NMR-Based Metabolomics for Geographical Discrimination of Adhatoda vasica Leaves

Adhatoda vasica (L.), Nees is a widespread plant in Asia. It is used in Ayurvedic and Unani medications for the management of various infections and health disorders, especially as a decoction to treat cough, chronic bronchitis, and asthma. Although it has a diverse metabolomic profile, this plant is particularly known for its alkaloids. The present study is the first to report a broad range of present compounds, e.g., α-linolenic acid, acetate, alanine, threonine, valine, glutamate, malate, fumaric acid, sucrose, β-glucose, kaempferol analogues, quercetin analogues, luteolin, flavone glucoside, vasicine and vasicinone, which were identified by NMR spectroscopy-based metabolomics. Multivariate data analysis was used to analyze 1H-NMR bucketed data from a number of Adhatoda vasica leave samples collected from eight different regions in Pakistan. The results showed large variability in metabolomic fingerprints. The major difference was on the basis of longitude/latitude and altitude of the areas, with both primary and secondary metabolites discriminating the samples from various regions.


Introduction
Adhatoda vasica (L.), Nees is a well-known medicinal plant in Ayurvedic and Unani medicine. It has been used for the treatment of various disorders, such as for respiratory tract ailments [1], radiomodulation [2], hypoglycaemic, cardiovascular prevention, antituberculosis, antiviral, hepatoprotective, antimutagenic and antioxidant [3]. The frequent use of A. vasica has resulted in its inclusion in the WHO manual "The Use of Traditional Medicine in Primary Health Care" which is intended for health workers in Southeast Asia to keep them informed on the utility of their surrounding flora; in the case of A. vasica, this concerns the treatment of cough, asthma and bleeding piles. This plant can be used both for adults and children for prolonged periods [4]. The Vasaka plant is known for pyrroquinazoline alkaloids, such as vasicine, vasicol and vasicinone. Vasicine, the major alkaloid, was shown to have acetylcholinesterase inhibitory activity [5].
The herbal medicine industry uses decoction to extract A. vasica leaves. This single plant extract contains hundreds of organic chemicals, and it will be difficult to characterize a single active ingredients from the complex decoction mixture [6]; even the occurrence of several compounds may interact for having the phytotherapeutic effect. Herbal medicines are sourced from a number different geographical locations, which means that they may have qualitative and quantitative differences in the spectra of compounds present in the plant, i.e., in the metabolome [7,8]. The identification of the active ingredient(s) and A total of sixteen compounds were identified: α-linolenic acid, acetate, alanine, threonine, valine, glutamate, malate, fumarate, sucrose, β-glucose, kaempferol analogues, quercetin analogues, luteolin, flavone glucoside, vasicine and vasicinone ( Table 2). All these compounds were confirmed through 1D and 2D NMR spectral analysis and comparison with the previously reported data as well. Vasicine and vasicinone are the most important known bioactive compounds of this plant, and they could be detected and identified in the 1 H-NMR spectrum of the crude extract by overlay the 1 H-NMR of the purified alkaloid fraction ( Figure 1) and by comparison with previously reported data of the pure alkaloids from A. vasica [17]. Table 2. 1 H chemical shifts (δ) and coupling constants (Hz) of metabolite of Vasaka leaves obtained using 1D and 2D NMR experiments (CD 3 OD-KH 2 PO 4 in D 2 O (pH 6.0).
Alanine  The NMR-based metabolomics coupled with multivariate data analysis has proved its strength for quality control that was limited to detection and/or quantification of targeted compounds [18]; however, the methodology as followed here has previously proved the similar results of chemometric methods and official conventional methods for the targeted compounds as well, and so no further separation is suggested [19,20]. The NMR-based metabolomics coupled with multivariate data analysis has proved its strength for quality control that was limited to detection and/or quantification of targeted compounds [18]; however, the methodology as followed here has previously proved the similar results of chemometric methods and official conventional methods for the targeted compounds as well, and so no further separation is suggested [19,20].
