An activity-integrated strategy of the identification, screening and determination of potential neuraminidase inhibitors from Radix Scutellariae

Small molecules isolated from herbal medicines (HMs) were identified as the potential neuraminidase inhibitors which are effective in influenza prevention and treatment. Unfortunately, current available screen methods of small molecules isolated from HMs are inefficient and insensitive. Here a novel Ultra Performance Liquid Chromatography coupled with diode-array detectors and auto-fraction collector / time-of-flight mass spectrometry (UPLC-DAD-FC/Q-TOF-MS) screening method with high efficiency was developed and validated to separate, collect, enrich, identify and quantify potential neuraminidase inhibitors from Radix Scutellariae. The results showed that 26 components with neuraminidase inhibitory activity were identified from Radix Scutellariae extracts. It was also found that the influence of origins on the quality of RS was more than that of cultivated time on the basis of the concentration of the effective components. These results brought novel insights into quality evaluation of Radix Scutellariae. It was demonstrated that new activity-integrated strategy was a suitable technique for the identification, screening and determination of potential neuraminidase inhibitors in herbal medicine and will provide novel potential strategies in other drug screening from herbal medicine.


Introduction
Influenza is a burgeoning infectious disease which causes severe respiratory illnesses [1]. The viral surface protein neuraminidase (NA) is widely used as a direct target to inhibit viral replication and block human body infection [2], which promotes the hydrolysis of terminal sialic acid residues of newly generated virus and host cell receptors [3]. Influenza neuraminidase inhibitors, such as Oseltamivir, Peramivir, Zanamivir, have been widely used for PLOS  influenza type-A and type-B virus treatment [4,5], but their applications are limited due to drug resistance and side effects including vomiting, nausea, and diarrhea [6]. Therefore, natural neuraminidase inhibitors have attracted increasing attention on the treatment of influenza [7]. It was reported that more than four hundred different kinds of herbal medicines (HMs) has been detected to have high efficiency of NA inhibitory activities [8]. Many of these HMs have been used to treat the Influenza for thousands of years in clinics in China. They are also regarded as an important source of novel pharmacologically active compounds and can be used to identify and purify of natural neuraminidase inhibitors in drug discovery and pharmaceutical research [9]. HMs was the complicated mixture of components [9]. It is therefore hard to screen and identify the effective functional components in HMs. It was very necessary to select some effective components as combinatorial markers for the quality control of these HMs in order to guarantee efficacy and safety. There are many combination of chemical analysis and biological tests methods reported by scholars with different features in order to solvent the quality problem of HMs [11][12][13], which have overcome to some extent the quality problems in HMs and screening active compounds of complex extracts as markers to evaluate quality of HMs. However, there were no chemical and biological analysis combined methods for rapid screening neuraminidase inhibitors from HMs in the literatures. The current bioactivity -guided fractionation and isolation method was mainly used to screen natural neuraminidase inhibitors as markers for quality control of HM. However, it is so time-consuming, laborintensive and expensive that screening the effective functional components was challenge in HMs. Recently, Ultra Performance Liquid Chromatography coupled with diode-array detectors and auto-fraction collector / time-of-flight mass spectrometry (UPLC-DAD-FC/ Q-TOF-MS) coupled with bioactive assays were used to screen and identify anti-diabetic ingredients from HMs in our previous studies [14]. In light of these, UPLC-DAD-FC/ Q-TOF-MS method was here proposed to screen natural neuraminidase inhibitors as combinatorial markers for the quality control of HMs.
Radix Scutellariae (Huangqin, RS), the dried root of Scutellariae baicalensis Georgi, has been used as one of HMs to treat diseases for thousands of years for its anti-febrile and detoxification functions. Pharmacology studies have demonstrated that Radix Scutellariae treatment is effective in anti-inflammatory [10], anti-oxidant [11], anti-cancer [12], anti-angiogenesis [13] and anti-viral activities [1,14]. Photochemistry studies showed that baicalin, wogonoside, baicalein, wogonin and oroxylin A were the major bioactive components in Radix Scutellariae [15]. Although RS is considered a potential HM for screening bioactive components of neuraminidase inhibitors, the current available screening and analysis methods are inefficient and insensitive. The inhibitory effects of baicalin and baicalein on neuramidase have been reported based on immobilized NA micro-reactor [16], but it only could be used to test NA inhibitory activities of ingredients or extracts of TCMs, but not rapidly screen the NA inhibitory activities of each component from the HMs extracts.
To our knowledge, UPLC-DAD-FC/Q-TOF-MS application for screening natural neuraminidase inhibitors from RS has not been reported in the literatures. In this study, UPLC-DAD-FC was used to separate, quantify, collect and enrich more components in the fractions of RS. Then neuraminidase inhibitory activities of all fractions were performed. Finally, the UPLC-DAD-Q-TOF-MS system was used to identify the components in fractions with neuraminidase inhibitory activities from RS (Fig 1). Given the importance of the natural neuraminidase inhibitors, hierarchical cluster analysis was used to distinguish quality control of the different RS samples from different areas basing on their contents. The new activityintegrated strategy of the identification, screening and determination of potential neuraminidase inhibitors in one-step work will become the advantageous tool for screening natural 26-35 min, 35-40% B. The flow rate was set at 0.3 mL min -1 , with a column temperature fixed at 35˚C. The injection volume was 1 μL and detection wavelength was 280 nm.
The fraction collection and enrichment of the active compounds were achieved with autofraction collector (FC) (BSZ-100, Shanghai Qingpu Huxi Instrument, Shanghai, China) and the UPLC system was employed in separation of the extract from RS and quantitative determination of the natural active components. Using a time span between different peaks in the chromatographic profile of the UPLC and a fixed-time interval collection based on the retention time, the various fractions were collected automatically. Accordingly, a time interval of 21 s was selected to collect the active compounds enriched by the UPLC separation. The samples were subsequently dried using nitrogen gas and the residues were dissolved in methanol for the activity assay.

