Plasma metabolomics reveal the correlation of metabolic pathways and Prakritis of humans.

BACKGROUND: Ayurveda, an ancient Indian medicinal system, has categorized human body constitutions in three broad constitutional types (prakritis) i.e. Vata, Pitta and Kapha.
OBJECTIVES: Analysis of plasma metabolites and related pathways to classify Prakriti specific dominant marker metabolites and metabolic pathways.
MATERIALS AND METHODS: 38 healthy male individuals were assessed for dominant Prakritis and their fasting blood samples were collected. The processed plasma samples were subjected to rapid resolution liquid chromatography-electrospray ionization-quadrupole time of flight mass spectrometry (RRLC-ESI-QTOFMS). Mass profiles were aligned and subjected to multivariate analysis.
RESULTS: Partial least square discriminant analysis (PLS-DA) model showed 97.87% recognition capability. List of PLS-DA metabolites was subjected to permutative Benjamini-Hochberg false discovery rate (FDR) correction and final list of 76 metabolites with p < 0.05 and fold-change > 2.0 was identified. Pathway analysis using metascape and JEPETTO plugins in Cytoscape revealed that steroidal hormone biosynthesis, amino acid, and arachidonic acid metabolism are major pathways varying with different constitution. Biological Go processes analysis showed that aromatic amino acids, sphingolipids, and pyrimidine nucleotides metabolic processes were dominant in kapha type of body constitution. Fat soluble vitamins, cellular amino acid, and androgen biosynthesis process along with branched chain amino acid and glycerolipid catabolic processes were dominant in pitta type individuals. Vata Prakriti was found to have dominant catecholamine, arachidonic acid and hydrogen peroxide metabolomics processes.
CONCLUSION: The neurotransmission and oxidative stress in vata, BCAA catabolic, androgen, xenobiotics metabolic processes in pitta, and aromatic amino acids, sphingolipid, and pyrimidine metabolic process in kapha Prakriti were the dominant marker pathways.

Introduction

Ayurveda is an ancient Indian medicinal system and is in practice since 1500 BC. It is extensively used in India, besides the contemporary medicine introduced only 250 years back. In Ayurvedic philosophy, five elements combine in pairs to form three dynamic forces or interactions called ‘doshas’ which are considered as the governing principles of living beings [1]. The doshas are divided into three classes and are described, viz. vata, pitta and kapha. These doshas refer to functions of motion, digestion & metabolism, anabolic activity & accumulation. Ayurveda attributes these constitutional characteristics – Prakriti of an individual to the preponderance of certain “doshas”. All doshas are present in every human being; however, only one is dominant. On the basis of dominant dosha, Prakriti i.e. vata, pitta and kapha are determined. Prakriti can be further divided into seven sub-types or mixed types of Prakritis [1]. Ayurveda describes physical, physiological, psychological and behavioral traits, to determine Prakriti subtypes. As Prakritis are based on dosha, it describes predisposition of individual to disease and response to treatment. All these descriptions of Prakritis seem to be based on genetics, later on proved by several studies. Studies have correlated single nucleotide polymorphisms, inflammation and oxidative stress related genes with different Prakritis. Authors also claimed the correlation of genetic constitution and Prakritis [2], [3].

Ayurveda suggests that the origin of vata, pitta and kapha is motion, digestion and accumulation respectively. It shows that these functions are closely related to metabolic events in body and hence to phenotypic expressions. Moreover, instant energy reserves and neurotransmitters in the body are in the form of metabolites [4]. Therefore, as genome, metabolome of an individual can be directly correlated with its Prakriti [5], [6], in this study, metabolomic approach was used to develop discrimination models to assess the Prakritis of individuals and to identify the dominant metabolic pathways of different dominating Prakritis.

Materials and methods

Chemicals and reagents

Internal and external calibrants and ESI tuning solution were purchased from Agilent Technologies. Solvents used for sample processing and in LCMS assembly: acetonitrile, methanol, formic acid and water were of LC–MS grade (Sigma–Aldrich). Standards of metabolites were purchased from Sigma–Aldrich.

Ethics statement

Human Ethics Committee of the PDDYP Ayurveda College, Pune, India approved the study wide letter no. RRI/2011/HEC/2023 dated 18-02-2011. All volunteers signed an informed consent to be a part of this study. This clinical trial was registered with the Clinical Trials Registry-India (CTRI) vide registration no. CTRI/2016/08/007187.

Development of questionnaire for Prakriti assessment

A questionnaire for clinical phenotyping was designed on the basis of Ayurvedic literature on phenotypes and methods of Prakriti assessment (Table S1). The physiological as well as psychological assessment of volunteers was done. The phenotypic classification included criteria for defining anatomical features like body built, body frame, size and symmetry of body parts, physiology, physical endurance and aptitudes.

Study design and Prakriti assessment

On the basis of dominant Prakriti types out of 205, 38 healthy male individuals were selected for the study. Volunteers were free from diabetes, hyperlipidemia, hypertension, and other chronic systemic disorders, smoking and alcohol consumption habits. As per the assessment the volunteers were categorized in vata–pitta (VP), pitta–kapha (PK) and kapha–pitta (KP) dual Prakriti categories where the earlier term refers the dominant Prakriti. For the sake of simplicity the groups were labeled as vata (V), pitta (P) and kapha (K). Individuals were asked to take a decided vegetarian diet and restricted from any kind of medicine for one week before collection of blood samples. On seventh day morning, fasting blood samples were collected in K+-EDTA vacutainer tubes (BD). Plasma fractions were separated out by centrifugation at 3000g for 15 min at 15 °C. Plasma samples (100 μL) were mixed with equal amounts of acetone and stored overnight at −80 °C. In the morning samples were centrifuged at 10,000 rpm for 15 min and supernatants were collected. Solvent from supernatants was evaporated using lyophilizer under reduced pressure. Lyophilized samples were dissolved in 100 μL acetonitrile:water (80:20 v/v) and kept overnight at −80 °C, centrifuged at 10,000 rpm for 15 min at 15 °C and supernatants were collected for further analysis [7].