The 1 H-NMR data were bucketed and analyzed by various multivariate data analysis methods: hierarchical cluster analysis (Figure 2), PCA ( Figure 3) and PLS-DA ( Figure 4). The cluster analysis clearly grouped all regions, proving a clear discrimination of all samples, but the series of samples from various regions was not in the sequence of their altitude (Table 1), which proves that this factor alone does not fully explain the clustering of plant samples. When various regions were aligned on the basis of longitude and latitude (the samples grouped on the basis of their region, especially longitude ( Figure 2, Table 2), we could draw a pattern in alterations up to metabolite level, accordingly (Figures 2-4). Similarly, the altitude of regions did not show a more defined relationship in our findings; however, it revealed an impact. The 1 H-NMR data were bucketed and analyzed by various multivariate data analysis methods: hierarchical cluster analysis (Figure 2), PCA ( Figure 3) and PLS-DA ( Figure 4). The cluster analysis clearly grouped all regions, proving a clear discrimination of all samples, but the series of samples from various regions was not in the sequence of their altitude (Table 1), which proves that this factor alone does not fully explain the clustering of plant samples. When various regions were aligned on the basis of longitude and latitude (the samples grouped on the basis of their region, especially longitude ( Figure 2, Table 2), we could draw a pattern in alterations up to metabolite level, accordingly (Figures 2-4). Similarly, the altitude of regions did not show a more defined relationship in our findings; however, it revealed an impact.      Here we report that the metabolomic composition of all samples from various regions is qualitatively similar but have clear discrimination in PCA ( Figure 3) and PLS-DA ( Figure 4) that reveals the quantitative differences among metabolites, including the  Here we report that the metabolomic composition of all samples from var is qualitatively similar but have clear discrimination in PCA ( Figure 3) a ( Figure 4) that reveals the quantitative differences among metabolites, in characteristic alkaloids of A. vasica. As in industrial applications, the plan collected from various geographic locations, producing the similar total yiel applications, but the biologically active ingredients matter the most, so these especially in case of bioactive secondary metabolites, i.e., the lower amount in case of A. vasica, impart negative impact on the nutraceutical quality of plant, which is a major concern for their use as medicine [21], while t metabolites, i.e., sugars and others, may also impart an impact on taste characteristics.

Vasicine
Studying the outcomes in detail reveals that the A. vasica samples from M (3) and Abbottabad (1) regions are grouped closely together in hierarchica ( Figure 2). This is also evident in principal component analysis ( Figure 3) and squares discriminant analysis ( Figure 4). Discrimination from other samples signals of α-linolenic acid, alanine, threonine and valine. It is evident (Figur the samples from Khanpur (4), Havalian (6) and Rawalpindi (5) were aligne other and clearly discriminated from others due to the signals of luteo glucoside, vasicine and vasicinone.
Although the samples collected from Taxila (3) were identified simi collected from Khanpur (4), Rawalpindi (5) and Havelian (6), as in hierarchic ( Figure 2), but these kept their identity within the same group by discrimina PLS-DA ( Figure 4) having higher signals for Kaempferol analogues an analogues. As aforementioned, the samples collected from Khanpur (4), Ra and Havelian (6) identified for luteolin and flavone glucoside as di metabolites. However, in the case of alkaloids, i.e., vasicine and vasicinone, i be the discriminating factor for the plants collected from aforementioned area plants from all these areas appear in similar groups as observed in hierarchic ( Figure 2). rt that the metabolomic composition of all samples from various regions ilar but have clear discrimination in PCA ( Figure 3) and PLS-DA veals the quantitative differences among metabolites, including the loids of A. vasica. As in industrial applications, the plant material is ious geographic locations, producing the similar total yield industrial he biologically active ingredients matter the most, so these differences, of bioactive secondary metabolites, i.e., the lower amount of alkaloids a, impart negative impact on the nutraceutical quality of a medicinal major concern for their use as medicine [21], while the primary ugars and others, may also impart an impact on taste and other utcomes in detail reveals that the A. vasica samples from Muzaffarabad d (1) regions are grouped closely together in hierarchical clustering lso evident in principal component analysis ( Figure 3) and partial least nt analysis ( Figure 4). Discrimination from other samples is due to the nic acid, alanine, threonine and valine. It is evident (Figures 2-4) that Khanpur (4), Havalian (6) and Rawalpindi (5) were aligned with each discriminated from others due to the signals of luteolin, flavone and vasicinone. samples collected from Taxila (3) were identified similar to those npur (4), Rawalpindi (5) and Havelian (6), as in hierarchical clustering se kept their identity within the same group by discriminating itself in 4) having higher signals for Kaempferol analogues and quercetin ementioned, the samples collected from Khanpur (4), Rawalpindi (5) ) identified for luteolin and flavone glucoside as discriminated ver, in the case of alkaloids, i.e., vasicine and vasicinone, it appears to ng factor for the plants collected from aforementioned areas, and so the se areas appear in similar groups as observed in hierarchical clustering re we report that the metabolomic composition of all samples from various regions itatively similar but have clear discrimination in PCA ( Figure 3) and PLS-DA 4) that reveals the quantitative differences among metabolites, including the eristic alkaloids of A. vasica. As in industrial applications, the plant material is d from various geographic locations, producing the similar total yield industrial tions, but the biologically active ingredients matter the most, so these differences, lly in case of bioactive secondary metabolites, i.e., the lower amount of alkaloids of A. vasica, impart negative impact on the nutraceutical quality of a medicinal which is a major concern for their use as medicine [21], while the primary lites, i.e., sugars and others, may also impart an impact on taste and other eristics. dying the outcomes in detail reveals that the A. vasica samples from Muzaffarabad Abbottabad (1) regions are grouped closely together in hierarchical clustering 2). This is also evident in principal component analysis ( Figure 3) and partial least discriminant analysis (Figure 4). Discrimination from other samples is due to the of α-linolenic acid, alanine, threonine and valine. It is evident (Figures 2-4) that ples from Khanpur (4), Havalian (6) and Rawalpindi (5) were aligned with each nd clearly discriminated from others due to the signals of luteolin, flavone de, vasicine and vasicinone. though the samples collected from Taxila (3) were identified similar to those d from Khanpur (4), Rawalpindi (5) and Havelian (6), as in hierarchical clustering 2), but these kept their identity within the same group by discriminating itself in (Figure 4) having higher signals for Kaempferol analogues and quercetin es. As aforementioned, the samples collected from Khanpur (4), Rawalpindi (5) avelian (6) identified for luteolin and flavone glucoside as discriminated lites. However, in the case of alkaloids, i.e., vasicine and vasicinone, it appears to iscriminating factor for the plants collected from aforementioned areas, and so the rom all these areas appear in similar groups as observed in hierarchical clustering 2).  Here we report that the metabolomic composition of all samples from various regions is qualitatively similar but have clear discrimination in PCA ( Figure 3) and PLS-DA ( Figure 4) that reveals the quantitative differences among metabolites, including the characteristic alkaloids of A. vasica. As in industrial applications, the plant material is collected from various geographic locations, producing the similar total yield industrial applications, but the biologically active ingredients matter the most, so these differences, especially in case of bioactive secondary metabolites, i.e., the lower amount of alkaloids in case of A. vasica, impart negative impact on the nutraceutical quality of a medicinal plant, which is a major concern for their use as medicine [21], while the primary metabolites, i.e., sugars and others, may also impart an impact on taste and other characteristics.
Studying the outcomes in detail reveals that the A. vasica samples from Muzaffarabad (3) and Abbottabad (1) regions are grouped closely together in hierarchical clustering ( Figure 2). This is also evident in principal component analysis ( Figure 3) and partial least squares discriminant analysis (Figure 4). Discrimination from other samples is due to the signals of α-linolenic acid, alanine, threonine and valine. It is evident (Figures 2-4) that the samples from Khanpur (4), Havalian (6) and Rawalpindi (5) were aligned with each other and clearly discriminated from others due to the signals of luteolin, flavone glucoside, vasicine and vasicinone.
