Study on the Mechanism of Gastrodia Rhizoma "FEATURE Identication based Quality Assessmen" based on Neuroprotection

Background: The theory of "Feature Identication based Quality Assessmen" is the essence of traditional Chinese medicine experience identication, and the theory is to identify the quality of Chinese Materia Medica by its properties. The mechanism of evaluating quality through "Feature " has not been claried. Methods: We used articial intelligence sensory technology electronic nose, electronic tongue and other instruments to quantitatively determine the “feature” include "Shape, Color, Qi, taste" of Gastrodia Rhizoma. The relationship between ngerprint of chemical constituents and protective effects on OGD/R injury of SH-SY5Y cells of 30 batches of medicinal materials was analyzed, and the Pharmacodynamic Components Group (six compounds) of Gastrodia Rhizoma were determined. In vitro and in vivo pharmacodynamic experiments conrmed that the Pharmacodynamic Components Group had good protective effects on OGD/R injury of SH-SY5Y cells, spinal injury of zebrash and cerebral ischemia of rats, and could be widely distributed in rats. Results: The Pharmacodynamic Components Group could represent more than 90% of the whole pharmacodynamic effect of Gastrodia Rhizoma, which showed that the Pharmacodynamic Components Group was basically equivalent to the crude drug. Through the correlation analysis of “feature” and Pharmacodynamic Components Group, we revealed that the mechanism of “Feature Identication based Quality Assessmen” of Gastrodia Rhizoma was related to the six components of the Pharmacodynamic Components Group. Conclusions: The mechanism of "Feature Identication based Quality Assessmen" of Gastrodia Rhizoma based on "Feature" is that "Shape", "Color", "Qi" and "Taste" of Gastrodia Rhizoma have signicant correlation with the content of Pharmacodynamic Components Group. This study provides a new way of thinking for the interpretation of the scientic connotation of " Feature Identication based Quality Assessmen" and the quality evaluation of traditional Chinese medicine through veried the vitro and vivo effects and absorption Through the correlation analysis of the of was explained. This study provides new ideas for the interpretation of the connotation of the theory of and standardization of new thinking modes and new technical routes for provides basis for determination of markers of traditional

on the above, this experiment is based on the neuroprotective effect, through the arti cial sensory intelligence technology, to quantify the "Shape", "Color", "Qi", "Taste" of GR. Secondly, HPLC and LC-MS were used to establish the ngerprint of GR and to identify and quantify the common components. Finally, pharmacodynamic components group (PCG) of GR which can represent the overall e cacy of medicinal materials were screened through the Spectrum-Effect relationship, and veri ed by the in vitro and in vivo effects and absorption distribution. Through the correlation analysis of "Feature-Quantity-Effect", the mechanism of "FIQA" of GR was explained. This study provides new ideas for the interpretation of the scienti c connotation of the theory of "FIQA" and the standardization of characters, as well as new thinking modes and new

Experimental drugs.
30 batches of GR (Table 1) was collected. They were identi ed as the dried tuber of Gastrodia Rhizoma Bl. by Professor Zhai Yanjun from the teaching and Research O ce of Liaoning University of traditional Chinese medicine.