HPLC-Q-TOF-MS analysis

Samples were resolved on Agilent 1290 Infinity Series UPLC interfaced to an Agilent 6538 Accurate-Mass Q-ToF MS. A volume of 15 μL of each sample was injected into an assembly of C18 ZORBAX (4.6 × 12.5 mm) guard column followed by ZORBAX 300SB-C18 (4.6 × 150 mm) HPLC column of particle size 5.0 μm. The solvent system had (A) 0.1% formic acid–water and (B) 0.1% formic acid–acetonitrile. The gradient mode {concentration/time (%/min)} used for solvent B was {0%/0; 1%/5; 100%/25; 100%/32; 1%/37}, with 0.4 mL/min flow rate. The mass spectrometer was operated in positive ion polarity mode with parameters: ionization type electrospray ionization (ESI), gas temperature 350 °C, nebulizer 45 psi, gas flow 8 L/min, capillary voltage 3500 V, octopole voltage 750 V, skimmer voltage 65 V and fragmentor voltage 175 V, fixed collision energy of 25.0 V and acquisition rate 4 spectra per cycle. The quadrupole was operated in the extended dynamic range (1700 m/z, 2 GHz). To assure the mass accuracy of recorded data, continuous internal calibration was performed using standards of signals of standard compounds. Blank samples (solvent mixture of acetonitrile and water) were also injected between the sample runs. The chromatograms of blank samples were used to study the carry-over effects in LC–MS runs.

Multivariate analysis

Mass Hunter Qualitative Analysis (B.04.00, Agilent Technologies) software was used for preliminary quality check of raw MS/MS data. Profinder software was used for baseline correction, noise reduction, background contaminants removal and extraction of molecular features. Molecular features having abundance >5000 cycles per second (cps), minimum absolute counts 10, relative height (5%) of largest peak, peak space tolerance, i.e. 0.0025 m/z and 5.0 ppm were extracted from mass spectra to avoid false molecular features. Profinder reduces acquired data size and complexity through the removal of redundant and non-specific information by identifying the important variables (features). The extracted data was imported in Mass Profiler Professional software (MPP, B.12.02, Agilent Technologies) and all the molecular features were aligned using mass error < 5.0 ppm and retention time (RT) variation < 0.2 min. Compounds present in less than 75% of samples of any of three groups and having p > 0.05, fold change < 2.0 and co-efficient of variance < 15 were removed from data sets and subjected to generate Box–Whisker plot and principal component analysis (PCA). Partial least-squares-discriminant analysis (PLS-DA), a class prediction algorithms was performed to distinguish 3 Prakriti groups and explore unique features present across the groups. Finally, data were subjected to one way ANOVA with permutative Benjamini–Hochberg FDR correction and Tukey HSD post hoc to minimize the FDR, keeping p value cutoff < 0.05 and fold change > 2.0 {for p value cutoff < 0.05, false discovery rate (FDR) become <5%}. Based on changes in expression levels (upregulated or downregulated) metabolites were arranged with respect to K vs P, K vs V and P vs V classes of comparison. Differentially expressed metabolites were identified by comparison of fragmentation pattern of the metabolites with standard human metabolome in Metlin library, HMDB and NIST databases. The flowchart indicating important steps during the analysis has been shown in Supplementary Fig. 1 (Fig. S1).

Significant pathway and biological processes identification

Cytoscape 3.3 was used for further analysis of data and to find out significantly altered pathways in all the three Prakritis. Data were imported in Metascape, a plugin of Cytoscape 3.3 and subjected to pathway analysis. All the metabolic pathways generated were further subjected to enrichment and topological analysis using another plugin JEPETTO. Only the pathways having significant q value were selected and further subjected to identification of biological processes associated with them using another plugin ClueGo with standard parameters.

Results

Prakriti assessment

The questionnaire (Table S1) was validated by pre-testing and the results obtained were confirmed by the clinical assessment of Prakriti independently by two Ayurveda physicians. More than 90% concordance was observed in Prakriti assessment by the two clinicians and questionnaire. Based on the responses of volunteers to standard questionnaire and physical examination carried out by an Ayurvedic physician; 38 healthy volunteers were categorized in 3 Prakriti categories as pitta (n = 16), kapha (n = 10) and vata (n = 12). Individuals who had thin and narrow body frame, weakly developed body build, with irregular appetite, food and bowel habits, difficulty in gaining weight, quick at physical activities, dry skin and hair, and less tolerance for cold temperature were considered as vata Prakriti. Individuals with moderately developed build, high frequency of appetite and thirst, good digestive power, perspiration tendency higher than normal, tolerance for cold weather, moderately mobile with moderate physical strength were identified as pitta Prakriti. Individuals who had broad body frames with well developed body build, tendency to gain weight, low appetite and digestion, preferred to be less mobile, less forgetful and with good healing power and cool temperament, were selected as kapha individuals (Table S2). Less than 20 basal metabolic rate (BMR) was observed in Vata category while 24.0–25.0 and >25.0 were observed for Pitta and Kapha respectively. Kapha Prakriti individuals had highest level of HDL, TG and TC (Table 1a). The number of compounds obtained for each individual in each group has been represented (Table 1b).

Table 1a

Basic characteristics of the data sets.

