Valorization of Edible Oil Industry By-Products Through Optimizing the Protein Recovery from Sunflower Press Cake via Different Novel Extraction Methods

Abstract

Sunflower press cake (SPC), a by-product of the edible oil industry, represents a promising source of plant-based protein. This study aimed to investigate and optimize protein recovery from SPC using conventional (CE) and advanced extraction techniques, including Ultrasound and Microwave-Assisted Extraction (UMAE), Pressurized Liquid Extraction (PLE) and Enzyme-Assisted Extraction (EAE). The protein content both in extracts and in the precipitated mass was measured through Lowry assay, while the amino acid profile of the extracted proteins under optimal conditions was analyzed via High-Performance Liquid Chromatography (HPLC). Extraction parameters were optimized using response surface methodology (RSM) for each method. Among the novel methods studied, UMAE and PLE demonstrated superior efficiency over CE, yielding higher protein recovery in significantly shorter extraction times. Optimal UMAE conditions (10 min, 0.03 g/mL, 450 W microwave power, and 500 W ultrasound power) yielded a precipitation yield (PY) of 21.2%, protein recovery in extract (PR_E) of 79.9%, and protein recovery in precipitated mass (PR_P) of 66.3%, with a protein content (PC_P) of 902.60 mg albumin eq./g. Similarly, optimal PLE conditions (6 min, 0.03 g/mL, and 50 °C) resulted in PY, PR_E, and PR_P of 17.7, 68.9, and 47.4%, respectively, with a PC_P of 932.45 mg albumin eq./g. EAE using Aspergillus saitoi protease was comparatively less effective. The amino acid profiling confirmed SPC as a valuable protein source, with glutamic acid, arginine, and aspartic acid being the most abundant. These results highlight the potential of UMAE and PLE as efficient strategies for valorizing edible oil industry by-products into high-quality protein ingredients for food and biotechnological applications.

1. Introduction

The efficient utilization of agricultural by-products is fundamental to achieving sustainable development, offering solutions to the pressing environmental, economic, and resource management challenges faced by modern agriculture. Within the edible oil industry, significant quantities of by-products are generated, with sunflower press cake (SPC)—the solid residue left after oil extraction—emerging as one of the most promising materials for valorization [1]. In Europe alone, SPC production reached 5.6 million tonnes in 2023, reflecting a 25% increase from 2020, driven by growing demand for sunflower oil [2]. This by-product is characterized by its rich nutritional composition, including protein (29.0–54.0%), fiber (4.3–45.0%), ash (5.9–8.8%), moisture (4.5–17.6%) and other nutrients, which vary depending on agricultural practices, seed genotype, and processing parameters [3]. To date, this by-product has been used as a high-quality livestock feed and organic fertilizer due to its essential amino acids and nutrient content [4]. However, beyond its conventional uses, SPC is being redefined as a significant raw material for innovative applications, particularly as a sustainable source of plant-based proteins.

The global demand for proteins is escalating due to population growth, increasing dietary awareness, and a shift toward sustainable and health-conscious food systems. While animal-derived proteins dominate human diets, their production is linked to significant environmental impacts, including greenhouse gas emissions, deforestation, and inefficient resource utilization. In contrast, plant-derived proteins offer a viable alternative with a lower environmental footprint, reduced microbial contamination risks, and greater alignment with global sustainability goals [5,6]. SPC, free of reported toxic or allergenic compounds and genetic modifications, is increasingly recognized as an untapped source of plant-based proteins, particularly in light of recommendations from global health authorities such as the World Health Organization (WHO) and World Cancer Research Fund (WCRF) advocating for plant-forward diets [7].

Recent advances in extraction technologies have opened new avenues for maximizing protein recovery from agricultural by-products such as SPC. To this end, innovative extraction methods, including Microwave-Assisted Extraction (MAE), Ultrasound-Assisted Extraction (UAE), Pressurized Liquid Extraction (PLE), and Enzymatic-Assisted Extraction (EAE), offer significant advantages for efficient protein recovery over conventional extraction techniques. These methods enhance efficiency by improving mass transfer, reducing processing times, and preserving protein quality under controlled conditions. Specifically, MAE uses microwave energy to rapidly heat the solvent and plant matrix, while UAE employs high-frequency sound waves to generate cavitation bubbles, leading to rapid cell wall disruption [8]. When used in combination (UMAE), these methods synergistically enhance protein release and reduce the need for harsh chemical conditions, making them attractive for industrial applications [9,10]. PLE, on the other hand, operates under elevated temperature and pressure to keep solvents in a liquid state above their boiling point, thereby enhancing solubility, reducing viscosity, and significantly improving extraction efficiency without prolonged exposure to high temperatures [11,12]. Finally, EAE utilizes specific enzymes to selectively break down cell wall components and complex macromolecules, facilitating the mild and efficient recovery of proteins while preserving their functional and nutritional properties [13]. The selection of these methods in the current study was guided by their growing relevance in green food processing and their demonstrated capacity to improve protein yield and recovery in oilseed by-products. Each technique represents a promising, scalable alternative to conventional methods, with distinct mechanisms that can target different structural and biochemical properties of SPC. By comparing these advanced methods under optimized conditions, this study aims to identify the most effective strategies for sustainable protein valorization, and contribute to the development of eco-efficient processing solutions for agro-industrial residues.

The effectiveness of these extraction methods in protein recovery has been demonstrated across various studies. For instance, Teixeira et al. (2024) [14] studied protein extraction from black beans using MAE, UAE, and PLE methods. The highest protein yield was achieved through PLE (23.9%), followed by UAE (17.9%) and MAE (16.0%). The protein contents of the extracts were 82.7%, 82.5%, and 61.3% for PLE, UAE, and MAE, respectively. Similarly, PLE was employed by Zhou et al. (2021) [15] to extract proteins from spirulina, achieving a maximum protein content of 47.9%, and by González-García et al. (2021) [16], who successfully extracted proteins from brewer’s spent grain. Additionally, Phongthai et al. (2016) [17] demonstrated efficient protein recovery from rice bran using MAE. The protein yield (4.37%) was 1.54 times higher than that of conventional extraction methods, with a protein content of 71.3% and recovery of 22.1%. Finally, Prandi et al. (2022) [18] carried out MAE and UAE experiments to recover proteins from coffee green beans, concluding that these techniques did not significantly impact the quality of the extracted proteins.

Despite the promising potential of SPC, research dedicated to optimizing protein extraction from this material remains limited. Previous studies have demonstrated the feasibility of protein recovery using conventional extraction methods under varying conditions. For instance, Kalpana et al. (2020) [19] reported protein yields ranging from 22.0% to 32.6%, with protein contents between 56.0% and 74.0%, under conditions including pH 9, 9% NaCl, and a 10% solid:liquid ratio. Similarly, Subaşı et al. (2020) [20] achieved 80% protein purity through alkaline extraction assisted by ultrasound probes, though the process lacked comprehensive optimization. However, existing studies often focus on isolated techniques, and fail to comprehensively optimize or compare innovative methods for extracting proteins from SPC.

This study seeks to bridge this research gap by systematically investigating and optimizing protein recovery from SPC using a range of conventional and advanced extraction techniques, including the synergistic application of UMAE, PLE, and EAE. The experiments were conducted under alkaline conditions, with protein precipitation performed at the isoelectric point (pI) determined under different acidic environments. Key parameters depending on the applied extraction technique (CE—solid:liquid ratio, extraction time; UMAE—solid:liquid ratio, MAE/UAE power; PLE—solid:liquid ratio, extraction time, temperature) were optimized using response surface methodology (RSM) to maximize protein yield and recovery. For EAE, the effect of enzyme activity was investigated. Furthermore, the protein content both in extracts and in the precipitated mass was measured through Lowry assay, while the amino acid profile of the extracted proteins under optimal conditions was analyzed via High-Performance Liquid Chromatography (HPLC). The specific objectives of this study were to (i) compare the efficiency of different extraction techniques for protein recovery from SPC, (ii) optimize critical process variables using RSM, (iii) evaluate the protein yield, recovery efficiency, protein content, and amino acid composition under optimal extraction conditions, and (iv) provide a comparative analysis of these techniques to support the development of sustainable, efficient valorization strategies for agro-industrial by-products.

2. Materials and Methods

2.1. Materials and Chemicals

SPC was provided by Agroinvest S. A. (Athens, Greece). According to the sunflower oil production process, sunflower seeds are mechanically pressed before the solvent extraction of sunflower oil with hexane is performed; the recovered SPC is then dried and finally ground before further use. Liquid hydrochloric acid (HCI) and liquid sodium hydroxide (NaOH), used for alkaline extraction, were sourced from Fischer Chemical (Leicestershire, UK). Protease from Aspergillus saitoi was sourced from Sigma-Aldrich (St. Louis, MO, USA), while Folin–Ciocalteu’s reagent was obtained from Carlo Erba (Emmendingen, Germany), and bovine serum albumin (BSA) was from Sigma-Aldrich (St. Louis, MO, USA). Sulfuric acid, potassium, and copper sulfate were supplied by Fischer Chemical (Leicestershire, UK). Distilled water was used in all experiments.

The Waters AccQ•Tag Ultra Derivatization Kit (Waters, Milford, MA, USA) was used for the derivatization of amino acids prior to analysis. The kit includes the AccQ•Tag Ultra Reagent Powder of 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC) derivatization reagent, AccQ•Tag Ultra Borate Buffer, and AccQ•Tag Ultra Reagent Diluent. Amino acid hydrolysis was performed using ultra-pure water (18.2 MΩ·cm, Milli-Q, Merck, Darmstadt, Germany), phenol crystals (≥99%, Sigma-Aldrich, St. Louis, MO, USA), and 6 M hydrochloric acid (ACS reagent grade, ≥37%, Sigma-Aldrich, St. Louis, MO, USA).

2.2. Extraction Methods for Protein Recovery from Sunflower Press Cake

2.2.1. Conventional Extraction

Protein extraction from SPC was carried out using a conventional alkaline extraction method. The SPC was mixed with distilled water at specific solid:liquid ratios and the pH was adjusted to 8 using 1 M NaOH solution. The mixture was stirred continuously at room temperature. To optimize the extraction parameters, a response surface methodology (RSM) with a full factorial design was used to evaluate the impacts of two independent variables—solid:liquid ratio (g/mL) and extraction time (min), with respective ranges of 0.03–0.10 g/mL and 30–120 min (

Table 1. Precipitation yield (PY, %), protein recovery in extract (PR_E, %), protein recovery in precipitated mass (PR_P, %) (dependent variables) responses of SPC using conventional extraction based on full factorial design.

	Variable Levels		Observed Values
Run	X1 (Solid:Liquid Ratio, g/mL)	X2 (Time, min)	PY (%)	PRE (%)	PRP (%)
1	0.10	30	11.6 ± 0.2 d	42.2 ± 0.5 g	23.9 ± 0.3 f
2	0.04	30	14.7 ± 0.3 c	59.1 ± 0.8 c	32.3 ± 0.5 c
3	0.03	30	15.2 ± 0.6 c	61.6 ± 0.9 b	33.7 ± 0.1 b,c
4	0.10	60	12.3 ± 0.1 d	49.7 ± 0.2 e	29.2 ± 0.2 d
5	0.04	60	15.2 ± 0.3 c	62.0 ± 1.1 b	37.1 ± 0.8 a
6	0.03	60	16.1 ± 0.2 a,b	64.1 ± 0.4 a	35.0 ± 0.7 b
7	0.10	120	12.0 ± 0.1 d	44.8 ± 0.2 f	26.9 ± 0.5 e
8	0.04	120	15.5 ± 0.4 b,c	55.2 ± 0.3 d	35.2 ± 0.4 b
9	0.03	120	16.5 ± 0.2 a	60.2 ± 1.2 b,c	37.4 ± 0.8 a

Statistical significance (p < 0.05) is indicated by differing letters.

