Characterization of Anti-Inflammatory and Antioxidant Constituents from Scutellaria baicalensis Using LC-MS Coupled with a Bioassay Method

An effective and previously demonstrated screening method for active constituents in natural products using LC-MS coupled with a bioassay was reported in our earlier studies. With this, the current investigation attempted to identify bioactive constituents of Scutellaria baicalensis through LC-MS coupled with a bioassay. Peaks at broadly 17–20 and 24–25 min on the MS chromatogram displayed an inhibitory effect on NO production in lipopolysaccharide-induced BV2 microglia cells. Similarly, peaks at roughly 17–19 and 22 min showed antioxidant activity with an 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS)/2,2-diphenyl-1- picrylhydrazyl (DPPH) assay. For confirmation of LC-MS coupled with a bioassay, nine compounds (1–9) were isolated from an MeOH extract of S. baicalensis. As we predicted, compounds 1, 8, and 9 significantly reduced lipopolysaccharide (LPS)-induced NO production in BV2 cells. Likewise, compounds 5, 6, and 8 exhibited free radical-scavenging activities with the ABTS/DPPH assay. In addition, the structural similarity of the main components was confirmed by analyzing the total extract and EtOAc fractions through molecular networking. Overall, the results suggest that the method comprised of LC-MS coupled with a bioassay can effectively predict active compounds without an isolation process, and the results of molecular networking predicted that other components around the active compound node may also be active.


Introduction
Natural products, which have a variety of chemical structures and are produced by diverse organisms (microorganisms, plants, animals, and humans), possess potential therapeutic properties and are fascinating drug leads [1]. The isolation of active components from natural products is normally achieved by bioassay-guided purification methodologies, but this often results in higher rates of unnecessary reisolation and dissipation of bioactive compounds from repeated processes [1,2].
Recently, liquid chromatography-mass spectrometry (LC-MS) or nuclear magnetic resonance (LC-NMR) methods were applied as dereplication tools to determine known compounds before beginning any isolation step [2]. Dereplication is an essential stage in the development of natural

LC-Quadrupole Time of Flight (QTOF) MS/MS Coupled with Bioassay
For LC-QTOF MS/MS coupled with bioassay, the first phase was to obtain the chemical profile of the MeOH extract of S. baicalensis ( Figure 1A,B, Figure S1 and Table 1), and the second phase was to collect the eluent through the column for 30 s per well in a 96-well plate. The collected sample was used for LC-MS coupled with a bioassay for NO production along with an ABTS/DPPH-determined free radical-scavenging activity assay ( Figure 1C-E). The results of the inhibition of NO production indicated that the constituents at 17-20 and 24-25 min on the MS chromatogram featured inhibitory activity. The results of the ABTS and DPPH free radical-scavenging activity assay indicated that the constituents at 17-19 and 22 min on the MS chromatogram showed scavenging activity. Peaks appearing during these times were expected to be active compounds.