Although the samples collected from Taxila (3) were identified similar to those collected from Khanpur (4), Rawalpindi (5) and Havelian (6), as in hierarchical clustering (Figure 2), but these kept their identity within the same group by discriminating itself in PLS-DA (Figure 4) having higher signals for Kaempferol analogues and quercetin analogues. As aforementioned, the samples collected from Khanpur (4), Rawalpindi (5) and Havelian (6) identified for luteolin and flavone glucoside as discriminated metabolites. However, in the case of alkaloids, i.e., vasicine and vasicinone, it appears to be the discriminating factor for the plants collected from aforementioned areas, and so the plants from all these areas appear in similar groups as observed in hierarchical clustering (Figure 2).  Here we report that the metabolomic composition of all samples from various regions is qualitatively similar but have clear discrimination in PCA ( Figure 3) and PLS-DA ( Figure 4) that reveals the quantitative differences among metabolites, including the characteristic alkaloids of A. vasica. As in industrial applications, the plant material is collected from various geographic locations, producing the similar total yield industrial applications, but the biologically active ingredients matter the most, so these differences, especially in case of bioactive secondary metabolites, i.e., the lower amount of alkaloids in case of A. vasica, impart negative impact on the nutraceutical quality of a medicinal plant, which is a major concern for their use as medicine [21], while the primary metabolites, i.e., sugars and others, may also impart an impact on taste and other characteristics.
Studying the outcomes in detail reveals that the A. vasica samples from Muzaffarabad (3) and Abbottabad (1) regions are grouped closely together in hierarchical clustering ( Figure 2). This is also evident in principal component analysis ( Figure 3) and partial least squares discriminant analysis (Figure 4). Discrimination from other samples is due to the signals of α-linolenic acid, alanine, threonine and valine. It is evident (Figures 2-4) that the samples from Khanpur (4), Havalian (6) and Rawalpindi (5) were aligned with each other and clearly discriminated from others due to the signals of luteolin, flavone glucoside, vasicine and vasicinone.
Although the samples collected from Taxila (3) were identified similar to those collected from Khanpur (4), Rawalpindi (5) and Havelian (6), as in hierarchical clustering ( Figure 2), but these kept their identity within the same group by discriminating itself in PLS-DA ( Figure 4) having higher signals for Kaempferol analogues and quercetin analogues. As aforementioned, the samples collected from Khanpur (4), Rawalpindi (5) and Havelian (6) identified for luteolin and flavone glucoside as discriminated metabolites. However, in the case of alkaloids, i.e., vasicine and vasicinone, it appears to be the discriminating factor for the plants collected from aforementioned areas, and so the plants from all these areas appear in similar groups as observed in hierarchical clustering (Figure 2).  Here we report that the metabolomic composition of all samples from various regions is qualitatively similar but have clear discrimination in PCA ( Figure 3) and PLS-DA ( Figure 4) that reveals the quantitative differences among metabolites, including the characteristic alkaloids of A. vasica. As in industrial applications, the plant material is collected from various geographic locations, producing the similar total yield industrial applications, but the biologically active ingredients matter the most, so these differences, especially in case of bioactive secondary metabolites, i.e., the lower amount of alkaloids in case of A. vasica, impart negative impact on the nutraceutical quality of a medicinal plant, which is a major concern for their use as medicine [21], while the primary metabolites, i.e., sugars and others, may also impart an impact on taste and other characteristics.
Studying the outcomes in detail reveals that the A. vasica samples from Muzaffarabad (3) and Abbottabad (1) regions are grouped closely together in hierarchical clustering ( Figure 2). This is also evident in principal component analysis ( Figure 3) and partial least squares discriminant analysis (Figure 4). Discrimination from other samples is due to the signals of α-linolenic acid, alanine, threonine and valine. It is evident (Figures 2-4) that the samples from Khanpur (4), Havalian (6) and Rawalpindi (5) were aligned with each other and clearly discriminated from others due to the signals of luteolin, flavone glucoside, vasicine and vasicinone.
Although the samples collected from Taxila (3) were identified similar to those collected from Khanpur (4), Rawalpindi (5) and Havelian (6), as in hierarchical clustering (Figure 2), but these kept their identity within the same group by discriminating itself in PLS-DA (Figure 4) having higher signals for Kaempferol analogues and quercetin analogues. As aforementioned, the samples collected from Khanpur (4), Rawalpindi (5) and Havelian (6) identified for luteolin and flavone glucoside as discriminated metabolites. However, in the case of alkaloids, i.e., vasicine and vasicinone, it appears to be the discriminating factor for the plants collected from aforementioned areas, and so the plants from all these areas appear in similar groups as observed in hierarchical clustering (Figure 2).  Here we report that the metabolomic composition of all samples from various is qualitatively similar but have clear discrimination in PCA ( Figure 3) and (Figure 4) that reveals the quantitative differences among metabolites, includ characteristic alkaloids of A. vasica. As in industrial applications, the plant ma collected from various geographic locations, producing the similar total yield in applications, but the biologically active ingredients matter the most, so these dif especially in case of bioactive secondary metabolites, i.e., the lower amount of a in case of A. vasica, impart negative impact on the nutraceutical quality of a m plant, which is a major concern for their use as medicine [21], while the metabolites, i.e., sugars and others, may also impart an impact on taste an characteristics.
