NMR- and MS-based multiplex approach for comparative metabolome profiling of Curcuma caesia versus curcuma longa rhizomes and in relation to in vitro biological effects Authors: Radwa H El-Akad, Tarik A Mohamed, Alexandru Nicolescu, Teshome Degfie Beshah, Sherif R Abdel-All, Andrei Mocan, Mohamed A Farag Published: 2025 ABSTRACT AbstractThe study presents the first integrated comparative metabolomics approach of Curcuma caesia known as blue curcuma, that is relatively less explored, compared with the well-known yellow curcuma Curcuma longa using a multiplex analytical approach viz., UV, NMR, UPLC/MS/MS, GC/MS and SPME/GC/MS. The study targeted fingerprinting and chemical characterization of their primary and secondary metabolites as well as annotation of volatiles affecting their aroma and culinary uses. Datasets were further modelled using multivariate data analyses (PCA and OPLS) for samples’ classification and discrimination. The study revealed several newly reported metabolites in blue curcuma and assigned discriminating biomarkers for each species i.e. sesquiterpenes in blue curcuma vs. curcuminoids in yellow curcuma. Bioactivity assays, including antioxidant and antidiabetic activities, showed moderate activity for C. caesia relative to C. longa, while both species were inactive as anti-obesity agents.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-30956-9.Keywords: Blue curcuma, C. caesia, C. longa, Metabolites profiling, UPLC/MS/MS, GC/MS, NMRSubject terms: Biochemistry, Biological techniques, Chemistry, Drug discovery, Plant sciences Introduction Curcuma is a prominent genus in the family Zingiberacea that comprises ~ 100 species; globally recognized for their applications in traditional medicine, culinary arts, food industry, therapeutic products, cosmetics, as well as sources for natural pigments1,2. Despite being native to tropical and subtropical areas in Asia, the word “Curcuma” is originally derived from the Arabic word “kurkum” which means “yellow” referring to the rhizome color2,3. Curcuma longa, known as yellow turmeric, is the most notable species that has a long ethnobotanical history in Ayruvedic and Unani medicine and reported for wide array of pharmacological activities attributed to its enriched phytochemical reservoir viz. curcuminoids, aroma compounds, and polyphenols2–4. In contrast, C. caesia, known as blue/black turmeric after its distinguished rhizome color, remains relatively less explored despite being traditionally used for the treatment of various ailments including gastrointestinal tract disorders, wound healing, respiratory disorders, fever, rheumatoid arthritis, among several others5,6. Previous phytochemical studies reported sesquiterpenes, flavonoids and volatile constituents in blue curcuma as well as positively reacting to screening assays for alkaloids, tannins and phenolics5,7–9. Both species are reported to have significant pharmacological activity against cancer, inflammation, bacterial and fungal infections, liver diseases, and other metabolic disorders2,3,5,6,8. Despite extensive studies on C. longa and increasing interest in C. caesia, an integrated and comprehensive comparative study between both species has yet to be performed that aids provide deeper insight on the metabolome of blue curcuma, in particular. This study presents the first multiplex analytical approach for a complementary metabolomics overview of both blue and yellow curcuma, using fingerprinting techniques i.e., ultraviolet absorbance (UV), nuclear magnetic resonance (NMR), versus hyphenated techniques i.e., ultra-performance liquid chromatography coupled to high resolution mass spectrometry (UPLC/HR-MS/MS), gas chromatography coupled to mass spectrometry (GC/MS) and solid phase microextraction-(SPME-GC/MS). The multi-platform metabolomic analyses presented in this study offer untargeted chemical characterization of polar, semi-polar and non-polar metabolites in addition to volatile constituents. Multivariate data analyses including principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were employed to classify among samples and biomarker assignment. Furthermore, antioxidant activity was investigated using different assays i.e. 1,1-diphenyl-2-picrylhydrazyl (DPPH), 2,2’-azino-bis(3-ethylbenzothiazoline) 6-sulfonic acid (ABTS) and ferric reducing antioxidant power (FRAP) to assess their free radical scavenging activity comparatively, alongside quantification of their total phenolic and flavonoid content. Further, in vitro assay of both turmeric samples against glucosidase and lipase enzymes was investigated as a marker for diabetes or obesity management. Materials and methods Plant material C. caesia and C. longa were collected from Chongzuo, Guangxi Province, during the winter of 2023. Dr. Kai Wang authenticated rhizomes at the Chinese Academy of Sciences, Beijing. Rhizomes were lyophilized overnight and stored in sealed containers until further assays. Specimen was deposited at the herbarium of Faculty of Pharmacy-Cairo University, voucher no.: 1−12-23-F. UV fingerprinting of C. caesia and C. longa chloroform extracts Three grams of each freeze-dried powdered Curcuma sample was macerated with 30 mL chloroform for 2 h, then centrifuged and filtered. An aliquot of 200 µL was used, prepared from four replicates in a 96-well plate for the Gen 5 Greener UV microplate reader (Gen 5, kitted with a 96-well quartz cell with 1 nm spectral resolution in the UV region). Aliquots of each curcuma extract (200 µL) were pipetted into the microplate wells (n = 4) of a Gen 5 Greener UV microplate quartz reader (BioTek Instruments, Inc., Winooski, VT, USA). The absorption spectra were recorded in the range of 200–500 nm 10. NMR fingerprinting of C. caesia and C. longa methanol extracts Five milliliters of the methanol extract were aliquoted and subjected to drying under a nitrogen stream till complete dryness, and pellet was further resuspended in 700 ul CD3OD containing 0.96 mM hexamethyldisilane (HMDS) as an internal standard. All spectra were obtained using a Varian/Agilent VNMRS 600 NMR spectrometer, operating at proton and carbon NMR frequencies of 599.83 MHz and 150.84 MHz, respectively, with a 5-mm inverse detection cryoprobe. The1H NMR spectra were recorded with a spectral resolution of 0.26 Hz/point (Spectral Width, SW = 8389 Hz, acquisition size = 64 K complex points, zero-filled to a final Fourier transform (FT) size of 128 K). The parameters used were as follows: pulse width (pw) = 6 µs (90°), relaxation delay = 22.3 s, acquisition time = 2.7 s, and 120 scans11,12. HMBC spectra were acquired with a bandwidth of 14 ppm in F2 (1H) and 235 ppm in F1 (13C), using two scans per 256 increments for F1 and 2516 complex data points in F2. The HMBC experiments were optimized for long-range couplings of 8 Hz, with a relaxation delay of 1 s and an acquisition time of 0.15 s, resulting in a total acquisition time of 22 min. To minimize degradation and artifact formation, all NMR spectra were recorded from samples prepared immediately before data acquisition, within a 6-h time frame. Spectra were automatically phase-corrected, with proton and carbon chemical shifts referenced to TSP at (δ1H = 0.0 ppm) and internal CD3OD (δ13C = 49.86 ppm), respectively11,12. Additionally, 2D-NMR spectra, including1H single bond1H total correlation spectroscopy (TOCSY), 1Hsingle bond13C heteronuclear single-quantum coherence (HSQC), and1H single bond13C HMBC, were recorded for selected samples using standard CHEMPACK 7.1 pulse sequences (zTOCSY, gHSQCAD, and gHMBCAD) implemented in the Varian VNMRJ 4.2 A spectrometer software. NMR quantification For metabolites quantification using NMR spectroscopy, the peak areas of selected proton signals belonging to the target compounds and the internal standard (TSP) were integrated manually for all samples11. The following equation was applied for absolute metabolite level calculations13: mT = MT X (IT/ISt) X (XSt/XT) X CSt X VSt. mT: mass of the target compound [µg/g] in the solution used for1H NMR measurement. MT: molecular weight of the target compound [g/mol]. IT: relative integral value of the1H NMR signal of the target compound. ISt: relative integral value of the1H NMR signal of the standard compound. XSt: number of protons belonging to the1H NMR signal of the standard compound. XT: number of protons belonging to the1H NMR signal of the target compound. CSt: concentration of internal standard (TSP) in the solution used for1H NMR measurement [mmol/L]. VSt: volume of solution used for1H NMR measurement [mL]. Metabolites profiling of C. caesia and C. longa extracts via UPLC/HR-MS/MS Briefly, 150 mg of each Curcuma fine powder specimen was homogenized with 5 mL MeOH (100% v/v) containing 10 µg/mL umbelliferone as an internal standard using an Ultra-Turrax mixer (IKA, Staufen, Germany) adjusted at 11,000 rpm, five times for 20 s periods, with intervals of 1 min between each mixing period to guard against temperature increases and heating effects. The resultant suspensions were then vortexed vigorously, centrifuged at 3000× g for 30 min, and filtered through a 22 μm