The dried plant material was purchased in a phytopharmacy shop in Split (Croatia): Calendula officinalis L. (flower) and aerial parts of Chelidonium majus L., Teucrium chamaedrys L. and Alchemilla vulgaris L. The plant material was identified in the Department of Biology, at the Faculty of Science (Zagreb, Croatia). The procedure for preparing the aqueous infusion was as follows: 15 g of the plant material was infused in 200 mL of boiling distilled water for 30 min with occasional stirring, filtered through Whatman paper No. 1 and concentrated to dryness under vacuum using a rotary evaporator (Rotavapor® R-200; Büchi, Flawil, Switzerland). The obtained residue was dissolved in distilled water to a final concentration of 60 mg/mL. The prepared samples were stored at –20 °C until analysis.

Total phenolic content (TPC) was determined using the Folin-Ciocalteu reagent and the results were expressed in milligram of gallic acid equivalents (GAE) per L (20). Briefly, this method is based on the capacity of Folin-Ciocalteu reagent (a phosphomolybdo-tungsten hetero acid) to react with phenols and generate intense blue colour. The absorbance was recorded by Perkin-Elmer Lambda EZ 201 (Waltham, MA, USA) at 756 nm.

The Agilent 1100 Series (Agilent Technologies, Waldbronn, Germany) system was used for reversed phase-high performance liquid chromatography (RP-HPLC) analyses, equipped with the guard column Zorbax Rx-C18 (4.6 mm×12.5 mm, d(particle)=5 μm) and a non-polar column Poroshell 120 SB C-18 (4.6 mm×75 mm, d(particle)=2.7 μm). The solvents used (acetonitrile, methanol and glacial acetic acid purchased from Sigma-Aldrich GmbH, Merck (Taufkirchen, Germany) were the same as in our previous study (21). The profile of solvents A and B was: 100 % A and 0 % B for 0 min, 91 % A and 9 % B for 6 min, 63 % A and 37 % B for 20 min, 0 % A and 100 % B for 22 min, 0 % A and 100 % B for 27 min, and 100 % A and 0 % B for 30 min. A volume of 5 µL of each extract was injected at a flow rate of 1.0 mL/min. The column temperature was 30 °C and the absorbance was measured at λ=360 nm. The components were identified based on their retention times, UV spectra compared to commercial standards, and co-injections with standards. For quantification purposes, eight known concentrations of the mixed standard solution were made and calibration curves were calculated. The calibration curves are shown in Table S1 along with their R² values. Results are expressed in mg/kg of plant dry mass with standard deviation (S.D.).

In order to analyse flavonoid aglycones, each extract was hydrolysed using 1.2 M HCl. The solutions were centrifuged three times using Hettich® Universal 320/320R centrifuge (Merck KGaA, Darmstadt, Germany) at 18 928×g for 5 min and the supernatants were stored at −20 °C until analysis.

The antioxidant activity of selected aqueous plant infusions was evaluated using three methods: DPPH, FRAP and Rancimat assay. The DPPH scavenging capacity of the samples was measured according to a previously described method (22). A volume of 50 mL of aqueous infusions of different concentrations (from 1 to 10 g/L) was placed in a cuvette with 1 mL of 6·10^–5 M ethanolic solution of DPPH. The decrease in absorbance at 517 nm was determined by Perkin-Elmer Lambda EZ 201. The obtained results are expressed as IC₅₀ values. The reducing potential (FRAP method) of the aqueous plant infusions was measured according to a previously described method (23). This method is based on the reduction of colourless ferric complex (Fe³⁺ tripyridyltriazine) to blue-coloured ferrous complex (Fe²⁺ tripyridyltriazine) by the action of electron-donating antioxidants at low pH. The reduction was monitored by measuring the change of absorbance at 593 nm by Perkin-Elmer Lambda EZ 201.

The effect of the infusions on the oxidative stability of extra virgin olive oil (Rancimat assay) was measured according to the described method and the results are expressed as relative protection factor (RPF) (24). Briefly, the induction period of the oil with and without the addition of plant infusions was determined with the Rancimat model 743 (Methrom, Herisau, Switzerland) at 120 °C and the airflow of 20 L/h. Conductivity was measured conductometrically as a function of the time.

Breast cancer cells MDA-MD-231, urinary bladder cancer cells T24 and lung cancer cells A549 were purchased from ATCC (LGC Standards, Teddington, UK) and cultured for a week in a humidified atmosphere with 5 % CO₂ at 37 °C in a Dulbecco's modified Eagle's medium (DMEM, Euroclone, Milan, Italy) containing 4.5 mg/mL glucose, 10 % foetal bovine serum (FBS) and 1 % antibiotics (penicillin and streptomycin; Euroclone).

