From waste to wonder: the ultrasonic-enzyme-fermentation process transforms virgin coconut oil press cake into a food ingredient

TL;DR

This study shows how to turn coconut oil waste into a nutritious food ingredient using sound waves, enzymes, and fermentation.

Contribution

A novel sequential bioprocess combining ultrasonication, enzymatic hydrolysis, and fermentation to upcycle coconut oil press cake.

Findings

Ultrasonication achieved 63% solubilization and reduced particle size to 0.35 µm.

Enzymatic hydrolysis increased soluble protein and free amino groups significantly.

Fermentation improved solubility and foaming functionality of the final product.

Abstract

•Sequential bioprocess efficiently transforms VCOPC into a valuable food ingredient.•Continuous ultrasonication achieved 63% solubilization and 0.35 µm size.•Enzymatic hydrolysis increased soluble protein and free amino groups.•Fermentation improved solubility and foaming functionality.

Sequential bioprocess efficiently transforms VCOPC into a valuable food ingredient.

Continuous ultrasonication achieved 63% solubilization and 0.35 µm size.

Enzymatic hydrolysis increased soluble protein and free amino groups.

Fermentation improved solubility and foaming functionality.

Virgin coconut oil press cake (VCOPC), a protein and fiber rich byproduct of coconut oil extraction, remains underutilized despite its nutritional potential. In this study, we developed a continuous bioprocessing strategy that integrates ultrasonication, enzyme-assisted hydrolysis, and lactic acid bacteria fermentation to transform VCOPC into a food ingredient with enhanced nutritional and functional properties. The ultrasonication process achieved a 63% solubilization of the press cake under optimized conditions (1.62 bar, 95 min), it effectively broke down the rigid polysaccharide–protein–lipid matrix and reducing particle size from 2.5 µm to 0.35 µm. Enzymatic treatment markedly increased soluble protein content (up to 137 mg/g) and free amino groups (up to 11.8 mmol/g), while fermentation with L. acidophilus and L. rhamnosus resulted in a 3-log increase in bacterial growth. Functional assessment of the various ingredients obtained by this bioprocessing approach showed high solubility and foaming capacity due to superior interfacial activity. This study demonstrates the successful conversion of coconut by-products into a nutritionally enriched food ingredient with enhanced techno-functional properties. Ultrasonication contributed to the highest yield, while enzymatic and microbial processes delivered the desired biochemical transformations. Combining initial mechanical disruption with enzyme hydrolysis and fermentation represents an effective and sustainable pathway for by-products upcycling and fosters circular innovation within the tropical food economy.

Figures

Process flow diagram illustrating the bioprocessing of virgin coconut oil press cake (VCOPC) into food ingredients through sequential ultrasonication, enzyme-assisted treatment, and lactic acid bacteria fermentation.

Microstructural alterations of the virgin coconut oil press cake (VCOPC) matrix during ultrasonication and structural organization of four food ingredients (FIs). Confocal micrographs show the distribution of cell wall polysaccharides (Calcofluor White, cyan), proteins (Rhodamine B, red), and lipids (BODIPY 505/515, green). Left: Progressive disintegration of the VCOPC matrix at different ultrasonication times (0–95 min). Right: Microstructures of four food ingredients; U-FI (ultrasonicated), E-FI (ultrasonicated and enzyme-treated), A-FI (ultrasonicated and fermented with *L. acidophilus*), a

Heatmap of volatile organic compounds (VOCs) identified in virgin coconut press cake (VCOPC) and food ingredients produced through ultrasonication (U-FI), enzymatic treatment (E-FI), and fermentation with *L. acidophilus* (A-FI) and *L. rhamnosus* (R-FI). Color intensity represents min–max normalized (0–1 scaled) relative abundance of each compound.

Full text

Introduction

1

The coconut (Cocos nucifera L.) is a highly versatile tropical crop that contributes significantly to both nutrition and industry across equatorial regions. According to the Food and Agriculture Organization, global coconut production reached approximately 62.5 million metric tons in 2022, with Indonesia, Philippines, Sri Lanka and India accounting for nearly 75% of total output. Production is expected to continue expanding modestly through 2034, supported by increasing demand for coconut-derived food ingredients, functional oils, and plant-based products [1]. This consistent growth underlines the crop’s importance as both a dietary staple and a renewable industrial raw material across tropical economies.

Among coconut value-added products, virgin coconut oil (VCO) is widely utilized in food, cosmetic and structured-lipid applications owing to its high oxidative stability and desirable medium-chain fatty acid profile. Recent work has shown that VCO can serve as a substrate for the biosynthesis of structured lipids enriched with long-chain polyunsaturated fatty acids (DHA/EPA) which would allow their inclusion in infant-formula [[2], [3]]. VCO production generates a nutrient dense by-product known as virgin coconut oil press cake (VCOPC). This residue, often discarded or used as animal feed, is rich in protein, dietary fiber, residual oil, and minor bioactive compounds. It is naturally free from food additives and is not exposed to harsh processing conditions during production. Its limited use in human food formulations arises mainly from its compact, insoluble structure and the entrapment of proteins within a rigid polysaccharide–fiber matrix. Transforming this underutilized by-product into a nutritionally valuable ingredient therefore represents a key opportunity for sustainable resource recovery within the circular food economy [[4], [5]]. Bioprocess tools offer a sustainable and effective means of unlocking the nutritional and functional potential of VCOPC, paving the way for its utilization in value-added food systems. A combination of physical, enzymatic, and microbial technologies offers a powerful route to restructure complex plant materials. Ultrasonication, through acoustic cavitation, can disrupt cell walls, reduce particle size, and improve solvent penetration, thereby facilitating the release of soluble proteins and polysaccharides. Enzyme-assisted hydrolysis, using proteases and carbohydrase such as cellulases and hemicellulases, can selectively target protein–carbohydrate linkages, promoting solubilization and the release of smaller peptides and fermentable sugars under mild conditions [[6], [7], [8]].

Complementary microbial processes further enrich the transformation by introducing enzymatic activity and metabolites that can enhance flavor, texture, and nutritional attributes. Integrating these methods provides a systematic approach for converting dense plant matrices into food ingredients with improved bio accessibility and application potential [[9], [10], [11]].

In this work, all processing stages were implemented using prototype continuous systems developed to closely emulate industrial-scale operation. Continuous ultrasonication, enzymatic hydrolysis, and bacterial fermentation were sequentially integrated into a single framework to ensure process uniformity, reproducibility, and scalability.

In this way we aim to convert VCOPC into a structurally modified, nutritionally enhanced matrix suitable for food formulations. The research focused on understanding how sequential continuous treatments influence protein solubilization, the release of soluble fractions, and compositional and structural transformations within the material.

Material and methods

2

Materials

2.1

The virgin coconut oil press cake (VCOPC) was supplied by SL Food Tech (Private) Limited (Colombo, Sri Lanka). The coconuts used originated from Sri Lanka and belong to the cultivar of CRIC 65. Flavourzyme® (1000 LAPU/g) enzyme was provided by Novozymes (Bagsværd, Denmark). The probiotic strain of Lactobacillus acidophilus LF08, and Lacticaseibacillus rhamnosus ATCC 53103, were provided by Probiotical S.p.A. (Novara, Italy). All other chemicals were analytical grade and obtained from Sigma or Merck (Merck Life Science NV, Amsterdam, The Netherlands).