Sr. no.	Parameters	Kapha	Pitta	Vata
1	Participants, n (%)	10 (26.3)	16 (42.1)	12 (31.5)
2	Age (years), Mean ± SD	36.77 ± 1.07	41.22 ± 1.38	39.66 ± 1.49
3	Smoker, n (%)	2	0	1
4	BMI (kg m−2)	>25.0	24.0–25.0	<20.0
5	HDL (mg dl−1), Mean ± SD	54.73 ± 1.98	20.92 ± 1.48	34.92 ± 1.42
6	Triglyceride TG (mg dl−1), Mean ± SD	190.58 ± 1.66	178.11 ± 1.51	176.49 ± 1.04
7	Cholesterol (mg dl−1), Mean ± SD	187.27 ± 1.36	171.53 ± 1.76	178.81 ± 1.57

Table 1b

Total number of compounds extracted from each sample falling under 3 study categories.

Sl. no.	Kapha (n = 10)	No. of compounds extracted	Pitta (n = 16)	No. of compounds extracted	Vata (n = 12)	No. of compounds extracted
1	Sample 1	2817	Sample 1	2470	Sample 1	3112
2	Sample 2	2769	Sample 2	2409	Sample 2	2568
3	Sample 3	2615	Sample 3	2248	Sample 3	2166
4	Sample 4	2389	Sample 4	2241	Sample 4	2166
5	Sample 5	2160	Sample 5	2231	Sample 5	3002
6	Sample 6	2543	Sample 6	2150	Sample 6	2922
7	Sample 7	2506	Sample 7	2148	Sample 7	2661
8	Sample 8	2461	Sample 8	2118	Sample 8	2447
9	Sample 9	2344	Sample 9	2117	Sample 9	2544
10	Sample 10	2130	Sample 10	2112	Sample 10	2290
11			Sample 11	2079	Sample 11	2080
12			Sample 12	2075	Sample 12	2020
13			Sample 13	2063
14			Sample 14	1956
15			Sample 15	2562
16			Sample 16	2501

Multivariate analysis of plasma metabolome

The uniform peak occurrence, spreading and lack of any outliers in chromatograms of individual groups indicated uniformity in metabolic composition in each sampling category and within group consistency of metabolites (Fig. S2, A–C). Further, only the compounds present universally in >75% of samples in each group were extracted for further processing to ensure the consistency of metabolomics profiles in each group. The chromatograms of blank samples contained only low intensity peaks, therefore, it was concluded that the carryover effect did not occur. The extracted base peak chromatograms (BPC) of kapha (K), pitta (P) and vata (V) samples showed distinct peaks (Fig. S3). Analysis of total ion chromatograms showed average 23 peaks containing around 2000 compounds. Variability in the peaks was normalized successfully using internal standards and Z-transforms. Reproducible and stable molecular features were subsequently used for statistical analysis. The well-established PLS-DA regression value 0.97 (R2 = 0.97) shows the accuracy of results. The result of the sample classification presented in terms of discrimination ability was found to be 97.87% accurate, representing the percentage of the samples correctly classified during model training and cross-validation. Data were further subjected to one way ANOVA with permutative (n = 100) and Benjamini–Hochberg multiple testing correction to validate the PLS-DA model and to further decrease false discovery rate (Table 2). Supervised PCA of P, V and K groups showed 46.0%, 7.3% and 5.96% variability along the X, Y and Z axes respectively. PCA is a most commonly used visualization to understand the metabolic differences in 3D space (Fig. 1A). The overall percent variance was represented as a box–whisker plot showing altered metabolites across all the 3 groups, i.e. P, V and K. Out of these, vata showed highest compound variability represented by elongated box shapes and whiskers towards either side to the grand mean (Fig. 1B). A Pearson heat map was also constructed to visualize the metabolomic variability among 3 study groups. All the groups showed considerable variability in metabolomic level and were clearly separated from each other (Fig. 2).

Table 2

Partial least square discriminant analysis (PLS-DA) model of class prediction was applied to study categories.

	Kapha (predicted)	Pitta (predicted)	Vata (predicted)	Accuracy (%)
Kapha (True)	9	0	0	100.00
Pitta (True)	0	29	0	100.00
Vata (True)	1	0	11	91.66
Overall accuracy (%)				97.87

Fig. 1

(A) PCA plot of variable metabolites obtained in positive mode showed significant differences among 3 sampling categories. X axis component value was 46% while for Y and Z it was 7.3%, 5.96% respectively. (B) The box-and-whisker plot is an exploratory graphic, used to show the distribution of metabolites in kapha (K), pitta (P) and vata (V) groups of individuals. The high variation in top whiskers of kapha and vata and low whisker of pitta, high variation in upper quartile of kapha and pitta and lower whisker of vata are representing metabolite variations in study groups, out of these vata has most diverse pool of metabolites.

Fig. 2

Pearson heat map alignment denoting fold changes of 76 listed metabolites across kapha (K), pitta (P) and vata (V) Prakritis. The columns correspond to different Prakriti categories and rows correspond to the altered metabolites. Correlation map is showing pitta (P) group as most discrete group and kapha (K) and vata (V) groups are more closely related.