Cells were resuspended in a diluted trypan blue solution (Thermo Fisher Scientific, Waltham, MA, USA) and counted with an inverted binocular microscope (AE30; Motic, Barcelona, Spain) using the Neubauer chamber. The cell count was calculated using the following equation:

Cells were then plated in 96-well plates at a density of 11 000 cell/well and incubated overnight. The cells were treated with the samples of selected aqueous plant infusions at concentrations of 0.1, 0.25, 0.5 and 1 mg/mL in triplicate for 4, 24, 48 and 72 h, respectively. Then, the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay was performed in such a manner that after the treatment with extracts, the cells were incubated with MTT at 0.5 g/L and 37 °C for 2 h. The medium was then removed and dimethylsulphoxide (DMSO, Sigma-Aldrich, Merck, Burlington, MA, USA) was added and incubated for a further 10 min at 37 °C with shaking. The degree of formazan formation, an indicator of living and metabolically active cells, was measured spectrophotometrically at λ=570 nm (HiPo MPP-96; Biosan, Riga, Latvia). The data were calculated in relation to the untreated control (100 %).

Statistical methods were used to interpret the results and draw conclusions from the experimental data, considering both univariate and multivariate approaches (25). We used the Statistica v. 13.3.1 programme (26). All experiments were performed in triplicate to ensure the reliability and reproducibility of the results. A one-way analysis of variance (ANOVA) was done to compare the mean values of the different samples. Duncan's new multiple range test (DNMRT) was used for post-hoc analysis, i.e. to determine which groups differed significantly from each other after the ANOVA. The results were considered statistically significant if the p-value was less than or equal to 0.05 (p≤0.05). Principal component (PC) analysis was performed to reduce dimensionality and visualise complex data sets, and highlight patterns and relationships in the data. Pearson’s correlation coefficients were calculated to quantify the linear relationship between phytochemical content, antioxidant capacity and species bioactivities.

A new RP-HPLC method was developed to achieve optimal separation of flavonoids in all tested extracts. The results of the analysis of non-hydrolysed infusions compared to the standard are shown in Fig. S1. The chromatogram of the standard flavonoid compounds is shown in Fig. S1a, while the peak designations, calibration curves and R² values are listed in Table S1. The chromatogram of the non-hydrolysed infusion of C. officinalis is shown in Fig. S1b. The predominant peak was identified as isorhamnetin-3-O-rutinoside (4), followed by isorhamnetin-3-O-glucoside (5) and quercetagetin (6). A lower peak was identified as quercetin-3-O-rutinoside (1) and only trace amounts of quercetin-3-β-d-glucoside (2) and kaempferol-3-O-rutinoside (3) were present. Isorhamnetin-3-O-rutinoside was present on dry mass basis at 8.34 mg/kg.

The RP-HPLC chromatogram of the non-hydrolysed infusion of C. majus is shown in Fig. S1c. The second most significant peak was identified as quercetin-3-O-rutinoside (1) on dry mass basis at 2.17 mg/kg, followed by the peak identified as isorhamnetin-3-O-rutinoside (4). Among the identified flavonoid glycosides, quercetin-3-O-rutinoside accounted for 75.1 %. Only small amounts of quercetin-3-β-d-glucoside (2) and kaempferol-3-O-rutinoside (3) were detected. Quercetin was also found as the predominant flavonoid aglycone in a C. majus extract from Poland, whereas isorhamnetin was not identified at all (27). Quercetin-3-O-rutinoside as the predominant compound and isorhamnetin-3-O-rutinoside as the second most abundant flavonol glycoside were detected in extracts of wild and cultivated C. majus (28). Fig. S1d shows the RP-HPLC profile of glycosides from the aqueous infusion of A. vulgaris. The main constituent of A. vulgaris was isorhamnetin-3-O-rutinoside (4). The second most abundant glycoside was kaempferol-3-O-rutinoside (3). The RP-HPLC chromatogram of the T. chamaedrys infusion is shown in Fig. S1e. Only one flavonoid glycoside was identified, namely quercetin-3-β-d-glucoside (2).

Isorhamnetin-3-O-rutinoside accounted for 69.9 % of the total flavonoid glycosides identified in the aqueous infusion of C. officinalis (Fig. 1a). The mass fraction on dry mass basis of the second most abundant glycoside (isorhamnetin-3-O-glucoside) was 0.92 mg/kg of C. officinalis, while for quercetin-3-O-rutinoside, quercetin-3-O-glucoside and aglycone quercetagetin it was 0.83, 0.27 and 1.47 mg/kg, respectively. After the acid hydrolysis, mass fraction of quercetin was the highest among the identified aglycones (Fig. 1a).