Sample Preparation
The GR(No.5) was extracted according to the literature method [13]. The extract was volatilized and added with the culture solution to prepare a solution with rst the concentration (0.4, 0.08, 0.008, 0.004, 0.002 and 0.001 g/ml), second concentration (high 0.25 mg/ml, medium 0.1 mg/ml and low 0.05 mg/ml), third concentration (1 g/ml) according to the calculation of the original medicinal materials for standby.
By using the method of preparing liquid phase, the PCG selected by "spectrum effect" analysis were collected and prepared into the PCG solution and with the concentration of 0.4, 0.08, 0.008, 0.004, 0.002 and 0.001 g/ml respectively.
The other components were collected after the PCG were removed by the method of preparing liquid phase. After volatilization, the culture medium was prepared into a solution with the concentration same as above.
2.3 Quantitative analysis of "Feature" of GR.
Macro "shape". Each batch of GR was randomly taken 6 pieces, and the longest, widest and thickest parts were measured with Vernier Caliper, and weighed on electronic balance. The results of each batch of GR were recorded and averaged.
Microcosmic "shape". The GR power passing No. 5 sieve was selected and accurately weighed 200.0 mg in 6 parts. Chloral hydrate was added to the GR powder of each group, which was grinded and transferred to 10 ml volumetric ask for many times. 7 ml of glycerol was added to each group of samples, and the volume was xed to the scale with chloral hydrate.
After fully shaking the sample solution before each sampling, 0.08 ml of the solution was accurately absorbed, and 50 pieces were made in parallel, and the sclerenchyma cells were counted under the microscope.
Calculate the Microscopic Characteristic Index (MCI) according to the following formula.
The GR powder passing No. 3 sieve was sprea in a measuring dish for measurement, and the measured chromaticity values L*, a*, b* were recorded respectively. Accurately weighed 1.5 g GR powder passing No. 5 sieve was placed in a 20 ml electronic nose special headspace bottle. After automatic injection, each sample was tested 3 times, and each sample was tested 3 times in parallel. After treatment as required, GR sample was directly placed in a special beaker (25 ml) for electronic tongue measurement. Each sample was tested three times, and the response values of seven sensors were obtained on the asree electronic tongue according to the test procedure.

Quantitative analysis of chemical constituents of GR.
HPLC Gradient elution method was used to determine the common peaks of ngerprints from different sources of GR and the content of related effective components [13]. The adjusted mass spectrometry conditions were used. The injection volume was 10 µL. The structure of the common components of GR were demarcated according to the results.

Protective effect on OGD/R injury of SH-SY5Y.
The cells were divided into blank control group, normal control group, model group, treatment group and positive control group. According to the method of literature [13], the model of Oxygen Glucose Deprivation and Reoxygenation (OGD/R) was copied, and chemical components of twelve ngerprints of GR were separated. The protective effect of common components on the injury of SH-SY5Y OGD/R was determined.
The preparation methods of blank group, model group, positive control group and test group were as "2.2".
According to the method of "2.4.1", the protective effect of PCG and other components on the injury of SH-SY5Y OGD/R were determined.
2.6 Protective effect on Zebra sh embryo OGD / R injury.
The embryos were incubated with water at 28 ℃. After Ao staining in vivo, the juveniles of Zebra sh were exposed to 0.64 mmol/L tricacaine methanesulfonic acid, and killed under anesthesia. The procedure of anesthesia was in accordance with the requirements of American Veterinary Association (AVMA). The model of Zebra sh embryo OGD/R injury was duplicated. Gastrodia Tuder Halimasch Tablets were ground and added with culture solution to prepare a solution of 0.005 g/ml. Ten 24 hpf Zebra sh were put into each hole of the 6-hole cell plate and randomly divided into normal control group, model group and treatment group. According to the literature operation [14][15][16][17], we used Image J processing software to process the photos taken, count the uorescence absorption intensity of the injured spinal nerve cells of Zebra sh, and calculated the uorescence intensity and neuroprotection rate. The medicinal material was GR 5.
2.7 Effect on infarct volume of midbrain in rats with local cerebral ischemia.
90 SD rats (half male and half female) were randomly divided into normal control group, model group, positive drug group, No.5 GR extract group (high, medium and low dose), and PCG group (high, medium and low dose). After intraperitoneal injection of Pentobarbital Sodium (dose 50 mg/kg) for 15 minutes, the cervical vertebrae of mice were dislocated when the respiratory rate was reduced and the touch reaction was basically unresponsive. After the brain was taken out, thick sections were prepared, stained with hematoxylin and eosin, and photos were taken. The infarct area was calculated by Image J image processing software. Infarct volume was calculated according to infarct area. Infarct volume = t (A1 + A2 + An) (t is slice thickness, A is infarct area).

Study on absorption and distribution in vivo.
SD rats were randomly divided into 10 groups (3 rats in each group) and given 4 ml of test solution by gavage. Blood samples (0.5 mL) were collected in heparinized tubes from the orbital vein, then the rats were killed at 0.5, 1, 2, 3, 4, 5, 6, 7, 8 and 10 h after administration. The brain, heart, liver, spleen, lung and kidney were collected. After treatment, HPLC was injected to determine the content of effective components at each time point [26].

