An efficient and clean extraction approach for hazelnut oil with high ceramide: Enhanced biological properties and abundant flavor complexity

Abstract

This study optimized the extraction of hazelnut oil (HO) using subcritical water extraction (SWE), achieving a maximum yield of 42.80% under conditions of 210 °C, 30 min, and a solid-to-liquid ratio of 1:10 g/mL. SWE-derived HO (SW-HO) was compared with commercial (P-HO), cold-pressed (CP-HO), and Soxhlet-extracted (S-HO) oils. All HO types contained over 92% unsaturated fatty acids (UFAs), with SW-HO showing the highest contents of palmitoleic and oleic acids. SW-HO exhibited distinct caramelized flavor, which was attributed to abundant furans and pyrazines. Lipidomics revealed significantly higher ceramide (Cer) content in SW-HO, which demonstrated superior antioxidant activity and significantly enhanced RAW264.7 macrophage proliferation at 400 μg/mL. These findings suggest that SWE effectively produces high-quality HO with elevated bioactive components (Cer, total phenol, and UFAs), enhancing functional properties and sensory attributes. This work provides both technical and theoretical support for the functional preparation and development of HO.

Highlights

• •SWE enables simultaneous heat pretreatment and extraction for hazelnut oil.
• •SW-HO had the highest content of unsaturated fatty acids (palmitoleic and oleic acids).
• •SW-HO, richest in aldehydes, ketones, furans, pyrazines, and pyrroles, endows desired flavor.
• •Lipidomics showed the highest ceramide content in SW-HO compared to other groups.
• •The high ceramide content of SW-HO promotes its functional properties.

1. Introduction

Hazelnuts (Corylus avellana L.) are a globally popular edible nut. They are an important eco-economic tree species in the Northeast of China and Turkey. Locally produced hazelnut products are relatively limited and are primarily consumed fresh or processed through roasting or baking, resulting in limited processing utilization. Hazelnuts are a globally popular oil-rich nut (Balballi & Karabulut, 2025), with hazelnut oil (HO) valued for its high UFA content, particularly oleic acid (C18:1) and linoleic acids (C18:2). These constituents exhibit significant nutritional and health-promoting properties. With the continuous expansion of the hazelnut industry and in-depth research on its nutritional value, HO has increasingly attracted attention in both academia and the consumer market because of its rich nutrients and bioactive components.

Extraction is one of the most crucial steps in the production of plant oils, as it significantly influences their flavor, quality, and functional properties (Wang et al., 2025). Traditional extraction methods present certain trade-offs: while mechanical pressing preserves oil quality and sensory attributes, it tends to yield low extraction efficiency (Bener et al., 2022; Geow et al., 2018). In contrast, Soxhlet extraction achieves higher oil yield but often results in organic solvent residues and requires complex refining steps (Xu et al., 2007). However, traditional extraction methods may not fully optimize the flavor profile or maximize the functional potential of HO. To achieve a superior balance between high oil quality and enhanced functional properties, selecting an advanced extraction approach is crucial. To simultaneously achieve high oil quality and desired functional properties, the appropriate extraction method is crucial. SWE is considered an eco-friendly technique that utilizes water under elevated temperature and pressure to reduce its dielectric constant and increase its ionic product, thereby enhancing its ability to extract bioactive compounds from plant materials (Zhang, Wen, et al., 2020). Although SWE represents a clean and efficient extraction alternative, its typical operation at high temperatures raises concerns. The potential of SWE to release distinctive aromatic compounds from HO is still uncertain, and elevated temperatures tend to degrade nutrients and promote lipid oxidation. Therefore, a comprehensive evaluation of the physicochemical properties and functional activities of HO extracted by SWE is essential. This study aims to fill this research gap and provide valuable insights and a reference for future studies on innovative plant oil extraction techniques.

This research aims to evaluate the quality and functional properties of HO obtained by the SWE method compared to commercial HO, oils extracted by traditional pressing and Soxhlet extraction techniques. Optimization will be conducted using Box–Behnken design and response surface experiment to determine the optimal SWE extraction conditions. The quality of the HO will be assessed based on nutrient content (fatty acids and total phenols), physicochemical properties (quality evaluation, Fourier-transform infrared spectroscopy (FTIR), ultraviolet–visible (UV) spectrum, differential scanning calorimetry (DSC), rheology, and colorimeter), sensory properties (electronic nose (E-nose) and gas chromatography-mass spectrometer (GC–MS)), lipidomics, biological properties (in vitro antioxidant properties-DPPH and ABTS free radical scavenging activity), and immune activities (cell viability, phagocytic activity, and the secretion of inflammatory signaling factors).

2. Materials and methods

2.1. Materials

The raw hazelnuts used in this study were of the “Yuzhui” cultivar, a type of Pingou hybrid hazelnut (Corylus heterophylla Fisch. × Corylus avellana L.), which is a unique Chinese cultivar and one of the main commercial hazelnut varieties grown in the region. The samples were obtained from Jilin Zhenzhen Manor Agricultural Co., Ltd. (Jilin, China). DPPH (2,2-Diphenyl-1-picrylhydrazyl) and ABTS (2,2-Azinobis-(3-ethylbenzthiazoline-6-sulphonate) were purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). All other reagents used in this study were of analytical grade. Lipopolysaccharide (LPS) from Escherichia coli strain 055:B5 was purchased from Beijing Solaibao Biotechnology Co., Ltd. (Beijing, China). The nitric oxide (NO) enzyme-linked immunosorbent assay (ELISA) kit was purchased from Yakeyin Biotechnology Co., Ltd. (Beijing, China), while tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6) ELISA kits were purchased from Shanghai Yuanxin Biotechnology Co., Ltd. (Shanghai, China).

2.2. Extraction of HO by subcritical water

2.2.1. Extraction process

The process of HO extraction is shown in Supplementary Fig. 1. Hazelnut powder (10.00 g) was mixed with a solvent at the specified solid-to-liquid ratio. The premixed slurry was transferred to the high-temperature and high-pressure reactor. The reactor was continuously sealed and exhausted with high-purity nitrogen (99.99%) for 5 min to establish an oxygen-free environment. The temperature was set as designed, and a temperature control system was used to increase the temperature. Extraction time will start after reaching the design temperature. Once the extraction time reaches the predetermined value, the reactor system is rapidly cooled to 60 °C using an internal cooling coil supplied with tap water. The extracted mixture was collected for solid-liquid separation by filtration through a 300-mesh nylon cloth. The filtrate was then subjected to liquid-liquid extraction with petroleum ether in a separating funnel. The upper organic phase was collected and concentrated by rotary evaporation. The organic phase is concentrated through rotary evaporation. The yield is calculated by weighing. The HO yield is calculated according to Eq. 1:

Y = m₁/m₀ × 100%(Eq. 1).

Where, Y represents the HO yield (%), m₁ denotes the weight of HO (g); and m₀ denotes the weight of hazelnut powder (g).

It's worth noting that the use of petroleum ether for extraction was merely for calculating the extraction yield of HO. In the subsequent sample preparation process, the extract was directly subjected to centrifugation to obtain SW-HO.

2.2.2. Single-factor experiment

A single-factor experiment was designed to optimize the yield of HO (Supplementary Table 1). The effects of three factors on the HO yield was determined by adjusting the extraction temperature (130, 150, 170, 190, 210 °C), extraction time (0, 10, 20, 30, 40 min), and solid-to-liquid ratio (1: 10, 1: 15, 1: 20, 1: 25, 1: 30).

2.2.3. Response surface experiment

Based on the results of the single-factor experiments, extraction temperature (A), extraction time (B), and solid-to-liquid ratio (C) were selected as the key independent variables (Supplementary Table 2), with the yield of HO extraction (Y) taken as the response value. The Box-Behnken design (BBD) involving three factors at three levels was employed to optimize the HO yield using Design Expert version 12 (Stat Ease Inc., Minneapolis, MN, USA).

$$\mathrm{Y}={\beta }_0+\sum _{i=1}^k{\beta }_i{X}_i+\sum _{i=1}^k{\beta }_{\mathit{ii}}{X}_i^2+\sum _{i<j}{\beta }_{\mathit{ij}}{X}_i{Y}_j$$ (2)

Where Y is the response variable, X_i and X_j are the independent variables, β0, βi, βii and βij is the intercept, linear, quadratic and interaction effect coefficients.

