Genotype selection for phytochemical content and pharmacological activities in ethanol extracts of fifteen types of Orthosiphon aristatus (Blume) Miq. leaves using chemometric analysis

Orthosiphon aristatus (Blume) Miq. of the Lamiaceae family, called as kumis kucing in Indonesia, is a valuable medicinal plant for their pharmacological properties. The present study comprised of fifteen genotypes of O. aristatus was undertaken to evaluate the genotypes based on phytochemical content and pharmacological activities of leaves ethanol extract. Chemometric analysis (correlation and principal component analysis) was also used to investigate the genetic variability based on phytochemical content and pharmacological activities of O. aristatus genotypes. Results of phytochemical characterization showed that total phenolic ranged from 1.48 to 36.08 (maximum in A15) mg GAE/g DW, total flavonoid ranged from 0.10 to 3.07 (maximum in A15) mg QE/g DW, sinensetin ranged from 0.36 to 4.02 (maximum in A11) mg/g DW, and rosmarinic acid ranged 0.06 to 7.25 (maximum in A7) mg/g DW. Antioxidant activity was tested using DPPH and FRAP assay. Antioxidant results showed that DPPH ranged from 1.68 to 15.55 (maximum in A15) μmol TE/g DW and FRAP ranged from 0.07 to 1.60 (maximum in A1 and A7) μmol TE/g DW. The genotype A8 showed the highest cytotoxic activities against HeLa (66.25%) and MCF-7 (61.79%) cell lines. Maximum α-glucosidase inhibitory activity was recorded in genotype A2 with the value of 62.84%. The genotypes A1, A2, A7, A11, and A15 were identified as superior based on their phytochemicals content and pharmacological activities coupled with chemometric analysis. This finding is important for breeding studies and also the pharmaceutical perspective of O. aristatus.


Results
Extraction yield. The results of ethanol extraction demonstrated the influence of genotypes on the extraction yield of O. aristatus leaves genotypes. The ethanol extraction yield of the studied genotypes varied from 1.73 to 16.91% DW ( Table 1). The genotype A1 had the highest extraction yield, which not significantly different from A11 (16.64% DW) and A15 (16.67% DW) at p < 0.05, and the genotype A6 had the lowest.
Phytochemical analysis. The results of the phytochemical analysis in ethanol extract of O. aristatus genotypes leaves are presented in Table 1. Total phenolic content (TPC), total flavonoid content (TFC), rosmarinic acid content (RAC), and sinensetin content (SC) have been evaluated in ethanol extract of 15 O. aristatus genotypes leaves. TPC varied significantly and ranged from 1.48 mg GAE/g DW to 36.08 mg GAE/g DW. The genotype A15 had the highest TPC, and the genotype A6 had lowest. TFC were observed statistically significant difference at p < 0.05, ranged from 0.10 mg QE/g DW to 3.07 mg QE/g DW among the O. aristatus genotypes. In this study, the maximum TFC was recorded with A15 and minimum was in A6. The highest RAC (7.25 mg/g DW) was found in genotype A7, meanwhile the lowest (0.06 mg/g DW) was observed in genotype A6. The SC ranged from 0.20 (A13) to 4.02 (A11) mg/g DW.
Pharmacological activities. Antioxidant activity in the studied genotypes were measured using DPPH and FRAP assays ( Table 2). In the DPPH assay, the antioxidant properties ranged between 1.68 (A6) to 15.55 (A15) μmol TE/g DW. DPPH value of A15 showed not significantly different with A1 (15.43 μmol TE/g DW) at p < 0.05. In FRAP assay, the genotype A1 and A7 showed maximum antioxidant properties with the same value of 1.60 mmol TE/g DW. In comparison, genotype A6 exhibited the lowest activity with a value of 0.07 mmol TE/g DW.