Compared to the flavonoid glycosides detected in the 70 % methanolic extract of C. officinalis from Tunisia (29), we found two additional glycosides in the water extract of Croatian C. officinalis – isorhamnetin-3-O-rutinoside (which was predominant) and kaempferol-3-O-rutinoside. The most likely reason for this difference is the use of different extraction solvents and a completely different collection area. In addition, plants can be collected and processed in different ways and in different forms. This is because the analysed samples were raw materials collected in different areas and non-standardised purchased products. Interestingly, ethanolic extracts of C. officinalis inflorescences from France, Egypt and Russia contained quercetin and isorhamnetin glycosides, but no kaempferol glycosides were detected (30-32). So far, the presence of kaempferol-3-O-glucoside and kaempferol has been detected in C. officinalis, but we have not found kaempferol-3-O-rutinoside in Croatian species (11). Therefore, to the best of our knowledge, this is the first report of the detection of kaempferol-3-O-rutinoside in the aqueous infusion of C. officinalis. In C. majus, quercetin-3-O-rutinoside, quercetin-3-β-d-glucoside, kaempferol-3-O-rutinoside and isorhamnetin-3-O-rutinoside were tentatively identified (Fig. 1b). The main constituent of T. chamaedrys was quercetin glucoside (Fig. 1c), while in A. vulgaris it was isorhamnetin-3-O-rutinoside, which accounted for 89.1 % of the identified flavonoid glycosides (Fig. 1d).

Acid hydrolysis of the aqueous infusion of C. officinalis released free aglycones (Fig. S2a). Although isorhamnetin (9) showed the highest peak in the chromatogram (Fig. S2a), among the aglycones in the hydrolysed extract, the mass fraction of quercetin (7) was the highest (Fig. 1a). After acid hydrolysis of C. majus, the peaks of free quercetin (7), kaempferol (8) and isorhamnetin (9) were detected (Fig. S2b). Quercetin glycoside was present at the highest mass fraction (Fig. 1b). As in the non-hydrolysed infusion of C. officinalis (Fig. S1b), the non-hydrolysed extract of A. vulgaris also contained the aglycone quercetagetin (6) (Fig. S1d). Interestingly, after acid hydrolysis, quercetin (7) was found to be the predominant component (Fig. S2c), suggesting that one or more of the glycosylated peaks in this extract correspond to quercetin glycosides, although they were not identified. Many quercetin glycosides have been found in various Alchemilla species from Turkey, which also supports our conclusion (33, 34). In addition, the flavonoids of Alchemilla species are usually quercetin derivatives (34). Our finding of isorhamnetin-3-O-rutinoside is, as far as we know, the first report of this flavonoid in A. vulgaris. After acid hydrolysis of T. chamaedrys, lower peaks corresponding to free quercetin (7), kaempferol (8) and isorhamnetin (9) were detected (Fig. S2d). This indicates that kaempferol and isorhamnetin glycoside forms were present in the non-hydrolysed extract, which differ from those available to us as standards.

After summarising all identified glycosides in non-hydrolysed extracts, their mass fraction was the highest in C. officinalis (Fig. S3a), while after hydrolysis the highest mass fraction of identified aglycones was found in A. vulgaris (Fig. S3b). According to the available data, there is no information so far on the presence of phenolic compounds in the aqueous infusion of T. chamaedrys.

Table 1 shows the total phenolic content and antioxidant activity of aqueous infusions of selected medicinal plants. The highest total phenolic content (expressed as GAE) was found in the infusion of T. chamaedrys ((2061±42) mg/L), while the infusion of C. officinalis had the lowest total phenolic content ((638±35) mg/L), which was 30 % lower than that of T. chamaedrys. The results are in agreement with those of Vlase et al. (35), who reported that the extract of T. chamaedrys contained the highest amount of flavonoid compounds compared to the extracts of Hyssopus officinalis and Ocimum basilicum. The highest free radical scavenging activity was found in A. vulgaris infusion (IC₅₀=0.8 mg/mL) using DPPH assay and in T. chamaedrys infusion (9798 mg/L, expressed as FeCl₂) using FRAP assay (Table 1). C. officinalis had the lowest antioxidant activity and the lowest concentration of total phenolics in all used methods. High antioxidant activity and lipid inhibition potential of the methanolic extract of A. vulgaris have been reported (17, 18). Özer (36) demonstrated a high antioxidant potential of the infusion and decoction of T. chamaedrys using DPPH and Cu(II) reducing antioxidant capacity (CUPRAC) assays. The ethanol extract of T. chamaedrys also showed a significant free radical scavenging activity (35). According to the available data, there were no reports so far on the influence of extracts of plants used in this study on the oxidative stability of edible oils measured by the Rancimat method. The protection factor (PF) of C. majus was significantly higher (1.19±0.03) than that of the other tested infusions (Table 1) and was similar to that of α-tocopherol, a commercially available antioxidant (37), so it was concluded that it could be used to preserve the oxidative stability of edible oils. The ability of medicinal plants to prolong the oxidative stability and shelf life of foods containing oil is important for the food industry and pharmaceutical sector. However, herbs and plants can be processed using different techniques and ingested in different ways and forms. Although this aspect was not the focus of this study, it may be interesting to conduct a comparative study on the use of different processing techniques of selected medicinal herbs to protect their biological activity.