2.3. Different samples preparation

SW-HO: extracted under the optimized conditions extracted by subcritical water.

S-HO: 2.00 g of hazelnut powder was placed into a filter tube and extracted with petroleum ether. A fully automatic Soxhlet extractor (SXT-6 A, Hangzhou Feiyue Instrument Co., Ltd., China) was used to extract the HO as a control under the following conditions: extraction temperature of 70 °C and extraction time of 1 h. After extraction, rotary distillation was performed to prepare S-HO.

CP-HO: The cold-pressed standardized commercial protocol was provided by the supplier (Linjiang City Zhenshangyuan Family Farm). After the hazelnuts are shelled, qualified and mold-free hazelnut kernels are selected and stir-fried at a temperature of 70 °C for 10 min, which is used to reduce kernel moisture and disrupt cell structures, thereby facilitating oil release and improving extraction efficiency without chemically altering the oil. Following this conditioning step, the kernels were allowed to cool to ambient temperature before being fed into the screw press, ensuring that no residual heat influenced the pressing process. The screw process was conducted in the ambient temperature. The extracted oil is allowed to stand and is filtered to remove the impurities, resulting in a relatively clear hazelnut oil, referred to as CP-HO.

P-HO: Purified commercial HO.

The physicochemical properties and nutritional value of four types of HOs were evaluated.

2.4. Nutrient content of HO

2.4.1. Determination of FA content

Potassium hydroxide (KOH)-methanol solution (2%, w/v, 2 mL) was added to 60 mg of HO, vortexed, and subjected to saponification at 60 °C for 30 min in a water bath to achieve methylation of the FA. After cooling to room temperature, boron trifluoride (14%, w/v, 2 mL) was added to the mixture, vortexed, and heated in a 70 °C water bath for 10 min. After cooling, 2 mL of n-hexane was added and shaken vigorously for 5 min. Following the addition of 4 mL of saturated saline solution, the upper n-hexane layer was collected for further instrumental analysis (Matei et al., 2021). GC–MS analysis (7890B, Agilent Technologies Inc., USA) equipped with an HP-88 capillary chromatography column (100 m × 0.25 mm × 0.20 μm) was performed. Helium was used as the carrier gas at a flow rate of 1.20 mL/min with a split ratio of 30:1. The injection port temperature was set to 250 °C. The oven temperature program started at 100 °C for 10 min, then increased to 240 °C at a rate of 4 °C/min, held for 20 min. Mass spectrometry conditions included an ion source temperature of 220 °C and an interface temperature of 260 °C. The solvent delay time was 8.78 min.

2.4.2. Determination of total polyphenol content (TPC)

TPC was determined using the Folin-Ciocalteu method according to the method (Velasco-Pérez & Ramos-Escudero, 2024) with some modifications. 5 mL of methanol was added to HO (1.0 g), vortexed for 5 min, and centrifuged at 6000 rpm for 6 min. The supernatant was collected in a 25 mL volumetric flask. This extraction process was repeated 5 times to extract the total phenols from the HO, and the combined extracts were finally brought to volume with methanol. Extraction solution (0.2 mL) was taken and mixed with 3 mL of distilled water, followed by the addition of 0.25 mL of Folin-Ciocalteu reagent and 0.75 mL of Na₂CO₃ solution (15%, w/v). The mixture was allowed to react for 1 h, after which the absorbance was measured at 760 nm. The calibration curve (y = 0.00645× + 0.01802, R² = 0.9999) was constructed using gallic acid as standard solutions ranging from 10 to 100 μg/mL.

2.5. Physicochemical properties of HO

2.5.1. Quality properties

Acid value (AV): the AV of P-HO, CP-HO, and S-HO was determined according to the Chinese national standard GB 5009.229–2016. The AV of SW-HO was measured following Manzoor et al. (2023) with some modifications due to its deep color. 3.0 g of SW-HO was accurately weighed into a conical flask, followed by the addition of 30 mL of an ethanol and ether mixture (1:1, v/v). The mixture was vortexed to ensure complete dissolution of the sample. Subsequently, 2–3 drops of phenolphthalein indicator (1% in ethanol) and 20 mL of saturated sodium chloride solution were added to improve phase clarity and enhance endpoint visibility. The solution was then titrated with 0.1 mol/L KOH standard solution under continuous shaking until a faint pink color persisted in the aqueous (saline) layer for 30 s. The AV was calculated as milligrams of potassium hydroxide required to neutralize the free FA in 1 g of oil (mg KOH/g oil).

Peroxide value (PV): All four HOs were measured using a portable acid value and peroxide value analyzer (YT-SG12Z, Yuntang Instrument Co., China) according to the manufacturer's instructions.

Iodine value (IV): was determined by the methodology of the Wijs method (Triyasmono et al., 2023).

Saponification value (SV): was determined according to the method described by Ajikumar et al. (2025). Sample (2.0 g) was placed into a conical flask. 25 mL of 0.5 mol/L KOH in ethanol solution was added to the flask, and the mixture was refluxed for 60 min to ensure complete saponification. The inner walls of the condenser were rinsed with 10 mL of neutralized ethanol (previously neutralized using phenolphthalein as an indicator) to collect any condensed liquid. Subsequently, 5 drops of phenolphthalein indicator were added to the flask. The excess potassium hydroxide was titrated with 0.5 mol/L hydrochloric acid standard solution under continuous swirling until the pink color disappeared and did not reappear within 30 s.

2.5.2. Spectral characteristic

FT-IR: The absorbance spectra of HO were measured by FT-IR (Vertex70, Bruker, Germany) with the attenuated total reflectance (ATR) method. HO was applied to the ATR crystal of zinc selenide, and the spectra were determined over the wavenumber range of 4000–400 cm⁻¹.

UV spectrum: HO (1 mL) was mixed with cyclohexane (9 mL), vortexed for 3 min, filtered through a membrane, and diluted. The UV absorption spectra were recorded and analyzed using ultraviolet–visible spectroscopy (UV-2600i, Shimadzu, Japan) over the range of 200–900 nm.

2.5.3. DSC

HO (10.00 mg) was weighed into an aluminum crucible and hermetically sealed, with an empty crucible used as the reference. The melting and crystallization behavior were analyzed via DSC using the following temperature program: (1) equilibrate at 20 °C for 2 min; (2) cool from 30 °C to −50 °C at a rate of 5 °C/min; (3) hold at −50 °C for 5 min; (4) finally heat back to 20 °C at 5 °C/min.

2.5.4. Rheological properties

The rheological properties of HO samples were measured using Discovery Hybrid Rheometer 1 (DHR 1, TA Instrument, USA) at 25 °C (Janesch et al., 2023).

2.5.5. Colorimeter analysis

The color parameters (lightness (L^⁎), redness (a^⁎), and yellowness (b^⁎)) of the four types of HO were determined using a colorimeter (CS-820, Hangzhou Caipu Technology Co., LTD, China). Distilled water was used to calibrate the colorimeter.

2.6. Sensory analysis

2.6.1. E-nose

The detection and analysis of volatile flavor (Supplementary Table 3) in HO were conducted (Yan et al., 2024). 5 mL of each sample was accurately transferred into a headspace vial and analyzed using a PEN3 E-nose detection system (Ensoul Technology Ltd., Airsence, Germany) via dynamic headspace sampling.

2.6.2. GC–MS analysis

The volatile compounds of hazelnut oils (P-HO, CP-HO, S-HO, and SW-HO) were analyzed using GC–MS. For each sample, 5 mL of oil was transferred into a 20 mL headspace vial. To facilitate compound identification and normalize detection sensitivity across the diverse flavor profiles, varying volumes of 2-octanol (as an internal standard) were added: 10 μL for P-HO, 200 μL for CP-HO, 15 μL for S-HO, and 600 μL for SW-HO. GC analysis was performed using a DB-WAX-UI silica capillary column(60.0 m × 0.25 mm × 0.25 μm), helium as the carrier gas, temperature programming: 40 °C (4 min) → 100 °C (4 °C/min, 2 min) → 150 °C (3 °C/min) → 230 °C (20 °C/min, 5 min). Mass spectrometry conditions: electron ionization (EI, 70 eV), ion source temperature 230 °C, scan range 30–500 m/z (Cebi et al., 2021). Volatile compounds were identified by comparing their mass spectra with the NIST library, and only those with a Similarity Index (SI) > 85 were considered. To ensure methodological robustness and comparability across samples, the results were expressed as relative abundance (%). This was calculated by dividing the peak area of each individual identified compound by the total peak area of all identified volatile compounds (SI > 85) within the sample.

