947, molecular formula C₂₈H₂₄O₁₇) indicated the presence of different fragment ions at m/z 479 ([M-H-152]⁻) after a loss of a galloyl moiety, 317 after a loss of a galloyl and hexose moieties [M-H-152-162]- and a basic peak at m/z 316 ([M-2H-152-162]^•−), corresponding to a myricetin derivative. The higher intensity of the ion ([M-2H-152-162]^•− than the one of [M-H-152-162]⁻helped to find the glycosylation site at 3-OH [42]. Thus, this compound was tentatively considered to be myricetin-3-O-(O-galloyl)-glucoside. In the MS/MS spectrum of the compound 45 ([M-H]⁻ at m/z 285), the main ions at m/z 151 (^1,3A⁻) and 133 (^1,3B⁻) were observed. Based on these results, we concluded that the compound 46 was the flavone luteolin. Compound 41 gave [M-H]⁻ ion at m/z 287. Its MS/MS spectrum gave ions at m/z 151.0390 (^0,3B⁻), 151.0026 (^1,3A⁻) and 135.0440 (^1,3B⁻) as the base peak, suggesting that the compound 41 is eriodictyol. Compound 43 exhibited [M-H]⁻ ion at m/z 637 and molecular formula C₂₉H₃₄O₁₆. The primary product ion at m/z 383 was registered, which corresponded to the neutral loss of C₁₃H₂₂O₁₁. Compound 43 produced also a fragment radical anion at m/z 268 [M-H-C₁₃H₂₂O₁₁-CH3]^•−, corresponding to the loss of a methyl radical from the aglycone. According to the fragmentation, compound 43 was identified as acacetin-O-heptosylglucoside.

3.2.8. Sesquiterpenoids

Six sesquiterpenoids were found in the extracts. Compounds 26 and 34 exhibited [M-H]⁻ ions at m/z 313 (molecular formula C₁₅H₂₂O₇), fragment ions at m/z 149 and base peaks at m/z 229 and 255, respectively. Four isomeric compounds (30, 31, 32 and 35) with [M-H]⁻ ions at m/z 297 (molecular formula C₁₅H₂₂O₆), were detected in ESI/MS. The fragment ions at m/z 177 and 175 were observed in their MS/MS spectra. Furthermore, fragment ions at m/z 149 in the spectra of the compounds 30, 31 and 35 were registered and daughter ions at m/z 133 in the spectra of 30, 31 and 32 were found. However, on the basis of their MS/MS spectra, we could not identify them.

3.2.9. Pentacyclic Triterpenoid Derivatives

Compounds 40 and 42 were considered to be isomers with [M-H]⁻ at m/z 711.3970 in the MS spectra, and their molecular formula was predicted to be C₃₇H₆₀O₁₃. The characteristic ion at m/z [M-H-208]⁻ (base peak) at m/z 503.3386 (C₃₀H₄₇O₆) was present in both MS/MS spectra, consistent with the neutral loss moiety of 2,3,4,5,6,7-hexahydroxyheptanoic acid. Other fragment ions were m/z 485.3278 [M-H-208-H2O]⁻, 453.3015 [M-H-208-H₂O-CH₃OH]⁻, 441.3412 [M-H-208-H₂O-CO₂]⁻ and 409.3109 [M-H-208-H₂O-CO₂-CH₃OH]⁻. By comparing with the previous data [35,43], the ion at m/z 503 was found to correspond to that of 19-α-hydroxyasiatic acid. Compounds 40 and 42 were finally identified as derivatives of 19-α-hydroxyasiatic acid and hexahydroxyheptanoic acid.