Bioprocessing of virgin coconut oil press cake

2.2

Four distinct food ingredients (FIs) were developed from VCOPC using integrated bioprocessing strategies that combined ultrasound-assisted pretreatment with either enzymatic hydrolysis or bacterial fermentation (Fig. 1). This process began with the establishment of a continuous ultrasonication unit and the optimization of ultrasonication conditions, as explained below.Fig. 1. Process flow diagram illustrating the bioprocessing of virgin coconut oil press cake (VCOPC) into food ingredients through sequential ultrasonication, enzyme-assisted treatment, and lactic acid bacteria fermentation.

VCOPC was dispersed in water and subjected to continuous ultrasonication under optimized conditions, after which the treated material was divided into batches for specific upstream processing. The four food ingredients (FIs) obtained were: sonicated (U-FI), enzyme-treated (E-FI), fermented with L. acidophilus (A-FI), and fermented with L. rhamnosus (R-FI).

Design of the Experiment

2.3

A response surface methodology (RSM) based on a central composite design (CCD) and fitted with a quadratic model was employed to investigate the influence of extraction parameters on the recovery of soluble matter. Screening trials were conducted to distinguish parameters requiring optimization from those to be held constant for the VCOPC matrix. Based on these screenings, the solid–liquid ratio was fixed, as changes beyond this level did not further improve the response, while the highest sonication amplitude tested resulted in the highest solubilization response and was therefore fixed for the RSM. Temperature variation did not further enhance the response and showed limited stability during prolonged continuous operation therefore, temperature was controlled.. Preliminary continuous screening across different pressure levels and treatment durations was used to define a practical operating window, balancing solubilization response, safe operation, and the feasibility of long-term continuous sonication. Pressure and sonication time within this window were therefore selected for RSM optimization. Two independent variables, pressure and sonication time were evaluated at three coded levels (−1, 0, and + 1). Additionally, five replicates at the center point were incorporated to assess experimental error and model adequacy, resulting in a total of 13 experimental runs.

The RSM was applied to determine the optimal extraction conditions and to describe changes in the response as a function of varying factor levels. The RSM predicts the simultaneous effects of three levels of each experimental factor and establishes the response variable as a function of the experimental factors, their interactions, errors, and quadratic effects. An empirical second-order polynomial model was used to obtain a regression model of the experimental data.

Arrangement of the Sonostation

2.4

The SonoStation (Hielscher Ultrasonics GmbH, Germany) is a self-contained ultrasonic setup comprising an agitated tank with a cooling jacket, a pump, and ultrasonic processors equipped with flow cells. The ultrasonication system consisted of an ultrasonic processor (UIP2000hdT, 2000 W, 20 kHz) integrated with a titanium transducer (IP65 grade) and a digitally controlled generator. The processor was fitted with a sonotrode (BS4d40) featuring a 40 mm tip diameter and was connected to the SonoStation’s 38 L stirred jacketed tank with flow cells and pumps, enabling continuous inline sonication. The jacket of the SonoStation and the ultrasonic flow cell reactor were connected to a digital heating circulating bath (Grant Optima TXF200-ST38, 38 L; UK), which was operated at 8 °C to maintain a constant processing temperature during sonication.

Ultrasonication of VCOPC

2.5

VCOPC and potable water were mixed in a 1:5 w/w ratio, in the SonoStation and overhead stirred at 200 rpm for 1 h to ensure complete hydration. Continuous sonication was performed according to the experimental design described in the Supplementary Material. The validated ultrasonication conditions produced sufficient sonicated VCOPC for subsequent bioprocessing steps.

Determination of dry matter content representing the soluble fraction

2.6

The dry matter content (DM) of the supernatant, representing the soluble fraction of VCOPC (DMSF), was determined. Samples collected after each run were centrifuged at 500 g for 15 min at room temperature. Subsequently, 2 mL of the supernatant was transferred into pre-weighed aluminium cups and dried at 105 °C until constant weight was achieved. The DMSF was calculated from the difference between the initial and final weights.

Enzyme assisted treatment

2.7

Ultrasonicated VCOPC was transferred to a 50 L fermentation vessel (Brew-Tek FV, BTFV 50 L, Australia) equipped with a hot-water circulation system. Once the temperature reached 50 °C, Flavourzyme® was added at 20 µL g^−1^ of ultrasonicated VCOPC, corresponding to an enzyme dosage of 20 LAPU g^−1^. The mixture was then stirred continuously at 100 rpm for 4 h at 50 °C.

Preparation of bacterial inoculum and bacterial fermentation

2.8

L. acidophilus and L. rhamnosus were inoculated into MRS broth and incubated at 37 °C for 24 h. Cells were harvested by centrifugation at 3200 g for 10 min at 4 °C, washed twice with Ringer solution, and resuspended. Plate counting on MRS agar was performed to determine the initial bacterial load.

Ultrasonicated VCOPC was batch-pasteurized in a jacketed kettle at 90 °C for 10 min under continuous agitation to reduce the initial microbial load prior to bacterial inoculation and was then hot-filled into pre-sterilized 20 L bioreactors. The contents were cooled to 37 °C by circulating cold water, and the headspace was flushed with N_2_. Once 37°C was reached, inocula of the two strains were aseptically transferred into separate bioreactors and fermented for 24 h with overhead mixing at 50 rpm. Each bioreactor received 110 µL of inoculum per gram of ultrasonicated sample, corresponding to an initial cell load of 4 –5 × 10^6^ CFU/mL. Samples were collected at the end of fermentation to determine the bacterial load.

Continuous centrifugation

2.9

The enzymatically treated and fermented VCOPC batches were subjected separately to continuous centrifugation using a continuous-flow rotor (Biofuge 17 RS, Heraeus Sepatech GmbH, Germany) at 500 g and 20 °C to separate the supernatant from the pellet. The supernatants were collected in pre-sterilized glass bottles.

High temperature short time (HTST) pasteurization

2.10

The soluble fractions were individually pasteurized using a miniature-scale high-temperature short-time/ultra-high-temperature (HTST/UHT) processing system (FT74XTS) including a feed pump, a pressurized water circulator and a plate heat exchanger coupled with a sterile filling unit (FT83XA) including a vertical laminar air flow cabinet with HEPA filter and aseptic filler (Armfield Limited, UK). Before the pasteurization, entire system was subjected to three-step Cleaning-In-Place (CIP) process flowed by Sterilizing-In-Place (SIP) with saturated steam at 121 ^0^C for 20 min and the sterile filling unit cabinet was disinfected using a UVC lamp.

Pasteurization was programmed for 95 ^0^C, 10 s in the holding tube with the help of stainless steel 316 concentric tubes split into product pre-heat, main heat, holding tube and cooling tubes. Pasteurized soluble contents were aseptically filled into pre-sterilized glass bottles at around 15 ^0^C and stored at 4 ^0^C. Samples were collected after pasteurization to evaluate the effectiveness of the thermal treatment.