2.7. Lipidomics

Lipids were extracted using the MTBE method. HO samples were mixed with 200 μL of ultrapure water and vortexed for 5 s. Subsequently, 240 μL precooled methanol (−20 °C) was added and vortexed for 30 s, then 800 μL MTBE was added and sonicated for 20 min, centrifuged (14,000 g for 15 min), and the upper organic phase was collected. The solvent was evaporated under nitrogen at 40 °C, and the residue was reconstituted in 200 μL of 90% isopropanol/acetonitrile (v/v), vortexed, centrifuged, and the supernatant was used for analysis. Quality control (QC) samples were prepared by pooling equal amounts of all samples.

Chromatography was performed using a UHPLC Nexera LC-30 A system equipped with a C18 column (1.7 μm, 2.1 mm × 100 mm, Waters) maintained at 45 °C, with a flow rate of 300 μL/min. Mobile phase A consisted of acetonitrile/water (6:4) with 0.1% (w/v) formic acid and 0.1 mM ammonium formate; mobile phase B consisted of acetonitrile/isopropanol (1: 9) with 0.1% (w/v) formic acid and 0.1 mM ammonium formate. The gradient program was as follows: 0–2 min, 30% B; 2–25 min, 30–100% B; 25–35 min, 30% B. Samples were kept at maintained at 10 °C in the autosampler and analyzed in random order. Mass spectrometry: ESI in both positive and negative modes was used on a Q Exactive mass spectrometer. Source temperature: 300 °C, capillary temperature: 350 °C, ion spray voltage: 3.0 kV, MS1 scan range: 200–1800 m/z.

2.8. Biological activity

2.8.1. Antioxidant ability

The DPPH radical scavenging assay was determined in light of Xu et al. (2011) with some modifications. P-HO, CP-HO, S-HO, and SW-HO were dissolved in ethyl acetate to prepare HO stock solutions at a concentration of 1 g/mL. DPPH reaction solution (0.2 mM, 2 mL) was mixed with HO solutions at different concentrations (10–100 mg/mL). All mixtures were incubated at room temperature for 30 min in the dark environment. The absorbance at 517 nm was measured and recorded. The DPPH radical scavenging activity was calculated as follows:

DPPH radical scavenging activity (%) = [1-(A_i-A_j)/A₀] × 100% (Eq. 3).

Where, A_i represents the absorbance of sample and DPPH reaction solution, Aj represents the absorbance of the mixture of the sample and ethyl acetate solution, and A_₀ refers to the absorbance of the mixture of ethyl acetate and DPPH solution.

ABTS radical scavenging activity was determined according to Zhang et al. (2022) with some modifications. ABTS reagent (7.4 mM) and potassium persulfate (140 mM) were mixed and diluted with 60% ethanol to achieve an absorbance of 0.70 ± 0.02 at 734 nm. The HO sample (2 mL) was reacted with 2 mL of the ABTS solution and incubated in the dark environment for 30 min. The absorbance at 734 nm was then measured and recorded. The ABTS radical scavenging activity was calculated using the following formula:

ABTS radical scavenging activity (%) = [1-(A_i-A_j)/A₀] × 100%(Eq. 4).

Where, A_i represents the absorbance of sample and DPPH reaction solution, Aj represents the absorbance of the mixture of the sample and ethyl acetate solution, and A_₀ refers to the absorbance of the mixture of ethyl acetate and ABTS solution.

2.8.2. Immune analysis

2.8.2.1. Cells and culture conditions

The NIH3T3 and RAW264.7 cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and double antibodies (100 U/mL penicillin and 100 μg/mL streptomycin). The cells were maintained in an incubator at 37 °C and 5% CO₂.

2.8.2.2. Preparation of sample solution

500 mg of HO was dissolved in anhydrous ethanol and stirred for 5 min to achieve complete dissolution, resulting in HO stock solutions at a concentration of 20 mg/mL. Then, 100 μL of this stock solution was thoroughly mixed with 1900 μL of DMEM medium to prepare a 1 mg/mL working solution. Considering the potential cytotoxic effects of anhydrous ethanol on RAW264.7 and NIH3T3 cells, all dilutions were prepared using DMEM as the diluent to minimize solvent interference. The 1 mg/mL working solution was serially diluted in DMEM to generate concentration gradients of 5, 10, 30, 60, 80, 100, 200, 300, and 400 μg/mL HO, which were used in subsequent experiments.

2.8.2.3. Cell cytotoxicity

Cell counting kit-8 (CCK-8) was used to evaluate the cell cytotoxicity of HO, dissolved in ethanol solution (5%, v/v) according to Grajzer et al. (2021). NIH3T3 cells in the logarithmic phase were prepared and seeded into a 96-well microplate at a density of 5 × 10⁴ cells per well. After a 24-h incubation to allow cell attachment, the old medium was aspirated from all wells. Then, 100 μL of fresh medium and sample solutions were added to each well and incubated for an additional 24 h. Subsequently, 10 μL of CCK reagent (Beyotime Biotechnology, China) was added to each well and incubated for 1 h at 37 °C in a 5% CO₂ atmosphere. Absorbance was measured using a microplate reader (Spark, Tecan Austria GmbH, Austria) at a wavelength of 450 nm. Culture medium without cells was used as a blank control. RAW264.7 cells were treated similarly, with samples added after a 12-h incubation.

2.8.2.4. The influence of HO on the phagocytic activity of RAW264.7 cell

Phagocytic ability of RAW264.7 cells was determined according to Wang et al. (2020). Following 24 h of co-culture with the test samples and LPS (10 μg/mL), RAW264.7 cells were incubated with 100 μL of 0.075% (w/v) Nneutral Rred staining solution for 1 h. The cells were then gently washed three times with phosphate-buffered saline (PBS) to remove unincorporated dye. After the final wash, 100 μL of ethanol-acetic acid lysis buffer (1:1, v/v) was added to each well to solubilize the internalized dye, and the plates were incubated for 30 min at room temperature. Absorbance was measured at 540 nm using a microplate reader.

2.8.2.5. The influence of HO on the secretion of inflammatory signaling factors

RAW 264.7 cells were incubated in 12-well plates and treated with 1000 μL of HO sample solutions or LPS (10 μg/mL). After 24 h of incubation, the cells were centrifuged, and the culture supernatant was collected. The concentrations of NO, TNF-α, and IL-6 were determined using ELISA kits according to the manufacturer's recommended protocols.

2.9. Statistical analysis

All the experiments were carried out in triplicate, and all data were performed in the form of mean ± standard deviation. All the pictures were plotted by GraphPad Prism 9 and Origin 2021. In addition, Tukey's test was used for multiple comparisons.

3. Results and discussion

3.1. Optimization of the extraction yield of HO

3.1.1. Single-factor experiment

The SWE method, characterized by its shorter extraction time, lower costs, and environmental friendliness, was used to extract plant seed oil (Daud et al., 2022). The extraction temperature (130–210 °C) is a key parameter that significantly affects the extraction yield of HO by altering the physical and chemical properties of water and the structure of hazelnut powder. When the temperature increased from 130 to 170 °C, the extraction yield of HO significantly rose from 15.20% to 29.60% (Supplementary Table 1). This result was mainly attributed to the decrease in the dielectric constant of water and the significant reduction in viscosity and surface tension in the subcritical state, thereby enhancing the solubility of non-polar lipids and the mass transfer efficiency (Zhang, Zeng, et al., 2020). The highest HO yield was observed at an extraction temperature of 210 °C, which was attributed to the enhanced oil release resulting from the disruption of the hazelnut flour cell wall structure (including hemicellulose hydrolysis and lignin softening) by the SWE method at high temperature.