3.2.10. Fatty Acids

Peak 33 with [M-H]⁻ at m/z 187.0968 was identified as nonanedioic acid according to its MS/MS spectrum. In it, a base peak at m/z 141 ([M-H-HCOOH]⁻) was observed as well as other fragment ions at m/z 125 ([M-H-CO₂-H₂O]-) and 123 ([M-H-CO₂-H₂O-H₂]⁻). Compounds 46, 49 and 50 were identified as trihidroxy fatty acids. In the negative MS/MS spectrum of compound 46 ([M-H]⁻ at m/z 327 and molecular formula C₁₈H₃₂O₅), the base peak was at m/z 171.1018 (C₉H₁₅O₃) and a high abundance ion at m/z 229.1448 (C₁₂H₂₁O₄) was also observed. Most probably, during the fragmentation, a C-C bond was broken at a carbon atom bounded to a hydroxyl group [44]. Based on this possibility, we considered that the hydroxyl groups are attached to the carbon atoms in positions 9 and 12. Fragmentation of 12,13-dihydroxy fatty acids generated an ion at m/z 183, while 9,10-dihydroxy fatty acids generated a daughter ion at m/z 201 [45]. In MS/MS spectrum of compound 46, an ion at m/z 183 was observed, but no ion was detected at m/z 201. The characteristic fragment ion at m/z 127.0752 (C₇H₁₁O₂) obtained upon cleavage of the C11-C12 bond was observed. The molecular structure of the compound 46 was found to be 9,12,13-trihydroxyoctadeca-10,15-dienoic acid, verified by comparing the daughter ion spectrum with the respective spectrum and data presented in MassBank of North America (Kong) [46]. The MS spectra of compounds 49 and 50 showed [M-H]⁻ ion at m/z 329, corresponding to the molecular formula C₁₈H₃₄O₅, which were trihydroxyoctadecenoic acid isomers. Fragment ions at m/z 229, 211, 183, 177 and 139 with high abundance were seen in the spectrum of compound 49. The daughter ion at m/z 183 indicated that hydroxyl groups were in positions 12 and 13. On the other hand, the ion at m/z 171 indicated that the other hydroxyl group is at position 9. By comparing our results to the previously published spectrum [46], we concluded that the substance was 9,12,13-trihydroxyoctadec-10-enoic acid. In the MS/MS spectrum of compound 50, ions at m/z 201 and 155 were observed, which is typical for 9,10-dihydroxy fatty acids. Ions at m/z 183 were not detected. This led us to assume that compound 50 was 9,10,13-trihydroxyoctadec-11-enoic acid. In the MS/MS spectrum of compound 53 with [M-H]⁻ at m/z 287 and molecular formula C₁₆H₃₂O₄, the base peak was at m/z 125 and high abundance ions at m/z 171 and 201 were observed. The ion at m/z 201 indicated that the position of hydroxyl groups was at 9 and 10. The base peak at m/z 125 was generated from the cleavage of the C8-C9 bond and elimination of H₂O. These data gave us a reason to decide that the compound was 9,10-dihydroxyhexadecanoic acid.

3.2.11. Polyalcohol

Peak 1 corresponded to the molecular formula C₆H₁₄O₆ due to the [M-H]⁻ ion at m/z 181. The compound was identified as hexan-1,2,3,4,5,6-hexol.

3.2.12. Other Compounds

The MS/MS-spectrum of compound 21 ([M-H]⁻ at m/z 225) displayed three fragment ions with high abundance at m/z 151, 149 and 147. Additionally, three fragment ions with very low abundance at m/z 163, 161 and 125 were found. The signal at m/z 125 appeared from a cleavage of the cyclopentanone ring to give C₇H₇O⁻. Based on this result, we recognized this compound as a 2-[hydroxypentenyl]-3-oxo-cyclopentyl] acetic acid.

The compound 54 exhibited a [M-H]⁻ ion at m/z 213, which fragmented into m/z 169 [M-H-CO₂]⁻ and a base peak with m/z 135 [M-H-C₆H₆]⁻. Several compounds–derivatives of benzoic acid or benzophenone are known to give such MS/MS spectra and, therefore, we could not identify compound 54.

3.2.13. Nitrogen-Containing Compounds

In this study, four N-containing compounds were found. The peak 27 with a molecular formula C₂₅H₄₆N₄O₆ was deduced from the [M-H]⁻ ion at m/z 497.3348. The characteristic fragment ions at m/z 242.1874 (C₁₂H₂₄N₃O₂), 225.1607 (C₁₂H₂₁N₂O₂), 224.1767 (C₁₂H₂₄N₃O₂; base peak) and 207.1500 (C₁₂H₁₉N₂O₂) were observed. Compounds 47 and 48 were found to be isomers with [M-H]⁻ ions at m/z 695.4022 (molecular formula C₃₈H₅₆N₄O₈). Formation of a base peak at m/z 487.3441 was a consequence of a neutral loss of 208 Da, corresponding to a moiety from 2,3,4,5,6,7-hexahydroxyheptanoic acid. In an experiment using a lower collision energy, a daughter ion at m/z 649.3966 ([M-H-HCOOH]⁻) was observed, indicating that the carboxyl group of hexahydroxyheptanoic acid was not covalently bound to an aglycone. Compound 52 showed [M-H]⁻ ion at 416.1618 (molecular formula C₂₄H₂₃N₃O₄). The diagnostic ions at m/z 249.0671 (C₁₅H₉N₂O₂), 196.0398 (C₁₂H₆NO₂) and 194.0603 (C₁₃H₈NO; base peak) were present in both MS/MS spectra. It is interesting to note that the three compounds (47, 48 and 52) were found only when using a biphasic extraction system. The available data both from our and previous studies were not enough to determine the structure of these compounds.