pore size filter to remove plant debris. Then, 1 mL was aliquoted and filtered, and placed in LCMS vials10. Three independent replicates (n = 3) were done for each sample to assess biological variance. The principal step of UPLC/HR-MS/MS analysis was conducted in triplicate (n = 3), with 2 µL introduced to an Dionex 3000 UPLC system (Thermo Fisher Scientific, Bremen, Germany) equipped with a HSS T3 column (100 × 1.0 mm, 1.8 μm; Waters®; column temperature: 40 °C), and a photodiode array detector (PDA, Thermo Fisher Scientific, Bremen). Binary gradient elution protocol at a flow rate of 150 µL/min using water/formic acid, 99.9/0.1 (v/v) (A) and acetonitrile/formic acid 99.9/0.1 (v/v) (B) varying between isocratic step for 1 min of 5% mobile phase B, then a linear increase of B from 5% to 100% over 11 min. The mobile phase was kept isocratic between 11 and 19 min at 100% B followed by decreasing towards 5% B within 1 min, and, finally, an additional 10 min, i.e., 20–30 min, for column re-equilibration using 5% B10. The UPLC system was coupled with a high-resolution mass spectrometer using an Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) equipped with a HESI electrospray ion source (spray voltage: positive ion mode 4 kV, negative ion mode 3 kV; source heater temperature: 250 °C; capillary temperature: 300 °C; FTMS resolution: 30,000). Nitrogen was used as sheath and auxiliary gas. The CID mass spectra (buffer gas: helium; FTMS resolution: 15,000) were recorded in data-dependent acquisition mode (dda) using a normalized collision energy (NCE) of 35% and 45%14. GC/MS analysis of silylated primary metabolites in C. caesia and C. longa Briefly, finely powdered samples (100 mg) were extracted with 5 mL 100% methanol, aided by sonication and for 30 min at room temperature using Branson CPX-952-518R (Branson Ultrasonics, Carouge, SA Switzerland). The extracts were then centrifuged (LC-04 C 80 − 2 C regen lab centrifuge, Zhejiang, China) at 12,000× g for 10 min. Three independent replicates (n = 3) were done for each sample to assess biological variance. Then, 100 µL of the methanol extract was evaporated to dryness in opened screw-cap vials by using stream of nitrogen gas. For derivatization, 150 µL of N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA) previously mixed (1:1) with anhydrous pyridine was added to the dried methanol extract and the mixture was incubated (Yamato Scientific DGS400 Oven, QTE TECHNOLOGIES, Hanoi, Vietnam) for 45 min at 60 °C. Silylated derivatives were separated on a Rtx-5MS Restek, Bellefonte, PA, USA (30-m length, 0.25-mm inner diameter and 0.25-m film)14,15.. SPME-GC/MS analysis of volatile constituents in C. caesia and C. longa Briefly, 20 mg of finely pulverized powder were placed in 1.5 ml SPME screw-cap vials quickly sealed and set on a tray that was kept at a constant 50 °C for 30 min. During that time, SPME fibers of stable flex coated with divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) 1 cm long of 50/30 µm purchased from Supelco® (Oakville, ON, Canada) were used for sampling by incubation in the headspace above the sample. The analysis of volatile compounds was carried out on a Schimadzu® GC-17 A gas chromatograph equipped with a DB-5 column (30 m x 0.25 mm i.d. x 0.25 μm film thickness; Supelco) and coupled to Schimadzu® QP5050A mass spectrometer14,15. Multivariate data analyses Both UPLC/MS/MS and GC-MS derived datasets for primary and secondary metabolites were subjected to unsupervised PCA and supervised OPLS-DA multivariate data analysis using SIMCA-P version 14.1 software package (Umetrics, Umeå, Sweden). All variables were mean-centered and scaled to Pareto variance. Unsupervised PCA was used to obtain a segregation pattern of the variance of metabolites among the samples. While the supervised OPLS-DA was used to verify PCA results, identify markers, and obtain detailed information on the distinctions among the studied specimens. The Chemometric models with several permutations were applied for models’ validation based on the main 2 indices, including R2 and Q2. Model predictability was indicated by Q2, and model goodness of fit was specified by R15. Quantification of total phenolic (TPC) and flavonoid contents (TFC) The TPC and TFC in rhizome methanolic extract of C. caesia and C. longa were analyzed using 96-well plates with a SPECTROstar® Nano Multi-Detection Microplate Reader (BMG Labtech, Ortenberg, Germany) following Folin-Ciocâlteu assay for TPC and aluminium chloride (AlCl3) colorimetric assay for TFC16,17. In detail, for TPC determination: 20 µl extract (2.0 mg/mL) was mixed with 100 µL Folin-Ciocâlteu reagent, and 80 µL sodium carbonate (7.5% w/v) ad incubated for 30 min at room temperature followed by measuring the absorbance at 760 nm using gallic acid as reference drug (0.011–0.154 mg/mL, R2 = 0.99). Results are expressed as mean of triplicates (n = 3) ± S.D as gallic acid equivalent per gram of freeze-dried powder (GAE/g w). For TFC determination: the absorbance of the mixture constituted of 40 µL extract (2.0 mg/mL), 40 µL sodium acetate (100 mg/mL) and 24 µL AlCl3 (25 mg/mL) in ethanol (total volume 200 µL) was measured at 430 nm (n = 3). Results were expressed as mean ± S.D. rutin equivalent per gram of freeze-dried powder (RE/g w). Calibration curve constructed of rutin at concentration range 0.025–0.125 mg/mL (R2 = 0.99). Bioactivity assays Antioxidant activity determination16 DPPH assay The assay was conducted following the same procedures described by Zayed et al.16,18,. Mixture of 30 µL extract and 270 µL 0.004% DPPH in methanol was incubated for 30 min. in darkness at room temperature followed by measuring its absorbance at 515 nm (n = 3). Results are expressed as mean ± S.D. mg trolox equivalent/gram weight of dried residue (mg TE/g w). ABTS assay The assay followed the exact procedures mentioned by Zayed et al.16,18,. The absorbance of the extract (2 mg/mL) and ABTS solution mixture was measured at 734 nm upon 30 min incubation at room temperature (n = 3). Results are expressed as mean ± S.D. mg trolox equivalent/gram weight of dried residue (mg TE/g w)18,19. FRAP assay The assay was performed as described by Zayed et al.16,18,. The absorbance of 20 µL extract (2 mg/mL) and 175 µL FRAP reagent mixture was measured at 593 nm upon 30 min. incubation at room temperature (n = 3). Results are expressed as mean ± S.D. mg trolox equivalent/gram weight of dried residue (mg TE/g w). Antidiabetic activity determination via α-glucosidase Inhibition The assay followed exact protocol in Zayed et al.18,20,. The mixture of extract (0.1–2 mg/mL), substrate, phosphate buffer, and yeast α-glucosidase was incubated at 37 °C before measuring the absorbance of the obtained color at 405 nm against blank and acarbose reference drug. % of Inhibition was calculated according to the following equation: The normalized logarithmic curve for % inhibition was used to express the data as IC50 ± SD (the concentration of the sample that could inhibit 50% of the enzyme for triplicates, n = 3). Anti-obesity assay determination via lipase Inhibition The assay was proceeded exactly as described by Zayed et al.18,. The crude porcine PL type II at 2.5 mg/mL (Sigma, EC 3.1.1.3) was suspended in 5 mM Tris–HCl buffer (pH 8.0, obtained using HCl). The mixture was centrifuged (6987.5×g, 8 min, 20.0 ± 1.0 °C), and the clear supernatant was isolated. Afterward, 40 µL of PL solution were pre-incubated for 15 min at 37 °C with 40 µL of varying concentrations (0.25–10 mg/mL) of each extract, then 4-nitrophenyl butyrate (p-NPB) substrate at a concentration of 10 mM in pure ethanol was added. The absorbance was measured spectrophotometrically at 405 nm in microplates, after another 5 min of incubation. Orlistat was used as a positive drug control for the assay18,21. Results and discussion UV fingerprinting of C. caesia and C. longa extracts Considering the clear color difference in both rhizome pulps that is blue versus yellowish, UV fingerprinting was employed initially for identification of the nature of coloring matter in blue curcuma. Blue curcuma was investigated for the presence of azulenes (conjugated 5- and 7-membered fused ring system) that is known to impart bluish coloration. Due to its non-polar nature, and lower solubility in methanol, extraction was attempted with chloroform for both yellow and blue curcuma rhizome powder for comparative analysis (Suppl. Fig. S1). Extraction using methanol failed to show absorbance spectra for these phytochemicals. From literature, azulenes show maximum absorbance at λmax 270–340 nm (< 350 nm) that corresponds to S₀ → S₂ transition (short wavelength)22, whereas, in this study blue curcuma showed an exclusive strong absorbance at longer wavelength ca. λmax 400 nm, that inferred the presence of guaiazulenes (alkylated derivatives of azulenes)23. Guaiazulenes are lipophilic bicyclic sesquiterpene hydrocarbons having distinctive bluish-purple color and common in planta23,24. The alkyl substitution in guaiazulenes viz. methyl and isopropyl moieties; donates electrons to the conjugated ring π- system, thus, reduces the