Extraction time has a significant influence on the extraction yield of HO as well (Supplementary Fig. 2). The extraction yield was increased with longer extraction times, which can be ascribed to two reasons: first, the hydrolytic effect of the subcritical water medium on the structure of plant cell walls (mainly acting on hemicellulose and pectin) significantly improves the mass transfer efficiency; second, as the extraction process progresses, the interfacial concentration gradient gradually diminishes, weakening the driving force for mass transfer (Khandelwal et al., 2024). Additionally, the content of extractable free lipids decreases over time. Moreover, the solid-liquid ratio on the extraction yield is significantly lower than that of the extraction temperature and time (Supplementary Table 1 and Supplementary Fig. 2C). Specifically, within the range of 1:10 to 1:25 g/mL, the extraction yield increased with the increase of liquid volume, reaching the highest extraction yield of 34.40% at 1:25 g/mL. This phenomenon can be attributed to the increased liquid volume reducing the viscosity of the system, enhancing the mass transfer efficiency, and simultaneously promoting the full swelling and rupture of the plant cell walls.

3.1.2. Box-Behnken design

The BBD experiment design with three factors and three levels was employed to systematically investigate the effects of the independent variables. Analysis of variance (ANOVA) was used to statistically evaluate the response surface model, and the model's reliability was verified by calculating the regression coefficients and significance levels (p-values). The results of the response surface test design, including actual and predicted yields, are presented in Supplementary Table 4.

There is a significant difference in extraction temperature (A, p < 0.0001) and extraction time (B, p < 0.01), but their quadratic term (A² and B²) was not significant. The solid-to-liquid ratio (C) did not show a significant effect, whereas its quadratic term (C²) was significant (p < 0.05) (Supplementary Table 5). These results are consistent with the conclusions of the single-factor experiment. Interaction analysis revealed that the interactions between A and B and between A and C, were significant (p < 0.05), while the interaction between B and C was not significant. The overall significance test of the model indicated that p < 0.05, and the lack of fit was not significant, confirming a good model fit. Based on these results, the insignificant terms (BC, A², and B²) were eliminated (Supplementary Table 6), and the optimized quadratic polynomial regression equation was established:

Y(%) = 28.67 + 5.76 A + 1.79B-0.9250C + 2.35AB + 2.35 AC + 3.42C²(Eq. 5).

This model can effectively predict the optimal process parameters for extracting HO by SWE, providing a theoretical basis for industrial production. The optimized regression model shows excellent predictive performance and reliability. The coefficient of variation was CV = 5.42 (Supplementary Table 6), confirming that the experiment results are highly repeatable and of high precision.

Three-dimensional response surfaces and contour lines can intuitively illustrate the extent of interaction effects on the extraction yield of HO. The three-dimensional response surface diagram (Fig. 1) analysis indicates that higher extraction yields are achieved at elevated extraction temperatures and longer extraction times. These findings are consistent with the ANOVA results presented in Supplementary Table 6. Based on the analysis of the second-order equation, the optimal extraction conditions were determined as follows: extraction temperature of 210 °C, extraction time of 30 min, and a solid-to-liquid ratio of 1:10 g/mL, and the predicted HO yield is 42.15%. Three parallel experiments were conducted, yielding an actual extraction rate of 42.80% ± 2.15%. This protocol ensures a high extraction yield while maintaining good economic efficiency and operational feasibility.

Fig. 1.

The effect of the interaction of extraction temperature and time (A, B), extraction temperature and solid-to-liquid ratio (C, D), and extraction time and solid-liquid ratio (E, F) on the yield of the hazelnut oil.

3.2. Nutrient content of HO

3.2.1. FA content

The functional properties and nutritional value of plant oil are primarily determined by the composition, specifically, the types and contents of FA present. The FA composition and the proportion of each FA in four types of HOs were analyzed (Table 1). FA are classified based on the number and configuration of carbon‑carbon double bonds in their molecular structures, primarily into saturated fatty acids (SFA) and UFAs, the latter consisted of monounsaturated fatty acids (MUFA) and polyunsaturated fatty acids (PUFA). The ratio of UFA to SFA is a key indicator for evaluating the nutritional quality of oils. A higher UFA/SFA ratio is generally associated with more favorable lipid metabolism properties. The FA in HO mainly consists of UFA, especially oleic acid and linoleic acid, which are similar to Geow et al. (2018). The UFA content in the four HOs was all above 92%. Among them, S-HO (94.73 ± 0.11%) and SW-HO (94.58 ± 0.14%) had the highest UFA content, with no significant difference between them. However, P-HO (92.11 ± 0.19%) had the lowest UFA content, which was significantly different from the others. These findings further demonstrate that S-HO and SW-HO samples possess higher nutritional value among the four types. Moreover, the content of linolenic acid and linoleic acid was highest in CP-HO, while its oleic acid content was the lowest among the four types of HO.

Table 1.

Fatty acid composition of P-HO, CP-HO, S-HO and SW-HO.

Fatty acids	P-HO	CP-HO	S-HO	SW-HO
Palmitic acid, C16:0 (%)	5.42 ± 0.14a	4.40 ± 0.14b	3.66 ± 0.11c	3.79 ± 0.10c
Palmitoleic acid, C16:1 (%)	0.42 ± 0.01a	0.48 ± 0.02a	0.45 ± 0.07a	0.44 ± 0.04a
Stearic acid, C18:0 (%)	2.48 ± 0.05a	1.97 ± 0.03b	1.60 ± 0.02c	1.63 ± 0.04c
Oleic acid, C18:1 (%)	74.99 ± 0.17b	68.64 ± 0.51b	78.78 ± 0.52a	78.26 ± 0.33a
Linoleic acid, C18:2 (%)	15.67 ± 0.14b	21.46 ± 0.55a	14.60 ± 0.38c	14.99 ± 0.39bc
Linolenic acid, C18:3 (%)	1.03 ± 0.09b	3.04 ± 0.13a	0.90 ± 0.06c	0.89 ± 0.06c
Saturated fatty acid, SFA (%)	7.89 ± 0.19a	6.37 ± 0.17b	5.27 ± 0.11c	5.42 ± 0.14c
Unsaturated fatty acid, UFA (%)	92.11 ± 0.19c	93.63 ± 0.17b	94.73 ± 0.11a	94.58 ± 0.14a
Monounsaturated fatty acid, MUFA (%)	75.41 ± 0.16b	69.12 ± 0.51c	79.23 ± 0.46a	78.70 ± 0.29a
Polyunsaturated fatty acids, PUFA (%)	16.70 ± 0.14b	24.51 ± 0.68a	15.50 ± 0.42c	15.88 ± 0.43c
Atherogenic index, AI	0.06 ± 0.00a	0.05 ± 0.00b	0.04 ± 0.00c	0.04 ± 0.00c
Thrombosis index, TI	0.94 ± 0.00b	0.73 ± 0.00c	0.98 ± 0.00a	0.97 ± 0.00a

Nutritional properties of HO contribute to significant biological activity effects in the prevention of cardiovascular diseases. The effects of SFA and UFA on cardiovascular diseases can be predicted using the atherosclerosis index (AI) and the thrombosis index (TI). Oils with a high content of UFA can reduce the thrombogenic and atherosclerotic properties of oil associated with oil consumption. A high content of oleic acid and α-linolenic acid may help prevent and control cardiovascular disease. The AI and TI values of these four types of HO, which are rich in UFA, are very low. This indicates that HO has potential nutritional benefits, consistent with the understanding that healthy oils should have low AI and TI values. Notably, the lowest AI was observed in S-HO and SW-HO, while the lowest TI was found in CP-HO. This further demonstrates that SWE plays a significant role in enhancing the nutritional value of plant oils and has the potential to be developed into a health food beneficial to human health.

3.2.2. TPC

A significant difference in TPC was observed among different types of HO (Fig. 2A). SW-HO exhibited the highest total phenol content (38.34 ± 2.99 mg/g), which was significantly higher than that of the other types: P-HO (19.74 ± 0.71 mg/g), CP-HO (13.27 ± 0.05 mg/g), and S-HO (12.92 ± 0.08 mg/g). Similar to Bener et al. (2022), hazelnuts are abundant in total phenols, especially tannins. The high TPC was ascribed to two factors: first, the presence of a large amount of phenolic compounds in the hazelnut peel (Pelvan et al., 2018), which originated from the raw material and could not be removed during processing; second, the SWE method was particularly effective at obtaining high TPC, especially at high temperature (Daud et al., 2022; Zhang et al., 2022). During the SWE process, the cell membranes of the raw materials were damaged, releasing the phenolic compounds bound in hazelnut powder (mainly hazelnut peel). Consequently, the TPC in SW-HO increased. Moreover, since phenolic compounds are associated with antioxidant properties, it's speculated that SW-HO extracted by the SWE method exhibits strong antioxidant activity, which can be verified in subsequent experiments.