Although alkaloids are generally found in plants, there is no evidence of such compounds to be detected in Epilobium genus, except for a single study in which a new alkaloid-angustifoline A has been identified in E. angustifolium [47]. However, in a previous report it is mentioned that alkaloids of yet unknown structure do exist in the herb E. parviflorum of Bulgarian origin [48]. In the present study, we confirm the occurrence of nitrogen-containing compounds in the extracts of the herb from different geographical regions in Bulgaria (compounds 27, 47 and 48; Supplementary Figure S3).

It should be noted that a comparison between our results and the previous data about the chemical composition of E. parviflorum shows forty compounds (1, 2, 6–10, 12, 14, 15, 17, 19–23, 25–27, 30–35, 38, 40–43, 45–54) not identified so far in E. parviflorum [2,4,40].

3.3. Enzymes Inhibition Assays

Peptide bonds formed by proline residues are stable to the action of most peptidases, including those with a broad specificity, which is due to the cyclic structure (pyrrolidine ring) of this amino acid [49]. Nevertheless, there is a comparatively small group of peptidases which are capable to cleave specifically peptide bonds at the carboxyl group of proline (Pro-Xaa). Those enzymes are called post-proline-specific peptidases (PPSPs) (or post-proline cleaving enzymes) and belong to the S9 family of serine peptidases. Most members of this family represent multi-domain proteins with an N-terminal β-propeller and a C-terminal α/β-hydrolase domain. Chief members of PPSPs are fibroblast activation protein alpha (FAP, EC 3.4.21.B28), dipeptidyl peptidase IV (DPPIV, EC 3.4.14.5) and prolyl oligopeptidase (POP, EC 3.4.21.26) [50]. This unique group of enzymes is essential because of their role in both activation and degradation of peptide hormones and neuropeptides [49]. Moreover, changes in the activity of PPSPs can result in different pathological conditions, including cancer [49]. PPSPs are regarded as potential therapeutic targets and, therefore, a search for selective inhibitors of individual enzymes is important [51]. In addition, effective drugs based on natural substances are always to be preferred.

Earlier studies show that penta-O-galloyl glucose and some isomers of tri- and tetra-O-galloyl glucoses are effective noncompetitive inhibitors of POP [52]. In our previous work [17], we obtained extracts from C. coggygria leaves with main components from penta- to nona-O-galloyl glucoses. Kinetic studies showed that those substances are also effective inhibitors of POP with a much lower inhibition activity to FAP. Another group of active POP inhibitors are ellagitannins [52], comprising high molecular weight compounds such as rugosin E. Furthermore, the ellagitannin oenothein B as a major component of the extracts from Epilobium species has a close molecular structure to rugosin E [40]. That is why we decided to check whether oenothein B can also inhibit POP and to determine its selectivity index to FAP.

Thus, kinetic studies were carried out with 80% aqueous ethanol and ethyl acetate/water extracts from aerial parts of E. parviflorum on the human recombinant POP using the fluorogenic substrate Z-Gly-Pro-AMC. The 80% aqueous ethanol extract was found to inhibit POP in a dose-dependent manner (Supplementary Materials Figure S4) with the IC₅₀ 1.72 μg/mL (Table 3). The LC-HRMS and RP-UPLC analyses (Supplementary Materials Figures S1A and S2) indicated that the major compounds in this extract were oenothein B (13) and myricitrin (29), which is in accordance with literature data for extracts from several Epilobium species [40]. In contrast, using the ethyl acetate/water extract, no inhibition of the enzyme was observed at the extract concentrations up to 5 μg/mL (Table 3). In the ethyl acetate/water extract, the major compounds were gallic acid (3) and myricitrin (29), while the amount of oenothein B (13) was very small. According to the previous studies [52], gallic acid does not inhibit POP. Based on the above results, we can ascertain that oenothein B is the substance effectively inhibiting POP.