Spray drying

2.11

The pasteurized soluble fractions were spray-dried using a mini spray dryer (Büchi B-290, Büchi Labortechnik, UK) to produce dry powders of the food ingredients, after which the weight of each powder was measured to determine the yield percentage. Powders were vacuum-packed in aluminium foil bags and stored at 4 °C and four different food ingredients were U-FI, E-FI, A-FI, and R-FI. Yield percentage was calculated on a dry weight basis according to the following equation:

$$[eqn]$$

Confocal laser scanning microscopy (CLSM) analysis of microstructure during ultrasonication and of the food ingredients

2.12

The CLSM observation of VCOPC, ultrasonicated samples collected at different time points, and the four food ingredients was conducted following the method described by Zahir et al., [12]. Briefly, cell walls, protein bodies, and oil bodies were stained with calcofluor white, rhodamine B, and BODIPY 505/515, respectively. The dye solutions were prepared by dilution in water to final concentrations of 0.002% (w/v) for calcofluor white and 0.001% (w/v) for both rhodamine B and BODIPY 505/515. Each sample was mixed with the dye mixture at a 1:1 (v/v) ratio, and 30 μL of the homogenized, stained suspension was placed on a glass slide. Visualization was performed using a CLSM Type 510 (Zeiss, Oberkochen, Germany) equipped with a 405 nm blue/violet diode laser for calcofluor white, a 543 nm HeNe laser for rhodamine B, and a 488 nm argon laser for BODIPY. Images were acquired using a 10×/20 × EC Plan-Neofluar/0.5 A objective lens and analyzed with ZEN Blue Edition software (Carl Zeiss Microscopy).

Changes of particle size and ζ-potential during sonication and of food ingredients

2.13

Particle size and ζ-potential (surface charge) of samples collected during sonication and of the food ingredients (FIs) were determined using a Zetasizer Ultra (Malvern Instruments Ltd., Worcestershire, UK), following the method described by Rodriguez-Loya et al. [13]. Measurements were conducted in disposable quartz cuvettes (Hellma, Müllheim, Germany). The FIs were hydrated to a concentration of 0.1% (w/w), and all samples, including those collected at different sonication time points, were filtered through 0.22 µm PTFE syringe filters (PerkinElmer, Shelton, CT, USA).

Particle size was measured by dynamic light scattering using a refractive index of 1.5 and an absorption coefficient of 0.001. All samples were equilibrated at 25 °C for 2 min and analyzed using backscattering detection at a scattering angle of 173°. Measurements were performed in triplicate.

Proximate characterization of food ingredients

2.14

The following analyses were performed for each food ingredient: moisture content was determined by drying 2.0 g of sample at 105 °C to constant mass and calculating mass loss (AOAC 950.46); total protein content was measured by the Dumas combustion method (AOAC 968.06) using approximately 15 mg of sample, with nitrogen converted to protein using a factor of 6.25; total fat content was determined by hot solvent extraction using an automated Soxtherm system (SOX 406; C. Gerhardt GmbH & Co. KG, Königswinter, Germany) according to the manufacturer’s instructions (Supplementary Materials).

Soluble protein content by BCA assay

2.15

Soluble protein content was determined by dispersing the samples in Milli-Q water at a 1:10 (w/w) ratio, followed by quantification using the Micro BCA™ Protein Assay Kit (Thermo Fisher Scientific, USA) according to the manufacturer’s instructions. An external bovine serum albumin (BSA) calibration curve ranging from 25 to 1000 µg/mL with good linearity (R2 > 0.99) was used for quantification.

Determination of free sugar concentration

2.16

Samples were prepared by mixing with warm Milli-Q water (50 °C) at a 1:10 (w/w) ratio, incubated at 50 °C for 30 min, and centrifuged at 2500 rpm for 10 min at room temperature. The supernatant was then filtered through a 0.2 µm CA filter, further diluted 50-fold, and transferred into HPLC vials for analysis. The concentrations of fructose, glucose, and sucrose were determined using an Acquity UPLC-H Class Plus System (Waters, Milford, MA, USA) equipped with an Acquity Evaporative Light Scattering Detector (ELSD). The UPLC–ELSD method is described in the Supplementary Materials.

Free amino-acids determination

2.17

Free amino groups in VCOPC and the four food ingredients were determined using the o-phthaldialdehyde (OPA) method. The assay was performed in microtiter plates by mixing 10 µL of sample with 200 µL of freshly prepared OPA reagent. After incubating at room temperature for 15 min, absorbance was measured at 340 nm using a microplate spectrophotometer (Benchmark Plus, Bio-Rad, UK). Free amino group concentration was calculated using an external L-serine calibration curve (0–1.5 mM; R^2^ > 0.99) and expressed as mmol L-serine equivalents per gram of sample.

Solubility

2.18

The solubility of the four food ingredients (FIs) was determined after pH adjustment. Each FI was dispersed in demineralised water at a concentration of 2% (w/w), and the pH was adjusted to 3, 5, or 7 using 1 M HCl or 1 M NaOH under continuous stirring at 100 rpm for 1 h. The pH was monitored every 15 min and readjusted as necessary to maintain the target value.

After pH stabilization, 500 µL of each suspension was weighed into pre-weighed aluminium trays to determine total dry matter content. In parallel, 2 mL of each suspension was transferred to Eppendorf tubes and centrifuged at 20,000 × g for 30 min at 4 °C. A 500 µL aliquot of the resulting supernatant was then transferred into pre-weighed aluminium trays, and the initial weights were recorded. All trays containing both total and supernatant samples were dried in an oven at 105 °C overnight, cooled in a desiccator for at least 20 min, and weighed to constant mass. Solubility (%) was calculated according to the following equation:

$$[eqn]$$

Foaming properties

2.19

The foaming properties of the four food ingredients were determined using a Foamscan™ instrument (Teclis, France). Each food ingredient was prepared at a concentration of 2% (w/w) in demineralised water adjusted to pH 7. A 60 mL aliquot of the sample solution was transferred into a glass cylinder, and nitrogen gas was injected at a flow rate of 400 mL/min for 30 s to generate foam. Foam formation and stability were monitored using an integrated camera system until the foam volume decreased to half of its maximum value at 20 ^0^C. The time required to reach this point was recorded as the foam half-life. Foam capacity was calculated by the Foamscan™ software as the ratio of total foam volume to gas volume after foam formation and expressed as a percentage.

Volatile compound analysis (VOC)

2.20

Two grams of each FI were placed into 5 mL headspace vials and sealed with screw caps. Automated headspace solid-phase microextraction (HS-SPME) was used for VOC sampling. Samples were equilibrated at 50 °C in the autosampler incubation chamber (CTC Analytics, Zwingen, Switzerland) and exposed for 45 min to a 50/30 µm DVB/CAR/PDMS fiber (Supelco, PA, USA). After extraction, VOCs were thermally desorbed at 250 °C in splitless mode and analyzed using a Clarus 500 GC–MS system (PerkinElmer, Waltham, MA, USA) equipped with an HP-Innowax column (30 m × 0.32 mm × 0.5 µm; Agilent Technologies, USA). Helium was used as carrier gas at a flow rate of 2 mL/min. The oven temperature program was set from 40 °C (held for 1 min) to 250 °C at 5 °C/min and held for 2 min. The MS operated in EI mode (70 eV) over an m/z range of 35–350. Compounds were identified by comparison with the NIST 14 library and verified using Kovats retention indices.

