Preparation of milk fat globule membrane ingredients enriched in polar lipids: Composition characterization and digestive properties

In this study, milk fat globule membrane (MFGM) ingredients enriched in polar lipids was prepared using membrane filtration, including microfiltration, diafil-tration, and ultrafiltration from butter serum powder. Polar lipids (phospholipids, sterols, and gangliosides) in prepared MFGM ingredients were analyzed by 31 P NMR, GC-MS, and UPLC-MS/MS, respectively. The lipolysis degree and microstructure of MFGM ingredient and soybean lecithin-emulsification during in vitro digestion were also analyzed. Microfiltration showed higher concentration efficiency than ultrafiltration which increased 2.16% and 2.73% in phospholipid, respectively. Moreover, diafiltration concentrated more polar lipids (6.39% of phospholipid) than microfiltration. MFGM ingredients had high level of sphingomyelin (1.27–1.36%) and ration of GD 3 than GM 3 of 9.25– 9.88-fold of total gangliosides. The different lipolysis behavior between MFGM ingredient-emulsification and soybean lecithin-emulsification were correlated with their different polar lipid compositions. Phospholipids from both MFGM ingredients and soybean lecithin could help maintain the initial structure during the gastric digestion. These results could provide a scientific basic for developing high-polar lipids food, particularly infant formulas and special functional foods.


INTRODUCTION
Milk fat globule membrane (MFGM) is a 3-layer bio-membrane covered the milk fat globules (MFGs) in mammalian milk (Mcpherson and Kitchen, 1983).The lipids of MFGM mainly consists of phospholipids, glycolipids, neutral lipids, and cholesterol.During the advanced dairy processing (particular high-pressure homogenization), the natural MFGs are destroyed and part of MFGM disperse into the aqueous phase.While MFGM fragments are preferentially enriched in byproducts of the dairy industry for example buttermilk, butter serum, and whey (Rombaut et al., 2007).These MFGM ingredients produced from them could be applied in infant formula as emulsifier to help increase the surface phospholipid completeness and enhance the stability (Wu et al., 2023).Recent studies indicated that MFGM ingredients also present nutritional functions (Miocinovic et al., 2014).Some raw materials are low value by-products originating from dairy processing, for instance, butter milk and butter serum (El-Loly and Mohamed, 2011) are enriched in MFGM fragments.The MFGM ingredients are first extracted using isolation method by 4 steps that were fat globule separation, cream washing, release of MFGM fragments and its collection (Dewettinck et al., 2008, Ali, 2021) in laboratory; which were successfully to extract MFGM ingredients from raw milk (Ali et al., 2020).Recently, membrane filtration has been one of the most important methods used in the modern industry (Huang et al., 2020).Membrane separation has been used to concentrate milk phospholipids, yielding an 11.1-20.0%purity of dry matter.Morin et al. (Morin et al., 2007) used microfiltration to increase the content of phospholipid in fresh butter milk from 1.43% to 2.5%.After microfiltration with diafiltration, phospholipid content increased to 4.1%.Since the particle size of casein micelles (0.02-0.3 μm) overlap the MFGM fragments (0.4-4 μm), sodium citrate, rennet or acid coagulation were also used to separate them during the membrane filtration (Fox andBrodkorb, 2008, Holzmüller andKulozik, 2016).
The composition of MFGM ingredients contained both polar lipids (30-75%), and protein (25-70%) which are naturally good emulsifiers (Huang et al., 2020).The MFGM ingredients have been reported to incubate with Caco-2 cells so as to regulate the accumulation of triacylglycerols (Martinez-Sanchez et al., 2023).The polar lipid fraction of MFGM consisted of glycerophospholipids and sphingolipids (SM).Dietary SM was the main polar lipids in human milk and showed obvious influences on infant cognitive development (Rombaut et al., 2007, Chenyu et al., 2021).As an interesting polar lipids in human milk, gangliosides were found important to neurological and cognitive development of infants (Gurnida et al., 2012).The digestion profile of these polar lipids is still not fully understood.Recent studies indicated that SM and gangliosides were resistant to simulated digestion (Chitchumroonchokchai et al., 2023).
Most of the recent commercial polar lipids are made from soybean and egg yolk.The addition of egg yolk lecithin with MFGM ingredients could enhance the lipid globule structure and the stability of the emulsion (Yu et al., 2022) as well as influenced lipolysis degree (Ma et al., 2024).Additionally, the differences of digestion properties between MFGM and plant-derived polar lipid-emulsification were highly correlated with their compositions, especially the SM (Mathiassen et al., 2015, Nejrup et al., 2017).The predominantly compounds in most of commercial MFGM ingredients are protein (mainly casein, accounting for more than 50%), while the MFGM ingredients enriched in polar lipids are rarely reported; with the influence on emulsifying effect and digestion properties are interesting but unrevealed.
This study aims to prepare the MFGM ingredients enriched in polar lipids from butter serum and to study the polar lipids composition during extraction.We also compared the prepared MFGM ingredients with commercial ones.Furthermore, we studied and compared the digestion properties including lipolysis degree, digestive products, and microstructure of O/W emulsions prepared by MFGM ingredients and soybean lectin.This study provides a preparation method to concentrate polar lipids from butter serum powder.The polar lipids data also provide scientific basis for developing high-polar functional lipids food for example infant formulas and special-functional foods.

