Antibacterial compounds against fish pathogenic bacteria from a combined extract of Angelica gigas and Artemisia iwayomogi and their quantitative analyses

The roots of Angelica gigas Nakai and aerial parts of Artemisia iwayomogi Kitamura were purchased in Kyungdong local market (Seoul, Korea). Both plant materials were identified in our own lab through comparison with reference medicinal plant materials provided by the Ministry of Food and Drug Safety of Korea. The plant materials have been verified by verification method of NHMI (National Herbal Medicine Information, https://nifds.go.kr), and detailed methods and results are described in Supplementary Material, 1. Verification of two medicinal plants.

First grade solvents (Daejung Chemicals & Metals, Siheung, Korea) were used for extraction, fractionation, and isolation. HPLC-grade solvents were purchased from Honeywell B & J (Morristown, NJ, USA). Sephadex^® LH-20 was purchased from GE Healthcare (Chicago, IL, USA). Silica gel column (Si 60) was purchased from Biotage^® (Uppsala, Sweden) and commercially prepared TLC plates (Silica gel 60, F₂₅₄, Merck, NJ, USA) were used. Brain Heart Infusion Agar (BHIA) and Broth (BHIB) were bought from Difco (Becton, NJ, USA).

In order to antibacterial-guided fractionation, dried roots of A. gigas and aerial parts of Ar. iwayomogi (670 g as a 1:1 mixture) were finely ground and extracted three times with 50% ethanol (13.4 L) at 40°C for 3 h using ultrasonic apparatus (Powersonic 420, Hwashin Tech., Gwangju-si, Gyeonggi-do, Korea). The extract was then filtered through a Whatman No. 4 filter paper (Maidstone, UK). The 50% ethanolic extract (CE) was evaporated to dryness under vacuum, yielding 225 g extract. This extract was then suspended in water and partitioned successively with n-hexane (Hex, 29.7 g), methylene chloride (MC, 12.9 g), ethyl acetate (EA, 7.7 g), water saturated n-butanol (W-S Bu, 40.3 g) and aqueous (126 g). These fractions were then tested for their antibacterial activity. Since the MC fraction presented the highest antibacterial activities among the five fractions (Table 1), active compounds were isolated from the MC fraction through various chromatographic techniques.

Silica gel chromatography of the MC fraction of CE (12.5 g) with Hex-acetone mixture (84:16 to 0:100) as an eluent afforded six fractions (MC-I to MC-VI). Among the six fractions, MC-II (662.9 mg) was subjected to silica gel chromatography with Hex-acetone mixture (84:16 to 0:100) and yielded four sub-fractions (MC-II_fr.1 to 4). MC-II_fr.2 (115.8 mg) was subjected to medium pressure liquid chromatography (MPLC) using silica gel chromatography with chloroform-methanol mixture (99:1 to 0:100), yielding 67 fractions. Among 67 fractions, 36 to 40 fractions were combined to give compound 3 (7.5 mg). Fractions 52 to 59 (88.4 mg) were further subjected to Sephadex^® LH-20 column chromatography eluted with methanol, yielding 130 fractions. Among 130 fractions, fractions 69 to 77 (15 mg) afforded compound 4 (9.2 mg) through semi-preparative high-performance liquid chromatography (S-4 µm, 8 nm, 250 × 10 mm; J’sphere ODS-H80, YMC, Kyoto, Japan). MC-II_fr.3 and 4 were combined, and the combined fraction (431.5 mg) was subjected to Sephadex^® LH-20 column chromatography using methanol-acetone mixture (4:6). Among the resulting 130 fractions, fractions 48 to 65 (176.1 mg) were subjected to Sephadex^® LH-20 column chromatography eluted with methanol, yielding compound 5 (13.2 mg).

Dried roots of A. gigas (40 g) were extracted four times with methanol (0.2 L) in an ultrasonic apparatus at 40°C for 2 h. After solvent removal under reduced pressure, the methanolic extract yielded was 12.1 g. The methanolic extract was suspended in water and partitioned successively with MC. The MC fraction was evaporated to dryness under vacuum and yielded 2.8 g.