Statistical analysis

2.21

All experiments were conducted in triplicate, and the results are presented as mean ± standard deviation. Statistical analysis was performed using RStudio 4.2.1 (Boston, Massachusetts, USA), employing two-way ANOVA for each sample combination. Post-hoc Tukey's tests were used to identify significant differences (p < 0.05) between samples.

Results and Discussion

3

Effect of ultrasonication conditions on soluble fraction

3.1

As shown in Table S1 (Supplementary Material), the response surface methodology (RSM) design included factorial, axial, and center points to evaluate the combined effects of sonication pressure (0.5–3.4 bar) and residence time (18.9–231 min) on the dry matter content in the soluble fraction (SF) of VCOPC. The dry matter content in the supernatant, representing the SF, was significantly influenced by both sonication pressure and time, while temperature (25 °C) and amplitude (100%) were maintained constant throughout the experiments. The SF yield ranged from 31% to 69% under the tested conditions. The maximum yield (69%) was achieved at 3.41 bar and 125 min, whereas the minimum yield (31%) was observed at 2 bar and 18.9 min, indicating that increasing both sonication pressure and time markedly enhanced SF recovery.

A second-order polynomial model was established to describe the relationship between sonication pressure (A) and time (B) on SF yield. ANOVA results (Table S2) confirmed the model’s strong statistical significance (F = 25.10, p = 0.0002), with both linear (A, B) and quadratic (A2, B2) terms being highly significant (p < 0.01), while the interaction term (AB) was not significant (p = 0.6611). The model demonstrated excellent fit with R2 = 0.9472, Adj. R2 = 0.9094, and a Pred. R2 = 0.7295, with the difference between adjusted and predicted values below 0.2, confirming strong predictive capability. A high adequate precision value (14.30) indicated a robust signal-to-noise ratio. Importantly, the lack of fitness was not significant (F = 2.38, p = 0.2103), further validating the model’s suitability.

The final quadratic model describing SF yield (%) as a function of pressure (A) and time (B) was:

$$[eqn]$$

where Y is the SF yield (%), A is the sonication pressure (bar), and B is the sonication time (min). Among the linear terms, both pressure (F = 42.41, p = 0.0002) and time (F = 22.80, p = 0.0003) exerted significant effects on SF yield. Significant quadratic effects for pressure (F = 14.80, p = 0.0063) and time (F = 19.80, p = 0.0030) revealed curvature in the response surface, indicating the presence of an optimal operating window.

Among the model terms, sonication pressure and time exhibited the strongest effects on SF yield, while the significant quadratic terms indicated the presence of curvature in the response surface and the existence of an optimal operating window. The 2D contour and 3D response surface plots (Fig. 2) further demonstrated that SF yield increased with higher pressure and longer sonication time up to a threshold, beyond which overprocessing resulted in a slight decline likely due to protein aggregation or reduced mass transfer.Fig. 22D contour (I) and 3D response surface (II) plots showing the combined effects of ultrasonication pressure (bar) and ultrasonication time (min) on the soluble fraction (SF) yield (%) of VCOPC during continuous ultrasonication.

Optimization and validation of the ultrasonication condition

3.2

To validate the predictive accuracy of the RSM model, experiments were conducted under the optimized conditions of 1.62 bar pressure and 95.45 min sonication time. The model predicted an SF yield of 54%, whereas the experimental value was 63%, falling within the 95% prediction interval (42–65%), confirming that the observed difference is statistically acceptable. This agreement between predicted and observed yields confirms the reliability and robustness of the model for practical applications. In summary, sonication pressure and sonication time are critical determinants of SF yield from VCOPC. The developed RSM-based quadratic model is statistically robust, predictive within the studied range, and the identified conditions represent the optimal operating point under the experimental constraints evaluated in this study.

Changes in microstructure during ultrasonication

3.3

The changes in microstructure over different time points of continuous ultrasonication with respect to the cell wall, protein, and lipids are shown in Fig. 3. Confocal microscopy demonstrated a clear time-dependent breakdown of the polysaccharide–protein–lipid matrix during ultrasonication. At 0 min, a dense and continuous fiber network encapsulated aggregated proteins and large lipid droplets, indicating a highly insoluble structure. Progressive ultrasonication from 16 min to 48 min led to gradual fiber fragmentation, reduced protein aggregation, and dispersion of lipid droplets, consistent with the onset of cavitation-induced disruption. By 63 min, fibers were largely disintegrated into short, dispersed fragments, proteins became diffusely distributed, and fat droplets were finely emulsified, reflecting solubilization of insoluble polysaccharides and improved interfacial organization. At 79–95 min, the matrix was fully deconstructed into a homogeneous, finely dispersed system where solubilized polysaccharide fragments, unfolded proteins, and nano-sized fat droplets formed a stable, integrated network.Fig. 3. Microstructural alterations of the virgin coconut oil press cake (VCOPC) matrix during ultrasonication and structural organization of four food ingredients (FIs). Confocal micrographs show the distribution of cell wall polysaccharides (Calcofluor White, cyan), proteins (Rhodamine B, red), and lipids (BODIPY 505/515, green). Left: Progressive disintegration of the VCOPC matrix at different ultrasonication times (0–95 min). Right: Microstructures of four food ingredients; U-FI (ultrasonicated), E-FI (ultrasonicated and enzyme-treated), A-FI (ultrasonicated and fermented with L. acidophilus), and R-FI (ultrasonicated and fermented with L. rhamnosus).

The gradual breakdown of the polysaccharide–protein–lipid matrix during ultrasonication aligns with reported mechanisms in which cavitation-induced microbubbles collapse violently, generating shear and shock waves that cleave glycosidic bonds, depolymerize polysaccharide chains, and reduce molecular weight and particle size. The progressive fragmentation of the fiber network in this study from intact, continuous bundles at 0 min to short, dispersed fragments at 63–95 min aligns with these findings and reflects efficient disruption of the carbohydrate matrix [14]. Simultaneously, protein aggregates became more diffuse and uniformly distributed, supporting previous evidence that ultrasonication can unfold protein structures, increase their surface activity, and promote interaction with solubilized polysaccharides. The reduction in fat droplet size observed at later stages corresponds with studies showing that ultrasound enhances emulsification by improving interfacial stabilization through protein–polysaccharide complexes. This transformation aligns with current research showing ultrasound-assisted modification enhances solubility, dispersion, and functionality of plant-based food matrices [[15], [16], [17], [18]].