Sample preparation
Butter serum whey.The schematic diagram of sample preparation with membrane filtration was shown in Figure 1.Nine kinds of raw materials, buttermilk (A, B, C), butter serum (A, B, C), and butter serum powder (A, B, C) obtained from butter production line were collected.Buttermilk A, B, and C were aqueous phase in the production of anhydrous cream from cream with 1, 2 or 3 centrifugations, respectively.Butter serum A, B, and C are replicate mixtures of aqueous phase in the production of butter serum with 3 centrifugations.Butter serum powder A, B, and C are replicate powder after spraying drying from butter serum.
Butter serum powder C was selected as the raw material after comprehensive consideration with the content of phospholipid transportation, storage, etc. Butter serum powder C was reconstituted with water (10%, w/v).Then the mixture was stirred for 15 min at room temperature and then stayed for 3-4 h for fully hydrated.The method of citric acid condensation to precipitate caseins was as follows: 0.1% CaCl 2 (wt/ wt) was added, heating at 70°C for 40 min.Afterward, citric acid (2 mol/L) was added to adjust the pH to 4.  or filter cloth (200 mesh, c) were used to remove caseins from butter serum whey 1 and 2, respectively.
Microfiltration and microfiltration with diafiltration.Microfiltration (2 μm, GCM-PFC-2000) was first used to remove caseins.The permeable liquid (permeate 1), which enriched in MFGM fragments was collected.Then microfiltration (30 nm, GCM-PFC-30) was used to concentrate the fragments to make the concentrate (concentrate 2) with more polar lipids.These 2 microfiltration runs were carried out at conditions of: 23°C, 1 bar transmembrane pressure and at the equipment power of 0.7 KW.MFGM ingredient 1 (MFGM-I1) was obtained after it was freeze-dried.Method b added diafiltration (30 nm) based on the former (method a) to concentrate further (concentrate 3).And MFGM ingredient 2 (MFGM-I2) was obtained after lyophilization.
Ultrafiltration.After filtering by filter cloth, the separator precipitated from caseins was obtained.Then ultrafiltration (UF-T203000-08-1-20-PVDF, φ = 8 mm) was used to concentrate it to obtain concentrate 4. The ultrafiltration run was carried out at 25°C -28°C with a transmembrane pressure of 3 bar.

Phospholipid composition analysis
The lipid in samples was extracted by Folch method (Folch et al., 1957).Briefly, 10 mL of each milk sample or powder (1 g of butter milk powder was dissolved into 10 mL deionized water) was mixed with 30 mL of CHCl 3 /MeOH (2:1, vol/vol), and oscillated for 3 min.Then, 9 mL of KCl was added to the sample and mixed for 3 min.The mixtures were centrifuged at 8000 rpm for 10 min.The lower layer was transferred into a new tube, and the upper layer was extracted for a second time with the same method described above.The final lower layer was dried under a nitrogen stream.
The phospholipid was extracted and its detection during the processing was analyzed by using 31 P NMR according to our previous study (Wei et al., 2019).Lipid (80 mg) was added to 0.5 mL MeOH, 0.5 mL EDTA-NaOH (0.2 mol/L, pH = 7.0), and 0.5 mL CDCl 3 with TPP (124 mg/L).The mixture was oscillated for 60 s and centrifuged at 5000 rpm for 3 min.Then the lower phase was transferred to the NMR tube for analysis.The 31 P NMR conditions were as follows: probe temperature, 25°C; excitation pulse, 90°; number of data points, 32 K; relaxation delay, 12.25 s; acquisition time, 3.22 s; pulse width, 11.66 μs; and spectrometer frequency, 161.98 Hz.The data were analyzed by using the MestReNova software (Mestrelab Research SL, Spain).

