Assessment of whole egg fractionation by tangential flow filtration: The problem of low-density lipoprotein aggregates

Available online at www.sciencedirect.com Food and Bioproducts Processing journal homepage: www.elsevier.com/locate/fbp Assessment of whole egg fractionation by tangential flow filtration: The problem of lowdensity lipoprotein aggregates Gema Puertas, Patricia Cazón, Manuel Vázquez ⁎ Department of Analytical Chemistry, Food Technology Area, Faculty of Veterinary, Campus Terra, University of Santiago de Compostela, 27002 Lugo, Spain article info Article history: Received 29 October 2022 Received in revised form 4 May 2023 Accepted 11 May 2023 Available online 13 May 2023 Keywords: Filtration Egg products Separation Cholesterol Low-density lipoproteins Food matrix abstract Eggs are a compelling source of bioactive compounds. Membrane separation is a green technology that allow components to be obtained while preserving their biological activity. The objective of this study was to assess the application of membrane separation technology to egg products. Whole egg plasma, egg yolk plasma and egg white were tangential flow filtrated evaluating the effect of the pore size and membrane materials. Transmission of dry matter and proteins from feed to filtrate was noticed in all the processes. Nevertheless, cholesterol and carotenoids were not detected at filtrates from 750 kDa to 0.2 µm when mixed ester or polyethersulfone were the membrane materials. The interactions between egg yolk and egg white components difficult the separation. Cholesterol was part of low-density lipoproteins. These lipoproteins together with carotenoids were filtrated with modified polyethersulfone at a minimum pore size of 0.2 µm. High uncertainty was obtained due to low-density lipoproteins aggregation. Three kinds of aggregates can justify these results: low-density lipoproteins aggregated themselves, low-density lipoproteins bound to ovomucin and other smaller egg white proteins (< 750 kDa), and low-density lipoproteins bound only to smaller egg white proteins. The detected chelate effect of egg white proteins could be important in vivo. Therefore, research should be led to assess the influence of cooking method and digestion of the different egg components on the bioaccessibility of nutrients and the generation of bioactive peptides. Moreover, the filtration processes analysed herein may be useful techniques for liquid food matrix studies. © 2023 The Author(s). Published by Elsevier Ltd on behalf of Institution of Chemical Engineers. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Hen eggs are a valuable source of essential amino acids and highly bioavailable proteins. Its high content in these makes the egg a benchmark for measuring the quality of food proteins. They also provide essential fatty acids, phospholipids, choline, sialic acids, minerals, carotenoids and all vitamins except vitamin C. This composition award health beneficials to eggs, including antioxidant properties, antimicrobials, anticancer, antihypertensive and immunomodulatory activities (Réhault-Godbert et al., 2019). In addition, eggs are a moderate energy source with a lot of culinary potential and generally available at an affordable price (CzarnowskaKujawska et al., 2021). Despite this great nutritional composition, their high cholesterol content of 415 mg / 100 g of edible portion (Puertas and Vázquez, 2021a) has typically dominated their nutritional evaluation. Several studies have focused on how egg intake modified blood cholesterol levels with controversial conclusions. The European Food Safety Authority https://doi.org/10.1016/j.fbp.2023.05.003 0960-3085/© 2023 The Author(s). Published by Elsevier Ltd on behalf of Institution of Chemical Engineers. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). ]]]] ]]]]]] ⁎ Corresponding author. E-mail address: [email protected] (M. Vázquez). Food and Bioproducts Processing 140 (2023) 99–109 (EFSA) recognised that there is a positive dose-dependent relationship between intake of dietary cholesterol with blood low density lipoproteins (LDL)-cholesterol concentrations. However, such authority decided not to propose a reference on cholesterol intake (Puertas and Vázquez, 2019a). Nowadays the recommendation for dietary cholesterol consumption is to be as low as possible without compromising the nutritional adequacy of the diet (U.S. Department of Agriculture and U.S. Department of Health and Human Services, 2020). Like that, eggs are included in a healthy dietary pattern. Cholesterol in egg is in the yolk exclusively. It is 95% linked to LDL and the remainder is bound to high density lipoproteins (HDL) or lipovitellin (Griffin, 1992). LDL are spherical nanoparticles with a lipid core of triglycerides and cholesterol esters surrounded by a monofilm of phospholipids, proteins and non-esterified cholesterol (Anton, 2013). About 90% of LDL particles had a diameter between 17 and 100 nm (Xie et al., 2020) The monofilm allows the solubility of lipids in the aqueous phase. Egg yolk is a natural oil in water emulsion where the continuous aqueous phase is called plasma and the dispersed phase are insoluble denser structures (0.3–2 mm) referred to as granules. Centrifugation of whole egg produces the separation of these phases. Two egg products are obtained: the pellet that encompasses the granules and the supernatant that contains egg yolk plasma and albumen (Puertas and Vázquez, 2021a). Egg yolk plasma contains LDL and livetins, while granules included HDL, phosvitin and LDL granules (Anton, 2013). The liquid nature of plasma makes it a good target for a membrane separation process. Multiple studies have been done to obtain the bioactive compounds of egg yolk and egg white independently (Li et al., 2022; Zhang et al., 2021). However, yolk and albumen are normally cooked and consumed together; and whole egg is the full source of nutrients and bioactive compounds. Nowadays, the important is the whole food and food matrix studies are being explored (Miller et al., 2023). As numerous nutrients in the food matrix can interact in a synergistic or antagonistic manner, the fundamental nutritional entity is the food containing all nutrients rather than the nutrient in isolation (Pérez-Jiménez et al., 2018). In the present, diets that focus on specific nutrients, such as fat, to