UPLC-Q-TOF-MS system
UPLC-Q-TOF-MS system (Agilent Co.) was used to identify the components in RS samples with the ESI-MS spectra obtained in both the negative and positive ion modes. The conditions of the capillary voltage (CV) of the positive ion mode and negative ion mode were +3.5 kV and -3.5 kV, respectively. The nebulization and auxiliary gas was high-purity nitrogen (>0.9999) with pressure of nebulizer gas of 35 psig and the flow rate of drying gas of 8.0 L min -1 at temperature of 350˚C. Fragmentor voltage (FV) and skimmer voltage (SV) were set at 175 and 65 V, respectively. The collision energy (CE) selected ranged from 15 V to 40 V. Data was collected for MS at MS scan ranging 100-1700 Da and for MS/MS at MS/MS scan ranging 50-1500 Da. The Mass Hunter software (Agilent Technologies) was employed to acquire data including the exact mass and to analyze the elemental compositions. The mobile phase used in UPLC-Q-TOF-MS analysis comprised formic acid aqueous solution (0.1%, v/v) (A phase) and acetonitrile (B phase). The separation condition was identical to that of the UPLC-UV-FC system. Preparation of sample solutions RS sample preparation. 0.1 g of the RS powder was mixed with 10 mL of 50% methanol (methanol-water) in ultrasonic bath for 30 min. The mixture was centrifuged at 14,000 rpm for 10 min and the supernatant was diluted four times for UPLC system analysis or screening assay.

Calibration curve, repeatability, LODs and LOQs
In constructing the calibration curves, each mixed standard solution was determined in triplicate at the various concentrations diluted with that of the standard stock solution. A plot of the peak areas (y) against the corresponding concentration (x, μg mL -1 ) showed the calibration curve of each standard. The repeatability, expressed as the relative standard deviation (RSD), was validated using six independent samples.
The LOD and the LOQ were defined as the lowest concentration for quantification and detection respectively, obtained by diluting the lowest concentration of the mixed standard solution. The LOD and LOQ were calculated at a signal-to-noise (S/N) ratio of 3 and 10, respectively.

Precision and stability
Precision was calculated by measuring the intra-day and inter-day variability and expressed as RSD. A preparation of the stock standard solutions at three different concentrations namely low, medium and high concentrations was analyzed and used to calculate the RSD. The mixed standard solutions were tested for six replicates within one day to estimate the intra-day precision. On the other hand, the inter-day variability assay was carried out by examining the mixed standard solutions in replicates for three days consecutively. At time intervals of 0, 2, 4, 6, 8, 12 and 24 h, the method stability was determined with the low, medium and high concentration of the mixed standard solutions by performing six repetitive injections of the samples. The accuracy ranging ±15% of the primitive concentration and RSD less than 5% were considered to be acceptable.