Sterol composition analysis
The sterol in samples was detected by using GC-MS (ISQ) equipped with DB-5 column (30 m × 0.25 mm × 0.25 μm) following our previous study (Zhang et al., 2020) with some modifications.Lipid (50 mg) was added to 0.5 mL 5α-cholesteryl alcohol (0.1 mg/mL), 3 mL KOH-ethanol (2 mol/L).The mixture was placed in an 85°C water bath for 1 h.After the sample was cooled to room temperature, water (2 mL) and n-hexane (5 mL) were added.Then it was vortexed for 60 s and centrifuged at 5000 rpm for 10 min.The upper phase was fried under a nitrogen steam.Finally, the sample was silanized using 200 μL BSTFA and TMCS (99:1, vol/ vol) at 75°C for 30 min, and then transferred to a gas vial with a 0.22 μm organic filter.The MS conditions was as follows: carrier gas, helium; flow rate, 1.2 mL/ min; shunt ratio, 100:1; heating program, from 200°C (0.5 min) to 300°C (18 min) at 10°C/min; temperatures of injection port and ion source were 280°C and 250°C, respectively; scan range, 50-550 m/z.

Ganglioside composition analysis
Ganglioside was extracted referring to Fong's method with slight modifications (Fong et al., 2011).A 0.5 mL sample (10%, wt/vol) was added to 1 mL water, 2.7 mL MeOH and 1.35 mL CHCl 3. The mixture was oscillated for 30 min, and centrifuged at 2000 rpm for 20 min.The upper phase was transferred to a new tube.Water (0.5 mL) and 2 mL CHCl 3 /MeOH (1:2, vol/vol) were added to the lower phase, then the mixture was centrifuged at 2000 rpm for 20 min.The upper phase was also transferred to the tube.Then 1.3 mL water was added, centrifuged again after gentle inversion 3-4 times.The upper phase was removed to a 10 mL volumetric flask.KCl (0.5 mL, 0.01 mol/L) was added to the lower phase and centrifuged to get the upper phase to mix in the same flask.It was dried under a nitrogen stream, and dissolved with 0.5 mL MeOH before analysis.
Ganglioside in samples was separated with an Aquity BEH C18 (50 mm, 2.1 mm, 1.7 μm).Mobile phase A was MeOH/ammonium acetate (100:1, vol/vol), and mobile phase B was MeOH/water/ammonium acetate (85:15:1, vol/vol/v).The flow rate was 0.2 mL/min with a gradient of solvent B from 100% to 0% over 10 min, and then to 100% at 12.1 min, followed by maintenance at 100% for 2.9 min, and 2 μL samples was injected for analysis by UPLC-Q-TOF-MS (Waters, Milford, MA, USA).The typical operating parameters were set as follows: the ion source and desolvation temperatures were 100°C and 400°C, respectively; argon as collision gas, and the desolvation gas was 700 L/h; survey scan range, 50-1500 m/z.

Gross composition analysis
The gross composition, including lactose, ash and protein were determined following the Chinese national standard.The detection of lactose was according to GB 5413.5-2010,pre-titration and accurate titration with Fehling's solution.Methyl blue solution was used as the indicator, the usage volume of Fehling's solution was used to calculated the content of lactose in samples.Ash content was determined referring to GB 5009.4-2016, the gravimetric method of incineration in a furnace at 550°C was used to measure the content of ash in samples.The protein content was detected following GB 5009.5-2016,nitrogen content was using the Kjeldahl method with an automatic kjeldahl apparatus and the crude protein content was calculated by multiplying the nitrogen content by 6.38.

Preparation of O/W emulsions
Coarse O/W emulsions were prepared composed of corn oil (10%, wt/wt), and 90% emulsifying agents, which were 1% MFGM-I2 or SL with 99% Milli-Q water.They were mixed for 6 min at first, and subjected to 3 cycles of homogenization 200 bar to form an emulsion with the average volume particle size (D 4, 3 ) of 1.51 ± 0.12 μm and 1.89 ± 0.06 μm, respectively.