address metabolic diseases are deemed suboptimal; thus, historical dietary approaches aimed at isolated nutrients have become obsolete (Miller et al., 2023). Therefore, tangential flow filtration (TFF) was applied to whole egg as a food matrix case. Two membrane separation processes in TFF, also called cross-flow, were studied in our work: ultrafiltration (UF) and microfiltration (MF). Both of them are pressure-driven. At each process, two products are obtained from the feed introduced into the membrane system. Retentate is the fraction retained by the membrane. Filtrate or permeate is the fraction that passes through the membrane. It is crossflow because the feed is pumped parallelly to the membrane, and it is possible to recirculate the retentate back to the feed flow. While ‘pressure-driven’ means that the main driving force of these processes is transmembrane pressure (TMP); the pressure discrepancy between retentate side and filtrate side (Dhineshkumar and Ramasamy, 2017). The use of membrane technology as a processing and separation method in food industry is gaining wide applications. It is considered a green technology. Because membrane separation processes are more efficient in economic and energetic terms compared to high temperatures treatments for pasteurization and sterilization. In addition, they could avoid the use of additives and chemicals (for example to extend the shelf-life or solvents for extraction), which is both better for environment and human health. Furthermore, they preserve the natural taste and nutritional value of food products in heat-sensitive components (Dhineshkumar and Ramasamy, 2017), such as eggs. Therefore, it has been explored as a method for food components fractionation. For example, egg white proteins have been obtained using UF, with the benefit of retaining their biological activity (Li et al., 2022). Other studies that ultrafiltrated eggs were, for example, to desalinate salted duck egg whites (Thammasena et al., 2020) or to isolate antioxidant peptides produced from egg yolk protein enzymatic hydrolysis (Chay Pak Ting et al., 2011). In fact, UF is already employed in food industry for whole egg and egg white concentration (Guiga and Lameloise, 2019). However, to our knowledge, MF to egg products had never been studied until our previous work (Puertas et al., 2023). At that publication it was demonstrated that the egg products obtained through TFF can be accurately quantified by UV–VIS–NIR spectroscopy combined with chemometrics. UF refers to membrane process with pores ranging approximately from 3 to 100 nm. MF concerns membranes with pore sizes ranging from 0.1 to 10 µm. Bacteria, yeasts and milk fat globules have appropriate sizes for MF while virus and soluble proteins for UF. Colloids and casein micelles are suitable for both filtrations (Guiga and Lameloise, 2019). Since UF mainly addresses macromolecules separation (such as proteins and peptides) is preferably expressed as molar mass. The membranes employed are classified according to the molecular weight cutoff (MWCO), in g/mol or Da (Dalton), that is defined as the molar mass of the molecule rejected at 90% (Guiga and Lameloise, 2019). However, the separation principle is not based on the pore sizes alone, the charge of the molecule/solutes and their affinity for the filtering membrane are also key aspects, especially in UF (Dhineshkumar and Ramasamy, 2017). Hydrophilic membranes are considered more advantageous over hydrophobic to filter an aqueous solution of proteins at neutral pH. Because negatively charged proteins are more repulsive against hydrophilic membranes resulting in less occurrence of membrane fouling. This phenomenon indicates the deposition of solutes/ particles on the membrane surface and/or into the membrane pores (Dhineshkumar and Ramasamy, 2017). It is described to be one of the key challenges restricting the practical application of UF (Mohammad et al., 2012). Thus, plasma is a better target than whole egg or egg yolk. A fouling mitigation technique employs for example in cell separation process is alternating tangential flow filtration (ATF) (Weinberger et al., 2022). In ATF, alternating pressure and vacuum are applied to the filtrate flow using a diaphragm pump. As a result, the reverse flow is used to reduce the fouling of the hollow fiber membrane (Matte, 2022). The objective of this study was to assess the application of TFF to egg products. Low pressures, no additives, and only water was employed to dilute whole egg plasma, egg yolk plasma and egg white before UF and MF. Cholesterol, dry matter, protein content and colour were analysed. Whole egg was chosen as the full source of bioactive compounds as case study. Yolk and white were employed to understand the mechanisms involved. 100 Food and Bioproducts Processing 140 (2023) 99–109 2. Materials and methods 2.1. Egg products preparation The egg products employed are described at our previous work (Puertas et al., 2023). Briefly, 11 starting egg products were prepared from 89 fresh eggs. These starting products were 7 liquid whole eggs (made with 8 or 10 eggs each), 2 liquid egg yolks and 2 liquid egg whites (made with 12 and 13 eggs manually separated). To simplify, liquid egg white is called egg white henceforth. All samples were homogenised using a high-performance homogenizer (Ultra Turrax®, IKA, Staufen, Germany) at 10,000 rpm for 1 min Next, liquid egg yolk was diluted 1:1 with tap water before centrifugation. Following, liquid whole egg and diluted liquid egg yolk samples were centrifuged at 12,000 g, 40 min and 8 ºC (Sorvall Contifuge Stratos, Thermo Fisher Scientific, Waltham, MA, USA). Conditions were settled according to previous studies (Puertas and Vázquez, 2021a, 2021b) to maximize granules separation and cholesterol in plasma. These settings caused some cholesterol precipitation inevitably due to some LDL precipitation in granules. Maximum granules’ yield was prioritised to minimize membrane fouling. After, plasmas were separated from granules manually. 