A comparative assessment of in vitro cytotoxic activity and phytochemical profiling of Andrographis nallamalayana J.L.Ellis and Andrographis paniculata (Burm. f.) Nees using UPLC-QTOF-MS/MS approach

Andrographis paniculata (Burm. f.) Nees and Andrographis nallamalayana J.L.Ellis have traditionally been used to treat various ailments such as mouth ulcers, intermittent fever, inflammation, snake bite. This study compares the comparative in vitro cytotoxic activity, and phytochemical profiling of methanol extract of A. nallamalayana (ANM) and A. paniculata (APM). UPLC-ESI-QTOF-MS/MS analysis has been performed. The cytotoxic activity of crude methanol extracts were evaluated against three different cancer cell lines (HCT 116, HepG2, and A549 cell line). Both plants' extract exhibited significant cytotoxic activity against tested cell lines in a dose-dependent manner. IC50 of ANM and APM in HCT 116 cell was 11.71 ± 2.48 μg ml−1 and 45.32 ± 0.86 μg ml−1 and in HepG2 cell line was 15.65 ± 2.25 μg ml−1 and 60.32 ± 1.05 μg ml−1 respectively. Cytotoxicity of these two extracts was comparatively similar in A549 cells. ANM induced cytotoxicity involved programmed cell death, externalisation of phosphatidylserine, ROS generation, up-regulation and down-regulation of major apoptotic markers. HRMS analysis of ANM and APM resulted in the identification of 59 and 42 compounds, respectively. Further, using the MS/MS fragmentation approach, 20 compounds, of which 18 compounds were identified for the first time from ANM, which belongs to phenolic acids, flavonoids, and their glycosides. Three known compounds, echioidinin, skullcapflavone I and 5,2′,6′-trihydroxy-7-methoxyflavone 2′-O-β-d-glucopyranoside, were isolated from A. nallamalayana and their crystal structures were reported for the first time. Subsequently, seven major compounds were identified in A. nallamalayana by direct comparison (retention time and UV-spectra) with authentic commercial standards and isolated compounds using HPLC-UV analysis. The cytotoxicity of phytochemicals from both the plants using in silico tools also justify their in vitro cytotoxic activity. It is the first report on the comparative characterisation of phytochemicals present in the methanolic extract of both the species of Andrographis, along with the cytotoxic activity of A. nallamalayana.


Introduction
Herbal medicines are the oldest kind of medicines the human race is aware of. Since time immemorial, plants have been used as sources for food, shelter and treating illness. Time and again, herbal drugs have been used by people all across the world. In India, they hold a special place. Due to the vast geographical difference, there is a drastic variety of medicinal plants, and they have been used for different ailments. 1 With a large variety of plants, Acanthaceae is considered one of the top nine families of medicinal plants, including 2500 species and 250 genera. 2 One of the important genera of Acanthaceae is Andrographis, which is widely used in the Indian medicine system. given orally from the third day of delivery/menses for four days and also for the treatment of leucorrhoea by the Adivasi in the Eastern Ghats of Andhra Pradesh, India. 22 Recent studies have shown that methanolic extract of A. nallamalayana used for antimicrobial, 23 anti-psoriatic, 24 anti-candidal 25 and antiproliferative, anti-inammatory and pro-apoptotic activities. 26 The preliminary phytochemical screening demonstrated the presence of avonoids, alkaloids, phenols, steroids and triterpenoids. Parlapally et al. using GC/MS and LC/MS analysis, reported the presence of chromones, avones/avanones and their glycosides. 24 UHPLC-ESI-QTOF-MS techniques have been used as a powerful analytical tool because of their high accuracy and sensitivity in characterising various complex natural products materials. The attained accurate mass spectra of elemental composition and tandem mass spectrometry (MS/MS) spectra allow detection and identication of the individual chemical structures. 27 The novel drug development is a very complicated and time-consuming process. However, nearly 40% of the drug applicants failed due to unanticipated toxicity and adverse drug reactions. For the preliminary stage of drug development, computer-aided in silico strategies have become vital as they support more cost-effectively. [28][29][30] For the development of bioactive phytoconstituents, the global research scenario recommends using virtual screening methods/technology. 31 Prediction of possible pharmacological activity via in silico approach is based on the structure-activity relationship, which is usually correlated with the experimental data. 32,33 In silico studies combined with biological activities would reduce the time and cost for the development of novel drugs. A. paniculata is a mine for bitter compounds for medicinal purposes, but the scarcity of literature studies on A. nallamalayana related to phytochemical proling and biological evaluation paves the way for this study. In this study, an attempt was made to determine the comparative in vitro cytotoxic activity and the phytochemical proling of A. nallamalayana and A. paniculata and in silico prediction of cytotoxic activity of identied compounds in order to validate the ethnopharmacological use of these plants in India. To the best of our knowledge, it was the rst report on the comparative characterisation of phytochemicals present in the methanolic extract of both the species of Andrographis, along with the cytotoxic activity of A. nallamalayana.  Yogi Vemana University, Kadapa, Andhra Pradesh, India, in consultation with Herbarium of Botanical Survey of India (BSI) Deccan Regional Centre, Hyderabad. The above voucher numbers given to the herbarium sheets, and the herbarium sheets were deposited in the Herbarium, Department of Botany, Yogi Vemana University, Kadapa, Andhra Pradesh, India (ESI Fig. 1 †).