gap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) i.e. ground and excited states, which in turn causes bathochromic shift in (S₀→S1) and (S₀→S₂) absorption22,24–26. NMR fingerprinting of C. caesia and C. longa methanol extracts NMR analysis was performed on C. caesia and C. longa extracts to elucidate major metabolites differences and also aided in standardization for quality control. Leveraging prior NMR-based literature, peaks corresponding to major metabolites in C. caesia (BT) and C. longa (YT) were identified. As summarized in Table 1, a diverse array of metabolites was classified into three major phytochemical groups, comprising 11 chemically distinct compounds: 5 sesquiterpenes (1–5), 3 curcuminoids (6–8), and 3 primary metabolites (9–11). The identified metabolites are shown in Suppl. Fig. S2. Table 1Resonance assignments with chemical shifts of constituents identified in Curcurma caesia (blue curcuma) and Curcurma longa (yellow curcuma) via 600 MHz1H NMR, HSQC and HMBC spectra.NoMetaboliteAssignment 1H (multiplicity)HSQCHMBCBlue curcumaYellow curcuma1. Sesquiterpenes 1 CurdioneCH3−130.84 d (J = 6.6 Hz)21.0C-7 (55.5), C-11 (30.7), C-12 (20.0)+-CH3−120.94 d (J = 7.0 Hz)20.0C-7 (55.5), C-11 (30.7), C-13 (21.0)CH3−140.94 d (J = 6.5 Hz)18.8C-3 (35.0), C-4 (47.0), C-5 (216.6) 2 XanthorrhizolCH3−91.17 d (J = 3.0 Hz)22.1C-5 (146.5), C-8 (43.0)C-10 (34.5)++CH3−14, 151.81 s18.7C-13 (134.2)CH3−72.07 s16.4C-2 (123.2), C-3 (139.5)H-82.48 m39.8H-125.34 m129.1 3 FuranodienoneH-151.26 s14.0C-9 (42.1), C-1 (132.2), C-10 (135.9)+-H-141.94 d (J = 1.2 Hz)17.6C-5 (133.2), C-4 (147.7)C-3 (41.0)H-132.06 s8.1C-11 (123.2), C-12 (139.5)H-15.16 dd (J = 5.0, 11.5 Hz)130.5H-55.87 s131.8C-6 (192.0), C-3 (41.0), C-14 (17.6) 4 CurcuzederoneH-132.06 s8.2C-11 (123.2), C-12 (139.5)++H-53.74 s55.1H-127.16 s139.5C-11 (123.2),C-8 (159.0) 5 Curcuminol GH-151.09 s18.5++H-141.83 s18.7H-131.87 s10.5C-11 (136.0), C-12 (180.0)2. Curcuminoids 6 Curcumin IH-4,4`7.55 d (J = 15.8 Hz)-+H-6,6`7.20 d (J = 1.7 Hz)H-10, 10`7.09 dd (J = 8.6, 1.7 Hz)H-3,3`6.59 d (J = 15.8 Hz) 7 Curcumin IIH-4,4`7.55 d (J = 15.8 Hz)-+H-67.20 d (J = 1.7 Hz)H-6`, 10`7.47 d (J = 8.6 Hz)H-7`, 9`6.80 d (J = 8.6 Hz)H-3,3`6.59 d (J = 15.8 Hz) 8 Curcumin IIIH-4,4`7.55 d (J = 15.8 Hz)-+H-6`, 10`7.47 d (J = 8.6 Hz)H-6, 107.47 d (J = 8.6 Hz)H-7`, 9`, 7, 96.79 d (J = 8.6 Hz)H-3,3`6.59 d (J = 15.8 Hz) 3. Primary Metabolites 9 MethionineS-CH32.07 s17.5CH2 (29.9)++NH2-CH3.30 m48.5 10 CholineN(CH3)33.18 s47.5++ 11 Fumaric acidCH = CH6.64 br s134.2++ Resonance assignments with chemical shifts of constituents identified in Curcurma caesia (blue curcuma) and Curcurma longa (yellow curcuma) via 600 MHz1H NMR, HSQC and HMBC spectra. C-5 (146.5), C-8 (43.0) C-10 (34.5) C-5 (133.2), C-4 (147.7) C-3 (41.0) C-11 (123.2), C-8 (159.0) Sesquiterpenes Sesquiterpenes (15-carbon terpenoids) represent a structurally diverse class of secondary metabolites. Curdione, exclusive to C. caesia, displays three methyl doublets (CH₃−13 at δH 0.84, J = 6.6 Hz; CH₃−12 and CH₃−14 at δ δH 0.94), confirming its germacrane skeleton. Key HMBC correlations from CH₃−14 to the ketone C-5 (δC 216.6) validate ring fusion27,28. Xanthorrhizol, present in both species, exhibits a trisubstituted olefin (H-12, δH 5.34) and characteristic methyl signals: a doublet (CH₃−9, δH 1.17, J = 3.0 Hz) and singlets (allylic CH₃−14/15, δH 1.81; vinylic CH₃−7, δH 2.07). HMBC correlations link CH₃−7 to olefinic C-2 (δC 123.2) and C-3 (δC 139.5), confirming its bisabolane core29,30. Furanodienone (both species) features a furan ring evidenced by H-5 (δH 5.87, s) and methyl groups (H-13, δH 2.06; H-14, δH 1.94, d, J = 1.2 Hz), with HMBC correlating H-14 to furan carbons C-3 (δC 41.0), C-4 (δC 147.7) and C-5 (δC 133.2) (Oon et al., 2016). Curcuzederone (both species) showed a phenolic structure with an aromatic proton (H-12, δH 7.16, s) and methoxy group (H-5, δH 3.74, s). Curcuminol G, exclusive to C. caesia, is identified by three methyl singlets (H-13–H-15, δH 1.09–1.87) and an HMBC correlation from H-13 to a carbonyl (C-12, δC 180.0), suggesting a sesquiterpenoid acid/ester (Fig. 1A and B, Suppl. Fig. S3 and S4, Table 1)27–32. 1H-NMR spectra (CD3OD, 500 MHz) of Curcuma caesia (A, B: δH: 0–8 ppm) and Curcuma longa (C, δH: 6.5–8 ppm) used in the curent study. The spectra are labelled with maor key metabolites, while peaks assignments are summerized in Table 1. Curcuminoids Curcuminoids, diarylheptanoid pigments exclusive to C. longa, share diagnostic trans-olefinic protons: H-3/H-3′ (δH 6.59, d, J = 15.8 Hz) and H-4/H-4′ (δH 7.55, d, J = 15.8 Hz), confirming the conjugated diene system (Khattab et al., 2022). Curcumin I displays asymmetric meta-substituted aromatics: H-6/H-6′ (δH 7.20, d, J = 1.7 Hz) and H-10/H-10′ (δH 7.09, dd, J = 8.6, 1.7 Hz). Curcumin II shows mixed substitution: one ring with meta-coupling (H-6, δH 7.20) and the other with ortho-coupling (H-6′/H-10′, δH 7.47, d, J = 8.6 Hz; H-7′/H-9′, δH 6.80, d, J = 8.6 Hz). Curcumin III exhibits symmetric ortho-substitution: equivalent aromatic protons for both rings (H-6/H-10 & H-6′/H-10′ at δH 7.47; H-7/H-9 & H-7′/H-9′ at δH 6.79). Their absence in C. caesia explains its lack of yellow pigmentation (Fig. 1C; Table 1)27,33,34. Primary metabolites Primary metabolites, conserved in both species, reflect essential cellular functions, and moreover nutritive value in both rhizomes. Methionine was identified by its sulfur-methyl group (S-CH₃, δH 2.07, s) and α-methine proton (NH₂-CH, δH 3.30, m), with HMBC correlating S-CH₃ to β-methylene (δC 29.9) (Wang et al., 2015). Choline showed a sharp singlet for its trimethylammonium group [N(CH₃)₃, δ 3.18, s]. Fumaric acid exhibits a broad singlet (δH 6.64) for its equivalent olefinic protons (CH = CH), consistent with its symmetric trans-structure. These compounds underpin universal metabolic roles: methionine in methylation, choline in membrane integrity, and fumaric acid in the Krebs cycle (Fig. 1A and B; Table 1)27,30,31. Quantitative1H-NMR analysis of key metabolites in C. caesia (BT), and C. longa (YT) 1H-NMR (qNMR) was employed to determine the absolute concentrations (µg/g) of key metabolites (7 metabolites out of 11 identified metabolites) underscores profound metabolic divergence between C. caesia (BT) and C. longa (YT), revealing species-specific biochemical strategies (Suppl.TableS1) BT’s exclusive production of curdione (1.1 µg/g, germacrane-type) and furanodienone (1.4 µg/g, furanosesquiterpenoid) aligns with its traditional use in anti-inflammatory therapies, while the absence of these in YT highlights evolutionary pathway specialization35. Conversely, xanthorrhizol is more abundant in BT (1.5 µg/g) than in YT (1.4 µg/g; bisabolane-type), whereas YT’s exclusive accumulation of curcumin I (0.93 µg/g) directly correlates with its antioxidant and iconic pigmentation absent in BT due to silenced diarylheptanoid biosynthesis (Jantan et al., 2012). Notably, Curcuzederone exhibits preferential accumulation in BT (0.9 µg/g vs. 0.3 µg/g in YT), suggesting differential regulation of phenolic sesquiterpenoid pathways, whereas near-identical curcuminol G levels (0.3 µg/g BT vs. 0.1 µg/g YT) imply conserved enzymatic machinery for this sesquiterpenoid acid. The exclusive detection of choline in BT (0.1 µg/g) further indicates distinct primary metabolism, potentially supporting membrane biosynthesis in stress adaptation. These quantified disparities rooted in proton-specific integrals (e.g., 3 H for methyl groups vs. 1H for olefinics) provide not only chemotaxonomic markers for species authentication but also a biochemical rationale for their ethnopharmacological differentiation, with BT as a sesquiterpene-rich analgesic, and YT as a curcuminoid-centric antioxidant36–38. Metabolites profiling of C. caesia (BT) and C. longa (YT) rhizome methanol extracts via UPLC/HR-MS/MS analysis in both ionization modes Metabolites profiling of BT and YT via UPLC/HR-MS/MS analysis in both negative and positive modes provided improved secondary metabolites coverage for less abundant secondary metabolites not detected using NMR39. Gradient system of water (0.1% formic acid): acetonitrile eluted metabolites from the most to the least polar throughout a 20 min. runtime. Annotation of metabolites was based on MS data, including molecular ion, product ions, respective elemental formulae (error < 5 ppm), and fragmentation pattern; in accordance with previously published literature and databases e.g., Reaxys, HMDB, FooDB, Pubchem, and phytochemical dictionary of natural products, in addition to predicted spectra generated by CFM-ID (https://cfmid.wishartlab.com). Base peak chromatograms (BPC) of Curcuma sp. in both modes are shown in Fig. 2. The identified metabolites are listed in Table 2 based on their classes. Structures of selected metabolites annotated in both species are demonstrated in Fig. 3. Base peak chromatograms of C. Caesia (blue curcuma) and C. longa (yellow curcuma) crude methanol extracts analyzed via UPLC/HR-MS/MS in both negative and positive modes. Table 2Metabolites profiling of Curcuma caesia (blue curcuma) and Curcuma longa (yellow