In vitro digestion model
The in vitro gastrointestinal digestion model used in this study was according to our previous study (Wang et al., 2022).And the conditions of digestive juices and activity of enzymes were referring to a previous study of in vitro dynamic term newborns (Oliveira et al., 2016).The samples were diluted with a content of 3% fat before they were digested.Lipids in samples at every 30 min were extracted according to Folch's method (Folch et al., 1957).And the lipolysis degree (LD) was detected by HPLC with a refractive index detector with a Sepax HP-Silica column (4.6 mm × 250 mm × 5 μm).Mobile phase was n-hexane/isopropyl alcohol/formic acid 15:1:0.003,vol/vol/v).The sample concentration was 20 mg/mL.The LD of these samples was calculated with the lipid class referring the following equation (Wang et al., 2022):

Particle size measurement
The particle size and distribution of samples were measured by laser light scattering using Zetasizer nano ZS (Malvern, US) and Microtrac S3500 (PA, USA).The refractive indexes used for the samples and water were 1.46 and 1.33, respectively (Ménard et al., 2010).Each sample was analyzed in triplicate.

Confocal laser scanning microscopy (CLSM)
A confocal Zeiss LSM880 microscope (Zeiss, Germany) with a 40 × magnification objective lens was used to observe the butter serum powder before and after treatment by citric acid, and the emulsion digestive samples.One mg/mL Rhodamine-PE (Rh-DOPE) was stained for amphiphilic compounds, 1 mg/mL NBD-PC for phosphatidycholine, and 0.1 mg/mL Nile Red for apolar lipids (Weng et al., 2021).The samples were stained with these fluorescent dyes and carried out for 20 min in the dark, the 5 μL samples were used to observe with the microscope.The images were analyzed by using the ZEN Lite black software (Zeiss, Germany).

Statistical analysis
All measurements were conducted in duplicate.The results are presented as the average with the standard deviation.IBM SPSS statistics (version 22.0, Chicago, IL, USA) was used to determine the data, in which Analysis of Variance (ANOVA) with significant differences (P < 0.05) was performed by Duncan's test.