Extraction and isolation
The leaves of A. nallamalayana (400 g) and A. paniculata (800 g) were shade dried for 7-8 days to achieve an optimum moisture content varied from 7% and 9%, respectively, before grinding to lesser particle size. The powdered leaves were defatted with petroleum ether (3 Â 48 h) and then extracted with methanol (3 Â 72 h) at room temperature using the cold maceration method. The methanol extracts of A. nallamalayana (53.38 g) and A. paniculata (87.25 g) were ltered through Whatman lter paper, and the ltrates were concentrated at 40 C under reduced pressure. The extracts were stored at 4 C in an airtight container until further use.

Characterization and structural determination
Characterization and structural determination of three compounds isolated from methanolic extract of A. nallamalayana were established mainly based on single-crystal X-ray crystallography, 1D NMR and mass spectral studies. The single-crystal X-ray diffraction (XRD) data was collected on a Bruker D8 Venture system (Bruker, Billerica, Massachusetts, United States) with microfocus optics using CuKa (l ¼ 1.54178) radiation. The data for three compounds were analysed and processed using Bruker Apex III soware suite, 34 incorporated with multiple tools such as cell_now and RLATT to determine unit cell, SAINT-plus for data reduction SADABS for absorption correction. The structure solution was performed with SHELXT, 35 and full matrix least-squares renements were performed using the SHELXL 36 suite of programs incorporated in either Apex III suite 34 or Olex 2.0-1.3-alpha. 37 A Bruker Avance-600 MHz superconducting FT-NMR spectrophotometer (Bruker, Billerica, Massachusetts, United States) with RT-TXI probe used to record 1 H and 13 C NMR spectra for the isolated compounds in DMSO-d 6 and tetramethylsilane (TMS) as an internal standard. HRMS of compounds was obtained on Agilent 6545B Q-TOF LC/ MS instrument (Agilent Technologies, Santa Clara, California, United States) in negative ionization mode.

Cell culture
HCT 116 (human colorectal carcinoma), HepG2 (hepatocellular carcinoma), A549 (human lung cancer), HEK 293 (human embryonic kidney cell) cells were grown in a humidied environment below 5% CO 2 in DMEM media combined with 10% FBS and 1% antibiotic (PSN) at 37 C. Cells were harvested with 0.5% trypsin and seeded at optimum density the day before the experiment was performed.

Cytotoxicity assay
Cytotoxicity of ANM and APM was determined by MTT [(3-(4,5dimethythiazol-2-yl)-2,5-diphenyltetrazolium bromide)] assay. 38 Cells were seeded into 96-well plates (1 Â 10 6 per well) and treated with different ANM and APM concentrations for 24 h before assessment using the MTT assay. Both the extracts were dissolved in 0.05% of DMSO to achieve extract concentrations of (10, 20, 40, 60, 80, 100 and 120 mg ml À1 ) and held in a humidied (5% CO 2 ) atmosphere and kept in the incubator for 24 h at 37 C. MTT (5 mg ml À1 ) was added aer incubation, and the plates were additionally incubated for another 4 h. Using an ELISA reader, the absorbance of DMSO-soluble intracellular formazan salt was measured at 595 nm. This experiment was carried out in triplicate. The percentage of cell death was determined by calculating the percentage inhibition and IC 50 value.

Determination of intracellular ROS generation (iROS)
Reactive oxygen species (ROS) generation was determined using the 2 0 ,7 0 -dichlorouorescein diacetate (H 2 DCFDA) dye which uses an increase in green uorescence intensity to quantify the intracellular ROS generation with respect to untreated cells. 39 The cells treated with the ANM (IC 50 ) were incubated at 37 C with 10 mM of H 2 DCFH-DA for 30 minutes following the ow cytometer determination (BD LSRFortessa, San Jose, CA, USA). The increase in DCF uorescence directly redirects the ROS produced inside the cells, representing the mean DCF uorescence intensity.

Detection of apoptosis by ow cytometry
Cell apoptosis is another critical parameter for the toxicity of materials. The determination of apoptosis and necrosis were analysed by ow cytometry using annexin-V-FITC/propidium iodide (PI) detection kit (Calbiochem, CA, USA). 40 Briey, in a six-well plate, HCT 116 cells were seeded and treated with ANM (IC 50 ) time-dependently up to 48 h and were stained with annexin-V/FITC-PI as per the direction of the manufacturer (Calbiochem, Merck Millipore, Burlington, Massachusetts, USA). The percentage of live, apoptotic (early and late), and necrotic cells were quantied using a ow cytometer (BD LSR Fortessa, San Jose, CA, USA).