Studying the outcomes in detail reveals that the A. vasica samples from Muza (3) and Abbottabad (1) regions are grouped closely together in hierarchical cl ( Figure 2). This is also evident in principal component analysis ( Figure 3) and par squares discriminant analysis (Figure 4). Discrimination from other samples is d signals of α-linolenic acid, alanine, threonine and valine. It is evident (Figures 2 the samples from Khanpur (4), Havalian (6) and Rawalpindi (5) were aligned w other and clearly discriminated from others due to the signals of luteolin, glucoside, vasicine and vasicinone.
Although the samples collected from Taxila (3) were identified similar collected from Khanpur (4), Rawalpindi (5) and Havelian (6), as in hierarchical cl (Figure 2), but these kept their identity within the same group by discriminating PLS-DA (Figure 4) having higher signals for Kaempferol analogues and q analogues. As aforementioned, the samples collected from Khanpur (4), Rawal and Havelian (6) identified for luteolin and flavone glucoside as discri metabolites. However, in the case of alkaloids, i.e., vasicine and vasicinone, it ap be the discriminating factor for the plants collected from aforementioned areas, an plants from all these areas appear in similar groups as observed in hierarchical cl (Figure 2). report that the metabolomic composition of all samples from various regions ly similar but have clear discrimination in PCA ( Figure 3) and PLS-DA at reveals the quantitative differences among metabolites, including the alkaloids of A. vasica. As in industrial applications, the plant material is various geographic locations, producing the similar total yield industrial but the biologically active ingredients matter the most, so these differences, case of bioactive secondary metabolites, i.e., the lower amount of alkaloids vasica, impart negative impact on the nutraceutical quality of a medicinal is a major concern for their use as medicine [21], while the primary i.e., sugars and others, may also impart an impact on taste and other s.
the outcomes in detail reveals that the A. vasica samples from Muzaffarabad ttabad (1) regions are grouped closely together in hierarchical clustering is is also evident in principal component analysis ( Figure 3) and partial least iminant analysis (Figure 4). Discrimination from other samples is due to the inolenic acid, alanine, threonine and valine. It is evident (Figures 2-4) that rom Khanpur (4), Havalian (6) and Rawalpindi (5) were aligned with each early discriminated from others due to the signals of luteolin, flavone sicine and vasicinone. h the samples collected from Taxila (3) were identified similar to those Khanpur (4), Rawalpindi (5) and Havelian (6), as in hierarchical clustering t these kept their identity within the same group by discriminating itself in ure 4) having higher signals for Kaempferol analogues and quercetin s aforementioned, the samples collected from Khanpur (4), Rawalpindi (5) n (6) identified for luteolin and flavone glucoside as discriminated owever, in the case of alkaloids, i.e., vasicine and vasicinone, it appears to inating factor for the plants collected from aforementioned areas, and so the ll these areas appear in similar groups as observed in hierarchical clustering ) and numbering is based on latitude.
Here we report that the metabolomic composition of all samples from various regions is qualitatively similar but have clear discrimination in PCA ( Figure 3) and PLS-DA ( Figure 4) that reveals the quantitative differences among metabolites, including the characteristic alkaloids of A. vasica. As in industrial applications, the plant material is collected from various geographic locations, producing the similar total yield industrial applications, but the biologically active ingredients matter the most, so these differences, especially in case of bioactive secondary metabolites, i.e., the lower amount of alkaloids in case of A. vasica, impart negative impact on the nutraceutical quality of a medicinal plant, which is a major concern for their use as medicine [21], while the primary metabolites, i.e., sugars and others, may also impart an impact on taste and other characteristics.