curcuma) rhizome methanol extracts via UPLC/HR-MS/MS analysis in both negative and positive ionization modes.PeakRt (min.)IdentificationMolecular ion [M-H]−/[M + H]+Elemental compositionError (ppm)Fragments MSn ionBlue CurcumaYellow Curcuma I- Sugars 10.59Hexose179.0547C6H11O6−1.6161, 135, 99, 87, 71, 59(+)(+)20.66Lactobionic acid357.1030C12H21O12−0.7339, 177, 119, 101, 89, 87, 71, 59(+)(+)30.68Dihexosyl-O-glycerol415.1444C15H27O13−0.4341, 235, 179, 161, 101, 89, 71(+)40.69Trihexoside503.1614C18H31O16−1.5341, 179, 161(+)50.7Dihexoside341.1082C12H21O11−1.1179, 161, 101, 89, 71(+)(+)60.71Sugar derivative267.0712C9H15O9−0.8249, 237, 219, 150, 125, 113, 101, 89, 71, 59(+)(+)71Dipentoside281.0875C10 H17 O9−2113, 96, 87, 85, 71, 59(+) II- Organic acids 80.9Malic acid133.0128C4H5O5−1.9115, 71(+)(+)90.96Citric acid191.0188C6H7O7−0.9173, 129, 111, 103, 87(+)(+)101.58Succinic acid117.0179C4H5O4−2.699, 73, 55(+)111.7Benzoic acid121.0280C7H5O2−2.5108, 101, 93, 91(+) III- Phenolics 121.6Salicylic acid137.0229C7H5O3−2101, 93, 87, 66(+)133.1Gallic acid-O-hexoside331.0663C13H15O10−1.1315, 168, 150, 125, 89(+)143.2Hydroxy-methoxyphenyl-O-hexoside301.0920C13H17O8−0.8138, 123, 97(+)153.53Protocatechuic acid-O-hexoside315.0715C13H15O9−1.1153, 135, 108, 91(+)163.7Hydroxy-dimethoxyphenyl-O-hexoside = (Dimethoxy-hydroquinone-O-hexoside) = Leonuriside A331.1027C14H19O9−0.4168, 153, 138, 125, 97, 93, 83, 71, 59(+)173.72Protocatechuic acid153.0183C7H5O4−0.4135, 108, 91(+)(+)184Salicin285.0973C13H17O7−1.5267, 195, 177, 123(+)194.06Hydroxyphenyl propanoic acid-O-hexoside343.1028C15H19O9−1.3181, 163, 137, 119, 101, 89, 71(+)204.08Syringic acid-O-hexoside359.0977C15H19O10−1.3197, 182, 167, 153, 138, 123, 71(+)(+)214.7Eugenol-O-hexoside325.1284C16H21O7−0.7163(+)(+) IV- Flavonoids/anthocyanins 224.6Epi/Catechin289.0710C15H13O6−1.4245, 205, 151(+)(+)235.65Malvidin-O-rutinoside639.1912C29H35O16+1493, 331, 150(+)245.7Quercetin-O-dirhamnoside593.1498C27H29O15−0.4300(+)256.88Di-O-methyl-epi/catechin-O-hexoside479.1552C23H27O11−0.48317, 285, 255, 151, 131(+)2610.3Naringenin271.0606C15H11O5−1.6253, 177, 151(+)2712.7Pinocembrin255.0654C15H11O4−1.07213, 151(+) V- Cinnamates 283.95Hydroxy-ferulic acid-O-hexoside373.1131C16H21O10−0.5211, 193, 165, 149, 123, 105, 73(+)294.85Caffeic acid179.0335C9H7O4−2135(+)(+)307.3Cinnamic acid147.0437C9H7O2−1129, 103, 91, 77(+)318.2Feruloyloxy-methoxycinnamic acid339.0862C19H15O6−0.1193(+)3213.35Methylbutenoxy ferulic acid261.1124C15H17O4−1.1217, 193, 189, 149, 137, 121, 105, 81, 79, 67(+) VI-Alkaloids 333.72Dihydroxyquinoline carboxylic acid-O-hexoside366.0824C16H16NO9−0.1204, 185, 173, 160, 130, 107, 102, 93, 87, 73, 68(+)(+)344Trihydroxyquinoline-2-carboxylic acid-O-hexoside382.0771C16H16NO10−0.6366, 351, 220, 176, 121(+)(+) VII- Di/Aryl alkanoids (Curcuminoids) 359.5Dihydroxy-(hydroxy-dimethoxyphenyl)-(hydroxy-methoxyphenyl) heptane405.1879C22H29O7−0.02390, 349, 192, 179, 165, 151, 135, 125, 109(+)(+)369.83Bis(hydroxyphenyl)-hydroxyheptenone313.1437C19H21O4+1207, 189, 163, 149, 119(+)379.86Calebin A385.1274C21H21O7+1.9193, 191, 177, 161, 149, 145, 133, 97(+)389.9(Hydroxyphenyl)-(Dihydroxyphenyl)-heptadiene-dione323.0916C19H15O5−0.6217, 189, 174, 161, 146, 133, 119, 109, 59(+)3910.32Curcumalongin A353.1015C20H17O6+1.5338, 321, 293, 269, 171, 153, 147(+)4010.35Hexahydrocurcumin373.1647C21H25O6−0.5193, 179, 165, 121(+)4110.48(Dihydroxybenzylidene)-(hydroxy-methoxystyryl) furanone353.1016C20H17O6+0.8338, 337, 177, 153, 150, 123(+)4210.6Curcumalongin B383.1117C21H19O7+1.8368, 350, 294, 177, 153, 145, 121(+)4310.7Curcumalongin C385.1274C21H21O7+1.8269, 219, 195, 193, 177, 161, 153, 145, 133, 117(+)4410.91(Hydroxy-methoxyphenyl)-(hydroxyphenyl)-hydroxyheptadieneone339.1230C20H19O5−1219, 189, 175, 161, 149, 134, 119, 69(+)4511.45Curcolone245.1174C15H17O3−1227, 203, 161(+)(+)4611.48Tetrahydrodemethoxycurcumin341.1384C20H21O5−0.4235, 205, 193, 179, 163, 135, 99(+)(+)4711.5Tetrahydrocurcumin371.1491C21H23O6−0.5235, 193, 178, 135, 99(+)(+)4811.7(Hydroxy-methoxybenzylidene)-(hydroxystyryl) furanone337.1073C20H17O5+1321, 305, 160, 147, 137, 123(+)4912.4Bis(hydroxy-methoxyphenyl)-heptatrien-3-one353.1378C21H21O5+1.4285, 259, 225, 197, 177, 161, 145, 137(+)5012.84Dihydrobisdemethoxycurcumin309.1124C19H17O4−0.85203, 189, 161, 145, 119, 93(+)5112.88Curcumin III, (Bisdemethoxycurcumin)309.1116C19H17O4−1.9189, 147, 131, 119(+)5212.95Bis(dimethoxyphenyl)-pentadienone355.1532C21H23O5+2265, 219, 201, 179, 177, 163, 145, 137(+)5313.05Curcumin II (Demethoxycurumin)339.1221C20H19O5+1.7269, 255, 223, 195, 177, 147, 119(+)(+)5413.1Curcumin I369.1326C21H21O6+0.4217, 177, 161, 145, 137, 117(+)(+)5513.25Methoxycurcumin397.1281C22H21O7−0.9247, 217, 203, 179, 173, 158, 149, 134(+)5614.53ar-Turmerone217.1579C15H21O+3119, 103, 91(+)5714.9Dihydro-ar-Turmerone219.174C15H23O+2.5121, 105, 93, 83, 79, 55(+) VIII- Sesquiterpenes 583.8Dihydro zedoarolide B-O-hexoside445.2061C21H33O10−0.6283, 265, 247, 239, 221, 219, 203, 185, 167, 161, 151, 133, 111, 71(+)594.16Trihydroxy-guaianenone429.2141C21H33O9−1.04411, 393, 267, 249, 231, 179, 161, 142, 139, 119, 89, 71(+)604.3Dihydro derivative of Zedoarolide A297.1331C15H21O6−0.1279, 261, 235, 207, 199, 165, 127, 83(+)614.4Zedoarolide B-O-hexoside443.1911C21H31O10−0.03281, 263, 249, 237, 223, 219, 179, 161, 89, 71(+)624.8Zedoalactone A/C/E-O-hexoside427.1966C21H31O9−0.1265, 247, 235, 207, 189, 161(+)635.02Zedoarolide A295.1178C15H19O6−0.7277, 259, 233, 215, 153, 125(+)645.05Hydroxy-(hydroxymethyl)-dimethyl-tetrahydroazulenofuranone (wenyujinin F)261.1126C15H17O4−1.8219, 217, 201, 199, 191, 175, 163, 159, 135, 123, 109, 95, 85, 65, 57(+)655.07Zedoalactone B/D279.1229C15H19O5−0.7261, 243, 235, 217, 199, 177, 163, 153, 137, 109, 95(+)(+)665.8Zedoarolide B281.1387C15H21O5−1.4263, 250, 237, 219, 207, 201, 189, 171, 139, 123, 83(+)(+)676.9Aerugidiol-O-hexoside411.2014C21H31O8−0.1249, 231, 213, 191, 179, 161,(+)687.1Unknown sesquiterpene lactone277.1074C15H17O5−1.4247, 233, 203, 185, 177, 161, 145, 133(+)698Dihydro derivative of Zedoarolide B283.1543C15H23O5−1.05265, 239, 221, 219, 203, 185, 167, 151, 133(+)7011.3Hexahydro-dihydroxy-trimethylazulenofuranone (wenyujinolides A/B)263.1281C15H19O4−1.4245, 201, 219, 191, 149,137, 125, 109, 97(+)7112.5Zedoalactone A/C/E265.1452C15H21O4−0.2221, 191, 147(+)7212.9Germacrone epoxide235.1685C15H23O2+2217, 207, 193, 189, 175, 161, 147, 135, 119, 107, 105, 93, 79, 69, 67(+)7312.97Furandiene217.1585C15H21O+0.8199, 189, 175, 161, 135, 119, 107, 105, 93, 79, 69(+)7413Confertin247.1330C15H19O3−0.1221, 203, 178, 149, 135, 111, 95, 57(+)7513.02Curcumenol235.1685C15H23O2+2.3217, 199, 189, 179, 177, 161, 149, 121, 119, 113, 107, 105, 97, 95, 93, 85, 79, 69, 65, 55(+)(+)7613.14Curcumenone235.16878C15H23O2+0.4207, 189, 165, 151, 149, 123, 119, 107, 83, 71(+)7713.32Zederone247.1319C15H19O3+3229, 201, 183, 139, 123, 121, 107, 95, 81, 67, 65(+)(+)7813.7Curdione237.1843C15H25O2+2.3221, 219, 201, 191, 135, 121, 109, 107, 93, 81, 69, 55(+)(+)7913.76Germacrone/Zerumbone219.1739C15H23O+1.9203, 159, 135, 121, 107, 97, 93, 79, 69(+)(+)8014Aerugidiol249.1493C15H21O3−3231, 223, 213, 205, 189, 184, 177, 165, 147, 113, 87, 79.9, 68, 55(+)(+)8114.14Epi/Curzerenone231.1374C15H19O2+2.1213, 203, 198, 189, 183, 173, 161, 149, 135, 119, 105, 95, 83(+) IX- Fatty acids/esters 826.61Suberic acid173.0806C8H13O4−0.8129, 111, 83, 67, 57(+)(+)837.99Azelaic acid187.0963C9H15O4−0.6169, 143, 125(+)8411.4Trihydroxyoctadecenoic329.2327C18H33O5−1.4229, 211, 183, 171(+)8512.8Trihydroxyoctadecanoic acid331.2482C18H35O5−1313, 295, 277, 185, 127(+)8613.3Gingerglycolipid A675.3589C33H55O14−0.4415, 397, 277, 255, 235, 161, 101(+)8713.34Dihydroxyoctadecadienoic acid311.2219C18H31O4−0.9293, 275, 249, 183, 171(+)8813.4Hydroxy-oxopentadecanoic acid271.1908C15H27O4−1.7253, 227, 209, 97(+)(+)8913.56Hydroxy-oxopentadecadienoic acid267.1594C15H23O4−1.4249, 223, 205, 197, 181, 137, 97, 85(+)9013.79Hydroxy-oxohexadecanoic acid285.2064C16H29O4−1.3267, 223, 59(+)9114.1Glyceropalmitoyl-O-dihexoside653.3743C31H57O14−0.07415, 397, 323, 255, 179, 89(+)9214.18Palmitoyl-sn-glycero-phosphoethanolamine (LysoPE(0:0/16:0)452.2775C21H43NO7P−0.8339, 255, 196(+)9314.2Oleoyl glycero-phosphoethanolamine478.2931C23H45NO7P−0.7339, 281, 196(+)9414.23Hydroxy octadecadienoic acid295.2271C18H31O3−1.2277, 195(+)(+)9515Hydroxypalmitic acid271.2270C16H31O3−1253, 225(+)9617Oleoyl-O-methyl-glycero-phosphate449.2666C22H42 O7P−0.8281, 167(+)9717.02Dilinoleoyl-glycero-phosphate695.4646C39H68 O8P−0.04415, 279, 153(+)9817.04Palmitoyl-linoleoyl-glycero-phosphoinositol833.5173C43H78 O13P−0.15553, 391, 279, 255(+)(+)9917.06Dioleoylglycerophosphate697.4793C39H69 O8P−0.53415, 280(+)(+)10017.07Linoleoyl-glycerophosphate433.2352C21H38 O7P−0.6279, 153(+)10117.1Palmitoyl-linoleoyl-glycero-phosphate671.4642C37H68 O8P−0.5415, 391, 279, 255(+)(+)10217.2Palmitoyl-oleoyl-glycerophosphoinositol835.5312C43H80O13P−2553, 281, 255(+)(+) X- Miscellaneous 1031.1Uridine243.0614C9H11O6N2−0.9183, 153, 140, 113, 110(+)1043.12Methyl Uridine257.0772C10H13O6N2−1.6228, 214, 183, 155, 139, 124, 109(+)1053.59Panthenol204.1227C9H18NO4−1.6125, 102, 71, 59(+)1063.71Pyroglutamic acid128.0339C5H6NO3−298, 