Dereplication through Molecular Networking
For dereplication, molecular networking was performed on the MeOH extract and EtOAc fraction of S. baicalensis, which contained most of the isolated compounds ( Figure 3A). Through clusters containing the identified peaks in the chemical profile, structural similarity was confirmed ( Figure 3B). Among many other clusters, molecular network (MN)1 was a cluster in which the overall nodes were included in the EtOAc fraction rather than the MeOH extract. The peaks c, k, and n contained in this MN1 were based on a structure with hydroxyl groups at C-5 and C-7, and peaks c and k were compounds 7 and 4 ( Figure 2), respectively. These structures are flavanones/flavanonols without double bonding in the C ring and are almost similar except for the difference of the presence or absence of a hydroxyl group on C-3. Peaks d and f-h, which occupy the MN2 cluster, all exhibited a value of C 6 H 8 O 6 missing from the molecular weight in the MS/MS fragments, which was the expressed glucuronide group. Therefore, most of the peaks exhibited a flavone structure. Among the nodes of MN3, peaks m-1 and o were compounds with the same molecular formula but different positions of a methoxy group. Peaks m-1 and o contained a value of CH 3 missing from the molecular weight in the MS/MS fragments, confirming that the methoxy group was present and its position was different. Peaks a and b were very similar in structure to sugars linked to chrysin, which is of the flavonoid family, and were confirmed to be in the MN4 cluster. Through these results, it was confirmed that the molecular network was grouped according to structural similarity. When these molecular networks and LC-MS data coupled with a bioassay method of Figure 1 were shown together, a correlation was observed between the antioxidant and anti-inflammatory components. As seen in Figure 1C-E, 17-20 min showed broad NO production-inhibition and ABTS and DPPH radical inhibitory activities, and the peaks d-j were distributed in the electrospray ionization (ESI)-MS chromatogram ( Figure 1A). The peaks d and f-h of MN2, whose structural similarity was verified through molecular networking, were confirmed. Specifically, peak i was in another cluster, but this was a computational result for MS/MS fragments of molecular networking, thereby generating these results. Therefore, it was predicted that the nodes of this MN2 cluster were components with antioxidant and anti-inflammatory activities. Similarly, 24-26 min showed NO production-inhibition activity, expressing a peak m-o in the MS chromatogram. The peaks m and o were distributed in the MN3 cluster, as before. Since the structural similarity of MSMS was confirmed, anti-inflammatory activity was expected from the node of the minor component present in the MN3 cluster.

Plant Material
The root of S. baicalensis was purchased from Dongwoodang Pharmaceutical Corporation (Youngcheon, Republic of Korea) and identified by Dr. Ki Yong Lee, a professor at the College of Pharmacy, Korea University. A voucher specimen (KUP-HD006) was deposited at the Laboratory of Pharmacognosy, College of Pharmacy, Korea University.

LC-QTOF MS/MS Coupled with Bioassay
The LC-QTOF MS/MS coupled with a bioassay method was composed of two phases. The first received the chemical profile of the sample by analysis with LC-QTOF MS/MS. The second phase was based on collecting the eluent through the column for 30 s per well with a 96-well plate. HPLC analysis was carried out a on Shiseido CapCell PAK C18 column particle size 5 µm (150 × 4.6 mm) using Agilent 126 series system. The mobile phase consisted of water with 0.1% formic acid (solvent A) and acetonitrile with 0.1% formic acid (solvent B), which were applied with the following gradient elution: 5% B (0-5 min) and 5-95% B (5-30 min). The first injection was 5 µL of S. baicalensis MeOH extract injected at 1 mg/mL and the second injection was 20 µL of the extract injected at 25 mg/mL. The flow rate was 0.6 mL/min. A UV chromatogram was recorded at 254 nm. In addition, an Agilent 6530 Q-TOF mass spectrometer (Agilent, Santa Clara, USA) was connected to an HPLC system via an electrospray ionization (ESI) interface and ionized in negative mode. The MS scan range was 50-1700 m/z. The MS/MS fragmentation was set at collision energies of 10, 20, and 30 eV. Data acquisition and processing were carried out by Mass Hunter Workstation software LC/MS Data Acquisition for 6530 series QTOF (version B.05.00). The collected sample in the plate was dried by a vacuum oven (JEIO Tech, OV-12) at 40 • C for 12 h to completely remove the mobile phase before assaying.

NO Production Inhibitory Assay and Cell Viability
BV2 mouse microglia cells were obtained from the College of Pharmacy and Research Institute of Pharmaceutical Science, Seoul National University, Seoul, Korea. Cells were maintained in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin at 37 • C in a humidified incubator under 5% CO 2 and 95% air. BV2 cells (2 × 10 5 cells/mL) were seeded in 96-well plates and cultured for 23 h. The dried samples collected through LC/MS were dissolved in 35 µL of DMEM without FBS. After the medium was replaced with serum-free DMEM, the cells were pretreated with samples for 1 h. Next, LPS (100 ng/mL) was added and the cells were incubated for 24 h. A total of 100 µL of supernatant in each well was mixed with 100 µL of Griess reagent (1% sulfanilamide, 0.1% naphthylethylenediamine dihydrochloride, 2% phosphoric acid). After incubation for 10 min at room temperature, the absorbance at 550 nm was measured using a spectrophotometer. NO production inhibitory activity for each well was calculated as follows: Inhibition activity (%) = 100 × (OD of LPS treated cultures − OD of LPS sample-treated cultures)/(OD of LPS-treated cultures − OD of control cultures). For cell viability, 10 µL of WST solution was added to 100 µL of the remaining supernatant. The plate was incubated for 2 h and the absorbance at 450 nm was measured. Cell viability for each well was calculated as follows: 100 × (OD of LPS sample-treated cultures)/(OD of control cultures).