). Following extraction, the mixture was centrifuged at 3500 rpm for 5 min, and the resulting supernatant was collected for protein precipitation.

2.2.2. Ultrasound and Microwave Synergy in Extraction Processes (UMAE)

Protein recovery from SPC was performed using a synergistic combination of ultrasound- and microwave-assisted extraction (UMAE) methods. The extraction was carried out using the XO-SM50 Ultrasonic Microwave Reaction System (Nanjing Xianou Instruments Manufacture Co., Ltd., Nanjing City, China). Initially, the effect of extraction time was examined under controlled conditions, with a solid:liquid ratio of 0.04 (g/mL), microwave power set to 200 W, and ultrasonic power set to 450 W. The SPC was mixed with distilled water, and the pH of the solution was adjusted to 8 using a 1 M NaOH solution. Extraction times of 5, 10 and 15 min were tested to determine the optimal duration. To further investigate the effects of extraction parameters on protein recovery, a Box–Behnken experimental design was employed. The independent variables—microwave power (W), ultrasound power (W) and solid:liquid ratio (g/mL)—were optimized within the ranges of 0–500 W, 0–700 W and 0.03–0.10 g/mL, respectively (

Table 2. PY (%), PR_E (%), and PR_P (%) (dependent variables) responses of SPC using UMAE based on Box–Behnken design.

	Variable Levels			Observed Values
Run	X1 (Solid:Liquid Ratio, g/mL)	X2 (Microwave Power, W)	X3 (Ultrasound Power, W)	PY (%)	PRE (%)	PRP (%)
1	0.10	0	450	11.1 ± 0.6 d	32.5 ± 2.8 h	24.8 ± 2.2 f
2	0.10	200	0	9.8 ± 1.1 d	29.5 ± 2.7 h	23.6 ± 2.3 f
3	0.10	200	700	12.0 ± 1.2 d	34.3 ± 2.6 g,h	21.5 ± 2.7 f
4	0.10	500	450	11.9 ± 0.5 d	44.1 ± 3.1 f,g	40.0 ± 2.9 d,e
5	0.04	0	0	3.9 ± 1.8 e	14.4 ± 0.7 i	5.1 ± 0.5 g
6	0.04	0	700	18.0 ± 1.7 b,c	57.5 ± 4.0 d,e	24.1 ± 2.0 f
7	0.04	200	450	18.7 ± 1.7 b	66.8 ± 4.1 b,c,d	53.8 ± 3.1 b
8	0.04	200	450	18.7 ± 1.7 b	66.8 ± 4.1 b,c,d	53.8 ± 3.1 b
9	0.04	200	450	18.7 ± 1.7 b	66.8 ± 4.1 b,c,d	53.8 ± 3.1 b
10	0.04	500	0	12.4 ± 1.4 d	44.5 ± 2.9 f,g	34.2 ± 2.3 e
11	0.04	500	700	19.7 ± 2.2 b	62.1 ± 4.0 c,d,e	51.4 ± 3.3 b,c
12	0.03	0	450	19.6 ± 2.1 b	70.0 ± 4.2 a,b,c	45.4 ± 3.2 c,d
13	0.03	200	0	13.8 ± 1.5 c,d	52.5 ± 3.4 e,f	34.8 ± 2.4 e
14	0.03	200	700	25.3 ± 0.9 a	77.2 ± 4.5 a,b	49.7 ± 2.8 b,c
15	0.03	500	450	21.2 ± 2.0 a,b	79.9 ± 4.3 a	66.3 ± 3.8 a

Statistical significance (p < 0.05) is indicated by differing letters.

). Following extraction, the mixture was centrifuged at 3500 rpm for 5 min and the resulting supernatant was collected for protein precipitation.

2.2.3. Pressurized Liquid Extraction (PLE)

PLE was employed for protein recovery from SPC, using the PLE^® Pressurized Liquid Extraction system (Fluid Management Systems, Watertown, MA, USA). The SPC was placed in a 100 mL stainless steel extraction cell, and distilled water at pH 8 was used as the solvent. The extraction pressure was maintained at 1750 psi. To optimize the extraction conditions, a Box–Behnken experimental design was used to evaluate the effect of solid:liquid ratio (g/mL), extraction temperature (°C), and extraction time (min), with optimization ranges of 0.03–0.10 g/mL, 50–150 °C and 3–10 min, respectively (

Table 3. PY (%), PR_E (%), and PR_P (%) (dependent variables) responses of SPC using PLE based on Box–Behnken design.

	Variable Levels			Observed Values
Run	X1 (Solid:Liquid Ratio, g/mL)	X2 (Temperature, °C)	X3 (Time, min)	PY (%)	PRE (%)	PRP (%)
1	0.10	100	3	10.2 ± 0.6 e,f	48.3 ± 0.5 h	25.9 ± 1.9 c,d
2	0.10	100	10	7.9 ± 0.8 f,g	43.5 ± 0.2 j	19.0 ± 1.6 d,e
3	0.10	50	6	10.5 ± 0.7 d,e,f	42.3 ± 0.6 j	27.0 ± 2.0 c
4	0.10	150	6	7.2 ± 0.75 e,f,g	68.2 ± 0.1 b,c	13.1 ± 1.4 e,f
5	0.04	50	3	13.1 ± 1.1 c,d	54.2 ± 0.1 f	40.7 ± 2.6 a,b
6	0.04	50	10	16.2 ± 1.4 a,b	60.3 ± 0.5 e	46.7 ± 2.9 a
7	0.04	100	6	14.5 ± 1.2 b,c	65.3 ± 0.2 d	39.3 ± 2.7 b
8	0.04	100	6	14.5 ± 1.2 b,c	65.3 ± 0.2 d	39.3 ± 2.7 b
9	0.04	100	6	14.5 ± 1.2 b,c	65.3 ± 0.2 d	39.3 ± 2.7 b
10	0.04	150	3	8.2 ± 0.5 f,g	46.6 ± 0.7 i	15.1 ± 1.3 e,f
11	0.04	150	10	6.2 ± 0.2 g	50.1 ± 0.3 g	11.2 ± 1.2 f
12	0.03	50	6	17.7 ± 0.9 a	68.9 ± 0.2 b	47.4 ± 3.0 a
13	0.03	100	3	16.5 ± 0.8 a,b	70.9 ± 0.5 a	47.5 ± 2.9 a
14	0.03	100	10	17.0 ± 1.0 a,b	67.4 ± 0.9 c	45.0 ± 3.1 a,b
15	0.03	150	6	11.1 ± 0.6 d,e	64.0 ± 0.5 d	24.5 ± 2.0 c,d

Statistical significance (p < 0.05) is indicated by differing letters.

). After extraction, the mixture was centrifuged at 3500 rpm for 5 min and the resulting supernatant was collected for protein precipitation.

2.2.4. Enzymatic-Assisted Extraction (EAE)

EAE was used for protein recovery from SPC. The raw material was mixed with distilled water at a solid:liquid ratio of 1:20, and the pH of the solution was adjusted to 7.5 using 1 M NaOH solution. The samples were incubated in a thermostatic bath at 37 °C. Various amounts of protease from Aspergillus saitoi (0.6 U/mg solid enzyme) were added to the samples—3, 12 and 24 U/g raw material—and the extraction process lasted for 240 min. The enzymatic reaction was terminated by cooling the samples to 4 °C for 15 min. The samples were then centrifuged at 3500 rpm for 5 min, and the supernatants were collected for protein precipitation.

2.3. Characterization of Extracts

2.3.1. Protein Precipitation and Precipitation Yield (PY, %)

The protein-rich fractions (extracts) obtained through the various extraction methods used were subjected to protein precipitation for the preparation of sunflower protein isolates (SPI) [19]. The pH of the supernatants collected from each method was adjusted to 4.40 (isoelectric point), as determined in 2.3.2 (pI determination), using a 1 M HCI solution. The mixtures were then stored at 4 °C for 24 hours to allow protein precipitation. Afterward, the precipitates were washed by centrifugation at 3500 rpm for 5 min, repeated twice, to remove excess acid. Finally, the protein precipitates were freeze-dried in a vacuum freeze-dryer (Biobase Biodustry, Co., Ltd., Jinan, Shandong, China) to obtain a powder product (SPI).

PY was calculated using the following equation

$$PY \left(\%\right)={\displaystyle \frac{M_1}{M_2}}·100,$$

(1)

where M₁ is the dry mass of the precipitated extract (g) and M₂ is the initial weight of SPC (g).

2.3.2. Isoelectric Point (pI) Determination

The pI of the protehins from SPC was determined by assessing the precipitation yield at different pH values, while keeping the solid:liquid ratio and extraction time constant. The SPC was mixed with distilled water at a solid:liquid ratio of 0.02 g/mL, and the pH of the solution was adjusted to 8 using a 1 M NaOH solution. Alkaline extraction was carried out by stirring the mixture for 60 min. Following extraction, the samples were centrifuged at 3500 rpm for 5 min. The supernatants were subjected to protein precipitation by adding 1 M HCl solution to adjust the pH to 3.80, 4.00, 4.20, 4.40, and 4.60. The subsequent steps followed the procedure outlined in 2.3.1. The pI was identified as the pH at which the highest precipitation yield was observed.

2.3.3. Protein Content

The protein contents of the liquid extracts and dried SPI samples obtained via the various extraction methods were quantified using the Lowry method [21]. For the dried SPI, the samples were first re-dissolved in distilled water to achieve a solid:liquid ratio of 0.4 mg/mL and the pH of the solution was then adjusted to 8 using a 1 M NaOH solution. Briefly, 0.1 mL of each sample was mixed with 1 mL of Lowry reagent and incubated in the dark for 15 min. After this incubation, 0.2 mL of 1 N Folin reagent was added, and the mixture was further incubated in the dark for 30 min. The absorbance was then measured at a wavelength of 750 nm using a Bel Photonics M51 UV-Vis Spectrometer (Bel Engineering s.r.l., Monza, Italy), and the protein concentration was calculated using a bovine serum albumin (BSA) calibration curve.

The protein content of raw material (SPC) was determined using the Kjeldahl method following the standard AOAC 984.13 method [22], which involves three steps: digestion, distillation, and titration. In the digestion step, 1 g of the sample was mixed with 20 mL of sulfuric acid, 10 g of potassium sulfate, 1 g of hydrated copper sulfate and boiling cores. The mixture was placed in a digestion system under vacuum conditions and heated to boiling to ensure complete dissolution and oxidation, converting nitrogen in the sample into ammonium bisulfate. At first, the heating was mild (150 °C for 45 min), but as digestion progressed, the temperature was increased to a maximum of 420 °C for 1 h. During distillation, 75 mL of sodium hydroxide (NaOH) and 75 mL of deionized water were added to the solution to convert ammonium ions into ammonia, which was then distilled and received in a 50 mL sample of a standard solution of 0.5 N sulfuric acid. Finally, the amount of ammonia was quantified by titration with a standard 0.5 N sodium hydroxide solution and a methyl-red methylene blue indicator. The protein content was calculated by multiplying the nitrogen content with the appropriate factor (5.30 for sunflower press cake).