2.2. Tangential flow filtration The egg products filtrated are identified in Table 1. A total of 11 runs were performed. Whole egg plasmas were analysed from run 1–7, egg yolk plasmas were employed at runs 8 and 9; while egg whites were studied at runs 10 and 11. Egg products were diluted 1:1 (w/w) with water before filtration. KrosFlo® Research IIi TFF System (Repligen, California, USA) was used. This system employs hollow fiber modules, and it was settled in a batch filtration configuration. As the aim was to preserve nutritional and functional properties of egg products organic polymeric membranes were considered over inorganic materials. They are available in a wide range of pore sizes, cheaper, more sensitive to washing chemicals and normally have a high packing density (Dhineshkumar and Ramasamy, 2017). The characteristics of the membranes employed are described in Table 1. MF was achieved with three pore sizes: 0.1 µm, 0.2 µm and 0.65 µm. While UF was performed with a MWCO of 750 kDa. Assuming the simplest shape of a sphere, this MWCO is equivalent to a protein diameter of about 12 nm minimum (Erickson, 2009). Three kinds of hydrophilic organic membranes were studied: Polyethersulfone (PES), modified polyethersulfone (mPES) and mixed ester (ME, consisting of cellulose acetate and cellulose nitrate). According to the hollow fiber inner lumen and their effective length, five surface areas were analysed from 500 cm 2 to 2600 cm 2 . Three parameters were calculated to quantify the membrane separation: transmission rate (Tr), concentration factor (CF) and weight reduction ratio (WRR). Transmission rate was expressed as a percentage according to the following equation: = ×Tr Concentration Concentration (%) 100 i Filtrate Retentate (1) Where i was the composition parameter studied because transmission rates were calculated for each composition parameter studied (see below). The concentrations of the composition parameters studied were measured at the end of the TFF in the filtrate and the retentate (Dhineshkumar and Ramasamy, 2017). CF is the ratio between the concentration in the retentate divided by initial concentration (Guiga and Lameloise, 2019). They were also calculated for each composition parameter: =CF Concentration Concentration i Retentate Initial (2) Whereas WRR is the ratio of the initial weight divided by final weight. This term is normally calculated as volume reduction ratio (VRR) (Guiga and Lameloise, 2019). In this work, initial and retentate samples were weighted: =WRR Weight Weight Initial Retentate (3) Initial concentration or weight considered were previous dilution. 2.3. Composition parameters The composition parameters studied on the initial feed (plasmas and egg white), retentate and filtrate were dry matter (DM), proteins and cholesterol. DM content was determined by gravimetric method. An aliquot of 2–5 g of sample was weighted, dried for 24 h in an oven at 105 ºC and Table 1 – Identification and characterization of the filtration systems studied. Run System Characteristics No. Sample Pore size / MWCO Membrane material Effective Fiber Length (cm) Surface area (cm 2 ) Fiber inner lumen (mm) 1 WEP 750 kDa mPES 41.5 1600 0.5 2 WEP 750 kDa mPES 41.5 1600 0.5 3 WEP 0.1 µm ME 20 720 0.63 4 WEP 0.2 µm PES 20.8 500 1 5 WEP 0.2 µm PES 20.8 500 1 6 WEP 0.2 µm mPES 65 2600 0.5 7 WEP 0.65 µm mPES 41.5 1075 0.75 8 EYP 750 kDa mPES 41.5 1600 0.5 9 EYP 0.65 µm mPES 41.5 1075 0.75 10 EW 750 kDa mPES 41.5 1600 0.5 11 EW 0.65 µm mPES 41.5 1075 0.75 WEP: whole egg plasma. EYP: egg yolk plasma. EW: egg white. MWCO: molecular weight cut-off. mPES: modified polyethersulfone. ME: mixed ester. PES: polyethersulfone. 101 Food and Bioproducts Processing 140 (2023) 99–109 then dried sample was weighted. DM content was expressed as g dry matter per 100 g fresh sample. Protein content was measured with the microvolume spectrophotometer Nanodrop 2000 (Thermo Scientific™, Waltham, MA, USA). The method followed is described elsewhere based on the Warburg and Christian method (Warburg and Christian, 1942). Briefly, it was an estimation of protein based on the maximum absorption at 280 nm that most proteins exhibit due primarily to the presence of tyrosine and tryptophan. As nucleic acids also absorb at that wavelength, a proportion is removed by calculation. Their contribution is eliminated by subtracting a proportion of the absorbance at 260 nm, where nucleotides absorb much more strongly (Layne, 1957). Protein content was expressed as mg per ml of sample. An enzymatic kit method (Enzytec™, RBiopharm AG, Darmstadt, Germany) was employed for cholesterol quantification based on previous studies (Puertas and Vázquez, 2019b). The sizes of the samples for analysis were 0.5 g for egg yolk, 1 g for liquid whole egg, 1.5–2 g for plasmas, 1–2.5 g for retentates and 5 g for filtrates. These quantities were decided based on preliminary studies to work within the detection limits stablished by manufacturer. Cholesterol content was expressed as mg per 1 g of egg product. All the composition determinations were done at least in duplicate, except for protein content that measurements were in triplicate. Initial pH of plasmas and egg whites were measured with Crison pH 25 + pH-meter (Crison Instruments S. A., Barcelona, Spain). 2.4. Colour parameters Spectrophotometer Jasco V670 (Jasco Inc., Hachioji, Tokyo, Japan) was employed to acquire the VIS spectra from 380 to 780 nm of each egg product. The cuvette employed was made of Quartz Suprasil ® 300 with a 1 mm light path (Hellma GmbH & Co. KG, Mulheim, Germany). Samples were measured at 22–24 °C. The spectral data were collected with Spectra Manager™ II software (Jasco Inc., Hachioji, Tokyo, Japan). This software calculated the CIE parameters L* , a* and b* established on the 2 Degree standard observer with light source D65. Duplicate values were achieved for each sample. The parameter L* expresses the degree of lightness (0 = black, 100 = white). The coordinate a* measures redness (−100 = green, 100 = red) and b* reflects yellowness (−100 = blue, 100 = yellow) (Puertas et al., 2023). 2.5. Statistical analysis Significant differences between the filtration results were determined by one-way analyses of variance (ANOVA) and post-hoc Tukey’s Honestly Significant Difference (HSD) test. Significant difference at p < 0.05 was defined. 