DPPH and ABTS Free Radical-Scavenging Assay
Antioxidant activity assay was carried out using a DPPH and ABTS assay. First, the DPPH assay was employed to measure the DPPH free radical-scavenging ability. Solutions of the collected sample through LC/MS were prepared in ethanol at a concentration of 300 µL per well. A total of 10 µL of dissolved sample was added to 190 µL of 15 µM DPPH solution in ethanol. Thereafter, the mixture was incubated in the dark at room temperature. After incubation for 30 min, absorbance at 517 nm was measured using a spectrophotometer. The inhibition of the radical-scavenging activity for each sample was calculated as follows: Inhibition activity (%) = 100 − [(S − S 0 )/(C − C 0 )] × 100, where C and S were the absorbance of the control and inhibitor after 30 min and S 0 and C 0 were the absorbance of the control and inhibitor in ethanol without DPPH solution.
The ABTS free radical was produced by a chemical oxidation reaction with potassium persulfate. A measured amount of 2.5 mM of ABTS was mixed with an equal volume of 2.45 mM potassium persulfate and vortexed for 30 s. Then, 100 µL of the dissolved sample was added to 200 µL of ABTS solution. The mixture was then incubated in the dark at room temperature. After incubation for 10 min, the absorbance at 718 nm was measured with a spectrophotometer. The inhibition of the radical-scavenging activity for each sample was calculated as follows: Inhibition activity (%) = 100 − [(S − S 0 )/(C − C 0 )] × 100, where C and S were the absorbance of the control and inhibitor, respectively, after 30 min, while S 0 and C 0 were the absorbance of the control and inhibitor, respectively, in distilled water without ABTS solution.

Molecular Networking
The total extract and EtOAc fraction were analyzed using molecular networking, a dereplication technique utilizing tandem MS data. Molecular networking was conducted on the Global Natural Products Social Molecular Networking (GNPS) web platform (https://gnps.ucsd.edu). The MS/MS data were converted to mzXML format using MS Convert software (Microsoft, Redmond, USA) and uploaded to GNPS using filezila software. The molecular network was created by uploading the MS/MS fragmentation similarity calculation under the following parameters: The precursor ion mass tolerance was m/z 0.02 Da, the fragment ion mass tolerance was m/z 0.02 Da, the minimum cosine score was 0.7, the minimum matched fragment ion was 4, the minimum cluster size was 2, the generated molecular network was schematically illustrated using Cytoscape 3.6.0 software, and dereplication was performed using MS/MS patterns of each cluster. The relevant molecular networking results can be accessed via the following link: https://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=521d25bff0bb481e94326c5225c91b5f

Conclusions
In conclusion, this study applied an LC-MS coupled with a bioassay method to target anti-inflammatory and antioxidant components from S. baicalensis more easily than previous bioassay-guided purification methods. In addition, structural similarity was confirmed with molecular networking as the dereplication identifier for the main constituents of S. baicalensis. Thus, the targeted components exhibited strong activity compared to other compounds. A correlation was shown between the components showing antioxidant and anti-inflammatory activity through LC-MS coupled with a bioassay method and molecular networking. Other nodes around the peak with demonstrated activity according to structural similarity were confirmed through molecular networking, showing the possibility of applying to other natural products. Therefore, these findings demonstrate that there is a foundation of future investigations in targeting and identifying bioactive components more easily and quickly using LC-MS coupled with a bioassay and molecular networking as part of dereplication.