Cytotoxic activities against HeLa and MCF-7 cell lines of 15 genotypes of O. aristatus extracts are exhibited in Table 2. Cytotoxic activity of O. aristatus genotypes against the HeLa cell line ranged from 7.33 to 66.25%. In the MCF-7 cell line, the cytotoxic activity of O. aristatus genotypes ranged from 21.96 to 61.79%. For both cell lines studied, genotype A8 showed the highest cytotoxic activity. The antiproliferative activity of genotype A8 against the HeLa cell line presented not significantly different from genotype A7 (60.37%). Meanwhile, in MCF-7 cell lines, cytotoxic activity of genotype A8 showed no difference with A2 (57.02%), A5 (58.56%), A14 (56.41%), and A15 (56.57%).
The α-glucosidase inhibition is used to select the O. aristatus varieties with the most antidiabetic potential. The 15 extracts of O. aristatus leaves from different genotypes had a different activity to inhibit α-glucosidase ( Table 2). The inhibition activity was varied from 6.28% to 62.84%. The highest inhibition found in genotypes of A2, whereas the lowest detected in A5.
Multivariate analysis. In this study, the correlation and principal component analysis (PCA) in O. aristatus studied genotypes were performed for multivariate data analysis. Figure 1 shows the correlation matrix chart of phytochemical content and pharmacological activities from the ethanol extract of O. aristatus studied genotypes. Extraction yield showed a significantly positive correlation   Figure 2 shows the loading plot of phytochemical content and pharmacological parameters. EY, TPC, TFC, SC, DPPH, and FRAP parameters had the significant effects on the PC1. Meanwhile, RAC and HeLa parameters had strong influences on the PC2. Biplot analysis of phytochemicals and biological properties created from the comparison of component one-two PCs revealed three distinct clusters of O. aristatus genotypes (Fig. 3). Four groups resulted, as shown in Fig. 1. Cluster I comprised of six genotypes viz. A1, A3, A4, A5, A7, and A8 with characterized by high rosmarinic acid and cytotoxic (HeLa & MCF-7) activity, intermediary for other phytochemical and antioxidant activity, and lowly GIA. Cluster II has composed seven genotypes: A2, A6, A9, A10, A12, A13, and A14. These genotypes characterized by low phytochemical content and biological activities, except high for GIA. Cluster III composed two genotypes (A11 and A15) and characterized by high total phenolic content, total flavonoid content, sinensetin content, extraction yield, and antioxidant activity. www.nature.com/scientificreports/

Discussion
The phytochemical content and pharmacological effects of a medicinal plant species will differ depending on the geographical conditions where the plant grows 25,26 . Therefore, to select medicinal plants that have superior performance for the plant breeding programs, limited environmental conditions are needed 27,28 . In this study, O. aristatus collected from Indonesia was investigated for the characterization of phytochemicals contents and pharmacological activities and obtained promising elite genotypes without environmental impacts. The phytochemical and pharmacological distinction of O. aristatus has economic significance as well as being accommodating in the development of pharmaceutical industries and plant breeding programs for new varieties. Fifteen genotypes of O. aristatus were assessed for the quality of phytochemicals and pharmacological activities in all records of this discussion. Phenolic is a group metabolite associated with several pharmacological activities; therefore, the content of phenolics (TPC) in the plant needs to be determined 29,30 . TPC ranged 1.48 mg GAE/g DW (A6) to 36.08 (A15) mg GAE/g DW (Table 1). TPC found in the leaves of O. aristatus is varied based on the different solvent used in the extraction process, different extraction methods, different extraction time and different environments the plant as well as the different of genotypes. Farhan et al. 31 , found the total phenolic content in the leaves as 230 mg GAE/g DW using methanol for extraction, Abdelwahab et al. 32 , reported TPC as 52.10 mg GAE/g DW from 60% Table 3. First five principle components and its eigenvalue and percent variation explained for phytochemical content and pharmacological activities of O. aristatus genotype. a For an explanation of parameter symbols, see Table 1 and Table 2.  www.nature.com/scientificreports/ methanol extract, and the lowest TPC content found is 20.03 mg GAE/g DW for extraction using 40% ethanol for 120 min at 65 ℃ 33 . The solvent for extraction used in this study is 70% ethanol, so the TPC of the extract in this study compares with the reports of Chew et al. 33 . The difference is our experiment without additional heat and Chew et al. using 65 ℃. Based on the dried weight of the leaves, the TPC of the genotype A15 is higher than reported by Chew et al. 33 . Phenolic compounds in O. aristatus leaves are from flavonoid groups, tannins, and simple phenolics. Simple phenolic compounds include rosmarinic acid (RAC) that is also determined in this study (Table 1). RAC ranged 0.06 mg/g DW (A6) to 7.25 mg/g DW (A7). There is no correlation between TPC and RAC (Fig. 1). It means the phenolic extracted by 70% ethanol is not only rosmarinic acid. Genotype A15 and A7 could be used to O. aristatus varieties development for high phenolics productivity in plant breeding program.