3. Results and discussion 3.1. Filtration parameters and composition results In this work, we investigated TFF as a method for egg compounds separation in the whole egg plasma matrix. Egg yolk plasma and egg white were also analysed to detect differences between food matrices and to comprehend the mechanisms involved. The separation process was analysed through changes in DM, proteins, cholesterol and colour parameters. Our approach aimed to preserve nutritional and functional properties of egg products. Therefore, organic hydrophilic membranes were analysed and no additives were employed, only water was added as diluent. Mean TMP and filtrate fluxes are described in Table 2. To avoid irreversible membrane fouling, filtration was stopped when pressure at the feed rose while filtrate weight did not. Except for run 9 where this membrane resistance was not detected, and separation was stopped when retentate weight was minimum. As a result, mean TMP achieved in this work (maximum almost 45 kPa in MF) were close to other studies where a minimum of 50 kPa was employed in MF (Weinberger and Kulozik, 2021a). All TFF developed with similar fluxes (p > 0.05); except for run 9 that achieved the highest flux. Egg yolk plasma MF (run 9) was accomplished quickly with low resistance. The fractionation effect of TFF, can be analysed by transmission rates. As this parameter considers concentrations at filtrates (Eq. (1)). Mean transmission rates are shown in Fig. 1. In contrast, to visualise the concentration effect of TFF on the egg products studied, CF and WRR were included in Fig. 2. A black dash-dotted line marks a limit at 1. According to Eq. (2), CF was above that black line when the concentration at the retentate was over the initial concentration, then TFF main accomplishment was concentration. Whereas, when the concentration at the retentate was lower than initial concentration, CF was below the black line; and did not achieve to concentrate at the retentate. The grey dash-dotted line marks a CF value of 0.5. CF around that line supposed that final concentration at retentate was close to that of the diluted sample introduced in the system; because they were diluted 1:1 before TFF. MF at run 3 (0.1 µm - ME) and run 4 (0.2 µm – PES) achieved the highest average TMP (Table 2). However, these TMP were not linked to the highest transmission rates (Fig. 1). On the contrary, from run 5, it was detected that protein and DM transmission rates increased (p < 0.05) when a lower TMP was applied (Table 2). Same TFF system was employed at run 4 and 5 (Table 1). This effect has been described before (Weinberger and Kulozik, 2021b). Protein fractionation was Table 2 – Results from operational variables. Run key data No MWCO / Pore size & material TMP (kPa) Filtrate flux (Lh −1 m −2 ) 1 750 kDa mPES 35.51 ± 17.24 b 2.46 ± 1.36 b 2 750 kDa mPES 35.4 ± 9.03 b 2.95 ± 1.33 b 3 0.1 µm ME 44.61 ± 15.17 a 4.07 ± 1.75 b 4 0.2 µm PES 42.26 ± 13.93 a 2.62 ± 1.16 b 5 0.2 µm PES 33.16 ± 11.24 b 3.34 ± 1.01 b 6 0.2 µm mPES 26.27 ± 13.72 bc 7.99 ± 7.07 b 7 0.65 µm mPES 29.64 ± 13.44 b 3.30 ± 2.00 b 8 750 kDa mPES 32.68 ± 8.41 b 2.40 ± 2.38 b 9 0.65 µm mPES 1.24 ± 1.86 e 24.81 ± 37.42 a 10 750 kDa mPES 18.34 ± 7.45 cd 6.72 ± 1.04 b 11 0.65 µm mPES 10.76 ± 4.41 de 4.56 ± 0.86 b MWCO: molecular weight cut-off. mPES: modified Polyethersulfone. ME: Mixed Ester. PES: Polyethersulfone. TMP: mean transmembrane pressure. Means with different letters within the column differed significantly from each other according to Tukey’s test (p < 0.05). Runs 1–7 employed whole egg plasma; runs 8 and 9 employed egg yolk plasma; runs 10 and 11 employed egg white. 102 Food and Bioproducts Processing 140 (2023) 99–109 negatively affected by high TMP due to deposit layer compaction and low protein transmission. In this work, the lower TMP described at run 5, let crossed more proteins (and DM) and this reduced CF of its retentate, being all around 1 (Fig. 2). Differences between CF from run 4 and 5 were statistically significant (p < 0.05). As cholesterol did not pass PES membrane, both CF were over one. On the other side, mPES membranes allowed the filtration of cholesterol, and more proteins and DM than PES; in agreement with the higher transmission rates (p < 0.05) achieved at runs 1, 2 and 6 despite pore sizes were below (750 kDa) or equal (0.2 µm) to runs 3–5. Cholesterol did not cross through UF membranes (750 kDa). It appeared at 0.2 µm filtrates. To understand the variations between whole egg plasma UF (run 1 and 2) in transmission rates (Fig. 1), although no differences at operational parameters were developed (Table 2), an analysis of runs 8 and 10 gave the answer. At run 8 plasma from egg yolk was ultrafiltrated with the same membrane of 750 kDa made of mPES. Retentate from run 8 had almost initial plasma composition: all CF were slightly above 1 (Fig. 2) and transmission rates were close to zero (Fig. 1). Mainly water from dilution crossed through the membrane; a concentration process was developed. At run 10, egg white was filtered through the same 750 kDa system. In this case, water, proteins and DM from egg white crossed through the membrane as shown WRR of 3.26 (Fig. 2) and transmission rates around 27% (Fig. 1). Hence DM and proteins detected at filtrates from runs 1 and 2, were mainly from egg white. At run 1, more egg white proteins filtrated, producing higher transmission rates (Fig. 1), even like run 10 (protein transmission rate: p = 0.42), and lower CF (Fig. 2). In contrast, it seems that whole egg plasma employed at run 2 had more egg yolk compared to that of run 1. Consequently, after UF, retentate was more concentrated, producing higher CF (Fig. 2); even protein CF was equal to run 8 (p = 0.63). But not higher transmission rates were detected (Fig. 1). Differences between both runs were mainly due to the biological intrinsic variability between initial eggs, food from animal origin. DM and proteins transmission rates of run 6 (0.2 µm) were higher (p < 0.05) than run 7 (0.65 µm); in contrast, cholesterol transmission rates, mean TMP and mean filtrate fluxes were not significantly different (p > 0.05) although run 6 had nearly 2.5 times more surface area (Table 1). These runs can be compared to egg yolk plasma (run 9) and egg white (run 11) MF. Filtrate and retentate from egg yolk plasma were close to reach an equilibrium of concentrations (Table 3), producing transmission rates close to 100% (Fig. 1) and CF around 0.5 (Fig. 2). Final concentration at retentate was near to that of the plasma diluted introduced in the hollow fibers. This mean that almost every compound at plasma egg yolk was able to cross through the membrane of 0.65 µm - mPES. Therefore, this MF for plasma egg yolk was poorly selective. In contrast, some egg white components were retained at 0.65 µm. At run 11, initial weight was reduced more than 4 times with a TMP of 10.76 kPa only, but transmission rates were not over 50% (Fig. 1); then it did not achieve an equilibrium concentration with filtrate. Juxtaposed to egg white (run 11) or egg yolk plasma MF (run 9), when whole egg plasma was microfiltrated with mPES and a minimum pore size of 0.2 µm (run 6 and 7), CF for DM and cholesterol were over 1.2 (Fig. 2). A concentration process was developed. Specially egg yolk lipids, because CF of proteins were about 1 and 1.1 for run 6 and 7, respectively. In Table 3, plasma from whole egg had similar DM to egg yolk plasma (p = 1), about 21%; and double DM compared to egg white. Unlike DM, whole egg plasma started with similar proteins (p = 0.21) to egg white (around 86 mg/ml) and nearly twice proteins compared to egg yolk plasma (about 47 mg/ ml). Filtration of plasma was hindered by the presence of proteins from egg whites. Random aggregation of egg white’s proteins was visualized, caused by only 0.5% of added egg yolk (Yao et al., 2014). In our work, these aggregates decreased transmission rates compared to egg yolk plasma (Fig. 1) and provoked differences in mean TMP (Table 2). 3.2. Colour results L*, a* and b* colour parameters were determined in egg whites, plasmas, water diluent, retentates and filtrates. Values from retentates were compared to those of initial samples. While filtrates were compared to water. Results are shown in Table 4. Means having same letter within the column did not differ significantly from each other according Fig. 1 – Transmission rates for dry matter (DM) in red, proteins in blue and cholesterol in green. Standard deviations are given as error bars (some are too small to be observed). For each parameter (N ≥ 2), means with different letters differed significantly from each other according to Tukey’s test (p < 0.05). Runs 1–7 employed whole egg plasma; runs 8 and 9 employed egg yolk plasma; runs 10 and 11 employed egg white. Run 1, 2, 8 and 10 employed membranes of 750 kDa made of mPES (modified polyethersulfone). Run 3: 0.1 µm of ME (mixed ester). Run 4 and 5: 0.2 µm of PES (polyethersulfone). Run 6: 0.2 µm of mPES. Run 7, 9 and 11: 0.65 µm of mPES. 103 Food and Bioproducts Processing 140 (2023) 99–109 to Tukey’s test, then p was over 0.05. Initial egg white had some compounds that gave a soft yellow appearance with a b* value around 3. Instead, egg yolk plasma had carotenoids that decreased lightness (L*) and increased redness (a*) and yellowness (b*). When these compounds were diluted with egg white, like it was in whole egg, a* and b* values decreased, and lightness increased as seen in Table 4. Egg yolk and white retentates were darker than initial samples (p < 0.05). Whereas whole egg retentates had similar brightness to initial whole egg plasma (p > 0.05). Only retentates with WRR over 3 (run 6 and 7) were significatively darker than initial plasma (p < 0.01). Egg white retentates were not significantly different in redness and yellowness (p > 0.05) due to the great differences with the others. However, when they were tested separately through Tukey HSD, significative differences were detected (p < 0.01). Egg white retentates were more than threefold yellowish (b* around 12) and twice redder (a* over 0) than initial albumen despite the already mentioned filtration of egg white components (Fig. 1). There were egg white coloured components that were concentrated by filtration. Concentration process should have been higher at 750 kDa (UF) than at 0.65 µm (MF). Because some of those components filtrated at higher pore size, although some were still retained at 0.65 µm. Differences in filtrates were not detected due to water dilution. All filtrates from egg white had a* negative values and b* around 1. They were significantly different to water (p < 0.05). Filtrate from egg yolk plasma UF (run 8) was not significantly different to water (p > 0.05). Then carotenoids were concentrated at retentate. This concentration resulted in a retentate as yellow as initial yolk (similar b*) (p > 0.05), but redder (higher a*) (p < 0.01). On the other side, at 0.65 µm Fig. 2 – Concentration factors (CF) for dry matter (DM) in red, proteins in blue and cholesterol in green. Standard deviations are given as error bars (some are too small to be observed). For each composition parameter (N ≥ 2), means with different letters differed significantly from each other according to Tukey’s test (p < 0.05). ME: Mixed Ester. PES: Polyethersulfone. mPES: modified Polyethersulfone. WRR: weight reduction ratio. Table 3 – Composition results of retentates, filtrates and initial samples. Run key data Dry matter (%) Proteins (mg/ml) Cholesterol (mg/g) No Pore size/ MWCO & material Retentate Filtrate Retentate Filtrate Retentate Filtrate 1 750 kDa mPES 20.24 ± 0.02 d 2.90 ± 0.01 g 70.72 ± 0.30 g 19.26 ± 0.36 e 4.55 ± 0.17 f 0 d 2 750 kDa mPES 28.43 ± 0.01 b 3.00 ± 0.0003 f 97.27 ± 0.60 b 16.69 ± 0.19 f 6.12 ± 0.12 cd 0 d 3 0.1 µm ME 28.97 ± 0.01 ab 1.58 ± 0.01 j 106.99 ± 0.92 a 8.50 ± 0.14 h 5.51 ± 0.08 e 0 d 4 0.2 µm PES 30.15 ± 0.01 a 1.66 ± 0.01 i 109.41 ± 0.98 a 9.20 ± 0.07 h 5.88 ± 0.06 de 0 d 5 0.2 µm PES 20.68 ± 0.00 d 1.86 ± 0.001 h 78.81 ± 0.61 e 10.48 ± 0.50 g 4.14 ± 0.09 fg 0 d 6 0.2 µm mPES 27.24 ± 0.03 b 8.28 ± 0.01 b 89.42 ± 0.36 c 35.94 ± 0.16 a 6.42 ± 0.12 bc 1.32 ± 0.03 b 7 0.65 µm mPES 28.64 ± 0.01 ab 6.62 ± 0.03 c 100.15 ± 0.73 b 32.93 ± 0.72 b 6.69 ± 0.03 ab 0.88 ± 0.02 c 8 750 kDa mPES 23.53 ± 0.01 c 0.46 ± 0.003 k 49.88 ± 0.22 h 2.16 ± 0.01 i 7.14 ± 0.13 a 0 d 9 0.65 µm mPES 10.69 ± 0.00 fg 10.09 ± 0.002 a 27.42 ± 0.34 i 25.20 ± 0.39 d 3.07 ± 0.11 h 2.89 ± 0.04 a 10 750 kDa mPES 14.01 ± 0.01 e 4.08 ± 0.01 e 105.81 ± 0.28 a 27.89 ± 0.79 c na na 11 0.65 µm mPES 9.80 ± 0.05 g 4.61 ± 0.01 d 74.52 ± 0.98 f 31.97 ± 1.24 b na na Whole egg plasma 21.14 ± 0.77 d 86.58 ± 2.95 d 3.77 ± 0.21 g Egg yolk plasma 21.26 ± 0.02 d 46.82 ± 0.89 h 6.16 ± 0.16 cd Egg white 11.82 ± 0.56 f 85.10 ± 2.36 d na MWCO: molecular weight cut-off. mPES: modified Polyethersulfone. ME: Mixed Ester. PES: Polyethersulfone. na: not applicable. Means with different letters within the column differed significantly from each other according to Tukey’s test (p < 0.05). Runs 1–7 employed whole egg plasma; runs 8 and 9 employed egg yolk plasma; runs 10 and 11 employed egg white. 