The inhibition potential of the above two extracts towards FAP was tested also using Z-Gly-Pro-AMC as a substrate. The results showed that both extracts (

Table 4. Inhibitory effects of the E. parviflorum extracts on post-proline-specific peptidases.

Extract	IC50 (μg/mL)
	POP	FAP
80% ethanol	1.72 ± 0.02	NI a
Ethyl acetate/water	NI b	NI a

^a No inhibition of the enzyme from extract at concentrations up to 21 μg/mL. ^b No inhibition of the enzyme from extract at concentrations up to 5 μg/mL. Values (mean ± SD) are average of three samples of each extract.

) did not inhibit the enzyme at concentrations up to 21 μg/mL. These results show that oenothein B is an effective POP inhibitor with a very high selectivity towards FAP.

3.4. Ligand-Protein Docking

The active site of POP is hidden in a large cavity at the interface of the catalytic and β-propeller domains. The central tunnel of the propeller is lined by hydrogen donors and acceptors which are all water solvated. The cavity is large enough to insert the loop at proline. Its volume is 8562 ų, which represents 8% of the volume of the whole enzyme [53]. Crystallographic structures and kinetic data show that POP exists in “open” and “closed” conformations. In the free status, the inter-domains interface is loose, thus providing access to the catalytic site. The inhibitor binding “locks” the two domains together [54], thus stabilizing the “closed” conformation. Considering these structural features of the active site of POP, as well as the molecular structure and molecular volume (about 1827 ų) of oenothein B, it can be explicated why this compound is an effective inhibitor for the enzyme.

In the structure of FAP, the organization of blades in the propeller domain leads to the formation of a central pore with a length of 27 Å and a diameter of 14 Å. The side opening has a diameter of 24 Å [18]. The specific molecular structure of oenothein B and its geometric parameters are probably the reason why this molecule cannot penetrate the central pore of the enzyme and/or does not interact well with the amino acid residues therein. In order to check out those rationales, we performed molecular docking tests.

Currently, computer molecular modeling methods are becoming an integral part of fundamental research aimed at studying the molecular mechanisms of protein functioning, as well as applied projects related to the rational design of novel medicinal compounds [55,56]. The method of molecular modeling, which aims at finding the most reliable orientation and conformation of a ligand in the binding site of a target protein, is called molecular docking. Molecular docking allows predicting the spatial structure of the receptor–ligand complex and the free energy of its formation based on data about the spatial structure of the receptor and the chemical structure of the ligand.

The structures of the enzyme–inhibitor complexes of oenothein B with POP and FAP were demonstrated by docking program AutoDock Vina for predicting binding models of a non-covalent inhibitor. The positions of oenothein B in active sites of selected enzymes with the highest binding affinity in representative 2D-docking diagrams are given in Figure 3 and Figure 4.

The binding positions of the ligand (oenothein B) with the two enzymes, generated by docking and corresponding to the complexes with the lowest energy are shown in Figure 3A and Figure 4A. In both complexes, the ligand does not form contacts with the catalytic cites of the enzymes. In POP–oenothein B complex, the ligand is at a distance 5.6 Å from catalytic Ser554, whereas in FAP–oenothein B complex, the ligand is at a distance 5.1 Å from catalytic Ser624. In enzyme–inhibitor complexes, a number of interactions occur between the ligand and the enzymes, such as Van der Waals forces, hydrogen bonds and others. The number of Van der Waals contacts between the ligand and the respective enzyme are similar. More hydrogen bonds are formed in the oenothein B–POP complex than in the oenothein B–FAP complex. Nevertheless, the binding energies are close: for POP–oenothein B complex −10.6 kcal/mol, respectively, −9.1 kcal/mol for FAP–oenothein B complex. From the results presented above, we can conclude that the reason for the effective inhibition of POP by oenothein B, unlike FAP, is due not to the energy of enzyme–inhibitor complex formation, but to another factor. As noted, upon binding of the ligand, the active site of POP closes and non-covalent interactions between the enzyme and oenothein B stabilize the “closed” conformation. The active site of FAP is partially open to the aqueous environment. On the other hand, oenothein is significantly hydrophilic (lgP = −3.30) and the polar interactions and hydrogen bonds it can form with water molecules favor the dissociation of the FAP–oenothein B complex [57].