Preparation and detection of milk fat globule membrane ingredients
Phospholipid composition in raw materials and its selection.Buttermilk, butter serum, and butter serum powder are widely used raw materials to extract MFGM ingredients.In this study, a total of 9 kinds of by-products in dairy industry were collected as raw materials.The phospholipid compositions of buttermilk (A, B, C), butter serum (A, B, C), and butter serum powder (A, B, C) are shown in Tables 1 and 2, respectively.The content of total phospholipid in butter serum is generally higher than that in butter serum powder and buttermilk.The content of total phospholipid in butter serum C reaches 1.14%, and it is larger than that of butter serum powder C (0.95%, P < 0.05) and buttermilk A (0.85%, P < 0.05).In previous studies, the content of phospholipid in buttermilk was lower than that in butter serum, which could be attributed to the different sources of raw materials (Rombaut et al., 2006a).Obvious difference of total phospholipid content in the same kind of raw materials were observed.Buttermilk A has significantly (P < 0.05) higher content of phospholipid than buttermilk C and B, which are 27.67 mg 100/mL and 19.37 mg/100 mL, respectively.This may be due to most of the smaller MFGs that existed during the 2 or 3 centrifugations were difficult to destroy.Butter serum powder C is evidently higher than butter serum powder B (0.74%) and A (0.54%, P < 0.05) in the content of total phospholipid.And butter serum C and B (106.57mg/100 mL) are significantly higher than butter serum A (P < 0.05).Compared with buttermilk and butter serum, the differences were weaker in butter serum powder, indicating their better stabilities of different batches.
In this study, butter serum powder which are obtained from spray drying was also analyzed.As shown in Table 2, the content of phospholipid of butter serum powder is 0.54-0.95%,which is lower than the study previously reported (Miocinovic et al., 2014).Butter serum has been reported to be the most promising raw material for purification for MFGM as its high polar lipid content (Rombaut and Dewettinck, 2006).However, in our study, after comprehensive consideration with phospholipid content, yield, storage, and transportation, butter serum powder C was selected as the raw materials to produce MFGM ingredient enriched in polar lipids.Moreover, 5 kinds of phospholipids, including phosphatidylcholine (PC), phosphatidylethanolamice (PE), SM, phosphatidylinositol (PI), and phosphatidylserine (PS) were detected in raw materials.In butter serum powder C, PC (0.34%) and PE (0.31%) were significantly higher than SM (0.18%, P < 0.05), followed by PS and PI.This result was consistent with the phospholipid composition in cow milk (Garcia et al., 2012).Additionally, the phospholipids in MFGM ingredients could play role in a metabolic rearrangement of lipids leading to accumulation of triacylglycerols during absorption (Martinez-Sanchez et al., 2023).
Preparation of MFGM ingredients enriched in polar lipids from butter serum powder C by membrane filtration.Microfiltration, microfiltration with diafiltration, and ultrafiltration were used to produce MFGM ingredients enriched in polar lipids (Figure 1).Studies have reported that using whey buttermilk as a starting material for concentration of MFGM components by microfiltration could help minimizing separation problems associated by the presence of caseins (Morin et al., 2006).And in our study, the content of protein in butter serum powder C (>20%) is evidently higher than that in buttermilk (<5%) and butter serum (<5%), in which casein accounts much (not shown).The protein content in butter serum powder is significantly higher than that in buttermilk and butter serum, and these results were consistent with other studies (Rombaut et al., 2006b, Miocinovic et al., 2014, Marie-Pierre et al., 2018).This condition might have occurred because of the greatly concentrated after spray drying.Moreover, in this study, citric acid was used to adjust pH to the isoelectric point of casein (4.6) to precipitate.Butter serum whey 1 and 2 were obtained after removing most of the proteins by staying overnight or filtration through a 200-mesh filter cloth, respectively (Figure 1).The content of protein in butter serum whey 1 and 2 (<2.5%, not shown) significantly decreased by about 10 times than butter serum powder C, illustrating the effect of casein removal by citric acid (Choemon and Dong-Hyun, 1990).
CLSM images and the particle size distribution of butter serum powder C (a) and its butter serum whey 1 (b) are shown in Figure 2. The amphiphilic compounds could be stained by the fluorescent dye of Rh-DOPE (yellow).We could see that polar lipids were less with large, and irregular spherical in butter serum powder C. On the contrary, the polar lipids in butter serum whey were more with smaller and denser particles.This result was similar with the particle size distribution, in which the average volume particle size of butter serum whey (about 100 nm) is smaller than that of butter serum powder C (about 200 nm).And the former had a narrower range of distribution, both of their particle sizes were within the range of studies reported previously (Huang et al., 2020).
Therefore, a diameter of 30 nm membrane of microfiltration was selected to concentrate the polar lipids in butter serum whey and ensure the flow rate.During the producing process, the distribution of phospholipid before and after concentration is shown in Figure 3.During the concentration process, butter serum whey 2 (0.55%) had a lower content of phospholipid than butter serum whey 1 (1.14%), indicating the loss of polar lipids trapped during extrusion.Although 46.5% of phospholipid was lost in permeate 1, they achieved a nearly 5-fold increase of phospholipid in concentrate 2 (5.82%) than butter serum whey 1 after microfiltration (30 nm).And its phospholipid concentration effect is better than that of ultrafiltration (c, 3.66%, P < 0.05), while worse than microfiltration with diafiltration (b, 6.39%, P < 0.05), indicating the more efficient with 2 or more combination of membrane filtration.And our results were consistent or higher concentration The values are represented as the means ± SD.PE represents phosphatidylethanolamice, SM represents sphingolipids, PS represents phosphatidylserine, PI represents phosphatidylinositol, PC represents phosphatidylcholine, and PL represents phospholipid.The "a-e" in an entry denotes significant differences (P < 0.05) between different raw materials.The values are represented as the means ± SD.PE represents phosphatidylethanolamice, SM represents sphingolipid, PS represents phosphatidylserine, PI represents phosphatidylinositol, PC represents phosphatidylcholine, and PL represents phospholipid.The "a, b, c" in an entry denotes significant differences (P < 0.05) between different raw materials.
efficiency with other studies (Miocinovic et al., 2014, María et al., 2020).Thus, in our study, microfiltration and/or with diafiltration are selected to concentrate MFGM ingredients enriched in polar lipids, MFGM-I1 and I2 were obtained from concentrate 2 and 3, respectively.Furthermore, during the producing process, PC was still the highest species of phospholipid, followed by PE, SM, PS, and PI, which was consistent with the order in butter serum powder C. Additionally, the gross compositions, mainly fat, protein, lactose, and ash of butter serum powder C, MFGM-I1, and MFGM-I2 were detected and shown in Table 3.In butter serum powder C, lactose (49.98 ± 0.04%) accounted the most, followed by protein, fat, and ash.After concentrated by microfiltration, there was a 30% of fat increased and 22% of protein decreased in MFGM-I1 (P < 0.05) compared with butter serum powder C.And after concentrated by microfiltration with diafiltration, there was a more about 25% of lactose (15.35 ± 1.22%, P < 0.05) decreased, and an approximately 25% of fat (64.22 ± 3.31%, P < 0.05) increased in MFGM-I2 compared with MFGM-I1.Diafiltration could remove lactose effectively and further concentrate polar lipids from MFGM-I1 with microfiltration, illustrating its important role on concentrating.