The O. aristatus leaves have a lot of types of flavonoids but the concentration is low. Methoxy flavonoid is the known type of flavonoids in this leaves. Sinensetin, salvigenin and eupatorine are methoxy-flavonoid groups in O. aristatus. The total flavonoid and sinensetin contents of all extracts are shown in Table 1. The report of Abdelwahab et al. 32 , showed that 60% methanolic extract of O. aristatus leaves had 17.46 mg RE/g extract. This report is equal to 8.69 mg QE/g extract. Different from the report of Chew et al. 33 , who reported TFC in dried leaves of O. aristatus is about 1611.9 mg CE/100 g DW which equal to 16.78 mg QE/g DW. Our best TFC content found in A15 is only about 3.07 mg QE/g DW. The sinensetin content in the dried materials of O. aristatus leaves, the highest was found on A11, while the lowest on A8. There is correlation between SC and TFC (Fig. 1). It means that the sinensetin is major flavonoid in studied extract genotypes. On the other hand, it was reported that sinensetin content will decrease when extraction was performed using polar solvent such as ethanol 70% 34 . This result indicated that the genotype A15 and A11 could be developed through a breeding program to get the high flavonoid and sinensetin contents of O. aristatus varieties.
Recently, an antioxidant is of particular concern, which played a critical role in preventing several human illnesses 35 . Thus, getting plant genotypes as a source of high antioxidant activities is essential 36 . In this study, DPPH and FRAP assays were used to evaluate the antioxidant activities in ethanol extract of O. aristatus studied genotypes. DPPH assay was used to analyses the scavenging capacities of extract, while the reducing power was determined by FRAP assay 37 . In present work, the O. aristatus genotypes studied showed variation for DPPH (1.68 to 15.55 μmol TE/g DW) and FRAP (0.07 to 1.60 µmol TEAC/g DW) ( Table 2). The highest antioxidant activities in DPPH and FRAP were found on genotype A15 and A7, respectively. In DPPH assay, our O. aristatus genotypes had lower antioxidant capacity compared with reported by Chew et al. 33 which about 21.81 µmol TEAC/g DW. Scavenging activity of genotype A15 recorded is not significantly different from A1, while genotype A7 and A1 were found not different in reducing power activities. This result indicated that the genotype A1 could be used to develop as O. aristatus plant producing high antioxidant properties. Furthermore, antioxidant activities strong correlated with TPC and TFC indicates that the polyphenol (phenolic and flavonoid) in O. aristatus genotypes are the major metabolites for antioxidant activity. Similarly, Saidan et al. 38 while evaluating the different genotypes, reported that the correlation between TPC and DPPH activity was about 0.90 and FRAP 0.78, while between TFC and DPPH was 0.73 and FRAP 0.71. Our results show that the rosmarinic acid content on O. aristatus leaves has a positive correlation with the DPPH activity and with FRAP activity of the extracts. The R 2 of the correlation between RAC and DPPH activity (0.30) is lower than with FRAP activity (0.54) (Fig. 1). This result is agreed with the report of Ho et al. 39 which concludes that radical scavenging of O. aristatus is related to the high www.nature.com/scientificreports/ content of rosmarinic acid. Since the R 2 is not higher than 70%, it means other compounds also responsible for the antioxidant activity. It can be understood that besides phenolic and flavonoid responsible to the antioxidant activity of O. aristatus, diterpene, triterpene, and saponin groups are also responsible for antioxidant activity 38,40 . Cytotoxic activity of 15 O. aristatus genotypes against HeLa and MCF-7 cell lines is summarized in Table 2. In this work, the O. aristatus genotypes studied showed cytotoxic variation for HeLa (7.33 to 66.25%) and MCF-7 (21.96 