Western blotting
Total protein isolation from HCT 116 cells was performed using cell lysis buffer, which is supplemented by phosphatase and protease inhibitor cocktail; proteins have been quantied by BCA assay kit (Thermo Fisher Scientic). 41 Using treated and non-treated cells, 40 mg of proteins were rst isolated and then separated electrophoretically into SDS polyacrylamide gel (12%) which was later transferred to PVDF membrane (Immobilon-P, Millipore Company, Bedford, MA, USA) by using wet trans-blot system (Transblot: wet transfer cell; Bio-Rad Laboratories, Inc., Hercules, CA, USA). The membranes were blocked with BSA for 2 h and then incubated with primary antibodies anti-bcl2 (SC-7392), anti-cleaved PARP 1 (SC-56196), anti-PUMA-a (SC-37701), anti-cleaved-caspase-9 (SC-56076) and anti-b-actin (SC-47778) with 11.707 AE 2.482 mg ml À1 , ANM (IC 50 ) for 0, 12, 24, and 48 h. Aer thorough washing, secondary antibodies were conjugated by the membranes and incubated with HRP. By adding ECL substrates, immunoreactive bands were visualised. b-Actin was used as loading endogenous control.

2.10
In vitro wound healing assay HCT 116 cells were seeded in 6-well plates and incubated at 37 C overnight for 24 h. Using pipette tip thrice washed with PBS, a straight wound was rendered onto the conuence cell layer. The cells in serum-free DMEM medium were then treated with ANM (IC 50 ). The wound repopulation gap width was measured and recorded at 0, 12, 24 and 48 h of incubation and was then compared to the original gap size at 0 h. The distance was calculated using the image processing system ImageJ, and the gap size was checked from the digital images at each point in time. 42

Determination of total phenolic and avonoid content
The Folin-Ciocalteu (F-C) colorimetric method was used to determine the total phenolic content described earlier. 43 Different concentrations of gallic acid (25 to 1000 mg ml À1 ) have been prepared and used to generate the calibration curve using a linear t (y ¼ 0.048x + 0.063, R 2 ¼ 0.987). Total phenolic content was represented as gallic acid equivalent (GAE) in mg g À1 of dried extract weight (mg of gallic acid per g dry weight). All the samples were done in triplicates. The aluminium chloride colourimetric method described by Chia-chi Chang et al. was used to determine total avonoid content. 44 Various quercetin concentrations ranging from 0 to 500 mg ml À1 have been prepared and used to generate the calibration curve. The total content of avonoid was calculated by using the calibration curve (y ¼ 0.063x + 0.131, R 2 ¼ 0.970) and expressed in quercetin equivalents (QE) per gram dry extract weight. All the other determinations were carried out in triplicate.