81, 65(+)1074.1Unknown271.1181C13H19O6−1.7193, 177, 135(+)(+)1084.9Unknown425.2019C18H33O11−0.5293, 263, 209(+)1095.81Hydroxy-trimethyl-cyclohexenoic acid-O-hexoside (picrocrocinic acid)345.1547C16H25O8−1.03179, 165, 161, 149, 137, 123, 101, 89(+)1105.82Dihydroxy-megastigmadien-one-O-hexoside (blumenol hexoside)385.1866C19H29O8−2.4223, 205, 161, 153, 101(+) Metabolites profiling of Curcuma caesia (blue curcuma) and Curcuma longa (yellow curcuma) rhizome methanol extracts via UPLC/HR-MS/MS analysis in both negative and positive ionization modes. Structures of selected metabolites tentatively identified via UPLC/HR-MS/MS analysis in C. caesia and C. longa extracts as shown in Table 1. Overall, UPLC/MS analysis revealed 110 metabolites including 7 sugars, 4 organic acids, 10 phenolics, 5 cinnamates, 5 flavonoids, 1 anthocyanin, 2 alkaloids, 23 di/aryl alkanoids, 24 sesquiterpenes, 21 fatty acids/glycerolipids, and 8 miscellaneous compounds (Table 2). Common fragmentation pathways in MS/MS spectra included dehydration (-H2O, 18 amu), deglycosilation (-C6H10O5, 162 amu for hexose, C5H8O4, 132 amu for pentose, -C6H10O4, 146 amu for deoxyhexose), decarboxylation (-COO, 44 amu), demethoxylation (-OCH3, 30 amu), decarbonylation (-CO, 28 amu), dealkylation, retro-Diels-Alder (RDA), and ring cleavage, as discussed in detail in the next subsections. UPLC/HR-MS/MS analysis provided insight into the chemical reservoir of C. caesia with several detected metabolites are reported herein for the first time, and adding to that taxa rich chemical composition. Organic acids Low molecular weight organic acids play a role in rhizomes to alleviate microbial stress and reduce heavy metals absorption40. Herein, 4 organic acids were detected in blue curcuma, viz. malic, citric, succinic and benzoic acids in peaks 1–4 early eluted at Rt range 0.9–1.7 min (Table 2)4. Observed fragments included decarboxylated and dehydrated product ions matching databases. For example, citric acid [M-H]−191.01881, C6H7O7− showed dehydrated ion at m/z 173, C6H5O6−, followed by decarboxylation at m/z 129, C5H5O4−, and further fragments at m/z 111, m/z 103, and m/z 87. Phenolics, flavonoids, and anthocyanins Phenolics alongside flavonoids are biosynthesized in plants as secondary metabolites mainly to filter or screen ultraviolet rays, alleviating its harmful oxidative stress, and as attractants for pollinators; thus, their abundance in rhizomes is limited relative to leaves41. In Curcuma, only 14 flavonoids were previously reported2. As shown in Tables 2 and 10 phenolics, 3 flavonoids, and 1 anthocyanin were annotated in C. caesia versus 3 phenolics and 3 flavonoids in C. longa; matching references reported on genus Curcuma3,6,42. Interestingly, a p-hydroquinone derivative was detected exclusively in C. caesia, peak 16, at [M-H]− 331.10275, C14H19O9− for the first time (Table 2). Such compounds are potent ferric-ion reductants to its GIT-absorbable ferrous form, decolorizing agents, skin lightening agents, anti-inflammatory and potent free radical scavengers43–47. Fragmentation pathways yielded deglycosilated product ion at m/z168, C8H8O4−, followed by sequential demethylation at m/z 153, C7H5O4− and m/z 138, C6H2O4−, further fragments at m/z 125, m/z 97, m/z 93, m/z 83, m/z 71 and m/z 59 corresponding to dehydroxylation and hydroquinone ring cleavage, matching the spectrum of leonuriside A, that is dimethoxy-hydroquinone-O-hexoside (Suppl.Fig.S5), as reported in literature and databases48. Flavans and flavanones were major flavonoid subclasses detected in C. caesia in peaks 22, 25 and 26, whereas, C. longa had additional flavonols in peaks 22, 24 and 27 (Table 2)2,42. Pinocembrin flavanone in peak 27 was annotated herein in C. longa extract only at [M-H]− 255.0654, C15H11O4−, known to exert several health benefits49,50. In contrast, malvidin-O-rutinoside, detected in peak 23 exclusively in C. caesia for the first time, is an anthocyanin that imparts violet color and was previously reported in C. alismatifolia pink bracts2, and could contribute to the blue coloration of C. caesia alongside azulenes (Suppl.Fig.S1). The molecular ion was detected only in positive ionization mode at [M + H]+ 639.19128, C29H35O16+ (Suppl. Fig. S6), which yielded fragments at m/z 493, C23H25O12+ and m/z 331, C17H15O7+ for the sequential deglycosilation of deoxyhexose and hexose sugars, respectively, followed by RDA fragmentation of malvidin at m/z 15351 (Table 2). Cinnamates In this study, 5 cinnamic acid derivatives were detected in peaks 28–32 (Table 2). MS/MS Spectra in peaks 28, 31 and 32 showed feruloyl fragment ion at m/z 193, C10H9O4−, while peaks 29 and 30 had molecular ion at [M-H]− 179.0335, C9H7O4− and [M-H]− 147.04379, C9H7O2−, identified as caffeic and cinnamic acids, respectively. Likewise, ferulic acid derivative in peak 31 annotated as feruloyloxy-methoxycinnamic acid at [M-H]−339.0862, C19H15O6−, was previously reported in Curcuma but first time in C. longa rhizome6. Another feruloyl derivative in peak 32 in C. caesia was annotated as oxyprenylated derivative of ferulic acid, namely, methylbutenoxy-ferulic acid, reported for the first time in Curcuma taxa. In detail, molecular ion was detected at [M-H]− 261.1124, C15H17O4− that yielded upon fragmentation several product ions viz. decarboxylated ion at m/z 217, C14H17O2−, feruloyl moiety at m/z 193, C10H9O4− upon the loss of the prenyl side chain (C5H8), m/z 149 C9H9O2− (-OCH3 & loss of oxyprenyl sidechain), m/z 121, C8H9O− and m/z 105 C8H9− (-CO2, -OCH3 &loss of pentenyl side chain) followed by dealkylation at m/z 79, C6H7− whereas, the isopentenyl side chain was detected at m/z 81, C5H5O− and its dealkylated fragment at m/z 67, C4H3O− (Table 1, Suppl.Fig.S7). Alkaloids In previous studies, C. longa was reported to yield hydroxyquinoline alkaloids2,6. Herein, 2 hydroxyquinoline alkaloids are reported for the first time in both blue and yellow curcuma as shown in peaks 33 and 34 at [M-H]− 366.0824, C16H16NO9− and [M-H]− 382.0771, C16H16NO10−annotated as di- and tri-hydroxyquinoline carboxylic acid-O-hexoside, respectively (Table 2). Fragments aiding in identification included deglycosilation (−162 amu, C6H10O5), decarboxylation (−44 amu, CO2) and quinoline ring cleavage matching spectra on databases52 (Suppl.Fig.S8). Similar structures were reported previously in Ephedra and cereals to which beneficial health promoting effect was attributed53,54. Di/Aryl alkanoids (Curcuminoids, bisabolane-sesquiterpene derivatives) Aryl and diaryl alkanoids are characteristic polyphenols in Curcuma, particularly in C. longa. The curcuminoids are diaryl-heptanoids, classified as bisabolane-type of sesquiterpenes. It imparts the distinguished yellow color and spice flavor of yellow curcuma and to which several therapeutic effects are attributed3. In this study, 23 di/aryl alkanoids were detected, predominantly in C. longa rhizome as shown in peaks 35–57 through Rt range 9.5–14.9 min, and were indeed previously reported in yellow curcuma2,3,6,42,50. In contrast, only 9 curcuminoids were detected in C. caesia as in peaks 35, 40, 45, 46, 47, 52, 53 and 54 (Table 2), and suggestive for activation of cucruminoids biosynthesis in C. longa as official drug in that taxa. Other diaryl alkanoids included furanone and pentadienone derivatives as in peaks 39, 41–43, 48 and 52 (Table 2). Identification was based on reported fragmentation pathways and diagnostic product ions as explained in detail by Rasheed, et al., and Quirós-Fallas, et al.3,50, (Suppl. Fig. S9). For example, C3- and C5- di-keto curcuminoids such as curcumin I, II, III and their analogues, undergo heptanoid chain cleavage at C4, followed by decarbonylation (-CO, 28 amu), demethylation and dehydroxylation (-CH3-OH, 32 amu) to yield diagnostic ions, depending on aryl substitution, commonly detected at m/z 177, m/z 147, m/z 145, m/z 1193,50. In case of a single C3-keto alkanoids as in peaks 49 and 52, diagnostic product ion at m/z 137 for the neutral loss of substituted aryl moiety was observed as a result of precursor ion rearrangement followed by chain cleavage at C1-C2 bond3,50. In mono- or di-hydroxylated heptanoids, dehydrated product ion was detected (-H2O, 18 amu). Whereas, cucruminoids having furanone ring in the central chain as curcumalongin A/B/C and others, as in peaks 39, 41–43 and 48 (Table 2), fragmentation pathway included cleavage of the 5-membered furanone ring preceded by precursor ion rearrangement to yield product ion at m/z 153 or m/z 123 3. Finally, the C3-ester containing curcuminoids as in calebin A, peak 37 (Table 2) had cleavage at the ester linkage releasing diagnostic ions at m/z 177 and m/z 207 3. The mono-aryl heptanoids were detected exclusively in yellow curcuma at [M + H]+ 217.1579; C15H21O+ and 219.174; C15H23O+ in peaks 56 and 57, and were identified as turmerone and dihydroturmerone having diagnostic ions at m/z 119 and m/z121, respectively50, (Table 2). Sesquiterpenes Sesquiterpenes are the major phytochemicals detected in genus Curcuma, where 328 compounds belonging to various subclasses were reported among 28 species2. In