Sequential solvent fractionation strategies modulate polyphenol distribution and antioxidant activity of Caryota mitis seed extracts

Abstract

Seeds of Caryota mitis represent an underexplored source of bioactive polyphenols. This study investigated the influence of four extraction workflows on phenolic/flavonoid enrichment, antioxidant activity, nitric oxide (NO) inhibitory potential, and LC-MS/MS chemical profiles of C. mitis seed extracts. Total phenolic content (TPC) was expressed using both gallic acid equivalents (GAE) and ferulic acid equivalents (FAE), while total flavonoid content (TFC) was expressed as catechin equivalents (CE). Among the tested ethanol concentrations, 96% ethanol (CME96) showed the highest mean TPC values under both GAE and FAE expressions and the highest mean TFC value, and was therefore selected for ethanol-based partitioning workflows. A2-EaM showed the highest TPC and TFC among the fractions, whereas B1-M displayed the most biologically promising profile, combining strong preliminary antioxidant responses with the lowest NO IC₅₀ among the tested fractions. LC-MS/MS profiling was re-curated to retain 21 tentatively annotated compounds/features supported by accurate precursor ions and diagnostic MS/MS fragments compared with MassBank and/or MoNA records. The revised results indicate that extraction workflow affected both bulk phenolic/flavonoid enrichment and qualitative metabolite composition. Biological activity was therefore interpreted as a workflow-dependent response that cannot be explained solely by total phenolic or flavonoid content.

---

1. Introduction

Caryota mitis Lour. (Arecaceae), commonly known as the clustering fishtail palm, is widely distributed in tropical regions of Asia and has been traditionally utilized in folk medicine. Various parts of the plant have been reported to contain bioactive secondary metabolites with potential pharmacological properties. Previous phytochemical studies mainly focused on the leaves and fruit peel, where flavonoids, phenolic compounds, and other metabolites were identified and associated with antioxidant and antimicrobial activities.^1–7 Recent metabolomics-driven investigations have further advanced the understanding of plant-based bioactivities.^8–10

Recent investigations on C. mitis have further demonstrated the biological potential of the species. Metabolomic profiling and compound isolation studies revealed several flavonoids with potential chemopreventive properties, while a newly discovered cerebroside isolated from the fruit peel exhibited chondrocyte proliferation activity.^2,5 Recent ethnopharmacological validation studies also suggested the therapeutic potential of C. mitis fruit peel in the management of osteoarthritis and gout.¹¹ Despite these advances, phytochemical and biological investigations of C. mitis seeds remain limited.

Polyphenols and flavonoids are widely recognized as important contributors to redox-related and inflammation-associated bioactivities in plant extracts, and their distribution is strongly governed by solvent polarity and extraction strategy.^12–16 Flavonoid-bearing plants have also been widely investigated as sources of structurally diverse antioxidant and enzyme-modulating secondary metabolites.¹⁷ Because different solvents and extraction strategies may selectively enrich specific metabolite subsets, comparative workflow-based investigation is useful for clarifying how extraction design shapes chemical and biological profiles.

Previous studies have largely focused on extraction yield or total phenolic content, while the role of extraction workflow in shaping qualitative metabolite composition and biological responses remains poorly understood. In this context, the present study systematically investigates the effect of four extraction workflows on the chemical composition and antioxidant/NO inhibitory activity of C. mitis seed extracts. Two direct sequential extraction workflows (A1 and A2) and two workflows involving ethanol extraction followed by liquid–liquid partitioning (B1 and B2) were compared. Representative fractions were further analyzed by LC-MS/MS to provide tentative chemical support for the observed bioactivity profiles.

2. Materials and methods

2.1 Plant material

Fruits of Caryota mitis Lour. were collected in Tien Giang Province, Vietnam, in August 2022. The plant material was authenticated by Dr Luu Hong Truong, and a voucher specimen (CM2022/D-1) was deposited at the Institute of Advanced Technology. The fruits were cleaned, air-dried, and separated to obtain the seeds. The dried seeds were ground into fine powder and stored in a dry environment at room temperature until extraction and analysis.

2.2 Chemicals and reagents

All analytical-grade reagents were purchased from Sigma-Aldrich and Merck.

2.3 Instruments

UV-vis spectra were recorded using a Shimadzu UV-1800 spectrophotometer (Japan). LC-MS/MS analysis was performed using an ExionLC™ UHPLC system coupled with an X500R QTOF mass spectrometer (AB SCIEX, USA). Chromatographic separation was achieved using a Hypersil GOLD C18 column (150 × 2.1 mm, 3 µm; Thermo Fisher Scientific, USA). Additional equipment included an analytical balance, vortex mixer, centrifuge, ultrasonic bath, and drying oven.

2.4 Extraction workflows

The extraction procedures are summarized in Fig. 1. Four extraction workflows were designed to investigate how extraction strategy influences the distribution of phenolics, flavonoids, and antioxidant/NO inhibitory responses.

Fig. 1 Schematic representation of extraction workflows used for Caryota mitis seed fractionation. Ethanol concentration screening was first performed using 30%, 50%, 70%, and 96% ethanol based on TPC and TFC values. Direct sequential extraction was conducted through workflows A1 and A2, whereas workflows B1 and B2 involved ethanol extraction followed by liquid–liquid partitioning. The resulting fractions were evaluated by TPC/TFC assays, antioxidant and NO inhibitory assays, and LC-MS/MS profiling.