UPLC-QTOF-MS and MS/MS analysis
Metabolite separation of A. nallamalayana and A. paniculata methanolic extract was performed on the Agilent 1290 Innity LC system. 1.0 mg of dried extract was dissolved in 1 ml of LC-MS-grade methanol containing 0.1% formic acid (v/v) and ltered through a 0.2 mm PTFE membrane lter before the analysis was performed. The chromatographic separation was achieved on Agilent ZORBAX SB-C18 column (2.1 Â 100 mm, 1. Before the next injection, the column was reconditioned for 5 minutes. 0.5 ml min À1 ow rate with a 0.5 ml injected volume was used for analysis, the UPLC system assembled with a diode array detector (DAD) and an autosampler.
The Agilent 1290 Innity LC device was coupled to Agilent 6545B Accurate-Mass Quadrupole Time-of-Flight (QTOF) for MS/MS study with Agilent Jet Stream Thermal Gradient Technology with electrospray ionisation (ESI) source. The analysis was performed in both positive and negative ionisation mode to obtain high-resolution mass spectra. The ESI parameters have been optimised as: the ow of drying gas (N 2 ), 8 l min À1 ; temperature of drying gas, 150 C. Other parameters were set as: fragmentor voltage, 150 V; skimmer voltage, 65 V; capillary voltage, 3500 V; nebuliser gas, 35 psig; nozzle voltage 1500 V. Fixed collision energies of 10, 20, 30, 40, 50 and 70 V were used for MS/MS analysis. The UPLC-QTOF data acquisition was performed using Agilent MassHunter Acquisition B.06.01 soware (Agilent Technologies, Santa Clara, CA, USA). With Agilent MassHunter Qualitative Analysis B.07.00 (MassHunterQual, Agilent Technologies), the data were deconvoluted into individual chemical peaks using Molecular Feature Extractor (MFE). The prediction of molecular formula and accurate molecular mass for putative molecules were screened in databases such as METLIN, CAS and MassBank. Agilent Technologies has provided an accurate mass MS/MS Library (PCDL) for the METLIN Personal Compound Database. METLIN PCDL contains all compounds and additionally accurate mass Q-TOF-MS/MS library reference spectra.
2.14 Prediction of the in silico biological activity 2.14.1 In silico prediction of anticancer activity using PASS. PASS (prediction of activity spectra for substances), a soware program, was used to obtain the identied compounds biological activity spectrum, including the anticancer activity. PASS is a widely used web tool (http://www.pharmaexpert.ru/ passonline) that contains more than 1 million biologically relevant compounds and can predict 7000 different pharmacological effects with 95% accuracy according to leave-one-out cross-validation (LOOCV) estimation. The program is based on the multilevel neighbourhoods of atoms (MNA). MNA descriptors were used to represent the chemical structure, and the prediction of activity is usually based on the structureactivity relationship (SAR) analysis of the training set according to the Bayesian algorithm. [45][46][47] PASS represents the activity spectrum as "probable activity (Pa) or probable inactivity (Pi)". The experimental value of Pa and Pi lies within the range of 0.000 to 1.000. When the value of Pa > Pi, i.e. if Pa > 0.7, then it represents that the compound would be experimentally active. A higher value of Pa reects the signicant biological effect experimentally and vice versa. 48,49 The structure of all the iden-tied phytoconstituents (20 compounds) were obtained from the PubChem database. The structures were submitted using the SMILES format and were subjected to the evaluation of the biological activity spectrum, including the anticancer activity.
2.14.2 In silico prediction of cell line cytotoxicity with CLC-Pred tool. CLC-Pred (Cell-Line Cytotoxicity Predictor), a freely available web-service for cell-line cytotoxicity prole prediction tool, was used to predict the cytotoxicity of the identied compounds in various cell lines (http://way2drug.com/Cell-line/). This prediction of cytotoxicity is based on the PASS (prediction of activity spectra for substances) program, which uses structure-cell line toxicity relationship using special training sets with leave-oneout cross-validation procedure. The predicted cytotoxicity is represented by Pa and Pi values; if Pa > 0.5, the probability of cytotoxicity would be high (active), whereas Pi value represents inactivity. 28,50 The structures were submitted in SMILES format and subjected for evaluation of cytotoxicity using the CLC-Pred tool.

Statistical analysis
Results were represented as mean AE SEM of the multiple data points. Statistical importance in the deference was calculated by the analysis of variance (ANOVA) and paired T test using GraphPad Prism (version 8.4.3) soware where p < 0.05 was considered as signicant.

Results and discussion
3.1 Assessment of in vitro cytotoxicity of crude methanolic extracts of A. nallamalayana and A. paniculata A comprehensive literature survey indicated only a few reports describe the cytotoxicity of A. nallamalayana, 26 whereas previous studies showed that A. paniculata exhibited cytotoxic activities against several tested cancer cell lines. [51][52][53][54] Motivated by these ndings, we were also interested in investigating the role of ANM as an anti-proliferative agent. Our study revealed that methanolic extract of A. nallamalayana (ANM) and A. paniculata (APM) showed signicant cytotoxicity towards three different types of cancer cell lines HepG2 (hepatocellular carcinoma), A549 (human lung cancer), HCT 116 (human colorectal carcinoma) in a dose-dependent manner as shown in Fig. 1A-C. Compared to APM, ANM effectively reduced cell viability in all tested cancer cell lines. The cytotoxicity of ANM was nearly four times higher than APM in HCT 116 and HepG2 cells. In HCT 116 cells, the IC 50 of ANM and APM was 11.71 AE 2.48 mg ml À1 and 45.32 AE 0.86 mg ml À1 , respectively, whereas, in HepG2 cells, it was 15.65 AE 2.25 mg ml À1 and 60.32 AE 1.05 mg ml À1 , respectively. Cytotoxicity of these two extracts was comparatively similar in A549 cells (Table 1). Both extracts did not show signicant cytotoxicity in HEK 293 cell line (human embryonic kidney cell) up to 120 mg ml À1 concentration (Fig. 1D). Andrographolide was used as the positive control, and the IC 50 value of andrographolide (42.723 AE 0.668 mg ml À1 ) in HCT 116 cells was similar to the methanolic extract of A. paniculata (Fig. 1E). Our results are consistent with the prior studies, where an alcoholic extract of A. paniculata exhibited cytotoxic activity against HT-29 (human colon) and IMR-32 (human neuroblastoma) cancer cell lines resulted in 51.25 AE 0.85 and 50.25 AE 1.6% inhibition at 200 mg ml À1 , respectively. 51 In another study, methanolic extract of A. paniculata demonstrated signicant anti-proliferative activity in MCF-7 (breast cancer) cell lines with minimum inhibition at a concentration of 31.25 mg ml À1 . 52 Dichloromethane fraction of methanol extract is also reported to maintain active compounds that contribute to the anticancer and immunostimulatory activity. The dichloromethane fraction signicantly reduces the proliferation of HT 29 cells (colon cancer) and increases the proliferation of human peripheral blood lymphocytes (HPBLs) at low concentrations. 53 Previously, the methanolic extract of A. nallamalayana reported for anti-proliferative activity against A375 and B16F10 melanoma cell lines. 26 The cytotoxic activity of A. nallamalayana and A. paniculata was categorise according to the guidelines of National Cancer Institute (NCI) as follows: IC 50 # 20 mg ml À1 ¼ highly active, IC 50 21-200 mg ml À1 ¼ moderately active, IC 50 201-500 mg ml À1 ¼ weakly active and IC 50 > 501 mg ml À1 ¼ inactive. 55,56 Following the NCI guidelines, it was concluded that both the extracts showed moderate to high activity in cancer cell lines. Further experiments were focused on exploring the mechanism of cytotoxicity of the methanolic extract of A. nallamalayana as it showed the better cytotoxicity as compared to methanolic extract of A. paniculata.