this study, 24 sesquiterpenes were tentatively identified in blue curcuma; peaks 58–81, several of which are reported for the first time in C. caesia. In contrast, only 7 of them were detected in yellow curcuma, peaks 65, 66,75, 77–80 (Table 2). Herein, sesquiterpenes were 5 germacrane (peaks 72, 73, 77–79), 1 elemane (peak 81), 1 carabrane (peak 76) and 17 guainane (guainolide) skeletons (Table 2) aside from the bisabaolane-type that was discussed in the previous subsection 3.2.52,9,55. C. longa had 2 types only i.e., 3 germacrane and 4 guainane. In contrast, C. caesia possessed the 4 subtypes of sesquiterpenes consistent with previously reported literature (Fig. 3)2,9,55,56. MS/MS Spectra distinguished between sesquiterpenes via observing product ions and their respective elemental formulae, that correspond to dehydration, decarboxylation, dealkylation and, most importantly, RDA fragmentation that allowed to discriminate between the different subtypes. For example, peaks 72, 75 and 76 had the same molecular ion and formula at [M + H] + 235.1685, C15H23O2+; the first, peak 72, showed fragments for the loss of water, carbonyl and isopropenyl side chain at m/z 217; C15H21O+ (-H2O, 18 amu), m/z 207; C14H23O+ (-CO, 28 amu), m/z 189; C14H21+ (-H2O and -CO), m/z 193; C12H17O2+ (- isopropenyl C3H6+) and m/z 147; C11H15+ (loss of H2O, CO and isopropenyl) followed by sequential delalkylation at m/z 135, m/z 121, m/z 107, m/z 93 and m/z 79 which is consistent with germacrone epoxide (germacrane type) (Suppl. Fig. S10A). The second, peak 75, showed dehydrated ion at m/z 217; C15H21O+, pentyl-ring cleavage at m/z 179; C11 H15O2+, and m/z 55; C4H7+, RDA fragmentation of heptyl-ring at m/z 113; C6H9O+ 2, m/z 97; C6H9O+ and m/z 69; C5H9+, which is consistent with the guainane-type sesquiterpene curcumenol (Suppl. Fig. S10B). Whereas, the carabrane-type curcumenone in peak 76 showed product ions for decarbonylation at m/z 207; C14H23O+ (-CO, 28 amu), butanone-side chain cleavage at m/z 165; C11H17O+ and m/z 71; C4H7O+, propyl-ring cleavage and loss of butanone-side chain at m/z 151, C10H15O+ and m/z 85, loss of isopropenyl and butanone side chain at m/z 123; C8H11O+ (Suppl. Fig. S10C). In peak 81, epi/curzerenone is an elemane sesquiterpene that was previously reported in C. caesia9. The precursor ion was detected at [M + H]+ 231.1374; C15H19O2+, which upon fragmentation resulted in diagnostic product ions at m/z 123; C7H7O2+ and m/z 109; C8H13+ for the hexyl-ring cleavage, m/z 173; C12H13O+ and m/z 55; C3H3O+ for furanone ring cleavage, m/z 161 C10H9O2+ for the loss of isopropenyl and ethenyl substituents followed by further dealkylation at m/z 149 and m/z 137. Furthermore, product ion corresponding to the furanone ring was detected at m/z 83; C5H7O+, and the hexanone ring moiety at m/z 95; C6H7O+. Zedoalactones and zedoarolides are guaianolides that were previously reported in Curcuma, including C. longa50. Whereas, the dihydro-derivative of zedoarolide B, peak 69, detected in C. caesia is, according to our knowledge, a newly detected metabolite (Table 2). The precursor ion was detected in negative ionization mode at [M-H]− 283.1543; C15H23O5−, having extra 2-hydogen atoms than zedoarolide B, peak 66, that was previously reported in C. phaeocaulis and C. zedoaria2,6,55. Fragmentation yielded product ions corresponding to decarboxylation (-CO2, 44 amu) and 3 successive dehydration (−3 × 18 amu) at m/z 265; C15H21O4−, m/z 239; C14H23O3−, m/z 221; C14H21O2−, m/z 203; C14H19O− and m/z 185; C14H17− (Suppl. Fig.S11). Other fragments included lactone ring cleavage at m/z 193, C12H17O2−, followed by sequential dealkylation and dehydration at m/z 169, C10H17O2−, m/z 151; C10H15O− and m/z 133, C10H13−. The myriad chemical reservoir of sesquiterpenes in C. caesia is a special feature that definitely has an impact on its therapeutic effects, particularly, anticancer, anti-inflammatory, antimicrobial and hepatoprotective bioactivities among several others that are reported on each of its subclasses via different mechanisms of action55. Further studies are recommended to compare the bioactivity of the crude extract containing mixture of sesquiterpenes with that of the isolated components or subfractions, as well as conducting chronic and acute toxicity studies to determine its recommended daily intake dose. Fatty acids (FA)/Glycerolipids (GL) Fatty constituents were one of the major classes detected herein represented by 21 metabolites in negative ionization mode as in peaks 82–103 and were eluted at wide Rt range 6.6–17.2 min according to their relative polarity i.e. short-chain dicarboxylated/oxygenated FA were eluted earlier in contrast to long-chain FA glycerates (Table 2). Several metabolites were derivatives of palmitic, linoleic and oleic acids as evidenced by product ions at m/z 255; C16H31O2−, m/z 279; C18H31O2− and m/z 281; C18H33O2−, respectively. Fragmentation included dehydration (-H2O, 18 amu), decarboxylation (-COO, 44 amu) in addition to deglycosilation in case of glycated FA as in peak 91 detected in blue curcuma (Table 2). In phosphethanolamine derivatives, diagnostic ion at m/z 196 was detected as in peaks 92 and 93 (Table 2). Hydroxylated-oxo-penta/hexa-decanoic acids were detected in both species in peaks 88–90, in which hydroxy-oxohexadecanoic acid, peak 90, was previously reported in yellow curcuma50. Primary metabolites profiling of C. longa and C. caesie via GC-MS (Post-Silylation) To assess variation of primary metabolites among C. longa and C. caesie to account for nutritive value and not readily ionized using ESI in UPLC-MS, GC–MS analysis was employed post-silylation (Suppl. Fig. S12). A total of 38 peaks were identified (Table 3) categorized into alcohols (2), fatty acids/esters (8), organic acids (5), sugars (7), alongside cinnamates (3), steroids/terpenes (13) as secondary metabolites. C. longa exhibited higher sugar level (40.5%), followed by fatty acids/esters (27.4%). In contrast, C. caesia was dominated by steroids/terpenes (42.5%), with sugars (29.9%) as the second major component (Fig. 4), and to rationalize why C. longa is more suited for culinary uses than C. caesia. Differences among each class is explained below in more details in the next subsections. Table 3Relative percentiles of silylated compounds detected in C. caesia and C. longa via GC/MS analysis.PeakRt (min.)KIIdentificationC. longaC. caesia Sugars 114.91814.5D-Fructose, 5TMS *0.02 ± 0.0030.05 ± 0.03215.31854Methyl.alpha.-Arabinofuranoside, 3TMS0.09 ± 0.050.32 ± 0.04315.901904.7Glucose, 5TMS *5.59 ± 0.445.86 ± 1.09416.11915Methyl pentopyranoside2.27 ± 0.231.21 ± 0.33516.71989.8b-Glucopyranose, 5TMS0.53 ± 0.050.35 ± 0.26616.71989Talose, 5TMS11.97 ± 1.1510.98 ± 2.35722.42674Sucrose, 8TMS *20.02 ± 1.1511.16 ± 1.56 Total sugars 40.5 29.93 Organic acids 813.616902-Methyl-2-butenedioic acid dimethyl ester2.63 ± 0.21.78 ± 1.2295.151063.1Lactic Acid, 2TMS*1.72 ± 0.710.84 ± 0.27107.871237.4Benzoic Acid, TMS0.14 ± 0.010.57 ± 0.541111.21487.5Malic acid, 3TMS1.37 ± 0.040.86 ± 0.211213.091641Glutaric acid, di(isobutyl) ester2.35 ± 0.131.56 ± 1.06 Total organic acids 8.2 5.61 Alcohols 138.31272.1Glycerol, 3TMS *1.79 ± 0.122.36 ± 0.671412.015531-Dodecanol, TMS0.45 ± 0.040.28 ± 0.24 Total Alcohols 2.24 2.64 Cinnamates 1514.31751Hydrocinnamic acid 2TMS0.09 ± 0.0050.11 ± 0.031616.119294-Coumaric acid, 2TMS0.16 ± 0.030.15 ± 0.091717.62082(Iso)ferulic acid, 2TMS4.22 ± 0.420.06 ± 0.03 Total cinnamates 4.47 0.32 Steroid/terpene 1811.31493.7Curzerene0.037 ± 0.011.55 ± 0.91912.61599.3Epicurzerenone0.09 ± 0.0222.29 ± 4.442013.21657aR-Turmerone7.13 ± 1.90.14 ± 0.052113.61695Curlone3.59 ± 0.840.21 ± 0.042213.71698.1Germacrone3.59 ± 0.840.21 ± 0.042313.91718Neocurdione00.01 ± 0.012414.11739b-Eudesmol, TMS0.17 ± 0.051.02 ± 0.32514.31718.4Curdione0.48 ± 0.040.68 ± 0.152615.181832.8Curcumenone1.25 ± 0.150.19 ± 0.032715.318454,5-Epoxygermacrone0.16 ± 0.0112.14 ± 2.492816.81994.3Zederone0.11 ± 0.071.89 ± 0.332917.72101.8Valerenic acid, TMS0.43 ± 0.142.16 ± 0.743028.23467b-Sitosterol, TMS0.06 ± 0.020.04 ± 0.02 Total steroid/terpene 17.1 42.54 Fatty acids/esters 3117.082025Palmitic Acid, TMS *14.21 ± 1.1610.24 ± 6.13218.62190Linoleic acid, TMS1.11 ± 0.040.82 ± 0.073318.62201.8Oleic Acid, (Z)-, TMS0.15 ± 0.010.1 ± 0.043418.712201.911-Octadecenoic acid, (Z)-, TMS1.11 ± 0.110.77 ± 0.53518.82219Stearic acid, TMS6.4 ± 0.594.68 ± 3.053621.725711-Monopalmitin, 2TMS2.42 ± 0.871.29 ± 0.533722.925711-Monolinolein, 2TMS0.18 ± 0.010.12 ± 0.043823.132763Glycerol monostearate, 2TMS1.92 ± 1.270.93 ± 0.64 Total fatty acid/ester 27.49 18.95 *denotes compounds confirmed using standards. Relative percentiles of silylated compounds detected in C. caesia and C. longa via GC/MS analysis. *denotes compounds confirmed using standards. Relative percentage of silylated metabolites detected in C. caesia (blue) and C. longa (yellow) rhizomes via GC/MS analysis. Sugars Sugars amounted as the main primary metabolite class in C. longa and C. caesie represented by a total of 7 peaks belonging to mono- and di-sugars. The relative higher sugar level in C. longa at 40.5% versus 29.93% in