Qualitative analysis and identification of variable metabolites

Metabolites with any missing values were filtered out from the statistically significant data obtained after analysis. Finally, 76 differentially expressed metabolites in all groups were obtained and identified using standard Metlin library and HMDB (Table 3). The variations in expression of these metabolites within K vs P, K vs V and P vs V classes were expressed in tabular format. These 76 metabolites were belonging to carnitines, nucleic acids, amino acids, vitamins and fatty acids/lipids. Out of these 76, 39 compounds had highest expression in pitta category, 6 were in vata while 31 were in kapha group. Four metabolites belonging to nucleotides class were 8-hydroxyguanine, an oxidative derivative of guanosine and biomarker of oxidative stress, 10-formyl tetrahydrofolyl l-glutamate is an useful intermediate in purine biosynthesis, 2′-deoxyuridine and 5-methylcytosine nucleotide analogues. 8-Hydroxyguanine and 2′-deoxyuridine were abundant in kapha while other two were in pitta Prakriti (Table 3). The lipid compounds were mostly belonging to kapha and pitta categories: PS(19:0/0:0), PA(14:0/14:0), DG(18:3(9Z,12Z,15Z)/20:2(11Z,14Z)/0:0), LysoPC(22:5(7Z,10Z,13Z,16Z,19Z)), 2,3-dinor-6-keto-prostaglandin F1α, a potent vasodilator and inhibitor of platelet aggregation whereas docosanamidoacetic acid and N-stearoylethanolamine, appear in various fatty acid oxidation disorders, were abundant in kapha Prakriti. 21-Hydroxypregnenolone useful in corticosterone and deoxycorticosterone synthesis, N-docosanoyl taurine, a novel taurine-conjugated fatty acids, LysoPC(18:0) useful in lipid signaling, lysoPE, (18:3(6Z,9Z,12Z)/0:0N-heptanoylglycine), appears in various fatty acid oxidation disorders, 2-hydroxy-13-O-d-glucuronoside-octadec-9Z-enoate is a linoleate intermediate were found upregulated in kapha Prakriti, whereas 16,17-epoxydeoxycorticosterone acetate; valerenic acid, a branched fatty acid; 2-methoxyestrone, a steroid intermediate; and glycochenodeoxycholate, a bile salt were abundant in pitta Prakriti. The 10,11-dihydro-20-trihydroxy-leukotriene B4, a metabolite of lipid omega-oxidation of leukotriene B4 and a diglyceride DG(14:0/14:1(9Z)/0:0) were abundant in vata category. Out of 11 differentially expressed amino acids and their derivatives, 6 were belonging to pitta category; l-valine is an essential branched chain amino acid (BCAA), critical to human life and is involved in stress, energy and muscle metabolism, L-2,4-diaminobutanoate an another BCAA metabolism component, 6-amino-2-oxohexanoate is a lysine degradation intermediate, capryloylglycine is used to diagnose disorders associated with mitochondrial inborn errors of metabolism. The dominant amino acid derivatives in kapha were argininosuccinic acid, a precursor for arginine and fumarate; kynurenine, a l-tryptophan breakdown product significant in numerous biological processes; and 2-oxopentanoic acid, an arginine catabolism intermediate (Table 3).

Table 3

The list of plasma metabolites obtained in +ESI, processed through Profinder and analyzed in MPP are arranged in ascending order of their mass range. Metabolites from the PLS-DA model without missing values were ranked according to the significance and variability and identified using various databases.