Statistical treatment.
The experimental results were analyzed by SPSS 19.0 software system, and the calculated data were expressed by mean ± standard deviation. If the variance is uniform, the data were designed with one-way ANOVA and LSD follow-up test. If the variance is not uniform, Welch was used for analysis, and Dunnett T3 was used for multiple comparison. P < 0.05 was the signi cant difference.

Results
3.1 Results of quantitative analysis of "Feature".
The results of macroscopic shape show that (Table 2), there were some differences in the length, width and thickness of GR from different producing areas. The length was 86.1-116.0 mm, the width was 19.88-34.62 mm, and the thickness was 8.18-22.72 mm. The results of microcosmic shape showed that (Table 3) there were signi cant differences in the microcosmic characteristic indexes of GR from different sources, among which the highest was No.7, reaching 74.66, and the lowest was No.14, 13.23. The results might be related to the quality of GR. The results (Table 4) of color analysis showed that the b* value of GR was greater than a * value, and the a * value color was close to the standard white, indicated that the color of GR was yellow and white, which was consistent with the description of "surface yellow white to yellow brown" in Chinese Pharmacopoeia (2015 Edition). It showed that it was feasible to determine the color of GR by using the Color Difference Instrument.    The results showed that ( Table 6) the response values of 7 sensors identi ed by electronic tongue were between 5.19 and 6.35. It could be seen that the "taste" of GR can be quanti ed by electronic tongue technology, the method was stable and feasible, and "Qi" could be used as the basis of "FIQA". In the ngerprint experiment, 12 common peaks were identi ed, and 30 batches of common peak areas of GR from different sources were obtained (Table 7). According to the results of liquid chromatography-mass spectrometry and related literature [27][28], 12 components were determined as: Citric acid, Methyl Citrate, Adenosine, Gastrodin, p-Hydroxybenzyl alcohol, Protocatechuic acid, p-Hydroxybenzaldehyde, Vanillin, Parishin B, Parishin C, Parishin A and 4,4 '-Dihydroxydibenzyl ether. The results of cell experiments of 12 common components showed that Adenosine, Gastrodin, p-Hydroxybenzyl alcohol, Protocatechuic acid, p-Hydroxybenzaldehyde, Vanillin, Parishin B, Parishin C and Parishin A had biological activities, so the contents of the above 9 components were determined [13]. And f1-f12 represented the above 12 components respectively  The partial correlation statistical method was used to analyze the spectral effect relationship with 12 components of GR as independent variables and the protective effect on OGD/R injury of SH-SY5Y as dependent variable The results of the protective effects of the common components on SH-SY5Y cells (Fig. 1) showed that 9 of the 12 common components had protective effects, which were Adenosine, Gastrodin, p-Hydroxybenzyl alcohol, Protocatechuic acid, p-Hydroxybenzaldehyde, Vanillin, Parishin B, Parishin C and Parishin A, respectively. 30 batches of GR ngerprint peak area and the results (Table 8) of SH-SY5Y protection [13] were analyzed by Spectrum-Effect relationship. The results showed that D3, D4, D5, D9, D10 and D11 were the main components affecting the OGD/R injury of SH-SY5Y cells. It was preliminarily determined that there were 6 effective components related to the neuroprotection of GR (Fig. 2). The structural formula of six components were shown in Fig. 3.
Based on spectral screening and neuroprotective activities of common peaks, six components were identi ed as Adenosine, Gastrodin, p-Hydroxybenzyl alcohol, Parishin B, Parishin C and Parishin A. The results showed that the selected PCG had good OGD/R injury protection effect, which could represent more than 95% of the overall e cacy of GE (Table 9). However, the e cacy of the remaining components of the PCG was only 46.4%, indicated that the protective effect of the PCG on OGD/R injury of SH-SY5Y cells could represent GE.  The results were shown in Fig. 4 and Table 10. The results showed that the average uorescence intensity of juveniles was 16.3 ± 1.8 in the normal group, 100.0 ± 2.5 in the model group, and 20.1-39.9 in the positive drug control group (0.05 g/ml).
The effective rate of neuroprotection of the PCG was more than 92%, while the effective rate of the remaining components was less than 40%, which proved that the PCG could represent GR in the protection of Zebra sh spinal injury.