C. caesia infers for the improved taste and nutritive value in C. longa. Among sugars, sucrose was the most abundant sugar detected at 20.02% and 11.16% in C. longa and C. caesia, respectively. The presence of sucrose at substantial levels suggest for its role as sweat attribute in both species29,57. Following sucrose, talose was another major sugar detected at 10–12%, less common in dietary sources, though of potential health benefits. Compared with di-sugars that showed differences among species, mono-sugar represented by glucose was present at comparable levels ca. 5–6%. These results fall in accordance with previous proximate analysis of C. caesia and C. longa revealing substantial sugar content (19–21%) of dry weight58, comparable to levels detected using GC/MS and extend here in to reveal for sucrose as the major form. Organic acids and alcohols Organic acids were detected in both species at 8.2% and 5.6% in in C. longa and C. caesia, respectively (Table 3). Organic acids contribute to food attributes as natural preservative, food digestion, and in protein utilization59, in addition to taste perception as a flavor enhancer. Major forms included 2-methyl-2-butenedioic acid dimethyl ester and glutaric acid, suggesting their potential use as food additive, preservation, and health-related effects. Compared to acids, alcohols, represented by glycerol and dodecanol, were detected at much lower levels of 2.2–2.6% (Table 3; Fig. 4). Previous studies reported the presence of various alcohols including glycerol and monoterpene alcohols, in Curcuma essential oils60. Cinnamates GC–MS analysis of C. longa and C. caesia extracts revealed much higher levels of cinnamates detected in C. longa than C. caesia at 4.5% and 0.3%, respectively. Examples of cinnamates included hydrocinnamic acid, 4-coumaric acid, and isoferulic/ferulic acid, with isoferulic/ferulic acid as major form in C. longa at 4.2%. The higher cinnamate level in C. longa suggests for stronger antioxidant and anti-inflammatory properties, as cinnamates are known for their role in human health61,62. Steroids/terpenes Steroids/terpenes were represented in C. long and C. caesia by 13 peaks, detected at much higher levels in C. caesia at 42.5% versus 17.1% in C. longa. Most notably, epicurzerenone was detected in C. caesia at 22.39% relative to C. longa at trace levels 0.09%. Epicurzerenone, a sesquiterpene, has been reported to exert potential antitumor and anti-inflammatory effects63. A study on essential oil analysis of C. caesia rhizomes reported less epicurzerenone (12.47%) comparable to levels detected here in this study64. Following epicurzerenone, 4,5-epoxygermacrone (12.14%) was the predominant sesquiterpene in C. caesia, suggesting a rich terpenoid profile associated with strong anti-inflammatory and antimicrobial activities65. Compared to C. caesia, C. longa is characterized by higher levels of αR-turmerone (7.13%) and curlone (3.59%), suggestive of differential terpenoids biosynthesis. Other sesquiterpenes detected at much lower levels included 4,5-epoxygermacrone (12.1%), valerenic acid (2%), zederone (1.9%), and curzerene (1.6%) in C. caesia, as well as curlone (3.6%), germacrone (3.6%), and curcumenone (1.3%) in C. longa. Fatty acids/esters Fatty acids and esters represented the second and third most abundant metabolite class in C. longa and in C. caesia, respectively (Table 3). The higher fatty acid level in C. longa at 27.5% versus 19% in C. caesia highlights for improved nutritive value in C. longa, and is concurrent with the higher sugar profile. Among fatty acids/esters, palmitic acid was the major form detected at 14.2% and 10.2% in C. longa and C. caesia, respectively. The presence of palmitic acid at considerable levels in both C. longa and C. caesia suggests its role in energy metabolism66. A study on fatty acids composition of C. longa oil reported oleic acid likewise as the major component (56–58%)67, slightly higher than that detected herein in alcohol extract using GC/MS.Following palmitic acid, stearic acid was the other major fatty acid at 4–6%. Linoleic and oleic unsaturated fatty acids were detected at much lower levels in both species at 0.8–1.1%, and suggestive that saturated fatty acids predominate in Curcuma species. Asides from free fatty acids, acyl esters were likewise detected such as 1-monopalmitin, 1-monolinolein, and glycerol monostearate in both species at 0.1–2.1%. Aroma profiling of C. longa and C. caesia rhizomes via SPME coupled to GC-MS SPME is well suited for the profiling of aroma and was employed for C. longa and C. caesia rhizomes, providing the real composition of their volatile blends, and is first time to be reported in both species via SPME-GC/MS (Suppl. Fig. S13). A total of 26 volatile constituents were detected in both C. longa and C. caesia rhizomes belonging to monoterpene hydrocarbon (6), oxide/ether (1), alcohol/aldehyde (2), sesquiterpene hydrocarbon (12), ketone (5), S-containing compounds (Suppl.Table 2). Both species showed comparable aroma composition, as seen in the relative percentile of each class in Fig. 5. In C. longa, ketones, and sesquiterpenes were the major volatile classes, detected at 33.8% and 29.17%, respectively, whereas sesquiterpenes were the most abundant in C. caesia (62.79%). Alcohol/aldehyde was detected in C. longa rhizomes at low levels (0.05%) versus monoterpenes detection at trace levels in C. caesia (1.18%). Tumerone was the major volatile in C. longa at 33.47% followed by β-sesquiphellandrene at 20.82%, while β-elemene was the major form in C. caesia at 18.64% followed by heptanol and zingiberene at %15.1 and 12.8%, respectively. Interestingly, turmerone ketones were reported to be responsible for C. longa aroma68, was detected at trace levels in C. caesia (0.01%). Turmerone was reported as a major constituent in the essential oil of C. longa at range 25.3–40.4%69, in accordance with our results using SPME. Likewise. Relative percentage of aroma compounds detected in C. caesia (blue) and C. longa (yellow) via SPME-GC/MS analysis. Multivariate data analyses of blue curcuma C. caesia and yellow curcuma C. longa derived metabolome profiles Classification models were attempted from the different analytical platforms i.e., UPLC/HR-MS/MS and GC-MS to identify discriminative metabolites for each species. Loading plots and S-plots identified biomarkers mediating for samples’ classification as denoted by their Mol.wt./Rt derived from respective PCA and OPLS-DA plots. PCA and OPLS-DA analyses of UPLC/HR-MS/MS dataset The unsupervised PCA model of UPLC/HR-MS/MS analyzed in negative ionization mode showed moderate prediction power (R2 = 0.7, Q2 = 0.55). Samples segregation was observed mostly along PC1 accounting for the major variance (Fig. 6A). PCA loading plot (Fig. 6B) revealed guainane-sesquiterpene “aerugidiol” as biomarker for C. caesia alongside other alkylbenzene sulfonate compounds at m/z 311 and m/z 325 (base peak m/z 183, C8H7O3S−). However, these are reported contaminants in irrigating waste water and pesticides used in agricultural field practice in China70. In contrast, C. longa was discriminated by exclusive presence of diarylheptanoids curcumin III and dihydrobisdemethoxycurcumin as well as abundance of curcumin I and curcumin II which was further confirmed using supervised OPLS-DA score plot (R2 = 0.9, Q2 = 0.9) and respective S-loading plot (Fig. 6C and D). Multivariate data analyses of UPLC/HR-MS/MS dataset of blue curcuma (C. caesia) and yellow curcuma (C. longa) in both negative (A-D) and positive (E-H) ionization modes (n = 3). Principle component analysis (PCA) score plots (A, E) and loading plots (B, F); orthogonal partial least squares discriminant analysis (OPLS-DA) score plots (C, G) and S-loading plots (D, H) with PC1 and PC2 represent > 60% and > 90% of total variation in data derived from negative and positive ionization modes, respectively. Metabolites were assigned based on Mol.wt/Rt as identified in Table 1. On the other side, the PCA model derived from UPLC/HR-MS/MS data analyzed in positive ionization mode showed stronger classification as evidenced by larger variance and prediction power (R2 = 0.9, Q2 = 0.9), where PC1 and PC2 accounted for ca. 90% of the variance (Fig. 6E). C. longa specimen was still discriminated by diaryl heptanoids curcumin I, curcumin II, and curcumin III (Fig. 6F) and in agreement with negative mode result (Fig. 6B), in addition to monoarylheptanoids ar-turmerone and dihydro-ar-turmerone revealed in positive mode, likely due to their improved ionization in that mode and highlighting the benefit of acquisition in both modes. In contrast, C. caesia was distinguished by guainane, germacrane, and elemane types of sesquiterpenes, namely curcumenol, germacrone epoxide, zederone, curdione, and epi/curzerenone. Supervised OPLS-DA was further employed for its superiority in biomarkers assignment (R2 = 0.9, Q2 = 0.9) (Fig. 6G). The S-loading plot of metabolites verified curcumin I, curcumin III, and dihydro-ar-turmerone as biomarkers of C. longa, while curdione and epi/curzerenone were the discriminatory metabolites