Polar lipids composition
Phospholipids.Phospholipid species of MFGM-I1 and I2 production after microfiltration and/or with diafiltration and 4 commercial MFGM ingredients (MFGM C1-C4) were analyzed using 31 P NMR, as shown in Table 4.The MFGM ingredients I1 and I2 are rich in phospholipids (5.82-6.39%)which are higher than those of most from the commercial ones (3.59-7.95%).Both MFGM-I1/I2 and commercial MFGM ingredients had a phospholipid content with the order from high to low of PC > PE > SM > PS > PI; which was consistent with cow milk.Compared with soybean or egg lecithin, which are common materials supplied in infant formula, SM in MFGM-I1/2 was obvious in high level.SM are the most abundant species in human milk (accounting for ~30-43% of total phospholipid) (Cilla et al., 2016).Although the relative content of SM in MFGM-I (20.75-24.81% of total phospholipid) was lower than that in human milk (Bitman et al., 1984, Sala-Vila et al., 2005), there are still much higher than soybean lectin and egg phospholipids which barely contain SM.
Sterols.Among the sterols, only cholesterol was detected in both MFGM ingredients and commercial MFGM ingredients (Figure 4) by GC-MS.The higher phospholipid content the higher cholesterol content was founds in all samples.The contents of total sterol in MFGM-I1 and MFGM-I2 were 1.74 mg/g and 2.45 mg/g, respectively, which were an 8.2-and 12-fold (not shown) higher than the raw material of butter serum powder C respectively.And the sterol in MFGM-I2 was significantly higher than that in commercial MFGM ingredients (0.62-1.79 mg/g).The differences of cholesterol between prepared and commercial MFGM ingredients may due to their different sources and producing process.
Studies have reported that the addition of SM could decrease the absorption of cholesterol in rats (Eckhardt et al., 2002).This condition have occurred because SM, which showed a very high affinity for cholesterol, decreased its micellar solubilization to decrease the cholesterol monomers for uptake by the enterocyte (Noh and Koo, 2004).In our present study, the range of the ratio between SM and cholesterol in commercial MFGM ingredients is 7.35-17.58:1,indicating the inconsistency of these materials in polar lipids.And the ratio in MFGM-I1 and MFGM-I2 was 7.30:1 and 5.55:1, respectively.The ratio of MFGM-I2 was lower than that in commercial MFGM ingredients, which would show different effects on the digestion and absorption of cholesterol when they were ingested.Moreover, it has been reported that several minor bioactive sterols, for example desmosterol, lanosterol, stigmasterol, and β-sitosterol were detected in the polar lipids of human milk (Benoit et al., 2010).Therefore, it was difficult to simulate sterols to human MFGM only by MFGM in-gredients.Polar lipids from plant (e.g., soybean or egg lecithin) were also good materials containing sterols, in which sitosterol and stigmasterol were 2 main sterols in soybean phosphatidylcholine membranes (Marsan et al., 1998).Their addition types, amount and mixture ratio need further study.
Gangliosides.There were 2 kinds of gangliosides determined in MFGM ingredients, which were disialoganglioside (GD 3 ) and monosialoganglioside (GM 3 ) which contain different content of sialic acids.These 2 kinds of gangliosides were reported as the main components in milk and dairy products, in which GD 3 is the main ganglioside in human colostrum, cow's milk and infant formulas, while GM 3 is present in trace amounts (Lacomba et al., 2010).After preparation by microfiltration and/or diafiltration, the content of gangliosides in MFGM-I1 and MFGM-I2 were a 13-and 21-fold higher than that in butter serum powder C (47.00 μg/g, not shown) respectively, indicating the high efficiency of microfiltration or with diafiltration for polar lipid concentration.Moreover, the content of ganglio-  The values are represented as the means ± SD.MFGM-I1 and MFGM-I2 represent milk fat globule membrane ingredient 1 and 2 respectively.The "a, b, c" in an entry denotes significant differences (P < 0.05) between different raw materials.side in MFGM ingredients produced in our study and from market was compared in Table 5.There was a big range of the content of ganglioside in commercial MFGM ingredients (76.47-443.59μg/mL), which could be attributed to their different material sources and producing processes.GD 3 accounted nearly 99.8% of the total ganglioside.More GM 3 was determined in MFGM ingredients produced in our study than that from market, and the content of ganglioside in MFGM-I1 and MFGM-I2 was 59.26 μg/mL and 92.61 μg/mL, respectively.GD 3 of MFGM-I1 and MFGM-I2 accounted for 90.81% and 90.25%, respectively.Studies have reported that the content of GM 3 increased up to 50% from human transitional milk to human mature milk, while the GM 3 content decreased (Li Pan and Izumi, 1999).And the relatively high content of ganglioside and the different proportion of GD 3 and GM 3 in our MFGM ingredients prepared by microfiltration or with diafiltration could meet the demands of ganglioside in different stages of infants.