Studying the outcomes in detail reveals that the A. vasica samples from Muzaffarabad (3) and Abbottabad (1) regions are grouped closely together in hierarchical clustering ( Figure 2). This is also evident in principal component analysis ( Figure 3) and partial least squares discriminant analysis (Figure 4). Discrimination from other samples is due to the signals of α-linolenic acid, alanine, threonine and valine. It is evident (Figures 2-4) that the samples from Khanpur (4), Havalian (6) and Rawalpindi (5) were aligned with each other and clearly discriminated from others due to the signals of luteolin, flavone glucoside, vasicine and vasicinone.
Although the samples collected from Taxila (3) were identified similar to those collected from Khanpur (4), Rawalpindi (5) and Havelian (6), as in hierarchical clustering ( Figure 2), but these kept their identity within the same group by discriminating itself in PLS-DA ( Figure 4) having higher signals for Kaempferol analogues and quercetin analogues. As aforementioned, the samples collected from Khanpur (4), Rawalpindi (5) and Havelian (6) identified for luteolin and flavone glucoside as discriminated metabolites. However, in the case of alkaloids, i.e., vasicine and vasicinone, it appears to be the discriminating factor for the plants collected from aforementioned areas, and so the plants from all these areas appear in similar groups as observed in hierarchical clustering (Figure 2).
Two locations, i.e., Haripur (2) and Gazi (1), were grouped near each other and were discriminated from others due to higher signals of β-glucose, sucrose, and fumaric acid. Longitude and altitude are contributing factors to define the environment of a region, maybe because of light (intensity, day length, UV) and temperature [22]. Besides this, it is evident that a difference in geology in geographical regions also imparts the metabolome, resulting in changes in the bioactivity of plants from the same species of plants [8]. In our results, the pattern of metabolomics discrimination is identified on the base of longitude, but it does not mean that it is the only variable impeding the biological processes in plants.
It is well documented that latitude is an important factor, such as in the case of a controlled experiment on Arabidopsis thaliana; a variation in plant size and relative growth rate (RGR) was observed along a latitudinal gradient. the plants at high latitudes are reported to have a smaller size for their seeds, cotyledon width and leaf area as compared to those from low latitudes [23]. Nevertheless, as it is the combination of altitude, latitude and longitude that effects the weather of a point on earth and so imparts metabolomic alteration in plants [24,25]. Unfortunately, the longitude of a position received less attention for biochemical alterations in plant materials; that is also important for specific positioning of a plant material.
Bioactive compounds occurring in plant material are part of a multi-component mixture, so their isolation, identification and structure elucidation remains a challenge [26]. Changes in such a complex matrix (including undesirable compounds) may alter the bioavailability and bio-efficacy of the active ingredients, e.g., through synergy [27]. Decoctions are used mainly as an extraction method for herbs in industry, so besides quantity of the desired active compound(s), the extraction efficiency of a diverse range of metabolites is also an important quality trait [28]. For example, the intracellular metabolite extraction efficiency was found to be dependent on the extraction method [29]. Based on our findings, it is recommended to use metabolomic analysis coupled with multivariate data analysis to define the plant material, i.e., the characteristics of the required plant material for an optimal reproducible pharmaceutical material and to describe protocols for the agricultural production and the subsequent extraction procedure to obtain this final product. Eventually the desired quality might be obtained by mixing different batches to arrive at the defined quality.

Collection of Plant Material
The plant leaves (fresh-looking top few leaves that were free from herbivory/infections and facing the sun side) were collected in the month of March from various areas, namely, Abbottabad (34 • (5) replications (each from different plant) from each location were collected. The botanical authentication was conducted at the Department of Botany, University of Malakand, Chakdara, Pakistan, and a voucher specimen (UOM/BGH/20/109) was deposited in the herbarium.

Drying of Leaves
A total of 50 g sample (leaves) for each replication was collected and dried in the shade for three days before further freezedried and stored in 50 mL polypropylene centrifuge tubes, until further used for extraction and/or NMR analysis.