Fig. 2.

Physiochemical properties of hazelnut oil with different extraction methods. Total phenol content (A); FT-IR analysis (B); UV spectra (C); melting diagram (D); crystallization diagram (E); rheological properties (F), and colorimeter analysis (G, H, I) of P-HO, CP-HO, S-HO, and SW-HO.

3.3. Physicochemical properties of HO

3.3.1. Quality properties

The PV is a key indicator for evaluating the degree of oxidation and deterioration in oils, particularly reflecting the accumulation of primary oxidation products. It is one of the core parameters used to measure the freshness and oxidative stability. All four types of HO met the qualification criteria (Table 2). During the complete autoxidation process, oils convert the generated peroxides and hydroperoxides into aldehydes, alcohols, and ketones. This transformation directly leads to the development of rancidity (Sarkar et al., 2024). AV reflects the content of free fatty acids (FFA) and serves as a key indicator for measuring the freshness and degree of hydrolysis in oils (Manzoor et al., 2023). As shown in Table 2, the AV of SW-HO was significantly higher than that of the other samples (p < 0.05). This increase may be attributed to the hydrolysis of triglycerides under the high-temperature conditions during SWE, resulting in elevated FFA content. Future studies could explore the addition of antioxidants, such as tea polyphenols, to the raw material prior to extraction to mitigate this effect.

Table 2.

Quality properties of P-HO, CP-HO, S-HO and SW-HO.

	P-HO	CP-HO	S-HO	SW-HO
Peroxide value	Qualified	Qualified	Qualified	Qualified
Acid value mg KOH/g	0.38 ± 0.00b	0.29 ± 0.01b	0.71 ± 0.03b	7.54 ± 0.41a
Iodine value g I2/100 g	97.08 ± 0.76a	99.46 ± 1.11a	88.13 ± 1.00b	87.53 ± 0.56b
Saponification value mg KOH/g	191.82 ± 3.03a	192.39 ± 5.20a	182.24 ± 2.79b	197.05 ± 0.40a

Note: Peroxide values were determined using a calibrated food quality analyzer. “Qualified” indicates values <0.25 g/100 g, depending on the standard used).

The IV directly reflects the total number of double bonds, which indicates the overall levels of PUFA and MUFA in oils. It serves as an important indicator for assessing the oxidation stability of oils. The IV of CP-HO was the highest; the IV of P-HO and CP-HO was significantly higher than those of S-HO and SW-HO. This finding demonstrates that the oil prepared by the cold-pressed method promotes the production of UFA and a greater number of double bonds, resulting in higher chemical reactivity and poorer oxidation stability; however, it may also offer better fluidity and nutritional properties. The SV reflects the chain length (molecular weight) of the FA in oils (Matei et al., 2021). There was no significant difference between P-HO, CP-HO, and SW-HO. However, all of these were significantly higher than that of S-HO (p < 0.05), indicating a relatively higher content of short-chain fatty acids in SW-HO. Ravber et al. (2015) demonstrated that FFA formation during SWE of sunflower oil is temperature-dependent: FFA levels increased significantly (p < 0.05) when the reaction temperature was raised to 240 °C. Notably, although SW-HO was produced under elevated temperatures, the extraction process was conducted in an oxygen-free environment, thereby eliminating the possibility of oxidative rancidity—oxygen exposure being a key factor in such degradation.

3.3.2. Spectral characteristics

Similar FA compositions result in comparable characteristic absorption patterns in the FT-IR spectra. The positions and relative intensities of the characteristic absorption peaks of the four samples were similar (Fig. 2B), indicating that the HO samples obtained from different hazelnut raw materials and various extraction methods have highly similar chemical compositions. This finding is consistent with the analysis presented in Section 3.2.1 (FA content). These results suggest that the main structural characteristics of HO exhibit good process stability across different extraction methods. The absorbance at 3007 cm⁻¹ is ascribed to the tensile vibration band of the C—H stretching in the cis double bond (Siddique et al., 2015), indicating the relatively high unsaturation present in the HO. The peak at 2925 cm⁻¹ corresponds to the vibrational absorption of the C—H stretching bond in the CH₂ group (Kocabay & Akkaya, 2020). The absorption at 1747 cm⁻¹ is attributed to the esterified carboxyl group. Moreover, the characteristic absorption peak at 1163 cm⁻¹ is assigned to the stretching vibration of the C—O bond in the esterified carboxyl group. The characteristic peak around 723 cm⁻¹ is attributed to the rocking vibration of the –CH₂ group (Siddique et al., 2015).

Unlike with FT-IR results, all HO samples exhibited significantly different ultraviolet absorption characteristics in the wavelength range of 200–400 nm (Fig. 2C). UV–Vis spectroscopy is widely used to provide spectral fingerprints of the electronic properties of molecules. Prior to data treatment, the region between 400 and 800 was removed due to its low signal-to-noise ratio values. Additionally, the region between 200 and 300 should be discarded because the absorbance for the samples was larger than 2 (A > 2) (El Haddad et al., 2023). Therefore, the characteristic absorption peaks of P-HO, CP-HO, S-HO, and SW-HO are concentrated in the wavelength range of 250–300 nm. All four HO samples exhibited varying degrees of UV characteristics, with SW-HO demonstrating the strongest ultraviolet absorption capacity. This finding is closely related to the conjugated double bonds presented in the UFA of the oils.

3.3.3. DSC

DSC is commonly used to analyze the thermal behavior of oils, etc. (Rajagukguk & Tomaszewska-Gras, 2025). All four HO samples exhibited a single melting peak and a single crystallization peak (Fig. 2D & E). Variance in the shape and position of the DSC curves indicates differences in thermodynamic behavior during phase transitions among the samples. During the melting process (Fig. 2D), all samples display a temperature-dependent transition over a range rather than a sharp melting point, a phenomenon associated with the gradual transformation of FA chains within triacylglycerol molecules from an ordered crystalline state to a disordered liquid state (Rajagukguk & Tomaszewska-Gras, 2025). Notably, the P-HO sample exhibits a relatively broader melting temperature range, which may indicate more complex polymorphic transitions occurring within its molecular structure (Siejak et al., 2025). During the crystallization process (Fig. 2E), SW-HO demonstrates significantly different phase transition behavior: it shows a higher crystallization onset temperature and a notably narrower peak width at half-height compared to the other samples. This indicates a greater crystallization driving force and faster crystallization kinetics. These differences may be attributed to the unique influence of the SWE process on intermolecular interactions among triacylglycerols, leading to distinct molecular chain packing arrangements and aggregation behaviors during cooling. Such thermodynamic characteristics are likely related to the composition of FA within the triacylglycerol molecules and their positional distribution on the glycerol backbone.

3.3.4. Rheological properties

HO obtained by different extraction methods all exhibited typical Newtonian fluid behavior (Fig. 2F). The apparent viscosity of each sample remained constant across the tested range, as determined by the slope of the shear stress–shear rate curves. The differences in viscosity among the four HO samples may be attributed to variations in extraction methods or raw materials. P-HO showed the highest viscosity, followed by CP-HO, SW-HO, and S-HO. The viscosity values of P-HO, CP-HO, and SW-HO were close to those reported by (Kim et al., 2010), who demonstrated that the contents of oleic acid and linoleic acid show a highly positive correlation with viscosity, indicating that these two major fatty acids play a decisive role in determining the rheological properties of HO.