It should be noted that the molecular docking simulations in this study were carried out using the 1-Click Docking tool. This simplified and automated platform is ap-propriate for rapid, preliminary screening of ligand–receptor interactions; however, it has certain methodological constraints. Proteins are dynamic molecular structures that constantly undergo structural changes, including to adapt to ligands (induced fit). 1-Click-docking programs often ignore this flexibility, which can lead to inaccurate predictions of binding. This approach is useful as a preliminary step before more advanced, detailed docking studies or molecular dynamics simulations and experimental confirmation.

The structure of oenothein B (macrocyclic ellagitannin) has limited flexibility of rotational bonds and large physical volume. This implies a relatively large number of contacts between the ligand and the active site of POP, which can be realized at different orientations of oenothein B, ultimately leading to closure of the active site and stabilization of the closed conformation.

The 80% ethanol extract of E. parviflorum was chosen for farther experiments because of its selective inhibitory effect on POP, which presumably contributes to the antitumor activity of the herb.

3.5. Antioxidant Activity of the 80% Ethanol Extract

Metabolic processes in the body are known to generate reactive oxygen species (ROS) which are toxic and, therefore, they are degraded by specialized enzymes [58]. The imbalance between generation and disposal of ROS may cause cellular dysfunction, degenerative processes and aging, along with different diseases, including cancer [59]. The excessive ROS can also be eliminated by exogenous antioxidants of plant origin, such as polyphenols, flavonoids, tannins and other secondary metabolites [60]. To determine the antioxidant properties of a particular plant extract, two types of methods are applied—hydrogen atom transfer (HAT) and single electron transfer (SET) or a combination of them. Although the end results from the two types of methods might be similar, kinetics and possible side reactions are different. We believe that the determination of the antioxidant potential of the extracts would provide additional information on the relationship with the mechanism of their antitumor action. In the study, we used three well-known methods to determine antioxidant capacity, however, based on different mechanisms.

The antioxidant capacity of extracts from Epilobium spp. is mainly associated with the presence of oenothein B [61]. The 80% ethanol extract obtained by us also contains as the main component oenothein B and a small amount of myricitrin. The results of the antioxidant activity of the extract are given in Figure 5. On the abscissa, the concentration of the extract is presented either in mg/mL or in logarithmic units; on the ordinate, the radical scavenging effect is given in percentages (

Table 5. Radical scavenging effect (SC₅₀) of 80% ethanol extract from the aerial parts of E. parviflorum.

Materials	SC50 µg/mL
	DPPH	ABTS	NBT
E. parviflorum extract	27.7 ± 1.71	68.8 ± 2.69	5.7 ± 0.26
Trolox	12.11 ± 0.64	9.67 ± 1.15	23.21 ± 0.38

). The values of SC₅₀ (the extract concentration necessary to scavenge 50% of the radicals) were calculated from the graphs. Troxol was used as a reference substance.

The DPPH test is based on measuring the neutralization of the stable free radical of 1,1-diphenyl-2-picrylhydrazyl (DPPH^•). The antioxidant capacity of E. parviflorum extract was measured in the concentrations range from 0.004 to 0.5 mg/mL. At the lowest concentration, the degree of scavenging of DPPH^• was 29%; at the highest concentration it was 78%. An excellent linear dependence of the radical scavenging rate and the logarithm of the extract concentration was found (R² = 0.9852) and the calculated SC₅₀ was 27.7 ± 1.71 µg/mL.

The ABTS test is based on neutralizing the free radical ABTS^•+ of the compound 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid), generated by potassium persulfate before the start of the test. The radical scavenging activity was examined in concentration range of 0.01 to 0.13 mg/mL. With the lowest tested concentration of the extract, the degree of scavenging of ABTS^•+ was 11%, and the maximal effect of 88% was achieved at 0.13 g/mL. In this case, a perfect linear dependence of the radical scavenging rate and the extract concentration was found (R² = 0.9980) and the calculated SC₅₀ was 68.8 ± 2.69 µg/mL.

The calculated SC₅₀ value at ABTS test is 2.5 times higher than the one obtained from the DPPH test for E. parviflourum extract. The reaction rates of DPPH^• with antioxidants depend on the steric accessibility of the radical site and on the ratio between SET and HAT mechanisms [62]. According to previous studies, kinetic curves of the interaction of ABTS^•+ with certain flavonoids were found to be of an atypical character and the reaction rates differed significantly [63]. Based on these facts, we suggest that the significant difference between the two tests is most likely due to different reaction rates with antioxidants caused by steric effects and the mechanism of the redox process.