Silica gel chromatography by MPLC (Isolera One, Biotage^®, Uppsala, Sweden) of the MC fraction of A. gigas (2.8 g) with a mixture of n-hexane (Hex)-acetone (95:5 to 0:100) as an eluent, and afforded four fractions (Fr-I to IV). Fr-II (81.2 mg) was subjected to Sephadex^® LH-20 column chromatography with methanol and yielded pure compound 1 (44 mg, purity 99.0%). Fr-III (195.4 mg) was subjected to silica gel chromatography with Hex-MC (50:50 to 0:100) and yielded six sub-fractions (fr-I to fr-VI). Among the six sub-fractions, fr-IV (144.2 mg) was subjected to Sephadex^® LH-20 column chromatography with methanol and yielded pure compound 2 (51.2 mg, purity 98.7%). The purities of standards were over 98% as determined by HPLC-UV with a wavelength of 210 nm (Supplementary Material, Fig. S1, tables and figures marked with S are found in the Supplementary Material). All compounds were identified using spectral data, such as HPLC-MSⁿ fragment pattern (HPLC: Surveyor MS Pump; Autosampler: Surveyor autosampler; MS: ThermoFinnigan LCQ Advantage MAX ion trap mass spectrometer, MA, USA), proton nuclear magnetic resonance (¹H-NMR), and carbon-13 nuclear magnetic resonance (¹³C-NMR) by direct comparison with published data of spectra and structures. NMR spectra (Figs. S2–S6) were recorded in chloroform-d, methanol-d₄, and dimethyl sulfoxide-d₆ using 400 MHz NMR spectrometer (AVANCE 400 FT-NMR, Bruker, MA, USA). All isolated compounds were screened for antibacterial activity.

Streptococcus iniae KCTC 3657, Vibrio anguillarum KCTC 2711, and Edwardsiella tarda KCTC 12267 were purchased from Korean Collection for Type Cultures (Daejeon, Korea). For antibacterial susceptibility test using microdilution method (Kang et al., 2008; Langfield et al., 2004), bacterial colonies taken directly from BHIA plates were incubated in BHIB at 28°C for 24 h. From this culture, a suspension equivalent to 0.5 McFarland standard in BHIB was prepared.

The fractions or compounds and an equal volume of bacteria at 1×10⁶ CFU/mL were mixed in a 96-well plate and incubated at 28°C for 24 h. The antibiotics amoxicillin and oxytetracycline were used as positive controls. The lowest concentration of antibiotics that visibly inhibited bacterial growth was considered as MIC. Each assay was repeated in three times.

Based on the standard deviation of the response (σ) and the slope (S) from calibration curve using Microsoft Excel, LOD and LOQ were calculated by following equations: LOD = 3.3 × σ/S and LOQ = LOD × 3.

To isolate the compounds responsible for the antibacterial activity of CE as active markers, an activity-guided fractionation strategy was performed throughout the separation procedure. The MC fraction of CE showed the strongest antibacterial activity against S. iniae and V. anguillarum (Table 1). Activity-guided fractionation of the active MC fraction yielded the sub-fractions MC-I and MC-II, which showed strong antibacterial activity against S. iniae and V. anguillarum with MIC values of 156.25 µg/mL and 78.13 to 312.5 µg/mL, respectively (Table 1). The major compounds of the most active sub-fractions (MC-I and II) were identified through LC-ion trap MS (Analytical conditions are in Supplementary Material, 2. Analytical conditions of LC-ion trap MS). LC-MSn analysis suggested that major compounds with the same characteristics as those of decursinol angelate (1) and decursin (2) were present in MC-I (Fig. 1 and Figs. S7–S9), and 1, 2, xanthotoxin (3), and 2,4-dihydroxy-6-methoxyacetophenone (5) were present in MC-II (Figs. S10–S13). Because 1 and 2 were identified as major compounds of the active fractions MC-I and II, they were separated from roots of A. gigas for massive and efficient isolation of two compounds. After isolation, 1 and 2 were identified as decursinol angelate and decursin, respectively, by analyses of the spectral data using ¹H-NMR (Yoo et al., 2008) and ESI-MS ([M+H]⁺ at m/z 329). In addition, three compounds were isolated from MC-II. Compounds 3, 4, and 5 were identified as xanthotoxin (Muller et al., 2004), demethylsuberosin (Masuda et al., 1998), and 2,4-dihydroxy-6-methoxyacetophenone (Brown, 1992), respectively, based on ¹H-NMR and ESI-MS data ([M+H]⁺ at m/z 217 of 3, [M+H]⁺ at m/z 231 of 4 and [M-H]^- at m/z 181 of 5). Of the five compounds, 1–4 are representative compounds reported in Angelica species, and 5 is reported to be present in Artemisia species. In the present study, only one of the compounds derived from Ar. iwayomogi was isolated in the active fractions. Essential oils of Ar. iwayomogi are known to have inhibitory effects on bacterial growth (Cha, 2007), and these essential oils are obtained by extraction with a highly polar solvent or distillation (Cha, 2007; Choi et al., 2017). Therefore, one of possible explanations for failure to isolate these essential oils with antibacterial activity is that essential oils are highly volatile, whereas CE was extracted with low-polarity solvents. This phenomenon coincided with a previous report that recovery of essential oils from plants is low at ethanol concentrations below 55% (Durling et al., 2007).

Fig. 1. Mass spectra and fragmentation patterns of the active compounds in CE. (A) ESI (positive) mass spectra of active compounds. (B) MS/MS fragmentation patterns of active compounds. CE, combined extract.