2.4.1 Direct sequential extraction. Two sequential extraction approaches were applied directly to the powdered seed material. Workflow A1 used hexane, ethyl acetate (EtOAc), and methanol (MeOH). Workflow A2 used a refined polarity gradient: hexane : EtOAc (1 : 1), EtOAc, EtOAc : MeOH (1 : 1), and MeOH. Each extract was filtered and the solvent was removed under reduced pressure to obtain the corresponding fraction.

2.4.2 Ethanol extraction followed by solvent partitioning. To determine the ethanol concentration used for ethanol-based workflows, 30%, 50%, 70%, and 96% ethanol were initially screened based on TPC and TFC. CME96 was selected for preparing the crude ethanol extract because it consistently belonged to the highest-performing group across TPC and TFC assays. The crude ethanol extract was then partitioned by two workflows: B1 (hexane → EtOAc → MeOH) and B2 (hexane : EtOAc, 1 : 1 → EtOAc → EtOAc : MeOH, 1 : 1 → MeOH). All fractions were concentrated under reduced pressure prior to analysis.

Extraction and fractionation yields were calculated and summarized in SI Table S1 to provide a quantitative comparison of process efficiency among workflows.

2.5 UV-vis spectral scanning

UV-vis spectral scanning was performed to describe the spectral behavior of the colorimetric reaction products used in TPC and TFC assays. The spectra were used only as SI to verify that extract responses were within the expected absorbance regions of the assays, and not as the sole criterion for selecting reference standards.

2.6 Determination of total phenolic content (TPC)

Total phenolic content was determined using the Folin–Ciocalteu method with slight modifications.¹⁸ The extract solution was mixed with Folin–Ciocalteu reagent followed by sodium carbonate solution. After incubation, absorbance was measured using a UV-vis spectrophotometer. TPC values were calculated on a dry extract weight basis and expressed as both milligrams of gallic acid equivalents and ferulic acid equivalents per gram of dry extract (mg GAE g⁻¹ extract and mg FAE g⁻¹ extract). Reporting both equivalents improved comparability and allowed assessment of whether the phenolic distribution pattern depended on the reference standard.

2.7 Determination of total flavonoid content (TFC)

Total flavonoid content was determined using the aluminum chloride colorimetric method as described previously.¹⁹ The extract solution was mixed sequentially with NaNO₂, AlCl₃, and NaOH, and the absorbance was measured using a UV-vis spectrophotometer. TFC values were calculated on a dry extract weight basis and expressed as milligrams of catechin equivalents per gram of dry extract (mg CE g⁻¹ extract).

2.8 Antioxidant and NO inhibitory assays

Antioxidant activities of crude extracts and fractions were evaluated using DPPH radical scavenging,²⁰ ABTS radical cation decolorization,²¹ reducing power (RP),²² ferric reducing antioxidant power (FRAP),²³ hydrogen peroxide scavenging,²⁴ and nitric oxide (NO) inhibitory assays.²⁵ Preliminary antioxidant screening was performed to select candidate fractions for IC₅₀ determination. DPPH and H₂O₂ screening were performed at fixed concentrations of 5 and 100 µg mL⁻¹, respectively, whereas RP and FRAP were used as screening indicators with different concentrations for weak nonpolar fractions and the remaining fractions. Selected fractions were then evaluated by IC₅₀ determination. Gallic acid was used as a reference antioxidant compound.

2.9 LC-MS/MS analysis

2.9.1 Sample preparation. Representative samples (CME96, A2-EA, A2-EaM, A2-M, B1-M, B2-EaM, and A1-M) were dissolved in methanol at 1 mg mL⁻¹. The solutions were sonicated for 10 min and filtered through a 0.22 µm PTFE membrane filter prior to analysis.

2.9.2 LC-MS/MS conditions and annotation strategy. Chromatographic separation was performed on an ExionLC™ UHPLC system coupled to an X500R QTOF mass spectrometer (AB SCIEX, USA). The Hypersil GOLD C18 column (150 × 2.1 mm, 3 µm) was used. Mobile phase A was water containing 0.1% formic acid, and mobile phase B was acetonitrile containing 0.1% formic acid. The gradient was 0–1 min, 2% B; 1–30 min, 2–98% B; and 30–36 min, 98% B. The flow rate was 0.4 mL min⁻¹ and the injection volume was 2 µL. Mass spectrometry was performed in negative electrospray ionization mode (ESI⁻) with a source temperature of 500 °C, ion spray voltage of −4.5 kV, and curtain gas/GS1/GS2 settings of 30/45/45 psi. TOF-MS and MS/MS spectra were acquired over m/z 70–2000 and m/z 50–1500, respectively.

Raw data were processed using SCIEX OS v1.2.0.4122 and converted to mzML using MSConvert. Compound annotation was performed by comparing accurate precursor ions and diagnostic MS/MS fragmentation patterns with MassBank and/or MoNA spectral records. Only features with clear extracted ion chromatogram (EIC) peaks at the corresponding retention time and precursor m/z were retained in the final annotation table. Because authentic standards were not used, all compounds/features were reported as tentatively annotated rather than definitively identified.^26–28

A limitation of the LC-MS/MS profiling is that online UV/DAD spectra were not acquired for individual chromatographic peaks. Therefore, LC peak annotations were based on accurate precursor ions, EIC behavior, and diagnostic MS/MS fragmentation patterns compared with spectral database records.