3.3.5. Colorimeter analysis

The sensory quality of HO, particularly its color, significantly influences consumers' perception of its overall quality. The color parameters of the four HO samples were determined (Fig. 2G-I). SW-HO exhibited the lowest L* value, indicating a noticeably darker color compared to the other samples (Supplementary Fig. 3). The a* value of SW-HO was higher than those of the other three samples. CP-HO showed the highest b* value, suggesting the most pronounced yellowish hue among the samples. The significantly darker appearance of SW-HO can be attributed to several complex thermochemical mechanisms occurring at temperatures exceeding 200 °C: 1) Caramelization : The high temperatures used in SWE facilitate the dehydration, isomerization, and polymerization of sugars (e.g., sucrose and glucose) leached from the hazelnut matrix. This caramelization process produces brown pigments and volatile aroma (Zhang et al., 2022); 2) Thermal protein degradation and Maillard reactions: Although the SWE environment is oxygen-free due to nitrogen purging, the presence of hazelnut proteins and reducing sugars under superheated conditions (140–200 °C) triggers the Maillard reaction; 3) Water-mediated thermal degradation of lipids. This process produces volatile carbonyl compounds, such as aldehydes and ketones, which possess inherent color and can further participate in secondary browning reactions (Plaza et al., 2010). 4) Additionally, the enhanced solubility of subcritical water likely promotes the co-extraction of dark-colored polyphenolic compounds from hazelnut skins, further intensifying the oil's hue. This observation is consistent with the findings of Pourali et al. (2009), who reported that elevated extraction temperatures promote the co-extraction of non-lipid components and induce browning reactions, thereby intensifying the oil's color. Refining processes could be applied to SW-HO to improve its sensory attributes in the future.

3.4. Sensory analysis

3.4.1. E-nose

E-nose technology enables the visualization of different volatile components and is widely used for food odor identification (Yan et al., 2024). The sensor response values indicated that all HO samples exhibited strong responses to the W5S and W1W sensors (Fig. 3A). Notably, SW-HO showed a significantly higher response to the W5S sensor compared to the other samples (p < 0.05), suggesting that the SWE process may enhance the formation of nitrogen oxides, sulfur-containing compounds, and pyrazines (Xu et al., 2025). These compounds could serve as key aroma markers for distinguishing between different types of HO. The principal component analysis (PCA) results (Fig. 3B) demonstrated that the first two principal components accounted for 86.99% of the total variance. SW-HO was completely separated from the other samples along the PC1 axis, indicating that the SWE method significantly altered the composition of volatile compounds in the oil. This phenomenon may be associated with the formation of Maillard and/or caramelization reactions that occurred during hydrothermal treatments. The combination of E-nose and PCA has proven effective in distinguishing odor profiles among different samples (Yan et al., 2024). SWE method enhances the presence of significant volatile odor compounds in the oil.

Fig. 3.

Sensory analysis of hazelnut oil with different extraction methods. The radar plot (A) and PCA plot (B) of the electronic nose results and GC–MS analysis (C) of P-HO, CP-HO, S-HO, and SW-HO.

3.4.2. GC–MS analysis

The aroma of HO is significantly influenced by volatile compounds (Fig. 3C). A total of 147 volatile compounds were identified across four types of HO (Supplementary Table 7). The numbers of the volatile compounds detected in P-HO, CP-HO, S-HO, and SW-HO were 54, 51, 49, and 38, respectively. Significant differences in volatile profiles were observed depending on the extraction treatment.

The classes of compounds contributing most significantly to the flavor characteristics varied across the four oils. Alcohols, aldehydes, ketones, alkanes, esters, and pyrazines were identified as the primary volatile constituents in hazelnuts, consistent with previous reports (Marzocchi et al., 2017). The highest relative abundance of alcohols, aldehydes and esters were observed in the P-HO. The volatile profile of CP-HO was primarily characterized by high relative abundances of esters, acids, and alcohols. In contrast, the volatile compounds of S-HO are relatively less complex, consisting mainly of alkanes. Furthermore, SW-HO exhibited a high relative abundance of aldehydes, ketones, furans, pyrazines, and pyrroles compared to the other three groups. These findings indicate that SWE promotes the generation of volatile compounds, particularly furans and pyrazines, which contribute to the distinctive aromatic profile of SW-HO.

Aldehydes, which are the primary degradation products of lipids and thermal degradation processes, including thermal degradation and the Maillard reaction, occur at high concentrations but have low odor thresholds, thereby making significant contributions to the aroma profile of food (Liu et al., 2022). Specifically, compounds, 5-Methyl-2-furancarboxaldehyde, α-ethylidene-benzeneacetaldehyde, 2-methyl-propanal, 2-methyl-butanal, furfural, 3-phenyl-furan, indole, and 5-methyl-2-phenyl-2-hexenal and most of pyrazines, furans and pyrizines, were detected only in the SW-HO sample. These compounds are primary thermal degradation products of carbohydrates, imparting a caramel-like odor to the product while simultaneously serving as constituents of caramel color. Although they contribute desirable caramel aromas, they exhibit pronounced bitterness and promote the progression of the Maillard reaction (Liu et al., 2022). Pyrazines, typically generated through the pyrolysis of β-hydroxy amino acids and Maillard reactions, were identified in all HO except S-HO, reaching their maximum relative abundance in SW-HO. The unique presence and high relative abundance of pyrazines and furans in SW-HO suggest that the SWE process significantly accelerates Maillard-derived and caramelization products flavor formation, endowing the distinctive aroma (roasting, nutty, and caramelized) of SW-HO. Combined E-nose and GC–MS analyses confirmed that pyrazine content in SW-HO increased significantly (p < 0.05). SWE more effectively promoted the caramelization and Maillard reaction, generating markedly higher levels of characteristic volatiles. These compounds contribute key roasted, nutty, and caramel notes to SW-HO, underscoring SWE as a promising alternative for producing HO with distinct aromatic profiles.

3.5. Lipidomics

The PCA results based on lipidomics are shown in Fig. 4A. QC samples are tightly clustered in the score plot, indicating good system stability and data reliability. P-HO, CP-HO, S-HO, and SW-HO show clear separation in the PCA space, demonstrating that both extraction methods and raw materials significantly influence the lipidomic profile of HO. A total of 2424 lipid molecules were identified in the HO sample, with 1341 in positive ion mode and 1083 in negative ion mode (Fig. 4B). Triglycerides (TG) are the most abundant, followed by monogalactosyldiacylglycerol (MGDG) and ceramides (Cer). The lipid composition and relative content of the four HO are analyzed (Fig. 4C-F). In P-HO, CP-HO, and S-HO, TG are the predominant lipid class, accounting for 69.912%, 58.015%, and 80.706%, respectively. In contrast, TG is the second most abundant component in SW-HO, representing only 40.481%. The lower TG content in SW-HO compared to the other three oils may be ascribed to thermal hydrolysis and degradation. At the elevated temperature (> 200 °C), water acts as both a solvent and a reactive species (exhibits higher polarity), these conditions can trigger the thermal hydrolysis of TGs into free fatty acids and glycerol and limit the solubility and mass transfer efficiency of non-polar TGs (Zheng et al., 2024). In SW-HO, Cer is the most abundant lipid class (47.443%), which is significantly higher than in the other three oils. Side reactions during SWE may have promoted ceramide formation. Additionally, diglyceride (DG) content in SW-HO (1.587%) was lower than the other three oils. The reduced levels of DG and TG in SW-HO likely result from thermal degradation and hydrolysis during extraction, producing FFA, which can further undergo β-scission, keto-enol tautomerization, or isomerization at carbon‑carbon double bonds to form various hydroperoxides (Yin et al., 2011; Zhou et al., 2023).

Fig. 4.

The PCA map (A); lipid species and number (B); lipid composition of SW-HO (C), P-HO (D), CP-HO (E), and S-HO (F); differential lipid volcano plots: SW-HO/P-HO (G), SW-HO/CP-HO (H), SW-HO/S-HO (I).