Protective effect on cerebral infarction in rats.
The results showed that (Fig. 5, Fig. 6), the PCG and GR extract had signi cant difference with the model group (P < 0.05).
The infarct volume of high, middle and low dose group of GR extract and PCG were lower than that of model group, which indicated that GR extract and PCG could signi cantly reduce the infarct volume of hypoxic-ischemic brain tissue. The effective rate of each concentration group of PCG was similar to that of GR, indicating that the protective effect of PCG on cerebral infarction in rats could represent GR.

Absorption and tissue distribution in rats.
The results [26] shown that three components could be detected in plasma within 0.5-10 h after oral administration. The content of Adenosine and p-Hydroxybenzyl alcohol reached the peak value in the plasma for 2 h, and the content of Paliscin C reached the peak value in 3 h. Previous experiments showed that Adenosine, Gastrodin, p-Hydroxybenzyl alcohol, Paliscin B, Paliscin C and Paliscin A were the effective components of GE. However, only Adenosine, p-Hydroxybenzyl alcohol and Paliscin C were detected in vivo. The structure of Gastrodin contained the aglycone phydroxybenzyl alcohol and glucose. After entering the body, Gastrodin was degraded to aglycone p-Hydroxybenzyl alcohol and glucose, so Gastrodin was not detected. Paliscin was a kind of compound formed by Gastrodin, p-Hydroxybenzyl alcohol and their derivatives and citric acid at different carboxyl sites, which was easy to metabolize in vivo [27]. In vivo, because the substitution positions of two Gastrodin of Paliscin B were close to each other, combining with Paliscin A of three Gastrodin molecules, it was not easy to decompose the target protein due to the intramolecular crowding caused by Gastrodin [28], while Paliscin B and Paliscin A were superior to Gastrodin in metabolism [29]. The two Gastrodins in Paliscin C were more dispersed and can better combine with the target protein, thus ensuring the stability, so it could be detected in animal blood and tissues.
The results of tissue distribution showed that three components could be detected in each tissue after oral administration.
3.4 The result of the mechanism of FIQA.
The quantitative values of macroscopical "shape", microcosmic "shape", "color", "gas", "taste" and content of each component in the PCG of 30 batches of GR samples were input into SPSS statistical software respectively, and Pearson correlation analysis was used to reveal the mechanism of "FIQA" of GR.
3.4.1 The results of correlation between "Shape" and PCG.
The results of Table 11 showed that the six pharmacodynamic components were related to the macroscopic "shape".
Gastrodin, Paliscin B, Paliscin C and Paliscin A were moderately related to the "Shape" of GR. Adenosine and phydroxybenzyl alcohol had low correlation with "Shape".

3.4.2
The results of correlation between "Color" and PCG. Table 12 showed that the brightness, yellow blue value and total color difference of GR appearance were moderately correlated with the content of Paliscin C. The brightness, red green degree and yellow blue degree of GR cross section were all moderately correlated with the content of Paliscin C. The brightness, red green value, total color difference of GR powder was correlated with the content of Gastrodin. 3.4.3 The results of correlation between "Qi" and PCG.
The results were shown in Table 13. It could be seen that the main components 2, 3, 4, 5 and 6 have good correlation with Paliscin C. The main components 2, 3 were high positive correlation with Paliscin C, 6 was high negative correlation with Paliscin C, and 4, 5 were medium correlation with Paliscin C. 3.4.4 The results of correlation between "Taste" and PCG.
The results were shown in Table 14. It could be seen that Gastrodin, Paliscin B and Paliscin A had good correlation with all "taste" indexes. The four key elements of TCM characters are Shape, Color, Qi and Taste, especially the latter three are the direct expression of chemical components in the characters. Therefore, the research object of this study is "form, color, gas and taste".
Among the microscopic characteristics of GR, including sclerenchyma, needle crystal, gelatinized polysaccharide, etc., sclerenchyma is the special characteristic of GR, which is the main difference between GR and its counterfeit. Therefore, sclerenchyma is selected as the microscopic "shape" of GR.