2.10 Statistical analysis

All experiments were performed in triplicate (n = 3), and results are expressed as mean ± standard deviation (SD). Statistical differences were analyzed using one-way ANOVA followed by Tukey's HSD post hoc test. For Table 2, statistical comparisons of TPC and TFC were performed within each extraction workflow. For Table 4, statistical comparisons were performed within each assay among the selected fractions and gallic acid. Significance was considered at p < 0.05.

3. Results and discussion

3.1 Spectral behavior and expression of TPC/TFC values

UV-vis spectral scanning was used to observe the spectral behavior of the TPC and TFC colorimetric reaction products (Fig. 2). This approach is consistent with recent studies that combine chemical profiling with biological activity assessments of plant extracts.²⁹ The spectra of the TFC reaction mixtures showed comparable absorption profiles among the CME extracts, while catechin and quercetin differed in signal intensity and profile. Considering the LC-MS/MS evidence for catechin/procyanidin-related flavonoid features and the comparable distribution trend obtained in preliminary equivalent calculations, TFC was finally expressed as catechin equivalents.

Fig. 2 UV-vis absorption spectra of Caryota mitis seed extracts and reference standards for the colorimetric assay products: (A) TFC determination and (B) TPC determination. The spectra were used to describe the spectral behavior of the assay reaction mixtures and were not used as the sole criterion for reference-standard selection.

For TPC determination, values were reported using both GAE and FAE. GAE was included because it is a widely used expression for the Folin–Ciocalteu assay and facilitates comparison with previous studies, whereas FAE was additionally retained to evaluate whether the phenolic distribution pattern was affected by the reference standard. The GAE- and FAE-based calculations showed consistent trends, supporting the robustness of the TPC comparison.

3.2 Effect of ethanol concentration on TPC and TFC extraction

To determine the ethanol concentration used for the ethanol-based workflows, 30%, 50%, 70%, and 96% ethanol were evaluated based on TPC and TFC (Table 1). CME96 showed the highest mean TPC values under both GAE and FAE expressions and also exhibited the highest mean TFC value expressed as CE. Although CME70 showed statistically comparable TPC values and CME50 showed comparable TFC, CME96 consistently belonged to the highest-performing group across TPC and TFC assays.

Table 1 Total phenolic content (TPC) and total flavonoid content (TFC) of Caryota mitis seed extracts obtained using different ethanol concentrations^a

Sample	TPC		TFC
	mg GAE g^−1 extract	mg FAE g^−1 extract	mg CE g^−1 extract
a Values are expressed as mean ± standard deviation (n = 3). Different letters (a–c) within the same column indicate statistically significant differences (one-way ANOVA followed by Tukey's HSD test, p < 0.05).
CME30	270.49 ± 21.26^c	371.88 ± 27.24^c	404.75 ± 4.30^b
CME50	368.86 ± 32.48^b	497.92 ± 41.61^b	455.11 ± 3.69^a
CME70	444.47 ± 6.14^a	594.79 ± 7.86^a	426.38 ± 11.51^b
CME96	490.81 ± 18.31^a	654.17 ± 23.45^a	469.65 ± 13.30^a

---

Accordingly, 96% ethanol was selected for preparing the crude extract used in the B1 and B2 liquid–liquid partitioning workflows. This selection was based on its high phenolic/flavonoid recovery and suitability for preparing a less aqueous crude extract for downstream partitioning.

3.3 Distribution of phenolics and flavonoids among extraction fractions

TPC and TFC values varied markedly among fractions within each extraction workflow (Table 2 and Fig. 3). Nonpolar fractions consistently showed the lowest TPC and TFC values, whereas medium-to-high polarity fractions were significantly enriched in phenolic and flavonoid constituents.

Table 2 Total phenolic content and total flavonoid content of fractions obtained from four extraction workflows of Caryota mitis seeds^a

Workflow	Fraction	TPC		TFC
		mg GAE g^−1 extract	mg FAE g^−1 extract	mg CE g^−1 extract
a Values are expressed as mean ± SD (n = 3). Different letters within the same workflow and the same column indicate statistically significant differences according to one-way ANOVA followed by Tukey's HSD test (p < 0.05). GAE: gallic acid equivalents; FAE: ferulic acid equivalents; CE: catechin equivalents.
A1	A1-H	2.07 ± 0.75^c	13.15 ± 0.78^c	13.57 ± 5.12^c
	A1-EA	519.27 ± 8.45^a	600.77 ± 8.88^a	226.01 ± 6.40^b
	A1-M	452.60 ± 12.28^b	530.68 ± 12.91^b	291.51 ± 5.07^a
A2	A2-Hea	68.24 ± 2.44^d	82.72 ± 2.56^d	35.90 ± 4.40^c
	A2-EA	407.07 ± 14.71^c	482.82 ± 15.60^c	152.75 ± 17.76^b
	A2-EaM	606.26 ± 12.42^a	692.22 ± 13.16^a	343.84 ± 11.45^a
	A2-M	573.74 ± 11.45^b	658.03 ± 12.12^b	325.62 ± 0.67^a
B1	B1-H	5.75 ± 0.49^c	11.53 ± 0.53^c	27.84 ± 2.62^c
	B1-EA	150.98 ± 4.22^b	213.59 ± 4.44^b	116.32 ± 19.33^b
	B1-M	511.14 ± 3.66^a	592.22 ± 3.92^a	305.47 ± 12.95^a
B2	B2-Hea	39.89 ± 0.50^d	47.43 ± 0.53^d	29.50 ± 1.78^c
	B2-EA	159.11 ± 8.57^c	222.14 ± 9.00^c	116.71 ± 6.40^b
	B2-EaM	535.53 ± 2.82^a	617.86 ± 2.96^a	285.31 ± 21.83^a
	B2-M	361.54 ± 3.66^b	434.96 ± 3.92^b	260.12 ± 20.17^a