Digestion behavior of corn oil covered by MFGM-I2 and SL
Lipolysis degree and lipid classes during the in vitro digestion.In the industrial production of infant formula, SL was once used widely as mainly emulsifier.However, as the composition, structure and function of human breast MFGM was revealed, more and more attention has been paid to MFGM ingredients, whose milk polar lipids were researched more similar to human MFGM (Gallier et al., 2015).Meanwhile, these 2 materials are derived from animal and plant, respectively, illustrating their representative and comparability.In this study, O/W emulsions with corn oil covered by MFGM-I2 and SL were prepared, and their lipolysis properties were compared with explore the different characterizations of these 2 materials on digestion.Compared with MFGM-I, the composition of polar lipids in SL contained more PC (17.06-29.43%)and PI (13.22-22.60%),but no SM.Additionally, no ganglioside was detected, while, more kinds of sterols, such as sitosterol (1.01-3.13mg/g) and stigmasterol (0.34-1.41 mg/g) were found (not shown).During the whole digestion, samples were obtained at 30, 60, 90, and 120 min from both the gastric and intestinal phase.The LDs of emulsions with MFGM-I2 (a) or SL (b) and their lipid classes (MFGM-I2, c; SL, d) were shown in Figure 5.We could find that there was a higher (P < 0.05) LD of MFGM-I2-emulsification than that of SL-emulsification during the first 60 min in gastric phase.This condition has occurred might be because i, MFGM-I2 had a similar spatial heterogeneity property with human MFGM, making the gastric lipase could attach fat droplets and hydrolyzed TAGs without inhibiting membrane (Bourlieu et al., 2014); ii, MFGM-I2 contained SM with more saturated fatty acids would affect gastric lipase activity (Mathiassen et al., 2015).The values are represented as the means ± SD.MFGM-I1-I2 represent milk fat globule membrane ingredient 1 and 2. MFGM-C1-C4 represent commercial milk fat globule membrane ingredient.The "a-d" in an entry denotes significant differences (P < 0.05) between different milk fat globule membrane ingredients.The values are represented as the means ± SD.MFGM-I1-I2 represent milk fat globule membrane ingredient 1 and 2. MFGM-C1-C4 represent commercial milk fat globule membrane ingredient.The "a-d" in an entry denotes significant differences (P < 0.05) between different milk fat globule membrane ingredients.
Moreover, there was no significant difference (P > 0.05) of LD between MFGM-I2-emulsification (22.48%) and SL-emulsification (21.10%, at the end of gastric digestion.However, a evident difference of LD (P < 0.05) was found when they were delivered into the intestinal phase, indicating the activity of pancreatin.And the LD of SL-emulsification was higher than MFGM-I2emulsification at I60 and I90 (P < 0.05), which could be due to the other constituents in MFGM-I2, e.g., proteins, whose proteolysis products would influence the anchoring of lipase (Luo et al., 2018).It was reported that bovine milk phospholipids-emulsification caused higher pancreatic lipase activity when the particles had been pre-treated with gastric lipase (Mathiassen et al., 2015).And in our present study, the LDs of MFGM-I2 and SL-emulsification reached 84.79% and 79.03% respectively at the end of the whole digestion (I120), indicating the higher hydrolysis degree of emulsion covered with MFGM-I2 than SL.
The lipid class of TAGs in these 2 emulsions decreased during the gastric digestion, while DAGs and FFAs increased.No MAGs were detected because of the RGE we used was tending to hydrolyze sn-3 FA, which is more homogenous with human gastric lipase  (Moreau et al., 1988).In the intestinal phase, MAGs were detected but with unstable content, which was attributed to the pancreatin lipase we used are sn-1/3 regiospecificity (Carrière et al., 1997).And DAGs could also be digested into MAGs and FFAs, which were consistent with studies previously reported (de Oliveira et al., 2016).
Particle size and microstructure during the in vitro digestion.The average volume diameter (D 4, 3 ) and their corresponding microstructure of these 2 emulsions during the whole digestion are shown in Figure 6.With the extension of gastric digestion, both MFGM-I2-(a) and SL-emulsification (b) had an increased trend of D 4, 3 .The particle size distribution of MFGM-I2emulsification showed mainly 2 peaks (not shown).At the first 30 min, SL-had the similar particle size distribution to MFGM-I2-emulsification with 0.01-1 μm and 1-10 μm.While MFGM-I2-emulsification had more stable particle size than that of SL-emulsification at 90 min (1-2 μm), indicating the glycoproteins on MFGM might prevent the coalescence of the globules, whose gastric digestion process was similar to the human milk fat digested in healthy infants (Xuan et al., 2020).In addition, the particle size distribution kept consistent with that in the gastric phase at the first 30 min in the intestinal phase.And there were large particle sizes (>10μm) and fat globule aggregation observed from 60 min, which would be attributed to the effects of phospholipid and protein interactions on the oil droplet microstructure and emulsion stability (Chen et al., 2020).And these phenomena were consistent with our previous study (Weng et al., 2021).
There was no significant but a tendency aggregation of both MFGM-I2 and SL-emulsification in the gastric phase, which could be because the acidic condition (Bourlieu et al., 2015).Phospholipids still existed and played an important role in protecting the MFGs from aggregation.The number of MFGs was significantly reduced and their regular spheres evidently destroyed when delivered to the intestinal phase, which could be attributed to the greatly hydrolyzed TAGs (Figure 5).And these results were consistent with the lipolysis degree and particle size shown above.Additionally, the large and aggregated MFGs observed from I30 might also be the formation of micelles/aggregates of bile salts with phospholipids (18-3000 nm or 3-10 μm) (Hofmann andBorgstrom, 1962, Rigler et al., 1988), which could play a crucial role in nutrition by transporting of the fat-soluble nutrients to the mucosa of the small intestine (Maldonado-Valderrama et al., 2011).