Changes in particle size, zeta potential, and pH during ultrasonication

3.4

Table 1 summarizes the changes in particle size and zeta potential during ultrasonication. The particle size of the supernatants at different time points during ultrasonication decreased from 2497 nm (0 min) to 351 nm (95 min) during sonication, while the zeta potential (−10.1 ± 0.7 mV) and pH (6.1 ± 0.1) remained stable throughout the process. The reduction in particle size observed during sonication is primarily driven by acoustic cavitation, a phenomenon in which ultrasound waves passing through a liquid medium generate microscopic bubbles that grow and collapse violently within microseconds. This collapse releases intense localized energy, producing high shear forces, turbulence, microstreaming, and shock waves that physically disrupt large particles and aggregates, breaking them into smaller, more uniformly dispersed particles. In addition to particle size reduction, these mechanical forces are powerful enough to rupture cellular structures, releasing intracellular components and enhancing the accessibility of bioactive compounds. In food systems, this dual action of particle breakdown and cell disruption significantly increases surface area, enhances extraction efficiency, and contributes to better dispersion and functionality of the final product [19]. Although particle size reduction can also be achieved through conventional mechanical approaches, ultrasonication was applied here as a cavitation-based processing strategy, with its effects discussed in the context of matrix disruption and solubilization [20]. Ultrasonic treatment did not significantly change the zeta potential, indicating that the surface charge and electrostatic repulsion remained largely unchanged throughout the process. The stable pH likely contributed to this, as pH strongly influences surface charge. The relatively low zeta potential (−10.1 ± 0.7 mV) below the ± 25 mV stability threshold, suggests that the system was electrostatically unstable and prone to aggregation. Therefore, the improved stability observed after ultrasonication is likely due to physical effects, such as particle size reduction and better dispersion, rather than changes in surface charge [21].Table 1. Particle size and zeta potential of samples measured at different ultrasonication times. Values are presented as mean ± standard deviation (n = 3). Particle size decreased significantly (p < 0.05) with increasing ultrasonication time. Zeta potential values showed no significant differences among treatments.TimeSize nmZeta mV0 min2497.33 ± 427.50−11.82 ± 3.1916 min2019.00 ± 251.73−8.80 ± 0.3932 min1877.50 ± 231.22−11.35 ± 1.5648 min1406.50 ± 79.90−9.96 ± 0.3663 min1165.00 ± 43.84−8.99 ± 0.9979 min681.95 ± 9.83−10.63 ± 0.1395 min351.20 ± 26.73−9.80 ± 1.04

Particle size and zeta potential of food ingredients

3.5

As shown in Table 2, the particle size and zeta potential of VCOPC and its bioprocessed food ingredients (FIs) varied depending on the applied treatment. As described in Section 3.4, ultrasonication markedly reduced the particle size of the supernatant to 351.20 ± 26.73 nm, while maintaining a relatively stable zeta potential of −9.80 ± 1.04 mV after 95 min of treatment. Following spray drying, the particle size of the sonicated sample (U-FI) increased to 567.13 ± 72.36 nm with a zeta potential of −8.388 ± 1.294 mV. In contrast, further processing steps led to a significant reduction in particle size and changes in surface charge, with the enzyme-treated sample (E-FI) showing a particle size of 230.45 ± 1.77 nm and a zeta potential of −3.9915 ± 0.921 mV, while the fermented samples (A-FI and R-FI) showed particle sizes of 197.60 ± 1.83 nm and 204.25 ± 1.77 nm, with corresponding zeta potentials of −2.0315 ± 0.370 mV and −4.2315 ± 0.503 mV, respectively.Table 2. Particle size (nm), polydispersity index and zeta potential (mV) of VCOPC and four food ingredients.SampleSize nmPolydispersity indexZeta mVVCOPC620.3 ± 100.10^a^0.74 ± 0.25^a^−8.682 ± 0.682^a^U-FI567.13 ± 72.36^a^0.51 ± 0.03^b^−8.388 ± 1.294^a^E-FI230.45 ± 1.77^b^0.36 ± 0.002^c^−3.9915 ± 0.921^b^R-FI204.25 ± 1.77^b^0.29 ± 0.0^d^−4.2315 ± 0.503^b^A-FI197.60 ± 1.83^b^0.38 ± 0.06^c^−2.0315 ± 0.370^c^Statistical differences were analyzed using ANOVA and Tukey's test. Results are expressed as mean ± standard deviation. Different letters above the values indicate statistically significant differences (p < 0.05.

After spray drying, the particle size of the sonicated sample (U-FI) increased to 567.13 ± 72.36 nm, suggesting partial re-agglomeration during the drying and reconstitution processes. Similar observations have been reported, as particle size reduction achieved by ultrasound in liquid systems can be partially reversed after spray drying due to dehydration-induced aggregation and the loss of stabilizing hydration layers around particles. Moreover, the removal of the continuous aqueous medium reduces electrostatic repulsion between particles, facilitating agglomeration upon rehydration [[22], [23]]. The reduced particle size of the enzyme treated and fermented samples can be attributed to the extensive breakdown of the cellular matrix and complex macromolecular structures during these bioprocessing steps. Enzymatic hydrolysis has been shown to cleave proteins and polysaccharides into smaller fragments, improving matrix disintegration and limiting aggregation during subsequent processing[[24], [25]]. Similarly, microbial fermentation facilitates further structural degradation through the action of microbial enzymes and metabolic activity, which alters the composition and organization of the matrix and enhances dispersion stability. These combined effects enable the particles to retain their reduced size even after spray drying, despite the process’s potential to induce aggregation [[26], [27]].

Importantly, interpretation of the mean particle size must be considered together with the polydispersity index (PDI), which reflects the breadth of the particle size distribution. VCOPC and U-FI exhibited high PDI values (>0.5), indicating highly heterogeneous and aggregation-dominated dispersions in which the mean particle size represents an effective hydrodynamic diameter rather than a uniform particle population. This high polydispersity explains the large variability associated with intensity-weighted size measurements and confirms that the observed size increase after spray drying is primarily driven by aggregation phenomena. In contrast, the enzyme-treated and fermented samples showed markedly lower PDI values (≤0.38), indicating narrower particle size distributions and improved dispersion homogeneity. The reduced PDI in these samples demonstrates that enzymatic hydrolysis and microbial fermentation not only decreased the mean particle size but also limited particle aggregation, allowing the reduced size to be retained after spray drying despite its known aggregation-inducing effects [28].

Bacterial growth during fermentation

3.6

The viable cell counts of L. acidophilus and L. rhamnosus were quantified before and after 24 h of fermentation. The initial load of L. acidophilus was 4.0 x 10^6^ CFU/mL, which increased to 2.7 x 10^9^ CFU/mL after fermentation, corresponding to an approximate 2.8 log unit increase. Similarly, L. rhamnosus exhibited an initial load of 5.0 x 10^6^ CFU/mL and reached 4.6 x 10^9^ CFU/mL following fermentation, representing a 3.0 log unit increase. These findings demonstrate substantial bacterial proliferation and confirm active growth and metabolic adaptation of both strains during the fermentation period. Complete LAB inactivation was achieved by combining HTST pasteurization and spray drying.