---

Fig. 3 Heatmap illustrating the distribution of total phenolic content (TPC) and total flavonoid content (TFC) across 14 fractions obtained from four extraction workflows (A1, A2, B1, and B2). Colors represent normalized values scaled from 0 (minimum, dark blue) to 1 (maximum, dark red), as shown by the color scale on the right. Fractions are arranged according to increasing solvent polarity: H (hexane), Hea (hexane : ethyl acetate, 1 : 1), EA (ethyl acetate), EaM (ethyl acetate : methanol, 1 : 1), and M (methanol). A2-EaM showed the highest absolute TPC and TFC values, whereas B1-M represented the most biologically promising fraction based on antioxidant screening and NO inhibitory activity.

A2-EaM showed the highest TPC under both GAE and FAE expressions and also had the highest absolute TFC value. Within A2, A2-EaM and A2-M formed the highest TFC statistical group. In B1, B1-M showed the highest TPC and TFC values, while in B2, B2-EaM showed the highest TPC and formed the highest TFC group together with B2-M. These results indicate that phenolic/flavonoid enrichment was strongly dependent on both solvent polarity and extraction workflow.

3.4 Preliminary antioxidant screening and IC₅₀-based evaluation

Table 3 presents the preliminary antioxidant screening used to select candidate fractions for IC₅₀ determination. DPPH and H₂O₂ scavenging activities were evaluated at fixed concentrations of 5 and 100 µg mL⁻¹, respectively, for all fractions. For RP and FRAP assays, nonpolar H/Hea fractions were screened at 500 µg mL⁻¹ because of their weak preliminary responses, whereas the remaining fractions were screened at 10 µg mL⁻¹. Therefore, RP and FRAP data were interpreted as screening indicators rather than direct concentration-normalized potency comparisons.

Table 3 Preliminary antioxidant screening of 14 fractions for selecting candidate fractions for IC₅₀ determination^a

Fr	RP/FRAP conc. (µg mL^−1)	RP, mg GAE g^−1 extract	FRAP, mg BHAE g^−1 extract	DPPH inhibition at 5 µg mL^−1 (%)	H2O2 inhibition at 100 µg mL^−1 (%)
a Values are expressed as mean ± SD (n = 3). DPPH radical scavenging activity and H2O2 scavenging activity were evaluated at fixed concentrations of 5 µg mL^−1 and 100 µg mL^−1, respectively, for all fractions. For RP and FRAP assays, nonpolar H/Hea fractions were screened at 500 µg mL^−1 because of their weak preliminary responses, whereas the remaining fractions were screened at 10 µg mL^−1. Therefore, RP and FRAP values were interpreted as preliminary screening indicators rather than direct concentration-normalized potency comparisons across all fractions. Gallic acid was used as a positive control for DPPH and H2O2 assays. GAE: gallic acid equivalents; BHAE: butylated hydroxyanisole equivalents.
A1-H	500	4.18 ± 1.42	2.69 ± 0.02	0.94 ± 0.33	25.69 ± 1.57
A1-EA	10	170.14 ± 1.88	268.99 ± 1.34	61.15 ± 3.27	31.81 ± 1.70
A1-M	10	189.53 ± 3.91	259.74 ± 0.00	70.06 ± 0.99	55.01 ± 4.19
A2-Hea	500	36.57 ± 9.04	31.76 ± 1.27	12.53 ± 6.69	35.14 ± 8.39
A2-EA	10	147.92 ± 12.89	245.46 ± 4.97	50.61 ± 4.13	36.35 ± 14.10
A2-EaM	10	223.81 ± 9.78	253.75 ± 0.73	84.06 ± 1.98	56.50 ± 3.69
A2-M	10	174.40 ± 9.27	273.51 ± 0.79	72.88 ± 0.30	52.64 ± 5.40
B1-H	500	25.22 ± 5.95	5.46 ± 0.30	4.02 ± 4.10	25.69 ± 8.11
B1-EA	10	92.60 ± 3.91	138.67 ± 6.71	46.27 ± 0.66	27.69 ± 0.35
B1-M	10	242.25 ± 12.39	254.03 ± 14.47	85.69 ± 0.87	62.15 ± 0.98
B2-Hea	500	30.66 ± 6.81	22.29 ± 0.23	14.56 ± 2.48	38.58 ± 1.64
B2-EA	10	112.93 ± 4.98	166.17 ± 1.73	39.12 ± 4.75	29.80 ± 1.19
B2-EaM	10	222.62 ± 7.67	206.10 ± 6.79	79.34 ± 2.53	56.54 ± 1.47
B2-M	10	203.24 ± 10.29	188.31 ± 3.01	67.23 ± 0.66	51.32 ± 0.34
Gallic acid	—	—	—	91.15 ± 0.54	91.44 ± 3.10

