Classification of Andrographis paniculata extracts by HPLC fingerprint analysis and chemometrics

Objective: Andrographis paniculata is widely used in Indonesia traditional medicines called jamu as an antidiabetic. The concentration of some chemical compound will be related to the level of therapeutic effect of A. paniculata and the solvent concentration for extraction affects the number of extracted chemical compound. Quality control method is needed to ensure the consistency level of chemical compound in A. paniculata. High-performance liquid chromatography fingerprint analysis combined with chemometrics was used for evaluation of sambiloto extract according to different solvent extraction. In addition, determination of the andrographolide (major bioactive compound in A. paniculata) and α-glucosidase inhibition activity were also performed. Result: Fingerprint chromatogram of A. paniculata extract with different solvent concentration have similar pattern with several typical peaks appear on each extract, only differ in the peak area and intensity value. Classification of each A. paniculata extract was done by using HPLC fingerprint and principal component analysis. Based on this classification, each extract is grouped in to their respective solvent extraction. The highest andrographolide content and α-glucosidase inhibition activity were in 50% ethanol extract and the lowest were in the water extract. HPLC fingerprint analysis could be used for identification of A. paniculata extract based on solvent extraction.

The composition and concentration of the chemical compounds in the plant will be affected by some factors, for exampale, genetics, environmental growth condition, harvest and postharvest, etc. The type of solvent extraction and its concentration play an important factor in the extraction of a bioactive compound that will be related to the level of biological activity because these differences will affect the amount of extracted bioactive compounds [12]. Quality control of medicinal plant is needed to ensure the consistency of its biological activity and related to its bioactive compound level. There are two approach mainly used for quality control of medicinal plant extract namely marker and fingerprint analysis [13]. The two approach has advantages and could be used in tandem to obtain good evaluation for quality control of medicinal plant. So, in this study we used the two-approach combined with chemometrics for classification of A. paniculata extracts.
Several previous studies have reported the A. paniculata extract in the composition and levels of chemical compounds and certain biological activities. However, no one has reported for A. paniculata extracted with different concentrations of solvents associated with levels of its marker compounds (andrographolide), chemical fingerprinting and inhibition of α-glucosidase. So, we took this opportunity to investigate the effect of solvent extraction concentration on the chemical compound extracted in A. paniculata using HPLC and inhibition of α-glucosidase.

Sample preparation and extraction
We used three months old of A. paniculata and before the extraction process, the sample was prepared by dry sorting and washed with water to clean the sample.
After that, the sample was dried and pulverized. About 10 gram of A. paniculata powdered sample was added 100 mL of solvent extraction and soaking it along with continuously stirring for 6 hours and without stirring for 12 hours. The solvents used for extraction were water, ethanol, 30%, 50%, and 70% ethanol. We collected the filtrate, concentrated with a rotary evaporator, and further dried with the freezedrying process.
The reaction was stopped by adding 100 µL Na 2 CO 3 0.2 M. Enzymatic hydrolysis of the substrate to produce p-nitrophenol was monitored at 410 nm in microplate reader (Epoch-BioTek. Winooski, USA). We also prepared the blank and measured each sample extract in triplicate analysis.

Preparation of sample solutions
The sample solution was prepared by weighing 10 mg of dried extract, diluted with 5 mL 50% methanol (HPLC grade), and sonicated for 1 hour. After that, the sample solutions were diluted with 10 mL of 50% methanol and filtered through a 0.45 μm membrane filter before injected into the HPLC system. Five different sample solutions were prepared and injected to the HPLC.

Determination of andrographolide
Andrographolide content was determined in each extract. A series of a standard solution with five concentrations of andrographolide was made in the range of 10-140 μg/mL to construct a calibration curve. Quantification of andrographolide is using the calibration curve with triplicate measurement.  Table 1. The highest yield when we extracted A. paniculata using 50% ethanol, while the lowest yield with water indicating that different concentrations of solvent extraction will affect the level of metabolite extracted.

Classification of A. paniculata extracts
The α-glucosidase inhibitory activity was determined in order to evaluate the effect of different solvent extraction to its biological activity. The assay principle is, αglucosidase will hydrolyze glucose in the substrate p-nitrophenyl-α-D-glucopyranoside to α-D-glucose and p-nitrophenol and inhibitory activity was measured based on p-nitrophenol produced. Table 2 showed the α-glucosidase inhibitory activity in each extract with inhibition activity was high in 50% ethanol extract follow by 70%, 30% ethanol, water and ethanol. The result showed that combination of water and ethanol could extract more polar and semipolar compounds that are predicted to have α-glucosidase inhibition activity.