Effect of bioprocessing on physicochemical characteristics

3.7

The physicochemical characteristics of VCOPC and the four food ingredients derived from it are presented in Table 3. The dry matter content was highest in ultrasonicated FI (96.50%), followed by ultrasonicated and fermented with L. rhamnosus FI (93.23%) and ultrasonicated and fermented with L. acidophilus FI (82.59%), while ultrasonicated enzyme-treated FI (81.96%) showed the lowest value. Total fat content ranged from 8.09% in ultrasonicated and enzyme-treated FI, while ultrasonicated and fermented with L. acidophilus FI (13.59%), ultrasonicated FI, and ultrasonicated and fermented with L. rhamnosus FI (≈13%) showed similar values. Total protein content increased compared to VCOPC (20.54%), with the highest values observed in ultrasonicated and fermented with L. acidophilus FI (28.21%) and ultrasonicated and enzyme-treated FI (27.37%), followed by ultrasonicated and fermented with L. rhamnosus FI (24.13%) and ultrasonicated FI (23.17%). Soluble protein content was highest in ultrasonicated and enzyme-treated FI (137.51 mg/g), followed by ultrasonicated FI (39.38 mg/g), ultrasonicated and fermented with L. acidophilus FI (28.85 mg/g), and ultrasonicated and fermented with L. rhamnosus FI (17.26 mg/g). Free amino groups were highest in ultrasonicated and enzyme-treated FI (11.82 mmol/g) and ultrasonicated and fermented with L. rhamnosus FI (9.98 mmol/g), followed by ultrasonicated FI (5.50 mmol/g) and ultrasonicated and fermented with L. acidophilus FI (4.82 mmol/g). Fructose was detected only in ultrasonicated and enzyme-treated FI (3.82 mg/g) and ultrasonicated and fermented with L. acidophilus FI (0.01 mg/g), while it was not detected in VCOPC, ultrasonicated FI, or ultrasonicated and fermented with L. rhamnosus FI. Glucose content was highest in ultrasonicated and enzyme-treated FI (2.42 mg/g), followed by ultrasonicated FI (0.39 mg/g), with the lowest level observed in VCOPC (0.18 mg/g). Glucose was not detected in the fermented samples. Sucrose content was highest in ultrasonicated FI (5.46 mg/g), followed by VCOPC (4.49 mg/g) and ultrasonicated and fermented with L. rhamnosus FI (1.32 mg/g). Lower sucrose contents were observed in ultrasonicated and fermented with L. acidophilus FI (0.20 mg/g), with the lowest level detected in ultrasonicated and enzyme-treated FI (0.05 mg/g).Table 3. Proximate composition, protein characteristics, sugar content, and fatty acid profile of virgin coconut press cake (VCOPC) and food ingredients obtained after sonication (U-FI), enzymatic treatment (E-FI), and fermentation with L. acidophilus (A-FI) and L. rhamnosus (R-FI).ParameterVCOPCU-FIE-FIA-FIR-FIDry matter %94.85 ± 0.19^b^96.50 ± 0.02^a^81.96 ± 0.22^d^82.59 ± 0.29^d^93.23 ± 0.05^c^Total fat %14.87 ± 0.07^a^13.77 ± 0.10^c^8.09 ± 0.04^d^13.59 ± 0.15^c^14.39 ± 0.12^b^Total protein %20.54 ± 0.05^d^23.17 ± 0.17^c^27.37 ± 0.20^a^28.21 ± 0.64^a^24.13 ± 0.26^d^Soluble protein mg/g0.99 ± 0.07^d^39.38 ± 1.09^b^137.51 ± 2.42^a^28.85 ± 2.66^bc^17.26 ± 0.04^c^Free amino groups (mmol L-serine equivalents g^−1^ sample)9.50 ± 0.26^a^5.50 ± 0.17^b^11.82 ± 1.03^a^4.82 ± 0.02^b^9.98 ± 0.89^a^Ash %5.22 ± 0.02^c^7.85 ± 0.14^b^9.74 ± 0.02^a^8.04 ± 0.18^b^8.14 ± 0.14^b^Sugars mg/gFructose003.82 ± 0.03^a^0.01 ± 0.00^b^0Glucose0.18 ± 0.00^c^0.39 ± 0.02^b^2.42 ± 0.01^a^00Sucrose4.49 ± 0.08^b^5.46 ± 0.33^a^0.05 ± 0.05^d^0.20 ± 0.00^d^1.32 ± 0.07^c^Yield %NA65395356Statistical differences were analyzed using one-way ANOVA followed by Tukey’s post hoc test. Results are expressed as mean ± standard deviation (n = 3). Different superscript letters within the same row indicate statistically significant differences (p < 0.05). NA denotes not applicable.

The higher dry matter content in ultrasonicated FI is likely due to structural changes caused by sonication, which create microchannels and improve moisture removal during spray drying. The lower dry matter in ultrasonicated and enzyme-treated FI and ultrasonicated and fermented with L. acidophilus FI indicates that enzymatic hydrolysis and fermentation increased the water-binding capacity of the material by producing smaller peptides, soluble sugars, and other hydrophilic compounds that retain moisture. The intermediate dry matter in ultrasonicated and fermented with L. rhamnosus FI suggests that fermentation with L. rhamnosus also enhanced water retention, but to a lesser extent. Both L. acidophilus and L. rhamnosus can produce exopolysaccharides (EPS) that provide hydrophilic sites and form gel-like structures, thereby increasing water-holding capacity. However, differences in their metabolic activity may explain the variation, as L. acidophilus tends to break down the matrix more extensively through acidification, leading to greater moisture retention, while L. rhamnosus forms a moderate EPS network with less matrix disruption [[26], [29], [30], [31], [32], [33]]. The lowest total fat content observed in ultrasonicated and enzyme-treated FI can be attributed to the extensive proteolytic activity of Flavourzyme, a mixture of endo- and exopeptidases that hydrolyses proteins into smaller peptides and free amino acids. This enzymatic action disrupts the structural integrity of the protein–lipid matrix, leading to the release of lipids that were previously bound or trapped within the cellular structure. As a result, a portion of the released oil can be separated or lost during subsequent processing steps [34]. Sonication contributed to an increase in total protein content and solubility by disrupting the matrix structure and releasing bound proteins, thereby improving their extractability. Enzymatic hydrolysis with Flavourzyme further enhanced this effect, as extensive proteolysis converted complex proteins into smaller peptides and free amino acids, resulting in the highest levels of both soluble protein and free amino groups. The rise in free amino groups reflects a higher degree of protein hydrolysis, indicating effective cleavage of peptide bonds and the generation of low-molecular-weight nitrogenous compounds. In contrast, fermentation resulted in a moderate increase in total protein but lower levels of soluble protein and free amino group levels, likely due to protein aggregation or utilization of free amino acids by microbial cells for growth and metabolism [[27], [35], [36]].

As per the tested sugars, sonication did not influence sugar release. This aligns with reports that ultrasound primarily enhances solubilization and molecular-weight reduction of carbohydrate matrices, often releasing oligosaccharide-range material [36]. Enzymatic treatment resulted in a clear increase in glucose and fructose with a corresponding decrease in sucrose. This change may be due to the release of endogenous enzymes during sonication, which could promote sucrose hydrolysis, the direct breakdown of sucrose into monosaccharides under the processing conditions, or the release of glucose and fructose that were originally trapped within the cellular matrix [24].

Fermentation led to almost complete depletion of glucose and fructose, indicating that both strains actively used these sugars as primary carbon sources. Sucrose utilization differed between the strains, with L. acidophilus consuming sucrose more effectively. These strain-specific differences are consistent with literature showing variability in sucrose hydrolysis and sugar metabolism among lactic acid bacteria [[10], [25]].