CONCLUSION
In this study, MFGM ingredients enrich in polar lipids were extracted from butter serum powder.Microfiltration resulted in higher phospholipid concentration compared with ultrafiltration.Prepared MFGM-I1/I2 and commercial MFGM-C1 to C4 showed a same phospholipid content with the order of PC > PE > SM > PS > PI, which were consistent with cow milk.MFGM ingredients had high level of sphingomyelin which are the polar lipids accounted highest content in human milk.Cholesterol was detected in MFGM ingredients as the kind of sterol.MFGM-I1 and MFGM-I2 had higher GM 3 than commercial MFGM ingredients, which could be used in different stages of infants with their different demands on gangliosides.MFGM-I2-emulsification had a different lipolysis behavior with SL-emulsification during digestion, which could be attributed to their polar lipid compositions.Phospholipids from both MFGM ingredients and soybean lecithin could protect the MFGs from aggregation during the gastric diges- tion.Comprehensive simulation of polar lipids between infant formula and human milk, including the source and adding ratio from animal or plant, and the digestion and absorption function would be researched in the future.
Figure 1.A schematic diagram of preparation of milk fat globule membrane ingredients enriched in polar lipids with membrane filtration: a) microfiltration; b) microfiltration with diafiltration; c) ultrafiltration.MFGM-I1 and MFGM-I2 represent milk fat globule membrane ingredient 1 and 2 respectively.

Figure 2 .
Figure 2. The CLSM images and particle size distribution of buttermilk powder C (a) and buttermilk whey 1 (b) after protein removal by citric acid condensation.

Figure 3 .
Figure 3.The content of phospholipids (%) in milk fat globule membrane ingredients before and after concentration process during different processing including (a) microfiltration; (b) microfiltration with diafiltration; (c) ultrafiltration.
Zhao et al.: Properties of MFGM enriched in polar lipids
Zhao et al.: Properties of MFGM enriched in polar lipids Zhao et al.: Properties of MFGM enriched in polar lipids

Table 1 .
Zhao et al.: Properties of MFGM enriched in polar lipids The content and comparison of phospholipids (mg 100/mL) in different raw milk materials of buttermilk A, B, C and butter serum A, B, C

Table 2 .
The content and comparison of phospholipids (%) in raw milk materials of butter serum powder A,

Table 3 .
The content and comparison of gross composition of fat, protein, lactose, and ash (%) in butter serum powder C and milk fat globule membrane ingredients

Table 4 .
The content and comparison of phospholipids (%) of milk fat globule membrane ingredient 1, 2 and from market

Table 5 .
The