---

Despite being tested at higher RP/FRAP concentrations, nonpolar H/Hea fractions showed weak responses. In contrast, A1-EA, A1-M, A2-EaM, B1-M, and B2-EaM showed stronger overall screening profiles and were selected for IC₅₀ determination. Among the screening results, B1-M showed high DPPH inhibition, the strongest RP response, and the highest H₂O₂ scavenging activity, whereas A2-M showed the highest FRAP value.

The IC₅₀ values of the five selected fractions are presented in Table 4. In the DPPH assay, B1-M, A2-EaM, B2-EaM, and A1-M showed comparable IC₅₀ values and formed the same statistical group, while gallic acid exhibited significantly stronger activity. In the ABTS assay, gallic acid was also significantly more active than the fractions, whereas the fractions showed relatively close IC₅₀ values.

Table 4 IC₅₀ values of five selected fractions and gallic acid in antioxidant and NO inhibitory assays^a

Sample	DPPH IC50, µg mL^−1	ABTS IC50, µg mL^−1	NO IC50, µg mL^−1	H2O2 IC50, µg mL^−1
a IC50 values are expressed as mean ± SD (n = 3). Lower IC50 values indicate stronger activity. Different letters within the same column indicate significant differences according to one-way ANOVA followed by Tukey's HSD test (p < 0.05). Gallic acid was used as the reference antioxidant compound.
A1-EA	4.67 ± 0.18^a	10.62 ± 0.19^a	137.03 ± 3.94^b	195.10 ± 9.08^a
A1-M	3.59 ± 0.32^b	9.68 ± 0.43^b	119.88 ± 13.56^b	162.27 ± 4.23^b
A2-EaM	3.27 ± 0.33^b	10.80 ± 0.54^a	216.54 ± 17.32^a	181.58 ± 3.09^a
B1-M	3.25 ± 0.38^b	10.25 ± 0.16 ^ab	68.43 ± 2.80^c	164.02 ± 3.97^b
B2-EaM	3.44 ± 0.30^b	10.34 ± 0.20 ^ab	131.64 ± 12.46^b	142.75 ± 9.52^c
Gallic acid	0.59 ± 0.03^c	0.68 ± 0.02^c	135.10 ± 13.30^b	5.92 ± 1.75^d

---

A notable difference was observed in the NO inhibitory assay. B1-M exhibited the lowest NO IC₅₀ value and was significantly more active than all other tested fractions and gallic acid in this in vitro NO inhibitory assay. This result indicates strong in vitro NO inhibitory potential, although it should not be interpreted as direct evidence of anti-inflammatory efficacy. In the H₂O₂ assay, gallic acid was the most active reference compound, while B2-EaM showed the lowest IC₅₀ among the tested fractions.

3.5 Curated LC-MS/MS annotation of representative fractions

To provide stronger chemical evidence for the representative fractions, the LC-MS/MS dataset was curated using EIC inspection, accurate precursor ions, and diagnostic MS/MS fragments. As a result, 21 compounds/features were tentatively annotated and summarized in Table 5. Representative BPCs, EICs, and MS/MS spectra are provided in Fig. S1A, S1B, and S2.

Table 5 Tentatively annotated compounds/features detected in representative Caryota mitis seed extracts by LC-MS/MS in negative ion mode ([M–H]⁻) based on accurate precursor ions and diagnostic MS/MS fragments^a