Reactive oxygen species (ROS) generation by crude methanolic extract of A. nallamalayana
It is well established that a rise in intracellular ROS (iROS) levels contribute to apoptosis-induced cell death, causing DNA damage and harm to other cell organelles. 57 ROS production can be measured by 2 0 ,7 0 -dichlorodihydrouorescein diacetate (H 2 DCFDA), a non-uorescent molecule. It was observed that following treatment with 11.707 AE 2.482 mg ml À1 , ANM (IC 50 ), the mean uorescence intensity of dihydro-dichlorouorescein (DCF) was increased signicantly over time, indicating that ROS generation is directly related to ANM-induced cytotoxicity (Fig. 2). The relative DCF uorescence intensity in ANM treated HCT16 cells increased in a time-dependent manner.

Annexin V-FITC/PI determination by ow cytometry of crude methanolic extract of A. nallamalayana
Activation of apoptosis is an important strategy in the treatment of cancer. The externalisation of phosphatidylserine (PS) from the inner membrane to the cell's outer membrane is the main characteristic of early apoptosis, and late apoptosis is achieved through DNA fragmentation. 58 To examine the possible induction of cell death (necrosis and/or apoptosis), experiment was performed using annexin V/propidium iodide assay by studying the exposed level of phosphatidylserine in the outer membrane of cells. 59 In this assay, Q3, Q4, Q2 and Q1 reect living cells, early apoptotic (EA), late apoptotic (LA), and necrotic, respectively. The percentage of apoptotic (early and late) cells were signicantly increased in a time-dependent manner (5.3% EA/28.1% LA for 12 h, 5.9% EA/30.2% LA for 24 h and 8.3% EA/54.1% LA for 48 h) compared to control cells (0.7% EA and 3.1% LA) in ANM (IC 50 ) treated cells (Fig. 3). A signicant number of annexin-V-FITC positive cells indicated that ANM induced cytotoxicity in HCT 116 cells were triggered through apoptosis. The understanding of apoptosis can be used to develop new targeted medicines that stop cancer cells from growing and spreading.

Regulation of apoptosis markers by crude methanolic extract of A. nallamalayana
We investigated the levels of protein expressions related to the induction of apoptosis in HCT 116 cells aer treatment with 11.707 AE 2.482 mg ml À1 , ANM (IC 50 ) for 0, 12, 24, and 48 h to investigate its effect on pro-and anti-apoptotic proteins. Previous studies in colon cancer cells showed that PUMA is a mitochondrial protein, and its mitochondrial position is necessary for apoptosis induction. 60 PUMA (p53 up-regulated apoptosis modulator) a member of Bcl-2 homology 3 (BH3)-  (10,20,40,60,80,100, and 120 mg ml À1 ) for 24 h, and cytotoxicity was determined by MTT assay. Data are expressed as mean AE SEM for triplicate experiments. Here *** denotes P value < 0.0001; ** denotes P value < 0.001.   64 It has been shown that the p24 fragment maintains its nucleolar localization, while p89 interacts with intact PARP-1 and blocks the PARP homodimerization, which is essential for enzyme activity. 65 Western blot analysis showed that the main markers of apoptosis such as cleaved PARP1, PUMA-a, and cleavedcaspase 9 and Bcl-2 were up-regulated and down-regulated in ANM treated HCT cells (Fig. 4).  Here *** denotes P value < 0.0001; ** denotes P value < 0.001, and ns indicates non-significant.