Download Original Figure

The five compounds were investigated for antibacterial activities against S. iniae, V. anguillarum, and E. tarda (Table 1). To the best of our knowledge, this is the first report on antibacterial activities of these five compounds against fish pathogenic bacteria. The isolated compounds showed MIC values of 31.25 to 1,000 µg/mL, and compounds 1-4 showed strong antibacterial activities against the test strains. In contrast, compound 5 was less active (500–1,000 µg/mL) than the other compounds. The positive controls amoxicillin and oxytetracycline showed MIC ranges of 0.0156 to 1 µg/mL and 0.156 to 0.625 µg/mL against the test strains, respectively. Similar susceptibilities of several terrestrial bacteria to the isolated compounds have been reported in other studies. According to previous studies (Lee et al., 2003c; Taechowisan et al., 2013), 2 exhibited antibacterial activity against Bacillus subtilis with MIC values of 12.5 and 64 µg/mL, and 1 has an MIC of 50 µg/mL. In addition, 1 and 2 possess strong antibacterial activities with MIC values of 6–200 µg/mL against Helicobater pylori (Bae et al., 1998). Compound 3 had MICs at a range of 500–2,000 µg/mL against various gram-positive and -negative strains (de Souza et al., 2005). Compound 5 at 800 µg/mL has previously been shown to exert a slight inhibitory activity against Clostridium perfringens (Ivarsen et al., 2014). This evidence suggest that major compounds of CE possess antibacterial activities against both gram-negative and -positive bacteria. In the present study, among the five compounds, 1 and 2 showed the most potent, broad-spectrum activities against S. iniae (62.5 and 250 µg/mL, respectively) and V. anguillarum (62.5 and 125 µg/mL, respectively). For this reason, 1 and 2 were selected as active marker compounds of CE and then quantified by HPLC-MS/MS.

A. gigas contains dihydropyranocoumarins, 1 and 2 as major compounds, and these compounds are structural isomers having the same formula (Fig. 2). To separate, detect, and quantify these two peaks, LC-UV methods using buffer elution have been reported (Ahn et al., 2008; Kim et al., 2009; Lee et al., 2003a). However, considering the complicated chemical composition of combined plant extracts, HPLC-MS/MS method may have advantages. Although previous studies performed quantification using HPLC-MS/MS (Park et al., 2013), direct comparisons between their and our results are inappropriate because they did not separate the two peaks. Recently, one study has reported separation of 1 and 2 by LC-MS/MS using columns of various sizes, but it was not possible to separate two isomeric peaks with a single column (Kim et al., 2018). In the present study, three different C₁₈ columns (50, 150, and 250 × 4.6 mm) were tested for the separation of 1 and 2. Among them, 250 × 4.6 mm column showed to be the most suitable and it provided good peak separation (Fig. 3). In contrast, other column showed poor peak separations at the same condition (Fig. S14). Therefore, 1 and 2 were quantified under optimized LC conditions using 250 × 4.6 mm column.

Fig. 2. Chemical structures of the compounds isolated from Angelica gigas and Artemisia iwayomogi.

Download Original Figure

Fig. 3. Representative MRM chromatograms of CE and its active compounds. (A) Chromatogram of blank solution. (B) Chromatogram of the active compounds in CE. (C) and (D) Chromatograms of standard compounds. CE, combined extract.

Download Original Figure

Method validation included tests of linearity, precision, accuracy, and specificity. Calibration curves for HPLC-MS/MS analyses were linear over a concentration range of 5 to 1,000 ng/mL (regression line: y = 395.39x – 193.52 with a correlation coefficient r² of 0.9999) for 1, and 5 to 2,000 ng/mL (regression line: y = 370.47x – 606.95 with a correlation coefficient r² of 0.9999) for 2. On the basis of the calibration curves, LOD and LOQ were calculated to be 0.72 and 2.18 ng/mL for 1, and 1.85 and 5.63 ng/mL for 2, respectively (Table 2).

Method reproducibility was evaluated by intra- (n = 3) and inter-day (n = 9) variability in three replicate analyses of sample solutions. Relative standard deviation (RSD) was less than 2%, showing good precision (Table 3).

The recoveries of 1 and 2 were assessed by spiking samples with high, medium, and low concentrations of each compounds (20, 100, and 800 ppb for 1; 30, 200, and 1,333 ppb for 2). Average recoveries ranged from 90.13% to 108.57% (Table 4). All the peaks of 1 and 2 in CE were identified by comparison of retention time, as well as parent and product ions with those of standards in MRM spectra. As a result, high specificity was shown (Fig. S15). Therefore, this method showed suitable precision, accuracy, reproducibility, and specificity.

Content analysis of 1 and 2 in CE was performed by HPLC-MS/MS (Fig. 1). Average contents of 1 and 2 were calculated to be 3.68 ± 0.07% and 6.14 ± 0.04%, respectively (Table 5). Considering the MICs (62.5 and 250 µg/mL) and contents of the two active compounds (sum of the two compounds: 9.82%), the antibacterial activity of CE with an MIC of 2,500 µg/mL can be explained.