The antiproliferative activity of the infusions of the selected medicinal plants tested against the cancer cell lines MD-MBA-231, T24 and A549 is shown in Table 2, Table 3 and Table 4 respectively in relation to the dose-dependent effect (1, 0.5, 0.25 and 0.1 mg/mL) and different incubation times (4, 24, 48 and 72 h). At a concentration of 1 mg/mL, A. vulgaris showed the highest inhibition of MDA-MD-231 cell proliferation at each time point (Table 2). At the same concentration, the inhibition rate of cancer cell proliferation by infusion of A. vulgaris was as high as 75 % after 72 h. At lower concentrations, A. vulgaris had the highest inhibition rate after 4 h, at later time points it was C. majus. The inhibition rate of T. chamaedrys infusion was below the detection limit at 0.50 mg/mL. The aqueous infusion of C. officinalis had a low antiproliferative effect against MDA-MD-231 cells (Table 2). C. majus aqueous infusion showed a high and dose-dependent antiproliferative effect against MDA-MD-231 cells (Table 2), which is consistent with other studies (38, 39). According to the available data, this is the first comparative report on the antiproliferative effect of A. vulgaris and C. majus infusions against MDA-MD-231 cells.

The antiproliferative activity against T24 cancer cells was the highest when using C. majus infusion (Table 3). At a concentration of 1 mg/mL, the antiproliferative activity of C. majus was as high as 96 % after 72 h of incubation. The antiproliferative activity of T. chamaedrys was already low at a concentration of 1 mg/mL, and at lower concentrations it was below the detection limit (data not shown). To the best of our knowledge, this is the first study on the antiproliferative activity of selected plant infusions against T24 cancer cells.

The antiproliferative activity of the infusions was lower in cancer cells A549 than in cancer cells MDA-MB-231 and T24 (Table 4). The antiproliferative activity of 1 mg/mL of C. officinalis and T. chamaedrys against A549 cells after 72 h incubation was low, while the activity of C. majus and A. vulgaris was moderate (51 and 46 %, respectively). The alcoholic extract of T. chamaedrys did not show any antiproliferative activity against A549 cells (40). According to the available data (38), the antiproliferative activity of the extract of C. majus against A549 cells was moderate, which is consistent with our results. To the best of our knowledge, there are no previous data on the comparative report of the antiproliferative activity of T. chamaedrys and A. vulgaris against A549 cells.

Based on the principal component (PC) analysis of the measured parameters, T. chamaedrys and C. majus were grouped together, meaning these were the most similar infusions, while A. vulgaris and C. officinalis were more distant from this group as well as from each other (Fig. 2a). The loading plot of the measured variables showed that quercetin glucoside and antioxidant capacity measured by the DPPH method contributed most to the separation of these infusions from the two others (Fig. 2b). The main contributors to the separation of A. vulgaris were quercetin, kaempferol, kaempferol-3-O-rutinoside, total identified aglycones and the rate of inhibition of proliferation of MD-MBA-231. The separation of C. officinalis was mainly due to the total identified glucosides, isorhamnetin, isorhamnetin-3-O-rutinoside and isorhamnetin-3-O-glucoside (Fig. 2b).

Hierarchical clustering shows the relationships between different data sets and indicates the degree of similarity/dissimilarity between samples. Based on the total amount and individual phenolics identified, antioxidant capacity and cytotoxicity, C. officinalis and C. majus were the least distant from each other and formed a cluster (Fig. 3). T. chamaedrys and A. vulgaris formed another cluster, but were more distant from each other than C. officinalis and C. majus.

According to Evans (41), Pearson’s correlation coefficient (r) between the measured parameters of the samples showed a very strong positive correlation between the antioxidant capacity (FRAP) and total phenolics (Table S2), between the antioxidant capacity measured with the Rancimat method and the inhibition rate of MD-MBA-123 and A549 cells, and between the inhibition rate of MD-MBA-123 and A549 cells.