Sr. no.	Compounds	Mass (Da)	RT (min)	p Value (corrected)	FC (K vs P)	FC (K vs V)	FC (P vs V)	Abundance based order	±PPM error
1	LysoPC(22:5(7Z,10Z,13Z,16Z,19Z))	569.3544	29.83	9.37E−04	2.75	1.00	−2.75	K = V > P	−3.81
2	Homoarginine	188.1524	5.35	9.37E−04	2.81	1.01	−2.77	K = V > P	−9.69
3	Unknown	1016.679	36.00	9.37E−04	2.85	1.05	−2.70	K = V > P	–
4	Unknown	278.1842	5.52	0.009072	2.31	−1.00	−2.32	K = V > P	–
5	Unknown	622.0015	5.39	4.28E−04	3.12	3.12	−1	K > P = V	–
6	1-O-Sinapoyl-beta-d-glucose	386.1439	5.77	0.01693	1.80	2.95	1.64	K > P > V	−7.64
7	Sphinganine	301.2291	19.68	0.00583	1.93	3.29	1.70	K > P > V	0.57
8	DG(18:3(9Z,12Z,15Z)/20:2(11Z,14Z)/0:0)	642.4563	33.70	0.002483	2.57	2.97	1.15	K > P > V	4.43
9	Hydroxyvalerylcarnitine	263.2615	25.14	3.74E−04	3.02	3.45	1.14	K > P > V	2.48
10	Unknown	751.6487	5.25	0.008485	1.66	3.28	1.97	K > P > V	–
11	N-Stearoylethanolamine	327.3137	27.82	0.002483	2.76	1.28	−2.16	K > V > P	−5.71
12	d-Threitol	122.0383	9.02	0.002113	2.85	1.40	−2.03	K > V > P	−0.34
13	21-Hydroxydydrogesterone glucuronide	504.2762	23.38	0.001943	2.90	1.47	−1.96	K > V > P	3.82
14	8-Hydroxyguanine	299.119	11.02	0.001819	2.96	2.05	−1.44	K > V > P	−11.99
15	2-Deoxyuridine	228.1109	5.79	1.90E−04	3.30	1.21	−2.74	K > V > P	5.02
16	PS(19:0/0:0)	539.3645	24.08	0.002217	2.82	1.35	−2.08	K > V > P	−4.78
17	2,3-Dinor-6-keto-prostaglandin F1α	342.2884	23.04	0.002003	2.87	1.42	−2.01	K > V > P	2.56
18	PA(14:0/14:0)	592.3261	22.66	0.001766	2.95	1.56	−1.88	K > V > P	−0.54
19	Docosanamidoacetic acid	397.356	29.87	0.002113	2.84	1.38	−2.05	K > V > P	−2.41
20	16-Hydroxypalmitic acid	271.2516	24.22	6.06E−04	3.31	1.91	−1.74	K > V > P	−3.77
21	Dibutyl phthalate	278.1553	28.78	1.45E−06	3.72	1.31	−2.87	K > V > P	−3.15
22	2-Oxopentanoic acid	116.0626	24.55	0.001766	2.98	2.08	−1.43	K > V > P	−4.77
23	Argininosuccinic acid	290.1134	23.41	0.001757	2.98	2.09	−1.42	K > V > P	−3.34
24	l-Kynurenine	208.1097	23.40	4.40E−04	3.33	1.67	−1.99	K > V > P	−7.95
25	1α,25-dihydroxy-11-(4-hydroxymethylphenyl)-9,11-didehydrovitamin D3	520.4126	33.94	0.002678	2.79	1.37	−2.02	K > V > P	−4.17
26	Unknown	287.2827	23.78	0.001819	2.96	2.04	−1.45	K > V > P	–
27	3-Hydroxyoctanoyl carnitine	303.2051	15.69	0.001701	2.99	2.11	−1.41	K > V > P	6.92
28	4-Methylthio-2-oxobutanoic acid	148.0159	35.41	0.017991	2.40	1.47	−1.63	K > V > P	0.56
29	Paired-like homeodomain transcription factor 3	8804.468	20.27	0.001903	2.95	2.03	−1.45	K > V > P	6.89
30	l-Octanoylcarnitine	287.2493	25.20	1.90E−04	3.30	1.20	−2.74	K > V > P	−7.77
31	Unknown	735.6437	35.91	0.001766	2.97	2.06	−1.43	K > V > P	–
32	2-Methoxyestrone	300.1383	9.037	9.37E−04	−2.76	1	2.76	P > K = V	−4.52
33	N-Docosanoyl taurine	447.3026	20.11	9.37E−04	−2.76	1	2.76	P > K = V	4.68
34	L-2,4-diaminobutanoate	118.0425	8.82	1.90E−04	−3.07	1	3.07	P > K = V	−5.95
35	Capryloylglycine	201.1733	8.75	9.37E−04	−2.76	−1.00	2.76	P > K = V	−4.84
36	Malonuric acid	146.0374	8.83	1.90E−04	−3.07	−1	3.07	P > K = V	5.92
	Transcription factor HFK1	15,119.04	17.82	9.37E−04	−2.75	−1	2.75	P > K = V	8.55
38	Unknown	303.2065	15.20	9.37E−04	−2.75	−1	2.75	P > K = V	–
39	Coumaric acid	164.0482	8.83	9.37E−04	−2.69	1.14	3.07	P > K > V	−8.61
40	5-Methylcytosine	125.0846	10.92	0.004456	−2.33	1.16	2.71	P > K > V	−8.74
41	LysoPC(18:0)	523.3677	26.96	0.017991	−1.85	1.41	2.62	P > K > V	−5.17
42	Benzoic acid	122.0374	8.83	9.37E−04	−2.65	1.15	3.06	P > K > V	−9.66
43	Traumatic acid	228.1488	8.79	9.37E−04	−2.76	−1.00	2.76	P > V = K	4.26
44	1,25-Dihydroxyvitamin D3 3-glycoside	578.2892	24.66	9.37E−04	−2.76	−1	2.76	P > V = K	0.15
45	(2E,6E,8E)N-Isobutyl-2,6,8-hexadecatriene-10-yneamide	301.2272	19.45	9.37E−04	−2.76	−1	2.76	P > V = K	2.42
46	Putative reverse transcriptase	6626.596	17.88	9.37E−04	−2.75	−1	2.75	P > V = K	7.45
47	Inositol cyclic phosphate	242.0381	11.36	0.004514	−2.70	−1.39	1.94	P > V > K	0.09
48	12α-Hydroxyerosone	368.1073	10.06	0.004862	−2.67	−1.42	1.87	P > V > K	−2.51
49	Guanidinoacetic acid	117.0583	12.14	0.004862	−2.66	−1.43	1.86	P > V > K	−8.61
50	Glycochenodeoxycholate	449.3177	23.00	0.005034	−2.66	−1.49	1.84	P > V > K	8.88
51	Melatonin	232.1593	13.85	0.005267	−2.65	−1.45	1.82	P > V > K	−1.97
52	10-Formyl tetrahydrofolyl l-glutamate	602.2897	24.73	9.37E−04	−2.76	−1	2.76	P > V > K	0.14
53	21-Hydroxypregnenolone	332.1452	31.32	0.006784	−2.61	−2.01	1.29	P > V > K	−8.83
54	LysoPE(18:3(6Z,9Z,12Z)/0:0)	475.2726	25.23	0.005976	−2.64	−1.93	1.36	P > V > K	−3.17
55	N-Heptanoylglycine	187.0642	12.11	0.004414	−2.71	−1.37	1.97	P > V > K	9.5
56	2-Hydroxy-13-O-d-glucuronoside-octadec-9Z-enoate	490.3318	13.84	0.004862	−2.67	−1.43	1.87	P > V > K	−8.2
57	Dehydrospermidine	143.0739	12.14	0.004862	−2.67	−1.42	1.88	P > V > K	−14.42
58	9-Hydroxyphenanthrene	194.0813	12.50	0.005181	−2.65	−1.43	1.85	P > V > K	7.51
59	Valine	117.0582	11.36	0.004414	−2.71	−1.37	1.96	P > V > K	−7.75
60	6-Amino-2-oxohexanoate	145.0533	12.14	0.004862	−2.66	−1.43	1.86	P > V > K	−4.15
61	Valerenic acid	234.1746	13.85	0.004862	−2.67	−1.42	1.87	P > V > K	−8.3
62	2-(3-Carboxy-3-aminopropyl)-l-histidine	256.1567	13.84	0.004862	−2.67	−1.42	1.87	P > V > K	−7.36
63	1α-hydroxy-2β-(5-hydroxypentoxy) vitamin D3	502.3023	13.62	0.004862	−2.68	−1.41	1.90	P > V > K	1.98
64	Unknown	287.2481	22.81	0.004514	−2.70	−1.38	1.94	P > V > K	–
65	l-Hexanoylcarnitine	259.18	15.11	0.004862	−2.67	−1.42	1.88	P > V > K	3.78
66	beta-Alanyl-l-arginine	245.164	12.58	9.37E−04	−3.02	−1.39	2.16	P > V > K	−6.56
67	Splicing factor 1 isoform 4	15,859.4	17.68	9.37E−04	−2.76	−1	2.76	P > V > K	6.88
68	Catechin	290.1154	23.17	0.004862	−2.66	−1.43	1.86	P > V > K	3.65
69	16,17-Epoxydeoxycorticosterone acetate	386.1762	22.61	0.004862	−2.67	−1.43	1.86	P > V > K	5.92
70	Unknown	1895.032	11.48	0.004537	−2.69	−1.39	1.93	P > V > K	–
71	l-Aspartate 4-semialdehyde	117.0789	5.52	0.001141	2.72	−1.00	−2.73	V > K > P	−4.05
72	1α,25-dihydroxy-26,27-dipropylvitamin D3	500.4923	28.41	9.37E−04	2.76	−1.03	−2.86	V > K > P	6.72
73	Unknown	471.8662	5.00	0.005393	1.10	−2.83	−3.13	V > K > P	–
74	Unknown	444.2072	23.98	0.0033	−2.56	−2.84	−1.11	V > P > K	–
75	10,11-Dihydro-20-trihydroxy-leukotriene B4	386.2481	22.15	9.37E−04	−2.76	−1	2.76	V > P > K	−10.48
76	DG(14:0/14:1(9Z)/0:0)	510.392	36.17	0.013222	−2.02	−2.86	−1.41	V > P > K	0.71