Microstructure of spray-dried food ingredients

3.8

Confocal microscopy (Fig. 3, right panel) showed that all spray-dried food ingredients, derived from 95 min of continuous ultrasonication, exhibited distinct microstructures shaped by ultrasonic pretreatment, enzymatic hydrolysis, and microbial fermentation. The ultrasonicated FI (U-FI) displayed a dispersed yet cohesive matrix of short fiber fragments, unfolded proteins, and small lipid droplets, consistent with cavitation-induced disruption [37]. The enzyme-treated FI (E-FI) showed a smoother and more homogeneous structure, where Flavourzyme hydrolysis reduced protein aggregation and enhanced lipid dispersion without altering the polysaccharide framework [35]. The L. acidophilus fermented FI (A-FI) presented a moderately porous matrix with protein clusters and emulsified lipids surrounded by protein–polysaccharide layers, while L. rhamnosus fermented FI (R-FI) showed the most uniform and integrated structure with minimal fiber aggregation and nano-sized lipid droplets. Overall, the transition from U-FI to R-FI mirrors the structural evolution seen during ultrasonication, where sequential bioprocessing further enhanced protein hydrolysis, solubilization, and interfacial organization within the biopolymer network [[17], [18]].

Yield percentage

3.9

As shown in Table 3, the yield (%) of the spray-dried soluble fractions varied markedly among treatments, reflecting the efficiency of each bioprocessing step in solubilizing and transforming the components of virgin coconut press cake (VCOPC). The highest yield was obtained for the ultrasonicated fraction (U-FI, 65%), followed by the fermented samples ultrasonicated and fermented with L. rhamnosus FI (56%) and ultrasonicated and fermented with L. acidophilus FI (53%), while the ultrasonicated and enzyme-treated sample exhibited the lowest yield (38%).

The high yield observed for ultrasonicated FI demonstrates the strong effect of ultrasonication on disintegrating the polysaccharide–protein–lipid matrix, thereby facilitating the release of soluble components into the supernatant. As described in 3.1, 3.2, 3.3, 3.4, cavitation-induced mechanical disruption enhanced the solubilization of proteins, polysaccharides, and lipids, leading to improved extractability and dispersion. This effect resulted in a larger recovery of soluble solids after centrifugation and spray drying, confirming the efficiency of ultrasonication in converting insoluble VCOPC components into soluble matter.

In contrast, the lowest yield obtained for ultrasonicated and enzyme-treated FI indicates that enzymatic hydrolysis by Flavourzyme caused extensive cleavage of macromolecules into low-molecular-weight peptides, free amino acids, and soluble sugars, many of which remained in the aqueous phase or were lost during downstream processing such as centrifugation and drying. The breakdown of structural polymers into hydrophilic fragments increased water retention and decreased the dry matter recovery, consistent with the higher moisture content and lower total solids reported in Table 2. Although ultrasonicated and enzyme-treated FI exhibited the highest soluble protein and free amino group contents, much of this solubilized material remained bound to water or was not effectively recovered, leading to a significantly lower dry yield.

The intermediate yields of ultrasonicated and fermented with L. acidophilus FI and ultrasonicated and fermented with L. rhamnosus can be attributed to a balance between substrate degradation and microbial biomass formation during fermentation. Both L. acidophilus and L. rhamnosus actively metabolized carbohydrates and nitrogen sources, converting part of the soluble fraction into organic acids, CO_2_, and other metabolites, thereby reducing the recoverable solids. The viable cell counts increased by approximately 2.8 log units for L. acidophilus and 3.0 log units for L. rhamnosus (Section 3.6 Bacterial growth), indicating substantial microbial proliferation and metabolic activity. The slightly higher log increase and moderate proteolysis observed in L. rhamnosus corresponded to its higher yield (56%) compared to L. acidophilus (53%).

Overall, the observed yield trend reflects the combined effects of physical disruption, enzymatic hydrolysis, and microbial metabolism on solubilization and solid recovery. Ultrasonication primarily enhanced solubilization and recovery through cavitation-driven matrix breakdown, whereas enzymatic and microbial treatments induced deeper molecular transformations that improved biofunctional properties but reduced dry yield. This trade-off between molecular modification and mass recovery demonstrates that while enzyme and fermentation-assisted processes enhance the biochemical and functional potential of the ingredient through peptide release and compositional modification, they inherently lower yield due to extensive hydrolysis and nutrient utilization during bioprocessing.

Solubility

3.10

Table 4 shows the solubility of VCOPC and four different food ingredient (FIs), determined at pH 3.0, 5.0, and 7.0. Solubility increased with increasing pH for all samples. The ultrasonicated and enzyme-treated FI showed the highest solubility, reaching around 90% at pH 5.0, followed by slightly lower values at pH 3.0 and 7.0. In contrast, VCOPC exhibited the lowest solubility, remaining below 50% even at neutral pH. The ultrasonicated FI and the two fermented samples, with L. rhamnosus and L. acidophilus showed intermediate solubility, ranging between 40–70%, with the highest values observed at pH 7.0. Among the fermented FI, the FI fermented with L. rhamnoses demonstrated slightly higher solubility than L. acidophilus. Overall, the enzymatic treatment proved most effective in enhancing protein solubility across all pH levels.Table 4. Solubility (%) of VCOPC and its bioprocessed food ingredients at different pH values, and foaming properties (foam expansion and foam capacity) of VCOPC and four food ingredients at pH 7.0.Solubility %Foaming propertiespH 3.0pH 5.0pH 7.0ExpansionFoam capacityVCOPC31.75±5.49^b^32.80 ± 1.47^b^45.21 ± 4.48^c^10.45 ± 5.39^ab^0.14 ± 0.02^b^U-FI41.59±4.91^b^38.56 ± 1.35^b^60.00 ± 8.07^b^8.93 ± 0.78^ab^0.95 ± 0.25^a^E-FI82.16±6.68^a^89.71 ± 6.36^a^80.24± 3.43^a^5.90 ± 0.70^b^0.17 ± 0.09^b^A-FI33.65±4.61^b^46.35 ± 3.28^b^63.22 ± 5.00^b^10.47 ± 3.23^ab^0.74 ± 0.06^a^R-FI32.79 ± 4.72^b^52.49 ± 5.05^b^69.80 ± 5.79^b^14.30 ± 4.29^a^0.69± 0.06^a^Values are expressed as mean ± standard deviation (n = 3). Different lowercase superscript letters within the same column indicate significant differences among samples (p < 0.05). Solubility data were analyzed by two-way ANOVA (sample × pH) followed by Tukey’s HSD test, whereas foaming properties measured at pH 7.0 were analyzed by one-way ANOVA followed by Tukey’s HSD test.

Solubility increased with rising pH across all samples, a common feature of plant proteins. Near their isoelectric point (pH 4–5), proteins have minimal net charge, leading to aggregation through hydrophobic interactions. As pH increases toward neutrality or alkalinity, deprotonation of acidic residues (–COOH) raises the net negative charge, enhancing electrostatic repulsion and protein water interactions while reducing aggregation. Partial unfolding at higher pH also exposes hydrophilic groups, further improving solubility and dispersion [38]. Enzymatic hydrolysis produced the greatest solubility improvement because peptide bond cleavage generates low-molecular-weight peptides with increased surface charge and hydrophilicity, which enhance protein–water interactions and minimize aggregation. This finding aligns with reports showing that enzymatic hydrolysis markedly improves solubility and functional properties of plant and coconut proteins by exposing charged and polar residues Vogelsang et al. [39]. Acoustic cavitation disrupts non-covalent interactions, reduces particle size, and unfolds protein structures, thereby increasing solvent accessibility [[40], [41]].