No.	Figure marker	RT (min)	Putative annotation	Class	Formula	Observed m/z	Library m/z	Error (ppm)	MS/MS fragments (observed/library m/z)	Database/reference	Found
a All compounds/features were tentatively annotated based on accurate precursor ions and diagnostic MS/MS fragmentation patterns compared with MassBank and/or MoNA spectral records. No authentic standards were used; therefore, the annotations should not be considered definitive structural confirmations. Only samples with clear extracted ion chromatogram (EIC) peaks at the corresponding retention time and precursor m/z were listed as detected.
1	P9	5.061	Protocatechuic acid	Phenolic acid	C7H6O4	153.01939	153.01933	0.3800	108.0208/108.0210; 91.0191/91.0190; 80.0276/80.0280; 65.0020/65.0020	MassBank/MoNA	A2-EA
2	P1	5.584	Procyanidin C1	Flavonoids	C45H38O18	865.19501	865.19854	−4.0700	713.1460/713.1456; 289.0689/289.0676; 161.0227/161.0227; 381.0945/381.0938; 449.0835/449.0846; 451.0999/451.0992	MassBank/MoNA	A1-M; A2-EA; A2-EaM; A2-M; B1-M; B2-EaM; CME96
3		5.603	Procyanidin B-type dimer isomer (B1-like)	Flavonoids	C30H26O12	577.13380	577.13515	−2.3400	289.0707/289.0720; 245.0817/245.0820; 179.0337/179.0340; 125.0236/125.0240; 109.0301/109.0300; 150.0299/150.0300	MassBank/MoNA	A1-M; A2-EA; A2-EaM; A2-M; B1-M; B2-EaM; CME96
4	P10	5.714	Procyanidin B-type dimer isomer (B2-like)	Flavonoids	C30H26O12	577.13247	577.13515	−4.6500	289.0705/289.0710; 245.0443/245.0450; 177.0189/177.0190; 137.0237/137.0240; 149.0236/149.0240; 151.0395/151.0400	MassBank/MoNA	A1-M; A2-EA; A2-EaM; A2-M; B1-M; B2-EaM
5	P2	5.965	Catechin	Flavonoids	C15H14O6	289.07056	289.07175	−4.0900	179.0341/179.0342; 123.0440/123.0440; 109.0286/109.0283; 81.0331/81.0333; 83.0125/83.0125; 93.0334/93.0333	MassBank/MoNA	A1-M; A2-EA; A2-EaM; A2-M; B1-M; B2-EaM; CME96
6	P3	6.608	Aromadendrin-O-glucoside	Flavonoids	C21H22O11	449.10810	449.10892	−1.8100	259.0603/259.0610; 83.0123/83.0125; 123.0078/123.0075; 149.0239/149.0233; 151.0029/151.0026; 153.0190/153.0184	MassBank/MoNA	A1-M; A2-EA; A2-EaM; A2-M; B1-M; B2-EaM; CME96
7		6.858	Taxifolin-O-glucoside	Flavonoids	C21H22O12	465.10263	465.10400	−2.9500	303.0507/303.0512; 285.0415/285.0416; 259.0631/259.0638; 178.9960/178.9952; 125.0244/125.0247; 152.0115/152.0119	MassBank/MoNA	A2-EA
8	P4	7.399	Isoquercitrin/quercetin-O-hexoside	Flavonoids	C21H20O12	463.08694	463.08820	−2.7300	301.0333/301.0335; 255.0322/255.0321; 151.0014/151.0007; 463.0853/463.0865; 163.0008/163.0005	MassBank/MoNA	A2-EA
9	P5	7.797	Astragalin	Flavonoids	C21H20O11	447.09262	447.09329	−1.5000	285.0380/285.0391; 255.0291/255.0293; 227.0344/227.0343; 447.0919/447.0927; 151.0038/151.0036	MassBank/MoNA	A2-EA
10		7.830	Kaempferol-O-glucoside isomer	Flavonoids	C21H20O11	447.09235	447.09328	−2.0800	285.0399/285.0404; 255.0297/255.0299; 227.0344/227.0349; 447.0922/447.0936; 151.0027/151.0026	MassBank/MoNA	A2-EA
11	P6	7.923	Piceatannol	Stilbenes	C14H12O4	243.06602	243.06590	0.4900	159.0443/159.0440; 201.0542/201.0550; 143.0490/143.0492; 243.0650/243.0659; 161.0229/161.0232; 172.0512/172.0517	MassBank/MoNA	A1-M; A2-EA; A2-EaM; A2-M; B1-M; B2-EaM; CME96
12		8.009	Naringenin-7-O-glucoside	Flavonoids	C21H22O10	433.11389	433.11402	−0.30	107.0154/107.0149; 119.0483/119.0481; 151.0023/151.0024; 177.0173/177.0180; 271.0595/271.0595; 313.0677/313.0688	MassBank/MoNA	A2-EA; A2-M; CME96
13		8.407	Aromadendrin	Flavonoids	C15H12O6	287.05579	287.05612	−1.1400	177.0545/177.0549; 153.0182/153.0188; 125.0230/125.0233; 57.0334/57.0332; 65.0035/65.0037; 83.0122/83.0125	MassBank/MoNA	A2-EA
14	P7	10.106	Naringenin	Flavonoids	C15H12O5	271.06133	271.06119	0.5100	151.0029/151.0026; 119.0491/119.0490; 143.0487/143.0491; 271.0602/271.0613; 121.0280/121.0283; 137.0230/137.0232	MassBank/MoNA	A2-EA
15	P11	10.270	Kaempferol	Flavonoids	C15H10O6	285.04011	285.04046	−1.2200	285.0390/285.0400; 229.0479/229.0490; 169.0655/169.0650; 65.0003/65.0000; 95.0139/95.0140; 109.0295/109.0290	MassBank/MoNA	A2-EA
16		10.404	Calealactone B (tentative)	Sesquiterpenoids	C21H26O9	421.14914	421.15039	−2.9800	99.0438/99.0437; 87.0438/87.0437; 159.0448/159.0441; 186.0685/186.0678; 211.0761/211.0760; 293.1025/293.1032	MassBank/MoNA	A1-M
17		11.848	Anisocoumarin H (tentative)	Terpene lactones	C19H22O4	313.14395	313.14453	−1.8500	206.0213/206.0215; 313.1434/313.1443; 174.0317/174.0312; 214.0617/214.0628; 183.1010/183.1014	MassBank/MoNA	A2-EA
18		12.122	Acanthospermolide (tentative)	Sesquiterpenoids	C20H26O6	361.16582	361.16565	0.4800	57.0318/57.0319; 61.9869/61.9869; 87.0077/87.0074; 89.0234/89.0236; 96.9592/96.9588; 218.8663/218.8669	MassBank/MoNA	A1-M; A2-EA; A2-EaM; A2-M; B1-M; B2-EaM
19		13.410	Emodin (tentative)	Anthraquinones	C15H10O5	269.04503	269.04500	0.1200	269.0443/269.0438; 225.0179/225.0168; 223.0363/223.0374; 201.0521/201.0517; 157.0663/157.0668; 83.0481/83.0482	MassBank/MoNA	A2-EA; CME96
20	P12	13.491	Apigenin (tentative)	Flavonoids	C15H10O5	269.04537	269.04550	−0.4700	201.0565/201.0559; 181.0666/181.0659; 169.0654/169.0658; 143.0497/143.0502; 157.0657/157.0658; 167.0496/167.0500	MassBank/MoNA	A2-EA; CME96
21	P8	15.470	6β,8β-Dihydroxyeremophilenolide (tentative)	Sesquiterpene lactone	C15H22O4	265.14539	265.14453	3.2400	96.9592/96.9588; 265.1455/265.1455; 61.9870/61.9870; 197.1595/197.1598	MassBank/MoNA	A1-M; A2-EA; A2-EaM; A2-M; B1-M; B2-EaM; CME96