Recovery
The recovery was tested as follows: spiking 100% quantities of the investigated compounds into a certain amount of the RS sample. Extraction and analysis were performed on the resultant samples using the methods mentioned above. The recovery was calculated by the formula: Recovery (%) = (found amount original amount)/spiked amount × 100%.

Bioassay of neuraminidase inhibition
Following the manufacture's instruction, the NA inhibition experiment was carried out in a 96-well microplate reader [17]. In each well, 70 μL of reaction buffer solution, 10 μL of NA and 10 μL of standard samples dissolved by methanol were mixed together. Mixture was vibrated for 1 min and then incubated at 37˚C for 2 min. After incubation, 10 μL of fluorescent substrate (2 0 -(4-Methylumbelliferyl)-α-D-N-acetylneuraminic acid sodium salt hydrate) was added. The mixture was subsequently vibrated for 1 min and then incubated at 37˚C for 30 min. The fluorescence intensity was measured by a fluorescence multifunctional microplate reader (Tecan GENios, Männedorf, Switzerland) with excitation at 485 nm and emission detection at 530 nm. Water was used in the control group instead of the standard samples.
The inhibition rate (R%) was calculated by a formula: (A1−A2)/A1×100. A1 is the absorbance of the control and A2 is the absorbance of the sample. The IC 50 was determined by plotting the percentage of NA activity against inhibitor concentration. All samples were determined in triplicates.

Cell culture, cell viability and transfection
Normal human primary lung fibroblasts (PLF) (ATCC 1 PCS201013™) were cultured in Fibroblast Basal Medium (ATCC 1 PCS201030™) supplemented with growth factors. HIEC-6 cells were cultured in the complete growth medium according to the protocols as described previously [18,19]. The HEK293T cells were cultured in DMEM medium with 10% FBS. The MTT assay was employed to test cell viability. In brief, cells were seeded into 96-well plates at a density of 1 × 10 5 cells well -1 . 24 hours after treatment of indicated drugs, the culture medium was replaced with 5 mg mL -1 MTT assay solution (Sigma-Aldrich) and incubated for 4 hours. The reaction was terminated in 150 μL anhydrous dimethyl sulfoxide (DMSO, Sigma-Aldrich). The absorbance at 570 nm was measured with the microplate reader. The viability value was calculated as average absorbance of experimental group/average absorbance of control group.
NA plasmid pNAHA was purchased from addgene (Plasmid #44169) [20], and transfection was performed by using Lipofectamine 3000 Reagent (Life Technologies) according to the manufacturer's protocol. HEK293T cells transfected with neuraminidase expression plasmid were grown in 24-well plate and treated by NA activity compound for 24h. Then the whole cell extract was used for NA inhibitory assay as mentioned in Bioassay of neuraminidase inhibition.

Optimization of chromatographic conditions
To improve peak resolution efficiently, UPLC chromatographic conditions were optimized, including mobile phases (acetonitrile-water, methanol-water), concentrations of formic acid (0.1% and 0.2%), column temperatures (30, 35 and 40˚C) and flow set (0.2, 0.3 and 0.4 mL min -1 ). By comparing the shapes of the peaks and the resolutions of the target compounds, the best results were obtained in the conditions with acetonitrile 0.1% formic acid as a mobile phase. The column temperature was 30˚C and flow was set at 0.3 mL min -1 (Fig 2).
The relative standard deviations (RSDs) of intra-day and inter-day precision were 0.17-2.62% and 0.47-2.84%, respectively. The intra-day and inter-day precision for the 9 standards accuracies were 95.0-107% and 92.6-108%. These standards were showed to be stable in 24 h. The relative standard deviations (RSDs) were < 3%. Furthermore, the accuracies ranged of stability was from 94.9% to 107% ( Table 3). The relative standard deviations (RSDs) of all recoveries were < 3% with the accuracy ranging of 98.7-104%, except for oroxylin A. Oroxylin A was not separated from skullcapflavone II by UPLC (Table 4). All these results indicated that the precision and stability of the developed method were reliable for the quantitative analysis of the naturally active components in RS.