The protein recovery of the liquid extract (PR_E, %) was calculated using the following equation:

$$P{R}_E \left(\%\right)={\displaystyle \frac{P{C}_E}{P{C}_{spc}}}·100,$$

(2)

Here, PC_E is the protein content in liquid extract (mg/g) and PC_SPC is the protein content of SPC (mg/g). The protein recovery in precipitated mass (PR_P, %) was calculated using the following equation:

$$P{R}_P\left(\%\right)={\displaystyle \frac{M_P·P{C}_P}{M_{SPC}·P{C}_{SPC}}}·100,$$

(3)

Here, M_P is the dry mass of the precipitated extract (g), M_SPC the initial weight of SPC (g), PC_P is the protein content in precipitated mass (mg/g), and PC_SPC the protein content of SPC (mg/g).

PR_E reflects the efficiency of the extraction process in transferring protein molecules from the SPC raw material into the solvent. In contrast, the PR_P represents the combined efficiency of the extraction and precipitation processes, as it quantifies the percentage of protein recovered in the precipitated mass relative to the total protein available in the SPC raw material.

2.3.4. Protein Hydrolysis and HPLC Amino Acid Profiling

The SPC protein was subjected to vacuum hydrolysis [23,24]. Specifically, 20 μg of protein from each optimally processed extract was transferred into 1 mL vacuum hydrolysis tubes (Thermo Fisher Scientific, Waltham, MA, USA) containing 200 μL of 6 M HCl (at constant boiling) and 1% (w/w) phenol crystals, which served as oxygen scavengers. The tubes were purged with nitrogen gas, and a vacuum was then applied using a three-way stopcock to transition from inert gas to vacuum. Hydrolysis was performed at 116 °C for 24 h using a Reacti-Therm Heating Module (Thermo Fisher Scientific). Following hydrolysis, excess acid was removed with a BÜCHI R-200 Rotavapor (BÜCHI Labortechnik, Flawil, Switzerland), and the residues were dissolved in 20 μL of 20 mM constant boiling HCl.

Amino acid was derivatized using the AccQ-Fluor reagent 6-Aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC; Waters, Milford, MA, USA) [25]. To derivatize, 20 μL of hydrolysate was combined with 20 μL of AQC reagent and 60 μL of borate buffer (pH 8.8), and the mixture was incubated at 55 °C for 10 min.

The derivatized amino acids were analyzed using an HPLC system with a Prominence-i LC-2030 3D gradient pump and diode array detector (Shimadzu, Kyoto, Japan). Separation was achieved on an AccQ-Tag C18 Amino Acid Column (60 Å, 4 μm, 3.9 mm × 150 mm; Waters, Milford, MA, USA) maintained at 37 °C [26]. The mobile phase included an aqueous acetate–phosphate buffer (solvent A), acetonitrile (solvent B), and water (solvent C). The gradient program started with 100% A and was adjusted as follows: 99% A/1% B/0% C for 0.5 min, 95% A/5% B/0% C for 18 min, 91% A/9% B/0% C for 19 min, 83% A/17% B/0% C for 29.5 min, and 0% A/60% B/40% C for 36 min. The system operated with a 5 μL injection volume and a 1 mL/min flow rate. Detection occurred at 254 nm. Data were acquired and processed using LabSolutions Workstation (Shimadzu). The identification of amino acids was accomplished by comparing UV-Vis spectra and retention times to those of a 17-amino acid standard derivatized with AccQ-Fluor (Waters, Milford, MA, USA).

2.4. Experimental Design and Statistical Analysis

Response Surface Methodology (RSM) was employed to analyze the experimental data and optimize each extraction process for protein recovery from SPC. All statistical analyses were performed using StatSoft STATISTICA 12.0 software (Hamburg, Germany).

For the conventional alkaline extraction, a full factorial design was utilized within the RSM framework. The independent variables considered in the analysis were the solid:liquid ratio (g/mL, X₁) and extraction time (min, X₂). In the case of the synergistic application of UMAE and PLE methods, a Box–Behnken design was applied within the RSM approach to determine the optimal conditions for the independent variables. For UMAE, the variables assessed included solid:liquid ratio (g/mL, X₁), microwave power (W, X₂) and ultrasound power (W, X₃). For PLE, the variables examined were solid:liquid ratio (g/mL, X₁), extraction temperature (°C, X₂) and extraction time (min, X₃). Each experiment was conducted in triplicate, with each variable coded at three levels: −1, 0, and 1. The dependent variables in the model consisted of PY (%), PR_E (%) and PR_P (%). Table 1 displays the responses obtained from 9 runs of the full experimental design for conventional extraction. Table 2 and Table 3 display the responses obtained from 15 runs of the full experimental design for UMAE and PLE method. The generalized mathematical model that relates the responses to the independent variables is as follows:

$$Y=b_0+{{\displaystyle \sum }}_{i=1}^4{b}_i{X}_i+{{\displaystyle \sum }}_{i=1}^4{b}_{ii}{X}_i^2+{{\displaystyle \sum }}_{i=1}^3{{\displaystyle \sum }}_{j=i+1}^4{b}_{ij}{X}_i{X}_j$$

(4)

Here, Y is the dependent variable (PY (%), PR_E (%) and PR_P (%)), b₀, b_i, b_ii and b_ij are regression coefficients of the intercept, linear, quadratic and interaction terms, respectively, and X_i and X_j are the independent variables.

An analysis of variance (ANOVA) was carried out to determine the statistical significance (p < 0.05) of the independent variables for precipitation yield (PY), protein recovery in extract (PR_E), and protein recovery in the precipitated mass (PR_P). Variables with the lowest p-values were considered the most influential (p < 0.05), while those with p-values above 0.05 were excluded from the final model—unless their quadratic or interaction terms were found to be statistically significant. This refined model, excluding non-significant linear effects, was used for interpreting the experimental outcomes. A 95% confidence level was set as the threshold for statistical significance. ANOVA was also employed to assess the relevance of each response variable, and a lack-of-fit test was used to evaluate the adequacy of the fitted models. The F-value provided insight into the model’s explanatory power, while the coefficient of determination (R²) measured its predictive accuracy. To illustrate the influences of key variables and their interactions, three-dimensional surface plots were generated, with a focus on the combinations that most strongly affected PY, PR_E, and PR_P. Additionally, one-way ANOVA was used to examine variations between treatment groups. Where applicable, Tukey’s post-hoc test (α = 0.05) identified significant differences among means. All statistical procedures were performed using STATISTICA software, version 12.0 (StatSoft, Hamburg, Germany).

3. Results and Discussion

3.1. SPC Protein Content and pI Determination

The protein content of the raw material of SPC was investigated through the Kjeldahl method, yielding a measurement of 289 ± 5 mg protein/g SPC. This is in agreement with what was found in the literature, which ranges from 22.3% to 63%, based on the oil extraction method [27,28]. Salgado et al. (2012) [29] found a protein content of 317 mg/g in the sunflower cake, which aligns with the findings of the present study. The pI of SPC was determined using conventional alkaline extraction with a solid:liquid ratio of 0.03 g/mL and an extraction time of 60 min, while varying the precipitation pH between 3.80 and 4.60. The corresponding PY values for each pH value are summarized in

Table 4. Determination of SPC pI.

Sample	Precipitation pH	PY (%)
1	3.80 ± 0.03	9.2 ± 0.1 e
2	4.00 ± 0.03	10.2 ± 0.1 d
3	4.20 ± 0.03	10.8 ± 0.1 c
4	4.40 ± 0.03	14.6 ± 0.2 a
5	4.60 ± 0.03	11.4 ± 0.1 b

Statistical significance (p < 0.05) is indicated by differing letters.

. The results reveal that the highest PY (14.6%) was achieved at a pH of 4.40, indicating it as the optimal pH for maximum precipitation. This finding is consistent with those from previous studies, such as that by Subasi et al. (2020) [20], which reported the pI of sunflower protein to be 4.40 ± 0.1.

3.2. Conventional Extraction of SPC

3.2.1. Optimization of Conventional Extraction Using RSM

The impact of the solid:liquid ratio (X₁) and extraction time (X₂) on PY, PR_E and PR_P values was investigated using a full factorial design across nine experimental runs. Table 1 summarizes the data collected under various extraction conditions, with solid:liquid ratios ranging from 0.03 to 0.10 g/mL, and extraction times from 30 to 120 min. The corresponding PY (%), PR_E (%), and PR_P (%) were systemically recorded for each run.

The results reveal that the PY ranged from 11.6% to 16.5%, with the highest values in runs 6 (0.03 g/mL, 60 min) and 9 (0.03 g/mL, 120 min), showing no significant statistical differences. PR_E values ranged from 42.2% to 64.1%, with the highest recovery recorded in run 6 (0.03 g/mL, 60 min). PR_P values varied from 23.9% to 37.4%, with the maximum value occurring in run 9 under the same conditions that yielded the highest PY.

Run 6 was identified as the optimal experiment, due to its efficiency in achieving the highest PR_E, which reflects superior protein extraction efficiency. Although run 9 exhibited slightly higher PY (16.5%) and PR_P (37.4%) than run 6 (PY = 16.1%, PR_P = 35.0%), the increase was minimal relative to the additional extraction time required (120 min vs. 60 min). Thus, run 6 represents the most favorable conditions, maximizing protein recovery in the extract while significantly reducing operational time.

3.2.2. Fitting Model to Data

The analysis of variance for the response surface of PY (%), PR_E (%), and PR_P (%) in the conventional extraction is presented in

Table A1. Analysis of variance for the response surface of PY (%), PR_E (%), and PR_P (%) using conventional extraction.

Source	Coefficients	Standard Error	Sum of Squares	DF	Mean Square	F-Value
PY (%) a
Model	16.78	0.85	5630.74	6	938.46	10,395.19 ***
X1	−6.19	1.79	70.94	1	70.94	785.82 ***
X12	3.57	0.78	0.34	1	0.34	3.81 NS
X2	1.30	0.66	2.20	1	2.20	24.37 ***
X22	−1.12	0.34	0.98	1	0.98	10.82 ***
X1X2	−0.53	0.20	0.65	1	0.65	7.23 *
Residual			1.90	21	0.09
Total			84.37	26
PRE (%) b
Model	70.84	3.01	86,087.79	6	14,347.97	12,756.70 ***
X1	−44.50	6.31	1351.56	1	1351.56	1201.66 ***
X12	30.95	2.76	20.30	1	20.30	18.05 ***
X2	9.95	2.34	0.38	1	0.38	0.33 NS
X22	−16.28	1.20	206.92	1	206.92	183.97 ***
X1X2	3.23	0.70	23.90	1	23.90	21.25 ***
Residual			23.62	21	1.12
Total			1807.57	26
PRP (%) c
Model	25.86	3.31	28,700.74	6	4783.46	3524.53 ***
X1	2.99	6.93	338.31	1	338.31	249.27 ***
X12	19.02	3.04	2.86	1	2.86	2.11 NS
X2	−3.73	2.57	39.28	1	39.28	28.95 ***
X22	−7.41	1.32	42.82	1	42.82	31.55 ***
X1X2	−0.68	0.77	1.05	1	1.05	0.77 NS
Residual			28.50	21	1.36
Total			531.49	26

* p < 0.05, *** p < 0.001, ^NS—not significant. ^a The coefficient of determination (R²) of model was 0.98. ^b The coefficient of determination (R²) of model was 0.98. ^c The coefficient of determination (R²) of model was 0.95.