Fermentation led to a moderate solubility increase, as seen in ultrasonicated and fermented FIs, which were higher than the VCOPC and ultrasonicated FI samples but lower than the enzymatically treated FI. This improvement can be attributed to microbial protease activity and acidification, which cause partial hydrolysis, deamidation, and slight shifts in the isoelectric point, thereby enhancing surface charge and protein–water interactions, particularly at higher pH. Since fermentation induces less specific and less extensive hydrolysis than enzymatic treatment, the overall effect is comparatively smaller. The strong pH dependence observed for U-FI reflects that ultrasound primarily induces physical modification through cavitation-driven disruption of non-covalent interactions, partial unfolding, and particle size reduction, increasing exposure of hydrophilic groups and enhancing solubility at neutral pH, whereas enzymatic hydrolysis generates low-molecular-weight peptides with increased surface charge and hydrophilicity, resulting in consistently high solubility across all pH values [42]. In addition, spray drying can be considered as a contributing factor to reduced solubility due to heat-induced aggregation. Similar behavior has been reported for fermented soy and pea proteins, where solubility increases relative to untreated and physically modified samples but remains below that of enzyme-hydrolyzed counterparts[[43], [44]].

Foaming properties

3.11

Table 4 summarizes the foaming properties of VCOPC and four different food ingredients at pH 7.0, including foam expansion and foam capacity. The foaming properties varied among the samples. VCOPC showed a foam expansion of 10.45% with a foam capacity of 0.04. The ultrasonicated FI exhibited a foam expansion of 8.93% and a much higher foam capacity of 0.95. Ultrasonicated and enzyme-treated FI, displayed lower foam expansion (5.90%) and a moderate foam capacity of 0.17. Ultrasonicated and fermented with L. acidophilus FI showed a foam expansion of 10.47% and a foam capacity of 0.74, while ultrasonicated and fermented with L. rhamnosus FI recorded the highest foam expansion of 14.30% and a foam capacity of 0.69. Overall, ultrasonicated and fermented with L. rhamnosus FI exhibited the greatest foam expansion, whereas ultrasonicated FI demonstrated the highest foam capacity among all samples.

The variations in foaming properties among samples can be attributed to structural and surface modifications induced by different treatments. The higher foam expansion observed in ultrasonicated and fermented with L. rhamnosus FI and ultrasonicated and fermented with L. acidophilus FI can be attributed to partial unfolding and improved flexibility of protein molecules, which facilitate film formation at the air–water interface. The superior foam capacity of ultrasonicated FI can be attributed to ultrasonication, which enhances protein dispersion, solubility, and interfacial activity, promoting efficient air incorporation. Conversely, the lower foaming ability of ultrasonicated and enzyme-treated FI may result from extensive hydrolysis, producing short peptides that are less effective in stabilizing foam. Similar findings have been reported for plant proteins such as soy and pea, where moderate unfolding or limited hydrolysis improves foaming, while excessive peptide breakdown reduces stability [[45], [46]].

Volatile organic compounds

3.12

The heatmap reported in Fig. 4 shows the relative abundance of volatile organic compounds (VOCs) in virgin coconut press cake (VCOPC) and the food ingredients, with color intensity indicating the level of each compound. The VCOPC sample mainly contained octanoic acid, benzyl alcohol, tetradecane, and 2H-pyran-2-one derivatives. The U-FI sample had a much simpler volatile profile, with benzyl alcohol, methoxyacetic acid ester, and dodecane as the main compounds. The E-FI sample showed new volatiles such as acetic acid, decane, and 4,6-O-ethylidene-α-D-glucose. Both A-FI and R-FI contained a wide range of volatiles, including formic acid, acetic acid, butanoic acid, decane, benzyl alcohol, and 2H-pyran-2-one derivatives. A-FI exhibited broader diversity, with additional nitrogen-containing and hydrocarbon compounds, while R-FI contained fewer types but higher levels of short-chain acids along with 4,6-O-ethylidene-α-D-glucose and 2,4,7,9-tetramethyl-5-decyn-4,7-diol.Fig. 4. Heatmap of volatile organic compounds (VOCs) identified in virgin coconut press cake (VCOPC) and food ingredients produced through ultrasonication (U-FI), enzymatic treatment (E-FI), and fermentation with L. acidophilus (A-FI) and L. rhamnosus (R-FI). Color intensity represents min–max normalized (0–1 scaled) relative abundance of each compound.

These results indicate that different processing methods significantly affected the volatile composition. The untreated VCOPC sample mainly consisted of native volatiles, while ultrasonication reduced the diversity, likely due to the loss or degradation of sensitive compounds [[47], [48]]. Enzymatic treatment led to the formation of new volatiles related to acid and sugar transformation, suggesting the breakdown of complex molecules. Fermentation produced the most diverse profile with various acids, alcohols, hydrocarbons, and nitrogen-containing compounds formed through microbial metabolism, reflecting active protein and lipid conversion. In the fermented samples, the volatile composition was dominated by fermentation-derived acids and metabolites formed through carbohydrate and lipid conversion, indicating more advanced transformation reactions and a metabolic focus on producing specific volatile groups [[45], [49], [50]].

Conclusion

4

In this study we developed a customized continuous pilot-scale bioprocessing approach integrating ultrasonication, enzymatic hydrolysis, and lactic acid bacteria fermentation. This bioprocessing strategy effectively transforms virgin coconut oil press cake (VCOPC) into a nutritionally enriched and functionally versatile food ingredient. Continuous ultrasonication disrupted the dense polysaccharide–protein–lipid matrix of VCOPC, achieving up to 63% solubilization under optimized conditions (1.62 bar, 95 min) and reducing particle size to the nanoscale, thereby improving the accessibility of matrix-entrapped nutrients. Enzyme-assisted hydrolysis using Flavourzyme further enhanced protein solubilization and peptide release, producing the highest levels of soluble protein (137 mg/g) and free amino groups (11.8 mmol/g). Subsequent fermentation with L. acidophilus and L. rhamnosus promoted microbial growth, matrix remodeling, and the formation of short-chain acids and volatile flavor compounds.

Functional evaluations revealed significant improvements in solubility (up to 90%), dispersion stability, and interfacial behavior, indicating enhanced applicability in food systems. Although enzymatic and microbial treatments reduced overall dry yield due to molecular degradation and moisture retention, they conferred superior nutritional and biochemical attributes compared to physical treatment alone.

Collectively, this research provides a scalable, sustainable processing framework that bridges laboratory innovation with industrial feasibility. The developed coconut-based food ingredient represents a high-value valorization route for coconut by-products, supporting circular food system design in tropical economies.

CRediT authorship contribution statement

Thisun Ranpatabendi: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Vishnu Priya Selvaraju: Writing – original draft, Investigation. Antonio Martins: Resources, Funding acquisition. Alberto Fiore: Supervision, Resources, Project administration, Funding acquisition, Conceptualization. Vincenzo Fogliano: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Conceptualization.

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.