Abbreviations: RT – retention time, p – calculated probability and FC – fold change.

Out of 4 vitamin D derivatives observed, 1,25-Dihydroxyvitamin D3 3-glycoside and 1α-hydroxy-2β-(5-hydroxypentoxy) vitamin D3 had highest expression in pitta Prakriti. Most of the carnitines were predominantly expressed in kapha category; these include l-octanoylcarnitine, 3-hydroxyoctanoyl carnitine and hydroxyvalerylcarnitine while hexanoylcarnitine an unusual acylcarnitine indicator of metabolic disorders was dominant in pitta Prakriti.

m-Coumaric acid and benzoic acid are polyphenol metabolites formed by the gut microflora, guanidoacetic acid, a creatine precursor that causes cardiovascular and skeletal problems, melatonin, an antioxidant, nucleic acid protector and regulator of body clock and dehydrospermidine that reduces amount of aging in human immune cells were highest in pitta category. Sphinganine, a blocker of postlysosomal cholesterol transport was abundant in kapha category.

The carbohydrate compounds, d-threitol, a main end product of d-xylose metabolism and an indicator of carbohydrate metabolism derangements, 1-O-sinapoyl-beta-d-glucose and 21-hydroxydydrogesterone glucuronide were highest in kapha Prakriti. Overall 11 compounds remained unidentified of which majority belonged to kapha category (Table 3).

Pathway analysis

Category-wise the highest abundant upregulated and downregulated metabolites were segregated and used to look at the significant metabolic pathways associated with each Prakriti group. Metascape plugin in Cytoscape 3.3 was used to explore significantly affected metabolomics pathways in different Prakritis. In kapha Prakriti, pathways with q value < 0.01 were found to be sphingolipid, glycerophospholipid, ether lipid, alanine, aspartate, glutamate, arginine, proline, pyrimidine metabolism and Fc gamma R-mediated phagocytosis. Pitta Prakriti showed enhanced valine, leucine and isoleucine degradation. Other pathways were found to be equally affected in vata and kapha Prakritis. Vata Prakriti showed upregulated arachidonic acid, beta-alanine, butanoate, catecholamine and glycerolipid metabolism (Table 4, Fig. 3).

Table 4

The list of significant metabolic pathways that are being regulated obtained from Cytoscape and its plugins (Metascape and JEPETTO) is represented. The metabolic pathways are arranged in ascending order of their respective q value, their impact in terms of XD score and regulation of pathways in different Prakritis have also been mentioned.

Sr. No.	Pathway or process	XD-score	q-Value	Overlap/size	Regulation in different Prakritis
1	Steroid hormone biosynthesis	1.29764	0.00001	5/15	P = V
2	Metabolism of xenobiotics by cytochrome P450	0.96431	0.00001	5/20	P = V
3	Sphingolipid metabolism	0.8734	0.00002	5/22	K
4	Tryptophan metabolism	0.73354	0.00003	5/26	P = V
5	Steroid hormone metabolism	1.05522	0.00173	3/11	P = V
6	Retinol metabolism	0.96431	0.00196	3/12	P = V
7	Glycerophospholipid metabolism	0.42145	0.00243	4/35	K
8	Ether lipid metabolism	0.71431	0.00377	3/16	K
9	Tyrosine metabolism	0.67673	0.00377	3/17	P = V
10	Drug metabolism – cytochrome P450	0.67019	0.00377	3/17	K = P = V
11	Glycerolipid metabolism	0.56431	0.0057	3/20	V
12	Alanine, aspartate and glutamate metabolism	0.53574	0.00611	3/21	K
13	Valine, leucine and isoleucine degradation	0.40876	0.01211	3/27	P
14	Arginine and proline metabolism	0.28864	0.02857	3/37	K
15	Fc gamma R-mediated phagocytosis	0.15943	0.03186	4/82	K
16	Pyrimidine metabolism	0.15255	0.03034	3/70	K
17	Arachidonic acid metabolism	1.04886	0.00000	7/26	V
18	beta-Alanine metabolism	0.77194	0.00148	3/15	V
19	Butanoate metabolism	0.54336	0.04580	2/14	V
20	Catecholamine metabolic process	0.53574	0.00377	3/21	V

Fig. 3

Figure showing significant pathways found altered in different Prakritis. All the metabolites were imported to Cytoscape from MPP and analysis was done with help of Metascape plugin and further significance was calculated using JEPETTO plugin.