In vitro wound healing assay of crude methanolic extract of A. nallamalayana
The study of cell migration is of particular importance in cancer, as metastatic progression is the primary cause of death in cancer patients. Cancer can grow and spread all across the body only if cancer cells can migrate and invade via extracellular matrix (ECM) and intravasate into the bloodstream, binding to a distant site and eventually extravasate to form distant foci. 66,67 The scratch wound assay was used to assess cell migration, a crucial step in forming metastatic foci. The scratch wound assay was carried out to detect ANM's inhibitory effect on HCT 116 cell migration. Aer treatment with ANM (IC 50 ), the HCT 116 cell migration was decreased in a time-dependent manner (0, 12, 24 and 48 h). The result showed fewer or no cells in the denuded region, indicating that ANM could reduce site-specic cell migration (Fig. 5). Decreased migration in HCT 116 cells could be described as decreased metastatic potential. Over cell proliferation and migration are hallmarks of cancerous cells. 68 The effectiveness of prospective cancer therapies is systematically estimated using in vitro cell-line proliferation screens. However, it is not clear whether tumour aggressiveness is more affected by the proliferative or migratory properties of cancer cells that make the therapy ineffective. 69 Thus, inhibition of cell proliferation and migration is considered necessary in order to treat cancer effectively. 70 3.6 Total phenolic and avonoid content in crude methanolic extract of A. nallamalayana and A. paniculata Phenolic and avonoid compounds act as antioxidants due to their redox properties. Total phenolic content could be used as a basis for rapid antioxidant screening because of hydroxyl  groups in phenolic compounds that facilitate free radical scavenging. 71 Total phenolic content was determined using the Folin-Ciocalteu method in each extract. The results were derived from a standard calibration curve (y ¼ 0.048x + 0.063, R 2 ¼ 0.987) of gallic acid (25 to 1000 mg ml À1 ) and expressed in gallic acid equivalents (GAE) per gram dry extract weight. Aluminium chloride colorimetric method was used to measure the avonoids content in each methanolic extract. The results were derived from the calibration curve (y ¼ 0.063x + 0.131, R 2 ¼ 0.970) of quercetin (0-500 mg ml À1 ) and expressed in quercetin equivalents (QE) per gram dry extract weight. The results were resumed in Table 2. The total phenolic content was found to be lower in ANM compared to APM while total avonoid content was higher in ANM than APM. Phenolic and avonoid are one of the most widely distributed secondary metabolites in the plant kingdom. Their anti-carcinogenic effects are primarily due to their ability to: (a) induce cell cycle arrest; 72,73 (b) inhibit oncogenic signalling cascades controlling cell proliferation, angiogenesis, and apoptosis; 74-77 (c) modulate ROS levels; 78-80 (d) promote tumour suppressor proteins such as p53; 81,82 and (e) halt cell migration. 83,84 Multiple studies clearly suggest that the anticancer and apoptosis-inducing properties of polyphenolic compounds is mainly due to their prooxidant action rather than antioxidant activity. 85 Flavonoids have a dual effect in terms of ROS homeostasis, acting as antioxidants under normal conditions and as powerful pro-oxidants in cancer cells, activating apoptotic pathways. 85,86 Because of the presence of phenolic hydroxyl groups, avonoids may directly scavenge ROS and chelate metal ions. 87,88 The indirect antioxidant effects of avonoids are associated with the activation of antioxidant enzymes, the repression of pro-oxidant enzymes, and stimulate the production of antioxidant enzyme and phase II detoxifying enzyme synthesis. 88 The anticancer activity of avonoid is mediated by both antioxidant and pro-oxidant activity. 89 The high avonoid and phenolic content could be responsible for the cytotoxicity of these crude extracts.