HPLC fingerprint and andrographolide content
Each extract of A. paniculata was analyzed using HPLC to know their differences in the composition of metabolite extracted using different solvent extraction. Figure 1 showed the fingerprint chromatogram of A. Paniculata extracts. About 23 peaks were detected in all extract with the percentage of the area more than 5%. Peak 15 (andrographolide) is the major peak in A. paniculata because the intensity and peak Another differences also showed in typical peaks that appear in each extract, such as peak 12 and 22 only appear in ethanol extract. So, this indicates the peak is typical for the fingerprint pattern of the extract. Also, peak 1 appears in 30%, 50% ethanol, and water extracts. It is also can be seen that the more polar extraction solvents will give more detected peaks. We found in Figure 1 water extract gives more detected peaks compared to other extracts. This result is following the previous study that more addition of water in ethanol, which means more polar solvent extraction, the yield is increased [14].
Andrographolide is one of the primary bioactive compounds present in A. paniculata. We have determined this compound in five extracts used in this study to see which extract has a higher andrographolide level. The andrographolide levels in each sample extract are shown in Table 1. Based on the result obtained, the highest andrographolide levels were found in the 50% ethanol extract, while the lowest level is in the water extract. The andrographolide content was shown in the following order, 50% ethanol > ethanol > 70% ethanol > 30% ethanol > water.
These results indicating the amount of andrographolide extracted depends on the polarity of the solvent extraction. As we know from earlier study, andrographolide has a lactone ring and this ring is very vulnerable, reactive and easily rearranged.
The opening of the lactone ring in andrographolide is the initial stage to begin the decomposition process. In water, ring opening will happen through hydrolysis, whereas in alcohol will happen through trans-esterification mechanism. The hydrolysis is estimated to be faster than trans-esterification. Therefore, the rate of andrographolide decomposition depends on the type of solvent. According to research conducted by Kumoro et al. [14], the addition of water will lead to the conversion of andrographolide into deoxyandrographolide through the hydrolysis process, so will reduce the andrographolide levels in the sample.

Classification of A. paniculata extract
The HPLC fingerprint chromatogram for A. paniculata extracts used in this work has a similar chromatogram pattern, only differ in the peak height and area correspond to the level of chemical extracted by different solvent extraction. To differentiate based on HPLC fingerprint chromatogram only will not easy, so we need an aid from chemometrics analysis. We used principal component analysis (PCA) for classification or grouping the extract according to its solvent extraction. The peak area of 8 major peaks (Peak 2, 7, 8, 10, 11, 13, 15, 21) were used as a variable.
Before subjected to the PCA, the variable was pretreated using autoscaling method.
Pretreatment of the data is an important step before the chemometric analysis to get a good result because the quality of input data greatly affects the quality of the output of the analysis. The common autoscaling method is applied by using standard deviation as a scaling factor and producing good analytical output using PCA chemometric analysis techniques [15].
Samples grouping to its solvent extraction was based on the chemical composition using PCA. This multivariate analysis is work to simplify the observed variables by reducing their dimensions and giving an overview of grouping sample through the principal component (PC) [16]. Figure 2A showed the PCA score plot of the A. paniculata extracts. As we can see in the PCA plot, the extracts were grouped according to the solvent extraction. Samples that show the similarity of the profile of the metabolite will be grouped together and the sample that shows the difference will form another group. The two principal components that are most often used in the analysis are the PC1 and PC2. Cumulative percentage of the two PCs used in this study is 89%. According to Varmuza (2001) [17], if the diversity amount of the main components one (PC1) and two (PC2) greater than 70%, the score plot shows good two-dimensional visualization.
PCA biplot is a combination of score plot and loading plot. Loading plot will give us information about how strong each variable affected on the principal component. Figure 2B showed the PCA biplot of A. paniculata extract with the most contribution variables affecting its grouping. We found, peak 5 and 7 gives strong contribution for grouping 50% and 70% ethanol extracts of A. paniculata.

Ethics approval and consent to participate
There is no human and animal study in this work.

Availability of data and materials
The data could be requested from the corresponding author upon reasonable request.  Average of 2 replication a , 3 replication b