The NBT test is based on the ability of antioxidants to react with the superoxide radical anion (dioxide(•1⁻)) O₂^•− generated in the reaction system. The disposal of this radical results in an inhibition of the reduction in NBT (nitro blue tetrazolium chloride), thus decreasing the amount of blue formazan formed. The activity of E. parviflorum extract was tested in concentrations from 0.002 to 0.033 mg/mL. It was demonstrated that at concentration of 0.002 mg/mL, the degree of scavenging effect is 9%, and the maximal effect of 91% was obtained at 0.033 mg/mL. A superb linear dependence of the radical scavenging rate and the logarithm of the extract concentration was found (R² = 0.9805) and the calculated SC₅₀ was 5.7 ± 0.26 µg/mL. Furthermore, the SC₅₀ value for Trolox was determined 23.21 ± 0.38 µg/mL, whereas the corresponding value for the extract was approximately fourfold lower, thereby demonstrating its markedly superior antioxidant activity.

The results obtained from the tests based on the scavenging of DPPH^• and ABTS^•+ are chemically very different from those that can be found in living organisms. These radicals exhibit high steric hindrance around its centered nitrogen atom and so they do not represent a good model for highly reactive radicals generated during the aerobic respiration process in which molecular oxygen is the final electron acceptor. On the other hand, O₂^•− are really generated during those processes in living organisms. Other radicals—HO^•, HOO^• and singlet oxygen—are formed from them, which are the main ROS responsible for the oxidative stress. In this sense, the test is applicable to mimic the antioxidant potential in in vivo conditions. The obtained experimental results show that the extract has a high antioxidant potential, better than the one of Trolox.

3.6. Cell Cycle Analysis

The effect of the 80% ethanol extract on the phases of the cell cycle of HT-29 cells was studied by flow cytometric analysis according to the standard protocol. Cells were treated with concentration of the plant extracts 31.2 μg/mL, which roughly corresponds to the IC₅₀ value as measured in our previous experiments [9]. Moreover, the activity of the extract is comparable to those of the standard cytostatic 5-fluorourcil established in the same cell line in our previous study [64]. The results are presented in Figure 6.

According to the results, cancer cells treated with the extract showed statistically significant changes in the percentage of cells in the individual stages of the cell cycle compared to the non-treated controls. Thus, the 24 h treatment led to a decrease in the number of cells in G1 phase by about 3%, indicating a certain block of the G0/G1 transition. This block was accompanied by an increase in the percentage of cells in S phase by more than 7% and a slight decrease in the cell number in G2/M. The obtained results can be interpreted as an accumulation of cells in S phase, demonstrating a constrained DNA synthesis.

A number of previous studies showed that the antitumor activity of various plant extracts of Epilobium sp. in prostate cancer cells were due to an inhibition of DNA synthesis with an arrest of G0/G1 transition [65]. Additionally, it was shown that the oenotein B-induced blockage at G1 caused a suppression of the growth of human non-small cell lung cancer cells line A549 [66]. According to our results, HT-29 human colorectal cancer cells also showed a certain restriction of DNA synthesis although they were more resistant to the activity of oenothein B in comparison to the other tumor cells mentioned above.

3.7. Apoptosis Assay

To quantify the pro-apoptotic potential of the studied extract (80% ethanol) from the aerial parts of E. parviflorum, apoptosis (early and/or late) and necrosis studies were performed on the HT-29 cell line by staining with fluorescent markers Annexin V-FITC and propidium iodide (PI). The cells were treated with the concentration corresponding to the IC₅₀ value 31.2 µg/mL for 24 h. Results are shown in Figure 7.

From the dot plots and graphs in Figure 7, it is seen that extract treatment causes an increase in early apoptosis about 24.7 ± 1.5% and a slightly lower increase in late apoptosis (2.3 ± 0.8%). The number of necrotic cells is also increased compared to the untreated sample, but these remain low compared to early apoptosis. These findings suggest that the extract exerts its cytotoxic activity primarily through the induction of early apoptosis rather than through nonspecific necrotic cell death.