NA inhibitory activity integration of chromatography
To integrate the compound concentration and bioactivity from the RS extract, the UPLC-DAD-FC system was used to separate, quantify, collect and enrich the compounds. One RS sample (labeled as S3 in Table 1) was analyzed by this method. 26 peaks of compounds in RS extract were detected by DAD detector. The fraction collector was set up at an interval time of 21s for the purity of most of the fractions. To make sure get enough amounts of components with low concentration in RS extract, enriching the samples of 10, 30, 50, 70 and 100  times were tested. The result shows for 10 times and 30 times enriching collection, some fractions were still undetectable for the NA inhibitory activity. We found that 50 times was the minimal enriching requirement for all fractions bioactivity to be detectable. The total 100 fraction were assayed for NA inhibitory activity, which was reflected with the inhibition rate (%) (Fig 3A). By integrating fraction bioactivity with chromatography analysis data, 26 components in Radix Scutellariae extracts were shown the neuraminidase inhibitory activity.
To searching for compounds with the highest potency of NA inhibitory activity, we integrated the NA inhibitory activity data with the chromatography quantity data. Interestingly, Table 3. Intra-day, inter-day precision and stability of the seven bioactive compounds (n = 6).

Compounds
Concentration (μg mL −1 ) Inter-day Intra-day Stability we found several components were identified to show relative high NA inhibitory activity. For example, the component (3',5,5',7-Tetrahydroxy-2',8-dimethoxyflavone, Peak 9) which is at low concentration detected in RS extract exhibited comparable high NA inhibitory activity. The components with low concentration but high NA inhibitory activity will be further studied in our future work. The quantification of component concentration and their NA inhibitory activity in this method could be finished at one step. The most current methods have to do this work at two individual steps, which takes time of months, but our method only takes 3 days for completion. Another advantage of this method was that a very little amount of herbal RS, such as 1 g, is sufficient for analysis, while several kilo grams of herbal RS is required in most other methods for detectable results and analysis. Moreover, in this method, the total of 26 compounds were identified at one time running, which is much more efficient than other methods that only detected several compounds at one time running [21].
Peak 26 produced the [M−H] − ion at m/z 343. Analyzing their fragments, an inference was made that both of them have three-OCH3 groups from observing the ions with m/z 328, m/z 313 and m/z298. Based on these results and previous knowledge on the flavones in RS, peak 26 was named tenaxin I [24]. All results were shown in Table 4.
This method combines UPLC-UV-FC and UPLC-Q-TOF-MS to assess both quantity and quality of components, as well as the corresponding bioactivity (Fig 3). From this result the highest potent NA inhibitors were identified, such as baicalin, wogonoside, etc.

Verification of the NA inhibitory activity of compounds in Radix Scutellariae
To confirm the activity-integrated method, inhibitory activity assays of the RS extracts and the 9 bioactive standards (purity > 98%) were performed. The IC 50 value of the RS extract was 745.5 ± 35.2 μg mL -1 . For the 9 standards the IC50 values were 321.8 ± 62.8 μg mL -1 for scutellarin, 664.5 ± 26.2 μg mL -1 for scutellarein, 267.3 ± 37.8 μg mL -1 for baicalin, 493.6 ± 64.2 μg mL -1 for chrysin-7-O-glucuronide, 428.3 ± 18.9 μg mL -1 for wogonoside, 397.7 ± 9.91 μg mL -1 for baicalein, 331.1 ± 49.4 μg mL -1 for wogonin, 402.80 ± 67.8 μg mL -1 for chrysin and the more than 1000 μgÁmL -1 for oroxylin A (Fig 4). Specifically, baicalin, baicalein, wogonoside and wogonin showed the high NA inhibitory activity, with the baicalin showing the highest NA inhibitory activity (Fig 4). All these compounds were shown high concentration in RS, and indicated their high potential in pharmacological values. Oroxylin A has the very weak NA inhibitory activity with the IC 50 of more than 1000 μgÁmL -1 , which is different from the result Identification, screening and determination of potential neuraminidase inhibitors of UPLC-UV-FC with oroxylin A showing the relative high NA inhibitory activity (Fig 3). The possible reason is that the skullcapflavone II, which override with oroxylin A in the same peak with mass spectrometry assay, was mixed in one fraction.
To assay the impacts of the components on cellular activity, the cell survival rate was detected by using both primary lung (PLF) and intestine (HIEC-6) cells lines treated with indicated components at corresponding IC 50 concentration. The result showed that baicalin, wogonoside and wogonin did not have significant effect on these cell survivals (Fig 5). To further confirm the bioactivity of the component, we treated HEK293T cells ectopically expressing viral neuraminidase with the indicated component of baicalin, wogonoside and wogonin at conditions of 50, 100, 200 and 500 μg mL -1 . The whole cell lysate was extracted for NA activity assay. Baicalin and wogonin were shown the high inhibition on NA in the cells at all tested concentrations (Fig 6). This result is in agreement with the previous data (Fig 4). Our method identified the major component of flavones in RS extract showing high NA inhibitory activity, suggesting the potential biomedical value of herbal medicine.