---

The tentatively annotated features included phenolic acids, procyanidins, flavonoid aglycones, flavonoid glycosides, stilbenes, and sesquiterpenoid-related compounds. The presence of procyanidin-related features, catechin, piceatannol, and several flavonoid glycosides in active fractions provides qualitative chemical support for their redox-related activity. However, direct attribution of activity to individual compounds was avoided because authentic standards, targeted quantification, and isolated-compound bioassays were not performed.

Although A2-EaM exhibited the highest bulk TPC/TFC values, it did not show the strongest NO inhibitory activity. Conversely, B1-M combined relatively high TPC/TFC values with strong preliminary antioxidant responses and the lowest NO IC₅₀. These findings suggest that the biological response was not solely determined by total phenolic or flavonoid content, but may reflect qualitative differences in metabolite composition generated by the extraction workflow. This interpretation remains tentative and requires further validation using isolated compounds or targeted quantification.

3.6 Limitations of the study

A limitation of the present study is the qualitative nature of the LC-MS/MS annotation. Since no authentic standards were used and targeted quantification was not performed, the presence/absence data of the 21 tentatively annotated compounds do not support multivariate statistical analyses such as PCA or PLS-DA. Consequently, the observed associations between metabolite profiles and bioactivities are interpreted descriptively and should be considered hypothesis-generating rather than confirmatory. Future studies employing targeted quantification or metabolomics with proper quality controls are needed to establish statistically robust correlations.

4. Conclusion

This study evaluated how extraction workflow modulates phenolic/flavonoid enrichment, antioxidant activity, NO inhibitory potential, and LC-MS/MS chemical profiles of C. mitis seed extracts. CME96 was selected for ethanol-based workflows because it showed the highest mean TPC and TFC values among the tested ethanol concentrations. Among the fractions, A2-EaM showed the highest TPC and TFC values, whereas B1-M was the most biologically promising fraction because it combined strong preliminary antioxidant responses with the lowest NO IC₅₀ among the selected fractions. Curated LC-MS/MS analysis retained 21 tentatively annotated compounds/features, providing qualitative support that extraction workflow shaped metabolite composition. Overall, the results indicate that optimizing extraction workflow is not merely a matter of maximizing total phenolic/flavonoid content, but a strategy to modulate qualitative chemical composition and biological response.

Author contributions

Dam Thi Thanh Hai: conceptualization, methodology, investigation, data curation, formal analysis, writing – original draft, funding acquisition. Le Thanh: investigation, methodology, data curation, formal analysis, writing – review & editing. Kieu Dang Minh Nhut, Doan Phu Quy, Quach Tong Hung: investigation, data analysis. Le Tien Dung: conceptualization, supervision, writing – review & editing.

Conflicts of interest

The authors declare that none of the work described in this research has been impacted by any known conflicting financial interests or personal relationships.

Data availability

The data supporting the findings of this study are available within the article and its supplementary information (SI). Supplementary information: extraction/fractionation yield data, representative LC-MS/MS chromatograms and spectra, and additional figures and tables. See DOI: https://doi.org/10.1039/d6ra02326f.

Acknowledgements

This research was funded by Petrovietnam University under grant code GV2503, project titled “Evaluation of Flavonoid and Polyphenol Distribution and Antioxidant Activity of Caryota mitis Fruit through Extraction Procedures and LC-MS/MS Analysis Combined with GNPS.”