.

• CE-Precipitation Yield (PY)

The fitted surface model for PY, based on the results presented in Table A1, is described by Equation (5):

$$PY \left(\%\right)=16.78-6.19X_1+3.57{X_1}^2+1.30X_2-1.12{X_2}^2-0.53X_1{X}_2$$

(5)

The model exhibited a strong fit to the experimental data with an R² value of 0.98 [30]. Analysis of variance (ANOVA) indicated that the linear term for solid:liquid ratio (X₁) had the most significant effect (F = 785.82, p < 0.001), followed by the linear term for extraction time (X₂) (F = 24.37, p < 0.001). Significant contributions were also observed from the quadratic term for extraction time (X₂²) and the interaction between solid:liquid ratio and extraction time (X₁X₂). In contrast, the quadratic term for solid:liquid ratio (X₁²) did not show a significant impact (p > 0.05). These results emphasize the critical roles of the solid:liquid ratio, extraction time, their interaction, and the quadratic effect of extraction time in influencing PY. Based on these findings, a refined model is expressed in Equation (6), excluding the non-significant terms:

$$PY \left(\%\right)=15.32-2.73X_1+3.57X_2-1.12{X_2}^2-0.53X_1{X}_2$$

(6)

This adjusted model remained statistically robust (F = 11,059.30, p < 0.001, R² = 0.97), with minimal deviations between predicted and actual values [31], as shown in Figure A1.

• CE-Protein Recovery in extract (PR_E)

The response surface model for PR_E showed an excellent fit with an R² value of 0.98, as shown in Table A1. The fitted model is expressed by Equation (7):

$$P{R}_E \left(\%\right)=70.84-44.50X_1+30.95{X_1}^2+9.95X_2-16.28{X_2}^2+3.23X_1{X}_2$$

(7)

The solid:liquid ratio (X₁) was the most influential factor (F = 1201.66, p < 0.001), followed by the quadratic term for extraction time (X₂²) (F = 183.97, p < 0.001). The interaction between solid:liquid ratio and extraction time (X₁X₂) and the quadratic term of solid:liquid ratio (X₁²) were also significant, though to a lesser extent. In contrast, the linear term for extraction time (X₂) was not significant, suggesting it has no linear effect on PR_E.

Based on these results, a refined regression model was developed by excluding the non-significant term as expressed by Equation (8):

$$P{R}_E \left(\%\right)=85.24-46.60X_1+9.95{X_1}^2-3.41{X_2}^2+5.22X_1{X}_2$$

(8)

This updated model remained statistically robust with an F-value of 2295.24, p < 0.001, and an R² of 0.91, indicating a strong predictive capability. As shown in Figure A2, the predicted versus observed values for PR_E align closely, confirming the model’s accuracy in predicting protein recovery efficiency under conventional extraction conditions.

• CE-Protein Recovery in Precipitated mass (PR_P).

The response surface model for PR_P also achieved a high R² value of 0.95 (Table A1). The fitted model is represented by Equation (9).

$$P{R}_P \left(\%\right)=25.86+2.99X_1+19.02{X_1}^2-3.73X_2-7.41{X_2}^2-0.68X_1{X}_2$$

(9)

The linear term for solid:liquid ratio (X₁) was the most influential factor (F = 249.27, p < 0.001), followed by the quadratic term for extraction time (X₂²) (F = 31.55, p < 0.001). In addition, the linear term for extraction time (X₂) demonstrated considerable importance. Conversely, non-significant terms included the quadratic term for solid:liquid ratio (X₁²) and the interaction between solid:liquid ratio and extraction time (X₁X₂). To improve model precision, a refined regression was performed by removing the non-significant terms. The updated model is shown in Equation (10).

$$P{R}_{\mathrm{P}} \left(\%\right)=30.76-7.70X_1+18.35X_2-7.41{X_2}^2$$

(10)

The updated model showed excellent statistical reliability with an F-value of 5091.43, p < 0.001, and an R² value of 0.94, indicating that it captured most of the variability in the data. As illustrated in Figure A3, the predicted values closely matched the observed values, confirming the model’s accuracy and consistency across the tested range.

In summary, the solid:liquid ratio (X₁) and extraction time (X₂) were the most influential factors affecting PY, as evidenced by their high F-values. These findings are consistent with those from previous studies, such as the research by Patra and Arun Prasath (2024) [32], which also identified these variables as crucial for optimizing precipitation yield in the conventional extraction of proteins from cassava (Manihot esculenta L.) leaves. Patra and Arun Prasath (2024) [32] demonstrated that reducing the solid:liquid ratio and extending the extraction time improved precipitation yields, aligning with this study’s results, wherein the solid:liquid ratio negatively affected yield while extraction time had a positive impact. The solid:liquid ratio was the factor with the most significant impact on the dependent variables (PY, PR_E, PR_P). These results are in agreement with the work of Firatligil-Durmus and Evranuz (2010) [33], who identified the solid:liquid ratio as a critical factor affecting protein extraction from red pepper seeds.

3.2.3. Graphical Interpretation of Model Predictions

The three-dimensional (3D) response surfaces and contour plots, shown in Figure 1, illustrate the relationship between the independent variables (solid:liquid ratio and extraction time) and the dependent variables (PY, PR_E, PR_P). These visual representations enable a clearer understanding of how changes in the independent variables influence the outcomes.

Figure 1a indicates that lower solid:liquid ratios and longer extraction times lead to higher PY values. The contour plot distinctly illustrates that a solid:liquid ratio of 0.03 g/mL combined with an extraction time exceeding 60 min yields the highest PY values. This observation aligns with the experimental findings, wherein run 9 achieved the highest PY (16.5%) under optimal conditions—a solid:liquid ratio of 0.03 g/mL and an extraction time of 120 min.

Figure 1b illustrates the response surface and contour plots regarding PR_E, and confirms that the highest recovery was achieved when the solid:liquid ratio was minimized and the extraction time was moderate, at around 50–90 min. This is consistent with the experimental results from run 6, wherein the highest PR_E (64.1%) was observed with a solid:liquid ratio of 0.03 g/mL and an extraction time of 60 min.

Similar studies using other plant materials have been reported. For, instance, Abas Wani et al. (2006) [34] worked on protein extraction from watermelon seed meal with an alkaline solution, highlighting the significant impact of the solid:liquid ratio on protein extraction. Their findings align with those of the current study, showing that reducing the solid:liquid ratio leads to higher protein yields. Research on protein extraction from SPC is relatively scarce. Kaur et al. (2024) [35] reported a higher protein recovery (75.32%) during conventional extraction from de-oiled sunflower meal under conditions of pH 11, a temperature of 50 °C, and a salt concentration of 0.2% for 60 min. The higher recovery observed in their study is likely due to the elevated pH and temperature, as well as the presence of salt, which can enhance protein solubility and extraction efficiency.

3.3. UMAE of SPC

3.3.1. Optimization of Extraction Time of UMAE

Initially, the extraction time of UMAE was examined for the protein extraction from SPC. Three different values of time (5, 10, and 15 min) were studied under controlled conditions and, specifically, a solid:liquid ratio of 0.04 (g/mL), a microwave power of 200 W, and an ultrasonic power of 450 W. According to the results presented in Figure 2, the optimum extraction time was 10 min, as the extract of that time showed significantly higher results for PY (27.33%) and PR_P (74.28%). In addition, the PR_E was similar to that of the 15 min extract, with no statistical difference.

3.3.2. Optimization of UMAE Using RSM

Further investigation of the effects of the other extraction parameters on protein recovery was employed, keeping the parameter of extraction time stable at 10 min. Ultrasound power ranged from 450 W to 700 W, and microwave power from 200 W to 500 W, with solid:liquid ratios between 0.03 and 0.10 g/mL. The corresponding PY (%), PR_E (%), and PR_P (%) were systemically recorded for each run.

The findings indicate that precipitation yield varied from 3.9 to 25.3%. The highest yield was achieved after applying 200 W microwave power and 700 W ultrasound power, combined with a ratio of 0.03 g/mL (run 14). PR_E and PR_P ranged from 14.4% to 79.9% and from 5.1% to 66.3%, respectively. The highest recoveries both in extract and precipitated mass were observed in run 15, which was conducted under 450 W microwave power, 500 W ultrasound power, and 0.03 g/mL ratio.

Run 15 was identified as the optimal experiment due to its maximum recoveries in both extract and precipitated mass. While the PY in run 15 (21.2%) was slightly lower than the highest value achieved in run 14 (25.3%), there was no statistical difference. Therefore, the optimal parameters for SPC protein extraction using the UMAE technique were determined to be 450 W ultrasound power, 500 W microwave power, and a solid:liquid ratio of 0.03 g/mL.

3.3.3. Fitting Model to Data

The analysis of variance for the response surface of PY (%), PR_E (%), and PR_P (%) in UMAE is presented in

Table A2. Analysis of variance for the response surface of PY (%), PR_E (%), and PR_P (%) for protein extracts derived through UMAE.

Source	Coefficients	Standard Error	Sum of Squares	DF	Mean Square	F-Value
PY (%) a
Model	21.96	3.31	12,289.09	10	1228.91	503.55 ***
X1	−30.60	5.92	382.19	1	382.19	156.60 ***
X12	8.32	1.19	64.16	1	64.16	26.29 ***
X2	15.30	1.43	50.86	1	50.86	20.84 ***
X22	11.15	2.17	56.55	1	56.55	23.17 ***
X3	−1.97	0.41	220.00	1	220.00	90.15 ***
X32	−2.46	0.67	32.87	1	32.87	13.47 ***
X1X2	−0.68	0.52	4.21	1	4.21	1.73 NS
X1X3	−3.94	0.64	91.54	1	91.54	37.51 ***
X2X3	−1.84	0.45	40.29	1	40.29	16.51 ***
Residual			85.42	35	2.44
Total			1344.43	44
PRE (%) b
Model	84.96	8.33	143,174.28	10	14,317.43	928.54 ***
X1	−115.50	14.89	6211.01	1	6211.01	402.81 ***
X12	27.53	3.00	785.75	1	785.75	50.96 ***
X2	59.79	3.60	1185.25	1	1185.25	76.87 ***
X22	39.02	5.47	645.58	1	645.58	41.87 ***
X3	−6.64	1.03	1360.33	1	1360.33	88.22 ***
X32	−16.79	1.69	1527.96	1	1527.96	99.09 ***
X1X2	−0.13	1.31	0.16	1	0.16	0.01 NS
X1X3	−10.59	1.62	661.29	1	661.29	42.89 ***
X2X3	−6.33	1.14	474.55	1	474.55	30.78 ***
Residual			539.67	35	15.42
Total			16,091.25	44
PRP (%) c
Model	39.10	6.61	79,286.11	10	7928.61	816.91 ***
X1	−55.15	11.81	2429.21	1	2429.21	250.29 ***
X12	30.66	2.38	201.00	1	201.00	20.71 ***
X2	57.82	2.86	2212.70	1	2212.70	227.98 ***
X22	19.73	4.34	774.70	1	774.70	79.82 ***
X3	−7.27	0.81	350.16	1	350.16	36.08 ***
X32	−23.55	1.34	3006.61	1	3006.61	309.78 ***
X1X2	−3.27	1.04	96.61	1	96.61	9.95 **
X1X3	−8.25	1.28	401.35	1	401.35	41.35 ***
X2X3	−0.70	0.90	5.73	1	5.73	0.59 NS
Residual			339.70	35	9.71
Total			11,815.62	44

** p < 0.01, *** p < 0.001, ^NS—not significant. ^a The coefficient of determination (R²) of model was 0.94. ^b The coefficient of determination (R²) of model was 0.97. ^c The coefficient of determination (R²) of model was 0.97.