Quantitative analysis of the different types of Radix Scutellariae by 26 NA inhibitory activity compounds
To determine the contents of each bioactive component in NA inhibitory of 18 RS samples, the UPLC-DAD was used. There are 18 RS samples which include three group of 1. biennially cultivated RS, 2. perennially (3 years or more) cultivated RS and 3. wild RS. The detailed information of the 18 RS samples was summarized in Table 1. Concentrations of the 26 components in 18 RS samples were shown in Table 6. Seven of active components were quantified with standards, including scutellarin, baicalin, chrysin-7-O-glucuronide, wogonoside. baicalein, wogonin and chrysin. Specifically, the corresponding concentration were 0.44-0.54 μg mg -1 , 244.23-421.18 μg mg -1 , 4.72-10.32 μg mg -1 , 52.91-104.89 μg mg -1 , 8.21-32.23 μg mg -1 and 1.60-2.73 mg mg −1 , respectively. The other 19 components were semi-quantitatively determined by the calibration curve of scutellarin due to lacking the corresponding standard.
The flavones (baicalin and wogonoside) are the most abundant component in RS extract among all the 18 samples with an average concentration of 116.54 mg g -1 . It is also found that the concentration of flavones (baicalin and wogonoside) in perennially (3 years or more) cultivated RS was significantly higher than that in biennially cultivated RS. Baicalin is the major component in total flavones, which reaches up to more than 5% of the gross total weight of RS sample Table 6, So that baicalin was expected to serve as the major contributor to NA inhibitory activity of RS extract.

Hierarchical cluster analysis of the different types of Radix Scutellariae by 26 NA inhibitory activity compounds
Hierarchical cluster analysis (HCA) is defined as a statistical method to analyze and cluster data by measuring the similarity in samples. HCA is preferential to cluster objects with highest similarities into one group. HCA was also applied to distinguish the different HM samples [25]. The 18 RS samples were assayed with HCA according their concentration of 26 compounds screened with UPLC/Q-TOF-MS-FC (Fig 3) HCA dendrogram (Fig 7) showed that 18 RS samples were classified into 4 groups based on concentration and NA inhibitory bioactivity of each sample (Fig 3, Table 6). The first group (group I) contains the most biennially cultivated RS, including S10, S14, S9, S1, S11, S2, and S12. Moreover, the second group (group II) contains the perennially (3 years or more) cultivated RS, inducing S3, S8, S6, S7, S4, S5 and S13. This result demonstrated that component concentration and NA inhibitory bioactivity in RS extract were distinct between biennially cultivated and the perennially (3 years or more) cultivated RS. Group III consisted of the three wild RS (S15, S16 and S17) samples. It implied that the effects of wildness on RS contents correlated with the age of RS. The sample S18, from plateau Area, was clustered into group IV individually. These results suggest that high altitude climate may have a special role on the chemical contents of RS. From HCA methods, it was indicated that the influence of origins on the quality of RS was more than that of cultivated time.

Conclusions
A new integrated UPLC-DAD-FC/UPLC-Q-TOF-MS method has been developed and successfully applied in screening natural neuraminidase inhibitors in RS extract. Using our new developed analysis methods, we found the main neuraminidase inhibitors in RS extract were baicalin, baicalein, wogonoside and wogonin. All of them have indicated the treatment effect in preventing influenza. The treatment effects of RS contents were more significantly detected in old RS than in the young one. Besides the concentration and components of the neuraminidase inhibitors in RS extract, the NA inhibitory activities of each component were also assayed by our fast method. Compared with traditional chromatographic separation and inhibitory activity methods, our neuraminidase inhibitory activity integrated UPLC-DAD-FC/UPLC-Q-TOF-MS method was more efficient, economic and fast in screening and identifying antiviral components from heat-clearing drugs in herbal medicines.