References

1. A. A. Islam, A. M. Abdel-baky and D. W. Bishay, J. Pharmacogn. Phytochem., 2017, 6, 2559–2562.
2. A. Hamed, A. Abou Zeid, H. El-Rafie, Z. Kandil, R. El-Akad and M. Farag, Egypt. J. Chem., 2020, 63, 5071–5086.
3. S. K. Pattanaik, S. Jena and D. Rath, Int. Res. J. Multidiscip. Scope, 2025, 06, 1168–1180.
4. R. H. El-Akad, A. H. Abou Zeid, H. M. El-Rafie, Z. A. Kandil and M. A. Farag, J. Food Biochem., 2021, 45, e13648.
5. P. T. N. Trinh, N. T. T. Quynh, M. D. Tri, P. N. Minh, N. T. Le Thuy, L. H. Truong, N. N. Tuan and L. T. Dung, Nat. Prod. Res., 2018, 32, 2632–2638.
6. S.-M. Huang, J.-F. Zhang, Yu Qin and L.-H. Mou, Food Sci., 2012, 33, 104–108.
7. A. A. Islam, A. E.-M. Mohamed, M. A. E. Afaf and D. W. Bishay, J. Med. Plants Stud., 2017, 5, 250–255.
8. I. M. Alanazi, H. H. Abo Nahas, D. I. Mohamed, N. Hosny, Y. H. A. Elewa, M. Agarwal, I. A. A. Ibrahim, A. H. Falemban, G. A. Bamagous, E. R. Sindi, T. A. Yousef, S. Rajkhowa, M. Alsunbul and E. M. Saied, Pharmaceuticals, 2025, 18, 1820.
9. A. R. Alzahrani, N. Hosny, D. I. Mohamed, H. H. Abo Nahas, A. Albogami, T. M. I. Al-Hazani, I. A. A. Ibrahim, A. H. Falemban, G. A. Bamagous and E. M. Saied, RSC Adv., 2024, 14, 19400–19427.
10. S. E. Elshaer, G. M. Hamad, S. E. Sobhy, A. M. G. Darwish, H. H. Baghdadi, H. H. Abo Nahas, F. M. El-Demerdash, S. S. A. Kabeil, A. S. Altamimi, E. Alolayan, M. Alsunbul, O. K. Docmac, M. Jaremko, E. E. Hafez and E. M. Saied, Front. Pharmacol., 2024, 15, 1378249.
11. P. T. Nhat Trinh, T. T. Danh, D. C. Tai, P. Van Trung, L. H. Truong, D. T. Le Hang, T. M. Dung and L. T. Dung, J. Ethnopharmacol., 2025, 350, 120043.
12. J.-H. Yoon and S. J. Baek, Yonsei Med. J., 2005, 46, 585.
13. S. Saha and R. J. Verma, J. Taibah Univ. Sci., 2016, 10, 805–812.
14. F. Shahidi and P. Ambigaipalan, J. Funct. Foods, 2015, 18, 820–897.
15. C. Santangelo, R. Varì, B. Scazzocchio, R. Di Benedetto, C. Filesi and R. Masella, Ann. Ist. Super. Sanita, 2007, 43, 394–405.
16. M. Dent, V. Dragović-Uzelac, M. Penić, M. Brnčić, T. Bosiljkov and B. Levaj, Food Technol. Biotechnol., 2013, 51, 84–91.
17. N. Kishore, D. Twilley, A. Blom van Staden, P. Verma, B. Singh, G. Cardinali, D. Kovacs, M. Picardo, V. Kumar and N. Lall, J. Nat. Prod., 2018, 81, 49–56.
18. V. L. Singleton and J. A. Rossi, Am. J. Enol. Vitic., 1965, 16, 144–158.
19. D. E. Djeussi, J. A. Noumedem Kenfack, C. N. Noumedem Anangmo, S. R. Simeni Njonnou, C. Ngongang Ouankou, I. B. Yossa Nzeuwa, C. Yamssi, F. Zambou Ngoufack, J.-R. Kuiate and V. Kuete, Eur. J. Appl. Microbiol. Biotechnol., 2022, 3, 75–80.
20. M. S. Blois, Nature, 1958, 181, 1199–1200.
21. R. Re, N. Pellegrini, A. Proteggente, A. Pannala, M. Yang and C. Rice-Evans, Free Radic. Biol. Med., 1999, 26, 1231–1237.
22. G.-C. Yen and H.-Y. Chen, J. Agric. Food Chem., 1995, 43, 27–32.
23. I. F. F. Benzie and J. J. Strain, Anal. Biochem., 1996, 239, 70–76.
24. D. Mukhopadhyay, P. Dasgupta, D. Sinha Roy, S. Palchoudhuri, I. Chatterjee, S. Ali and S. Ghosh Dastidar, Free Radic. Antioxid., 2016, 6, 124–132.
25. J. B. Habu and B. O. Ibeh, Biol. Res., 2015, 48, 16.
26. M. Bala, P. K. Verma, S. Awasthi, N. Kumar, B. Lal and B. Singh, Nat. Prod. Commun., 2015, 10, 43–48.
27. S. Upendra, B. Manju, K. Praveen, R. Geetanjali, K. Neeraj, S. Bikram and A. Saroj, J. Med. Plants Res., 2010, 4, 2488–2494.
28. M. Bala, S. Kumar, K. Pratap, P. K. Verma, Y. Padwad and B. Singh, Nat. Prod. Res., 2019, 33, 622–627.
29. M. M. Yousef, A.-N. A. Zohri, A. M. G. Darwish, A. Shamseldin, S. A. Kabeil, A. Abdelkhalek, R. Binsuwaidan, M. Jaremko, H. A. Alshwyeh, E. E. Hafez and E. M. Saied, Front. Microbiol., 2023, 14, 1219823.

---