Metabolite proling by UPLC-QTOF-MS (HRMS) analysis of methanolic extract of A. nallamalayana and A. paniculata aerial parts
Since the phytochemical analysis showed that the extracts were rich in phenolic and avonoid contents, they were used to identify and characterise metabolites using UPLC-QTOF-MS analysis. Accurate mass values (m/z) of all the primary ions identied in UPLC-MS analysis were screened against databases such as Metlin, MassBank and HMDB and literature within ve ppm accuracy. Peak identication was carried out by matching retention times (Rt) and mass spectra with literature data and databases. The comparative phytochemical investigation revealed that both species have different chemical constituents. In UPLC-ESI-QTOF-MS analysis, 42 compounds were identied with andrographolides as the major constituents of A. paniculata, whereas a total of 59 compounds were identied from the methanolic extract of A. nallamalayana.
Most of the compounds were identied for the rst time from this species. Among all the identied compounds from both species, eight compounds were similar, i.e. chlorogenic acid, andrographidine B, 1,3-dicaffeoylquinic acid, apigenin 7-O-bglucuronide, andrographidine D, andropaniculoside A, skull-capavone I, oroxylin A (ESI Fig. 2 †). The phytochemicals characterisation revealed that the identied compounds belong to phenolic acids, diterpenoids, avonoids, and their glycosides. The names of the identied compounds, molecular formula, experimental mass (m/z), peak height, the retention time (min), score and adduct/ion species are summarised in ESI Tables S1 and S2. †  Table 3. 3.8.1 Identication of phenolic acids. In the methanolic extract of A. nallamalayana, four caffeoylquinic acids were identied, but their correct identication is a rather tricky job due to widespread isomerism (geometrical and regional). In quinic acid, the linkage position of caffeoyl groups for monoacyl caffeoylquinic acids could be identied based on their characteristic fragmentation pathways. Diagnostic fragmentation ions (DFIs), e.g. m/z 173, 179 and 191 corresponds to [quinic acid-H-H 2 O], [caffeic acid-H] À and [quinic acid-H] À respectively, were suggested or calculated from the analysis of fragmentation pattern for each chemical class of chlorogenic acids (CGAs). 91 Compound 1 (peak 1) displayed a deprotonated molecular ion peak at m/z 353.09 [M-H] À . In mass fragmentation, the base peak at m/z 191.0556 was obtained from the moiety of the quinic acid, [quinic acid-H] À . Additionally, the secondary peak at m/z 179.0344 (C 9 H 7 O 4 ) was derived from the moiety of caffeic acid, [caffeic acid-H] À , together with a daughter ion at m/z 161.025. Compound 1, based on the fragmentation patterns and pseudo molecular ion at m/z 353 in the MS/MS experiment, was tentatively characterised and identied as chlorogenic acid. 92 No distinct difference was observed in the MS/MS spectra of compound 1 and compound 2, but a secondary peak at m/z 135.0448 (C 8 H 7 O 2 ), [caffeic acid-H-CO 2 ] À displayed by compound 1, which was absent in the spectra of compound 2. Based on the fragmentation pattern and previous literature reports, compound 2 (peak 2) was tentatively identied as 1-Ocaffeoylquinic acid. 93  According to the fragmentation pattern and literature data compound, 10 was tentatively iden-tied as 3-p-coumaroylquinic acid. 95 3.8.2 Identication of avones. Compound 12 (peak 12) displayed a deprotonated molecular ion peak at m/z 581.1331 [M-H] À . The fragmentation pattern suggests that this compound may be a combination of orientin and vitexin derivatives because the retention time and molecular masses were identical. The fragment ions produced were found to be identical to those of the type II avone C-glycosides. Compounds 12 were tentatively classied as 2 00 -O-vanilloylvitexin based on earlier literature studies. 96   HCO loss, respectively. The peak at m/z 117 corresponds to the generation of the B-ring fragments. Fragment ions formed by Aring at m/z 165 and m/z 121 indicate that methoxyl substituent occurs at the 8 th position. Finally, compound 20 was tentatively identied as 7,2 0 -dihydroxy-5-methoxyavone. Compound 5 was tentatively identied as hispidulin 7-glucoside (homoplantaginin). 104 Similarly, Compound 6 (peak 6) displayed a deprotonated molecular ion at m/z 447 [M-H] À . The mass spectrum showed the high abundance of fragments at m/z 285 [M-H-162] À is due to the loss of a hexose unit. The deprotonated aglycone fragment at m/z 285 suggested that it was originated from luteolin or kaempferol. Characteristic fragments at m/z 175, 151 and 133 conrmed luteolin as aglycone. Thus compound 6 was tentatively designated as luteolin 4 0 -glucoside. 105 The molecular ion peak at m/z 567 [M-H] À along with the characteristic fragment ion at m/z 285 [luteolin-H] À supports the previously proposed structure. Compound 7 (peak 7) was, thus, tentatively identied as neobignonoside [luteolin-7-O-(6 00 -p-hydroxybenzoyl-b-D-glucopyranoside)] from its negative ESI-MS/MS analysis. 106  The structures of three known compounds were characterised as echioidinin 113 (compound 1), skullcapavone I 114 (compound 2), and 5,2 0 ,6 0 -trihydroxy-7-methoxyavone 2 0 -O-b-D-glucopyranoside 108 (compound 3) based on their single-crystal X-ray diffraction (XRD) ( Table 4) and comparison of their spectral data with literature ( Table  5). The crystal structures of echioidinin (CCDC deposition no. 2072153); skullcapavone I (CCDC deposition no. 2072155), and 5,2 0 ,6 0 -trihydroxy-7-methoxyavone 2 0 -O-b-D-glucopyranoside (CCDC deposition no. 2072714) were reported for the rst time (Fig. 8). The crystal of compound 3 was twinned and treated accordingly with HKL5 format. The nal residual factors or discrepancy indices (R 1 values) of compound 3 was 6.95% which was due to the limited quality of the data. All crystals' thermal ellipsoid plot was represented in ESI along with the CheckCif alerts (Fig. S5, S11, and S17). †