Differentially abundant lipid molecules were identified using the criteria of p < 0.05 and fold change (FC) > 1.5 (up-regulated) or FC < 0.67 (down-regulated) (Fig. 4G-I). Among the top 10 up-regulated lipids, Cer predominated. Compared to the other three types of HO, Cer species in SW-HO were predominantly up-regulated. A total of 265, 310, and 218 significantly different lipid molecules were identified in the SW-HO/P-HO, SW-HO/CP-HO, and SW-HO/S-HO groups, respectively (Supplementary Table 8). These significantly altered lipids belong to five major classes: sphingolipids (SP), glycerolipids (GL), glycerophospholipids (GP), saccharolipids (SL), and fatty acyls, with SP accounting for more than half of the metabolites. SP in HO is primarily formed through the reaction of FA with Cer. SPs may play positive roles in reducing inflammation and promoting skin repair (Kendall et al., 2017). The bubble plot (Supplementary Fig. 4 A-C) visually displays the significantly different lipid molecules identified in each group. Similarly, Cer are the predominant type among these differentially abundant lipids (Supplementary Table 9), which may influence the biological activity and could serve as potential biomarkers for distinguishing different types of HO. Among all significantly altered lipid molecules, the top 10 metabolites with the lowest p-values were identified in the SW-HO/P-HO, SW-HO/CP-HO, and SW-HO/S-HO groups, and the results are summarized in. Notably, Cer(d18:0_18:2) (Cer-1), Cer(d18:0_16:0) (Cer-2), Cer(t18:1_18:2) (Cer-3), and Cer(t18:1_18:1) (Cer-4) showed significant contributions to the SW-HO group (p < 0.05), indicating that these Cer species are likely the most prominent potential markers for differentiating SW-HO from the other HO.

3.6. Biological properties

3.6.1. Antioxidant ability

The physicochemical properties and chemical composition of oils may influence their biological activities. The DPPH radical scavenging activity of four HO showed a positive correlation with their concentration (Fig. 5A). SW-HO exhibited significantly stronger DPPH radical scavenging capacity compared to P-HO, CP-HO, and S-HO. At a low concentration of 10 mg/mL, SW-HO achieved a scavenging rate of 51.28%, which was already significantly higher than the maximum scavenging rate of P-HO (26.01%) observed at the highest tested concentration, demonstrating the superior antioxidant properties of SW-HO. A relatively low concentration of SW-HO can achieve a desirable antioxidant effect. The IC₅₀ values of SW-HO were significantly lower than those of other groups, further indicating that SW-HO possesses the strongest antioxidant capacity among the tested samples. The ABTS radical scavenging capacity of the four HO samples was generally consistent with the DPPH results (Fig. 5B). Additionally, the IC₅₀ value for the ABTS radical scavenging capacity of SW-HO was significantly lower than those of the other groups, indicating that SW-HO exhibited the strongest ABTS radical scavenging activity. These findings demonstrate that SWE can be effectively used to enhance the antioxidant capacity of the oil system, which is similar to that of Guo et al. (2021). This enhancement could be attributed to two potential mechanisms: first, this technique enables efficient enrichment of plant-derived phenolic compounds; second, the high temperatures involved in the extraction process may induce Maillard reactions, thereby promoting the formation of antioxidant-active products such as melanoidins (Y. Zhang et al., 2020).

Fig. 5.

Biological properties of hazelnut oil with different extraction methods. DPPH radical scavenging activities (A), IC₅₀ values of DPPH radical scavenging activities (B), ABTS radical scavenging activities (C), and IC₅₀ values of DPPH radical scavenging activities (D). Effect of P-HO, CP-HO, S-HO, and SW-HO on the phagocytic activity (E), NO secretion (F), TNF-α secretion (G), and IL-6 secretion (H) of RAW264.7 cells.

3.6.2. Cell cytotoxicity

The cell cytotoxicity of the four types of HO samples on NIH3T3 cells (Supplementary Fig. 5) and RAW264.7 cells (Supplementary Fig. 6) was evaluated. To ensure the solubility of the lipophilic components, HO was initially dissolved in anhydrous ethanol before being diluted in DMEM to the target concentrations (5–400 μg/mL). For the NIH3T3 cells, viability remained approximately 90% across the entire concentration range for all four HO types, and no dose-dependent relationship between cell viability and concentration was observed. Analogously, within the concentration range of 5–400 μg/mL, the cell viability of RAW264.7 cells treated with the four HO samples remained above 100%, indicating that none of HO exhibited cell cytotoxic effects, nor did the accompanying ethanol vehicle (even in the highest concentration group) exert cytotoxic effects. Furthermore, cell viability of RAW264.7 cells was increased with rising concentrations, suggesting that HO may promote cell proliferation. Among the samples, SW-HO demonstrated a relatively stronger effect in enhancing the proliferation of RAW264.7 cells. Based on these cytotoxicity evaluation results, a concentration range of 0–400 μg/mL was deemed to be safe and was selected for use in subsequent experiments.

3.6.3. Immune analysis

3.6.3.1. Phagocytic activity

Phagocytic ability is a critical immune defense mechanism that plays a key role in clearing pathogenic microorganisms as well as apoptotic and necrotic cells (Dong et al., 2020). RAW264.7 macrophages, widely used for their significant immunological functions, participate in pathogen elimination and the maintenance of immune homeostasis through their phagocytic ability. Since macrophages often require priming by immune activators to induce measurable functional responses in vitro, whether hazelnut oil (HO) could exert a potentiating effect on RAW264.7 macrophages under simulated inflammatory conditions (co-culture with LPS) was evaluated. A blank control group was utilized to establish a baseline, confirming that the observed increases in phagocytic rates were specifically induced by the HO samples or LPS stimulation rather than spontaneous cellular fluctuations.

The effect of HO on the phagocytic activity of these cells was evaluated (Fig. 5E). All four HO types possessed the potential to enhance the phagocytic activity of macrophages in a dose-dependent manner, suggesting the possible immunomodulatory or immune-activating effects of HO. Notably, P-HO, CP-HO, and S-HO samples at a concentration of 400 μg/mL exhibited the highest phagocytic activity, which was significantly different from other concentrations (5–300 μg/mL) (p < 0.05). And SW-HO demonstrated the most potent phagocytic activation compared to the other HO samples (p < 0.05), indicating that the unique composition of SWE-extracted oil may be more effective in priming immune cells. Beyond evaluating immune-enhancing potential, the phagocytosis assay served as a critical screening tool to identify the optimal dosage for subsequent functional studies. Based on the robust phagocytic performance and the statistical significance observed, three specific concentrations: 5 μg/mL (low), 100 μg/mL (medium), and 400 μg/mL (high) were selected for further investigation into cytokine secretion and other downstream immunological markers.

3.6.3.2. The secretion of inflammatory signaling factors

NO is a critical signaling molecule in macrophages, mediating cellular activation and immune regulation, which are essential for defense against pathogenic microorganisms and tumor cells (Dong et al., 2020). As shown in Fig. 5F, LPS stimulation significantly increased NO release to 28.94 μmol/L, which was significantly higher than that of the blank control group (p < 0.05). Regarding the HO treatments, P-HO, CP-HO, and S-HO at low (5 μg/mL) and medium (100 μg/mL) concentrations maintained NO secretion at relatively low levels, with no significant difference observed at low levels of NO secretion (< 20 μmol/L) between the 5–100 μg/mL doses within these groups. However, all HO samples exhibited a significant, dose-dependent increase in NO production, peaking at the maximum at 400 μg/mL of HO. Notably, SW-HO demonstrated a superior capacity to induce NO secretion compared to the other three HO groups across all tested concentrations. At the highest concentration (400 μg/mL), the NO secretion induced by SW-HO was comparable to that of the LPS-treated group (p > 0.05). These findings suggest that SW-HO can promote NO secretion by effectively activating cell activity and proliferation in a dose-dependent manner, which is consistent with the analysis of cell viability, potentially enhancing its immunomodulatory functions.

TNF-α, a prototypical pro-inflammatory cytokine, plays a critical role in immune regulation. The LPS group exhibited a significantly higher TNF-α secretion level compared to the blank control group. HO promoted TNF-α release from cells in a concentration-dependent manner (Fig. 5G). Notably, TNF-α levels in the SW-HO groups were higher than those in the LPS group (188.91 ± 2.49 ng/L) at all tested concentrations and also exceeded those observed in the P-HO, CP-HO, and S-HO groups. RAW264.7 cells participate in immune regulation by IL-6, a cytokine that plays a crucial role in antibody production and allergic responses. Similarly, SW-HO induced significantly greater IL-6 production than the LPS at all three concentration levels and other HO group at equivalent concentrations (Fig. 5H).