104 Food and Bioproducts Processing 140 (2023) 99–109 (MF), carotenoids passed through the membrane. Retentate was as red as initial plasma (p > 0.05) and filtrate obtained a* like retentate (around 9). The described equilibrium concentration was also measurable at colour parameters. Filtrate was slightly yellower than retentate, but both were nearly half yellowish of initial plasma. Between retentate and filtrate, carotenoids were separated rather evenly. This colour separation was neither detected at whole egg plasma filtration. Despite carotenoids also filtrated at 0.2 and 0.65 µm with mPES membranes (run 6 and 7), retentates were redder (p < 0.01) and even yellower (only run 6) than initial plasma. As aforementioned, the interactions between whole egg plasma components not described for egg white or egg yolk plasma may be the reason. Furthermore, differences between filtration products from runs 6 or 7 could indicate that carotenoids and other coloured compounds from egg white may separate different at 0.2 µm or 0.65 µm with mPES membranes. The low transmission rates previously described for PES and ME membranes (run 3–5) explained the small differences detected at colour Tukey HSD test. Due to concentration, retentates were redder than initial plasma (a* p < 0.01). A light-yellow coloration (b* around 1) was detected at filtrates. This coloration was already described at egg white filtrates and not detected at egg yolk plasma UF (run 8). It seems that some egg white-coloured compounds filtrated at all the membranes studied. Colour differences between retentates from run 1 and 2 supported the hypothesis of an initial whole egg plasma with more egg yolk at feed 2. Because retentate 2 was redder and yellower than initial plasma (p < 0.01). While the colour of retentate 1 was not significantly different to plasma (p > 0.05). The lack of international standards for egg colour measurements (Milovanovic et al., 2021) was also evident in this work. However, the objective method employed for colour assessment was precise and allowed samples comparations. Although it cannot be used to measure initial whole egg and egg yolk due to their low transmittance, it was suitable for plasmas and albumen. The values from egg white differs from literature mainly in L* and b* (Milovanovic et al., 2021). Plasma from egg yolk or whole egg differs with literature not only because of discrepancies in standards measurements, but also because centrifugation conditions change colour parameters as carotenoids separate differently (Puertas and Vázquez, 2021b). In our work only carotenoids that remained in plasma were studied. 3.3. Interaction of whole egg compounds The components of egg yolk plasma were mainly isolated between 750 kDa and 650 nm. While egg white proteins were obtained along the different pore sizes. Small pore size membranes made of mPES were useful to concentrated cholesterol with carotenoids and other proteins. Then, results confirmed that centrifugation and filtration conditions did not disrupt LDL because free cholesterol was not at plasmas neither at filtration products. Cholesterol was part of LDL particles. LDL and carotenoids were indirectly appreciated. LDL through cholesterol quantification and carotenoids from colour parameters. Carotenoids are molecules that could have crossed membranes of 750 kDa. Their main absence at those filtrates together with their presence at those in which LDL was detected (b* values over 20 at runs 6, 7 and 9) suggest they were bound. Proteins of LDL present a large proportion of hydrophobic amino acids and, consequently, have a high ability of interacting by this way (Speroni et al., 2005). Other studies that removed cholesterol from egg yolk with chelates or absorbents also detected carotenoids extraction (Puertas and Vázquez, 2019a). LDL in egg yolk plasma crossed membranes of 650 nm cutoff almost freely. It nearly achieved an equilibrium concentration at both sides of the membrane. Therefore, LDL did not separate selectively. These results showed that LDL aggregates had a molecular size mainly below 0.65 µm, as described in literature (Xie et al., 2020). However, cholesterol concentrations in retentates from run 7 and 9 (Table 3) were statistically different (p < 0.01). This difference highlighted a possible LDL aggregation at whole egg plasma that avoided LDL filtration freely. Firstly, it was described the formation of large aggregates of LDL (over 1000 nm) when pH was 8 (Speroni et al., 2005). The average pH measured at initial feed were: 8.38 ± 0.12 for whole egg plasma, 6.20 ± 0.29 for egg yolk plasma and 9.29 ± 0.06 for Table 4 – Colour results for the retentates, filtrates and initial samples. Run key data Retentate Run Filtrate No Pore size/ MWCO & material L* a* b* No L* a* b* 1 750 kDa mPES 68.81 ± 0.32 cd 3.64 ± 0.06 efg 58.97 ± 0.13 bc 1 96.40 ± 0.02 bc 0.05 ± 0.00 b 1.15 ± 0.01 de 2 750 kDa mPES 62.58 ± 0.23 d 5.32 ± 0.05 ce 69.56 ± 0.13 a 2 96.22 ± 0.11 bc 0.02 ± 0.01 b 1.05 ± 0.00 de 3 0.1 µm ME 63.37 ± 0.02 d 5.98 ± 0.02 cde 58.68 ± 0.04 bc 3 95.54 ± 0.08 de 0.04 ± 0.01 b 1.24 ± 0.01 de 4 0.2 µm PES 60.52 ± 0.25 de 5.88 ± 0.00 cde 60.47 ± 0.20 ac 4 95.90 ± 0.06 cd 0.00 b 1.46 ± 0.01 d 5 0.2 µm PES 59.13 ± 0.25 de 5.46 ± 0.01 cde 57.10 ± 0.13 bc 5 96.41 ± 0.01 bc 0.00 b 0.86 ± 0.01 e 6 0.2 µm mPES 51.92 ± 0.25 e 7.04 ± 0.02 bc 63.42 ± 0.22 ab 6 87.14 ± 0.03 f -1.87 ± 0.01 e 29.67 ± 0.02 b 7 0.65 µm mPES 39.81 ± 0.08 f 7.70 ± 0.04 bc 51.18 ± 0.04 c 7 84.40 ± 0.06 g -0.37 ± 0.01 d 22.23 ± 0.01 c 8 750 kDa mPES 39.91 ± 0.95 f 13.25 ± 0.03 a 57.75 ± 0.95 bc 8 96.74 ± 0.13 ab 0.00 b 0.19 ± 0.01 f 9 0.65 µm mPES 24.49 ± 0.02 g 9.44 ± 0.48 b 30.46 ± 0.49 d 9 40.48 ± 0.44 h 9.10 ± 0.18 a 39.36 ± 0.57 a 10 750 kDa mPES 74.43 ± 0.06 bc 1.28 ± 0.03 gh 15.70 ± 0.06 e 10 95.32 ± 0.25 e -0.18 ± 0.01 c 1.07 ± 0.01 de 11 0.65 µm mPES 79.86 ± 0.19 b 0.67 ± 0.01 h 11.65 ± 0.01 ef 11 95.44 ± 0.08 de -0.25 ± 0.02 cd 1.35 ± 0.01 de Whole egg plasma 66.43 ± 3.41 d 3.36 ± 0.71 f 53.12 ± 4.16 c Water 96.89 ± 0.03 a -0.02 ± 0.01 b 0.13 ± 0.03 f Egg yolk plasma 61.88 ± 5.64 d 7.58 ± 1.55 bd 63.54 ± 1.34 ab Egg white 93.91 ± 0.46 a -0.31 ± 0.06 h 3.79 ± 0.25 f MWCO: molecular weight cut-off. mPES: modified Polyethersulfone. ME: Mixed Ester. PES: Polyethersulfone. Means with different letters within the column differed significantly from each other according to Tukey’s test (p < 0.05). Runs 1–7 employed whole egg plasma; runs 8 and 9 employed egg yolk plasma; runs 10 and 11 employed egg white. 