HPLC-UV analysis of methanolic leaf extract of A. nallamalayana
A simple RP HPLC method with a gradient of acetonitrile and water was used for the simultaneous identication of secondary metabolites of A. nallamalayana. The HPLC-UV chromatogram of methanolic leaf extract of A. nallamalayana showed 13 major  and minor peaks (Fig. 9). Seven major peaks, peaks 1 and 2, respectively, identied as phenolic acids that correspond to chlorogenic acid (l ¼ 217, 240sh, 324 nm, rt: 3.08 min), and 3, HPLC-UV analysis revealed that avonoids and phenolics were the main components of methanolic leaf extract of A. nallamalayana, which were also characterised by UPLC-QTOF-MS/MS analysis. Flavonoids are the most studied class of plant's secondary metabolites with well-dened physical and chemical properties. Flavonoids give a characteristic UV absorption pattern, making their UV/PDA spectra very distinctive and UV spectroscopy a preferred analytical tool for identication. 115 Two characteristic bands observed in UV spectra of avonoids, band I (l max 300-380 nm) is caused by ring B absorption, while band II (l max 240-280 nm) is caused by ring A absorption. These bands' location provides details on the class of avonoids as well as their substitution pattern; hence, UV spectroscopy has been used as the primary tool for the quantication and identication of avonoids for years. 116 All peaks were identied by direct comparison (retention time and UV-spectra) with authentic commercial standards and isolated compounds.
3.11 Prediction of the in silico biological activity 3.11.1 In silico anticancer activity prediction using PASS program. In order to mitigate the complexity and expenses of experimental in vivo screening of anticancer agents using tens of millions of natural and synthetic chemical compounds, in silico phenotypic screening methods are required. 50 We used the successfully reported PASS (Prediction of Activity Spectra for Substances) algorithm to predict the anticancer activity of all the compounds identied from the methanolic extract of A. nallamalayana. 20 compounds were analysed by the PASS program for antitumor effects. The results obtained by the PASS prediction are shown in ESI (Table S3). † Among the screened compounds, 17 compounds showed signicant probable anticancer/antineoplastic activity (Pa $ 0.9), whereas three compounds were inactive (Pa < 0.9). Isorhamnetin 3-glucoside showed the highest Pa values (0.974/0.001), whereas chlorogenic acid showed the minimum Pa value (0.846/0.004). The majority of compounds belong to phenolic acids, avonoids and their glucoside which are well known for their anticancer activities. [117][118][119] As evident from the ndings, the scores for probable activity (Pa) were very close to 1, and the scores for probable inactivity (Pi) were very close to 0, indicating that these compounds are highly likely to be active in the in vitro/in vivo studies.
3.11.2 In silico cell line cytotoxicity prediction using CLC-Pred tool. CLC-Pred, a well-known tool used in chemoinformatics and medicinal chemistry to predict the in silico cytotoxicity for tumour and non-tumour cell lines, was used to predict the cytotoxicity of the identied compounds. The estimated results that have been presented in Pa values, which is >0.5, are probably more active with the predicted cancer cell line. From the 20 compounds selected, 14 compounds showed aspirated outcome, and barely six compounds displayed negative results (Table S4 †). The cytotoxicity represented by the compounds identied from the methanolic extract of A. nallamalayana matches the present study and literature survey. [119][120][121]

Conclusion
In the present study, the methanolic extract of two Andrographis species, A. nallamalayana and A. paniculata, showed signicant cytotoxicity towards three different cancer cell in a dosedependent manner. The cytotoxicity of ANM was nearly four times higher than APM in HCT 116 and HepG2 cells, whereas both extracts showed comparatively similar cytotoxicity in A549 cells and no or very less cytotoxicity in HEK 293 cells. Furthermore, ANM induced cell death involves apoptotic changes and inhibition of migration, ROS generation, up-regulation and down-regulation of main apoptotic markers as seen in HCT 116 cells. Although both species showed promising cytotoxic activity, the comparative phytochemical investigation revealed that both species had different chemical constituents. Both species were found to contain signicant quantities of phenolics and avonoids. Using UPLC-QTOF-MS (HRMS) analysis, andrographolides were identied as the major compounds of A. paniculata. Interestingly, andrographolides were not found in A. nallamalayana.
Further, using the MS/MS fragmentation approach, 20 compounds were characterized/identied from A. nallamalayana; out of 20, 18 compounds were identied for the rst time from this species. Three known compounds, echioidinin, skullcapavone I and 5,2 0 ,6 0 -trihydroxy-7-methoxyavone 2 0 -Ob-D-glucopyranoside, were isolated from A. nallamalayana and their crystal structures were reported for the rst time. Subsequently, seven major compounds were identied in A. nallamalayana by direct comparison (retention time and UV-spectra) with authentic commercial standards and isolated compounds using HPLC-UV analysis. The prediction of anticancer activity using in silico tools also justies the evaluation of the in vitro cytotoxic activity. Our experimental studies have validated the traditional use of A. nallamalayana and A. paniculata as an anticancer herbal drug. However, more studies are required to explore the role of A. nallamalayana in different in vivo cancer models so that it can contribute to the successful treatment of cancer in future.

Conflicts of interest
There are no conicts to declare.