Go biological processes

Further, in order to explore the biological processes associated with differentially expressed metabolites, ClueGo plugin in Cytoscape 3.3 was used. In kapha Prakriti mitochondrial genome maintenance, pyrimidine, tryptophan, kynurenine, membrane lipid, aromatic amino acid family, NAD biosynthetic metabolism, and cellular biogenic amine catabolic processes were found to be dominant (Fig. 4A). In pitta Prakriti retinoid metabolic, aminoglycan biosynthetic, uronic acid metabolic, estrogen metabolic process, branched-chain amino acid metabolic, toxin metabolic, heparin biosynthetic and regulation of cellular ketone metabolic processes were found to be positively regulated. Whereas, cellular carbohydrate, lipid, and cellular glucuronidation metabolic processes regulation were found to be downregulated (Fig. 4B). In vata Prakriti, very long-chain fatty acid, sulfur amino acid, glutamate, catecholamine, leukotriene, estrogen metabolic, vitamin K, hydrogen peroxide metabolic processes and ketone, phylloquinone, dopamine, sodium ion homeostasis and cellular detoxification catabolic processes were found to be dominant (Fig. 4C) (Fig. S4, A–C).

Fig. 4

Go Biological process in all the three Prakritis analyzed using Metscape and ClueGo modules of Cytoscape 3.3 (A) Kapha, (B) Pitta, and (C) Vata.

Discussion

Ayurveda is an ancient system of personalized medicine documented and practiced in India since 1500 B.C. In Ayurveda, it is believed that the human body constitution determines predisposition, prognosis of diseases, therapy and lifestyle regime. The entire description of human physiology in Ayurveda is based on the theory of Tridosha [8]. According to Ayurveda, vata, pitta, and kapha Prakriti are found to have unique metabolic activities, kapha is slow, pitta is fast, and vata is considered to have variable metabolism [9]. Vata is associated with bone, pitta with blood, while kapha is associated with other tissues related to structure and storage such as adipose tissue [10], [11].

Kapha governs the nervous and musculo-skeletal systems that maintain body mass, shape, and flexibility [12]. The association of kapha dosha with anabolic processes and co-ordination of genes delayed manifestation of aging and provides a longer life span [13], [14]. In the current study, mitochondrial genome maintenance, pyrimidine, membrane lipid, NAD biosynthetic processes were found to be dominant in kapha Prakriti (Fig. 4A). It leads to slow catabolism of lipids and therefore, storage, and weight gaining. Kapha Prakriti individuals are prone to disorders of the respiratory system and therefore, continuous mitochondrial genome maintenance is required. The results supported by Hankey who reported that the peptide coenzyme A is associated with lipid metabolism and dominant in kapha Prakriti [15]. As noticed in the present study (Table 1a, Table 1b), higher levels of markers of metabolic syndrome, such as triglyceride (TG), total cholesterol, LDL, VLDL and HDL were also reported in kapha Prakriti people [16]. Accumulations of lipids also lead to a variety of cardiovascular, neurological processes and diseases [17]. Increased levels of capryloylglycine, 8-hydroxyguanine, d-threitol, hexanoylcarnitine and Kynurenine; the well-established markers of cardiovascular diseases, metabolic irregularities and oxidative stress, confirm the biological processes established in the study (Table 3). Docosanamidoacetic acid, a known marker of fatty acid oxidation disorders was abundantly found in kapha people (Table 3). Amino acid derivatives 2-oxopentanoic acid, a metabolite of arginine catabolism; homoarginine, an inhibitor of liver alkaline phosphohydrolase; and argininosuccinic acid, an arginine precursor were observed in kapha. Presence of valine and arginine precursors may be linked with body's response to stress induced by kapha defects by enhancing BCAA and urea metabolism. Overall, lipid and pyrimidine metabolic and aromatic amino acid family catabolic processes are dominant in kapha Prakriti people (Fig. 4A).

Pitta is primarily responsible for metabolism, actions of enzymes, growth factors, hormones, thermo-regulation, energy homeostasis, digestion, and cellular or sub-cellular metabolism [18]. Ghodke et al., have shown that the extensive metabolizer (EM) genotype of the drug metabolizing enzyme CYP2C19 was found only in pitta Prakriti [19]. In the current study, enhanced metabolism of xenobiotics by cytochrome P450 (Table 4) and the toxin metabolic process was observed in pitta Prakriti (Fig. 4B). The pitta predominance Prakriti type individuals have a high basal metabolic rate (BMR) and energy consumption leading to tissue destruction and premature aging and average life span [9]. LysoPE(18:3(6Z,9Z,12Z)/0:0) and LysoPC(18:0) are useful in lipid signaling, 2-hydroxy-13-O-d-glucuronoside-octadec-9Z-enoate, a linoleate intermediate and N-heptanoylglycine, a marker of mitochondrial fatty acid beta-oxidation disorders were abundant in pitta Prakriti. Along with these, levels of glycochenodeoxycholate, a bile acid salt that solubilizes fats for absorption; were found to be high in pitta Prakriti people. All of these metabolites correspond to the enhanced lipid and energy metabolism and therefore, BMI.