105 Food and Bioproducts Processing 140 (2023) 99–109 egg white. Subsequently, a higher amount of these large aggregates in run 7 could explain differences; besides minute amounts of them in run 9 could justify transmission rates below 100% at run 9. These large aggregates could have formed a gel/cake layer on top of the membrane by adsorption and subsequent compression by smaller particles. This cake layer led to membrane blockage (Nikolay et al., 2020). Secondly, the presence of egg white proteins seems to be essential to explain differences between run 7 and 9. The formation of a complex between a yolk component with ovomucin was described as the cause for the decrease of the albumen foaming ability when even small quantities of egg yolk were presented (Lomakina and Míková, 2006). And as aforementioned, other authors reported these aggregates. Ovomucin is the biggest egg white protein. It is a sulfated glycoprotein that consists of a carbohydrate poor subunit (βovomucin) and a carbohydrate-rich subunit (ß-ovomucin). These subunits have molecular weights below 750 kDa (Omana et al., 2010). However, ovomucin size showed large variations according to different conditions. For example, storage time has demonstrated to vary ovomucin size between 1 and 1000 nm (Shan et al., 2020). While the effect of different pH and ionic strengths detected variations around 3–1124 nm (Sun et al., 2018). When Sun et al. (2018) analysed ovomucin hydrodynamic size at high salt concentration (150 mM NaCl), it was below 25 nm. Because the salts decreased the repulsive forces and therefore reduced ovomucin particle size. Whereas when conditions were like those employed in our work (neutral pH and low ionic strength) more than 70% of ovomucin was over 0.65 µm. At this pH, ovomucin had net negative charge from aspartic and glutamic acids in protein part, and sialic acid and sulfate in the glycan part. These negative charges cause repulsion that could stretch ovomucin chains. In our study, salts were not added, so ovomucin average size may have been over 0.65 µm. Subsequently, much of the ovomucin did not filtrate at none of the filtrations studied. Then, it was probably that this glycoprotein formed aggregates as well with LDL. These aggregates would explain the lower transmission rates described for whole egg plasma (runs 6–7) compared to egg yolk plasma (run 9) (Fig. 1, p < 0.05). Moreover, ovomucin had shown hypocholesterolemic action in vivo (Nagaoka et al., 2002). It was detected that this egg white protein inhibited cholesterol uptake which could significantly reduce serum cholesterol in rats. As part of the mechanism of cholesterol lowering induced by ovomucin, it was described the decrease of solubility of cholesterol micelles that may inhibit cholesterol absorption through the direct interaction between cholesterol mixed micelles and ovomucin in jejunal epithelium (jejunal effect) (Tu et al., 2020). This association also corroborate the aggregation detected at the membrane separation processes. This effect can explain other studies were dietary cholesterol contained in whole egg was not well absorbed and did not acutely affect plasma total cholesterol concentration. However, same research detected that triglycerides in plasma rose after whole egg consumption (Kim and Campbell, 2018). And triglycerides are part of LDL particles (Anton, 2013). Therefore, further research is needed to understand the fate of LDL along the digestive tract. On the other side, it cannot be discarded the interaction of LDL with other egg white proteins forming protein aggregates. Egg white proteins consist of five major proteins, ovalbumin (54%), ovotransferrin (12%), ovomucoid (11%), lysozyme (3.5%), ovomucin (3.5%) and several minor proteins (Avramescu et al., 2008). Ovalbumin in its natural state usually exists in hydrophilic form and has poor binding ability with hydrophobic bioactive components. But under appropriate conditions, it could unfold easily, exposing its internal hydrophobic groups (Liu et al., 2022). These could bind with LDL particles increasing their size and preventing them to cross the membrane or producing a partial or complete pore blocking. Although ovomucin accounts for around 3.5% of egg white proteins and the other major proteins are below 750 kDa (Avramescu et al., 2008), protein transmission rates at runs 10 and 11 were only 26% and 43%, respectively. It is described that ovomucin is usually present in egg white as a complex with lysozyme and other egg white proteins (Omana et al., 2010). Then, ovomucin interacted with other egg white proteins, avoiding their filtration. If these complexes were similar or bigger than membrane pore size, they could have produced a partial or completed pore blocking and gel/cake layer formation. Both mechanisms could produce membrane fouling and stop filtration (Nikolay et al., 2020). Fig. 3 summarizes LDL aggregates that could be found at whole egg plasma: particles of LDL aggregated themselves, LDL bound to ovomucin and other smaller egg white proteins (< 750 kDa), and LDL bound only to smaller egg white proteins. LDL was also bound to carotenoids presented in plasmas. All these interactions were plausible due to the large proportion of hydrophobic amino acids abovementioned present at LDL apoproteins. Hydrophobic forces were predominant over electrostatic interactions because at pH studied all main proteins except lysozyme were negatively charged (Avramescu et al., 2008; Chalamaiah et al., 2017; Navidghasemizad et al., 2015). Then, electrostatic repulsion between main proteins and membranes was achieved as desired. Because it avoided protein adsorption to membrane and reduced filter cake formation (Nikolay et al., 2020). Moreover, it has been suggested that hydrophobic interactions are the driving force for protein binding. Because at protein−protein interfaces nonpolar residues are typically concentrated at the center and charged residues are localized at the rim (Zhou and Pang, 2018). Fig. 3 – Types of LDL aggregates at whole egg plasma: particles of LDL aggregated themselves, LDL bound to ovomucin