Notably, the SW-HO group displayed a superior capacity to elicit TNF-α and IL-6 production, with levels significantly exceeding those of the LPS-positive control and the other three different HO treatments (P-HO, CP-HO, and S-HO) across the entire concentration range. These results demonstrate that apparent SW-HO possesses immune-activating effects on RAW264.7 macrophages. Combined with the analysis of the phagocytic activity, these results demonstrated that SW-HO not only promotes macrophage proliferation but also significantly enhances the secretion of key inflammatory mediators (NO, TNF-α, IL-6), thereby boosting macrophage activity. Moreover, SW-HO (possibly a water-based method or a specific polar solvent extraction) may have enriched polar components, such as glycolipids, lipopolysaccharide analogues, or specific phytosterols/phenols, which can be recognized by pattern recognition receptors on the surface of macrophages. High concentrations of active substances can more extensively bind to receptors, activating downstream NF-kB or MAPK signaling pathways, resulting in the large-scale transcription of pro-inflammatory factor genes. Furthermore, this dose-dependent duality (immune-enhancing at low levels but pro-inflammatory at high concentrations) aligns with the known hormetic effects of plant polyphenols. Although many vegetable oils have been studied for their “anti-inflammatory” effects, many natural extracts exhibit the properties of immunostimulants under specific conditions. The “bidirectional regulation” range of the sample: Many natural products exhibit anti-inflammatory properties at low concentrations, but at high concentrations, they may become pro-inflammatory due to cytotoxicity or excessive activation. Given the dual role in immune defense and cytotoxicity of HO, further investigation of SW-HO's immunopathological potential is warranted.

3.7. Correlation analysis

To investigate the potential relationships between the physicochemical properties and biological activities, a correlation analysis (Pearson correlation, p < 0.05, n = 4) was performed (Fig. 6).

Fig. 6.

Correlation analysis between physicochemical properties and antioxidant activity (A) and physicochemical properties and immunological activity (B). Note: Pearson correlation, p < 0.05. Note: Cer −1, Cer(d18:0_18:2); Cer-2, Cer(d18:0_16:0); Cer-3, Cer(t18:1_18:2); Cer-4, Cer(t18:1_18:1).

Significant correlations were observed between the physicochemical properties and antioxidant activities of HO (Fig. 6A). Ceramide lipids and TPC showed negative correlations with the IC₅₀ of free radical scavenging activities. Specifically, Cer exhibited negative correlations with the IC₅₀ of DPPH, while Cer-1, Cer-2, and Cer-4 were significantly negatively correlated with the IC₅₀ of ABTS values (p < 0.05), TPC was negatively correlated with the IC₅₀ values of DPPH and ABTS as well, respectively. These results indicate that these TPC and ceramide compounds in SW-HO effectively enhance its antioxidant capacity. Cer-1, Cer-2, Cer-3, and Cer-4 were significant positively correlated with the TPC (p < 0.05). Moreover, AV was positively correlated with Cer (p < 0.05) and UFA/MUFA, suggesting that oils with a higher degree of unsaturation may be more prone to partial hydrolysis or oxidation during extraction. The correlation analysis between physicochemical properties and immunological activities is also presented (Fig. 6B). TPC and Cer-3 showed a significant positive correlation with NO secretion (p < 0.05). SV and SFA (palmitic and stearic acids) were negatively correlated with TNF-α and IL-6 secretion, whereas TPC, Cer, UFA, MUFA, and C18:1 were positively correlated with NO, TNF-α, and IL-6 secretion, which may be ascribed to high-content Cer being convenient for cell activity and proliferation.

3.8. Research limitations and future perspective

3.8.1. Research limitations about energy consumption

Although SWE demonstrates superior efficacy compared to conventional extraction methods, yielding hazelnut oil with higher extraction efficiency and enhanced retention of volatile aroma compounds, the current SWE process remains energy-intensive due to the high-temperature heating and subsequent cooling cycles required. However, significant potential exists for energy recovery and process optimization. For instance, waste heat from the cooling water or the hot reactor effluent can be captured and repurposed to preheat the incoming feed water or extraction medium, thereby substantially reducing net energy demand. Moreover, transitioning from a batch-based system to a continuous-flow reactor configuration would further enhance energy efficiency. Continuous operation facilitates steady-state thermal management and enables more effective heat integration across process stages, minimizing thermal losses and improving overall process sustainability. Future research should therefore focus on integrating heat recovery systems and scaling up SWE to continuous-flow operation to improve its economic and environmental viability for industrial-scale hazelnut oil production.

3.8.2. Potential valorization pathways of oil residue

To align with the principles of a circular bioeconomy and achieve a zero-waste process, the side streams generated during extraction, specifically the solid residue post-filtration and the aqueous phase post-liquid-liquid extraction (LLE), must be strategically reintegrated into the value chain. Due to the abundant hazelnut oil residue at high protein, the residue serves as a premium feedstock for plant protein concentrates or plant protein isolates and animal nutrition and aquaculture. And the aqueous phase post-LLE is enriched with polar components leached during SWE, this aqueous phase represents a promising source of value and could be further utilized as follows: Bioactives and flavor enhancers can be repurposed as a natural antioxidant syrup or a flavor enhancer, utilizing Maillard reaction precursors such as free amino acids and reducing sugars; Microbial Feedstock: the abundance of sugars and nitrogenous compounds makes this an ideal fermentation medium and anaerobic digestion. However, some safety & toxicological constraints, especially thermal degradation products during the SWE process could be monitored.

3.8.3. Potential hazards

PV alone is insufficient to fully characterize oxidative or thermal degradation, where radical-mediated, sufficient thermal degradation pathways dominate under high-temperature SWE conditions. Due to PV being assessed as a parameter for primary oxidation products, p-anisidine (p-AV), thiobarbituric acid reactive substances (TBARS), Total Oxidation (TOTOX) were always used to analyze secondary oxidation compounds. Therefore, the critical indices of p-AV, TBARS, and TOTOX are necessary for assessing the accumulation of secondary oxidation products, particularly aldehydes, which are directly linked to rancidity, off-flavors, and potential health implications. Moreover, some heat-induced contaminants, including trans fatty acids, oxidized lipid polymers, and heterocyclic aromatic amines, can be generated during the SWE process. These compounds may form through thermal isomerization, radical coupling, or reactions with amino groups during heating, even under oxygen-free conditions. Further studies should focus on toxicological risk assessments and the development of methods to identify and quantify such contaminants in plant oils extracted by SWE.

4. Conclusion

In this study, the optimal process conditions of SWE for HO were determined through a combination of single-factor experiments and response surface methodology. Compared to other extraction method, SW-HO was found to contain a large amount of special flavor substances (furans, pyrazines, pyridines, pyrroles and thiophenes), which endowed the unique roasting, nutty, and caramelized aroma. Lipidomics analysis revealed that a significantly higher Cer content behaved in SW-HO than that in other three types of HO. These differential lipids can serve as potential markers to distinguish among the four HO. The high TPC, Cer and UFA content likely contribute to the enhanced nutritional and bioactive properties of SW-HO. This study successfully developed a novel extraction method for producing HO with a distinctive flavor profile, high ceramide content, and promising antioxidant and immune activation ability. However, future research should focus on mild refining technologies (e.g., selective membrane adsorption) to balance color improvement with nutrient retention, particularly preserving the high content of ceramides and phenolic antioxidants in SW-HO. Moreover, SW-HO is ideally suited as a functional ingredient for food products or cosmetic applications. The findings not only provide a theoretical basis for developing clean and efficient extraction techniques for HO but also lay the foundation for future research on aroma formation and metabolite byproducts of HO under high-temperature processing.

CRediT authorship contribution statement

Fan Zhang: Writing – review & editing, Writing – original draft, Investigation, Funding acquisition, Formal analysis, Data curation. Lichen Zhang: Methodology, Investigation, Formal analysis, Data curation. Xujiao Zheng: Methodology, Investigation, Data curation. Feng Qiao: Investigation, Data curation. Ruhui Liu: Investigation, Formal analysis. Wenna Ma: Investigation, Formal analysis. Yitong Hou: Investigation. Yirun Zhou: Software. Wenchuan Guo: Formal analysis. Xinhua Zhu: Data curation. Lisong Liang: Supervision, Resources. Xin Lü: Writing – review & editing, Supervision, Funding acquisition. Xin Wang: Writing – review & editing, Funding acquisition, Formal analysis.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.