to 61.79%). In both cell lines, genotype A8 recorded the highest in cytotoxic activity. The previous report on the cytotoxic study of O. aristatus leaves extracts concludes that this extract is toxic against breast cancer (MCF-7, 53.4% inhibition) and colon cancer (HCT116, 43.7% inhibition) cell lines 38 . Our results show better cytotoxicity compared with the previous report of Saidan et al. 38 . O. aristatus extract has also been reported to have cytotoxic activity against prostate cancer (DU145) with no harmful effect against normal fibroblast cell of HSFF 1184 34 . Saidan et al. 38 conclude that the cytotoxic activity of O. aristatus extract had a positive correlation with TPC and TFC. This phenomenon is not found in our results. There is no correlation between TPC and TFC with cytotoxic activity against MCF-7 and HeLa. However, MCF-7 presented weakly positive correlation with TFC and SC (Fig. 1). These indicate that the flavonoid and sinensetin compounds are responsible for cytotoxic activity in O. aristatus extract. Previous work has shown that methoxylated flavonoids (eupatorine and sinensetin) have contributed greatly to the cytotoxic activity of O. aristatus extract 41,42 . Our results also show that there is a correlation between cytotoxic activity and rosmarinic acid, which in line with the previous report of Suhaimi et al. 34 . Suhaimi et al. 34 said that the highest content of rosmarinic acid showed an anti-proliferative effect against prostate cancer. Results indicated that the genotype A8 could be developed through a breeding program to get the high anticancer capacities of O. aristatus varieties.
Multivariate analysis, namely chemometric, is the statistical procedure to enhance the understanding of metabolite information and to associate quality traits to data of analytical instrument. It has been used to study in the correlation of the phytochemical component with pharmacological activities in the plant 20,28,43 . In this study, we used two chemometric methods namely correlation and principal component analysis (PCA) to evaluate the phytochemical and pharmacological properties of the O. aristatus. Antioxidant activity (DPPH & FRAP) found a significantly positive correlation with total phenolic and total flavonoid contents (Fig. 1). These indicate that they are significant selections parameters in cultivating the O. aristatus. Furthermore, these results indicated that the polyphenols are the main compounds for antioxidant capacities in O. aristatus. The results are in line with previous reports in O. aristatus and other different crops 38,44,45 . Besides, several reports shows association between polyphenol contents and antioxidant properties, but no one highlighted the mechanism of the association between polyphenol compounds and antioxidants in O. aristatus. This work showed a strong correlation between rosmarinic acid & FRAP assay and sinensetin & DPPH assay. Results indicated that the antioxidant mechanism of rosmarinic acid from O. aristatus as reducing power while free radical scavenging mechanism found in the sinensetin compound. Interestingly, in PCA using phytochemical contents and pharmacological parameters, it was resulted three clusters. Two O. aristatus genotypes viz. A11 and A15 had high TPC, TFC, SC, EY, DPPH, and FRAP, indicating a correlation of high polyphenol in these genotypes with antioxidant activities and extraction yields. Six O. aristatus genotypes viz. A1, A3, A4, A5, A7, and A8 had high rosmarinic acid and cytotoxic activities (HeLa & MCF-7), indicating the correlation of high rosmarinic acid content in these genotypes with cytotoxic properties. In the future, combining with anticancer activities such as A8 genotype with high phytochemical contents such as A15 genotype will be the main goal for new varieties development as anticancer sources with valuable phytochemical-rich in O. aristatus.