and other smaller egg white proteins (< 750 kDa), and LDL bound only to smaller egg white proteins. Carotenoids were mainly bound to LDL. 106 Food and Bioproducts Processing 140 (2023) 99–109 The random formation of these LDL aggregates together with differences in surface areas justified the disparities described at runs 6 and 7. Therefore, egg components filtration cannot be predicted with the conditions employed. Addition of salts or pH modifications could decrease aggregation. Consequently, further studies could be led changing conditions. But protein activity should be monitored. In addition, these aggregates explained the low transmission rates detected at 0.1 µm ME and 0.2 µm PES membranes (runs 3–5). Those membranes must lack in pores with a size enough for LDL and LDL aggregates filtration. It has been shown that cut-off described by manufacturer can differ considerably from effective cut-off and pore size distribution (Nikolay et al., 2020). The possible chelate effect of egg white proteins should be deeply studied. This could be studied in vitro using dynamic gastric models that have already been used in food and pharmaceutical research (Réhault-Godbert et al., 2019). These models could assess the influence of cooking method and digestion on the bioaccessibility of nutrients and the generation of bioactive peptides. But differences between whole egg, egg yolk and egg white intake must be considered. Hollow-fiber membranes with small pore sizes (5–10 kDa cutoff) are already employed to calculate the bioaccessibility of nutrients in different in vitro digestion approaches. Dialysis through these artificial membranes is used to simulate absorptive intestinal compartments (jejunum and ileum) (Marze, 2017). However, it is known that food matrix plays a key role in nutrient bioaccessibility. The term ‘food matrix’ remarks that nutrients are contained in a medium where they may interact with other components and structures also present. Therefore, the physical state of the matrix is essential in the release, mass transfer, accessibility and biochemical stability of many food components (Oliveira et al., 2018). For example, Nimalaratne et al. (2015) demonstrated that bioaccessibility of carotenoids (lutein and zeaxanthin) was lower from scrambled egg yolks compared to boiled egg yolks. Chemical and structural changes of proteins and lipoproteins in egg yolk during different cooking conditions were underlined as a probable reason. Considering our results, attention can be drawn to possible differences between LDL aggregates. Nevertheless, as yolks were separated before stirring, the effect of egg white proteins was not studied. This work has demonstrated physical differences between whole egg plasma cholesterol and egg yolk plasma cholesterol in their matrices. Consequently, on the one hand, it reflected the importance of food matrix with egg cholesterol as an example. On the other hand, UF and MF may be useful techniques to understand other liquid food matrices. 4. Conclusion Egg products were studied through different membrane separation processes. The interactions between egg yolk and egg white components made separation difficult. Cholesterol was part of LDL particles. These particles together with carotenoids were only filtrated with mPES membranes and a minimum pore size of 0.2 µm. High uncertainty was obtained due to LDL aggregation. Three types of aggregates can justify the results: particles of LDL aggregated themselves, LDL bound to ovomucin and other smaller egg white proteins (< 750 kDa), and LDL bound only to smaller egg white proteins. As pH adjustments or salt addition could reduce aggregation, additional research could be performed to implement TFF for the separation of whole egg components. Further studies are needed to understand egg white proteins interactions with LDL and their aggregates composition. UF and MF can be useful techniques to understand liquid food matrices. Funding This work was supported by the Spanish National Plan for Scientific and Technical Research and Innovation. A University Professor Education grant (FPU 16/05128) by the Spanish Ministry of Education, Culture and Sport to author Gema Puertas is gratefully acknowledged. CRediT authorship contribution statement Gema Puertas: Investigation, Methodology, Writing – original draft, Visualization. Patricia Cazón: Validation, Writing – review & editing. Manuel Vázquez: Conceptualization, Formal analysis, Writing – review & editing, Supervision. 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. References Anton, M., 2013. Egg yolk: structures, functionalities and processes. J. Sci. Food Agric. 93, 2871–2880. https://doi.org/10. 1002/jsfa.6247 Avramescu, M.E., Borneman, Z., Wessling, M., 2008. Particleloaded hollow-fiber membrane adsorbers for lysozyme separation. J. Memb. Sci. 322, 306–313. https://doi.org/10.1016/j. memsci.2008.06.013 Chalamaiah, M., Esparza, Y., Temelli, F., Wu, J., 2017. Physicochemical and functional properties of livetins fraction from hen egg yolk. Food Biosci. 18, 38–45. https://doi.org/10. 1016/j.fbio.2017.04.002 Chay Pak Ting, B.P., Mine, Y., Juneja, L.R., Okubo, T., Gauthier, S.F., Pouliot, Y., 2011. Comparative composition and antioxidant activity of peptide fractions obtained by ultrafiltration of egg yolk protein enzymatic hydrolysates. Membr. (Basel) 1, 149–161. https://doi.org/10.3390/membranes1030149 Czarnowska-Kujawska, M., Draszanowska, A., Gujska, E., Klepacka, J., Kasinska, M., 2021. Folate content and yolk color of hen egss from different farming systems. Molecules 26, 1–14. https://doi.org/10.3390/molecules26041034 Dhineshkumar, V., Ramasamy, D., 2017. Review on membrane technology applications in food and dairy processing. J. Appl. Biotechnol. Bioeng. 3, 399–407. https://doi.org/10.15406/jabb. 2017.03.00077 Erickson, H.P., 2009. Size and shape of protein molecules at the nanometer level determined by sedimentation, gel filtration, and electron microscopy. Biol. Proced. Online 11, 32–51. https://doi.org/10.1007/s12575-009-9008-x Griffin, H.D., 1992. Manipulation of egg yolk cholesterol: A physiologist’s view. Worlds Poult. Sci. J. 48, 101–112. https://doi. org/10.1079/WPS19920010 Guiga, W., Lameloise, M.L., 2019, Membrane separation in food processing, in: Green Food Processing Techniques: 107 Food and Bioproducts Processing 140 (2023) 99–109