The high metabolic turnover in pitta Prakriti was shown by valine, leucine, isoleucine biosynthesis as well as degradation and vitamin metabolism. The BCAA are critical to human life and are particularly involved in stress, energy and muscle metabolism, as valine goes solely to carbohydrates, and valine deficiency is marked by neurological defects in the brain [20]. At the same time, melatonin, a crucial hormone that decides the body's basal metabolism rate and sleep–wake cycle was also found to be highest in pitta Prakriti. Protein network analysis has shown that BCCA metabolic, heparin biosynthetic and cellular ketone metabolic processes were positively regulated in pitta Prakriti (Fig. 4B).

In pitta Prakriti, 2-methoxyestrone, a steroid intermediate of androgen and estrogen metabolic pathway that determines the physical constitution of humans was found to be upregulated. These steroid hormones are known to modify brain structures, implicated in behavioral conditions, and libido, hence can be first marker for Tridosha system. Moreover, in pitta Prakriti, retinoid metabolic, aminoglycan biosynthetic, uronic acid metabolic, cellular carbohydrate, lipid, and cellular glucuronidation metabolic processes were found to be affected (Fig. 4B).

Vata contributes to manifestation of shape, cell division, signaling, movement, excretion of wastes, cognition and also regulates the activities of kapha and pitta. Vata dosha is responsible for movement, cell growth and apoptosis. Vata body types can have propensity to develop neurological problems, dementia, movement and speech disorders, arrhythmias, and related chronic diseases [21]. In vata Prakriti very long-chain fatty acid, sulfur amino acid, glutamate, catecholamine, leukotriene, estrogen metabolic, vitamin K, hydrogen peroxide metabolic processes and ketone, phylloquinone, dopamine, sodium ion homeostasis and cellular detoxification catabolic processes were found to be dominant (Fig. 4C). In the present study, LysoPC(22:5(7Z,10Z,13Z,16Z,19Z)), lysophospholipid have role in lipid signaling, 10,11-dihydro-20-trihydroxy-leukotriene B4, a metabolite of lipid omega-oxidation of leukotriene B4 and a diglyceride DG(14:0/14:1(9Z)/0:0) were abundant in vata category. Amino acid metabolism and l-Aspartate 4-semialdehyde that is involved in lysine and homoserine synthesis also were most abundant in vata. Increased amounts of dopamine in vata Prakriti people may lead to neurological problems as well as insulin resistance. Insulin resistance, cytokines and inflammatory markers have got positive relation with V and K group both [15].

During the human metabolism studies, it is an established fact that average metabolite concentration varies by ±50% [22]. This variation is due to age, gender, genetics, health status, activity level and diurnal variations [23]. Ayurveda describes the predominance of kapha in childhood, pitta in young and vata in old age. It is apparent that children are more prone to certain type of diseases due to provocation of kapha. Similarly certain types of diseases are very common in younger and older age. This concept is called as vayoanupatinee Prakriti [24], [25]. Hence, the age may have a profounding effect on metabolic expression that can't be assessed in present setup and needs to be explored further. Ayurveda considers all these factors during Prakriti analysis. However, current methods for Prakriti analysis are inconsistent and some scientifically developed methods lack informational ground. Ayurmetabolomics may provide a platform to discriminate different Prakriti types with constitutional information. In our study, functional categories of metabolites showing differential expression among Prakriti types were significantly enriched in core biological processes like amino acids, vitamins, lipids and nucleic acid metabolism. The relative closeness between vata and kapha Prakritis has been confirmed in PCA plot and heat map (Figs. 1A and 2).

Conclusion

The present study, presented an idea that plasma level of some simple metabolites, i.e. carnitines, 4-aminohippuric acid, androgen, melatonin, prostaglandins, l-kynurenine, amino acids, vitamins and lipids; can classify the human phenotypes, i.e. Prakritis. Metabolites are the end players in body metabolism and signaling molecules to initiate metabolic and catabolic pathways in the body. Metabolomics deals with the levels of metabolites in the body fluids and tissues in different conditions. Phenotypes of individuals mainly depends on the body metabolism. Tridosha classification in Ayurveda mainly depends upon the phenotypic and psychological characteristics of individuals. Hence, integration of Ayurveda with metabolomics holds potential and promise for future predictive, individualized medicine. Present study indicates that dominating kapha Prakriti individual are predisposed to respiratory problems, hence continuous mitochondrial genome maintenance is the dominating biological process in response to stress slow lipid catabolism. It also leads to lipid storage and hence, weight gain. Moreover, upregulated NAD biosynthetic processes also confirmed the slow lipid metabolism. The increased lipid levels may lead to oxidative stress and cardiovascular problems. In the study, the levels of cardiovascular and neurological diseases markers i.e. capryloylglycine, 8-hydroxyguanine, d-threitol, hexanoylcarnitine and kynurenine were found to be increased. Hence, study showed the predisposition of kapha dominant people to cardiovascular and neurological diseases as described in Ayurveda, but continuous maintenance of respiratory system help to control these diseases. In pitta Prakriti metabolism of xenobiotics and toxins was observed through cytochrome P450. In this group, beta-oxidation disorders are more common, may be because of increased lipid absorption through the intestine as increased levels of bile acids were observed in the study. Melatonin and 2-methoxyestrone, hormones to maintain circadian biological clock, basal metabolism rate, and physical constitution, were found to be dominant. All these processes can be correlated with fast metabolism of pitta Prakriti. In vata Prakriti cellular detoxification process along with neurotransmitters i.e. dopamine and glutamate was found to be dominant. These processes are important to excrete waste and cognition. Hence, vata individuals may have predisposition to neurological problems and insulin resistance. The present study provides a framework to understand Prakriti in terms of metabolites and also to prognose associated disorders. Experimentally, this is limited data and needs further exploration with large data sets. However, data provides a useful benchmark and provides correlation in phenotype to metabolic signatures of different Prakritis.

Sources of funding

None.

Conflict of interest

None.