.

• UMAE Precipitation Yield (PY)

The response surface model for the PY of SPC using UMAE demonstrated a strong fit to the experimental data, with an R² value of 0.94 [31]. The fitted model for PY is represented by the following equation:

$$P{\rm Y} \left(\%\right)=21.96-30.60X_1+8.32X_1^2+15.30X_2+11.15X_2^2-1.97X3-2.46X_3^2-0.68X_1{X}_2-3.94X_1{X}_3-1.84X_2{X}_3$$

(11)

Among the variables, the linear term of solid:liquid ratio (X₁) had the greatest effect (F = 382.19, p < 0.001), followed by the linear term of ultrasound power (X₃) (F = 220.00, p < 0.001). The quadratic term of solid:liquid ratio (X₁²), the linear and quadratic terms of microwave power (X₂ and X₂²), the quadratic term of ultrasound power (X₃²), the interaction between solid:liquid ratio and ultrasound power (X₁X₃), and the interaction between microwave power and ultrasound power (X₂X₃) were also significant. In contrast, the interaction between the solid:liquid ratio and microwave power (X₁X₂) did not show a significant impact (p > 0.05). These results demonstrate that solid:liquid ratio and ultrasound power are critical in influencing the protein yield of SPC. A refined model was developed by excluding the non-significant term, resulting in the updated Equation (12):

$$P{\rm Y} \left(\%\right)=22.62-31.27X_1+11.16X_1^2+7.66X_2-1.96X_2^2+15.28X3-2.41X_3^2-3.96X_1{X}_3-1.90X_2{X}_3$$

(12)

This revised model remained highly predictive, with an F-value of 503.55, a p-value below 0.001, and R² = 0.93. As shown in Figure A4, the predicted and observed values aligned closely along the diagonal, indicating minimal error and confirming the model’s accuracy and reliability in predicting PY during UMAE.

• UMAE Protein Recovery in Extract (PR_E).

The response surface model for protein recovery in the extract (PR_E) during UMAE showed an excellent fit, with an R² value of 0.97 (Table A2). The fitted model is represented by Equation (13),

$$P{R}_E\left(\%\right)=84.96-115.50X_1+27.53X_1^2+59.79X_2+39.02X_2^2-6.64X3-16.79X_3^2-0.13X_1{X}_2-10.59X_1{X}_3-6.33X_2{X}_3$$

(13)

The F-value and p-value emphasize the significance of the solid:liquid ratio (X₁) (F = 402.81, p < 0.001), which was the most significant factor. The quadratic term of the solid:liquid ratio (X₁²), the linear and quadratic terms of microwave power (X₂ and X₂²), the linear and quadratic terms of ultrasound power (X₃ and X₃²), the interaction between solid:liquid ratio and ultrasound power (X₁X₃) and the interaction between microwave power and ultrasound power (X₂X₃) were also significant, showing the combined effects of all factors on PR_E. Compared to PY above, the interaction between solid:liquid ratio and microwave power (X₁X₂) was not significant (p > 0.05). A refined model was developed by removing the non-significant interaction term, resulting in Equation (14):

$$P{R}_E \left(\%\right)=85.09-115.63X_1+39.02X_1^2+27.40X_2-6.64X_2^2+59.79X3-16.78X_3^2-10.59X_1{X}_3-6.34X_2{X}_3$$

(14)

This updated model demonstrated strong predictive accuracy, with an F-value of 928.54, a p-value below 0.001, and R² = 0.97. As shown in Figure A5, the predicted and observed values align closely along the diagonal, indicating minimal deviation and confirming the model’s robustness in predicting PR_E during UMAE.

• UMAE Protein Recovery in Precipitated mass (PR_P)

For the PR_p, the model fit was satisfactory, with an R² value of 0.97 (Table A2). The quadratic model for PR_p follows the equation:

$$P{R}_p\left(\%\right)=39.10-55.15X_1+30.66X_1^2+57.82X_2+19.73X_2^2-7.27X3-23.55X_3^2-3.27X_1{X}_2-8.25X_1{X}_3-0.70X_2{X}_3$$

(15)

The most significant factors were the linear term of solid:liquid ratio (X₁) (F = 250.29, p < 0.001), the linear term of microwave power (X₂) (F = 227.98, p < 0.001), and the quadratic term of ultrasound power (X₃²) (F = 309.78, p < 0.001). The quadratic term of solid:liquid ratio (X₁²), the quadratic term of microwave power (X₂²), the linear term of ultrasound power (X₃), the interaction between solid:liquid ratio and microwave power (X₁X₂) and the interaction between solid:liquid ratio and ultrasound power (X₁X₃) also had significant effects, while the interaction between microwave and ultrasound power (X₂X₃) did not have a significant impact (p > 0.05). These results highlight that all factors studied play a crucial role in influencing the PR_P of SPC extracts. To improve model precision, a refined regression was performed excluding the non-significant interaction. The revised model is given in Equation (16):

$$P{R}_p \left(\%\right)=39.68-54.93X_1+19.65X_1^2+30.05X_2-7.28X_2^2+57.07X3-23.56X_3^2-3.34X_1{X}_2-8.19X_1{X}_3$$

(16)

The updated model remained highly predictive, with an F-value of 816.91, p < 0.001, and R² = 0.97. As shown in Figure A6, predicted and observed PR_P values closely aligned along the diagonal, indicating strong agreement and confirming the model’s accuracy in describing PR_P behavior under UMAE conditions.

In summary, the solid:liquid ratio (X₁) emerged as the most significant factor influencing all examined response surfaces, as evidenced by the high F-values, followed by ultrasound power (X₃) and microwave power (X₂). Furthermore, all parameters, including their quadratic terms and interactions, significantly affected the extraction outcomes.

Notably, there is a lack of research investigating the synergistic effects of microwave and ultrasound on extraction processes. To address this gap, the findings of this study are compared with those from similar studies employing either MAE or UAE techniques. For instance, Tang et al. (2010) [36] and Arik Kibar and Aslan (2024) [37] explored ultrasound-assisted protein extraction from brewer’s spent grain and chickpea, respectively, highlighting the crucial role of ultrasound power and, most importantly, the solid:liquid ratio in maximizing protein yield. Similar findings were reported by Wang et al. (2020) [38], who examined protein recovery in the precipitated mass of pea flour. Ochoa-Rivas et al. (2017) [39] demonstrated improved protein yield and protein purity from peanut flour extract using both MAE and UAE. However, combining UAE followed by MAE did not result in significant improvements in these variables.

3.3.4. Graphical Interpretation of Model Predictions

The response surface plots demonstrate the effect of MW power, US power, and solid:liquid ratio on the PY (%), PR_E (%), and PR_P (%) of SPC extracts through the UMAE technique.

Figure 3a shows that the highest precipitation yields are presented for the lowest ratios (0.03 g/mL). Also, the presence of US positively impacts PY, as for US power above 250 W, the yield was significantly increased. Regarding Figure 3b, the increases in both US and MW power positively affected PY. Figure 4a,b reveals that proteins can be extracted more efficiently into the extract when a low ratio and high MW and US power are applied. Specifically, it seems that PR_E reaches its maximum values for US and MW power above 300 W. In addition, the lowest studied ratio (0.03 g/mL) gave significantly higher values of PR_E, regardless of the MW power applied. Figure 5 shows the interaction of MW power (a) and US power (b) versus ratio, and it seems that in both plots, lower solid:liquid ratios lead to increased PR_P. However, the behaviors of MW and US in each plot, respectively, are different. Elevated MW power consistently enhanced protein recovery, while US power exhibited a threshold effect. Below approximately 400 W, US power positively impacted protein recovery, but beyond this level, the protein recovery in the precipitated mass decreased.

The application of ultrasound and microwave treatments positively influences protein precipitation and recovery. The mechanism underlying UAE is primarily driven by cavitation, which generates localized energy within the medium. Similarly, MAE provides energy through heat generation, facilitated by dipole rotation and ionic conduction. These mechanisms contribute to the disruption of plant tissue, leading to alterations in its chemical and physical properties, thereby enhancing solvent penetration through cell membranes. Consequently, these processes facilitate the release of cellular matrix components and improve mass transfer efficiency [40,41].

From Figure 3, Figure 4 and Figure 5, we can see that the optimal extraction conditions for maximizing the selected protein extraction variables through the combined application of MW and US involve low solid:liquid ratios (below 0.04 g/mL) and moderate to high levels of US and MW power (between 400 and 700 W for US and 300 and 500 W for MW). These results align well with the optimal experimental results obtained and presented above (US power, 450 W; MW power, 500 W; ratio, 0.03 g/mL). In addition, these conditions effectively optimize PY, PR_E, and PR_P, achieving a balance without necessitating excessive energy input or prolonged processing time, while avoiding reduced efficiency.

Due to the limited research on protein extraction from SPC using the UMAE method, a comparative analysis with individual techniques was conducted. For example, Dabbour et al. (2018) [42] applied UAE to extract protein from sunflower meal, achieving a PR_E of 54.26% under optimal conditions. In contrast, the current study achieved a considerably higher PR_E of 79.9% using the UMAE approach. This improvement can be attributed to the synergistic interaction between ultrasound and microwave energy, which enhances cell wall disruption and mass transfer more effectively than either technique alone. Similarly, Sert et al. (2022) [43] investigated ultrasound treatment under alkaline conditions for protein recovery from pumpkin press cake, which is a different substrate with its own compositional characteristics, reporting a protein recovery percentage (PR_P) of 57.8%. While the reported PR_P values were moderate, our study obtained a higher PR_P of 66.3% from SPC, again demonstrating the advantage of combining technologies. Additionally, Putri et al. (2025) [44] utilized the MAE method for protein extraction from soybean, achieving a PR_P ranging from 47.5 to 77.8%. Although these values are partly comparable to those in our work, it is important to note that soybeans inherently contain higher baseline protein levels. Therefore, achieving similarly high recovery from SPC—a high-protein by-product that remains underexplored in advanced extraction processes—emphasizes the novelty and efficiency of the UMAE technique in valorizing such agro-industrial residues. Moreover, previous studies on oilseed meals such as lupin primarily employed single techniques (e.g., UAE or enzymatic treatment) with longer processing times and lower recovery efficiency, and they often lacked comprehensive optimization using RSM. For example, Fadimu et al. (2022) demonstrated that ultrasound pretreatment improved the bioactivity of enzymatically hydrolyzed lupin protein, but did not employ RSM or assess protein recovery efficiency in the context of multiple extraction methods [45]. Similarly, Aguilar-Acosta et al. (2020) applied ultrasound during lupin protein extraction, and reported moderate improvements in yield (14% increase) without integrated multi-factor optimization [46]. In contrast, our study integrates a dual-mode UMAE process with RSM-driven optimization, achieving superior results under milder conditions.

3.4. PLE of SPC

3.4.1. Optimization of PLE Using RSM

The protein extraction from SPC using the PLE technique was examined under various conditions of extraction temperature and time, combined with different solid:liquid ratios. Specifically, the extraction temperature ranged from 50 to 150 °C, the extraction time from 3 to 10 min and the solid:liquid ratio from 0.03 to 0.10 g/mL. The corresponding PY (%), PR_E (%), and PR_P (%) values were systemically recorded for each run.