of C. caesia (Fig. 6H). The results are consistent with previous studies reporting that C. caesia is among 4 Curcuma species only that biosynthesize mixture of 4 different types of sesquiterpenes (guainane, germacrene, elemane and carabrane), whereas diaryl heptanoids are detected in 10 Curcuma species including C. longa9. These results confirm that chemotaxonomic classification in Curcuma taxa is based on diarylheptanoids versus sesquiterpenes biosynthesis. PCA and OPLS-DA analyses of GC/MS silylated metabolite dataset PCA model of silylated compounds to encompass a total of 39 peaks analyzed via GC/MS showed good variance and prediction power (R2 = 0.9, Q2 = 0.8) where PC1 and PC2 accounted for > 70% of metabolites (Fig. 7A). Loading plot of metabolites assigned epicurzerenone sesquiterpene as biomarker for blue curcuma C. caesia, while yellow curcuma C. longa was discriminated by abundance of palmitic acid, iso/ferulic acid, germacrone and arylheptanoid turmerone (Fig. 7B). These results are in agreement with UPLC/HR-MS/MS results shown in Table 1and Fig. 6E and H. To verify classification, OPLS-DA was conducted, where S-plot assigned epicurzerenone as C. caesia discriminatory metabolite alongside epoxygermacrone. In contrast, sucrose was abundant in C. longa (Fig. 7C and D), and to account for the improved taste of C. longa and culinary use compared to C. caesia. Multivariate data analyses of GC/MS dataset of blue curcuma (C. caesia) and yellow curcuma (C. longa) of silylated (A-D) and volatile (E-H) components (n = 3). Principle component analysis (PCA) score plots (A, E) and loading plots (B, F), orthogonal partial least squares discriminant analysis (OPLS-DA) score plots (C, G) and S-loading plots (D, H) with PC1 and PC2 represent > 70% and > 90% of total variation in data of silylated and volatile components, respectively. Metabolites were assigned based on Mol.wt/Rt as identified in Tables 2 and 3. PCA and OPLS-DA analyses of SPME-GC/MS volatiles dataset With regards to volatile components analyzed via SPME-GC/MS, PCA and OPLS-DA score plots showed likewise good variance and prediction power (R2 = 0.9, Q2 = 0.8–0.9) representing > 90% of the total variance (Fig. 7E). Both loading and S-plots verified that turmerone and sesquiphellandrene are the discriminating volatiles of C. longa (Fig. 7F and H). PCA loading plot detected further enrichment of C. caesia aroma profile with alcohol i.e., heptanol, sesquiterpene hydrocarbons β-elemene, humulene, zingiberene and oxygenated ketone sesquiterpene epi/curzerenone (Fig. 7F). The S-plot verified further enrichment of C. caesia in heptanol and epi/curzerenone and considered its volatile biomarkers (Fig. 7H). It’s noteworthy that epi/curzerenone imparts a mushroom-like rooty odor of blue curcuma that is balanced with sweet and fresh herbal aroma attributed to sesquiterpenes hydrocarbons71. In contrast, turmerone essentially imparts the distinctive turmeric spicy odor in C. longa72. In conclusion, epi/curzerenone oxygenated sesquiterpene of elemane type were recognized as a discriminating metabolite for blue curcuma C. caesia based on UPLC/HR-MS/MS, post- silylated GC/MS, and SPME-GC/MS analyses. In contrast, C. longa is recognized with its mono-aryl and diaryl-heptanoids, in addition to its further enrichment in fatty acids, sugars, and cinnamates (Figs. 6 and 7). Quantification of total phenolic (TPC) and flavonoid contents (TFC) in C. caesia and C. longa rhizomes The total phenolic and flavonoid contents were quantified in both turmeric samples for standardization purposes. The results revealed significant differences in both contents, where TPC in yellow curcuma (106.26 ± 3.53 mg GAE/g w) was ~ 7 times higher than that in blue curcuma (16.64 ± 0.14 mg GAE/g w) quantified as gallic acid equivalent. Similarly, TFC in yellow curcuma (18.45 ± 0.3 mg RE/g w) exhibited triple the flavonoid content of blue curcuma (6.43 ± 0.27 mg RE/g w) quantified as rutin equivalent. The high TPC in yellow curcuma is compatible with its further enrichment with curcuminoids (Di/phenyl heptanoids) as revealed from UPLC/HR-MS/MS profiling and subsequent multivariate data analyses in contrast to blue curcuma (Table 2; Fig. 6). Bioactivity assays Antioxidant activity determination Scavenging of free radicals is known to mediate for several bioactivities and serve as a good prediction for exploring the potentials of natural resources, thus both species were comparatively assayed for their free radical scavenging potential using 3 different in vitro assays viz. DPPH, ABTS and FRAP, measured as trolox equivalent. This ensured to have better validated evaluation of both turmeric extracts according to their lipophilic and hydrophilic constituents using different action mechanisms73. Yellow curcuma exhibited higher scavenging activity by DPPH (37.36 ± 0.99 mg TE/g w) and ABTS (34.37 ± 0.74 mg TE/g w) assays than the blue curcuma (20.78 ± 0.74 mg TE/g w and 18.31 ± 0.18 mg TE/g w, respectively) which still showed moderate antioxidant activity. In contrast, FRAP assay which measures the reducing power or electron-donating capacity of compounds, revealed comparable antioxidant potential of blue (34.12 ± 0.33 mg TE/g w) and yellow curcuma (31.44 ± 0.93 mg TE/g w). The obtained results are compatible with the enriched phenolic content in yellow curcuma that are potent scavengers of free radicals, as well as, the presence of hydroquinone derivatives detected via UPLC/HR-MS/MS in blue curcuma that are potent ferric reductants, thus reflected in the FRAP assay45 (Table 2). Antidiabetic activity determination via α-glucosidase Inhibition Both blue and yellow curcuma are reported to exhibit in vivo antidiabetic activity74,75. In this study their α-glucosidase inhibition activity was determined comparably against acarbose reference drug. C. longa showed more potent activity evidenced by lower IC50 (0.325 ± 0.025 mg/ml) than that observed for C. caesia rhizome (1.04 ± 0.064 mg/mL) and acarbose (0.4891 ± 0.1 mg/mL). This is justified by the further abundance of phenolics in C. longa which interfere with the active sites of α-glucosidase, thus, preventing it from binding with carbohydrates76. Moreover, curcuminoids are proven by clinical and preclinical trials to act synergistically against diabetes via different mechanism of action6,77. On the other side, terpenes of C. caesia showed high in silico binding affinity to α-glucosidase enzyme that contributes in the herein observed antidiabetic activity78. Anti-obesity activity determination via lipase Inhibition Rhizome extracts of C. longa and C. caesia were investigated for their anti-obesity activity using orlistat reference drug via lipase inhibition. Both samples showed no significant inhibitory activity against lipase at the maximal test concentration (10 mg/mL) and its dilutions (orlistat IC50 = 0.16 ± 0.08 mg/mL). Conclusion The study presented the first integrated comparative metabolome profiling of C. caesia and C. longa using multiplex approach aided by multivariate data analyses, uncovering distinct metabolic divergence among both species and highlighting several first-time reported metabolites in the less explored turmeric blue curcuma. The analyses revealed for 11 metabolites annotated via NMR, 110 tentatively identified metabolites by UPLC/HR-MS/MS, 38 silylated compounds and 26 volatiles detected by GC/MS analysis. Samples’ classification and discriminatory chemotaxonomic biomarkers were assigned for each species by PCA and OPLS data analyses emphasizing diverse sesquiterpenes in C. caesia vs. curcuminoids and sugar enrichment in C. longa aligning with its established culinary and therapeutic uses. Antioxidant and antidiabetic bioassays showed superior activity of C. longa compatible with its higher quantified contents of total phenolics and flavonoids, whereas both species were inactive as anti-obesity agents compared to orlistat reference drug. Targeted isolation of novel sesquiterpenes annotated in C. caesia is recommended for therapeutic evaluation to elucidate its potential in pharmaceutical and food industries. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements Authors would like to thank Dr. Kai Wang at the Chinese Academy of Sciences, Beijing for sample collection and authentication. Author contributions R.H.: performed chemical profiling via UPLC/MS, wrote and revised the main manuscript, and prepared figuresT.A.M.: performed NMR profiling and quantification, prepared figures, wrote the main manuscriptM.A.F.: conceptualization, team management, writing, and reviewing the main manuscriptS.R.: GC/MS data interpretation and prepared its figures, wrote main manuscriptT.D.B.: contributed to manuscript writingA.N. and A.M.: biological activity evaluation. Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB). Data availability The raw files dataset used and/or analyzed during the current study available from the corresponding author on reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The raw files dataset used and/or analyzed during the current study available from the corresponding author on reasonable request.