Quantitative analysis of chemical constituents of GR.
According to the analysis of the effective components of GR from different sources, there was a certain correlation among Gastrodin, p-Hydroxybenzyl alcohol and p-Hydroxybenzaldehyde. In terms of structure, p-Hydroxybenzyl alcohol is the aglycone of Gastrodin, and p-Hydroxybenzaldehyde is the oxidation product of p-hydroxybenzyl alcohol, which has a direct conversion relationship. From the analysis results, the content of p-Hydroxybenzyl alcohol and p-Hydroxybenzaldehyde in most of the GR plants with high content of Gastrodin was higher. This might be related to the different growth cycle of GR or the different degree of mutual transformation of effective components in different habitats.

Determination of PCG.
The neuroprotective effects of 12 common components showed that Gastrodin, Adenosine, p-Hydroxybenzyl alcohol, Paliscin B, Paliscin C and Paliscin A had better effects. According to the results of Spectrum-Effect relationship analysis, six components were selected as the PCG, which were Adenosine, Gastrodin, p-Hydroxybenzyl alcohol, Paliscin A, Paliscin B and Paliscin C.The PCG had good protective effect on OGD/R injury of SH-SY5Y cells. The effect of PCG can reach more than 90% of the whole effect of GR, the average value was 91.01%, which showed that the effective components group selected could replace the whole effect of GR. The activity of the remaining components was less than 30% of the whole effect of GR after removing the PCG, which indicated the rationality of the PCG. GR extract had a signi cant protective effect on the damage of Zebra sh OGD/R. The effect of the PCG was equivalent to that of the whole GR, but the effect of removing the remaining components of the PCG is relatively weak. The experimental results of infarct volume of brain tissue in the model of local cerebral ischemia showed that GR and PCG both had good protective effect. The pharmacodynamic effect of the PCG was equivalent to 98-105% of the whole effect of GR. The above experiments veri ed that the PCG was reasonable. Preliminary experiments showed that PCG such as Adenosine, p-Hydroxybenzyl alcohol and Paliscin C could be detected in vivo. All of the above had shown that the neuroprotective effect of PCG could represent the whole quality of GR.
4.4 Analysis of the mechanism of GR's "FIQA".
Macroscopically, Gastrodin and Paliscin A were negatively correlated with each other, and were moderately negatively correlated with Paliscin B, and moderately positively correlated with Paliscin C. The microcosmic "Shape" was highly correlated with Gastrodin in the PCG, while Paliscin B was highly correlated with Paliscin A. It could be seen that macroscopical "Shape" and microcosmic "Shape" can re ect the quality of medicinal materials. Most of the indexes of GR appearance color were moderately correlated with the content of Paliscin C. The color index of GR cross section was negatively correlated with the content of Paliscin C. The four indexes of powder color of GR have low correlation with Gastrodin. The "Qi" index of GR had a strong correlation with the group of effective components, mainly re ected in the good correlation between the main components 2, 3, 4, 5, 6 and the content of Paliscin C. The "Taste" of GR had a good correlation with the active ingredient group, which was embodied in the fact that the acid, salty, sweet and bitter taste of GR were mostly related to Gastrodin, Paliscin B and Paliscin A. Among them, acid, salty and sweet had a medium correlation with the three components, while bitter had a low correlation with the three components.

Conclusions
The six active components in the PCG are Adenosine, Gastrodin, p-Hydroxybenzyl alcohol, Paliscin A, Paliscin B and Paliscin C. Their effect is close to the whole effect of GR, which can represent the whole effect of GR and re ect the quality of GR. There is a signi cant correlation between the "Shape", "Color", "Qi" and "Taste" of GR and the PCG, which can fully demonstrate the mechanism of "FIQA".   Brain slice of rats. (1-5 were blank group, model group, positive drug group, GR and PCG group respectively).

Figure 6
Protective effect of drugs in each group on cerebral infarction in rats. (∆:P<0.05 was signi cant difference between model group and the blank group. *: P<0.05 was signi cant difference between treatment groups and the model group. **: P<0.01 was signi cant difference between treatment groups and the model group.)