Fifteen O. aristatus genotypes had phytochemical and biological activities diversity. Genotypes of A11 and A15 produced high total phenolic and flavonoid contents, genotype A7 produced high rosmarinic acid content, and genotype A11 produced high sinensetin content. The genotype A1 and A7 had the best antioxidant activity against FRAP, while genotype A15 had the highest antioxidant activity against DPPH assay. The genotype of A2 had the best cytotoxic and α-glucosidase inhibition activities. The selected genotypes with different purposes could be developed in plant breeding.

Methods
Plant materials and sample preparation. Plant materials consisted of 15 different O. aristatus genotypes, belonging to dissimilar plant height (short, medium and tall), anthocyanin coloration on stem (absent or very weak, weak, medium and strong), leaf shape (narrow elliptic, medium elliptic, and medium ovate), flower colour (white and purple), and maturity group (early and late) ( Table 4). Fifteen types of O. aristatus were cultivated in the experimental fields of Tropical Biopharmaca Research Center, Bogor Agricultural University, West Java, Indonesia (6°32′25.47″ N, 106°42′53.22″ E, 142.60 m above the sea level) with the randomized complete block design in three replications. For the evaluation of phytochemicals and biological activities, leaves were harvested at three months after planting. After harvesting, the leaves of each genotype were dried in the oven (50 °C). The powder (100 mesh) samples were extracted in 70% ethanol using the maceration method described by Abdullah et al. 46 , with slight modification. Briefly, 15 g leaves samples were taken and their extraction was performed in 150 ml ethanol for 24 h at 150 rpm rotation speed (Eyela multi shaker MMS-Germany) and room temperature. The solution was filtered with Whatman no. 4. Then, rotary evaporator was applied to evaporate the extracts. To calculate the extraction yield, the extracts were gathered and weighed. Resulted extracts were kept at 4 °C until further usage.

Scientific Reports
| (2020) 10:20945 | https://doi.org/10.1038/s41598-020-77991-2 www.nature.com/scientificreports/ Determination of total phenolic and flavonoid content. Total phenolic and flavonoid content were determined using spectrophotometrically suggested by Khumaida et al. 47 . Total phenolic content in each extract was determined using the Folin-Ciocalteu reagent and gallic acid was used as an external standard. About 10 μl extract sample was mixed with distilled water (160 μl), 10% Folin-Ciocalteu reagent (10 μl), and 10% Na 2 CO 3 (20 μl). After incubation at room temperature for 30 min, the absorbance was measured using the microplate reader (Epoch BioTek, USA) at 750 nm. The total phenolic content was expressed as gallic acid equivalent (mg GAE/ g DW). Total flavonoid content was measured using aluminium chloride reagent and quercetin was used as an external standard. In a 96-well microplate, a 10 μl extract sample was mixed with methanol (60 μl), 10 μl of 10% aluminium chloride, 10 μl of 1 M potassium acetate, and 120 μl of distilled water. Finally, after incubation at room temperature for 30 min, the absorbance was measured using the microplate reader (Epoch BioTek, USA) at 415 nm. Results expressed as quercetin equivalent (mg QE/ g DW).