Extraction of Leaves for Mixture Analysis and NMR Measurements
All of the solvents and reagents were purchased from Sigma-Aldrich, Germany. Extraction of freeze-dried plant material by using 50% methanol-d 4 in D 2 O (KH 2 PO 4 buffer, pH 6.0) containing 0.05% TSP (trimethyl silyl propionic acid sodium salt, w/v) and NMR measurements was carried out by using a 500 MHz Bruker DMX-500 NMR spectrometer (Bruker, Karlsruhe, Germany) operating at a proton NMR frequency of 500.13 MHz, with the same protocol as reported previously [18]. Compounds were identified by 1D ( Figure 1) and 2D (Figures S1-S3; Supplementary Data) NMR spectral analysis and confirmed by comparison with the previously reported data as well [30,31].

Extraction Alkaloids
All the solvents and reagents were purchased from Sigma-Aldrich, Germany. Extraction and isolation of alkaloid fraction was carried out as reported previously through liquid-liquid (acid/base) extraction followed by preparatory TLC for extraction of crude alkaloid fraction [32]. The freeze-dried ground leaves (1 kg) were extracted with ethanol (3 times with 3 L, each times), and the extract was evaporated to dryness that yielded a gummy green material which was further extracted with hot, double deionized water (3 times with 500 mL, each times), cooled and filtered; where the chlorophyll part was discarded, the aqueous solution was further extracted with chloroform, (3 times with 250 mL, each times). The aqueous layer was basified with 5% NaOH (pH 8-9) and further extracted with chloroform (CHCI 3 ) (3 times with 250 mL, each time). Furthermore, this chloroform layer was extracted with 5% HCI (3 times with 200 mL, each time), and the acidic solution was basified with NaOH and extracted once again with chloroform until the organic layer was free of alkaloids. The alkaloid fraction was separated gradually, dried, confirmed through Dragendorffs reagent and further analyzed by NMR to clearly identify various alkaloids in it, whereas this data set was also used further to support our findings in mixture analysis by NMR.

Computational Processing of NMR Data
The Mnova (v. 6.2.1-7569, Mesterlab Research S.L., Santiago de Compostela, Spain) software was used for 1 H-NMR and 2D NMR spectral analysis of the alkaloid fraction and the total leave extract. Results of previous reports on this plant were referenced for identification of metabolites [18,33]. Superimposing the NMR spectrum of the alkaloid fraction (as identified and mentioned above) on the leave extract spectrum was used to confirm the identification of the alkaloids in the crude leave extract.
As second step, the bucketing of the one-dimensional proton NMR spectrum was performed by the AMIX software (Bruker), where the spectrum was normalized manually with solvent peak (CD 3 OD) by hundred and binned to equal width (0.04) throughout the spectral width of δ 0.3-10.0, followed by transforming the bucketed files to an ASCII file. The scaling was conducted by total intensity, whereas regions of δ 4.75-4.9 and δ 3.28-3.34 were excluded from the analysis because of the residual signal of solvents [18,34]. The hierarchical clustering analysis principal component analysis (PCA) and partial least square-discriminant analysis (PLS-DA) were performed with the SIMCA-P software (v. 11.0, Umetrics, Umea, Sweden) by using unit variance method, respectively [13,35].

Conclusions
Our results are evident for the geographic alteration in the metabolomic profile of the studied plant material. The NMR-based analysis as non-targeted analysis of plant material, coupled with multivariate data analysis, effectively described the sample variability and clustering corresponding to metabolites contribution as specific to geographic region. The plants from Taxila (3), Khanpur (4), Rawalpindi (5) and Havelian (6) regions are discriminated due to the ingredients of interest from the A. vasica, especially the characteristic alkaloids of this plant, i.e., vasicine and vasicinone.
However, the plants collected from Taxila (3) were discriminated by kaempferol analogues and quercetin analogues from those of Khanpur (4), Rawalpindi (5) and Havelian (6) that were discriminated by luteolin and flavone glucoside. However, plants from all other regions were more or less discriminated by amino acids, organic acids, sugars, etc., in generic terms.
Geographic positioning of plant growth matters a lot, and plant materials can be tracked for their geographic positioning as well as bioactivity potential in industry through a metabolomic approach. This report proves that NMR metabolomic profiling coupled with multivariate data analysis has a great potential to be used for quality control of herbs, such as in this case of Adhatoda vasica. This also demonstrates that care and proper analysis must be conducted by herbal processors while importing herbs from different regions or countries.