The results show that PY ranged from 6.2% to 17.7%, with the highest value observed in run 12, which was conducted under the conditions of 50 °C, 6 min, and a solid:liquid ratio of 0.03 g/mL. PR_E and PR_P ranged from 54.2% to 70.9% and 11.2% to 47.5%, respectively, with the maximum values observed in run 13. The conditions for run 13 were 100 °C, 3 min, and a ratio of 0.03 g/mL.

Run 12 was identified as the optimal experiment due to its efficiency in achieving the highest PY at 17.7%, while also PR_E and PR_P values of 68.9% and 47.4%, respectively. Although run 13 exhibited slightly higher PR_E and PR_P values, these were achieved at a higher temperature, making run 12 the more favorable choice. The conditions of run 12 (50 °C, 6 min, and 0.03 g/mL) represent the optimal balance, maximizing protein recovery while operating at a lower temperature and maintaining energy efficiency.

3.4.2. Fitting Model to Data

The analysis of variance for the response surface of PY (%), PR_E (%), and PR_P (%) in the PLE is presented in

Table A3. Analysis of variance for the response surface of PY (%), PR_E (%), and PR_P (%) for protein extracts derived through PLE.

Source	Coefficients	Standard Error	Sum of Squares	DF	Mean Square	F-Value
PY (%) a
Model	18.34	2.87	7466.79	10	746.68	825.89 ***
X1	−25.89	3.69	270.05	1	270.05	298.70 ***
X12	15.12	2.64	32.27	1	32.27	35.70 ***
X2	14.28	2.33	158.51	1	158.51	175.33 ***
X22	7.91	1.32	66.18	1	66.18	73.20 ***
X3	−9.79	1.14	3.12	1	3.12	3.45 NS
X32	−3.67	0.94	13.78	1	13.78	15.24 ***
X1X2	2.73	0.70	13.77	1	13.77	15.23 ***
X1X3	−2.11	0.62	10.57	1	10.57	11.69 **
X2X3	−4.60	0.98	19.97	1	19.97	22.08 ***
Residual			31.64	35	0.90
Total			643.62	44
PRE (%) b
Model	83.80	9.82	159,094.49	10	15,909.45	1501.11 ***
X1	−101.08	12.64	1817.15	1	1817.15	171.45 ***
X12	14.81	9.02	355.84	1	355.84	33.57 ***
X2	62.45	7.99	242.30	1	242.30	22.86 ***
X22	26.25	4.53	242.93	1	242.93	22.92 ***
X3	−18.75	3.92	4.92	1	4.92	0.46 *
X32	−26.04	3.22	694.97	1	694.97	65.57 ***
X1X2	25.30	2.40	1178.95	1	1178.95	111.24 ***
X1X3	−4.72	2.11	52.86	1	52.86	4.99 *
X2X3	−3.08	3.35	8.94	1	8.94	0.84 NS
Residual			370.95	35	10.60
Total			4290.69	44
PRP (%) c
Model	56.13	9.52	53,470.46	10	5347.05	536.67 ***
X1	−66.82	12.25	2409.65	1	2409.65	241.85 ***
X12	44.92	8.75	175.09	1	175.09	17.57 ***
X2	27.82	7.74	2477.33	1	2477.33	248.64 ***
X22	18.41	4.39	840.82	1	840.82	84.39 ***
X3	−34.88	3.80	51.56	1	51.56	5.18 *
X32	−7.00	3.12	50.20	1	50.20	5.04 *
X1X2	9.67	2.33	172.03	1	172.03	17.27 ***
X1X3	−4.86	2.05	55.89	1	55.89	5.61 *
X2X3	−9.78	3.25	90.21	1	90.21	9.05 **
Residual			348.72	35	9.96
Total			7541.95	44

* p < 0.05, ** p < 0.01, *** p < 0.001, ^NS—not significant. ^a The coefficient of determination (R²) of model was 0.95. ^b The coefficient of determination (R²) of model was 0.91. ^c The coefficient of determination (R²) of model was 0.95.

.

• PLE-Precipitation Yield (PY)

The response surface model for PY under PLE showed a strong fit, with an R² value of 0.95 (Table A3). The fitted quadratic model is expressed by Equation (17),

$$P{\rm Y} \left(\%\right)=18.34-25.89X_1+15.12X_1^2+14.28X_2+7.91X_2^2-9.79X3-3.67X_3^2+2.73X_1{X}_2-2.11X_1{X}_3-4.60X_2{X}_3$$

(17)

The linear term of solid:liquid ratio (X₁) had the greatest effect (F = 298.70, p < 0.001), followed by the linear term of extraction temperature (X₂) (F = 175.33, p < 0.001). The quadratic term of solid:liquid ratio (X₁²), the quadratic term of extraction temperature (X₂²), the quadratic term of extraction time (X₃²), the interaction between solid:liquid ratio and extraction temperature (X₁X₂), the interaction between solid:liquid ratio and extraction time (X₁X₃), and the interaction between extraction temperature and extraction time (X₂X₃) were also significant. In contrast, the linear term of extraction time did not show a significant impact (p > 0.05). These results demonstrate that solid:liquid ratio and extraction temperature are critical in influencing the protein yield of SPC. A revised model excluding the non-significant term was developed, as shown in Equation (18).

$$P{\rm Y} \left(\%\right)=28.06-28.25X_1+8.37 X_1^2+11.26X_2-9.25X_2^2+1.22X_3^2+2.82X_1{X}_2-1.05X_1{X}_3-2.08X_2{X}_3$$

(18)

The model remained statistically robust with an F-value of 825.89, p < 0.001, and R² = 0.90, indicating high predictive power. As shown in Figure A7, the predicted values closely matched the observed data, with points clustering tightly along the diagonal. This confirms the model’s accuracy and reliability in estimating PY under PLE conditions.

• PLE-Protein Recovery in extract (PR_E)

The response surface model for PR_E under PLE showed a strong fit, with an R² value of 0.91 (Table A3). The fitted model is expressed by Equation (19),

$$P{R}_E\left(\%\right)=83.80-101.08X_1+14.81X_1^2+62.45X_2+26.25X_2^2-18.75X3-26.04X_3^2+25.30X_1{X}_2-4.72X_1{X}_3-3.08X_2{X}_3$$

(19)

The most significant factors were the solid:liquid ratio (X₁) (F = 402.81, p < 0.001) and the interaction between solid:liquid ratio and extraction temperature (X₁X₂) (F = 111.24, p < 0.001). The quadratic term of the solid:liquid ratio (X₁²), the linear and quadratic terms of extraction temperature (X₂ and X₂²), the linear and quadratic terms of ultrasound power (X₃ and X₃²), and the interaction between solid:liquid ratio and extraction time (X₁X₃) were also significant, showing the combined effects of all factors on PR_E. In contrast, the interaction between extraction temperature and extraction time (X₂X₃) was not significant (p > 0.05). After excluding the non-significant variable, the refined model is presented in Equation (20):

$$P{R}_E \left(\%\right)=86.97-101.18X_1+26.25X_1^2+11.64X_2-18.75X_2^2+59.37X3-26.04X_3^2+25.40X_1{X}_2-4.72X_1{X}_3$$

(20)

This model was statistically robust, with an F-value of 1501.11, p < 0.001, and R² = 0.91, confirming its predictive reliability. As shown in Figure A8, the predicted values closely aligned with observed data along the diagonal, indicating low error and validating the model’s effectiveness in predicting PR_E for the PLE process.

• PLE-Protein Recovery in Precipitated mass (PR_P)

The response surface model for PR_p under PLE demonstrated a good fit, with an R² value of 0.95 (Table A3). The fitted quadratic model is expressed by Equation (21):

$$P{R}_P\left(\%\right)=56.13-66.82X_1+44.92X_1^2+27.82X_2+18.41X_2^2-34.88X3-7.00X_3^2+9.67X_1{X}_2-4.86X_1{X}_3-9.78X_2{X}_3$$

(21)

The results indicate that all factors studied contributed significantly to PR_p, with the linear term of solid:liquid ratio (X₁) (F = 241.85, p < 0.001) and the linear term of extraction temperature (X₂) (F = 248.64, p < 0.001) emerging as the most influential.

The model’s F-value of 536.67, p < 0.001, and R² = 0.95 confirm its strong statistical significance in predicting PR_P under PLE conditions. As shown in Figure A9, the predicted values closely align with the observed data along the diagonal, demonstrating minimal error and confirming the model’s accuracy and reliability.

In summary, the solid:liquid ratio (X₁) and extraction temperature (X₂) were identified as the most influential factors for all evaluated variables in the PLE method for SPC, as evidenced by their high F-values. These findings align with existing literature on protein extraction from other raw materials. For instance, studies by de la Fuente et al. (2021) [47], González-García et al. (2021) [16], and Ho et al. (2007) [11] reported that extraction temperature was among the most critical parameters for maximizing protein content in extracts or solid residues after precipitation analyses of seabass side streams, Brewer’s spent grain, and flaxseed meal, respectively. Similarly, Hernández-Corroto et al. (2020) [48] demonstrated that higher extraction temperatures enhanced protein content in pomegranate peel extracts, while Zhou et al. (2021) [15] observed that increasing the extraction temperature during spirulina protein extraction also elevated the protein content in the solid residue post-precipitation. In contrast, extraction time exhibited minimal or no significant influence in all the studies mentioned above.

3.4.3. Graphical Interpretation of Model Predictions

The response surface plots (Figure 6, Figure 7 and Figure 8) demonstrate the effects of extraction temperature, extraction time, and solid:liquid ratio on the PY (%), PR_E (%), and PR_P (%) of SPC extracts through the PLE technique.

Figure 6a–c shows the effects of the different parameters on PY. Increased yields were observed at low solid:liquid ratios (below 0.04 g/mL) and moderate to low temperatures (50–100 °C). Conversely, extraction time had little to no significant effect on the results. A similar trend in the parameters’ combined effect was also evident for the PR_P, as shown in Figure 8a–c.

In Figure 7a, the interaction between extraction temperature and ratio is depicted regarding protein recovery in the extract. The highest recoveries were achieved at ratios below 0.04 g/mL and temperatures below 120 °C. High recoveries were also attainable at an increased ratio (0.10 g/mL) combined with increased temperatures (above 120 °C), although the range of recovery values is narrower. Finally, Figure 7b reveals that moderate extraction time (4–9 min), combined with low ratios, yielded high PR_E values.

Considering all the data presented in the plots, the best extraction conditions for SPC extract through PLE are low solid:liquid ratios and moderate to low temperatures, with the extraction time having minimal effect. These findings confirm the corresponding ones derived from a mathematical model analysis, which also identified the solid:liquid ratio and temperature as the major factors for PLE application, while extraction time plays an insignificant role. In general, elevated temperatures enhance protein recovery by improving solvent properties and accelerating mass transfer. As temperature increases, the viscosity and surface tension of water decrease, allowing better solvent penetration into the matrix and facilitating the formation of solvent cavities. This promotes the more efficient solubilization of proteins. Additionally, higher temperatures increase diffusion rates and reduce the time required to reach extraction equilibrium [11].