The pro-apoptotic effects of Epilobium species have been increasingly reported in recent years. For example, it is demonstrated that aqueous extracts from E. angustifolium, E. parviflorum and E. hirsutum induce apoptosis in prostate cancer cells (LNCaP) in a dose-dependent manner [40]. This effect has been associated with mitochondrial membrane depolarization and activation of caspase-3. Similarly, a recent study reports E. parviflorum extract treatment of melanoma cells, which has resulted in mitochondrial depolarization, upregulation of pro-apoptotic proteins, and caspase activation [7]. The authors attribute these effects in part to macrocyclic ellagitannins, particularly oenothein B, which can induce intrinsic apoptosis via reactive oxygen species (ROS) generation.

Since our 80% ethanol extract from E. parviflorum contains oenothein B as a main component, it can be supposed that its significant pro-apoptotic effect on HT-29 cells is also due to mitochondrial depolarization and activation of caspases pathway.

3.8. Comet Assay of DNA Damage Detection

The genotoxic potential of the studied extract against HT-29 cancer cells was investigated using the alkaline comet assay by determining the percentage of DNA that migrated from the head of the comet to the tail. The cells were treated with extract concentrations of 80 μg/mL for 4 and 24 h. The treatment dose was doubled in order to enhance the effect of the extract in inducing DNA breaks to measurable values. As a negative control we used untreated cells, whereas the positive controls were cells treated with the well-known genotoxic agent hydrogen peroxide (H₂O₂) at concentration 150 μM for 3 min. The results are given in Figure 8.

It can be seen that 4 h after the start of treatment, DNA Damage Index increases twice in comparison to the untreated cells and more than three times after 24 h of treatment (Figure 8E). For the positive control, the index was very high under the selected conditions. Obviously, the tested extract showed a substantial genotoxic effect on the HT-29 cells. DNA breaks are visible in Figure 8, which correspond to the increase in the percentage of cells, especially in low and very high DNA damage. This DNA damage causes about 25% of the cells to enter early apoptosis at 24th h after the start of treatment, as determined by the Annexin-V/PI assay (Figure 7). When comparing the results of the two methods (flow cytometry analysis of apoptosis and comet assay), it can be summarized that they confirm the same trend of cell behavior after treatment with the corresponding concentrations of the studied extract. Thus, in both assays, a decrease in the percentage of live cells is observed. Furthermore, the increase in the percentage of cells in early apoptosis, detected by Annexin-V/PI assay, corresponds to the increase in cells with low and medium DNA damage, discovered by the comet assay. Additionally, the percentage of late apoptotic and necrotic cells in both methods rises slowly (2.5% in apoptosis assay and 3% of cells with high and very high DNA damage in comet assay) compared to the untreated cells.

Thus, the results of the two experiments are consistent and lead to the conclusion that the E. parviflorum extract has a pronounced pro-apoptotic effect on the HT-29 cells.

It is important to note that while hydrogen peroxide induces massive DNA damage, the E. parviflorum extract causes a moderate but significant increase, suggesting that it does not inflict indiscriminate cytotoxicity but rather induces controlled genotoxic stress capable of triggering apoptosis without excessive necrosis or inflammation. This selective mechanism is a desirable feature in anticancer drug development.

4. Conclusions

Based on the above results, the following conclusions can be drawn:

The aqueous monophasic extraction of the aerial parts (leaves and twigs) of E. parviflorum leads to high-yield extracts, whereas 80% ethanol efficiently concentrates oenothein B.

The LC-HRMS analyses show the presence of fifty-four compounds in all the obtained extracts, which belong to thirteen groups based on their chemical structure. Forty of those compounds have not been identified so far in E. parviflorum and are probably characteristic only for the herb of Bulgarian origin.

The 80% aqueous ethanol extract of E. parviflorum inhibits POP in a dose-dependent way (IC₅₀ = 1.72 μg/mL). Moreover, the extract does not inhibit FAP at concentrations up to 21 μg/mL, meaning that oenothein B is a highly specific inhibitor of POP. This selectivity can be explained by the high probability of dissociation of the FAP–oenothein B complex as shown using the molecular docking method.

Based on the NBT test, the 80% ethanol extract shows a high ability to scavenge superoxide radical anion O2^•− with an SC₅₀ as low as 5.7 µg/mL. Since O2^•− is the main product in living organisms from which key ROS responsible for the oxidative stress are synthesized, the extract can be used successfully as an antioxidant for preventing severe inflammations and cancer.

The antitumor activity of the 80% ethanol extract on HT-29 colorectal carcinoma cells may include different mechanisms such as inhibition of POP—an enzyme involved in tumor progression and metastases; partial restriction of DNA synthesis (S-block); substantial DNA damage ability leading to cell orientation towards apoptosis.