Determination of sinensetin and rosmarinic acid content. Sinensetin  Determination of antioxidant activity. Two in-vitro methods were applied to analyse antioxidant activities. DPPH and FRAP methods were used to evaluate the free radical scavenging and reducing power activities, respectively. Antioxidant activity was expressed as Trolox Equivalent with Trolox (vitamin E analogues) as an antioxidant standard. Trolox was frequently used as an α-tocopherol model compound 49 . Furthermore, Trolox is used to express the antioxidant capacities of food, chemical compounds and biological matrices, as a standard antioxidant compound in terms of Trolox equivalent antioxidant capacity (TEACH) 50 . For this reason, Trolox can be considered as the proper counterpart of Vitamin E to examine its actions in aqueous radical environments on its own and in combination with other antioxidant compounds. For 2,2-diphenyl picrylhydrazyl (DPPH) method proposed by Nurcholis et al. 51 , with modification. Briefly, 100 µl O. aristatus ethanol extract was added to 100 µl of 125 µM DPPH in methanol at the 96-well microplate (Costar-USA). After incubation in darkroom temperature for 30 min, the absorbance was measured using the microplate reader (BMG Labtech, Germany) at 517 nm. Results expressed in µmol Trolox Equivalent (TE/g DW).
FRAP was determined according to the assay of Benzie and Strain 52 with modification. FRAP reagent was prepared by mixing 300 mM acetate buffer pH 3.6, 10 mM TPTZ in 40 mmol HCl, and 20 mmol FeCl 3 in the ratio www.nature.com/scientificreports/ of 10:1:1 (v/v/v). One hundred µl O. aristatus ethanol extract was added to 300 µl of FRAP reagent in the 96-well microplate (Costar-USA). After incubation at 37 °C, the absorbance was measured using the microplate reader (BMG Labtech, Germany) at 593 nm. Antioxidant activity was expressed as µmol Trolox Equivalent (TE/g DW).
Determination of cytotoxic activity. Cytotoxic activity was measured by the hemocytometer method according to previous reported by Fitria et al. 53 . MCF-7 and HeLa cancer cell lines, obtained from Veterinary Medicine Faculty, IPB University, were used cytotoxic evaluation. Cell lines were grown in medium which consist of 850 μl Dulbecco's minimum Eagle's medium (Gibco, Rockville, MD, USA), 30 μl fetal bovine serum (10%; Sigma-Aldrich, St. Louis, MO, USA), and 10 μl fungizone (Gibco, Rockville, MD, USA). In brief, 50 μl cell line were exposed to a sample of 100 μg/ml. Also included, the control group was an untreated cell line. The cell incubated at 37 °C in 5% CO 2 for 3 days. After incubation, 100 μl cells were added to 5 μl trypan blue. After homogenized, 10 μl cells were placed in a hemocytometer for calculation number of cells. The cytotoxic activity was determined based on the treated cell and untreated cells in percentage inhibition.
Determination of α-glucosidase inhibitory activity. The α-glucosidase inhibitory activity in O.
aristatus genotype was performed according to the assay previous reported 54 . The extract sample (10 μl) of concentration 1000 μg/ml was a mixed with 0.1 M phosphate buffer pH 7.0 (50 μl), 0.5 mM pNPG (25 μl), and 0.2 unit/ml α-glucosidase solution (25 μl). Then, the mixture incubated at 37 °C for 30 min. Finally, the reaction was stopped by 200 mM Na 2 CO 3 (100 μl) and the absorbance was measured using the microplate reader (Epoch Biotech, USA). Results expressed as percentage inhibition of α-glucosidase activity. The percentage of α-glucosidase inhibition (GIA) was calculated using formula (1): where the control reaction absorbance (with all reagents but one of its extract) is A control and the extract absorbance tested in the reaction mixture is A sample .
Statistical analysis. Data were expressed as the mean ± SD from three replicates. ANOVA was analysed using SPSS version 25. R was used for multivariate analyses using correlation and principal component analysis. The Pearson correlation coefficients between phytochemical and pharmacological parameters were generated using PerformanceAnalytics packages in R 55 . The FactoMineR packages in R was used to create PCA analysis using data matrix of phytochemical and pharmacological variables 56 .