Compared to prior studies, the present work introduces several methodological advancements in the application of PLE for protein recovery from SPC. For instance, Neves et al. (2024) [49] reported a protein content of 133.5 mg/g of SPC using extraction conditions of 150 °C for 15 min. Although effective, such high-temperature conditions are considerably more energy-intensive than those employed in the current study. Under optimized conditions (50 °C, 6 min, 0.03 g/mL), our approach yielded a PR_E of 68.9%, demonstrating comparable results with significantly milder thermal input and shorter processing time. In another study, Ebru et al. (2010) [33] observed a PR_E ranging from 35.4% to 52.4% when extracting proteins from red pepper seed flour. Similarly, de la Fuente et al. (2021) [47] demonstrated that PR_E values varied widely, between approximately 20% and 80%, depending on the specific type of fish side stream utilized. However, their work was focused exclusively on animal-based substrates and did not incorporate a statistical optimization strategy. To the best of our knowledge, this study represents the first systematic investigation of RSM-optimized PLE applied specifically to SPC. The integration of response surface methodology enabled efficient process parameter tuning, resulting in high protein recovery under low-temperature and short-duration conditions. These findings underscore the potential use of PLE as a sustainable and scalable method for valorizing underutilized, protein-rich agro-industrial residues.

Consequently, the optimum conditions (temperature, 50 °C; time, 6 min; ratio, 0.03 g/mL) are consistent with both the experimental results and model predictions. Regarding PLE, these conditions effectively enhance PY, PR_E, and PR_P, achieving a balance that minimizes energy consumption and processing time while maintaining high efficiency.

3.5. Enzymatic-Assisted Extraction (EAE) of SPC

Protein extraction from SPC using the EAE method was examined under different enzyme activities. The amount of protease derived from A. Saitoi varied from 0 to 24 U/g raw material. The extraction process was conducted under optimal temperature and pH conditions to maximize enzyme activity. The results for PY (%), PR_E (%), and PR_P (%) are presented in Figure 9a–c.

As shown in Figure 9a,b, the PY and PR_E values did not vary significantly across different enzyme activities. The results indicate that the PY ranged from 15.4% to 17.3%, with the highest value recorded at an enzyme activity of 24 U/g raw material. Although increasing the enzyme activity from 3 to 24 U/g raw material led to a slight increase in PY, the increase was minimal relative to the enzyme amount. A similar trend was noted for PR_E, suggesting that the enzyme had little effect on improving the extraction effectiveness. PR_E values ranged from 53.6% to 61.4%. The highest PR_E value was obtained at 24 U/g raw material, while the lowest was recorded in the control sample (0 mg enzyme). PR_P values (Figure 9c) varied between 34.5% and 46.0%, with the maximum value also occurring at an enzyme activity of 24 U/g raw material. The results indicate that increasing enzyme activity generally improved PR_P, except for the 3 U/g raw material concentration, which showed a lower PR_P value (33.4%) than the control sample. This suggests that the presence of the enzyme, particularly at higher concentrations, had a positive effect on the precipitation of proteins extracted from SPC.

In general, proteases facilitate protein extraction through proteolysis. Since this study focuses on alkaline protein extraction, an alkaline protease derived from A. saitoi was selected for the experiments. The addition of protease can enhance alkaline extraction by reducing protein size and making extraction more efficient [50]. Previous research has explored enzyme-assisted protein extraction, using proteases, from sunflower meal and other sources. For instance, Yust et al. (2003) [51] improved the alkaline extraction process by treating sunflower meal with Alcalase, which significantly increased protein yield (87.4% vs. 57.5%) compared to the conventional method. Similarly, Baurin et al. (2022) [52] increased the protein extraction yield from sunflower meal up to 72.5% using alkaline proteases such as Protex 40E, Protex 6L, Protex 7L and Protex 51FP. Another study by Sari et al. (2013) [53] examined protein extraction from rapeseed, soybean and microalgae meals using the alkaline protease Protex 5L. Their findings show that while the enzyme had little impact on soybean meal extraction, it positively influenced protein extraction from rapeseed and microalgae meals. In another study, the use of Alcalase significantly increased the protein yield from sesame bran to 79.3%, compared to 24.5% achieved with conventional alkaline extraction [54]. Furthermore, combining enzyme-assisted and ultrasound-assisted extraction led to the highest yield (87.9%) [54].

Building on these findings, our study applied the EAE method to SPC using protease from A. saitoi, a substrate–enzyme combination that remains underexplored in the literature. Unlike previous studies that reported substantial gains in total protein recovery, EAE in our system did not significantly enhance PY or PR_E compared to the conventional method. However, it had a notable positive effect on protein recovery in PR_P, particularly at higher enzyme activity levels. This suggests that while the overall extraction yield may be substrate- or enzyme-limited, targeted fractionation benefits can still be achieved. These results highlight the selective potential of EAE in improving specific protein recovery stages from SPC, thereby contributing novel insights into the use of protease treatments in agro-industrial protein valorization.

3.6. Optimum Extracts—Comparison

Each extraction method was evaluated under various parameters, including extraction time and solid:liquid ratio, which were examined across all methods. The optimal conditions for each method, along with the results for PY %, PR_E %, PR_P %, PC_E (mg albumin eq./g raw material) and PC_P (mg albumin eq./g prec. mass) are summarized in Figure 10.

The results indicate that a solid:liquid ratio of 0.03 g/mL was optimal for all the extraction methods studied, while in contrast, the optimal extraction time varied across methods. A decreased solid:liquid ratio corresponds to a larger solvent volume relative to the extracted mass, leading to a lower protein concentration in the solvent. This difference enhances the concentration gradient between the solid and solvent phases, thereby increasing the driving force for mass diffusion during the extraction process [55,56]. Consequently, water–protein interactions are favored over protein–protein interactions in the larger solvent volume, improving protein solubility [40].

For PY, UMAE achieved the highest value (21.23%), followed by PLE, EAE, and CE, with yields of 17.70%, 17.27%, and 16.06%, respectively. Notably, the optimal yields for PLE and UMAE were obtained in significantly shorter extraction times—6 and 10 min, respectively—than EAE (240 min) and CE (60 min). A similar trend was observed for PR_E and P_E. UMAE and PLE demonstrated statistically higher values in comparison with CE and EAE, and thus showed a marked reduction in extraction time. UMAE yielded the highest PR_E (79.86%) and PC_E (230.80 mg albumin equivalents/g SPC).

These results contrast with those observed for protein recovery in precipitated mass, as all innovative techniques exhibited higher PR_P and greater PC_P compared to CE. This is because PR_P reflects not only the efficiency of the extraction method but also the precipitating potential of the extracted proteins. Specifically, UMAE achieved the highest recovery (66.30%), followed by PLE (47.40%) and EAE (45.96%), although EAE required 4 h. In terms of PC_P, PLE achieved the highest value (932.45 mg albumin equivalents/g precipitated mass), followed by UMAE and EAE. Compared to CE, PLE demonstrated a 48.2% increase in protein content with a 90% reduction in extraction time, and UMAE showed a 43.5% increase with an 83.3% reduction in time. Kalpana et al. (2020) [19] obtained protein isolates from de-oiled sunflower cake using conventional alkaline extraction, reporting PR_P values ranging from 56 to 74%. These results align with our study; however, the higher PR_P values observed with EAE, UMAE, and PLE demonstrate the superior efficiency of these innovative methods for protein extraction. In conclusion, these findings emphasize the advantages of novel extraction techniques in terms of yield, recovery, and reduced processing time.

The differences observed between extract and precipitated mass data can be attributed to the composition and solubility of the two primary storage proteins in SPC—helianthinin (11S globulin) and 2S albumins. While 2S albumins are smaller, highly soluble across a broad pH range, and easier to extract, they have limited precipitation when at an acidic pH. In contrast, helianthinin, which constitutes 70–80% of SPC proteins, requires specific conditions for solubilization, but precipitates efficiently at its isoelectric point [57,58,59]. Innovative methods such as UMAE and PLE appear to enhance helianthinin extraction by disrupting SPC tissue and protein aggregates, making larger polypeptides more accessible. The energy applied in UMAE may also fragment proteins into smaller, more soluble molecules, while elevated temperature and pressure in PLE promote protein release. Consequently, these methods increase helianthinin content in the extract, improving protein precipitation and resulting in higher PR_P and PC_P values. This underscores the advantages of UMAE and PLE in enhancing both protein extraction and recovery from SPC.

Finally, the protein purity in the precipitated mass, inferred from the PC_P values, is particularly noteworthy. All studied methods demonstrated protein purities exceeding 60%, with the innovative techniques UMAE and PLE achieving purities surpassing 90%, classifying them as highly purified proteins. Additionally, these innovative methods also exhibited high protein recovery efficiency (PR_E), particularly UMAE, which achieved values exceeding 65%. This combination of high purity and recovery efficiency underscores the effectiveness and selectivity of the alkaline extraction method and precipitation protocol. High-purity proteins hold significant value in the food industry due to their superior functional properties, nutritional benefits, and versatile applications in product development [60,61,62].

Comparing the results of this study with those reported in the literature highlights the efficacy of innovative extraction techniques in enhancing protein recovery and purity. Tang et al. (2025) [63] utilized the UMAE technique for protein recovery from lupin and investigated its efficiency compared to ultrasound and microwave techniques alone. Their results indicate an improved extraction yield and protein recovery due to the combined benefits of both technologies, representing a promising approach for protein extraction from plant sources. Tang et al. (2024) [64] investigated various UAE methods for pea protein extraction, demonstrating that innovative approaches improve protein extraction efficiency. Using extraction times ranging from 5 to 60 min and a solid:liquid ratio of 0.02 g/mL, they reported PY values ranging from 11.98% to 14.22%, PR_P from 46.72% to 52.53%, and PC_P between 727.2 and 787.6 mg protein/g raw material. CE, by comparison, resulted in values of 13.94% PY, 45.28% PR_P, and 659.4 mg protein/g raw material. Further studies optimized UAE for Brewer’s spent grain, where Tang et al. (2010) [36] achieved a protein content of 96.4 mg/g raw material at a solid:liquid ratio of 0.02 g/mL and an extraction time of 81.4 min. Similarly, Ochoa-Rivas et al. (2017) [39] examined peanut flour, reporting a 15 min UAE application yielding a PY of 65% and a PC_E of 850 mg/g raw material, while an 8 min MAE application achieved a PY of 55% and 920 mg protein/g raw material. Teixeira et al. (2024) [14] compared UAE, MAE, and PLE for broken black beans (BBB) under optimized conditions (15 min, 60 °C, solid:liquid ratio of 0.04 g/mL). UAE yielded a PY of 17.9% with 825 mg protein/g BBB, MAE achieved 16.0% PY with 613 mg protein/g BBB, and PLE resulted in the highest PY at 23.9% with 827 mg protein/g BBB. Similarly, Rudke et al. (2024) [65] compared UAE, MAE and PLE for protein recovery from peach seeds under alkaline conditions. The highest PR_P was observed for the PLE sample (50.3%), followed by MAE and UAE samples, with PR_P values of 37.9% and 24.6%, respectively. In another study, Galván et al. (2022) [66] investigated PLE for malt rootlets, achieving a protein content in the precipitated mass of 730 mg protein/g raw material in just 6 min by incorporating 33% ethanol.

From an economic and practical standpoint, each extraction method offers distinct advantages and limitations. CE is low-cost and simple in terms of materials and equipment, but it requires longer processing times and produces moderate yields, limiting its efficiency for industrial applications. UMAE, while requiring more sophisticated and energy-intensive equipment, enables significantly higher protein recovery in much shorter times, potentially reducing operational costs in high-throughput scenarios [67]. PLE also showed high efficiency and required less thermal energy, but its reliance on high-pressure systems may increase capital investment [6]. EAE, though environmentally friendly and selective, entails higher enzyme costs and prolonged incubation, making it less economically viable for large-scale operations [68]. These trade-offs underscore the importance of balancing extraction efficiency, scalability, and cost when selecting the most appropriate method for valorizing agro-industrial by-products.

Beyond economic considerations, each extraction technique also presents method-specific limitations that must be acknowledged. CE suffers from limited extraction efficiency due to slower mass transfer and less effective cell disruption, which may result in incomplete protein solubilization. In the case of UMAE, while the combination of ultrasound and microwave accelerates extraction kinetics, it also introduces the risk of localized overheating or protein denaturation if energy input is not precisely controlled [69]. PLE, although effective at solubilizing proteins under subcritical conditions, can be constrained by challenges in maintaining stable high-pressure and temperature environments, which may affect protein integrity or increase equipment wear over time [70]. For EAE, enzyme specificity and activity can vary significantly depending on substrate composition and extraction conditions, requiring careful optimization; in addition, residual enzymatic activity may interfere with downstream processing or product stability [3]. These technical constraints highlight the importance of not only optimizing process parameters, but also evaluating the robustness and reproducibility of each method under varying conditions.

Overall, in contrast to most previous studies that examined individual extraction techniques in isolation and often under narrowly optimized conditions, the present work provides a systematic, side-by-side evaluation of four distinct protein extraction methods applied to the same substrate (SPC), using uniform optimization protocols. This integrated approach not only enables a more robust comparison of efficiency and purity metrics, but also reveals how method-specific mechanisms—such as thermal–mechanical synergy in UMAE or subcritical fluid behavior in PLE—contribute to the recovery of distinct protein fractions. Unlike conventional methods that focus predominantly on yield, this study demonstrates how innovative techniques can simultaneously enhance PR_P and PC_P, particularly by enhancing the extraction of helianthinin, the dominant but less accessible globulin in SPC. These results position UMAE and PLE as promising, scalable alternatives to conventional and enzymatic methods, offering superior yield, processing time, and product quality. Furthermore, the economic and technical trade-offs evaluated herein provide essential insights for selecting appropriate technologies for the valorization of SPC and similar agro-industrial by-products.

3.7. HPLC Analysis of Extracts at Optimum Conditions

The amino acid (AA) profile obtained via HPLC-DAD analysis, presented in Figure 11, demonstrates the presence of essential and non-essential amino acids in the recovered protein fractions. The analysis confirms the presence of aspartic acid (ASP), serine (SER), glutamic acid (GLU), glycine (GLY), histidine (HIS), arginine (ARG), threonine (THR), alanine (ALA), proline (PRO), cysteine (CYS), tyrosine (TYR), valine (VAL), methionine (MET), lysine (LYS), isoleucine (ILE), leucine (LEU), and phenylalanine (PHE).

Table 5. Amino acid concentrations (mg/g) in SPC protein extracts obtained using different extraction methods CE, EAE, UMAE, and PLE. Concentrations are expressed as mg of amino acid per g of raw material.

AA	CE	EAE	UMAE	PLE
	mg/g Raw Material	mg/g Raw Material	mg/g Raw Material	mg/g Raw Material
ALA	7.28 ± 0.43 b	8.04 ± 0.34 b	10.8 ± 0.5 a	7.68 ± 0.42 b
ARG	23.0 ± 1.3 b	25.2 ± 0.9 b	33.2 ± 1.1 a	24.2 ± 1.5 b
ASP	13.9 ± 0.8 c	17.3 ± 1.1 a,b	20.2 ± 1.6 a	14.7 ± 0.9 b,c
CYS	2.57 ± 0.15 b	2.90 ± 0.22 b	4.08 ± 0.31 a	2.75 ± 0.17 b
GLU	31.6 ± 1.8 b	34.6 ± 1.2 b	45.4 ± 1.4 a	33.2 ± 1.3 b
GLY	7.12 ± 0.41 b	7.87 ± 0.56 b	10.5 ± 0.42 a	7.51 ± 0.51 b
HIS	2.89 ± 0.17 b	3.26 ± 0.23 b	4.54 ± 0.18 a	3.09 ± 0.20 b
ILE	5.82 ± 0.34 b	6.45 ± 0.44 b	8.69 ± 0.31 a	6.15 ± 0.40 b
LEU	10.00 ± 0.6 b	11.10 ± 0.8 b	14.70 ± 0.5 a	10.60 ± 0.7 b
LYS	5.33 ± 0.31 b	5.91 ± 0.26 b	8.00 ± 0.45 a	5.64 ± 0.27 b
MET	2.25 ± 0.14 b	2.55 ± 0.23 b	3.62 ± 0.26 a	2.41 ± 0.19 b
PHE	6.63 ± 0.38 b	7.33 ± 0.45 b	9.82 ± 0.51 a	7.00 ± 0.39 b
PRO	7.44 ± 0.43 b	8.22 ± 0.51 b	11.00 ± 0.55 a	7.85 ± 0.41 b
SER	7.61 ± 0.44 b	8.39 ± 0.23 b	11.20 ± 0.36 a	8.02 ± 0.22 b
THR	5.98 ± 0.35 b	6.62 ± 0.36 b	8.92 ± 0.39 a	6.32 ± 0.33 b
TYR	4.03 ± 0.24 b	4.49 ± 0.28 b	6.15 ± 0.33 a	4.28 ± 0.26 b
VAL	7.12 ± 0.41 b	7.87 ± 0.42 b	10.50 ± 0.49 a	7.51 ± 0.38 b

Values labeled with different letters indicate significant differences between different techniques for each AA (p < 0.05).

presents the amino acid concentrations in SPC protein extracts produced using various extraction methods, highlighting differences in yield across CE, EAE, UMAE, and PLE techniques.

The amino acid profile of SPC obtained through different extraction methods aligns with findings in the literature, confirming its potential as a valuable protein source. The identified amino acids in the present study include ASP, SER, GLU, GLY, HIS, ARG, THR, ALA, PRO, CYS, TYR, VAL, MET, LYS, ILE, LEU, and PHE, which are consistent with previous studies on sunflower proteins [71,72]. The data confirm that GLU, ARG, and ASP are the most abundant amino acids in SPC, in agreement with reports by Sarrazin et al. (2003) [73], who noted that GLU dominates sunflower protein composition, followed by ASP and ARG. In this study, GLU reached up to 45.4 ± 1.4 mg/g in raw material extracted with UMAE, while ARG concentrations peaked at 33.2 ± 1.1 mg/g in UMAE, confirming their high presence in SPC proteins.

The SPC is known for its essential amino acid content. Still, it is often noted for its relatively low LYS concentrations compared to other protein sources, which could limit its use as a sole protein source [74]. The results confirm this observation, with LYS concentrations ranging from 5.33 ± 0.31 mg/g in CE to 8.00 ± 0.45 mg/g in UMAE, values that remain lower than those found in legumes and some cereal-based protein sources. Despite this, sunflower proteins provide other essential amino acids, such as LEU, VAL, and MET, which contribute to their nutritional quality. LEU was measured at 14.7 ± 0.5 mg/g in raw material, while LYS concentrations ranged from 5.33 ± 0.31 mg/g in CE to 8.00 ± 0.45 mg/g in UMAE. VAL showed a similar trend, reaching 10.5 ± 0.49 mg/g in raw material, reinforcing the value of advanced extraction methods in increasing protein concentration.

The extraction process plays a significant role in determining amino acid recovery. Studies have shown that processing conditions, such as germination and hydrolysis, can enhance certain amino acid levels, particularly LEU and MET [75]. The results also indicate higher recoveries of these amino acids in PLE and UMAE compared to CE, with MET reaching 3.62 ± 0.26 mg/g in UMAE versus 2.25 ± 0.14 mg/g in CE, highlighting the benefits of advanced extraction. These trends align with previous research showing that sunflower protein retains its amino acid integrity under different processing conditions [73]. The concentration of sulfur-containing amino acids, including CYS and MET, was lower than that of other essential amino acids, a pattern observed in previous studies [76]. This reinforces that supplementation with other protein sources may be necessary for complete nutritional balance. Sulfur-containing amino acids, MET and CYS, were found in lower concentrations across all methods, with methionine ranging from 2.25 ± 0.14 mg/g in CE to 3.62 ± 0.26 mg/g in UMAE. Their lower abundance may be attributed to oxidative degradation, particularly under high-energy conditions. PRO, which plays a key role in protein folding and stability, showed increased recovery with UMAE, reaching 11.00 ± 0.55 mg/g in raw material, reinforcing the effectiveness of pressure-assisted techniques in extracting hydrophobic amino acids.

The amino acid profile obtained in this study confirms the consistency of the nutritional values of SPC with what has been found in prior research, with GLU, ARG, and ASP being the most abundant amino acids. Although LYS remains a limiting factor, other essential amino acids, such as LEU, VAL, and MET, enhance its value as a protein ingredient. These findings underline the superior efficiency of these methods in maximizing protein extraction from SPC, making them promising approaches for food, feed, and biotechnological applications. Further optimization could improve the bioavailability of limiting essential amino acids such as LYS and MET, ensuring the development of high-quality sunflower protein ingredients.

4. Conclusions

The study systematically investigated and optimized protein recovery from SPC using CE and advanced techniques such as UMAE, PLE, and EAE. Among these methods, UMAE was identified as the most optimal, demonstrating superior performance by achieving the highest PY, PR_E, and PR_P, while significantly reducing extraction time compared to CE. For CE, the optimal conditions (0.03 g/mL solid:liquid ratio, 60 min) resulted in PY—16.1%, PR_E—64.1%, and PR_P—35.0%. UMAE, under optimal conditions (0.03 g/mL, 450 W MW, 500 W US), achieved the highest values, of PY—21.2%, PR_E—79.9%, and PR_P—66.3%, with only 10 min of extraction. PLE also showed strong results, with PY—17.7%, PR_E—68.9%, and PR_P—47.4% at 0.03 g/mL, 6 min, and 50 °C. In contrast, EAE, though selective and environmentally friendly, required longer extraction (4 h), yielding PY—17.3%, PR_E—61.4%, and PR_P—46.0% at 24 U/g enzyme activity.

Across all extraction methods, a solid:liquid ratio of 0.03 g/mL was optimal, maximizing extraction efficiency by enhancing mass diffusion and protein solubility. Among the studied techniques, UMAE and PLE clearly outperformed CE and EAE, delivering higher efficiency and shorter processing times. Their enhanced ability to extract helianthinin by disrupting protein aggregates likely contributed to improved protein recovery and precipitation. Notably, both methods yielded protein purities exceeding 90%, making them promising for food and nutraceutical applications.

The amino acid profile analysis confirmed that SPC-derived proteins are rich in essential amino acids, with GLU, ARG, and ASP being the most abundant. While LYS levels were relatively low, the presence of LEU, VAL, and MET contributed to its nutritional value. Notably, UMAE significantly enhanced amino acid recovery compared to CE. These findings highlight the effectiveness of advanced extraction technologies in producing high-quality sunflower-derived protein ingredients for applications in food, feed, and biotechnology.

Future research should focus on scaling these extraction techniques for industrial application, conducting detailed techno-economic and life cycle assessments, and exploring protein functionality in food systems. Furthermore, a comparative evaluation of the structural, techno-functional, and bioactive properties of proteins obtained through each extraction method will provide deeper insights into their application potential in different product formulations. In addition, future studies should investigate the bioavailability of amino acids recovered through different extraction techniques to better assess their nutritional efficacy and suitability for food and nutraceutical product development.